VOLUME 1

MODERN
ELECTROCHEMISTRY
VOLUME 1

MODERN
ELECTROCHEMISTRY
An Introduction to an Interdisciplinary Area

John O'M. Bockris

Professor of Electrochemistry
University of Pennsylvania, Philadelphia, Pennsylvania

and
Amulya K. N. Reddy
Professor of Electrochemistry
Indian Institute of Science, Bangalore, India

A Plenum/Rosetta Edition
 Library of Congress Cataloging in Publication Data

6ockris, John O'M.

Modern electrochemistry.

"A Plenum/Rosetta edition."

1. Electrochemistry. I. Reddy, Amulya K. N., joint author. II. Title.
00553.663 1973 541'.37 73-13712

20 19 18 17 16 15 14 13 12 11 10

A Plenum/Rosetta Edition
Published by Plenum Publishing Corporation
227 West 17th Street, New York, N.Y. 10011

@ 1970 Plenum Press, New York

A DivisIOn of Plenum Publishing Corporation

United Kingdom edition published by Plenum Press, london

A Division of Plenum Publishing Company, ltd.
Davis House (4th Floor),8 Scrubs lane, Harlesden, london, NW10 6SE, England

All rights reserved

ISBN-13: 978-1-4615-8602-9 e-1SBN-13: 978-1-4615-8600-5

DOl: 10.1007/ 978-1-4615-8600-5
PREFACE

This book had its nucleus in some lectures given by one of us (J.O'M.B.)
in a course on electrochemistry to students of energy conversion at the
University of Pennsylvania. It was there that he met a number of people
trained in chemistry, physics, biology, metallurgy, and materials science,
all of whom wanted to know something about electrochemistry. The concept
of writing a book about electrochemistry which could be understood by
people with very varied backgrounds was thereby engendered. The lectures
were recorded and written up by Dr. Klaus Muller as a 293-page manuscript.
At a later stage, A.K.N.R. joined the effort; it was decided to make a
fresh start and to write a much more comprehensive text.
Of methods for direct energy conversion, the electrochemical one is
the most advanced and seems the most likely to become of considerable
practical importance. Thus, conversion to electrochemically powered trans-
portation systems appears to be an important step by means of which
the difficulties of air pollution and the effects of an increasing concentration
in the atmosphere of carbon dioxide may be met. Corrosion is recognized
as having an electrochemical basis. The synthesis of nylon now contains
an important electrochemical stage. Some central biological mechanisms
have been shown to take place by means of electrochemical reactions. A
number of American organizations have recently recommended greatly
increased activity in training and research in electrochemistry at universities
in the United States. Three new international journals of fundamental
electrochemical research were established between 1955 and 1965.
In contrast to this, physical chemists in U.S. universities seem-perhaps
partly because of the absence of a modern textbook in English-out of touch
with the revolution in fundamental interfacial electrochemistry which has

v
vi PREFACE

occurred since 1950. The fragments of electrochemistry which are taught

in many U.S. universities belong not to the space age of electrochemically
powered vehicles, but to the age of thermodynamics and the horseless
carriage; they often consist of Nernst's theory of galvanic cells (1891)
together with the theory of Oebye and Hiickel (1923).
Electrochemistry at present needs several kinds of books. For example,
it needs a textbook in which the whole field is discussed at a strong theo-
reticallevel. The most pressing need, however, is for a book which outlines
the field at a level which can be understood by people entering it from
different disciplines who have no previous background in the field but
who wish to use modern electrochemical concepts and ideas as a basis for
their own work. It is this need which the authors have tried to meet.
The book's aims determine its priorities. In order, these are:

1. Lucidity. The authors have found students who understand advanced

courses in quantum mechanics but find difficulty in comprehending a field
at whose center lies the quantum mechanics of electron transitions across
interfaces. The difficulty is associated, perhaps, with the interdisciplinary
character of the material: a background knowledge of physical chemistry
is not enough. Material has therefore sometimes been presented in several
ways and occasionally the same explanations are repeated in different
parts of the book. The language has been made informal and highly ex-
planatory. It retains, sometimes, the lecture style. In this respect, the authors
have been influenced by The Feynmann Lectures on Physics.
2. Honesty. The authors have suffered much themselves from books
in which proofs and presentations are not complete. An attempt has been
made to include most of the necessary material. Appendices have been
often used for the presentation of mathematical derivations which would
obtrude too much in the text.
3. Modernity. There developed during the 1950's a great change in
emphasis in electrochemistry away from a subject which dealt largely with
solutions to one in which the treatment at a molecular level of charge
transfer across interfaces dominates. This is the "new electrochemistry,"
the essentials of which, at an elementary level, the authors have tried to
present.
4. Sharp variation is standard. The objective of the authors has been
to begin each chapter at a very simple level and to increase the level to
one which allows a connecting up to the standard of the specialized mono-
graph. The standard at which subjects are presented has been intentionally
 PREFACE vii

variable, depending particularly on the degree to which knowledge of the

material appears to be widespread.
5. One theory per phenomenon. The authors intend a tcaching book,
which acts as an introduction to graduate studies. They have tried to
present, with due admission of the existing imperfections, a simple version
of that model which seemed to them at the time of writing to reproduce
the facts most consistently. They have for the most part refrained from
presenting the detailed pros and cons of competing models in areas in
which the theory is still quite mobile.

In respect to references and further reading: no detailed references to

the literature have been presented, in view of the elementary character of
the book's contents, and the corresponding fact that it is an introductory
book, largely for beginners. In the "further reading" lists, the policy is to
cite papers which are classics in the development of the subject, together
with papers of particular interest concerning recent developments, and in
particular, reviews of the last few years.
It is hoped that this book will not only be useful to those who wish
to work with modern electrochemical ideas in chemistry, physics, biology,
materials science, etc., but also to those who wish to begin research on
electron transfer at interfaces and associated topics.
The book was written mainly at the Electrochemistry Laboratory in
the University of Pennsylvania, and partly at the Indian Institute of Science
in Bangalore. Students in the Electrochemistry Laboratory at the University
of Pennsylvania were kind enough to give guidance frequently on how they
reacted to the clarity of sections written in various experimental styles and
approaches. For the last four years, the evolving versions of sections of
the book have been used as a partial basis for undergraduate, and some
graduate, lectures in electrochemistry in the Chemistry Department of
the University,
The authors' acknowledgment and thanks must go first to Mr. Ernst
Cohn of the National Aeronautics and Space Administration. Without his
frequent stimulation, including very frank expressions of criticism, the book
might well never have emerged from the Electrochemistry Laboratory.
Thereafter, thanks must go to Professor B. E. Conway, University of
Ottawa, who gave several weeks of this time to making a detailed review
of the material. Plentiful help in editing chapters and effecting revisions
designed by the authors was given by the following: Chapters IV and V,
Dr. H. Wroblowa (Pennsylvania); Chapter VI, Dr. C. Solomons (Penn-
sylvania) and Dr. T. Emi (Hokkaido); Chapter VII, Dr. E. Gileadi (Tel-
viii PREFACE

Aviv); Chapters VIII and IX, Prof. A. Despic (Belgrade), Dr. H. Wroblowa,
and Mr. J. Diggle (Pennsylvania); Chapter X, Mr. J. Diggle; Chapter XI,
Dr. D. Cipris (Pennsylvania). Dr. H. Wroblowa has to be particularly
thanked for essential contributions to the composition of the Appendix on
the measurement of Volta potential differences.
Constructive reactions to the text were given by Messers. G. Razumney,
B. Rubin, and G. Stoner of the Electrochemistry Laboratory. Advice was
often sought and accepted from Dr. B. Chandrasekaran (Pennsylvania),
Dr. S. Srinivasan (New York), and Mr. R. Rangarajan (Bangalore).
Comments on late drafts of chapters were made by a number of the
authors' colleagues, particularly Dr. W. McCoy (Office of Saline Water),
Chapter II; Prof. R. M. Fuoss (Yale), Chapter III; Prof. R. Stokes (Armi-
dale), Chapter IV; Dr. R. Parsons (Bristol), Chapter VII; Prof. A. N.
Frumkin (Moscow), Chapter VIII; Dr. H. Wroblowa, Chapter X; Prof.
R. Staehle (Ohio State), Chapter XI. One of the authors (A.K.N.R.) wishes
to acknowledge his gratitude to the authorities of the Council of Scientific
and Industrial Research, India, and the Indian Institute of Science, Ban-
galore, India, for various facilities, not the least of which were extended
leaves of absence. He wishes also to thank his wife and children for sacrificing
many precious hours which rightfully belonged to them.
CONTENTS

VOLUME 1
CHAPTER 1
Electrochemistry
1.1 Introduction ............................................. . 1
1.2 Electrons at and across Interfaces ....................... . 3
1.2.1 Many Properties of Materials Depend upon Events Occurring at Their
Surfaces ..................................................... . 3
1.2.2 Almost All Interfaces Are Electrified ............................ . 3
1.2.3 The Continuous Flow of Electrons across an Interface: Electrochemical
Reactions .................................................... . 7
1.2.4 Electrochemical and Chemical Reactions ........................ . 8
1.3 Basic Electrochemistry ................................... . 12
1.3.1 Electrochemistry before 1950 ................................... . 12
1.3.2 The Treatment of Interfacial Electron Transfer as a Rate Process:
The 1950's ................................................... . 17
1.3.3 Quantum Electrochemistry: The 1960's .......................... . 19
1.3.4 Ions in Solution, as well as Electron Transfer across Interfaces ..... . 22
1.4 The Relation of Electrochemistry to Other Sciences ..... . 26
1.4.1 Some Diagrammatic Presentations .............................. . 26
1.4.2 Some Examples of the Involvement of Electrochemistry in Other Sciences 28
1.4.3 Electrochemistry as an Interdisciplinary Field, Apart from Chemistry? 29
1.5 Electrodics and Electronics .............................. . 31
1.6 Transients 32

ix
x CONTENTS

1.7 Electrodes are Catalysts .................................. 34

1.8 The Electromagnetic Theory of Light and the Examination of
Electrode Surfaces ........................................ 35
1.9 Science, Technology, Electrochemistry, and Time......... 38
1.9.1 Do Interfacial Charge-Transfer Reactions Have a Wider Significance
Than Has Hitherto Been Realized? .............................. 38
\.9.2 The Relation between Three Major Advances in Science, and the Place
of Electrochemistry in the Developing World ..................... 39

CHAPTER 2
lon-Solvent Interactions
2.1 Introduction ............................................. . 45
2.2 The Nonstructural Treatment of Ion-Solvent Interactions .. 48
2.2. t A Quantitative Measure of Ion-Solvent Interactions ............... 48
2.2.2 The Born Model: A Charged Sphere in a Continuum. ..... ...... .. 49
2.2.3 The Electrostatic Potential at the Surface of a Charged Sphere ...... 52
2.2.4 On the Electrostatics of Charging (or Discharging) Spheres. . . .. . . . . 54
2.2.5 The Born Expression for the Free Energy of Ion-Solvent Interactions 56
2.2.6 The Enthalpy and Entropy of Ion-Solvent Interactions ............ 59
2.2.7 Can One Experimentally Study the Interactions of a Single Ionic Species
with the Solvent? .............................................. 61
2.2.8 The Experimental Evaluation of the Heat of Interaction of a Salt and
Solvent....................................................... 64
2.2.9 How Good Is the Born Theory? ............ . . . . . . . . . . . . . . . . . . . . . 68
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.3 Structural Treatment of the Ion-Solvent Interactions ..... . 72
2.3.1 The Structure of the Most Common Solvent, Water....... ......... 72
2.3.2 The Structure of Water near an Ion .................. . . . . . . . . . . . . 76
2.3.3 The Ion-Dipole Model of Ion-Solvent Interactions ................ 80
2.3.4 Evaluation of the Terms in the lon-Dipole Approach to the Heat of
Solvation ..................................................... 88
2.3.5 How Good Is the Ion-Dipole Theory of Solvation? ................ 93
2.3.6 The Relative Heats of Solvation of Ions on the Hydrogen Scale ..... 95
2.3.7 Do Oppositely Charged Ions of Equal Radii Have Equal Heats of
Solvation? .................................................... 96
2.3.8 The Water Molecule Can Be Viewed as an Electrical Quadrupole ... 98
2.3.9 The Ion-Quadrupole Model of Ion-Solvent Interactions ............ 99
2.3.10 Ion-Induced-Dipole Interactions in the Primary Solvation Sheath.. 102
2.3.11 How Good Is the Ion-Quadrupole Theory of Solvation? ........... 103
2.3.12 The Special Case of Interactions of the Transition-Metal Ions with Water 108
2.3.13 Some Summarizing Remarks on the Energetics of Ion-Solvent Inter-
actions ....................................................... 113
Further Reading........................ ....... ........................ 116
2.4 The Solvation Number ................................... 117
2.4.1 How Many Water Molecules Are Involved in the Solvation of an Ion? 117
 CONTENTS xi

2.4.2 Static and Dynamic Pictures of the Ion-Solvent Molecule Interaction 120
2.4.3 The Meaning of Hydration Numbers ............................ . 123
2.4.4 Why Is the Concept of Solvation Numbers Useful? ............... . 124
2.4.5 On the Determination of Solvation Numbers .................... . 125
Further Reading ...................................................... . 132
2.5 The Dielectric Constant of Water and Ionic Solutions ..... . 132
2.5.1 An Externally Applied Electric Field Is Opposed by Counterfields
Developed within the Medium ................................. . 132
2.5.2 The Relation between the Dielectric Constant and Internal Counterfields 136
2.5.3 The Average Dipole Moment of a Gas-Phase Dipole Subject to Electrical
and Thermal Forces .......................................... . 139
2.5.4 The Debye Equation for the Dielectric Constant of a Gas of Dipoles 142
2.5.5 How the Short-Range Interactions between Dipoles Affect the Average
Effective Moment of the Polar Entity Which Responds to an External
Field ........................................................ . 145
2.5.6 The Local Electric Field in a Condensed Polar Dielectric .......... . 147
2.5.7 The Dielectric Constant of Liquids Containing Associated Dipoles .. . 152
2.5.8 The Influence of Ionic Solvation on the Dielectric Constant of Solutions 155
Further Reading ...................................................... . 158
2.6 Ion-Solvent-Nonelectrolyte Interactions ................. . 158
2.6.1 The Problem .................................................. 158
2.6.2 The Change in Solubility of a Nonelectrolyte Due to Primary Solvation 159
2.6.3 The Change in Solubility Due to Secondary Solvation............ 160
2.6.4 The Net Effect on Solubility of Influences from Primary and Secondary
Solvation ..................................................... 163
2.6.5 The Case of Anomalous Salting in .............................. 164
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Appendix 2.1 Free Energy Change and Work ............... . 168
Appendix 2.2 The Interaction between an Ion and a Dipole .. . 169
Appendix 2.3 The Interaction between an Ion and a Water Quad-
rupole ...................................... . 171

CHAPTER 3
lon-Ion Interactions
3.1 Introduction ............................................. . 175
3.2 True and Potential Electrolytes .......................... . 176
3.2.1 Ionic Crystals Are True Electrolytes ............................. 176
3.2.2 Potential Electrolytes: Nonionic Substances Which React with the
Solvent to Yield Ions .......................................... 176
3.2.3 An Obsolete Classification: Strong and Weak Electrolytes .......... 177
3.2.4 The Nature of the Electrolyte and the Relevance of Ion-Ion Interactions 180
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
3.3 The Debye-Hiickel (or Ion-Cloud) Theory oflon-Ion Inter-
actions ................................................... 180
3.3.1 A Strategy for a Quantitative Understanding of Ion-Ion Interactions 180
xii CONTENTS

3.3.2 A Prelude to the Ionic-Cloud Theory ........................... . 183

3.3.3 How the Charge Density near the Central Ion Is Determined by Electro-
statics: Poisson's Equation ..................................... . 186
3.3.4 How the Excess Charge Density near the Central Ion Is Given by a Clas-
sical Law for the Distribution of Point Charges in a Coulombic Field .. 187
3.3.5 A Vital Step in the Debye-Hlickel Theory of the Charge Distribution
around Ions: Linearization of the Boltzmann Equation ........... . 189
3.3.6 The Linearized Poisson-Boltzmann Equation .................... . 190
3.3.7 The Solution of the Linearized P-B Equation .................... . 191
3.3.8 The Ionic Cloud around a Central Ion .......................... . 193
3.3.9 How Much Does the Ionic Cloud Contribute to the Electrostatic Po-
tential 'Pr at a Distance r from the Central Ion? ................... 199
3.3.10 The Ionic Cloud and the Chemical-Potential Change Arising from Ion-
Ion Interactions ............................................... 201
Further Reading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
3.4 Activity Coefficients and Ion-Ion Interactions ........... . 202
3.4.1 The Evolution of the Concept of Activity Coefficient .............. . 202
3.4.2 The Physical Significance of Activity Coefficients ................. . 204
3.4.3 The Activity Coefficient of a Single Ionic Species Cannot Be Measured 206
3.4.4 The Mean Ionic Activity Coefficient ............................ . 207
3.4.5 The Conversion of Theoretical Activity-Coefficient Expressions into a
Testable Form ............................................... . 209
Further Reading ...................................................... . 212
3.5 The Triumphs and Limitations of the Debye-Htickel Theory
of Activity Coefficients .................................. . 212
3.5.1 How Well Does the Debye-Hiickel Theoretical Expression for Activity
Coefficients Predict Experimental Values? ....................... . 212
3.5.2 Ions Are of Finite Size, Not Point Charges ...................... . 219
3.5.3 The Theoretical Mean Ionic-Activity Coefficient in the Case of Ionic
Clouds with Finite-Sized Ions .................................. . 222
3.5.4 The Ion-Size Parameter a ...................................... . 224
3.5.5 Comparison of the Finite-Ion-Size Model with Experiment ........ . 227
3.5.6 The Debye-Hiickel Theory of Ionic Solutions: An Assessment ..... . 230
3.5.7 On the Parentage of the Theory of Ion-Ion Interactions .......... . 237
Further Reading ...................................................... . 238
3.6 Ion-Solvent Interactions and the Activity Coefficient ..... . 238
3.6.1 The Effect of Water Bound to Ions on the Theory of Deviations from
Ideality ....................................................... 238
3.6.2 Quantitative Theory of the Activity of an Electrolyte as a Function of
the Hydration Number ........................... '" .. .. .. .. .. . 240
3.6.3 The Water-Removal Theory of Activity Coefficients and Its Apparent
Consistency with Experiment at High Electrolytic Concentrations ... 243
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
3.7 The So-Called "Rigorous" Solutions of the Poisson-Boltz-
mann Equation ........................................... 246
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
3.8 Temporary Ion Association in an Electrolytic Solution: For-
mation of Pairs, Triplets, etc. ... . . .. . . . . . . . . . . . . . . . . . . . . . . . 251
 CONTENTS xiii

3.8.1 Positive and Negative Ions Can Stick Together: Ion-Pair Formation 251
3.8.2 The Probability of Finding Oppositely Charged Ions near Each Other 251
3.8.3 The Fraction of Ion Pairs, According to Bjerrum .................. 253
3.8.4 The Ion-Association Constant KA of Bjerrum ..................... 257
3.8.5 Activity Coefficients, Bjerrum's Ion Pairs, and Debye's Free Ions ... 260
3.8.6 The Fuoss Approach to Ion-Pair Formation ...................... 261
3.8.7 From Ion Pairs to Triple Ions to Clusters of Ions. . . . . . . . . . . . . . . . . . 265
Further Reading ................................. : . . . . . . . . . . . . . . . . . . . . . 266
3.9 The Quasi-Lattice Approach to Concentrated Electrolytic So-
lutions .................................................... 267
3.9.1 At What Concentration Does the Ionic-Cloud Model Break Down? 267
3.9.2 The Case for a Cube-Root Law for the Dependence of the Activity
Coefficient on Electrolyte Concentration ......................... . 269
3.9.3 The Beginnings of a Quasi-Lattice Theory for Concentrated Electrolytic
Solutions .................................................... . 271
Further Reading ...................................................... . 272
3.10 The Study of the Constitution of Electrolytic Solutions. . . . 273
3.10.1 The Temporary and Permanent Association of Ions. . . . . . . . . . . . . . . . 273
3.10.2 Electromagnetic Radiation, a Tool for the Study of Electrolytic Solutions 274
3.10.3 Visible and Ultraviolet Absorption Spectroscopy .................. 275
3.10.4 Raman Spectroscopy........................................... 276
3.10.5 Infrared Spectroscopy .......................................... 278
3.10.6 Nuclear Magnetic Resonance Spectroscopy ....................... 278
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
3.11 A Perspective View on the Theory of Ion-Ion Interactions. 279
Appendix 3.1 Poisson's Equation for Spherically Symmetrical
Charge Distribution .............................. 282
Appendix 3.2 Evaluation of the Integral f;:ct r(xr)(I<r) d(I<r) .... 283
Appendix 3.3 Derivation of the Result 1+ = {fVt + 1':.._)I/v 284
Appendix 3.4 To Show That the Minimum in the Pr versus r Curve
Occurs at r = Al2 ................................ 284
Appendix 3.5 Transformation from the Variable r to the Variable
Y = Air........................................... 285
Appendix 3.6 Relation Between Calculated and Observed Activity
Coefficients ....................................... 285

CHAPTER 4
Ion Transport in Solutions
4.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
4.2 Ionic Drift under a Chemical-Potential Gradient: Diffusion 289
4.2.1 The Driving Force for Diffusion ............................... . 291
4.2.2 The "Deduction" of an Empirical Law: Fick's First Law of Steady-
State Diffusion ............................................... . 293
xiv CONTENTS

4.2.3 On the Diffusion Coefficient D ................................. . 296

4.2.4 Ionic Movements: A Case of the Random Walk .................. . 299
4.2.5 The Mean Square Distance Traveled in a Time t by a Random-Walking
Particle ...................................................... . 301
4.2.6 Random-Walking Ions and Diffusion: The Einstein-Smoluchowski
Equation .................................................... . 304
4.2.7 The Gross View of Non-Steady-State Diffusion .................. . 307
4.2.8 An Often Used Device for Solving Electrochemical Diffusion Problems:
The Laplace Transformation ................................... . 309
4.2.9 Laplace Transformation Converts the Partial Differential Equation
Which Is Fick's Second Law into a Total Differential Equation ..... . 312
4.2.10 The Initial and Boundary Conditions for the Diffusion Process Stimu-
lated by a Constant Current (or Flux) ........................... . 313
4.2.11 The Concentration Response to a Constant Flux Switched on at t = 0 317
4.2.12 How the Solution of the Constant-Flux Diffusion Problem Leads On to
the Solution of Other Problems ................................. . 323
4.2.13 Diffusion Resulting from an Instantaneous Current Pulse .......... . 328
4.2.14 What Fraction of Ions Travels the Mean Square Distance <x2 ) in the
Einstein-Smoluchowski Equation? ............................. . 332
4.2.15 How Can the Diffusion Coefficient Be Related to Molecular Quantities? 338
4.2.16 The Mean Jump Distance t, a Structural Question ................ . 339
4.2.17 The Jump Frequency, a Rate-Process Question ................... . 340
4.2.18 The Rate-Process Expression for the Diffusion Coefficient .......... . 342
4.2.19 Diffusion: An Overall View ................................... . 342
Further Reading ...................................................... . 345
4.3 Ionic Drift under an Electric Field: Conduction .......... . 345
4.3.1 The Creation of an Electric Field in an Electrolyte ................ . 345
4.3.2 How Do Ions Respond to the Electric Field? .................... . 349
4.3.3 The Tendency for a Conflict between Electroneutrality and Conduction 351
4.3.4 The Resolution of the Electroneutrality-versus-Conduction Dilemma:
Electron-Transfer Reactions .................................... . 351
4.3.5 The Quantitative Link between Electron Flow in the Electrodes and
Ion Flow in the Electrolyte: Faraday's Law ..................... . 353
4.3.6 The Proportionality Constant Relating the Electric Field and the Current
Density: The Specific Conductivity ............................. . 354
4.3.7 Molar Conductivity and Equivalent Conductivity ................. . 357
4.3.8 The Ec;.uivalent Conductivity Varies with Concentration ........... . 360
4.3.9 How the Equivalent Conductivity Changes with Concentration: Kohl-
rausch's Law ................................................. . 363
4.3.10 The Vectorial Character of Current: Kohlrausch's Law of the Inde-
pendent Migration of Ions ..................................... . 364
Further Reading ...................................................... . 367
4.4 The Simple Atomistic Picture of Ionic Migration ......... . 367
4.4.1 Ionic Movements under the Influence of an Applied Electric Field .. . 367
4.4.2 What Is the Average Value of the Drift Velocity? ................. . 368
4.4.3 The Mobility of Ions ......................................... . 369
4.4.4 The Current Density Associated with the Directed Movement of Ions
in Solution, in Terms of the Ionic Drift Velocities ................. . 371
4.4.5 The Specific and Equivalent Conductivities in Terms of the Ionic Mo-
bilities ....................................................... . 373
4.4.6 The Einstein Relation between the Absolute Mobility and the Diffusion
Coefficient ................................................... . 374
 CONTENTS xv

4.4.7 What Is the Drag (or Viscous) Force Acting on an Ion in Solution? .. 377
4.4.8 The Stokes-Einstein Relation .................................. . 379
4.4.9 The Nernst-Einstein Equation .................................. . 381
4.4.10 Some Limitations of the Nernst-Einstein Relation ................ . 382
4.4.11 A Very Approximate Relation between Equivalent Conductivity and
Viscosity: Walden's Rule .................................... , .. 385
4.4.12 The Rate-Process Approach to Ionic Migration .................. . 387
4.4.13 The Rate-Process Expression for Equivalent Conductivity ......... . 391
4.4.14 The Total Driving Force for Ionic Transport: The Gradient of the
Electrochemical Potential ...................................... . 394
Further Reading ...................................................... . 399
4.5 The Interdependence of Ionic Drifts ...................... . 399
4.5.1 The Drift of One Ionic Species May Influence the Drift of Another . 399
4.5.2 A Consequence of the Unequal Mobilities of Cations and Anions, the
Transport Numbers ........................................... . 400
4.5.3 The Significance of a Transport Number of Zero ................. . 402
4.5.4 The Diffusion Potential, Another Consequence of the Unequal Mobil-
ities of Ions .................................................. . 406
4.5.5 Electroneutrality Coupling between the Drifts of Different Ionic Species 410
4.5.6 How Does One Represent the Interaction between Ionic Fluxes? The
Onsager Phenomenological Equations .......................... . 411
4.5.7 An Expression for the Diffusion Potential ....................... . 413
4.5.8 The Integration of the Differential Equation for Diffusion Potentials:
The Planck-Henderson Equation' ............................... . 417
Further Reading ...................................................... . 420

4.6 The Influence of Ionic Atmospheres on Ionic Migration ... 420

4.6.1 The Concentration Dependence of the Mobility of Ions ............ 420
4.6.2 Ionic Clouds Attempt to Catch Up with Moving Ions. . . . . . . . . . . . . . 422
4.6.3 An Egg-Shaped Ionic Cloud and the "Portable" Field on the Central Ion 423
4.6.4 A Second Braking Effect of the Ionic Cloud on the Central Ion: The
Electrophoretic Effect .......................................... 424
4.6.5 The Net Drift Velocity of an Ion Interacting with Its Atmosphere. . . . 425
4.6.6 The Electrophoretic Component of the Drift Velocity .............. 427
4.6.7 The Procedure for Calculating the Relaxation Component of the Drift
Velocity ...................................................... 427
4.6.8 How Long Does an Ion Atmosphere Take to Decay? .................... 428
4.6.9 The Quantitative Measure of the Asymmetry of the Ionic Cloud Around
a Moving Ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
4.6.10 The Magnitude of the Relaxation Force and the Relaxation Component
of the Drift Velocity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
4.6.11 The Net Drift Velocity and Mobility of an Ion Subject to Ion-Ion
Interactions ................................................... 432
4.6.12 The Debye-Htickel-Onsager Equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
4.6.13 The Theoretical Predictions of the Debye-Htickel-Onsager Equation
versus the Observed Conductance Curves ........................ 435
4.6.14 A Theoretical Basis for Some Modifications of the Debye-Htickel-
Onsager Equation ............................................. 438
Further Reading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439

4.7 Nonaqueous Solutions: A New Frontier in Ionics? ....... . 440

4.7.1 Water Is the Most Plentiful Solvent ............................ . 440
xvi CONTENTS

4.7.2 Water Is Often Not an Ideal Solvent ............................. . 441

4.7.3 The Debye-Hilckel-Onsager Theory for Nonaqueous Solutions .... . 442
4.7.4 The Solvent Effect on the Mobility at Infinite Dilution ............ . 443
4.7.5 The Slope of the A versus d Curve as a Function of the Solvent ... . 445
4.7.6 The Effect of the Solvent on the Concentration of Free Ions: Ion As-
sociation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
4.7.7 The Effect of Ion Association upon Conductivity.. .. ... .. . .. .. . . . . 448
4.7.8 Even Triple Ions Can Be Formed in Nonaqueous Solutions. .. . .. .. . 450
4.7.9 Some Conclusions about the Conductance of Nonaqueous Solutions
of True Electrolytes ........................................... 452
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
Appendix 4.1 The Mean Square Distance Traveled by a Random-
Walking Particle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
Appendix 4.2 The Laplace Transform of a Constant ... . . . . . . . . . . 454
Appendix 4.3 A Few Elementary Ideas on the Theory of Rate
Processes ......................................... 455
Appendix 4.4 The Derivation of Equations (4.257) and (4.258) .. 458
Appendix 4.5 The Derivation of Equation (4.318) ............... 460

CHAPTER 5

Protons in Solution
5.1 The Case of the Nonconforming Ion: The Proton ......... . 461
5.2 Proton Solvation ......................................... . 462
5.2.1 What Is the Condition of the Proton in Solution? ................. 462
5.2.2 Proton Affinity ................................................ 466
5.2.3 The OveraIl Heat of Hydration of a Proton ...................... 467
5.2.4 The Coordination Number of a Proton. . . . .. .. .. . . . ... ... . ..... . 468
Further Reading. ... .. .. ... .. ... .. . .. ..... . .. .. .. ... .. ... .. .... ... ... .. 470
5.3 Proton Transport ........................................ . 470
5.3.1 The Abnormal Mobility of a Proton. .. . .. . .. .. .. .. . .. . ... .. ... . 470
5.3.2 Protons Conduct by a Chain Mechanism ........................ 474
5.3.3 Classical Proton Jumps and Proton Mobility ..................... 476
5.3.4 Do Proton Jumps Obey Classical Laws? ......................... 478
5.3.5 Quantum-Mechanical Proton Jumps and Proton Mobility ......... 480
5.3.6 Water Reorientation, a Prerequisite for Proton Jumps ............. 481
5.3.7 The Rate of Water Reorientation and Proton Mobility. . .. . .. .. ... 482
5.3.8 A Picture of Proton Mobility in Aqueous Solutions ............... 484
5.3.9 The Rate-Determining Water-Rotation Model of Proton Mobility and
the Other Anomalous Facts .................................... 485
5.3.10 Proton Mobility in Ice ......................................... 486
5.3.11 The Existence of the Hydronium Ion from the Point of View of Proton
Mobility ...................................................... 487
5.3.12 Why Is the Mechanism of Proton Mobility So Important? ......... 487
Further Reading ..... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
 CONTENTS xvii

