Tips - Principles of Geochemistry PDF

Principles of Geochemistry
Principles of
Geochemistry
GIULIO OT TONELLO
Columbia University Press / New York

Columbia University Press
Publishers Since 1893
New York Chichester, West Sussex
Copyright 䉷 1997 Columbia University Press

All rights reserved
Italian edition, Principi di geochimica 䉷 1991 Zanichelli Editore S.p.A.
Library of Congress Cataloging-in-Publication Data
Ottonello, Giulio.
[Principi di geochimica. English]
Principles of geochemistry / Giulio Ottonello.
p. cm.
Includes bibliographical references (p. 817) and index.
ISBN 0-231-09984-3 (acid-free paper)
1. Geochemistry. I. Title.
QE515.08813 1997
551.9—dc20 96–23987
CIP
Casebound editions of Columbia University Press books

are printed on permanent and durable acid-free paper.
Printed in the United States of America
c 10 9 8 7 6 5 4 3 2 1
CONTENTS
Preface xi
PA R T O N E
Geochemistry of Crystal Phases
CHAPTER ONE
Elemental Properties and Crystal Chemistry 3
1.1 General Information on Atomic Structure and the Nature
of Electrons 3
1.2 Periodic Properties of the Elements 10
1.3 Generalities on the Concept of Chemical Bond in Solids 15
1.4 Ionic Bond and Ionic Radius 16
1.5 Pauling’s Univalent Ionic Radii 19
1.6 Covalent Bond and Covalent Radius 20
1.7 Electronegativity and Fractional Ionic Character 21
1.8 Ionic Poliarizability and van der Waals Bond 26
1.9 Crystal Radius and Effective Distribution Radius (EDR) 29
1.10 Effective Ionic Radii 32
1.11 Forces Interacting Between Two Atoms at Defined
Distances 43
1.12 Lattice Energy of an Ionic Crystal 45
1.13 Born-Haber-Fayans Thermochemical Cycle 52
1.14 Internal Energy of an Ionic Crystal and Temperature 54
1.15 Internal Energy and Pressure 58
1.16 Crystal Field Theory 63
1.17 Ligand Field Theory 73
1.18 The Gordon-Kim Modified Electron Gas (MEG) Model 81
1.19 Polarization Energy 87
/ v
vi / Contents
CHAPTER TWO
Concepts of Chemical Thermodynamics 91
2.1 General Concepts and Definitions 91
2.2 Gibbs Free Energy and Chemical Potential 92
2.3 Gibbs Free Energy of a Phase as a Function of the
Chemical Potential of Its Components 93
2.4 Relationships Between Gibbs Free Energy and Other
Thermodynamic Magnitudes 97
2.5 Partial Molar Properties 100
2.6 Gibbs Phase Rule 101
2.7 Stability of the Polymorphs 103
2.8 Solid State Phase Transitions 107
2.9 Standard State and Thermodynamic Activity 113
2.10 Calculation of Gibbs Free Energy of a Pure Phase in
Various P-T Conditions 117
2.11 Gibbs-Duhem Equation 118
2.12 Local Equilibrium, Affinity to Equilibrium, and Detailed
Balancing 119
CHAPTER THREE
Thermochemistry of Crystalline Solids 122
3.1 Entropy, Heat Capacity, and Vibrational Motion of
Atoms in Crystals 122
3.2 Heat Capacity at Constant P and Maier-Kelley Functions 131
3.3 Entropy and Heat Capacity from Vibrational Spectra:
Kieffer’s Model 135
3.4 Empirical Estimates of Heat Capacity at Constant
Pressure 141
3.5 Empirical Estimates of Standard State Entropy 148
3.6 Estimates of Enthalpy and Gibbs Free Energy of
Formation from Elements by Additivity: the
⌬O2⫺ Method 151
3.7 Gibbs Free Energy of a Phase at High P and T, Based on
the Functional Forms of Heat Capacity, Thermal
Expansion, and Compressibility 155
3.8 Solid Mixture Models 157
3.9 General Equations of Excess Functions for Nonideal
Binary Mixtures 168
3.10 Generalizations for Ternary Mixtures 170
3.11 Solvus and Spinodal Decomposition in Binary Mixtures 173
Contents / vii
3.12 Spinodal Decomposition and Critical Nucleation Radius 178

3.13 Chemical Solvus and Coherent Solvus 182
CHAPTER FOUR
Some Concepts of Defect Chemistry 185
4.1 Extended Defects 185
4.2 Point Defects 187
4.3 Intrinsic Disorder 187
4.4 Extrinsic Disorder 189
4.5 Point Impurities 192
4.6 Energy of Formation of Point Defects 192
4.7 Defect Concentration as a Function of Temperature
and Pressure 196
4.8 Associative Processes 198
4.9 Stability of a Crystalline Compound in the Presence of
Defect Equilibria: Fayalite as an Example 202
4.10 Diffusion in Crystals: Atomistic Approach 205
4.11 Diffusion and Interdiffusion 212
CHAPTER FIVE
Silicates 217
5.1 General Information 217
5.2 Olivines 223
5.3 Garnets 248
5.4 Pyroxenes 266
5.5 Amphiboles 299
5.6 Micas 321
5.7 Feldspars 345
5.8 Silica Minerals 371
5.9 Thermobarometric Properties of Silicates 375
PA R T T W O
Geochemistry of Silicate Melts
CHAPTER SIX
Thermochemistry of Silicate Melts 411
6.1 Structural and Reactive Properties 411
6.2 Structure of Natural Melts 419
viii / Contents
6.3 Effects of P and T Intensive Variables on Melting 424

6.4 Effects of Chemical Composition on Melting (Equation
of Lowering of Freezing Point) 426
6.5 Mixing Properties in Melts with Constant Silica Contents 428
6.6 Thermodynamic Properties of Complex Silicate Melts 433
6.7 Miscibility of Natural Melts: Ghiorso-Carmichael Model 439
CHAPTER SEVEN
Introduction to Petrogenetic Systems 449
7.1 Phase Stability Relations in Binary Systems (Roozeboom
Diagrams) 449
7.2 Extension to Ternary Systems 466
7.3 Crystallization and Fusion in an Open System: Diopside-
Albite-Anorthite as an Example 471
PA R T T H R E E
Geochemistry of Fluids
CHAPTER EIGHT
Geochemistry of Aqueous Phases 479
8.1 General Information on Structure and Properties
of Water 479
8.2 Electrolytic Nature of Aqueous Solutions:
Some Definitions 491
8.3 Models of “Ionic Coupling-Complexation” and
“Specific Interations” 491
8.4 Activities and Activity Coefficients of Electrolytes in
Aqueous Solutions 492
8.5 Ionic Strength and Debye-Hückel Theory 494
8.6 The “Mean Salt” Method 497
8.7 Activity of Dissolved Neutral Species 499
8.8 Activity of Solvent in Aqueous Solutions 501
8.9 Speciation 502
8.10 Carbonate Equilibria 510
8.11 Thermodynamic Properties of Solutes Under High P and
T Conditions: the Helgeson-Kirkham-Flowers
Approach 520
8.12 Redox Conditions, pH, and Thermodynamic Stability of
Aqueous Solutions 539
Contents / ix
8.13 Relative Predominance Limits of Aqaueous Solutes and

Eh-pH Diagrams 546
8.14 “Electronic Activity,” “System Eh,” and Other Redox
Parameters 550
8.15 Eh-pH Diagrams for Main Anionic Ligands 553
8.16 Complex Eh-pH Diagrams: Examples of Iron,
Manganese, and Copper 556
8.17 Notes on Metal-Organic Complexation 560
8.18 Redox State and Biologic Activity 575
8.19 Degree of Saturation of Condensed Phases and Affinity
to Equilibrium in Heterogeneous Systems 578
8.20 Activity Diagrams 582
8.21 Reaction Kinetics in Hydrous Environments 587
8.22 Chemistry of Seawater: Salinity, Chlorinity, Alkalinity,
and pH 600
8.23 Steady State Models and the Concept of “Time of
Residence” 607
CHAPTER NINE
Geochemistry of Gaseous Phases 612
9.1 Perfect Gases and Gaseous Mixtures 612
9.2 Real Gases at High P and T Conditions 616
9.3 The Principle of Corresponding States and Other
Equations of State for Real Gases 620
9.4 Mixtures of Real Gases: Examples in the C-O-H System 623
9.5 Volcanic Gases 626
9.6 Solubilities of Gaseous Species in Silicate Melts 631
9.7 Fluid Equilibria in Geothermal Fields 643
PA R T F O U R
Methods
CHAPTER TEN
Trace Element Geochemistry 657
10.1 Assimilation of Trace Elements in Crystals 657
10.2 Raoult’s and Henry’s Laws 657
10.3 High Concentration Limits for Henry’s Law in Crystalline
Solutions 659
10.4 Lower Concentration Limits for Henry’s Law Behavior in
Silicate Crystals 666
x / Contents
10.5 Fractionation Operated by Crystalline Solids: Onuma

Diagrams 672
10.6 Assimilation of Trace Elements in Silicate Melts 674
10.7 Limits of Validity of Henry’s Law for Trace Elements in
Silicate Melts 680
10.8 Conventional and Normalized Partition Coefficients 680
10.9 Trace Element Distribution Models for Magma Genesis
and Differentiation 688
10.10 Presence of Fluid Phases 701
10.11 “Hygromagmatophile” and “Incompatible” Elements 702
CHAPTER ELEVEN
Isotope Geochemistry 707
11.1 Isotopes in Nature 707
11.2 Nuclear Energy 713
11.3 Nuclear Decay 715
11.4 Isotopic Decay 722
11.5 Growth of Radioactive Products 723
11.6 Isotope Fractionation 726
11.7 Application I: Radiometric Dating 741
11.8 Application II: Stable Isotope Geothermometry 767
APPENDIX ONE
Constants, Units, and Conversion Factors 801
APPENDIX TWO
Review of Mathematics 805
A2.1 Exact Differentials and State Functions 805
A2.2 Review of Implicit Functions 807
A2.3 Integration on Exact Differentials 808
A2.4 Operations on Vectors 809
A2.5 Lagrange Equations, Hamiltonian 811
A2.6 Error Function 814
A2.7 York’s Algorithm for Isochrons 815
References 817
Index 863
PREFACE
A quick look at the reference list of this textbook (initially conceived for Italian
university students) is sufficient to appreciate the leading role played by North
American scientists in the development of geochemistry, and it is a great honor
for me to present my appraisal of this fascinating discipline to readers whose
mother tongue is English.
The fascination of geochemistry rests primarily on its intermediate position
between exact sciences (chemistry, physics, mathematics) and natural sciences.
The molding of the quantitative approach taken in physical chemistry, thermody-
namics, mathematics, and analytical chemistry to natural observation offers enor-
mous advantages. These are counterbalanced, however, by the inevitable draw-
backs that have to be faced when writing a textbook on geochemistry: 1) the
need to summarize and apply, very often in a superficial and incomplete fashion,
concepts that would require an entire volume if they were to be described with
sufficient accuracy and completeness; 2) the difficulty of overcoming the diffi-
dence of nature-oriented scientists who consider the application of exact sciences
to natural observations no more than models (in the worst sense of that term).
To overcome these difficulties I gradually introduce the various concepts, first
presenting them in a preliminary way, and then, whenever possible, repeating
them more rigorously and in greater detail.
The organization of this book follows the various states of aggregation of the
earth’s materials, in an order that reflects their relative importance in geology.
Five chapters deal with the crystalline state. The first chapter is preparatory, the
second and third are operative. The fourth summarizes some concepts of defect
chemistry, the role of which in geochemistry is becoming more and more impor-
tant as studies on kinetics and trace element applications advance. The fifth chap-
ter is a (necessarily concise) state-of-the-art appraisal of the major silicate min-
erals.
Two chapters are devoted to silicate melts. One is an introduction to petro-
genetic diagrams, extensively treated in petrology. Aqueous solutions are covered
in a single chapter that basically deals with electrolyte solution theory and its
applications, since any further subdivision seemed unnecessary. A single chapter
was deemed sufficient to describe the up-to-date information about gases. The
decision not to treat chemistry and equilibria in the earth’s atmosphere was dic-
/ xi
xii / Preface
tated by the consideration that such treatment would involve a preliminary evalu-
ation of organic chemistry concepts that could not be adequately covered owing
to lack of space. The last two chapters concern trace element geochemistry and
isotope geochemistry.
As regards trace elements, whose widespread use is due to the simplicity of
operation, I have attempted to show that their simplicity is often only apparent
and that deductions on a planetary scale must be made with caution. Isotope
geochemists may find the last chapter rather concise, but again, space restrictions
did not allow more extended treatment, and I tried to favor the basic concepts,
which should always be borne in mind when dealing with global systematics.
This book is dedicated to my wife Elisabeth and my son Giulio Enrico.
Principles of Geochemistry
PART ONE
Geochemistry of Crystal Phases
CHAPTER ONE
Elemental Properties
and Crystal Chemistry
1.1 General Information on Atomic Structure

and the Nature of Electrons
1.1.1 Atom Building

An atom is composed of a nucleus of protons and neutrons surrounded by an
electron cloud. Theoretically, electrons may be found at any distance from the
nucleus, although they preferentially rotate around low-energy orbits or levels.
Within a single level, various sublevels can be distinguished. [The term “level”
corresponds to “electron shell” in the Bohr model. The terminological analogy
is: shell K ⫽ level I (n ⫽ 1); shell L ⫽ level II (n ⫽ 2); shell M ⫽ level III (n ⫽
3); shell N ⫽ level IV (n ⫽ 4); and so on.] Electron levels are established according
to four quantum numbers:
1. The principal quantum number (n) is related to the energy of the electron
levels. An increase in quantum number n corresponds to an increase in
the distance of the electron from the nucleus.
2. The orbital or “azimuthal” quantum number (l ) defines “form” (i.e., “ec-
centricity of elliptical orbit”; cf. Pauling, 1948) and indicates which sub-
level is occupied by the electron. It assumes integer values between 0 and
n ⫺ 1.
3. The magnetic quantum number (ml ) indicates which orbital of a given
sublevel is occupied by the electron. For a given orbital, quantum number
l assumes integer values between ⫺1 and ⫹1. An electron with a magnetic
quantum number ml subjected to a magnetic field B increases its energy
by a term E ⫽ ␮bmlB, where ␮b is Bohr’s magneton (Zeeman effect).
4. The spin quantum number (ms ) gives the projection of the intrinsic angular
momentum of the electron, with values ⫹12 and ⫺12. Two oppositely ori-
ented spins on the same orbital are graphically represented by arrows in
opposite directions (↑↓).
/ 3
4 / Geochemistry of Crystal Phases
The development of quantum mechanics has led to the formulation of various

principles governing the structure of atoms:
1. The Aufbau or “build-up” principle states that electrons occupy orbitals

of progressively increasing energies.
2. The Pauli exclusion principle states that no more than two electrons may
occupy the same orbital, and they must have opposite spins. Based on
this principle, 2n is the maximum number of electrons compatible with a
given level.
3. Hund’s rule states that, before two opposite-spin electrons can occupy the
same orbital of a given sublevel, all the other orbitals of the same sublevel
must have been occupied by single electrons of equal spin.
Table 1.1 shows the quantum numbers for the first four levels. Table 1.2 lists
the complete electron configurations of all the elements.
1.1.2 Wavelike Behavior of Electrons

1
Like any particle, an electron has a definite mass, corresponding to 1850 of the
mass of the hydrogen atom. Nevertheless, experiments with electron beams accel-
erated by known voltage differences show that a wavelength ␭ can be associated
with the mass of the electron, so that
h
λ= , (1.1)
mv
where h is Planck’s constant and v is velocity. Equation 1.1, which is one of the
fundamental results of quantum mechanics, establishes the dual nature of elec-
trons as particles and wavelike elements.
The propagation of a wave in one dimension is obtained by the partial derivative of

the wave function
x 
ψ = ψ 0 sin 2π  − vt , (1.2)
λ 
where v is frequency (v ⫽ v / ␭ ) and t is time. Function 1.2 is periodic in t when x is

constant (each point vibrates) and periodic in x when t is constant (instantaneous picture
of a wave). Partial differentiation of function 1.2 on x results in
∂ 2ψ 4π
= − 2 ψ. (1.3)
∂ x2 λ
Elemental Properties and Crystal Chemistry / 5
Table 1.1 Quantum numbers.
n l ml ms
1 (level K) 0 (sublevel s) 0 ⫹12 ⫺12
2 (level L) 0 (sublevel s) 0 ⫹12 ⫺12
1 (sublevel p) ⫹1 ⫹12 ⫺12
0 ⫹12 ⫺12
⫺1 ⫹12 ⫺12
3 (level M) 0 (sublevel s) 0 ⫹12 ⫺12
1 (sublevel p) ⫹1 ⫹12 ⫺12
0 ⫹12 ⫺12
⫺1 ⫹12 ⫺12
2 (sublevel d ) ⫹2 ⫹12 ⫺12
⫹1 ⫹12 ⫺12
0 ⫹12 ⫺12
⫺1 ⫹12 ⫺12
⫺2 ⫹12 ⫺12
4 (level N) 0 (sublevel s) 0 ⫹12 ⫺12
1 (sublevel p) ⫹1 ⫹12 ⫺12
0 ⫹12 ⫺12
⫺1 ⫹12 ⫺12
2 (sublevel d ) ⫹2 ⫹12 ⫺12
⫹1 ⫹12 ⫺12
0 ⫹12 ⫺12
⫺1 ⫹12 ⫺12
⫺2 ⫹12 ⫺12
3 (sublevel f ) ⫹3 ⫹12 ⫺12
⫹2 ⫹12 ⫺12
⫹1 ⫹12 ⫺12
0 ⫹12 ⫺12
⫺1 ⫹12 ⫺12
⫺2 ⫹12 ⫺12
⫺3 ⫹12 ⫺12
The three-dimensional analog of function 1.2 is
 ( x2 + y2 + z 2 ) 1 2 
ψ = ψ 0 sin 2π  − vt  , (1.4)
 λ 
and it can be shown that partial derivation of equation 1.4 over the three dimensions
results in
∂ 2ψ ∂ 2ψ ∂ 2ψ 4π
+ + 2 = − 2 ψ. (1.5)
∂x 2
∂y 2
∂z λ
Let us now consider a particle with mass m, total energy E, and potential
energy ⌽. The kinetic energy of particle Ek will be
Table 1.2 Electron configuration of elements.
1 2 3 4 5 6 7
Z Element s s p s p d s p d f s p d f s p d f s
1 H 1
2 He 2
3 Li 2 1
4 Be 2 2
5 B 2 2 1
6 C 2 2 2
7 N 2 2 3
8 O 2 2 4
9 F 2 2 5
10 Ne 2 2 6
11 Na 2 2 6 1
12 Mg 2 2 6 2
13 Al 2 2 6 2 1
14 Si 2 2 6 2 2
15 P 2 2 6 2 3
16 S 2 2 6 2 4
17 Cl 2 2 6 2 5
18 Ar 2 2 6 2 6
19 K 2 2 6 2 6 1
20 Ca 2 2 6 2 6 2
21 Sc 2 2 6 2 6 1 2
22 Ti 2 2 6 2 6 2 2
23 V 2 2 6 2 6 3 2
24 Cr 2 2 6 2 6 5 1
25 Mn 2 2 6 2 6 5 2
26 Fe 2 2 6 2 6 6 2
27 Co 2 2 6 2 6 7 2
28 Ni 2 2 6 2 6 8 2
29 Cu 2 2 6 2 6 10 1
30 Zn 2 2 6 2 6 10 2
31 Ga 2 2 6 2 6 10 2 1
32 Ge 2 2 6 2 6 10 2 2
33 As 2 2 6 2 6 10 2 3
34 Se 2 2 6 2 6 10 2 4
35 Br 2 2 6 2 6 10 2 5
36 Kr 2 2 6 2 6 10 2 6
37 Rb 2 2 6 2 6 10 2 6 1
38 Sr 2 2 6 2 6 10 2 6 2
39 Y 2 2 6 2 6 10 2 6 1 2
40 Zr 2 2 6 2 6 10 2 6 2 2
41 Nb 2 2 6 2 6 10 2 6 4 1
42 Mo 2 2 6 2 6 10 2 6 5 1
43 Tc 2 2 6 2 6 10 2 6 6 1
44 Ru 2 2 6 2 6 10 2 6 7 1
45 Rh 2 2 6 2 6 10 2 6 8 1
46 Pd 2 2 6 2 6 10 2 6 10
Table 1.2 (continued)
1 2 3 4 5 6 7
47 Ag 2 2 6 2 6 10 2 6 10 1
48 Cd 2 2 6 2 6 10 2 6 10 2
49 In 2 2 6 2 6 10 2 6 10 2 1
50 Sn 2 2 6 2 6 10 2 6 10 2 2
51 Sb 2 2 6 2 6 10 2 6 10 2 3
52 Te 2 2 6 2 6 10 2 6 10 2 4
53 I 2 2 6 2 6 10 2 6 10 2 5
54 Xe 2 2 6 2 6 10 2 6 10 2 6
55 Cs 2 2 6 2 6 10 2 6 10 2 6 1
56 Ba 2 2 6 2 6 10 2 6 10 2 6 2
57 La 2 2 6 2 6 10 2 6 10 2 6 1 2
58 Ce 2 2 6 2 6 10 2 6 10 2 2 6 2
59 Pr 2 2 6 2 6 10 2 6 10 3 2 6 2
60 Nd 2 2 6 2 6 10 2 6 10 4 2 6 2
61 Pm 2 2 6 2 6 10 2 6 10 5 2 6 2
62 Sm 2 2 6 2 6 10 2 6 10 6 2 6 2
63 Eu 2 2 6 2 6 10 2 6 10 7 2 6 2
64 Gd 2 2 6 2 6 10 2 6 10 7 2 6 1 2
65 Tb 2 2 6 2 6 10 2 6 10 9 2 6 2
66 Dy 2 2 6 2 6 10 2 6 10 10 2 6 2
67 Ho 2 2 6 2 6 10 2 6 10 11 2 6 2
68 Er 2 2 6 2 6 10 2 6 10 12 2 6 2
69 Tm 2 2 6 2 6 10 2 6 10 13 2 6 2
70 Yb 2 2 6 2 6 10 2 6 10 14 2 6 2
71 Lu 2 2 6 2 6 10 2 6 10 14 2 6 1 2
72 Hf 2 2 6 2 6 10 2 6 10 14 2 6 2 2
73 Ta 2 2 6 2 6 10 2 6 10 14 2 6 3 2
74 W 2 2 6 2 6 10 2 6 10 14 2 6 4 2
75 Re 2 2 6 2 6 10 2 6 10 14 2 6 5 2
76 Os 2 2 6 2 6 10 2 6 10 14 2 6 6 2
77 Ir 2 2 6 2 6 10 2 6 10 14 2 6 7 2
78 Pt 2 2 6 2 6 10 2 6 10 14 2 6 9 1
79 Au 2 2 6 2 6 10 2 6 10 14 2 6 10 1
80 Hg 2 2 6 2 6 10 2 6 10 14 2 6 10 2
81 Tl 2 2 6 2 6 10 2 6 10 14 2 6 10 2 1
82 Pb 2 2 6 2 6 10 2 6 10 14 2 6 10 2 2
83 Bi 2 2 6 2 6 10 2 6 10 14 2 6 10 2 3
84 Po 2 2 6 2 6 10 2 6 10 14 2 6 10 2 4
85 At 2 2 6 2 6 10 2 6 10 14 2 6 10 2 5
86 Rn 2 2 6 2 6 10 2 6 10 14 2 6 10 2 6
87 Fr 2 2 6 2 6 10 2 6 10 14 2 6 10 2 6 1
88 Ra 2 2 6 2 6 10 2 6 10 14 2 6 10 2 6 2
89 Ac 2 2 6 2 6 10 2 6 10 14 2 6 10 2 6 1 2
90 Th 2 2 6 2 6 10 2 6 10 14 2 6 10 2 6 2 2
91 Pa 2 2 6 2 6 10 2 6 10 14 2 6 10 2 2 6 1 2
92 U 2 2 6 2 6 10 2 6 10 14 2 6 10 3 2 6 1 2
continued
1 2 3 4 5 6 7
93 Np 2 2 6 2 6 10 2 6 10 14 2 6 10 4 2 6 1 2
94 Pu 2 2 6 2 6 10 2 6 10 14 2 6 10 6 2 6 2
95 Am 2 2 6 2 6 10 2 6 10 14 2 6 10 7 2 6 2
96 Cm 2 2 6 2 6 10 2 6 10 14 2 6 10 7 2 6 1 2
97 Bk 2 2 6 2 6 10 2 6 10 14 2 6 10 8 2 6 1 2
98 Cf 2 2 6 2 6 10 2 6 10 14 2 6 10 10 2 6 2
99 Es 2 2 6 2 6 10 2 6 10 14 2 6 10 11 2 6 2
100 Fm 2 2 6 2 6 10 2 6 10 14 2 6 10 12 2 6 2
101 Md 2 2 6 2 6 10 2 6 10 14 2 6 10 13 2 6 2
102 No 2 2 6 2 6 10 2 6 10 14 2 6 10 14 2 6 2
103 Lw 2 2 6 2 6 10 2 6 10 14 2 6 10 14 2 6 1 2
1
EK = mv 2 = E − Φ. (1.6)
2
From equation 1.6 we obtain
m2 v 2 = 2 m E − Φ ,( ) (1.7)
which, when combined with equation 1.1, gives
1
=
2m E − Φ( ) (1.8)
λ 2
h2
Substituting the right side of equation 1.8 into wave equations 1.3 and 1.5 leads,
respectively, to
d 2ψ 8π 2 m
dx 2
+
h2
E−Φψ =0 ( ) (1.9)
and
8π 2 m
∇2ψ +
h2
(E − Φ ψ = 0, ) (1.10)
where ∇ 2 is the Laplacian operator
∂2 ∂2 ∂2
∇2 = + + (1.11)
∂ x2 ∂ y2 ∂ z2
(see appendix 2). Equation 1.9 is the Schrödinger equation for one-dimensional
stationary states, and equation 1.10 is the tridimensional form. Differential equa-
tions 1.9 and 1.10 possess finite single-valued solutions, within the domain of
squared integrable functions, only for definite values of energy (eigenvalues). For
vibrational systems, such values are given by
 1
E ( n ) =  n +  hv , (1.12)
 2
where n is the vibrational quantum number.
Applying to equation 1.9 the potential of the monodimensional harmonic oscillator

(i.e., a particle constrained to a certain equilibrium position by a force that is proportional
to its displacement from the origin), we obtain
d 2ψ ( x ) 8π 2 m  KF x2 
+ E −  ψ ( x ) = 0. (1.13)
dx 2
h2  2 
If position x of the particle corresponds to its displacement from the origin, the restoring
force acting on the particle by virtue of the harmonic potential is given by Hooke’s law:
F = − KF x, (1.14)
where KF is the force constant.

Because, in a conservative field, the force is linked to the gradient of the potential by
dΦ( x )
F =− , (1.15)
dx
by integration of equation 1.15 it follows that
KF x2
Φ( x ) = , (1.16)
2
which, substituted into equation 1.9, leads to equation 1.13.

The eigenvalues of the energy that satisfy equation 1.13 are given by
12
 K h2   1
E( n ) =  F 2  n +  . (1.17)
 4π m   2
We will later adopt function 1.17 when treating the vibrational energy of crystalline sub-
stances and its relationship to entropy and heat capacity.
A more detailed account of the Schrödinger equation can be found in physical

chemistry textbooks such as those of Hinshelwood (1951) and Atkins (1978), or
in more specialized texts such as that of Hirschfelder et al. (1954). An excellent
review of the applications of quantum mechanics to geochemistry has recently
been proposed by Tossell and Vaughan (1992).
1.2 Periodic Properties of the Elements
The periodic law, discovered by Dmitri Mendeleev, establishes “an harmonic peri-
odicity of properties of chemical individuals, dependent on their masses” (Men-
deleev, 1905). To highlight the existence of periodic properties, the various ele-
ments are arranged, according to their increasing masses, in groups (8),
subgroups (8), and periods (7) (table 1.3). A given group comprises elements with
similar properties. The periodicity of elemental properties is easily understood if
we recall the laws governing the occupancy of atomic orbitals. The periodic table
shows the electron structures of the various elements. For the sake of brevity, the
structure of the inner shells is indicated by referring to that of the preceding inert
gas. An inert gas displays no chemical reactivity (i.e., no tendency to enter into
other states of combination). This “stable” configuration corresponds to comple-
tion of all sublevels of the principal quantum number n. As we can see, a period
begins with the occupancy of a new level; elements in this position are ready to
lose one electron, with the formation of a univalent positive ion. At the end of
the period, immediately preceding the inert gas, the atoms lack a single electron
to complete the stable configuration. Such atoms tend to be transformed into
univalent negative ions by electron capture. The number of electrons that com-
plete the period is readily calculated from the following summation:
n −1
∑ ( 4l + 2 ) = 2 n 2 (1.18)
l =0
—i.e., two electrons in the first period (n ⫽ 1), eight electrons in the second
(n ⫽ 2), 18 electrons in the third (n ⫽ 3), and so on.
The first period is completed by only two elements (H, He), because the first
level has a single sublevel that can be occupied by two electrons with opposite
spin (because n ⫽ 0, l is also zero; ml may vary from ⫺l to ⫹l and is also zero;
ms ⫽ ⫹12 and ⫺12 ). The second period begins with an electron occupying the lowest
energy orbital of sublevel 2s (Li) and terminates with the addition of the sixth
electron of sublevel 2p (Ne). The pile-up procedure is analogous for the third
period, but in the fourth period this regularity is perturbed by the formation of
transition elements (Sc-Zn) and by progressive pile-up of electrons in the 3d orbit-
als, whereas the fourth shell is temporarily occupied by one or two electrons. Pile-
up in the fourth level starts again with Ga, and the period terminates with the
completion of the stable octet (Kr). The same sequence is observed in the fifth
period (Y-Cd transition elements). In the sixth period, the transition is observed
over sublevel f of the fourth shell (4 f ), generating a group of elements (rare
earths, La-Lu) of particular importance in geochemistry for the regularity of pro-
gression of reactive properties within the group (the most external shells are prac-
tically identical for all elements in the series; the progressive pile-up in the inner
fourth shell induces shrinkage of the atomic radius, known as “lanthanidic con-
traction”). In the seventh period, the transition begins with actinium (Ac) and
thorium (Th), whose external electrons occupy sublevel 6d, and continues with
the subsidiary series of protactinium (Pa-Lw), with electrons progressively occu-
pying sublevel f of the fifth shell (5 f ).
1.2.1 Periodic Properties, Ionization Potentials, and Energy

of External Orbitals
In terms of energy, the existence of periodicity is easily understood by analyzing
the expressions that relate the energy of external orbitals to quantum number n.
Under the hydrogen ion approximation (ions of nuclear charge Ze with a single
electron progressively occupying the various orbital levels), the potential acting
on the electron by virtue of the nuclear charge is given by
Ze 2
ΦC = − , (1.19)
r
where Z is the atomic number and r is the distance from the nucleus expressed in
Cartesian coordinates:
( )
12
r = x2 + y2 + z2 . (1.20)
Substitution of equation 1.19 into equation 1.9 leads to
8π 2 m  Ze 2 
∇2ψ +  E +  ψ = 0. (1.21)
h2  r 
The eigenvalues of equation 1.21 are
2π 2 me 4 Z 2
E( n ) = − , (1.22)
n2 h2
where E(n) is the energy of the various orbital levels in the hydrogen ion approxi-
mation.
Table 1.3 Mendeleev’s periodic chart and electron configuration of elements.
Figure 1.1 First ionization potentials as functions of atomic number. Adapted from Har-
vey and Porter (1976).
The energy of external orbitals in real atoms (which is numerically equivalent

to the first ionization potential expressed in eV) is reproduced by an empirical
expression of the form:
(Z − S )
2
e Ce 2
Z eff Ce
E = = (1.23)
n2 n2
where Se is a screen constant, depending on the number of inner electrons, and

Ce is a conversion factor. Equation 1.23 establishes that, with an increase in the
quantum number (table 1.2), we should observe a jump in the energy values of
the external orbitals. We recall for this purpose that the external or “valence”
electrons are those involved in the formation of atomic bonds (because the abso-
lute values of their orbital energies are much lower than those pertaining to the
inner electrons). As figure 1.1 shows, the first ionization potentials* plotted as
functions of atomic number Z exhibit marked discontinuities, corresponding to
the variations in the principal quantum number, and hence corresponding to the
change of period in Mendeleev’s periodic chart. Within the same period, we ob-
* The “ionization potential” is the amount of energy that is necessary to subtract an electron
from an atom and to bring it at infinite distance at rest. The “first ionization energy” is defined as
the amount of energy necessary to subtract a valence electron from a neutral atom. The “second
ionization energy” is the amount of energy required to subtract a second electron from an ion from
which a first electron has already been subtracted, and so on. The “total electronic energy” is the
sum of all ionization potentials for a given atom. The “electron affinity” is the amount of energy
required to add an electron to a neutral atom or to an anion. Ionization energies for all atoms are
listed in table 1.6.
serve a progressive increase in the first ionization potential, with minor disconti-
nuities for transition elements (due to the progressive filling of inner shells, which
modifies the value of screen constant Se in eq. 1.23). Recalling the definition of
“valence electron” and considering that bonding energies are related to external
orbitals, the periodicity of the reactive properties of the elements is readily under-
stood in light of equations 1.22 and 1.23.
1.3 Generalities on the Concept of Chemical Bond

in Solids
“There is a chemical bond between two atoms or groups of atoms in the case that
the forces acting between them are such as to lead to the formation of an aggre-
gate with sufficient stability to make it convenient for the chemist to consider it
as an independent molecular species” (Pauling, 1948).
We will consider four distinct types of bonds:
1. ionic bonds
2. covalent bonds
3. van der Waals bonds
4. metallic bonds
Although in nature it is rare to see solid compounds whose bonds are completely
identifiable with one of the types listed above, it is customary to classify solids
according to the dominant type of bond—i.e., ionic, covalent, van der Waals, and
metallic solids.
In ionic solids, the reticular positions are occupied by ions (or anionic
groups) of opposite charge. In NaCl, for instance (figure 1.2A), the Cl ⫺ ions are
surrounded by six Na⫹ ions. The alternation of net charges gives rise to strong
coulombic interactions (see section 1.11.1).
In covalent solids, two bonded atoms share a pair of electrons. Covalent
bonds are particularly stable, because the sharing of two electrons obeys the
double duty of completing a stable electronic configuration for the two neigh-
boring atoms. In diamond, for instance (figure 1.2B), each carbon atom shares
an electron pair with four surrounding carbon atoms, generating a stable giant
molecule that has the dimension of the crystal.
In van der Waals solids (e.g., solid Ne; figure 1.2C), the reticular positions
are occupied by inert atoms held together by van der Waals forces. The van der
Waals bond is rather weak, and the bond does not survive a high thermal energy
(Fm3m solid Ne, for instance, is stable below 20 K). The van der Waals bond also
exists, however, in molecular compounds (e.g., solid CO2 ). The reticular positions
in this case are occupied by neutral molecules (CO2 ). Within each molecule, the
bond is mainly covalent (and interatomic distances are considerably shorter), but
Figure 1.2 Various types of solids. For each solid, a sketch of the bond is given.
the various molecules are held together by van der Waals forces whose nature will
be discussed later.
In metallic solids, the reticular positions are occupied by cations immersed in
a cloud of delocalized valence electrons. In solid Na, for instance (figure 1.2D),
the electron cloud or “electron gas” is composed of electrons of the s sublevel of
the third shell (cf. table 1.2). Note that the type of bond does not limit the crystal
structure of the solid within particular systems. For example, all solids shown in
figure 1.2 belong to the cubic system, and two of them (NaCl, Ne) belong to the
same structural class (Fm3m).
1.4 Ionic Bond and Ionic Radius
The concept of “ionic radius” was derived from the observation of crystal struc-
tures understood as “regular arrays of ions of spherical undeformable symmetry.”
The structure of ionic crystals obeys pure geometrical rules and corresponds to
Figure 1.3 Interatomic distances in alkali halides, arranged in cationic and anionic series.
the energetically optimal condition of an ion surrounded by the maximum num-

ber of ions of opposite charge. The assumption of “undeformability” makes it
possible to consider interionic distance as a constant, at parity of ionic charge.
The concept of ionic radius is thus related to that of “interionic distance” (di ) by
simple addition, according to
di = r+ + r− , (1.24)
where r⫹ and r⫺ are, respectively, the ionic radii of a cation and an anion at the
interionic distance observed in a purely ionic crystal.
The concept that the ionic radius is relatively independent of the structure of
the solid arose intuitively from experimental observations carried out on alkali
halides, which are ionic solids par excellence. Figure 1.3 shows the evolution of
interatomic distances in alkali halides as a function of the types of anion and
cation, respectively. Significant parallelism within each of the two families of
curves may be noted. This parallelism intuitively generates the concept of con-
stancy of the ionic radius.
The first attempt to derive empirical ionic radii by geometrical rules applied
to ionic compounds (and metals) was performed by Bragg (1920), who proposed
a set of ionic radii with a mean internal precision of 0.006 Å. Further tabulations
Figure 1.4 Electron density (in electrons per Å3) in the (100) plane of NaCl. Reprinted
from Witte and Wölfel (1955), with kind permission of R. Oldenbourg Verlag GmbH,
München, Germany.
were proposed by Landé (1920), Wasastjerna (1923), and Goldschmidt et al.

(1926).
However, we have already seen that the Schrödinger equation denies the exis-
tence of definite limits in the geometry of atomic orbitals, and that the presence
of an electron in a particular portion of the space surrounding the nucleus is a
probabilistic concept. These two facts apparently conflict with the concept of an
ion as a spherical undeformable entity. But this conflict is only apparent, as may
be deduced from the “electron density maps” obtained for crystalline compounds
by X-ray diffraction techniques. Figure 1.4 illustrates the electron density distribu-
tion for the (100) plane in NaCl, as determined by Witte and Wölfel (1955). Con-
centric lines delineate zones of equal electron density (electrons per Å3 ), increas-
ing toward the nuclei (to a maximum of 70 electrons per Å3 for Na⫹ and 130
electrons per Å3 for Cl ⫺ ). The absolute minimum electron density between the
two nuclei defines the limits (hence the “dimensions”) of the two ions. By integra-
tion of the electron densities, it can be calculated that 17.70 electrons are present
in the region surrounding the nucleus of Cl ⫺ and 10.05 electrons are present in
the region surrounding the nucleus of Na⫹. These values, compared with the theo-
retical limits of 18 and 10 electrons, respectively, indicate that, within experimen-
tal approximation, there has been the complete transfer of one electron—and,
hence, that a completely “ionic bond” is present in the cubic structure of NaCl.
We also observe that the electron density map defines almost perfect sphericity
for both ions. The “ionic radii” of Na⫹ and Cl ⫺ in the crystalline cubic compound
NaCl thus correspond to the radii (radial extensions) of the two elements in the
condition of “ionic bond” when there is complete transfer of one electron from
Na to Cl.
1.5 Pauling’s Univalent Ionic Radii
The first attempt to derive ionic radii independently of the geometrical concept
of simple additivity was made by Pauling (1927a). His method was based on the
assumption that the dimension of an ion is determined by the distribution of its
outer electrons. As we have already seen (eq. 1.23), the “effective nuclear charge”
of an ion (Zeff ) is lower than its atomic number, due to the shielding effect of
inner electrons, expressed by screening constant Se . Values of Se may be obtained
theoretically (Pauling, 1927b), and partially by results of X-ray diffraction studies
(Pauling and Sherman, 1932). Based on such values, the univalent ionic radius
(r⫹⫺ ) can be expressed as
Cn C
r+ − = = n , (1.25)
Z − Se Z eff
where Cn is a constant determined by the principal quantum number of the outer-

most electrons in the inert gas configuration. Applying equation 1.25 to couples
of isoelectronic ions such as NaF, KCl, RbBr, and CsI, which have, respectively,
the same Cn , Pauling determined the dimension ratios of the isoelectronic ions in
the corresponding solids and derived the corresponding ionic radii by combining
obtained with observed values of interionic distance. For ions of the neon struc-
ture, for example, Se has a value of 4.52, and the values of Zeff for Na⫹ and F⫺
are thus 6.48e and 4.48e, respectively. Because Cn is identical for the two ions,
knowledge of the interionic Na⫹⫺F⫺ distance (2.31 Å) gives the univalent radii
solving the system
rF − Z eff,Na + 6.48
= = , (1.26)
rNa + Z eff,F − 4.48
rNa + + rF − = 2.31 (1.27)
which gives 1.36 Å for F⫺ and 0.95 Å for Na⫹. If we now continue to apply
equation 1.25 to multivalent ions, assuming that Cn remains unchanged at the
value of the intermediate inert gas, we obtain values for all ions with He, Ne, Ar,
Kr, and Xe structures. These radii, known as “univalent ionic radii in VI-fold
co-ordination,” “do not have absolute values such that their sums are equal to
equilibrium inter ionic distances” (Pauling, 1960) but rather “are the radii the
Table 1.4 Ionic radii in VI-fold coordination, based on univalent ionic

radii of Pauling (Huheey, 1975).
Z Ion rZ (Å) Z Ion rZ (Å) Z Ion rZ (Å)

3 Li⫺ 0.60 24 Cr6⫹ 0.52 48 Cd2⫹ 0.97
4 Be2⫹ 0.31 25 Mn2⫹ 0.80 49 In3⫹ 0.81
5 B3⫹ 0.23 25 Mn3⫹ 0.66 50 Sn2⫹ 0.93
6 C4⫹ 0.15 25 Mn4⫹ 0.60 50 Sn4⫹ 0.71
7 N5⫹ 0.13 26 Fe2⫹ 0.74 51 Sb5⫹ 0.62
11 Na⫹ 0.95 26 Fe3⫹ 0.66 55 Cs⫹ 1.69
12 Mg2⫹ 0.65 27 Co2⫹ 0.72 56 Ba2⫹ 1.35
13 Al3⫹ 0.50 28 Ni2⫹ 0.69 57 La3⫹ 1.14
14 Si4⫹ 0.42 29 Cu⫹ 0.96 58 Ce3⫹ 1.07
15 P5⫹ 0.35 29 Cu2⫹ 0.72 58 Ce4⫹ 0.94
16 S6⫹ 0.30 30 Zn2⫹ 0.74 72 Hf 4⫹ 0.78
17 Cl7⫹ 0.27 31 Ga3⫹ 0.62 74 W5⫹ 0.62
19 K⫹ 1.33 32 Ge4⫹ 0.53 80 Hg2⫹ 1.10
20 Ca2⫹ 0.99 33 As5⫹ 0.46 81 Tl⫹ 1.47
21 Sc3⫹ 0.81 34 Se6⫹ 0.42 81 Tl3⫹ 0.95
22 Ti3⫹ 0.76 37 Rb⫹ 1.47 82 Pb2⫹ 1.20
22 Ti4⫹ 0.68 38 Sr2⫹ 1.13 82 Pb4⫹ 0.84
23 V2⫹ 0.88 39 Y3⫹ 0.92 90 Th4⫹ 1.02
23 V3⫹ 0.74 40 Zr4⫹ 0.79 92 U4⫹ 0.97
23 V4⫹ 0.63 41 Nb5⫹ 0.69 92 U6⫹ 0.80
23 V5⫹ 0.59 42 Mo6⫹ 0.62
24 Cr3⫹ 0.63 47 Ag⫹ 1.26
multivalent ions would possess if they were to retain their electron distributions
but enter into Coulomb interaction as if they were univalent” (Pauling, 1960).
Univalent radii may be opportunely modified to obtain effective ionic radii in real
structures (Pauling, 1960; Huheey, 1975):
rZ = r+ − z (
−2 n −1 ), (1.28)
where n is the Born exponent (cf. section 1.11.2) and z is the ionic charge. Ionic
radii in VI-fold coordination derived by Huheey (1975) are listed in table 1.4.
1.6 Covalent Bond and Covalent Radius
According to Lewis’s (1916) definition, a “covalent bond” exists whenever two

atoms are linked in a stable molecule by sharing two outer electrons. We distin-
guish between “homopolar” and “heteropolar” covalent bonds on the basis of
whether or not the bonded atoms are of the same type. Pauling (1960) has shown
that in covalent compounds the distance between two elements A and B, indepen-
dent of the nature of the two atoms, is the same in all compounds containing
Table 1.5 Nonpolar covalent radii of Sanderson (1960).
Z Ion rZ (Å) Z Ion rZ (Å) Z Ion rZ (Å)

1 H 0.32 23 V 1.18 49 In 1.44
2 He 0.93 24 Cr 1.17 50 Sn 1.40
3 Li 1.23 25 Mn 1.17 51 Sb 1.40
4 Be 0.89 26 Fe 1.17 52 Te 1.36
5 B 0.82 27 Co 1.16 53 I 1.33
6 C 0.77 28 Ni 1.15 54 Xe 2.09
7 N 0.75 29 Cu 1.17 55 Cs 2.35
8 O 0.73 30 Zn 1.25 56 Ba 1.98
9 F 0.72 31 Ga 1.26 57 La 1.69
10 Ne 1.31 32 Ge 1.22 58 Ce 1.65
11 Na 1.54 33 As 1.20 59 Pr 1.64
12 Mg 1.36 34 Se 1.17 72 Hf 1.44
13 Al 1.18 35 Br 1.14 73 Ta 1.34
14 Si 1.11 36 Kr 1.89 74 W 1.30
15 P 1.06 37 Rb 2.16 79 Au 1.34
16 S 1.02 38 Sr 1.91 80 Hg 1.49
17 Cl 0.99 39 Y 1.62 81 Tl 1.48
18 Ar 1.99 40 Zr 1.45 82 Pb 1.47
19 K 2.03 41 Nb 1.34 83 Bi 1.46
20 Ca 1.74 42 Mo 1.30 86 Rn 2.14
21 Sc 1.44 47 Ag 1.34 90 Th 1.65
22 Ti 1.22 48 Cd 1.48 92 U 1.42
both elements, regardless of the state of aggregation of the compound (i.e., solid,
liquid, or gaseous). Covalent radii are easily calculated by dividing by 2 the in-
teratomic distances observed in homopolar covalent compounds (e.g., P4, S8 ,
etc.). An internally consistent tabulation was derived by Sanderson (1960) from
interatomic distances in heteropolar covalent gaseous molecules (e.g., AlCl3 ,
CH4 , SiF4, etc.). The covalent radii of Sanderson (table 1.5) are defined as “non-
polar,” because they are obtained from compounds that do not have charge po-
larity.
1.7 Electronegativity and Fractional Ionic Character
The concept of “electronegativity” is widely used in geochemistry to define the

combinational properties of the various elements in a qualitative manner. How-
ever, it is important to stress that this concept is largely empirical, because it
varies from author to author and cannot be univocally quantified.
According to Pauling’s (1932) definition, electronegativity is “the capacity of
an atom of a molecule to attract the electrons of another atom.” The electronega-
tivity scale of Pauling (1960) is based on the differences in the binding energies
in diatomic molecules AB with respect to homeopolar compounds AA and BB,
according to
 
χ A − χ B = γ 1 2  E AB −

1
2
(
E AA + E BB  ,

) (1.29)
where ␹A and ␹B are the electronegativities of atoms A and B, EAB, EAA, and EBB
are the binding energies of compounds AB, AA, and BB, and operative constant
␥ is a term that constrains the adimensional ␹A and ␹B to change by 0.5 with each
unit valence change in the first row of the periodic table ( ␥1/2 ⫽ 0.208).
According to Mulliken (1934, 1935), electronegativity is the algebraic mean
of the first ionisation potential and of electron affinity.
According to Gordy (1946), electronegativity is represented by the value of
the potential resulting from the effect of the nuclear charge of an unshielded atom
on a valence electron located at a distance corresponding to the covalent radius
of the atom.
According to Sanderson (1960), electronegativity represents the mean elec-
tron density per unit volume of an inert element:
3Z
De = , (1.30)
4π rc3
where Z is the atomic number and rc is the covalent nonpolar radius.

Iczkowski and Margrave (1961) consider electronegativity as the local value
of a potential defined by the partial derivative of the energy of the atom W with
respect to ionic charge z:
∂W
χ( z) = . (1.31)
∂z
The most immediate use of the concept of electronegativity is in the definition

of “fractional ionic character” ( fi ). The more or less ionic character of a bond is
determined by the distribution of charges within the diatomic molecule. For a
purely ionic bond, there is complete charge transfer between the two neighboring
atoms, whereas in a purely covalent bond there is no charge transfer. The relation-
ship between electronegativity and percentage of ionicity in diatomic molecules
has been proposed by several authors in different ways. According to Pauling
(1932, 1960), the more or less ionic character of a bond is expressed by parameter
fi , derived as follows:
 1 
( )
2
f i = 1 − exp  − χ A − χB . (1.32)
 4 
According to Gordy (1950), the same parameter can be obtained by applying

fi =
1
2
(
χ A − χB . ) (1.33)
Hinze et al. (1963) suggest that charge transfer in diatomic molecules obeys
the principle of equality of energies involved in the process. If we adopt equation
1.31 as a definition of electronegativity and imagine that the amount of energy
involved in the transfer of a fraction of charge dz from atom A to atom B is
quantifiable as
 ∂WA 
dEA =   dz , (1.34)
 ∂z  z
A
then, by application of the equality principle, we have
 ∂WA   ∂WB 
dEA =   dz =   dz = dE B . (1.35)
 ∂z  z  ∂z  z
A B
A similar concept was developed by Sanderson (1960).

Electronegativity values according to Pauling and Sanderson are listed in
table 1.6, together with the first four ionization potentials and the electron affini-
ties of the various elements.
At present, the most widely used definition of fractional ionic character in solid state
chemistry is that of Phillips (1970), based on a spectroscopic approach. Phillips defined
fractional ionic character as
Ei2
fi = , (1.36)
Ei2 + E c2
where Ei (“ionic energy gap”) is obtained from the “total energy gap” (Eg ) between bond-
ing and antibonding orbitals (see section 1.17.1):
បωP
Eg =
(ε − 1) (1.37)
12
∞
( )
12
Ei = E g2 − E c2 , (1.38)
where ប␻P is the plasma frequency for valence electrons, ␧∞ is the optic dielectric constant,
and Ec corresponds to Eg for the nonpolar system in the same row of the periodic table,
with a correction for interatomic spacing.
Table 1.6 Elemental properties: first four ionization potentials (I1, I2, I3, I4, from
Samsonov, 1968; values expressed in eV; values in parentheses are of doubtful reliability);
electron affinity (e.a., from Samsonov, 1968; eV); Pauling’s electronegativity (P, from
Samsonov, 1968; eV); Sanderson’s electronegativity (S, from Viellard, 1982; adimensional).
Ionization Potential Electronegativity
Z Atom I1 I2 I3 I4 e.a. P S
1 H 13.595 – – – 0.747 2.15 3.55
2 He 24.58 54.40 – – ⫺0.53 – –
3 Li 5.39 75.62 122.42 – 0.82 1.0 0.74
4 Be 9.32 18.21 153.85 217.66 ⫺0.19 1.5 2.39
5 B 8.296 25.15 37.92 259.30 0.33 2.0 2.84
6 C 11.264 24.376 47.86 64.48 1.24 2.5 3.79
7 N 14.54 29.60 47.426 77.45 0.05 3.0 4.49
8 O 13.614 35.15 54.93 77.39 1.465 3.5 5.21
9 F 17.418 34.98 62.65 87.23 3.58 4.0 5.75
10 Ne 21.559 41.07 63.5 97.16 ⫺0.57 – –
11 Na 5.138 47.29 71.8 98.88 0.84 0.9 0.70
12 Mg 7.644 15.03 78.2 109.3 ⫺0.32 1.2 1.99
13 Al 5.984 18.82 28.44 119.96 0.52 1.5 2.25
14 Si 8.149 16.34 33.46 45.13 1.46 1.8 2.62
15 P 10.55 19.65 30.16 51.35 0.77 2.1 3.34
16 S 10.357 23.4 34.8 47.29 2.07 2.5 4.11
17 Cl 13.01 23.80 39.9 53.3 3.76 3.0 4.93
18 Ar 15.755 27.6 40.90 59.79 ⫺1.0 – –
19 K 4.339 31.81 45.9 61.1 0.82 0.8 0.41
20 Ca 6.111 11.87 51.21 67.3 – 1.0 1.22
21 Sc 6.56 12.89 24.75 73.9 – 1.3 1.30
22 Ti 6.83 13.57 28.14 43.24 – 1.5 1.40
23 V 6.74 14.2 29.7 48.0 – 1.6 1.60
24 Cr 6.764 16.49 31 (51) – 1.6 1.88
25 Mn 7.432 15.64 33.69 (53) – 1.5 2.07
26 Fe 7.90 16.18 30.64 (56) – 1.8 2.10
27 Co 7.86 17.05 33.49 (53) – 1.7 2.10
28 Ni 7.633 18.15 36.16 (56) – 1.8 2.10
29 Cu 7.724 20.29 36.83 (59) 0.29 1.9 2.60
30 Zn 9.391 17.96 39.70 (62) – 1.6 2.84
31 Ga 6.00 20.51 30.70 64.2 – 1.6 3.23
32 Ge 7.88 15.93 34.21 45.7 – 2.0 3.59
33 As 9.81 18.7 28.3 50.1 – 2.0 3.91
34 Se 9.75 21.5 32.0 42.9 3.7 2.4 4.25
35 Br 11.84 21.6 35.9 47.3 3.54 2.9 4.53
36 Kr 13.996 24.56 36.9 52.5 – – –
37 Rb 4.176 27.56 40 52.6 – 0.8 0.33
38 Sr 5.692 11.026 43.6 57.1 – 1.0 1.00
39 Y 6.38 12.23 20.5 61.8 – 1.2 1.05
40 Zr 6.835 12.92 24.8 33.97 – 1.4 1.10
41 Nb 6.88 13.90 28.1 38.3 – 1.6 –
42 Mo 7.131 15.72 29.6 46.4 – 1.8 –
43 Tc 7.23 14.87 31.9 (43) – 1.9 –
44 Ru 7.36 16.60 30.3 (47) – 2.2 –
Ionization Potential Electronegativity
Z Atom I1 I2 I3 I4 e.a. P S
45 Rh 7.46 15.92 32.8 (46) – 2.2 –
46 Pd 8.33 19.42 (33) (49) – 2.2 –
47 Ag 7.574 21.48 36.10 (52) 2.5 1.9 2.57
48 Cd 8.991 16.904 44.5 (55) – 1.7 2.59
49 In 5.785 18.86 28.0 68 – 1.7 2.86
50 Sn 7.332 14.6 30.7 46.4 – 1.8 3.10
51 Sb 8.64 16.7 24.8 44.1 ⬎2.0 1.9 3.37
52 Te 9.01 18.8 31.0 38 3.6 2.1 3.62
53 I 10.44 19.0 33 (42) 3.29 2.5 3.84
54 Xe 12.127 21.2 32.1 (45) – – –
55 Cs 3.893 25.1 34.6 (46) – 0.7 0.29
56 Ba 5.810 10.00 37 (49) – 0.9 0.78
57 La 5.614 11.433 19.166 (52) – 1.1 0.88
58 Ce 6.54 12.31 19.870 36.7 – 1.1 –
59 Pr (5.76) (11.54) (20.96) – – 1.1 –
60 Nd (6.31) (12.09) (20.51) – – 1.1 –
62 Sm 5.56 11.4 (24.0) – – 1.1 –
63 Eu 5.61 11.24 (24.56) – – 1.1 –
64 Gd 5.98 (12) (23) – – 1.1 –
65 Tb (6.74) (12.52) (22.04) – – 1.2 –
66 Dy 5.80 (12.60) (21.83) – – 1.2 –
67 Ho 5.85 (12.7) (22.1) – – 1.2 –
68 Er 6.11 (12.5) (23.0) – – 1.2 –
69 Tm 5.85 (12.4) (24.1) – – 1.2 –
70 Yb 5.90 12.10 (25.61) – – 1.2 –
71 Lu 6.15 14.7 (21.83) – – 1.2 –
72 Hf 5.5 14.9 (21) (31) – 1.3 1.05
73 Ta 7.7 16.2 (22) (33) – 1.5 –
74 W 7.98 17.7 (24) (35) – 1.7 1.39
75 Re 7.87 16.6 (26) (38) – 1.9 –
76 Os 8.7 17 (25) (40) – 2.2 –
77 Ir 9.2 17.0 (27) (39) – 2.2 –
78 Pt 8.96 18.54 (29) (41) – 2.2 –
79 Au 9.223 20.5 (30) (44) 2.1 2.4 2.57
80 Hg 10.434 18.751 34.2 (46) 1.54 1.9 2.93
81 Tl 6.106 20.42 29.8 50 2.1 1.8 3.02
82 Pb 7.415 15.03 31.93 39.0 – 1.8 3.08
83 Bi 7.277 19.3 25.6 45.3 ⬎0.7 1.9 3.16
84 Po 8.2 19.4 27.3 (38) – 2.0 –
85 At 9.2 20.1 29.3 (41) – 2.2 –
86 Rn 10.745 21.4 29.4 (44) – – –
87 Fr 3.98 22.5 33.5 (43) – 0.7 –
88 Ra 5.277 10.144 (34) (46) – 0.9 –
89 Ac 6.89 11.5 – (49) – 1.1 –
90 Th 6.95 11.5 20.0 28.7 – 1.3 –
91 Pa – – – – – 1.5 –
92 U 5.65 14.63 25.13 – – 1.7 –
Catlow and Stoneham (1983) have shown that ionic term Ei corresponds to the
difference of the diagonal matrix elements of the Hamiltonian of an AB molecule in a
simple LCAO approximation (HAA ⫺ HBB; cf. section 1.18.1), whereas the covalent energy
gap corresponds to the double of the off-diagonal term HAB —i.e.,
(H − HBB ) + 4HAB .
2
Eg = AA (1.39)
1.8 Ionic Polarizability and van der Waals Bond
The electron cloud of an ion subjected to an electric field undergoes deformations

that may be translated into displacement “d ” of the baricenters of negative
charges from the positions held in the absence of external perturbation, which
are normally coincident with the centers of nuclear charges (positive). The non-
coincidence of the two centers causes a “dipole moment,” determined by the
product of the displaced charge (Ze ) and the displacement d. The displacement
is also proportional to the intensity of the electrical field (F). The proportionality
factor ( ␣ ) is known as “ionic polarizability”:
µ d = Z e d = Fα . (1.40)
Let us consider two neighboring atoms A and B (figure 1.5) and imagine that
the electronic charge distribution is, for an infinitesimal period of time, asymmet-
rical. In this period of time, atom A becomes a dipole and induces a distortion
in the electron cloud of atom B by attracting electrons from its positive edge:
atom B is then transformed, in turn, into a dipole. All the neighboring atoms
Figure 1.5 Polarization of charges in neighboring atoms.

undergo the same process, generating attractive forces among themselves. Due to
electron orbital motion, the induced forces fluctuate with a periodicity dictated
by the rotational velocity of electrons. The van der Waals bond is the result of the
induced attractions between fluctuating dipoles. The attractions diminish with
increased molecular distance and, when distances are equal, increase with the
number of electrons within the molecule.
Free-ion polarizability calculations have been proposed by various authors,
based on the values of molecular refraction of incident light (Rm ), according to
3Rm
α = , (1.41)
4πN 0
where N 0 is Avogadro’s number.

Free-ion polarizabilities, arranged in isoelectronic series, are listed in table
1.7. A semilogarithmic plot of the listed values (figure 1.6) reveals a functional
dependence on atomic number. This dependence, quite marked, allows us to esti-
mate the free-ion polarizability for ions for which there are no precise experimen-
tal data (values in parentheses in table 1.7; estimates according to Viellard, 1982).
Figure 1.6 Free-ion polarizability as a function of atomic number. Curves are drawn for
isoelectronic series. Reprinted from Viellard (1982), Sciences Geologiques, Memoir n°69,
Université Louis Pasteur, with kind permission of the Director of Publication.
Table 1.7 Free–Ion polarizability (␣f ) arranged in isoelectronic series. Data in Å3.
N ⫽ number of electrons (adapted from Viellard, 1982).
Ion N ␣f Ion N ␣f Ion N ␣

H⫺ 2 16.86 Sc0 21 26.8 Se2⫺ 36 10.63
He0 2 0.221 Ti⫹ 21 1.67 Br⫺ 36 4.82
Li⫹ 2 0.030 V2⫹ 21 (0.55) Kr0 36 2.475
Be2⫹ 2 0.008 Cr3⫹ 21 (0.33) Rb⫹ 36 1.415–1.704
B3⫹ 2 0.003 Mn4⫹ 21 (0.24) Sr2⫹ 36 0.978
C4⫹ 2 0.0014 Y3⫹ 36 0.550
N5⫹ 2 0.007 Zr4⫹ 36 0.3704
Ti0 22 13.603–22.4 Nb5⫹ 36 0.293
N3⫺
10 28.80 V⫹ 22 1.41 Mo6⫹ 36 0.19
O2⫺ 10 2.75–3.92 Cr2⫹ 22 (0.50)
F⫺ 10 0.978 Mn3⫹ 22 (0.32)
Ne0 10 0.4162 Pd0 46 4.928–6.91
Na⫹ 10 0.148–0.54 V 0
23 11.395–19.10 Ag⫹ 46 1.856–2.682
Mg2⫹ 10 0.069–0.075 Cr⫹ 23 1.141–1.16 Cd2⫹ 46 0.926–1.012
Al3⫹ 10 0.0419 Mn2⫹ 23 0.499–0.7172 In3⫹ 46 0.702
Si4⫹ 10 0.0255 Fe3⫹ 23 0.234–0.2475 Sn4⫹ 46 0.453
P5⫹ 10 0.014 Sb5⫹ 46 0.326
S6⫹ 10 0.0112 Cr0 24 6.8–16.6 Te6⫹ 46 0.259
Cl7⫹ 10 0.008 Mn⫹ 24 1.13 I7⫹ 46 0.206
Fe2⫹ 24 0.6875
P3⫹ 12 1.35 Co3⫹ 24 0.548 Cd0 48 6.001–9.187
S4⫹ 12 0.792 In⫹ 48 –
Cl5⫹ 12 0.502 Mn0 25 8.61–14.8 Sn2⫹ 48 (3.98)
Fe⫹ 25 1.03 Co2⫹ 25 (0.66) Sb3⫹ 48 (2.88)
Si4– 18 376.9 Te4⫹ 48 (1.995)
P3– 18 41.66 I5⫹ 48 1.449
S2– 18 10.32 Fe0 26 12.60
Cl⫺ 18 3.556 Co⫹ 26 0.8446–0.896 I⫺ 54 7.41
Ar0 18 1.674 Ni2⫹ 26 0.771 Xe0 54 4.03
K⫹ 18 0.811 Cs⫹ 54 2.415
Ca2⫹ 18 0.511 Co0 27 6.80–11.113 Ba2⫹ 54 1.689
Sc3⫹ 18 0.320 Ni⫹ 27 1.329 La3⫹ 54 1.05
Ti4⫹ 18 0.200 Cu2⫹ 27 0.971–1.289 Ce4⫹ 54 0.74
V5⫹ 18 0.123
Cr6⫹ 18 0.084 Ni0 28 6.51–10.00 Er0 68 31.9
Mn7⫹ 18 0.0756 Cu⫹ 28 0.982–1.115 Tm⫹ 68 –
Zn2⫹ 28 0.4186–0.726 Yb2⫹ 68 1.718
K0 19 37.30–42.97 Ga3⫹ 28 0.213–0.477 Lu3⫹ 68 –
Ca⫹ 19 8.34 Ge4⫹ 28 0.1269–0.347 Hf 4⫹ 68 0.750
Sc2⫹ 19 – As5⫹ 28 0.083–0.243 Ta5⫹ 68 –
Ti3⫹ 19 (0.46) Se6⫹ 28 0.190 W6⫹ 68 0.4603
V4⫹ 19 (0.32) Br7⫹ 28 0.043–0.132
Au⫹ 78 2.792
Ca0 20 25.04–33.80 As3⫹ 30 1.572 Hg2⫹ 78 1.615
Sc⫹ 20 2.06 Se4⫹ 30 Tl3⫹ 78 1.074
Ti2⫹ 20 (0.72) Br5⫹ 30 0.8054 Pb4⫹ 78 0.775
V3⫹ 20 (0.40) Bi5⫹ 78 0.590
Ion N ␣f Ion N ␣f Ion N ␣

Cr4⫹ 20 –
Hg0 80 9.129
Ti⫹ 80 5.187
Pb2⫹ 80 3.439
Bi3⫹ 80 2.47
Th4⫹ 86 1.75
U6⫹ 86 1.141
1.9 Crystal Radius and Effective Distribution

Radius (EDR)
The first satisfactory definition of crystal radius was given by Tosi (1964): “In an
ideal ionic crystal where every valence electron is supposed to remain localised
on its parent ion, to each ion it can be associated a limit at which the wave func-
tion vanishes. The radial extension of the ion along the connection with its first
neighbour can be considered as a measure of its dimension in the crystal (crystal
radius).” This concept is clearly displayed in figure 1.7A, in which the radial elec-
tron density distribution curves are shown for Na⫹ and Cl ⫺ ions in NaCl. The
nucleus of Cl ⫺ is located at the origin on the abscissa axis and the nucleus of Na⫹
is positioned at the interionic distance experimentally observed for neighboring
ions in NaCl. The superimposed radial density functions define an electron den-
sity minimum that limits the dimensions or “crystal radii” of the two ions. We
also note that the radial distribution functions for the two ions in the crystal
(continuous lines) are not identical to the radial distribution functions for the
free ions (dashed lines).
The “crystal radius” thus has local validity in reference to a given crystal
structure. This fact gives rise to a certain amount of confusion in current nomen-
clature, and what it is commonly referred to as “crystal radius” in the various
tabulations is in fact a mean value, independent of the type of structure (see
section 1.11.1). The “crystal radius” in the sense of Tosi (1964) is commonly
defined as “effective distribution radius” (EDR). The example given in figure 1.7B
shows radial electron density distribution curves for Mg, Ni, Co, Fe, and Mn on
the M1 site in olivine (orthorhombic orthosilicate) and the corresponding EDR
radii located by Fujino et al. (1981) on the electron density minima.
Table 1.8 shows the differences among EDR radii, ionic radii, and crystal
radii for selected elements.
It also must be stressed that EDR may assume different values along various crystal-
lographic directions. This is clearly evident if we compare electron density maps obtained
Figure 1.7 (A) Radial electron density distribution curves for Na⫹ and Cl ⫺ in NaCl.
(B) Radial electron density distributions for Mg, Ni, Co, Fe, and Mn on M1 site in olivine.
Arrows indicate electron density minima defining EDR radii. Reprinted from Tosi (1964),
with kind permission of Academic Press Inc., Orlando, Florida (A); and from Fujino
et al. (1981), with kind permission of The International Union of Crystallography (B).
by calculation of electron density functions of neutral, spherically averaged atoms (known

as Independent Atom Model or IAM) with observed total electron densities. Figure 1.8,
for instance, makes this comparison for BeO: subtraction of IAM densities (B) from total
electron densities (A) (pseudoatom approximation; see Downs, 1991 for details) results in
a deformation density chart (C), which shows how the electron density of the IAM refer-
ence model has migrated due to bonding effects. (Solid contours in the figure are positive
accumulations of charges; short dashes are negative; long dashes mark unperturbed
zones.)
Table 1.8 Comparison of EDR radii (Fujino et al., 1981) with ionic radii
(IR) and crystal radii (CR) (Shannon, 1976) in various structures (adapted
from Fujino et al., 1981; values in Å).
Cation Phase Site EDR IR CR Coordination

Li ␣-spodumene M2 0.83 0.92 1.06 VIII
Na Giadeite M2 1.30 1.18 1.32 VIII
Mg Periclase 0.92 0.72 0.86 VI
Mg Forsterite M1 0.93 0.72 0.86 VI
Mg Forsterite M2 0.94 0.72 0.86 VI
Mg Diopside M1 0.91 0.72 0.86 VI
Mg Enstatite M1 0.89 0.72 0.86 VI
Mg Enstatite M2 0.96 0.72 0.86 VI
Ca Diopside M1 1.15 0.83 0.97 VI
Mn Tephroite M2 1.16 0.83 0.97 VI
Fe Fayalite M1 1.10 0.78 0.92 VI
Fe Fayalite M2 1.12 0.78 0.92 VI
Co CoO 1.00 0.745 0.883 VI
Co Co-olivine M1 1.03 0.745 0.883 VI
Co Co-olivine M2 1.07 0.745 0.883 VI
Ni Bunsenite 1.08 0.69 0.83 VI
Ni Ni-olivine M1 1.04 0.69 0.83 VI
Ni Ni-olivine M2 1.05 0.69 0.83 VI
Si Forsterite 0.96 0.26 0.40 IV
Si ␣-spodumene 0.96 0.26 0.40 IV
Figure 1.8 Electron density plots along (110) plane of BeO: (A) effective total electron
density (pseudoatom approximation); (B) total electron density of IAM; (C) deformation
density (pseudoatom-IAM). From Downs (1992). Reprinted with permission of Springer-
Verlag, New York.
1.10 Effective Ionic Radii
For crystal chemical purposes, it is useful to have at hand tabulations of “effective

ionic radii” to allow the reproduction of interionic distances in real crystals, inde-
pendently of precise knowledge of the structure of interest. The construction of
such tables is a stepwise process that needs to be outlined to some extent.
1.10.1 Ionic Radii and Coordination

Goldschmidt (1923) was the first to note that ionic radii vary as a function of
coordination number—i.e., with the number of first neighbors surrounding the
ion. The observed variation is fundamentally identical for all cations and, in the
case of cations coordinated by anionic ligands, results in a progressive increase in
ionic radius with increasing coordination number. As an example, figure 1.9
shows the increases in ionic radius for the various rare earth elements (and In3⫹,
Y3⫹, and Sc3⫹ ) as a function of coordination number with O2⫺.
1.10.2 Ionic Radii and Ionization Potentials

Ahrens (1952) proposed the first extended tabulation of ionic radii, partially mod-
ifying the univalent radii in the VI-fold coordination of Pauling (1927a) on the
basis of the observed correlation between ionic radius (r) and ionization potential
(I ), which can be expressed in the forms
1
I ∝ (1.42)
r
and
1
r∝ (1.43)
I2
for ions of homologous charge, and in the form
I n ∝ Zc , (1.44)
where Zc is valence and n ⱕ 2, for isoelectronic series.

By interpolation techniques based on equations 1.42 to 1.44, Ahrens deter-
mined several ionic radii not known at that time (Ag2⫹, At7⫹, Au3⫹, Br5⫹, Cl 5⫹,
Cu2⫹, Fr⫹, Ge2⫹, I5⫹, N3⫹, Np7⫹, P3⫹, Pa5⫹, Pd 2⫹, Po6⫹, Pt2⫹, Re7⫹, S4⫹, Se4⫹,
Sn2⫹, Tc7⫹, U6⫹, V2⫹ ); compare these with the preceding tabulation of Pauling
Figure 1.9 Ionic radii of rare earth ions (and In3⫹, Y3⫹, and Sc3⫹) as a function of coordi-
nation number with oxygen vc. Reprinted from Shannon (1976), with kind permission
from the International Union of Crystallography.
(table 1.4). Ahrens’s (1952) tabulation was a common tool for geochemists until
the 1970s, when the extended tabulations of Shannon and Prewitt (1969), Shan-
non (1976), and Whittaker and Muntus (1970) appeared.
1.10.3 Ionic Radii, Bond Strength, and Degree of Covalence

Pauling (1929) introduced the concept of “mean strength of the bond” (S l ) as a
ratio of the valence with coordination number vc . He also formulated the “prin-
ciple of electrostatic valence,” which establishes that the sum of the mean bond
strengths for a given coordination polyhedron (P) coincides with the valence
state—i.e.,
∑ Sl = Zc .
(1.45)
P
This last principle is not strictly observed in natural compounds, and, particularly
in silicates (Baur, 1970), deviations of up to 40% over theoretical values have been
observed. These deviations must be attributed to partial covalence of the bonds.
Pauling (1947) has also shown that, in purely covalent bonds, modification of
bond distance within a given coordination polyhedron may be associated with
the “bond number” (nl ; number of shared electrons per bond) through arbitrary
constant K l —i.e.,
∆R l = R l − R1,l = −2 K l log n l , (1.46)
where R1,l is characteristic distance.

Combination of the above principles led Donnay (1969) and Donnay and
Allmann (1970) to propose a generalized scheme for variation of bond length
with bond strength in mixed bonds (not purely ionic or purely covalent). The
equation proposed by the above authors is
Nl
R 
Sl = S 0,l  l  , (1.47)
 Rl 
where Sl is the strength of a bond with length R l , S0,l is the ideal bond strength
of a bond of length R l , coinciding with the mean of bond distances within the
coordination polyhedron, and Nl is a constant that has a single value for each
cation-anion pair and, in some instances, for each single coordination number
within a fixed cation-anion pair.
A family of empirical covariance curves between bond strength and bond
length was then derived by Brown and Shannon (1973) on the basis of equation
1.47 coupled with analysis of a large number of crystal structures. In solving the
system by means of statistical interpolation procedures, Brown and Shannon kept
the bond strength sum for a given polyhedron as near as possible to the theoreti-
cal value postulated by Pauling’s (1929) “electrostatic valence principle” (cf. eq.
1.45). They also showed that the degree of covalence for a given bond and its
relative bond strength are related by an empirical equation in the form
f cl = α l SlM l , (1.48)
where ␣l and Ml depend on the number of electrons of the cation and are constant
for isoelectronic series. The ␣l and Ml constants for cation-oxygen bonds are listed
in table 1.9.
Figure 1.10 shows the relationship between bond strength and degree of cova-
lence in cation-oxygen bonds for cations with 18, 36, and 54 electrons (note that
in table 1.9 the parameters of equation 1.48 are identical for the three isoelectronic
series). By combining equation 1.48 with a modified form of Donnay’s (1969)
equation:
Table 1.9 Parameters relating covalence

to bond strength and bond length in
equations 1.48 and 1.50 (adapted from
Brown and Shannon, 1973).
e ␣l Ml MlNl
0 0.67 1.8–2 4
2 0.60 1.73 7.04
10 0.54 1.64 7.04
18 0.49 1.57 7.04
28 0.60 1.50 9.08
36 0.49 1.57
46 0.67 1.43
54 0.49 1.57
Table 1.10 General parameters for

relationship between bond strength and
bond length in cation-oxygen bonds
(adapted from Brown and Shannon, 1973).
Cation e R1,l Nl
H 0 0.86 2.17
Li, Be, B 2 1.378 4.065
Na, Mg, Al, Si, P, S 10 1.622 4.290
K, Ca, Sc, Ti, V, Cr 18 1.799 4.483
Mn, Fe 23 1.760 5.117
Zn, Ga, Ge, As 28 1.746 6.050
−Nl
 R 
S0 , l = l  , (1.49)
 R1,l 
where R1,l and Nl are general parameters valid for isoelectronic series (with the
same anion; see table 1.10), Brown and Shannon obtained an empirical equation
relating bond length and degree of covalence:
−Ml Nl
 R 
f cl = αl  l  . (1.50)
 R1,l 
Figure 1.10 Degree of covalence ( f cl ) vs bond strength (So,l) in M-O bonds for cations
with 18, 36, and 54 electrons. Values are in valence units and scales are logarithmic. Re-
printed from Brown and Shannon (1973), with kind permission from the International
Union of Crystallography.
1.10.4 Present-Day Tabulations: Effective Ionic Radii and

Crystal Radii
Table 1.11 lists the effective ionic radii (IRs) and crystal radii (CRs) obtained by
Shannon (1976), updating the preceding tabulation of Shannon and Prewitt
(1969). In deriving ionic radii based on the values of interionic distances for more
Table 1.11 Effective crystal radii (CR) and ionic radii (IR) of Shannon (1976). CN ⫽
coordination number; SP ⫽ spin; sp ⫽ square planar; py ⫽ pyramidal; HS ⫽ high spin;
LS ⫽ low spin; data in Å.
Ion CN SP CR IR Ion CN SP CR IR
Ac3⫹ VI 1.26 1.12 Bi5⫹ VI 0.90 0.76
Ag1⫹ II 0.81 0.67 Bk3⫹ VI 1.10 0.96
IV 1.14 1.00 Bk4⫹ VI 0.97 0.83
IVsp 1.16 1.02 VIII 1.07 0.93
V 1.23 1.09 Br1⫺ VI 1.82 1.96
VI 1.29 1.15 Br3⫹ IVsp 0.73 0.59
VII 1.36 1.22 Br5⫹ IIIpy 0.45 0.31
VIII 1.42 1.28 Br7⫹ IV 0.39 0.25
Ag2⫹ IVsp 0.93 0.79 VI 0.53 0.39
VI 1.08 0.94 C4⫹ III 0.06 ⫺0.08
3⫹
Ag IVsp 0.81 0.67 IV 0.29 0.15
VI 0.89 0.75 VI 0.30 0.16
Al3⫹ IV 0.53 0.39 Ca2⫹ VI 1.14 1.00
V 0.62 0.48 VII 1.20 1.06
VI 0.675 0.535 VIII 1.26 1.12
Am2⫹ VII 1.35 1.21 IX 1.32 1.18
VIII 1.40 1.26 X 1.37 1.23
IX 1.45 1.31 XII 1.48 1.34
Am3⫹ VI 1.115 0.975 Cd2⫹ IV 0.92 0.78
VIII 1.23 1.09 V 1.01 0.87
Am4⫹ VI 0.99 0.85 VI 1.09 0.95
VIII 1.09 0.95 VII 1.17 1.03
As3⫹ VI 0.72 0.58 VIII 1.24 1.10
As5⫹ IV 0.475 0.335 XII 1.45 1.31
VI 0.60 0.46 Ce3⫹ VI 1.15 1.01
At7⫹ VI 0.76 0.62 VII 1.21 1.07
Au1⫹ VI 1.51 1.37 VIII 1.283 1.143
Au3⫹ IVsp 0.82 0.68 IX 1.336 1.196
VI 0.99 0.85 X 1.39 1.25
Au5⫹ VI 0.71 0.57 XII 1.48 1.34
B3⫹ III 0.15 0.01 Ce4⫹ VI 1.01 0.87
IV 0.25 0.11 VIII 1.11 0.97
VI 0.41 0.27 X 1.21 1.07
Ba2⫹ VI 1.49 1.35 XII 1.28 1.14
VII 1.52 1.38 Cf3⫹ VI 1.09 0.95
VIII 1.56 1.42 Cf4⫹ VI 0.961 0.821
IX 1.61 1.47 VIII 1.06 0.92
X 1.66 1.52 Cl1⫺ VI 1.67 1.81
XI 1.71 1.57 Cl5⫹ IIIpy 0.26 0.12
XII 1.75 1.61 Cl7⫹ IV 0.22 0.08
Be2⫹ III 0.30 0.16 VI 0.41 0.27
IV 0.41 0.27 Cm3⫹ VI 1.11 0.97
VI 0.59 0.45 Cm4⫹ VI 0.99 0.85
Bi3⫹ V 1.10 0.96 VIII 1.09 0.95
VI 1.17 1.03 Co2⫹ IV HS 0.72 0.58
VIII 1.31 1.17 V 0.81 0.67
continued
VI LS 0.79 0.65 Eu3⫹ VI 1.087 0.947
VI HS 0.885 0.745 VII 1.15 1.01
VIII 1.04 0.90 VIII 1.206 1.066
Co3⫹ VI LS 0.685 0.545 IX 1.260 1.120
HS 0.75 0.61 F1⫺ II 1.145 1.285
Co4⫹ IV 0.54 0.40 III 1.16 1.30
Cr2⫹ VI LS 0.87 0.73 IV 1.17 1.31
VI HS 0.94 0.80 VI 1.19 1.33
Cr3⫹ VI 0.755 0.615 F7⫹ VI 0.22 0.08
Cr4⫹ IV 0.55 0.41 Fe2⫹ IV HS 0.77 0.63
V 0.72 0.58 IVsp HS 0.78 0.64
VI LS 0.69 0.55 VI 0.75 0.61
HS 0.785 0.645 HS 0.920 0.780
VIII 0.92 0.78 VIII HS 1.06 0.92
VI LS 0.69 0.55 Fe3⫹ IV HS 0.63 0.49
5⫹
Cr IV 0.485 0.345 V 0.72 0.58
VI 0.63 0.49 VI LS 0.69 0.55
VIII 0.71 0.57 HS 0.785 0.645
Cr6⫹ IV 0.40 0.26 VIII HS 0.92 0.78
VI 0.58 0.44 Fe4⫹ VI 0.725 0.585
Cs1⫹ VI 1.81 1.67 Fe6⫹ IV 0.39 0.25
VIII 1.88 1.74 Fr1⫹ VI 1.94 1.80
IX 1.92 1.78 Ga3⫹ IV 0.61 0.47
X 1.95 1.81 V 0.69 0.55
XI 1.99 1.85 VI 0.760 0.620
XII 2.02 1.88 VII 1.14 1.00
Cu1⫹ II 0.60 0.46 Gd3⫹ VI 1.078 0.938
IV 0.74 0.60 VIII 1.193 1.053
VI 0.91 0.77 IX 1.247 1.107
Cu2⫹ IVsp 0.71 0.57 Ge2⫹ VI 0.87 0.73
V 0.79 0.65 Ge4⫹ IV 0.530 0.390
VI 0.87 0.73 VI 0.670 0.530
Cu3⫹ VI LS 0.68 0.54 H1⫹ I ⫺0.24 ⫺0.38
D1⫹ II 0.04 0.10 II ⫺0.04 ⫺0.08
Dy2⫹ VI 1.21 1.07 Hf 4⫹ IV 0.72 0.58
VII 1.27 1.13 VI 0.85 0.71
VIII 1.33 1.19 VII 0.90 0.76
Dy3⫹ VI 1.052 0.912 VIII 0.97 0.83
VII 1.11 0.97 Hg1⫹ III 1.11 0.97
VIII 1.167 1.027 VI 1.33 1.19
IX 1.223 1.083 Hg2⫹ II 0.83 0.69
3⫹
Er VI 1.030 0.890 IV 1.10 0.96
VII 1.085 0.945 VI 1.16 1.02
VIII 1.144 1.004 VIII 1.28 1.14
Eu2⫹ VI 1.31 1.17 Ho3⫹ VI 1.041 0.901
VII 1.34 1.20 VIII 1.155 1.015
VIII 1.39 1.25 IX 1.212 1.072
IX 1.44 1.30 X 1.26 1.12
X 1.49 1.35 I1⫺ VI 2.06 2.20
I5⫹ IIIpy 0.58 0.44 Mo5⫹ IV 0.60 0.46
VI 1.09 0.95 VI 0.75 0.61
I7⫹ IV 0.56 0.42 Mo6⫹ IV 0.55 0.41
VI 0.67 0.53 V 0.64 0.50
In3⫹ IV 0.76 0.62 VI 0.73 0.59
VI 0.940 0.800 VII 0.87 0.73
VIII 1.06 0.92 N3⫺ IV 1.32 1.46
3⫹
Ir VI 0.82 0.68 N3⫹ VI 0.30 0.16
Ir4⫹ VI 0.765 0.625 N5⫹ III 0.044 0.104
Ir5⫹ VI 0.71 0.57 IV 0.27 0.13
K1⫹ IV 1.51 1.37 Na1⫹ IV 1.13 0.99
VII 1.60 1.46 V 1.14 1.00
VIII 1.65 1.51 VI 1.16 1.02
IX 1.69 1.55 VII 1.26 1.12
X 1.73 1.59 VIII 1.32 1.18
XII 1.78 1.64 IX 1.38 1.24
La3⫹ VI 1.172 1.032 XII 1.53 1.39
VII 1.24 1.10 Nb2⫹ VI 0.86 0.72
VIII 1.300 1.160 Nb4⫹ VI 0.82 0.68
IX 1.356 1.216 VIII 0.93 0.79
X 1.41 1.27 Nb5⫹ IV 0.62 0.48
XII 1.50 1.36 VI 0.78 0.64
Li1⫹ IV 0.730 0.590 VII 0.83 0.69
VI 0.90 0.76 VIII 0.88 0.74
VIII 1.06 0.91 Nd2⫹ VIII 1.43 1.29
3⫹
Lu VI 1.001 0.861 IX 1.49 1.35
VIII 1.117 0.977 Nd3⫹ VI 1.123 0.983
IX 1.172 1.032 VIII 1.249 1.109
Mg2⫹ IV 0.71 0.57 IX 1.303 1.163
V 0.80 0.66 XII 1.41 1.27
VI 0.860 0.720 Ni2⫹ IV 0.69 0.55
VIII 1.03 0.89 IVsp 0.63 0.49
Mn2⫹ IV HS 0.80 0.66 V 0.77 0.63
V HS 0.89 0.75 VI 0.830 0.690
VI LS 0.81 0.67 Ni3⫹ VI LS 0.70 0.56
HS 0.970 0.830 HS 0.74 0.60
VII HS 1.04 0.90 Np2⫹ VI 1.24 1.10
VIII 1.10 0.96 Np3⫹ VI 1.15 1.01
Mn3⫹ V 0.72 0.58 Np4⫹ VI 1.01 0.87
VI LS 0.72 0.58 VIII 1.12 0.98
HS 0.785 0.645 Np5⫹ VI 0.89 0.75
4⫹
Mn IV 0.53 0.39 Np6⫹ VI 0.86 0.72
VI 0.670 0.530 Np7⫹ VI 0.85 0.71
Mn5⫹ IV 0.47 0.33 O2⫺ II 1.21 1.35
Mn6⫹ IV 0.395 0.255 III 1.22 1.36
Mn7⫹ IV 0.39 0.25 IV 1.24 1.38
VI 0.60 0.46 VI 1.26 1.40
Mo3⫹ VI 0.83 0.69 VIII 1.28 1.42
Mo4⫹ VI 0.790 0.650 OH1⫺ II 1.18 1.32
continued
III 1.20 1.34 Pu3⫹ VI 1.14 1.00
IV 1.21 1.35 Pu4⫹ VI 0.94 0.80
VI 1.23 1.37 VIII 1.10 0.96
Os4⫹ VI 0.770 0.630 Ra2⫹ VIII 1.62 1.48
Os5⫹ VI 0.715 0.575 XII 1.84 1.70
Os6⫹ V 0.63 0.49 Rb1⫹ VI 1.66 1.52
VI 0.685 0.545 VII 1.70 1.56
Os7⫹ VI 0.665 0.525 VIII 1.75 1.61
Os8⫹ IV 0.53 0.39 IX 1.77 1.63
P3⫹ VI 0.58 0.44 X 1.80 1.66
P5⫹ IV 0.31 0.17 XI 1.83 1.69
V 0.43 0.29 XII 1.86 1.72
VI 0.52 0.38 XIV 1.97 1.93
Pa3⫹ VI 1.18 1.04 Re4⫹ VI 0.77 0.63
Pa4⫹ VI 1.04 0.90 Re5⫹ VI 0.72 0.58
VIII 1.15 1.01 Re7⫹ IV 0.52 0.38
Pa5⫹ VI 0.92 0.78 VI 0.67 0.53
IX 1.09 0.95 Rh3⫹ VI 0.805 0.665
2⫹
Pb IVpy 1.12 0.98 Rh4⫹ VI 0.74 0.60
VI 1.33 1.19 Rh5⫹ VI 0.69 0.55
VII 1.37 1.23 Ru3⫹ VI 0.82 0.68
VIII 1.43 1.29 Ru4⫹ VI 0.760 0.620
IX 1.49 1.35 Ru5⫹ VI 0.705 0.565
X 1.54 1.40 Ru7⫹ IV 0.52 0.38
XI 1.59 1.45 Ru8⫹ IV 0.50 0.36
XII 1.63 1.49 S2⫺ VI 1.70 1.84
Pb4⫹ IV 0.79 0.65 S4⫹ VI 0.51 0.37
V 0.87 0.73 S6⫹ IV 0.26 0.12
VI 0.915 0.775 VI 0.43 0.29
VIII 1.08 0.94 Sb3⫹ IVpy 0.90 0.76
1⫹
Pd II 0.73 0.59 V 0.94 0.80
Pd2⫹ IVsp 0.78 0.64 VI 0.90 0.76
VI 1.00 0.86 Sb5⫹ VI 0.74 0.60
3⫹
Pd VI 0.90 0.76 Sc3⫹ VI 0.885 0.745
Pd4⫹ VI 0.755 0.615 VIII 1.010 0.870
Pm3⫹ VI 1.11 0.97 Se2⫺ VI 1.84 1.98
VIII 1.233 1.093 Se4⫹ VI 0.64 0.50
IX 1.284 1.144 Se6⫹ IV 0.42 0.28
Po4⫹ VI 1.08 0.94 VI 0.56 0.42
VIII 1.22 1.08 Si4⫹ IV 0.40 0.26
6⫹
Po VI 0.81 0.67 VI 0.540 0.400
Pr3⫹ VI 1.13 0.99 Sm2⫹ VII 1.36 1.22
VIII 1.266 1.126 VIII 1.41 1.27
IX 1.319 1.179 IX 1.46 1.32
Pr4⫹ VI 0.99 0.85 Sm3⫹ VI 1.098 0.958
VIII 1.10 0.96 VII 1.16 1.02
Pt2⫹ IVsp 0.74 0.60 VIII 1.219 1.079
VI 0.94 0.80 IX 1.272 1.132
Pt4⫹ VI 0.765 0.625 XII 1.38 1.24
Pt5⫹ VI 0.71 0.57 Sn4⫹ IV 0.69 0.55
V 0.76 0.62 Tm3⫹ VI 1.020 0.880
VI 0.830 0.690 VIII 1.134 0.994
VII 0.89 0.75 IX 1.192 1.052
VIII 0.95 0.81 U3⫹ VI 1.165 1.025
Sr2⫹ VI 1.32 1.18 U4⫹ VI 1.03 0.89
VII 1.35 1.21 VII 1.09 0.95
VIII 1.40 1.26 IX 1.19 1.05
IX 1.45 1.31 XII 1.31 1.17
X 1.50 1.36 U5⫹ VI 0.90 0.76
XII 1.58 1.44 VII 0.98 0.84
Ta3⫹ VI 0.86 0.72 U6⫹ II 0.59 0.45
Ta4⫹ VI 0.82 0.68 IV 0.66 0.52
Ta5⫹ VI 0.78 0.64 VI 0.87 0.73
VIII 0.88 0.74 VII 0.95 0.81
Tb3⫹ VI 1.063 0.923 VIII 1.00 0.86
VII 1.12 0.98 V2⫹ VI 0.93 0.79
VIII 1.180 1.040 V3⫹ VI 0.780 0.640
Tb4⫹ VI 0.90 0.76 V4⫹ V 0.67 0.53
VIII 1.02 0.88 VI 0.72 0.58
Tc4⫹ VI 0.785 0.645 VIII 0.86 0.72
Tc5⫹ VI 0.74 0.60 V5⫹ IV 0.495 0.355
Tc7⫹ IV 0.51 0.37 V 0.60 0.46
VI 0.70 0.56 VI 0.68 0.54
Te2⫺ VI 2.07 2.21 W4⫹ VI 0.80 0.66
Te4⫹ III 0.66 0.52 W5⫹ VI 0.76 0.62
IV 0.80 0.66 W6⫹ IV 0.56 0.42
VI 1.11 0.97 V 0.65 0.51
Te6⫹ IV 0.57 0.43 VI 0.74 0.60
VI 0.70 0.56 Xe8⫹ IV 0.54 0.40
4⫹
Th VI 1.08 0.94 VI 0.62 0.48
VIII 1.19 1.05 Y3⫹ VI 1.040 0.900
IX 1.23 1.09 VII 1.10 0.96
X 1.27 1.13 VIII 1.159 1.019
XI 1.32 1.18 IX 1.215 1.075
XII 1.35 1.21 Yb3⫹ VI 1.008 0.868
Ti2⫹ VI 1.00 0.86 VII 1.065 0.925
Ti3⫹ VI 0.810 0.670 VIII 1.125 0.985
Ti4⫹ IV 0.56 0.42 IX 1.182 1.042
V 0.65 0.51 Zn2⫹ IV 0.74 0.60
VI 0.745 0.605 V 0.82 0.68
VIII 0.88 0.74 VI 0.880 0.740
Tl1⫹ VI 1.64 1.50 VIII 1.04 0.90
VIII 1.73 1.59 Zr4⫹ IV 0.73 0.59
XII 1.84 1.70 V 0.80 0.66
Tl3⫹ IV 0.89 0.75 VI 0.86 0.72
VI 1.025 0.885 VII 0.92 0.78
VIII 1.12 0.98 VIII 0.98 0.84
Tm2⫹ VI 1.17 1.03 IX 1.03 0.89
VII 1.23 1.09
than 1000 crystal structures, Shannon and Prewitt (1969) made two distinct as-
sumptions:
1. The oxygen radius in VI-fold coordination is effectively 1.40 Å, and is

thus identical to the value proposed by Pauling and Ahrens.
2. The oxygen radius in VI-fold coordination is 1.26 Å. This value is consis-
tent with the theoretical estimates of Fumi and Tosi (1964) for F⫺, given
the assigned difference between the ionic radii of F⫺ and O2⫺.
Shannon and Prewitt (1969) named the ionic radii derived from the first assump-
tion “effective ionic radii” (IRs) and those derived from the second assumption
“crystal radii,” thus stating that they are more appropriate to real structures.
Shannon’s (1976) tabulation, reported in the preceding pages, is a simple up-
dating of the previous tabulation of Shannon and Prewitt (1969), and obeys the
following assumptions:
1. The additivity of cationic and anionic radii in reproducing the interionic

distance (eq. 1.24) is valid, provided that coordination number, electron
spin, degree of covalence, repulsive forces, and polyhedral distortion (eq.
1.49) are all taken into account.
2. According to the above limitations, effective ionic radii and crystal radii
are independent of the type of structure.
3. Both cationic and anionic radii vary with coordination number.
4. With the same anion, unit-cell volumes in isostructural series are propor-
tional (but not necessarily in a linear fashion) to the cationic volume.
5. Covalence effects on the shortening of metal-oxygen and metal-fluorine
distances are not comparable.
6. The mean interatomic distance for a coordination polyhedron within a
given structure varies in a reproducible fashion with the degree of poly-
hedral distortion and with coordination number.
We focus attention on the fact that the crystal radii (CRs) for the various
cations listed in table 1.11 are simply equivalent to the effective ionic radii (IRs)
augmented by 0.14 Å. Wittaker and Muntus (1970) observed that the CR radii
of Shannon and Prewitt (1969) conform better than IR radii to the “radius ratio
principle”* and proposed a tabulation with intermediate values, consistent with
the above principle (defined by the authors as “ionic radii for geochemistry”), as
particularly useful for silicates. It was not considered necessary to reproduce the
* The so-called “radius ratio principle” establishes that, for a cation/anion radius ratio lower
than 0.414, the coordination of the complex is 4. The coordination numbers rise to 6 for ratios
between 0.414 and 0.732 and to 8 for ratios higher than 0.732. Actually the various compounds
conform to this principle only qualitatively. Tossell (1980) has shown that, if Ahrens’s ionic radii are
adopted, only 60% of compounds conform to the “radius ratio principle.”
tabulation here, because the actual incertitude of the interpolated values is ap-
proximately equal to the CR-IR gap.
1.11 Forces Interacting Between Two Atoms

at Defined Distances
1.11.1 Coulombic Attraction
The attractive force exerted by a cation of pointlike charge Z⫹ on an anion of

pointlike charge Z⫺ at distance r is given by
Z+Z−e2
FC = . (1.51)
r2
where e is the electronic charge. The coulombic potential at distance r is obtained

by integration of equation 1.51 in dr from infinite distance to r:
Z+Z−e2
ΦC = ∫ − FC dr = r
. (1.52)
The potential energy progressively decreases with decreasing distance between

the two ions. Thus, theoretically, the two ions could collapse if opposite forces
did not oppose their excessive proximity.
1.11.2 Short-Range Repulsion

First of all, let us note that short-range coulombic repulsion exists, due to the
homologous nature of the electronic charges of the two atoms at short distances
from each other. However, the main repulsive force is quantum-mechanistic and
is a manifestation of Heisenberg’s indetermination principle. The more electrons
are confined in a limited region of space, the more their angular momentum and
hence their kinetic energy increases (of section 1.18).
The increase of energy with decreasing distance is exponential. Born first
suggested a potential in the form
b
ΦR = , (1.53)
rn
where b is a constant and n (Born exponent) is an integer (between 5 and 12) that
depends on the nature of interacting ions. Equation 1.53 is then replaced by
 −r 
ΦR = b exp   , (1.54)
 ρ
which expresses the exponential dependency of the radial functions better. In

equation 1.54, b and ␳ are two constants that usually are defined, respectively, as
“repulsion factor” and “hardness factor.”
1.11.3 Dispersive Forces

If electron clouds of cations and anions are deformable in an electric field and
thus have polarizabilities of ␣⫹ and ␣⫺ , the value of the dipole moment induced
in the cations by the presence of anions ( ␮⫺⫹ ) and induced in the anions by the
presence of cations ( ␮⫹⫺ ) are, respectively, proportional to ␣⫹ /r3 and ␣⫺ /r3. The
electric field acting on the cations is proportional to ⫺ ␣ ⫹␮ 2⫺⫹ /r 6, where ␮ 2⫺⫹ is
the mean square dipole moment of the cations.
Polarization has been formally represented as the movement of a charge Ze through

a displacement distance d (cf. Section 1.8). The displacement is due to field F, and the
drop in potential energy caused by movement in the field is Fd. If the force acting on
charge Ze by virtue of the parent nucleus is
F = − KF d (1.55)
(see eq. 1.14), the resultant fall in potential energy will be
1 1 1
− KF d 2 + KF d 2 = − KF d 2 − αF2 . (1.56)
2 2 2
The interaction of the two dipoles can thus be associated with a potential energy propor-
tional to the square of the electric field F. Moreover, F itself is proportional to r⫺3, so that
the functional dependency of the dispersive energy over the distance will be proportional
to r⫺6.
Accounting for the instantaneous higher moments of the charge distributions

of the atoms leads to an inverse eighth-power functional form (dipole-quadrupole
interactions). The bulk dispersive potential is represented as shown by Mayer
(1933):
dd + − dq + −
ΦD = 6
+ . (1.57)
r r8
The dipole-dipole (dd⫹⫺ ) and dipole-quadrupole (dq⫹⫺ ) coefficients can be de-

rived from free-ion polarizability ␣ and the mean excitation E by applying
3  E+E− 
dd + − = α +α −   (1.58)
2  E+ + E− 
and
27  α + α −   E + E −   α + E + α − E − 
dq + − =    +
n − 
(1.59)
8  e 2   E + + E −   n+
(Mayer, 1933; Boswara and Franklyn, 1968), where n⫹ and n⫺ are effective elec-
trons (see Tosi, 1964, for an extended discussion of the significance of the vari-
ous terms).
1.12 Lattice Energy of an Ionic Crystal
The lattice energy of an ionic crystal is the amount of energy required at absolute
zero temperature to convert one mole of crystalline component into constituent ions
in a gaseous state at infinite distance. It is composed of the various forms of ener-
gies, as shown above. The calculation is in fact somewhat more complex because
of the presence of various ions of alternating charges in a regular tridimensional
network.
1.12.1 Madelung Constant

To approach the complexity of a real tridimensional structure, let us first consider
the case of a monodimensional array of alternating positive and negative charges,
each at distance r from its first neighbor. We will assume for the sake of simplicity
that the dispersive potential is negligible. The total potential is therefore
Φ total = Φ C + Φ R . (1.60)
The potential at the (positive) central charge in figure 1.11 is that generated by
the presence of two opposite charges at distance r, by two homologous charges
at distance 2r, by two opposite charges at distance 3r, and so on. The coulombic
potential is then represented by a serial expansion of the type
2e 2 2e 2 2e 2 2e 2 2e 2  1 1 1 
ΦC = − + − + −L= −  1 − + − + L . (1.61)
r 2r 3r 4r r  2 3 4 
The term in parentheses in equation 1.61, which in this particular case corre-
sponds to the natural logarithm of 2 (ln 2 ⫽ 0.69), is known as the Madelung
constant (M ) and has a characteristic value for each structure.
Figure 1.11 Monodimensional array of alternating charges.
Let us now consider the coulombic potential at a positive central charge in a

regular tridimensional network of alternating charges. The cubic compound NaCl
is exemplified in this respect. Taking the central charge on a Na⫹ ion, we find six
Cl ⫺ ions as first neighbors at distance r0, corresponding to 12 the cell edge a 0 (cf.
figure 1.2). The second neighbors are 12 Na⫹ ions at distance r 0冑 2, the third
neighbors are eight Cl ⫺ ions at distance r 0冑 3, the fourth neighbors are six Na⫹
ions at distance 2r0, the fifth neighbors are 24 Cl ⫺ ions at distance r 0冑 5, and so
on. The coulombic potential acting on the cation is
Z +2 − e 2  12 8 6 24 
Φ C , Na+ = − 6 − + − + − L . (1.62)
r0  2 3 4 5 
where Z⫹⫺ is the largest common factor of the pointlike charges. If we now repeat
the calculation adopting Cl ⫺ as the central charge, we obtain a coulombic poten-
tial that (for this particular structure) is identical to that acting at the cationic
site. The bulk coulombic potential is given by the summation of the cationic and
anionic parts divided by 2:
ΦC , Na + + ΦC ,Cl − Z +2 − e 2 M
ΦC = =− . (1.63)
2 r0
Let us now go back to the monodimensional array of figure 1.11. The repul-
sive potential at the central charge is composed of
  −r   −2r   −3r    −r 
ΦR = 2b  exp   + exp   + exp   + L . ≈ 2b exp   (1.64)
  ρ  ρ   ρ    ρ
(the effect of repulsive forces vanishes rapidly with increasing distance and, for
practical purposes, extended effects beyond the first neighbors at first sight can
be neglected). We then have
−1.38Z +2 − e 2  −r 
Φtotal = + 2b exp   . (1.65)
r  ρ
The ions in the array arrange themselves in order to minimize the bulk energy of
the substance. The equilibrium condition is analytically expressed by equating to
zero the first derivative of the potential energy over distance:
 ∂ Φtotal 
  = 0. (1.66)
 ∂ r  r =r
0
Applying the above condition and solving over b, we obtain
−1.38Z +2 − e 2  ρ
Φtotal( r = r0 ) = 1 −  . (1.67)
r0  r0 
Equation 1.67 is known as Born-Landé equation.

If we now consider the effect of the crystal lattice, by introducing Madelung
constant M and multiplying by Avogadro’s number N 0 (i.e., the number of atoms
per weight formula unit), we get
− MN 0 Z +2 − e 2  ρ
Φtotal( r = r0 ) = 1 −  . (1.68)
r0  r0 
Equation 1.68 is the Born-Mayer equation.
1.12.2 Zero-Point Energy

The lattice energy of a crystalline substance U with purely ionic bonds and negli-
gible polarization effects is given by equation 1.68, changed in sign and with the
subtraction of an energetic term known as “zero-point energy”:
U = − Φtotal − E 0 . (1.69)
To understand the significance of this last term, we must go back to the monodi-
mensional harmonic oscillator and quantum mechanics.
Atoms in a crystal are in fact in vibrational motion even at zero temperature.
Under these thermal conditions, vibrational quantum number n is at the lowest

possible value. When we substitute the value obtained by the vibrational fre-
quency for force constant KF of the harmonic oscillator (eq. 1.14):
K F = 4 π 2 mv 02 , (1.70)
equation 1.17 becomes:
 1
E ( n ) =  n +  hv 0 , (1.71)
 2
which, with n ⫽ 0, gives the zero-point energy—i.e., the vibrational energy pres-
ent even at the zero point:
1
E0 = hv0 . (1.72)
2
However, for a real crystal at zero temperature it is impossible to group all vibra-
tional motions on the lowest single vibrational mode and, if the crystal behaves
as a Debye solid (see later), zero-point energy is expressed as:
9
E0 = N 0 hv max , (1.73)
4
where vmax is the highest vibrational mode occupied in the crystal.
1.12.3 Ladd and Lee’s Equation

Accounting for all the various forms of energy outlined above (including disper-
sive forces) and solving for b at equilibrium conditions we obtain the equation of
Ladd and Lee (1958):
M N 0 Z +2 − e 2  ρ  N0 dd + −  6 ρ  N 0 dq + −  8ρ 
U = 1 −  + 1 − + 1 − 
r0  r0  r06  r0  r08  r0 
(1.74)
9
− N 0 hv max .
4
1.12.4 Quantitative Calculations: an Example

Adopting the equation of Catti (1981), the lattice energy of an ionic crystal may
be expressed by means of the double summations
Table 1.12 Lattice energies of olivine end-members; values in kJ/mole (Ottonello, 1987).
EBHF ⫽ energy of the thermochemical cycle; EC ⫽ coulombic energy; ER ⫽ repulsive energy;
EDD and EDQ ⫽ dispersive energies.
Compound EBHF Hf,Tr,Pr ⫺U EC ER EDD EDQ
Mg2SiO4 18456.4 ⫺2175.7 ⫺20632.1 ⫺24131.3 3835.3 ⫺259.4 ⫺76.0

Fe2SiO4 19268.4 ⫺1481.6 ⫺20750.3 ⫺23934.6 3779.3 ⫺442.1 ⫺152.9
Ca2SiO4 17608.5 ⫺2316.2 ⫺19925.1 ⫺23105.4 3546.9 ⫺282.4 ⫺84.3
Mn2SiO4 18825.4 ⫺1728.1 ⫺20553.5 ⫺23734.8 3641.3 ⫺348.5 ⫺111.6
Co2SiO4 19443.9 ⫺1406.7 ⫺20850.6 ⫺24040.8 3855.4 ⫺490.5 ⫺174.6
Ni2SiO4 19609.9 ⫺1405.2 ⫺21015.1 ⫺24219.8 4027.4 ⫺596.4 ⫺226.2
Zn2SiO4 19326.0 ⫺1390.4 ⫺20716.4 ⫺24068.0 3875.2 ⫺393.9 ⫺129.6
n n n n n n n n
U = ∑ ∑ Zi Z j CijEL + ∑ ∑ bij CijR + ∑ ∑ dd ij CijDD + ∑ ∑ dq ij CijDQ (1.75)
i =1 j =i i =1 j =i i =1 j =i i =1 j =i
extended to all ions in the lattice.

Coefficients bij , ddij , and dqij in equation 1.75 have the same significance as
coefficient b in equation 1.54 and dd⫹⫺ and dq⫹⫺ in equation 1.57. Coefficients
C EL R DD DQ
ij , C ij , C ij , and C ij depend on crystal symmetry and on interionic distances
and are obtained by means of standard mathematical routines that cannot be
detailed here (see appendix A in Tosi, 1964 for exhaustive treatment). Table 1.12
lists as examples the lattice energy values obtained through application of equa-
tion 1.75 to orthorhombic silicates (olivine, spatial group Pbnm). Values are ex-
pressed in kilojoules per mole. We note that lattice energy is mainly (about 90%)
composed of the attractive coulombic term. Repulsive energy is about 15%
of bulk energy and is of opposite sign. Dispersive energy is greatly subordinate
(1–3%).
Figure 1.12 shows the potential well for Ni2SiO4 obtained by expanding and/
or compressing isotropically the bond distances among all the ions in the struc-
ture, with respect to the equilibrium position. The stable state condition of the
solid compound is at the minimum of the potential well (⌽0 at r 0 ; see eq. 1.66).
We note that the energy modification is asymmetric with respect to the minimum
of the potential well (dashed line in figure 1.12), due to the different functional
dependencies of r on the various forms of energy. As we will see later, this asym-
metry (which is responsible for the thermal expansion of the substance) may be
translated into an anharmonicity of vibrational motions of the ions in the
structure.
1.12.5 Semiempirical Calculations: the Kapustinsky Method

As already noted, the estimate of the lattice energy of a crystalline compound
requires detailed knowledge of its structure. When the structure is not known
Figure 1.12 Potential well for Ni2SiO4 olivine. Abscissa values are fractional values of
isotropic compression and/or expansion.
with sufficient precision, or when calculation of the Madelung constant is particu-

larly time-consuming, alternative semiempirical calculations can be carried out.
The Kapustinsky method is perhaps the most popular among the semiquantitative
approaches. The difference between the Madelung constant of a given crystal
and the Madelung constant of the same compound with the structure of sodium
Table 1.13 Thermochemical radii for polyanionic complexes (Greenwood, 1970).
Complex r⬘ Complex r⬘ Complex r⬘ Complex r⬘

BF4 2.28 BeF4 2.45 ClO4 2.36 IO4 2.49
ClO4 2.36 SO42⫺ 2.30 SeO42⫺ 2.43 TeO42⫺ 2.54
CrO42 2.40 MoO42⫺ 2.54 PO43⫺ 2.38 AsO43⫺ 2.48
SbO43 2.60 BiO43⫺ 2.68 SiO44⫺ 2.40 OH 1.40
SH 1.95 SeH 2.15 CN 1.82 CNO 1.59
CNS 1.95 NH2⫺ 1.30 NO2 1.55 HCO3 1.58
CH3CO2⫺ 1.58 HCO3 1.63 CO32⫺ 1.85 NO3 1.89
ClO3 2.00 BrO3 1.91 IO3 1.89 O32⫺ 1.80
chlorine is compensated by the difference between the actual interionic distances

in the crystal in question and those obtained from the ionic radii in VI-fold coor-
dination with oxygen. Thus derived is a set of “thermochemical radii” that repre-
sent the energy counterpart of a given ion (or polyanion) appropriate to any
structure. The Kapustinsky equation reads as follows:
vF Z +2 −  0.345   Kcal 
U = 287.2 1 − ,   (1.76)
r+′ + r−′  r+′ + r−′   mole 
where vF is the number of ions per formula unit and r ⫹⬘ and r ⬘⫺ are the thermo-
chemical radii. Table 1.13 lists typical thermochemical radii for the most impor-
tant polyanionic groups. Coupling the thermochemical radius of the polyanionic
group SiO44⫺ with the ionic radius of Mg2⫹ or Fe2⫹ in VI-fold coordination with
oxygen, the lattice energy of Mg2SiO4 or Fe2SiO4 is readily derived by application
of equation 1.76.
An extended form of the Kapustinsky equation has been proposed by Saxena
(1977). Its formulation is a compromise between equation 1.76 and the Ladd and
Lee equation (eq. 1.74):
b′  0.345  c′  6 ⋅ 0.345 
U = a′ + 1 − + 1 − 
r+′ + r−′  r+′ + r−′ 
( )  r+′ + r−′ 
6
r+′ + r−′
(1.77)
d′  8 ⋅ 0.345 
+ 1 − 
(r ′ + r ′ )  r+′ + r−′ 
8
+ −
The third and fourth terms on the right side of equation 1.77 take into account
the effect of dispersive potential. Constants a⬘, b⬘, c⬘, and d⬘ in equation 1.77 have
no direct physical meaning and are derived by interpolation procedures carried
out on the main families of crystalline compounds.
1.13 Born-Haber-Fayans Thermochemical Cycle
In calculating the lattice energy of a crystal, we adopt an arbitrary reference con-

dition (two isolated ions in the gaseous state and at infinite distance) to which we
assign a zero potential. It is worth stressing that this condition is not equivalent
to the standard state commonly adopted in thermochemical calculations, which
is normally that of element at stable state at reference P,T.
The relationships between the two different states and between the “enthalpy
of formation from the elements at standard state” (H 0f ) and the “lattice energy”
(U ) are easily understood by referring to the Born-Haber-Fayans thermochemical
cycle. In this cycle, the formation of a crystalline compound from isolated atoms
in the gaseous state is visualized as a stepwise process connecting the various
transformations. Let us follow the condensation process of a crystal MX formed
from a metal M and a gaseous molecule X2 :
1. First we transform metal M into a gaseous atom M(gas) and the gaseous
molecule 12X2 into a gaseous atom X(gas):
E
M (metal)  
S
→ M (gas) (1.78)
1
1 ED
X 2 (gas)  
2
→ X (gas) . (1.79)
2
The energies required by these processes are composed of the sublimation

energy of the metal (ES ) and the dissociation energy of the gaseous mole-
cule ( 12ED ).
2. We now transform the gaseous atoms into gaseous ions:
I
M (gas)  
M
→M+ . (1.80)
(gas)
−E
X (gas)  
→ X −(gas) .
X
(1.81)
The required energies are, respectively, those of first ionization (IM ) and
electron affinity (⫺EX ).
3. To transform the two gaseous ions into a crystalline molecule, we must
expend energy corresponding to the lattice energy of the crystalline sub-
stance (i.e., the lattice energy with its sign changed):
−U
+
M (gas) + X (gas)
−
→ MX (crystal) . (1.82)
4. We then close the cycle by transforming crystal MX into its component

elements in the standard state. To do this, we must furnish energy corre-
sponding to the enthalpy of formation from the elements:
−H0 1
MX (crystal)  

f
→ M (metal) + X 2 (gas) . (1.83)
2
The complete cycle is visualized in the following scheme:
1
1 Es + ED
M (metal) + X 2 (gas)  → M (gas) + X (gas)
2
2
↑ 0
−H f ↓ IM − E x . (1.84)
MX (crystal) ←−
U
 +
M (gas) + X (gas)
−
By applying the Hess additivity rule to the involved energy terms, we have:
1
H 0f = E S + ED + IM − EX − U . (1.85)
2
Equation 1.85 can be generalized for polyatomic molecules as follows:
n
n
H 0f , MX = E S +
n 2
ED + ∑ IM − nE X − U + (1 + n )RT , (1.86)
1
where ∑ n1I M is the summation of the first n ionization potentials of M.

Lastly, note that equation 1.86 includes the expansion work [(1 ⫹ n)RT ] on
the gaseous ions from absolute zero to standard state temperature (this energy
term should be included in equations 1.80 and 1.81, but the energy variation
associated with this term is negligible, being well within the experimental incerti-
tude involved in magnitudes IM , ES , ED , EX , and U ).
The tabulated values of EBHF for olivine end-members (table 1.12) correspond
to the summation of the first four terms on the right side of equation 1.86. Table
1.14 lists the various energy terms of the Born-Haber-Fayans thermochemical
Table 1.14 Born-Haber-Fayans energy terms for alkali halides. Values in

k/mole (adapted from Grenwood, 1970).
n
Compound ⫺H 0f Es 1
2 ED ∑ IM Ex U
1
LiCl 408.8 138.9 121.3 520.1 351.5 837.6

NaCl 410.9 100.4 121.3 495.8 360.7 767.8
KCl 436.0 81.2 121.3 418.8 354.8 702.4
RbCl 430.5 77.8 121.3 402.9 354.8 677.8
CsCl 433.0 70.3 121.3 375.7 359.8 640.6
cycle for alkali halides. We emphasize that the electron affinity of chlorine should
be constant in all compounds and that observed differences must be imputed to
experimental approximation.
1.14 Internal Energy of an Ionic Crystal

and Temperature
The internal energy of a crystal at zero temperature is equivalent to its lattice

energy. When heat is applied to an ionic crystal, its internal energy is increased
by vibrational effects. If ionic bonds are considered in light of the harmonic ap-
proximation, this increase in energy can be viewed as a statistical increase in the
amplitude of vibrational motion (vmax in eq. 1.73). As already noted (see also
section 3.1), the ionic bond does not conform perfectly to the harmonic approxi-
mation, because anharmonic vibrational terms are present (i.e., the crystal has a
definite thermal expansion). Although the treatment of anharmonicity in vibra-
tional studies results in a complex approach (see Kittel, 1971 for a detailed ac-
count of this problem), generally speaking the variation of potential energy of a
pair of ions displaced from their equilibrium position by thermal agitation can
be represented by a polynomial expansion in x of the following type (Dega-
nello, 1978):
U ( x ) = cx 2 − gx 3 − fx 4 , (1.87)
where c, g, and f are positive. The term ⫺gx3 reflects the asymmetricity of repul-
sive effects, and ⫺ fx4 accounts for the attenuation of vibrational motions at high
amplitude. With increasing temperature, gx3 and fx4 become more and more
significant. Thus, generally, thermal expansion is not a linear function in T.
1.14.1 Internal Energy and Thermal Expansion

The relationship between the increase in internal energy, due to the first term on
the right in equation 1.87, and thermal expansion is easily understood if we con-
sider the form of the potential well for a crystal (cf. figure 1.12). If we heat the
crystal and inhibit any thermal expansion, we observe a sudden increase in energy,
caused mainly by repulsive effects between neighboring ions: the increase in en-
ergy is identical to that observed on the compressional side of the potential well
in figure 1.12 (harmonic condition). Indeed, Debye formulated a principle stating
that linear thermal expansion is simply related to heat capacity at constant vol-
ume (CV ). By means of the Born repulsive model (eq. 1.53), it can be shown that,
for a cubic crystal with a single vibrational frequency, the Debye principle can be
translated as follows:
CV  n + 2 
αl =  , (1.88)
2U  n 
where ␣l is the linear thermal expansion coefficient and n is the Born exponent
(see Das et al., 1963 for an explanatory application of eq. 1.88).
1.14.2 Linear and Volumetric Thermal Expansion of

Crystalline Solids
The thermal expansion of a crystal with increasing temperature is generally de-
scribed by a second-order polynomial in T of type
VT = V0 + aT + bT 2 , (1.89)
where V0 is the volume at reference condition and a and b are experimentally

determined factors, varying from substance to substance. Equation 1.89 is purely
functional and cannot be used in thermochemical calculations that require a gen-
eralized equation. This is provided by linear isobaric thermal expansion coeffi-
cient ␣l and volumetric isobaric thermal expansion coefficient ␣V, defined as
follows:
1  ∂d 
αl =   (1.90)
d  ∂T  P ,X
1  ∂V 
αV =   . (1.91)
V  ∂T  P ,X
We have already seen that ␣l can be related to U by means of the Debye principle.
The usefulness of ␣V will be evident later, when we treat the thermodynamic prop-
erties of crystalline substances at high T.
Another coefficient generally used in thermal expansion calculations is
the mean polyhedral linear expansion coefficient between temperatures T1 and T2
( ␣ T1⫺T2 ):
2  d 2 − d1 
α T1 −T2 =  . (1.92)
d1 + d 2  T2 − T1 
Examining the mean polyhedral linear thermal expansion coefficients of a

large number of crystalline compounds, Hazen and Prewitt (1977) formulated
two important generalizations:
Table 1.15 Relationship among mean polyhedral linear thermal expansion (␣20–1000),
Pauling’s bond strength, and ionicity factor (adapted from Hazen and Finger, 1982).
S 2i Z⫹Z⫺ S 2i Z⫹Z⫺
Cation Anion ␯c ␣20–1000 Cation Anion ␯c ␣20–1000
␯c ␯c
Re6⫹ O2⫺ 6 1 1 Li⫹ O2⫺ 4 1/4 22
Bi5⫹ O2⫺ 6 5/6 0 Li⫹ O2⫺ 6 1/6 20
Si4⫹ O2⫺ 4 1 0 K⫹ O2⫺ 6 1/6 21
Ti4⫹ O2⫺ 6 2/3 8 Na⫹ O2⫺ 6 1/6 17
Zr4⫹ O2⫺ 7 4/7 8 Na⫹ O2⫺ 7 1/7 35
Hf 4⫹ O2⫺ 7 4/7 8 Na⫹ O2⫺ 8 1/8 22
U4⫹ O2⫺ 8 1/2 10 Na⫹ O2⫺ 9 1/9 18
Th4⫹ O2⫺ 8 1/2 9 Li⫹ F⫺ 6 0.125 46
Ce4⫹ O2⫺ 8 1/2 13 Na⫹ Cl⫺ 6 0.125 51
Al3⫹ O2⫺ 4 3/4 1 K⫹ Cl⫺ 6 0.125 46
Al3⫹ O2⫺ 5 3/5 6 K⫹ Br⫺ 6 0.125 49
Al3⫹ O2⫺ 6 1/2 9 K⫹ I⫺ 6 0.125 47
V3⫹ O2⫺ 6 1/2 13 Na⫹ F⫺ 6 0.125 45
Fe3⫹ O2⫺ 6 1/2 9 Rb⫹ Br⫺ 6 0.125 44
Cr3⫹ O2⫺ 6 1/2 8 Ca2⫹ F⫺ 8 0.19 21
Ti3⫹ O2⫺ 6 1/2 8 Cs⫹ Br⫺ 8 0.094 68
Bi3⫹ O2⫺ 6 1/2 9 Mn2⫹ S2⫺ 6 0.27 18
Be2⫹ O2⫺ 4 1/2 9 Pb2⫹ S2⫺ 6 0.27 22
Zn2⫹ O2⫺ 4 1/2 7 Pb2⫹ Te2⫺ 6 0.27 20
Ni2⫹ O2⫺ 6 1/3 14 Pb2⫹ Se2⫺ 6 0.27 21
Co2⫹ O2⫺ 6 1/3 14 Zn2⫹ S2⫺ 4 0.40 9
Fe2⫹ O2⫺ 6 1/3 13 Zn2⫹ Te2⫺ 4 0.40 10
Mn2⫹ O2⫺ 6 1/3 15 Zn2⫹ Se2⫺ 4 0.40 9
Cd2⫹ O2⫺ 6 1/3 13 Al3⫹ As3⫺ 4 0.56 5
Mg2⫹ O2⫺ 6 1/3 14 Ga3⫹ As3⫺ 4 0.56 7
Mg2⫹ O2⫺ 8 1/4 13 B3⫹ N4⫺ 4 0.45 13
Ca2⫹ O2⫺ 6 1/3 15 Ti4⫹ N4⫺ 6 0.53 9
Ca2⫹ O2⫺ 7 2/7 16 U4⫹ N4⫺ 6 0.53 9
Ca2⫹ O2⫺ 8 1/4 14 Nb4⫹ C4⫺ 6 0.53 7
Sr2⫹ O2⫺ 6 1/3 14 Ta4⫹ C4⫺ 6 0.53 7
Ba2⫹ O2⫺ 6 1/3 15 Ti4⫹ C4⫺ 6 0.53 8
Ba2⫹ O2⫺ 9 2/9 15 Zr4⫹ C4⫺ 6 0.53 7
Pb2⫹ O2⫺ 12 1/6 23 C4⫹ C4⫺ 4 0.80 3.5
1. For each polyhedron, the mean polyhedral linear thermal expansion is

independent of the type of bonding of the polyhedron with the other
structural units of the crystal.
2. All oxygen-based polyhedra with the same Pauling bond strength (cat-
ionic charge/coordination number; see section 1.10.3) have the same
mean polyhedral linear expansion.
These generalizations led the authors to propose the following empirical relation,
valid for oxygen-based polyhedra:
Figure 1.13 Expansion and/or compression effects on crystal structures (planar section).
A ⫽ shared edge; B ⫽ shared corner. From R. M. Hazen and L. W. Finger, Comparative
Crystal Chemistry, copyright 䉷 1982 by John Wiley and Sons. Reprinted by permission
of John Wiley & Sons.
 Z 
α 20 −1000 = 32.9  0.75 − +  × 10 −6 ° C −1 . (1.93)
 vc 
Equation 1.93 was later superseded by a more general equation proposed by Ha-
zen and Finger (1979):
 4v 
α 20 −1000 =  2 C  × 10 −6 ° C −1 . (1.94)
 Si Z + Z − 
The term S 2l is defined by the authors as an “ionicity factor” and assumes a value
of 0.5 for oxides and silicates, 0.75 for halides, 0.40 for calcogenides, 0.25 for
phosphides and arsenides, and 0.2 for nitrides and carbides (Z⫺ is the anion
charge). Equation 1.94 is based on the thermal expansion data listed in table 1.15.
It must be emphasized that the bulk thermal expansion of a crystal is not
simply related to the mean expansion of the various polyhedra within the struc-
ture, because progressive readjustments of polyhedral orientations are often dic-
tated by geometric constraints. As an example, figure 1.13 shows the different
behaviors of crystals with polyhedra sharing an edge (case A) or a corner (case
B). In case B, the thermal expansion is accompanied by a substantial modification
of the angular value of the shared corner. The same type of reasoning is valid for
compressional effects.
1.15 Internal Energy and Pressure
The equation of state of a cubic crystal under hydrostatic pressure is given by
 ∂A 
P = −  , (1.95)
 ∂V  T ,X
where A is the Helmholtz free energy, composed of internal energy U plus a vibra-
tional term (A vib ):
A( V ,T ,X ) = U ( V ,X ) + Avib(V,T,X ) . (1.96)
Adopting the harmonic approximation, the vibrational term is not explicitly de-
pendent on volume, and equation 1.95 may be reexpressed as follows:
 ∂U 
P = −  (1.97)
 ∂P  T ,X
We now introduce two new parameters that describe the changes in interionic
distances and volume with pressure: isothermal linear compression coefficient ␤l ,
and isothermal volumetric compression coefficient ␤V :
1 ∂d
βl = −   (1.98)
d  ∂P  T , X
1  ∂V 
βV = −   . (1.99)
V  ∂P  T ,X
The analogy with thermal expansion coefficients is evident when one compares
equations 1.98 and 1.99 with equations 1.90 and 1.91, respectively. For compress-
ibility, as for thermal expansion, a mean coefficient that defines the volumetric
variation (V2 ⫺ V1 ) for a finite pressure range (T2 ⫺ T1 ) may be introduced. This
coefficient, the “mean volumetric isothermal compressibility,” is given by
2  V2 − V1 
βV = −  . (1.100)
V1 + V2  P2 − P1 
The reciprocal of ␤V, expressed in units of pressure, is called “bulk modulus” (K):
1
K = . (1.101)
βV
Combining equations 1.97 and 1.99, we obtain
−1
1  ∂ 2U 
βV = −   . (1.102)
V  ∂V 2  T ,X
As already noted, equation 1.102 is valid when vibrational energy does not depend explic-
itly on volume. In a more refined treatment of the problem, the vibrational energy of the
crystal can be treated as purely dependent on T. This assumption brings us to the Hilde-
brand equation of state and to its derivative on V:
∂U α
= −P + T V (1.103)
∂V βV
 ∂ 2U  1 T  ∂β  α  ∂β  
V  = + 2  V2  + V  V . (1.104)
 ∂V 2
βV βV  ∂ T  P, X βV  ∂P  T , X 
 
Equation 1.104 is obtained by application of thermodynamic relations (see chapter 2):
 ∂S  αV
  = (1.105)
 ∂V  T βV
 ∂α  1  ∂βV 
V V  =   . (1.106)
 ∂V  T βV  ∂P  T
An even more precise treatment, based on the assumption that the vibrational Helm-
holtz free energy of the crystal, divided by temperature, is a simple function of the ratio
between T and a characteristic temperature dependent on the volume of the crystal, leads
to the Mie-Gruneisen equation of state (see Tosi, 1964 for exhaustive treatment):
∂U A α
= −P + vib V . (1.107)
∂V CV βV
The ratio ␣v /CV␤V multiplied by the reciprocal of the density of the substance
(1/ ␳ ) is called the Gruneisen thermodynamic parameter ( ␥G ):
αV 1
γG = . (1.108)
CV βV ρ
In a more or less complex fashion, equations 1.97, 1.103, and 1.107 describe the re-
sponse of the crystal, in terms of energy, to modification of the intensive variable P. We
note that, in all cases, the differential forms combine internal energy and volume and
are thus analytical representations of the potential diagram seen above for Ni2SiO4 in
figure 1.12.
For simple compounds such as NaCl, where a single interionic distance (r) is sufficient
to describe the structure, if we adopt the simplified form of equation 1.97 and combine
the partial derivatives of U and V on r, we obtain
3
 ∂V 
 
 ∂r
βV = . (1.109)
 ∂V ∂ 2U ∂U ∂ 2V 
V − 
 ∂ r ∂ r2 ∂ r ∂ r2 
The error introduced by neglecting the term T ␣v / ␤v (cf. eq. 1.97 and 1.103) is, in the
case of alkali halides, about 0.5 to 1.5% of the value assigned to U (Ladd, 1979).
For cubic structures with more than one interionic distance—as, for instance, spinels
(multiple oxides of type AB2O4 with A and B cations in tetrahedral and octahedral coordi-
nation with oxygen, respectively)—it is still possible to use equation 1.109, but the partial
derivatives must be operated on the cell edge, which is, in turn, a function of the various
interionic distances (Ottonello, 1986).
1.15.1 Pressure-Volume Empirical Relationships

for Crystalline Solids
As early as 1923, Bridgman had observed that the compressibility of metals is

proportional to their molar volume raised to 43. Anderson and coworkers (cf. An-
derson, 1972, and references therein) extended Bridgman’s empirical relationship
to various isostructural classes of compounds, noting that, within a single isos-
tructural class, the following relation is valid:
βV V = constant . (1.110)
Hazen and Finger (1979) extended equation 1.110 to mean polyhedral compress-
ibility ␤ VP (mean compressibility of a given coordination polyhedron within a
crystal structure), suggesting that it is related to the charge of ions in the polyhe-
dron through an ionicity factor, analogous to what we have already seen for ther-
mal expansion—i.e.,
Table 1.16 Polyhedral compressibility moduli in selected natural and synthetic compounds
(adapted from Hazen and Finger, 1982). Data are expressed in megabars (1Mbar ⫽ 106 bar).
Str ⫽ type of structure. K ⫽ 1/␤Vp.
Polyhedron Str K Polyhedron Str K Polyhedron Str K

Ni-O 1 1.96 Ca-O 10 0.85 Mn-F 5 0.9
Co-O 1 1.85 Na-O 12 0.32 Ti-C 1 1.9
Fe-O 1 1.53 Zr-O 13 2.8 U-C 1 1.6
Mn-O 1 1.43 Si-O 13 ⬎2.5 Zr-C 1 1.9
Ca-O 1 1.10 Li-F 1 0.66 C-C 16 5.8
Sr-O 1 0.91 Na-F 1 0.45 C-C 17 5.9
Ba-O 1 0.69 K-F 1 0.293 Ca-S 1 0.43
Be-O 2 2.5 Rb-F 1 0.273 Sr-S 1 0.40
Zn-O 2 1.4 Li-Cl 1 0.315 Ba-S 1 0.35
U-O 3 2.3 Na-Cl 1 0.240 Pb-S 1 0.48
Th-O 3 1.93 K-Cl 1 0.180 Cd-S 2 0.61
Al-O 4 2.4 Rb-Cl 1 0.160 Zn-S 2 0.77
Fe-O 4 2.3 Li-Br 1 0.257 Zn-S 15 0.76
Cr-O 4 2.3 Na-Br 1 0.200 Sn-S 18 1.2
V-O 4 1.8 K-Br 1 0.152 Zn-Se 15 0.60
Si-O 5 3.2 Rb-Br 1 0.138 Ca-Se 1 0.49
Ge-O 5 2.7 Li-I 1 0.188 Sr-Se 1 0.43
Ti-O 5 2.2 Na-I 1 0.161 Ba-Se 1 0.36
Ru-O 5 2.7 K-I 1 0.124 Pb-Se 1 0.34
Mn-O 5 2.8 Rb-I 1 0.111 Cd-Se 2 0.54
Sn-O 5 2.3 Cs-Cl 14 0.182 Ca-Te 1 0.42
Si-O 6 ⬎5 Cs-Br 14 0.155 Sr-Te 1 0.334
Ge-O 6 ⬎4 Cs-I 14 0.129 Ba-Te 1 0.305
Mg-O 7 1.3 Th-Cl 14 0.236 Pb-Te 1 0.41
Al-O 7 2.2 Th-Br 14 0.225 Sn-Te 1 0.42
Si-O 7 3 Cu-Cl 15 0.40 Cd-Te 15 0.42
Ca-O 7 1.15 Ag-l 15 0.243 Hg-Te 15 0.44
Al-O 7 2.2 Ca-F 3 0.86 Zn-Te 15 0.51
Si-O 7 3 Ba-F 3 0.57 Ga-Sb 1 0.56
Mg-O 8 1.5 Pb-F 3 0.61 In-Sb 1 0.47
Ni-O 9 1.5 Sr-F 3 0.70 Ga-As 1 0.75
Si-O 9 ⬎2.5 Mg-F 5 1.0 In-As 1 0.58
1 ⫽ NaCl 2 ⫽ Zincite 3 ⫽ Fluorite 4 ⫽ Corundum 5 ⫽ Rutile 6 ⫽ Quartz 7 ⫽ Garnet

8 ⫽ Olivine 9 ⫽ Spinel 10 ⫽ Clinopyroxene 11 ⫽ Mica 12 ⫽ Feldspar 13 ⫽ Zircon 14 ⫽ CsCl
15 ⫽ Cubic ZnS 16 ⫽ Diamond 17 ⫽ Graphite 18 ⫽ CdI
0.044 d 3
βVP = ( Mbar −1 ). (1.111)
Si2 Z + Z −
S 2i in equation 1.111 has the same significance and values already seen in equation
1.94, and d is the bond distance.
Experimental values of mean polyhedral compressibility modulus in various
compounds are listed in table 1.16.
Figure 1.14 Mean polyhedral compressibility for different coordination states. Interpo-
lant is equation 1.111. From R. M. Hazen and L. W. Finger, Comparative Crystal Chemis-
try, copyright 䉷 1982 by John Wiley and Sons. Reprinted by permission of John Wi-
ley & Sons.
The reproducibility of equation 1.111 may be deduced from figure 1.14, based
on the tabulated values. Consistent deviations are noted only for some VIII-fold
coordinated complexes.
1.15.2 Effect of Pressure on Compressibility

The compressibility of crystals varies with pressure. The simplest functional form
that can be adopted to describe the P-V relations for a solid is a polynomial
expansion in P, as we have already seen for temperature (cf. eq. 1.89):
VP = V 0 + aP + bP 2 . (1.112)
An expansion of the series to the third term is not sufficiently accurate to repro-
duce the actual behavior of solids and it is preferable to use more physically sound
functional forms, such as the first-order equation of state of Murnaghan:
K0  V  K ′ 
P=  0  − 1 = P0 (1.113)
K′  V  
 
or the first-order Birch form:
 V  7 3  V  5 3    V  2 3 
3    3  
P = K0   −  
0 0
1 − ( 4 − K ′ )  
0
− 1  + P0 , (1.114)
2  V   V   4  V  
   
where K 0 is the bulk modulus at zero pressure, V0 is the molar volume at zero
pressure, and K⬘ is the first derivative of the bulk modulus in dP.
Equations 1.113 and 1.114 are the most frequently used in geochemical stud-
ies. However, the most physically sound equation of state for solids is perhaps the
Born-Mayer form:
 2 3   −1 3   V  4 3 
 V0   r0   V0   
P = 3K 0   exp  −    − 1  −  0  
 V   ρ   V    V  

 
(1.115)
−2
 r  
×  0  − 2  + P0 .
 ρ  
Although not normally adopted in geochemistry, this form best takes into ac-
count the functional form of short-range repulsive forces (cf. eq. 1.54).
Second-order equations of state (i.e., equations involving the second deriva-
tive of the bulk modulus in P) have also been proposed (see for this purpose the
synoptic table in Kim et al., 1976). However these high-order equations are of
limited application in geochemistry because of the still large incertitude involved
in high-P compressional studies.
1.16 Crystal Field Theory
For better comprehension of the crystal field theory, we must reconsider to some
extent the wavelike behavior of electrons.
1.16.1 Recalls on Atomic Orbitals

The wave function of the hydrogen atom in the ground state (n ⫽ 1) is given by
1
 1 2  r 
ψ1 =   exp − , (1.116)
 π a 03   a0 
where
h2
a0 = . (1.117)
4π 2 me 2
In equation 1.116, a 0 is Bohr’s radius for the hydrogen ion in the ground state
(a 0 ⫽ 0.529 Å). The probability distribution relative to equation 1.116 is
1  −2r 
ρ 1 = ψ 12 = exp   (1.118)
π a 03  a0 
and is spherically symmetrical with respect to the center of coordinates, located

in the nucleus. The probability functions for the excited states of the hydrogen
atom have a more complex spatial dependency.
We have already seen (section 1.1.2) that the wave equation (eq. 1.2, 1.4) is function
of both time and spatial coordinates. However, the wave function may be rewritten in
the form
ψ = ψ 0 exp( 2π i vt ), (1.119)
where i is the imaginary number (i ⫽ 冑 ⫺1). In equation 1.119, ␺ 0 is a function only of

spatial coordinates, whereas the exponential part represents the periodic variation with
time.
The so-called “conjugate of ␺,” written ␺, is
ψ = ψ 0 exp( −2π i vt ). (1.120)
Then
ψ ψ = ψ 20 . (1.121)
The product in equation 1.121, which is independent of time, represents the density of
distribution—i.e., “the probability of finding a given particle in the region specified by the
spatial coordinates of ␺.” The application of wave equation 1.10 to hypothetical atoms
consisting of a nucleus of charge Z and a single electron progressively occupying the vari-
ous orbitals (hydrogenoid functions) requires transformation to polar coordinates (system
sketched in figure 1.15). The corresponding transformation of equation 1.10 is
1 ∂  ∂ψ  1 ∂  ∂ψ 
r + 2  sinθ 
2
r ∂r
2
 ∂r  r sinθ ∂θ  ∂θ 
(1.122)
1 ∂ 2ψ 8π 2 m ( E − Φ )
+ + ψ = 0.
r 2 sin 2θ ∂φ 2 h2
As already shown (Section 1.2.1), the energy eigenvalues that satisfy wave equation 1.10
are given by equation 1.22:
2π 2 me 4 Z 2
E(n) = − . (1.123)
n2 h2
In equation 1.123, n must be a positive integer.

Posing
n = n′ + l + 1 (1.124)
(Hinshelwood, 1951), it is evident that l cannot be greater than n ⫺ 1. It is also clear that
E (n) does not depend on the total value of n but on the separate values of n⬘ and l. The
correspondence between n and the principal quantum number is thus obvious: the energy
levels follow the prediction of Bohr’s model, assuming distinct values according to quan-
tum numbers n, l, and ml , which lead to configurations s, p, and d. Equation 1.122 is then
satisfied by a product of separate functions, one dependent on l and a function of angular
coordinates ␪ and ␾, and the other dependent on n, l, and a function of radial distance r:
ψ = ψ l (θ ,φ ) ψ n, l (r ) . (1.125)
In quantum states with n ⬎ 1, quantum number l assumes different values;

for instance, for n ⫽ 3, l ⫽ 0, 1, 2. When l is equal to or greater than 1, several
independent wave functions exist (2l ⫹ 1). An electron of the second level, sub-
level p, can occupy three 2p orbitals of the same energy, described by three distinct
wave functions. These orbitals, which, in the absence of external perturbation,
rigorously have the same energy, are called “atomic degenerate orbitals” (ADs).
The analytical expressions of the various wave functions, for the ground state
(n ⫽ 1) and for orbitals 2p and 3d, are listed in figure 1.15. The same figure also
shows the “form” (actually, the contour surface) attained by the five ADs of sub-
level 3d and, for comparative purposes, the three ADs of sublevel 2p and the AD
1s for the ground state (n ⫽ 1). The orientation of the atomic orbitals depends
on angular factor ␺l and not on principal quantum number n. Orbitals of the
same l thus have the same orientation, regardless of the value of n.
Figure 1.15 1s, 2p, and 3d degenerate atomic orbitals (ADs) and corresponding wave
functions (cf. eq. 1.125). Upper left: spherical coordinate system. Adapted from Harvey
and Porter (1976).
1.16.2 Crystal Field Effect

Let us now consider a coordination polyhedron, in which the cation (with its
nucleus at the center of the spherical polar coordinates) is surrounded by six
anionic ligands, located two-by-two on the polar axes (“octahedral field,” cf. fig-
ure 1.16). If we move the six ligands from infinite distance toward the nucleus of
the central cation, we generate interaction forces of the type described in section
1.11. In particular, short-range repulsion between the electron clouds of the an-
ionic ligands and the electrons of the d orbitals of the central atom gives rise to
a generalized increase in orbital energies. However, the repulsion is more marked
for orbitals oriented along the Cartesian axes (d z2 and d x2⫺y2; cf. figures 1.15 and
1.16) and less marked for orbitals transversally oriented with respect to the
Figure 1.16 Positions of anionic ligands for various configurational states: A ⫽ square-
planar coordination; B ⫽ octahedral coordination; C ⫽ tetrahedral coordination. Cation
nuclei are located in center of Cartesian coordinates.
Cartesian axes (dxy, dxz , and dyz ). The degeneracy (equal energy) of the five d
orbitals is thus partly eliminated. The energy gap between the first and second
groups of orbitals (defined by convention d␥ , d␧ or eg , t2g , respectively) is known
as “crystal field splitting” and is usually denoted by the symbol ⌬ (or 10Dq).
According to electrostatic theory, the magnitude of crystal field splitting is
given by
5 e µa 4
∆= , (1.126)
d6
where e is the electronic charge, ␮ is the dipole moment of the ligand (cf. section
1.8), a is the mean radial distance of d orbitals, and d is the cation-to-anion dis-
tance. With respect to a spherical field, where the degeneracy of d orbitals is
maintained, the increase in energy of orbitals eg is 35⌬, and the decrease in energy
of orbitals t2g is 35⌬. Orbitals eg and t2g are therefore called, respectively, “destabi-
lized” and “stabilized.”*
* The crystal field effect is due primarily to repulsive effects between electron clouds. As we
have already seen, the repulsive energy is of opposite sign with respect to coulombic attraction and
the dispersive forces that maintain crystal cohesion. An increase in repulsive energy may thus be
interpreted as actual destabilization of the compound.
Figure 1.17 Absorption spectrum of a forsteritic olivine under polarized light. Ordinate
axis represents optical density (relative absorption intensity, I/I0). From R. G. Burns
(1970), Mineralogical Applications of Crystal Field Theory. Reprinted with the permission
of Cambridge University Press.
With the cation and the coordination number held constant, the magnitude
of ⌬ depends on the type of ligand and increases according to the spectrochemi-
cal series
Br − < Cl − < F − < OH − < O 2 − < H 2 O < EDTA 4 −

(1.127)
< NH 2 − < NO − < CN − .
A “weak field” is usually defined as one generated by ligands at the beginning of

the spectrochemical series; a “strong field” is caused by ligands at the end of
the series.
The value of ⌬ may be deduced by absorption of light measurements on the

various complexes. The frequency of the maximum of the first absorption band
corresponds to the transference of an electron from a t2g orbital to an eg orbital
and directly gives the value of ⌬ expressed in cm⫺1 (see eq. 1.128, below). As an
example, figure 1.17 shows the absorption spectrum, under polarized light, of
an olivine of composition (Mg0.88Fe0.12 )2SiO4 (88% forsterite, 12% fayalite). The
position of the absorption maximum varies slightly as a function of the orienta-
tion of the polarized light with respect to optical axes, as an effect of crystal
symmetry. The frequency of the absorption maximum is related to the ⌬ parame-
ter according to
∆
va = (1.128)
h
where h is Planck’s constant.
1.16.3 Octahedral Site Preference Energy

Figure 1.18 shows energy levels for d orbitals in crystal fields of differing symme-
try. The splitting operated by the octahedral field is much higher than that of the
tetrahedral field (⌬t ⫽ 49⌬ O ) and lower than the effect imposed by the square
planar field. If we define as “crystal field stabilization energy” the decrease in
Figure 1.18 Crystal field splitting for d orbitals: A ⫽ square-planar field; B ⫽ octahedral
field; C ⫽ tetrahedral field.
Table 1.17 Electronic configuration and crystal field stabilization energies for 3d electrons in
transition elements. (Ar) ⫽ argon core: 1s22s22p63s23p6.
Electronic Octahedral Tetrahedral

Ion Configuration Spin Field Field OSPE
Sc3⫹; Ti4⫹ (Ar)3d 0 0 0 0

Ti3⫹ (Ar)3d 1 ↑ 2
5
12
45
6
45
V3⫹ (Ar)3d 2 ↑ ↑ 4
5
24
45
12
45
Cr3⫹ (Ar)3d 3 ↑ ↑ ↑ 6
5
16
45
38
45
Mn2⫹; Fe3⫹ (Ar)3d 5 ↑ ↑ ↑ ↑ ↑ 0 0 0

Fe2⫹ (Ar)3d 6 ↑↓ ↑ ↑ ↑ ↑ 2
5
12
45
6
45
Co2⫹ (Ar)3d 7 ↑↓ ↑↓ ↑ ↑ ↑ 4
5
24
45
12
45
Ni2⫹ (Ar)3d 8 ↑↓ ↑↓ ↑↓ ↑ ↑ 6
5
16
45
38
45
Cu2⫹ (Ar)3d 9 ↑↓ ↑↓ ↑↓ ↑↓ ↑ 3
5
8
45
19
45
Zn2⫹ (Ar)3d 10 ↑↓ ↑↓ ↑↓ ↑↓ ↑↓ 0 0 0
energy with respect to a spherical field and compare the values for the tetrahedral
and octahedral fields, we see that, for all configurations (besides 3d 0 , 3d 5, and
3d 10 , for which crystal field stabilization is obviously zero), the octahedral field
has a higher stabilization effect, called “Octahedral Site Preference Energy”
(OSPE) (cf. table 1.17).
In 1957, almost contemporaneously, McClure (1957) and Dunitz and Orgel
(1957) proposed precise calculations of crystal field stabilization energies in tetra-
hedral and octahedral fields, and the corresponding OSPE values. Their data are
summarized in table 1.18. The ⌬ o values reported by McClure were actually de-
rived for hydrated ions (although the OH⫺ groups are near O2⫺ in spectrochemi-
cal series 1.127), whereas those listed by Dunitz and Orgel were obtained for
crystalline oxides and glasses from spectroscopic data on absorption of light. As
noted by Dunitz and Orgel, the ⌬ o values for O2⫺ and OH⫺ ligands are identical
for bivalent ions but very different for trivalent ions (8 to 12% lower for O2⫺ with
respect to OH⫺ ligands). In both works, the ⌬t value was not experimentally
obtained but was assumed to correspond to 49⌬ 0 , in agreement with theory. The
estimated OSPE values thus have a high level of incertitude. In some instances,
for trivalent ions, Dunitz and Orgel imposed the “strong field” condition (cf. sec-
tion 1.16.2).
Both McClure (1957) and Dunitz and Orgel (1957) derived their OSPE values
with the aim of explaining the intracrystalline distribution of transition elements
observed in multiple oxides of cubic structure (spinels). According to these au-
thors, a more or less high OSPE value was crucial in intracrystalline partitioning.
Their hypothesis was later taken up again by several authors. In particular, Burns
(1970) furnished several examples of natural compounds, calculating the crystal
field stabilization energy terms from adsorption spectra under polarized light (cf.
figure 1.17).
Nevertheless, the application of crystal field theory to natural compounds
Table 1.18 Crystal field stabilization energies for 3d transition elements according to
McClure (1957) (1) and Dunitz and Orgel (1957) (2).
Octahedral Tetrahedral
⌬o ⌬t Stabilization Stabilization OSPE
Ion (cm⫺1) (cm⫺1) Energy (kJ/mole) Energy (kJ/mole) (kJ/mole)
Ti3⫹ (1) 20,300 9000 96.7 64.4 32.3
Ti3⫹ (2) 18,300 8130 87.4 58.6 28.8
V3⫹ (1) 18,000 8400 128.4 120.0 8.4
V3⫹ (2) 16,700 7400 160.2* 106.7 53.5*
V2⫹ (1) 11,800 5200 168.2 36.4 131.8
Cr3⫹ (1) 17,600 7800 251.0 55.6 195.4
Cr3⫹ (2) 15,700 6980 224.7 66.9* 157.8*
Cr2⫹ (1) 14,000 6200 100.4 29.3 71.1
Mn3⫹ (1) 21,000 9300 150.2 44.4 105.8
Mn3⫹ (2) 18,900 8400 135.6 40.2 95.4
Mn2⫹ (1) 7,500 3300 0 0 0
Fe3⫹ (1) 14,000 6200 0 0 0
Fe2⫹ (1) 10,000 4400 47.7 31.4 16.3
Fe2⫹ (2) 10,400 4620 49.8 33.1 16.7
Co3⫹ (1) – 7800 188.3 108.8 79.5
Co2⫹ (1) 10,000 4400 71.5 62.7 8.8
Co2⫹ (2) 9,700 4310 92.9* 61.9 31.0*
Ni2⫹ (1) 8,600 3800 122.6 27.2 95.4
Ni2⫹ (2) 8,500 3780 122.2 36.0* 86.2*
Cu2⫹ (1) 13,000 5800 92.9 27.6 65.3
Cu2⫹ (2) 12,600 5600 90.4 26.8 63.6
*Valid for strong field conditions.
must be carried out with extreme caution. OSPE is only a tiny fraction of the
cohesive energy of the crystal. For spinels in particular, it can be shown that the
transition from normal structure (type [A]T [BB]OO4 , where [A]T is a bivalent ion
in a tetrahedral site and [B]O is a trivalent ion in an octahedral site) to inverse
structure ( [B]T [AB]OO4 ) corresponds in terms of lattice energy to less than 200
kJ/mole and also depends on other energy effects. As an example, table 1.19 com-
pares lattice energy terms (⫺U ⫽ EC ⫹ ER ), the modulus of the bulk energy gap
between normal and inverse conditions ( | ⌬U | ), and the modulus of the crystal
field stabilization energy gap ( | ⌬CFSE | ) for some spinel compounds. In the
light of the contents of table 1.19 and of the above discussion, it is not surprising
that CFSE predictions (without detailed consideration of all forms of static ener-
gies) are denied in most cases by experimental observations.
The CFSE contribution to lattice energy is almost insignificant for meta- and
orthosilicates in which normal and distorted octahedral coordinations are present
(M1 and M2 sites, respectively). As shown in table 1.20, the CFSE gap between
normal and distorted octahedral fields is in fact only a few kJ/mole.
Table 1.19 Lattice energy and CFSE for selected spinel compounds.
Values of U, Ec , and Er from Ottonello (1986). CFSE values calculated
with values proposed by McClure (1957). Data in kJ/mole; x ⫽ degree of
inversion.
Compound x U EC ER | ⌬U | | ⌬CFSE |
CoFe2O4 1 18,712 ⫺21,730 3018 270 8.8
CoCr2O4 0 18,918 ⫺22,090 3172 300 186.6
CuFe2O4 1 18,805 ⫺21,697 2892 150 65.3
FeAl2O4 0 19,029 ⫺22,844 3815 400 16.3
FeFe2O4 1 18,655 ⫺21,664 3009 220 16.3
FeCr2O4 0 18,810 ⫺22,029 3219 260 179.1
NiCr2O4 0 18,948 ⫺22,116 3168 200 100.0
NiFe2O4 0 18,793 ⫺21,829 3036 220 95.4
Table 1.20 CFSE for M1 and M2 sites in some

silicates. Data from Burns (1970) and Wood (1974),
expressed in kJ/mole.
Compound Ion Site M1 Site M2 M1 ⫺ M2

2⫹
Fayalite Fe 57.7 53.6 4.1
Forsterite Fe2⫹ 61.7 54.8 6.9
Bronzite Fe2⫹ 48.1 49.0 ⫺0.9
Ferrosilite Fe2⫹ 46.0 47.3 ⫺1.3
Ni-Mg olivine Ni2⫹ 114.2 107.5 6.7
1.16.4 Field Strength and Spin Condition

Hund’s rule establishes that all orbitals of a given sublevel must have been occu-
pied by single electrons with parallel spins before two electrons of opposite spins
can occupy a single sublevel. Let us consider, for instance, a metallic ion with
d 5 configuration. Based on Hund’s rule, the first three electrons occupy the t2g
degenerate orbitals with parallel spins. The remaining two electrons then have
two possibilities:
1. to occupy the two eg degenerate orbitals with parallel spins or

2. to fill two of the three t2g degenerate orbitals with opposite spins.
These two configurations are energetically determined by the influence of the

crystal field and are denoted, respectively, as conditions of “high” or “low” spin.
In the presence of a weak octahedral field (generated by ligands at the beginning
of the spectrochemical series), the value of ⌬o is low and the resulting electron
configuration is that sketched in figure 1.19A. In the presence of a strong field
(ligands at the end of the spectrochemical series), crystal field splitting is high
and electrons tend to occupy the t2g stabilized orbitals, opposing the spin and
Figure 1.19 Electron configurations under high spin (A) and low spin (B), corresponding,
respectively, to weak and strong field conditions.
leaving the destabilized eg orbitals free (figure 1.19B). The magnetic properties of
the compound (which depend on the electron spin condition) are thus influenced
by the crystal field. It should be recalled here that crystal field theory was origi-
nally developed mainly to account for the magnetic properties of crystalline com-
pounds (Bethe, 1929).
1.17 Ligand Field Theory
1.17.1 Concepts of Molecular Orbitals

Like atomic orbitals (AOs), molecular orbitals (MOs) are conveniently described by quan-
tum mechanics theory. Nevertheless, the approach is more complex, because the interac-
tion involves not simply one proton and one electron, as in the case of AOs, but several
protons and electrons. For instance, in the simple case of two hydrogen atoms combined
in a diatomic molecule, the bulk coulombic energy generated by the various interactions
is given by four attractive effects (proton-electron) and two repulsive effects (proton-
proton and electron-electron; cf. figure 1.20):
 1 1 1 1   1 1 
ΦC = − e 2  + + +  +e 
2
+  . (1.129)
 dp e
A A
dp e dp e
B B
dp e 
A B
 dp p
B A
de e 
A B A B
To obtain the wave function appropriate to the case examined in figure 1.20, we must
introduce ⌽C in the general equation (eq. 1.10). However, the complexity of the problem
does not allow an analytical solution, and the wave function is thus derived by application
Figure 1.20 Coulombic interactions in a diatomic molecule composed of hydrogen ions;

pA and eA are the proton and electron of atom A, and pB and eB are those of atom B.
of the “variational principle” (Hirschfelder et al., 1954). The general equation is rewritten
in the form
 h2 
Φ − ∇2  ψ = Eψ . (1.130)
 8π 2 m 
The term in parentheses in equation 1.130 is defined as the “quantum-mechanic hamilto-

nian” (H ; cf. appendix 2):
Hψ = Eψ . (1.131)
Multiplying both terms in equation 1.131 by ␺, extracting E, and integrating over the
Cartesian coordinates, we obtain the energy of molecular orbitals (MOs):
E =
∫ ψ H ψ dτ ,
(1.132)
∫ ψ 2 dτ
where
dτ = dx ⋅ dy ⋅ dz. (1.133)
Equation 1.132 has the property of never giving an energy that is lower than the true
energy (resonance principle; cf. Pauling, 1960). This property allows us to assign to the
MO wave functions that are obtained by linear combination of the AO functions of the
separate atoms (Linear Combination of Atomic Orbitals, or LCAO, method), by progres-
sive adjustment of the combinatory parameters, up to achievement of the lowest energy.
The LCAO approximation may be expressed as:
ΨMO = ψ A + λψ B , (1.134)
where ␺A and ␺B are wave functions of the AOs of atoms A and B, respectively. The ␭
constant denotes the bond prevalence and is an expression of the polarity of MO. For
instance, with ␭ ⫽ 0, the MO of the AB molecule coincides with the AO of atom A; the
electron of B is completely displaced on A and the bond is purely ionic. With ␭ ⫽ 1, the
MO is a weighted mean of the AOs of A and B, the electrons rotate along a bicentric orbit
around the two nuclei, and the bond is perfectly covalent.
If we apply the variational procedure to hydrogen atoms at ground state, we find that
the lowest orbitals in order of increasing energy are
Ψg = ψ A(l s ) + ψ B(l s ) ( λ = 1) (1.135)
and
Ψu = ψ A(l s ) − ψ B(l s ) ( λ = −1) (1.136)
( g ⫽ gerade ⫽ even parity; u ⫽ ungerade ⫽ odd parity).

Orbital ⌿g has lower energy with respect to the two constituent AOs and thus forms
a stable molecule (cf. section 1.3). This orbital is thus called a “bonding orbital” (or “at-
tractive state”). Orbital ⌿u has higher energy with respect to the two AOs and is called an
“antibonding orbital” (or “repulsive state”) (cf. figure 1.21). The above discussion is valid
for any kind of molecule—i.e., “the combination of any couple of AOs leads always to
the formation of one bonding and one antibonding MO”.
The two MOs of a diatomic molecule are classified as ␴, ␲, ␦, etc., according to
whether the projection of angular momentum along the molecular axis is 0, ⫾1, ⫾2, etc.
(in units of h/2␲ ). Moreover, the antibonding condition is indicated by an asterisk; for
Figure 1.21 Bonding and antibonding conditions for a hydrogen molecule. Adapted from
Harvey and Porter (1976).
instance, ␴ *1s is the antibonding MO with rotational symmetry of the hydrogen molecule
(cf. figure 1.21).
The choice of the two AOs that can efficiently combine to form the MO may be based
on three principles:
1. The energies of the two AOs ␺A and ␺B must be of similar magnitude.

2. The two AOs must overlap as much as possible.
3. The two AOs must have the right symmetry with respect to the internuclear axis.
Based on the above principles, only some combinations among the AO of the cation
and those of the ligands are possible. Plausible combinations are selected as follows:
1. Once the symmetry of the complex is known, the AO of the central cation that
best conforms to the studied case is selected—i.e., for an octahedral complex, we
select the AO directed along the Cartesian axes [ns, npx , npy , npz , (n ⫺ 1) d z2 , (n
⫺ 1)d x2 ⫺y 2 ].
2. For each ligand, the six AOs appropriate to the symmetry of the central cation
(denoted by symbols ␾1 , ␾2 , ␾3 , ␾4 , ␾5 , ␾6 ) are selected.
3. By applying the LCAO method and the principle of lowest energy, the selected
AOs are combined to form the ⌿ function of the corresponding MO. Figure
1.22A shows an example of combination of AO dxy with ␲⫺bonding orbitals, and
figure 1.22B shows a probable combination between d x2 ⫺y 2 orbitals of the central
cation with ␴⫺bonding orbitals. The function corresponding to the latter case is
12
 1 − α2 
Ψx 2 − y 2 = α d x 2 − y 2 +  ( φ1 + φ2 − φ 3 − φ 4 ) (1.137)
 4 
and the antibonding condition is expressed by
α
Ψx*2 − y 2 = (1 − α 2 ) 1 2 d x 2 − y 2 − ( φ1 + φ2 − φ 3 − φ 4 ) (1.138)
2
(a complete tabulation of LCAO-MO wave functions can be found in classical

physical chemistry textbooks). In equations 1.137 and 1.138, the ␣ term denotes
the degree of overlapping of the AOs of the anionic ligands with those of the
central metal cation. This term is analogous to the ␭ factor in equation 1.134.
With ␣2 ⫽ 0, the bond is purely ionic; with ␣2 ⫽ 12, it is purely covalent.
The energy levels for the ␴ orbitals of the octahedral complex are sketched in figure
1.23A. For each orbital, both bonding (⌿) and antibonding (⌿*) conditions are repre-
sented. Also, the energies of the ⌿*z 2 and ⌿*x 2⫺y 2 MOs are identical so that, in analogy
with AO nomenclature, they are defined as “double degenerate” (the ⌿pz , ⌿py , and ⌿pz
MOs and their corresponding antibonding conditions are thus “triple degenerate”). Orbit-
als dxy , dxz , and dyz do not form any ␴ bonds and thus maintain their original energies.
Figure 1.22 Examples of ␲-bonding orbitals (A) and ␴-bonding orbitals (B) for d elec-
trons.
1.17.2 Ligand Field Effect

Let us now imagine locating the electrons on the various sublevels according to
the Aufbau principle, applying Hund’s rule to degenerate orbitals. As an octa-
hedral complex has 12 electrons pertaining to the ligands, they occupy two-by-
two the ⌿ z2, ⌿x2⫺y2, ⌿s , ⌿px , ⌿py , and ⌿pz bonding MOs; consequently six d
electrons (dxy, dxz , and dyz ) remain on their original AOs but four are displaced
on antibonding degenerate orbitals ⌿* x 2 and ⌿*x 2 ⫺y 2 . We therefore reach the same
condition described by crystal field theory: three (triply) degenerate t2g orbitals
(AOs dxy, dxz , and dyz ) and two (doubly) degenerate eg orbitals (antibonding
MOs ⌿* z 2 and ⌿*x 2 ⫺y 2 ).
Figure 1.23 General energy level diagram for an octahedral complex (A) and energy lev-
els of pyrite (B). Part (B) from Burns and Vaughan (1970). Reprinted with permission of
The Mineralogical Society of America.
The analogy between the two theories is only formal. Crystal field theory is
a purely electrostatic approach that does not take into consideration the forma-
tion of MOs and the nature of the bond. According to crystal field theory, optical
and magnetic properties are ascribed to crystal field splitting between two AOs,
whereas in ligand field theory energy splitting occurs between AOs (dxy, dxz , and
dyz ) and antibonding MOs ⌿* x 2 ⫺y 2 . Lastly, note that the ␴ condition
z 2 and ⌿*
implies combinations between AOs that are oriented in the same direction, and
crystal field theory assigns the highest energy to such configurations.
The fact that the ligand field theory can account for the fractional character
of the bond through the ␣ parameter (cf. eq. 1.137 and 1.138) leads to important
applications for covalent compounds, for which electrostatic theory is not ade-
quate.
The most classical example of application is that of sulfur compounds. Figure
1.23B shows the energy level scheme developed by Burns and Vaughan (1970) for
pyrite (FeS2 ).
The structure of FeS2 is similar to that of NaCl, with Fe2⫹ occupying the
position of Na⫹ and that of Cl ⫺ occupied by the mean point of a (S-S) group.
Each iron atom is surrounded by six sulfur atoms, pertaining to six distinct
(S-S) groups.
The energy level diagram of figure 1.23B is similar to that of figure 1.23A
(although the formal representation of degeneracy is different), but two distinct
conditions of ligand field separation, corresponding to ␴ and ␲ symmetry, are
considered in this case. The ␴ orbitals of the ligands are composed of the six
hybrid orbitals 3s, 3p of sulfur (one for each ligand at the corners of the coordina-
tion polyhedron). The 3d orbitals of Fe2⫹ do not form MOs and remain un-
changed on the AO of the metal cation. Burns and Vaughan (1970) define ⌬cov⫺␴
as the crystal field separation between the 3d AO of Fe2⫹ and the t2g antibonding
MOs, implying that the bond is purely covalent (we recall for this purpose that
the condition of perfect covalence is reached when ␣2 ⫽ 12; see, for instance, eq.
1.138). These authors also suggest that, in FeS2 compounds, bonding and anti-
bonding MOs with ␲ symmetry may be formed between the 3d orbitals of the
(S-S)2⫺ group and those of Fe2⫹. In this case, energy splitting should be higher
than for the ␴ symmetry and therefore this should be the type of bonding present
in pyrite. When comparing figures 1.23A and B, it must be noted that although
the two representations of degenerate orbitals are different (on the same line in
part A and piled in part B), they have rigorously the same significance. Besides
the quoted work of Burns and Vaughan (1970) on transition element disulfides
(CuS2 , NiS2 , CoS2 , FeS2 ), ligand field theory has been successfully applied to
thiospinels (AB2S4 compounds with cubic structures) with a full account of sev-
eral of their physical properties (hardness, reflectivity, mutual solubility, density;
Vaughan et al., 1971).
1.17.3 Jahn-Teller Effect

“All nonlinear nuclear configurations are unstable for an orbitally degenerate elec-
tronic state.” This principle, formulated on theoretical grounds by Jahn and Teller
(1937) by application of group theory to the electronic states of polyatomic mole-
cules, has profound consequences for the significance of local symmetry in crystal
lattices. This principle establishes that the perfect symmetric condition of a regu-
lar coordination polyhedron is always unstable with respect to a distorted state,
whenever the central cation (or the cation-ligand MO) has orbital degeneracy.
Jahn and Teller (1937) explicitly stated that the effect of instability applies to
degenerate electrons that contribute appreciably to molecular binding. Inner
shells are thus excluded from the effect of nuclear displacement.
Too often the Jahn-Teller principle is presented in a reversed fashion—i.e.,
“the distortion of a coordination polyhedron eliminates orbital degeneracy.” This
formulation renders the principle ineffective, because the distortion is at first sight
attributed to external forces, whereas in fact it is the result of the intrinsic instabil-
ity of the symmetric condition.
The Jahn-Teller principle finds applications both in the framework of crystal
field theory and in the evaluation of energy levels through the LCAO-MO ap-
proach. In both cases, practical applications are restricted to 3d transition ele-
ments.
Octahedrally coordinated transition ions with odd numbers of 3d electrons in
the destabilized eg orbitals (i.e., d 4, d 9, and low-spin d 7 configurations) are subject
Figure 1.24 Energy splitting of antibonding MOs in tetragonally distorted octahedral

sites: (left) elongation along axis z; (right) compression along axis z.
to marked Jahn-Teller distortion. The same is true for d 3, d 4, d 8, and d 9 configu-

rations in tetrahedral coordination. Let us consider, for instance, Cu2⫹ in octahe-
dral coordination with oxygen ligands. Cu2⫹ has a (t2g )6(eg )3 configuration. If the
octahedron is regular, the eg orbitals are degenerate, but if it is elongated in direc-
tion z, the short-range repulsion between the d z2 AO of the central cation and the
AO of oxygen is lower in direction z and higher on the x-y plane.
Orbitals d x2⫺y2 are thus destabilized with respect to d z2, and two electrons
occupy d z2 in a low-spin configuration. The opposite is observed in the case of
compressional distortion along axis z. However, this “crystal field” interpretation
is more appropriately translated in terms of ligand field LCAO-MO theory by
stating that ligand field splitting occurs between antibonding orbitals ⌿*. For a
d 9 complex with three electrons in the antibonding MOs, the alternative configu-
2 1 1 2
rations (⌿* z 2 ) (⌿*
x 2 ⫺y 2 ) and (⌿*
z 2 ) (⌿*
x 2 ⫺y 2 ) may be attained. The first con-
figuration leads to a higher antibonding character (higher energy) in direction z
(left side of figure 1.24); the second leads to higher energy along x and y (right
side of figure 1.24). Higher energy is equivalent to a weaker bond and hence to
a greater bond distance: the Jahn-Teller effect thus results in distortion of the
coordination polyhedron.
In geochemistry, the Jahn-Teller effect is relevant for metals Fe2⫹ and Cu2⫹
in octahedral complexes and for Cr3⫹ and Ni2⫹ in tetrahedral complexes. Other
transition ions (e.g., Cr2⫹ and Co2⫹ ) require unusual oxidation or low-spin condi-
tions that can be reached only under extreme pressure.
Distortion of coordination polyhedra is commonly observed in silicates. Be-
cause distortion of the coordination field implies elimination of degeneracy, it is
important to evaluate to what extent the nondegenerate condition contributes to
the stability of the phase. Systematic MO calculations in this direction have been
carried out by Wood and Strens (1972) on the basis of the AOM (Angular Over-
lap Model) integrals of Ballhausen (1954).
1.18 The Gordon-Kim Modified Electron Gas

(MEG) Model
The ionic models discussed in section 1.12 involve some sort of empiricism in
the evaluation of repulsive and dispersive potentials. They thus need accurate
parameterization based on experimental values. They are useful in predicting in-
teraction energies within a family of isostructural compounds, but cannot safely be
adopted for predictive purposes outside the parameterized chemical system or in
cases involving structural changes (i.e., phase transition studies).
However, the energies of crystalline substances can be evaluated from first
principles with reasonable precision. This goal, which is now in sight as a result
of impressive improvements in automated computing capabilities, will lead to
significant advances in theoretical computational geochemistry, as excellently
outlined by Tossell and Vaughan (1992).
We will limit ourselves here to introducing the simplest of the quantum me-
chanics procedures: the Modified Electron Gas (MEG) treatment of Gordon and
Kim (1971). This procedure is on the borderline between the classical atomistic
approach and quantum mechanics ab initio calculations that determine energy by
applying the variational principle. A short introduction to MEG treatment
should thus be of help in filling the conceptual gap between the two theories.
MEG treatment is based on the following main assumptions:
1. Each atom is represented by its electronic charge density. No re-

arrangement or distortion of separate atomic densities takes place when
atoms are brought together.
2. In the overlap region between two atoms (A and B), densities are additive
and the total electron density is
ρ = ρ A + ρB . (1.139)
3. Total interaction energy is represented by a coulombic term (VC ) calcu-

lated between all charges (both electrons and nuclei), plus an electron gas
term (Veg ) related to local density in the overlap region:
Vtotal = VC + Veg. (1.140)
4. Only the outer regions of the atoms in which atomic densities overlap
contribute significantly to the interaction.
5. The electron densities of the separate atoms are described by Hartree-
Fock wave functions approximated by analytic extended (Slater-type) ba-
sis sets.
The coulombic energy between interacting atoms A and B at distance R is

the summation of four terms:
Z AZB [ ρ A ( r1 ) + ρ B ( r1 )][ ρ A ( r2 ) + ρ B ( r2 )]
ΦC =
R
+ ∫∫ r12
dr1 dr2
(1.141)
[ ρ A ( r1 ) + ρ B (r1 )] ρ A ( r1 ) + ρ B ( r1 )]
−Z A ∫ r1 A
dr1 − Z B ∫ r1B
dr1 ,
where ZA and ZB are nuclear charges of atoms A and B, r12 is the distance between
two electrons, and r1A and r1B are the distances between electrons and nuclei.
The first term on the right in equation 1.141 is repulsion between the two
nuclei, the second term represents electron-electron repulsion, and the third and
fourth terms are attraction effects between electrons and nuclei. If we subtract the
coulombic energies of separate atoms from equation 1.141:
1  ρ A ( r1 ) ρ A ( r2 )   ρ A ( r1 ) 
ΦC , A =
2 ∫ 
 r12
 dr1 dr2 −Z A

∫ 
 r1 A 
 dr1 (1.142)
1  ρ B ( r1 ) ρ B ( r2 )   ρ B ( r1 ) 
ΦC ,B =
2 ∫ 
 r12
 dr1 dr2 −Z B

∫ 
 r1B 
 dr1 , (1.143)
we obtain the coulombic interaction VC:
Z AZB ρ A ( r1 ) ρ B ( r2 ) ρ A ( r1 )
VC =
R
+ ∫∫ r12
dr1 dr2 − Z B ∫ r1B
dr1
(1.144)
ρ B ( r1 )
−Z A ∫ r2 A
dr2 ,
All integrals in equation 1.144 may be evaluated analytically but, in the case
of multielectron atoms, the calculations are quite complex and involve many-term
summations. Gordon and Kim (1971) thus suggest a solution by two-dimensional
(Gauss-Laguerre or Gauss-Legendre) numerical quadrature after reduction of
equation 1.144 into a single integrand that, in the case of two interacting univalent
ions, takes the form
VC = − R −1 + ∫∫ ρ A ( r1 ) ρ B ( r2 )
 ZA Z B + 1 ZB ZA 
 R −1 + r12−1 − r1−B1 − r2−A1  dr1 dr2 . (1.145)
(
 A Z − 1 )( Z B + 1) Z B + 1 ZA − 1 
The form of equation 1.145 is obtained by introducing
∫ ρ A ( r ) dr = ZA − 1 (1.146)
and
∫ ρB ( r ) dr = ZB − 1. (1.147)
Similar forms can be obtained for coulombic interaction between neutral atoms or be-
tween atoms and ions by appropriate consideration of the integrated electron densities of
the two species (in the case of interaction between two neutral atoms, the right sides of
equations 1.146 and 1.147 are, respectively, equivalent to the nuclear charges of the two
species: ZA and ZB ).
The energy of the electron gas is composed of two terms, one Hartree-Fock
term (⌽HF ) and one correlation term (⌽corr ). The Hartree-Fock term comprises
the zero-point kinetic energy density and the exchange contribution (first and
second terms on the right in equation 1.148, respectively):
13
3 4  3
ΦHF ( ρ ) = ( 3π 2 ) 2 3 ρ 2 3 −   ρ 1 3. (1.148)
10 3 π
The correlation term (expressed in Hartree) is accurately reproduced at high elec-

tron density (rs ⱖ 10) by
Φcorr = 0.0311 ln(r s ) − 0.048 + 0.009r s ln(r s ) − 0.018r s (1.149)
and at low electron density (rs ⱕ 0.7) by
−1 −3 2 −2 −5 2
Φcorr = − 0.438r s + 1.325r s − 1.47r s − 0.4r s , (1.150)
where rs is related to electron density through
3
rs = 3 , (1.151)
4π a 03 ρ
in which a 0 is the Bohr radius.

In the intermediate electron density range, Gordon and Kim (1971) used the
interpolant expression:
Φcorr = − 0.06156 + 0.01898 ln(r s ) (1.152)
(alternative equations have been proposed, following the Gordon-Kim introduc-

tory paper; see Clugston, 1978 for a discussion).
The electron gas functional is given by the summation of the two terms:
Φeg = ΦHF + Φcorr (1.153)
and the electron gas contribution to the interatomic interaction is
Veg = ∫ {[ ] }
dr ρ A ( rA ) + ρ B ( rB ) Φeg ( ρ A + ρ B ) − ρ A ( rA )Φeg ( ρ A ) − ρ B ( rB )Φeg ( ρ B )
(1.154 )
Because electron densities are positive-definite, the contribution of the elec-

tron gas to interaction energy takes the same sign as the electron gas functional
(see Clugston, 1978). Hence, the kinetic energy term is repulsive (cf. eq. 1.148
and 1.154), exchange energy is attractive (cf. eq. 1.148 and 1.154), and the correla-
tion term is also attractive (cf. eq. 1.149, 1.150, and 1.152).
Since its appearance in 1971, the MEG model has been improved to various
extents. The most effective modification was perhaps that suggested by Rae
(1973), which introduced a correction factor to lower the exchange energy and
added dispersive series (these modifications were partly embodied in the MEG
calculation routine by Green and Gordon, 1974; see Clugston, 1978 for a careful
treatment). The main advantage of the Gordon-Kim MEG treatment with re-
spect to more sophisticated ab initio or density functional theories is undoubtedly
its computational economy.
The MEG model has been extensively used to determine lattice energies and
interionic equilibrium distances in ionic solids (oxides, hydroxides, and fluorides:
Mackrodt and Stewart, 1979; Tossell, 1981) and defect formation energies (Mack-
rodt and Stewart, 1979). Table 1.21 compares the lattice energies and cell edges
of various oxides obtained by MEG treatment with experimental values.
Table 1.21 MEG values of lattice energy and lattice parameter for
various oxides, compared with experimental values. Source of data:
Mackrod and Stewart (1979). UL is expressed in kJ/mole; a0 (in Å)
corresponds to the cell edge for cubic substances, whereas it is the lattice
parameter in the a plane for Al2O3, Fe2O3, and Ga2O3 and it is the lattice
parameter parallel to the sixfold axis of the hexagonal unit cell in rutiles
CaTiO3 and BaTiO3.
UL a0
Oxide MEG Experimental MEG Experimental

Li2O 3,049 2,904 2.30 2.31
BeO 4,767 4,526 4,603 2.72 2.70
MgO 3,932 3,937 3,898 2.18 2.11
␣-Al2O3 15,622 15,477 13.2 13.0
CaO 3,474 3,570 3,483 2.46 2.41
TiO2 11,811 12,158 4.52 4.59
MnO 3,618 3,815 2.36 2.22
␣-Fe2O3 14,349 15,082 14.3 13.7
Ga2O3 14,445 15,188 15,613 14.3 13.4
SrO 3,271 3,310 3,261 2.62 2.58
CdO 3,348 3,879 2.58 2.35
SnO2 10,962 11,869 11,367 4.87 4.74
BaO 3,088 3,126 3,078 2.78 2.76
CeO2 10,392 10,344 2.72 2.71
PbO2 10,585 11,753 11,570 5.05 4.95
ThO2 9,794 9,939 10,103 2.87 2.80
UO2 9,881 10,296 2.84 2.73
CaTiO3 15,352 15,622 3.90 3.84
BaTiO3 14,908 15,265 4.04 4.01
MgAl2O4 19,646 19,366 4.12 4.04
Although the MEG model is essentially ionic in nature, it may also be used
to evaluate interactions in partially covalent compounds by appropriate choices
of the wave functions representing interacting species. This has been exemplified
by Tossell (1985) in a comparative study in which MEG treatment was coupled
with an ab initio Self Consistent Field-Molecular Orbital procedure. In this way,
Tossell (1985) evaluated the interaction of CO32⫺ with Mg2⫹ in magnesite
(MgCO3 ). Representing the CO32⫺ ion by a 4–31G wave function, holding its
geometry constant with a carbon-to-oxygen distance of 1.27 Å, and calculating
its interaction energy at various Mg-C distances, Tossell was able to reproduce
the experimental structure with reasonable precision. Calculations also showed
that the covalently bonded CO32⫺ anion has a smaller repulsive interaction with
Mg2⫹ than that of the overlapping free ions C4⫹(O2⫺ )3 . This results from the fact
that the electron densities along the C-O internuclear axis are higher near the
midregion of the C-O bond than the sum of overlapping spherical free-ion den-
Figure 1.25 Electron density in CO32⫺ minus sum of superimposed spherical free-ion
densities along C-O internuclear axis. Reprinted from Tossell (1985), with permission of
Elsevier Science B.V.
sities, whereas they are lower outside the oxygen nucleus (figure 1.25). Because
electron density falls in the peripheral zone of the CO32⫺ group, short-range re-
pulsion falls with approaching cations.*
* Incidentally it must be noted that the interaction energy of Mg2⫹ with CO32⫺ does not corre-
spond to the lattice energy of the crystalline substance, but is about 1⁄10 of its value. Most of the
cohesive energy is due to the covalent bonds within the CO32⫺ group.
1.19 Polarization Energy
We have already introduced the concept of ionic polarizability (section 1.8) and
discussed to some extent the nature of dispersive potential as a function of the
individual ionic polarizability of interacting ions (section 1.11.3). We will now
treat another type of polarization effect that is important in evaluation of defect
energies (chapter 4).
Let us imagine subtracting a ⫺e charge at a given point in the crystal lattice:
at that point, effective charge ⫹e exerts polarization energy of the type
1
EP = − eΦe , (1.155)
2
where ⌽e is the electric potential induced by the polarized medium at charge ⫹e

taken at the origin. If the medium is composed of i ions, each at distance ri and
with polarizability ␣i , equation 1.155 will be expressed as
1 µ 1 F
EP = − e ∑ 2i = − e ∑ α i 2i , (1.156)
2 i ri 2 i r
where Fi is the component directed along ri (toward effective charge ⫹e) of total
electric field ⌽e on the ith ion and ␮i is the corresponding dipole moment.
Applying differential operators (cf. appendix 2), we can write
 e γ 
Fi =  2 + F j i , (1.157)
 ri ri 
where ␥i is the modulus of dipole ij at distance ri from charge ⫹e and Fj is the

summation of the gradients of the electric potential from the various j ions toward
i—i.e.,
Fj = ∑ ∇ i Φe , j (1.158)
j
and
 ∂ Φe, j ∂ Φe, j ∂ Φe, j 

∇ i Φe, j =  , , .
 ∂x ∂y ∂ z  (1.159)
The solution of the system shown in equations 1.156 to 1.159 leads to a large
number of equations (cf. Rittner et al., 1949; Hutner et al., 1949), and simplifying
Figure 1.26 Induced polarization effects arising from a cationic vacancy in a crystal (neg-
ative net charge). Arrows mark induced dipoles at various lattice positions. From Lasaga
(1981c). Reprinted with permission of The Mineralogical Society of America.
assumptions must be introduced to solve the calculations properly. The generally

adopted method is that of Mott and Littleton (1938), which solves the polariza-
tion energy between two ions (i and j) through the system
1
EP = − eΦe (1.160)
2
 1 1 
Φe = − ea 3  M i ∑ r4
+Mj ∑ r4 (1.161)
 i j 
2α i 1  1
Mi = 1 −  (1.162)
α i + α j 4π  εo 
2α j 1  1
Mj = 1 −  , (1.163)
αi + α j 4π  εo 
where a3 is the volume occupied by the pair of ions and ␧ 0 is the dielectric con-
stant of the medium.
Table 1.22 Optical (␣e) and static (␣T) polarizabilities for silicates. Calculated values
obtained by additivity principle starting from polarizability of oxide constituents (adapted
from Lasaga and Cygan, 1982).
␣e ␣T
Mineral Formula Calculated Experimental Calculated Experimental

Forsterite Mg2SiO4 6.44 6.35 11.48 11.55
Fayalite Fe2SiO4 8.80 8.33 12.62 12.12
Orthoclase KAlSi3O8 13.31 13.11 23.83 25.98
Enstatite MgSiO3 4.65 4.56 8.16 7.96
Diopside CaMgSi2O6 10.48 9.85 18.20 18.78
Tephroite Mn2SiO4 8.54 8.26 13.76 14.45
Phenacite Be2SiO4 5.48 5.46 9.14 8.51
Willemite Zn2SiO4 8.62 8.08 13.78 14.35
Monticellite CaMgSiO4 7.62 7.52 13.36 14.56
Rhodonite MnSiO3 5.70 5.62 9.30 10.45
Zircon ZrSiO4 7.50 7.55 12.46 11.67
Edembergite CaFeSi2O6 11.66 10.95 18.77 22.89
Giadeite NaAlSi2O6 9.61 8.84 17.35 17.96
Leucite KAlSi2O6 10.45 10.46 18.99 23.09
Nefeline NaAlSiO4 6.75 6.71 12.51 13.46
Microcline KAlSi3O8 13.31 13.23 23.83 26.61
Albite NaAlSi3O8 12.47 12.25 22.19 26.37
Anorthite CaAl2Si2O8 12.88 13.35 22.58 26.49
Almandine Fe3Al2Si3O12 21.68 20.07 33.89 23.93
Spessartine Mn3Al2Si3O12 21.29 20.02 35.60 32.20
Sillimanite Al2SiO5 7.05 7.36 12.54 15.22
Beryl Be3Al2Si6O18 25.28 26.32 43.19 53.02
Sphene CaTiSiO5 10.96 10.42 17.22 19.18
Wollastonite CaSiO3 5.83 5.62 10.04 11.35
The Mott-Littleton approach has various degrees of approximation (up to 4). In the
zero-order approximation, polarization is regarded as the result of a series of induced
dipoles at the various lattice sites (cf. figure 1.26). In I-, II-, III-, and IV-order approxima-
tions, Mott and Littleton performed precise calculations, respectively, for the I, II, III,
and IV series of cells surrounding the origin, treating the remaining part of the crystal in
the same manner as the zero-order approximation (see Rittner et al., 1949a,b for a detailed
account of the method). Adopting the zero-order Mott-Littleton approximation, Lasaga
(1980) proposed polarization calculations for silicates. For the Mg2SiO4 component, the
system shown in equations 1.160 to 1.163 is rewritten as follows:
1
EP = − eΦe (1.164)
2
Vcell  1 1 1
Φe = −e  M Mg ∑ r 4 + M Si ∑ 4
+ MO ∑ 4  (1.165)
4  Mg Si r O r 
α iT 1  1
Mi =  1− T . (1.166)
2α Mg
T
+ α Si + 4α O 4π 
T T
ε0 
The ␣T terms in equation 1.166 represent total ionic polarizability, composed of electronic
polarizability ␣e plus an additional factor ␣d, defined as a “displacement term,” due to the
fact that the charges are not influenced by an oscillating electric field (as in the case of
experimental optical measurements) but are in a “static field” (Lasaga, 1980):
αT = α e + α d . (1.167)
Based on this concept, the dielectric constant is modified by application of the Clausius-
Mosotti relation:
ε 0T − 1 π
= ( 2α Mg
T
+ α TSi + 4α OT ). (1.168)
ε 0 + 2 3Vcell
T
Table 1.22 lists the values of optical polarizability ␣e and static polarizability
␣T, calculated by Lasaga and Cygan (1982) for various silicates on the basis of
the additivity principle
α Mg
T
= 2α MgO
T
+ α TSiO . (1.169)
2 SiO 4 2
Stemming from a comparison between calculated and experimental methods, the

proposed calculations appear quite precise and thus open up new perspectives in
polarization calculations for natural crystalline phases. As we will see in chapter
4, polarization energy is of fundamental importance in the evaluation of defect
equilibria and consequent properties.
CHAPTER TWO
Concepts of Chemical
Thermodynamics
2.1 General Concepts and Definitions
In geology it is customary to consider systems in which the intensive variables

pressure (P) and temperature (T ) are characteristic of the ambient and, therefore,
are prefixed and constant. In these conditions, the Gibbs free energy of the system
(G ) is at minimum at equilibrium. The treatments presented in this chapter are
based on this fundamental principle. Let us first introduce in an elementary fash-
ion some fundamental definitions.
SYSTEM – “Whatever part of the real world which is the subject of a thermody-
namic discussion is generally defined system” (Lewis and Randall, 1970). As far
as we are concerned, a system may be a group of atoms, or minerals, or rocks.
The limit of the system can be defined in any way one wishes (for instance, an
outcrop, a hand specimen, or a thin section). We generally define a system in
such a way that all phases within it are in thermodynamic equilibrium. The modi-
fications that take place in a system may or may not imply interactions with mat-
ter external to the system itself, so we consider a system that does not exchange
matter or energy with the exterior as “isolated,” a system that is able to exchange
energy but not matter with the exterior as “closed,” and a system that exchanges
both matter and energy as “open.” A system is also called “heterogeneous” if it
is composed of several phases.
PHASE – A phase is defined as a fraction (region) of a system that has peculiar
and distinguishable chemical and physical properties. For instance, a “magmatic”
system may be composed of mineral solid phases, fluids (H2O, CO2 , etc.), and
melts.
COMPONENT – Each phase in a system is in turn composed of chemical compo-
nents. The choice of the various components is, under certain provisos, arbitrary.
For instance, the olivine solid mixture (Mg, Fe)2SiO4 may be considered to be
composed of fayalite (Fe2SiO4 ) and forsterite (Mg2SiO4 ) end-members, of oxides
/ 91
(MgO; FeO; SiO2 ), or of elements (Mg; Fe; O2 ). As we will see in section 2.6, the
choice of the various components is no longer arbitrary whenever the “phase
rule” is applied.
EQUILIBRIUM – This term is used to indicate a condition in which the various
components are subjected to dynamic and reversible exchanges among various
phases within the heterogeneous system. Stable equilibrium, dictated by the prin-
ciple of minimization of Gibbs free energy, must be distinguished from metastable
equilibrium or apparent equilibrium conditions, which are often encountered in
geological systems. The classic example of metastable equilibrium is the secretion
of carbonate shells by marine organisms, composed of the denser high-P stable
aragonite phase of the CaCO3 component (see table 2.1 and figure 2.6). In the P-T
conditions of synthesis, the calcite polymorph, which has lower Gibbs free energy,
should be formed instead of aragonite. Another example of metastable equilib-
rium is silica solubility in water: at room P-T conditions the stable phase is quartz;
however, because this phase does not nucleate easily, metastable equilibrium be-
tween amorphous silica and a saturated solution is observed. There is apparent
equilibrium in a system whenever two or more of its phases do not apparently
interact to any measurable extent and it is impossible to determine whether stable
equilibrium exists. The local equilibrium in a system represents a zone of complete
reversibility of exchanges (or “zero affinity”; see section 2.12) restricted to a re-
gion sufficiently large for compositional fluctuations to be negligible.
SOLUTION and MIXTURE – There is some confusion between these two terms
in geological literature. According to the I.U.P.A.C. (International Union for Pure
and Applied Chemistry), the term mixture must be adopted whenever “all compo-
nents are treated in the same manner”, whereas solution is reserved for cases in
which it is necessary to distinguish a “solute” from a “solvent.” This distinction
in terminology will be more evident after the introduction of the concept of “stan-
dard state.” It is nevertheless already evident that we cannot treat an aqueous
solution of NaCl as a mixture, because the solute (NaCl) in its stable (crystalline)
state has a completely different aggregation state from that of the solvent (H2O)
and, because NaCl is a strong electrolyte (see section 8.2), we cannot even imag-
ine pure aqueous NaCl.
2.2 Gibbs Free Energy and Chemical Potential
To understand fully the chemical reaction processes that take place in rock assem-
blages, it is necessary to introduce the concept of chemical potential. Much the
same as in a gravitational potential, in which an object tends to fall from a high
to a low altitude, in a chemical potential field the reaction or “flow direction of
components” always tends to proceed from a high to a low chemical potential
region.
Concepts of Chemical Thermodynamics / 93
The Gibbs free energy of a system (Gtotal ) is composed of all the Gibbs free
energies of the various phases in the system, defined by the chemical potentials
of their components. Algebraically, this is expressed as
G total = ∑ µi n i , (2.1)
i
where ni is the number of moles of the ith component, ␮i is its chemical potential,
and summation is extended to all components.
For a system at constant T and P,
 ∂ G total 
  = µi . (2.2)
 ∂ n i  P ,T
As already stated, chemical reactions proceed spontaneously in the direction of

lowering of chemical potentials. Based on equation 2.1, minimization of chemical
potentials corresponds to minimization of the Gibbs free energy of the system.
2.3 Gibbs Free Energy of a Phase as a Function of

the Chemical Potentials of Its Components
The Gibbs free energy of a phase is a measurable property that depends on pres-
sure, temperature, structure, and composition. Because the chemical potential is
the partial derivative of the Gibbs free energy with respect to composition, it
graphically represents the tangent to the Gibbs free energy curve at any given
position of the compositional field. For instance, figure 2.1 shows the chemical
Figure 2.1 Relationships between chemical potentials of two components and Gibbs free
energy of a binary phase.
potential of component 1 in a solid mixture ␣ composed of two components (1

and 2) and analytically defined by
∂ Gα
µ 1,α = . (2.3)
∂ n1
Analogously, for component 2,
∂ Gα
µ 2 ,α = . (2.4)
∂n2
Reasoning in terms of molar Gibbs free energy (G, the energy of one mole of the
given phase) and considering a homogeneous system composed of one phase with
two components, because
X 1 + X 2 = 1, (2.5)
where X1 and X2 are the mole fractions of the components in the system, we have
dG α µ 1,α − G α dG α µ 2 ,α − G α
= and = , (2.6)
dX 1 1 − X1 dX 2 1 − X2
from which we obtain
µ1,α = Gα + 1 − X 1( ) dG
dX
α
. (2.7)
1
and
dGα
µ 2 ,α = Gα + X 1 .
(
d 1 − X1 ) (2.8)
It is also evident from figure 2.1 that the molar Gibbs free energy of a pure phase
composed of a single component is equivalent to the chemical potential of the
component itself.
Let us now consider two coexisting phases ␣ and ␤, each having components
1 and 2 and the Gibbs free energy relationships outlined in figure 2.2. The follow-
ing four identities apply:
(
µ1,α = Gα + 1 − X 1,α ) dX
dG α
(2.9)
1, α
Figure 2.2 Relationships between chemical potential and Gibbs free energy in a two-
phase system with partial miscibility.
dGα
µ 2 ,α = Gα + X 1,α
(
d 1 − X 1,α ) (2.10)
dG β
(
µ1, β = G β + 1 − X 1, β ) dX (2.11)
1, β
dG β
µ 2 , β = G β + X 1, β .
(
d 1 − X 1, β ) (2.12)
It is clear from figure 2.2 that the chemical potentials of the two phases at equilib-
rium are equal:
µ1,α = µ1, β (2.13)
µ 2 ,α = µ 2 , β . (2.14)
We thus write
dG β
(
Gα + 1 − X 1,α ) dX
dGα
(
= G β + 1 − X 1, β ) dX (2.15)
1, α 1, β
and
dGα dG β
Gα + X 1,α = G β + X 1, β .
(
d 1 − X 1,α ) (
d 1 − X 1, β ) (2.16)
Equations 2.15 and 2.16 are the general equilibrium relations for two coexisting
phases.
We can now examine phase stability in the compositional field: for amounts
of component 2 in ␣ higher than X␣ the tangent to the Gibbs free energy curve
of phase ␣ never intercepts the loop of phase ␤ (in other words, the potentials of
the two components in the two phases are never identical); only phase ␣ is stable
in such conditions. Analogously for X1 ⬎ X␤ , only phase ␤ is stable, whereas
stable coexistence of the two phases is the case for X␤ ⬍ X1 ⬍ X␣ and X␣ ⬍ X2
⬍ X␤ . Although equations 2.15 and 2.16 imply that only two phases may form in
the binary system, this constraint is generally not valid, and the conformation of
the Gibbs free energy minimum curve in the G-X space becomes more complex
as the number of phases that may nucleate in the system increases. Although there
is no analytical solution to this problem, the graphical outline of the equilibrium
is quite simple. Let us consider, for instance, the isobaric-isothermal G-X plot
depicted in figure 2.3: four phases with partial miscibility of components may
form in the system (i.e., ␣, ␤, ␥, and ␦ ), but the principle of minimization of the
Gibbs free energy and the equality condition of potentials must be obeyed over
the entire compositional range.
The conformation of the G ⫽ f (X) curve that satisfies these conditions can
be determined by draping a rope under the G -X loops of the single phases and
pulling it up on the ends, as shown in figure 2.3A. The resulting minimum Gibbs
Figure 2.3 Conformation of the minimum Gibbs free energy curve in the binary compo-
sitional field (modified from Connolly, 1992).
free energy curve may be conceived as composed of two kinds of regions (figure
2.3B): linear regions, in which the curve spans the G -X loops of the single phases,
and nonlinear regions, in which the curve overlaps the G -X loop of a single phase.
Figure 2.3 highlights three important facts (Connolly, 1992):
1. Linear regions (constant slope, hence constant potentials) define the

composition interval over which a two-phase assemblage is stable.
2. Because the minimum Gibbs free energy curve of the system is never
convex, the chemical potential of any component will always increase
with the increase of its molar proportion in the system.
3. Because P and T are arbitrarily fixed and the conformations of the G -X
loops of the various phases are mutually independent, no more than two
phases may be stable in any part of the compositional space.
These arguments may be extended to an n-compositional space: the straight

paths will be replaced by planes in a three-component system, by hyperplanes in
a four-component system, and so on. Moreover, in an n-component system at
fixed P and T, no more than n phases may be stable at the same P-T-X condition,
and their compositions are uniquely determined at any combination of P, T,
and X.
2.4 Relationships Between Gibbs Free Energy

and Other Thermodynamic Magnitudes
The internal energy of a system is a “state function”—i.e., the variation of the

internal energy of the system when passing from state A to state B does not de-
pend on the reaction path but only on initial and final state conditions (cf. appen-
dix 2).
For a crystalline phase, we have already seen that internal energy corresponds
to lattice energy at the zero point. At higher T, the internal energy of the phase
increases as a result of the increase in vibrational motion of all the atoms in
the lattice.
We have also seen (in chapter 1) that enthalpy and lattice energy are related
through the Born-Haber-Fayans thermochemical cycle, on the basis of the energy
additivity principle of Hess. The enthalpy or “heat content” of a phase (H ) is
composed of the internal energy U at the T of interest and the PV product:
H = U + PV . (2.17)
Equation 2.17 is general and can be extended to a heterogeneous system.

The Gibbs free energy of a phase (or a system) is composed of its enthalpy
minus the TS product:
G = H − TS. (2.18)
The concept of the entropy (S ) of a system is too often defined cryptically, for
the following reasons:
1. The definition is an operative concept.

2. The operative concept is essentially based on the relationship between
work and energy, and accounts for the reversibility of processes in ther-
modynamic cycles (in relation to the development of thermal engines in
the 19th century, or “steam age”: cf. Kittel, 1989).
The first satisfactory definition of entropy, which is quite recent, is that of

Kittel (1989): “entropy is the natural logarithm of the quantum states accessible
to a system.” As we will see , this definition is easily understood in light of Boltz-
mann’s relation between configurational entropy and permutability. The defini-
tion is clearly “nonoperative” (because the number of quantum states accessible
to a system cannot be calculated). Nevertheless, the entropy of a phase may be
experimentally measured with good precision (with a calorimeter, for instance),
and we do not need any “operative” definition. Kittel’s definition has the merit
to having put an end to all sorts of nebulous definitions that confused causes with
effects. The fundamental P-V-T relation between state functions in a closed sys-
tem is represented by the exact differential (cf. appendix 2)
dU = T dS − P dV . (2.19)
Differentiating equation 2.17, we obtain
dH = dU + P dV + V dP , (2.20)
which, combined with equation 2.19, gives
dH = T dS + V dP. (2.21)
Differentiating equation 2.18 and combining it with equation 2.21, we also obtain
dG = − S dT + V dP. (2.22)
The differential equations above are valid for reversible processes taking place in
closed systems in which there is no flow of matter among the various phases.
The bulk differential of the Gibbs free energy for a compositionally variable
phase is
 ∂G   ∂G   ∂G 
dG =   dT +   dP +  dni ,
 ∂ T  P, n  ∂P  T, n  ∂ ni  T, P, n (2.23)
j j j
which gives
dG = − S dT + V dP + ∑ µ i dni .
(2.24)
i
In equation 2.24, contrary to equation 2.22, the Gibbs free energy of the phase is
a function not only of the intensive variables T and P, but also of composition.
Equation 2.24 is thus of more general validity and can also be used in open sys-
tems or whenever there is flow of components among the various phases in the
system. Like the exact differential dG, we can reexpress exact differentials dH and
dU as
 ∂H 
dH = T dS + V dP +   dni (2.25)
 ∂ ni  T, P, n
j
and
 ∂U 
dU = T dS − P dV +   dni . (2.26)
 ∂ ni  T, P, n
j
The “Gibbs equations” 2.24, 2.25, and 2.26 (Gibbs, 1906) are exact differentials,
composed of the summations of the partial derivatives over all the variables of
which derived magnitude is a function (cf. appendix 2).
The partial derivatives not explicit in equations 2.25 and 2.26 are
 ∂G 
  = −S (2.27)
 ∂ T  P, n
j
and
 ∂G 
  = V. (2.28)
 ∂P  T, n
j
The need to make explicit the compositional derivations in equations 2.25 and
2.26 is dictated by the fact that, in the Gibbs (1906) sense, all partial derivatives
over composition are called “chemical potentials” and are symbolized by ␮, re-
gardless of the kind of magnitude that undergoes partial derivation (i.e., not only
the partial derivative of Gibbs free energy G, as in equation 2.23, for instance, but
also ∂H/ ∂ ni and ∂U/ ∂ ni ). This fact gives rise to some confusion in several text-
books.
Deriving the G/T ratio in dT, we obtain

(
∂ G T )  =−
H
.
 ∂T  T2 (2.29)
 P,n j
Moreover, if we derive molar enthalpy H (the enthalpy of one mole of the sub-
stance) with respect to T, we have
 ∂H 
  = CP , (2.30)
 ∂T  P , n
j
where CP is the specific heat of the substance at constant P—i.e., “the amount
of heat that is needed to increase by 1 kelvin one mole of the substance at the P
and T of interest.”
2.5 Partial Molar Properties
We have defined the “chemical potential of a component” as the partial derivative

of the Gibbs free energy of the system (or, for a homogeneous system, of the
phase) with respect to the number of moles of the component at constant P and
T—i.e.,
 ∂G 
µi =   . (2.31)
 ∂ n i  P ,T
As we have already seen, chemical potential varies with phase composition. If we

derive this chemical potential with respect to T and P at constant composition,
we obtain
 ∂µ i   ∂S 
  = −  = − si (2.32)
 ∂ T  P, n  ∂ ni  P , T
i
 ∂µ i   ∂V 
  =  = vi (2.33)
 ∂P  T , n  ∂ ni  P ,T
i

(
 ∂ µi T )  =−
hi
,
 ∂T  T2 (2.34)
  P , ni
where magnitudes si , vi , and hi are defined as the “partial molar properties” of the
ith component in the phase. These magnitudes should not be confused with molar
properties S, V, and H , with which they nevertheless coincide in the condition of
a pure component in a pure phase (we have also seen that, in the same condition
␮i ⬅ G).
“Partial molar properties,” like molar properties, are intensive magnitudes
(i.e., they do not depend on the extent of the system), and the same relationships
that are valid for molar properties hold for them as well. For instance,
µ i = h i − Tsi . (2.35)
As we will see , partial molar properties are of general application in the thermo-
dynamics of mixtures and solutions.
2.6 Gibbs Phase Rule
The phase rule formulated by Gibbs establishes the degree of freedom (variance)
as a function of the number of phases and of the numbers of components and of
intensive (P, T ) and condition variables (gravitational field, magnetic field, elec-
tric field, etc.). This rule is intuitively assimilated to the degree of determination
of linear systems in elementary algebra: the variance of a system of variables
correlated by linear equations is equivalent to the number of independent vari-
ables less the number of correlation equations. If we consider an algebraic system
with two variables (x and y) that are not mutually dependent (i.e., there are no
correlation equations linking the two variables), the variance of the system is 2;
if y ⫽ f (x), the variance is 1 because a fixed value of y corresponds to a given
value of x. If there are two correlation equations between x and y, the variance is
0 (the system is completely determined).
Let us now consider a heterogeneous thermodynamic system at equilibrium.
If there are ⌽ phases in the system, it can easily be seen that ⌽ ⫺ 1 equations of
type 2.15 and 2.16 apply for each component in the system. Hence, if there are n
components, the number of equations will be n(⌽ ⫺ 1). Moreover, the following
mass-balance equation holds for each phase:
∑ X i ,α = 1,
(2.36)
i
where Xi,␣ is the molar fraction of the ith component in phase ␣.

The number of condition equations is therefore
( )
n Φ − 1 + Φ. (2.37)
The system variables are composed of n ⌽ compositional terms plus ambient vari-
ables that are usually two in number: temperature and pressure (hydrostatic and/
or lithostatic-isotropic pressure). The variance (V ) of the system is readily ob-
tained by subtracting the number of condition equations from the total number
of variables (n ⌽ ⫹ 2):
( ) [( )
V = n Φ + 2 − n Φ − 1 + Φ = n − Φ + 2. ] (2.38)
When applying the Gibbs phase rule, it must be remembered that the choice of
components is not arbitrary: the number of components is the minimum number
compatible with the compositional limits of the system.
Consider, for instance, the polymorphic reaction
CaCO3 ⇔ CaCO3 .
calcite aragonite (2.39)
There are two phases in reaction and, apparently, three components, correspond-
ing to atoms Ca, C, and O. However, the compositional limits of the system are
such that
Ca = C (2.40)
and
O = 3C. (2.41)
Because conditions 2.40 and 2.41 hold for all phases in the system, the minimum
number of components necessary to describe phase chemistry is 1.
Consider again the exchange reaction
2 MgSiO3 + Fe2SiO4 ⇔ 2 FeSiO3 + Mg2SiO4 .

pyroxene olivine pyroxene olivine (2.42)
Because in all phases
O = Mg + Fe + 2 (2.43)
and
Si = 1, (2.44)
there are only two independently variable components (Fe and Mg).
2.7 Stability of Polymorphs
Polymorphism occurs whenever a given component exists under different aggre-

gation states as a function of P and T. The stable state requires that the chemical
potential of the component (hence, the Gibbs free energy of the phase for a phase
composed of a single component) be at minimum at equilibrium. Figure 2.4
shows examples of G-T plots for Al2SiO5 in various P conditions.
Figure 2.4 G-T plots for Al2SiO5 in various P conditions (schematic).

Al2SiO5 occurs in nature in the solid state in three different aggregation forms:
kyanite, which is triclinic, dense, and stable at high P, and sillimanite and andalu-
site, which are orthorhombic, less dense, and stable at lower P (see table 2.1). If
we examine the G-T diagram for P ⫽ 10 kbar in figure 2.4, we note that the
kyanite polymorph has the lowest Gibbs free energy over a wide T range (T ⬍
780 °C). At T ⬎ 780 °C, sillimanite has the lowest energy. At P ⫽ 2 kbar, the T
range in which kyanite has the lowest energy is restricted to T ⬍ 320 °C, andalu-
site has the lowest energy for 320 °C ⬍ T ⬍ 600 °C, and sillimanite has the lowest
energy at T ⬎ 600 °C. At P ⫽ 3.75 kbar, we see that the Gibbs free energy is equal
for all three polymorphs at T ⬵ 500 °C (504 ⫾ 20 °C according to Holdaway and
Mukhopadhyay, 1993).
By combining the various observations obtained from the G-T diagrams in
different P conditions, we can build up a P-T diagram plotting the stability fields
of the various polymorphs, as shown in figure 2.5. The solid dots in figures 2.4
and 2.5 mark the phase transition limits and the triple point, and conform to the
experimental results of Richardson et al. (1969) (A, R, B⬘, C⬘) and Holdaway
(1971) (A, H, B, C). The dashed zone defines the uncertainty field in the
Table 2.1 Examples of polymorphism in geology.
Density,
Compound Phase g/cm3 Symmetry Elemental Coordination
SiO2 Trydimite 2.27 Hexagonal Si(IV)

Cristobalite 2.35 Cubic Si(IV)
Quartz 2.654 Trigonal Si(IV)
Cohesite 3.01 Monoclinic Si(IV)
Stishovite 4.28 Tetragonal Si(VI)
CaCO3 Calcite 2.715 Trigonal C(III) Ca(VI)
Aragonite 2.94 Orthorhombic C(III) Ca(IV)
Al2SiO5 Andalusite 3.15 Orthorhombic Al(IV) Al(V) Si(IV)
Sillimanite 3.25 Orthorhombic Al(IV) Al(IV) Si(IV)
Kyanite 3.6 Triclinic Al(IV) Al(VI) Si(IV)
C Graphite 2.25 Hexagonal C(III)
Diamond 3.51 Cubic C(IV)
Mg2SiO4 Forsterite 3.23 Orthorhombic Mg(VI) Mg(VI) Si(IV)
Spinel 3.58 Cubic Mg(IV) Mg(VI) Si(IV)
MgSiO3 Clinoenstatite 3.19 Monoclinic Mg(VI) Si(IV)
Enstatite 3.20 Orthorhombic Mg(VI) Si(IV)
Al2O3H2O Bohemite 3.02 Orthorhombic Al(VI) H(II)
Diaspore 3.4 Orthorhombic Al(VI) H(II)
ZnS Wurtzite 4.0 Hexagonal Zn(IV)
Sphalerite 4.09 Cubic Zn(IV)
HgS Metacinnabar 7.60 Trigonal Hg(IV)
Cinnabar 8.18 Cubic Hg(VI)
Figure 2.5 Polymorphs of Al2SiO5. Dashed area covers uncertainty range among the
various experiments. Marked dots B, B⬘, C, C⬘, and H are those plotted in figure 2.4.
andalusite-kyanite and kyanite-sillimanite transitions, arising from the compari-

son of the two sets of data. In figure 2.4 (dashed area), the uncertainty is regarded
as an error progression involving the value of the Gibbs free energy of sillimanite
with increasing T. In fact, as discussed by Newton (1987), the bias must be im-
puted to the formation of metastable fibrolite in the andalusite-sillimanite transi-
tion in the experimental runs of Richardson et al. (1969), and to difficulty in
reaching equilibrium for the kyanite-sillimanite transition.*
If we now recall the phase rule, it is evident that, at the P-T conditions repre-
sented by point D in figure 2.5, slight variations in the P or T values will not
induce any change in the structural state of the phase (there are one phase and
one component: the variance is 2). At point A in the same figure, any change in
one of the two intensive variables will induce a phase transition. To maintain the
coexistence of kyanite and andalusite, a dP increment consistent with the slope
of the univariant curve (there are two phases and one component: the variance
* A recent study by Holdaway and Mukhopadhyay (1993) essentially confirms the stability
diagram of Holdaway (1971). However, it is of interest to show how even slight errors in the assigned
Gibbs free energy of a phase drastically affect the stability fields of polymorphs.
is then 1) would correspond to each dT variation. At the triple point R (or H),
all three polymorphs coexist stably, but any minimal change in the P-T conditions
results in the disappearance of one or two phases (variance is 0).
Figure 2.6 shows P-T stability diagrams for several components exhibiting
polymorphism in geology (the Co2SiO4 orthosilicate, which is not a major constit-
uent of rock-forming minerals, is nevertheless emblematic of phase transitions
observed in the earth’s mantle; cf. section 5.2.3).
Stability fields conform qualitatively to the following general rules:
1. With increasing T, low-density structural forms are tendentially more

stable.
2. Increasing P increases the stability of high-density polymorphs.
3. High-T phase transitions frequently proceed in the direction of increas-
ing crystalline symmetry.
The significance of the first two rules will be more evident in sections 2.10 and
3.7, when we discuss calculation of the Gibbs free energy of a phase at various
P-T conditions.
Figure 2.6 Selected cases of polymorphism in geology (simplified).

2.8 Solid State Phase Transitions
We saw in the preceding section that, at solid state in nature, a given component
may assume various structural states as a function of P and T variables. In given
P-T conditions, there is a given stable form. If one of the intensive variables is
changed, at a certain moment the univariant equilibrium of phase transition is
reached, and we then observe a solid state transformation leading to the new
aggregation state that is stable in the new P-T conditions. The Gibbs free energy
modification connected with the solid state phase transition is zero at equilibrium,
but the partial derivatives of the Gibbs free energy with respect to the P and T
variables are not zero.
We now distinguish solid state transformations as “first-order transitions” or
“lambda transitions.” The latter class groups all high-order solid state transforma-
tions (second-, third-, and fourth-order transformations; see Denbigh, 1971 for
exhaustive treatment). We define first-order transitions as all solid state transfor-
mations that involve discontinuities in enthalpy, entropy, volume, heat capacity,
compressibility, and thermal expansion at the transition point. These transitions
require substantial modifications in atomic bonding. An example of first-order
transition is the solid state transformation (see also figure 2.6)
C ⇔ C.
graphite diamond (2.45)
This reaction requires breaking of at least one bond and modification of the coor-
dination state of carbon (cf. table 2.1). During phase transformation, an interme-
diate configuration with high static potential is reached. In order to reach this
condition it is necessary to furnish the substance with a great amount of energy
(mainly in thermal form). First-order transitions, as a result of this energy re-
quirement, are difficult to complete, and metastable persistence of the high-energy
polymorph may often be observed. Diamond, for instance, at room P-T condi-
tions is a metastable form of carbon; aragonite, as already stated, at room P-T
conditions is a metastable form of CaCO3 , and so on.
In lambda transitions, no discontinuity in enthalpy or entropy as a function
of T and/or P at the transition zone is observed. However, heat capacity, thermal
expansion, and compressibility show typical perturbations in the lambda zone,
and T (or P) dependencies before and after transition are very different.
Alpha-beta ( ␣-␤ ) transitions of the condensed forms of SiO2 quartz, try-
dimite, and cristobalite may all be regarded as lambda transformations. Their ki-
netics are higher than those of quartz-trydimite, quartz-cristobalite, and
quartz-coesite, which are first-order transformations. Figure 2.7 plots in detail the
evolution of enthalpy, entropy, heat capacity, and volume at the transition zone
Figure 2.7 Enthalpy, entropy, heat capacity, and volume modification in ␣-quartz/
␤-quartz transition region. Reprinted from H. C. Helgeson, J. Delany, and D. K. Bird,
American Journal of Science, 278A, 1–229, with permission.
between ␣-quartz and ␤-quartz. The lambda point at 1 bar for this transition
occurs at T ⫽ 848 K. (The ␣-quartz/ ␤-quartz transition in fact implies minimal
variations in enthalpy and entropy at the transition point and is therefore inter-
preted by many authors as an overlapping first-order-plus-lambda transition. We
will discuss this transition more extensively in section 5.8.)
As we have already stated, the Gibbs free energy modification connected with
a polymorphic transition is zero. The univariant equilibrium along the transition
curve is described by the Clapeyron equation:
dP ∆S transition
0
= = K transition , (2.46)
dT ∆Vtransition
0
where ⌬ S 0transition and ⌬ V 0transition are, respectively, the standard molar entropy and
standard molar volume of transition, and K transition is the Clapeyron slope of the
solid-solid transition. For a first-order transition at P ⬍ 10 kbar, K transition is al-

most constant at all temperatures. With several substitutions from equation 2.46,
we derive
 ∂ ∆V transition
0   ∂ ∆V transition
0 
∆Cp transition
0
= 2 K transition T   + K 2
transition T   . (2.47)
 ∂T P  ∂P T
Equation 2.47 describes the interdependence of thermal expansion, compressibil-

ity, and heat capacity of a first-order transition and furnishes a precise tool for
the evaluation of the internal consistency of experimental data in solid state tran-
sition studies (see Helgeson et al., 1978 for a careful application of eq. 2.47).
2.8.1 Phase Transitions and Landau Theory

The application of Landau theory to rock-forming minerals has been promoted
by Ekhard Salje and his coworkers in an attempt to achieve better quantification
of complex transition phenomena (mainly in feldspars, but also in pyroxenes and
spinels; Salje et al., 1985; Salje, 1985, 1988; Carpenter and Salje, 1994a,b; Carpen-
ter, 1988).
If “excess specific heat,” defined as “that part of the specific heat that is due
to the structural phase transition” (cf. Salje, 1988), is denoted by ⌬CP,excess,trans ,
the excess entropy associated with transition within the upper and lower limits of
CP perturbation is given by
Tu
∆C P,excess,trans
∆S excess,trans = ∫ T
dT , (2.48)
Tl
and the excess enthalpy is given by
Tu
∆H excess,trans = ∫ ∆C P,excess,trans dT . (2.49)

Tl
The excess thermodynamic properties correlated with phase transitions are

conveniently described in terms of a macroscopic order parameter Q. Formal
relations between Q and the excess thermodynamic properties associated with a
transition are conveniently derived by expanding the Gibbs free energy of transi-
tion in terms of a Landau potential:
( )
2
∆G excess,trans,T ,P , µ ,Q = G ∇Q − HQ + AQ 2 + BQ 4 + CQ 6 , (2.50)
(Landau and Lifshitz, 1980), where H is the conjugate field of the order parame-
ter, coefficient A is a function of temperature:
(
A = a T − Tc , ) (2.51)
and the remaining coefficients are not explicitly dependent on T. Tc in equation

2.51 is the critical temperature (e.g., the Curie temperature in the case of mag-
netic transitions).
In the case where transitions involve low-symmetry phases, general equation
2.50 may be recast into the unidimensional form
∆G excess,trans =
1
2
( 1
) 1
a T − Tc Q 2 + BQ 4 + CQ 6
4 6
(2.52)
(Carpenter, 1988). Because the excess Gibbs free energy of transition must always
be at a minimum with respect to the macroscopic order parameter Q, i.e.:
 ∂ ∆G excess,trans 
  = 0, (2.53)
 ∂Q  T ,P
the various types of transitions are evaluated on the basis of condition 2.53 by
setting Q ⫽ 0 at 0 K. The type of transition depends essentially on the signs of
B and C.
For a second-order transition, B ⬎ 0 and C ⫽ 0, and condition 2.53 leads to
a (T − Tc )Q2 + BQ 4
1 1
∆Gexcess,trans = (2.54)
2 4
∂ ∆Gexcess,trans 1
∆Sexcess,trans = − = − aQ2 (2.55)
∂T 2
1 1
∆H excess,trans = ∆Gexcess,trans − T ∆Sexcess,trans = − aTcQ2 + BQ 4 (2.56)
2 4
a
∆C P, excess,trans = T, T ≤ Tc (2.57)
2B
B
Tc = . (2.58)
a
The equilibrium temperature of transition is coincident with the critical temperature:
Ttrans = Tc . (2.59)
For a tricritical transition (i.e., a high-order transition crossing a first-order phase

boundary), B ⫽ 0 and C ⬎ 0, and condition 2.53 leads to
a (T − Tc )Q2 + CQ6
1 1
2 6
1
∆Sexcess,trans = − aQ2 (2.61)
2
1 1
∆H excess,trans = − aTcQ2 + CQ6 (2.62)
2 6
aT
(T − T)
−1 2
∆C P, excess,trans = , T < Tc
4 Tc
c (2.63)
C
Tc = (2.64)
a
Ttrans = Tc . (2.65)
For a first-order transition, B ⬍ 0 and C ⬎ 0, and condition 2.53 leads to
a (T − Tc )Q2 + BQ 4 + CQ6
1 1 1
2 4 6
1   3  T − Tc   
∆Sexcess,trans = − aQ2 1 + 1 −  
4  Ttrans − Tc    (2.67)
 
3

1 1 1
∆H excess,trans = − aTcQ2 + BQ 4 + CQ6 (2.68)
2 4 6
3 B2
Ttrans = Tc + . (2.69)
16aC
The latent heat of the transformation L is
1
L= 2
aQtrans Ttrans, (2.70)
2
where Qtrans is the order parameter at Ttrans.
Figure 2.8 shows the behavior of the macroscopic order parameter Q as a

function of T/Tc (T/Ttrans for first-order transitions).
For a second-order transition,
12
 T
Q = 1 −  , T < Tc . (2.71)
 Tc 
For a tricritical transition,
14
 T
Q = 1 −  , T < Tc . (2.72)
 Tc 
For a first-order transition, the macroscopic order parameter shows a jump

at Ttrans from Q ⫽ 0 to Q ⫽ Qtrans —i.e., below Ttrans,
Figure 2.8 The order parameter Q as a function of T/Tc for second-order (a) and tricriti-
cal (b) transitions, and as a function of T/Ttrans for a first-order transition (c) with Tc ⫽
0.99Ttrans and Qtrans ⫽ 0.4. From Carpenter (1988). Reprinted by permission of Kluwer
Academic Publishers.
 
12 
2 2  3  T − Tc  
Q 2 = Q trans 1 + 1 −   , (2.73)
3   4  T trans − T c   
and at Ttrans,
12
 4 a(Ttrans − Tc ) 
Q trans = ±   . (2.74)
 b 
Curve c in figure 2.8, which is valid for a first-order transition, has been calcu-
lated by Carpenter (1988) for a relatively small difference between Ttrans and Tc
(Tc ⫽ 0.99Ttrans ) and a relatively high Qtrans (Qtrans ⫽ 0.4), resulting in a small
jump at Ttrans.
Obviously, if we know experimentally the behavior of the macroscopic order-
ing parameter with T, we may determine the corresponding coefficients of the
Landau expansion (eq. 2.52). However, things are not so easy when different
transitions are superimposed (such as, for instance, the displacive and order-
disorder transitions in feldspars). In these cases the Landau potential is a summa-
tion of terms corresponding to the different reactions plus a coupling factor asso-
ciated with the common elastic strain.
We will see detailed application of Landau theory to complex superimposed
transition phenomena when we treat the energetics of feldspars in chapter 5.
2.9 Standard State and Thermodynamic Activity
From what we have seen up to this point, we can perform “relative” evaluations
of the modification of Gibbs free energy with composition (section 2.2) and with
state conditions of the system (section 2.4). However, in order to calculate ther-
modynamic equilibrium, we need a rigorous energy scale—one that refers to a
firmly established “starting level” of the chemical potential. This level, or standard
state, which is kept constant as calculations develop, allows us to evaluate quanti-
tatively the change in chemical potential that takes place, for instance, in a solid
state transition such as those discussed in the preceding section. For the same
purpose, we introduce the concept of thermodynamic activity, which reflects the
difference in chemical potential for a given component in a given phase at fixed
P-T conditions with respect to chemical potential at standard state.
The algebraic formulation of the relationship among chemical potential ␮i ,
standard state chemical potential ␮ 0i , and thermodynamic activity ai is
µ i ,α = µ i0,α + RT ln a i ,α , (2.75)
where ␮i,␣ is the chemical potential of the ith component in phase ␣ and R is the
gas constant. The reversed expression for thermodynamic activity is therefore
 µ i ,α − µ i0,α 
ai ,α = exp  . (2.76)
 RT 
It must be emphasized that the choice of a particular standard state of refer-

ence has no influence on the result of equilibrium calculations and is only a matter
of convenience. It is generally convenient to adopt as standard state for solid
components the condition of “pure component in the pure phase at the P and T
of interest” or, alternatively, the condition of “pure component in the pure phase
at P ⫽ 1 bar and T ⫽ 298.15 K.”
2.9.1 Standard State of Pure Component in Pure Phase

at P and T of Interest
We can identify this reference condition by means of the notation
µ i0,α = µ i ,α ( X i = 1) . (2.77)
The relationship between the thermodynamic activity and molar fraction of a

given component i in mixture ␣ is described, over the whole compositional
range, by
ai , α = X i , α γ i , α , (2.78)
where ␥i,␣ is the rational activity coefficient, which is a function of P, T, and com-
position and embodies interaction effects among the component of interest and
all the other components in the mixture. It is obvious from equation 2.75 that, at
standard state, the activity of the component is equal to 1. If the adopted standard
state is that of “pure component,” we also have
X i , α = 1 ⇒ γ i , α = 1. (2.79)
Combining equation 2.78 with equation 2.75, we derive the general activity-
concentration relation that is valid over the entire compositional range:
µ i ,α = µ i0,α + RT ln X i ,α + RT ln γ i ,α . (2.80)
Based on equation 2.80, we can distinguish three different compositional fields:

Figure 2.9 Activity–molar fraction relationship for component 1 in mixture ␣.
1. Compositional range of validity of Raoult’s law

2. Compositional range of validity of Henry’s law
3. Intermediate field
These fields are shown in a semilogarithmic plot in figure 2.9, relative to compo-
nent 1 in mixture ␣.
Compositional Range of Validity of Raoult’s Law

This field is extended from the condition of “pure component” (X1,␣ ⫽ 1) to a
given concentration value, and is characterized by a straight-line correlation with
slope RT on the semilogarithmic plot:
µ1,α = µ10,α + RT ln X 1,α . (2.81)
Chemical interactions among component 1 and all other components in the mix-
ture are virtually absent, because the mixture is composed essentially of compo-
nent 1 itself ( ␥1,␣ ⫽ 1). If we compare equations 2.81 and 2.75, we see that, in
this compositional field, we have
α 1,α = X 1,α . (2.82)
Compositional Range of Validity of Henry’s Law

This field ranges from a given (generally low) value of the molar concentration
of component 1 down to infinite dilution (X1,␣ ⫽ 0). In this case, too, the relation-
ship between chemical potential and the natural logarithm of the molar fraction
is straight, with slope RT, but, unlike the previous case, it also involves an inter-
cept term:
µ1,α = µ10,α + RT ln X 1,α + RT ln h1,α , (2.83)
where h1,␣ is Henry’s constant for component 1 in mixture ␣ and depends on the
P, T, and composition of the mixture, but not on the amount of component 1 in
the mixture. In this case, the thermodynamic activity of component 1 is defined,
in terms of equation 2.78, by
γ 1,α = h1,α . (2.84)
Intermediate Field
Equation 2.78 is still valid, and the chemical potential of component 1 is defined
by equation 2.80, with ␥ varying with P, T, and composition (see figure 2.9).
2.9.2 Standard State of Infinite Dilution of Component

This standard state is adopted for solutions in which, as already stated, the condi-
tion of solute is distinguished from that of solvent. Moreover, it is (implicitly)
adopted in trace element geochemistry (see chapter 10).
If we embody solute-solvent interactions (observed at low solute concentra-
tions) in the standard state condition, i.e.:
µ1*,α = µ10,α + RT ln h1,α , (2.85)
in the field of Henry’s law we have
µ1,α = µ1*,α + RT ln X 1,α , (2.86)
where ␮ 1*,␣ is the chemical potential for Henry’s standard state. In this case, too,
as shown in figure 2.9, we observe in the field of Henry’s law a straight-line rela-
tion with slope RT between chemical potential and the natural logarithm of the
molar concentration, but the intercept is now ␮* 1 , ␣ . Again, we introduce an activ-
ity coefficient varying with X1,␣ for the intermediate concentration field, and we
have
µ1,α = µ1*,α + RT ln a1,α (2.87)
a1,α = X 1,α γ 1*,α , (2.88)

γ 1*,α = f (X 1,α ). (2.89)
We will see in chapter 8 that, in the case of aqueous solutions, it is convenient

to adopt the condition of “hypothetical 1-molal solution at P ⫽ 1 bar and T ⫽
298.15 K” as the standard state. Most experimental data on aqueous solutions
conform to this reference condition. In this case, the resulting activity coefficient
is defined as the “practical activity coefficient” and must not be confused with
the rational activity coefficient of general relation 2.80.
2.10 Calculation of Gibbs Free Energy of a Pure

Phase in Various P-T Conditions
From the differential equations (expressed in terms of molar properties)
 ∂H   ∂H 
dH =   dT +   dP (2.90)
 ∂T  P  ∂P  T
and
 ∂S   ∂S 
dS =   dT +   dP , (2.91)
 ∂T  P  ∂P  T
where
 ∂H   ∂V 
  = V −T  (2.92)
 ∂P  T  ∂T  P
and
 ∂S   ∂V 
  = −  , (2.93)
 ∂P  T  ∂T  P
recalling equation 2.30 and integrating, we obtain
  ∂V  
T P
H T, P = H T0 , P +
r r ∫ PC dT + ∫ V − T  ∂T   dP (2.94)
Tr Pr  P
and
T P
dT  ∂V 
ST, P = ST0 , P +
r r ∫ CP
T
− ∫   dP.
 ∂T  P (2.95)
Tr Pr
Because equation 2.18 holds at all P-T conditions, we have
G T, P = H T, P − TST, P . (2.96)
The molar Gibbs free energy of the phase at P and T of interest is obtained by
combining equations 2.94, 2.95, and 2.96 for known values of molar enthalpy
and molar entropy at Tr and Pr reference conditions (H 0Tr ,Pr and S 0Tr ,Pr respec-
tively).
2.11 Gibbs-Duhem Equation
The general form of the Gibbs-Duhem equation for an n-component system can
be expressed as
X 1 d ln a1 + X 2 d ln a2 + L + X n d ln a n = 0. (2.97)
For a binary system, we have
X 1 d ln a1 + X 2 d ln a2 = 0, (2.98)
which, by application of the properties of the exact differentials (see appendix 2),
may be expressed as
∂ ln a1 ∂ ln a2
X1 + X2 = 0. (2.99)
∂X 2 ∂X 2
Based on equation 2.99, the corresponding activity trend of component 2 may be

deduced, whenever the activity trend of component 1 in the mixture with the
increase in the molar fraction of component 2 is known.
An analogous expression holds for the activity coefficients:
∂ ln γ 1 ∂ ln γ 2
X1 + X2 = 0. (2.100)
∂X 2 ∂X 2
In integrated form, equation 2.100 becomes

X2
X 2 ∂ ln γ 2
ln γ 1 = − ∫ dX 2 , (2.101)
0
1 − X 2 ∂X 2
which, with integration by parts, may be transformed as shown by Wagner (1952):
X2
X2 ln γ 2
ln γ 1 = −
1 − X2
ln γ 2 + ∫ (1 − X 2 ) 2
dX 2 . (2.102)
0
Evaluation of the integral in equation 2.102 may be performed graphically or by

polynomial interpolation.
In the case of ternary or higher-order mixtures, solution of the Gibbs-Duhem
equation is again based on application of the properties of the exact differentials
(Lewis and Randall, 1970):
µ1 dn1 + µ 2 dn2 + L + µ n dn n = dG. (2.103)
The solution of the problem requires integration procedures along pseudo-binary

lines that result in the combination of integral forms of the type in equation 2.101.
In this context it is unnecessary to proceed further with detailed treatment, for
which reference may be made to Lewis and Randall (1970).
2.12 Local Equilibrium, Affinity to Equilibrium,

and Detailed Balancing
The “principle of local equilibrium in disequilibrium processes” (Prigogine, 1955)

establishes that every natural system contains a macroscopic region combining a
number of molecules sufficiently large for microscopic fluctuations to be negli-
gible, within which chemical exchanges are reversible and equilibrium is complete.
At local equilibrium, there is no flux of components among coexisting phases;
moreover, the spatial extent of the local equilibrium zone is time-dependent (this
fact renders the principle of local equilibrium a formidable tool in the investiga-
tion of geological systems). The degree of attainment of local equilibrium may
be evaluated by introducing the concept of thermodynamic affinity: because a
reaction proceeds in the sense of minimization of the Gibbs free energy of the
system, for a generic reaction j taking place at constant temperature and pressure,
thermodynamic affinity (Aj ) is defined as
 ∂G  Qj
A j = −  = −RT ln , (2.104)
 ∂ξ j  P ,T Kj
where ␰j is the progress variable for the jth reaction, Kj is the equilibrium constant,
and Qj is the observed activity product of the (macroscopic) reaction:
∏ ai , j
vi , j
Qj = , (2.105)
i
where ai, j is the activity of the ith term in the jth reaction, and vi, j is the stoichio-
metric coefficient of the ith term in the jth reaction (positive for products and
negative for reactants).
If the equilibrium condition of a system is perturbed by modifications in in-
tensive variables P and T, or by exchanges of energy or matter with the exterior,
the system will tendentially achieve a new Gibbs free energy minimum through
modification of the aggregation states in its various regions ( phases) or through
modification of phase compositions by flow of components. In a heterogeneous
system, several reactions will generally concur to restore equilibrium. The overall
chemical affinity to equilibrium of a heterogeneous system (Asystem ) is related to
the individual affinities of the various reactions through
Asystem = ∑ σ j Aj , (2.106)
j
where ␴j is the relative rate of the jth reaction with respect to the overall reaction
progress variable ( ␰ ) (Temkin, 1963):
dξ j
σj = (2.107)
dξ
Because, in the zone of local equilibrium,
Asystem = ∑ Aj = 0,
(2.108)
j
and because activities can always be translated into compositional variables (mol-
alities in fluids, or molar concentrations in condensed phases), we may represent
local equilibria as discrete points in a compositionally continuous trend of locally
reversible but overall irreversible exchanges (figure 2.10); i.e., for a given compo-
nent i, in terms of total number of moles ni ,
ξ =1
 ∂n i , j 
n i , end − n i , start = ∫  ∑j ∂ξ j  dξ .