5.4 Homogeneous Proton-Transfer Reactions and Potential Elec-

trolytes ................................................... 488
5.4.1 Acids Produce Hydrogen Ions and Bases Produce Hydroxyl Ions: The
Initial View ................................................... 488
5.4.2 Acids Are Proton Donors, and Bases' Are Proton Acceptors: The
Bronsted View ............................................... . 489
5.4.3 The Dissolution of Potential Electrolytes and Other Types of Proton-
Transfer Reactions ............................................ . 491
5.4.4 An Important Consequence of the Bronsted View: Conjugate Acid-
Base Pairs ................................................... . 493
5.4.5 The Absolute Strength of an Acid or a Base ..................... . 494
5.4.6 The Relative Strengths of Acids and Bases ..................... . 495
5.4.7 Proton Free-Energy Levels .................................... . 500
5.4.8 The Primary Effect of the Solvent upon the Relative Strength of an Acid 504
5.4.9 A Secondary (Electrostatic) Effect of the Solvent on the Relative
Strength of Acids ............................................. . 507
Further Reading ...................................................... . 511

CHAPTER 6
Ionic liquids
6.1 Introduction 513
6.1.1 The Limiting Case of Zero Solvent: Pure Liquid Electrolytes ..... 513
6.1.2 The Thermal Dismantling of an Ionic Lattice. . . .. . .. . .. . .. . . . . . .. 514
6.1.3 Some Features of Ionic Liquids (Pure Liquid Electrolytes) ......... 515
6.1.4 Liquid Electrolytes Are Ionic Liquids. . . . . . . . . . . . . . .. . . . . . . .. .. . . 517
6.1.5 The Fundamental Problems in Pure Liquid Electrolytes. . .. .. . . . . . . .. 518
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
6.2 Models of Simple Ionic Liquids ......................... . 522
6.2.1 The Origin of Liquid Electrolyte Models. . . . . . .. .. . . . ... .. .. . .. . . 522
6.2.2 Lattice-Oriented Models ........................................ 523
6.2.2a The Experimental Basis for Model Building ............... 523
6.2.2b The Need to Pour Empty Space into a Fused Salt ......... 523
6.2.2c The Vacancy Model: A Fused Salt Is an Ionic Lattice with
Numerous Vacancies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
6.2.2d The Hole Model: A Fused Salt Is Full of Holes like Swiss Cheese 527
6.2.3 Gas-Oriented Models for Liquid 'Electrolytes ..................... 529
6.2.3a The Cell-Theory Approach .............................. 529
6.2.3b The Free Volume Belongs to the Liquid and Not to the Particles:
The Liquid Free-Volume Model ......................... 530
6.2.4 A Summary of the Models for Liquid Electrolytes ................ 532
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
6.3 Quantification of the Hole Model for Liquid Electrolytes. . 533
6.3.1 An Expression for the Probability That a Hole Has a Radius between r
and r + dr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
6.3.2 The Furth Approach to the Work of Hole Formation ............. 536
6.3.3 The Distribution Function for the Size of the Holes in a Liquid Elec-
trolyte ........................................................ 537
xviii CONTENTS

6.3.4 What Is the Average Size of a Hole? . ... .. .. ... . .. . .. . ... .... ... .. 539
Further Reading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 541
6.4 Transport Phenomena in Liquid Electrolytes ............. . 541
6.4.1 Some Simplifying Features of Transport in Fused Salts ............. 541
6.4.2 Diffusion in Fused Salts ....................................... 542
6.4.2a Self-Diffusion in Pure Liquid Electrolytes: It May Be Revealed
by Introducing Isotopes ................................. 542
6.4.2b Results of Self-Diffusion Experiments .................... 544
6.4.3 The Viscosity of Molten Salts .................................. 547
6.4.4 What Is the Validity of the Stokes-Einstein Relation in Ionic Liquids? 550
6.4.5 The Conductivity of Pure Liquid Electrolytes ..................... 553
6.4.6 The Nernst-Einstein Relation in Ionic Liquids .................... 555
6.4.6a The Nernst-Einstein Relation: Its Degree of Applicability ... 555
6.4.6b The Gross View of Deviations from the Nernst-Einstein Equa-
tion ................................................... 557
6.4.6c Possible Molecular Mechanisms for Nernst-Einstein Deviations 560
6.4.7 Transport Numbers in Pure Liquid Electrolytes. .. . .. . ... .. . .. .. . . 564
6.4.7a Some Ideas about Transport Numbers in Fused Salts ....... 564
6.4.7b The Measurement of Transport Numbers in Liquid Electrolytes 566
6.4.7c A Radiotracer Method of Calculating Transport Numbers in
Molten Salts ........................................... 571
6.4.7d A Stokes' Law Approach to a Rough Estimate of Transport
Numbers .............................................. 572
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573
6.5 The Atomistic View of Transport Processes in Simple Ionic
Liquids .................................................. . 574
6.5.1 Holes and Transport Processes ................................. 574
6.5.2 What Is the Mean Lifetime of Holes in Fused Salts? ............... 576
6.5.3 Expression for Viscosity in Terms of Holes ....................... 577
6.5.4 The Diffusion Coefficient from the Hole Model ................... 577
6.5.5 A Critical Test of a Model for Ionic Liquids Is a Rationalization of the
Heat of Activation of 3.7 RTm for Transport Processes ............. 580
6.5.6 An Attempt to Rationalize En = E~ = 3.7RTm ................... 581
6.5.7 The Hole Model, the Most Consistent Present Model for Liquid Elec-
trolytes ....................................................... 584
Further Reading.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 587
6.6 Mixture of Simple Ionic Liquids-Complex Formation ... 587
6.6.1 Mixtures of Simple Ionic Liquids May Not Behave IdeaIly ......... 587
6.6.2 Interactions Lead to Nonideal Behavior ......................... 588
6.6.3 Can One MeaningfuIly Refer to Complex Ions in Fused Salts? ..... 589
6.6.4 Raman Spectra, and Other Means of Detecting Complex Ions. . . . . 590
Further Reading ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593
6.7 Mixtures of Liquid Oxide Electrolytes. . . . . . . . . . . . . . . . . . . . . 594
6.7.1 The Liquid Oxides. .. . . . .. .. . . . .. .. . .. .. . .. .. . . ... ... .. ... ... .. 594
6.7.2 Pure Fused Nonmetallic Oxides Form Network Structures Like Liquid
Water ........................................................ 594
6.7.3 Why Does Fused Silica Have a Much Higher Viscosity Than Do Liquid
Water and the Fused Salts? ............................... ,..... 597
6.7.4 The Solvent Properties of Fused Nonmetallic Oxides .............. 601
 CONTENTS xix

6.7.5 Ionic Additions to the Liquid-Silica Network: Glasses.. ... .... ... . 603
6.7.6 The Extent of Structure Breaking of Three-Dimensional Network Lat-
tices and Its Dependence on the Concentration of Metal Ions ....... 604
6.7.7 The Molecular and Network Models of Liquid Silicate Structure. .. . 606
6.7.8 Liquid Silicates Contain Large Discrete Polyanions ............... 610
6.7.9 The "Iceberg" Model .......................................... 615
6.7.10 Fused-Oxide Systems in Metallurgy: Slags ....................... 616
Further Reading ...................................................... 618
Appendix 6.1 The Effective Mass of a Hole..................... 619
Appendix 6.2 Some Properties of the Gamma Function ......... 620
Appendix 6.3 The Kinetic Theory Expression for the Viscosity of
a Fluid........................................... 621
Index ..................................................... , xxxiii
CONTENTS

VOLUME 2

CHAPTER 7

The Electrified Interface

7.1 Electrification of an Interface............................. 623
7.1.1 The Electrode-Electrolyte Interface: The Basis of Electrodics ....... 623
7.1.2 New Forces at the Boundary of an Electrolyte. . . . . . . . . . . . . . . . . . . . 623
7.1.3 The Interphase Region Has New Properties and New Structures. . . . . 626
7.1.4 An Electrode Is Like a Giant Central Ion... . ... ....... ... ... . ... 626
7.1.5 The Consequences of Compromise Arrangements: The Electrolyte Side
of the Boundary Acquires a Charge ............................... 627
7.1.6 Both Sides of the Interface Become Electrified: The So-Called "Electrical
Double Layer" ................................................ 629
7.1.7 Double Layers Are Characteristic of All Phase Boundaries. . . . . . . . . . 630
7.1.8 A Look into an Electrified Interface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639
7.2 Some Problems in Understanding an Electrified Interface. . 640
7.2.1 What Knowledge Is Required before an Electrified Interface Can Be Re-
garded as Understood? ........................................ 640
7.2.2 Predicting the Interphase Properties from the Bulk Properties of the
Phases........................................................ 641
7.2.3 Why Bother about Electrified Interfaces? ........................ 642
7.2.4 The Need to Clarify Some Concepts........ . ... .... ... ... ...... 643
7.2.5 The Potential Difference across Electrified Interfaces .............. 644
7.2.5a What Happens when One Tries to Measure the Absolute Po-
tential Difference across a Single Electrode-Electrolyte Interface 644

xxi
xxii CONTENTS

The Absolute Potential Difference across a Single Electrified

7.2.5b
Interface Cannot Be Measured ........................... 648
7.2.5c Can One Measure Changes in the Metal-Solution Potential
Difference? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
7.2.5d The Extreme Cases of Ideally Nonpolarizable and Polarizable
Interfaces .............................................. 653
7.2.5e The Development of a Scale of Relative Potential Differences 655
7.2.5/ Can One Meaningfully Analyze an Electrode-Electrolyte Po-
tential Difference? ...................................... 659
7.2.5g A Thought Experiment Involving a Charged Electrode in
Vacuum............................................... 660
7.2.5h The Test Charge Must Avoid Image Interactions with the
Charged Electrode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 660
7.2.5i The Outer Potential tp of a Material Phase in Vacuum ...... 663
7.2.5j What is the Relevance of the Outer Potential to Double-Layer
Studies? ............................................... 665
7.2.5k Another Thought Experiment Involving an Uncharged, Dipole-
Covered Phase ......................................... 667
7.2.5/ The Dipole Potential Difference MLlsX across an Electrode-
Electrolyte Interface .................................... 670
7.2.5m The Sum of the Potential Differences Due to Charges and
Dipoles: The Absolute Electrode-Electrolyte (or Galvani)
Potential Difference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 670
7.2.5n The Outer, Surface, and Inner Potential Differences ........ 673
7.2.50 An Apparent Contradiction: The Sum of the Ll<p's across a
System of Interfaces Can and the Ll<p across One Interface
Cannot Be Measured ................................... 674
7.2.5p What Deeper Understanding Has Been Hitherto Gained Re-
garding the Absolute Potential Difference Across an Electrified
Interface? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677
7.2.6 The Accumulation and Depletion of Substances at an Interface. . . . . 679
7.2.6a What Would Represent Complete Structural Information Re-
garding an Electrified Interface? ......................... 679
7.2.6b The Concept of Surface Excess .......................... 680
7.2.6c Does Knowledge of the Surface Excess Contribute to Knowledge
of the Distribution of Species in the Interphase Region? . . . . . 683
7.2.6d Is the Surface Excess Equivalent to the Amount Adsorbed? 684
7.2.6e Is the Surface Excess Measurable? ....................... 685
7.2.6/ The Special Position of Mercury in Double-Layer Studies ... 687
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 687
7.3 The Thermodynamics of Electrified Interfaces............. 688
7.3.1 The Measurement of Interfacial Tension as a Function of the Potential
Difference across the Interface ................................... 688
7.3.2 Some Basic Facts about Electrocapillary Curves .................. 690
7.3.3 A Digression on the Electrochemical Potential .................... 693
7.3.3a Definition of Electrochemical Potential ................... 693
7.3.3b Can the Chemical and Electrical Work Be Determined Sepa-
rately? ................................................ 695
7.3.3c A Criterion of Thermodynamic Equilibrium between Two
Phases: Equality of Electrochemical Potentials ............ 696
7.3.3d Nonpolarizable Interfaces and Thermodynamic Equilibrium 697
7.3.4 Some Thermodynamic Thoughts on Electrified Interfaces .......... 698
 CONTENTS xxiii

7.3.5 Interfacial Tension Varies with Applied Potential: Determination of

the Charge Density on the Electrode ............................ 701
7.3.6 Electrode Charge Varies with Applied Potential: Determination of the
Electrical Capacitance of the Interface ........................... 703
7.3.7 The Potential at Which an Electrode Has a Zero Charge.. . . . . . .. . . 706
7.3.8 Surface Tension Varies with Solution Composition: Determination of
the Surface Excess ............................................ 707
7.3.9 Reflections on Electrocapillary Thermodynamics .................. 714
7.3.10 Retrospect and Prospect in the Study of Electrified Interfaces ..... 715
Further Reading.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 717
7.4 The Structure of Electrified Interfaces .................... . 718
7.4.1 The Parallel-Plate Condenser Model: The Helmholtz-Perrin Theory 718
7.4.2 The Double Layer in Trouble: Neither Perfect Parabolas nor Constant
Capacities .................................................... . 719
7.4.3 The Ionic Cloud: The Gouy-Chapman Diffuse-Charge Model of the
Double Layer ................................................ . 722
7.4.4 Ions under Thermal and Electric Forces near an Electrode ........ . 724
7.4.5 A Picture of the Potential Drop in the Diffuse Layer ............. . 728
7.4.6 An Experimental Test of the Gouy-Chapman Model: Potential Depend-
ence of the Capacitance, but at What Cost? ..................... . 732
7.4.7 Some Ions Stuck to the Electrode, Others Scattered in Thermal Disarray:
The Stern Model .............................................. . 733
7.4.8 A Consequence of the Stern Picture: Two Potential Drops across an
Electrified Interface ........................................... . 734
7.4.9 Another Consequence of the Stern Model: An Electrified Interface Is
Equivalent to Two Capacitors in Series .......................... . 735
7.4.10 The Relative Contributions of the Helmholtz-Perrin and Gouy-Chap-
man Capacities ............................................... . 737
7.4.11 Some Questions Regarding the Sticking of Ions to the Electrode .... . 738
7.4.12 An Electrode Is Largely Covered with Adsorbed Water Molecules .. . 739
7.4.13 Metal-Water Interactions ...................................... . 740
7.4.14 The Orientation of Water Molecules on Charged Electrodes ...... . 741
7.4.15 How Close Can Hydrated Ions Come to a Hydrated Electrode? .... . 741
7.4.16 Is It Only Desolvated Ions which Contact-Adsorb on the Electrode? 742
7.4.17 The Free-Energy Change for Contact Adsorption ............... . 742
7.4.18 What Determines the Degree of Contact Adsorption? ............ . 743
7.4.19 How Is Contact Adsorption Measured? ........................ . 745
7.4.20 Contact Adsorption, Specific Adsorption, or Superequivalent Adsorption 748
7.4.21 Contact Adsorption: Its Influence of the Capacity of the Interface .. . 749
7.4.22 Looking Back to Look Forward ............................... . 752
7.4.23 The Complete Capacity-Potential Curve ....................... . 753
7.4.24 The Constant-Capacity Region ................................ . 753
7.4.24a The So-Called "Double Layer" Is a Double Layer ........ . 753
7.4.24b The Dielectric Constant of the Water between the Metal and the
Outer Helmholtz Plane .................................. 756
7.4.24c The Position of the Outer Helmholtz Plane and an Interpreta-
tion of the Constant Capacity .......................... . 757
7.4.25 The Capacitance Hump ........................................ . 761
7.4.26 How Does the Population of Contact-Adsorbed Ions Change with
Electrode Charge? ............................................ . 762
7.4.27 The Test of the Population Law for Contact-Adsorbed Ions ....... . 769
7.4.28 The Lateral-Repulsion Model for Contact Adsorption ............ . 776
xxiv CONTENTS

7.4.29 Flip-Flop Water on Electrodes ................................. 779

7.4.30 Calculation of the Potential Difference Due to Water Dipoles. . . . . . . 781
7.4.31 The Excess of Flipped Water Dipoles over Flopped Water Dipoles. . . 782
7.4.32 The Contribution of Adsorbed Water Dipoles to the Capacity of the
Interface ...................................................... 788
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 790
7.5 The Competition between Water and Organic Molecules at
the Electrified Interfaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 791
7.5.1 The Relevance of Organic Adsorption. . . . . .. . . . .. . . . .. . .... . .. .. 791
7.5.2 The Forces Involved in Organic Adsorption ...................... 792
7.5.3 Does Organic Adsorption Depend on Electrode Charge? .......... 793
7.5.4 The Examination of the Water Flip-Flop Model for Simple Cases of
Organic Adsorption ........................................... 797
7.5.5 At What Potential Does Maximum Organic Adsorption Occur? . . . . . . 798
Further Reading ...................................................... 801
7.6 Electrified Interfaces at Metals Other than Mercury. . . . . . . . 801
Further Reading ...................................................... 803
7.7 The Structure of the Semiconductor-Electrolyte Interface.. 803
7.7.1 How Is the Charge Distributed inside a Solid Electrode? ........... 803
7.7.2 The Band Theory of Crystalline Solids .......................... 804
7.7.3 Conductors, Insulators, and Semiconductors ..................... 806
7.7.4 Some Analogies between Semiconductors and Electrolytic Solutions 811
7.7.5 The Diffuse-Charge Region inside an Intrinsic Semiconductor: The
Garrett-Brattain Space Charge ................................. 813
7.7.6 The Differential Capacity Due to the Space Charge.. . ... ... .. . . ... 816
7.7.7 Impurity Semiconductors, n Type and p Type.. .. . .. . .. . . . .. .. . .. 818
7.7.8 Surface States: The Semiconductor Analogue of Contact Adsorption... ... 821
7.7.9 Semiconductor Electrochemistry: The Beginnings of the Electrochem-
istry of Nonmetallic Materials .................................. 823
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 823
7.8 A Bird's-Eye View of the Structure of Charged Interfaces. . 824
7.9 Double Layers between Phases Moving Relative to Each Other 826
7.9.1 The Phenomenology of Mobile Electrified Interfaces: Electrokinetic
Properties .................................................... . 826
7.9.2 The Relative Motion of One of the Phases Constituting an Electrified
Interface Produces a Streaming Current ........................ . 829
7.9.3 A Potential Difference Applied Parallel to an Electrified Interface Pro-
duces an Electro-osmotic Motion of One of the Phases Relative to the
Other ........................................................ . 831
7.9.4 Electrophoresis: Moving Solid Particles in a Stationary Electrolyte .. 832
Further Reading ...................................................... . 835
7.10 Colloid Chemistry ....................................... . 835
7.10.1 Colloids: The Thickness of the Double Layer and the Bulk Dimensions
Are of the Same Order ........................................ . 835
7.10.2 The Interaction of Double Layers and the Stability of Colloids ..... . 836
7.10.3 Sols and Gels ................................................ . 839
Further Reading ...................................................... . 841
Appendix 7.1 Measurement of the Electrode-Solution Volta Po-
tential Difference ................................ . 841
 CONTENTS xxv

CHAPTER 8

Electrodics
8.1 Introduction 845
8.1.1 The Situation Thus Far ........................................ 845
8.1.2 Charge Transfer: Its Chemical and Electrical Implications ......... 846
8.1.3 Can an Isolated Electrode-Solution Interface Be Used as a Device? 849
8.1.4 Electrochemical Systems Can Be Used as Devices ................ 851
8.1.5 An Electrochemical Device: The Substance Producer............. 851
8.1.6 Another Electrochemical Device: The Energy Producer. . . . . . . . . . . . 855
8.1.7 The Electrochemical Undevice: The Substance Destroyer and Energy
Waster ....................................................... 859
8.1.8 Some Basic Questions ......................................... 861
8.2 The Basic Electrodic Equation: The Butler-Volmer Equation 862
8.2.1 The Instant of Immersion of a Metal in an Electrolytic Solution ..... 862
8.2.2 The Rate of Charge-Transfer Reactions under Zero Field: The Chemical
Rate Constant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 865
8.2.3 Some Consequences of Electron Transfer at an Interface ........... 868
8.2.4 What Is the Rate of an Electron-Transfer Reaction under the Influence
of an Electric Field? .......................................... 869
8.2.5 The Two-Way Electron Traffic across the Interface. . . . . . . . . . . . . . . . 873
8.2.6 The Interface at Equilibrium: The Equilibrium Exchange-Current
Density io .................................................... 876
8.2.7 The Interface Departs from Equilibrium: The Nonequilibrium Drift-
Current Density i ............................................. 879
8.2.8 The Current-Producing (or Current-Produced) Potential Difference:
The Overpotential 1] . . . . . . . . . • . . . • . • . . • . . . . . . . . . . . . . . . . . . . . . . . . 880
8.2.9 The Basic Electrodic (Butler-Volmer) Equation: Some General and
Special Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 883
8.2.10 The High-Field Approximation: The Exponential i versus 1] Law ... 888
8.2.11 The Low-Field Approximation: The Linear i versus 1] Law........ 892
8.2.12 Nonpolarizable and Polarizable Interfaces ....................... 894
8.2.13 Zero Net Current and the Classical Law of Nernst ................ 897
8.2.14 The Nernst Equation .......................................... 901
8.2. I 5 The Nernst Equation: Its Sphere of Relevance .................... 906
8.2.16 Looking Back................................................. 908
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 909

8.3 The Butler-Volmer Equation: Further Details............. 910

8.3.1 The Need for a Careful Look at Some Quantities in the Butler-Volmer
Equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 910
8.3.2 The Relation between Structure at the Electrified Interface and the Rate
of Charge-transfer Reactions ................................... 911
8.3.3 The Interfacial Concentrations May Depend on Ionic Transport in
the Electrolyte. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 916
8.3.4 What Is the Physical Meaning of the Symmetry factor f3? .......... 917
8.3.4a The Factor f3 Is at the Center of Electrode Kinetics ........ 917
8.3.4b A Preliminary to a Second Theory of f3: Potential-Energy-
Distance Relations of Particles Undergoing Charge Transfer 918
8.3.4c A Simple Picture of the Symmetry Factor ................. 922
xxvi CONTENTS

8.3.4d Is the fJ in the Butler-Volmer Equation Independent of Over-

potential? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 926
8.3.5 Summing-up of Further Details on the Butler-Volmer Equation ... 928
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 929
8.4 The Current-Potential Laws at Other Types of Charged Inter-
faces...................................................... 930
8.4.1 Semiconductor n-p Junctions ................................... 930
8.4.2 The Current across Biological Membranes ....................... 937
8.4.3 The Hot Emission of Electrons from a Metal into Vacuum ......... 942
8.4.4 The Cold Emission of Electrons from a Metal into Vacuum.. .. . .. . 944
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 946
8.5 The Quantum Aspects of Charge-Transfer Reactions at Elec-
trode-Solution Interfaces ................................ . 947
8.5.1 A Few Words on the Mechanics of Electrons .................... . 947
8.5.2 The Penetration of Electrons into Classically Forbidden Regions .. . 950
8.5.3 The Probability of Electron Tunneling through Barriers .......... . 953
8.5.4 The Distribution of Electrons among the Energy Levels in a Metal .. . 956
8.5.5 Under What Conditions Do Electrons Tunnel between the Electrode and
Ions in Solution? .............................................. 959
8.5.6 The Tunneling Condition and the Proton-Transfer Curve. . . . .. .. . . 966
8.5.7 Electron Tunneling and the De-electronation Reaction. . . .. . .. . .. .. 973
8.5.8 A Perspective View of Charge-Transfer Reactions at an Electrode. . . 974
8.5.9 The Symmetry Factor fJ: A Better View ......................... 975
8.5.10 Quantifying the Charge-Transfer Picture ......................... 977
8.5.11 Some Desirable Refinements and Generalizations ................. 981
8.5.12 Surveying the Progress. . . . . .. .. . .. .. .. .. . .. . .. . .. . ... . .. . .. .. .. 983
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 986
8.6 Electrodic Reactions and Chemical Reactions ........... . 986
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 989
Appendix 8.1 The Number of Electrons Having Energy EF Strik-
ing the Surface of a Metal from the Inside ....... 990

CHAPTER 9
Electrodics: More Fundamentals
9.1 Multistep Reactions ...................................... . 991
9.1.1 The Question of Multistep Reactions ........................... . 991
9.1.2 Some Ideas on Queues, or Waiting Lines ........................ . 992
9.1.3 The Overpotential 'Y) Is Related to the Electron Queue at an Interface 993
9.1.4 A Near-Equilibrium Relation between the Current Density and Over-
potential for a Multistep Reaction ............................. . 994
9.1.5 The Concept of a Rate-Determining Step ....................... . 997
9.1.6 Rate-Determining Steps and Energy Barriers for Multisfep Reactions . 1002
9.1.7 How Many Times Must the Rate-Determining Step Take Place for the
Overall Reaction to Occur Once? The Stoichiometric Number v .... . 1004
9.1.8 The Order of an Electrodic Reaction ........................... . 1008
 CONTENTS xxvii

9.1.9 Blockage of the Electrode Surface during Charge Transfer: The Sur-
face-Coverage Factor .......................................... 1014
Further Reading....................................................... 1017
9.2 The Transient Behavior of Interfaces ..................... 1017
9.2.1 The Interface under Equilibrium, Transient, and Steady-State Conditions 1017
9.2.2 How an Interface Is Stimulated to Show Time Variations ......... . 1019
9.2.3 Some Ideas on the Understanding of Transients ................. . 1020
9.2.4 Intermediates in Electrodic Reactions and Their Effects on Potential-
Time Transients .............................................. . 1026
9.2.5 Experimental Methods for the Determination of Partial Coverage, with
Adsorbed Entities, of the Surface of Electrocatalysts ............. . 1029
9.2.5a Radiotracer Method ................................... . 1030
9.2.5b Galvanostatic Transient Method ........................ . 1030
9.2.5c Potentiostatic Transients ................................ . 1032
9.2.5d The Potential-Sweep, or Potentiodynamic, Method ....... . 1033
Further Reading ...................................................... . 1035
9.3 Transport in the Electrolyte Effects Charge Transfer at the
Interface ................................................. . 1036
9.3.1 Ionics Looks after the Material Needs of the Interface ............ . 1036
9.3.2 How the Transport Flux Is Linked to the Charge-Transfer Flux: The
Flux-Equality Condition ....................................... . 1038
9.3.3 Appropriations from the Theory of Heat Transfer ................ . 1040
9.3.4 A Qualitative Study of How Diffusion Affects the Response of an
Interface to a Constant Current ................................ . 1041
9.3.5 A Quantitative Treatment of How Diffusion to an Electrode Affects the
Response with Time of an Interface to a Constant Current ........ . 1044
9.3.6 The Concept of Transition Time ............................... . 1047
9.3.7 Convection Can Maintain Steady Interfacial Concentrations ...... . 1050
9.3.8 The Origin of Concentration Overpotential ..................... . 1052
9.3.9 The Diffusion Layer .......................................... . 1055
9.3.10 The Limiting Current Density and Its Practical Importance ....... . 1059
9.3.lOa Polarography: The Dropping-Mercury Electrode ......... . 1060
9.3.lOb The Rotating-Disc Electrode .......................... . 1070
9.3.11 The Steady-State Current-Potential Relation under Conditions of Trans-
port Control ................................................ . 1072
9.3.12 Transport-Controlled De-electronation Reactions ................ . 1074
9.3.13 What Is the Effect of Electrical Migration on the Limiting Diffusion-
Current Density? 1075
9.3.14 Some Summarizing Remarks on the Transport Aspects of Electrodics 1076
Further Reading ...................................................... . 1079
9.4 Determining the Stepwise Mechanism of an Electrodic Re-
action ................................................... . 1080
9.4.1 How One Tries to Determine the Reaction Mechanism ........... . 1080
9.4.2 Which Is the Rate-Determining Step in the Iron Deposition and Dis-
solution Reaction? ........................................... . 1083
9.4.3 The Transfer Coefficient a and Reaction Mechanisms ............ . 1089
9.4.4 Summarizing Remarks Concerning Mechanistic Studies ........... . 1090
Further Reading ..................................................... . 1093
9.5 More on Mechanism Determination ..................... . 1093
9.5.1 Why Review Mechanism Determination? ........................ 1093
xxviii CONTENTS

9.5.2 What [s Mechanism Determination? ........................... . 1094

9.5.3 Stages in the Elucidation of a Reaction Mechanism ............... . 1095
9.5.4 The Elucidation of the Overall Reaction, the Entities in Solution, and
the Surface Coverage ......................................... . 1096
9.5.5 Some Techniques for Mechanism Determination ................. . 1099
9.5.5a The Determination of Reaction Order ................... . 1099
9.5.5b The Determination of the Transfer Coefficients ........... . 1100
9.5.5c The Determination of the Stoichiometric Number ........ . 1101
9.5.5d Auxiliary Methods ..................................... . 1104
9.5.6 Mechanism Determination for Saturated Hydrocarbons ........... . 1107
Further Reading ..................................................... . 1110
9.6 Current-Potential Laws for Electrochemical Systems ..... . 1111
9.6.1 The Potential Difference across an Electrochemical System ........ . 1111
9.6.2 The Equilibrium Potential Difference across an Electrochemical Cell 1114
9.6.3 The Problem with Tables of Standard Electrode Potentials ........ . 1115
9.6.4 The pH-Potential Diagrams: A General Representation of Equilibrium
Potential Differences across Cells ............................... . ! 121
9.6.5 Are Equilibrium Cell Potential Differences Useful? .............. . 1124
9.6.6 Electrochemical Cells: A Qualitative Discussion of the Variation of
Cell Potential with Current .................................... . 1129
9.6.7 Electrochemical Cells in Action: Some Quantitative Relations between
Cell Current and Cell Potential ................................ . 1132
Further Reading ...................................................... . 1137
9.7 The Grand Divide........................................ 1138

CHAPTER 10
Electrodic Reactions of Special Interest
10.1 Electrocatalysis .......................................... . 1141
10.1.1 A Chemical Catalyst and an Electrocatalyst ..................... . 1141
10.1.2 At What Potential Should Electrocatalysts Be Compared? ........ . 1143
10.1.3 Electrocatalysis in Simple Redox Reactions ..................... . 1146
1O.1.3a How Does the Electrocatalytic Rate Depend upon the Substrate
Work Function at the Reversible Potential? .............. . 1146
1O.l.3b Can the Exchange-Current Density Depend upon the Work
Function? ............................................. . 1149
10.1.4 Electrocatalysis in Reactions Involving Adsorbed Species ......... . 1153
1O.1.4a Electrocatalysis in the Hydrogen-Evolution and - Dissolution
Reaction .............................................. . 1153
1O.1.4b The Electrocatalytic De-electronation of Hydrocarbons .... . 1156
1O.1.4c The Dependence of the Rate upon Substrate for the Oxidation
of Ethylene ........................................... . 1161
1O.1.4d The Special Position of Platinum as an Electrocatalyst ..... . 1166
10.1.5 Special Features of Electrocatalysis ............................ . 1168
1O.1.5a The Effect of the Electric Field ........................ .. 1168
1O.1.5b Reactivity at Low Temperatures ......................... 1169
10.I.5c The Activation of an Electrocatalyst ..................... 1170
1O.1.5d Increasing the Power Output by Changing the Reaction Path 1171
10.1.5e The Use of Porous Electrodes.......................... 1171
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1172
 CONTENTS xxix