(2.109)
ξ=0
As exemplified in figure 2.10, in the attainment of new equilibrium a single

reaction may proceed in opposite directions (i.e., forward and backward). In hy-
drolytic equilibria, the dissolution process is conventionally defined as moving in
Figure 2.10 Schematic representation of concept of local equilibrium in disequilibrium

processes. Reprinted from P. F. Sciuto and G. Ottonello, Geochimica et Cosmochimica
Acta, 59, 2207–2214, copyright 䉷 1995, with kind permission from Elsevier Science Ltd.,
The Boulevard, Langford Lane, Kidlington OX5 1 GB, UK.
a forward direction and the precipitation process as moving in a backward or

“reverse” direction—i.e., in the case of silica,
forward( k + ) →
SiO 2(solid) + 2 H 2 O → H 4 SiO 04 (aqueous) . (2.110)
← backward( k − )
Because at equilibrium there is no flow of components, the rate constants of

forward reaction (k⫹ ) and reverse reaction (k⫺ ) must counterbalance. Moreover,
because affinity to equilibrium is zero, activity product Q must be equivalent to
equilibrium constant K. For a generic reaction j at equilibrium, we have
k+
= K nj (2.111)
k−
where n is a positive coefficient related to the orders of forward and backward

reactions (see section 8.21.1) and to stoichiometry (see Dembigh, 1971, for an
appropriate treatment of the actual significance of n). Equation 2.111, which is
the fundamental link between thermodynamics and kinetics, establishes that local
equilibrium is not associated with cessation of processes but rather with a dy-
namic balance between flows moving in opposite directions.
CHAPTER THREE
Thermochemistry of Crystalline Solids
Systematic analysis of the mixing properties of crystalline compounds, coupled

with impressive experimental efforts in the last three decades, has produced a
qualitative leap forward in our comprehension of the chemical-physical processes
operating in the earth’s lithosphere. Excellent works of systematic revision and
theoretical framing (to which we refer for detailed treatment) have also appeared
(Thompson, 1969; Saxena, 1973; Helgeson et al., 1978; Ganguly and Saxena,
1987; Navrotsky, 1994; Anderson, 1995). In this chapter, only those concepts that
are operatively most significant in crystal energetics will be outlined, after an
elementary digression on the quantum-mechanic interpretation of entropy and
heat capacity. The quantum-mechanic interpretation of the vibrational motion of
atoms clarifies the real significance of the macroscopic thermodynamic properties
of crystalline substances and allows us to rationalize the plethora of spectroscopic
data for thermochemical interpretation. Interested readers will find adequate
treatments in specialized volumes such as that edited by Kieffer and Navrotsky
(1985).
3.1 Entropy, Heat Capacity, and Vibrational Motion

of Atoms in Crystals
Let us consider a diatomic molecule and assume that it behaves as a harmonic

oscillator with two masses, m1 and m2 , connected by an “ideal” (constant-force)
spring. At equilibrium, the two masses are at a distance X 0 ; by extending or com-
pressing the distance by an amount X, a force F will be generated between the
two masses, described by Hooke’s law (cf. equation 1.14):
F = −K F X. (3.1)
The indefinite integration in dX of equation 3.1 gives the potential energy of the
harmonic oscillator (cf. equation 1.16):
122 /
Thermochemistry of Crystalline Solids / 123
KF X 2
Φ( X ) = . (3.2)
2
Force constant KF thus represents the curvature of the potential with respect
to distance:
∂ 2 Φ( X )
KF = . (3.3)
∂X 2
Most diatomic molecules have a force constant in the range 102 to 103 N ⭈ m⫺1. A
common tool for the calculation of KF in diatomic molecules (often extended to couples
of atoms in polyatomic molecules) is “Badger’s rule”:
KF =
186
(N ⋅ m ) ,
−1
(X )
3 (3.4)
0 − dij
where dij is a constant relative to i and j atoms. The equation of motion for the harmonic
oscillator is, in the case of diatomic molecules,
d 2X
m* = − KF X , (3.5)
dt 2
where m* is the reduced mass (typically 10⫺26 to 10⫺27 kg)
m1m2
m* = . (3.6)
m1 + m2
The frequency of the harmonic oscillator (cycles per unit of time) is given by
1 KF
v0 = (3.7)
2π m*
and is typically about 10⫺12 to 10⫺14 s⫺1 (McMillan, 1985).
The Schrödinger equation, as we have already seen (section 1.17.1), can be

written in the general form
Hψ = Eψ , (3.8)
where H is the quantum-mechanic hamiltonian
h2
H =Φ− ∇2. (3.9)
8π 2 m
Introducing the potential of the harmonic oscillator (eq. 3.2) in the monodimen-
sional equivalent of equation 3.9 (i.e., the Schrödinger equation for one-
dimensional stationary states; see eq. 1.9), we obtain
d 2ψ ( X ) 8π 2 m *  KF X 2 
+  E − ψ ( X ) = 0.
2 
(3.10)
dX 2 h2 
The admitted energy levels correspond to the eigenvalues of equation 3.10 for
the various quantum numbers n (see section 1.1.2):
12
 K h2   1
E ( n) = F  n +  (3.11)
 4π 2 m *   2
and may be expressed as functions of the characteristic oscillation frequencies of

the harmonic oscillator (v 0 ) by substituting equation 3.7 in equation 3.11:
 1
E ( n ) = hv0  n +  . (3.12)
 2
We thus obtain a set of energy levels separated by an energy gap ⌬E ⫽ hv 0 and

bounded by the potential function 3.2 (figure 3.1).
Based on equation 3.12, the lowest vibrational mode (n ⫽ 0) also has a certain
amount of energy, known as zero-point energy:
1
E0 = hv0 . (3.13)
2
This energy contributes to the lattice energy of the substance at T ⫽ 0, in the

terms outlined in section 1.12.2.
The vibrational functions, corresponding to the energy eigenvalues of equation 3.11

or “vibrational eigenfunctions,” are given by
 y2 
ψ ( n ) = C( n )H( n, y ) exp −  , (3.14)
 X
where
 4π 2 m * K 
y= F
 X . (3.15)
 h 2
 
C(n) is the “normalization factor”:
( )
−1 2
C ( n ) = 2 n n! π (3.16)
and H(n,y) is the Hermite polynomial:
H ( 0, y) = 1
H (1 , y ) = 2 y
(3.17)
H ( 2 , y) = 4 y2 − 2
H( 3, y ) = 8 y 3 − 12 y.
The vibrational motion of atoms in diatomic molecules and, by extension, in

crystals cannot be fully assimilated to harmonic oscillators, because the potential
well is asymmetric with respect to X 0 . This asymmetry is due to the fact that
the short-range repulsive potential increases exponentially with the decrease of
interionic distances, while coulombic terms vary with 1/X (see, for instance, fig-
ures 1.13 and 3.2). To simulate adequately the asymmetry of the potential well,
empirical asymmetry terms such as the Morse potential are introduced:
[ ]
2
Φ ( X ) = De 1 − exp( − aX ) (3.18)
where De is the depth of the potential well and
2 m*
a = π v0 . (3.19)
De
The anharmonicity of atomic vibration in crystals may also be approximated

by adding higher-order terms to the potential energy function of equation 3.2:
KF K
Φ( X ) = X 2 + F X 3 + L. (3.20)
2 3
Figure 3.1 Energy levels and wave functions of harmonic oscillator. Heavy line:
bounding potential (3.2). Light solid lines: quantum-mechanic probability density distri-
butions for various quantum vibrational numbers (␺ 2(n), see section 1.16.1). Dashed lines:
classical probability distribution; maximum classical probability is observed in the zone
of inversion of motion where velocity is zero. From McMillan (1985). Reprinted with
permission of The Mineralogical Society of America.
The energy eigenvalues for equation 3.20 (McMillan, 1985) are
 1  1
2
 1
3 
E ( n) = hve  n +  − X e  n +  + Ye  n +  + L , (3.21)
 2  2  2 
where ve replaces v 0 in the presence of anharmonicity and Xe and Ye are anhar-

monicity constants.
Zero-point energy, in the presence of anharmonicity (eq. 3.20), is
1 X Y 
E 0 = hve  − e + e  (3.22)
2 4 8
and is lower with respect to the zero-point energy of the harmonic oscillator (Xe
and Ye are generally positive with 1 ⬎ ⬎ Xe ⬎ ⬎ Ye; cf. McMillan, 1985).
Figure 3.2 The experimental potential of H2 molecule (heavy line) compared with har-
monic fit (dashed line) and Morse potential (dotted line). From McMillan (1985). Re-
printed with permission of The Mineralogical Society of America.
Figure 3.2 shows the experimental potential well for the H2 molecule, com-
pared with the harmonic fit and the Morse potential. Horizontal lines represent
quantized energy levels. Note that, as vibrational quantum number n increases,
the energy gap between neighboring levels diminishes and the equilibrium dis-
tance increases, due to the anharmonicity of the potential well. The latter fact is
responsible for the thermal expansion of the substance.
At high T, when the spacing of vibrational energy levels is low with respect
to thermal energy, crystalline solids begin to show the classical behavior predicted
by kinetic theory, and the heat capacity of the substance at constant volume (CV )
approaches the theoretical limit imposed by free motion of all atoms along three
directions, in a compound with n moles of atoms per formula unit (limit of Dulong
and Petit):
CV = 3nR . (3.23)
At low T, thermal energy is not enough to ensure statistical occupancy of all

energy levels accessible to each atom, and the heat capacity (either CV or CP )
approaches zero as absolute zero is approached.
The relationship among heat capacity, entropy, and temperature in crystalline
solids may be understood on the basis of two fundamental concepts: the Boltz-
mann factor and the partition function (or “summation over the states,” from the
German term Zustandsumme). Consider a system in which energy levels E 0 , E1 ,
E2 , E3 , . . . , En are occupied and the components in the various excited states

may be recognized. The ratio of the atoms of a given component in excited state
Ei , (Ni ), with respect to all the other atoms of the same component at ground
state E 0 , (N v ), is given by the Boltzmann factor:
0
Ni  ∆E i 
= exp − , (3.24)
Nv 0
 kT 
where ⌬Ei is the energy gap between excited state Ei and E 0 . The partition func-
tion is the summation of the relative statistical population over all accessible
quantum states:
∞
 ∆E i 
Q= ∑ exp − kT 
 (3.25)
i =0
and total energy is given by
∞
U = ∑ Ni Ei . (3.26)
i =0
The constant-volume heat capacity of the substance (CV ) represents the vari-
ation of total energy U with T in the harmonic approximation (X is constant over
T), and the integration of CV over T gives the (harmonic) entropy of the sub-
stance:
T
 ∂U  dT
CV =  ;
 ∂T 
S( T ) = ∫ CV T
. (3.27)
0
If a single energy level exists in the ground state, then clearly the partition func-
tion tends toward 1 as T decreases toward zero (nondegenerate ground state).
However, because several (n) configurations may occur at ground state (degener-
ate ground state), the system has a nonzero entropy even at the zero point:
S 0 = R ln( n). (3.28)
The extension of the quantum-mechanic interpretation of the vibrational mo-

tion of atoms to a crystal lattice is obtained by extrapolating the properties of the
diatomic molecule. In this case there are 3N independent harmonic oscillators (N
is here the number of atoms in the primitive unit cell—e.g., fayalite has four
formula units Fe2SiO4 in the unit cell, which gives 3N ⫽ 3 ⫻ 4 ⫻ 7 ⫽ 84 vibra-

tional normal modes*), each with a characteristic vibrational frequency:
vt = λ1t 2 . (3.29)
The energy of the system is then established, in terms of characteristic frequencies

and of a set of vibrational quantum numbers for the various levels:
 1  1  1
E =  n1 +  hv1 +  n2 +  hv2 +  n3 +  hv3 + L. (3.30)
 2  2  2
The energy of the nondegenerate ground state is given by the summation of the
various oscillators over the characteristic frequency:
3N
1
E0 = h ∑ vt . (3.31)
2 t =1
Planck showed that the mean energy of a great number of oscillators, each with
a characteristic angular frequency ␻t ⫽ 2␲vt , is given by
បωt
E = ,
 ប ω t kT  (3.32)
e −1
 
where
h
ប= . (3.33)
2π
Einstein (1907) applied equation 3.32 (originally conceived by Planck for the
quantization of electromagnetic energy) to the quantization of particle energy,
describing the internal energy of a solid composed of N 0 atoms as
3N 0
U = .
(e បω t kT
−1 ) (3.34)
CV is then obtained from equation 3.34 by partial derivation in dT:
* The three translational and three rotational degrees of freedom become normal modes when
the “unit cell molecule” is embedded in the crystal.
 ∂U  3N 0 kX 2 e X
CV =   = = 3RE ( X )
 ∂T 
( ) (3.35)
2
eX − 1
where
hvt បωt
X = = (3.36)
kT kT
and
X 2 eX X2
E( X ) = = .
(e ) 2X
2
X
−1 4 sin   (3.37)
 2
Equation 3.37, known as the Einstein function, is tabulated for various X-values
(see, for instance, Kieffer, 1985). In the Einstein function, the characteristic fre-
quency ␻t (and the corresponding characteristic temperature ␪E ; see, for instance,
eq. 3.40) has an arbitrary value that optimizes equation 3.35 on the basis of high-
T experimental data. Extrapolation of equation 3.35 at low temperature results
in notable discrepancies from experimental values. These discrepancies found a
reasonable explanation after the studies of Debye (1912) and Born and Von Kar-
man (1913).
In the Debye model, the Brillouin zone (see section 3.3) is replaced by a
sphere of the same volume in the reciprocal space (cf. eq. 3.57):
( )
3
4 2π
πK max
3
= . (3.38)
3 VL
Maximum wave vector Kmax has maximum angular frequency ␻D :
ω D = VM K max , (3.39)
where V M is the mean velocity of acoustic waves in the crystal.

The characteristic frequency ␻D leads to the definition of a characteristic tem-
perature for the solid, known as the Debye temperature (whose nature is analo-
gous to that of ␪E ):
បωD
θD = . (3.40)
k
The Debye temperature of the solid defines the form of the vibrational spectrum
in the acoustic zone (low frequency) and is related to the molar volume of the
solid V and to the mean velocity of acoustic waves V M through
13
ប  6π 2 N 0 
θD =  VM ,
k  ZV 
(3.41)
where Z is the number of formula units per unit cell.

The heat capacity at constant volume in the Debye model is given by
3 θD T
T  eX θ 
C v = 9 nR   ∫ X 4 dX = 3 nRD  D  , (3.42)
 θD  (e ) T 
3
0 X
−1
where D( ␪D /T ) is a tabulated function known as Debye heat capacity (see table 3

in Kieffer, 1985).
3.2 Heat Capacity at Constant P

and Maier-Kelley Functions
As we have already seen, the heat capacity is the amount of heat that must be
furnished to raise the temperature of a given substance by 1 K at the T of interest.
If constant pressure is maintained during heat transfer, heat capacity is defined
as “heat capacity at constant P” (CP ). As already seen in chapter 2, CP is the
partial derivative of the enthalpy of the substance at constant P and com-
position—i.e.,
 ∂H 
CP =   . (3.43)
 ∂ T  P, n
If heat transfer takes place at constant volume, the magnitude is defined as “heat
capacity at constant volume” (CV ) and is equivalent, as we have seen, to the
partial derivative of the internal energy of the substance at constant volume and
composition:
 ∂U 
CV =   . (3.44)
 ∂T  V ,n
CP and CV are related through

α V2
C P = CV + TV , (3.45)
βV
where ␣V and ␤V are, respectively, the isobaric thermal expansion coefficient and
the isothermal compressibility coefficient of the substance (see also section 1.15).
We have seen in chapter 2 that the heat capacity at constant P is of fundamen-
tal importance in the calculation of the Gibbs free energy, performed by starting
from the standard state enthalpy and entropy values
T
H T ,Pr = H T0 ,P +
r r ∫ CP dT , (3.46)
Tr
T
dT
ST ,Pr = ST0 ,P +
r r ∫ CP T
, (3.47)
Tr
and
G T , Pr = H T , Pr − TST , Pr . (3.48)
With increasing T, CP, like CV, approaches a limit imposed by the Dulong and
Petit rule:
CV = 3nR (3.49)
and
α V2
C P = 3nR + TV , (3.50)
βV
where n is the number of atoms in the compound formula.

The CP value of crystalline substances is obtained directly from calorimetric
measures operated at various T. CP dependency on T is commonly represented
by polynomial expansions of the type:
C P = K 1 + K 2 T A + K 3T B + L + K nT N , (3.51)
where K1 , K2 , K3 , . . . , Kn are interpolation constants and A, B, . . . , N are fixed

exponents. Already in 1932, Maier and Kelley had proposed an interpolation of
the type
C P = K 1 + K 2 T + K 3T −2 . (3.52)
Since then, several other polynomial expansions have been proposed. Haas and
Fisher (1976), for instance, proposed an interpolation with five terms in the form
C P = K 1 + K 2 T + K 3T −2 + K 4T 2 + K 5T −1 2 . (3.53.1)
The thermodynamic tables of Robie et al. (1978) use the Haas-Fisher polynomial
(eq. 3.53.1). Helgeson et al. (1978) use the Maier-Kelley expansion, changing sign at the
third term:
C P = K1 + K2T − K 3T −2 . (3.53.2)
Barin and Knacke (1973) and Barin et al. (1977) adopt a four-term polynomial:
C P = K1 + K2T + K 3T −2 + K 4T 2 . (3.53.3)
More recently, Berman and Brown (1985) proposed an interpolation of the type
C P = K1 + K2T − 1 2 + K 3T −2 + K 4T −3 , (3.54)
Figure 3.3 Interpolation properties of Maier-Kelley functions, exemplified by compound

clinoenstatite. Experimental data from Robie et al. (1978). B-B ⫽ Berman-Brown polyno-
mial; F-S ⫽ Fei-Saxena polynomial. Dashed line: Dulong and Petit limit. Reprinted from
Y. Fei and S. K. Saxena, Geochimica et Cosmochimica Acta, 51, 251–254, copyright 䉷
1987, with kind permission from Elsevier Science Ltd., The Boulevard, Langford Lane,
Kidlington OX5 1GB, UK.
where K2 ⱕ 0 and K3 ⱕ 0. Polynomial 3.54 seems to obey the limits imposed by vibrational
theory better than the preceding forms (cf. eq. 3.50).
Even more recently, Fei and Saxena (1987) proposed the more complex functional
form
( ) ( )
CP = 3nR 1 + K 1T −1 + K 2 T −2 + K 3T −3 + A + BT + C P′ , (3.55)
where A and B are coefficients related, respectively, to thermal expansion and compress-
ibility, and C⬘P is the deviation from the theoretical Dulong and Petit limit due to anhar-
monicity, internal disorder, and electronic contributions. Figure 3.3 shows how the
Berman-Brown and Saxena-Fei equations conform with the experimental data of Robie
et al. (1978) for clinoenstatite (MgSiO3 ).
As has been repeatedly observed by various authors (see, for instance, Robie
et al., 1978; Haas et al., 1981; Robinson and Haas, 1983), Maier-Kelley and Haas-
Figure 3.4 Comparison of interpolation properties of Maier-Kelley (MK), Haas-Fisher

(HF), and Berman-Brown (BB) functions for low albite. Dashed line: Dulong and Petit
limit (eq. 3.50). From R. G. Berman and T. H. Brown, Heat capacity of minerals in the
system Na2O-K2O-CaO-MgO-FeO-Fe2O3-A12O3-SiO2-TiO2-H2O-CO2: representation,
estimation, and high temperature extrapolation, Contributions to Mineralogy and Petrol-
ogy, 89, 168–183, figure 3, copyright 䉷 1985 by Springer Verlag. Reprinted with the per-
mission of Springer-Verlag GmbH & Co. KG.
Fisher polynomials can be used only within the T range for which they were
created. Extrapolation beyond the T limits of validity normally implies substan-
tial error progression in high-T entropy and enthalpy calculations. For instance,
figure 3.4 compares Maier-Kelley, Haas-Fisher, and Berman-Brown polynomials
for low albite. As can be seen, the first two interpolants, if extended to high T,
definitely exceed the Dulong and Petit limit. The Berman-Brown interpolant also
passes this limit, but the bias is less dramatic.
3.3 Entropy and Heat Capacity from Vibrational

Spectra: Kieffer’s Model
It has already been stated that, theoretically, N atoms in a crystal have 3N possible
vibrational modes. Obviously, if we knew the energy associated with each vibra-
tional mode at all T and could sum the energy terms in the manner discussed in
section 3.1, we could define the internal energy of the crystal as a function of T,
CV could then be obtained by application of equation 3.27, and (harmonic) en-
tropy could also be derived by integration of CV in dT/T.
Although the number of atoms in a crystal is extremely high, we can imagine
the crystal as generated by a spatial reproduction of the asymmetric unit by means
of symmetry operations. The calculation can thus be restricted to a particular
portion of space, defined as the Brillouin zone (Brillouin, 1953).
If we consider a primitive Bravais lattice with cell edges defined by vectors a1 , a2 , and
a3 , the corresponding reciprocal lattice is defined by reciprocal vectors b1 , b2 , and b3
so that
b1 =
( 2π )a 2 × a3
; b2 =
( 2π )a 3 × a1
; b3 =
( 2π )a 1 × a2
. (3.56)
a1 ⋅ a2 × a 3 a2 ⋅ a 3 × a1 a 3 ⋅ a1 × a2
The volume of Brillouin zone VB corresponds to (2␲ )3 times the reciprocal of cell volume
VL commonly adopted in crystallography, and the origin of the coordinates is at the center
of the cell (and not at one of the corners):
( 2π )
3
VB = . (3.57)
VL
Wave vectors K( ␩) are defined in the reciprocal space according to
K ( η ) = η1 b 1 + η 2 b 2 + η 3 b 3 , (3.58)
where ␩1 , ␩2 , and ␩3 are integers (Kieffer, 1979a).

Figure 3.5 Phonon dispersion diagram for a complete unit cell with 3N degrees of free-
dom. From Kieffer (1985). Reprinted with permission of The Mineralogical Society of
America.
N is thus replaced by the number of atoms in the corresponding primitive

unit cell (N ; see section 3.1). We now (theoretically) have two possibilities:
1. We can compute from first principles all possible vibrational modes for
3N oscillators in the cell unit, solving the Schrödinger equation with ap-
propriate atomic (and/or molecular) wave functions.
2. We can deduce the vibrational modes of atoms from experimental data
on the interaction of radiation of various frequencies (photons and pho-
nons) with the crystalline matter.
The first approach is still computationally prohibitive because of the chemical-

structural complexity of the crystalline materials of our interest. The second type
of approach is also rather complex, however, and necessitates several operative as-
sumptions.
Figure 3.5 shows the path of the vibrational frequencies of the crystal along
a given direction in the reciprocal lattice. For each value of K there are 3N degrees
Figure 3.6 Diatomic chain with two force constants KF1 and KF2 and different atomic
masses (m1 ⬎ m2). Reprinted with permission from Kieffer (1979a), Review of Geophysics
and Space Physics, 17, 1–19, copyright 䉷 1979 by the American Geophysical Union.
of freedom, and the vectors are defined in the Brillouin zone from K ⫽ 0 to K ⫽
Kmax. In this phonon dispersion diagram, each vibrational mode, for a given value
of wave vector K, pertains to a branch represented by a curve in the figure. Three
acoustic branches (which approximate zero frequency at K ⫽ 0) and 3N ⫺ 3 opti-
cal branches (the three at the highest frequency being Einstein oscillators) can
be recognized.*
The nature of the wave vectors sketched in figure 3.5 is not dissimilar from
what can be deduced for a harmonic oscillator with two force constants K F1 and
K F2.
Consider a diatomic chain in which the atoms, of distinct masses m1 and m2 , are
positioned at distance a (figure 3.6). The repetition distance of the chain is 2a, and the
Brillouin zone falls between ⫺␲ /2a and ␲ /2a. If only the interactions between first neigh-
bors are significant, the equation of motion for atom r at position ␮ is given by
d 2 µr
Fr = KF1 ( µ r − 1 − µ r ) + K F2 ( µ r − 1 − µ r ) = m1 (3.59)
dt 2
(Kieffer, 1985) and, for the atom at position r ⫹ 1,
d 2 µ r +1
Fr + 1 = KF1 ( µ r − µ r + 1 ) + KF2 ( µ r + 2 − µ r + 1 ) = m2 . (3.60)
dt 2
Because we are in a single dimension, we can adopt a scalar notation (i.e., K instead of K).
* The motion of atoms in the lattice can be depicted as a wave propagation (phonon). By
“dispersion” we mean the variation in the wave frequency as reciprocal space is traversed. The propa-
gation of sound waves is similar to the translation of all atoms of the unit cell in the same direction;
hence the set of translational modes is commonly defined as an “acoustic branch.” The remaining
vibrational modes are defined as “optical branches,” because they are capable of interaction with
light (see McMillan, 1985, and Tossell and Vaughan, 1992, for more exhaustive explanations).
The solution of equations 3.59 and 3.60 is a wave equation of amplitudes ␰1 and ␰2 :
µ r = ξ2 exp(irKa − iω t) (3.61)
[
µ r + 1 = ξ1 exp i (r + 1) Ka − iω t , ] (3.62)
where ␻ is the angular frequency
ω = 2π ν (3.63)
The system comprised of equations 3.62 and 3.63 has solutions only when the determinant
of coefficients ␰1 and ␰2 (secular determinant) is zero. The coefficient matrix, or dynamic
matrix, is
m1ω 2 − KF1 − KF2 K F1 exp(iKa ) + KF2 exp( −iKa )

.
KF1 exp( −iKa ) + KF2 exp(iKa ) m2ω 2 − KF1 − KF2 (3.64)
The general solution of matrix 3.64 is
  
12

KF + KF2  16KF1 KF2 m*2 2  Ka   
ω2 = 1 1 ± 1 − sin   , (3.65)

( )  2 
2
2m *  KF1 + KF2 m1m2 
   
where m* is the reduced mass (see section 3.1 and eq. 3.6).
For low values of K, sin2 Ka ⬵ K 2a2, and equation 3.65 has roots
(
ω 0 ≈ KF1 + KF2 )( m 1 + m2 ) K 2 a 2
−1
(3.66)
and
 1 1 
(
ω 1 ≈ KF1 + KF2  + ).
 m1 m2 
(3.67)
The first solution represents acoustic vibrational modes, and the second represents optical
vibrational modes.
A typical vibrational spectrum of a crystalline phase appears as a section of

the dispersion diagram along the ordinate axis. Figure 3.7 shows a generalized
Figure 3.7 Schematic representation of a vibrational spectrum of a crystalline phase.

Dotted curves: acoustic branches and optical continuum. Solid line: total spectrum. ␻E 1
and ␻E 2: Einstein oscillators. Reprinted with permission from Kieffer (1979c), Review of
Geophysics and Space Physics, 17, 35–39, copyright 䉷 1979 by the American Geophysi-
cal Union.
diagram of the density of states: in this case the ordinate axis represents the num-
ber of frequencies between ␻ and ␻ ⫹ d␻. Acoustic modes generally prevail at
low frequencies, the optical continuum occurs at intermediate frequencies, and
Einstein’s oscillators are found at high frequencies. According to Kieffer (1985),
the density of states for acoustic branches may be represented by three sinusoi-
dal functions:
[ ( )]
2
 2  1 arcsin ω / ω i
3 3
g (ω ) = ∑ 3N 0   + g 0(ω i ) , (3.68)
π Z
( )
12
i =1 ω i2 − ω
where ␻i is the maximum frequency of each acoustic branch (see Kieffer, 1979c).
Posing X ⫽ ប␻ /kT (cf. eq. 3.36), equation 3.68 (dispersed acoustic function; cf.
Kieffer, 1985) can be rewritten as
[arcsin (X X )] X
2
2
 2
3 Xi
i e 2 dX
S ( Xi ) =  ∫ . (3.69)
π
( ) ( )
12 2
0 Xi 2 − X 2 eX −1
In the optical continuum, frequency distribution is approximately constant (see

figure 3.7):
g (ω ) = g 0 for ω > ω3. (3.70)
The corresponding vibrational function in the optical continuum frequency field

has the form
Xu
X 2 e X dX
K ( X l ,X u ) = ∫ (3.71)
(X ) (e )
X 2
Xl u − Xl X
−1
(Kieffer, 1979c, 1985), where Xl and Xu correspond, respectively, to the lower and
upper bounds.
For the Einstein oscillators,
g (ω ) = g 0 + g E for ω = ωE , (3.72)
and the corresponding vibrational function is
X E2 e XE
E( X ) =
(e ) (3.73)
E 2
XE
−1
(cf. eq. 3.37). The molar heat capacity at constant V, normalized to a monatomic
equivalent (CV*), based on vibrational functions 3.69, 3.71, and 3.73, is
3N 0 k 3  1 
CV* = ∑
N i =1
S ( X ) + 3 N 0 k 1 −
i  N
− q K ( Xl , Xu ) + 3N 0 kqE( XE ) , (3.74)

where N ⫽ nZ is the number of atoms in the primitive cell and q is the proportion
of vibrational modes due to Einstein oscillators.
Because N 0 k ⫽ R, the unnormalized heat capacity (CV ) is
3R 3  1 
CV = ∑
Z i =1
S( X ) + 3nR1 −
i  N
− q K ( Xl , Xu ) + 3nRqE ( XE ) .

(3.75)
Because the dispersed acoustic function 3.69, the optic continuum function 3.71,
and the Einstein function 3.73 may be tabulated for the limiting values of undi-
mensionalized frequencies (see tables 1, 2, 3 in Kieffer, 1979c), the evaluation of
CV reduces to the appropriate choice of lower and upper cutoff frequencies for
the optic continuum (i.e., Xl and Xu limits of integration in eq. 3.71), of the three
zone edges for the dispersed acoustic function (i.e., three values of the Xi upper
limit of integration in eq. 3.69), and of the frequency bands representing Einstein
oscillators (i.e., XE in eq. 3.73).
The nondimensionalized frequencies are related to linear and angular frequencies by

equation 3.36. The conversion factor from linear frequencies in cm⫺1 to undimension-
alized frequencies is ch/k ⫽ 1.4387864 cm (where c is the speed of light in vacuum). Acous-
tic branches for the various phases of interest may be derived from acoustic velocities
through the guidelines outlined by Kieffer (1980). Vibrational modes at higher frequency
may be derived by infrared (IR) and Raman spectra. Note incidentally that the tabulated
values of the dispersed sine function in Kieffer (1979c) are 3 times the real ones (i.e., the
listed values must be divided by 3 to obtain the appropriate S (X l ) value for each acoustic
branch; see also Kieffer, 1985).
To appreciate the predictive properties of Kieffer’s model, it is sufficient to

compare calculated and experimental entropy values for several phases of geo-
chemical interest in table 3.1, which also lists entropy values obtained through
application of Debye’s and Einstein’s models. One advantage of Kieffer’s model
with respect to the two preceding formulations is its wider T range of applicability
(Debye’s model is appropriate to low frequencies and hence to low T, whereas
Einstein’s model is appropriate to high frequencies and hence to high T ).
The bias observed between experimental measurements and Kieffer’s model
predictions is due to the relative paucity of experimental data concerning cutoff
frequencies of acoustic branches, and also to the assumption that the frequencies
of the lower optical branches are constant with K and equivalent to those detected
by Raman and IR spectra (corresponding only to vibrational modes at K ⫽ 0).
Indeed, several of these vibrational modes, and often the most important ones,
are inactive under Raman and IR radiation (Gramaccioli, personal communica-
tion). The limits of the Kieffer model and other hybrid models with respect to
nonempirical computational procedures based on the equation of motion of the
Born–Von Karman approach have been discussed by Ghose et al. (1992).
3.4 Empirical Estimates of Heat Capacity

at Constant Pressure
For many crystalline compounds, CP values are experimentally known and tabu-
lated at various T conditions. In several cases, the data are also presented in the
interpolated form of a heat capacity function through Maier-Kelley-type poly-
nomials.
Whenever the heat capacity CP or heat capacity function CP ⫽ f (T ) for a
given substance is not known, reasonable estimates can be afforded through addi-
tivity algorithms. The simplest procedure is decomposition of the chemical for-
Table 3.1 Entropy values obtained by application of Einstein

(E), Debye (D), and Kieffer (K) models, compared with
experimental data at three different temperatures. Data are
expressed in J/(mole ⫻ K). Values in parentheses are Debye
temperatures (␪D) and Einstein temperatures (␪E) adopted in
the respective models (from Kieffer, 1985).
Experimental
Mineral/T(K) Value E D K
Halite (210) (306)

298.15 34.8 34.2 33.3 35.1
700 55.0 55.1 54.0 55.9
1000 63.1 63.9 62.8 64.7
Periclase (567) (942)
298.15 13.2 12.4 9.9 13.2
700 31.2 30.9 26.9 31.2
1000 39.2 39.4 35.3 39.7
Brucite (1184) (697)
298.15 12.6 2.4 15.2 14.4
400 17.5 5.4 21.2 19.4
Corundum (705) (1026) (model 1)
298.15 10.0 8.6 8.6 10.3
700 26.8 25.8 25.0 27.3
1000 34.7 34.2 33.3 35.6
Spinel (680) (879) (model 2)
298.15 11.3 9.2 11.0 11.3
700 28.5 26.6 28.5 28.5
1000 36.9 35.0 36.9 36.9
Quartz (815) (567)
298.15 13.7 6.4 19.3 13.9
700 29.1 22.5 38.9 29.7
␣-Cristobalite (819) (519)
298.15 14.3 6.4 21.2 13.8
400 19.1 11.0 27.8 18.5
Cohesite (779) (676) (model 1)
298.15 13.5 7.0 15.8 14.0
700 29.5 23.5 34.7 30.2
1000 37.4 31.8 43.3 38.2
Stishovite (757) (921) (model 1)
298.15 9.0 7.5 10.2 10.6
700 24.5 24.2 27.5 26.9
1000 32.0 32.5 35.2 35.1
Rutile (580) (781) (model 1)
298.15 16.6 11.9 13.1 16.6
700 34.0 30.3 31.3 34.5
1000 42.1 38.8 39.8 42.9
Experimental
Albite (753) (472)

298.15 15.9 7.6 23.3 15.9
700 32.3 24.3 43.4 32.2
1000 40.2 32.6 52.1 40.2
Microcline (746) (460)
298.15 16.5 7.7 23.9 16.2
700 32.7 24.5 44.0 32.4
1000 40.8 32.8 52.7 40.5
Anorthite (757) (518)
298.15 15.3 7.5 21.3 14.8
700 32.3 24.2 41.1 31.0
1000 40.7 32.5 49.8 39.2
Clinoenstatite (691) (719)
298.15 13.4 8.9 14.6 13.6
700 29.9 26.2 33.2 30.6
1000 37.9 34.6 41.8 38.9
Orthoenstatite (705) (719)
298.15 13.1 8.6 14.6 13.1
700 – 25.8 33.2 30.1
1000 – 34.2 41.8 38.3
Diopside (680) (654)
298.15 14.2 9.2 16.4 14.5
700 31.2 26.6 35.5 31.5
1000 39.5 35.0 44.1 39.8
Jadeite (711) (724)
298.15 13.2 8.4 14.5 13.7
700 29.7 25.6 33.1 30.5
1000 37.5 34.0 41.6 38.7
Tremolite (763) (547)
298.15 13.3 7.4 20.1 14.1
700 30.1 24.0 39.8 30.8
1000 38.3 32.3 48.5 38.9
Muscovite (860) (520)
298.15 14.6 5.7 21.2 13.8
700 31.0 21.3 41.0 30.0
1000 39.2 29.4 49.7 38.0
Talc (882) (525)
298.15 12.4 5.4 21.0 12.5
700 28.6 20.8 40.8 28.5
Calcite (766) (468)
298.15 18.3 7.3 23.5 19.2
700 35.3 23.9 43.6 35.3
1000 43.3 32.2 52.3 43.3
continued
Experimental
Zircon (694) (601)

298.15 14.0 8.82 18.1 14.4
700 31.0 26.2 37.5 31.2
1000 39.3 34.5 46.2 39.5
Forsterite (670) (747)
298.15 13.4 9.4 13.9 14.0
700 30.2 27.0 32.3 30.2
1000 38.1 35.4 40.9 39.5
Pyrope (697) (794)
298.15 13.3 8.8 12.8 13.6
700 28.2 26.1 30.9 30.5
1000 36.5 34.4 39.4 38.7
Grossular (704) (821)
298.15 12.7 8.6 12.2 12.6
700 29.7 25.8 30.1 29.5
1000 37.9 34.2 38.6 37.7
Kyanite (757) (916)
298.15 10.4 7.5 10.3 10.8
700 26.6 24.2 27.6 27.2
1000 34.2 32.5 36.0 35.4
Andalusite (754) (761)
298.15 11.5 7.5 13.5 12.0
700 27.5 24.3 32.0 28.4
1000 35.1 32.6 40.4 36.5
Sillimanite (789) (782)
298.15 12.0 6.9 13.0 12.3
700 28.3 23.2 31.3 28.5
1000 36.1 31.5 39.8 36.6
mula into the oxide constituents, followed by stoichiometric summation of the CP

values of such constituents. The procedure is generalized by the expression
C P0
r ,Tr ,i
= ∑ v j ,iCP0r ,Tr , j , (3.76)
j
where C 0Pr ,Tr ,i is the standard state heat capacity for mineral i, C 0Pr ,Tr ,j is the stan-
dard state heat capacity for the jth constituent oxide, and vj,i is the stoichiometric
number of moles of the jth oxide in mineral i.
For instance, general equation 3.76 applied to forsterite gives
CP0r ,Tr ,Mg 2 SiO4 = 2CP0r ,Tr ,MgO + CP0r ,Tr , SiO2 . (3.77)
Table 3.2 Coefficients of Berman-Brown polynomials for the commonest

oxide constituents of rock-forming minerals. The resulting CP is in J/(mole
⫻ K) (from Berman and Brown, 1985).
Oxide K1 K2 ⫻ 10⫺2 K3 ⫻ 10⫺5 K4 ⫻ 10⫺7
Na2O 95.148 0.0 ⫺51.0405 83.3648

K2O 105.140 ⫺5.7735 0.0 0.0
CaO 60.395 ⫺2.3629 0.0 ⫺9.3493
MgO 58.196 ⫺1.6114 ⫺14.0458 11.2673
FeO 77.036 ⫺5.8471 0.0 0.5558
Fe2O3 168.211 ⫺9.7572 0.0 ⫺17.3034
TiO2 85.059 ⫺2.2072 ⫺22.5138 22.4979
SiO2 87.781 ⫺5.0259 ⫺25.2856 36.3707
Al2O3 155.390 ⫺8.5229 ⫺46.9130 64.0084
H2Oa 106.330 ⫺12.4322 0.0 9.0628
H2Ob 87.617 ⫺7.5814 0.0 0.5291
CO2 119.626 ⫺15.0627 0.0 17.3869
a
“Structural” water. b“Zeolitic” water.
Because equation 3.76 is valid at the various T conditions, by application of the

additivity of polynomials, it follows that the coefficients of Maier-Kelley-type
functions for mineral i can also be derived from the corresponding coefficients of
the constituent oxides—i.e.,
K 1,i = ∑ v j ,i K1, j (3.78.1)

j
K 2 ,i = ∑ v j ,i K 2 , j (3.78.2)
j
K n,i = ∑ v j ,i K n, j . (3.78.3)
j
Table 3.2 lists the optimal values of the interpolation coefficients estimated
by Berman and Brown (1987) for the most common oxide constituents of rock-
forming minerals. These coefficients, through equations 3.78.1, 3.78.2, and 3.78.3,
allow the formulation of polynomials of the same type as equation 3.54, whose
precision is within 2% of experimental CP values in the T range of applicability.
However, the tabulated coefficients cannot be applied to phases with lambda tran-
sitions (see section 2.8).
Because the heat capacities of crystalline solids at various T are related to the
vibrational modes of the constituent atoms (cf. section 3.1), they may be expected
to show a functional relationship with the coordination states of the various
atoms in the crystal lattice. It was this kind of reasoning that led Robinson and
Table 3.3 Coefficients of “structural components” valid for polynomial 3.79. Coefficient K6
appears only at integration for calculation of entropy. The resulting CP is in J/(mole ⫻ K)
(from Robinson and Haas, 1983).
Structural
component K1 K2 K3 K6 K4 K5
Al2O3-IV 156.985 6.4774E-3 0.00 ⫺992.000 0.0000 ⫺1372.21

Al2O3-V 205.756 ⫺7.82311E-3 0.00 ⫺1349.63 0.0000 ⫺2084.06
Al2O3-VI 222.740 ⫺8.20451E-3 0.00 ⫺1507.24 0.0000 ⫺2464.56
CaO-VI 78.8255 ⫺1.91875E-3 0.00 ⫺480.538 0.0000 ⫺622.865
CaO-VII 78.8255 ⫺1.91875E-3 0.00 ⫺471.709 0.0000 ⫺622.865
CaO-VIII 83.6079 ⫺2.97891E-3 19,661.5 ⫺515.167 0.0000 ⫺716.401
Fe2O3-IV 318.412 ⫺4.89380E-3 417,088 0.00000 2.6E-5 ⫺3307.95
Fe2O3-VI 318.412 ⫺4.89380E-3 417,088 0.00000 2.6E-5 ⫺3307.95
FeO-VI 81.1612 0.00000000 0.00 ⫺485.209 0.0000 651.941
F 13.9627 1.28265E-2 0.00 ⫺52.2387 0.0000 0.00000
H 2O 56.9125 0.00000000 0.00 ⫺300.702 0.0000 ⫺263.847
OH 129.124 ⫺6.01221E-3 632,070 ⫺886.693 0.0000 ⫺1645.32
K2O-VIII 7.71711 5.27163E-2 0.00 108.368 0.0000 656.875
K2O-VI 42.4609 1.70942E-2 0.00 ⫺125.937 0.0000 171.435
MgO-IV 43.0846 7.44796E-4 0.00 ⫺206.902 0.0000 0.00000
MgO-VI 89.9331 ⫺3.19321E-3 0.00 ⫺588.796 0.0000 ⫺872.529
MgO-VIII 47.8300 0.00000000 ⫺810,599 ⫺245.321 0.0000 0.00000
Na2O-VI 58.0738 1.24598E-2 0.00 ⫺226.355 0.0000 ⫺45.8234
Na2O-VII 58.0738 1.24598E-2 0.00 ⫺251.172 0.0000 ⫺45.8234
Na2O-VIII 58.0738 1.24598E-2 0.00 ⫺259.204 0.0000 ⫺45.8234
SiO2-IV 109.383 ⫺2.77591E-3 0.00 ⫺704.147 0.0000 ⫺1083.05
Haas (1983) to formulate a set of coefficients valid for “structural components”

that take into account the structural states of the various elements in the phase
of interest (table 3.3). The coefficients are then applied to a modified form of the
Haas-Fisher polynomial—i.e.,
C P = K 1 + 2 K 2 T + K 3T −2 + K 4T 2 + K 5T −1 2 . (3.79)
Using the structural coefficients in table 3.3 implies knowledge of the condi-
tions of internal disorder of the various cations in the structure. In most cases
(especially for pure components), this is not a problem. For instance, table 3.4
shows the decomposition into “structural components” of the acmite molecule,
in which all Na⫹ is in VIII-fold coordination with oxygen, Fe3⫹ is in VI-fold
coordination, and Si4⫹ is in tetrahedral coordination. In some circumstances,
however, the coordination states of the various elements are not known with
sufficient precision, or, even worse, vary with T. A typical example is spinel
(MgAl2O4 ), which, at T ⬍ 600 °C, has inversion X ⫽ 0.07 (i.e., 7% of the tetrahe-
drally coordinated sites are occupied by Al 3⫹ ) and, at T ⫽ 1300 °C, has inversion
X ⫽ 0.21 (21% of tetrahedral sites occupied by Al 3⫹ ). Clearly the calculation of
Table 3.4 Coefficients of modified Haas-Fisher polynomial 3.79 for acmite component of
pyroxene (NaFeSi2O6), after application of structural coefficients in table 3.3. The resulting
CP is in J/(mole ⫻ K) (adapted from Robinson and Haas, 1983).
Partial Contribution to the Coefficients
Component Moles K1 K2 K3 K4 K5
Na2O-VIII 0.5 29.0370 6.22990E-3 0.0 0.0 ⫺0.02291

Fe2O3-VI 0.5 159.206 2.44690E-3 208,544 1.2856E-5 ⫺1653.98
SiO2-IV 2.0 218.766 ⫺5.55182E-3 0.0 0.0 ⫺2166.10
NaFeSi2O6 1.0 407.009 3.12498E-3 208,544 1.2856E-5 ⫺3842.99
the Haas-Fisher coefficients of the CP function through the structural compo-

nents is different for the two different occupancy states. The resulting Haas-Fisher
polynomials thus have only local validity and cannot be extended to the whole T
range of stability of the phase (note, however, that, according to Berman and
Brown, 1985, the structural effect on CP at high T is quite limited, affecting less
than 2% of the bulk amount).
A method alternative to the simple additive procedures described above is
outlined by Helgeson et al. (1978). An exchange reaction is written, involving the
compound of interest and an isostructural compound for which the CP function
is known with sufficient precision. It is then assumed that the ⌬CP of the reaction
is negligible. Consider for instance the exchange reaction between the wollastonite
and clinoenstatite components of pyroxene:
CaSiO3 + MgO ⇔ MgSiO3 + CaO .

wollastonite clinoenstatite
(3.80)
The heat capacities of MgSiO3 , CaO, and MgO being known, we derive the heat
capacity of CaSiO3 by applying
CP CaSiO3 = CP MgSiO3 + CP CaO − CP MgO . (3.81)
According to Helgeson et al. (1978), the method is quite precise and leads to
estimates better than those provided by simple additivity procedures.
Table 3.5 compares estimates made by the different methods: the procedure
of Helgeson et al. (1978) effectively leads to the best estimates. According to
Robinson and Haas (1983), however, the result of this procedure is path-
dependent, because it is conditioned by the selected exchange components:
different exchange reactions (with different reference components) can lead to
appreciably different results for the same component.
Table 3.5 Comparison of simple additivity of oxide constituents

(column II) and exchange method of Helgeson et al. (1978)
(column I), as methods of estimating heat capacity for crystalline
components. Experimental values are shown for comparison in
column III. Lower part of table: adopted exchange reactions (for
which it is assumed that ⌬CP reaction ⫽ 0). Data in J/(mole ⫻ K)
(adapted from Helgeson et al., 1978).
Mineral Formula I II III
Wollastonite CaSiO3 84.1 87.4 85.4

Clinoenstatite MgSiO3 80.3 82.4 79.1
Forsterite Mg2SiO4 116.7 120.9 118.0
Fayalite Fe2SiO4 141.8 143.9 133.1
Diopside CaMgSi2O6 164.4 169.5 156.9
Akermanite Ca2MgSi2O7 208.8 212.5 212.1
Merwinite Ca3MgSi2O8 253.6 255.2 252.3
Low albite NaAlSi3O8 208.8 207.5 205.0
Microcline KalSi3O8 212.5 215.1 202.9
CaSiO3 ⫹ MgO ⇔ MgSiO3 ⫹ CaO
MgSiO3 ⫹ CaO ⇔ CaSiO3 ⫹ MgO
Mg2SiO4 ⫹ 2CaO ⇔ ␣-Ca2SiO4 ⫹ 2MgO
Fe2SiO4 ⫹ 2MgO ⇔ Mg2SiO4 ⫹ 2FeO
CaMgSi2O6 ⇔ MgSiO3 ⫹ CaSiO3
Ca2MgSi2O7 ⫹ ␣-Al2O3 ⇔ Ca2Al2SiO7 ⫹ MgO ⫹ SiO2
Ca3MgSi2O8 ⇔ Ca2MgSi2O7 ⫹ CaO
NaAlSi3O8 ⫹ CaO ⫹ 0.5␣-Al2O3 ⇔ CaAl2Si2O8 ⫹ SiO2 ⫹ 0.5Na2O
KAlSi3O8 ⫹ 0.5Na2O ⇔ NaAlSi3O8 ⫹ 0.5K2O
3.5 Empirical Estimates of Standard State Entropy
As shown by Helgeson et al. (1978), satisfactory estimates of standard state molar

entropy for crystalline solids can be obtained through reversible exchange reac-
tions involving the compound of interest and an isostructural solid (as for heat
capacity, but with a volume correction). Consider the generalized exchange re-
action
M v X + v M i O ⇔ M iv X + v MO, (3.82)
where Mv X and M iv X are isostructural solids, and the molar entropies of Mv X,

Mi O, and MO are known; X is the common anionic radical, and v is the number
of moles of oxide components in the isostructural solids.
Posing
S S0 = ∆S reaction + SM0 = S 0i + vSMO

0
− vS 0 i (3.83)
vX Mv X MO
and
VS0 = ∆Vreaction + VM0 X = VM0 vi X + vVMO

0
− vV 0
, (3.84)
v Mi O
where S 0 and V 0 are, respectively, standard molar entropies and standard molar
volumes, the standard molar entropy of the compound of interest is given by
S S0 VS0 + V 0 
Miv X 
S 0i = . (3.85)
Mv X 2VS0
For nonisostructural solids, satisfactory estimates can be obtained with the

method proposed by Fyfe et al. (1958):
 
S 0i
MvX
= ∑ v j ,i S j0 + K VM0 X − ∑ v j ,iV j 0  ,
i
v
(3.86)
j j
where vj,i is the number of moles of the jth oxide per formula unit of compound
i, V 0j is the standard molar volume of the jth oxide, and S 0j is the standard molar
entropy of the jth oxide. Constant K is related to the partial derivative of entropy
with respect to volume, corresponding to the ratio between isobaric thermal
expansion ␣ and isothermal compressibility ␤:
 ∂S   ∂P  α
  =  = . (3.87)
 ∂V  T  ∂ T  V β
Holland (1989) reconsidered the significance of constant K in light of Ein-

stein’s model for the heat capacity of solids (see eq. 3.35 and 3.45):
 ∂S  nRX 2
  = ,
 ∂V  298 V 0
Miv X
( )(
e X − 1 1 − e −X ) (3.88)
where n is the number of atoms in the formula unit and X is the nondimension-
alized frequency (see eq. 3.36). As shown by Holland (1989), the mean value
of ( ∂S/ ∂V )298, obtained by application of equation 3.88 to a large number of
compounds, is 1.07 ⫾ 0.11 J ⭈ K⫺1 ⭈ cm⫺3. An analogous development using
Debye’s model (more appropriate to low-T conditions; see section 3.1) leads to a
mean partial derivative ( ∂S/ ∂V )298 of 0.93 ⫾ 0.10 J ⭈ K⫺1 ⭈ cm⫺3. Based on the
above evidence, Holland (1989) proposed a set of S 0j ⫺ V 0j finite differences for
Table 3.6 Comparison of predictive capacities of various equations in estimating standard
molar entropy (T ⫽ 298.15 K, P ⫽ 1 bar). Column I ⫽ simple summation of standard molar
entropies of constituent oxides. Column II ⫽ equation 3.86. Column III ⫽ equation 3.86
with procedure of Holland (1989). Column IV ⫽ equation 3.85. Values are in J/(mole ⫻ K).
Lower part of table: exchange reactions adopted with equation 3.85 (from Helgeson et al.,
1978) and S̄ 1j finite differences for structural oxides (Holland, 1989).
Mineral Formula Unit Experimental I II III IV
Clinoenstatite MgSiO3 67.8 70.3 61.5 67.2 67.8

Wollastonite CaSiO3 82.0 81.2 82.4 79.3 83.7
Diopside CaMgSi2O6 143.1 149.4 131.0 144.2 143.9
Giadeite NaAlSi2O6 133.5 145.6 120.1 134.8 136.0
Tephroite Mn2SiO4 163.2 160.7 159.4 162.1* 162.3
Ca-olivine Ca2SiO4 120.5 120.9 128.0 120.5 120.9
Phenakite Be2SiO4 64.4 69.5 64.4 – 69.0
Willemite Zn2SiO4 131.4 128.4 131.4 – 131.8
Akermanite Ca2MgSi2O7 209.2 189.1 195.8 208.0 209.2
Merwinite Ca3MgSi2O8 253.1 228.9 227.2 252.2 250.6
Low albite NaAlSi3O8 207.1 187.0 203.8 206.8 203.3
Nefeline NaAlSiO4 124.3 104.2 120.1 125.8 125.9
Tremolite Ca2Mg5Si8O22(OH)2 548.9 584.9 554.8 553.2 551.5
Caolinite Al2Si2O5(OH)4 202.9 213.8 216.7 188.4 201.7
MgSiO3 ⫹ MnO ⇔ MnSiO3 ⫹ MgO
CaSiO3 ⫹ MnO ⇔ MnSiO3 ⫹ CaO
CaMgSi2O6 ⇔ MgSiO3 ⫹ CaO ⫹ SiO2
NaAlSi2O6 ⫹ CaO ⫹ MgO ⇔ CaMgSi2O6 ⫹ 0.5Na2O ⫹ 0.5Al2O3
Mn2SiO4 ⫹ 2MgO ⇔ Mg2SiO4 ⫹ 2MnO
Ca2SiO4 ⫹ 2BeO ⇔ Be2SiO4 ⫹ 2CaO
Be2SiO4 ⫹ 2MgO ⇔ Mg2SiO4 ⫹ 2BeO
Zn2SiO4 ⫹ 2MgO ⇔ Mg2SiO4 ⫹ 2ZnO
Ca2MgSi2O7 ⫹ Al2O3 ⇔ Ca2Al2SiO7 ⫹ MgO ⫹ SiO2
Ca3MgSi2O8 ⫹ CaO ⇔ 2Ca2SiO4 ⫹ MgO
NaAlSi3O8 ⫹ 0.5K2O ⇔ KAlSi3O8 ⫹ 0.5Na2O
NaAlSiO4 ⫹ 0.5K2O ⇔ KAlSiO4 ⫹ 0.5Na2O
Ca2Mg5Si8O22(OH)2 ⇔ Mg3Si4O10(OH)2 ⫹ 2CaMgSi2O6
Al2Si2O5(OH)4 ⫹ 2SiO2 ⇔ Al2Si4O10(OH)2 ⫹ H2O
Component S 1j Component S 1j Component S 1j

[4] [8] [8]
SiO2 17.45 CaO 27.37 Na2O 56.32
[6] [8–10] [9–12]
Al2O3 22.60 CaO 34.37 Na2O 79.49
[4] [gt] [a]
Al2O3 28.89 CaO 17.86 K2O 79.55
[6] [4–8] [b]
MgO 15.75 FeO 30.78 K2O 87.96
[4] [gt]
MgO 18.77 FeO 36.50 H2O(a) 15.71
[7] [6]
MgO 21.06 MnO 33.41 H2O(b) 7.44
[gt] [6]
MgO 26.06 TiO2 32.63
[6] [6]
CaO 21.94 Fe2O3 50.24
[6]
⫽ VI-fold coordination
[a]
⫽ framework positions
[b]
⫽ interlayer positions and cavities
⫽ valid for garnets
[gt]
H2O(a) ⫽ diaspore, gibbsite, serpentine, dioctahedral micas

H2O(b) ⫽ chlorites, talc, trioctahedral micas, amphiboles
*29.8 J⋅ K⫺1 ⭈ mole⫺1 added to account for electronic contribution to configurational disorder.
structural oxides, to be used in conjunction with equation 3.86 at K ⫽ 1. These

terms, in units of J ⭈ K⫺1 ⭈ mole⫺1, are reported in the lower part of table 3.6.
Standard molar entropy estimates obtained by means of equation 3.85 are generally
within 1% deviation of the calorimetric entropy for the majority of silicates, provided
that the application does not involve ferrous compounds. Estimates obtained by applying
equation 3.86 are generally less precise. When applied to ferrous compounds, either equa-
tion (3.85 or 3.86) leads to values that are generally higher than calorimetric ones. Ac-
cording to Helgeson et al. (1978), the bias is around 10 J per mole of FeO in the com-
pound. According to Burns (1970) and Burns and Fyfe (1967), the lower entropy of
ferrous compounds is due to the nonspherical symmetry of Fe2⫹ ions in oxides and sili-
cates. More precisely, the entropy differences are due to the different electronic contribu-
tions of the d orbitals in the various coordination states (see equations 4 to 7 in Wood,
1980). To account for such contributions, appropriate correction factors for nonspherical
ions can be introduced in equation 3.85. With the method of Holland (1989), equation
3.86 reduces to S M 0
i
vX
⫽ ∑ j v j,iS 1j ⫹ V 0MivX where the S j1 terms are finite differences
S j ⫺ V j for structural oxides. The precision of the method is comparable to that of equa-
0 0
tion 3.85 (average absolute deviation ⫽ 1.41 J ⭈ K⫺1 ⭈ mole⫺1 over 60 values).
Comparative evaluation of the predictive properties of equations 3.85 and 3.86 is
given in table 3.6. Column I lists values obtained by simple summation of the entropies
of the constituent oxides. This method (sometimes observed in the literature) should be
avoided, because it is a source of significant errors. The lower part of table 3.6 lists the
adopted exchange reactions and the S j1 terms of Holland’s model.
3.6 Estimates of Enthalpy and Gibbs Free Energy

of Formation from Elements by Additivity:
the ⌬O2⫺ Method
Both the enthalpy of formation from the elements and the Gibbs free energy of
formation from the elements of various crystalline compounds at the standard
state of 298.15 K, 1 bar, can be predicted with satisfactory approximation through
the linear additivity procedures developed by Yves Tardy and colleagues (Tardy
and Garrels, 1976, 1977; Tardy and Gartner, 1977; Tardy and Viellard, 1977;
Tardy, 1979; Gartner, 1979; Viellard, 1982). Tardy’s method is based on the defi-
nition of the ⌬O2⫺ parameter, corresponding to the enthalpy (⌬H O2⫺ ) or the
Gibbs free energy (⌬GO2⫺ ) of formation of a generic oxide M2/zO(crystal) from its
aqueous ion:
2
∆ H O 2 − M z + = ∆H 0f M 2 / z O ( crystal) − ∆H 0f M z( +aqueous) (3.89)
z
2
∆ G O 2 − M z + = ∆G 0f M 2 z O ( crystal) − ∆G 0f M z + . (3.90)
z ( aqueous)
Table 3.7 Values of ⌬GO2⫺for various cations. Data in

kJ/mole (from Tardy and Garrels, 1976).
Ion ⌬GO2⫺ Ion ⌬GO2⫺ Ion ⌬GO2⫺
H⫹ ⫺223.63 Fe2⫹ ⫺154.22 Y3⫹ ⫺143.09

Li⫹ ⫹27.36 Co2⫹ ⫺151.46 La3⫹ ⫺108.66
Na⫹ ⫹147.19 Ni2⫹ ⫺166.48 Ce3⫹ ⫺118.87
K⫹ ⫹245.68 Cu2⫹ ⫺192.17 Eu3⫹ ⫺142.38
Rb⫹ ⫹273.63 Zn2⫹ ⫺174.47 Ti3⫹ ⫺244.22
Cs⫹ ⫹289.62 Cd2⫹ ⫺150.50 V3⫹ ⫺210.33
Cu⫹ ⫺246.77 Hg2⫹ ⫺222.84 Cr3⫹ ⫺205.31
Ag⫹ ⫺165.06 Pd2⫹ ⫺250.62 Mn3⫹ ⫺241.42
Tl⫹ ⫺82.51 Ag2⫹ ⫺257.32 Fe3⫹ ⫺236.02
Be2⫹ ⫺225.10 Sn2⫹ ⫺231.04 Au3⫹ ⫺234.60
Mg2⫹ ⫺114.64 Pb2⫹ ⫺165.02 Ge4⫹ ⫺246.98
Ca2⫹ ⫺50.50 Al3⫹ ⫺203.89 Sn4⫹ ⫺263.26
Sr2⫹ ⫺2.51 Ga3⫹ ⫺228.74 Pb4⫹ ⫺260.75
Ba2⫹ ⫹32.22 In3⫹ ⫺207.78 Zr4⫹ ⫺259.24
Ra2⫹ ⫹71.13 Tl3⫹ ⫺246.98 Hf4⫹ ⫺259.41
V2⫹ ⫺168.62 Bi3⫹ ⫺225.81 U4⫹ ⫺226.35
Mn2⫹ ⫺135.56 Sc3⫹ ⫺205.64 Th4⫹ ⫺222.59
The ⌬O2⫺ parameter is thus in linear relation with the solution energy of a given
crystalline oxide M2/zO(crystal) in water, through
2 z+
2 H + + M 2 z O ( crystal) ⇔ M (aqueous) + H 2 O ( liquid) (3.91)
z
∆G reaction
0
= − ∆G O 2 − + ∆G 0f H 2 O ( liquid) (3.92)
∆H reaction
0
= − ∆ H O 2 − + ∆H 0f H 2 O ( liquid) . (3.93)
Table 3.7 lists the ⌬GO2⫺ parameters derived by Tardy and Garrels (1976) on
the basis of the Gibbs free energy of formation of oxides, hydroxides, and aqueous
ions. The ⌬O2⫺ parameters allow us to establish linear proportionality between
the enthalpy (or Gibbs free energy) of formation of the compound from the con-
stituent oxides within a given class of solids. If we define ⌬Hcompound and ⌬Gcom-
pound as, respectively, the enthalpy of formation and the Gibbs free energy of for-
mation of a given double oxide, starting from the oxide components M 2 / z ⫹O and 1
N 2 / z O, the general proportionality relations between these two magnitudes and the
2
⌬O2⫺ parameters of the corresponding cations have the form
∆H compound = −α H
n1n2
n1 + n 2
(
∆ HO2− Mz + ∆ HO2− N z
+
1
+
2
) (3.94)
and
∆G compound = −α G
n1 n2
n1 + n2
(
∆GO2 − M z + ∆GO2 − N z
+
1
+
2
) (3.95)
where n1 and n2 are the numbers of oxygen ions linked, respectively, to the M z1⫹
and N z2⫹ cations, and ␣H and ␣G are the correlation parameters for a given class
of compounds (double oxides). Adding to the ⌬Hcompound (or ⌬Gcompound ) the
enthalpy (or Gibbs free energy) of formation from the elements of the oxides in
question, the enthalpy (or Gibbs free energy) of formation from the elements of
the compound of interest is readily derived.
For instance, let us consider the compound clinoenstatite. Its enthalpy of
formation from the constituent oxides is
∆H MgSiO3 = H 0f ,MgSiO − H f,MgO

0
− H 0f , SiO . ( 3.96.1)
3 2
n1 n2 2
With = ; α H = −1.3095 ( 3.96.2 )
n1 + n2 3
and ∆ H O 2 − Mg 2 + = −139.95 ; ∆ H O 2 − Si 4 + = −193.33, ( 3.96.3)
we obtain
∆H MgSiO3 = − 46.60; H 0f ,MgSiO = − 46.60 − 601.49 − 910.70

3
(3.97)
= −1558.79.
The experimental value is ⫺1547.45 (kJ/mole). The estimate thus has an approxi-
mation of about 10 kJ/mole.
As figure 3.8 shows, equations 3.94 and 3.95 appear as straight lines of slope
␣H (or ␣G in the case of Gibbs free energy) in a binary plot ⌬compound ⫽ f (⌬O2⫺ ).
The intercept value on the abscissa, corresponding to zero on the ordinate, repre-
sents the ⌬HO2⫺ of the “reference cation” based on equation 3.94. Thus, in the
case of metasilicates (figure 3.8B), we have ⌬HO2⫺ Si4⫹ ⫽ ⫺193.93 kJ/mole (see
also equations 3.96.3 and 3.97), and, for orthosilicates (figure 3.8A),
⌬HO2⫺Si4⫹ ⫽ ⫺204.53 kJ/mole.
Table 3.8 lists values of ␣H and ⌬HO 2⫺
reference cation for silicates and aluminates.
The tabulated parameters, as simple interpolation factors, have good correlation
coefficients, confirming the quality of the regression.
Table 3.9 lists general regression parameters between ⌬GO2⫺ and the Gibbs
free energy of formation from constituent oxides, for various classes of sub-
stances. The values conform to the general equation
∆G compound = a × ∆ G O 2 − M z + + b. (3.98)
Figure 3.8 Relationship between ⌬Hcompound and ⌬H O2⫺ in orthosilicates (A) and meta-
silicates (B). The ⌬H O2⫺ value of the reference cation is found at a value of zero on the
ordinate. Reprinted from Viellard (1982), Sciences Geologiques, Memoir n°69, Université
Louis Pasteur, with kind permission of the Director of Publication.
Table 3.8 Regression parameters ⌬H O2⫺ for silicates and aluminates. Values in
kJ/mole (from Tardy and Garrels, 1976, 1977; Tardy and Viellard, 1977; Tardy
and Gartner, 1977).
Compound ␣H ⌬H Oref.
2⫺
cation Points R2 Precision
MSiO3 ⫺1.3095 ⫺193.93 13 0.9987 ⫾9.84

M2SiO4 ⫺1.305 ⫺204.53 14 0.9961 25.3
M3Si2O7 ⫺1.213 ⫺201.92 3 0.9999 8.4
MSi2O5 ⫺1.258 ⫺181.26 4 0.9993 6.2
M3SiO5 ⫺1.297 ⫺193.91 3 0.9703 73.5
MAl2O4 ⫺0.797 ⫺224.52 12 0.9374 22.3
Table 3.9 General regression parameters between ⌬H O2⫺ and ⌬Gcompound.

Values in kJ/mole (from Tardy and Garrels, 1976, 1977; Tardy and Viellard,
1977; Tardy and Gartner, 1977).
Compound a b Points R2 Precision
M(OH)x ⫺0.206 ⫺46.02 104 0.975 10.46

MxSiO3 ⫺0.673 ⫺126.90 0.998 11.72
MxSiO4 ⫺1.01 ⫺189.12 0.985 16.74
MxPO4 ⫺1.285 ⫺418.48 18 0.972 49.37
MxSO4 ⫺0.541 ⫺159.79 37 0.982 57.74
MxNO3 ⫺0.541 ⫺159.79 20 0.984 31.80
MxCO3 ⫺0.771 ⫺199.41 32 0.987 35.15
3.7 Gibbs Free Energy of a Phase at High P and T,

Based on the Functional Forms of Heat Capacity,
Thermal Expansion, and Compressibility
We may first assume that isothermal compressibility ␤V and isobaric thermal

expansion coefficient ␣V are independent, respectively, of T and P. Equations 1.91
and 1.99, integrated on T and P, respectively, give
VTr ,P = VT0 ,P exp − βV P − Pr

r r [ ( )] (3.99)
and
VT,P = VT0 ,P exp α V T − Tr

r r r [ ( )] . (3.100)
The final state of volume does not depend on the integration path (see appendix
2), and thus
r r [ (
VT ,P = VT0 ,P exp α V T − Tr )] exp[ − β (P − P )] .
V r (3.101)
The ⌬P values for which the change of volume is significant are quite high,
and reference pressure Pr (normally Pr ⫽ 1) can be ignored in the development of
calculations. The integrals on pressure in equations 2.94 and 2.95 take the forms
  ∂V  
P
∫
Pr
V − T 


[ (
1 
)]
  dP = PVTr, Pr exp α V T − Tr  1 − βV P  1 − α V T
 ∂T  P 
0
 2 
( )
( 3.102 )
and
Pr
∫
 ∂V 

 ∂T  P
0 

1
2
[ (

 dP = α V PVTrPr exp α V T − Tr  1 − βV P  .

)] (3.103)
We have seen (section 3.2) that heat capacity at constant P can be expressed
in a functional form representing its T-dependency. Adopting, for instance, the
Haas-Fisher polynomial:
C P = K 1 + K 2 T + K 3T −2 + K 4T 2 + K 5T −1 2 , (3.104)
the integrals on T in equations 2.94 and 2.95 take the forms
∫ CP dT (
= K 1 T − Tr + ) K2
2
(
T 2 − Tr2 − K 3 T −1 − Tr−1) ( )
Tr
(3.105)
K
( )
+ 4 T 3 − Tr3 + 2 K 5 T 1 2 − Tr1 2
3
( )
and
T
T 
∫ CP
dT
T
K
(
= K 1 ln   + K 2 T − Tr − 3 T −2 − Tr−2
 Tr  2
) ( )
Tr
(3.106)
K
( )
+ 4 T 2 − Tr2 − 2 K 5 T −1 2 − Tr1 2 .
2
( )
The enthalpy of the phase at the P and T of interest is derived from the
standard state values of enthalpy and volume (H 0Tr ,Pr and V 0Tr ,Pr ) through
T
H T,P = H T0 ,P +
r r ∫ CP dT
Tr
(3.107)
+ PVT 0,P exp α V T − Tr
r r [ ( )]  1 
 1 − βV P  1 − α V T .
 2 
( )
Analogously, for entropy we have
T
ST,P = ST0 ,P +
r r ∫ CP
Tr
dT
T r r


1
2
[ (
+ αV PVT0,P exp αV T − Tr 1 − βV P 

)] (3.108)
and for Gibbs free energy we have
T T
GT,P = GT0 ,P − ST0 ,P T − Tr +
r r r r
( ) ∫C P dT − T ∫ CP
dT
T
Tr Tr
(3.109)
 
+ PVT ,0P
r r

(1
exp 1 − αV T − Tr − βV P  .
2 
)
3.8 Solid Mixture Models
We will now consider in some detail the mixture models that can be applied to
rock-forming minerals. These models are of fundamental importance for the cal-
culation of equilibria in heterogeneous systems. In the examples given here we
will refer generally to binary mixtures forming isomorphous compounds over the
entire compositional field. Each of these mixtures also has a well-defined crystal
structure, with two or more energetically distinguishable structural sites. Exten-
sion of the calculations to ternary mixtures will also be given. The discussion will
be developed in terms of molar properties (i.e., referring to one mole of sub-
stance). The term “molar” will be omitted throughout the section, for the sake
of simplicity.
3.8.1 Ideal Mixture

An “ideal” binary crystalline mixture obeys the equations
H mixing = 0 (3.110)
and
(
S mixing = −R X A ln X A + X B ln X B , ) (3.111)
where XA and XB are the molar fractions of components A and B in a binary

mixture. The entropy of mixing described by equation 3.111 derives from the
Boltzmann equation:
S configuration = k ln Q , (3.112)
where k is the Boltzmann constant:
R
k= (3.113)
N0
and Q is “permutability”—i.e., the total number of possible configurational ar-

rangements in the mixture. Q can be calculated through the combinatorial
formula
N0 !
Q= .
( N A ) ! ( N B) !
0 0
(3.114)
Equations 3.112 and 3.114 give
 
S configuration = k ln 
( )(
N0 !
 N0 A ! N0 B ) ! 
[ ( ) (
 = k ln N0 ! − ln N0 A ! − ln N0 B ! . )] ( 3.115)
By application of Stirling’s approximation for large numbers (note that N 0 ⫽

6.022 ⫻ 1023 ):
ln N0 ! = N0 ln N0 − N0 , (3.116)
as
N0 = N0 A + N0 B , (3.117)
we have
[ ( )
S mixing = k N0 ln N0 − N0A ln N0 A − N0B ln N0 B ( )]
 N0 A  N0 A  N0B  N0B 
= − kN0  ln  + ln  .
 N0 A + N0 B  N0 A + N0 B  N0 A + N0 B  N0 A + N0 B  
( 3.118 )
Because
N0 A N0 B
= XA and = XB , (3.119)
N0 A + N 0 B N0 A + N0 B
equation 3.111 is readily derived. Moreover, because
G mixing = H mixing − TS mixing , (3.120)

then
G mixture = X A µ A0 + X B µ B0 + G mixing (3.121)
and, for an ideal mixture,
(
G mixture = X A µ A0 + X B µ B0 + RT X A ln X A + X B ln X B . ) (3.122)
Equations 3.121 and 3.122 distinguish the bulk Gibbs free energy of the mixture
(Gmixture ) from the Gibbs free energy term involved in the mixing procedure
(Gmixing ).
Let us now consider in detail the mixing process of two generic components
AN and BN, where A and B are cations and N represents common anionic radi-
cals (for instance, the anionic group SiO 44⫺ ):
AN + BN → ( A, B) N (3.123)
G mixing = ∆G123 . (3.124)
If mixture (A,B)N is ideal, mixing will take place without any heat loss or heat
production. Moreover, the two cations will be fully interchangeable: in other
words, if they occur in the same amounts in the mixture, we will have an equal
opportunity of finding A or B over the same structural position. The Gibbs free
energy term involved in the mixing process is
(
G mixing = ∆G123 = RT XA ln XA + XB ln XB )
(3.125)
[ (
= RT XA ln XA + 1 − XA ln 1 − XA) ( )] .
Adopting as standard state the condition of “pure component at T and P of

interest,” for whatever composition in the binary field, we have
 ∂ ∆G 123 
µ A − µ A0 =   = RT ln X A (3.126)
 ∂X A  P ,T
and
 ∂ ∆G 123 
µ B − µ B0 =  
 ∂X B  P ,T
= RT ln X B = RT ln 1 − X A .( ) (3.127)
Recalling now the relationship linking chemical potential to thermodynamic ac-

tivity (eq. 2.75 in section 2.9), for all compositions in the binary field we will
obtain
a A = XA and (
a B = XB = 1 − XA . ) (3.128)
The concepts above can be extended to multicomponent mixtures (i ⬎ 2),

representing the Gibbs free energy of the mixture as
G mixture = ∑ µ i0 Xi + G mixing .
(3.129)
i
The Gibbs free energy of mixing can also be expressed as
G mixing = ∑ (µi )
− µ i0 Xi
(3.130)
i
and for an “ideal” multicomponent mixture we can write
G mixing = G ideal mixing = RT ∑ X i ln Xi . (3.131)

i
Figure 3.9A shows that the Gibbs free energy of ideal mixing is symmetric-
concave in shape, with a maximum depth directly dependent on T. Note that the
ideal Gibbs free energy of mixing term is always present even in nonideal mixtures
(see figure 3.9B, C, and D, and section 3.11).
3.8.2 Generalities on Nonideal Mixtures

Maintaining the same standard state previously adopted for the ideal case, for
nonideal mixtures we have
 ∂∆G 123   ∂ G mixing 

µ i − µ i0 =   =  = RT ln a i , (3.132)
 ∂X i  P ,T  ∂X i  P ,T
where ai ⫽ Xi␥i and ␥i is the activity coefficient.

The Gibbs free energy of mixing is given by
G mixing = RT ∑ Xi ln a i = RT ∑ Xi ln X i + RT ∑ Xi ln γ i
i i i
(3.133)
= G ideal mixing + G excess mixing ,
where Gexcess mixing is the excess Gibbs free energy of mixing term.
Figure 3.9 Conformation of Gibbs free energy curve in various types of binary mixtures.
(A) Ideal mixture of components A and B. Standard state adopted is that of pure compo-
nent at T and P of interest. (B) Regular mixture with complete configurational disorder:
W ⫽ 10 kJ/mole for 500 ⬍ T(K) ⬍ 1500. (C) Simple mixture: W ⫽ 10 ⫺ 0.01 ⫻ T(K) (kJ/
mole). (D) Subregular mixture: A0 ⫽ 10 ⫺ 0.01 ⫻ T (kJ/mole); A1 ⫽ 5 ⫺ 0.01 ⫻ T (kJ/
mole). Adopting corresponding Margules notation, an equivalent interaction is obtained
with WBA ⫽ 15 ⫺ 0.02 ⫻ T (kJ/mole); WAB ⫽ 5 (kJ/mole).
It is clear from equation 3.133 that the thermodynamic properties of any

nonideal mixture are always related to the corresponding properties of an ideal
mixture through excess terms:
 ∂ G excess mixing 
S excess mixing = −   =
 ∂T 
(3.134)
 ∂ ln γ A ∂ ln γ B 
= − RT  XA
 ∂T
+ XB
∂ T 
(
− R XA ln γ A + XB ln γ B )
 ∂ G excess mixing T 
H excess mixing = −T 2  
 ∂T 
(3.135)
 ∂ ln γ A ∂ ln γ B 
= − RT  XA 2
+ XB
 ∂T ∂ T 
 ∂ G excess mixing   ∂ ln γ A ∂ ln γ B 
V excess mixing =   = RT  X A + XB , (3.136)
 ∂P   ∂P ∂P 
where ␥A and ␥B are the activity coefficients of components AN and BN, respec-
tively, in the (A,B)N mixture.
Let us now review the main types of nonideal mixtures.
3.8.3 Athermal Mixtures

In this type of nonideal mixture, deviations from ideality are solely due to en-
tropic terms:
H excess mixing = 0 (3.137)
G excess mixing = H excess mixing . (3.138)
3.8.4 Regular Mixtures with Complete Configurational

Disorder (Zeroth Approximation)
Deviation from ideality is entirely due to enthalpic terms. The entropy of mixing
is that of an ideal mixture:
S excess mixing = 0 (3.139)

G excess mixing = −TS excess mixing . (3.140)
Excess enthalpy is parameterized through interaction parameter W:
G excess mixing = X A XBW (3.141)
W = N0 w (3.142)
2w AB − w AA − w BB
w= . (3.143)
2
The terms wAB, wAA, and wBB are, respectively, the ionic interactions between
A-B, A-A, and B-B atoms in the (A,B)N mixture. Note that, in this model, W is
not dependent on T and P (see also figure 3.9B). The condition of complete disor-
der is often defined as “approximation of the Zeroth principle.”
3.8.5 Simple Mixtures

Interaction parameter W I does depend on T and P, and the excess Gibbs free
energy of mixing is described as in the preceding model:
G excess mixing = X A XB W I . (3.144)
T-dependency implies an entropic contribution to the nonideality of the mixing

process:
∂W I
− S excess mixing = X A XB (3.145)
∂T
 ∂W I 
H excess mixing = X A XB  W I − T . (3.146)
 ∂T 
The activity coefficients may be related to the molar concentration through the ex-
pressions
WI 2
ln γ A = XB (3.147)
RT
and
WI 2
ln γ B = X . (3.148)
RT A
It can easily be deduced from equations 3.147 and 3.148 that the activity-
composition relationships for the two components are symmetric in the composi-
tional field (see also figure 3.9C).
3.8.6 Quasi-Chemical Model

The interaction between components in a mixture is regarded as the result of
short-range electrostatic and repulsive interactions among the various ions. Ion
A tends to be coordinated by A ions, if this condition minimizes the electrostatic-
plus-repulsive potential of the structure, or, alternatively, by B ions, if this is the
more favorable configuration in terms of energy.
The molar fractions of components in mixture XA ⫹ XB are converted into
associative fractions ␾A and ␾B through contact factors qA and qB:
XAq A
φ A = 1 − φB = (3.149)
X A q A + XB q B
(see the concept of “associate” in section 4.8).

The tendency toward configurational disorder is expressed by parameter ␻q :
[ ]
12
ω q = 1 + 4φ Aφ B ( λ − 1) , (3.150)
where
 2W II 
λ = exp . (3.151)
 ZRT 
(Green, 1970). W II in equation 3.151 is the interaction parameter of the quasi-

chemical model and Z is the coordination number.
The activity coefficients of components in mixtures are expressed by
( )
Zq 2
 φ B ωq − 1  A
γA = 1 +  (3.152)

 ( )
φ A ωq + 1 

and
( )
Zq 2
 φ A ωq − 1  B
γB = 1 +  (3.153)

 ( )
φ B ωq − 1 

The excess Gibbs free energy of mixing is given by
G excess mixing =
Z



RT  X A q A ln 1 +
φ B ωq − 1  ( ) 
 + XB q B ln 1 +
( )
φ A ωq − 1  
 .
2  

φ A ωq + 1 
 ( ) 
 ( )
φ B ωq + 1  

( 3.154 )
Guggenheim (1952) named W II “cooperative energy”, intuitively associating this

term with an exchange reaction of the type
AA + BB ⇔ AB + AB. (3.155)
It can easily be seen that, when W II ⫽ 0 in equation 3.151, then ␭ ⫽ 1 and also
␻q ⫽ 1 (cf. eq. 3.150); it follows that activity coefficients ␥A and ␥B are 1 at all
concentrations (cf. eq. 3.152 and 3.153). Nevertheless, if 2W II /RT ⬎ 0, then ␻q is
greater than 1; in this condition configuration AA-BB is more stable than AB-
AB. There is then a tendency toward clustering of ions A-A and B-B, which is
phenomenologically preliminary to unmixing (the activity coefficients are corre-
spondingly greater than 1). If 2W II /RT ⬍ 0, then ␻q is less than 1, and there is a
tendency toward mixing.
The quasi-chemical model was derived by Guggenheim for application to or-
ganic fluid mixtures. Applying it to crystalline solids is not immediate, because it
necessitates conceptual modifications of operative parameters, such as the above-
mentioned “contact factor.” Empirical methods of derivation of the above param-
eters, based on structural data, are available in the literature (Green, 1970; Sax-
ena, 1972). We will not treat this model, because it is of scanty application in
geochemistry. More exhaustive treatment can be found in Guggenheim (1952)
and Ganguly and Saxena (1987).
3.8.7 Multisite and Reciprocal Mixtures

Let us consider a multisite mixture of type (A, B, C, . . .) v1 (M, N, O, . . .) v2Z,
where cations of types A, B, C and M, N, O, respectively, occupy energetically
distinct sites present in one mole of substance in the stoichiometric amounts v1
and v2 , and Z is the common anionic group. Applying the permutability concept
to each distinct site and assuming random mixing and the absence of interactions
on sites, the activity of component Av1M v2Z in the mixture may be expressed as
a Av M v 2Z
= X Av1 X Mv2 . (3.156)
1
Equation 3.156 follows the general treatment proposed by Temkin (1945) for
fused salts.
If site-interaction energy terms are nonnegligible, equation 3.156 must be

modified to account for site-activity coefficients, as follows:
a Av M v 2Z
= X Av1 γ Av1 X Mv 2 γ M
v2
. (3.157)
1
If, however, site interactions of atoms on energetically equivalent sites are equal
and the standard molar volumes of mixing components are not dissimilar (i.e.,
within 5 to 10% difference), equation 3.157 may be simplified and the activity of
any component i in the mixture (ai ) may be expressed in a generalized fashion as
shown by Helgeson et al. (1978):
ai = ki ∏ ∏ X j ,ss, j , i ,
v
(3.158)
s j
where vs, j,i is the number of sth energetically equivalent sites occupied by the jth
atom in one mole of the ith component, Xj,s is the molar fraction of atom j on
site s, and ki is a proportionality constant imposed by the limit
lim ai = 1.
Xi → 1 (3.159)
The constant ki may be also expressed in a generalized fashion, as shown by

Helgeson et al. (1978):
− v s, j , i
 v s , j ,i 
ki = ∏ ∏   , (3.160)
s j  v s ,i 
where vs,i is the number of sites of type s in the ith component:
v s ,i = ∑ v s , j ,i . (3.161)
j
Equation 3.158 is consistent with
 
Gmixing = RT ∑ X i  ln ki + ∑ ∑ vs, j,i ln X j, s  (3.162)
i s j
(Kerrick and Darken, 1975; Helgeson et al., 1978) and is strictly valid only for isovolu-
metric mixtures.
It must be noted that, when only one kind of atom occupies the s sites in the end-
member formula, as in our compound Av1M v 2Z, for instance, then vs, j,i ⬅ vs, i and ki
reduces to 1. In silicate end-members of geochemical interest, we often have two types
of atoms occupying the identical sublattice. For example, the end-member clintonite of
trioctahedral brittle micas has formula Ca(Mg2Al) (SiAl3 )O10 (OH)2 , with Al and Mg
occupying three energetically equivalent octahedral sites and Si and Al occupying four
energetically equivalent tetrahedral sites (cf. section 5.6). We thus have, for octahedral
sites, vs, i ⫽ 3, vs, j,i ⫽ 2 for Mg, and vs, j,i ⫽ 1 for Al. For tetrahedral sites, vs,i ⫽ 4, vs, j,i ⫽
1 for Si, and vs, j,i ⫽ 3 for Al. The resulting proportionality constant is
−2 −1 −1 −3
 2  1  1  3 
ki =         = 64. (3.163)
 3  3  4  4
In some cases, attribution of coefficients vs, i and vs, j,i is not so straightforward as it
may appear at first glance. This is the case, for instance, for K-feldspar (KAlSi3O8 ), with
Al and Si occupying T1 and T2 tetrahedral sites. Assuming, for the sake of simplicity,
that T1 and T2 sites are energetically equivalent (which is not the case; see section 5.7.2),
we have vs, i ⫽ 4, vs, j,i ⫽ 1 for Al, and vs, j,i ⫽ 3 for Si, so that
−1 −3
 1  3
ki =     = 9.481. (3.164)
 4  4
Adopting topologic symmetry, with two T1 sites energetically different from the two T2
sites, and assuming Al and Si to be randomly distributed in T1 and T2, we obtain
−0.5 −1.5 −0.5 −1.5

 0.5   1.5   0.5   1.5 
ki =         = 6.158. (3.165)
 2   2   2   2 
Assuming all T sites to be energetically different (which happens at low temperatures,

when T1o, T1m, T2o, and T2m nonenergetically equivalent sites exist and Al is stabilized
in site T1o; see section 5.7.2), we have
−1 −1 −1 −1
 1  1  1  1
ki =         = 1. (3.166)
 1  1  1  1
Because activity is related to molar concentration through the rational activity

coefficient, from equation 3.157 we may derive
γ Av Mv 2 Z = γ Av1 γ M
v2
(3.167.1)
1
XAv Mv 2 Z = X Av1 X M
v2
(3.167.2)
1
a Av Mv 2 Z = a Av1 aM
v2
. (3.167.3)
1
Equations 3.167.1, 3.167.2, and 3.167.3 find their analogs in the theory of electro-
lyte solutions, as we will see in detail in section 8.4.
Let us now consider again the mixture (A, B, C, . . .) v1(M, N, O, . . .) v2Z.
We can identify four components, Av1M v2Z, Av1N v2Z, B v1M v2Z, and B v1N v2Z, ener-
getically related by the reciprocal reaction
Av1Mv2 Z + Bv1 Nv2 Z ⇔ Bv1Mv2 Z + Av1 Nv2 Z. (3.168)
According to equation 3.168, the energy properties of the four components are
not mutually independent but, once the Gibbs free energy is fixed for three of
them, the energy of the fourth is constrained by the reciprocal energy term ⌬G168
⬅ ⌬Greciprocal. Accepting this way of reasoning, the relationship between rational
activity coefficient and site-activity coefficients is no longer given by equation
3.167.1 but, according to Flood et al. (1954), involves a reciprocal term ( ␥reciprocal )
whose nature is analogous to the cooperative energy term expressed by Guggen-
heim (1952) in the quasi-chemical model (see preceding section):
aAv
1
Mv 2 Z (
= X Av1 γ Av1 X M
v2 v2
)
γ M γ reciprocal (3.169)
 ∆G reciprocal
0 
γ reciprocal

(
= exp  1 − X A 1 − XM
RT
)( ,

) (3.170)
where ⌬ G 0reciprocal is the Gibbs free energy change of reciprocal reaction 3.168.
3.9 General Equations of Excess Functions

for Nonideal Binary Mixtures
So far, we have seen that deviation from ideal behavior may affect one or more
thermodynamic magnitudes (e.g., enthalpy, entropy, volume). In some cases, we
are able to associate macroscopic interactions with real (microscopic) interactions
of the various ions in the mixture (for instance, coulombic and repulsive interac-
tions in the quasi-chemical approximation). In practice, it may happen that none
of the models discussed above is able to explain, with reasonable approximation,
the macroscopic behavior of mixtures, as experimentally observed. In such cases
(or whenever the numeric value of the energy term for a given substance is more
important than actual comprehension of the mixing process), we adopt general
(and more flexible) equations for the excess functions.
Let us consider a generic mixture of components 1 and 2 in the binary compo-
sitional field X1 ⫽ 0 → 1, X2 ⫽ 1 → 0. The excess Gibbs free energy of mixing is

expressed as a polynomial function of the composition:
( )
G excess mixing = X 1X 2  A0 + A1 X 1 − X 2 + A2 X 1 − X 2 ( ) + L,
2
(3.171)
 
and the activity coefficients are defined by
 ∂ G excess mixing 
RT ln γ 1 = G excess mixing + X 2   =
 ∂X 1 
(3.172)
= X 22 [A0 ( ) (
+ A1 3X 1 − X 2 + A2 X 1 − X 2 5X 1 − X 2 + L)( ) ]
[ ( ) (
RT ln γ 2 = X 12 A0 − A1 3X 2 − X 1 + A2 X 2 − X 1 5X 2 − X 1 + L . )( ) ] (3.173)
Guggenheim’s polynomial expansion (equation 3.171; Guggenheim, 1937) and

the two Redlich-Kister equations (3.172 and 3.173; Redlich and Kister, 1948) are
of general applicability for any type of mixture:
1. If A 0 ⫽ A1 ⫽ A2 ⫽ 0, the excess Gibbs free energy of mixing is zero

throughout the compositional field and the mixture is ideal.
2. If A 0 ⬆ 0 and A1 ⫽ A2 ⫽ 0, the mixture is regular:
A0 ≡ W ⇒ G excess mixing = X 1 X 2W . (3.174)
3. If all the odd coefficients in the polynomial expansions (i.e., A1 , A3 , . . .)

are zero, the mixture is symmetric (i.e., ␥1 and ␥2 assume identical values
for symmetric compositions in the binary field; see, for instance, figure
3.9C).
4. If A 0 ⬆ 0 and A1 ⬆ 0, the excess functions do not exhibit symmetrical
properties over the compositional field. In this case, the mixture is defined
as subregular (see, for instance, figure 3.9D).
The subregular model approximates several silicate mixtures with sufficient

precision, as we will see in chapter 5. For a subregular mixture, we have
[
G excess mixing = X 1 X 2 A0 + A1 X 1 − X 2( )] (3.175)
[ (
RT ln γ 1 = X 22 A0 + A1 3X 1 − X 2 )] (3.176)
[ (
RT ln γ 2 = X 12 A0 − A1 3X 2 − X 1 . )] (3.177)
Van Laar parameters A 0 and A1 can be translated into the corresponding Mar-
gules subregular parameters W12 and W21 , posing
W12 = A0 − A1 ; W21 = A0 + A1 (3.178)
and
W21 + W12 W21 − W12

A0 = ; A1 = . (3.179)
2 2
The Margules subregular model then becomes
(
G excess mixing = X 1 X 2 W12 X 2 + W21 X 1 ) (3.180)
[ (
RT ln γ 1 = X 22 W12 + 2 X 1 W21 − W12 )] (3.181)
[ (
RT ln γ 2 = X 12 W21 + 2 X 2 W12 − W21 . )] (3.182)
Figure 3.9D shows the form of the curve of the excess Gibbs free energy of mixing
obtained with Van Laar parameters variable with T: the mixture is subregular—
i.e., asymmetric over the binary compositional field.
3.10 Generalizations for Ternary Mixtures
So far, we have seen several ways of calculating the Gibbs free energy of a two-
component mixture. To extend calculations to ternary and higher-order mixtures,
we use empirical combinatory extensions of the binary properties. We summarize
here only some of the most popular approaches. An extended comparative ap-
praisal of the properties of ternary and higher-order mixtures can be found in
Barron (1976), Grover (1977), Hillert (1980), Bertrand et al. (1983), Acree (1984),
and Fei et al. (1986).
Because the ideal Gibbs free energy of mixing contribution is readily general-
ized to n-component systems (cf. eq. 3.131), the discussion involves only excess
terms.
3.10.1 Wohl Model

According to Wohl (1953), the excess Gibbs free energy of mixing for a three-
component mixture is represented by
(
G excess mixing = X 1 X 2 W12 X 2 + W21 X 1 )
( )
+ X 2 X 3 W23 X 3 + W32 X 2 + X 1 X 3 W13 X 3 + W31 X 1 ( ) (3.183)
1 
2
(
+ X 1 X 2 X 3  W12 + W21 + W13 + W31 + W23 + W32 ) − C ,

where X1 , X2 , and X3 are the molar fractions of components 1, 2, and 3 in the

mixture, W12 , W21 , W13 , W31 , W23 , and W32 are binary interaction parameters
between the components (see eq. 3.180), and C is a ternary interaction constant
whose value must be determined experimentally.
The activity coefficient of component 1 in a ternary mixture is defined by
[ (
RT ln γ 1 = X 22 W12 + 2 X 1 W21 − W12 )] + X [W2
3 13 (
+ 2 X 1 W31 − W13 )] +
1
( )
X 2 X 3  W12 + W21 + W13 + W31 − W23 − W32 + X 1 W21 − W12 + W31 − W13 +
2
( )
(X 2 )( ) (
− X 3 W23 − W32 − 1 − 2 X 1 C . ) ] ( 3.184 )
(To obtain the corresponding function for component 2, substitute 2 for 1, 3 for
2, and 1 for 3, and for component 3, substitute 3 for 1, 1 for 2, and 2 for 3.)
3.10.2 Hillert Model

According to the Hillert model (Hillert, 1980), the excess Gibbs free energy of
mixing of a ternary mixture is given by
 X2  
G excess mixing =  [ (
  X3  ×
  X 1* X 2* A012 + A112 X 1* − X 2*  + 
 1 − X1 

 1 − X1 
)]
 1 3 013
 [
X * X * A + A X * − X * 
113 1 3 

( )] (3.185)
[
+ X 2 X 3 A023 + A123 V23 − V32 , ( )]
where
V23 =
1
2
(
1 + X2 − X3 ) (3.186)
and
V32 =
1
2
(
1 + X3 − X2 . ) (3.187)
The A 012, A 013, A 023, A 112, A 113, and A 123 parameters in equation 3.185 are anal-
ogous to the Van Laar parameters in eq. 3.175 and represent the binary interac-
tions between components 1 and 2, 1 and 3, and 2 and 3, respectively. X1 , X2 ,
and X3 are the molar fractions of components 1, 2, and 3 in the mixture, and
X 1* , X 2* , and X 3* are the corresponding scaled fractions in the binary field.
3.10.3 Kohler Model

The Kohler model is a general model based on linear combination of the binary
interactions among the components in a mixture, calculated as if they were pres-
ent in binary combination (relative proportions) and then normalized to the ac-
tual molar concentrations in the multicomponent system. The generalized expres-
sion for the excess Gibbs free energy is
∑ ( Xi )
2
G excess mixing = + Xj ⋅ Gij , binary , (3.188)
i≠ j
where Gij,binary is the excess Gibbs free energy calculated for i and j components
as if they were in binary combination—i.e., for a two-component subregular Mar-
gules notation:
(
Gij , binary = Xi* X j* Wij X j* + Wji Xi* ) (3.189)
Xi
Xi* = (3.190)
Xi + X j
Xj
X *j = . (3.191)
Xi + X j
Consider, for instance, a three-component system (1, 2, 3). We have
( ) ( )
2 2
G excess mixing = X 1* + X 2* G12 , binary + X 1* + X 3* G13, binary
(3.192)
( )
2
+ X 2* + X 3* G23, binary
(
G12 , binary = X 1* X 2* W12 X 2* + W21 X 1* ) (3.193)
X1 X2
X 1* = ; X 2* = , (3.194,195)
X1 + X 2 X1 + X 2
and so on.
The application of multicomponent mixing models to the crystalline state is
a dangerous exercise. For instance, Fei et al. (1986) have shown that, adopting
the same binary interaction parameters between components in a mixture, exten-
sions to the ternary field through the Wohl and Hillert models do not lead to the
same results. Moreover, from a microscopic point of view, we cannot expect ter-
nary interactions to be linear combinations of binary terms simply because the
interionic potential field in a crystal is not a linear expression of the single proper-
ties of the interacting ions. We have seen (in chapter 1) how the lattice energy of a
crystalline substance is a complex function of individual ionic properties (effective
charge, repulsive radius, hardness factor, polarizability, etc.) and of structure (co-
ordination, site dimension, etc.)—even in a simple pair-potential static approach
(however, see also relationships between electron density and coulombic, kinetic,
and correlation energy terms in section 1.18). Undoubtedly, multicomponent
mixing models cannot be expected to have heuristic properties when applied to
the crystalline state, but must be simply adopted as parametric models to describe
mixing behavior (experimentally observed or calculated from first principles) in
multicomponent space in an easy-to-handle fashion (see, as an example, Otto-
nello, 1992).
3.11 Solvus and Spinodal Decomposition

in Binary Mixtures
In chapter 2 we defined equilibrium conditions for coexisting phases:
1. Equality of chemical potentials of the various components in all coexist-

ing phases
2. Minimization of the Gibbs free energy of the system
Let us now imagine that we are dealing with a regular mixture (A,B)N with an
interaction parameter W ⫽ ⫹20 kJ/mole. The Gibbs free energy of mixing at
various temperatures will be
G mixing = G ideal mixing + G excess mixing

(3.196)
(
= RT X A ln X A + XB ln XB )+X A XBW .
The Gibbs free energy of mixing curves will have the form shown in figure
3.10A. By application of the above principles valid at equilibrium conditions, we
deduce that the minimum Gibbs free energy of the system, at low T, will be
Figure 3.10 Solvus and spinodal decomposition fields in regular (B) and subregular (D)
mixtures. Gibbs free energy of mixing curves are plotted at various T conditions in upper
part of figure (A and C, respectively). The critical temperature of unmixing (or “consolute
temperature”) is the highest T at which unmixing takes place and, in a regular mixture
(B), is reached at the point of symmetry.
reached by the coexistence of two phases: one rich in component (A)N, and the
other rich in (B)N. The loci of tangency points (or “binodes”) at the various
values of T will define a compositional field, in a T-X space, within which the two
phases will coexist stably at equilibrium—the “solvus field of the solid mixture”
(figure 3.10B). In this case, the solvus is symmetrical with respect to composition.
Instead, if we have a solid mixture whose excess Gibbs free energy of mixing
is approximated by a subregular Margules model with WAB ⫽ 5 kJ/mole and
WBA ⫽ 35 kJ/mole, the Gibbs free energy of mixing at the various values of T is
G mixing = G ideal mixing + G excess mixing

(3.197)
( )
= RT X A ln X A + XB ln XB + X A XB WAB XB + WBA X A ( )
(figure 3.10C). The resulting solvus field is asymmetric (figure 3.10D).
Figure 3.10 also shows the fields of “spinodal decomposition”* defined by
the loci of the points of inflection in the Gibbs free energy of mixing curves. These
points obey the following general conditions:
 ∂ 2 G mixing 
  =0 (3.198)
 ∂X
2

and
 ∂ 3G mixing 
  = 0. (3.199)
 ∂X
3

Recalling that
G mixing = G ideal mixing + G excess mixing , (3.200)
we can write
∂ 2 G excess mixing ∂ 2 G ideal mixing RT

=− =−
∂X 2 ∂X 2 (
X 1− X ) (3.201)
and
∂ 3G excess mixing
=−
∂ 3G ideal mixing
=−
(
RT 2 X − 1 ).
(3.202)
∂X ∂X
(1 − X )
3 3 2
X2
* The term “spinode” was proposed by van der Waals, who formulated it in analogy with the
shape of a thorn (“spine”) given by the intersection of the tangent plane to the ␮ ⫽ f (P) function
for gaseous phases at the critical point (see figure 1 in van der Waals, 1890; see also Cahn, 1968, for
an extended discussion of the etymology of the term).
The critical unmixing condition is obtained from equations 3.201 and 3.202
by substituting for the left-side terms the algebraic form of the excess function
appropriate to the case under consideration. For a simple mixture, we have, for
instance (cf. eq. 3.144),
∂ 2 G excess mixing
= −2W I (3.203)
∂X 2
and
= 0, (3.204)
∂X 3
which, combined with equations 3.201 and 3.202, give
RT
2W I =
(
X 1− X ) (3.205)
and
(
RT 2 X − 1 ) = 0.
(3.206)
(1 − X )
2
X2
Because the simple mixture has symmetric properties, the system defined by equa-
tions 3.205 and 3.206 can be easily solved. Setting X ⫽ 0.5, we have
WI
T = Tconsolute = . (3.207)
2R
If the mixture is subregular, definition of the limits of spinodal decomposition

is more complex. For a subregular Margules model (figure 3.10C and D), we have
∂ 2 G excess mixing  
∂X 2 
1
2 
(
= − 4 W21 − W12  + 6 X W21 − W12 ) (3.208)
and
∂X 3
(
= 6 W21 − W12 ) (3.209)
Figure 3.11 Miscibility gaps in compositional field CaCO3–MgCO3. Reprinted from

J. R. Goldsmith and H. C. Heard (1961), Journal of Geology, 69, 45–74, copyright 䉷 1961
by The University of Chicago, with permission of The University of Chicago Press.
and, for a two-coefficient Van Laar model,
∂ 2 G excess mixing
∂X 2
(
= −2 A0 + 6 A1 2 X − 1) (3.210)
and
= 12 A1 . (3.211)
∂X 3
The combination of equations 3.208, 3.209 or 3.210, and 3.211 with general
equations 3.201 and 3.202 gives rise to transcendental equations that must be
solved using iterative procedures. In the treated cases, we have seen a single solvus
field that occupies a limited portion of the compositional field (figure 3.10B), and
an asymmetric solvus field that extends to the condition of pure component (fig-
ure 3.10D). In nature, there are also cases in which two or more solvi may be
observed over the same binary range. One example is the CaCO3–MgCO3 system
(figure 3.11). Actually, within the compositional field, we observe second-order
phase transitions (see section 2.8): the spatial group dolomite (Ca0.5Mg0.5CO3 ;
group R3 ) is a subgroup of calcite (CaCO3 ; group R3 c), and the definition “mis-
cibility gap” or even “pseudo-solvus” instead of “solvus field” would be prefer-
able (Navrotsky and Loucks, 1977). We note that the consolute temperatures for
the two miscibility gaps are markedly different; moreover, the gap at low Mg
content is strongly asymmetric.
3.12 Spinodal Decomposition and Critical

Nucleation Radius
Let us again consider a solid mixture (A,B)N with a solvus field similar to the
one outlined in the T-X plot in figure 3.10B, and let us analyze in detail the “form”
of the Gibbs free energy of mixing curve in the zone between the two binodes
(shaded area in figure 3.10A).
We can identify the binodal tangency point that marks the equality of poten-
tials for the coexisting phases at a given composition (Xb in figure 3.12A). We
also note that the Gibbs free energy of mixing curve, after the binode, increases
its slope up to an inflection point called the spinode (Xs ). Beyond the spinode,
the slope of the curve decreases progressively. If we now analyze the geometrical
relationships between composition and Gibbs free energy of mixing of two ge-
I II
neric points intermediate between binode and spinode (X (1 ) and X (1) in figure
3.12A), we see that the algebraic summation of the Gibbs free energies at the two
compositional points is always higher than the Gibbs free energy of the mean com-
position X(1)m. Based on the Gibbs free energy minimization principle, a generic
mixture of composition X(1)m apparently will be unable to unmix in two phases,
because any compositional fluctuation in the neighborhood of X(1)m would result
in an increase in the Gibbs free energy of the system. This apparently contrasts
with the statements of the preceding paragraphs—i.e., within the solvus field, the
minimum Gibbs free energy condition is reached with splitting in two phases.
Indeed, if we analyze the bulk Gibbs free energy of the system when unmixing is
completed, we see that it is always lower than the corresponding energy of the
homogeneous phase. The increase in energy related to compositional fluctuations
in the zone between binode Xb and spinode Xs must be conceived of as an “energy
threshold of activation of the unmixing process” that must be overcome if the
process is to terminate. The spinodal field defined by the inflection points in the
Gibbs free energy of mixing curve thus represents a “kinetic limit” and not a
phase boundary in the proper sense of the term.
The Gibbs free energy variation connected with compositional fluctuations
in the vicinity of X(1)m (figure 3.12B) peaks near the so-called “critical nucleation
radius” (rc ) and then decreases progressively with the increase of (compositional)
distance. Let us analyze the significance of rc from a microscopic point of view.
A generic mixture (A,B)N in which AN is present at (macroscopic) concentration
X(1)m and BN is at a concentration of (1 ⫺ X(1)m ) may exhibit compositional
fluctuations around X(1)m, on the microscopic scale. If these fluctuations are such
that ⌬XAN at X(1)m is higher than ⌬X rc, the energy threshold of the process will
Figure 3.12 Energy relationships between solvus and spinodal decompositions. (A) Por-
tion of Gibbs free energy of mixing curve in zone between binodal (Xb) and spinodal (Xs)
points. (B) Gibbs free energy variation as a consequence of compositional fluctuations
around intermediate points X(1)m and X(2)m.
be overcome; a nucleus of composition (A X(1 )m ⫹ ⌬X )N with ⌬X ⬎ ⌬X rc will be

formed and “homogeneous nucleation” will take place. The process will be one
of “heterogeneous nucleation” if the compositional step necessary to overcome
the critical nucleation is favored by the presence of extended defects, such as limits
of grains and/or dislocations. If we consider the equation relating composition
and Gibbs free energy in the compositional field between the two spinodes, any
compositional fluctuation in the vicinity of X(2)m will result in a decrease in the
Gibbs free energy of the system (figure 3.12A and B): the mixture is therefore
intrinsically unstable in this compositional range, and will tend to decompose
spontaneously. The compositional region between the solvus and the spinodal
fields is thus one of metastability, whereas the region within the spinodes is one
of intrinsic instability.
For analytical comprehension of the kinetics of spinodal decomposition processes,

we must be able to evaluate the Gibbs free energy of a binary mixture of nonuniform
composition. According to Cahn and Hilliard (1958), this energy can be expressed by the
linear approximation
 g(X ) + K ( ∇X )2  dV ,
G= ∫   (3.212)
V
where the first term, g(X ), represents the Gibbs free energy of an infinitesimal volume of
mixture with uniform composition, and the second term, K(∇X )2, defines a local energy
gradient that is always positive because K is inherently positive. The expansion of g(X ) in
the vicinity of a mean composition X 0 is given by the following expansion truncated at
the third term:
 dg  2  d 2g 
g ( X ) = g ( X 0 ) + ( X − X 0 )  + (X − X 0 ) 
1
 + L. (3.213)
 dX  X 2  dX 2  X
0 0
The Gibbs free energy difference between the system with compositional fluctuations
and an analogous homogeneous system of composition X 0 is given by
 ∆g(X ) + K ( ∇X )2  dV ,
∆G = ∫   (3.214)
V
where
∆g(X ) = g(X ) − g(X 0 ) . (3.215)
Compositional fluctuations may be represented through a Fourier expansion, and are

sufficiently approximated (for our interests) by the first sinusoidal term
X − X 0 = A sin βz , (3.216)
where ␤z is the scalar component of the wave vector ␤ along the decomposition direction
z, and A is a constant term. The Gibbs free energy variation due to periodic fluctuation is
A2  d 2 g  
∆G =  2 
+ 2 Kβ 2  . (3.217)
4  dX  
Because K is inherently positive, the stability conditions of the mixture with compositional
fluctuations can be deduced by the value of the differential term d 2g/dX 2. The spinodal
locus is defined by d 2g/dX 2 ⫽ 0; if d 2g/dX 2 ⬎ 0, then ⌬G ⬎ 0, and the mixture is stable
with respect to infinitesimal fluctuations (i.e., region between binodes and spinodes, point
X(1) in figure 3.12A). However, if d 2g/dX2 ⬍ 0, then ⌬G ⬍ 0 for compositional fluctuations
of wavelength overcoming the critical value
12
 
2π 8π 2 K
λ critical = = − 2  .
β critical

(
 d g dX 2 ) X0


(3.218)
Because there is no energy barrier opposed to compositional fluctuations for com-

positions within the spinodal field, the decomposition mechanism is entirely a diffusional
process. Unmixing effectively begins with a minimal compositional fluctuation that ex-
tends over the entire volume of the system.
The flow of matter as a function of the local chemical potential along a given direction
is described by the Cahn (1968) equation
 d ( µ A − µB ) 
JA − JB = M  , (3.219)
 dX 
Figure 3.13 Schematic evolution of concentration profiles in decomposition processes.

Upper part: nucleation and growth. Lower part: spinodal decomposition. Reprinted from
Cahn (1968), with kind permission of ASM International, Materials Park, Ohio.
where ␮A and ␮B are the chemical potentials of components A and B, JA and JB are their
fluxes, and M is the atomic mobility per unit volume. The diffusion equation along z is
dX dJ  d 2g   d 2X   d 4X 
=− = M    − 2 MK  4 , (3.220)
dt dz  dX 2   dz 2   dz 
whereJ is the interdiffusional flux (J ⫽ JA ⫽ ⫺JB ) (note that eq. 3.220 can be generalized
to all directions by substituting the ∇ 2X operator for the second derivative in dz; see eq.
8 in Cahn, 1968).
Because mobility term M is always positive, the diffusion coefficient M(d 2g/dX 2 ) de-
pends on the sign of the second derivative of the Gibbs free energy. Within the spinodal
field, (d 2g/dX 2 ) is negative and the diffusion coefficient is thus also negative. This means
that there is “uphill” diffusion: the atoms of a given species migrate from low to high
concentration zones. The final result is a clustering of atoms of the same species. Neverthe-
less, if we are in the compositional zone between spinode and binode, (d 2g/dX 2 ) is posi-
tive, the diffusion coefficient is then also positive, and diffusion takes place from high to
low concentration zones (nucleation and growth). The two different accretion processes are
shown in figure 3.13 as a function of time (and distance).
Let us now consider the process from a thermal point of view (figure 3.14)
and imagine freezing the system from point p to point p⬘ (corresponding to tem-
perature T⬘): we begin to see unmixing phenomena in p⬘. Actually it may happen
that the mixture remains metastably homogeneous (i.e., the critical nucleation
radius is not overcome) down to a point p⬙ (temperature T⬙) at which we effec-
tively observe decomposition processes taking place (at p⬙ the system enters the
Figure 3.14 Stability relations in a binary mixture (A,B)N as a function of temperature.

Heavy, solid line: activity trend for component (B)N in the case of binodal decomposition.
Dashed line: activity trend in the case of spinodal decomposition.
spinodal field). In terms of thermodynamic activity of components, we observe a

progressive increase in the activity of component (B)N in mixture, with an in-
crease in its molar concentration, in the metastability region between the solvus
and spinodal decomposition fields, and a progressive decrease (due to decomposi-
tion) within the spinodal field. There is therefore a maximum and a minimum in
the transition zone between metastability and instability. Nevertheless, if decom-
position already takes place in p⬘ (at the limits of the solvus field), the activity of
component (B)N remains constant throughout the compositional range lying—
at that temperature—within the solvus field (figure 3.14).
3.13 Chemical Solvus and Coherent Solvus
The exsolution process in either binodal or spinodal decomposition may lead to

coherent or incoherent interfaces between the unmixed phases (figure 3.15). The
Figure 3.15 Types of interfaces between exsolving phases.
Figure 3.16 Solvus and spinodal decomposition fields as a function of elastic strain.
(a) “Strain-free” or “chemical” solvus; (b) strain-free spinodal; (c) “coherent” solvus;
(d) coherent spinodal. From Ganguly and Saxena (1992). Reprinted with permission of
Springer-Verlag, New York.
representation of the Gibbs free energy of a binary mixture, as a continuous func-

tion of composition, requires coherence between the unmixed phases. This coher-
ence is obtained by adding to the Gibbs free energy of the mixture an elastic
strain term, represented by
η2 E y
(X − X )
2
∆g elastic = ,
(1 − v ) 0 (3.221)
where ␩ is linear expansion per unit of compositional variation, Ey is Young’s

modulus for X 0 , and v is Poisson’s ratio. The Gibbs free energy modification due
to periodic fluctuation (eq. 3.217) is thus modified to include this elastic strain
term:
A2  d 2 g  η2 E y 
∆G =   + 2 Kβ 2
+ .
4  dX 2  1− v( ) 
(3.222)
The energy of elastic strain modifies the Gibbs free energy curve of the mixture,
and the general result is that, in the presence of elastic strain, both solvus and
spinodal decomposition fields are translated, pressure and composition being
equal, to a lower temperature, as shown in figure 3.16.
CHAPTER FOUR
Some Concepts of Defect Chemistry
When investigating the thermochemical properties of crystalline substances, we,

more or less consciously, adopt an idealization of the phase of interest: we imagine
a crystal as being composed of a regular network of atoms, with interionic dis-
tances dictated by the symmetry operations of the spatial group. In this idealized
picture, a crystal is an unperturbed tridimensional replication of the asymmetric
unit. Actually, this hypothetical homogeneity-structural continuity is, in real
phases, interrupted by local irregularities called lattice defects. Lattice defects are
furthermore subdivided into two main groups: extended defects and point defects.
4.1 Extended Defects
Extended defects are primarily composed of linear dislocations, shear planes, and
intergrowth phenomena. Figure 4.1A and B, for example, show two types of lin-
ear dislocation: an edge dislocation and a screw dislocation.
Extended defects interrupt the continuity of the crystal, generating crystal
subgrains whose dimensions depend, in a complex fashion, on the density of
extended defects per unit area. Table 4.1 gives examples of reported dislocation
densities and subgrain dimensions in olivine crystals from the San Carlos perido-
tite nodules (Australia). Assuming a mean dislocation density within 1.2 ⫻ 105
and 6 ⫻ 105 cm⫺2, Kirby and Wegner (1978) deduced that a directional strain
pressure of 35 to 75 bar acted on the crystals prior to their transport to the surface
by the enclosing lavas.
The presence of extended defects does not significantly affect the energetic
properties of crystalline compounds, because the concentration of extended de-
fects necessary to create significant energy modifications is too high. For instance,
Wriedt and Darken (1965) observed deviations in the solubility of nitrogen in
steels induced by an extremely high dislocation density: 1011 cm⫺2 was the limit
at which the solubility modification could be measured. Such densities are not
normally observed in natural phases, whose dislocation densities normally range
between 104 and 109. Kohlstedt and Vander Sande (1973) observed dislocation
/ 185
Figure 4.1 Examples of extended defects in crystals: edge dislocation (A) and screw dis-
location (B).
Table 4.1 Dislocation densities (cm⫺2) and mean subgrain dimensions (cm) in San Carlos
olivines (from Kirby and Wegner, 1978). n.d. ⫽ not determined.
Mean Subgrain Dimension Dislocation Density
Sample [100] [010] [001] Surficial Internal
(1) 0.07 ⫾ 0.04 – 0.22 ⫾ 0.08 5.85 ⫻ 10 6

5 ⫾ 4 ⫻ 105
(2) 0.03 ⫾ 0.02 – 0.05 ⫾ 0.02 2.74 ⫻ 106 1.2 ⫾ 0.4 ⫻ 105
(3) 0.26 ⫾ 0.02 – 0.04 ⫾ 0.03 n.d. n.d.
(4) 0.02 ⫾ 0.01 – 0.02 ⫾ 0.01 n.d. 3 ⫾ 2 ⫻ 105
(5) 0.02 ⫾ 0.01 0.02 ⫾ 0.01 – n.d. 6 ⫾ 3 ⫻ 105
(6) – 0.02 ⫾ 0.01 0.02 ⫾ 0.01 n.d. 1.3 ⫾ 0.3 ⫻ 105
Some Concepts of Defect Chemistry / 187
densities of 106 to 107 cm⫺2 in natural orthopyroxenes, and dislocation densities

of 105 to 109 cm⫺2 have been observed in olivine (Kohlstedt and Goetze, 1974;
Buiskool Toxopeus and Boland, 1976; Kirby and Wegner, 1978).
Anomalously high extended defect concentrations may be achieved in crystals
by submitting them to directional stress. For example, Willaime and Gaudais
(1977) induced dislocation densities of 109 cm⫺2 in sanidine crystals (KAlSi3O8 ;
triclinic), and Ardell et al. (1973) reached a dislocation density of 1012 cm⫺2 in
quartz with the same method.
Extended defects such as dislocations cannot exist in the crystal at thermody-
namic equilibrium (Friedel, 1964), and their energy contribution must be evalu-
ated through the elastic theory. Calculations in this sense (Holder and Granato,
1969) lead to energies between 1 and 22 eV per atomic unit length. At a defect
energy of about 20 eV and a dislocation density of 109 cm⫺2, the resulting energy
is about 50 J/mole. This is lower than experimental error in most experimental
measurements of crystal energetics, and may virtually be neglected in calculations
of both mixing properties and intercrystalline equilibria.
4.2 Point Defects
Point defects exist in crystals at equilibrium conditions. They primarily affect some
chemical-physical features of crystalline substances, such as electronic and ionic
conductivity, diffusivity, light absorption and light emission, etc., and, in subordi-
nate fashion, their thermodynamic properties. As already noted, our idealized
picture of a crystal has little to do with reality. As extended defects interrupt the
symmetry of crystals, point defects modify their chemistry significantly, inducing
defect equilibria with coexisting phases and modifying stoichiometry. The disor-
der induced by the presence of point defects is grouped into two main categories:
intrinsic disorder and extrinsic disorder.
4.3 Intrinsic Disorder
Intrinsic disorder is observed in conditions of perfect stoichiometry of the crystal.

It is related to two main defect equilibria: Schottky defects and Frenkel defects.
Let us consider a generic oxide MO; in conditions of perfect stoichiometry,
one mole of MO is composed of one mole of M and half a mole of O2 . If a cation
is subtracted from a cationic position, creating a doubly ionized cationic vacancy
||
(V M ), then, in order to maintain crystal electroneutrality, one oxygen ion must
be subtracted from a lattice site pertaining to the anion, thus creating an anionic
vacancy (VO··). Using the Kröger-Vink notation, generally adopted to describe
point defect equilibria, the process, defined as “Schottky defectuality,” is repre-
sented as follows:
MO → VM

+ VO⋅⋅ (4.1)
or
×
MM + O O× → VM

+ VO⋅⋅ (4.2)
or
0 → VM

+ VO⋅⋅ , (4.3)
where the superscript ⫻ indicates neutrality relative to the ideal crystal, | stands
for negative excess charge taken by the cation vacancy, · stands for a positive
excess charge taken by the oxygen vacancy, and the subscripts indicate the lattice
position of the defect.
Let us now imagine moving a generic cation M from its normal position M
(M ⫻
M ) to an interstitial position i, leaving the previously occupied cationic site
empty:
×
MM → M ⋅⋅i + VM

. (4.4)
The equilibrium thus established is a “Frenkel defect.” In both the Schottky and
Frenkel equilibria, the stoichiometry of the crystal is unaltered (figure 4.2). As-
suming that the thermodynamic activity of the various species obeys Raoult’s law,
thus corresponding to their molar concentrations (denoted hereafter by square
brackets), the constant of the Schottky process is reduced to
[ ][ ]
K S = VM

VO⋅⋅ , (4.5)
and that of the Frenkel defect is
Figure 4.2 Intrinsic Schottky (A) and Frenkel (B) defects in a MO crystal.
[ ][ ]
K F = M ⋅⋅i VM

(4.6)
(because the defect concentrations are extremely low—i.e., generally lower than
10⫺4 in terms of molar fractions—the activity of ionic species at their normal
positions is assumed to be 1).
4.4 Extrinsic Disorder
Let us imagine immersing the crystal MO in an atmosphere in which the partial

pressure of oxygen (P O2 ) is lower than the partial pressure of intrinsic stability
(i.e., the partial pressure of oxygen at which crystal MO is perfectly stoichiomet-
ric: P* O 2 ). We observe two distinct defect processes:
1
O O× → O 2 + VO⋅⋅ + 2e  (4.7)
2 (gas)
and
1
×
MM + O O× → O 2 + M ⋅⋅i + 2e . (4.8)
2 (gas)
The first process produces doubly ionized positive oxygen vacancies and electrons
(e|) and the second produces doubly ionized positive cation interstitials and electrons.
The equilibrium constants of the two processes are given by
[ ][ ]
K 7 = VO⋅⋅ e  PO1 2
2
2
(4.9)
and
[ ][ ]
K 8 = M ⋅⋅i e  PO1 2 .
2
2
(4.10)
Because the bulk crystal is electroneutral, we can apply the “electroneutrality

condition.” For process 4.7 we write
[ ] [ ]
2 VO⋅⋅ = e  (4.11)
(i.e., the molar concentration of electrons must be twice the molar concentration
of doubly ionized positive vacancies). From equations 4.9 and 4.11 we obtain
[e ] = 2
 13
K 71 3 PO−1 6
2 (4.12)
and
K 71 3 −1 6
[ ]
VO⋅⋅ = P .
41 3 O2
(4.13)
Applying the electroneutrality condition to the equilibrium 4.8, we have
[ ] [ ]
2 M ⋅⋅i = e  , (4.14)
which, combined with equation 4.10, gives
[e ] = 2
 13
K 81 3 PO−1 6
2 (4.15)
and
13
[M⋅⋅ ] = K4
i
8
13
PO−1 6 .
2
(4.16)
At PO2 higher than P*O 2 , we observe the formation of doubly ionized cationic
vacancies and positive electronic vacancies (usually defined as electron holes by
symbol h· ):
1
O2 → O O× + VM

+ 2h ⋅ . (4.17)
2 (gas)
The constant of equilibrium 4.17 is
[ ][ ]
h ⋅ PO−1 2
2
K 17 = VM

(4.18)
2
and defect concentrations are given by
[h ⋅ ] = 2 13 13 16
K 17 PO
2 (4.19)
and
13
[ ]

VM =
K 17
P1 6 .
41 3 O2
(4.20)
Figure 4.3 plots defect concentrations as a function of the partial pressure of

oxygen, ranging from low PO2 to intrinsic pressure P*O 2 to high PO 2 .
Figure 4.3 Defect concentration as a function of partial pressure of oxygen. Magnesium

Frenkel defects are assumed to predominate at stoichiometric composition. At low PO2,
electronic conductivity is of type n (negative charge transport) and proportional to P⫺1/6O2 .
At high PO2, electronic conductivity is of type p (positive charge transport) and propor-
tional to P1/6
O2 . From R. Morlotti and G. Ottonello, Solution of rare earth elements in
silicate solid phases: Henry’s law revisited in light of defect chemistry, Physics and Chemis-
try of Minerals, 8, 87–97, figure 1, copyright 䉷 1982 by Springer Verlag. Reprinted with
the permission of Springer-Verlag GmbH & Co. KG.
Diagrams such as that presented in figure 4.3 are commonly used in the inter-
pretation of the defects in crystalline substances. Based on the equations above,
the defect concentration depends on PO2 in an exponential fashion (proportional
to PO⫺12 / 6 and PO1/26 in low and high PO2 ranges, respectively; because the plot is
logarithmic, the exponent becomes the slope of the function). Outside the intrin-
sic stability range, the stoichiometry of the crystal, as we have seen, is altered by
exchanges with the surrounding atmosphere. However, the above defect scheme
is an oversimplification of the complex equilibria actually taking place in hetero-
geneous systems. Indeed, defect processes are not simply ruled by the partial pres-
sure of oxygen, but by the chemistry of all phases present in the system. For
example, in the presence of a silica-rich phase, forsterite (Mg2SiO4 ) can incorpo-
rate SiO2 in excess with respect to its normal stoichiometry through the defect
equilibrium
2 SiO 2 ⇔ 2 VMg

+ Si ⋅⋅⋅⋅
i + Si Si + 4 O O
× ×
(4.21)
or through the equilibrium
SiO 2 ⇔ 2 VMg

+ Si ×Si + 2 VO⋅⋅ + 2 O O× (4.22)
(cf. Smith and Stocker, 1975). Based on the experimental data of Pluschkell and
Engell (1968), the silica excess in this way reaches 2.4% at 1400 °C.
4.5 Point Impurities
Let us imagine equilibrating a fayalite crystal (Fe2SiO4 ) in an atmosphere suffi-

ciently oxidizing to allow a defect equilibrium of the following type to proceed
toward the right:
×
2 Fe Fe → VFe

+ 2 Fe ⋅Fe (4.23)
or consider that a limited amount of component Sm4(SiO4 )3 may dissolve into a

forsteritic olivine according to the process
Sm 4 (SiO 4 ) 3 + 4 Mg Mg
×
+ 2 Si ×Si + 8 O O× ⇔ 4 Sm ⋅Mg + 2 VM

+ SiO 2 (4.24)
(Morlotti and Ottonello, 1984). In both cases, the crystal maintains its electroneu-
trality, but locally we observe the presence of cations whose charge (3⫹ in this
case) is not that of the normal constituent (2⫹) and which, in the case of equilib-
rium 4.24, do not correspond to a major component. These “point impurities,”
normally present in the crystal at equilibrium, significantly affect the chemical-
physical properties of the phase, acting, for instance, as activators in light
emission processes or as centers of heterogeneous nucleation (see section 3.12).
Processes such as 4.24 also regulate the solubility of trace components in rock-
forming minerals; the implications of these processes in trace element geo-
chemistry are discussed in chapter 10.
4.6 Energy of Formation of Point Defects
We have seen that defect concentrations in crystals may be described with the aid
of electroneutral equilibria involving species of differing charges. We have also
seen that defect concentrations are related to the partial pressure of oxygen in the
surrounding atmosphere and to the chemistry of coexisting phases in a hetero-

geneous system. Combining equations of the same types as equations 4.12, 4.13,
4.15, 4.16, 4.19 and 4.20 with experimental data on the actual defect population
at a given temperature, we may derive the values of the constants of the various
processes and, through them, evaluate the Gibbs free energy involved in each
process.
Let us imagine, for instance, that we “measured” the defect population of a
generic oxide MO composed essentially of Schottky defects (such a measure
could be obtained, for instance, by thermogravimetry at various T and PO2 condi-
tions or by electronic and ionic conductivity measurements at various values of
T and PO2 ). We can then write
0 → VM

+ VO⋅⋅ (4.25)
and
K S = VM

[ ][ ]
VO⋅⋅ , (4.26)
where [V M ] ⫽ [VO··] is the concentration of vacant sites at temperature T. The

||
Gibbs free energy variation connected with the defect process is*
∆G S = − kT ln K S , (4.27)
where k is the Boltzmann constant (k ⫽ R/N 0 ).

An alternative (and probably more precise) method for evaluating defect en-
ergies is based on the calculation of lattice energy potentials.
Consider a Schottky equilibrium taking place in a forsterite crystal and in-
volving Mg2⫹ cations and O2⫺ anions, displacing them from their lattice positions
and transporting them to the surface of the crystal:
×
Mg Mg + O O× → VM

+ VO⋅⋅ + MgO (surface) . (4.28)
To subtract the cation Mg2⫹ from its lattice position in the crystal and to bring
it to the surface, we must work against the static potentials (coulombic plus repul-
sive plus dispersive) at the Mg site. In terms of energy, this work corresponds to
half the lattice contribution of Mg2⫹ (in the Mg site of interest—i.e., M1 or M2;
see section 5.2) to the bulk static energy of the phase (see also section 1.12):
1
∆EU ,i = − ( EC ,i + E R,i + E D,i ). (4.29)
2
* Because the energy is defined at the atom scale, the standard state is implicit in the definition
and the zero superscript is hereafter omitted.
Subtracting a charge in a given lattice position also contributes to the defect en-
ergy with an induced polarization term whose significance was described in sec-
tion 1.19:
1
∆E P = − eΦe . (4.30)
2
Table 4.2 lists defect energies calculated with the method described above in fayal-
ite and forsterite crystals. Note that the energies obtained are on the magnitude
of some eV—i.e., substantially lower than the energies connected with extended
defects. Note also that, ionic species being equal, the defect energies depend on
Table 4.2 Defect energies in forsterite and fayalite based on lattice energy calculations. I ⫽
ionization potential; E ⫽ electron affinity; Ed ⫽ dissociation energy for O2; ⌬H ⫽ enthalpy of
defect process (adapted from Ottonello et al., 1990).
Process ⌬EC ⌬ER ⌬ED ⌬CFSE ⌬EP I E Ed ⌬H
FAYALITE
(1) 22.50 ⫺4.89 0.89 0.31 ⫺15.31 - - - 3.50
(2) 24.60 ⫺4.81 0.87 0.28 ⫺14.84 - - - 6.10
(3) ⫺27.34 6.14 ⫺1.07 - 12.44 - 7.46 2.79 0.42
(4) ⫺27.32 7.11 ⫺1.15 - 12.65 - 7.46 2.79 1.54
(5) ⫺52.20 12.8 ⫺2.06 - 24.68 - 14.92 5.58 3.72
(6) 11.60 5.59 ⫺0.87 0.28 ⫺42.94 30.64 - - 4.30
(7) 15.80 5.75 ⫺0.72 0.22 ⫺42.00 30.64 - - 9.77
(8) ⫺18.67 ⫺6.70 1.01 0.62 ⫺7.66 30.64 - - ⫺0.76
(9) ⫺20.17 ⫺6.59 0.99 0.56 ⫺7.42 30.64 - - ⫺1.99
(10) ⫺1.07 - - - - - - - ⫺1.07
(11) ⫺1.03 - - - - - - - ⫺1.03
FORSTERITE
(12) 23.44 ⫺4.59 0.11 - ⫺17.14 - - - 1.82
(13) 23.44 ⫺6.29 2.08 - ⫺17.14 - - - 2.09
(14) 25.16 ⫺4.46 0.13 - ⫺16.46 - - - 4.37
(15) 55.39 ⫺12.71 1.48 - ⫺26.22 - ⫺7.46 ⫺2.79 7.69
(16) 54.73 ⫺14.51 1.54 - ⫺26.22 - ⫺7.46 ⫺2.79 4.89
(17) 52.73 ⫺13.10 1.50 - ⫺26.00 - ⫺7.46 ⫺2.79 4.89
⫻
(1) Fe M1 → V||M1 ⫹ Fe2⫹ (surface)
⭈ ⫹ Fe |Si → {FeM1
(10) Fe M1 ⭈ Fe|Si}⫻
(2) Fe M2 → V||M2 ⫹ Fe2⫹
⫻ ⭈ ⫹ Fe|Si → {Fe M2
(11) Fe M2 ⭈ Fe|Si}⫻
(surface)
(3) 21 O2(gas) → O2⫺

(O1.surface)
(12) MgM1 → VM1 ⫹ Mg2⫹
⫻ ||
(surface)
(4) 21 O2(gas) → O2⫺

(O2.surface)
⫻
(13) FeM1 → V||M1 ⫹ Fe2⫹
(surface)
(5) 21 O2(gas) → O2⫺

(O3.surface)
(14) MgM2 → V||M2 ⫹ Mg2⫹
⫻
(surface)
⫻
(6) Fe M1 ⫹ h⭈ → V||M1 ⫹ Fe|Si ⫹ Si4⫹
(surface)
⫻
(15) OO1 ⫹ 2e| → V||O1 ⫹ 21O2(gas)
⫹ h⭈ → V||M2 ⫹ Fe|Si ⫹ Si 4⫹ ⫹ 2e| → V||O2 ⫹ 21O2(gas)
⫻
⫻
(7) Fe M2 (surface)
(16) OO2
⫹ h⭈ → FeM1 ⭈ ⫹ 2e| → V||O3 ⫹ 21O2(gas)
⫻
⫻
(8) Fe M1 (17) OO3
⫻
(9) Fe M2 ⫹ h⭈ → FeM2 ⭈
Table 4.3 Energy of Schottky and Frenkel equilibria in halides, oxides, and sulfides. (1)
Kröger (1964); (2) Barr and Liliard (1971); (3) Greenwood (1970).
⌬HD , ⌬HD , K 0,
Compound Process Calculated (eV) Observed (eV) Observed
AgBr (1) 0 → V|Ag ⫹ V⭈Br - 0.9-2.2 8000-20,000

AgBr (1) ⫻
Ag Ag → V|Ag ⫹ Ag⭈i - 0.87–1.27 157–1500
AgCl (1) Ag Ag → V|Ag ⫹ Ag⭈i
⫻
- 1.08–1.69 36–2300
AgI (1) ⫻
Ag Ag → V|Ag ⫹ Ag⭈i 0.6–0.8 0.69 -
CsCl (2) 0 → V|Cs ⫹ V⭈Cl - 1.87 -
CsBr (2) 0 → V|Cs ⫹ V⭈Br - 2.00 -
CsI (2) 0 → V|Cs ⫹ V⭈I - 1.91 -
CdS (1) 0 → V|Cd ⫹ V⭈S - 2.7 -
CdTe (1) CdCd⫻
→ V|Cd ⫹ Cd⭈i - 1.4 0.00017
KBr (1) 0 → V|K ⫹ V⭈Br 1.92 1.96–2.0 -
KCl (1) 0 → V|K ⫹ V⭈Cl 2.18–2.21 2.1–2.4 45
KI (2) 0 → V|K ⫹ V⭈I - 1.61 -
LiBr (1) 0 → V|Li ⫹ V⭈Br - 1.80 815
LiCl (1) 0 → V|Li ⫹ V⭈Cl - 2.12 1700
LiF (1) 0 → V|Li ⫹ V⭈F - 2.68 500
LiI (1) 0 → V|Li ⫹ V⭈I - 1.34 500
NaBr (1) 0 → V|Na ⫹ V⭈Br - 1.66–1.68 -
NaCl (1) 0 → V|Na ⫹ V⭈Cl 1.92–2.12 2.02–2.12 5.4
PbS (1) 0 → V|Pb ⫹ V⭈S - 1.75 0.27
ZnS (1) 0 → V||Zn ⫹ V ··
S 4–6 - -
MgO (3) 0 → V||Mg ⫹ V ··O - 4.77 -
SrO (1) 0 → V||Sr ⫹ V ··
O 3–5 - -
BaO (1) 0 → V||Ba ⫹ V ··
O 4.2 3.7 -
CaO (3) 0 → V||Ca ⫹ V ··
O 3–5 4.34 -
UO2 (1) OO⫻
→ V ··
O ⫹ Oi
||
- 3.44 -
the lattice positions of the defects themselves. For instance, creation of a cationic
|| ||
vacancy on site M (V M1 ) requires less energy than that necessary to create V M2.
Also, the amount of energy required to create an oxygen vacancy on site O3
·· ) is lower than that necessary to create VO1
(VO3 ·· or VO2
·· .
Table 4.2 lists energy values assigned to point impurities such as FeM1 and
FeM2 . In the calculation of their energies, an ionization term necessary to subtract
one electron from Fe2⫹ has also been added (ionization potential). With defect
notation, this process can be expressed as
⋅ ⋅
1 + h → Fe M 1 .
×
Fe M (4.31)
An alternative way of calculating defect energies on the basis of static poten-

tials is that outlined by Fumi and Tosi (1957) for alkali halides, in which the
energy of the Schottky process is seen as an algebraic summation of three terms:
Na ×Na → VNa

+ Na (gas)
+
( + 4.80 eV) (4.32)
×
Cl Cl ⋅ + Cl −
→ VCl ( +5.14 eV)
(gas) (4.33)
+
Na (gas) + Cl (gas)
−
→ Na ×Na + Cl Cl
×
( −7.82 eV) (4.34)
0 → VNa
 ⋅
+ VCl ( +2.12 eV) (4.35)
Note that in this case the energies of processes 4.32 and 4.33 are doubled with
respect to the energy necessary to bring the ion to the surface of the crystal;
moreover, the energy of process 4.34 is equal to the lattice energy of the crystal.
For comparative purposes, table 4.3 lists defect energies (enthalpies) of
Schottky and Frenkel processes in halides, oxides, and sulfides. The constant K 0
appearing in the table is the “preexponential factor” (see section 4.7) raised to a
power of 1/2.
4.7 Defect Concentration as a Function of

Temperature and Pressure
As we have seen, a given enthalpy term can be associated with each defect process
(tables 4.2 and 4.3). The defect process generally also involves an entropic term,
which may be quantified through
 ∆S 
exp  D  = g1 g 2 f , (4.36)
 k 
where g1 and g2 are the electronic degeneracies of the atoms participating in the
defect process and f is a vibrational term whose nature will be discussed later.
Because a defect entropy (⌬SD ), a defect enthalpy (⌬HD ), and a defect vol-
ume (⌬VD ) exist, the Gibbs free energy of the defect process is also defined:
∆GD = ∆HD − T∆SD + P∆VD . (4.37)
The enthalpic term ⌬HD is the dominant one in equation 4.37 and determines the
variation of the defectual concentration with T. For instance, for a Schottky de-
fect we can write
 ∆HS 
K S = K 0 exp  − , (4.38)
 kT 
in which the so-called “preexponential factor” K 0 is given by

 ∆S 
K 0 = exp  S  , (4.39)
 k 
where ⌬SS is the amount of entropy associated with the Schottky process.
If we assume, at first approximation, that ⌬HS does not change with T, it is
obvious that the increase in defect concentration with T is simply exponential.
Table 4.4, for instance, lists Schottky defect concentrations calculated in this way
at various T by Lasaga (1981c) for NaCl and MgO, assuming defect enthalpies
of 2.20 and 4.34 eV, respectively.
The approximation discussed above is in fact rather rough, and there are actually two
general cases.
1. A linear variation of the Gibbs free energy of the defect process with T, of type
∆GD = ∆GD0 + β ′T , (4.40)
in which ␤⬘ is the slope of the function. In this case, an Arrhenius plot of the
constant of the defect process results in a straight line:
 1
ln K D ∝   . (4.41)
T 
The preexponential factor also depends on T according to
 −β ′ 
K 0 = g1g2 f exp . (4.42)
 k 
Table 4.4 Schottky defect concentrations in NaCl and MgO at various T

(from Lasaga, 1981c).
NaCl MgO
| | ||
T (K) T (°C) [V Na] [V Na]/cm3 [V Mg] [V ||Mg]/cm3
0 ⫺273 0 0 0 0
25 298 2.5 ⫻ 10⫺19 5.6 ⫻ 103 2.1 ⫻ 10⫺37 1.1 ⫻ 10⫺14
200 473 1.9 ⫻ 10⫺12 4.2 ⫻ 1010 7.9 ⫻ 10⫺24 4.2 ⫻ 10⫺1
400 673 5.9 ⫻ 10⫺9 1.3 ⫻ 1014 5.8 ⫻ 10⫺17 3.1 ⫻ 106
600 873 4.5 ⫻ 10⫺7 1.0 ⫻ 1016 3.0 ⫻ 10⫺13 1.6 ⫻ 1010
800 1073 6.8 ⫻ 10⫺6 1.5 ⫻ 1017 6.5 ⫻ 10⫺11 3.5 ⫻ 1012
1000 1273 Melt 2.6 ⫻ 10⫺9 1.4 ⫻ 1014
1500 1773 Melt 6.9 ⫻ 10⫺7 3.7 ⫻ 1016
2000 2273 Melt 1.6 ⫻ 10⫺5 8.6 ⫻ 1017
2. A linear variation with T of the enthalpy of the process:
∆H D = ∆H D0 + βT . (4.43)
In this case, the constant of the process does not lead to a straight function in an
Arrhenius plot and the preexponential factor varies with T according to
K 0 = g1 g2 f T β k . (4.44)
Generally, nevertheless, the defect concentration tendentially increases with T.
Reexpressing the constant of the Schottky process as
∆GS
ln K S = − (4.45)
kT
and deriving in P, we obtain
 ∂ ln K S  1  ∂∆GS  ∆V
  =−   =− S , (4.46)
 ∂P  T ,X kT  ∂P  T ,X kT
where ⌬VS is the volume variation associated with the formation of the defect.
The value of ⌬VS is generally positive for Schottky defects in ionic crystals, be-
cause the lattice around the vacant sites undergoes relaxation due to static effects.
The corresponding ⌬VF is generally lower for Frenkel defects. It follows that the
P effect on the defect population is more marked for Schottky disorder than for
Frenkel disorder. Because ⌬VS is inherently positive, the defect population gener-
ally decreases with increasing P.
4.8 Associative Processes
Point defects of opposite charges tend to be reciprocally attracted, forming the

so-called associate defects.
Let us consider the general case of two point defects A and B forming an
associate defect (AB) according to the process
A + B → AB .( ) (4.47)
If we imagine that both defects A and B and associate defect (AB) are distributed
in conditions of complete configurational disorder, and if we substitute thermody-
namic activity for the molar concentration, we have
K A,B =
[( AB)] .
[ A ][B]
(4.48)
The Gibbs free energy modification due to associative processes (disregarding

volumetric effects) is
∆G AB = − kT ln K AB = ∆H AB − T∆SAB . (4.49)
The enthalpy change related to associative process (⌬HAB ) is due essentially to

coulombic interactions and, subordinately, to polarization, repulsion, covalent
bonding, elastic interactions, and vibrational effects. The latter two causes are
generally negligible and may have some effects only at low T.
The coulombic energy of associative processes is conveniently described by
q2
∆H AB,coulombic = − , (4.50)
ε sr
where q is the charge of interacting defects, r is the distance of shortest approach,

and ␧s is the static dielectric constant of the crystal.* Evaluation of the coulombic
interaction through equation 4.50 usually leads to overestimation of the energy
of the associative process (see table 4.5).
The entropy variation connected with associative processes (⌬SAB ) may be
ascribed to two different contributions: configurational and vibrational. The con-
figurational term is given by
Z
∆S AB,conf = k ln   , (4.51)
σ
where Z is the number of ways in which the associate may be formed and ␴ is a
symmetry factor. For an associate defect (AB), Z is the number of equivalent
sites that can be occupied by B at the shortest distance from A (or by A at the
shortest distance from B). For the same associate defect (AB), ␴ ⫽ 1, but becomes
2 for an associate dimer (AA), 3 for a trimer (AAA), and so on.
The vibrational entropy (⌬SAB,vibr ) contributing to the bulk entropy of defect
process arises from the variation of the vibrational spectrum of the crystal in the
neighborhood of the associate:
* Actually, as pointed out by Kröger (1964), at small distances the use of the static dielectric
constant based on a continuum approach to interaction energy is not justified, because one would
rather adopt the permittivity of a vacuum. However, using the static dielectric constant leads to
underestimation of the binding energy by 10 to 15%, counterbalancing the fact that short-range
repulsive forces are neglected.
∆S AB,vibr = k ln f (4.52)
v0, j
f = ∏ v ′j
, (4.53)
j
where v 0,j and v⬘j are vibrational frequencies related, respectively, to free and asso-
ciated defects (see Kröger, 1964, for a more detailed treatment).
For a pair of Schottky defects with, respectively, x and y first neighbors and v⬘ and v⬙
vibrational frequencies, vibrational term f has the form
x y
v  v 
f =  0  0 . (4.54)
 v ′   v ′′ 
For instance, for a Schottky defect pair in NaCl, we have
v0 v
= 0 = 2 (4.55)
v′ v ′′
x = y = 6, (4.56)
from which we obtain f ⫽ 4096. If the vibrational frequency of the defects increases as a
result of association, then f ⬎ 1, and the vibrational entropy also increases. The opposite
happens when f ⬍ 1.
In the case of a Frenkel defect pair, if vi is the frequency of interstitial atom i and v⬘i
is the vibrational frequency of the Z first neighbors surrounding it, we have
z x
 v  v   v 
f =  0  0   0  . (4.57)
 vi   vi′   v ′ 
Combining equations 4.49, 4.51, and 4.52, we obtain
Zf  ∆H AB 
K AB = exp − . (4.58)
σ  kT 
For the simple associate defect (AB), ␴ ⫽ 1 and the relative concentrations of
associate defect (AB) and free defects A and B are ruled by
[( AB)] + [ A] = [ A] total (4.59)
and
[( AB)] + [B] = [B] total

. (4.60)
Introducing fractions
β AB =
[( AB)] ; βA =
[A] ; βB =
[B] ,
[A] [A] [B]
(4.61)
total total total
we have
β AB  ∆H AB 
βA βB
= A [ ] total
Zf exp  −
 kT 
. (4.62)
For a generic defect concentration c, such as
c= A[ ] total
[ ]
= B
total
, (4.63)
at the point where half the defects are associated, we have
β AB
=1 (4.64)
β A βB
and
 ∆H AB 
  = ln c + ln Zf . (4.65)
 kT  0.5
Figure 4.4 plots relative concentrations of associate defect (AB) and free de-
fects A and B calculated as functions of ⌬HAB /kT for various values of c, with
Z ⫽ 4 and f ⫽ 1. Note that when the enthalpy change connected with the associa-
tion process is low (and/or T is high) all defects are practically free, whereas for
high (negative) values of ⌬HAB /kT all defects are associated. The dashed line in
figure 4.4 represents the equality condition outlined by equation 4.65.
Table 4.5 lists the energies of associative processes in halides and simple sul-
fides. As already mentioned, the enthalpy modification resulting from coulombic
interactions is generally higher with respect to the values resulting from a detailed
calculation. Table 4.5 also lists association energies involving point impurities. We
will see in chapter 10, when treating the stabilization of trace elements in crystals,
that association energy plays an important role in trace element distribution pro-
cesses.
Figure 4.4 Relative concentrations of free and associate defects for various values of total
defect concentration c, with Z ⫽ 4 and f ⫽ 1 (from Kröger, 1964; redrawn). Dashed line:
equality conditions in populations of free and associate defects.
4.9 Stability of a Crystalline Compound in the

Presence of Defect Equilibria: Fayalite as an Example
We have already noted that the concept of “stoichiometric crystal” is an extreme

idealization of an effectively more complex reality. In the presence of extrinsic
disorder, stoichiometry varies as a function of the chemistry of the coexisting
phases and of T and P. To clarify this concept better, the procedure developed by
Nakamura and Schmalzried (1983) to describe fayalite may be briefly recalled.
We define ␰ and ␩ as nonstoichiometry parameters describing the chemistry
of fayalite according to
n Si 1
ξ = − (4.66)
n Si + nFe 3
Table 4.5 Energies of associative processes in halides and simple sulfides. Values in eV
(from Kröger, 1964; modified).
⫺⌬HAB , ⫺⌬HAB ,
Associative Coulombic Total ⫺⌬HAB ,
Compound Process Calculated Calculated Experimental K0
AgBr Cd⭈Ag ⫹ VAg → (Cd⭈AgVAg)⫻

| |
0.22 - 0.16 -
AgCl Cd⭈Ag ⫹ V|Ag → (Cd⭈AgV|Ag)⫻ 0.12 - 0.18 -
AgCl V|Ag ⫹ V⭈Cl → (V|AgV⭈Cl)⫻ - - 0.43 -
CaF2 Y⭈Ca ⫹ F|i → (Y⭈CaF|i)⫻ - - 1.43 -
CaF2 Na|Ca ⫹ V⭈F → (Na|CaV⭈F)⫻ - - 0.07 -
KBr Ca⭈K ⫹ V|K → (Ca⭈KV|K)⫻ - - 0.56 -
KCl Ca⭈K ⫹ V|K → (Ca⭈KV|K)⫻ 0.69 0.32 0.52 0.6
KCl Cd⭈K ⫹ V|K → (Cd⭈KV|K)⫻ 0.69 0.32 - -
KCl Sr⭈K ⫹ V|K → (Sr⭈KV|K)⫻ 0.69 0.39 0.21-0.42 -
KCl V|K ⫹ V⭈Cl → (V|KV⭈Cl)⫻ 0.98 0.58–0.72 - -
LiF Mg⭈Li ⫹ V|Li → (Mg⭈LiV|Li)⫻ - - 0.7 -
NaCl Ca⭈Na ⫹ V|Na → (Ca⭈NaV|Na)⫻ 0.6 0.38 0.67 0.04
NaCl Cd⭈Na ⫹ V|Na → (Cd⭈NaV|Na)⫻ 0.6 0.38–0.44 0.3–0.34 -
NaCl Sr⭈Na ⫹ V|Na → (Sr⭈NaV|Na)⫻ 0.6 0.45 - -
NaCl Mn⭈Na ⫹ V|Na → (Mn⭈NaV|Na)⫻ - - 0.39–0.7 1.3
NaCl V|Na ⫹ V⭈Cl → (V|NaV⭈Cl)⫻ 0.91 0.44–0.60 - -
ZnS V||Zn ⫹ V ·· ·· ⫻
S → (VZnV S )
||
- 0.3–1.0 - -
and
nO 4
η= − , (4.67)
n Si + nFe 3
where the n-terms are the moles of the various elements. Obviously, in the absence
of extrinsic disorder, parameters ␰ and ␩ take on a value of zero. Figure 4.5A
shows the Gibbs phase triangle for the Fe-Si-O system. The (stoichiometric) com-
pound Fe2SiO4 , coexisting with SiO2 and Fe, with Fe and FeO, or with FeO and
Fe3O4 , is represented by a dot. Actually this scheme is not of general validity, and,
in the presence of extrinsic disorder, the compositional field of intrinsic stability is
expanded, as shown in figure 4.5B.
Based on thermogravimetric experiments on the compound Fe2SiO4 at vari-
ous T and P O2 conditions, Nakamura and Schmalzried (1983) established that
the extrinsic disorder of fayalite is conveniently represented by the equilibrium
×
×
10 Fe Fe + Si Si
×
+ 2 O 2(gas) ⇔ 3VFe

+ 6 Fe⋅Fe +  Fe Si

Fe⋅Fe  + FeSiO 4 . (4.68)
The defect scheme shown in equation 4.68 was later confirmed by electromotive
force measurements with galvanic cells (Simons, 1986) and by diffusivity mea-
Figure 4.5 Nonstoichiometry of compound Fe2SiO4 (fayalite) in the Fe-Si-O system. (A)
Usual representation of Fe2SiO4 as a discrete point in Gibbs space. (B) Defect equilibria
expand intrinsic stability field of phase through variation in nonstoichiometry parameters.
From A. Nakamura and H. Schmalzried, On the nonstoichiometry and point defects in
olivine, Physics and Chemistry of Minerals, 10, 27–37, figure 1, copyright 䉷 1983 by
Springer Verlag. Reprinted with the permission of Springer-Verlag GmbH & Co. KG.
surements (Brinkmann and Laqua, 1985) and is in reasonable agreement with

the defect energies reported in table 4.2.
As shown in figure 4.5B, extrinsic disorder (nonstoichiometry) may vary con-
siderably, as a function of solid paragenesis, between the low P O2 stability limit
defined by equilibrium
2 Fe + SiO2 + O2(gas) ⇔ Fe 2 SiO4 (4.69)
and the high P O2 stability limit defined by
3 Fe 2 SiO4 + O2(gas) ⇔ 2 Fe 3O 4 + 3 SiO2 . (4.70)

Table 4.6 shows the energy of extrinsic disorder calculated for the solid mixture
(Fe, Mg)2SiO4 at T ⫽ 1200 °C, based on the defect scheme of equation 4.68 and
on the defect energies of table 4.2.
We note that the defect energy contribution associated with extrinsic disorder
varies considerably as a function of the partial pressure of oxygen of the system.
These energy amounts may significantly affect the intracrystalline disorder, with
marked consequences on thermobarometric estimates based on intracrystalline
distribution. As we will see in detail in chapter 10, most of the apparent complexi-
ties affecting trace element distribution may also be solved by accurate evaluation
of the defect state of the phases.
4.10 Diffusion in Crystals: Atomistic Approach
Diffusion processes in crystals are markedly conditioned by point defects. The

presence of vacancies (or interstitials) allows ions to jump from one site to a
vacant neighboring position, leaving behind a vacancy: in this way an elemental
diffusion process is created (together with complementary vacancy migration) as a
summation of single displacement steps. During its movement from the occupied
to the vacant site, the ion performs a saddle-shaped trajectory, of the type exem-
plified in figure 4.6, to overcome the repulsive barriers of neighboring ions. Work
against the repulsive potential requires a certain amount of energy (mainly acces-
sible in thermal form). This energy, which is the only energy required during dis-
placement, is defined as “migration energy” (actually, not only the repulsive po-
tential but also all the static energy forms present in the crystal field play a role
in determining the amount of migration energy, although the repulsive term is
the dominant one).
Table 4.7 lists the migration enthalpies of some halides and simple oxides
Table 4.6 Energy contributions deriving from extrinsic

disorder of mixture (Fe,Mg)2SiO4 at T ⫽ 1200 °C. Values
expressed in J/mole; PO2 is in bar (from Ottonello et al.,
1990).
(1) (2)
XFe2SiO4 PO2 ⌬GD PO2 ⌬GD
0.2 1.96E-14 ⫺6.21 2.65E-4 ⫺401.80

0.3 4.41E-14 ⫺10.35 2.32E-5 ⫺377.72
0.4 7.85E-14 ⫺16.66 4.14E-6 ⫺397.71
0.5 1.23E-13 ⫺26.37 1.08E-6 ⫺445.34
0.6 1.77E-13 ⫺41.26 3.63E-7 ⫺517.86
0.7 2.40E-13 ⫺64.06 1.44E-7 ⫺617.18
(1) ⫽ maximum reduction limit (equilibrium 4.69)

(2) ⫽ maximum oxidation limit (equilibrium 4.70)
Figure 4.6 Migration of a cation M from an occupied site M⫻ ||

M to a vacant site V M in an
ionic crystal MO. Trajectory passes through centers of equilateral triangles O1-O4-O3 (X1)
and O2-O4-O3 (X2).
for vacancy migration and interstitial migration. Note that the required energies
are subordinate with respect to the energies of defect formation (cf. tables 4.3
and 4.7).
Let us now consider the crystal MO. If the diffusion takes place by migration
of cationic vacancies, the number of atoms that undergo the process depends on
||
the vacancy concentration [V M ] and the thermal state of single atoms M (the
jump takes place only whenever atom M in the neighborhood of the vacancy
has sufficient energy to perform it). The diffusion coefficient associated with the
vacancy migration process is given by
 H  
D=
d 2 v0
3
[ ]

VM exp  −
 kT
VM

 (4.71)
 
(Jost, 1960; Lasaga, 1981c), where d is the interionic distance between vacant and
occupied sites, and v 0 is a specific vibrational frequency of the crystal (Einstein
frequency, or a fraction of the Debye frequency; see Lasaga, 1981c).
A similar relation holds for anionic vacancies and for diffusion through inter-
||
stitial migration (replace [V M ] and HV M
||
by [VO··] and HVO·· , and so on).
Because at high T the vacancy concentration depends exponentially on T, i.e.
Table 4.7 Migration enthalpy in alkali halides and simple

oxides. Values in eV. Data from Barr and Liliard (1971) (1)
and Greenwood (1970) (2).
Compound HVM HVX HMi HXi
LiF 0.70(1) 0.65(1) - -

CaF2 - - - 1.65(1)
SrF2 - - - 1.00(1)
BaF2 - - - 0.78(1)
LiCl 0.39(1) - - -
NaCl 0.71(1) 1.00(1) - -
AgBr - - 0.17(1) -
KCl 0.71(1) 1.00(1) - -
KBr 0.65(1) 0.91(1) - -
NaBr 0.78(1) 1.17(1) - -
KI 1.21(1) - - -
AgCl - - 0.13(1) -
PbCl2 - 0.35(1) 1.56(2) -
PbBr2 - 0.30(1) - -
MgO 3.38(1) 2.69(2) - -
CaO 3.51(1) - - -
PbO 2.91(1) - - -
BeO 2.43(1) - - -
UO2 1.21(2) - - -
ZrO2 1.08(2) - - -
HVM ⫽ migration enthalpy of cationic vacancy

HVX ⫽ migration enthalpy of anionic vacancy
HMi ⫽ migration enthalpy of cationic interstitial
HXi ⫽ migration enthalpy of anionic interstitial
 ∆HS   ∆SS 
[V ] = K
M
 12
S = exp  −  exp 
 2 kT   2k 
, (4.72)
equation 4.71 can be reexpressed as
 Q 
D = D0 exp  − D  , (4.73)
 kT 
where QD represents the “bulk activation energy of the diffusion process,” com-
posed of the enthalpy of vacancy formation plus the enthalpy of vacancy mi-
gration:
∆H S
QD = + HVM
 , (4.74)
2
and D 0 groups the preexponential terms:

d 2 v0  ∆S 
D0 = exp  S  . (4.75)
3  2k 
Equation 4.75 finds its application in the region of intrinsic disorder (a similar
equation can be developed for Frenkel defects), where Schottky and Frenkel de-
fects are dominant with respect to point impurities and nonstoichiometry.
With decreasing temperature, as we have seen, the intrinsic defect population
decreases exponentially and, at low T, extrinsic disorder becomes dominant.
Moreover, extrinsic disorder for oxygen-based minerals (such as silicates and ox-
ides) is significantly affected by the partial pressure of oxygen in the system (see
section 4.4) and, in the region of intrinsic pressure, by the concentration of point
impurities. In this new region, term QD does not embody the enthalpy of defect
formation, but simply the enthalpy of migration of the defect—i.e.,
QD = HVM
 . (4.76)
In an Arrhenius plot of diffusivity, the transition from condition 4.74 to con-

dition 4.76 corresponds to a change of slope, which in turn corresponds to transi-
tion from intrinsically dominated defects (high T ) to extrinsically dominated de-
fects (low T ). Comparison of the slopes of the function in the two regions also
allows evaluation of the respective values of ⌬HS and HV M || (or ⌬H
F and HM ··
i
).
For instance, figure 4.7 shows diffusivity measurements in (Mg, Fe)2SiO4 olivine
mixtures, after Buening and Buseck (1973). We see a change in the slope at about
1125 °C, ascribable to the above transition. This transition is also clearly detect-
able in Co-Mg interdiffusion experiments (Morioka, 1980; T ⫽ 1300 °C) and in
Fe-Mg interdiffusion experiments (Misener, 1974; T ⫽ 1100 °C). T values being
equal, Buening and Buseck (1973) also observed a marked anisotropy of diffusion
as a function of crystallographic direction: diffusion was higher along axis c, indi-
cating preferential migration by vacancy diffusion on M1 sites, in agreement with
the energies of table 4.2 (the same anisotropy was observed in Ni diffusion by
Clark and Long, 1971 and Morioka, 1980 for Co-Mg interdiffusion, and by Mis-
ener, 1974 for Fe-Mg interdiffusion).
Lastly, diffusivity appears to increase progressively with the amount of Fe
in the system, in agreement with the defect scheme outlined in section 4.9 (see
equation 4.68).
Based on the data of Buening and Buseck (1973), the energy of diffusion
activation along axis c is 2.735 eV (corresponding to 264 kJ/mole) in the intrinsic
high-T region, and 1.378 eV (corresponding to 133 kJ/mole) in the extrinsic low-
T region.
Applying equation 4.75, we obtain an enthalpy of formation of Schottky de-
fects of 2.64 eV, which is somewhat higher than the value derived by Ottonello
et al. (1990) with static potential calculations (see table 4.2):
Figure 4.7 Arrhenius plot of cationic diffusivity in olivines. Reprinted with permission
from Buening and Buseck (1973), Journal of Geophysical Research, 78, 6852–6862, copy-
right 䉷 1973 by the American Geophysical Union.
×
Fe M 1 → VM 1 + Fe (surface)
 2+
( +2.09 eV ) . (4.77)
(To evaluate the energy of the Schottky process, based on the energies of table
4.2, the concomitant process of anion vacancy formation and the recombination
energies of cation and anion on the surface of the crystal must be added to pro-
cess 4.77.)
Not all crystalline compounds exhibit slope modifications similar to those
shown for olivine in figure 4.7. For instance, figure 4.8 shows diffusivity plots
for feldspars according to the revision of Yund (1983). The temperature ranges
investigated by various authors are sufficiently wide, but in any case we do not
observe any slope modifications imputable to modified defect regimes.
However, it must be emphasized that interpretation of elemental diffusion in
feldspars is complicated by the structural state of the polymorphs, which vary in a
complex fashion with temperature, chemistry and re-equilibration kinetics. These
complexities also account for the controversies existing in the literature regarding
diffusion energy in these phases (see also, incidentally, figure 4.8). Elemental
diffusivity data for rock-forming silicates are listed in table 4.8.
Whenever diffusivity rates are not experimentally known and cannot be esti-
mated by static potential calculations, approximate values can be obtained by
empirical methods. The most popular of these methods establishes a linear rela-
tionship between the enthalpy of the Schottky process (and enthalpy of migra-
tion) and the melting temperature of the substance (expressed in K):
Figure 4.8 Arrhenius plot of elemental diffusivity in feldspars. From Yund (1983). Re-
∆H S ≈ 2.14 × 10 −3 Tf ( eV ) (4.78)
and
∆H VM ≈ 0.014Tf ( eV ) . (4.79)
Equations 4.78 and 4.79 are valid for ionic solids. The analogous relations for
metals are
∆H S ≈ 7.5 × 10 − 4Tf ( eV ) (4.80)
and
∆H VM ≈ 6 × 10 − 4Tf ( eV ) . (4.81)
The significance of empirical equations 4.78, 4.79, 4.80, and 4.81 has not yet
been clarified in a satisfactory fashion. The enthalpy of Frenkel processes in met-
als can be related to the Debye temperature of the solid through
Table 4.8 Elemental diffusivity in silicates (from Lasaga, 1981c; Morioka and Nagasawa,
1991).
T-Range, D0, QD ,
Compound Element (°C) (cm2s⫺1) (eV) Reference
Mg2SiO4 FeM1 1125–1200 5.6 ⫻ 10⫺2 2.73 Buening and Buseck (1973)
Mg2SiO4 FeM1 1100–1125 5.9 ⫻ 10⫺7 1.38 Buening and Buseck (1973)
Mg2SiO4 FeM2 900–1000 1.1 ⫻ 10⫺2 2.61 Misener (1974)
Mg2SiO4 Ni 1250–1450 2.2 ⫻ 102 4.29 Morioka (1981)
Mg2SiO4 Ni 1149–1234 1.1 ⫻ 10⫺5 2.00 Clark and Long (1971)
Mg2SiO4 Co 1150–1400 2.8 ⫻ 10⫺1 3.24 Morioka (1980, 1981)
Mg2SiO4 Ca 1200–1400 7.3 ⫻ 101 4.31 Morioka (1981)
Mg2SiO4 MgM1 1300–1400 1.5 ⫻ 103 4.60 Morioka (1981)
Mg2SiO4 MgM2 1000–1150 1.8 ⫻ 10⫺8 1.48 Sockel and Hallwig (1977)
Mg2SiO4 OO1 1150–1600 1 ⫻ 10⫺4 3.32 Jaoul et al. (1980)
Mg2SiO4 OO2 1275–1628 3.5 ⫻ 10⫺3 3.86 Reddy et al. (1980)
Mg2SiO4 OO3 1472–1734 2.9 ⫻ 10⫺2 4.31 Ando et al. (1981)
Mg2SiO4 Si 1300–1700 1.5 ⫻ 10⫺6 3.94 Jaoul et al. (1981)
Fe2SiO4 MgM1 900–1100 1.5 ⫻ 10⫺2 2.16 Misener (1974)
Fe2SiO4 MgM1 1125–1200 2.9 ⫻ 10⫺2 2.55 Buening and Buseck (1973)
Fe2SiO4 MgM1 1000–1125 7.6 ⫻ 10⫺7 1.28 Buening and Buseck (1973)
Co2SiO4 Si 1200–1300 8.9 ⫻ 10⫺3 3.44 Schmaizried (1978)
KAlSi3O8 O 350–700 4.5 ⫻ 10⫺8 1.11 Giletti et al. (1978)
Adularia O 400–700 5.3 ⫻ 10⫺7 1.28 Yund and Anderson (1974)
Orthoclase Na 500–800 8.9 2.29 Foland (1974)
Orthoclase K 500–800 16.1 2.96 Foland (1974)
Orthoclase Rb 500–800 38 3.17 Foland (1974)
Orthoclase Sr 800–870 6 ⫻ 10⫺4 1.79 Misra and Venkatasubramanian
(1977)
Microcline O 400–700 2.8 ⫻ 10⫺6 1.28 Yund and Anderson (1974)
Microcline K 600–800 133.8 3.04 Lin and Yund (1972)
Microcline Sr 800–870 5 ⫻ 10⫺4 1.67 Misra and Venkatasubramanian
(1977)
NaAlSi3O8 O 350–800 2.3 ⫻ 10⫺9 0.92 Giletti at al. (1978)
NaAlSi3O8 O 600–800 2.5 ⫻ 10⫺5 1.61 Anderson and Kasper (1975)
NaAlSi3O8 Na 600–940 5 ⫻ 10⫺4 1.52 Bailey (1971)
NaAlSi3O8 Na 200–600 2.3 ⫻ 10⫺6 0.82 Lin and Yund (1972)
Biotite K 550–700 2.7 ⫻ 10⫺10 0.91 Hofman and Giletti (1970)
Anorthite O 350–800 1.4 ⫻ 10⫺7 1.14 Giletti et al. (1978)
12
 ∆H F 
θD = 34.3  , (4.82)
 MV 2 3 
where ⌬HF is in cal/mole. M is the molar weight and V is the molar volume of
the crystal (Mukherjee, 1965). Equation 4.82 is related to the preceding empirical
formulations through the Lindemann equation:
Tf ∝ θ D . (4.83)
Detailed experimental studies on rock-forming minerals (see Chakraborty

and Ganguly, 1991, and references therein) have shown that the pressure effect
on elemental diffusivity cannot be neglected in petrologic studies, in which highly
variable pressure regimes are commonly involved.
This effect can be analytically expressed by the partial derivative
∂ ln D ∆V +
=− , (4.84)
∂P kT
where ⌬V ⫹ is the activation volume of diffusion—i.e., “the volume change associ-

ated with raising the system from a potential energy minimum to the activated
state” (Ganguly and Saxena, 1987). Equation 4.84 implies that the activation
energy of the diffusion process is a function of pressure:
Q D( P ) = Q D( P
r)
(
+ ∆V + P − Pr , ) (4.85)
where Pr is the P condition of reference.

In light of equation 4.85, and adopting Pr ⫽ 1, equation 4.73 is more conve-
niently rewritten in the form
D( P )
( )
 −Q D + P − 1 ∆V +
= D0 exp 

. (4.86)
 kT 
 
Table 4.9 gives a summary of diffusion data for divalent cations in aluminous
garnets, with the relative activation volume. Although estimation of activation
volume is still largely uncertain, its evaluation is essential when dealing with the
wide baric regimes encountered in petrologic studies. As shown in the third col-
umn of table 4.9, the presence of ⌬V ⫹ implies substantial modifications of QD on
the kbar scale of pressure.
4.11 Diffusion and Interdiffusion
The diffusion coefficient for a given ion in a crystal is determined, as we have seen,
by the atomistic properties of the ion in the structural sites where the vacancy (or
interstitial) participating in the migration process is created (see eq. 4.71). The
units of diffusion (and/or “self-diffusion”) are usually cm2sec⫺1. Fick’s first law
relates the diffusion of a given ion A ( JA ) to the concentration gradient along a
given direction X:
 ∂C 
JA = − DA  A  , (4.87)
 ∂X  T ,P
Table 4.9 Summary of self-diffusion data for divalent cations in

aluminosilicate garnets (after Chakraborty and Ganguly, 1991; energy data
converted to eV scale, P in kbar). See Chakraborty and Ganguly, 1991, for
references.
Cation D0 (cm2s⫺1) ⌬V ⫹ (cm3mole⫺1) Q(P)(eV)
Mn2⫹ 5.15 ⫻ 10⫺4 6.04 ⫾ 2.93 2.626 ⫹ 0.0063P

Fe2⫹ 6.36 ⫻ 10⫺4 5.63 ⫾ 2.87 2.854 ⫹ 0.0059P
Mg2⫹a 1.11 ⫻ 10⫺3 5.27 ⫾ 2.96 2.948 ⫹ 0.0055P
Mg2⫹b 2.79 ⫻ 10⫺4 3.22 ⫾ 2.49 2.838 ⫹ 0.0034P
T range, 1100–1475 °C.

a
T range, 750–1475 °C.

b
where ( ∂CA / ∂X )T,P is the concentration gradient along X at temperature T and

pressure P, and JA is the flux of A in the same direction. Equation 4.87 can be
applied only at stationary state—i.e., whenever the concentration gradient does
not change with time. If it does change with time, its modification is described by
Fick’s second law:
 ∂ CA  ∂   ∂ CA  
  =  DA  . (4.88)
 ∂ t  T ,P ∂X   ∂X  
If there is simultaneous diffusion of more than one component in the crystal,

the flux of A in direction X depends on the individual diffusivities of all diffusing
components (Darken, 1948), and the individual diffusivity coefficient in equa-
tions 4.87 and 4.88 is replaced by interdiffusion coefficient D̃; i.e., for the simulta-
neous diffusion of two ions A and B,
 ∂C   ∂C 
JA = − D˜  A  ; JB = − D˜  B  (4.89)
 ∂X  T ,P  ∂X  T ,P
and
 ∂ CA  ∂   ∂ CA    ∂ CB  ∂   ∂ CB  
  =  D˜  ;   =  D˜    . (4.90)
 ∂ t  T ,P ∂X   ∂X    ∂ t  T ,P ∂X   ∂X  
Let us now consider the diffusion of two neutral species A and B in two
portions of metal slag welded at one interface (figure 4.9A). If the diffusivity of
A is higher than that of B, the portion on the right in figure 4.9A will tendentially
grow, whereas the portion on the left will become smaller (figure 4.9B). However,
if we keep the metal slag in a fixed position, the interface will be displaced to
the left (Kirkendall effect, figure 4.9C). The interdiffusion coefficient is related to
individual diffusivities according to
Figure 4.9 Kirkendall effect in a slag for metals A and B, with DA ⬎ DB. (A) Initial
configuration of diffusion couple. Position of interface fixed (dashed line) during diffusion
(accretion of right part and shrinking of left part of slag block). (C) Position of slag block
fixed during diffusion (interface moves leftward). From P. Haasen, Physical Metallurgy,
copyright 䉷 1978 by Cambridge University Press. Reproduced with modifications by per-
mission of Cambridge University Press.
D˜ = XA DB + X B DA . (4.91)
If, however, the diffusing species are electrically charged, the net flux at the inter-
face is effectively zero, to maintain the charge neutrality. Particularly, if A and B
are of the same charge, the flux of A will equal that of B:
∑ Ji = 0.
(4.92)
i
Imposing condition 4.92 on Fick’s first law results in
DA DB
D˜ = (4.93)
X A DA + XB DB
(see Ganguly and Saxena, 1987 for a more detailed derivation of equations 4.91
and 4.93).
Equation 4.90 is of particular interest when treating diffusion in nonmetallic
solids, where diffusion is primarily limited to ionic species. Note that, based on
equations 4.91 and 4.93, D̃ → DB when XB → 0. Thus, in a diluted binary mixture,
the interdiffusion coefficient assumes the same value as the diffusion coefficient
for the minor component.
In interdiffusion studies involving ions of the same charge, the dependence
of the single-ion diffusivity on concentration is obtained by means of annealing
experiments involving single oriented crystals of different compositions kept in
contact along a smooth surface. The interface coincides with the Boltzmann-
Matano plane—i.e., “a plane such that the gain of a diffusing species on one side
equals its loss on the other side” (Crank, 1975):
C0
∫ X dC = 0, (4.94)
0
where C 0 is the initial concentration in the diffusing couple and X is the distance
from the Boltzmann-Matano plane.
Figure 4.10 shows profiles of Mn2⫹ and Mg2⫹ distribution in the
Figure 4.10 Distribution profiles of Mn2⫹ and Mg2⫹ along axis c in the olivine mixture
(MnxMg1⫺x)2SiO4 after couple annealing at 1200 °C for 277 hours. From Morioka and
Nagasawa (1991). Reprinted with permission of Springer-Verlag, New York.
(MnxMg1⫺x )2SiO4 olivine mixture obtained by Morioka (1981). The concentra-

tion profiles are symmetric, indicating that the diffusivities are independent of the
concentration of the diffusing ion. On the contrary, possible asymmetry in the
diffusion profiles indicate that concentration depends significantly on the diffus-
ing cations. In this case, the interdiffusion coefficient can be obtained by the
Boltzmann-Matano equation:
C1
1 ∂X
D˜ C =C 1 = −
2t ∂C ∫ X dC , (4.95)
0
where D̃ C⫽C1 is the interdiffusion coefficient at concentration C1 intermediate be-

tween infinite dilution and initial concentration C 0 (0 ⱕ C1 ⱕ C 0 ), and t is the
duration of the diffusion run.
CHAPTER FIVE
Silicates
5.1 General Information
Qualitatively and quantitatively, compounds of silicon and oxygen are the class
of substances of greatest importance in the earth’s crust and mantle, in regard to
both mass and variety of structural forms.
In silicon and oxygen compounds, Si4⫹ has coordination IV or VI. The SiO44⫺
tetrahedron is energetically quite stable, with mixed ionic-covalent bonds.
The tetrahedral coordination of the [SiO4 ]4⫺ group could be regarded as a
direct consequence of the capability of the silicon atom to form four hybrid orbit-
als sp3, each directed toward one of the apexes of a regular tetrahedron, as shown
in figure 5.1. The energy gap between 3s and 3p orbitals at Z ⫽ 14 is limited, and
thus the combination is at first sight energetically plausible. Detailed quantum
mechanics calculations confirm that the tetrahedral symmetry corresponds to the
minimum energy for Si-O bonds.
“Hybrid orbitals” may be considered as “perfectioned AOs,” adopted in the calcula-

tion of localized MOs in polyatomic molecules, with the LCAO method (cf. section
1.17.1). In the case of hybrid orbitals sp3, the four linear combinations of s and p orbitals
(Te1 , Te2 , Te3 , Te4 ) that lead to tetrahedral symmetry are
Te1 =
1
2
(s + px + py + pz ) (5.1)
Te2 =
1
2
(
s + px − py − pz ) ( 5.2 )
Te3 =
1
2
(
s − px + py − pz ) ( 5.3 )
Te4 =
1
2
( )
s − px − py + pz . ( 5.4 )
/ 217
Figure 5.1 Tetrahedral hybrid orbitals sp3. (A) Section along development axis. (B) Spa-
tial form and orientation of hybrid orbital Te1 (from Harvey and Porter, 1976; redrawn
and reproduced with permission).
However, things are more complex when we look at the MO energy diagram (fig-
ure 5.2) developed by Tossell et al. (1973) for the [SiO4 ]4⫺ cluster on the basis of
a density functional molecular method (MS-SCF X␣: Multiple Scattering Self
Consistent Field). The orbital with the lowest binding energy is 1t1 , which is
essentially the nonbonding AO 2p of oxygen, followed in order of increasing
energy by the two 1e and 5t2 MOs with ␲ symmetry (⯝5 eV) formed by combina-
tion of the AOs 3s and 3p of silicon with the AO 2p of oxygen, and the 4t2 MO
(⯝7 eV) also formed by combination of the AO 3p of silicon with the AO 2p of
oxygen but with ␴ symmetry (high-energy orbitals in the inner valence region are
predominantly AO 2s in character).
Determination of electron density maps for the ␣-quartz polymorph estab-
lishes that the charge transfer between silicon and oxygen is not complete and
that a residual charge of ⫹1.0 (⫾0.1) electron units (e.u.) remains localized on
silicon, whereas a charge of ⫺0.5 (⫾0.1) e.u. is localized on each oxygen atom.
The interpretation of this fact in terms of the bond ionicity is not as univocal as
it may appear at first glance.
According to Pauling (1980), electron density maps confirm the fractional ionic char-
acter ( fi ⫽ 51%) deduced from electronegativity (see eq. 1.32 in section 1.7). Pauling’s
reasoning is as follows: the length of the Si-O bond in all compounds in which the [SiO4 ]4⫺
group shares the oxygen atoms with other tetrahedrally coordinated ions is 1.610 Å
(⫾0.003; cf. Baur, 1978). Because the observed Si-O distance is about 0.20 Å lower than
the sum of the single-bond radii and there are four Si-O bonds, the bond number 1.55 is
obtained and covalence is 6.20 e.u. The charge transfer from oxygen to silicon is thus 2.2
valence electrons (i.e., 6.2 to 4). Based on the net charge of ⫹1.0 e.u. on silicon, the frac-
tional ionic character is therefore
Silicates / 219
Figure 5.2 MO energy levels for the [SiO4]4⫺ cluster based on MS-SCF X␣ calculations.
Reprinted from Tossell et al. (1973), with permission of Elsevier Science B.V.
1 + 2.2
fi = = 0.5161 ≈ 51.6%, ( 5.5)
6.2
which is in good agreement with the value of 51% deduced from electronegativity.
However, starting from the same experimental evidence, Stewart et al. (1980) adopt
the following definition of fractional ionic character:
100 × residual charge 100 × 1

fi = = = 25. ( 5.6 )
formal charge 4
According to Stewart et al. (1980), the fractional ionic character thus defined is in
agreement with the value deduced from electronegativity theory for molecular orbitals.
The energy of a generic atom A is defined as
WA = a A + bAnA + cAnA2 , ( 5.7 )
where aA, bA, and cA are constants valid for atom A, and nA is the number of electrons in
the orbital: 0, 1, or 2. Constants bA and cA are related to ionization potential I and to the
electron affinity E:
bA = I A ( 5.8 )
cA =
1
4
(EA − I A ) . ( 5.9 )
The definition of Iczkowski and Margrave (1961) for electronegativity (see section 1.7)
gives
∂WA
χA = = bA + 2cAnA . ( 5.10 )
∂n
According to the treatment of Hinze et al. (1963), the charge transfer obeys the equal-
ity principle outlined in equation 1.31 (section 1.7). The transfer of fractional charges nA
and nB for two atoms A and B results in a stable bond with a fractional ionic character:
1  χ A − χB 
fi = , ( 5.11)
2  cA − cB 
where ␹A and ␹B are the initial electronegativities of the AOs that form the bond. In the
case of ␣-quartz, applying the electronegativity values of orbitals ␹ A and ␹B and coeffi-
cients cA and cB, both known for states sp and sp2 and evaluable for hybrid orbitals sp3 on
the basis of bond angles (for the Si-O-Si bond between neighboring tetrahedral groups,
the bond angle is 143.68°; cf. Le Page and Donnay, 1976; Hinze and Jaffe, 1962), Stewart
et al. (1980) obtain
1  2.25 − 5.80 
fi =   = 0.2536, ( 5.12 )
2  −2.26 − 4.74 
which is in good agreement with the definition of electronegativity expressed by equa-

tion 5.6.
Based on the controversies outlined above, discussion on the degree of ionicity in
natural silicates appears highly conjectural and superfluous, because there is no agreement
even for building group [SiO4 ]4⫺. Therefore, no mention will be made hereafter of the
degree of ionicity (or covalence) in these compounds.
Disregarding the type of bond, the tetrahedral group [SiO4 ]4⫺ is usually repre-
sented in the conventional ways outlined in figure 5.3, with the oxygen atoms at
the corners and silicon at the center of the tetrahedron, and the structure of sili-
cates is normally described as a function of the relative arrangements of the
Silicates / 221
Figure 5.3 Conventional ways of representing group [SiO4]4⫺.
Table 5.1 Structural arrangements of groups [SiO4 ]4⫺ and [SiO6 ]8⫺;
x ⫽ known, 0 ⫽ not known (from Liebau, 1982).
[SiO4 ]4⫺ [SiO6 ]8⫺

Arrangement of Groups (Tetrahedral) (Octahedral)
Isolated x x
Corner sharing x x
Edge sharing x x
Face sharing 0 0
[SiO4 ]4⫺ groups. Besides coordination IV, silicon also assumes coordination VI
with oxygen. Although the number of phases containing [SiO6 ]8⫺ groups is lim-
ited (about 30), the importance of octahedral coordination becomes more and
more evident with the advance of studies on high-P compounds. Both groups,
[SiO4 ]4⫺ and [SiO6 ]8⫺, exhibit various structural arrangements (see table 5.1).
For [SiO4 ]4⫺ groups, corner sharing and presence in isolated groups are the
most common structural arrangements (only fibrous SiO2 seems to exhibit an
edge-sharing configuration, although this arrangement is rather suspect; S. Mer-
lino, personal communication). For [SiO6 ]8⫺ groups, the edge-sharing configura-
tion (as in stishovite, for instance) is as common as the corner-sharing ar-
rangement.
Several elements (generally those with high field strength) exhibit isomor-
phous substitution of silicon in the tetrahedron (Be2⫹, Al 3⫹, Fe3⫹, Ga3⫹, B3⫹,
Ge4⫹, Ti4⫹, P5⫹ ). The most important diadochy is that of aluminum, which forms
pentavalent tetrahedral groups [AlO4 ]5⫺. Tetrahedral groups with isomorphous
substitution at the center of the tetrahedron are usually designated [TO4 ].
The chemical classification of silicates is based on their “multiplicity” and
type of structural arrangement. It partly differs from the classification commonly
adopted in mineralogy (table 5.2).
In mineralogy, monosilicates are called “nesosilicates” and all other oligosili-
cates are called “sorosilicates”; moreover, the term “polysilicates” is replaced by
“inosilicates.”
The multiplicity of the groups establishes the number m of [TO4 ] groups that
Table 5.2 Chemical classification of silicates.
Oligosilicates: monosilicates, disilicates, trisilicates, tetrasilicates

Cyclosilicates: monocyclosilicates, dicyclosilicates
Polysilicates: monopolysilicates, dipolysilicates, tripolysilicates, tetrapolysilicates, pentapolysilicates
Phyllosilicates: monophyllosilicates, diphyllosilicates
Tectosilicates
Table 5.3 Kostov’s classification of silicates. A ⫽ axial; P ⫽

planar; I ⫽ isometric (from Liebau, 1982).
Classes
Cations XSi ⬎ 4 3 ⬎ XSi ⬎ 1 XSi ⬍ 1

Be, Al, Mg, (Fe) A-P-I A-P-I A-P-I
Zr, Ti, (Sn), Nb A-P-I A-P-I A-P-I
Ca, (REE), Mn, Ba A-P-I A-P-I A-P-I
Zn, Cu, Pb, U A-P-I A-P-I A-P-I
link to form multiple tetrahedra. Multiplicities m ⫽ 2, 3, 4, 8, 9, and 10 have been

observed. When several [TO4 ] groups are linked to form rings, the number of
mutually linked rings is defined as multiplicity: the maximum multiplicity ob-
served in this case is 2.
If [TO4 ] groups are linked to form chains, multiplicity is the number of linked
chains; the observed multiplicities in this case are m ⫽ 2, 3, 4, and 5. Analogously,
multiplicity is defined as the number of mutually linked sheets. If several [TO4 ]
groups are linked in a chain, periodicity ( p) is the number of groups that defines
the structural motive (i.e., after p groups, the chain is obtained by simple transla-
tion). Dimensionality of [TO4 ] groups is the degree of condensation of a crystal
structure. The condensation of an infinite number of tetrahedra may lead to infi-
nite unidimensional chains (d ⫽ 1), bidimensional sheets (d ⫽ 2), and tridimen-
sional lattices (d ⫽ 3). The dimensionality of an isolated [TO4 ] tetrahedron is
zero (d ⫽ 0).
The alternative classification of Kostov (1975) is based on the “degree of
silicification” (XSi ), expressed by the ratio
Si + Al
X Si = , (5.13)
M′
where M⬘ ⫽ M⫹ 2⫹ 3⫹ 4⫹
2 ; M ; M 0.666; M0.5 .
The so-called degree of silicification is inversely proportional to the Si:O
atomic ratio and represents the degree of condensation of the [TO4 ] groups. Three
classes with distinct silicification degrees are thus distinguished, and each class is
further subdivided into groups, as a function of the geochemical affinity of the
elements, and into subclasses, as a function of the morphology of the substance
Silicates / 223
(table 5.3). We refer readers to Liebau (1982) for a more exhaustive discussion
on classification schemes for silicates.
5.2 Olivines
Olivines are monosilicates (“nesosilicates” in the mineralogical classification) and

crystallize in the orthorhombic system (spatial group Pbnm). The structural ar-
rangement is shown in figure 5.4: isolated [SiO4 ]4⫺ groups (nesos means “island”)
are mutually linked by divalent cations. This arrangement gives rise to two non-
equivalent octahedral sites, usually designated M1 and M2, in which the cations
are located, respectively, between the bases of tetrahedra (M1 sites) and between
a basis and an edge of two neighboring tetrahedra (M2 sites). As figure 5.4A
shows, tetrahedral apexes are alternate in opposite directions; the dashed line in
the figure contours the elementary cell. The formula unit usually adopted for
olivine is M2SiO4 , with M as generic divalent cation. The elementary cell is com-
posed of four formula units. Figure 5.4B details the conformation of the two
octahedral sites. The M1 octahedron shares six of its 12 edges with the neigh-
boring polyhedra (four with octahedra and two with tetrahedra), whereas the M2
octahedron shares only three edges (two with octahedra and one with a tetra-
hedron). Because shared edges are usually shorter than nonshared edges, to mini-
mize repulsion between neighboring cations (Pauling’s third rule), the octahedral
symmetry in both sites is generally distorted (although the degree of distortion is
more evident for the M2 site).
Figure 5.4 (A) Olivine structure: dashed line contours elementary cell (axes c and b). (B)
Details of coordination state and distortion of M1-M2 sites; O(1), O(2), O(3) oxygens
occupy nonequivalent positions; distortion from perfect octahedral symmetry is more
marked for M2 site.
5.2.1 Chemistry and Occurrence

The chemistry of natural olivines is conveniently described by the four main com-
ponents Mg2SiO4 (forsterite), Fe2SiO4 (fayalite), Mn2SiO4 (tephroite), and
Ca2SiO4 (Ca-olivine), whose molar fractions generally sum up to more than 99%
of the total composition. To these major components must be added the minor
terms Ni2SiO4 , Co2SiO4 , Zn2SiO4 , and Cr2SiO4 .
Table 5.4 lists analyses of some natural specimens. In some instances, there
is a silicon deficiency with respect to the stoichiometric formula and diadochy of
Al 3⫹, Ti4⫹, and Fe3⫹ on the tetrahedral sites (cases 1, 2, 3, and 4). In other cases,
the amount of silicon is near the theoretical value and only divalent cations are
virtually present in the M1 and M2 positions (cases 5 and 6).
Olivine is the main mineral, in terms of mass and volume, of the earth’s upper mantle
(peridotites), where it is present as a mixture of the main components forsterite and fayal-
ite, in molar proportions of 85 to 95% and 15 to 5%, respectively. The intrinsic stability
limit of the mixture is assumed to coincide with the seismic discontinuity observed at a
depth of about 400 km, where olivine undergoes transition to the ␤- and ␥-spinel poly-
morphs (Ringwood, 1969). Olivine is also a primary phase of mafic rocks (basalts, gab-
bros, dolerites), where it occurs with amounts of forsterite ranging from Fo80 to Fo50 . Its
importance (in terms of relative abundance) decreases considerably in acidic rocks (gran-
ites, syenites, rhyolites, trachytes), where fayalite prevails. Monticellite (CaMgSi2O6 ) is
normally found as a constituent in Si-Mg marbles, where it is a marker of the sanidine
facies (and probably also of the granulitic facies) of thermal regional metamorphism
(Brown, 1970). Terms of the tephroite-fayalite mixture are commonly found in Fe-Mn ore
deposits associated with skarns and in metamorphosed manganiferous sediments.
The Ni2SiO4 content in natural olivines rarely exceeds 1% in mole (weight % of NiO
⬍ 0.5), and is usually directly correlated to the forsterite content (see, for instance, Davis
and Smith, 1993). The inverse correlation is observed for Mn2SiO4 (weight % of MnO ⬍
4). Figure 5.5 shows the relative distributions of Mn2SiO4 , Ni2SiO4 , and Ca2SiO4 in effu-
sive and plutonic rocks, based on the synthesis of Simkin and Smith (1970): marked con-
trol of the mineralization environment can be seen for Ca2SiO4 . This effect is less evident
for the manganous term and is virtually absent for Ni. The relationship between amount
of Ca2SiO4 and crystallization conditions may be imputed to the pressure effect on the
solvus field existing for the join Mg2SiO4-CaMgSiO4 (cf. Finnerty, 1977; Finnerty and
Boyd, 1978; see section 5.2.5).
Table 5.5 lists silicate components that are known to crystallize with the oliv-
ine structure, according to the compilation of Ganguly (1977). The pure compo-
nents Zn2SiO4 and Cr2SiO4 have never been synthesized in the laboratory, but are
known to form the mixtures (Mg,Zn)2SiO4 and (Mg,Cr)2SiO4 , in which they
reach maximum molar concentrations of 24% (Sarver and Hummel, 1962) and
5% (Matsui and Syono, 1968; Ghose and Wan, 1974), respectively. It must be
noted that the presence of chromium as a minor component of terrestrial olivines
is restricted to kimberlitic rocks, with a maximum content of 0.15 weight %
Cr2O3 , although greater abundances (⬎0.4% Cr2O3 ) have been observed in lunar
Silicates / 225
Table 5.4 Olivine major element compositions (in weight %). Samples occur in different
types of rocks: (1) ⫽ forsterite from a metamorphosed limestone: (2) ⫽ hortonolite from an
olivine gabbro: (3) ⫽ fayalite from a pantelleritic obsidian; (4) ⫽ fayalite from an Fe-gabbro;
(5) ⫽ forsterite from a cumulitic peridotite; (6) ⫽ forsterite from a tectonitic peridotite.
Samples (1) to (4) from Deer et al. (1983); sample (5) from Ottonello et al. (1979); sample (6)
from Piccardo and Ottonello (1978).
Samples
Oxide (1) (2) (3) (4) (5) (6)

SiO2 41.07 34.04 30.56 30.15 39.52 40.00
TiO2 0.05 0.43 0.72 0.20 - -
Al2O3 0.56 0.91 0.09 0.07 - -
Fe2O3 0.65 1.46 0.10 0.43 - -
FeO 3.78 40.37 60.81 65.02 13.27 11.11
MnO 0.23 0.68 3.43 1.01 - -
MgO 54.06 20.32 3.47 1.05 46.36 47.55
CaO - 0.81 1.13 2.18 - -
NiO - - - - 0.07 0.46
H2O(⫹) 0.05 0.09 - - - -
Total 100.45 99.11 100.31 100.11 99.22 99.12
Cations Ionic Fractions on a Four-Oxygens Basis
Si 0.979 0.990 0.996 1.002 0.992 0.997

Al 0.016 0.032 0.004 0.003 - -
Ti 0.001 0.009 0.018 0.005 - -
Fe3⫹ 0.012 0.032 0.004 0.011 - -
Total IV 1.008 1.063 1.022 1.021 0.992 0.997
Mg 1.920 0.881 0.169 0.052 1.735 1.766

Fe2⫹ 0.075 0.983 1.659 1.808 0.279 0.232
Ni - - - - 0.001 0.009
Mn 0.005 0.017 0.094 0.028 - -
Ca - 0.025 0.039 0.078 - -
Total VI 2.000 1.906 1.961 1.966 2.015 2.017
specimens. Because chromium is present in nature in the valence states Cr2⫹ and
Cr3⫹, but occurs in olivine only (or predominantly) in the Cr2⫹ form (Burns,
1975; Sutton et al. 1993), the paucity of Cr2SiO4 in terrestrial olivines must be
ascribed to the oxidation state of the system, which is generally too high to allow
the formation of an olivine compound with stoichiometry Cr2SiO4 . Lastly, it must
be noted that the synthesis of compounds of type LiLnSiO4 or NaLnSiO4 (in
which Ln is a generic rare earth with atomic number not lower than that of Ho)
apparently contrasts with natural evidence. Although microprobe analyses report
Figure 5.5 Relative distributions of minor components Mn2SiO4, Ni2SiO4, and Ca2SiO4
in fayalite-forsterite mixtures (weight %).
sufficient amounts of Na2O in natural olivines, the rare earth element (REE)
abundances are generally at ultratrace levels (i.e., REE ⬍ 1 ppm; see, for instance,
Ottonello, 1980; Frey, 1982) and the relative distribution of the various terms of
the series does not reveal any crystal chemical control.
5.2.2 Structural Properties

Table 5.6 lists the structural properties of the various olivine components ac-
cording to the compilation of Smyth and Bish (1988). Note the conspicuous vari-
Silicates / 227
Table 5.5 Olivine compounds, known and synthesized.
Component Name Component

Mg2SiO4 Forsterite CaCoSiO4
Fe2SiO4 Fayalite LiScSiO4
Ca2SiO4 Larnite LiYSiO4
Mn2SiO4 Tephroite LiLnSiO4 (Ln ⫽ Ho-Lu)
Co2SiO4 Co-olivine LiInSiO4
Ni2SiO4 Liebembergite NaYSiO4
CaMgSiO4 Monticellite NaLnSiO4 (Ln ⫽ Ho-Lu)
CaFeSiO4 Kirschsteinite
CaMnSiO4 Glaucocroite
Table 5.6 Structure of olivine compounds (from Smyth and Bish, 1988). Molar volume is
expressed in cm3/mole; molar weight in g/mole; density in g/cm3; cell edges and cell volume are in
Å and Å3, respectively.
Formula units per unit cell: 4

System: Orthorhombic
Class: mmm
Spatial group: Pbnm
Phase Forsterite Fayalite Monticellite Kirschsteinite

Component Mg2SiO4 Fe2SiO4 MgCaSiO4 FeCaSiO4
Molar weight 40.708 203.778 156.476 188.011
Density 3.227 4.402 3.040 3.965
Molar volume 43.603 46.290 51.472 47.415
Cell parameters:
a 4.7534 4.8195 4.822 4.844
b 10.1902 10.4788 11.108 10.577
c 5.9783 6.0873 6.382 6.146
Volume 289.58 307.42 41.84 314.89
Reference Fujino et al. (1981) Fujino et al. (1981) Onken (1965) Brown (1970)
Phase Larnite Tephroite Co-olivine Liebembergite

Component Ca2SiO4 Mn2SiO4 Co2SiO4 Ni2SiO4
Molar weight 172.244 201.960 209.950 209.503
Density 2.969 4.127 4.719 4.921
Molar volume 58.020 48.939 44.493 42.574
Cell parameters:
a 5.078 4.9023 4.7811 4.726
b 11.225 10.5964 10.2998 10.118
c 6.760 6.2567 6.0004 5.913
Volume 385.32 325.02 295.49 282.75
Reference Czaya (1971) Fujino et al. (1981) Brown (1970) Lager and Meagher (1978)
ation of density between larnite (2.969 g/cm3 ) and liebembergite (4.921 g/cm3 ),
which results from the fact that a low molecular weight is associated to the highest
molar volume (larnite) and vice versa (liebembergite).
Figure 5.6 plots the molar volumes of various olivine mixtures against the
cubes of the ionic radii of the major cations in VI-fold coordination with oxygen,
according to the compilation of Brown (1982). The figure has been partially modi-
fied to correct the volume of Ca2SiO4 on the basis of the data of Czaya (1971).
Based on figure 5.6, olivine compounds seem to obey the simple proportionality
established by Vegard’s rule (with the exception of the pure component larnite):
Vcell ∝ r 3 . ( 5.14 )
Brown (1970) calculated the following equation, which is valid for olivine
compounds (ionic radii of Shannon and Prewitt, 1969):
Vcell = 188.32r 3 + 220.17 ( Å ).

3
( 5.15)
Equation 5.15 is based on 13 values, has a correlation coefficient R2 ⫽ 0.99, and

does not include the cell volume of the pure compound Ca2SiO4 (see figure 5.6).
It can be opportunely modified to give molar volumes according to
× 10 −24
V 0
(
= 188.32r 3 + 220.17 )N 0
v
( cm 3
)
mole , ( 5.16 )
where N 0 is Avogadro’s number and v is the number of formula units per unit cell.
Equations 5.15 and 5.16 give only approximate values for the volumes of
olivine compounds. More accurate study of binary mixtures outlines important
deviations from ideal behavior, which result in slight curves in cell parameter vs.
composition plots, as shown in figure 5.7. The greatest deviations regard cell
edges and are particularly evident for the (Mg,Ni)2SiO4 mixture.
The volume properties of crystalline mixtures must be related to the crystal chemical
properties of the various cations that occupy the nonequivalent lattice sites in variable
proportions. This is particularly true for olivines, in which the relatively rigid [SiO4 ]4⫺
groups are isolated by M1 and M2 sites with distorted octahedral symmetry. To link the
various interionic distances to the properties of cations, the concept of ionic radius is in-
sufficient; it is preferable to adopt the concept of crystal radius (Tosi, 1964; see section
1.9). This concept, as we have already noted, is associated with the radial extension of the
ion in conjunction with its neighboring atoms. Experimental electron density maps for
olivines (Fujino et al., 1981) delineate well-defined minima (cf. figure 1.7) marking the
maximum radial extension (rmax ) of the neighboring ions:
rmax = min 4π r 2 ρ ( r ) , ( 5.17 )

Silicates / 229
Figure 5.6 Relationship between cell volume and cubed ionic radius of main cation in
VI-fold coordination with oxygen. From Brown (1982). Reprinted with permission of The
Mineralogical Society of America.
where ␳(r) is electron density at distance r. Equation 5.17 may be operatively translated
to the form
C1 C1
rmax = = , ( 5.18 )
Zeff Zn − S e
where C1 is a constant determined by the electronic configuration of the inner shells, and
effective nuclear charge Zeff is the actual nuclear charge minus screening constant Se (cf.
section 1.5).
The energy of the atomic orbitals in the hydrogenoid approximation is proportional
to the square of the effective charge, and is (numerically) equivalent to the ionization po-
tential
Ry hcz 2 2
ZeffC2
E( n ) = − 2
= , ( 5.19 )
n n2
Figure 5.7 Molar volumes and cell edges in binary olivine mixtures (from Akimoto et al., 1976; modified). Note conspicuous deviations from linearity
in cell edges of (Mg,Ni)2SiO4 mixture, satisfactorily reproduced by site-occupancy calculations based on parameters in table 5.7.
Silicates / 231
where R y is Rydberg’s constant, h is Planck’s constant, c is the speed of light, and z is the
ionic charge.
In light of equations 5.18 and 5.19, a relationship can be derived between ionization
potential (I ) and crystal radius (Della Giusta et al., 1990):
 
C
rmax =  − 112  I −1 2 . ( 5.20 )
 (
 C2 n ) 

Equation 5.20 is general and differs from the proportionality of Ahrens, which is valid for
ions of homologous charge:
r ∝ I −2 . ( 5.21)
Because the effective charge may also be related to free ion polarizability ␣ fi:
−4
α i f = 4n 4 a 30 Zeff , ( 5.22 )
where a 0 is Bohr’s radius (0.529 Å), the crystal radius may be related to polarizability
through
 C1 
rmax =  α i f 1 4 . ( 5.23 )
(
 1.4142n a 30 4
 ) 

(Della Giusta et al., 1990). Experimental observations show that equations 5.20 and 5.23
both hold.
In olivine, 24 independent interatomic distances must be evaluated in order to define
the structure. For all 24 distances (Dij ) plus cell edges and volume, regressions of the
following type may be proposed:
Dij = ∑ ∑ ω ij, sX i, s + ξ j . ( 5.24 )

s i
In equation 5.24, ␻ij,s and ␰j are adjustable coefficients valid for a given distance j in a
given site s, and Xi,s are the atomic fractions of i ions in site s.
Equation 5.24 is quite precise. The maximum bias between observed and predicted
distances for a population of 55 samples is in fact 0.005 Å, practically coinciding with the
experimental error. The coefficients of equation 5.24 may be directly derived from the
ionization potential and ion polarizability, based on equations 5.20 and 5.23:
−1 2
 z 
∑ Im
1 4
ω ij , s = K 1 j , s + K 2 j , s α if + K 3 j ,s . ( 5.25)
 m =1  i
In equation 5.25, [∑zm⫽1Im ]i is the summation of the ionization potentials of ion i up to

valence z, and K1j,s, K2j,s and K3j,s are numerical constants valid for site s. The values of
Table 5.7 Constants of equation relating coefficients ␻ij ,s to ionization potential and
polarizability of cations in M1 and M2 sites in olivine. R is the correlation coefficient
(from Della Giusta et al., 1990).
Distance K1 K2 K3 Site of Reference R

O2A-O3A 1.4060 0.0720 1.8584 M1 0.936
M1-O1B 4.9132 0.1563 ⫺0.7328 M1 0.986
M1-O2A 6.1811 0.2281 ⫺0.1680 M1 0.991
M1-O3A 7.6123 0.2705 ⫺0.3521 M1 0.986
Mean M1-O 6.2303 0.2204 ⫺0.8553 M1 0.991
O1B-O2A 8.8115 0.3572 0.3179 M1 0.988
O1B-O3A 6.1686 0.1827 0.9582 M1 0.966
O1B-O3C 11.552 0.4195 0.752 M1 0.993
O2A-O3C 16.151 0.5740 ⫺0.6596 M1 0.905
O1B-O2B 6.9951 0.1934 1.0069 M1 0.991
M2-O1A 8.4392 0.3092 ⫺1.4253 M2 0.982
M2-O3C 7.3525 0.2576 ⫺0.7356 M2 0.968
M2-O3D 5.6366 0.1738 ⫺0.7300 M2 0.967
M2-O3E 6.4770 0.1854 ⫺1.0222 M2 0.987
Mean M2-O 6.6693 0.2149 ⫺1.0299 M2 0.994
O1A-O3E 16.395 0.5200 ⫺3.6118 M2 0.998
O3D-O3F 11.072 0.3221 ⫺1.7030 M2 0.999
O3E-O3F 10.732 0.3251 ⫺2.3447 M2 0.990
O2C-O3D 11.591 0.4004 ⫺2.4462 M2 0.981
O2C-O3E 6.3085 0.2119 ⫺1.4248 M2 0.964
a0 7.3633 0.2328 2.1574 M1 0.986
a0 1.7883 0.0634 0.1920 M2 0.957
b0 25.462 0.7845 ⫺0.7220 M2 0.994
c0 11.430 0.2874 1.3087 M1 0.991
c0 11.322 0.3398 ⫺0.8387 M2 0.997
V 1255.84 38.3462 ⫺3.7834 M1 0.992
V 1466.70 44.3668 ⫺499.44 M2 0.995
these constants for the principal interatomic distances (excluding the [SiO4 ]4⫺ tetrahe-
dron) are listed in table 5.7.
Table 5.8 lists cell edges and fractional atomic coordinates obtained for an olivine
mixture of composition Mg1.4Ni0.6SiO4 with 53.2% of the Ni distributed in M1 and 46.8%
in M2. The values obtained by calculation almost coincide with natural observation (see
also figure 5.7).
5.2.3 Phase Stability as a Function of P and T

With increasing T and constant P, olivines expand their octahedral sites M1 and
M2, whereas the tetrahedral group [SiO4 ]4⫺ remains virtually unaffected. Table
5.9 lists linear thermal expansion coefficients for M1-O and M2-O polyhedra in
some olivine compounds, according to Lager and Meagher (1978). These values
are generally higher than those obtained by the method of Hazen and Prewitt
(1977) (see eq. 1.93 and section 1.14.2). In particular, thermal expansion is anom-
Silicates / 233
Table 5.8 Measured (m) structural parameters for mixture Mg1,4Ni0,6SiO4

compared with results of calculations (c) based on equation 5.24 and observed
site occupancies (Ottonello et al., 1989)
a0 b0 c0 V
m c m c m c m c
4.7458 4.7473 10.1986 10.1968 5.9563 5.9559 288.35 288.33
x y z
m c m c m c
M1 0.0 0.0 0.0 0.0 0.0 0.0

M2 0.9900 0.9902 0.2761 0.2760 0.2500 0.2500
Si 0.4257 0.4258 0.0936 0.0936 0.2500 0.2500
O(1) 0.7660 0.7665 0.0924 0.0926 0.2500 0.2500
O(2) 0.2202 0.2202 0.4461 0.4460 0.2500 0.2500
O(3) 0.2763 0.2764 0.1626 0.1627 0.0326 0.0322
Table 5.9 Linear thermal expansion coefficients in

olivines (°C⫺1 ).
Compound ␣lM1⫺O (⫻ 105) ␣lM2⫺O (⫻ 105)

Ni2SiO4 1.54 1.31
Mg2SiO4 1.93 1.72
Mg0.69Fe0.31SiO4 1.17 1.22
Fe2SiO4 1.26 1.50
CaMgSiO4 1.74 1.33
CaMnSiO4 1.54 1.52
Mean value 1.53 1.43
Equation 1.93 1.38 1.38
alously high for Mg2⫹-bearing polyhedra. With increasing T, the rigidity of

[SiO4 ]4⫺ groups and the marked expandibility of octahedral sites result in in-
creased distortion of M1-O and M2-O polyhedra. However, at whatever T, the
M1 site remains less distorted than the M2 site.
Octahedral distortion prevents calculation of the volumetric thermal expan-
sion coefficient of the phase stemming from the linear polyhedral thermal expan-
sion values. However, volumetric thermal expansions may be obtained by least-
squares regression of the observed PVT relations. Table 5.10 lists, for instance,
the coefficients of a thermal expansion relation of the type
α V = α 0 + α 1T + α 2T −2 . (5.26)
These coefficients are valid for a pressure of 1 bar, within the stability field of
the phase.
Table 5.10 Isobaric thermal expansion of olivine compounds. Regressions for

Mg2SiO4 and Fe2SiO4 from Fei and Saxena (1986).
Compound ␣0 ⫻ 104 ␣1 ⫻ 108 ␣2 Reference

Mg2SiO4 0.2660 0.8736 ⫺0.2487 Suzuki et al. (1981)
Fe2SiO4 0.2711 0.6885 ⫺0.5767 Suzuki et al. (1981)
CaMgSiO4 0.3160 0.7008 ⫺0.1469 Skinner (1966)
CaFeSiO4 0.2260 1.4456 ⫺0.1652 Skinner (1966)
Incidentally, it must be noted that, although most olivine compounds are

stable (at P ⫽ 1 bar) up to the melting temperature, in some compounds polymor-
phic transitions are observed at temperatures below the melting point (for in-
stance, Ca2SiO4 has a transition to a monoclinic polymorph at 1120 K). Forsterite
at 1 bar melts at Tf ⫽ 2163 K, whereas fayalite melts at a much lower temperature
(Tf ⫽ 1490 K). The significance of this difference, as shown by Hazen (1977),
must be ascribed to different polyhedral thermal expansions for polyhedra con-
taining Mg2⫹ and Fe2⫹ cations. Hazen (1977) pointed out that, if the curve of
volumes of the members of the (Mg,Fe)2SiO4 mixture is extrapolated at various
T conditions (VT ) up to the melting point of each compound, the same limiting
volume is reached (VT f ⫽ 319 Å3; cf. figure 5.8). According to Hazen this volume
represents a sort of critical condition, beyond which the thermal expansion of
M1-O and M2-O polyhedra is not compatible (notwithstanding distortion) with
the rigidity of the groups: the bonds break and the compound melts. This inter-
pretation is extremely convincing and its rationalization of the melting behavior
of olivines should be tentatively extended to other silicate phases. We will examine
the thermodynamic significance of solid-liquid equilibria in the melting process
of the mixture (Mg,Fe)2SiO4 (section 6.5), observing that the phase relations give
rise to Roozeboom diagrams of type I (section 7.1.3). We will also see that the
observed phase boundaries may be ascribed to essentially ideal behavior, in both
the solid and liquid aggregation states of the mixture.
Concerning the effect of pressure on the structure of olivine, experimental
studies show that compression affects only M1-O and M2-O polyhedra (the
[SiO4 ]4⫺ group is virtually incompressible). Pressure thus has the same (although
inverse) effect as temperature. It must be noted, however, that, unlike tempera-
ture, the increase of pressure do not seem to cause further distortion of octahedral
sites (Brown, 1982). This is quite reasonable: octahedral distortion is largely a
result of the fact that one (in the M2 site) and two (in the M1 site) of the octahe-
dral edges are shared with tetrahedra, and are thus shorter. Because the [SiO4 ]4⫺
tetrahedron is substantially rigid, increased T causes increased dissimilarity of
the polyhedral edges, whereas increased P renders the octahedral edges more sim-
ilar in length (S. Merlino, personal communication).
The polyhedral compressibility of the various olivine compounds conforms
satisfactorily to the generalizations of Anderson (1972) and Hazen and Finger
Silicates / 235
Figure 5.8 Relationship between cell volume and temperature for compounds of mixture
(Mg,Fe)2SiO4. Extrapolation to melting point (Tf ) identifies the same limiting volume
(VTf ⫽ 319 Å3) for the various terms of the mixture. From Hazen (1977). Reprinted with
permission of The Mineralogical Society of America.
(1979) (cf. section 1.15.1 and eq. 1.110 and 1.111): there is an inverse linear rela-
tion between polyhedral compressibility and polyhedral volume.
Table 5.11 lists experimental bulk moduli for some olivine end-members.
Combining thermal expansion and compressibility data for compounds in
the (Mg,Fe)2SiO4 mixture, Hazen (1977) proposed the following equation of state
for olivines of the series Mg2SiO4-Fe2SiO4 :
 
(
Vcell + 290 + 0.17XFa − 0.006T + 0.000006T 2 1 −

) P
 , (5.27)
1350 − 0.16T 
where T is in °C, P is in bar, and V is in Å3.

We have already noted that the (Mg,Fe)2SiO4 mixture at high P undergoes
polymorphic transitions to the forms ␤ (modified spinel) and ␥ (spinel) by pro-
gressive thickening, which leads first to the [SiO4 ]4⫺ tetrahedra sharing a corner,
thus forming sorosilicate groups [Si2O7 ]6⫺ ( ␤ form; Moore and Smith, 1970), and
then to the maximum close packing (eutaxis) of oxygen ions (cubic form ␥ ).
Figure 5.9A summarizes the phase relations observed by Akimoto et al.
Table 5.11 Bulk modulus K0 (Mbar) and its first derivative

on P (K⬘) for some olivine end-members.
Compound K0 (Mbar) K⬘ Reference

Mg2SiO4 1.379 5.0 Sumino et al. (1977)
Fe2SiO4 1.600 5.0 Sumino (1979)
Mn2SiO4 1.310 5.0 Sumino (1979)
Co2SiO4 1.502 5.0 Sumino (1979)
Figure 5.9 Phase stability relations in binary systems Mg2SiO4-Fe2SiO4 (A) and
Mg2SiO4-Co2SiO4 (B) at T ⫽ 1000 °C.
(1976) at T ⫽ 1000°C in the binary system Mg2SiO4-Fe2SiO4 . The phase diagram

is essentially identical to that previously proposed by Ringwood and Major
(1970). Three two-phase fields are observed: ␣ ⫹ ␥, ␣ ⫹ ␤, and ␤ ⫹ ␥. Phase ␤
is stable only for molar Fe2SiO4 amounts under 30%. The peculiar “rabbit ears”
conformation of the binary loops is also observed in the Mg2SiO4-Co2SiO4 system
(figure 5.9B; Ringwood and Major, 1970), but in this case the “ears” are more
developed and phase stability is extended to 50% in moles of the component
Co2SiO4 . The transition from ␤ (orthorhombic) to ␥ (cubic) is first-order. Assum-
ing that upper-mantle olivine has a composition Fo89Fa11 , Ringwood and Major
(1970) deduced that the phase transition to forms ␤ and ␥ involves a portion of
mantle about 27 km thick. With a geothermal gradient of 30 bar/°C for the ␤–␥
transition, the discontinuity in the velocity of seismic waves, observed at about
400 km depth, may be ascribed to this structural rearrangement. The pressure in
the transition zone is 132 Kbar and the temperature 1600°C (the discontinuity
is observed at 370–420 km depth and p-waves increase their velocity by 0.6–1
km/s).
Silicates / 237
5.2.4 Thermodynamic Properties of End-Members

Table 5.12 reports a compilation of thermochemical data for the various olivine
components (compound Zn2SiO4 is fictitious, because it is never observed in na-
ture in the condition of pure component in the olivine form). Besides standard
state enthalpy of formation from the elements (Tr ⫽ 298.15 K; Pr ⫽ 1 bar; pure
component), the table also lists the values of bulk lattice energy and its constit-
uents (coulombic, repulsive, dispersive). Note that enthalpy of formation from
elements at standard state may be derived directly from bulk lattice energy,
through the Born-Haber-Fayans thermochemical cycle (see section 1.13).
Standard state entropy values and Maier-Kelley coefficients of heat capacity
at constant P, with respective T limits of validity, are listed for the same compo-
nents in table 5.13. The adopted polynomial expansion is the Haas-Fisher form:
Table 5.12 Thermochemical data for various olivine end-members (Tr ⫽ 298.15 K; Pr ⫽ 1 bar).
Listed values in kJ/mole (from Ottonello, 1987). H 0f , Tr , Pr ⫽ standard state enthalpy of formation
from the elements; UL ⫽ lattice energy; EBHF ⫽ energy of Born-Haber-Fayans cycle; EC ⫽
coulombic energy; EDD ⫽ dipole-dipole interactions; EDQ ⫽ dipole-quadrupole interactions; ER ⫽
repulsive energy.
Compound EBHF H f0, Tr , Pr UL EC EDD EDQ ER

Mg2SiO4 18,456.4 ⫺2175.7(1) 20,632.1 ⫺24,131.3 ⫺259.4 ⫺76.8 3835.4
⫺2170.4(2) 20,626.8 3840.6
⫺2176.9(3) 20,633.3 3834.0
⫺2163.6(4) 20,620.0 3847.3
⫺2171.9(8) 20,628.3 3839.2
Fe2SiO4 19,268.7 ⫺1481.6(1) 20,750.3 ⫺23,934.6 ⫺442.1 ⫺152.9 3779.3
⫺1479.4(2) 20,748.0 3781.6
⫺1479.9(3) 20,748.5 3781.1
⫺1478.8(8) 20,747.5 3782.1
Ca2SiO4 17,608.5 ⫺2316.2(2) 19,925.1 ⫺23,105.4 ⫺282.4 ⫺84.3 3546.9
Mn2SiO4 18,825.4 ⫺1728.1(2) 20,553.5 ⫺23,734.8 ⫺348.5 ⫺111.6 3641.3
⫺1727.2(3) 20,552.6 3642.2
⫺1732.7(8) 20,558.1 3636.7
Co2SiO4 19,443.9 ⫺1406.7(3) 20,850.6 ⫺24,040.8 ⫺480.5 ⫺174.7 3855.4
Ni2SiO4 19,609.9 ⫺1405.2(3) 21,015.1 ⫺24,219.8 ⫺596.5 ⫺226.2 4027.4
⫺1393.1(5) 21,004.0 4038.5
⫺1405.8(6) 21,015.7 4026.7
Zn2SiO4 19,326.0 ⫺1390.4(7) 20,716.4 ⫺24,068.0 ⫺393.9 ⫺129.6 3875.2
(1) Helgeson et al. (1978)
(2) Robie et al. (1978)
(3) Barin and Knacke (1973); Barin et al. (1977)
(4) JANAF (1974–1975)
(5) Levitskii et al. (1975)
(6) Mah and Paneratz (1976)
(7) Ottonello (1987) extrapolated value
(8) Holland and Powell (1990)
Table 5.13 Selected standard state entropy (S 0T r , P r ; Tr ⫽ 298.15 K, Pr ⫽ 1 bar) and Maier-Kelley
coefficients of heat capacity function. References as in table 5.12. Data in J/(mole ⫻ K).
Compound S T0 r , P r K1 K2 ⫻ 103 K3 ⫻ 10⫺5 K4 ⫻ 105 K5 T Limit Reference
Mg2SiO4 95.19 149.83 27.36 ⫺35.648 0 0 1800 (1)

Fe2SiO4 148.32 152.76 39.162 ⫺28.033 0 0 1800 (1)
Ca2SiO4 120.50 132.57 52.51 ⫺19.049 0 0 1200 (2)
Mn2SiO4 163.20 512.52 182.73 46.026 5.2058 ⫺6640.4 1800 (2)
Co2SiO4 158.57 157.40 22.05 ⫺26.694 0 0 1690 (3)
Ni2SiO4 110.04 163.18 19.748 ⫺24.309 0.88282 0 1818 (3)
Zn2SiO4 122.04 ⫺7.921 111.04 ⫺70.715 ⫺1.7405 3108.3 1800 (7)
C P = K 1 + K 2T + K 3T −2 + K 4T 2 + K 5T −1 2 . (5.28)
The polynomial expansion in equation 5.28 allows the calculation of the high-T enthalpy
and entropy of the compound of interest, from standard state values, through the equa-
tions
HT0r , Pr ≡ H f0 , Tr , Pr ( 5.29 )
T
HT , Pr = HT0r , Pr + ∫ C P dT ( 5.30 )
Tr
T
dT
ST , Pr = ST0r , Pr + ∫ CP T
( 5.31)
Tr
(see section 3.7 and eq. 3.105 and 3.106 for the form of integrals 5.30 and 5.31).
We recall also that the Gibbs free energy of the compound at the T and P of interest
may be obtained by applying
P
GT , P = HT , Pr − TST , Pr + ∫ VT , P dP. ( 5.32 )
Pr
To solve equation 5.32, we must first calculate the volume of the compound at the T of
interest and P ⫽ Pr. Adopting polynomial expansions of the type shown in equation 5.26,
we have
 T 
VT , Pr = VT0r , Pr  1 + ∫ α (T ) dT 
 Tr 
(5.33)
 α1 2 
= VT0r , Pr (
1 + α 0 (T − Tr ) + 2 T − Tr − α 2 T − Tr
2 −1 −1
) ( ) .
 
Silicates / 239
Integration on P depends on the form of the equation of state. For instance, adopting the
Murnaghan form:
−1 K ′
 K′ 
VT , P = VT , Pr  1 + K P , ( 5.34 )
 0 
we have
P P
∫ VT , P dP = ∫ VT , P dP
Pr 1
  ( K ′ − 1) K0 
(5.35)
 K 0    K′  
= VT , Pr   1 + P − 1  .
 K ′ − 1  K0  

We have already seen that the binary mixture (Mg,Fe)2SiO4 undergoes poly-
morphic transition with increasing pressure. The univariant curves in figure 5.9
define the loci of equality of potential for the components in the two phases at
equilibrium. The Gibbs free energy of polymorph ␣⫺␤ (or polymorphs ␣⫺␥ and
␤⫺␥) is thus identical at each point on the univariant curve. On the basis of this
principle, Fei and Saxena (1986) proposed a list of thermochemical values that
(within the limits of experimental approximation) reproduce P/X stability fields
(T ⫽ 1000 °C) for the system Mg2SiO4-Fe2SiO4 . These values are reported in
table 5.14.
When adopting the values in table 5.14, note that they refer to half a mole of
Table 5.14 Compilation of thermodynamic data for compounds ␤-Mg2SiO4, ␤-Fe2SiO4,

␥-Mg2SiO4 and ␥-Fe2SiO4, after Fei and Saxena (1986). Tr ⫽ 298.15; Pr ⫽ 1 bar. Note that
each listed value refers to one-half mole of compound (see first column).
CP J/(mole ⫻ K)
H 0f,T r ,Pr S 0T r ,Pr V To r ,Pr
Compound kJ/mole J/(mole ⫻ K) 3
cm /mole K1 K3 ⫻ 10⫺7 K5 ⫻ 10⫺3 K6 ⫻ 10⫺8
␤-MgSi0.5O2 ⫺1072.450 44.53 20.270 108.218 0 ⫺0.7362 ⫺2.7895
␤-FeSi0.5O2 ⫺736.473 70.93 21.575 136.332 0 ⫺1.3198 2.2412
␥-MgSi0.5O2 ⫺1069.300 42.75 19.675 100.096 ⫺0.066 ⫺0.5184 ⫺2.4582
␥-FeSi0.5O2 ⫺738.334 68.44 21.010 130.986 0 ⫺1.2041 1.3419
Compressibility
Thermal Expansion (K⫺1) (Mbar)
Compound ␣0 ⫻ 104 ␣1 ⫻ 108 ␣2 K0 K⬘
␤-MgSi0.5O2 0.2711 0.6885 ⫺0.5767 1.600 4.00

␤-FeSi0.5O2 0.2319 0.7117 ⫺0.2430 1.660 4.00
␥-MgSi0.5O2 0.2367 0.5298 ⫺0.5702 2.130 4.00
␥-FeSi0.5O2 0.2455 0.3591 ⫺0.3703 1.970 4.00
orthosilicate component (i.e., MgSi0.5O2 , FeSi0.5O2 ) and must thus be doubled to

be consistent with the values in tables 5.12 and 5.13. Note also that Fei and Sax-
ena 1986) use the Berman-Brown polynomial for CP, in the form
C P = K 1 + K 3T −2 + K 5T −1 2 + K 6T −3 (5.36)
(see section 3.2). It is also important to note that ␤-Fe2SiO4 is fictitious because
it never occurs in nature as a pure compound (cf. figure 5.9A).
To reproduce the binary loops of the system Mg2SiO4-Fe2SiO4 accurately, Fei
and Saxena (1986) use a (symmetric) regular mixture model for ␣-␤-␥ phases
with the following interaction parameters (cf. section 3.8.4): W␣ ⫽ ⫺8.314 kJ/
mole; W␤ ⫽ ⫺9.977 kJ/mole; W␥ ⫽ ⫺11.210 kJ/mole. These parameters also
refer to one-half mole of substance and must thus be multiplied by 2 to be consis-
tent with the usual stoichiometry.
Alternative thermochemical tabulations to reproduce the P/X stability fields
of the system Mg2SiO4-Fe2SiO4 may be found in Bina and Wood (1987) and Na-
vrotsky and Akaogi (1984).
5.2.5 Mixing Properties and Intracrystalline Disorder

If we consider the seven components in tables 5.12 and 5.13 as representative of
the chemistry of natural olivines, it is clear that 21 regular binary interaction
parameters (disregarding ternary and higher-order terms) are necessary to de-
scribe their mixing properties, through a combinatory approach of the Wohl or
Kohler type (cf. section 3.10). In reality, the binary joins for which interactions
have been sufficiently well characterized are much fewer. They are briefly de-
scribed below.
(Mg,Fe)2SiO4 Mixture
Mixing properties of the binary join Mg2SiO4-Fe2SiO4 have been extensively in-
vestigated by various authors. There is general consensus on the fact that the
mixture is almost ideal and that the excess Gibbs free energy of mixing is near
zero. Electrochemical measurements by Nafziger and Muan (1967) and Kitayama
and Katsura (1968) at T ⫽ 1200 °C and T ⫽ 1204 °C, respectively (P ⫽ 1 bar),
suggest slight deviation from ideality, which can be modeled with an interaction
parameter of about 1.5 to 2 kJ/mole (see also Williams, 1971; Driessens, 1968;
Sack, 1980). Calorimetric measurements by Sahama and Torgeson (1949) indi-
cate zero heat of mixing at 25 °C. However, more recent measurements by Wood
and Kleppa (1981) show the existence of positive enthalpy of mixing at T ⫽ 970
K and P ⫽ 1 bar, reproducible through a subregular mixing model of the form
( )
H mixing = 2 WH , FeMg X Fe X Mg + WH ,MgFe X Mg X Fe , (5.37)
Silicates / 241
Figure 5.10 Experimentally observed intracrystalline disorder in (Mg,Fe)2SiO4 mixture,

compared with theoretical distribution curves generated by interionic potential calcula-
tions. ● ⫽ Aikawa et al. (1985); ◆ ⫽ Smyth and Hazen (1973); ✸ ⫽ Brown and Prewitt
(1973); 䡺,䡬,䉭 ⫽ Ottonello et al. (1990). From G. Ottonello, F. Princivalle, and A. Della
Giusta, Temperature, composition and f02 effects on intersite distribution of Mg and Fe2⫹
in olivines, Physics and Chemistry of Minerals, 17, 301–12, copyright 䉷 1990 by Springer
Verlag. Reprinted with the permission of Springer-Verlag GmbH & Co. KG.
where WH,FeMg ⫽ 1 kcal/mole; WH,MgFe ⫽ 2kcal/mole, and XMg and XFe are mole
fractions of magnesian and ferrous components in the mixture.
As regards intracrystalline disorder on M1 and M2 sites, there is agreement
on the fact that Fe2⫹ and Mg2⫹ are almost randomly distributed between the two
sites. In figure 5.10, the intracrystalline distribution coefficient
FeM1 Mg M 2
KD = , ( 5.38 )
FeM 2 Mg M1
experimentally observed for various terms of the mixture at various T conditions

and P ⫽ 1 bar, is plotted against the molar fraction of the ferrous end-member.
The same figure also shows distribution curves calculated on the basis of an in-
terionic potential model. These curves define the conditions of internal disorder
in which the Gibbs free energy of the mixture is minimized and thus represent
stable equilibrium. There is satisfactory agreement between high-T experimental
results and theoretical calculations. It is interesting to note that the interionic
potential calculations predict KD values lower than 1 at low T and higher than 1
at high T. Starting from a temperature of about 600 °C (which represents a prob-
able kinetic low-T limit in intracrystalline exchanges), we see progressive “in-
crease of order” with increasing T, which is also apparent in the experimental
study of Aikawa et al. (1985). However, because intracrystalline distribution is
virtually random at all P and T conditions of interest, the entropy of mixing,
within reasonable approximation, may approach the ideal case (i.e., Sexcess mixing ⫽
0). We may therefore adopt for the mixture of interest a regular (or subregular)
mixture model of the zeroth principle. For this purpose, Chatterjee (1987) sug-
gests a Margules model:
( )
G excess mixing = X Fe X Mg WFeMg X Mg + WMgFe X Fe , (5.39)
where WFeMg ⫽ WMgFe ⫽ ⫹9 kJ/mole. This model agrees with the calorimetric
data of Wood and Kleppa (1981) but contrasts with the interaction parameter
adopted by Fei and Saxena (1986) on the basis of the polymorphic transitions
observed in the same system (cf. section 5.2.4).
Recent experiments by Akamatsu et al. (1993) show that pressure has a significant
effect on Fe-Mg intracrystalline distribution ( ∂KD /∂P ⬵ 2 Mbar⫺1 ). At the upper pressure
limit of stability of olivine (P 艐 0.12 Mbar, corresponding to a depth of about 400 km;
p-wave seismic discontinuity, cf. section 5.2.3), the increase in KD with respect to the P ⫽
1 bar condition, T and composition being equal, is about 0.24. For a Fo80 olivine, this KD
variation is similar to that induced by a T increase of about 250 °C (cf. figure 5.11).
(Mg,Ca)2SiO4 Mixture
The formation of the intermediate compound CaMgSiO4 (monticellite), with lim-
ited solubility toward the end-member components Mg2SiO4 and Ca2SiO4 , is well
known in this system. Warner and Luth (1973) experimentally investigated the
miscibility between CaMgSiO4 and Mg2SiO4 in the T range 800 to 1300 °C at
P ⫽ 2.5 and 10 kbar. The miscibility gap was parameterized by the authors with
a simple mixture model in which the interaction parameter (kJ/mole) varies with
T (K) and P (kbar) according to
2
 T   T   PT 
WT , P = 85.094 − 56.735   + 15.175   − 1.487P + 1.292   . (5.40)
 1000   1000   1000 
The miscibility gap extends over practically the whole compositional field from
CaMgSiO4 to Mg2SiO4 , and the maximum amount of monticellite miscible in
Mg2SiO4 is lower than 5% in molar proportions. Based on equation 5.40, the
miscibility gap should expand with increasing pressure. Figure 5.11 shows the
Gibbs free energy of mixing curves calculated at T ⫽ 600 °C and P ⫽ 1 bar, with
a static interionic potential approach. Based on the principles of minimization of
Gibbs free energy at equilibrium and of equality of chemical potentials at equilib-
Silicates / 243
Figure 5.11 Gibbs free energy of mixing in binary join Mg2SiO4-Ca2SiO4 at T ⫽ 600 °C
and P ⫽ 1 bar, calculated with a static interionic potential approach. Reprinted from
G. Ottonello, Geochimica et Cosmochimica Acta, 3119–3135, copyright 䉷 1987, with kind
permission from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5
1GB, UK.
rium, the plotted curves define two solvi (cf. sections 2.3 and 3.11): one extending
from (Mg0.02Ca0.98 )2SiO4 to (Mg0.49Ca0.51 )2SiO4 and the other extending from
(Mg0.51Ca0.49 )2SiO4 to pure forsterite. According to the model, monticellite is thus
virtually insoluble in forsterite.
(Mg,Ni)2SiO4 Mixture
According to the experiments of Campbell and Roeder (1968), at T ⫽ 1400 °C
the (Mg,Ni)2SiO4 mixture is virtually ideal. More recent measurements by Seifert
and O’Neill (1987) at T ⫽ 1000 °C seem to confirm this hypothesis. Because the
intracrystalline distribution of Mg2⫹ and Ni2⫹ on M1 and M2 sites is definitely
nonrandom (Ni2⫹ is preferentially stabilized in M1), the presumed ideality im-
plies that negative enthalpic terms of mixing counterbalance the positive excess
entropy arising from the configurational expression
 
S configuration = − R ∑ X i ,M1 ln X i ,M1 + ∑ X i ,M 2 ln X i ,M 2  ( 5.41)
 i i
—i.e., Sconfiguration does not reach the maximum value of equipartition (11.526
J/(mole ⫻ K)) and consequently an excess term in S appears.
Figure 5.12 Comparison between calculated and experimentally observed intracrystal-

line disorder in (Mg,Ni)2SiO4 mixture at various T, and P ⫽ 1 bar. Theoretical distribution
trends calculated with two values of hardness factor (␳ ⫽ 0.20 and 0.21). From Ottonello
et al. (1989). Reprinted with permission of The Mineralogical Society of America.
Figure 5.12 shows how Ni2⫹ and Mg2⫹ ions are distributed between M1 and
M2 sites in the (Mg, Ni)2SiO4 mixture at various T (P ⫽ 1 bar). Experimentally
observed distributions are compared with the results of static interionic potential
calculations carried out at two different values of the hardness factor ␳ (cf. sec-
tion 1.11.2).
Lastly, as far as volume is concerned, the mixture (Mg,Ni)2SiO4 exhibits
conspicuous deviations from ideality (see section 5.2.2): at high pressure, these
deviations presumably induce unmixing phenomena, not observed at P ⫽
1 bar.
For this purpose, it is of interest to recall that electrochemical galvanic cell measure-
ments indicate additional complexities of the system: Ottonello and Morlotti (1987) ob-
Silicates / 245
Table 5.15 Activity of component Ni2SiO4 in (Mg, Ni)2SiO4

mixture: (1) results of interionic potential model (Ottonello et
al., 1989); (2) Campbell and Roeder, (1968).
T ⫽ 1000 °C,
T ⫽ 1400 °C, P ⫽ 1 bar P ⫽ 1 bar
XNi2SiO4 (1) (2) XNi2SiO4 (1)
0.18 0.19 0.18 0.18 0.18

0.22 0.25 0.26 0.21 0.23
0.35 0.37 0.39 0.33 0.36
0.40 0.41 0.42 0.42 0.41
0.52 0.53 0.53 0.51 0.51
0.72 0.70 0.75 0.63 0.62
0.89 0.94 0.93 0.65 0.63
0.71 0.66
0.79 0.76
0.81 0.80
0.90 0.98
served two isoactivity regions in the binary join, one at approximately XNi ⫽ 0.2 and the
other between 0.4 and 0.6. Ottonello and Morlotti attributed this complex thermodynamic
feature to the formation of the intermediate compound (Ni0.25Mg0.75 )2SiO4 , with limited
solubility toward the end-member components Mg2SiO4 and Ni2SiO4 . Actually, as shown
by Campbell and Roeder (1968), in the presence of excess silica, an olivine mixture of
this limiting composition reacts with SiO2 to form pyroxene. Because the galvanic cell
compartment contains (Mg,Ni)2SiO4 ⫹ SiO2 ⫹ Ni, the two isoactivity regions indicate
the coexistence of a (reacted) pyroxene-type (Mg,Ni)SiO3 phase with a (Mg,Ni)2SiO4
mixture. The activity data derived from electromotive force (EMF) measurements cannot
thus be considered representative of the (Mg,Ni)2SiO4 mixture, and the most reliable ac-
tivity data are those of Campbell and Roeder (1968), which can be reproduced by appro-
priate parameterization of the repulsive potentials in the mixture, as shown in table 5.15.
(Mg,Mn)2SiO4 Mixture
According to Glasser (1960), the (Mg,Mn)2SiO4 mixture is almost ideal in the
temperature range T ⫽ 1300 to 1600 °C (P ⫽ 1 bar). The experiments of Maresch
et al. (1978) at T ⫽ 700 to 1100 °C (P ⫽ 2 kbar) indicate the presence of unmixed
phases in the composition range Fo40 to Fo75 . According to lattice energy calcula-
tions, there are two solvi at P ⫽ 1 bar, extending from intermediate composition
toward the two end-members (figure 6 in Ottonello, 1987). As regards intracrys-
talline disorder, Mn2⫹ appears preferentially stabilized in M2 at all investigated
T conditions (Francis and Ribbe, 1980; Urusov et al., 1984).
(Fe,Ca)2SiO4 Mixture
This mixture shows several analogies with the Mg2SiO4-Ca2SiO4 system. Also
in this case, an intermediate compound (kirschsteinite, FeCaSiO4 ) forms, with
limited solubility toward the two end-members. Mukhopadhyay and Lindsley

(1983) show that, at T ⫽ 800 °C, the solvus field in the Ca-rich side of the join
extends from (Fe0.08Ca0.92 )2SiO4 to (Fe0.42Ca0.58 )2SiO4 . Bowen et al. (1933) also
observed an extended miscibility gap between FeCaSiO4 and Fe2SiO4 . These mix-
ing properties are confirmed by lattice energy calculations (Ottonello, 1987). In
(Fe,Ca)2SiO4 , as in the (Mg,Ca)2SiO4 mixture, Ca2⫹ is stabilized at all T condi-
tions in the M2 site with more distorted octahedral symmetry.
(Fe,Mn)2SiO4 Mixture
According to Schwerdtfeger and Muan (1966), this mixture is essentially ideal at
T ⫽ 1150 °C, P ⫽ 1 bar, although the intracrystalline distribution measurements
of Annersten et al. (1984) and Brown (1970) show that Fe2⫹ is preferentially dis-
tributed on the M1 site.
(Co,Fe)2SiO4 Mixture
According to Masse et al. (1966), this mixture shows slight positive deviations
from ideality of activity values at T ⫽ 1180 °C, P ⫽ 1 bar. There are no experi-
mental data on intracrystalline distribution.
(Mg,Fe,Ca)2SiO4 Mixture
Only the quadrilateral portion of the Mg2SiO4-Ca2SiO4-Fe2SiO4 system with
XCa2SiO4 ⱕ 0.5 is of relevant interest in geochemistry. Mixing behavior in the oliv-
ine quadrilateral was investigated by Davidson and Mukhopadhyay (1984). As
figure 5.13 shows, the solvus between Ca-rich and Ca-poor compositional terms
extends within the quadrilateral, progressively shrinking toward the Fe-Ca join.
Multicomponent mixing behavior has been parameterized by Davidson
and Mukhopadhyay (1984) with a nonconvergent site-disorder model. The
interacting components were first expanded in a set of six site-ordered end-
members: FeM1MgM2SiO4 , MgM1FeM2SiO4 , FeM1FeM2SiO4 , MgM1MgM2SiO4 ,
FeM1CaM2SiO4 , and MgM1CaM2SiO4 . It was then assumed for simplicity that
mixing of Fe2⫹-Mg2⫹ is ideal in both M1 and M2 sites and that mixing of Fe2⫹-
Ca2⫹ and Mg2⫹-Ca2⫹ is nonideal in the M2 site. The authors then followed the
guidelines of Thompson (1969, 1970) and expanded the Gibbs free energy of mix-
ing of the mixture into two distinct contributions arising from configurational
entropy and enthalpic interaction terms:
G mixture = G end − members + G mixing (5.42)
G mixing = H mixing − TS mixing (5.43)
G end − members = µ MgCa

0
XMg,M 1 XCa,M 2 + µ FeCa
0
XFe,M 1 XCa,M 2
+ µ MgMg
0
XMg,M 1 XMg,M 2 + µ FeFe
0
XFe,M1 XFe,M 2 (5.44)
+ µ MgFe
0
XMg,M 1 XFe,M 2 + µ FeMg
0
XFe,M 1 XMg,M 2
Silicates / 247
Figure 5.13 Mixing properties of (Mg,Fe,Ca)2SiO4 olivines at P ⫽ 1 kbar and various

temperatures, after Davidson and Mukhopadhyay (1984). Tie lines connect coexisting un-
mixed phases according to the proposed mixture model. Experimentally determined com-
positions are also shown. From P. M. Davidson and D. K. Mukhopadhyay, Ca-Fe-Mg
olivines: phase relations and a solution model, Contributions to Mineralogy and Petrology,
86, 256–263, figure 3, copyright 䉷 1984 by Springer Verlag. Reprinted with the permission
of Springer-Verlag GmbH & Co. KG.
H mixing = WMgCaXMg,M 2 XCa,M 2 + WFeCaXFe,M 2 XCa,M 2 (5.45)
Smixing = −R( XMg,M1 ln XMg,M1 + XFe,M1 ln XFe,M1

(5.46)
+ XMg,M 2 ln XMg,M 2 + XFe,M 2 ln XFe,M 2 + XCa,M 2 ln XCa,M 2 )
The mixing properties were then solved by minimizing the Gibbs free energy of
the mixture with respect to ordering parameter t:
t = XFe,M 2 − XFe,M1 (5.47)
∂ G mixture  X Fe,M 1 X Mg,M 2 

∂t
= 0 = − RT ln  (
 − X Ca,M 2 WMgCa − WFeCa
 X Mg,M 1 X Fe,M 2 
)
(5.48)
∗
0
( )
+ ∆G 0.5X C a,M 2 + t + F − ∆ 0
G E0 (1 − 0.5X Ca,M 2 )
and by applying the concept of equality of chemical potential for components in
coexisting (unmixed) phases at equilibrium (tie lines in figure 5.13; cf. also sec-
tions 2.3 and 3.11).
Five excess terms appear in equation 5.48: two of them are the usual regular
interaction parameters (i.e., WMgCa ⫽ 32.9 kJ, WFeCa ⫽ 21.4 kJ); the other three
are related to standard-state potentials as follows:
∆G E0 = µ Fe
0
− µ Mg
0
= − 0.84 kJ ( 5.49 )
M 1Mg M 2 M 1FeM 2
∆G 0 = µ Mg
0
− µ Fe
0
= 7.0 ± 3.9 kJ ( 5.50 )
∗ M 1FeM 2 M 1Mg M 2
(
F 0 = 2 µ Mg
0
M1Ca M 2
− µ Fe
0
M1 Ca M 2
)+µ 0
Fe M1Fe M 2
(5.51)
− µ Mg
0
= 12.7 ± 1.6 kJ .
M1 Mg M 2
Note that equation 5.42 is analogous to the canonical expression
Gmixture = ∑ µ i0Xi + Gideal mixing + Gexcess mixing. ( 5.52)

i
However, Davidson and Mukhopadhyay (1984) did not use the macroscopic fractions Xi
but thermodynamic fractions arising from pair probabilities and consistent with the six–
end-member identification (see section 3.8.7). Moreover, nonrandom configurational con-
tributions were accounted for in the entropy term (which is not ideal, because Ca is absent
in the M1 site). In Thompson’s (1969, 1970) notation, entropic effects are treated in the
same manner, and the standard state Gibbs free energies of the pure end-members and all
nonconfigurational effects are grouped in a vibrational Gibbs energy term (G*):
G = −TSIC + G∗. ( 5.53)
The word “vibrational,” is in this case somewhat misleading, because enthalpic interac-
tions are essentially static in nature and are purely static whenever the zeroth approxima-
tion is appropriate (see Della Giusta and Ottonello, 1993). Lastly, it must be noted that
the excess term F 0 represents a reciprocal energy arising from the noncoplanarity of the
four quadrilateral end-members (cf. section 3.8.7). The model of Davidson and Mukho-
padhyay (1984) was reconsidered by Hirschmann (1991), who proposed an analogous
treatment for (Ni,Mg,Fe)2SiO4 mixtures.
5.3 Garnets
The structure of garnet was determined for the first time by Menzer (1926). The
general formula unit for garnet may be expressed as X3Y2Z3O12 . Garnet has a
body-centered-cubic cell (spatial group Ia3d ) and 8 formula units per unit cell.
Cations X, Y, and Z occupy nonequivalent sites with, respectively, dodecahedral,
octahedral, and tetrahedral symmetry. The main IV-fold coordinated cation is
Si4⫹. Figure 5.14 shows that in garnet, too, the [TO4 ] groups are isolated from
Silicates / 249
Figure 5.14 Structure of garnet. Positional angle defines degree of rotation on 4 axis.
one other (nesosilicates). The octahedra are also isolated, and are linked by single
apexes to tetrahedra. Dodecahedra share edges either with tetrahedra and octahe-
dra or with other dodecahedra. With respect to other silicates, the structure of
garnet is very dense as regards oxygen packing. Assuming a spherical oxygen ion
with an ionic radius of 1.38 Å, in noncalcic garnets oxygen occupies 70% of the
volume of the unit cell—a value near the theoretical value of maximum packing
(eutaxis; 74%; cf. Meagher, 1982).

The variety of symmetries in the garnet structure (coordinations 4, 6, and 8)
allows considerable compositional range. Table 5.16 lists the elements commonly
present in positions X, Y, and Z. The diadochy of Al, Ti4⫹, and Fe3⫹ in the tetra-
hedral site has been confirmed by Mössbauer spectroscopy on natural Fe-Ti-
bearing garnets (Schwartz and Burns, 1978), and the presence of phosphorus in
these sites, observed in upper mantle garnet, is attributable, according to Bishop
et al. (1976), to coupled substitutions of the type
Ca X + Si Z ⇔ Na X + PZ . ( 5.54 )
Table 5.16 Elements present at major, minor, and trace levels in the garnet phase.
X-site Y-site Z-site

(Dodecahedral (Octahedral (Tetrahedral
Abundance Symmetry) Symmetry) Symmetry)
Major Ca, Mn, Fe2⫹, Mg Al, Cr3⫹, Fe3⫹ Si

Minor Zn, Y, Na Ti3⫹, Ti4⫹, V3⫹, Fe2⫹, Zr, Sn, Al, Ti4⫹, Fe3⫹, P
Trace Li, Be, B, F, Sc, Cu, Ga,Ge, Sr, Nb, Ag, Cd, In, REE
Table 5.17 lists analyses of several types of natural garnet, after Deer et al.
(1983). The calculation of ionic fractions on the basis of 24 oxygens shows effec-
tive diadochy of Al on Si in tetrahedral site Z, whereas Ti4⫹ and Fe3⫹ appear to
be stabilized in octahedral site Y. Deviation from ideal closure in sites Y and X
(6 and 4, respectively) may be attributed in some instances to erroneous estima-
tion of ferric iron (samples 4, 5, and 6), but in others it is difficult to interpret
(samples 1, 2, and 3).
As suggested by Winchell (1933), the main garnet components may be sub-
divided into two groups with complete miscibility of their members: the
calcic group of ugrandites, comprising uvarovite (Ca3Cr2Si3O12 ), grossular
(Ca3Al2Si3O12 ), and andradite (Ca3Fe2Si3O12 ); and the group of pyralspites,
including pyrope (Mg3Al2Si3O12 ), almandine (Fe3Al2Si3O12 ), and spessartine
(Mn3Al2Si3O12 ). Six other less common compositional terms (which, however,
become more and more important in phase stability studies at high P) may be
associated with these six end-members. The names and formula units of these six
terms are listed in table 5.18 (note that all the natural garnets listed in table 5.17
have compositions that may be reexpressed as molar proportions of the various
compositional terms in table 5.18).
Table 5.19 summarizes the main occurrences of the various compositional
terms. Because garnets are solid mixtures, the listed components merely indicate
the main compositional terms of the phase.
Using multiple regression techniques, Novak and Gibbs (1971) obtained linear
equations that give cell edge and oxygen positional parameters as functions of
the radii of cations occupying sites X, Y, and Z. The equation of Novak and
Gibbs for cell edge is
a = 9.04 + 1.61rX + 1.89rY ( Å ). ( 5.55)
Equation 5.55 was later modified by Novak and Colville (1975) to include the
dimension of site Z:
Silicates / 251
Table 5.17 Compositions (in weight %) of some natural garnets (from Deer et al., 1983): (1)
almandine from quartz-biotite gneiss; (2) andradite from metamorphosed andesite; (3)
grossular from garnet-bearing gneiss; (4) pyrope from eclogite; (5) spessartine from a
cornubianitic rock; (6) uvarovite from a garnet-tremolite-pyrrothite vein.
Samples
Constituent
Oxide (1) (2) (3) (4) (5) (6)
SiO2 38.03 37.03 38.69 41.52 35.84 35.88

TiO2 - 0.04 0.55 trace 0.03 -
Al2O3 22.05 8.92 18.17 23.01 20.83 1.13
Cr2O3 - - - 0.22 - 27.04
Fe2O3 0.88 18.34 5.70 1.22 0.65 2.46
FeO 29.17 2.25 3.78 12.86 1.78 -
MnO 1.57 1.09 0.64 0.33 33.37 0.03
MgO 6.49 0.83 0.76 16.64 2.48 0.04
CaO 1.80 30.26 31.76 4.71 5.00 33.31
冧冧
H2O (⫹) - 0.48 0.13 -
0.16 0.18
(⫺)
H 2O - 0.16 0.06 -
Total 99.99 99.40 100.24 100.67 99.98 100.07
Ionic fractions on 24-oxygens basis
Si 5.951 6.043 5.966 5.999 5.808 5.964

Al 0.049 - 0.034 0.001 0.192 0.036
Total site Z 6.000 6.043 6.000 6.000 6.000 6.000
Al 4.019 1.716 3.268 3.912 3.786 0.186

Cr 0.102 - - 0.026 - 3.554
Fe3⫹ - 2.253 0.662 0.132 0.080 0.308
Ti - 0.005 0.064 - 0.004 -
Total site Y 4.121 3.974 3.994 4.070 3.870 4.048
Mg 1.513 0.201 0.175 3.580 0.598 0.010

Fe2⫹ 3.818 0.307 0.487 1.552 0.241 -
Mn 0.203 0.151 0.083 0.040 4.581 0.004
Ca 0.302 5.292 5.248 0.730 0.868 5.934
Total site X 5.836 5.951 9.993 5.902 6.288 5.948
Molar proportions of main components (%)
Almandine 65.4 5.2 8.1 26.1 4.0 -

Andradite 2.7 56.9 18.2 3.3 2.2 7.7
Grossular 2.5 32.0 69.4 8.9 11.2 2.4
Pyrope 25.8 3.4 2.9 60.3 7.9 0.2
Spessartine 3.6 2.5 1.4 0.7 74.7 0.1
Uvarovite - - - 0.7 - 89.6
Table 5.18 Chemistry and terminology of main components of the garnet phase.
Name Formula Unit References on Structure

Pyrope Mg3Al2Si3O12 Novak and Gibbs (1971)
Almandine Fe3Al2Si3O12 Skinner (1966), Geiger et al. (1987)
Spessartine Mn3Al2Si3O12 Skinner (1966)
Grossular Ca3Al2Si3O12 Novak and Gibbs (1971)
Uvarovite Ca3Cr2Si3O12 Huckenholtz and Knittel (1975)
Andradite Ca3Fe2Si3O12 Novak and Gibbs (1971)
Knorringite Mg3Cr2Si3O12 Irefune et al. (1982)
Calderite Mn3Fe2Si3O12 Lattard and Schreyer (1983)
Skiagite Fe3Fe2Si3O12 Karpinskaya et al. (1983), Woodland and O’Neill (1993)
Khoharite Mg3Fe2Si3O12 McConnell (1966)
Mn-Cr garnet Mn3Cr2Si3O12 Fursenko (1981)
Fe-Cr garnet Fe3Cr2Si3O12 Fursenko (1981)
Table 5.19 Occurrence of garnets in various types of rocks.
Mineral Occurrence
Almandine Eclogites (alm-pir), gneisses, metamorphic schists, granites, pegmatites,
Fe3Al2Si3O12 granulites (mineral index)
Grossular Typical of contact-metamorphosed limestones (skarns, marbles, cornubianites)
Ca3Al2Si3O12
Andradite In contact-metamorphic zones, in association with magnetite and Fe-rich
Ca3Fe2Si3O12 silicates (feldspathoid-bearing igneous rocks)
Uvarovite Mainly in chromite deposits
Ca3Cr2Si3O12
Spessartine Pegmatites and other igneous acidic rocks, metamorphic rocks and Mn-rich
Mn3Al2Si3O12 clay deposits
Pyrope Basic and ultrabasic rocks (peridotites, serpentinites), kimberlitic xenolites,
Mg3Al2Si3O12 rodingites, eclogites
a = 8.44 + 1.71rX + 1.78rY + 2.17rZ ( Å ). ( 5.56 )
Both equations are based on the ionic radii of Shannon and Prewitt (1969). The
method of Novak and Gibbs (1971) was then reexamined by Hawthorne (1981b),
and finally extended by Basso (1985) to hydrogarnets. Basso’s (1985) equations
are shown in table 5.20. They determine cell edge, fractional atomic coordinates
of oxygen, and mean metal-to-oxygen bond distances with satisfactory precision.
For even better precision, it is good practice to submit the obtained parameters
to a further optimization procedure, based on crystallographic constraints (i.e.,
Distance-Least-Squares, or DLS, treatment).
Table 5.21 shows structural data on various garnet end-members, compared
Silicates / 253
Table 5.20 Equations for structural simulation of garnet (Basso, 1985).
Fractional coordinates of oxygen

XO ⫽ 0.0258rX ⫹ 0.0093rY ⫺ 0.0462rZ ⫹ 0.0171
YO ⫽ ⫺0.0261rX ⫹ 0.0261rY ⫹ 0.0310rZ ⫹ 0.0514
ZO ⫽ ⫺0.0085rX ⫹ 0.0323rY ⫺ 0.0237rZ ⫹ 0.6501
Metal-to-oxygen distances
X1 ⫺ O ⫽ 0.558rX ⫹ 0.298rY ⫹ 0.244rZ ⫹ 1.447
X2 ⫺ O ⫽ 0.702rX ⫹ 0.083rY ⫹ 1.673
Y ⫺ O ⫽ 0.154rX ⫹ 0.754rY ⫹ 0.132rZ ⫹ 1.316
Z ⫺ O ⫽ 1.026rX ⫹ 1.366
Cell edge
a⫽
5
冦
8 3(X1 ⫺ O)2
2
⫹ 2(X2 ⫺ O)2 ⫹
7(Y ⫺ O)2 3(Z ⫺ O)2
5
⫹
2
⫹
冦冤3(X1 2⫺ O) ⫹ 2(X2 ⫺ O) ⫹ 7(Y ⫺5 O) 3(Z ⫺ O)2

冥
2 2 2
⫹ 2
⫹
2
⫺20冤冢X1 ⫺ O冣 ⫺ 冢Z ⫺ O冣冥
2 2 2
冤冢X1 ⫺ O冣 ⫺ 4冢X2 ⫺ O冣 ⫹ 2冢Y ⫺ O冣 ⫹ 冢Z ⫺ O冣冥

2 2 2 2 2
5
⫺
4
⫺5冤冢X1 ⫺ O冣 ⫺ 2冢Y ⫺ O冣 ⫺ 冢Z ⫺ O冣冥冧冧
2 2 2 2 1/2 1/2
with the values obtained by the method of Basso (1985), followed by the DLS
procedure of Baerlocher et al. (1977). The optimization procedure determinates
cell parameters for the six components not pertaining to the ugrandite and pyral-
spite series.
On the basis of equations of type 5.55 and of analogous expressions for posi-
tional parameters, Novak and Gibbs (1971) proposed several guidelines to estab-
lish whether a “garnet” compound may form and remain stable, according to the
dimensions of cations occupying sites X, Y, and Z. These guidelines consider the
following three factors.
1. The maximum distance compatible with Si-O bonds

2. The O-O distance of the nonshared octahedral edge
3. The mean dimensions of sites X and Y compared with the simple summa-
tions of rX ⫹ rO2⫺ and rY ⫹ rO2⫺, respectively.
On the basis of these principles, Novak and Gibbs (1971) proposed a map of
compatible rX and rY dimensions for the formation of a garnet phase. This map
(figure 5.15) shows that the commonly observed components plot in the low-
volume zone of the compatibility area. Theoretically, compounds with higher vol-
umes may form and remain stable at appropriate P and T conditions.
Table 5.21 Structural data on garnet components
Calculated Positional Parameter

Cell Edge (Å) of Oxygen
Component Formula Unit Observed Calculated XO YO ZO

Pyrope Mg3Al2Si3O12 11.459 11.4591 0.03291 0.05033 0.65364
Almandine Fe3Al2Si3O12 11.526 11.5263 0.03960 0.04925 0.65330
Spessartine Mn3Al2Si3O12 11.621 11.6210 0.03545 0.04770 0.65281
Grossular Ca3Al2Si3O12 11.845 11.8453 0.03902 0.04399 0.65163
Uvarovite Ca3Cr2Si3O12 11.996 11.9962 0.03982 0.04664 0.65463
Andradite Ca3Fe2Si3O12 12.058 12.0583 0.04016 0.04772 0.65588
Knorringite Mg3Cr2Si3O12 - 11.6040 0.03387 0.05290 0.65670
Calderite Mn3Fe2Si3O12 - 11.8288 0.03672 0.05137 0.65715
Skiagite Fe3Fe2Si3O12 11.728 11.7320 0.03529 0.05288 0.65769
Khoharite Mg3Fe2Si3O12 - 11.6637 0.03427 0.05395 0.65806
Mn-Cr garnet Mn3Cr2Si3O12 - 11.7683 0.03635 0.05030 0.65588
Fe-Cr garnet Fe3Cr2Si3O12 - 11.6720 0.03489 0.05183 0.65640
Cell Volume Molar Volume Molar Weight Density

Component (Å3) (cm3/mole) (g/mole) (g/cm3)
Pyrope 1504.67 113.287 403.122 3.559
Almandine 1531.21 115.284 496.668 4.308
Spessartine 1569.39 118.159 495.021 4.189
Grossular 1661.90 125.124 450.450 3.600
Uvarovite 1726.27 129.971 500.480 3.851
Andradite 1753.18 131.997 508.180 3.850
Knorringite 1562.51 117.641 453.152 3.852
Calderite 1655.09 124.612 552.751 4.436
Skiagite 1614.79 121.578 554.398 4.560
Khoharite 1586.75 119.466 460.852 3.858
Mn-Cr garnet 1629.83 122.710 545.051 4.442
Fe-Cr garnet 1590.14 119.722 546.698 4.566
5.3.3 Compressibility and Thermal Expansion

The experimental data of Meagher (1975) on pyrope and grossular, and of Rakai
(1975) on spessartine and andradite, show that Si-O distances do not vary appre-
ciably with increasing temperature and that thermal expansion affects both octa-
hedral and dodecahedral sites in an almost linear fashion. The different thermal
behaviors of the various symmetries lead to rotation of the [SiO4 ] groups on the
4 axis ( positional angle; cf. figure 5.14). With increasing T we also observe pro-
gressive increases in the shared O-O octahedral distances, whereas nonshared
O-O edges remain virtually unchanged.
Because of the importance of garnet in upper-mantle parageneses, the com-
pressibility of this phase has been the subject of many experimental studies. Ha-
zen and Finger (1978), studying the compressibility of pyrope, observed that the
Silicates / 255
Figure 5.15 Dimensional compatibility field for stable garnets according to Novak and
Gibbs (1971). Points correspond to 12 main components listed in table 5.21. Molar vol-
umes are also indicated.
[MgO8 ] dodecahedron is more compressible than the [AlO6 ] octahedron, and that
the [SiO4 ] group, unlike that of olivine, is also compressible. Thermal expansion
and compressibility are thus not perfectly antipathetic in this case. Notwithstand-
ing experimental studies on the compressibilities of the various compositional
terms, complete agreement among authors does not exist (Isaak and Graham,
1976; Sato et al., 1978; Leitner et al., 1980; Babuska et al., 1978; Hazen and
Finger, 1978; Bass, 1986). Calculation of garnet compressibility and thermal
expansion based on structure simulations at various pressures and temperatures
gives bulk moduli that are within the range of experimental values and may be
extended to not previously investigated compounds. Values (listed in table 5.22)
are relative to the Birch-Murnaghan equation of state. The same table also com-
pares calculated thermal expansion values with the experimental values of Skin-
ner (1956), Suzuki and Anderson (1983), Isaak et al. (1992), and Armbruster and
Geiger (1993) reconducted to the usual polynomial expansion in T:
1  ∂V 
α V ,T =   = α 0 + α 1T + α 2 T .
2
( 5.57 )
V  ∂T  P
Table 5.22 Bulk modulus and thermal expansion for various compositional terms of the
garnet phase. The experimental range for bulk moduli is relative to a common value of
K⬘ ⫽ 4. The calculated values are from Ottonello et al. (1996). References for thermal
expansion are as follows: (1) Ottonello et al. (1996); (2) Skinner (1956); (3) Suzuki and
Anderson (1983); (4) Isaak et al. (1992); (5) Armbruster and Geiger (1993).
Experimental
Range
Calculated Values K0 (K⬘ ⫽ 4)
Compound Formula Unit K0 (Mbar) K⬘
Pyrope Mg3Al2Si3O12 1.776 4.783 1.66–1.79
Almandine Fe3Al2Si3O12 1.837 4.795 1.75–1.80
Spessartine Mn3Al2Si3O12 1.797 4.833 1.72–1.75
Grossular Ca3Al2Si3O12 1.673 4.887 1.39–1.69
Uvarovite Ca3Cr2Si3O12 1.743 4.977 1.43–1.65
Andradite Ca3Fe2Si3O12 1.726 4.919 1.38–1.59
Knorringite Mg3Cr2Si3O12 1.876 4.842 -
Calderite Mn3Fe2Si3O12 1.818 4.823 -
Skiagite Fe3Fe2Si3O12 1.867 4.762 -
Khoharite Mg3Fe2Si3O12 1.857 4.799 -
Mn-Cr garnet Mn3Cr2Si3O12 1.852 4.894 -
Fe-Cr garnet Fe3Cr2Si3O12 1.902 4.831 -
Thermal Expansion
Compound ␣0 ⫻ 105 K⫺1 ␣1 ⫻ 109 K⫺2 ␣2 K Reference

Pyrope 2.431 5.649 ⫺0.2205 (1)
2.311 5.956 ⫺0.4538 (2)
2.244 9.264 ⫺0.1761 (3)
Almandine 1.781 11.558 ⫺0.2000 (1)
1.776 12.140 ⫺0.5071 (2)
Spessartine 2.962 2.398 ⫺0.8122 (1)
2.927 2.726 ⫺1.1560 (2)
Grossular 2.224 5.272 ⫺0.4119 (1)
1.951 8.089 ⫺0.4972 (2)
2.610 3.390 ⫺0.7185 (4)
Uvarovite 2.232 5.761 ⫺0.2329 (1)
Andradite 2.208 6.019 ⫺0.2005 (1)
2.103 6.839 ⫺0.2245 (2)
1.869 9.909 ⫺0.1486 (5)
Knorringite 2.894 2.9642 ⫺0.7748 (1)
Calderite 2.898 3.7316 ⫺0.5455 (1)
Skiagite 2.336 7.0286 ⫺0.2950 (1)
Khoharite 2.626 5.3687 ⫺0.4521 (1)
Mn-Cr garnet 2.862 4.0015 ⫺0.5160 (1)
Fe-Cr garnet 2.584 3.7865 ⫺0.4282 (1)
Silicates / 257
Leitner et al. (1980) showed that the bulk moduli of terms of pyrope-almandine-
grossular and andradite-grossular mixtures vary with chemical composition in a
perfectly linear fashion (within the range of experimental incertitude). This be-
havior is also implicit in the generalization of Anderson (1972).

Table 5.23 lists thermochemical data for various garnet endmembers. The pro-
posed enthalpy values for garnet reported in the literature are affected by large
discrepancies that reach 60 kJ/mole in the case of almandine and 45 kJ/mole for
andradite (see table 7 in Anovitz et al., 1993; and table 4 in Zhang and Saxena,
1991). The calculated lattice energy values shown in table 5.23 are consistent with
the selected values of standard state enthalpy of formation from the elements,
through the Born-Haber-Fayans thermochemical cycle (cf. section 1.13). The cal-
culated repulsive energies are also consistent with the Huggins-Mayer formula-
tion (see Tosi, 1964) for a common value of the hardness parameter ( ␳ ⫽ 0.48).
Table 5.23 also lists enthalpy of formation values for unknown compounds calcu-
lated by Ottonello et al. (1996) with an interionic potential model. These values
are affected by the same range of incertitude observed for ugranditic and pyrals-
pitic terms.
Table 5.24 lists selected data concerning entropy and isobaric heat capacity,
covering andradite, grossular, pyrope, and almandine terms, compared with re-
sults of calculations based on the Kieffer model (Ottonello et al., 1996).
Table 5.23 Thermochemical data for various endmembers of the garnet series.
ECFS ⫽ crystal field stabilization energy. The other column heads are defined as in table 5.12.
Data in kJ/mole.
0
Compound EBHF Hf,Tr ,Pr
UL EC ER EDD EDQ ECFS
Pyrope 59,269.2 ⫺6290.8(1) 65,560.0 ⫺77,060.6 12,287.8 ⫺599.8 ⫺187.4 0.0
Almandine 60,487.6 ⫺5261.3(2) 65,611.0 ⫺76,916.7 12,328.7 ⫺769.6 ⫺253.4 ⫺138.0
Spessartine 59,822.8 ⫺5686.9(3) 65,509.8 ⫺76,704.9 12,097.9 ⫺684.2 ⫺218.6 0.0
Grossular 57,997.4 ⫺6632.8(4) 64,630.2 ⫺76,148.6 12,360.9 ⫺645.3 ⫺197.2 0.0
Uvarovite 58,356.4 ⫺6034.6(5) 63,919.0 ⫺75,446.1 12,396.7 ⫺660.6 ⫺209.1 ⫺472.0
Andradite 58,489.9 ⫺5758.9(6) 64,248.8 ⫺75,160.5 11,779.0 ⫺660.5 ⫺206.8 0.0
Knorringite 59,628.2 ⫺5576.1(7) 64,732.4 ⫺76,294.3 12,390.4 ⫺623.9 ⫺204.5 ⫺472.0
Calderite 60,315.3 ⫺4719.8(7) 65,035.1 ⫺75,661.1 11,556.8 ⫺701.7 ⫺229.1 0.0
Skiagite 60,980.1 ⫺4276.2(7) 65,118.4 ⫺75,856.7 11,783.8 ⫺783.5 ⫺262.0 ⫺138.0
Khoharite 59,761.7 ⫺5299.5(7) 65,061.2 ⫺75,981.6 11,750.0 ⫺626.5 ⫺203.1 0.0
Mn-Cr garnet 60,181.8 ⫺5007.1(7) 64,716.9 ⫺75,965.8 12,182.0 ⫺701.5 ⫺231.7 ⫺472.0
Fe-Cr garnet 60,846.7 ⫺4555.3(7) 64,792.0 ⫺76,162.5 12,420.4 ⫺784.6 ⫺265.3 ⫺610.0
References: (1) Charlu et al. (1975); (2)Anovitz et al. (1993); (3)Boeglin (1981); (4)Charlu et al. (1978); (5)
Kiseleva et
al. (1977); (6)Helgeson et al. (1978); (7)Ottonello et al. (1996)
Table 5.24 Entropy and heat capacity function values for the various garnet end-members compared
with calorimetric values.
CP ⫽ K1 ⫹ K2T ⫹ K3T⫺2 ⫹ K5T ⫺1/2 ⫹ K4T 2
Compound S 0Tr ,Pr K1 ⫻ 10⫺2 K2 ⫻ 10 K3 ⫻ 10⫺5 K5 ⫻ 10⫺3 K4 ⫻ 105 T Range References

Pyrope 266.283 8.7293 ⫺1.4750 ⫺4.7243 ⫺8.6643 4.1084 298.15-m.p. (1)
266.27 8.025 ⫺0.958 ⫺23.52 ⫺7.2877 0 350–1000 (2,3)
8.72988 ⫺1.3744 0.045 ⫺8.7943 3.3408 820–1300 (4)
Almandine 342.562 8.7094 ⫺1.6067 ⫺1.6953 ⫺8.5176 5.2364 298.15-m.p. (1)
342.6 2.9274 3.4652 ⫺72.099 0.7163 15.317 400–1000 (5)
Spessartine 331.499 8.7486 ⫺1.4680 ⫺4.7045 ⫺8.6060 4.0201 298.15-m.p. (1)
Grossular 260.125 8.7296 ⫺1.4846 ⫺4.6095 ⫺8.6330 3.9643 298.15-m.p. (1)
260.12 (2)
Uvarovite 286.81 8.7266 ⫺1.4730 ⫺3.1499 ⫺8.5940 4.0649 298.15-m.p. (1)
Andradite 316.425 8.7479 ⫺1.4858 3.4215 ⫺8.6066 4.1040 298.15-m.p. (1)
316.4 8.0924 ⫺0.7025 ⫺6.789 ⫺7.403 0 300–1000 (6)
Knorringite 305.17 8.7557 ⫺1.4656 ⫺4.5706 ⫺8.6212 4.0758 298.15-m.p. (1)
Calderite 414.588 8.7683 ⫺1.4606 6.9140 ⫺8.5977 4.1656 298.15-m.p. (1)
Skiagite 369.871 8.7074 ⫺1.4476 ⫺4.0045 ⫺8.6526 4.3207 298.15-m.p. (1)
Khoharite 344.417 8.7616 ⫺1.4715 5.2712 ⫺8.6516 4.2446 298.15-m.p. (1)
Mn-Cr garnet 367.791 8.7736 ⫺1.4637 5.2303 ⫺8.6415 4.2123 298.15-m.p. (1)
Fe-Cr garnet 357.955 8.7447 ⫺1.4813 ⫺1.5757 ⫺8.5953 4.0796 298.15-m.p. (1)
(1) Ottonello et al. (1996); (2) Haselton and Westrum (1980); (3) Newton et al. (1977); (4) Tequı̀ et al. (1991);
(5) Anovitz et al. (1993); (6) Robie et al. (1987)
In the Kieffer model (Kieffer 1979a, b, and c; cf. section 3.3), the isochoric heat capac-
ity is given by
3R
3
 1 
CV =
Z
∑ S( X i ) + 3nR1 −
 N
− q K ( X l ,X u ) + 3nRq E( XE ) ,

( 5.58)
i =1
where S(Xi ) is the dispersed acoustic function, K (Xl,Xu ) represents the optic continuum, and
E(XE ) is the Einstein function (see eq. 3.69, 3.71, and 3.73 in section 3.3). The expression
for third-law harmonic entropy is:
[arcsin(X X )] X dX
3 2
Xi
3R  2 
3
∑∫
i
S =  
Z π ( ) ( )
12
i =1 X −X 0 e −1 i
2 2 X
[arcsin(X X )] ln(1 − e ) dx
3 2
Xi
3R  2 
3
∑∫
i
−   −X
Z π ( )
12
i =1 X −X 0 i
2 2
(5.59)
 3   3 
3R1 − − q 3R1 − − q
   
Xu Xu
∫ ln(1 − e ) dX
3s X dX 3s
+
Xu − Xl
∫( eX − 1
−
) Xu − Xl
− XE
Xl Xl
+
(
3RqX E
e XE − 1 )
− 3Rq ln 1 − e − X E , ( )
Silicates / 259
Table 5.25 Parameters of the Kieffer model for calculation of CV and harmonic entropy for the 12
garnet end-members in the system (Mg, Fe, Ca, Mn)3(Al, Fe, Cr)2Si3O12. ␻*l,Kmax is the lower cutoff
frequency of the optic continuum optimized on the basis of the experimental value of entropy at
T ⫽ 298.15 °C. ␪D is the Debye temperature of the substance. Sm and San are, respectively, the
magnetic spin and anharmonicity contributions to third-law entropy expressed in J/(mole ⫻ K).
See section 3.3 for the significance of the various frequencies (cm⫺1) (from Ottonello et al., 1996).
Compound ␪D ␻1 ␻2 ␻3 ␻l,K⫽0 ␻l,Kmax ␻*l,Kmax ␻u m1/m2 E1 E2 Sm San

Pyrope 793 73 73 130 149 115.7 124.0 638 1.519 975 875 0 3.316
Almandine 731 68 68 122 117 103.2 78.4 635 3.491 975 875 40.146 2.280
Spessartine 743 69 69 121 113 99.6 - 631 3.434 965 865 44.694 2.845
Grossular 821 74 74 126 180 152.2 141.9 619 2.505 950 850 0 2.295
Uvarovite 733 71 71 124 170 143.7 - 609 2.505 930 830 23.053 3.098
Andradite 723 70 70 123 132 111.6 121.2 589 2.505 925 825 29.796 3.174
Knorringite 701 64 64 118 139 107.9 - 628 1.519 955 855 23.053 2.931
Calderite 642 59 59 108 65 57.2 - 600 3.434 940 840 74.490 3.875
Skiagite 631 58 58 108 69 60.8 - 695 3.491 950 850 69.942 3.316
Koharite 690 64 64 116 101 78.4 - 608 1.519 950 850 29.796 3.406
Mn-Cr garnet 652 60 60 109 103 90.6 - 621 3.434 945 845 67.747 3.904
Fe-Cr garnet 640 59 59 110 107 94.3 - 625 3.491 955 855 63.199 3.331
where
បω
X = , ( 5.60 )
kT
␻ is angular frequency, ប is Planck’s constant (h/2␲ ), R is the Gas constant,

N ⫽ nZ is the number of atoms in the primitive cell, and q is the total proportion of
vibrational modes ascribed to Einstein’s oscillators.
The relevant parameters of the model for the various end-members adopted by Otto-
nello et al. (1996) are listed in table 5.25. The ␻l limit is the most critical parameter for
the estimation of macroscopic thermodynamic functions. Values of ␻l at K ⫽ 0 (center of
the Brillouin zone) based on infrared spectra may be reduced to ␻l,Kmax at K ⫽ Kmax
by applying
−1 2
 m 
ω l , K max = ω l , K = 0  1 + 2  , ( 5.61)
 m1 
where m1 and m2 are reduced masses of intervening ions (oxygen and “heavy” cation,
respectively, in our case; see Kieffer, 1980). ␻l,Kmax values for pyrope, almandine, spessar-
tine, grossular, and andradite are those of Kieffer (1980). For uvarovite, a value of 170
cm⫺1 for ␻l,K⫽0 was estimated from the spectrum of Moore et al. (1971), and the corre-
sponding ␻l,Kmax was obtained by application of equation 5.61.
The model parameters relative to the uncommon garnet end-members were obtained
by Bokreta (1992) through linear regression involving the atomistic properties of the in-
tervening ions. Equation 5.59 and the various vibrational function terms in eq. 5.58 were
calculated by gaussian integration. Values on nondimensionalized frequencies ␻l,Kmax were
adjusted, stemming from the initial guess values with a trial-and-error procedure, match-
ing the calculated S with experimental third-law entropy values at T ⫽ 298.15 K and
P ⫽ 1 bar, after correction for a magnetic spin entropy contribution of the form
Sm = R ln( 2 q + 1) , ( 5.62 )
where q is 32 for Cr3⫹, 52 for Mn3⫹ and Fe2⫹, 2 for Fe3⫹, and correction for anharmonicity,
according to
T
San = ∫ Vα 2 K 0 dT . ( 5.63 )
0
The procedure of Ottonello et al. (1996a) is operationally different from that indicated by
Kieffer (1980), which suggested that the calculated CV be fitted to the calorimetric CP at
100 K after subtraction of the anharmonicity contribution
C P = CV + TVα 2 K 0 . ( 5.64 )
Figure 5.16A shows, for example, how calculated CP values for pyrope compare with
low-T calorimetric measurements of Haselton and Westrum (1980). Adopting the initial
guess value of ␻l,Kmax (115.7 cm⫺1; see table 5.25) leads to an entropy estimate at Tr ⫽
298.15 of 272.15 J/(mol ⫻ K) (268.834 harmonic plus 3.316 anharmonic). The calorimet-
ric value being 266.27, the initial guess value of ␻l,Kmax is too low and must be raised to
124.0 cm⫺1 to match the calorimetric evidence. As we see in the upper part of figure 5.16
also the resulting CP in the low-T region compares better with calorimetric values after
adjustment of the nondimensionalized frequency at the limit of the Brillouin zone. In
the lower part of figure 5.16B, model predictions for pyrope are compared with high-T
calorimetric data of Newton et al. (1977) and Tequı̀ et al. (1991), indicating a reasonable
agreement with the latter.
5.3.5 Mixing Properties

The existence of unmixing processes in the garnet phase was long masked by its
isotropic nature, which inhibits mineralogical investigation under polarized light,
and was first observed by Cressey (1978) by transmission electron microscopy on
a garnet of composition (Mg0.21Fe0.49Ca0.29Mn0.01)3Al2Si3O12 . Indeed, the exis-
tence of positive interactions among garnet components in binary mixtures im-
plies the opening of solvus fields at given values of the consolute temperature (Tc;
see section 3.11). For instance, if we adopt an interaction parameter of 3.0 kcal/
mole for the (Mg,Mn)3Al2Si3O12 mixture, as proposed by Chakraborty and Gan-
guly (1992), the consolute temperature is
Tc =
W
=
3000
2 R 2 × 1.98726
( )
= 756 K = 483 ˚C . ( ) ( 5.65)
Silicates / 261
Figure 5.16 Heat capacity of pyrope at constant P, as determined from the Kieffer vibra-
tional model, compared with low-T (upper part of figure) and high-T (lower part of figure)
experimental evidences. From Ottonello et al. (1996). Reprinted with permission of The
This temperature may be somewhat enhanced at high-pressure regimes in light of

the observed positive deviations from Vegard’s law (Ganguly et al., 1993).
In spite of a conspicuous effort spent in deciphering the interaction properties
of binary garnet mixtures, agreement in the literature is still scanty. Figure 5.17
shows, for example, enthalpic interactions in binary aluminiferous garnet mix-
tures, according to various authors, represented as subregular Margules models
and compared with calorimetric values and results of interionic potential model
calculations. The corresponding subregular Margules interaction parameters are
Agreement is also poor concerning entropy and volume excess terms. Because
divalent cations (Mg, Ca, Fe, Mn) occupy only dodecahedral sites whereas octa-
hedral sites are reserved for trivalent cations (Cr, Fe3⫹, Al), each cation has only
one site at its disposal and permutability is fixed by stoichiometry (cf. section
3.8.1). As regards the occupancy on tetrahedral positions, we have already seen
that analyses of natural specimens show silicon deficiencies, compensated by Al 3⫹
Figure 5.17 Enthalpic interactions in the various binary joins of aluminiferous garnets.
䡺: calorimetric data; 䡵: results of interionic potential calculations. The corresponding
subregular Margules interaction parameters are listed in table 5.26 (from Ottonello et al.,
in prep.).
in tetrahedral coordination (cf. table 5.17). This fact should be carefully evaluated
in studies of the mixing properties of aluminiferous garnets. Unfortunately,
agreement among the various experimental sources is so poor that it prevents
appreciation of the effect of Al-Si substitutions in natural mixtures (cf. tables
5.26 and 5.27). Hence, the existence of excess entropy terms in mixtures must be
provisionally ascribed either to substantial short-range ordering on sites or to
vibrational effects.
Particularly, for the join (Ca,Fe)3Al2Si3O12 , the recent data of Koziol (1990)
confirm that excess entropy of mixing is virtually absent (cf. table 5.27). Existing
experimental data on the Mg3Al2Si3O12-Ca3Al2Si3O12 system show that the excess
entropy of mixing terms are probably symmetric and coupled to a slightly positive
excess volume. Newton and Wood (1980) suggested that the molar volume of the
Silicates / 263
Table 5.26 Subregular Margules parameters for enthalpic interactions in aluminiferous garnets
(from Ottonello et al., in prep.). Values in kJ/mole on a three-cation basis. The subregular Margules
model is Hmixing, ij ⫽ XiXj (Wij XiXj ⫹ WjiXjXi) with (1) Mg3Al2Si3O12, (2) Mn3Al2Si3O12,
(3) Ca3Al2Si3O12, and (4) Fe3Al2Si3O12.
WH12 WH21 WH13 WH31 WH14 WH41 WH23 WH32 WH24 WH42 WH34 WH43 Reference
- - 93.6 93.6 - - - - - - - - (1)
- - 25.1 47.9 - - - - - - - - (2)
- - 48.1 48.1 - - 0 0 0 0 - - (3)
4.5 4.5 - - - - - - - - - - (4)
- - - - 6.4 2.1 - - - - - - (5)
- - 12.6 50.8 31.4 2.5 - - - - ⫺7.9 58.0 (6)
- - 12.6 50.8 - - - - - - - - (7)
- - 15.0 56.5 - - - - - - - - (8)
- - - - 36.2 ⫺15.8 - - - - ⫺9.1 13.7 (9)
- - - - - - - - 0 0 - - (10)
12.6 12.6 - - - - - - - - - - (11)
- - - - - - 0 0 - - ⫺3.3 7.8 (12)
30.3 30.3 25.9 59.3 6.4 2.1 1.4 1.4 1.9 1.9 2.6 20.3 (13)
0 0 21.6 69.2 3.7 0.2 0 0 0 0 2.6 20.3 (14)
46.3 46.3 25.9 25.9 6.4 2.1 0 0 2.3 2.3 2.6 20.3 (15)
17.7 15.1 38.6 42.4 18.3 14.1 ⫺0.4 14.6 ⫺1.2 ⫺3.2 12.6 2.0 (16)
(1) Hensen et al., 1975; phase equilibration; (2) Newton et al., 1977; calorimetric; (3) Ganguly and Kennedy, 1974;
phase equilibration; (4) Wood et al., 1994; phase equilibriation; (5) Hackler and Wood, 1989; phase equilibration;
(6) Ganguly and Saxena, 1984; multivariate analysis; (7) Haselton and Newton, 1980; calorimetric; (8) Wood,
1988; phase equilibration; (9) Geiger et al., 1987; calorimetric; (10) Powenceby et al., 1987; phase equilibration;
(11) Chakraborty and Ganguly, 1992; diffusion experiments; (12) Koziol, 1990; phase equilibration; (13) Ganguly
and Cheng, 1994; multivariate analysis; (14) Berman, 1990; multivariate analysis; (15) Ganguly, 1995; personal
communication; (16) Ottonello et al. (in prep.); structure-energy calculations.
(Mg,Ca)3Al2Si3O12 and (Fe,Ca)3Al2Si3O12 mixtures has a sigmoidal path (posi-

tive excess volume over most of the binary join, coupled with slight negative devi-
ations observed in the vicinity of the Mg or Fe end-member). However according
to Dempsey (1980), this could be indicative of a displacive transition to space
group I213. Indeed, if we compare the experimental results of Newton and Wood
(1980) and Ganguly et al. (1993) for the same binary join, the assignment of
negative excess volume becomes highly conjectural. The experiments of Koziol
(1990) on the (Ca,Mn)3Al2Si3O12 join indicate that the binary mixture is virtually
ideal. According to Woodland and O’Neill (1993), molar volumes of mixing are
also ideal for the Fe3(Al,Fe)2Si3O12 join. Proposed subregular entropic and volu-
metric Margules parameters for aluminiferous garnets are presented in table 5.27.
As regards the calcic garnets, there is general consensus that the
Ca3(Al,Fe3⫹ )2Si3O12 mixture is essentially ideal (Holdaway, 1972; Ganguly, 1976;
Perchuk and Aranovich, 1979; Bird and Helgeson, 1980). For the Ca3(Cr,Al)2-
Si3O12 mixture, the calorimetric measurements of Wood and Kleppa (1984) indi-
cate negative enthalpy of mixing terms, which can be modeled with an enthalpic
interaction parameter WH falling between 0 and ⫺3 kcal/mole. According to
Table 5.27 Entropic (WS , cal/mole ⫻ K) and volumetric (WV ; cm3/mole)
terms of binary subregular interaction parameters for garnets in the
system (Fe, Mg, Ca, Mn)3Al2Si3O12.
WS,12 WS,21 WV,12 WV,21 Reference

Mg(1)-Ca(2) 4.5 4.5 - - (1)
11.19 0 3.0 0 (2)
2.51 2.51 0.58 0.12 (4)
- - 1.73 0.36 (5)
Fe(1)-Ca(2) 4.5 4.5 - - (1)
0 0 - - (1)
0 0 3.0 0 (2)
0 0 0 0 (3)
1.21 1.21 0.3 0.3 (4)
Fe(1)-Mg(2) 0 0 - - (1)
0 0 ⫺0.09 0.9 (2)
0 0 0.03 0.2 (4)
Ca(1)-Mn(2) 0 0 - - (4)
Mg(1)-Mn(2) 3.73 3.73 0 0 (4)
Fe(1)-Mn(2) 0 0 0 0 (4)
(1) Ganguly and Saxena (1984); (2) Newton et al. (1986); (3) Koziol (1990); (4) Gan-
guly (personal communication); (5) Ganguly et al. (1993).
Table 5.28 Macroscopic interaction parameters for garnets in

the three four-component systems (Mg, Mn, Ca, Fe)3Al2Si3O12,
(Mg, Mn, Ca, Fe)3Cr2Si3O12, and (Mg, Mn, Ca,Fe)3Fe2Si3O12
(Wohl’s 1946 formulation). Terms in parentheses are Berman’s
(1990) thernary interaction parameters (from Ottonello et al., in
prep.). X cation: (1) ⫽ Mg; (2) ⫽ Mn; (3) ⫽ Ca; (4) ⫽ Fe2⫹.
Y Cation
Al3⫹ Cr3⫹ Fe3⫹

W12 17.673 14.758 ⫺0.773
W21 15.112 11.480 5.236
W13 38.624 32.928 38.111
W31 42.418 61.085 46.801
W14 18.316 11.741 27.985
W41 14.142 18.767 6.726
W23 ⫺0.392 7.904 19.090
W32 14.549 21.798 ⫺1.067
W24 ⫺1.156 ⫺12.002 12.765
W42 ⫺3.193 6.984 ⫺14.429
W34 12.605 11.107 30.821
W43 1.963 ⫺3.189 ⫺4.458
W123 ⫺23.254 (58.825) 21.389 ⫺13.313
W124 11.904 (45.424) 24.347 ⫺65.766
W134 19.892 (11.470) ⫺12.967 ⫺27.540
W234 2.654 (1.975) ⫺4.676 104.290
W1234 0 (0) 0 0
Silicates / 265
Figure 5.18 Enthalpy of mixing surface for aluminiferous garnets in the system
(Mg,Mn,Ca,Fe)3Al2Si3O12 based on the Wohl’s model parameters in table 5.28 (from Otto-
nello et al., in prep.).
Mattioli and Bishop (1984), the mixture is of regular type but has a positive bulk
interaction parameter W between 0.6 and 3 kcal/mole. Extension to multicompo-
nent mixtures was attempted by Berman (1990), Ganguly (personal communica-
tion), and Ottonello et al. (in prep.). Berman (1990) applied the model of Berman
and Brown (1984), originally conceived for silicate melts, to aluminiferous garnets
and deduced the magnitude of ternary interaction parameters by applying
Wi , j , k =
(W
i ,i , j + Wi , j , j + Wi ,i , k + Wi , k, k + W j , j , k + W j , k, k ) −C (5.66)
i, j,k
2
and neglecting ternary term Ci, j,k (note that eq. 5.66 corresponds to the part on
the right in square brackets in Wohl’s equation 3.183). Ganguly (personal com-
munication) applied the formulation of Cheng and Ganguly (1994) and found
the ternary interaction terms to be insignificant. Wohl’s model parameters for
enthalpic interactions in aluminiferous, ferriferous, and chromiferous garnets ob-

tained by static interionic potential calculations are shown in table 5.28. Al-
though the three systems investigated are strictly quaternary, there is no need for
a quaternary interaction term because the enthalpy of mixing surface (figure 5.18)
does not exhibit an upward-convex conformation (contrary to what has been
observed for clinopyroxenes; see section 5.4.5).
5.4 Pyroxenes
Pyroxenes are polysilicates (“inosilicates” in the mineralogical classification),

crystallizing, respectively, in the monoclinic (clinopyroxenes; spatial groups
C2/c, P21 /c, P2/n) and orthorhombic systems (orthopyroxenes; spatial groups
Pbca, Pbcn). Chain periodicity p is 2 and defines the length of the c edge of the
unit cell (about 5.3 Å). Multiplicity m is 1 (cf. section 5.1). The peculiar charac-
teristic of pyroxene structures is the bonding of a tetrahedral group [SiO4 ] with
two neighboring groups, through bridging oxygens, to form infinite silicate chains
of composition [SiO3 ]n . Figure 5.19A shows the geometry of these chains, pro-
jected along the various axes and in perspective. Figure 5.19B shows more gener-
ally the relative positions of the chains in the structure: each chain has an apical
and a basal front. Crystallographic sites located between two neighboring basal
fronts are larger (coordination VI or VIII with oxygen) and are generally defined
as M2 positions. Sites between two neighboring apical fronts are smaller (VI-fold
coordination) and are usually defined as M1 positions. Neighboring chains may
be displaced relative to one another by tilting of the tetrahedra on the bridging
oxygens, thus allowing incorporation of different kinds of metal cations in M2
sites, as a function of the degree of distortion. This is the origin of the polymor-
phic transitions encountered in nature under various P, T, and X conditions.
The most representative spatial group of clinopyroxenes is C2/c, this structure
having been resolved as long ago as 1928 by Warren and Bragg. It is characterized
by sheets of single tetrahedral chains (parallel to axis c), alternating with octahe-
dral sheets with VI-fold or VIII-fold coordinated cations. The M1 octahedron is
quite regular, whereas M2 has distorted symmetry coordinated by eight oxygens.
The spatial group P21 /c (pigeonite, clinoenstatite, clinoferrosilite) differs from
C2/c in having two types of tetrahedral chains distinct in symmetry (A and B).
In the polymorphic P21 /c-C2/c transition (e.g., clinoferrosilite-ferrosilite) one ob-
serves doubling of cell edge a and orthogonalization of angle ␤ between edges a
and b. The Pbca structure was resolved by Warren and Modell (1930), who noted
its close correspondence with the C2/c structure (from which it differs essentially
by doubling of edge a of the unit cell and by the presence of two symmetrically
distinct tetrahedral chains). This structure contains octahedral sheets with alter-
nating orientations (⫾), composed of M1 and M2 sites with, respectively, regular
and distorted symmetry.
Silicates / 267
Figure 5.19 (A) Schematic representation of [SiO3]n chains of pyroxenes: (a) projection
on plane (100): (b) projection along axis Z; (c) projection along axis Y; (d) perspective
representation. (B) Relative arrangement of tetrahedral chains in pyroxene structure (seen
in their terminal parts with intercalation of M1 and M2 positions. From Putnis and
McConnell (1980). Reproduced with modifications by permission of Blackwell Scientific
Publications, Oxford, Great Britain.

The general formula for pyroxenes is XYZ2O6 where X is a generic cation in site
M2 (VIII-fold coordination in clinopyroxene and VI-fold in orthopyroxene), Y is
a generic cation in M1, and Z is a tetrahedrally coordinated cation. The most
important diadochy in tetrahedral positions is that of Al 3⫹, replacing Si4⫹, al-
though diadochy of Fe3⫹ has been observed under particular T and P conditions
(Huckenholtz et al., 1969). Al 3⫹ and Fe3⫹ also occur in site M1, together with
Cr3⫹ and Ti4⫹. Mg2⫹, Fe2⫹, and Mn2⫹ occupy both M1 and M2 sites, whereas
Li⫹, Na⫹, and Ca2⫹ occupy M2 positions exclusively.
According to Cameron and Papike (1982), pyroxenes contain Cr2⫹ and Ti3⫹
in rocks equilibrated at low f O2 (lunar specimens, meteorites). However, spec-
troscopic evidence is ambiguous and insufficient for a safe attribution (Rossman,
1982). Some authors (Bocchio et al., 1979; Ghose et al., 1986; Griffin and Mot-
tana, 1982) report the presence of Mn3⫹ in M1 sites in clinopyroxene. Davoli
(1987) reexamined this hypothesis, proposing precise structural criteria to detect
the presence of Mn3⫹ in the monoclinic phase (the ratio Mn3⫹ /Mn2⫹ may be a
potential f O2 barometer).
The chemical classification of pyroxene components disregards to some ex-
tent the represented structural groups, being based essentially on the chemistry
of major cations. As an example, table 5.29 reports the classification of Deer et
al. (1978), which subdivides pyroxene components into five groups (Mg-Fe, Ca,
Ca-Na, Na, and Li pyroxenes). In this classification, the term “enstatite” identi-
fies a component crystallizing in different structural classes (Pbca, P21 /c, Pbcn).
Table 5.29 Chemical classification of pyroxenes according to

Deer et al. (1978).
Name Stoichiometry Structural Class

Mg-Fe PYROXENES
Enstatite Mg2Si2O6 Pbca, P21/c, Pbcn
Ferrosilite Fe2Si2O6 Pbca, P21/c
Orthopyroxene (Mg,Fe,Ca)2Si2O6 Pbca
Pigeonite (Mg,Fe,Ca)2Si2O6 P21/c, C2/c
Ca PYROXENES
Augite (Ca,R2⫹)(R2⫹,R3⫹,Ti4⫹)(Si,Al)2O6 C2/c
Diopside CaMgSi2O6 C2/c
Hedembergite CaFeSi2O6 C2/c
Johannsenite CaMnSi2O6 C2/c
Ca-Na PYROXENES
Omphacite (Ca,Na)(R2⫹,Al)Si2O6 C2/c, P2/n, P2
Augitic aegirine (Ca,Na)(R2⫹,Fe3⫹)Si2O6 C2/c
Na PYROXENES
Jadeite NaAlSi2O6 C2/c
Acmite NaFeSi2O6 C2/c
Ureyite NaCrSi2O6 C2/c
Li PYROXENES
Spodumene LiAlSi2O6 C2, C2/c
R2⫹ in table 5.29 represents divalent cations (Mn2⫹, Fe2⫹, Mg2⫹ ), and R3⫹ repre-
sents trivalent cations (Fe3⫹, Cr3⫹, Al 3⫹ ). Thus, from the crystal chemical point
of view, the terms “augite,” “omphacite,” and “augitic aegirine” do not identify
precise stoichiometry. Analogously, the general term “orthopyroxene,” according
to Deer et al. (1978), represents a mixture of magnesian and ferroan components,
and the term “pigeonite” identifies an Mg-Fe-Ca mixture.
In geochemistry it is necessary to describe the composition of pyroxene by
end-members that are compositionally simple and stoichiometrically well defined.
It is also opportune to distinguish the various structural classes, because P-T
stability and reactivity vary greatly with type of polymorph. This inevitably re-
quires the formulation of “fictitious” components (i.e., components that have
never been synthesized as pure phases in the structural form of interest, but that
are present as members of pyroxene mixtures). For example, table 5.30 gives the
list of pyroxene geochemical components proposed by Ganguly and Saxena
(1987) (partly modified here).
In compiling table 5.30, the “orthopyrope” component (Mg1.5Al1.0Si1.5O6 ) of
Ganguly and Saxena was discarded, because it may be obtained as a mixture of
Mg-Tschermak and enstatite. Moreover, Cr- and Li-bearing terms (ureyite and
spodumene) are added to the C2/c class and a Co-bearing term to the Pbca class.
The manganous term (Mn2Si2O6 ; Tokonami et al., 1979) has been added to the
P21 /c class. It must be noted that natural pigeonites are mixtures of compositional
Silicates / 269
Table 5.30 Pyroxene components that may be adopted in geochemistry. x ⫽ actually

existing; o ⫽ fictive.
Component Stoichiometry System Structural Class Occurrence

Clinoenstatite Mg2Si2O6 Monoclinic P21/c x
Clinoferrosilite Fe2Si2O6 Monoclinic P21/c x
Mn clinopyroxene Mn2Si2O6 Monoclinic P21/c x
Diopside CaMgSi2O6 Monoclinic C2/c x
Ca-Tschermak CaAl2SiO6 Monoclinic C2/c x
Jadeite NaAlSi2O6 Monoclinic C2/c x
Hedembergite CaFeSi2O6 Monoclinic C2/c x
Al clinopyroxene Al2Al2O6 Monoclinic C2/c o
Wollastonite Ca2Si2O6 Monoclinic C2/c x
Acmite NaFeSi2O6 Monoclinic C2/c x
Ureyite NaCrSi2O6 Monoclinic C2/c x
Spodumene LiAlSi2O6 Monoclinic C2/c x
Enstatite Mg2Si2O6 Orthorhombic Pbca x
Ferrosilite Fe2Si2O6 Orthorhombic Pbca x
Orthodiopside CaMgSi2O6 Orthorhombic Pbca o
Mg-Tschermak MgAl2Si2O6 Orthorhombic Pbca o
Orthohedembergite CaFeSi2O6 Orthorhombic Pbca o
Al orthopyroxene Al2Al2O6 Orthorhombic Pbca o
Co orthopyroxene Co2Si2O6 Orthorhombic Pbca x
terms Mg2Si2O6 , Fe2Si2O6 , CaMg2Si2O6 , and CaFe2Si2O6 , crystallizing in the

spatial group P21 /c of the monoclinic system. Two additional terms with structure
P21 /c (CaMg2Si2O6 and CaFe2Si2O6 ) must then be added for complete represen-
tation of pigeonites. The same is true for components Mg2Si2O6 and Fe2Si2O6 of
the C2/c class. In geochemistry, nevertheless, it is usual to assume that the ener-
getic properties of the two monoclinic classes differ solely by a coherence term.
The energy of coherence (which may be assimilated to a strain energy, cf. section
3.13) is embodied in the Gibbs free energy of mixing terms, thus avoiding the
formulation of a redundant number of fictious components.
Pyroxenes (as feldspars) occupy a position that is “chemically central in the
realm of rock compositions” (Robinson, 1982). They are therefore found ubiqui-
tously and virtually in any kind of paragenesis developing in the P-T-X conditions
typical of earth’s mantle and crust. Table 5.31 lists the main occurrences.
Table 5.32 lists chemical compositions of natural pyroxenes. Pyroxene chem-
istry is quite complex and analyses must follow precise stoichiometric rules based
on site occupancies. Cameron and Papike (1982) follow four rules in the selection
of “acceptable” analyses among the plethora of values reported in the literature:
1. The sum of Si and tetrahedrally coordinated Al must not be higher than

2.02 or lower than 1.98.
2. The sum of M1 cations (Al, Fe3⫹, Cr 3⫹, Ti4⫹, Mg, Fe2⫹, Mn) must be
higher than 0.98.
Table 5.31 Occurrence of main compositional terms of pyroxenes in various types of rocks.
Term Stoichiometry Occurrence

ORTHORHOMBIC PYROXENES
Enstatite-bronzite (Mg,Fe)2Si2O6 Gabbros; Mg-rich ultramafic rocks;
serpentinites; meteorites
Hyperstene (Mg,Fe)2Si2O6 Charnockites; norites; Fe-rich
0.3 ⬍ Fe2Si2O6 ⬍ 0.5 ultramafic rocks; andesites;
rocks with high-P metamorphism
MONOCLINIC PYROXENES
Diopside-hedembergite (Ca,Mg,Fe)2Si2O6 Cr-rich ultramafic rocks; contact
metamorphism (cipollines,
cornubianites)
Pigeonite (Mg,Fe,Ca)2Si2O6 Basalts and dolerites;
Ca2Si2O6 ⬍ 0.12 kimberlites; meteorites
Augite (Ca,Mg,Fe,Al)2(Si,Al)2O6 Basalts, andesites, tephrites,
limburgites gabbros, dolerites,
ultramafic rocks. Sometimes in granites
Aegirine NaFeSi2O6 Nepheline syenites; hypersodic granites
Omphacite (Ca,Mg,Na,Al)2(Al,Si)2O6 Eclogites
Spodumene LiAlSi2O6 Pegmatites
3. The sum of M2 cations must range between 0.98 and 1.02.

4. The residual of charges in the equilibrium X ⫹ Y ⫽ Z with X ⫽ NaM2 ,
Y ⫽ AlM1 ⫹ CrM1 ⫹ Fe3⫹ M1 ⫹ 2TiM1 , and Z ⫽ AlT , must be lower than
4⫹
0.030.
Based on these four rules, Cameron and Papike (1982) selected 175 analyses out
of 405 reported by Deer et al. (1978) and discarded 230 compositions. This selec-
tion is extremely rigorous and does not take into account either the possible stabi-
lization of Fe3⫹ in tetrahedral sites or the existence of extrinsic disorder (cf. chap-
ter 4). Robinson (1982) showed that, by accepting a limited amount of cationic
vacancies in M2 sites and assuming possible stabilization of Fe3⫹ in tetrahedral
positions, 117 additional analyses out of the 230 discarded by Cameron and Pa-
pike (1982) may be selected.
Samples 6 and 7 in table 5.32 are from the Zabargad peridotite (Red Sea)
and are representative of the chemistry of upper mantle pyroxenes (Bonatti et al.,
1986). The absence of Fe2O3 in these samples is due to the fact that microprobe
analyses do not discriminate the oxidation state of iron, which is thus always
expressed as FeO. It must be noted here that the observed stoichiometry (based
on four oxygen ions) is quite consistent with the theoretical formula and that no
Fe3⫹ is required to balance the negative charges of oxygen.
Silicates / 271
Table 5.32 Compositions (in weight %) of natural pyroxenes (samples 1–5 from Deer et al.,
1983; samples 6 and 7 from Bonatti et al., 1986): (1) enstatite from a pyroxenite; (2) ferrosilite
from a thermometamorphic iron band; (3) hedembergite; (4) chromian augite from a
gabbroic rock of the Bushveld complex; (5) aegirine from a riebeckite-albite granitoid;
(6) diopside from a mantle peridotite; (7) enstatite from a mantle peridotite.
Samples
Oxide (1) (2) (3) (4) (5) (6) (7)

SiO2 57.73 45.95 48.34 52.92 51.92 50.87 54.03
TiO2 0.04 0.10 0.08 0.50 0.77 0.61 0.15
Al2O3 0.95 0.90 0.30 2.80 1.85 6.47 3.97
Fe2O3 0.42 0.31 1.50 0.85 31.44 - -
Cr2O3 0.46 - - 0.88 - 0.68 0.26
FeO 3.57 41.65 22.94 5.57 0.75 2.62 7.11
MnO 0.08 5.02 3.70 0.15 - 0.08 0.18
NiO 0.35 - - 0.10 - 0.01 0.02
MgO 36.13 3.49 1.06 16.40 - 15.13 33.13
CaO 0.23 1.43 21.30 19.97 - 21.79 0.67
Na2O - - 0.14 0.35 12.86 1.23 0.01
K2O - - 0.03 0.01 0.19 0.07 0.05
H2O(⫹) 0.52 0.65 0.46 0.10 0.17 - -
H2O(⫺) 0.04 0.09 - 0.07 - - -
Total 100.52 99.59 99.85 100.67 99.95 99.56 99.58

Si 1.972 1.972 1.988 1.929 1.986 1.858 1.886
Al 0.028 0.028 0.012 0.071 0.014 0.142 0.114
Total T 2.00 2.00 2.00 2.00 2.00 2.00 2.00

Al 0.010 0.018 0.006 0.049 0.069 0.136 0.049
Ti 0.001 0.003 0.002 0.014 0.022 0.017 0.004
Fe3⫹ 0.010 0.010 0.046 0.024 0.905 - -
Cr 0.012 - - 0.026 - 0.020 0.007
Mg 1.839 0.223 0.065 0.891 - 0.824 1.724
Ni 0.010 - - 0.003 - 0.000 0.001
Fe2⫹ 0.102 1.495 0.789 0.170 0.024 0.080 0.208
Mn 0.002 0.183 0.129 0.005 - 0.002 0.005
Ca 0.008 0.066 0.939 0.780 - 0.853 0.025
Na - - 0.011 0.024 0.953 0.087 0.001
K - - 0.001 0.000 0.008 0.003 0.002
Total M1 ⫹ M2 1.99 2.00 1.99 1.99 1.98 2.02 2.02

Mg/(Mg ⫹ Fe ⫹ Ca) 93.9 12.4 3.3 47.6 - 46.9 88.1
Fe/(Mg ⫹ Fe ⫹ Ca) 5.7 83.9 49.0 10.7 - 4.6 10.6
Ca/(Mg ⫹ Fe ⫹ Ca) 0.4 3.7 47.7 41.7 - 48.5 1.3
Figure 5.20 Pyroxene quadrilateral. Solid lines: compositionally continuous series;

dashed tie lines connect compositional terms coexisting on opposite sides of miscibility
gap. From G. M. Brown (1967). Mineralogy of basaltic rocks, in Basalts, H. H. Hess and
A. Poldervaart, eds., copyright 䉷 1967 by John Wiley and Sons. Reproduced with modifi-
cations by permission of John Wiley & Sons.
Most natural pyroxenes fall in the compositional field Mg2Si2O6-

CaMg2Si2O6-CaFe2Si2O6-Fe2Si2O6 which is usually called the “pyroxene quadri-
lateral” (fig. 5.20). This system contains one compositionally continuous series
of monoclinic C2/c pyroxenes (augites, Fe-augites) and another compositionally
continuous series of orthorhombic Pbca pyroxenes (bronzite-hyperstene). The
two series are separated by a miscibility gap whose nature will be discussed later.
An intermediate series with structure P21 /c (pigeonites) is also stable in the
system.

Table 5.33 lists the structural characters of various pure components of pyroxene
crystallizing in the stable state, according to the synthesis of Smyth and Bish
(1988). Note the doubling of cell edge a in the transition from monoclinic to
orthorhombic structure, composition being equal (i.e., clinoenstatite-enstatite;
clinoferrosilite-ferrosilite). This effect may be represented by the equation
a opx = 2 × a cpx sin β , (5.67)
where ␤ is the angle between edges a and c.

It should be stressed that doubling of cell edge a in the transition from
P21 /c (or C2/c) to Pbca structure does not imply doubling of the molar volume,
because the number of unit formulas per unit cell (Z) is also doubled (see table
5.33).
Silicates / 273
Table 5.33 Structure of pure pyroxene components in stable state. Molar volume in cm3/
mole; molar weight in g/mole; cell edges in Å; cell volume in Å3. Adapted from Smyth and
Bish (1988).
Phase Diopside Hedembergite Jadeite Acmite

Formula unit CaMgSi2O6 CaFeSi2O6 NaAlSi2O6 NaCrSi2O6
Molar weight 216.560 248.095 202.140 231.005
Density 3.279 3.656 3.341 3.576
Molar volume 66.039 67.867 60.508 64.606
Z 4 4 4 4
System Monoclinic Monoclinic Monoclinic Monoclinic
Class 2/m 2/m 2/m 2/m
Space group C2/c C2/c C2/c C2/c
Cell parameters:
a 9.746 9.845 9.423 9.658
b 8.899 9.024 8.564 8.795
c 5.251 5.245 5.223 5.294
␤ 105.63 104.70 107.56 107.42
Cell volume 438.58 450.72 401.85 429.06
Reference Cameron et Cameron et Cameron et Clark et al.
al. (1973 b) al. (1973 b) al. (1973 b) (1969)
Phase Ureyite Spodumene Ca-Tschermak

Formula unit NaCrSi2O6 LiAlSi2O6 CaAlAlSiO6
Molar weight 227.154 186.089 218.125
Density 3.592 3.176 3.438
Molar volume 63.239 58.596 63.445
Z 4 4 4
System Monoclinic Monoclinic Monoclinic
Class 2/m 2/m 2/m
Space group C2/c C2/c C2/c
Cell parameters:
a 9.579 9.461 9.609
b 8.722 8.395 8.652
c 5.267 5.218 5.274
␤ 107.37 110.09 106.06
Cell volume 419.98 389.15 421.35
Reference Cameron et al. Sasaki et al. Okamura et al.
(1973 b) (1980) (1974)
Phase Enstatite Ferrosilite Co-orthopyroxene
Formula unit Mg2Si2O6 Fe2Si2O6 Co2Si2O6

Molar weight 200.792 263.862 270.035
Density 3.204 4.002 4.222
Molar volume 62.676 65.941 63.963
Z 8 8 8
System Orthorhombic Orthorhombic Orthorhombic
continued
Phase Enstatite Ferrosilite Co-orthopyroxene

Class mmm mmm mmm
Space group Pbca Pbca Pbca
Cell parameters:
a 8.2271 8.4271 8.296
b 8.819 9.076 8.923
c 5.179 5.237 5.204
␤ 90 90 90
Cell volume 832.49 875.85 849.58
Reference Sasaki et al. Sasaki et al. Sasaki et al.
(1982) (1982) (1982)
Phase Clinoenstatite Clinoferrosilite Mn-clinopyroxene

Formula unit Mg2Si2O6 Fe2Si2O6 Mn2Si2O6
Molar weight 200.792 263.862 262.044
Density 3.188 4.005 3.819
Molar volume 62.994 65.892 68.608
Z 4 4 4
Class 2/m 2/m 2/m
Space group P21/c P21/c P21/c
Cell parameters:
a 9.626 9.7085 9.864
b 8.825 9.0872 9.179
c 5.188 5.2284 5.298
␤ 108.33 108.43 108.22
Cell volume 418.36 437.60 455.64
Reference Morimoto Burnham Tokonami
et al. (1960) (1967) et al. (1979)
Concerning the volume properties of the various polymorphs with respect to

their chemical compositions, existing information on Pbca and P21 /c structures
indicates linear dependence of mean M1 and M2 site dimensions on the ionic
radii of the occupying cations (algebraic mean). As figure 5.21 shows, this linear-
ity is more marked for M1 than for M2, which has distorted symmetry.
Figure 5.22 shows cell edges and volumes of synthetic pyroxenes in the
Mg2Si2O6-CaMg2Si2O6-CaFe2Si2O6-Fe2Si2O6 quadrilateral, according to Tur-
nock et al. (1973). The various parameters do not appear to evolve in linear fash-
ion with the chemistry of the system, and the relationship is particularly complex
for edge c.
The various lattice site dimensions in C2/c clinopyroxenes depend in a complex fash-
ion on the bulk chemistry of the phase. Nevertheless, the various interionic distances (and
cell edges and volumes) may be expressed as functions of cationic occupancies of lattice
Silicates / 275
Figure 5.21 Mean cation-to-oxygen distances in sites M1 (A) and M2 (B) plotted against
ionic radii of occupying cations (algebraic mean). Adopted radii are those of Shannon
(1976). Data from Cameron and Papike (1981).
Figure 5.22 Cell parameters of synthetic pyroxenes in the Mg2Si2O6-CaMg2Si2O6-

CaFe2Si2O6-Fe2Si2O6 quadrilateral. From Turnock et al. (1973). Reprinted with permis-
sion of The Mineralogical Society of America.
sites, through multiple regression parameters analogous to those proposed by Lumpkin

and Ribbe (1983) and Ottonello (1987) for olivines. These parameters (table 5.34) allow
estimations of a precision near that of experimental error. The mean discrepancy between
estimated and experimental values is 0.002 Å for T-O, M1-O, and M2-O distances and
0.004 Å for O-O distances. To obtain, for instance, distance T-O1 we apply
DT − O1 = 0.4894rZ − 0.1205rX − 0.0090rY + 1.4479, ( 5.68 )

Table 5.34 Multiple regression parameters relating site population to interionic distances in
C2/c pyroxenes.
Distance Z (T) X (M2) Y (M1) Constant

T-O1 0.4894 ⫺0.1205 ⫺0.0090 1.4479
T-O2 0.6058 0.0042 0.0009 1.2680
T-O3A1 0.4130 0.1645 0.0368 1.2879
T-O3A2 0.5045 0.2151 0.0736 1.1807
Mean T-O 0.5033 0.0659 0.0256 1.2959
O1A1-O2A2 0.9590 ⫺0.1717 ⫺0.0538 2.4170
O1A1-O3A1 0.9994 0.2413 0.1001 1.8700
O2A1-O3A1 0.8560 0.0922 0.0656 2.0755
O1A1-O3A2 1.0174 0.2637 0.0604 1.8977
O2A1-O3A2 0.5151 0.0241 0.0581 2.2260
O3A1-O3A2 0.4842 0.1592 ⫺0.0053 2.2888
M1-O1A2 0.1981 0.8682 0.2433 1.0684
M1-O1A1 0.2117 0.6872 ⫺0.1261 1.6636
M1-O2 ⫺0.1021 0.8627 ⫺0.0973 1.5851
Mean M1-O 0.3488 0.8976 0.0719 1.1705
O1A1-O1B1 ⫺0.1164 0.2724 ⫺0.1609 2.8378
O1B1-O1A2 0.2976 2.0568 0.0461 1.1266
O1A1-O1A2 0.6910 0.6797 ⫺0.0893 2.3024
O1A1-O2C1 0.0447 1.5730 ⫺0.1913 2.0718
O1B2-O2C1 ⫺0.3369 1.0620 0.7987 1.4958
O1A2-O2C1 ⫺0.0050 0.9138 ⫺0.5247 2.8250
O2C1-O2D1 0.2763 0.7692 0.0378 2.2430
M2-O1 0.5574 0.1390 0.8411 1.0302
M2-O2 0.6639 0.0160 1.3030 0.5274
M2-O3C1 0.2968 0.9035 ⫺0.7956 2.6450
M2-O3C2 ⫺1.4111 ⫺0.0657 ⫺0.3851 3.9561
Mean M2-O 0.0251 0.2471 0.2406 2.0416
O1A1-O2C2 0.0588 ⫺0.4499 0.7313 2.6130
O1A1-O3C2 ⫺0.9130 ⫺0.0194 0.5252 3.5457
O2C2-O3C2 ⫺0.5931 ⫺0.6688 0.9047 3.9398
O2C2-O3D2 0.3201 0.4011 0.9837 1.8535
O3C1-O3D2 ⫺0.2204 0.8072 0.1395 2.2979
O3C1-O3D1 2.1568 1.5370 0.8506 0.1710
a 1.4187 1.8303 0.3395 7.3108
b 0.3150 1.8880 0.0721 7.3158
c 0.7793 0.1775 0.1483 4.5631
␤ ⫺13.2743 ⫺12.5773 ⫺8.3044 131.2997
Volume 171.9523 217.2755 51.4963 136.2097
where
rZ = rAl IV X Al,T + rSiX Si,T ( 5.69 )
rX = rMgX Mg,M 2 + rCaX Ca,M 2 ( 5.70 )
rY = rAlVI X Al,M 1 + rFe3+ X Fe3+ , M 1 + rCr 3+ X Cr 3+ , M 1 + L . ( 5.71)

Silicates / 277
The ionic radii in equations 5.69 to 5.71 (rAlIV, rAlVI, rSi, . . .) are those of Shannon (1976)
in the appropriate coordination state with oxygen and XAl,T , XSi,T , . . . are the atomic
occupancies in the various sites.
5.4.3 Effect of Intensive Variables on Structure and

Phase Stability
The structural behavior of pyroxenes with the increase of T (pressure and chemi-
cal composition being equal) is to some extent similar to what we saw previously
for olivine and garnet. The tetrahedral sites are relatively insensitive to tempera-
ture, whereas M1 and M2 sites expand appreciably with T. Site expansion is a
function of the occupying cations, and site M2 generally exhibits more marked
thermal expansion with respect to M1, as a result of the lower mean charge of its
site population and the higher coordination number (VI-VIII as opposed to VI).
Because the “inert” tetrahedra are linked to form chains (unlike monosilicates)
and share corners with two sites with different polyhedral thermal expansions,
with increasing T there is progressive structural decoupling, partially adsorbed
by rotation of the tetrahedra at the point of junction (on axis c; cf. figure 5.23)
or by minimal rotation out of the chain plane. At high T, the chains are stretched
to a greater length, as shown in figure 5.23, as a consequence of the increased
angle ␪.
Experimental studies on C2/c pyroxenes (Cameron et al., 1973b; Finger and
Ohashi, 1976) at temperatures up to T ⫽ 1000 °C show that the structure remains
stable without transition phenomena, and that the increase of the angle ␪ is 2 to
3°. Analogous studies on Pbca pyroxenes (hyperstene, ferrosilite; Smyth, 1973)
show phase stability in the T range 25 to 1000 °C but a more conspicuous increase
in the angle ␪ (i.e., 10° for chain A and 15° for chain B in ferrosilite; 5° for chain
A and 10° for chain B in hyperstene). However, phase transitions have been ob-
served in P21 /c pyroxenes. Brown et al. (1972) observed a transition to the form
Figure 5.23 Extension of tetrahedral chains of pyroxenes as a function of temperature.

Table 5.35 Molar volume (cm3/mole), isobaric thermal expansion (K-1), and isothermal bulk
modulus (Mbar) of pyroxene end-members according to Saxena (1989)
Thermal Expansion Compressibility
Phase Component VTr ,Pr ␣0⫻104 ␣1⫻108 ␣2 K0 K⬘
C2/c ⫹ P21/c CLINOPYROXENES

Clinoenstatile Mg2Si2O6 62.91 0.20303 1.28049 0.0010 1.10 4.20
Clinoferrosilite Fe2Si2O6 66.28 0.23326 ⫺0.10413 ⫺0.0059 1.00 4.20
Diopside CaMgSi2O6 66.10 0.18314 1.41273 0.0011 1.00 4.20
Ca-Tschermak CaAl2SiO6 63.56 0.2470 0.00000 0.0000 1.20 4.20
Jadeite NaAlSi2O6 60.40 0.13096 2.23308 0.17442 1.33 4.20
Hedembergate CaFeSi2O6 68.26 0.18314 1.41273 0.0011 1.00 4.20
Al-elinopyroxene Al2Si2O6 52.4 0.20303 1.28050 0.0010 1.10 4.20
Wollastonite Ca2Si2O6 79.86 0.1000(?) - - 1.00 4.00
Pbca ORTHOPYROXENES
Enstatite Mg2Si2O6 62.66 0.20303 1.28049 0.0010 1.07 4.20
Ferrosilite Fe2Si2O6 65.90 0.23326 ⫺0.10413 ⫺0.0059 1.01 4.20
Orthodiopside CaMgSi2O6 69.09 0.18314 1.41273 0.0011 1.07 4.20
Orthohedembergite CaFeSi2O6 69.22 0.18314 1.41273 0.0011 1.075 4.20
Al-orthopyroxene Al2Si2O6 53.56 0.20303 1.28050 0.0010 1.00 4.20
Pbcn ORTHOPYROXENES
Protoenstatite Mg2Si2O6 64.88 (?) - - 1.10 4.00
C2/c at about 960 °C in pigeonite, and Smyth (1974) observed a first-order transi-
tion in the low-T clinopyroxene at T ⫽ 725 °C.
Pressure has an antithetical effect on the interionic distances of pyroxenes
with respect to temperature, and the observed compressibilities of M2, M1, and T
sites obey the generalizations of Anderson (1972) and Hazen and Prewitt (1977).
The properties above indicate that the responses of the various pyroxene poly-
morphs to modifications of P and T intensive variables are different. It is essen-
tially this fact that determines the opening of miscibility gaps during the cooling
stage from “magmatic” P-T regimes. To rationalize this phenomenon in thermo-
dynamic terms, we must distinguish the structural properties of the various com-
positional terms crystallizing in the various spatial groups with appropriate vol-
ume factors. Table 5.35 lists the values of molar volume, isobaric thermal
expansion, and isothermal bulk modulus for pyroxene polymorphs in the thermo-
dynamic database of Saxena (1989). Note that, for several components, Saxena
(1989) adopts identical thermal expansion parameters for the structural classes
C2/c and Pbca. This assumption renders the calculations straightforward in phase
transition studies but is not compulsory.
Let us examine in detail the phase stability relationships of the four end-members of
the pyroxene quadrilateral. Figure 5.24A shows the P and T stability ranges of the various
polymorphs of the Mg2Si2O6 compound, according to Lindsley (1982). The stable form
Silicates / 279
Figure 5.24 Phase stability relations for end-members of pyroxene quadrilateral. Melting
curves refer to anhydrous conditions. Solidus curves for CaMgSi2O6 in saturated vapor
phase conditions are also shown for various CO2/H2O ratios in the vapor phase. Dashed
lines are extrapolated. From Lindsley (1982). Reprinted with permission of The Mineral-
ogical Society of America.
at low T is C2/c (clinoenstatite; CEn in figure 5.24A). At a given T between 566 °C (1

bar) and about 720 °C (40 kbar), there is a first-order transition to the orthorhombic form
Pbca (orthoenstatite; OEn in figure 5.24A). At P ⫽ 1 bar and T ⫽ 985 °C, there is transi-
tion from orthoenstatite (Pbca) to protoenstatite (Pbcn; PEn in figure 5.24A). The ther-
mobaric gradient for this polymorphic transition is quite high and, with increasing P, the
T stability range of the Pbcn polymorph decreases progressively and vanishes at about
P ⫽ 15 kbar. At higher temperatures, the Pbca form is thus stable at all T conditions, up
to congruent melting of the compound. At P ⫽ 1 bar, protoenstatite is stable up to T ⫽
1557 °C, at which incongruent melting occurs, with formation of forsterite plus liquid.
Incongruent melting has been observed only at low pressure, but the high-P limit of the
process is not known precisely.
At low P, the Fe2Si2O6 component is extrinsically unstable with respect to ␣-quartz
plus fayalite paragenesis:
Fe 2 Si 2 O6 ⇔ Fe 2 SiO 4 + SiO2 . ( 5.72 )
This reaction proceeds spontaneously toward the right, up to pressures that are not known
precisely but that are estimated to be around 5 to 8 kbar for T in the range 200 to 400 °C.
At higher P and low T, a monoclinic phase is stable (clinoferrosilite; probably in P21 /c
form; CFs in figure 5.24B), and at high P and T the orthorhombic form Pbca is stable
(OFs in the same figure). According to Lindsley (1982), below the melting point the ortho-
rhombic phase has another transition to a pyroxenoid phase (ferrosilite III; Fs-III in figure
5.24B) or to the monoclinic C2/c polymorph.
At high P, CaMgSi2O6 undergoes congruent melting (figure 5.24C). The melting tem-
perature at P ⫽ 1 bar is about 1390 °C and (not shown in figure) the melting process at
low P is incongruent with the formation of forsterite:
CaMgSi 2 O6 ⇔ Mg2 SiO 4 + melt . ( 5.73 )
The T stability of CaMgSi2O6 is greatly affected by fluid partial pressure. In the presence
of H2O-rich fluids, the thermobaric gradient of the melting reaction (which is positive in
anhydrous conditions, as observed for most crystalline silicates; cf. chapter 6) is reversed,
according to the experiments of Rosenhauer and Eggler (1975) (figure 5.24C).
The stable form of the CaFeSi2O6 compound is monoclinic C2/c in a wide
P-T range (hedembergite; Hd in figure 5.24D). At P ⬍ 13 kbar, hedembergite has a transi-
tion to the triclinic form bustamite (Bu in figure 5.24D). At low P, bustamite melts incon-
gruently to produce silica plus melt. Nevertheless, the melting reaction is congruent at P
higher than 2 to 3 kbar.

We have already noted the necessity in geochemistry for complete definition of
the thermodynamic properties of all compositional terms required to describe the
chemistry of natural pyroxenes in their various structural forms (C2/c; P21 /c;
Pbca; Pbcn). Because the pyroxene quadrilateral is usually described as a quater-
nary system (it is in fact a portion of the ternary system Mg2Si2O6-Ca2Si2O6-
Fe2Si2O6 ) and because four polymorphs form in the system, we should theoreti-
cally attribute 16 thermodynamic parameters for each magnitude to describe
quantitatively phase stability relations at the compositional limits (plus mixing
properties inside the quadrilateral; see section 5.4.5). Actually, the Pbcn structure
is important only for the magnesian end-member protoenstatite. Moreover, C2/c
and P21 /c are often treated as a single polymorph, under the assumption that
Silicates / 281
no apparent discontinuities are observable in the Gibbs free energy curve of the
crystalline mixtures as a function of their chemical composition (see, for instance,
Buseck et al., 1982). Table 5.36 lists the thermodynamic data of end-members at
standard state of pure component according to Saxena (1989). These data are
internally consistent and, combined with the appropriate mixing properties (see
section 5.4.5), quantitatively describe the phase relations observed within the
quadrilateral at the various P and T conditions of interest (the database of Sax-
ena, 1989 is given preference here with respect to the more recent revision of
Saxena et al., 1993). For comparative purposes, the same table also lists entropy,
enthalpy of formation from the elements, and the heat capacity function adopted
by Berman (1988) and by Holland and Powell (1990).
Table 5.37 lists the various terms of the lattice energy of some clinopyroxene
components. Data refer to the C2/c spatial group. Components Mg2Si2O6 and
Fe2Si2O6 , which crystallize in P21 /c, must thus be considered “fictitious.”
5.4.5 Mixing Properties and Subsolidus Phase Relations in

the Pyroxene Quadrilateral Mg2Si2O6-CaMgSi2O6-
CaFeSi2O6-Fe2Si2O6
During crystallization of basaltic magmas, orthopyroxene (Pbca), pigeonite
(P21 /c), and/or augite (C2/c) form. Figure 5.25A reports the solidus relations
drawn by Huebner (1982) on the basis of existing petrologic data. As outlined by
that author, figure 5.25A cannot actually be considered as a phase diagram, be-
cause the P and T values are not specified and the natural systems in which pyrox-
ene phases form by crystallization are chemically heterogeneous. With these pro-
visions, figure 5.25A furnishes a realistic picture of the solidus stability relations
of the various pyroxene polymorphs. There is a miscibility gap between orthopy-
roxene and augite, extending from the iron-free join to an Fe/(Fe ⫹ Mg) molar
ratio of 0.45. For more ferroan compositions, augite coexists stably with pigeonite
and the miscibility gap shrinks progressively with increasing iron content in the
system. Three pyroxene phases coexist stably in the transition zone between the
two miscibility gaps (i.e., 0.40 ⬍ Fe/(Fe ⫹ Mg) ⬍ 0.60). This three-phase field is
larger for pyroxenes of stratified complexes, as shown in figure 5.25B [0.25 ⬍ Fe/
(Fe ⫹ Mg) ⬍ 0.70]. The coexistence of three pyroxenes in magmatic systems at
solidus conditions is imputed to the heterogeneous reaction (Ross and Huebner,
1975):
orthopyroxene + liquid ⇔ pigeonite + augite .

Pbca P21 /c C2/c (5.74)
In this interpretation, the persistence of the orthorhombic polymorph must

be imputed to metastability associated to low reaction kinetics. At subsolidus
temperatures, characteristic of most metamorphic regimes (including the granu-
Table 5.36 Thermodynamic properties of pure pyroxene components in their various structural forms according to Saxena (1989) (1),
Berman (1988) (2), and Holland and Powell (1990) (3) database. S 0Tr ,Pr ⫽ standard state entropy of pure component at Tr ⫽ 298.15 K and
Pr ⫽ 1 bar (J/mole); H 0f,Tr ,Pr ⫽ enthalpy of formation from elements at same standard state conditions. Isobaric heat capacity function CP is
CP ⫽ K1 ⫹ K2T ⫹ K3T ⫺2 ⫹ K4T 2 ⫹ K5T ⫺3 ⫹ K6T ⫺1/2 ⫹ K7T ⫺1.
K3 K5 K6 K7
Phase Component Reference H 0f,Tr ,Pr S 0Tr ,Pr K1 K2 ⫻ 104 K4
106 108 102 104
C2/c ⫹ P21/c CLINOPYROXENES

Clinoenstatite Mg2Si2O6 (1) ⫺3088.0 134.5 288.90 37.64 ⫺2.700 0 9.224 0 ⫺3.876
(2) ⫺3091.9 132.7 279.92 0 8.8004 0 10.7142 ⫺9.94 0
(3) ⫺3083.2 135.0 356.2 ⫺29.9 ⫺0.5969 0 0 ⫺31.8503 0
Diopside CaMgSi2O6 (1) ⫺3201.8 143.1 221.21 328 ⫺6.586 0 0 0 0
(2) ⫺3200.6 142.5 305.41 0 ⫺7.166 0 9.2184 ⫺16.049 0
(3) ⫺3200.2 142.7 314.5 0.41 ⫺2.7459 0 0 ⫺20.201 0
Ca-Tschermak CaAl2SiO6 (1) ⫺3306.0 139.3 280.08 26.85 ⫺5.344 0 9.611 0 ⫺2.711
(2) ⫺3298.8 140.8 310.7 0 ⫺0.4827 0 1.666 ⫺16.577 0
(3) ⫺3305.6 138.0 347.6 ⫺69.74 ⫺1.7816 0 0 ⫺27.575 0
Jadeite NaAlSi2O6 (1) ⫺3029.9 134.7 265.66 138.6 3.294 0 ⫺4.93 0 ⫺3.826
(2) ⫺3025.1 108.6 311.29 0 ⫺5.3503 0 6.6257 ⫺20.051 0
(3) ⫺3029.9 133.5 301.1 101.43 ⫺2.2393 0 0 ⫺20.551 0
Hedembergite CaFeSi2O6 (1) ⫺2838.8 170.3 229.33 341.84 ⫺6.28 0 0 0 0
(3) ⫺2843.4 175.0 310.4 125.70 ⫺1.8460 0 0 ⫺20.400 0
K3 K5 K6 K7
Phase Component Reference H 0f,Tr ,Pr S 0Tr ,Pr K1 K2 ⫻ 104 K4
106 108 102 104
Al-clinopyroxene Al2Al2O6 (1) ⫺3289.8 116.0 160.72 740 0 0 0 0 0
Wollastonite Ca2Si2O6 (1) ⫺3269.5 162.1 280.42 36. 262.81 0 ⫺2.708 0 ⫺3.816
(2) ⫺3263.0 163.6 298.14 0 ⫺7.3186 0 9.687 ⫺13.806 0
(3) ⫺3266.3 163.4 165.1 ⫺18.41 ⫺0.7933 0 0 ⫺11.998 0
Pbca ORTHOPYROXENES
Enstatite Mg2Si2O6 (1) ⫺3092.6 132.5 288.9 37.64 ⫺2.70 0 9.224 0 ⫺3.876
(2) ⫺3091.1 132.3 333.16 0 ⫺4.5412 0 5.583 ⫺2.4012 0
(3) ⫺3089.4 132.5 356.2 ⫺29.90 ⫺0.5969 0 0 ⫺31.853 0
Ferrosilite Fe2Si2O6 (1) ⫺2390.4 189.1 263.78 1.2428 ⫺9.894 0 0.285 0 0.04638
(2) ⫺2388.8 191.8 338.12 0 ⫺4.1942 0 5.850 ⫺23.86 0
(3) ⫺2388.2 192.0 357.4 ⫺27.56 ⫺0.7111 0 0 ⫺29.926 0
Orthodiopside CaMgSi2O6 (1) ⫺3182.1 143.1 221.21 328 ⫺6.586 0 0 0 0
(3) ⫺3192.8 148.5 314.5 0.41 ⫺2.7459 0 0 ⫺20.201 0
Orthohedembergite CaFeSi2O6 (1) ⫺2833.6 169.8 229.33 341.84 ⫺6.28 0 0 0 0
Al-orthopyroxene Al2Al2O6 (1) ⫺3282.0 124.8 160.72 740 0 0 0 0 0
Pbcn ORTHOPYROXENES
Protoenstatite Mg2Si2O6 (1) ⫺3086.2 137.6 246.436 132.44 ⫺24.06 0 39.18 0 1.1197
(2) ⫺3087.9 134.9 333.16 0 ⫺4.5412 0 5.583 ⫺24.012 0
Table 5.37 Lattice energy terms for C2/c pyroxenes. Values in kJ/mole. EBHF ⫽ energy of Born-
Haber-Fayans thermochemical cycle; UL ⫽ lattice energy; EC ⫽ coulombic energy; ER ⫽ repulsive
energy; EDD ⫽ dipole-dipole interactions; EDQ ⫽ dipole quadrupole interactions; H 0f,Tr ,Pr ⫽
enthalpy of formation from the elements at 298.15 K, 1bar reference conditions
Phase Component EBHF H 0f,Tr ,Pr UL EC EDD EDQ ER

Clinoenstatite Mg2Si2O6 30,543.6 ⫺3087.1 a
33,639.1 ⫺39,215.7 ⫺398.0 ⫺122.5 6097.1
⫺3094.2i 33,637.8 6098.4
⫺3088.0b 33,631.6 6104.6
Clinoferrosilite Fe2Si2O6 31,355.9 ⫺2387.2c 33,743.1 ⫺39,254.6 ⫺590.0 ⫺203.8 6305.4
⫺2361.9d 33,717.8 6330.7
Diopside CaMgSi2O6 30,119.6 ⫺3210.8a 33,330.4 ⫺38,878.6 ⫺437.7 ⫺132.6 6118.5
⫺3205.5ef 33,325.1 6123.8
⫺3203.2g 33,322.8 6126.1
⫺3201.8hi 33,321.4 6127.5
Hedembergite CaFeSi2O6 30,525.8 ⫺2849.2j 33,375.0 38,766.0 ⫺533.9 ⫺174.5 6099.3
⫺2838.2gh 33,364.6 6109.7
Ca-Tschermak CaAl2SiO6 28,301.7 ⫺3306.0n 31,604.4 ⫺37,152.9 ⫺464.4 ⫺141.9 6157.8
⫺3280.2ng 31,578.6 6183.6
Jadeite NaAlSi2O6 31,928.5 ⫺3030.7c 35,280.3 ⫺41,115.0 ⫺459.0 ⫺145.5 6439.2
⫺3029.9k 35,279.5 6440.0
⫺3029.4a 35,279.0 6440.5
⫺3023.8i 35,272.9 6446.1
⫺3011.5g 35,261.1 6458.4
⫺3010.8f 35,260.4 6459.0
Acmite NaFeSi2O6 32,174.8 ⫺2593.7l 34,768.5 ⫺40,588.7 ⫺468.3 ⫺150.2 6438.6
CaMnSi2O6 30,304.2 ⫺2947.2m 33,251.4 ⫺38,472.7 ⫺466.5 ⫺144.8 5832.7
Ureyite NaCrSi2O6 32,108.1 ⫺2900.7d 35,008.7 ⫺40,728.4 ⫺470.4 ⫺150.2 6340.2
Ca-Ti-Tschermak CaTiAl2O6 27,244.1 ⫺3553.1d 30,797.2 ⫺36,066.6 ⫺490.1 ⫺153.7 5913.1
a f k
Robie et al. (1978) Naumov et al. (1971) Hemingway et al. (1981)
b g l
Saxena and Chatterjee (1986) Helgeson et al. (1978) Calculated from the enthalpy of
c
Saxena et al. (1986) h
Saxena (1989) formation from the oxides of Viellard
(1982)
d i m
Ottonello et al. (1992) Berman (1988) Newton and McCready (1948)
e j n
Wagman et al. (1981) Stull and Prophet (1971) Chatterjee (1989)
lite facies), pigeonite disappears and a unique miscibility gap extends from the
orthorhombic Pbca polymorph to the C2/c monoclinic form (figure 5.25C). The
gap covers a wide compositional range, from the iron-free join up to Fe/(Fe ⫹
Mg) ⬵ 0.80. For higher relative amounts of Fe, pyroxene is extrinsically unstable
with respect to the olivine plus liquid paragenesis (Huebner, 1982).
In order to assess appropriately the above phase stability relations, we must
rely on experimental evidence concerning the limiting binary join and, particu-
larly, attention must focus on the Mg2Si2O6-CaMgSi2O6 and Fe2Si2O6-CaFeSi2O6
joins. Figure 5.26A shows phase relations in the system Mg2Si2O6-CaMgSi2O6 at
P ⫽ 15 kbar, based on the experiments of Lindsley and Dixon (1976) (LD in
figure 5.26A) and Schweitzer (1977) (S). A wide miscibility gap is evident between
Silicates / 285
Figure 5.25 Phase stability relations for natural pyroxenes in the quadrilateral. (A) Mag-
matic pyroxenes (not reequilibrated at low T). (B) Three-phase region for pyroxenes of
stratified complexes and magmatic series. (C) Pyroxenes reequilibrated in subsolidus con-
ditions. 䡺 ⫽ Pbca orthopyroxene; ∇ ⫽ augite (C2/c); ⌬ ⫽ pigeonite (P21/c); ◊ ⫽ olivine.
From Huebner (1982). Reprinted with permission of The Mineralogical Society of
America.
Figure 5.26 (A) Phase stability relations along binary join Mg2Si2O6-CaMgSi2O6 at P ⫽
15 kbar (anhydrous conditions) based on experiments of Lindsley and Dixon (1976) (LD)
and Schweitzer (1977) (S). En ⫽ enstatite solid mixture (Pbca); Di ⫽ diopside solid mix-
ture (C2/c); Pig ⫽ pigeonite solid mixture (P21/c). From Lindsley (1982). Reprinted with
permission of The Mineralogical Society of America. (B) Schematic representation of
Gibbs free energy properties responsible for opening of a miscibility gap between ortho-
rhombic (opx) and monoclinic (cpx) forms (Saxena, 1973). Coherence is assumed between
P21/c and C2/c polymorphs, which are thus represented by a single Gibbs free energy
curve. Note distinction between “solvus field” (intrinsic instability) and “miscibility gap”
(extrinsic instability).
the orthorhombic (opx) and monoclinic (cpx) polymorphs. This gap is compli-
cated by the appearance at T ⫽ 1450 °C of pigeonite. At T higher than 1450 °C,
two distinct miscibility gaps occur (Pbca-P21 /c and Pbca-C2/c). With increasing
P, the gap widens and the stability field of pigeonite progressively disappears (P ⫽
25 kbar). As shown in part B of figure 5.26, the opx-cpx gap may be conceived
as the result of the combined intrinsic instabilities of the two polymorphs: i.e.,
the “miscibility gap” between cpx and opx covers the two “solvi” valid, respec-
tively, for the opx and cpx phases. A third phase (pigeonite) appears at the T-X
zone of intersection of the two solvus limbs on the Ca-poor side of the binary
join. Lindsley et al. (1981), on the basis of all existing experimental data on the
Mg2Si2O6-CaMgSi2O6 join, proposed a model valid for the pressure range 1 bar
⬍ P ⬍ 40 kbar. This model does not distinguish spatial groups P21 /c and C2/c
and treats the equilibria as stability relations between a generic orthorhombic
phase and a generic monoclinic phase. In this model, the Gibbs free energy curves
for the monoclinic mixture thus imply the existence of a certain amount of elastic
strain energy (to maintain coherence) the nature of which was described in sec-
tion 3.13.
To solve the energy model, Lindsley et al. (1981) based their calculations on the prin-
ciple of equality of chemical potentials of components in mixture for phases at equilibrium
(cf. section 2.3):
Silicates / 287
µ Mg 2 Si 2 O6 , cpx = µ Mg 2 Si 2O6 , opx ( 5.75)
µCaMgSi2 O6 , cpx = µCaMgSi2 O6 , opx ( 5.76 )
applied at every point in the P-T-X space of interest. Application of the usual relations
between chemical potential and thermodynamic activity to identities 5.75 and 5.76 gives
µ Mg 2 Si 2 O6 , cpx = µ Mg
0
2 Si 2 O6 , cpx
+ RT ln a Mg 2 Si 2 O6 , cpx ( 5.77 )
µ Mg 2 Si 2 O6 , opx = µ Mg
0
2 Si 2 O6 , opx
+ RT ln a Mg 2 Si 2 O6 , opx ( 5.78 )
µCaMgSi2 O6 , cpx = µCaMgSi

0
2 O6 , cpx
+ RT ln aCaMgSi2 O6 , cpx ( 5.79 )
µCaMgSi2 O6 , opx = µCaMgSi

0
2 O6 , opx
+ RT ln aCaMgSi2 O6 , opx ( 5.80 )
a Mg 2 Si 2 O6 , opx
∆µ Mg
0
2 Si 2 O6
= RT ln (5.81)
a Mg 2 Si 2 O6 , cpx
aCaMgSi2 O6 , opx
∆µCaMgSi
0
2 O6
= RT ln . (5.82)
aCaMgSi2 O6 , cpx
Adopting for both components the standard state of pure component at the Tr and Pr of
reference, as for a pure component:
µ 0 ≡ GTr , Pr , ( 5.83 )
Table 5.38 Model of Lindsley et al. (1981) for binary system Mg2Si2O6-CaMgSi2O6.
Reaction (a) Mg2Si2O6 (opx) ⇔ Mg2Si2O6 (cpx)

Reaction (b) CaMgSi2O6 (opx) ⇔ CaMgSi2O6 (cpx)
⌬Ha,Tr , P ⫽ 3.561 kJ/mole Mixing model:
{
r
⌬Sa,Tr, P ⫽ 1.91 J/(mole ⫻ K) opx: W ⫽ 25 kJ/mole

r
⌬Va,Tr,P ⫽ 0.0355 J/(bar ⫻ mole) (regular mixture)

r
 WH,12 ⫽ 25.484 kJ/mole

⌬Hb,Tr ,Pr ⫽ ⫺21.178 kJ/mole cpx:  WH,21 ⫽ 31.216 kJ/mole
V,21 ⫽ 0.0812 J/bar ⫻ mole
WWV,12 ⫽ ⫺0.006 J/bar ⫻ mole

⌬Sb,Tr , P ⫽ ⫺8.16 J/(mole ⫻ K)
r
 W12 ⫽ WH,12 ⫹ PWV,12
⌬Vb,Tr , P ⫽ ⫺0.0908 J/(bar ⫻ mole)
r
 with 1 ⫽ Mg2Si2O6, 2 ⫽ CaMgSi2O6
Figure 5.27 Phase stability relations along binary join CaFeSi2O6-Fe2Si2O6 under various
P conditions. From Lindsley (1982). Reprinted with permission of The Mineralogical So-
ciety of America.
Table 5.39 Model of Lindsley (1981) for binary system Fe2Si2O6-CaFeSi2O6.
Reaction (a) Fe2Si2O6 (opx) ⇔ Fe2Si2O6 (cpx)

Reaction (b) CaFeSi2O6 (opx) ⇔ CaFeSi2O6 (cpx)
⌬Ha,Tr ,Pr ⫽ 1.843 kJ/mole Mixing model:
⌬Sa,Tr ,Pr ⫽ 1.68 J/mole ⫻ K
⌬Va,Tr ,Pr ⫽ 0.0369 J/(bar ⫻ mole)
opx:
{ W ⫽ 15 kJ/mole
(regular mixture)
WH,12 ⫽ 16.941 kJ/mole

⌬Hb,Tr ,Pr ⫽ ⫺5.261 kJ/mole cpx: WH,21 ⫽ 20.697 kJ/mole
V,12 ⫽ 0.0059 J/bar ⫻ mole
WWV,21 ⫽ ⫺0.00235 J/bar ⫻ mole

⌬Sb,Tr ,Pr ⫽ 0.53 J/(mole ⫻ K) W12 ⫽ WH,12 ⫹ PWV,12
⌬Vb,Tr ,Pr ⫽ ⫺0.0951 J/(bar ⫻ mole) with 1 ⫽ Fe2Si2O6, 2 ⫽ CaFeSi2O6
we also have
∆µ 0 ≡ ∆GTr , Pr . ( 5.84 )
The ⌬GTr ,Pr of reaction between pure components may be split into enthalpic, entropic,
and volumetric parts:
∆GTr , Pr = ∆HTr , Pr − Tr ∆STr , Pr + Pr ∆VTr , Pr (5.85)
The parameters of the thermodynamic model of Lindsley et al. (1981) are

reported in table 5.38. The mixing properties described are those of a regular
mixture for orthopyroxene and of a subregular mixture (asymmetric with Mar-
Silicates / 289
Figure 5.28 Phase relations in pyroxene quadrilateral for P ⫽ 15 kbar at two subsolidus
temperatures. From Lindsley (1982). Reprinted with permission of The Mineralogical So-
ciety of America.
gules notation) for clinopyroxene. Moreover, for the monoclinic phase, the inter-
action is split into enthalpic and volumetric terms, with no excess entropic term
of mixing. (In the interpretation of Lindsley et al. (1981), because Ca never occurs
in M1 sites, the configurational entropy is that of ideal mixing of Ca and Mg in
M2 sites, with no excess terms.)
Figure 5.27 shows phase stability limits in the (pseudo)binary system
CaFeSi2O6-Fe2Si2O6 in various P conditions, as reported by Lindsley (1982). At
high P (i.e., above 11–12 kbar), the stability relations are similar to those pre-
viously seen for the CaMgSi2O6-Mg2Si2O6 join: there is a miscibility gap between
monoclinic (cpx) and orthorhombic (opx) phases. Moreover, within a limited P
range, pigeonite forms. In the iron-rich side of the join, the stability diagram
shows additional complexities resulting from the extrinsic instability of ferrosilite
with respect to the fayalite plus liquid paragenesis (see also figure 5.24D).
For the binary join CaFeSi2O6-Fe2Si2O6 , Lindsley (1981) proposed a thermo-
Table 5.40 Binary interaction parameters for pyroxenes. Parameters refer to an ionic mixing
model in which n is the number of sites over which permutability is calculated—i.e., Gmixing ⫽
nRT (X1 ln X1 ⫹ X2 ln X2 ). Data in J/mole (H ), J/(mole ⫻ K) (S), and J/(bar ⫻ mole) (V ),
respectively.
Margules Paramters
Mixture H S V n Reference
CLINOPYROXENES (C2/c; P21 /c)
Mg2Si2O6-Fe2Si2O6 W12 ⫽ ⫺960.5 0.041 0.0557 2 (1)
W21 ⫽ 1190.1 ⫺0.081 ⫺0.0180
W12 ⫽ ⫺1558.6 24.389 0.0282 1 (1)
W21 ⫽ 2518.6 24.651 0.0226
Mg2Si2O6-CaMgSi2O6 W12 ⫽ 25,484 0.0 0.0812 1 (2)
W21 ⫽ 31,216 0.0 ⫺0.0061
Mg2Si2O6-CaFeSi2O6 W12 ⫽ 52,971 15.987 0.0416 2 (1)
W21 ⫽ 29,085 3.538 0.0294
W12 ⫽ 93,300 45.0 0.0 1 (3)
W21 ⫽ 20,000 ⫺28.0 0.0
Fe2Si2O6-CaMgSi2O6 W12 ⫽ 46,604 14.996 0.019 2 (1)
W21 ⫽ 22,590 3.940 0.072
W12 ⫽ 15,000 0.0 0.0 1 (3)
W21 ⫽ 24,000 0.0 0.0
Fe2Si2O6-CaFeSi2O6 W12 ⫽ 16,941 0.0 0.0059 1 (4)
W21 ⫽ 20,697 0.0 ⫺0.023
CaMgSi2O6-CaFeSi2O6 W12 ⫽ 13,984 ⫺0.001 0.0068 1 (1)
W21 ⫽ 17,958 0.004 ⫺0.0084
W12 ⫽ 12,000 0.0 0.0 1 (3)
W21 ⫽ 12,000 0.0 0.0
Mg2Si2O6-NaAlSi2O6 W12 ⫽ 30,000 0.0 0.0 1 (5)
W21 ⫽ 30,000 0.0 0.0
CaFeSi2O6-NaAlSi2O6 W12 ⫽ 25,200 17.05 0.0 1 (5)
W21 ⫽ 0.0 1.85 0.0 1 (5)
CaMgSi2O6-CaAl2SiO6 W12 ⫽ 30,918 18.78 0.146 1 (6)
W21 ⫽ 35,060 ⫺44.72 0.469
Redlich-Kister Parameters
CLINOPYROXENES (C2/c; P21 /c)
CaMgSi2O6-NaAlSi2O6 A0 ⫽ 12,600 9.45 0.0 1 (7)
A1 ⫽ 12,600 7.60 0.0
A2 ⫽ ⫺21,400 ⫺16.2 0.0
A0 ⫽ 19,800 ⫺4.4 0.0 1 (8)
A1 ⫽ 9600 9.1 0.0
A2 ⫽ ⫺8200 ⫺16.4 0.0
Silicates / 291
Redlich-Kister Parameters
NaAlSi2O6-CaAl2SiO6 A0 ⫽ 7000 13.0 0.0 1 (8)

A1 ⫽ 9300 0.0 0.0
A2 ⫽ ⫺1100 0.0 0.0
Ideal Mixtures
NaAlSi2O6-NaFeSi2O6 (9)
CaMgSi2O6-NaFeSi2O6 (10)
Mg2Si2O6-Fe2Si2O6 (3)
Margules Parameters
ORTHOPYROXENES (Pbca)
Mg2Si2O6-Fe2Si2O6 W12 ⫽ 13,100 15.0 0.0 1 (1) T ⬍ 873 K
W21 ⫽ 3370 5.0 0.0
Mg2Si2O6-CaMgSi2O6 W12 ⫽ 25,000 0.0 0.0 1 (2)
W21 ⫽ 25,000 0.0 0.0
Mg2Si2O6-CaFeSi2O6 W12 ⫽ 60,000 0.0 0.0 1 (3)
W21 ⫽ 30,000 0.0 0.0
Fe2Si2O6-CaMgSi2O6 W12 ⫽ 16,000 0.0 0.0 1 (3)
W21 ⫽ 20,000 0.0 0.0
Fe2Si2O6-CaFeSi2O6 W12 ⫽ 30,000 0.0 0.0 1 (4)
W21 ⫽ 33,000 0.0 0.0
CaMgSi2O6-CaFe2Si2O6 W12 ⫽ 5000 0.0 0.0 1 (3)
W21 ⫽ 7000 0.0 0.0
Ideal Mixtures
Mg2Si2O6-Fe2Si2O6 T ⬎ 873 K (3)
References:
(1) Ottonello et al. (1992); based on interionic potential calculations
(2) Lindsley et al. (1981); experimental
(3) Saxena et al. (1986); based on phase equilibria
(4) Lindsley et al. (1981); experimental
(5) Chatterjee (1989); based on phase equilibria
(6) Gasparik and Lindsley (1980); estimate on the basis of various experiments
(7) Gasparik (1985); estimate on the basis of various experiments
(8) Cohen (1986); estimate on the basis of various experiments
(9) Ganguly and Saxena (1987); estimate on the basis of various experiments
(10) Ganguly (1973); based on crystal-chemical arguments
Figure 5.29 Excess Gibbs free energy surface at T ⫽ 1000 °C and P ⫽ 1 bar obtained
with Wohl model parameters of table 5.41 and consistent with results of static interionic
potential calculations in C2/c pyroxene quadrilateral. From G. Ottonello, Contributions to
Mineralogy and Petrology, Interactions and mixing properties in the (C2/c) clinopyroxene
quadrilateral, 111, 53–60, figure 6, copyright 䉷 1992 by Springer Verlag. Reprinted with
permission of Springer-Verlag GmbH & Co. KG.
dynamic model similar to that of Lindsley et al. (1981) for the CaMgSi2O6-
Mg2Si2O6 join. The model parameters are listed in table 5.39.
Figure 5.28 shows stability relations in the pyroxene quadrilateral, based on
experimental evidence, at P ⫽ 15 kbar. The miscibility gap contours depend
markedly on the mixing properties observed along the binary joins CaMgSi2O6-
Mg2Si2O6 and CaFeSi2O6-Fe2Si2O6 (cf. figures 5.26 and 5.27). Figure 5.28 con-
firms petrological observations on natural pyroxenes (figure 5.25)—i.e., at high
T the miscibility gap between monoclinic (C2/c) and orthorhombic (Pbca)
phases is small, with the formation, at intermediate compositions, of pigeon-
ite (P21 /c), which in turn causes the formation of two separate miscibility gaps
(C2/c-P21 /c and P21 /c-Pbca); at low T, pigeonite disappears and the single misci-
bility gap Pbca-C2/c expands throughout the compositional field.
Table 5.40 lists some of the binary interaction parameters that may be used
to describe the mixing properties of monoclinic and orthorhombic pyroxenes.
Readers are referred to sections 3.8 and 3.9 for the meanings of these parameters.
Some of the interaction parameters in table 5.40 are deduced from experimental
Silicates / 293
observations; part are assumed or calculated by interionic potential models

(Ottonello et al., 1992) and still require confirmation.
We have seen from the comparison of figures 5.25 and 5.28 that the properties
of the MgSi2O6-CaMgSi2O6-CaFeSi2O6-Fe2Si2O6 quadrilateral in synthetic sys-
tems can be extended to natural pyroxenes as well. It may thus be deduced that
the effect on the phase energetics of minor components in natural mixtures, such
as NaAlSi2O6 , Al2Al2O6 , CaAl2SiO6 , NaFeSi2O6 , NaCrSi2O6 , and LiAlSi2O6 ,
are limited. However, some of these components may predominate in peculiar
parageneses (see table 5.31). In this case, their interaction properties must be
carefully evaluated in order to define the energy of the mixture. A more extensive
discussion on the binary interaction properties of components outside the pyrox-
ene quadrilateral may be found in Ganguly and Saxena (1987).
Concerning multicomponent interactions in the pyroxene quadrilateral, fig-
ure 5.29 shows the excess Gibbs free energy surface generated at T ⫽ 1000 °C
and P ⫽ 1 bar by a Wohl model simulation of calculated interaction energies in
C2/c pyroxenes, on the basis of interionic static potentials (Ottonello, 1992). Ex-
cess energy terms are slightly asymmetric in the compositional field and reach a
maximum at Fe/(Fe ⫹ Mg) 艐 0.4 and Ca/(Fe ⫹ Mg) 艐 0.6.
The adopted Wohl model formulation is
( ) (
Gexcess mixing = X1X2 W12X2 + W21X1 + X1X 3 W13X 3 + W31X1 )
( )
+ X1X 4 W14X 4 + W41X1 + X2X 4 W24X 4 + W42X2 ( )
+ X X (W X
3 4 34 4 +W X )
43 3
1 
+ X1X2X 3  (W12 + W21 + W13 + W31 + W23 + W32 ) − W123 
2 
1 
+ X1X2X 4  (W12 + W21 + W14 + W41 + W24 + W42 ) − W124 
2 
(5.86)
1 
+ X1X 3X 4  (W13 + W31 + W14 + W41 + W34 + W43 ) − W134 
2 
1 
+ X2X 3X 4  (W23 + W32 + W24 + W42 + W34 + W43 ) − W234 
2 
1
(
+ X1X2X 3X 4  W12 + W21 + W13 + W31 + W14 + W41 + W23
2

+ W32 + W24 + W42 + W34 + W43 − W1234 . ) 
The excess terms refer to a strict macroscopic formulation in which the ideal Gibbs free
energy of mixing term is given by
Gideal mixing = −TSideal mixing = RT ∑ X i ln(X i ) , ( 5.87 )

i
Table 5.41 Macroscopic interaction

parameters for C2/c quadrilateral
pyroxenes. Resulting excess Gibbs free
energies are in J/mole (from Ottonello,
1992).
W12 ⫽ ⫺1558.6 ⫺24.398T ⫹ 0.0282P

W21 ⫽ 2518.64 ⫺ 24.651T ⫹ 0.0226P
W13 ⫽ 23,663.1 ⫺ 0.016T ⫹ 0.0208P
W31 ⫽ 32,467.3 ⫹ 0.007T ⫹ 0.0066P
W14 ⫽ 50,774.2 ⫺ 38.907T ⫹ 0.0771P
W41 ⫽ 27,841.2 ⫺ 24.609T ⫹ 0.0510P
W23 ⫽ 49,128.0 ⫺ 41.017T ⫺ 0.0040P
W32 ⫽ 17,326.7 ⫺ 23.620T ⫹ 0.1375P
W34 ⫽ 13,984.2 ⫹ 0.001T ⫹ 0.0068P
W43 ⫽ 17,958.0 ⫺ 0.004T ⫺ 0.0084P
W24 ⫽ 16,109.3 ⫺ 0.033T ⫹ 0.0453P
W42 ⫽ 20,099.8 ⫹ 0.034T ⫹ 0.0530P
W123 ⫽ ⫺810.1 ⫺ 26.123T ⫺ 0.0226P
W124 ⫽ ⫺87,866.4 ⫺ 20.612T ⫹ 0.0188P
W134 ⫽ ⫺39,017.2 ⫺ 34.475T ⫺ 0.0935P
W234 ⫽ 55,293.4 ⫺ 37.629T ⫺ 0.1014P
W1234 ⫽ 74,713.2 ⫺ 316.028T ⫹ 0.5061P
(1) ⫽ Mg2Si2O6; (2) ⫽ Fe2Si2O6; (3) ⫽

CaMgSi2O6; (4) ⫽ CaFeSi2O6; T is in K,
P in bar
where Xi represents molar fractions of the components in the mixture. Using a strictly
macroscopic formulation leads to negative entropic interactions along binary joins in
which intracrystalline disorder occurs in two separate sites (i.e., M1 and M2: cf. table
5.41). A nonnegligible quaternary interaction term results (W1234 in eq. 5.86), because the
system is treated as quaternary whereas it is in fact ternary. The complete set of interaction
parameters of the model is listed in table 5.41.
5.4.6 Exsolutive Processes

Owing to interaction energies among the various compositional terms in mixture
(table 5.40), the principle of minimimization of the Gibbs free energy of the sys-
tem at equilibrium does not allow the phase to remain homogeneous with de-
creasing temperature. Extension and completion of exsolutive processes depend
on the chemistry of the (initially) homogeneous phase and on the kinetics of the
cooling event. The information gained through the study of exsolutive processes
is thus essentially kinetic.
In pyroxenes, exsolutive processes proceed either by nucleation and growth
or by spinodal decomposition (see sections 3.11, 3.12, and 3.13). Figure 5.30B
shows the spinodal field calculated by Saxena (1983) for Ca0.5Mg0.5SiO3 (diop-
side) and MgSiO3 (clinoenstatite) in a binary mixture, by application of the sub-
regular Margules model of Lindsley et al. (1981):
Silicates / 295
Figure 5.30 (A) Simplified Gibbs free energy curves for various polymorphs along
enstatite-diopside join at T ⫽ 1300 °C. (B) Resulting solvus, spinodal field, and miscibility
gap compared with experimental data of McCallister and Yund (1977) on pyroxene un-
mixing kinetics (part B from Ganguly and Saxena (1992). Reprinted with permission of
Springer-Verlag, New York).
W12 = 25, 484 − 0.0812P (J mole ) (5.88)
W21 = 31, 216 − 0.061P (J mole ) (5.89)
where P is expressed in bar.

The spinodal field, calculated at P ⫽ 1 bar on the basis of equations 3.201
and 3.202 (section 3.11) conforms quite well to the experimental observations of
McCallister and Yund (1977), obtained by heating clinopyroxenes to various T

and observing the effects of thermal annealing by transmission electron micros-
copy (TEM). Filled symbols in figure 5.30B indicate unmixing evidences after
heating for 19 to 186 hours. Open circles indicate homogeneous composition (no
evidence of unmixing after prolonged heating), and open squares indicate unmix-
ing after prolonged thermal treatment (i.e., more than 186 hours). Spinodal de-
composition takes place spontaneously (19 to 186 hours, as a function of T )
for initial compositions inside the calculated spinodal field, whereas metastable
persistence of the homogeneous phase is observed outside the field.
To understand fully the nature of the unmixing process, we must consider the
form of the Gibbs free energy curves for T ⫽ 1300 °C for the various polymorphs.
Figure 5.30A shows the Gibbs free energy curves for the orthorhombic phase
(Pbca; G). For the monoclinic polymorph, two separate curves are drawn: one
for a monoclinic phase grouping the two classes C2/c and P21 /c (G⬘ through a
coherence energy term (see section 3.13), and one in the absence of coherence
energy (C2/c; G⬙, strain-free).
Let us now consider a composition C1 intermediate between binodal point b⬘
and spinodal point s⬘ on the G⬘ curve in figure 5.30A. Unmixing requires marked
compositional fluctuation, to overcome the critical nucleation radius of the phase
(cf. section 3.12). The result is a coherent solvus unmixing (or, in the absence of a
coherence energy term, a chemical solvus unmixing). If the compositional fluctua-
tion is sufficiently large, two phases (Pbca and C2/c) form: pigeonite plus diopside
in the first case and orthopyroxene plus diopside in the second. It must be noted
that the intrinsic instability field is located by the tangent to the minima of the
Gibbs free energy curve of the single phase C2/c (G⬙) and that the minimum
Gibbs free energy is indeed achieved here by coexistence (i.e., tangency of the
minima) of the two phases Pbca (G) and C2/c (G⬙, i.e., extrinsic instability), all
other conditions being metastable with respect to this arrangement.
Marked compositional fluctuation requires high ionic diffusivity within the
phase, and we know that elemental diffusivities decrease exponentially with de-
creasing T. The probability of attainment (and completion) of the process is thus
greater at high T than at low T. The process is also enhanced by the presence
of nucleation centers (heterogeneous nucleation) composed of extended defects
(dislocations, grain boundaries, etc.). If the cooling rate of the system is too high
with respect to ionic diffusivity, the process does not take place and the phase
remains metastably homogeneous (as experimentally observed for composition
C1; cf. figure 5.30B).
Let us now consider an initial composition C2. We are within the two spino-
dal points on the Gibbs free energy coherent curve G⬘: decomposition takes place
spontaneously throughout the entire phase, with modulated compositional fluc-
tuations whose amplitudes increase as the process advances. We always obtain
pigeonite plus diopside, although theoretically the orthopyroxene plus diopside
paragenesis, also in this case, corresponds to the minimum energy of the system.
Silicates / 297
Figure 5.31 TTT diagram for a pyroxene of composition similar to C2 in figure 5.30.
Curves a, b, c, and d: cooling rates of a lava flow at increasing depths from surface (a) to
1 m inside the body (d ), based on calculations of Provost and Bottinga (1974). From Bu-
seck et al. (1982). Reprinted with permission of The Mineralogical Society of America.
Exsolution kinetics are appropriately described with the aid of TTT (Time-
Temperature-Transformation) diagrams, in which the cooling rates of the system cross the
kinetic boundaries of the beginning and end of the process. For example, figure 5.31 is a
TTT diagram for a pyroxene with a composition like that of C2 in figure 5.30 (Buseck et
al., 1982). Curves a, b, c, and d are cooling rates calculated by Provost and Bottinga (1974)
for a basaltic lava: curve a is the cooling rate at the surface of the lava flow, and curves b,
c, and d are those at increasing depths beneath the surface. These curves encounter the
kinetic curves of the beginning and end of heterogeneous nucleation (cpx-opx miscibility
gap or chemical solvus in figure 5.30B), homogeneous nucleation (coherent solvus), and
spinodal decomposition (spinodal field) at various values of time and temperature. Note
that the three processes take place under different temperature conditions, times being
equal.
Let us examine the significance of figure 5.31 in some detail: cooling rate a does not
cross any transformation curve; a pyroxene of composition similar to C2 at the surface of
the lava flow thus remains metastably homogeneous throughout the entire cooling process.
Cooling rate b encounters the starting curve for spinodal decomposition but not the curve
of the end of the process; the cooling rate of the system is still too high to allow completion
of the exsolutive process. Cooling rate c encounters the starting curve for heterogeneous
nucleation but not its end curve; the process starts but does not reach completion. At
lower temperatures, cooling rate curve c crosses the TT field of homogeneous nucleation;
a second process of nucleation and growth develops until completion. Cooling rate d,
which corresponds to a depth of about 1 meter beneath the lava flow surface, brings the
process of heterogeneous nucleation to completion.
TEM investigation of pyroxenes identifies the type of exsolutive process as

a function of phase microstructure. Spinodal decomposition initially produces
characteristic “tweed microstructures” [compositional modulations oriented in a
crossed fashion along the (100) and (001) planes] or modulated microstructures
with compositional bands of sinusoidal form whose wavelength depends on the
degree of advancement of the process [plane (001)]. The process then evolves
with the formation of tiny intercalated lamellae (艐 1000 Å), internally coherent
but of contrasting composition (i.e., at opposite sides of the spinodal field). Het-
erogeneous nucleation often develops at grain boundary limits and is character-
ized by incoherent interfaces between phases of different structures. Recognition
of homogeneous nucleation is not so immediate.
Independent of the type of exsolutive process, pyroxene exolutions usually
develop on the (001) and (100) planes, probably because coherence energy is mini-
mal along the axis perpendicular to these planes (development of compositional
fluctuations is maximum along the direction that minimizes the value of G; cf. eq.
3.212; see also Morimoto and Tokonami, 1969, for an example of calculation).
For a more detailed discussion on exsolutive processes in pyroxenes, there is
an excellent synthesis in Buseck et al. (1982).
5.5 Amphiboles
Amphiboles are dipolysilicates (multiplicity ⫽ 2; periodicity ⫽ 3; dimensional-

ity ⫽ 1) and have several structural analogies with pyroxenes, from which they
differ primarily by virtue of a doubling of tetrahedral chains [T4O11 ]n . Like those
of pyroxenes, their cationic structural positions are characterized by the relative
arrangement of the apical and basal parts of tetrahedral groups [SiO4 ] (compare
figures 5.19 and 5.32): sites occurring between two basal parts of neighboring
tetrahedral chains are called M4 (like M2 sites in pyroxenes), and sites between
apical parts of neighboring chains (more restricted) are called M1, M2, and M3
(like M1 sites in pyroxenes). Sites M1, M2, and M3 are octahedrally coordinated,
whereas sites M4 may be either VI-fold or VIII-fold coordinated. The double-
chain arrangement leads to the formation of another nonequivalent position A,
located between two rings formed by the basal parts of opposing neighboring
tetrahedral chains (figure 5.32). OH⫺ groups or halogens (F⫺, Cl ⫺ ) are also lo-
cated at the centers of hexagonal rings at apical levels of tetrahedral chains.
Silicates / 299
Figure 5.32 Structural features of amphiboles. (A) Double chain [T4O11]n seen along axis
c (a) and in perspective (b). (B) Double chain seen from terminal part and various cationic
positions (compare with figure 5.19, for analogies with the pyroxene structure).

The general formula for amphiboles may be expressed as
(
W0−1X 2Y5Z 8 O 22 OH, F, Cl , ) 2
(5.90)
where W is in site A, generally occupied by alkalis (Na, K) or by cationic vacan-

cies V A| ; X is in site M4, generally occupied by univalent and divalent cations; Y
groups the M1, M2, and M3 positions; and Z groups two nonequivalent tetrahe-
dral positions (T1 and T2). Table 5.42 lists the cationic occupancies of major
elements on the various nonequivalent sites of the structure.
Concerning minor and trace elements, in joesmithite Be2⫹ is found in site
T1. Moreover, Ge-amphiboles have been synthesized in which Ge4⫹ completely
replaces Si4⫹ in sites T1 and T2 (Grebenshchikov et al., 1974). Transition ele-
ments Co, Ni, and Cr3⫹ appear to occupy sites M1, M2, M3, and M4, whereas
Zn is stabilized essentially in sites M1, M2, and M3 (Klein and Ito, 1968).
The chemical classification of amphiboles groups the various terms according
to their chemical compositions independent of the represented structural classes.
Table 5.43 furnishes a simplified classification, with the various members
composed of mixtures of Mg,Fe2⫹ and Al,Fe3⫹ terms: for instance, barroisite
is a generic monoclinic (C2/m) mixture CaNa(Mg,Fe)3(Al,Fe3⫹ )2Si7AlO22(OH)2
composed of the end-members aluminous barroisite [CaNaMg3Al2Si7Al
O22(OH)2 ], ferrous-aluminous barroisite [CaNaFe3Al2Si7AlO22(OH)2 ], ferric
barroisite [CaNaMg3Fe3⫹ 2 Si7AlO22(OH)2 ] and ferrous-ferric barroisite [CaNaFe3
3⫹
Fe2 Si7AlO22(OH)2 ].
The structure of amphibole admits a large number of isomorphous substitu-
tions, and the group shows a marked heterogeneity in natural specimens. Table
Table 5.42 Cationic occupancies in amphiboles. (?) ⫽ dubious

assignment
W X Y Z
Cation A M4 M1 M2 M3 T1 T2
Si ⫻ ⫻
Al ⫻ ⫻ ⫻ ⫻ ⫻
Fe3⫹ ⫻ ⫻ ⫻ ⫻(?) ⫻(?)
Ti ⫻ ⫻ ⫻ ⫻(?) ⫻(?)
Fe2⫹ ⫻ ⫻ ⫻ ⫻
Mn ⫻ ⫻ ⫻ ⫻
Mg ⫻ ⫻ ⫻ ⫻
Ca ⫻
Li ⫻ ⫻ ⫻ ⫻ ⫻
Na ⫻ ⫻
K ⫻
Table 5.43 Amphibole terminology and chemical classification. After Hawthorne (1981a), with
modifications. Note that X and Y here do not correspond to identification 5.90.
Name Stoichiometry Structural Class

Fe-Mg-Mn AMPHIBOLES
Anthophyllite
冦 NaX (Mg,Mn,Fe)7⫺Y AlY (AlX⫹Y ,Si8⫺X⫺Y) O22 (OH,F,Cl)2
X ⫹ Y ⬍ 1.00
Pnma
Gedrite
冦 NaX (Mg,Mn,Fe)7⫺Y AlY (AlX⫹Y ,Si8⫺X⫺Y) O22 (OH,F,Cl)2
X ⫹ Y ⱖ 1.00
Pnma
Holmquistite Li2(Mg,Mn,Fe)3(Fe3⫹,Al)2Si8O22(OH,F,Cl)2 Pnma; C2/m

Cummingtonite (Mg,Mn,Fe)7Si8O22(OH,F,Cl)2 P21 /m; C2/m
CALCIC AMPHIBOLES
Actinolite Ca2(Mg,Fe)5Si8O22(OH)2 C2/m
Edenite NaCa2(Mg,Fe)5Si7AlO22(OH)2 C2/m
Pargasite NaCa2(Mg,Fe)4AlSi6Al2O22(OH)2 C2/m
Hastingsite NaCa2(Mg,Fe)4Fe3⫹Si6Al2O22(OH)2 C2/m
Tschermakite Ca2(Mg,Fe)3(Al,Fe3⫹)2Si6Al2O22(OH)2 C2/m
Aluminous hornblende Ca2(Mg,Fe)4AlSi7AlO22(OH)2 C2/m
Kaersutite NaCa2(Mg,Fe)4TiSi6Al2(O,OH)24 C2/m
SODIC-CALCIC AMPHIBOLES
Richterite NaCaNa(Mg,Fe)5Si8O22(OH)2 C2/m
Winchite CaNa(Mg,Fe)4(Al,Fe3⫹)Si8O22(OH)2 C2/m
Barroisite CaNa(Mg,Fe)3(Al,Fe3⫹)2Si7AlO22(OH)2 C2/m
Katophorite NaCaNa(Mg,Fe)4(Al,Fe3⫹)Si7AlO22(OH)2 C2/m
Taramite NaCaNa(Mg,Fe)3(Al,Fe3⫹)2Si6Al2O22(OH)2 C2/m
ALKALINE AMPHIBOLES
Glaucophane Na2(Mg,Fe)3Al2Si8O22(OH)2 C2/m
Riebeckite Na2(Mg,Fe)3Fe23⫹Si8O22(OH)2 C2/m
Eckermannite NaNa2(Mg,Fe)4AlSi8O22(OH)2 C2/m
Arfvedsonite NaNa2(Mg,Fe)4Fe3⫹Si8O22(OH)2 C2/m
Kozulite NaNa2Mn4(Al,Fe3⫹)Si8O22(OH)2 C2/m
Silicates / 301
Table 5.44 Compositions (in weight %) of some natural amphiboles (from Deer et al.,
1983): (1) ⫽ anthophyllite from a serpentinite; (2) gedrite from a gedrite-kyanite-garnet
paragenesis; (3) cummingtonite from a oligoclase-biotite schist; (4) common hornblende from
a tonalite; (5) pargasite from a metamorphic limestone; (6) basaltic hornblende from a latite;
(7) glaucophane from a glaucophane schist.
Samples
Oxide (1) (2) (3) (4) (5) (6) (7)

SiO2 58.48 44.89 51.53 44.99 48.10 45.17 57.73
TiO2 0.03 0.67 0.31 1.46 0.10 2.11 -
Al2O3 0.57 17.91 5.02 11.21 11.05 7.68 12.04
Fe2O3 0.58 0.67 0.82 3.33 0.67 14.30 1.16
FeO 7.85 13.31 16.91 13.17 1.65 2.81 5.41
MnO 0.27 0.37 0.22 0.31 - 0.41 -
MgO 29.25 18.09 20.84 10.41 20.60 13.44 13.02
CaO 0.14 0.40 1.34 12.11 12.50 11.18 1.04
Na2O 0.08 1.45 0.65 0.97 2.54 1.35 6.98
K2O 0.02 0.05 0.00 0.76 1.24 1.09 0.68
H2O(⫹) 2.60 2.02 2.15 1.48 0.71 0.19 2.27
H2O(⫺) 0.20 0.00 0.64 0.04 0.11 0.06 -
F⫺ - - - - 1.90 0.35 -
100.20 99.87 100.43 100.41 101.17 100.14 100.33
O2⫺ ⫹ F⫺ ⫹ Cl⫺ - - - - 0.80 0.14 -
Total 100.20 99.87 100.43 100.41 100.37 100.00 100.33
Ionic fractions on the basis of 24 (O⫽ ⫹ OH⫺ ⫹ F⫺)
Si 7.885 6.325 7.364 6.669 6.760 6.728 7.789

Al 0.090 1.675 0.636 1.331 1.240 1.272 0.211
Total T 7.98 8.00 8.00 8.00 8.00 8.00 8.00

Al - 1.301 0.209 0.629 0.592 0.076 -
Ti 0.003 0.071 0.033 0.163 0.011 0.236 0.118
Fe3⫹ 0.058 0.070 0.087 0.370 0.071 1.602 2.618
Mg 5.876 3.799 4.438 2.300 4.315 2.984 0.611
Fe2⫹ 0.885 1.569 2.022 1.633 0.194 0.350 -
Mn 0.031 0.044 0.027 0.039 - 0.051 -
Na 0.020 0.396 0.180 0.278 0.693 0.390 1.826
Ca 0.020 0.060 0.205 1.923 1.882 1.784 0.150
K 0.004 0.008 - 0.144 0.221 0.208 0.117
Total X ⫹ Y ⫹ W 6.90 7.32 7.20 7.48 7.98 7.68 7.14

⫺
OH 2.338 1.898 2.049 1.462 0.667 0.190 2.043
F⫺ - - - - 0.845 0.165 -
5.44 lists chemical compositions of natural amphiboles (Deer et al., 1983), to-
gether with the corresponding structural formulae on a 24-anion basis (O2⫺,
OH⫺, F⫺, Cl ⫺ ). Because of the complex chemistry of natural amphiboles, calcu-
lation of the structural formula from a complete set of chemical analyses should
begin with an evaluation of the relative amounts of H2O, F, and Cl. If the analysis
Table 5.45 Occurrence of amphiboles in rocks. Note that X and Y here do not correspond
to site identification 5.90.
ORTHORHOMBIC AMPHIBOLES
Name: anthophyllite
Formula: NaX (Mg,Mn,Fe)7⫺Y AlY(AlX⫹Y ,Si8⫺X⫺Y)(OH,F,Cl)2 X ⫹ Y ⬍ 1.00
Occurrence: crystalline schists, peridotites, serpentinites
Name: gedrite
Formula: NaX (Mg,Mn,Fe)7⫺Y AlY(AlX⫹Y ,Si8⫺X⫺Y)(OH,F,Cl)2 X ⫹ Y ⱖ 1.00
Occurrence: amphibolites, eclogites, cordierite-bearing and garnet-bearing gneisses
MONOCLINIC AMPHIBOLES
Name: cummingtonite
Formula: (Mg,Mn,Fe)7Si8O22(OH,F,Cl)2
Occurrences: cornubianites, low-metamorphic schists, micaschists, (iron ores)
Name: actinote (tremolite, actinolite)
Formula: Ca2(Mg,Fe)5Si8O22(OH)2
Occurrence: crystalline schists, calcic cornubianites, cipollines
Name: common hornblende
Formula: NaX Ca2(Mg,Fe)5⫺XAlY (AlX⫹Y ,Si8⫺X⫺Y)O22(OH,F,Cl)2
Occurrence: calc-alkaline granites, sienites, diorites and corresponding effusive terms,
lamprophires, amphibolites, amphibole-bearing hornfels.
Name: pargasite
Formula: NaCa2(Mg,Fe)4AlSi6Al2O22(OH,F,Cl)2
Occurrence: paragneisses, methamorphosed limestones, amphibole peridotites
Name: basaltic hornblende
Formula: NaX Ca2(Mg,Fe)5⫺X(Al,Fe3⫹)Y(AlX⫹Y ,Si8⫺X⫺Y)O22(OH,F,Cl)2
Occurrence: basic volcanites, gabbros, and ultramafic rocks
Name: glaucophane
Formula: Na2(Mg,Fe)3Al2Si8O22(OH,F,Cl)2
Occurrence: glaucophanites, eclogites, prasinites
Name: arfvedsonite
Formula: NaNa2(Mg,Fe)4Fe3⫹Si8O22(OH,F,Cl)2
Occurrence: hypersodic granites, nepheline sienites, metamorphic rocks
is very accurate, one can begin directly with a structural calculation based on 24
anions (O2⫺, OH⫺, F⫺, Cl ⫺ ). However, if there are any doubts about the internal
precision of the data set, it is preferable to assume that the bulk charges of OH⫺,
F⫺, and Cl ⫺ amount to 2 and that the cationic charges are balanced by these two
negative charges plus 22 oxygens (O2⫺ ).
According to Robinson et al. (1982), the correct procedure for assigning site occupan-
cies is as follows.
Sites T (T1 ⫹ T2)

Add Si and Al, followed by Fe3⫹ and Ti4⫹ up to 8. If Si is higher than 8, there is an
analytical problem. If Si ⫹ Al ⫹ Fe3⫹ ⫹ Ti4⫹ does not reach 8, there is again an analyti-
cal problem.
Silicates / 303
Sites Y (M1 ⫹ M2 ⫹ M3)

Add the remaining parts of Al, Ti, and Fe3⫹ with Cr3⫹, Mg, Fe2⫹, Zn, Mn, Ca, Li, and
Na up to 5.0. If Ca and Na are necessary to reach 5.0, this may indicate an analytical
problem, although the presence of Ca in limited amounts in Y sites is reasonable in some
sodic amphiboles (Hawthorne, 1976).
Site X (M4)
Add the remaining parts of Mg, Fe2⫹, Zn, Mn, Ca, Li, and Na in this precise order up to
2.0. If the sum is substantially lower than 2.0, there is an analytical problem or, alterna-
tively, a high number of cationic vacancies on the site.
Site W
Add the remaining part of Ca with Li, Na, and K up to 1.0. If some Ca still remains after
this operation, the analysis is suspect. There are also problems if the sum exceeds 1.0.
Crystal-chemical data indicate that Na ⫹ K occupancies generally vary between 0 and 1,
and it is precisely this variability that sometimes renders structural reconstruction prob-
lematic.
Table 5.45 lists the main occurrences of the various compositional terms. The
name of the mineral reflects the dominant component in the mixture.

The structural classes occurring in amphiboles are those of the monoclinic
(C2/m, P21 /m, P2a) and orthorhombic (Pnma, Pnmn) systems. The most com-
mon space groups are C2/c and P21 /m (monoclinic) and Pnma (orthorhombic;
cf. table 5.43). Table 5.46 lists structural data for some compositional terms be-
longing to two of the spatial groups of interest (C2/m and Pnma). As in pyrox-
enes, note doubling of cell edge a and orthogonalization of cell angle ␤ in the
transition from monoclinic to orthorhombic symmetry. Note also that densities
are on average lower than those of corresponding pyroxene compounds (cf. tables
5.33 and 5.46).
Evaluation of the effects of chemistry on volume properties of amphiboles is
complicated by the high number of sites in which isomorphous substitutions take
place. Limiting ourselves to the main structural classes C2/m, Pnma, and P21 /m,
we present the following evaluations.
C2/m Amphiboles
There are three sites with pseudo-octahedral coordination (M1, M2, and M3)
whose mean dimensions exhibit straight correlation with the ionic radii of cations
occupying these lattice sites. The relationship is complicated by the fact that, for
sites M1 and M3, two of the coordinating anions may be OH⫺, F⫺, or Cl ⫺ instead
of O2⫺. Figure 5.33A shows the relationship between mean 〈M-O〉 distance and
Table 5.46 Structural data on amphibole end-members (from Smyth and Bish, 1988). Molar
volume in cm3/mole; molar weight in g/mole; cell edges in Å; cell volume in Å3.
Compound Cummingtonite Pargasite Tremolite
{
W - Na0.63K0.30 Na0.38K0.12
X Fe1.7Mg0.3 Ca2 Ca1.8Mg0.2
Formula
Y Fe0.6Mg4.4 Fe1.10Mg3.25Al0.55 Mg5
Z Si8 Al1.86Si6.14 Al0.23Si7.8
Molar weight 853.72 870.30 824.57
Density 3.142 3.165 3.010
Molar volume 271.68 274.94 273.93
Z 2 2 2
Class 2/m 2/m 2/m
Group C2/m C2/m C2/m
Cell parameters:
a 9.51 9.910 9.863
b 18.19 18.022 18.048
c 5.33 5.312 5.285
␤ 101.92 105.78 104.79
Volume 902.14 912.96 909.60
Reference Fisher (1966) Robinson et al. Hawthorne and
(1973) Grundy (1976)
Compound Glaucophane Anthophyllite Gedrite
{
W Na0.04 - Na0.05Ca0.03
冧冧
X Na1.80Ca0.20 X X
Formula Fe1.47Mg5.53 Fe1.1Mg4.5Al1.2
Y Fe1Mg2.38Al1.58Ti0.06 Y Y
Z Al0.08Si7.92 Si8 Al1.8Si6.3
Molar weight 822.03 827.23 827.42

Density 3.135 3.111 3.184
Molar volume 262.25 265.88 259.84
Z 2 4 4
System Monoclinic Orthorhombic Orthorhombic
Class 2/m mmm mmm
Group C2/m Pmna Pmna
Cell parameters:
a 9.541 18.560 18.531
b 17.740 18.013 17.741
c 5.295 5.2818 5.249
␤ 103.667 90.0 90.0
Volume 870.83 1765.82 1725.65
Reference Papike and Clark Finger (1970) Papike and Ross
(1968) (1970)
Silicates / 305
Figure 5.33 Correlations between algebraic mean of ionic sizes of intervening ions in M
positions (rM) and mean dimension of M sites (M1, M2, M3) in C2/m (A) and Pnma (B)
amphiboles. From Hawthorn (1981a). Reprinted with permission of The Mineralogical
Society of America.
mean ionic radius of intervening cations. This may be described by the two equa-
tions
M − O = 1.527 + 0.764rM ( Å) ( 5.91)
and
M − O = 1.017 + 0.817rM + 0.354rO 3 ( Å) , ( 5.92 )
where rM is the mean radius (algebraic mean) of cations occupying sites M1,
M2, and M3, and equation 5.92 explicitly takes into account the dimensions of
coordinating O3 anions (rO3 ).
Concerning tetrahedral sites, several authors agree that a straight correlation
exists between the amount of Al in the tetrahedral site and the mean site dimen-
sion. From the equations proposed by Robinson et al. (1973) we may derive
T 1 − O = 0.0013 + 1.6184 X Al,T 1 ( Å) ( 5.93)
T 2 − O = 0.0008 + 1.6292 X Al,T 2 ( Å) , ( 5.94 )
where XAl,T1 is the occupancy of Al 3⫹ cations in tetrahedral site T1. Equations

5.93 and 5.94 have a correlation coefficient R2 ⫽ 0.98. Similar correlations may
be found in Ungaretti (1980). Concerning site M4, Ungaretti et al. (1981) showed
that M4-site occupancy markedly affects cell edges with linear proportionalities
of the type
a c = 1.8653 − 0.0258 X Na,M 4 (R 2 = 0.994) ( 5.95)
b + c = 23.3338 − 0.1647 X Na,M 4 (R 2 = 0.992) . ( 5.96 )
Based on equations 5.91 to 5.96, we may expect C2/m amphibole mixtures to have nearly
ideal behavior in regard to volume properties.
Any deviations from ideal behavior may be attributable to a high site distortion, rep-
resented in linear terms by parameter ⌬:
 
n  l − l 2 
∆ = ∑  l   × 10 4 ,
i m
   (5.97)
i =1
 m
 n 
 
where li is the individual length of a single bond in the polyhedron, lm is the mean length
of polyhedral bonds, and n is the number of bonds in the polyhedron. Site distortion may
also be described in angular terms by parameter ␴ 2:
(Θ − Θm )
2
n
∑
i
σ2 = , (5.98)
i =1 n −1
where ⌰i is the individual bond angle and ⌰m is the ideal bond angle.
As table 5.47 shows, the observed distortions of the various sites are in some cases
quite high and there is also considerable variability in the degrees of polyhedral distortion
of the various compositional terms.
Hawthorne (1976) showed that the distortion parameter of site M2 is a function of
the formal charge on site M4. Because site M2 shows the most severe distortion, the simple
relationships connecting site M4 occupancy and cell edges may be complicated to some
extent by the distortion induced on site M2.
Detailed investigations of the volumes of binary C2/m mixtures with Fe-Mg,

Na-K, and Fe3⫹-Al coupled substitutions invariably show substantial ideality of
mixing. The systems studied comprise:
NaCa2Mg4AlSi6Al2O22(OH)2-NaCa2Fe4AlSi6Al2O22(OH)2
(pargasite–ferro-pargasite; Charles, 1980)
Na2Mg3Al2Si8O22(OH)2-Na2Fe3Fe2Si8O22(OH)2
(glaucophane-riebeckite; Borg, 1967)
Na2CaMg5Si8O22(OH,F)2-KNaCaMg5Si8O22(OH,F)2
(sodium–magnesio-richterite–potassium-magnesio-richterite; Huebner and
Papike, 1970)
Silicates / 307
Na2CaMg5Si8O22(OH,F)2-Na2CaFe5Si8O22(OH,F)2
(magnesio-richterite–ferro-richterite; Charles, 1974)
Na2Mg3Fe2Si8O22(OH)2-Na2Fe3Al2Si8O22(OH)2
(magnesio-riebeckite–ferro-glaucophane; Borg, 1967)
For the first four systems, the experimental data cover the entire compositional
field, whereas for magnesio-riebeckite–ferro-glaucophane, the data are limited to
the field XNa2Mg3Fe2Si8O22(OH)2 ⫽ 1.0 → 0.5. Within the studied compositional
range and within experimental approximation, the volume of the mixture is ideal.
Pnma Amphiboles
In orthorhombic amphiboles, as in monoclinic C2/m amphiboles, site dimensions
are linearly dependent on the ionic radii of occupying cations (figure 5.33B).
However, no empirical equations exist to derive M-O mean dimensions from cat-
ionic occupancy, because of the paucity of experimental information. The same
is true for sites M4, T1, T2, and A.
Concerning cell volume, the experimental data of Popp et al. (1976) on the
system Mg7Si8O22(OH)2-Fe7Si8O22(OH)2 indicate substantial ideality of mixing.
P21 /m Amphiboles
Volume properties in the system Mg7Si8O22(OH)2-Fe7Si8O22(OH)2 (cumming-
tonite-grunerite) show well-defined linearity in the compositional range
XFe7Si8O22(OH)2 ⫽ 0.60 → 1.00 (Klein, 1964; Viswanathan and Ghose, 1965); Ve-
gard’s law probably holds over the entire compositional field.
Investigations of T effects on amphibole volume are complicated by dehy-

droxylation. Experimental data show that loss of OH⫺ groups is favored by the
presence of Fe2⫹, based on equilibrium:
1
Fe 2 + + OH − ⇔ Fe 3+ + O 2 − + H2 (5.99)
2
(as already suggested by Barnes, 1930).

Instead, in magnesian end-members, the configuration (MgMgMg)OH re-
mains stable up to high temperatures (Ernst and Wai, 1970). Detailed studies on
tremolite and other chemically similar amphiboles (Cameron et al., 1973a, b)
show maximum expansion along cell edge a and minimum expansion on cell edge
c (in this respect, amphiboles differ from pyroxenes, which show maximum
expansion along edge b; cf. Sueno et al., 1973).
High-T structural refinements indicate that mean polyhedral expansions
Table 5.47 Polyhedral distortion parameters for C2/m amphiboles

(from Hawthorne, 1981a).
T1 T2 M1
Phase ⌬ ␴2 ⌬ ␴2 ⌬ ␴2
Grunerite 0.29 0.8 0.78 15.8 2.25 36.0
Glaucophane 0.05 0.6 1.88 17.4 0.21 69.5
Tremolite 0.54 5.0 4.14 19.9 0.15 35.6
Fluor-richterite 2.06 14.3 5.60 22.4 0.62 43.1
Fluor-tremolite 0.38 6.2 3.45 19.3 0.01 46.9
Potassic pargasite 0.59 6.6 1.42 19.1 1.25 50.5
Ferro-tschermakite 0.63 4.8 0.61 17.0 4.00 57.3
Potassic ossi-kaersutite 0.29 5.8 1.16 18.9 14.35 47.9
Potassic arfvedsonite 1.05 11.4 5.54 21.4 0.51 44.6
Potassic ferri-taramite 1.25 3.6 1.53 13.8 2.45 45.8
Fluor-riebeckite 0.02 4.4 2.02 12.7 0.29 47.9
Ferro-glaucophane 0.06 0.7 2.04 14.8 0.39 70.5
M2 M3
Phase ⌬ ␴2 ⌬ ␴2
Grunerite 2.79 43.2 0.11 60.9
Glaucophane 15.77 35.0 0.34 85.0
Tremolite 5.52 22.9 0.09 43.6
Fluor-richterite 12.54 33.1 0.97 43.3
Fluor-tremolite 5.76 21.9 1.03 47.9
Potassic pargasite 5.98 24.3 0.01 75.7
Ferro-tschermakite 5.00 19.6 0.33 106.0
Potassic ossi-kaersutite 7.06 31.2 0.05 56.8
Potassic arfvedsonite 17.27 44.5 0.60 67.6
Potassic ferri-taramite 8.53 28.7 0.08 82.3
Fluor-riebeckite 15.48 39.1 0.26 68.7
Ferro-glaucophane 14.36 31.9 1.14 97.4
differ markedly according to coordination number (figure 5.34). In this phase,

too, Si tetrahedra are almost inert to T modifications, whereas octahedral sites
show maximum thermal expansion. Concerning site M4, the structural response
to T modifications is linear for VIII-fold coordination but not for VI-fold coordi-
nation. The contrasting behavior of octahedral and tetrahedral sites results in
decoupling between tetrahedral and octahedral chain moduli, partially compen-
sated by rotation of tetrahedral groups (this type of distortion directly affects the
dimensions of sites A; see figure 5.35).
Heating experiments on P21 /m amphiboles (cummingtonite) show transition
to the C2/m polymorph at low T (50 to 100 °C; Prewitt et al., 1970; Sueno et al.,
1972) in analogy to the behavior of pigeonitic pyroxenes (P21 /c), which show
transition to the form C2/c (cf. section 5.4.3). Phase stability limits are markedly
conditioned by the chemistry of the system and, particularly, by the partial pres-
Silicates / 309
Figure 5.34 Effect of temperature on mean dimensions of various coordination polyhe-

dra in tremolite (C2/m). From Hawthorn (1981a). Reprinted with permission of The Min-
eralogical Society of America.
Figure 5.35 Structure of tremolite projected on (001) plane. Dashed lines on [T4O11]n
tetrahedral chains: displacement (highly exaggerated for illustrative purposes) due to ther-
mal decoupling. From Sueno et al. (1973). Reprinted with permission of The Mineralogi-
cal Society of America.
Table 5.48 Thermal expansion and compressibility of some amphibole end-members according to
Saxena et al. (1993) (1) and Holland and Powell (1990) (2).
Phase Stoichiometry ␣0 ⫻ 105 ␣1 ⫻ 109 ␣2 ⫻ 103 ␣3 Reference

Anthophyllite Mg7Si8O22(OH)2 6.44229 9.4909 4.715 ⫺4.97167 (1)
2.7882 0 0 0 (2)
Fe-anthophyllite Fe7Si8O22(OH)2 2.8345 0 0 0 (2)
Cummingtonite Mg7Si8O22(OH)2 2.7956 0 0 0 (2)
Grunerite Fe7Si8O22(OH)2 10.67 24.8 ⫺30.67 1.6558 (1)
2.8417 0 0 0 (2)
Tremolite Ca2Mg5Si8O22(OH)2 7.46263 5.2307 ⫺13.168 ⫺0.56802 (1)
Fe-tremolite Ca2Fe5Si8O22(OH)2 3.1117 0 0 0 (2)
Glaucophane Na2Mg3Al2Si8O22(OH)2 4.416 9.727 ⫺15.43 2.226 (1)
2.6871 0 0 0 (2)
Fe-glaucophane Na2Fe3Al2Si8O22(OH)2 3.0462 0 0 0 (2)
Edenite NaCa2Mg5AlSi7O22(OH)2 3.1008 0 0 0 (2)
Hornblende Ca2Mg4Al2Si7O22(OH)2 3.0752 0 0 0 (2)
Fe-hornblende Ca2Fe4Al2Si7O22(OH)2 3.1183 0 0 0 (2)
Magnesio-riebeckite Na2Mg3Al2Si8O22(OH)2 2.6907 0 0 0 (2)
Pargasite NaCa2Mg4Al3Si6O22(OH)2 3.0871 0 0 0 (2)
Phase Stoichiometry ␤0 ⫻ 106 ␤1 ⫻ 1010 ␤2 ⫻ 1014 ␤3 ⫻ 1017 Reference

Anthophyllite Mg7Si8O22(OH)2 1.86582 4.3803 2.3045 3.27547 (1)
1.2434 0 0 0 (2)
Fe-anthophyllite Fe7Si8O22(OH)2 1.1481 0 0 0 (2)
Cummingtonite Mg7Si8O22(OH)2 1.2467 0 0 0 (2)
Grunerite Fe7Si8O22(OH)2 1.447 5.042 ⫺25.05 12.35 (1)
1.1511 0 0 0 (2)
Tremolite Ca2Mg5Si8O22(OH)2 1.8257 0.48749 23.491 ⫺2.8108 (1)
Fe-tremolite Ca2Fe5si8O22(OH)2 1.3083 0 0 0 (2)
Glaucophane Na2Mg3Al2Si8O22(OH)2 1.4563 1.1979 8.47 0.42052 (1)
1.1132 0 0 0 (2)
Fe-glaucophane Na2Fe3Al2Si8O22(OH)2 1.1658 0 0 0 (2)
Edenite NaCa2Mg5AlSi7O22(OH)2 1.1074 0 0 0 (2)
Hornblende Ca2Mg4Al2Si7O22(OH)2 1.2968 0 0 0 (2)
Fe-hornblende Ca2Fe4Al2Si7O22(OH)2 1.2903 0 0 0 (2)
Magnesio-riebeckite Na2Mg3Al2Si8O22(OH)2 1.1058 0 0 0 (2)
Pargasite NaCa2Mg4Al3Si6O22(OH)2 1.1025 0 0 0 (2)
sure of fluids (fH2O ) and by redox state (fO2 ). An appropriate discussion of such
limits is beyond the scope of this text, and readers are referred to the work of
Gilbert et al. (1982).
Table 5.48 summarizes thermal expansion and compressibility data for am-
phibole end-members according to the databases of Holland and Powell (1990)
and Saxena et al. (1993). Isobaric thermal expansion ( ␣, K⫺1 ) and isothermal
compressibility ( ␤, bar⫺1 ) may be retrieved from the listed coefficients by applying
the polynomial expansions
Silicates / 311
α (T ) = α 0 + α 1 × T + α 2 × T −1 + α 3 × T −2 (5.100)
and
β ( T ) = β 0 + β1 × T + β 2 × T 2 + β 3 × T 3 . (5.101)
The P-derivative of the bulk modulus (K⬘; cf. section 1.15.2) is 4.0 in all cases.

Thermodynamic data on amphibole end-members are mostly derived from phase
equilibria, because of the difficulty of synthesizing pure components for direct
calorimetric investigation. Table 5.49 lists existing values for some pure compo-
nents. Most data come from the internally consistent set of Holland and Powell
(1990).
5.5.4 Mixing Properties and Intracrystalline Disorder

The chemistry of many natural amphiboles is conveniently described by a compo-
sitional quadrilateral like that seen for pyroxenes. The quadrilateral components
are anthophyllite [Mg7Si8O22(OH)2 ], grunerite [Fe7Si8O22(OH)2 ], tremolite
[Ca2Mg5Si8O22(OH)2 ], and ferro-tremolite [Ca2Fe5Si8O22(OH)2 ] (figure 5.36).
There is complete miscibility among the calcic terms with C2/m structures.
These terms correspond to C2/c calcic pyroxenes (augites). Orthoamphiboles
with Pnma structures (anthophyllite series) are equivalent to Pbca orthopyrox-
enes, although with a much more restricted Fe2⫹ /Mg compositional range. Cum-
mingtonites, with C2/m and P21 /m structures, are analogous to pigeonites with
C2/c or P21 /c structures. As for pyroxenes, amphiboles have an extended miscibil-
ity gap between calcic monoclinic terms and Ca-poor orthorhombic terms. More-
over, the three-pyroxene paragenesis (augite-orthopyroxene-pigeonite; cf. figures
5.25B and 5.28A) is similar to the three-amphibole paragenesis (hornblende-
anthophyllite-cummingtonite) observed in metamorphic terrains. Notwithstand-
ing the above analogies, the chemistry of amphiboles is much more complex than
that of their pyroxene counterparts, and the study of mixing behavior is compli-
cated by the possibility of isomorphous substitutions over a large number of non-
equivalent sites (A, M4, M1, M2, M3, T1, T2).
Besides the amphibole quadrilateral discussed above, a potentially useful in-
terpretative analysis is based on the M4-site population projected on the composi-
tional triangle Na – Ca – (Fe2⫹⫹Mg⫹Mn). A plot of this type groups metamor-
phic amphiboles into three main assemblages (mafic, calcic, and sodic), as shown
in figure 5.37. The same figure also shows the miscibility gaps responsible for this
compositional splitting, together with the tie lines connecting coexisting amphi-
boles at the borders of the miscibility gaps.
Table 5.49 Thermodynamic data of amphibole end-members, Tr ⫽ 298.15 K; Pr ⫽ 1 bar; H f,T
0
,
r Pr
in kJ/mole; S 0Tr ,Pr in J/(mole ⫻ K); V 0Tr ,Pr in cm3/mole. (1) Helgeson et al. (1978); (2) Berman
(1988); (3) Saxena et al. (1993); (4) Holland and Powell (1990). Heat capacity equation: CP ⫽ K1 ⫹
K2T ⫹ K3T ⫺2 ⫹ K4T 2 ⫹ K5T ⫺3 ⫹ K6T ⫺1/2 ⫹ K7T ⫺1.
Phase Stoichiometry Structure H f,0 T r , P r S T0 r , P r V T0 r , P r Reference

Anthophyllite Mg7Si8O22(OH)2 Pnma ⫺12,071.0 535.19 265.44 (3)
⫺12,069 535.20 265.60 (2)
⫺12,064.14 537.0 265.40 (4)
Fe-anthophyllite Fe7Si8O22(OH)2 Pnma ⫺9625.86 729.0 278.70 (4)
Cummingtonite Mg7Si8O22(OH)2 P21/m ⫺12,075.73 542.0 264.70 (4)
Grunerite Fe7Si8O22(OH)2 P21/m ⫺9631.5 714.6 278.73 (3)
683.2 256.67 (1)
⫺9614.14 734.0 278.0 (4)
Tremolite Ca2Mg5Si8O22(OH)2 C2/m ⫺12,305.6 551.2 272.68 (2)
⫺12,302.50 550.0 272.70 (4)
Fluor-tremolite Ca2Mg5Si8O22(F)2 C2/m 541.4 270.45 (1)
Fe-tremolite Ca2Fe5Si8O22(OH)2 C2/m ⫺10,848.9 684.1 282.80 (1)
⫺10,527.10 705.0 282.80 (4)
Edenite NaCa2Mg5AlSi7O22(OH)2 C2/m ⫺12,580.53 599.0 270.90 (4)
Hornblende Ca2Mg4Al2Si7O22(OH)2 C2/m ⫺12,420.29 551.0 269.90 (4)
Fe-hornblende Ca2Fe4Al2Si7O22(OH)2 C2/m ⫺10,999.97 690.0 279.00 (4)
Mg-riebeckite Na2Mg3Al2Si8O22(OH)2 C2/m ⫺11,087.79 602.0 271.30 (4)
Pargasite NaCa2Mg4Al3Si6O22(OH)2 C2/m ⫺12,719.83 591.0 272.10 (4)
Fe-pargasite NaCa2Fe4AlAl2Si6O22(OH)2 C2/m 776.1 279.89 (1)
Glaucophane Na2Mg3Al2Si8O22(OH)2 C2/m 543.9 269.70 (1)
⫺11,963.86 535.0 260.50 (4)
Fe-glaucophane Na2Fe3Al2Si8O22(OH)2 C2/m ⫺10,901.10 624.0 265.90 (4)
Richterite Na2CaMg5Si8O22(OH)2 C2/m 575.7 272.80 (1)
Phase K1 K2 ⫻ 103 K3 ⫻ 10⫺5 K4 K5 ⫻ 10⫺7 K6 ⫻ 10⫺2 K7 ⫻ 10⫺3 Reference
Anthophyllite 1128.7 38.86 50.43 0 ⫺76.42 0 ⫺151 (3)

1219.31 0 ⫺347.661 0 440.09 ⫺57.665 0 (2)
1277.3 25.825 ⫺97.046 0 0 ⫺90.747 0 (4)
Fe-anthophyllite 1383.1 30.669 ⫺42.247 0 0 ⫺112.576 0 (4)
Cummingtonite 1277.3 25.825 ⫺97.046 0 0 ⫺90.747 0 (4)
Grunerite 763.704 356.55 ⫺178.27 0 28.902 0 5.5255 (3)
831.9 254.93 ⫺165.48 0 0 0 0 (1)
1383.1 30.669 ⫺42.247 0 0 ⫺112.576 0 (4)
Tremolite 1229.36 0 ⫺320.899 0 420.881 ⫺64.019 0 (2)
1214.4 26.528 ⫺123.62 0 0 ⫺73.885 0 (4)
Fluor-tremolite 768.8 231.17 ⫺182.97 0 0 0 0 (1)
Fe-tremolite 828.1 246.65 ⫺172.26 0 0 0 0 (1)
1290.0 29.991 ⫺84.475 0 0 ⫺89.47 0 (4)
Edenite 1264.9 24.090 ⫺125.60 0 0 ⫺77.040 0 (4)
Hornblende 1229.6 25.438 ⫺121.635 0 0 ⫺77.503 0 (4)
Fe-hornblende 1290.0 28.209 ⫺90.319 0 0 ⫺89.971 0 (4)
Mg-riebeckite 1701.5 ⫺115.65 70.216 0 0 ⫺185.336 0 (4)
Pargasite 1280.2 22.997 ⫺123.595 0 0 ⫺80.658 0 (4)
Fe-pargasite 893.6 179.87 ⫺197.86 0 0 0 0 (1)
Glaucophane 796.0 249.49 ⫺209.24 0 0 0 0 (1)
1717.5 ⫺121.07 70.75 0 0 ⫺192.72 0 (4)
Fe-glaucophane 1762.9 ⫺118.992 94.237 0 0 ⫺202.071 0 (4)
Richterite 815.0 255.64 ⫺193.09 0 0 0 0 (1)
Silicates / 313
Figure 5.36 Amphibole quadrilateral.
Figure 5.37 Locations of metamorphic amphiboles in compositional triangle Na – Ca –

(Fe2+⫹Mg⫹Mn) (atom proportions). Miscibility gaps are simplified. Solid lines are micro-
probe analyses; dashed lines are wet chemical analyses. From Robinson et al. (1982). Re-
According to Ghose (1982), there are three general conditions (rules) leading
to immiscibility between two amphibole components:
1. Site M4 is occupied in one of the two components by a cation with a

large ionic radius (e.g., Ca2⫹ ) and by cations of much shorter radii (e.g.,
Mg2⫹ and Fe2⫹ ) in the other component (hornblende-cummingtonite or
actinolite-cummingtonite).
Figure 5.38 Structural parameters of amphiboles belonging to various chemical groups.

From Hawthorn (1981a). Reprinted with permission of The Mineralogical Society of
America.
2. Site A is partially or fully occupied in one component and completely

empty in the other (hornblende-actinolite, hornblende-cummingtonite,
gedrite-anthophyllite, richterite-riebeckite).
3. Site M4 is occupied by Ca2⫹ in one component and by Na⫹ in the other
(actinolite-riebeckite).
Based on these rules, mixing properties appear to be strongly conditioned by

the charges and dimensions of intervening cations in positions M4 and A. Indeed,
figure 5.38 shows that structures and dimensions of monoclinic amphiboles are
substantially affected by occupancies on site M4, which appear to condition not
only cell edge a but also angle ␤: the four fields outlined in figure 5.38 are emblem-
atic confirmations of general rules 1 and 3 proposed by Ghose (1982).
According to Robinson et al. (1982), a convenient way of rationalizing mixing
properties of amphiboles in the system SiO2-Al2O3-Fe2O3-TiO2-MgO-FeO-CaO-
Na2O-H2O is to consider tremolite [䡺Ca2Mg5Si8O22(OH)2 ] (where 䡺 symbolizes
a cationic vacancy on site A) as a base component. Tremolite is then transformed
into the other components of interest by a set of independent exchange reactions,
as summarized in table 5.50. The resulting components are 8 (including tremo-
lite). In this compositional field there are 28 binary joins, and calculation of mix-
ing properties in the multicomponent system is thus prohibitive. Even limiting
Silicates / 315
Table 5.50 C2/m amphibole end-member formulation of Robinson et al. (1982).
Base Component
(Tremolite) Exchange Resulting Component
䡺 Ca2Fe5Si8O22(OH)2
䡺 Ca2Mg5Si8O22(OH)2 Fe⇔Mg Ferro-actinolite
䡺 NaCa2Mg5AlSi7O22(OH)2
䡺 Ca2Mg5Si8O22(OH)2 䡺 Si⇔NaAAlIV Edenite
䡺 Ca2Mg3Al2Si6O22(OH)2
䡺 Ca2Mg5Si8O22(OH)2 MgSi⇔AlVIAlIV Tschermak
䡺 Ca2Mg3Fe23⫹Si6Al2O22(OH)2
䡺 Ca2Mg5Si8O22(OH)2 MgSi⇔Fe 3⫹,VI
Al IV
Ferri-Tschermak
䡺 Na2Mg3Al2Si8O22(OH)2
䡺 Ca2Mg5Si8O22(OH)2 CaMg⇔NaM4AlVI Glaucophane
䡺 Ca2Mg5AlSi6TiAlO22(OH)2
䡺 Ca2Mg5Si8O22(OH)2 MgSi2⇔TiIVAlIVAlVI Ti-Tschermak
䡺 Mg7Si8O22(OH)2
䡺 Ca2Mg5Si8O22(OH)2 Ca⇔MgM4 Cummingtonite
ourselves to the calculation of mixing properties in the amphibole quadrilateral,

there are still six binary joins for which mixing properties must be assessed (disre-
garding ternary and quaternary interaction terms).
To our knowledge, direct experimental data on amphibole mixtures have been
obtained only for the (pseudo)binary system actinolite-cummingtonite (Cam-
eron, 1975) at Ptotal ⫽ PH2O ⫽ 2 kbar and for the (pseudo)binary system
tremolite-pargasite at Ptotal ⫽ PH2O ⫽ 1 kbar (Oba, 1980). In both cases, an ex-
tended miscibility gap (or solvus field in the second case), is evident at low T (i.e.,
600 to 800 °C), which is indicative of strong positive interactions in the solid
mixtures. Unmixing of other compositional terms is also evident in microprobe
investigations (see Ghose, 1982 for an appropriate discussion).
Information on amphibole mixing properties comes from intracrystalline and inter-

crystalline Fe2⫹-Mg distribution studies. Mueller (1960, 1961) described the coexistence
of actinolite and cummingtonite in ferriferous metamorphosed bands of Bloom Lake
(Quebec, Canada) based on the exchange equilibrium
2 Fe 5 Si 8O22 (OH )2 + 2 Mg 5 Si 8O 22 (OH )2

1 1
Ca M 4 Y
MgM 4 Y
2 7
Fe-actinolite Mg-cummingtonite
(5.102)
2 Mg 5 Si 8 O22 (OH )2 + Fe 2 Fe 5 Si 8O22 (OH )2 ,
1 1 M4 Y
⇔ Ca M 4 Y
2 7
Mg-actinolite Fe-cummingtonite
where superscripts M4 and Y define the structural position of the element. According
to Mueller (1961), the observed intracrystalline partitioning of Fe2⫹ and Mg2⫹ may be
attributable to ideal mixing behavior in actinolite and to regular mixing behavior in cum-
mingtonite. Later, in order to assess the mixing properties of cummingtonite better,
Mueller (1962) proposed an intracrystalline exchange of the type
Fe M 4 + MgY ⇔ MgM 4 + FeY ( 5.103 )
in which mixing on sites Y (M1, M2, M3) is ideal, but not on site M4. The constant of
equilibrium 5.103 may be expressed as
X Mg,M 4 X Fe,Y γ Mg,M 4 γ Fe,Y

K103 = × , (5.104)
X Fe,M 4 X Mg,Y γ Fe,M 4 γ Mg,Y
where the ␥ terms are site-activity coefficients.

Adopting the model of Mueller (1962), equation 5.104 reduces to
X Mg,M 4 X Fe,Y γ Mg,M 4

K103 = × . (5.105)
X Fe,M 4 X Mg,Y γ Fe,M 4
In a regular mixture, the activity coefficient of component i is associated with the

interaction parameter W through
W
(1 − X i )
2
ln γ i = (5.106)
RT
thus equation 5.104 may be reexpressed in terms of interaction parameters on sites M4

and Y (WM4 , WY ) as
 
(
X Fe,Y 1 − X Fe,M 4 ) 
(
exp  1 − 2X Fe,Y )W Y
RT 
K103 = × .
(
X Fe,M 4 1 − X Fe,Y ) 
(
exp  1 − 2X Fe,M 4 ) WM 4  (5.107)
 RT 
Hafner and Ghose (1971) solved equation 5.107 on the basis of Mueller’s (1960) data
and obtained K103 ⫽ 0.19; WM4 /RT ⫽ 1.58, and WY /RT ⫽ 0.54. Their results thus imply
a certain nonideality of mixing on sites Y as well.
Ghose and Weildner (1971, 1972) experimentally studied the intracrystalline distribu-
tion of Fe2⫹-Mg2⫹ in natural cummingtonites annealed at various T conditions and Pto-
tal ⫽ 2 kbar in the presence of H2O (3 weight %). The observed distribution of Fe
2⫹
between M4 and Y sites may be due either to regular mixing (with K103 ⫽ 0.26, WM4 ⫽
1.09 kcal/mole, WY ⫽ 1.23 kcal/mole at T ⫽ 600 °C; cf. Ghose and Weidner, 1971; and
figure 5.39A) or to ideal mixing on both M4 and Y sites, as shown in figure 5.39B.
Silicates / 317
Figure 5.39 Intracrystalline distribution of Fe2+-Mg2+ between Y (M1, M2, M3) and M4
sites in natural cummingtonites. (A) Regular site-mixing model. (B) Ideal site-mixing
model. (C) Arrhenius plot of equilibrium constant.
As we see by comparing figure 5.39A and B, the choice of the site-mixing

model is, within certain limits, arbitrary: an ideal model gives hyperbolic distribu-
tion curves such as those shown in figure 5.39B. Any nonzero site interaction
parameters perturb the hyperbolic distribution and induce fluctuations whose
magnitude depends directly on the absolute magnitude of W. The choice of an
appropriate mixing model thus rests on the sensitivity of the experimenter.
Plotting on an Arrhenius diagram (figure 5.39C) the equilibrium constants
measured on several natural cummingtonites, Ghose and Weidner (1972) deduced
the standard enthalpy variation associated with the intracrystalline exchange
5.103 ( ⌬H 0 ⫽ 3.6 kcal/mole). Figure 5.39C may be utilized as a potential geother-
mometer to define closure temperatures for intracrystalline exchanges in natural
cummingtonites.
In the example above, we did not make any distinction among M1, M2, and M3 site
populations, because these sites were grouped in the Y notation. Studying in detail the
structure of sodic amphiboles of the riebeckite-glaucophane series, Ungaretti et al. (1978)
observed that the Fe2⫹-Mg2⫹ population is not randomly distributed between M1 and M3
sites, being Fe2⫹ preferentially stabilized in M3. Considering the intracrystalline exchange
Fe 2M+3 + MgM1 ⇔ Fe 2M+1 + MgM 3 , ( 5.108 )
the resulting equilibrium constant, based on the simply hyperbolic distribution of cations
(figure 5.40), may be reconducted to an ideal site-mixing behavior:
K108 =
a Mg,M 3 aFe,M1
≡
(1 − X Fe,M 3 )X Fe,M1
= 0.403.
aFe,M 3 a Mg,M1 (
X Fe,M 3 1 − X Fe,M1 ) (5.109)
Figure 5.40 Intracrystalline distribution of Fe2+-Mg2+ between M1 and M3 sites in sodic

amphiboles. Reprinted from Ungaretti et al. (1978) with author’s permission.
We recall to this purpose that “ideal site mixing” does not necessarily mean that the mix-
ture as a whole is ideal, but that the intracrystalline site distribution may be described by
a constant (without the necessity of introducing site-interaction parameters; cf. equation
5.109, for example). Generally speaking, if the adopted constant differs from 1 (as in the
case of sodic amphiboles in figure 5.40 and of cummingtonites in figure 5.39), the mixture
as a whole is obviously nonideal because site permutability does not reach the maximum
value attained by the ideal condition (cf. section 3.8.1).
The nature of the miscibility gap between orthorhombic and monoclinic am-
phiboles has been recently reconsidered by Will and Powell (1992) with an ex-
tended application of Darken’s Quadratic Formalism (Darken, 1967). This for-
malism assumes that, within a restricted compositional field (“simple regions”),
excess energy terms may be described by simple mixing models (regular mixture,
ideal mixture), even if the intermediate compositions necessitate more complex
treatment. Under this proviso, a general binary join may be considered to be
composed of two simple regions, each including one end-member, connected by
a transitional zone, as depicted schematically in figure 5.41. The complex activity-
composition relationship observed within the simple region is then reconducted
to a regular (or even ideal) mixture between the real end-member in the simple
region and a “fictive” end-member on the opposite side of the join, by appropriate
selection of the standard state properties of the fictive term.
Assuming ideal mixing within the simple regions, the principle of equality of
potentials between phases at equilibrium reduces to
Silicates / 319
Figure 5.41 G-X diagram illustrating application of Darken’s Quadratic Formalism to a

binary join. Although mixing behavior of components 1 and 2 in phases ␣ and ␤ is noni-
deal (heavy lines), in each of the simple regions it is modeled by an ideal mixing model
(light lines) by means of an appropriate choice of the fictive standard state potential ␮*.
From Will and Powell (1992). Reprinted with permission of The Mineralogical Society
of America.
∆µ i0 = − RT ln K D,i = µ i*,α − µ i0, β , (5.110)
where ␮t,*␣ is the fictive standard state chemical potential of component i in phase
␣, and KD,i is the distribution coefficient KD,i ⫽ Xi,␣ /Xi,␤ .
Consideration of the actual extension of miscibility gaps in natural systems
led Will and Powell (1992) to establish systematic relationships between the actual
free energy of pure components and their fictive potentials in the phases of inter-
est, as listed in table 5.51. Note that, amphiboles being multisite phases, their
ideal activity in a chemically complex phase is expressed in terms of multiple
product of site ionic fractions (see section 3.8.7). For anthophyllite
(䡺Mg2Mg3Mg2Si4Si4O22(OH)2 ), for instance, we have
aanthophyllite ≡ X anthophyllite = X ▫, A X Mg,

2 3 2 4
M 4 X Mg,M1 − M 3 X Mg,M 2 X Si,T 2 ( 5.111)
(it is assumed that Al-Si mixing takes place only on T 2 sites and that M1 to M3
sites are energetically equivalent).
Table 5.51 Energy terms relating standard state chemical potentials of pure components (␮0i,␣) to fictive potentials in the
host phase (␮*
i,␤) in the Will and Powell (1992) application of Darken’s Quadratic Formalism to amphiboles.
Cummingtonite-actinolite series Calcic-sodic amphibole series Orthoamphibole-actinolite series

T ⬵ 650 °C T ⬵ 500 °C T ⬵ 650 °C
␮*cumm,act ⫽ ␮0cumm,cumm ⫹ 51.5 (⫾4.5) ␮*gl,act ⫽ ␮0gl,gl ⫹ 48.7 (⫾6.7) (low P) ␮*anth,act ⫽ ␮0anth,oam ⫹ 35.4 (⫾4.0)
␮*tr,cumm ⫽ ␮0tr,act ⫹ 16.0 (⫾3.8) ␮*gl,act ⫽ ␮0gl,gl ⫹ 31.8 (⫾2.9) (high P) ␮*tr,oam ⫽ ␮0tr,act ⫹ 34.5 (⫾6.5)
␮*hb,cumm ⫽ ␮0hb,act ⫹ 62.6 (⫾8.6) ␮*rich,act ⫽ ␮0rich,gl ⫹ 25.0 (⫾3.0) ␮*hb,oam ⫽ ␮0hb,act ⫹ 40.1 (⫾4.8)
␮*ed,cumm ⫽ ␮0ed,act ⫹ 41.1 (⫾3.7) ␮*tr,gl ⫽ ␮0tr,act ⫹ 46.3 (⫾3.4) ␮*ed,oam ⫽ ␮0ed,act ⫹ 44.3 (⫾4.1)
␮*parg,cumm ⫽ ␮0parg,act ⫹ 12.6 (⫾1.3) ␮*hb,gl ⫽ ␮0hb,act ⫹ 32.1 (⫾2.0) ␮*parg,oam ⫽ ␮0parg,act ⫹ 54.5 (⫾4.4)
␮*ed,gl ⫽ ␮0ed,act ⫹ 59.8 (⫾3.8)
Orthoamphibole-actinolite series ␮*parg,gl ⫽ ␮0parg,act ⫹ 46.0 (⫾2.5)
T ⬵ 650 °C
cumm ⫽ cummingtonite (P21/m) 䡺 Mg2Mg3Mg2Si4Si4O22(OH)2
tr ⫽ tremolite (C2/m) 䡺 Ca2Mg3Mg2Si4Si4O22(OH)2
hb ⫽ hornblende (C2/m) 䡺 Ca2Mg3(MgAl)(Si3Al)Si4O22(OH)2
ed ⫽ edenite (C2/m) NaCa2Mg3Mg2(Si3Al)Si4O22(OH)2
parg ⫽ pargasite (C2/m) NaCa2Mg3(MgAl)(Si2Al2)Si4O22(OH)2
gl ⫽ glaucophane (C2/m) 䡺 Na2Mg3(AlAl)Si4Si4O22(OH)2
rich ⫽ richterite (C2/m) Na(CaNa)Mg3Mg2Si4Si4O22(OH)2
anth ⫽ anthophyllite (Pnma) 䡺 Mg2Mg3Mg2Si4Si4O22(OH)2
act ⫽ actinolite; oam ⫽ orthoamphibole
Silicates / 321
5.6 Micas
Micas are diphyllosilicates (multiplicity ⫽ 2, periodicity ⫽ 3, dimensionality ⫽

2). They are composed of mixed sheets in a 2:1 ratio (“talc layer”) containing two
tetrahedral sheets [T2O5 ]n interposed by one octahedral sheet. The tetrahedral
sheets are formed from three shared corners of the tetrahedral groups [SiO4 ],
which results in a bidimensional lattice with hexagonal meshes, as shown in figure
5.42A. At the center of each hexagon is a hydroxyl group OH or F (or S in
anandite). Figure 5.42B shows the stacking that leads to the mixed 2:1 talc layer.
The apexes of the tetrahedral groups are opposed to form junction planes in com-
mon with the octrahedral sheet.
Micas are classified according to the charge of the mixed layer per formula
unit. This depends on the diadochy level of Al 3⫹ (or Fe3⫹ or, more rarely, Be2⫹ )
with Si4⫹ in tetrahedral groups, and on the charge (and occupancy) of the cation
in the intermediate octahedral sheet.
In each 2:1 mixed layer, the upper tetrahedral sheet is translated by a/3 with
respect to the lower one, thus creating the octahedral oxygen coordination around
the cations of the intermediate sheet. Translation may occur along any positive or
negative direction defined by structural axes X1 , X2 , and X3 of a pseudohexagonal
lattice, as shown in figure 5.43.
The degree of freedom in the translational properties of the various sheets
gives the polytypism of the phase. (Polytypes are structures with layers of essen-
tially identical composition but differing in translational sequence. Polytypes usu-
ally have different symmetry and periodicity along axis Z). Smith and Yoder
Figure 5.42 Structural scheme of micas. (A) Tetrahedral sheet with tetrahedral apexes
directed upward. (B) Structure of mixed layers along axis Y; a, b, and c are edges of ele-
mentary cell unit.
Figure 5.43 Sliding between tetrahedral sheets of mixed 2:1 talc layer. (a) Set of positions
1 occupied. (b) Set of positions II occupied. From Bailey (1984a). Reprinted with permis-
sion of The Mineralogical Society of America.
(1956) have shown that six standard ideal polytypes may be formed, with period-
icity between one and six layers: 1M, space group C2/m, ␤ ⫽ 100; 2M, space
group C2/c, ␤ ⫽ 95; 3T, space group P3112; 2Or, space group Ccmm, ␤ ⫽ 90;
2M2, space group C2/c, ␤ ⫽ 98; 6H, space group P6122 (first number identifies
periodicity—i.e., polytype 3T ⇒ periodicity ⫽ 3; polytype 1M ⇒ periodicity ⫽
1, etc.). In trioctahedral micas, the most frequent polytype is 1M, followed by
3T and 2M1. In dioctahedral micas, the most frequent is 2M1, followed by
1M and 3T (Bailey, 1984a). The diadochy of Al (of Fe3⫹ ) on Si4⫹ normally
has the following ratios: 0:4, 0.5:3.5, 1:3, 2:2, and 3:1. Thus, polyanionic
groups composed of two tetrahedral sheets may be of these types:
[(OH)2Si4O10 ]6⫺; [(OH)2Al0.5Si3.5O10 ]6.5⫺; [(OH)2AlSi3O10 ]7⫺; [(OH)2Al2Si2O10 ]8⫺;
[(OH)2Al3SiO10 ]9⫺.
The intermediate octahedral sheet is normally made up of cations of charge
2 or 3 (Mg, Al, Fe2⫹, Fe3⫹, or, more rarely, V, Cr, Mn, Co, Ni, Cu, Zn), but in
some cases cations of charge 1 (Li) and 4 (Ti) are also found. In the infinite
octahedral sheet, formed by the sharing of six corners of each octahedron, there
may be full occupancy of all octahedral sites (“trioctahedral micas”); alterna-
tively, one site out of three may be vacant (“dioctahedral micas”). Nevertheless,
the primary classification of micas is based on the net charge of the mixed 2:1
layer. In “common micas” this charge is close to 1, whereas in “brittle micas” it
Silicates / 323
is close to 2. In common (or “true”) micas, the interlayer cations (coordination

XII) counterbalancing the net charge of the mixed layers are normally K or Na
(or, more rarely, Rb, Cs, NH4 , or H3O); in brittle micas, such cations are Ca or
Ba (or, more rarely, Sr). Table 5.52 lists the structural characters of some tri- and
dioctahedral micas. Note the doubling of cell edge c from polytype 1M to poly-
types 2M1 and 2M2, resulting in doubling of the cell volume and modification of
the angle ␤.
Table 5.53 lists the general classification of micas with their main composi-
tional terms. Stoichiometry obeys the general formula XY2–3Z4O10 (OH,F)2 ,
where X ⫽ interlayer cations, Y ⫽ octahedrally coordinated cations of the 2:1
mixed layer, and Z ⫽ tetrahedrally coordinated cations of the 2:1 mixed layer. It
must be noted that several compositional terms are indeed solid mixtures of more
elementary components. In particular, glauconite has a complex chemistry and
an Al:Si diadochy of 0.33:3.67. (R2⫹ and R3⫹ terms in table 5.53 identify generic
divalent and trivalent cations, respectively.)

Micas are primary phases in several types of igneous, metamorphic, and sedimen-
tary rocks. Table 5.54 furnishes a rough scheme of main mica occurrences.
The main terms of trioctahedral common micas are phlogopite and biotite. Table 5.53
represents biotite as a mixture of Fe2⫹- and Mg2⫹-bearing terms, in variable proportions
on site Y, and phlogopite as a pure end-member. Current nomenclature is somewhat arbi-
trary; trioctahedral common micas with Mg:Fe2⫹ ratios less than 2:1 are generically de-
fined as “biotites,” trioctahedral common micas with Mg:Fe2⫹ ratios higher than 2:1 are
called “phlogopites.” Phlogopite is a marker of thermometamorphism in magnesian lime-
stones, where it is formed by reaction between dolomite and potash feldspar in hydrous
conditions, or by reaction between muscovite and dolomite according to
3CaMg(CO3 )2 + KAlSi 3O8 + H2O ⇔ KMg3(Si 3Al)O10(OH)2 + 3CaCO3 + 3CO2

( 5.112.1)
dolomite K - feldspar phlogopite calcite
and
4CaMg(CO3)2 + KAl2(Si3Al)O10(OH)2 ⇔ 4CaCO3 + KMg3(Si3Al)O10(OH)2 + MgAl2O4 + 4CO2

dolomite muscovite calcite phlogopite spinel
( 5.112.2)
Moreover, phlogopite is a primary mineral in leucite-bearing rocks and also sometimes

occurs in ultramafic rocks. Biotite is a primary phase in several types of igneous and meta-
morphic rocks. In intrusive igneous rocks, it is present in granites, granodiorites, tonalites,
diorites, norites, and nepheline-syenites. In effusive rocks it is less frequent, but it is still
sometimes present in ryolites, trachytes, dacites, latites, andesites, and basalts, almost in-
variantly partially altered, with evident resorption phenomena and reaction rims. The
Table 5.52 Structural parameters for some trioctahedral and dioctahedral micas
(from Smyth and Bish, 1988).
TRIOCTAHEDRAL MICAS
Name Phlogopite Polylithionite Zinnwaldite
Polytype 1M 1M 1M
Interlayer X cations K0.77Na0.16Ba0.05 K0.89Na0.06Rb0.05 K0.9Na0.05
Octahedral Y cations Mg3 Al1.3Li1.7 Al1.05Fe0.93Li0.67
Tetrahedral Z cations Al1.05Si2.95 Al0.6Si3.4 Al0.91Si3.09
Hydroxyls and fluorine (OH)0.7F1.3 (OH)0.5F1.5 (OH)0.8F1.2
Molar weight (g/mole) 421.831 396.935 428.995
Molar volume (cm3/mole) 146.869 140.5 144.55
Density (g/cm3) 2.872 2.825 2.986
Formula units per unit cell 2 2 2
Class 2/m 2/m 2
Group C2/m C2/m C2
Cell parameters (Å):
a 5.3078 5.20 5.296
b 9.1901 9.01 9.140
c 10.1547 10.09 10.096
␤ 100.08 99.28 100.83
Volume (Å3) 487.69 466.6 480.00
Reference Hazen and Sartori (1976) Guggenheim and
Burnham (1973) Bailey (1977)
DIOCTAHEDRAL MICAS
Name Muscovite Paragonite Margarite

Polytype 2M1 2M1 2M2
Interlayer X cations K Na0.92K0.04Ca0.02 Na0.19K0.01Ca0.81
Octahedral Y cations Al2 Al1.99Fe0.03Mg0.01 Al2
Tetrahedral Z cations Al1Si3 Al1.06Si2.94 Al1.89Si2.11
Hydroxyls and fluorine (OH)2 (OH)2 (OH)2
Molar weight (g/mole) 398.314 384.31 395.91
Molar volume (cm3/mole) 140.57 132.13 129.33
Density (g/cm3) 2.834 2.909 3.061
Formula units per unit cell 4 4 4
Class 2/m 2/m m
Group C2/c C2/c Cc
Cell parameters (Å):
a 5.1918 5.128 5.1038
b 9.1053 8.898 8.8287
c 20.0157 19.287 19.148
␤ 95.735 94.35 95.46
Volume (Å3) 933.56 877.51 858.89
Reference Rothbauer (1971) Lin and Bailey (1984) Guggenheim and
Bailey (1975, 1978)
Silicates / 325
Table 5.53 Classification and compositional terms of micas (from Bailey 1984a; modified).
Subgroup Species Stoichiometry
COMMON MICAS
Trioctahedral Phlogopite KMg3(Si3Al)O10(OH,F)2
Na-phlogopite NaMg3(Si3Al)O10(OH,F)2
Biotite K(Mg,Fe)3(Si3Al)O10(OH,F)2
Annite KFe32⫹(Si3Al)O10(OH,F)2
Ferri-annite KFe32⫹(Si3Fe3⫹)O10(OH,F)2
Polylithionite K(Li2Al)Si4O10(OH,F)2
(ex-lepidolite)
Taeniolite K(Mg2Li)Si4O10(OH,F)2
Siderophyllite K(Fe2⫹Al)(Si2Al2)O10(OH,F)2
Zinnwaldite K(Li,Fe,Al)3(Si,Al)4O10(OH,F)2
Muscovite KAl2(Si3Al)O10(OH,F)2
Paragonite NaAl2(Si3Al)O10(OH,F)2
Phengite K[Al1.5(Mg,Fe2⫹)0.5](Si3.5Al0.5)O10(OH,F)2
2⫹
Dioctahedral Illite K0.75(Al1.75R0.25 )(Si3.5Al0.5)O10(OH,F)2
Celadonite K(Mg,Fe,Al)2Si4O10(OH,F)2
2⫹ 3⫹
Glauconite K(R1.33 R0.67 )(Si3.67Al0.33)O10(OH,F)2
BRITTLE MICAS
Trioctahedral Clintonite Ca(Mg2Al)(SiAl3)O10(OH,F)2
Bityite Ca(Al2Li)(Si2AlBe)O10(OH,F)2
Anandite BaFe32⫹(Si3Fe3⫹)O10(OH)S
Dioctahedral Margarite CaAl2(Si2Al2)O10(OH,F)2
biotites of effusive rocks are also generally richer in Fe3⫹ and Ti (and poorer in Fe2⫹ ) with
respect to biotites of intrusive counterparts.
In thermal metamorphism, biotite occurs in the clorite-sericite facies as a dispersed
phase in the argillitic matrix and is stable up to low-grade cornubianites. In regional meta-
morphism, biotite is typical of argillitic and pelitic rocks up to the staurolite-garnet facies
(biotite, biotite-sericite, biotite-chlorite, and albite-biotite schists, and garnet-staurolite
micaschists).
Muscovite is the main term of the common dioctahedral micas, and it is found in
acidic intrusive rocks (biotite and biotite-muscovite granites), although in subordinate
quantities with respect to biotite. It is a common mineral in aplitic rocks and is peculiar to
fluorine metasomatism in the contact zones between granites and slates (“greisenization”).
In metamorphic rocks, muscovite occurs in low-grade terrains of the regional meta-
morphism (albite-chlorite-sericite schists). It must be noted here that the term “sericite”
identifies fine-grained white micas (muscovite, paragonite).
Table 5.55 lists analyses of natural micas taken from the collection of Deer et
al. (1983).
5.6.2 Intrinsic Stability Limits

The intrinsic stability of micas as a function of P and T intensive variables and
chemistry is essentially related to the capability of the tetrahedral and octahedral
Table 5.54 Occurrences of main compositional terms of mica in igneous, metamorphic and
sedimentary rocks.
Term Igneous Rocks Metamorphic Rocks Sedimentary Rocks

Phlogopite Peridotites Metamorphosed
limestones, dolomites
Biotite Gabbros, norites, Phyllites, chlorite-sericite
diorites, granites, schists, biotite facies
pegmatites of the regional
metamorphism, gneisses
Muscovite Granites, aplites Phyllites, schists, gneisses Detritic and
authigenic sediments
Zinnwaldite Pegmatites and high-T
Polylithionite hydrothermal veins
Glauconite Marine sediments
(green sands)
sheets of the 2:1 mixed layer to accommodate the metal-to-oxygen interionic dis-
tances, with positional rotations of tetrahedra on the basal plane of oxygens, as
discussed by Hazen and Wones (1972).
Figure 5.44 shows the conformation of the hexagonal rings on the tetrahedral
sheet for limiting values of the rotational angle ( ␣r ⫽ 0° and 12°, respectively).
Due to structural constraints, the value of ␣r depends directly on the mean cation-
to-oxygen distances in [TO4 ] tetrahedra (dt ) and in octahedra (do ), according to
3 3 d o sin ψ
cos α r = , (5.113)
4 2 dt
where ␺ is the “octahedral flattening” angle (cf. Hazen and Finger, 1982), which
is also a function of do :
sin ψ ≈ 1.154 − 0.144d o . (5.114)
As already noted, the structural limits of the rotational angle are 0° and 12°.
This implies that the ratio of mean octahedral and tetrahedral distances is re-
stricted in the range
do
1.235 ≤ ≤ 1.275 (5.115)
dt
(Hazen and Finger, 1982). Mean do and dt distances vary as a function of the
ionic radii of intervening cations and, chemical composition being equal, vary
Silicates / 327
Table 5.55 Chemical analyses of natural micas (from Deer et al., 1983). Note that ionic
fractions are retrieved on a 24-anion basis—i.e. double the canonical formula. (1) Muscovite
from a low-grade metamorphic prasinite schist; (2) glauconite from a sandstone; (3)
phlogopite from a marble; (4) biotite from a quartz-bearing latite; (5) lepidolite from a
pegmatite.
Oxide (1) (2) (3) (4) (5)

SiO2 48.42 49.29 40.95 39.14 49.80
TiO2 0.87 0.12 0.82 4.27 0.00
Al2O3 27.16 3.17 17.28 13.10 25.56
Fe2O3 6.57 21.72 0.43 12.94 0.08
FeO 0.81 3.19 2.38 5.05 0.00
MnO - Traces Traces 0.14 0.38
MgO Traces 3.85 22.95 12.75 0.22
CaO Traces 0.74 0.00 1.64 0.00
Na2O 0.35 0.12 0.16 0.70 0.40
K2O 11.23 6.02 9.80 6.55 9.67
F Traces - 0.62 1.11 6.85
H2O(⫹) 4.31 7.21 4.23 2.41 0.38
H2O(⫺) 0.19 4.60 0.48 0.58 0.50
99.91 100.35 100.13 100.38 102.96

O⬅F - - 0.26 0.46 2.89
Total 99.91 100.35 99.87 99.92 100.07

IONIC FRACTIONS ON A 24-ANION BASIS (O,OH,F)
Si 6.597 7.634 5.724 5.790 6.750
Al 1.403 0.366 2.276 2.210 1.250
Total Z 8.00 8.00 8.00 8.00 8.00

Al 2.959 0.213 0.562 0.074 2.834
Ti 0.089 0.014 0.084 0.474 -
Fe3⫹ 0.672 2.532 0.340 1.440 0.008
Fe2⫹ 0.091 0.413 0.276 0.625 -
Mn - - - 0.017 0.044
Mg - 0.889 4.776 2.811 0.044
Total Y 3.81 4.06 6.04 5.44 6.17

Ca - 0.123 - 0.260 -
Na 0.092 0.036 0.034 0.199 0.106
K 1.952 1.190 1.746 1.236 1.674
Total X 2.04 1.35 1.78 1.70 2.02

F - - 0.278 0.519 2.936
OH 3.916 4.00 3.946 2.378 0.344
Total (OH,F) 3.92 4.00 4.22 2.90 3.28

Figure 5.44 Sketch of tetrahedral rotational angle ␣r for two limiting conditions, ␣r ⫽ 0°
and ␣r ⫽ 12°. From Hazen and Wones (1972). Reprinted with permission of The Mineral-
with T and P conditions according to the mean individual and compressibilities

of the various polyhedra (see sections 1.14.2 and 1.15.1). Because the tetrahedra
are relatively insensitive to P and T modifications, the effects of the two intensive
variables are primarily registered by the do distances.
In common trioctahedral micas with Al:Si diadochy of 1:3
[(KR2⫹3 (Si3Al)O10 (OH)2 ], the mean T-O distance is 1.649 Å (Hazen and Burn-
ham, 1973). Based on this value and on equations 5.113 and 5.114, we obtain
2.035 ≤ d o ≤ 2.110. (5.116)
Let us calculate the effects of intensive variables on the do distance, considering phlog-
opite [KMg3(Si3Al)O10 (OH)2 ] as an example. The octahedral cation is Mg2⫹; the mean
octahedral distance at P ⫽ 1 bar and T ⫽ 25 °C is about 2.06 Å (Hazen and Burnham,
1973). The mean polyhedral linear thermal expansion is ␣l ⫽ 1.4 ⫻ 10⫺5(°C⫺1 ), and the
mean polyhedral linear compressibility is ␤l ⫽ 1.7 ⫻ 10⫺4 (kbar⫺1 ). We know that the
upper stability limit is do ⫽ 2.110 Å. We can solve the limiting P and T conditions by
applying the equality
d o(T , P ) = d o(Tr , Pr ) (1 + α lT − β l P ) = 2.110, ( 5.117 )
( )
2.06 1 + 1.4 × 10 −5 T − 1.7 × 10 −4 P = 2.110 ( 5.118 )
and
Silicates / 329
Table 5.56 P and T intrinsic stability limits of mica

according to Hazen and Finger (1982).
R Cation Lower Limit Upper Limit

␣r ⫽ 0° ␣r ⫽ 12°
General Formula KR32⫹ (Si3Al)O10(OH)2
Ni T ⫽ 11P ⫹ 200 T ⫽ 11P ⫹ 2800
Mg T ⫽ 12P ⫺ 900 T ⫽ 12P ⫹ 1700
Co T ⫽ 12P ⫺ 1700 T ⫽ 12P ⫹ 900
Fe T ⫽ 13P ⫺ 2900 T ⫽ 13P ⫺ 300
Mn T ⫽ 14P ⫺ 4600 T ⫽ 14P ⫺ 2000
2⫹
General Formula K(R2.5 Al0.5)(Si2.5Al1.5)O10(OH)2
Ni T ⫽ 11P ⫹ 2100 T ⫽ 11P ⫹ 3400
Mg T ⫽ 12P ⫹ 1000 T ⫽ 12P ⫹ 3400
Co T ⫽ 12P ⫹ 300 T ⫽ 12P ⫹ 2800
Fe T ⫽ 13P ⫺ 700 T ⫽ 13P ⫹ 1700
Mn T ⫽ 14P ⫺ 2400 T ⫽ 14P
T (˚C ) ≈ 12P(kbar ) + 1700. ( 5.119 )
Analogously, for the lower limit,
( )
2.06 1 + 1.4 × 10 −5 T − 1.7 × 10 −4 P = 2.035 ( 5.120 )
and
( )
T o C ≈ 12P(kbar ) − 900. ( 5.121)
Table 5.56 lists intrinsic stability limits for common trioctahedral micas with gen-
eral formulas KR2⫹ 2⫹
3 (Si3Al)O10 (OH)2 and K(R 2.5 Al0.5 )(Si2.5Al1.5 )O10 (OH)2 , cal-
culated by Hazen and Finger (1982) with the method described above. Based on
these calculations, trioctahedral common micas with general formula
KR2⫹3 (Si3Al)O10 (OH)2 are structurally unstable for cations Fe
2⫹
and Mn2⫹,
which are too large and exceed the limiting value do /dt ⫽ 1.275. To keep the ratio
within limits, it is necessary to increase the value of dt , by means of a coupled
substitution involving Al (cf. table 5.56):
Si IV + R 2 + , VI ⇔ Al VI + Al IV. ( 5.122 )
Other elemental exchanges capable of reducing the do /dt ratio (Hazen and Finger,
1982) are
( )
R 2 + , VI + OH − ⇔ R 3 + , VI + O 2 − ( 5.123)
Al IV ⇔ Fe 3 + ,IV (5.124)
2R 2 + , VI ⇔ Ti 4 + + VY ( 5.125)
(with V Y|| ⫽ bi-ionized cationic vacancy on site Y ).

The structural model of Hazen and Wones has been confirmed experimen-
tally. For example, the pure “annite” end-member [KFe2⫹ 3 (Si3Al)O10 (OH)2 ] ap-
pears to be geometrically unstable at all P and T conditions (table 5.56). Indeed,
pure annite has never been synthesized, although slightly oxidized annites of com-
position K(Fe2⫹ 3⫹
2.7 Fe0.3 )(Si3Al)O10.3(OH)1.7 have been prepared, with a do /dt ratio
of 1.275, at the limit of intrinsic stability.
5.6.3 Extrinsic Stability

Biotite-Phlogopite
Biotite is frequently found as a primary mineral in plutonic rocks, but its occur-
rence in effusive counterparts is quite rare and almost invariantly affected by evi-
dent signs of instability. The experimental explanation of this fact may be found
in the classic paper by Yoder and Eugster (1954), which compared the P-T stabil-
ity curve KMg3(Si3Al)O10 (OH)2 (phlogopite) with the curves of incipient melting
of granite and basalt. As figure 5.45 shows, whereas the stability curve of phlogo-
pite crosses the curve of incipient melting of basalt at intermediate P and T condi-
tions, it always plots to the right of the curve of incipient melting of granite at all
P and T conditions (except at extremely low pressure). Phlogopite may thus form
as a stable primary phase in granitic magmas, but is stable in basaltic melts only
under high thermobaric conditions (i.e., above about 2.5 kbar and 1100 °C).
Further experimental studies on the stability of the magnesian phlogopite component

in silica undersaturated and oversaturated systems have been carried out by various au-
thors. Wones (1967) investigated the equilibrium
(
2KMg 3 Si 3 Al O 10 OH) ( ) 2
⇔ KAlSi 2 O 6 + KAlSiO 4 + 3Mg 2 SiO 4 + 2H 2 O.
(5.126)
phlogopite leucite kalsilite forsterite
Hewitt and Wones (1984) have shown that the equilibrium constant for the above reaction
may be expressed as a polynomial function of P (bar) and T (K):
log K 126 = 2 log fH2 O = −16633T −1 + 18.1 + 0.105(P − 1)T −1. ( 5.127 )
Wood (1976a) studied the stability of phlogopite in the presence of quartz:
( )
KMg 3 Si 3 Al O 10 OH ( ) 2
+ 3 SiO 2 ⇔ KAlSi 3 O 8 + 3 MgSiO 3 + H 2 O .
( 5.128 )
phlogopite quartz sanidine enstatite
Silicates / 331
Figure 5.45 P-T stability curve of phlogopite compared with the incipient melting curves
of granite and basalt. Reprinted from H. S. Yoder and H. P. Eugster, Geochimica et Cos-
mochimica Acta, 6, 157–185, copyright 䉷 1954, with kind permission from Elsevier Sci-
ence Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
According to Hewitt and Wones (1984), the equilibrium constant may be expressed in
the form
log K 128 = log fH2 O = −2356T −1 + 4.76 + 0.077(P − 1)T −1. ( 5.129 )
With increasing P and T, the equilibrium curve of equation 5.129 reaches an invariant
condition, determined by the appearance of silicate melt (T ⫽ 835 °C, P ⫽ 0.45 kbar;
Wones and Dodge, 1977).
Concerning the ferriferous term annite [KFe32⫹(Si3Al)O10 (OH)2 ], the most important
experimentally studied reactions are as follows:
(
2 KFe 3 Si 3 Al O 10 OH ) ( ) 2
+ H 2 O ⇔ 2 KAlSi 3 O 8 + 3 Fe 2 O 3 + 3 H 2
( 5.130 )
annite sanidine hematite
(Eugster and Wones, 1962);
KFe 3 (Si 3 Al ) O10 (OH )2 ⇔ KAlSi 3O8 + Fe 3O 4 + H2 ,

annite sanidine magnetite (5.131)
with
log K 131 = log fH2 = −8113T −1 + 9.59 + 0.0042(P − 1)T −1 ( 5.132 )

(Hewitt and Wones, 1981, 1984);
(
2 KFe3 Si 3Al O10 OH ) ( ) 2
⇔ 3 Fe2SiO4 + KAlSi2O6 + KAlSiO4 + 2 H2O,
( 5.133)
annite fayalite leucite kalsilite
which is analogous to reaction 5.126 valid for the magnesian component, with
log K133 = 2 log fH2 O = −17705T −1 + 22.44 + 0.112(P − 1)T −1 ( 5.134 )
(Eugster and Wones, 1962; Hewitt and Wones, 1984);
3 KFe3(Si 3Al) O10 OH ( ) 2

⇔ 3 Fe2SiO4 + 3 KAlSi2O6 + Fe 3O4 + 2 H2O+ 2 H2 ,
( 5.135)
annite fayalite leucite magnetite
with
log K 135 = 2 log f H 2O + log fH 2 = −26044T −1 + 32.41 + 0.075 P − 1 T −1 ( 5.136 ) ( )

(Eugster and Wones, 1962; Hewitt and Wones, 1984); and
KFe 3 (Si 3 Al ) O10 (OH )2 + H2 ⇔ Fe 2 SiO 4 + KAlSi 2 O6 + Fe + 2 H2 O

fayalite leucite metal
( 5.137 )
annite
( ) ( )
KFe3 Si 3Al O10 OH
2
+ 3 H2 ⇔ KAlSi 3O8 + 3 Fe+ 4 H2O .
(5.138)
annite sanidine metal
Figure 5.46 shows the stability field of the annite end-member at 1 kbar total
pressure, based on equilibria 5.131, 5.133, and 5.135, as a function of temperature
and hydrogen fugacity in the system. The figure also shows the positions of the
iron-wuestite (IW) and wuestite-magnetite buffer (WM).
Muscovite-Paragonite
The P-T stability curve of muscovite intersects the incipient melting curve of
granite in hydrous conditions at about 2.3 kbar total pressure and T ⫽ 650 °C.
Thus, muscovite may crystallize as a primary phase from granitic melts above
these P and T conditions (interstitial poikilitic crystals) or may form by reaction
with pristine solid phases at lower P and T (muscovite as dispersed phase within
feldspars, for instance). In this second type of occurrence, the following two equi-
libria are of particular importance:
( )
KAl 2 Si 3 Al O 10 OH ( ) 2
⇔ KAlSi 3 O 8 + Al 2 O 3 + H 2 O
(5.139)
muscovite sanidine corundum
Silicates / 333
Figure 5.46 Stability field of annite at Ptotal ⫽ 1 kbar, as a function of T and fO2. From
Hewitt and Wones (1984). Reprinted with permission of the Mineralogical Society of
America.
( ) ( )
KAl2 Si3 Al O10 OH
2
+ SiO 2 ⇔ KAlSi3O 8 + Al2 SiO 5 + H2 O .
(5.140)
muscovite quartz K- feldspar
Equilibrium 5.139 has been investigated by several authors. According to Hewitt and
Wones (1984), the equilibrium constant may be expressed as a polynomial function of P
(bar) and T (K):
log K 139 = log fH2 O = 8.5738 − 5126T −1 + 0.0331(P − 1)T −1. ( 5.141)
The univariant equilibrium 5.139 for Ptotal ⫽ PH2O, based on the experimental data of
Chatterjee and Johannes (1974), is compared in figure 5.47A with the experimental results
of Ivanov et al. (1973) for XH2O ⫽ XCO2 ⫽ 0.5. Figure 5.47A also superimposes the melting
curves of granite in the presence of a fluid phase of identical composition. Note that the
composition of the fluid dramatically affects stability relations: if the amount of H2O in
the fluid is reduced by half (XH2O ⫽ XCO2 ⫽ 0.5), the stability field of muscovite is re-
stricted to approximately P ⬎ 5.5 kbar and T ⬎ 680 °C.
According to the experiments of Chatterjee and Johannes (1974), the equilibrium
constant for equation 5.140 may be expressed within the stability field of andalusite in
the form
Figure 5.47 Extrinsic stability curves for muscovite and paragonite, based on (A) equilib-
ria 5.139 and 5.144 (quartz absent) and (B) equilibria 5.140 and 5.145 (quartz present).
Incipient melting curves of granite for XH2O ⫽ 0.5 to 1, and stability curves of Al2SiO5
polymorphs according to Richardson et al. (1969), are superimposed.
log K 140(andalusite) = log fH2 O

(5.142)
= 8.3367 − 4682T −1 + 0.0163(P − 1)T −1
and, within the stability field of sillimanite, in the form
log K 140(sillimanite) = log fH2O

(5.143)
(
= 8.9197 − 5285T −1 + 0.0248 P − 1 T −1. )
The validity limits for equations 5.142 and 5.143 (Al2SiO5 polymorphic transitions) are
those of Richardson et al. (1969) (triple point at P ⫽ 5.5 kbar, T ⫽ 620 °C). In figure 5.48
we see the univariant equilibrium 5.140 plotted by Kerrick (1972) for different com-
positions of the fluid phase (XCO2 ⫹ XH2O ⫽ 1). The effect of lowering the H2O partial
pressure in the fluid is analogous to previous observations for equilibrium 5.139. However,
Kerrick (1972) adopted the andalusite-kyanite-sillimanite stability limits of Holdaway
(1971) (P ⫽ 3.8 kbar, T ⫽ 500 °C).
The extrinsic stability relations of the sodic term paragonite [NaAl2(Si3Al)

O10 (OH)2 ] have been investigated as have those of the potassic end-member mus-
covite:
( ) ( )
NaAl2 Si3 Al O10 OH
2
⇔ NaAlSi3O 8 + Al2 O 3 + H2 O
(5.144)
paragonite albite corundum
Silicates / 335
Figure 5.48 Extrinsic stability limits of muscovite, based on equilibrium 5.140 for vari-
able amounts of H2O component in fluid. Also shown are incipient melting curves of
granite for various XH2O isopleths and stability curves of Al2SiO5 polymorphs, according
to Holdaway (1971) From Kerrick (1972), American Journal of Science, 272, 946–58. Re-
printed with permission of American Journal of Science.
(
NaAl2 Si3Al ) O (OH)
10 2
+ SiO2 ⇔ NaAlSi3O8 + Al2SiO5 + H2O .
(5.145)
paragonite quartz albite
Evaluation of equilibria 5.144 and 5.145 is complicated by the formation of poly-

type 1M, which is metastable with respect to 2M1. The 1M ⇔ 2M1 transforma-
tion is slow and often is not completed at low T. The most consistent data are
those of Chatterjee (1970), who performed reversal experiments. The position of
univariant equilibrium 5.144 in the P-T field according to this author is shown in
figure 5.47A. Part B of that figure plots the position of the univariant equilibrium
5.145 according to Chatterjee (1972), compared with the stability limits for the
potassic term. Note, by comparing equilibria 5.139 and 5.140 with equilibria
5.144 and 5.145, that the sodic terms are extrinsically stable, pressure being equal,
at lower T (about 100 °C) with respect to their potassic counterparts. When
studying the crystallization of dioctahedral micas from natural melts, it is there-
fore important to evaluate carefully the role of the Na/K ratio in the system.
Figure 5.49 Stability field of margarite plus quartz in the CaO-Al2O3-SiO2-H2O system
for various values of aH2O, fluid. From Chatterjee (1974). Reprinted with permission of The
Margarite
Chatterjee (1974) studied in detail the equilibrium
( )
CaAl2 (Si2 Al2 ) O10 OH ⇔ CaAl2Si2O8 + Al2O3 + H2O
2 (5.146)
margarite anorthite corundum
in the pressure range 1 to 7 kbar. According to Hewitt and Wones (1984), the
equilibrium constant varies with P and T as
log K 146 = log f H 2O = 8.8978 − 4758T −1 − 0.0171 P − 1 T −1 . ( ) (5.147)
In the presence of quartz, the most significant equilibria are
( ) ( )
4CaAl2 Si2 Al2 O10 OH + 3SiO2 ⇔ 2Ca2 Al3Si3O12 OH + 5Al2SiO5 + 3H2O
2
( )
margarite quartz zoisite
(5.148)
and
( ) ( )
CaAl2 Si2 Al2 O10 OH + 3SiO2 ⇔ CaAl2Si2O8 + Al2SiO5 + 3H2O.
2 (5.149)
margarite quartz anorthite
Figure 5.49 shows the P-T stability field of margarite plus quartz in the system
CaO-Al2O3-SiO2-H2O, defined by Chatterjee (1976) for various values of H2O
activity in the fluid.
All these phase stability relations may be expressed, as we have seen, as func-
Silicates / 337
Figure 5.50 Chemistry of natural biotites as a function of the redox state of the system.
Solid lines delineate the compositional trend imposed by the ruling buffer. From Speer
(1984). Reprinted with permission of The Mineralogical Society of America.
tions of the fugacity of a gaseous component in the fluid phase (i.e., fH2O, fH2;
see, however, section 9.1 for the significance of the term “fugacity”). It must be
emphasized here that the stability and chemical composition of natural micas are
also affected by the redox state of the system, which may itself be expressed as
the fugacity of molecular oxygen in fluid phases. Wones and Eugster (1965) have
shown that the oxidation state of biotites may be described as the result of mixing
of three compositional terms: KFe2⫹ 3 AlSi3O10 (OH)2 , KMg3AlSi3O10 (OH)2 , and
3⫹
KFe3 AlSi3O12H1 . In a ternary field of this type (figure 5.50), the compositional
trend of micas depends on the buffer capability of the system (for the meaning
of the term “buffer”, see section 5.9.5).
Figure 5.51A shows in greater detail how the stability field of biotite (phlogo-
pite-annite pseudobinary mixture) is affected by fO2-T conditions for a bulk pres-
Figure 5.51 fO2-T stability field of biotite. (A) Pseudobinary phlogopite-annite mixture.
Numbers at experimental points indicate observed Fe/(Fe ⫹ Mg) atom ratio. (B) Annite.
HM ⫽ hematite-magnetite buffer; NNO ⫽ Ni-NiO buffer; MW ⫽ magnetite-wuestite
buffer; QFM ⫽ quartz-fayalite-magnetite buffer. From Wones and Eugster (1965). Re-
sure of 2.07 kbar. Numbers at the various experimental points indicate the com-
position of the mixture. Tie lines connect points of identical composition at
opposite sides of the stability region. Figure 5.51B shows the stability field of the
annite end-member for the same bulk pressure. We have already noted that, for
structural reasons, pure annite is intrinsically unstable at any P-T condition, and
that the presence of Fe3⫹ is required to constrain the angular distortion of [SiO4 ]
groups within structurally acceptable limits. The lower f O 2-T stability limit of
annite must thus be ascribed to this structural constraint (the amount of Fe3⫹ in
oxidized annite obviously diminishes with decreasing f O 2 in the system; see also
Wones and Eugster, 1965).
Silicates / 339
5.6.4 Thermodynamic Properties of End-Members and

Mixing Behavior
Table 5.57 lists the thermodynamic data for mica end-members according to Hel-
geson et al. (1978), Berman (1988), Holland and Powell (1990) and Saxena et al.
(1993). The substantial agreement among the last three sources may be ascribed
to the similar optimization procedures adopted by the authors in assessing the
energy properties of end-members through multivariate analysis. However, much
experimental effort must still be made in order to gain proper knowledge of the
energetics of micaceous components.
It is of interest here to recall the method suggested by Tardy and Garrels (1974) for
the empirical estimate of the Gibbs free energy of layer silicates. This method, which is
even simpler than the combinatory procedures mentioned in section 3.6, has an accuracy
comparable to the discrepancy of existing experimental data.
According to Tardy and Garrels (1974), the Gibbs free energy of formation of layered
silicates may be expressed as the simple sum of fictive energies ascribable to the constituent
oxides, after suitable “decomposition” of the chemistry of the phase. Consider, for in-
stance muscovite and paragonite: the Gibbs free energy of formation from the elements
at T ⫽ 298.15 K and P ⫽ 1 bar may be expressed as the sum of the fictive silicate compo-
nents (Gf,sil; see table 5.58):
1 0 3
G 0f ,KAl AlSi O = G f , sil,K O + G 0f , sil, Al O + 3G 0f , sil, SiO + G 0f , sil,H O
3 10(OH) 2
2
2 2
2 2 3 2 2
=
1
2
(
−786.6 +
3
2
) ( ) (
−1600.0 + 3 −856.0 + −247.7 ) ( ) (5.150)
= −5609.0
1 0
G 0f , NaAl2 AlSi3O10(OH)2 = G +L
2 f , sil,Na2O (5.151)
= −5556.3 (kJ/mole ).
Note that the Gibbs free energy of formation from the elements of the fictive constituent
oxides does not correspond to the actual thermodynamic value. It differs from it by an
empirical term ( ⌬G 0f,silicification ) that accounts for the structural difference between the ox-
ide in its stable standard form and as a formal entity present in the layered silicate (cf.
table 5.58).
Table 5.59 lists Gibbs free energies of formation from the elements of mica end-
members obtained with the procedure of Tardy and Garrels (1974). For comparative pur-
poses, the same table lists Gibbs free energies of formation from the elements derived from
the tabulated H 0f,Tr ,Pr and S 0Tr ,Pr values (same sources as in table 5.57) by application of
S 0f , Tr , Pr = ST0r , Pr − ∑ vi Si0, Tr , Pr ( 5.152 )

i
Table 5.57 Thermodynamic properties of mica end-members in databases of Helgeson et al.

(1978) (1), Saxena et al. (1993) (2), Berman (1988) (3), and Holland Powell (1990) (4). H 0ƒ,Tr ,Pr ⫽
enthalpy of formation from the elements at standard state (Tr ⫽ 298.15 K, Pr ⫽ 1 bar; kJ/mole);
S 0Tr ,Pr ⫽ standard state entropy [J/(mole ⫻ K)]; V 0Tr ,Pr ⫽ standard state volume (cm3/mole); ␣0 ⫽
isobaric thermal expansion (K⫺1); ␤0 ⫽ isothermal compressibility (bar⫺1). Heat capacity function:
CP ⫽ K1 ⫹ K2T ⫹ K3T ⫺2 ⫹ K4T 2 ⫹ K5T ⫺3 ⫹ K6T ⫺1/2 ⫹ K7T ⫺1.
End-Member Formula Reference H ƒ0 , Tr , P r S T0 r , P r V T0 r , P r

Phlogopite KMg3AlSi3O10(OH)2 (1) ⫺6226.072 318.40 149.66
(2) ⫺6203.900 334.600 149.66
(3) ⫺6207.342 334.158 149.77
(4) ⫺6211.76 325.0 149.64
Na-phlogopite NaMg3AlSi3O10(OH)2 (4) ⫺6173.64 315.0 144.50
F-phlogopite KMg3AlSi3O10(F)2 (1) ⫺6419.474 317.57 146.37
Annite KFe3AlSi3O10(OH)2 (1) ⫺5155.504 398.317 154.32
(2) ⫺5125.500 440.910 154.32
(4) ⫺5149.32 414.0 154.32
Muscovite KAl2AlSi3O10(OH)2 (1) ⫺5972.275 287.86 140.71
(2) ⫺5976.300 306.40 140.82
(3) ⫺5976.740 293.157 140.87
(4) ⫺5981.63 289.0 140.83
Paragonite NaAl2AlSi3O10(OH)2 (1) ⫺5928.573 277.82 132.53
(2) ⫺5951.0 277.100 131.88
(3) ⫺5944.208 277.70 132.16
(4) ⫺5948.15 276.0 131.98
Celadonite KMgAlSi4O10(OH)2 (1) 313.38 157.1
(4) ⫺5834.27 297.0 139.60
Fe-celadonite KFeAlSi4O10(OH)2 (4) ⫺5484.96 328.0 140.00
Margarite CaAl2Al2Si2O10(OH)2 (3) ⫺6236.603 265.084 129.58
(4) ⫺6241.64 265.0 129.64
Eastonite KMg2AlAl2Si2O10(OH)2 (4) ⫺6336.56 315.0 147.51
Siderophyllite KFe2AlAl2Si2O10(OH)2 (4) ⫺5628.27 375.0 150.63
End-Member Reference K1 K2 ⫻ 103 K3 ⫻ 10⫺5 K4 K5 ⫻ 10⫺8

Phlogopite (1) 420.952 120.416 ⫺89.956 0 0
(2) 554.42 27.835 ⫺16.458 0 ⫺1.0629
(3) 610.38 0 ⫺215.33 0 28.4104
(4) 770.3 ⫺36.939 ⫺23.289 0 0
Na-phlogopite (4) 773.5 ⫺40.229 ⫺25.979 0 0
F-phlogopite (1) 402.21 111.88 ⫺85.395 0 0
Annite (1) 445.303 124.558 ⫺80.793 0 0
(2) 626.50 62.63 132.76 0 ⫺7.947
(4) 809.2 ⫺70.250 ⫺6.789 0 0
Muscovite (1) 408.191 110.374 ⫺106.441 0 0
(2) 462.6 65.98 ⫺333.74 0 47.388
Silicates / 341
End-Member Reference K1 K2 ⫻ 103 K3 ⫻ 10⫺5 K4 K5 ⫻ 10⫺8
(3) 651.49 0 ⫺185.232 0 27.4247

(4) 756.4 ⫺19.84 ⫺21.70 0 0
Paragonite (1) 407.647 102.508 ⫺110.625 0 0
(2) 502.209 32.215 ⫺196.23 0 26.177
(3) 577.57 0 ⫺322.144 0 50.5008
(4) 803.0 ⫺31.58 2.17 0 0
Celadonite (1) 335.77 105.855 ⫺77.571 0 0
(4) 741.2 ⫺18.748 ⫺23.688 0 0
Margarite (3) 699.80 0 ⫺68.077 0 7.3432
(4) 744.4 ⫺16.8 ⫺20.744 0 0
Eastonite (4) 785.5 ⫺38.031 ⫺21.303 0 0
Siderophyllite (4) 815.8 ⫺36.645 ⫺5.645 0 0
End-Member Reference K6 ⫻ 10⫺2 K7 ⫻ 10⫺4 T Limit (K) ␣0 ⫻ 105 ␤0 ⫻ 106

Phlosopite (1) 0 0 1000 - -
(2) 0 ⫺5.5447 - 2.0 1.5128
(3) ⫺20.838 0 1000 - -
(4) ⫺65.316 0 - 2.6063 1.6373
Na-phlogopite (4) ⫺65.126 0 - 2.9758 1.6609
F-phlogopite (1) 0 0 1000 - -
Annite (1) 0 0 1000 - -
(2) 0 ⫺11.67 - 1.7 1.3897
(4) ⫺74.03 0 - 2.7864 0.9720
Muscovite (1) 0 0 1000 - -
(2) 0 1.2027 - 2.5 1.1546
(3) ⫺185.232 0 967 - -
(4) ⫺69.792 0 - 2.7693 1.065
Paragonite (1) 0 0 1000 - -
(2) 0 ⫺2.1204 - 1.5 1.1620
(3) ⫺14.728 0 719 - -
(4) ⫺81.51 0 - 3.1823 1.2881
Celadonite (1) 0 0 1000 - -
(4) ⫺66.169 0 - 2.7937 1.1461
Margarite (3) ⫺55.871 0 996 - -
(4) ⫺67.832 0 - 2.7769 1.0028
Eastonite (4) ⫺68.937 0 - 2.5761 1.6270
Siderophyllite (4) ⫺75.171 0 - 2.7219 1.1950
Table 5.58 Gibbs free energy of formation from elements

at T ⫽ 298.15 K and P ⫽ 1 bar for fictive structural oxide
components of layered silicates (Tardy and Garrels, 1974)
compared with the actual thermodynamic values of stable
oxides (data in kJ/mole).
Component G 0ƒ,oxide G 0ƒ,sil G 0ƒ,silicification

H2O ⫺237.2 ⫺247.7 ⫺10.5
Li2O ⫺561.9 ⫺797.5 ⫺235.6
Na2O ⫺377.4 ⫺681.2 ⫺303.8
K2O ⫺322.2 ⫺786.6 ⫺464.4
CaO ⫺604.2 ⫺764.8 ⫺160.6
MgO ⫺569.4 ⫺624.3 ⫺54.9
FeO ⫺251.5 ⫺268.2 ⫺16.7
Fe2O3 ⫺743.5 ⫺743.5 0
Al2O3 ⫺1582.0 ⫺1600.0 ⫺18.0
SiO2 ⫺856.0 ⫺856.0 0
Table 5.59 Gibbs free energies of formation from the elements of micaceous components
according to method of Tardy and Garrels (1974), compared with tabulated values. References as
in table 5.57. Data in kJ/mole.
G 0ƒ,obs
Component G 0ƒ,calc ∑ v i S 0i (1) (2) (3) (4)

i
KMg3AlSi3O10(OH)2 ⫺5881.9 1571.46 ⫺5852.5 ⫺5835.1 ⫺5838.4 ⫺5840.1

NaMg3AlSi3O10(OH)2 ⫺5818.7 1558.08 - - - ⫺5803.0
KFe3AlSi3O10(OH)2 ⫺4813.6 1555.26 ⫺4810.6 ⫺4793.3 - ⫺4809.1
KAl2AlSi3O10(OH)2 ⫺5609.0 1567.74 ⫺5590.7 ⫺5600.2 ⫺5596.7 ⫺5600.4
NaAl2AlSi3O10(OH)2 ⫺5556.3 1554.36 ⫺5547.9 ⫺5570.2 ⫺5563.6 ⫺5567.0
KMgAlSi4O10(OH)2 ⫺5489.3 1562.53 - - - ⫺5456.9
KFeAlSi4O10(OH)2 ⫺5133.2 1557.13 - - - ⫺5118.5
CaAl2Al2Si2O10(OH)2 ⫺5924.5 1554.23 - - ⫺5852.2 ⫺5857.3
KMg2AlAl2Si2O10(OH)2 ⫺6001.6 1553.26 - - - ⫺5967.4
KFe2AlAl2Si2O10(OH)2 ⫺5928.4 1603.49 - - - ⫺5262.0
KFe32⫹Fe3⫹Si3O10(OH)2 ⫺4385.4 1591.81 - - - -
KLi2AlSi4O10(OH)2 ⫺5662.5 1588.09 - - - -
G 0f , Tr , Pr = H 0f , Tr , Pr − Tr S 0f , Tr , Pr , ( 5.153 )
where ∑iviS 0i,Tr ,Pr is the sum of standard state entropies of the elements at stable state
(Robie et al., 1978), multiplied by their stoichiometric factors.
The method of Tardy and Garrels (1974) gives approximate estimates

of Gibbs free energy values for mica compounds regardless of their chemical
complexity. Nevertheless, the linearity of the method cannot account for the
energy of mixing contributions. Experimental data on mica mixtures are also
Silicates / 343
Figure 5.52 (A) Molar volumes of the (Na,K)Al2Si3AlO10(OH)2 mixture at various P

conditions. (B) Corresponding excess volumes of mixing. From Blencoe (1977). Reprinted
with permission of The Mineralogical Society of America.
limited to a few binary joins, and experimental observations are complicated

by polymorphism (polytypism) and by metastable persistence at low T of
polytypes that do not correspond to the minimum energy structure at the P,
T, and X of interest. For instance, figure 5.52A shows the molar volumes of
NaAl2Si3AlO10 (OH)2-KAl2Si3AlO10 (OH)2 (paragonite-muscovite) mixture plot-
ted by Blencoe (1977) according to existing experimental data. Note the clear
discrepancies in the molar volume of the same composition and conspicuous devi-
ations from ideal mixing behavior. Figure 5.52B shows the corresponding excess
volumes of the mixture. According to Blencoe (1977), volumetric excess terms
may be represented by a subregular Margules model, with the parameters listed in
Table 5.60 Volumetric [WV ; J/(bar ⫻ mole)] enthalpic (WH ; kJ/mole), and entropic [WS ; J/(mole ⫻
K] terms of subregular Margules model for (Na,K)Al2Si3AlO10(OH)2 binary mixture, according to
various authors.
References WH,12 WH,21 WS,12 WS,21 WV,12 WV,21

Blencoe (1977) 1M, 2 kbar - - - - 0.3628 0.3151
Blencoe (1977) 1M, 4 kbar - - - - 0.4502 0.0594
Blencoe (1977) 1M, 8 kbar - - - - 0.5192 ⫺0.1519
Blencoe (1977) 2M, 8 kbar - - - - 0.3138 0.2389
Blencoe and Luth (1973) - - - - ⫺0.1423 ⫺0.3222
Eugster et al. (1972) 12.895 17.422 0.711 1.653 0.344 0.527
Chatterjee and Froese (1975) 12.230 19.456 0.710 1.654 0.665 ⫺0.456
(1) ⫽ NaAl2Si3AlO10(OH)2 and (2) ⫽ KAl2Si3AlO10(OH)2
W12 ⫽ WH,12 ⫺ TWS,12 ⫹ PWV,12
W21 ⫽ WH,21 ⫺ TWS,21 ⫹ PWV,21
Gexcess mixing ⫽ X1X2(W12X2⫹W21X1)
table 5.60. The same table also lists enthalpic and entropic subregular interaction
parameters for the same mixture according to Eugster et al. (1972) and Chatterjee
and Froese (1975).
Because excess volume terms are generally positive (with the exception of the
values indicated by Chatterjee and Froese, 1975 and Blencoe and Luth, 1973) and
because Margules parameters are also generally positive, the mixing model of
table 5.60 implies the existence of a solvus field in the binary range, whose exten-
sion should widen with increasing pressure [at T ⫽ 500 °C and Ptotal ⫽ P H2O ⫽
4 kbar, the solvus field should cover the range 0.1 ⬍ KAl2AlSi3O10 (OH)2 ⬍ 0.85;
Chatterjee and Froese, 1975].
According to Wones (1972) and Mueller (1972), the mixture KMg3AlSi3O10
(OH)2-KFe3AlSi3O10 (OH)2 (phlogopite-annite) is virtually ideal.
Munoz and Ludington (1974, 1977) experimentally measured the ionic ex-
change reaction
OH −mica + HF ⇔ Fmica
−
+ H2 O ( 5.154 )
in muscovite, siderophyllite, annite, and phlogopite in systems F-buffered by the

solid assemblage anorthite-fluorite-sillimanite-quartz at 673 ⬍ T (K) ⬍ 1000 and
Pfluid ⫽ 1 to 2 kbar. For all four phases, the T-dependency of the equilibrium
constant was identical—i.e.:
( )
KMg 3 AlSi 3 O 10 OH, F( ) 2
→ log K 154 = 2100T −1 + 1.52 (5.155)
( ) (
KFe 3 AlSi 3 O 10 OH, F ) 2
→ log K 154 = 2100T −1 + 0.41 (5.156)
Silicates / 345
( )( )
K Fe 22 + Al Al 2 Si 2 O 10 OH, F ( ) 2
→ log K 154 = 2100T −1 + 0.20 (5.157)
( )
KAl 2 AlSi 3 O 10 OH, F ( ) 2
→ log K 154 = 2100T −1 − 0.11. (5.158)
Let us now consider the exchange reaction
(
KMg3 AlSi3 O10 OH) ( ) 2
+ ( ) ()
KFe 3 AlSi3 O10 F ⇔
2
(
KMg3 AlSi3 O10 F) ()
2
+ ( ) ( )
KFe3 AlSi3 O10 OH .
2
OH- phlogopite F-annite F- phlogopite OH-annite
(5.159)
The constant of equilibrium 5.159 is given by the ratio of the constant relative to
exchanges 5.155 and 5.156—i.e. in logarithmic notation,
log K 159 = log K 155 − log K 156 = 1.11. (5.160)
Independently of T, reaction 5.159 thus appears to proceed spontaneously toward

the right, in agreement with the “principle of Fe-F incompatibility” in silicates,
postulated long ago by Ramberg (1952). This fact also implies the nonideality of
the system. According to Munoz (1984), the activity coefficients of OH-
phlogopite and F-phlogopite in mixture are defined by
log γ OH − phlogopite = −X Fe X F log K 159 (5.161)
log γ F − phlogopite = X Fe X OH log K 159 . ( 5.162 )
Hinrichsen and Schurmann (1971) and Franz et al. (1977) detected the existence
of a solvus field between paragonite [NaAl2(Si3Al)O10 (OH)2 ] and margarite
[CaAl2(Si2Al2 )O10 (OH)2 ] by X-ray diffractometry and IR spectroscopy. Mea-
surements in the pressure range 1 ⱕ Ptotal (kbar) ⱕ 6 show a two-phase region
that covers the compositional field 0.20 ⬍ XCaAl2( Si2Al2 )O10( OH )2 ⬍ 0.55 at T ⫽
400 °C. The mixture appears to be homogeneous at T ⬎ 600 °C.
5.7 Feldspars
Feldspars are tectosilicates (multiplicity ⫽ 4, periodicity ⫽ 4, dimensionality ⫽

3): the tetrahedral groups [TO4 ] share all their corners with neighboring [TO4 ]
units, thus forming a rigid tridimensional network. In feldspars, 25 to 50% of the
silicon atoms are replaced by Al 3⫹. The basic structure of the network is made up
of four-member rings of [TO4 ] groups, two-by-two in an upward- and downward-
Figure 5.53 Idealized feldspar structure (topologic symmetry). (A) Projection from axis
a, showing conformation of four-member rings composed of two nonequivalent upward-
directed T1-T2 tetrahedra and two nonequivalent T1-T2 downward-directed tetrahedra.
(B) Double “gooseneck” chain seen along axis a. Periodicity along axis a is 4 (about 8.4
Å).
directed (UUDD; Up-Up-Down-Down) configuration and linked to form a

chain along axis c (figure 5.53B). The chains are interconnected, as shown in
figure 5.53A. Inside the ring there is an interstitial site M, with VII-fold to IX-
fold coordination. The structure as a whole is relatively rigid (or “inflexible” in
the sense of J. V. Smith; cf. Merlino, 1976), because the [TO4 ] groups of the same
chain are alternately linked to two neighboring chains. This sort of connection
prevents any cooperative rotational movement of the four-member rings.