10.2 The Electrogrowth of Metals on Electrodes .............. . 1173

10.2.1 The Two Aspects of Electrogrowth ............................. . 1173
10.2.2 The Reaction Pathway for Electrodeposition .................... . 1175
10.2.3 Stepwise Dehydration of an Ion; the Surface Diffusion of Adions .. . 1177
10.2.4 Mechanism Determination on Surfaces Which Change with Time .. . 1182
10.2.5 The Time Variation of the Average Adion Concentration in Response
to the Switching on of a Constant Current ...................... . 1185
10.2.6 The Contributions of Double-Layer Charging and Faradaic Reaction
to the Total Deposition-Current Density ....................... . 1190
10.2.7 The Time Variation of the Overpotential and the Rate-Determining
Step in Electrodeposition ...................................... . 1192
10.2.8 The Contribution of Charge Transfer and Surface Diffusion to the
Total Overpotential for Electrodeposition at Steady State ......... . 1199
10.2.9 From Deposition to Crystallization ............................ . 1202
10.2.10 Some Devices for Building Lattices from Adions: Screw Dislocations
and Spiral Growths ........................................... . 1203
10.2.11 Microsteps and Macrosteps ................................... . 1207
10.2.12 How Steps from a Pair of Screw Dislocations Interact ............ . 1210
10.2.13 Crystal Facets Form .......................................... . 1212
10.2.14 Deposition on Single-Crystal and Polycrystal Substrates .......... . 1218
10.2.15 How the Diffusion of Ions in Solution May Affect E1ectrogrowth .. . 1218
10.2.16 Organic Additives and Electrodeposits ......................... . 1221
10.2.17 The Simultaneous Deposition of More Than One Metal: Alloy De-
position. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1223
10.2.18 The Sometimes Unavoidable Complication: Hydrogen Codeposit ion 1227
Further Reading... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1230
10.3 The Hydrogen-Evolution Reaction........................ 1231
10.3.1 A Reaction with a Special History.............................. 1231
10.3.2 What Are the Possible Paths for the Hydrogen-Evolution Reaction? 1233
10.3.3 What Mechanisms Are Possible in Hydrogen Evolution? .......... 1235
10.3.4 How One Determines the Path and Rate-Determining Step of the
Hydrogen-Evolution Reaction................................... 1237
10.3.4a The Determination of the Exchange-Current Density ...... 1238
1O.3.4b The Determination of the Transfer Coefficient.. . ... .. . .. . . 1238
10.3.4c The Determination of Reaction Order with Respect to Hydrogen
Ions in Solution................... . . . . . . . . . . . . . . . . . . . . . 1243
1O.3.4d The Stoichiometric Number v ........................... 1244
1O.3.4e The Determination of Hydrogen Coverage. ..... . . . . .... .. 1245
1O.3.4[ The Heat of Adsorption of Atomic Hydrogen on the Electrode 1247
10.3.4g Isotopic Separation Factors ............................. 1248
1O.3.4h What Are the Probable Mechanisms for Hydrogen Evolution? 1250
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1250
10.4 The Electronation of Oxygen .............................. 1251
10.4.1 The Importance of the Oxygen-Electronation Reaction ............ 1251
10.4.2 The Evaluation of One of the Mechanisms of Oxygen Electronation . . 1253
10.4.3 Catalysis and the Oxygen Reaction ............................. 1256
10.4.4 Some Special Difficulties with Electrodic Reactions Having Small Ex-
change-Current Densities ................ :..................... 1259
10.4.5 An Electrodic Method of Purifying Solutions ..................... 1261
10.4.6 Observing Very Slow Reactions near Equilibrium ...................... 1263
Further Reading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1263
xxx CONTENTS

CHAPTER 11
Some Electrochemical Systems of Technological Interest
11.1 Technological Aspects of Electrochemistry. . . . . . . . . . . . . . . . . 1265
11.2 Corrosion and the Stability of Metals. . . . . . . . . . . . . . . . . . . . . . 1267
11.2.1 Civilization and Surfaces ...................................... . 1267
11.2.2 Charge-Transfer Reactions Are the Origin of the Instability of a Surface 1268
11.2.3 A Corroding Metal is Analogous to a Short-Circuited Energy-Producing
Cell ......................................................... . 1269
11.2.4 The Mechanism of the Corrosion of Ultrapure Metals ............ . 1273
11.2.5 What Is the Electronation Reaction in Corrosion? ............... . 1275
11.2.6 Thermodynamics and the Stability of Metals .................... . 1277
11.2.7 Potential-pH (or Pourbaix) Diagrams: Uses and Abuses ......... . 1281
11.2.8 The Corrosion Current and the Corrosion Potential .............. . 1285
11.2.9 The Basic Electrodics of Corrosion in the Absence of Oxide Films .. . 1287
11.2.10 An Understanding of Corrosion in Terms of Evans Diagrams ..... . 1291
11.2.11 Which Step in the Corrosion Process Controls the Corrosion Current? 1296
11.2.12 Metals, pH, and Air ......................................... . 1297
11.2.13 Some Common Examples of Corrosion ......................... . 1301
11.2.14 Electrodic Approaches to Increasing the Stability of Metals ....... . 1306
11.2.14a Corrosion Inhibition by the Addition of Substances to the
Electrolytic Environment of a Corroding Metal ......... . 1306
11.2.14b Corrosion Prevention by Charging the Corroding Metal with
Electrons from an External Source ..................... . 1309
11.2.15 Passivation: The Transformation from a Corroding and Unstable Sur-
face to a Passive and Stable Surface ............................ . 1315
11.2.16 The Mechanism of Passivation ................................ . 1318
11.2.17 The Dissolution-Precipitation Model for Film Formation ......... . 1321
11.2.18 Spontaneous Passivation: Nature's Method of Stabilizing Surfaces .. . 1323
11.2.19 A Competition in Models for Passivation? ...................... . 1324
11.2.20 The Thermodynamics of Passivation ............................ . 1326
11.2.21 Hydrogen Diffusion into a Metal .............................. . 1328
11.2.22 The Preferential Diffusion of Absorbed Hydrogen to Regions of Stress
in a Metal .................................................. . 1330
11.2.23 Interstitial Hydrogen Can Crack Open a Metal Surface ............ . 1333
11.2.24 Surface Instability and the Internal Decay of Metals: Stress-Corrosion
Cracking ..................................................... . 1335
11.2.25 Surface Instahility and Internal Decay of Metals: Hydrogen Embrittle-
ment ......................................................... 1338
11.2.26 Charge Transfer and the Stability of Metals . . . . . . . . . . . . . . . . . . . . . . . 1345
11.2.27 The Cost of Corrosion ........................................ 1346
11.2.28 A Bird's-Eye View of Corrosion ................................ 1347
Further Reading ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1349
11.3 Electrochemical Energy Conversion....................... 1350
11.3.1 The Present Situation in Energy Consumption ................... . 1350
11.3.2 How Are the Hydrocarbon Fuels Used at Present? ............... . 1352
11.3.3 The Pollution of the Atmosphere with Products from Internal-Combus-
tion Reactions and Its Possible Effect on World Temperature and Sea
Levels ....................................................... . 1353
11.3.3a Products of Combustion Other than Carbon Dioxide ..... . 1353
11.3.3b Carbon Dioxide ...................................... . 1355
 CONTENTS xxxi

11.3.3c Uncertainties in Predicting the Future Pollution of the At-

mosphere ............................................. 1357
11.3.4 Thermal-Combustion Engines Waste the Chemical Energy Available
from Burning Hydrocarbons in Air ............................. 1357
11.3.5 Direct Energy Conversion ..................................... 1358
11.3.6 Direct Energy Conversion by Electrochemical Means .............. 1361
11.3.7 The Maximum Intrinsic Efficiency in Electrochemical Conversion of the
Energy of a Chemical Reaction to Electric Energy ................. 1361
11.3.8 The Actual Efficiency of an Electrochemical Energy Converter ...... 1366
11.3.9 The Physical Interpretation of the Absence of the Carnot Efficiency
Factor in Electrochemical Energy Conversion .................... 1366
11.3.10 Cold Combustion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1369
11.3.11 Making V near Ve Is the Central Problem of Electrochemical Energy
Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1369
11.3.12 The Electrochemical Quantities Which Must Be Optimized for Good
Energy Conversion............................................. 1374
11.3.13 The Power Output of an Electrochemical Energy Converter. ... . . .. 1376
11.3.14 The Electrochemical Engine .................................... 1378
11.3.15 Was the Wrong Path Taken in the Development of Power Sources at
the End of the Nineteenth Century? ............................. 1379
11.3.16 Electrodes Burning Oxygen from Air ............................ 1382
11.3.17 The Special Configurations of Electrodes in Electrochemical Reactors 1382
11.3.18 Electrochemical Electricity Producers: The Two Basic Types. . .... . 1385
11.3.19 Examples of Electrochemical Generators ......................... 1386
11.3.19a The Hydrogen-Oxygen Cell ............................ 1388
11.3.19b Reformer-Supplied Hydrogen-Air Cells ................. 1389
11.3.19c Hydrocarbon-Air Cells.. . . ... ..... . .. . . .. . . .. . . ... . . ... 1391
11.3.19d Dissolved-Fuel Fuel Cells .............................. 1393
11.3.1ge Natural Gas and CO-Air Cells ........................ , 1393
11.3.20 The Relations between Electrochemical Energy Conversion and the
Future Dominance of Atomic Energy as the Source of Power..... . . 1395
11.3.20a Will Atomic Power Sources Compete for Any of the Uses
Foreseen for Electrochemical Power Sources? ............ 1395
11.3.20b Will Electrochemical Means Be Used to Convert Nuclear
Power to Electricity? .................................. 1396
11.3.20c What Is the Relation between Electricity Storage and Atomic
Energy? .............................................. 1397
11.3.21 A Summary of the Direct Conversion of Chemical Energy to Electricity 1398
Further Reading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1400
11.4 Electricity Storage ....................................... . 1401
11.4.1 Conventional and Descriptive Terminology in Energy Conversion and
Storage ................ , ............................ , ....... , . 1403
11.4.2 The Important Quantities in Electricity Storage .................. . 1404
11.4.2a Electricity Storage Density ............................ . 1404
11.4.2b Energy Density ....................................... . 1407
11.4.2c Power ............................................... . 1410
11.4.2d Desirable Trends ..................................... . 1412
11.4.3 Classical Electricity Storers .................................... . 1413
11.4.3a The Lead-Acid Storage Battery ...................... .. 1413
11.4.3b A Dry Cell .......................................... . 1415
11.4.3c Two Relatively New Electricity Storers ................. . 1416
xxxii CONTENTS

I 1.4.4 The Large Gap between the Maximum Feasible and the Present Actual
Energy Densities of Electricity Storers ........................... 1418
11.4.5 Outlines of Some Possible Future Electricity Storers ............... 1420
11.4.5a Electricity Storage in Hydrogen ......................... 1420
ll.4.5h Storage by Using Alkali Metals.. .. . .. . .. . .. . ... ... . ... . 1422
11.4.5c Storers Involving Nonaqueous Solutions ................. 1424
11.4.5d Storers with Zinc in Combination with an Air Electrode. . . 1427
11.4.6 The Respective Realms of Applicability of Electrochemical Energy
Converters and Electricity Storers ............................... 1428
11.4.7 Electrochemical Electricity Storage in a Nutshell.................. 1430
Further Reading .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1432

Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxix
VOLUME 1

MODERN
ELECTROCHEMISTRY
CHAPTER 1

ELECTROCHEnllSTRY

1.1. INTRODUCTION

It is hoped by the writers of this book that many of its readers will
become electrochemists, i.e., full-time researchers in the field which deals
with chemical transformations produced by electric currents and with the
production of such currents from the transformation of chemical substances.
If that hope is to be realized, then the reader has much to learn for electro-
chemistry is a field which is interdisciplinary. Electrochemistry may be con-
sidered to be born out of Chemistry by Electricity, but it involves far more
than a knowledge of the chemistry of substances and the physics of electric
fields.
The reader and future electrochemist must also realize that, even if
he digests all that is in this book, he will still be walking on a set of shifting
logs, for modern electrochemistry is a rapidly developing branch of know-
ledge, mostly awakened to life in the 1950's and only beginning to call out
lustily in the 1960's.
This book attempts to present a perspective view of the contemporary
scene in the field. Much of the theoretical material presented here is bound
to be improved, and some revised, during the next few decades. This is
especially true because the book intentionally contains a dearth of solidly
sure, but un stimulating, thermodynamic treatments and an excess of con-
ceptual, model-oriented approaches which, though associated with a tem-
porary and uncertain character, undoubtedly stimulate criticisms and new
developments.
2 CHAPTER 1

@a phase
Other
phase

»'"' ' '

contains
ions

~
Fig. 1.1. The fundamental act in electro-
chemistry. (Often, the electron-containing
phase is a metal; and the ion-containing
phase, an aqueous solution. But germanium
in contact with a molten salt would also in-
volve electron- and ion- containing systems.)

What should one choose to give as introductory material in a field

as wide as electrochemistry? Should one startby considering the quantum
mechanics of the exchange of electrons between a metal and ions in the
solution with which the metal is in contact (Fig. 1.1)? Solve the quantum
mechanical equations in the first 10 pages, and obtain the interfacial reaction
rate as a function of the potential difference across the interphase?t This
would not be a good way, we think, to teach electrochemistry at present,
and for two reasons. Firstly, the quantum mechanical formulation of the
reaction rate across the interface-the heart of electrochemistry-has grown
slowly, and a lot of scaffolding is still up. Secondly, because the viewpoint
taken here may be expressed by the statement "Electrochemistry is the
study of phenomena at electrified interfaces," is an unconventional one
for Anglo-American readers (though in accord with a viewpoint long
accepted in Europe), we do need to have a fairly detailed understanding of
the surroundings of the solid-solution interphase before we attempt to
find out about electric charges crossing it. In presenting the preliminary

t It is essential in electrochemistry to distinguish between the terms interface and inter-

phase. An interface formed by two phases is the apparent two-dimensional surface
of contact of these two phases. It is an apparent surface because, in reality, when
two phases come together, there is a region in which there is a continuous transition
from the properties of one phase to the properties of the other. If one aims to refer
specifically to this three-dimensional transition region, then it is more appropriate
to use the term interphase.
 ELECTROCHEMISTRY 3

IONICS ELECTRODICS

Concerns ions in solution and in Concerns the region between an

the liquids arising from melting electronic and an ionic conductor
solids composed of ions. and the transfer of electric charges
across it.

Fig. 1.2. A way to divide the two quite different aspects of the field of electrochem-
istry. In this book, the point of view is presented that the electrodic area should be
the realm associated with electrochemistry. lonics is a necessary adjunct field (just
as is the theory of electrons in metals and semiconductors, which is adequately dealt
with in books on the solid state).

(electrolyte-oriented) material in the first part of our book, we are presenting

material which was formerly understood to constitute electrochemistry
itself. (For, as explained in Section 1.3.1, there was a hiatus in the develop-
ment of interfacial electrochemistry from about 1910 to 1950, and re-
searchers became so interested in the study of the ionic solution surrounding
the interface that they identified electrochemistry with this study and turned
away from the events at the interface.)
So, we have deemed it better to go stepwise and to retain a certain
aura of conventionality in arrangement, first treating the preliminary sub-
ject of the behavior of ions in solutions-the field which we suggest should
be called "ionics"-then considering the main part of electrochemistry-
"electrodics"-, the theory of charged interfaces and the conditions govern-
ing the transfer of charges across them (see Fig. 1.2).
In this chapter, an attempt is made to explain a little of what present-
day fundamental electrochemistry is about.

1.2. ELECTRONS AT AND ACROSS INTERFACES

1.2.1. Many Properties of Materials Depend upon Events Oc-

curring at Their Surfaces

The great interdisciplinary area of materials science treats the properties

of materials. This composite field of study is made up of metallurgy along
with the physics and chemistry of the solid state. It seeks to interpret, for
example, their mechanical properties predominantly in terms of the inter-
actions of the atoms and molecules which constitute the bulk of the material.
This approach has respectable precedents because the beginning of our
understanding of the constitution of materials came with the X-ray work
4 CHAPTER 1

of Laue in 1912 and the quantum theory of specific heats proposed by

Einstein in 1907. Such a bulk-oriented approach to solid materials has been
the origin of most of the progress which has been hitherto made toward
understanding their properties. For example, it is possible to derive theo-
retically the heat contents of solids, or their electrical conductivity. Further,
much of the work on solids has contributed to the growing appreciation
of the theory of the bonds between atoms (the essence of chemistry) and
the energy states of electrons in solids.
Such studies of the bulk properties can give, however, only a very
partial interpretation of the observable behavior of materials. Some phe-
nomena which escape interpretation concern important practical aspects
of materials. For example, metals sometimes undergo unexpected breakage
when strained, and such sudden failures are not always explicable in terms
of the bulk properties of the metal. Such phenomena may be explicable
in terms of certain events associated with the surface, and these events
cannot be predicted from a knowledge of the bulk properties. It may be
found that an atomic-scale defect at the surface, caused by a surface reaction,
is the source of a spreading crack which, assisted by an applied or residual
stress, may work its way spontaneously through several millimeters of
solid in a few hours.
Further, there is a general fact (one of the more important from the
viewpoint of our limited resources) which we have to interpret, namely,
the decay of materials. Metals, with the exception of those that we, there-
fore, call noble, have a limited life when they are in contact with moist air
and tend to revert to the oxides from which many of them were extracted.
Little of this general decay of materials (which limits the practical life of a
ship to some 30 years or perhaps causes the undercarriage of an aircraft
to break even when subjected to normal loads) is due to erosion, i.e., the
wearing away by rubbing, as of rocks by rain. Most of the decay is due to
much more subtle electrochemical surface phenomena, knowledge of the
basic principles of which did not begin to spread until the late 1940's.
Lastly, it is important to recall that very many of the molecules which
govern the mechanics of biological processes are colloidal. This means that
their size is between about 100 and 10,000 A and that their stability as
separate entities depends predominantly upon the forces which exist be-
tween their surfaces and the surrounding ionic solutions, for it is the repulsive
interaction between the electrified interfaces of these particles which prevents
them from coalescing when they approach each other. Correspondingly,
these repulsive interactions govern the behavior at the many interfaces
within biological systems. Proteins and other biological macromolecules
 ELECTROCHEMISTRY 5

MATERIALS SCIENCE SUR FACE SCIENCE

Has dealt hitherto largely with the Many properties of materials-for

bulk properties of solids and with example, their stability or instabil-
the constitution of molecules. This ity in contact with other materials-
approach is the origin of most of cannot be explained adequately in
our present knowledge of the prop- terms of bulk properties. Much of
erties of materials. this needed, future knowledge of
materials is in the realm of surface
science.

Fig. 1.3. Many properties of materials depend primarily upon events occurring on
their surfaces.

such as salts of deoxyribonucleic acid (DNA) are substances of this kind,

in which the surface situation dominates the other influences in determining
stability.
Thus, some important properties of materials and the behavior of
entities which determine the properties of living systems depend on surface
phenomena and surface properties, i.e., surface science (Fig. 1.3).

1.2.2. Almost All Interfaces Are Electrified

Consider the bulk of a solid material, e.g., a metal. It may be pictured,

in an elementary way, as consisting of charged particles, positive ions and
electrons, with the total negative charge balancing the total positive charge
so that there is no net electric charge in the bulk of the material.
Imagine, now, a thought experiment in which a limitingly thin knife
-perhaps a narrow laser beam-cuts through a piece of metal kept in
vacuum. The cutting process bares two surfaces. What happens at each
freshly created metal-vacuum interface? There is a small degree of spillover
of the electrons from the metal into vacuum (cf Section 7.2) resulting in a
disturbance to the former balance of electrical charges in the surface region
of the metal. Having lost some electronic charge, the metal surface becomes
positively charged; and having gained some electronic charge, the vacuum
side of the interface becomes negatively charged. Thus, the metal-vacuum
interface has becomes electrified (Fig. 1.4).
If air t is allowed to come into contact with the metal, further possibil-
ities arise. What one is dealing with is, in fact, the interface between a

t In practice, the term air often means moist air.

6 CHAPTER 1

metal and a moisture film; and, on the moisture side of this interface, there
are water molecules, dissolved oxygen molecules, and perhaps hydrogen
ions (H+). Hence, the interface under consideration may be that between an
electronic conductor and an ionic conductor. The water molecules, which
are electrical dipoles (cf Section 2.3.1), may form an oriented layer on
the metal surface. Such an oriented dipole layer is equivalent to two sheets
of charge (Fig. 1.4). Another possibility is the exchange of electrons between
the metal and particles in the moisture film. For example, the electron
transfer may proceed thus,

and thereafter in the opposite direction, thus,

These electron flows in the two directions-from metal to hydrogen ions

and from hydrogen molecules to metal-only become equal at equilibrium,
by which stage there would have been a net electron flow in one of the
directions. If, for example, this net electron flow is from metal to solution,
then the metal has suffered a certain depletion of its negative charge and,
thus, its surface acquires a net positive charge. Correspondingly, the solution
side of the interface acquires a net negative charge.
It is seen therefore that, irrespective of whether one is considering a
spillover of electrons from a metal surface into vacuum, a net orientation
of dipolar neutral molecules, or the net flow of electrons in electron transfer
processes, there are two results. Firstly, there is the charging of the two
sides of the interface between a metal and another phase (Fig. 1.4), and,
secondly, there is the development of a potential difference across the inter-

m:=
Vacuum
Solution

m
/~~++ _ Ionic
~".""

Due to electron
spill-over
Two sheets of charge
m\(Due to net electron flow
in electrochemical reactions
due to oriented dipoles

Fig. 1.4. Electrified interfaces.

 ELECTROCHEMISTRY 7

phase. The charging proceeds until the charges on the two phases are equal
in magnitude but opposite in sign.
The argument about a net charge of one sign arising on each surface
(of the two phases meeting at an interface) has been couched in terms
of a metal and moist air. But the argument can be generalized. Almost all
interfaces, therefore, are electrified, and the surfaces of material carry
excess charges. It is these excess charges which affect the surface properties
of materials and often make them deviate from those of the bulk.
There are widespread consequences of the fact that the surfaces of
nearly all phases carry a net electric charge and form one side of an electric
double layer. t Among them is the fact that the electrical aspect of the sur-
face forces is an important factor in the determination of properties of the
interphase. Correspondingly, of course, those properties of solids which de-
pend on happenings at the surface depend on the electric charge separation
and electric field strength across the interphase. Very many aspects of the
behavior of surfaces, in practice, involve electrochemical considerations.

1.2.3. The Continuous Flow of Electrons across an Interface:

Electrochemical Reactions

It has been argued in the above section that almost all surfaces carry
an excess electric charge and that interfaces are electrified. However, the
argument was made by considering an isolated piece of material uncon-
nected to a source or sink of electrons.
Suppose now that the metal, an electronic conductor, is connected
to a power supply, t i.e., to a source of electrons so large in capacity that
the 1019 to 1020 electrons* drawn from the source leave it unaffected in any
significant way. To make the discussion specific, consider that the elec-
tronic conductor is a platinum plate and the ionically conducting phase is
an aqueous solution of HI.
Then, by connecting the power source to the metal plate, it becomes
possible for electrons to flow from the source to the surface of the platinum
plate. Before this was done, the electrified platinum-solution interface was

t Some of these consequences concern semiconductor devices, now the central aspect
of solid-state electronics.
t Actually, a power supply has two terminals, and one must also consider how the metal-
electrolyte interface is connected to the other terminal, but this consideration is post-
poned to the next section.
* An Avogadro number (~ 1023 ) of electrons produce 1 gram equivalent of substance
in an electrochemical reaction; hence 1019 to 1020 electrons produce 10- 4 to 10- 3 g-eq.
8 CHAPTER 1

in equilibrium. Under these equilibrium conditions, the platinum plate had

a certain net surface charge, and the ionically conducting solution, an equal
and opposite excess charge; and, further, the electron flows across the inter-
face, associated with the electron-transfer reactions, were occurring to an
equal extent in the two directions. What happens when the disequilibrating
shower of extra charges from the power source arrives at the surface of
the platinum? The details of what happens-the mechanism-is a long
story, told partly in Chapters 8 and 10. However, the essence of it is that
the new electrons overflow, as it were, the electron capacity of the metal
plate and cross the surface to strike and neutralize the ions, e.g., hydrogen
ions, in the solution phase. This process can proceed continuously because
the power source supplying the electrons has been thought of as infinite
in capacity and the ionic conductor also has an abundance of ions in it;
these tend to migrate up to the metal surface and appear there for the
purpose of capturing some of the overflowing electrons.
What is being described here is an electrochemical reaction; i.e., it is
a chemical transformation involving a net electron transfer, and it can be
written in familiar style as
2H+ + 2e~ H 2 •

The hydrogen ions are discharged on the electrode and there is an evolution
of hydrogen gas.
The simplicity of such a formulation should not obscure the fact that
what has been described is a remarkable and distinctive part of chemistry.
An electric current, a controllable electron stream, has been made in a
controlled way to react with a chemical substance and produce another
new chemical substance. t That is what a good deal of electrochemistry is
about-it is about the electrical path for producing chemical transformations.
Much of electrochemistry is also connected with the other side of this
penny, namely, the production of electric currents and therefore electric
power directly from changes in chemical substances, which is one of the
newer methods of producing electrical energy (fuel cells, Section 11.2).

1.2.4. Electrochemical and Chemical Reactions

There is another aspect of the electrochemical reaction which has just
been described. It concerns the happenings to the iodide ions of the hy-

t There is not much limitation on what kind of chemical substance; e.g., it does not
have to be an ion. C.H4 + 4H.O -+ 2CO. + 12H+ + 12e is as much an electrochemical
reaction as is 2H + + 2e -+ H •.
 ELECTROCHEMISTRY 9

drogen iodide, which must also have been present in the solution of HI
in water. Where do they go while the hydrogen ions are being turned into
hydrogen molecules?
The 1- ions have not yet appeared in the picture because only half
of the picture has been shown. Electrical sources have two terminals. The
consideration of a power source pumping electrons into a platinum plate
in contact with an ionic solution is essentially a thought experiment. In the
real situation, one immerses another electronic conductor in the same
solution and connects this second electronic conductor to the other terminal
of the power source (Fig. 1.5). Then, whereas electrons from the power
source pour into the platinum plate, they would flow away from the second
electronic conductor (made, e.g., of rhodium) and back to the power
source. It is clear that, if we want a system which can carryon for some
time with hydrogen ions receiving electrons from the platinum plate, then
iodide ions have got to give up electrons to the rhodium plate at the same
rate as electrons are given up from the platinum. Thus, the whole system
can function smoothly without the loss of electroneutrality which would
arise were the hydrogen ions to receive electrons from the platinum without
a balancing event to occur at the other plate. Such a process would be
required to remove the negatively charged ions which would become excess
ones directly the positively charged hydrogen ions had been removed from
the solution.
The assembly, or system, consisting of one electronic conductor which
acts as an electron source for particles in the ionic conductor and another
electronic conductor acting as an electron sink receiving electrons from
the ionic conductor is known as an electrochemical cell, or electrochemical
system.

2e 2e

- 2e

Pt Rh

Fig. 1.5. The electrochemical reactor.

10 CHAPTER 1

We have seen that, at one charged plate, electrochemical electron-

transfer reactions can occur. What happens if one takes into account the
second plate? There, the electron transfer is from the solution to the other
electronic conductor. Thus, if we consider the two electronic conductor-
ionic conductor interfaces, there is no net electron transfer. The electron
outflow from one electronic conductor equals the inflow to the other; i.e.,
a purely chemical reaction (one not involving net electron transfer) can be
carried out in an electrochemical cell. Such net reactions in an electro-
chemical cell are formally identical to the familiar thermally induced chem-
ical reactions in which molecules collide with each other and form new
species with new bonds. There are, however, very fundamental differences
between the ordinary chemical way of effecting a reaction and the less
familiar electrical or electrochemical way, in which the reactants collide
not with each other but with separated "charge-transfer catalysts," as the
two plates which serve as electron-exchange areas might be called. One of
the differences, of course, pertains to the facility with which the rate of
a reaction in an electrochemical cell can be controlled; all one has to do
is to control electronically the power source. This facile control arises
because the electrochemical reaction rate is the rate at which the power
source pushes out and receives back electrons through the electrochemical
cell.
Thus, one could write the electrochemical happenings as

2HI in
solution
, 2H+ + 21-
2H+ + 2e at Pt
plate
, H2 t

21-
at Rh
plate
, 12 + 2e
2HI ' H2 + 12
Thus, from an overall point of view, this net cell reaction is identical
to that which would occur if one heated hydrogen iodide and produced
hydrogen and iodine by a purely chemical, or thermal, reaction.
There is another way in which the electrical method of carrying out
chemical reactions is distinct from the other methods for achieving chemical
changes (Fig. 1.6). "Ordinary" chemical reactions, like the homogeneous
combination of H2 and 12 or the heterogeneous combinations of H2 and
O 2 , occur because thermally energized molecules occasionally collide and,
during the small time they stay together, change around some bonds to
 ELECTROCHEMISTRY 11

H I
f:i. ·1
I + I
Thermal
H I
Bond •
collisillnal
• H··r breaking
activation

Chemical (or Thermal) Reaction

HI
(gas)

Electrochemical (or Electric) Reaction

Fig. 1.6. The chemical and electrochemical

ways of carrying out reactions. In the electro-
chemical way, the particles do not collide with
each other but with separated sources and
sinks of electrons.

form a new arrangement. Correspondingly, photochemical reactions occur

by mechanisms in which photons strike molecules and give them extra
energy so that they break up and form new compounds. In a similar way,
the high-energy particles emanating from radioactive substances can ener-
gize molecules, which then react. The electrical method of causing chemical
transformations is different from the other two general methods of provok-
ing chemical reactivity in that the overall electrochemical cell reaction is
composed of two separate electron-transfer reactions which occur at spa-
tially separated electrode-electrolyte interfaces and are susceptible to elec-
trical control.
If electrical energy provokes and controls chemical reactions, chemical
reactions can presumably give rise to a flow of electricity. Thus, two reac-
tant substances may be allowed to undergo spotaneous electron transfer at
the separated (electrode) sites characteristic of the electrochemical way of
bringing about chemical reactions, and then the electrons transferred in the
two reactions at the two interfaces can be made to flow through an electrical
load, for example, the circuit of an electric motor (Fig. 1.7). In this reverse
12 CHAPTER 1

4e 4e

Fig. 1.7. Grove was the first to obtain elec-

tric power directly from a chemical reaction.

process, too, there is a unique aspect when one compares it with the produc-
tion of available energy from thermally induced chemical reactions. It can
be shown (cf Section I 1.3) that the fraction of the total energy of the chem-
ical reaction which can be converted to mechanical energy turns out to be
intrinsically much greater in the electrical than in the chemical way of
producing electricity.