Feldspars have the general formula XZ4O8 . The X cation is constituted essentially
of alkalis (Na⫹, K⫹, Rb⫹ ) or alkaline earths (Ca2⫹, Ba2⫹, Sr2⫹ ). The net charge
of the [Z4O8 ] radical depends on the Al:Si diadochy level, and varies between 1⫺
and 2⫺. The occupancy of the M site (univalent and/or divalent X cations) is thus
determined by the net charge of the four-member rings. The relative amplitude of
the M site and the need to limit the charge to 1⫹ or 2⫹ results in relatively simple
chemistry because high-field-strength ions, transition elements, and Mg2⫹ are a
priori excluded.
There are three main components of feldspar: NaAlSi3O8 (albite), KAlSi3O8
(K-feldspar) and CaAl2Si2O8 (anorthite). These main components may also, for
completeness, be associated with BaAl2Si2O8 (celsian) and RbAlSi3O8 , although
these components are subordinate in natural mixtures (see table 5.61). The capa-
bility of feldspars to bear vacant sites in M positions is clearly demonstrated by
experimental evidence. Grundy and Ito (1974), for instance, observed the exis-
Silicates / 347
tence of 13% vacant sites in M positions in an Sr-feldspar studied by single crystal

X-ray diffraction. Extrinsic disorder (deviation from ideal XZ4O8 stoichiometry)
also occurs as silica and alumina excesses: Kim and Burley (1971) observed a 5%
SiO2 excess in albite synthesized at P ⫽ 5 kbar and T ⫽ 670 °C; Goldsmith (1980)
observed an Al deficiency with respect to the canonical formula in anorthite coex-
isting with corundum and synthesized at P ⬎ 10 kbar. As regards minor element
and trace element levels, natural feldspars have Fe3⫹ and Ti4⫹ substituting for
Si4⫹ in tetrahedral positions and Li⫹, Rb⫹, Cs⫹, Sr2⫹, Ba2⫹, Fe2⫹, and Mg2⫹ in
the M sites.
Table 5.61 summarizes synthetically the observed occurrences of various
trace elements in feldspars, based mainly on the indications of Smith (1983).
Feldspars are the most abundant minerals of igneous rocks, where their ubiq-
uity and abundance of their components influence normative classifications. They
are also abundant in gneisses, and may be observed in several facies of thermal
and regional metamorphic regimes. Notwithstanding their alterability, they are
ubiquitously present in sedimentary rocks, as authigenic and/or detritic phases.
Only in carbonaceous sediments is their presence subordinate.
Table 5.62 lists the chemical compositions of natural feldspars, taken from
Deer et al. (1983). Feldspar chemistry is solved on a 32-oxygen anhydrous basis,
corresponding to four formula units. The X cation summation is quite close to the
stoichiometric value (4); the Z cations are also close to the theoretical value (16).
5.7.2 Structural Properties and Intracrystalline Disorder

Feldspars crystallize in the monoclinic system (spatial groups C2/m and I2/m)
and the triclinic system (spatial groups C1 and P1 . We have already noted that
the topologic symmetry (i.e., the highest symmetry reached by the structure in
the absence of distortion) identifies two sets of nonequivalent tetrahedra T1 and
T2. In this unperturbed condition, the structure is monoclinic (e.g., high-T sani-
dine, KAlSi3O8 ). In the absence of distortion, because the tetrahedra are occupied
by either Si4⫹ or Al 3⫹ (in sanidine, the Al:Si proportion is 1:3), the T1 and T2
site populations are statistically identical and correspond to a stoichiometric ratio
of 1:3. Whenever the Al 3⫹ and Si4⫹ ions tend to order (as happens, for instance,
at parity of chemical composition, with decreasing temperature), a new structural
configuration is reached, with duplication of nonequivalent tetrahedral positions
(i.e., from T1 and T2 to T1o , T1m , T2o , and T2m ). In the condition of maximum
ordering, Al 3⫹ occupies only one of these sites (T1o ), and silicon occupies the
remaining three nonequivalent positions (microline, KAlSi3O8 ; figure 5.54A).
The site preference of aluminum for T1o may be ascribed to simple electrostatic
reasons (oxygens surrounding T1 are more tightly linked to X cations than are
oxygens surrounding T2, and the occupancy of T1 by Al 3⫹ is thus energetically
favored). In KAlSi3O8 , the intracrystalline diffusion of Al 3⫹ and Si4⫹ necessary
to attain ordered distribution is difficult to achieve at a reasonable rate, because
Table 5.61 Trace elements in feldspars, according to Smith (1983).
Mean
Element Concentration Notes
B 100 ppm Forms pure component NaBSiO3 (reedmergnerite); maximum observed

amount in mixture corresponds roughly to 1 weight % B2O5
(Desborough, 1975).
Ga 10–100 ppm Tendencially correlated to Na; highest concentration observed in
pegmatites.
Ge 1–10 ppm -
Ti 30–60 ppm Occasionally high concentrations (up to 600–700 ppm) in intermediate
members of plagioclase series (labradorite; bytownite).
P 20–200 ppm Tendencially increases with increasing albite component. High
concentrations observed by ion-probe analyses in perthites (Mason,
1982).
Be 1–10 ppm Marked correlation with Na (Steele et al., 1980a). Locally high
concentrations in Be-rich pegmatites.
Sn 1–10 ppm Much higher values observed in greisens.
Fe 100–1000 ppm Forms pure component KFeSi3O8 and is mainly present in valence
state 3⫹.
Mg 100–3000 ppm Positively correlated with amount of Ca2⫹.
Mn 10–100 ppm -
Li 1–50 ppm Marked correlation with Na (Steele et al., 1980b, 1981).
Rb 1 ppm to % Forms pure component RbAlSi3O8 (cf. table 5.63); marked correlation
with K.
Cs (?) Majority of data is dubious because of presence of mica impurities
(Smith, 1983).
Ta 0.01–30 ppm Highest concentrations in alkaline feldspars.
Sr 10–100 ppm In alkaline feldspars of pegmatites.
100–5000 ppm In other occurrences. Marked correlation with Na in comagmatic series
(Steele and Smith, 1982).
Ba 100–1000 ppm Forms pure component BaAl2Si2O8 (celsian; cf. table 5.63); positively
correlated with K. Exceptionally high values observed in Mn deposit,
where celsian occurs as main component.
Pb 1–100 ppm Exceptionally high concentrations (up to 10,000 ppm) in K-feldspars
from pegmatites.
Cu 0–30 ppm Marked correlation with amount of anorthite in mixture (Ewart et al.,
1973).
REE - REE amounts in feldspars have been carefully studied, particularly for
presence of a positive anomaly in Eu concentration, associated with
presence of Eu2⫹. This anomaly allows estimates of the redox state of
the system and is discussed in chapter 10.
NH4 1–200 ppm Forms pure component NH4Al3Si3O8 (buddingtonite), sporadically
found in hydrothermally altered layers.
Silicates / 349
Table 5.62 Chemical analyses of some natural feldspars (Deer et al., 1983). (1) orthoclase
(Mogok, Burma); (2) Adularia (St. Gottard, Switzerland); (3) anorthoclase: inclusion in
augite (Euganean Hills NE Italy); (4) sanidine from a nepheline-leucitite; (5) albite from a
pegmatite; (6) anorthite from a calc-silicatic rock.
Sample
Oxide (1) (2) (3) (4) (5) (6)

SiO2 63.66 64.28 63.70 63.62 67.84 43.88
TiO2 - - - 0.08 0.00 -
Al2O3 19.54 19.19 21.83 19.12 19.65 36.18
Fe2O3 0.10 0.09 0.18 0.47 0.03 0.08
FeO - - - - 0.02 0.00
MgO - 0.10 0.14 0.05 0.04 -
BaO - 0.11 - 1.56 - -
CaO 0.50 0.11 2.75 0.05 0.00 19.37
Na2O 0.80 0.92 7.55 2.66 11.07 0.22
K2O 15.60 15.30 3.75 12.09 0.29 0.00
H2O(⫹) - 0.36 0.19 0.11 0.56 0.28
H2O(⫺) - - - 0.00 0.30 0.08
Total 100.20 100.46 100.09 99.81 99.80 100.10
Ionic fractions on a 32 oxygens basis
Si 11.759 11.852 11.383 11.770 11.964 8.126

Al 4.254 4.170 4.598 4.169 4.085 7.898
Fe3⫹ 0.014 0.012 0.024 0.096 0.004 0.011
Ti - - - 0.011 - -
Total Z 16.03 16.03 16.00 16.05 16.05 16.03

Mg - 0.028 0.038 0.013 0.011 -
Na 0.286 0.329 2.616 0.954 3.785 0.079
Ca 0.099 0.022 0.526 0.010 - 3.884
K 3.676 3.599 0.855 2.855 0.066 -
Ba - 0.008 - 0.113 - -
Total X 4.06 3.99 4.03 3.95 3.87 3.92

KAlSi3O8 90.5 90.5 21.2 75.5 1.7 -
NaAlSi3O8 7.1 8.3 64.8 23.9 98.0 2.0
CaAlSi2O2 2.4 1.2 14.0 0.6 0.3 98.0
the energy of activation of the diffusive process (which embodies reticular distor-
tion effects) is quite high as a result of the rigidity of the structure. Low-T persis-
tency of metastable forms with intermediate intracrystalline disorder (adularia
and orthose) is thus often observed.
In NaAlSi3O8 , because the ionic size of Na⫹ is much smaller than that of
⫹
K , reticular distortion is implicit in the structure for simple topologic reasons,
and migration of tetrahedral ions is energetically easier, not implying symmetry
Figure 5.54 (A) Cationic occupancies in tetrahedral positions in case of complete disor-
der (monoclinic structure; upper drawing) and complete order (triclinic structure; lower
drawing). (B) Condition of complete order in microcline and low albite with Al:Si ⫽ 1:3,
compared with cationic ordering in anorthite (Al:Si ⫽ 2:2). Note doubling of edge c in an-
orthite.
modifications (in other words, the energy of activation of the migration process
is lower). “High albite” and “low albite,” both crystallizing in the triclinic system,
represent the two opposite arrangements of maximum disorder and maximum
order, respectively. The high-T monoclinic form “monalbite” (isostructural with
sanidine) is stable only in the vicinity of the melting point (i.e., 980 to 1118 °C at
P ⫽ 1 bar).
In tectosilicates, the Al-O-Al bonds are energetically unstable with respect to
the Al-O-Si arrangement, which better conforms to Pauling’s electrostatic valence
principle. It is essentially this fact that leads to the alternate ordering of silicon
and aluminum in Al:Si⫽1:1 feldspars (“aluminum avoidance principle”; Loew-
enstein, 1954). Thus, in anorthite (CaAl2Si2O8 ), because Al:Si⫽1:1, at all T con-
ditions we observe a natural invariant ordering of [AlO4 ] tetrahedral groups
alternating with [SiO4 ] groups. As shown in figure 5.54, this alternation implies
doubling of cell edge c.
Calorimetric data on sodium and calcium aluminosilicate crystals and glasses

(Navrotsky et al., 1982, 1985) indicate that, in an exchange reaction of type
2−
[L Si − O − Si] + [L Al − O − Al] [ ]
1−
⇔ 2 L Si − O − Al , ( 5.163 )
Silicates / 351
the gain in energy falls between ⫺36 and ⫺100 kJ per mole of substance, depending on
the type of nontetrahedral coordinating cation. This value agrees quite well with two-body
and three-body interionic potential lattice energy calculations for zeolites (⫺40 kJ/mole;
Bell et al., 1992) and with more recent ab initio SCF estimates (Tossell, 1993; ⫺40
kJ/mole)
Intracrystalline ordering in alkali feldspars (Al 3⫹ occupancy on T sites) pro-

ceeds according to different kinetics:
1. One-step ordering path. Diffusion of Al 3⫹ on T1o takes place at the same

rate, independent of starting position (T1m , T2o , or T2m ).
2. Ideal two-step ordering path. Starting from maximum disorder at high
temperature (high sanidine), all Al 3⫹ initially moves from T2 to T1, pro-
ducing a phase that may virtually be defined as “low sanidine” or “or-
dered orthoclase” (cf. Martin, 1974). From this initial condition, com-
plete ordering of Al 3⫹ on T1o is reached by T1m → T1o migration.
3. Intermediate two-step ordering path. The most common process in K-rich
and Na-rich feldspars. It initially proceeds as in the case above, up to
Al 3⫹ occupancy on T1 sites of around 0.75; once this concentration is
reached, T1m → T1o migration begins.
It must be emphasized that the inverse path (induced by heating of natural

samples) is more complex, because of kinetic reasons related to the length of
heating (see, for instance, Blasi et al., 1983).
In plagioclases (NaAlSi3O8-CaAl2Si2O8 mixtures), ordering with decreasing
T is further complicated by the Al avoidance principle: chemically homogeneous
feldspars with anorthite molar amounts exceeding 2% cannot achieve perfectly
ordered distribution because of excess Al 3⫹. This fact leads to exsolution phe-
nomena, whose nature varies with the amount of anorthite (An) in the mixture.
For instance, “peristerites” are unmixing products of initially homogeneous pla-
gioclase mixtures of composition An2-An16 , constituted at low T by a fully or-
dered albite-rich phase (An0–2 ) and a structurally more complex phase with mean
composition An25 .
Table 5.63 lists the structural characters of the various feldspar polymorphs,
according to Smyth and Bish (1988). Note duplication of cell edge c in anorthite
and celsian and the progressive increase in volume that accompanies the lowering
of symmetry in KAlSi3O8 .
In the absence of distortion, feldspar cell edges may be obtained directly from
tetrahedral edges, as suggested by Megaw (1974). Denoting l the length of the
tetrahedral edge, we have
 2 
a≈ + 3 l (5.164)
 3 
Table 5.63 Structural data of main feldspar polymorphs (from Smyth and Bish, 1988).
Name Sanidine Orthoclase Microcline Rb-Feldspar

Formula unit KAlSi3O8 KAlSi3O8 KAlSi3O8 RbAlSi3O8
Molar weight (g/mole) 278.337 278.337 278.337 324.705
Density (g/cm3) 2.578 2.571 2.567 2.930
Molar volume (cm3/mole) 107.958 108.283 108.425 110.825
Z 4 4 4 4
System Monoclinic Monoclinic Triclinic Monoclinic
Class 2/m 2/m 1 2/m
Group C2/m C2/m C1 C2/m
Cell parameters:
a 8.539 8.561 8.560 8.820
b 13.015 12.996 12.964 12.992
c 7.179 7.192 7.215 7.161
␣ 90.65
␤ 115.99 116.01 115.83 116.24
␥ 87.70
Volume (Å3) 717.15 719.13 720.07 736.01
Reference Phillips and Colville Brown and Gasperin (1971)
Ribbe and Ribbe Bailey
(1973) (1968) (1964)
Name High Albite Low Albite Anorthite Celsian

Formula unit NaAlSi3O8 NaAlSi3O8 CaAl2Si2O8 BaAl2Si2O8
Molar weight (g/mole) 262.225 262.225 278.210 375.470
Density (g/cm3) 2.610 2.621 2.765 3.400
Molar volume (cm3/mole) 100.452 100.054 100.610 110.440
Z 4 4 8 8
System Triclinic Triclinic Triclinic Monoclinic
Class 1 1 1 2/m
Group C1 C1 P1 I2/m
Cell parameters:
a 8.161 8.142 8.173 8.627
b 12.875 12.785 12.869 13.045
c 7.110 7.159 14.165 14.408
␣ 93.53 94.19 93.113 90
␤ 116.46 116.61 115.913 115.22
␥ 90.24 87.68 91.261 90
Volume (Å3) 667.12 664.48 1336.35 1166.90
Reference Winter et Harlow Wainwright and Newnham
al. (1979) and Brown Starkey (1971) and
(1980) Megaw (1960)
Silicates / 353
 2
b = 3 + 2 l
3 
(5.165)

(
c = 1+ 3 l ) (5.166)
(
β = arcsin 1 + 2 ) al (5.167)
(Ribbe, 1983a), where l varies between 2.62 Å ( [SiO4 ]4⫺ tetrahedra) and 2.88 Å
( [AlO4 ]5⫺ tetrahedra).
In feldspars with Al:Si ⫽ 1:3 stoichiometry, the length of the tetrahedral edge
is 2.8 Å. Introducing this value into the equations above gives a ⫽ 8.1, b ⫽ 13.0,
c ⫽ 7.62 Å, and ␤ ⫽ 123°. Comparison with experimental values shows good
agreement for edge b, but the approximation is too rough for edges a and b and
the angle ␤ to propose equations 5.164 to 5.167 as a determinative method.
Concerning the effect of P and T intensive variables on feldspar structure, we
must distinguish structural modifications associated with intracrystalline disorder
from pure displacive effects associated with compression and thermal expansion.
It may generally be stated that the triclinic structure of low-T feldspars has a
sufficient degree of freedom to modify interionic distances without incurring
polymorphic transitions—i.e., the (relatively inert) [TO4 ] groups are free to sat-
isfy the polyhedral thermal expansion and/or compressibility of M cations with-
out structural constraints. This fact results in the so-called “inverse relation” (i.e.,
structural modifications induced by cooling of the substance are similar to those
induced by compression; cf. Hazen and Finger, 1982).
Table 5.64 lists isobaric thermal expansion and isothermal compressibility
coefficients for feldspars. Due to the clear discrepancies existing among the vari-
ous sources, values have been arbitrarily rounded off to the first decimal place.
We may envisage the almost linear variation of thermal expansion and com-
pressibility with the amount of anorthite in the plagioclase mixture. We also see
that the three polymorphs of the KAlSi3O8 component have substantially similar
compressibilities, within uncertainties.

Polymorphic transitions render the thermodynamic description of feldspar end-
members rather complex. According to Helgeson et al. (1978), the transition be-
tween the monoclinic and triclinic forms of NaAlSi3O8 may be regarded as the
overlap of two high-order transitions (cf. section 2.8), one associated with displac-
ive structural modifications and the other with ordering on tetrahedral sites. The
analysis of calorimetric data by Holm and Kleppa (1968) concerning the enthalpy
Table 5.64 Isobaric thermal expansion (K⫺1) and isothermal compressibility (bar⫺1)
of feldspars
Compound ␣0 ⫻ 105 ␤0 ⫻ 106 Reference

High albite 2.7 1.6 Holland and Powell (1990)
Albite - 2.0 Birch (1966)
2.7 1.6 Holland and Powell (1990)
2.5 1.4 Saxena et al. (1993)
Ab99An1 1.9 - Skinner (1966)
Ab78An22 - 1.7–1.8 Birch (1966)
Ab77An23 1.2 - Skinner (1966)
Ab56An44 1.3 - Skinner (1966)
Ab52An48 - 1.5 Birch (1966)
Ab48An52 - 1.4 Birch (1966)
Ab5An95 1.2 - Skinner (1966)
Anorthite - 1.1 Liebermann and Ringwood (1976)
2.3 0.9 Saxena et al. (1993)
Orthoclase - 2.1 Birch (1966)
K-feldspar 1.9 1.8 Holland and Powell (1990)
Sanidine - 1.8 Birch (1966)
1.4 1.4 Saxena et al. (1993)
Microcline (Or91Ab9) - 1.9 Birch (1966)
Microcline (Or83.5Ab16.5) (0.7) - Skinner (1966)
Adularia (Or88.3Ab9.3An2.4) 1.4 - Skinner (1966)
variation between monalbite and low albite shows that the enthalpy change
associated with displacive phenomena is around 0.9 kcal/mole at T of transition
(Ttransition ⫽ 965 °C ⫽ 1338 K; Thompson et al., 1974). According to Thompson
et al. (1974), the intracrystalline disorder of the NaAlSi3O8 end-member in tri-
clinic form may be expressed by an ordering parameter Qod , directly obtainable
through
( ) (
Q od = X Al,T1o + X Al,T1m − X Al,T 2 o + X Al,T 2m ) (5.168)
= −12.523 − 3.4065 b + 7.9454c ,
where X Al,T1o is the atomic fraction of Al 3⫹ in T1o site and b and c are cell edges.
Equation 5.168 may be reconducted to equations 5.165 and 5.166. According to
Helgeson et al. (1978), displacive disorder is virtually absent at T ⬍ 350 °C; more-
over, the enthalpy variation connected with the displacive process ( ⌬ H 0di ) may be
described as a function of ordering parameter Qdi (it is assumed that Qdi ⫽ 0.06
at T ⫽ 623 K), according to
[ (
∆H di0 = 2.47 − 2.63 Q di − 0.06 )] (T − 623) (cal mole ) . (5.169)
Silicates / 355
The standard molar enthalpy change connected with intracrystalline disorder

( ⌬ H 0od ) is also a function of ordering parameter Qod :
∆H od
0
= H Q0
od
− H Q0
od =1
(
= 2630 1 − Q od ) ( cal mole) . (5.170)
According to equation 5.170, the completely disordered form of the NaAlSi3O8

end-member (i.e., monalbite, Qod ⫽ 0) has an enthalpy about ⫹2.6 kcal/mole
higher than that of the fully ordered form (low albite, Qod ⫽ 1) as a result of
simple substitutional effects.
The phase transition is energetically constituted of displacive and substitutional

terms, so that we obtain
∆H trans
0
= ∆H di0 + ∆H od
0
≈ 2.6 + 0.9 (kcal mole) . ( 5.171)
Moreover, because
 ∂ ∆H od
0 
∆C P0r , od =   ( 5.172 )
 ∂T  P
r
and
 ∂ ∆H di0 
∆C P0r , di =   , ( 5.173 )
 ∂T  P
r
equation 5.171 must be consistent with the similar equation for the heat capacity function:
∆C P0r , trans = ∆C P0r , di + ∆C P0r , od . ( 5.174 )
Nevertheless, we have already noted that the transition of NaAlSi3O8 from monoclinic to
triclinic form may be regarded as the overlap of two high-order transitions, and that high-
order transition implies continuity in S, H, and G at the transition point and discontinuity
in the CP function (see section 2.8). The thermodynamic parameters for the two poly-
morphs on the opposite sides of the transition zone must thus be constrained to fulfill the
above requirements. The method followed by Helgeson et al. (1978) is as follows.
1. Extrapolation at the Tr and Pr of reference of the change in enthalpy between low

albite and high albite, determined as a continuous function of ordering parame-
ter Q.
2. Adoption of the ordered form as a stable polymorph at the Tr and Pr of reference.
3. Summation of ⌬C 0Pr ,trans, determined as a continuous function of ordering param-
eter Q, to the CP function of the fictive “stable” polymorph.
We thus obtain thermodynamic parameters of a generic albite phase that, with T, progres-
sively modifies its substitutional and displacive states from low-T to high-T forms.
Because for T ⬍ 473 K, ⌬C 0Pr ,trans ⫽ 0, Helgeson et al. (1978) adopt in the T range
298.15 to 473 K the original CP function tabulated by Kelley (1960) valid for low albite.
The albite CP function is described, starting from 473 K, by a second set of coefficients
(cf. table 5.65).
Salje (1985) interpreted overlapping (displacive plus Al-Si substitutional)

phase transitions in albite in the light of Landau theory (see section 2.8.1), as-
signing two distinct order parameters Qdi and Qod to displacive and substitutional
disorder and expanding the excess Gibbs free energy of transition in the appro-
priate Landau form:
∆G trans =
1
2
( ) 1
4
1
( )
a di T − Tc , di Q di2 + B di Q di4 + aod T − Tc ,od Qod
2
2
(5.175)
1 1
+ Bod Qod4 + C od Qod
6
+ λQ di Qod .
4 6
Coefficients a, B, and C in equation 5.175 have the usual meanings in the Landau
expansion (see section 2.8.1) and for the (second-order) displacive transition of
albite assume the values adi ⫽ 1.309 cal/(mole ⫻ K) and Bdi ⫽ 1.638 kcal/mole
(Salje et al., 1985). Tc,di is the critical temperature of transition (Tc,di ⫽ B/a ⫽
1251 K). The corresponding coefficients of the ordering process are aod ⫽ 9.947
cal/(mole ⫻ K), Bod ⫽ ⫺2.233 kcal/mole, Cod ⫽ 10.42 kcal/(mole ⫻ K), and
Tc,od ⫽ 824.1 K. With all three coefficients being present in the Landau expansion
relative to substitutional disorder it is obvious that Salje et al. (1985) consider
this transition first-order. ␭ is a T-dependent coupling coefficient between displac-
ive and substitutional energy terms (Salje et al., 1985):
λ = −0.519 − 0.727T − 3.75 × 10 −4 T 2

(5.176)
+ 5.04 × 10 −7 T 3 (cal mole ) .
For a C2/m ⇔ C1 transition, the relevant components of the spontaneous strain
tensor are x4 and x6 . The Gibbs free energy change associated with elastic strain is
∆Gtrans,ε = DdiQdiε + DodQod ε + Eε 2 . ( 5.177 )
If ⭸ ⌬G/⭸x4 ⫽ 0 and ⭸ ⌬G/⭸x6 ⫽ 0, the strain components are linearly dependent on Qdi
and Qod (Salje, 1985):
x4 = A′ Qdi + B ′Qod ( 5.178.1)
x6 = C ′Qdi + D ′Qod . ( 5.178.2 )

Silicates / 357
Replacing the generalized strain ␧ with strain components x4 and x6 and adding the elastic
energy term in the Landau expansion results in equation 5.175.
As discussed by Salje et al. (1985) and also evident from equations 5.178.1 and
5.178.2, the two order parameters Qdi and Qod are not independent of each other (although
the displacive order parameter Qdi is largely determined by x4 whereas the substitutional
disorder Qod depends more markedly on x6 ). Their mutual dependence is outlined in fig-
ure 5.55.
Note that, displacive disorder being associated with a first-order transition, Qdi exhib-
its a jump at Ttrans whose magnitude decreases from fully ordered albite (i.e., Qod ⫽ 1) to
analbite (i.e., Qod ⫽ 0) (lower part of figure 5.5; see also section 2.8.1). Moreover, the
order-disorder phase transition is stepwise for Qdi ⫽ 0 (front curve in upper part of figure
5.5) and becomes smooth for higher Qdi values.
The Cp change at the transition temperature resulting from displacive disor-

der is, according to the Landau expansion above, ⌬ C 0Pr ,di ⫽ 0.66 cal/(mole ⫻ K).
The resulting entropy change associated with Al-Si substitutional disorder is
⫺4.97 cal/(mole ⫻ K), near the maximum configurational effect expected by the
random distribution of 3Al ⫹ 1Si atoms on four tetrahedral sites.
According to Hovis (1974), the intracrystalline disorder in monoclinic sani-
dine may be derived directly from cell edges by applying
( )
Q od = 2 X Al,T1 − X Al,T 2 = 7.6344 − 4.3584 b + 6.8615c , (5.179)
where XAl,T1 and XAl,T2 are the atomic fractions of Al 3⫹ on T1 and T2 sites.
Note that the ordering parameter adopted by Hovis (1974) differs from the
form adopted for the triclinic phase of the NaAlSi3O8 end-member (eq. 5.168).
Helgeson et al. (1978) have shown that the enthalpy of disordering in sanidine
is similar to that observed in albite—i.e.,
∆H od
0
(
= 2650 1 − Q od ) (cal mole ) . (5.180)
However, the structural transition between the triclinic and monoclinic forms
(maximum microcline and high sanidine, respectively) “has no apparent effect on
its thermodynamic behavior” (cf. Helgeson et al., 1978); in other words, no gain
or loss of heat and no entropy variation is associated with the purely structural
modification. In this respect, the KAlSi3O8 end-member differs considerably from
NaAlSi3O8 , for which the contribution of structural disorder to bulk transforma-
tion (displacive plus substitutional) is about one-third in terms of energy. This
difference is evident if we compare the heat capacity functions of the two com-
pounds: a cusp is seen for albite in the triclinic-monoclinic transition zone (T ⫽
1238 K; cf. figure 5.56B), whereas no apparent discontinuity can be detected at
the transition point (T ⫽ 451 ⫾ 47 °C at P ⫽ 1 bar; Hovis, 1974) for KAlSi3O8
(figure 5.56A).
Also for KAlSi3O8 , it is convenient to account for phase transition effects
Figure 5.55 Mutual dependence of Qdi and Qod order parameters. In the upper part of
the figure is outlined the T dependence of substitutional disorder Qod for different values
of Qdi and, in the lower part, the T dependence of the displacive disorder parameter Qdi
for different values of Qod. The heavy lines on the surface of local curves represent the
solution for thermal equilibrium. From E. Salje and B. Kuscholke, Thermodynamics of
sodium feldspar II: experimental results and numerical calculations, Physics and Chemis-
try of Minerals, 12, 99–107, figures 5–8, copyright 䉷 1985 by Springer Verlag. Reprinted
with the permission of Springer-Verlag GmbH & Co. KG.
in the heat capacity function, as previously discussed for NaAlSi3O8 . A fictive

polymorph, K-feldspar, is thus generated, representing KAlSi3O8 in its stable state
of displacive and substitutional disorder at all T conditions.
Table 5.65 lists H 0f,Tr ,Pr, S 0Tr, Pr, V 0Tr ,Pr and the heat capacity function for
NaAlSi3O8 , KAlSi3O8 , and CaAl2Si2O8 polymorphs according to Helgeson et al.
(1978). Note that the enthalpy of formation from the elements at Tr ⫽ 298.15 K
and Pr ⫽ 1 bar for albite and K-feldspar are identical to those of low-T poly-
morphs (low albite and maximum microcline), for the previously noted reasons.
Note also that the heat capacity expressions for all NaAlSi3O8 polymorphs are
initially identical in the low-T range. Moreover, as we see in figure 5.56B, the
Silicates / 359
Figure 5.56 Maier-Kelley heat capacity functions for various structural forms of the
KAlSi3O8 (A) and NaAlSi3O8 (B) feldspar end-members (solid curves), compared with T
derivative of H 0T -H0Tr finite differences at various T (filled symbols) (from Helgeson et al.,
1978; redrawn and reproduced with permission).
second heat capacity function for the fictive form albite (valid for T ⬎ 473 K)
conforms quite well to experimental observations (H 0Pr ,T ⫺ H 0Pr ,473 finite differ-
ence derivatives) up to the T of triclinic-monoclinic transition (T ⫽ 1238 K).
It is generally assumed that CaAl2Si2O8 , based on the Al avoidance principle,
has complete ordering of Al and Si on tetrahedral sites at all temperatures. In-
deed, some authors suggest that ordering is not complete at high T (see, for in-
stance, Bruno et al., 1976). Moreover, anorthite at high T (⬎2000 K) in the disor-
dered state has I1 symmetry, unlike the P1 symmetry at low T (Bruno et al., 1976;
Chiari et al., 1978). According to Carpenter and MacConnell (1984), the P1 ⇔
I1 transition is of high order (as in the previous cases). Transition temperatures
between the P1 ⇔ I1 forms for pure anorthite at P ⫽ 1 bar are estimated by
Carpenter and Ferry (1984) to be around 2000 to 2250 K and (implicitly) the
authors consider structural transition as coincident with substitutional order ⇔
Table 5.65 Thermodynamic properties of feldspar end-members in various structural

forms according to Helgeson et al. (1978). Heat capacity function: CP ⫽ K1 ⫹ K2T ⫹
K3T ⫺2; H f,T
0
r ,Pr
in kcal/mole; S 0Tr ,Pr in cal/(mole ⫻ K); V T0r ,Pr in cm3/mole.
End-Member Form H 0f,Tr ,Pr S 0Tr ,Pr V 0Tr ,Pr

NaAlSi3O8 Low albite ⫺939.680 49.51 100.07
NaAlSi3O8 High albite ⫺937.050 52.30 100.43
NaAlSi3O8 Albite ⫺939.680 49.51 100.25
KAlSi3O8 Maximum microcline ⫺949.188 51.13 108.74
KAlSi3O8 High sanidine ⫺946.538 54.53 109.01
KAlSi3O8 K-feldspar ⫺949.188 51.13 108.87
CaAl2Si2O3 Anorthite ⫺1007.772 49.10 100.79
CP
End-Member Form K1 K2 ⫻ 103 K3 ⫻ 10⫺5 Tlimit

NAlSi3O8 Low albite 61.70 13.90 ⫺15.01 < 1400
NAlSi3O8 High albite 61.70 13.90 ⫺15.01 > 623
64.17 13.90 ⫺15.01 < 1400
NAlSi3O8 Albite 61.70 13.90 ⫺15.01 < 473
81.880 3.554 ⫺50.154 < 1200
KAlSi3O8 Maximum microcline 63.83 12.90 ⫺17.05 < 1400
KAlSi3O8 High sanidine 63.83 12.90 ⫺17.05 < 1400
KAlSi3O8 K-feldspar 76.617 4.311 ⫺29.945 < 1400
CaAl2Si2O8 Anorthite 63.311 14.794 ⫺15.44 < 1700
disorder transition. This implicit consideration leads to an estimated transition

enthalpy ( ⌬Htrans ) of 3.7 ⫾ 0.6 kcal/mole and a transition entropy ( ⌬Strans ) of 1.4
to 2.2 cal/(mole ⫻ K). The data in table 5.65 on anorthite are specific to the P1
polymorph and do not comprise the transition effect that is outside the T limit
of validity of the CP function.
5.7.4 Mixing Properties

Most natural feldspars are chemically represented by the ternary system
NaAlSi3O8-KAlSi3O8-CaAl2Si2O8 . Miscibility among alkaline terms is complete,
like the miscibility between NaAlSi3O8 and CaAl2Si2O8 (“plagioclase” series),
whereas an extended miscibility gap (whose limbs vary with T and P) is observed
between KAlSi3O8 and CaAl2Si2O8 . The miscibility gap field extends within the
ternary system, as shown in figure 5.57, which also lists names conventionally
assigned to the various compositional terms of the plagioclase series.
For better comprehension of the mixing properties in the ternary system,
we must first examine in detail the binary mixtures NaAlSi3O8-KAlSi3O8 ,
NaAlSi3O8-CaAl2Si2O8 , and KAlSi3O8-CaAl2Si2O8 .
Silicates / 361
Figure 5.57 Mixing properties in ternary system NaAlSi3O8-KAlSi3O8-CaAl2Si2O8 at

Ptotal ⫽ PH2O ⫽ 1 kbar and various T, based on data from Seck (1971).
NaAlSi3O8-KAlSi3O8 Mixture
Figure 5.58 shows subsolidus stability relationships in the NaAlSi3O8-KAlSi3O8
system at various Ptotal ⫽ Pfluid. At low P, the system is pseudobinary, because
KAlSi3O8 melts incongruently to form leucite (KAlSi2O6 ) plus liquid. However,
melting relations are not of interest here and will be treated to some extent later.
We note that, at subsolidus, an extended miscibility gap exists between Na-rich
(Absm ) and K-rich (Orsm ) terms. This gap also expands remarkably with increas-
ing P, due to the volume properties of NaAlSi3O8-KAlSi3O8 mixtures.
As a matter of fact, the volumes of NaAlSi3O8-KAlSi3O8 mixtures are not
ideal and have positive excess terms. Deviations from simple Vegard’s rule propor-
tionality are mainly due to edge b. Kroll and Ribbe (1983) provide three polyno-
mial expressions relating the volumes of mixtures to the molar fractions of the
potassic component (XOr ) valid, respectively, for monoclinic symmetry (complete
disorder), triclinic symmetry of perfect order (low albite and maximum micro-
cline), and the intermediate structural state:
XOr = −584.6683 + 2.58732V − 3.83499 × 10 −3V 2

(5.181.1)
+ 1.90428 × 10 −6 V 3
XOr = −1227.8023 + 5.35958V − 7.81518 × 10 −3V 2

(5.181.2)
+ 3.80771 × 10 −6 V 3
Figure 5.58 Phase stability relations in (pseudo) binary system NaAlSi3O8-KAlSi3O8 at

various P and T conditions (Le ⫽ leucite; Or ⫽ orthoclase; Sa ⫽ sanidine; L ⫽ liquid;
V ⫽ vapor). Here, “orthoclase” identifies simply an intermediate state of substitutional
disorder.
XOr = −929.1523 + 4.07032V − 5.96146 × 10 −2 V 2

(5.181.3)
+ 2.91994 × 10 −5V 3 .
The subsolidus diagram of figure 5.58 is extremely simplified although based on

various experimental evidence. It does not refer to the structural form of the
various phases at the P, T, and X of interest. Actually, in order to describe appro-
priately the mixing properties of NaAlSi3O8-KAlSi3O8 at subsolidus, their struc-
tural state must be taken into account. The stable form of NaAlSi3O8 immediately
below the melting temperature (T ⫽ 1118 °C; Greig and Barth, 1938) is mono-
clinic (“monalbite”). If monalbite is cooled rapidly, a triclinic polymorph forms,
without modification of Al and Si intracrystalline disorder, by simple collapse of
the coordinating oxygens around Na as a result of attenuation of thermal vibra-
tional motions (displacive transformation). The resulting form is called “anal-
bite” and is unstable at all temperatures. If the cooling process is sufficiently slow,
monalbite changes into high albite (triclinic) as a result of a diffusive process
involving Al and Si. At P ⫽ 1 bar, the transition takes place at T ⫽ 980 °C (Kroll
et al., 1980). The presence of KAlSi3O8 in the mixture lowers the temperature in
an almost linear fashion (Kroll et al., 1980):
( )
Ttrans oC = 978 − 19.2 X Or . ( 5.182 )
According to Merkel and Blencoe (1982), the relationship among the amount of
KAlSi3O8 , pressure, and temperature of transition is
Silicates / 363
Ttrans
XOr = 0.474 − 0.361 + 25.2Ptrans , ( 5.183)
1000
where Ttrans is in K and Ptrans is in bar.

Pure KAlSi3O8 behaves essentially like NaAlSi3O8 , with high-order transfor-
mations that progressively modify the structure. With the decrease of T, high
sanidine (monoclinic, perfectly disordered) is transformed into low sanidine and/
or orthoclase (medium disorder) and then into microcline (triclinic, perfectly or-
dered). The high-T miscibility between the sodic and potassic terms is thus essen-
tially that between a triclinic phase (high albite) and a monoclinic phase (sani-
dine): because miscibility is not complete, we reserve the term “miscibility gap”
for the compositional extension of unmixed phases. At low T, both end-members
are triclinic: if the components unmix, we have a coherent solvus (and a coherent
spinodal field) in the presence of elastic strain energy, and a strain-free or “chemi-
cal” solvus (and a chemical spinodal field) in the absence of strain energy. The
difference between the two unmixing phenomena is discussed in detail in sections
3.11, 3.12, and 3.13.
The petrological and mineralogical literature contains a plethora of names identifying

unmixing products in the NaAlSi3O8-KAlSi3O8 system. However, the adopted terminol-
ogy does not distinguish clearly between the structural states of the unmixed phases, which
is necessary to refer the phenomena to a well-defined thermodynamic process.
The unmixing products of alkaline feldspars are usually defined as “perthites,” which
are further subdivided as follows:
1. Macroperthites. When the unmixed K-rich and Na-rich zones may be observed
and characterized directly under the microscope;
2. Microperthites. When the existence of unmixing is anticipated under the micro-
scope by widespread turbidity or the presence of spots, but is clearly detectable
only by X-ray investigation;
3. Cryptoperthites. When unmixing is detected only by X-ray investigation.
Perthites are also subdivided according to their chemical composition into:
1. Perthites sensu stricto. Unmixing on an initially homogeneous K-rich phase.

2. Antiperthites. Unmixing on an initially homogeneous Na-rich phase.
3. Mesoperthites. Unmixing on an initially homogeneous phase of intermediate
composition.
“The compositional relations of the sodic and potassic phases in perthites

depend on whether the phases are coherent or not. If the perthitic phases are
non-coherent (or if the rock consists of separate grains of sodic and potassic
feldspar) their equilibrium compositions are given by the strain-free solvus. This
is the solvus which has been studied extensively in the past.” These sentences,
Figure 5.59 Effects of structural and coherence state on extent of unmixing in the NaAl-
Si3O8-KAlSi3O8 binary join. (A) Maximum microcline–low albite, based on data from
Bachinski and Müller (1971) and Yund (1974). (B) Sanidine–high albite, based on data
from Thompson and Waldbaum (1969) and Sipling and Yund (1976).
taken entirely from Yund and Tullis (1983), lead us to a careful evaluation of the
real nature of perthitic unmixing in rocks and of the role of coherence energy in
determining feldspar phase relations. Figure 5.59A shows the approximate posi-
tion of the chemical solvus between maximum microcline and low albite ac-
cording to Bachinski and Müller (1971), compared with the coherent solvus de-
termined by Yund (1974). Figure 5.59B shows the miscibility gap between
sanidine and high albite and the corresponding coherent solvus, based on data
from Thompson and Waldbaum (1969) and Sipling and Yund (1976). Note that
the locations of the unmixing fields depend markedly on the structural state of
the components (compare parts A and B) and, structural state being equal, on
coherence energy. It is thus essential to determine both structure and coherence
of unmixed phases in order to assess the P and T conditions of unmixing.
Table 5.66 furnishes a synthesis of Margules subregular interaction parame-
ters according to various authors for the NaAlSi3O8-KAlSi3O8 join, collected by
Ganguly and Saxena (1987). The various sets of values show important differ-
ences that lead to consistent discrepancies in the calculated Gibbs free energy
values of the mixtures. This nonhomogeneity partly reflects the terminological
and operative confusion existing in the literature in the treatment of feldspar ther-
modynamics, and partly is the result of experimental artifacts.
NaAlSi3O8-CaAl2Si2O8 Mixture
Components NaAlSi3O8 and CaAl2Si2O8 show complete miscibility over a wide
range of P and T at subsolidus. Figure 5.60 shows the classical “lens diagram” of
phase stability of plagioclases. This conformation is indicative of ideality of mix-
ing in both liquid and solid states, and will be explained in detail in chapter 7. As
Silicates / 365
Table 5.66 Margules subregular parameters for the NaAlSi3O8-KAlSi3O8

(Ab-Or) binary join (Ganguly and Saxena, 1987). (1) Brown and Parson
(1981); (2) Thompson and Waldbaum (1969); (3) Hovis and Waldbaum
(1977), Thompson and Hovis (1979); (4) Haselton et al. (1983). Note that
W12 ⫽ WH,12 ⫺ TWS,12 ⫹ PWV,12 and Gexcess mixing ⫽ X1X2 (W12X2 ⫹ W21X1),
with 1 ⫽ NaAlSi3O8, 2 ⫽ KAlSi3O8, WH cal/mole, WS cal/(mole ⫻ K),
and WV ⫽ cal/(mole ⫻ bar).
(1) (2) (3) (4)
WH,12 2890 6327 7474 4496

WH,21 6100 7672 4117 6530
WH,12 (T<1000 K) 2359 1695 2273
WH,21 (T<1000 K) 3810 3377 4117
WS,12 0.531 4.632 5.200 2.462
WS,21 2.290 3.857 0.000 2.462
WV,12 0.079 0.092 0.086 0.087
WV,21 0.168 0.112 0.086 0.087
figure 5.60 shows, the structural form of mixtures at subsolidus is not unique
over the entire compositional range, but can be schematically subdivided into two
domains (P1 and I1) limited by a straight boundary between An59 at T ⫽ 1000
°C and An77 at T ⫽ 1400 °C (P ⫽ 1 bar; Carpenter and Ferry, 1984). If this
subdivision is adopted, the two-phase field (solid plus liquid) is also necessarily
subdivided into two domains: one constituted of P1 mixture plus liquid, and the
other of I1 mixture plus liquid, as shown in the figure.
Newton et al. (1980) calorimetrically measured the enthalpy of the NaAl-
Si3O8-CaAl2Si2O8 mixture at T ⫽ 970 K and P ⫽ 1 bar and found a positive
excess enthalpy reproduced by a subregular Margules model:
( )
H excess mixing = X 1X 2 WH ,12 X 2 + WH , 21X 1 , (5.184)
where 1 ⫽ NaAlSi3O8 , 2 ⫽ CaAl2Si2O8 , WH,12 ⫽ 6.7461 kcal/mole, and WH,21 ⫽

2.0247 kcal/mole. Concerning the entropy of the mixture, Kerrick and Darken
(1975) proposed a simple calculation of the configurational effect of mixing based
on the Al avoidance principle.
In anorthite, tetrahedral groups [TO4 ] lying parallel to the (123) plane exhibit perfect
alternation of Al-bearing and Si-bearing strata. Let us define NT as the total number of
tetrahedra occupied by Al and Si:
NT = NAl + N Si . ( 5.185)
We also define the sites occupied by Al as ␣ and those occupied by Si as ␤.

If we replace half the Al with Si on a tetrahedral plane, we have random mixing over
1
2NT sites. This sort of mixing does not contradict the Al avoidance principle. Let us now
Figure 5.60 X-T phase stability diagram for the NaAlSi3O8-CaAl2Si2O8 system at P ⫽ 1
bar, showing transition between forms P 1 and I 1 at subsolidus and within two-phase
field. Note that form of disorder phase is usually “mediated” into crystalline class C 1,
hence in the literature the phase transition is usually described as C 1 ⇔ I 1. From M. A.
Carpenter and J. M. Ferry, Constraints on the thermodynamic mixing properties of pla-
gioclase feldspars, Contributions to Mineralogy and Petrology, 87, 138–48, figure 3, copy-
right 䉷 1984 by Springer Verlag. Reprinted with the permission of Springer-Verlag
GmbH & Co. KG.
consider the exchange of one Si atom in a ␤ site with one Al atom in an ␣ site: the exchange
is possible only when the resulting configuration does not result in Al-O-Al alternation.
Because the probability that an Si atom in a ␤ site is surrounded by four other Si neighbors
is ( 12)4 ⫽ 16
1
, the total number of sites available for Al-Si mixing, based on the Al avoidance
principle, is
1
NAl ⇔ Si,α = NT ( 5.186 )
2
1 1
NAl ⇔ Si, β = × NT . ( 5.187 )
2 16
Therefore, in practice, mixing takes place over ␣ sites only. Applying this condition to the
stoichiometry of the binary mixture, we obtain
N Si,α = N Na ( 5.188 )
NAl = 2 NCa + N Na ( 5.189 )
N Si,α + NAl = 2( NCa + N Na ) ( 5.190 )
(cf. Kerrick and Darken, 1975). If we now recall the equation for permutability Q (section
3.8.1) and apply the calculation to the four atoms in the mixture (Si, Al, Na, Ca), because
Na and Ca mix over one M site and Si and Al mix essentially over ␣ sites, we obtain
Silicates / 367
Q=
(N Na + NCa ) !
×
(N Si,α + N Al ! )
N Na! NCa! N Si,α ! N Al !
[ 2( N ]
(5.191)
(N Na + NCa ) ! Ca + N Na ) !
= × .
N Na! NCa! N Na! ( 2 NCa + N Na )!
Application of Stirling’s formula to equation 5.191 and comparison with the configura-
tional state of pure components lead to the definition of a configurational entropy of
mixing term in the form
  X (1 + X ) 2 
  . ( 5.192 )
S mixing conf [ 2
]
= − R X Ab ln X Ab (1 − X An ) + X An ln 

An
4
An

  
In the absence of enthalpic contributions, the activities of the components in the mix-
ture become
a Ab = X Ab
2
( 2 − XAb ) ( 5.193 )
and
X An ( 1 + XAn ) .
1 2
a An = ( 5.194 )
4
(In eq. 5.193 and 5.194, Ab or An simply identifies the component without any implication
about its structural state.)
Figure 5.61B shows how the entropy of mixing derived by application of the
Al avoidance principle compares with the one-site ideal mixing contribution. Also
shown are entropy of mixing curves obtained by summing, up to the ideal one-
site contribution, additional terms related to phase transition effects [variable be-
tween 1 and 2.5 cal/(mole ⫻ K)] (Carpenter and Ferry, 1984). Figure 5.61C com-
pares the Gibbs free energy of mixing experimentally observed by Orville (1972)
at T ⫽ 700 °C and P ⫽ 1 bar with Gibbs free energy values obtained by combining
the entropy effect of the Al avoidance principle and the enthalpy of mixing experi-
mentally observed by Newton et al. (1980) (see also figure 5.61A):

( ) 
2
X2 1 + X2
G mixing = H mixing


[ (
+ RT X 1 ln X 12 2 − X 1 )] + X 2 ln
4

.

(5.195)
 
As figure 5.61C shows, the equation of Kerrick and Darken (1975), coupled with
the enthalpy of mixing terms of Newton et al. (1980), satisfactorily fits the experi-
Figure 5.61 Mixing properties of plagioclases. (A) Solution enthalpy in Pb2B2O5 at T ⫽

970 K and P ⫽ 1 bar after Newton et al. (1980). Enthalpy of mixing of substance is
obtained by subtracting value of solution enthalpy actually measured for substance from
ideal solution enthalpy (dashed line). (B) Entropy of mixing of Kerrick-Darken model
compared with one-site mixing configurational entropy. Dashed lines: ordering contribu-
tions that may arise from C 1 ⇔ P 1 phase transition effects (Carpenter and Ferry, 1984).
(C) Experimental Gibbs free energy of mixing curve at T ⫽ 970 K and P ⫽ 1 bar (Orville,
1972; solid line) compared with various models.
mental observations of Orville (1972). The same values cannot be reproduced

either by an ideal one-site mixing model or by a regular mixture, as shown in the
figure. Moreover, the phase transition contribution postulated by Carpenter and
Ferry (1984) does not appear to affect the mixing properties of the substance to
any significant extent.
Silicates / 369
Saxena and Ribbe (1972) have shown that the excess Gibbs free energy of
mixing of the mixture, based on the data of Orville (1972), may be reproduced
by a subregular Margules model:
( )
G excess mixing = X 1X 2 W12 X 2 + W21X 1 , (5.196)
where 1 ⫽ NaAlSi3O8 , 2 ⫽ CaAl2Si2O8 , W12 ⫽ 252 cal/mole, and W21 ⫽ 1682

cal/mole.
The mixing properties of plagioclases discussed above refer to high-T struc-
tural forms. The corresponding properties for low-T forms are extremely complex
and only a few (and contrasting) thermochemical data exist at present. For in-
stance, according to Kotelnikov et al. (1981), at T ⫽ 700 °C and P ⫽ 1 kbar, a
miscibility gap exists between An67 and An92 (i.e., “Huttenlocher gap”; cf. Smith,
1983), although this contrasts with the analogous study by Orville (1972) at P ⫽
2 kbar (same T). Moreover, we have already mentioned the existence of the
“peristerite gap” at low An amounts (see Smith, 1983, for a detailed account of
low-T unmixing processes in plagioclases).
KAlSi3O8-CaAl2Si2O8 Mixtures
The subsolidus properties of the KAlSi3O8-CaAl2Si2O8 system are essentially
those of a mechanical mixture (cf. section 7.1). According to Ghiorso (1984), the
miscibility gap is not influenced by pressure and may be described by a strongly
asymmetric Margules model:
( )
G excess mixing = X 1X 2 W12 X 2 + W21X 1 , (5.197)
where 1 ⫽ KAlSi3O8 , 2 ⫽ CaAl2Si2O8 , and
W12 = WH ,12 − TWS ,12 , (5.198)
where WH,12 ⫽ 16,125 cal/mole, WH,21 ⫽ 6688 cal/mole, WS,12 ⫽ 2.644 cal/(mole
⫻ K), and WS,21 ⫽ ⫺4.830 cal/(mole ⫻ K).
NaAlSi3O8-KAlSi3O8-CaAl2Si2O8 Ternary Mixtures

The most complete set of experiments concerning mixing properties in the
NaAlSi3O8-KAlSi3O8-CaAl2Si2O8 system is given by Seck (1971) (cf. figure 5.57).
Already in 1973, Saxena had rationalized the experimental results of Seck (1971)
in thermodynamic form, representing ternary interactions in the framework of
Wohl’s model (see section 3.10.1). Later, Barron (1976) evaluated the reliability
of Wohl’s and Kohler’s models in reproducing the data of Seck (1971) and de-
duced that Wohl’s model better reproduced the slope of tie lines connecting the
compositions of unmixed phases on opposite sides of the miscibility gap limb.
Figure 5.62 Experimentally observed distribution of NaAlSi3O8 component between

plagioclases and alkaline feldspars (Seck, 1971) compared with results of Wohl’s and
Kohler’s models. Reproduced from Barron (1976), with modifications.
Figure 5.62 shows in detail the distribution of the NaAlSi3O8 component between
unmixed phases at T ⫽ 900 °C and P H2O ⫽ 0.5 kbar, as observed by Seck (1971)
and compared with Wohl’s and Kohler’s predictions. In both models, the binary
interaction parameters are those deduced by Saxena (1973) on the basis of Or-
ville’s (1972) data for the NaAlSi3O8-CaAl2Si2O8 system, coupled with the experi-
mental observations of Seck (1971).
A more recent model (Ghiorso, 1984) is based on the binary interaction pa-
rameters of Thompson and Hovis (1979) for the NaAlSi3O8-KAlSi3O8 join and
on the experimental results of Newton et al. (1980), coupled with the Al avoid-
ance principle of Kerrick and Darken (1975) extended to the ternary field. Ghi-
orso (1984) expressed the excess Gibbs free energy of mixing in the form
 1   1 
G excess mixing = W12 X 1X 2  X 2 + X 3  + W21X 1X 2  X 1 + X 3 
 2   2 
 1   1 
+ W13X 1X 3  X 3 + X 2  + W31X 1X 3  X 1 + X 2  (5.199)
 2   2 
 1   1 
+ W32 X 3X 2  X 2 + X 1  + W23X 3X 2  X 3 + X 1  + W123X 1X 2 X 3 ,
 2   2 
where 1 ⫽ NaAlSi3O8 , 2 ⫽ CaAl2Si2O8 , and 3 ⫽ KAlSi3O8 .

Binary interaction parameters W23 and W32 and ternary interaction term W123
were obtained by Ghiorso (1984) by imposing isoactivity on the various compo-
Silicates / 371
Table 5.67 Interaction parameters for ternary feldspar mixtures

according to Ghiorso (1984). 1 ⫽ NaAlSi3O8; 2 ⫽ CaAl2Si2O8;
3 ⫽ KAlSi3O8.
AB WH,AB , cal/mole WS,AB , cal/(mole ⫻ K) WV,AB , cal/(mole ⫻ bar)

12 6,746 0 0
21 2,025 0 0
13 7,404 5.12 0.086
31 4,078 0 0.086
23 6,688 ⫺4.83 0
32 16,126 2.64 0
123 ⫺3,315 ⫺3.50 0
W ⫽ WH,AB ⫺ T WS,AB ⫹ P WV,AB
nents in the unmixed phases of Seck (1971). Table 5.67 reports the resulting inter-
action parameters.
5.8 Silica Minerals
Silica has 22 polymorphs, although only some of them are of geochemical inter-
est—namely, the crystalline polymorphs quartz, tridymite, cristobalite, coesite,
and stishovite (in their structural modifications of low and high T, usually desig-
nated, respectively, as ␣ and ␤ forms) and the amorphous phases chalcedony
and opal (hydrated amorphous silica). The crystalline polymorphs of silica are
tectosilicates (dimensionality ⫽ 3). Table 5.68 reports their structural properties,
after the synthesis of Smyth and Bish (1988). Note that the number of formula
units per unit cell varies conspicuously from phase to phase. Also noteworthy is
the high density of the stishovite polymorph.
Quartz is a primary phase in acidic igneous rocks (both intrusive and effusive) and
in metamorphic and sedimentary rocks. Tridymite is a high-T form found in porosities of
rapidly quenched effusive rocks (mainly andesites and trachites). Cristobalite is rarely
found in volcanic rocks and sometimes occurs in opal gems. Coesite is a marker of high-
P regimes. It has been found coexisting with garnet in the Dora-Maira Massif (Western
Alps) and in various other high-P metamorphic terrains (the former Soviet Union,
China). Stishovite is stable above about 80 kbar (cf. Liu and Bassett, 1986, and references
therein) and is found exclusively in “impactites” (i.e., rocks formed by meteorite impact).
Chalcedony and opal are amorphous phases generated by hydrothermal deposition from
SiO2-saturated solutions.
Figure 5.63A shows the relative positions of Si atoms in the ␣-quartz structure
projected along axis z. In this simplified representation, note the hexagonal ar-
rangement of Si atoms, which can be internally occupied by univalent cations
through coupled substitution involving Al 3⫹ in tetrahedral position:
Table 5.68 Structural properties of SiO2 polymorphs (from Smyth and Bish, 1988).
Polymorph Quartz Coesite Stishovite Cristobalite Tridymite

Formula unit SiO2 SiO2 SiO2 SiO2 SiO2
Molar weight 60.085 60.085 60.085 60.085 60.085
(g)
Density (g/cm3) 2.648 2.909 4.287 2.318 2.269
Molar volume 22.688 20.657 14.017 25.925 26.478
(cm3)
Z 3 16 2 4 48
System Trigonal Monoclinic Tetragonal Tetragonal Monoclinic
Class 32 2/m 4/mmm 422 m
Group P3221 C2/c P42/mnm P41212 Cc
Cell parameters:
a 4.9134 7.1464 4.1790 4.978 18.494
b 4.9134 12.3796 4.1790 4.978 4.991
c 5.4052 7.1829 2.6651 6.948 25.832
␤ 120.283 117.75
Volume (Å3) 113.01 548.76 46.54 172.17 2110.2
Reference Le Page Smyth et Baur and Peacor (1973) Kato and
et al. (1980) al. (1987) Kahn (1971) Nukui (1976)
Figure 5.63 Quartz structure projected along axis z. Filled dots are Si atoms (oxygen
atoms not shown). ␣-quartz (A) - ␤-quartz (B) transition takes place by simple rotation
of tetrahedra on ternary helicogyre (adapted from Gottardi, 1972).
SiT× + ’ ⇔ Na ⋅’ + AlT ( 5.200 )
(the hexagonal ring is actually a spiral with atoms positioned on separate planes).
Figure 5.63B shows the structure of high-T ␤-quartz. The ␣-quartz–␤-quartz
transition, which may be described as the overlap of a first-order and a ␭ transi-
tion, takes place by simple rotation of tetrahedra on the ternary helicogyre.
Silicates / 373
Figure 5.64 Effect of cooling rate on SiO2 polymorphic transitions. From Putnis and
McConnell (1980). Reproduced with modifications by permission of Blackwell Scientific
Publications, Oxford, Great Britain.
The structural arrangements of [SiO4 ] groups in quartz, tridymite, and cristo-

balite are different, and the corresponding first-order phase transitions are diffi-
cult to complete, often giving rise to metastability. At P ⫽ 1 bar, for instance, ␤-
cristobalite is stable only at T ⬎ 1470 °C. With decreasing T, we should observe
transition to tridymite, but this transformation requires substantial re-
arrangement of tetrahedra and is rather sluggish. If the cooling rate is high, we
observe metastable persistence of ␤-cristobalite down to T ⫽ 230 °C, where the
␤-cristobalite–␣-cristobalite ␭ transition takes place. The same metastability is
observed for the tridymite polymorph with respect to quartz, as shown in figure
5.64.
Table 5.69 lists thermal expansion and compressibility factors for some SiO2
polymorphs, according to the databases of Saxena et al. (1993) and Holland and
Powell (1990). Table 5.70 lists thermodynamic data for the various SiO2 poly-
morphs according to various sources.
The amount of H2O in amorphous silica (number n of H2O molecules per
unit formula) varies between 0.14 and 0.83 (Frondel, 1962). Nevertheless, the
thermodynamic properties of the phase are not particularly affected by the value
of n (Walther and Helgeson, 1977). The molar volume of opal is 29 cm3 /mole.
The same volume of ␣-quartz may be adopted for chalcedony; see table 5.68 for
the other polymorphs.
We have already stated that the ␣-␤ transition of quartz may be described as a ␭
transition overlapping a first-order transition. The heat capacity function for the two poly-
morphs is thus different in the two stability fields, and discontinuities are observed in the
H and S values of the phase at transition temperature (Ttrans cf. section 2.8). For instance,
to calculate the thermodynamic properties of ␤-quartz at T ⫽ 1000 K and P ⫽ 1 bar, we
Table 5.69 Thermal expansion at P ⫽ 1 bar (K-1) and bulk modulus at T ⫽ 25°C (Mbar) for some
SiO2 polymorphs in the databases of Saxena et al. (1993) (1) and Holland and Powell (1990) (2).
Thermal expansion at various T values of interest is obtained by applying
␣(T) ⫽ ␣0 ⫹ ␣1T ⫹ ␣2T ⫺1 ⫹ ␣3T ⫺2
Thermal Expansion Bulk Modulus
Polymorph ␣0 ⫻ 105 ␣1 ⫻ 108 ␣2 ⫻ 104 ␣3 ⫻ 102 K0 K⬘ Reference

␣-quartz 2.7513 2.9868 0.05572 9.1181 0.3097 6.4 (1)
3.5260 - - - 0.3846 - (2)
␤-quartz 2.0604 3.4694 1.30754 ⫺163.76 0.5149 5.3 (1)
0.9107 - (2)
Coesite 0.543 0.76 0 0 0.9135 8.4 (1)
1.066 - - - 0.969 - (2)
Tridymite 2.1797 ⫺1.53 0 0 0.25 6.0 (1)
Cristobalite 0.5 0 0 0 0.1949 6.0 (1)
Stishovite 0.23 1.2 62.0 ⫺113.0 3.1268 6.0 (1)
Table 5.70 Thermodynamic data for silica polymorphs. Stishovite and tridymite
from Saxena et al. (1993); remaining polymorphs from Helgeson et al. (1978).
H 0f, Tr , P r in kJ/mole; S 0Tr , P r in J/(mole ⫻ K). Heat capacity function is
C P ⫽ K 1 ⫹ K 2 T ⫹ K 3T ⫺2 ⫹ K 4T 2 ⫹ K 5T ⫺3 ⫹ K 6 T ⫺1/2 ⫹ K 7T ⫺1.
Polmorph Formula Unit H f,0 T r , P r S T0 r , P r K1 K2 ⫻ 103 K3 ⫻ 10⫺5

Quartz (␣ and ␤) SiO2 ⫺910.648 41.34 46.944 34.309 ⫺11.297
60.291 8.117 0
Cristobalite (␣ and ␤) SiO2 ⫺906.903 43.396 58.492 13.975 ⫺15.941
72.760 1.297 ⫺41.380
Tridymite SiO2 ⫺906.913 45.116 66.6993 5.2779 ⫺21.323
Coesite SiO2 ⫺906.313 40.376 46.024 34.309 ⫺11.297
59.371 8.117 0
Stishovite SiO2 ⫺864.000 29.50 58.12 7.002 ⫺126.89
Chalcedony SiO2 ⫺909.108 41.337 46.944 34.309 ⫺11.297
Amorphous silica SiO2 ⫻ nH2O ⫺897.752 59.998 24.811 197.485 ⫺95.312
Polymorph K4 ⫻ 107 K5 ⫻ 10⫺8 K6 ⫻ 10⫺7 K7 ⫻ 10⫺4 Tlimit (K)

Quartz (␣ and ␤) 0 0 0 0 848
0 0 0 0 2000
Cristobalite (␣ and ␤) 0 0 0 0 543
0 0 0 0 2000
Tridymite ⫺3.5478 0 0 0 –
Coesite 0 0 0 0 848
0 0 0 0 2000
Stishovite 0 17.928 0 1.7012 –
Chalcedony 0 0 0 0 848
Amorphous silica 0 0 0 0 622
Silicates / 375
Table 5.71 Transition energy terms for quartz and cristobalite: ⌬S 0trans ⫽ molar
apparent transition entropy at P ⫽ 1 bar [ J/(mole ⫻ K)]; ⌬H 0trans ⫽ molar
apparent transition enthalpy at P ⫽ 1 bar (kJ/mole); ⌬V 0trans ⫽ molar apparent
transition volume at P ⫽ 1 bar (cm3/mole).
Transition Ttrans ⌬S 0trans ⌬H trans

0
⌬V trans
0
␣-quartz → ␤-quartz 848 1.431 1.213 0.372

␣-cristobalite → ␤-cristobalite 543 2.472 1.343 (?)
proceed as follows: calculate enthalpy and entropy at Ttrans ⫽ 848 K with the aid of the
heat capacity expression valid for ␣-quartz (integral between Tr ⫽ 298.15 K and T ⫽
Ttrans ⫽ 848 K); to the obtained values add the apparent standard molar ␣-␤ transition
enthalpy and entropy at Ttrans; calculate the integrals between Ttrans and T ⫽ 1000 K with
the CP function valid for the ␤-phase and add the results to the previously obtained en-
thalpy and entropy values. Note that, at the transition point, the two polymorphs coexist
and thus have the same Gibbs free energy. Table 5.71 lists ␣-␤ transition enthalpy, entropy,
and volume values for quartz and cristobalite. These values are consistent with the
␣-phase of Helgeson et al. (1978). An extensive discussion on the energetics of SiO2 poly-
morphs may be found in Helgeson et al. (1978) and Berman (1988).
5.9 Thermobarometric Properties of Silicates
Estimates of T and P conditions of equilibrium for a given mineralogical paragen-

esis are of fundamental importance in understanding geologic processes. From
the first attempts to define metamorphic regimes (P, T, anistropic P) through
index minerals, advances in physical geochemistry progressively allowed quanti-
fication of the reaction, exchange, and ordering processes in minerals that lead to
consistent thermobarometric applications. Thermobarometry is now a common
tool in geochemistry. We will later examine briefly the various types of processes
usually considered in thermobarometric studies. As a premise, reasoning in terms
of “equilibrium,” the deduced P and T values correspond to the last condition at
which equilibrium was complete, further exchanges being prevented by kinetic
reasons (i.e., low elemental diffusivites and/or high cooling rates).
5.9.1 Intercrystalline Exchange Geothermometry

Let us consider the exchange reaction of two ions, A and B, between two phases,
␣ and ␤, composed of binary mixtures (A,B)M ( ␣ ) and (A,B)N ( ␤ ):
A α + B β ⇔ A β + Bα . ( 5.201)
The Gibbs free energy change involved in reaction 5.201 is given by the sum of the
chemical potentials of components in reaction ( ␮i ), multiplied by their respective
stoichiometric factors (␯i )—i.e.,
∆G 201 = ∑ ν i µi . (5.202)
i
In the case of equation 5.201, stoichiometric factors are 1 for all components;
equation 5.202 may thus be reduced to
∆G 201 = ∑ µi . (5.203)
i
For each component,
µ = µ i0 + RT ln a i (5.204)
is valid, where ␮0i is the standard state chemical potential and ai is thermody-
namic activity.
Applying equation 5.203 to reaction 5.201 and recalling equation 5.204, we
obtain
∆G 201 = ∆G 201
0
+ RT ln K 201, P,T , (5.205)
where ⌬G 0201 is the Gibbs free energy change involved in reaction 5.201 at the
standard state of reference condition and K201,P,T is the equilibrium constant at
the P and T of interest. Because, at equilibrium, ⌬G of the reaction is zero, we
may rewrite equation 5.205 as
∆G 201
0
= − RT ln K 201, P,T . (5.206)
It is evident from equation 5.204 that the intrinsic significance of equation

5.206 is closely connected with the choice of standard state of reference. If the
adopted standard state is that of the pure component at the P and T of interest,
then ⌬G 0201 is the Gibbs free energy of reaction between pure components at the P
and T of interest. Deriving in P the equilibrium constant, we obtain
 ∂ ln K 
= − 
(
 ∂ ∆G 0 RT )  =−
∆V 0
  . (5.207)
 ∂P  T  ∂P  RT
 T
Nevertheless, if the standard state of reference is that of the pure component at

P ⫽ 1 bar and T of interest, then
Silicates / 377
 ∂ ln K 
  = 0. (5.208)
 ∂P  T
Concerning temperature, if the standard state of reference is that of the pure

component at the T and P of interest, we have
 ∂ ln K   ∂ ∆G 0 RT ( )  ∆H 0
  = − = (5.209)
 ∂T  P  ∂T  RT 2
 P
or, deriving in the inverse of T,
 ∂ ln K  ∆H 0
  =− . (5.210)
( )
 ∂ 1 T 
P
R
However, if the standard state is that of the pure component at P ⫽ 1 bar and T
of interest,
∂ ln K ∆H 0
= ( 5.211)
∂T RT 2
not dependent on P.
If we now consider the equilibrium constant expressed as activity product,
then
K 201, P,T = ∏ a iν i
. (5.212)
i
In the case of reaction 5.201, stoichiometric coefficients are 1 and equation 5.212
may be reduced to
a A, β aB,α  X A, β X B,α   γ A, β γ B,α 

K 201,P ,T = =  × . ( 5.213)
a A,α aB, β  X A,α X B, β   γ A,α γ B, β 
The first term in parentheses on the right side of equation 5.213 is the distribution
coefficient (KD ), and the second groups activity coefficients related to the mixing
behavior of components in the two phases. The equilibrium constant is thus re-
lated to the interaction parameters of the two phases at equilibrium. For example,
the equilibrium between two regular mixtures is defined as
Wβ
ln K 201,P ,T = ln K D −
Wα
RT
(
1 − 2 X A ,α +
RT
)
1 − 2 X A ,β , ( ) (5.214 )
where W␣ and W␤ are interaction parameters for phases ␣ and ␤, respectively, or,
for two asymmetric Van Laar mixtures,
A0,α A1,α
ln K201,P ,T = ln K D +
RT
(X A, α − XB, α + ) RT
(6 XB, α )
X A,α − 1
(5.215)
A0, β A1, β
+
RT
(X
B, β − X A, β + ) RT
(6X A, β XB, β − 1 . )
where A0,␣, A1,␣, A0,␤ , and A1,␤ are Guggenheim’s parameters for phases ␣ and ␤,
respectively (Saxena, 1973).
Let us now examine two natural phases, ␣ and ␤, coexisting at equilibrium,
for which we know in detail both mixing behavior (hence W␣ and W␤ , or A0,␣,
A1,␣, A0,␤ , and A1,␤ ) and composition (molar fractions of AM, BM, AN, and
BN components, or the analogous ionic fractions XA,␣, XA,␤, XB,␣, and XB,␤ ).
Application of equation 5.206 allows us to define the loci of P and T conditions
that satisfy equilibrium, provided that in the definitions of both ⌬G 0 and the
thermodynamic activity of components in the mixture we are consistent with the
adopted standard state. It is obvious that the more ⌬G 0 is affected by P and T
intensive variables, and the more the interaction parameters describing the mixing
behavior are accurate, the more efficient and precise is the thermobarometric ex-
pression. By stating “the loci of P and T conditions that satisfy equilibrium,” we
emphasize that we are dealing with univariant equilibria. In a P and T space, the
slope of the univariant curve is defined by equations 5.207 and 5.209; if the vol-
ume of reaction is negligible, then obviously the equilibrium constant is not
affected by P, and the elemental exchange furnishes only thermometric infor-
mation.
Exchange geothermometry generally considers couples of elements of identi-
cal charge showing the property of diadochy—for instance, Na⫹-K⫹, Mg2⫹-Fe2⫹,
or Fe3⫹-Cr3⫹. Intercrystalline exchange thermometers have been developed
mainly on the Mg2⫹-Fe2⫹ couple, and the investigated mineral phases comprise
olivine, orthopyroxene, clinopyroxene, garnet, spinel, ilmenite, cordierite, biotite,
hornblende, and cummingtonite. Table 5.72 lists bibliographic sources for these
geothermometers. More detailed references may be found in Essene (1982) and
Perchuk (1991).
Let us consider, for instance, the Fe2⫹-Mg2⫹ exchange reaction between garnet and
clinopyroxene:
1 1
Mg3Al2Si 3O12 + CaFeSi2O6 ⇔ Fe 3Al2Si 3O12 CaMgSi2O6 .
3 pyrope hedembergite
3 almandine diopside
(5.216)
Silicates / 379
Table 5.72 Fe2⫹-Mg2⫹ exchange geothermometers.
Equilibrium References
Olivine-orthopyroxene-spinel Ramberg and Devore (1951); Nafziger and Muan (1967);
Speidel and Osborn (1967); Medaris (1969); Nishizawa and
Akimoto (1973); Matsui and Nishizawa (1974); Fujii (1977);
Engi (1978)
Ilmenite-orthopyroxene Bishop (1980)
Ilmenite-clinopyroxene Bishop (1980)
Orthopyroxene-clinopyroxene Mori (1977); Herzberg (1978b)
Garnet-Olivine Kawasaki and Matsui (1977); O’Neill and Wood (1979)
Garnet-Cordierite Currie (1971); Hensen and Green (1973); Thompson (1976);
Hensen (1977); Holdaway and Lee (1977); Perchuk (1991)
Garnet-clinopyroxene Saxena (1979); Banno (1970); Oka and Matsumoto (1974);
Irving (1974); Raheim and Green (1974, 1975); Raheim (1975,
1976); Mori and Green (1978); Slavinskiy (1976); Ellis and
Green (1979); Ganguly (1978); Saxena (1979); Dahl (1980);
Perchuk (1991); Krog (1988).
Garnet-biotite Saxena (1969); Thompson (1976); Perchuk (1977, 1991);
Goldman and Albee (1977); Ferry and Spear (1978).
Garnet-chlorite Ghent et al. (1987); Perchuk (1991)
Garnet-staurolite Perchuk (1991) and references therein
Garnet-amphibole Perchuk (1991) and references therein
Garnet-chloritoid Perchuk (1991) and references therein
Biotite-chloritoid Perchuk (1991) and references therein
Chlorite-chloritoid Perchuk (1991) and references therein
The constant of reaction 5.216 is plotted in a semilogarithmic Arrhenius diagram in figure

5.65. In this short of diagram (Ganguly, 1979), the equilibrium constant assumes an essen-
tially straight path, indicating that the reaction enthalpy does not vary significantly with
T at constant pressure (cf. eq. 5.210). Figure 5.65 also shows the distributive constant
XFe,garnet XMg,cpx
KD = , (5.217)
XMg,garnet XFe,cpx
which was experimentally measured by Wood (1976b).

By applying the ionic solution model, we may express the relationship between ionic
fraction and activity in the garnet phase as
( )
3
aMg 3Al 2Si 3O12 ≡ X Mgγ Mg , ( 5.218)
garnet
where ␥Mg is the activity coefficient of ion Mg2⫹ in the structural site and 3 is the stoichio-
metric number of sites over which substitution takes place. Analogously, for clinopy-
roxene,
Figure 5.65 Garnet-clinopyroxene geothermometric exchange (Fe2+-Mg2+). Thermody-

namic constant K is compared with distribution coefficient KD: the difference between the
two terms represents the effect of interactions in mixtures. Reprinted from J. Ganguly,
Geochimica et Cosmochimica Acta, 43, 1021–1029, copyright 䉷 1979, with kind permission
from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
aCaMgSi2 O6 ≡ X Mgγ Mg ( ) cpx

. ( 5.219 )
Applying equation 5.212 to the equilibrium constant, we have
[( ]
13
) (X ) (X ) (γ )
3
X Feγ Fe Mg γ Mg Fe X Mg Fe γ Mg
garnet
K216 = × = × = KD K γ
cpx garnet garnet
(X ) (X ) (γ )
[( ] Feγ Fe γ Mg
13
) X Mg
3
X Mg γ Mg cpx Fe cpx Fe cpx
garnet
( 5.220 )
and, in logarithmic notation,
ln K216 = ln KD + ln Kγ . ( 5.221)
As figure 5.65 shows, the term ln K␥ definitely affects the equilibrium. This term is consis-
tent with a subregular Margules formulation for excess Gibbs free energy terms in both
phases (Ganguly, 1979; see also sections 5.3.5 and 5.4.5). Combining the various experi-
mental evidence, Ganguly (1979) calibrated the distributive function of equation 5.217
over T and P, and proposed the following thermometric expression:
4100 + 11.07P
ln KD = + c, ( 5.222 )
T
where P is in kbar, T is in K, c ⫽ ⫺2.4 for T ⬍ 1333 K, and c ⫽ ⫺2.93 for

T ⬎ 1333 K.
Silicates / 381
As we see, equation 5.222 is ready to use for petrologic applications. Geothermome-

tric expressions of this type are extremely useful but must be applied with caution, with
careful evaluation of the P and T limits of applicability and the effect of the presence
of minor components in the mixture. For instance, for the garnet-clinopyroxene geother-
mometer, the presence of grossular (besides pyrope and almandine) and jadeite (besides
hedembergite-diopside) has been studied by Saxena (1969), Banno (1970), Oka and Mat-
sumoto (1974), and Irving (1974), but thermometric calibration is generally based on ex-
perimental runs in compositionally simple synthetic systems. The extension to chemically
complex natural phases normally involves large degrees of incertitude. Also, some geother-
mometric expressions operate well under certain P and T regimes but cannot be applied
to P and T conditions outside the calibration range. For instance, the formulation of
Ellis and Green (1979) for the geothermometer 5.216 is particularly suitable for granulitic
terrains, because it expressly accounts for the Ca amount in garnet:
3104X Ca,garnet + 3030 + 10.86P

T = ( 5.223 )
ln KD + 1.9039
(T in K and P in kbar); however, its application to other thermobaric conditions is not so

appropriate. The most satisfactory formulations for the exchange reaction 5.216 are those
of Slavinskiy (1976), Ganguly (1979), Saxena (1979), and Dahl (1980), which are based
on a large body of thermodynamic data, petrologic studies, and natural observations.
Besides caution regarding the P-T range of applicability and compositional

limits for a given intracrystalline exchange expression, the role of kinetics in the
attainment of equilibrium must also be emphasized. Reasoning in simple thermo-
dynamic terms, the P and T values deduced from an ionic exchange reaction
represent the final state condition. Moreover, phases at equilibrium must be com-
positionally homogeneous, and this corresponds (in kinetic terms) to sufficiently
high ionic diffusivity for the thermometric couple in both phases.
Figure 5.66A shows the concentration profiles in an ideal intercrystalline ex-
change geothermometer where exchange kinetics do not play any role: elemental
concentrations are constant at all distances from the interface. Figure 5.66B
shows how diffusive kinetics perturb the concentration profiles of the thermomet-
ric couple: elemental diffusion is too slow with respect to the cooling rate of the
system, resulting in marked compositional zonation as a function of the distance
from the interface plane between the two exchanging phases. Obviously, in this
case, application of a given thermometric exchange relation will result in con-
trasting T deductions, depending on the compositional zone of application.
The implications of diffusion kinetics for ionic exchange geothermometry
have been discussed in detail by Lasaga (1983).
Ionic diffusivity varies exponentially with temperature, according to
 −Ea 
D( T ) = A exp , ( 5.224 )
 RT 
Figure 5.66 Elemental concentration profiles in a thermometric couple. (A) Ideal case;
elemental concentration is constant from interface to nucleus of crystal. (B) Nonideal
case: slow diffusivity generates concentration gradients from interface to nucleus of
crystal.
where A is the preexponential factor and Ea is the energy of activation of the diffusive
process.
Diffusivity at time t may be related to initial diffusivity at time t0 by associating with
t the corresponding temperature T:
 E  1 1  
D( t ) = D(0t 0 ) exp  a  − 0  . ( 5.225)
 R  T( t ) T( t 0 )  
Assuming a linear decrease in temperature with time,
T( t ) ≈ T( t0 ) − v t , ( 5.226 )
0
where v is the cooling rate of the system. Equation 5.225 is reduced to
D( t ) = D(0t 0 ) exp(−γ t) , ( 5.227 )
where
Ea v
γ = 2
. ( 5.228 )
RT( t0 )
0
Parameter ␥ is extremely important, because it defines the time within which elemen-
tal exchanges proceed, beginning at the initial time t 0 when temperature was T 0 . If, for
instance, the energy of activation of the diffusion process is 50 kcal/mole, the cooling rate
Silicates / 383
of the system is 2 °C/106 years, and T 0 ⫽ 1000 K, the resulting ␥ is (0.05 ⫻ 10⫺6 ) years⫺1.
The time elapsed from the initial condition (T 0 , t 0 ) to the (kinetic) closure of elemental
exchanges is
t′ =
1
γ
[
1 − exp( −γ t) ≤
1
γ
]
≤ 2 × 10 7 years (5.229)
(Lasaga, 1983).
As regards the effect of temperature on the distribution constant, if both phases are
ideal, the distribution constant is equivalent to the thermodynamic constant, and we can
write
 ∆H 0  1 1 
K D = K D 0 exp   − 0 , ( 5.230 )
 R T T 
where ⌬H 0 is the enthalpy of the exchange reaction at standard state and KD0 is the distri-
bution constant at the initial temperature T 0 . We can define KD as a function of time, as
was done previously for diffusivity (cf. equation 5.227):
KD = K D 0 exp( − ε ′t ) , ( 5.231)
(t)
where
∆H 0 v
ε′ = 2
. ( 5.232 )
RT 0
( t0 )
Lasaga (1979) has shown that the molar concentration of an exchanging element A
at the interface between two phases ␣ and ␤ at time t (XA,␣,0,t ) obeys the law
X A, α , 0, t = X A0 , α exp( − ε t) . ( 5.233 )
Applying Fick’s law on diffusive fluxes at the interface between the two phases (cf. sec-
tion 4.11):
∂X A, α ∂X A, β
− D˜ α , t = − D˜ β , t ( 5.234 )
∂χ ∂χ
and
∂X B , α ∂X B , β
− D˜ α , t = − D˜ β , t , ( 5.235)
∂χ ∂χ
where ␹ is the distance from the interface, and D̃␣,t and D̃␤,t are interdiffusion coefficients
for elements A and B in phases ␣ and ␤, Lasaga (1983) has shown that ␧ is related to
␧⬘ through
D˜ β0 D˜ α0
ε = ε′ ,
 X A0 , α   X A0 , α X A0 , α  (5.236)
D˜ β0 D˜ α0  1 +  X 0 + X 0 
 X 0
B, α   A, β B, β 
where X 0A,␣ is the concentration of A in phase ␣, initially homogeneous at temperature T 0 .

If D̃0␤ ⬎⬎ D̃0␣ and X 0B,␣ ⬎ X A,
0
␣, then
ε ≈ ε′ ( 5.237 )
and
ε ∆H 0
≈ . ( 5.238 )
γ Eα
If D̃0␤ ⬍ D̃0␣ and X 0B,␣ ⬎ X A,

0
␣, then
ε = ε ′ D˜ β0 D˜ α0 ( 5.239 )
and
ε
≈ ∆ H 0. ( 5.240 )
γ
Analysis of existing data shows that ratio ␧ / ␥ in the main thermometric phases ranges
between 0.01 and 0.2. The highest ␧ / ␥ values correspond to minerals with high ionic
diffusivity. Knowledge of ratio ␧ / ␥ in phases used in thermometric studies allows us to
establish the reliability of the obtained temperature, as a function of the grain size of the
phases and the cooling rate of the system. Besides ratio ␧ / ␥, Lasaga (1983) also defined
quantity ␥⬘ as
E α va 2
γ′= , ( 5.241)
D˜ α0 RT (0t )
2
where D0␣ is the interdiffusion coefficient at temperature T 0 in phase ␣ and a is the length
of the crystal orthogonal to the interface. ␥⬘ allows us to establish whether or not the
nuclei of crystals undergoing equilibration maintain the original compositions they had at
crystallization (t 0 , T 0 ). Practical calculations show that this happens for ␥⬘ ⬎ 10. Figure
5.67 shows the combined effects of parameter ␥⬘ and ratio ␧ / ␥: for low values of ␥⬘, the
Silicates / 385
Figure 5.67 Concentration profiles in a crystal, for various values of ␥⬘ and ␧/␥. Abscissa:
fractional distance from crystal edge (␹ ⫽ 0 at edge; ␹ ⫽ 1 at center). Reproduced with
modifications from Lasaga (1983), with permission of Springer-Verlag, New York Inc.
initial composition of the nucleus is not preserved even in phases with slow diffusivity
(low ␧ / ␥ ). We also see that, at constant ␥⬘, the concentration profiles are more or less
altered with respect to the initial condition, depending on diffusion rate. Lasaga (1983)
emphasized the fact that the core of the crystal may be homogeneous at both high (i.e.,
⬎ 100) and low (⬍ 0.01) values of parameter ␥⬘. Hence, contrary to what is commonly
assumed, the homogeneity of crystal cores does not necessarily mean preservation of the
initial composition acquired at crystallization.
Table 5.73 list iso-␥⬘ temperatures preserved by some minerals at various

cooling rates of the system (Lasaga, 1983). For each phase, T 0 is the initial tem-
perature of homogeneous composition preserved at cooling rate v for ␥⬘ ⫽ 10
(crystal size ⫽ 1 mm). Consider, for instance, pyroxene: for low cooling rates (i.e.,
⬍ 10 °C/106 y), temperatures in the range 900 to 1000 °C may be effectively
considered as quenched equilibrium temperatures (i.e., either crystallization or
recrystallization). However, preservation of higher T (for instance, 1100 to 1300
°C) requires very high cooling rates, to prevent alteration of concentration pro-
files down to the crystal nuclei. The data in table 5.73 may thus be used to evaluate
the reliability of a thermometric calculation, depending on the kinetic properties
of the phase.
Table 5.73 Iso-␥ ⬘ temperatures for some minerals as a function of

cooling rate of system. ␥ ⫽ 10; a ⫽ 1 mm (from Lasaga, 1983).
(a) Based on diffusion data of Lasaga et al. (1977). (b) Based on
diffusion data of Elphick et al. (1981).
T (°C)
Garnet
6
v (°C/10 y) Pyroxene (a) (b) Olivine
0.01 805 565 785 -
0.1 865 610 840 -
1 935 665 907 315
10.0 1010 720 980 370
100.0 1100 790 1050 430
1000.0 1200 860 1155 500
10000.0 1320 945 1260 585
5.9.2 Intracrystalline Exchange Geothermometry

Let us consider a solid mixture (A,B)N with two lattice positions M1 and M2 in
which intracrystalline exchanges take place:
A M1 + BM2 ⇔ A M2 + BM1 . ( 5.242 )
This is the case, for instance, for the enstatite-ferrosilite solid mixture [orthopy-
roxene: (Fe,Mg)SiO3 ]:
Mg M1 + Fe 2M+2 ⇔ Mg M 2 + Fe 2M+1 . (5.243)
If mixing on sites may be approximated by a regular model, we have
ln K 243 = ln K D +
WM1
RT
( )
W
(
1 − 2 X Fe,M1 − M 2 1 − 2 X Fe,M 2 .
RT
) (5.244 )
It is important to emphasize here that, theoretically, if a solid mixture is ideal,

intracrystalline distribution is completely random (cf. section 3.8.1) and, in these
conditions, the intracrystalline distribution constant is always 1 and coincides
with the equilibrium constant. If the mixture is nonideal, we may observe some
ordering on sites, but intracrystalline distribution may still be described without
site interaction parameters. We have seen in section 5.5.4, for instance, that the
distribution of Fe2⫹ and Mg2⫹ on M1 and M3 sites of riebeckite-glaucophane
amphiboles may be approached by an ideal site mixing model—i.e.,
Mg M1 + Fe 2M+3 ⇔ Mg M 3 + Fe 2M+1 (5.245)

Silicates / 387
K 245 =
aMg,M3 aFe2 +,M1
=
(1 − X Fe,M 3 )X Fe,M1
= 0.403. (5.246 )
aMg,M1aFe2 +,M 3 (1 − X Fe,M1 )XFe,M 3
In the case of reaction 5.243, Saxena and Ghose (1971) experimentally deter-
mined the distribution constant over a wide range of temperatures. According to
these authors, it has an exponential dependency on T (K) of the type
 10 3 
− RT ln K 243 = ∆G 243
0
= 4479 − 1948
 T 
 (cal mole ) . (5.247)
In the absence of site interaction energy terms, the equality
K 243 = K D (5.248)
holds. We would therefore observe simple hyperbolic distribution curves similar

to those for amphiboles in figure 5.39B. The presence of site interaction energy
terms causes perturbations of the hyperbolic distribution curves, whose fluctua-
tions become more marked as the values of W site parameters become higher.
According to Saxena (1973), the observed distribution of Fe2⫹ and Mg2⫹ in M1
and M2 sites of orthopyroxene requires site interaction parameters that vary with
T (K) according to
 10 3 
WM1 = 3525 
 T 
 − 1667 (cal mole ) (5.249)
and
 10 3 
WM2 = 2458 
 T 
 − 1261 ( cal mole ) . ( 5.250 )
The form of the distribution curves, based on equation 5.244 and application of
conditions 5.247, 5.249, and 5.250, is displayed in figure 5.68.
More recently, Ganguly and Saxena (1987) reconsidered equation 5.243 in
the light of all existing experimental data, proposing
 1561.81 WM 1
ln K D =  0.1435 −
 T 
+
RT
1 − 2 X Fe,M 1 ( )
(5.251)
W
(
− M 2 1 − 2 X Fe,M 2 ,
RT
)
Figure 5.68 Fe2+-Mg2+ intracrystalline distribution between M1 and M2 sites in natural

orthopyroxenes of composition near a binary mixture (Fe,Mg)SiO3. From Saxena and
Ghose (1971). Reprinted with permission of The Mineralogical Society of America.
with WM1 ⫽ 1524 (cal/mole) and WM2 ⫽ ⫺1080 (cal/mole). Note that the first
term in parentheses in equation 5.251 corresponds to ln K243 in equation 5.244.
The diagrammatic form of figure 5.68 is that commonly adopted to display
intracrystalline distributions (see also figures 5.39 and 5.40). However, this sort
of plot has the disadvantage of losing definition as the compositional limits of
the system are approached. A different representation of intracrystalline disorder
is that seen for olivines (figures 5.10 and 5.12; section 5.2.5): the distribution
constant is plotted against the molar fraction of one of the components in the
mixture.
The effects of kinetics on intracrystalline disorder are less marked than those
affecting intercrystalline exchanges, essentially because intracrystalline exchanges
take place on the Å scale, whereas intercrystalline equilibration requires displace-
ments on the mm scale. As a result, temperatures registered by intracrystalline
exchange thermometers are generally lower than those derived from intercrystal-
line equilibria. Lastly, it must be noted that the effects of pressure on intracrystal-
line partitioning are not negligible, as generally assumed, because the volume
properties of phases are definitely affected by the state of internal disorder.
5.9.3 Solvus Thermobarometry

Solvus thermobarometry uses the unmixing phenomena of crystalline com-
pounds, retrieving the T and P conditions of unmixing from the compositions
Silicates / 389
(and structures) of unmixed phases. The various solvi are located experimentally
by physical observation of the mixing-unmixing phenomenon induced on a fixed
composition by modifying the T and P. For the sake of accuracy, the experiment
is usually “reversed,” the miscibility gap limb being approached from opposite
directions in the T-P space.
The thermodynamic significance of unmixing was stressed in chapter 3 (sec-
tions 3.11, 3.12, and 3.13), together with the intrinsic differences among solvus
field, spinodal field, and miscibility gap, and the role of elastic strain on the T-X
extension of the coherent solvus and coherent spinodal field with respect to chem-
ical solvus and chemical spinodal fields, respectively. Although the significance
of the various compositional fields is clear in a thermodynamic sense, very often
the stability relations of unmixed phases do not seem as clear in microscopic
investigation of natural assemblages. The confusion in terms used to describe
unmixing phenomena in natural phases was discussed in some detail in section
5.7.4.
Feldspars
Feldspars, as we have already seen, are mixtures of albite (NaAlSi3O8 ),
K-feldspar (KAlSi3O8 ), and anorthite (CaAl2Si2O8 ) components (see section
5.7). Albite and anorthite are miscible to all extents (plagioclases), as are albite
and K-feldspar (alkali feldspars), but the reciprocal miscibility between the two
series is limited for T ⬍ 800 °C. Stormer (1975) calculated the distribution of the
NaAlSi3O8 component between plagioclases and alkali feldspars, accounting for
the nonideality of the NaAlSi3O8-KAlSi3O8 solid mixture (Parson, 1978) and for
P effects on the solvus field. Parson’s (1978) model reproduces quite well the ex-
perimental observations of Seck (1971) at T ⫽ 650 °C for anorthite amounts
between 5% and 45% in moles. Modifications of Stormer’s (1975) model were
suggested by Powell and Powell (1977a). Whitney and Stormer (1977) also pro-
posed a low-T geothermometer based on K-feldspar amounts in mixtures. The
isothermal distributions of Stormer’s (1975) model at various P conditions are
compared with the experimental evidence of Seck (1972) in figure 5.69. When
applying this sort of thermometer, we must bear clearly in mind the significance
of feldspar unmixes in the various assemblages, particularly with respect to the
degree of coherence of unmixed zones. Stormer’s geothermometer defines the loci
of chemical solvi (i.e., absence of strain energy) and so cannot be applied to coher-
ent exsolution or to spinodal decomposition. It is also worth noting that the alkali
feldspar series unmixes itself partially during cooling, so that the compositions of
unmixed grains must be reintegrated in order to evaluate the original composition
of the solid mixture on the solvus limb. Application of Stormer’s (1975) geother-
mometer may simply be graphical [plotting the molar amounts of NaAlSi3O8 in
the two series, for a selected value of P (figure 5.69) and comparing the results
with the isotherms] or analytical, through the appropriate thermometric
equation.
Figure 5.69 Feldspar geothermometry. From Stormer (1975). Reprinted with permission
of The Mineralogical Society of America.
Stormer’s (1975) model is based on the assumption that the potassic end-member has
no influence on the mixing behavior of the plagioclase series and that the calcic compo-
nent does not affect the mixing behavior of the K-feldspar series to any extent. With
these assumptions, the problem of equilibrium between two ternary feldspars, normally
represented by the equalities
µ NaAlSi 3O 8 , Pl = µ NaAlSi 3O 8 , Kf ( 5.252 )
µ KAlSi 3O 8 , Pl = µ KAlSi 3O 8 , Kf ( 5.253 )
µ CaAl 2 Si 2O 8 , Pl = µ CaAl 2 Si 2O 8 , Kf , ( 5.254 )
where Pl ⫽ plagioclases and Kf ⫽ alkali feldspars, is reduced to evaluation of the simple

equilibrium of the albitic component (eq. 5.252). The chemical potential of the pure com-
ponent in the mixture refers to the standard state potential through
Silicates / 391
µ NaAlSi = µ NaAlSi
0
+ RT ln a NaAlSi 3O 8, Pl ( 5.255)
3O 8, Pl 3O 8, Pl
and
µ NaAlSi = µ NaAlSi
0
+ RT ln a NaAlSi 3O 8 , Kf . ( 5.256 )
3O 8 , Kf 3O 8 , Kf
If we adopt for both phases the standard state of the pure component at the T and P of
interest (and if the two phases are strictly isostructural), because
µ NaAlSi
0
= µ NaAlSi
0
, ( 5.257 )
3O 8 , Pl 3O 8 , Kf
from equations 5.255 and 5.256, and by application of equation 5.252, we obtain
a NaAlSi 3O 8 , Pl = a NaAlSi 3O 8 , Kf . ( 5.258 )
Note that the isoactivity condition of equation 5.258 is valid only for all the loci of the
solvus limb (see, for instance, figure 3.10) but not for a spinodal limit or for a miscibility
gap limb, because in these cases the structural state of unmixes is different, so that equa-
tion 5.257 does not hold.
If we call the distribution constant KD , as a ⫽ X ⭈ ␥, from equation 5.258 we derive
X NaAlSi 3O 8 , Kf γ NaAlSi 3O 8 , Pl
KD = = . ( 5.259 )
X NaAlSi 3O 8 , Pl γ NaAlSi 3O 8 , Kf
Moreover, if the plagioclase series is ideal (which is the case in the compositional range
Ab100 to Ab45 ; cf. Orville, 1972), equation 5.259 may be further reduced to
1
KD = . ( 5.260 )
γ NaAlSi 3O 8 , Kf
Adopting a subregular Margules model for the NaAlSi3O8-KAlSi3O8 (Ab-Or) binary mix-
ture and assuming that the activity coefficient of the albite component is not affected by
the presence of limited amounts of the third component in the mixture (i.e., CaAl2Si2O8 ),
equation 5.260 may be transformed into
ln K D = − ln γ Ab, Kf
(5.261)
1
(1 − X ) [W ( )] .
2
=− Ab, Kf Ab − Or + 2X Ab, Kf WOr − Ab − WAb − Or
RT
Using the interaction parameters of Thompson and Waldbaum (1969) (cf. second column
of parameters in table 5.66), i.e.,
WAb− Or = 6326.7 + 0.0925P − 4.6321T ( 5.262 )

WOr− Ab = 7671.8 + 0.1121P − 3.8565T , ( 5.263 )
and combining equations 5.262 and 5.263 with equation 5.260, we obtain the thermomet-
ric equation of Stormer (1975):
6326.7 − 9963.2 X Ab , Kf + 943.3 X Ab2 , Kf + 2690.2 X Ab3 , Kf

T (K) =
 X Ab , Kf  
 −1.9872 ln   + 4.6321 − 10.815 X Ab , Kf + 7.7345 X Ab2 , Kf − 1.5512 X Ab3 , Kf 
 X Ab , Pl
 
(5.264)
(0.0925 − 0.1458X Ab , Kf
+ 0.0141X
2
Ab , Kf
+ 0.0392 X
3
Ab , Kf
)P
+ .
 X Ab , Kf  
 −1.9872 ln   + 4.6321 − 10.815 X Ab , Kf + 7.7345 X Ab2 , Kf − 1.5512 X Ab3 , Kf 
 X Ab , Pl
 
Pyroxenes
As we have already seen in section 5.4.6, unmixing between monoclinic and or-
thorhombic pyroxenes is conveniently interpreted in the framework of the quadri-
lateral field diopside-hedembergite-enstatite-ferrosilite (CaMgSi2O6-CaFeSi2O6-
Mg2Si2O6-Fe2Si2O6 ). Indeed, the compositional field is ternary (CaSiO3-MgSiO3-
FeSiO3 ), but the molar amount of wollastonite (CaSiO3 ) never exceeds 50% in
the mixture. Initial studies on the clinopyroxene-orthopyroxene miscibility gap
considered only the magnesian end-members (enstatite and diopside; see, for in-
stance, Boyd and Schairer, 1964). Extension to the quadrilateral field was later
attempted by Ross and Huebner (1975). Figure 5.70 shows the pyroxene quadri-
lateral at 5, 10, and 15 kbar total pressure according to Lindsley (1983).
As figure 5.70 shows, with increasing T the miscibility gap shrinks toward
the iron-free join; moreover, on the Ca-poor side of the quadrilateral, miscibility
relations are complicated by the formation of pigeonite over a restricted P-T-X
range (cf. section 5.4.6). Once the compositions of coexisting clinopyroxene and
orthopyroxene are known, figure 5.70 is a practical geothermometer.
To calculate the molar proportions of components in the unmixed phase, we proceed

as follows:
Clinopyroxene
If the analysis reports the amount of ferric iron, this value is accepted and aluminum is
partitioned between VI-fold and IV-fold coordinated sites, according to
Al VI + Fe VI
3+
+ CrVI + 2 Ti VI = Al IV + Na M 2 ( 5.265)
and
Al VI + Al IV = Al total. ( 5.266 )
Silicates / 393
Figure 5.70 Lindsley’s (1983) pyroxene geothermometer.
If the analysis reports only total iron expressed as FeO (for example, microprobe data),
the amount of IV-fold coordinated aluminum is assumed to obey
Al IV = 2 − Si . ( 5.267 )
3⫹
FeVI is then defined by equations 5.265 and 5.266.
We then evaluate the molar amounts of the various clinopyroxene components as

follows:
(1) Acmite ⫽ NaFe3⫹Si2O6 ⫽ smaller amount between Na and Fe3⫹

(2) Jadeite ⫽ NaAlSi2O6 ⫽ smaller amount between AlIV and remaining Na
(3) Fe-Ca-Tschermak ⫽ CaFeAlSiO6 ⫽ remaining Fe3⫹
(4) Cr-Ca-Tschermak ⫽ CaCrAlSiO6 ⫽ Cr
(5) Al-Ca-Tschermak ⫽ CaAl2SiO6 ⫽ remaining AlVI
(6) If the pyroxene is an augite, then <<Ca>> ⫽ Ca; if it is a pigeonite, then
<<Ca>> ⫽ 2 ⫺ (Fe2⫹ ⫹ Mg)
(7) X ⫽ Fe2⫹ /(Fe2⫹ ⫹ Mg)
(8) Wollastonite ⫽ Ca2Si2O6 ⫽ (<<Ca>> ⫹ acmite ⫺ Fe-Ca-Tschermak ⫺ Cr-Ca-
Tschermak ⫺ Al-Ca-Tschermak)/2
(9) Clinoenstatite ⫽ Mg2Si2O6 ⫽ (1 ⫺ wollastonite) (1 ⫺ X)
(10) Clinoferrosilite ⫽ Fe2Si2O6 ⫽ (1 ⫺ wollastonite)X.
The resulting molar amounts are already normalized to 1 and may be plotted in figure
5.70 for thermometric deductions.
Orthopyroxene
(1) AlIV ⫽ 2 ⫺ Si
(2) AlVI ⫽ Altotal ⫺ AlIV
(3) Fe3⫹ calculated by applying equation 5.265
(4) R3⫹ ⫽ AlVI ⫹ Cr ⫹ Fe3⫹
(5) R2⫹ ⫽ (1 ⫺ X)Mg ⫹ X Fe2⫹
(6) NaR3⫹Si2O6 ⫽ smaller amount between Na and R3⫹
(7) NaTiAlSiO6 ⫽ smaller amount between Ti and AlIV or remaining Na
(8) R2⫹TiAl2O6 ⫽ smaller amount between Ti and AlIV /2
(9) R2⫹R3⫹AlSiO6 ⫽ smaller amount between remaining R3⫹ and AlIV
(10) Remaining Ca, Fe2⫹, and Mg normalized to form wollastonite ⫹ enstatite ⫹
ferrosilite.
On the basis of the experimental data of Lindsley and Dixon (1976) on the
CaMgSi2O6-Mg2Si2O6 binary join and accounting for the observed compositions
of natural pyroxenes coexisting at equilibrium, Kretz (1982) proposed two empiri-
cal thermometric equations applicable to clinopyroxene and valid, respectively,
for T ⬎ 1080 °C and T ⬍ 1080 °C:
( )
T K =
1000
( 5.268 )
(
0.468 + 0.246 X Fe,cpx − 0.123 ln 1 − 2 X Ca,cpx )
and
Silicates / 395
( )
T K =
1000
, ( 5.269 )
(
0.054 + 0.608 X Fe,cpx − 0.304 ln 1 − 2 X Ca,cpx )
where XCa,cpx ⫽ Ca/(Ca ⫹ Mg) and XFe,cpx ⫽ Fe/(Fe ⫹ Mg) in clinopyroxene.
An analogous formulation, also based on the experiments of Lindsley and
Dixon (1976), came from Wells (1977):
( )
T K =
7341
 aMg 2 Si2 O6 , cpx 
,
3.355 + 2.44 X Fe,opx − ln  (5.270)

 aMg 2 Si2 O6 ,opx 
where XFe,opx ⫽ Fe/(Fe ⫹ Mg) in orthopyroxene.

In Wells’ (1977) thermometer, the activity of the magnesian end-member in
the two unmixed phases is attributable to ideal mixing of Mg and Fe on M1 and
M2 sites, i.e.,
aMg 2 Si2 O6 , cpx = X Mg,M 1 ⋅ X Mg,M 2 , ( 5.271)
after attribution of all Ca to site M2 and all Al to site M1 (cf. Wood and Banno,
1973). This choice must be considered as purely operational, because it is incon-
sistent with the real nature of the mixture. It must also be emphasized that both
the formulations of Kretz (1982) and Wells (1977) are based on equilibrium distri-
bution of the magnesian component between the unmixed terms, although
Kretz’s (1982) formulation is explicitly derived for one of the two polymorphs. In
both formulations, the effect of adding iron to the system (i.e., moving from the
CaMg-Mg join into the quadrilateral) is regarded as simply linear (cf. eq. 5.268,
5.269, and 5.270), owing to the fact that, in natural clinopyroxenes, there is a
linear relationship between XFe,cpx and ln (1 ⫺ 2XCa,cpx ) (see Kretz, 1982).
5.9.4 Solid-Solid Reactions

The coexistence of various solids at equilibrium in a heterogeneous system may
be opportunely described by chemical reactions involving the major components
of the solid phases. If we observe, for instance, the coexistence of pyroxene, pla-
gioclase, and quartz, we can write an equation involving sodic terms:
NaAlSi 3 O 8 ⇔ NaAlSi 2 O 6 + SiO 2

albite jadeite quartz (5.272)
or an equation involving calcic terms:

CaAl 2 Si 2 O 8 ⇔ CaAl 2 SiO 6 + SiO 2 .

anorthite Ca- Tschermak quartz (5.273)
If we observe the coexistence of two polymorphs of the same compound (e.g.,

Al2SiO5 ), we can also write
Al 2 SiO 5 ⇔ Al 2 SiO 5 .
andalusite kyanite (5.274)
Coexistence of phases on opposite sides of the reaction equation implies that

they are in equilibrium and that the ⌬G of reaction is zero—i.e.,
∆G P0,T = − RT ln K . (5.275)
If the heat capacity functions of the various terms in the reaction are known and
their molar enthalpy, molar entropy, and molar volume at the Tr and Pr of refer-
ence (and their isobaric thermal expansion and isothermal compressibility) are
also all known, it is possible to calculate ⌬G 0P,T at the various T and P conditions
of interest, applying to each term in the reaction the procedures outlined in sec-
tion 2.10, and thus defining the equilibrium constant (and hence the activity prod-
uct of terms in reactions; cf. eq. 5.272 and 5.273) or the locus of the P-T points
of univariant equilibrium (eq. 5.274). If the thermodynamic data are fragmentary
or incomplete—as, for instance, when thermal expansion and compressibility
data are missing (which is often the case)—we may assume, as a first approxima-
tion, that the molar volume of the reaction is independent of the P and T intensive
variables. Adopting as standard state for all terms the state of pure component
at the P and T of interest and applying
( )
∆G P0,T = ∆H P0,T − T∆S P0,T + P − Pr ∆VP0,T = − RT ln K , (5.276)
we may also identify the P-T locus of points where the equilibrium constant takes
on the identical value. It is obvious from equation 5.276 that equilibrium among
solids involving a conspicuous volume of reaction ( ⌬V 0 ) will be a good baromet-
ric function, and that one involving consistent ⌬H 0 (and limited ⌬V 0 ) will be a
good thermometric function.
Table 5.74 lists the main solid-solid reactions commonly adopted in geochem-
istry, with their respective ⌬ H 0Pr ,Tr, ⌬ S P0 r ,Tr, ⌬ V 0Pr ,Tr, and ⌬ G 0Pr ,Tr, calculated with
the INSP and THERMO computer packages (Saxena, 1989).
Some peculiar aspects of the various reactions are briefly summarized below.
The discussion is taken from Essene (1982), to whom we refer for more exhaus-
tive treatment.
Silicates / 397
Table 5.74 Main solid-solid reaction adopted in geothermobarometry. Values of ⌬ H 0P r , T r (kJ/

mole), ⌬S 0Pr ,Tr [ J/(mole ⫻ K)], ⌬V 0P r , T r (cm3/mole), and ⌬ G 0P r , Tr (kJ/mole) calculated with INSP
and THERMO computer packages (Saxena, 1989).
Reaction
Paragenesis Reaction Number
Andalusite-kyanite Al2SiO5 ⇔ Al2SiO5 1

Kyanite-sillimanite Al2SiO5 ⇔ Al2SiO5 2
Andalusite-sillimanite Al2SiO5 ⇔ Al2SiO5 3
Orthopyroxene- Fe2Si2O6 ⇔ Fe2SiO4 ⫹ SiO2 4
olivine-quartz
Orthopyroxene- Mg2Si2O6 ⇔ Mg2SiO4 ⫹ SiO2 5
olivine-quartz
Clinopyroxene-olivine CaMgSi2O6 ⫹ Mg2SiO4 ⇔ CaMgSiO4 ⫹ Mg2Si2O6 6
Clinopyroxene- NaAlSi3O8 ⇔ NaAlSi2O6 ⫹ SiO2 7
plagioclase-quartz
Clinopyroxene- CaAl2Si2O8 ⇔ CaAl2SiO6 ⫹ SiO2 8
plagioclase-quartz
Garnet-plagioclase- 2Al2SiO5 ⫹ Ca3Al2Si3O12 ⫹ SiO2 ⇔ 3CaAl2Si2O8 9
sillimanite-quartz
1 2
Garnet-plagioclase- CaAl2Si2O8 ⫹ Fe2SiO4 ⇔ Ca3Al2Si3O12⫹ Fe3Al2Si3O12 10
3 3
olivine
1 2
Garnet-plagioclase- CaAl2Si2O8 ⫹ Mg2SiO4 ⇔ Ca3Al2Si3O12 ⫹ Mg3Al2Si3O12 11
3 3
olivine
1 2
Garnet-plagioclase- CaAl2Si2O8 ⫹ Fe2Si2O6 ⇔ Ca3Al2Si3O12 ⫹ Fe3Al2Si3O12 ⫹ SiO2 12
3 3
orthopyroxene-quartz
1 2
Garnet-plagioclase- CaAl2Si2O8 ⫹ Mg2Si2O6 ⇔ Ca3Al2Si3O12 ⫹ Mg3Al2Si3O12 ⫹ SiO2 13
3 3
orthopyroxene-quartz
Garnet-Fe cordierite- 3Fe2Al4Si5O18 ⇔ 2Fe3Al2Si3O12 ⫹ 4Al2SiO5 ⫹5SiO2 14
sillimanite-quartz
Garnet-Mg cordierite- 3Mg2Al4Si5O18 ⇔ 2Mg3Al2Si3O12 ⫹ 4Al2SiO5 ⫹ 5SiO2 15
sillimanite-quartz
Garnet-spinel- Fe3Al2Si3O12 ⫹ 2Al2SiO5 ⇔ 3FeAl2O4 ⫹ 5SiO2 16
sillimanite-quartz
Garnet-spinel- Mg3Al2Si3O12 ⫹ 2Al2SiO5 ⇔ 3MgAl2O4 ⫹ 5SiO2 17
sillimanite-quartz
Garnet-rutile-ilmenite- 3FeTiO3 ⫹ Al2SiO5 ⫹ 2SiO2 ⇔ Fe3Al2Si3O12 ⫹ 3TiO2 18
sillimanite-quartz
Reaction
Number ⌬H 0Pr ,Tr ⌬S 0Pr ,Tr ⌬V 0Pr ,Tr ⌬G 0Pr ,Tr
1 ⫺4.1 ⫺9.1 ⫺7.4 ⫺1.4
2 8.2 13.5 5.7 4.1
3 4.1 4.4 ⫺1.7 2.7
4 1.6 3.4 3.0 ⫺2.9
5 7.5 3.1 3.7 6.5
6 41.56 7.69 4.61 39.27
7 ⫺11.0 ⫺44.1 ⫺17.3 2.1
8 19.3 ⫺18.8 ⫺13.9 24.9
continued
Reaction
Number ⌬H 0Pr ,Tr ⌬S 0Pr ,Tr ⌬V 0Pr ,Tr ⌬G 0Pr ,Tr
9 12.5 105.7 52.8 ⫺18.8

10 ⫺14.3 ⫺39.7 ⫺27.9 0.9
11 5.2 ⫺29.5 ⫺26.6 14.0
12 ⫺12.7 ⫺36.4 ⫺24.8 ⫺1.9
13 12.7 ⫺26.4 22.9 20.5
14 ⫺116.8 ⫺150.2 ⫺159.8 ⫺72.3
15 6.8 ⫺109.9 ⫺160.2 39.1
16 14.6 ⫺1.7 20.7 15.0
17 ⫺7.9 ⫺6.1 19.6 ⫺6.2
18 ⫺0.5 ⫺17.7 ⫺18.6 5.1
Reactions 1, 2, and 3. Al2SiO5 polymorphs are widely adopted as index minerals in meta-
morphic terranes and are easily recognizable even on the macroscopic scale. Their P-T
stability fields have long been subjected to some uncertainty because of disagreement
among authors on the exact location of the ternary invariant point (P ⫽ 6.5 kbar and
T ⫽ 595 °C according to Althaus, 1967; 5.5 kbar and 620 °C according to Richardson et
al., 1969; 3.8 kbar and 600 °C according to Holdaway, 1971). Essene (1982) reports the
substantial preference of petrologists for Holdaway’s (1971) triple point, which better con-
forms to thermobarometric indications arising from other phases. As already outlined in
section 2.7, the recent study of Holdaway and Mukhopadhyay (1993) confirms that the
triple point location of Holdaway (1971) is the most accurate.
Reactions 4 and 5. Bohlen and Boettcher (1981) studied the effect of the presence of
magnesian and manganoan terms in the mixture on the distribution of ferrous component
between orthopyroxene and olivine according to the reaction
Fe 2 Si 2 O6 ⇔ Fe 2 SiO 4 + SiO2 .
opx olivine quartz (α or β ) (5.277)
Figure 5.71A shows univariant equilibrium curves for various molar amounts of ferrous
component in the orthopyroxene mixture. The P-T field is split into two domains, corre-
sponding to the structural state of the coexisting quartz ( ␣ and ␤ polymorphs, respec-
tively). If the temperature is known, the composition of phases furnishes a precise estimate
of the P of equilibrium for this paragenesis. Equation 5.277 is calibrated only for the
most ferriferous terms, and the geobarometer is applicable only to Fe-rich rocks such as
charnockites and fayalite-bearing granitoids.
Reaction 6. This reaction was investigated in detail by Finnerty and Boyd (1978) and
was more recently recalibrated by Köhler and Brey (1990). Because the Ca content in
olivine is P-dependent, due essentially to a large solvus between monticellite and forsterite
that expands with pressure (cf. section 5.2.5), this reaction should act as a sensitive baro-
metric function. However, the enthalpy of reaction is quite high and the effect of T on the
equilibrium is also marked. The calibrated P-T slope has an inflection, the origin of which
is not clear at first glance.
Silicates / 399
Figure 5.71 Thermobarometric equilibria plotted in P-T field. Numbers on univariant

curves in parts B, C, and D are base-10 logarithms of respective equilibrium constants
(log K ⫽ 0 for pure components). From Bohlen et al. (1983). Reprinted with permission
of The Mineralogical Society of America.
Reactions 7 and 8. Thermobarometric estimates are complicated by order-disorder tran-

sition in albite and by the difficulty of evaluating the activity of jadeite and Ca-Tschermak
in natural clinopyroxenes (cf. Essene, 1982).
Reaction 9. This reaction is the most popular geobarometer, because of the ubiquity of
garnet-Al2SiO5-plagioclase-quartz paragenesis in medium- and high-grade metamorphic
terranes. It may be affected by considerable error progression if the specimens are low in
Ca (i.e., small amounts of anorthite in plagioclase and of grossular in garnet). Moreover,
the slopes of the univariant curves in the P-T space are quite marked, and the T of equilib-
rium must be known with good approximation to avoid large errors in the derived P value.
For practical applications, see Ghent (1976), Newton and Haselton (1981), and Perkins
(1983).
Reactions 10 and 11. The equilibrium
(Ca 0.333 )
Fe 0.666 Al2Si 3O12 ⇔ CaAl2Si2O8 + Fe2SiO4
3
( 5.278)
was calibrated by Green and Hibberson (1970). The reaction is clearer if we split the
garnet compound into the pure terms grossular and almandine (figure 5.71B):
Ca3 Al2Si3O12 + 2 Fe 3Al2Si 3O12 ⇔ 3 CaAl2Si2O8 + 3 Fe2SiO4 .

grossular almandine anorthite fayalite
(5.279)
In this case the equilibrium constant is
3
a CaAl ⋅ a Fe
3
2 Si 2O 8 2 SiO 4
K 279 = . (5.280)
a Ca 3 Al 2 Si 3O12 ⋅ a 2
Fe 3 Al 2 Si 3O 12
The analogous reaction involving magnesian components is
Ca3 Al2Si3O12 + 2 Mg3Al2Si3O12 ⇔ 3 CaAl2Si2O8 + 3 Mg2SiO4

grossular pyrope anorthite forsterite
(5.281)
and is particularly applicable to garnet-bearing metagabbros. Consistent information may

be obtained by contemporaneous application of equations 5.279 and 5.281. For all terms
in the reaction, thermodynamic activity must be evaluated using appropriate mixing mod-
els. Comparison of thermobarometric information obtained with different compositional
terms (e.g., eq. 5.279 and 5.281) indirectly indicates the appropriateness of the mixing
models used. Incidentally, it must be noted here that interaction energies in the solid mix-
tures of interest are generally positive, so that activity coefficients are higher than 1 for all
terms. Thus, eventual errors in the evaluation of activity coefficients may be mutually
compensated in the evaluation of the equilibrium constant (eq. 5.280).
Reactions 12 and 13. The coexistence of plagioclase, orthopyroxene, garnet, and quartz
may be expressed as an equilibrium involving ferrous and/or magnesian components, as
seen in the preceding case:
Ca3 Al2Si3O12 + 2 Fe3 Al2Si3O12 + 3 SiO2 ⇔ 3 CaAl2Si2O8 + 3 Fe2Si2O6

grossular almandine quartz anorthite ferrosilite
(5.282)
Ca3 Al2Si3O12 + 2Mg3 Al2Si3O12 + 3 SiO2 ⇔ 3 CaAl2Si2O8 + 3Mg2Si2O6

grossular pyrope quartz anorthite enstatite (5.283)
Reaction 5.282 was calibrated by Bohlen et al. (1982) and is particularly useful in granu-
lites and garnet-bearing amphibolites. Figure 5.71C shows the loci of P-T points for which
equilibrium constant K282 assumes identical values (base-10 logarithm). The univariant
Silicates / 401
curve for pure end members (log K ⫽ 0) is in boldface. The corresponding reaction involv-
ing magnesian terms was calibrated by Newton and Perkins (1982).
Reactions 14 and 15. There are still several problems regarding the calibration of these
equilibria, which are discussed in detail by Essene (1982). Based on the energy terms listed
in table 5.74, both T and P markedly affect reaction 14, whereas reaction 15 furnishes
precise barometric indications.
Reactions 16 and 17. The reaction
Fe3 Al2Si3O12 + 2 Al2SiO5 ⇔ 3 FeAl2O4 + 5 SiO2

almandine sillimanite hercynite quartz
(5.284)
is a potentially useful thermobarometer in high-grade metamorphic terranes, where spinel,

when present, is a mixture of the major components gahnite and hercynite. Experimental
data concerning this reaction were produced by Richardson (1968), Holdaway and Lee
(1977), and Wall and England (1979). No quantitative applications have hitherto been
afforded.
Reaction 18. The reaction
3 FeTiO3 + Al2SiO5 + 2 SiO2 ⇔ Fe3 Al2Si3O12 + 3 TiO2

ilmenite sillimanite quartz almandine rutile
(5.285)
was calibrated by Bohlen et al. (1983). The loci of identical value of equilibrium constant
K285 are plotted in the P-T space of figure 5.71D (base-10 logarithm), which also shows
the position of the univariant curve for pure end members (log K ⫽ 0) and, superimposed,
the kyanite-andalusite-sillimanite primary phase fields. As we can see, because the equilib-
rium is not markedly affected by temperature, it is a precise geobarometer.
An important warning regarding thermobarometric deductions based on the

mineral chemistry of solid phases in heterogeneous systems concerns the correct
application of the principle of equality of chemical potentials at equilibrium.
Thermobarometric deductions are based essentially on the fact that in an n-
component system the compositions of phases in an n-phase equilibrium are
uniquely determined at any P-T condition. In the simplest case, as we have seen
in section 5.9.1, the actual n-component system may be reduced to a binary sub-
system of major components, assuming more or less implicitly that the only sig-
nificant chemical variations occur in this subsystem (for example, magnesian and
ferroan components in garnet-clinopyroxene equilibria; cf. eq. 5.216). Also in this
case, however, the principle of equality of chemical potentials at equilibrium de-
mands that the chemical potentials of both components be the same in the two
phases at equilibrium (cf. section 2.3). Relating the chemical potential to the stan-
dard state chemical potential (pure component at P and T of interest; cf. section
2.9.1), it may be easily seen that the equality principle applied to each component
results in two equations in X (composition of the two phases) in two unknowns
(P and T ). The system could thus be theoretically solved (as, unfortunately, it
Figure 5.72 Effects of an erroneous application of the principle of equality of chemical

potentials at equilibrium (from Connolly, 1992).
often is) on the basis of a single component. This, however, is a dangerous ex-
ercise.
As shown in figure 5.72, the equality of potential is satisfied for component
a in mixtures ␣ and ␤ at the measured compositions. Neglecting the energy infor-
mation arising from component b, one could deduce that the two phases, with
the measured composition (open circles), coexist stably at the estimated P and T
of equilibrium. However, as we can see, ␮b,␣ ⬆ ␮b,␤ and equilibrium at the de-
duced P and T conditions would demand different compositions (solid circles).
Things may be even more dramatic with increasing chemical complexity of the
system: correct solution of equilibria of type 5.282 and 5.283 in terms of chemical
potentials involves, for example, five equalities at the univariant equilibrium (i.e.,
10 equations in six variables, SiO2 being a pure phase).
5.9.5 Solid-Fluid Reactions
So far we have considered only compositional terms present in solid mixtures

and/or pure phases. When we wish to evaluate the chemistry of a fluid coexisting
at equilibrium with a given solid paragenesis, a compositional term of the fluid
phase must explicitly appear in the reaction equation. The term “fluid” here
means a generic liquid or gaseous phase or a subcritical liquid-plus-gas assem-
blage. The state of the phase is not important when a particular standard state of
reference is adopted, provided that, once the choice has been made, the adopted
notation is maintained throughout the thermodynamic study. The most important
molecular species appearing in solid-fluid reactions of geochemical interest are
Silicates / 403
O2 , H2O, CO2 , H2 , and S2 . Of the reactions involving molecular oxygen (O2 ), the
following buffer equilibria are of great importance:
1
FeO ⇔ Fe + O 2 (IW) log a O 2 = 6.80 − 27568T −1
wuestite metal
2 gas (5.286)
1
NiO ⇔ Ni + O 2 (NNO) log aO2 = 9.31 − 24810T −1
bunsenite metal
2 gas (5.287)
1
Fe3O 4 ⇔ 3FeO + O2 (MW) log aO2 = 12.92 − 32638T −1
magnetite wuestite
2 gas (5.288)
3SiO2 + 2Fe3O 4 ⇔ 3Fe2SiO 4 + O2 (QFM) log aO2 + 10.50 − 26913T −1

quartz magnetite fayalite gas
(5.289)
1
3Fe2O3 ⇔ 2Fe3O 4 + O2 (HM) log aO2 = 14.41 − 24912T −1
2 gas (5.290)
hematite magnetite
The significance of the term “buffer” stems from the control operated by the
solid paragenesis on the activity of the O2 component in the gaseous phase. Con-
sider, for instance, the buffer magnetite-wuestite. Applying the usual relations
among chemical potential, standard state chemical potential, and thermody-
namic activity, we obtain
3
a FeO × a O1 2
∆G 288
0
= − RT ln 2
. (5.291)
a Fe 3O 4
If we imagine the solid phases to be composed of the pure terms FeO and Fe3O4 ,
equation 5.291 may be reduced to
∆G 288
0
= − RT ln a O1 2 (5.292)
2
and, by rearranging, to
 2 ∆G288
0 
aO2 = exp  − . ( 5.293)
 RT 
Because in geological literature it is customary to express the buffer effect in

terms of “fugacity” of the gaseous species, it should be stressed that the adopted
terminology is often misleading. As we see in equation 5.293, the buffer reaction
determines the thermodynamic activity of the component in the gaseous phase;
“activity” corresponds to “fugacity” only when the adopted standard state for
the gaseous component is that of “pure perfect gas at P ⫽ 1 bar and T ⫽ 298.15
K” because, in this condition, the standard state fugacity is 1 (see section 9.1 for
a discussion on the significance of the various terms).
The buffer reactions of equations 5.286 to 5.290 have been accurately cali-
brated by various authors, and the “fugacity” (actually “activity”) of O2 is usually
expressed as a semilogarithmic function of absolute T. The buffer functions above
are taken from Simons (1986) (eq. 5.286, 5.288, and 5.289), Huebner and Sato
(1970) (eq. 5.287), and Eugster and Wones (1962) (eq. 5.290).
One peculiar oxygen buffer reaction is the well-known thermobarometric
equation of Buddington and Lindsley (1964), which furnishes information on
both T and the activity of molecular oxygen. The reaction uses the equilibrium
among titanomagnetite (or “ulvospinel”), magnetite, hematite, and ilmenite com-
ponents in the hemo-ilmenite and spinel phases:
1
Fe 2TiO 4 + Fe 3O 4 + O2 ⇔ FeTiO3 + 2 Fe2O3 .
2 (5.294)
spinel spinel gas hemo- ilmenite hemo- ilmenite
Equation 5.294 may be split into two partial equilibria, as shown by Powell and
Powell (1977b): the exchange reaction
Fe2 TiO 4 + Fe2O3 ⇔ FeTiO3 + Fe3O 4,

titanomagnetite hematite ilmenite magnetite
(5.295)
which is independent of the gaseous phase, and the HM buffer (eq. 5.290).
Figure 5.73 shows the loci of aO2-T points obtained by intersecting the iso-
pleths of the titaniferous terms (XFeTiO3 and XFe2TiO4 ) in the coexisting spinel and
hemo-ilmenite phases. At T ⬍ 600 °C, the thermobarometer is inapplicable, be-
cause of the solvus field existing between magnetite and ulvospinel in the spinel
phase. Note in the same figure the position of the HM buffer, corresponding to
the absence of Ti in the system. The Buddington-Lindsley thermobarometer finds
application in both igneous (see Duchesne, 1972) and metamorphic terranes
(with the above low-T limits; see also Bohlen and Essene, 1977 for a discussion).
A multiple buffer of particular importance in volcanic gaseous equilibria (see
also section 9.5) is
Fe3O 4 + 3S2 ⇔ 3FeS2 + 2O2

magnetite gas pyrite gas
(5.296)
Silicates / 405
Figure 5.73 Isopleths of titaniferous components ilnenite (FeTiO3) and ulvospinel

(Fe2TiO4) in spinel and hemo-ilmenite coexisting at equilibrium in various aO2-T condi-
tions. Position of hematite-magnetite buffer (HM) is also shown. From Spencer and Lind-
sley (1981). Reprinted with permission of The Mineralogical Society of America.
Assuming unitary activity of components in solid phases (pure phases and stan-
dard state ⫽ pure component at P and T of interest), we have
aO2  ∆G296
0 
K 296 = 2
= exp  − . ( 5.297 )
a S3
2
 RT 
Let us now consider the effect of pressure on equilibrium. We assume for the
sake of simplicity that fluid pressure is equal to total pressure and that gaseous
components mix ideally. From
fO 2
a O2 = (5.298)
fO0
2
and
fS 2
a S2 = , (5.299)
fS0
2
0
where f O2 is the fugacity of O2 and f O 2
is the standard state fugacity of O2 :
fO 2 = PO 2 Γ O 2 (5.300)
fS 2 = PS 2 Γ S 2 , (5.301)
where PO2 is the partial pressure of O2 and ⌫O2 is the fugacity coefficient of O2 :
PO 2 = PX O 2 (5.302)
PS 2 = PX S 2 , (5.303)
where XO2 is the molar fraction of O2 in the gaseous phase, we obtain
3
f S0 Γ O2 X O2
PK 296 = 2 2 2
. ( 5.304 )
02
f O Γ 3S X S3
2 2 2
If we adopt as the standard state for gaseous components the state of pure perfect
gas at P ⫽ 1 bar and T ⫽ 298.15 K (f 0O2 ⫽ f 0S2 ⫽ 1) and neglect for simplicity
the fugacity coefficients, equation 5.304 combined with equation 5.297 gives
X O2  ∆G296
0 
2
= P exp  − . ( 5.305)
X S3
2
 RT 
There are three variables in equation 5.305: composition (and speciation) of the
fluid, P, and T. Once the chemistry of the fluid and P are known, we can deduce
T; if we know the chemistry of the fluid and T, we can deduce P; and if we know
P and T, we can deduce the chemistry of the fluid. This trivariance is a formidable
obstacle in the application of thermobarometry in the presence of fluid phases
and is often underevaluated in geology. Direct sampling of the fluid to constrain
the chemical factor is in most cases impossible, or, when it is possible (e.g., for
volcanic gases, geothermal gases, and fluid inclusions), the speciation state must
still be evaluated and, as we will see in detail in chapter 9, speciation in the fluid
phase is complex and quite variable as a function of P, T, and composition. Never-
theless, in the case of a simple oxygen buffer such as IW, NNO, WM, QFM, HM,
Silicates / 407
or ilmenite-ulvospinel, the estimate of P is not as crucial, because aO2 varies by

several orders of magnitude as a function of T (e.g., if aO2 ⫽ 10⫺25 at T ⫽ 600 °C
and P ⫽ 1 kbar, it will be 10⫺24 at T ⫽ 600 °C and P ⫽ 10 kbar, 10⫺23 at P ⫽
100 kbar, and so on).
In the case of the ilmenite-ulvospinel thermobarometer, aO2 of 10⫺25 at T ⫽ 600 °C is

reached with Ti-rich compositions. If we recall that Avogadro’s number is 6.022 ⫻ 1023
molecules/gfw, we may deduce that, at P ⫽ 1 bar and T ⫽ 600 °C, less that one O2 mole-
cule per mole of fluid is present in the system. Things are even more dramatic at lower T
(with the IW buffer, for instance, aO2 ⬵ 10⫺41 at T ⫽ 300 °C and P ⫽ 1 bar) or at constant
T with increasing Ptotal. In light of the relative abundance of atomic oxygen (which is
dominant in our systems, in most cases), these considerations force us to reflect carefully
on the enormous importance of speciation in the fluid phase.
Concerning estimation of the activity of the H2O component in the fluid, we

can recall, for example, all the extrinsic stability reactions presented in section
5.6.3 for micaceous components (eq. 5.126, 5.128, 5.133, 5.139, 5.140, 5.144,
5.145, 5.146, 5.148, and 5.149) and the respective f H2O calibration functions (eq.
5.127, 5.129, 5.134, 5.141, 5.142, 5.143, and 5.147; the provisos previously out-
lined for O2 are valid for the term “fugacity”). Mica may be also used to deduce
the activity of H2 in the fluid (eq. 5.131) or the combined activities of H2O and
H2 (eq. 5.130, 5.135, 5.137, and 5.138). The application of all these functions is
complicated by difficulties arising when one attempts to evaluate the activity of
micaceous components in mica polymorphs, for which data on mixing behavior
are fragmentary and incomplete (cf. section 5.6.4).
Concerning estimation of the activity of CO2 in the fluid phase, besides
carbonate equilibria (discussed in some detail in section 8.10), equilibria in the
CaO-SiO2-CO2 system are also important. In the medium-low metamorphism of
limestones, the equilibrium
1
CaCO 3 + SiO 2 ⇔ Ca 2 Si 2 O 6 + CO 2 ,
2 (5.306)
calcite quartz wollastonite fluid
investigated in detail by Greenwood (1967), is of particular interest.

PART TWO
Geochemistry of Silicate Melts
CHAPTER SIX
Thermochemistry of Silicate Melts
6.1 Structural and Reactive Properties
The geochemical interpretation of silicate melts is essentially based on three

main concepts:
1. Ionic character of melts

2. Polymeric nature of the anion matrix
3. Lux-Flood acid-base properties of dissolved oxides.
6.1.1 Ionic Character and Quasi-Reticular Interpretation
Electrical conductivity measurements on silicate melts indicate an essentially

ionic conductivity of unipolar type (Bockris et al., 1952a,b; Bockris and Mellors,
1956; Waffe and Weill, 1975). Charge transfer is operated by cations, whereas
anionic groups are essentially stationary. Transference of electronic charges (con-
ductivity of h- and n-types) is observed only in melts enriched in transition ele-
ments, where band conduction and electron hopping phenomena are favored. We
may thus state that silicate melts, like other fused salts, are ionic liquids.
In an ionic melt, coulombic forces between charges of opposite sign lead to
relative short-distance ordering of ions, with anions surrounded by cations and
vice versa. The probability of finding a cation replacing an anion in such ordering
is effectively zero and, from a statistical point of view, the melt may be considered
as a quasi-lattice, with two distinct reticular sites that we will define as anion
matrix and cation matrix.
Let us consider a mixture of two fused salts A⫹Z⫺ and B⫹Y⫺, with A⫹ and
B⫹ cations and Y⫺ and Z⫺ anions. The bulk configurational entropy of the mix-
ture is given by the sum of the configurational entropies of the two matrices:
S mixing = Sanionmixing + S cation mixing . (6.1)
/ 411
412 / Geochemistry of Silicate Melts
If cations and anions are respectively of similar type, the ideal condition of maxi-
mum disorder will be reached in each matrix and the bulk entropy of mixing may
be expressed by
[( ) ( )]
S mixing = − R X Z ln X Z + X Y ln X Y + X A ln X A + X B ln X B , (6.2)
where XA is the ionic fraction of A in the appropriate matrix (i.e., XA ⫹ XB ⫽ 1).

The activity of component AZ in the ideal mixture of the two fused salts AZ
and BY is expressed as
a AZ,melt = X A ⋅ X Z . (6.3)
This “quasi-lattice” formulation of fused salts is known as Temkin’s equation. Its

application to silicate melts was provided by Richardson (1956), but it is inade-
quate for the compositional complexity of natural melts, mainly because, in a
compositionally complex melt, the types of anions and consequently the entity
of the anion matrix vary in a complicated way with composition.
Substantial further development in the theory was achieved by Toop and
Samis (1962a,b) and Masson (1968), and consisted of the application of polymer
theory to fused silicates.
6.1.2 Polymeric Nature of Anion Matrix: Toop-Samis and

Masson Models
In polymeric models for silicate melts, it is postulated that, at each composition,
for given values of P and T, the melt is characterized by an equilibrium distribu-
tion of several ionic species of oxygen, metal cations, and ionic polymers of mo-
nomeric units SiO4⫺4 .
The charge balance of a polymerization reaction involving SiO4⫺ 4 monomers
may be formally described by a homogeneous reaction involving three forms of
oxygen: singly bonded O⫺, doubly bonded O 0 (or “bridging oxygen”), and free
oxygen O2⫺ (Fincham and Richardson, 1954):
2 O − ⇔ O 0 + O 2− .
melt melt melt
(6.4)
In fact, equation 6.4 is similar to a reaction between SiO4⫺

4 monomers:
SiO 44− + SiO 44− ⇔ Si 2 O 76 − + O 2 − , (6.5)
which in stereochemical representation may be conceived as follows:

Thermochemistry of Silicate Melts / 413
O− O− O− O−
O − − Si − O − + O − − Si − O − ⇔ O − − Si − O − Si − O − + O 2 − . (6.6)
− − − −
O O O O
Polymer chemistry shows that the larger the various polymers become, the
more their reactivity becomes independent of the length of the polymer chains.
This fact, known as the “principle of equal reactivity of cocondensing functional
groups,” has been verified in fused polyphosphate systems (which, for several
properties, may be considered as analogous to silicate melts; cf. Fraser, 1977) with
polymeric chains longer than 3 PO3⫺4 units (Meadowcroft and Richardson, 1965;
Cripps-Clark et al., 1974). Assuming this principle to be valid, the equilibrium
constant of reaction 6.4:
K4 =
(O )(O )
0 2−
(6.7)
(O )
2
−
(in which the terms in parentheses represent the number of moles in the melt) is
always representative of the polymerization process, independent of the effective
length of the polymer chains.
Toop and Samis (1962a,b) showed that, in a binary melt MO-SiO2 , in which
MO is the oxide of a basic cation completely dissociated in the melt, the total
number of bonds per mole of melt is given by
( ) ( )
2 O 0 + O − = 4 N SiO2 , (6.8)
where NSiO2 are the moles of SiO2 in the MO-SiO2 melt. The number of bridging
oxygens in the melt is thus
4 N SiO2 − O − ( ).
(O ) =
0
2
(6.9)
Mass balance gives the number of moles of free oxygen per mole of melt:
(O ) , −
(O ) = (1 − N )
2−
SiO 2 −
2
(6.10)
where (1 ⫺ NSiO2 ) are the moles of basic oxide in the melt. Equations 6.7 to
6.10 yield
Table 6.1 Terms of quadratic equation 6.12 for various

values of NSiO2 (Toop and Samis, 1962a).
NSiO2 a(O⫺)2 ⫹ b(O⫺) ⫹ c ⫽ 0
0.10 (4K4 ⫺ 1)(O⫺)2 2.2(O⫺) ⫺0.72

0.20 (4K4 ⫺ 1)(O⫺)2 2.4(O⫺) ⫺1.28
⫺ 2 ⫺
0.30 (4K4 ⫺ 1)(O ) 2.6(O ) ⫺1.68
⫺ 2 ⫺
0.40 (4K4 ⫺ 1)(O ) 2.8(O ) ⫺1.92
⫺ 2 ⫺
0.50 (4K4 ⫺ 1)(O ) 3.0(O ) ⫺2.00
0.60 (4K4 ⫺ 1)(O⫺)2 3.2(O⫺) ⫺1.92
⫺ 2 ⫺
0.70 (4K4 ⫺ 1)(O ) 3.4(O ) ⫺1.68
⫺ 2 ⫺
0.80 (4K4 ⫺ 1)(O ) 3.6(O ) ⫺1.28
⫺ 2 ⫺
0.90 (4K4 ⫺ 1)(O ) 3.8(O ) ⫺0.72
K4 =
[4N SiO 2 ( )] [ 2 − 2 N
− O− SiO 2 − O− ( )] . (6.11)
4 (O )
2
−
Thus, (O⫺ ) is given by the quadratic equation
(O ) ( 4 K ) ( )( ) ( )
2
−
4 − 1 + O − 2 + 2 N SiO2 + 8N SiO2 N SiO2 − 1 = 0, (6.12)
which may be solved for discrete values of K4 and NSiO2 (see table 6.1).
Figure 6.1 shows the solution of the system for K4 ⫽ 0.06: note that the distri-
bution of the three forms of oxygen is asymmetric over the compositional space.
Moreover, bridging oxygen is the only form present in the SiO2 monomer.
The Gibbs free energy change involved in equation 6.4 is
∆G 40 = − RT ln K 4 (6.13)
Because two moles of O⫺ produce one mole of O 0 and one of O2⫺, the Gibbs
free energy of mixing per mole of silicate melt is given by
∆G mixing =
(O ) RT ln K
−
4. ( 6.14 )
2
Figure 6.2 shows Gibbs free energy of mixing values obtained by application of
equation 6.14. They are remarkably similar to the Gibbs free energy of mixing
values experimentally observed on binary synthetic systems of appropriate com-
position. This means that the Gibbs free energy of mixing in an MO-SiO2 melt is
Figure 6.1 Equilibrium distribution of (O0), (O⫺), and (O2⫺) in binary field MO-SiO2 for
K4 ⫽ 0.06. Reprinted from G. W. Toop and C. S. Samis, Canadian Metallurgist Quarterly,
1, 129–152, copyright 䉷 1962, with kind permission from Elsevier Science Ltd., The Bou-
levard, Langford Lane, Kidlington 0X5 1GB, UK.
Figure 6.2 Gibbs free energy of mixing in the quasi-chemical model of Toop and Samis
(1962a,b), compared with values experimentally observed in PbO-SiO2 and CaO-SiO2
melts. Reprinted from Toop and Samis (1962b), with kind permission of ASM Interna-
tional, Materials Park, Ohio.
essentially attributable to interactions among silica and the various forms of

oxygen.
If we apply Gibbs-Duhem equations (cf. section 2.11) we may deduce that
the activity of the dissolved basic oxide is entirely represented by the activity of
free oxygen in the system, so that
∆G mixing
(1 − N ) log a
SiO 2 O2−
+ N SiO2 log a SiO2 =
2.303 RT
. (6.15)
In Masson’s model (Masson, 1965, 1968, 1972), polymerization reactions are vis-
ualized as a stepwise process of the type
SiO 44− + SiO 44− ⇔ Si 2 O 76 − + O 2 − (6.16)
Si 2 O 76 − + SiO 44− ⇔ Si 3O 10
8−
+ O 2− , (6.17)
which may be generalized in the form
Si nO (32nn++1 2 ) − + SiO 44− ⇔ Si n+1O (32nn++44) − + O 2 − . (6.18)
The constants of equations 6.16 and 6.17 are identical, on the basis of the prin-
ciple of equal reactivity of cocondensing functional groups. Adopting Temkin’s
model for fused salts, and thus assuming that molar fractions represent activities
over the appropriate matrix, we obtain
X Si ⋅ X O2 − X Si ⋅ X O2 − X ⋅ X O2 −
6−
2O7
8−
3O10 Si n + 1O(32nn++44 ) −
= =K . ( 6.19 )
X SiO4 − ⋅ X SiO4 − X SiO4 − ⋅ X Si 6− X SiO4 − ⋅ X
4 4 4 2O7 4 Si nO(32nn++12 ) −
Simplifying this infinite set of equilibria, Masson obtained an equation relating

the molar fraction of silica in an SiO2-MO melt (with MO basic oxide completely
dissociated at standard state to give M2⫹ ⫹ O2⫺ ) to the activity of dissolved
oxide aMO:
1 1 1
=2+ − .
X SiO2 1 − a MO  1  (6.20)
1 + a MO  − 1
 K 16 
Equation 6.20 is valid only for open or “bifunctional” chains (cf. Fraser, 1977).
An extension of the model to open ramified polymer chains leads to the equation
1 1 3
=2+ − .
X SiO2 1 − a MO  3  (6.21)
1 + a MO  − 1
 K 16 
(Whiteway et al., 1970).

Figure 6.3 shows the oxide activities of Masson’s model compared with exper-
imental values observed in MO-SiO2 systems. As already described in chapter 2,
the significance of activity is intimately connected with the choice of a standard
state of reference, which in Masson’s model is that of completely dissociated basic
oxide. The more the actual behavior of the metal oxide in the melt differs from
the reference condition (i.e., the more the metal oxide is acidic in the Lux-Flood
sense of the term; see section 6.1.3), the more the thermodynamic activity of the
oxide increases in the melt, silica amounts being equal (see figure 6.3). Here, as
in the model of Toop and Samis (figure 6.2), we observe a direct relationship
between degree of acidity of the dissolved oxide and polymerization constant K16 .
The significance of this correlation will become clearer in the next section.
6.1.3 Acid-Base (Lux-Flood) Character of Dissolved Oxide

and Its Bearing on Structure of Melt
The reactive properties of oxides have always been related to their capacity to
form salts—i.e., “basic” oxides such as CaO, FeO, and MgO may form salts by
reaction with “acidic” oxides such as SiO2 and CO2 :
Acid + Base ⇔ Salt. (6.22)
For instance,
SiO 2 + 2 MgO ⇔ Mg2 SiO 4 ( 6.23)
CO 2 + CaO ⇔ CaCO 3 . ( 6.24 )
Moreover, in geology, a rock with a high SiO2 content is usually defined as

“acidic”. Although the underlying concepts are obvious, the terminology used
deserves some clarification.
The acid-base behavior of a substance in aqueous solutions and other proton-
ated systems is conveniently described by the Brönsted-Lowry equation:
Acid ⇔ Base + H + (6.25)
—i.e.,
Figure 6.3 Experimental oxide activities in MO-SiO2 melts compared with estimates of
Masson’s model. Reprinted from Masson (1968), with kind permission of American Ce-
ramic Society Inc., Westerville, Ohio.
H2 SO 4 ⇔ HSO −4 + H + . ( 6.26 )
In silicate melts and other nonprotonated solvents, the Brönsted-Lowry equation

is not applicable and is conveniently replaced by the Lux-Flood acid-base defini-
tion (Lux, 1939; Flood and Förland, 1947), according to which free oxygen O2⫺
replaces H⫹. A “basic oxide” is one capable of furnishing oxygen ions, and an
“acidic oxide” is one that associates oxygen ions:
Base ⇔ Acid + O 2 (6.27)
—i.e.,
SiO 44− ⇔ SiO 23 − + O 2 − (6.28.1)
SiO 23 − ⇔ SiO 2 + O 2 − . (6.28.2)
A generic oxide MO will dissociate as a basic oxide when
MO ⇔ M 2 + + O 2 − (6.29)
or as an acidic oxide when

Table 6.2 Relationship between field

strength of dissolved cation and
polymerization constant in MO-SiO2 melts
(data from Hess, 1971).
ZMZ⫹
Cation K4
rMZ⫹ ⫹ rO2⫺
K⫹ 0.13 0.001
Na⫹ 0.18 0.001
Li⫹ 0.23 0.001
Pb2⫹ 0.30 0.002–0.01
Ca2⫹ 0.35 0.002
Mn2⫹ 0.41 0.04
Fe2⫹ 0.44 0.05
Co2⫹ 0.44 0.10
Ni2⫹ 0.46 0.35
MO + O 2 − ⇔ MO 22 − . (6.30)
An “amphoteric oxide” is an oxide component with reaction properties interme-

diate between those of equations 6.29 and 6.30.
It is obvious that the acid-base property of a dissolved oxide may markedly
affect the structure of a silicate melt. A dissolved acidic oxide associates the free
oxygen, thus displacing reaction 6.4 toward the right, resulting in a marked corre-
lation between the field strength of the dissolved cation and the polymerization
constant of the melt. This correlation is shown in the values listed in table 6.2.
For instance, a quite marked effect on the structure of the silicate melt is
produced by the basic oxide H2O. Addition of water causes a drastic decrease in
the viscosity of silicate melts: adding 6.4 weight % of H2O to a granitic melt at
T ⫽ 1000 °C causes a decrease in viscosity of about six orders of magnitude (i.e.,
from about 10⫺1 0 to about 10⫺4 poise⫺1; cf. Burnham, 1975). The dissolution
mechanism of H2O is important in magma rheology and will be discussed more
extensively in section 9.6.1.
6.2 Structure of Natural Melts
The structure of the anion matrix of silicate melts is conveniently represented in

a ternary plot whose apexes are the SiIV, O 0 , and O⫺ constituents of the anion
groups. In this ternary representation, all the possible types of anions are aligned
along a straight polymerization path, from the SiO4⫺ 4 monomer to the SiO2 poly-
mer (Toop and Samis, 1962a). For a given amount of silica, the value assumed
by polymerization constant K4 determines the exact position along the path (fig-
ure 6.4).
Figure 6.4 Ternary plot of singly bonded oxygen O⫺, silicon SiIV, and bridging oxygen
O0 in each kind of polyanion. Shaded areas: compositional ranges 6.31 and 6.33. Re-
printed from Ottonello (1983), with kind permission of Theophrastus Publishing and Pro-
prietary Co.
In natural melts, the presence of high field strength ions such as Al 3⫹, Fe3⫹,
Ti , and P5⫹, which, like silicon, preferentially assume a tetrahedral coordination
4⫹
with oxygen, complicates the structure, and the constitution of the anion matrix
may not be deduced on the basis of the equations in section 6.1.2. Structural
parameters valid for compositionally complex melts were proposed by Mysen et
al. (1980) and Virgo et al. (1980) on the basis of the results of Raman spectros-
copy. These parameters are NBO/Si and NBO/T (NBO ⫽ Non-Bridging Oxygen;
T groups all tetrahedrally coordinated cations—i.e., Si4⫹, Al 3⫹, Fe3⫹, Ti4⫹, P5⫹,
. . .). Comparing the Raman spectra of silicate melts with various NBO/Si and
NBO/T ratios with the Raman spectra of crystals of the same chemical com-
position, the above authors subdivided the anion structure of silicate melts into
three compositional ranges:
4 > NBO Si > 2 (6.31)
2 > NBO Si > 1 (6.32)
1 > NBO Si > 0.1; NBO T < 0.75 (6.33)

The following disproportionation reactions dominate, respectively, in the three

ranges:
2 Si 2 O 76 − ⇔ 2 SiO 44− + Si 2 O 64− (6.34)
3 Si 2 O 64− ⇔ 2 SiO 44− + 2 Si 2 O 52 − (6.35)
2 Si 2 O 52 − ⇔ Si 2 O 64− + 2 SiO 2 (6.36)
T2 O 52 − ⇔ TO 23 − + TO 2 (6.37)
These equations do not imply any progression in the polymerization process and
reduce to identities in the Fincham-Richardson formalism (cf. table 6.3). Never-
theless, the experimental evidence of Mysen et al. (1980) and Virgo et al. (1980)
may be explained by the progressive polymerization steps listed in table 6.3. These
equilibria are consistent with the Fincham-Richardson formalism (of which they
constitute simple multiples) and obey the proportionality rules of figure 6.4.
Whiteway et al. (1970) derived an equation defining the fractional amount
NX of the various polymer units of extension X as a function of the ratio between
bridging oxygen (O 0 ) and singly bonded oxygen in an initially totally depolymer-
ized melt (O⫺*) composed entirely of SiO4⫺4 monomers (i.e., ␣ ⫽ O /O *):
0 ⫺
X −1 2 X +1
 2α   2α 
NX = ωX   1 −  (6.38)
 3   3 
3X
ωX = . ( 6.39 )
( )
2X + 1 ! X!
Table 6.4 lists the molar fractions of the various polymer units calculated for ␣
values between 0.1 and 0.5.
Note the close connection between the values deduced by application of
equation 6.38 and the ternary representation of figure 6.4. The polymerization
field 0.1 ⬍ ␣ ⬍ 0.5 corresponds roughly to the range 4 ⬎ NBO/Si ⬎ 2. The
polymer units in this range are mainly monomers, dimers, and ramified tetramers
13 ). With increasing ␣, we progressively move along the polymerization
(Si4O10⫺
path, the molar fraction of monomers decreases, and the mean length of the
polymer units increases. For values of ␣ higher than 0.5, equation 6.38 is no
longer valid and is replaced by
Table 6.3 Reaction schemes for coexistence of various polymers in natural melts (from
Ottonello, 1983).
Compositional Disproportionation Polymerization

Range Scheme Reactions
4 ⬎ NBO/Si ⬎ 2 2Si2O76⫺ ⇔ 2SiO44⫺ ⫹ Si2O64⫺ 2SiO44⫺ ⇔ Si2O76⫺ ⫹ O2⫺

⫺
12O⫺ ⫹ 4O0 ⇔ 12O ⫹ 4O0 Si2O76⫺ ⇔ Si2O64⫺ ⫹ O2⫺
8O⫺ ⇔ 6O⫺ ⫹ O0 ⫹ O2⫺
6O⫺ ⫹ O0 ⇔ 4O⫺ ⫹ 2O0 ⫹ O2⫺
2 ⬎ NBO/Si ⬎ 1 3Si2O64⫺ ⇔ 2SiO44⫺ ⫹ 2Si2O52⫺ 2SiO44⫺ ⇔ Si2O64⫺ ⫹ 2O2⫺
⫺
12O⫺ ⫹ 6O0 ⇔ 12O ⫹ 6O0 2Si2O64⫺ ⇔ Si2O52⫺ ⫹ O2⫺
8O⫺ ⇔ 4O⫺ ⫹ 2O0 ⫹ 2O2⫺
4O⫺ ⫹ 2O0 ⇔ 2O⫺ ⫹ 3O0 ⫹ O2⫺
1 ⬎ NBO/Si ⬎ 0.1 2Si2O52⫺ ⇔ Si2O64⫺ ⫹ 2SiO2 Si2O64⫺ ⇔ 2SiO2 ⫹ 2O2⫺
⫺
NBO/T ⬍ 0.75 4O⫺ ⫹ 3O0 ⇔ 4O ⫹ 3O0 Si2O64⫺ ⇔ Si2O52⫺ ⫹ O2⫺
4O⫺ ⫹ 2O0 ⇔ 4O0 ⫹ 2O2⫺
4O⫺ ⫹ 2O0 ⇔ 2O⫺ ⫹ 3O0 ⫹ O2⫺
Table 6.4 Relationship between relative amounts of bridging oxygen ␣

and molar proportions of various polymer units (from Henderson, 1982).
␣ Monomer Dimer Trimer Tetramer Other
0.1 0.813 0.142 0.033 0.009 0.003

0.2 0.651 0.196 0.078 0.036 0.039
0.3 0.512 0.197 0.101 0.059 0.131
0.4 0.394 0.170 0.123 0.064 0.249
0.5 0.296 0.132 0.078 0.053 0.441
X −1 2 X +1
 2α   2α 
NX = ωX 
 3 
 1 −


3 
(1 − N ) ,
O2−
(6.40)
where NO2⫺ is the molar fraction of free oxygen.

As regards the coordination states of the various cations in the melt, the main
experimental information comes from Mössbauer spectroscopy and infrared ab-
sorption spectroscopy on silicate glasses. Because the glass-melt transition in-
volves partial rearrangement of the structure, the interpretation of experimental
evidence is not straightforward (cf. Boon and Fyfe, 1972). Treating the melt as
chaotic packing of rigid spheres all of the same dimension, Bernal (1964) showed
that the main coordination of the cavities between spheres is tetrahedral (73%),
followed by the half-octahedron (20.3%), trigonal prism (3.2%), tetragonal do-
decahedron (3.1%), and Archimedean antiprism (0.4%). Moreover, the mean co-
Figure 6.5 Coordination state of Ti4+ in glasses of TiO2-SiO2 system. Reprinted from
R. B. Greegor, F. W. Lytle, D. R. Sanstrom, J. Wang, and P. Schultz. Investigation of
TiO2-SiO2 glasses by X-ray absorption spectroscopy, Journal of Non-Crystalline Solids, 55,
27–43, copyright 䉷 1983, with kind permission of Elsevier Science-NL, Sara Burgerhart-
straat 25, 1055 KV Amsterdam, The Netherlands.
ordination number (VIII-IX) is lower than that observed in regular packing of

rigid spheres (XII). Indeed, existing experimental studies on glasses show that
the coordination number varies between IV and VIII-IX according to type of
cation, chemistry of the system, and temperature. For instance, in alkaline glasses
(Li2O-SiO2 , Na2O-SiO2 and Rb2O-SiO2 ), Ni2⫹ exhibits coordination IV and VI.
Moreover, the fraction of tetrahedrally coordinated sites increases with increasing
T, composition being equal, and decreases with the increasing field strength of
the major dissolved cation (i.e., from Rb to Li) at the same T. Experimental inves-
tigations on the structures of TiO2-SiO2 glasses by EXAFS (Extended X-ray Ab-
sorption Fine Structure) and XANES (X-ray Absorption Near Edge Structure)
carried out by Greegor et al. (1983) show that Ti4⫹ occurs in both coordination
IV and coordination VI. The proportions of the two sites vary in a complex fash-
ion with the chemistry of the system (figure 6.5): in a pure silica glass, up to 0.05
weight % of TiO2 (X ⫽ 0.0004) Ti4⫹ occupies exclusively tetrahedral sites. For
greater amounts of TiO2 (X ⫽ 0.07), Ti4⫹ substitutes for Si4⫹ and the VI-fold
coordinated fraction remains below 5%. Above 9 weight % of dissolved TiO2 , the
ratio in VI-fold and IV-fold coordination increases appreciably, and at 15% of
dissolved TiO2 a structural arrangement similar to rutile is observed (figure 6.5).
Another group of elements attentively investigated for their coordination
states in silicate glasses and melts is the rare earth elements (REE). Existing stud-
ies show that Ce3⫹, Nd 3⫹, Eu3⫹, Er3⫹, and Yb3⫹ occupy mainly octahedral sites,
although coordinations VII and VIII may occasionally be observed (cf. Rob-
inson, 1974, and references therein).
6.3 Effects of P and T Intensive Variables on Melting
Let us consider one mole of a pure crystalline phase at equilibrium with one mole
of melt of the same composition—e.g., pure forsterite at P ⫽ 1 bar and T ⫽ 2163
K. Because the two phases are at equilibrium, we have
µ Mg 2SiO4 ,olivine = µ Mg 2SiO4 , melt . (6.41)
If we vary the temperature and/or pressure, equilibrium will be maintained

through identical modifications of the chemical potential of the Mg2SiO4 compo-
nent in the two phases:
dµ Mg 2 SiO4 ,olivine = dµ Mg 2 SiO4 , melt . ( 6.42 )
Because for a pure single-component phase,
dµ = dG = − S dT + V dP , (6.43)
we can write
− S Mg dT + V Mg dP = − S Mg dT + V Mg dP. (6.44)
2 SiO4 , olivine 2 SiO4 , olivine 2 SiO4 , melt 2 SiO4 , melt
Collecting the terms in dP and dT, respectively, we have
(
dT S Mg SiO
2 4 , melt
− S Mg SiO
2 4 ,olivine
) = dP(V Mg 2SiO 4 , melt )
− VMg 2SiO4 ,olivine . (6.45)
Denoting the two terms in brackets, respectively, as ⌬S fusion and ⌬V fusion and
rearranging, we obtain the Clausius-Clapeyron equation:
dT ∆Vfusion
= . (6.46)
dP ∆S fusion
Because ⌬G ⫽ 0 for the two phases at equilibrium, we also have
∆ H fusion
∆ S fusion = (6.47)
T
Figure 6.6 Experimentally observed univariant dT/dP equilibria for several silicates of
petrogenetic interest. Melting curves are satisfactorily reproduced by Simon equation with
parameters listed in table 6.5.
dT Tfusion ⋅ ∆Vfusion
= . (6.48)
dP ∆ H fusion
The Clausius-Clapeyron equation describes the univariant equilibrium between

crystal and melt in the P-T field. Because molar volumes and molar entropies of
molten phases are generally greater than their crystalline counterparts, the two
terms ⌬S fusion and ⌬V fusion are both positive and we almost invariably observe an
increase in Tfusion with increasing pressure. Although extension to multicompo-
nent systems is somewhat more complex, we can now understand why adiabatic
decompression acting on a rock system generally induces partial melting.
Univariant melting curves for several pure components are shown in figure
6.6. Because silicate melts are more compressible than their crystalline counter-
parts, ⌬V fusion decreases with increasing P, thus causing curvature in the univari-
ant line.
Table 6.5 Parameters of Simon equation for some rock-forming silicates (after Bottinga, 1985).
Tmax and Pmax are maximum conditions of applicability.
Tmax TSimon Pmax PSimon a b

Compound (K) (K) (kbar) (kbar) (kbar) (adimensional)
Mg2SiO4 2736 2163 150 0 108.33 3.7

Mg3Al2Si3O12 2410 2073 100 40 19.79 9.25
Mg2SiO3 2220 1830 46.7 0 28.65 5.01
CaMgSi2O6 2172 1670 50 0 34.71 3.40
NaAlSi2O6 1817 1609 43 29 7.97 8.35
NaAlSi3O8 1625 1373 30 0 60.95 2.38
Table 6.5 lists the parameters of the Simon equation simulating the dT/dP
evolution of the univariant curve. The Simon equation:
 b 
T 
P = a   − 1 + PSimon (6.49)
 TSimon  
 
is a purely empirical fitting of experimental evidence and must be used with cau-
tion (cf. Bottinga, 1985).
6.4 Effects of Chemical Composition on Melting

(Equation of Lowering of Freezing Point)
The chemical potential of the component of interest in a multicomponent phase

must be referred to its potential at standard state. Equation 6.41 is thus reex-
pressed as follows:
µ Mg
0
+ RT ln a Mg 2SiO4 ,olivine = µ Mg
0
+ RT ln a Mg 2SiO4 , melt . (6.50)
2SiO 4 ,olivine 2SiO 4 , melt
Gathering together the logarithmic terms, we obtain
a Mg 2SiO4 , melt µ Mg
0
− µ Mg
0
2SiO 4 ,olivine 2SiO 4 melt
ln = . (6.51)
a Mg 2SiO4 ,olivine RT
Because
µ0 H0 S0
= − , (6.52)
RT RT R
differentiating equation 6.51 in dT yields:

∂  a Mg 2SiO4 , melt  H Mg
0
− H Mg
0
2SiO 4 , melt 2SiO 4 ,olivine
 ln  = . (6.53)
∂ T  a Mg 2SiO4 ,olivine  RT 2
The term (H 0Mg2SiO4,melt ⫺ H 0Mg2SiO4,olivine ) is the molar enthalpy of solution of

pure forsterite in a pure Mg2SiO4 melt, and is equivalent to the molar enthalpy
of fusion ⌬H fusion at T ⫽ Tfusion.
Because the variation of enthalpy of solution between two temperatures may
be expressed by the integral
T T
∫ d∆H solution = ∫ ∆C P dT , (6.54)

Tfusion Tfusion
if we assume ⌬CP to be constant within the integration limits, equation 6.53 be-
comes
a Mg 2SiO4 , melt 1
T
(
∆H fusion + ∆C P T − Tfusion ) dT ,
ln
a Mg 2SiO4 ,olivine
=
RT ∫ T 2
(6.55)
fusion
which, when integrated, gives
a Mg 2 SiO4 , melt 1   1 1  T 
ln
a Mg 2 SiO4 ,olivine
= (
 ∆H fusion − ∆CP Tfusion 
R   Tfusion T 
)
−  + ∆CP ln  .
 Tfusion  
( 6.56 )
(see Saxena, 1973 and references therein; Wood and Fraser, 1976).
Equation 6.56 is known as the “equation of lowering of freezing point” and
is valid for solid mixtures crystallizing from multicomponent melts. Like the
Clausius-Clapeyron equation, it tells us how the system behaves, with changing
T, to maintain equilibrium on the univariant curve. However, whereas in the
Clausius-Clapeyron equation equilibrium is maintained with concomitant
changes in P, here it is maintained by appropriately varying the activity of the
component of interest in the melt and in the solid mixture.
Recalling the relationship between thermodynamic activity and molar con-
centration,
ai = X i γ i , (6.57)
we see that, for practical purposes, the response of the system to modifications in
T consists of a flow of matter between the two phases at equilibrium.
Lastly, it must be noted that, whenever a pure single-component phase crys-
tallizes from the melt, the denominator of the term on the left side of equation
6.56 is reduced to 1 and the equation directly furnishes the activity of the dis-
solved component in the melt.
6.5 Mixing Properties in Melts with Constant

Silica Contents
We have already seen that the degree of polymerization of the melt is controlled
by the amount of silica in the system (see, for instance, figure 6.4). If we mix two
fused salts with the same amount of silica and with cations of similar properties,
the anion matrix is not modified by the mixing process and the Gibbs free energy
of mixing arises entirely from mixing in the cation matrix—i.e.,
S mixing = S cation, mixing . (6.58)
This hypothesis, advanced by Richardson (1956), has received some experimen-

tal confirmation (Belton et al., 1973; Rammensee and Fraser, 1982; Fraser et
al., 1983, 1985). In particular, in a series of experimental studies using Knud-
sen cell mass spectrometry, Fraser et al. (1983, 1985) showed that the melts
NaAlSi2O6-KAlSi2O6 , NaAlSi3O8-KAlSi3O8 , NaAlSi4O1 0 -KAlSi4O1 0 , and
NaAlSi5O12-KAlSi5O12 mix ideally over the entire binary range. Table 6.6 lists the
activity coefficients of the jadeite component NaAlSi2O6 in the molten mixture
NaAlSi2O6-KAlSi2O6 obtained at various T by Fraser et al. (1983). Within experi-
mental incertitude, these values approach 1.
The results of the experimental study of Fraser et al. (1983) on the
NaAlSi2O6-KAlSi2O6 join are at first sight surprising, because the crystalline
counterparts of the pure end-members (i.e., jadeite and leucite) have very differ-
Table 6.6 Activity coefficients of jadeite component NaAlSi2O6 in molten (and/or over cooled)
mixture NaAlSi2O6-KAlSi2O6. Terms in parentheses indicate experimental incertitude at various T
(from Fraser et al., 1983).
␥NaAlSi2O6
XNaAlSi2O6 1200 °C 1300 °C 1400 °C 1500 °C 1600 °C
0.4 1.032 (⫾0.03) 1.043 (⫾0.02) 1.043 (⫾0.02) 1.050 (⫾0.02) 0.998 (⫾0.04)
0.5 1.022 1.039 1.046 1.057 1.041
0.6 1.014 1.035 1.048 1.063 1.069
0.7 1.008 1.031 1.051 1.069 1.086
0.75 1.005 1.029 1.051 1.070 1.089 (⫾0.02)
0.8 1.004 1.018 1.032 1.045 1.056
0.9 1.001 1.005 1.006 1.011 1.014
1.0 1.000 1.000 1.000 1.000 1.000
Figure 6.7 Excess partial molar properties of jadeite component in NaAlSi2O6-KAlSi2O6

molten mixture. (A) Excess partial molar entropy. (B) Excess partial molar enthalpy. Re-
printed from Fraser et al. (1983), Bulletin Mineralogique, 106, 111–117, with permission
from Masson S.A., Paris, France.
ent structural arrangements. Evidently the hypothesis of Richardson (1956) is

valid and the melt does not carry any structural remnant of the crystalline aggre-
gates after the melting event.
Nevertheless, if we analyze the excess partial molar properties of the terms in
the mixture, we see well-defined minima at XNaAlSi2O6 ⫽ 0.75 (Fraser et al., 1983).
And in figure 6.7, there are marked minima in the excess partial molar entropy
and excess partial molar enthalpy of the jadeite component in the mixture at
this concentration level. The compositional ratio Na:K ⫽ 3:1 corresponds to the
eutectic minimum in the low-SiO2 portion of the SiO2-KAlSiO4-NaAlSiO4 system
(T ⫽ 1020 °C at P ⫽ 1 bar; Schairer, 1950). Moving from the K-rich side of the
join toward the sodic terms, at Na:K ⫽ 3:1, we find the field of Na-K ␤-nepheline,
with two sites of different sizes in the ratio 3:1, occurring at liquidus.
Table 6.7 lists experimental activity coefficient values for albite and orthoclase
components in the NaAlSi3O8-KAlSi3O8 molten mixture. The albite component
mixes ideally at all T in the compositional range Ab100 to Ab2 0 Or8 0 , but deviates
slightly from ideality for more potassic compositions.
Analyzing the partial molar properties of the terms in the mixtures, we ob-
serve a sudden change in slope for the partial molar enthalpy of the two compo-
nents at XAb ⫽ 0.2 (figure 6.8), whereas the integral enthalpy of mixing remains
near zero over the entire compositional range. Rammensee and Fraser (1982)
interpret this as a result of incipient ordering in the melt structure, which is neces-
sary to approach the structural arrangement observed in leucite, prior to its crys-
tallization.
From the various experimental evidence available on the investigated systems,
Table 6.7 Activity coefficients of albite (Ab) and orthoclase (Or) components in molten mixtures
and over cooled liquids in NaAlSi3O8-KAlSi3O8 system (after Rammensee and Fraser, 1982).
1200 °C 1300 °C 1400 °C 1500 °C
XAb Ab Or Ab Or Ab Or Ab Or
0 1.56 1.00 1.41 1.00 1.27 1.00 1.11 1.00

0.05 1.30 1.01 1.20 1.00 1.09 1.00 1.00 1.00
0.1 1.12 1.02 1.08 1.01 1.00 1.01 1.00 1.00
0.2 1.00 1.04 1.00 1.02 1.00 1.01 1.00 1.00
0.3 1.00 1.04 1.00 1.02 1.00 1.01 1.00 1.00
0.4 1.00 1.04 1.00 1.02 1.00 1.01 1.00 1.00
0.5 1.00 1.04 1.00 1.02 1.00 1.01 1.00 1.00
1.0 1.00 1.04 1.00 1.02 1.00 1.01 1.00 1.00
Figure 6.8 Partial molar enthalpies and integral enthalpy of mixing in pseudobinary
NaAlSi3O8-KAlSi3O8 system. Reprinted from W. Rammensee and D. G. Fraser, Geochim-
ica et Cosmochimica Acta, 46, 2269–2278, copyright 䉷 1982, with kind permission from
Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
Fraser et al. (1983, 1985) deduced that, in several instances, the structure of the
silicate melt should mimic the structural arrangement of the solid phase at liq-
uidus. In incongruent melting processes, the structure of the liquid differs sub-
stantially from that of the solid phase of identical stoichiometry.
The quasi-ideality of several other molten systems with constant amounts of
silica may be deduced by applying the lowering of freezing point equation. A very
instructive example in this sense concerns the Fe2SiO4-Mg2SiO4 system (Saxena,

1973 and references therein; Wood and Fraser, 1976). Adopting Richardson’s
model to describe the activity of fayalite (Fe2SiO4 : Fa) and forsterite (Mg2SiO4 :
Fo) components in the binary melt, we obtain
a Fo, melt = X Mg,

2
cations (6.59)
a Fa, melt = X Fe,

2
cations . (6.60)
Because the solid Fe2SiO4-Mg2SiO4 mixture is ideal (cf. section 5.2.5), the activi-
ties of components in the solid are given by
a Fo,olivine = X Mg,olivine
2
(6.61)
a Fa,olivine = X Fe,olivine
2
. (6.62)
Substituting equations 6.59, 6.60, 6.61, and 6.62 into the lowering of freezing
point equation, we obtain
 X Mg, melt  1
T
( )
∆H fusion,Fo + ∆C P ,Fo T − Tfusion,Fo
2 ln   = ∫
 X Mg,olivine  R Tfusion,Fo T2
dT ( 6.63)
 X Fe, melt  1
T
( )
∆H fusion,Fa + ∆C P,Fa T − Tfusion,Fa
2 ln   = ∫
 X Fe,olivine  R Tfusion,Fa T2
dT . (6.64)
Wood and Fraser (1976) assigned the following molar values to the various pa-
rameters: ⌬H fusion,Fo ⫽ 34,240 (cal); ⌬H fusion,Fa ⫽ 25,010 (cal); ⌬CP,Fo ⫽ ⌬CP,Fa ⫽
7.45 (cal/K); Tfusion,Fa ⫽ 1479.5 (K); Tfusion,Fo ⫽ 2164 (K).
Ratios XMg,melt /XMg,olivine and XFe,melt /XFe,olivine necessary to maintain equilib-
rium between the coexisting phases (solid plus melt) can be calculated by means
of equations 6.63 and 6.64. The results are shown in table 6.8.
If we let a represent the XFe,melt /XFe,olivine ratio and let b represent the XMg,melt /
XMg,olivine ratio, then, because
X Fe, melt = a ⋅ X Fe,olivine (6.65)
and
Table 6.8 Relative amounts of Fe and Mg in olivine and melt at equilibrium, based on equations
6.63 and 6.64 (after Wood and Fraser, 1976).
T (K )
Ratio 1479.5 1548 1600 1683 1738 1768
XFe, melt / XFe, olivine (a) 1.0 1.21 1.39 1.70 1.93 2.06
XMg, melt / XMg, olivine (b) – 0.23 0.27 0.34 0.40 0.43
Figure 6.9 Melting relations in Mg2SiO4-Fe2SiO4 system. Experimental data confirm va-
lidity of Richardson’s (1956) ideal model for liquid. Reprinted from B. J. Wood and D. G.
Fraser, Elementary Thermodynamics for Geologists, 1976, by permission of Oxford Univer-
sity Press.
(1 − X Fe, melt ) = b(1 − X Fe,olivine ), (6.66)
we obtain
X Fe, melt =
(
a 1− b ) (6.67)
a−b
and
1− b
X Fe,olivine = . (6.68)
a−b
Figure 6.9 shows how the calculated phase boundaries compare with the ex-
perimental observations of Bowen and Schairer (1935) on the same binary join.
The satisfactory reproduction of phase assemblages clearly indicates that the
Mg2SiO4-Fe2SiO4 melt is effectively ideal, in agreement with Richardson’s

(1956) assumptions.
6.6 Thermodynamic Properties of Complex

Silicate Melts
We now examine the thermodynamic properties of natural melts, after a brief

digression on the differences between vitreous and molten states, which is neces-
sary for better understanding of the properties of interest.
6.6.1 Relationships Between Molten and Vitreous States
If we rapidly quench a silicate melt, we may obtain a glass without the formation
of any crystalline aggregate. The temperature of transition between melt and glass
may be operatively defined as “that temperature at which the differing thermody-
namic properties of melt and glass intersect” (Berman and Brown, 1987). The
supercooling phenomenon, which may easily be observed in melts with molar
amounts of silica higher than 50%, involves discontinuities in the values of heat
capacity, enthalpy, entropy, and Gibbs free energy. Moreover, the various thermo-
dynamic properties for the vitreous state (and for the transition temperature) vary
with the quenching rate. Figure 6.10 shows, for instance, the enthalpy and heat
capacity for a CaMgSi2O6 melt compared with the corresponding properties in
the crystalline and vitreous states at various T. Note in figure 6.10A that the
dH/dT slope is different for vitreous and molten states. Moreover, the glass-melt
transition is observed at two different temperatures, depending on quenching rate
(T ⫽ 1005 or 905 K), and the quenching rate also affects the enthalpy of the
vitreous state, which is not the same in the two cases. The enthalpy of the crystal-
line counterpart in all T conditions is lower than the molten and vitreous states.
At the temperature of fusion Tf , the enthalpy difference between crystal and melt
is the enthalpy of fusion of CaMgSi2O6 (⌬H fusion ), whereas the enthalpy differ-
ences at lower T are enthalpies of solution of pure diopside in a pure CaMgSi2O6
melt (⌬H solution ) and enthalpies of vitrification (⌬H vitrification ). Note also in figure
6.10B that the heat capacity of the vitreous state (CPv ) is very similar to the heat
capacity of the crystal (CPc ) and that both, T being equal, are notably lower than
the heat capacity of the molten state (CPf ).
Lastly, we observe that the glass-melt transition involves a marked discontinu-
ity at T of transition, reflecting the increase of vibrational freedom (rotational
components appear). This discontinuity involves some complexity in the extrapo-
lation of calorimetric data (usually obtained on glasses) from vitreous to molten
states, discussed in detail by Richet and Bottinga (1983, 1986).
Figure 6.10 Temperature dependence of enthalpy (A) and heat capacity (B) for CaMg-
Si2O6 component in crystalline, vitreous, and molten states at various T conditions. Re-
printed with permission from Richet and Bottinga (1986), Review of Geophysics and Space
Physics, 24, 1–25, copyright 䉷 1986 by the American Geophysical Union.
6.6.2 Heat Capacity of Silicate Melts

We have already seen (section 6.5) that the mixing behavior of melt components
with identical amounts of silica is essentially ideal. Ideal mixing implies that ex-
tensive properties of the melt, such as heat capacity at constant pressure CP, is a
linear function of the molar properties of the end-members—i.e.,
Table 6.9 Parameters for calculation of heat capacity of silicate melts: (1) after Carmichael et al.
(1977) and Stebbins et al. (1984); (2) model of Richet and Bottinga (1985)
Constituent Oxide
Model Parameter SiO2 TiO2 Al2O3 Fe2O3 FeO MgO CaO Na2O K2O
(1) K1 80.0 111.8 157.6 229.0 78.9 99.7 99.9 102.3 97.0
⫾0.9 ⫾5.1 ⫾3.4 ⫾18.4 ⫾4.9 ⫾7.3 ⫾7.2 ⫾1.9 ⫾5.1
(2) K1 81.37 75.21 27.21 199.7 78.94 85.78 86.05 100.6 50.13
K2 ⫻ 103 0.0 0.0 94.28 0.0 0.0 0.0 0.0 0.0 15.78
K3 ⫻ 10⫺5 0.0 875.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0
(1) CP 冢moleJ⫻ K冣 ⫽ ∑ K
i
1,iX i ; (2) CP 冢moleJ⫻ K冣 ⫽ ∑ X (K i
i 1,i ⫹ K2,i T ⫹ K3,i T ⫺2) ⫹ CP,excess
C P , melt = ∑ Xi CP0,i , ( 6.69 )

i
where Xi is the molar fraction of component i in the molten mixture and C 0P,i is
its molar specific heat at the standard state of pure melt. On the basis of calori-
metric measurements carried out on natural and synthetic molten mixtures in the
T range 1000 to 1600 K, Carmichael et al. (1977) suggested that the ideal model
postulated for binary mixtures may be extended to multicomponent systems.
They also observed that the heat capacity of the various components in the mix-
ture may be considered independent of T, within the limits of experimental incer-
titude, through the suitable choice of standard state values (i.e., a single term in
Maier-Kelley-type functions; cf. section 3.2). However, further investigations by
Stebbins et al. (1984) and Richet and Bottinga (1985, 1986) showed significant
deviations from the ideal mixing model. For instance, in the pseudoternary system
CaAl2Si2O8-NaAlSi3O8-CaMgSi2O6 , positive deviations from ideality occur
along the NaAlSi3O8-CaMgSi2O6 join and negative ones in CaMgSi2O6-
CaAl2Si2O8 molten mixtures.
Table 6.9 lists the parameters of the model of Carmichael et al. (1977) for the
heat capacity of silicate melts, recalibrated by Stebbins et al. (1984) and valid for
the T range 1200 to 1850 K. To obtain the heat capacity of the melt at each T
condition, within the compositional limits of the system, it is sufficient to combine
linearly the molar proportions of the constituent oxides multiplied by their re-
spective CP values (cf. equation 6.69).
Table 6.9 also lists the parameters of Richet and Bottinga’s (1985) model, which,
devised primarily for Al-free melts, accounts for the nonideality of molten sili-
cate systems containing K by introducing an excess term that represents Si-K inter-
actions:
CP, melt = ∑ X iCP,i + CP, excess ( 6.70 )

i
 J 
CP, excess = WSi − K X SiO2 X K2 O ; W Si − K = 151.7  . ( 6.71)
2  mol e × K 
The model of Richet and Bottinga (1985) considers the heat capacity of the melt to
be variable with T and allows better reproducibility of experimental evidence in composi-
tionally simple systems. The Stebbins-Carmichael model seems to reproduce the experi-
mental observations on aluminosilicate melts better (Berman and Brown, 1987). Applica-
tion of both models to natural melts gives substantially identical results (i.e., differences
of about 1%, within the range of data uncertainty; cf. Berman and Brown, 1987).
6.6.3 Relationships Between Enthalpy-Entropy of Fusion and

Enthalpy-Entropy of Formation of Melt Components
at Standard State
Although in crystalline phases determination of the enthalpy of formation from
the constituent elements (or from constituent oxides) may be carried out directly
through calorimetric measurements, this is not possible for molten components.
If we adopt as standard state the condition of “pure component at T ⫽ 298.15
K and P ⫽ 1 bar”, it is obvious that this condition is purely hypothetical and not
directly measurable. If we adopt the standard state of “pure component at P and
T of interest”, the measurement is equally difficult, because of the high melting
temperature of silicates.
The molar enthalpy of molten components in the standard reference condi-
tions of T ⫽ 298.15 K and P ⫽ 1 bar is usually obtained indirectly, by adding
first the molar enthalpy of fusion to the molar enthalpy of the crystalline compo-
nent at its melting point (see also figure 6.10):
H melt ,Tf = H crystal,Tf + ∆ H fusion . ( 6.72 )
Once the enthalpy of fusion is known, because solid and liquid are at equilibrium
at the melting point, the entropy of the molten component at Tf is also known
through
∆G fusion = 0 ⇒ ∆ S fusion = ∆ H fusion Tf . ( 6.73)
If the heat capacity of the molten component is known (cf. section 6.6.2), molar
enthalpy and molar entropy at the standard state of the pure melt at T ⫽ 298.15
and P ⫽ 1 bar may be readily derived by applying
Tf
0
H melt , 298.15 K ,1 bar = H melt,T f − ∫ C P , melt dT ( 6.74 )
298.15
Table 6.10 Parameters for calculation of molar volume of silicate melts at various T
conditions (after Stebbins et al., 1984). Resulting volume is in cm3/mole.
Component ai bi ⫻ 103 Component ai bi ⫻ 103
SiO2 26.27 ⫺0.42 MgO 12.39 0.21

TiO2 24.22 0.63 CaO 17.68 0.15
Al2O3 37.89 0.38 Na2O 30.54 0.18
Fe2O3 44.93 0.22 K 2O 48.92 0.41
FeO 14.50 0.19
and
Tf
dT
0
S melt , 298.15K ,1 bar = S melt,T f − ∫ C P, melt
T
. (6.75)
298.15
Because H fusion is difficult to measure as a result of the high value of Tf , it may

be derived indirectly through calculations involving the vitreous state (see Berman
and Brown, 1987) or through the Clausius-Clapeyron equation for the crystal-
melt equilibrium (cf. equation 6.48 and section 6.3).
6.6.4 Molar Volume, Thermal Expansion, and

Compressibility of Silicate Melts
Like heat capacity, the molar volume of chemically complex silicate melts may
also be obtained through a linear combination of the molar volumes of molten
oxide components—i.e.,
Vmelt = ∑ Xi Vi 0 . ( 6.76 )
i
Bottinga and Weill (1970) applied equation 6.76 extensively, and it was later recal-
ibrated by Stebbins et al. (1984). In the latter model, the molar volume of the
melt and its modification with T are calculated through the equation
Vmelt,T = ∑ a i X i + (T − 1873) ∑ bi X i , (6.77)

i i
whose coefficients are listed in Table 6.10. Equation 6.77 also implies that thermal
expansion is a simple additive function of the linear expansions of the various
oxide components.
Concerning melt compressibility, existing experimental data are rather scanty.
On the basis of solid/liquid univariant equilibria for six silicate components
(Mg2SiO4 , MgSiO3 , CaMgSi2O6 , Mg3Al2Si3O12 , NaAlSi2O6 , and NaAlSi3O8 )
Table 6.11 Isothermal bulk modulus and first derivative

for silicate melt components (after Bottinga, 1985).
Component Treference K0 (Mbar) K⬘
Mg2SiO4 2494 0.58815 3.75

MgSiO3 2067 0.20555 9.79
CaMgSi2O6 1970 0.24218 5.31
Mg3Al2Si3O12 2290 0.11905 23.33
NaAlSi2O6 1735 0.04096 23.97
NaAlSi3O8 1515 0.42502 20.86
Figure 6.11 Compressibilities of silicate melt components (logarithmic scale) as a func-

tion of P. Values in parentheses give T of reference. Dashed area: compressibility of basal-
tic liquids according to Stolper et al. (1981). Reprinted from Bottinga (1985), with kind
permission from Elsevier Science Publishers B.V, Amsterdam, The Netherlands.
and existing thermodynamic data, Bottinga (1985) calculated isothermal com-

pressibility for the corresponding melts in the P range 0 to 150 kbar. This calcula-
tion, which is based on the Simon equation (eq. 6.49), gives the bulk moduli listed
in table 6.11 and valid for the Birch-Murnaghan equation of state.
The calculations of Bottinga (1985) show that the compressibility of silicate
melts decreases rather abruptly with increasing P. This fact, which has profound
implications in petrogenesis, is clearly shown in figure 6.11. Note particularly the
rapid decrease in compressibility for the liquid pyrope component Mg3Al2Si3O12
in the range 40 ⬍ P (kbar) ⬍ 60, and for the jadeite component NaAlSi2O6 in
the range 29 ⬍ P ⬍ 43. The compressibilities of the corresponding solids are
lower, and the first derivative on P of their bulk moduli is also generally lower
(compare tables 6.11 and 5.11, 5.14, 5.22, 5.35). Because at 1 bar the density of
silicate melts is lower than that of the corresponding solids, the calculations of
Bottinga (1985) imply that the density contrast decreases with increasing P. This
fact would tendencially obstaculate the separation of molten portions from the
residual solid, with increasing depth within the earth’s crust and mantle.
6.7 Miscibility of Natural Melts:

Ghiorso-Carmichael Model
The existence of miscibility gaps in silicate melts has been experimentally known
for decades (Greig, 1927; Bowen, 1928; Bowen and Schairer, 1938; Schairer and
Bowen, 1938; Roedder, 1951, 1956; Holgate, 1954) and has been directly observed
both in terrestrial rocks (Anderson and Gottfried, 1971; Ferguson and Currie,
1971; De, 1974; Gelinas et al., 1976; Philpotts, 1979; Coltorti et al., 1987) and in
lunar basalts (Roedder and Wieblen, 1971).
Figure 6.12 combines experimental evidence and natural observations in the
form of a ternary SiO2 - (CaO⫹MgO⫹FeO⫹TiO2 ) - (Na2O⫹K2O⫹Al2O3 ) plot.
The glassy portions of lunar basalts have compositions that conform quite well
to the experimental miscibility gap of Gelinas et al. (1976), whereas terrestrial
volcanites have smaller gaps.
Immiscibility phenomena in silicate melts imply positive deviations from ide-
ality in the mixing process. Ghiorso et al. (1983) developed a mixing model appli-
cable to natural magmas adopting the components listed in table 6.12. Because
all components have the same standard state (i.e., pure melt component at the T
and P of interest) and the interaction parameters used do not vary with T, we are
dealing with a regular mixture of the Zeroth principle (cf. sections 2.1 and 3.8.4):
1
G melt, P,T = ∑ ni µ i0 + RT ∑ ni ln X i +
2
n total ∑ ∑ Wij X i X j , (6.78)
i i i j
where ␮0i is the standard state chemical potential of the pure ith melt component
at the P and T of interest, ni is the number of moles of the ith component, Xi is
the molar ratio of the ith component, and Wij is the interaction parameter between
components i and j.
In applying the model above (which is an extension of the previous version of Ghiorso
and Carmichael, 1980, integrated with new experimental observations), some caution
must be adopted, because of the following considerations:
1. If the heat capacity of a chemically complex melt can be obtained by a linear

summation of the specific heat of the dissolved oxide constituents at all T (i.e.,
Stebbins-Carmichael model), the melt is by definition ideal. The addition of ex-
cess Gibbs free energy terms thus implies that the Stebbins-Carmichael model
calculates only the ideal contribution to the Gibbs free energy of mixing.
Figure 6.12 Miscibility gap in natural and synthetic silicate melts. Dots: unmixing for a
tholeiitic magma, based on model of Ghiorso et al. (1983). ∑ FeO: total iron as FeO.
Table 6.12 Melt components in Ghiorso-Carmichael model. Calculation first involves

computation of molar fractions of various oxides, and then their combination.
Si4O8 ⫽ 0.25 ⫻ [SiO2⫺0.5(FeO⫹MnO⫹MgO⫹NiO⫹CoO⫹CaO)⫺Na2O⫺K2O]

Ti4O8 ⫽ 0.25 ⫻ TiO2
Al16/3O8 ⫽ 0.375 ⫻ Al2O3
Fe16/3O8 ⫽ 0.375 ⫻ Fe2O3
Cr16/3O8 ⫽ 0.375 ⫻ Cr2O3
Fe4Si2O8 ⫽ 0.25 ⫻ FeO
Mn4Si2O8 ⫽ 0.25 ⫻ MnO
Mg4Si2O8 ⫽ 0.25 ⫻ MgO
Ni4Si2O8 ⫽ 0.25 ⫻ CoO
Ca4Si2O8 ⫽ 0.25 ⫻ CaO
Na16/3Si8/3O8 ⫽ 0.375 ⫻ Na2O
K16/3Si8/3O8 ⫽ 0.375 ⫻ K2O
P16/5O8 ⫽ 0.625 ⫻ P2O5
Sr8O8 ⫽ 0.125 ⫻ SrO
H 2O ⫽ 1 ⫻ H2O
2. Calculation of the excess Gibbs free energy of mixing (third term on right side of
eq. 6.78) involves only binary interactions. Although there is no multiple interac-
tion model that can be reduced to the simple summation of binary interactions
used here (cf. Acree, 1984; see also section 3.10), this choice is more than ade-
quate for the state of the art, which does not allow precise location of the miscibil-
ity gap in the chemical space of interest.
3. Because the excess Gibbs free energy of mixing is independent of P, the volume,
compressibility, and thermal expansion of the melt are obviously regarded as lin-
ear combinations of the values of its components (see section 6.6.4). As table
6.12 shows, the model uses a set of ad hoc components. The choice of components
is dictated by simple convenience—i.e., the possibility of deriving realistic ther-
modynamic data for melt constituents through univariant equilibria involving
their crystalline counterparts or through additive procedures.
The first step in the model is a normative calculation (table 6.12) to establish
the molar amounts of the various melt components. For each component, the
molar Gibbs free energy at the standard state of pure component at T and P of
interest is given by
P
Gi 0,P ,T , melt = H i0,P
r ,T , melt
− TSi 0,P
r ,T , melt
+ ∫ Vi ,0P ,T , melt dP , ( 6.79 )
Pr
where
Tf
H i0,P ,T , melt
r
= H i0,P ,T , crystal
r r
+ ∫ CP ,i , crystal dT
Tr (6.80)
+ ∆H i , fusion + CP ,i , melt T − T f ( )
and
Tf
dT
Si 0,P
r ,T , melt
= Si 0,P
r ,T r , crystal
+ ∫ CP ,i , crystal T
Tr
(6.81)
T 
+ ∆Si , fusion + CP ,i , melt ln  .
 Tf 
Tables 6.13 and 6.14 list sets of internally consistent thermodynamic data for
crystalline and liquid components, for use in equations 6.79 to 81. Although equa-
tions 6.80 and 6.81 are simply based on the Clausius-Clapeyron approach, Ghi-
orso et al. (1983) use a semiempirical formulation for the volume of melt compo-
nents. Its development is
Table 6.13 Thermodynamic parameters of crystalline components in Ghiorso-Carmichael model.

Heat capacity function: C 0P ⫽ K 1 ⫹ K 2 T ⫹ K 3 T ⫺ 2 ⫹ K 4 T ⫺ 1 / 2 .
C P0
H i0, Pr , Tr , solid S i0, Pr , Tr , solid
Component (kcal/mole) (cal/(mole ⫻ K)) K1 K2 ⫻ 10⫺3 K3 ⫻ 105 K4
Si4O8 ⫺867.448 48.61 69.556 1.2432 ⫺39.503 0

Ti4O8 ⫺909.543 48.08 60.304 10.810 ⫺9.4289 ⫺5.3692
Al16/3O8 ⫺1069.406 32.45 100.293 0.45827 ⫺12.090 ⫺629.73
Fe16/3O8 ⫺519.023 80.99 ⫺698.32 173.79 ⫺65.259 21644.0
Cr16/3O8 ⫺723.200 51.73 75.856 0.60525 ⫺9.1899 ⫺2.1698
Fe4Si2O8 ⫺704.730 70.90 73.020 18.720 ⫺1.3400 0
Mn4Si2O8 ⫺826.038 78.02 78.510 7.4953 ⫺18.229 0
Mg4Si2O8 ⫺1036.701 45.50 108.977 1.6319 ⫺4.2733 ⫺833.94
Ca4Si2O8 ⫺1109.070 36.44 98.0 0 0 0
Na16/3Si8/3O8 ⫺987.623 72.56 83.040 25.600 ⫺17.253 0
K16/3Si8/3O8 ⫺990.064 93.15 79.280 44.933 ⫺9.5467 0
P16/5O8 ⫺575.488 44.18 13.400 86.402 0 0
Sr8O8 ⫺1129.048 106.16 106.44 10.443 ⫺5.3164 ⫺298.14
Table 6.14 Thermodynamic parameters of melt components in Ghiorso-Carmichael model

(⌬H 0fusion ⫽ Tf ⌬S 0fusion).
V i0, melt
(cal/(mole ⫻ bar))
Tf ⌬ S 0 fusion ⌬ H 0 fusion 0
C P, i , melt
Component (K) (cal/(mole ⫻ K)) (kcal/mole) (cal/(mole ⫻ K)) a V, i b V, i ⫻ 10 4
Si4O8 1996 3.908 7.800 83.16 2.604 ⫺0.1294

Ti4O8 2143 39.20 84.006 106.96 1.202 5.740
Al16/3O8 2327 32.08 74.650 65.71 1.767 3.397
Fe16/3O8 1895 34.80 65.946 122.27 2.173 3.659
Cr16/3O8 2603 31.76 82.671 100.00 1.854 -
Fe4Si2O8 1490 29.57 44.059 114.60 1.920 2.834
Mn4Si2O8 1620 26.46 42.865 116.20 2.584 -
Mg4Si2O8 2163 37.61 81.350 128.06 2.238 0.8797
Ca4Si2O8 2403 20.81 50.006 118.96 1.859 5.923
Na16/3Si8/3O8 1362 24.24 33.015 113.01 2.859 4.170
K16/3Si8/3O8 1249 25.62 31.999 114.67 3.456 7.138
P16/5O8 853 10.68 9.110 93.60 2.272 -
Sr8O8 2938 49.02 144.021 128.00 3.955 -
Vi ,0P = Vi ,0T = 0 + aV ,i T + bV ,i T 2 ( 6.82 )

r ,T , melt
∫ Vi ,0P ,T , melt dP =
Pr
(6.83)
     
( 2
)
1 1
2
(
 6
1
)
= Vi ,0P ,T , melt  P − Pr − cV ,i  P 2 − Pr2 − cV ,i P 2   P 3 − Pr3  ,

( )

r
where aV,i and bV,i are the thermal expansion coefficients listed in table 6.14 and
cV,i is a compressibility factor of the form
1 × 10 −6
cV ,i = . ( 6.84 )
2.76 0
0.7551 + Vi ,P ,T , melt
ni r
Equation 6.83 is not valid for component Si4O8 , for which the following equation
is used:
∫ VSi0 O ,P ,T , melt dP = VSi0 O ,P ,T , crystal (P − 1)

4 8 4 8 r
Pr (6.85)
(
− 4.204 × 10 −6 P 2 − Pr2 . )
Once the standard state potentials at the P and T of interest have been calcu-
lated ( ␮0i ⬅ G i0 for a pure single-component phase), the ideal and excess Gibbs
free energy of mixing terms are easily obtained on the basis of the molar fractions
of the various melt components and the binary interaction parameters listed in
table 6.15 (cf. eq. 6.78).
It must be noted here that, because there are 16 melt components in the
model, there should be 120 binary interaction parameters. Only 55 of these are
listed in table 6.15, because the remaining ones are considered virtually equal to
zero. Lastly, it must be noted that the Gibbs free energy of hydrated magmas
cannot be obtained simply by application of equations 6.78 to 6.85, but requires
additional considerations that cannot be outlined in this context (see Ghiorso and
Carmichael, 1980, and Ghiorso et al., 1983, for detailed treatment of the method).
As already noted, the model can define the existence of miscibility gaps in
natural magmas. Let us imagine that we have calculated the Gibbs free energy of
mixing of a magmatic series with increasing amounts of component i:
G mixing,melt = G ideal mixing + G excess mixing . (6.86)
We can then derive the calculated Gibbs free energy of mixing with respect to the
molar amount of the component of interest, thus obtaining the difference between
the chemical potential of the component in the mixture and its chemical potential
at standard state:
 ∂G mixing,melt 
  = RT ln a i , P,T , melt = µ i , P,T , melt − µ i0, melt . (6.87)
 ∂ X i  P, T , X
j
Table 6.15 Binary interaction parameters in Ghiorso-Carmichael model (cal/mole). Si4O8 ⫽ 1;

Ti4O8 ⫽ 2; Al16/3O8 ⫽ 3; Fe16/3O8 ⫽ 4; Fe4Si2O8 ⫽ 5; Mn4Si2O8 ⫽ 6; Mg4Si2O8 ⫽ 7; Ca4Si2O8 ⫽ 8;
Na16/3Si8/3O8 ⫽ 9; K16/3Si8/3O8 ⫽ 10; H2O ⫽ 11.
1 2 3 4 5 6 7 8 9 10
2 ⫺29,364
3 ⫺78,563 ⫺67,350
4 2,638 ⫺6,821 1,240
5 ⫺9,630 ⫺4,595 ⫺59,529 4,524
6 5,525 ⫺2,043 ⫺1,918 212 ⫺703
7 ⫺30,354 12,674 ⫺48,675 ⫺1,277 ⫺57,926 ⫺2,810
8 ⫺64,068 ⫺102,442 ⫺98,428 1,520 ⫺59,356 699 ⫺78,924
9 ⫺73,758 ⫺101,074 ⫺135,615 ⫺3,717 ⫺36,966 780 ⫺92,611 ⫺62,780
10 ⫺87,596 ⫺40,701 ⫺175,326 284 ⫺84,580 ⫺61 ⫺45,163 ⫺27,908 ⫺18,130
11 ⫺412 ⫺196 ⫺71,216 ⫺103,024 7,931 310 ⫺20,260 ⫺38,502 ⫺49,213 ⫺23,296
When the partial derivative reaches zero, the two potentials coincide and compo-
nent i is present unmixed as a pure term. If unmixing takes place in solvus condi-
tions, the thermodynamic activity of component i remains constant for the entire
solvus field. However, in the case of spinodal decomposition, the activity of i
within the spinodal field plots within a maximum and minimum (cf. sections 3.11
and 3.12).
For instance, figure 6.13 shows the behavior of the partial derivative of the
Gibbs free energy of mixing with respect to the molar amount of Si4O8 com-
ponent:
 ∂ G mixing,melt 
  = RT ln a Si 4O 8 , P,T , melt
 ∂X Si 4O 8, melt  P,T , X
j (6.88)
(
= 4 µ SiO2 , P,T , melt − µ SiO
0
2 , melt
).
On the SiO2-rich side of the diagram, spinodal decomposition (points b and d in
figure 6.13) is seen at 1700 K. Ghiorso et al. (1983) showed that the topology of
the Gibbs free energy of mixing surface in a multicomponent space is rather com-
plex, with a saddle-shaped locus of spinodal decomposition. The calculations of
Ghiorso et al. (1983) show that the immiscibility dome in a T-X space intersects
the liquidus field of a tholeiitic magma. Table 6.16 gives the composition of a
tholeiite (Rattlesnake Hill basalt) investigated by Philpotts (1979) because it was
evidently an unmixed magma. Philpotts (1979) remelted the rock and was able to
observe the unmixing phenomenon experimentally. During the cooling stage, two
liquids were produced by unmixing at 1313 °C. Analysis of the unmixed portions
indicated that one of the unmixed terms was richer in TiO2 , total iron (as FeO),
Figure 6.13 Partial derivative of Gibbs free energy of mixing of melt with respect to Si4O8
component plotted against molar amount of silica in melt. Enlarged area: spinodal field
of critical immiscibility (spinodal decomposition). From M. S. Ghiorso and I. S. E. Car-
michael, A regular solution model for meta-aluminous silicate liquids: applications to
geothermometry, immiscibility, and the source regions of basic magma, Contributions to
Mineralogy and Petrology, 71, 323–342, figure 3, copyright 䉷 1980 by Springer Verlag.
Reproduced with modifications by permission of Springer-Verlag GmbH & Co. KG.
Table 6.16 Rattlesnake Hill tholeiite (Philpotts, 1979)

and its unmixes according to Ghiorso-Carmichael
model (weight %; * ⫽ total iron as FeO).
Oxide Tholeiite Liquid 1 Liquid 2

SiO2 52.29 45.87 53.21
TiO2 1.17 0.00 1.26
Al2O3 14.75 9.06 15.28
FeO* 12.25 25.24 11.47
MnO 0.22 0.95 0.17
MgO 5.30 7.98 5.16
CaO 9.89 8.99 10.04
Na2O 2.60 0.59 2.76
K2O 0.33 0.00 0.36
P2O5 0.16 0.65 0.13
Figure 6.14 Gibbs free energy of mixing surface calculated for Archean basalts of Piumhi
(Minas Gerais, Brazil) in SiO2-Na2O compositional space.
MnO, MgO, CaO, and P2O5 , and the other was richer in SiO2 , Na2O, and K2O.
Table 6.16 lists the compositions of these unmixes according to the Ghiorso-
Carmichael model (also plotted in the ternary diagram of figure 6.12): the two
liquids have the same compositional features described in the experiment of Phil-
potts (1979).
Figure 6.14 shows in stereographic form the values of the Gibbs free energy
of mixing obtained for the Archean basalts of Piumhi (Minas Gerais, Brazil) on
the basis of the data of Coltorti et al. (1987): note the pronounced saddle defining
the spinodal decomposition field (unmix compositions also plotted in ternary
diagram of figure 6.12).
The capability of evaluating with sufficient accuracy the Gibbs free energy of
a silicate melt in various P-T-X conditions has obvious petrogenetic implications,
besides those already outlined. For instance, the P-T loci of equilibrium of a
given crystal with the melt can be determined with good approximation. Let us
consider, for example, the equilibrium
Mg 2 SiO 4 ⇔ Mg 2 SiO 4 .
olivine melt
(6.89)
Figure 6.15 Examples of thermometric calculations based on Ghiorso-Carmichael

model for silicate liquids. From M. S. Ghiorso, I. S. E. Carmichael, M. L. Rivers, and
R. O. Sack, The Gibbs free energy of mixing of natural silicate liquids: an expanded regu-
lar solution approximation for the calculation of magmatic intensive variables, Contribu-
tions to Mineralogy and Petrology, 84, 107–145, figures 1–2, copyright 䉷 1983 by Springer
Verlag. Reprinted with the permission of Springer-Verlag GmbH & Co. KG.
For the pure component, reaction 6.89 has a constant (K89 ) whose value under
various P-T conditions gives the activity ratio of the magnesian component in
the two phases:
 ∆G890  aMg 2 SiO4 , melt

K 89 = exp  −  = . ( 6.90 )
 RT  aMg 2 SiO4 ,olivine
For a given value of P, there is only one T condition at which equation 6.90
is satisfied. Hence, once the activity of the component in the two phases is known,
the P-T loci of equilibrium are also known. This concept can be generalized for
solid phases whose crystalline components (M) are stoichiometric combinations
of melt components (C) according to
n
M solid = ∑ viC i , (6.91)
i =1
where the vi terms are stoichiometric factors. The generalized calculation is

n
RT ln K = ∑ vi RT ln a i , melt − RT ln a i ,solid . (6.92)
i =1
Figure 6.15 gives examples of solid-liquid geothermometry relative to the olivine-

liquid (A), orthopyroxene-liquid (A), and plagioclase-liquid (B) pairs. There is
satisfactory agreement between experimental and calculated values, and the ther-
mometric procedure may thus be applied with confidence.
CHAPTER SEVEN
Introduction to Petrogenetic Systems
7.1 Phase Stability Relations in Binary Systems

(Roozeboom Diagrams)
Knowledge of mineral-melt equilibria in magmas is based on experimental inves-

tigation of multicomponent synthetic systems, the simplest of which are binary
joins. The interpretation of phase stability fields experimentally defined by P-T-
X diagrams is straightforward, being based on equilibrium thermodynamics. The
principles ruling phase stability relations are in fact only two: (1) minimization of
the Gibbs free energy of the system at equilibrium, and (2) equality of chemical
potentials of components in coexisting phases at equilibrium. From binary joins
the extension to more complex systems is also straightforward, based on the prin-
ciples listed above and on mass conservation.
Figure 7.1 shows the Gibbs free energy curves for three distinct phases that
differ in mixing behavior. Phase ␣ shows complete miscibility of components 1
and 2 in all proportions; the Gibbs free energy curve has a single downward-open
concavity. The Gibbs free energy of the phase at each composition is defined by
the sum of the chemical potentials of components (we recall that the chemical
potential is in fact the partial derivative of the Gibbs free energy of the phase with
respect to the number of moles of the component of interest):
 ∂ Gα 
µ1,α =   ( 7.1)
 ∂ n1,α  P ,T
 ∂ Gα 
µ 2 ,α =   ( 7.2 )
 ∂ n2 ,α  P ,T
Gα = ∑ ni , α µ i , α ( 7.3)
i
/ 449
Figure 7.1 Relationships between chemical potential and composition in binary phases
with different miscibility behavior: ␣ ⫽ complete miscibility; ␤ ⫽ partial miscibility; ␥ ⫽
lack of miscibility or “mechanical mixture.”
Gα = ∑ X i ,α µ i ,α = X 1,α µ1,α + X 2 ,α µ 2 ,α . ( 7.4 )

i
In equation 7.4, G ␣ is the molar Gibbs free energy of phase ␣ at the composi-
tional point defined by fractional amounts X1,␣ and X2,␣. The Gibbs free energy
curve of phase ␤ is saddle-shaped, indicating the existence of a miscibility gap.
Applying the principles of minimization of Gibbs free energy and of equality of
potentials at equilibrium, we can deduce that the compositions of the two un-
mixed portions ␤⬘ and ␤⬙ are determined by the points of tangency (and identi-
cal slope):
µ1, β ′ = µ1, β ′′ ; µ 2 , β ′ = µ 2 , β ′′ ( 7.5)
G β ′ = X 1, β ′ µ1, β ′ + X 2, β ′ , µ 2, β ′ (7.6)
Introduction to Petrogenetic Systems / 451
G β ′′ = X 1, β ′′ µ1, β ′′ + X 2, β ′′ , µ 2, β ′′ . (7.7)
Denoting X␤⬘ and X␤ ⬙ the molar fractions of the unmixed portions, we also have
G β = X β ′G β ′ + X β ′′G β ′′ . (7.8)
The Gibbs free energy of phase ␥ is represented by a straight line connecting

the standard state potentials of the two end-members in the mixture. Because we
use the term “mixture,” it is evident that the standard state of both end-members
is the same and is that of “pure component.” The two components are totally
immiscible in any proportion and the aggregate is a “mechanical mixture” of the
two components crystallized in form ␥:
G γ = X 1,γ µ10,γ + X 2,γ µ 20,γ . (7.9)
Recalling the general formulation of the Gibbs free energy of a mixture:
G mixture = ∑ ni µ i0 + G ideal mixing + G excess mixing , (7.10)

i
it is evident that phase ␣ has an excess Gibbs free energy sufficiently low (or
negative) to avoid the formation of a solvus. In the case of phase ␤, Gexcess mixing
is positive and sufficiently high to induce a zone of immiscibility. In the case of
phase ␥, the terms Gideal mixing and Gexcess mixing are mutually compensated over the
whole compositional range.
7.1.1 Crystallization of Phases with Immiscible Components

Let us call the melt phase ␣ and the solid phase with complete immiscibility of
components ␥. P is constant and fluids are absent. The Gibbs free energy relation-
ships at the various T for the two phases at equilibrium are those shown in figure
7.2, with T decreasing downward from T1 to T6. The G-X relationships observed
at the various T are then translated into a T-X stability diagram in the lower part
of the figure.
At T1, the Gibbs free energy of phase ␣ (i.e., melt) at all compositions is lower
than that of mechanical mixture ␥⬘ ⫹ ␥ ⬙: phase ␣ is then stable over the whole
compositional range. At T2, the chemical potential of component 1 in ␣ is identi-
cal to the chemical potential of the same component in ␥⬘. Moreover, the equality
condition is reached at the standard state condition of the pure component: T2 is
thus the temperature of incipient crystallization of ␥⬘. At T3, the Gibbs free energy
of ␣ intersects mechanical mixture ␥⬘ ⫹ ␥ ⬙ on the component 1–rich side of the
diagram and touches it at the condition of pure component 2. Applying the prin-
Figure 7.2 G-X and T-X plots for a binary system with a molten phase with complete
miscibility of components at all T conditions and a solid phase in which components are
totally immiscible at all proportions (mechanical mixture, ␥ ⫽ ␥⬘ ⫹ ␥⬙).
ciples of equality of potentials at equilibrium and Gibbs free energy minimization,

we can deduce that the zone of coexistence of ␣ and ␥⬘ is given by the straight
segment originating at ␮0␥⬘ and tangent to ␣. Moreover, T3 is the temperature of
incipient crystallization of ␥ ⬙. With decreasing T, the Gibbs free energy curve of
␣ is progressively moves upward (i.e., is less negative) and, consequently, the two-
phase fields identified by the tangency zones in the potential vs. composition plot
progressively expand (see condition T4 in figure 7.2). At T5, we reach a peculiar
T-X condition, the eutectic point, where the Gibbs free energy of the mechanical
mixture is identical (at the single point of tangency) to that of the melt (phase
␣ ). At this point, the three phases ␣, ␥⬘, and ␥ ⬙ coexist stably at equilibrium. Note
that, because there are two components in the system, based on the Gibbs phase
rule the equilibrium is invariant: i.e., the eutectic point is defined at a single eutec-
tic temperature. (The term “eutectic” was coined by Guthrie, 1884, to identify
the lowest temperature of melting of the system.) At T6, the Gibbs free energy of
the mechanical mixture is lower than that of phase ␣ over the entire binary range:
only the mechanical mixture ␥⬘ ⫹ ␥ ⬙ is stable at equilibrium.
Let us now consider the cooling process of a system of composition C in
figure 7.2. At high T, the system is completely molten (phase ␣, condition T1 ). As
T is progressively lowered, condition Tc is reached, at which ␥⬘ begins to crystal-
lize. The crystallization of ␥⬘ enriches the composition of the residual melt of
component 2. At T4, the melt has composition C⬙, while crystals ␥⬘ continue to
form. The molar proportions of crystals and melts may be deduced by application
of baricentric coordinates at temperature T4 (relative lengths of the two composi-
tional segments in the two-phase field ␣ ⫹ ␥⬘, or “lever rule”). We reach eutectic
point E at T5: phase ␥ ⬙ forms at equilibrium in stable coexistence with ␣ and ␥⬘;
the temperature now remains constant during the crystallization of ␥ ⬙, until all
the residual melt is exhausted.
At T6, the melt is absent and ␥⬘ and ␥ ⬙ coexist stably in a mechanical mixture
of bulk composition C. The crystallization process in a closed system is thus com-
pletely defined at all T (and P) conditions by the Gibbs free energy properties of
the various phases that may form in the compositional field of interest.
7.1.2 The Diopside-Anorthite System as an Example

Figure 7.3 depicts phase stability relations in the pseudobinary system
CaMgSi2O6-CaAl2Si2O8 (diopside-anorthite). The original study of Bowen (1915)
described crystallization behavior identical to the previously discussed case: a
mechanical mixture (Di-An) in equilibrium with a completely miscible melt. A
later investigation (Osborn, 1942) showed that the system is not strictly binary
Figure 7.3 CaMgSi2O6-CaAl2Si2O8 (diopside-anorthite) system, after Bowen (1915) and

Osborn (1942). Abscissa axis: percentile weight of both components.
(i.e., cannot be defined simply on the basis of the two components CaMgSi2O6
and CaAl2Si2O8 ), but is rather “pseudobinary,” because a limited amount of
Al2O3 may be dissolved in CaMgSi2O6 . However, we will neglect this limited solu-
bility here and will hereafter treat the system as purely binary.
Pure components CaMgSi2O6 and CaAl2Si2O8 at P ⫽ 1 bar melt, respectively,
at Tf,Di ⫽ 1391.5 °C and Tf,An ⫽ 1553 °C. The eutectic temperature is 1274 °C.
The progressive decrease of incipient crystallization, from 1553 °C to 1274 °C on
the An-rich side of the join and from 1391.5 °C to 1274 °C on the Di-rich side, is
a result of the lowering of the freezing point expression (see chapter 6: eq. 6.56,
section 6.4). If we cool a system of initial composition C, we observe incipient
crystallization at T ⫽ 1445 °C (figure 7.3). At this temperature, the entire system
is still in the liquid form. At T ⫽ 1400 °C, some anorthite crystals are already
formed. We can determine their weight amounts by applying baricentric coordi-
nates (or the “lever rule”) to compositional segments An and L in figure 7.3—i.e.,
An 1
X An = = = 0.20 ( 7.11)
An + L 1 + 4
L 4
XL = = = 0.80. ( 7.12 )
An + L 1 + 4
Note that the resulting fractional amounts are in weight percent, because the
abscissa axis of the phase diagram reports the fractional weights of the two com-
ponents (similar application of baricentric coordinates to a molar plot of type
7.2 would have resulted in molar fractions of phases in the system). Applying the
lever rule at the various T, we may quantitatively follow the crystallization behav-
ior of the system (i.e., at T ⫽ 1350 °C, XAn 艐 0.333 and XL 艐 0.666; at T ⫽ 1300
°C, XAn 艐 0.44 and XL 艐 0.56; and so on). It is important to recall that the
baricentric coordinates lose their significance at the eutectic point, because solid/
melt proportions depend on time and on the dissipation rate of the heat content
of the system. Below the eutectic temperature, the lever rule can again be applied
to the couple Di-An, and the resulting proportions are those of the initial system
C (XAn ⫽ 0.70, XDi ⫽ 0.30). For a closed system (see section 2.1), the melting
behavior of an initial composition C exactly follows the reverse path of the crys-
tallization process (i.e., beginning of melting at T ⫽ 1274 °C/displacement along
univariant curve/abandonment of univariant curve at T ⫽ 1450 °C).
7.1.3 Crystallization of a Binary Phase with Complete

Miscibility of Components (Roozeboom Types I, II,
and III)
The intersections of the Gibbs free energy curves of the melt with those of the
solid mixture of components 1 and 2 at the various T (both phases having com-
Figure 7.4 Gibbs free energy curves and phase stability relations for two binary mixtures
with complete miscibility of components (types I, II, and III of Roozeboom, 1899).
plete miscibility) may give rise to three distinct configurations of the phase dia-
gram, as outlined in figure 7.4.
Roozeboom Type I
Roozeboom type I is obtained whenever the Gibbs free energy curve of the melt
initially touches that of the mixture in the condition of pure component. This
takes place at the melting temperature of the more refractory component (T2 in
figure 7.4). With decreasing T, the intersections of the Gibbs free energy curves
of the solid and liquid mixtures define tangency zones spanning the entire binary
range. The two curves finally touch at the condition of pure component 1 (T5 ):
T5 is thus the melting temperature of the less refractory component in the mixture.
Projection of the two-phase zones in a T-X space gives the phase diagram shown
at the bottom of figure 7.4: liquidus and solidus curves limit a single lens-shaped
field of stable coexistence of solid and liquid ( ␣ ⫹ L).
Roozeboom Type II
The Gibbs free energy curves of both solid mixture and melt initially touch at an
intermediate composition of the binary join (T1 in the central column of figure
7.4). With decreasing T, the Gibbs free energy loop of crystalline mixture ␣ is
progressively lowered, crossing the Gibbs free energy loop of the melt in two
points. At T3, the chemical potential of pure component 2 is identical in the solid
and liquid phases. At this temperature (that of melting of pure component 2), the
Gibbs free energy curve of the solid phase still intersects the Gibbs free energy
loop of the melt in another part of the join. At T4 (melting temperature of pure
component 1), the two Gibbs free energy curves intersect at the condition of pure
component 1. At lower T, the entire Gibbs free energy loop of the solid lies below
the Gibbs free energy curve of the melt: the entire system is crystallized. The
resulting T-X phase diagram (figure 7.4, bottom center) shows two distinct two-
phase lenses ( ␣ ⫹ L) departing from a peak on the liquidus curve.
Roozeboom Type III

This is analogous to the preceding case, but the sequence of conditions T1, T2,
and T3 of type II is inverted. With decreasing T, we reach the melting temperature
of pure component 2 (T2 ); at this T, the two Gibbs free energy loops also intersect
in another zone of the join (corresponding to condition T3 of the preceding case,
see figure 7.4, center). At T3, we observe a double intersectioning (corresponding
to T2 ) and at T4 a single point of tangency (i.e., T1 ). Projection of the tangency
zones on a T-X space leads to two two-phase lenses converging toward a mini-
mum on the solidus curve (figure 7.4, bottom right).
7.1.4 The Albite-Anorthite (Roozeboom Type I) and Albite-

Sanidine (Roozeboom Type II) Systems as Examples
Figure 7.5 shows phase stability relations in the binary system NaAlSi3O8-
CaAl2Si2O8 (albite-anorthite), based on the experiments of Bowen (1913), Greig
and Barth (1938), and Osborn (1942).
Pure anorthite (An) at P ⫽ 1 bar melts at T ⫽ 1553 °C (Osborn, 1942), and
pure albite (Ab) melts at T ⫽ 1118 °C (Greig and Barth, 1938). The phase stabil-
ity relations in the Ab-An join are Roozeboom type I. Cooling a melt of molar
Figure 7.5 Phase stability relations in binary join CaAl2Si2O8-NaAlSi3O8 (anorthite-

albite).
composition Ab33 (C in figure 7.5), we observe incipient crystallization at T ⫽

1500 °C. As crystallization advances, the residual liquid becomes more and more
enriched in the Ab component. The molar proportions of the solid and liquid
mixtures can be derived by application of the lever rule at the T of interest (seg-
ments L and Pl in figure 7.5). That is, at T ⫽ 1450 °C, the molar proportions of
the two mixtures are approximately 57% plagioclase and 43% melt, and the melt
has composition L⬘ and is in equilibrium with a solid of composition C⬘. At T ⫽
1400 °C, the proportion of liquid (L⬙ in figure 7.5) decreases to about 10% (90%
of solid with composition C⬙) and completely disappears at about 1375 °C. Below
this temperature, only the solid mixture is stable and has a molar composition
corresponding to the initial system C.
Figure 7.6 shows the phase stability relations in the NaAlSi3O8-KAlSi3O8
(Ab-Sa) system. Part A of the figure has a restricted T range, to facilitate compre-
hension of metastability phenomena on the K-rich side of the diagram. In part
B, the temperature scale is more extended, to encompass subsolidus relations.
The system is strictly binary only for T ⬍ 1078 °C. In K-rich compositions
at higher T, the sanidine solid mixture (Sas.m. ) is metastable with respect to the
leucite solid mixture (K, Na)AlSi2O6 (Les.m. ). On the Na-rich side of the join
(Abs.m. ), the crystallization trend of an initial liquid mixture of composition C1
(80% Ab by weight) is similar to that of the previous example. On the K-rich side
of the join, cooling of an initial composition C2 (50% Ab by weight) leads to
initial formation of Les.m. crystals; the residual liquid moves along the liquidus
curve until it reaches the peritectic reaction point P (T ⫽ 1078 °C). At P, Les.m.
converts into Sas.m. while T remains constant. As shown in figure 7.6B, the
NaAlSi3O8-KAlSi3O8 system in subsolidus conditions shows limited solubility
Figure 7.6 Phase stability relations in pseudobinary system NaAlSi3O8-KAlSi3O8, after

Waldbaum and Thompson (1969). (A) High-T relations: loop of metastable persistency
of sanidine like a Roozeboom type III. (B) Expanded T-range: with decreasing T, a solvus
field opens downward.
with a consolute temperature Tc ⫽ 660 °C, below which the homogeneous solid
mixture (Na, K)AlSi3O8 unmixes into an Na-rich (Abs.m. ) and a K-rich (Ors.m. )
composition. This further complexity introduces us to the third type of phase re-
lations.
7.1.5 Crystallization of Phases with Limited Miscibility

(Roozeboom Types IV and V)
There are two possible geometrical configurations between a mixture with a
saddle-shaped Gibbs free energy curve (indicative of unmixing) and a phase with
a concave curve, indicating complete miscibility of components (figure 7.7).
Roozeboom Type IV
The Gibbs free energy curve of the solid phase with partial miscibility initially
touches the curve of the liquid in the condition of pure component 1. This hap-
pens at the melting temperature of that component (i.e., T1 ⫽ Tf,1 ). With decreas-
ing T, the Gibbs free energy loop of the solid is progressively displaced downward
(i.e., becomes more negative) and the equality conditions of potentials of compo-
nents 1 and 2 in the liquid and in unmixing lobe ␤⬙ are reached first (T2, T3 ). As
T further decreases, a particular condition (T4 ) is reached: the chemical potentials
of the two components are identical in the liquid and in the two lobes ␤⬘ and ␤⬙:
the three phases ␤⬘, ␤⬙, and L coexist stably at equilibrium. At T5, the solvus of
the solid expands (tangency segment ␤⬘-␤⬙) and there still exists a solid-liquid
zone of coexistence on the component 2–rich side of the join (segment L-␤⬘). At
T6, the chemical potential of pure component 2 is identical in the solid and liquid
phases: T6 is hence the melting temperature of the less refractory component
in the join (T6 ⫽ Tf,2 ). Below this temperature, only the solid phases are stable:
␤⬘ ⫹ ␤⬙ within the solvus field, and the homogeneous mixture ␤ outside it. Projec-
tion of the various isothermal G-X geometries onto the T-X space gives the phase
diagram (figure 7.7, bottom right). The resulting phase stability relations may be
conceived as a combination of a Roozeboom type I diagram with a solvus field
intersecting it along the solidus curve.
Roozeboom Type V
Initially like Roozeboom type IV, Roozeboom type V differs from it at tempera-
ture T5, where the tangency condition of the two unmixing lobes ␤⬘ and ␤⬙ and
liquid L finds the Gibbs free energy curve of the liquid in an intermediate position
between the unmixing lobes, rather than in an external position, as in the previous
case. The resulting T-X stability relations may be conceived as a Roozeboom type
III diagram with a solvus field intersecting it along the solidus curve (cf. figures
7.6 and 7.7).
Lastly, let us consider the last column in figure 7.7, depicting a case in which
Figure 7.7 Gibbs free energy curves and T-X phase stability relations between a phase
with complete miscibility of components (silicate melt L) and a binary solid mixture with
partial miscibility of components (crystals ␤).
the compounds are completely miscible in the solid phase at high T, but become
partially immiscible with decreasing T (this is the general case, from the thermo-
chemical point of view, for the crystalline solids of our interest). Because un-
mixing begins at T5, where the Gibbs free energy of the solid is entirely above
that of the liquid phase, the solvus does not intersect the solidus line in the re-
sulting T-X plot. We thus have a Roozeboom type III diagram with a solvus
opening at subsolidus conditions, which was the case in figure 7.6B for the
NaAlSi3O8-KAlSi3O8 system.
The unmixing fields shown in figure 7.7 are solvi (i.e., defined by binodes at
the various T), but in nature we may also observe the metastable persistency of a
homogeneous phase until a critical condition of unmixing is reached. The loca-
tion of a spinodal field on a T-X diagram follows the guidelines outlined in section
3.11—i.e., the spinodes at the various T plot on the inflection point on the Gibbs
free energy curve of the solid. Spinodal fields thus occupy an inner and more
restricted portion of the T-X space with respect to the corresponding solvi.
7.1.6 Formation of Crystalline Compounds with

Fixed Stoichiometry
Let us now consider a case in which components 1 and 2 form an intermediate

crystalline compound (C) with precise, invariant stoichiometry (e.g., 60% of com-
ponent 2 and 40% of component 1). If the chemical composition of the intermedi-
ate compound is fixed, it behaves as a mechanical mixture with respect to its pure
components (i.e., zero miscibility). The presence of the intermediate compound
subdivides the compositional join into two fields: mechanical mixture 2-C ( ␥ ⬙)
and mechanical mixture C-1 ( ␥⬘). The resulting crystallization path may assume
two distinct geometrical configurations, as shown in figure 7.8.
In the first column of the figure, we see the energy conditions leading to the
formation of two eutectic points separated by a thermal barrier. At T1 ⫽ Tf,1, the
Gibbs free energy curve of the liquid touches the curve of the solid in the condi-
tion of pure component 1. At T2, the chemical potential of pure crystalline com-
ponent 1 is lower than that of pure liquid 1, and a zone of tangency is thus defined.
As T further decreases, the Gibbs free energy curve of the liquid touches interme-
diate component C: at this temperature, compound C begins to crystallize (i.e.,
T3 ⫽ Tf,C ). At T5, we observe the first invariant point: the potentials of the two
components coincide in liquid L, intermediate compound C, and pure component
1; T5 is thus the eutectic temperature for the 1-C side of the join. At T6, we observe
the second invariant condition, this time for the C-2 side of the join. The resulting
T-X phase relations may be conceived as a simple combination of two phase dia-
grams relative to crystallization of mechanical mixtures ␥ ⬙ and ␥⬘ (compare figure
7.8, bottom left, and figure 7.2).
Column 2 of figure 7.8 defines the Gibbs free energy relations leading to a
Figure 7.8 Gibbs free energy curves and T-X phase relations for an intermediate com-
pound (C), totally immiscible with pure components. Column 1: Gibbs free energy rela-
tions leading to formation of two eutectic minima separated by a thermal barrier. Column
2: energy relations of a peritectic reaction (incongruent melting). To facilitate interpreta-
tion of phase stability fields, pure crystals of components 1 and 2 coexisting with crystals
C are labeled ␥⬘ and ␥⬙, respectively, in T-X diagrams; same notation identifies mechanical
mixtures 2-C and C-1 in G-X plots.
peritectic crystallization event. At T1, the Gibbs free energy of crystalline compo-
nent 1 is identical to that of the liquid (i.e., T1 ⫽ Tf,1 ). With decreasing T (T2, T3 ),
zones of tangency between the Gibbs free energy curves of the liquid and compo-
nent 1 are observed, but the chemical potential of intermediate compound C
decreases at the same time. Lastly, at T4 ⫽ Tf,C, we reach an invariant condition
with the coexistence of three phases: compound C, liquid L, and the pure crystals
of component 1. A peritectic reaction now takes place at constant T: the pure
crystals of component 1 react with the liquid, forming intermediate compound C.
Only when pure crystals 1 are completely exhausted does the temperature again
decrease. This decrease leads to another invariant point: the eutectic point of the
system, at which the liquid coexists stably with C and the pure crystals of compo-
nent 2 (T6 ). Below this T, the liquid is no longer stable and only the two solid
phases (crystals 2 ⫹ C) are stable at equilibrium. The resulting phase diagram is
shown at the bottom of figure 7.8.
7.1.7 Examples of Formation of Intermediate Compounds

with Fixed Stoichiometry: Nepheline-Silica and
Kalsilite-Silica Joins
Figure 7.9A shows the NaAlSiO4-SiO2 (nepheline-silica) system, after Schairer
and Bowen (1956). Let us first examine the SiO2-rich side of the join. At P ⫽ 1
bar, the pure component SiO2 crystallizes in the cristobalite form (Cr) at T ⫽
1713 °C (cf. figure 2.6). At T ⫽ 1470 °C, there is a phase transition to tridymite
(Tr), which does not appreciably affect the form of the liquidus curve, which
reaches the eutectic point at T ⫽ 1062 °C.
In the central part of the diagram, two moles of SiO2 react with one mole of
NaAlSiO4 to form albite (Ab):
NaAlSiO4 + 2 SiO2 ⇔ NaAlSi 3 O8 .

nepheline albite (7.13)
NaAlSiO4 is immiscible with either SiO2 or NaAlSi3O8 (in other words, its stoichi-
ometry is fixed in the binary join of interest). When expressed in weight percent,
equation 7.13 involves 54% NaAlSiO4 and 46% SiO2 (the original diagram of
Schairer and Bowen, 1956, like most phase diagrams of that period, uses a weight
scale rather than a molar range, simply because mixtures were dosed in weight
amounts). Crystallization of any composition within NaAlSi3O8 and SiO2 follows
the classical behavior of a mechanical mixture, as discussed in section 7.1.1 (cf.
figure 7.2). For example, an initial composition C1 in figure 7.9 (corresponding
to 50% NaAlSiO4 and 50%SiO2 by weight) crystallizes at about 1110 °C, forming
NaAlSi3O8 (Ab). With decreasing T, the progressive crystallization of Ab dis-
places the composition of the residual liquid toward eutectic point E⬘, which is
Figure 7.9 NaAlSiO4-SiO2 (A) and KAlSiO4-SiO2 (B) system at P ⫽ 1 bar, after Schairer
and Bowen (1955, 1956).
reached at 1062 °C. The crystallization of an initial composition C2, slightly

poorer in SiO2 (i.e., 45%), displaces the composition of the residual liquid in
the opposite direction: the liquid becomes more and more depleted in SiO2 as
crystallization proceeds, until eutectic point E⬙ is reached at T ⫽ 1068 °C. Al-
though the system may be treated as binary for SiO2 amounts exceeding 20% by
weight, phase relations are more complex for less acidic compositions. Crystalliz-
ing pure NaAlSiO4 in the laboratory at T ⫽ 1526 °C leads to the formation of
the crystalline polymorph carnegieite (Cg), which is not observed in nature. Car-
negieite is indeed a solid mixture (Cgs.m. in figure 7.9A), accepting limited
amounts of SiO2 in solution. At the peritectic temperature Tr ⫽ 1280 °C, the solid
mixture carnegieite is transformed into solid nepheline. Nepheline, too, is a solid
mixture (Nes.m. ) whose maximum amount of SiO2 is reached at eutectic point E⬙
(T ⫽ 1068 °C). Note also in figure 7.9A that the two-phase fields Cgs.m. ⫹ L and
Nes.m. ⫹ L are bounded on the left by solidus curves separating them from the
corresponding single-phase fields Cg and Nes.m.. Moreover, a limited two-phase
field Ne ⫹ Cg occurring between the single-phase fields Cg and Nes.m. is observed
experimentally at subsolidus. The Nes.m. ⫹ L field is limited downward by the
solidus curve at eutectic point E⬙ (T ⫽ 1068 °C), below which we find the Ne ⫹
Ab solvus.
Let us now follow the crystallization of an initial composition C3 (20% Ne
and 80% SiO2 by weight). The solid mixture Cgs.m. initially forms at about 1300
°C. As crystallization proceeds, the liquid becomes enriched in SiO2 , moving
along the liquidus curve. At T ⫽ 1280 °C, crystallization stops until the congruent
reaction Cgs.m. ⇔ Nes.m. is completed (there is invariance because there are three
phases and two components). Once the congruent reaction is complete, crystalli-
zation proceeds again: both the solid mixture Nes.m. and the liquid become more
and more enriched in SiO2 (moving along solidus and liquidus curves, respec-
tively) until eutectic point E⬙ is reached and the remaining liquid is converted into
albite plus nepheline.
Figure 7.9B shows the binary diagram for the KAlSiO4-SiO2 (kalsilite-silica)
system. On the SiO2-rich side of the join, phase boundaries are similar to those
of their sodic counterpart, but two compounds with fixed stoichiometry form
in the intermediate zone: sanidine (KAlSi3O8 ; Sa) and leucite (KAlSi2O6 ; Le).
Crystallising an initial composition C4 (40% SiO2 and 60% KAlSiO4 by weight),
we first observe the production of leucite crystals (T ⫽ 1580 °C). The liquid be-
comes more and more enriched in SiO2 as crystallization proceeds, until the in-
variant peritectic point is reached at Tr and leucite is transformed into sanidine
by the incongruent reaction
KAlSi 2 O6 + SiO 2 ⇔ KAlSi 3 O8 .

leucite sanidine (7.14)
However, all the SiO2 component in the liquid is exhausted before reaction 7.14
is completed, and the resulting subsolidus assemblage is composed of Le ⫹ Sa
crystals. If the initial molar amount of SiO2 in the liquid is higher than the value
dictated by the molar proportion
KAlSiO4 : SiO2 = 1 : 2, ( 7.15)
peritectic reaction 7.14 achieves completion and the crystallization process con-
tinues with the production of sanidine at eutectic point E⬘. If, instead, the initial
molar amount of SiO2 in the liquid is lower than the value dictated by the mo-
lar proportion
KAlSiO4 : SiO2 = 1 : 1 ( 7.16 )
(as, for instance, C5 in figure 7.9B), crystallization proceeds with the formation
of liquids increasingly impoverished in the SiO2 component until eutectic E⬙ is
reached.
Comparing figures 7.9 and 7.8, we can fully understand the nature of phase
stability relations in the examined systems: the binary KAlSiO4-SiO2 system is a
simple combination of the two configurations displayed in figure 7.8, whereas the
NaAlSiO4-SiO2 system shows some additional complexities on the Na-rich side
of the join. However, it is important to stress that the apparent complexity of T-
X phase stability diagrams always arises from the (quite simple) configuration of
the Gibbs free energy curves in the various phases forming in the system. This
indicates how important it is in geochemistry to establish accurately the mixing
properties of the various phases involved in petrogenetic equilibria.
We have seen that the crystallization of initial compositions respectively
poorer (C2) and richer (C1) in SiO2 than NaAlSi3O8 stoichiometry follows oppo-
site paths in the binary NaAlSiO4-SiO2 system. This occurs essentially because
compound NaAlSi3O8 melts at a local T maximum (or, in thermochemical terms,
with decreasing T the chemical potential of NaAlSi3O8 reaches the Gibbs free
energy curve of the liquid before the two eutectic configurations; cf. figure 7.8).
Analogously, in the KAlSiO4-SiO2 system, we have a local T maximum (T ⫽ 1686
°C) at the melting temperature of the stoichiometric compound KAlSi2O6 .
These apparently insignificant facts actually have profound consequences for
magma petrology, because they constitute thermal barriers below which magma
differentiation proceeds in opposite chemical directions (silica undersaturation or
silica oversaturation), depending on the initial composition of the system.
7.2 Extension to Ternary Systems
7.2.1 Ternary Mechanical Mixtures

In petrology, the usual representation of phase equilibria in the crystallization
process of ternary systems is based on a projection from the T axis onto the
compositional plane. The latter is triangular, with each side representing a binary
join of the ternary system. The projection shows the liquidus surface and, at each
point of the diagram, highlights the condition of incipient crystallization of the
melt. To facilitate interpretation of the crystallization path followed by any initial
composition in the T-X field, the isotherms are drawn on the liquidus surface.
Figure 7.10 shows a ternary diagram generated by the projection from the T
axis of the phase limits for three mechanical mixtures of components 1, 2, and 3
(1–2, 1–3, and 2–3, respectively). Crystallization of an initial composition C1 will
follow the path shown in figure 7.10. Composition C1 is within “primary phase
field” ␥ ⬙. With decreasing T, once the value of the isotherm passing through C1
is reached, crystallization begins with the formation of ␥ ⬙ crystals of pure compo-
nent 2. As a result of crystallization of ␥ ⬙, the composition of the residual liquid
moves away from C1 along the direction obtained by connecting corner 2 with
the initial composition. During this step, the equilibria are divariant (because P
is fixed, the variance is given by: number of components ⫺ number of phases ⫹
1). Once the residual liquid reaches cotectic line E-EIII, determined by the inter-
section of primary phase field ␥ ⬙ with primary phase field ␥⬘, crystals ␥⬘ (com-
Figure 7.10 Phase stability relations in a ternary system in which components are totally
immiscible at solid state, and relationships with three binary joins.
posed of pure component 1) begin to form in conjunction with ␥ ⬙. The com-

position of the residual liquid now moves along the cotectic line in a univariant
equilibrium (at each T, the chemistry of the residual liquid is now fixed, indepen-
dent of initial composition). Crystallization finally ends at ternary eutectic point
EIII, where pure crystals ␥ ⵮ (component 3) are formed in conjunction with ␥⬘ and
␥ ⬙ until the residual liquid is completely exhausted.
Note that the crystallization path in the T-X space does not necessarily follow
the steepest descent (i.e., the trajectory is not necessarily orthogonal to the direc-
tion of the isotherms), but is determined by conjunction 2-C1 between the pri-
mary phase and the initial composition. In the case of mechanical mixtures, the
trajectory in the primary phase field is thus always straight. Note also that the
amount of crystals at each T is dictated by the baricentric coordinates: at T ⫽
1150 °C (point X in figure 7.10), the amount of crystals (in percent by weight or
percent in moles, depending on the scale used) is given by the ratio of C1-X and
2-C1 segments. A melting process in a closed system on a ternary diagram such
as that shown in figure 7.10 will follow the T-X path exactly opposite to that of
the crystallization event.
7.2.2 Formation of an Intermediate Compound with

Fixed Stoichiometry
Figure 7.11 shows a ternary diagram for a system with components 1–3 and 2–3
immiscible in all proportions at solid state. However, here components 1 and 2
form an intermediate compound with fixed stoichiometry (ratio 30:70).
Let us examine the crystallization process of an initial composition C1: the
first crystals to form are those of the pure component 2 ( ␥ ⬙, T ⫽ 1175 °C). Crys-
tallization of ␥ ⬙ drives the residual liquid radially away from corner 2 along direc-
tion 2-C1, until peritectic curve PP is reached. Here, crystals ␥ ⬙ react with the
liquid and are transformed into intermediate compound Ci. Displacement along
the peritectic curve terminates when all phase ␥ ⬙ is completely resorbed. This
occurs geometrically when trajectory Ci-C1 encounters peritectic curve PP. Crys-
tallization of Ci then proceeds, while the residual liquid moves radially away from
Ci. The liquid path then reaches cotectic curve E⬘-EIII and crystals ␥⬘ begin to
precipitate in conjunction with Ci. Crystallization continues along the cotectic
line and, at ternary eutectic EIII, crystals ␥ ⵮ precipitate together with Ci and ␥⬘
until the residual liquid is exhausted.
Let us now consider the crystallization process operating on a composition
C2, initially poorer in component 1: this composition falls into the compositional
triangle Ci-2–3. Crystallization begins in primary phase field ␥ ⬙ (T ⫽ 1225 °C).
Precipitation of ␥ ⬙ crystals drives the residual liquid radially away from C2 along
direction 2-C2. At T ⫽ 1100 °C, the composition of the liquid reaches cotectic
line EIII-P, where crystals ␥ ⵮ begin to form together with ␥ ⬙. At peritectic point P,
crystals ␥ ⬙ are partially resorbed by reaction with the liquid to form intermediate
compound Ci. However, the peritectic reaction is not completed, because of ex-
Figure 7.11 Phase stability relations in a ternary system with components 1–3 and 2–3
totally immiscible at solid state and components 1–2 forming an intermediate compound
with fixed stoichiometry (Ci).
haustion of the residual melt, and at the end of crystallization the solid assem-
blage is composed of crystals ␥ ⬙, ␥ ⵮, and Ci.
The compositional paths of liquids for melting processes in closed systems
of initial composition C1 or C2 are exactly the opposite of those observed in
crystallization events.
7.2.3 Ternary Systems with Phases of Differing Miscibility

Figure 7.12 shows the liquidus surface of a ternary system with complete miscibil-
ity at solid state between components 1–2 and complete immiscibility at solid
state between components 1–3 and 2–3. Note also that components 1 and 2 form
a lens-shaped two-phase field, indicating ideality in the various aggregation states
(Roozeboom type I).
The ternary system of interest closely resembles the phase relations observed
in the CaAl2Si2O8-CaMgSi2O6-NaAlSi3O8 system. We will examine this system in
some detail later, when dealing with the concepts of f ractional melting and f rac-
Figure 7.12 Phase stability relations in a ternary system with complete solid state misci-
bility of components 1–2 and complete immiscibility at solid state for components 1–3
and 2–3.
tional crystallization. For the moment, we limit ourselves to describing crystalliza-

tion in a closed system (called “equilibrium crystallization” in geochemical litera-
ture) of generic composition C1 (component 1 ⫽ 50%; component 2 ⫽ 35%;
component 3 ⫽ 15%). Unlike the preceding cases, in which the components were
immiscible at solid state and the crystallization of pure components resulted in
straight paths in the T-X space (see figures 7.10 and 7.11), here we have complete
miscibility of components 1 and 2 in all aggregation states. The formation of
compositionally continuous 1–2 crystalline mixtures results in a curved T-X tra-
jectory for the residual liquid, the curvature depending on the shapes of the liq-
uidus and solidus surfaces of the system. If we progressively cool the initial com-
position C1, at T1 ⫽ 1180 °C, we encounter the liquidus surface, and crystals X1
begin to form. The residual liquid moves radially away from X1 in the direction
X1-C1. The newly formed crystals have compositions more enriched in compo-
nent 2 (X2), and the liquid (C2) moves radially away from C2 along X2-C2. The
entire compositional path of the liquid is the result of the various instantaneous
compositional vectors at each crystallization step and has the curved shape shown
in figure 7.12. This curved trajectory brings the residual liquid to cotectic line E⬘-
E⬙ at point C3, and the crystals at equilibrium with the liquid have composition
X3. Crystallization continues along the cotectic line with simultaneous precipita-
tion of the 1–2 crystalline mixture and of crystals ␥ ⵮ composed of pure compo-
nent 3. Crystallization terminates at point C4, where the crystalline mixture 1–2
has composition X4, coinciding (in terms of relative proportions of components
1 and 2) with initial liquid C1. If the initial composition of the liquid falls in
primary phase field ␥ ⵮ (C5), the initial crystallization path is initially straight
until it intersects the cotectic line. Crystallization then proceeds as in the preced-
ing case.
7.3 Crystallization and Fusion in an Open System:

Diopside-Albite-Anorthite as an Example
In all the preceding cases, crystallization takes place in a closed system. The loss
of heat toward the exterior causes a decrease in temperature, total pressure and
bulk chemistry being equal, and this modifies the Gibbs free energy relationships
among the various phases of the system under the new P and T conditions. In
petrology, however, it is also of interest to investigate the effects of the combined
exchange of matter and energy toward the exterior. The processes taking place in
this case are usually defined as “fractional crystallization” (loss of energy and
matter) and “fractional melting” (gain of energy, loss of matter), whereas the
corresponding terms “equilibrium crystallization” and “equilibrium melting” are
applied when there is no exchange of matter with the exterior. The counterposi-
tion “fractionation-equilibrium” implicit in this terminology may give rise to con-
fusion, and it is worth stressing here that fractional melting and fractional crystal-
lization processes also take place at equilibrium, but that equilibrium is local.
We will therefore use the corresponding terms “melting in an open system” and
“crystallization in an open system” to avoid any ambiguity.
Let us now consider, as an example, crystallization in an open system in pla-
gioclases (figure 7.13). Let us imagine cooling the initial composition C until the
liquidus curve of the system is reached (C1); the crystals in equilibrium with
liquid C1 have composition X1. With decreasing T, we reach condition C2, where
crystals of composition X2 coexist with a liquid of composition L2. If the system
is further cooled by a simple heat loss, we reach point C3. The crystals that
formed during cooling at point C2 should now react with the liquid to reach
composition X3. Let us now imagine that the kinetic rate of reaction is lower
than the rate of heat loss toward the exterior: the inner portions of the crystals of
composition X2 do not react with the liquid, and the final composition of the
whole crystals is not X3 but intermediate between X3 and X2. The net result is
that part of the system (inner parts of crystals) plays no part in the equilibrium:
Figure 7.13 Compositional trends in plagioclase binary system during crystallization in

open or closed systems.
the new bulk chemistry is no longer that of the initial system C but is displaced
toward the less refractory component (i.e., more albitic). Proceeding in this way,
we define a compositional path of fractionation corresponding to progressive sub-
traction of the inner portions of the crystals as crystallization proceeds. Note that,
as a net result, crystallization is not complete at the solidus temperature of initial
system C (as in the case of crystallization in a closed system), but proceeds down
to the melting temperature of the low-melting component (i.e., in our case,
Tf,Ab ⫽ 118 °C). The resulting crystals exhibit wide compositional zoning, inde-
pendent of the initial composition of the system (in our case, crystal nuclei have
composition An80 and the surfaces of crystals have composition An 0 ).
The compositional relations discussed above are explained by Rayleigh’s dis-
tillation process (Rayleigh, 1896), choosing a suitable partition coefficient for the
component of interest—i.e., for component An, we can write
 1 − F LK An 
X An,solid = X An,initial  , ( 7.17 )
 1 − FL 
whereX An,solid is the mean composition of the crystals, FL is the fractional amount
of residual liquid in the system, and KAn is the solid/liquid partition coefficient at
interface for component An (see section 10.9.1 for derivation of eq. 7.17).
Let us now treat compositional relations in the ternary system Di-An-Ab.