1.3. BASIC ELECTROCHEMISTRY

1.3.1. Electrochemistry before 1950

Certain fields develop logically, usually. from an initial empirical dis-
covery. Such a one is the production of energy from the heat evolved in
nuclear reactions. The harnessing of nuclear energy through controlled
nuclear reactions in a reactor (Fermi et al., 1942) was the culmination of
the work of Becquerel (1896) on radioactivity, of Einstein (1905) on mass-
energy equivalence, of Rutherford (1911) on the nuclear atom model, of
Cockroft and Walton (1930) on the artificial transmutation of elements,
and of Hahn and Meitner (1939) on fission.
Electrochemistry made a much earlier and, indeed, even more promis-
ing, start than did nuclear chemistry. Luigi Galvani described the first
observations in a field which, nearly 180 years later, has come again to
focus upon its electrical aspects. Galvani wrote (1791):
I had dissected a frog ... and had placed it upon a table on which there
was an electric machine, while I set about doing certain other things. The
frog was entirely separated from the conductor of the machine and indeed
was at no small distance away from it. While one of those who were as-
 ELECTROCHEMISTRY 13

sisting me touched lightly and by chance the point of his scalpel to the
internal crural nerves of the frog, suddenly all the muscles of its limbs
were seen to be so contracted that they seemed to have fallen into tonic
convulsions. Another of my assistants, who was making ready to take up
certain experiments in electricity with me, seemed to notice that this hap-
pened only at the moment when a spark came from the conductor of the
machine. He was struck with the novelty of the phenomenon, and imme-
diately spoke to me about it, for I was at the moment occupied with other
things and mentally preoccupied. I was at once tempted to repeat the
experiment so as to make clear whatever might be obscure in it. For this
purpose I took up the scalpel and moved its point close to one or the
other of the crural nerves of the frog while at the same time one of my
assistants elicited sparks from the electric machine. The phenomenon
happened exactly as before. Strong contractions took place in every muscle
of the limb, and at the very moment when the sparks appeared the animal
was seized as it were with tetanus.
This observation of Galvani's was closely followed by an observation
by Volta, who wrote thus in 1800:
I have the pleasure of communicating to you, and through you to the
Royal Society, some striking results I have obtained in pursuing my ex-
periments on electricity excited by the mere mutual contact of different
kinds of metal, and even by that of other conductors, also different from
each other, either liquid or containing some liquid, to which they are
properly indebted for their conducting power. The principle of these results
is the construction of an apparatus having a resemblance in its effects to
the Leyden flask. The apparatus to which I allude, and which will, no
doubt, astonish you, is only the assemblage of a number of good con-
ductors of different kinds arranged in a certain manner. Thirty or more
pieces of copper, or, better, silver, applied each to a piece of tin or zinc,
which is much better, and as many strata of salt water or any other conduct-
ing liquid, or pasteboard, skin, etc., well soaked in these liquids; such
strata interposed between every pair of two different metals and always in
the same order are all that is necessary for constructing my new instrument.
To this apparatus, much more similar to the natural electric organ of
the torpedo or the electric eel, etc., than to the Leyden flask, I would
wish to give the name of the "Artificial Electric Organ." ...
All the facts which I have related in this long paper in regard to the
action which the electric fluid excited and, when moved by my apparatus,
exercises on the different parts of our body which the current attacks and
passes through; an action which is not instantaneous, but which lasts, and is
maintained during the whole time that this current can follow the chain
not interrupted in its communications; in a word, an action the effects of
which vary according to the different degrees of excitability in the parts,
14 CHAPTER 1

as has been seen; all these facts, sufficiently numerous, and others which
may still be discovered by multiplying and varying the experiments of
this kind, will open a very wide field for reflection, and of views, not only
curious, but particularly interesting to Medicine. There will be a great deal
to occupy the anatomist, the physiologist, and the practitioner.
These discoveries by Galvani and Volta near the beginning of the
nineteenth century not only emphasized the role of electrochemistry in the
production of electricity but also demonstrated the importance of electro-
chemical phenomena in biology.
This brilliant and very early start to electrochemistry was carried on
by the discovery by Faraday in 1834 of the laws which govern the quantity
of a chemical substance formed by the passage of a certain amount of
electricity across an interface such as that at the platinum-hydrogen ion-
containing solution of Fig. 1.6. The laws of Faraday are among the most
exactly verified laws of science and led to the realization by Stoney (1891)
that electricity, like matter, is atomic in nature. Stoney proposed the name
electron for the "atom," or unit, of electric charge.
A fourth major advance in electrochemistry at this very early time
consisted of the first attainment by Grove of electrochemical energy
conversion. In 1834, Grove produced electrical energy from the union of
hydrogen and oxygen in an electrochemical cell (Fig. 1.7). The far-reaching
significance of the achievement of Grove's was realized by Ostwald as
early as 1894. At the Bunsen Gesellschaft meeting of that year, Ostwald
pointed out that the direct conversion of chemical energy to electrical
power could be achieved electrochemically with a possible conversion ef-
ficiency of nearly 90%. In contrast, the (indirect) thermal method (a chem-
ical reaction produces heat which expands a gas which pushes a cylinder
which drives a wheel which turns a generator), however, was subject to the
Carnot restriction on the maximum possible efficiency with which the heat
energy liberated in a chemical reaction can be converted into mechanical
work. The restriction rests on intrinsic thermodynamic reasoning and argues
for a practical upper limit of about 40% for the efficiency of conversion of
chemical energy into electricity by heat (Fig. 1.8).

THERMAL REACTION OF ATOMS ELECTRIC REACTION OF ATOMS

---+ HEAT ---+ MECHANICAL ENERGY ---+ ELECTRICITY
---+ ELECTRICITY

Fig. 1.8. Ostwald pointed out that electrically achieved chemical reactions could
convert a higher fraction of the energy difference between their initial and final states to
electricity than could thermally achieved reactions.
 ELECTROCHEMISTRY 15

The progress of electrochemistry in these early times thus seemed

excellent. For example, it appeared to have been relatively greater than
the corresponding progress in the field of homogeneous reaction kinetics
and had obtained a head start over, e.g., thermionics. Furthermore, funda-
mental advances were made in the early years of the present century by
Le Blanc, who made an early application of ideas of nonstationary potentials
across interfaces, and by Tafel, who, in 1905, established what could
perhaps be called the most used law in electrochemistry, namely, that the
potential difference L1<p across the interface at which an electrochemical
reaction is occurring changes linearly with the logarithm of the current
density i, or
L1<p = a - b In i

where a and b are independent of L1<p, or

This empirical law of Tafel presents an early example of a series of ex-

perimental laws, involving the rate at which events occur, such as that
found by Arrhenius for the relation of reaction velocity and flow velocity
to temperature and that by Richardson for the dependence of electron
emission on the thermionic work function of the electrode.
The discoveries by Galvani, Volta, Faraday, LeBlanc, and Tafel in-
dicated a promising future for electrochemistry with a rapid development
in the understanding of the rates of electrochemical cell reactions as a
function of the potential differences across cells. But this promise was
not quickly fulfilled. A somewhat stultifying influence settled upon electro-
chemistry. This influence arose from the great success achieved at about
the turn of the century by the thermodynamic treatment of electrochemical
cells in equilibrium, i.e., the treatment of the situation when no net current
passes across the interfaces in a cell and, therefore, when no net cell reaction
occurs. This thermodynamic view considered the electrical energy lost or
gained when an electric charge was, in a thought experiment, taken around
a circuit consisting of an electrochemical cell with its two electrode-solution
interfaces. The electrical energy change is the algebraic sum of the potential
differences in the cell, multiplied by the charge transferred in the reactions
at each interface. Such an electrical energy sum could then be placed equal
to the change in free energy which occurs in the net chemical reaction
which takes place in an electrochemical cell.
This brilliant thermodynamic analysis made by Nernst (1891) opened
up excellent ways of getting at the free energy changes in chemical reactions
16 CHAPTER 1

and solving the problem of distinguishing between the change in the heat
content of reactants and products and the change in the free energy be-
tween them. Great as these advances were in the history of chemistry, they
had also weighty, if indirect, negative consequences upon the development
of electrochemistry. Two of these were:
I. They made the viewpoint in the treatment of electrochemical reac-
tions across interfaces "potential-centric" and not "current (or rate t)_
centric." Thus, when the current was changed, it was thought that the
corresponding (unrationalized) changes in electrode potential were due to
some illicit t malfunction somewhere in the cell. The change was sometimes,
particularly by the polarographers, ascribed to a delay in transporting
reactants through the ionically conducting solution up to the interfaces,
so that the electrode surface sees a different concentration of ions from
that which it would have seen at zero current. The thought that there was
a net flow of electrons across the interfaces only because the potential
difference across them had been made to differ from the equilibrium one
of Ncrnst was suppressed in the zeal of the potential-centric and rate-
indifferent thermodynamic approach.
2. Apart from making them regard the potential as the variable de-
pendent on the current, the overwhelming success of the thermodynamic
treatment-and its emphasis in all textbooks of physical chemistry for the
next 50 years- seems to have awed electrochemists from about 1910
onward. They did not, with a very few exceptions, try to develop a rational
kinetic treatment of the interfacial charge transfer. As late as 1947, most of
them were still trying to do the impossible, i.e., to treat the highly (thermo-
dynamically) irreversible electrode reactions by a series of misconceptions
and approximations on the basis of reversible thermodynamics. This
fundamental error and lack of conceptualization held a dead hand on the
electric mode of achieving chemical reactions and on the direct conversion of
chemical to electrical energy, for some 40 to 50 years. This period in elec-
trochemistry could well be called "The Great Nernstian Hiatus" (Fig. 1.9).

t The current density across an interface is the rate of electron passage across it per
unit area. Thus, it is usually written as amp cm- 2 •
t The attitude of post-Nernstian electrochemists toward changes in potential away from
that described by Nernst's equations seems sometimes to have been tinged with a
derogatory moral fervor. The word polarized, used to denote in a vague way the non-
equilibrium condition of electrodes, had a downing and critical kind of flavor even
in the late 1940's. It will be seen later, however, that polarization is an essential factor
associated with electrochemical reactions out of equilibrium and, without polarization,
the reactions cannot occur at a net rate.
 ELECTROCHEMISTRY 17

Nernsfs law:

Tafel's law in its potentiocentric form: 6.¢ = P + Q In i

Tafel's law in its current-centric form:

Fig. 1.9. The great Nernstian hiatus. Here t:,.¢ is the potential across the interface;
8i is the activity of a species in the solution; and i is the current density (t:,.¢o, P, Q,
A. and B are constants). Affected by the successful thermodynamic thinking of
Nernsfs theory of galvanic cells at equilibrium, electrochemists regarded the current
passing in cells out of equilibrium as affecting the potential differences at the inter-
faces; they did not for a long time understand that the current flowed across them
only because the potential changed away from that of equilibrium, and possessed
an overpotential, 1/.

One probable consequence of great socio-economic importance can be

mentioned. It will be seen in Chapter II that the electric mode of bringing
about reactions can be reversed so that electricity can be produced directly
from chemical reactions in an electrochemical system or cell. The efficiency
of the conversion of chemical energy into electricity and the power which an
electrochemical converter produces depend on the nonthermodynamic phe-
nomenon of gross departures of the potential difference across electro-
chemical cells from the equilibrium cell potential (the degree of this depar-
ture is called the overpotentia.l). It was of course impossible to rationalize the
overpotential in the early fuel cells, or electrochemical converters, so that
a practical fuel cell was not produced until the 1960's, and not even then,
one capable of producing more than a few kilowatts. It is of interest to
speculate that, had the Nernstian hiatus not inhibited the development
of the underlying subject of electrode kinetics, then, perhaps, practical elec-
trochemical energy converters would have been produced several decades
earlier. A consequence of such a development, had it become possible to
build fuel cells with megawatt output, would then have been the absence
of the atmospheric pollution which has arisen from the production of
power by the chemical combustion reaction of hydrocarbons with oxygen
to carbon dioxide, together with other more complex products, some of
which are the base substances of smog.

1.3.2. The Treatment of Interfacial Electron Transfer as a Rate

Process: The 1950' s

By about 1950 onward, most electrochemists in Europe and America

began generally turning toward the view that the immediate cause of the
18 CHAPTER 1

current flowing across the interface in an electrochemical cell was the fact
that the outside power source caused the electric potential difference across
the electric double layer at the interfaces to change from the values that would
correspond, for given electrochemical reactions, to the zero-current situation
of equilibrium, i.e., to assume values such that there was an overpotential.
Such a view had been implied in a rudimentary way earlier by Butler
(1924) and stated explicitly by Volmer and Erdey-Gruz in 1930. Only in
Russia, t under the leadership of Frumkin, had the new attitude been consist-
ently adopted during the 1930's and 1940's. Most other electrochemists
during the 1930's and 1940's thought of the potential at the interface as a
quantity given by thermodynamics only. It took until the 1950's to get
electrochemists generally moving along in the direction of relating current
densities to deviations of the electrode potentials from the value which
they possessed when the interface was at equilibrium, i.e.,
i = Ae B "

where the overpotential (the departure of the potential from that charac-
teristic of equilibrium) 1] was the current-provoking quantity, and not as
Lfrp = a - b In i
where the attitude was that the potential was basically that given by equilib-
rium thermodynamics but disturbed by the passage of electrons across the
interface.
The change of attitude among electrochemists in the 1950's was more
than a formal one because it compelled them to think about the inter-
phasic region in a more realistic way. Electrochemistry, after a delay of
35 to 40 years, became molecular kinetic and structural in attitude. Its
researchers began to talk in terms of the molecular structure at the interface
and the effect of the interfacial field on the transfer of electrons between
the electrode and the particles in the adjacent layer of solution. The under-
standing of the kinetic interpretation of equilibrium became widespread,
i.e., equilibrium corresponds to electrons crossing the interface at the same
rates in both directions; the methods were developed by which the various
steps in electrochemical reactions could be elucidated. Electronics became
an indispensable tool to the study and control of electrochemical reactions.
There was sufficient work at the end of the decade for the first t general

t Isolated examples of this attitude elsewhere certainly existed, e.g., in the work of
Audubert in France and of Bowden in England.
t A brief text in Russian, due to Frumkin's school, dates from 1952, but is almost
unknown in the West.
 ELECTROCHEMISTRY 19

I
I
I
I
I
Electronic I Interphase Ionic
conductor : region conductor
I
I
I
ELECTRODICS IONICS

Fig. 1.10. lonics and electradics.

textbook on the kinetics of electrochemical reactions to be written (Vetter,

1961) and to mark the end of the first stage of the rebirth and growth of
electrochemistry (Figs. 1.10 and 1.11).

1.3.3. Quantum Electrochemistry: The 1960's

Electrochemistry by the mid-1950's showed much vigor arising from

the new attitudes; workers in electrochemistry were returning from other
occupations (see below) to the attractive task of increasing knowledge

Interface
8
8 88
8
8
G G e G
e
//~= e0
o
0
0 0 e

::t.~~IIY
/conducting /
e
+
00 0
e
0
0 8
e 0
8
0 Ionically
conducting
e 0 phose

%
o e e0
e o 00
o 0 e
o e 0
~-
Fig. 1.11. Electrochemistry.
20 CHAPTER 1

concerning interfacial charge transfer. Electrochemistry was no longer a

frozen field.
But the electrochemists of the mid-1950's had not been sufficiently
discerning in the type of electron transfer which they considered. In therm-
ionic electron emission at high temperatures, the electrons have to climb
up and over an energy cliff, and significant rates of transfer are obtained
by raising the temperature, often to 1000 to 2000oK, so that, in Richardson's
expression
i = APe-<f>RIT

(where rp is the thermionic work function and A is a constant for a given

substance), the exponential term is made sufficiently low.
In the situations treated by electrochemistry, the transfer of electrons
across interfaces, the most common one is that of a metal in contact with
an aqueous ion-containing solution. Here, the value of the work function
rp gets changed from that value which it has in a vacuum and which applies
essentially to the thermionic emission case. But, when, in the 1960's, elec-
trochemists tackled the more detailed task of considering what kind of
electron transfer would correspond to the range of rates of charge transfer
observed at the metal-solution interfaces, it turned out that classical electron
emissions (electrons climbing over barriers to get in and out of the electro-
nically conducting phase) would not do. Such electron transfers provided
rates many orders of magnitude smaller than those experimentally observed.
How does one know whether classical mechanics will apply effectively
to a problem? This is a fairly easy question to answer; it is covered by the
correspondence principle, according to which the predictions of the quantal
laws correspond more and more to those of the classical laws under certain
limiting conditions. For example, in the emission spectra of atoms, classical
and quantum theory tend to agree if the quantum number considered is
sufficiently big. Correspondingly, in situations in which atomic vibrations
are concerned, the classical theory becomes applicable if the frequency of
vibration is sufficiently small. The most basic example of this rule is that
if the happenings described concern motions much larger than their de
Broglie wavelengths, the treatment may be well approximated by classical
mechanics.
The energy of a particle which tunnels must be the same on both sides
of the barrier (for otherwise, radiation will occur). Hence,

P P22
2
_1_
2m
+ VI = --
2m-
+ V.
 ELECTROCHEMISTRY 21

where PI and P2 are the momenta of the electrons inside the metal and the
atom which receives the electron, respectively; U1 is the potential energy
of the electron inside the metal and U2 that of the electron in an atom in
the solution. The proper calculation of U1 and U2 is an exercise in the
theory of electron transfer at interfaces (see Chapter VIII). Suppose U2 = 0
and U1 = C/>, the thermionic work function. Then, with the de Broglie
wavelength

But PI is the momentum of electrons in the Fermi level and p,2/2m e

;;;:; kinetic energy ~ the Fermi energy of an electron in the metal. But the
Fermi energy of an electron differs from its work function only by the
surface potential, which is much less than the work function. Hence,
2
~
2m -
C/> or
J'-. P1 2 '" 2mC/>

Hence,

Taking C/> as some 5 X 10- 12 erg atom-I,

A J'-. 10- 3 cm

Thus, the relevant de Broglie wavelength is several thousand angstroms.

However, at a given interface, the distance over which an electron jumps
in an electrode reaction is always of the order of a few angstroms. Hence,
the transition with which one is associated in an interfacial electron transfer
is much less than the length of the relevant de Broglie wave. The transitions
must be treated quantally.
It is of interest to note that the first paper to treat the electrode-solution
interface quantally was published by Gurney in 1931, but the matter was
then allowed to rest there for some 20 years instead of having been exploit-
ed as one of the first pieces of quantum chemistry to be published. Basic
electrochemistry (and all that it implies in respect to the properties of
materials, the direct conversion of chemical energy to mechanical work,
and the functioning of biological systems) must be worked out in quantum,
rather than classical, mechanics.
22 CHAPTER 1

1.3.4. Ions in Solution, as well as Electron Transfer across

Interfaces
From what has been said so far, it might be thought that the field
encompassed by electrochemistry concerns the functions of interfacial
charge transfer only. This is certainly not the impression one would get if
one looked at other books on electrochemistry. In fact, in so far as most
of these were published in the 1940's or earlier, one gets almost the reverse
impression; electron transfer across interfaces was dealt with sparingly
indeed in those days.
There is a good historical reason for this reversing disbalance. When
the development of interfacial electrochemistry along modern lines became
restricted by the overthermodynamic attitude of its adherents in the pre-
1950 days, much attention was diverted to what had seemed previously to
some extent the accompanying side issues, i.e., the physical chemistry of
the bulk solution adjoining the double layer. This had concentrated upon
an interest in deviations in the behavior of solutions from laws derived on
the assumption that interactions between particles in them were negligible.
Thus, the famous Debye-Htickel theory of such interactions-sophisticat-
ed and successful-attracted the attention of electrochemists away from
the blocked interfacial studies. Thus, from about 1920 to 1950, the majority
of research workers in the electrochemical field were occupied with deter-
mining, e.g., activity coefficients of salts in dilute aqueous solutions, the
electrical conductance of molten salts, or electrostatic effects on the dis-
sociation constants of acids or bases in aqueous solutions, etc. (Figs. 1.12
and 1.13).
The situation from the 1950's is that, with the opening up of the rate
process and quantum mechanical approaches to interfacial reactions, most
attention in electrochemistry concerns mechanisms at the interface. But we
must not forget to research the fields concerned with ions in solution, too.
It will be seen in Chapter 3, for example, that a great deal remains to be
done before we can say that we have a sophisticated picture of ionic solutions
at practical concentrations. The pure ionic systems, such as liquid sodium
chloride or silicate, require deeper examination, particularly from a theor-
etical viewpoint, for at present it is only possible to provide a quantitative
calculation of transport properties if knowledge of an experimental para-
meter, the melting point, is available. Besides these feelings of a desire to
finish the job which was begun in 1930 and 1940, there is a reasoned case
for including "ionics" in the field of electrochemistry. If electrochemistry
is basically the study of the electrified interphasic region between an elec-
tronic and an ionic conductor and of the transfer of electrons across such
Period Electrochemistry Becomes
Electrochemistry Becomes
of Early Nernstian Hiatus Rise of lonics Predominantly Concerned
an Interdisciplinary Area
Discovery with Interfaces

Muscles Overthermodynamical Study of ions in so- The kinetics of electrical re- Electrochemistry plays a role in
move when approach prevents de- lutions, adjunct to actions at interfaces is for- space power, stability of mate-
currents velopment of charge electrochemical studies, mulated, and some basic pro- rials, functioning of biological
pass transfer kinetics at in- becomes identified with gress in investigating its mech- cells, nylon synthesis, vehicular
terfaces electrochemistry itself an isms is made transportation
in England and America

1791 1910-1940 1920-1940 1950's 1960's

m
Fig. 1.12. Electrochemistry, a perspective from afar. e-
m
n
-i
:D
o
n
:x:
m
s:
CJ>
-i
:D
-<
N
W
24 CHAPTER 1

Date Period Concerned Main Events

1791-1830 The initial period of Muscular movements are associated

great experimental with electric currents; there is a "pro-
discoveries found connection" of biological and
electrical events; a definite amount of
current produces a definite weight of
deposited material; electric energy can
be produced directly from chemical
reactions

1890-1905 Electrode kinetics makes Tafel's equation connects current and

a hesitant beginning, but interfacial potential
is potentiocentric and
soon fades

1891-1947 The great Nernstian Nernst's first paper on the thermo-

hiatus dynamics of galvanic cells (1891);
thermodynamically oriented electrode
"kinetics" still dominates Faraday
meeting of 1947

1920-1940 The (substitutional) rise The theory of Oebye and Huckel is the
(and fall) of ionics do- first statistical-mechanical theory of
minates electrochemistry solutions to give quantitative success.
But it is shown to be intrinsically limit
ed to dilute solutions. No one breaks
through this barrier.

1940-1950 lonics, as well as elec- lonics stuck in mathematical difficul-

trodics, becalmed ties for concentrated solutions. Elec-
trodics not yet fully unfrozen from
Nernstian hiatus, except in Russia,
where it is developed through 1940's.

1924-1941 Isolated papers give ba- Butler interprets Nernstian potentials

sis to awakening of kinetically; Volmer gives a theoretical
1950's formulation of current-potential rela-
tion; Gurney introduces a quantum-
mechanical approach to charge trans-
fer, but it is attacked and forgotten;
Frumkin connects current to double-
layer structure; Horiuti formulates a
statistical mechanics of interfacial rela-
tions; Eyring formulates a current-
potential relation in terms of the theory
of absolute reaction rates.

Fig, 1.13. Electrochemistry, a perspective from a medium distance.

 ELECTROCHEMISTRY 25

Date Period Concerned Main Events

1949-1960 Electrochemistry awa- International Committee of Electro-

kes; fundamental re- chemical Thermodynamics and Kinetics
search becomes wide- (CITCE) formed, 1949; Understanding
spread and intense. It of concept of overpotential and
is strongly oriented to exchange current density becomes
ki netics of process at general among fundamental electro-
interfaces chemists; methods of examining the
mechanism of electrochemical reac-
tions becomes established; two inter-
national journals of electrochemistry
are founded; NASA chooses electro-
chemical path for space power; the
silver-zinc cell provides high electro-
chemical power density; Vetter writes a
general textbook on electrode kinetics.

1960 Electrochemistry be- Quantum mechanics receives increased

comes realized as an attention: Gurney's view is revived.
interdisciplinary area. Space vehicles powered electrochem-
Its applications spread ically; quantitative verification of
widely. electrochemical theory of failure under
stress; Del Duca proposes an electro-
chemical theory of the biological cell;
nylon synthesis by an electrochemical
route commences; Ford announces in-
tention to introduce electrochemically
powered cars.

Fig. 1.13. (continued).

a region, then it is not possible to do much in the theoretical direction unless

one knows the structure of both of the phases on either side of the inter-
phase. The structure of the electronically conducting phase-the metal or
semiconductor-is well taken care of in the physics of the solid state. Who
is to study the solution side except the electrochemists? It is they, primarily,
who need to know about it.
Shall such an attitude of keeping ionics as a part of electrochemistry
be permanent?
There are some difficulties to it. The subjects of electron transfer across
interfaces and of the interactions of ions in solutions differ considerably.
They have different parentage, the latter in the theory of solutions and bulk
properties of liquids, the former in interfacial chemistry and electrostatics.
Although ionics is a field of great breadth of application, its fundamentals
26 CHAPTER 1

are not interdisciplinary. It is a conceptually straightforward part of the

applications of statistical mechanics to the interactions of particles in liq-
uids, and its difficulties are largely mathematical. It might better be treated
under the "physical chemistry of solutions."
Finally, what turns out to be regarded as a field in a given period of
time cannot be decreed. It works itself out. Let us wait another decade or
two to determine the wisdom of keeping ionics as well as electrodics within
electrochemistry. Here, we take the attitude that we need a knowledge of
ionically conducting fluid phases if we are to understand the structure of
electrified interfaces and deal with interfacial charge-transfer reactions.
Nearly half of what we shall describe here as electrochemistry will be about
ions in solutions and in pure ionic liquids. But, it should be understood,
we regard ionics as the supportive aspect in electrochemistry, and we
wish to solicit the reader's support in thinking of our subject primarily as
the study of the structural and other properties of the interphasic region
and of charge transfer across interfaces.

1.4. THE RELATIONS OF ELECTROCHEMISTRY TO OTHER

SCIENCES

1.4.1. Some Diagrammatic Presentations

Let us look at Figure 1.14 to see something of the parentage of con-

ventional electrochemistry, in both the ionic and electrodic aspects.
We could also look at these relations in a different way and derive
electrochemistry with the dominant emphasis on charge transfer at inter-

Fig. 1.14. Physical chemistry and electrochemistry

Thermodynamics Metallurgy of defects on Quantum mechanics
of this situation at metal surfaces of transfer of electrons
equilibrium through barrier at interface
Fick's second law diffusion
Crystallography of surface theory of time dependence of
concentration

Electronics of I- -----1 Physical chemistry of solu-

circuitry to control tions
potential across I Electronic e Ionic I
interfaces ED Statistical mechanics of par-
conductor conductor 1
I ticle distribution near interface
in field
1- 1
Hydrodynamics of flow of so-
Physics of energy Surface chemistry of Spectroscopy of acceptor lution, transports ions to surface
levels in metals and intermediate radicals on particles, gives energy
semiconductors surface; and adsorption levels for electrons

Surface physics Physical chemistry

of electron overlap of surface reactions
,...m
potential very near m
(")
surface of metal Metallurgy, e.g., sputtered -l
::0
film formation on surface o(")
:t:
m
Optics of examination of surfaces s:
en
-l
::0
-<
Fig. 1.15. Some disciplines invoved in the study of charge transfer at interfaces.

"'"
28 CHAPTER 1

faces and the interdisciplinary character of the fields involved in studying it.
Such a view is implied in Figure 1.15, where space limits mention of the
number of disciplines which are associated with the study of electrified
interfaces.
Apart from the large number of areas of knowledge which are needed
to advance modern electrochemistry, there are many areas to which it
contributes or even plays the essential role. Because much surface chemistry
under real conditions concerns electrified interfaces and double layers, the
systems to which electrochemical concepts are relevant are nearly as wide
as surface chemistry itself. This, together with the fact that the subject
embraces interactions between electric currents and materials-i.e., between
two large areas of physics and chemistry -, signifies the widespread charac-
ter of the fields of knowledge which electrochemical considerations underly
(Fig. 1.15).

1.4.2. Some Examples of the Involvement of Electrochemistry

in Other Sciences

Chemistry. Chemistry itself contains much basic material which origin-

ated in electrochemistry. The third law of thermodynamics was conceived
on the basis of the temperature variations observed in the heat-content and
free-energy change in cell reactions. The concepts of pH and dissociation
constant were formerly studied as part of the electrochemistry of solutions.
Ionic reaction kinetics in solution and the Bronsted-Bjerrum relation is
expressed in terms of the electrochemical Debye and Huckel theory. Elec-
trolysis, metal deposition, syntheses at electrodes, plus perhaps as much as
half of the modern methods of analysis in solution depend on electrode
kinetics. All colloid chemistry is essentially dependent on the physical
chemistry of electrified interfaces.

Metallurgy. An electrochemical aspect has already been mentioned in

respect to the properties of materials and stability. The extraction of metals
from ores dissolved in molten salts, the separation of metals from mixtures
in solution, and the formation of single crystals from metals in solution
are among others.

Engineering. Examples abound because of the interest in driving ve-

hicles by electrochemical power sources. Part of this interest arises because
electrochemical power sources avoid the buildup of CO 2 in our atmosphere
and thus avoid the deleterious effects of this CO 2 buildup (cf Section 11.2).
Electrochemical energy converters, the fuel cells, provided the onboard
 ELECTROCHEMISTRY 29

power for the first space vehicles, and there are general prospects of evolution
away from the thermal to the electrochemical method of realizing the energy
of chemical reactions. One of the more certain aspects of the future is the
cheapening of electric power, with a sharp fall in the cost when controlled
hydrogen fusion becomes practical. Such changes would clearly favor
economically the increasing adoption of electrochemical processes. Elec-
trochemical engineering systems include those of a large proportion of the
nonferrous metals industries, in particular, the production of aluminum by
deposition from a molten salt containing the oxide.

Biology. Food is converted to mechanical work by biochemical mech-

anisms which have an efficiency much greater than that from some corre-
sponding form of energy conversion involving the heat-engine principle.
The mechanism involves electrochemical reactions. Correspondingly, the
mechanism of the transmission of impulses through nerves, as well as that
of the stability of blood, and probably many of the macromolecules involved
in biological processes, depend on aspects of electrochemistry which concern
electrochemical charge transport and double-layer interaction.

Geology. An example of electrochemistry in geology concerns certain

types of soil movements. The movement of earth under stress depends on
the viscosity of the earth regarded as a slurry, i.e., a viscous mixture of
suspended solid in water with a consistency of very thick cream. Such
mixtures of material exhibit thixotropy which depends on the interactions
of the double layers between colloidal particles. These depend -on the con-
centration of ions which affect the field across the double layer and cause
the colloidal structures upon which the soil's consistency depends to repel
each other and remain stable. Thus, addition of ionic solutions to soils in
certain conditions may cause a radical increase in their ability to flow.

1.4.3. Electrochemistry as an Interdisciplinary Field, Apart from

Chemistry?

All fields in chemistry, e.g., that of the liquid state or of reaction kinetics,
have connections to other parts of the general subject; and, indeed, all
subjects treated under chemistry tend to be subject, as time goes on, to
treatments at the more advanced degree of sophistication attained under
the aspect of the science we call physics. Chemists dare to make approximate
treatments of relatively complicated subjects not yet simple enough for
physicists to tackle in a more exact way. But the involvement, e.g., of the
study of liquids and gaseous reaction kinetics is largely with the parent
 30 CHAPTER 1

I
Electronics:
Concerns interfacial n-tYP~m 0 ;;;ype
charge transfer between G"~"i'~mOOi,m
two electronic conductors.

Electrodics:
Concerns interfacial /Pla/t~inum/
///
/:
'%
charge transfer between eHYdrOChlOric
acid solution
an electronic and an ionic
conductor.

0
Fig. 1.16. Electronics and electrodics.

areas of chemistry itself (see Fig. 1.14)-e.g., statistical and quantum

mechanics. A direct connection to areas right outside chemistry does not
exist for the two examples of other fields of physical chemistry given above.
In electrochemistry, however, there is an immediate definitional con-
nection to the physics of current flow and electric fields. Further, it is difficult
to pursue interfacial electrochemistry without an acquaintance with some
principles of theoretical structural metallurgy and electronics as well as of
hydrodynamic theory. Conversely (see Section 1.4.2), the range of fields in
science in which the important steps are controlled by the electrical proper-
ties of interfaces and the flow of charge across them is great and exceeds
the range to which other areas of what is specifically physical chemistryt
maintain relevance. In fact, so great is it that a worker who is concerned,
e.g., with the creation of passive films on metals and their resistance to
environmental attack is scarcely in intellectual contact, e.g., with the man
who is interested in attempting to find the model for why blood clots or
with him who is seeking to extract a metal from sea water.
These widespread involvements with other areas of science, in respect
to both the family of disciplines which compose the considerations involved
in electrochemical thoughts, and the breadth of applications of electrical
aspects of surfaces and surface reactions throughout the sciences connected

t As apart from areas of basic science, e.g., quantum mechanics, which primarily originate
in physics and underlie all chemistry, including, of course, electrochemistry.
 ELECTROCHEMISTRY 31

with materials, may suggest that electrochemistry be handled increasingly

in the future as an interdisciplinary area rather than a branch of physical
chemistry.
Correspondingly, there exists a general tendency at the present time
to break down the older formal disciplines of physical, inorganic, and
organic chemistry and to make new groupings. That of materials science
-the solid-state aspects of metallurgy, physics, and chemistry-is one.
Energy conversion-the energy-producing aspects of nuclear fission, elec-
trochemistry, photovoltaics, thermionic emission, magnetohydrodynamics,
etc.-is another. Electrochemistry would be concerned with the part played
by electrically charged interfaces and interfacial charge transfer in chemistry,
metallurgy, biology, engineering, etc. Perhaps it would also involve areas
which are directly related to such charge transfer, as is the physical chemistry
of ionic solutions (see Section 1.3.4).
The precise subject matter considered covered by the term electro-
chemistry will evolve. But the divisions of science into physics, chemistry,
metallurgy, biology, etc., are purely arbitrary and were made with a nine-
teenth century perspective. It may be that a time for regrouping and for
the recognition of new areas, homogenized by a common aspect of sufficient
importance and breadth of relevance in other sciences and to technology,
is here-or may be overdue.

1.5. ELECTRODICS AND ELECTRONICS

Aristotle wrote in his Poetics: "What is important to master is meta-

phor. It is the mark of genius." The direct metaphor seeks things which
may be likened to each other. Much may be learned from such likenesses.
Electrochemical systems are like electronic systems. What are electrodic
[sic] systems? They are two-phase systems in which one phase, an elec-
tronic conductor, is connected by an electric field (across which electrons
sometimes pass) to the other phase, an ionic conductor. What are electronic
systems? They are two-phase systems, both phases electronically conducting,
between which an electron passes, as shown in Figure 1.16.
In both subjects, interfacial electron transfer is the essential event.
In one, electrodics, it takes place from a phase in which electrons are the
principal charge carriers to a phase in which ions are the principal charge
carriers. In the other, electronics, the transfer of charge is to a phase in
which the electron does not remain localized in one particle but is free
to continue movement in the new phase. Superficially, there is not a great
32 CHAPTER 1

deal of difference between the initial theory of electronics and electrodics.

This can be seen by comparing the basic equation in both fields. In elec-
trodics, it is

where i is the rate of passage of electrons across the interface, A and fJ are
constants for the system, Cion is the concentration of ions in contact with
the surface of the electronic conductor, and 'YJ is the disturbance of the
potential difference across the interface from that corresponding to electron-
transfer equilibrium. In electronics, the initial equations is that for ther-
mionics, i.e.,
i = A'e- MJ / RT

where the terms have analogous meanings, ,d(]) being the difference in
energy of the electron inside and outside the electronic conductor from
which it is ejected.
A similar analogy exists between the behavior of electrons and holes
in semiconductors and the behavior of H+ and OH- ions in electrolytes.
Insights of this type are important to make and should be sought because
it is clear that "all Science is one," and, in a remote sense, all phenomena
take place in consistency with the same general laws of physics. It is helpful,
then, to try to break down the barriers between fields distinguished largely
by having different names and, unfortunately, different terminologies.
Often, one finds parts of one's problem have been solved in another field
or the other field has a problem to which one's own field has a solution,
or nearly so. Hence, electronics should be regarded as a neighboring subject
to electrodics not only because electronic circuitry external to the electrodic
system is essential in helping electrodics but because there is a fundamental
closeness in electrodic, and some ionic, systems to those studied under the
terms electronics and thermionics.

1.6. TRANSIENTS

Hitherto, we have referred to the interfacial transfer of electrons as

though this were always a steady-state process, i.e., one occurring at a rate
invariant with time. The corollary is, of course, that, because of the law
(cf Fig. 1.9)
i = Ae Bn

the interfacial potential difference across the double layer must be constant.
 ELECTROCHEMISTRY 33
Reference
vollaoe

1---'""\ Cell

Voltaoe
level
Fig. 1.17. An electronic circuit which produces transient
behavior in an electrodic circuit.

It follows therefore that, if, between the power source and the cell in the
circuit in Figure 1.17, we add a device which enables us to vary the 'I] (the
variable part of the potential difference across the interface) with time,
the rate of charge transfer will vary. Further, the delay time between a
change of potential and the resulting current is (1) small and (2) charac-
teristic of the details of the type of process occurring across the interface.
(For example, does the rate at which the electrons are being transferred
control itself in a relatively simple way, e.g., just vary at the behest of the
potential as the above equation indicates, or is the charge-transfer reaction
rate indirectly controlled by something else, such as the adsorption of some
entity from solution which might conceivably partly block the transfer site,
with the degree of blockage also a function of potential but a different
function from that of charge transfer?)
In the last section, an analogy was made between the fundamental
processes involved in charge transfer at an electronic-conductor-ionic-
conductor interface and those at an electronic-conductor-electronic-con-
ductor interface. It has not yet been shown, however, that electronics is
connected with electrochemistry in another way, i.e., as a tool, the value
of which can hardly be overestimated. Thus, consider the matter of trans-
ients. By electronic circuits of appropriate design, one may impose upon
the metal electrode electric-energy pulses of virtually any shape or size
and with widely varying time durations. One can make the interfacial current
dance to the tune being played by the electronic circuit. The time ranges in
which one can operate are as low as tens of nanoseconds.
Such ability to program circuitry to make metal-solution interfaces
34 CHAPTER 1

do things at command and very quickly has great significance for research
work on the electrode-solution interface. It allows us to change the electrical
interfacial conditions from a given condition so quickly that, before the
particles on the surface (intermediate radicals in some reaction, say) have
had time to adjust to the new order, they have been hit by a change of
potential which promotes a reaction which, e.g., electrically burns them
up and wipes them off the surface. Then, by finding out the current passing
during this process and integrating it with time, one has a measurement
of the total amount of electricity equivalent to the radicals on the surface
and hence a measure of the degree of coverage of the surface with radicals.
Such a quantity is often a vital piece of knowledge in research on reaction
mechanisms and eventually, e.g., on energy converters.
Such a use of transient techniques makes a good example of why the
mechanism of electrical interfacial reactions has better prospects of being
understood than have classical heterogeneous "chemical" reactions. Elec-
trochemists have known of transient techniques to obtain the concentration
of radicals on noble metal surfaces for about thirty years, but the correspond-
ing technique of the heat pulse in thermal kinetics has had a much shorter
history. Flash photolysis, a transient technique in gas kinetics, was intro-
duced twenty years after the effective and more powerful transient technique
available for radical investigation and control in electrochemical processes.
The advantage of the electrochemical measurements is that one has really
just to count electrons, and, for that, there are available a large number
of rather sophisticated techniques.

1.7. ELECTRODES ARE CATALYSTS

Here again, as with electrodics and electronics, there is a comparison
which is informative. When chemical reactions occur on surfaces-heter-
ogeneous reactions, as they are called-, there is generally a great dependence
of the rate upon the nature of the solid surface concerned. On some sur-
faces, the rate may be so small that the reaction is said not to take place
at all. On others-they may be metals but are often also oxides-the rate
is large; such surfaces are said to catalyze the reaction concerned, i.e., they
make it go faster without themselves undergoing any change in the reaction.
The electronic conductors which are the source and sink of electrons
in electrochemical interfacial electron-transfer reactions are called elec-
trodes, and a few experimental facts quickly show that electrodes are
catalysts. For example, see Table 1.1 of rates (expressed as the current
passing equally in both directions across the surface when this is held at
 ELECTROCHEMISTRY 35

TABLE 1.1
The Rate of the Hydrogen Evolution Reaction on a Series of Metals at
the Equilibrium Potential

log rate, log rate, log rate,

Metal Metal Metal
amp cm- 2 amp cm- 2 amp cm- 2

Pd -3.0 Ag -6.1 Sn - 8.2

Pt -3.1 Nb -6.4 TI -10.2
Rh -3.2 Mo -6.5 Cd -10.7
Ir -3.7 Cu -6.7 Mn -10.7
Ni -5.2 Ta -7.0 Ga -11.0
Fe -5.2 Bi -8.1 Pb -11.3
Au -5.7 Al -8.0 Hg -12.0
W -5.9 Ti -8.2

the potential which corresponds to equilibrium) for the electrochemical

reaction of hydrogen evolution

Thus, for this reaction, there is a very great dependence of the reaction
rate upon the surface chosen. t Electrodes, then, are catalysts; but the science
of electrocatalysis is in its infant stages because, like so much else in elec-
trochemistry, the concept of electrodes as catalysts is only just becoming
a concept-it was first written of in 1963 by Grubb. The rate at which
a reaction occurs for a given departure from equilibrium in the electric
potential applied across the interface depends on the constant A of the
equation i = Ae Bn , and A must therefore be a kind of chemical rate constant,
within the electrochemical reaction, upon which is superimposed the in-
fluence of the interfacial electric field.

1.8. THE ELECTROMAGNETIC THEORY OF LIGHT AND THE

EXAMINATION OF ELECTRODE SURFACES

It will be recalled that electrochemistry is the study of electrified inter-

faces with special reference to electron transfer across them. A well-estab-
lished piece of physics deals with the band structure of the energy levels

t In fact, the variation of some 1010 times is far greater than the range of variation of
rates observed for a given reaction on various catalysts in thermal catalytic reactions.
36 CHAPTER 1

in the electronic conductor. A good picture of the solution side is also

available for dilute solutions (though some of the theory of the acceptor
states in the molecules in solution had to be developed by electrochemists).
What is little understood is the state of the electrode surface in contact
with a solution. Of course, the surface of the liquid metal mercury is fairly
well understood, but work on mercury surfaces in electrochemistry has
become increasingly relegated to chemists interested less in fundamental
advances in, e.g., electrocatalysis and more in using electrochemistry as a
tool, perhaps for the analysis of compounds dissolved in solution. The
modern trend is toward the study of solid surfaces. To work with them
without knowing their structure is a little like attempting to determine the
mechanism of decomposition of a solid without first knowing the crystal
structure of the substance concerned.
It has been pointed out that there is an advantage in the electro-
chemical way of studying the mechanism of reactions compared with the
chemical way because, in the former, one can make a determination of
the degree of coverage of the surface by radicals and sometimes distinguish
one radical from another. However, this advantage would be counteracted
if one had no way of looking at surfaces on electrodes in situ (one can look
at catalysts used outside solutions, e.g., those used in classical heterogeneous
reactions, by means of electron microscopy). The necessity for in situ
examination of electrode surfaces gains added significance because the
presence of adsorbed layers and oxide films on surfaces is the rule rather
than the exception for reactions on electrode catalysts.
It is of interest to remind ourselves why high-energy radiations, e.g.,
electron beams or X rays, cannot be used for looking at electrode surfaces
in situ. It is because they interact strongly with the solution and thus get
absorbed. The energy necessary to get electrons to pass through even a
few millimeters of aqueous solution and still have energy to send back a
message after reflection from the electrode is too much to make such methods
practical. Thus, there is the significant difficulty that the electron-optical
and X-ray methods used for surface studies in solid-state chemistry cannot
be used in the case of electrode catalysts, the seat of the charge-transfer
reactions which are the foundation of electrochemistry.
This difficulty can be compensated by using visible light, which can
pass with facility through the solution, be reflected by the electrode surface,
and emerge out from the solution (Fig. 1.18). Ifthe incident light is polar-
ized, then the process of reflection produces changes in the state of polar-
ization. These changes are determined essentially by the state of the sur-
face-the existence and nature of the film on the electrode-and can be
 ELECTROCHEMISTRY 37

Monochromatic
light source
Photomultiplier

Reference
electrode

Quarter-wove
plate

Fused
Mirror quartz
electrode cell

Fig. 1.18. Determination of certain properties of film-covered surfaces

in solution by ellipsometry.

understood by using the electromagnetic theory of light and, in particular,

Fresnel's equations for the reflection of polarized light.
This technique of ellipsometry was originated by Tronstad as early
as 1937, but it has only been in the past few years that new and powerful
ways of using ellipsometry to study electrode surfaces have developed. It
seems likely that its use will spread. Applied to the detection and measure-
ment of adsorbed films, it should give an optical means of following the
concentration of radicals on the surface while the electronic instrumentation
follows the rate of the charge transfer. Further, by using different frequencies
of light, a kind of ellipsometric spectroscopy may be devised.

CJ
A

--+---B
F E D

G
Fig. 1.19. Schematic of polarized-light microscopy on a
surface in solution; A, light source; B, polarizer; C, semi-
silvered mirror; D, Wolloston prism; E, microscope ens;
F, object; G, analyzer.
38 CHAPTER 1

Other methods of using polarized light involve interferometry with

polarized light by which irregularities of a few hundred angstroms in the
surface structure can be observed. Interference contrast microscopy, also
using polarized light, lets us observe entities on the surface which are about
half the wavelength of light in extent (Fig. 1.19). Both these kinds of
microscopy use visible light and not electrons; hence, they can be used
in the presence of solutions (but, conversely, do not resolve the picture as
well as electrons because the wavelength, and hence resolving power, of
visible light is in the range of thousands of angstroms, and that of electrons
may be made to be a few angstroms.)
Electron microscopy can be and is used for obtaining the state of the
surface of an electrode if the latter is removed suddenly from the solution
while the current is still passing. Replicas of the surface structure which
correspond to that at the moment of removal can then be made. If the
current were just switched off and the electrode left in the solution, it
would have a surface which would be difficult to associate with any clear
condition of potential across the interface because of the growth which
may continue after cessation of current. As this further growth involves
strictly charge-transfer mechanisms and these demand the presence of a
solution, the removal of the electrode from the solution leads to a cessation
of current and prevention of further growth.
The use of electron microscopy in this way (with the electrode surface
frozen at a certain moment and then examined) is of course hardly a conven-
ient way of following changes on the electrodes surface, e.g., with potential.

1.9. SCIENCE, TECHNOLOGY, ELECTROCHEMISTRY, AND

TIME
1.9.1. Do Interfacial Charge-Transfer Reactions Have a Wider
Significance than Has Hitherto Been Realized?
It is informative, in this chapter, to make some attempt to fit electro-
chemistry among the sciences to obtain the relative measure of its signifi-
cance. It is difficult to do this for a field, the modern phase of which dates
back only 10 to 15 years. Let us try, but let us realize that what we are
doing here is making a speculation, although we shall give some of the
reasoning which makes it seem, at this time anyway, a probable one.
Firstly, we can ask a relative question: Is interfacial electrochemistry
simply a special aspect of reaction kinetics, somewhat analogous to photo-
chemistry? There, one might say, one studies the effect of energy packets
striking molecules by means of radiations; and, in electrochemistry, one
 ELECTROCHEMISTRY 39

studies the effect of striking molecules dissolved in solution with electrons

emitted from electrically charged conductors.
Or is the significance of interfacial electrochemistry something more?
The evidence [or a point of view tending to this latter direction is as
follows:
1. Interfacial electrochemistry has a relatively high degree of ubiquity
compared with those of other branches of knowledge outside physics. A
way to appreciate this is to realize how often one is concerned with electro-
chemical phenomena outside the laboratory. For example, one shaves, starts
the car, and listens to its radio on battery power; television pictures of
the moon are transmitted from space vehicles to the earth by fuel-cell
power; the office building may be partly made of electrochemically extracted
aluminum; the water which is in the coffee may be obtained by electro-
chemical deionization from impure or brackish water; one may ride a
car with a magnesium-alloy engine block produced from seawater by an
electrochemical process; one uses clothes of nylon produced from adipo-
nitrile, which is electrochemically synthesized; the gas in one's car contains
lead tetraethyl, probably produced electrochemically; one may take a
tranquilizer which exerts an effect by an electrochemical mechanism; one
adds an inhibitor to the radiator fluid of a car to reduce electrochemical
corrosion. Finally, one thinks by electrochemical mechanisms, and one's
blood remains functional insofar as the electrochemical potential at the
interface between its corpuscles and their solution remains sufficiently high.
2. This ubiquitous role of electrified interfaces throughout many aspects
of science suggests that electrochemistry should not be regarded as a branch
of chemistry. Rather, while most chemists have concentrated upon ther-
mally activated reactions and their mechanisms, with electrochemical reac-
tions as some special academic subcase, there is a parallel type of chemistry
not based on the collisions of molecules and the energy transfers which
underlie these collisions but on interfacial electron transfer. It is this chem-
istry of interfacial electron-transfer processes that seems to underlie much
of what goes on around us.
Some examples of alternative thermal and electrical approaches to
common chemical happenings may illustrate this (Table 1.2).

1.9.2. The Relation between Three Major Advances in Science,

and the Place of Electrochemistry in the Developing World
If one stands sufficiently away from one's specialization in the sciences
and looks as far back as the nineteenth century, then three great scientific
40 CHAPTER 1

TABLE 1.2
Examples of the Alternative Thermal and Electrochemical Paths
in ChemiGal Happenings

Phenomenon or process Thermal Electrochemical

The determination of free- Determine equilibrium con- Determine thermodynamic

energy changes and equi- stant and use cell potential and use
librium constants in chem- LlGO = -RTln K LlGo = -nFE
ical reactions

Synthesis, e.g., water from Occurs heterogeneously pre- Occurs in electrochemical

hydrogen and oxygen sumably by noncharge- cell by reactions
transfer collisional proces- H2 --)0 2H+ + 2e
ses H2 + t02 --)0 H20 t02 + 2H+ + 2e --)0 H 20
H2 + to. --)0 H.O

Biochemical digestion Series enzyme-catalyzed Some enzymatic reactions

chemical reactions may act through elec-
trochemical mechanisms
analogous to local cell
theory of corrosion

Many so-called chemical TiCI.+2Mg --)0 Ti+2MgCI. 2Mg --)0 2Mg++ + 4e

reactions, e.g., chemical (apparently a thermal col- TiH + 4e --)0 Ti
synthesis of Ti lisional reaction) 4CI- -)- 4CI-

Production of electrical H. + to, explodes, pro- H, and 0, ionize on elec-

energy duces heat, expands gas, trodes, as above in this
causes piston to move, column, and produce
and drives generator current

Storage of electrical energy Electricity pumps water up Allow to cause some elec-
to height and allows it to trochemical change, e.g.,
fall on demand to drive Cd++ + 2e --)0 Cd, Ni++
generator --)0 NiH + 2e, which will

be reversed on demand

Deionize water Distill it and leave behind Electrodialyze through

nonvolatile ions membrane

Syntheses of inorganic and 2AI,Oa + 3C --)0 3C02+ 6AI; A!3+ + 3e --)0 Al

organic material, e.g., AI, Tetrahydrofuran --)0 cathodic
H.C = HC - C N --
coupling
adiponitrile I,4-dichlorobutane --)0
NC - (CH.). - CN
adiponitrile
 ELECTROCHEMISTRY 41

TABLE 1.2 (continued)

Phenomenon or process Thermal Electrochemical

Spreading of cracks through Amount of stress at the Bottom of crack dissolves

metal apex of crack per unit anodically, obtaining cur-
area is so high that crack rent from local cell formed
is propagated into metal with surface (which IS an
bulk electron donor, probabl~
to O2 from air)

Practical enhancement of Cover surface with coating; Appyl electron-donating

stability of metal prevent access of air field, reduce electron ac-
ceptance, and hence re-
duce M -+ M2+ + 2e

contributions stand out in the matter of their impact on science and tech-
nology. They are:
1. The electromagnetic theory of light due to Maxwell (nineteenth
century)
2. The theory of relativistic mechanics due to Einstein (nineteenth
to twentieth centuries) t
3. The theory of quantum mechanics originating with the work of
Planck, Einstein, and Bohr and developed by SchrOdinger, Heisen-
berg, Born, and Dirac (twentieth century)
It is important to recall why these contributions to physics and chem-
istry are regarded as so outstandingly important. Maxwell's nineteenth
century theory is the basis for a large fraction of twentieth century tech-
nology. It provides the basis for the transfer of energy and communication
across distances, and the delivery of mechanical power at command as a
consequence of the controlled application of electric currents. The signifi-
cance of relativistic mechanics is that its theory of mass-energy equivalence
constitutes the basis of the generation of nuclear power which, in association
with electrical and electronic devices, makes it possible in decades rather
than centuries to end man's necessity of having to work to be able to live.
The quantum theory is the basis of the solid-state device and transistor
technology and has given us such a great revolution in thinking, e.g., the

t Einstein's first paper was of 1905, but the thought mode and many consequences were
based on thoughts typical of the nineteenth century, in particular, the time necessary
for the propagation of radiation.
42 CHAPTER 1

knowledge that macroscopic and microscopic systems behave in funda-

mentally different ways and insight into many phenomena at a molecular
level (particularly the understanding of the permanence of genes), that it
takes its place with the greatest of the contributions to knowledge.
For how is weighed the eventual magnitude of a contribution in science?
Is it not the degree to which the applications which arise from it eventually
change everyday life? Is not the essence of our present civilization the
controlled development of the surroundings at man's command?
It is in this light that one may estimate the relative significance of the
theory of electrified interfaces and thus of electrochemistry. It is of interest
to note how interfacial charge-transfer theories are based on a combination
of the electric currents of Maxwell's theory and the quantum-mechanical
tunneling of electrons through energy barriers.
A number of illustrations have been given to support the statement
that electrochemical mechanisms are relevant in many fields of science.
The nineteenth century contributed to physics the theory of electromag-
netism. The twentieth century contributed to physics the relativistic theory,
an implication of which is the provision of limitless cheap energy, and the
quantum theory, which underlies solid-state electronics and has promoted
a fundamental revision of thinking in chemistry and biology. In the twenty-
first century, it seems reasonable to assume that the major preoccupations
will be in the direction of taking up and working out the consequences of
these developments toward a predominantly electrical postindustrial tech-
nology from which noncreative work will have been eliminated.
Two very general types of probable advances can be made out. One
is in the direction noted, for example, the development of practical photo-
galvanic energy-conversion devices for the direct conversion of light to
electricity, or the development of replacement of parts in the body by
artificial devices connected with body circuits.
The other type will be those developments necessary as corollaries of
the interference with nature of the last 50 years, for example, electrically
powered vehicular transportation to avoid increasing the CO2 content of
the atmosphere and processes for the reduction of the presently polluted
state of much useful water, and of the present atmosphere.
Electrochemistry is the basic science upon which many of these elec-
trically oriented advances of the next few decades will be founded. It under-
lies electrically powered synthesis, extractions (including fresh from brackish
water), machining, stabilization of materials, storage of energy in the form
of electricity, efficient conversion to electricity of the remaining fossil fuels,
and the basis for practical developments in molecular biology.
 ELECTROCHEMISTRY 43

It is worthwhile thinking, too, that the mode of organization of urban

living is likely to develop as a function of available electricity from atomic
sources. Corresponding to such a basic development will tend to come an
increasing need to invest resources in recovering from atmospheric and
aqueous pollution (even of the sea) and an increasing exhaustion of
supplies.
In such a situation, the town will tend to be built around the reactor.
The only form of energy used will be electrical. The town will be very
largely self-contained. Little material mass will leave or enter it. The proc-
esses on which it will be run will be all electrical, and those involving
matter, therefore, electrochemical. Transportation will use electrochemic-
ally stored electricity; manufacturing and machining processes and recovery
of materials used or degraded will all be electrochemical. Polluted liquids
will be electrodialytically regenerated. Sewage will be processed electro-
chemically. Medical electronics-the electronic-electrodic combination in
medical research-will be highly developed.
Thus, by the end of the present century, it seems reasonable to expect
the achievement of several electrochemically based innovations: a great
extension in the use of cordless devices; the provision of cheap heat elec-
trically from storers charged during off-peak times; electrochemically power-
ed vehicles and perhaps ships; a viable photogalvanic solar conversion
system; extensive use of electrochemical machining and electrochemically
based tools; the electrochemical production of metals of great strength;
an internal fuel-ceIl-powered heart; and electrometallurgical extractions of
materials on a large scale from the moon. Many more areas of probable
advance could be mentioned. A large, immediately developable area lies
in the electrochemical aspects of molecular biology and in the development
of circuitry which will join the brain and its probably electrochemical
mechanisms to artificial limbs with their electrochemical functions and,
perhaps, even to circuits not connected to the body.
Thus, a press of advances in the electrical aspects of matter, and there-
fore in electrochemistry, seems likely to develop, along with the transition
of the principal attention of material scientists from the bulk to the surface.
During the next few decades an increasing fraction of our surroundings will
have an electrochemical aspect.
Let us, therefore, read this book with the realization that it is written
early in an era in which the electrical aspects of existence will bear great
increase, that, often in that time, the detailed interpretive positions describ-
ed below as probable will be modified and improved, and that all those
prospects which lie before us in electrochemical applications depend greatly
44 CHAPTER 1

for their realization upon the fundamental work by which the relevant basic
mechanisms become understood.
To realize this vista, contributions from scientists of varied fields of
original training will be necessary. They have to learn about the field of
electrochemistry. It will be necessary to train electrochemists and electro-
chemical engineers in far greater numbers than is now done. The stimulus
for the writing of this book is the need to provide these workers with
material which assumes a background knowledge only of freshman physics
and chemistry and yet takes the student in most of the areas considered
to a position in which he can comprehend the specialist monograph in
electrochemistry.
CHAPTER 2

ION-SOL VENT
INTERACTIONS

2.1. INTRODUCTION

An electrochemical system (Fig. 2.1) includes two interfaces, at each

one of which an electronic conductor is in contact with an ionic conductor
(or electrolyte t ). The electronic conductor is generally a metal but may
well be a semiconductor. The ionic conductor, as the term suggests, is a
material which consists of mobile ions.
How does one produce a medium of mobile ions? One method is
based on the fact that certain substances, which, in the pure form, do not
contain any significant concentration of ions, are able to interact to produce
ions. This is how neutral, i.e., nonionic, molecules of water and of acetic
acid interact to give an electrolytic solution of hydrogen ions and acetate
ions (Fig. 2.2 and Table 2.1). This chemical method of producing an ionic
conductor will be studied in Chapter 5.
Another approach is based on starting off with a solid ionic crystal
and reducing the forces which hold the ions together. A stage is reached
when the cohesive forces are so weakened that the ions, which could only

t The term electrolyte is used in electrochemistry to refer not only to the idnically con-
ducting medium through which electricity is passed but also to the substances which,
when dissolved (or melted), give rise to a conducting medium.

45
46 CHAPTER 2

E lec tronic Ionic Electronic

Co nductor ~ Conductor / Conductor

V
Interfaces

External
Electron
Source or
Sink

Fig. 2.1. The essential parts of an electrochemical system.

vibrate in the solid, acquire a new degree of freedom-the freedom of

translational motion.
There are two distinct ways in which the interionic forces in a crystal
can be overcome. One method is based simply on an agitational effect.
Heat energy is used to increase the tempo of the ionic vibrations in the
solid until thermal forces prevail and the long-range order of the ionic
arrangement in the crystal lattice is wiped out-the ionic crystal "melts"
(see Chapter 6).
One is left with a pure liquid electrolyte, a molten material teeming
with positive and negative ions and with free space which is far more
plentiful than in the solid. These ions are in ceaseless random motion
and ready to respond to applied electric fields by conducting electricity.
What has been described is a thermal method of obtaining a pure liquid
electrolyte.
There is, however, another way of overcoming the interionic forces in
an ionic crystal and producing mobile ions. This is with the aid of a solvent.
A crystal of potassium chloride, e.g., is placed in water. Soon it becomes
apparent that the crystal as an entity has disappeared. The solvent has

?o ~/H Proton - transfer

H3 C - C" + 0" reaction
O-H H

Neutral ocetic Neutral Acetate ion Hydrogen ion

acid molecule water
molecule
Fig. 2.2. The chemical method of producing ionic solutions.
 ION-SOLVENT INTERACTIONS 47

TABLE 2.1

Ionic Concentrations in Pure Water. Pure Acetic Acid.

and Acetic Acid Solution

Ionic concentration
g-ions liter-1 at 25 DC

Pure water 10- 7

Pure acetic acid 10-6.5

O.lN acetic acid solution 10-3

enticed the ions out of the solid so that they can wander off into the solvent
(Fig. 2.3). (The Greek word for wanderer is ion.) One has witnessed the
process of dissolution of an ionic crystal.
What are the influences which the solvent brings to bear upon the
ions of the crystal? What are the ion-solvent forces which overcome the
ion-ion forces holding together the crystal?
It is obvious that questions such as these are of central significance to
the understanding of ionic solutions and, hence, the electrochemical proc-
esses which occur in them. For the questions imply that ions in solution
are constantly affected by ion-solvent forces, and that, to understand the
behavior of ions inside an electrolytic solution, one has to reckon with
the forces arising from the presence of the solvent. One must understand
ion-solvent interactions.