Figure 7.14A shows the projection of the liquidus surface of the system onto the
compositional plane; the locations of cotectic line E⬘-E⬙ and of the isotherms are
based on the experiments of Bowen (1913), Osborn and Tait (1952), and Schairer
and Yoder (1960).
The system is not strictly ternary, because of the limited solubility of Al2O3 in
diopside, but we will neglect this minor complication here. Because the maximum
number of phases in the system is three (diopside plus plagioclase mixtures plus
liquid) and the components also number three, the minimum variance of the sys-
tem is 1, at constant P.
Figure 7.14 Crystallization and melting in an open system: ternary join CaMgSi2O6-
CaAl2Si2O8-NaAlSi3O8 (Di-An-Ab). Crystallization (B) and melting trends (C) are taken
from Morse (1980).
In the previously discussed cases of crystallization in a closed system, at every

moment of the crystallization process the residual liquid was always located on
the straight line connecting the initial composition with the solid. This geometri-
cal condition is not true for an open system, and each liquid at a given crystalliza-
tion state behaves like an “initial liquid.” At each point in the primary phase field
it is thus essential to locate the composition of crystals in instantaneous equilib-
rium with the liquid and the direction of displacement of the residual liquid when
an infinitesimal amount of solid is subtracted from equilibrium. These composi-
tional vectors, or “fractionation lines,” have been experimentally determined for
the various systems of interest (cf. Hamilton and McKenzie, 1965). In the primary
phase field of the Ab-An mixture, the fractionation lines depart from high-melting
apex An and follow the trends shown in figure 7.14A. Figure 7.14B shows the
fractionation line passing through initial composition C1 (85% plagioclase An60 ,
15% diopside, by weight). During crystallization in an open system of initial com-
position C1, and within the primary phase field of plagioclases, the residual liquid
moves along the fractionation line passing through C1. The instantaneous solid
in equilibrium with initial composition C1 is An84 . The liquid reaches the cotectic
line at C2, moving along the fractionation line. Connecting points C2 and C1,
and applying the lever rule, we can deduce that the mean composition of the bulk
solid precipitated up to C2 is An78 and that the fractional amount of residual
liquid at that point (with respect to the mass of the initial system) is FL ⫽ 0.49.
Note that the composition of the last crystal in instantaneous equilibrium with a
liquid of composition C2 is An65 —i.e., much richer in Ab than the mean com-
position of the precipitated solids. As soon as the composition of the residual
liquid reaches the cotectic line, diopside begins to precipitate in conjunction with
plagioclases. From this point onward, the composition of the liquid moves along
the cotectic line and the bulk composition of the fractionally accumulated solids
is defined at any moment by prolonging the fractionation lines of diopside (which
are straight, Di being immiscible with Ab and An in any proportion) within the
primary phase field of plagioclases and intersecting them with the lines connect-
ing residual and initial liquids. Application of baricentric coordinates to segments
C3-C1-X3, C4-C1-X4, and C5-C1-X5 allows us to determine the fractional
amount of residual liquid in the system (FL ⫽ 0.43, 0.33, and 0.17, respectively)
at any point.
Melting in an open system of the same initial composition C1 begins at a
temperature of 1213 °C on the cotectic line (Morse, 1980; figure 7.14C). Instanta-
neous subtraction of the resulting liquid drives the composition of the residual
solid across the primary phase field of plagioclases until point C2 is reached on
the Ab-An baseline. This happens at T ⫽ 1229 °C, when all the diopside in the
solid paragenesis has been exhausted. The fractionally accumulated liquid (out-
side the system) now has bulk composition L2; the Di component (exhausted in
the solid and externally accumulated) is no longer present in the system, which is
thus now binary. Melting stops until the temperature reaches the solidus isotherm
of composition C2 in the binary Ab-An join (T ⫽ 1408 °C). Once this tempera-
ture is reached, melting starts again, generating liquids that are increasingly richer
in An component, up to pure An. The composition of fractionally accumulated
liquids now moves along the dotted line in figure 7.14C. The amount of this liquid
may be determined by applying the baricentric coordinates to segments L2-C1-
C2, L3-C1-C3, and L4-C1-C4 (FL ⫽ 0.52, 0.59, and 0.79, respectively). Note that
the compositional paths of fractionally accumulated liquids and solids produced
during open-system melting and crystallization are not opposite, as in the case of
closed systems.
PART THREE
Geochemistry of Fluids
CHAPTER EIGHT
Geochemistry of Aqueous Phases
The importance of aqueous solutions in geochemistry can be appreciated if we

recall that two-thirds of the surface of our planet is covered by water and that
aqueous fluids of various salinities are determinant in the development of volcanic
and metamorphic processes in the earth’s upper mantle and crust.
Table 8.1 shows the general distribution of water masses in the hydrosphere.
Although H2O occurs mainly in oceans, 19% of the earth’s H2O mass is still
trapped in lithospheric rocks. A nonnegligible H2O mass (about 1%) is also fixed
in crystalline form as ice.
8.1 General Information on Structure and Properties

of Water
8.1.1 Structure of the H2O Molecule
There are two main ways of representing the H2O molecule: the point-charge
model and molecular orbital representation.
Figure 8.1A shows the point-charge model of Verwey (1941). Point charges
are located at the equilibrium distance (0.99 Å) and form bond angles of 105°
between two H-O bond directions. There are six proton charges (⫹6e), located
on oxygen and half a proton charge (⫹12e) on each of the two hydrogen ions. A
charge of ⫺7e (resulting from the electroneutrality condition) is arbitrarily placed
at 0.024 Å from the oxygen nucleus along the bisector of the bond angle. The
resulting dipole moment ( ␮ ⫽ 1.87 ⫻ 10⫺18 esu ⫻ cm) thus agrees with experi-
mental observations.
Figure 8.1B shows an experimental contour map of electron density for the
H2O molecule in plane y-z, after Bader and Jones (1963). The electron density is
higher around the nuclei and along the bond directrix. The experimental electron
density map conforms quite well to the hybrid orbital model of Duncan and Pople
(1953) with the LCAO approximation.
/ 479
480 / Geochemistry of Fluids
Table 8.1 Mass distribution of H2O in hydrosphere (g).
Oceans and seas 1.37 ⫻ 1024

Lithosphere (hydration plus pore water) 3.2 ⫻ 1023
Ice 1.65 ⫻ 1022
Evaporation plus precipitation 4.5 ⫻ 1020
Lakes and rivers 3.4 ⫻ 1019
Atmospheric water 1.05 ⫻ 1019
Figure 8.1 Structural models for H2O molecule. (A) Electrostatic point-charge model
(from Eisemberg and Kauzmann, 1969; redrawn). (B) Electron density map (from Bader
and Jones, 1963). (C) Formation of MOs ⌿(b⬘) and ⌿(b⬙) starting from AOs 2p and 2s of
oxygen and 1s of hydrogen.
Based on the model of Duncan and Pople (1953), two molecular orbitals (b⬘ and b⬙)
may form as linear combinations of hybrid orbitals 2p and 2s of oxygen and orbital 1s
of hydrogen:
[ ]
Ψ( b′ ) = λ cos ε bψ O(2 s ) + sin ε bψ O(2 p, b′ ) + ξψ H′ (1s ) ( 8.1)
[ ]
Ψ( b′′ ) = λ cos ε bψ O(2 s ) + sin ε bψ O(2 p, b′′ ) + ξψ H′′ (1s ) ( 8.2)
and two “lone pair” orbitals (l⬘ and l⬙) may form as combinations of hybrid orbitals 2s
and 2p of oxygen:
Geochemistry of Aqueous Phases / 481
Ψ( l ′ ) = cos ε lψ O(2 s ) + sin ε lψ O(2 p, l ′ ) ( 8.3)
Ψ( l ′′ ) = cos ε lψ O(2 s ) + sin ε lψ O(2 p, l ′′ ) ( 8.4)
The significance of term ␭ was defined in section 1.17.1. Factor ␰ is analogous (ratio ␭ / ␰
defines the polarity of the bond) and parameters ␧b and ␧l describe the hybridization state
of the orbitals. Because cos ␧b ⫽ 0.093 (Duncan and Pople, 1953), molecular hybrid orbit-
als ⌿(b⬘) and ⌿(b ⬙) are composed essentially of the wave functions of atomic orbitals 2p of
oxygen and 1s of hydrogen. The value 0.578 obtained for cos ␧l also indicates that orbitals
⌿(l⬘) and ⌿(l ⬙ ) are essentially of type sp3. Figure 8.1C shows the formation of hybrid MOs
⌿(b⬘) and ⌿(b ⬙ ) on plane y-z, starting from the AOs 2p and 1s of hydrogen.
8.1.2 Structure of Water: Models of Distortion of the

Hydrogen Bond
An exhaustive discussion on the structure of water is beyond the scope of this
textbook. The term “structure” itself, when applied to a liquid state, demands
several clarifications that cannot be afforded here. We will therefore refer hence-
forth only very briefly to models of distortion of the hydrogen bond that, for their
simplicity do not require extensive explanations.
Such models consider that transition from the crystalline to the liquid state
mainly involves distortion of the hydrogen bond, but no bond rupture. The model
of Pople (1951) assumes that, in the liquid state, each H2O molecule is linked to
three other molecules; distanceR o between neighboring oxygen nuclei is constant,
but the distance between next neighbors varies according to the degree of distor-
tion of the hydrogen bond. As figure 8.2A shows, in Pople’s model the hydrogen
bond is considered undistorted when the O-H bond direction of a molecule falls
along the line connecting neighboring oxygen nuclei. As the degree of distortion
increases ( ␾ angles in figure 8.2A), the internal energy of the liquid also increases.
This may be quantified by
(
∆U = K φ 1 − cos φ . ) ( 8.5)
In equation 8.5, K␾ is the constant of distortion of the hydrogen bond. The degree
of distortion of the hydrogen bond in water varies with temperature [from 26° to
30° for 0 ⬍ T (°C) ⬍ 100] and is zero in ice, where H2O molecules are linked to
form perfect tetrahedra.
The “random network model” of Bernal (1964) is a development of Pople’s
model. In Bernal’s version, the water molecules are conceived as forming an irreg-
ular network of rings (not open chains, as in the preceding case). Most of the
rings have five members, because the O-H bond in a single H2O molecule is near
the angle (108°) characteristic of a five-membered ring, but four, six, seven, or
even more H2O molecules may be present in some rings.
Figure 8.2 Models of distortion of hydrogen bond for water. Note that bond angle be-
tween two O-H bonds of a single molecule is 105°; formation of a five-membered ring
(part B) thus implies distortion of 3°.
When we will discuss the effects of solvent collapse in solute-solvent interac-

tions (section 8.11.2), we will mean local modifications of the water structure
(degree of distortion of the oxygen bond; distance between neighboring oxygen
nuclei) induced by the presence of electrolytes in solution. We refer to the classical
text of Eisemberg and Kauzmann (1969) for a more detailed discussion on the
various aggregation states of the H2O compound.
8.1.3 Dielectric Constant of Water

The concept of dipole moment and its relationship to ion polarizability were dis-
cussed in section 1.8. Section 1.19 introduced the concept of dielectric constant
of a crystalline solid and its relationship with the polarizability of its constituting
ions (see eq. 1.168). The dielectric constant of a liquid solvent such as water
represents the capacity of the solvent’s molecules to shield the charges of ion
Figure 8.3 Dielectric constant of water as a function of P and T conditions. Reprinted

from T. M. Seward, Physics and Chemistry of Earth, 13, 113–132, copyright 䉷 1981, with
kind permission from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington
0X5 1GB, UK.
solutes by means of the solvation process. The dielectric constant of solvent ␧

may be related to molecular polarizability ␣m and dipole moment ␮ of the single
solvent molecules through Kirkwood’s equation (Bockris and Reddy, 1970):
(ε − 1)( 2ε + 1) = α +
µ 2 g  4πN 0 d
 m × , ( 8.6 )
gε  3kT  3W
where N 0 and k are, respectively, Avogadro’s number and Boltzmann’s constant,

W is the molecular weight of the solvent, d is the density, and g is a function of
absolute temperature and density. It may be derived for water by applying Pitzer’s
(1983) polynomial:
 a  a 4 
g H2 O 
= 1 + a1 d + a 2 d  
3
5
− 1 , ( 8.7 )
 T  
 
with a1 ⫽ 2.68, a2 ⫽ 6.69, a3 ⫽ 565, and a4 ⫽ 0.3.

The dielectric constant is directly proportional to the density of the solvent
(hence to pressure) and inversely proportional to temperature, as shown in figure
8.3. Discrete values of the dielectric constant of water at high P and T conditions
are listed in table 8.2.
Table 8.2 Dielectric constant of water at various T and P conditions, after Pitzer (1983).
Temperature (°C)
P (kbar) 400 450 500 550 600 650 700 750 800
1 – 12.49 9.43 7.04 5.34 4.24 3.52 3.04 2.69
2 – 16.46 13.87 11.69 9.87 8.37 7.14 6.16 5.37
3 21.78 18.72 16.20 14.10 12.31 10.80 9.51 8.42 7.49
4 23.46 20.35 17.83 15.73 13.95 12.44 11.14 10.02 9.05
5 24.82 21.63 19.63 16.94 15.17 13.65 12.35 11.21 10.22
8.1.4 Ionic Dissociation of Water

Water is partly ionized as a result of the ion dissociation process:
H2O ⇔ H+ + OH−. (8.8)
Dissociation constant KH2O:
aH + ⋅ aOH −
K H 2O = ( 8.9 )
a H 2O
varies directly with T and inversely with P, as shown in figure 8.4. Discrete values
of the ionic dissociation constant for T between 0 and 60 °C and P ⫽ 1 bar are
Eugster and Baumgartner (1987) suggested that the ion dissociation constant
can be numerically reproduced in a satisfactory way by the polynomial
B CP
log K H2O = A + + + D log d , (8.10)
T T
where d is water density at the T of interest and constants A, B, C, and D have

the following values: A ⫽ ⫺4.247; B ⫽ ⫺2959.4; C ⫽ 0.00928; D ⫽ 13.493.
8.1.5 Critical Temperature and P-T Stability of Water

Figure 8.5 shows the projection on the P-V plane of the state diagram of H2O
near the critical point. This figure distinguishes four separate portions of the
P-V space, limited, respectively, by the following:
1. The liquid saturation boundary curve

2. The vapor saturation boundary curve
3. The critical point
4. The critical T isotherm (Tc ⫽ 374 °C).
Figure 8.4 Ion dissociation constant of water as a function of P and T conditions. From
Eugster and Baumgartner (1987). Reprinted with permission of The Mineralogical Society
of America.
Table 8.3 Ionic dissociation constant of water for T between

0 and 60 °C at P ⫽ 1 bar.
T (°C) KH2O ⫻ 1014 T (°C) KH2O ⫻ 1014

0 0.1139 5 0.1846
10 0.2920 15 0.4505
20 0.6809 25 1.008
30 1.469 35 2.089
40 2.919 45 4.018
50 5.474 55 7.297
60 9.614
Within the PV region delimited by the two saturation boundary curves, liquid
and vapor phases coexist stably at equilibrium. To the right of the vapor satura-
tion curve, only vapor is present: to the left of the liquid saturation curve, vapor
is absent. Let us imagine inducing isothermal compression in a system composed
of pure H2O at T ⫽ 350 °C, starting from an initial pressure of 140 bar. The H2O
will initially be in the gaseous state up to P ⬍ 166 bar. At P ⫽ 166 bar, we
reach the vapor saturation curve and the liquid phase begins to form. Any further
Figure 8.5 P-V projection of state diagram for H2O near critical point. Pc and Vc are
critical pressure and critical volume of compound.
compression on the system will not induce any increase in P, because the appear-
ance of a new phase has decreased the degree of freedom of the system, which is
now invariant. Compression thus simply induces an increase in the bulk density
of the vapor-plus-liquid assemblage, until the liquid saturation curve is reached.
At this point, the vapor phase disappears and the system is again univariant: any
further compression now induces a dramatic increase in P (because the liquid
phase is relatively incompressible with respect to the gaseous counterpart: the
slopes of the isotherms on the gas and liquid sides of the saturation boundaries
are very different).
If we now repeat compression along the critical isotherm Tc, we reach (at the
critical point) a flexus condition, analytically defined by the partial derivatives
 ∂P 
  =0 ( 8.11.1)
 ∂V  T
c
 ∂ 2P
  = 0. ( 8.11.2 )
 ∂V 2  T
c
At T higher than Tc, a liquid H2O phase can never be formed, regardless of how
much pressure is exerted on the system.
Figure 8.6A is an enlarged portion of the P-T plane for H2O in the vicinity
of the critical point (Tc ⫽ 373.917 °C, Pc ⫽ 220.46 bar).
Five P-T regions of H2O stability are distinguished:
1. The “liquid” region, limited by the vaporization boundary (i.e., the liquid
saturation and vapor saturation boundary curves) and the critical isobar.
2. The “supercritical liquid” region, limited to supercritical pressures and
subcritical temperatures.
3. The “supercritical fluid” region, which encompasses all conditions of su-
percritical temperatures and pressures.
4. The “supercritical vapor” region, below the critical isobar and above the
critical isotherm.
Figure 8.6 (A) P-T regions of H2O stability. (B) Extended P-T field showing location of
the critical region. The critical isochore is ␳ ⫽ 0.322778 g/cm3. From Johnson and Norton
(1991), American Journal of Science, 291, 541–648. Reprinted with permission of Ameri-
can Journal of Science.
5. The “vapor” region, limited by the vaporization boundary and the criti-
cal isotherm.
In geochemistry, unspecified liquid-plus-gas assemblages are commonly defined as

“fluids.” Note, however, that a “supercritical fluid” is necessarily composed of a single
gaseous phase. Another origin of confusion is use of the term “vapor” instead of “gas.”
The two terms in fact have exactly the same meaning. The term “vapor” is preferred when-
ever the compound of interest is liquid in the stable state at ambient conditions—i.e., we
say “H2O vapor” or “Hg vapor” because both H2O and Hg are liquid in the stable state,
just as we say “oxygen gas” or “hydrogen gas” because both are gaseous in the stable state.
Note also that “saturated vapor” means H2O gas stably coexistent with water (i.e., within
the two-phase field) and “unsaturated vapor” or “dry vapor” means gaseous H2O to the
right of the vapor saturation boundary in the PV space or under supercritical conditions.
Figure 8.6B shows a wider P-T portion with the location of the “critical re-
gion” for H2O, bound by the 421.85 °C isotherm and the ␳ ⫽ 0.20 and 0.42 g/cm3
isochores. The PVT properties of H2O within the critical region are accurately
described by the nonclassical (asymptotic scaling) equation of state of Levelt
Sengers et al. (1983). Outside the critical region and up to 1000 °C and 15 kbar,
PVT properties of H2O are accurately reproduced by the classical equation of
state of Haar et al. (1984). An appropriate description of the two equations of
state is beyond the purposes of this textbook, and we refer readers to the excellent
revision of Johnson and Norton (1991) for an appropriate treatment.
The nonclassical equation of state of Levelt Sengers et al. (1983) for the critical region
of H2O may be expressed in terms of reduced parameters (cf. section 9.2) as follows:
 ∂P˜ 
ρr =   , ( 8.12 )
 ∂µ˜  T˜
where ␳r is the reduced density ␳ / ␳c , with ␳c ⫽ 0.322778 g/cm3 (critical density). The other
dimensionless parameters T̃, P̃, and ␮˜ are given by
T 
T˜ = −Tr−1 = −  c  ( 8.13.1)
T
P T  P
P˜ = r = c   ( 8.13.2 )
Tr Pc  T 
ρ c Tc  µ 
µ˜ =  , ( 8.13.3 )
Pc T 
where Tc and Pc are the critical temperature and critical pressure.

A classical equation of state is normally composed of a truncated Taylor series in the

independent variables, normalized to the critical point conditions (e.g., van der Waals,
virial expansion, etc.). All these sorts of equations yield similar (so-called “classical”)
asymptotic behavior in their derivative properties at the critical point.
The classical equation of state of Haar et al. (1984), valid outside the critical region,
may be expressed as follows:
−1 2
 ∂A 
( )
12
ρ = PCM   ( 8.14 )
 ∂ρ  T
(Johnson and Norton, 1991), where M is the molar weight of H2O (18.0152 g/mol), C is
a conversion factor (0.02390054 cal bar⫺1cm⫺3 ) and A is the molal Helmholtz free en-
ergy function:
( ) ( )
A( ρ , T ) = Abase ρ , T + Aresid ρ , T + Aideal ρ , T ( ) ( 8.15)
(see Johnson and Norton, 1991, for a detailed account on the significance of the various
terms in eq. 8.15). Note that the molar volume may be readily derived from equations
8.12 and 8.14 by applying V ⫽ M/ ␳.
Reversing equations 8.11.1 and 8.11.2, we obtain
−1 −1
 ∂P   ∂ 2P 
  =∞ and   = ∞. ( 8.16 )
 ∂V  T  ∂V 2  T
c c
Moreover, recalling what we have already seen in section 1.15 (namely, eq.
1.95, 1.96, 1.97, and 1.103), we may also derive
−1 −1
 ∂ 2A   ∂ 3A 
  = −∞ and   = −∞, ( 8.17 )
 ∂V 2  T  ∂V 3  T
c c
where A is the Helmholtz free energy of the substance, and
−1 −1
 ∂ 2U   ∂ 3U 
  = −∞ and   = −∞. ( 8.18 )
 ∂V 2  T  ∂V 3  T
c c
Equations 8.16 and 8.17 or 8.16 and 8.18 show that, at the critical point, “the
specific first and second derivative properties of any representative equation of
state will be divergent” (Johnson and Norton, 1991). This inherent divergency
has profound consequences on the thermodynamic and transport properties of
H2O in the vicinity of the critical point. Figure 8.7 shows, for example, the behav-
Figure 8.7 (A) Isobaric thermal expansion, (B) its first T-derivative, (C) isothermal com-
pressibility, and (D) isobaric heat capacity of H2O within the critical region, based on the
equation of state of Levelt Sengers et al. (1983). From Johnson and Norton (1991), Ameri-
can Journal of Science, 291, 541–648. Reprinted with permission of American Journal
of Science.
ior of the isobaric thermal expansion ( ␣ ) of H2O and of its first derivative on T
within the critical region, computed by Johnson and Norton (1991) on the basis
of the equation of state of Levelt Sengers et al. (1983). The divergent asymptotic
behavior of ( ∂␣ / ∂T ) in the vicinity of the critical point has been arbitrarily trun-
cated within the ⫺5 ⫻ 10⫺3 K⫺2 and 5 ⫻ 10⫺3 K⫺2 lower and upper limits for
illustrative purposes. Note also in parts C and D of figure 8.7 the huge domes in
the isothermal compressibility ␤ and isobaric heat capacity CP of H2O within the
critical region.
8.2 Electrolytic Nature of Aqueous Solutions:

Some Definitions
Now that we have examined to some extent the properties of water as a “solvent,”
let us discuss the nature of aqueous solutions.
From the point of view of chemical modeling, aqueous solutions are treated
as “electrolytic solutions”—i.e., solutions in which solutes are present partially
or totally in ionic form. “Speciation” is the name for the characteristic distribu-
tion of ion species in a given aqueous solution in the form of simple ions, ionic
couplings, and neutral molecules. Solutes in aqueous solutions are defined as
“electrolytes” and may be subdivided into “nonassociated” and “associated.”
Nonassociated electrolytes are also defined as “strong” and mainly occur in the
form of simple or simply hydrated ions. An example of a strong electrolyte is the
salt NaCl, which, in aqueous solution of low ionic strength, occurs in the form
of completely dissociated Na⫹ and Cl ⫺ ions.
Associated electrolytes are further subdivided into two groups: “weak electro-
lytes” and “ionic couplings.” A weak electrolyte occurs both in the form of cova-
lent molecules and in ionic form. Examples of weak electrolytes are various acids,
some bases, and a few inorganic salts. Aqueous solutions of carbonic acid H2CO3 ,
for instance, contain both undissociated neutral molecules H2CO03 and carbonate
and bicarbonate ions CO32⫺ and HCO3⫺. Solutions of strong inorganic acids such
as HF, when present in high molar concentration, also contain neutral molecules
HF 0 besides H⫹ and F⫺. The term “ionic coupling” is used when the association
of oppositely charged ions is due not to the formation of a strong covalent bond
(as in the case of weak electrolytes) but to electrostatic interactions. For instance,
in aqueous solutions of calcium sulfate CaSO4 , besides Ca2⫹ and SO42⫺ ions,
neutral particles CaSO4 0 are also present. In aqueous solutions of calcium hy-
droxide, some of the solutes are in the form of ionic couplings Ca(OH)⫹. The
distinction of weak electrolytes and ionic couplings in aqueous solutions is some-
what problematic, because in some cases both types of electrolytes have the same
stoichiometry. Lastly, the term “complex ion” identifies all nonmonatomic solute
ions, independent of the nature of the chemical bond. Examples of complex ions
are thus Ca(OH)⫹, CO32⫺, HCO3⫺, HS⫺, ZnCl42⫺, and B3O3(OH)4⫺.
8.3 Models of “Ionic Coupling–Complexation” and

“Specific Interactions”
The approach usually adopted in geochemistry to describe the speciation state of

aqueous solutions is that of ionic coupling–complexation, based on the classical
model developed by Garrels and Thompson (1962) for seawater. This model con-
siders the various aqueous ions as species physically present in solution, in molal
proportions determined by the coexistence of multiple homogeneous equilibria
of the type
Ca2 + + HCO 3− ⇔ CaHCO 3+ , ( 8.19 )
and related to the energy balance through application of appropriate individual

ionic activity coefficients (see section 8.9).
The approach of specific interactions, developed primarily by Pitzer (1973)
and Whitfield (1975a,b), considers all salts, from a purely formal point of view,
as completely dissociated, and embodies the effects of specific interactions into
particular activity coefficients, defined as “total activity coefficients” or “stoichio-
metric activity coefficients,” with symbol ␥T. For instance, for ion i,
a i = m iT γ T
i = miγ i , ( 8.20)
where mTi is the total molal concentration of the species and mi is the molal con-
centration of the free ion:
 mi 
γ T
=γi  . ( 8.21)
i  mT
 i 
The main advantage of the specific interactions model lies in the simplicity of its
calculations. Also, considering dissolved salts (in their neutral salt stoichiome-
try—e.g., NaCl) as “components,” activities and total activity coefficients are
experimentally observable magnitudes.
However, the ionic coupling–complexation model is more appropriate to the
actual complexity of aqueous solutions, as is evident from spectral absorption
and ionic conductivity studies.
8.4 Activities and Activity Coefficients of Electrolytes

in Aqueous Solutions
When speaking of “solutions,” we implicitly state that the H2O “solvent” has a
different standard state of reference with respect to “solutes” (note that the term
“mixture” is used when all components are treated in the same way; see section
2.1). The standard state generally used for the solvent in aqueous solutions is that
of “pure solvent at P and T of interest” (or P ⫽ 1 bar and T ⫽ 298.15 K). For
solutes, the “hypothetical one-molal solution referred to infinite dilution, at P
and T of interest” (or P ⫽ 1 bar and T ⫽ 298.15 K) is generally used. This choice
is dictated by practical considerations.
Activity ae of a generic electrolyte e ⫽ Cv⫹Av⫺, dissociating according to the

scheme
C v + Av − ⇔ v +C + v −A , ( 8.22 )
where C ⫽ cation and A ⫽ anion, is described at high dilution by Henry’s law:
a e = K e m ev , ( 8.23)
where
v = v+ + v− . ( 8.24 )
Let us now imagine plotting the activity of electrolyte e as a function of mol-

ality, elevated to stoichiometric factor v (mve ), as shown in figure 8.8. Graphically
in this sort of plot, Henry’s law constant Ke represents the slope of equation 8.23
at infinite dilution. Extending the slope of Henry’s law up to value mve ⫽ 1 (dashed
line in figure 8.8) and arbitrarily fixing the ordinate scale so that activity is 1 at
the point defined by the extension of the Henry’s law slope at mve ⫽ 1, we can
construct the condition of “hypothetical one-molal solution referred to infinite
dilution.” This condition not only obeys the unitary activity implicit in the stan-
dard state definition, but also results in an activity coefficient of 1 at infinite dilu-
tion (i.e., ae ⫽ mve ). The term “hypothetical” emphasizes the fact that the adopted
reference condition does not correspond to the energy properties of the actual
solution at the same concentration level (see figure 8.8).
The activity of electrolyte ae may be related to the individual ionic activities
of constituting ions a⫹ and a⫺ :
Figure 8.8 Construction of standard state of “hypothetical one-molal solution referred

to infinite dilution.”
( ) (a )
v+ v−
ae = a+ − ( 8.25 )
which, in turn, determine “mean ionic activity” a⫾
1v
( ) ( )
a± =  a+ 
v+ v−
a− = a1e v . ( 8.26 )
 
Similarly, “mean ionic molality” is defined as
1v
m± =  m+ ( ) (m )
v+ v−  1v
= me , ( 8.27 )
 − 
and the “individual ionic activity coefficients” are expressed by
a+ a−
γ+ = and γ− = , ( 8.28 )
m+ m−
where m⫹ and m⫺ are moles of dissociated ions.

Lastly, “mean ionic activity coefficient” is given by
1v
( ) (γ )
=  γ+
a± v+ v− 
γ = = γ 1e v . ( 8.29 )
m±  
± −
Among the above-defined thermodynamic entities, the individual ionic activ-

ity coefficients are particularly useful, because they allow practical calculation of
the speciation state of an aqueous phase, linking individual ionic molalities to the
energy balance. We will see in the following section how these coefficients may
be derived.
8.5 Ionic Strength and Debye-Hückel Theory
The concept of ionic strength, which allows individual ionic activity coefficients
to be estimated, was developed by Lewis and Randall (1921). The ionic strength
of a solution is given by
1
I=
2
∑ m i Z i2 , ( 8.30.1)
i
where mi is the molality, Zi is the charge of the ith ion, and summation is extended
to all ions in solution.
The ionic strength of a one-molal solution of CaCl2 is thus
1 m
( )  = 3.
2
I= × 2 2 + m Cl − × −1 ( 8.30.2 )
2  Ca 2 + 
Clearly, ionic strength corresponds to molality in the case of a uni-univalent

strong electrolyte—i.e., for a one-molal solution of NaCl:
1 m
( )  = 1.
2
I= × 1 2 + m Cl − × −1 ( 8.30.3)
2  Na − 
The concept of ionic strength is very useful in quantifying the effects of electro-
static charges of all ions present in solution. Its relationship with the nature of
electrostatic interactions was stressed by Lewis and Randall: “In diluted solutions
the activity coefficient of a given strong electrolyte is the same in all solutions
having the same ionic strength.” In fact, in light of the Debye-Huckel theory, the
relationship between individual ionic activity coefficient ␥i of ion i and the ionic
strength of solution I is given by
AZ i2 I
log γ i = − . ( 8.31)
1 + åi B I
A and B in equation 8.31 are characteristic parameters of the solvent (in this case,
water), which can be derived in the various T and P conditions of interest by
application of
1.824829 × 10 6 ρ 1 2
A= (kg1 2 × mole −1 2 ) ( 8.32.1)
(εT )
32
and
50.291586 × 10 8 ρ 1 2
B= (kg1 2 × mole −1 2 × cm −1 ) ( 8.32.2 )
(εT )
12
(Helgeson and Kirkham, 1974). In equations 8.32.1 and 8.32.2, ␳ is solvent den-
sity, ␧ is the dielectric constant of the solvent, and åi is the “effective diameter”
of ion i in solution. The latter parameter must not be confused with the ionic
diameter of the ion in a condensed aggregation state, but represents the range of
electrostatic interaction with the solvent molecules (see section 8.11.3).
Table 8.4 lists discrete values of solvent parameters A and B for H2O in the
Table 8.4 Debye-Huckel constants for

H2O in T range 0 to 300 °C at saturation P
(from Helgeson and Kirkham, 1974). A is
in kg1/2 ⫻ mole1/2 and B is in kg1/2 ⫻
mole1/2 ⫻ cm⫺1.
T (°C) A B ⫻ 10⫺8
0 0.4911 0.3244
25 0.5092 0.3283
50 0.5336 0.3325
75 0.5639 0.3371
100 0.5998 0.3422
125 0.6416 0.3476
150 0.6898 0.3533
175 0.7454 0.3592
200 0.8099 0.3655
225 0.8860 0.3721
250 0.9785 0.3792
275 1.0960 0.3871
300 1.2555 0.3965
Table 8.5 Effective ion diameters (Å) for various ions and inorganic complexes in
aqueous solution (after Kielland, 1937).
å Simple Ions and Aqueous Inorganic Complexes
2.5 Rb⫹, Cs⫹, NH⫹ ⫹

4 , Tl , Ag
⫹
3 K⫹, Cl⫺, Br⫺, I⫺, CN⫺, NO⫺ ⫺

2 , NO3 ,
3.5 OH⫺, F⫺, NCS⫺, NCO⫺, HS⫺, ClO⫺ ⫺ ⫺ ⫺ ⫺

3 , ClO4 , BrO3 , IO4 , MnO4
4 Hg 22⫹, SO42⫺, S2O32⫺, S2O82⫺, SeO42⫺, CrO42⫺, HPO42⫺, S2O62⫺, PO43⫺,
Fe(CN)63⫺, Cr(NH3)63⫹, Co(NH3)53⫹, Co(NH3)5H2O3⫹
4–4.5 Na⫹, CdCl⫹, ClO⫺ ⫺ ⫺ ⫺ ⫺ ⫺
2 , IO3 , HCO3 , H2PO4 , HSO3 , H2AsO4 , Co(NH3)4(NO2)2
⫹
4.5 2⫹ 2⫺ 2⫺ 2⫺ 2⫹
Pb , CO3 , SO3 , MoO4 , Co(NH3)5Cl , Fe(CN)5NO 2⫺
5 Sr2⫹, Ba2⫹, Ra2⫹, Cd2⫹, Hg2⫹, S2⫺, S2O4⫺2, WO42⫺, Fe(CN)64⫺

6 Li⫹, Ca2⫹, Cu2⫹, Zn2⫹, Sn2⫹, Mn2⫹, Fe2⫹, Ni2⫹, Co2⫹, Co(S2O3)(CN)54⫺
8 Mg2⫹, Be2⫹
9 H⫹, Al3⫹, Fe3⫹, Cr3⫹, Sc3⫹, Y3⫹, La3⫹, Ce3⫹, Pr3⫹, Nd3⫹, Sm3⫹, In3⫹, Co(SO3)2(CN)45⫺
11 Th4⫹, Zr4⫹, Ce4⫹, Sn4⫹
T range 0 to 300 °C (saturation P), and table 8.5 lists effective diameter values
for various ions in aqueous solution.
When the ionic strength of a solution is low, the denominator in equation
8.31 tends toward 1, and the individual ionic activity coefficient is well approxi-
mated by
log γ i ≈ − AZ i2 I . (8.33)
8.6 The “Mean Salt” Method
The mean salt method derives implicitly from the concept of ionic strength, as
expressed by Lewis and Randall (1921). It is assumed that, within the range of
the ionic strength of interest, for a standard uni-univalent electrolyte such as KCl,
the following assumption is valid:
γ K+
=γ Cl −
( 8.34 )
From equation 8.34, with a reasonable approximation (see eq. 8.29), we can
derive
[( )(γ )]
1 2
γ ± KCl = γ K+ Cl −
=γ K+ =γ Cl −
. ( 8.35 )
Because the ␥⫾ values for the various salts in solution may be experimentally
obtained with a satisfactory precision, equation 8.35 is used to derive the corre-
sponding values of individual ionic activity coefficients from them. For instance,
from a generic univalent chloride MCl, the ␥⫾ of which is known, we may derive
the ␥⫹ of M⫹ by applying
1 2 1 2
γ ± MCl

=γ

( M+ )(γ ) Cl −




=γ

( M+ )(γ ± KCl ) 


( 8.36.1)
γ M+ = γ 2
± MCl γ ± KCl . (8.36.2)
For a divalent chloride MCl2 ,
13 13
( )(
γ ± MCl 2 =  γ M 2 + γ Cl − ) 
(
=  γ M 2 + γ ± KCl )( ) 
2 2
 (8.37.1)
  
γ M 2+ = γ 3
± MCl γ ±2 KCl . (8.37.2)
Similarly, the same method and the same mean salt furnish values of individual
ionic activity coefficients for other ions—e.g., the sulfate group SO2⫺
4 :
(γ ) (γ )
13 13
(
γ ± K 2SO 4 =  γ K + ) (
=  γ ± KCl )
2 2
(8.38.1)
 SO 24 −  SO 24 −
γ SO2 − = γ 3
± K2 SO 4
γ ±2 KCl . ( 8.38.2 )
4
For a salt with both cation and anion different from those of the mean salt (e.g.,
CuSO4 ), double substitution is required—i.e.,
( )
12
γ ± CuSO4 =  γ Cu2 +  γ SO 2 −  ,
 4 
and, stemming from equation 8.38.2,
γ ±2 CuSO γ ±2 KCl
γ Cu2 + = 4
. (8.40)
γ ±3 K
2SO 4
Figure 8.9 shows mean activity values for various salts plotted according to
the ionic strength of the solution. Note how, at low I (I ⬍ 0.05), the values con-
verge for salts of the same stoichiometry (i.e., uni-univalent: NaCl, KCl: uni-
divalent: Na2SO4 , K2SO4 ).
Figure 8.10 shows the comparison between the mean salt method and the
Debye-Huckel model in the estimate of individual ionic activity coefficients. The
two methods give concordant results for low ionic strength values (I ⬍ 0.1). For
ions of the same net charge, convergence toward identical values occurs at zero
ionic strength. At high ionic strength, the individual ionic activity coefficients of
the cations increase exponentially with I. Estimates at I ⬎ 5 are affected by a high
degree of uncertainty.
Individual ionic activity coefficients may also be estimated more precisely
Figure 8.9 Mean activity coefficients for chlorides and sulfides, plotted following ionic
strength of solution. Reprinted from Garrels and Christ (1965), with kind permission from
Jones and Bartlett Publishers Inc., copyright 䉷 1990.
Figure 8.10 Comparison between individual ionic activity coefficients obtained with
Debye-Hückel equation and with mean salt method for various ionic strength values. Re-
printed from Garrels and Christ (1965), with kind permission from Jones and Bartlett
Publishers Inc., copyright 䉷 1990.
with Helgeson’s (1969) equation, which introduces a corrective factor B̊ in the

Debye-Hückel equation:
AZ i2 I ˚ .
log γ i = − + BI ( 8.41)
1 + åi B I
8.7 Activity of Dissolved Neutral Species
Figure 8.11 shows activity coefficients for gaseous molecules dissolved in aqueous
NaCl solutions at various molalities (T ⫽ 15 and 25 °C; P ⫽ 1 bar). The coeffi-
cients converge on 1 at zero ionic strength. Each species also has distinct values,
at any I.
Figure 8.12 shows that, for a single neutral species (CO2 ), the activity coeffi-
cient varies according to type of dissolved electrolyte and increases with increas-
ing ionic strength of the solution. This phenomenon is known as “salting out,”
and generally takes place for nonpolar neutral species such as CO2 , O2 , H2 , and
N2 dissolved in aqueous solutions. For polar neutral species such as CaSO04 and
MgSO04, the inverse phenomenon, called “salting in,” is observed: in this case, the
Figure 8.11 Activity coefficients for neutral gaseous molecules dissolved in aqueous solu-
tions of various ionic strengths. All values are for T ⫽ 25 °C and P ⫽ 1 bar, except for
hydrogen (T ⫽ 15 °C, P ⫽ 1 bar). Reprinted from Garrels and Christ (1965), with kind
permission from Jones and Bartlett Publishers Inc., copyright 䉷 1990.
Figure 8.12 Salting-out phenomenon for aqueous CO2. Activity coefficient of neutral
species increases with increasing salinity, determining decreased solubility of aqueous CO2
in water, T and P conditions being equal. Reprinted from Garrels and Christ (1965), with
kind permission from Jones and Bartlett Publishers Inc., copyright 䉷 1990.
activity coefficient of the neutral species in solution decreases with increasing

ionic strength. In the development of speciation calculations (see section 8.9), it
is often assumed that all neutral species in solution have the same activity coeffi-
cient as aqueous CO2 , calculated with a polynomial in I of the type:
log γ i = C1,CO2 I + C 2 ,CO2 I 2 + C 3,CO2 I 3 + C 4,CO2 I 4 , ( 8.42 )
where C1,CO2, C2,CO2, C3,CO2, and C4,CO2 are known constants.
8.8 Activity of Solvent in Aqueous Solutions
The activity of H2O solvent aH2O in aqueous solutions decreases with increasing
molality of solutes. Figure 8.13 shows the effects of progressive addition of dis-
solved chlorides in solution on aH2O.
In theory, once the activity of an electrolyte in solution is known, the activity
of the solvent can be determined by the Gibbs-Duhem integration (see section
2.11). In practice, the calculation is prohibitive, because of the chemical complex-
ity of most aqueous solutions of geochemical interest. Semiempirical approxima-
tions are therefore preferred, such as that proposed by Helgeson (1969), con-
sisting of a simulation of the properties of the H2O-NaCl system up to a solute
Figure 8.13 Activity of H2O solvent in aqueous solutions of various solute molalities.
Reprinted from Garrels and Christ (1965), with kind permission from Jones and Bartlett
Publishers Inc., copyright 䉷 1990.
molality of 2. The approximation adopts parameter IE , defined as “equivalent

stoichiometric ionic strength” and numerically equivalent to the molal concentra-
tion of the Na⫹ or Cl ⫺ in solution. The calculation proceeds as follows:
−2 Iϕ ′
log a H 2 O = , ( 8.43)
2.303ω H 2 O
where ␸⬘ is the osmotic coefficient of water and ␻H2O is the number of moles of
H2O per 103 g of solvent ( ␻H2O ⫽ 1000
18 ). The osmotic coefficient is then calculated
with the polynomial expansion
1 1 1
ϕ ′ = 1 − D′J ′ + W 2 I E + W 3 I E2 + W 4 I E3 , ( 8.44 )
2 3 4
where
2.303A
D′ = ( 8.45 )
W13 I E
1
J ′ = B ′ − 2 ln B ′ − ( 8.46 )
B′
B ′ = 1 + W1 I E1/2 , ( 8.47 )
with W1, W2, W3, and W4 being experimentally determined constants.
8.9 Speciation
As already stated, “speciation” is the characteristic distribution of various ionic

and/or neutral species in an aqueous solution. Speciation calculation, allowing
practical estimation of the reactive properties of an aqueous solution, acidity,
redox state, the degree of saturation of the various solids, and so on, is carried
out on a thermodynamic basis starting from the chemical composition of the
solution of interest and using the reaction constants of the various equilibria of
the type seen in equation 8.19.
When performing the chemical analysis of an aqueous solution, we obtain a
set of values representing the bulk concentration of dissolved components, but
we do not discriminate the various forms of solutes in which a given species is
partitioned. For instance, we can measure the total molal amount of calcium
mTCa or fluorine mTF, but this value is the sum of all partial molalities of ionic and
neutral species containing Ca or F—i.e.,
T
mCa = mCa2 + + mCaOH + + mCaCO0 + mCaHCO + + mCaHPO0 + L ( 8.48 )
3 3 4
mFT = mF − + mHF 0 + 2 mH 0 + 2 mHF − + 3mAlF 0 + L ( 8.49 )

2 F2 2 3
Because we are generally able to define the chemistry of an aqueous solution

containing n chemical elements by analytical procedures, n equations such as 8.48
and 8.49 exist, relating the bulk concentration of a given element mTi to all species
actually present in solution. Associated with mass balance equations of this type
may be a charge balance equation expressing the overall neutrality of the solution:
∑ Z i m i = 0, ( 8.50)
i
where Zi is the net charge of species i, and mi is its partial molality.

Besides the above-quoted equations, a set of homogeneous reactions also ex-
ists, relating the individual ionic activities of all the complex ions in solution—
e.g.,
CaOH+ ⇔ Ca 2 + + OH − ( 8.51)
HF 0 ⇔ H + + F − ( 8.52 )
AlF30 ⇔ Al 3 + + 3 F − . ( 8.53)
These equilibria may be expressed in the forms
log K 51 = log a Ca 2 + + log a OH − − log a CaOH + = log mCa 2 + + log mOH −

(8.54)
− log mCaOH + + log γ Ca 2 + + log γ OH − − log γ CaOH +
log K 52 = log aH + + log aF − − log aHF 0 = log mH + + log mF −

(8.55)
− log mHF 0 + log γ H + + log γ F − − log γ HF 0
log K 53 = log a Al 3 + + 3 log aF − − log aAlF 0 = log mAl 3 + + 3 log mF −

3
(8.56)
− log mAlF 0 + log γ Al 3 + + 3 log γ F − − log γ AlF 0 .
3 3
Because the thermodynamic constants of the various equilibria of the type exem-
plified by equations 8.51 to 8.53 can easily be calculated from the partial molal
thermodynamic properties of the solutes, and the individual ionic activity coeffi-
cients may also be calculated, the unknowns in equations 8.54 to 8.56 are repre-
sented simply by partial molalities. When equations such as 8.48 and 8.49 are
combined with equations such as 8.54 to 8.56 and with equation 8.50, the system
is determined and can be solved by iterative procedures.
8.9.1 Generalities on Complexing of Metal Cations as a

Function of Their Elemental Properties
The simplest approach to describing the interactions of metal cations dissolved
in water with solvent molecules is the Born electrostatic model, which expresses
solvation energy as a function of the dielectric constant of the solvent and,
through transformation constants, of the ratio between the squared charge of the
metal cation and its effective radius. This ratio, which is called the “polarizing
power” of the cation (cf. Millero, 1977), defines the strength of the electrostatic
interaction in a solvation-hydrolysis process of the type
H  H 
/  / 
→  M − O
M Z + O solvation
  hydrolysis
 →[M − O − H] ( Z −1) + + H + . (8.57)
\  \ 
H  H 
The stronger the polarizing power of the cation, the stronger the tendency of
reaction 8.57 to proceed rightward. Figure 8.14 shows the relationship between
polarizing power and the degree of hydrolysis of aqueous cations in a solution
with ionic strength I ⫽ 0.65, pH ⫽ 8.2, T ⫽ 25 °C, and P ⫽ 1 bar according to
Turner et al. (1981). In the logarithmic plot of figure 8.14, the ordinate axis is the
coefficient of partial reaction (i.e., ratio of hydrolyzed cation with respect to total
concentration). The abscissa axis does not properly indicate the polarizing power,
because authors have adopted Shannon and Prewitt’s (1969) ionic radii (see
Turner et al., 1981, for details).
Five elemental groups can be distinguished on the basis of a diagram such as
figure 8.14:
Group 1: cations with Z2 /r ⬍ 2.5 Å⫺1

Group 2: cations with 2.5 ⬍ Z2 /r ⬍ 7 Å⫺1
Group 3: cations with 7 ⬍ Z2 /r ⬍ 11 Å⫺1
Group 4: cations with 11 ⬍ Z2 /r ⬍ 25 Å⫺1
Group 5: cations with Z2 /r ⬎ 25 Å⫺1.
Observe the sharp transition from zero hydrolysis to complete hydrolysis in the
region 7 ⬍ Z2 /r ⬍ 25 Å⫺1. The fully hydrolyzed elements belong to groups IV
Figure 8.14 Effect of polarizing power on degree of hydrolysis of cations in water. Re-
printed from D. R. Turner, M. Whitfield, and A. G. Dickson, Geochimica et Cosmochimica
Acta, 45, 855–881, copyright 䉷 1981, with kind permission from Elsevier Science Ltd.,
The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
and V. For these elements (except for uranium, present as uranyl ion UO2⫹2 , whose
speciation is dominated by a strong interaction with carbonate ion CO2⫺ 3 ), com-
plexing with any other anionic ligands that may be present in aqueous solution is
negligible. Speciation of the remaining cations with the various anionic ligands is
related to their tendency to form partially covalent bondings. The arrangement
of the various groups on the periodic chart, according to the classification of
Turner et al. (1981), is shown in figure 8.15.
Ahrland’s (1973) classification distinguishes incompletely hydrolyzed ions
into types a and b. Cations of type a form stable complexes with electronic donors
of the first row of the periodic chart through electrostatic interactions; those of
Figure 8.15 Arrangement of various ions from classification of Turner et al. (1981) on
periodic chart. Reprinted from D. R. Turner, M. Whitfield, and A. G. Dickson, Geochim-
ica et Cosmochimica Acta, 45, 855–881, copyright 䉷 1981, with kind permission from
type b form stable complexes with electronic donors of the second, third, and
fourth rows (P, S, Cl: As, Se, Br; Sb, Te, I) through essentially covalent bondings.
Turner et al. (1981) suggested a classification based on the different stabilities of
the chloride and fluoride complexes of the cations. Denoting the stability differ-
ence as ⌬K:
∆K = log K MF − log K MCl , ( 8.58)
where KMF and KMCl are complexing constants at reference conditions T ⫽ 25 °C

and P ⫽ 1 bar, four distinct cation groups arise:
Type a cations: ⌬K ⬎ 2
Type a⬘ cations: 2 ⬎ ⌬K ⬎ 0
Type b cations: 0 ⬎ ⌬K ⬎ ⫺2
Type b⬘ cations: ⌬K ⬍ ⫺2.
Figure 8.16 shows how the complexing constants of ions vary with the polar-
izing power of the cation, ligand being equal. Note also here that the abscissa
Figure 8.16 Complexing constants of various cations with different anionic ligands ar-
ranged according to ratio between squared charge and ionic radius (plus a constant term),
where ␴ is the standard deviation on regression. Reprinted from D. R. Turner, M. Whit-
field, and A. G. Dickson, Geochimica et Cosmochimica Acta, 45, 855–881, copyright 䉷
1981, with kind permission from Elsevier Science Ltd., The Boulevard, Langford Lane,
Kidlington 0X5 1GB, UK.
axis does not strictly represent polarizing power, because the effective radius is
replaced by the ionic radius plus a constant term.
We will now examine in detail the progression of hydrolysis reactions and
their influence on the stability in solution of hydrolyzed cations.
8.9.2 Energies of Hydrolysis and Solvation Processes

The initial step in the hydrolysis of a generic cation M is usually described by
the equilibrium
MZ + + H2 O ⇔ MOH( Z −1) + + H+ . ( 8.59 )
A more realistic representation of the actual process is given by
( ) ( )( )
Z+ ( Z −1 ) +
M OH 2 ⇔ M OH OH 2 + H+ ( 8.60)
n n −1
(Baes and Mesmer, 1981), where there is a proton loss by solvation of a water
molecule. In light of equation 8.60, it is not surprising to observe that the enthalpy
change associated with the process is close to the dissociation enthalpy of water
(13.3 kcal/mole at T ⫽ 25 °C and P ⫽ 1 bar).
If the metal concentration in solution is sufficiently low to remain below the
solubility product, increasing the pH, with molal concentration, T, and P re-
maining equal, hydrolysis proceeds by further solvation of H2O molecules, with
proton loss according to the scheme
( ) ( )
( Z −1 ) + (Z −2 )+
M Z + → M OH → M OH . ( 8.61)
2
The process can be generalized as follows:
( ) ( )
( Z −Y ) + ( Z −Y −1) +
M OH + H2 O → M OH + H+ , ( 8.62 )
Y (Y +1)
with Y increasing from zero.

The constant of reaction 8.62 is
 ( ) (Y +1 )  [ ]
 M OH ( Z −Y −1 ) +  H +
K 62 = , ( 8.63)
 ( )
 M OH ( Z −Y ) + 
Y 
where the terms in square brackets denote thermodynamic activity.

Baes and Mesmer (1981) observed a simple relationship between the entropy change
involved in equilibrium 8.62 at reference T and P conditions ( ⌬S62,Tr ,Pr ) and the charge of
species in reaction (Z-Y)⫹:
 cal 
∆S62 , Tr , Pr = −17.8 + 12.2( Z − Y )  . ( 8.64 )
 mole × K 
Because
∆G62 , Tr , Pr = ∆H 62 , Tr , Pr − Tr ∆S62 , Tr , Pr = −2.303RT log K 62 , Tr , Pr , ( 8.65)
combining equations 8.64 and 8.65 gives
 kcal 
∆H 62 , Tr , Pr = −1.36 log K 62 , Tr , Pr − 5.3 + 3.64( Z − Y )  . ( 8.66 )
 mole 
At cation molal concentrations exceeding 10⫺3 moles per kilogram of solvent,

polynuclear species may be formed, with processes of the type
( )
( XZ −Y ) +
X MZ + + Y H2O ⇔ MX OH + Y H+ , (8.67)
Y
whose equilibrium constant is
K 67 = M X OH ( ) [ ] [M ]
 H+
( XZ −Y ) + Y −X
Z+
. (8.68)
 Y 
The concentrations of the various hydrolyzed species at saturation are con-

trolled by the solubility product, with equilibria of the type
( )
M OH
Z
+ Z H+ ⇔ MZ + + Z H2O (8.69)
[ ][ ]
−Z
K 69 = M Z + H + (8.70)
( ) + (XZ − Y ) H ( ) ( )
( XZ −Y ) +
X M OH +
⇔ MX OH + XZ − Y H2O (8.71)
Z Y
K 71 = M X OH ( ) [ ]
 H+
( XZ −Y ) + − ( XZ −Y )
. (8.72)
 Y 
Figure 8.17 shows equilibrium relations between hydrolyzed cations and their
saturation limits as a function of pH. In this semilogarithmic plot, the saturation
Figure 8.17 Equilibria in solution between various hydrolyzed species. From Baes and
Mesmer (1981), American Journal of Science, 281, 935–62. Reprinted with permission of
American Journal of Science.
limit defined by solubility product K71 is a straight line with slope XY-Z as a
function of pH (disregarding minor effects of the activity coefficients). This is
true for all species forming during the various hydrolysis steps with respect to
their saturation limits. The overall saturation curve (bold line) is given by the
envelope of the saturation limits valid for each single species. The predominance
limits of the hydrolyzed species (isoactivity condition of neighboring species) are
simply defined by pH (see eq. 8.59 and 8.62). At high metal concentrations, a
field of predominance of polynuclear species opens, whose boundaries are defined
by equilibria of the type exemplified by equation 8.67.
The above-discussed equilibria are of fundamental importance for an under-
standing of alteration processes, with special emphasis on the role of pH in water-
rock interaction events.
8.10 Carbonate Equilibria
Besides controlling the redox conditions of seawater, as we will see in some detail
in section 8.10.1, carbonate equilibria are important when investigating sedimen-
tary rocks, of which carbonates are major components. The procedure normally
followed in geochemistry in establishing such equilibria is the classical approach
of Garrels and Christ (1965), later taken up by other authors (e.g., Helgeson,
1969; Holland, 1978; Stumm and Morgan, 1981). Five separate equilibria are
taken into consideration in this model, relating aqueous (a), gaseous ( g), and
condensed (c) phases:
H +( a ) + CO 23 (−a ) ⇔ HCO−3 ( a ) ( 8.73)
H(+a ) + HCO3−( a ) ⇔ H2CO∗3( a ) (8.74)
H2O( a ) ⇔ H+( a )+ OH−( a ) (8.75)
CO2( g) + H2O( a ) ⇔ H2CO∗3( a ) (8.76)
CaCO3 ( c ) ⇔ Ca 2( a+) + CO 23 (−a ) . ( 8.77 )
The first three equilibria relate the activities of the main aqueous species, the
fourth establishes the relationship between gaseous and aqueous phases, and the
fifth establishes the relationship between aqueous species and the condensed
phase.
The speciation state determined by the five equilibria depends on the type of
system (closed or open) and imposed conditions ( fCO2, pH). Two cases are partic-
ularly relevant in geochemistry:
1. Equilibrium between CaCO3(c) and an aqueous phase with fCO2 buffered

by an external reservoir (atmosphere)
2. Equilibrium between CaCO3(c) and an aqueous phase with a constant
quantity of solutes and fixed pH.
Before we examine these two cases in detail, it is necessary to introduce some precau-
tionary considerations.
1. Equilibria are normally treated by considering calcite as the only carbonate poly-
morph present in the system. Actually, as already mentioned, CaCO3(c) crystal-
lizes in two forms: calcite (trigonal) and aragonite (orthorhombic). Aragonite,
which at low pressure is metastable with respect to the less dense calcite (cf. table
2.1), is in fact synthesized by several marine organisms that fix CaCO3(c) in this
form in their shells. Nevertheless, the difference in the Gibbs free energy of forma-
tion of the two polymorphs is so limited (0.2 to 0.5 kcal/mole) that the choice of
type of phase does not influence calculations.
2. Carbonates in sedimentary rocks are mixtures of several components: CaCO3 ,
CaMg(CO3)2 , FeCO3 , SrCO3 , and BaCO3 . Calculations are usually developed
assuming that the condensed phase is pure calcite and assigning unitary activity
to the CaCO3 component. For accurate calculations, we would introduce an ap-
propriate activity value, depending on the actual chemistry of the solid and bear-
ing in mind that carbonates are not at all ideal mixtures but have wide miscibility
gaps (cf. figure 3.11).
3. Aqueous solutions of carbon dioxide contain both CO2(a) and H2CO3(a) neutral
molecules. The CO2(a) molecule has a linear structure (O⫽C⫽O), whereas
that of H2CO3(a) is planar trigonal because of the molecular orbitals constituted
of hybrid AO sp2. Conversion from the linear form CO2(a) to the trigonal-planar
form H2CO3(a) —i.e.,
CO2( a ) + H2O( a ) → H2CO3( a ) ( 8.78)
—is relatively slow (about 3% of CO2(a) is converted into H2CO3(a) per second in
room conditions; cf. Kern, 1960). The dehydration process
H2CO3( a ) → CO2( a ) + H2O( a ) ( 8.79)
is relatively faster (net rate constant Kr ⫽ 2.0 ⫻ 10⫺3 sec⫺1 at 0 °C, according to
Jones et al., 1964).
The equilibrium kinetics of reactions 8.78 and 8.79 show that, at room P and T
and in stationary conditions, the CO2(a) /H2CO3(a) ratio is near 600 (Kern, 1960).
Notwithstanding this fact, in geochemistry it is customary to represent aqueous
CO2 as if it were entirely in the form of carbonic acid H2CO3(a), not distinguishing
planar from trigonal forms (i.e., H2CO* 3(a) in eq. 8.74). This has little connection
with reality and may lead to errors unless all equilibria are explicitly defined ac-
cording to a single value of the Gibbs free energy of formation for aqueous neu-
tral species.
8.10.1 Calcite in Equilibrium with an Aqueous Phase at Low

Salinity and with a Gaseous Phase at Constant CO2
Fugacity
Because the electrolyte H2CO3(a) does not directly influence the acidity of the
aqueous phase, it is preferable to avoid the lack of precision inherent in the defini-
tion of the form of the aqueous CO2 by replacing equation 8.76 by
CO2 ( g) + H2O( a ) ⇔ HCO3−( a ) + H(+a ) (8.80)
CO2 ( g) + H2O( a ) ⇔ CO23(−a ) + 2 H(+a ) . (8.81)
The amount of H2CO* 3(a) may then be determined through equation 8.74.
Table 8.6 lists equilibrium constants for reactions 8.73, 8.77, 8.80, and 8.81
along the water-vapor univariant curve. We use values of T ⫽ 25 °C and P ⫽ 1
bar in the following calculations.
Let us equilibrate some pure calcite crystals with an aqueous phase at T ⫽
25 °C and P ⫽ 1 bar. Calcite partially dissolves to reach the solubility product
(eq. 8.77).
Table 8.6 Constants of carbonate equilibria at various P and T conditions along univariant
water-vapor boundary
P (bar) T (°C) log K77 log K77(*) log K73 log K80 log K81
1 25 ⫺8.616 ⫺8.451 10.329 ⫺7.818 ⫺18.147
0.006 0 ⫺8.485 ⫺8.317 10.553 ⫺7.683 ⫺18.236
0.123 50 ⫺8.771 ⫺8.608 10.123 ⫺7.988 ⫺18.111
1.013 100 ⫺9.072 ⫺8.912 9.759 ⫺8.385 ⫺18.144
4.758 150 ⫺9.359 ⫺9.198 9.434 ⫺8.831 ⫺18.265
15.537 200 ⫺9.652 ⫺9.487 9.115 ⫺9.323 ⫺18.438
39.728 250 ⫺9.987 ⫺9.817 8.751 ⫺9.890 ⫺18.641
85.805 300 ⫺10.459 ⫺10.282 8.235 ⫺10.626 ⫺18.861
165.125 350 ⫺11.480 ⫺11.297 7.117 ⫺11.973 ⫺19.090
( )
* With aragonite as condensed phase at equilibrium.
Because the activity of component CaCO3(c) in the pure calcite phase is 1, we

can write
[Ca ][CO ] = K
2+ 2−
3 77 = 10 −8.62. (8.82)
Based on equations 8.80 and 8.81 and assuming unitary fugacity for CO2(g) at
standard state (see section 9.1), we can write
[ HCO ][ H ] ⋅ f
−
3
+ −1
CO 2( g )
= K 80 = 10 −7.82 (8.83)
[CO ][ H ]
2
2−
3
+
⋅ fCO
−1
= K 81 = 10 −18.15 (8.84)
2( g )
and, based on equation 8.73,
[ HCO ][ H ] [CO ]
−1 −1
−
3
+ 2−
3 = K 73 = 1010.33. (8.85)
We also recall that the ionic dissociation constant of pure water under room con-
ditions is
[ H ][OH ] = 10
+ − −14
. (8.86)
Let us now express the activities of the main ionic species in solution as a
function of the activity of hydrogen ions and f CO2(g). Combining equations 8.82
to 8.86, we obtain
[Ca ] = 10 [ H ]
2
2+ 9.53 +
⋅ fCO
−1
(8.87)
2( g )
[ HCO ] = 10 [ H ]
−1
−
3
−7.82 +
⋅ fCO 2( g ) (8.88)
[CO ] = 10 [H ]
−2
2−
3
−18.15 +
⋅ fCO 2( g ) (8.89)
[OH ] = 10 [H ]
−1
− −14.00 +
. (8.90)
System 8.87→8.90 is composed of four equations with six unknowns. If we fix

the fugacity of gaseous CO2 , the unknowns are reduced to five. Note, moreover,
that equation 8.85 has virtually no influence, because it can be obtained by adding
equations 8.80 and 8.81. To determine the system, we introduce the electroneu-
trality condition
2 m Ca 2 + + m H + = 2 m CO 2 − + m HCO− + m OH− ( 8.91)

3 3
(cf. section 8.9).

Let us now assume that, at first approximation, all activity coefficients of
aqueous species are 1. Equation 8.91 becomes
[ ] [ ] [ ] [
2 Ca 2 + + H + = 2 CO 23 − + HCO −3 + OH − . ] [ ] (8.92)
Combining equation 8.92 with the system 8.87→8.90, we obtain a polynomial in

[H⫹ ] that may be solved with iterative procedures for various values of f CO2(g):
[ ] + [H ]
4 3
10 9.83 ⋅ f CO
−1
⋅ H+ +
2( g )
(8.93)
( )[ ]
− 10 −7.82 ⋅ f CO2 ( g ) + 10 −14 H + − 10 −17.85 ⋅ f CO2 ( g ) = 0.
Figure 8.18 shows how equation 8.93 behaves in the vicinity of zero. Equilibrium
pH corresponds to the zero value on the ordinate axis.
If we adopt the present-day atmospheric f CO2(g) value (10⫺3.48 ), the corre-
sponding pH, obtained through application of equation 8.93, is 8.2. It is well
known that the amount of atmospheric CO2 is steadily increasing, as a result
of intensive combustion of fossil fuels. It has been deduced, from present-day
exploitation rates and projections, that the amount of CO2(g) in the atmosphere
will double in the next 40 to 70 years (i.e., fCO2(g) will increase from 10⫺3.48 to
10⫺3.17 ). Correspondingly, the pH will fall by about 0.2 unit.
Equation 8.93 is asymptotic, with a flat flexus near zero (figure 8.18). Obviously, this
type of equation does not allow precise pH determinations, because even slight approxi-
Figure 8.18 Relationship between pH of aqueous solution and fCO2 of gaseous phase in
equilibrium with water and calcite. Equilibrium pH is in correspondance with zero on
ordinate axis.
mations in the equilibrium constants generate large variations in the solving value (the
small pH differences between the procedure of Garrels and Christ, 1965, and the one
developed here at quasi-parity of fCO2(g) must be ascribed to this sort of problem). The
method therefore cannot be used for high ionic strength or chemically complex solutions,
in which the electroneutrality condition cannot be simplified in the form of equation 8.91.
In seawater, for instance, only 88% of calcium is in the form of Ca2⫹ and 12% is in the
aqueous sulfate form of CaSO04. Similarly, the bicarbonate ion constitutes only 73% of
total carbonate in solution: 13% is present as magnesium bicarbonate (MgHCO3⫹ ), 5.5%
as magnesium carbonate (MgCO03 ), 2.8% as calcium bicarbonate (CaHCO3⫹ ), and only
2.1% as carbonate ion. Clearly, for seawater the carbonate equilibria of equations
8.73→8.77 constitute only a particular subsystem of the n equilibria that actually deter-
mine the speciation state.
Nevertheless, it is of interest to note that precise speciation calculation conducted
with an automated routine clearly indicates that, at the pH observed in seawater (8.22),
the activities of Ca2⫹ and CO32⫺ in solution are near the solubility product of calcite and
fCO2(g) is near the atmospheric value (i.e., fCO2(g) ⫽ 10⫺3.35 ). We can thus deduce that car-
bonate equilibria are effectively determinant in seawater and that they are conditioned by
the atmospheric value of fCO2(g). Table 8.7 shows that the increase in CO2 in the atmo-
sphere determines a progressive decrease of pH in seawater, in line with what was pre-
viously deduced for an aqueous solution of low salinity. In light of this marked control by
atmospheric CO2 on the acidity of aqueous solutions, during geochemical samplings of
deep waters, direct contact with the atmosphere must be avoided before pH measurements
are taken, because otherwise the pH turns out to be intermediate between the in situ value
and the new equilibrium situation. Note here that, according to the calculation of Garrels
and Christ (1965), the pH of an aqueous solution in equilibrium with calcite at T ⫽ 25
Table 8.7 Relationships between ƒ C O 2 ( g ) and pH in carbonate equilibria at T ⫽ 25 ° C and

P ⫽ 1 bar. (1) ⫽ calculated through equation 8.93; (2) ⫽ calculated with EQ3NR/EQ6
routine; (3) ⫽ Garrels and Christ (1965). A ⫽ seawater in equilibrium with calcite and
atmospheric CO2; B ⫽ seawater in equilibrium with calcite; C ⫽ seawater in equilibrium
with atmospheric CO2.
A B C
Parameter (1) (1) (3) (3) (2) (2) (2)

log ƒCO2(g) ⫺3.17 ⫺3.48 ⫺3.50 - ⫺3.17 ⫺3.35 ⫺3.48
pH 8.00 8.20 8.40 9.90 8.06 8.22 8.32
log[Ca2⫹] ⫺3.30 ⫺3.39 ⫺3.40 ⫺3.90 ⫺2.65 ⫺2.65 ⫺2.65
log[HCO3⫺] ⫺2.99 ⫺3.10 ⫺3.00 ⫺4.05 ⫺2.95 ⫺2.98 ⫺3.00
log[CO32⫺] ⫺5.32 ⫺5.23 ⫺4.90 ⫺4.40 ⫺5.22 ⫺5.08 ⫺5.01
log[OH⫺] ⫺6.00 ⫺5.80 ⫺5.60 ⫺5.10 ⫺5.94 ⫺5.78 ⫺5.68
°C and P ⫽ 1 bar and in the absence of atmospheric CO2 is 9.9—i.e., much higher than
the value obtained for equilibrium with atmosphere (cf. table 8.7).
8.10.2 Calcite in Equilibrium with an Aqueous Phase at

Fixed pH
This is the case generally encountered in geochemical investigations: a deep water
is sampled and total carbonates and pH are determined to solve the question
of whether or not the sampled water was in equilibrium with calcite (or other
carbonates) at the T and P of sampling. The system is clearly determined; the
equilibria to be considered in this case are those of equations 8.73, 8.74, and 8.77
(eq. 8.76, 8.80, and 8.81 have no influence).
As a specific example of the problem, let us calculate the equilibria for an
actual case study: a deep water from the Sarcidano region (Sardinia, Italy) in
equilibrium with Mesozoic dolomites (Bertorino et al., 1981). The compositions
in mEq/l of water sampled in a drilled well are listed in table 8.8. The in situ
temperature is 21 °C; we assume here that the in situ T is 25 °C at 1 bar, to
simplify calculations. We also assume for the sake of simplicity that the main ion
species in solution are HCO3⫺, Mg2⫹, Ca2⫹, CO2⫺ ⫺ ⫹
3 , OH , and H , and that all
Ca and Mg are in the ionic forms Ca2⫹ and Mg2⫹.
The ionic strength of the solution is I ⫽ 0.014. Applying the Debye-Hückel
equation (eq. 8.41), the appropriate solvent parameters A and B, and the effective
diameters for the ions in solution listed in table 8.5, we obtain the following activ-
ity coefficients: ␥HCO⫺ 3
⫽ 0.90, ␥Mg2⫹ ⫽ 0.68, ␥Ca2⫹ ⫽ 0.66, ␥CO32⫺ ⫽ 0.65, ␥OH⫺ ⫽
0.90, and ␥H⫹ ⫽ 0.91.
Let us admit that total inorganic carbon represents the summation of par-
tial molalities:
mH * + mHCO− + mCO2 − = 8.1 × 10 −3 , ( 8.94 )

2 CO 3( a ) 3( a ) 3( a )
with
mH * = mH2 CO 3( a ) + mCO2 ( a ) . ( 8.95)

2 CO 3( a )
Equilibrium 8.74 has constant
γH * ⋅ mH *
2 CO 3 2 CO 3
K 74 = = 10 6.4. (8.96)
γ H + ⋅ mH + ⋅ γ HCO− ⋅ mHCO−
3 3
With
γH * = 1, γ H + ⋅ mH + = 10 −7.3 , and γ HCO− = 0.90,

2 CO 3 3
we obtain
mH * = 10 −0.94 ⋅ mHCO− . ( 8.97 )

2 CO 3 3
Equation 8.73 yields
m HCO− ⋅ γ HCO−3
K 73 = 3
= 1010.33 . (8.98)
γ H+
⋅ m H+ ⋅ γ CO 23 −
⋅ m CO 2 −
3
Table 8.8 Compositions of superficial and deep waters of the Sarcidano region (Sardinia, Italy)
equilibriated with Mesozoic dolomite limestones (Bertorino et al., 1981). Values in mEq/l. CT total
inorganic carbon in mmol/l.
Sample T (°C) Type pH Ca Mg Na K CT SO4 Cl

27 14 Superficial 7.1 4.4 4.7 0.9 0.04 10 1.1 0.9
28 14 Superficial 7.1 3.5 3.5 0.6 0.02 8.1 0.7 0.7
29 16 Superficial 7.6 4.2 3.8 1.0 0.04 8.4 1.0 0.9
30 15 Superficial 7.1 3.7 4.6 1.1 0.03 9.4 0.6 1.1
36 14 Superficial 7.1 4.2 4.5 0.9 0.04 9.7 1.0 1.0
37 13 Superficial 6.9 4.8 5.2 0.9 0.05 12 1.0 1.0
38 14 Superficial 7.1 3.4 3.3 0.6 0.02 7.5 1.1 0.7
39 13 Superficial 7.1 3.9 3.9 0.6 0.02 9.1 0.2 0.7
40 15 Superficial 7.1 3.2 3.2 0.6 0.02 7.3 1.0 0.6
79 21 Well 7.3 3.5 3.5 1.5 8.1 8.1 0.9 1.4
With ␥CO2⫺
3
⫽ 0.65, we obtain
m CO 2 − = 10 −2.888 × m HCO− . ( 8.99)

3 3
From the equation of total inorganic carbon (eq. 8.94), combined with equa-
tions 8.97 and 8.99, we derive
10 −0.94 × mHCO− + mHCO− + 10 −2.888 × mHCO− = 10 −2.092 . ( 8.100 )

3 3 3
Thus, mHCO⫺ 3
⫽ 10⫺2.139, mH2CO*3 ⫽ 10⫺3.079, and mCO2⫺
3
⫽ 10⫺5.027.
From the total molality of Ca, assuming all calcium to be present in solution
in the ionic form Ca2⫹, we obtain
γ Ca 2 +
⋅ m Ca 2 + = 0.66 × 0.00175 = 10 −2.937 . ( 8.101)
The activity product derived for Ca2⫹ and CO2⫺

3 is thus
[ ][ ]
Q = Ca 2 + CO 23 − = 0.65 × 10 −5.027 × 10 −2.937 = 10 −8.151. (8.102)
The solubility product of calcite at T ⫽ 25 °C and P ⫽ 1 bar is K ⫽ 10⫺8.62.

The log Q/K ⫽ 0.469 value indicates slight oversaturation in calcite with an
affinity to precipitation (cf. section 2.12):
Q  kcal 
AP = RT ln = 0.640  . (8.103)
K  mole 
Table 8.9 shows the results of detailed speciation calculations carried out with
the automated routine EQ3NR/EQ6 (Wolery, 1983). The molalities of the ions
in solution resulting from this calculation are tabulated in order of decreasing
abundance.
In the total inorganic carbon balance of equation 8.94, we neglected the presence of
some species. Table 8.9 shows that this approximation is reasonable although, for accurate
calculations, we should also have considered bicarbonates CaHCO3⫹ and MgHCO3⫹ and
the soluble carbonate CaCO30. Note also that sulfates occur in negligible molar propor-
tions and that carbonate alkalinity coincides with total alkalinity.
The precise speciation calculation indicates that four minerals are at almost perfect
saturation:
Calcite: log Q/K ⫽ 0.359; AP ⫽ 0.490 kcal/mole

Disordered dolomite: log Q/K ⫽ 0.195; AP ⫽ 0.265 kcal/mole
Table 8.9 Speciation of a deep water from Sarcidano

(sample 79 in table 8.8) according to EQ3NR/EQ6 procedure.
Species log m log ␥ log a

HCO3⫺ ⫺2.1548 ⫺0.0466 ⫺2.2014
Mg2⫹ ⫺2.7570 ⫺0.1697 ⫺2.9267
Ca2⫹ ⫺2.7570 ⫺0.1796 ⫺2.9366
Na⫹ ⫺2.8239 ⫺0.0473 ⫺2.8712
H2CO3 ⫺3.1553 0.0013 ⫺3.1540
CaHCO3⫹ ⫺3.8423 ⫺0.0456 ⫺3.8879
K⫹ ⫺3.8539 ⫺0.0488 ⫺3.9027
MgHCO3⫹ ⫺3.9945 ⫺0.0473 ⫺4.0418
CaCO30 ⫺4.9678 0.0 ⫺4.9678
CO32⫺ ⫺5.0417 ⫺0.1878 ⫺5.2295
MgCO3 ⫺5.1750 0.0 ⫺5.1750
OH⫺ ⫺6.6393 ⫺0.0481 ⫺6.6874
NaCO3⫺ ⫺7.0938 ⫺0.0466 ⫺7.1404
H⫹ ⫺7.2591 ⫺0.0409 ⫺7.3000
Mg(OH)⫹ ⫺7.3537 ⫺0.0473 ⫺7.4010
Cl⫺ ⫺7.4036 ⫺0.0488 ⫺7.4524
Ca(OH)⫹ ⫺8.1772 ⫺0.0473 ⫺8.2245
SO42⫺ ⫺8.4411 ⫺0.1907 ⫺8.6318
CaSO40 ⫺9.2570 0.0 ⫺9.2570
MgSO40 ⫺9.3109 0.0 ⫺9.3109
NaOH ⫺10.2586 0.0 ⫺10.2586
NaSO4⫺ ⫺10.7365 ⫺0.0469 ⫺10.7834
NaCl ⫺11.2362 0.0 ⫺11.2362
KSO4⫺ ⫺11.6340 ⫺0.0481 ⫺11.6821
HSO4⫺ ⫺13.8794 ⫺0.0473 ⫺13.9267
HCl ⫺20.8863 0.0 ⫺20.8863
O2(a) ⫺22.9945 0.0013 ⫺22.9932
H2(a) ⫺34.6176 0.0013 ⫺34.6163
HS⫺ ⫺108.2457 ⫺0.0481 ⫺108.2937
H2S2(a) ⫺108.6132 0.0 ⫺108.6132
S2⫺ ⫺117.9757 ⫺0.1850 ⫺118.1607
Aragonite: log Q/K ⫽ 0.194; AP ⫽ 0.264 kcal/mole

Magnesite: log Q/K ⫽ ⫺0.249; AP ⫽ ⫺0.339 kcal/mole.
Our approximation in the calculation is 150 cal/mole, which is quite satisfactory. The fact
that dolomite is the mineral phase nearest to complete equilibrium is in agreement with
the geology of the area and was in fact already evident at first sight from the chemical
analyses, indicating similar molalities for Ca and Mg in solution. The resulting fCO2(g) is
10⫺1.67, which is much higher than the atmospheric value. The previous recommendation
about pH measurements is particularly obvious in this case.
In geology we often encounter conditions that are intermediate with respect

to the two limiting cases mentioned above. For instance, a rainwater initially at
equilibrium with atmospheric CO2 percolates through nonreactive soils and
reaches a carbonate layer: this is clearly a combination of the two cases discussed
in this section and in section 8.10.1. Calculations on a semiquantitative basis
in such a case have only academic interest (see Garrels and Christ, 1965, for a
detailed treatment).
8.11 Thermodynamic Properties of Solutes

Under High P and T Conditions:
the Helgeson-Kirkham-Flowers Approach
Study of water-rock interactions occurring within the continental crust and along
mid-ocean ridges, where large masses of water are involved in convective trans-
port within the lithosphere, requires accurate estimates of the partial molal prop-
erties of solutes under high P and T conditions. In other words, in order to deter-
mine the reactivity of a given aqueous solution toward a given mineralogical
assemblage with which it comes into contact, a detailed speciation calculation at
the P and T of interest is necessary. As we have already seen in section 8.9, here
we need knowledge of the various equilibrium constants among solutes, and these
constants, to be determined, in turn require knowledge of the partial molal Gibbs
free energies of the solutes at the P and T of interest. The energy model developed
by Helgeson, Kirkham, and Flowers (1981) and later revised by Tanger and Hel-
geson (1988) allows accurate calculation of the partial molal properties of ions in
aqueous solutions up to a maximum P of 5 kbar and a maximum T of 1000 °C.
A brief outline of the procedure is given here (see the works of Helgeson and co-
authors for more exhaustive explanations).
8.11.1 Standard State Conventions and the

Additivity Principle
The standard state adopted in the Helgeson-Kirkham-Flowers (HKF) model for aqueous
electrolytes and ions in solution is the classical one of a “hypothetical one-molal solution
referred to infinite dilution at each P and T condition” (see section 8.4). The standard
state for the solvent is that of a “pure solvent at each P and T condition.” Let us recall
briefly the significance of the “partial molal property” of a solute in an aqueous solution.
Chapter 2 introduced the concept of the chemical potential of a component in mixture as
partial derivative of the Gibbs free energy of the phase with respect to the number of
moles of component (eq. 2.3). Because the Gibbs free energy of the mixture is determined
by the standard state energy of the pure components in their molar proportions plus the
mixing terms (ideal plus excess), the chemical potential of a given component in mixture
is not a constant, but varies in a more or less complex fashion with the chemistry of the
phase as a function of interactions of all the components in that mixture. We have also
seen that we can consider the chemical potential of the component as made up of partial
molar properties (eq. 2.59, 2.60, 2.61). These properties (i.e., partial molar enthalpy, par-
tial molar entropy, partial molar volume) represent the discrete values assumed by the
molar property at any particular composition of the phase in the system. Partial molar
properties are related, in the terms outlined by equations 2.31 to 2.34 and 2.1, to the Gibbs
free energy of the phase (and of the system). The significance of the partial molal proper-
ties of solutes in aqueous solutions are obviously similar, although some distinction must
be made:
1. These properties are referred to the molality of the solute (i.e., moles of solute
per 103 g of solvent) and not to molarity.
2. “Absolute” properties are distinguished from “conventional” properties; i.e., a
generic standard partial molal property of an aqueous ion j ( ⌶ 0j ) is related to the
absolute property by
Ξ j0 = Ξ 0j abs − Z j Ξ 0 +abs , (8.104)

H
where Zj is the formal charge of the jth aqueous ion and ⌶ 0abs
H⫹ is the correspond-
ing absolute property of the hydrogen ion.
3. An “additivity principle” is adopted (Millero, 1972), according to which the con-
ventional standard partial molal property of a generic electrolyte K is related to
the absolute properties of its constituent ions through
Ξ K0 = ∑v j, K Ξ 0j abs ,
(8.105)
j
where vj,K is the stoichiometric number of moles of ion j in electrolyte K.

From equations 8.104 and 8.105, applying the electroneutrality condition, it follows
that the additivity principle is also valid for conventional properties—i.e.,
Ξ K0 = ∑v j, K Ξ j0 .
(8.106)
j
Because the conventional standard partial molal properties of a hydrogen ion is zero, it
follows that the conventional standard partial molal properties of a generic anion A are
identical to the experimental values of the corresponding acid electrolyte. Moreover, based
on equation 8.104, the standard partial molal properties of a generic cation C can be
calculated, once the experimental values for aqueous electrolytes Hv⫹A and Cv⫹Av⫺ are
known. Note here the close analogy between the additivity principle and the mean salt
method, previously discussed as a method of calculating ionic activity coefficients (sec-
tion 8.6).
8.11.2 Generalities of Approach

Each partial molal property of a jth aqueous ion in solution is considered to be composed
of two terms: the intrinsic property of the solute ( ⌶ 0i, j ) and an electroconstriction contribu-
tion ( ⌬⌶ 0e, j ):
Ξ 0j abs = Ξ i0, abs

j + ∆Ξ e0,abs
j . (8.107)
The electroconstriction contribution derives from the structural collapse of the solvent in
the immediate neighborhood of the ion ( ⌬⌶ 0abs c, j ) and by the ion-solvation process
( ⌬⌶ 0abs
s, j ) (electrostatic solute-solvent interactions):
∆Ξ e0,abs
j = ∆Ξ c0,abs
j + ∆Ξ s0, abs
j . (8.108)
Equations 8.107 and 8.108, referring to absolute properties, are equally valid for conven-
tional properties based on equations 8.105 and 8.106. Moreover, the additivity principle
is applied to all partial contributions: thus, for a generic electrolyte K, we have
∆Ξ e0, K = ∑v j, K ∆Ξ e0, j ,
(8.109)
j
∆Ξ c0, K = ∑v j, K ∆Ξ c0, j ,
(8.110)
j
∆Ξ s0, K = ∑v j, K ∆Ξ s0, j ,
(8.111)
j
and so on.
The HKF model groups the intrinsic properties of the solute and the effects of struc-
tural collapse of the solvent into a single term, which is defined as the contribution of
“nonsolvation”:
∆Ξ n0,abs
j = Ξ i0, abs
j + ∆Ξ c0,abs
j . (8.112)
The partial molal property of the solute is thus composed of one nonsolvation term
( ⌬⌶ 0abs
n, j ) plus one solvation ( ⌬⌶ s, j ) term; i.e., for a generic ion j,
0abs
Ξ 0j abs = ∆Ξ n0,abs
j + ∆Ξ s0, abs
j (8.113)
Ξ 0j = ∆Ξ n0, j + ∆Ξ s0, j . (8.114)
Similarly, for a generic electrolyte K, he have
Ξ K0 = ∆Ξ n0, K + ∆Ξ s0, K . (8.115)
8.11.3 Effective Electrostatic Radius, Born Coefficient, and

Solvation Energy
We have already mentioned the significance of the “effective diameter” of an ion in solu-
tion, stating that it represents the distance within which a given ion in solution affects the
structure of the solvent (section 8.5). We must now introduce the concept of effective
electrostatic radius. For this purpose, we recall that the Gibbs free energy change ⌬G 0abs
s, j
involved in the solvation of a generic ion j in solution is defined by the Born equation
1 
∆Gs0, abs = ω abs
j  − 1 , (8.116)
j
ε 
(Born, 1920), where ␧ is the dielectric constant of the solvent and ␻abs
j is the absolute Born
coefficient for the jth ion in solution:
N 0 e 2 Z 2j
ω abs
j = , (8.117)
2re , j
where N 0 is Avogadro’s number, e is the elementary charge, Zj is the formal charge of ion
j, and re, j is its electrostatic radius. Posing
1  Å ⋅ cal 
η= N 0 e 2 = 1.66027 × 10 5  , (8.118)
2  mole 
equation 8.117 can be reexpressed as
ηZ 2j
ω abs
j = . (8.119)
re , j
Equation 8.116 is also valid for conventional magnitudes:
1 
∆Gs0, j = ω j  − 1 . (8.120)
ε 
The conventional Born coefficient ␻j is related to the absolute coefficient ␻abs

j by
ω j = ω abs
j − Z j ω abs+ , (8.121)
H
where ␻ abs
H⫹ is the absolute Born coefficient of the hydrogen ion at the P and T of interest
( ␻ abs
H ⫹ ⫽ 0.5387 ⫻ 105 cal/mole at T ⫽ 25 °C and P ⫽ 1 bar; cf. Helgeson and Kirkham,
1974).
Stemming from the general relations developed in the preceding section, the solvation
energy of a generic electrolyte K can be related to that of its constituting ions j through
the Born coefficient ␻K (Tanger and Helgeson, 1988):
v j , K ηZ 2j
ωK = ∑v j, K ω j
abs
= ∑ re , j
= ∑v j ,K
ωj (8.122)
j j j
1 
∆Gs0, K = ω K  − 1 . (8.123)
ε 
The solvation energy described by the Born equation is essentially electrostatic in

nature. Born equations 8.116 and 8.120 are in fact similar to the Born-Landé equation
(1.67) used to define the electrostatic potential in a crystal (see section 1.12.1). In light of
this analogy, the effective electrostatic radius of an ion in solution re, j assumes the same
significance as the “equilibrium distance” in the Born-Landé equation. We may thus ex-
pect a close analogy between the crystal radius of an ion and the effective electrostatic
radius of the same ion in solution.
This analogy is expressed by the empirical equation
re , j = rx , j + Z j Γ ± , (8.124)
where rx, j is the crystal radius of generic ion j and
Γ ± = K Z + g. ( 8.125)
KZ in equation 8.125 represents a charge constant that is zero for anions and 0.94 Å for
cations, and g is a solvent parameter dependent on P and T (see equations 36A, B, and C
in Tanger and Helgeson, 1988).
Table 8.10 lists electrostatic radii of aqueous ions at T ⫽ 25 °C and P ⫽ 1 bar (Shock
and Helgeson, 1988).
8.11.4 Standard Partial Molal Volume, Compressibility, and

Thermal Expansion
The standard partial molal volume of a generic ion (or electrolyte) in solution can be
expressed, along the lines of the preceding section, through the summation of “nonsolva-
tion” and “solvation” contributions—i.e., omitting the subscripts ( j and/or K):
V 0
= ∆Vn 0 + ∆Vs 0 . ( 8.126 )
Partial molal volumes can be related to the corresponding Gibbs free energy terms
through the partial derivatives on P (see equations 2.28 and 2.33).
 ∂∆Gn0 
∆Vn 0 =   (8.127)
 ∂P  T
 ∂∆Gs 0 
∆Vs 0 =   . (8.128)
 ∂P  T
The solvation contribution has the form

Table 8.10 Electrostatic radii (Å) of various aqueous ions at T ⫽ 25 °C and P ⫽ 1 bar,
after Shock and Helgeson (1988).
Ion re Ion re Ion re Ion re

Li⫹ 1.62 La3⫹ 3.96 Re4⫹ 4.48 BrO⫺ 1.77
Na⫹ 1.91 Ce3⫹ 3.89 Ru4⫹ 4.44 BrO3⫺ 3.29
K⫹ 2.27 Pr3⫹ 3.88 S4⫹ 4.13 BrO4⫺ 4.51
Rb⫹ 2.41 Nd3⫹ 3.86 Se4⫹ 4.26 Br3⫺ 5.35
Cs⫹ 2.61 Sm3⫹ 3.82 Si4⫹ 4.18 IO⫺ 1.50
Ag⫹ 2.20 Eu3⫹ 3.80 Sn4⫹ 4.47 IO3⫺ 2.51
Tl⫹ 2.41 Gd3⫹ 3.79 Tb4⫹ 4.57 IO4⫺ 5.77
Cu⫹ 1.90 Tb3⫹ 3.75 Te4⫹ 4.46 I3⫺ 7.41
NH4⫹ 2.41 Dy3⫹ 3.74 Th4⫹ 4.78 H2VO4⫺ 2.55
VO2⫹ 1.34 Ho3⫹ 3.73 Ti4⫹ 4.44 HCrO4⫺ 3.29
Au⫹ 2.31 Er3⫹ 3.71 U4⫹ 4.73 MnO4⫺ 4.17
Fr⫹ 2.74 Tm3⫹ 3.69 V4⫹ 4.39 ReO4⫺ 4.59
Mg2⫹ 2.54 Yb3⫹ 3.68 W4⫹ 4.46 BeO22⫺ 2.45
Ca2⫹ 2.87 Lu3⫹ 3.67 Zr4⫹ 4.55 CO32⫺ 2.87
Sr2⫹ 3.00 Sc3⫹ 3.63 F⫺ 1.33 SiF62⫺ 3.92
Ba2⫹ 3.22 Am3⫹ 3.89 Cl⫺ 1.81 N2O22⫺ 3.26
Ra2⫹ 3.31 As3⫹ 3.40 Br⫺ 1.96 HPO42⫺ 2.93
Mn2⫹ 2.68 Au3⫹ 3.67 I⫺ 2.20 H2P2O72⫺ 4.30
Fe2⫹ 2.62 B3⫹ 3.05 OH⫺ 1.40 HAsO42⫺ 3.10
Ag2⫹ 2.77 Bi3⫹ 3.78 HS⫺ 1.84 SO32⫺ 2.96
Cu2⫹ 2.60 Cr3⫹ 3.45 AlO2⫺ 1.40 SO42⫺ 3.21
Zn2⫹ 2.62 Mn3⫹ 3.48 BO2⫺ 1.36 S22⫺ 6.8
Cd2⫹ 2.85 Np3⫹ 3.92 BF4⫺ 3.78 S2O32⫺ 3.51
Pb2⫹ 3.08 Pa3⫹ 3.95 HCO3⫺ 2.26 S2O42⫺ 3.69
Ni2⫹ 2.57 Pm3⫹ 3.88 CN⫺ 2.21 S2O52⫺ 3.78
Co2⫹ 2.60 Pu3⫹ 3.90 NO3⫺ 2.97 S2O62⫺ 3.95
Hg2⫹ 2.98 Rh3⫹ 3.50 NO2⫺ 2.57 S2O82⫺ 5.31
Sn2⫹ 2.81 Sb3⫹ 3.58 N3⫺ 2.37 S32⫺ 3.50
Sm2⫹ 2.88 Ti3⫹ 3.58 H2PO4⫺ 2.18 S3O62⫺ 4.06
Eu2⫹ 3.06 V3⫹ 3.56 H3P2O7⫺ 5.22 S42⫺ 3.77
Yb2⫹ 2.88 Am4⫹ 4.68 H2AsO4⫺ 2.49 S4O62⫺ 5.52
VO2⫹ 2.53 Ce4⫹ 4.70 H2AsO3⫺ 2.40 S52⫺ 4.08
UO22⫹ 2.67 Ge4⫹ 4.29 SbS4⫺ 1.40 S5O62⫺ 4.34
Hg22⫹ 3.50 Hf 4⫹ 4.54 HO2⫺ 1.65 SeO32⫺ 3.18
Be2⫹ 2.23 Ir4⫹ 4.44 HSO3⫺ 2.84 SeO42⫺ 3.42
Ge2⫹ 2.61 Mn4⫹ 4.36 HSO4⫺ 2.54 TeO32⫺ 3.18
Pd2⫹ 2.68 Mo4⫹ 4.46 HSO5⫺ 5.15 HVO42⫺ 3.20
Pt2⫹ 2.68 Nb4⫹ 4.50 HSe⫺ 2.07 CrO42⫺ 3.40
V2⫹ 2.76 Np4⫹ 4.71 HSeO3⫺ 2.76 Cr2O72⫺ 5.59
Al3⫹ 3.33 Os4⫹ 4.45 HSeO4⫺ 3.02 MnO42⫺ 3.45
Fe3⫹ 3.46 Pa4⫹ 4.74 HTeO3⫺ 2.83 MoO42⫺ 3.26
Ga3⫹ 3.44 Pb4⫹ 4.60 HF2⫺ 2.20 WO42⫺ 3.34
In3⫹ 3.36 Pd4⫹ 4.41 ClO⫺ 1.77 PO43⫺ 3.74
Tl3⫹ 3.77 Pr4⫹ 4.68 ClO2⫺ 2.29 HP2O73⫺ 4.93
Co3⫹ 3.45 Pt4⫹ 4.41 ClO3⫺ 3.30 AsO43⫺ 3.95
Y3⫹ 3.74 Pu4⫹ 4.69 ClO4⫺ 3.85 P2O74⫺ 5.59
ω  ∂ ln ε   1   ∂ω 
∆Vs0 = −   +  − 1   . (8.129)
ε  ∂P  T  ε   ∂P  T
The nonsolvation contribution can be expressed as a function of T through
 1 
∆Vn 0 = σ + ξ  , (8.130)
T − θ 
where ␪ is a solvent constant that, for water, assumes the value ␪ ⫽ 228 K (singular tem-
perature of supercooled water; Angell, 1982), and ␴ and ␰ are coefficients not dependent
on T but variable with P, based on
 1 
σ = a1 + a2   (8.131)
 ψ + P
 1 
ξ = a3 + a4  . (8.132)
 ψ + P
In equations 8.131 and 8.132, ␺ is another solvent constant that, for water, has the value
␺ ⫽ 2600 bar, and a1 , a2 , a3 , and a4 are characteristic parameters of ions and electrolytes
in solution. These parameters, constant over P and T, are listed in table 8.11.
Combining equations 8.126 to 8.132, the standard partial molal volume can be ex-
pressed as
 1    1   1 
0
= a1 + a2   + a + a4    
 ψ + P   3  ψ + P    T − θ 
V
(8.133)
 1   ∂ω 
− ωQ +  − 1   ,
ε   ∂P  T
where Q is the Born function:
1  ∂ ln ε 
Q=   . (8.134)
ε  ∂P  T
The compressibility of a generic ion or electrolyte in solution is given by the variation

with P of the nonsolvation and solvation terms:
 ∂∆Vn 0   ∂∆Vs 0 
−β 0 =   +  . (8.135)
 ∂P  T  ∂P  T
Adopting the same parameters previously used for calculating the standard partial molal
volume, compressibility can be expressed as
Table 8.11 Volume parameters of HKF model (from Shock
and Helgeson, 1988). a1 ⫽ cal/(mole ⫻ bar); a2 ⫽ cal/mole;
a3 ⫽ (cal ⫻ K)/(mole ⫻ bar); a4 ⫽ (cal ⫻ K)/mole
Ion a1 ⫻ 10 a2 ⫻ 10⫺2 a3 a4 ⫻ 10⫺4

H⫹ 0 0 0 0
Li⫹ ⫺0.0237 ⫺0.0690 11.5800 ⫺2.7761
Na⫹ 1.8390 ⫺2.2850 3.2560 ⫺2.726
K⫹ 3.5590 ⫺1.4730 5.4350 ⫺2.712
Rb⫹ 4.2913 ⫺0.9041 7.4070 ⫺2.7416
Cs⫹ 6.1475 ⫺0.1309 4.2094 ⫺2.7736
NH4⫹ 3.8763 2.3448 8.5605 ⫺2.8759
Ag⫹ 1.7285 ⫺3.5608 7.1496 ⫺2.6318
Tl⫹ 4.3063 2.7333 4.6757 ⫺2.8920
VO2⫹ ⫺2.5825 ⫺14.0871 11.2869 ⫺2.1966
Cu⫹ 0.7835 ⫺5.8682 8.0565 ⫺2.5364
Au⫹ 3.3448 0.3856 5.5985 ⫺2.7949
Mg2⫹ ⫺0.8217 ⫺8.5990 8.3900 ⫺2.390
Ca2⫹ ⫺0.1947 ⫺7.2520 5.2966 ⫺2.4792
Sr2⫹ 0.7071 ⫺10.1508 7.0027 ⫺2.3594
Ba2⫹ 2.7383 ⫺10.0565 ⫺0.0470 ⫺2.3633
Pb2⫹ ⫺0.0051 ⫺7.7939 8.8134 ⫺2.4568
Mn2⫹ ⫺0.1016 ⫺8.0295 8.9060 ⫺2.4471
Co2⫹ ⫺1.0748 ⫺12.9948 16.4112 ⫺2.2418
Ni2⫹ ⫺1.6942 ⫺11.9181 10.4344 ⫺2.2863
Cu2⫹ ⫺1.1021 ⫺10.4726 9.8662 ⫺2.3461
Zn2⫹ ⫺1.0677 ⫺10.3884 9.8331 ⫺2.3495
Cd2⫹ 0.0537 ⫺10.7080 16.5176 ⫺2.3363
Be2⫹ ⫺1.0684 ⫺10.3901 9.8338 ⫺2.3495
Ra2⫹ 0.6285 ⫺6.2469 8.2053 ⫺2.5208
Sn2⫹ 0.0786 ⫺7.5895 8.7330 ⫺2.4652
VO2⫹ ⫺1.7351 ⫺12.0180 10.4736 ⫺2.2822
Fe2⫹ ⫺0.7803 ⫺9.6867 9.5573 ⫺2.3786
Pd2⫹ ⫺0.6079 ⫺9.2658 9.3919 ⫺2.3960
Ag2⫹ ⫺0.4299 ⫺8.8310 9.2210 ⫺2.4139
Hg2⫹ ⫺0.5280 ⫺9.0707 9.3152 ⫺2.4040
Hg22⫹ 4.0126 2.0164 4.9575 ⫺2.8624
Sm2⫹ 1.4002 ⫺4.3625 7.4647 ⫺2.5987
Eu2⫹ 1.8333 ⫺3.3050 7.0490 ⫺2.6424
Yb2⫹ 0.7707 ⫺5.8995 8.0688 ⫺2.5351
Al3⫹ ⫺3.3404 ⫺17.1108 14.9917 ⫺2.0716
La3⫹ ⫺2.7880 ⫺14.3824 10.9602 ⫺2.1844
Ce3⫹ ⫺2.9292 ⫺14.9338 11.6196 ⫺2.1616
Pr3⫹ ⫺3.2406 ⫺14.1998 8.1257 ⫺2.1920
Nd3⫹ ⫺3.3707 ⫺14.5452 8.3211 ⫺2.1777
Sm3⫹ ⫺3.2065 ⫺15.6108 11.8857 ⫺2.1337
Eu3⫹ ⫺3.1037 ⫺15.3599 11.7871 ⫺2.1440
Gd3⫹ ⫺2.9771 ⫺15.0506 11.6656 ⫺2.1568
Tb3⫹ ⫺2.9355 ⫺14.9491 11.6257 ⫺2.1610
Dy3⫹ ⫺3.0003 ⫺15.1074 11.6879 ⫺2.1545
Ho3⫹ ⫺3.1198 ⫺15.3992 11.8026 ⫺2.1424
Er3⫹ ⫺3.3041 ⫺15.8492 11.9794 ⫺2.1238
continued
Ion a1 ⫻ 10 a2 ⫻ 10⫺2 a3 a4 ⫻ 10⫺4

Tm3⫹ ⫺3.2967 ⫺15.8312 11.9724 ⫺2.1245
Yb3⫹ ⫺3.4983 ⫺16.3233 12.1658 ⫺2.1042
Lu3⫹ ⫺3.5630 ⫺16.4812 12.2279 ⫺2.0977
Ga3⫹ ⫺3.4573 ⫺16.2232 12.1264 ⫺2.1083
In3⫹ ⫺3.1646 ⫺15.5086 11.8456 ⫺2.1379
Tl3⫹ ⫺2.8195 ⫺14.6658 11.5143 ⫺2.1727
Fe3⫹ ⫺3.1784 ⫺15.5422 11.8588 ⫺2.1365
Co3⫹ ⫺3.3247 ⫺15.8994 11.9992 ⫺2.1217
Au3⫹ ⫺3.0567 ⫺15.2450 11.7419 ⫺2.1488
Sc3⫹ ⫺3.1236 ⫺15.4084 11.8062 ⫺2.1420
Y3⫹ ⫺3.0140 ⫺15.1408 11.7010 ⫺2.1531
AlO2⫺ 3.7280 3.9800 ⫺1.5170 ⫺2.9435
HCO3⫺ 7.5621 1.1505 1.2346 ⫺2.8266
NO3⫺ 7.3161 6.7824 ⫺4.6838 ⫺3.0594
NO2⫺ 5.5864 5.8590 3.4472 ⫺3.0212
H2PO4⫺ 6.4875 8.0594 2.5823 ⫺3.1122
OH⫺ 1.2527 0.0738 1.8423 ⫺2.7821
HS⫺ 5.0119 4.9799 3.4765 ⫺2.9849
HSO3⫺ 6.7014 8.5816 2.3771 ⫺3.1338
HSO4⫺ 6.9788 9.2590 2.1108 ⫺3.1618
F⫺ 0.6870 1.3588 7.6033 ⫺2.8352
Cl⫺ 4.0320 4.8010 5.5630 ⫺2.847
ClO3⫺ 7.1665 9.7172 1.9307 ⫺3.1807
ClO4⫺ 8.1411 17.3059 ⫺12.2254 ⫺3.4944
Br⫺ 5.2690 6.5940 4.7450 ⫺3.143
BrO3⫺ 6.9617 9.2173 2.1272 ⫺3.1600
I⫺ 7.7623 8.2762 1.4609 ⫺3.1211
IO3⫺ 5.7148 6.1725 3.3240 ⫺3.0342
MnO4⫺ 7.8289 11.3346 1.2950 ⫺3.2476
ReO4⫺ 8.6513 13.3428 0.5057 ⫺3.3306
BO2⫺ ⫺2.2428 ⫺6.2065 ⫺6.3216 ⫺2.5224
BF4⫺ 7.9796 11.7027 1.1504 ⫺3.2628
CN⫺ 5.5125 5.6786 3.5181 ⫺3.0138
H2AsO4⫺ 6.9755 9.2509 2.1140 ⫺3.1614
H2AsO3⫺ 5.7934 6.3646 3.2485 ⫺3.0421
HO2⫺ 3.0671 ⫺0.2924 5.8649 ⫺2.7669
HSO5⫺ 8.9938 14.1790 0.1771 ⫺3.3652
HSe⫺ 4.8181 3.9831 4.1845 ⫺2.9437
HSeO3⫺ 6.5703 8.2615 2.5029 ⫺3.1205
HSeO4⫺ 6.3749 7.7844 2.6904 ⫺3.1008
HF2⫺ 5.2263 4.9797 3.7928 ⫺2.9849
ClO⫺ 3.6325 1.0881 5.3224 ⫺2.8240
ClO2⫺ 5.5036 5.6569 3.5266 ⫺3.0129
BrO⫺ 3.6325 1.0881 5.3224 ⫺2.8240
BrO4⫺ 8.5987 13.2144 0.5562 ⫺3.3253
Br3⫺ 9.0992 14.4364 0.0759 ⫺3.3758
IO⫺ 2.1432 ⫺2.5482 6.7516 ⫺2.6737
IO4⫺ 9.2968 14.9189 ⫺0.1138 ⫺3.3957
I3⫺ 9.8500 16.2697 ⫺0.6447 ⫺3.4516
Ion a1 ⫻ 10 a2 ⫻ 10⫺2 a3 a4 ⫻ 10⫺4

H2VO4⫺ 6.1355 7.1997 2.9202 ⫺3.0766
HCrO4⫺ 8.1112 12.0239 1.0241 ⫺3.2761
H3P2O7⫺ 9.0334 14.2757 0.1391 ⫺3.3692
CO32⫺ 2.8524 ⫺3.9844 6.4142 ⫺2.6143
HPO42⫺ 3.6315 1.0857 5.3233 ⫺2.8239
SO42⫺ 8.3014 ⫺1.9846 ⫺6.2122 ⫺2.6970
S2O32⫺ 6.6685 12.4951 ⫺7.7281 ⫺3.2955
S2O82⫺ 13.3622 24.8454 ⫺4.0153 ⫺3.8061
CrO42⫺ 5.5808 5.8454 3.4525 ⫺3.0206
MoO42⫺ 6.9775 9.2559 2.1121 ⫺3.1616
WO42⫺ 7.2074 8.7934 4.4896 ⫺3.1425
BeO22⫺ ⫺0.3210 ⫺8.5653 9.1166 ⫺2.4249
SiF62⫺ 8.5311 13.0492 0.6211 ⫺3.3185
H2P2O72⫺ 9.8215 16.2000 ⫺0.6173 ⫺3.4487
HAsO42⫺ 4.6321 3.5288 4.3630 ⫺2.9249
SO32⫺ 3.6537 0.3191 7.3853 ⫺2.7922
S22⫺ 5.5797 5.8426 3.4536 ⫺3.0205
S2O42⫺ 7.5815 10.7306 1.5324 ⫺3.2226
S2O52⫺ 7.9775 11.6976 1.1524 ⫺3.2626
S2O62⫺ 8.6225 13.2724 0.5334 ⫺3.3277
S32⫺ 6.7661 8.7396 2.3150 ⫺3.1403
S3O62⫺ 9.0313 14.2706 0.1410 ⫺3.3689
S42⫺ 7.9381 11.6012 1.1902 ⫺3.2586
S4O62⫺ 12.7850 23.4362 ⫺3.4614 ⫺3.7479
S52⫺ 9.1107 14.4645 0.0649 ⫺3.3770
S5O62⫺ 9.9535 16.5224 ⫺0.7440 ⫺3.4620
SeO32⫺ 5.0791 4.6204 3.9340 ⫺2.9700
SeO42⫺ 5.6180 5.9362 3.4168 ⫺1.2986
HVO42⫺ 5.2115 4.9437 3.8069 ⫺2.9834
Cr2O72⫺ 12.5200 22.7889 ⫺3.2070 ⫺3.7211
MnO42⫺ 6.5291 8.1609 2.5424 ⫺3.1164
PO43⫺ ⫺0.5259 ⫺9.0654 9.3131 ⫺2.4042
2
  1   1   ∂Q 
β = a 2 + a 4 
0
   + ω 
  
T − θ  ψ + P   ∂P  T
(8.136)
 ∂ω   1   ∂ 2ω 
+ 2Q   −  − 1  2  .
 ∂P  T  ε   ∂P  T
The thermal expansion of a generic ion or electrolyte in solution is given by the varia-
tion with T of the nonsolvation and solvation terms:
 ∂∆Vn 0   ∂∆Vs 0 
α0 =   +  . (8.137)
 ∂T  P  ∂T  P
Adopting the same parameters previously used for calculating the standard partial molal
volume (eq. 8.133) and compressibility (eq. 8.136), thermal expansion can be expressed as
2
  1   1   ∂Q   ∂ω 
α 0 = − a 3 + a 4     − ω   − Q 
  ψ + P    T − θ   ∂T  P  ∂T  P
 1   ∂ (∂ω ∂P ) T
(8.138)
 ∂ω  
−Y   +  − 1   ,
 ∂P  T ε  ∂T  P

where Y is the Born function:
1  ∂ ln ε 
Y =   . (8.139)
ε  ∂T  P
8.11.5 Heat Capacity and Standard Partial Molal Entropy

Heat capacity of solutes, as partial molal volume, compressibility, and thermal expansion,
is composed of the contributions of nonsolvation ( ⌬C 0P,n ) and solvation ( ⌬C 0P,s ):
CP0 = ∆CP0, n + ∆CP0, s.
The variation of ⌬C 0P,n with T at reference pressure has an asymptotic form that can be
described by two coefficients, constant over T and P and characteristic for each ion and
electrolyte in solution (see table 8.12):
2
 1 
∆C P0, n = c1 + c2   . (8.141)
T − θ
The variation of ⌬C 0P,n with P and T is described by
2
 1 
∆C P0, n = c1 + c2  
T − θ
(8.142)
 1    ψ + P 
3
− 2T   a 3 (P − Pr ) + a 4 ln  .
T − θ   ψ + Pr  
The solvation contribution is given by
 ∂ω   1   ∂ 2ω 
∆C P0, s = ωTX + 2TY   − T  − 1  2  , (8.143)
 ∂T  P ε   ∂T  P
Table 8.12 Standard state heat capacity (Tr ⫽ 25 °C, Pr ⫽ 1 bar),
coefficients of heat capacity, and Born coefficient in HKF model
0
(from Shock and Helgeson, 1988). CP ⫽ cal/(mole ⫻ K); c1 ⫽ cal/
(mole ⫻ K); c2 ⫽ (cal ⫻ K)/mole; ␻ ⫽ cal/mole.
c2 ⫻ 10⫺4 ␻ ⫻ 10⫺5
0
Ion CP c1
H⫹ 0 0 0 0
Li⫹ 14.2 19.2 ⫺0.24 0.4862
Na⫹ 9.06 18.18 ⫺2.981 0.3306
K⫹ 1.98 7.40 ⫺1.791 0.1927
Rb⫹ ⫺3.0 5.7923 ⫺3.6457 0.1502
Cs⫹ ⫺6.29 6.27 ⫺5.736 0.0974
NH4⫹ 15.74 17.45 ⫺0.021 0.1502
Ag⫹ 7.9 12.7862 ⫺1.4254 0.2160
Tl⫹ ⫺4.2 5.0890 ⫺3.8901 0.1502
VO2⫹ 31.1 30.8449 3.3005 0.7003
Cu⫹ 13.7 17.2831 ⫺0.2439 0.3351
Au⫹ 0.6 8.1768 ⫺2.9124 0.1800
Mg2⫹ ⫺5.34 20.80 ⫺5.892 1.5372
Ca2⫹ ⫺7.53 9.00 ⫺2.522 1.2366
Sr2⫹ ⫺10.05 10.7452 ⫺5.0818 1.1363
Ba2⫹ ⫺12.30 3.80 ⫺3.450 0.9850
Pb2⫹ ⫺12.70 8.6624 ⫺5.6216 1.0788
Mn2⫹ ⫺4.1 16.6674 ⫺3.8698 1.4006
Co2⫹ ⫺7.8 15.2014 ⫺4.6235 1.4769
Ni2⫹ ⫺11.7 13.1905 ⫺5.4179 1.5067
Cu2⫹ ⫺5.7 20.3 ⫺4.39 1.4769
Zn2⫹ ⫺6.3 15.9009 ⫺4.3179 1.4574
Cd2⫹ ⫺3.5 15.6573 ⫺3.7476 1.2528
Be2⫹ ⫺1.3 22.9152 ⫺3.2994 1.9007
Ra2⫹ ⫺14.4 6.2858 ⫺5.9679 0.9290
Sn2⫹ ⫺11.2 11.4502 ⫺5.3160 1.2860
VO2⫹ ⫺12.0 13.3910 ⫺5.4790 1.5475
Fe2⫹ ⫺7.9 14.9632 ⫺4.6438 1.4574
Pd2⫹ ⫺6.3 15.3780 ⫺4.3179 1.4006
Ag2⫹ ⫺5.3 15.2224 ⫺4.1142 1.3201
Hg2⫹ 2.2 18.0613 ⫺2.5865 1.151
Hg22⫹ 17.1 23.7433 0.4487 0.8201
Sm2⫹ 3.7 19.6533 ⫺2.2809 1.2285
Eu2⫹ 6.0 19.7516 ⫺1.8124 1.0929
Yb2⫹ 0.7 17.8951 ⫺2.8920 1.2285
Al3⫹ ⫺32.5 10.7 ⫺8.06 2.8711
La3⫹ ⫺37.2 4.2394 ⫺10.6122 2.1572
Ce3⫹ ⫺38.6 4.0445 ⫺10.8974 2.2251
Pr3⫹ ⫺47.7 ⫺1.1975 ⫺12.7511 2.2350
Nd3⫹ ⫺43.2 1.6236 ⫺11.8344 2.2550
Sm3⫹ ⫺43.3 1.9385 ⫺11.8548 2.2955
Eu3⫹ ⫺36.6 6.0548 ⫺10.4900 2.3161
Gd3⫹ ⫺35.9 6.5606 ⫺10.3474 2.3265
Tb3⫹ ⫺40.5 4.2522 ⫺11.2844 2.3685
Dy3⫹ ⫺31.7 9.5076 ⫺9.4919 2.3792
Ho3⫹ ⫺33.3 8.6686 ⫺9.8178 2.3899
continued
c2 ⫻ 10⫺4 ␻ ⫻ 10⫺5
0
Ion CP c1
Er3⫹ ⫺34.3 8.2815 ⫺10.0215 2.4115
Tm3⫹ ⫺34.3 8.4826 ⫺10.0215 2.4333
Yb3⫹ ⫺36.4 7.3533 ⫺10.4493 2.4443
Lu3⫹ ⫺32.0 9.5650 ⫺9.7160 2.4554
Ga3⫹ ⫺30.8 13.2451 ⫺9.3086 2.7276
In3⫹ ⫺34.9 8.7476 ⫺10.1437 2.5003
Tl3⫹ ⫺39.3 4.7607 ⫺11.0400 2.3474
Fe3⫹ ⫺34.1 11.0798 ⫺9.9808 2.7025
Co3⫹ ⫺32.3 12.2500 ⫺9.6141 2.7150
Au3⫹ ⫺36.2 7.5724 ⫺10.4085 2.4554
Sc3⫹ ⫺35.4 8.4546 ⫺10.2456 2.5003
Y3⫹ ⫺35.7 7.1634 ⫺10.3067 2.3792
AlO2⫺ ⫺11.9 19.1 ⫺6.2 1.7595
HCO3⫺ ⫺8.46 12.9395 ⫺4.7579 1.2733
NO3⫺ ⫺16.4 7.70 ⫺6.725 1.0977
NO2⫺ ⫺23.3 3.4260 ⫺7.7808 1.1847
H2PO4⫺ ⫺7.0 14.0435 ⫺4.4605 1.3003
OH⫺ ⫺32.79 4.15 ⫺10.346 1.7246
HS⫺ ⫺22.17 3.42 ⫺6.27 1.4410
HSO3⫺ ⫺1.4 15.6949 ⫺3.3198 1.1233
HSO4⫺ 5.3 20.0961 ⫺1.9550 1.1748
F⫺ ⫺27.23 4.46 ⫺7.488 1.7870
Cl⫺ ⫺29.44 ⫺4.40 ⫺5.714 1.4560
ClO3⫺ ⫺12.3 8.5561 ⫺5.5401 1.0418
ClO4⫺ ⫺5.5 22.3 ⫺8.9 0.9699
Br⫺ ⫺30.42 ⫺3.80 ⫺6.811 1.3858
BrO3⫺ ⫺20.6 3.7059 ⫺7.2308 1.0433
I⫺ ⫺28.25 ⫺6.27 ⫺4.944 1.2934
IO3⫺ ⫺16.2 7.7293 ⫺6.3345 1.2002
MnO4⫺ ⫺1.8 13.7427 ⫺3.4013 0.9368
ReO4⫺ ⫺0.2 14.3448 ⫺3.0753 0.9004
BO2⫺ ⫺41.0 ⫺1.6521 ⫺11.3863 1.7595
BF4⫺ ⫺5.6 11.8941 ⫺4.1753 0.9779
CN⫺ ⫺32.7 ⫺1.1135 ⫺9.6956 1.2900
H2AsO4⫺ ⫺0.7 16.8622 ⫺3.1772 1.2055
H2AsO3⫺ ⫺2.9 15.8032 ⫺3.6253 1.2305
HO2⫺ ⫺30.2 2.7007 ⫺9.1863 1.5449
HSO5⫺ 29.2 31.2126 2.9134 0.8611
HSe⫺ ⫺12.6 11.1345 ⫺5.6012 1.3408
HSeO3⫺ 4.9 19.5432 ⫺2.0365 1.1402
HSeO4⫺ ⫺41.9 ⫺8.3616 ⫺11.5696 1.0885
HF2⫺ ⫺33.2 ⫺1.3751 ⫺9.7974 1.2934
ClO⫺ ⫺49.2 ⫺9.0630 ⫺13.0566 1.4767
ClO2⫺ ⫺30.5 ⫺0.0659 ⫺9.2474 1.2637
BrO⫺ ⫺49.2 ⫺9.0630 ⫺13.0566 1.4767
BrO4⫺ 0.6 14.8727 ⫺2.9124 0.9068
Br3⫺ 5.6 17.2705 ⫺1.8939 0.8490
IO⫺ ⫺64.1 ⫺16.2398 ⫺16.0918 1.6455
IO4⫺ 7.6 18.2345 ⫺1.4865 0.8264
I3⫺ 13.1 20.8712 ⫺0.3661 0.7628
c2 ⫻ 10⫺4 ␻ ⫻ 10⫺5
0
Ion CP c1
H2VO4⫺ 0.6 17.4795 ⫺2.9124 1.1898
HCrO4⫺ 20.4 26.9872 1.1209 0.9622
H3P2O7⫺ 29.6 31.4072 2.9949 0.8568
CO32⫺ ⫺69.5 ⫺3.3206 ⫺17.1917 3.3914
HPO42⫺ ⫺58.3 2.7357 ⫺14.9103 3.3363
SO42⫺ ⫺64.38 1.64 ⫺17.998 3.1463
S2O32⫺ ⫺57.3 ⫺0.0577 ⫺14.7066 2.9694
S2O82⫺ ⫺25.0 12.9632 ⫺8.1271 2.3281
CrO42⫺ ⫺59.9 ⫺1.0175 ⫺15.2362 3.0307
MoO42⫺ ⫺47.5 7.0224 ⫺12.7103 3.1145
WO42⫺ ⫺44.5 8.3311 ⫺12.0992 3.0657
BeO22⫺ ⫺100.0 ⫺18.0684 ⫺23.5879 3.7880
SiF62⫺ ⫺47.1 4.0970 ⫺12.6289 2.7716
H2P2O72⫺ ⫺20.3 18.4241 ⫺7.1697 2.6218
HAsO42⫺ ⫺51.8 5.4710 ⫺13.5863 3.2197
SO32⫺ ⫺76.1 ⫺7.8368 ⫺18.5362 3.3210
S22⫺ ⫺65.1 ⫺3.3496 ⫺16.2955 3.1083
S2O42⫺ ⫺52.9 1.6707 ⫺13.8103 2.8772
S2O52⫺ ⫺50.5 2.6824 ⫺13.3215 2.8343
S2O62⫺ ⫺46.5 4.3301 ⫺12.5066 2.7587
S32⫺ ⫺57.9 ⫺0.3595 ⫺14.8288 2.9749
S3O62⫺ ⫺44.1 5.3169 ⫺12.0178 2.7131
S42⫺ ⫺50.7 2.6081 ⫺13.3622 2.8390
S4O62⫺ ⫺21.3 14.6933 ⫺7.3734 2.2805
S52⫺ ⫺43.6 5.5361 ⫺11.9159 2.7051
S5O62⫺ ⫺38.5 7.6266 ⫺10.8770 2.6076
SeO32⫺ ⫺68.1 ⫺4.5783 ⫺16.9066 3.1658
SeO42⫺ ⫺60.2 ⫺1.2986 ⫺15.2973 3.0192
HVO42⫺ ⫺48.3 6.9055 ⫺12.8733 3.1527
Cr2O72⫺ ⫺20.4 15.0820 ⫺7.1901 2.2654
MnO42⫺ ⫺59.3 ⫺0.9267 ⫺15.1140 3.0024
PO43⫺ ⫺114.9 ⫺9.4750 ⫺26.4397 5.6114
where X is the Born function:
1  ∂ 2 ln ε   ∂ ln ε  
2
X =   −    (8.144)
ε  ∂ T 2  P  ∂ T  P 
 
(Helgeson and Kirkham, 1974). Summing equations 8.142 and 8.143, we obtain
c2
 
2T
C P0 = c1 + − ×
(T − θ )  (T − θ ) 3 
2
 
(8.145)
  ψ + P   ∂ω  1   ∂ 2ω 
( )
 a3 P − Pr + a 4 ln    + ωTX + 2TY 
 ψ + Pr  
 − T  − 1 
∂T  P ε
 .
  ∂T 2  P

Solvation entropy contribution ⌬S 0s is given by the integral in dT/T of ⌬C 0P,s (equation

8.143):
 1   ∂ω 
∆S s0 = ωY −  − 1   , (8.146)
ε  ∂T  P
and nonsolvation contribution ⌬S 0n is given by the integral in dT/T of ⌬C 0p,n (equation

8.142):
T 
c2  1   1  1  Tr (T − θ )   
∆S n0 = c1 ln −   −  + ln  
Tr θ  T − θ   Tr − θ  θ  T (Tr − θ )  
 
(8.147)
 1    ψ + P 
2
+  a (P − Pr ) + a 4 ln 
 T − θ   3 .
 ψ + Pr  
Combining the solvation and nonsolvation contributions, the partial molal entropy
change between reference state conditions Pr and Tr and the P and T of interest is given by
r r
(
S P0, T − S P0 , T = ∆S n0 − ∆S n0, P , T
r r
) + ( ∆S s
0
− ∆S s0, P
r , Tr
)
 T  c2  1   1  1  Tr (T − θ )  
= c1 ln  −   +  + ln  
 Tr  θ  T − θ   Tr − θ  θ  T (Tr − θ )  
(8.148)
 1    ψ + P 
2
+  a 3 (P − Pr ) + a 4 ln  
T − θ   ψ + Pr  
 1   ∂ω 
+ ωY −  − 1   − ω Pr , TrY Pr , Tr .
ε   ∂T  P
8.11.6 Enthalpy and Gibbs Free Energy of Formation of

Aqueous Solutes Under High P and T Conditions
We now have all the parameters and analytical forms necessary to perform enthalpy and
Gibbs free energy calculations at high P and T conditions:
  ∂V 0  
T P
H P0, T = H P0 , T +
r r ∫ C P0 dT +
r ∫ V

0
− T   dP
 ∂T  P  (8.149)
Tr Pr T
and
GP0, T = GP0 , T − S P0 , T T − Tr
r r r r
( )
T T P
(8.150)
+ ∫ C P0 dT − T ∫ C P0 d ln T +
r r ∫V 0
dP.
Tr Tr Pr
Note that thermodynamic tabulations do not normally report the standard partial
molar properties of solutes H 0Tr , Pr and G 0Tr , Pr, but rather the enthalpy of formation
H 0f, T r , Pr and the Gibbs free energy of formation from the elements at stable state under Tr
and Pr reference conditions G 0f, T r , Pr (note, however, that H 0T r , Pr and H 0f,T r ,Pr coincide,
because the enthalpy of elements at stable state and Tr ⫽ 25 °C and Pr ⫽ 1 bar is conven-
tionally set at zero). Because it is also true that
H f0, P , T − H f0, P , T = H P0, T − H P0 , T ( 8.151)

r r r r
and
G f0, P , T − G f0, P , T = GP0, T − GP0 , T , ( 8.152 )

r r r r
substituting the various coefficients (a1 , a2 , a3 , a4 , c1 , c2 ) and the Born functions ␻ , Y,

and X of the HKF model in equations 8.149 and 8.150 and taking into account equations
8.151 and 8.152, we obtain
 1   1  
H f0, P , T = H f0, P , T + c1 T − Tr − c2  ( )  −   + a 1 (P − Pr )
 T − θ   Tr − θ  
r r
 ψ + P   2T − θ  
 a 3 (P − Pr ) + a 4 ln  ψ + P  
+ a 2 ln  +
 ψ + Pr   (T − θ ) 2   ψ + Pr  
  (8.153)
1   1   ∂ω   1 
+ ω − 1 + ωTY − T  − 1   − ω Pr , Tr  − 1
ε  ε   ∂T  P  ε Pr , Tr 
− ω Pr , Tr TrY Pr , Tr
and
 T 
(
G f0, P , T = G f0, P , T − S P0 , T T − Tr − c1 T ln )  − T + Tr 
 Tr 
 
r r r r
ψ +P   1   1    θ − T 
(
+ a 1 P − Pr + a 2 ln  )  − c2    −   
 ψ + Pr    T − θ   Tr − θ    θ 
 Tr (T − θ )    1  
(8.154)
 ψ + P 
 a 3 (P − Pr ) + a 4 ln 
T
− ln   +  
θ 2  T (Tr − θ )    T − θ    ψ + Pr  

1   1 
+ ω − 1 − ω Pr , Tr  − 1 + ω Pr , TrY Pr , Tr (T − Tr ).
ε   ε Pr , Tr 
Table 8.13 lists G 0f, Pr ,Tr , H 0f,Pr ,Tr , S 0Pr ,Tr, and V 0Pr ,Tr values for various aqueous ions,
after Shock and Helgeson (1988). These values, together with the coefficients in tables
Table 8.13 Standard partial molal Gibbs free energy of formation
from the elements (G 0f, P r , Tr ; cal/mole), standard partial molal
enthalpy of formation from the elements ( H 0f, P r , Tr ; cal/mole),
standard partial molal entropy ( S 0P r , T r ; cal/(mole ⫻ K)), and
standard partial molal volume ( V 0P r , T r ; cm3/mole) for aqueous ions
at Tr ⫽ 25 °C and Pr ⫽ 1 bar (Shock and Helgeson, 1988).
Ion G f,0 P r , Tr H f,0 P r , Tr S P0 r , T r V P0 r , T r
H⫹ 0 0 0 0
Li⫹ ⫺69,933 ⫺66,552 2.70 ⫺0.87
Na⫹ ⫺62,591 ⫺57,433 13.96 ⫺1.11
K⫹ ⫺67,510 ⫺60,270 24.15 9.06
Rb⫹ ⫺67,800 ⫺60,020 28.8 14.26
Cs⫹ ⫺69,710 ⫺61,670 31.75 21.42
NH4⫹ ⫺18,990 ⫺31,850 26.57 18.13
Ag⫹ 18,427 25,275 17.54 ⫺0.8
T1⫹ ⫺7,740 1,280 30.0 18.2
VO2⫹ ⫺140,300 ⫺155,300 ⫺10.1 ⫺33.5
Cu⫹ 11,950 17,132 9.7 ⫺8.0
Au⫹ 39,000 47,580 24.5 11.1
Mg2⫹ ⫺108,505 ⫺111,367 ⫺33.00 ⫺21.55
Ca2⫹ ⫺132,120 ⫺129,800 ⫺13.5 ⫺18.06
Sr2⫹ ⫺134,760 ⫺131,670 ⫺7.53 ⫺17.41
Ba2⫹ ⫺134,030 ⫺128,500 2.3 ⫺12.60
Pb2⫹ ⫺5,710 220 4.2 ⫺15.6
Mn2⫹ ⫺54,500 ⫺52,724 ⫺17.6 ⫺17.1
Co2⫹ ⫺13,000 ⫺13,900 ⫺27.0 ⫺24.4
Ni2⫹ ⫺10,900 ⫺12,900 ⫺30.8 ⫺29.0
Cu2⫹ 15,675 15,700 ⫺23.2 ⫺24.6
Zn2⫹ ⫺35,200 ⫺36,660 ⫺26.2 ⫺24.3
Cd2⫹ ⫺18,560 ⫺18,140 ⫺17.4 ⫺15.6
Be2⫹ ⫺83,500 ⫺91,500 ⫺55.7 ⫺25.4
Ra2⫹ ⫺134,200 ⫺126,100 13 ⫺10.6
Sn2⫹ ⫺6,630 ⫺2,100 ⫺3.8 ⫺15.5
VO2⫹ ⫺106,700 ⫺116,300 ⫺32 ⫺29.4
Fe2⫹ ⫺21,870 ⫺22,050 ⫺25.3 ⫺22.2
Pd2⫹ 42,200 42,080 ⫺22.6 ⫺20.8
Ag2⫹ 64,300 64,200 ⫺21.0 ⫺19.3
Hg2⫹ 39,360 40,670 ⫺8.68 ⫺19.6
Hg22⫹ 36,710 39,870 ⫺15.66 14.4
Sm2⫹ ⫺123,000 ⫺120,500 ⫺6.2 ⫺5.7
Eu2⫹ ⫺129,100 ⫺126,100 ⫺2.4 ⫺2.2
Yb2⫹ ⫺128,500 ⫺126,800 ⫺11.2 ⫺10.3
Al3⫹ ⫺115,377 ⫺126,012 ⫺75.6 ⫺44.4
La3⫹ ⫺164,000 ⫺169,600 ⫺52.0 ⫺38.6
Ce3⫹ ⫺161,600 ⫺167,400 ⫺49 ⫺39.8
Pr 3⫹ ⫺162,600 ⫺168,800 ⫺50 ⫺42.1
Nd3⫹ ⫺160,600 ⫺166,500 ⫺49.5 ⫺43.1
Sm3⫹ ⫺159,100 ⫺165,200 ⫺50.7 ⫺42.0
Eu3⫹ ⫺137,300 ⫺144,700 ⫺53.0 ⫺41.3
Gd3⫹ ⫺158,600 ⫺164,200 ⫺49.2 ⫺40.4

Tb3⫹ ⫺159,500 ⫺166,900 ⫺54 ⫺40.2
Dy3⫹ ⫺158,700 ⫺166,500 ⫺55.2 ⫺40.7
Ho3⫹ ⫺161,400 ⫺169,000 ⫺54.3 ⫺41.6
Er 3⫹ ⫺159,900 ⫺168,500 ⫺58.3 ⫺43.0
Tm3⫹ ⫺159,900 ⫺168,500 ⫺58.1 ⫺43.0
Yb3⫹ ⫺153,000 ⫺160,300 ⫺56.9 ⫺44.5
Lu3⫹ ⫺159,400 ⫺167,900 ⫺63.1 ⫺32.0
Ga3⫹ ⫺38,000 ⫺50,600 ⫺79.0 ⫺44.9
In3⫹ ⫺23,400 ⫺25,000 ⫺63.0 ⫺42.2
Tl3⫹ 51,300 47,000 ⫺46.0 ⫺39.3
Fe3⫹ ⫺4,120 ⫺11,850 ⫺66.3 ⫺42.8
Co3⫹ 32,000 22,000 ⫺73 ⫺43.9
Au3⫹ 103,600 96,930 ⫺57.9 ⫺41.3
Sc3⫹ ⫺140,200 ⫺146,800 ⫺61 ⫺41.9
Y 3⫹ ⫺163,800 ⫺170,900 ⫺60 ⫺40.8
AlO2⫺ ⫺198,465 ⫺222,261 ⫺8.4 10.0
HCO3⫺ ⫺140,282 ⫺164,898 23.53 24.6
NO3⫺ ⫺26,507 ⫺49,429 35.12 29.0
NO2⫺ ⫺7,700 ⫺25,000 29.4 25.0
H2PO4⫺ ⫺270,140 ⫺309,820 21.6 31.3
OH⫺ ⫺37,595 ⫺54,977 ⫺2.56 ⫺4.18
HS⫺ 2,860 ⫺3,850 16.3 20.65
HSO3⫺ ⫺126,130 ⫺149,670 33.4 33.3
HSO4⫺ ⫺180,630 ⫺212,500 30.0 35.2
F⫺ ⫺67,340 ⫺80,150 ⫺3.15 ⫺1.32
Cl⫺ ⫺31,379 ⫺39,933 13.56 17.79
ClO3⫺ ⫺1,900 ⫺24,850 38.8 36.9
ClO4⫺ ⫺2,040 ⫺30,910 43.5 44.2
Br⫺ ⫺24,870 ⫺29,040 19.40 24.85
BrO3⫺ 4,450 ⫺16,030 38.65 35.4
I⫺ ⫺12,410 ⫺13,600 25.50 36.31
IO3⫺ ⫺30,600 ⫺52.900 28.3 25.9
MnO4⫺ ⫺106,900 ⫺129,400 45.7 42
ReO4⫺ ⫺166,000 ⫺188,200 48.1 48.1
BO2⫺ ⫺162,240 ⫺184,600 ⫺8.9 ⫺14.5
BF4⫺ ⫺355,400 ⫺376,400 43.0 43.0
CN⫺ 41,200 36,000 22.5 24.2
H2AsO4⫺ ⫺180,010 ⫺217,390 28 35.1
H2AsO3⫺ ⫺140,330 ⫺170,840 26.4 26.4
HO2⫺ ⫺16,100 ⫺38,320 5.7 5.7
HSO5⫺ ⫺152,370 ⫺185,380 50.7 50.7
HSe⫺ 10,500 3,800 19 ⫺19.0
HSeO3⫺ ⫺98,340 ⫺122,980 32.3 32.3
HSeO4⫺ ⫺108,100 ⫺139,000 35.7 31.0
HF2⫺ ⫺138,160 ⫺155,340 22.1 22.1
ClO⫺ ⫺8,800 ⫺25,600 10 10.0
ClO2⫺ 4,100 ⫺15,900 24.2 24.2
BrO⫺ ⫺8,000 ⫺22,500 10 10.0
continued
BrO4⫺ 28,200 3,100 47.7 47.7

Br3⫺ ⫺25,590 ⫺31,370 51.5 51.5
IO⫺ ⫺9,200 ⫺25,700 ⫺1.3 ⫺1.3
IO4⫺ ⫺14,000 ⫺36,200 53 53.0
I3⫺ ⫺12,300 ⫺12,300 57.2 57.2
H2VO4⫺ ⫺244,000 ⫺280,600 29 29.0
HCrO4⫺ ⫺182,800 ⫺209,900 44.0 44.0
H3P2O7⫺ ⫺483,600 ⫺544,100 51 51.0
CO32⫺ ⫺126,191 ⫺161,385 ⫺11.95 ⫺5.02
HPO42⫺ ⫺260,310 ⫺308,815 ⫺8.0 5.4
SO42⫺ ⫺177,930 ⫺217,400 4.50 13.88
S2O32⫺ ⫺124,900 ⫺155,000 16 28.5
S2O82⫺ ⫺266,500 ⫺321,400 58.4 79.0
CrO42⫺ ⫺173,940 ⫺210,600 12.0 20.4
MoO42⫺ ⫺199,900 ⫺238,500 6.5 30.4
WO42⫺ ⫺219,150 ⫺257,100 9.7 32.2
BeO22⫺ ⫺153,000 ⫺189,000 ⫺38 ⫺24.6
SiF62⫺ ⫺525,700 ⫺571,000 29.2 42.6
H2P2O72⫺ ⫺480,400 ⫺544,600 39 52.4
HAsO42⫺ ⫺170,790 ⫺216,620 ⫺0.4 13.0
SO32⫺ ⫺116,300 ⫺151,900 ⫺7 5.6
S22⫺ 19,000 7,200 6.8 ⫺65.1
S2O42⫺ ⫺143,500 ⫺180,100 22 35.4
S2O52⫺ ⫺189,000 ⫺232,000 25 38.4
S2O62⫺ ⫺231,000 ⫺280,400 30 43.3
S32⫺ 17,600 6,200 15.8 29.2
S3O62⫺ ⫺229,000 ⫺279,000 33 46.4
S42⫺ 16,500 5,500 24.7 38.1
S4O62⫺ ⫺248,700 ⫺292,600 61.5 74.9
S52⫺ 15,700 5,100 33.6 47.0
S5O62⫺ ⫺229,000 ⫺281,000 40 53.4
SeO32⫺ ⫺88,400 ⫺121,700 3. ⫺16.4
SeO42⫺ ⫺105,500 ⫺143,200 12.9 20.7
HVO42⫺ ⫺233,000 ⫺277,000 4. 17.4
Cr2O72⫺ ⫺311,000 ⫺356,200 62.6 73.0
MnO42⫺ ⫺119,700 ⫺156,000 14.0 27.4
PO43⫺ ⫺243,500 ⫺305,300 ⫺53. ⫺30.6
8.11 and 8.12, allow the calculation of the corresponding partial molar properties in high
P and T conditions.* Practical calculations are performed by the various releases of the
* Actually, the various equations listed in this section are insufficient to perform the complete
calculation since one would first calculate the density of H2O through eq. 8.12 or 8.14. Equation
8.14 in its turn involves the partial derivative of the Helmholtz free energy function 8.15. Moreover,
the evaluation of electrostatic properties of the solvent and of the Born functions ␻, Q, Y, X involve
additional equations and variables not given here for the sake of brevity (eqs. 36, 40 to 44, 49 to 52
and tables 1 to 3 in Johnson et al., 1991). In spite of this fact, the decision to outline here briefly
the HKF model rests on its paramount importance in geochemistry. Moreover, most of the listed
thermodynamic parameters have an intrinsic validity that transcends the model itself.
automated routine SUPCRT, distributed by the Laboratory of Theoretical Geochemistry

of the University of California, Berkeley (Johnson et al. 1991).
8.12 Redox Conditions, pH, and Thermodynamic

Stability of Aqueous Solutions
Redox equilibrium is defined as a process characterized by the flow of electrons

from the substance being oxidized (“reducing medium”) to the substance being
reduced (“oxidizing medium”). For instance, ionic iron in aqueous solutions is
present in two valence states, related by the redox equilibrium
reduction→
Fe 3 + + e − ⇔ Fe 2 + . (8.155)
←oxidation
In a “reduction process” the flow of electrons is from left to right (i.e., from
oxidizing to reducing media). A Gibbs free energy change ⌬G 0155 is associated
with equilibrium 8.155, so that
a Fe 2 +
∆G155
0
= − RT ln K 155 = − RT ln . (8.156)
a Fe 3+ a e −
Term ae⫺ in equation 8.156 is the “activity of the hypothetical electron in solu-
tion,” which, by convention, is always assumed to be unitary.
Let us now consider a redox equilibrium involving metallic iron Fe and fer-
rous iron Fe2⫹ (one-molal solution at Tr ⫽ 25 °C and Pr ⫽ 1 bar):
Fe ⇔ Fe 2 + + 2 e − .
metal aqueous (8.157)
The Gibbs free energy change involved in equation 8.157 corresponds to the par-
tial molal Gibbs free energy of formation of Fe2⫹ from the element at stable state
(cf. table 8.13):
∆G157
0
=G 0
. (8.158)
f , Tr , Pr , Fe 2 +
In the case of oxidation of hydrogen, starting from the diatomic gaseous molecule,
we have
1
H2 ⇔ H+ + e − .
2 (8.159)
gaseous aqueous
⌬G 0159 is the Gibbs free energy change associated with equation 8.159, corre-
sponding to the formation of the aqueous hydrogen ion H⫹ from the gaseous
molecule H2 . This amount of energy is, by convention (see, for instance, table
8.13), fixed at zero.
To the reactions of equations 8.155, 8.157, and 8.159, which imply electron
transfer from the oxidized to the reduced state, the concept of electrical potential
is associated through Faraday’s law:
∆G = − nFE , (8.160)
where n is the number of electrons involved in the redox equilibrium, F is Fara-

day’s constant [F ⫽ 96484.56 J/(volt ⫻ mole)], and E is the electrical potential
expressed in volts.
Because the Gibbs free energy change ⌬G159 0
involved in the redox equilibrium
of equation 8.159 at standard state is zero, it follows from equation 8.160 that the
potential associated with this equilibrium at standard state is also zero. This fact
allows us to establish a “scale of standard potentials” for the various redox cou-
ples, where zero corresponds to the oxidation of gaseous hydrogen (“normal hy-
drogen electrode”). The minus sign in equation 8.160 is due to the fact that, by
convention, in the definition of standard potentials for metal-aqueous ion cou-
ples, the oxidizing medium is placed to the left of the reaction sign and the reduc-
ing medium to the right (i.e., a reduction process; cf. eq. 8.155), whereas the
Gibbs free energy of formation of an aqueous cation from the stable metal implies
the opposite (i.e., oxidation; cf. eq. 8.157 or 8.159). It must also be emphasized
that absolute potentials for the various redox couples cannot be determined, and
that the “standard potentials” therefore represent the potential gap measured at
the terminals of a galvanic cell in which one of the electrodes is the “normal
hydrogen electrode.” For instance, the standard potential of zinc is assessed with
a cell of this type:
H2 , H + Zn, Z 2n + ,
(8.161)
gas aq metal aq
where H2 ,H⫹ means a one-molal solution of H⫹ ions at T ⫽ 25 °C and P ⫽ 1

bar, in equilibrium with an atmosphere whose H2 fugacity is unitary at standard
state—i.e. (cf. section 9.1),
f H 2 gas f H 2 , gas
a H 2 , gas = = = 1. (8.162)
f H0 gas 1
2
In equation 8.161, Zn,Zn2⫹ means metallic zinc Zn in equilibrium with a one-

molal solution of Zn2⫹ under the same T and P conditions, and | | represents a
membrane with an electron transport number of 1, in the absence of diffusive

potential (the “unitary electron transport number” means that the electric con-
ductivity is purely electronic, because ions Zn2⫹ and H⫹ cannot cross the mem-
brane).
Because, as we have already seen, the standard potential of hydrogen is zero,
the electromotive force of the galvanic cell (eq. 8.161) directly gives the value of
the standard potential for the Zn,Zn2⫹ redox couple. Table 8.14 lists the standard
potentials for various aqueous ions. The listed values are arranged in decreasing
order and are consistent with the standard partial molal Gibbs free energies of
table 8.13.
Let us now imagine a redox process involving two metal-ion couples:
Table 8.14 Standard potentials for various redox couples (aqueous cation–metal; element–aqueous
anion) arranged in order of decreasing E 0. Values, expressed in volts, are consistent with standard
partial molal Gibbs free energy values listed in table 8.13. E 0 value for sulfur is from Nylen and
Wigren (1971).
Equilibrium E 0 (V ) Equilibrium E 0 (V )
Li⫹ ⫹ e⫺ → Li ⫺3.034 Al3⫹ ⫹ 3e⫺ → Al ⫺1.668
Cs⫹ ⫹ e⫺ → Cs ⫺3.023 Mn2⫹ ⫹ 2e⫺ → Mn ⫺1.182
Rb⫹ ⫹ e⫺ → Rb ⫺2.940 Zn2⫹ ⫹ 2e⫺ → Zn ⫺0.763
K⫹ ⫹ e⫺ → K ⫺2.928 Ga2⫹ ⫹ 2e⫺ → Ga ⫺0.549
Sr 2⫹ ⫹ 2e⫺ → Sr ⫺2.922 S ⫹ 2e⫺ → S2⫺ ⫺0.480
Ra2⫹ ⫹ 2e⫺ → Ra ⫺2.910 Fe2⫹ ⫹ 2e⫺ → Fe ⫺0.474
Ba2⫹ ⫹ 2e⫺ → Ba ⫺2.906 Cd2⫹ ⫹ 2e⫺ → Cd ⫺0.402
Ca2⫹ ⫹ 2e⫺ → Ca ⫺2.865 In3⫹ ⫹ 3e⫺ → In ⫺0.338
Eu2⫹ ⫹ 2e⫺ → Eu ⫺2.799 Tl⫹ ⫹ e⫺ → Tl ⫺0.336
Yb2⫹ ⫹ 2e⫺ → Yb ⫺2.786 Co2⫹ ⫹ 2e⫺ → Co ⫺0.282
Na⫹ ⫹ e⫺ → Na ⫺2.714 Ni2⫹ ⫹ 2e⫺ → Ni ⫺0.236
Sm2⫹ ⫹ 2e⫺ → Sm ⫺2.667 Sn2⫹ ⫹ 2e⫺ → Sn ⫺0.144
La3⫹ ⫹ 3e⫺ → La ⫺2.371 Pb2⫹ ⫹ 2e⫺ → Pb ⫺0.124
Y3⫹ ⫹ 3e⫺ → Y ⫺2.368 Fe3⫹ ⫹ 3e⫺ → Fe ⫺0.059
Mg2⫹ ⫹ 2e⫺ → Mg ⫺2.353 2H⫹ ⫹ 2e⫺ → H2 0.000
Pr3⫹ ⫹ 3e⫺ → Pr ⫺2.350 Cu2⫹ ⫹ 2e⫺ → Cu 0.340
Ce3⫹ ⫹ 3e⫺ → Ce ⫺2.336 Co3⫹ ⫹ 3e⫺ → Co 0.462
Ho3⫹ ⫹ 3e⫺ → Ho ⫺2.333 Cu⫹ ⫹ e⫺ → Cu 0.518
Nd3⫹ ⫹ 3e⫺ → Nd ⫺2.321 I2 ⫹ 2e⫺ → 2I⫺ 0.538
Er3⫹ ⫹ 3e⫺ → Er ⫺2.311 Tl3⫹ ⫹ 3e⫺ → Tl 0.742
Tm3⫹ ⫹ 3e⫺ → Tm ⫺2.311 Ag⫹ ⫹ e⫺ → Ag 0.799
Tb3⫹ ⫹ 3e⫺ → Tb ⫺2.306 Hg2⫹ ⫹ 2e⫺ → Hg 0.853
Lu3⫹ ⫹ 3e⫺ → Lu ⫺2.304 Pd2⫹ ⫹ 2e⫺ → Pd 0.915
Dy3⫹ ⫹ 3e⫺ → Dy ⫺2.294 Br2 ⫹ 2e⫺ → 2Br⫺ 1.078
Gd3⫹ ⫹ 3e⫺ → Gd ⫺2.293 Cl2 ⫹ 2e⫺ → 2Cl⫺ 1.361
Yb3⫹ ⫹ 3e⫺ → Yb ⫺2.212 Ag2⫹ ⫹ 2e⫺ → Ag 1.394
Sc3⫹ ⫹ 3e⫺ → Sc ⫺2.027 Au3⫹ ⫹ 3e⫺ → Au 1.498
Eu3⫹ ⫹ 3e⫺ → Eu ⫺1.985 Au⫹ ⫹ e⫺ → Au 1.691
Be2⫹ ⫹ 2e⫺ → Be ⫺1.810 F2 ⫹ 2e⫺ → 2F⫺ 2.920
Fe + Cu 2 + ⇔ Fe 2 + + Cu
metal aqueous aqueous metal
(8.163)
or even
Zn + Cu 2 + ⇔ Zn 2 + + Cu . (8.164)
metal aqueous aqueous metal
We can visualize the reactions 8.163 and 8.164 as composed of partial equilib-
ria—i.e., for equation 8.163,
Fe ⇔ Fe 2 + + 2 e − (0.474 V ) (8.165)
metal aqueous
Cu 2 + + 2 e − ⇔ Cu (0.340 V ) . (8.166)
aqueous metal
Analogously, for the redox process of equation 8.164,
Zn ⇔ Zn 2 + + 2 e −
metal aqueous
(0.763 V ) (8.167)
Cu 2 + + 2 e − ⇔ Cu
aqueous metal
(0.340 V ) . (8.168)
The partial equilibria of equations 8.165→8.168 reveal the usefulness of standard

potentials: the Gibbs free energy change ⌬G of the redox equilibrium is always
given by applying Faraday’s equation to the algebraic sum of the standard poten-
tials of the redox couples in question. For equation 8.163, the bulk potential is
thus
0
E163 = 0.474 + 0.340 = 0.814 (V) , (8.169)
and for equation 8.164 it is
0
E164 = 0.763 + 0.340 = 1.103 (V). (8.170)
Based on Faraday’s law (equation 8.160), the ⌬G of the reaction is in both cases
negative: reactions then proceed spontaneously toward the right, with the dissolu-
tion of Fe and Zn electrodes, respectively, and deposition of metallic Cu at the
cathode.
Let us now consider a galvanic cell with the redox couples of equation 8.164.
This cell may be composed of a Cu electrode immersed in a one-molal solution
of CuSO4 and a Zn electrode immersed in a one-molal solution of ZnSO4 (“Dan-
iell cell” or “Daniell element”). Equation 8.170 shows that the galvanic potential
is positive: the ⌬G of the reaction is negative and the reaction proceeds toward
the right. If we short-circuit the cell to annul the potential, we observe dissolution
of the Zn electrode and deposition of metallic Cu at the opposite electrode. The
flow of electrons is from left to right: thus, the Zn electrode is the anode (metallic
Zn is oxidized to Zn2⫹; cf. eq. 8.167), and the Cu electrode is the cathode (Cu2⫹
ions are reduced to metallic Cu; eq. 8.168):
e−
 →
Zn, Zn 2 + , SO 24 − Cu, Cu 2 + , SO 24 − . (8.171)
䊞䊝
As we have already seen, the standard potentials are relative to standard refer-
ence conditions—i.e., one-molal solutions at Tr ⫽ 25 °C and Pr ⫽ 1 bar, in equilib-
rium with pure metals or pure gases. Applying the Nernst relation to a redox
equilibrium such as reaction 8.163 and assuming unitary activity for the con-
densed phases (i.e., pure metals), we have
a Fe 2 + ,aqueous
∆G163 = ∆G163
0
+ RT ln . (8.172)
a Cu2 + ,aqueous
Combining equation 8.172 with the Faraday relation (eq. 8.160), we obtain
RT aFe2 + ,aqueous
E163 = E163
0
− ln
nF aCu2 + ,aqueous
(8.173)
RT m 2+ RT γ Fe2 +
= E163
0
− ln Fe − ln .
nF mCu2 + nF γ Cu2 +
Equation 8.173 establishes that in an electrolytic solution the relative proportions

of oxidized and reduced species are controlled by the redox state of the system
(we will see in section 8.19 that in actual fact the redox equilibrium among the
various redox couples in natural, chemically complex, aqueous solutions is rarely
attained). Imposing on equation 8.173 the value of the ruling redox potential
(Eh)—i.e.,
E 163 = Eh, ( 8.174)

the activity ratio conforms to the difference between the imposed Eh and the
standard state potential:
RT a Fe 2 + ,aqueous
Eh − E 163
0
=− ln . (8.175)
nF a Cu2 + ,aqueous
Based on equation 8.173, the new equilibrium is attained through a change of

molality of ions in solution.
Let us now consider redox limits for the thermodynamic stability of aqueous
solutions. Maximum oxidation is defined by the dissociation of water molecules,
with the formation of hydrogen ions and gaseous oxygen—i.e.,
O2 + 4 H+ + 4 e − ⇔ 2 H2O .
gas aqueous
(8.176)
The standard potential of equation 8.176 is E 0176 ⫽ 1.228 V. At standard state,

the activity of gaseous oxygen is 1 by definition, and standard potential E 0176 thus
refers to H2O in equilibrium with an atmosphere of pure O2 at T ⫽ 25 °C and
P ⫽ 1 bar. Applying the Nernst and Faraday relations to equation 8.176 and
transforming natural logarithms into base 10 logarithms, we obtain
2.303RT 2.303RT
Eh = 1.228 + 4 log aH + + log f O2
nF nF (8.177)
= 1.228 + 0.0591 log aH + + 0.0148 log f O2 .
The maximum reduction limit for the thermodynamic stability of water is

defined by the reduction of hydrogen ions in solution with the formation of di-
atomic gaseous molecules H2 (normal hydrogen electrode):
2 H + + 2 e − ⇔ H2 .
aqueous gas
(8.178)
The redox potential dictated by equation 8.178 is related to standard potential

E 0178 through
Eh = 0.000 + 0.0591 log a H + − 0.0295 log f H 2 . ( 8.179)
We now recall that the ionic dissociation of water is
H2O ⇔ H+ + OH− (8.180)

The constant of equation 8.180 at T ⫽ 25 °C and P ⫽ 1 bar (cf. table 8.3) is
K 180 = a H + a OH− = 1.008 × 10 −14 . ( 8.181)
Expressed in logarithmic form, equation 8.181 becomes
log a H + + log a OH− ≈ −14.0. ( 8.182)
The definition of pH is
pH = − log a H + , (8.183)
hence
pH = 14.0 + log a OH− . ( 8.184)
Based on equation 8.184, we can subdivide the acid-base property of an aqueous

solution into two fields according to the relative predominance of hydroxyls OH⫺
and hydrogen ions H⫹ —i.e.,
a H + > a OH − ; pH < 7 (acidic solutions )
a H + = a OH − ; pH = 7 ( neutral solutions ) (8.185)

a H + < a OH − ; pH > 7 ( basic, or " alkaline," solutions ).
In analogy with acid-base neutrality at pH ⫽ 7, we can also define redox neutral-

ity through the molecular dissociation process:
2 H2O ⇔ 2 H2 + O2 . (8.186)
Redox neutrality corresponds to the condition
f H2 = 2 fO . (8.187)
(gas) 2 (gas)
Combining equation 8.187 with equations 8.176 and 8.178 and with the acid-
base neutrality concept (eq. 8.184 and 8.185), we obtain the absolute neutrality
condition for a diluted aqueous solution at T ⫽ 25 °C and P ⫽ 1 bar—i.e.,
pH = 7.0; log fH2 = −27.61; log fO2 = −27.91; E 0 = 0.4 ( V). (8.188)
Figure 8.19 Thermodynamic stability field for aqueous solutions at T ⫽ 25 °C and P ⫽

1 bar.
Figure 8.19 shows the thermodynamic stability field for diluted aqueous solu-
tions at T ⫽ 25 °C and P ⫽ 1 bar. The pH-Eh range is bounded upward by the
maximum oxidation limit, with a slope of ⫺0.0591 and an intercept on the ordi-
nate axis depending on the partial pressure of oxygen: several limiting curves are
drawn for various values of f O2. We recall here that the partial pressure of oxygen
in the earth’s atmosphere at sea level is about 0.2 bar, so that, assuming a fugacity
coefficient of 1 ( f O2 ⫽ PO2 ⫽ 0.2), the intercept on the ordinate axis at pH ⫽ 0 is
Eh ⫽ 1.22. The lower stability limit is defined, as we have already seen, by equa-
tion 8.179 for the various values of f H2. The absolute neutrality point subdivides
the Eh-pH field of thermodynamic stability into four parts:
1. Oxidizing acidic solutions

2. Oxidizing basic solutions
3. Reducing basic solutions
4. Reducing acidic solutions.
8.13 Relative Predominance Limits of Aqueous

Solutes and Eh-pH Diagrams
We have already seen that the molality ratio of species involved in redox equilibria
is buffered by the redox state of the system with respect to the standard potential
of the redox couple. To understand this concept better, let us examine the various
speciation states of a polyvalent element such as cerium in aqueous solution.

Cerium, in the absence of other anionic ligands besides OH⫺ groups, is mainly
present in aqueous solutions in three forms, Ce3⫹, Ce(OH)3⫹, and Ce(OH)2⫹ 2 ,
related by the homogeneous equilibria
( )
3+
Ce OH + H+ + e − ⇔ Ce3 + + H2O (8.189)
( )
2+
Ce OH + 2 H+ + e − ⇔ Ce3 + + 2 H2O (8.190)
2
( ) ( )
3+ 2+
Ce OH + H2O ⇔ Ce OH + H+ . (8.191)
2
The ruling equilibria between aqueous species and condensed forms are
Ce3 + + 3 e − ⇔ Ce
aqueous metal
(8.192)
Ce3 + + 3 H2O ⇔ Ce OH + 3 H+ ( ) 3 (8.193)

aqueous crystal
( ) ( )
2+
Ce OH + H2O+ e − ⇔ Ce OH + H+
2 3 (8.194)
aqueous crystal
( )
2+
CeO2 + 2 H+ ⇔ Ce OH ,
2 (8.195)
crystal aqueous
to which we can add the following equilibria among condensed forms:
( )
Ce OH + 3 H+ + 3 e − ⇔ Ce+ 3 H2O
3 (8.196)
crystal metal
CeO2 + H2O + H+ + e − ⇔ Ce OH . ( ) 3 (8.197)

crystal crystal
If we apply the Nernst and Faraday relations to equations 8.189 and 8.190 as
before (cf. eq. 8.177 and 8.179), we obtain
Ce OH
( ) 
3+
 
Eh = 0
E189 − 0.0591 pH + 0.0591 log (8.198)
Ce 3+ [ ]
and
Ce OH
( ) 
2+
 2 
Eh = E190
0
− 0.1182 pH + 0.0591 log , (8.199)
Ce 3+ [ ]
where terms in square brackets denote thermodynamic activities of
solutes, E 0189 ⫽ 1.715 V, and E 0190 ⫽ 1.731 V.
On an Eh-pH plot (figure 8.20), the isoactivity conditions for solute species:
(
Ce OH
) [
 = Ce 3+
]
3+
(8.200)
 
Figure 8.20 Eh-pH diagram for the Ce-H2O system (modified from Pourbaix, 1966). Fig-
ures on limiting curves are base 10 logarithms of solute activity: unitary activity (i.e., one-
molal solution) is identified by zero.
Ce OH
( ) [
 = Ce 3+
]
2+
(8.201)
 2 
are defined, respectively, by the equations
Eh = 1.715 − 0.0591 pH ( 8.202)
Eh = 1.731 − 0.1182 pH . ( 8.203)
Equations 8.202 and 8.203 define the “boundaries of relative predominance” of

solutes. Obviously, in the case of eq. 8.191, in which there is no electron transfer-
ence, the relative predominance limits of Ce(OH)3⫹ and Ce(OH)2⫹ 2 are deter-
mined solely by the acidity of the solution and not by its redox state—i.e.,
( )
Ce OH 
2+
 2 
log = −0.29 + pH . (8.204)
Ce (OH )
3+
The limit (equation 8.204) is thus parallel to the ordinate axis (pH ⫽ 0.29; cf.
figure 8.20).
As we did for equilibria between solute species, we can also define the bound-
aries between solute species and condensed phases. Assuming the condensed
forms to be pure phases (i.e., assuming unitary activity), in the presence of metal-
lic cerium we have
Eh = E 192
0
+ 0.0197 log Ce 3 + , [ ] ( 8.205)
with E 0192 ⫽ ⫺2.483 V; in the presence of Ce(OH)3 we have
+ 0.0591 pH+ 0.0591 log Ce OH ( ) ,

2+
Eh = E194
0
(8.206)
 2 
with E 0194 ⫽ 0.422; in the presence of Ce(OH)3 we also have
[ ]
log Ce 3+ = 22.15 + 3 pH, (8.207)
and in the presence of CeO2 we have
(
log Ce OH )  = 19.22 − 2 pH .
2+
 2  (8.208)
As figure 8.20 shows, the boundary between condensed phases and solute species
is dictated by the activity of solutes, T being equal.
Lastly, the boundary between Ce and Ce(OH)3 (eq. 8.196) is given by
Eh = E 196
0
− 0.0591 pH, ( 8.209)
with E 0196 ⫽ 2.046, and the boundary between Ce(OH)3 and CeO2 by
Eh = E197
0
− 0.0591 pH, (8.210)
with E 0197 ⫽ 1.559.

Usually, in plots such as those in figure 8.20, for the sake of simplicity, the
boundaries between solutes and condensed phases are not drawn for the various
molal concentrations in solution, as seen for cerium, but for a selected bulk molal
amount, which in geochemistry is normally 10⫺6. This choice is dictated by the
fact that, at this bulk molal concentration, the solid phase can be considered
effectively inert from the point of view of reactivity (cf. Garrels and Christ, 1965).
In the various heterogeneous equilibria involving aqueous solutions and con-
densed or gaseous phases, it is nevertheless opportune always to specify the molal
concentrations to which the various boundaries refer.
The main usefulness of Eh-pH diagrams consists in the immediacy of qualita-
tive information about the effects of redox and acid-base properties of the system
on elemental solubility. Concerning, for instance, cerium, figure 8.20 immediately
shows that, within the stability field of water, delimited upward by oxidation
boundary curve o and downward by reduction boundary curve r, the element (in
the absence of other anionic ligands besides OH⫺ groups) is present in solution
mainly as trivalent cerium Ce3⫹ and as soluble tetravalent hydroxide Ce(OH)2⫹ 2 .
It is also evident that, with increasing pH, cerium precipitates as trivalent hydrox-
ide Ce(OH)3 .
The interpretation of Eh-pH diagrams implies assumption of complete equi-
librium among the various solutes and condensed forms. Although this assump-
tion is plausible in a compositionally simple system such as that represented in
figure 8.20, it cannot safely be extended to more complex natural systems, where
the various redox couples are often in apparent disequilibrium. It is therefore
necessary to be cautious when dealing with the concept of the “system Eh” and
the various redox parameters.
8.14 “Electronic Activity,” “System Eh,” and Other

Redox Parameters
As repeatedly noted, standard potentials are linked to the standard molal Gibbs
free energy of formation from the elements through Faraday’s equation. Let us
now reconsider the equilibrium between metallic iron and aqueous ion at T ⫽
25 °C and P ⫽ 1 bar:
Fe ⇔ Fe2 + + 2 e − .
metal aqueous
(8.211)
At standard state, equation 8.211 obviously concerns standard state stable com-
ponents—i.e., pure Fe metal at T ⫽ 25 °C, P ⫽ 1 bar, and a hypothetical
one-molal Fe2⫹ solution referred to infinite dilution, at the same P and T condi-
tions (cf. section 8.4). The chemical potentials of components in reactions
(aFe(metal), aFe2⫹(aqueous), ae⫺(aqueous) ) are those of standard state; hence, by defini-
tion, the activity of all components in reaction is 1—i.e.,
a Fe (metal) = 1; a Fe 2 + = 1; a e− = 1, ( 8.212)
(aqueous) (aqueous)
where ae⫺ is “the activity of the hypothetical electron in solution,” which is 1, as

we have seen at standard state. Because, by convention, the partial molal Gibbs
free energy of formation of the hypothetical aqueous electron is assumed to be
zero at all P, T, and molality conditions—i.e.,
G f e − = G f0 e − = 0 ( 8.213)
—it follows that the activity of the hypothetical electron in solution ae⫺ is always
1 under all P, T, and molality conditions. We have already seen that, based on
this consideration, ae⫺ is never explicit in Nernst equations (as in eq. 8.172 and
8.175, for example).
However, in the literature we often encounter the redox parameter “pe,”
which, like pH, represents the cologarithm of the activity of the hypothetical elec-
tron in solution:
pe = − log ae− . (8.214)
The relationship between pe and Eh is
F
pe = Eh (8.215)
2.303RT
(cf. Stumm and Morgan, 1981). As outlined by Wolery (1983), equation 8.215
implies that the Gibbs free energy of formation of the hypothetical electron in
solution is not always zero, as stated in equation 8.213, but is given (cf. eq.
8.159) by
1
−
G f e (aqueous) = G H 2(gas) − G H + . (8.216)
2 (aqueous)
The incongruency between equations 8.216 and 8.213 leads us to discourage the
use of parameter pe, which is formally wrong from the thermodynamic point of
view. We also note that, based on equation 8.215, pe is simply a transposition in
scale and magnitude of Eh (pe is adimensional, Eh is expressed in volts).
For the sake of completeness, we recall that analogies with the concept of pH
do not concern solely the pe factor. In biochemistry, for instance, O2 and H2
fugacities in gaseous phases are often described by the rO and rH parameters, re-
spectively:
rO = − log f O 2 (gas) ( 8.217)
rH = − log f H 2 (gas) . ( 8.218)
In the description of redox equilibria in aqueous solutions, we adopted the

term “ruling Eh” and not “system Eh.” By “ruling Eh” we meant the Eh value
imposed on the redox couple to attain a determinate activity ratio between re-
duced and oxidized forms. As already noted, natural aqueous solutions often
contain dissolved elements in different and contrasting redox states, clearly in-
compatible with a single value for the “system Eh.” For instance, in almost all
natural aqueous solutions, organic carbon and dissolved oxygen coexist (Wolery,
1983). In several marine sediments, methane CH4 and bicarbonate ion HCO⫺ 3
coexist with dissolved sulfides and sulfates (Thorstenson, 1970). Berner (1971)
quotes the presence in surface seawaters of nitrogen and dissolved nitrates coexist-
ing with free oxygen. Table 8.15 (from Jenne, 1981) lists Eh potentials for seawa-
ters of different salinities in San Francisco Bay (California), calculated from the
activities of dissolved redox species and compared with the measured Pt-electrode
potential (Jenne et al., 1978).
Table 8.15 Eh values in seawaters of San

Francisco Bay relative to various redox
couples, compared with measured Eh
values (from Jenne, 1981).
Redox Couple Eh Potential (V)

[NO3]/[NO2] 0.44-0.43
[NO3]/[NH4] 0.33-0.32
[Fe3⫹]/[Fe2⫹] 0.12-0.05
[SO4]/[S2⫺] ⫺0.17-⫺0.20
Pt-electrode 0.39-0.36
Table 8.15 stresses the existence of marked redox disequilibria among the
various couples. In this case (and also for the previously quoted cases), the con-
cept of “system Eh” is inappropriate. Nevertheless, table 8.15 shows that the mea-
sured Eh is near the value indicated by the redox couple [NO3 ]/[NO2 ]. We can
thus assume that the redox equilibrium between [NO3 ] and [NO2 ] determines the
ruling Eh.
Of the various factors that cause redox disequilibria, the most effective are
biologic activity (photosynthesis) and the metastable persistence of covalent com-
plexes of light elements (C, H, O, N, S), whose bonds are particularly stable and
difficult to break (Wolery, 1983). For the sake of completeness, we can also note
that the apparent redox disequilibrium is sometimes actually attributable to ana-
lytical error or uncertainty (i.e., difficult determination of partial molalities of
species, often extremely diluted) or even to error in speciation calculations (when
using, for instance, the redox couple Fe3⫹ /Fe2⫹, one must account for the fact
that both Fe3⫹ and Fe2⫹ are partly bonded to anionic ligands so that their free
ion partial molalities do not coincide with the bulk molality of the species).
8.15 Eh-pH Diagrams for Main Anionic Ligands
Knowledge of the redox states of anionic ligands is fundamental for understand-

ing complexation processes operating in various geologic environments. The re-
dox states of the main anionic ligands under various Eh-pH conditions within
the stability range of water will be briefly described here. The predominance limits
appearing in figures 8.21 to 8.24 define the isoactivity conditions of the dissolved
species (cf., for instance, eq. 8.200 and 8.201). The diagrams are taken from Broo-
kins (1988), to whom we refer readers for more exhaustive explanations. The
adopted standard molal Gibbs free energy values of the various species do not
differ appreciably from the tabulated values of Shock and Helgeson (1988) (table
8.13) and come mainly from Wagman et al. (1982). The two series of data are
compared in table 8.16. The observed discrepancies do not greatly alter the pre-
dominance limits drawn in the figures.
Carbon is present in nature in three oxidation states (4⫹, 0, and 4⫺). The
predominance limits of figure 8.21A relate to a molal concentration of total dis-
solved C of 10⫺3. Carbonic acid H2CO3 and its ionization products occur over
most of the Eh-pH stability field of water. At pH higher than approximately 6.4,
the bicarbonate ion HCO3⫺ predominates, and at pH higher than 10.3, the car-
bonate ion CO2⫺ 3 predominates. Native carbon C appears to be stable at reducing
conditions in a very restricted Eh range, below which aqueous methane CH4(aq)
occurs. The development of gaseous methane is observed under reducing condi-
tions corresponding to the lower limit of stability of water (r).
Table 8.16 Standard molal Gibbs free energies of formation from the elements for aqueous ions
and complexes and condensed phases, partly adopted in constructing the Eh-pH diagrams in figure
8.21. Data in kcal/mole. Values in parentheses: Shock and Helgeson’s (1988) tabulation. Sources of
data: (1) Wagman et al. (1982); (2) Garrels and Christ (1965); (3) Pourbaix (1966); (4) Berner (1971)
Ion or Ion or
Complex G f,0 T r , P r Reference Complex G f,0 T r , P r Reference
BO2 ⫺162.26 (⫺162.24) (1) S2⫺ 20.51 (19.00) (1)
B2O3(cr) ⫺285.29 (1) HS⫺ 2.89 (2.86) (1)
B4O72⫺ ⫺622.56 (1) H 2S ⫺6.65 (1)
BH4⫺ 27.33 (1) SO32⫺ ⫺116.28 (⫺116.30) (1)
B(OH)4⫺ ⫺275.61 (1) HSO3⫺ ⫺126.13 (⫺126.13) (1)
H3BO3 ⫺231.54 (1) SO42⫺ ⫺177.95 (⫺177.93) (1)
H2BO3⫺ ⫺217.60 (3) HSO4⫺ ⫺180.67 (⫺180.63) (1)
HBO32⫺ ⫺200.29 (3) F⫺ ⫺66.63 (⫺67.34) (1)
BO33⫺ ⫺181.48 (3) Cl⫺ ⫺31.36 (⫺31.38) (1)
CH4 ⫺8.28 (1) ClO3⫺ ⫺1.90 (1.90) (1)
CH2O ⫺31.00 (2) ClO4⫺ ⫺2.04 (⫺2.04) (1)
H2CO3 ⫺149.00 (1) Br⫺ ⫺24.85 (24.87) (1)
HCO3⫺ ⫺140.24 (⫺140.28) (1) BrO3⫺ 4.46 (4.45) (1)
CO32⫺ ⫺126.15 (⫺126.19) (1) BrO4⫺ 28.23 (28.20) (1)
NO3 ⫺25.99 (⫺26.51) (1) I⫺ ⫺12.33 (⫺12.41) (1)
NH4⫹ ⫺18.96 (⫺18.99) (1) IO3⫺ ⫺30.59 (⫺30.60) (1)
NH3(gas) ⫺3.98 (4) IO4⫺ ⫺13.98 (⫺14.00) (1)
H3PO4 ⫺273.07 (1) H2PO4⫺ ⫺270.14 (1)
HPO42⫺ ⫺260.31 (⫺260.41) (1) PO42⫺ ⫺243.48 (⫺243.50) (1)
Figure 8.21B shows the Eh-pH diagram for nitrogen, calculated for fN2 ⫽ 0.8
bar (total molality of dissolved N species ⫽ 10⫺3.3 ). The gaseous molecule N2 is
stable over most of the Eh-pH field. The ammonium ion NH⫹ 4 predominates un-
⫺
der reducing conditions, and the nitrate ion NO3 predominates under oxidizing
conditions. Gaseous ammonia forms only under extremely reducing and alkaline
conditions. The fields of ammonium and nitrate ions expand metastably within
the stability field of gaseous nitrogen (the limit of metastable predominance of
the two species is drawn as a dashed line within the N2 field in figure 8.21B).
Figure 8.21C shows the Eh-pH diagram for phosphorus at a solute total mol-
ality of 10⫺4. Within the stability field of water, phosphorus occurs as orthophos-
phoric acid H3PO4 and its ionization products. The predominance limits are dic-
tated by the acidity of the solution and do not depend on redox conditions.
Sulfur (figure 8.21D) is present in aqueous solutions in three oxidation states
(2⫺, 0, and 6⫹). The field of native S, at a solute total molality of 10⫺3, is very
limited and is comparable to that of carbon (for both extension and Eh-pH
range). Sulfide complexes occupy the lower part of the diagram. The sulfide-
sulfate transition involves a significant amount of energy and defines the limit of
predominance above which sulfates occur.
The extension of the field of native elements (S, C) depends on the adopted
Figure 8.21 Eh-pH diagrams for main anionic ligands. From Brookins (1988). Reprinted
with permission of Springer-Verlag, New York.
molality values of solutes (i.e., it expands with increasing, and shrinks with de-
creasing, molal concentrations). The molal concentrations used for constructing
figure 8.21 are mean values for natural waters, but they may vary considerably.
The above-discussed anionic ligands are the most important ones from the
viewpoint of redox properties. For the sake of completeness, we can also consider
the redox behavior of halides and boron. Halides stable in water are the simple
ions Cl ⫺, Br⫺, F⫺, and I⫺. At high Eh conditions, IO⫺

3 ion forms; the predomi-
nance limit between I⫺ and IO⫺ 3 is given by
Eh = 1.09 − 0.0591 pH . ( 8.219)
For all the other halides, Eh-pH conditions have no influence. Boron occurs in
water mainly as boric acid H3BO3 and its progressive ionization products at in-
creasing pH. Redox conditions do not affect the speciation state of boron.
8.16 Complex Eh-pH Diagrams: Examples of Iron,

Manganese, and Copper
Information on the speciation states of solutes and their equilibria with con-
densed phases furnished by Eh-pH diagrams is often simply qualitative and
should be used only in the initial stages of investigations. The chemical complexity
of natural aqueous solutions and the persistent metastability and redox disequi-
librium induced by organic activity are often obstacles to rigorous interpretation
of aqueous equilibria.
Let us now consider the effects of chemical complexity on the predominance
limits of solutes at environmental P and T conditions. The elements in question
are emblematic for their particular importance in ore formation.
Figure 8.22A shows the Eh-pH diagram of iron in the Fe-O-H system at T ⫽
25 °C and P ⫽ 1 bar. The diagram is relatively simple: the limits of predominance
are drawn for a solute total molality of 10⫺6. Within the stability field of water,
iron is present in the valence states 2⫹ and 3⫹. In figure 8.22A, it is assumed that
the condensed forms are simply hematite Fe2O3 and magnetite Fe3O4 . Actually,
in the 3⫹ valence state, metastable ferric hydroxide Fe(OH)3 and metastable goe-
thite FeOOH may also form, and, in the 2⫹ valence state, ferrous hydroxide
Fe(OH)2 may form. It is also assumed that the trivalent solute ion is simply Fe3⫹,
whereas, in fact, various aqueous ferric complexes may nucleate [i.e., Fe(OH)2⫹,
Fe(OH)⫹ 2 , etc.].
Figure 8.22B shows the effect of adding carbon to the system (molality of
solute carbonates ⫽ 10⫺2 ): in reducing alkaline states, the field of siderite FeCO3
opens; in the other Eh-pH conditions, the limits are unchanged.
Adding sulfur to the system (molality of solutes ⫽ 10⫺6; figure 8.22C), a wide
stability field opens for pyrite FeS2 in reducing conditions and, almost at the
lower stability limit of water, a limited field of pyrrhotite FeS is observed.
Figure 8.22 shows the effect of silica in the system (aqueous solution saturated
with respect to amorphous silica): under reducing alkaline conditions, the meta-
silicate FeSiO3 forms at the expense of magnetite through the equilibrium
Fe 3 O 4 + 3 SiO 2 + 2 H + + 2 e − ⇔ 3 FeSiO 3 + H 2 O ( 8.220 )

Figure 8.22 Eh-pH diagrams for iron-bearing aqueous solutions. Chemical complexity
increases from A to F.
with a limit defined by
Eh = 0.272 − 0.0591 pH . ( 8.221)
Reaction 8.220 takes place in the silicate bands present in iron ores: iron silicates
occur in such bands in coexistence with opaline amorphous silica (Klein and
Bricker, 1972). A complex Eh-pH diagram for the Fe-C-S-O-H system is shown
in figure 8.22E. The stability field of pyrrhotite FeS disappears and is replaced by
siderite FeCO3 , which also limits the field of pyrite. When silicon is also present
in the system (figure 8.22F), magnetite does not nucleate as a stable phase.
The predominance limits shown in figure 8.22 are analytically summarized in
table 8.17. Compare figures 8.22 and 8.21 to better visualize the redox state of the
anionic ligands at the various Eh-pH conditions of interest (particularly the
sulfide-sulfate transition and carbonate limits). We remand to Garrels and Christ
(1965) for a more detailed account on the development of complex Eh-pH dia-
grams.
Let us now consider the effects of Eh-pH conditions on the speciation state
and solubility of manganese in aqueous solutions. Manganese complexes have
been carefully studied in the last decade, owing to the discovery on ocean floors
of economically important metalliferous deposits (nodules and crusts) in which
Mn compounds are dominant.
Figure 8.23A shows a simplified Eh-pH diagram for the Mn-O-H system.
Within the stability field of water, manganese occurs in three valence states (2⫹,
3⫹, and 4⫹). Figure 8.23A shows the condensed phases relative to the three va-
lence states as the hydroxide pyrochroite Mn(OH)2 (2⫹), multiple oxide haus-
mannite Mn3O4 (2⫹, 3⫹), sesquioxide Mn2O3 (3⫹), and oxide pyrolusite
MnO2 (4⫹).
A wide field is occupied by the Mn2⫹ ion. The condensed phases nucleate
only in alkaline solutions, and the stability fields of the various solids expand with
increasing Eh. The precipitation of Mn oxides and hydroxides is actually a com-
plex phenomenon involving the initial nucleation of metastable compounds. Hem
and Lind (1983) showed that, at fO2 near the upper stability limit of water and at
pH in the range 8.5 to 9.5, the type of nucleating solid depends on temperature:
at T ⫽ 25 °C metastable hausmannite Mn3O4 forms, whereas at T near 0 °C the
mixed hydroxides feitknechtite, manganite, and groutite [all with formula unit
MnO(OH)] precipitate. All these compounds are metastable in experimental Eh-
pH conditions and transform progressively into the tetravalent compound MnO2 .
Figure 8.23 Eh-pH diagrams for manganese. From Brookins (1988). Reprinted with per-
mission of Springer-Verlag, New York.
Table 8.17 Main predominance limits of aqueous complexes and saturation limits between
solutes and condensed phases in iron-bearing aqueous solutions (see figure 8.22). Standard
state Gibbs free energies of formation of species are listed in table 8.18. (c) ⫽ crystalline;
(a) ⫽ aqueous; (g) ⫽ gaseous; (l) ⫽ liquid. [ ] ⫽ thermodynamic activity.
System Fe-O-H; T⫽ 25 °C; P ⫽ 1 bar; solutes molality ⫽ 10⫺6

Fe2O3(c) ⫹ 6H⫹
(a) → 2Fe(a) ⫹ 3H2O(1)
3⫹
log[Fe3⫹] ⫽ ⫺0.720 ⫺ 3pH

3⫹
Fe(a) ⫹ e⫺ → Fe(a)
2⫹
Eh ⫽ 0.78
2⫹
2Fe(a) ⫹ 3H2O(l) → Fe2O3(c) ⫹ 6H⫹
(a) ⫹ 2e
⫺
Eh ⫽ 0.728 ⫺ 0.0591 log[Fe2⫹] ⫺ 0.177pH

2⫹
3Fe(a) ⫹ 4H2O(l) → Fe3O4(c) ⫹ 8H⫹
(a) ⫹ 2e
⫺
Eh ⫽ 0.980 ⫺ 0.0885 log[Fe ] ⫺ 0.236pH

2⫹
2 Fe3O4(c) ⫹ H2O(l) → 3 Fe2O3(c) ⫹ 2H⫹

(a) ⫹ 2e
⫺
Eh ⫽ 0.221 ⫺ 0.0591pH
System Fe-C-O-H; ƒco2(g) ⫽ 10⫺2; [H2CO3] ⫽ 10⫺2

log[H2CO3] ⫽ ⫺2.0
log[HCO3⫺] ⫽ ⫺6.4 ⫹ log[H2CO3] ⫹ pH
FeCO3(c) ⫹ 2H⫹
(a) → Fe(a) ⫹ CO2(g) ⫹ H2O(l)
2⫹
log[Fe2⫹] ⫽ 7.47 ⫺ log ƒco2(g) ⫺ 2pH

3FeCO3(c) ⫹ 4H2O(l) → Fe3O4(c) ⫹ 3HCO⫺ ⫹
3(a) ⫹ 5H(a) ⫹ 2e
⫺
Eh ⫽ 1.010 ⫺ 0.148pH ⫹ 0.0885 log[HCO3⫺]
System Fe-S-O-H; solutes molality ⫽ 10⫺6

FeS(c) ⫹ H⫹ ⫺
(a) → Fe(a) ⫹ HS(a)
2⫹
log[Fe2⫹] ⫽ ⫺4.40 ⫺pH ⫺ log[HS⫺]

2H2S(a) ⫹ Fe(a)
2⫹
→ FeS2(c) ⫹ 4H⫹ ⫹ 2e⫺
Eh ⫽ 0.057 ⫺ 0.118pH ⫺ 0.0591 log[H2S] ⫺ 0.0295 log[Fe2⫹]
FeS2(c) ⫹ 8H2O(l) → 2HSO⫺ ⫹
4(a) ⫹ Fe(a) ⫹ 14H(a) ⫹ 14e
2⫹ ⫺
Eh ⫽ 0.339 ⫺ 0.0591pH ⫹ 0.0084 log[HSO4⫺] ⫹ 0.0042 log[Fe2⫹]

FeS(c) ⫹ H2S(a) → FeS2(c) ⫹ 2H⫹
(a) ⫹ 2e
⫺
Eh ⫽ 0.133 ⫺ 0.0591pH ⫺ 0.0295 log[H2S]

3FeS2(c) ⫹ 28H2O(l) → Fe3O4(c) ⫹ 6SO4(a)
2⫺
⫹ 56H⫹
(a) ⫹ 44e
⫺
Eh ⫽ 0.384 ⫺ 0.075pH ⫹ 0.0080 log[SO4 ] 2⫺
FeS(c) ⫹ HS⫺ ⫹
(a) → FeS2(c) ⫹ H(a) ⫹ 2e
⫺
Eh ⫽ ⫺0.340 ⫺ 0.0295pH ⫺ 0.0295 log[HS⫺]
System Fe-Si-O-H; unitary activity of amorphous silica

2FeSiO3(c) ⫹ H2O(l) → Fe2O3(c) ⫹ 2SiO2(amorphous) ⫹ 2H⫹ ⫹ 2e⫺
Eh ⫽ 0.258 ⫺ 0.0591pH
FeSiO3(c) ⫹ 2H⫹
(a) → Fe(a) ⫹ SiO2(amorphous) ⫹ H2O(l)
2⫹
log[Fe ] ⫽ 8.03 ⫺ 2pH

2⫹
Figure 8.24 Eh-pH diagrams for copper. From Brookins (1988). Reprinted with permis-
sion of Springer-Verlag, New York.
Figure 8.23B shows the effect of adding sulfur and carbon in solution (solute
molality ⫽ 10⫺3 ). In reducing-alkaline conditions, both rhodochrosite MnCO3
and alabandite MnS are stable. However, the field of alabandite is quite restricted,
which explains why this mineral is rarely found in nature (Brookins, 1988).
Figure 8.24A shows the Eh-pH diagram for copper in the Cu-O-H system
(solute molality ⫽ 10⫺6 ). Within the stability field of water, copper assumes the
valence states 0, ⫹1, and ⫹2. Native copper occupies much of the diagram, en-
compassing all acidity states and extending to Eh conditions higher than the
sulfide-sulfate transition (cf. figure 8.21). In acidic solutions, native copper oxi-
dizes to cupric ion Cu2⫹. In alkaline solutions, with increasing Eh, cuprite Cu2O
and tenorite CuO form. Only in extremely alkaline solutions is the soluble com-
pound CuO2⫺ 2 stable. If carbon and sulfur are added to the system (figure 8.24B;
solute molality ⫽ 10⫺3 ), the field of native copper is reduced, owing to the pres-
ence of the sulfides calcocite Cu2S and covellite CuS. At high Eh, we find the
hydrated carbonate malachite Cu2(OH)2CO3 replacing tenorite. In the presence
of iron, the diagram becomes much more complex, with the formation of impor-
tant ore-forming phases, such as chalcopyrite CuFeS2 and bornite Cu5FeS4 (see,
for instance, figures 7.24 and 7.25 in Garrels and Christ, 1965).
The sequence of diagrams shown in figures 8.22, 8.23, and 8.24 is only an
example of an approach to the actual complexity of natural solutions. The infor-
mation given by such diagrams, although qualitative, furnishes a basis for more
rigorous calculations and is therefore extremely precious in the comprehension
of natural phenomena. Table 8.18 lists the standard state Gibbs free energy values
used in the construction of these diagrams.
8.17 Notes on Metal-Organic Complexation
Although the predominant species forming in the C-H-O system (besides H2O)
are methane CH4 , carbon dioxide CO2 and its aqueous complexes HCO3 and
Table 8.18 Standard state (T ⫽ 25 °C; P ⫽ 1
bar) Gibbs free energy of formation for iron,
copper, and manganese compounds, partly
adopted in constructing Eh-pH diagrams of
figures 8.22, 8.23, and 8.24. References: (1)
Wagman et al. (1982); (2) Garrels and Christ
(1965); (3) Robie et al. (1978); (4) Hem et al.
(1982); (5) Shock and Helgeson (1988); (6)
Bricker (1965).
Fe ⫽ 0.0
Fe2⫹ ⫽ ⫺20.30 (2)
Fe3⫹ ⫽ ⫺2.52 (2)
Fe(OH)2⫹ ⫽ ⫺55.91 (2)
Fe(OH)⫹2 ⫽ ⫺106.2 (2)
FeO2H⫺ ⫽ ⫺90.6 (2)
Fe0.95O (wustite) ⫽ ⫺58.4 (2)
Fe2O3 (hematite) ⫽ ⫺177.1 (2)
Fe3O4 (magnetite) ⫽ ⫺242.4 (2)
Fe(OH)2 ⫽ ⫺115.57 (2)
Fe(OH)3 ⫽ ⫺166.0 (2)
FeS (pyrrhotite) ⫽ ⫺23.32 (2)
FeS2 (pyrite) ⫽ ⫺36.00 (2)
FeCO3 (siderite) ⫽ ⫺161.06 (2)
Cu ⫽ 0.00
Cu⫹ ⫽ 11.94 (1)
Cu2⫹ ⫽ 15.65 (1)
Cu2S ⫽ ⫺20.60 (chalcocite) (1)
CuS ⫽ ⫺12.81 (covellite) (1)
Cu2O ⫽ ⫺34.98 (cuprite) (2)
CuO ⫽ ⫺31.00 (tenorite) (1)
CuO22⫺ ⫽ ⫺43.88 (1)
Cu2(CO3)(OH)2 ⫽ ⫺213.58 (malachite) (1)
Cu3(CO3)2(OH)2 ⫽ ⫺314.29 (azurite) (1)
Mn2⫹ ⫽ ⫺54.52 (1)
Mn(OH)⫹ ⫽ ⫺96.80 (1)
Mn(OH)3⫺ ⫽ ⫺177.87 (1)
MnSO4 ⫽ ⫺232.5 (1)
MnO4⫺ ⫽ ⫺106.9 (5)
MnO42⫺ ⫽ ⫺119.7 (5)
MnO ⫽ ⫺86.74 (1)
Mn(OH)2 ⫽ ⫺146.99 (1); ⫺147.14 (6) (pyrochroite)
MnCO3 ⫽ ⫺195.20 (1); ⫺195.04 (3) (rhodochrosite)
MnS ⫽ ⫺52.20 (1)
Mn3O4 ⫽ ⫺306.69 (1) (hausmannite)
Mn2O3 ⫽ ⫺210.59 (1)
Mn(OH)3 ⫽ ⫺181.0 (2)
␤ MnO(OH) ⫽ ⫺129.8 (4) (feitknechtite)
␥ MnO(OH) ⫽ ⫺133.3 (6) (manganite)
␣ MnO2 ⫽ ⫺108.3 (6) (birnessite)
␤ MnO2 ⫽ ⫺111.17 (3) (pyrolusite)
␥ MnO2 ⫽ ⫺109.1 (6) (␥-nsutite)
CO2⫺3 , and graphite C, at low P and T conditions, several organic complexes form
metastably, mainly as the result of bacterial activity, and persist at P and T condi-
tions that may exceed those achieved at a burial depth on the order of 5 km,
which is generally considered the limit of complete degradation.
Organic matter suspended in solution and present in sediments is an efficient
complexing agent for most divalent and trivalent cations. Most organic matter in
aquatic and sedimentary environments is in the form of humic and fulvic acids.
About 80% of soluble organic matter in lakes and rivers is composed of fulvic
acids, whereas the abundance of humic acids varies markedly from region to re-
gion (from 0.1 to about 10 mg/l of solution; cf. Reuter and Perdue, 1977). In
classical terminology, “humic acids” are substances extracted from soils and sedi-
ments by alkaline solutions and precipitated through acidification, whereas “ful-
vic acids” refer to organic matter remaining in solution after acidification. These
substances are practically composed of polyelectrolytes varying in colour from
yellow to black, with molecular weights ranging from few a hundred to several
hundred thousand grams per mole. The relationships between macroscopic and
reactive properties for these classes of substances are outlined in figure 8.25.
The capacity of complexing of humic substances is ascribed to their oxygen-
based functional group (table 8.19).
The most common apical structure complexing metal cations in solution is
Figure 8.25 Classification and properties of humic substances. Reproduced with modifi-
cations from Stevenson (1983), with kind permission of Theophrastus Publishing and Pro-
prietary Co.
Table 8.19 Oxygen-based functional groups in humic and fulvic acids (from Stevenson,
1982). Concentrations expressed in mEq/100 g.
Acid Type Total Acidity COOH OH-Acid OH-Alcoholic C⫽O

Humic 560–890 150–570 210–570 20–496 10–560
Fulvic 640–1420 520–1120 30–570 260–950 120–420
Figure 8.26 (A) Functional groups of humic substances. (B) Complexing schemes for
fulvic substances. Reprinted from Stevenson (1983), with kind permission of Theophras-
tus Publishing and Proprietary Co.
shown in figure 8.26A. Figure 8.26B shows the most important complexing path-
ways followed by fulvic substances on copper, through the OH-phenolic and
COOH functional groups.
As shown in figure 8.25, the acidic-exchange capacity of humic substances
varies widely; a representative mean value is in the range 1.5 to 5 mEq/g. Assum-
ing a mean carbon content of 56% in polymer units, Stevenson (1983) deduced
that the ratio between complexed divalent cations and carbon atoms was around
1:20 to 1:60. Complexing capability is also greatly affected by the pH and ionic
strength of the solution (Guy et al., 1975). Generally, moreover, for elements
present in trace concentration levels, complexing by humic substances is more
efficient in pure water than in saline solutions, where organic matter is partly
bonded to the major dissolved cations (Ca2⫹, Mg2⫹; see Mantoura et al., 1978).
Table 8.20 Metal-organic complexing (log ␤0) for some trace elements in
aqueous solution at pH ⫽ 8.2, T ⫽ 25 °C. I: ionic strength of solution.
References: (1) Turner et al. (1981); (2) Mantoura et al. (1978); (3)
Stevenson (1976).
Ion I⫽0 I ⫽ 0.02 I ⫽ 0.65 % Chelate Metal

Mn2⫹ 5.00 (1) 4.53 (2) 3.61 (1) 0.01
Co2⫹ 5.20 (1) 4.73 (2) 3.81 (1) 0.01
Ni2⫹ 5.78 (1) 5.31 (2) 4.39 (1) 0.04
Cu2⫹ 9.84 (1) 9.37 (2) 8.45 (1) 47
Zn2⫹ 5.73 (1) 5.26 (2) 4.34 (1) 0.03
Cd2⫹ 5.23 (1) 4.76 (2) 3.84 (1) 0.001
Hg2⫹ 20.00 (1) 19.53 (2) 18.61 (1) 0.0
Pb2⫹ 8.70 (1) - 7.31 (1) 2
Table 8.20 lists metal-organic complexing constants for some trace metals in wa-
ters of various ionic strengths, according to Stevenson (1976), Mantoura et al.
(1978), and Turner et al. (1981). The last column in table 8.20 lists the percentage
of chelate metal in seawater (I ⫽ 0.65), which is significant for Cu2⫹. According
to Turner et al. (1981), the metal-organic complexing in seawater is also signifi-
cant for rare earths and fully hydrolyzed elements forming stable complexes with
carbonate ions (cf. figure 8.16B).
In addition to their complexing capability, humic substances can also reduce
oxidized forms of metal cations and polyanions (Fe3⫹ → Fe2⫹; MoO2⫺ 4 → Mo ;
5⫹
VO3 → VO ; see Szilagyi, 1971). The reduced forms are then embodied in the
2⫹
structure of humic substances by functional-group fixation.

Besides humic and fulvic substances in surface waters and soils, organic com-
pounds occur in a variety of geochemical environments. Their role is determinant
in the following processes.
1. Metal complexation (and fixation) in soils and sedimentary basins

2. Deposition of ore metals (reducing agents)
3. Formation of graphite in metamorphic processes
4. Complexation of metal cations and transport in ore-forming solutions
derived from sedimentary basins by organic acid anionic complexes pres-
ent in oil field brines
5. Generation, transport, and deposition of petroleum
6. Formation of secondary porosity in oil reservoir formations
7. Origin of life.
Impressive improvements in our knowledge of thermodynamic properties of

the various organic ligands and of our capability of calculating metal-organic
complexation constants at the various P and T conditions of interest come from
the systematizations of Shock and Helgeson (1990) and Shock and Koretsky
(1995), which evaluated equation of state parameters to be used in the revised
HKF model (see section 8.11). The main observations on which the generaliza-
tions were based are as follows.
1. The standard partial molal properties of aqueous electrolytes consisting

of inorganic cations and organic anions exhibit P-T-dependent behavior
similar to that of inorganic aqueous electrolytes (i.e., at pressures corre-
sponding to liquid-vapor equilibrium, with increasing T they reach a
maximum at intermediate T values and then approach ⫺∞ as T ap-
proaches the critical point of H2O; see figure 8.27A).
2. The standard partial molal properties of neutral organic aqueous mole-
cules behave similarly to their inorganic counterparts (i.e., at pressures
corresponding to liquid-vapor equilibrium, with increasing T they reach
a minimum at intermediate T values and then approach ⫹∞ as T ap-
proaches the critical point of H2O; see figure 8.27B).
3. The Born coefficients of the HKF model for the various organic species
are simply related to their standard partial molal entropies at 25 °C and
1 bar through the effective charge (Shock et al., 1989).
4. The standard partial molal volumes (V 0 ), heat capacities (C0P ), and entro-
pies (S 0 ) of aqueous n-polymers, together with their standard partial
molal enthalpies ( ⌬H0f ) and Gibbs free energies of formation from the
elements ( ⌬G0f ), are linear functions of the number of moles of carbon
atoms in the alkyl chains (figure 8.28).
T(°C) T(°C)
Figure 8.27 Experimental values of standard partial molal heat capacity of sodium ace-
tate (A) and ethylene (B) in water as a function of T(°C) at pressures corresponding to
water-vapor equilibrium. Interpolating curves generated by the HKF model equations.
Reprinted from E. L. Shock and H. C. Helgeson, Geochimica et Cosmochimica Acta, 54,
915–946, copyright 䉷 1990, with kind permission from Elsevier Science Ltd., The Boule-
vard, Langford Lane, Kidlington 0X5 1GB, UK.
Figure 8.28 Correlation of standard partial molal properties of aqueous n-polymers with
the number of moles of carbon atoms in the alkyl chain. Reprinted from E. L. Shock and
H. C. Helgeson, Geochimica et Cosmochimica Acta, 54, 915–946, copyright 䉷 1990, with
0X5 1GB, UK.
The various polymerization steps along the series of alcohols may be visual-
ized in stereochemical representation as follows:
H H H H H H H H
       
H− C − OH+ C ⇔ H− C − C − OH+ C ⇔ H− C− C− C− OH.
        (8.222)
H H H H H H H H
methanol ethanol propanol
The linearity in the standard partial molal properties of the aqueous polymers
as a function of the length of the chain (in terms of C atoms) observed by Shock
and Helgeson (1990) is the result of the principle of equal reactivity of co-
condensing functional groups (which, as we have already seen in section 6.1.2,
also holds for silica polymers). This principle is, however, strictly valid only when
the length of the chain is sufficiently elevated and small departures are observed
for chains with one or two carbon atoms (cf. figure 8.28).
The generalized behavior is consistent with an equation of the form
Ξ 0j = mn + p ( 8.223)
(Shock and Helgeson, 1990), where ⌶0j is the regressed standard state partial
molal property of the jth aqueous polymer, n is the number of moles of carbon
atoms in the alkyl chain, and m and p are, respectively, slope and intercept of the
regression lines shown in figure 8.28. Values of the regression constants for vari-
ous aqueous polymers according to Shock and Helgeson (1990) are listed in
table 8.21.
Table 8.21 Correlation parameters between the standard partial molal properties of
aqueous n-polymers and number of moles of carbon in the alkyl chain ( T ⫽ 25 °C, P ⫽ 1
bar; Shock and Helgeson, 1990). See section 8.11 and tables 8.11, 8.12, and 8.13 for
identification of symbols and units.
Property S P0 r , T r C P0 V P0 r , T r G ƒ0 , P r , T r H ƒ0 , P r , T r
Slope 6.7 21.2 15.8 2.05 ⫺5.67

Group INTERCEPTS
n-alkanes 12.8 47.0 19.6 ⫺7.6 ⫺13.5
1-alkenes - - - 11.9 -
n-alkynes - - - 42.2 -
n-alkylbenzenes 31.3 81.3 81.6 28.2 8.6
1-alcohols 23.2 19.0 22.6 ⫺47.8 ⫺58.4
Aldehydes - - - ⫺36.7 -
2-ketones 23.2 - 18.9 ⫺44.8 ⫺45.2
1-amines 20.0 14.4 26.8 1.8 ⫺12.3
n-carboxylic acid 29.3 ⫺2.4 20.0 - ⫺105.0
n-amino acids - - - ⫺95.4 ⫺115.5
Methyl alkanoates - - - - ⫺97.8
Table 8.22 Partial molal thermodynamic properties of organic aqueous species at 25 °C and 1 bar and the parameters of the revised
HKF model, after Shock and Helgeson (1990).
Species G ƒ0 , P r , T r H ƒ0 , P r , T r S P0 r , T r C P0 V P0 r , T r a1 ⫻ 10 a2 ⫻ 10⫺2 a3 a4 ⫻ 10⫺4 c1 c2 ⫻ 10⫺4 ␻ ⫻ 10⫺5

Methane ⫺8,234. ⫺21,010. 20.99 66.3 37.3 6.7617 8.7279 2.3212 ⫺3.1397 42.0941 10.4707 ⫺0.3179
Ethane ⫺3,886. ⫺24,650. 26.81 88.3 51.2 8.6340 13.3011 0.5205 ⫺3.3288 54.1755 14.9521 ⫺0.4060
Propane ⫺1,963. ⫺30,490 33.37 110.6 67.0 10.7625 18.4948 ⫺1.5133 ⫺3.5435 66.3294 19.4946 ⫺0.5053
n-butane 36. ⫺36,230. 40.02 133.9 82.8 12.8905 23.6960 ⫺3.5683 ⫺3.7585 79.0569 24.2408 ⫺0.6061
n-pentane 2,130. ⫺41,560. 47.5 153. 98.6 15.0143 28.8771 ⫺5.5949 ⫺3.9727 89.2070 28.1315 ⫺0.7193
n-hexane 4,420. ⫺47,400. 52.9 175.2 114.4 17.1487 34.0927 ⫺7.6530 ⫺4.1883 101.4642 32.6536 ⫺0.8011
n-heptane 6,470. ⫺52,950. 60.0 196.4 130.2 19.2745 39.2810 ⫺9.6879 ⫺4.4028 112.8980 36.9721 ⫺0.9086
n-octane 8,580. ⫺59,410. 63.8 217.6 146.0 21.4171 44.5126 ⫺11.7437 ⫺4.6191 124.7924 41.2905 ⫺0.9662
Ethylene 19,450. 8,570. 28.7 62.5 45.5 7.8560 12.6391 ⫺1.8737 ⫺3.3014 39.1 9.7 ⫺0.4
1-propene 17,910. ⫺290. 36.7 83.7 61.3 9.9655 16.5518 ⫺0.7564 ⫺3.4632 50.0997 14.0151 ⫺0.5558
1-butene 20,310. ⫺5,635. 43.4 104.9 77.1 12.0932 21.7476 ⫺2.7991 ⫺3.6779 61.5894 18.3335 ⫺0.6572
1-pentene 22,470. ⫺11,200. 50.1 126.1 92.9 14.2210 26.9431 ⫺4.8413 ⫺3.8927 73.0791 22.6520 ⫺0.7587
1-hexene 24,370. ⫺17,025. 56.8 147.3 108.7 16.3488 32.1387 ⫺6.8838 ⫺4.1075 84.5689 26.9704 ⫺0.8602
1-heptene 26,450. ⫺22,670. 63.5 168.5 124.5 18.4766 37.3342 ⫺8.9260 ⫺4.3223 96.0586 31.2889 ⫺0.9616
1-octene 28,720. ⫺28,120. 70.2 189.7 140.3 20.6044 42.5297 ⫺10.9682 ⫺4.5371 107.5483 35.6073 ⫺1.0631
Ethyne 51,890. 50,700. 30.0 53.0 49.3 8.3577 12.6272 0.7833 ⫺3.3009 33.0423 7.7615 ⫺0.4543
1-propyne 47,880. 38,970. 36.7 74.2 65.1 10.4855 17.8227 ⫺1.2588 ⫺3.5157 44.5321 12.0799 ⫺0.5558
1-butyne 50,030. 33,400. 43.4 95.4 80.9 12.6132 23.0185 ⫺3.3011 ⫺3.7305 56.0218 16.3984 ⫺0.6572
1-pentyne 52,160. 27,800. 50.1 116.6 96.7 14.7410 28.2140 ⫺5.3433 ⫺3.9453 67.5117 20.7168 ⫺0.7587
1-hexyne 54,420. 22,340. 56.8 137.8 112.5 16.8688 33.4096 ⫺7.3857 ⫺4.1601 79.0012 25.0353 ⫺0.8602
1-heptyne 56,730. 16,980. 63.7 159.0 128.3 18.9956 38.6015 ⫺9.4245 ⫺4.3747 90.4632 29.3537 ⫺0.9647
1-octyne 58,860. 11,330. 70.2 180.2 144.1 21.1244 43.8006 ⫺11.4705 ⫺4.5896 101.9808 33.6721 ⫺1.0631
Benzene 32,000. 12,230. 35.5 86.3 83.5 13.1244 18.1143 11.8225 ⫺3.5277 80.8626 1.7799 ⫺0.1976
Toluene 30,260. 3,280. 43.9 102.8 97.71 15.0259 21.9153 12.1349 ⫺3.6849 93.9240 2.8953 ⫺0.3248
Ethylbenzene 32,440. ⫺2,500. 49.8 120.5 113.8 17.1974 26.2574 12.4885 ⫺3.8644 108.3695 4.0918 ⫺0.4142
n-propylbenzene 34,170. ⫺8,630. 56.1 144.8 129.6 19.3272 30.5152 12.8372 ⫺4.0404 128.4523 5.7345 ⫺0.5096
n-butylbenzene 36,110. ⫺14,430. 62.8 166.0 145.4 21.4550 34.7658 13.1931 ⫺4.2162 145.8054 7.1676 ⫺0.6110
n-pentylbenzene 38,210. ⫺19,750. 69.5 187.2 161.2 23.5828 39.0164 13.5490 ⫺4.3918 163.1584 8.6007 ⫺0.7125
Table 8.22 (continued )
n-hexylbenzene 40,390. ⫺25,590. 76.2 208.4 177. 25.7106 43.2732 13.8894 ⫺4.5678 180.5115 10.0338 ⫺0.8140
n-heptylbenzene 42,640. ⫺31,090. 82.9 229.6 192.8 27.8384 47.5238 14.2453 ⫺4.7435 197.8642 11.4670 ⫺0.9154
n-octylbenzene 44,690. ⫺36,760. 89.6 250.8 208.6 29.9662 51.7744 14.6016 ⫺4.9193 215.2173 12.9001 ⫺1.0169
Methanol ⫺42,050. ⫺58,870. 32.2 37.8 38.17 6.9383 5.5146 11.4018 ⫺3.0069 39.4852 ⫺1.4986 ⫺0.1476
Ethanol ⫺43,330. ⫺68,650. 35.9 62.2 55.08 9.2333 9.9581 12.1445 ⫺3.1906 60.0175 0.1507 ⫺0.2037
1-propanol ⫺41,910. ⫺75,320. 41.4 84.3 70.70 11.3426 14.3252 12.1079 ⫺3.3711 78.3142 1.6447 ⫺0.2869
1-butanol ⫺38,840. ⫺80,320. 46.9 104.4 86.60 13.4902 17.9400 14.1816 ⫺3.5205 94.8858 3.0034 ⫺0.3702
1-pentanol ⫺38,470. ⫺87,730. 53.4 125.2 102.68 15.6573 22.6230 13.6464 ⫺3.7141 111.9216 4.4095 ⫺0.4687
1-hexanol ⫺35,490. ⫺92,690. 64.8 144.4 118.65 17.7843 27.4289 12.5890 ⫺3.9128 126.8936 5.7074 ⫺0.6413
1-heptanol ⫺32,000. ⫺97,270. 72.1 174.2 133.43 19.7695 31.3977 12.9129 ⫺4.0769 151.5815 7.7219 ⫺0.7519
1-octanol ⫺30,250. ⫺103,060. 72.3 195. 149.2 21.9264 35.7093 13.2760 ⫺4.2551 169.4962 9.1280 ⫺0.7549
Phenol ⫺12,585. ⫺36,640. 45.8 75.3 86.17 13.4370 18.7373 11.8793 ⫺3.5535 69.9366 1.0363 ⫺0.3536
Acetone ⫺38,500. ⫺61,720. 44.4 57.7 66.92 10.8100 13.4912 11.4350 ⫺3.3366 54.9496 ⫺0.1534 ⫺0.3324
2-butanone ⫺36,730. ⫺67,880. 50.3 80.4 82.52 12.9145 17.6967 11.7839 ⫺3.5105 73.7083 1.3810 ⫺0.4217
2-pentanone ⫺34,390. ⫺73,460. 56.3 101.6 98.0 15.0021 21.8653 12.1372 ⫺3.6828 91.1588 2.8142 ⫺0.5126
2-hexanone ⫺32,480. ⫺79,220. 63.4 122.8 113.7 17.1141 26.0886 12.4806 ⫺3.8574 108.4560 4.2473 ⫺0.6201
2-heptanone ⫺30,430. ⫺84,890. 70.1 144.0 129.5 19.2419 30.3392 12.8366 ⫺4.0331 125.8091 5.6804 ⫺0.7216
2-octanone ⫺28,380. ⫺90,560. 76.8 165.2 145.3 21.3697 34.5960 13.1769 ⫺4.2091 143.1620 7.1135 ⫺0.8231
Formic acid ⫺88,982. ⫺101,680. 38.9 19.0 34.69 6.3957 4.6630 10.7209 ⫺2.9717 22.1924 ⫺3.1196 ⫺0.3442
Acetic acid ⫺94,760. ⫺116,100. 42.7 40.56 52.01 8.8031 12.4572 3.5477 ⫺3.2939 40.8037 ⫺0.9218 ⫺0.2337
Propanoic acid ⫺93,430. ⫺122,470. 49.4 56.0 67.9 10.9213 13.7115 11.4582 ⫺3.3457 54.313 ⫺0.246 ⫺0.4
Butanoic acid ⫺91,190. ⫺127,950. 56.1 80.5 84.61 13.1708 18.2050 11.8362 ⫺3.5315 72.9853 1.3878 ⫺0.5096
Pentanoic acid ⫺89,210. ⫺133,690. 62.8 103.3 100.5 15.3109 22.4858 12.1803 ⫺3.7085 91.7185 2.9291 ⫺0.6110
Hexanoic acid ⫺87,080. ⫺139,290. 69.5 125.2 116.55 17.4729 26.8090 12.5312 ⫺3.8872 109.6753 4.4095 ⫺0.7125
Heptanoic acid ⫺85,150. ⫺145,080. 76.2 146.4 132.3 19.5938 31.0465 12.8851 ⫺4.0624 127.0284 5.8426 ⫺0.8140
Octanoic acid ⫺83,400. ⫺151,050. 82.9 167.6 148.1 21.7216 35.2971 13.2410 ⫺4.2381 144.3812 7.2758 ⫺0.9154
Formate ⫺83,862. ⫺101,680. 21.7 ⫺22.0 26.16 5.7842 6.3405 3.2606 ⫺3.0410 17.0 ⫺12.4 1.3003
Acetate ⫺88,270. ⫺116,180. 20.6 6.2 40.5 7.7525 8.6996 7.5825 ⫺3.1385 26.3 ⫺3.86 1.3182
continued
Table 8.22 (continued )

Propanoate ⫺86,770. ⫺122,630. 26.5 30.9 54.95 9.6992 15.9017 ⫺0.5003 ⫺3.4363 52.3 ⫺4.2 1.2276
Butanoate ⫺84,610. ⫺127,930. 31.8 44.5 70.3 11.7724 20.9657 ⫺2.4947 ⫺3.6456 42.8123 6.0300 1.1469
Pentanoate ⫺82,600. ⫺134,410. 38.3 78.8 86.31 13.9304 26.2319 ⫺4.5589 ⫺3.8633 62.0175 13.0170 1.0496
Hexanoate ⫺80,450. ⫺139,890. 45.3 100. 102.21 16.0700 31.4589 ⫺6.6186 ⫺4.0794 73.4573 17.3354 0.9427
Heptanoate ⫺78,390 ⫺145,560. 52.0 112.2 118.6 18.2787 36.8480 ⫺8.7285 ⫺4.3022 79.6770 19.8205 0.8417
Octanoate ⫺76,340. ⫺151,230. 58.7 133.4 134.4 20.4065 42.0436 ⫺10.7711 ⫺4.5170 91.1664 24.1390 0.7402
Methanamine 5,040. ⫺16,320. 30.5 37.0 41.88 7.4547 6.7824 10.8884 ⫺3.0593 39.0324 ⫺1.5527 ⫺0.1219
Ethanamine 6,300. ⫺23,830. 33.6 56.8 58.61 9.7281 11.3257 11.2651 ⫺3.2471 55.6799 ⫺0.2142 ⫺0.1688
1-propanamine 7,010. ⫺30,680. 40.8 78. 74.15 11.8178 15.5013 11.6113 ⫺3.4197 72.9633 1.2188 ⫺0.2799
1-butanamine 9,410. ⫺36,110. 47.2 101. 89.80 13.9266 19.7205 11.9476 ⫺3.5941 91.9109 2.7736 ⫺0.3748
1-pentanamine 12,800. ⫺40,650. 53.5 123. 105.7 16.0701 24.0048 12.3010 ⫺3.7713 110.0099 4.2608 ⫺0.4702
1-hexanamine 14,860. ⫺46,320. 60.2 144. 121.6 18.2115 28.2822 12.6611 ⫺3.9481 127.1903 5.6804 ⫺0.5717
1-heptanamine 16,910. ⫺51,990. 66.9 162.8 137.6 20.3667 32.5921 13.0098 ⫺4.1263 142.4729 6.9513 ⫺0.6731
1-octanamine 18,960. ⫺57,660. 73.6 184.0 153.2 22.4671 36.7896 13.3574 ⫺4.2998 159.8259 8.3844 ⫺0.7746
Glycine ⫺88,618. ⫺122,846. 37.84 9.4 43.25 7.6046 7.0825 10.9119 ⫺3.0717 14.1998 ⫺3.4185 ⫺0.2330
Alanine ⫺88,800. ⫺132,130. 40.0 33.8 60.45 9.9472 11.7629 11.3023 ⫺3.2652 34.9465 ⫺1.7690 ⫺0.2658
␣-ABA ⫺87,120. ⫺138,180. 46.7 53.1 76.5 12.1092 16.0862 11.6529 ⫺3.4439 50.6605 ⫺0.4643 ⫺0.3672
Valine ⫺85,300. ⫺147,300. 42.6 72.2 90.79 14.0856 20.0343 11.9842 ⫺3.6071 67.7090 0.8267 ⫺0.3051
Leucine ⫺82,000. ⫺151,070. 51.5 95.1 107.57 16.3362 24.5312 12.3587 ⫺3.7930 86.2214 2.3748 ⫺0.4399
Isoleucine ⫺82,200. ⫺150,900. 52.8 91.6 105.45 16.0395 23.9414 12.3010 ⫺3.7686 83.0208 2.1382 ⫺0.4596
Serine ⫺122,100. ⫺170,800. 46.5 28.1 60.62 9.9372 11.7454 11.2943 ⫺3.2645 29.1225 ⫺2.1543 ⫺0.3642
Threonine ⫺120,000. ⫺179,100. 53.2 50.2 76.86 12.1251 16.1194 11.6532 ⫺3.4453 47.2520 ⫺0.6604 ⫺0.4657
Aspartic acid ⫺172,400. ⫺226,370. 54.8 30.4 71.79 11.4232 14.7124 11.5470 ⫺3.3871 29.9488 ⫺1.9989 ⫺0.4899
Glutamic acid ⫺173,000. ⫺232,000. 70.5 42.3 89.36 13.7471 19.3606 11.9216 ⫺3.5793 38.0235 ⫺1.1944 ⫺0.7276
Asparagine ⫺128,650. ⫺186,660. 55.2 29.9 77.18 12.1587 16.1812 11.6715 ⫺3.4478 29.4616 ⫺2.0327 ⫺0.4959
Glutamine ⫺126,600. ⫺192,330. 61.9 44.7 94.36 14.4753 20.8173 12.0386 ⫺3.6395 41.2938 ⫺1.0322 ⫺0.5974
Phenylalanine ⫺49,500. ⫺110,080. 52.9 91.8 121.92 18.2927 28.4440 12.6752 ⫺3.9548 83.1794 2.1517 ⫺0.4611
Tryptophan ⫺26,900. ⫺97,800. 36.6 100.4 144.0 21.3975 34.6481 13.1912 ⫺4.2113 92.8723 2.7330 ⫺0.2143
Tyrosine ⫺87,300. ⫺157,400. 45.5 71.5 123. 18.4784 28.8127 12.7114 ⫺3.9700 66.7005 0.7794 ⫺0.3490
Methionine ⫺120,200. ⫺177,600. 65.7 70. 105.3 15.9529 23.7688 12.2862 ⫺3.7615 62.5883 0.6780 ⫺0.6550
Table 8.23 Standard partial molal thermodynamic properties of aqueous metal complexes of monovalent organic acid ligands at
25°C and 1 bar and the parameters of the revised HKF model, after Shock and Koretsky (1995). See section 8.11 and tables 8.11,
8.12, and 8.13 for identification of symbols and units. For: formate; Prop: propanoate: But: n-butanoate; Pent: n-pentanoate; Glyc:
glicoate; Lac: lactate; Gly: glycinate; Ala: alanate. The residual charge on the ligand refers to the metal-organic complex, i.e.
Mn(For)+ means Mn(COOH)+ and Mn(For)2 means Mn(COOH)2.
Ligand G ƒ0 , P r , T r H ƒ0 , P r , T r S P0 r , T r C P0 V P0 r , T r a1 ⫻ 10 a2 ⫻ 10⫺2 a3 a4 ⫻ 10⫺4 c1 c2 ⫻ ⫺4 ␻ ⫻ 10⫺5

MANGANESE
(For)⫹ ⫺140,680. ⫺155,740. 7.4 38.4 16.1 4.1094 2.2545 4.8590 ⫺2.8721 32.7312 4.7962 0.4379
(For)2 ⫺226,030. ⫺259,600. 26.8 77.8 53.0 9.0055 14.2060 0.1701 ⫺3.3662 51.4842 12.8126 ⫺0.0300
(Prop)⫹ ⫺143,020. ⫺176,110. 12.2 136.6 44.8 8.0255 11.8128 1.1111 ⫺3.2672 89.6329 24.7960 0.3686
(Prop)2 ⫺230,820. ⫺300,040. 37.9 312.3 113.9 17.3347 34.5476 ⫺7.8345 ⫺4.2071 188.8879 60.5707 ⫺0.0300
(But)⫹ ⫺140,790. ⫺181,340. 17.5 161.9 60.2 10.0986 16.8770 ⫺0.8847 ⫺3.4766 103.6776 29.9377 0.2873
(But)2 ⫺226,377. ⫺310,010 50.9 372.5 146.3 21.7755 45.3895 ⫺12.0924 ⫺4.6553 224.2130 72.8487 ⫺0.0300
(Pent)⫹ ⫺138,600 ⫺187,650. 24.0 225.5 76.2 12.2563 22.1440 ⫺2.9511 ⫺3.6943 140.0855 42.9054 0.1895
(Pent)2 ⫺222,010. ⫺322,030. 65.1 524.5 180.2 26.4073 56.6981 ⫺16.5352 ⫺5.1228 313.3047 103.8147 ⫺0.0300
(Glyc)⫹ ⫺177,830. ⫺208,590. 11.9 91.3 29.7 5.9609 6.7761 3.0811 ⫺3.0590 63.1381 15.5712 0.3735
(Glyc)2 ⫺300,160. ⫺364,740. 37.2 204.1 81.9 12.9661 23.8786 ⫺3.6359 ⫺3.7660 125.5108 38.5424 ⫺0.0300
(Lac)⫹ ⫺178,980. ⫺217,760. 17.6 128.8 46.1 8.1746 12.1772 0.9667 ⫺3.2823 84.2783 23.2081 0.2832
(Lac)2 ⫺302,970. ⫺383,050. 50.3 293.6 116.6 17.7108 35.4655 ⫺8.1945 ⫺4.2450 177.9788 56.7789 ⫺0.0300
(Gly)⫹ ⫺134,770. ⫺165,800. 25.0 67.0 33.7 6.4299 7.9186 2.6372 ⫺3.1063 47.0493 10.6185 0.1739
(Gly)2 ⫺214,100. ⫺278,850. 64.6 146.1 90.2 14.1002 26.6481 ⫺4.7253 ⫺3.8805 91.4846 26.7157 ⫺0.0300
(Ala)⫹ ⫺133,580. ⫺173,180. 29.0 112.3 51.8 8.8873 13.9167 0.2842 ⫺3.3542 73.0237 19.8433 0.1124
(Ala)2 ⫺212,070. ⫺294,250. 71.5 254.2 128.5 19.3395 39.4414 ⫺9.7539 ⫺4.4094 154.8617 48.7440 ⫺0.0300
COBALT
⫹
(For) ⫺99,400. ⫺118,150. ⫺5.4 30.1 7.9 3.0614 ⫺0.3037 5.8638 ⫺2.7663 29.6264 3.1004 0.6305
(For)2 ⫺184,930. ⫺223,370. 10.0 61.6 43.9 7.7655 11.1773 1.3615 ⫺3.2410 41.9649 9.5039 ⫺0.0300
(Prop)⫹ ⫺101,560. ⫺138,350. ⫺0.6 128.3 36.7 6.9775 9.2545 2.1152 ⫺3.1615 86.5252 23.1002 0.5608
(Prop)2 ⫺189,410. ⫺263,490. 21.1 296.0 104.8 16.0946 31.5192 ⫺6.6433 ⫺4.0819 179.3685 57.2620 ⫺0.0300
(But)⫹ ⫺99,980. ⫺144,230. 4.7 153.5 52.1 9.0507 14.3185 0.1200 ⫺3.3708 100.5727 28.2419 0.4799
(But)2 ⫺186,140. ⫺274,660. 33.3 356.3 137.2 20.5354 42.3610 ⫺10.9009 4.5301 214.6936 69.5400 ⫺0.0300
(Pent)⫹ ⫺97,930. ⫺150,670. 11.2 217.2 68.1 11.2020 19.5901 ⫺1.9551 ⫺3.5888 136.9956 41.2096 0.3837
(Pent)2 ⫺182,030. ⫺286,940. 48.3 508.3 171.1 25.1672 53.6696 ⫺15.3437 ⫺4.9976 303.7853 100.5060 ⫺0.0300
(Glyc)⫹ ⫺136,870. ⫺171,330. ⫺0.9 83.0 21.6 4.9137 4.2158 4.0944 ⫺2.9532 60.0521 13.8754 0.5681
(Glyc)2 ⫺260,100. ⫺329,560. 20.4 187.9 72.9 11.7260 20.8501 ⫺2.4444 ⫺3.6408 115.9914 35.2337 ⫺0.0300
continued

(Lac)⫹ ⫺137,390. ⫺179,860. 4.8 120.5 38.0 7.1281 9.6212 1.9722 ⫺3.1766 81.2111 21.5123 0.4799
(Lac)2 ⫺261,460. ⫺346,410. 33.5 277.4 107.5 16.4707 32.4370 ⫺7.0030 ⫺4.1198 168.4594 53.4702 ⫺0.0300
(Gly)⫹ ⫺95,200. ⫺129,080. 15.0 58.7 25.5 5.3683 5.3273 3.6539 ⫺2.9991 43.5722 8.9227 0.3260
(Gly)2 ⫺175,960. ⫺243,430. 55.0 129.8 81.2 12.8601 23.6197 ⫺3.5341 ⫺3.7553 81.9652 23.4070 ⫺0.0300
(Ala)⫹ ⫺94,100. ⫺136,250. 20.0 104.0 43.6 7.8202 11.3148 1.2990 ⫺3.2467 69.3955 18.1475 0.2482
(Ala)2 ⫺173,790. ⫺259,270. 60.0 238.0 119.4 18.0995 36.4127 ⫺8.5626 ⫺4.2842 145.3423 45.4353 ⫺0.0300
NICKEL
⫹
(For) ⫺97,310. ⫺117,570. ⫺10.6 21.3 2.8 2.3868 ⫺1.9558 6.5236 ⫺2.6980 25.2125 1.3130 0.7096
(For)2 ⫺182,850. ⫺223,290. 3.2 44.4 38.2 6.9840 9.2687 2.1133 ⫺3.1621 31.9308 6.0164 ⫺0.0300
(Prop)⫹ ⫺99,640. ⫺137,940. ⫺5.8 119.5 31.6 6.3025 7.6097 2.7546 ⫺3.0935 82.1010 21.3127 0.6388
(Prop)2 ⫺187,630 ⫺263,710. 14.3 278.9 99.1 15.3131 29.6105 ⫺5.8911 ⫺4.0030 169.3347 53.7744 ⫺0.0300
(But)⫹ ⫺97,930. ⫺143,690. ⫺0.5 144.8 46.9 8.3767 12.6711 0.7713 ⫺3.3027 96.1758 26.4544 0.5608
(But)2 ⫺184,120. ⫺274,630. 26.5 339.2 131.5 19.7540 40.4521 ⫺10.1489 4.4512 204.6596 66.0525 ⫺0.0300
(Pent)⫹ ⫺95,870. ⫺150,130. 6.0 208.4 62.9 10.5339 17.9393 ⫺1.3007 ⫺3.5205 132.5696 39.4221 0.4615
(Pent)2 ⫺180,000. ⫺286,890. 41.5 491.2 165.4 24.3858 51.7607 ⫺14.5917 ⫺4.9187 293.7512 97.0185 ⫺0.0300
(Glyc)⫹ ⫺135,150. ⫺171,130. ⫺6.1 74.2 16.5 4.2391 2.5699 4.7380 ⫺2.8851 55.6378 12.0879 0.6472
(Glyc)2 ⫺258,710. ⫺330,150. 13.6 170.8 67.2 10.9446 18.9412 ⫺1.6924 ⫺3.5619 105.9573 31.7462 ⫺0.0300
(Lac)⫹ ⫺135,600. ⫺179,580. ⫺0.4 111.7 32.9 6.4541 7.9800 2.6079 ⫺3.1088 76.8142 19.7248 0.5608
(Lac)2 ⫺259,960. ⫺346,900. 26.8 260.3 101.8 15.6892 30.5284 ⫺6.2512 ⫺4.0409 158.4254 49.9827 ⫺0.0300
(Gly)⫹ ⫺94,540. ⫺129,290. 12.0 49.9 20.4 4.6813 3.6515 4.3097 ⫺2.9299 38.8214 7.1352 0.3686
(Gly)2 ⫺176,530. ⫺246,060. 48.0 112.7 75.4 12.0787 21.7107 ⫺2.7819 ⫺3.6764 71.9311 19.9195 ⫺0.0300
(Ala)⫹ ⫺93,530. ⫺137,130. 15.0 95.2 38.5 7.1451 9.6640 1.9536 ⫺3.1784 64.9703 16.3600 0.3260
(Ala)2 ⫺174,550. ⫺262,970. 50.0 220.8 113.8 17.3180 34.5040 ⫺7.8105 ⫺4.2053 135.3083 41.9478 ⫺0.0300
COPPER
⫹
(For) ⫺70,890. ⫺88,300. ⫺0.2 34.8 7.7 3.0050 ⫺0.4430 5.9211 ⫺2.7606 31.6869 4.0629 0.5536
(For)2 ⫺156,550. ⫺193,180. 16.8 70.8 43.7 7.7315 11.0979 1.3851 ⫺3.2377 47.3676 11.3819 ⫺0.0300
(Prop)⫹ ⫺74,120. ⫺109,578. 4.6 133.0 36.5 6.9197 9.1126 2.1719 ⫺3.1556 88.5486 24.0627 0.4799
(Prop)2 ⫺163,000. ⫺235,270. 27.9 305.2 104.5 16.0606 31.4336 ⫺6.6036 ⫺4.0784 184.7714 59.1399 ⫺0.0300
(But)⫹ ⫺71,850. ⫺114,770. 10.0 158.3 51.8 8.9947 14.1844 0.1682 ⫺3.3653 102.6486 29.2044 0.4046
(But)2 ⫺158,470. ⫺245,180. 40.1 365.5 137.0 20.5014 42.2754 ⫺10.8613 ⫺4.5266 220.0963 71.4180 ⫺0.0300
(Pent)⫹ ⫺69,820. ⫺121,220. 16.4 221.9 67.8 11.1516 19.4472 ⫺1.8933 ⫺3.5828 139.0314 42.1721 0.3041
(Pent)2 ⫺154,380. ⫺257,470. 55.1 517.5 170.8 25.1332 53.5902 ⫺15.3200 ⫺4.9943 309.1882 102.3839 ⫺0.0300

(Glyc)⫹ ⫺109,440. ⫺142,560. 4.3 87.7 21.4 4.8555 4.0750 4.1470 ⫺2.9474 62.0660 14.8378 0.4862
(Glyc)2 ⫺233,020. ⫺300,660. 27.2 197.1 72.6 11.6921 20.7705 ⫺2.4205 ⫺3.6376 121.3943 37.1116 ⫺0.0300
(Lac)⫹ ⫺110,350. ⫺151,480. 10.0 125.2 37.8 7.0703 9.4856 2.0147 ⫺3.1710 83.2379 22.4748 0.3993
(Lac)2 ⫺235,050. ⫺318,180. 40.3 286.6 107.3 16.4367 32.3514 ⫺6.9634 ⫺4.1163 173.8623 55.3481 ⫺0.0300
(Gly)⫹ ⫺71,300. ⫺102,410. 25.0 63.4 25.3 5.2864 5.1238 3.7413 ⫺2.9907 44.9395 9.8852 0.1739
(Gly)2 ⫺156,480. ⫺221,770. 63.0 139.0 80.9 12.8262 23.5338 ⫺3.4944 ⫺3.7518 87.3679 25.2850 ⫺0.0300
(Ala)⫹ ⫺70,600. ⫺109,970. 30.0 108.7 43.4 7.7387 11.1165 1.3762 ⫺3.2385 70.7759 19.1100 0.0974
(Ala)2 ⫺154,650. ⫺237,360. 70.0 247.2 119.2 18.0655 36.3271 ⫺8.5230 ⫺4.2807 150.7452 47.3132 ⫺0.0300
ZINC
(For)⫹ ⫺121,480. ⫺140,700. ⫺4.3 33.5 8.0 3.0712 ⫺0.2795 5.8532 ⫺2.7673 31.4547 3.7879 0.6143
(For)2 ⫺206,910. ⫺245,730. 11.4 68.1 44.1 7.7824 11.2204 1.3417 ⫺3.2428 45.8240 10.8453 ⫺0.0300
(Prop)⫹ ⫺123,680. ⫺160,939. 0.5 131.7 36.8 6.9879 9.2834 2.0962 ⫺3.1627 88.3708 23.7877 0.5464
(Prop)2 ⫺211,460. ⫺285,920. 22.5 302.6 104.9 16.1116 31.5558 ⫺6.6473 ⫺4.0834 183.2279 58.6033 ⫺0.0300
(But)⫹ ⫺121,820. ⫺166,540. 5.8 156.9 52.2 9.0597 14.3386 0.1167 ⫺3.3717 102.3810 28.9294 0.4615
(But)2 ⫺207,670. ⫺296,560. 34.7 362.9 137.4 20.5524 42.4038 ⫺10.9205 ⫺4.5319 218.5528 70.8814 ⫺0.0300
(Pent)⫹ ⫺119,680 ⫺172,900. 12.3 220.6 68.2 11.2174 19.6055 ⫺1.9494 ⫺3.5894 138.7888 41.8971 0.3636
(Pent)2 ⫺203,420. ⫺308,690. 49.8 514.9 171.2 25.1842 53.7125 ⫺15.3636 ⫺4.9994 307.6444 101.8474 ⫺0.0300
(Glyc)⫹ ⫺159,620. ⫺194,550. 0.2 86.4 21.7 4.9216 4.2387 4.0776 ⫺2.9541 61.8303 14.5628 0.5464
(Glyc)2 ⫺283,310. ⫺353,140. 21.8 194.5 73.0 11.7430 20.8930 ⫺2.4643 ⫺3.6426 119.8505 36.5751 ⫺0.0300
(Lac)⫹ ⫺160,730. ⫺200,060. 18.0 123.9 38.1 7.0755 9.4969 2.0126 ⫺3.1715 81.3400 22.1998 0.2792
(Lac)2 ⫺285,380. ⫺364,730. 55.0 284.0 107.7 16.4877 32.4799 ⫺7.0229 ⫺4.1216 172.3186 54.8116 ⫺0.0300
(Gly)⫹ ⫺117,820. ⫺151,610. 18.0 62.1 25.6 5.3677 5.3226 3.6628 ⫺2.9989 45.1185 9.6102 0.2792
(Gly)2 ⫺199,140. ⫺267,410. 55.0 136.4 81.3 12.8771 23.6625 ⫺3.5537 ⫺3.7571 85.8243 24.7484 ⫺0.0300
(Ala)⫹ ⫺116,610. ⫺160,450. 17.0 107.4 43.8 7.8514 11.3891 1.2746 ⫺3.2497 71.8106 18.8350 0.2956
(Ala)2 ⫺197,110. ⫺283,390. 60.0 244.5 119.6 18.1164 36.4558 ⫺8.5823 ⫺4.2860 149.2015 46.7767 ⫺0.0300
CADMIUM
⫹
(For) ⫺104,930. ⫺121,320. 7.7 40.0 17.7 4.3381 2.8097 4.6479 ⫺2.8951 33.5224 5.0712 0.4379
(For)2 ⫺191,520. ⫺226,400. 27.1 80.4 54.9 9.2604 14.8299 ⫺0.0792 ⫺3.3920 53.0278 13.3492 ⫺0.0300
(Prop)⫹ ⫺107,910. ⫺142,340. 12.5 138.0 46.5 8.2526 12.3660 0.8957 ⫺3.2901 90.3786 25.0710 0.3636
continued
Ligand G ƒ0 , P r , T r H ƒ0 , P r , T r S P0 r , T r C P0 r , T r V P0 r , T r a1 ⫻ 10 a2 ⫻ 10⫺2 a3 a4 ⫻ 10⫺4 c1 c2 ⫻ ⫺4 ␻ ⫻ 10⫺5

(Prop)2 ⫺196,520. ⫺267,040. 38.2 314.9 115.7 17.5895 35.1655 ⫺8.0682 ⫺4.2326 190.4317 61.1072 ⫺0.0300
(But)⫹ ⫺105,290 ⫺147,170. 17.8 163.2 61.9 10.3259 17.4297 ⫺1.0970 ⫺3.4994 104.4311 30.2127 0.2832
(But)2 ⫺191,460. ⫺276,420. 50.4 375.2 148.2 22.0303 46.0136 ⫺12.3417 ⫺4.6811 225.7566 73.3853 ⫺0.0300
(Pent)⫹ ⫺103,380. ⫺153,760. 24.3 226.9 77.9 12.4829 22.6985 ⫺3.1719 ⫺3.7173 140.8184 43.1804 0.1832
(Pent)2 ⫺187,390. ⫺288,730. 65.5 527.2 182 26.6621 57.3222 ⫺16.7845 ⫺5.1486 314.8482 104.3513 ⫺0.0300
(Glyc)⫹ ⫺142,280. ⫺174,380. 12.2 92.7 31.4 6.1880 7.3293 2.8656 ⫺3.0819 63.8836 15.8461 0.3686
(Glyc)2 ⫺265,389. ⫺331,280. 37.5 206.7 83.8 13.2210 24.5024 ⫺3.8851 ⫺3.7918 127.0546 39.0789 ⫺0.0300
(Lac)⫹ ⫺143,410. ⫺183,520. 17.9 130.2 47.8 8.4019 12.7361 0.7390 ⫺3.3054 85.0322 23.4831 0.2792
(Lac)2 ⫺267,700. ⫺349,090. 50.7 296.3 118.5 17.9656 36.0834 ⫺8.4279 ⫺4.2706 179.5224 57.3155 ⫺0.0300
(Gly)⫹ ⫺100,240. ⫺132,090 27.0 68.4 35.3 6.6477 8.4525 2.4231 ⫺3.1283 47.5446 10.8935 0.1417
(Gly)2 ⫺180,580. ⫺246,610. 65.0 148.7 92.1 14.3550 27.2722 ⫺4.9746 ⫺3.9063 93.0282 27.2523 ⫺0.0300
(Ala)⫹ ⫺100,080. ⫺141,020. 29.2 113.7 53.4 9.1151 14.4742 0.0630 ⫺3.3773 73.7914 20.1183 0.1098
(Ala)2 ⫺179,940. ⫺263,420. 71.9 256.8 130.4 19.5944 40.0653 ⫺10,0034 ⫺4.4352 156.4053 49.2806 ⫺0.0300
LEAD
(For)⫹ ⫺92,150. ⫺100,690. 37.1 19.1 17.7 4.1867 2.4389 4.7962 ⫺2.8797 17.2647 0.8546 ⫺0.0100
(For)2 ⫺177,720. ⫺202,040. 65.7 40.0 54.9 9.2604 14.8299 ⫺0.0792 ⫺3.3920 29.3579 5.1222 ⫺0.0300
(Prop)⫹ ⫺95,670. ⫺122,250. 41.9 117.3 46.5 8.1013 12.0011 1.0303 ⫺3.2750 74.1249 20.8544 ⫺0.0838
(Prop)2 ⫺184,200. ⫺244,160. 76.8 274.5 115.7 17.5895 35.1655 ⫺8.0682 ⫺4.2326 166.7618 52.8802 ⫺0.0300
(But)⫹ ⫺92,820. ⫺126,860 47.2 142.5 61.9 10.1748 17.0643 ⫺0.9612 ⫺3.4843 88.1802 25.9961 ⫺0.1639
(But)2 ⫺179,080. ⫺253,470. 89 334.8 148.2 22.0303 46.0136 ⫺12.3417 ⫺4.6811 202.0867 65.1583 ⫺0.0300
(Pent)⫹ ⫺90,790. ⫺133,320. 53.7 206.9 77.9 12.3324 22.3315 ⫺3.0281 ⫺3.7021 124.5859 38.9638 ⫺0.2620
(Pent)2 ⫺175,020. ⫺265,790. 104 486.8 182 26.6621 57.3222 ⫺16.7845 ⫺5.148 291.1785 96.1242 ⫺0.0300
(Glyc)⫹ ⫺130,020. ⫺154,270. 41.6 72.0 47.8 8.2809 12.4416 0.8529 ⫺3.2932 47.6306 11.6296 ⫺0.0788
(Glyc)2 ⫺253,620. ⫺308,950. 76.1 166.4 118.5 17.9656 36.0834 ⫺8.4279 ⫺4.2706 103.3846 30.8519 ⫺0.0300
(Lac)⫹ ⫺131,350. ⫺163,610. 47.3 109.5 47.8 8.2516 12.3686 0.8843 ⫺3.2902 68.8031 19.2665 ⫺0.1656
(Lac)2 ⫺256,300. ⫺327,120. 89.2 255.9 118.5 17.9656 36.0834 ⫺8.4279 ⫺4.2706 155.8527 49.0884 ⫺0.0300
(Gly)⫹ ⫺88,450. ⫺112,310. 56.9 47.7 35.3 6.4949 8.0790 2.5712 ⫺3.1129 31.2483 6.6769 ⫺0.3103
(Gly)2 ⫺168,350. ⫺222,990. 106.3 108.3 92.1 14.3550 27.2722 ⫺4.9746 ⫺3.9063 69.3582 19.0253 ⫺0.0300
(Ala)⫹ ⫺87,190. ⫺120,280. 58.7 93.0 53.4 8.9640 14.1088 0.1992 ⫺3.3622 57.5413 15.9017 ⫺0.3372
(Ala)2 ⫺166,270. ⫺239,190. 110.4 216.4 130.4 19.5944 40.0653 ⫺10.0034 ⫺4.4352 132.7356 41.0535 ⫺0.0300
Discrete values of the HKF model parameters for various organic aqueous
species are listed in table 8.22. Table 8.23 lists standard partial molal thermody-
namic properties and HKF model parameters for aqueous metal complexes of
monovalent organic acid ligands, after Shock and Koretsky (1995).
8.18 Redox State and Biologic Activity
We have already mentioned that photosynthesis and other biochemical processes

are the main causes of disequilibrium in aqueous solutions. The conversion of
luminous energy into chemical energy (formation of stable covalent bonds) in-
volves local lowering of the redox state. For instance, the conversion of carbon
dioxide into glucose:
1 1 1
CO 2(gas) + H + + e − ⇔ C 6 H12 O 6 + H 2 O (8.224)
4 24 4
is a reducing process whose standard potential is E 0224 ⫽ ⫺0.426 V. This low Eh

value is normally attained in photosynthetic processes. However nonphotosyn-
thesizing organisms tend to reestablish equilibrium, with catalytic decomposition
of photosynthesis products. The energy gained in the redox process is partly ex-
ploited in the synthesis of new cells. Figure 8.29 represents the main redox pro-
cesses taking place through mediation of microbes in electron transfer from re-
duced to oxidized phase. The sequence of reactions (top to bottom) is that which
may be expected on thermochemical bases, according to the Gibbs free energy
changes involved in the various processes (table 8.24).
Table 8.24 lists the main redox exchanges mediated by biological activity and
their equilibrium constants at pH ⫽ 7.
In a closed aqueous system containing organic matter (exemplified by formal-
dehyde CH2O), oxidation of organic matter initially implies reduction of free oxy-
gen, followed by reduction of NO⫺3 . These processes involve a progressive de-
crease in Eh. In the presence of manganese, contemporaneously with the
reduction of nitrates there is also reduction of MnO2 . At lower Eh, reduction of
iron complexes (FeOOH(crystal) to ferrous ion Fe2⫹ ) takes place. If Eh decreases
sufficiently, fermentation reactions take place, accompanied by reduction of sul-
fates and carbonates. The sequence of reactions in table 8.24 reflects the vertical
distribution of components in an aqueous system rich in nutrients (eutrophic)
and can be compared with an “ecological sequence” of microorganisms that act
as mediators of redox processes (aerobic–heterotropic–denitrificant–fermentant–
sulfate reducer–methaniferous bacterials; cf. Stumm and Morgan, 1981).
Let us now briefly review the effects of biologic activity on the various
systems.
Figure 8.29 Sequence of redox equilibria mediated by biologic activity. From W. Stumm
and J. J. Morgan (1981), Aquatic Chemistry, copyright 䉷 1981 by John Wiley and Sons.
Reprinted by permission of John Wiley & Sons. The various equilibrium constants are
8.18.1 Effects on Carbon

It is estimated that photosynthesis annually fixes about 54 ⫻ 1014 moles of C.
About 99.95% of the fixed carbon is produced by living organisms and only 0.05%
(0.025 ⫻ 1014 moles) is buried in sedimentary processes in the form of native
carbon. The organic complexes of carbon present in aqueous solutions are ther-
modynamically unstable (with the exception of methane CH4 ). The decomposi-
tion of acetic acid CH3COOH and formaldehyde CH2O into graphite, methane,
and carbon dioxide are spontaneous thermochemical processes, although their
reaction kinetics are rather sluggish, leading to metastable persistence of re-
actants (Stumm and Morgan, 1981):
Table 8.24 Redox equilibria mediated by biological activity in eutrophic systems (from
Stumm and Morgan, 1981). pH ⫽ 7; [HCO3⫺] ⫽ 10⫺3; (g) ⫽ gaseous, (s) ⫽ solid.
Reduction log K
(A) 1/4 O2(g) ⫹ H⫹ ⫹ e⫺ ⇔ 1/2 H2O ⫹13.75
(B) 1/5 NO3⫺ ⫹ 6/5 H⫹ ⫹ e⫺ ⇔ 1/10 N2(g) ⫹ 3/5 H2O ⫹12.65
(C) 1/2 MnO2(s) ⫹ 1/2 HCO3⫺ ⫹ 3/2 H⫹ ⫹ e⫺ ⇔ 1/2 MnCO3(s) ⫹ H2O ⫹8.9
(D) 1/8 NO3⫺ ⫹ 5/4 H⫹ ⫹ e⫺ ⇔ 1/8 NH4⫹ ⫹ 3/8 H2O ⫹6.15
(E) FeOOH(s) ⫹ HCO3⫺ ⫹ 2H⫹ ⫹ e⫺ ⇔ FeCO3(s) ⫹ 2 H2O ⫺0.8
(F) 1/2 CH2O ⫹ H⫹ ⫹ e⫺ ⇔ 1/2 CH3OH ⫺3.01
(G) 1/8 SO42⫺ ⫹ 9/8 H⫹ ⫹ e⫺ ⇔ 1/8 HS⫺ ⫹ 1/2 H2O ⫺3.75
(H) 1/8 CO2(g) ⫹ H⫹ ⫹ e⫺ ⇔ 1/8 CH4(g) ⫹ 1/4 H2O ⫺4.13
(J) 1/6 N2(g) ⫹ 4/3 H⫹ ⫹ e⫺ ⇔ 1/3 NH4 ⫺4.68
Oxidation log K
⫹ ⫺
(L) 1/4 CH2O ⫹ 1/4 H2O ⇔ 1/4 CO2(g) ⫹ H ⫹ e ⫹8.20
(L-1) 1/2 HCOO⫺ ⇔ 1/2 CO2(g) ⫹ 1/2 H⫹ ⫹ e⫺ ⫹8.33
(L-2) 1/2CH2O ⫹ 1/2H2O ⇔ 1/2 HCOO⫺ ⫹ 3/2H⫹ ⫹ e⫺ ⫹7.68
(L-3) 1/2 CH3OH ⇔ 1/2 CH2O ⫹ H⫹ ⫹ e⫺ ⫹3.01
(L-4) 1/2 CH4(g) ⫹ 1/2 H2O ⇔ 1/2 CH3OH ⫹ H⫹ ⫹ e⫺ ⫺2.88
(M) 1/8 HS⫺ ⫹ 1/2H2O ⇔ 1/8 SO42⫺ ⫹ 9/8 H⫹ ⫹ e⫺ ⫹3.75
(N) FeCO3(s) ⫹ 2 H2O ⇔ FeOOH(s) ⫹ HCO3⫺ ⫹ 2H⫹ ⫹ e⫺ ⫹0.8
(O) 1/8 NH4+ ⫹ 3/8 H2O ⇔ 1/8 NO3⫺ ⫹ 5/4 H⫹ ⫹ e⫺ ⫺6.15
(P) 1/2 MnCO3(s) ⫹ H2O ⇔ 1/2 MnO2(s) ⫹ 1/2 HCO3⫺ ⫹ 3/2 H⫹ ⫹ e⫺ ⫺8.9
 kJ 
CH 3 COOH ⇔ 2 H 2 O + 2 C (solid)  ∆G = −103  (8.225)
 mole 
 kJ 
CH 3 COOH ⇔ CH 4(gas) + CO 2 (gas)  ∆G = −51  (8.226)
 mole 
 kJ 
CH 2 O ⇔ C (solid) + H 2 O  ∆G = −107 . (8.227)
 mole 
A process of bacterial degradation of carbon compounds can be conceived

essentially as “a conversion from a CO2-predominant to a CH4-predominant sys-
tem” (Stumm and Morgan, 1981). The production of organic methane by micro-
bial activity is a typical example of such conversion. Organic matter is first trans-
formed into organic acids, which are then decomposed into acetic acid, gaseous
hydrogen, and CO2 , and finally recombined as CH4 :
4 H 2(gas) + CO 2(gas) → CH 4 + 2 H 2 O
Complex organic matter → organic acids
CH 3 COOH → CH 4 + CO 2(gas)
( 8.228 )
8.18.2 Effects on Nitrogen

We have already seen that reduction of gaseous nitrogen N2(g) into ammonium
ion NH⫹ 4 takes place at low Eh but still within the stability field of water. Blue-
green algae are the organisms that most efficaciously fix free nitrogen into ammo-
nium ion complexes, operating under the redox conditions established by photo-
synthesis (pH being equal, carbonate reduction takes place at lower Eh; cf. figure
8.21A and B). The fact that N2 fixation is not a process as extensive as the reduc-
tion of atmospheric CO2 (and of soluble complexes H2CO3 , HCO⫺ 2⫺
3 , and CO3 )
is attributable to kinetic problems connected with the bond stability of the N2
molecule (Stumm and Morgan, 1981). NH⫹ ⫺
4 and NO3 are also stable covalent
complexes that may persist metastably under Eh-pH conditions pertinent to the
N2 molecule (see figure 8.21B).
8.18.3 Effects on Sulfur

The sulfate → sulfide reduction requires quite low Eh conditions (figure 8.21D).
The process takes place through enzymatic mediation (enzymes oxidize organic
matter at the Eh of interest). At low pH, the reducing process may result in the
formation of native sulfur, as an intermediate step of the process
 kJ 
SO24 − + 2 H+ + 3 H 2 S ⇔ 4 S(crystal) + 4 H2O  ∆G = −27.7 . (8.229)
 mole 
As already mentioned, the Eh-pH extension of the field of native sulfur depends
on the total sulfur concentration in the system. With solute molality lower than
approximately 10⫺6 moles/kg, the field of native sulfur disappears.
8.19 Degree of Saturation of Condensed Phases and

Affinity to Equilibrium in Heterogeneous Systems
Let us consider a generic hydrolytic equilibrium between a crystalline substance—

such as K-feldspar, for instance—and an aqueous solution:
KAlSi3O8 + 4H+ + 4H2O ⇔ K+ + Al3 + + 3H4SiO04. (8.230)
We can calculate the equilibrium constant of reaction 8.230 under any given P-T
condition by computing the standard molal Gibbs free energy values of the vari-
ous components at the P-T of interest—i.e.,
 − ∆G230
0 
K 230,P ,T = exp  . (8.231)
 RT 
Applying the mass action law and assuming unitary activity for the water solvent
and the condensed phase, we obtain
K 230,P ,T = aK + ⋅ a Al 3 + ⋅ a 3 ⋅ a − 4+ . ( 8.232 )
H 4 SiO 04 H
The activities of solute species in an aqueous solution in equilibrium with K-

feldspar at the P and T of interest will be those dictated by equation 8.232. Let
us now imagine altering the chemistry of the aqueous solution in such a way that
the activities of the aqueous species of interest differ from equilibrium activities.
New activity product Q:
( ) ( )
3 −4
Q = a K′ + ⋅ a Al
′ 3+ ⋅ a H′ 0 ⋅ a H′ + (8.233)
4SiO 4
will differ from equilibrium constant K, and the Q/K ratio will define the distance
from equilibrium of the new fluid with respect to K-feldspar at the P and T of
interest. In a generalized form, we can write
Q= ∏ ai′ v i
,
(8.234)
i
where i is a generic solute species and vi is its corresponding stoichiometric co-

efficient in reaction (positive for products and negative for reactants).
It follows from equation 8.234 that
Q  
log   = log  ∏ ai′ vi  − log K . (8.235)
K  i 
Ratio Q/K is adimensional: if it is higher than 1 (i.e., log Q/K ⬎ 0), the aqueous
solution is oversaturated with respect to the solid phase; if it is zero, we have
perfect equilibrium; and if it is lower than 1 (i.e., log Q/K ⬍ 0), the aqueous phase
is undersaturated with respect to the solid phase.
Recalling now the concept of thermodynamic affinity (section 2.12), we can,
in a similar way, define an “affinity to equilibrium” as
Q
A = −2.303 RT log  . (8.236)
K
This parameter has the magnitude of an energy (i.e., kJ/mole, kcal/mole) and
represents the energy driving toward equilibrium in a chemical potential field—
i.e., the higher A becomes, the more the solid phase of interest is unstable in
solution. In a heterogeneous system at equilibrium, based on the principle of
equality of chemical potentials, all phases in the system are in mutual equilibrium.
It follows that equation 8.236 is of general applicability to all condensed phases,
and the value of affinity must be zero for all solids in equilibrium with the fluid.
It is immediately evident that equations 8.235 and 8.236 furnish a precise measure
of geochemical control of the degree of equilibrium attained by the fluid with
respect to a given mineralogical paragenesis. However, because the equilibrium
constant is a function of intensive variables T and P, plotting the affinity value
for various minerals under different T (or P) conditions against the T (or P) of
calculation provides precise thermometric (or barometric) information concern-
ing the equilibrium state of the fluid with respect to the reservoir. Figure 8.30A
shows a log Q/K ⫽ f (T ) diagram in the case of complete equilibrium between
mineralogical paragenesis and fluid at a given T condition (in this case, T ⫽
250 °C): note how the saturation curves for all solid phases converge to log
Q/K ⫽ 0 at the temperature of equilibrium.
Detailed speciation calculations for the fluid coupled with mineralogical in-
vestigation of the solid paragenesis generally allow a sufficiently precise estimate
of the T of equilibrium. As figure 8.30A shows, the chemistry of the fluid is
buffered by wall-rock minerals, so that the saturation curves of other phases not
pertaining to the system of interest are scattered (figure 8.28B).
The interpretation of log Q/K ⫽ f (T ) plots may be complicated by additional
processes affecting the chemistry of the fluid. The main complications result from
two processes that take place during the ascent of the fluid toward the surface:
adiabatic flashing and dilution by external aquifers.
Figure 8.30C shows the boiling effect on the fluid in equilibrium with the
paragenesis of figure 8.30A, calculated by Reed and Spycher (1984) assuming
subtraction of a gaseous phase amounting to 0.2% by weight of the initial fluid
and composed of 99.525% H2O, 0.451% CO2 , 0.013% H2S, and 0.011% H2 : the
increased molal concentration of dissolved silica displaces the intersection of the
saturation curve of quartz with the equilibrium condition log Q/K ⫽ 0 to a higher
T. Moreover, the change in pH, altering the speciation state of aluminum, iron
sulfides, and carbonates, modifies the shapes of the saturation curves of alkali
feldspars, muscovite, pyrite, and calcite.
The effect of dilution by shallow aquifers can also be quantified. It invariably
results in scattered saturation curves, proportional to the extent of contamination.
To the above considerations must be added the fact that equations of type
8.235 apply to pure compounds, whereas natural solids are mixtures of several
components. It follows that the activity of the solid component in hydrolysis is
never 1. Equation 8.235 must therefore be modified to take account of this fact:
Q  
log   = log  ∏ ai′ vi  − log K − log a K . (8.237)
K  i 
Figure 8.30 Degree of saturation of various phases as a function of T in a heterogeneous

system. A ⫽ complete equilibrium; B ⫽ disequilibrium; C ⫽ dispersion of saturation
curves as a result of boiling. Reprinted from M. Reed and N. Spycher, Geochimica et
Cosmochimica Acta, 48, 1479–1492, copyright 䉷 1984, with kind permission from Elsevier
Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
aK in eq. 8.237 is the activity of the solid component in the mixture for which
hydrolytic equilibrium has been calculated. It must be emphasized that the effect
of dilution in a solid mixture is often quite limited: a change from activity ⫽ 1.0
to activity ⫽ 0.5 involves a displacement of the log Q/K factor of 0.3 (T, P, and
chemistry of the fluid being equal).
8.20 Activity Diagrams
In the investigation of mineral-fluid equilibria, of fundamental importance are

the “activity diagrams,” depicting equilibria at various P and T of interest in terms
of activity ratios of solutes or fugacity of gaseous components. Let us consider
the system MgO-CaO-FeO-Na2O-K2O-Al2O3-SiO2-CO2-H2O, as representative
of the chemistry of most rock-forming minerals. Of the various oxides in the
system, only H2O, CO2 , and SiO2 do not dissociate appreciably in aqueous solu-
tion. For the remaining oxides, dissociation is almost complete and can be repre-
sented as follows:
Na2O+ 2 H+ ⇔ 2 Na+ + H2O (8.238)
CaO+ 2 H+ ⇔ Ca2 + + H2O (8.239)
Al2O3 + 6 H+ ⇔ 2 Al3 + + 3 H2O . (8.240)
The dissociation of a generic oxide Mv⫹Ov⫺ (where v+ and v⫺ are, respectively, the
cationic and anionic stoichiometric factors in the formula unit) can be expressed
by the general equilibrium
( )( )
 v+ Z +
Mv +Ov − + v + Z + H+ ⇔ 
( ) ( )  H O + (v + ) M Z+
,
2 (8.241)
 2 
where Z⫹ is the cationic charge.