GI -
6>-:\
~---.....
",/
~
/ _ ..
/
= /
N+ -@.
-
+
",
/ =
+

(@
.I""" , -
./
Na+ ~
r-- " -

rtf!J~ '\ Ion in solut ion

(9~~ \

~~ ~
of NoG I Water molecule

Fig. 2.3. Dissolution of an ionic crystal by the action of a solvent.

48 CHAPTER 2

2.2. THE NONSTRUCTURAL TREATMENT OF ION-SOLVENT

INTERACTIONS

2.2.1. A Quantitative Measure of lon-Solvent Interactions

A field of study often undergoes a qualitative change when the concepts

used can be associated with numbers and made quantitative. The problem,
therefore, is to develop a quantitative measure of ion-solvent interactions.
This type of problem is a common one in chemistry. It is often solved by
considering two situations or states, one where the interactions operate
(are "switched on") and the other where they do not exist (are "switched
off"), and then computing the free-energy difference LlG1 - S between the
two states (Fig. 2.4).
In the case of ion-solvent interactions, the state in which the interac-
tions exist is an obvious one; it is the situation in which ions are inside
the solvent. Ions are charged particles, and charges interact with other
charges. So there will also be ion-ion, as well as ion-solvent, interactions
in the solution. But the former are excluded in the quantitative analysis
of ion-solvent interactions; they will be given separate consideration later
on (Chapter 3).
Now, what is a situation in which there are no ion-solvent interac-
tions? Obviously, one in which there is no solvent. Hence, one must consider
an initial state in which there are large spaces between individual ions,
and nothing else present. The initial state, therefore, is that of ions in
vacuum at an infinitely low pressure.
The problem, therefore, is to consider the free-energy change for the
transfer of ions from vacuum to solution (Fig. 2.5).

Ions in vacuum -->- Ions in solution.

Recall, however, the thermodynamic relation (cf Appendix 2.1) which

states that, in a reversible process taking place at constant temperature and
pressure, the free-energy change is equal to the net work done on the sys-
tem, i.e., the total work done other than the work of producing a volume
change.

Initial state Final state

Free energy
No ion- solvent Ion-solvent
change. LlG I-S
Interactions Interactions

Fig. 2.4. The free-energy change arising from ion-solvent interactions.

 ION-SOLVENT INTERACTIONS 49

Ion

Free energy change = free energy

Vaccum of ion - solvent interactions

Solvent

Fig. 2.5. The free energy of ion-solvent interactions

is the free-energy change resulting from the transfer of
ions from vacuum into solution.

Hence, the basic problem of deriving an expression for the free energy
of ion-solvent interaction can be defined as follows. What is the work done
when one transfers an ion from vacuum into a position deep inside the
solvent? This work will include the energy of all the interactions between
the ion and the surrounding solvent, for example, water.

2.2.2. The Born Model: A Charged Sphere in a Continuum

A moment's thought will reveal that, to work out exactly all the ion-
solvent interactions, one must know the structure of the solvent, i.e., the
dispositions of all the particles constituting the solvent and the forces
between the ion and these particles. But the solvent, e.g., water, may have
a fairly complex structure. To understand this structure, one must be able
to answer a vast number of questions. For example, are there discrete
solvent molecules, or are they associated to such an extent that one should
not speak of separate molecules? What do the ions do to the solvent struc-
ture? Do they disrupt it, or are there spaces inside the structure so that
ions can be smuggled in but cause little damage to it?
The problem seems insuperable, but one can resort to modelistic think-
ing. Models are simplified representations of the real microstructure of
nature, often as mental pictures derived from the macroscopic world. They
are intended to reproduce approximately the essential features of the real
situation. The better they are able to predict experimental quantities, the
better do they serve as aids to thinking about how nature really works.
An example of a very crude and approximate model for ion-solvent
interactions is that suggested by Born in 1920. In the Born model, an ion
50 CHAPTER 2

Net
(Zj
Char~e
eo ) .,
Ij'
is equivalent to
~c h,,,.'" 'ph,,,
~. Ion

Nucleus Electron cloud

(a )

Structure-less
is equivalent to continuum
(with dielectric
constant.O!" s )

(h )

Fig. 2.6. The Born model for ion-solvent interactions considers (a) an ion
equivalent to a charged sphere and (b) the structured solvent equivalent to a
structureless continuum.

is viewed as a rigid sphere (of radius ri) bearing a charge ZieO (eo is the
electronic charge), and the solvent is taken to be a structureless continuum
(Fig. 2.6). Thus, the problem of ion-solvent interaction assumes the follow-
ing form: What is the work done in transferring a charged sphere from
vacuum into a continuum (Fig. 2.7)?
By considering a charged sphere equivalent to an ion, the Born model
is assuming that it is only the charge on the ion (or charged sphere) that is
responsible for ion-solvent interactions. The interactions between the solvent
and the ion are considered to be solely electrostatic in origin.
The Born model suggests a simple thought process for calculating the
free energy LlG1 - S of ion-solvent interactions, i.e., the work of transferring
an ion from vacuum into the solvent (Fig. 2.8). One uses a thermodynamic
cycle. The basic idea behind a thermodynamic cycle is the law of the con-
servation of energy. If one starts with a certain system (say, an ion in
vacuum) and then goes through a hypothetical cycle of changes, ending up
with the starting condition (i.e., the ion in vacuum), then the algebraic
sum of all the energies involved in the various steps must be zero. The
particular cycle that will be used is the following: (1) The ion (or charged
sphere) is first considered in a vacuum, and the work WI of stripping it
of its charge Zieo is computed. (2) This uncharged sphere is slipped into the
solvent; this process will involve no work, i.e., W 2 = 0 because the only
 ION-SOLVENT INTERACTIONS 51

Charged sphere

Vocuum

Free energy of Work of

ion-sol vent tronsfer
interoctions

is equivolent to

Continuum
(dielectric
constont*"s)

Solvent
molecules

Fig. 2.7. The Born model views the free energy of ion-solvent interactions as
equal to the work of transferring a charged sphere (of radius 'i and charge ZieO) from
vacuum into a continuum (of dielectric constant liS)'

)+---Charged sphere

rk of di schorg ing

)+---Uncharged sphere

No electrostatic
wark

Solvent---=--

Work of chorging

......- - Charged sphere

Fig. 2.B. Method of calculating the work of transferring a

charged sphere from vacuum into the solvent by a thermo-
dynamic cycle.
52 CHAPTER 2

interactional work is assumed to arise from the charge on the ion. t (3) Then,
the charge on the sphere inside the solvent is restored to the full value
zieO-one says, the sphere is charged up to the value ZieO-, and the charg-
ing work W3 is computed. (4) Finally, the ion is transferred from the
solvent to vacuum. Since this transfer process is opposite to that involved
in the definition of the free energy LlG1 _ S of ion-solvent interactions, the
work W4 associated with this last step of the cycle, i.e., the transfer of an
ion from the solvent to vacuum, yields -LlG1 - S '
Now, if the algebraic sum of the work terms associated with the steps
of the cycle is set equal to zero, one gets

or
Work of discharging Work of charging
. .
IOn III vacuum
+0+ ion in solvent
- Ll GI -s = 0

I.e.,
Work of discharging Work of charging
LlG1 - S = .
IOn
.
III vacuum
+ ion in solvent
(2.1)

2.2.3. The Electrostatic Potential at the Surface of a Charged

Sphere

In considering the work of charging up a sphere in a vacuum, one

starts off from the definition of electrostatic potential. To facilitate the
definition, it is assumed that there exists a reservoir of charge at an infinite
distance away from the sphere under consideration.
The electrostatic potential "P at a point in space is then defined as the
work done to transport a unit positive charge from infinity up to that point.
Thus, the potential "Pr at a distance r from a charged sphere is the work
done to transport a unit positive charge from infinity up to a distance r
from the sphere. The reason there is a need to do work is that the charged
sphere exerts an electric force on the charge being transported. The mag-
nitude of the potential, I "Pr I, i.e., the work done on the unit charge, is
given by the electric field Xr (i.e., the electric force operating on the unit
charge) times the distance r through which the charge is carried

I "Pr I = Xr·r (2.2)

t What happens when charges cross interfaces is discussed in Section 7.2.

 ION-SOLVENT INTERACTIONS 53

4-.-----Direction of i ncreasing potential r

Direction of Direc tion of field

• tronsport of chorge
•
Unit positive charge
Posit ively cho rged sphere

Fig. 2.9. The relative directions of the field due to a charged sphere, of the
movement of the test charge, and of increasing electrostatic potential.

The sign of the potential 1jJr is thought out as follows. Suppose the
sphere is charged positively. Then it exerts a repulsive force on the unit
positive charge, and the potential 1jJr' the work which has to be done by
an external agency in transporting the unit positive charge, i.e., overcoming
the repulsive interaction, will be taken to be positive. But the electric force
of the charged sphere on the unit positive charge, i.e., the field, acts in a
direction opposite to that in which the charge is being moved (Fig. 2.9).
Since both the field and the direction of transport are vectors (quantities
with direction and magnitude) and since the vectors point in opposite
ways, their product is negative. t Hence, to relate the positive potential 1jJr
to the product of the field and distance, it is necessary to state that

1jJr = -Xr·r (2.3)

Equation (2.2) for the electrostatic potential is valid only if the field
Xr acting on the unit positive charge remains the same, independent of the
distance of the unit charge from the source of the field. Suppose, however,
as will be seen to be the case with the field due to a charged sphere, the
field varies with distance from the source of the field. Then one must allow
for the inconstancy of the field in the definition of the potential at a point.
What one does is to take the field Xr as a constant over an infinitesimally
short distance dr. In this case, the electrostatic potential 1jJr at a point r is
obtained by summing up all the little bits of work, Xr dr, as the unit charge
is carried from infinity up to the point r in steps of length dr, i.e.,

1jJr = - J~ Xr dr (2.4)

t The product of two vectors A and B is AB cos (), where () is the angle between the
two vectors. If the vectors are in opposite directions, () = 7l and cos () = -1 and the
product is - AB.
64 CHAPTER 2

By inserting an upper limit of ri in the integration, one can indicate

that the unit charge has been brought up to the surface of the sphere.
Thus, the electrostatic potential at the surface of the charged sphere is

"P ' j = - I ri

00
Xdr
r
(2.5)

The electric force X, operating on a unit charge in vacuum is obtained

from Coulomb's law for the electric force F between two charges ql and
q2, i.e.,
(2.6)

where r is the distance between the charges. Thus, by setting ql = q and

q2 = 1, the electric force per unit charge (i.e., the electric field X,) due to a
charge q becomes
q
X , =r2
- (2.7)

Substituting for X, in equation (2.5), one gets for the potential at the
surface of the sphere
"Prj = - Ioorq dr
Ti
2

(2.8)

2.2.4. On the Electrostatics of Charging (or Discharging)

Spheres

The electrostatic potential at the surface of the sphere pertains to the

work of transporting a unit charge to the sphere; hence, the work done in
transporting a charge of any other magnitude is simply given by the product
of the potential and the magnitude of that charge. It will be noticed, how-
ever, that the electrostatic potential at the surface of the sphere varies with
the charge q on the sphere. So the work of adding on any charge to the
sphere depends upon how much charge q the sphere already has. This is
awkward. So the best thing to do is start with an uncharged sphere (q = 0)
and add charge onto it in little driblets or infinitesimal amounts, dq, each
of which requires an infinitesimal amount of work, dw, given by the product
of the potential and the infinitesimal charge dq, i.e.,

dw = "Prj dq (2.9)

This procedure is known as a charging process.

 ION-SOLVENT INTERACTIONS 55

If, therefore, one starts with an uncharged sphere of radius ri in a

vacuum and slowly builds up the charge from zero to a final value which
can be taken as ZieO, corresponding to a charge on an ion containing Zi
electronic charges, then the total work consists of all the little elements of
work, dw, i.e.,
W = f dw = f ·
zoeo

o
1jlr/q

= f -dq
z;eo
o
q
ri
=
[ - q2 ]z;eo
2ri 0

(2.10)

Obviously, the work of discharging a charged sphere in a vacuum is the

negative of the charging work because, in the discharging process, one is
taking away charge from a charged sphere, i.e.,

(2.11 )

Now that the process of discharging a sphere in a vacuum has been

analyzed, one can consider the charging process when the sphere is placed
inside the solvent. The question is: can one use the vacuum formula for
the electrostatic potential at the surface of the sphere, i.e., Eq. (2.S)?

The answer is no, because this formula was obtained from the expression
for the electric force between two charges in a vacuum and it is known that
the electric force between two charges depends on the medium between
them. The electric force in the presence of a material medium is less than
that which operates when only a vacuum is present. A simple explanation
of this phenomenon is given later on (cf Section 2.5). The ratio of the force
in vacuum to the force in the medium is a characteristic of the medium
and is known as its dielectric constant f (Fig. 2.10)

Electric force in vacuum

(2.12)
f = Electric force in medium

Hence, the coulombic force between two charges in a medium of dielectric

constant f is
(2.13)
56 CHAPTER 2

Medium Vacuum

0-----0 0-------8

Electric force Is £ times Electric force

q. q- q. q-
in medium =---;;.- f - less than !-------
in vacuum =---;:.-

Fig. 2.10. The electric force between two charges q + and q _ in vacuum and in
a medium.

and the electric field becomes

(2.14)

Hence, the potential at the surface of the sphere of radius ri placed in a

medium of dielectric constant E is

(2.15)

In terms of this expression for the electrostatic potential at the surface

of a charged sphere situated in a medium of dielectric constant E, the elec-
trostatic work of charging a sphere becomes

(2.16)

2.2.5. The Born Expression for the Free Energy of lon-Solvent

Interactions

Now that the basic electrostatics of charging and discharging spheres

has been presented, it can be immediately applied to the model suggested
by Born for the calculation of the free energy of ion-solvent interactions.
It has been argued in Section 2.2.2 (see also Fig. 2.8) that the free
energy of ion-solvent interactions, LI G1 - S , is given by the sum of the work
done to discharge an ion in vacuum and the work done to charge it up in
 ION-SOLVENT INTERACTIONS 57

the solvent of dielectric constant Bs' Since, according to the Born model,
a sphere (of radius ri and charge ZieO) is considered to be equivalent to an
ion of radius ri and charge Zieo, it follows that the work of discharging an
ion in vacuum is equal to the work of discharging the equivalent sphere
in vacuum and the work of charging an ion in the solvent is equal to the
work of charging the equivalent sphere in the solvent. Hence [ef Eq. (2.1 )],

L1G 1- S = Work of discharging equivalent sphere in vacuum

+ Work of charging equivalent sphere in solvent
(ZieO)2 (ZieO)2
=--
2rj
- +2B.r;
-- per ion

= _ (Zi eO)2 (1 __1) per ion

2ri Bs

per mole of ions (2.17)

where NA is the Avogadro number (Fig. 2.11).

Thus, by considering that the free energy of ion-solvent interactions
is given by the net electrostatic work of discharging a sphere (of the same
size and charge as the ion). in a vacuum, of transferring the discharged

Ion Charged
sphere

Free energy of
ion - salven'
interactions

is equivalent to

Continuum
(dielectric
constant,t"sl

Structured
solvent

Fig. 2.11. The free-energy change resulting from the transfer of an ion from
a vacuum into the solvent.
58 CHAPTER 2

120r-------------,...--,
,
011
"0
~o 80
u
.x.

0.8

Fig. 2.12. The free energy of ion-solvent

interactions as a function of the reciprocal
of the ionic radius.

sphere into a medium with the same dielectric constant as the solvent,
and of then charging the sphere till it has the same charge as the ion, the
Born model has yielded the free-energy change resulting from the transfer
of ions from a vacuum to solvent.
What is the importance of this free-energy change? The importance
derives from the fact that systems in nature try to attain a state of minimum
free energy. Thus, if the L'1G1 - S is negative, then ions exist more stably
in the solvent than in vacuum. Since the dielectric constant of any medium
is greater than unity, I > 1/10., and, therefore, L'1G1 - S is always negative;
hence, the Born equation (2.17) shows that all ions would rather be involved
in ion-solvent interactions than be left in vacuum. The Born equation
predicts that the smaller the ion (smaller rJ and the larger the dielectric
constant lOs, the greater will be the magnitude of the free-energy change
in the negative direction (Fig. 2.12).
If one stands back and looks at the situation with regard to ions and
their existence in solvents before and after the theory of Born (1920),
several points emerge. One set out to discover the interactions of ions with
a solvent, and one ended up doing a problem in electrostatics. This illustrates
a feature of electrochemistry-it often involves the application of electro-
statics to chemistry. The basis of this link is of course that electrochemistry
is involved with ions and charged interfaces, and these can be most simply
represented in models by charged spheres and charged plates, the stuff
with which electrostatics deals.
One has also seen in the Born theory of ion-solvent interactions an
example of very simple thinking based on models. A complicated situation
has been reduced to a simple one by the choice of a simple model. In the
 ION-SOLVENT INTERACTIONS 59

case of ion-solvent interactions, once the analogies between an ion and a

charged sphere and between a structured solved and a dielectric continuum
are stressed, the rest is easy.
It will be shown later that, not unexpectedly, the Born model over-
simplifies the problem, but one must see the model in its historical perspec-
tive. It was proposed at a time when the very existence of charged particles
in solution was questioned. Indeed, the Born approach to ion-solvent
interactions and the fact that it gave answers of the same order of magnitude
as experiment (cf Section 2.2.9) helped to confirm the hypothesis that
ions exist in solution. Seen in historical perspective, the simple Born model
may be recognized as an important step forward.

2.2.6. The Enthalpy and Entropy of lon-Solvent Interactions

Before finding out about the experimental testing of the Born theory,
it is preferable to recover from the theoretical expression for the free energy
LlG1 - S , the enthalpy (heat) and entropy changes associated with ion-
solvent interactions. This is because it is the heat of ion-solvent interac-
tions, rather than the free energy, which is obtained directly from the ex-
perimentally measured heat changes observed to occur when solids con-
taining ions are dissolved in a solvent, i.e., when ion-solvent interactions
are provoked.
By making use of the combined first and second laws (dE = T dS
- p dV) in G = H - TS = E + PV - TS, one gets

dG = VdP - SdT (2.18)

and, at constant pressure,

(!£)
aT p
=-S (2.19)

Thus, applying (2.19) to a transformation from state I to state 2

results in

(2.19a)

Hence
aLlG = _ LIS (2.19b)
aT
60 CHAPTER 2

Hence, all one has to do to get the entropy changes associated with
ion-solvent interactions is to differentiate L1G1 - S [given by Eq. (2.17)]
with respect to temperature. During this differentiation, the question arises
whether the dielectric constant should be treated as a constant or as a
variable with temperature. At this stage of the presentation, one does not
have a feel for dielectric constants to be able to answer the question (see,
however, Section 2.5); so one has to appeal to experiment. It turns out
that the dielectric constant does vary with temperature (Table 2.2) and
must therefore be treated as a variable in differentiating Eq. (2.17) with
respect to temperature.
Thus, the entropy change due to ion-solvent interactions is

(2.20)

and from
(2.21)

one has for the heat change:

(2.22)

Now that one has a theoretical expression for a heat change, it is time
to think of comparing the predictions of the Born theory with experiment.
There are, however, a few conceptual questions first to be considered.

TABLE 2.2
Variation of Dielectric Constant of Water with Temperature

Temperature, Temperature,
DC Dielectric constant fW
DC Dielectric constant fW

0 87.74 50 69.91
10 83.83 60 66.81
20 80.10 70 63.85
25 78.30 80 61.02
30 76.54 90 58.31
40 73.15 100 55.72
 ION-SOLVENT INTERACTIONS 61

2.2.7. Can One Experimentally Study the Interactions of a Single

Ionic Species with the Solvent?

The Born theory has indicated an elementary, modelistic, and over-

simplified way of calculating the free energy, entropy, and enthalpy changes
associated with the interaction of a single ionic species with the solvent.
But how can these quantities be experimentally measured so that one can
check on the degree of correctness of the Born model?
It is in connection with the actual process of measuring the heat of the
ion-solvent interactions that there arises a difficulty which is best brought out
by the following thought experiment. Suppose one imagines that a beam
of anyone species of ions (Fig. 2.13), say K+ ions, is introduced into a
solvent. What will be the result? The solvent will become positively charged.
So the work of introducing an ion into a charged solution will consist not
only of the work arising from ion-solvent interactions but also of the
coulombic work of bringing an ion into a charged medium. One has to
conclude that, even if one could experimentally measure the heat change
in the imaginary process of shooting a beam of one ionic species into a
solvent, it would not correspond to the calculated heat of ion-solvent
interactions [cf Eq. (2.22)] because the experimental value would include
the extra energy arising from the interactions of the charged ions with
the charged solution, a serious complicating factor.
In contrast to the thought experiment referred to in the preceding
paragraph the conventional method of introducing ions into a solvent is
to dissolve a salt which must contain both positive and negative ions.
Thus, one can dissolve KCI to introduce K+ ions into the solvent. Since
one introduces a negatively charged Cl- ion along with every positively

4 Ions of one sign

@:---being introduced Solution charged
@ into solvent positive

<t)+
<t> <t> ® +
(t) <t> <t> ® <±>
Neutral solvent

Fig. 2.13. The result of introducing ions of one sign

into a neutral solvent is a charged solution.
62 CHAPTER 2

charged K+ ion, the solution never gets charged; it maintains its electro-
neutrality.
At the same time, however, one has generated another problem. Two
ionic species have been introduced into the solvent, and each one is going
to have its own work of interactions so what is experimentally measured is
actually the sum of the contributions of the two ionic species to ion-solvent
interactions. How, then, can the sum of the heats of ion-solvent interactions
of the positive and negative ions, i.e., the heat of salt-solvent interactions,
be separated into the individual contributions of each ionic species?
This matter of attempting to extract from an experimental measure-
ment on a salt (which consists of both positive and negative ions) the value
of the property concerned for one of the ions constituting the salt is a
frequent one in electrochemistry. Thus, for the kind of reason shown up
by the thought experiment described above and pictured in Fig. 2.13, only
the combined effects of the contributions of positive and negative ions can be
experimentally measured, the separation of the effects into the contributions
of the individual ionic species being an experimentally insoluble problem
(c/, for example, the question of individual ionic activity coefficients, Sec-
tion 3.4).
A safe but rather timid way out of the situation is to remain content
with determining only the effects of electrolytes (salts). One could even
seek to justify this attitude by saying that one should only think about
things which can be measured without ambiguity. A corollary of this
viewpoint is that experimental heats of ion-solvent interactions of individual
ionic species are neither meaningful nor significant.
There is, however, entirely objective evidence, free from any taint of
the inaccuracies of a given model, for individual heats of ion-solvent
interactions. One can tabulate heats of salt-solvent interactions (their ex-
perimental determination is described in Section 2.2.8) for pairs of salts
with a common ion, e.g., KCI and NaCl, KBr and NaBr, and KI and Nal.
It is seen (Table 2.3) that there are almost constant differences in the heats
of interactions of these different salts with water. This is just what one
would expect if each of the ionic species was making a characteristic con-
tribution to the heats of interaction of the salt with the solvent. Then,
because, in each pair of salts in Table 2.3, the differing ions are lithium
and sodium or sodium and potassium, respectively, the constant difference
would be the difference of the heats of interaction of these individual ionic
species with the solvent. Thus, it is meaningful to try to obtain experimental
values of the heats of solvation of the individual ionic species.
There are several approaches which could be made to the problem
 ION-SOLVENT INTERACTIONS 63

TABLE 2.3
Constant Differences in the Heats of Solvation of Pairs of Salts with a
Common Ion

LlHs -H 2 O,
Salt Difference
kcal moie-1

LiF -245.2
-27.4
NaF -217.8
LiCI -211.2
-27.4
NaCl -183.8
LiBr -204.7
-27.4
NaBr -177.3
LiI -194.9
-27.4
NaI -167.5

NaCl -183.8
-20.0
KCI -163.8
NaBr -177.3
-20.0
KBr -157.3
NaI -167.5
-20.0
KI -147.5

of obtaining heats of interaction of individual ions with the solvent from

the experimental measurements (ef Section 2.2.8) of heats of salt-solvent
interactions. A crude one is based on the realization that, according to the
Born model, the heat of ion-solvent interactions is inversely proportional
to the ionic radius [ef Eq. (2.22)]. Hence, if one considers a salt in which
the positive and negative ions are of equal radii, then the heat of the salt-
solvent interactions can be split equally between the two species. The usual
pair chosen for this purpose is KF because the radii of the K + and F-
ions are almost equal (Table 2.4). Of course, once the heat of ion-solvent
interactions is obtained for any single ionic species, then, one can quickly
get the heats for all other ionic species. For example, if the heat of ion-
solvent interactions is known for K + (because of the above procedure
applied to KF), then the heat for Cl- can be obtained from the measured
heat for KCI because the heat of interaction of KCl with the solvent is
the sum of that due to the K + and Cl- ions.
64 CHAPTER 2

TABLE 2.4
Some Ionic Radii

Ion Ionic radius, A Ion Ionic radius, A

Li+ 0.60 Fe++ 0.76

Na+ 0.95 Co++ 0.74
K+ 1.33 Ni++ 0.72
Rb+ 1.48 Cu++ 0.72
Cs+ 1.69 Al3+ 0.50
Ag+ 1.26 Sc3+ 0.81
Be++ 0.31 La3+ 1.15
Mg++ 0.61 Fe3+ 0.64
Ca++ 0.99 F- 1.36
Sr++ 1.13 Cl- 1.81
Ba++ 1.35 Be 1.95
Zn++ 0.74 1- 2.16
Mn++ 0.80

The approach just described for the separation of the heat of salt-
solvent interactions into its component heats of ion-solvent interactions
is a very simple one. More complicated methods have been suggested.
But almost all of them have one important defect in common: they start
with ideas based on a specific model (e.g., the idea following from Born's
simple view that ions of the same radius will have the same heat of inter-
action with the solvent) and then split the experimental heat of salt-solvent
interactions into heats of interaction of individual ions with the solvent.
Hence, one generally has only so-called "experimental" values of the heat
of ion-solvent interactions. The heats of interaction with the solvent for
single ionic species can never be as "clean" and unambiguous as those for
electrolytes (salts); the latter are given by pure experiment, untainted by
any (always approximate) model-oriented speculations.

2.2.8. The Experimental Evaluation of the Heat of Interaction

of a Salt and Solvent

It remains therefore to describe how the heat of salt-solvent interaction

is experimentally measured. If one takes a solid salt and dissolves it in a
solvent, it is only in exceptional cases that marked temperature changes,
i.e., thermal effects, are noticed (Table 2.5). In general, the dissolution
process is accompanied by only a slight heating or cooling (usually < 10 kcal
mole-I).
 ION-SOLVENT INTERACTIONS 65

TABLE 2.5
He~1; of Solution I1Hso ln • the Heat Absorbed During Dissolution of Some
Ionic Crystals at 25 0 C
kcal mole- 1

Anions
Cations
F- Cl-

Li+ +1.1
Na+ +0.1 +0.9 -0.2
K+ -4.2 +4.1 +4.8 +4.9
Rb+ -6.3 +5.2 +6.2
Cs+ -9.0 +7.9

These heat changes are very small compared with the heat of ion-
solvent interactions (of an order of magnitude of 100 kcal mole- 1 for singly-
charged ions) calculated by the Born model. What is the reason for the
big difference? The answer is not difficult to see. The Born model calculates
the heat change when isolated ions are transferred from vacuum into the
solvent, in contrast to the heat of solution which is the measure of the
heat change when ions from an ionic crystal are transferred into the solvent.
But, in the crystal lattice, the ions are engaged in interactions with each
other. This means that two processes are occurring simultaneously when a
crystal is dissolved. Firstly, the lattice is being dismantled and the ions
separated from each other, and, secondly, the ions are entering into inter-
actions with the solvent. It is only this second part of the dissolution process
in which ion-solvent interactions playa part.
To get at this ion-solvent part of the process, one must know the
ion-ion interactions in the crystal lattice. In other words, one must know
how much energy is required to dismantle the lattice and take the ions so
far from each other that they do not interact any more. Then, the difference
between the heat change associated with breaking up the lattice and produc-
ing a gas of ions extremely far from each other and that associated with the
dissolution process is equal to the heat of interaction of the salt with the
solvent.
To be specific, consider a potassium fluoride crystal. First, this ionic
lattice must be conceptually disassembled and a very dilute (to cut down
ion-ion interaction) gas of K + and F- ions produced. The heat-content
66 CHAPTER 2

Fig. 2.14. Thermodynamic cycle of hypo-

thetical changes to obtain the heat of salt-
solvent interactions.

change associated with this process is called the lattice energy L1 Hlattice .
Then, the ions, which are infinitely far apart, are introduced into the solvent,
in which process the heat-content change is the heat of salt-solvent inter-
actions, L1Hs - s . Finally, the ions in the solvent are assembled back into
the crystal; this is the opposite of the process of dissolution and is therefore
associated with minus the heat of solution, - L1Hsoln ' The crystal is back
where it started (Fig. 2.14), a thermodynamic cycle has been completed,
and the various heat changes must all algebraically add up to zero, i.e.,

L1Hlattice + L1Hs-s - L1Hsoln = 0 (2.23)

or
L1Hs-s = L1Hsoln - L1Hlattice (2.24)

Both the quantities on the right-hand side are experimentally accessible

(cf Table 2.6), and, hence, one can obtain experimental values of the heat
of salt-solvent interactions (Table 2.7).
If, now, the L1Hs - s for KF is taken and divided equally (cf Section
2.2.7) between K+ and F- ions because the radii of these ions are almost
equal, one gets the heat of interaction of these two individual ions with
the solvent. Once these values are obtained, one can construct a table of
so-called t "experimental" heats of ion-solvent interactions (Table 2.8).

t So-called because their evaluation has involved the postulate that ions which have
equal crystallographic radii also have equal interaction with the solvent.
 ION-SOLVENT INTERACTIONS 67

TABLE 2.6
Lattice Energy I1Hlattice of Ionic Crystals and Heat of Solution. t I1H so l n
kcal mole- 1 at 25°C

Salt LlHlattice LlHsoln

LiF +246.3 +1.1

NaF +217.9 +0.1
KF + 193.6 -4.2
RbF +186.4 -6.3
CsF +177.9 -0.9
NaCI +184.7 +0.9
KCI +167.9 +4.1
NaBr +177.1 -0.2
KBr + 162.1 +4.8
RbBr +157.4 +5.2
KI + 152.4 +4.9
RbI +148.6 +6.2
CsI + 144.5 +7.9

t These are the measured molar heats of solution extrapolated to infinite dilution to avoid
ion-ion interactions (c{. Chapter 3).