Applying the mass action law to equilibrium 8.238, we can write
a2 ⋅ a H 2O
Na +
K 244 = . (8.242)
a Na 2O ⋅ a 2 +
H
The generalized form of equation 8.242 is given by
a v +Z + ⋅ a
[ (v + )(Z + ) 2 ]
M H2 O
K = . (8.243)
(v + )(Z + )
aMv + Ov − ⋅ a
H+
Let us now rewrite equation 8.242 in logarithmic notation, assigning unitary ac-
tivity to the solvent:
a + 
ln a Na2 O = 2 ln  Na  − ln K 238 . (8.244)
 aH + 
For the generic oxide Mv⫹Ov⫺ , we have
a 
( ) Z+
ln aMv + Ov − = v + ln  M  − ln K .
 a (Z + ) 
(8.245)
 H+ 
The relationship between thermodynamic activity and chemical potential is
µ Na2 O = µ Na
0
+ RT ln a Na2 O . ( 8.246 )
2O
For two generic oxides i and j, the partial derivatives of equation 8.246 can be
combined (Helgeson, 1967) to give
  a Z+  
 d ln  M( i )  
v(+i )
 Z +( i )
 aH 
dµ i d ln ai   + 
= = +  . (8.247)
dµ j d ln a j v( j )   a M Z +  
 d ln Z + ( j )  
( j)
  aH+  
  
Equation 8.247 can be rearranged in the form
 aM Z + 
d ln  Z + ( i ) 
(i )
 aH+  v(+j ) v(+j ) df ( i )

  dµ i
= ⋅ = + ⋅ , (8.248)
 aM Z +  v(+i ) dµ j v( i ) df ( j )
d ln  Z + ( j ) 
( j)
 aH+ 
 
where the last term on the right is the fugacity ratio of gaseous components.
Equation 8.248 is the basis for the construction of activity diagrams.
Let us now imagine a process of hydrothermal alteration of arkose sandstones
composed of Mg-chlorite, K-feldspar, K-mica, and quartz. Because precipitating
SiO2 during alteration is amorphous, we will assume the presence of amorphous
silica instead of quartz, and we will consider MgO as the generic oxide i and K2O
as the generic oxide j.
The alteration of muscovite to kaolinite is described by the reaction
( )
2 KAl 3 Si3O10 OH + 2 H+ + 3 H2O ⇔ 3 Al 2 Si2O5 OH + 2K+.
2
( ) 4 (8.249)
muscovite kaolinite
Assigning unitary activity to the condensed phases and to solvent H2O, equilib-
rium constant K249 takes the form
a2 +
K 249 = K
, (8.250)
a2 +
H
and, in logarithmic notation,
1
log K 249 = log aK + − log aH + = log aK + + pH. (8.251)
2
The equilibrium between K-feldspar and muscovite is expressed as
( )
3 KAlSi3O8 + 2 H+ ⇔ KAl3Si3O10 OH + 2K+ + 6 SiO2 ,
2 (8.252)
K- feldspar muscovite
which yields
1
log K 252 = log aK + − log aH + = log aK + + pH. (8.253)
2
The equilibrium between Mg-chlorite and kaolinite is given by
Mg 5 Al 2 Si 3 O 10 OH( ) 8
+ 10 H + ⇔ Al 2 Si 2 O 5 OH( ) 4
+ 5 Mg 2 + + SiO 2 + 7 H 2 O,
Mg- chlorite kaolinite
( 8.254 )
1
log K 254 = log a Mg 2 + + 2 pH . (8.255)
5
The equilibrium between Mg-chlorite and muscovite may be depicted as
( )
3Mg5 Al 2 Si 3O10 OH + 2K + + 28H+ ⇔ 2KAl 3Si 3O10 OH
8
( ) 2
+ 15Mg2 + + 3SiO2 + 24H2O.
Mg- chlorite muscovite
(8.256)
The equilibrium constant of reaction 8.256 is
a 15
Mg 2 +
K 256 = . (8.257)
a2+ ⋅ a 28+
K H
Multiplying and dividing by a2H⫹ and transforming into logarithmic notation,

we obtain
a +   aMg 2 + 
2 log  K  + log K 256 = 15 log  2  . (8.258)
 aH +   a
 H+ 
Lastly, the equilibrium between Mg-chlorite and K-feldspar is depicted by
( )
Mg5 Al2 Si3O10 OH + 2 K+ + 8 H+ + 3 SiO2 ⇔ 2 KAlSi3O8 + 5 Mg2 + + 8 H2O .
8 (8.259)
Mg- chlorite K- feldspar
The equilibrium constant is
a5
Mg 2 +
K 259 = . (8.260)
a2+ ⋅ a8 +
K H
Again, multiplying and dividing by a2H⫹ and transforming into logarithmic nota-
tion, we obtain
a +   aMg 2 + 
2 log  K  + log K 259 = 5 log  2  . (8.261)
 aH +   a
 H+ 
We can now build up the activity diagram, adopting as Cartesian axes log
aK⫹ ⫹ pH (abscissa) and log aMg2⫹ ⫹ 2pH (ordinate; see figure 8.31). The kaolin-
ite–muscovite equilibrium in such a diagram is defined by a straight line parallel
to the ordinate, with an intercept at 12 log K249 on the abscissa. Analogously, the
equilibrium between muscovite and K-feldspar is along a straight line, parallel to
the preceding one and with an intercept at 12 log K252 ; the kaolinite–Mg-chlorite
equilibrium lies along a straight line parallel to the abscissa, with intercept at
1
5 log K254 .
Let us now consider Mg-chlorite–muscovite and Mg-chlorite–K-feldspar

equilibria. Because they depend on the activity of both Mg2⫹ and K⫹ in solution,
besides H⫹, the slope of the univariant curve is dictated by the stoichiometric
2
coefficients of ions in reaction: for reaction 8.256, the slope is 15 and the intercept
Figure 8.31 Example of activity diagram.
1
on the ordinate axis is 15 log K256 (cf. eq. 8.258). Analogously, for the Mg-
chlorite–K-feldspar equilibrium, the univariant curve has slope 25 and intercept
at 15 log K259 (cf. eq. 8.261). As shown in figure 8.31, the intersections of univariant
equilibria contour the stability fields of the various solids in reaction. A fluid of
composition A in figure 8.31 is in equilibrium with kaolinite; fluid B, falling on
the kaolinite–muscovite boundary at T ⫽ 350 °C, is clearly buffered by the alter-
ation of mica; and a fluid of composition C is in invariant equilibrium with the
original paragenesis. Translating the above observations into log Q/K ⫽ f (T )
notation (section 8.19), for a fluid of composition C we observe convergence of
the saturation curves of muscovite, K-feldspar, and Mg-chlorite toward log Q/
K ⫽ 0 at T ⫽ 350 °C (note that, because amorphous silica has fixed unitary
activity, the components are reduced to three in the computation of the degree of
Table 8.25 Equilibrium constants used for construction of figure 8.31 (activity diagram).
T (°C) P(bar) K249 K252 K254 K256 K259

25 1 6.841 1.967 67.223 194.828 63.632
250 39.728 4.009 4.869 31.271 89.803 26.688
300 85.805 3.576 5.242 26.217 75.075 21.531
350 165.125 3.019 5.470 18.356 52.048 13.703
freedom and, because P and T are fixed, three phases determine invariance in a
pseudoternary system).
To construct the diagram, we must know the various equilibrium constants
at the P and T of interest. It is thus necessary to compute the standard molal
Gibbs free energies of water, solid phases and aqueous ions in reaction. For solids,
the methods described in section 3.7 must be adopted. For aqueous ions, the
HKF model (section 8.11) or similar types of calculations are appropriate. To
avoid error progression and inconsistencies, the database used in calculating the
Gibbs free energies of both aqueous species and solid compounds must be inter-
nally consistent. In the case of figure 8.31, the constants were calculated at differ-
ent T and P conditions along the liquid-vapor univariant curve using the database
of Helgeson et al. (1978) (computer package SUPCRT ). The resulting values are
listed in table 8.25. It must be emphasized that extensive compilations of activity
diagrams in chemically complex systems are already available in the literature.
For instance, the compilation of Helgeson et al. (1976) for the system MgO-CaO-
FeO-Na2O-K2O-Al2O3-SiO2-CO2-H2O-H2S-H2SO4-Cu2S-PbS-ZnS-Ag2S-HCl
presents activity diagrams drawn for various T conditions (0, 25, 60, 100, 150,
200, 250, 300°C; P ⫽ 1 bar) and is an excellent tool for interpreting water-rock
equilibria at shallow depths.
8.21 Reaction Kinetics in Hydrous Environments
The arguments treated in the two preceding sections were developed in terms of
simple equilibrium thermodynamics. The “weathering” of rocks at the earth’s
surface by the chemical action of aqueous solutions, and the complex water-rock
interaction phenomena taking place in the upper crust, are irreversible processes
that must be investigated from a kinetic viewpoint. As already outlined in section
2.12, the kinetic and equilibrium approaches are mutually compatible, both being
based on firm chemical-physical principles, and have a common boundary repre-
sented by the “steady state” condition (cf. eq. 2.111).
8.21.1 Compositional Dependence of Rate Constants

The fundamentals of the kinetic approach to water-rock interaction are consti-
tuted by the quantitative assessment of reaction kinetics for mineral hydrolysis
and precipitation. These may be expressed in terms of rates of change in molality
of one of the main ions involved in the process. Denoting as mi the molality of
solute ion i and as ni the number of moles of i in solution, molality is related to
the mass of solvent by
ni
mi = Ω, (8.262)
nW
where nw are the moles of water solvent and ⍀ are the moles of H2O in 1 kg of
water (⍀ ⫽ 55.51). Differentiating equation 8.262 with respect to time t gives
dmi  Ω   dni  ni  dnW 

=  −  , (8.263)
dt  nW   dt  nW  dt 
(Delany et al., 1986), where dnw /dt is the consumption (or production) rate of
water molecules in the system. This term is often negligible with respect to the
other elemental balances.
The rate of a generic reaction j is represented by its absolute velocity v j, which
may be obtained from experiments, provided that the mass of solvent is held
constant throughout the experiment and no additional homogeneous or heteroge-
neous reactions concur to modify the molality of the ion in solution (Delany et
al., 1986):
nW dmi
vj ≅ × , (8.264)
Ωvi , j dt
where vi, j is the stoichiometric reaction coefficient of ion i as it appears in reaction

j (positive for products and negative for reactants; i.e., positive in the case of a
hydrolysis reaction).
Using a molar concentration notation instead of molality (ci ⫽ ni /Vs; cf. ap-
pendix 1), equation 8.264 is translated into
Vs dc
vj ≅ × i. (8.265)
vi , j dt
When more than one reaction affects the mass of the ion i in solution (e.g., disso-
lution and/or precipitation of heterogeneous solid assemblages; homogeneous re-
actions in solution), the overall change in mass of component i is given by the
summation of individual absolute reaction velocities v j, multiplied by their respec-
tive stoichiometric reaction coefficients:
T
dni
dt
= ∑ vi , j vj . (8.266)
j =1
The contribution of each reaction to the mass change is
 dni 
  = vi , j vj . (8.267)
 dt  j
Based on equation 8.266, the molality change induced by T simultaneous reac-

tions involving ion i is
dmi  Ω T   vW , j 
=  ∑ vi , j −    vj (8.268)
dt  nW  j =1   nW  
(Delany et al., 1986), where vW, j is the number of water molecules involved in the
jth reaction.
The absolute reaction velocity of an irreversible reaction can be related to an individ-

ual progress variable ␰j by
dξ j
vj = . (8.269)
dt
In section 2.12 we recalled the significance of the “overall reaction progress variable”
␰ which represents the extent of advancement of simultaneous irreversible processes in
heterogeneous systems (Temkin, 1963). The overall progress variable is related to progress
variables of individual reactions by
T
ξ = ∑ ξj (8.270)
j =1
(Delaney et al., 1986). It follows from equations 8.269 and 8.270 that we can also define
an “overall reaction rate” v, which is related to individual velocities by
T
v= ∑ vj (8.271)
j =1
(Delany et al., 1986). The “relative rate of the jth reaction” ␴j with respect to the overall
reaction progress variable ␰ is given by
dξ j
σj = (8.272)
dξ
(Temkin, 1963). From the various equations it follows that

vj
σj = (8.273)
v
and
dξ
v= . (8.274)
dt
We can generally group the rate laws adopted to describe dissolution and
precipitation kinetics into three main categories (Delany et al., 1986):
1. Simple rate laws in meq or ceq

2. Activity term rate laws
3. Rate laws based on transition state theory.
Simple Rate Laws in meq or c eq

When forward and backward reaction rates are fast enough to achieve a steady
state, reaction kinetics can be expressed as a function of steady state (or “equilib-
rium”) concentration (mieq or cieq ). The general equation of dissolution in terms
of molality change can be written as
dmiT  s 
( ),
n
=   k + mieq − mi (8.275)
dt Ω
where dmTi is the differential increment of total molality of ion i in solution (free
ion plus complexes), k⫹ is the rate constant of forward reaction (i.e., dissolution;
cf. eq. 2.110), s is the active surface area, and n is an integer corresponding to the
degree of reaction. The corresponding equation for precipitation is
dmiS  s 
( ),
n
− =   k − mi − mieq (8.276)
dt  Ω
where mSi is the total molality of i in the precipitating solid and k is the rate
constant of the backward reaction.
The simplest case of compositional dependence is the zero-order reaction, in
which the concentration gradient is not affected by concentration. Denoting the
molar concentration of the ith element (or component) as ci (and neglecting sur-
face area and volume of solution effects), we have
 ∂ ci 
  = −k+ . (8.277)
 ∂ t  T,P
Integration in dt of equation 8.277 yields
ci ,t = ci0 − tk + , (8.278)
where c0i , is the initial molar concentration of i at time t 0 .

In first-order reactions, the concentration gradient depends on the local con-
centration at time t:
 ∂ ci 
  = − kci ,t . (8.279)
 ∂ t  T,P
Integration in dt leads to the exponential dependency of molar concentration

on time:
( )
ci ,t = ci0 exp − kt . ( 2.280)
In this case, the rate constant has units of t⫺1 and is usually termed “half-life of
the reaction” (t1/2 ), the amount of time required to reduce the initial concentra-
tion c0i , to half:
ln 2
t1 2 = . (8.281)
k
As we will see in detail in chapter 11, the production of radiogenic isotopes fol-
lows perfect first-order reaction kinetics.
In molar notation, and referencing to the equilibrium concentration cieq as-
suming n ⫽ 1 (i.e., first-order reaction), equation 8.275 can be translated into
 s 
dci
( ).
n
=   k + cieq − ci (8.282)
dt  Vs 
Assuming that s and Vs remain constant throughout the process, the solution of
a first-order equation in the form of equation 8.282 (n ⫽ 1) is
 c eq − ci   s 
ln  ieq = −  k+ t ,
0   Vs 
(8.283)
 ci − ci 
where c0i is the initial concentration in solution at t ⫽ 0. If ci0 ⫽ 0, a semilogarith-

mic plot of (cieq ⫺ ci )/ci versus t should exhibit a constant slope corresponding to
the rate constant of the forward reaction k⫹ multiplied by ⫺s/Vs. As we can see
in figure 8.32, the rate of dissolution of biogenic opal is found to obey such a
Figure 8.32 Rates of dissolution of biogenic opal in seawater at T ⫽ 25 °C, for various
amounts of suspended opal (percent mass). Reprinted from Hurd (1972), with kind per-
mission from Elsevier Science Publishers B.V, Amsterdam, The Netherlands.
relationship. T and Vs being equal, the negative slope depends on the percent
amount of suspended opal (hence on the active surface area), but remains con-
stant throughout the process.
By “active surface area” we meant the kinetically active part of the total surface area.
According to Helgeson et al. (1984), this area is restricted to etch pits. Alternative esti-
mates of surface areas may be obtained from measurements of specific surface area s#
whenever solid particles have a narrow size range. The specific surface area for spherical
particles is given by
6
s# = , (8.284)
ρD
where ␳ is grain density and D is diameter. For cubic particles, the same equation applies
after substitution of edge length a for diameter:
6
s# = . (8.285)
ρa
Experimental measurements of surface area are usually carried out by BET gas absorp-
tion techniques (see Delany et al., 1986, for an appropriate treatment).
In some case the rate laws in meq or ceq are of high order—i.e., n ⬎ 1. If ci is
lower than the equilibrium concentration cieq, the solution of equation 8.282 is
( n − 1)s k t.
(c ) ( )
1− n 1− n
i
eq
− ci − cieq − ci0 = + (8.286)
Vs
To deduce the degree of reaction, one may use equation 8.286 with different inte-
ger values of n. The value of n that yields a straight line in a (cieq ⫺ ci )1⫺n versus
t plot is the order of reaction.
The equilibrium concentration cieq to be introduced for this purpose in equa-
tion 8.286 is the one asymptotically achieved in a ci versus t plot, at high values
of t (see Lasaga, 1981a, for a more appropriate treatment of high-order reaction
kinetics).
Activity Term Rate Laws

The simplest activity term expressions of rate laws are those involving the acidity
of the system. Because, generally, dissolution of a silicate can be schematically
depicted as
Mineral + vH + → SiO 02(aq) + cations ± Al-hydroxides (aq) ( 2.287 )
(Lasaga, 1984), the rates of dissolution can be expected to increase with aH⫹ in
acidic solutions. The experiments, in fact, denote the pH dependency of rate con-
stants as simple proportionalities of type:
( )
v+
k + ∝ a H+ , (8.288)
where v⫹ is a real number, depending on kinetics (not to be confused with the

stoichiometric integer v in eq. 8.287). Experimentally observed values of v⫹ are
More complex forms of activity term rate laws involve various ions and com-
plexes in solution. Dissolution and precipitation phenomena are in this approach
regarded as a summation of individual reactions taking place at the surface of
the solid. The net absolute rate is obtained as a summation of individual terms i,
each with its specific rate constant ki and activity product (Delany et al., 1986):
Table 8.26 pH dependence of mineral dissolution rates

(after Lasaga, 1984; with integrations).
Phase ␯⫹ pH Range References

K-feldspar 1.0 pH ⬍ 7 Helgeson et al. (1984)
1.0 pH ⬍ 5 Tole and Lasaga (unpublished results)
Nepheline ⫺0.20 pH ⬎ 7 Tole and Lasaga (unpublished results)
Diopside 0.7 2 ⬍ pH ⬍ 6 Schott et al. (1981)
Enstatite 0.8 2 ⬍ pH ⬍ 6 Schott et al. (1981)
Forsterite 1.0 3 ⱕ pH ⱕ 5 Grandstaff (1980)
Chrysotile 0.24 7 ⱕ pH ⱕ 10 Bales and Morgan (1985)
Quartz 0.0 pH ⬍ 7 Rimstidt and Barnes (1980)
Anorthite 0.54 2 ⬍ pH ⬍ 5.6 Fleer (1982)
Sr-feldspar 1.0 pH ⬍ 4 Fleer (1982)
⫺0.28 pH ⬎ 6 Fleer (1982)
v j = s ∑ ki ∏ ak
v k,i
. (8.289)
i k
Plummer et al. (1978) utilized, for instance, an equation in the form of equation
8.289 to describe the dissolution of calcite. According to the authors, the simulta-
neous reactions concurring to define the absolute velocity of dissolution are
CaCO3 + H+ ⇔ Ca2 + + HCO3− (8.290.1)
CaCO3 + H2CO03 ⇔ Ca2 + + 2HCO3− (8.290.2)
CaCO3 + H2O ⇔ Ca2 + + HCO3− + OH−. (8.290.3)
The absolute velocity of the bulk reaction is
v j = s  k1 aH + + k2 aH CO 0 + k3 aH2 O − k 4 aCa2 + aHCO−  . (8.291)

2 3 3
Figure 8.33 shows in detail the effect of the single rate constants on the for-
ward velocity at the various pH-PCO2 conditions (T ⫽ 25 °C). Although the indi-
vidual reactions 8.290.1–8.290.3 take place simultaneously over the entire compo-
sitional field, the bulk forward rate is dominated by reactions with single species
in the field shown: away from steady state, reaction 8.290.1 is dominant, within
the stippled area the effects of all three individual reactions concur to define the
overall kinetic behavior, and along the lines labeled 1, 2, and 3 the forward rate
corresponding to one species balances the other two.
Figure 8.33 Reaction mechanism contribution to the total rate of calcite dissolution re-
action as a function of pH and PCO2 at 25 °C. From L. N. Plummer, T. M. L. Wigley, and
D. L. Pankhurst (1978), American Journal of Science, 278, 179–216. Reprinted with per-
mission of American Journal of Science.
The backward velocity k4 is obtained by Plummer et al. (1978) by application of the

principle of detailed balancing. Denoting as K1 , K2 , KC , and KW, respectively, the equilib-
rium constants of the processes
H2CO03 ⇔ H+ + HCO−3 ( 8.292.1)
HCO −3 ⇔ H + + CO 23 −
CaCO 3 ⇔ Ca 2 + + CO 23 − ( 8.292.3 )
H2O ⇔ H+ + OH− , (8.292.4)

we have
k1 K 2 k2 K 2 k3 K 2
k4 = + a HCO − + a OH− , (8.293)
KC KC K1 3 K C KW
K2  
k4 =  k1 +
K C 
1
a H+ 2 3
(
k2 a H CO 0 + k 3  .

) (8.294)
The forward rate constants (cm/sec) are semilogaritmic functions of reciprocal tem-
perature (K) according to
444
log k1 = 0.198 − (8.295.1)
T
2177
log k2 = 2.84 − (8.295.2)
T
log k 3 = −5.86 −
317
T
(T < 298 K ) (8.295.3)
log k 3 = −1.10 −
1737
T
(T > 298 K ) . (8.295.4)
The corresponding activation energies deduced from the Arrhenius slope (see section
8.21.2) are listed in table 8.27.
Rate Laws Based on Transition State Theory

The Transition State Theory (TST) treatment of reaction kinetics accounts for
the fact that the true (microscopic) behavior of a dissolution reaction may involve
the production of an intermediate complex as an essential step of the bulk process.
Consider, for instance, the dissolution of silica in water. The macroscopic reaction
may be written as
SiO 2 → SiO 02 .
solid aqueous
(8.296)
The true (microscopic) reaction involves, however, the formation of an activated

complex in the intermediate step (Delany et al., 1986):
[
SiO2 + v H2O → SiO2 ⋅ v H2O * → SiO20 + v H2O, ] (8.297)
solid aqueous
where v molecules of water act as a catalyst of the process.

Because the energy of the activated complex is higher than the energies of
reactants and of the final reaction products, it is crucial to establish how many
activated complexes may form at given P-T-X conditions. This can be done by
application of quantum chemistry concepts.
Limiting the system to all accessible states of energy for the molecules of interest, the
probability of finding an individual molecule B in the state of energy ␧j is
Pj =
(
exp − ε j kT ),
(8.298)
Q
where k is Boltzmann’s constant and Q is the partition function
 − Ei 
Q= ∑ exp 
 kT 
 (8.299)
i
(cf. section 3.1). The partition function Q is related to the partition function qB relative to
the individual molecule B by
 −ε j 
qB = ∑ exp 
 kT 
 (8.300)
j
N
Q = q B B = QB N B ! (8.301)
(Lasaga, 1981b), where NB is the total number of B molecules in the system. Because
the chemical potential is related to the partial derivative of the Helmholtz free energy at
constant volume:
 ∂F   ∂ ln(Q N B !) 
µB =   = − kT   (8.302)
 ∂N B  T ,V  ∂N B  T ,V
recalling the Stirling approximation (eq. 3.116) and applying equation 8.301, we obtain:
 qB 
µ B = − kT ln   (8.303)
 NB 
Consider now a generic reaction of type
v A A + vB B → vC C + vD D. ( 8.304 )
Applying equation 8.303 to each term, we get, at equilibrium,

vA vB vC vD
q  q  q  q 
− kT ln  A  − kT ln  B  = − kT ln  C  − kT ln  D  . (8.305)
 NA   NB   NC   ND 
Rearranging equation 8.305, we obtain
N Cv C ⋅ N Dv D q Cv C ⋅ q Dv D
= . (8.306)
v v
N A A ⋅ N Bv B q A A ⋅ q Bv B
Finally, dividing both sides of equation 8.306 by the volume of the system, we obtain
(q ) ⋅ (q )
vC vD
c Cv C ⋅ c Dv D C V D V
K = = . (8.307)
c v A ⋅ c Bv B (q V ) ⋅ (q V)
vA vB
A
A B
Equation 8.307 is the basis of all applications founded on the TST approach.
Delany et al. (1986) give the following expressions of TST-based absolute

reaction rates:
  A 
v j = sk + q + 1 − exp  +   (8.308)
  ηRT  
  A 
− v j = sk − q − 1 − exp  +   . (8.309)
  ηRT  
Equation 8.308, representing the forward reaction rate, is written in terms of for-
ward direction parameters: A⫹ is thermodynamic affinity (cf. section 2.12), q⫹ is
the kinetic activity product for the left side of the microscopic reaction (i.e., in
the case of reaction 8.297, q⫹ ⫽ aSiO2 ⭈ avH2O ), and ␩ is a stoichiometric factor
relating the stoichiometric coefficient of the solid reactants in the macroscopic
and microscopic reaction notations (in the case of reactions 8.296 and 8.297, this
term is 1, because both reactions have the same stoichiometric coefficient.
8.21.2 Temperature and Pressure Dependence of

Rate Constants
If equilibrium in a system is perturbed for some reason, attainment of the new
equilibrium requires a certain amount of time t, depending on the rates of the
single reactions, and a certain amount of energy spent in activating the reaction
process. The relationship between reaction rate k and activation energy Ea has
the exponential form
Figure 8.34 Arrhenius plot of reaction rates for equilibrium of analcite ⫹ quartz ⇔ al-
bite in NaCl, and NaDS-bearing solutions, after Matthews (1980). Preexponential factors
and activation energies can be deduced from the fitting expression. NaDS ⬅ Na2Si2O5.
 −Ea 
k = k 0 exp  . (8.310)
 RT 
In equation 8.310, k 0 is the preexponential factor. Treating the preexponential

factor as not dependent on T, with a partial derivative in d(1/T) we have
∂ ln k E
=− a.
( )
∂1T R (8.311)
Plotting the natural logarithm of the reaction rate against the reciprocal of abso-
lute temperature (Arrhenius plot), the activation energy of the reaction is repre-
sented by the slope of equation 8.311 at the T of interest multiplied by ⫺R. If Ea
is constant over T, the semilogarithmic plot gives rise to a straight line, as illus-
trated in figure 8.34.
Activation energies of some important geochemical reactions are listed in
table 8.27.
Table 8.27 Activation energies of geochemical reactions (kJ/mole) (after Lasaga, 1981a, 1984)
Reaction Ea T Range (°C) Reference

H4SiO4 → SiO2 ⫹ 2H2O 50 0–300 Rimstidt and Barnes (1980)
quartz
SiO2 ⫹ 2H2O → H4SiO4 67–75 0–300 Rimstidt and Barnes (1980)
quartz
SiO2 ⫹ 2H2O → H4SiO4 61–65 0–300 Rimstidt and Barnes (1980)
(am)
Porcelanite → Quartz 97 300–500 Ernst and Calvert (1969)
(chert) (hydrothermal)
NaAlSi2O6 䡠 H2O ⫹ SiO2 → NaAlSi3O8 ⫹ H2O 186 300–500 Matthews (1980)
analcite quartz albite
Anorthite dissolution 35 25–70 Fleer (1982)
Nepheline dissolution 54–71 25–80 Tole and Lassaga (unpublished)
Mg2SiO4 ⫹ 4H⫹ → 2Mg2⫹ ⫹ H4SiO4 38 – Grandstaff (1980)
forsterite
Mg0.77Fe0.23SiO3 dissolution 44 – Grandstaff (1977)
orthopyroxene
Diopside dissolution 50–150 – Berner et al. (1980)
Enstatite dissolution 50 – Berner et al. (1980)
Augite dissolution 79 – Berner et al. (1980)
CaCO3 ⫹ H⫹ → Ca2⫹ ⫹ HCO3⫺ 8.4 5–48 Plummer et al. (1978)
calcite
CaCO3 ⫹ H2CO30 → Ca2⫹ ⫹ 2HCO3⫺ 42 5–48 Plummer et al. (1978)
calcite
CaCO3 ⫹ H2O → Ca2⫹ ⫹ HCO3⫺ ⫹ OH⫺ 6.3 5–25 Plummer et al. (1978)
calcite 33 25–48
CaCO3 → Ca2⫹ ⫹ CO32⫺ 35 5–50 Sjoberg (1976)
calcite
Ca2C2O6 ⫹ Mg2⫹ → CaMgC2O6 ⫹ Ca2⫹ 205 252–295 Katz and Matthews (1977)
calcite dolomite
FeS ⫹ S → FeS2 71 – Rickard (1975)
(am) pyrite
32
SO4 ⫹ H234S→ 34SO4 ⫹ H232S 7.5 pH ⬍ 3 Ohmoto and Lasaga (1982)
126 4 ⬍ pH ⬍ 7
201 pH ⬎ 7
Because reaction rates are also pressure-dependent (although they are much
less P-dependent than T-dependent), the volume of activation of reaction Va can
be deduced in an analogous fashion from the partial derivative
∂ ln k V
=− a . (8.312)
∂P RT
8.22 Chemistry of Seawater: Salinity, Chlorinity,

Alkalinity, and pH
The main saline constituents dissolved in seawater are Cl ⫺, SO2⫺ 2⫹

4 , Mg , K ,
⫹
Ca2⫹, and Na⫹. Their concentration ratios are remarkably constant, as is the bulk
Table 8.28 Composition and chlorinity of seawater for

a salinity of 35 (Stumm and Morgan, 1981). (1): weight
concentration in g/kg; (2): weight concentration/
chlorinity; (3): molarity/chlorinity.
Solute (1) (2) (3)

Na⫹ 10.77 0.556 0.0242
Mg2⫹ 1.29 0.068 0.0027
Ca2⫹ 0.4121 0.02125 0.000530
K⫹ 0.399 0.0206 0.000527
Sr 2⫹ 0.0079 0.00041 0.0000047
Cl⫺ 19.354 0.9989 0.0282
SO42⫺ 2.712 0.1400 0.0146
HCO3⫺ 0.1424 0.00735 0.00012
Br ⫺ 0.0673 0.00348 0.000044
F⫺ 0.0013 0.000067 0.0000035
B 0.0045 0.000232 0.0000213
Total 35.1605 1.8163 0.0710
amount of salts. This parameter is usually expressed as salinity (S) and represents
the percent by weight of dissolved inorganic matter per 103 g of seawater. In the
salinity calculation, the amount of Br⫺ and I⫺ is replaced by an equivalent
amount of Cl ⫺, and carbonate ions HCO⫺ 2⫺
3 and CO3 are converted into oxides.
The chlorinity of seawater (Cl ) is the amount of Cl⫺ and other halogens content
(expressed in g per kg of seawater), equivalent to the total amount of alkalis. This
parameter is practically determined by titration with AgNO3 , and the current
definition of chlorinity is “weight in g of Ag necessary to precipitate the halogens
(Cl ⫺, Br⫺, I⫺ ) in 328.5233 g of sea water” (Stumm and Morgan, 1981). Table
8.28 lists the mean composition of seawater at salinity 35. The first data column
lists the amount of salts in g per kg of solution. In column (2), weight is related
to chlorinity; column (3) shows the molarity/chlorinity ratio.
The salinity of seawater ranges from 33 to 37. Its value can be derived directly
from chlorinity through the relationship
S = 1.805Cl + 0.30. (8.313)
However, salinity values are easily obtained with a salinometer (which measures
electrical conductivity and is appropriately calibrated with standard solutions and
adjusted to account for T effects). The salinity of seawater increases if the loss of
H2O (evaporation, formation of ice) exceeds the atmospheric input (rain plus
rivers), and diminishes near deltas and lagoons. Salinity and temperature concur
antithetically to define the density of seawater. The surface temperature of the
sea reflects primarily the latitude and season of sampling. The vertical thermal
profile defines three zones: surface (10–100 m), where T is practically constant;
thermoclinal (100–1000 m), where T diminishes regularly with depth; and abyssal
(⬎ 1000 m), with constant low T (2–4 °C). Density at atmospheric level is usually
denoted by parameter ␴:
 kg 
σ = Density 3
− 1 × 1000 (8.314)
 dm 
(thus, for seawater with density 1.025 at atmospheric level, ␴ ⫽ 25).

The density of seawater varies from a maximum of ␴ ⫽ 29, observed in deep
antarctic waters, to a minimum of ␴ ⫽ 25 in subtropical oceanic thermoclinal
waters. The high density of polar waters causes them to sink beneath subtropical
waters and constitutes the driving force of deep oceanic circulation.
The HCO⫺ 3 amount listed in table 8.28 actually represent carbonate alkalinity, calcu-
lated neglecting CO32⫺. The alkalinity [Alk] of a solution represents its capacity to neutral-
ize strong acids in solution. Alkalinity can be defined as “the sum of equivalents of all
species whose concentrations depend on the concentration of H⫹ ions in solution, minus
the concentration of H⫹ ions in solution.” For a solution containing borates and carbon-
ates, alkalinity can be expressed as
[ Alk] = [ HCO −3 ] + 2[ CO 23 − ] + [ B(OH) −4 ] + [OH − ] − [H + ]. (8.315)
where the terms in square brackets denote molality. If we consider only carbonate equilib-
ria, equation 8.315 reduces to
[ Alk] = [ HCO −3 ] + 2[ CO 23 − ] + [OH − ] − [H + ] , (8.316)
and, in the absence of CO32⫺ (at the normal pH of seawater, the carbonate ion is virtually
absent; see figure 8.21),
[ Alk] = [ HCO −3 ] + [OH − ] − [H + ]. (8.317)
[ HCO −3 ] ≅ [ Alk] . (8.318)
Figure 8.35 shows the redox state and acidity of the main types of seawaters.
The redox state of normal oceanic waters is almost neutral, but they are slightly
alkaline in terms of pH. The redox state increases in aerated surface waters. Sea-
waters of euxinic basins and those rich in nutrients (eutrophic) often exhibit Eh-
pH values below the sulfide-sulfate transition and below carbonate stability limits
(zone of organic carbon and methane; cf. figure 8.21). We have already seen (sec-
tion 8.10.1) that the pH of normal oceanic waters is buffered by carbonate equi-
libria. At the normal pH of seawater (pH ⬵ 8.2), carbonate alkalinity is 2.47
mEq per kg of solution.
Figure 8.35 Mean Eh-pH values for various types of seawaters. A ⫽ normal oceanic
waters; B ⫽ oxidized surface waters; C ⫽ euxinic basins; D ⫽ eutrophic waters. Dashed
line: field of natural waters according to Baas Becking et al. (1960).
The relative constancy in the chemistry of the main saline components of

seawater does not extend to minor and trace constituents, which are somewhat
variable.
Interpretation of the bulk chemistry of seawater is based on five types of
general considerations:
1. Chemical equilibrium between seawater and sediments

2. Exchanges with the atmosphere
3. Water-rock interactions
4. Interactions between biologic and mixing cycles
5. Kinetics of the inputs of primary constituents.
The above-listed factors concur simultaneously to define the stationary chemistry

of seawater. Moreover, the control operated by the various phenomena is selective
for the various types of elements: for instance, the amounts of Ca are largely
controlled by precipitation of carbonates, those of Na and K by silicate hydroly-
Table 8.29 Trace element solubility control at T ⫽ 25 °C and P ⫽ 1 bar (from

Turekian, 1969).
Expected Observed Observed

Concentration Concentration Concentration in
Insoluble in Seawater, in Seawater, Reducing Muds,
Element Salt log(mole/l) log(mole/l) log(mole/l)
Lanthanum LaPO4 ⫺11.1 ⫺10.7 -

Thorium Th3(PO4)4 ⫺11.8 ⫺11.7 -
Cobalt CoCO3 ⫺6.5 ⫺8.2 ⫺12.1
Nickel Ni(OH)2 ⫺3.2 ⫺6.9 ⫺10.7
Copper Cu(OH)2 ⫺5.8 ⫺7.3 ⫺26.0
Silver AgCl ⫺4.2 ⫺8.5 ⫺19.8
Zinc ZnCO3 ⫺3.7 ⫺6.8 ⫺14.1
Cadmium CdCO3 ⫺5.0 ⫺9.0 ⫺16.2
Mercury Hg(OH)2 ⫹1.9 ⫺9.1 ⫺43.7
Lead PbCO3 ⫺5.6 ⫺9.8 ⫺16.6
Table 8.30 Mean composition of seawater for a salinity of 35 (Turekian, 1969). Values
expressed in ppb.
H 1.10 ⫻ 108 Ti 1.0 Ba 21

He 0.0072 V 1.9 La 0.0034
Li 170 Cr 0.2 Ce 0.0012
Be 0.0006 Mn 0.4 Pr 0.00064
B 4.450 Fe 3.4 Nd 0.0028
C (inorganic) 28,000 Co 0.39 Sm 0.00045
C (organic solute) 500 Ni 6.6 Eu 0.000130
N (dissolved as N2) 15.500 Cu 0.9 Gd 0.00070
N (as NO2⫺, NO3⫺, NH4⫹, 670 Zn 0.003 Tb 0.00014
and inorganic solute)
O (dissolved as O2) 6,000 As 2.6 Ho 0.00022
O (as H2O) 8.33 ⫻ 108 Se 0.090 Er 0.00087
F 1,300 Br 67,300 Tm 0.00017
Ne 0.120 Kr 0.21 Yb 0.00082
Na 1.08 ⫻ 107 Rb 120 Lu 0.00015
Mg 1.29 ⫻ 106 Sr 8,100 Hf ⬍ 0.008
Al 1 Y 0.0013 Ta ⬍ 0.0025
Si 2,900 Zr 0.026 W ⬍ 0.001
P 88 Nb 0.015 Rh 0.0084
S 9.04 ⫻ 105 Mo 10 Au 0.011
Cl 1.94 ⫻ 107 Ru 0.0007 Hg 0.15
Ar 450 Ag 0.28 Pb 0.03
K 3.92 ⫻ 105 Cd 0.11 Bi 0.02
Ca 4.11 ⫻ 105 Sn 0.81 Ra 1 ⫻ 10⫺7
Sc ⬍ 0.004 Sb 0.33 Th 0.0004
I 0.64 Xe 0.047 Pa 2 ⫻ 10⫺10
Cs 0.30 U 3.3
Figure 8.36 Changes in quantities of phosphorus, nitrogen, and silicon in seawater with
depth.
sis, SiO2 by silicate hydrolysis plus biogenic fixation (plankton), Mn by water-

rock interactions in mid-ocean ridge zones, S by evaporite precipitation, and
so on.
It is evident from the considerations above that we cannot expect the minor
saline constituents of seawater to reflect the solubility product of their primary
salts. For example, silver, lead, mercury, and copper form chlorides that are highly
soluble in water. Nevertheless, their observed concentrations in seawater are much
lower (i.e., by 3 or 4 orders of magnitude) than would be expected from the solu-
bility products. Analogously, the solubility products of carbonates and insoluble
hydroxides of transition elements suggest elemental concentrations higher than
those actually observed. Table 8.29 compares hypothetical concentrations derived
from the solubility products of insoluble salts with actual concentrations in oxi-
dized surface waters and reducing muds (below the sulfide-sulfate transition).
The expected concentration values were calculated by Turekian (1969), adopt-
ing the following activities of anionic ligands in solution: log[PO2⫺ 4 ] ⫽ ⫺9,3,
⫺
log[CO2⫺ 3 ] ⫽ ⫺5,3; log[OH ] ⫽ ⫺6; log[S2⫺
] ⫽ ⫺9.
Note that the observed concentrations of La and Th in seawater are in fact
near the values expected from the solubility of phosphates (La and Th are effec-
tively fixed as phosphates in almyrolithic exchanges between biogenic sediments
and seawater), but the concentrations of the remaining elements are far lower
than the values dictated by the solubility products.
Figure 8.37 REE concentration profiles with depth in Atlantic and Pacific Oceans.
Table 8.30 shows the chemistry of seawater compiled by Turekian (1969) for
major, minor, and trace constituents, expressed in parts per billion (ppb) at a
mean salinity of 35. The listed values are estimates of mean amounts in solution,
whereas elemental concentrations actually vary with depth. The most conspicu-
ous variations are observed in the first 200 m from the surface, where photosyn-
thetic processes are dominant and phosphorus and nitrogen are fixed by plankton
and benthos, as well as silica and calcium, which constitute, respectively, the skel-
etons of planktonic algae (diatom) and the shells of foraminifera and mollusks.
Figure 8.36 shows how the elemental amounts of P (generally present as com-
⫺
plexes of the phosphate ion PO3⫺ 4 ), N (present as NO3 complexes), and Si vary
with depth in the various oceanic basins. Concentrations are expressed in micro-
moles per liter of solution. Note the relative constancy of concentrations below
1000 m.
Trace elements also exhibit systematic changes in concentration with depth.
Figure 8.37 shows, for instance, concentration profiles of rare earth elements
(REE) determined by De Baar et al. (1985) in Pacific and Atlantic waters. Note
that the concentration profiles differ for the various elements in the series; in
particular, the amount of Ce is quite high in surface waters of the Atlantic Ocean.
Figure 8.38 Ce anomalies in normalized REE patterns of seawater at various depths.

Reproduced with modifications from H. J. W. DeBaar, M. P. Bacon, P. G. Brewer, and
K. W. Bruland, Geochimica et Cosmochimica Acta, 49, 1943–1960, copyright 䉷 1985, with
0X5 1GB, UK.
Figure 8.38 shows that the anomalous concentration of Ce results in a posi-

tive Ce anomaly in normalized REE patterns (relative abundances arranged in
order of increasing atomic number). This anomaly, which may be ascribed to
scavenging by Mn-Fe hydroxides, decreases progressively and inverts with depth.
8.23 Steady State Models and the Concept of “Time

of Residence”
The kinetic interpretation of the chemistry of oceanic waters (kinetics of inputs

of primary constituents; interactions between biologic and mixing cycles) leads
to the development of steady state models, in which the relatively constant chemis-
try of seawater in the recent past (i.e., Phanerozoic; cf. Rubey, 1951) represents a
condition of kinetic equilibrium among the dominant processes. In a system at
steady state, the inputs of matter exactly balance the outputs. In the “seawater
system,” inputs are constituted primarily by:
1. Dissolved and particulate matter of rivers

2. Mid-ocean-ridge thermal springs and submarine eruptions
3. Atmospheric inputs.
Primary outputs are produced essentially by sedimentation and (to a much

lower extent) by emissions in the atmosphere. The steady state models proposed
for seawater are essentially of two types: “box models” and “tube models.” In box
models, oceans are visualized as neighboring interconnected boxes. Mass transfer
between these boxes depends on the mean residence time in each box. The differ-
ence between mean residence times in two neighboring boxes determines the rate
of flux of matter from one to the other. The box model is particularly efficient
when the time of residence is derived through the chronological properties of
first-order decay reactions in radiogenic isotopes. For instance, figure 8.39 shows
the box model of Broecker et al. (1961), based on 14C. The 14C/ 12C ratio, normal-
ized to the atmospheric value, is determined for each box. According to the mea-
sured 14C/ 12C ratios, the maximum time of residence in each box is 1000 years,
whereas the bulk time of residence in the oceans (time elapsed between precipita-
tion plus river inputs and evaporation plus transport to continents) is 40,000
years.
The “tube models” consider the oceanic mass of water as subdivided into
columns. Mass transfer between columns takes place by advection and diffusion.
Examples of tube models may be found in Munk (1966) and Bieri et al. (1966),
to whom we refer readers for further clarification.
The thermal motion of molecules of a given substance in a solvent medium causes

dispersion and migration. If dispersion takes place by intermolecular forces acting within
a gas, fluid, or solid, “molecular diffusion” takes place. In a turbulent medium, the migra-
tion of matter within it is defined as “turbulent diffusion” or “eddy diffusion.” Diffusional
flux Jd is the product of linear concentration gradient dC/dX multiplied by a proportional-
ity factor generally defined as “diffusion coefficient” (D) (see section 4.11):
dC
J d = −D × . (8.319)
dX
“Advection” is generally defined as the displacement of a portion of matter with respect

to the point of observation under the influence of force fields (Lerman, 1979). An advec-
tion flux [Ja ; g/(sec ⫻ cm2 )] of a material with density ␳ (g/cm3 ) moving at velocity V
(cm/sec) with respect to the point of observation is given by
Ja = ρ × V . ( 8.320 )
Figure 8.39 Box model for steady state chemistry of seawater. Numbers in boxes: 14C/
12
C ratio normalized to atmospheric value. L: specific transfer rate between atmosphere
and ocean; R: transfer rates between oceanic reservoirs. Reprinted from W. S. Broecker,
R. D. Gerard, M. Ewing, and B. C. Heezen, Geochemistry and physics of ocean circula-
tion. In Oceanography, M. Sears, ed., copyright 䉷 1961 American Association for the
Advancement of Science. Reprinted with permission of the American Association for the
Advancement of Science, Washington.
Steady state models are based on the concept of the mean time of residence
of the elements in the system. Mean time of residence ␶ for a steady state is
defined as
mass in the system mass in the system

τ = = . (8.321)
unit time input unit time output
Indexing as Ji the various input and output fluxes, we can write
M
τ = ,
∑ Ji (8.322)
i
where mass M and fluxes have the same units (e.g., M ⫽ moles, Ji ⫽ moles/year).
The “fractional time of residence” is relative to a given input:
M
τi = , (8.323)
Ji
Table 8.31 Times of residence of elements in seawater in years (data from

Brewer, 1975; Stumm and Morgan, 1981).
Element(s) t log t Element(s) t log t

Cl, Br 1 ⫻ 108 8 Ni 9.0 ⫻ 104 5.0
Na 6.8 ⫻ 107 7.8 V, Hg 8.0 ⫻ 104 4.9
B 1.3 ⫻ 107 7.1 As 5.0 ⫻ 104 4.7
Mg 1.2 ⫻ 107 7.1 Sc, Ag, Ba 4.0 ⫻ 104 4.6
K 7.0 ⫻ 106 6.8 Cu, Zn, Se 2.0 ⫻ 104 4.3
Rb, Sr 4.0 ⫻ 106 6.6 Si 1.8 ⫻ 104 4.2
U 3.0 ⫻ 106 6.5 Ti 1.3 ⫻ 104 4.1
Li, N 2.3 ⫻ 106 6.4 Ga, Mn 1.0 ⫻ 104 4.0
Ca 1.0 ⫻ 106 6.0 Sb 7.0 ⫻ 103 3.8
Cs 6.0 ⫻ 105 5.8 Cr 6.0 ⫻ 103 3.8
F 5.2 ⫻ 105 5.7 La 6.0 ⫻ 102 2.8
I 4.0 ⫻ 105 5.6 Pb 4.0 ⫻ 102 2.6
Mo, Au 2.0 ⫻ 105 5.3 Fe 2.0 ⫻ 102 2.3
P 1.8 ⫻ 105 5.2 Al 1.0 ⫻ 102 2.0
W 1.2 ⫻ 105 5.1 Th 6.0 ⫻ 101 1.8
and the “mean time of residence” is given by
1 1 1 1
= + +L . (8.324)
τ τ1 τ 2 τn
Indexing input and output fluxes as Jin and Jout, respectively, the mass varia-
tion in time dt is given by
dM
= J in − J out . (8.325)
dt
Assuming the removal rate of a given element from the system to be proportional
(at first approximation) to the elemental abundance in the system, through pro-
portionality constant Kt :
J out = K t × M , (8.326)
at steady state we have
dM
= 0; J in = K t × M , (8.327)
dt
1 M
= = τ. (8.328)
Kt J in
The time of residence thus represents the reciprocal of the constant of re-
moval from the system. The times of residence of elements in the “seawater sys-
tem” are generally evaluated on the basis of equation 8.327. The main input flux is
fluvial: the fractional time of residence based on fluvial input is thus the primary
constituent of the mean time of residence (cf. eq. 8.322). An alternative way of
evaluating the time of residence is based on output flux due to sedimentation,
applying equation 8.326 (the two methods give concordant results). Table 8.31
lists times of residence of elements in seawater, expressed in years and arranged
in decreasing series. Long times of residence (e.g., those of Cl, Br, Na, B, and
Mg) are indicative of low reactivity in the system; short times of residence are
connected with oversaturation and high reactivity (e.g., La, Pb, Al, Fe, and Th).
CHAPTER NINE
Geochemistry of Gaseous Phases
9.1 Perfect Gases and Gaseous Mixtures
The equation of state of perfect gases:
PV = nRT , (9.1)
where n is the number of moles of component, can be applied to a gas phase

composed of molecules moving freely as a result of thermal agitation, in the ab-
sence of mutual interactions. For a mixture of perfect gases the same law holds,
taking into account the fact that, in this case, n represents the summation of all
molecular species in mixture—i.e.,
n= ∑ ni (
= n X1 + X 2 + X 3 + L + X i , ) (9.2)
i
where X1 , X2 , . . . ,Xi are the molar fractions of the various gaseous species.
The activity of a perfect gas, as for any substance, is unitary, by definition, at
standard state. Moreover, for a perfect gas, activity is (numerically) equivalent to
pressure, at all pressures. Let us consider the relationship existing, with T held
constant, between the chemical potential of component i in gaseous phase g at 1
bar ( ␮i,1,T,g ) and at pressure P ( ␮i,P,T,g ):
P
µ i , P,T , g = µ i ,1,T , g + ∫ Vi , g dP. (9.3)
1
Because for a pure phase partial molar volume V i,g is equivalent to molar volume
V g , and since we have defined as standard state the condition of the pure compo-
nent at P ⫽ 1 bar and T of interest:
µ i ,1 , T , g ≡ µ i0, g , ( 9.4 )
612 /
Geochemistry of Gaseous Phases / 613
we have
P
µ i , P,T , g = µ i0, g + ∫ Vg dP. (9.5)
1
Because
RT ln a i , g = µ i , P , T , g − µ io, g , ( 9.6 )
from equations 9.1, 9.5, and 9.6 for n ⫽ 1, we obtain
P
dP
RT ln a i , g = ∫ RT
P
, (9.7)
1
which, when integrated, gives
RT ln a i , g = RT ln P. ( 9.8)
Hence
a i , g = P. ( 9.9)
In the more general case of a standard state pressure of reference different

from 1, we have
P
RT ln a i , g = RT ln   (9.10)
 Pr 
and
P
ai,g = . (9.11)
Pr
A real gas does not obey the equation of state of perfect gases (eq. 9.1) be-
cause of the interactions among gaseous molecules. The higher the pressure and
the lower the temperature, the greater the deviation from ideality. At very low P,
intermolecular distances are great and interactions between gaseous molecules
are virtually absent. For almost all gaseous species of geochemical interest, the
approximation to a perfect gas is acceptable only over a very small P range.
Interaction forces among gaseous molecules are generally subdivided into
Figure 9.1 Molecular interaction potentials in Stockmayer’s (1941) model for H2O vapor.
(a): antiparallel dipolar moments; (b): parallel dipolar moments. Reprinted from D.
Eisemberg and W. Kauzmann, The Structures and Properties of Water, 1969, by permission
of Oxford University Press.
long-range forces (electrostatic, inductive, and dispersive forces) and short-range

forces (repulsive forces). Figure 9.1 shows the energy potentials of two gaseous
molecules of H2O (A-B), according to the classical model of Stockmayer (1941).
Bulk potential UAB obeys the equation
µ 2 f 4 εσ 6 4 εσ 12
U AB = − − + . (9.12)
R3 R6 R 12
The first term on the right in the sum of potentials in equation 9.12 is the electro-
static term (represented here as a simple dipole moment). Factor f is the angular
function of dipole orientation:
( )
f = sin θ A sin θ B cos φ A − φ B − 2 cos φ A cos φ B . (9.13)
Angles ␪ and ␾ in equation 9.13 define the dipole orientation with respect to a
coordinate system centered on the oxygen nuclei. The second term on the right
in equation 9.12 represents the dispersive energy (independent of molecular ori-

entation), and the third term represents the repulsive interactions (also indepen-
dent of orientation). ␴ is the “collision diameter of the gaseous molecule” and
represents the distance at which attractive dipole moments and repulsive energy
terms mutually compensate (i.e., when R ⫽ ␴, the series of potentials of eq. 9.12
reduces to the first term), and ␧ represents the “characteristic energy” of the gas-
eous molecule. It is of interest to recall that the ␴ value observed by Stockmayer
(1941) for gaseous H2O is identical to the hydrogen-bond distance in ice I (Eisem-
berg and Kauzmann, 1969).
Note in figure 9.1 that bulk potential UAB depends greatly on the mutual
orientation of dipoles: molecules with parallel dipoles form more stable bonds
than molecules with antiparallel dipoles. Note, moreover, as already stated, that
repulsive and dispersive forces are not affected by the state of dipole orientation.
Deviations from the equation of state of perfect gases may be expressed in
terms of either pressure or volume. It is obvious that specific interactions among
the molecules of a real gas are translated into modifications of volume with re-
spect to the volume occupied by n moles of an ideal perfect gas at the same
temperature T, or, T and volume being held constant, into modifications of pres-
sure exercised by the gaseous system toward the exterior. Analytically, this may
be expressed through the concept of “fugacity.” Fugacity f i of a real pure gas i
is obtained through
P
RT ln fi = RT ln P + ∫ (Vi )
− Vperfect dP , (9.14)
Pr
where V i is the molar volume of the real gas and V perfect is the volume occupied
by one mole of perfect gas under the same P and T conditions. The relationship
between fugacity f i and pressure Pi of a real gas i is given by
fi = Pi Γi . (9.15)
In equation 9.15, ⌫i is the “fugacity coefficient”:
( )
 P Vi − Vperfect dP 
Γi = exp  ∫ . (9.16)
P RT 
 r 
Because ⌫i is adimensional, it is evident that fugacity has the dimension of pres-

sure. It is important to emphasize that it is conceptually erroneous to consider
the fugacity of a gaseous species as the equivalent of the thermodynamic activity
of a solid or liquid component (as is often observed in literature). The relationship
between activity ai and fugacity f i of a gaseous component is given by
fi
ai = , (9.17)
fi 0
where fi0 is fugacity at standard state. Fugacity fi0 ⫽ 1 only when Pi ⫽ 1 bar and
T and P are such that ⌫i ⬅ 1 (cf. eq. 9.15). If the adopted standard state is, for
instance, pure gas at the P and T of interest and P is high (or T is low), fi0 ⬆ 1,
and hence ai ⬆ f i .
9.2 Real Gases at High P and T Conditions
Figure 9.2 shows how fugacity coefficient ⌫i of H2 , O2 , and CO2 behaves with
increasing P at various T (0 °C, 100 °C, 200 °C). Tables 9.1 and 9.2 list discrete
values attained by ⌫i for H2O and CO2 gases under various P and T conditions.
Values of ⌫i for gases of geochemical interest for which experimental data are
Figure 9.2 Fugacity coefficients of H2, O2, and CO2 plotted as functions of P at various
T. Reprinted from Garrels and Christ (1965) with kind permission from Jones and Bartlett
Publishers Inc. Copyright 䉷 1990.
Table 9.1 Fugacity coefficients of CO2 under various P and T conditions (Mel’nik, 1972).
T (K)
P (atm) 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500
500 0.57 0.85 1.02 1.11 1.15 1.17 1.19 1.20 1.20 1.20 1.20 1.20
1,000 0.63 0.95 1.16 1.28 1.34 1.37 1.38 1.38 1.37 1.37 1.37 1.37
1,500 0.82 1.20 1.44 1.56 1.62 1.62 1.62 1.59 1.59 1.57 1.55 1.53
2,000 1.16 1.60 1.85 1.95 2.00 1.95 1.91 1.86 1.82 1.78 1.76 1.71
3,000 2.46 3.00 3.18 3.15 2.95 2.08 2.46 2.34 2.29 2.19 2.09 2.04
4,000 5.49 5.60 5.60 5.18 4.67 3.89 3.55 3.31 3.16 2.95 2.75 2.63
5,000 12.5 11.3 9.88 8.52 6.92 5.75 5.01 4.57 4.27 3.89 3.63 3.39
6,000 28.4 22.1 17.4 13.9 10.5 8.32 7.24 6.31 5.75 5.13 4.68 4.27
7,000 64.2 42.7 30.2 22.4 16.2 12.6 10.5 8.91 7.94 6.92 6.17 5.50
8,000 143 81.4 51.8 35.7 24.6 18.6 15.1 12.6 10.7 9.12 7.94 7.08
9,000 319 154 88.1 56.3 38.0 28.8 22.4 17.8 14.8 12.3 10.5 9.33
10,000 703 289 148 80.9 56.2 41.7 31.6 24.6 20.0 16.2 13.5 11.8
Table 9.2 Fugacity coefficients of H2O ( ⌫i ⫻ 102; Helgeson and Kirkham, 1974).
P (kbar)
T (°C) 0.5 1 2 3 4 5 6 7 8 9
200 3.66 2.36 0.19 0.20 0.24 0.30 0.38 0.50 0.66 0.89
250 8.59 5.46 0.43 0.45 0.51 0.62 0.77 0.97 1.25 1.62
300 16.59 10.50 0.82 0.83 0.92 10.9 13.2 16.3 20.5 25.9
350 27.63 17.52 13.5 13.5 14.8 17.1 20.3 24.6 30.3 37.5
400 40.83 26.22 20.2 19.9 21.6 24.5 28.7 34.3 41.4 50.4
450 54.16 35.94 27.8 27.3 29.3 32.9 38.0 44.7 53.3 64.0
500 64.81 45.89 36.1 35.3 37.6 41.8 47.8 55.6 65.4 77.6
550 72.59 55.28 44.5 43.5 46.1 51.0 57.7 66.5 77.5 90.9
600 78.48 63.58 52.7 51.7 54.6 60.0 67.5 77.1 89.0 103.4
650 82.97 70.57 60.3 59.5 62.8 68.6 76.7 87.0 99.0 115.0
700 86.46 76.33 67.3 66.8 70.4 76.7 85.3 96 109.5 125.4
800 91.37 84.87 78.9 79.5 83.9 91.0 100.4 112 126.3 142.9
not available in the literature can be obtained through the “universal chart of
gases” (figure 9.3). Construction of this chart is based on the fact that the thermo-
dynamic properties of all real gases are similar functions of P and T, when the
parameters of the equation of state are reduced with respect to the values of
critical temperature Tc, critical pressure Pc, and critical volume Vc of the com-
pound of interest. We can thus introduce the concepts of reduced pressure Pr ,
reduced temperature Tr , and reduced volume Vr:
Pi
Pr,i = (9.18)
Pc,i
Figure 9.3 Universal chart of gases. From Garrels and Christ (1965); reprinted with kind permission from Jones and Bartlett Publishers
Inc. Copyright 䉷 1990.
Table 9.3 Critical temperatures (Tc ) and critical pressures

(Pc ) for gaseous species of geochemical interest (from Garrels
and Christ, 1965).
Tc
Compound (°C) (K) Pc (atm)

CO2 31.1 304.3 73
CS2 273 546 76
CO ⫺139 134 35
COS 105 378 61
Cl2 144.0 417.2 76.1
He ⫺267.9 5.3 2.26
H2 ⫺239.9 33.3 12.8
HBr 90 363 84
HCl 51.4 324.6 81.6
HI 151 424 82
H2Se 138 411 88
H2S 100.4 373.6 88.9
Hg ⬎ 1550 ⬎ 1823 ⬎ 200
CH2 ⫺82.5 190.7 45.8
N2 ⫺147.1 126.1 33.5
O2 ⫺118.8 154.4 49.7
SiF4 ⫺1.5 271.7 50
SiH4 ⫺3.5 269.7 48
SnCl4 318.7 591.9 37.0
S 1040 1313 –
SO2 157.2 430.4 77.7
SO3 218.3 491.5 83.6
H2O 374.0 647.2 217.7
Ti
Tr,i = (9.19)
Tc,i
Vi
Vr,i = . (9.20)
Vc,i
Values of critical temperature and critical pressure for gaseous species of geo-
chemical interest are listed in table 9.3.
The ⌫i values allow the calculation of fugacity and, thence, the chemical po-
tential of the gaseous component at the P and T of interest, applying
 f 
µ i , g = µ i0, g + RT ln  i0  . (9.21)
 fi 
9.3 The Principle of Corresponding States and Other

Equations of State for Real Gases
As we have already seen, the universal chart of gases assumes that all gaseous
species exhibit the same sort of deviation from ideal behavior at the same values
of Tr , Pr , and Vr . This fact, known as the principle of corresponding states, is ana-
lytically expressed by the “deviation parameter” (or “compressibility factor”) Zg .
For n ⫽ 1,
Zg =
PV
RT
(
= f Pr , Tr . ) (9.22)
Confirmation of the validity of the principle of corresponding states is shown

in figure 9.4A, where we see the analogies in the reduced isotherms of H2O, CO2 ,
and N2 according to reduced pressure. However, figure 9.4B shows that the prin-
ciple of corresponding states is only an approximation of actual behavior in the
vicinity of the critical point: in the case of H2O, there are in fact discrepancies
between predicted and actual behavior.
Deviation parameter Zg is operationally expressed as a virial expansion on P
or V:
Z g ( P,T ) = 1 + BP + CP 2 + DP 3 + L (9.23)
Z g (V ,T ) = 1 + B ′V + C ′V −2 + D ′V −3 + L. (9.24)
Figure 9.4 (A) Reduced isotherms of N2, CO2, and H2O vapor according to reduced
pressure. (B) Comparison between experimental isotherms of H2O and Tr predicted by
principle of corresponding states.
Virial expansion coefficients B, C, D and B⬘, C⬘, D⬘ in equations 9.23 and 9.24 are
simple functions of T and are specific to each gaseous species. Virial expansion
equations such as equations 9.23 and 9.24 do not reproduce the behavior of real
gases in the vicinity of the critical point and under high P and T conditions with
satisfactory precision.
Deviations from ideality of real gases can also be expressed through the van
der Waals equation:
(P + aV )(V − b) = RT ,
−2
(9.25)
where term aV⫺2 is defined as “internal pressure” or “cohesive pressure” and

term b as “co-volume.” Cohesive pressure is the result of attractive forces among
nonideal gaseous molecules; co-volume reflects the intermolecular distance at
which repulsion between neighboring molecules becomes appreciable (see sec-
tion 9.1).
If we define the distance between the centers of two gaseous molecules at the
collision point as d, we have
4
b= πd 3 (9.26)
3
and, for a rigid-sphere model,
32 3
b= π rg , (9.27)
3
where rg is the radius of the gaseous molecule, conceived as a rigid sphere.

An alternative (and more accurate) form of equation of state for real gases is
the Redlich-Kwong equation:
 
P +
a
( )
 V − b = RT .
 ( )
V V + b T 1 / 2 
(9.28)
Parameters a and b in equation 9.28 have the same significance as in equation 9.25. How-
ever, whereas in the van der Waals equation these parameters are constant for a given
gaseous species, in the Redlich-Kwong formulation they are often assumed to be variable.
For instance, Holloway (1977) adopted a coefficient a varying with T:
a = f ( T ). ( 9.29 )
Touret and Bottinga (1979) and Bottinga and Richet (1981) proposed
 a  3  a  6 
a = a 1  3  −  3   + a 2 (9.30)
 V  V  

and
 V   −1
b = ln   + b1 b2 , (9.31)
  a3  
where a1 , a2 , a3 , b1 , and b2 are adjustable coefficients for each gaseous species.

Kerrick and Jacobs (1982) adopted the Carnahan-Starling modification of the
Redlich-Kwong equation, with polynomial expansions of coefficient a on V and T:
a = c + dV −1 + eV −2 ( 9.32 )
c = c1 + c2 T + c 3 T 2 ( 9.33 )
d = d 1 + d 2 T + d 3T 2 (9.34)
e = e1 + e2 T + e 3 T 2 (9.35)
Halback and Chatterjee (1982) considered coefficient a as a function of T and coefficient

b as a function of P:
a = a 1 + a 2 T + a 3T −1 (9.36)
b=
(1 + b P + b P + b P ) .
1 2
2
3
3
(b + b P + b P )
4 5 6
2 (9.37)
More recently, Saxena and Fei (1987) proposed quantifying deviations from ideality
of real gases through an expression they defined as “nonvirial”:
Z g = A + BP + CP 2 + L. (9.38)
The difference between equation 9.38 and the corresponding virial expansion of equation
9.23 lies in term A, which substitutes 1 and which is a function of T, as are the remaining
coefficients. Equation 9.38 also finds application in a “corresponding states” notation:
Z g = A + BPr + CPr2 + L. (9.39)
Table 9.4 lists the parameters of the nonvirial expansion of equation 9.38 for H2O
and the parameters of the corresponding states notation for the remaining gaseous species.
The equation of state appropriate to a given gaseous species under high P

and T conditions must be selected with care. It may generally be stated that,
Table 9.4 Parameters of nonvirial expansion of Saxena-Fei (from Ganguly and Saxena, 1987).
H2O Corresponding States Corresponding States

(P ⬎ 1 kbar; T ⬎ 400 K) (P ⬎ 1 kbar; T ⬎ 400 K) (P ⬍ 1 kbar)
A ⫺0.7025 ⫹ 1.16 ⫻ 10⫺3T 1 ⫺ 0.5917Tr⫺2 1

⫹ 99.6799T ⫺1
B 0.2143T ⫺1 ⫺ 3.1423 ⫻ 10⫺14T 3 0.09122Tr⫺1 0.09827T r⫺1 ⫹ 0.2709T r⫺3
C ⫺2.249 ⫻ 10⫺6T ⫺ 0.1459T ⫺3 ⫺1.46 ⫻ 10⫺4Tr⫺2 ⫺ 2.8349 0.01472Tr⫺4 ⫺ 0.00103T r⫺1.5
⫹ 2.169 ⫻ 10⫺15T 2 ⫻ 10⫺6 ln Tr
although the modified Redlich-Kwong equation and the nonvirial equation of

Saxena-Fei are more precise than other formulations, calculations of the Redlich-
Kwong equation are more difficult to perform. For instance, to evaluate the fugac-
ity of a given gaseous phase at high P, based on
P
RT ln fi = ∫ (Vi )
− Vperfect dP + RT ln P , (9.40)
1
it is necessary to perform an integration on volumes that in the case of the

Redlich-Kwong formulation, is particularly complicated. As suggested by Gan-
guly and Saxena (1987), the Redlich-Kwong equation should be reserved for H2O,
and van der Waals-type equations should be adopted preferentially for CO2 , Ar,
N2 , and CH4 (Ross and Ree, 1980). Ganguly and Saxena (1987) also suggested
adopting the model of Halback and Chatterjee (1982) for H2O and the model of
Bottinga and Richet (1981) or Kerrick and Jacobs (1982) for CO2 at P ⬍ 1 kbar
and T ⬍ 400 K. According to these authors, the nonvirial Saxena-Fei equation
finds application under higher P and T conditions.
9.4 Mixtures of Real Gases: Examples

in the C-O-H System
The calculation of the Gibbs free energy of a gaseous mixture can be performed,
starting from the chemical potentials of the various pure gaseous components at
the P and T of interest, assuming that the mixture is ideal:
Gg = ∑ µ i0, g X i , g + G ideal mixing

(9.41)
i
G ideal mixing = RT ∑ X i , g ln X i , g . (9.42)

i
However, this approximation is not valid in geochemistry, because the ther-

mobaric regimes of interest are so wide that they prevent the assumption of ide-
Figure 9.5 Activity-concentration relationships in binary gaseous H2O-CO2 mixtures.

Experimental data from Shmulovich et al. (1982).
ality. Interactions among components in mixture cannot be neglected whenever

the confining pressure of the system is particularly high (and the corresponding T
not sufficiently high). Moreover, interactions at equal P and T conditions increase
between nonpolar-nonpolar, polar-nonpolar, and polar-polar species, respec-
tively (by “polar” we mean here a gaseous molecule with charge polarity).
Of particular interest in geochemistry are gaseous mixtures in the C-O-H
system. The main gaseous species forming are H2O, CO2 , O2 , CO, CH4 , and H2 .
Among these, only H2O exhibits charge polarity; the remaining ones are practi-
cally nonpolar (the polarity of CO is negligible for practical purposes). Figure
9.5 shows the activity diagram of the H2O-CO2 mixture, based on the experimen-
tal data of Shmulovich et al. (1982). The left part of the figure shows how, with
P held constant, interactions decrease with increasing T, and the right part shows
how, with T held constant, they increase with increasing P. The significance of
this sort of deviation from ideal behavior has already been outlined in section 9.2.
As we see in figure 9.5, the mixing properties are slightly asymmetric and
interactions are nonnegligible. To simulate the mixing properties of real gaseous
species such as those of figure 9.5, we can combine the parameters of Stockmay-
er’s potentials (see, for instance, Prausnitz et al., 1986) more or less empirically.
For instance, if we define 1 and 2 as the interacting gaseous species, parameters
R , ␴, ␧, and ␮ in equation 9.12 are obtained by algebraic and geometric means:
R 12 =
1
2
(
R 11 + R 22 ) (9.43)
σ 12 =
1
2
(
σ 11 + σ 22 ) (9.44)
( )
12
ε 12 = ε 11ε 22 (9.45)
( )
12
µ12 = µ11 µ 22 . (9.46)
Once potential U12 , with substitution of the various parameters in equation 9.12,
has been derived, an interaction parameter W12 can be obtained by applying
 

1
W12 = z U 12 − U 11 + U 22 
2 
( ) (9.47)
(Guggenheim, 1952), where z is the number of reciprocal coordinations of species

1 and 2 in the gaseous mixture.
Saxena and Fei (1988) have shown that the observed deviations from ideality in
H2O-CO2 mixtures can be obtained from interaction parameter W12 by applying
RTW12 X 1X 2
Gexcess mixing = ,
(X q 1 1 + X2 q 2 ) (9.48)
where q1 and q2 are the “efficacious volumes” of the gaseous molecules H2O and CO2 .
Because these parameters cannot be derived easily from first principles, the authors
adopted the simplifications
q1 V1
= (9.49)
q2 V2
and
q 1 + q 2 = 1, ( 9.50 )
where V1 and V2 are the volumes of molecules 1 and 2.

The activity coefficients arising from equation 9.48 are
2
 q 1X 1 
ln γ 1 = q 1W12  1 +  (9.51)
 q2 X 2 
and
2
 q X 
ln γ 2 = q 2W12  1 + 2 2  . (9.52)
 q 1X 1 
Note that these coefficients are real “activity coefficients” and not “fugacity coefficients”
(see section 9.1).
Table 9.5 Thermodynamic parameters for main gaseous species of geochemical interest.
H 0f, P r , Tr in kJ/mole: S 0P r , T r in J/(mole ⫻ K). CP ⫽ K1 ⫹ K2T ⫹ K3T 2 ⫹ K4T ⫺1/2 ⫹ K5 T ⫺2
(data from Robie et al., 1978). Tr ⫽ 298.15 K, Pr ⫽ 1 bar
Species H 0f,Pr ,Tr S 0Pr ,Tr K1 K2 ⫻ 103 K3 ⫻ 106 K4 ⫻ 10⫺2 K5 ⫻ 10⫺5

H2O ⫺241.814 188.83 7.368 27.468 ⫺4.8117 3.6174 ⫺2.2316
CO2 ⫺393.510 213.79 87.82 ⫺2.6442 0 ⫺9.9886 7.0641
CO ⫺110.530 197.67 45.73 ⫺0.097115 0 ⫺4.1469 6.6270
O2 0 205.15 48.318 ⫺0.69132 0 ⫺4.2066 4.9923
H2 0 130.68 7.4424 11.707 ⫺1.3899 4.1017 ⫺5.1041
CH4 ⫺74.810 186.26 78.976 43.186 ⫺10.598 ⫺1.3202 18.836
H 2S ⫺20.627 205.80 26.356 26.497 ⫺6.0244 0.43559 2.6599
N2 0 191.61 23.941 11.068 ⫺6.5518 0 1.9064
NH3 ⫺45.940 192.78 29.735 39.119 ⫺8.2274 ⫺1.4378 2.9243
Once the excess Gibbs free energy of mixing is known, the Gibbs free energy
of the gaseous mixture is calculated as follows:
Gg = ∑ µ i0, g X i , g + G ideal mixing + G excess mixing .

(9.53)
i
The first term implies knowledge of the chemical potentials of the gaseous com-
ponents at standard state. Thermodynamic parameters for the main gaseous spe-
cies of geochemical interest at T ⫽ 298.15 K and P ⫽ 1 bar are listed in table 9.5.
The molar volume in these T-P conditions for all species is that of an ideal gas—
i.e., 24,789.2 cm3 (or 2478.92 J/bar). The limiting T for the CP function is 1800
K. Data are from Robie et al. (1978).
9.5 Volcanic Gases
The study of volcanic gases is of fundamental importance in understanding erup-

tive processes. The evolution of these processes is primarily conditioned by the
solute states of the various volatile species in the magma, by the relationship
between bulk pressure and partial pressure of the gaseous species above the solu-
bility product, and by the (strongly exothermic) interactions between aquifers and
rising magma (phreatomagmatic eruptions).
To understand fully the relationships existing between the dynamics of erup-
tive processes and the chemistry of gaseous phases, it is essential to assess the
solubility relations of the various gaseous species in magmas. To do this, proper
thermochemical models for silicate melts must be developed, in light of the pleth-
ora of petrologic information existing on such systems, focusing attention on the
reciprocal effects of the presence of fluid phases on melt structure and of the
chemistry of the melt on the solubility product of the fluid.
Figure 9.6 Ternary atomic composition diagram for volcanic gases. Crosses: samples
from Mount Etna. From Gerlach and Nordlie (1975a), American Journal of Science, 275,
353–76. Reprinted with permission of American Journal of Science.
9.5.1 Atomic Compositions of Volcanic Gases

The atomic compositions of volcanic gases are conveniently described by the C-
O-H-S system, because these four atoms concur to form almost all the gaseous
molecules associated with magmas. Besides the above species, chlorine follows in
order of abundance, mostly in the form of hydrochloric acid HCl, often conspicu-
ous in low-T (i.e., below 600 °C) fumaroles. The amount of HCl in high-T volca-
nic gases does not generally exceed 1% in moles. Other atoms forming gaseous
molecules in volcanic emissions are F, N, and B. Fluorine is chiefly present in the
form of hydrofluoric acid HF, typically in a ratio of 1:10 with respect to HCl.
Nitrogen is mainly present as N2 and NH3 and, although bulk amounts of N
and B are extremely low, the presence of N2 is sometimes indicative of probable
atmospheric contamination. Figure 9.6 shows a ternary diagram with the atomic
species O, H, and C ⫹ S at the apexes.
As figure 9.6 shows, most gaseous emissions associated with basaltic lavas
have compositions lying along a tie line connecting stoichiometry
H : O = 2 :1 (9.54)
with stoichiometry
(C + S) : O = 2 : 1. ( 9.55)
Figure 9.7 O-H-S-C compositional tetrahedron. From Gerlach and Nordlie (1975a),
American Journal of Science, 275, 353–76. Reprinted with permission of American Journal
of Science.
This indicates that the major molecular species present in volcanic gases are H2O,
CO2 , and SO2 . Some samples (mainly from Mount Etna’s emissions in 1970) also
contain appreciable amounts of H2 molecules.
Figure 9.7 shows the O-H-S-C compositional tetrahedron. In this representa-
tion, the compositional limits of volcanic gases are composed of vertical planes
with atomic ratios
C :S = 1: 2 (9.56)
and
C : H = 1 : 1, (9.57)
respectively, the H-O-C plane, and the surfaces established by the hematite-
magnetite (HM) and quartz-fayalite-magnetite (QFM) oxygen buffers:
1
3 Fe2O3 ⇔ 2 Fe3O4 + O2 (9.58)
2
3 SiO2 + 2 Fe3O 4 ⇔ 3 Fe2SiO 4 + O2 . (9.59)
The upper surface in the compositional tetrahedron of figure 9.7 corresponds

to fO2 conditions determined by buffer equation 9.58 at T ⫽ 250 °C and represents
the highest oxidation limit observable in terrestrial volcanic gases. The surface
immediately below the upper surface is defined by the same buffering equilibrium
at T ⫽ 800 °C, and the lower surface (buffer eq. 9.59) is attained at maximum
reduction. According to Gerlach and Nordlie (1975a,b,c), this compositional
space is representative of the chemistry of all terrestrial volcanic gases.
Although, as already stated, the major molecular species present in volcanic
gases are H2O, CO2 , and SO2 , several other molecules also occur, although in
subordinate amounts. Figure 9.8 shows the complete speciation in gaseous phases
with different C/H and C/S relative amounts at T ⫽ 1250 °C, as a function of
fO2. As we can see, molecules S2 , H2 , H2S, COS, S, SO3 , CS2 , and O2 are also
present in molar proportions dependent on fO2.
Besides buffer equations 9.58 and 9.59, figure 9.8 shows the locations of
nickel–nickel oxide (NNO) and hematite-wuestite (HW) fO2-buffering equilibria:
1
Ni+ O2 ⇔ NiO (9.60)
2
1
2 FeO+ O2 ⇔ Fe2O3 . (9.61)
2
Another equilibrium that controls the chemistry and speciation of volcanic

gases is the multiple buffer involving gaseous species S2 and O2 :
Fe3O 4 + 3 S2 ⇔ 3 FeS2 + 2 O2 . (9.62)
Application of mass-action law to equation 9.62 requires knowledge of the activ-

ity of solid components magnetite and pyrite in the respective solid phases. More-
over, in order to translate partial pressures of gaseous components into activities,
the corresponding fugacity coefficients at standard state and at the state of inter-
est, and the activity coefficients in the mixture, must be evaluated (see section
9.1). Application of the buffering equilibrium of equation 9.62 to the atomic com-
positions of the gaseous emissions of Kilawea (Hawaii) (Gerlach and Nordlie,
1975c) results in a speciation state rather similar to that actually observed by
analysis (table 9.6). In calculating mass balance through equilibrium constant
K62 , the authors assumed O2 and S2 to behave as perfect gases—i.e., fO2 ⫽ PO2
and fS2 ⫽ PS2. This assumption is plausible, because T is high and bulk pressure
is low (see sections 9.1 and 9.2).
Figure 9.8 Speciation of volcanic gases with different C/H and C/S relative amounts as
a function of fO2 at T ⫽ 1250 °C. From Gerlach and Nordlie (1975a), American Journal
of Science, 275, 353–76. Reprinted with permission of American Journal of Science.
Table 9.6 Calculated speciation for gases of Alae Crater (Hawaii),

compared with analytical results of gaseous emissions of Kilawea
(Hawaii) from Gerlach and Nordlie, 1975c.
T ⫽ 800 °C, P ⫽ 1 bar, C/S ⫽ 1.5.
Alae Crater Kilawea Emissions

Gaseous Species (calculated) (measured; Nordlie, 1971)
H2O 4.8 ⫻ 10⫺1 3.1 ⫻ 10⫺1

CO2 3.1 ⫻ 10⫺1 4.1 ⫻ 10⫺1
SO2 2.0 ⫻ 10⫺1 2.7 ⫻ 10⫺1
H2S 3.6 ⫻ 10⫺3 2.3 ⫻ 10⫺3
H2 1.2 ⫻ 10⫺3 6.9 ⫻ 10⫺4
S2 2.0 ⫻ 10⫺3 2.3 ⫻ 10⫺3
CO 7.2 ⫻ 10⫺4 8.6 ⫻ 10⫺4
COS 6.9 ⫻ 10⫺5 8.9 ⫻ 10⫺5
SO3 4.3 ⫻ 10⫺8 6.6 ⫻ 10⫺8
CS2 3.5 ⫻ 10⫺9 4.4 ⫻ 10⫺9
O2 7.0 ⫻ 10⫺14 8.6 ⫻ 10⫺14
S8 1.3 ⫻ 10⫺15 2.5 ⫻ 10⫺15
9.6 Solubilities of Gaseous Species in Silicate Melts
Before examining in detail the present-day knowledge of the solubilities of various

gaseous species in magmas, it is necessary to describe solubility in terms of reac-
tivity of gaseous species with melt components. A simple heterogeneous equilib-
rium between gas and melt involves the inert gas Ar:
Ar ⇔ Ar .
gas melt (9.63)
This can be conceived as “mechanical solubility,” which does not imply either
reactivity or modifications of the molecular structure of the compound of inter-
est. Several gaseous species do not exhibit solubility of the type exemplified by
equation 9.63, but more or less complex reactions with melt components. For
instance, in the case of CO2 ,
CO 2 + O 2 − ⇔ CO 23 − .
gas melt melt
(9.64)
Equilibrium 9.64 is in fact composed of two partial reactions:
CO 2 ⇔ CO 2
gas melt
(9.65)
and
CO 2 + O 2 − ⇔ CO 23 − .
melt melt melt
(9.66)
Homogeneous equilibrium 9.66 modifies the stoichiometry, structure, and proper-

ties of the component of interest.
Generally, in the case of mechanical solubility, an extended Henry’s law field
is observed, whereas homogeneous equilibria of type 9.66 usually imply preco-
cious deviations from ideality (Battino and Clever, 1966; Wilhem et al., 1977).
There are several ways of expressing the solubilities of gaseous species in melts or
liquids. The Bunsen coefficient ␣g defines “the volume of reduced gas at T ⫽ 273.15 K,
P ⫽ 1.013 bar adsorbed by a unit volume of solvent at T of interest and P ⫽ 1.013 bar.”
The equation used to calculate the Bunsen coefficient is
 273.15 Pg  1    1.013 
α g = Vg ×    ,
1.013   Vmelt    Pg 
(9.67)
 T
where Vg is the volume of reduced gas adsorbed at the T of interest, Vmelt is the volume of
solvent (melt, in our case), and Pg is the partial pressure of the gaseous species in equilib-
rium with the solvent. Note that equation 9.67 reduces to
Vg 273.15
αg = × . (9.68)
Vmelt T
The Ostwald coefficient Lg is “the ratio between volume of adsorbed gas and volume
of solvent, measured at the same temperature”:
Vg T
Lg = = αg . (9.69)
Vmelt 273.15
Henry’s constant Kh of the gas represents the ratio between the partial pressure of the
species in the gaseous phase and the molar fraction of the species in solution:
Pg
Kh = . (9.70)
X g , melt
Note that Henry’s constant differs from equilibrium constant K, which represents the ac-
tivity ratio between the two phases at equilibrium:
a g , gas Γg Pg 1 Γg
K = = × = Kh × , (9.71)
a g , melt f g0, gas X g , melt γ g , melt fg 0, gas γ g , melt
where ⌫g is the fugacity coefficient in the gas phase and f 0g is the standard state fugacity.
In the case of homogeneous equilibria such as equilibrium 9.66, the relationship between
thermodynamic constant and Henry’s constant is obviously more complex.
In geochemistry, the solubilities of gaseous species in melts are usually expressed
through Henry’s constant or as fractional solubilities in moles (molar fraction of gaseous
species in the gas phase over molar fraction of gaseous species in the melt).
9.6.1 Solubility of H2O

The H2O molecule reacts with the bridging oxygens of the polymeric units in
silicate melts to form hydroxyl groups OH on their free terminations (Fraser,
1977):
H 2 O + O 0 ⇔ 2OH.
gas melt melt
(9.72)
Because H2O is a basic oxide (i.e., it tendentially dissociates free oxygen; cf. sec-
tion 6.1.3), reaction 9.72 may appear contradictory. However, reaction 9.72
clearly represents the summation of partial reactions:
H 2 O ⇔ H 2 O.
gas melt
(9.73)
H 2 O ⇔ 2H + + O 2 −
melt melt melt
(9.74)
O 2 − + O 0 ⇔ 2O −
melt melt melt
(9.75)
2O − + 2H+ ⇔ 2OH.
melt melt melt
(9.76)
If we consider a dimer, in a stereochemical representation, the bulk reaction of

equation 9.72 is
O− O− H O− O−
  /  
O − − Si − O − Si − O − + O ⇔ O − − Si − OH + O − − Si − OH. (9.77)
  \  
O− O− H O− O−
Contrary to common assumptions, H2O is not present in the melt completely

in the form of OH groups linked to polymer chains, but is also partially present
as discrete H2O molecules. In other words, the homogeneous reaction
Figure 9.9 Amounts of molecular H2O and OH hydroxyls determined in hydrated silicate
glasses by IR spectroscopy. Reprinted from E. Stolper, Geochimica et Cosmochimica Acta,
46, 2609–2620, copyright 䉷 1982, with kind permission from Elsevier Science Ltd., The
Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
H 2 O + O 0 ⇔ 2OH
melt melt melt
(9.78)
is not completely displaced rightward. Stolper (1982) has shown that the weight
proportions of H2O and OH in melts vary nonlinearly with increasing bulk
amount of H2O. When this exceeds 5%, molecular water is dominant. Figure 9.9
shows the proportions of OH and molecular H2O in melts with increasing
amounts of dissolved H2O, experimentally determined by infrared (IR) spectros-
copy. The observed distributions are consistent with an equilibrium constant K78
between 0.1 and 0.3.
The value of constant K78 is determined in the following manner (Stolper, 1982). The
constant of equilibrium 9.78 is given by the activity ratio
2
a OH, melt
K 78 = . (9.79)
a H 2O, melt a O 0 , melt
It is assumed, at first approximation, that the activities of the three species in homoge-
neous equilibrium correspond to the relative molar fractions
nH 2O, melt
a H 2O, melt ≡ X H 2O, melt = (9.80)
nH 2O, melt + nOH, melt + nO 0 , melt
nOH, melt
a OH, melt ≡ X OH, melt = (9.81)
Figure 9.10 Molar fractions of components H2O, OH and O0 in melt calculated by

Stolper (1982) for a value of K78 ⫽ 0.2. Calculated quantities are in good agreement with
experimental observations. Reprinted from E. Stolper, Geochimica et Cosmochimica Acta,
2609–2620, copyright 䉷 1982, with kind permission from Elsevier Science Ltd., The Bou-
levard, Langford Lane, Kidlington 0X5 1GB, UK.
nO 0 , melt
a O 0 , melt ≡ X O 0 , melt = . (9.82)
If N1 moles of component H2O mix with N2 moles of an anhydrous melt whose stoichio-
metric formula contains r moles of oxygen (i.e., r ⫽ 8 for NaAlSi3O8 , r ⫽ 2
for SiO2 , and so on), because one mole of O 0 is necessary to form two moles of OH
hydroxyls, we obtain
nO 0 , melt = rN 2 − 0.5nOH, melt ( 9.83 )
nH 2O, melt = N 1 − 0.5nOH, melt . ( 9.84 )
Substituting in equation 9.79, we obtain
2 2
X OH, melt nOH, melt
K 78 = = ,
X H 2O, melt ⋅ X O 0 , melt (N 1 )(
− 0.5nOH, melt rN 2 − 0.5nOH, melt ) (9.85)
which can be solved easily for discrete values of K78 .

Figure 9.10 shows molar fractions of H2O, OH, and O 0 calculated by Stolper (1982)
with K78 ⫽ 0.2 and plotted as a function of N1(N1 ⫹ rN2 ) (relative amount of oxygen in
melt due to H2O component with respect to total amount).
Figure 9.11 Relationship between fugacity of gaseous component H2O and amount of
same component dissolved in an albitic melt at equilibrium at T ⫽ 1100 °C (K78 ⫽ 0.2).
Interpolants in various P conditions determine value of K72, based on equation 9.87. Re-
printed from E. Stolper, Geochimica et Cosmochimica Acta, 2609–2620, copyright 䉷 1982,
with kind permission from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidling-
ton 0X5 1GB, UK.
Figure 9.11 shows the effect of pressure on the solubility of H2O in an albitic
melt at T ⫽ 1100 °C, based on the experiment of Burnham and Davis (1974). At
low H2O amounts, a marked linear correlation between fugacity of H2O in the
gaseous phase and the squared molar fraction of H2O component in the melt is
observed. Note that the abscissa axis in figure 9.11 is X 2H2Ototal, melt and thus differs
from the same axis in figure 9.9. Let us now again consider equations 9.72, 9.73,
and 9.78. Clearly, the constant of heterogeneous equilibrium 9.72 is given by the
product of partial equilibria 9.73 and 9.78—i.e.,
K 72 = K 73 × K 78 . (9.86)
Figure 9.12 Arrhenius plot of equilibria 9.72 and 9.73.
If equilibrium 9.78 is displaced rightward (i.e., the relative amount of hydroxyls

is dominant with respect to molecular H2O), it can be shown that
  2
K 72  rN 22 − N 12  X H 2O,melt
× = (9.87)
4 f H0 O,gas  N + N 2 
2 (
 1 2 ) 
f H 2O,gas
(Stolper, 1982). If the total amount of H2O in the melt (XH2Ototal, melt ) is suffi-
ciently low, it is also true that N1 ⬍⬍ N2 and the relationship between
X 2H2Ototal, melt and f H2O,gas is simply linear (cf. eq. 9.87 and figure 9.11).
The interpolation of experimental points in figure 9.11 allows constant K72 to
be determined at each pressure. This exercise, when repeated at the various T of
interest, leads to the construction of the Arrhenius plot of figure 9.12 (Fraser,
1975b). The same figure also shows the Arrhenius trend relative to equation 9.73,
calculated for K78 ⫽ 0.2 at T ⫽ 1100 °C and K78 ⫽ 0.1 at T ⫽ 700 °C (Stolper,
1982)—i.e.,
ln K 73 = ln K 72 − ln K 78 . (9.88)
Note that constant K73 is the reciprocal of Henry’s constant when fugacity co-
efficient ⌫g ⫽ 1.
Lastly, figure 9.13 shows that the linear relationship between the square root
of activity in the gaseous phase and the amount of H2O component in the melt
Figure 9.13 Relationship between activity of H2O in gaseous phase and molar amount
of H2O in melt at equilibrium. Experimental data from Burnham and Davis (1974) (䡵);
Fraser (1975b) (䊱); Kurkjian and Russel (1958) (䢇). Reprinted from B. J. Wood and D. G.
Fraser, Elementary Thermodynamics for Geologists, 1976, by permission of Oxford Univer-
sity Press.
is generally observed in all types of silicate (and also phosphate) melts (Kurkjian
and Russel, 1958; Burnham and Davis, 1974; Burnham, 1975; Fraser, 1975b,
1977).
Neglecting N1 in equation 9.87, we obtain
X H 2O,melt r 12
= K 72 . (9.89)
a 1H2O,gas 2
2
The angular coefficient of straight-line regressions in figure 9.13 thus depends on

the stoichiometry of the silicate melt component (moles of oxygen per mole of
anhydrous melt) and on the value of constant K72 .
9.6.2 Solubility of CO2 and SO2

Carbon dioxide behaves as an acidic oxide and associates free oxygen accord-
ing to
CO2 + O2 − ⇔ CO23 − .
melt melt melt
(9.90)
However, not all the CO2 is present in the melt in the form of carbonate ion:
molecular CO2 is also present, based on the heterogeneous equilibrium
Figure 9.14 Weight concentration of molecular CO2, carbonate ion CO32⫺, and total CO2
dissolved in albitic melts as a function of P, under various T conditions. Reproduced with
modifications from E. Stolper, Geochimica et Cosmochimica Acta, 46, 2609–2620, copy-
right 䉷 1982, with kind permission from Elsevier Science Ltd., The Boulevard, Langford
Lane, Kidlington 0X5 1GB, UK.
CO2 ⇔ CO2 .
gas melt
(9.91)
The complete equilibrium between gaseous phase and melt is a combination

of partial equilibria 9.90 and 9.91:
( ) ( )
CO2 + X O 2 − ⇔ X CO 23 −+ 1 − X CO2 , ( ) (9.92)
gas melt melt melt
where X and (1 ⫺ X ) are molar fractions of molecular CO2 and carbonate ion in
the melt, respectively, with respect to the total amount of CO2 . IR spectroscopy
measurements on albitic glasses (Stolper et al., 1987) show that, at constant P,
the solubility of molecular CO2 in the melt decreases with increasing T, whereas
the solubility of carbonate ion increases: the net result is that bulk solubility is
practically unaffected by temperature. This fact, based on equilibrium 9.92, can
be interpreted as increased reactivity of free oxygen in the melt, with increasing
T. As P increases, with T held constant, we observe increased solubility of carbon-
ate ion and, more appreciably, of molecular CO2 (figure 9.14).
The solubility of sulfur dioxide in melts may be ascribed to two concomi-
tant processes:
3
SO2 + O2 − ⇔ S2 − + O2
2 (9.93)
gas melt melt gas
Table 9.7 Linear equations relating

Henry’s constant of noble gases in
silicate melts to density of Liquid (␳)
(from Lux, 1987).
Kh, He ⫽ (⫺267 ⫾ 19)␳ ⫹ (765 ⫾ 49)

Kh, Ne ⫽ (⫺140 ⫾ 53)␳ ⫹ (400 ⫾ 140)
Kh, Ar ⫽ (⫺49 ⫾ 4)␳ ⫹ (137 ⫾ 10)
Kh, Kr ⫽ (⫺39 ⫾ 3)␳ ⫹ (108 ⫾ 8)
Kh, Xe ⫽ (⫺23 ⫾ 3)␳ ⫹ (63 ⫾ 7)
Figure 9.15 Solubility of noble gases in silicate melts as a function of kinetic radius of
gaseous molecule (A) and temperature (B). Ordinate axis in part A: natural logarithm of
Henry’s constant; in part B, natural logarithm of equilibrium constant. Reprinted from
G. Lux, Geochimica et Cosmochimica Acta, 51, 1549–1560, copyright 䉷 1987, with kind
1GB, UK.
and
1
SO2 + O2 + O2 − ⇔ SO24 − .
2 (9.94)
gas gas melt melt
Equilibria 9.93 and 9.94 are similar to those proposed by Fincham and Richard-
son (1954) for native sulfur. According to these equilibria, the solute states of
sulfur compounds depend on oxygen fugacity; the total concentration of sulfur
dissolved in the melt initially decreases with increasing fO2, based on equation
9.93, and then increases again when equilibrium 9.94 becomes dominant.
9.6.3 Solubility of Noble Gases

The solubility of noble gases in silicate melts obeys Henry’s law but varies mark-
edly with the chemistry and physical properties of the melt: it increases with the
Table 9.8 Standard molar enthalpies (⌬H 0i ) and standard entropies (⌬S 0i ) of solution of noble
gases in silicate melts, after Lux (1987). (1) leucite-basanite; (2) tholeiite; (3) alkali-olivine basalt.
( ⌬H i0 ) (kcal/mole) ( ⌬S i0 ) (cal/mole)
Gas (1) (2) (3) (1) (2) (3)

He 1.5 ⫾ 4.3 4.9 ⫾ 4.2 0.5 ⫾ 4.2 ⫺25.0 ⫾ 2.5 ⫺23.1 ⫾ 2.5 ⫺26.0 ⫾ 2.5
Ne 5.1 ⫾ 5.8 1.3 ⫾ 5.8 12.3 ⫾ 5.8 ⫺23.8 ⫾ 3.5 ⫺26.8 ⫾ 3.5 ⫺20.4 ⫾ 3.5
Ar 4.5 ⫾ 2.2 6.8 ⫾ 2.2 14.2 ⫾ 2.2 ⫺26.7 ⫾ 1.3 ⫺26.0 ⫾ 1.3 ⫺22.2 ⫾ 1.3
Kr 3.7 ⫾ 2.2 9.4 ⫾ 2.2 19.0 ⫾ 2.2 ⫺27.7 ⫾ 1.4 ⫺27.7 ⫾ 1.4 ⫺24.9 ⫾ 1.4
Xe 4.4 ⫾ 4.4 2.9 ⫾ 4.3 14.4 ⫾ 4.4 ⫺29.2 ⫾ 2.7 ⫺30.6 ⫾ 2.7 ⫺24.8 ⫾ 2.7
amount of SiO2 and decreases with the amounts of MgO and CaO in the melt,
increases with melt viscosity (in agreement with the chemical effect; White et al.,
1989) and decreases with increasing density. The latter fact is in agreement with
essentially “mechanical” solubility (cf. section 9.6)—i.e., solubility is a function
of the free volume available within the liquid which in turn is (approximately)
inversely proportional to density. Lux (1987) furnished the parameters of a linear
equation which reproduces the Henry’s constant of the various gases, based on
density of melt, within the field 2.1 to 2.7 g/cm3 (Table 9.7). The resulting Henry’s
constant is in cm3 of reduced volume per g of melt/atm ⫻ 10⫺5. With P, T, and
composition held constant, solubility decreases with increasing atomic number
and can be correlated directly to the kinetic radius of noble gas ri through an
empirical equation of the type
ln K h,i = a + bri2 ( 9.95)
(Kirsten, 1968), as shown in figure 9.15A. With P and composition held constant,
solubility increases slightly with T. Figure 9.15B shows an Arrhenius plot of Hen-
ry’s constant for an alkaline basalt melt in the T range 1250 ⬍ T (°C) ⬍ 1500
(Lux, 1987). From equality of potentials at equilibrium:
µ i , Pr ,T , gas = µ i , Pr ,T , melt , (9.96)
we can derive
∆H i0 ∆Si 0
ln K i = − + . (9.97)
RT R
By applying equation 9.97 to plots of the type shown in figure 9.15B, Lux (1987)
obtained standard molar enthalpies ⌬H 0i and standard molar entropies ⌬S i0 of
solution of noble gases in melts, as listed in table 9.8. Enthalpies of solution are
positive but close to zero, within error approximation.
Lastly, figure 9.16 shows the effects of pressure on the solubility of argon in
Figure 9.16 Effects of pressure on solubility of argon in silicate melts. From White et al.
(1989). Reprinted with permission of The Mineralogical Society of America.
Table 9.9 Interpolation parameters giving solubility of Ar in silicate melts as a function of P.

Standard state of reference for gaseous phase is pure gas Ar at P ⫽ 15 kbar and T ⫽ 1600 °C.
0
Concentration XAr,Pr,Tr
is adopted standard state for Ar in melt. V 0 is molar volume of Ar in melt,
assumed to be constant with T and P (from White et al., 1989). ⌬HP0 r,Tr is the molar enthalpy of
solution at P and T of reference.
Melt Pr (kbar) Tr (°C) 0

XAr,Pr ,Tr
⌬HP0 r ,Tr (kJ/mole) V 0 (cm3/mole)
NaAlSi3O8 15 1600 0.0057 ⫾ 0.0004 8.2 ⫾ 0.8 22.8 ⫾ 0.3
KAlSi3O8 15 1600 0.0067 ⫾ 0.0004 7.2 ⫾ 1.0 20.6 ⫾ 0.3
CaAl2Si2O8 15 1600 0.0008 ⫾ 0.0001 14.1 ⫾ 6.2 22.5 ⫾ 0.5
CaMgSi2O6 15 1600 0.0004 ⫾ 0.0001 14.3 ⫾ 6.5 21.5 ⫾ 0.6
K2Si4O9 15 1600 0.0079 ⫾ 0.0004 12.0 ⫾ 0.8 22.6 ⫾ 0.3
Granite 15 1600 0.0081 ⫾ 0.0009 9.3 ⫾ 0.5 23.1 ⫾ 0.3
Basalt 15 1600 0.0012 ⫾ 0.0002 12.9 ⫾ 6.0 23.7 ⫾ 0.5
silicate melts of various compositions, based on the experiments of White et al.

(1989). The interpolating curves are based on
∆H 0 1 1 V0
ln K P ,T − ln K Pr ,Tr = −
R
 − −
 T Tr  RT
P − Pr ( ) (9.98)
for a standard state of pure gas Ar at P ⫽ 15 kbar and T ⫽ 1600 °C, and assuming
Ar to mix ideally in the liquid over a matrix composed of Ar and O2⫺. The inter-
polation parameters of equation 9.98 are listed in Table 9.9.
9.7 Fluid Equilibria in Geothermal Fields
9.7.1 Solubility of Gases in Aqueous Solutions

We have already seen that the concept of the “solubility” of a gaseous species in a
liquid must be clarified by distinguishing between “purely mechanical solubility,”
attained whenever the gaseous molecule maintains its form in the liquid phase (as,
for instance, in the case of noble gases dissolved in silicate melts), and “reactive
solubility,” attained whenever the gaseous molecule reacts with the host phase,
forming stable complexes different from the gaseous form (see also the concept
of “chemical bond” in section 1.3). Figure 9.17 shows the behavior of noble gases
in liquid H2O, within the T range between 35 °C and the critical temperature at
1 bar total pressure (373 °C). In all cases, solubility (which is purely mechanical
in the case of noble gases) does not follow an Arrhenius trend but shows more
or less marked upward convex patterns, reaching a maximum at intermediate T
conditions and then decreasing toward zero, which is (necessarily) attained at the
critical temperature of the liquid. In the case of reactive gases (such as N2 in figure
Figure 9.17 Solubility of various gas-forming molecular species in water as a function of

T. Ordinate axis: Kh ⫻ 10⫺4.
Table 9.10 Constants of Potter-Clynne equation for noble gases.
Gas A0 A1 A2 A3 A4 A5
He 6830.52 ⫺187,444.82 1.974201 ⫺2738.127 ⫺687.751 889.593
Ne 1421.175 ⫺55,338.362 0.185867 ⫺516.808 ⫺20.0201 27.2899
Ar 3145.791 ⫺87,149.49 ⫺516.808 ⫺1270.795 ⫺468.632 615.388
Kr 1098.849 ⫺32,333.73 ⫺20.0201 ⫺442.515 ⫺230.154 310.802
Xe 1074.854 ⫺34,604.31 27.2899 ⫺420.535 ⫺113.399 147.401
9.17, but also O2 , H2 , CH4 , NH3 , H2S, and CO2 ), the two solubility branches of
the trend, ascending and descending, are virtually straight. Because early experi-
ments on the solubility of gaseous species in water were limited to low T (typically
below 50 °C), solubility was described by means of empirical relationships accu-
rately reproducing low-T behavior—for instance, the Benson-Krause equation:
( ) Si
( )
2
ln K h,i = A T1 /T − 1 − T1 /T − 1 , (9.99)
R
where T1 is the hypothetical temperature at which Kh,i ⫽ 1; A ⫽ ⫺36.855; and S i

is the partial molal entropy of the gaseous species in aqueous solution at T1 .
More recently, extension of experiments to high T has allowed the derivation
of empirical equations valid over the entire stability field of water. The equation
of Potter and Clynne (1978) for noble gases, for instance, is valid in the T range
298 ⬍ T (K) ⬍ 647:
   1   
2
 A1  1
K h,i = A0 + + A2T + A3 log T + A 4 log   + A5  log    
 T  ρ   ρ    (9.100)

× 10 4 × Γi .
In equation 9.100, A 0 , A1 , A2 , A3 , A4 , and A5 are empirical constants determined

for each noble gas (Table 9.10), ␳ is the density of water, and ⌫i is the fugacity
coefficient of gas with deviation parameter Zg , which can be calculated as a func-
tion of the compressibility factor, as follows (see also eq. 9.16 and 9.22 in sec-
tion 9.1):
(
ln Γi = − ∫ 1 − Z g ) dP
P
. (9.101)
0
Still more recently, Giggenbach (1980) proposed simple linear equations re-
lating the molar distribution constants of the main reactive gases to T expressed
in Celsius:
Table 9.11 Coefficients for Giggenbach

equation (eq. 9.102) for various gaseous
species in T-range 100 ⬍ T (°C) ⬍ 340
(from Giggenbach, 1980).
Species A B
NH3 1.4113 ⫺0.00292
H2S 4.0547 ⫺0.00981
CO2 4.7593 ⫺0.01092
CH4 6.0783 ⫺0.01383
H2 6.2283 ⫺0.01403
N2 6.4426 ⫺0.01416
 X i,g 
log K i′ ≡ log   = A + BT . (9.102)
 X i ,l 
These constants, listed in Table 9.11, were derived by Giggenbach (1980) from
the respective Henry’s constant by applying:
Z g , H 2O
K i′ = K h,i × , (9.103)
PH 2O
where Zg,H2O is the compressibility factor of gaseous H2O. Because the resulting
values of K⬘ vary in a simple exponential fashion with T (Figure 9.18), the maxi-
mum in Henry’s constant observed for the various noble gases in Figure 9.17 can
be simply ascribed, in light of equation 9.103, to the compressibility factor of
H2O vapor.
To understand fully the relationship between Henry’s constant and the molar distribu-
tion constant of a given species between a gas and a liquid phase, we must recall the
concept of the “compressibility factor” of a gas (see section 9.3):
PV
Z g ,i = i i
. (9.104)
ni RT
The molar fraction of gaseous species i in gaseous mixture (i ⫹ H2O(g) ) is given by
ni
X i, g = . (9.105)
n i + n H 2O ( g )
If gaseous component i is at trace concentration level, we can also write
ni ni
X i, g = ≈ . (9.106)
n i + n H 2O ( g ) n H 2O ( g )
Figure 9.18 Molar distribution constants of various gaseous species as a function of

T. Reprinted from W. F. Giggenbach, Geochimica et Cosmochimica Acta, 44, 2021–2032,
copyright 䉷 1980, with kind permission from Elsevier Science Ltd., The Boulevard, Lang-
ford Lane, Kidlington 0X5 1GB, UK.
Reexpressing ni and nH2O(g) through their respective compressibility factors, we obtain
Pi Z g ,H 2O
X i, g = × . (9.107)
PH 2O Z g ,i
Moreover, dividing by the molar fraction of species i in the liquid, we have
X i, g Pi Z g ,H 2O 1 Z g ,H 2O 1
K i′ = = × × = K h, i × × . (9.108)
X i, l X i, l Z g ,i PH 2O Z g ,i PH 2O
In the transition K⬘ → Kh,i, the compressibility factor of the trace component in the gas-
eous phase is usually neglected because, with the concentration so low, it is virtually 1 (see
eq. 9.23).
9.7.2 Compositions of Geothermal Gases

The development of a gaseous phase in geothermal fields is associated with boil-
ing of aquifers, generally induced by decompression during convective transport
toward the surface or by heat transfer from reservoir rocks exceeding the T limit
of the boiling curve at all P. Analysis of temperature profiles in deep geothermal
fields, between 0.5 and 2 km deep, almost invariantly reveals P-T conditions along
the boiling curve of water (i.e., at all depths, the temperature of the fluid is that
of the aqueous solution at the P of interest; cf. Table 9.12). There are two main
reasons for this.
Table 9.12 Thermal profile of water along boiling curve

(from Haas, 1971).
T (°C) P (bar) Depth (m) Density (g/cm3)

100 1.0 0.0 0.958
110 1.4 4.5 0.951
120 2.0 10.4 0.943
130 2.7 18.2 0.935
140 3.6 28.2 0.926
150 4.8 40.9 0.917
160 6.2 56.8 0.907
170 7.9 76.4 0.897
180 10.0 100.5 0.887
190 12.6 129.7 0.876
200 15.6 164.7 0.865
210 19.1 206.8 0.853
220 23.2 256.4 0.840
230 28.0 314.9 0.827
240 33.5 383.3 0.814
250 39.8 462.9 0.799
260 46.9 555.2 0.784
270 55.1 661.8 0.768
280 64.2 784.6 0.751
290 74.4 925.6 0.732
300 85.9 1088 0.712
310 98.7 1273 0.691
320 122.9 1487 0.667
330 128.6 1732 0.640
340 146.1 2017 0.609
350 165.4 2350 0.573
360 186.7 2746 0.525
370 210.5 3243 0.446
1. Rocks are extremely efficient heat insulators.

2. The liquid-vapor transition for H2O involves quite high evaporation
enthalpy, which controls the thermal state of the system.
At shallower depths, the vapor phase separates from the liquid phase and
migrates toward the surface, sometimes heating superficial aquifers and finally
giving rise to fumarole activity. The gaseous phase is dominated by H2O mixed
with subordinate amounts of CO2 , H2S, CH4 , H2 , N2 , and NH3 . Table 9.13 lists
representative analyses of geothermal gases from the main geothermal fields of
the world. It also lists the pressure of separation of the gaseous phase from the
liquid, the steam fraction in the fluid (FH2O, f ), and the “fraction of gas in steam”
(Xg ). The latter definition is purely operative, because H2O vapor is a gas to all
intents and purposes and the gaseous phase is not unmixed at the various P and
T of interest. The need to distinguish the total amount of non-H2O gaseous com-
Table 9.13 Gas composition in geothermal fluids (from Henley, 1984). Psep ⫽ pressure of
separation of gaseous phase. nd ⫽ not determined.
Concentration in Gas
(millimoles/mole total gases)
Locality Psep (bar) FH2O, ƒ Xg 冢mole

millimoles
of steam
冣 CO2 H 2S CH4 H2 N2 NH3
(1) 2 0.3 0.2 917 44 9 8 15 6

(2) 9.8 0.2 1.2 936 64 nd nd nd nd
(3) 11 0.19 10.04 956 18.4 11.8 1.01 8.89 4.65
(4) 2.87 0.19 24.8 945 11.7 28.1 3.0 2.1 10.2
(5) 7.6 0.289 5.88 822 79.1 39.8 28.6 5.1 23.1
(6) 8.6 0.414 2.95 932 55 4.1 3.6 1.2 4.3
(7) 20 0.135 0.248 962 29 1 2 6 nd
(8) 1 0.38 0.507 957 43.9 nd nd nd nd
(9) – 1.0 20.0 941 16 12 23 8 nd
(10) – 1.0 5.9 550 48 95 150 125 nd
(1) Wairakei, New Zealand; (2) Tauhara, New Zealand; (3) Broadlands, New Zealand; (4) Ngawha, New
Zealand; (5) Cerro Prieto, Mexico; (6) Mahio-Tongonan, Philippines; (7) Reykjanes, Iceland; (8) Salton
Sea, California; (9) Geysers, Wyoming; (10) Larderello, Italy.
ponents from gaseous H2O (i.e., “steam”) is dictated by the fact that, as the gas-
eous phase flows through superficial aquifers, further enrichment in H2O may
take place by simple evaporation.
9.7.3 Thermobarometry of Gaseous Phases

The chemical composition of the gaseous phase, in terms of minor components,
can be related to the kinetics of the process of separation from the liquid through
simple distribution laws such as
Cl 1
=
C 0 1 + Fg K ′ − 1 ( ) (9.109)
Cg K′
=
C0 (
1 + Fg K ′ − 1 ) (9.110)
[ ( )]
Cl −n
= 1 + ∆ Fg K ′ − 1 (9.111)
C0
Cg K′
=
[1 + ∆F ( K ′ − 1)] (9.112)
n
C0
g
Cl
C0
≈ exp − K ′F g( ) (9.113)
Cg
C0
(
≈ K ′ exp − K ′F g . ) (9.114)
Equations 9.109 and 9.110 are valid for a boiling process in a closed system: the
gaseous phase develops from a liquid with initial concentration C 0 , Fg is the mass
fraction of developed gas, and K ⬘ is the mass distribution constant of the compo-
nent of interest between gas and liquid. Equations 9.111 and 9.112 describe a
multistage separation process in which n is the number of separation stages, ⌬Fg
is the mass fraction of gas separated in each stage, and K ⬘ is the mean mass distri-
bution constant of the process. Equations 9.113 and 9.114 refer to a boiling pro-
cess in an open system: the gas is continuously removed from the system as the
process advances.
The following assumptions are implicitly accepted when adopting equations
9.109 to 9.114.
1. The mass distribution constant is not affected by the concentration of the

component of interest in either liquid or gas, which is a reasonable as-
sumption when the concentration is sufficiently low (Nernst’s law range).
2. The mass distribution constant does not vary appreciably with the chem-
istry of the liquid phase.
Assumption 2 may lead to substantial errors, because the salinity of the liquid
normally increases during boiling. The consequent increase in ionic strength
modifies the activity coefficients of solutes. In particular, if the gaseous species
are polar and nonpolar, the activity coefficients of solutes diverge as the process
advances (see section 8.7).
Let us now consider speciation in the gaseous phases. According to the litera-
ture, there are three main reactions controlling speciation in the gaseous phase:
CH4 + 2H2O ⇔ CO2 + 4H2

gas gas gas gas
(9.115)
2NH 3 ⇔ N 2 + 3H 2
gas gas gas
(9.116)
H2 S ⇔ S + H2 .
gas solid gas
(9.117)
Giggenbach (1980) expressed the thermodynamic constants of equilibria 9.115

and 9.116 as semilogarithmic functions of absolute T:
9323
log K 115 = 10.76 − (9.118)
T
5400
log K 116 = 11.81 − . (9.119)
T
Assuming the gaseous phase to be ideal, we can substitute partial pressures for
the thermodynamic activities of the components:
PCO2 ⋅ PH4
K 115 = 2
(9.120)
PCH 4 ⋅ PH2
2O
PN2 ⋅ PH3
K 116 = 2
2
. (9.121)
PNH
3
Substituting molar fractions for partial pressures—i.e.,
Pi = X i , g Pt , (9.122)
where Pt is total pressure, we obtain
X CO2 ⋅ X H4
K 115 = 2
⋅ Pt2 (9.123)
X CH 4 ⋅ X H2
2O
X N2 ⋅ X H3
K 116 = 2
2
⋅ Pt2 . (9.124)
X NH
3
For the evaluation of Pt, Giggenbach (1980) suggested an approximation based

on the mean solubility of the main gaseous component (usually CO2 ), because
Pt = PH 2O + ∑ Pi , (9.125)
i
where the Pi terms are the partial pressures of gaseous components other than
H2O, assuming
∑ Pi = Pt X g F g K CO
′ 2,
(9.126)
i
where Xg is the fraction of gas (non-H2O) in the system and Fg is the fraction of
vapor plus non-H2O gas in the system, one has:
PH 2O
Pt = .
(1 − X ′ 2
g F g K CO ) (9.127)
The vapor pressure along the univariant curve of pure water can be expressed as
a semilogarithmic function of the reciprocal of absolute temperature (Giggen-
bach, 1980):
2048
log PH2 O = 5.51 − . (9.128)
T
For aqueous solutions with mean saline amounts, a similar expression holds:
1820
log PH2 O = 4.90 − (9.129)
T
(Giggenbach, 1987). Rearranging equations 9.123 and 9.124 in light of equations

9.126 and 9.127, and accounting for equations 9.118, 9.119, and 9.128, we obtain
( )
T K =
5227
(
log X CH 4, g − log X CO 2, g − 4 log X H 2, g − 0.26 + 4 log 1 − X g F g K CO
′ 2 )
(9.130)
( )
T K =
1304
2 log X NH
3, g
− log X N
2, g
(
− 3 log X H2 , g + 0.79 + 2 log 1 − X g F g K CO
′ 2 )
( 9.131)
The precision of thermobarometric equations 9.130 and 9.131 (once T is

known, P is also fixed by the water-vapor univariant curve) depends on the accu-
racy of the last term on the right, which becomes more precise as the fractional
amount of gas in vapor Xg falls. Rearranging equations 9.130 and 9.131 with the
introduction of mass distribution constants of the type defined in equation 9.102,
Giggenbach (1980) transformed equations 9.130 and 9.131 into thermobaromet-
ric functions based on the chemistry of the fluid.
More recently, Chiodini and Cioni (1989) proposed a modification of equilibrium

9.115:
CH4 + 3CO2 ⇔ 4CO + 2H2O

gas gas gas liquid
(9.132)
CO2 + H2 ⇔ CO + H2O
gas gas gas liquid
(9.133)
(substituting H2O vapor for liquid H2O in the two equilibria and subtracting from equilib-
rium 9.132 equilibrium 9.133 multiplied by 4, we return to equilibrium 9.115).
The substitution of H2O vapor for liquid H2O simplifies the equilibrium. Because the
activity of liquid H2O is almost 1, we obtain
4
X CO, g
K 132 = .
X CH 4 , g ⋅ X CO
3 (9.134)
2,g
Based on thermodynamic data (Barin and Knacke, 1973), Chiodini and Cioni (1989) ob-
tained
13606
log K132 = 8.065 − , (9.135)
T
with T in K, which, combined with equation 9.134, gives
( )
T K =
log X CH
13, 606
+ 3 log X CO 2 , g − 4 log X CO, g + 8.065
.
(9.136)
4 ,g
As we see, Pt does not appear in equation 9.136 but can be calculated, based on equilib-
rium 9.133 and assuming
Pt ≈ PH 2O + PCO 2 (9.137)
2094
log Pt = 9.083 − + log XCO, g − log XH2, g (9.138)
T
(Chiodini and Cioni, 1989).
Let us now consider equilibrium 9.117. The formation of H2S obviously im-
plies the presence of sulfur in the system (but not necessarily native sulfur). The
main crystalline sulfur compound in geothermal fields is always pyrite FeS2 . Equi-
librium 9.117 can thus be rewritten as follows:
2H2S + Fe2 + ⇔ 2H+ + H2 + FeS2.

gas aqueous aqueous gas crystal
(9.139)
To be solved in thermometric fashion, equilibrium 9.139 requires knowledge of

the pH of the solution and the amount of dissolved ferrous iron.
D’Amore and Panichi (1980) proposed an empirical geothermometer based
on a modified form of the decomposition reaction of methane:
C + CO2 + 6H2 ⇔ 2CH4 + 2H2O

crystal gas gas gas liquid
(9.140)
Table 9.14 Standard partial molal properties and coefficients of HKF model for neutral molecules
0
in aqueous solution. Gf,Pr,Tr
0
and Hf,Pr,Tr
in cal/mole; SP0 r,Tr in cal/(mole ⫻ K)); VP0 r,Tr in cm3/mole
(Shock et al., 1989).
Property Ne(a) Kr(a) Rn(a) He(a) Ar(a) Xe(a)
G0f,Pr,Tr 4565 3554 2790 4658 3900 3225

0
Hf,Pr,Tr
⫺870 ⫺3650 ⫺5000 ⫺150 ⫺2870 ⫺4510
SP0 r,Tr 16.74 15.06 16.0 14.02 14.30 14.62
Cp 39.7 57.6 83.0 37.3 52.8 64.8
VP0 r,Tr 20.4 33.5 54.1 13.0 31.71 42.6
a1 4.4709 6.2721 9.0862 3.4722 6.0003 7.5196
a2 ⫻ 10⫺2 3.1352 7.5333 14.4045 0.6967 6.8698 10.5793
a3 4.5177 2.7891 0.0884 5.4762 3.0499 1.5919
a4 ⫻ 10⫺4 ⫺2.9086 ⫺3.0904 ⫺3.3745 ⫺2.8078 ⫺3.0630 ⫺3.2163
c1 27.0977 37.8226 52.5774 26.0707 35.6471 42.1036
c2 ⫻ 10⫺4 5.0523 8.6985 13.8725 4.5634 7.3160 10.1652
␻ ⫻ 10⫺5 ⫺0.2535 ⫺0.2281 ⫺0.2423 ⫺0.2123 ⫺0.3073 ⫺0.2214
Property N2(a) SO2(a) H2(a) O2(a) NH3(a) CO2(a)

0
Gf,Pr,Tr
4347 ⫺71,980 4236 3954 ⫺6383 ⫺92,250
0
Hf,Pr,Tr
⫺2495 ⫺77,194 ⫺1000 ⫺2900 ⫺19,440 ⫺98,900
SP0 r,Tr 22.9 38.7 13.8 26.04 25.77 28.1
Cp 56.0 46.6 39.9 56.0 17.9 58.1
VP0 r,Tr 33.3 38.5 25.2 30.38 24.43 32.8
a1 6.2046 6.9502 5.1427 5.7889 5.0911 6.2466
a2 ⫻ 10⫺2 7.3685 9.1890 4.7758 6.3536 2.7970 7.4711
a3 2.8539 2.1383 3.8729 3.2528 8.6248 2.8136
a4 ⫻ 10⫺4 ⫺3.0836 ⫺3.1589 ⫺2.9764 ⫺3.0417 ⫺2.8946 ⫺3.0879
c1 35.7911 31.2101 27.6251 35.3530 20.3 40.0325
c2 ⫻ 10⫺4 8.3726 6.4578 5.0930 8.3726 ⫺1.17 ⫺8.8004
␻ ⫻ 10⫺5 ⫺0.3468 ⫺0.2461 ⫺0.2090 ⫺0.3943 ⫺0.05 ⫺0.02
Property HF(a) H2S(a) B(OH)3(a) HNO3(a) SiO2(a) H3PO4(a)
0
Gf,Pr,Tr
⫺71,662 ⫺6673 ⫺231,540 ⫺24,730 ⫺199,190 ⫺273,100
0
Hf,Pr,Tr
⫺76,835 ⫺9001 ⫺256,820 ⫺45,410 ⫺209,775 ⫺307,920
SP0 r,Tr 22.5 30.0 37.0 42.7 18.0 38.0
Cp 14.0 42.7 12.0 18.0 ⫺76.1 23.6
VP0 r,Tr 12.5 34.92 39.22 – 16.1 48.1
a1 3.4753 6.5097 7.0643 – 1.9 8.2727
a2 ⫻ 10⫺2 0.7042 6.7724 8.8547 – 1.7 12.4182
a3 5.4732 5.9646 3.5844 – 20.0 0.8691
a4 ⫻ 10⫺4 ⫺2.8081 ⫺3.0590 ⫺3.1451 – ⫺2.7 ⫺3.2924
c1 14.3647 32.3 11.3568 15.2159 29.1 17.9708
c2 ⫻ 10⫺4 ⫺0.1828 4.73 ⫺0.5902 1.2635 ⫺51.2 1.7727
␻ ⫻ 10⫺5 ⫺0.0007 ⫺0.10 ⫺0.2 ⫺0.36 0.1291 ⫺0.22
and on a multiple buffer involving carbonates, oxides, sulfides, and sulfates:
1 7
CaSO4 + FeS2 + 3H2O + CO2 ⇔ CaCO3 + Fe3O 4 + 3H2S + O2 .
crystal crystal liquid gas crystal
3 crystal gas
3 gas (9.141)
The thermometric equation of D’Amore and Panichi (1980) has the form
( )
T K =
 XCH , g   XH , g 
24, 775
X 
.
2 log 4
− 6 log 2
− 3 log H2 S, g  − 7 log PCO 2 + 36.05
 XCO , g   XCO , g   XCO , g 
 2   2   2 
(9.142)
In table 9.14, for the sake of completion, we list the thermodynamic parameters
of the HKF model concerning neutral molecules in solution (Shock et al., 1989).
Calculation of partial molal properties of solutes (see section 8.11), combined
with calculation of thermodynamic properties in gaseous phases (Table 9.5),
allows rigorous estimates of the various equilibrium constants at all P and T
of interest.
PART FOUR
Methods
CHAPTER TEN
Trace Element Geochemistry
10.1 Assimilation of Trace Elements in Crystals
The presence of trace elements in minerals can be ascribed to at least four main
phenomena (McIntire, 1963):
1. Surface adsorption. Foreign ions are kept in a diffuse sheet at the surface
of the crystal, as a result of electrostatic interaction with surface atoms
whose bonds are not completely saturated.
2. Occlusion. Adsorbed impurities on the surface of the crystal are trapped
by subsequent strata during crystal accretion.
3. Substitutional solid solution with a major component. The trace element
substitutes for a major element in a regular position of the crystal lattice.
4. interstitial solid solution. Similar to the preceding phenomenon, but here
the trace element occupies an interstitial position in the crystal lattice.
The first two processes may be operative at ultratrace concentration levels. The
first one is relevant whenever the mineral has a high surface-to-mass ratio, as
in the case of colloids. The last two processes are by far the most important in
geochemistry and can be appropriately described in thermodynamic terms.
10.2 Raoult’s and Henry’s Laws
As we saw in section 3.8.1, Raoult’s law describes the properties of an ideal solu-
tion, in which the thermodynamic activity of the component is numerically equiv-
alent to its molar concentration:
a i = Xi . (10.1)
In dilute solutions, the chemical potential of the trace component is given by
/ 657
658 / Methods
Figure 10.1 (A) Activity–molar concentration plot. Trace element concentration range
is shown as a zone of constant slope where Henry’s law is obeyed. Dashed lines and ques-
tion marks at high dilution: in some circumstances Henry’s law has a limit also toward
infinite dilution. The intercept of Henry’s law slope with ordinate axis defines Henry’s law
standard state chemical potential. (B) Deviations from Nernst’s law behavior in a logarith-
mic plot of normalized trace/carrier distribution between solid phase s and ideal aqueous
solution aq. Reproduced with modifications from Iiyama (1974), Bullettin de la Societée
Francaise de Mineralogie et Cristallographie, 97, 143–151, by permission from Masson
S.A., Paris, France. ⌬ in part A and log ⌬ in part B have the same significance, because
both represent the result of deviations from Henry’s law behavior in solid.
µ i = µ i0 + RT ln γ i X i , (10.2)
where ␥i is the activity coefficient deriving from solute-solvent interactions. In the

concentration field of validity of Henry’s law, activity coefficient ␥h,i is constant,
owing to the fact that, although a solute-solvent interaction exists, it is not
affected by the solute concentration, which is too low to influence the thermo-
chemical properties of the solid. We can thus embody the interaction effects at
concentration levels within the Henry’s law field into a particular standard state
condition called “Henry’s law standard state,” which gives
µ i = µ i∗ + RT ln Xi . (10.3)
The relationship between the Henry’s law activity coefficient of component i

␥h,i and the Henry’s law standard state potential ␮i* is then
 µ∗ − µ0   ∆µ ∗ 
γ h,i = exp  i i
 ≡ exp 
i
. (10.4)
 RT   RT 
   
Trace Element Geochemistry / 659
The excess Gibbs free energy term ⌬␮i* corresponds to the amount of energy
required to transfer one mole of trace component i from an ideal solid solution
to the solid solution of interest. As shown in figure 10.1, the more ␮i* differs from
␮0i , the more angular coefficient ␥h,i differs from 1, which represents the condition
of Raoult’s law. Trace element distribution between phases of geochemical interest
is governed by Henry’s law only within restricted solid solution ranges (here we
use the term “solution” instead of “mixture,” because the Henry’s law standard
state of the trace element differs from the standard state adopted for the carrier,
which is normally that of pure solvent at the P and T of interest: cf. section 2.1).
10.3 High Concentration Limits for Henry’s Law

in Crystalline Solutions
Let us consider the distribution of a trace element A and a carrier B between a

crystal and an ideal aqueous solution at equilibrium. Defining
XA
Xs = (10.5)
XB
the molar ratio in the solid and
mA
Xaq = (10.6)
mB
as the molality ratio in aqueous solution, the equilibrium distribution for two
ideal solutions is given by
X aq = K ( P,T ) Xs , (10.7)
where K(P,T) is the thermodynamic constant valid at given P-T conditions. If

solute-solvent interactions in the solid are embodied in the definition of Henry’s
law standard state, equation 10.7 is also valid within the concentration limits of
Henry’s law in the solid. In a logarithmic plot of concentrations (figure 10.1B),
the validity zone of equation 10.7 is represented by a straight line with an angular
coefficient of 1, or “Nernst’s law range,” and the intercept of its prolongation on
the ordinate axis at log Xs ⫽ 0 defines the value of K(P,T). Deviations from Henry’s
law behavior in solid are functions of the trace element concentration in the sys-
tem and can be expressed as
(
log ∆ = log X aq − log K + log Xs . ) (10.8)
660 / Methods
If the solid solution is regular, with an interaction parameter W (cf. section

3.8.4), the equilibrium distribution curve is defined by
(
ln X aq = ln K + ln X s + 1 − 2X A,s ) RT
W
(10.9)
(Iiyama, 1974). Posing K ⫹ W/RT ⫽ K⬘ and transforming the logarithms into

common logarithms, we obtain
2W
log Xaq = log K′ + log Xs − XA, s log e . (10.10)
RT
Adopting equation 10.10, Iiyama (1974) calculated values of W necessary to

justify observed deviations from Henry’s law behavior for trace elements in sili-
cates. The resulting theoretical equilibrium distributions are shown in Figure
10.2A and B. As pointed out by the author, deviations from linearity in log-log
distributions observed at molar concentration levels around 10⫺3 to 10⫺2 (as fre-
quently happens in mineral-solution hydrothermal equilibria; see, for example,
Roux, 1971a,b; Iiyama, 1972, 1974; Volfinger, 1976) imply extremely high values
of W—i.e., strong interactions between solute and solvent in the crystal structure.
Alternative ways of accounting for precocious deviations from ideality in
solid-liquid distributions involve consideration of configurational entropy effects
in the solid solution process. Two of these are the “two ideal sites” model (Matsui
and Banno, 1962; Grover and Orville, 1969) and the “local lattice distortion”
model (Iiyama, 1974).
10.3.1 Two Ideal Sites Model

If substitution of a trace element for a carrier takes place in two distinct lattice
positions in the crystal, the distribution equation
X aq = KX s (10.11)
is replaced by
X I,s = K I X aq (10.12)
X II,s = K II X aq , (10.13)
where subscripts I and II identify magnitudes relative to sites of types I and II in

the crystal. If we call m and n, respectively, the relative proportions of sites of
Figure 10.2 Deviations from Nernst’s law in crystal–aqueous solution equilibria, as ob-
tained from application of various thermodynamic models. (A and B) Regular solution
(Iiyama, 1974). (C) Two ideal sites model (Roux, 1971a). (D) Model of local lattice distor-
tion (Iiyama, 1974). Reprinted from Ottonello (1983), with kind permission of Theophras-
tus Publishing and Proprietary Co.
types I and II in the crystal (m ⫹ n ⫽ 1), the relationship between partial distribu-
tion constants KI, KII, and K is
mK I + nK II + X aq K I K II
K = .
(
1 + X aq nK I + mK II ) (10.14)
Eliminating Xaq from equations 10.11 and 10.14, we obtain an expression of

K as a function of Ks:
662 / Methods
Figure 10.3 Two ideal sites model. Normalized Rb/Na distribution between nepheline
(Xs) and hydrothermal solution (Xaq) (data from Roux, 1974).
 12 
1
(
K 0 1 − aXs +  1 − aXs ) + bXs 
2
K = , (10.15)
 
2  
where
K 0 = mK I + nK II (10.16 )
mK II + nK I
a= (10.17 )
K0
K I K II
b= . (10.18)
K 02
Values of equation 10.15 are plotted in figure 10.2C for various KII /KI ratios
between the limiting cases:
K II
K = K 0 when =1 (10.19 )
KI
and
 n  K
K = K 0  1 − X s  when II = 0 . (10.20)
 m  KI
As we see, the more KI differs from KII and the more Xs approaches m/n, the more
marked deviation from ideality becomes.
It was shown by Roux (1974) that the two ideal sites model applies perfectly
to partitioning of Rb distribution between nepheline and hydrothermal solutions.
Based on the experimental work of Roux (1974), deviation from ideality in the
normalized Rb/Na distribution between nepheline (Xs ) and hydrothermal solu-
tion (Xaq ) is detectable for Xs values higher than 10⫺2 (figure 10.3A) Distribution
constant K in the Nernst’s law range is K 0 ⫽ 0.82 (Roux, 1974), and the modifica-
tion of K with increasing Xs is well described by equation 10.20, within experi-
mental approximation (figure 10.3B).
It may be noted in figure 10.3B that prolongation of the interpolant encoun-
ters the abscissa axis at Xs ⫽ 13, which is exactly the proportion of the two struc-
tural sites where Rb/Na substitutions take place in nepheline. According to figure
10.3B and equation 10.20, it is plausible to admit that Rb is fixed almost exclu-
sively in the larger of the two structural sites.
10.3.2 Local Lattice Distortion Model

This model (Iiyama, 1974) can be applied whenever trace-carrier substitution
takes place in a single site within the crystal. According to this model, when
trace element A substitutes for carrier B in a given lattice position, it causes local
deformation of the structure in the neighborhood of the site where substitution
takes place. This deformation, which increases as the crystal-chemical properties
of trace element and carrier differ, prevents further substitutions from occurring
in the vicinity of the site. A “forbidden zone” of r sites is thus created, clearly
modifying the configurational properties of the phase.
In a crystal containing N sites occupied by B and A atoms, the number of possible

configurations as a function of dimensions of the forbidden zone (in terms of r site posi-
tions) is given by
Q=
[ (
N N − r +1 )] [ N − 2(r + 1)] L [ N − ( N A )(
−1 r +1)]
(10.21)
NA !
(cf. section 3.8.1). The entropy of mixing is
S mixing = k ln Q, ( 10.22 )
where k is Boltzmann’s constant. Applying the equality
N B = N − NA ( 10.23 )
and the Stirling approximation

664 / Methods
(
ln N ! = N ln N − 1 , ) ( 10.24 )
the entropy of mixing is
 NA  N B − rNA   N B − rNA  
S mixing = − k  NA ln +  ln  . (10.25)
 N  r +1   N 
Considering an athermal solution (Hmixing ⫽ 0 and Gmixing ⫽ ⫺T ⫻ Smixing; see section

3.8.3) composed of nA moles of AM and nB moles of BM, where M is the common anionic
group, we obtain
  n A   nB − rn A   nB − rn A  
Gmixing = RT n A ln   +  ln  . (10.26)
  n A + nB   r + 1   n A + nB  
The excess Gibbs free energy of mixing is
Gexcess mixing = G mixing − Gideal mixing , ( 10.27 )
where Gideal mixing is the Gibbs free energy of mixing when r ⫽ 0:
 nB − rn A   nB − rn A   nB  
Gexcess mixing = RT   ln   − nB ln  . (10.28)
  r + 1   n A + nB   n A + nB  
Differentiating equation 10.28 with respect to nA,we obtain the activity coefficient
of component AM in the solid solution (BX ,A1⫺X )M:
 ∂ G excess mixing  −r
ln γ A = 
∂ nA  = (
ln X B − X A , ) (10.29)
  r +1
where
NA nA
XA = = (10.30)
N n A + nB
and
NB nB
XB = = . (10.31)
N nA + nB
Figure 10.2D shows normalized distributions of elements A and B for various

values of r. The more the forbidden zone (in terms of r sites) is large, the more
precocious are deviations from ideality. Although the approximation to athermal
solutions is not strictly valid in the case of trace elements in crystals, solid/liquid
Figure 10.4 Applications of the local lattice distortion model to mineral-hydrothermal

solution equilibria. Experimental data from Iiyama (1972) for Ba/K and Sr/K, Volfinger
(1970) for Li/K, and Roux (1971b) for Rb/Na. Reprinted from Iiyama (1974), Bullettin de
la Societée Francaise de Mineralogie et Cristallographie, 97, 143–151, with permission from
Masson S.A., Paris, France.
distributions calculated with the local lattice distortion model reproduce experi-
mental evidence fairly well. As described by Iiyama (1974), this occurs because
any enthalpic contribution to the Gibbs free energy of mixing in the solid phase
(Hmixing ), arising from solute-solvent interactions, varies linearly with XA in the
Henry’s law concentration field.
Figure 10.4 shows normalized Ba/K and Sr/K distributions between sanidine
and a hydrothermal solution (Iiyama, 1972), Li/K between muscovite and a hy-
drothermal solution (Voltinger, 1970), and Rb/Na between nepheline and a hy-
drothermal solution (Roux, 1971b), interpreted through the local lattice distor-
tion model, by an appropriate choice of the Nernst’s law mass distribution
constant K and the lattice distortion propagation factor r.
Table 10.1 lists r values calculated for normalized distributions of alkalis and
alkaline earths between silicates and hydrothermal solutions and for normalized
distributions of rare earth elements (REE) between pyroxenes and silicate melts.
666 / Methods
Table 10.1 Linear propagation of lattice distortion (in terms

of number of sites r) calculated for silicate/hydrothermal
solutions (liyama, 1974) and silicate/melt equilibria (Ottonello
et al., 1978).
Mineral Trace Element Carrier r

Muscovite Li K 2.45
Sanidine Sr K 5.65
Albite Ba K 2.15
Albite Sr Na 3.22
Nepheline Rb Na 7.00
Clinopyroxene Nd Ca 10.00
Clinopyroxene Ce Ca 10.00
Clinopyroxene Lu Ca 75.00
Orthopyroxene Lu Mg 70.00
Orthopyroxene La Mg 165.00
Table 10.2 lists upper concentration limits for Henry’s law behavior of various
trace elements in silicate crystals for specific P and T conditions. In some cases,th-
ese limits are only approximate or conjectural. The table is not exhaustive, be-
cause it reflects the state of the art in the early 1980s and is subject to substantial
modifications in response to the advance of experimental knowledge since that
time.
10.4 Lower Concentration Limits for Henry’s Law

Behavior in Silicate Crystals
Henry’s law in crystals has low concentration limits (figure 10.1A), ascribable to
modifications in the solid solution process, which, at high dilution, is greatly
affected by the intrinsic and extrinsic defects in the crystal (Wriedt and Darken,
1965; Morlotti and Ottonello, 1982). Figure 10.5 shows some experimental evi-
dence concerning solid solutions of REE in silicates. Deviations from Nernst’s
law observed in figure 10.5 may be ascribed to defect equilibria between REE and
cationic vacancies present in the solid (De Vore, 1955; Wood, 1976c; Harrison
and Wood, 1980; Morlotti and Ottonello, 1982; see also chapter 4).
In logarithmic plots of simple weight concentrations, as in figure 10.5, the
Nernst’s law region appears as a straight line with an angular coefficient of 1. At
the low concentration limits for Henry’s law behavior in the solid, we observe a
sudden change in slope, and, at high dilution, the new distribution has an angular
coefficient lower than 1 (i.e., ␣ ⬍ 45°). Both the slope rupture point and the
angular coefficient valid at high dilution are dictated by the extrinsic defects in
the solid (Morlotti and Ottonello, 1982).
Correct evaluation of solution mechanisms of a given trace element in silicate
Table 10.2 Upper concentration limits for Henry’s law behavior in silicates and oxides.
Abbreviations: Ab ⫽ albite; Sa ⫽ sanidine; Pl ⫽ plagioclase; Fel ⫽ alkaline feldspar; Rb-fel ⫽
Rb-feldspar; Ne ⫽ nepheline; Mu ⫽ muscovite; Ol ⫽ olivine; Di ⫽ diopside; Cpx ⫽ clinopyroxene;
Oxp ⫽ orthyporoxene; Amph ⫽ amphibole; Par ⫽ pargasite; Gr ⫽ garnet; Ilm ⫽ ilmenite;
Arm ⫽ armalcolite (from Ottonello, 1983).
Equilibrium T (°C) P (bar) Trace Carrier Henry’s Law Limit Reference

Ab–hydrothermal solution 600 2000 Rb Na (Rb/Na)Ab ⫽ 3 ⫻ 10⫺4 1
600 1000 Rb Na (Rb)Ab ⫽ 33 ppm 6
600 880 Rb Na (Rb)Ab ⫽ 1600 ppm 3
600 2000 Cs Na (Cs/Na)Ab ⫽ 3 ⫻ 10⫺5 1
600 1000 Cs Na (Cs)Ab ⫽ 5 ppm 6
600 880 Cs Na (Cs)Ab ⫽ 500 ppm 3
600 1000 Sr Na (Sr)Ab ⫽ 1340 ppm 6
600 1000 Ca Na (Ca)Ab ⫽ 914 ppm 6
600 880 K Na (K)Ab ⫽ 150 ppm 3
Sa–hydrothermal solution 800 1000 Cs K (Cs/K)Sa ⫽ 10⫺2 7
700 1000 Cs K (Cs)Sa ⫽ 60 ppm 8
600 1000 Cs K (Cs)Sa ⫽ 716 ppm 6
500 1000 Cs K (Cs/K)Sa ⫽ 5 ⫻ 10⫺3 7
400 1000 Cs K (Cs/K)Sa ⫽ 10⫺2 9
600 800 Cs K (Cs)Sa ⫽ 480 ppm 3
600 1000 Rb K (Rb)Sa ⫽ 32,300 ppm 6
600 800 Rb K (Rb)Sa 艐 10,000 ppm 3
600 1000 Ba K (Ba)Sa ⫽ 4920 ppm 6
600 1000 Sr K (Sr)Sa ⫽ 157 ppm 6
600 1000 Na K (Na)Sa ⫽ 8300 ppm 6
600 800 Na K (Na)Sa ⫽ 8500 ppm 3
600 1000 Ca K (Ca)Sa ⫽ 72 ppm 6
Pl–melt 1190 1 Sm Ca (Sm)Pl ⫽ 3800 ppm 4
1190 1 Sm Ca (Sm)Pl 艐 5000 ppm 5
1190 1 Ba Ca (Ba)Pl ⫽ 14,500 ppm 4
1190 1 Sr Ca (Sr)Pl ⫽ 70,200 ppm 4
Fel–melt 780 8000 Rb K (Rb)Fel 艐 4000 ppm 10
780 8000 Sr K?Ca? (Sr)Fel 艐 3000 ppm 10
780 8000 Ba K?Ca? (Ba)Fel 艐 40,000 ppm 10
Rb-fel–hydrothermal solution 600 800 Na Rb (Na)Rb-fel ⫽ 4000 ppm 3
600 800 K Rb (K)Rb-fel ⫽ 12,000 ppm 3
600 800 Cs Rb (Cs)Rb-fel ⫽ 400 ppm 3
Ne–hydrothermal solution 600 2000 Rb Na (Rb/Na)Ne ⫽ 10⫺3 1
600 2000 Cs Na (Cs/Na)Ne ⫽ 10⫺4 1
Mu–hydrothermal solution 600 1000 Li K (Li/K)Mu ⫽ 7 ⫻ 10⫺3 11
400 1000 Cs K (Cs/K)Mu ⫽ 10⫺4 9
Ol–melt 1184 1 Ni Mg (Ni)Ol 艐 47,000 ppm 12
1350 1 Ni Mg (Ni)Ol 艐 18,000 ppm 12
1025 20,000 Ni Mg (Ni)Ol 艐 700 ppm* 13
1075 10,000 Ni Mg (Ni)Ol 艐 1000 ppm* 13
1025 20,000 Sm ? (Sm)Ol 艐 0.7 ppm? 14
continued
668 / Methods
Equilibrium T (°C) P (bar) Trace Carrier Henry’s Law Limit Reference
Di–melt 1300 1 Ni Mg (Ni)Di 艐 80,000 ppm 15

1300 1 Co Mg (Co)Di ⫽ 30,000 ppm 15
Cpx–melt 1025 20,000 Ni Mg (Ni)Cpx 艐 20 ppm 16
950 10,000 Sm Ca (Sm)Cpx ⫽ 17 ppm 17
Opx–melt 1025 20,000 Ni Mg (Ni)Opx 艐 100 ppm? 16
1075 10,000 Sm Mg (Sm)Opx ⫽ 1 ppm 14
Amph–melt 1000 15,000 Ni Mg (Ni)Amph 艐 400 ppm? 16
Par–melt 1000 15,000 Sm unknown (Sm)Par ⫽ 1.8 ppm 17
Gr–melt 1025 20,000 Ni unknown (Ni)Gr 艐 50 ppm 13
950 20,000 Sm unknown (Sm)Gr ⫽ 8 ppm 18
Ilm–melt 1128 1 Zr unknown (Zr)Ilm ⫽ 0.5 ppm 19
1127 1 Nb unknown (Nb)Ilm ⫽ 1.5 ppm 19
Arm–melt 1127 1 Zr unknown (Zr)Arm ⫽ 2 ppm 19
1127 1 Nb unknown (Nb)Arm ⫽ 2.5 ppm 19
(1) Roux (1971a) (2)Iiyama (1972) (3) Lagache and Sabatier (1973)
(4) Drake and Weill (1975) (5) Drake and Holloway (1978) (6) Iiyama (1973)
(7) Eugster (1954) (8) Lagache (1971) (9) Volfinger (1976)
(10) Long (1978) (11) Volfinger (1970) (12) Leeman and Lindstrom (1978)
(13) Mysen (1976a) (14) Mysen (1976b) (15) Lindstrom and Weill (1978)
(16) Mysen (1977a) (17) Mysen (1978) (18) Mysen (1977b)
(19) McCallum and Charette (1978)
艐 Approximated values.
? Suggested values.
* Contrasting evidence (see Drake and Holloway, 1978).
crystals thus requires knowledge of the defects in the phases. As we saw in chapter
4, defect equilibria in crystals are conveniently described by referring to the f O2
condition. For instance, oxygen deficiency may generate in the crystal doubly
ionized Schottky vacancies VO·· or interstitials Mi·· wheras oxygen excess may result
||
in cationic vacancies V M. Intrinsic disorder is also always present under intrinsic
oxygen partial pressure conditions P* O2. Table 10.3 shows the energetically plausi-
ble defect equilibria that may affect the distribution of trivalent REE between
pyrope Mg3Al2Si3O12 and silicate melt (see Kröger, 1964, for exhaustive treatment
of the defect state of crystals; see also chapter 4 for the significance of the adopted
symbols). The proposed equilibria, which refer to the various f O2 conditions and
trace element concentrations of interest, can be adapted to other silicate crystals
with opportune modifications in the “carrier” notations (Mg, in this case). All
the equilibria ruling the REE distribution within the various concentration ranges
are then translated into a bulk distribution equation whose single terms on the
right each apply to a specific concentration range. For instance, constant K4 in
Figure 10.5 Experimentally observed deviations from Nernst’s law behavior in solid/melt
REE distribution. (A) Weight distribution of Sm and Tm between garnet and melt, experi-
mental sources specified in figure. (B) Weight distribution of Sm between clinopyroxene
and silicate melt. (C) Weight distribution of Sm and Tm between plagioclase and silicate
melt. Systems are An55Di45 and Ab60An30Di10 at T ⫽ 1300 °C and P ⫽ 1 bar. From R.
Morlotti and G. Ottonello, Solution of rare earth elements in silicate solid phases: Henry’s
law revisited in light of defect chemistry, Physics and Chemistry of Minerals, 8, 87–97,
figures 2–3, 1982, copyright 䉷 1982 by Springer Verlag. Reprinted with the permission of
Springer-Verlag GmbH & Co. KG.
table 10.5, relative to equilibrium (4) in the same table, describes defect process
involving preexisting intrinsic defects, whereas constant K7 describes the creation
of extrinsic defects that take place when intrinsic defects are saturated (i.e., at
higher concentrations).
Deviations from Nernst’s law in log-distribution diagrams such as figure 10.5
are represented by the simple equation:
X∗ 1 X∗
ln γ A = ln  A  − 
A
, (10.32)
 X A  tan α  XA 
   
where XA is the trace element concentration in the solid, X A * is the lower concen-
tration limit for Nernst’s distribution, and tan ␣ is the slope of non-Nernstian
distribution at high dilution (see table 10.4). Equation 10.32 can be interpreted
in light of the defect equilibria listed in table 10.3, assuming that slope rupture
670 / Methods
Table 10.3 Defect equilibria affecting REE solubility in pyrope and their effects in bulk garnet/melt
REE distribution (Morlotti and Ottonello, 1982)
A: Oxygen fugacity corresponding to the intrinsic oxygen partial pressure: ƒ O2 ⫽ P O

*
2
Defect equilibrium: ⫻
MgMg ⇔ Mg··
i ⫹ VMg
||
(1)
Equilibrium constant: KF ⫽ [Mg·· ||

i ] [VMg]* (2)
Defect concentration: [VMg]* ⫽ KF1/2

||
(3)
Relative concentrations Corresponding equilibria and their constants
[REE·Mg] ⬍ ⬍ [V||Mg]* REE(l3⫹) ⫹ V||Mg ⇔ REE·Mg (4)
KD ⫽ [REE·Mg] / [REE(l3⫹) ] ⫽ K4KF1/2 (5)
[REE·Mg] ⱕ [VMg]*
||
KD ⫽ K4KF1/2 ⫺ K4 [REE·Mg] (6)
3 ⫻ 3 1
[REE·Mg] ⱖ [V||Mg]* REE(l3⫹) ⫹ MgMg ⇔ Mg(l2⫹) ⫹ REE·Mg ⫹ V||Mg (7)
2 2 2
KD ⫽ K7[REE·Mg]⫺1/2 (8)
[REE·Mg] ⱖ [V||Mg]* REE(l3⫹) ⇔ REE·Mg ⫹ e⫺ (9)

⫺1
KD ⫽ K9[REE·Mg] (10)
⫻ ⫻
[REE·Mg] ⬎ ⬎ [V||Mg]* REE(l3⫹) ⫹ Al(l3⫹) ⫹ MgMg ⫹ SiSi ⇔ REE·Mg ⫹ Al|Si ⫹ Mg(l2⫹) ⫹ Si(l
4⫹
) (11)
⫺1 ⫺1
KD ⫽ K11KD, AlKD, MgKD, Si (12)
Bulk distribution:
KD ⫽ K4KF1/2 ⫺ K4 [REE·Mg] ⫹ K7[REE·Mg]⫺1/2 ⫹ K9[REE·Mg]⫺1 ⫹ K11KD, AlKD,

⫺1 ⫺1
MgKD, Si (13)
B: Oxygen fugacity higher than the intrinsic oxygen partial pressure: ƒO2 ⬎ P*O2
1 ⫻
Defect equilibrium: O2 ⇔ OO ⫹ V||Mg ⫹ 2h· (14)
2 (gas)
⫺1/2
Equilibrium constant: K14 ⫹ 4[V||Mg]3 ƒO 2
(15)
⫺1/3
Defect concentration: [VMg] ⫽ [VMg]* ⫹ 4
|| || 1/3 1/6
K14 ƒO2 (16)
Bulk distribution:
KD ⫽ K4KF1/2 ⫺ K4 [REE·Mg] ⫹ 4⫺1/3 K4K14 ƒO2 ⫹ K7[REE·Mg]⫺1/2 ⫹ K9[REE·Mg]⫺1 ⫹ K11KD, AlKD,

1/3 1/6 ⫺1 ⫺1
MgKD, Si
(17)
C: Oxygen fugacity lower than the intrinsic oxygen partial pressure: ƒO2 ⬍ P*O2
1 ⫺
Defect equilibrium: OO⫻ ⇔ O2(gas) ⫹ V ··
O ⫹ ⫹2e (18)
2
Equilibrium constant: K18 ⫽ 4[V ·· 3 1/2
O] ƒO2 (19)
⫺1/3 1/3 ⫺1/6
Defect concentration: [VMg] ⫽ [V Mg]* ⫺ 4
|| ||
K18 ƒO2 (20)
Bulk distribution:
KD ⫽ K4KF1/2 ⫺ K4 [REE·Mg] ⫺ 4⫺1/3 K4K18 ƒO2 ⫹ K7[REE·Mg]⫺1/2 ⫹ K9[REE·Mg]⫺1 ⫹ K11K4KD, AlKD,

1/3 1/6 ⫺1 ⫺1
MgKD, Si
(21)
Table 10.4 Least-squares fits of type ln Xsolid ⫽ a ln Xliquid ⫹ b relating REE concentrations in silicate crystals and melts, as derived from existing
experimental data. Corresponding values of R2 correlation parameter are also listed (from Morlotti and Ottonello, 1982).
Low Concentration High Concentration
Element Solid T (°C) P (bar) a b ␣ R2 a b ␣ R2

Ce Grossular 1300 30,000 0.8444 0.4091 40.18 0.9998 1.0086 ⫺0.4544 45.25 0.9998
Ce Pyrope 1300 30,000 0.7159 ⫺1.6725 35.60 0.99998 0.9827 ⫺2.6263 44.50 1
Sm Grossular 1300 30,000 0.8305 0.5522 39.71 0.9986 0.9813 ⫺0.1343 44.46 0.9999
Sm Pyrope 1300 30,000 0.8010 ⫺1.3151 38.70 0.9986 1.0086 ⫺2.2824 45.25 0.99998
Sm Pyrope 1500 30,000 0.7079 ⫺0.7346 35.29 0.99998 0.9712 ⫺1.6399 44.16 0.9997
Sm Garnet s.s. 950 20,000 – – 0.9469 1.21 43.44 0.991
Tm Grossular 1300 30,000 0.9387 1.8198 43.19 0.9997 0.9878 1.6158 44.65 0.99999
Tm Pyrope 1300 30,000 0.8535 ⫺1.1110 40.48 0.9991 0.9537 ⫺1.4400 43.64 0.9994
Sm Clinopyroxene 950 20,000 0.786 ⫺0.125 38.17 0.997 0.837 ⫺0.249 39.93 0.994
Sm Diopside 1400 30,000 0.8385 ⫺1.177 39.98 0.990 0.9181 ⫺1.442 42.56 0.9997
Sm Anorthite 1300 1 0.2745 ⫺0.4857 15.35 0.996 1.0389 ⫺3.046 46.09 0.997
Sm Plagioclase 1300 1 0.1931 ⫺0.224 10.92 0.68 0.964 ⫺2.379 43.95 0.998
Tm Anorthite 1300 1 0.580 ⫺1.786 30.11 0.94 0.975 ⫺3.284 44.27 0.999
Tm Plagioclase 1300 1 – – 0.976 ⫺2.219 44.30 0.998
672 / Methods
points X A
* represent saturation of intrinsic and extrinsic defects. In the case of
samarium, for instance,
2  ∗
∗ =
X Sm
3
[ ]
V Mg (10.33)
and
 2  ∗ 
X Sm =  V Mg
3
[ ] [ ]
− V Mg

.

(10.34)
Equation 10.32 can thus be rewritten in the form
 
 1 − tan α   3X Sm 
ln γ Sm =  ln  . (10.35)
 tan α   ∗
 [ ] 
2 V Mg

10.5 Fractionation Operated by Crystalline Solids:

Onuma Diagrams
As outlined in section 10.1, the presence of trace elements in crystals is attribut-

able to several processes, the most important one being the formation of substitu-
tional solid solutions. The “ease” of substitution depends on the magnitude of
interactions between trace element and carrier. We have already seen (section
3.8.4) that macroscopic interaction parameter W can be related to microscopic
interactions in a regular solution of the zeroth principle:
 2w − wAA − w BB 
W = N 0  AB . (10.36)
 2 
In qualitative terms, microscopic interactions are caused by differences in crystal

chemical properties of trace element and carrier, such as ionic radius, formal
charge, or polarizability. This type of reasoning led Onuma et al. (1968) to con-
struct semilogarithmic plots of conventional mass distribution coefficients K⬘ of
various trace elements in mineral/melt pairs against the ionic radius of the trace
element in the appropriate coordination state with the ligands. An example of
such diagrams is shown in figure 10.6.
The arrangement of elements of the same charge gives rise to more or less
smooth curves with maxima located at “optimum site dimensions” for trace-
Figure 10.6 Onuma diagrams for crystal/melt trace element distributions. Ionic radii of
Whittaker and Muntus (1970). (A) Augite/matrix distribution, data of Onuma et al.
(1968). (B) Plagioclase/matrix distribution, data of Dudas et al. (1971), Ewart and Taylor
(1969), and Philpotts and Schnetzler (1970). Reprinted from B. Jensen, Geochimica et Cos-
mochimica Acta, 37, 2227–2242, copyright 䉷 1973, with kind permission from Elsevier
Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
carrier substitutions in the crystal of interest (Onuma et al., 1968; Jensen, 1973).
Figure 10.6A, for instance, shows the mass distribution of various trace elements
between augitic pyroxene and the silicate matrix (assumed to represent the silicate
melt), based on the data of Onuma et al. (1968). There are two maxima, at 0.79
and 1.01 Å, interpreted respectively as “optimum dimensions” of sites M1 and
M2 in the crystal (Jensen, 1973). The peaks are more clearly defined for divalent
ions, and the partition coefficients decrease more rapidly for elements with higher
formal charges, departing from the optimum dimension. Figure 10.3B shows an
Onuma diagram focused on the mass distribution of alkaline earths between pla-
gioclase and the matrix. The distribution curves indicate the marked composi-
tional control exerted by the solid: because the carrier is Ca, the mass distribution
coefficients progressively increase with the amount of anorthite component in
solid solution, whereas fractionation is not particularly affected. The main advan-
tage of Onuma diagrams is that they can elucidate the marked fractionation in-
duced by the crystal structure in crystal/melt equilibria. However, their interpreta-
674 / Methods
tion is merely qualitative and does not currently allow translation of the observed
phenomena into solution models for the solid phases of interest.
10.6 Assimilation of Trace Elements in Silicate Melts
As we will see in section 10.9, trace element geochemistry has important applica-
tions in the investigation of magmatic processes. However, in partial melting and
partial crystallization models based on trace elements, the role ascribed to the
liquid phase is often subordinate to that of the solids undergoing melting or crys-
tallization. In such models, in fact, the fractionation properties of solids are
clearly defined, but the silicate melt is often regarded as a sort of reservoir that
does not exercise any structural control on trace partitioning, where trace ele-
ments are merely discharged or exhausted, depending on their relative affinities
with the solids. This sort of reasoning is fundamentally wrong and may result in
prejudice and vicious circles.
10.6.1 Compositional Effects of the Melt

As described in chapter 6, the main factors determining the solubility of a given
element in a silicate melt are the Lux-Flood acidity of its oxide and the relative
proportions of the cations of different field strengths (cation charge over squared
sum of cation plus ligand radii Z/A2 ) or charge densities (cation charge over ionic
radius Z/r).
More properly concerning trace elements, although we may hypothesize a
Henry’s law solubility (i.e., molar proportions are too low to affect the extension
and relative proportions of cationic and anionic matrices in the melt), it is true
that their solubility is a complex function of the chemistry of the liquid phase
(and of P-T conditions), as already seen for major elements.
Studying the unmixed portions of the miscibility gap between granitic melt
(quartz plus feldspar normative) and ferropyroxenitic melt (pyroxene normative),
Ryerson and Hess (1978) showed that high-field-strength elements such as Ti,
REE, and Mn are preferentially stabilized in low polymerized liquids (e.g., ferro-
pyroxenitic melt) with respect to highly polymerized liquids, whereas the opposite
is true for low-field-strength ions such as alkaline earths. Analogous indications
were obtained by Watson (1976) in the K2O-Al2O3-FeO-SiO2 system. This experi-
mental evidence confirms that trace partitioning between silicate crystals and
melts is definitely affected by the structure of the liquid. According to Ryerson
and Hess (1978), for instance, the mass distribution coefficients of the various
REE between clinopyroxene and ferropyroxenitic liquid are 10 times lower than
the clinopyroxene/granitic liquid mass distribution coefficients. Significant effects
have also been observed in other elements (Ti, Mn, K), whereas still other ele-
ments are less affected (e.g., Sr, Ba: Watson, 1976; Ryerson and Hess, 1978).
Figure 10.7 shows the effects of temperature and melt composition on the
Figure 10.7 Compositional effects of melt on olivine/liquid mass distribution of Mn2+.

Reprinted from E. B. Watson, Geochimica et Cosmochimica Acta, 41, 1363–1374, copy-
mass distribution of manganese between olivine and silicate liquid: with decreas-
ing melt acidity (expressed here as the Si/O atomic ratio), conventional mass
distribution coefficient K⬘ decreases for increased solubility of the element in the
liquid phase, as a result of the increased relative extension of the anion matrix
(see section 6.1.3).
According to Ryerson and Hess (1978), the augmented stabilization of high-field-

strength elements in slightly polymerized melts is attributable to their tendency to associ-
ate with O⫺, to minimize electrostatic repulsion with neighboring cations, and the prefer-
ential stabilization of low-field-strength ions in highly polymerized melts is attributable to
coupled substitutions with tetrahedrally coordinated aluminum to maintain charge bal-
ance. However, we saw in chapter 6 that the preferential association with singly bonded
oxygen involves low-field-strength ions (for example, preferential association of Ca2⫹ with
O⫺ in the presence of Fe2⫹ ions). We also know that high-field-strength ions favor the
polymerization reaction through acidic complexation of free oxygen O2⫺. It is thus evident
that the bulk solubility of the trace element in the liquid depends on its Lux-Flood acidity
and on the relative extensions of the cation and anion matrices of the liquid phase.
10.6.2 Fractionation Effects in Melts: the Example of REE
We may now argue whether or not a silicate melt can fractionate elements of the
same group. The question is rather important for REE, whose relative fraction-
676 / Methods
ation is usually interpreted by petrologists as a petrogenetic indicator of mag-

matic evolution.
According to Fraser (1975a), rare earth oxides may dissociate into the melt
with both basic and acidic behavior—i.e.,
REE 2O3 ⇔ 2REE 3 − + 3O 2 − (10.37)
and
REE 2O3 + O 2 − ⇔ 2REEO 2− . (10.38)
The relative proportions of cationic (REE3⫹ ) and polyanionic (REEO⫺

2 ) are es-
tablished according to the homogeneous equilibrium
REE 3( l+) + 2O 2( l−) ⇔ REEO 2−( l ) , (10.39)
whose Gibbs free energy change is related to partial equilibria 10.37 and 10.38 by
∆G 39 =
1
2
(
∆G 38 − ∆G 37 . ) (10.40)
Rewriting equilibrium 10.39 in terms of thermodynamic activity of the involved

species, we obtain
 a 
REEO−
∆G39 = − RT ln 
2( l ) .
 a 3+ a 2 2 −  (10.41)
 REE( l ) O( l ) 
Writing the bulk molar amount of trivalent REE in melt n in terms of partial
amounts of polyanions (n1 ) and cations (n2 ):
n REE III ( l ) = n1 REEO 2−( l ) + n2 REE 3( l+) , (10.42 )
and adopting Temkin’s model (Temkin, 1945) to define thermodynamic activity

over the appropriate matrix, after some passage, one arrives at a mean activity
coefficient ␥REE (relating mean thermodynamic activity to mean concentration)
that is a function of the disproportionation constant K39 and of the activity of
free oxygen in the silicate melt (Ottonello, 1983):
2 aO2 − ⋅ K 39
γ REE =
(l )
. (10.43)
1 + aO2 − ⋅ K 39
(l )
The Gibbs free energy change involved in disproportionation equilibrium

10.39 varies linearly with the field strengths of the various elements of the series:
 Z 
∆G 39 = a + b  2  , (10.44)
 AREE 
where a and b are constants. This amount of energy (hence, the various dispropor-
tionation constants K39 ) can be evaluated from experimental observations on liq-
uid/liquid REE partitioning (see Ottonello, 1983, for details of calculations).
Table 10.5 shows a comparison between measured (m: Watson, 1976; Ryerson
and Hess, 1978, 1980) and calculated (c: Ottonello, 1983) mass distributions on
the basis of the amphoteric solution model. The K39 values obtained from the
model are as follows: La ⫽ 7.5, Ce ⫽ 9, Pr ⫽ 11, Nd ⫽ 14, Sm ⫽ 21, Eu ⫽ 27,
Gd ⫽ 33, Tb ⫽ 40, Dy ⫽ 50, Ho ⫽ 62, Er ⫽ 76, Tm ⫽ 95, Yb ⫽ 120, and Lu ⫽
150. Applications of these K39 values to equation 10.43 leads to the mean activity
coefficient trends observed in figure 10.8.
The plot in figure 10.8 clearly shows that silicate melts may exert marked
fractionation on REE: acidic melts with relatively greater free oxygen activity
display the highest mean activity coefficients in light REE (LREE). LREE are
thus less soluble in these melts than heavy REE (HREE). The opposite is ob-
served in basic melts, which have the highest mean activity coefficients at the
HREE level. In other words, REE relative concentrations in silicate melts simply
reflect their relative solubilities, as we saw for crystal structures.
10.6.3 Effects of Melt Composition on Solid/Liquid

Distribution of Altervalent Elements
The Eu2⫹ /Eu3⫹ ratio in silicate melts is controlled by the redox equilibrium
4 Eu 3( l+) + 2 O (2l−) ⇔ 4 Eu 2( l+) + O 2 (gas) . (10.45)
Morris and Haskin (1974) observed an increase in the Eu2⫹ /Eu3⫹ ratio with in-
creasing (Al⫹Si)/O and high field-strength ions in the melt. This effect may be
ascribed to the different acid-base properties of Eu2O3 and EuO (Fraser,
1975a,b). Equilibrium 10.45 may in fact be rewritten in the form
1
Eu2O3( l ) ⇔ 2 EuO ( l ) + O 2 (gas) . (10.46)
2
The activities of the two oxides in the liquid are determined by the type of
dissociation. For EuO, basic dissociation is plausible:
Table 10.5 Measured (m) and calculated (c) basic liquid/acidic liquid REE mass distribution constants (Ottonello, 1983).
La Sm Dy Yb Lu
Reference(s) m c m c m c m c m c
Watson (1976) 3.70–4.31 3.68 4.19–5.15 4.15 – – – – 4.18–7.13 5.69
Ryerson and Hess (1978) 3.08–6.90 6.86 – – 2.20–4.64 10.9 3.03–4.18 18.2 – –
(system a)
Ryerson and Hess (1978, 1980)
(system b) 13.8 14.5 – – 16.57 27.37 – – – –
Figure 10.8 Mean activity coefficients of various REE in silicate melts with different
values of activity of dissolved free oxygen aO2⫺, plotted as a function of increasing atomic
number. Reprinted from Ottonello (1983), with kind permission of Theophrastus Publish-
ing and Proprietary Co.
EuO ⇔ Eu 2 + + O 2 − . (10.47 )
However, Eu2O3 behaves as an amphoteric oxide and dissociates according to the

two reaction schemes already proposed for all trivalent REE (cf. eq. 10.37 and
10.38). Denoting Ka and Kb , respectively, as the acidic and basic dissociation
constants, the Eu3⫹ /Eu2⫹ oxidation ratio varies with the silica amount in the melt
(1 ⫺ XMO ) according to
Eu 3+
=
a O1 4
2(gas)

 K a1 2 ⋅ a 3 22− ⋅
( )
1 − X MO 

( )
(10.48)
Eu 2 + c  O(l )
X MO + 0.25 1 − X MO 
(Wood and Fraser, 1976), where c is a constant and XMO is related to (Al⫹Si)/O
in the systems investigated by Morris and Haskin (1974) through (Fraser, 1975b):
 O   O  
X MO =  − 2   − 2 + 1 . (10.49)
 Al + Si   Al + Si  
It must be noted that the increased stabilization of the oxidized form with the
basicity of the melt, experimentally observed by Morris and Haskin (1974) for
680 / Methods
Table 10.6 Henry’s law concentration

limits for trace elements in silicate melts.
? ⫽ suggested values (from Ottonello,
1983).
Element Henry’s Law Limit (ppm)

Ba 5,000
Sr 5,000
La ⬎ 6,000
Sm ⬎ 11,000
Lu 3,000?
Mn ⬎ 7,000
Ti 5,000
Zr ⬎ 10,000
Ta ⬎ 10,000
europium, is a generalized phenomenon for all altervalent elements present at

trace levels (it was observed, for instance, by Paul and Douglas, 1965a,b, and
Nath and Douglas, 1965, for Ce3⫹ /Ce4⫹, Cr3⫹ /Cr6⫹, and Fe2⫹ /Fe3⫹ ). This fact
confirms that the acid-base treatment proposed by Fraser (1975a,b) can be ex-
tended to all trace elements present in melts in various oxidation states.
10.7 Limits of Validity of Henry’s Law for Trace

Elements in Silicate Melts
There are few data on this point. Most information comes from the experimental
work of Watson (1976) on trace partitioning between immiscible liquids in the
K2O-Al2O3-FeO-SiO2 system at 1180 °C and 1 bar. Table 10.6 shows some con-
centration limits based on this experimental work. At present, it is impossible to
establish whether or not deviations from Henry’s law take place in the acidic or
basic liquids at equilibrium.
10.8 Conventional and Normalized

Partition Coefficients
We have already seen that, within the range of Nernst’s law, the solid/liquid parti-
tion coefficient differs from the thermodynamic constant by the ratio of the Hen-
ry’s law activity coefficients in the two phases—i.e.,
γ h,i ,l
K i′ = Ki . (10.50)
γ h,i , s
Although this equation reduces to an identity whenever solute-solvent interac-

tions are embodied in the definition of the “Henry’s law standard state” (cf. sec-
tion 10.2), it must be noted that K⬘i is the molar ratio of trace element i in the two
phases and not the weight concentration ratio usually adopted in trace element
geochemistry. As we will see later in this section, this double conversion (from
activity ratio to molar ratio, and from molar ratio to weight concentration ratio)
complicates the interpretation of natural evidence in some cases. To avoid ambi-
guity, we define here as “conventional partition coefficients” (with the same sym-
bol K⬘) all mass concentration ratios, to distinguish them from molar ratios and
equilibrium constants.
In an attempt to minimize the compositional effects of phases on trace parti-
tioning, Henderson and Kracek (1927) also introduced the concept of “normal-
ized partition coefficient” D, which compares the relative trace/carrier (Tr/Cr)
mass distributions in the two phases at equilibrium—i.e.,
D=
( Tr Cr) s
.
( Tr Cr) l
(10.51)
The formulation of this coefficient derives from the consideration that solid/liquid
trace element distribution can be ascribed to the existence of simple exchange
equilibria of the type
CrM + Tr ⇔ TrM + Cr,

solid liquid solid liquid (10.52)
where M is the common ligand. For instance, an exchange reaction of this type
may be responsible for the isomorphic substitution of Sr in plagioclase:
CaAl 2 Si 2 O 8 + Sr 2 + ⇔ SrAl 2 Si 2 O 8 + Ca 2 + .
solid melt solid melt
(10.53)
In this case, trace element and carrier occupy the same structural position both
in the solid phase and in the melt and are subject to the same compositional
effects in both phases (i.e., extension of the cation matrix in the melt and amount
of anorthite component in the solid). Figure 10.9A shows the effect of normaliza-
tion: the conventional partition coefficient of Sr between plagioclase and liquid
varies by about one order of magnitude under equal P-T conditions, with increas-
ing anorthite component in solid solution, whereas normalized distribution co-
efficient D is virtually unaffected. Figure 10.9B shows the same effect for the
Ba-Ca couple.
It is important to note that, in the example given above, trace element (Sr or
Ba) and carrier (Ca) have the same formal charge. In the case of substitution of
682 / Methods
Figure 10.9 Effect of normalization on trace element distribution of Sr (A) and Ba (B)
between plagioclase and silicate liquid. Distribution curves are based on various experi-
mental evidence (Ewart et al., 1968; Ewart and Taylor, 1969; Berlin and Henderson, 1968;
Carmichael and McDonald, 1961; Philpotts and Schnetzler, 1970; Drake and Weill, 1975).
Ab: albite; Or: orthoclase; An: anorthite.
elements with different valences, the exchange equilibrium is more complex and
the adoption of normalized partition coefficients does not eliminate the composi-
tional effects. For instance, in the case of stabilization of a rare earth in diopside,
we can write
CaMgSi 2 O 6 + REE 3 + + Al 3 + ⇔ REEMgSiAlO 6 + Ca 2 + + Si 4 + .

pyroxene melt melt pyroxene melt melt
(10.54)
In this case, trace-carrier substitution REE3⫹-Ca2⫹ is associated with coupled

substitution Si4⫹-Al 3⫹, necessary to maintain charge balance. Simple normaliza-
tion to the Ca2⫹ amount in the solid and liquid phases at equilibrium is ineffective
in this case.
We have already noted that the double conversion from activity ratio to
weight concentration ratio implicit in trace element geochemistry may involve
complexities that must be carefully evaluated in the interpretation of natural evi-
dence. Let us consider, for instance, the distribution of Ni between clinopyroxene
and silicate liquid. Figure 10.10A shows the effect of temperature on the conven-
Figure 10.10 Arrhenius plot of conventional partition coefficient (A) and thermody-
namic constant (B) for clinopyroxene/liquid distribution of Ni.
tional weight concentration ratio, according to the data of Lindstrom and Weill
(1978) and Häkli and Wright (1967): the Arrhenius plot defines two distinct
arrays, with different slopes and intercepts.
If we now consider solid/liquid Ni distribution from a thermodynamic point
of view, as suggested by Lindstrom and Weill (1978):
NiO + CaO + 2SiO 2 ⇔ CaNiSi 2 O 6 ,

liquid liquid liquid pyroxene
(10.55)
the thermodynamic constant of equilibrium 10.55 is given by
K 55 =
[CaNiSi O ] 2 6
, (10.56)
[NiO ][CaO ][SiO ]
2
2
where the terms in square brackets denote thermodynamic activity.

We can adopt the approximations
[CaNiSi O ] = X
2 6 Ni,M 1 (10.57)
[SiO ] = X
2 SiO 2 , l (10.58)
[CaO ] = X CaO,l (10.59)

684 / Methods
[NiO ] = X NiO,l (10.60)
(Lindstrom and Weill, 1978), where XNi,M1 is the atomic fraction of Ni in the M1
site of pyroxene, and the other X terms represent the molar fractions of oxide in
the melt. As shown in figure 10.10B, the plot of thermodynamic constant K55
obtained in this way defines a single Arrhenius trend. Evidently, in this case, the
double conversion implicit in the adoption of conventional partition coefficients
introduces an apparent complexity that is indeed ascribable to the different com-
positions of the investigated systems. These effects must be carefully evaluated
when studying trace element distributions in petrogenetic systems.
Recent appraisals of the state of art in experimental trace partitioning studies
relevant to igneous petrogenesis may be found in the review articles of Green
(1994) and Jones (1995).
10.8.1 Effects of Temperature on Partition Coefficients

Let us consider the chemical potentials of a trace component (TrM) and a carrier
component (CrM) in a solid mixture and in the liquid at equilibrium:
µ TrM,s = µ TrM,
0
s + RT ln a TrM,s (10.61)
µ CrM, s = µ CrM,
0
s + RT ln a CrM, s (10.62)
µ TrM,l = µ TrM,
0
l + RT ln a TrM,l (10.63)
µ CrM,l = µ CrM,
0
l + RT ln a CrM,l . (10.64)
Equations 10.61 to 10.64 yield
a TrM,l  µ TrM,l − µ TrM,

0
l − µ TrM, s + µ TrM, s
0 
= exp   (10.65)
a TrM, s  RT 
and
a CrM,l  µ CrM,l − µ CrM,

0
l − µ CrM, s + µ CrM, s
0 
= exp   (10.66)
a CrM, s  RT 
Because at equilibrium ␮TrM,s ⫽ ␮TrM,l and␮CrM,s ⫽ ␮CrM,l, equations 10.65 and

10.66 reduce to Vant Hoff isotherms:
a TrM,l  µ TrM,
0
s − µ TrM,l
0 
= exp   (10.67)
a TrM, s  RT 
and
a CrM,l  µ CrM,
0
s − µ CrM,l
0 
= exp  . (10.68)
a CrM, s  RT 
Substituting molar fractions X and molalities m for activities in the solid and
in the liquid phase, respectively, we obtain
a TrM,l m TrM,l ⋅ γ TrM,l

K TrM = = (10.69)
a TrM, s X TrM, s ⋅ γ TrM, s
and
a CrM,l mCrM,l ⋅ γ CrM,l

K CrM = = . (10.70)
a CrM, s X CrM, s ⋅ γ CrM, s
As we have already seen, within Henry’s law range the activity coefficient of the
trace component in the solid can be expressed as
 ∆µ ∗ 
γ TrM, s = exp 
 RT 
, (10.71)
 
where ⌬␮* is the molar Gibbs free energy amount involved in solute-solvent inter-
actions. The (molar) partition coefficient thus has the form
 ∆µ ∗   ∆µ TrM
0  −1
K ′TrM = exp 
 RT   ⋅ exp   ⋅ γ TrM,l . (10.72)
   RT 
Analogously, assuming the carrier component to obey Raoult’s law in the

solid, we obtain the following equation for normalized molar distribution Ds/l:
 ∆µ ∗   ∆µ TrM
0   ∆µ CrM
0  γ CrM,l
D s l = exp   ⋅ exp   ⋅ exp  − ⋅ .
RT  γ TrM,l
 RT  (10.73)
   RT  
Because
686 / Methods
∆µ ∗TrM = ∆ h ∗TrM − T∆s ∗TrM + P∆v ∗TrM (10.74 )
∆µ TrM
0
= ∆ hTrM
0
− T∆ s TrM
0
+ P∆v TrM
0
(10.75)
∆µ CrM
0
= ∆ h CrM
0
− T∆s CrM
0
+ P∆v CrM
0
, (10.76 )
where terms in h, s, and v are, respectively, partial molar enthalpies, partial molar
entropies, and partial molar volumes (cf. section 2.5), substituting equations
10.74 to 10.76 in equations 10.72 and 10.73 and deriving the logarithmic forms
in dT, we obtain
 ∂ ln K ′TrM 

 ∂T 
 P, X
=
1  0
RT 2 
( ) ∗ 
hTrM, s − hTrM,l − ∆ hTrM
 (10.77)
 ∂ ln D s l 
 ∂T 
  P, X
= (
1  0
RT 2 
) (
hTrM, s − hTrM,l − hCrM,
0
) ∗ 
s − hCrM,l − ∆ h TrM  .
 (10.78)
The terms in parentheses in equations 10.77 and 10.78 represent the enthalpies
of solution of TrM and CrM, respectively, in the melt at the T of interest (cf.
section 6.4). If the enthalpy of solution does not vary appreciably with T, because
the interaction term ⌬h* TrM cannot also be expected to vary appreciably with T,
integration of equation 10.77 leads to a linear form of the type
ln K ′TrM = A + BT −1 , (10.79)
where A and B are constants.

Linearity is more marked for normalized distributions, because possible mod-
ifications with T of the heats of solution of trace and carrier components compen-
sate each other.
The behavior described above is exemplified in figure 10.11, where Arrhenius
plots of solid/liquid conventional partition coefficients for transition elements re-
veal more or less linear trends, with some dispersion of points ascribable to com-
positional effects.
10.8.2 Effects of Pressure on Partition Coefficients

Substituting equations 10.74 to 10.76 in equations 10.72 and 10.73 and deriving
in dP the logarithmic forms, we obtain
 ∂ ln K ′TrM 

 ∂P 
 T,X
=
1  0
RT 
( ) ∗ 
v TrM, s − v TrM,l − ∆v TrM
 (10.80)
Figure 10.11 Effects of temperature on conventional solid/liquid partition coefficients of

various transition elements. Sources of data are also listed.
and
 ∂ ln D s l 
 ∂P 
  T,X
=
RT 

(
1  0
) (
v TrM, s − v TrM,l − vCrM,
0
) ∗ 
s − v CrM,l − ∆v TrM  .(10.81)

The trace component in solid solution usually has partial molar volumes not
far from those of the carrier, for crystal-chemical reasons, and the terms in paren-
theses in equation 10.81 compensate each other. The term ⌬v* TrM, expressing the
differences in partial molar volume arising from solute-solvent interactions in the
solid, is also usually negligible. As a result, the effect of P on normalized partition
coefficients is usually very restricted. The same cannot be said for conventional
partition coefficient K⬘TrM, because the partial molar volumes of the trace compo-
nent in the solid and liquid phases (and hence their differences) may vary appreci-
ably with P (McIntire, 1963).
10.8.3 Effects of Oxygen Fugacity on the Solid/Liquid

Distribution
Oxygen fugacity f O2 directly affects the redox states of trace elements in melts
(see, for instance, equation 10.45). Figure 10.12 shows the effects of f O2 on the
oxidation state of Ti and V in an Na2Si2O5 melt at T ⫽ 1085 °C, according to the
688 / Methods
Figure 10.12 Effects of oxygen fugacity on oxidation state of Ti and V in Na2Si2O5 melt
at T ⫽ 1085 °C (experimental data from Johnston, 1964, 1965, and Johnston and
Chelko, 1966).
experiments of Johnston (1964, 1965) and Johnston and Chelko (1966). These
effects, which must not be confused with the previously described compositional
effect, markedly influence solid/liquid trace element distribution, because crystal
structures dramatically discriminate elements in various oxidation states. Figure
10.13 shows, for example, the effect of f O2 on the conventional partition coeffi-
cient of Cr between olivine and melt (figure 10.13A) and between subcalcic py-
roxene and melt (figure 10.13B). In both cases, the effect of f O2 is nonlinear and
varies with T in a rather complex way.
10.9 Trace Element Distribution Models for Magma

Genesis and Differentiation
The evolution of an igneous system is controlled by changes in intensive variables

P and T over time and by the capability of the system to exchange heat or matter
with the exterior (i.e., “open,” “closed,” or “isolated” systems; cf. section 2.1).
Note that the term “exterior” must be intended in its full thermodynamic sense:
it may be the neighboring zone of a lithospheric portion undergoing melting,
where liquids are conveyed, as well as the inner parts of newly formed crystals,
unable to exchange matter with the residual liquid for purely kinetic reasons,
during magma differentiation. We can consider here two main cases:
Figure 10.13 Effect of oxygen fugacity on conventional partition coefficient of Cr. (A)
Olivine/liquid partitioning; experimental data of Bird (1971), Weill and McKay (1975),
Huebner et al. (1976), Lindstrom (1976), and McKay and Weill (1976). (B) Subcalcic py-
roxene/liquid partitioning; experimental data of Schreiber (1976). Reprinted from A.J. Ir-
ving, Geochimica et Cosmochimica Acta, 42, 743–770, copyright 䉷 1978, with kind permis-
sion from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB,
UK.
1. Complete equilibrium between solid and liquid

2. Interface equilibrium between accreting (or dissolving) crystals and
liquid
In the first case, the diffusion rates of the trace component in all phases of the
system must be sufficiently high with respect to rates of crystal accretion or melt-
ing. Figure 10.14 shows experimental data on elemental diffusivities of alkalis in
a rhyolitic melt. T conditions being equal, the diffusivities of Na and K in the
melt are four to six orders of magnitude greater in the liquid than in feldspars
(compare figures 10.14 and 4.8). We can thus expect the formation of chemically
zoned crystals, whereas possible chemical zoning in the liquid may more reason-
ably be ascribed to advection.
The partition coefficient of trace element i between a given solid phase s and
690 / Methods
Figure 10.14 Elemental diffusivities of alkalis in a rhyolitic melt (experimental data of

Gabis et al., 1979).
liquid l in a closed system is valid on each portion of the two phases at equi-
librium:
X i,s
K′ = . (10.82)
X i ,l
In the case of interface equilibrium (open system conditions), the partition co-
efficient is valid only at the interface between solid and liquid (or at zero distance
from the interface) and at time of crystallization (or melting) t:
X i , s, 0,t
K′ = . (10.83)
X i ,l , 0,t
It is important to stress that, when measuring the simple concentration ratios of

mean abundances in the solid and liquid phases formed at interface equilibrium
(as is typically done in comparative studies), the resulting (apparent) partition
coefficient KA is related to partition coefficient K⬘, valid at the interface, by
t
K′
t ∫ X i ,l , 0,t dt
KA = L
0
(10.84)
1
L ∫ X i ,l , x,t dx
0
(Albarede and Bottinga, 1972), where t is time since the beginning of the process
(at constant accretion or dissolution rates), Xi,l,x,t is the trace element concentra-
tion measured at distance x from the interface and at time t, and L (in the case of
fractional crystallization) is half the mean distance between centers of accreting
crystals less half the mean thickness of the crystals. We will see later the effects
of erroneous evaluation of KA and K⬘ when modeling fractional crystallization
processes.
10.9.1 Crystallization in an Open System, or

“Fractional Crystallization”
During Rayleigh’s crystallization process (formation of crystals with equilibrium
limited to the solid/liquid interface), the relative concentration of a trace element
in liquid Cl with respect to the initial concentration in system C 0 is described by
the Doerner-Hoskins equation (Doerner and Hoskins, 1925; we will omit here-
after subscripts in the indicization of coefficients, for the sake of simplicity):
[ ]
K ′ −1
Cl = C 0 1 − F , (10.85)
where F is the mass fraction of solid in the system.
Distribution equation 10.91 has been developed independently by various authors,

starting from the differential equation valid for a single component:
dm M − m
= K′  , (10.86)
ds  Q−S 
where Q is the initial mass of the system, M is the mass of trace elements in the system,
S is the mass of the solid, and m is the mass of trace element in the solid. According to
Neumann et al. (1954), the integrated form of equation 10.86 is
  S 
K′
m = M 1 −  1 −   (10.87)
  Q 

(when S ⫽ 0, m ⫽ 0).
Differentiating equation 10.87 in ds yields
K ′ −1
dm M S
= K′ 1 −  . (10.88)
ds Q  Q
Translated into our formalism, equation 10.88 becomes

692 / Methods
[ ]
K ′ −1
Cs = C0K′ 1 − F ( 10.89 )
and, applying equation 10.82,
[ ]
K ′ −1
Cl = C0 1 − F . (10.90)
It must be noted that equation 10.89 is valid only at the solid/liquid interface, whereas
equation 19.90 is valid over the entire crystallization range.
The mean concentration in solid C s may be obtained by integration of equa-

tion 10.86 (Albarede and Bottinga, 1972):
[ ]
K′
1− 1− F
Cs = C0 . (10.91)
F
Clearly, equation 10.91 can be derived directly from equation 10.87 by substitut-
ing concentrations for masses.
Equation 10.91 also can be expressed as
[ ]
K ′ −1
C s = K AC 0 1 − F (10.92)
(the identity between eq. 10.92 and eq. 10.89 at the interface where KA ⫽ K⬘
is obvious).
The ratio between equations 10.91 and 10.90 gives the relationship between
apparent partition coefficient KA and conventional partition coefficient at inter-
face K⬘ (i.e., the integrated form of eq. 10.84):
[ ]
K′
C 1− 1− F
KA = s = . (10.93)
[ ]
K ′ −1
Cl F 1− F
Equation 10.93 clearly shows that, if apparent partition coefficient KA is adopted

instead of conventional partition coefficient K⬘ actually valid at the solid/liquid
interface to model Rayleigh’s crystallization, errors arise whose magnitudes in-
crease the more K⬘ differs from 1 and the longer the process advances. This is
clearly shown in figure 10.15, in which fractional differences (KA ⫺ K⬘)/K⬘ are
plotted as functions of F for various values of K⬘.
In the case of compositional gradients in the liquid resulting from advection
or slow diffusivity, the relationship between KA and K⬘ is more complex and de-
pends both on the accretion rate of crystals vA and the diffusion (or removal) rate
in liquid vD (see eq. 10.84). Values of KA for various vD /vA ratios with K⬘ ⫽ 2 are
Figure 10.15 Fractional differences (KA ⫺ K⬘)/K⬘ plotted as functions of degree of frac-
tional crystallization F for various values of K⬘. Reprinted from F. Albarede and Y. Bot-
tinga, Geochimica et Cosmochimica Acta, 36, 141–156, copyright 䉷 1972, with kind per-
mission from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5
1GB, UK.
Table 10.7 Relationships between apparent partition

coefficient KA and actual partition coefficient K⬘ for various
vD /vA ratios (from Albarede and Bottinga, 1972).
KA
vD / vA K⬘ F ⫽ 0.1 F ⫽ 0.3 F ⫽ 0.5

10 2 2.04 2.31 2.78
0.1 2 1.54 1.38 1.36
0.001 2 1.15 1.14 1.14
694 / Methods
10.9.2 Crystallization in a Closed System, or “Equilibrium

Crystallization”
Whenever diffusion rates within crystals and melt are high enough to ensure equi-
librium in all portions of the system, the distribution equations are
K ′C 0
Cs =
[1 − (1 − K ′ )F ] (10.94.1)
and
C0
Cl = ,
[1 − (1 − K ′ )F ] (10.94.2)
where K⬘ is valid in all portions of the system (and not only at the interface, as
in the preceding case).
10.9.3 Particular Cases: (I) Resorption of Solid Phases

During Rayleigh’s Crystallization
The distribution equations for this particular case—corresponding, for instance,
to a peritectic reaction taking place during cooling of a magma chamber—were
evaluated by Neumann et al. (1954) starting from a modified form of general
equation 10.86:
dm  M − m + C PS 
= K′  Tr
,
ds (
 Q − 1 − P S  ) (10.95)
where P is the mass ratio between resorbed and precipitated phases and C Tr is the
mean concentration of the trace element in the resorbed phase (see eq. 10.86 for
the remaining symbols). Neumann et al. (1954) gave the following distribution
equations, which are valid at the interface:
[ K ′ (1−P ) −1]  K ′M [ K ′ (1−P ) −1]

Cs =
K ′C Tr P 
 1 − 1 − P  − 1 +

1 −
( )
1− P S
 (10.96)
1 − P − K′  Q   Q  Q 
 
Cs
Cl = . (10.97)
K′
10.9.4 Particular Cases: (II) Simultaneous Crystallization

of Several Phases
Conventional partition coefficient K⬘ is replaced by bulk solid/liquid mass distri-
bution coefficient Ds/l, corresponding to the weighted mean of the conventional
partition coefficients for each solid/liquid pair:
Xα Xβ Xγ
Ds l = + + + L. (10.98)
K l′ α K l′ β K l′ γ
In equation 10.98, X␣, X␤, and X␥ are the relative mass fractions of crystallizing
phases (X␣ ⫹ X␤ ⫹X␥ ⫹ . . . ⫽ 1), and K⬘l/ ␣, K⬘l/ ␤, and K⬘l/ ␥ are their respective
conventional partition coefficients.
Ds/l can be substituted for K⬘ in equations 10.89 and 10.90 and equations
10.94.1 and 10.94.2.
10.9.5 Particular Cases: (III) Presence of

Intercumulus Liquids
Like the preceding case, this case is also quite common in geology. During crystal
settling at the bottom of magma chambers, portions of liquid are trapped between
crystals and are not totally removed by successive filter-pressing processes. The
trace element concentration in the zone of accumulus (Cacc ) is given by the
weighted summation of the relative concentrations in the solid liquid (Xacc ) and
trapped liquid (1 ⫺ Xacc ). That is, for an equilibrium crystallization process,
K ′C 0
C acc = × X acc +
C0
(
× 1 − X acc )
[1 − (1 − K ′)F ] [1 − (1 − K ′ )F ] (10.99)
and, for a case of Rayleigh’s crystallization,
(
C acc = C 0 K A 1 − F ) ( ) ( )
KA −1 
× X acc + C 0 1 − F
KA −1 
× 1 − X acc . (10.100)
   
The effects of the presence of intercumulus liquid on the relative distribution

of trace elements during cooling of a magma chamber were investigated in detail
by Albarede (1976). Figure 10.16 shows the effects of different portions of intercu-
mulus liquid (arbitrary ␣ values of 1 ⫺ Xacc ) on the relative proportions of two
trace elements in the accumulus zone. Note that relative fractionation of the two
elements evolves along simple logarithmic trends whenever the mass proportion
of trapped liquid remains approximately constant throughout the crystallization
696 / Methods
Figure 10.16 Effects of presence of intercumulus liquid on relative distribution of trace

elements i and j during partial crystallization. ␣, ␣1, and ␣2 are arbitrary proportions of
intercumulus liquid. Reprinted with modifications from F. Albarede, Geochimica et Cos-
mochimica Acta, 40, 667–673, copyright 䉷 1976, with kind permission from Elsevier Sci-
ence Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
process. Note also, however, that the field of possible trace element concentrations
is quite wide (dashed area in figure 10.16), depending on the efficiency of filter-
pressing.
10.9.6 Melting in a Closed System, or

“Equilibrium Melting”
Following Shaw (1970), we can distinguish two types of melting of a heteroge-

neous solid in a closed system:
1. Modal melting. Solid phases melt in the same modal proportion in which
they occur in the initial solid, so that the modal composition of the resid-
ual solid remains unchanged throughout the process.
2. Nonmodal melting. The melting proportions of solid phases differ from
the modal proportions in the initial solid, so that the modal composition
of the residuum is progressively modified throughout the process.
In the first case, the distribution equations are

C0
Cl =
[D 0 (
+ F 1 − D0 )] (10.101)
and
C 0D 0
Cs = ,
[D 0 (
+ F 1 − D0 )] (10.102)
where F is the mass fraction of melt in the system (not the fraction of solid, as in
the preceding cases) and D 0 is the initial solid/liquid distribution at the beginning
of melting:
X 0,α X 0, β X 0,γ
D0 = + + + L. (10.103)
K l′ α K l′, β K l′ γ
However, “modal melting” is highly unlikely in geological systems, because the

eutectic normally differs from the modal constitution of the initial rock. Solid/
liquid distribution Ds/l thus changes progressively throughout melting (case 2),
based on the melting proportions P␣, P␤, P␥, . . . of the solid phases:
D 0 − PF
Ds l = , (10.104)
1− F
where
Pα Pβ Pγ
P= + + + L. (10.105)
K l′ α K l′ β K l′ γ
In this case the distribution equations are
C0
Cl =
[D 0 + F 1− P ( )] (10.106)
and
C 0D s l
Cs = .
[D 0 + F 1 − D0( )] (10.107)
Because P␣,␤,␥ and K⬘l / ␣,␤,␥ represent final state conditions, equations 10.106
and 10.107 can be applied only to the entire melting range whenever melting
proportions and conventional mass distribution coefficients remain constant
698 / Methods
throughout the process. In all other cases, the equations above must be associated
with functions describing the evolution of P␣,␤,␥ and K⬘l/ ␣,␤␥ with the degree of
partial melting F.
10.9.7 Melting in an Open System, or “Fractional Melting”

Several models have been proposed (Gast, 1968; Shaw, 1970; Hertoghen and
Gijbels, 1976), but the most satisfactory are those based on iterative calculations,
because they allow the evaluation of discontinuous events with sudden modifica-
tions of melting proportions, which are common in geological open systems (note
that such discontinuities, or even the evolution of melting proportions, cannot be
accounted for, if the distribution equations are obtained by integration at the
limit). The following distribution model (Ottonello and Ranieri, 1977) is a partial
modification of Gast’s (1968) approach:
n
C s,i = C 0 ∏
(
1 − Di + i − 1 ∆ F)
(10.108)
i =1 1− F
Di −1 − Pi −1 ∆ F
Di = (10.109)
1 − ∆F
C l , i = D i C s , i −1 (10.110)
n
1
C l ,i = ∑ C l ,i
i i =1
(10.111)
P = f (F ) (10.112)
K l′ α , β ,γ = f ( F ), (10.113)
where Cs,i is trace element concentration in the residual solid after removal of the
ith increment of liquid, Cl,i is trace element concentration in the ith increment of
liquid, C l,i is mean trace element concentration in i accumulated increments, and
⌬F is the fractional mass increment of melting.
Many natural processes are closer to fractional melting than to equilibrium
melting conditions. For instance, the leucosome bands in gneisses may be assimi-
lated to fractional melting liquids, being progressively removed from the system
as soon as they are produced. In such cases, the melting proportions of the solid
phases may change dramatically as the process advances (cf. Presnall, 1969) and
must be carefully evaluated for rigorous modeling of trace element distributions.
For this purpose, it is opportune to compare the natural system to a synthetic
analog, the melting behavior of which is known with sufficient precision. As we
Figure 10.17 Evolution of liquid and solid during fractional melting of initial com-
position C in Di-Ab-An system. Curves ILC(1) and ILC(2): fractional melting liquids;
TSC: residual solids; dotted line: fractionally accumulated liquids at quoted degrees of
partial melting (from Ottonello and Ranieri, 1977).
saw in chapter 7, the melting proportions of synthetic systems can be properly

quantified by application of petrogenetic rules.
Let us consider, for example, the system CaMgSi2O6 (Di)–NaAlSi3O8 (Ab)–
CaAl2Si2O8 (An) (figure 10.17).
Fractional melting of initial composition C produces instantaneous liquids evolving
initially along curve ILC(1) and then along curve ILC(2), when all diopside in the residual
solid is exhausted. The evolution of fractional melting liquids during the melting event
may be described by two systems of equations P ⫽ f (F ) and P ⫽ g(F ), which are valid,
respectively, in two melting ranges. For 0 ⬍ F ⬍ 0.51 [evolution along cotectic line
ILC(1)],


Pα = PAb =
[13 − (9 + 40 F )
12
]
20


P = f ( F ) ≡ Pβ = PAn =
(1 − Pα ) (10.114)
 2.17


Pγ = PDi = 1.17Pβ .

700 / Methods
For 0.52 ⬍ F ⬍ 1 [evolution along Ab-An edge ILC(2)],
Pα = PAb = 1.32(1 − F )


P = g( F ) ≡ Pβ = PAn = 1 − Pα (10.115)
P = P = 0.
 γ Di
The equations above were determined graphically using a few points on the ILC
curves of figure 10.17 and allow us to follow trace element distribution during fractional
melting of initial system C (Di15An51Ab34 ). The heavy line (TSC) in figure 10.17 describes
the evolution of the residual liquid during melting; the dotted curve represents the modal
evolution of fractionally accumulated liquids calculated by Morse (1976) with application
of the lever rule (see chapter 7). This last curve can be obtained directly by summing the
compositions of instantaneous liquids obtained through the two systems of equations
10.114 and 10.115.
10.9.8 Zone Melting

Zone melting is an industrial process used to purify slags and organic compounds,
and its application to geology was initially suggested by Harris (1957). The forma-
tion of a magma by zone melting is associated with the relative displacement of
a portion of lithosphere with respect to a zone of high heat flow. In such cases,
the solid/liquid distribution is ascribable to two distinct processes (fractional
melting and Rayleigh’s crystallization) taking place, respectively, at the face and
rear front of the high heat zone. During relative displacement of the molten zone
along trajectory 0 → X, trace element concentrations in the melt are dictated by
fractionation processes occurring at the two fronts:
 1  1  X 
Cl = C0  −
 Ds l  Ds l
− 1 ⋅ exp − Ds
 L
( l ) . (10.116)

In equation 10.116, L is the length of the molten zone and X is the length of
relative displacement. The application of zone melting to geological systems must
be carried out with caution, because the modal proportions in the original solid,
in the melt, and in the precipitating solid all generally differ markedly. Moreover,
initial concentration C 0 cannot be expected to remain constant over long X tra-
jectories. Nevertheless, zone melting probably operates in the earth’s upper man-
tle: it has been suggested that the existence of the low-velocity layer is a result of
such a process. Radiogenic elements such as U, Th, and K may be efficaciously
extracted by zone melting from convecting lithospheric portions and contribute
feedback to the process by heat release associated with radioactive decay.
10.10 Presence of Fluid Phases
During magma genesis and differentiation, fluid phases may develop. The fluids
may exolve from the melt for two main reasons: progressive increase of concentra-
tion of gaseous species that reach the solubility product of the fluid in question,
or decreased solubility in the melt during ascent to the surface (adiabatic decom-
pression). If we call the mass fractions of solid, melt, and fluid in the system Fs ,
Fl , and Ff , respectively, the mass balance equation relating initial trace element
concentration C 0 to the concentrations in solid, melt, and fluid is
C 0 = C l Fl + C s F s + C f F f . (10.117)
For melting in a closed system, equation 10.117 gives
C0
Cl =
[Fl ( )
+ D s l 1 − Fl − F f + D f l F f ] (10.118)
C 0D s l
Cs =
[Fl ( )
+ D s l 1 − Fl − F f + D f l F f ] (10.119)
Cf = D f l Cl . (10.120 )
For melting in an open system, Shaw (1978) proposed the equations
(1 Ps l −1 )
C0  Fl Ps f 
Cf = 1 −  (10.121)
D0, s f + v  D0, s f + v 
Cs =
(
C f D0, s f − Ps f F f ) (10.122)
1 − Fl
Cf
Cl = (10.123)
Df l
 1 Ps l 
C0   Fl Ps f 
,
Cl = 1− 1−
Fl   D0, s f + v   (10.124)
 
702 / Methods
where ␯ is the constant fractional amount of fluid with respect to the initial mass
of the system.
If the fluid is present during fractional crystallization, the trace element con-
centration in the liquid is dictated by
C l = C 0 Fl
( Ds l + D f l G −1) (10.125)
(Allegre et al., 1977). In equation 10.125, G is the concentration of gaseous com-

ponents* dissolved in magma, which, following Holland (1972), is assumed to
remain constant throughout the process.
10.11 “Hygromagmatophile” and

“Incompatible” Elements
Some trace elements have a marked tendency to partition into the melt during
magma genesis and differentiation. Elements showing the highest affinity for the
melt (i.e., Th, Ta, Zr, Hf, La) were defined by Treuil (Treuil, 1973; Treuil and
Varet, 1973; Treuil and Joron, 1975; Treuil et al., 1979) as “hygromagmatophile,”
and those preferentially partitioned into the melt (although not so markedly as
the hygromagmatophile ones) were defined as “incompatible.”
The hygromagmatophile character of elements is determined according to Treuil by

the tendency to form stable complexes with anionic ligands in the melt (Treuil et al., 1979).
Let us consider the formation of a complex between hygromagmatophile element Mx⫹
and ligand L y⫺ present in the melt through homogeneous equilibrium of the type
M x + + nL y − ⇔ MLzn − , ( 10.126 )
whose constant is ␤. If we admit that solid/melt trace partitioning only involves cationic
species Mx⫹ (in agreement with the concept of “noncrystallizable groupings” of Ubbel-
hode, 1965, and Boon, 1971):
K =
[M ] x+
s
,
[M ] x+
l
(10.127)
the resulting conventional partition coefficient is
* The sentence “the gas phase is not considered as an additional phase” adopted by Allegre et
al. (1977) for this purpose is rather cryptic. They probably meant that the fractionation laws valid
for solid phases are also valid for the gaseous phase, regardless of the different state of aggregation.
In our view, the gas phase is a real phase to all intents and purposes (i.e., a region of the system with
peculiar and distinguishable chemical and physical properties; cf. section 2.1). Hence, the term G is
operationally analogous (although not identical) to v in equations 10.121 and 10.124.
Figure 10.18 Conventional trace element partition coefficient as a function of factor

␤(Lyl ⫺)n. Reprinted from Treuil et al. (1979), Bullettin Mineralogique, 102, 402–409, with
permission from Masson S.A., Paris, France.
K′s l =
[M ] γ
x+ −1
x+
s M ,s
.
[M x+
] γ + [ML ] γ
−1
x+
l M ,l
z− −1
z−
l ML , l
(10.128)
Introducing constant ␤ (eq. 8 in Treuil et al., 1979):
K
K ′s l = .
γ M x+ , s γ M x+ , l γ n
L y− ,l
( ) (10.129)
n
+ β Lyl −
γ M x +, l γ MLz −, l
Assuming ideal solution behavior by Mx⫹, Ly⫺, and MLz⫺ in the melt, equation 10.129
reduces to
K
K ′s = .
( )
l n (10.130)
1 + β Lyl −
Based on equation 10.130 (figure 10.18), Treuil et al. (1979) suggested that hygromagmato-
phile behavior is restricted to the field ␤(L ly⫺ )n ⬎⬎ 1, where K⬘s /l ⬍⬍ K, whereas for values
of (L ly⫺ )n near 1 the highest variability of K⬘s /l occurs, corresponding to the field of incom-
patible elements.
Figure 10.18 shows that coordination number n of the complex (which depends on
704 / Methods
the relative charges of cation and ligand) has a strong influence on K⬘s/l. The higher the
value of n (i.e., the higher the charge of the cation for similar ligands: for example, Zr4⫹,
Hf 4⫹, Th4⫹ ), the wider the field of hygromagmatophile behavior. Moreover, if the anionic
ligand is a complex of type CO32⫺, F⫺, Cl ⫺, or OH⫺, its concentration in the melt in
saturation conditions depends on the solubility product of the gaseous phase. Modifica-
tions in fluid partial pressure may thus result in sudden changes in K⬘s/l during magma
differentiation (Treuil et al., 1979).
According to Treuil, the constant concentration ratio of two hygromagmato-

phile elements is indicative of fractional crystallization. From the Doerner-
Hoskins equation:
C l = C 0F ( ),
Ds l −1
(10.131)
where F is the fraction of melt, if Ds/l ⬍⬍ 1 we obtain
Cl
≈ F −1 (10.132)
C0
(Anderson and Greenland, 1969) and, for two hygromagmatophile elements

i and j,
C i .l C
= i .0 = constant. (10.133)
C j ,l C j,0
Error ⌬ associated with equation 10.132 can be expressed as a function of F

(Allegre et al., 1977):
1 
∆ =  − F ( s l −1 )  F
D
(10.134)
F 
and is plotted for various values of Ds/l in figure 10.19.

However, equation 10.133 is not sufficient to ascribe magmatic evolution to
fractional crystallization if it is not supported by petrologic evidence. Adopting
the formalism of Shaw (1970), it may be shown that the distribution of hygromag-
matophile elements in fractionally accumulated liquids leads to the same sort of
linear array. The equation
[
C l = C 0 1 − F 1 Ds l F −1 ,] (10.135)
where F is the fraction of residual solid, in fact reduces to equation 10.132 when-
ever Ds/l is negligible. In the case of fractional melting, errors associated with
Figure 10.19 Fractional error associated with equation 10.132. (A) Rayleigh’s crystalliza-
tion and fractional melting. (B) Equilibrium melting.
equation 10.132 are symmetric with respect to errors associated with fractional
crystallization:
1 
∆ =  − F ( s l ) F
l D
(10.136)
 F 
(see dashed line calculated with Ds/l ⫽ 0.3 in figure 10.19A).