TABLE 2.7
Experimental Values of the Heats of Interaction between a Salt and Water
I1H8 _ 8
kcal mole- 1 at 25°C

F- Cl- Br- 1-

Li+ -245.2 -211.2 -204.7 -194.9

Na+ -217.8 -183.8 -177.3 -167.5
K+ -197.8 -163.8 -157.3 -147.5
Rb+ -192.7 -158.7 -152.2 -142.4
Cs+ -186.9 -152.9 -146.4 -136.6
68 CHAPTER 2

TABLE 2.8
The Experimental Heats 6.H1- HzO of Interaction between Individual Ions
and Water. Assuming 6.HK+-HzO = 6.HF --HzO = !6.HKF-HzO

Heat of ion-water interactions,

Ion
kcal mole-1 at 25°C

Li+ -146.3
Na+ -118.9
K+ - 98.9
Rb+ - 93.8
Cs+ - 88.0
F- - 98.9
C!- - 64.9
Br- - 58.4
1- - 48.6

2.2.9. How Good Is the Born Theory?

In making numerical calculations based on Eq. (2.22) so that a compar-

ison between the predictions of the Born theory and the results of experiment
can be made, the first question to decide is: What value of the radius of
the ions shall be substituted into the Born equation (2.22)? Since the exper-
imental values of the heat of ion-solvent interactions are based on dis-
mantling an ionic crystal and then plunging the ions so produced into the
solvent, the obvious radius to choose is the radius of ions obtained from
X-ray measurements on the corresponding ionic crystal, i.e., the crystallo-
graphic radius.
When the crystal radii are inserted into the Born expressions for the
free energy and enthalpy of ion-solvent interactions, the resulting values
turn out to be of the correct order (Table 2.9). An important conclusion
can be drawn from this order of magnitude comparison: Ion-solvent inter-
actions arise largely (at least for the ions concerned) from coulombic
forces for those are the only forces reckoned with in the Born model.
However, a detailed examination of Table 2.9 reveals that the Born
values for the heats of ion-solvent interactions are numerically too high,
in some cases nearly 50% too high.
It also turns out that the experimental heats of interaction between
ions and solvent do not vary inversely as the radius, as predicted by the
 ION-SOLVENT INTERACTIONS 69

TABLE 2.9
Ionic Radii and the Born Free Energy LlG f - H20 and Enthalpy LlHf - H20 of
lon-Water Interactions at 25 0 C

LlGf - H2 0,
Ion
A kcal mole- 1
Calc. from Eq. (2.22) Exp. from Table 2.8

Li+ 0.60 -273.2 -277.7 -146.3

Na+ 0.95 -172.6 -175.5 -118.9
K+ 1.33 -123.2 -125.3 - 98.9
Rb+ 1.48 -110.8 -113.1 - 93.8
Cs+ 1.69 - 97.0 - 98.6 - 88.0
F- 1.36 -120.5 -122.6 - 98.9
Cl- 1.81 - 90.6 - 92.1 - 64.9
Br- 1.95 84.1 85.5 - 58.4
1- 2.16 - 75.9 - 77.2 - 48.6

Born equation (2.22) (Fig. 2.15). Of course, one can arbitrarily adjust the
values of the radii to differ from the crystallographic radii and then obtain
better fit between theory and experiment. Indeed, it was found some years
ago by Latimer, Pitzer, and Slansky that, by adding 0.85 A to the radii of
the positive ions and 0.1 A to those of the negative ions, one can "remove"
the discrepancy between the calculated and observed values (Fig. 2.16 and

270

'Q'0I 210
E
~ 150
~

o
/90
:X::.:.
~ 30
2.2

Fig. 2.15. Experimental heats of ion-water

interactions do not vary inversely as the
ionic radius.
70 CHAPTER 2

~ ...~40
'\I
I

0.6 0.8
Reciprocal of ''corrected'' ionic radius
UNA)
Fig. 2.16. By adding 0.85 and 0.10 A
to the crystallographic radii of positive and
negative ions, respectively, the calculated
Born free energies of ion-water interactions
vary inversely with the corrected tonic radii.

Table 2.10). But then one has to give a theory of where the magic numbers
0.85 A and 0.1 A come from. Pending the proposal of some such theory,
one is back at the starting point, i.e., the Born theory suggests that ion-
solvent interactions are much stronger than experiment shows them to be
(Table 2.9).

TABLE 2.10

Calculated Born Values of llHI-H.O after Adding 0.85 to the Radii of the A
Positive Ions and 0.10 A
to Those of Negative Ions

LlHI-H20, kcal mole-1

Ion Corrected radius, A
Calc. with corrected radii Exp. from Table 2.8

Li+ 1.45 -lI5.8 -146.3

Na+ 1.80 - 92.6 -lI8.9
K+ 2.18 - 98.5 - 98.9
Rb+ 2.33 - 90.1 - 93.8
Cs+ 2.54 - 65.6 - 88.0
F- 1.46 -lI4.1 - 98.9
Cl- 1.91 - 87.3 - 64.9
Br- 2.05 -81.3 - 58.4
1- 2.26 - 73.8 - 48.6
 ION-SOLVENT INTERACTIONS 71

Another approach at explaining the discrepancy between the Born

theory and experiment revealed in Fig. 2.15 centers around the value chosen
for the dielectric constant fs to be used in numerical calculations with Eq.
(2.22). The Born theory uses the experimentally measured value for the
bulk solvent, e.g., 80 for water. But this bulk value is what one would obtain
if one put the solvent (e.g., water) between the plates of a capacitor and
measured the reduction factor for the electric force compared with the
value in vacuum (Fig. 2.17). What one has found out in this measurement
is the average dielectric constant of the solvent, i.e., of the solvent taken
in bulk quantities. It is this bulk value fs which has been used in the Born
equations, but it is the effective value feff near the ion which can be consider-
ed of greater significance to the charging process (cf Section 2.2.4). Now,
if feff near the ion is not equal to the bulk value but is much less than it,
they, by using feff instead of fs in the Born equation (2.22), the dHr - s will
be pulled down in the right direction, i.e., toward the experimental value.
But at this stage of the present discussion of ion-solvent interactions,
there is no rationale for considering to what extent the values of dielectric
constants near an ion may be different from the bulk values. In fact, the
atomistic origins of dielectric constants have yet to be explored. The fs
which has appeared in the Born theory has been taken from experiment,
and, in this aspect, the theory has a somewhat nonstructural, thermodynamic
flavor.
To go further, one has to think of the structure of the solvent near
an ion. After all, how can one discuss ion-solvent interactions if, as pres-
cribed by the Born model, one turns a blind eye to the structure of the

Hh",. H
solvent?

Vacuum Solvent

-q

Charge ,+ q r. em I.cm

Electric force q" Electric force q'

in vacuum =7 with solvent =~s I '

M-- Dielectric constont

~OfbUlk solvent=~s

Fig. 2.17. The bulk dielectric constant of a solvent.

72 CHAPTER 2

If one knew the arrangement of the particles in the solvent and the
forces operating between them and the ion, then one could make at least
approximate calculations of the ion-solvent interaction on a particulate
basis rather than on the continuum basis used by Born. Thus, a new strategy
for understanding ion-solvent interactions must be mapped. At first, one
must understand the structure of the solvent in the bulk far away from the
ion; then, one must understand the structure near the ion. It is a mix of
the ion-solvent forces from both regions which determines the energy of
the ion-solvent interactions.

Further Reading

1. M. Born, Z. Physik, 1: 45 (1920).

2. W. M. Latimer, K. S. Pitzer, and C. M. Slansky, J. Chern. Phys., 7: 108 (1939).
3. E. J. W. Verwey, Rec. Trav. Chim., 60: 887 (1961); 61: 127 (1942).
4. B. E. Conway and J. O'M. Bockris, "Ionic Solvation" in: J. O'M. Bockris,
ed., Modern Aspects of Electrochemistry, No.1, Butterworth's Publications,
Inc., London, 1954.
5. K. J. Laidler and C. Pegis, Proc. Roy. Soc. (London), A241: 80 (1957).
6. R. H. Stokes, J. Am. Chern. Soc., 86: 979, 982, and 2337 (1964).
7. E. Glueckauf, Trans. Faraday Soc., 60: 572 (1964).
8. W. A. Millen and D. W. Watts, J. Am. Chern. Soc., 89: 6051 (1967).
9. K. Ross, Aust. J. Phys., 21: 597 (1968).

2.3. STRUCTURAL TREATMENT OF THE ION-SOLVENT IN-

TERACTIONS

2.3.1. The Structure of the Most Common Solvent, Water

In the first instance, one can examine the structure of water in its gaseous
form. Water vapor consists of separate water molecules. Each of these is a
bent molecule, the H-O-H angle being about 105° (Fig. 2.18). In the
gaseous oxygen atom, there are six electrons in the second shell (two 2s
electrons and four 2p electrons). When the oxygen atoms enter into bond

Fig. 2.18. A water molecule is nonlinear.

 ION-SOLVENT INTERACTIONS 73

formation with the hydrogen atoms, there is a blurring of the distinction

between the sand p electrons. The six electrons from oxygen and the two
from hydrogen interact, and it is found that four pairs of electrons tend
to distribute themselves so that they are most likely to be found in four
approximately equivalent directions in space. Since the motion of electrons
is described by quantum mechanics, according to which one cannot specify
precise orbits for the electrons, one talks of the regions where the electrons
are likely to be found as orbitals, or blurred orbits. The electron orbitals
in which the electron pairs are likely to be found are arranged approximately
along the directions joining the oxygen atom to the corners of a tetrahedron
(Fig. 2.19). The eight electrons around the oxygen are neither s nor p
electrons; they are Sp3 hybrids. Of the four electron orbitals, two are used
for the O-H bond, and the remaining two remain as free orbitals for the
lone pair of electrons. Because of the repulsion of the electron pairs, the
H-O-H angle is not exactly equal to the tetrahedral angle (109 0 28')
but is less than that.
The free orbitals in which are found the electron lone pairs confer an
interesting property (Fig. 2.20) on the water molecule. The center of "grav-
ity" of the negative charge in the water molecule does not coincide with
the center of gravity of the positive charge. In other words, there is a charge
separation within the electrically neutral water molecule; it can be considered
an electric dipole. The moment of a dipole is defined by the product of the
full electronic charges at either end times the distance between the centers
of electrical charge. The dipole moment of water is 1.87 X 10-18 esu in
the gas phase (but becomes larger when the water molecule is associated
wi th other water molecules).
The availability of the free orbitals (with lone electron pairs) on the
oxygen atom contributes not only to the dipolar character of the water
molecule but also to another interesting consequence. The two lone pairs
can be used for electrostatic bonding onto two other hydrogen atoms. This
is what happens in a crystal of ice. The oxygen atoms lie in layers with each
layer consisting of a network of open, puckered hexagonal rings (Fig. 2.21).
Each oxygen atom is tetrahedrally surrounded by four other oxygen atoms
(Fig. 2.22). In between any two oxygen atoms is a hydrogen atom which
provides a hydrogen bonding (Fig. 2.23). At any instant, the hydrogen
atoms are not situated exactly halfway between two oxygens. Each oxygen
has two hydrogen atoms near it (the two hydrogen atoms of the water
molecule) at an estimated distance of about 1.75 A. Such a network structure
of associated water molecules contains interstitial regions (between the
tetrahedra) which are larger than the dimensions of a water molecule (Fig.
74 CHAPTER 2

Fig. 2.19. The hybrid orbitals

of an oxygen atom.

Water molecule Electric dipole

Center of
H Negallve ./ p\ ..
oSltlve
"" ~ charge ..3harge
l..: __" is electrically equivalent to.
\If~

Center of oxygen nudeus qd=p =I.B7debyes

Fig. 2.20. A water molecule can be considered elec-

trically equivalent to a dipole.

~PUCkered.
: ~ : : : hexagonal
~~ , ring
~~
I
II I :

~
Fig. 2.21. The oxygen atoms in ice, which
are located at the intersections of the lines
in the diagram, lie in a network of open,
puckered hexagonal rings.

~
1 jCOOrdinoting
.',gens
Centrol oxygen
Fig. 2.22. Each oxygen atom in ice is tet-
rahedrally coordinated by four other oxygen
atoms. The hydrogen atoms are not shown
in the diagram.
 ION-SOLVENT INTERACTIONS 75

~):D
'96-I.02!
Hydrogen
bond
1.74-I.~A
Fig. 2.23. The hydrogen bond between
two oxygen atoms (the oxygen and hy-
drogen atoms are indicated by 0 and .,
respectively) .

2.24). Hence, a free nonassociated water molecule can enter the interstitial
regions with little disruption of the network structure.
Structural research, originating from a classic paper by Bernal and
Fowler, has shown that liquid water, under most conditions, is best described
as a somewhat broken-down, slightly expanded (Table 2.11) form of the ice
lattice, but this statement must not be taken to mean that there is no asso-
ciation of water molecules in water. X-ray and other techniques indicate
that, in water, there is a considerable degree of short-range order charac-
teristic of the tetrahedral bonding in ice. Thus, liquid water partly retains

Framework water molecule

Fig. 2.24. The structure of ice with interstitial spaces

large enough to accommodate a free, unassociated water
molecule (based on a diagram from General Chemistry by
Linus Pauling).
76 CHAPTER 2

TABLE 2.11
The Structure of Ice and Liquid Water

Ice Liquid water

Mean 0-0 distance 2.76 A 2.92 A

Number of oxygen nearest neighbors 4 4.4--4.6

the tetrahedral bonding and resulting network structure characteristic of

the crystalline structure of ice.
In addition to the water molecules which are part of the network,
there can be a certain fraction of structurally free, un associated water
molecules in interstitial regions of the network (Fig. 2.25). When a network
water molecule breaks its hydrogen bonds with the network, it can move
into interstitial regions as an interstitial water molecule which can rotate
freely. Thus, the classification of the water molecules into network water
and free (or interstitial) water is not a static one. It is dynamic. As argued
by Frank, clusters of water molecules are cooperating to form networks,
and, at the same time, the networks can break down. A water molecule
may be free in an interstitital position at one instant, and, in the next
instant, it may become held as a unit of the network. t
What happens to this picture of liquid water when ions enter it?

2.3.2. The Structure of Water near an Ion

The aim here is to take a structural microscopic view of the ion inside
the solvent. The central consideration is that ions orient dipoles. The spher-
ically symmetrical electric field of the ion may tear water dipoles out of
the water lattice and make them point (like compass needles toward a
magnetic pole) with the appropriate charged end toward the central ion.
Hence, viewing the ion as a point charge and the solvent molecules as

t Knowledge of the structure of water is likely to increase greatly in the near future.
There is need to regard the present view-as usual-as part of an evolving picture, one
which will become more complete with an increase of research. For example, there is
evidence which suggests that normal water, described here, contains a small quantity
of a second form of water, which has a much higher boiling point and lower freezing
point than normal water. The new form of water may be separated from normal
water by passage through some types of capillary tubes.
 ION-SOLVENT INTERACTIONS 77

Unassociated
water molecules

Fig. 2.25. Schematic diagram to show

that, in liquid water, there are networks
of associated water molecules and also a
certain fraction of free, unassociated water
molecules.

electric dipoles, one comes out with a picture of ion-dipole forces as the
principal basis of ion-solvent interactions.
Due to the operation of these ion-dipole forces, a certain number of
water molecules in the immediate vicinity of the ion (more about just how
many, later) may be trapped and oriented in the ionic field. Such water
molecules cease to associate with other water molecules to form the net-
works characteristic of water (cf Section 2.3.1). They are immobilized
except in so far as the ion moves, in which case the sheath of immobilized
water molecules moves with the ion. In other words, the ion and its water
sheath are a single kinetic entity. (More discussion of this is in Section 2.4.3).
Thus, the picture (Fig. 2.26) is of ions enveloped by a solvent sheath of
oriented, immobilized water molecules.
What about the situation far away from the ion? At a sufficient dis-
tance away from the ion, its influence is negligible because the ionic fields

Fig. 2.26. An ion enveloped by a

sheath of oriented solvent molecules.
78 CHAPTER 2

have become attenuated virtually to zero. The normal structure of water

is undisturbed; it is that of bulk water.
In the region between the solvent sheath (where the ionic influence
determines the water orientation) and the bulk water (where the ionic in-
fluence has ceased to affect the orientation of water molecules), the orienting
influences of the ion and the water network operate; the former tries to
align the water dipoles parallel to the spherically symmetrical ionic field,
and the water network tries to make the water in the in-between region
continue the tetrahedral arrangement (Fig. 2.27). Caught between the two
types of influences, the in-between water adopts a compromise structure
that is neither completely oriented nor disoriented. The compromising
water molecules are not close enough to the ion to become oriented perfect-

/
~/m)/
In - between region
/
/

/J
Possible orientation
to suit bulk water
/
/

In-between
water molecule

Fig. 2.27. Schematic diagram to indicate that, in the (hatched)

region between the primary solvated ion and bulk water, the in-
between water molecules must compromise between an orien-
tation which suits the ion (oxygen-facing ion) and an orien-
tation which suits the bulk water (hydrogen-facing ion).
 ION-SOLVENT INTERACTIONS 79

Iy around it, and neither are they sufficiently far away from the ion to be
part of the structure of bulk water; hence, depending on their distance
from the ion, they orient out of the water network to varying degrees.
In this intermediate region, the water structure is said to be partly broken
down.
One can summarize this description of the structure of water near an
ion by referring to three regions (Fig. 2.28). In the primary, or structure-
enhanced, region next to the ion, the water molecules are immobilized and
oriented by the ionic field; they move as and where the ion moves. Then,
there is a secondary, or structure-broken, region, in which the normal bulk
structure of water is broken down to varying degrees. The in-between water
molecules, however, do not partake of the translational motion of the ion.
Finally, at sufficient distance from the ion, the water structure is unaffected
by the ion and displays the tetrahedrally bonded networks characteristic
of bulk water.
The three regions just described differ in their degree of sharpness.
The primary region-to be discussed in greater detail below-in which
there are (at least for some ions) water molecules which share the transla-
tional motion of the ion, is a fairly sharply defined region. In contrast, the
secondary region, stretching from the termination of the primary region to

Primary region with

ompletel y oriented
woter

Secondory region with

port Iy - oriented woter

Bulk region with

unoriented water

Fig. 2.28. The neighborhood of an ion may be con-

sidered to consist of three regions with differing solvent
structures: (1) the primary or structure-forming region,
(2) the secondary or structure-breaking region, and (3)
the bulk region.
80 CHAPTER 2

the resumption of the normal bulk structure, cannot be sharply defined;

the bulk properties and structure are asymptotically approached.
These structural changes in the primary and secondary regions are
generally referred to as solvation (or as hydration when water is the solvent).
Since they result from interactions between the ion and the surrounding
solvent, one often uses the terms solvation and ion-solvent interactions
synonymously; the former is the structural result of the latter.

2.3.3. The lon-Dipole Model of lon-Solvent Interactions

The above description of the solvent surrounding an ion can now be

used as the basis of a structural treatment of ion-solvent interactions,
initiated by Bernal and Fowler (1933).
Consider an isolated ion in the gas phase above the solvent. The total
work done to transfer this ion from a very dilute gas of ions to the inside
of the solvent defines the free energy of solvation (cf Section 2.2.1), i.e.,
the free-energy change arising from ion-solvent interactions, il GI -8' This
free-energy change is composed of both enthalpy changes and entropy
changes. The latter arise from changes in the degrees of freedom (transla-
tional, rotational, and vibrational) experienced by the water molecules
as they come out of the water structure and associate with an ion. In this
simplified treatment, only the enthalpy changes will be treated.

Ion Solvent molecule not

(±) n - Solvent used to form primary

i
Remove (n + I )solvent
F"m ~,::,":'~O
~V
solvent sheath
~ Condensation

.0 'oto "h"'col cO't'l

molecules from spherical Transfer primary solvateb
volume

-Solvent- - -- +
o
(a) (h)
Fig. 2.29. A thought experiment to separate out various aspects of ion-solvent
interactions.
 ION-SOLVENT INTERACTIONS 81

Vacuum

Work done to remove ~

(JJ+ I) solvent
molecules =WCF ~+I) solvent
molecules

Cluster of (n+1)
solvent molecules

Fig. 2.30. The formation of a cavity in the solvent by the removal of

n + 1 solvent molecules.

The ion-solvent interactions consist of several contributions. There is,

for example, the interaction between the ion and the n nearest neighbors
which surround the ion and make up the primary solvent sheath (see Fig.
2.28). Then, there is the energy used up for the structure breaking in the
secondary region (Fig. 2.28), etc.
To separate out the various aspects of the total interaction, one can
consider a thought experiment (Fig. 2.29) proposed by Eley and Evans
in which the proper number of solvent dipoles are taken from the solvent
to the gas phase and there oriented around the ion (by ion-dipole forces).
Finally, the primary solvated ion is transferred into the solvent, upon
which structure breaking, etc., occurs.
The steps of this thought experiment will now be described more
elaborately.
1. One starts the thought experiment with the knowledge gained from
various types of experiment that the primary solvated ion will occupy a
volume corresponding to the volume of n primary solvent molecules plus
one more to make room for the bare ion. t This volume corresponding to

t Note that, as a first approximation, it is assumed that the volume of a water molecule
is the same as that of a bare ion. For some ions, this is a reasonable approximation.
Thus, the radius of a water molecule is 1.38 A and that of K+ is 1.33 A.
82 CHAPTER 2

Vacuum ~

~ ~~,~~ :?. ~~
~
Cluster
G
Fig. 2.31. Dissociation of a cluster of n + 1
molecules by breaking the bonds holding them
together.

n + I solvent molecules must be made available in the solvent for immers-

ing a primary solvated ion. Hence, n + I molecules will be removed from
the solvent and taken into the vacuum phase (Fig. 2.30). Thus, the cavity
which is left in the solvent will be large enough to accommodate an ion plus
n molecules in its primary solvation sheath. Let this work of cavity formation
be represented by WCF '
2. Before the n + I solvent molecules just removed from the solvent
can orient around the ion in the gas phase, they must be detached from
the cluster of n + I molecules and made free to orient around the ion.
To make this feasible, the bonds holding together these n + I solvent
dipoles in the cluster are broken asunder (Fig. 2.31), i.e., the group is
dissociated in the gas phase into n + I separate molecules. This dissociation
will involve an amount of work represented by W D'
3. Next, ion-dipole bonds are forged between the ion and n out of
the n + I solvent dipoles, and, thus, the primary solvent sheath is formed.
The work of interaction between an ion and a dipole (of moment fls

@;"
and radius rs) for the configuration shown in Fig. 2.32 is approximately

C?.-L
Dipole

Negative ion

Fig. 2.32. The minimum interaction-energy

orientation of a dipole to an ion.
 ION-SOLVENT INTERACTIONS 83

Ion-e Vacuum

~
Free solvent
o
f-:\Work ot
ion-dipole
Interaction =
dipole
~ ~-D Primary

Q solvated
ion

Fig. 2.33. Formation of a primary solvated ion.

given by (cf Appendix 2.2)t

Zieof.l.
(ri + r.)2
But, it is n solvent molecules that are involved in the primary solvent sheath.
Hence, per mole of ions, the ion-dipole interaction work (Fig. 2.33) is t

(2.25)

4. Now, the ion together with its primary solvent sheath is transferred
from vacuum into the cavity in the solvent (Fig. 2.34). What work is in-
volved in this transfer? A simple way to look at it is to imagine that the
solvated ion in the gas phase is discharged and then, still preserving its
solvent sheath, is sneaked into the cavity formed in step 1 of the thought
experiment (cf Fig. 2.30), whereafter the discharged but still solvated ion
is charged up to its normal value Zjeo . What has been described is simply
a Born charging process (cf Section 2.2.5). There is, however, an important
difference between the Born charging done here and that previously describ-
ed (Fig. 2.35). It is not a bare ion but a primary solvated ion which under-
goes the charging process. Hence, the radius to be used in the Born ex-
pression (2.22) is no longer the crystallographic radius rj but the radius
of a solvated ion, i.e., ri + 2r•.
Since it has been decided to deal only with enthalpies (or heat-content

t Note that the dielectric constant does not appear in this expression because there is
only vacuum between the dipole (i.e., the water molecule) and the (adjacent) ion.
t Note that the ion-dipole work always contributes a negative quantity to the heat of
solvation, independently of the sign of Zi, because the dipole always orients so that
that pole is in contact with the ion which makes the interaction attractive.
84 CHAPTER 2

Primary solvated ion

Vocuum

Work of transferring
primary solvated ion

c:'\ from vacuum to cavity

o
in solvent =K6~
Solvent

Fig. 2.34. Transfer of a primary solvated

ion from vacuum into a cavity in the solvent.

Bore
Pr imary
solvated
ion Vacuum Vacuum

8
Solvent

Solvent

(a) (b)

Fig. 2.35. The difference between the Born charging process in (a)
the ion-dipole model of solvation in which a primary solvated ion of ra-
dius 'j + 2rs is transferred into the solvent (Section 2.3.3) and in (b)
the nonstructural model of Born (ct. Section 2.2.5) in which a bare ion
of radius rj is involved.
 ION-SOLVENT INTERACTIONS 85

changes), one can set the work of transferring a solvated ion from vacuum
into a cavity in the solvent equal to the Born heat of solvation. This con-
tribution to the total heat of ion-solvent interactions shall be called the
Born charging contribution, W Be . Thus, per mole of ions,

(2.26)

Is it reasonable to use here an equation based on the Born model

even though one motivation for this structural treatment of solvation is
to get away from the Born nonstructural approach? The justification is
as follows. The radius ri + 2rs has been precisely defined (its ambiguity
was a problem in the Born model), the water outside the cavity is, at this
stage of the thought experiment, normal and undisturbed, and, therefore,
its dielectric constant (another ambiguity of the Born model) should be
that of bulk water. Thus, by considering the process of ion-solvent inter-
actions as occurring in steps with corresponding heat-content changes,
one of the steps, namely, the introduction of a primary solvated ion
into an undisturbed solvent, has been made to resemble a Born charging
process.
5. Once the cavity is filled up with the solvated ion and the Born
charging is carried out, one must ask whether the solvated ion leaves the
surrounding water undisturbed. It does not (Figs. 2.28 and 2.36). The
introduction of the primary solvated ion into the cavity does lead to some
Work of structure breaking
in secondary region =~B

J..TSI-_ _ Primary
solvated
ion

Region of
"structure
Solvent
breaking"

Fig. 2.36. The introduction of a primary solvated ion

into the cavity causes disturbance to the structure of the
solvent in the immediate vicinity of the solvated ion.
 86 CHAPTER 2

Solvent molecule not used

to form primary solvent
sheath

Work of condensation = UG

Solvent

Fig. 2.37. The condensation of a water

molecule left behind in vacuum because it
was not used to form the primary solvent
sheath.

disturbance of the structure of the surrounding solvent. In fact, this is the

structure breaking that has been referred to in dealing with the secondary
region, between the primary solvent sheath and the bulk water far away
from the ion (Fig. 2.28). Let this work of structure breaking be repre-
sented by WSB '
6. One must check up on the cycle now. Have all the solvent molecules
(taken out of the solvent into vacuum to create the cavity) been returned
to the solvent? Of the n + 1 solvent molecules removed, only n have return-
ed in the company of the ion as members of its solvation sheath. The one
water molecule which did not become part of the solvation sheath of the
ion and which has been left behind in vacuum, has to be returned to the
solvent to complete the cycle (Fig. 2.37). The work involved in this process
is equal to the work of condensation, We.
Now, all the solvent molecules which were removed from the solvent
in the thought experiment have returned to the solvent. In addition, the
ion which was in vacuum at the beginning of the thought experiment has
been transferred into the solvent. Hence, any work resulting from plunging
the ion into the solvent must result purely from ion-solvent interactions.
This work (or heat)t of solvation or ion-solvent interactions is therefore

t Note that, as already stated, there is an approximation being made here: It is the
free-energy change which is exactly equal to the work done (Appendix 2.1). One has
 ION-SOLVENT INTERACTIONS 87

r---------------~+

Fig. 2.38. How the total heat D.HI_s of ion-solvent interactions

has been separated in a thought experiment into the various steps
of cavity formation WCF' cluster dissociation Wo, formation of
primary solvated ion WI_O, Born charging WBC' structure break·
ing Ws B , and condensation, We.

given by (Fig. 2.38) the sum of all the pieces of work performed in each
step, i.e.,
iJH1 - s = WCF+ W D + W1 - D + WBC + WSB + We (2.27)

= W + W1 - D + W BC (2.28)
where
W = WCF + W D + WSB + We (2.29)

neglected T LIS, where LIS is the change of entropy during the solvation process. Struc·
tural theories of the entropy of hydration are known but will not be discussed here.
The error introduced by the approximation is about 10%.
88 CHAPTER 2

where the Avogadro number has been introduced to get the heat of solva-
tion per mole of ions.

2.3.4. Evaluation of the Terms in the lon-Dipole Approach to

the Heat of Solvation

The Born term, i.e., the last term in Eq. (2.30), can be easily calculated.
One uses the crystallographic radius ri of the ion, the radius r s of the solvent
molecules, and the bulk dielectric constant Es of the solvent. The ion-
dipole term, i.e., the second term in the expression (2.30) for the heat of
solvation, can also be calculated without difficulty provided one knows-or
estimates-the number n of solvent molecules which coordinate (or are
nearest neighbors to) the ion.
The first term in Eq. (2.30), however, is more awkward. It will be
recalled [Eq. (2.29)] that it consists of WCF , the work of forming a cavity
in the solvent by the removal into the gas phase of a cluster of n + I solvent
molecules; W D, the work of splitting up the cluster and separating to in-
finity the n + I solvent molecules; W::;u, the work of altering the orientation
of the solvent molecules in the solvent around the primary solvated ions;
and We, the work of condensing the one solvent molecule (from the cluster)
which is not used in the solvation of the ion.
The work W D of breaking the cluster and separating the n + 1 solvent
molecules can be considered either as the work of separating dipoles, i.e.,
the work arising from dipole-dipole forces or, in the case of hydrogen-
bonded liquids such as water, the work of breaking hydrogen bonds (Fig.
2.39). Since about 5 kcal mole-1 is required to break hydrogen bonds, the
value of W D depends on the value of n in the cluster of n + I solvent mole-

Fig. 2.39. Four hydrogen bonds (which

are numbered) must be broken to separate
the cluster of 4 + 1 = 5 water molecules.
 ION-SOLVENT INTERACTIONS 89

cules which are removed from the solvent to make room for an ion and its
n nearest neighbors. If, for example, an ion surrounds itself with four water
molecules in a tetrahedral configuration, then the cluster consists of
4 + I = 5 water molecules, and four hydrogen bonds must be broken per
cluster to separate the water molecules. Since I mole of cluster must be
removed from the solvent for the solvation of I mole of ions, it is necessary
to break 4 moles of hydrogen bonds per mole of ions. This requires
4 x 5 = 20 kcal mole-1.
The work We of condensing one solvent molecule per ion, or I mole
of solvent molecules per mole of ions, can be taken from the experimental
latent heat of condensation (Fig. 2.40); it is about -10 kcal mole-1•
The cavity formation work WCF and the structure-breaking work W SB
can be only roughly calculated. When the n + I water molecules are remov-
ed to form the cavity, a certain number of hydrogen bonds linking these
molecules to those outside the cavity are broken (Fig. 2.41). When the
primary solvated ion is introduced into the cavity, some of the solvent
molecules surrounding the solvated ion have to reorient. This reorientation
leads to the breakage of some hydrogen bonds and the formation of others.
Thus, if one considers the combined steps of cavity formation and structure
breaking, a certain net number of hydrogen bonds will be broken. Once
this number is known, one can easily get WCF + W SB by multiplying the
net number of hydrogen bonds broken by 5 kcal mole-1 .
A simple way of getting this number is to look at the water structure
before and after the solvated ion is introduced into the cavity. A careful
study of Fig. 2.42 shows that, whereas 12 hydrogen bonds are broken in
the cavity formation step involving the removal of 4 + I = 5 water mole-
<
Work of returnirYJ a
solvent molecule = Latent heat of condensation
per molecule

Solvent

Fig. 2.40. The work of condensing a water molecule is

equal to the latent heat of condensation per molecule.
 90 CHAPTER 2

Tetrahedral
cluster

Fig. 2.41. A total of 12 hydrogen bonds are broken when a tetrahedral clus-
ter of water molecules is removed from the solvent to form the cavity (num-
bers represent broken hydrogen bonds).