But equation 10.132 cannot be applied to equilibrium melting. The intro-
duced errors:
1 1 
∆= − F (10.137)
 F D sl + F + FD s l 
are not negligible for any reasonable value of F (figure 10.19B) and for two hygro-
magmatophile elements i and j the relative abundances will not generally obey
equation 10.133. This is especially true for low melting conditions in which Ds/l
is comparable in magnitude to F.
The differentiation index defined by equation 10.132 is a formidable tool in
the petrologic investigation of volcanic suites. Because types and modal propor-
tions of crystallizing phases change discontinuously during fractional crystalliza-
tion (cf. Presnall, 1969), plotting the various chemical parameters as a function of
F (defined through eq. 10.132) generally gives smooth trends with discontinuities
located at particular F values corresponding to the appearance or exhaustion of
a given solid. This was exemplified quite convincingly by Barberi et al. (1975) in
the Boina magmatic series (Afar, Ethiopia). Figure 10.20 shows major element
706 / Methods
Figure 10.20 Major element concentrations in effusive products of Boina series plotted
as functions of degree of fractional crystallization, based on equation 10.132. I ⫽ first
discontinuity (transition from olivine-dominated to plagioclase-dominated fractionation);
II ⫽ second discontinuity (appearance of Fe-Ti oxides); III ⫽ third discontinuity (field
of SiO2 oversaturated trachytes: apatite and Mn-oxides precipitate); IV ⫽ beginning of
peralkalinity field. From Barberi et al. (1975), Journal of Petrology, 16, 22–56. Reproduced
with modifications by permission of Oxford University Press.
concentrations plotted as functions of the degree of fractional crystallization F

determined on the relative abundances of Ce (CCe,0 /CCe ).
The Boina emission products show a more or less continuous differentiation
trend from transitional basalt to pantellerite. In the nonperalkaline field, the tran-
sition from basalt to ferro-basalt is dominated first by olivine (F ⬎ 0.65) and then
by plagioclase (F ⬎ 0.45), with minor clinopyroxene. A second differentiation
step with the appearance of Fe-Ti oxide crystals begins at F ⫽ 0.45: Fe and Ti
decrease abruptly. Less marked discontinuities are also observed at F ⫽ 0.3 (silica-
oversaturated trachytes) and F ⫽ 0.15 (peralkalinity field).
CHAPTER ELEVEN
Isotope Geochemistry
11.1 Isotopes in Nature
Atomic nuclei are composed mainly of protons, with a mass of 1.67262 ⫻ 10⫺24
g, or 1.007276 atomic mass units (amu, relative to 12C ⫽ 12.000000), and neutrons,
with a mass of 1.008665 amu (see appendix 1 for a complete energy list of nu-
clear properties).
The number of protons Z in the nucleus establishes the identity of a chemical
element. The mass number A of a given nucleus is established by the number of
protons Z and neutrons N, added together. Neutrons are “neutral” in charge, and
their relative numbers with respect to protons are not generally fixed. Generally,
stable elements with even numbers of protons tolerate greater variability in terms
of numbers of neutrons per nucleus than elements with odd numbers of protons.
Ag atoms, for instance, are stable only when they have 47 protons and 60 or 62
neutrons. Each possible combination of nuclear charge Z and neutron number N
is referred to as a nuclide. Nuclides of the same nuclear charge (same element)
with the same number of protons Z and a different numbers of neutrons N are
defined as isotopes of that element (iso ⫽ “same,” topos ⫽ “place”—i.e., they
occupy the same position on the periodic chart of the elements). Nuclides of the
same mass number A but with different atomic numbers Z are called isobars, and
nuclides with the same number of neutrons N but different masses A are isotones.
The difference N ⫺ Z (corresponding to A ⫺ 2Z) is referred to as an isotopic
number. Finally, isomers are two or more nuclides having the same mass number
A and atomic number Z but existing for measurable times in different quantum
states, with different energies and radioactive properties.
The term “isotope” was coined by Soddy (1914) to define two or more substances of
different masses occupying the same position in the periodic chart of the elements. Soddy’s
hypothesis was adopted to explain apparent anomalies in the relative positions of three
couples of elements (Ar-K, Co-Ni, and Te-I) in the periodic chart. For instance, pot-
assium is present in nature with three isotopes with masses of 39, 40, and 41, respectively,
in the following proportions: 39K ⫽ 93.26, 40 K ⫽ 0.01, and 41K ⫽ 6.73. Because the pro-
portion of 39K is dominant, the element K occurs in nature with mass 39.102, near the
/ 707
708 / Methods
mass of isotope 39K. Ar also has three isotopes: 36Ar (0.34%), 38Ar (0.06%), and 40 Ar
(99.60%) and, obviously, the atomic weight of element Ar (39.948) is near that of isotope
40
Ar, which is the most abundant. Argon, which has an atomic number lower than that of
potassium, has a higher mass, and this explains the (apparent) anomaly of their respective
positions on Mendeleev’s periodic chart.
Figure 11.1 is a chart of nuclides with N as the ordinate and Z as the abscissa.
In this representation, isotones appear along horizontal lines and isotopes along
the same vertical line. The opposite sort of representation is known as a “Segré
chart.”
All stable nuclides fall above the N ⫽ Z line, with the exception of 1H and
3
He. Up to 157 N, the number of neutrons does not exceed the number of protons
by more than 1. Above 15 7 N, the stable nuclides diverge from the N/Z ⫽ 1 line
upward, as a result of a progressive increase in the proportion of neutrons.
Hydrogen atoms and part of 4He are believed to have been created during the “Big
Bang” by proton-electron combinations. Most nuclides lighter than iron were created by
nuclear fusion reactions in stellar interiors (cf. table 11.1). Nuclides heavier than the Fe-
group elements (V, Cr, Mn, Fe, Co, Ni) were formed by neutron capture on Fe-group seed
nuclei. Two types of neutron capture are possible: slow (s-process) and rapid (r-process).
Figure 11.2 shows two enlarged portions of the Segré chart where s-process and
r-process are effective. Every box in the figure represents a possible nuclide, but only the
white boxes are occupied by stable nuclides. The diagonally ruled boxes are unstable (half-
lives of hours, minutes, or seconds) and the crosshatched ones are highly unstable. In the
s-process (upper chain in figure), neutrons generated during He burning in stellar interiors
are added to Fe-group seed nuclei. Neutron capture (n) displaces the nuclear composition
rightward until it reaches an unstable box. ␤⫺ decay (electron emission, see later) then
takes place and a neutron is transformed into a proton. This results in a diagonal move-
ment in the chart of nuclides (upward and leftward).
The lower chain in figure exemplifies the r-process: several neutrons are added in a
rapid sequence until a highly unstable Z-N combination is attained (crosshatched boxes),
␤⫺ decay then takes place.
The p-process takes place in supernovas by addition of protons followed by ␤⫹ decay
(positron emission). s-process nuclides are also transformed into nuclides of higher Z by
photodisintegration reactions (gamma rays eject neutrons). Photodisintegration (by cos-
mic ray particles) is also responsible for the formation of the light nuclides 6Li, 9Be, 10 B,
and 11B by fragmentation of the nuclei of carbon, nitrogen, and oxygen. Table 8.1 lists in
an extremely concise fashion the main nuclear reactions that lead to the formation of the
light nuclides.
All elements with Z ⬎ 83 (Bi) are unstable and belong to chains of radioactive
decay, or decay series. Three decay series include all radioactive elements in the
Z ⬎ 83 part of the chart of nuclides—namely, 4n, 4n ⫹ 2, and 4n ⫹ 3 (because
the decay takes place by ␣ emission with mass decrease of four units, or by ␤
emission with a negligible mass decrease, all nuclides within a series differ by
Figure 11.1 Chart of nuclides. •: stable, °: unstable. Reproduced with modifications from
Rankama (1954).
710 / Methods
Figure 11.2 Enlarged portions of Segré chart of nuclides, showing s-process (upper
chain) and r-process (lower chain). White boxes: stable nuclides; diagonally ruled boxes:
unstable nuclides; crosshatched boxes: highly unstable nuclides.
multiples of four, and hence they are in a 4n-type series, with n an integer). The
major branches of the three decay series are shown in figure 11.3. The parent
nuclide of 4n series is 232Th, which has 208Pb as its stable end-product. The 4n ⫹
2 series has 238U as parent nuclide and 206Pb as stable end-product, and the 4n ⫹
3 series has 235U as parent nuclide and 207Pb as stable end-product.
All naturally occurring radioactive substances (besides those specified in the
Isotope Geochemistry / 711
Table 11.1 Nucleosynthesis and the main nuclear reactions taking place in stars. * ⫽ unstable;
n ⫽ neutron; ␯ ⫹ ⫽ neutrino; ␤ ⫹ ⫽ positron; ␥ ⫽ gamma radiation.
NUCLEOSYNTHESIS (3 minutes after the Big Bang → 1 hour after the Big Bang; log T ⬵ 9;
1
H ⫹ n → 2D ⫹ ␥ 2
D ⫹ 1H → 3He ⫹ ␥ 3
He ⫹ 3 He → 4He ⫹ 1H ⫹ 1H
HYDROGEN BURNING (first-generation stars; log T ⬵ 7)
1
H ⫹ 1H → 2D ⫹ ␤ ⫹ ⫹ ␯ ⫹ 2
D ⫹ 1H → 3He ⫹ ␥ 3
He ⫹ 3He → 4He ⫹ 1H ⫹ 1H
3
He ⫹ 4He → 7Be ⫹ ␥
7
Be ⫹ e⫺ → 7Li ⫹ ␯ ⫹ 7
Li ⫹ 1H → 8Be* → 4He ⫹ 4He
7
Be ⫹ 1H → 8B ⫹ ␥ 8
B → 8Be* ⫹ ␤ ⫹ ⫹ ␯ ⫹ 8
Be* → 4He ⫹ 4He
CARBON-NITROGEN CYCLE (second- and later-generation stars)
12
C ⫹ 1H → 13N ⫹ ␥ 13
N → 13C ⫹ ␤ ⫹ ⫹ ␯ ⫹
13
C ⫹ 1H → 14N ⫹ ␥ 14
N ⫹ 1H → 15O ⫹ ␥ 15
O → 15N ⫹ ␤ ⫹ ⫹ v ⫹
15
N ⫹ 1H → 12C ⫹ 4He
HELIUM-BURNING (red giants; log T ⬵ 8)
4
He ⫹ 4He → 8Be* 8
Be* ⫹ 4He → 12C ⫹ ␥ 12
C ⫹ 4He → 16O ⫹ ␥
16
O ⫹ 4He → 20Ne ⫹ ␥
CARBON-BURNING reactions (log T ⬵ 8.8)
12
C ⫹ 12C → 24Mg ⫹ ␥
12
C ⫹ 12C → 23Mg ⫹ n
12
C ⫹ 12C → 23Na ⫹ 1H
12
C ⫹ 12C → 20Ne ⫹ ␣
12
C ⫹ 12C → 16O ⫹ 2␣
OXYGEN-BURNING reactions (log T ⬵ 9.4)
16
O ⫹ 16O → 31S ⫹ n
16
O ⫹ 16O → 31P ⫹ 1H 31
P ⫹ ␥ → 30Si ⫹1H 30
Si ⫹ ␥ → 29Si ⫹ n
29
Si ⫹ ␥ → 28Si ⫹ n
16
O ⫹ 16O → 28Si ⫹ ␣
decay series of figure 11.3) are shown in table 11.2, together with their half-lives,
isotopic abundances, and decay products. All the listed nuclides with A ⱖ 40 are
the results of early cosmic events. Lighter radionuclides are continuously pro-
duced in the lower atmosphere by interaction of neutrons (produced by cosmic
rays and slowed to thermal energies) with atmospheric nitrogen, oxygen, and
argon nuclei. Note that ␤⫺ emission and ␣ decay are by far the most common
decay processes for natural radionuclides. Positron emission ( ␤⫹ ) is observed only
in the decay of 26Al, 36C1, 59Ni and in a small branch of 40 K (0.001%; see section
11.7.3). Electron capture (EC) is also a quite limited process.
Figure 11.3 Major branches of decay chains of heavy radionuclides. heavy arrows: ␣ decay: light arrows: ␤ decay. a ⫽ years, d ⫽ days,
h ⫽ hours, s ⫽ seconds.
Table 11.2 Naturally occurring radioactive substances. a ⫽ years. d ⫽ days.
Radionuclide Decay Process Half-Life Isotopic Abundance (%) Stable End-Product

3
H ␤⫺ 12.26 a 1.38 ⫻ 10⫺4 3
He
7 7
Be EC 53.3 d – Li
10
Be ␤⫺ 1.5 ⫻ 106 a 19.9 10
B
14
C ␤⫺ 5730 a – 14
N
26
Al ␤ ⫹, EC 7.16 ⫻ 105 a – 26
Mg
32
Si ␤⫺ 276 a – 32
P
36
Cl ␤ ⫹, EC, ␤ ⫺ 3.08 ⫻ 105 a – 36
S, 36Ar
37 37
Ar EC 34.8 d – Cl
39
Ar ␤⫺ 269 a – 39
K
40
K ␤⫺, EC, ␤ ⫹ 1.25 ⫻ 109 a 0.0117 40
Ca, 40Ar
50
V ␤⫺, EC 6 ⫻ 1015 a 0.25 50
Cr, 50Ti
53
Mn EC 3.7 ⫻ 106 a – 57
Cr
59
Ni EC, ␤ ⫹ 8 ⫻ 104 a – 59
Co
81
Kr EC 2.13 ⫻ 105 a – 81
Br
85
Kr ␤⫺ 10.6 a – 85
Rb
87
Rb ␤⫺ 4.88 ⫻ 1010 a 27.835 87
Sr
113
Cd ␤⫺ 9 ⫻ 1015 a 12.22 113
In
115
In ␤⫺ 5.1 ⫻ 1014 a 95.7 115
Sn
123
Te EC 1.2 ⫻ 1013 a 0.908 123
Sb
138
La EC, ␤⫺ 2.69 ⫻ 1011a 0.089 138
Ba, 138Ce
142
Ce ␣ 艐5 ⫻ 1016 a 11.07 138
Ba
144
Nd ␣ 2.1 ⫻ 1015 a 23.80 140
Ce
147
Sm ␣ 1.06 ⫻ 1011 a 15.0 143
Nd
148
Sm ␣ 7 ⫻ 1015 a 11.3 144
Nd
149
Sm ␣ ⬎1 ⫻ 106 a 13.8 145
Nd
152
Gd ␣ 1.1 ⫻ 1014 a 0.20 148
Sm
174
Hf ␣ 2.0 ⫻ 1015 a 0.162 170
Yb
176
Lu ␤⫺ 3.57 ⫻ 1010 a 2.59 176
Hf
187
Re ␤⫺ 4.23 ⫻ 1010 a 62.60 187
Os
190
Pt ␣ 6 ⫻ 1011 a 0.013 186
Os
232
Th Decay chain 1.40 ⫻ 1010 a 100 208
Pb
234
U Decay chain 2.44 ⫻ 105 a 0.0055 206
Pb
235
U Decay chain 7.04 ⫻ 108 a 0.720 207
Pb
238
U Decay chain 4.47 ⫻ 109 a 99.2745 206
Pb
11.2 Nuclear Energy
Nuclear mass is always lower than the sum of proton and neutron masses because
of mass-energy conversion, which transforms part of the mass into binding en-
ergy. For example, 4He has mass 4.002604, whereas the total mass of two protons
and two neutrons is 4.032981. Based on Einstein’s equation
E = mc 2 (11.1)
(where c is the speed of light: 2.99792458 ⫻ 108 m ⭈ s⫺1 ), the mass differential
714 / Methods
Figure 11.4 Mean binding energy per nucleon as a function of A. From G. Friedlander,
J. W. Kennedy, E. S. Macias, and J. M. Miller, Nuclear and Radiochemistry, copyright
1981 by John Wiley and Sons. Reprinted by permission of John Wiley & Sons.
dm = 4.032981 − 4.002604 = 0.030377 (11.2)
corresponds to a binding energy of 28.3 MeV (i.e., about 7.1 MeV per nuclear
constituent). The average binding energy per nucleon is roughly constant (7.4 to
8.8 MeV), except in a few of the lightest nuclei. It increases slightly from light
elements to A ⫽ 60 (iron and nickel nuclei) and then decreases more or less
linearly with increasing A (figure 11.4).
Both protons and neutrons in the nucleus have an intrinsic angular momen-
tum resulting from spinning of nucleons,
h
P=I × , (11.3)
2π
and an orbital angular momentum
L = m×v (11.4 )
that is an integral multiple of h/2␲, being associated with motion along nuclear
orbits with angular frequency ␻ ⫽ 2␲v. Because the intrinsic spin of protons and
neutrons is I ⫽ 12, the nuclear spin of nuclides with odd A is 12; if A is even, it is
zero or an integer. Because rotation of a charged particle produces a magnetic
moment, nuclides with nuclear spins different from zero also have a magnetic
moment of magnitude ZeP/2mc. The magnetic moment of a single proton is
1
taken as the unit of nuclear magnetic moment, corresponding to 658 of the Bohr
magneton (magnetic moment of electron).
Because experimental evidence indicates that the volume and total binding
energy of nuclei are proportional to the number of nucleons, it can be deduced
that nuclear matter is incompressible and that each nucleon is able to interact
only with a limited number of surrounding nucleons. This evidence leads to for-
mulation of the binding energy and mass of nuclei in terms of a “charged drop”
model—i.e., the nuclei are assimilated to charged liquid drops with surface ten-
sion. Total binding energy EB is the result of various energy contributions, each
with simple functional dependence on mass A and charge Z of the nucleus (Myers
and Swiatecki, 1966):
  N − Z    N − Z 
2 2
E B = c1 A1 − k    − c 2 A 1 − k   
2 3
  A     A  
  (11.5)
− c 3Z 2 A −1 3 + c 4Z 2 A −1 + δ .
Although equation 11.5 has only six adjustable parameters, it is able to reproduce
the energy of the approximately 1200 nuclides of known mass. When binding
energy is expressed in MeV, the empirically adjusted coefficients have the follow-
ing values: c1 ⫽ 15.677 MeV, c2 ⫽ 18.56 MeV, c3 ⫽ 0.717 MeV, c4 ⫽ 1.211 MeV,
and k ⫽ 1.79. ␦ is associated with the number of protons (if A is odd, ␦ ⫻ A ⫽
0; if it is even, ␦ ⫻ A ⫽ ⫾132).
The first term on the right in equation 11.5 is the volume energy, proportional
to the number of protons and neutrons in the nucleus. It contains a symmetry
correction term, proportional to (N ⫺ Z)2 /A2, accounting for the fact that nuclear
forces are at maximum when N ⫽ Z and decrease symmetrically on both sides of
N ⫽ Z. The second term is the surface energy, which accounts for the fact that
nucleons at the surface of the nucleus have unsaturated forces that reduce binding
energy proportionally to A2/3 (and hence to the surface, A being proportional to
volume). The third term is the electrostatic Coulomb interaction, accounting for
Coulomb repulsion between protons, which have homologous charge. This term
also reduces the binding energy of the nucleus. Because the Coulomb interaction
is proportional to Z2, it becomes increasingly important with the increase of Z
and is responsible for the observed progressive relative increase of N/Z (see figure
11.1). The fourth term in equation 11.5 is a correction for Coulomb energy, re-
sulting from the nonuniform charge distribution (or “diffuse boundary”).
11.3 Nuclear Decay
All nuclides are generally subdivided into four types, depending on whether they
contain even or odd numbers of protons and neutrons, as shown in table 11.3.
716 / Methods
Table 11.3 General classification of nuclides: significance of parity is related to symmetry

properties of nuclear wave functions. A nuclide is said to have odd or even parity if the sign
of the wave function of the system respectively changes or not with changing sign in all
spatial coordinates (see Friedlander et al., 1981 for more detailed treatment). The value
assigned to ␦xA is appropriate for A ⬎ 80. For A ⬍ 60, a value of ⫾65 is more appropriate.
Z-N Combination Type Parity Example Number of Nuclides ␦xA

Even-even 1 Even 16
8O 164 ⫺132
9
Even-odd 2 Odd 4 Be 55 0
10
Odd-even 3 Odd 5B 50 0
Odd-odd 4 Even 7
3 Li 4 ⫺132
The occurrence of nuclides in nature reflects relative stability. In general, nu-

clides whose mass numbers are multiples of 4 are exceptionally stable, because a
group of two protons and two neutrons forms a closed shell (nuclear shells are to
some extent analogous to the electron shells discussed in chapter 1).
11.3.1 Beta Decay

Beta decay is the most common decay process (either natural or artificial): a neu-
tron is transformed into a proton by emission of a ␤⫺ particle (electron):
n → p + β−. (11.6)
Alternatively, a proton is transformed into a neutron by emission of a ␤⫹ parti-

cle (positron):
p → n + β+. (11.7 )
Mass number A does not change in either case.

Because the mass of the nucleus is related to the masses of protons Mp and
neutrons Mn through
(
M = ZMP + A − Z Mn − E B , ) (11.8)
by combining equations 11.5 and 11.8 we obtain
( )( ) ( )( ) ( )
2
M = f1 A Z − ZA + f 2 A Z − ZA − δ A . (11.9)
Because, for a constant value of A, f 1(A), f 2(A), and ␦(A) are also constant,
equation 11.9 is a parabola (for a given value of ␦ ) with maximum at Z ⫽ ZA :
∂M
=0 → ZA = −
( ).
f2 A
2 f ( A)
(11.10)
∂Z 1
Figure 11.5 shows the nuclear energy parabolas for odd (A ⫽ 125) and even
(A ⫽ 128) nuclides. These parabolas represent sections of the nuclear energy sur-
face on the Z-EB plane. Because ␦ is zero at any odd value of A, there is a single
parabola, whereas for an even A there are two parabolas, separated along the
energy axis by 2␦ /A.
The usefulness of plots such as those in figure 11.5 stems from the fact that
they give approximate values of the energy available for ␤ decay between neigh-
boring isobars. For odd A (figure 11.5A) there is only one stable nuclide—i.e.,
the one nearest to the minimum of the curve. Note here that, as is customary,
energy parabolas are drawn in a reverse fashion, with the minima actually corre-
sponding to maxima in the Z-EB plane. Note also that in some instances the
energy minimum may be achieved at a noninteger value of Z, because equation
11.5 is continuous in Z. In these cases, the actual minimum is attained at the
integer value of Z nearest to the calculated one.
A nuclide of type 1 emitting ␤⫹ or ␤⫺ particles is transformed into a type 4
nuclide, lying on a parabola translated by 2␦/A (figure 11.5B). After a new ␤⫹ or
␤⫺ emission, the type 4 nuclide is reconverted into a type 1 nuclide, and so on.
In a ␤ emission chain, nuclides of different nuclear charges alternate between the
two energy parabolas in the energy–nuclear charge field. The ␤ decay process
operating on even A parent nuclides may result in two or more stable isotopes of
the even-even type. For example, in figure 11.5B, both 128 128
52 Te and 54 Xe can be
considered stable, having a binding energy lower than that of the odd parent 128 53 I.
The energy associated with ␤ emission during nuclear decay is not constant,
but varies from a minimum value near zero to a maximum corresponding to the
energy gap between the parent nuclide and the radiogenic nuclide. This apparently
contrasts with the principle of energy conservation. However, as shown by Pauli
(1933), a third particle participates in the process. This particle is called a “neu-
trino” ( ␯ because of its very limited mass and charge. The existence of the neu-
trino (and of its corresponding “antineutrino,” ␯ ), postulated on purely theoreti-
cal grounds by Pauli, has only recently received experimental confirmation.
Beta particles do not transfer all the binding energy freed by the decay pro-
cess; part of it resides in neutrinos. The mean ␤ emission frequency is one-third
of the energy gap involved in nuclear decay; in other words, most emitted ␤ parti-
cles have kinetic energy corresponding to one-third of the energy gap. Only a few
particles are emitted with a frequency corresponding to maximum energy. Emit-
ted ␤ particles may penetrate surrounding matter before losing most of their ki-
netic energy. Secondary electrons hit by emitted ␤ particles also have high energy
and may be expelled by their localized orbitals, causing successive ionizations.
Figure 11.5 Energy versus nuclear charge plots for odd parity (A) and even parity (B) nuclides. Experimentally observed values of binding
energy differences are reported in parentheses for comparative purposes. From Nuclear and Radiochemistry, G. Friedlander and J. W. Kennedy,
Copyright 1956 by John Wiley and Sons. Reprinted by permission of John Wiley & Sons, Ltd.
Besides the production of positive ions, secondary electrons may be captured by

saturated atoms or molecules, resulting in negative ions.
If the available decay energy (mass difference between decaying and produced
nuclides) exceeds 1.02 MeV, one positron and one electron may be annihilated
reciprocally, with the emission of two ␥ photons of energy corresponding to the
masses of the annihilated particles—i.e.,
E = 2 mc 2 = 1.02 MeV. (11.11)
Let us consider, for instance, the decay process of the unstable oxygen isotope
14 14 14
8 O, which is transformed into stable 7 N (figure 11.6). The 8 O decays, emitting
⫹
two series of ␤ particles of different energies. In decay process (1) in figure 11.6,
a 14
7 N isomer forms in an unstable excited state, and reaches stability by emission
of ␥ rays of different energies. The first ␥ ray has an energy of 2.313 MeV and
leads to an excited state 1.02 MeV above the ground level. The second decay step
involves positron-electron interaction and leads to emission of two ␥ photons
with energies of 1.02 MeV, which reduce the energy of the daughter nuclide at the
⫹
7 N. The energy of the ␤
ground state of stable 14 particles involved in decay pro-
cess (2) is higher (4.1 MeV) and leads directly to the excited state preceding
positron-electron interaction.
We can summarize the ␤ decay process with two examples:
40
19 K → 40
20 Ca + β − + v + 1.312 MeV (11.12 )
and
8 O + β + v + 1.655 MeV.
→ 18
18 +
9F (11.13)
Figure 11.6 Energy scheme for ␤ + decay of unstable 148O.

720 / Methods
11.3.2 Electron Capture

Electron capture is a particular form of ␤ decay, described for the first time by
Alvarez (1938) and practically equivalent to positron emission. When the decay
energy of the nuclide is not sufficient to attain the reaction
p → n + β+ (11.14 )
(i.e., 1.02 MeV; cf. eq. 11.11), an electron of the inner K level enters the nucleus,
leaving behind a vacancy that is occupied by a more external electron. Vacancy
migration proceeds toward the external orbitals and is associated with X-ray
emission. An example of electron capture (EC) decay is
57 La + 56 Ba + hv ( γ ) − hv ( X ).
e − → 138
138
(11.15)
11.3.3 Alpha Decay

Alpha particles are composed of two protons and two neutrons. Thus they have
Z ⫽ 2, N ⫽ 2, and A ⫽ 4 and correspond to a helium nucleus 42He. The emission
of ␣ particles thus produces a decrease of 4 units in A. An unstable nuclide under-
going ␣ decay may emit ␣ particles of various energy and thus directly reach the
ground level of the stable product. Alternatively, as in ␤ emission, an intermediate
excited state is reached, followed by ␥ emission. Figure 11.7 shows, for example,
the decay process of 228 224
90 Th, which may directly attain the ground level of 88 Ra by
emission of ␣ particles of energy 5.421 MeV or intermediate excited states by

emission of ␣ particles of lower energy, followed by ␥ emission.
Each ␥-ray energy emission connecting two excited states corresponds to the
difference between the disintegration energy associated with the two ␣ decays,
which lead to the two limiting excited conditions.
11.3.4 Gamma Transitions

As we saw above, when a radiogenic nuclide (or “radionuclide”) in an excited
state is produced as an intermediate step of the decay process, the ground level is
often attained by emission of electromagnetic ␥ radiation of high energy (gener-
ally higher than 100 keV). Gamma radiations interact with surrounding matter,
giving rise to the photoelectric effect, the Compton effect, and internal conversion.
Photoelectric Effect
Emitted ␥ radiation interacts with electrons of the surrounding matter. If the fre-
quency of the emitted radiation exceeds the energy level corresponding to the
ionization potential of the element, the electron may be expelled from its localized
Figure 11.7 Energy scheme for ␣ decay of 228

90 Th.
orbital (K-level electrons are normally involved in the process). Because hv is the
energy of the ␥ radiation and W is the energy necessary to expel the electron from
its orbital (or “work function of the metal,” an important parameter in mass
spectrometry), expulsion will take place only when hv is greater than W. In this
case, the expelled electron achieves kinetic energy corresponding to the surplus:
1
mv 2 = hv − W . (11.16)
2
Compton Effect
When ␥ radiation hits an electron, it is deviated from its original trajectory and,
losing energy, changes its frequency. The increase in frequency consequent upon
anelastic scattering is only a function of the angle between incident and deflected
rays and does not depend on the energy of the incident radiation. The importance
of this fact, known as the Compton effect, increases as the atomic number of the
element decreases.
Internal Conversion
Expelled electrons belong to the K-level of the emitting atom. Also in this case,
as in K-capture, the created vacancy is progressively occupied by more external
electrons and is associated with X-ray emission.
722 / Methods
11.4 Isotopic Decay
The decay rate of a given radionuclide is generally constant and independent of

the past history of the nuclide and of the P, T (at least below 108 K), and chemis-
try of the system. The probability P of the radionuclide disintegrating over time
⌬t is
P = λ ∆t, (11.17)
where ␭ is the proportionality constant. Conversely, the probability of the radio-

nuclide not disintegrating in the same lapse of time is
1 − P = 1 − λ ∆t. (11.18)
This probability remains unchanged even if the radionuclide has survived the first
elapsed time increment. Combining the single probabilities for n elapsed time
intervals, we obtain
n
N  t
= 1 − λ  , (11.19)
N0  n
where N 0 is the number of initial radionuclides and N is the number of remaining

radionuclides after disintegration at time t:
t = n × ∆t. (11.20)
Reducing time interval ⌬t to an infinitesimal amount dt and applying to equation

11.19 the limit
n
lim 1 + x  = exp( x ) ,
n → ∞   (11.21)
n
we obtain
N
= exp( − λ t) (11.22)
N0
—i.e., the rate of a “first-order reaction” (cf. section 8.21.1).

The amount of time t1/2 required to reduce by one-half the initial number N 0
of radionuclides is called the half-life of the species:
 N   1
ln   = ln   = − λ t1 2 (11.23)
 N0  2
ln 2 0.69315
t1 2 = = . (11.24)
λ λ
The average life ␶ of a given radionuclide corresponds to the inverse of the decay
constant—hence, to 1/ln 2 times the half-life t1/2, as obtained by integration of
the time of existence of all radionuclides divided by the initial number N 0 :
t=∞
1 t1 2 1
τ =−
N0
∫ tdN =
ln 2
=
λ
. (11.25)
t=0
11.5 Growth of Radioactive Products
Because in nature the ground state of a stable nuclide is often attained by decay
chains involving intermediate species decaying at different rates, it is worth evalu-
ating the implications of the relative magnitudes of the various decay constants
on the isotopic composition of the element.
If we call N1 the number of nuclides of the first species, decaying at rate ␭1 ,
the number of disintegrated nuclides of type 1 per increment of time dt is
dN1
= − λ 1N1 . (11.26)
dt
The population of type 2 nuclides produced by decay of N1 is affected both by

the rate of production (corresponding to the decay rate of species 1—i.e., eq.
11.26 changed in sign) and by its own rate of disintegration:
dN 2
= λ 1N1 − λ 2 N2 . (11.27)
dt
Several substitutions yield the following equation, relating the population of the
second decaying species to elapsed time t:
λ1 N0,1
N2 =
λ 2 − λ1
[ exp (− λ t ) − exp (−λ t )] + N
1 2 0,2 ( )
exp − λ 2 t , (11.28)
where N0,1 and N0,2 are the populations of species 1 and 2 at time t ⫽ 0.
724 / Methods
We can now distinguish three general cases, depending on whether the first
decaying species has a longer, a much longer, or a shorter half-life than that of
the daughter nuclide. These three cases are transient equilibrium, secular equilib-
rium, and nonequilibrium.
11.5.1 Transient Equilibrium

Figure 11.8A shows the radioactivity emitted by a mixture of two independently
decaying species 1 and 2, detected with detection coefficients c1 and c2 (indepen-
dent and not necessarily identical, but identical here, for the sake of simplicity).
Total radioactivity R T is the sum of separate radioactivities of the two species
and depends on their half-lives (t1/2,1 ⫽ 8 hr and t1/2,2 ⫽ 0.8 hr for the cases
exemplified in figure 11.8) according to
R T = R 1 + R 2 = c1 λ1 N1 + c2 λ 2 N2 . (11.29 )
As shown in figure 11.8A, in the case of independently decaying species, compos-

ite decay curve a attains the limiting behavior imposed by the long-lived nuclide.
Let us consider now species 1 and 2 linked in a decay chain with the parent
nuclide 1 being longer-lived than the daughter nuclide 2 (i.e., ␭1 ⬍ ␭2 ). After a
relatively short time, the terms exp(⫺␭2t) and N0,2 exp(⫺␭2t) become negligible
with respect to exp(⫺␭1t). As a result, equation 11.28 is reduced to
λ1 N 0,1
N2 =
λ 2 − λ1
(
exp − λ1t ) (11.30)
and, because
( )
N 0,1 exp − λ1t = N1 , (11.31)
the final result is
N1 λ − λ1
= 2 . (11.32)
N2 λ1
As shown in figure 11.8B, because the ratio N1 /N2 becomes constant, the
slopes of the combined decay curves of the two radionuclides attain a constant
value corresponding to the half-life of the longer-lived term (curves a and b in
figure 11.8B). Moreover, assuming identical detection coefficients for the two spe-
cies, their radioactivity ratio also attains a constant value of
R 1 λ 2 − λ1
= . (11.33)
R2 λ2
Figure 11.8 Composite decay curves for (A) mixtures of independently decaying species,
(B) transient equilibrium, (C) secular equilibrium, and (D) nonequilibrium. a: composite
decay curve; b: decay curve of longer-lived component (A) and parent radio nuclide (B,
C, D); c: decay curve of short-lived radionuclide (A) and daughter radionuclide (B, C, D);
d: daughter radioativity in a pure parent fraction (B, C, D); e: total daughter radioactivity
in a parent-plus-daughter fraction (B). In all cases, the detection coefficients of the various
species are assumed to be identical. From Nuclear and Radiochemistry, G. Friedlander and
J. W. Kennedy, Copyright 1956 by John Wiley and Sons. Reprinted by permission of
John Wiley and Sons Ltd.
11.5.2 Secular Equilibrium

If the decay constant of the parent radionuclide is much lower than that of the
daughter radionuclide (i.e., ␭1 ⬍⬍ ␭2 ), equation 11.32 can be reasonably re-
duced to
726 / Methods
N1 λ
= 2. (11.34)
N2 λ1
Assuming identical detection coefficients for the two species, the radioactivity
ratio obviously reduces to 1. This condition, known as “secular equilibrium,” is
illustrated in figure 11.8C for t1/2,1 ⫽ ∞ and t1/2,2 ⫽ 0.8 hr. Secular equilibrium
can be conceived of as a limiting case of transient equilibrium with the angular
coefficient of decay curves progressively approaching the zero slope condition
attained in figure 11.8C.
11.5.3 Nonequilibrium
The case of “nonequilibrium” occurs whenever the decay constant of the parent
nuclide is higher than the decay constant of the daughter nuclide (i.e., ␭1 ⬎ ␭2 )
and is illustrated in figure 11.8D for ␭1 / ␭2 ⫽ 10. As the parent decays, the number
of daughter nuclides progressively rises, reaching a maximum at time tm. It then
decreases in a constant slope, characteristic of its own half-life. Time tm is found
graphically by intersecting the decay curve of the parent (curve b in figure 11.8D)
and the extrapolation of the final decay curve to time zero (curve c). The same
parameter can be obtained analytically by applying the equation
ln λ 2 − ln λ1
tm = . (11.35)
λ 2 − λ1
11.6 Isotope Fractionation
11.6.1 Equilibrium Exchanges
Isotopes of the same element may fractionate during homogeneous or heteroge-

neous chemical reactions. The extent of fractionation increases with decreasing
T and increasing mass contrast, and is favored by changes in coordination, the
force constant being equal (see later in this section). The fractionation of isotopes
of the same element during chemical exchanges is a quantum effect. In quantum
chemistry, partition function Q (cf. section 3.1) may in a first approximation be
written as a product of translational, rotational, and vibrational terms:
Q = Q trans × Q rot × Q vib . (11.36 )
The translational partition function Qtrans is equal to the classical one at all
temperatures
 ε k ,i , x ε k ,i , y ε k ,i , z 
Q trans = ∑ exp  − − −
kT 
(11.37 )
i  kT kT
where ␧k,i,x is the kinetic energy of molecule i along direction x. Also the rota-
tional partition function Qrot is equal to the classical one, with the exception of
hydrogen (see section 11.8.2).
In the harmonic approximation, Qvib is given by
(
exp − Xi 2 )
Q vib = ∏ 1 − exp( − X )
(11.38)
i i
(Bigeleisen and Mayer, 1947) where
hvi បωi
Xi = = (11.39)
kT kT
(see section 3.3 for the significance of the symbols).

At high T (low Xi ), Qvib reduces to ⌸i(1/Xi ). Because the potential energies
of molecules differing only in isotopic constituents are alike, one can define a
“separative effect” based on the partition function ratio f of isotopically heavy
•
and light molecules (Q and Q°, respectively) in such a way that the rotational
and translational contributions to the partition function cancel out:
 s •   Q •   m° 
32
 s• 
 o  f =  o  o  • 
s   s  Q  m 
(11.40)
exp − Xi • 2 1 − exp − Xi •  
Xi •     
= ∏ .
i Xi ° ( )[
exp − Xi ° 2 1 − exp − Xi ° ( )]
•
The ratio of symmetry numbers s /s° in equation 11.40 merely represents the
relative probabilities of forming symmetrical and unsymmetrical molecules, and
•
m and m° are the masses of exchanging molecules (the translational contribution
to the partition function ratio is at all T equal to the 32 power ratio of the inverse
molecular weight). Denoting as ⌬Xi the vibrational frequency shift from isotopi-
cally heavy to light molecules (i.e., ⌬Xi ⫽ X °i ⫺ X •i ) and assuming ⌬Xi to be
intrinsically positive, equation 11.40 can be transated into
 s•  1 − exp  −  Xi • + ∆Xi 
 
Xi •  ∆Xi  
 f = ∏ exp  . (11.41)
 s°  i Xi • + ∆Xi  2  1 − exp − Xi • 
 
728 / Methods
The Helmholtz free energy of each component in reaction is related to the

partition function Q:
F = − kT ln Q (11.42 )
Based on equation 11.41, the difference between the Helmholtz free energies of
formation of two isotopic molecules with respect to their gaseous atoms depends
on the shift of vibrational frequencies between heavy and light isotope-bearing
compounds—i.e., according to Bigeleisen and Mayer (1947),

 1   1 − exp  − Xi •  
∆F • − ∆F ° ∆ Xi  
= ∑  − ∆X i + ln  1 +  + ln 
kT i  2  Xi •

 (
1 − exp − Xi  °
)
  (11.43)
 s• 
+ ln   ,
 s° 
•
where ⌬F and ⌬F° are the Helmholtz free energies of formation of isotopically
heavy and light molecules. If ⌬Xi is small, which is the case for all isotopes except
hydrogen, equation 11.43 can be reduced to
 
∆F • − ∆ F ° 1 1 1  s• 
= −∑  − +  ∆ Xi + ln  °  , (11.44)
kT i  2 Xi • exp  Xi •  − 1  s 
   
•
and the separative effect (s /s°) f becomes
 
 s•  1 1 1 
 °  f = 1+ ∑  − +  ∆Xi . (11.45)
s  i  2 Xi • exp Xi •  − 1 
   
•
If X •i is also small, then the terms in brackets in equation 11.45 approach X i /12
and the separative effect reduces to
•
 s•  Xi ∆Xi
 °  f = 1+
s 
∑ 12
. (11.46)
i
•
Because ⌬Xi is positive, (s /s°) f is always greater than 1, and the heavy isotope
preferentially stabilizes in the condensed phase whereas the light isotope favors
the gaseous phase.
Figure 11.9 Separative effect per unit shift, plotted against nondimensionalized frequen-
cies X ⫽ h␯ /kt ⫽ ប␻ /kT. Reprinted from Bigeleisen and Mayer (1947), with permission
from the American Institute of Physics.
At low temperatures and light frequencies, the separative effect per unit shift
(the terms in brackets in eq. 11.44 and 11.45) approaches 12 (figure 11.9) and the
Helmholtz free energy difference approaches the differences in zero-point ener-
gies. At high T (low frequencies), the separative effect per unit shift approaches
•
zero and the total separative effect (s /s°) f approaches 1, so that no isotopic frac-
tionation is observed.
A useful approximation for the isotopes of heavy elements (Bigeleisen and
Mayer, 1947) is
 s•  M ∆m 2
 °  f = 1+ X s n, (11.47)
s  24 m 2
where m is the mass of the central atom (isotope) surrounded by n ligands of mass
M, ⌬m is the difference in isotopic masses, and nondimensionalized frequency Xs
is related to the totally symmetric frequency arising from the stretching
of bonds with central atom ns by
hνs h 2πC
Xs = = (11.48)
kT kT M
where C is a force constant.

Table 11.4 lists the results of calculations based on equation 11.47 for isotopic
exchange reactions at T ⫽ 300 K for tetrahedrally and octahedrally coordinated
“heavy” isotopes:
730 / Methods
Table 11.4 Separative effect f and isotopic

fractionation constant K for heavy isotopes,
computed through equation 11.47. ␻s is angular
totally symmetric stretching frequency derived
from Raman spectra (see Bigeleisen and Mayer,
1947 for references).
Molecule ␻s f K
SiF4 800 1.111
1.002
SiF62⫺ 600 1.109
SnCl4 367 1.00256
1.00025
SnCl62⫺ 314 1.00281
28
SiF4 + 30
SiF62 − ⇔ 30
SiF4 + 28
SiF62 − (11.49 )
120
SnCl 4 + 118 SnCl 26 − ⇔ 118 SnCl 4 + 120 SnCl 26 − . (11.50 )
Note that the separative effect is quite significant, even for the relatively “heavy”
isotopes of tin. Note also the large cancellation effect on the isotopic fractionation
constant, arising from superimposed partial reactions such as
120
SnCl 4 + 118 Sn → 118 SnCl 4 + 120 Sn (11.51)
and
118
SnCl 26 − + 120 Sn → 120 SnCl 26 − + 118 Sn . (11.52 )
This cancellation (always marked in exchanges of acid-base type) can be ascribed

to the decrease in force constant C of the central atom with increasing coordina-
tion number n, and is sometimes not so significant in other types of isotopic
exchange reactions.
As suggested by Urey (1947), an isotopic exchange can be expressed by a
reaction such as
a A ° + b B • ⇔ a A• + b B ° , (11.53)
• •
where A°, A and B°, B are molecules of the same type that differ only in the
isotopic composition of a given constituent element (open and full symbols: light
and heavy), and a and b are stoichiometric coefficients.
In light of equation 11.42, the equilibrium constant for the isotopic exchange
reaction of equation 11.53 reduces to
(Q )
a
•
A Q A°
K 53 = . (11.54)
(Q )
b
•
B Q B°
The equivalent formulation of equation 11.38 for a crystalline solid is
3N − 6
(
exp − Xi 2 ),
Q= ∏ 1 − exp( − X )
(11.55)
i =1 i
where N is the number of atoms in the crystal, and the ratio of partition functions
of the isotopically heavy and light compounds (Kieffer, 1982) is
Q• 3N − 6
(
 exp − Xi • 2 )  1 − exp( −X )  . °
∏
i
= 
Q° i =1 1 − exp − Xi •
 ( )   exp( −X 2)  i
° (11.56)
•
Introducing the masses of exchanging isotopes (m and m°, respectively), the
ratio of partition functions for crystalline components can be related to that of
the primitive unit cell (Kieffer, 1982), thus defining “reduced” partition function
ratio f (whose formulation is equivalent to that obtained by Bigeleisen and
Mayer, 1947, and Urey, 1947, for gaseous molecules):
( 3 2 )r
Q • ′ Q •  m° 
f = =   , (11.57)
Q °′ Q °  m• 
where r is the number of exchanging isotopes per unit formula.

Kieffer (1982) proposed detailed calculation of partition function ratio f in
crystalline solids through direct evaluation of the Helmholtz free energies of isoto-
pically light and heavy compounds:
ωl
   − ប ω   
F = 3 nNβ + kT ∫ ln 1 −

exp 

   g (ω )dω ,
 kT   
(11.58)
0
where
3 nN
1
3nNβ = −U0 −
2
∑ ប ω. (11.59)
r =1
In equation 11.58, n is the number of atoms in the chemical formula, g( ␻ ) is the

frequency distribution of normal modes, and ␻l is the highest lattice vibrational
732 / Methods
Figure 11.10 Schematic representation of vibrational spectra for an isotopically light

compound (heavy bottom line) and an isotopically heavy compound (light upper line) of
the same bulk stoichiometry and structure. Reprinted with permission from Kieffer
(1982), Review of Geophysics and Space Physics, 20, 827–849, copyright 1982 by the
American Geophysical Union.
frequency. In equation 11.59, U0 is the potential energy at the bottom of the po-
tential well, and the summation term is zero-point energy.
As shown schematically in figure 11.10, the frequency distribution for an iso-
topically light compound g°(v) [and the corresponding g°( ␻ )] can be expected to
•
differ from the frequency distribution of the isotopically heavy counterpart g (v)
•
[or g ( ␻ ].
Kieffer (1982) suggests a set of rules describing expected frequency shifts for different
types of vibrational modes upon substitution of a heavy isotope into the mineral structure.
Kieffer’s (1982) rules for 18O-16O substitution in silicates are as follows.
Rule 1 (antisymmetric (Si,Al)-O stretching modes)

If ␯1 and ␯3 modes can be identified, their frequencies are assumed to decrease by 0.943
and 0.977, respectively, upon substitution of 18O for 16O. If ␯1 and ␯3 cannot be separately
identified and enumerated, an average shift of 0.96 is adopted.
Rule 2 (symmetric (Si,Al)-O-Si stretching modes)

The frequency shift of symmetric (Si,Al)-O-Si stretching modes upon substitution of 18O
for 16O is taken to be 0.99, for modes involving mainly Si motion.
Rule 3 (Si-O bending modes in orthosilicates)

If the Si-O bending modes in orthosilicates can be identified and enumerated, their fre-
quency shift for substitution of 18O for 16O is taken to be 0.95.
Rule 4 (acoustic modes)

The ratio of frequencies upon substitution of 18O for 16O is taken to be proportional to
the square root of the ratio of the mean molecular masses.
Rule 5 (remaining undifferentiated modes)

The frequency shift of the remaining modes for substitution of 18O for 16O must be consis-
tent with zero fractionation at high T—i.e., at high T the frequency shifts must obey
( 3 2 )r
3s
Xi °  m• 
∏ Xi •
= 
 m° 
, (11.60)
i =1
where s is the number of atoms per unit cell.
Because
F° − F• 3  m° 
ln f = + r ln  •  , (11.61)
kT 2 m 
•
calculations of F° and of the corresponding F for isotopically heavy compounds
(through application of the above general rules for frequency shifts induced by
isotopic substitution) allows determination of partition function ratio f and of
the corresponding “reduced partition function” (103 /r) ln f (see section 11.8.1).
11.6.2 Reaction Kinetics

Let us consider again an isotopic exchange reaction of the type shown in equation
11.53, such as, for instance, the already treated equilibrium between stannous
chloride and an aqueous complex
120
SnCl 4 + 118SnCl 26 − ⇔ 118SnCl 4 + 120SnCl 26 − . (11.62 )
Reaction 11.62 can be conceived of as composed of two partial reactions—i.e.,
120
SnCl 4 + 2 Cl − ⇔ 120SnCl 26 − (11.63)
and
118
SnCl 4 + 2 Cl − ⇔ 118SnCl 26 − . (11.64 )
734 / Methods
To each of these two reactions will apply a reaction constant (K63 and K64 , respec-
tively) that is related to the forward and backward reaction rates by
k + ,63
K 63 = (11.65.1)
k − ,63
and
k + ,64
K 64 = . (11.65.2)
k − ,64
Because at equilibrium,
k k 
∆G 62
0
= − RT ln K 62 = − RT ln  + ,63 × − ,64  , (11.66)
 k + ,64 k − ,63 
and because we have seen that the isotopic fractionation constant at low T differs
from 1 (i.e., K62 ⫽ 1.0025 at T ⫽ 300 K; cf. table 11.4), it follows that isotopically
heavy and light molecules cannot be expected to have identical reaction rates at
low T (even in the case of relatively “heavy” isotopes such as 118Sn and 120 Sn).
The effects discussed above may be particularly important when the fractional
mass difference between isotopically light and heavy molecules is high (see the
calculations of Sutin et al., 1961). A quantitative approach to this problem was
afforded by Bigeleisen (1949) on the basis of transition state theory.
Adopting the same notation developed in the previous section, the reaction
rates of isotopically heavy and light molecules are related to the differences in
•
their respective activation energies of reaction E , E°, through
(
∂ ln k k °
•
)=E •
− E°
, (11.67)
∂T kT 2
where k is Boltzmann’s constant, and to shifts of vibrational spectra, through
( •
∂ ln k k ° )= ∂ ln τ ( •
τ° )+X •2
s − X s°
2
∂T ∂T 12T
 
3 N −6 °
(
)
 1 + Xi − 1 exp Xi
°
1  ∆ Xi*
(11.68)
+ ∑  −  ,
[ ( ) ]
2
i =1  exp X i° − 1 2 T
 
•
where ␶ and ␶° are transmission coefficients (which, for asymmetric reactions
such as reactions 11.63 and 11.64, must be evaluated experimentally), and ⌬X i*
is the vibrational shift induced by formation of the activated complex.
11.6.3 Diffusion
Some aspects of thermally activated diffusion were discussed in section 5.9.1 in
regard to intracrystalline exchange geothermometry. Concerning more precisely
isotope diffusion, there are two main aspects that may be relevant in geochem-
istry:
1. Diffusion-induced isotope fractionation

2. Significance of closure temperature.
Diffusion-Induced Isotope Fractionation

Some aspects of this phenomenon are still controversial, because diffusion-
induced isotopic fractionation effects may be erroneously ascribed to differential
reaction kinetics (see section 11.6.2), and vice versa.
Basically, whenever isotopic exchanges occur between different phases (i.e.,
heterogeneous equilibria), isotopic fractionations are more appropriately de-
scribed in terms of differential reaction rates. Simple diffusion laws are neverthe-
less appropriate in discussions of compositional gradients within a single phase—
induced, for instance, by vacancy migration mechanisms, such as those treated in
section 4.10—or whenever the isotopic exchange process does not affect the ex-
trinsic stability of the phase.
Giletti et al. (1978) interpreted the results of laboratory experiments involving
oxygen isotope exchanges between feldspars and hydrothermal fluids at various
T in terms of self-diffusion of 18O. The experiment of Giletti et al. (1978) is quite
elegant: the mineral is equilibrated for a certain amount of time with isotopically
altered water (40% H18 2 O) and the reacted crystals are analyzed by ion micro-
probe, using the primary ion beam as a drilling device and determining the 18O/
16
O ratio of the sputtered material (corresponding to various depths within the
crystal) by ionization and mass spectrometry. Figure 11.11 shows the experimen-
tal results of Giletti et al. (1978) for adularia. The initial 18O/ 16O ratio of the
natural material is 0.002, and the solid curve is the fitting diffusion equation. The
induced isotopic fractionation is interpreted in terms of “one-dimension diffusion
in a semi-infinite medium with constant surface concentration” (Crank, 1975):
C x − C1  x 
= erf  , (11.69)
C 0 − C1
( )
12
 2 Dt 
736 / Methods
Figure 11.11 18O/ 16O relative concentration profile as a function of depth in adularia
hydrothermally equilibrated with an 18O-enriched fluid (H218O ⫽ 40%). Reprinted from
B. J. Giletti, M. P. Sennet, and R. A. Yund, copyright 1978, with kind permission from
where Cx is the 18O concentration at distance x from the surface of the crystal,
C 0 is the concentration at infinite distance, C1 is the concentration at the surface,
t is time, and D the self-diffusion coefficient.
Equation 11.69 was reexpressed by Giletti et al. (1978) in terms of 18O/ 16O
ratios ␳x , ␳ 0 , and ␳1 as
A × erf(Y ) + ρ1
ρx = , (11.70)
1 − A × erf(Y )
where
ρ 0 − ρ1
A= (11.71)
1 + ρ0
and
x
Y = . (11.72)
2(Dt )1 2
Results were then interpreted in terms of the usual form for thermally activated
diffusion:
 −Ea 
D = D 0 exp  , (11.73)
 RT 
where D 0 is the preexponential factor and Ea is the activation energy of diffusion

(see sections 4.10 and 5.9.1). The results indicate that all feldspars have very simi-
lar oxygen exchange behavior, independent of composition and structural state,
with low activation energy, averaging 24 kcal per gram formula weight of oxygen.
Unfortunately, the interpretation of Giletti et al. (1978) does not solve the problem
of differential diffusivities of 18O and 16O. To do this, their experimental results should be
interpreted in terms of interdiffusion of 18O and 16O. Application of Fick’s first law to
interdiffusion of the two species would in fact lead to the definition of an interdiffusion
coefficient D , so that
D18O D16O
D = (11.74)
X 18O D18O + X 16O D16O
(cf. section 4.11). Because, when X18O → 0,D → D18O, interpretation of the relative concen-
tration profile in terms of self-diffusion is appropriate for the inner parts of the crystals,
where 16O is dominant, but it loses its significance near the borders of grains, where inter-
diffusion effects cannot be neglected (see figure 11.11).
The experiment of Giletti et al. (1978) has important implications for the
isotopic reequilibration of feldspars in hydrous environments. Assuming the
diffusion of 18O to obey diffusion equation 11.73, with D0 ⫽ 2.0 ⫻ 10⫺8 cm2 /s
and Ea ⫽ ⫺24 kcal/mole, and defining F as the fractional net 18O exchange oc-
curring with respect to total net exchange at equilibrium, Giletti et al. (1978)
calculated the length of time required to achieve F ⫽ 0.1 and F ⫽ 0.9 according
to temperature and particle size. The results of their calculations are shown in
figure 11.12.
Any feldspar particle of radius less than 10 ␮m is able to exchange with a
hydrothermal fluid, essentially to completion (i.e., Fⱖ 0.9), in less than 500 years,
if the temperature exceeds 300 °C. However, exchanges for larger crystals may
span periods of time from one million to several hundred million years, depending
on size and T, before isotopic equilibrium is attained. Note also that, because of
the rather low activation energy, exchanges have no single “closure temperature.”
This fact introduces us to the next aspect of the problem.
738 / Methods
Figure 11.12 Time required for diffusional exchange of 18O between feldspar and hydro-
thermal fluid according to particle size and temperature. F ⫽ 0.1 can be taken as condition
of absence of exchanges and F ⫽ 0.9 as one of completion of exchanges. Reprinted from
B. J. Giletti, M. P. Sennet, and R. A. Yund, copyright 䉷 1978, with kind permission from
Significance of Closure Temperature

The concept of closure temperature is particularly important in geochronology.
According to Dodson (1973), closure temperature Tc can be defined as “the tem-
perature of the system at the time corresponding to its apparent age.”
The fundamental difference between the concentration gradients of stable
and radiogenic isotopes can be expressed in terms of the simple equations
∂ C Dt  − t  
= 2 exp   ∇ 2 C 
0
(11.75)
∂t a  τ  
and
Figure 11.13 Relationship between “geochronological closure” (A) and “frozen equilib-
rium” (B). Diffusion parameters are identical, and T ⫺1 is assumed to increase linearly
with time. From M. H. Dodson, Closure temperature in cooling geochronological and
petrological systems, Contributions to Mineralogy and Petrology, 40, 259–274, figure 2,
1973, copyright 䉷 1973 by Springer Verlag. Reprinted with the permission of Springer-
Verlag GmbH & Co. KG.
∂ C* Dt*  − t  
= 2 exp   ∇ 2 C *  + L*
0
(11.76)
∂t a  τ *  
(Dodson, 1973). Both of these equations are developed assuming a linear de-
crease of temperature T with time t (see the analogy of treatment in section 5.9.1).
␶ is a time constant corresponding to the time taken for D to diminish by a factor
of exp(⫺1), a is the characteristic dimension of the phase, and the asterisk de-
notes the radiogenic isotope.
Equations 11.75 and 11.76 both obey boundary conditions Cx ⫽ C0 every-
where at t0 and C ⫽ C1 at the surface for t ⱖ 0. In equation 11.76, L* is the rate
of production of the radiogenic isotope.
As we see in figure 11.13, for a stable isotope such as 18O, the mean concentra-
tion in a phase attains a constant value after the “closure” of exchanges (“closure
temperature” Tc in this case is graphically determined by intersecting the prolon-
gations of surface concentration and mean concentration; see upper dashed lines
in figure 11.13). However, the radiogenic daughter isotope progressively increases
its concentration, as a result of the virtual cessation of exchanges, beyond the
value of Tc (in this case, Tc and the corresponding tc are graphically determined
740 / Methods
by the prolongation of the daughter concentration curve to C * ⫽ 0; see lower

dashed line in figure 11.13).
The analytical representation of closure temperature for a geochronological
system (Dodson, 1973) is
R
Tc = .
 Aτ Dt 0  (11.77)
E a ln  2 
 a 
In equation 11.77, A is a numerical constant depending on the geometry and

decay constant of the parent radionuclide. If the half-life of the parent is long in
comparison with the cooling period, A takes a value of 55, 27, or 8.7 for volume
diffusion from a sphere, cylinder, or plane sheet, respectively. If decay rates are
faster, A progressively diminishes (see table 1 in Dodson, 1973, for numerical
values).
Ganguly and Ruitz (1986) investigated the significance of closure temperature
in terms of simple equilibrium thermodynamics. Assuming Rb-Sr exchanges be-
tween phases ␣ and ␤ to be representable in terms of the equilibrium
87
Rb α + 87
Srβ ⇔ 87
Srα + 87
Rb β , (11.78 )
the Rb/Sr fractional amounts in the two phases can be related to the equilibrium
constant through
 − ∆G780   ( 87
Rb 87
Sr ) β

⋅K .
K 78 = exp   =
 RT  

( 87
Rb 87
Sr )
α


γ , 78 (11.79)
The term within the square brackets in equation 11.79 is the normalized (equilib-
rium) distribution coefficient between the minerals ␣ and ␤ (cf. section 10.8) at
the closure condition of the mineral isochron. Ganguly and Ruitz (1986) have
shown it to be essentially equal to the observed (disequilibrium) distribution co-
efficient between the two minerals as measured at the present time. K␥,78 can be
assumed to be 1, within reasonable approximation. Equation 11.79 can be cali-
brated by opportunely expanding ⌬G 078 over P and T:
∆G 780 = ∆ H 780 − T∆S780 + P∆V780 . (11.80)
Based on equations 11.79 and 11.80, the measured DRb/Sr value for each mineral
of a given isochron must furnish concordant indications in terms of the T and P
of equilibrium. Moreover, the deduced parameters can be assumed to correspond
to the condition of closure of exchanges.
Table 11.5 Some estimates of closure temperature, related to diffusion of

daughter isotope. Large discrepancies and wide Tc ranges can be ascribed to
differences in cooling rates of system and grain size dimensions.
Phase Element Tc (°C) Reference

Biotite Ar 373 ⫾ 21 Berger and York (1981)
Sr 250–311 Dodson (1979)
Sr 280–300 Dodson (1973)
Sr 300 Hart (1964)
Sr 350 ⫾ 50 Jäger et al. (1967)
Sr 350–720 Hanson and Gast (1967)
Sr 400 Verschure et al. (1980)
Sr 531–833 Ganguly and Ruitz (1986)
Muscovite Sr 308–726 Ganguly and Ruitz (1986)
Sr 500 Jäger et al. (1967)
Hornblende Ar 490 Harrison (1981)
Ar 685 ⫾ 53 Berger and York (1981)
K-feldspar Ar 230 ⫾ 18 Berger and York (1981)
Sr 314–352 Dodson (1979)
Sr 531–925 Ganguly and Ruitz (1986)
Plagioclase Ar 176 ⫾ 54 Berger and York (1981)
Sr 567 Ganguly and Ruitz (1986)
Microcline Ar 132 ⫾ 13 Harrison and McDougall (1982)
Discrete values of closure temperatures in various minerals and mineral cou-

ples for the 40 Ar-39Ar and Rb-Sr geochronological systems are reported in table
11.5.
11.7 Application I: Radiometric Dating
In this section we will briefly examine the various methods adopted in geochro-
nology, all based essentially on the constancy of the decay rates of radiogenic
nuclides, which, within reasonable limits, are unaffected by the physicochemical
properties of the system. For more detailed treatment, specialized textbooks such
as Rankama (1954), Dalrymple and Lanphere (1969), Faure and Powell (1972),
Jäger and Hunziker (1979), Faure (1986), and McDougall and Harrison (1988)
are recommended.
11.7.1 Rb-Sr
The element rubidium has 17 isotopes, two of which are found in nature in the
following isotopic proportions: 85Rb ⫽ 0.72165 and 87Rb ⫽ 0.27835. Strontium
has 18 isotopes, four of which occur in nature in the following average weight
742 / Methods
proportions 84Sr ⫽ 0.0056, 86Sr ⫽ 0.0986, 87Sr ⫽ 0.0700, and 88Sr ⫽ 0.8258. 87Rb
decays to 87Sr by ␤⫺ emission, according to
38 Sr + β
→ 87 + v,
87 −
37 Rb (11.81)
with a decay constant of ␭ ⫽ 1.42 ⫻ 10⫺11 a (a ⫽ anna, the Latin word for
“years”), corresponding to a half-life t1/2 of about 4.9 ⫻ 1010 a.
Because the quantity of unstable parent radionuclides P remaining at time t
is related to the decay constant (cf. eq. 11.22) through
P = N 0, P exp ( − λt ) , (11.82)
and the quantity of daughter radionuclides D produced at time t is
[
D = P exp (λt ) − 1 , ] (11.83)
the amount of radiogenic strontium 87Srrad produced at time t is
87
Srrad = 87
[
Rb exp ( λt ) − 1 . ] (11.84 )
The bulk amount of 87Sr at time t is
87
Srt = 87
Sr0 + 87
[
Rb exp ( λt ) − 1 , ] (11.85)
where 87Sr 0 is the amount of 87Sr present at t ⫽ 0.

In practice, because in radiometric determinations based on mass spectrome-
try it is more convenient to measure mass ratios than absolute values, equation
11.85 is usually rewritten as follows, normalized to the abundance of the stable
isotope 86Sr:
 Sr   Sr 
[ ]
87 87 87
Rb
  =  + exp ( λt ) − 1 . (11.86)
 86 
Sr t  86
Sr  0 86
Sr
Whenever it is possible to analyze in a given rock at least two minerals that crys-
tallized at the same initial time t ⫽ 0, equation 11.86 can be solved in t. On a
Cartesian diagram with coordinates 87Sr/ 86Sr and 87Rb/ 86Sr, equation 11.86 ap-
pears as a straight line (“isochron”) with slope exp( ␭t) ⫺ 1 and intercept ( 87Sr/
86
Sr) 0 . As shown in figure 11.14A, all minerals crystallized at the same t from the
same initial system of composition ( 87Sr/ 86Sr) 0 rest on the same isochron, whose
slope exp( ␭t) ⫺ 1 increases progressively with t.
Figure 11.14 (A) Internal Rb-Sr isochron for a system composed of three crystalline
phases of initial compositions A 0, B0, and C0 formed at time t ⫽ 0 and thereafter closed
to isotopic exchanges up to time of measurement t, when they acquired compositions
At , Bt , and Ct . (B) Effects of geochronological resetting resulting from metamorphism
or interaction with fluids. X1, X2, and X3: bulk isotopic compositions of the three rock
assemblages. In cases of short-range isotopic reequilibration, the three assemblages define
crystallization age and original (87Sr/ 86Sr)0 of the system; the three internal isochrons (con-
cordant in this example) define resetting age.
From the relative positions of phases A, B, and C in figure 11.14, it is obvious

that the last one is richest in 87Rb, so that the production of radiogenic 87Sr is
highest in this phase. A hypothetical phase with no initial 87Rb at time t ⫽ 0
would lie at zero on the abscissa axis and would maintain its initial ( 87Sr/ 86Sr) 0
unaltered.
Whenever the phases in the system are reopened to isotopic exchanges, as a
result of a thermal event or interaction with permeating fluids that enhance ad-
vective exchanges, time is reset to zero. To highlight isotopic remobilization
events, it is useful to compare age determinations obtained from internal isochrons
(minerals within a single rock specimen) with indications arising from whole rock
isochrons (isotopic compositions of bulk rock specimens within a single outcrop).
In the case of chronological resetting, the slope of the isochron, based on bulk
rock samples from different portions of the same outcrop, may define the actual
age of crystallization, if isotopic resetting was limited to short-range intercrystal-
line exchanges (figure 11.14B). In this case, the internal isochrons based on the
isotopic compositions of the various phases within a single rock specimen define
the age of resetting.
744 / Methods
11.7.2 Sm-Nd, Lu-Hf, and La-Ce

Geochronological methods based on REE (plus Hf) have been developed only in
the last two decades, because of the high precision required in analysis. Natural
samarium has seven isotopes, occurring in the following proportions:144Sm ⫽
0.0309, 147Sm ⫽ 0.1497, 148Sm ⫽ 0.1124, 149Sm ⫽ 0.1383, 150 Sm ⫽ 0.0744,
152
Sm ⫽ 0.2672, and 154Sm ⫽ 0.2271. Three of these isotopes undergo ␣ decay,
being transformed into Nd isotopes:
60 Nd + α
→ 143 t1 2 = 1.06 × 1011 a
147
62 Sm (11.87 )
60 Nd + α
→ 144 t1 2 = 7.0 × 1015 a
148
62 Sm (11.88 )
60 Nd + α
→ 145 t1 2 = 1.0 × 1016 a
149
62 Sm (11.89 )
Neodymium also has seven isotopes, occurring in nature in the following pro-
portions: 142Nd ⫽ 0.2711, 143Nd ⫽ 0.1217, 144Nd ⫽ 0.2385, 145Nd ⫽ 0.0830,
146
Nd ⫽ 0.1719, 148Nd ⫽ 0.0573, and 150 Nd ⫽ 0.0562. Sm-Nd dating is based on
the decay of 147Sm, whose half-life is the only one sufficiently short for geochrono-
logical purposes. The isochron equation is normalized to the 144Nd abundance:
 143 Nd   143 Nd 
[ ]
147
Sm
 144  =  144  + exp (λt ) − 1 . (11.90)
 Nd  t  Nd  0 144
Nd
Lu-Hf dating is based on the ␤⫺ decay process of 176Lu into 176Hf:
72 Hf + β
→ 176 + v.
176 −
71 Lu (11.91)
The first applications of the Lu-Hf method were attempted about 20 years ago.
Lutetium has 22 isotopes, two of them occurring in nature in the following pro-
portions: 175Lu ⫽ 0.9741 and 176Lu ⫽ 0.0259. Hafnium has 20 isotopes, six of
which occur in nature in the following proportions: 174Hf ⫽ 0.0016, 176Hf ⫽
0.0520, 177Hf ⫽ 0.1860, 178Hf ⫽ 0.2710, 179Hf ⫽ 0.1370, and 180 Hf ⫽ 0.3524. The
isochron equation is normalized to the 177Hf abundance:
 Hf   Hf 
[ ]
176 176 176
Lu
  =  + exp ( λt ) − 1 . (11.92)
 177
Hf  t  177
Hf  0 177
Hf
The decay constant is not known experimentally but has been deduced by com-
parisons with radiometric ages obtained by other methods. The obtained ␭ value
is (1.94 ⫾ 0.07) ⫻ 10⫺11 a⫺1, corresponding to a half-life of (3.57 ⫾ 0.14) ⫻
1010 a.
The first application of La-Ce dating to earth sciences was proposed by Ta-
naka and Masuda (1982), following improvements in the technology of solid

source mass spectrometry. The La-Ce method is based on the combined ␤⫺-EC
decay of 138La into 138Ce and 138Ba:
58 Ce + β
→ 138 +v
138 −
57 La (11.93)
57 La + 56 Ba + hv ( γ ) + hv ( X ).
e − → 138
138
(11.94 )
Lanthanum has 19 isotopes with A between 126 and 144. Only two occur in
nature, in the following proportions: 138La ⫽ 0.0009 and 139La ⫽ 0.9991. The low
relative amount of 138La and its low decay rate ( ␭ ⫽ ␭EC ⫹ ␭␤⫺ ⫽ 6.65 ⫻ 10⫺12
a⫺1 ) prevented earlier application of this dating method. Cerium has 19 isotopes,
with masses between 132 and 148, and two isomers at A ⫽ 137 and 139. Four
isotopes are found in nature, in the following proportions: 136Ce ⫽ 0.0019,
138
Ce ⫽ 0.0025, 140 Ce ⫽ 0.8848, and 142Ce ⫽ 0.1108. The isochron equation is
normalized to the 142Ce abundance:
 138Ce   138Ce   ␭β − 
[ ]
138
La
 142  =  142  +   exp( ␭t ) − 1 . (11.95)
 Ce  t  Ce  0  ␭  142
Ce
where ␭␤⫺ ⫽ 2.24 ⫻ 10⫺12 a⫺1.

Figure 11.15 shows some examples of geochronological application of REE
decay processes. In figure 11.15A we see that Sm-Nd dating of gabbroic rocks of
the Stillwater Complex gives consistent results when based on both mineral and
total rock isochrons. This indicates that the LREE were not remobilized during
the various hydrous alteration processes undergone by the host rocks. Figure
11.15B shows the first Lu-Hf isochron obtained by Patchett (1983) on eucrite-
type meteorites, whose age of formation (4.55 ⫻ 109 a) was adopted to define
t1/2. Figure 11.15C shows the large uncertainty associated with 138Ce/ 142Ce deter-
minations, which makes the application of this sort of dating problematic.
11.7.3 K-Ar
Potassium has 10 isotopes, three of which occur in nature in the following propor-
tions: 39K ⫽ 0.9326, 40 K ⫽ 0.0001, and 41K ⫽ 0.0673. Argon (from the Greek
verb argéos—i.e., to be inactive) is present in the earth’s atmosphere (0.934% in
mass). It is composed of eight isotopes, three of which are present in nature in
the following proportions: 36Ar ⫽ 0.0034, 38Ar ⫽ 0.0006, and 40 Ar ⫽ 0.9960.
K-Ar geochronology is based on the natural decay of 40 K into 40 Ar (EC,
␤ ⫽ 10.5%). However, as shown in figure 11.16, 40 K also decays into 40 Ca ( ␤⫺ ⫽
⫹
89.5%). The 40 K-to-40 Ar decay branch is dominated by electron capture, al-

though a limited amount of positron emission of low energy is observed (about
0.001%).
Figure 11.15 Geochronological applications of REE decay processes. (A) Total rock and
mineral Sm-Nd isochrons for Stillwater Complex, Montana. (B) Lu-Hf isochron for eu-
crites. (C) La-Ce dating of gabbros of Bushveld Complex, South Africa. Reprinted from
D. J. De Paolo and G. J. Wasserburg, Geochimica et Cosmochimica Acta, 43, 999–1008,
copyright 䉷 1979; J. P. Patchett, Geochimica et Cosmochimica Acta, 47, 81–91, copyright
䉷 1983, with kind permission from Elsevier Science Ltd., The Boulevard, Langford Lane,
Kidlington 0X5 1GB, UK; and from T. Tanaka and A. Masuda, Nature, 300, 515–518,
copyright 䉷 1982 Macmillan Magazines Limited.
The decay constant of the first-order reaction transforming 40 K into 40 Ar is

␭EC ⫽ 0.581 ⫻ 10⫺10 a⫺1 (89.3%), and the ␤⫺ decay constant producing 40 Ca is
␭␤ ⫽ 4.962 ⫻ 10⫺10 a⫺1 (10.7%). The resulting “branching ratio” ␭EC / ␭␤ is 0.117
(⫾0.001), and the half-life for the dual decay is 1.250(⫾0.002) ⫻ 109 a⫺1.
Because there is a double decay involved in parent P disintegration, the decay
equation corresponding to the general formula of equation 11.83 reads as follows:
 λ EC 
40
18 Art = 18
40
Ar0 + 
λ
 EC + λ β 
40
{ [(
 × 19 K exp λ EC + λ β t − 1 , )] } (11.96)
and the corresponding age equation (obtained under the assumption that no
40
18Ar 0 is present in the mineral at the time of its formation) is
  λ EC + λ β  40 
1 18 Ar
t= ln 1 +  
λ EC + λ β   λ EC  40
19 K 

(11.97)
 40 
(
= 1.804 × 10 9 ) ln  1 + 9.54

18 Ar
40 .

19 K
K is usually determined by conventional analytical method such as atomic

absorption spectroscopy, flame photometry, X-ray fluorescence, neutron activa-
tion, etc., and the corresponding amount of 40 K is deduced from the fractional
isotopic abundance ( 40 K ⫽ 0.0118%). Analysis of 40 Ar is more complex, because
of the high amount of 40 Ar present in the earth’s atmosphere. It is conducted by
mass spectrometry on gaseous portions extracted from minerals and rocks by
melting in radiofrequency or induction furnaces in high-vacuum lines (higher
than 10⫺8 torr). The released Ar is purified of reactive gaseous components such
as H2 , N2 , CO2 , O2 , and H2O by chemical traps, and a “spike” of known isotopic
composition enriched in 38Ar is added. The total 40 Ar released by a given sample
is composed of radiogenic plus atmospheric amounts:
40
18 Artotal = 40
18 Arrad + 40
18 Aratm . (11.98 )
Because the 40 Ar/ 36Ar ratio of the earth’s atmosphere is 295.5 (constant), the
correction for atmospheric argon is made by measuring the amount of 36Ar:
40
18 Arrad = 18 Artotal −
40
295.5 × 36
18 Ar . (11.99 )
In practice, mass spectrometric measurements are carried out on the 40 Ar/ 38Ar
and 38Ar/ 36Ar ratios, for the reasons previously outlined (see Dalrymple and
Lanphere, 1969 for the form of the relevant chronological equation).
748 / Methods
Figure 11.16 Decay scheme for 40K, showing double decay to 40Ca and 40Ar.
In conventional K-Ar dating based on equation 11.97, it is assumed that no

40
initial 40
18Ar 0 is present and that nonradiogenic Ar results entirely from atmo-
spheric contamination (appropriately corrected using eq. 11.99). However, an ad-
ditional source of nonradiogenic argon may be the magma from which the crystal
grew ( 40 Ar originated from the earth’s mantle or from an old K-rich crust). Deep
strata affected by thermal metamorphism may also release 40 Ar, which becomes
an “excess component” in crystals equilibrated at shallower levels. The problem
of excess 40 Ar may be overcome by the K-Ar isochron method of dating. Equa-
tion 11.96 is normalized to the 36Ar abundance:
 40
Ar   Ar 
40
  =  
 36
Ar  t  Ar  0
36
(11.100)
 λ EC  K
{ [( )] }
40
+   exp λ EC + λ β t − 1 .
 λ EC + λ β  Ar 
36
The intercept term ( 40 Ar/ 36Ar) 0 , which accounts for igneous, metamorphic, or
atmospheric sources, is regarded as the excess contribution present at time t ⫽ 0,
whereas the second term is the radiogenic component accumulating in the various
minerals of the isochron by decay of 40 K. If all the minerals used to construct the
isochron underwent the same geologic history and the same sort of contamination
by excess 40 Ar, the slope of equation 11.100 would have a precise chronological
Figure 11.17 Whole rock K-Ar isochron of Tuff IB strata from Olduvai Gorge, Tanzania.
From G. Faure (1986), Principles of Isotope Geology, 2nd edition, copyright 䉷 1986 by
John Wiley and Sons. Reprinted by permission of John Wiley & Sons.
significance. An example of a K-Ar isochron is given in figure 11.17. It is based

on whole rock analyses of Tuff IB strata in the Olduvai Gorge, Tanzania (a level
of paramount importance in dating the early history of humans). The slope of
the isochron was recalculated by Fitch et al. (1976) on the basis of the original
data of Curtis and Hay (1972) with York’s (1969) algorithm (see appendix 2) and
gives an age of 2.04 ⫾ 0.02 Ma. The excess argon intercept term is 276 ⫾ 29.
Besides the presence of excess 40 Ar, the main problem in K-Ar geochronology
is the mobility of Ar, which may be released from the crystal lattice even under
ambient P and T conditions. The Ar retention capacities of silicates seem to im-
prove in the following sequence: feldspathoids–feldspars–micas–amphiboles–
pyroxenes. Thermal events and/or alteration processes partially remove radio-
genic argon from the crystal lattice, resulting in “apparent ages” younger than
the actual ones. A classical study by Hart (1964) on the Precambrian gneisses of
the Idaho Springs Formation, regionally metamorphosed at about 1400 Ma, re-
veals the Ar release effect induced by a young quartz monzonite (Eldora stock)
that intruded the gneiss at about 55 Ma (figure 11.18). As figure 11.18 shows, the
ages were completely reset in the contact zone and partially rejuvenated up to a
distance of more than 1 km from the contact. Moreover, amphibole was less
affected by resetting, at the same distance, with respect to mica and feldspar.
40
11.7.4 Ar/ 39Ar
The 40 Ar/ 39Ar dating method is based on the artificial production of 39

Ar by
neutron irradiation of K-bearing samples in nuclear reactors:
750 / Methods
Figure 11.18 Apparent K-Ar ages of minerals from Idaho Springs Formation (Front
Range, Colorado, 1350–1400 Ma) in zone subjected to contact metamorphism by intru-
sion of a quartz monzonite (Eldora stock, 55 Ma). Reprinted from S. R. Hart, Journal of
Geology, (1964), 72, 493–525, copyright 䉷 1964 by The University of Chicago, with per-
mission of The University of Chicago Press.
19 K + n→ + p.
39 39
18 Ar (11.101)
The number of 39Ar atoms formed depends on irradiation time ⌬t, density of neutron
flux ␸( ⑀ ) as a function of energy ␧, and cross-section of 39K atoms ␴( ⑀ ):
39
18 Ar = 39
19 [
K ∆ t ∫ ϕ( ε ) σ( ε ) dε .] (11.102)
If we denote the flux production factor in square brackets in equation 11.102 as ⌽, we can
define a parameter J related to 39Ar production in a monitor sample of known age tm:
−1 −1
 40
K  λ EC 
J = 19
 m  Φ
 19 K 
39
 λ EC + λ β 
(11.103)
 Ar 
{exp [(λ ) ] }
39
= 18
 + λβ tm − 1 .
 Ar  m
40 EC
18
The 40 Ar/ 39Ar ratio of a given sample after neutron irradiation is

Table 11.6 Interfering nuclear reactions during neutron irradiation of mineral

samples. Boldface: principal reaction for 40Ar/ 39Ar dating procedure (from Brereton,
1970).
Produced
Isotope Calcium Potassium Argon Chlorine
36
Ar 40
Ca(n, n␣) – – –
37
Ar 40
Ca(n, ␣) 39
K(n, n d ) 36
Ar(n, ␥ ) –
38
Ar 42
Ca(n, n␣) 39
K(n, d) 40
Ar(n, n d, ␤ ⫺) 37
Cl(n, ␥, ␤ ⫺)
41
K(n, ␣, ␤ ⫺)
39
Ar 42
Ca(n, ␣) 39
K(n, p) 38
Ar(n, ␥ ) –
43
Ca(n, n␣) 40
K(n, d) 40
Ar(n, d, ␤ ⫺)
40
Ar 43
Ca(n, ␣) 40
K(n, p) – –
44
Ca(n, n␣) 41
K(n, d)
40
Ar
=
[(
exp λ EC + λ β t − 1 )] , (11.104)
39
Ar J
and the age of the sample is
 1   40
Ar  
t=  ln   J + 1 . (11.105)
 λ EC + λ β   39
Ar  
Although 39 18Ar is unstable, its half-life (269 a) is long enough to be considered
stable with respect to the time normally involved in analyses. Ages obtained
through application of equation 11.105 are normally referred to as “total argon
release dates” (cf. Faure, 1986) and are subject to the same uncertainties as con-
ventional K-Ar ages. Moreover, corrections for atmospheric Ar through equation
11.99 are complicated by the production of 36Ar by 40 Ca(n,n␣ ) reaction during
irradiation (see table 11.6). More generally, neutron irradiation is responsible for
a complex alteration of the pristine isotopic composition of Ar by nuclear reac-
tions involving the neighboring isobars, as summarized in table 11.6.
According to Dalrymple and Lanphere (1971), the appropriate 40 Ar/ 39Ar
ratio to be introduced in age equation 11.105 can be derived from the measured
ratio ( 40 Ar/ 39Ar)meas by application of
40
Ar  40
Ar   36
Ar  
=   − C1   + C1C 2 D − C 3 
 Ar  meas  Ar  meas
39 39 39
Ar 
(11.106)
( )
−1
× 1 − C4 D ,
752 / Methods
Figure 11.19 The first 40Ar/ 39Ar age spectrum. Reprinted from Turner et al. (1966), with
kind permission from Elsevier Science Publishers B.V, Amsterdam, The Netherlands.
where C1 ⫽ 295.5 is the 40 Ar/ 36Ar ratio in the atmosphere, C2 ⫽ 2.72 ⫾ 0.014 ⫻
10⫺4 is the 36Ar/ 37Ar ratio induced by interfering nuclear reactions involving Ca,
C3 ⫽ 5.9 ⫾ 0.42 ⫻ 10⫺3 is the 40 Ar/ 39Ar ratio induced by interfering nuclear
reactions involving K, C4 ⫽ 6.33 ⫾ 0.04 ⫻ 10⫺4 is the 39Ar/ 37Ar ratio induced by
interfering nuclear reactions involving Ca, and D is the 37Ar/ 39Ar ratio in the
sample, after correction for decay of 37Ar (t1/2 ⫽ 35.1 days). We refer readers to
Brereton (1970) and to Dalrymple and Lanphere (1971) for detailed accounts of
the necessary corrections.
The main virtue of the 40 Ar/ 39Ar dating method is that the isotopic com-
position of the various argon fractions released during stepwise heating of the
sample (or “age spectrum”) furnishes precious information about the nature of
excess argon and the thermal history of the sample. The first interpretation in this
sense was that of Turner et al. (1966), who analyzed a sample of the disturbed
Bruderheim chondrite and arranged the mass spectrometric 40 Ar/ 39Ar data in a
series corresponding to the cumulative percentage of released 39Ar (figure 11.19).
The data reported in figure 11.19 refer to two separate irradiations, but their
distribution is consistent with the model age spectrum of a sample composed of
spheres with a lognormal distribution of radii that underwent a brief episode of
radiogenic 40 Ar* outgassing (90%) 0.5 Ga before present (solid line in figure
11.19). According to the single-site diffusion model of Turner (1968), if we as-
sume homogeneous distribution of 39Ar, a single (thermally activated) transport
mechanism, identical diffusion rates for all Ar isotopes, and a radiogenic 40 Ar*
boundary concentration of zero, a set of model age curves can be parametrically
generated, corresponding to the various degrees of outgassing. As shown in figure
Figure 11.20 Theoretical age spectra based on Turner’s (1968) single-site diffusion model.
True age of sample is 4.5 Ga, and outgassing episode took place 0.5 Ga ago. Reprinted
from Turner (1968).
11.20, the theoretical age spectra thus obtained differ greatly according to the
size distribution of the outgassing particles (uniform radius or lognormal distri-
bution) if the 40 Ar* loss exceeds 40%. The first infinitesimal increment of gas
extracted represents the age of a brief thermal disturbance that affected the
sample, causing a partial loss of radiogenic argon, whereas the true age is attained
at maximum 39Ar release.
Note that Turner’s theoretical spectra for uniform spheres indicate that, if
40
Ar* loss exceeds 10%, the maximum age achieved at complete 39Ar release does
not correspond to the true age of the sample—i.e., the asymptotic rise of the
curves does not reach 40 Ar/ 39Ar ⫽ 1 for both lognormal and uniform distribution
of radii. In these cases, assessment of 40 Ar* loss must be established from the
form of the experimental age spectrum to correct appropriately for the lowering
of maximum age. For example, in the case exemplified in figure 11.20, a 40 Ar*
loss of 60% would lead to an apparent maximum age of 3.9 Ga for an aggregate
of uniform spheres, whereas the age perturbation would be greatly reduced for
an aggregate of spheres with a lognormal distribution of radii (apparent age 艐
4.4 Ga). Partial modifications of Turner’s (1968) single-site diffusion model were
presented by Harrison (1983) to account for multiple 40 Ar* losses in samples with
particularly complex thermal histories.
Further complications in 40 Ar/ 39Ar age spectra arise whenever samples con-
tain excess 40 Ar not produced by in situ decay of 40 K. In these cases, the presence
of 40 Ar violates the zero concentration boundary condition of Turner’s model,
and the first infinitesimal increments of extracted argon often display extremely
754 / Methods
Figure 11.21 Age spectra of hornblendes of a Paleozoic gabbro (367 Ma) intruded by a
granitic body during the Cretaceous (114 Ma). Samples underwent permeation of 40Ar*
from lower crustal portions and differential losses of radiogenic argon (32, 57, and 78%
respectively), proportional to distance from contact (0.3, 1, and 2.5 km). Reprinted from
T. M. Harrison and I. McDougall, Geochimica et Cosmochimica Acta, 44, 2005–2020,
copyright 䉷 1980, with kind permission from Elsevier Science Ltd., The Boulevard, Lang-
ford Lane, Kidlington 0X5 1GB, UK.
old apparent ages, which decrease rapidly in the subsequent stepward releases.
An emblematic representation of such complex age spectra is given in the study
of Harrison and McDougall (1980) on hornblendes from Paleozoic gabbro in-
truded by a granitic body during the Cretaceous. The observed 40 Ar/ 39Ar spectra
reveal loss profiles compatible with Turner’s model for an aggregate of spheres
with lognormal distribution of radii (figure 11.21) with partial losses of 31, 57,
and 78% of radiogenic argon and a true age of 367 Ma (coinciding with an isoch-
ron age of 366 ⫾ 4 Ma; Harrison and McDougall, 1980). However, the presence
of intergranular excess argon is evident from the initial steeply descending relative
concentration profiles (which in some samples of the same outcrop generate initial
apparent ages as high as 3.5 Ga; cf. Harrison and McDougall, 1980).
Additional complexities arise whenever the permeating excess argon diffuses
from intergranular boundaries to lattice sites or whenever the solid phase under-
Figure 11.22 (A) Saddle-shaped age spectra of calcic plagioclases from amphibolites,
Broken Hill, Australia. (B) Arrhenius plots of reactor-produced isotopes for two of the
five samples, defining existence of three diffusion domains corresponding to albite-rich
lamellae (domain 1) and anorthite-rich lamellae of different widths (domains 2 and 3).
Reproduced with modifications from T. M. Harrison and I. McDougall (1981), with kind
permission from Elsevier Science Publishers B.V, Amsterdam, The Netherlands.
goes subsolidus unmixing, generating domains with differing diffusional proper-

ties. In these cases, the age spectra have typical “saddle-shaped” conformations,
like those shown for plagioclases in figure 11.22A.
In the case exemplified in figure 11.22A, the saddle-shaped conformation
arises from the presence of three distinct structural domains corresponding to the
products of subsolidus unmixing of plagioclase: a modulated “e” structure of
alternating albite-rich and anorthite-rich lamellae (domains 1 and 2) and a
coarser transitional anorthite-rich zone (domain 3). As shown in figure 11.22B,
these three structural zones are reflected in three diffusion domains shown by the
Arrhenius plots of the reactor-produced isotopes. The slopes for the albite-rich
and anorthite-rich lamellar zones differ, indicating different activation energies of
the diffusion process, whereas the anorthite-rich lamellae and coarse anorthite-
rich transitional zone have the same slope (corresponding to an activation energy
of about 35 kcal/mol). Their offset is attributable to a factor 10 difference in
lamellar width (Harrison and McDougall, 1981). Excess 40 Ar and radiogenic
40
Ar* reside in different sites (anion and cation sites, respectively. At high extrac-
tion temperatures, the relatively low diffusion jump probability of excess argon in
anion sites causes a retarded movement of 40 Ar in the crystal, resulting in ex-
tremely high apparent ages in the final extractions (see McDougall and Harrison,
1988 for an extended discussion on the diffusional mechanisms of Ar).
756 / Methods
11.7.5 K-Ca
The K-Ca dating method uses the ␤⫺ decay branch of 40 K into 40 Ca (cf. figure
11.16). Calcium has six stable isotopes occurring in nature in the following abun-
dances: 40 Ca ⫽ 0.969823, 42Ca ⫽ 0.006421, 43Ca ⫽ 0.001334, 44Ca ⫽ 0.020567,
46
Ca ⫽ 0.000031, and 48Ca ⫽ 0.001824. The high relative abundance of 40 Ca (a
result of its mass number A, which is a multiple of 4 and thus exceptionally stable;
cf. section 11.3) is one of the two main problems encountered in this sort of dating
(in Ca-rich samples, the relative enrichment in 40 Ca resulting from 40 K decay is
low with respect to bulk abundance). The other problem is isotopic fractionation
of calcium during petrogenesis (and also during analysis; see for this purpose
Russell et al., 1978). These two problems prevent extensive application of the K-
Ca method, which requires extreme analytical precision. The isochron equation
involves normalization to the 42Ca abundance
 40
Ca   40
Ca 
  =  
 42
Ca  t  42
Ca  0
(11.107)
 
{ [( }
λβ K
)]
40
+    exp λ EC + λ β t − 1 .
 λ EC + λ β 
42
Ca 
A substantial improvement in K-Ca dating was achieved by Marshall and De

Paolo (1982), who combined the double-spike technique of Russell et al. (1978)
with high-precision measurements on chromatographically separated sample ali-
quots.
The ( 40 Ca/ 42Ca)t ratio to be introduced into equation 11.107 (hereafter, Rcorr ) was
derived from the measured ratio (Rmeas ) by application of the “exponential law”:
P
m 
Rmeas,ij = Rcorr,ij  i  (11.108)
 mj 
with exponent
 Rmeas, jk 
ln  
 Rnorm, jk 
P= , (11.109)
 mj 
ln  
 mk 
where pedices i, j, and k identify nuclides of mass numbers 40, 42, and 44, respectively.
Rnorm,jk is the normalizing ratio ( 42Ca/ 44Ca) taken to be 0.31221 (Russell et al., 1978), and
mi , mj, mk are, respectively, the masses of the corresponding nuclides ( 40 Ca ⫽ 39.962591,
42
Ca ⫽ 41.958618, and 44Ca ⫽ 43.955480).
Figure 11.23 K-Ca isochron for granite batholith of Pikes Peak (Colorado). Aliquot-
spiked samples not included in age calculation. Reprinted from B. D. Marshall and D. J.
De Paolo, Geochimica et Cosmochimica Acta, 46, 2537–2545, copyright 䉷 1982, with kind
1GB, UK.
Figure 11.23 shows the isochron obtained by Marshall and De Paolo (1982)
for the granite batholith of Pikes Peak (Colorado). The effectiveness of the
double-spike technique is evident, especially when we see that aliquot-spiked
samples do not fall on the best-fit interpolant (York’s algorithm; York, 1969).
The obtained age (1041 ⫾ 32 Ma) is consistent with that previously obtained with
Rb-Sr whole rock analyses (1008 ⫾ 13 Ma; see Marshall and De Paolo, 1982, for
references). The initial ratio ( 40 Ca/ 42Ca) 0 of 151.0 is identical, within the range
of uncertainty, to upper mantle values, indicating negligible contamination by old
crust components: the relative K/Ca abundance in the earth’s mantle is about
0.01, a value too low to alter the “primordial” ( 40 Ca/ 42Ca) 0 composition.
11.7.6 U-Th-Pb
Uranium has 15 isotopes, all unstable, with A-values from 227 to 240, and two
isomers at A ⫽ 235. Natural uranium is composed of 238U (99.2745%), 235U
(0.720%), and 234U (0.0055%). Thorium has 12 isotopes, all unstable, with A-
values from 223 to 234. Lead has 29 isotopes, four of them present in nature in
the following atomic proportions: 204Pb ⫽ 1.4%, 206Pb ⫽ 24.1%, 207Pb ⫽ 22.1%,
and 208Pb ⫽ 52.4%. The 4n, 4n ⫹ 2, and 4n ⫹ 3 decay series can be schematically
represented as follows:
758 / Methods
90 Th → 82 Pb +
232 208
6α + 4 β − λ 232 = 4.9475 × 10 −11 a −1 (11, 110.1)
238
92 U → 82 Pb +
206
8α + 6 β − λ 238 = 1.55125 × 10 −10 a −1 (11.110.2 )
235
92 U → 82 Pb +
207
7α + 4 β − λ 235 = 9.8485 × 10 −10 a −1 , (11.110.3)
where ␭232, ␭238, and ␭235 are the combined decay constants. Equation 11.83 ap-
plied to the three parent nuclides gives
208
Pb = 232
[ (
Th exp λ 232 t − 1 ) ] (11.111.1)
206
Pb = 238
[ (
U exp λ 238 t − 1 ) ] (11.111.2 )
207
[ (
Pb = 235 U exp λ 235 t − 1 . ) ] (11.111.3)
204
Normalization to the stable isotope Pb leads to the chronological equa-
tions
 Pb   Pb   Th 
[ ( ) ]
208 208 232
  =  +  exp λ 232 t − 1 (11.112.1)
 204
Pb  t  204
Pb  0  204
Pb 
 Pb   Pb   U
[ ( ) ]
206 206 238
  =  +  exp λ 238 t − 1 (11.112.2 )
 204
Pb  t  204
Pb  0  204
Pb 
 Pb   Pb   U
[ ( ) ]
207 207 235
  =  +  exp λ 235 t − 1 . (11.112.3)
 204
Pb  t  204
Pb  0  204
Pb 
Because of the different decay rates of 235U and 238U and their constant relative
proportion in nature ( 235U/ 238U ⫽ 0.00725), a fourth chronological equation can
be derived, based on the combination of equations 11.112.3 and 11.112.2:
 207
Pb   207
Pb 
∗   − 
 207
Pb   204
Pb  t  204
Pb  0
  =
 206
Pb   206
Pb   206
Pb 
  − 
 204
Pb  t  204
Pb  0 (11.113)
= 0.00725
[ exp (λ t ) − 1] .
235
[ exp (λ t ) − 1]
238
Table 11.7 Values of [exp(␭238 t) ⫺ 1] and [exp(␭235 t) ⫺ 1]

and of (207Pb/ 206Pb)* radiogenic ratio as a function of age t
expressed in Ga (1 Ga ⫽ 1 ⫻ 109 a) (from Faure, 1986).
t (Ga) [exp( ␭235 t) ⫺ 1] [exp( ␭238 t) ⫺ 1] ( 207Pb/ 206Pb)*

0 0.0000 0.0000 0.04604
0.2 0.2177 0.0315 0.05012
0.4 0.4828 0.0640 0.05471
0.6 0.8056 0.0975 0.05992
0.8 1.1987 0.1321 0.06581
1.0 1.6774 0.1678 0.07250
1.2 2.2603 0.2046 0.08012
1.4 2.9701 0.2426 0.08879
1.6 3.8344 0.2817 0.09872
1.8 4.8869 0.3221 0.11000
2.0 6.1685 0.3638 0.12298
2.2 7.7292 0.4067 0.13783
2.4 9.6296 0.4511 0.15482
2.6 11.9437 0.4968 0.17436
2.8 14.7617 0.5440 0.19680
3.0 18.1931 0.5926 0.22266
3.2 22.3716 0.6428 0.25241
3.4 27.4597 0.6946 0.28672
3.6 33.6556 0.7480 0.32634
3.8 41.2004 0.8030 0.37212
4.0 50.3878 0.8599 0.42498
4.2 61.5752 0.9185 0.48623
4.4 75.1984 0.9789 0.55714
4.6 91.7873 1.0413 0.63930
Equation 11.113 is transcendental and must be solved by iterative procedures or

by interpolation on tabulated values or plots representing the expected values of
relative radiogenic lead abundances ( 207Pb/ 206Pb)* with respect to t (table 11.7
and figure 11.24).
Note that, because equation 11.113 is indeterminate at the limit of t ⫽ 0, its solution
under boundary conditions involves application of De l’Hopital’s rule:
 f (t )   f ′ (t ) 
lim   = lim   (11.114)
t→0
 g( t )  t → 0
 g ′(t ) 

lim 0.00725
[ exp(λ t ) − 1]  = lim 0.00725 [λ
235 235 ( )]
exp λ 235t 


t→0
 [ exp(λ t ) − 1]  
238
t→0
[λ 238 exp( λ t )] 
238 
(11.115)
λ 235
= 0.00725 .
λ 238
760 / Methods
Figure 11.24 (207Pb/ 306Pb)* radiogenic ratio of a U-bearing system as a function of age
t (Ga).
Ages obtained through radiometric equations 11.112.1 to 11.112.3 and equa-

tion 11.113 may be concordant, if the system remained closed to all isotopic ex-
changes from t ⫽ 0, or discordant. The second case is more commonly observed
in nature and the obtained apparent ages usually have the following progression
t232Th-208Pb ⬍ t238U-206Pb ⬍ t235U-207Pb ⬍ t207Pb-206Pb. Age determinations involving estima-
tion of residual 238U parent amounts are often subject to large error, because of
the relatively easy mobilization of this element as a uranyl complex during hy-
drous alteration events. An emblematic representation of U-leaching effects on
U-Th-Pb age systematics is given in figure 11.25. Basing their determinations on
the isotopic composition of Pb (virtually unaffected by partial Pb losses), Rosholt
et al. (1973) found an age of 2.79 ⫾ 0.08 Ga for rock samples from the Granite
Mountains of Wyoming. Although these rocks underwent variable gains or losses
of Th and Pb, the imposed Pb-Pb age isochron is not far from the best interpolant
of the 232Th-208Pb system (figure 11.25A), whereas conspicuous deviations are
observed when this is superimposed on the 238U-206Pb diagram because of U
losses ranging from 37 to 88% (figure 11.25B).
Limiting ourselves to the observation of isotopic abundances determined by
the 4n ⫹ 2 and 4n ⫹ 3 decay series, we can construct a “concordia diagram”
(Wetherill, 1956) relating ( 206Pb/ 238U)* and ( 207Pb/ 235U)* ratios developing at
various t in a closed system (figure 11.26):
 206
Pb   206
Pb 
∗   − 
 Pb   Pb  t  Pb  0
[ ( ) ]
206 204 204
  = = exp λ 238 t − 1 (11.116.1)

 238
U  238
U
 
 204
Pb  t
Figure 11.25 Th-Pb and U-Pb isotopic compositions of whole rocks from Granite Moun-
tains Formation (Wyoming) based on data of Rosholt et al. (1973). 2.79 Ga isochrons
based on Pb-Pb age are superimposed, to highlight effect of differential elemental remobi-
lization.
Figure 11.26 Concordia diagram. (A) Interpretation of straight discordia path as a result
of a single episode of Pb loss. (B) Continuous diffusion model of Tilton (1960) applied to
world minerals of a t ⫽ 2800 Ma “common” age. Reproduced with modifications from
G. Faure (1986), Principles of Isotope Geology, 2nd edition, copyright 䉷 1986 by John
Wiley and Sons, by permission of John Wiley & Sons and from Tilton (1960) by permis-
sion of the American Geophysical Union.
762 / Methods
 207
Pb   207
Pb 
∗   − 
 Pb   Pb  t  Pb  0
[ ( ) ]
207 204 204
  = = exp λ 235 t − 1 . (11.116.2 )

 235
U  235
U
 
 204
Pb  t
Each age t therefore corresponds to a certain 206Pb/ 238U, 207Pb/ 235U couple, and
their time evolution gives rise to a curved path or “concordia curve” in the binary
field (figure 11.26A and B). The age interpretation of samples falling outside the
concordia line is controversial. In the simplest acceptance, straight-line arrange-
ments (or “discordia lines”) below the concordia curve are interpreted in terms
of Pb losses or U gains in the system (an eventual U loss from the system would
extend the straight discordia path above the concordia curve). Figure 11.26A
shows, for instance, the discordant compositions of three zircons from a gneiss
of Morton and Granite Falls (Southern Minnesota; Catanzaro, 1963): because
the partial loss of Pb does not significantly affect the 206Pb/ 207Pb ratio, the dis-
cordia line is straight and intersects the concordia curve at two points, t and t⬘,
which are interpreted, respectively, as the true age of the rock (t) and the age of
remobilization (t⬘). However, figure 11.26B shows that U-bearing minerals from
five continents with 206Pb/ 207Pb ages greater than 2300 Ma plot on a single dis-
cordia line, thus indicating an apparent common crystallization age of t ⫽ 2800
Ma, followed by a common Pb loss episode at t⬘ ⫽ 600 Ma (Tilton, 1960). How-
ever, as shown by Tilton (1960), the common straight-path arrangement of the
various minerals is a result of continuous diffusion of Pb from crystals. Assuming
that diffusion obeys Fick’s first law (cf. section 4.11) and affects only Pb, the
daughter-to-parent ratios of the various minerals (assimilated to spheres of radius
r with identical and constant diffusion coefficient D and uniform U distribution)
evolve along a curved trajectory that may be approximated by a straight line for
D/r2 ⬍ 50 ⫻ 10⫺12 a⫺1. Obviously, in this case, the age of rejuvenation t⬘ is only
apparent, although initial age t maintains the same significance as in the previous
model (see Faure, 1986 for alternative interpretations of the concordia diagram).
11.7.7 Re-Os
Rhenium has 12 isotopes, from A ⫽ 185 to A ⫽ 196, and two isomers at A ⫽ 186
and A ⫽ 190. Natural rhenium is composed of 185Re (0.37398) and 187Re
(0.62602). Osmium has 28 isotopes from A ⫽ 169 to A ⫽ 196, with isomers at
A ⫽ 181, 183, 190, 191, and 192. Natural osmium has seven isotopes, all stable:
184
Os (0.00024), 186Os (0.01600), 187Os (0.01510), 188Os (0.13286), 189Os
(0.16252), 190 Os (0.26369), and 192Os (0.40958) (fractional atomic abundances
after Luck and Allègre, 1983).
The 187Re isotope is transformed into 187Os by ␤⫺ decay:
76 Os + β
→ 187 + v + 0.0025 MeV .
187 −
75 Re (11.117 )
The low decay energy prevents accurate determinations of half-life by direct

counting. Reported half-lives range from 3 to 6.6 ⫻ 1010 a. The most recent direct
determination (Lindner et al., 1989) assigns a half-life of (4.23 ⫾ 0.13) ⫻ 1010 a
to the decay process of equation 11.117, which is fairly consistent with the indirect
estimates of Hirt et al. (1963) [(4.3 ⫾ 0.5) ⫻ 1010 a]. The isochron equation is
normalized to the 186Os abundance:
 Os   Os 
[ ( ) ]
187 187 187
Re
  =  + exp λt − 1 . (11.118)
 186
Os  t  186
Os  0 186
Os
Equation 11.118 finds practical application in cosmological studies and in geol-

ogy (dating of sulfide deposits and sediments). Figure 11.27A shows, for instance,
the Re-Os isochron for iron meteorites and the metallic phase of chondrites, ob-
tained by Luck and Allègre (1983). The fact that all samples fit the same isochron
within analytical uncertainty has three important cosmological implications:
1. The primordial isotopic composition in the solar nebula was homoge-

neous ( 187Os/ 186Os ⫽ 0.805 ⫾ 0.006);
2. The ages of all samples are sufficiently concordant, indicating that they
formed within a relatively short interval of time (about 90 Ma);
3. Because the present-day composition of the earth’s mantle falls on the
same isochron, the earth and the parent bodies of meteorites must have
formed at about the same time from the same primordial source.
Note that the decay rate derived by Luck and Allègre (1983) from the isoch-
ron of figure 11.27, assuming a common age of 4550 Ma for all investigated
samples ( ␭Re ⫽ 1.52 ⫾ 0.04 ⫻ 10⫺11 a⫺1 ), corresponds to a half-life of 4.56 ⫾
0.12 ⫻ 1010 a, which is somewhat higher than the direct determination of Lindner
et al. (1989) but consistent with the direct determination of Payne and Drever
(1965) [(4.7 ⫾ 0.5) ⫻ 1010 a].
Figure 11.28 shows the rhenium and osmium isotopic compositions of black
shales and sulfide ores from the Yukon Territory (Horan et al., 1994). The black
shale and sulfide layers are approximately isochronous. The superimposed refer-
ence isochrons bracket the depositional age of the enclosing shales. One reference
line represents the minimum age (367 Ma) with an initial ( 187Os/ 186Os) 0 ratio
of one, consistent with the mantle isotopic composition at that age (see later).
The other reference isochron is drawn for a maximum age of 380 Ma, with
( 187Os/ 186Os) 0 ⫽ 12 (the maximum value measured in terrigenous sediments).
Further examples of application of Re-Os dating of sediments can be found in
Ravizza and Turekian (1989).
764 / Methods
Figure 11.27 Re-Os isochron for iron meteorites and metallic phase of chondrites and
earth’s mantle. Reprinted with permission from J. M. Luck and C. J. Allègre, Nature, 302,
130–132, copyright 䉷 1983 Macmillan Magazines Limited.
Figure 11.28 Re-Os isochron for black shales and sulfide deposits of Yukon Territory
(Canada). Reprinted from M. F. Horan et al., Geochimica et Cosmochimica Acta, 58, 257–
265, copyright 䉷 1994, with kind permission from Elsevier Science Ltd., The Boulevard,
Langford Lane, Kidlington 0X5 1GB, UK.
14
11.7.8 C
Carbon has eight isotopes, from A ⫽ 9 to A ⫽ 16. Natural carbon is composed
of the two stable isotopes 12C (0.9889) and 13C (0.0111) and of radiogenic 14C,
continuously produced in the earth’s atmosphere mainly by (n,p) reaction of slow
(or “thermal”) neutrons with 14N:
7 N+ n → 14
6 C + p.
14
(11.119)
The 14C isotope reacts directly with atmospheric oxygen to produce 14CO2 and
14
CO, and undergoes isotopic exchanges with the preexisting 12CO2-12CO and
13
CO2-13CO molecules. Mixing of the various molecules is rapidly achieved in the
atmosphere (less than 2 years). Atmospheric carbon (1.4% of exchangeable car-
bon, mainly as CO2 ) equilibrates readily with oceanic carbon (H2CO3, HCO⫺ 3,
the various soluble carbonates, and very subordinate CO2⫺ 3 , amounting in total
766 / Methods
to 95% of exchangeable carbon) and with carbon in the terrestrial biosphere and
humus (3.2% of exchangeable carbon).
The decay of 14C to 14N takes place by ␤⫺ emission:
7 N+ β
→ 14 + v + 0.156 MeV .
14 −
6 C (11.120 )
The relatively rapid production rate of 14C (approximately 2s⫺1 cm⫺2 ) coupled
with high exchange kinetics (the mean time of residence in the atmosphere is
relatively short—i.e. 5 to 10 years) and a comparatively long half-life of the decay
process of equation 11.120 (5730 ⫾ 40 a; Godwin, 1962) ensure that a steady
state in the relative concentrations of the three carbon isotopes is quite rapidly
achieved. The steady state isotopic composition of carbon results in specific ra-
dioactivity of 14.9 dpm/g in the atmosphere and 13.6 dpm/g in the biosphere (the
lowering of specific radioactivity is a result of exchange kinetics among atmo-
spheric, oceanic, and biogenic reservoirs and of minor isotope fractionation
effects).
Living animals and plants acquire and maintain a constant level of 14C radio-
activity during their lifetimes as a result of continuous reequilibration with atmo-
sphere and biosphere. However, when they die, these continuous exchanges cease
and 14C radioactivity progressively decreases according to the exponential law
R t = R 0 exp( − λt ) , (11.121)
where R t is 14C radioactivity measured at time t after the death of the living organ-
ism, and R 0 is 14C radioactivity in the biosphere during its lifetime. Through some
substitutions, equation 11.121 can be rewritten as the chronological equation
R 
t = 8.2666 × 10 3 ln  0  . (11.122)
 Rt 
The range of applicability of equation 11.122 depends on the limits of detection of

14
C in the sample. The current maximum age attained by direct radioactivity counting is
about 4 ⫻ 104 a. To measure residual 14C radioactivity, the total carbon in the sample is
usually converted to CO2 and counted in the gas phase, either as purified CO2 or after
further conversion to C2H2 or CH4 . To enhance the amount of counted carbon, with the
same detection limit (about 0.1 dpm/g), 14C counters attain volumes of several liters and
operate at several bars. More recent methods of direct 14C detection (selective laser excita-
tion; Van de Graaff or cyclotron acceleration) has practically doubled the range of deter-
minable ages (Muller, 1979).
Using equation 11.122 implies knowledge of specific 14C radioactivity in the

biosphere at time t ⫽ 0. Calibration of 14C radioactivity over the past 7500 years
has been achieved by comparison of apparent radiocarbon ages with true ages
obtained by dendrochronology (measurement of annual growth rings in trees).

Discrepancies are translated into differences between the specific R 0 radioactivi-
ties necessary to achieve concordant ages and specific present-day radioactivity,
and are shown in figure 11.29 as per-mil deviations plotted against t.
Long-term sinusoidal fluctuations (period ⫽ 10,000 a, amplitude 艐 10 a) are
correlated with modifications of the earth’s dipole moment (solid line in figure
11.29; Damon et al., 1978). Short-term fluctuations (inset in figure 11.29; Suess,
1965) can be ascribed to variations in solar activity. Both phenomena affect the
intensity of the flux of cosmic rays striking the earth: both direct increases in the
magnetic field of the earth and enhanced magnetic activity associated with sun-
spots cause enhanced deflection of the low-energy cosmic rays responsible for
producing thermal neutrons. The effect of solar activity is well documented for
the last three centuries. In particular, the maximum positive deviations in ⌬‰
14
C-radioactivity around .. 1700 corresponds to the “Maunder minimum,” a
period of almost complete absence of sunspots from 1649 to 1715. Note also in
the inset of figure 11.29 the sudden present-day decrease in 14C radioactivity be-
cause of massive combustion of fossil fuels (virtually 14C-free), which results in
world-scale dilution of 14C with respect to stable isotopes 13C and 12C (Suess,
1965), partly offset by anthropic 14C production associated with the explosion of
nuclear weapons.
11.8 Application II: Stable Isotope Geothermometry
Although the scope of this book does not allow an appropriate treatment of stable
isotope compositions of earth’s materials (excellent monographs on this subject
can be found in the literature—e.g., Hoefs, 1980; Faure, 1986), we must neverthe-
less introduce the significance of the various compositional parameters adopted
in the literature before presenting the principles behind stable isotope geother-
mometry.
Let us consider again a generic isotopic exchange as represented by a reaction
of the type
a A ° + bB • ⇔ aA • + bB ° (11.123)
• •
(Urey, 1947), where A°, A and B°, B are, respectively, molecules of the same type
differing only in the isotopic composition of a given constituting element (open
and full symbols: light and heavy), and a and b are stoichiometric coefficients.
The “fractionation factor” ␣ represents the relative distribution of heavy and
light isotopes in the two phases at equilibrium (somewhat similar to the normal-
ized distribution coefficient adopted in trace element geochemistry; cf. section
10.8):
Figure 11.29 Specific radioactivity of 14C expressed as per-mil deviation (⌬‰) from present-day radioactivity over the last 7500 years,
derived from comparisons with dendrochronological studies. Reprinted from Damon et al. (1978a), with permission, from The Annual
Review of Earth and Planetary Sciences, Volume 6, copyright 1978 by Annual Reviews Inc. and from Suess (1965), Journal of Geophysi-
cal Research, 70, 5937–5952, copyright 1965 by the American Geophysical Union.
X•
 °
X A
α = , (11.124)
X•
 °
 X B
•
where (X /X °)A is the ratio of isotopic abundance in compound A. Assuming
random distribution of isotopes in the exchanging compounds, the fractionation
factor is related to equilibrium constant K (Epstein, 1959) by
α = K1 r , (11.125)
where r is the number of exchanging isotopes per formula unit. In the simplest
case, exchange reactions are written in such a way that only one isotope is ex-
changed. In this case, the fractionation factor obviously identifies with the equi-
librium constant (i.e., ␣ ⫽ K ).
Consider, for instance, the exchange
1 1 1 1
MgSi 16O3+ Mg2Si 18O4 ⇔ MgSi 18O3+ Mg2Si 16O4 .
3 4 3 4 (11.126)
opx ol opx ol
The equilibrium constant of the isotopic exchange is
[ O] [ O]
1 3 1 4
18 16
opx ol
K 126 = , (11.127)
[ O] [ O]
1 3 1 4
16 18
opx ol
where the terms in brackets denote activities
[ O]
18
opx
= X 183
O, opx
≡ X•
n
(11.128.1)
[ O]
18
ol
= X 418
O, o l
≡ X•
m
(11.128.2)
[ O]
n
16
= X 163 ≡ X° (11.128.3)
opx O, opx
[ O]
m
16
= X 164 ≡ X° . (11.128.4)
ol O, o l
770 / Methods
Assuming the same notation adopted in equation 11.123, we have
r = a × n = b × m = 1, ( 11.129 )
and hence
α = K 126 . ( 11.130 )
Because the modifications affecting the stable isotope compositions of natural

materials are rather limited (and mainly attributable to the processes discussed in
sections 11.6.1 to 11.6.3), they are usually expressed in terms of ⌬ per-mil varia-
tions:
(
 X• X° ) 
∆ AB = • A
(
− 1 × 10 3 = α − 1 × 10 3. )
(
 X X° ) B


(11.131)
The isotopic measurements are actually performed in the laboratories quantifying

the ‰ difference with respect to a standard material, expressed by the ␦ parame-
ter. Denoting as st the standard material of reference, we have
δA
(
 X• X°
= •
) A

− 1 × 10 3 (11.132.1)
 (
 X X° ) st


and
δB =  •
(
 X• X° ) B

− 1 × 10 3. (11.132.2)
 (
 X X° ) st


Standard materials generally adopted to define the isotopic compositions of

hydrogen and oxygen are the Standard Mean Composition of Oceanic Water
(SMOW; Craig, 1961), the NBS distilled water (National Bureau of Standards),
and the Standard Light Antarctic Precipitation (SLAP; International Atomic En-
ergy Agency). For carbon (and for oxygen in carbonates), the standard materials
are the PDB standard CO2 , produced by the University of Chicago from Creta-
ceous belemnites (Belemnitella Americana) of the Peedee Formation (South Car-
olina) and other working standards whose ␦13C with respect to PDB have been
defined by Craig (1957). For sulfur, the commonly adopted reference standard is
CD troilite (FeS) from the Canyon Diablo iron meteorite.
Based on the equations above, the relationships existing between the fraction-
ation factor ␣ and the ␦ and ⌬ parameters are
Table 11.8 Comparison among the various parameters

adopted to define the stable isotope compositions of natural
materials (after Hoefs, 1980).
␦A ␦B ⌬AB ␣AB 103 ln ␣AB

1.00 0 1 1.001 0.9995
10.00 0 10 1.01 9.95
20.00 0 20 1.02 19.80
10.00 5.00 4.98 1.00498 4.96
20.00 15.00 4.93 1.00493 4.91
30.00 20.00 9.80 1.00980 9.76
30.00 10.00 19.80 1.01980 19.61
δA 10 3 + 1 δA + 10 3
α AB = = (11.133)
δ B 10 3 + 1 δ B + 10 3
and
 δ A + 10 3 
∆ AB = − 1 × 10 3 . (11.134)
 δ B + 10 3 
Because
 X 
10 3 ln  1 +  ≅ X, (11.135)
 1000 
the following approximations are often adopted in stable isotope geochemistry:
δ A − δ B ≅ ∆ AB ≅ 10 3 ln α AB . (11.136 )
The closeness of these approximations depend on the modulus of the first term
|␦A ⫺ ␦B|, as shown in table 11.8.
11.8.1 Oxygen
Oxygen is present in nature with three stable isotopes: 16O (0.997630), 17O
(0.000375), and 18O (0.001995). The first application of oxygen isotopic fraction-
ation studies to geological problems was suggested by Urey (1947), who related
the temperature of precipitation of calcium carbonate in seawater to the (T-
dependent) isotopic fractionation induced by the exchange reaction between H2O
molecules and carbonate ion CO2⫺ 3 :
C 16 O 23 − + 3H18
2 O ⇔ C O 3 + 3H 2 O.
18 2 − 16
(11.137)
772 / Methods
Figure 11.30 Reduced partition function for various minerals calulated by Kieffer (1982)
through equation 11.61 plotted against T ⫺2. Heavy curve labeled H2O(l) is reduced parti-
tion function of water according to Becker (1971). Dashed curve is a T ⫺2 extrapolation of
high-T reduced partition curve for quartz. Mineral abbreviations: Qtz (quartz), Calc
(calcite), Albt (albite), Musc (muscovite), Enst (clinoenstatite), Anor (anorthite), Diop
(diopside), Pyrp (pyrope), Gros (grossular), Zron (zircon), Fors (forsterite), Andr (andra-
dite), Rutl (rutile). Reprinted with permission from Kieffer (1982), Review of Geophysics
and Space Physics, 20, 827–849, copyright 1982 by the American Geophysical Union.
Detailed vibrational calculations (Bottinga, 1968; Becker, 1971; Kieffer, 1982) as

well as accurate experiments (Clayton et al., 1989; Chiba et al., 1989, and refer-
ences therein) and semiempirical generalizations (Zheng, 1991, 1993a,b) have
been made since the work of Urey (1947) and ensure satisfactory knowledge of
the 18O/ 16O fractionation properties of minerals and water and mineral-mineral
couples as a function of T.
Figure 11.30 shows the reduced partition function (1/r) ⫻ 1000 ln f calcu-
lated by Kieffer (1982) by application of equation 11.61 and plotted against T ⫺2.
The adoption of this form of plot is dictated by the fact that the natural logarithm
of the partition function ratio is asymptotically proportional to T ⫺2 at high tem-
Table 11.9 Third-order polynomials relating reduced

partition functions to inverse of squared absolute
temperature x ⫽ 106 ⫻ T ⫺2 (from Clayton and
Kieffer, 1991).
Phase a b c
Calcite 11.781 ⫺0.420 0.0158
Quartz 12.116 ⫺0.370 0.0123
Albite 11.134 ⫺0.326 0.0104
Anorthite 9.993 ⫺0.271 0.0082
Diopside 9.237 ⫺0.199 0.0053
Forsterite 8.326 ⫺0.142 0.0032
Magnetite 5.674 ⫺0.038 0.0003
peratures (Urey, 1947; Bigeleisen and Mayer, 1947; see also section 11.6.1). In-
deed, a cubic function in T ⫺2 fits the calculated reduced partition functions within
⫾0.02 per mil, for all temperatures above 400 K.
The Kieffer model correctly predicts the systematic change of the reduced
partition functions of various minerals with structure, as indicated by Taylor and
Epstein (1962). For anhydrous silicates, the decrease in the sequence framework-
chain-orthosilicate reflects the decreasing frequency of antisymmetric Si-O
stretching modes. The internal frequencies of the carbonate ion give a high re-
duced partition function at all T. The value for rutile is low because of the low
frequencies of the Ti-O modes (Kieffer, 1982).
Combining the results of Kieffer’s model and of laboratory experiments,
Clayton and Kieffer (1991) obtained a set of third-order equations relating the
reduced partition functions of various minerals to the inverse of the squared abso-
lute temperature (table 11.9) according to
 1000   Q • 
fA =   ln   = ax + bx + cx
2 3
(11.138)
 r  Q°  A
x = 10 6 × T −2 . (11.139 )
Oxygen isotopic fractionation between two minerals A and B at any given T

can easily be obtained by subtracting their respective reduced partition function
equations—i.e.,
∆ AB = fA − f B . (11.140)
Equations 11.138 to 11.140 supersede the simple one-parameter equation pro-

posed by Clayton et al. (1989) and Chiba et al. (1989):
774 / Methods
Table 11.10 Comparison of polynomial and linear fits for mineral pairs
at 1000 K. Values in italics refer to equation 11.141 (from Clayton and
Kieffer, 1991).
Qz Ab An Di Fo Mt
Ca ⫺0.38 0.56 1.65 2.33 3.19 5.74
⫺0.38 0.56 1.61 2.37 3.29 5.91
Qz 0.94 2.03 2.72 3.57 6.12
0.94 1.99 2.75 3.67 6.29
Ab 1.09 1.78 2.63 5.18
1.05 1.81 2.73 5.35
An 0.69 1.54 4.09
0.76 1.68 4.30
Di 0.86 3.41
0.92 3.54
Fo 2.55
2.62
10 6
∆ AB = a AB × . (11.141)
T2
A comparison of the two methods for T ⫽ 1000 K is given in table 11.10. For
each entry, a positive value indicates 18O enrichment in the phase on the left. The
upper entry is relative to equations 11.138 to 11.140 and the lower entry to equa-
tion 11.141. Note that the table implicitly gives the parameters of the equation of
Clayton et al. (1989), because at 1000 K equation 11.141 reduces to ⌬AB ⫽ aAB.
Figure 11.31 shows the results of equations 11.138 to 11.141 applied to the
calcite-diopside couple compared with experimental evidence and with the indi-
cations of Kieffer’s (1982) model.
Reduced partition functions for oxides were calculated by Zheng (1991) on
the basis of the “modified increment method.”
This method derives from the observation that the degree of 18O enrichment in a set
of cogenetic silicate minerals (expressed as oxygen isotope index I-18O) can be related
to bond strengths in minerals (Taylor, 1968). The oxygen isotope index for phase A is
defined by
(M M ) *
3 2
n ct
( ) ∑Z
16 18
I− 18
O = × i ct′ − O , (11.142)
2 n (M M )
A 3 2 ct
1
O 16 18 A
where the asterisk denotes a reference phase. M16 and M18 are the formula weights of
isotopically light and heavy compounds, nO and nct are the stoichiometric numbers of
oxygens and cations per formula unit. Zct is the cationic charge, and the summation ex-
tends over all cations in the formula unit, i⬘ct-O is the “normalized 18O-increment” of a
cation-oxygen bond (Zheng, 1991):
Figure 11.31 Oxygen isotopic fractionation between calcite and diopside as obtained
from equations 11.131 to 11.133. Experimental data points are from Chiba et al. (1989).
Dotted line: indications from vibrational calculations following Kieffer’s (1982) procedure.
From Clayton and Kieffer (1991), reprinted with kind permission of The Geochemical
Society, Pennsylvania State University, University Park, Pennsylvania.
( )
q
i ct′ − O = i ct − O i Si − O . ( 11.143 )
In equation 11.143, q is a parameter introduced by Zheng (1993a) to account for different

types of bonds:
q = Z ctk k = −1, 0, + 1. ( 11.144 )
Integer k assumes a value of ⫺1 in complex anions (e.g., Al 3⫹ and Si4⫹ in feldspars), ⫹1

in weakly bonded complex cations (e.g., Na⫹ and Ca2⫹ in feldspars), and 0 for other
complex cations (e.g., Fe2⫹ and Mg2⫹ in garnets).
The 18O increment of a cation-oxygen bond is given by
i = Sct − O ln Wct − O . ( 11.145)
Zheng (1991) defined S ct-O “bond-strength” as
Z ct
Sct − O = . (11.146)
v ct Rct − O
Although equation 11.146 agrees with the principle of electrostatic valence of Pauling
(1929), normalization to the cation-to-oxygen distance Rct-O to account for bond variation
776 / Methods
with the degree of ionicity is not in line with commonly adopted generalizations (see sec-
tion 1.10.3).
The term Wct-O is related to the masses of cations and oxygen isotopes:
Wct − O =
(m ct )
+ m 18O m 18O
.
(m )m
(11.147)
ct + m 18O 16O
The reduced partition function ratio of the mineral is derived from the equations above
through
 1000   Q • 
fA =  (
 ln  = I −18O ) × exp
( (
 ∆E 1 − I−18 O ) ) 
A
 r   Q°  A A  RT 
 
(11.148)
 1000   Q  •
×  ln  .
 r   Q°  *
In equation 11.148, ⌬E is regarded as an Arrhenius-slope energy term that has a value of

•
1 kJ/mol, and (1000/r) ln(Q /Q°)* is the reduced partition function ratio for the reference
phase (quartz).
Calculated 18O increments for various oxides with different coordination

states are listed in table 11.11.
Calculation of reduced partition normalized fractionation factors, through
the modified increment method, allows the derivation of T-dependent mineral-
mineral oxygen isotopic fractionations expressed in the polynomial form
∆ AB = a × 10 6 T −2 + b × 10 3 T −1 + c. (11.149)
The T-dependency selected by Zheng (1991, 1993a,b) differs from the equation of
Clayton and Kieffer (1991), and the two methods do not give concordant results,
especially at low T, as shown in figure 11.32.
Zheng (1993b) also extended the modified increment method to the evalua-
tion of reduced partition function ratios in hydroxyl-bearing silicate minerals
through
2nO − nOH 18
( I− O )
18
OH− A
=
2 nO
I− O ,
A
( ) (11.150)
where nO and nOH represent the numbers of oxygens and hydroxyl groups in the
hydrous mineral and (I⫺18O)A is the isotope index in the anhydrous counterpart.
Although largely empirical, the method of Zheng (1991, 1993a,b) has the
Table 11.11 Calculation of 18O increments for silicates (from Zheng, 1993a).
Bond Zct nct nO Rct-O mct Wct-O Sct-O ict-O

Si-O 4 4 2 1.61 28.09 1.03748 0.62112 0.02285
Si-O 4 4 3 1.62 28.09 1.03748 0.61728 0.02271
Si-O 4 4 4 1.64 28.09 1.03748 0.60976 0.02244
Si-O 4 6 3 1.76 28.09 1.03748 0.37881 0.01394
Al-O 3 4 2 1.74 26.98 1.03690 0.43103 0.01562
Al-O 3 4 3 1.75 26.98 1.03690 0.42861 0.01553
Al-O 3 4 4 1.77 26.98 1.03690 0.42373 0.01535
Al-O 3 5 3 1.84 26.98 1.03690 0.32612 0.01182
Al-O 3 6 3 1.89 26.98 1.03690 0.26455 0.00959
Al-O 3 6 4 1.91 26.98 1.03690 0.26178 0.00949
Fe-O 3 6 3 1.96 55.85 1.04631 0.25541 0.01156
Fe-O 3 6 4 1.98 55.85 1.04631 0.25253 0.01126
Fe-O 2 6 3 2.06 55.85 1.04631 0.16181 0.00733
Fe-O 2 6 4 2.08 55.85 1.04631 0.16064 0.00728
Fe-O 2 8 4 2.16 55.85 1.04631 0.11574 0.00524
Mg-O 2 6 3 2.08 24.31 1.03537 0.16026 0.00557
Mg-O 2 6 4 2.10 24.31 1.03537 0.15873 0.00552
Mg-O 2 8 4 2.25 24.31 1.03537 0.11111 0.00386
Mn-O 2 6 3 2.11 54.94 1.04613 0.15798 0.00712
Mn-O 2 6 4 2.13 54.94 1.04613 0.15649 0.00706
Mn-O 2 8 4 2.31 54.94 1.04613 0.10823 0.00488
Ca-O 2 6 3 2.36 40.08 1.04234 0.14124 0.00586
Ca-O 2 6 4 2.38 40.08 1.04234 0.14006 0.00581
Ca-O 2 7 3 2.42 40.08 1.04234 0.11806 0.00489
Ca-O 2 8 3 2.48 40.08 1.04234 0.10081 0.00418
Ca-O 2 8 4 2.50 40.08 1.04234 0.10000 0.00415
Ba-O 2 9 3 2.80 137.34 1.05381 0.07937 0.00416
K-O 1 6 3 2.74 39.10 1.04205 0.06283 0.00251
K-O 1 9 3 2.90 39.10 1.04205 0.03831 0.00158
K-O 1 10 3 2.92 39.10 1.04205 0.03425 0.00141
K-O 1 12 3 2.96 39.10 1.04205 0.02815 0.00116
Na-O 1 6 3 2.38 22.99 1.03454 0.07003 0.00238
Na-O 1 8 3 2.52 22.99 1.03454 0.04960 0.00168
Na-O 1 8 4 2.54 22.99 1.03454 0.04921 0.00167
Li-O 1 6 3 2.10 6.94 1.01729 0.07937 0.00136
Be-O 2 4 3 1.63 9.01 1.02064 0.30675 0.00627
Zn-O 2 4 3 1.96 65.37 1.04798 0.25510 0.01196
Cr-O 3 6 4 1.995 52.00 1.04551 0.25063 0.01115
Ni-O 2 6 4 2.07 58.71 1.04674 0.16103 0.00736
Ti-O 4 6 3 1.965 47.90 1.04455 0.33927 0.01479
Sn-O 4 6 3 2.05 118.69 1.05300 0.32520 0.01679
Th-O 4 8 3 2.40 232.04 1.05655 0.20833 0.01146
U-O 4 8 3 2.38 238.03 1.05665 0.21008 0.01158
Zr-O 4 8 3 2.20 91.22 1.05090 0.22727 0.01128
Hf-O 4 8 3 2.19 178.49 1.05525 0.22831 0.01228
Figure 11.32 Oxygen isotopic fractionation factors between quartz and common rock-
forming minerals, as obtained with modified increment method (Zheng 1993a), compared
with indications based on fractionaton equations of Clayton and Kieffer (1991) (dashed
lines). Reproduced with modifications from Y. F. Zheng, Geochimica et Cosmochimica
Acta, 57, 1079–1091, copyright 1993, with kind permission from Elsevier Science Ltd.,
The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
advantage of furnishing quite extensive estimates of fractionation factors in

mineral-mineral and mineral-water couples, which are valid over a wide T range
(i.e., 0 to 1200 °C; see table 11.12).
11.8.2 Hydrogen
Hydrogen has two stable isotopes, 11H (0.99985) and 21H (0.00015), and a short-
lived radioactive isotope 31H (tritium) produced in the atmosphere by interaction
of 14N with cosmic ray neutrons:
14
7 N + n → 13 H + 12
6 C. (11.151)
The concentration of 31H in the atmosphere (measured in Tritium Units, or TU;

1 TU ⫽ 31H/ 11H ⫻ 1018 ) has been rapidly increasing (from 1952 to 1962) because
of nuclear tests. Tritium decays to stable 32He by ␤⫺ emission (t1/2 ⫽ 12.26 a; see
Kamensky et al., 1991 for an account of 3H-3He dating of groundwaters).
Isotope 21H (deuterium), discovered by Urey et al. (1932), is usually denoted by sym-
bol D. The large relative mass difference between H and D induces significant fraction-
ation ascribable to equilibrium, kinetic, and diffusional effects. The main difference in the
calculation of equilibrium isotopic fractionation effects in hydrogen molecules with re-
spect to oxygen arises from the fact that the rotational partition function of hydrogen is
nonclassical. Rotational contributions to the isotopic fractionation do not cancel out at
high T, as in the classical approximation, and must be accounted for in the estimates of
the partition function ratio f .
The classical expression for Qrot is
* = hcB ប2
Qrot = , (11.152)
kT 4 I kT
where B is the rotational constant in cm⫺1, I is the moment of inertia, and the remaining
symbols have their usual meanings. Bigeleisen and Mayer (1947), to account for the non-
classical behavior of hydrogen molecules, suggested expansion of the deviation from the
classical rotational partition function in the polynomial form
Qrot  σ σ2 σ3 
= 1 + + + + L ,
∗
Qrot  3 15 315  (11.153)
where ␴ ⫽ 1/Q*rot.
Because the partition function ratio f is defined in such a way that the classical rota-
tional and translational contributions are canceled, equations 11.40, 11.41 and 11.43 must
be modified by introducing the ratio of the deviations from classical rotational behavior
of heavy and light hydrogen molecules. For small values of ␴, this contribution reduces to
780 / Methods
Table 11.12 Oxygen isotope fractionation in quartz-mineral, mineral-water, and calcite-mineral

couples, according to modified increment method of Zheng (1991, 1993a,b).
⌬ A B ⫽ a ⫻ 10 6 T ⫺ 2 ⫹ b ⫻ 10 3 T ⫺ 1 ⫹ c (from Zheng, 1991, 1993a,b).
Quartz-Mineral Mineral-Water Calcite-Mineral
Mineral a b c a b c a b c
Quartz 4.48 ⫺4.77 1.71 ⫺0.47 0.10 0
Stishovite 0.22 1.88 ⫺0.78 4.26 ⫺6.64 2.08 ⫺0.25 1.98 ⫺0.78
K-feldspar 0.16 1.50 ⫺0.62 4.32 ⫺6.27 2.00 ⫺0.30 1.60 ⫺0.62
Albite 0.15 1.39 ⫺0.57 4.33 ⫺6.15 1.98 ⫺0.32 1.49 ⫺0.57
Anorthite 0.36 2.73 ⫺1.14 4.12 ⫺7.50 2.24 ⫺0.11 2.83 ⫺1.14
Leucite 0.22 1.91 ⫺0.79 4.26 ⫺6.67 2.08 ⫺0.24 2.01 ⫺0.79
Nepheline 0.37 2.77 ⫺1.15 4.11 ⫺7.53 2.24 ⫺0.10 2.89 ⫺1.15
Diopside 0.56 3.67 ⫺1.53 3.92 ⫺8.43 2.40 0.10 3.78 ⫺1.53
Enstatite 0.51 3.45 ⫺1.44 3.97 ⫺8.22 2.37 0.05 3.55 ⫺1.44
Hedenbergite 0.55 3.61 ⫺1.51 3.93 ⫺8.37 2.40 0.08 3.71 ⫺1.51
Ferrosilite 0.48 3.33 ⫺1.39 3.99 ⫺8.09 2.35 0.02 3.43 ⫺1.39
Jadeite 0.31 2.42 ⫺1.01 4.17 ⫺7.19 2.18 ⫺0.16 2.52 ⫺1.01
Acmite 0.30 2.38 ⫺0.99 4.18 ⫺7.14 2.17 ⫺0.17 2.45 ⫺0.99
Wollastonite 0.67 4.11 ⫺1.72 3.81 ⫺8.87 2.49 0.21 4.21 ⫺1.72
Rhodonite 0.63 3.94 ⫺1.65 3.85 ⫺8.71 2.46 0.16 4.04 ⫺1.65
Beryl 0.50 3.40 ⫺1.42 3.98 ⫺8.16 2.34 ⫺0.03 3.50 ⫺1.42
Cordierite 0.38 2.85 ⫺1.19 4.10 ⫺7.62 2.26 ⫺0.08 2.95 ⫺1.19
Almandine 0.72 4.26 ⫺1.79 3.76 ⫺9.02 2.52 0.25 4.36 ⫺1.79
Pyrope 0.73 4.30 ⫺1.80 3.75 ⫺9.07 2.52 0.26 4.40 ⫺1.80
Spessartine 0.73 4.31 ⫺1.81 3.75 ⫺9.07 2.52 0.26 4.41 ⫺1.81
Grossular 0.74 4.35 ⫺1.82 3.74 ⫺9.11 2.52 0.27 4.45 ⫺1.82
Andradite 0.72 4.29 ⫺1.80 3.76 ⫺9.05 2.52 0.26 4.38 ⫺1.80
Uvarovite 0.72 4.28 ⫺1.80 3.76 ⫺9.05 2.52 0.26 4.39 ⫺1.80
Titanite 0.67 4.11 ⫺1.72 3.81 ⫺8.87 2.49 0.21 4.21 ⫺1.72
Malayaite 0.60 3.83 ⫺1.60 3.88 ⫺8.60 2.43 0.14 3.94 ⫺1.60
Zircon 0.72 4.26 ⫺1.79 3.76 ⫺9.03 2.52 0.25 4.37 ⫺1.79
Thorite 0.79 4.50 ⫺1.89 3.69 ⫺9.27 2.55 0.32 4.61 ⫺1.89
Fayalite 0.84 4.69 ⫺1.97 3.64 ⫺9.46 2.59 0.38 4.79 ⫺1.97
Forsterite 0.93 4.95 ⫺2.09 3.55 ⫺9.72 2.64 0.46 5.05 ⫺2.09
Tephroite 0.86 4.75 ⫺2.00 3.62 ⫺9.51 2.60 0.39 4.85 ⫺2.00
Willemite 0.69 4.18 ⫺1.75 3.79 ⫺8.94 2.50 0.23 4.28 ⫺1.75
Phenacite 1.06 5.34 ⫺2.26 3.42 ⫺10.11 2.70 0.60 5.45 ⫺2.26
Sillimanite 0.22 1.89 ⫺0.78 4.26 ⫺6.65 2.07 ⫺0.25 1.99 ⫺0.78
Andalusite 0.28 2.26 ⫺0.94 4.20 ⫺7.02 2.15 ⫺0.19 2.36 ⫺0.94
Kyanite 0.32 2.52 ⫺1.05 4.16 ⫺7.29 2.20 ⫺0.14 2.63 ⫺1.05
Muscovite 0.38 2.84 ⫺1.18 4.10 ⫺7.61 2.25 ⫺0.08 2.97 ⫺1.18
Paragonite 0.37 2.77 ⫺1.15 4.11 ⫺7.54 2.24 ⫺0.10 2.87 ⫺1.15
Margarite 0.38 2.84 ⫺1.18 4.10 ⫺7.61 2.25 ⫺0.08 2.94 ⫺1.18
Illite 0.34 2.60 ⫺1.08 4.14 ⫺7.36 2.21 ⫺0.13 2.70 ⫺1.08
Phengite 0.35 2.64 ⫺1.10 4.13 ⫺7.41 2.22 ⫺0.12 2.74 ⫺1.10
Glauconite 0.49 3.34 ⫺1.39 3.99 ⫺8.11 2.34 0.02 3.44 ⫺1.39
Phlogopite 0.62 3.92 ⫺1.64 3.86 ⫺8.68 2.45 0.16 4.02 ⫺1.64
Annite 0.59 3.79 ⫺1.59 3.89 ⫺8.56 2.43 0.13 3.89 ⫺1.59
Biotite 0.64 3.99 ⫺1.67 3.84 ⫺8.76 2.46 0.18 4.10 ⫺1.67
Lepidolite 0.38 2.81 ⫺1.17 4.10 ⫺7.58 2.25 ⫺0.09 2.91 ⫺1.17
Hornblende 0.59 3.80 ⫺1.59 3.89 ⫺8.56 2.43 0.13 3.90 ⫺1.59
Tremolite 0.53 3.52 ⫺1.47 3.95 ⫺8.28 2.38 0.06 3.62 ⫺1.47
Actinolite 0.52 3.48 ⫺1.45 3.96 ⫺8.25 2.37 0.05 3.58 ⫺1.45
Cummingtonite 0.52 3.48 ⫺1.45 3.96 ⫺8.25 2.37 0.05 3.58 ⫺1.45
Grunerite 0.50 3.40 ⫺1.42 3.98 ⫺8.17 2.36 0.03 3.50 ⫺1.42
Glaucophane 0.45 3.18 ⫺1.32 4.03 ⫺7.94 2.31 ⫺0.02 3.28 ⫺1.32
Riebeckite 0.43 3.08 ⫺1.28 4.05 ⫺7.85 2.30 ⫺0.04 3.18 ⫺1.28
Pargasite 0.71 4.22 ⫺1.77 3.77 ⫺8.99 2.51 0.24 4.32 ⫺1.77
Anthophyllite 0.52 3.49 ⫺1.46 3.96 ⫺8.26 2.38 0.06 3.60 ⫺1.46
Gedrite 0.63 3.96 ⫺1.66 3.85 ⫺8.72 2.46 0.17 4.06 ⫺1.66
Kaolinite 0.19 1.68 ⫺0.69 4.29 ⫺6.44 2.03 ⫺0.28 1.78 ⫺0.69
Lizardite 0.51 3.42 ⫺1.43 3.97 ⫺8.19 2.36 0.04 3.52 ⫺1.43
Amesite 0.53 3.53 ⫺1.47 3.95 ⫺8.30 2.38 0.06 3.63 ⫺1.47
Pyrophyllite 0.08 0.86 ⫺0.35 4.40 ⫺5.62 1.87 ⫺0.38 0.96 ⫺0.35
Talc 0.28 2.27 ⫺0.94 4.20 ⫺7.04 2.14 ⫺0.19 2.38 ⫺0.94
Serpentine 0.49 3.35 ⫺1.40 3.99 ⫺8.12 2.35 0.02 3.45 ⫺1.40
Pennine 0.51 3.43 ⫺1.43 3.97 ⫺8.19 2.36 0.04 3.53 ⫺1.43
Clinochlore 0.51 3.43 ⫺1.43 3.97 ⫺8.19 2.36 0.04 3.53 ⫺1.43
Chamosite 0.42 3.04 ⫺1.27 4.06 ⫺7.81 2.30 ⫺0.04 3.14 ⫺1.27
Thuringite 0.58 3.73 ⫺1.56 3.90 ⫺8.50 2.42 0.11 3.84 ⫺1.56
Zoisite 0.43 3.07 ⫺1.28 4.05 ⫺7.84 2.30 ⫺0.04 3.18 ⫺1.28
Epidote 0.42 3.05 ⫺1.27 4.05 ⫺7.81 2.29 ⫺0.04 3.15 ⫺1.27
Vesuvianite 0.93 4.97 ⫺2.09 3.55 ⫺9.74 2.63 0.47 5.07 ⫺2.09
Ilvaite 1.05 5.32 ⫺2.24 3.43 ⫺10.08 2.69 0.58 5.42 ⫺2.24
Norbergite 0.86 4.75 ⫺2.00 3.62 ⫺9.51 2.60 0.39 4.85 ⫺2.00
Chondrodite 0.74 4.34 ⫺1.82 3.73 ⫺9.11 2.52 0.27 4.45 ⫺1.82
Humite 0.71 4.24 ⫺1.78 3.77 ⫺9.01 2.51 0.24 4.34 ⫺1.78
Clinohumite 0.70 4.19 ⫺1.75 3.78 ⫺8.95 2.50 0.23 4.29 ⫺1.75
Staurolite 0.39 2.90 ⫺1.21 4.09 ⫺7.66 2.27 ⫺0.07 3.00 ⫺1.21
Topaz 0.30 2.41 ⫺1.00 4.18 ⫺7.18 2.17 ⫺0.16 2.52 ⫺1.00
Datolite 0.15 1.44 ⫺0.60 4.33 ⫺6.21 1.99 ⫺0.31 1.54 ⫺0.60
Chloritoid 0.49 3.35 ⫺1.40 3.99 ⫺8.11 2.35 0.02 3.45 ⫺1.40
Tourmaline 0.27 2.22 ⫺0.92 4.21 ⫺6.99 2.14 ⫺0.20 2.32 ⫺0.92
Axinite 0.28 2.26 ⫺0.94 4.20 ⫺7.02 2.15 ⫺0.19 2.36 ⫺0.94
Pectolite 0.41 2.95 ⫺1.23 4.08 ⫺7.71 2.28 ⫺0.06 3.05 ⫺1.23
Prehnite 0.30 2.40 ⫺1.00 4.18 ⫺7.17 2.18 0.17 2.50 ⫺1.00
Rutile 0.79 6.82 ⫺3.59 3.45 ⫺10.60 2.55
Pyrolusite 0.64 6.16 ⫺3.23 3.61 ⫺9.94 2.19
Cassiterite 0.56 5.80 ⫺3.04 3.68 ⫺9.58 2.00
782 / Methods
Plattnerite 0.56 5.80 ⫺3.04 3.68 ⫺9.58 2.00
Cerianite 1.67 9.32 ⫺4.97 2.57 ⫺13.09 3.93
Thorianite 1.66 9.30 ⫺4.96 2.58 ⫺13.07 3.92
Uraninite 1.61 9.19 ⫺4.90 2.63 ⫺12.97 3.86
Ilmaite 1.36 8.61 ⫺4.57 2.88 ⫺12.38 3.53
Geikelite 1.46 8.85 ⫺4.70 2.78 ⫺12.63 3.66
Hematite 1.55 9.05 ⫺4.82 2.69 ⫺12.82 3.78
Corundum 2.00 9.94 ⫺5.32 2.24 ⫺13.71 4.28
Magnetite 1.22 8.22 ⫺4.35 3.02 ⫺12.00 3.31
Jacobsite 1.23 8.24 ⫺4.37 3.02 ⫺12.02 3.33
Magnesioferrite 1.29 8.43 ⫺4.47 2.95 ⫺12.20 3.43
Ulvöspinel 1.43 8.78 ⫺4.67 2.81 ⫺12.56 3.63
Chromite 1.63 9.22 ⫺4.91 2.62 ⫺12.99 3.87
Magnesiochromite 1.76 9.49 ⫺5.07 2.48 ⫺13.27 4.03
Hercynite 1.92 9.80 ⫺5.24 2.32 ⫺13.58 4.20
Spinel 2.06 10.04 ⫺5.38 2.18 ⫺13.81 4.34
( )
• 2
•
σ −σ° σ −σ° σ
•2
−σ°
2
1+ + + . (11.154)
3 18 90
In a different approach, Bottinga (1969a) evaluated the nonclassical rotational parti-

tion function for water vapor, molecular hydrogen, and methane through the asymptotic
expansion of Strip and Kirkwood (1951):
Qrot = 
 π 
12

1 +
[
2s x + 2s y + 2s z − (s x s y s z ) − (s y s z s x ) − (s z s x s y ) 
+ L,
]

 sxsy sz   12 
( 11.155)
where sx is the classical rotational partition function contribution about the

x-axis:
∗ = ប2
s x = Qrot, x , (11.156)
4 I x kT
and Ix is the moment of inertia about the x-axis.
Bottinga (1969a) calculated the isotopic fractionation relative to the ex-

change reactions
1 1
H 2O+ D ⇔ D 2O+ H (11.157)
2 2
1 1
CH 4 + D ⇔ CD 4 + H (11.158)
4 4
1 1 1 1
H 2 O + CD 4 ⇔ D 2 O + CH 4 . (11.159)
2 4 2 4
Results of his calculations are listed in table 11.13 in terms of fractionation factors
␣ and ⌬A* B. The ⌬A*B factors for methane–water vapor and hydrogen–water vapor
couples are plotted in figure 11.33 as a function of T (°C).
Note that Bottinga (1969a) defines ⌬AB * as
∆∗AB = (α − 1) × 1000 (11.160)
(see eq. 11.133 for the significance of ␣ ). Taking the water vapor as phase B, the
parameter ⌬AB * is equal to the deuterium enrichment in phase A with respect to
water, and hence is equivalent to the ␦ notation adopted in the geochemical litera-
ture, under the assumption of equilibrium.
The hydrogen isotope fractionation between OH-bearing minerals and water
is a sensitive function of T. Figure 11.34 shows the experimental results of Suzu-
oki and Epstein (1976). The interpolant expressions for muscovite-water, biotite-
water, and hornblende-water in the T range 450 to 850 °C are
22.1 × 10 6
10 3 ln α muscovite− water = − + 19.1 (11.161.1)
T2
21.3 × 10 6
10 3 ln α biotite− water = − − 2.8 (11.161.2)
T2
23.9 × 10 6
10 3 ln α hornblende− water = − + 7.9 (11.161.3)
T2
Because the slope coefficient in the equations above is practically constant, within
experimental uncertainty, the T-dependency of the hydrogen fractionation factor
for mica and amphibole can be generalized as
22.4 × 10 6
10 3 ln α mineral − water = −
T2 (11.162)
+ 28.2 + 2 X Al − 4 X Mg − 68 X Fe
(Suzuoki and Epstein, 1976), where XAl, XMg, and XFe are molar fractions of
sixfold coordinated cations. Equation 1.162 implies that hydrogen isotope frac-
tionation among coexisting mica and amphibole is temperature independent and
784 / Methods
Table 11.13 Deuterium fractionation in the water-hydrogen-methane

system, after Bottinga (1969) (wv ⫽ water vapor, me ⫽ methane,
hy ⫽ hydrogen).
ln ␣ 103 ln ␣ ⌬AB
*
T (°C) wv-hy me-hy wv-me me-wv hy-wv

0 1.395 1.387 7.3 ⫺7.3 ⫺752.1
10 1.355 1.322 13.3 ⫺13.3 ⫺736.9
20 1.280 1.261 18.9 ⫺18.7 ⫺721.9
25 1.253 1.232 21.8 ⫺21.2 ⫺714.5
30 1.228 1.204 24.0 ⫺23.7 ⫺707.1
40 1.179 1.151 28.7 ⫺28.2 ⫺692.5
50 1.134 1.101 33.0 ⫺32.4 ⫺678.2
60 1.091 1.054 36.9 ⫺36.3 ⫺664.1
70 1.051 1.010 40.6 ⫺39.8 ⫺650.3
80 1.012 0.968 43.9 ⫺43.0 ⫺636.7
90 0.977 0.930 47.0 ⫺45.9 ⫺623.4
100 0.943 0.893 49.9 ⫺48.6 ⫺610.4
120 0.880 0.825 54.8 ⫺53.4 ⫺585.2
140 0.824 0.765 59.0 ⫺57.3 ⫺561.1
160 0.773 0.710 62.5 ⫺60.6 ⫺538.2
180 0.726 0.661 65.4 ⫺63.3 ⫺516.3
200 0.684 0.616 67.8 ⫺65.5 ⫺495.4
220 0.645 0.576 69.6 ⫺67.2 ⫺475.5
240 0.610 0.539 71.1 ⫺68.6 ⫺456.5
260 0.577 0.505 72.2 ⫺69.7 ⫺438.4
280 0.547 0.474 73.0 ⫺70.4 ⫺421.2
300 0.519 0.445 73.5 ⫺70.9 ⫺404.8
320 0.493 0.419 73.8 ⫺71.2 ⫺389.2
340 0.469 0.395 73.9 ⫺71.3 ⫺374.3
360 0.446 0.372 73.8 ⫺71.2 ⫺360.1
380 0.425 0.352 73.5 ⫺70.9 ⫺346.5
400 0.406 0.333 73.2 ⫺70.5 ⫺333.6
450 0.362 0.291 71.7 ⫺69.2 ⫺303.9
500 0.325 0.255 69.6 ⫺67.3 ⫺277.5
550 0.293 0.226 67.2 ⫺65.0 ⫺253.9
600 0.265 0.201 64.7 ⫺62.3 ⫺232.9
650 0.241 0.179 61.9 ⫺60.0 ⫺214.1
700 0.220 0.161 59.1 ⫺57.4 ⫺197.3
cannot be used as a geothermometer. Moreover, based on equation 11.162, T

being equal, minerals in which Al-OH bonds predominate (i.e., muscovite, for
example) are richer in D with respect to phases with prevalent Mg-OH bonds
(biotite), and minerals with the same Mg/Fe ratio have the same D-enrichment
(biotite-hornblende).
Equation 11.162 cannot be extended to other OH-bearing phases such as
serpentine, chlorite, and clay minerals (illite, smectite). Concerning serpentine,
the experiments conducted by Sakai and Tsutsumi (1978) on clinochrysotile at 2
Figure 11.33 Calculated oxygen isotope fractionation factors ⌬ *AB for methane–water va-
por and hydrogen–water vapor couples plotted against T (°C). Reprinted from Y. Bottinga,
Geochimica et Cosmochimica Acta, 33, 49–64, copyright 䉷 1969, with kind permission
from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
kbar water pressure in the T range 100 to 500 °C lead to the following polyno-
mial expansion:
27.5 × 10 6 7.69 × 10 4
10 3 ln α serpentine− water = − + 40.8. (11.163)
T2 T
Note that the T-dependency in equation 11.163 is inverse with respect to equation
11.162 (see, however, Wenner and Taylor, 1971).
Concerning clay minerals (illite and smectite), Capuano (1992) proposed a
simple linear dependency of the mineral-water fractionation factor over recipro-
cal temperature:
45.3 × 10 3
10 3 ln α clay − water = − + 94.7. (11.164)
T
Equation 11.164 is valid in the T range 0 to 150 °C. The same sort of T-
dependency (although less marked) was envisaged by Yeh (1980) (slope coeffi-
cient ⫽ ⫺19.6 ⫻ 103, intercept ⫽ 25).
One of the main applications of hydrogen and oxygen isotope thermometry in
geochemistry is the estimation of the reservoir temperatures of active geothermal
systems or the evaluation of the ruling T conditions during deposition or alter-
Figure 11.34 Fractionation factors versus T ⫺2 for mica-water and amphibole-water
couples. Reprinted from T. Suzuoki and S. Epstein, Geochimica et Cosmochimica Acta,
40, 1229–1240, copyright 䉷 1976, with kind permission from Elsevier Science Ltd., The
Boulevard, Langford Lane, Kidlington 0X5 1GB, UK.
ation of mineral assemblages in sedimentary basins and fossil geothermal sys-

tems. Such estimates are, however, subject to serious errors if salinity effects on
the activity coefficient of HDO or H218O solvent molecules in the aqueous fluids
are not attentively evaluated (the same errors could also affect generalizations of
the type exemplified by eq. 11.163 and 11.164).
Because HDO or H218O molecules cannot be expected to have the same solu-
bility in the brine of the main solvent species H216O (hereafter simply denoted
H2O), their differential behavior must be accounted for by introduction of the
appropriate activity coefficient ratio (or “isotope salt effect”; cf. Horita et al.,
1993a,b) ⌫:
γ HDO
ΓD = (11.165.1)
γ H2 O
γ 18
H2 O
ΓO = . (11.165.2)
γ H 2O
Denoting as Ractivity the activity ratio [HDO]/[H2O] or [H218O]/[H2O] and as

Rcomposition corresponding molar ratios (HDO)/(H2O) and (H218O)/(H2O), the
isotope salt effect can also be expressed as
R activity 1 + 10 −3 δ activity
Γ = = (11.166)
R composition 1 + 10 −3 δ composition
(Horita et al., 1993a,b), where ␦ is the conventional delta value (‰). Combining
equations 11.165 and 11.166, through some substitutions, we can derive
10 3 ln α A − brine(activity) = 10 3 ln α A − brine( comp ) − 10 3 ln Γ brine . (11.167 )
Both the oxygen and hydrogen isotope salt effects were determined by Horita
et al. (1993a) in single-salt (NaCl, KCl, MgCl2 , CaCl2 , Na2SO4 , and MgSO4 )
aqueous solutions of various molalities, and their results can be represented in
the simple linear form
 b
10 3 ln Γ = m a +  , (11.168)
 T
where m is molality, and a and b are fitting parameters whose values are listed in
table 11.14.
788 / Methods
Table 11.14 Calibration of “isotope salt effect” in single-salt solutions

(eq. 11.161), after Horita et al. (1993a).
Salt Isotope a (mol⫺1 ) b (mol⫺1 ⫻ K) T Range (°C)

NaCl (D) ⫺2.89 1503.1 10–100
(18O) ⫺0.015 0 25–100
KCl (D) ⫺5.10 2278.4 20–100
(18O) ⫺0.612 230.83 25–100
CaCl2 (D) ⫺2.34 2318.2 50–100
(18O) ⫺0.368 0 50–100
MgCl2 (D) ⫹4.14 0 50–100
(18O) ⫹0.841 ⫺582.73 25–100
Na2SO4 (D) ⫹0.86 0 50–100
(18O) ⫺0.143 0 50–100
MgSO4 (D) ⫹8.45 ⫺2221.8 50–100
(18O) ⫹0.414 ⫺432.33 0–100
Neglecting the isotope salt effect leads to erroneous deductions concerning

the T of equilibration between minerals and brines. Figure 11.35 shows, for ex-
ample, the isotope salt effect in the case of kaolinite-water equilibrium. In the
case illustrated, the analyzed isotopic compositions of kaolinite and aqueous so-
lution yield a fractionation factor 103 ln ␣ of ⫺15 for deuterium. If the kaolinite–
pure water fractionation line of Liu and Epstein (1984) is utilized, this value
would suggest an equilibrium temperature of 88 °C. Imagine, however, the ana-
lyzed aqueous solution to be rich in saline components (as are the majority of
geothermal fluids); in this case the actual T of equilibrium would be 81 °C (1m
NaCl solution), 64 °C (3m NaCl solution), or 29 °C in the case of the Salton Sea
brine (3.14m NaCl, 0.882m CaCl2 , 0.588m KCl). In the last case, the error in-
duced by neglecting the isotope salt effect would thus be on the order of 300%.
The same problems encountered in mineral-water isotopic thermometry
could be present in T estimates based on isotopic equilibrium between brine and
dissolved species, such as SO2⫺4 . Isotope geothermometers based on mineral cou-
ples or gaseous couples (e.g., CH4-CO2 ), however, are not affected by isotopic
salt effect in brines (Horita et al., 1993b).
11.8.3 Carbon
Fractionation factors for the two stable carbon isotopes 12C and 13C have been
calculated by Bottinga (1969a,b) for exchanges among graphite, calcite, carbon
dioxide, and methane and for graphite-CO2 and diamond-CO2 equilibria. Results
of vibrational calculations of Bottinga (1969a,b) in the T range 0 to 1000 °C are
listed in table 11.15 and graphically displayed in figure 11.36.
The main points arising from the study of Bottinga (1969a) are as follows:
Figure 11.35 Estimation of formation temperature from measured ␦ D values of coexist-

ing kaolinite and aqueous solution on the isotope composition scale. The kaolinite–pure
water line is from Liu and Epstein (1984). Dashed lines show the shift induced by increas-
ing salinity of solution. Reprinted from J. Horita, D. R. Cole, and D. J. Wesolowski, Geo-
chimica et Cosmochimica Acta, 57, 4703–4711, copyright 䉷 1993, with kind permission
from Elsevier Science Ltd., The Boulevard, Langford Lane, Kidlington 0X5 1GB UK.
1. The carbon isotope fractionation between CO2 and methane and between
graphite and methane is a sensitive function of temperature.
2. The isotope composition of graphite does not characterize its origin be-
cause it is not unique to any simple method of condensation.
3. The carbon fractionation factors between gaseous CO2 and CH4 are
rather large, especially at low T. Temperatures deduced by application of
13
C fractionation geothermometers are, however, reasonable for either
CO2 or CH4 produced by bacterial decomposition of organic matter and
present in gaseous emissions associated with geothermal fields (cf. tables
9 and 10 in Bottinga, 1969a).
790 / Methods
Table 11.15 Calculated fractionation factors (1000 ln ␣) for 13C exchange among carbon dioxide,
calcite, graphite, diamond, and methane (Bottinga 1969a,b). Cc ⫽ calcite; Gr ⫽ graphite;
Dm ⫽ diamond.
1000 ln ␣AB
T (°C) Cc-CH4 CO2-CH4 Gr-CH4 Cc-Gr CO2-Gr CO2-Cc CO2-Dm Dm-Gr

0 90.6 77.7 63.4 27.2 14.3 ⫺13.0 2.8 11.5
10 85.5 73.7 59.3 26.2 14.4 ⫺11.8 - -
20 80.7 70.1 55.5 25.2 14.6 ⫺10.7 - -
25 78.5 68.3 53.7 24.8 14.6 ⫺10.1 - -
30 76.3 66.7 52.0 24.3 14.7 ⫺9.5 - -
40 72.2 63.6 48.8 23.4 14.8 ⫺8.6 - -
50 68.4 60.7 45.8 22.6 14.9 ⫺7.7 7.0 7.8
60 64.9 58.0 43.0 21.8 15.0 ⫺6.9 - -
70 61.6 55.5 40.5 21.1 15.0 ⫺6.1 - -
80 58.5 53.2 38.1 20.4 15.1 ⫺5.4 - -
90 55.6 51.0 35.9 19.7 15.1 ⫺4.7 - -
100 52.9 48.9 33.8 19.1 15.1 ⫺4.0 9.6 5.9
120 48.1 45.2 30.2 17.9 15.0 ⫺2.9 - -
140 43.8 41.9 26.9 16.9 14.9 ⫺1.9 - -
150 - - - - - - 10.9 3.9
160 40.0 38.9 24.1 15.9 14.8 ⫺1.1 - -
180 36.6 36.3 21.6 15.0 14.6 ⫺0.4 - -
200 33.6 33.9 19.5 14.2 14.4 0.2 11.5 2.9
220 31.0 31.7 17.5 13.4 14.2 0.8 - -
240 28.6 29.8 15.8 12.7 13.9 1.2 - -
250 - - - - - - 11.6 2.1
260 26.4 28.0 14.3 12.1 13.7 1.5 - -
280 24.5 26.3 13.0 11.5 13.4 1.8 - -
300 22.8 24.8 11.8 11.0 13.1 2.1 11.4 1.7
320 21.2 23.5 10.7 10.5 12.8 2.3 - -
340 19.8 22.2 9.7 10.0 12.5 2.4 - -
350 - - - - - - 11.0 1.3
360 18.5 21.0 8.9 9.6 12.2 2.6 - -
380 17.3 20.0 8.1 9.2 11.9 2.6 - -
400 16.3 19.0 7.4 8.8 11.5 2.7 10.5 1.1
450 14.0 16.8 6.0 8.0 10.8 2.8 9.9 0.9
500 12.2 14.9 4.9 7.3 10.1 2.8 9.3 0.8
550 10.7 13.4 4.0 6.7 9.4 2.7 8.7 0.7
600 9.5 12.1 3.3 6.2 8.8 2.6 8.2 0.6
650 8.5 10.9 2.7 5.8 8.2 2.4 7.7 0.6
700 7.7 10.0 2.3 5.4 7.7 2.3 7.2 0.5
750 - - - - - - 6.7 0.5
800 - - - - - - 6.3 0.5
850 - - - - - - 5.9 0.5
900 - - - - - - 5.5 0.5
1000 - - - - - - 4.4 0.4
Figure 11.36 Carbon isotope fractionation among various phases in the Ca-C-O-H sys-
tem, plotted as a function of T. Calcite-graphite, CO2-graphite, CO2-graphite, graphite-
CH4, and CO2-CH4 after Bottinga (1969a); diamond-graphite, CO2-diamond, and CO2-
calcite after Bottinga (1969b). Superimposed symbols denote experimental results on
CO2-calcite fractionation after Chacko et al. (1991) (solid boxes) and Romanek et al.
(1989). Reprinted from Bottinga (1969a), with kind permission from Elsevier Science Pub-
lishers B.V, Amsterdam, The Netherlands.
792 / Methods
Moreover, as we can see in figure 11.36, the fractionation factor between CO2
and graphite is practically constant below 200 °C (1000 ln ␣ 艐 14.5) and falls
progressively with increasing T. Also, the carbon fractionation factor between
CO2 and calcite has a maximum near 460 °C (1000 ln ␣ 艐 2.85) and a crossover
at 192 °C, confirmed by direct observations of natural occurrences in geothermal
areas by Craig (1963).
According to the calculations of Bottinga (1969b) for the C-O system, the
high-density form diamond should be preferentially enriched in 13C with respect
to the low-density form graphite (although the enrichment is quite limited; cf.
figure 11.36). Moreover, the CO2-diamond carbon fractionation curve exhibits a
broad maximum near 300 °C (1000 ln ␣ 艐 11).
Bottinga’s (1969b) calculations of the carbon fractionation factors between carbon

polymorphs and coexisting gaseous phases are carried out along the guidelines outlined
in section 11.6.1. The phonon frequency distribution in the solid phases is, however, ob-
tained by application of lattice dynamics to crystalline polymorphs. After an opportune
selection of static potentials describing the interaction between neighboring carbon atoms
in the structure, as a function of interatomic distances and angles, the equations of motion
are obtained for all atoms in the unit cell, and the normal modes of vibration of all r
particles in the crystal composed of N unit cells:
ω max
∫ g( ω ) dω = 3 Nr (11.169)
0
are calculated by solving the 3r ⫻ 3r dynamical matrix obtained by general solution of

wave vectors. The vibrational partition function is then obtained by applying
kT gn
X n+1
 exp( − X 2 ) 
ln Qvib = ∑ × ∫ ln   dX ,
 1 − exp( − X ) 
(11.170)
hc n ω n +1 − ω n Xn
where the various gn (i.e., the eigenvalues of the dynamic matrix) are the numbers of modes
in the vibrational frequency interval ␻n⫹1 ⫺ ␻n (see sections 3.1, 3.3, and 11.6.1 for the
significance of the remaining symbols).
Vibrational spectra generated by Bottinga (1969b) for graphite and diamond with this
procedure are shown in figure 11.37. The spectrum for graphite is based on the interionic
potential model of Yoshimori and Kitano (1956) and adopts the force constants of Young
and Koppel (1965) (assumed to be identical for 12C and 13C). The histogram of figure
11.37A represents 1.79 ⫻ 106 frequencies in the irreducible segment ( 24 1
) of the Brillouin
zone, with a frequency interval ␻n⫹1 ⫺ ␻n of 1 cm⫺1. Figure 11.37B shows the normalized
phonon spectra for diamond, based on the interionic potential model of Aggarwal (1967)
and the dynamic matrix equations of Herman (1959) ( 48 1
of the Brillouin zone; 3.46 ⫻ 106
frequencies). The corresponding Debye spectrum for ␪D ⫽ 1860 K is also shown for com-
parative purposes. The jagged appearance of the phonon spectra is partly attributable to
incomplete root sampling of phonon frequencies and to the narrow bandwidth adopted
in the two histograms (Bottinga, 1969b).
Figure 11.37 Normalized phonon spectra for 12C and 13C in graphite (A) and diamond
(B). Reprinted from Bottinga (1969b), with kind permission from Elsevier Science Pub-
lishers B.V, Amsterdam, The Netherlands.
Concerning calcium carbonate polymorphs calcite and aragonite, the experi-

ments of Romanek et al. (1992) at low T (10 to 40 °C) seem to indicate that the
enrichment factor of 13C in the solid phase for CaCO3-CO2 partitioning is reduced
by about two units in the aragonite polymorph with respect to the calcite counter-
part, T being equal (e.g., at T ⫽ 10 °C, 1000 ln ␣ 艐 13 for calcite-CO2 and 11 for
794 / Methods
aragonite-CO2 ; at T ⫽ 40 °C, 1000 ln ␣ 艐 9 for calcite-CO2 and 7 for

aragonite-CO2 ).
Concerning isotopic equilibria between solute carbonates and gaseous CO2 ,
Deines et al. (1974), on the basis of all the existing data, derived linear equations
relating the fractionation factor to temperature:
6.3 × 10 3
1000 ln α H * = − 0.91 (11.171.1)
2 CO 3 − CO2 ( g ) T2
1.099 × 10 6
1000 ln α HCO− − CO = − 4.54 (11.171.2 )
3 ( )
2 g T2
8.7 × 10 5
1000 ln α CO2 − − CO = − 3.4 . (11.171.3)
3 ( )
2 g T2
Equations 11.171.1 to 11.171.3 are, however, of limited practical application be-

cause they demand precise knowledge of the state of speciation of carbonates in
aqueous solution during solid phase condensation (or late exchanges). The fact
that different carbonate solute species distinctly fractionate 13C is masterfully out-
lined by the experiments of Romanek et al. (1992), which indicate a marked con-
trol by solution pH of the fractionation between total dissolved inorganic carbon
(DIC) and gaseous CO2 (figure 11.38).
With decreasing pH, the proportion of aqueous CO2 in solution (i.e.,
H2CO* 3 if one does not distinguish between simple planar O⫽C⫽O and trigonal
solvated form; cf. section 8.10) increases drastically, causing a marked decrease
in 13␦DIC.
11.8.4 Sulfur
Sulfur has four stable isotopes: 32S (0.9502), 33S (0.0075), 34S (0.0421), and 36S
(0.0002). The 34S/ 32S sulfur isotope fractionation in sulfide minerals and solute
sulfur species is a sensitive function of temperature (Sakai, 1968; Ohmoto, 1972).
Detailed vibrational calculations of the partition function ratio and correspond-
ing isotopic fractionation factors for solute sulfur species were first attempted by
Sakai (1968). The results of his calculations are shown in figure 11.39 in terms of
the ⌬34S enrichment factor:
∆ AB ≅ 1000 ln a AB ≅ δ 34 S A − δ 34 S B (11.172)
with H2S as reference phase B.

Figure 11.39 makes evident the marked role played by the oxidation state of
Figure 11.38 Relationship between average 13C enrichment factor and pH in solute car-
bonates–gaseous CO2 equilibria (⌬DIC⫺CO2 ⫽ 13␦DIC⫺CO2 ). Reprinted from C. S. Romanek,
E. L. Grossman, and J. W. Morse, Geochimica et Cosmochimica Acta, 56, 419–430, copy-
the complex (i.e., strong positive relative enrichment factors for sulfate ion
SO2⫺ 2⫺
4 with respect to H2S and negative enrichment factors for sulfide ion S ; see
later in this section). Moreover, it is also evident that a simple inverse relation
with the squared temperature holds.
Because the fractionation effect between condensed sulfates and SO2⫺ 4 ions is
negligible, it can be assumed that all sulfate complexes in solution (e.g.,
HSO⫺ ⫺ ⫺
4 , KSO4 , NaSO4 , . . .) have similar ⌬ S enrichment factors, T being equal
34
(Ohmoto, 1972).
Sulfur isotope relative fractionation factors between various minerals and py-
rite are shown in figure 11.40. As we can see, the isotopic compositions of sulfates
reflect the relative fractionation effects induced by the SO2⫺ 4 groups (compare
figures 11.40 and 11.39).
Ohmoto and Rye (1979), through a critical examination of all the existing
data, proposed a set of fractionation equations relating the fractionation factors
between important ore-forming sulfides and H2S(aq) to T of formation. The pro-
posed equations obey a simple linear dependence over the inverse squared tem-
perature (K), passing through the origin:
A
10 3 ln α i − H2 S(aq) = . (11.173)
T2
The temperature of equilibrium can be readily derived by applying

796 / Methods
Figure 11.39 ⌬34S enrichment factors for solute sulfur species based on the data of Sakai
(1968). Reproduced with modifications from Ohmoto, Economic Geology, 1972, Vol. 67,
p. 533.
A
T (K ) = . (11.174)
10 ln α i − H2 S aq
3
( )
The slope coefficients A for the various sulfides are listed in table 11.16.
Coupling thermometric equations such as equation 11.174 for different min-
erals, one obtains solid-solid fractionation expressions of the type
TAB ( K ) =
(AA − AB ).
(11.175)
10 ln α AB
3
For example, for the pyrite-galena couple, we have
0.40 + 0.63
Tpyrite − galena ( K ) = . (11.176)
10 3 ln α AB
Uncertainties associated with T estimates based on equation 11.176 range from

⫾20 to 25 K (pyrite-galena) to ⫾40 to 55 K (pyrite-pyrrhotite).
Figure 11.40 ⌬34S relative enrichment factors for various sulfur minerals with respect to
pyrite. Reproduced with modifications from Rye and Ohmoto, Economic Geology, 1974,
Vol. 69, p. 828.
Table 11.16 Slope coefficients relating sulfur isotopic

fractionation factors between sulfides and aqueous H2S
to temperature (K) (after Ohmoto and Rye, 1979).
Phase Composition Slope Coefficient A

Molybdenite MoS2 0.45
Pyrite FeS2 0.40
Sphalerite ZnS 0.10
Pyrrhotite FeS 0.10
Chalcopyrite CuFeS2 ⫺0.05
Covellite CuS ⫺0.40
Galena PbS ⫺0.63
Chalcosite Cu2S ⫺0.75
Argentite Ag2S ⫺0.80
798 / Methods
Figure 11.41 Variation of ␦ 34S of sulfate, H2 S(aq), and sulfide minerals in response to
H2S/SO42⫺ variations in the hydrothermal solution at T ⫽ 200 °C and ␦ 34S∑S ⫽ 0‰. Repro-
duced with modifications from Rye and Ohmoto, Economic Geology, 1974, Vol. 69, p. 828.
The large isotopic fractionations observed between oxidized and reduced

forms of sulfur compounds and aqueous complexes require accurate appraisal of
the effective cogeneticity of sulfur minerals utilized as geothermometric couples,
and of their equilibrium condition, to avoid erroneous deductions. In fact, besides
temperature, the isotopic composition of sulfur minerals is also affected by the
bulk isotopic composition of the sulfur in the system ( ␦ 34S⌺S ) (which is controlled
by the source of sulfur) and by the proportion of oxidized and reduced sulfur
species in solution (which is controlled by Eh and pH conditions and total sulfur
in the system; see section 8.15).
Figure 11.41 shows, for example, the effect of oxidation state on the isotopic
compositions of sulfur minerals at T ⫽ 200 °C assuming ␦ 34S⌺S ⫽ 0‰. At T ⫽
200 °C, the sulfur isotopic fractionation between SO2⫺ 4 and H2S is about 32‰. If
the sulfur in the ore fluid were distributed between H2S(aq) and SO2⫺4 and the H2S/
2⫺
SO4 ratio were changed from 1:9 to 9:1, in response to a decrease in f O2 sulfide
minerals such as sphalerite and galena eventually precipitating from such fluids
would attain the different ␦ 34S values shown in figure 11.41. At high f O2, the
dominant sulfate ion concentrates 34S, and H2S is relatively enriched in 32S, and
hence the precipitating sulfides have low ␦ 34S values. The opposite is true at low
f O2, where the relative predominance of sulfides causes the precipitating ores to
be relatively richer in 34S. A similar effect is induced by pH: because, with T held
constant, HS⫺ and S2⫺ ions are depleted in 34S with respect to H2S, and because
the aqueous speciation of sulfide ions is controlled by
H2 S (aq) ⇔ H + + HS − (11.177 )
Figure 11.42 ␦ 34S contours superimposed on the stability fields of Fe-S-O minerals and
barite at T ⫽ 250 °C and ␦ 34S∑S ⫽ 0‰ Solid lines and their dashed extensions: ␦ 34S con-
tours for pyrite (values in brackets) and barite (values in parentheses). Dashed lines: Fe-
S-O mineral boundaries for a 10⫺1 molal concentration of sulfur in the system. Small
dashed lines: Fe-S-O mineral boundaries for a 10⫺3 molal concentration of sulfur in the
system. Dash-and-dot line: solubility product of barite. Reproduced from Ohmoto, Eco-
nomic Geology, 1972, Vol. 72, p. 555.
HS − ⇔ H + + S 2 − , (11.178)
the increase in pH favors the formation of the relatively 34S-depleted ions HS⫺
and S2⫺, and hence the precipitating sulfides attain higher ␦34S values, T and Eh
being equal.
The combined effects of f O2 and pH variation on the sulfur isotopic com-
positions of ore-forming minerals are shown in figure 11.42 for a temperature of
250 °C and ␦ 34S⌺S ⫽ 0‰. The dashed lines delineate the stability limits of con-
densed phases in the Fe-S-O system for two different total sulfur concentrations
(10⫺1 and 10⫺2 molal). On these lines are superimposed ␦ 34S contours (solid lines
800 / Methods
and their dashed extensions) for coexisting pyrite (values in brackets) and barite
(values in parentheses). ␦ 34S for pyrite can range from ⫹5 to ⫺27 per mil and
for barite from 0 to ⫹32 per mil within geologically plausible Eh-pH ranges. In
acidic-reducing conditions at the limit of the pyrite stability field, the ␦ 34S value
of sulfide is similar to ␦ 34S⌺S and rather insensitive to Eh-pH changes. Neverthe-
less, at high Eh-pH conditions, near the pyrite-hematite and pyrite-magnetite sta-
bility limits, the ␦ 34S value of sulfide is largely variable in response to limited Eh-
pH changes and much dissimilar from ␦ 34S⌺S as a result of the presence of sulfate
ion in solution, which fractionates 34S.
Figure 11.42 elucidates in a rather clear fashion the marked control exerted
by the chemistry of the fluid on the sulfur isotopic compositions of ore-forming
minerals. At the T and ␦ 34S⌺S conditions delineated in the figure, an increase in
f O2 or in pH of one log unit can cause decreases in ␦ 34S values of sulfide minerals
by as much as 20‰.
Although these chemical effects are important in deciphering the genesis of
ore minerals, it must be emphasized that the differences in the ␦ 34S values among
coexisting condensed phases (hence the fractionation factor) at equilibrium are
constant in each case because they depend only on T.
APPENDIX ONE
Constants, Units, and

Conversion Factors
The International Union for Pure and Applied Chemistry (IUPAC) recommends
the adoption in scientific publications of the SI system notation. Tables A1.1 and
A1.2 present the base units and some derived units.
In geochemical literature, one often encounters non-SI units (e.g., bar, atmo-
sphere, calorie, angstrom, electron-volt, etc.), which can be converted to SI units
by the use of simple multiplicative factors, as shown in table A1.3.
Table A1.1 SI Base Units
Magnitude Unit Symbol
Length meter m
Mass kilogram kg
Time second s
Electric current ampere A
Thermodynamic temperature kelvin K
Amount of substance mole mol, mole
Intensity of light candle cd
Table A1.2 SI Derived Units
Magnitude Unit Symbol SI Derivation
Energy joule J m ⭈ kg ⭈s ⫺ 2
2
Force newton N m ⭈k g⭈ s ⫺ 2
Power watt W J⭈s⫺1
Pressure pascal Pa m ⫺1 ⭈ kg ⭈s ⫺ 2
Electric charge coulomb C A⭈s
Electric potential volt V m 2 ⭈ kg ⭈s ⫺ 3 ⭈A ⫺ 1
Capacitance farad F m ⫺2 ⭈ kg ⫺ 1 ⭈s 4 ⭈A 2
Resistance ohm ⍀ m 2 ⭈ kg ⭈s ⫺ 3 ⭈A ⫺ 2
Magnetic flux weber Wb m 2 ⭈ kg ⭈s ⫺ 2 ⭈A ⫺ 1
Magnetic flux density tesla T k g⭈ s ⫺ 2 ⭈A ⫺ 1
Frequency hertz Hz s ⫺1
/ 801
802 / Appendix 1
Table A1.3 Non-SI units and their factors for conversion to the SI system
Magnitude Unit Symbol SI Conversion
Length angstrom Å 10⫺10 m

Mass ton t 10⫺3 kg
Pressure bar bar 105 Pa
Pressure atmosphere atm 101,325 Pa
Energy calorie cal 4.184 J
Energy electron-volt eV 1.602189 ⫻ 10⫺19 J
Kinematic viscosity stokes st 10⫺4 m2s⫺1
Dynamic viscosity poise p 10⫺1 Pa⭈s
Of particular practical interest are the general energy conversion factors

(Lewis and Randall, 1970) presented in table A1.4.
As we have already seen, the “mole” is the mass unit in the SI system. The
molar concentrations of components in geochemical systems assume different
significances as a function of the adopted reference unit:
Mole ⫽ amount of substance, expressed in weight (g), containing Avogadro’s

number of molecules of the substance of interest
Molality ⫽ moles of solute per kg of solvent
Molarity ⫽ moles of solute per 10⫺3 m3 of solution
Formality ⫽ moles of solute per kg of solution
Molar fraction ⫽ moles of component of interest divided by the sum of moles
of all components in the phase.
We recall that chemical analyses generally report the composition of a given

phase as “weight percent (%) of constituting oxides,” or as parts per mil in weight
(‰) or parts per million in weight (ppm) or parts per billion in weight (ppb). To
switch from weight percent to ppm, one simply multiplies the observed value by
104. Analogously, to switch from weight percent to ppb, the value must be
multiplied by 107.
Finally, table A1.5 lists the principal physical constants, and table A1.6 pres-
ents some energy equivalents of atom units.
Table A1.4 Energy conversion factors
cal⭈mole⫺1 joule⭈mole⫺1 cm3⭈atm⭈mole⫺1 cm⫺1 eV erg⭈molecule⫺1

1 cal⭈mole⫺1 1 4.1840 41.292 0.34974 4.3361 ⫻ 10⫺5 6.9465 ⫻ 10⫺17
1 joule⭈mole⫺1 0.23901 1 9.8692 0.08359 1.0364 ⫻ 10⫺5 1.6602 ⫻ 10⫺17
1 cm3⭈atm⭈mole⫺1 0.024218 0.10133 1 8.470 ⫻ 10⫺3 1.0501 ⫻ 10⫺6 1.6823 ⫻ 10⫺18
1 cm⫺1 2.8593 11.963 118.07 1 1.2398 ⫻ 10⫺4 1.0862 ⫻ 10⫺16
1 eV 23.062 96.493 9.523 ⫻ 105 8065.7 1 1.6020 ⫻ 10⫺12
1 erg⭈molecule⫺1 1.4396 ⫻ 1016 6.023 ⫻ 1016 5.944 ⫻ 1017 5.0348 ⫻ 1015 6.2422 ⫻ 1011 1
804 / Appendix 1
Table A1-5 Main physical constants
Constant Symbol SI Value Non-SI Value
Gas constant R 8.31451 J⭈K⫺1⭈mole⫺1 1.98726 cal⭈K⫺1⭈mole⫺1

Faraday’s constant F 96485.3 J⭈V⫺1 23062.3 cal⭈V⫺1
Avogadro’s number N0 6.022137 ⫻ 1023 mole⫺1 -
Boltzmann’s constant k 1.380658 ⫻ 10⫺23 J⭈K⫺1 1.380658 ⫻ 10⫺16 erg⭈K⫺1
Planck’s constant h 6.626075 ⫻ 10⫺34 J⭈s 6.626075 ⫻ 10⫺27 erg⭈s
Planck’s constant
(ប ⫽ h/2␲) ប 1.0545727 ⫻ 10⫺34 J⭈s 1.0545727 ⫻ 10⫺27 erg⭈s
Elementary charge e 1.6021773 ⫻ 10⫺19 J -
Atomic mass unit (amu) u 1.6605402 ⫻ 10⫺27 kg -
Electron rest mass me 5.4657990 ⫻ 10⫺4 u -
Proton rest mass mP 1.007276470 u -
Neutron rest mass mn 1.008664904 u -
Bohr’s radius a0 0.52917725 ⫻ 10⫺10 m -
Bohr’s magneton ␮B 9.274015 ⫻ 10⫺24 J⭈kg⫺1⭈s2⭈A -
Rydberg’s constant Ry 1.097373153 ⫻ 107 m⫺1 -
Vacuum permittivity ␧0 8.85418782 ⫻ 10⫺12 F⭈m⫺1 -
Speed of light in vacuum c 2.99792458 ⫻ 108 m⭈s⫺1 -
Table A1.6 Energy conversions of atom units
Atom Unit Energy Equivalent
Atomic mass unit (amu) 931.4943 MeV

Electron rest mass 0.5109991 MeV
Proton rest mass 938.2723 MeV
Neutron rest mass 939.5656 MeV
Temperature corresponding to 1 eV 1.160450 ⫻ 104 K
Photon wavelength associated to 1 eV 1.239852 ⫻ 10⫺6 m
APPENDIX TWO
Review of Mathematics
In the development of the various treated arguments, we adopted infinitesimal

calculation procedures, vectorial notations (gradient, Laplacian), and some
mathematical functions (Lagrange equations, Hamiltonians, error functions) that
should already be known to the reader but that perhaps would be useful to review
in a concise way.
A2.1 Exact Differentials and State Functions
Let F be a function of the independent variables:
F = F ( x1 , x 2 , K , x n ) (A 2.1)
and let f 1 , f 2 , . . . f n be functions of the single independent variables x1 , x2 , . . .

,xn , such that
f1 dx1 + f2 dx 2 + K + fn dx n = ∑ fi dx i . (A2.2)
i
Whenever the differential equation A2.2 can be equated with the differential of
function F:
 ∂F   ∂F   ∂F 
df =   dx1 +   dx 2 + L   dx n , (A2.3)
 ∂ x1  x  ∂ x2  x ,Kx  ∂xn 
2 ,K x n 1 n x1 , x2 , K
then dF is an exact differential and
 ∂F   ∂F   ∂F 
f1 =   ; f2 =   ; fn =   . (A2.4)
 ∂ x1  x  ∂x 2  x  ∂x n  x
2, K x n x
1 ,K n 1, x 2, K
/ 805
806 / Appendix 2
The state functions presented in section 2.4 are examples of exact differential
functions of the type exemplified by equation A2.3. Generally, reversible transfor-
mations in physical systems are described by equations of this type.
For two generic independent variables xg and xh , such that
 ∂F 
fg =   (A2.5)
 ∂ xg  x
1,K
and
 ∂F 
fh =   (A2.6)
 ∂ xh  x
1,K
we have
 ∂ fg  ∂ 2F
∂x  = (A2.7)
 hx ∂ x g∂ x h
1,K
and
 ∂ fh  ∂ 2F
 ∂x  = . (A2.8)
 g x ∂ xh ∂ xg
1, K
Applying Schwarz’s theorem on sequential differentiation of mixed derivatives:
∂ 2F ∂  ∂F  ∂  ∂F  ∂ 2F
=   =   = , (A2.9)
∂ x∂ y ∂ x  ∂ y  ∂ y  ∂ x  ∂ y∂ x
we obtain
 ∂ fg   ∂f 
 ∂x  = h . (A2.10)
 h  x ,K  ∂ xg  x ,K
1 1
Equation A2.10 represents a necessary and sufficient condition for equation A2.2
to be an exact differential and is valid for any couple of conjugate variables
( f g ,xg ) and ( f h ,xh ).
Review of Mathematics / 807
A2.2 Review of Implicit Functions
Let f ⫽ f (x,y,z) be a function of three variables, such that
( )
f x, y, z = 0. (A 2.11)
The function
(
z = z x, y ) (A 2.12 )
is said to be an “implicitly defined function” of equation A2.11, because it is the

solution in z of such an equation. Differentiating equation A2.11 produces
∂f ∂f ∂f
dx + dy + dz = 0 (A2.13)
∂x ∂y ∂z
and, from equation A2.13, keeping constant one of the three variables and assum-
ing the partial derivatives of f to be non-null, we obtain
 ∂ x  ∂ y  ∂f   ∂f 
  = 1   = −    (A2.14.1)
 ∂ y z  ∂ x z  ∂ y  ∂ x
 ∂ z  ∂ y  ∂f   ∂f 
  = 1   = −    (A2.14.2)
 ∂ y x  ∂ z x  ∂ y  ∂ z
 ∂z  ∂ x  ∂f   ∂f 
  = 1   = −   . (A2.14.3)
 ∂ x y  ∂z y  ∂ x  ∂ z
From equations A2.14.1 to A2.14.3, further relationships among partial deriva-

tives can be obtained, such as
 ∂ x  ∂ y  ∂ z
  ⋅   ⋅   = −1 (A2.15)
 ∂ y z  ∂ z x ∂x y
or
 ∂ x  ∂ z  ∂ z
−  =     . (A2.16)
 ∂ y z  ∂ y x  ∂ x y
808 / Appendix 2
Let us now consider the exact differential:
 ∂ x  ∂ x
dx =   dy +   dz (A2.17)
 ∂ y z  ∂ z y
and a further function of the two variables x and y, ␸(x,y), such that
 ∂ϕ   ∂ϕ 
dϕ =   dx +   dy. (A2.18)
 ∂ x y  ∂ y x
If we let x and y vary while holding z constant, we get, from the partial derivative
in dx of equation A2.18,
 ∂ϕ   ∂ϕ   ∂ϕ   ∂ y
  =  +    . (A2.19)
 ∂ x z  ∂ x y  ∂ y  x  ∂ x z
Equation A2.19 is one of the possible ways to switch from a couple of variables
to another conjugate couple. Most transforms adopted in the thermodynamics
of systems with two independent variables are based on equation A2.19 and its
modifications (Legendre transforms, cf. section 2.4).
A2.3 Integration on Exact Differentials
For a function F:
(
F = F x1 , x 2 , L , x n ) (A 2.20)
such that dF is its exact differential:
dF = f1 dx1 + f2 dx 2 + L + fn dx n , (A 2.21)
where
 ∂F 
f1 =   ,
 ∂ x1  x
2 ,K x n
etc. (cf. eq. A2.4), we have
∫ ( f1 dx1 + f2 dx 2 + L + fn dx n = 0 ) (A 2.22.2)
—i.e., the cyclic integral of such a function is zero. Moreover, the value of a
generic contour integral between points A and B in a field of existence described
by variables x1 , x2 , . . . ,xn does not depend on the path followed in integration.
That is, if
(
F = F x1 , x 2 ) (A 2.23)
defined in a field x1, x2 is such that
 ∂F   ∂F 
dF =   dx1 +   dx 2 , (A2.24)
 ∂ x1  x  ∂ x2  x
2 1
we have
B B
 ∂F   ∂F 
∫ dF = ∫   dx1 + 
 ∂ x1  x
 dx 2
 ∂ x2  x
A A 2 1
(A2.25)
B
= ∫ f1 dx1 + f 2 dx 2 = F( B ) − F( A)
A
independent of the followed integration path.
A2.4 Operations on Vectors
A vector is defined by length and direction. The length V of a generic vector V is

given by its modulus:
V = V. (A 2.26)
With V being the modulus of vector V, its unit vector U is such that
V = V U. (A 2.27)
The inner product a ⭈ b of vectors a and b is the inner quantity
a ⋅ b = ab cos α ab , (A 2.28)
where a and b are the moduli of the respective vectors and ␣ab is the minimum
angle between the representatives of the unit vectors.
810 / Appendix 2
The vectorial product a ⫻ b of vectors a and b is a vector c whose modulus

c is
c = ab sin α ab . (A 2.29)
The direction of c is orthogonal to a and b and depends on the rotation required

for a to overlap on b (the “right-hand rule”).
By decomposing the vector into its directional components, it can be shown
that the modulus of the vectorial product of two vectors a and b is given by
the determinant
Ux Uy Uz
a × b = ax ay az (A2.30)
bx by bz
and the mixed product of three vectors a, b, and c is given by the determinant
ax ay az
a × b ⋅ c = bx by bz . (A2.31)
cx cy cz
The vectorial operator “gradient” (with symbol nabla ∇) allows the passage
from scalar to vectorial fields. For scalar ⌽ the vector ∇⌽ (gradient of ⌽) is
given by
∂Φ ∂Φ ∂Φ
∇Φ = , , . (A2.32)
∂x ∂y ∂z
In section 1.19, for example, ∇i⌽e, j was the vectorial component of the electrical
potential of the f th dipole ⌽e, j directed toward i.
The differential operator “divergence” allows the passage from vectorial to
scalar fields. For a vector ␸,
∂ϕ ∂ϕ ∂ϕ
div ϕ = ∇ϕ = , , . (A2.33)
∂x ∂y ∂z
The “curl” of a vector function (or “rotation” with symbol “rot”) is a vector
that is formally the cross product of the operator and the vector. For a vector V,
Ux Uy Uz
∂ ∂ ∂
rot V = ∇ × V = . (A2.34)
∂x ∂y ∂z
vx vy vz
The Laplacian operator ∇2 performs the differential operation
∂2 ∂2 ∂2
∇2 = + + (A2.35)
∂ x 2 ∂ y 2 ∂ z2
and is applied to either scalars or vectors.

In sections 1.1.2, 1.19, and 3.1, we have seen applications of the Laplacian
operator to the vectorial field.
A2.5 Lagrange Equations, Hamiltonians
The bases of classical mechanics are established by Newton’s laws. In classical

mechanics, the dynamic behavior of a set of particles can be determined, on the
basis of Newton’s laws, if one knows the initial Cartesian coordinates of each
particle (Xl ) at time t 0 and their velocities (dXl /dt).
In quantum mechanics, the spatial variables are constituted by “generalized
coordinates” (qi ), which replace the individual Cartesian coordinates of all single
particles in the set. The Lagrangian equations of motion are the Newtonian equa-
tions transposed to the generalized coordinate system.
Let us consider a particle of mass m located at a point p and subjected to a
force F given by the gradient of potential ⌽ in p. Denoting Xl, where l ⫽ 1, 2 and
3, the Cartesian coordinates in p, and denoting qj the corresponding Lagrangian
coordinates, the two systems of spatial coordinates are related by:
∂Xl
Xl = ∑ ∂qj
dq j . (A2.36)
j
For an infinitesimal displacement dq we have
∂p
dp = ∑ ∂ qj
dq j . (A2.37)
j
Velocity V, which is the derivative of displacement on time, is expressed as

812 / Appendix 2
dp ∂p ∂qj ∂p
V=
dt
= ∑ ⋅
∂qj ∂t
+
∂t
. (A2.38)
j
If constraints are fixed, the term ∂p/ ∂t is zero and we have
∂V ∂p ∂q j
= , where ∂ q˙ j = . (A2.39)
∂ q˙ j ∂q j ∂t
Because acceleration a is the partial derivative of velocity on time, through

some substitutions, and posing
1
T= mV 2 , (A2.40)
2
we obtain
 d ∂T ∂T
ma dp = ∑  dt ∂ q˙ − ∂ q  dq j . (A2.41)
j  j j 
Because to a displacement dp corresponds a potential difference ⫺d⌽ according

to
F dp = − dΦ (A2.42)
we also have
∂Φ
F dp = − dq j . (A2.43)
∂qj
Let us consider now Newton’s second law for conservative systems*:
∑ ( Fj )
− ma j dp j = 0
(A2.44)
j
Substituting equations A2.41 and A2.43 into A2.44, we get
* If the work exerted by a force field on a particle moving from point A to point B is indepen-
dent of the path followed by the particle between the two points, the force field is said to be “conser-
vative.”
 ∂Φ d ∂T ∂T
∑  − ∂ q − dt ∂ q + ∂ q  dq j = 0, (A2.45)
j  j j j 
from which it follows that
d ∂T ∂T ∂Φ
− =− . (A2.46)
dt ∂ q˙ j ∂ q j ∂qj
Posing
L = T − Φ, (A 2.47)
equation A2.46 takes the form
d ∂L ∂L
− = 0. (A2.48)
dt ∂ q j ∂ q j
˙
This sort of equation, known as a “Lagrange equation of motion,” is generally

valid for systems with unconstrained degrees of freedom. If a system has n degrees
of freedom, n second-order Lagrange equations will exist (functions of q̇j, qj , and
time: quadratic in q̇j ).
We have seen in section 3.1 the application of the concept of the harmonic
oscillator in the interpretation of vibrational properties of crystals. For a unidi-
mensional harmonic oscillator, there is a single degree of freedom:
dx d 2x
q = x; x˙ = = V; x˙˙ = =a (A2.49.1)
dt dt 2
1 KF 2
T= mx˙ 2 ; Φ= x (A2.49.2)
2 2
1 K
L= mx˙ 2 − F x 2 . (A2.49.3)
2 2
The corresponding Lagrange equation is
d ∂L ∂L
− =0 (A2.50.1)
dt ∂ x ∂ x
d
mx˙ + KF x = 0 (A2.50.2)
dt
814 / Appendix 2
mx˙˙ + KF x = m a + KF x = 0. (A 2.50.3)
The Hamiltonian operator is
∂L
H= ∑ ∂ q˙ j
dq˙ j − L. (A2.51)
j
From the definition
L = T − Φ, (A 2.52)
the Hamiltonian becomes
∂T
H= ∑ ∂ q˙ j
dq˙ j − T + Φ, (A2.53)
j
but, because
∂T
∑ ∂ q˙ j
dq˙ j = 2 T , (A2.54)
j
we obtain
H = T + Φ. (A 2.55)
As we have seen in sections 1.17 and 3.1, the quantum mechanics Hamilto-
nian in the Schrödinger equation has the form
h2
H=− ∇ 2 + Φ. (A2.56)
8π 2 m
In the unidimensional form, the Laplacian operator ∇ 2 is replaced by the simple

second derivative in x— i.e. ( ∂ 2␺(x) / ∂x2 ).
A2.6 Error Function
The error function erf x and its complementary function erfc x appear in the
solution of differential equations describing diffusive processes (see, for instance,
section 11.6.3). They are defined by the integrals
x
erf x =
2
∫ ( )
exp − y 2 dy (A2.57)
π 0
and
∞
erfc x =
2
∫ ( )
exp − y 2 dy. (A2.58)
π x
The limits of error function erf x are
( )
erf −∞ = −1; erf 0 = 0 ; erf ∞ = 1 (A 2.59)
Because
erfc x = 1 − erf x, (A 2.60)
the limits of erfc x are
( )
erfc −∞ = 2; erfc 0 = 1; erfc ∞ = 0. (A 2.61)
Values of functions erf x and erfc x are tabulated (cf. Gautschi, 1964) but can
also be rapidly obtained by the polynomials
( )
−8
erfc x = 1 + a 1 x + a 2 x 2 + a 3 x 3 + a 4 x 4 + a 5 x 5 (A 2.62)
(Gautschi, 1964), where a1 ⫽ 0.14112821, a2 ⫽ 0.08864027, a3 ⫽ 0.002743349,

a4 ⫽ ⫺0.00039446, a5 ⫽ 0.00328975 for 0 ⱕ x ⱕ 3, and
erfc x =
( )
exp − x 2  1 1 3 15 105 945 10, 365 
 − + 2 5 − 3 7 + 4 9 − 5 11 + 6 13 
π  x 2x 3
2 x 2 x 2 x 2 x 2 x 
(A 2.63)
for x ⱖ 3. The function erf x is then obtained by applying equation A2.60.
A2.7 York’s Algorithm for Isochrons
The least-squares treatment of the fitting of an isochron must account for the fact
that the error in the y-coordinate of each point is correlated with the error in the
x-coordinate.
With each point in the isochron being subjected to the requirement
yi = a + bx i i = 1, K , n, (A 2.64)
816 / Appendix 2
and denoting the observations as Xi and Yi, the adjusted values of these observa-
tions as xi and yi , the weights of the various observations as ␻(Xi ) and ␻(Yi ), and
the correlations between the x and y errors as ri , the best slope (i.e., “age”) can
be found by means of the equation
 U bVi rV 
∑ Zi2Vi  ω Yi + − i i

i  ( )
i ω Xi ( ) αi 

b= , (A2.65)
 U bVi brU 
∑ Z i2U i  i + − i i 
i  ω Yi( )ω ( X i ) αi 
where
Zi =
( ) ( )
ω Xi ω Yi
b ω (Y ) + ω (X ) − 2 br α
2 (A2.66)
i i i i
U i = Xi − X (A 2.67)
Vi = Yi − Y (A 2.68)
∑ Zi X i
X = i
(A2.69)
∑ Zi
i
∑ ZiYi
Y = i
(A2.70)
∑ Zi
i
( ) ( )
α i = ω Xi ω Yi . (A2.71)
Because Ui , Vi, and Zi contain b, the solution of the system A2.65 →A2.71 is
obtained by an iterative procedure, inserting an approximate value of b into these
terms and calculating a new b from equation A2.65. The best intercept is found
by use of the equation
a = Y − bX . (A 2.72)
REFERENCES
Ackermann T. and Schreiner F. (1958). Molwärmen und Entropien einiger Fettsäuren und
ihrer Anionen in wässriger Lösung. Zeits. Elektroch., 62:1143–1151.
Acree W. E. Jr. (1984). Thermodynamic Properties of Nonelectrolyte Solutions. New York: Aca-
demic Press.
Aggarwal K. G. (1967). Lattice dynamics of diamond. Proc. Phys. Soc., 91:381–389.
Ahrens L. H. (1952). The use of ionization potentials, 1: Ionic radii of the elements. Geochim.
Cosmochim. Acta, 2:155–169.
Ahrland S. (1973). Thermodynamics of the stepwise formation of metal-ion complexes in aque-
ous solution. Struc. Bonding, 15:167–188.
Ahrland S. (1975). Metal complexes present in sea water. In The Nature of Sea Water, E. D.
Goldberg, ed. Dahlem Konferenzen.
Aikawa N., Kumazawa M., and Tokonami N. (1985). Temperature dependence of intersite
distribution of Mg and Fe in olivine and the associate change of lattice parameters.
Phys. Chem. Minerals, 12:1–8.
Akamatsu T., Kumazawa M., Aikawa N., and Takei H. (1993). Pressure effect on the divalent
cation distribution in nonideal solid solution of forsterite and fayalite. Phys. Chem. Min-
erals, 19:431–444.
Akimoto S., Komada E., and Kushiro I. (1967). Effect of pressure on the melting of olivine
and spinel polymorph of Fe2SiO4. J. Geophys. Res., 72:679–689.
Akimoto S., Matsui S., and Syono S. (1976). High-pressure crystal chemistry of orthosilicates
and the formation of the mantle transition zone. In The Physics and Chemistry of Miner-
als and Rocks, R. J. G. Strens, ed. New York: John Wiley.
Albarede F. (1976). Some trace element relationships among liquid and solid phases in the
course of fractional crystallization of magmas. Geochim. Cosmochim. Acta, 40:667–673.
Albarede F. and Bottinga Y. (1972). Kinetic disequilibrium in trace element partitioning be-
tween phenocrysts and host lava. Geochim. Cosmochim. Acta, 36:141–156.
Allègre C. J., Treuil M., Minster J. F., Minster B., and Albarede F. (1977). Systematic use of
trace element in igneous process, Part I: Fractional crystallization processes in volcanic
suites. Contrib. Mineral. Petrol., 60:57–75.
Allred G. C. and Woolley E. M. (1981). Heat capacities of aqueous acetic acid, sodium acetate,
ammonia, and ammonium chloride at 283.15 K, 298.15 K, and 313.15 K: ⌬C 0P for ion-
ization of acetic acid and for dissociation of ammonium ion. J. Soln. Chem., 14:549–560.
Althaus E. (1967). The triple point andalusite-sillimanite-kyanite. Contrib. Mineral. Petrol.,
13:31–50.
Alvarez L. W. (1938). The capture of orbital electrons by nuclei. Phys. Rev., 54:486–497.
Anderson A. T. and Gottfried D. (1971). Contrasting behaviour of P, Ti and Nb in a differenti-
ated high-alumina olivine-tholeite and a calc-alkaline suite. Geol. Soc. Amer. Bull.,
82:1929–1942.
Anderson A. T. and Greenland L. P. (1969). Phosphorous fractionation diagram as a quantita-
tive indicator of crystallization differentiation of basaltic liquid. Geochim. Cosmochim.
Acta, 33:493–505.
/ 817
818 / References
Anderson O. L. (1972). Patterns in elastic constants of minerals important to geophysics. In

Nature of the Solid Earth, E. C. Robinson, ed. New York: McGraw-Hill.
Anderson O. L. (1995). Equations of State of Solids for Geophysics and Ceramic Science. New
York: Oxford University Press.
Anderson T. F. and Kasper R. B. (1975). Oxygen self-diffusion in albite under hydrothermal
conditions. Trans. Amer. Geophys. Union. 56:459.
Ando K., Kurokawa H., Oishi Y., and Takei H. (1981). Self-diffusion coefficient of oxygen in
single-crystal forsterite. J. Amer. Ceram. Soc., 64:2.
Angell C. A. (1982). Supercooled water. Amer. Rev. Phys. Chem., 34:593–630.
Annersten H., Adetnuji J., and Filippidis A. (1984). Cation ordering in Fe-Mn silicate olivines.
Amer. Mineral., 69:1110–1115.
Anovitz L. M. and Essene E. J. (1982). Phase relations in the system CaCO3-MgCO3-FeCO3.
Trans. Amer. Geophys. Union, 63:464.
Anovitz L. M., Essene E. J., Metz G. W., Bohlen S. R., Westrum E. F. Jr., and Hemingway
B. S. (1993). Heat capacity and phase equilibria of almandine, Fe3Al2Si3O12. Geochim.
Cosmochim. Acta, 57: 4191–4204.
Ardell A. J., Christie J. M., and Tullis J. A. (1973). Dislocation substructures in experimentally
deformed quartzites. Cryst. Lattice Defects, 4:275–285.
Armbruster T. and Geiger C. A. (1993). Andradite crystal chemistry: Dynamic X-site disorder
and structural strain in silicate garnets. Eur. J. Min., 5:59–71.
Armbruster T., Geiger C. A., and Lager G. A. (1992). Single-crystal X-ray structure study of
synthetic pyrope almandine garnets at 100 and 293 K. Amer. Mineral., 77:512–521.
Atkins P. W. (1978). Physical Chemistry. Oxford: Oxford University Press.
Baas Becking L. G. M., Kaplan I. R., and Moore O. (1960). Limits of the natural environments
in terms of pH and oxidation-reduction potentials. J. Geol., 68:243.
Babuska V., Fiala J., Kumazawa M., and Ohno I. (1978). Elastic Properties of Garnet Solid-
Solution series. Phys. Earth Planet. Int., 16:157–176.
Bachinski S. W. and Müller G. (1971). Experimental determination of microcline-low albite
solvus. J. Petrol., 12:329–356.
Bader R. F. W. and Jones G. A. (1963). The electron density distribution in hydride molecules,
I: The water molecule. Can. J. Chem., 41:586–606.
Baerlocher Ch., Hepp A., and Meier W. M. (1977). DLS—76. A program for Simulation of
Crystal Structures by Geometry Refinement. Zurich: Institut of Crystallography and
Petrography ETH.
Baes C. F. Jr. and Mesmer R. E. (1981). The thermodynamics of cation hydrolysis. Amer. Jour.
Sci., 281:935–962.
Bailey A. (1971). Comparison of low-temperature with high-temperature diffusion of sodium
in albite. Geochim. Cosmochim. Acta, 35:1073–1081.
Bailey S. W. (1984a). Classification and structures of the micas. In Reviews in Mineralogy, vol.
13, P. H. Ribbe (series ed.), Mineralogical Society of America.
Bailey S. W. (1984b). Crystal chemistry of true micas. In Reviews in Mineralogy, vol. 13, P. H.
Ribbe (series ed.), Mineralogical Society of America.
Bales R. C. and Morgan J. J. (1985). Dissolution kinetics of chrysotile at pH 7 to 10. Geochim.
Ballhausen C. J. (1954). Studies of absorption spectra, II: Theory of copper (II)—spectra.
Danske Videnskab. Selskab. Mat. Fys. Medd., 29:4–18.
Banno S. (1970). Classification of eclogites in terms of physical conditions of their origin. Phys.
Earth Planet Int., 3:405–421.
Barberi F., Ferrara G., Santacroce R., Treuil M., and Varet J. (1975). A transitional basalt-
pantellerite sequence of crystallization: The Boina Centre (Afar Rift, Ethiopia). J. Pet-
rol., 16:22–56.
Barin I. and Knacke O. (1973). Thermochemical Properties of Inorganic Substances. Berlin-
Heidelberg-New York: Springer-Verlag.
References / 819
Barin I., Knacke O., and Kubaschewsky O. (1977). Thermochemical Properties of Inorganic
Substances: Supplement. Berlin-Heidelberg-New York: Springer-Verlag.
Barnes V. E. (1930). Changes in hornblend at about 800 °C. Amer. Mineral., 5:393–417.
Barr L. W. and Liliard A. B. (1971). Defects in ionic crystals. In Physical Chemistry: An Ad-
vanced Treatise, vol. 10, W. Jost (series ed.).
Barron L. M. (1976). A comparison of two models of ternary excess free energy. Contrib. Min-
eral. Petrol., 57:71–81.
Bass, J. D. (1986). Elasticity of uvarovite and andradite garnets. J. Geophys. Res., 91:7505–7516.
Bass J. D. (1989). Elasticity of grossular and spessartite garnets by Brillouin spectroscopy.
J. Geophys. Res., 94:7621–7628.
Basso R. (1985). Crystal chemical and crystallographic properties of compounds with garnet
or hydrogarnet structure. Neues Jahrb. Miner. Monat., 3:108–114.
Battino R. and Clever H. L. (1966). The solubilities of gases in liquids. Chem. Rev., 66:395–463.
Baur W. H. (1970). Bond length variation and distorted coordination polyedra in inorganic
crystals. Trans. Amer. Cryst. Assoc., 6:129–155.
Baur W. H. (1978). Variation of mean Si-O bond lengths in silicon-oxygen tetrahedra. Acta
Cryst., B34:1754–1756.
Baur W. H. and Kahn A. A. (1971). Rutile-type compounds, IV: SiO2, GeO2, and a comparison
with other rutile-type compounds. Acta Cryst., B27:2133–2139.
Becker R. H. (1971). Carbon and oxygen isotope ratios in iron-formation and associated rocks
from the Hamersley Range of Western Australia and their implications. Ph.D. diss.,
University of Chicago.
Bell R. G., Jackson R. A., and Catlow C. R. A. (1992). Loewenstein’s rule in zeolite A: A com-
putational study. Zeolites, 12:870–871.
Belton G. R., Suito H., and Gaskell D. R. (1973). Free energies of mixing in the liquid iron-
cobalt orthosilicates at 1450 °C. Met. Trans., 4:2541–2547.
Berger G. W. and York D. (1981). Geothermometry from 40Ar/39Ar dating experiments. Geo-
chim. Cosmochim. Acta, 45:795–811.
Berlin R. and Henderson C. M. B. (1968). A re-interpretation of Sr and Ca fractionation trends
in plagioclases from basic rock. Earth Planet. Sci. Lett., 4:79–83.
Berman R. G. (1988). Internally consistent thermodynamic data for minerals in the system
Na2O-K2O-CaO-MgO-FeO-Fe2O3-Al2O3-SiO2-TiO2-H2O-CO2. J. Petrol., 29:445–522.
Berman R. G. (1990). Mixing properties of Ca-Mg-Fe-Mn garnets. Amer. Mineral., 75:
328–344.
Berman R. G. and Brown T. H. (1984). A thermodynamic model for multicomponent melts
with applications to the system CaO-Al2O3-SiO2. Geochim. Cosmochim. Acta, 48:
661–678.
Berman R. G. and Brown T. H. (1985). Heat capacity of minerals in the system Na2O-K2O-
CaO-MgO-FeO-Fe2O3-Al2O3-SiO2-TiO2-H2O-CO2: Representation, estimation, and
high temperature extrapolation. Contrib. Mineral. Petrol., 89:168–183.
Berman R. G. and Brown T. H. (1987). Development of models for multicomponent melts:
Analysis of synthetic systems. In Reviews in Mineralogy. vol. 17, P. H. Ribbe (series ed.),
Bernal J. D. (1964). The structure of liquids. Proc. Roy. Soc. London, A280:299–322.
Berner R. A. (1971). Principles of Chemical Sedimentology. New York: McGraw-Hill.
Berner R. A. (1980). Early Diagenesis: A Theoretical Approach. Princeton: Princeton Univer-
sity Press.
Berner R. A., Sjoberg E. L., and Schott J. (1980). Mechanisms of pyroxene and amphibole
weathering, I: Experimental studies. In Third International Symposium on Water-Rock
Interaction: Proceedings, Edmonton: Alberta Research Council.
Berthelot M. (1872). On the law which governs the distribution of a substance between two
solvents. Ann. Chim. Phys., 4th. ser., 26:408–417.
Bertorino G., Caboi R., Caredda A., Cidu R., Fanfani L., and Zuddas P. (1981). Caratteri
820 / References
idrogeochimici delle acque naturali della Sardegna meridionale, 3: Le acque della Mar-
milla e del Sarcidano. Rend. S.I.M.P, 37:951–966.
Bertrand G. L., Acree W. E. Jr., and Burchfield T. (1983). Thermodynamical excess properties
of multicomponent systems: Representation and estimation from binary mixing data.
J. Solution. Chem., 12:327–340.
Bethe H. (1929). Splitting of terms in crystals. Amer. Phys., 3:133–206.
Bieri R., Koide M., and Goldberg E. D. (1966). Noble gases contents of Pacific seawater.
J. Geophys. Res., 71:5243–5265.
Bigeleisen J. (1949). The relative reaction velocities of isotopic molecules. J. Chem. Phys.,
17:675–678.
Bigeleisen J. and Mayer M. G. (1947). Calculation of equilibrium constants for isotopic ex-
change reactions. J. Chem. Phys., 13:261–267.
Biggerstaff D. R. (1986). The thermodynamic properties of aqueous solutions of argon, ethyl-
ene, and xenon up to 720 K and 34 MPa. Ph.D. diss., University of Delaware.
Bina C. R. and Wood B. J. (1987). The olivine-spinel transitions: Experimental and thermody-
namic constraints and implications for the nature of the 400 km seismic discontinuity.
J. Geophys. Res., 92:4853–4859.
Birch F. (1966). Compressibility. In Handbook of Physical Constants, S. P. Clark Jr., ed. Memoir
97, The Geological Society of America.
Bird D. K. and Helgeson H. C. (1980). Chemical interaction of aqueous solution with
epidolite-feldspar mineral assemblage in geologic systems, I: Thermodynamic analysis
of phase relations in the system CaO-FeO-Fe2O3-Al2O3-SiO2-H2O-CO2. Amer. Jour.
Sci., 280:907–941.
Bird M. L. (1971). Distribution of trace elements in olivines and pyroxenes: An experimental
study. Ph.D. diss., University of Missouri, Rolla.
Bish D. L. (1981). Cation ordering in synthetic and natural Ni-Mg olivine. Amer. Mineral.,
66:770–776.
Bishop F. C. (1980). The distribution of Fe⫹⫹ and Mg between coexisting ilmenite and pyrox-
ene with application to geochemistry. Amer. Jour. Sci., 280:46–47.
Bishop F. C., Smith J. V., and Dawson J. B. (1976). Na, P, Ti, and coordination of Si in garnet
from peridotite and eclogite xenoliths. Nature, 260:696–697.
Blasi A., Brajkovic A., and De Pol Blasi (1983). Dry-heating conversions of low microcline to
high sanidine via a one-step disordering process. Abstr. NATO Advanced Study Inst. on
Feldspars and Feldspatoids, Rennes, France.
Blencoe J. G. (1977). Molal volumes of synthetic paragonite-muscovite micas. Amer. Mineral.,
62:1200–1215.
Blencoe J. G. and Luth W. C. (1973). Muscovite-paragonite solvi at 2, 4, and 8 kbar pressure.
Geol. Soc. Amer. Abstr. with Programs, 5:553–554.
Bocchio R., Bondi M., and Domeneghetti M. C. (1979). Reexamination of shefferite, urbanite,
and lindesite from central Sweden, Part I: Chemistry and crystal structure. Per. Min-
eral., 48:181–193.
Bockris J. O’M. and Mellors G. W. (1956). Electric conductance in liquid lead silicates and
borates. J. Phys. Chem., 60:1321–1328.
Bockris J. O’M. and Reddy A. K. N. (1970). Modern Electrochemistry. New York: Plenum
Press.
Bockris J. O’M., Kitchener J. A., and Davies A. E. (1952a). Electric transport in liquid silicates.
Trans. Faraday Soc., 48:536–548.
Bockris J. O’M., Kitchener J. A., Ignatowicz S., and Tomlinson J. W. (1952b). The electrical
conductivity of silicate melts: Systems containing Ca, Mn, Al. Discuss. Faraday Soc.,
4:281–286.
Boeglin J. L. (1981). Mineralogie et geochimie des gisements de manganese de Conselheiro
Lafaiete au Bresil et de Moanda au Gabon. These 3eme Cycle, Toulouse, 1608.
Bohlen S. R. and Boettcher A. L. (1981). Experimental investigations and geological applica-
tions of orthopyroxene geobarometry. Amer. Mineral., 66: 951–964.
References / 821
Bohlen S. R. and Essene E. J. (1977). Feldspar and oxide thermometry of granulites in the
Adirondack Highlands. Contrib. Mineral. Petrol., 62:153–169.
Bohlen S. R., Wall V. J., and Boettcher A. L. (1982). The system albite-water-carbon dioxide: A
model for melting and activities of water at high pressures. Amer. Mineral., 67:451–462.
Bohlen S. R., Wall V. J., and Boettcher A. L. (1983). Experimental investigation and geological
application of equilibria in the system FeO-TiO2-Al2O3-SiO2-H2O. Amer. Mineral.,
68:1049–1058.
Bokreta M. (1992). Energetics of garnets: Computational model of the thermochemical and
thermophysical properties. Ph.D. diss., University of Philadelphia.
Bokreta M. and Ottonello G. (1987). Enthalpy of formation of end-member garnets. EOS,
68:448.
Bonatti E. (1965). Palagonite, hyaloclastites, and alteration of volcanic glass in the ocean. Bull.
Volcanol., 28:251–269.
Bonatti E., Ottonello G., and Hamlyn P. R. (1986). Peridotites from the Island of Zabargad
(St. John)., Red Sea: Petrology and geochemistry. J. Geophys. Res., 91:599–631.
Boon J. A. (1971). Mössbauer investigations in the system Na2O-FeO-SiO2. Chem. Geol.,
7:153–169.
Boon J. A. and Fyfe W. S. (1972). The coordination number of ferrous ions in silicate systems.
Chem. Geol., 10:287–298.
Borg I. Y. (1967). Optical properties and cell parameter in the glaucophane-riebeckite series.
Contrib. Mineral. Petrol., 15:67–92.
Born M. (1920). Volumen und Hydratationswarme der ionen. Zeitschr. Physic, 1:45–48.
Born M. and Von Karman T. (1913). Hüber die Verteilung der Eigenschwingungen von Punkt-
gittern. Physik. Zeitschr., 14:65–71.
Boström D. (1987). Single-crystal X-ray diffraction studies of synthetic Ni-Mg olivine solid
solutions. Amer. Mineral., 66:770–776.
Boström D. (1988). Experimental studies of (Ni,Mg)—and (Co,Mg)—olivine solid solutions
by single crystal X-ray diffraction and solid state emf method. Ph.D. thesis, Department
of Inorganic Chemistry, University of Umeå, Umeå, Sweden.
Boström D. and Rosen E. (1988). Determination of activity-composition in (Ni,Mg)2SiO4 solid
solution at 1200–1600 K by solid state emf measurements. Acta Chim. Scand.,
A42:149–155.
Boswara I. M. and Franklyn A. D. (1968). Theory of the energetics of simple defects in oxides.
In Mass Transport in Oxides, J. B. Wachtman and A. D. Franklyn, eds. Natl. Bur. Stand.
(U.S.). Spec. Publ., 296.
Bottinga Y. (1968). Calculation of fractionation factors for carbon and oxygen isotopic ex-
changes in the system calcite-carbon dioxide-water. J. Phys. Chem., 72:800–807.
Bottinga Y. (1969a). Calculated fractionation factors for carbon and hydrogen isotope ex-
change in the system calcite-carbon dioxide-graphite-methane-hydrogen-water vapor.
Geochim. Cosmochim. Acta, 33:49–64.
Bottinga Y. (1969b). Carbon isotope fractionation between graphite, diamond and carbon di-
oxide. Earth Planet. Sci. Letters, 5:301–307.
Bottinga Y. (1985). On the isothermal compressibility of silicate liquids at high pressure. Earth
Planet. Sci. Letters., 74:350–360.
Bottinga Y. E. and Richet P. (1981). High pressure and temperature equation of state and calcu-
lation of the therodynamic properties of gaseous carbon dioxide. Amer. Jour. Sci.,
281:615–660.
Bottinga Y. and Weill D. F. (1970). Densities of liquid silicates systems calculated from partial
molar volumes of oxide components. Amer. Jour. Sci., 269:169–182.
Bowen N. L. (1913). Melting phenomena in the plagioclase feldspar. Amer. Jour. Sci.,
35:577–599.
Bowen N. L. (1915). The crystallization of haplobasaltic, haplorhyolitic and related magmas.
Amer. Jour. Sci., 4th ser., 33:551–573.
Bowen N. L. (1928). The Evolution of Igneous Rocks. Princeton University Press, Princeton, N.J.
822 / References
Bowen N. L. and Schairer J. F. (1935). The system MgO-FeO-SiO2. Amer. Jour. Sci.,
29:151–217.
Bowen N. L. and Schairer J. F. (1938). Crystallization equilibrium in nepheline-albite-silica
mixtures with fayalite. J. Geol., 46:397–411.
Bowen N. L., Shairer J. F., and Posniak E. (1933). The system Ca2SiO4-Fe2SiO4. Amer. Jour.
Sci., 26:273–297.
Boyd F. R. (1973). The pyroxene geothermometry. Geochim. Cosmochim. Acta, 37:2533–2546.
Boyd F. R. and Schairer J. F. (1964). The system MgSiO3-CaMgSi2O6. J. Petrol., 5:275–309.
Bragg W. L. (1920). The arrangement of atoms in crystals. Phil. Mag., 40:169–172.
Brereton N. R. (1970). Corrections for interfering isotopes in the 40Ar/ 39Ar dating method.
Earth Planet. Sci. Letters, 8:427–433.
Brewer P. G. (1975). Minor elements in seawater. In Chemical Oceanography. vol. 2, J. P. Riley
and G. Skirrow, eds. New York: Academic Press.
Bricker O. (1965). Some stability relations in the system Mn-O2-H2O at 25° and one atmosphere
total pressure. Amer. Mineral., 57:284–287.
Brillouin L. (1953). Wave Propagation in Periodic Structures. New York: Dover.
Brinkmann U. and Laqua W. (1985). Decomposition of fayalite (Fe2SiO4). in an oxygen poten-
tial gradient at 1418 K. Phys. Chem. Minerals, 12:283–290.
Broecker W. S., Gerard R. D., Ewing M., and Heezen B. C. (1961). Geochemistry and physics
of ocean circulation. In Oceanography, M. Sears, ed. Amer. Assn. Adv. Sci., Wash-
ington.
Brookins D. G. (1988). Eh-pH Diagrams for Geochemistry. Berlin-Heidelberg-New York:
Springer-Verlag.
Brown B. E. and Bailey S. W. (1964). The structure of maximum microcline and the sanidine-
microcline series. Norsk Geol. Tidsskr., 42/2:25–36.
Brown G. E. (1970). The crystal chemistry of the olivines. Ph.D. Dissertation, Virginia Polytech-
nic Inst. and State University, Blacksburg, Virginia.
Brown G. E. (1982). Olivines and silicate spinels. In Reviews in Mineralogy, vol. 5, P. H. Ribbe
(series ed.), Mineralogical Society of America.
Brown G. E. and Prewitt C. T. (1973). High temperature crystal chemistry of hortonolite. Amer.
Mineral., 58:577–587.
Brown G. E., Prewitt C. T., Papike J. J., and Sueno S. (1972). A composition of the structures
of low and high pigeonite. J. Geophys. Res., 77:5778–5789.
Brown G. M. (1967). Mineralogy of basaltic rocks. In Basalts, H. H. Hess and A. Poldervaart,
eds., New York: John Wiley.
Brown I. D. and Shannon R. D. (1973). Empirical bond-strength—bond-length curves for
oxides. Acta Cryst., A29:266–282.
Brown W. L. and Parson I. (1981). Towards a more practical two-feldspar geothermometer.
Bruno E., Chiari G., and Facchinelli A. (1976). Anorthite quenched from 1530 °C. I. Structure
refinement. Acta Cryst., B32:3270–3280.
Buddington A. F. and Lindsley D. H. (1964). Iron-titanium oxide minerals and synthetic equiv-
alents. J. Petrol., 5:310–357.
Buening D. K. and Buseck P. K. (1973). Fe-Mg lattice diffusion in olivine. J. Geophys. Res.,
78:6852–6862.
Buiskool Toxopeus J. M. A. and Boland J. M. (1976). Several types of natural deformation in
olivine: An electron microscope study. Tectonophysics, 32:209–233.
Burnham C. W. (1967). Ferrosilite. Carnegie Inst. Wash. Yb., 65:285–290.
Burnham C. W. (1975). Water and magmas: A mixing model. Geochim. Cosmochim. Acta,
39:1077–1084.
Burnham C. W. and Davis N. F. (1974). The role of H2O in silicate melts, II: Thermodynamic
and phase relations in the system NaAlSi3O8-H2O to 10 kbar, 700 °C to 1000 °C. Amer.
Jour. Sci., 274:902–940.
References / 823
Burns R. G. (1970). Mineralogical Application of Crystal Field Theory Cambridge University

Press, Cambridge.
Burns R. G. (1975). On the occurrence and stability of divalent chromium in olivines included
in diamonds. Contrib. Mineral. Petrol., 51:213–221.
Burns R. G. and Fyfe W. S. (1967). Crystal field theory and geochemistry of transition ele-
ments. In Researches in Geochemistry, vol. 2, P. H. Abelson (series ed.). New York:
John Wiley.
Burns R. G. and Vaughan D. J. (1970). Interpretation of the reflectivity behavior of ore miner-
als. Amer. Mineral., 55:1576–1586.
Buseck P. R., Nord G. L., and Veblen D. R. (1982). Subsolidus phenomena in pyroxenes. In
Reviews in Mineralogy, vol. 7, 2d ed., P. H. Ribbe (series ed.), Mineralogical Society
of America.
Cahn J. W. (1959). Free energy of a nonuniform system, II: Thermodynamic basis. J. Chem.
Phys., 30:1121–1124.
Cahn J. W. (1968). Spinodal decomposition. Trans. Met. Soc. AIME, 242:166–180.
Cahn J. W. and Hilliard J. E. (1958). Free energy of a nonuniform system, I: Interfacial free
energy. J. Chem. Phys., 28:258–267.
Cameron K. L. (1975). An experimental study of actinolite-cummingtonite phase relations with
notes on the synthesis of Fe-rich anthophyllite. Amer. Mineral., 60:375–390.
Cameron M. and Papike J. J. (1981). Structural and chemical variations in pyroxenes. Amer.
Mineral., 66:1–50.
Cameron M. and Papike J. J. (1982). Crystal chemistry of silicate pyroxenes. In Reviews in
Mineralogy. vol. 7, 2d ed., P. H. Ribbe (series ed.), Mineralogical Society of America.
Cameron M., Sueno S., Prewitt C. T., and Papike J. J. (1973a). High-temperature crystal chem-
istry of K-fluor-richterite (abstract). Trans. Amer. Geophys. Union, 54:457–498.
Cameron M., Sueno S., Prewitt C. T., and Papike J. J. (1973b). High temperature crystal chem-
istry of acmite, diopside, hedembergite, jadeite, spodumene and ureyte. Amer. Min-
eral., 58:596–618.
Campbell F. E. and Roeder P. (1968). The stability of olivine and pyroxene in the Ni-Mg-Si-O
system. Amer. Mineral., 53:257–268.
Capuano R. M. (1992). The temperature dependence of hydrogen isotope fractionation be-
tween clay minerals and water: Evidence from a geopressured system. Geochim. Cos-
mochim. Acta, 56:2547–2554.
Carmichael I. S. E. and McDonald A. (1961). The geochemistry of some natural acid glasses
from the North Atlantic tertiary volcanic province. Geochim. Cosmochim. Acta,
25:189–222.
Carmichael I. S. E., Nicholls J., Spera F. J., Wood B. J., and Nelson S. A. (1977). High tempera-
ture properties of silicate liquids: applications to the equilibration and ascent of basic
magma. Phil. Trans. Roy. Soc. London, A286:373–431.
Carpenter M. A. (1985). Order-disorder transformations in mineral solid solutions. In Reviews
in Mineralogy, P. H. Ribbe (series ed.), Mineralogical Society of America, 14:187–
223.
Carpenter M. A. (1988). Thermochemistry of aluminium/silicon ordering in feldspar minerals.
In Physical Properties and Thermodynamic Behaviour of Minerals, E. K. H. Salje, ed.
265–323, D. Reidel Publishing Company.
Carpenter M. A. and Ferry J. M. (1984). Constraints on the thermodynamic mixing properties
of plagioclase feldspars. Contrib. Mineral. Petrol., 87:138–148.
Carpenter M. A. and McConnell J. D. (1984). Experimental delineation of the transformation
in intermediate plagioclase feldspars. Amer. Mineral., 69:112–121.
Carpenter M. A. and Salje E. K. H. (1994a). Thermodynamics of nonconvergent cation order-
ing in minerals, II: Spinels and the orthopyroxene solid solution. Amer. Mineral.,
79:1068–1083.
Carpenter M. A. and Salje E. K. H. (1994b). Thermodynamics of nonconvergent cation order-
824 / References
ing in minerals, III: Order parameter coupling in potassium feldspar. Amer. Mineral.,
79:1084–1093.
Carpenter M. A., Powell R., and Salje E. K. H. (1994). Thermodynamics of nonconvergent
cation ordering in minerals, I: An alternative approach. Amer. Mineral., 79:1053–1067.
Catanzaro E. J. (1963). Zircon ages in southwestern Minnesota. J. Geophys. Res., 68:2045–2048.
Catlow C. R. A. and Stoneham A. M. (1983). Ionicity in solids. J. Phys., C16:4321–4338.
Catti M. (1981). A generalized Born-Mayer parametrization of the lattice energy in orthorhom-
bic ionic crystals. Acta Cryst., A37:72–76.
Chacko T., Mayeda T. K., Clayton R. N., and Goldsmith J. R. (1991). Oxygen and carbon
isotope fractionations between CO2 and calcite. Geochim. Cosmochim. Acta, 55:2867–
2882.
Chakraborty S. and Ganguly J. (1991). Compositional zoning and cation diffusion in garnets.
In Advances in Physical Geochemistry, vol. 8, J. Ganguly, ed. New York-Heidelberg-
Berlin: Springer-Verlag.
Chakraborty S. and Ganguly J. (1992). Cation diffusion in aluminosilicate garnets: Experimen-
tal determination in spessartine-almandine diffusion couples, evaluation of effective bi-
nary diffusion coefficients, and applications. Contrib. Mineral. Petrol., 111:74–86.
Charles R. W. (1974). The physical properties of the Mg-Fe richerites. Amer. Mineral.,
59:518–528.
Charles R. W. (1980). Amphiboles on the join pargasite-ferropargasite. Amer. Mineral., 65:996–
1001.
Charlu T. V., Newton R. C., and Kleppa O. J. (1975). Enthalpies of formation at 970 K of
compounds in the system MgO-Al2O3-SiO2 from high temperature solution calorimetry.
Charlu T. V., Newton R. C., and Kleppa O. J. (1978). Enthalpy of formation of lime silicates
by high temperature calorimetry, with discussion of high pressure phase equilibrium.
Chatillon-Colinet C., Newton R. C., Perkins D., and Kleppa O. J. (1983). Thermochemistry of
(Fe⫹⫹,Mg)SiO3 orthopyroxene. Geochim. Cosmochim. Acta, 47:1597–1603.
Chatterjee N. (1987). Evaluation of thermochemical data on Fe-Mg olivine, orthopyroxene,
spinel and Ca-Fe-Mg-Al garnet. Geochim. Cosmochim. Acta, 51:2515–2525.
Chatterjee N. (1989). An internally consistent thermodynamic data base on minerals: Applica-
tions to the earth’s crust and upper mantle. Ph.D. diss., City University of New York.
Chatterjee N. D. (1970). Synthesis and upper stability of pargasite. Contrib. Mineral. Petrol.,
27:244–257.
Chatterjee N. D. (1972). The upper stability limit of the assemblage paragonite ⫹ quartz and
its natural occurrences. Contrib. Mineral. Petrol., 34:288–303.
Chatterjee N. D. (1974). Synthesis and upper thermal stability limit of 2M-margarite,
CaAl2[Al2Si2O10(OH)2]. Schweiz. Mineral. Petrogr. Mitt., 54:753–767.
Chatterjee N. D. (1976). Margarite stability and compatibility relations in the system CaO-
Al2O3-SiO2-H2O as a pressure-temperature indicator. Amer. Mineral., 61:699–709.
Chatterjee N. D. and Froese E. (1975). A thermodynamic study of the pseudobinary join
muscovite-paragonite in the system KAlSi3O8-NaAlSi3O8-Al2O3-SiO2-H2O. Amer. Min-
eral., 61:699–709.
Chatterjee N. D. and Johannes W. (1974). Thermal stability and standard thermodynamic
properties of synthetic 2M, muscovite KA1[AlSi3O10(OH)2]. Contrib. Mineral. Petrol.,
48:89–114.
Cheng W. and Ganguly J. (1994). Some aspects of multicomponent excess free energy models
with subregular binaries. Geochim. Cosmochim. Acta, 58:3763–3767.
Chiari G., Facchinelli A. and Bruno E. (1978). Anorthite quenched from 1630 °C. Discussion.
Acta Cryst., B34:1757–1764.
Chiba H., Chacko T., Clayton R. N., and Goldsmith J. R. (1989). Oxygen isotope fraction-
ations involving diopside, forsterite, magnetite, and calcite: Application to geother-
mometry. Geochim. Cosmochim. Acta, 53:2985–2995.
References / 825
Chiodini G. and Cioni R. (1989). Gas geobarometry for hydrothermal systems and its applica-
tions to some Italian geothermal areas. Appl. Geochem., 4:465–472.
Clark A. M. and Long J. V. P. (1971). Anisotropic diffusion of nickel in olivine. In Graham
Memorial Symposium on Diffusion Processes. New York: Gordon & Breach.
Clark J. R., Appleman D. E., and Papike J. J. (1969). Crystal-chemical characterization of cli-
nopyroxenes based on eight new structure refinements. Min. Soc. Amer. Spec. Paper,
2:31–50.
Clayton R. N. and Kieffer S. W. (1991). Oxygen isotopic thermometer calibrations. In Stable
Isotope Geochemistry: A Tribute to Samuel Epstein, H. P. Taylor Jr., J. R. O’Neil, and
I. R. Kaplan, eds., The Geochemical Society, Special Publication n. 3.
Clayton R. N., Goldsmith J. R., and Mayeda T. K. (1989). Oxygen isotope fractionations in
quartz, albite, anorthite and calcite. Geochim. Cosmochim. Acta, 53:725–733.
Clugston M. J. (1978). The calculation of intermolecular forces: A critical examination of the
Gordon-Kim model. Adv. Phys., 27:893–912.
Cohen R. E. (1986). Thermodynamic solution properties of aluminous clinopyroxenes: Nonlin-
ear least squares refinements. Geochim. Cosmochim. Acta, 50:563–575.
Coltorti M., Girardi V., and Schoerster J. (1987). Liquid immiscibility in the Archean Green-
stone Belt of Piumhi (Minas Gerais, Brazil). Lithos, 20:77–91.
Colville A. A. and Ribbe P. H. (1968). The crystal structure of an adularia and a refinement of
a structure of orthoclase. Amer. Mineral., 53:25–37.
Connolly J. A. D. (1992). Phase diagram principles and computations: A review. In Proceedings
of the V Summer School of Earth and Planetary Sciences, University of Siena Press,
Siena.
Connolly J. A. D. (1994). Computer calculation of multidimensional phase diagrams: Why and
how? In Proceedings of the VII Summer School of Earth and Planetary Sciences, Univer-
sity of Siena Press, Siena.
Connolly J. A. D. and Kerrick D. M. (1987). An algorithm and computer program for calculat-
ing composition phase diagrams. CALPHAD, 11:1–55.
Coulson C. A. (1961). Valence. (2d ed.) Oxford University Press, London.
Craig H. (1957). Isotopic standards for carbon and oxygen and correction factors for mass-
spectrometric analysis of carbon dioxide. Geochim. Cosmochim. Acta, 12:133–149.
Craig H. (1961). Standards for reporting concentrations of deuterium and oxygen-18 in natural
waters. Science, 133:1833–1934.
Craig H. (1963). The isotopic geochemistry of water and carbon in geothermal areas. In Nu-
clear Geology of Geothermal Areas, E. Tongiorgi, ed. Spoleto.
Crank J. (1975). The Mathematics of Diffusion Oxford University Press, London.
Cressey G. (1978). Exsolution in almandine-pyrope-grossular garnet. Nature, 271:533–534.
Cressey G., Schmid R., and Wood B. J. (1978). Thermodynamic properties of almandine-
grossular garnet solid solutions. Contrib. Mineral. Petrol., 67:397–404.
Cripps-Clark C. J., Sridhar R., Jeffes J. H. E., and Richardson F. D. (1974). Chain distribution
and transition temperatures for phosphate glasses. In Physical Chemistry of Process
Metallurgy. J. H. E. Jeffes and R. J. Tait, eds. Inst. Mining Met.
Currie K. L. (1971). The reaction 3 cordierite ⫽ 2 garnet ⫹ 4 sillimanite ⫹ 5 quartz as a
geological thermometer in the Ipinicon Lake Region, Ontario. Contrib. Mineral. Pet-
rol., 33:215–226.
Currie K. L. and Curtis L. W. (1976). An application of multicomponent solution theory to
jadeitic pyroxenes. J. Geol., 84:179–194.
Curtis G. H. and Hay R. L. (1972). Further geological studies and potassium-argon dating at
Olduvai Gorge and Ngorongoro Crater. In Calibration of Hominoid Evolution, W. W.
Bishop and J. A. Miller, eds. Edinburgh: Scottish Academic Press.
Czaya R., (1971). Refinement of the structure of ␣⫺Ca2SiO4. Acta Cryst., B27:848–849.
Dahl P. S. (1980). The thermal compositional dependence of Fe⫹⫹–Mg distributions between
co-existing garnet and pyroxene: Applications to geothermometry. Amer. Mineral.,
65:852–866.
826 / References
Dal Negro A., Carbonin S., Molin G. M., Cundari A., and Piccirillo E. M. (1982). Intracrystal-
line cation distribution in natural clinopyroxenes of tholeiitic, transitional and alkaline
basaltic rocks. In Advances in Physical Geochemistry. vol. 1, S. K. Saxena (series ed.),
New York: Springer-Verlag.
Dal Negro A., Carbonin S., Domeneghetti C., Molin G. M., Cundari A., and Piccirillo E. M.
(1984). Crystal chemistry and evolution of the clinopyroxene in a suite of high pressure
ultramafic nodules from the newer volcanism of Victoria, Australia. Contrib. Mineral.
Petrol., 86:221–229.
Dalrymple G. B. and Lanphere M. A. (1969). Potassium-Argon Dating W. H. Freeman, San
Francisco.
Dalrymple G. B. and Lanphere M. A. (1971). 40Ar/ 39Ar dating technique of K-Ar dating: A
comparison with the conventional technique. Earth Planet. Sci. Letters, 12:300–308.
Damon P. E., Lerman J. C., and Long A. (1978). Temporal fluctuations of 14C: Casual factors
and implications. Ann. Rev. Earth Planet. Sci., 6:457–464.
D’Amore F. and Panichi C. (1980). Evaluation of deep temperatures of hydrothermal systems
by a new gas geothermometer. Geochim. Cosmochim. Acta, 44:549–556.
Darken L. S. (1948). Diffusion mobility and their interrelations through free energy in binary
metallic systems. Trans. Met. Soc. AIME, 175:184–201.
Darken L. S. (1967). Thermodynamics of binary metallic solutions. Trans. Met. Soc. AIME,
239:80–89.
Das C. D., Keer H. V., and Rao R. V. G. (1963). Lattice energy and other properties of some
ionic crystals. Z. Physic. Chem., 224:377–383.
Davidson P. M. and Mukhopadhyay D. K. (1984). Ca-Fe-Mg olivines: Phase relations and a
solution model. Contrib. Mineral. Petrol., 86:256–263.
Davis L. L. and Smith D. (1993). Ni-rich olivine in minettes from Two Buttes, Colorado: A
connection between potassic melts from the mantle and low Ni partition coefficients.
Davoli P. (1987). A crystal chemical study of aegirin-augites and some evaluations on the oxida-
tion state of Mn. Neues Jahrb. Miner. Abh., 158:67–87.
De A. (1974). Silicate liquid immiscibility in the Deccan Traps and its petrogenetic significance.
Geol. Soc. Amer. Bull., 85:471–474.
De Baar H. J. W., Bacon M. P., Brewer P. G., and Bruland K. W. (1985). Rare earth elements
in the Pacific and Atlantic Oceans. Geochim. Cosmochim. Acta, 49:1943–1960.
Debye P. (1912). Zur Theorie der spezifischen warmer. Ann. Physik, 39(4).: 789–839.
Deer W. A., Howie R. A., and Zussman J. (1978). Rock Forming Minerals vol. 2A, (2d ed.),
New York: John Wiley.
Deer W. A., Howie R. A., and Zussman J. (1983). An Introduction to the Rock-Forming Miner-
als. Longman, Harlow Essex, England.
Deganello S. (1978). Thermal expansion from 25°C to 500°C of a few ionic radii. Zeit. Krist.,
147:217–227.
Deines P., Langmuir D., and Harmon R. S. (1974). Stable carbon isotope ratios and the exis-
tence of a gas phase in the evolution of carbonate ground waters. Geochim. Cosmochim.
Acta, 38:1147–1164.
Delany J. M., Puigdomenech I., and Wolery T. J. (1986). Precipitation Kinetics Option for the
EQ6 Geochemical Reaction Path Code Lawrence Livermore National Laboratory, Liv-
ermore, Cal., UCRL-53642.
Della Giusta A. and Ottonello G. (1993). Energy and long-range disorder in simple spinels.
Della Giusta A., Ottonello G., and Secco L. (1990). Precision estimates of interatomic distances
using site occupancies, ionization potentials and polarizability in Pbnm silicate olivines.
Acta Cryst., B46:160–165.
Dempsey M. J. (1980). Evidence for structural changes in garnet caused by calcium substitu-
tion. Contrib. Mineral. Petrol., 7:281–282.
References / 827
Denbigh K. G. (1971). The Principles of Chemical Equilibrium. 3d ed., Cambridge University

Press, Cambridge.
De Paolo D. J. and Wasserburg G. J. (1979). Sm-Nd age of the Stillwater complex and the
mantle curve for neodymium. Geochim. Cosmochim. Acta, 43:999–1008.
Desborough G. A. (1975). Authigenic albite and potassium feldspar in the Green River forma-
tion, Colorado and Wyoming. Amer. Mineral., 60:235–239.
De Vore G. W. (1955). The role of absorption in the fractionation and distribution of elements.
J. Geol., 63:159–190.
Dodson M. H. (1973). Closure temperature in cooling geochronological and petrological sys-
tems. Contrib. Mineral. Petrol., 40:259–274.
Dodson M. H. (1979). Theory of cooling ages. In Lectures in Isotope Geology, E. Jäger and
J. C. Hunziker, eds., Berlin: Springer-Verlag.
Doerner H. A. and Hoskins W. M. (1925). Coprecipitation of radium and barium sulfates.
J. Amer. Chem. Soc., 47:662–675.
Donnay G. (1969). Further use for the Pauling-bond concept. Carnegie Inst. Wash. Yb.,
68:292–295.
Donnay G. and Allmann R. (1970). How to recognize O2⫺, OH, and H2O in crystal structures
determined by X-rays. Amer. Mineral., 55:1003–1015.
Downs J. W. (1991). Electrostatic properties of minerals from X-ray diffraction data: A guide
for accurate atomistic models. In Advances in Physical Geochemistry, vol. 8, J. Ganguly,
ed. Berlin-Heidelberg-New York: Springer-Verlag.
Drake M. J. and Holloway J. R. (1978). Henry’s law behavior of Sm in a natural plagioclase/
melt system: Importance of experimental procedure. Geochim. Cosmochim. Acta,
42:679–683.
Drake M. J. and Weill D. F. (1975). The partioning of Sr, Ba, Ca, Y, Eu2⫹, Eu3⫹. and other
REE between plagioclase feldspar and magmatic silicate liquid: An experimental study.
Driessens F. C. M. (1968). Thermodynamics and defect chemistry of some oxide solid solu-
tions. Parts I and II. Ber. Buns. Phys. Chem., 72:754–772.
Duchesne J. C. (1972). Iron-titanum oxide minerals in the Bjerkrem-Sogndal massif, south-
western Norway. J. Petrol., 13:57–81.
Dudas M. J., Schmitt R. A., and Harward M. E. (1971). Trace element partitioning between
volcanic plagioclases and dacitic pyroclastic matrix. Earth Planet. Sci. Letters, 11:440–
446.
Duncan A. B. F. and Pople J. A. (1953). The structure of some simple molecules with lone pair
electrons. Trans. Faraday Soc., 49:217–227.
Dunitz J. D. and Orgel L. E. (1957). Electronic properties of transition metal oxides, II: Cation
distribution amongst octahedral and tetrahedral sites. J. Phys. Chem. Solids, 3:318–323.
Einstein A. (1907). Die Plancksche Theorie der Strahlung und die Theorie der spezifischen
Wärme. Ann. Physik, 22:180–190.
Eisemberg D. and Kauzmann W. (1969). The Structures and Properties of Water Oxford Univer-
sity Press, London.
Ellis A. J. and Golding R. M. (1963). The solubility of carbon dioxide above 100°C in water
and sodium chloride solutions. Amer. Jour. Sci., 261:47–60.
Ellis D. J. and Green D. H. (1979). An experimental study of the effect of Ca upon garnet-
clinopyroxene exchange equilibria. Contrib. Mineral. Petrol., 71:13, 22.
Elphick S. C., Ganguly J., and Leonis T. P. (1981). Experimental study of Fe-Mg interdiffusion
in aluminumsilicate garnet (abstract). EOS Trans. Amer. Geophys. Union, 62:411.
Engi M. (1978). Mg-Fe Exchange equilibria among spinel, olivine, orthopyroxene and cordier-
ite. Ph.D. diss., n. 6256, Eidgenossische Technische Hochschule, Zurich, Switzerland.
Epstein S. (1959). The variation of the 18O/16O ratio in nature and some geologic applications.
In Researches in Geochemistry, P. H. Abelson, ed. New York: John Wiley.
Ernst W. G. and Calvert S. E. (1969). An experimental study of the recrystallization of porcela-
828 / References
nite and its bearing on the origin of some bedded cherts. Amer. Jour. Sci., 267A:114–
133.
Ernst W. G. and Wai C. M. (1970). Infrared, X-ray and optical study of cation ordering and
dehydrogenation in natural and heat-treated sodic amphiboles. Amer. Mineral.,
55:1226–1258.
Essene E. J. (1982). Geologic thermometry and barometry. In Reviews in Mineralogy, vol. 10,
P. H. Ribbe (series ed.), Mineralogical Society of America.
Eugster H. P. (1954). Distribution of cesium between sanidine and a hydrous fluid. Carnegie
Inst. Wash. Yb., 53:102–104.
Eugster H. P. and Baumgartner L. (1987). Mineral solubilities and speciation in supercritical
metamorphic fluids. In Reviews in Mineralogy. vol. 17, P. H. Ribbe (series ed.), Mineral-
Eugster H. P. and Wones D. R. (1962). Stability relations of the ferruginos biotite, annite.
J. Petrol., 3:82–125.
Eugster H. P., Albee A. L., Bence A. E., Thompson J. B. Jr., and Walbaum D. R. (1972). The
two phase region and excess mixing properties of paragonite-muscovite crystalline solu-
tions. J. Petrol., 13:147–179.
Ewart A. and Taylor S. R. (1969). Trace element geochemistry of rhyolitic volcanic rocks, Cen-
tral North Island, New Zealand. Phenocryst data. Contrib. Mineral. Petrol., 22:127–138.
Ewart A., Bryan W. B., and Gill J. B. (1973). Mineralogy and geochemistry of younger volcanic
islands of Tonga S. W. Pacific. J. Petrol., 14:429–465.
Ewart A., Taylor S. R., and Capp A. C. (1968). Trace and minor element geochemistry of the
rhyolitic volcanic rocks, Central North Island, New Zealand. Contrib. Mineral. Petrol.,
18:76–85.
Faure G. (1986). Principles of Isotope Geology. 2d ed. New York: John Wiley.
Faure G. and Powell J. L. (1972). Strontium Isotope Geology. New York: Springer-Verlag.
Fei Y. and Saxena S. K. (1986). A thermochemical data base for phase equilibria in the system
Fe-Mg-Si-O at high pressure and temperature. Phys. Chem. Minerals., 13:311–324.
Fei Y. and Saxena S. K. (1987). An equation for the heat capacity of solids. Geochim. Cos-
mochim. Acta, 51:251–254.
Fei Y., Saxena S. K., and Eriksson G. (1986). Some binary and ternary silicate solution models.
Ferguson J. and Currie K. L. (1971). Evidence of liquid immiscibility in alkaline ultrabasic
dikes at Callender Bay, Ontario. J. Petrol., 12:561–585.
Ferry J. M. and Spear F. S. (1978). Experimental calibration of the partitioning of Fe and Mg
between biotite and garnet. Contrib. Mineral. Petrol., 66:113–117.
Fincham C. J. B. and Richardson R. F. (1954). The behaviour of sulfur in silicate and aluminate
melts. Proc. Roy. Soc. London, A223:40–61.
Finger L. W. (1970). Refinement of the crystal structure of an anthophyllite. Carnegie Inst.
Wash. Yb., 68:283–288.
Finger L. W. and Ohashi Y. (1976). The thermal expansion of diopside to 800 °C and a refine-
ment of the crystal structure at 700 °C. Amer. Mineral., 61:303–310.
Finnerty T. A. (1977). Exchange of Mn, Ca, Mg and Al between synthetic garnet, orthopyrox-
ene, clinopyroxene and olivine. Carnegie Inst. Wash. Yb., 68:290–292.
Finnerty T. A. and Boyd F. R. (1978). Pressure dependent solubility of calcium in forsterite
coexisting with diopside and enstatite. Carnegie Inst. Wash. Yb., 77:713–717.
Fisher K. (1966). A further refinement of the crystal structure of cummingtonite, (Mg, Fe)7
(Si4O11)2(OH)2. Amer. Mineral., 51:814–818.
Fitch F. J., Miller J. A., and Hooker P. J. (1976). Single whole rock K-Ar isochron. Geol.
Mag., 113:1–10.
Fleer V. N. (1982). The dissolution kinetics of anorthite (CaAl2Si2O8) and synthetic strontium
feldspar (SrAl2Si2O8) in aqueous solutions at temperatures below 100°C: With applica-
tions to the geological disposal of radioactive nuclear wastes. Ph.D. diss., Pennsylvania
State University, University Park.
References / 829
Flood H. and Förland T. (1947). The acidic and basic properties of oxides. Acta Chem.
Scand., 1:952–1005.
Flood H., Förland T., and Gzotheim K. (1954). Über den Zusammenhang zwischen Konzen-
trazion und Aktivitäten in geschmolzenen Salzmischungen. Zeit. Anorg. Allg. Chem.,
276:290–315.
Foland K. A. (1974). Alkali diffusion in orthoclase. In Geochemical Transport and Kinetics,
Hoffman, Giletti, Yoder, and Yund, eds. Carnegie Institution of Washington.
Francis C. A. and Ribbe P. H. (1980). The forsterite-tephroite series, I: Crystal structure re-
finement. Amer. Mineral., 65:1263–1269.
Franz G., Hinrichsen T., and Wannermacher E. (1977). Determination of the miscibility gap
on the mean of the infrared spetroscopy. Contrib. Mineral. Petrol., 59:207–236.
Fraser D. G. (1975a). Activities of trace elements in silicate melts. Geochim. Cosmochim.
Acta, 39:1525–1530.
Fraser D. G. (1975b). An investigation of some long-chain oxi-acid systems. D.Phil. diss., Uni-
versity of Oxford.
Fraser D. G. (1977). Thermodynamic properties of silicate melts. In Thermodynamics in Geol-
ogy. D. G. Fraser, ed. Reidel, Dortrecht-Holland.
Fraser D. G., Rammensee W., and Hardwick A. (1985). Determination of the mixing properties
of molten silicates by Knudsen cell mass spectrometry, II: The system (Na-K)AlSi4O10
and (Na-K)AlSi5O12. Geochim. Cosmochim. Acta, 49:349–359.
Fraser D. G., Rammensee W., and Jones R. H. (1983). The mixing properties of melts in the
system NaAlSi3O6-KAlSi2O6 determined by Knudsen-Cell Mass Spectrometry. Bull.
Mineral., 106:111–117.
Frey F. A. (1982). Rare earth element abundances in upper mantle rocks. In Rare Earth Element
Geochemistry, P. Henderson, ed. Elesevier, Amsterdam.
Friedel J. (1964). Dislocations. Addison-Wesley, Reading, Massachusetts.
Friedlander G. and Kennedy J. W. (1956). Nuclear and Radiochemistry. New York: John Wiley.
Friedlander G., Kennedy J. W., Macias E. S., and Miller J. M. (1981). Nuclear and Radiochemis-
try. New York: John Wiley.
Frondel C. F. (1962). The System of Mineralogy of James Dwight Dana and Edward Salysbury
Dana. 7th ed., vol. 8, Silica Minerals, New York: John Wiley.
Fujii T. (1977). Fe-Mg partitioning between olivine and spinel. Carnegie Inst. Wash. Yb.,
76:563–569.
Fujino K., Sasaki S., Takeuchi Y., and Sadanaga R. (1981). X-ray determination of electron
distribution in forsterite, faialite and tephroite. Acta Cryst., B37:513–518.
Fumi F. G. and Tosi M. P. (1957). Naor relations between Madelung constants for cubic ionic
lattices. Phil. Mag., 2:284–285.
Fumi F. G. and Tosi M. P. (1964). Ionic sizes and Born repulsive parameters in the NaCl-type
alkali halides, I: The Huggins-Mayer and Pauling forms. J. Phys. Chem. Solids, 25:31–43.
Fung P. C. and Shaw D. M. (1978). Na, Rb and Tl distributions between phlogopite and sani-
dine by direct synthesis in a common vapour phase. Geochim. Cosmochim. Acta,
42:703–708.
Fursenko B. A. (1981). High pressure synthesis of the chromium bearing garnets
Mn3Cr2Si3O12. Doklady Acad. Sci. USSR, 250:176–179.
Fyfe W. S., Turner F. J., and Verhoogen J. (1958). Metamorphic reactions and metamorphic
facies. Geol. Soc. Amer. Mem., vol. 75, 253 pp.
Gabis V., Abelard P., Ildefonse J. P., Jambon A., and Touray J. C. (1979). La diffusion a haute
temperature dans les systems d’interet geologique. In Haute Temperatures et Science
de la Terre. Centre Regional de Publications de Toulouse, Editions du C.N.R.S., 163–
174.
Ganguly J. (1973). Activity-composition relation of jadeite in omphacite pyroxenes: Theoretical
deductions. Earth Planet. Sci. Letters, 19:145–153.
Ganguly J. (1976). The energetics of natural garnet solid solution, II: Mixing of the calcium
silicate end members. Contrib. Mineral. Petrol., 55:81–90.
830 / References
Ganguly J. (1977). Crystal chemical aspects of olivine structures. N. Jahrb Miner. Abh.,
130:303–318.
Ganguly J. (1979). Garnet and clinopyroxene solid solutions and geothermometry based on
Fe-Mg distribution coefficient. Geochim. Cosmochim. Acta, 43:1021–1029.
Ganguly J. and Cheng W. (1994). Thermodynamics of (Ca,Mg,Fe,Mn)-garnet solid solution:
New experiments, optimized data set, and applications to thermo-barometry. I.M.A.,
16th General Meeting (Abstracts)., Pisa, Italy.
Ganguly J. and Kennedy G. C. (1974). The energetics of natural garnet solid solutions, I: Mix-
ing of the aluminosilicate end members. Contrib. Mineral. Petrol., 48:137–148.
Ganguly J. and Ruitz J. (1986). Time-temperature relation of mineral isochrons: A thermody-
namic model, and illustrative examples for the Rb-Sr system. Earth Planet. Sci. Let-
ters, 81:338–348.
Ganguly J. and Saxena S. K. (1984). Mixing properties of aluminosilicate garnets: Constraints
from natural and experimental data and applications to geothermo-barometry. Amer.
Mineral., 69:79–87.
Ganguly J. and Saxena S. K. (1987). Mixtures and Mineral Reactions. Berlin-Heidelberg-New
York: Springer-Verlag.
Ganguly J., Cheng W., and O’Neill H. St. C. (1993). Syntheses, volume, and structural changes
of garnets in the pyrope-grossular join: Implications for stability and mixing properties.
Amer. Mineral., 78:583–593.
Garrels R. M. and Christ C. L. (1965). Solutions, Minerals, and Equilibria. New York: Harper
and Row.
Garrels R. M. and Thompson M. E. (1962). A chemical model for seawater at 25°C and one
atmosphere total pressure. Amer. Jour. Sci., 260:57–66.
Gartner L. (1979). Relations entre enthalpies en enthalpies libres de formation des ions, des oxydes
de formule MnNmO2. Utilization des frequences de vibration dans l’infra-rouge Doct. Ing.
Université de Strasbourg, 193 pp.
Gasparik T. (1985). Experimentally determined compositions of diopside-jadeite pyroxene in
equilibrium with albite and quartz at 1200–1350 °C and 15–34 kbar. Geochim. Cos-
Gasparik T. and Lindsley D. H. (1980). Phase equilibria at high pressure of pyroxenes con-
taining monovalent and trivalent ions. In Reviews in Mineralogy, vol. 7, P. H. Ribbe
Gasperin M. (1971). Crystal structure of rubidium feldspar, RbAlSi3O8. Acta Cryst.,
B27:854–855.
Gasperin J. and McConnel J. D. (1984). Experimental delineations of the C1̄–I1̄ trasformation
in intermediate plagioclase feldspar. Amer. Mineral., 69:112–121.
Gast P. W. (1968). Trace element fractionation and the origin of tholeiitic and alkaline magma
types. Geochim. Cosmochim. Acta, 32:1057–1086.
Gautschi W. (1964). Error function and Fresnel integrals. In Handbook of Mathematical Func-
tions, M. Abramowitz and I. A. Stegun, eds. National Bureau of Standards, Wash-
ington.
Geiger C. A., Newton R. C. and Kleppa O. J. (1987). Enthalpy of mixing of synthetic
almandine-grossular and almandine-pyrope garnets from high-temperature solution
calorimetry. Geochim. Cosmochim. Acta, 51:1755–1763.
Gelinas L., Brooks C., and Trzcienski W. E. Jr. (1976). Archean variolites-quenched immiscible
liquids reexamined: A reply to criticisms. Can. Jour. Earth Sci., 14:2945–2958.
Gerlach T. M. and Nordlie B. E. (1975a). The C-O-H-S- gaseous systems, Part I: Composition
limits and trends in basaltic cases. Amer. Jour. Sci., 275:353–376.
Gerlach T. M. and Nordlie B. E. (1975b). The C-O-H-S-gaseous systems, Part II: Temperature,
atomic composition, and molecular equilibria in volcanic gases. Amer. Jour. Sci.,
275:377–394.
Gerlach T. M. and Nordlie B. E. (1975c). The C-O-H-S-gaseous systems, Part III: Magmatic
References / 831
gases compatible with oxides and sulfides in basaltic magmas. Amer. Jour. Sci.,
275:395–410.
Ghent E. D. (1976). Plagioclase-garnet-Al2SiO5-quartz: A potential geobarometer-
geothermometer. Amer. Mineral., 61:710–714.
Ghent E. D., Stout M. Z., Black P. M., and Brothers R. N. (1987). Chloritoid bearing rocks
associated with blueschists and eclogites, northern New California. J. Met. Geol.,
5:239–254.
Ghiorso M. S. (1984). Activity/composition relation in the ternary feldspars. Contrib. Mineral.
Petrol., 87:282–296.
Ghiorso M. S. and Carmichael I. S. E. (1980). A regular solution model for meta-aluminous
silicate liquids: Applications to geothermometry, immiscibility, and the source regions
of basic magma. Contrib. Mineral. Petrol., 71:323–342.
Ghiorso M. S., Carmichael I. S. E., Rivers M. L., and Sack R. O. (1983). The Gibbs free energy
of mixing of natural silicate liquids; an expanded regular solution approximation for the
calculation of magmatic intensive variables. Contrib. Mineral. Petrol., 84:107–145.
Ghose S. (1982). Subsolidus reactions and microstructures in amphiboles. In Reviews in Miner-
alogy, vol. 9A, P. H. Ribbe (series ed.), Mineralogical Society of America.
Ghose S. and Wan C. (1974). Strong site preference of Co2⫹ in olivine Co1.10Mg0.90SiO4. Con-
trib. Mineral. Petrol., 47:131–140.
Ghose S. and Weidner J. R. (1971). Mg2⫹-Fe2⫹ isotherms in cummingtonites at 600 and 700
°C (abstract). Trans. Amer. Geophys. Union, 52:381.
Ghose S. and Weidner J. R. (1972). Mg2⫹-Fe2⫹ order-disorder in cummingtonite (Mg,Fe)7Si8-
O22(OH)2. A new geothermometer. Earth Planet. Sci. Letters, 16:346–354.
Ghose S., Choudhury N., Chaplot S. L., and Rao K. R. (1992). Phonon density of states and
thermodynamic properties of minerals. In Advances in Physical geochemistry, vol. 10,
S. K. Saxena (series ed.). Berlin-Heidelberg-New York: Springer-Verlag.
Ghose S., Kersten M., Langer K., Rossi G. and Ungaretti L. (1986). Crystal field spectra and
Jahn-Teller effect of Mn3⫹ in clinopyroxene and clinoamphiboles from India. Phys.
Chem. Minerals, 13:291–305.
Gibbs J. W. (1906). On the equilibrium of heterogeneous substances. In The Scientific Papers of
J. Willard Gibbs. Vol. 1: Thermodynamics. Longmans Green (reprinted by Dover, 1961).
Giggenbach W. F. (1980). Geothermal gas equilibria. Geochim. Cosmochim. Acta, 44:2021–
2032.
Giggenbach W. F. (1987). Redox processes governing the chemistry of fumarolic gas discharges
from White Islands, New Zealand. Appl. Geochem., 2:143–161.
Gilbert M. C., Helz R. T., Popp R. K., and Spear F. S. (1982). Experimental studies of amphi-
bole stability. In Reviews in Mineralogy, vol. 9B, P. H. Ribbe (series ed.), Mineralogical
Society of America.
Giletti B. J., Sennet M. P., and Yund R. A. (1978). Studies in diffusion, III: Oxygen in feld-
spars—an ion microprobe determination. Geochim. Cosmochim. Acta, 42:45–57.
Glasser F. P. (1960). Einige Ergebnisse von Phasengleichgewwichtsuntersuchungen in der Sys-
temen MgO-MnO-SiO2 und CaO-MnO-SiO2. Silikattechnik, 11:362–363.
Godwin H. (1962). Half-life of radiocarbon. Nature, 195:984.
Goldman D. S. and Albee A. L. (1977). Correlation of Mg/Fe partitioning between garnet and
biotite with 18O/16O partitioning between quartz and magnetite. Amer. Jour. Sci.,
277:750–767.
Goldschmidt V. M. (1923). Geochemical laws of the distribution of the elements. Videnskaps.
Skrift. Mat.-Natl. Kl., 3:1–17.
Goldschmidt V. M., Barth T., Lunde G., and Zachariasen W. (1926). Geochemische Verteilung-
sgesetze der Elemente, VII: Die Gesetze die Kristallchemie. Det. Norske. Vid. Akad.
Oslo I, Mat. Natl. Kl., 2:1–117.
Goldsmith J. R. (1980). The melting and breakdown reactions of anorthite at high pressures
and temperatures. Amer. Mineral., 66:1183–1188.
832 / References
Goldsmith J. R. and Heard H. C. (1961). Subsolidus phase relations in the system CaCO3-
MgCO3. J. Geol., 69:45–74.
Goldsmith J. R. and Newton R. C. (1969). P-T-X relations in the system CaCO3-MgCO3 at
high temperatures and pressures. Amer. Jour. Sci., 267A:160–190.
Gordon R. G. and Kim Y. S. (1971). Theory of the forces between closed-shell atoms and mole-
cules. J. Chem. Phys., 56:3122–3133.
Gordy W. (1946). A new method of determining electronegativity from other atomic properties.
Phys. Rev., 69:604–607.
Gordy W. (1950). Interpretation of nuclear quadrupole couplings in molecules. J. Chem.
Phys., 19:792–793.
Gottardi G. (1972). I Minerali. Serie Geologia, Boringhieri, Bologna.
Gramaccioli C. M. and Filippini G. (1983). Lattice dynamical calculations for orthorhombic
sulfur: A non-rigid molecular model. Chem. Phys. Letters, 108:585–588.
Grandstaff D. E. (1977). Some kinetics of bronzite orthopyroxene dissolution. Geochim. Cos-
mochim. Acta, 41:1097–1103.
Grandstaff D. E. (1980). The dissolution rate of forsterite olivine from Hawaiian beach sand.
In Third International Symposium on Water-Rock Interaction: Proceedings, Alberta Re-
search Council, Edmonton.
Graves J. (1977). Chemical mixing in multicomponent solutions. In Thermodynamics in Geol-
ogy, D. E. Fraser, ed. Reidel, Dordrecht-Holland.
Grebenshchikov R. G., Romanov D. P., Sipovskii D. P., and Kosulina G. I. (1974). Study of a
germanate hydroxyamphibole. Zhur. Prikl. Khimii, 47:1905–1910.
Greegor R. B., Lytle F. W., Sanstrom D. R., Wang J., and Schultz P. (1983). Investigation of
TiO2-SiO2 glasses by X-ray absorption spectroscopy. J. Non. Cryst. Solids, 55:27–43.
Green D. H. and Hibberson W. (1970). The instability of plagioclase on peridotite at high pres-
sure. Lithos, 3:209–221.
Green E. J. (1970). Predictive thermodynamic models for mineral systems, I: Quasi-chemical
analysis of the halite-sylvite subsolidus. Amer. Mineral., 55:1692–1713.
Green S. H. and Gordon R. G. (1974). POTLSURF: A program to compute the interaction
potential energy surface between a closed-shell molecule and an atom. Quantum Chemis-
try Program Exchange No. 251, Indiana University.
Green T. H. (1994). Experimental studies of trace-element partitioning applicable to igneous
petrogenesis—Sedona 16 years later. Chem. Geol., 117:1–36.
Greenwood H. J. (1967). Wollastonite: Stability in H2O-CO2 mixtures and occurrence in a con-
tact metamorphic aureole near Salmo, British Columbia, Canada. Amer. Mineral.,
52:1669–1680.
Greenwood N. N. (1970). Ionic Crystals, Lattice Defects and Non-stoichiometry. New York:
Chemical Pub. Co.
Greig J. W. (1927). Immiscibility in silicate melts. Amer. Jour. Sci., 5th Ser., 13(73).: 1–44 and
13(74).: 133–154.
Greig J. W. and Barth T. F. W. (1938). The system Na2O-Al2O3-2SiO2 (nepheline, carnegieite).-
Na2O-Al2O3-6SiO2. (albite). Amer. Jour. Sci., 5th. ser., 35A:93–112.
Griffin W. L. and Mottana A. (1982). Crystal chemistry of clinopyroxenes form St. Marcel
manganese deposit, Val d’Aosta, Italy. Amer. Mineral., 67:568–586.
Grover J. (1977). Chemical mixing in multicomponent solutions: An introduction to the use of
Margules and other thermodynamic excess functions to represent non-ideal behaviour.
In Thermodynamics in Geology, D. G. Fraser, ed. D. Reidel, Dordrecht-Holland.
Grover J. E. and Orville T. M. (1969). The partitioning of cations between coexisting single
and multisite phases with application to the assemblage: Orthopyroxene-clinopyroxene
and orthopyroxene-olivine. Geochim. Cosmochim. Acta, 33:205–226.
Grundy H. D. and Ito J. (1974). The refinement of the crystal structure of a synthetic non-
stoichiometric Sr-feldspar. Amer. Mineral., 68:1319–1326.
Guggenheim E. A. (1937). Theoretical basis of Raoult’s Law. Trans. Faraday Soc., 33:205–226.
References / 833
Guggenheim E. A. (1952). Mixtures. Clarendon Press, Oxford.

Guggenheim S. and Bailey S. W. (1975). Refinement of the margarite structure in subgroup
symmetry. Amer. Mineral., 60:1023–1029.
Guggenheim S. and Bailey S. W. (1977). The refinement of zinnwaldite-IM in subgroup symme-
try. Amer. Mineral., 62:1158–1167.
Guggenheim S. and Bailey S. W. (1978). Refinement of the margarite structure in subgroup
symmetry: Correction, further refinement, and comments. Amer. Mineral., 63:186–187.
Guidotti C. V., Herd H. H., and Tuttle C. L. (1973). Composition and structural state of K-
feldspars from K-feldspars ⫹ sillimanite grade rocks in northwestern Maine. Amer.
Mineral., 58:705–716.
Guthrie F. (1884). On eutexia. Philos. Mag., 17–5:462–482.
Guy R. D., Chakrabarti C. L., and Schraumm L. L. (1975). The application of a simple chemi-
cal model of natural waters to metal fixation in particulate matter. Canadian Jour.
Chem., 53:661–669.
Haar L., Gallager J. G., and Kell G. S. (1979). Thermodynamic properties of fluid water. In
Contributions to the 9th Int. Conf. on Properties of Steam, Munich, Western Germany.
Haar L., Gallager J. G., and Kell G. S. (1984). NBS/NRC Steam Tables: Thermodynamic and
transport properties and computer programs for vapor and liquid states of water in SI units.
New York: Hemisphere Pub. Co., McGraw-Hill.
Haas I. L. Jr., Robinson G. R. Jr., and Hemingway B. S. (1981). Thermodynamic tabulations
for selected phases in the system CaO-Al2O3-SiO2-H2O at 101.325 kpa (1 atm). between
273.15 and 1800 K. J. Phys. Chem. Ref. Data, 10:576–669.
Haas J. L. (1971). Effect of salinity on the maximum thermal gradient of a hydrothermal system
at hydrostatic pressure. Econ. Geol., 66:940–946.
Haas J. L. Jr. and Fisher J. R. (1976). Simultaneous evaluation and correlation of thermody-
namic data. Amer. Jour. Sci., 276:525–545.
Haasen P. (1978). Physical Metallurgy. Cambridge University Press, London.
Hackler R. T. and Wood B. J. (1989). Experimental determination of Fe and Mg exchange
between garnet and olivine and estimation of Fe-Mg mixing properties in garnet. Amer.
Mineral., 74:994–999.
Hafner S. S. and Ghose S. (1971). Iron and magnesium distribution in cummingtonites. Zeit.
Kristallog., 133:301–326.
Häkli T. and Wright T. L. (1967). The fractionation of nickel between olivine and liquid as a
geothermometer. Geochim. Cosmochim. Acta, 31:877–884.
Halback H. E. and Chatterjee N. D. (1982). An empirical Redlich-Kwong-type equation state
for water to 1000°C and 200 kbar. Contrib. Mineral. Petrol., 79:337–345.
Hamilton D. L. and MacKenzie W. S. (1965). Phase equilibrium studies in the system
NaAlSiO4 (nepheline)-KAlSiO4 (kalsilite)-SiO2-H2O. Min. Mag., 34:214–231.
Hanson G. N. and Gast P. W. (1967). Kinetic studies in contact metamorphic zones. Geochim.
Harlow, G. E. and Brown G. E. Jr. (1980). Low albite: An X-ray and neutron diffraction study.
Harris P. G. (1957). Zone refining and the origin of potassic basalts. Geochim. Cosmochim.
Acta, 12:195–208.
Harrison T. M. (1981). Diffusion of 40Ar in hornblende. Contrib. Mineral. Petrol., 78:324–331.
Harrison T. M. (1983). Some observations on the interpretation of 40Ar/39Ar age spectra. Isot.
Geosci., 1:319–338.
Harrison T. M. and McDougall I. (1980). Investigations of an intrusive contact, northwest
Nelson, New Zealand, II: Diffusion of radiogenic and excess 40Ar in hornblende re-
vealed by 40Ar/39Ar age spectrum analysis. Geochim. Cosmochim. Acta, 44:2005–2020.
Harrison T. M. and McDougall I. (1981). Excess 40Ar in metamorphic rocks from Broken Hill,
New South Wales: Implications for 40Ar/39Ar age spectra and the thermal history of the
region. Earth Planet. Sci. Letters, 55:123–149.
834 / References
Harrison T. M. and McDougall I. (1982). The thermal significance of potassium feldspar K-Ar
ages inferred from 40Ar/39Ar age spectrum results. Geochim. Cosmochim. Acta, 46:1811–
1820.
Harrison W. J. (1977). An experimental study of the partitioning of samarium between garnet
and liquid at high pressures. In Papers Presented to the International Conference on Ex-
perimental Trace Elements Geochemistry, Sedona, Arizona.
Harrison W. J. (1978). Rare earth element partitionings between garnets, pyroxenes and melts
at low trace element concentration. Carnegie Inst. Wash. Yb., 77:682–689.
Harrison W. J. (1981). Partition coefficients for REE between garnets and liquids: Implications
of non-Henry’s law behavior for models of basalt origin and evolution. Geochim. Cos-
mochim. Acta, 45:1529–1544.
Harrison W. J. and Wood B. J. (1980). An experimental investigation of the partitioning of REE
between garnet and liquid with reference to the role of defect equilibria. Contrib. Min-
eral. Petrol., 72:145–155.
Hart R. (1970). Chemical exchange between seawater and deep ocean basalts. Earth. Planet.
Sci. Letters., 9:269–279.
Hart S. R. (1964). The petrology and isotopic-mineral age relations of a contact zone in the
Front Range, Colorado. J. Geol., 72:493–525.
Harvey K. B. and Porter G. B. (1976). Introduzione alla Chimica Fisica Inorganica. Piccin Edi-
tore, Padova.
Haselton H. T. and Newton R. C. (1980). Thermodynamics of pyrope-grossular garnets and
their stabilities at high temperatures and pressures. J. Geophys. Res., 85:6973–6982.
Haselton H. T. and Westrum E. F. Jr. (1980). Low-temperature heat capacities of synthetic
pyrope, grossular, and pyrope60grossular40. Geochim. Cosmochim. Acta, 44:701–709.
Haselton H. T., Hovis G. L., Hemingway B. S., and Robie R. A. (1983). Calorimetric investiga-
tion of the excess entropy of mixing in analbite-sanidine solid solutions: Lack of evi-
dence for Na, K short range order and implications for two feldspar thermometry. Amer.
Mineral., 68:398–413.
Hawthorne F. C. (1976). The crystal chemistry of the amphiboles, V: The structure and chemis-
try of arfvedsonite. Canadian Mineral., 346–356.
Hawthorne F. C. (1981a). Crystal chemistry of amphiboles. In Reviews in Mineralogy, vol. 9A,
P. H. Ribbe (series ed.), Mineralogical Society of America.
Hawthorne F. C. (1981b). Some systematics of the garnet structure. J. Solid State Chem.,
37:157–164.
Hawthorne F. C. and Grundy H. D. (1976). The crystal chemistry of the amphiboles, IV: X-ray
and neutron refinement of the crystal structure of tremolite. Canadian Mineral.,
14:334–345.
Hazen R. M. (1977). Effects of temperature and pressure on the crystal structure of ferromag-
nesian olivine. Amer. Mineral., 62:286–295.
Hazen R. M. and Burnham C. W. (1973). The crystal structure of 1 layer phlogopite and annite.
Hazen R. M. and Finger L. W. (1978). Crystal structure and compressibilities of pyrope and
grossular to 60 kbar. Amer. Mineral., 63:297–303.
Hazen R. M. and Finger L. W. (1979). Bulk-modulus volume relationship for cation-anion
polyhedra. J. Geophys. Res., 84:6723–6728.
Hazen R. M. and Finger L. W. (1982). Comparative Crystal Chemistry. New York: John Wiley.
Hazen R. M. and Prewitt C. T. (1977). Effects of temperature and pressure on interatomic
distances in oxygen-based minerals. Amer. Mineral., 62:309–315.
Hazen R. M. and Wones D. R. (1972). The effect of cation substitutions on the physical proper-
ties of trioctahedral micas. Amer. Mineral., 57:103–129.
Helgeson H. C. (1967). Solution chemistry and metamorphism. In Research in Geochemistry,
vol. 2, P. H. Abelson (series ed.). New York: John Wiley.
Helgeson H. C. (1969). Thermodynamics of hydrothermal system at elevated temperatures and
pressures. Amer. Jour. Sci., 267:729–804.
References / 835
Helgeson H. C. and Kirkham D. H. (1974). Theoretical prediction of thermodynamic behavior

of aqueous electrolytes at high pressures and temperatures, I: Summary of the thermo-
dynamic/electrostatic properties of the solvent. Amer. Jour. Sci., 274:1089–1198.
Helgeson H. C., Brown T. H., and Leeper R. H. (1976). Handbook of Theoretical Activity Dia-
grams Depicting Chemical Equilibria in Geologic Systems Involving an Aqueous Phase at
1 atm and O to 300 °C. Freeman, Cooper, San Francisco.
Helgeson H. C., Delany J., and Bird D. K. (1978). Summary and critique of the thermodynamic
properties of rock-forming minerals. Amer. Jour. Sci., 278A:1–229.
Helgeson H. C., Kirkham D. H., and Flowers G. C. (1981). Theoretical prediction of the ther-
modynamic behavior of aqueous electrolytes at high pressures and temperatures, IV:
Calculation of activity coefficients, osmotic coefficients and apparent molal and stan-
dard and relative partial molal properties to 600 °C and 5 kbar. Amer. Jour. Sci.,
281:1249–1516.
Helgeson H. C., Murphy W. M., and Aagaard P. (1984). Thermodynamic and kinetic con-
straints on reaction rates among minerals and aqueous solutions, II: Rate constants,
effective surface area, and the hydrolysis of feldspar. Geochim. Cosmochim. Acta.,
48:2405–2432.
Hem J. D. and Lind C. J. (1983). Nonequilibrium models for predicting forms of precipitated
manganese oxides. Geochim. Cosmochim. Acta, 47:2037–2046.
Hem J. D., Roberson C. E., and Fournier R. B. (1982). Stability of -MnOOH and manganese
oxide deposition from spring water. Water. Resour. Res., 18:563–570.
Hemingway B. S., Krupka K. M., and Robie R. A. (1981). Heat capacities of the alkali feld-
spars between 350 and 1000 K from differential scanning calorimetry, the thermody-
namic functions of alkali feldspars from 298.15 to 1400 K, and the reaction quartz ⫹
jadeite ⫽ albite. Amer. Mineral., 66:1202–1215.
Henderson L. M. and Kracek F. C. (1927). The fractional precipitation of barium and chro-
mates. J. Amer. Chem. Soc., 49:739–749.
Henderson P. (1982). Inorganic Geochemistry. Pergamon Press, Oxford.
Henley R. W. (1984). Chemical structure of geothermal systems. In Reviews in Economic Geol-
ogy, vol. 1, J. M. Robertson (series ed.), Society of Economic Geologists.
Hensen B. J. (1977). Cordierite garnet bearing assemblages as geothermometers and barome-
ters in granulite facies terranes. Tectonophysics, 43:73–88.
Hensen B. J. and Green D. H. (1973). Experimental study of the stability of cordierite and
garnet in pelitic compositions at high pressures and temperatures. Contrib. Mineral.
Petrol., 38:151–166.
Hensen B. J., Schmid R., and Wood B. J. (1975). Activity-composition relationship for pyrope-
grossular garnet. Contrib. Mineral. Petrol., 54:161–166.
Herman F. (1959). Lattice vibrational spectrum of germanium. J. Phys. Chem. Solids, 8:405–418.
Hertoghen J. and Gijbels K. (1976). Calculation of trace element fractionation during partial
melting. Geochim. Cosmochim. Acta, 40:313–322.
Herzberg C. T. (1978a). The bearing of phase equilibria in simple and complex systems on the
origin and evolution of some garnet websterites. Contrib. Mineral. Petrol., 66:375–382.
Herzberg C. T. (1978b). Pyroxene geothermometry and geobarometry: Experimental and ther-
modynamic evaluation of some subsolidus phase relations involving pyroxenes in the
system CaO-MgO-Al2O3-SiO2. Geochim. Cosmochim. Acta, 42:945–957.
Herzberg C. T. and Lee S. M. (1977). Fe-Mg cordierite stability in high grade pelitic rocks
based on experimental, theoretical, and natural observations. Contrib. Mineral. Pet-
rol., 63:175–193.
Hess P. C. (1971). Polymer models of silicate melts. Geochim. Cosmochim. Acta, 35:289–306.
Hess P. C. (1977). Structure of silicate melts. Canadian Mineral., 15:162–178.
Hewitt D. A. and Wones D. R. (1981). The annite-sanidine-magnetite equilibrium GAC-MAG
Joint Annual Meeting, Calgary, Abstracts, 6:A-66.
Hewitt D. A. and Wones D. R. (1984). Experimental phase relations of the micas. In Reviews
in Mineralogy. vol. 13, P. H. Ribbe (series ed.), Mineralogical Society of America.
836 / References
Hillert M. (1980). Empirical methods of predicting and representing thermodynamic properties

of ternary solution phases. CALPHAD, 4:1–12.
Himmelbau D. M. (1960). Solubilities of inert gases in water. J. Chem. Eng. Data, 5:10–15.
Hinrichsen T. and Schurmann K. (1971). Synthese und stabilität von glimmern im System CaO-
Na2O-Al2O3-SiO2-H2O. Forschr. Mineral., 49:21.
Hinshelwood C. N. (1951). The Structure of Physical Chemistry Oxford University Press,
London.
Hinze J. A. and Jaffe H. H. (1962). Electronegativity, I: Orbital electronegativity of neutral
atoms. J. Amer. Chem. Soc., 84:540–546.
Hinze J. A., Whitehead M. A., and Jaffe H. H. (1963). Electronegativity, II: Bond and orbital
electronegativity. J. Inorg. Chem., 38:983–984.
Hirschfelder J. O., Curtiss C. F., and Bird R. B. (1954). The Molecular Theory of Gases and
Liquids. New York: John Wiley.
Hirschmann M. (1991). Thermodynamics of multicomponent olivines and the solution proper-
ties of (Ni,Mg,Fe)2SiO4 and (Ca,Mg,Fe)2SiO4 olivines. Amer. Mineral., 76:1232–1248
Hirt B., Herr W., and Hoffmeister W. (1963). Age determinations by the rhenium-
osmium method. In Radioactive Dating, International Atomic Energy Agency, Vienna.
Hoefs J. (1980). Stable Isotope Geochemistry. Berlin-Heidelberg-New York: Springer-Verlag.
Hofman A. W. and Giletti B. J. (1970). Diffusion of geochronologically important nuclides in
minerals under hydrothermal conditions. Eclogae Geol. Helv., 63:141–150.
Holdaway M. J. (1971). Stability of andalusite and the aluminium silicate phase diagram. Amer.
Jour. Sci., 271:97–131.
Holdaway M. J. (1972). Thermal stability of Al-Fe epidote as a function of f O2 and Fe content.
Holdaway M. J. and Lee S. M. (1977). Fe-Mg cordierite stability in high grade pelitic rocks
based on experimental, theoretical and natural observations. Contrib. Mineral. Petrol.,
63:175–198.
Holdaway M. J. and Mukhopadhyay B. (1993). A reevaluation of the stability relations of anda-
lusite: Thermochemical data and phase diagram for the aluminum silicates. Amer. Min-
eral., 78:298–315.
Holder J. and Granato A. V. (1969). Thermodynamic properties of solid containing defects.
Phys. Rev., 182:729–741.
Holgate N. (1954). The role of liquid immiscibility in igneous petrogenesis. J. Geol., 62:439–306.
Holland H. D. (1972). Granites-solutions and base metal deposits. Econ. Geol., 67:281–301.
Holland H. D. (1978). The Chemistry of Atmosphere and Oceans. New York: Wiley Interscience.
Holland T. J. B. (1989). Dependence of entropy on volume for silicate and oxide minerals: A
review and predictive model. Amer. Mineral., 74:5–13.
Holland T. J. B. and Powell R. (1990). An enlarged and updated internally consistent thermo-
dynamic dataset with uncertainties and correlations: The system K2O-Na2O-CaO-
MgO-MnO-FeO-Fe2O3-Al2O3-TiO2-SiO2-C-H2-O2. J. Metamorphic Geol., 8:89–124.
Holloway J. R. (1977). Fugacity and activity of molecular species in supercritical fluids. In
Thermodynamics in Geology, D. G. Fraser, ed. Reidel, Dordrecht.
Holloway J. R. (1987). Igneous fluids. In Reviews in Mineralogy, vol. 17, P. H. Ribbe (series
ed.), Mineralogical Society of America.
Holm J. L. and Kleppa O. J. (1968). Thermodynamics of the disordering process in albite.
Hoover J. D. (1978). The distribution of samarium and tulium between plagioclase and liquid in
the system An-Di and Ab-An-Di at T ⫽ 1300°C. Carnegie Inst. Wash. Yb., 77:703–706.
Horan M. F., Morgan J. W., Grauch R. I., Coveney R. M. Jr., Murowchick J. B., and Hulbert
L. J. (1994). Rhenium and osmium isotopes in black shales and Ni-Mo-PGE-rich sulfide
layers, Yukon Territory, Canada, and Hunan and Guizhou provinces, China. Geochim.
Horita J., Wesolowski D. J., and Cole D. R. (1993a). The activity-composition relationship of
References / 837
oxygen and hydrogen isotopes in aqueous salt solutions, I: Vapor-liquid water equilibra-
tion of single salt solutions from 50 to 100°C. Geochim. Cosmochim. Acta, 57:2797–2817.
Horita J., Cole D. R., and Wesolowski D. J. (1993b). The activity-composition relationship of
oxygen and hydrogen isotopes in aqueous salt solutions, II: Vapor-liquid water equili-
bration of mixed salt solutions from 50 to 100°C and geochemical implications. Geo-
Hovis G. L. (1974). A solution calorimetric and X-ray investigation of Al-Si distribution in
monoclinic potassium feldspars. In The feldspars, W. S. MacKenzie and J. Zussman, eds.
Manchester: Manchester University Press.
Hovis G. L. and Waldbaum D. R. (1977). A solution calorimetric investigation of K-Na mixing
in a sanidine-analbite join-exchange series. Amer. Mineral., 53:1965–1979.
Hsu L. C. (1968). Selected phase relationships in the system Al-Mn-Fe-Si-O-H: A model for
garnet equilibria. J. Petrol., 9:40–83.
Huckenholtz H. G. and Knittel D. (1975). Uvarovite: Stability of uvarovite-grossularite solid
solutions at low pressure. Contrib. Mineral. Petrol., 49:211–232.
Huckenholtz H. G., Schairer J. F., and Yoder H. S. Jr. (1969). Synthesis and stability of ferri-
diopside. Mineral. Soc. Amer. Spec. Paper, 2:163–177.
Huebner J. S. (1982). Phyroxene phase equilibria at low pressure. In Reviews in Mineralogy, vol.
7 (2d ed.), P. H. Ribbe (series ed.), Mineralogical Society of America.
Huebner J. S. and Papike J. J. (1970). Synthesis and crystal chemistry of sodium-potassium
richterite, (Na, K)NaCaMg5Si8O22(OH,F)2. Amer. Mineral., 55:1973–1993.
Huebner J. S. and Sato M. (1970). The oxygen fugacity-temperature relationships of manganese
oxide and nickel oxide buffers. Amer. Mineral., 55:934–952.
Huebner J. S., Lipin B. R., and Wiggins L. B. (1976). Partitioning of chromium between silicate
crystal and melts. Proc. Seventh Lunar Sci. Conf., 1195–1220.
Huheey J. E. (1975). Inorganic Chemistry: Principles of Structure and Reactivity. New York:
Harper and Row.
Hurd D. C. (1972). Factors affecting solution rate of biogenic opal in sea water. Earth Planet.
Sci. Letters, 15:411–417.
Hutner R. A., Rittner E. S., and Du Pré F. K. (1949). Concerning the work of polarization in
ionic crystals of the NaCl type, II: Polarization around two adjacent charges in the rigid
lattice. J. Chem. Phys., 17:204–208.
Iczkowski R. P. and Margrave J. L. (1961). Electronegativity. J. Amer. Chem. Soc., 83: 3547–
3551.
Iiyama J. T. (1972). Fixation des elements alcalinoterreux, Ba, Sr et Ca dans les feldspaths;
etude experimentale. Proc. 24 Congres Geol. Internat., Section 10, 122–130.
Iiyama J. T. (1973). Behavior of trace elements in feldspars under hydrothermal conditions. In
The Feldspars, W. S. MacKenzie and J. Zussmann, eds. Manchester: Manchester Univer-
sity Press.
Iiyama J. T. (1974). Substitution, deformation locale de la maille et equilibre de distribution
des elements en tracs entre silicates et solution hydrothermale. Bul. Soc. Fr. Mineral.
Cristallog., 97:143–151.
Irefune T., Ohtani E., and Kumazawa I. (1982). Stability field of knorringite at high pressure
and its application to the occurrence of Cr-rich pyrope in the upper mantle. Phys. Earth
Planet. Interiors, 27:263–272.
Irving A. J. (1974). Geochemical and high-pressure experimental studies of garnet pyroxenite
and granulite xenoliths from the Delegate basaltic pipes, Australia. J. Petrol., 15:1–40.
Irving A. J. (1978). A review of experimental studies of crystal/liquid trace element parti-
tioning. Geochim. Cosmochim. Acta. 42:743–770.
Isaak D. G. and Graham E. K. (1976). The elastic properties of an almandine-spessartite gar-
net and elasticity in the garnet solid solutions series. J. Geophys. Res., 81:2483–2489.
Isaak D. G., Anderson L., and Oda H. (1992). High-temperature thermal expansion and elas-
ticity of calcium-rich garnets. Phys. Chem. Minerals, 19:106–120.
838 / References
Ivanov I. P., Potekhin V. Y., Dmitriyenko L. T., and Beloborodov S. M. (1973). An experimen-
tal study of T and P conditions of equilibrium of reaction: Muscovite-Kfeldspar ⫹ co-
rundum ⫹ H2O at P(H2O)⬍P(total). Geokhimiya, 9:1300–1310.
Jäger E. and Hunziker J. C. (1979). Lectures in Isotope Geology Berlin-Heidelberg-New York:
Springer-Verlag.
Jäger E., Niggli E., and Wenk E. (1967). Alterbestimmungen an Glimmern der Zentralalpen.
Beitr. Geol. Karte Schweitz, NF134, Bern.
Jahn H. A. and Teller E. (1937). Stability of polyatomic molecules in degenerate electronic
states. Proc. Roy. Soc. London, A161:220–235.
JANAF (1974–1975). Thermochemical Tables U.S. Department of Commerce, National Bureau
of Standards, Institute for Applied Technology, Supplements.
Jaoul O., Froidevaux C., Durham W. B., and Michaud M. (1980). Oxygen self-diffusion in
forsterite: Implications for the high temperature creep mechanism. Earth. Planet. Sci.
Letters, 47:391–397.
Jaoul O., Poumellec M., Froidevaux C., and Havette A. (1981). Silicon diffusion in forsterite:
A new constraint for understanding mantle deformation. In Anelasticity in the Earth,
F. D. Stacey et al., eds. Godyn Ser., vol. 4, AGU, Washington, D.C.
Jenne E. A. (1981). Geochemical modeling: A review. In Waste/Rock Interactions Technology
Program, Pacific Northwest Laboratory, Publication 3574.
Jenne E. A., Girovin D. C., Ball J. W., and Burchard J. M. (1978). Inorganic speciation of silver
in natural waters-fresh to marine. In Environmental Impacts of Nucleating Agents Used
in Weather Modification Programs, D. A. Klein, ed. Dowden-Hutchison and Ross, Stras-
burg, Pennsylvania.
Jensen B. (1973). Patterns of trace elements partitioning. Geochim. Cosmochim. Acta, 37:2227–
2242.
Johnson J. W. and Norton D. (1991). Critical phenomena in hydrothermal systems: State, ther-
modynamic, electrostatic, and transport properties of H2O in the critical region. Amer.
Jour. Sci., 291:541–648.
Johnson J. W., Oelskers E. H., and Helgeson H. C. (1991). SUPCRT92: A Software Package for
Calculating the Standard Molal Thermodynamic Properties of Minerals, Gases, Aqueous
Species, and Reactions from 1 to 5000 bars and 0° to 1000°C Earth Sciences Department,
L-219, Lawrence Livermore National Laboratory, Livermore, California.
Johnston W. D. (1964). Oxidation-reduction equilibria in iron-containing glass. J. Amer. Ceram.
Soc., 47:198–201.
Johnston W. D. (1965). Oxidation-reduction equilibria in molten Na2O-2SiO2 glass. J. Amer.
Ceram Soc., 48:184–190.
Johnston W. D., and Chelko A. (1966). Oxidation-reduction equilibria in molten Na2O-2SiO2
glass in contact with metallic copper and silver. J. Amer. Ceram. Soc., 49:562–564.
Jones J. H. (1995). Experimental trace element partitioning. In Rock Physics and Phase Rela-
tions. A Handbook of Physical Constants, T. J. Ahrens, ed., AGU, Washington, D.C.
Jost W. (1960). Diffusion in Solid State Physics. New York: Academic Press.
Kamensky I. L., Tokarev I. V., and Tolstikhin I. N. (1991). 3H-3He dating: A case for mixing
of young and old groundwaters. Geochim. Cosmochim. Acta, 55:2895–2899.
Karpinskaya T. B., Ostrovsky I. A., and Yevstigneyeva O. (1983). Synthetic pure iron garnet
skiagite. Internat. Geol. Rev., 25:1129–1130.
Kato K. and Nukui A. (1976). Die Kristallstructur des monoclinen Tief-trydymits. Acta Cryst.,
B32:2486–2491.
Katz A. and Matthews A. (1977). The dolomitization of CaCO3: An experimental study at
252–295°C. Geochim. Cosmochim. Acta, 41:297–308.
Kawasaki T. and Matsui Y. (1977). Partitioning of Fe2⫹ and Mg2⫹ between olivine and garnet.
Earth. Planet. Sci. Letters, 37:159–166.
Kelley K. K. (1960). Contributions to the data on theoretical metallurgy, XIII: High tempera-
ture heat content, heat capacity and entropy data for the elements and inorganic com-
pounds. U.S. Bur. Mines Bull., 584, 232 pp.
References / 839
Kern D. M. (1960). The hydration of carbon dioxide. J. Chem. Educ., 37:14–23.

Kerrick D. M. (1972). Experimental determination of muscovite ⫹ quartz stability with P(H2O)
⬍ P (total). Amer. Jour. Sci., 272:946–958.
Kerrick D. M. and Darken L. S. (1975). Statistical thermodynamic models for ideal oxide and
silicate solid solutions, with applications to plagioclase. Geochim. Cosmochim. Acta,
39:1431–1442.
Kerrick D. M. and Jacobs G. K. (1982). A modified Redlich-Kwong equation for H2O, CO2
and H2O-CO2 mixtures at elevated pressures and temperatures. Amer. Jour. Sci.,
281:735–767.
Kieffer S. W. (1979a). Thermodynamics and lattice vibrations of minerals, 1: Mineral heat ca-
pacities and their relationships to simple lattice vibrational models. Rev. Geophys. Space
Phys., 17:1–19.
Kieffer S. W. (1979b). Thermodynamics and lattice vibrations of minerals, 2: Vibrational char-
acteristics of silicates. Rev. Geophys. Space Phys., 17:20–34.
Kieffer S. W. (1979c). Thermodynamics and lattice vibrations of minerals, 3: Lattice dynamics
and an approximation for minerals with application to simple substances and frame-
work silicates. Rev. Geophys. Space Phys., 17:35–59.
Kieffer S. W. (1980). Thermodynamics and lattice vibrations of minerals, 4: Application to
chain and sheet silicates and orthosilicates. Rev. Geophys. Space Phys., 18:862–886.
Kieffer S. W. (1982). Thermodynamics and lattice vibrations of minerals, 5: Applications to
phase equilibria, isotopic fractionation, and high-pressure thermodynamic properties.
Rev. Geophys. Space Phys., 20:827–849.
Kieffer S. W. (1985). Heat capacity and entropy: Systematic relations to lattice vibrations. In
Reviews in Mineralogy, vol. 14, P. H. Ribbe (series ed.), Mineralogical Society of
America.
Kieffer S. W. and Navrotsky A. (1985). Scientific perspective. In Reviews in Mineralogy, vol. 14,
P. H. Ribbe (series ed.)., Mineralogical Society of America.
Kielland J. (1937). Individual activity of ions in aqueous solutions. J. Amer. Chem. Soc.,
59:1675–1678.
Kim K. and Burley B. J. (1971). Phase equilibria in the system NaAlSi3O8-NaAlSiO4-H2O with
special emphasis on the stability of analcite. Canadian Jour. Earth Sci., 8:311–337.
Kim K. Y., Chabildas L. C., and Ruoff A. L. (1976). Isothermal equations of state for lithium
fluoride. J. Appl. Phys., 47:2862–2866.
Kirby S. H. and Wegner M. W. (1978). Dislocation substructure of mantle-derived olivine as
revealed by selective chemical etching and transmission electron microscopy. Phys.
Chem. Minerals, 3:309–330.
Kirsten T. (1968). Incorporation of rare gases in solidifying enstatite melts. J. Geophys. Res.,
73:2807–2810.
Kiseleva I. A. (1977). Gibbs free energy of formation of calcium garnets. Geokhimiya,
5:705–715.
Kitayama K. and Katsura T. (1968). Activity measurements in orthosilicate and metasilicate
solid solutions, I: Mg2SiO4-Fe2SiO4 and MgSiO3-FeSiO3 at 1204 °C. Bull. Chem. Soc.
Jpn., 41:1146–1151.
Kittel C. (1971). Introduction to Solid State Physics. New York: John Wiley.
Kittel C. (1989). How to define entropy. Nature, 339:170.
Klein C. and Bricker O. P. (1977). Some aspects of the sedimentary and diagenetic environment
of proterozoic banded iron formation. Econ. Geol., 72:1457–1470.
Klein C. Jr. (1964). Cummingtonite-grunerite series: A chemical optical and X-ray study. Amer.
Mineral., 49:963–982.
Klein C. Jr. and Ito J. (1968). Zincian and manganoan amphiboles from Franklin, New Jersey.
Amer. Mineral., 53:1264–1275.
Kohler F. (1960). Zur Berechnung der thermodynamischen Daten eines ternären Systems aus
den zugehörigen binären Systemen. Monatsh. Chem., 96:1228–1251.
Köhler T. P. and Brey G. P. (1990). Calcium exchange between olivine and clinopyroxene cali-
840 / References
brated as a geothermobarometer for natural peridotites from 2 to 60 kb with applica-

tions. Geochim. Cosmochim. Acta, 54:2375–2388.
Kohlstedt D. L. and Goetze C. (1974). Low-stress high-temperature creep in olivine single crys-
tals. J. Geophys. Res., 79:2045–2051.
Kohlstedt D. L. and Vander Sande J. B. (1973). Transmission electron microscopy investigation
of defects microstructure of four natural orthopyroxenes. Contrib. Mineral. Petrol.,
42:169–180.
Kostov I. (1975). Crystal chemistry and classification of silicate minerals. Geochem. Mineral.
Petrol., 1:5–14.
Kotelnikov A. R., Bychkov A. M., and Chernavina N. I. (1981). Experimental study of calcium
distribution in granitoid plagioclase and water-salt fluid at 700°C and Pf1 ⫽ 1000 atm.
Geokhimiya, 21:707–721.
Koziol A. M. (1990). Activity-composition relationship of binary Ca-Fe and Ca-Mn garnets
determined by reversed, displaced equilibrium experiments. Amer. Mineral., 75:319–327.
Kretz R. (1982). Transfer and exchange equilibria in a portion of the pyroxene quadrilateral as
deduced from natural and experimental data. Geochim. Cosmochim. Acta, 46:411–422.
Kröger F. A. (1964). The Chemistry of Imperfect Crystals. North Holland Pub. Co., Am-
sterdam.
Krogh E. J. (1988). The garnet-clinopyroxene Fe-Mg geothermometer—a reinterpretation of
existing experimental data. Contrib. Mineral. Petrol., 99:44–48.
Kroll H. and Ribbe P. H. (1983). Lattice parameters, composition and Al, Si order in alkali
feldspars. In Reviews in Mineralogy, vol. 2 (2d ed.), P. H. Ribbe (series ed.), Mineralogi-
cal Society of America.
Kroll H., Bambauer H. U., and Schirmer U. (1980). The high albite-monalbite and analbite-
monalbite transitions. Amer. Mineral., 65:1192–1211.
Kumar M. D. (1987). Cation hydrolysis and the regulation of trace metal composition in seawa-
ter. Geochim. Cosmochim. Acta, 51:2137–2145.
Kurkjian C. R. and Russel L. E. (1958). Solubility of water in molten alkali silicates. Jour. Soc.
Glass Technol., 42:130–144.
Ladd M. F. C. (1979). Structure and Bonding in Solid State Chemistry. New York: Ellis Har-
wood Limited.
Ladd M. F. C. and Lee W. H. (1958). Solubility of some inorganic halides. Trans. Faraday
Soc., 54:34–39.
Ladd M. F. C. and Lee W. H. (1959). Calculation of lattice energies. J. Inorg. Nucl. Chem.,
11:264–271.
Ladd M. F. C. and Lee W. H. (1964). Lattice energy and related topics. In Progress in Solid
State Chemistry, vol. 1, H. Reiss, ed. New York: MacMillan.
Lagache M. (1971). Etude experimental de la rertition du cesium entre les felspaths sodi-
potassiques et des solutions hydrothermales a 700°C, 1 kb. C. R. Acad. Sci. Ser. D.
272:1328–1330.
Lagache M. and Sabatier G. (1973). Distribution des elements Na, K, Rb et Cs a l’etat de
trace entre feldspaths alcalins et solutions hydrothermales a 650°C, 1 kbar: Données
experimentales et interpretation thermodynamique. Geochim. Cosmochim. Acta, 37:
2617–2640.
Lager G. A. and Meagher D. M. (1978). High temperature structural study of six olivines.
Landau L. D. and Lifshitz E. M. (1980). Statistical Physics Pergamon Press, Oxford.
Landé A. (1920). The magnitude of the atom. Z. Physik, 1:191–197.
Lasaga A. C. (1979). Multicomponent exchange and diffusion in silicates. Geochim. Cos-
Lasaga A. C. (1980). Defects calculations in silicates: Olivine. Amer. Mineral., 65:1237–1248.
Lasaga A. C. (1981a). Rate laws of chemical reactions. In Reviews in Mineralogy, vol. 8, P. H.
Ribbe (series ed.), Mineralogical Society of America.
References / 841
Lasaga A. C. (1981b). Transition state theory. In Reviews in Mineralogy, vol. 8, P. H. Ribbe

Lasaga A. C. (1981c). The atomistic basis of kinetics: Defects in minerals. In Reviews in Miner-
alogy, vol. 8, P. H. Ribbe (series ed.), Mineralogical Society of America.
Lasaga A. C. (1983). Geospeedometry: An extension of geothermometry. In Advances in Physi-
cal Geochemistry, vol. 3, S. K. Saxena (series ed.). Berlin-Heidelberg-New York:
Springer-Verlag.
Lasaga A. C. (1984). Chemical kinetics of water-rock interactions. J. Geophys. Res., 89:4009–
4025.
Lasaga A. C. and Cygan R. T. (1982). Electronic and ionic polarizabilities of silicate minerals.
Lasaga A. C. and Kirkpatrick R. J. (1981). Kinetics of Geochemical processes. In Reviews in
Mineralogy, vol. 8, P. H. Ribbe (series ed.), Mineralogical Society of America.
Lasaga A. C., Richardson S. M., and Holland H. D. (1977). The mathematics of cation diffu-
sion and exchange between silicate minerals during retrograde metamorphism. In Ener-
getics of Geological Processes, S. K. Saxena and S. Bhattacharji, eds. New York:
Springer-Verlag.
Lattard D. and Schreyer W. (1983). Synthesis and stability of the garnet calderite in the system
Fe-Mn-Si-O. Contrib. Mineral. Petrol., 84:199–214.
Leeman W. P. and Lindstrom D. J. (1978). Patritioning of Ni2⫹ between basaltic and synthetic
melts and olivines—an experimental study. Geochim. Cosmochim. Acta, 42:801–816.
Leitner B. J., Weidner D. J., and Liebermann R. C. (1980). Elasticity of single crystal pyrope
and implications for garnet solid solution series. Phys. Earth Planet. Interiors., 22:111.
Le Page Y. and Donnay G. (1976). Refinement of the crystal structure of low-quartz. Acta
Cryst., B32:2456–2459.
Le Page Y., Calvert L. D., and Gabe E. J. (1980). Parameter variation in low-quartz between
94 and 298 K. Phys. Chem. Solids, 41:721–725.
Lerman A. (1979). Geochemical Processes: Water and Sediment Environments. New York:
John Wiley.
Levelt Sengers J. M. H., Kamgar-Parsi B., Balfour F. W., and Sengers J. V. (1983). Thermody-
namic properties of steam in the critical region. J. Phys. Chem. Ref. Data, 12:1–28.
Levitskii V. A., Golovanova Yu. G., Popov S. G., and Chentsov V. N. (1975). Thermodynamics
of binary oxide system: Thermodynamic properties of nickel orthosilicate. Russ. J. Phys.
Chem., 49:971–974.
Lewis G. N. (1916). The atom and the molecule. J. Amer. Chem. Soc., 38:762–785.
Lewis G. N. and Randall M. (1921). The activity coefficient of strong electrolytes. J. Amer.
Chem. Soc., 43:1112–1154.
Lewis G. N. and Randall M. (1970). Termodinamica. Leonardo Edizioni Scientifiche, Roma.
Liebau F. (1982). Classification of silicates. In Reviews in Mineralogy. vol. 5, P. H. Ribbe (series
Liebermann R. C. and Ringwood A. E. (1976). Elastic properties of anorthite and the nature
of the lunar crust. Earth Planet. Sci. Letters, 31:69–74.
Lin T. H. and Yund R. A. (1972). Potassium and sodium self-diffusion in alkali feldspar. Con-
trib. Mineral. Petrol., 34:177–184.
Lin Y. C. and Bailey S. W. (1984). The crystal structure of paragonite—2M1. Amer. Mineral.,
19:122–127.
Lindner M., Leich D. A., Russ G. P., Bazan J. M., and Borg R. J. (1989). Direct determination
of the half-life of 187Re. Geochim. Cosmochim. Acta, 53:1597–1606.
Lindsley D. H. (1981). The formation of pigeonite on the join hedembergite-ferrosilite at 11, 5
and 15 kbar: Experiments and a solution model. Amer. Mineral., 66:1175–1182.
Lindsley D. H. (1982). Phase equilibria of pyroxenes at pressure ⬎ 1 atmosphere. In Reviews
in Mineralogy, vol. 7 (2d ed.), P. H. Ribbe (series ed.), Mineralogical Society of America.
Lindsley D. H. (1983). Pyroxene thermometry. Amer. Mineral., 68:477–493.
842 / References
Lindsley D. H. and Anderson D. J. (1982). A two pyroxene thermometer. Proc. 13th Lunar
Planet. Sci. Conf., 887–906.
Lindsley D. H. and Dixon S. A. (1976). Diopside-enstatite equilibria at 850 to 1400°C, 5 to 35
kbar. Amer. Jour. Sci., 276:1285–1301.
Lindsley D. H., Grover J. E., and Davidson P. M. (1981). The thermodynamics of the
Mg2Si2O6-CaMgSi2O6 join: A review and a new model. In Advances in Physical Geo-
chemistry, vol. 1, S. K. Saxena (series ed.), New York: Springer-Verlag.
Lindstrom D. J. (1976). Experimental study of the partitioning of the transition metals between
clinopyroxene and coexisting silicate liquids. Ph.D. diss., University of Oregon.
Lindstrom D. J. and Weill D. F. (1978). Partitioning of transition metals between diopside and
coexisting silicate liquids, I: Nickel, cobalt, and manganese. Geochim. Cosmochim.
Acta, 42:801–816.
Liu K. K. and Epstein S. (1984). The hydrogen isotope fractionation between kaolinite and
water. Chem. Geol. (Isot. Geosci. Sec.)., 2:335–350.
Liu L. and Bassett A. (1986). Elements oxides and silicates. In Oxford Monographs on Geology
and Geophysics, vol. 4, P. Allen, H. Charnock, E. R. Oxburg, and B. J. Skinner, eds. New
York: Oxford University Press.
Loewenstein W. (1954). The distribution of aluminum in the tetrahedra of silicates and alumi-
nates. Amer. Mineral., 39:92–96.
London F. (1930). Properties and application of molecular forces. Z. Phys. Chem., B11:
222–251.
Long P. E. (1978). Experimental determination of partition coefficients for Rb, Sr and Ba be-
tween alkali feldspar and silicate liquid. Geochim. Cosmochim. Acta, 42:833–846.
Luck J. M. and Allègre C. J. (1983). 187Re-187Os systematics in meteorites and cosmological
consequences. Nature, 302:130–132.
Lumpkin G. R. and Ribbe P. H. (1983). Composition, order-disorder and lattice parameters of
olivines: Relationships in silicate, germanate, beryllate, phosphate and borate olivine.
Lux G. (1987). The behavior of noble gases in silicate liquids: Solution, diffusion, bubbles and
surface effects, with application to natural samples. Geochim. Cosmochim. Acta,
51:1549–1560.
Lux H. Z. (1939). Acid and bases in fused salt bath: The determination of oxygen-ion concen-
tration. Z. Elektrochem., 45:303–309.
Mackrodt W. C. and Stewart R. F. (1979). Defect properties of ionic solids, II: Point defect
energies based on modified electron gas potentials. J. Phys., 12C:431–449.
Mah A. D. and Pancratz L. B. (1976). Thermodynamic properties of nickel and its inorganic
compounds. U.S. Bur. Mines Bull., 668:1–125.
Maier C. G. and Kelley K. K. (1932). An equation for the representation of high temperature
heat content data. J. Amer. Chem. Soc., 54:3243–3246.
Manacorda T. (1985). Elementi di Meccanica Ondulatoria. Istituto di Matematiche Applicate
Ulisse Dini, Pisa.
Mantoura R. C. F., Dickson A., and Riley J. P. (1978). The complexation of metals with humic
materials in natural waters. Estuarine Coastal Mar. Sci., 6:387–408.
Maresch M. V., Mirwald P. W., and Abraham K. (1978). Nachweis einer Mischlüke in der Olivi-
nreihe Forsterit (Mg2SiO4)-Tephroit (Mn2SiO4). Fortschr. Miner., 56:89–90.
Marshall B. D. and De Paolo D. J. (1982). Precise age determination and petrogenetic studies
using the K-Ca method. Geochim. Cosmochim. Acta, 46:2537–2545.
Marshall W. L. and Franck E. U. (1981). Ion product of water substance, 0–1000°C, 1–10,000
bars: New international formulation and its background. J. Phys. Chem. Ref. Data,
10:295–303.
Martin R. F. (1970). Cell parameters and infra-red absorption of synthetic high to low albites.
Martin R. F. (1974). Controls of ordering and subsolidus phase relations in the alkali feldspars.
References / 843
In The feldspars, W. S. MacKenzie and J. Zussman, eds. Manchester: Manchester Uni-

versity Press.
Mason R. A. (1982). Trace element distributions between the perthite phases of alkali feldspars
from pegmatites. Min. Mag., 45:101–106.
Masse D. P., Rosèn E., and Muan A. (1966). Activity composition relations in Co2SiO4-Fe2SiO4
solid solution at 1180°C. J. Amer. Ceram. Soc., 49:328–329.
Masson C. R. (1965). An approach to the problem of ionic distribution in liquid silicates. Proc.
Roy. Soc. London, A287:201–221.
Masson C. R. (1968). Ionic equilibrium in liquid silicates. J. Amer. Ceram. Soc., 51:134–143.
Masson C. R. (1972). Thermodynamics and constitution of silicate slags. Jour. Iron Steel
Inst., 210:89–96.
Matsui Y. and Banno S. (1962). Intercrystalline exchange equilibrium in silicate solid solutions.
Proc. Japan Acad., 41:461–466.
Matsui Y. and Nishizawa O. (1974). Iron magnesium exchange equilibrium between coexisting
synthetic olivine and orthopyroxene. Amer. Jour. Sci., 267:945–968.
Matsui Y. and Syono Y. (1968). Unit cell dimension of some synthetic olivine group solid solu-
tions. Geochem. Jour., 2:51–59.
Matthews A. (1980). Influences of kinetics and mechanism in metamorphism: A study of albite
crystallization. Geochim. Cosmochim. Acta, 44:387–402.
Mattioli G. S. and Bishop F. (1984). Experimental determination of chromium-aluminum mix-
ing parameter in garnet. Geochim. Cosmochim. Acta, 48:1367–1371.
Mayer J. G. (1933). Dispersion and polarizability and the Van Der Waals potential in the alkali
halides. J. Chem. Phys., 1:270–279.
McCallister R. H. and Yund R. A. (1977). Coherent exsolution in Fe-free pyroxenes. Amer.
Mineral., 62:721–726.
McCallum I. S. and Charette M. P. (1978). Zr and Nb partition coefficients: Implications for
the genesis of mare basalts, KREEP and sea floor basalts. Geochim. Cosmochim. Acta,
42:859–870.
McClure D. S. (1957). The distribution of transition metal cations in spinels. J. Phys. Chem.
Solids, 3:311–317.
McConnell D. (1966). Propriétes physique des grenats:calcul de la dimension de la maille unite
a partir de la composition chimique. Geochim. Cosmochim. Acta, 31:1479–1487.
McDougall I. and Harrison T. M. (1988). Geochronology and Thermochronology by the
40
Ar/39Ar Method. Oxford Monographs on Geology and Geophysics, no. 9. New York:
Oxford University Press.
McGregor I. D. (1974). The system MgO-Al2O3-SiO2: Solubility of Al2O3 in enstatite for spinel
and garnet peridotite composition. Amer. Mineral., 59:110–119.
McIntire W. L. (1963). Trace element partition coefficients—a review of theory and applica-
tions to geology. Geochim. Cosmochim. Acta, 27:1209–1269.
McKay G. A. and Weill D. F. (1976). Petrogenesis of KREEP. Proc. Seventh Lunar Sci. Conf.,
2427–2447.
McMillan P. (1985). Vibrational spectroscopy in the mineral sciences. In Reviews in Mineralogy,
vol. 14, P. H. Ribbe (series ed.), Mineralogical Society of America.
Meadowcroft T. R. and Richardson F. D. (1965). Structural and thermodynamic aspects of
phosphate glasses. Trans. Faraday Soc., 61:54–70.
Meagher E. P. (1975). The crystal structures of pyrope and grossularite at elevated tempera-
tures. Amer. Mineral., 60:218–228.
Meagher E. P. (1982). Silicate garnets. In Reviews in Mineralogy, vol. 5, P. H. Ribbe (series ed.),
Medaris L. G. (1969). Partitioning of Fe2⫹ and Mg2⫹ between coexisting synthetic olivine and
orthopyroxene. Amer. Jour. Sci., 267:945–968.
Megaw H. D. (1974). The architecture of the feldspars. In The Feldspars, W. S. MacKenzie and
J. Zussman, eds. Manchester: Manchester University Press.
844 / References
Mel’nik Y. P. (1972). Thermodynamic parameters of compressed gases and metamorphic reac-

tions involving water and carbon dioxide. Geochem. Int., 9:419–426.
Mendeleev D. (1905). Principles of Chemistry London.
Mendeleev D. (1908). Versuch einen chemisschen Aufflassung des Weltäthers St. Petersburg.
Menzer G. (1926). Die kristallstructur von granat. Zeit. Kristallog., 63:157–158.
Merkel G. A. and Blencoe J. G. (1982). Thermodynamic procedures for treating the mono-
clinic/triclinic inversions as a high-order phase transition in equations of state for binary
analbite-sanidine feldspars. In Advances in Physical Geochemistry, vol. 2, S. K. Saxena
(series ed.). New York: Springer-Verlag.
Merlino S. (1976). Framework silicates. Izvj. Jugoslav. Centr. Krist., 11:19–37.
Merrihue C. and Turner G. (1966). Potassium-argon dating by activation with fast neutrons.
J. Geophys. Res., 71:2852–2857.
Millero F. J. (1972). The partial molal volumes of electrolytes in aqueous solutions. In Water
and Aqueous Solutions, R. A. Horne (series ed.), New York: Wiley Interscience.
Millero F. J. (1977). Thermodynamic models for the state of metal ions in sea water. In The
Sea, vol. 6, E. D. Goldberg et al., eds. New York: Wiley Interscience.
Misener D. J. (1974). Cationic diffusion in olivine to 1400°C and 35 kbar. In Geochemical Trans-
port and Kinetics, A. W. Hoffmann, B. J. Giletti, H. S. Yoder Jr., and R. A. Yund, eds.,
Carnegie Institution of Washington.
Misra N. K. and Venkatasubramanian V. S. (1977). Strontium diffusion in feldspars—a labora-
tory study. Geochim. Cosmochim. Acta, 41:837–838.
Moore P. B. and Smith J. V. (1970). Crystal structure of ␤-Mg2SiO4: Crystal chemical and geo-
physical implications. Phys. Earth Planet. Int., 3:166–177.
Moore R. K., White W. B., and Long T. V. (1971). Vibrational spectra of the common silicates,
I: The garnets. Amer. Mineral., 56:54–71.
Mori T. (1977). Geothermometry of spinel Iherzolites. Contrib. Mineral. Petrol., 59:261–279.
Mori T. and Green D. H. (1978). Laboratory duplication of phase equilibria observed in natu-
ral garnet Iherzolites. J. Geol., 86:83–97.
Morimoto N. and Tokonami M. (1969). Oriented exsolution of augite in pigeonite. Amer. Min-
eral., 54:1101–1117.
Morimoto N., Appleman D. E., and Evans H. T. Jr. (1960). The crystal structures of clinoens-
tatite and pigeonite. Zeit. Kristallogr., 67:587–598.
Morioka M. (1980). Cation diffusion in olivine, I: Cobalt and magnesium. Geochim. Cos-
mochim. Acta., 44:759–762.
Morioka M. (1981). Cation diffusion in olivine, II: Ni-Mg, Mn-Mg, Mg, and Ca. Geochim.
Cosmochim. Acta., 45:1573–1580.
Morioka M. and Nagasawa H. (1991). Ionic diffusion in olivine. In Advances in Physical Geo-
chemistry, vol. 8, J. Ganguly, ed. New York-Heidelberg-Berlin: Springer-Verlag.
Morlotti R. and Ottonello G. (1982). Solution of rare earth elements in silicate solid phases:
Henry’s law revisited in light of defect chemistry. Phys. Chem. Minerals, 8:87–97.
Morlotti R. and Ottonello G. (1984). The solution of trace amounts of Sm in fosteritic olivine:
An experimental study by EMF galvanic cell measurements. Geochim. Cosmochim.
Acta, 48:1173–1181.
Morris R. V. and Haskin L. A. (1974). EPR measurement of the effect of glass composition
on the oxidation states of europium. Geochim. Cosmochim. Acta, 38:1435–1445.
Morse S. A. (1976). The lever rule with fractional crystallization and fusion. Amer. Jour. Sci.,
276:330–346.
Morse S. A. (1980). Basalts and Phase Diagrams. New York, Heidelberg, and Berlin:
Springer-Verlag.
Mott N. F. and Littleton M. S. (1938). Conduction in polar crystals, I: Electrolitic conduction
in solid salts. Trans. Faraday Soc., 34:485–499.
Mueller R. F. (1960). Compositional characteristics and equilibrium relation in mineral assem-
blages of a metamorphosed iron formation. Amer. Jour. Sci., 258:449–497.
References / 845
Mueller R. F. (1961). Analysis of relations among Mg, Fe and Mn in certain metamorphic

minerals. Geochim. Cosmochim. Acta, 25:267–296.
Mueller R. F. (1962). Energetics of certain silicate solid solutions. Geochim. Cosmochim.
Acta, 26:581–598.
Mueller R. F. (1972). Stability of biotite: A discussion. Amer. Mineral., 57:300–316.
Mukherjee K. (1965). Monovacancy formation energy and Debye temperature of close-packed
metals. Phil. Mag., 12:915–918.
Mukhopadhyay D. K. and Lindsley D. H. (1983). Phase relations in the join kischsteinite
(CaFeSiO4)-faialite (Fe2SiO4). Amer. Mineral., 62:1089–1094.
Muller R. A. (1979). Radioisotope dating with accelerators. Phys. Today, 32:23–28.
Mulliken R. S. (1934). A new electroaffinity scale; together with data on valence states and on
valence ionization potentials and electron affinities. J. Chem. Phys., 2:782–793.
Mulliken R. S. (1935). Electronic structure of molecules, XI: Electroaffinity, molecular orbitals,
and dipole moments. J. Chem. Phys., 3:573–591.
Munk W. H. (1966). Abyssal reciper. Deep Sea Res., 13:707–730.
Munoz J. L. (1984). F-OH and Cl-OH exchange in micas with applications to hydrothermal
ore deposits. In Reviews in Mineralogy. vol. 13., P. H. Ribbe (series ed.), Mineralogical
Society of America.
Munoz J. L. and Ludington S. D. (1974). Fluorine-hydroxil exchange in biotite. Amer. Jour.
Sci., 274:396–413.
Munoz J. L. and Ludington S. D. (1977). Fluorine-hydroxil exchange in synthetic muscovite
and its application to muscovite-biotite assemblages. Amer. Mineral., 62:304–308.
Myers W. D. and Swiatecki W. J. (1966). Nuclear masses and deformations. Nucl. Phys.,
81:1–16.
Mysen B. O. (1976a). Nickel partitioning between upper mantle crystals and partial melts as a
function of pressure, temperature, and nickel concentration. Carnegie Inst. Wash. Yb.,
75:662–668.
Mysen B. O. (1976b). Rare earth partitioning between crystals and liquid in the upper mantle.
Carnegie Inst. Wash. Yb., 75:656–659.
Mysen B. O. (1977a). Partitioning of nickel between liquid, pargasite and garnet peridotite
minerals and concentration limit of behavior according to Henry’s law at high pressure
and temperature. Amer. Jour. Sci., 278:217–243.
Mysen B. O. (1977b). Experimental determination of cesium, samarium and thulium parti-
tioning between hydrous liquid, garnet peridotite minerals and pargasite. Carnegie Inst.
Wash. Yb., 76:588–594.
Mysen B. O. (1978). Limits of solution of trace elements in minerals according to Henry’s law:
Review of experimental data. Geochim. Cosmochim. Acta, 42:871–886.
Mysen B. O., Virgo D., and Scarfe C. M. (1980). Relation between the anionic structure and
viscosity of silicate melts: A Raman spectroscopic study. Amer. Mineral., 65:690–710.
Nafziger R. H. and Muan A. (1967). Equilibrium phase compositions and thermodynamic
properties of olivines and pyroxenes in the system MgO-FeO-SiO2. Amer. Mineral.,
52:1364–1384.
Nakamura A. and Schmalzried H. (1983). On the nonstoichiometry and point defects in oliv-
ine. Phys. Chem. Minerals, 10:27–37.
Nakamura A. and Schmalzried H. (1984). On the Fe2⫹-Mg2⫹ interdiffusion in olivine. Phys.
Chem., 88:140–145.
Nath P. and Douglas R. W. (1965). Cr3⫹/Cr6⫹ equilibrium in binary alkali silicate glasses. Phys.
Chem. Glasses, 6:197–202.
Naumov G. B., Ryzhenko B., and Khodakovsky I. L. (1971). Handbook of Thermodynamic
Data Atomizdat, Moscow.
Navrotsky A. (1985). Crystal chemical constraints on the thermochemistry of minerals. In Re-
views in Mineralogy, vol. 14, P. H. Ribbe (series ed.), Mineralogical Society of America.
Navrotsky A. (1994). Physics and Chemistry of Earth Materials Cambridge University Press.
846 / References
Navrotsky A. and Akaogi M. (1984). a-b-c phase relations in Fe2SiO4-Mg2SiO4 and Co2SiO4-
Mg2SiO4: Calculation from thermochemical data and geophysical applications. J. Geo-
phys. Res., 89:10135–10140.
Navrotsky A. and Loucks D. (1977). Calculations of subsolidus phase relations in carbonates
and pyroxenes. Phys. Chem. Minerals, 1:109–127.
Navrotsky A., Geisinger K. L., McMillan P., and Gibbs G. V. (1985). The tetrahedral frame-
work in glasses and melts: Inferences from molecular orbital calculations and implica-
tions for structure, thermodynamics and physical properties. Phys. Chem. Minerals,
11:284–298.
Navrotsky A., Peraudeau G., McMillan P., and Coutures J. P. (1982). A thermochemical study
of glasses and crystals along the joins silica-calcium aluminate and silica-sodium alumi-
nate. Geochim. Cosmochim. Acta, 46:2039–2047.
Nernst W. (1891). Distribution of a substance between two solvents and between solvent and
vapor. Z. Phis. Chem., 8:110–139.
Neumann H., Mead J., and Vitaliano C. J. (1954). Trace element variation during fractional
crystallization as calculated from the distribution law. Geochim. Cosmochim. Acta,
6:90–99.
Newnham R. E. and Megaw H. D. (1960). The crystal structure of celsian (barium feldspar).
Acta Cryst., 13:303–312.
Newton R. C. (1987). Thermodynamic analysis of phase equilibria in simple mineral system.
In Reviews in Mineralogy. vol. 17, P. H. Ribbe (series ed.), Mineralogical Society of
America.
Newton R. C. and Haselton H. T. (1981). Thermodynamics of the garnet-plagioclase-Al2SiO5-
quartz geobarometer. In Thermodynamics of Minerals and Melts, R. C. Newton, A. Na-
vrotsky, and B. J. Wood, eds. New York: Springer-Verlag.
Newton R. C. and Perkins D. (1982). Thermodynamic calibration of geobarometers based on
the assemblages garnet-plagioclase-orthopyroxene (clinopyroxene)-quartz. Amer. Min-
eral., 67:203–222.
Newton R. C. and Wood B. J. (1980). Volume behavior of silicate solid solutions. Amer. Min-
eral., 65:733–745.
Newton R. C., Charlu T. V., and Kleppa O. J. (1977). Thermochemistry of high pressure gar-
nets and clinopyroxenes in the system CaO-MgO-Al2O3-SiO2. Geochim. Cosmochim.
Acta, 41:369–377.
Newton R. C., Charlu T. V., and Kleppa O. J. (1980). Thermochemistry of high structural state
plagioclases. Geochim. Cosmochim. Acta, 44:933–941.
Newton R. C., Geiger C. A., Kleppa O. J., and Brousse C. (1986). Thermochemistry of binary
and ternary garnet solid solutions. I.M.A., Abstracts with Program, 186.
Newton W. and McCready N. (1948). Thermodynamic properties of sodium silicates. C. Phys.
Coll. Chem., 52:1277–1283.
Nishizawa O. and Akimoto S. (1973). Partitioning of magnesium and iron between olivine and
spinel, and between pyroxene and spinel. Contrib. Mineral. Petrol., 41:217–240.
Nordlie B. E. (1971). The composition of the magmatic gas of Kilauea and its behavior in the
near surface environment. Amer. Jour. Sci., 271:417–473.
Novak G. A. and Colville A. (1975). A linear regression analysis of garnet chemistries versus
cell parameters (abstract). Geol. Soc. Amer. S. W. Section Meeting, Los Angeles, 7:359.
Novak G. A. and Gibbs G. V. (1971). The crystal chemistry of silicate garnets. Amer. Min-
eral., 56:791–825.
Nylen P. and Wigren N. (1971). Stechiometria CEDAM, Padova.
Oba T. (1980). Phase relations in the tremolite-pargasite join. Contrib. Mineral. Petrol.,
71:247–256.
Ohmoto H. (1972). Systematics of sulfur and carbon isotopes in hydrothermal ore deposits.
Econ. Geol., 67:551–578.
Ohmoto H. and Lasaga A. C. (1982). Kinetics of reactions between aqueous sulfates and sul-
fides in hydrothermal systems. Geochim. Cosmochim. Acta, 46:1727–1745.
References / 847
Ohmoto H. and Rye R. O. (1979). Isotopes of sulfur and carbon. In Geochemistry of Hydrother-
mal Ore Deposits, 2d ed., H. L. Barnes, ed. New York: John Wiley.
Oka Y. and Matsumoto T. (1974). Study on the compositional dependence of the apparent
partitioning coefficient of iron and magnesium between coexisting garnet and clinopyro-
xene solid solutions. Contrib. Mineral. Petrol., 48:115–121.
Okamura F. P. S., Ghose S., and Ohashi H. (1974). Structure and crystal chemistry of calcium
Tschermak’s pyroxene CaAlAlSiO6. Amer. Mineral., 59:549–557.
O’Neill H.St. C. and Wood B. J. (1979). An experimental study of Fe-Mg partitioning between
garnet and olivine and its calibration as a geothermometer. Contrib. Mineral. Petrol.,
70:59–70.
O’Neill M. J. and Fyans R. L. (1971). Design of differential scanning calorimeters and the
performance of a new system. Norwalk, Connecticut, Perkin-Elmer Corp., 38 pp.
O’Nions R. K. and Smith D. G. W. (1973). Bonding in silicates: An assessment of bonding in
orthopyroxene. Geochim. Cosmochim. Acta, 37:249–257.
Onken H. (1965). Verfeinerung der Kristallstruktur von Monticellite. Tsch. Min. Petr. Mitt.,
10:34–44.
Onuma N., Higuchi H., Wakita H., and Nagasawa H. (1968). Trace element partition between
two pyroxenes and host volcanic rocks. Earth Planet. Sci. Letters, 5:47–51.
Orville P. M. (1972). Plagioclase cation exchange equilibria with aqueous chloride solution:
Results at 70°C and 2000 bars in the presence of quartz. Amer. Jour. Sci., 272:234–272.
Osborn E. F. (1942). The system CaSiO3-diopside-anorthite. Amer. Jour. Sci., 240:751–788.
Osborn E. F. and Tait D. B. (1952). The system diopside-forsterite-anorthite. Amer. Jour. Sci.,
Bowen Volume, 413–434.
Ottonello G. (1980). Rare earth abundance and distribution in some spinel peridotite xenoliths
from Assab (Ethiopia). Geochim. Cosmochim. Acta, 44:1885–1901.
Ottonello G. (1983). Trace elements as monitors of magmatic processes (1) limits imposed by
Henry’s law problem and (2) Compositional effect of silicate liquid. In The Significance
of Trace Elements in Solving Petrogenetic Problems and Controversies, S. Augusthitis, ed.
Theophrastus Publications, Athens.
Ottonello G. (1986). Energetics of multiple oxides with spinel structure. Phys. Chem. Miner-
als., 13:79–90.
Ottonello G. (1987). Energies and interactions in binary (Pbnm) orthosilicates: A Born parame-
trization. Geochim. Cosmochim. Acta, 51:3119–3135.
Ottonello G. (1992). Interactions and mixing properties in the (C2/c) clinopyroxene quadrilat-
eral. Contrib. Mineral. Petrol., 111:53–60.
Ottonello G. and Morlotti R. (1987). Thermodynamics of nickel-magnesium olivine solid solu-
tion. J. Chem. Thermodyn., 19:809, 818.
Ottonello G. and Ranieri G. (1977). Effetti del controllo petrogenetico sulla distribuzione del
Ba nel processo di anatessi crustale. Rend. S.I.M.P., 33:741–753.
Ottonello G., Bokreta M. and Sciuto P. F. (1996). Parameterization of energy and interactions
in garnets: End-member properties. Amer. Mineral., 81:429–447.
Ottonello G., Bokreta M., and Sciuto P. F. (in preparation). Parameterization of energy and
interactions in garnets: Mixing properties.
Ottonello G., Della Giusta A., and Molin G. M. (1989). Cation ordering in Ni-Mg olivines.
Ottonello G., Piccardo G. B. and Ernst W. G. (1979). Petrogenesis of some Ligurian peridotites,
II: Rare earth element chemistry. Geochim. Cosmochim. Acta, 43:1273–1284.
Ottonello G., Princivalle F., and Della Giusta A. (1990). Temperature, composition and f O2
effects on intersite distribution of Mg and Fe2⫹ in olivines. Phys. Chem. Minerals,
17:301–312.
Ottonello G., Della Giusta A., Dal Negro A., and Baccarin F. (1992). A structure-energy model
for C2/c pyroxenes in the system Na-Mg-Ca-Mn-Fe-Al-Cr-Ti-Si-O. In Advances in
Physical Geochemistry, S. K. Saxena (series ed.), vol. 10. Berlin-Heidelberg-New York:
Springer-Verlag.
848 / References
Ottonello G. Piccardo G. B., Mazzucotelli A., and Cimmino F. (1978). Clinopyroxene-

orthopyroxene major and REE partitioning in spinel peridotite xenoliths from Assab
(Ethiopia). Geochim. Cosmochim. Acta., 42:1817–1828.
Papike J. J. and Clark J. R. (1968). The crystal structure and caution distributions of glauco-
phane. Amer. Mineral., 53:1156–1173.
Papike J. J. and Ross M. (1970). Gedrites: Crystal structures and intracrystalline cation distri-
butions. Amer. Mineral., 55:1945–1972.
Parson I. (1978). Alkali feldspars: Which solvus? Phys. Chem. Minerals, 2:199–213.
Patchett J. P. (1983). Importance of the Lu-Hf isotopic system in studies of planetary chronol-
ogy and chemical evolution. Geochim. Cosmochim. Acta, 47:81–91.
Paul A. and Douglas R. W. (1965a). Ferrous-ferric equilibrium in binary alkali silicate glasses.
Phys. Chem. Glasses, 6:207–211.
Paul A. and Douglas R. W. (1965b). Cerous-ceric equilibrium in binary alkali silicate glasses.
Phys. Chem. Glasses, 6:212–215.
Pauli W. (1933). Die allgemeinen Prinzipien der Wellenmechanik. Handbuch der Physik, 24/
1:83–227.
Pauling L. (1927a). The sizes of ions and the structure of ionic crystals. Jour. Amer. Chem.
Soc., 49:765–792.
Pauling L. (1927b). The theoretical prediction of the physical properties of many-electron
atoms and ions. Mole refraction, diamagnetic susceptibility, and extension in space.
Proc. Roy. Soc., A114:181–189.
Pauling L. (1929). The principles determining the structure of complex ionic crystals. Jour.
Amer. Chem. Soc., 51:1010–1026.
Pauling L. (1932). Nature of the chemical bond. IV. The energy of single bonds and the relative
electronegativity of atoms. Jour. Amer. Chem. Soc., 54:3570–3582.
Pauling L. (1947). Atomic radii and interatomic distances in metals. Jour. Amer. Chem. Soc.,
69:542–553.
Pauling L. (1948). The Nature of the Chemical Bond. Cornell University Press, 2d ed., Ithaca,
N.Y.
Pauling L. (1960). The Nature of the Chemical Bond. Cornell University Press, Ithaca, N.Y.
Pauling L. (1980). The nature of silicon-oxygen bonds. Amer. Mineral., 65:321–323.
Pauling L. and Sherman J. (1932). Screening constants for many-electrons atoms: The calcula-
tion and interpretation of X-ray term values and the calculation of atomic scattering
factors. Zeit. Krist., 81:1–29.
Payne J. A. and Drever R. (1965). An investigation of the beta decay of rhenium to osmium
with high temperature proportional counters. Ph.D. diss., University of Glasgow.
Peacor D. R. (1973). High temperature single-crystal study of the cristobalite inversion. Zeit.
Krist., 138:274–298.
Perchuk L. L. (1977). Thermodynamic control of metamorphic processes. In Energetics of Geo-
logical Processes, S. K. Saxena and S. Batacharji, eds. New York: Springer-Verlag.
Perchuk L. L. (1991). Derivation of a thermodynamically consistent set of geothermometers
and geobarometers for metamorphic and magmatic rocks. In Progress in Methamorphic
and Magmatic Petrology, L. L. Perchuk, ed. Cambridge University Press, London.
Perchuk L. L. and Aranovich L. Ya. (1979). Thermodynamics of variable composition:
Andradite-grossularite and pistacite-clinozoisite solid solutions. Phys. Chem. Minerals,
5:1–14.
Perkins D. (1983). The stability of Mg-rich garnet in the system CaO-MgO-Al2O3-SiO2 at 1000
to 1300°C and high pressure. Amer. Mineral., 68:355–364.
Phillips J. C. (1970). Ionicity of the chemical bond in crystals. Rev. Mod., Phys., 42(3).:317.
Phillips M. W. and P. H. Ribbe (1973). The structure of monoclinic potassium-rich feldspars.
Philpotts A. R. (1979). Silicate immiscibility in tholeiitic basalts. J. Petrol., 20:99–118.
Philpotts J. A. and Schnetzler C. C. (1970). Phenocryst-matrix partition coefficients for K, Rb,
References / 849
Sr and Ba, with application to anorthosite and basalt genesis. Geochim. Cosmochim.
Acta, 34:307–322.
Piccardo G. B. and Ottonello G. (1978). Partial melting effects on coexisting mineral com-
positions in upper mantle xenoliths from Assab (Ethiopia). Rend. S.I.M.P., 34:499–526.
Pitzer K. S. (1973). Thermodynamics of electrolytes. I Theoretical basis and general equations.
J. Phys. Chem., 77:268–277.
Pitzer K. S. (1983). Dieletric constants of water at very high temperature and pressure. Proc.
Natl. Acad. Sci. U.S.A., 80:4575–4576.
Plummer L. N., Wigley T. M. L. and Parkhurst D. L. (1978). The kinetics of calcite dissolution
in CO2-water system at 5° to 60°C and 0.0 to 1.0 atm CO2. Amer. Jour. Sci., 278:179–216.
Pluschkell W. and Engell H. J. (1968). Ionen und Elektronenleitung in Magnesiumorthosilikat.
Ber. Dtsch. Keram. Ges., 45:388–394.
Poldervaart A. and Hess H. H. (1951). Pyroxenes in the crystallization of basaltic magmas.
J. Geol., 59:472–489.
Pople J. A. (1951). The structure of water and similar molecules. Proc. Roy. Soc., A202:323–336.
Popp R. K., Gilbert M. C. and Craig J. R. (1976). Synthesis and X-ray properties of Fe-Mg
orthoamphiboles. Amer. Mineral., 61:1267–1279.
Potter R. W. and Clynne M. A. (1978). The solubility of noble gases He, Ne, Ar, Kr and Xe in
water up to the critical point. J. Soln. Chem., 7:837–844.
Pourbaix M. (1966). Atlas of Electrochemical Equilibria in Aqueous Solutions Pergamon Press,
Oxford.
Powell M. and Powell R. (1977a). Plagioclase-alkali feldspar thermometry revisited. Min.
Mag., 41:253–256.
Powell R. and Powell M. (1977b). Geothermometry and oxygen barometry using iron-titanium
oxides: A reappraisal. Min. Mag., 41:257–263.
Powenceby M. I., Wall V. J. and O’Neill H. St. C. (1987). Fe-Mn partitioning between garnet
and ilmenite: Experimental calibration and applications. Contrib. Mineral. Petrol.,
97:116–126.
Prausnitz J. M., Lichtenhaler R. N., and De Azvedo E. G. (1986). Molecular Thermodynamics
of Fluid Phase Equilibria. New York: Prentice-Hall.
Presnall D. C. (1969). The geometrical analysis of partial fusion. Amer. Jour. Sci., 267:1178–
1194 Prewitt C. T., Papike J. J. and Ross M. (1970). Cummingtonite. A reversible, non-
quenchable transition from P21/m to C2/m symmetry. Earth. Planet. Sci. Letters,
8:448–450.
Prigogine I. (1955). Introduction to Thermodynamics of Irreversible Processes. 2d ed. New York:
Interscience Publishers.
Provost A. and Bottinga Y. (1974). Rates of solidification of Apollo 11 basalts and hawaiian
tholeiite. Earth. Planet. Sci. Letters, 15:325–337.
Putnis A. and McConnell J. D. C. (1980). Principles of Mineral Behavior. Blackwell Scientific
Publications, Oxford.
Rae A. I. M. (1973). A theory for the interactions between closed shell systems. Chem. Phys.
Lett., 18:574–577.
Raheim A. (1975). Mineral zoning as a record of P, T history of Precambrian metamorphic
rocks in W. Tasmania. Lithos, 8:221–236.
Raheim A. (1976). Petrology of eclogites and surrounding schists from the Lyell Highway–
Collingwood River area, W. Tansmania. Geol. Soc. Austr. Jour., 23:313–327.
Raheim A. and Green D. H. (1974). Experimental determination of the temperature and pres-
sure dependence of the Fe-Mg partition coefficient for coexisting garnet and clinopyro-
xene. Contrib. Mineral. Petrol., 48:179–203.
Raheim A. and Green D. H. (1975). P, T paths of natural eclogites during metamorphism, a
record of subduction. Lithos, 8:317–328.
Rakai R. J. (1975). Crystal structure of spessartite and andradite at elevated temperatures.
M.Sc. diss., University of British Columbia.
850 / References
Ramberg H. (1952). Chemical bonds and the distribution of cations in silicates. J. Geol.,
60:331–355.
Ramberg H. and Devore G. (1951). The distribution of Fe2⫹ and Mg in coexisting olivines and
pyroxenes. J. Geol., 59:193–210.
Rammensee W. and Fraser D. G. (1982). Determination of activities in silicate melts by Knud-
sen cell mass spectrometry, I: The system NaAlSi3O8-KAlSi3O8. Geochim. Cosmochim.
Acta, 46:2269–2278.
Rankama K. (1954). Isotope Geology Pergamon Press, London.
Ravizza G. and Turekian K. K. (1989). Application of the 187Re-187Os system to black shale
geochronometry. Geochim. Cosmochim. Acta, 53:3257–3262.
Rayleigh J. W. S. (1896). Theoretical considerations respecting the separation of gases by diffu-
sion and similar processes. Philos. Mag., 5th ser., 42:493–498.
Reddy K. P. R., Oh S. M., Major L. D. and Cooper A. R. (1980). Oxygen diffusion in forsterite.
J. Geophs. Res., 85:322–326.
Redlich O. and Kister A. T. (1948). Thermodynamics of non-electrolyte solutions: x-y-t rela-
tions in a binary system. Ind. Eng. Chem., 40:341–345.
Reed M. and Spycher N. (1984). Calculation of pH and mineral equilibria in hydrothermal
waters with application to geothermometry and studies of boiling and dilution. Geo-
Reuter J. H. and Perdue E. M. (1977). Importance of heavy metal-organic matter interactions
in natural waters. Geochim. Cosmochim. Acta, 41:325–334.
Ribbe P. H. (1983a). The chemistry, structure and nomenclature of feldspars. In Reviews in
Mineralogy, vol. 2 (2d ed.), P. H. Ribbe (series ed.), Mineralogical Society of America.
Ribbe P. H. (1983b). Aluminium-silicon order in feldspars: Domain textures and diffraction
patterns. In Reviews in Mineralogy, vol. 2 (2d ed.), P. H. Ribbe (series ed.), Mineralogical
Society of America.
Richardson F. D. (1956). Activities in ternary silicate melts. Trans. Faraday Soc., 52:1312–1324.
Richardson S. W. (1968). Staurolite stability in a part of the system Fe-Al-Si-O-H. J. Petrol.,
9:467–488.
Richardson S. W., Gilbert M. C. and Bell P. M. (1969). Experimental determination of kyanite-
andalusite and andalusite-sillimanite equilibria: The aluminum silicate triple point.
Amer. Jour. Sci., 267:259–272.
Richet P. and Bottinga Y. (1983). Verres, liquides, et transition vitreuse. Bull. Mineral.,
106:147–168.
Richet P. and Bottinga Y. (1985). Heat capacity of aluminium-free liquid silicates. Geochim.
Richet P. and Bottinga Y. (1986). Thermochemical properties of silicate glasses and liquids: A
review. Rev. Geophys. Space Phys., 24:1–25.
Rickard D. T. (1975). Kinetics and mechanism of pyrite formation at low temperatures. Amer.
Jour. Sci., 275:636–652.
Rimstidt J. D. and Barnes H. L. (1980). The kinetics of silica-water reactions. Geochim. Cos-
mochim. Acta, 44:1683–1699.
Ringwood A. E. (1969). Phase transformations in the mantle. Earth Planet. Sci. Letters,
5:401–412.
Ringwood A. E. and Major A. (1970). The system Mg2SiO4-Fe2SiO4 at high pressures and
temperatures. Phys. Earth Planet. Int., 3:89–108.
Rittner E. S., Hutner R. A. and Du Pré F. K. (1949). Concerning the work of polarization in
ionic crystals of the NaCl type. II. Polarization around two adjacent charges in the rigid
lattice. J. Chem. Phys., 17:198–203.
Robie R. A., Hemingway B. S. and Fisher J. R. (1978). Thermodynamic properties of minerals
and related substances at 298.15 K and 1 bar (105 pascal) pressure and at higher temper-
atures. U.S.G.S. Bull., 1452, 456 pp.
Robie R. A., Bethke P. M., Toulmin M. S. and Edwards J. L. (1966). X-ray crystallographic
References / 851
data, densities and molar volumes of minerals. In Handbook of Physical Constants,

S. Clark, ed. Geol. Soc. Amer. Mem., 97:27–73.
Robie R. A., Zhao-Bin, Hemingway B. S. and Barton M. D. (1987). Heat capacity and thermo-
dynamic properties of andradite garnets, Ca3Fe2Si3O12, between 10 and 1000 K and
revised values for ⌬Gof (298.15 K) of hedembergite and wollastonite. Geochim. Cos-
mochim. Acta, 51:2219–2236.
Robinson C. C. (1974). Multiple sites for Er3⫹ in alkali glasses. Jour. Non-Cryst. Solids, 15:1–10.
Robinson G. R. Jr. and Haas J. L. Jr. (1983). Heat capacity, relative enthalpy and calorimetric
entropy of silicate minerals: An empirical method of prediction. Amer. Mineral.,
68:541–553.
Robinson K., Gibbs G. V., Ribbe P. H. and Hall M. R. (1973). Cation distribution in three
hornblendes. Amer. Jour. Sci., 273A:522–535.
Robinson P. (1982). The composition space of terrestrial pyroxenes. Internal and external lim-
its. In Reviews in Mineralogy, vol. 7, P. H. Ribbe (series ed.), Mineralogical Society of
America.
Robinson P., Spear F. S., Schumacher J. C., Laird J., Klein C., Evans B. W. and Doolan B. L.
(1982). Phase relations of metamorphic amphibole, natural occurrence and theory. In
Reviews in Mineralogy, vol. 9B, P. H. Ribbe. (series ed.), Mineralogical Society of
America.
Roedder E. (1951). Low temperature liquid immiscibility in the system K2O-FeO-Al2O3-SiO2.
Roedder E. (1956). The role of liquid immiscibility in igneous petrogenesis: A discussion.
J. Geol., 64:84–88.
Roedder E. and Wieblen P. W. (1971). Lunar petrology of silicate melt inclusions, Apollo 11
rocks. Proc. Apollo 11 Lunar Sci. Conf., 1:801–837.
Romanek C. S., Grossman E. L. and Morse J. W. (1989). Carbon isotope fractionation in ara-
gonite and calcite: Experimental study of temperature and kinetic effects. Geol. Soc.
Amer. Abstr. Prog., 21:A76.
Romanek C. S., Grossman E. L. and Morse J. W. (1992). Carbon isotopic fractionation in syn-
thetic aragonite and calcite: Effects of temperature and precipitation rate. Geochim. Cos-
Roozeboom P. (1899). Erstarrungpunkte der Mischkrystalle Zweier Stoffe. Zeitschr. Phys.
Chem., 385:30–32.
Rosenbaum J. M. (1994). Stable isotope fractionation between carbon dioxide and calcite at
900°C. Geochim. Cosmochim. Acta, 58:3747–3753.
Rosenhauer M. and Eggler D. H. (1975). Solution of H2O and CO2 in Diopside melt. Carnegie
Inst. Wash. Yb., 74:474–479.
Rosholt J. N., Zartman R. E. and Nkomo I. T. (1973). Lead isotope systematics and uranium
depletion in the Granite Mountains, Wyoming. Geol. Soc. Amer. Bull., 84:989–1002.
Ross M. and Huebner J. S. (1975). A pyroxene geothermometer based on composition tempera-
ture relationship of naturally occurring orthopyroxene, pigeonite and augite. In Interna-
tional Conference on Geothermometry and Geobarometry, The Pennsylvania State Uni-
versity.
Ross M. and Ree F. H. (1980). Repulsive forces of simple molecules and mixtures at high den-
sity and temperature. J. Chem. Phys., 73:6146–6152.
Rossmann G. R. (1982). Pyroxene spectroscopy. In Reviews in Mineralogy, vol. 7, P. H. Ribbe
Rothbauer R. (1971). Untersuchung eines 2M-Muskovit mit Neutronenstrahlen. N. Jahrb. Min.
Monat., 4:143–154.
Roux J. (1971a). Fixation du rubidium et du cesium dans la nepheline et dans l’albite a 600°C
dans les conditions hydrothermales. C. R. Acad. Sci. Fr. Ser. D, 222:1469–1472.
Roux J. (1971b). Etude experimentale sur la distribution des elemets alcalines Cs, Rb et K entre
l’albite, la néphéline et les solutions hdrothermales These 3 cycle, Université de Paris Sud.
852 / References
Roux J. (1974). Etude des solutions solides des néphélines (Na,K)AlSiO4 et (Na,Rb)AlSiO4.
Rubey W. W. (1951). Geologic history of seawater. Geol. Soc. Amer. Bull., 62:1111–1147.
Russell W. A., Papanastassiou D. A. and Tombrello T. A. (1978). Ca isotope fractionation on
the earth and other solar system materials. Geochim. Cosmochim. Acta, 42:1075–1090.
Rye R. O. and Ohmoto H. (1974). Sulfur and carbon isotopes and ore genesis: A review. Econ.
Geol., 69:826–842.
Ryerson F. J. and Hess P. C. (1978). Implications of liquid distribution coefficients to mineral-
liquid partitioning. Geochim. Cosmochim. Acta, 42:921–932.
Ryerson F. J. and Hess P. C. (1980). The role of P2O5 in silicate melts. Geochim. Cosmochim.
Acta, 44:611–624.
Sack R. O. (1980). Some constraints on the thermodynamic mixing properties of Fe-Mg ortho-
pyroxenes and olivines. Contrib. Mineral. Petrol., 71:257–269.
Sahama Th. G. and Torgeson D. R. (1949). Some examples of applications of thermochemistry
to petrology. J. Geol., 57:255–262.
Sakai H. (1968). Isotopic properties of sulfur compounds in hydrothermal processes. Geochem.
J., 2:29–49.
Sakai H. and Tsutsumi M. (1978). D/H fractionation factors between serpentine and water at
1000° to 500°C and 2000 bar water pressure and the D/H ratios of natural serpentines.
Salje E. (1985). Thermodynamics of sodium feldspar I: Order parameter treatment and strain
induced coupling effects. Phys. Chem. Minerals, 12:93–98.
Salje E. (1988). Structural phase transitions and specific heat anomalies. In Physical Properties
and Thermodynamic Behaviour of Minerals, E. K. H. Salje, ed. Reidel Publishing
Company.
Salje E., Kuscholke B., Wruck B., and Kroll H. (1985). Thermodynamics of sodium feldspar,
II: Experimental results and numerical calculations. Phys. Chem. Minerals, 12:99–107.
Samsonov V. (1968). Handbook of the Physicochemical Properties of the Elements. New York-
Washington: IFI/Plenum.
Sanderson R. T. (1960). Chemical Periodicity. New York: Reinhold.
Sanderson R. T. (1966). Bond energies. J. Inorg. Nucl. Chem., 28:1553–1565.
Sanderson R. T. (1967). Inorganic Chemistry. New York: Reinhold.
Sartori F. (1976). The crystal structure of 1M lepidolite. Tsch. Min. Petr. Mitt., 23:65–75.
Sarver J. V. and Hummel F. A. (1962). Solid solubility and eutectic temperature in the system
Zn2SiO4-Mg2SiO4. J. Amer. Ceram. Soc., 45:304–314.
Sasaki S., Fujino K., Takeuchi Y., and Sadanaga R. (1980). On the estimation of atomic charges
by the X-ray method for some oxides and silicates. Acta Cryst., A36:904–915.
Sasaki S. Y., Takeuchi Y., Fujino K., and Akimoto S. (1982). Electron density distribution of
three orthopyroxenes, Mg2Si2O6, Co2Si2O6 and Fe2Si2O6. Zeits. Krist., 158:279–297.
Sato Y., Akaogi M., and Akimoto S. (1978). Hydrostatic compression of the synthetic garnets
pyrope and almandine. J. Geophys. Res., 83:335–338.
Saxena S. K. (1968). Distribution of iron and magnesium between coexisting garnet and clino-
pyroxene in rocks of varying metamorphic grade. Amer. Mineral., 53:2018–2021.
Saxena S. K. (1969). Silicate solid solution and geothermometry. 3. Distribution of Fe and Mg
between coexisting garnet and biotite. Contrib. Mineral. Petrol., 22:259–267.
Saxena S. K. (1972). Retrival of thermodynamic data from a study of intercrystalline and intra-
crystalline ion-exchange equilibrium. Amer. Mineral., 57:1782–1800.
Saxena S. K. (1973). Thermodynamics of Rock-Forming Crystalline Solutions. Berlin-
Heidelberg-New York: Springer-Verlag.
Saxena S. K. (1977). A new electronegativity scale for geochemists. In Energetics of Geological
Processes, S. K. Saxena and S. Bhattacharji, eds. New York-Heidelberg-Berlin:
Springer-Verlag.
Saxena S. K. (1979). Garnet-clinopyroxene geothermometer. Contrib. Mineral. Petrol.,
70:229–235.
References / 853
Saxena S. K. (1983). Exsolution and Fe2⫹-Mg order-disorder in pyroxenes. In Advances in

Physical Geochemistry, vol. 3, S. K. Saxena (series ed.). Berlin-Heidelberg-New York:
Springer-Verlag.
Saxena S. K. (1989). INSP and THERMO Computer Routines and Corresponding Data-base
Uppsala University, Sweden.
Saxena S. K. and Ghose S. (1971). Mg2⫹-Fe2⫹ order disorder and the thermodynamics of the
orthopyroxene-crystalline solution. Amer. Mineral., 56:532–559.
Saxena S. K. and Ribbe P. H. (1972). Activity-composition relations in feldspars. Contrib. Min-
eral. Petrol., 37:131–138.
Saxena S. K. and Chatterjee N. (1986). Thermochemical data on mineral phases I. The system
CaO-MgO-Al2O3-SiO2. J. Petrol., 27:827–842.
Saxena S. K. and Fei Y. (1987). Fluids at crustal pressures and temperatures I. Pure species.
Saxena S. K. and Fey Y. (1988). Fluid mixtures in the C-H-O system at high pressure and
temperature. Geochim. Cosmochim. Acta, 52:505–512.
Saxena S. K., Sykes J., and Eriksson G. (1986). Phase equilibria in the pyroxene quadrilateral.
J. Petrol., 27:843–852.
Saxena S. K., Chatterjee N., Fei Y., and Shen G. (1993). Thermodynamic Data On Oxides and
Silicates. New York: Springer-Verlag.
Schairer J. F. (1950). The alkali feldspar join in the system NaAlSiO4-KAlSiO4-SiO2. J. Geol.,
58:512–517.
Schairer J. F. and Bowen N. L. (1938). The system leucite-diopside-silica. Amer. Jour. Sci.,
Ser.5, 35:507–528.
Schairer J. F. and Bowen N. L. (1955). The system K2O-Al2O3-SiO2. Amer. Jour. Sci.,
253:681–747.
Schairer J. F. and Bowen N. L. (1956). The system Na2O-Al2O3-SiO2. Amer. Jour. Sci.,
254:129–195.
Schairer J. F. and Yoder H. S. (1960). The nature of residual liquids from crystallization, with
data on the system nepheline-diopside-silica. Amer. Jour. Sci., 258A:273–283.
Schilling G. J. and Winchester J. W. (1966). Rare earths in Hawaian basalts. Science,
153:867–869.
Schmalzried H. (1978). Reactivity and point defects of double oxides with emphasis on simple
silicates. Phys. Chem. Minerals, 2:279–294.
Schott J., Berner R. A. and Sjoberg E. L. (1981). Mechanism of pyroxene and amphibole
weathering, I. Experimental studies of iron-free minerals. Geochim. Cosmochim. Acta,
45:2123–2135.
Schreiber H. D. (1976). The experimental determination of redox states, properties, and distri-
bution of Chromium in synthetic silicate phases and application to basalt petrogenesis.
Ph.D. diss., University of Wisconsin.
Schwartz K. B. and Burns R. G. (1978). Mossbauer spetroscopy and crystal chemistry of natu-
ral Fe-Ti garnets. Trans. Amer. Geophys. Union, 59:395–396.
Schweitzer E. (1977). The reactions pigeonite ⫽ diopside ⫹ enstatite at 15 kbar. M.S. diss.,
State University of New York, Stony Brook.
Schwerdtfeger K. and Muan A. (1966). Activities in olivine and pyroxenoid solid solution of
the system Fe-Mn-Si-O at 1150°C. Trans. Met. Soc. AIME, 236:201–211.
Sciuto P. F. and Ottonello G. (1995a). Water-rock interaction on Zabargad Island (Red Sea).,
a case study: I). application of the concept of local equilibrium. Geochim. Cosmochim.
Acta, 59:2187–2206.
Sciuto P. F. and Ottonello G. (1995b). Water-rock interaction on Zabargad Island (Red Sea).,
a case study: II). from local equilibrium to irreversible exchanges. Geochim. Cosmochim.
Acta, 59:2207–2214.
Seck H. A. (1971). Koexisterend Alkali-feldspate und Plagioclase im System NaAlSi3O8-
KAlSi3O8-CaAl2Si2O8-H2O bei temperaturen von 650°C bis 900°C. N. Jahrb. Min.
Abh., 155:315–395.
854 / References
Seck H. A. (1972). The influence of pressure on the alkali-feldspar solvus from peraluminous
and persilicic materials. Fortschr. Mineral., 49:31–49.
Seifert S. and O’Neill H.St. C. (1987). Experimental determination of activity-composition re-
lations in Ni2SiO4-Mg2SiO4 and Co2SiO4-Mg2SiO4 olivine solid solutions at 1200 K and
0.1 MPa and 1573 K and 0.5 GPa. Geochim. Cosmochim. Acta, 51:97–104.
Seward T. M. (1981). Metal complex formation in aqueous solutions at elevated temperatures
and pressures. Phys. Chem. Earth, 13:113–132.
Shannon R. D. (1976). Revised effective ionic radii and systematic studies of interatomic dis-
tances in halides and chalcogenides. Acta Cryst., A32:751–767.
Shannon R. D. and Prewitt C. T. (1969). Effective ionic radii in oxides and fluorides. Acta
Cryst., B25:925–946.
Shaw D. M. (1970). Trace element fractionation during anatexis. Geochim. Cosmochim. Acta,
34:237–243.
Shaw D. M. (1978). Trace element behaviour during anatexis in the presence of a fluid phase.
Shmulovich K. I., Shmonov V. M., and Zharikov V. A. (1982). The thermodynamics of super-
critical fluid systems. In Advances in Physical Geochemistry. vol. 2, S. K. Saxena (series
ed.). New York: Springer-Verlag.
Shock E. L. and Helgeson H. C. (1988). Calculations of the thermodynamic and transport
properties of aqueous species at hight pressures and temperatures: Correlation algo-
rithms for ionic species and equations of state predictions to 5Kb and 1000°C. Geochim.
Shock E. L. and Helgeson H. C. (1990). Calculations of the thermodynamic and transport
properties of aqueous species at high pressures and temperatures: Standard partial
molal properties of organic species. Geochim. Cosmochim. Acta, 54:915–946.
Shock E. L. and Koretsky C. M. (1995). Metal-organic complexes in geochemical processes:
Estimation of standard partial molal thermodynamic properties of aqueous complexes
between metal cations and monovalent organic acid ligands at high pressures and tem-
peratures. Geochim. Cosmochim. Acta, 59:1497–1532.
Shock E. L., Helgeson H. C. and Sverjensky B. A. (1989). Calculation of thermodynamic and
transport properties of aqueous species at high pressures and temperatures: Standard
partial molal properties of inorganic neutral species. Geochim. Cosmochim. Acta,
53:2157–2183.
Simkin T. and Smith J. V. (1970). Minor element distribution in olivine. J. Geol., 78:304–325.
Simons B. (1986). Temperatur und Druckabhangigkeit der Fehlstellenkonzentration der Olivine
und Magnesiowustite Habilitationsschrift, Christian-Albrechts Universitat, Kiel.
Sipling P. J. and Yund R. A. (1976). Experimental determination of the coherent solvus for
sanidine-high albite. Amer. Mineral., 61:897–906.
Sjoberg E. L. (1976). A fundamental equation for calcite dissolution kinetics. Geochim. Cos-
Skinner B. J. (1956). Physical properties of end-members of the garnet group. Amer. Mineral.,
41:428–436.
Skinner B. J. (1966). Thermal expansion. In Handbook of Physical Constants, S. P. Clark Jr.,
ed. Memoir 97, The Geological Society of America.
Slavinskiy V. V. (1976). The clinopyroxene-garnet geothermometer. Dokl. Akad. Nauk USSR,
231:181–184.
Smith D. M. and Stocker R. L. (1975). Point defects and non-stoichiometry in forsterite. Phys.
Earth. Planet. Int., 10:183–192.
Smith J. V. (1983). Some chemical properties of feldspars. In Reviews in Mineralogy, vol. 2 (2d
ed.), P. H. Ribbe (series ed.), Mineralogical Society of America.
Smith J. V. and Yoder H. S. Jr. (1956). Experimental and theoretical studies of the mica poly-
morphs. Min. Mag., 31:209–235.
Smyth J. R. (1973). An orthopyroxene structure up to 850 °C. Amer. Mineral., 58:636–648.
References / 855
Smyth J. R. (1974). The high temperature crystal chemistry of clinohyperstene. Amer. Min-
eral., 59:1069–1082.
Smyth J. R. and Bish D. L. (1988). Crystal Structures and Cation Sites of the Rock-Forming
Minerals Allen and Unwin, Boston.
Smyth J. R. and Hazen R. M. (1973). The crystal structure of forsterite and hortonolite at
several temperatures up to 900°C. Amer. Mineral., 58:588–593.
Smyth J. R., Smith J. V., Artioli G., and Kvick A. (1987). Crystal structures of coesite at 15 K
and 298 K from neutron and X-ray single crystal data: Test of bonding models. J. Phys.
Chem., 91:988–992.
Sockel H. G. and Hallwig D. (1977). Ermittlung kleiner Diffusionkoeffizienten mittels SIMS in
oxydischen Verbindungen. Mikrokim. Acta, 7:95–107.
Soddy F. (1914). The Chemistry of the Radio-Elements. Pt. II Longmans & Green, London.
Speer J. A. (1984). Micas in igneous rocks. In Reviews in Mineralogy, vol. 13, P. H. Ribbe (series
Speidel D. H. and Osborn E. F. (1967). Element distribution among coexisting phases in the
system MgO-FeO-Fe2O3-SiO2 as a function of temperature and oxygen fugacity. Amer.
Mineral., 52:1139–1152.
Spencer K. J. and Lindsley D. H. (1981). A solution model for coexisting iron-titanium oxides.
Amer. Mineral., 66:1189–1201.
Stebbins J. F., Carmichael I. S. E. and Moret L. K. (1984). Heat capacities and entropies of
silicate liquids and glasses. Contrib. Mineral. Petrol., 86:132–148.
Steele I. M. and Smith J. V. (1982). Ion probe analysis of plagioclase in three howardites and
three eucrites. Geochim. Cosmochim. Acta, 42:959–971.
Steele I. M., Hutcheon I. D. and Smith J. V. (1980a). Ion microprobe analysis and petrogenetic
interpretations of Li, Mg, Ti, K, Sr, Ba in lunar plagioclase. Geochim. Cosmochim. Acta,
Suppl. 14, 571–590.
Steele I. M., Hutcheon I. D. and Smith J. V. (1980b). Ion microprobe analysis of plagioclase
feldspar (Ca1⫺xNaxAl2⫺xSi2⫹xO8). for major, minor and trace elements. VIII Int. Congr.
X-ray Optics Microanalysis, Pendell Pub. Co, Midland, Michigan.
Steele I. M., Smith J. V., Raedeke L. D. and McCallum I. S. (1981). Ion probe analysis of Still-
water plagioclase and comparison with lunar analyses. Lunar Planet. Sci., 12:1034–1036.
Stevenson F. J. (1976). Stability constants of Cu2⫹, Pb2⫹ and Cd 2⫹ complexes with humic acids.
Soil. Sci. Soc. Amer. Jour., 40:665–672.
Stevenson F. J. (1982). Humus Chemistry: Genesis, Composition, Reactions. New York: Wiley
Interscience.
Stevenson F. J. (1983). Trace metal-organic matter interaction in geologic environment. In The
Significance of Trace Elements in Solving Petrogenetic Problems and Controversies, S. Au-
gustitis, ed. Theophrastus Publications, Athens.
Stewart D. B. and Ribbe P. H. (1969). Structural explanation for variations in cell parameters
of alkali feldspars with Al/Si ordering. Amer. Jour. Sci., 267A:444–462.
Stewart R. F., Whitehead M. A. and Donnay D. (1980). The ionicity of the Si-O bond in
quartz. Amer. Mineral., 65:324–326.
Stockmayer W. H. (1941). Second virial coefficients of polar gases. J. Chem. Phys., 9:398–402.
Stolper E. (1982). The speciation of water in silicate melts. Geochim. Cosmochim. Acta,
46:2609–2620.
Stolper E., Fine G., Johnson T., and Newmann S. (1987). Solubility of carbon dioxide in albitic
melt. Amer. Mineral., 72:1071–1085.
Stolper E., Walker D., Hager B. H. and Hays J. F. (1981). Melt segregation from partially mol-
ten source regions: The importance of melt density and source region size. J. Geophys.
Res., 86:6261–6271.
Stormer J. C. Jr. (1975). A practical two feldspar thermometer. Amer. Mineral., 60:667–674.
Strip K. F. and Kirkwood J. G. (1951). Asymptotic expansion of the rotational partition func-
tion of the asymmetric top. J. Chem. Phys., 19:1131–1133.
856 / References
Stull D. R. and Prophet H. (1971). JANAF Thermochemical Tables Data series, Washington
D.C., 37:1–1141.
Stumm W. and Morgan J. J. (1981). Acquatic Chemistry. New York: John Wiley.
Sueno S., Papike J. J. and Prewitt C. T. (1973). The high temperature crystal chemistry of trem-
olite. Amer. Mineral., 58:649–664.
Sueno S., Papike J. J., Prewitt C. T. and Brown G. E. (1972). Crystal structure of high cumming-
tonite. J. Geophys. Res., 77:5767–5777.
Suess H. (1965). Secular variations of the cosmic-ray produced carbon-14 in the atmosphere
and their interpretation. J. Geophys. Res., 70:5937–5952.
Sumino Y. (1979). The elastic constants of Mn2SiO4, Fe2SiO4 and Co2SiO4, and the elastic
properties of olivine group minerals at high temperature. J. Phys. Earth, 27:209–238.
Sumino Y., Nishizawa O., Goto T., Ohno I., and Ozima I. (1977). Temperature variation of
elastic constants of single crystal forsterite between 190 and 400°C. J. Phys. Earth,
28:273–280.
Sutin N., Rowley J. K. and Dodson W. (1961). Chloride complexes of iron (III) ions and the
kinetics of chloride-catalyzed exchange reactions between iron (II) and iron (III) in light
and heavy water. J. Phys. Chem., 65:1248–1252.
Sutton S. R., Jones K. W., Gordon B., Rivers M. L., Bajt S., and Smith J. V. (1993). Reduced
chromium in olivine grains from lunar basalt 15555: X-ray Absorption Near Edge Struc-
ture (XANES). Geochim. Cosmochim. Acta, 57:461–468.
Suzuki I. and Anderson O. L. (1983). Elasticity and thermal expansion of a natural garnet up
to 1000 K. J. Phys. Earth, 31:125–138.
Suzuki I., Seya K., Takei H., and Sumino Y. (1981). Thermal expansion of fayalite, Fe2SiO4.
Suzuoki T. and Epstein S. (1976). Hydrogen isotope fractionation between OH-bearing miner-
als and water. Geochim. Cosmochim. Acta, 40:1229–1240.
Szilagyi M. (1971). Reduction of Fe3⫹ ion by humic acid preparations. Soil Sci., 3:233–235.
Tanaka T. and Masuda A. (1982). La-Ce geochronometer: A new dating method. Nature,
300:515–518.
Tanger J. C. and Helgeson H. C. (1988). Calculation of the thermodynamic and transport prop-
erties of aqueous species at high pressures and temperatures: Revised equation of state
for the standard partial molal properties of ions and electrolytes. Amer. Jour. Sci.,
288:19–98.
Tardy Y. (1979). Relationship among Gibbs energies of formation of compounds. Amer. Jour.
Sci., 279:217–224.
Tardy Y. and Garrels R. M. (1974). A method of estimating the Gibbs energies of formation
of layer silicates. Geochim. Cosmochim. Acta, 38:1101–1116.
Tardy Y. and Garrels R. M. (1976). Prediction of Gibbs energies of formation: I-Relationships
among Gibbs energies of formation of hydroxides, oxides and aqueous ions. Geochim.
Tardy Y. and Garrels R. M. (1977). Prediction of Gibbs energies of formation of compounds
from the elements, II: Monovalent and divalent metal silicates. Geochim. Cosmochim.
Acta, 41:87–92.
Tardy Y. and Gartner L. (1977). Relationships among Gibbs energies of formation of sulfates,
nitrates, carbonates, oxides and aqueous ions. Contrib. Mineral. Petrol., 63:89–102.
Tardy Y. and Viellard Ph. (1977). Relationships among Gibbs energies and enthalpies of forma-
tion of phosphates, oxides and aqueous ions. Contrib. Mineral. Petrol., 63:75–88.
Taylor H. P. Jr. (1968). The oxygen isotope geochemistry of igneous rocks. Contrib. Mineral.
Petrol., 19:1–71.
Taylor H. P. Jr. and Epstein S. (1962). Relation between 18O/16O ratio in coexisting minerals of
igneous and metamorphic rocks. Part I. Principles and experimental results. Geol. Soc.
Amer. Bull., 73:461–480.
References / 857
Temkin M. (1945). Mixtures of fused salts as ionic solutions. Acta Phys. Chim. URSS, 20:411–
420.
Temkin M. (1963). The kinetics of stationary reactions. Akad. Nauk. SSSR Doklady,
152:782–785.
Tequı̀ C., Robie R. A., Hemingway B. S., Neuville D. R. and Richet P. (1991). Melting and
thermodynamic properties of pyrope (Mg3Al2Si3O12). Geochim. Cosmochim. Acta,
55:1005–1010.
Tetley N. W. (1978). Geochronology by the 40Ar/39Ar technique using HIFAR reactor Ph.D. Dis-
sertation, Australia National University, Canberra.
Thompson A. B. (1976). Mineral reactions in pelitic rocks, II: Calculations of some P-T-X (Fe-
Mg). phase relations. Amer. Jour. Sci., 276:425–444.
Thompson J. B. Jr. (1969). Chemical reactions in crystals. Amer. Mineral., 54:341–375.
Thompson J. B. Jr. (1970). Chemical reactions in crystals: Corrections and clarification. Amer.
Mineral., 55:528–532.
Thompson J. B. Jr. and Hovis G. L. (1978). Triclinic feldspars: Angular relations and the repre-
sentation of feldspar series. Amer. Mineral., 63:981–990.
Thompson J. B. Jr. and Waldbaum D. R. (1969). Mixing properties of sanidine crystalline solu-
tions. III Calculations base on two phase data. Amer. Mineral., 54:811–838.
Thompson J. B. Jr., Waldbaum D. R. and Hovis G. L. (1974). Thermodynamic properties re-
lated to ordering in end member alkali feldspars. In The Feldspars, W. S. MacKenzie
and J. Zussman, eds. Manchester: Manchester University Press.
Thompson P. and Grimes N. W. (1977). Madelung calculation for the spinel structure. Phil.
Mag., 36:501–505.
Thorstenson D. C. (1970). Equilibrium distribution of small organic molecules in natural wa-
ters. Geochim. Cosmochim. Acta, 34:745–770.
Tilton G. R. (1960). Volume diffusion as a mechanism for discordant lead ages. J. Geophys.
Res., 65:2933–2945.
Tokonami M., Horiuchi H., Nakano A., Akimoto S., and Morimoto N. (1979). The crystal
structure of pyroxene type MnSiO3. Min. Jour., 9:424–426.
Toop G. W. and Samis C. S. (1962a). Some new ionic concepts of silicate slags. Can. Met.
Quart., 1:129–152.
Toop G. W. and Samis C. S. (1962b). Activities of ions in silicate melts. Trans. Met. Soc.
AIME, 224:878–887.
Tosi M. P. (1964). Cohesion of ionic solids in the Born model. Solid State Phys., 16:1–120.
Tossell J. A. (1980). Theoretical study of structures, stabilities and phase transition in some
metal dihalide and dioxide polymorphs. J. Geophys. Res., 85:6456–6460.
Tossell J. A. (1981). Structures and cohesive properties of hydroxides and fluorides calculated
using the modified electron gas ionic model. Phys. Chem. Minerals, 7:15–19.
Tossell J. A. (1985). Ab initio SCF MO and modified electron gas studies of electron deficient
anions and ion pairs in mineral structures. Physica, 131B:283–289.
Tossell J. A. (1993). A theoretical study of the molecular basis of the Al avoidance rule and of
the spectral characteristics of Al-O-Al linkages. Amer. Mineral., 78:911–920.
Tossell J. A. and Vaughan D. J. (1992). Theoretical Geochemistry: Application of Quantum Me-
chanics in the Earth and Mineral Sciences. New York-Oxford: Oxford University Press.
Tossell J. A., Vaughan D. J. and Johnson K. H. (1973). X-ray photoelectron, x-ray emission
and UV spectra of SiO2 calculated by the SCF-X␣. scattered wave method. Chem. Phys.
Lett., 20:329–334.
Tossell J. A., Vaughan D. J. and Johnson K. H. (1974). The electronic structure of rutile, wustite
and hematite from molecular orbital calculations. Amer. Mineral., 59:319–334.
Touret J. and Bottinga Y. (1979). Equation d’etat pour le CO2; application aux inclusion car-
boniques. Bull. Mineral., 102:577–583.
Treuil M. (1973). Critères petrologiques geochimiques et structuraux de la genèse et de la
858 / References
differenciation des magmas basaltiques: Exemple de l’Afar. Ph.D. diss., University of

Orleans.
Treuil M. and Joron J. L. (1975). Utilisation des elements hydromagmatophiles pour la sempli-
fication de la modelisation quantitative des processes magmatiques. Exemples de l’Afar
et de la dorsale medioatlantique. Rend. SIMP, 31:125–174.
Treuil M. and Varet (1973). Criteres volcanologiques, petrologiques, et geochimiques de la ge-
nèse et de da differenciation des magmas basaltiques: Exemple de l’Afar. Bull. Soc. Geol.
France, 15:506–540.
Treuil M., Joron J. L., Jaffrezic H., Villemant B., and Calas G. (1979). Geochimie des elements
hydromagmatophiles, coefficiens de partage mineraux/liquides et proprietes structurales
de ces elements dans les liquides magmatiques. Bull. Mineral., 102:402–409.
Turekian K. K. (1969). The oceans, streams, and atmosphere. In Handbook of Geochemistry,
vol. 1, K. H. Wedepohl, ed. Berlin-Heidelberg-New York: Springer-Verlag.
Turner D. R., Whitfield M., and Dickson A. G. (1981). The equilibrium speciation of dissolved
components in fresh water and seawater at 25°C and 1 atm pressure. Geochim. Cos-
Turner G. (1968). The distribution of potassium and argon in chondrites. In Origin and Distri-
bution of the Elements, L. H. Ahrens, ed. Pergamon Press, London.
Turner G., Miller J. A. and Grasty R. L. (1966). The thermal history of the Bruderheim meteor-
ite. Earth Planet. Sci. Letters, 1:155–157.
Turnock A. C., Lindsley D. M. and Graver J. E. (1973). Synthesis and unit cell parameters of
Ca-Mg-Fe pyroxenes. Amer. Mineral., 58:50–59.
Ubbelhode A. R. (1965). Melting and Crystal Structure Oxford University Press, London.
Ungaretti L. (1980). Recent developments in X-ray single crystal diffractometry applied to the
crystal-chemical study of amphiboles. 15th Conf. Yugoslav. Centre Crystallog., Bar, Yu-
goslavia.
Ungaretti L., Mazzi F., Rossi G., and Dal Negro A. (1978). Crystal-chemical characterization
of blue amphiboles. Proceedings of The 1th I. M.A. General Meeting, Novosibirsk.
Ungaretti L., Smith D. C. and Rossi G. (1981). Crystal-chemistry by X-ray structure refinement
and electron microprobe analysis of a series of sodic calcic to alkali-amphiboles from
the Nybo eclogite pod, Norway. Bull. Mineral., 104:400–412.
Urey H. C. (1947). The thermodynamic properties of isotopic substances. J. Chem. Soc. (Lon-
don). 1:562–581.
Urey H. C., Brickwedde F. G. and Murphy G. M. (1932). An isotope of hydrogen of mass 2
and its concentration (abstract). Phys. Rev., 39:864.
Urusov V. S., Lapina I. V., Kabala Yu. K. and Kravchuk I. F. (1984). Isomorphism in the
forsterite-tephrolite series. Geokhimiya, 7:1047–1055.
van der Waals J. D. (1890). Molekulartheorie eines Korpers, der aus zwei verschiedenen Stoffen
besteth. Zeit. Phys. Chem., 5:133–167.
Vaughan D. J., Burns R. G. and Burns V. M. (1971). Geochemistry and bonding of thiospinel
minerals. Geochim. Cosmochim. Acta. 35:365–381.
Verschure R. H., Andriessen P. A. M., Boelrijk N. A., Hebeda E. H., Maijer C., Priem
H. N. A. and Verdurmen E. A. (1980). On the thermal stability of Rb-Sr and K-Ar
biotite systems. Contrib. Mineral. Petrol., 74:245–252.
Verwey E. J. W. (1941). The charge distribution in the water molecule and calculation of the
intermolecular forces. Recl. Trav. Chim. Pays Bas Belg., 60:887–896.
Viellard P. (1982). Modele de calcul des energies de formation des mineraux bati sur la connais-
sance affinee des structures cristallines. CNRS Memoirs, 69, Université Louis Pasteur,
Strasbourg.
Virgo D., Mysen B. O. and Kushiro I. (1980). Anionic constitution of 1-atmosphere silicate
melts: Implications for the structure of igneous melts. Science, 208:1371–1373.
Viswanathan K., and Ghose S. (1965). The effect of Mg2⫹-Fe2⫹ substitution on the cell dimen-
sion of cummingtonite. Amer. Mineral., 30:1106–1112.
References / 859
Volfinger M. (1970). Partage de Na et Li entre sanidine muscovite et solution hydrothermale a

600°C et 1000 bars. C. R. Acad. Sci. Fr. Ser. D., 271:1345–1347.
Volfinger M. (1976). Effect de la temperature sur les distributions de Na, Rb et Cs entre la
sanidine la muscovite la phlogopite et une solution hydrothermale sous une pression de
1 kbar. Geochim. Cosmochim. Acta, 40:267–282.
Waffe H. S. and Weill D. F. (1975). Electrical conductivity of magmatic liquid effects of temper-
atures, oxygen fugacity and composition. Earth Planet. Sci. Letters, 28:254–260.
Wagman D. D., Evans W. H., Parker W. B., Halow I., Bailey S. M. and Stumm R. H. (1981).
Selected values of the chemical thermodynamic properties. NBS Tech. Note, 270-
8:1–134.
Wagman D. D., Evans W. H., Parker W. B., Stumm R. H., Halow I., Bailey S. M., Churney
K. L. and Nuttall R. L. (1982). The NBS tables of chemical thermodynamic properties.
Selected values for inorganic and C1 and C2 organic substances in SI units. J. Phys.
Chem. Ref. Data, 11, Suppl. 2.
Wagner R. C. (1952). Thermodynamics of Alloys Addison-Wesley, Reading, Mass.
Wainwright J. E. and Starkey J. (1971). A refinement of the structure of anorthite. Z. Kristal-
logr., 133:75–84.
Wall V. J. and England R. N. (1979). Zn-Fe spinel-silicate-sulphide reactions as sensors of
metamorphic intensive variables and process. Geol. Soc. Amer. Abstr. Progr., 11:534.
Walther J. V. and Helgeson H. C. (1977). Calculation of the thermodynamic properties of aque-
ous silica and the solubility of quartz and its polymorphs at high pressures and tempera-
tures. Amer. Jour. Sci., 277:1315–1351.
Warner R. D. and Luth W. C. (1973). Two-phase data for the join monticellite (CaMgSiO4)-
forsterite (Mg2SiO4): Experimental results and numerical analyses. Amer. Mineral.,
58:998–1008.
Warren B. and Bragg W. L. (1928). XII. The structure of diopside CaMg(SiO3)2. Z. Kristal-
logr., 69:168–193.
Warren B. and Modell D. I. (1930). 1. the structure of enstatite MgSiO3. Z. Kristallogr.,
75:1–14.
Wasastjerna J. (1923). On the radii of ions. Soc. Sci. Fenn. Comm. Phys. Math., 38:1–25.
Watson E. B. (1976). Two-liquid partition coefficients: Experimental data and geochemical im-
plications. Contrib. Mineral. Petrol., 56:119–134.
Watson E. B. (1977). Partitioning of manganese between forsterite and silicate liquid. Geochim.
Weill D. F. and McKay G. A. (1975). The partitioning of Mg, Fe, Sr, Ce, Sm, Eu and Yb in
lunar igneous system and a possible origin of KREEP by equilibrium partial melting.
Proc. Sixth Lunar Sci. Conf., 1143–1158.
Wells P. R. A. (1977). Pyroxene thermometry in simple and complex systems. Contrib. Mineral.
Petrol., 62:129–139.
Wenner D. B. and Taylor H. P. Jr. (1971). Temperatures of serpentinization of ultramafic rocks
based on 18O/16O fractionation between coexisting serpentine and magnetite. Contrib.
Mineral. Petrol., 32:165–185.
Wetherill G. W. (1956). Discordant uranium-lead ages. Trans. Amer. Geophys. Union,
37:320–326.
White B. S., Breakley M., and Montana A. (1989). Solubility of argon in silicate liquids at high
pressures. Amer. Mineral., 74:513–529.
Whiteway S. G., Smith I. B. and Masson C. R. (1970). Theory of molecular size distribution in
multichain polymers. Can. Jour. Chem., 48:33–45.
Whitfield M. (1975a). An improved specific interaction model for sea water at 25°C and 1
atmosphere total pressure. Mar. Chem., 3:197–213.
Whitfield M. (1975b). The extension of chemical models for sea water to include trace compo-
nents at 25°C and 1 atm. pressure. Geochim. Cosmochim. Acta, 39:1545–1557.
Whitney J. A. and Stormer J. R. (1977). The distribution of NaAlSi3O8 between coexisting mi-
860 / References
crocline and plagioclase and its effect on geothermometric calculations. Amer. Min-
eral., 62:687–691.
Whittaker E. J. W. and Muntus R. (1970). Ionic radii for use in geochemistry. Geochim. Cos-
Wilhem E., Battino R., and Wilcock R. J. (1977). Low pressure solubility of gases in liquid
water. Chem. Rev., 77:219–262.
Will T. M. and Powell R. (1992). Activity-composition relationships in multicomponent amphi-
boles: An application of Darken’s quadratic formalism. Amer. Mineral., 77:954–966.
Willaime C. and Gaudais M. (1977). Electron microscope study of plastic defects in experimen-
tally deformed alkali feldspar. Bull. Soc. Fr. Mineral. Cristallogr., 100:263–271.
Williams R. J. (1971). Reaction constants in the system Fe-MgO-SiO2-O2 at 1 atm between 900
and 1300°C: Experimental results. Amer. Jour. Sci., 270:334–360.
Winchell A. N. (1933). Elements of Optical Mineralogy. New York: John Wiley.
Winter J. K., Okamura F. P. and Ghose S. (1979). A high temperature structural study of high
albite, monalbite, and the analbite-monalbite phase transition. Amer. Mineral.,
64:409–423.
Witte H. H. and Wölfel E. (1955). Electron distribution in rock salt. Zeit. Phys. Chem.,
3:296–329.
Wohl K. (1946). Thermodynamic evaluation of binary and ternary liquid system. Trans. Amer.
Inst. Chem. Eng., 42:215–249.
Wohl K. (1953). Thermodynamic evaluation of binary and ternary liquid systems. Chem. Eng.
Progr., 49:218–219.
Wolery T. J. (1983). EQ3NR. A Computer Program for Geochemical Aqueous Speciation-
Solubility Calculations: User’s Guide and Documentation Lawrence Livermore Labora-
tory, Livermore, Cal., UCRL-53414.
Wolery T. J. (1986). Some Forms of Transition State Theory, Including Non-Equilibrium Steady
State Forms Lawrence Livermore Laboratory, Livermore, Cal., UCRL-94221.
Wones D. R. (1967). A low pressure investigation of the stability of phlogopite. Geochim. Cos-
mochim. Acta, 31:2248–2253.
Wones D. R. (1972). Stability of biotite: A reply. Amer. Mineral., 57:316–317.
Wones D. R. and Dodge F. C. W. (1977). The stability of phlogopite in the presence of quartz
and diopside. In Thermodynamics in Geology, D. G. Fraser, ed. Dordrecht-Holland,
Reidel.
Wones D. R. and Eugster H. P. (1965). Stability of biotite: Experiment theory and application.
Amer. Mineral., 50:1228–1272.
Wood B. J. (1974). Crystal field spectrum of Ni2⫹ in olivine. Amer. Mineral., 59:244–248.
Wood B. J. (1976a). The reaction phlogopite ⫹ quartz ⫽ enstatite ⫹ sanidine ⫹ H2O. In Prog-
ress in Experimental Petrology, G. M. Biggar, ed. Natural Environment. Res. Coun-
cil:Publ. series D.
Wood B. J. (1976b). The partitioning of iron and magnesium between garnet and clinopyro-
xene. Carnegie Inst. Wash. Yb., 75:571–574.
Wood B. J. (1976c). Samarium distribution between garnet and liquid at high pressure. Carne-
gie Inst. Wash. Yb., 75:659–662.
Wood B. J. (1980). Crystal field electronic effects on the thermodynamic properties of Fe2⫹
minerals. In Advances in Physical Geochemistry, vol. 1, S. K. Saxena (series ed.), New
York-Heidelberg-Berlin: Springer-Verlag.
Wood B. J. (1987). Thermodynamics of multicomponent systems containing several solid solu-
tions. In Reviews in Mineralogy. vol. 17, P. H. Ribbe (series ed.), Mineralogical Society
of America.
Wood B. J. (1988). Activity measurements and excess entropy-volume relationships for pyrope-
grossular garnets. J. Geol., 96:721–729.
Wood B. J. and Banno S. (1973). Garnet-orthopyroxene and orthopyroxene-clinopyroxene rela-
tionships in simple and complex systems. Contrib. Mineral. Petrol., 42:109–124.
References / 861
Wood B. J. and Fraser D. G. (1976). Elementary Thermodynamics for Geologists Oxford Univer-
sity Press.
Wood B. J. and Kleppa O. J. (1981). Thermochemistry of forsterite-fayalite olivine solutions.
Wood B. J. and Kleppa O. J. (1984). Chromium-aluminum mixing in garnet: A thermochemical
study. Geochim. Cosmochim. Acta, 48:1373–1375.
Wood B. J. and Strens R. G. J. (1972). Calculation of crystal field splittings in distorted coordi-
nation polyhedra: Spectra and thermodynamic properties of minerals. Min. Mag.,
38:909–917.
Wood B. J., Hackler R. T. and Dobson D. P. (1994). Experimental determination of Mn-Mg
mixing properties in garnet, olivine and oxide. Contrib. Mineral. Petrol., 115:438–448.
Wood J. A. (1979). The Solar System Prentice-Hall, Englewood Cliffs, N.J.
Woodland A. B. and O’Neill H. St. C. (1993). Synthesis and stability of garnet and phase rela-
tions with solutions. Amer. Mineral., 78:1002–1015.
Wriedt H. A. and Darken L. S. (1965). Lattice defects and solubility of nitrogen in deformed
ferritic steel. Trans. Met. Soc. AIME, 233:111–130.
Yeh H. W. (1980). D/H ratios and late-stage dehydration of shales during burial. Geochim. Cos-
mochim. Acta, 341–352.
Yoder H. S. and Eugster H. P. (1954). Phlogopite synthesis and stability range. Geochim. Cos-
York D. (1969). Least-squares fitting of a straight line with correlated errors. Earth Planet. Sci.
Letters, 5:320–324.
Yoshimori A. and Kitano Y. (1956). Theory of the lattice vibration of graphite. J. Phys. Soc.
Japan, 11:352–361.
Young J. A. and Koppel J. U. (1965). Phonon spectrum of graphite. J. Chem. Phys., 42:357–364.
Yund R. A. (1974). Coherent exsolution in the alkali feldspars. In Geochemical Transport and
Kinetics, Hofmann, Giletti, Yoder, and Yund, eds. New York: Carnegie Inst. Washing-
ton and Academic Press.
Yund R. A. (1983). Diffusion in feldspars. In Reviews in Mineralogy, vol. 2 (2d ed.), P. H. Ribbe
Yund R. A. and Anderson T. F. (1974). Oxygen isotope exchange between potassium feldspar
and KCl solution. In Geochemical Transport and Kinetics, Hofmann, Giletti, Yoder, and
Yund, eds. New York: Carnegie Inst. Washington and Academic Press.
Yund R. A. and Tullis J. (1983). Subsolidus phase relations in the alkali feldspars with emphasis
on coherent phases. In Reviews in Mineralogy, vol. 2 (2d ed.), P. H. Ribbe (series ed.),
Zhang Z. and Saxena S. K. (1991). Thermodynamic properties of andradite and application to
skarn with coexisting andradite and hedenbergite. Contrib. Mineral. Petrol.,
107:255–263.
Zheng Y. F. (1991). Calculation of oxygen isotope fractionation in metal oxides. Geochim. Cos-
mochim. Acta, 55:2299–2307.
Zheng Y. F. (1993a). Calculation of oxygen isotope fractionation in anhydrous silicate minerals.
Zheng Y. F. (1993b). Calculation of oxygen isotope fractionation in hydroxyl-bearing silicates.
INDEX
Absolute individual ionic, for aqueous solutes,

reaction rates, in the transition state the- 493, 497
ory, 598 mean ionic, for electrolytes in solution,
thermodynamic properties, in the HKF 494, 497–98
model, 521 of a gaseous component in mixture, 625
velocity, of a reaction, 588 of a trace component in solution, 658,
Acetic acid. See CH3COOH 664
Acidity: Lux-Flood, of oxides in melts, of compositional terms in mixture, 162
417–19, 674, 676 Activity diagrams, 582
Acmite availability of in literature, 587
coulomb energy, 284 construction of, 585
dispersive energy, 284 Activity product: in hydrolitic equilibria,
enthalpy of formation from the elements, 579
284 Adiabatic flashing: effect on fluid’s chemis-
heat capacity, 146 try, 580
lattice energy, 284 Adularia
molar volume, 273 compressibility, 354
repulsive energy, 284 thermal expansion, 354
structural class, 268–69 Advection: definition, 608
structural properties, 273 Affinity to equilibrium
Acoustic and solubility product, 518
branches, 137 in hydrolitic equilibria, 579
vibrational modes, definition, 137 overall, of a system, 120
Actinolite Age equation
14
Fe-Mg intracrystalline distribution, 315 6C, 766
formula unit, 300 K-Ar, 747
mixing with cummingtonite, 315 of total Ar release, 751
structural class, 300 Al-clinopyroxene
Activation energy bulk modulus, 278
and reaction rate, 598 enthalpy of formation from the elements,
of reactions, 599–600 283
Active surface area: definition, 592 heat capacity function, 283
Activity coefficient molar volume, 278
in a simple mixture, 163 standard state entropy, 283
in reciprocal mixtures, 168 structural class, 269
in Redlich-Kister expansions, 169 thermal expansion, 278
in subregular mixtures, 169 Al-orthopyroxene
in the Henry’s law range, 116 bulk modulus, 278
in the quasi-chemical model, 164 enthalpy of formation from the elements,
in Wohl’s model, 171 283
/ 863
864 / Index
Al-orthopyroxene (continued) Amphiboles

heat capacity function, 283 alkaline, classification, 300
molar volume, 278 cell edges and site dimensions in C2/m
standard state entropy, 283 structures, 306
structural class, 269 chemical classification, 299
thermal expansion, 278 chemistry, 299
Al2SiO5: polymorphic transitions, 334 compressibility, 310
Al3⫹: coordination, in melts, 420 Fe-Mg intracrystalline distribution in
Alabandite: Eh-pH stability limits, 560 sodic mixtures, 317
Albite Fe-Mg-Mn, classification, 300
compressibility, 354 intracrystalline exchange geothermome-
enthalpy of formation from the elements, try, 386
360 M-site dimensions in C2/m structures,
heat capacity function, 360 305
kinetic equilibrium with quartz, analcite, miscibility among C2/m calcic terms, 311
and fluid, 599 miscibility gaps, 311, 318
melting temperature, 456 miscibility rules, 313
molar volume, 360 mixing properties, 311
standard state entropy, 360 monoclinic, occurrences, 302
thermal expansion, 354 orthorombic, occurrences, 302
Alkali feldspars quadrilateral, 311
crystallization of, 457 site dimensions in Pnma structures, 307
intracrystalline disorder in, 351 site distortion in C2/m structures, 306
unmixing, 459 size mismatch in P21/m structures, 308
Alkalinity sodic, classification, 300
balance of, 518 sodic-calcic, classification, 300
definition,602 stoichiometry rules, 302
of carbonates, 518 structural properties, 303
total, 516 T-site dimensions in C2/m structures, 305
Almandine thermal expansion, 310
bulk modulus, 256 thermodynamic properties, 311
coulomb energy, 257 volume of mixing in binary C2/m mix-
crystal field stabilization energy, 257 tures, 306
dispersive energy, 257 volume of mixing in binary P21/m mix-
enthalpy of formation from the elements, tures, 307
257 Analbite: intrinsic instability of, 362
heat capacity function, 258 Analcite: kinetic equilibrium with quartz, al-
lattice energy, 257 bite, and fluid, 599
molar volume, 254 Anandite: formula unit, 325
repulsive energy, 257 Andalusite
structural properties, 254 extrinsic stability, 104
thermal expansion, 256 in equilibrium with kyanite, 397, 398
third-law entropy, 258 in equilibrium with sillimanite, 397, 398
Aluminous-hornblende Andradite
formula unit, 300 bulk modulus, 256
structural class, 300 coulomb energy, 257
Aluminum avoidance principle, 350 dispersive energy, 257
in plagioclases, 365 enthalpy of formation from the elements,
Amorphous silica 257
enthalpy of formation from the elements, heat capacity function, 258
374 lattice energy, 257
heat capacity function, 374 molar volume, 254
standard state entropy, 374 repulsive energy, 257
Index / 865
structural properties, 254 enthalpy of formation from the elements,

thermal expansion, 256 312
third-law entropy, 258 formula unit, 300, 312
Angle heat capacity function, 312
of octahedral flattening, in micas, 326 molar volume, 304, 312
positional in garnets, 249 standard state entropy, 312
rotational, in micas, 326 structural class, 300
Angular Overlap Model, 81 structural properties, 304
Anharmonicity thermal expansion, 310
constants of, 126 Antineutrino: discovery of, 717
of vibrational motions, 49; 125 Antiperthites: definition, 363
Anion matrix: in melts, 411 Apparent
Anionic vacancy: definition, 187 age
Annite in K-Ar dating, 749, 750
compressibility, 341 in Pb-Pb dating, 762
enthalpy of formation from the elements, equilibrium: definition, 92
340 partition coefficient
fO2-T stability, 333, 338 and conventional partition coefficient,
formula unit, 325, 340 692–93
Gibbs free energy of formation from the definition, 690
elements, 342 Aqueous n-polymers: thermodynamic prop-
heat capacity function, 340 erties, 565, 567
in equilibrium with fayalite, leucite, and Aqueous solutions
iron, 332 absolute neutrality, 545
in equilibrium with fayalite, leucite, and acid-base neutrality, 545
kalsilite, 332 redox neutrality, 545
in equilibrium with fayalite, leucite, and Ar
magnetite, 332 age spectra, 752–55
in equilibrium with sanidine and hema- aqueous, thermodynamic properties, 653
tite, 331 atmospheric contamination in radiomet-
in equilibrium with sanidine and iron, ric dating, 747
332 diffusion in crystals, 749–52
in equilibrium with sanidine and magne- Henry’s constant in silicate melts, 640
tite, 331 isotopic composition, 745
intrinsic instability, 330 solubility in silicate melts, P effect, 642
mixing with phlogopite, 344 solubility in water, T effect, 643
molar volume, 340 solution enthalpy in silicate melts, 641
standard state entropy, 340 solution entropy in silicate melts, 641
thermal expansion, 341 Arfvedsonite
Anorthite formula unit, 300
compressibility, 354 structural class, 300
enthalpy of formation from the elements, Arkose sandstones: hydrothermal alteration,
360 538
formula unit, 352 Associate defects, 198–202
heat capacity function, 360 Associative fractions: in the quasi-chemical
intracrystalline disorder, 350 model, 164
melting temperature, 454, 456 Asymmetricity: of the potential well, 125
molar volume, 352, 360 Atom
standard state entropy, 360 build-up rules, 4
structural properties, 352 energy conversion factors, 804
thermal expansion, 354 energy of external orbitals, 14
Anthophyllite nuclear binding energy, 715
activity by site ionic fractions, 319 nuclear composition, 707
866 / Index
Atom (continued) conventional coefficient in the HKF

nuclear energy,713 model, 523, 531
quantum numbers, 5 equation of, 523, 524
screen constant, 14 exponent of, 20, 43, 55
structure, 3 function Q, in the HKF model, 526
Atomic orbitals function X, in the HKF model, 533
and hydrogenoid wave functions, 64 function Y, in the HKF model, 530
and molecular orbitals in the SiO44⫺ Bornite: Eh-pH stability limits, 560
cluster, 218 Branching ratio: in isotopic decay processes,
Attractive state. See Molecular orbitals 747
Aufbau principle, 4 Brillouin zone, 130, 135, 259
Augite: structural class, 268 volume of, 135
Augitic aegirine: structural class, 268 Brittle micas: classification, 325
Azurite Buffer: multiple of S2 and O2 activity in
Formula unit, 561 gases, 629
Gibbs free energy of formation from the Bulk distribution coefficient
elements, 561 definition, 695
in modal melting episodes, 697
B(OH)3: aqueous, thermodynamic proper- in nonmodal melting episodes, 697
ties, 653 Bulk modulus
Ba definition, 59
plagioclase/melt partitioning, 682 of silicate melt components, 438
sanidine/fluid partitioning, 665 Bunsen coefficient: definition, 632
Badger’s rule, 123 Bustamite: P-T stability range, 280
Barroisite
formula unit, 300 C
structural class, 299, 300 Eh-pH stability limits, 553
Binary joins isotopes, ␣ fractionation factors and tem-
(Ca,Fe)2Si2O6 perature, 790–91, 794
mixing model, 293 isotopes, normalized phonon spectra,
phase relations, 289 793
(Ca,Mg)2Si2O6 isotopes, vibrational partition function,
mixing model, 286 792
phase relations, 284 isotopic composition, 765
Binodes: definition, 174 isotopic fractionation, 788
Biotite reduction to CH4, 652
fO2-T stability, 337 Ca: isotopic composition, 756
formula unit, 325 Ca-olivine. See Larnite
occurrences, 323, 326 Ca-Ti-Tschermak
oxidation state, 337 Coulomb energy, 284
Birnessite dispersive energy, 284
Formula unit, 561 enthalpy of formation from the elements,
Gibbs free energy of formation from the 284
elements, 561 lattice energy, 284
Bityite: formula unit, 325 repulsive energy, 284
Bohr: radius of, 64 Ca-Tschermak
Boltzmann bulk modulus, 278
constant of, 157 coulomb energy, 284
equation of, 157 dispersive energy, 284
factor of, 127 enthalpy of formation from the elements,
Boltzmann-Matano plane, 215 282, 284
Born heat capacity function, 282
absolute coefficient in the HKF model, lattice energy, 284
523 molar volume, 273, 278
Index / 867
repulsive energy, 284 Gibbs free energy of formation from the

standard state entropy, 282 elements, 342
structural class, 269 heat capacity function, 341
structural properties, 273 molar volume, 341
thermal expansion, 278 standard state entropy, 341
CaCO3 polymorphs, 104, 511 thermal expansion, 341
Calcite Celsian
dissolution kinetics, 594 formula unit, 352
in equilibrium with a pH buffered aque- molar volume, 352
ous phase, 516 structural properties, 352
in equilibrium with an aqueous phase CH2O: decomposition by bacterial media-
and a CO2-buffered gaseous phase, tion, 575
512 CH3COOH: decomposition by bacterial
in equilibrium with quartz, wollastonite, mediation, 576
and fluid, 407 CH4
solubility product, 518 gas, thermodynamic properties, 626
Calcocite in geothermal fluids, 648
Eh-pH stability limits, 560 oxidation to C, 652
Formula unit, 561 by bacterial degradation of organic mat-
Gibbs free energy of formation from the ter, 577
elements, 561 solubility in water, 645
Calcopyrite Chalcedony
Eh-pH stability limits, 560 enthalpy of formation from the elements,
formula unit, 560 374
Calderite heat capacity function, 374
bulk modulus, 256 molar volume, 373
coulomb energy, 257 occurrences, 371
dispersive energy, 257 standard state entropy, 374
enthalpy of formation from the elements, Chemical bond
257 bond number, 34
heat capacity function, 258 covalent, 15, 20
lattice energy, 257 covalent-heteropolar, 20
molar volume, 254 covalent-homeopolar, 20
repulsive energy, 257 degree of covalence, 33
structural properties, 254 fractional ionic character of, 22–23
thermal expansion, 256 ionic, 15, 18
third-law entropy, 258 mean length of, 34
Carbonate equilibria, 510 mean strength of, 33
Carnegieite: melting temperature, 465 metallic, 15
Cation matrix: in melts, 411 Pauling’s strength of, 56
Cationic vacancies van der Waals, 15, 26
definition, 187 Chemical potential
in feldspars, 346 and Gibbs free energy, 94
Cd: thermodynamic properties of organic and Helmholtz free energy, 597
complexes, 573 and thermodynamic activity, 113, 160
Ce as a partial molar property, 101
aqueous equilibria, 547 at standard state, 113
concentration anomaly in seawater, 607 in an ideal mixture, 159
isotopic composition, 745 in Gibbs’ acception, 99
Celadonite of Henry’s law standard state, 116,
compressibility, 341 658
enthalpy of formation from the elements, of silicate melt components, 443
341 Chemical solvus
formula unit, 325, 340 definition, 182
868 / Index
Chemical solvus (continued) Co-olivine

in feldspars, 364 bulk modulus, 236
in pyroxenes, 297 coulomb energy, 237
Chlorinity dispersive energy, 237
and salinity, 601 enthalpy of formation from the elements,
definition, 601 49, 237
Clausius-Mosotti: relation of, 90 heat capacity function, 238
Clinoenstatite lattice energy, 49, 237
coulomb energy, 284 repulsive energy, 237
dispersive energy, 284 standard state entropy, 238
enthalpy of formation from the elements, structural properties, 227
282, 284 Co-orthopyroxene
enthalpy of formation from the oxides, molar volume, 273
153 structural class, 269
heat capacity function, 282 structural properties, 273
lattice energy, 284 Covolume: of a gaseous phase, definition,
molar volume, 274; 278 621
P-T stability range, 279 CO2
repulsive energy, 284 amount in the earth’s atmosphere, 514
standard state entropy, 282 aqueous, thermodynamic properties, 653
structural classes, 269 conversion into glucose, 575
structural properties, 274 critical pressure, 619
thermal expansion, 278 critical temperature, 619
Clinoferrosilite fugacity coefficient, at various P and T,
bulk modulus, 278 616, 617
coulomb energy, 284 gas, thermodynamic properties, 626
dispersive energy, 284 in geothermal fluids, 648
enthalpy of formation from the elements, solubility in silicate melts, 638
284 solubility in water, 643, 645
lattice energy, 284 Coefficient
molar volume, 274; 278 of interdiffusion, 213, 216
P-T stability range, 280 of isobaric thermal expansion, 132
repulsive energy, 284 of isothermal compressibility, 132
spatial group, 266 of oxide constituent in CP functions, 145
structural class, 269 of structural components in CP func-
structural properties, 274 tions, 146
thermal expansion, 278 Coesite
Clinopyroxene bulk modulus, 374
in equilibrium with olivine, 397, 398 enthalpy of formation from the elements,
in equilibrium with plagioclase and 374
quartz, 397, 399 heat capacity function, 374
stoichiometric calculations, 392 molar volume, 372
Clintonite occurrences, 371
formula unit, 325 standard state entropy, 374
permutability calculation, 167 structural properties, 372
CO thermal expansion, 374
critical pressure, 619 Coherence energy: in pyroxenes, 286, 297
critical temperature, 619 Coherent solvus
gas, thermodynamic properties, 626 definition, 182
Co: thermodynamic properties of organic in pyroxenes, 297
complexes, 571 Cohesive pressure. See Internal pressure
Common micas: classification, 325
Index / 869
Complex ion: definition, 491 in the MEG approach, 83

Component: definition, 92 potential, 43
Compressibility Covalence energy
non-solvation term, in the HKF model, and lattice energy, 86
526 in the ligand field theory, 79
of a solute, in the HKF model, 526 Covellite
of silicate melts, 437–38 Eh-pH stability limits, 560
solvation term in the HKF model, 526 formula unit, 561
Compressibility factor. See Deviation pa- Gibbs free energy of formation from the
rameter elements, 561
Compressibility of crystalline solids Cr
and bulk modulus, 59 in natural olivines, 224
isothermal linear coefficient, 58 olivine/melt partitioning, 689
isothermal volumetric coefficient, 58 pyroxene/melt partitioning, 689
mean volumetric isothermal coefficient, Cr2⫹: in pyroxenes, 267
58 Cristobalite
pressure dependence, 62 bulk modulus, 374
Compton effect, 721 enthalpy of formation from the elements,
Concordia diagram, 760–61 374
Conservative system: definition, 812 heat capacity function, 374
Consolute temperature, 176 melting temperature, 463
Constant molar volume, 372
of distortion of the hydrogen bond, 481 occurrences, 371
of the associative process, 199–200 standard state entropy, 374
of the Schottky process, 198 structural properties, 372
Contact factor: in the quasi-chemical thermal expansion, 374
model, 164–65 Critical
Conventional partition coefficient nucleation radius, 178
and equilibrium constant, 683 temperature, of geochemically important
definition, 680, 690 gaseous species, 619
P effect on, 686 unmixing. See Spinodal decomposition
T effect on, 684–87 Cryptoperthites: definition, 363
Conventional thermodynamic properties: Crystal field
in the HKF model, 521 effect, 66–69
Conversion: from linear to nondimension- energy and absorption spectra, 69
alized frequency, 141 splitting parameter, 67
Cooling rate: effects on intercrystalline stabilization energies, 70
exchange geothermometry, 386 strong, 68
Cooperative energy, 165 theory, 63–69
in reciprocal mixtures, 168 weak, 68
Coordination number, 164 Crystal radius, 24, 36
Correlation energy: in the MEG approach, and electron density in olivine, 228
83 and ionization potential, 231
Cotectic line, 468, 471, 474 and polarizability, 231
significance of, 467 Cu
Coulomb complexation by fulvic acids, 563
energy thermodynamic properties of organic
and molecular orbitals, 73 complexes, 572
in the MEG approach, 82 Cummingtonite
of the associative process, 199 enthalpy of formation from the elements,
interactions 312
in atomic nuclei, 715 Fe-Mg intracrystalline distribution, 315
formula unit, 300, 312
870 / Index
Cummingtonite (continued) Diffusion

heat capacity function, 312 and point defects, 205
molar volume, 304, 312 and temperature, in olivine, 208
standard state entropy, 312 in garnets, 213
structural class, 300 in silicates, 211
structural properties, 304 coefficient
T-induced phase transitions, 308 and concentration gradient, 212
thermal expansion, 310 and defect equilibria, 206
Cuprite and diffusional flux, 608
Eh-pH stability limits, 560 and interdiffusion, 737
formula unit, 561 fluxes, in unmixing processes, 180
Gibbs free energy of formation from the Diffusivity: and temperature, 381
elements, 561 Dimensionality
definition, 222
Daniell cell, 543 in amphiboles, 299
Darken’s Quadratic Formalism: applied to in feldspars, 345
amphiboles, 318 in micas, 321
De l’Hopital: rule of, 759 of silica polymorphs, 371
Debye Diopside
heat capacity function of, 131 activity by ionic solution model, 380
temperature of, and acoustic frequencies, bulk modulus, 278
130 coulomb energy, 284
temperature of, and enthalpy of Frenkel dispersive energy, 284
process, 210 enthalpy of formation from the elements,
Defect 282, 284
concentration, and oxygen partial pres- heat capacity function, 282
sure, 191 lattice energy, 284
energy, by thermochemical calculations, melting temperature, 454
196 molar volume, 273, 278
equilibria P-T stability range, 280
in fayalite, 203 repulsive energy, 284
role of, in silicate melt REE parti- standard state entropy, 282
tioning, 668 structural class, 268, 269
Degeneracy: of orbital energies, 67 structural properties, 273
Degenerate ground state, 128 thermal expansion, 278
Degree of silicification, 222 Dipole
Density of states: of acoustic branches, 139 moment, 26, 87
Destabilized orbitals, 67 -dipole interactions, 44
Deviation parameter quadrupole interactions, 44
non-virial expansion of, 622 Dislocation density: in natural crystals, 186
of a gaseous phase, definition, 620 Dispersed acoustic function: in the Kieffer
of steam, 645 model, 139, 258
virial expansion of, 620 Dispersion: of wave frequencies, 137
Diadochy Displacement term: in polarization effects,
and intercrystalline exchanges, 378 90
in tetrahedral sites of amphiboles, 299 Disproportionation
in tetrahedral sites of garnets, 249 equilibria, of REE in silicate melts, 677
in tetrahedral sites of micas, 321 reactions, in melts, 421–22
in tetrahedral sites of pyroxenes, 267 Dissociation energy, 52
in tetrahedral sites of silicates, 221 Distribution coefficient: in intercrystalline
of Al3⫹ for Si4⫹ in feldspars, 346 exchanges, 377
Diffuse boundary: nuclear energy contribu- Divergence: of a vector, definition, 810
tion, 715 DLS: refinement of garnet’s structure, 253
Index / 871
Dolomite: miscibility with calcite, 177 Electroconstriction contribution: in the

Dynamic matrix, 138 HKF model, 521
Electron
of valence, 14
Eastonite orbital levels, 3
compressibility, 341 orbital sub-levels, 3
enthalpy of formation from the elements, rest mass of, 4
341 wavelike behavior of, 4
formula unit, 340 affinity, 52
Gibbs free energy of formation from the data, 24
elements, 342 definition, 14
heat capacity function, 341 density radii, 29, 31
molar volume, 341 gas, 16
standard state entropy, 341 holes, in defect equilibria, 189–90
thermal expansion, 341 transport number, 541
Eckermannite Electronegativity
formula unit, 300 according to Gordy, 22
structural class, 300 according to Mulliken, 22
Eddy diffusion. See Turbulent diffusion according to Pauling, 21, 24
Edenite according to Sanderson, 22, 24
enthalpy of formation from the elements, and ionization energy, 22
312 Electroneutrality
formula unit, 300, 312 in defect equilibria, 189
heat capacity function, 312 in the HKF model, 521
molar volume, 312 Electronic polarizability. See Optical polariz-
standard state entropy, 312 ability
structural class, 300 Elements
thermal expansion, 310 effective nuclear charge of, 14
Edge dislocation, 185 electron configuration of, 6
Effective periodic properties of, 10
crystal radii, 31, 37 Energy
diameter, 495–96 conversion factors, 803
electrons, 45 eigenvalues of potential energy func-
electrostatic radius, in the HKF model, tions, 126
522, 524, 525 eigenvalues of wave functions, 124
ionic radii, 31–32, 36 eigenvalues, and quantum numbers,
nuclear charge, 19 9–11, 65
Eh-pH diagrams eigenvalues, of wave functions, 65
C, 553 of activation of the diffusion process,
Ce, 548 207, 382, 735
Cu, 560 of activation of the unmixing process,
Fe, 556 178
Mn, 558 of associative processes in halides and
N, 554 simple sulphides, 203
P, 554 of extended defects, 187
S, 554 of extrinsic disorder in fayalite, 205
Einstein of the non-degenerate ground state, 129
characteristic temperature of, 130 of vacancy migration, 205
function of, 130, 258 Enstatite
mass-energy conversion equation, 713 bulk modulus, 278
oscillators of, 137, 259 enthalpy of formation from the elements,
Elastic strain energy. See Coherence energy 283
Electrical conductivity: in melts, 411 heat capacity function, 283
872 / Index
Enstatite (continued) of silicate melt components, 441

molar volume, 273, 278 of the defect process, 196
standard state entropy, 283 solvation, 534
structural class, 269 standard partial molal of solutes, in the
structural properties, 273 HKF model, 536
thermal expansion, 278 vibrational, of the associative process,
Enthalpy 199
and internal energy, 97 Equation
and isothermal heat capacity, 132 of B̊ of Helgeson, 499
estimates by the ⌬O2⫺ method, 151 of Benson-Krause, 644
excess, of transition, 109 of Boltzmann-Matano, 216
of a phase, at the P,T of interest, 117, of Born, 524
156 of Born-Landé, 47, 524
of fusion, 424–25, 431, 433, 436, 442 of Born-Mayer, 47
of reaction, and thermobarometry, 396 of Brönsted-Lowry, 417
of silicate melt components, 436, 441 of Clausius-Clapeyron, 424, 437
of solution in melt, 427, 433 of Debye-Hückel, 495, 516
of the defect process, 196 of Doerner-Hoskins, 691, 704
of the Schottky process, and T of melt- of Gibbs-Duhem, 118, 416
ing, 209 of Kapustinsky, 51
of vacancy migration of Kirkwood, 483
and T of melting, 209 of Ladd and Lee, 48
in alkali halides and simple oxides, of lowering of freezing point, 426, 454
207 of Potter-Clynne, 644
of vitrification, 433 of Simon, 426, 438
standard partial molal of formation of of Temkin, 412
solutes, in the HKF model, 535–36 Equation of state
Entropy of Birch, first order, 63
and isothermal heat capacity, 132 of Birch-Murnaghan, 438
and volume, 149 applied to garnets, 255
anharmonicity contribution in garnets, of Born-Mayer, 63
260 of Carnahan-Starling, for real gases, 622
at standard state, empirical estimates, of Hildebrand, 59
148 of Mie-Gruneisen, 59
configurational of Murnaghan, first order, 62, 239
in melts, 411 of perfect gases, 612
in the presence of lattice distortion, of Redlich-Kwong, for real gases, 621
663 of van der Waals, for real gases, 621
of an ideal mixture, 158 Equilibrium
of the associative process, 199 definition, 92
definition, 98 constant, and conventional partition
empirical estimates for ferrous com coefficient, 683
pounds, 151 crystallization, definition, 470, 471, 694
estimates by the Kieffer model, 142 melting
excess, of transition, 109 and trace partitioning, 696
harmonic, of a substance, 128 definition, 471
magnetic spin contribution in garnets, trace partitioning in the presence of
260 fluid, 701
non-solvation, 534 Error function: definition, 814
of a phase, at the P,T of interest, 118, Eu
156 amphoteric behavior in silicate melts,
of fusion, of silicate melt components, 679
442 oxidation state in silicate melts, 677
Index / 873
Eutectic enthalpy of formation from the elements,

line, 473 312
point, definition, 452 formula unit, 312
energetic significance of, 452, 463 heat capacity function, 312
in ternary systems, 468 molar volume, 312
Exact differentials standard state entropy, 312
definition, 805 thermal expansion, 310
integration of, 808 Fe-celadonite
Excess compressibility, 341
enthalpy of mixing, 162–63 enthalpy of formation from the elements,
entropy of mixing, 162–63 341
Gibbs free energy of mixing, 160, 163 formula unit, 340
for trace components in crystals, 664 Gibbs free energy of formation from the
in gaseous mixtures, 625 elements, 342
in Guggenheim’s expansions, 169 heat capacity function, 341
in Hillert’s model, 171 molar volume, 341
in Kohler’s model, 172 standard state entropy, 341
in subregular mixtures, 169 thermal expansion, 341
in the quasi-chemical model, 165 Fe-Cr garnet
in Wohl’s model, 170 bulk modulus, 256
volume of mixing, 162 coulomb energy, 257
Exolution microstructures, 299 crystal field stabilization energy, 257
Extended defects: in crystals, 185 dispersive energy, 257
Extrinsic enthalpy of formation from the elements,
disorder 257
in crystals, 189 heat capacity function, 258
in fayalite, 203 lattice energy, 257
in feldspars, 347 molar volume, 254
instability, 297 repulsive energy, 257
structural properties, 254
Faraday: law of, 540, 543 thermal expansion, 256
Fayalite third-law entropy, 258
amount in olivines of Earth’s mantle, Fe-glaucophane
224 enthalpy of formation from the elements,
bulk modulus, 236 312
coulomb energy, 237 formula unit, 312
defect energies in, 194 heat capacity function, 312
defect equilibria in, 203 molar volume, 312
dispersive energy, 237 polyhedral distortion, 308
enthalpy of formation from the elements, standard sate entropy, 312
49, 234, 237 thermal expansion, 310
extrinsic disorder in, 192 Fe-hornblende
heat capacity function, 238 enthalpy of formation from the elements,
lattice energy, 49, 237 312
molar volume, 227 formula unit, 312
repulsive energy, 237 heat capacity function, 312
standard state entropy, 238 molar volume, 312
structural properties, 227 standard state entropy, 312
thermal expansion, 234 thermal expansion, 310
thermodynamic properties of ␤ and ␥ Fe-olivine. See Fayalite
polymorphs, 239 Fe-pargasite
Fe-Mg exchange geothermometry, 379 enthalpy of formation from the elements,
Fe-anthophyllite 312
874 / Index
Fe-pargasite (continued) Fick

formula unit, 312 first law of diffusion, 212, 383, 737
heat capacity function, 312 second law of diffusion, 213
molar volume, 312 Field strength, 419
standard state entropy, 312 and spin condition, 72
Fe-tremolite and solution behavior in melts, 675
enthalpy of formation from the elements, First order transition: definition, 107
312 Fluor-phlogopite
formula unit, 312 compressibility, 341
heat capacity function, 312 enthalpy of formation from the elements,
molar volume, 312 340
standard state entropy, 312 formula unit, 340
thermal expansion, 310 heat capacity function, 340
Fe-tschermakite molar volume, 340
polyhedral distortion, 308 standard state entropy, 340
Fe3⫹: coordination, in melts, 420 thermal expansion, 341
Feitknechtite Fluor-richterite: polyhedral distortion, 308
Eh-pH stability limits, 558 Fluor-riebeckite: polyhedral distortion, 308
Formula unit, 561 Fluor-tremolite
Gibbs free energy of formation from the enthalpy of formation from the elements,
elements, 561 312
Feldspars formula unit, 312
cell edges, generalizations, 351 heat capacity function, 312
chemistry of, 346 molar volume, 312
intracrystalline disorder, 347 polyhedral distortion, 308
miscibility gap in, 360 standard state entropy, 312
mixing properties, 360 Forbidden zone
mixing properties and thermobarometry, and deviations from ideality, 664
389 in substitutional solid solutions, 663
occurrences, 347 Force
stoichiometry, 347 and polarization, 44
Stormer’s mixing model, 390 and potential, 9, 43, 123
structural classes, 347 dispersive, 44
structure of, 345 Formaldehyde. See CH2O
thermodynamic properties, 353 Formality: definition, 802
thermometric equation, 392 Forsterite
Ferri-annite amount in olivines of Earth’s mantle,
formula unit, 325 224
Gibbs free energy of formation from the bulk modulus, 236
elements, 342 coulomb energy, 237
Ferrosilite defect energies in, 194
bulk modulus, 278 dispersive energy, 237
enthalpy of formation from the elements, enthalpy of formation from the elements,
283 49, 237
heat capacity function, 283 heat capacity function, 238
in equilibrium with silica and magnetite, lattice energy, 49, 237
556 molar volume, 227
molar volume, 273, 278 repulsive energy, 237
P-T stability range, 280 silica excess in, 191
spatial group, 266 standard state entropy, 238
standard state entropy, 283 structural properties, 227
structural class, 269 thermal expansion, 234
structural properties, 273 thermodynamic properties of ␤ and ␥
thermal expansion, 278 polymorph, 239
Index / 875
Fraction of gas in steam: definition, 647 in equilibrium with plagioclase and

Fractional olivine, 397, 400
crystallization in equilibrium with plagioclase, orthopy-
definition, 471, 691 roxene, and quartz, 397, 400
trace partitioning in the presence of in equilibrium with plagioclase, siliman-
fluid, 702 ite, and quartz, 397, 399
ionic character, of Si-O bonds, 218 in equilibrium with rutile, ilmenite,
melting sillimanite, and quartz, 397, 401
and trace partitioning, 698 in equilibirum with spinel, sillimanite,
definition, 471 and quartz, 397, 401
trace partitioning in the presence of intercrystalline exchanges with clino-
fluid, 701 pyroxene, 378
time of residence, definition, 609 mixing properties, 260
Fractionation lines: significance of, 474 multicomponent mixing models, 265
Free ion polarizability, 28 structural simulation, 253
Frenkel defects structure of, 248
definition, 188 subregular Margules models for aluminif-
enthalpy of formation, in oxides, halides, erous binary joins, 264
and sulphides, 195 thermal expansion, 254
Frequency Gaseous
characteristic angular of an oscillator, mixtures, intermolecular potentials, 624
129 molecules
maximum angular of the Debye model, angular function of dipole orienta-
130 tion, 614
Fugacity characteristic energy, 615
and activity, 404 collision diameter, 615
and chemical potential, 619 efficacious volume, 625
and partial pressure, 615, 623 long-range interactions, 614
and thermodynamic activity, 615 potential energy, 614
definition, 615 short-range interactions, 614
coefficient Gedrite
and pressure, 616 formula unit, 300
definition, 615 molar volume, 304
Fulvic acids: definition, 562 structural class, 300
Functional groups structural properties, 304
in fulvic acids, 563 Generalized coordinates: in quantum
in humic acids, 562, 563 mechanics, 811
Geothermal
fluids, composition, 648
Gamma-nsutite gases
Formula unit, 561 composition, 656
Gibbs free energy of formation from the speciation in the C-H-O system, 649,
elements, 561 651–52
Garnet speciation in the N-H system, 649
cell edge and cationic occupancies, 250 speciation in the S-H system, 649
chemistry of, 249 thermobarometry, 648
compressibility, 254 Geothermometry
enthalpic interactions in binary mixtures, based on solid-fluid reactions, 402–7
262–63 based on solid-liquid equilibria, 447–48
heat capacity according to Kieffer’s based on solid-solid reactions, 395–402
model, 258 of intercrystalline exchange, 375–86
heat capacity calculations, 260 of intracrystalline exchange, 386–88
in equilibrium with cordierite, silliman- of solvus, 388–95
ite, and quartz, 397, 401 Gibbs equations, 99
876 / Index
Gibbs free energy structural class, 300

and chemical potential, 93, 449 structural properties, 304
and enthalpy-entropy, 98 thermal expansion, 310
and extrinsic stability, 96, 466 Gradient: of a vector, definition, 810
and miscibility, 450 Grossular
and standard potentials, 550 bulk modulus, 256
change, in redox equilibria, 539, 543 coulomb energy, 257
change, of a reciprocal reaction, 168 dispersive energy, 257
elastic strain term, 184 enthalpy of formation from the elements,
excess, of transition, 109 257
in relation with enthalpy and entropy, heat capacity function, 258
molar, of a phase, 94 molar volume, 254
of a compositionally non-uniform mix- repulsive energy, 257
ture, 179 structural properties, 254
of a gaseous mixture, 626 thermal expansion, 256
of a mechanical mixture, 451 third-law entropy, 258
of a phase, at P,T of interest, 118, 157 Groutite
of a silicate melt, 439 Eh-pH stability limits, 558
of an ideal mixture, 159 Formula unit, 561
of formation from the elements, by the Gibbs free energy of formation from the
⌬O2⫺ method, 151 elements, 561
of formation of aqueous solutes, 554 Gruneisen: thermodynamic parameter of,
of mixing, 158–59 59
ideal, for gaseous components, 623 Grunerite
in a regular mixture, 173 enthalpy of formation from the elements,
in a subregular mixture, 175 312
in melts, 414, 428, 446 formula unit, 312
multisite mixtures, 166 heat capacity function, 312
of silicate melt components, 441 molar volume, 312
of the associative process, 199 polyhedral distortion, 308
of the defect process, 193, 196 standard state entropy, 312
of unmixing, in solvus decomposition, thermal expansion, 310
178
standard partial molal of formation of
solutes, in the HKF model, 535–36 H2
total, of a system, 92 aqueous, thermodynamic properties, 653
Gibbs phase rule, 101 fugacity coefficient, P-dependence, 616
Glasses gas, thermodynamic properties, 626
enthalpy of, in relation with molten in geothermal fluids, 648
state, 434 solubility in water, 645
heat capacity of, in relation with molten H2CO3: reaction kinetics in solution, 512
state, 434 H 2O
Glauconite amount in amorphous silica, 373
formula unit, 325 as a solvent, 492
occurrences, 326 classical equation of state, 489
Glaucophane critical pressure, 619
enthalpy of formation from the elements, critical region, 488
312 critical temperature, 486, 619
formula unit, 300, 312 dielectric constant, 495
heat capacity function, 312 electron density distribution, 479
molar volume, 304, 312 fugacity coefficient, at various P and T,
polyhedral distortion, 308 617
standard state entropy, 312 ideal gas, thermodynamic properties, 626
Index / 877
ionic dissociation, 544 He

isobaric heat capacity in the critical re- aqueous, thermodynamic properties, 653
gion, 490 critical pressure, 619
isobaric thermal expansion in the critical critical temperature, 619
region, 490 Henry’s constant in silicate melts, 640
isothermal compressibility in the critical solubility in water, T effect, 643
region, 490 solution enthalpy in silicate melts, 641,
liquid saturation boundary, 484 642
mass distribution in the earth, 479 solution entropy in silicate melts, 641
maximum oxidation limit, 544 Heat: latent, of transformation, 112
maximum reduction limit, 544 Heat capacity
molecular structure, 479 and enthalpy, 131
non-classical equation of state, 488 and internal energy, 130, 131
redox stability, 544 excess, of transition, 109
solubility in silicate melts, 633, 636 isobaric, 131
solvent parameters in Debye-Hückel as a function of T, 141
equation, 495 determined by exchange reactions,
thermal profile along boiling curve, 647 147
vapor saturation boundary, 484 empirical estimates of, 141
H2S functional forms of, 132
aqueous, thermodynamic properties, 653 Haas-Fisher interpolation, 133, 156
critical pressure, 619 Maier-Kelley interpolation, 132
critical temperature, 619 isochoric
gas, thermodynamic properties, 626 and internal energy, 128
in geothermal fluids, 648 definition, 131
solubility in water, 645 in the Debye model, 131
H3PO4 : aqueous, thermodynamic proper- in the Kieffer model, 140
ties, 653 non-solvation contribution, in the HKF
Half-life: of a reaction, 591, 722 model, 530
Hamiltonian operator of silicate melt components, 436, 442
definition, 814 of solutes, in the HKF model, 530, 531
quantum-mechanic, 74, 124 of solution in melt, 427
and ionic energy gap, 26 of transition, 109
Hardness factor, 44 relationship between constant volume
Harmonic and constant pressure terms, 131
entropy, in the Kieffer model, 258 solvation contribution, in the HKF
oscillator, 9 model, 530
characteristic oscillation frequency of, Heat content. See Enthalpy
124 Hedembergite
equation of motion of, 123 bulk modulus, 278
force constant of, 48, 122, 137 coulomb energy, 284
frequency of, 123 dispersive energy, 284
potential energy of, 122 enthalpy of formation from the elements,
wave functions of, 126 282, 284
Hartree-Fock energy: in the MEG ap- heat capacity function, 282
proach, 83 lattice energy, 284
Hastingsite molar volume, 273, 278
formula unit, 300 P-T stability range, 280
structural class, 300 repulsive energy, 284
Hausmannite standard state entropy, 282
Eh-pH stability limits, 558 structural class, 268, 269
Formula unit, 561 structural properties, 273
Gibbs free energy of formation from the thermal expansion, 278
elements, 561
878 / Index
Helmholtz free energy, 58 formula unit, 312

and isotope separative effect, 728 heat capacity function, 312
and partition function, 728 molar volume, 312
by vibrational calculations, 731 standard state entropy, 312
Hematite thermal expansion, 310
Eh-pH stability limits, 556 Humic acids
Formula unit, 561 acidic-exchange capacity, 563
Gibbs free energy of formation from the definition, 562
elements, 561 Hund’s rule, 4
Henry Huttenlocher gap, 369
constant Hybrid orbitals: of silicon, 217
and equilibrium constant, 632 Hydrogen
and molar distribution, 646 bond, in water, 481
in gas-melt equilibria, definition, 632 ion approximation, 11
law isotopes
and activity coefficient, 685 ␣ factors, 784
and trace elements, 658 ␣ factors and temperature, 783, 786
compositional range of validity, 115 ⌬*AB factors, 783, 784
concentration limits in silicate melts, ⌬*AB factors and temperature, 784–85
680 salt effect, 787–89
for electrolytes in solution, 493 isotopic composition, 779
lower concentration limits in silicates, isotopic fractionation, 782–84
666 rotational partition function, 779
upper concentration limits in silicates, Hydrolysis
667 energy of, 508
Hermite polynomial, 125 enthalpy change involved in, 509
Heterogeneous nucleation, 179, 297 entropy change involved in, 509
HF: aqueous, thermodynamic properties, in relation with atomistic properties, 505
653 Hygromagmatophile element
Hf: isotopic composition, 744 as a fractionation index, 706
High albite definition, 702
compressibility, 354 in fractional crystallization processes,
360 in fractional melting processes, 704
formula unit, 352
heat capacity function, 360 Illite: formula unit, 325
intracrystalline disorder in, 350 Implicit functions: definition, 807
kinetics of formation, 362 Incompatible element: definition, 702
molar volume, 352, 360 Inert gas: electronic structure of, 10
standard state entropy, 360 Input flux, 610
structural properties, 352 Interaction parameter
thermal expansion, 354 and pair interactions, 163
High sanidine: P-T stability, 363 and trace element distribution, 660
HNO3: aqueous, thermodynamic properties, of a regular mixture, 163
653 of a simple mixture, 163
Holmquistite of silicate melt components, 444
formula unit, 300 of the quasi-chemical model, 164
structural class, 300 Intercumulus liquid: effect of, in trace parti-
Homogeneous nucleation, 179 tioning, 695–96
Hooke’s law, 9, 122 Interdiffusion
Hornblende coefficient, 384
enthalpy of formation from the elements, in olivine, 215
312 in unmixing processes, 181
Index / 879
Interface equilibrium: and trace parti- Isotopes

tioning, 690 and reaction kinetics, 733
Interionic distance, 17 closure temperature of exchanges, 738,
Internal energy 741
and lattice energy, 54 definition, 707
and pressure, 58 differences in vibrational spectra, 732
and temperature, 54 diffusion-induced fractionation, 735
and thermal expansion, 54 fractionation of, 726
and vibrational frequencies in the Ein- reduced partition function, 731, 733
stein model, 129 separative effect, 727, 729
of water, in relation with structure, 481 stable ⌬‰ variation, 770
Internal stable, ␣ fractionation factor, 767,
isochron, definition, 743 770–71
pressure of a gaseous phase, definition, stable, ⌬ parameter, 770
621 stable, standard reference materials, 770
39
Interstitial solid solution: of a trace compo- 18Ar
nent in a crystal, 657 half life, 751
Intrinsic disorder: in crystals, 187–89 production by neutron irradiation,
Intrinsic instability, 297 750
40
field, of a mixture, 179, 182 18Ar
Ionic radiogenic excess component, 748
14
couplings 6C
and complexation model, 491 decay process, 766
terminology, 491 detection limit, 766
polarizability, 26, 87 half life, 766
and polarization energy, 88 production in the earth’s atmosphere,
radii, 16 765
and coordination, 32 radioactivity calibration, 766, 768
18
and ionization potentials, 32 9F: decay process, 719
2
in VI-fold coordination, 20 1H: discovery of, 779
3
of Shannon, 37 1H
strength, 494, 498, 516 concentration in the earth’s atmo-
equivalent stoichiometric, 502 sphere, 779
Ionicity factor, 56–57 decay process, 779
Ionization potential, 52 half life, 779
data, 24 production in the earth’s atmosphere,
definition, 14 779
39
role of, in defect equilibria, 195 19K: artificially induced decay, 750
40
Isobars: definition, 707 19K
Isochron equation decay process, 719

207
Pb-206Pb, 758, 761 double decay process, 745, 748
232
Th-204Pb, 758 half life, 747
138
235
U-204Pb, 758 57La: decay process, 720, 745
176
238
U-204Pb, 758 71Lu
K-Ar, 748, 749 decay process, 744

K-Ca, 756, 757 half life, 744
14
La-Ce, 745, 746 8O: decay process, 719
18
Lu-Hf, 746 O
Rb-Sr, 742 diffusional exchanges between feld-
Re-Os, 763, 764, 765 spar and fluid, 738
Sm-Nd, 744, 746 self-diffusion in feldspars, 735
18
Isomers: definition, 707 O/16O: diffusion-induced fractionation
Isotones: definition, 707 in adularia, 736
880 / Index
Isotopes (continued) KAlSi3O8

87 heat capacity function, 357
37Rb
decay process, 742 incongruent melting of, 361
half life, 742 ordering parameter, 357
187
75Re
Kaolinite
decay process, 763 in equilibrium with Mg-chlorite and
half life, 763 fluid, 584
147
62Sm: decay process, 744
in equilibrium with muscovite and fluid,
232 583
90Th: decay series, 710–12, 758
238 Katophorite
92U: decay process, 720–21
235 formula unit, 300
92U: decay series, 710–12, 758
structural class, 300
Khoharite
Jadeite bulk modulus, 256
activity coefficient, in melts, 428 coulomb energy, 257
bulk modulus, 278 dispersive energy, 257
coulomb energy, 284 enthalpy of formation from the elements,
dispersive energy, 284 257
enthalpy of formation from the elements, heat capacity function, 258
282, 284 lattice energy, 257
heat capacity function, 282 molar volume, 254
lattice energy, 284 repulsive energy, 257
molar volume, 273, 278 structural properties, 254
repulsive energy, 284 thermal expansion, 256
standard state entropy, 282 third-law entropy, 258
structural class, 268, 269 Kinetic
structural properties, 273 of crystallization, 471
thermal expansion, 278 of intercrystalline exchanges, 381
Jahn-Teller effect, 79 of intracrystalline exchanges, 388
and ligand field interpretation, 80 rate constant
in transition elements, 80 compositional dependence of, 587
Johannsenite: structural class, 268 of a backward reaction, 734
of a forward reaction, 591, 734
K P-T dependence of, 598
diffusivity in silicate melts, 690 rate laws
isotopic composition, 745 based on transition state theory, 596
K-feldspar first order, 591, 722
compressibility, 354 high order, 593
enthalpy of formation from the elements, of dissolution of minerals, pH depen-
360 dence, 594
heat capacity function, 360 of the activity terms, 593
hydrolysis of, 578 simple, in meq or ceq, 590
in equilibrium with Mg-chlorite and zero-order, 590
fluid, 585 Kirkendall effect, 213–14
in equilibrium with muscovite and fluid, Kirschsteinite
molar volume, 360 solubility in Fayalite and Larnite, 245
permutability calculation, 167 structural properties, 227
standard state entropy, 360 thermal expansion, 234
thermal expansion, 354 Knorringite
Kaersutite bulk modulus, 256
formula unit, 300 coulomb energy, 257
structural class, 300 crystal field stabilization energy, 257
Index / 881
dispersive energy, 257 LCAO approximation, 26, 74–76, 217, 479

enthalpy of formation from the elements, Legendre transforms, 98
257 Lepidolite. See Polylithionite
heat capacity function, 258 Leucite: melting temperature, 465
lattice energy, 257 Lever rule, 468
molar volume, 254 application, 454
repulsive energy, 257 in ternary systems, 475
structural properties, 254 Li: muscovite/fluid partitioning, 665
thermal expansion, 256 Liebembergite
third-law entropy, 258 activity of, in Mg-Ni mixtures, 245
Kozulite amount in natural olivines, 224
formula unit, 300 coulomb energy, 237
structural class, 300 dispersive energy, 237
Kr enthalpy of formation from the elements,
aqueous, thermodynamic properties, 653 49, 237
Henry’s constant in silicate melts, 640 heat capacity function, 238
solubility in water, T effect, 643 lattice energy, 49, 237
solution enthalpy in silicate melts, 641 molar volume, 227
solution entropy in silicate melts, 641 repulsive energy, 237
Kyanite standard state entropy, 238
extrinsic stability, 104 structural properties, 227
in equilibrium with sillimanite, 397–98 Ligand field theory, 73–79
and molecular orbital energy, 77, 78
La: isotopic composition, 745 applied to sulphur compounds, 78
Lagrange: equation of motion, 813 Limit
Lambda transition: definition, 107 of Dulong and Petit, 127, 132
Landau theory, 109–13 of predominance of a species in solution,
applied to feldspars, 356 510
Laplacian operator, 8 Lindemann: equation of, 211
definition, 811 Liquidus
in Schrödinger’s equation, 814 curve, definition, 456
Larnite surface, significance of, 467
amount in natural olivines, 224 Local equilibrium, 119–21
coulomb energy, 237 definition, 92
dispersive energy, 237 Low albite
enthalpy of formation from the elements, enthalpy of formation from the elements,
49, 237 360
heat capacity function, 238 formula unit, 352
lattice energy, 49, 237 heat capacity function, 360
molar volume, 227 intracrystalline disorder in, 350
polymorphic transitions, 234 molar volume, 352, 360
repulsive energy, 237 standard state entropy, 360
structural properties, 227 Low sanidine: P-T stability, 363
Lattice Lu: isotopic composition, 744
defects, 185
distortion Macroperthites: definition, 363
linear propagation of, 666 Macroscopic order parameter, 109, 112
energy, 45, 48 Madelung constant: and coulombic poten-
and Born-Haber cycle, 52 tial, 45
and enthalpy, 53 Magnetite
of olivine end-members, 49 Eh-pH stability limits, 556
semiempirical estimates, 51 Formula unit, 561
882 / Index
Magnetite (continued) MEG model, 81–86

Gibbs free energy of formation from the and interionic distances in ionic solids,
elements, 561 85
in equilibrium with pyrite and gas, 404 and lattice energy of ionic solids, 85
Malachite application to carbonates, 85
Eh-pH stability limits, 560 main assumptions, 81
Formula unit, 561 Melts
Gibbs free energy of formation from the compressibility, 437–39, 443
elements, 561 heat capacity of, 434–36
Manganite in univariant equilibrium with solids,
Eh-pH stability limits, 558 446–48
Formula unit, 561 miscibility gap in, 439–41, 445
Gibbs free energy of formation from the miscibility of, 439–41
elements, 561 molar volume, 437, 442–43
Margarite structure of, 417–19, 422
compressibility, 341 thermal expansion, 443
enthalpy of formation from the elements, Mendeleev: periodic chart, 10
341 Mesoperthites: definition, 363
extrinsic stability, 336 Metal organic
formula unit, 325, 340 complexes, thermodynamic properties,
Gibbs free energy of formation from the 571
elements, 342 complexation, 560–75
heat capacity function, 341 in the HKF model, 564–75
in equilibrium with anorthite and corun- Metastability: field, of a mixture, 179, 182
dum, 336 Metastable equilibrium: definition, 92
in equilibrium with anorthite, quartz and Methane. See CH4
Al2SiO5, 336 Mg-chlorite
in equilibrium with zoisite, quartz and in equilibrium with K-feldspar and fluid,
Al2SiO5, 336 585
molar volume, 324, 341 in equilibrium with kaolinite and fluid,
polytypes, 324 584
standard state entropy, 341 in equilibrium with muscovite and fluid,
structural class, 324 584
thermal expansion, 341 Mg-olivine. See Forsterite
Margules: subregular parameters, 170 Mg-riebeckite
Maunder minimum: and 146C radioactivity, enthalpy of formation from the elements,
767 312
Mean formula unit, 312
activity coefficient heat capacity function, 312
of REE in silicate melts, 676, 679 molar volume, 312
concentration standard state entropy, 312
of a trace element in a fractionally thermal expansion, 310
crystallized solid, 692 Mg-Tschermak: structural class, 269
excitation, 44 Micas
salt method, 497–99 chemistry of, 323
time of residence extrinsic stability of, 330, 335
definition, 609 hydroxyls-fluorine exchanges in, 344
of elements in seawater, 610 P-T stability, 325, 326
Mechanical solubility polytypes, 321, 335
and density, in a melt, 641 stoichiometry, 323
and Henry’s law range, 632 structure of, 321
definition, 641 thermodynamic properties, 339
of a gas in aqueous solutions, 643 Microcline
of noble gases in melts, 631 compressibility, 354
Index / 883
enthalpy of formation from the elements, (Fe,Mn)2SiO4: mixing properties, 246

360 (Fe,Mn)2Al2Si3O12: interaction parame-
formula unit, 352 ters, 264
heat capacity function, 360 Fe2Si2O6-CaMgSi2O6: interaction param-
intrinsic disorder in, 350 eters, 290
molar volume, 352, 360 H2O-CO2 gaseous, mixing properties,
P-T stability, 363 624
standard state entropy, 360 (K,Ca)(Al,Si)4O8
structural properties, 352 interaction parameters, 369
thermal expansion, 354 mixing properties, 369
Microperthites: definition, 363 K(Mg,Fe)3AlSi3O10(OH)2: mixing
Miscibility gap properties, 344
in feldspars, 364 Mg2Si2O6-CaFeSi2O6: interaction
in silicate melts, 443, 674 parameters, 290
in thernary feldspars, 360 Mg2Si2O6-NaAlSi2O6: interaction
of carbonates, 177 parameters, 290
Mixture (Mg,Ca)2Si2O6: interacton parameters,
athermal, 162 290
definition, 92 (Mg,Ca)2SiO4: miscibility gap in, 242
ideal, 157 (Mg,Ca)3Al2Si3O12
mechanical, crystallization of, 451, 461 entropy of mixing, 262
mechanical, nature of, 451 interaction parameters, 264
regular of the Zeroth principle, 162 volume of mixing, 263
regular, application to silicate melts, 439 (Mg,Fe)2Si2O6: interaction parameters,
simple, 163 290
subregular, 169 (Mg,Fe)2SiO4
symmetric, 169 enthalpy of mixing, 240
ternary, crystallization path in a closed equation of state, 235
system, 470 Gibbs free energy, 242
Ca3(Al,Fe3⫹)2Si3O12: mixing properties, intracrystalline disorder, 241
263 mixing in molten state, 431
Ca3(Cr,Al)2Si3O12: calorimetric enthalpy, polymorphic transitions, 235, 239
263 (Mg,Fe,Ca)2SiO4: mixing model, 246
CaFeSi2O6-NaAlSi2O6: interaction (Mg,Mn)2SiO4: mixing properties, 245
parameters, 290 (Mg,Mn)3Al2Si3O12
(Ca,Fe)3Al2Si3O12: entropy of mixing, consolute temperature, 260
262 interaction parameters, 264
CaMgSi2O6-CaAl2SiO6: interaction (Mg,Ni)2SiO4
parameters, 290 configurational entropy of, 243
Ca(Mg,Fe)Si2O6: interaction parameters, mixing properties, 244
290 volume of mixing, 244
CaMgSi2O6-NaAlSi2O6: interaction NaAlSi2O6-CaAl2SiO6: Redlich-Kister
parameters, 290 interaction parameters, 292
(Ca,Mn)3Al2Si3O12 (Na,Ca)(Al,Si)4O8. See Plagioclases
interaction parameters, 264 (Na,Ca)Al2(Si,Al)4O10(OH)2: mixing
mixing properties, 263 properties, 345
(Co,Fe)2SiO4: mixing properties, 246 (Na,K)2Si3AlO10(OH)2
Fe3(Al,Fe)2Si3O12: mixing properties, 263 interaction parameters, 344
(Fe,Ca)2SiO4: mixing properties, 245 volume of mixing, 343
(Fe,Ca)3Al2Si3O12 (Na,K)AlSi2O6: mixing in molten state,
interaction parameters, 264 428
volume of mixing, 263 (Na,K)AlSi3O8
(Fe,Mg)3Al2Si3O12: interaction parame- interaction parameters, 364, 391
ters, 264 miscibility gap, 361
884 / Index
Mixture (continued) antibonding, 75

mixing in molten state, 428, 430 bonding, 75
mixing properties, 361 energy of, 74
volume of mixing, 361 energy of, in the SiO44⫺ cluster, 218
(Na,K)AlSi4O10: mixing in molten state, for the H2O molecule, 480
428 in the LCAO approximation, 76
(Na,K)AlSi5O12: mixing in molten state, Monalbite
428 intrinsic stability, 350
(Na,K,Ca)(Al,Si)4O8 P-T stability, 362
interaction parameters, 371 Monticellite
mixing model, 370 molar volume, 227
mixing properties, 369 natural occurrences, 224
Mn solubility in Larnite and Forsterite, 242
olivine/melt partitioning, 675 structural properties, 227
thermodynamic properties of organic thermal expansion, 234
complexes, 571 Morse potential, 125
Mn-clinopyroxene Multiplicity
molar volume, 274 definition, 221
structural class, 269 in amphiboles, 299
structural properties, 274 in feldspars, 345
Mn-Cr garnet in micas, 321
bulk modulus, 256 in pyroxenes, 266
coulomb energy, 257 Muscovite
crystal field stabilization energy, 257 compressibility, 341
dispersive energy, 257 enthalpy of formation from the elements,
257 extrinsic stability, 332
heat capacity function, 258 formula unit, 325, 340
lattice energy, 257 Gibbs free energy of formation from the
molar volume, 254 elements, 342
repulsive energy, 257 heat capacity function, 340
structural properties, 254 in equilibrium with K-feldspar and fluid,
thermal expansion, 256 584
third-law entropy, 258 in equilibrium with kaolinite and fluid,
Mn-olivine. See Tephroite 583
Mn3⫹: in pyroxenes, 267 in equilibrium with Mg-chlorite and
Modal melting: definition, 696 fluid, 584
Modified Electron Gas model. See MEG in equilibrium with quartz, K-feldspar,
model and Al2SiO5, 333
Molality in equilibrium with sanidine and corun-
definition, 521, 802 dum, 332
mean ionic of electrolytes in solution, molar volume, 324, 340
494 occurrences, 326
Molar P-T stability, 332
fraction, definition, 802 polytypes, 324
properties, definition, 157 standard state entropy, 340
volume, of silicate melt components, 442 structural class, 324
Molarity: definition, 802 thermal expansion, 341
Mole: definition, 802
Molecular
diffusion, definition, 608 N2
orbitals, 73 aqueous, thermodynamic properties, 653
and atomic orbitals, 75 critical pressure, 619
Index / 885
critical temperature, 619 thermodynamic properties of organic

fixation by blue-green algae, 578 complexes, 572
gas, thermodynamic properties, 626 Ni-olivine. See Liebembergite
in geothermal fluids, 648 Ni2⫹: coordination state in alkaline glasses,
solubility in water, 643, 645 423
Na: diffusivity in silicate melts, 690 NO3⫺: metastability in aqueous solution,
Na-phlogopite 578
compressibility, 341 Noble gases
enthalpy of formation from the elements, Henry’s constant in melts, 641
340 Henry’s constant in water, 644
formula unit, 325, 340 kinetic radii, 641
Gibbs free energy of formation from the solubility in silicate melts, 640
elements, 342 solubility in water, T effect, 643
heat capacity function, 340 solution enthalpy in silicate melts, 641
molar volume, 340 solution entropy in silicate melts, 641
standard state entropy, 340 Non-degenerate ground state, 128
thermal expansion, 341 Non-modal melting: definition, 696
NaAlSi3O8 Non-polar covalent radii, 21
enthalpy of transition, 354 Non-solvation contribution: in the HKF
heat capacity function, 359 model, 522
ordering parameter, 354 Non-stoichiometry: in crystals, 202
substitutional and displacive disorder in, Normal hydrogen electrode, 540
356 Normalisation factor: in vibrational eigen-
transition entropy, 357 functions, 125
Nd: isotopic composition, 744 Normalized partition coefficient: definition,
Ne 681
aqueous, thermodynamic properties, 653 Nucleation and growth, 181
Henry’s constant in silicate melts, 640 in pyroxenes, 295
solubility in water, T effect, 643 Nucleons
solution enthalpy in silicate melts, 641 binding energy, 714
solution entropy in silicate melts, 641 intrinsic angular momentum, 714
Nepheline: solubility of SiO2 in, 465 orbital angular momentum, 714
Nernst law, 543 Nucleosynthesis, 711
concentration range of, 659 Nuclides
limits, in silicate/melt REE partitioning, decay series, 708
669 definition, 707
upper concentration limits for REE in magnetic moment, 714
silicate crystals, 669 nuclear energy, 717, 718
Neutrino: and energy conservation prin- origin of, 708
ciple, 717 stability of, 708, 710, 715
Neutron
mass of, 707 O2
capture, 708, 710 aqueous, thermodynamic properties, 653
Newton: laws of, 811–12 buffer
NH3 hematite-magnetite, 403, 628
aqueous, thermodynamic properties, 653 hematite-wuestite, 629
gas, thermodynamic propeties, 626 iron-wuestite, 402
in geothermal fluids, 648 magnetite-wuestite, 402
solubility in water, 645 quartz-fayalite-magnetite, 402, 628
NH4⫹: metastability in aqueous solution, Ni-NiO, 402, 629
578 fugacity coefficient, P-dependence, 616
Ni thermodynamic properties, 626
clinopyroxene/melt partitioning, 682–83 Occlusion: of a trace element, 657
886 / Index
Octahedral site preference energy. intracrystalline exchange geothermome-

See OSPE try, 386
Olivine stoichiometric calculations, 394
cell volume, 228 Os: isotopic composition, 762
chemistry of, 224–25 OSPE
compressibility, 234 according to Dunitz-Orgel, 71
in Earth’s mantle, 224 according to McClure, 71
Mg-Mn interdiffusion in, 216 and intracrystalline disorder in spinels, 71
natural and synthetic end members, 227 and splitting parameter, 70
natural occurrences, 224 Ostwald coefficient: definition, 632
structure of, 223, 226 Output flux, 610
T-stability and size mismatch, 234 Overall
thermal expansion, 233 reaction progress variable, 120
Omphacite: structural class, 268 reaction rate and individual velocities,
Opal 589
molar volume, 373 Oxides
occurrences, 371 acidity and basicity in melts, 417
rates of dissolution in seawater, 592 amphoteric properties in melts, 419
Optical Lux-Flood character, 418
continuum, in the Kieffer model, 258 Oxidizing medium: definition, 539
branches, 137 Oxygen
polarizability, of silicates, 89 bridging, in melts, 412, 420, 422
vibrational modes, definition, 137 free, in melts, 412
Orbital degeneracy: and field strength, 72 fugacity
Organic ligands: thermodynamic properties, and speciation in gaseous phase, 630
568 role of, in trace element partitioning,
Orthoclase 687
compressibility, 354 isotopes
formula unit, 352 ⌬AB‰ factors, 773–76, 778, 780
molar volume, 352 isotope salt effect, 787, 788
structural properties, 352 modified increment method, 774
thermal expansion, 354 normalized 18O-increment, 774, 777
Orthodiopside reduced partition function, 776
bulk modulus, 278 isotopic
enthalpy of formation from the elements, composition, 771
283 fractionation, 771
heat capacity function, 283 non-bridging, in melts, 412, 420
molar volume, 278 partial pressure
standard sate entropy, 283 and defect equilibria in fayalite, 204
structural class, 269 of intrinsic stability, 189
thermal expansion, 278 reduced partition function, 772–73
Orthoenstatite: P-T stability range, 279 singly bonded. See non-bridging
Orthohedembergite
bulk modulus, 278 Paragonite
enthalpy of formation from the elements, compressibility, 341
283 enthalpy of formation from the elements,
molar volume, 278 extrinsic stability, 334
standard state entropy, 283 formula unit, 325, 340
structural class, 269 Gibbs free energy of formation from the
thermal expansion, 278 elements, 342
Orthopyroxene heat capacity function, 341
in equilibrium with olivine and quartz, in equilibrium with albite and corun-
397, 398 dum, 334
Index / 887
in equilibrium with albite, quartz, and Peritectic

Al2SiO5, 335 curve: significance of, in ternary systems,
mixing with muscovite, 343 468
molar volume, 324, 341 point
polytypes, 324 energetic significance of, 463
standard state entropy, 341 in the (Na,K)AlSi3O8 system, 457
structural class, 324 Permutability
thermal expansion, 341 and lattice distortion, 663
Pargasite calculations, in plagioclases, 366
enthalpy of formation from the elements, definition, 158
312 Perthites: definition, 363
formula unit, 300, 312 pH: definition, 545
heat capacity function, 312 Phase
molar volume, 304, 312 definition, 91
standard sate entropy, 312 transitions
structural class, 300 and Landau theory, 109
structural properties, 304 first order, 111
thermal expansion, 310 second order, 110
Partial tricritical, 111
distribution constant, for trace elements Phengite: formula unit, 325
in multisite solutions, 661 Phlogopite
molar properties, definition, 520 compressibility, 341
molar enthalpy, as T- derivative of chemi- enthalpy of formation from the elements,
cal potential, 101 340
molar entropy, definition, 100 extrinsic stability, 330
molar properties, definition, 100 formula unit, 325, 340
of trace elements in solution, 686 Gibbs free energy of formation from the
Particles elements, 342
kinetic energy of, 5 heat capacity function, 340
potential energy of, 5 in equilibrium with leucite, kalsilite and
Partition function, 127 forsterite, 330
and isotopic fractionation, 726, 730 in equilibrium with quartz, sanidine and
in transition state theory, 597 enstatite, 330
rotational, 727 molar volume, 324, 340
translational, 726 occurrences, 323, 326
vibrational, 727 P-T stability, 328, 330
Pauli: exclusion principle, 4 polytypes, 324
Pauling standard state entropy, 340
electrostatic valence principle, 33, structural class, 324
350 thermal expansion, 341
univalent ionic radii, 19 Phonon dispersion: diagram of, 137
third rule, 223 Photoelectric effect, 720
Pb: thermodynamic properties of organic Photosynthesis: role of, in redox equilibria,
complexes, 574 575
Periodicity Pigeonite
definition, 222 P-T-X stability in the quadrilateral, 293
in amphiboles, 299 spatial group, 266
in feldspars, 345 structural classes, 268
in micas, 321 Pigeonites: chemistry of, 268
in pyroxenes, 266 Plagioclases
Peristerites closed-system crystallization, 456
definition, 351 compressibility, 354
miscibility gap, 369 enthalpy of mixing, 365
entropy of mixing, 367
888 / Index
Plagioclases (continued) of equality of chemical potentials at equi-

Gibbs free energy of mixing, 367 librium, 95, 403, 449, 452, 579–80
interaction parameters, 369 of minimization of Gibbs free energy at
intracrystalline disorder, 351 equilibrium, 449, 452
mixing properties, 364 of the corresponding states, 620
open-system crystallization, 471 of radius ratio, 42
phase transitions, 366 of resonance, in MOs energy calcula-
significance of compositional zoning, tions, 74
472 Process: of solvation-hydrolysis, 504
thermal expansion, 354 Progress variable: in kinetic processes, 589
Point defects Protoenstatite
and trace partitioning, 666 bulk modulus, 278
energy of formation, 192 enthalpy of formation from the elements,
in crystals, 187 283
Poisson’s ratio, 184 heat capacity function, 283
Polarization energy, 87 molar volume, 278
in defect processes, 194 P-T stability range, 279
in orthosilicates, 89 standard state entropy, 283
in the Mott-Littleton approximation, 88 structure of, 280
Polarizing power: definition, 504 thermal expansion, 278
Polyhedral compressibility, 61 Proton
Polylithionite magnetic moment, 714
formula unit, 325 mass of, 707
Gibbs free energy of formation from the Pseudo-solvus. See Miscibility gap
elements, 342 Pyralspites, 250
molar volume, 324 Pyrite
occurrences, 326 buffering effect on S bearing fluids, 652
polytype, 324 Eh-pH stability limits, 556
structural class, 324 Formula unit, 561
Polymerization Gibbs free energy of formation from the
constant, 413, 419 elements, 561
of aqueous organic polymers, 567 Pyrochroite
reactions in melts, 416, 422 Eh-pH stability limits, 558
Polymorphs: in Geology, 104 formula unit, 561
Polytypism: in micas, 321 Gibbs free energy of formation from the
Potassic elements, 561
arfvedsonite, polyhedral distortion, 308 Pryolusite
ferri-taramite, polyhedral distortion, 308 Eh-pH stability limits, 558
ossi-kaersutite, polyhedral distortion, Formula unit, 561
308 Gibbs free energy of formation from the
pargasite, polyhedral distortion, 308 elements, 561
Potential well: for the H2 molecule, 127 Pyrope
Pre-exponential activity by ionic solution model, 379
factor, in diffusivity equation, 382 bulk modulus, 256
term coulomb energy, 257
in defect equilibria, 196 dispersive energy, 257
in diffusion equations, 207 enthalpy of formation from the elements,
Principle 257
of additivity, in the HKF model, 521, heat capacity function, 258, 261
of detailed balancing, 595 molar volume, 254
of equal reactivity of co-condensing func- repulsive energy, 257
tional groups, 413, 567 structural properties, 254
Index / 889
thermal expansion, 256 decay energy, 719

third-law entropy, 258 disintegration probability, 722
Pryoxenes electron capture, 720
chemistry of, 267, 269 gamma transitions, 720
effect of P on structure, 278 half life, 713, 722
effect of T on structure, 277 internal conversion, 721
exolutive processes, 295 natural occurrence, 713
interionic distances and site occupancies, non-equilibrium, 725, 726
274–75 rate of production, 723
mixing properties and thermobarometry, secular equilibrium, 725
392 transient equilibrium, 724, 725
occurrences of, 269 Raoult’s Law
P-T stability, 277 and trace elements, 657
quadrilateral, 272–74 compositional range of validity, 115
mixing properties, 281 Rate constant: and equilibrium constant,
multicomponent mixing model, 294 121
stability relations, 278 Rational activity coefficient, 114
stochiometry rules, 269 Rayleigh
structure of, 266 distillation process of, 472
thermodynamic properties, 280 crystallization. See Fractional crystalli-
thermometric equations, 394 zation
Pyrrothite Rb
Eh-pH stability limits, 556 isotopic composition, 741
Formula unit, 561 nepheline/fluid partitioning, 663, 665
Gibbs free energy of formation from the Rb-Feldspar
elements, 561 formula unit, 352
molar volume, 352
Quantum numbers structural properties, 352
and periodic properties of the elements, Re: isotopic composition, 762
10 Reactive solubility
magnetic, 3 of a gas in aqueous solutions, 643
orbital or azimuthal, 3 of CO2 in melts, 631
principal, 3 Reciprocal reactions, 168
spin, 3 Redox
vibrational, 9 equilibria
Quartz and biologic activity, 577
␣-bulk modulus, 374 definition, 539
␣-thermal expansion, 374 Fe-Cu, 541
␤-bulk modulus, 374 in natural solutions, 553
␤-thermal expansion, 374 Zn-Cu, 542
coupled substitutions in, 371 parameters
enthalpy of formation from the elements, pe, 551
374 rH, 552
heat capacity function, 374 rO, 552
molar volume, 372 potential: and standard potential,
occurrences, 371 543–44
standard state entropy, 374 Reduced
structural properties, 372 mass, 123, 259
P, T and V, definition, 617
Radionuclides Reducing medium: definition, 539
alfa decay, 720 REE
average life, 723 as amphoteric oxides, 675
beta decay, 716, 719 clinopyroxene/melt partitioning, 674, 682
890 / Index
REE (continued) standard state entropy, 360

coordination state in glasses, 423 structural properties, 352
fractionation in silicate melts, 677–78 thermal expansion, 354
in natural olivines, 226 Schottky defects
pyrope/melt partitioning, 670 concentrations in NaCl and MgO, 197
radiometric dating, 744 definition, 187
seawater concentration patterns, 606 energy of formation, 193
solubility in silicate melts, 675 enthalpy of formation, in oxides, halides,
Relative and sulphides, 195
predominance limits of aqueous species, in forsterite, 193
546, 559 Schrödinger equation, 123
rate of a reaction, 120, 589 for one-dimensional stationary states, 9,
Repulsion factor, 44 124
Repulsive for three-dimensional stationary states, 8
potential, 43 Schwarz: theorem of, 806
state. See Molecular orbitals Screw dislocation, 185
Rhodochrosite Seawater
Eh-pH stability limits, 560 acidity, 602
Formula unit, 561 carbonate alkalinity, 602
Gibbs free energy of formation from the chemical variations with depth, 606
elements, 561 chlorinity, 601
Richterite composition, 600, 601, 604
enthalpy of formation from the elements, density, 602
312 redox state, 602
formula unit, 300, 312 salinity, 601
heat capacity function, 312 steady-state
molar volume, 312 chemistry, 603
standard state entropy, 312 models, 607
structural class, 300 Secular determinant, 138
Riebeckite Segré chart, 708, 710
formula unit, 300 Sericite: occurrences, 325
structural class, 300 Short-range
Rn: aqueous, thermodynamic properties, 653 interactions, in the MEG approach, 84
Roozeboom diagrams, 454–56, 459–63 repulsion, 43
Rotation: of a vector, definition, 810 Si isotopes
separative effect, 730
S isotopic fractionation, 730
Eh-pH stability limits, 554 SI system
isotopes base units, 801
␣ fractionation factors and tempera- conversion to, 802
ture, 795–97 derived units,801
⌬AB fractionation factors, 794 main physical constants, 804
effect of Eh on fractionation, 798–99 use in geochemistry, 801
effect of pH on fractionation, 799 Si4⫹: coordination, in melts, 420
isotopic composition, 794 Siderite
production through enzymatic media- Eh-pH stability limits, 556
tion, 578 Formula unit, 561
Sanidine Gibbs free energy of formation from the
compressibility, 354 elements, 561
enthalpy of formation from the elements, Siderophyllite
360 compressibility, 340
formula unit, 352 enthalpy of formation from the elements,
molar volume, 352, 360 formula unit, 325, 340
Index / 891
Gibbs free energy of formation from the coherent, in feldspars, 363

elements, 342 energetic significance in Roozeboom
heat capacity function, 341 diagrams, 459
molar volume, 341 field, definition, 174
standard state entropy, 341 thermobarometry, 388
thermal expansion, 341 Speciation
Silica calculations, 502, 512
aqueous, thermodynamic properties, 653 definition, 491
dissolution kinetics in water, 596 in volcanic gases, 631
polymers, in melts, 412 Specific
polymorphs, 371 heat. See Heat capacity
phase transition, 373, 463 interactions model, 492
transition energies, 375 Spectrochemical series, 68
Sillimanite: extrinsic stability, 104 Spessartine
Site bulk modulus, 256
dimension, and trace element parti- coulomb energy, 257
tioning, 672 dispersive energy, 257
interaction parameters, in orthopyrox- enthalpy of formation from the elements,
ene, 387 257
activity coefficient, 166 heat capacity function, 258
Skiagite lattice energy, 257
coulomb energy, 257 repulsive energy, 257
crystal field stabilization energy, 257 structural properties, 254
dispersive energy, 257 thermal expansion, 256
enthalpy of formation from the elements, third-law entropy, 258
257 Spinodal decomposition
heat capacity function, 258 in a simple mixture, 176
lattice energy, 257 in a subregular mixture, 176
molar volume, 254 in pyroxenes, 295, 299
repulsive energy, 257 of silicate melts, 444
structural properties, 254 Spinodal field
thermal expansion, 256 chemical, in feldspars, 363
third-law entropy, 258 coherent, in feldspars, 363
Sm definition, 175
isotopic composition, 744 Spinode
pyrope/melt partitioning, 672 in van der Waals acception, 175
Sn location on Gibbs free energy curves,
isotope fractionation, 730, 733 178
isotopes, separative effect, 730 Spodumene
So2 molar volume, 273
aqueous, thermodynamic properties, 653 structural class, 268, 269
critical pressure, 619 structural properties, 273
critical temperature, 619 Sr
solubility in silicate melts, 639 isotopic composition, 742
SO3 plagioclase/melt partitioning, 681, 682
critical pressure, 619 sanidine/fluid partitioning, 665
critical temperature, 619 Stability: of polymorphs, 103–6
Solidus curve: definition, 456 Stabilized orbitals, 67
Solution: definition, 92 Stable equilibrium: definition, 92
Solvation: energy of, 508 Standard
in the HKF model, 522–23 molar entropy of transition, 108
Solvus molar volume of transition, 108
chemical, in feldspars, 363, 389 redox potentials, 540–41
892 / Index
Standard state Tenorite

definition, 113 Eh-pH stability limits, 560
for aqueous solutes, 492, 520 Formula unit, 561
for the solvent, 520 Gibbs free energy of formation from the
in the mixture notation, 159, 451 elements, 561
of Henry’s law, 658 Tephroite
types of, 114 bulk modulus, 236
State function: definition, 97, 806 coulomb energy, 237
Static polarizability: of silicates, 89 dispersive energy, 237
Stirling approximation, 158, 597, 663 enthalpy of formation from the elements,
Stishovite 49, 237
bulk modulus, 374 heat capacity function, 238
enthalpy of formation from the elements, lattice energy, 49, 237
heat capacity function, 374 natural occurrences, 224
molar volume, 372 repulsive energy, 237
occurrences, 371 standard state entropy, 238
structural properties, 372 Th: isotopic composition, 757
thermal expansion, 374 Thermal barrier
Stockmayer: static potentials for gaseous energetic significance of, 461
molecules, 614, 624 petrogenetic significance of, 466
Strain energy: in pyroxenes, 269 Thermal expansion
Strain-free solvus. See Chemical solvus and heat capacity, 55
Sublimation energy, 52 non-solvation term, in the HKF model,
Substitutional solid solution, 657 529
Sulfates: reduction by enzymatic mediation, of crystalline solids
578 isobaric linear coefficient, 55
Surface isobaric volumetric coefficient, 55, 56
adsorption, of a trace element, 657 mean polyhedral linear coefficient,
energy, of atomic nuclei, 715 55
System of silicate melts, 437
definition, 91 of solutes, in the HKF model, 529
CaMgSi2O6-CaAl2Si2O8-NaAlSi3O8 solvation term, in the HKF model, 529
crystallization in open conditions, Thermobarometry
473, 699 of intercrystalline exchange, 375–86
melting in open conditions, 474 of intracrystalline exchange, 386–88
phase relations, 473 of solid-fluid equilibria, 402–7
CaMgSi2O6-CaAl2Si2O8 of solid-melt equilibria, 446–48
eutectic temperature, 454 of solid-solid equilibria, 395–402
phase relations, 453 of solvus, 388–95
KAlSiO4-SiO2 Thermochemical cycle, 52
eutectic point, 466 Thermodynamic
peritectic point, 465 activity
phase relations, 465 and chemical potential, 114, 583
NaAlSiO4-SiO2 and fugacity of a gaseous component,
eutectic point, 463, 465 615
peritectic point, 465 definition, 113
phase relations, 463 in an ideal mixture, 160
in reciprocal mixtures, 168
Taeniolite: formula unit, 325 mean ionic of electrolytes in solution,
Taramite 494
formula unit, 300 of a component in a multisite mix-
structural class, 300 ture, 165
Index / 893
of a perfect gas, 612 Tschermakite

of electrolytes in solution, 493 formula unit, 300
of fused salts, 412 structural class, 300
of H2O solvent in solution, 501 TTT diagrams: of pyroxenes, 298
of neutral species in solution, 499 Turbulent diffusion: definition, 608
of the hypothetical electron in solu-
tion, 539, 551 U: isotopic composition, 757
affinity, definition, 119 Ugrandites, 250
stability, of aqueous solutions, 546 Ulvospinel: in equilibrium with magnetite,
Ti: oxidation state in silicate melts, 688 hematite, and ilmenite, 404
Ti3⫹: in pyroxenes, 267 Universal chart of gases, 617
Ti4⫹ Unmixing: in the quasi-chemical model, 165
coordination state in glasses, 423 Uphill diffusion, 181
coordination, in melts, 420 Ureyite
Total coulomb energy, 284
electronic energy, definition, 14 dispersive energy, 284
ionic polarizability, 90 enthalpy of formation from the elements,
Trace elements 284
amount in seawater, 604 lattice energy, 284
in feldspars, 347 molar volume, 273
metal organic complexation, 564 repulsive energy, 284
Transition state theory structural class, 268, 269
and kinetic rate laws, 596–98 structural properties, 273
and isotope fractionation, 734 Uvarovite
Transition temperature bulk modulus, 256
and critical temperature, 111 coulomb energy, 257
in first order transitions, 112 crystal field stabilization energy, 257
in second order transitions, 111 dispersive energy, 257
in tricritical transitions, 111 enthalpy of formation from the elements,
Tremolite 257
as a base component in amphibole stoi- heat capacity function, 258
chiometry, 314 lattice energy, 257
enthalpy of formation from the elements, molar volume, 254
312 repulsive energy, 257
formula unit, 312 structural properties, 254
heat capacity function, 312 thermal expansion, 256
mixing with pargasite, 315 third-law entropy, 258
molar volume, 304, 312
polyhedral distortion, 308 V: oxidation state in silicate melts, 688
standard state entropy, 312 Vacancy migration, 205
structural properties, 304 Van Laar parameters, 170, 172
T-effect on site dimensions, 309 Variance, 101–2, 106, 473
thermal decoupling in, 309 Variational principle, 74, 75
thermal expansion, 310 Vectors
Tridymite definitions, 809
bulk modulus, 374 operations, 809
enthalpy of formation from the elements, operators, 810
374 Vibrational
heat capacity function, 374 eigenfunctions, 124
molar volume, 372 energy, of zero point, 48
occurrences, 371 frequencies, in defect equilibria, 200
standard state entropy, 374 function
structural properties, 372 of Einstein oscillators, 140
thermal expansion, 374 of the optical continuum, 140
894 / Index
Vibrational (continued) Winchite

modes formula unit, 300
acoustic, in the Kieffer model, 138 structural class, 300
and isotope fractionation, 732 Wollastonite
highest, 48 bulk modulus, 278
normal, in a crystal, 129 enthalpy of formation from the elements,
optical, in the Kieffer model, 138 283
Volcanic gases: atomic composition, 627 heat capacity function, 283
Volume molar volume, 278
and isothermal compressibility of a standard state entropy, 283
phase, 155 structural class, 269
and thermal expansion of a phase, 155 thermal expansion, 278
change associated to a defect process, Work function of a metal, 721
196, 198 Wuestite
energy, of atomic nuclei, 715 Formula unit, 561
non-solvation contribution, in the HKF Gibbs free energy of formation from the
model, 524 elements, 561
of activation of a reaction, 600
of activation of diffusion, 212 Xe
of fusion, 424 aqueous, thermodynamic properties, 653
of melts, 437 Henry’s constant in silicate melts, 640
of reaction, and thermobarometry, 396 solubility in water, T effect, 643
of silicate melt components, 442 solution enthalpy in silicate melts, 641
parameters of HKF model, 527 solution entropy in silicate melts, 641
solvation contribution, in the HKF
model, 524
standard partial molal of solutes, in the York: algorithm for isochrons, 749, 815
HKF model, 524, 536 Young’s modulus, 184
Water Zero point energy, 124

dielectric constant, 482, 484 definition, 47
ionic dissociation, 484 in anharmonic potential wells, 126
osmotic coefficient, 502 in the MEG approach, 83
P-T stability, 484 Zeroth principle: approximation of, 163
random network model, 481 Zinnwaldite
singular temperature of supercooled, 526 formula unit, 325
structure of, 481 molar volume, 324
vaporization boundary, 487 occurrences, 326
Wave polytypes, 324
functions structural class, 324
and spatial coordinates, 65–66 Zn: thermodynamic properties of organic
density of distribution, 64 complexes, 573
equation of monodimensional propa- Zn: olivine, 224
gation, 4 coulomb energy, 237
hydrogenoid, 64–65 dispersive energy, 237
probability distribution, 64 enthalpy of formation from the elements,
three-dimensional, 5 49, 237
vectors heat capacity function, 238
in the reciprocal space, 135 lattice energy, 49, 237
Wavelength: critical, of compositional fluc- repulsive energy, 237
tuations, 180 standard state entropy, 238
Whole rock isochron: definition, 743 Zone melting, 700

Tips - Principles of Geochemistry PDF

Uploaded by

Copyright:

Available Formats

Tips - Principles of Geochemistry PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tips - Principles of Geochemistry PDF

Uploaded by

Copyright:

Available Formats

Principles of Geochemistry

Columbia University Press / New York

Copyright 䉷 1997 Columbia University Press

Library of Congress Cataloging-in-Publication Data

Casebound editions of Columbia University Press books

3.12 Spinodal Decomposition and Critical Nucleation Radius 178

6.3 Eﬀects of P and T Intensive Variables on Melting 424

8.13 Relative Predominance Limits of Aqaueous Solutes and

10.5 Fractionation Operated by Crystalline Solids: Onuma

1.1 General Information on Atomic Structure

1.1.1 Atom Building

The development of quantum mechanics has led to the formulation of various

1. The Aufbau or “build-up” principle states that electrons occupy orbitals

1.1.2 Wavelike Behavior of Electrons

The propagation of a wave in one dimension is obtained by the partial derivative of

where v is frequency (v ⫽ v / ␭ ) and t is time. Function 1.2 is periodic in t when x is

Table 1.1 Quantum numbers.

The three-dimensional analog of function 1.2 is

Table 1.2 Electron configuration of elements.

Table 1.2 (continued)

Table 1.2 (continued)

From equation 1.6 we obtain

which, when combined with equation 1.1, gives

where ∇ 2 is the Laplacian operator

where n is the vibrational quantum number.

Applying to equation 1.9 the potential of the monodimensional harmonic oscillator

where KF is the force constant.

by integration of equation 1.15 it follows that

which, substituted into equation 1.9, leads to equation 1.13.

A more detailed account of the Schrödinger equation can be found in physical

1.2 Periodic Properties of the Elements

1.2.1 Periodic Properties, Ionization Potentials, and Energy

Substitution of equation 1.19 into equation 1.9 leads to

The eigenvalues of equation 1.21 are

The energy of external orbitals in real atoms (which is numerically equivalent

where Se is a screen constant, depending on the number of inner electrons, and

1.3 Generalities on the Concept of Chemical Bond

1.4 Ionic Bond and Ionic Radius

the energetically optimal condition of an ion surrounded by the maximum num-

were proposed by Landé (1920), Wasastjerna (1923), and Goldschmidt et al.

1.5 Pauling’s Univalent Ionic Radii

where Cn is a constant determined by the principal quantum number of the outer-

rNa + + rF − = 2.31 (1.27)

Table 1.4 Ionic radii in VI-fold coordination, based on univalent ionic

Z Ion rZ (Å) Z Ion rZ (Å) Z Ion rZ (Å)

1.6 Covalent Bond and Covalent Radius

According to Lewis’s (1916) definition, a “covalent bond” exists whenever two

Table 1.5 Nonpolar covalent radii of Sanderson (1960).

Z Ion rZ (Å) Z Ion rZ (Å) Z Ion rZ (Å)

1.7 Electronegativity and Fractional Ionic Character

The concept of “electronegativity” is widely used in geochemistry to define the

where Z is the atomic number and rc is the covalent nonpolar radius.

The most immediate use of the concept of electronegativity is in the definition

According to Gordy (1950), the same parameter can be obtained by applying

then, by application of the equality principle, we have

A similar concept was developed by Sanderson (1960).

Ionization Potential Electronegativity

Ionization Potential Electronegativity

1.8 Ionic Polarizability and van der Waals Bond