II

II- -n.....

Solvated ion
Cluster with its broken hydrogen bonds
(a)

Fig. 2.42. Schematic diagram to show that, out of four coordinating water mol-
ecules [I, II, III, and IV in (a)] in a tetrahedral cluster removed from the cavity, two
water molecules [I and II in (b)] reorient in the formation of a primary solvated positive
ion, and, therefore, only 10 H bonds [see (a)] are remade when the solvated positive
ion is introduced into the cavity.
 ION-SOLVENT INTERACTIONS 91

cules, only 10 hydrogen bonds are remade when the primary solvated
positive ion is introduced into the cavity. That is, a net number of 2 hy-
drogen bonds are broken per ion in the combined process of cavity formation
and structure breaking. The corresponding heat change W CF + W SB is
2 X 5 = 10 kcal mole- I of ions.
It is now possible to write down for a tetrahedrally coordinated positive
ion an approximate value for the work term [cf Eq. (2.29)] W = W D
+ We + WCF + W SB ' Using the arguments just presented, i.e., W D = 20
kcal mole-I, We = -10 kcal mole-I, and WCF + W SB = 10 kcal mole-I,
one has for four-coordinated positive ions

W = W D + We + W CF + W SB
= 20 - 10 + 10
= 20 kcal mole- I (2.31 )

Now consider negative ions. If, once again, tetrahedral coordination

is considered, then W D continues to be the work required to break up a
cluster of five water molecules, i.e., it is 20 kcal mole-I. The latent heat of
condensation of a water molecule obviously remains the same (-10 kcal
mole-I) for positive and negative ions. But a perusal of Fig. 2.43 shows
that the negative ion differs from the positive ion in that more water mole-
cules have to reorient when the primary solvated ion is introduced into the
cavity. In other words, the orientation of water molecules around a primary
solvated ion is less compatible with the water molecules in the primary
solvation shell of negative ions than with those of positive ions. Thus,
of the 12 hydrogen bonds broken in forming the cavity, only 8 are remade
when the cavity is filled up with a solvated ion (see Fig. 2.43). That is, the
net number of hydrogen bonds broken in the combined process of cavity
formation and structure breaking is four in the case of tetrahedrally
coordinated negative ions; and the corresponding work WCF + W SB is
4 x 5 = 20 kcal mole-I. Consequently, the work W [cf Eq. (2.29)] for
four coordinated negative ions is given by

W = W D + We + W CF + W SB
=20-10+20
= 30 kcal mole-I (2.32)

Now that the work W has been evaluated, it can be introduced into
Eq. (2.30) for the heat of hydration. Thus, for four coordination, one has
92 CHAPTER 2

~_-->...---I

_,,+--_ m II:

Cluster with its broken H-bonds Solvated negative ion

(a) (h)

(e) "'-
Solvated negative ion

Fig. 2.43. Schematic diagram similar to Fig. 2.42 except that a negative ion is
being considered here. Thus, two water molecules [III and IV in Fig. 2.42 (b)] re-
orient in the formation of a primary solvated negative ion; and, therefore, only 8 H
bonds [1 to 7 and 10 in Fig. 2.42 (c)] out of 12 H bonds [see Fig. 2.42 (a)] are remade
when the solvated ion is introduced into the cavity.

for positive ions, and

for negative ions.

By analyzing the structure breaking for octahedral (n = 6) coordin-
ation, one can develop expressions for the heat of solvation of ions with
six solvent molecules in their primary solvent shells. In this case, the ex-
pression is
 ION-SOLVENT INTERACTIONS 93

2.3.5. How Good Is the lon-Dipole Theory of Solvation?

The heats of ion-solvent interactions calculated on the basis of the

ion-dipole approach [Eq. (2.33) and (2.34)] can now be compared with
the experimental values used to test the Born theory. The comparison is
therefore made with values obtained by equally dividing the experimental
heat of solvation of the salt KF between K + and F- ions and then using
the individual values thus gained in data for the experimental hydration
heats of other salts (cf Section 2.2.7).
The comparison (Table 2.12) shows that the ion-dipole model is a
considerable improvement over the rudimentary Born continuum model.
The improvement indicates that the ion-dipole model is on the right track
in considering that an ion sees the solvent in contact with it as consisting
of discrete water dipoles which orient around it. It is only the solvent lying
farther out which the ion views Born-wise as a dielectric continuum. Thus,
by assuming that the solvent has the bulk dielectric constant right up to
the surface of the ion, the Born model missed the work of orientation of
water dipoles around the ion and the related change of dielectric constant
of water near the ion. t
When one considers numerically the various contributions to the heats
of ion-solvent interactions calculated from Eqs. (2.33) and (2.34), it can
be seen (Table 2.13) that the main contributions come from the ion-dipole
and Born charging terms, i.e.,
NAnZieoflw
(ri + rw)2
and

respectively. This fact must be taken to mean that ion-solvent interactions

are essentially electrostatic in origin. The ion behaves like a charged sphere
to the water outside the primary solvent sheath and like an orienting
attracting charge to the water molecules inside the primary solvent shell.
The approximately ± 10% agreement between the calculated and ex-
perimental values for the heats of solvation of ions should normally be
cause for jubilation, but the situation here is abnormal. The so-called ex-
perimental values have been obtained by splitting the unambiguous experi-

t The connection between the orientation of water dipoles around ions and the dielectric
constant of the medium will be looked into much further in Section 2.5.
94 CHAPTER 2

TABLE 2.12
Comparison between the Experimental Heats of lon-Solvent Interactions
and Those Calculated on the Basis of the lon-Dipole Approach

LJHI _8 calc. with Eqs. (2.33) Exp. LJHI _8 with

Ion
and (2.34), kcal mole- 1 LJHKLH.O = LJHF--H.O = tLJHKF-H.O

Li+ -160.1 -146.3

Na+ -119.2 -118.9
K+ - 90.5 - 98.9
Rb+ - 81.9 - 93.8
Cs+ - 71.8 - 88.0
F- - 78.6 - 98.9
Cl- - 56.8 - 64.9
Br- - 51.6 - 58.4
1- - 44.7 - 48.6

mental value of the heat of solvation of KF equally between K + and F-

ions, i.e., by assuming that ions of equal radius but opposite signs have
equal heats of solvation. This assumption, based on the Born non structural
model, taints the so-called experimental values of the individual heats of
ionic solvation, which are being used to test the numerical values calculated

TABLE 2.13
Various Contributions to the lon-Solvent Interactions

Heat of Born charging, Heat of ion-dipole interactions,

Ion
kcal mole- 1 kcal mole-1

Li+ -49.6 -130.5

Na+ -45.0 - 94.2
K+ -40.8 - 69.9
Rb+ -39.3 - 62.6
Cs+ -37.5 - 54.3
F- -40.5 - 68.1
Cl- -36.5 - 50.3
Br- -35.4 - 46.2
1- -33.9 - 40.8
 ION-SOLVENT INTERACTIONS 95

by the structural ion-dipole theory. One has therefore to scrutinize the

assumption more carefully in order to assess the progress made in the
understanding of the heats of solvation of individual ions.

2.3.6. The Relative Heats of Solvation of Ions on the Hydrogen

Scale

Consider the unambiguous experimental value Ll H nx -H 20, the heat of

interaction between HX and water. It is made up of the heats of solvation
of H+ ions and X- ions t

(2.36)

The heat of solvation of X- ions relative to that of H+ ions, i.e., LlHx- (reI)
can be defined by considering LlHH+ an arbitrary zero in Eq. (2.36)

LlHx- (reI) = LlHx- (abs) + LlHH+ (abs) = LlHnx (2.37)

where the notation (reI) and (abs) has been inserted to distinguish be-
tween the relative LlHx- value of X- ions on a arbitrary scale of LlHH+
(abs) = 0 and the absolute or true LlHx- values. From equation (2.37),

LlHx- (reI) = LlHux (2.38)

and, since iJHux can be experimentally obtained to a precision determined

by measuring techniques, one can see that the relative heats of solvation,
LlHx- (reI), are clear-cut experimental quantities.
Relative heats of solvation can also be defined for positive ions. One
writes
iJHMx = iJHM+ (abs) + LlHx- (abs) (2.39)

and substitutes for LlHx- (abs) from Eq. (2.37). Thus, one has

LlHx- (abs) = LlHx- (reI) - LlHu+ (abs) (2.40)

= LlHHX - LlHu+ (abs) (2.41 )

t One should, strictly speaking, write

but the -H 20 will be dropped out in the subsequent text to make the notation less
cumbersome.
96 CHAPTER 2

TABLE 2.14
Relative Heats of Hydration of Individual Ions. llHH+ (abs) = 0

Ion Relative heats of hydration

Li+ +136.34
Na+ +163.68
K+ + 183.74
Rb+ +188.80
Cs+ +194.60
F- -381.50
CI- -347.50
Br- -341.00
1- -331.20

which, when inserted into Eq. (2.39), gives

!JHMX = !JHM+ (abs) + !JHHX - !JHH+ (abs) (2.42)
or
(2.43)

Taking !JHH+ (abs) as an arbitrary zero in this equation permits the defin-
ition of the relative heat !JHM+ (reI) of solvation of positive ions
!JHM+ (reI) = !JHM+ (abs) - !JHH+ (abs) (2.44)
= !JHMX - !JHHX (2.45)

Since .!JHMX and !JHHX are unambiguous experimental quantities, so are

the relative heats of solvation, !JHM+ (reI), of positive ions.
On this basis, a table of relative heats of solvation of individual ions
can be drawn up (cf Table 2.14). These relative heats will now be used to
gauge the assumption that ions of equal radii and opposite charge have
equal heats of solvation.

2.3.7. Do Oppositely Charged Ions of Equal Radii Have Equal

Heats of Solvation?
Consider two ions Mi+ and X i- of equal radius ri but opposite charge.
If their absolute heats of solvation are equal, then, one expects that
(2.46)
 ION-SOLVENT INTERACTIONS 97

But, from the definition of the relative heats of solvation of positive ions
[Eq. (2.45)] and of negative ions [Eq. (2.40)], one has by subtraction

LJHMi+ (abs) - LJHXi - (abs) = [LJHMi+ (reI) - LJHxc (reI)] + 2LJHH + (abs)
(2.47)

If, therefore, the left-hand side is zero, then one should find, since
LJHH + (abs) is a constant, that

(2.48)

This prediction can easily be checked. One makes a plot of the experi-
mentally known relative heats of solvation of positive and negative ions as
a function of ionic radius. By erecting a perpendicular at a radius ri, one
can get the difference [LJHMi+(rel) - LlHxc(rel)] between the relative heats
of solvation of positive and negative ions of radius ri' By repeating this
procedure at various radii, one can make a plot of the differences [LJHM+ (reI)
- LJHx- (reI)] as a function of radius. If oppositely charged ions of the
same radius have the same absolute heats of hydration, then, [LJHM+ (rel)
- LJHx- (reI)] should have a constant value independent of radius. It does
not (Fig. 2.44).
When, however, one examines the terms in Eqs. (2.33) and (2.34)
for the heat of ion-solvent interaction, it is clear that neither the Born
charging term nor the ion-dipole term depends on the sign of the charge
on the ion. The only term which does depend on the sign of the ionic charge
is the first term W, which is 20 kcal mole-1 for positive ions and 30 kcal

300
c::

-.,-
~ I
290
~Q)
1-

:t;~ ~ 280
"l-
I 8
.>' 270
0
~
;-
260
:t;:t
..sl 250 0
~
4
r in A
Fig. 2.44. Plot of the difference between the relative heats of
hydration of oppositely charged, equiradii ions versus ionic radius.
98 CHAPTER 2

Oxygen
atoms
is equivalent to

+q
f?q

-q
,(1
""""'M'I/y
Hydrogen +q +q
atoms

Fig. 2.45. The electrical equivalence between a water molecule and a quadrupole.

mole- 1 for negative ions having n = 4. Since this argues for a constant
difference between the heats of solvation of positive ions and negative ions,
W cannot explain why [LtHM+ (reI) - LtHx- (rel)] is not a constant but
varies with radius. t One has to seek an alternative explanation.

2.3.8. The Water Molecule Can Be Viewed as an Electrical Qua-

drupole

The structural approach to ion-sol vent interactions has been developed

so far by considering that the electrical equivalent of a water molecule is
an idealized dipole, i.e., two charges of equal magnitude but opposite sign
separated by a certain distance. Is this an adequate representation of the
charge distribution in a water molecule?
Consider an ion in contact with the water-molecule; this is the situation
obtained in the primary hydration sheath. The ion is close enough to see
one positively charged region near each hydrogen nucleus and two negatively
charged regions corresponding to the lone pairs near the oxygen atom.
In fact, from this intimate viewpoint, the charge distribution in the water
molecule can be represented (Fig. 2.45) by a model with four charges of

t The assumption has been made that W depends only on the sign of the charge on
the ion and not on the radius of the ion. While this assumption is reasonable in an
approximate evaluation of W, a closer examination of the expression for W, i.e.,
W = Wp + We + (WCF + Ws B ), reveals that the volume of the cavity that has to
be formed and the extent of structure breaking do depend on ionic radius. However,
the work of cavity formation has a sign opposite to that of structure breaking, and,
therefore, there should be some cancellation of these radius-dependent effects. Further,
if W+ for positive ions has a radius dependence, so has W_ for negative ions, and it
is likely that, in taking the difference W+ - W_ for two ions of opposite charge but
equal radius, there is no significant radius dependence of W+ - W_. This, however,
is an assumption which has to be substantiated by exact analysis. A similar statement
applies to the assumption that n is independent of radius, and this must clearly break
down for sufficiently large ions.
 ION-SOLVENT INTERACTIONS 99

equal magnitude q-a charge of +q near each hydrogen atom, and two
charges each of value -q near the oxygen atom. Thus, rather than consider
that the water molecule can be represented by a dipole (an assembly of
two charges), a better approximation, suggested by Buckingham (1957),
is to view it as a quadrupole, i.e., an assembly of four charges. What may
this increase in realism of model do to the remaining discrepancies in the
theory of ion-solvent interactions?

2.3.9. The lon-Quadrupole Model of lon-Solvent Interactions

It will be recalled that the structural calculation of the heat of ion-

solvent interactions (cf Sections 2.3.3 and 2.3.4) involved the following
cycle of hypothetical steps: (1) A cluster of n + I water molecules is
removed from the solvent to form a cavity; (2) the cluster is dissociated
into n + 1 independent water molecules; (3) n out of n + I water molecules
are associated with an ion in the gas phase through the agency of ion-dipole
forces; (4) the primary solvated ion thus formed in the gas phase is plunged
into the cavity; (5) the introduction of the primary solvated ion into the
cavity leads to some structure breaking in the solvent outside the cavity;
and (6), finally, the water molecule left behind in the gas phase is condensed
into the solvent. The heat changes involved in these six steps are WCI!"
WD , WI - D , WBC , WSB , and We, respectively, where, for n =4.

W = WCF + WD + WSB + We = +20 for positive ions

= +30 for negative ions (2.29)
W - - 4NA z i eof-lw (2.25)
I-V - ( + )2
r'i 'w

and the total heat of ion-water interactions is

(2.28)

If one scrutinizes the various steps of the cycle, it will be realized that
only for one step, namely, step 3, does the heat content change [Eq. (2.25)]
depend upon whether one views the water molecule as an electrical dipole
or quadrupole. Hence, the expressions for the heat changes for all steps
except step 3 can be carried over as such into the theoretical heat of ion-
100 CHAPTER 2

Fig. 2.46. Improvement in the calculation

of the ion-water molecule interactions by
altering the model of the water molecule
from a dipole to a quadrupole.

water interactions, LlH1- H • O , derived earlier. In step 3, one has to replace

the heat of ion-dipole interactions, W1- D [Eq. (2.25)] with the heat of
ion-quadrupole interactions (Fig. 2.46).
But what is the expression for the energy of interaction between an ion
of charge ZieO and a quadrupole? The derivation of a general expression
requires sophisticated mathematical techniques, but, when the water mole-
cule assumes a symmetrical orientation (Fig. 2.47) to the ion, the ion-
quadrupole interaction energy can easily be shown to be (Appendix 2.3)

(2.49)

E =_ Zj 90 #w + Zj 90Pw
1-0 r2 2r3

Ion Quadrupole

Fig. 2.47. The symmetrical orientation of a quadrupole to an ion.

 ION-SOLVENT INTERACTIONS 101

where the + in the ± is for positive ions, and the - is for negative ions,
and the Pw is the quadrupole moment (3.9 X 10-26 esu) of the water mole-
cule. It is at once clear that a difference will arise for the energy of interac-
tion of positive and negative ions with a water molecule, a result hardly
forseeable from the rudimentary Born viewpoint and hence probably ac-
countable for the result of Fig. 2.44.
The first term in this expression [Eq. (2.49)] is the dipole term, and
the second term is the quadrupole term. It is obvious that, with increasing
distance r between ion and water molecule, the quadrupole term becomes
less significant. Or, in other words, the greater the value of r, the more
reasonable it is to represent the water molecule as a dipole. But, as the
ion comes closer to the water molecule, the quadrupole term becomes
significant, i.e., the error involved in retaining the approximate dipole
model becomes more significant.
When the ion is in contact with the water molecule, as is the case in
the primary solvation sheath, the expression (2.49) for the ion-quadrupole
interaction energy becomes

(2.50)

The quantity E1 - Q represents the energy of interaction between one

water molecule and one ion. If, however, four water molecules surround
one ion and one considers a mole of ions, the heat change W1 - Q involved
in the formation of a primary solvated ion through the agency of ion-
quadrupole forces is given by

(2.51)

where, as before, the + in the ± refers to positive ions and the - to neg-
ative ions.
Substituting this expression for W1- Q in place of W1- D in expression
(2.28) for the heat of ion-water interactions, one has

NA.(Zi eO)2
2(ri + 2rw)

(2.52)
102 CHAPTER 2

for positive ions and

iJH = 30 _ NAZjeo/-lw _ 4NA z jeOpw N A(z.je o)2

r -H 20 (ri + rw)2 2(ri + rw)3 +
2(ri 2rw)

(2.53)

for negative ions.

2.3.10. lon-Induced-Dipole Interactions in the Primary Solva-

tion Sheath

If one compares Eqs. (2.52) and (2.53) with Eqs. (2.33) and (2.34),
it is clear that the ion-quadrupole calculation of L1Hr - H2o differs from the
ion-dipole calculation of the same quantity in only one respect: In repre-
senting the water molecule by a quadrupole, one is making a more refined
assessment of the interactions between the ion and the water molecules
of the primary solvation sheath. At this level of sophistication, one
wonders whether there are other subtle interactions which one ought to
consider.
For instance, when the water molecule is in contact with the ion, the
field of the latter tends to distort the charge distribution in the water mole-
cule. Thus, if the ion is positive, the negative charge in the water molecule
tends to come closer to the ion and the positive charge to move away. This
implies that the ion tends to induce an extra dipole moment in the water
molecule over and above its permanent dipole moment. For small fields,
one can assume that the induced dipole moment /-lind is proportional to
the inducing field X
/-lind = aX (2.54)

where a, the proportionality constant, is known as the deformation polariz-

ability and is a measure of the "distortability" of the water molecule along
its permanent dipole axis.
Thus, one must consider the contribution to the heat of formation
of the primary solvated ion, i.e., step 3 of the cycle used in the theoretical
calculation presented above, arising from interactions between the ion and
the dipoles induced in the water molecules of the primary solvent sheath.
The interaction energy between a dipole and an infinitesimal charge dq is
-/-l dq/r2, or, since dq/r2 is the field dX due to this charge, the interaction
energy can be expressed as -/-l dX. Thus, the interaction energy between
the dipole and an ion of charge ZjeO , exerting a field zje o/r2 can be found
 ION-SOLVENT INTERACTIONS 103

by performing the integration - f~ieO/r21l dX. In the case of permanent

dipoles, Il does not depend on the field X and one gets the result (cf
Appendix 2.2)
(2.55)

For induced dipoles, however, Ilind = aX, and, hence,

Considering a mole of ions and four water molecules in contact with an

ion, the heat of ion-induced-dipole interactions is
4NAa(z;eo)2
2(r.; + rw)4

Introducing this induced dipole effect into the expression for the heat
of ion-solvent interactions [Eqs. (2.52) and (2.53)], one has

for positive ions, and

4NAz;eo/lw 4NA z.e Opw
JH1 - H20 30 -
(ri + rw)2 2(r.; + rw)J
= -;---'-~

NA(z;e o)2
- 2(ri + 2rw)
(lIT 8cw)
cw 2 aT -
4N Aa(z;eo)2
(2.58)
- lOW - 2(ri + rw)4

for negative ions.

2.3.11. How Good Is the lon-Quadrupole Theory of Solvation 1

A simple test for the validity of these theoretical expressions (2.57) and
(2.58) can be constructed. Consider two ions M/ and X i- of equal radius
but opposite charge. The difference JHMi+ (abs) - JHxc (abs) in their
absolute heats of hydration is obtained by subtracting Eq. (2.58) from
Eq. (2.57). Since the signs of the dipole term, namely,
4NA z i eoflw
(ri + rw)2
104 CHAPTER 2

the Born charging term, namely,

and the induced dipole term, namely,

4NA (Zi eO)2

a 2(ri + rw)4

are invariant with the sign of the charge of the ion, they cancel out in the sub-
traction (so long as the orientation of a dipole near a cation is simply the
mirror image of that near an anion). The quadrupole term, however, does
not cancel out because it is positive for positive ions, and negative for
negative ions. Hence, one obtains t

(2.59)

It is seen from this equation that the quadrupolar character of the water
molecule would make oppositely charged ions of equal radii have radius-
dependent differences in their heats of hydration (cf Fig. 2.44). Further,
Eq. (2.47) has given

iJHMi+ (abs) - iJHxc (abs) = iJHMi+ (rel) - iJHxj - (rel) + 2iJHH + (abs)
(2.47)
By combining Eqs. (2.47) and (2.59), the result is

(2.60)

Thus, the ion-quadrupole model of ion-solvent interactions predicts

that, if the experimentally available differences iJHMi+ (reI) - iJHXi - (reI)
in the relative heats of solvation of oppositely charged ions of equal radii
ri are plotted against (rj + rw)-3, one should get a straight line with a
slope +4NA z j eOpw. From Fig. 2.48, it can be seen that the experimental
points do give a straight line except as the ionic radius falls below about
1.3 A. Further, the theoretical slope (1078 kcal mole-l A3) is in fair agree-
ment with the experimental slope (909 kcal mole- l A3).

t The expression (2.59) is based on the assumption of the radius independence of

w+ - w_ = 20 - 30 = - 10 and the constancy of n with radius over the interval
concerned (cf footnote on p. 98).
 ION-SOLVENT INTERACTIONS 105

It can therefore be concluded that, by considering a quadrupole model

for the water molecule, one can not only explain why oppositely charged
ions of equal radius have differing heats of hydration (cf Fig. 2.44) but
also predict quantitatively the way these differences in the heats of hydration
vary with radius.
It remains to test the values of the absolute heat of hydration calculated
from Eqs. (2.57) and (2.58). The question is: With what experimental values
must the calculated values be compared? Obviously, the theoretical values
cannot be checked with so-called experimental values obtained-following
the rudimentary Born view of Eq. (2.22)-by dividing L1HKF-H20 by two
to get L1HKLH20 and L1HF-- H20 because this procedure assumes incorrectly
that oppositely charged, equiradii ions should have equal heats of interac-
tion with water molecules. Hence, one has to develop a set of absolute
heats of hydration which is based on the fact that oppositely charged ions
of equal radii have differences in their heats of hydration and that these
differences are proportional to (ri + rw)-3.
An elegant method of obtaining such experimental values is available.
Starting from the experimentally proved linearity of L1HM,.+(rel) - L1Hx ,.-(rel)
versus (ri + rw)-3 (cf Fig. 2.48), one can take Eq. (2.60)

and, following Halliwell and Nyburg (1963), extrapolate the L1HM+ , (rel)
- L1HXi- (reI) versus (ri + rw)-3 plot to infinite radius, i.e., to (ri + rw )-3 ---+ 0
The intercept which is 522 kcal mole-1 is then equal to -2LlHH + (abs) - 10,
or
L1HH+ (abs) = 266 kcal mole-1

Once one has obtained thus the experimental absolute heat of hydration
of the proton, one has the heat of hydration of one individual species but
on a much better basis than that gained by splitting the heats of hydration
of KF in half. This semiabsolute value can now be introduced into the
heats of hydration of HX compounds to yield individual heats of hydration
of X-, which can then be used with the experimental heats of hydration
of MX to give the individual heats of hydration of other ions (Table 2.15).
When these experimental values are used to check the values of the
absolute heat of hydration calculated by theory, i.e., by Eqs. (2.57) and
(2.58), it is seen (Table 2.16) that there is agreement between theory and
experiment, with an average disagreement of about 5%.
106 CHAPTER 2

2500~~1~~1~~1~~1~_L-1~~1
0.02 004 0.06
(Ij+ 1.3sf3 inA- 3

Fig. 2.48. The plot of half the difference in the

relative heats of hydration of positive and negative
ions of the same radii versus (rj + 1.38)-3. The solid
line is through experimental points and the dotted
line is the extrapolation of the straight line going
through the experimental points.

TABLE 2.15
Quasi-Experimental Absolute Heats of Hydration of Various Individual Ions

Ion Absolute heat of hydration

Li+ -129.7
Na+ -102.3
K+ - 82.3
Rb+ - 77.2
Cs+ - 71.4
F- -115.5
C1- -81.5
Br- 75.0
1- - 65.2
 TABLE 2.16
Comparison of !J.H1_H,O Calculated from Equations (2.57) and (2.58) with Experimental Values t

Ion Born term Ion-dipole term Ion-quadrupole term Ion-induced-dipole term Totalt Experimental Deviation, %

Li+ -49.6 -130.5 +69.5 -62.4 -153.0§ -129.7 -18

Na+ -45.0 94.2 +42.6 -32.7 -109.3 -102.3 6.8

K+ -40.8 69.7 +27.1 -19.2 82.6 83.3 0.4

Rb+ -39.3 62.6 +23.1 -14.6 73.4 77.2 + 5

Cs+ -37.5 54.3 +18.7 -10.5 63.6 71.4 +11

F- -40.5 68.1 -26.2 -16.5 -121.3 -115.5 5

o
Cl- -36.5 50.3 -16.6 9.4 82.8 81.5 2 Z
I
CJ)
o
Br- -35.4 46.2 -14.6 7.9 74.1 75.0 + 1.2 ~
m
Z
-I
1- -33.9 40.8 -12.2 6.4 63.3 65.2 + 3
Z
-I
m
:rJ
»
t All values in kilocalories per mole. ~
t These totals include the +20 kcalmole- 1
for positive ions and +30 kcal
mole- 1 for negative ions [cf. EQs. (2.57) and (2.58)]. o
z
CJ)
§ Some authors [ef A. D. Buckingham, Discussions Faraday Soc., 24: 151 (1957)] have used a Li+ ionic radius of 0.781, in which case the cal-
culated LlHI-H20 = 144.3 kcal mole-I, corresponding to a deviation of 11%. The figure of -153.0 corresponds to an ionic radius of 0.601.
...o
...
 108 CHAPTER 2

j"
~ - 550 r----------------------,
E
"3 -500 p',c:r~ oQ..,-(
l( d"
'!. ,p"---o.''O'/
+f-450~ /
+ "
tl:'::I( -400t- d'
"I i I I I
Co Sc Ti V Cr Mn Fe eo Ni Cu Zn
(a I

,
.!!
o
E
c 1200
u
l( J;r---------O
~ -1100 ,I:r ""'O-'r:t"
:I: s::/
:1 ,0"
+... -1000 /
tl:"" ,
"I d
- 900 Sc!:--:T;"'i-:v';-;C~r-:M!-:-n--=Fe--=eo-:l-N::-i..L...L.~G-o-l

Fig. 2.49. Plot of the heats of hydration

of transition-metal ions (and their imme-
diate neighbors) versus atomic number; (a)
divalent ions and (b) trivalent ions.

2.3.12. The Special Case of Interactions of the Transition-Metal

Ions with Water

Theoretical considerations on the heats of ion-solvent interactions

have been restricted thus far to alkali-metal and halide ions. For these
ions, it has been shown that the ion-quadrupole theory is in good agreement
with experiment. t In the case of the transition-metal ions, some complica-
tions arise.
A simple way of presenting these complications is to plot the experi-
mental heats of hydration of transition-metal ions versus their atomic
number. It is seen that, in the case of both divalent and trivalent ions, the
heats of hydration lie on double-humped curves [Fig. 2.49(a) and (b)].
Now, if the transition-metal ions had spherical charge distributions,
then one would expect that, with increasing atomic number, there would

t The theory is almost, but not quite, as good for doubly charged alkaline-earth ions.
 ION-SOLVENT INTERACTIONS 109

Fig. 2.50 The five 3d orbitals.

be a decreasing ionic radius t and thus a smooth and monotonic increase

of the heat of hydration in the negative direction. The double-humped
curve implies therefore the operation of factors which make transition-
metal ions deviate from the behavior of charged spheres. What are
these factors?
In the case of transition-metal ions, it is the shapes of the 3d orbitals
which contribute the special properties. The 3d orbitals are not spherically
symmetrical; in fact, they are as shown in Fig. 2.50.
In a gaseous ion (i.e., a free unhydrated ion), all the 3d orbitals are
equally likely to be occupied because they all correspond to the same
energy. Now, consider what happens when the ion becomes hydrated by
six+ water molecules situating themselves at the corners of an octahedron
enveloping the ion. The lone electron pairs of the oxygen atoms (of the
water molecules) exert a repulsive force on the valence electrons of the ion
(Fig. 2.51).
This repulsive force acts to the same extent on all the p orbitals, as
may be seen from Fig. 2.52. The d orbitals, however, can be classified in
two types: (1) those that are directed along the X, Y, Z axes-these are
known as the de orbitals-, and (2) those that are directed between the
axes-these are known as the dy orbitals. It is clear (cf Fig. 2.53) that the
repulsive field of the lone electron pairs of the oxygen atoms acts more
strongly on the de orbitals than on the dy orbitals. Thus, under the electrical

t The radius of an ion is determined mainly by the principal quantum number and the
effective nuclear charge. As the atomic number increases in the transition-metal series,
the principal quantum number remains the same, but the effective charge seen by the
valence electrons increases; hence, the ionic radius should decrease with atomic number.
t The figure of six, rather than four, is used because of the experimental evidence that
transition-metal ions undergo six coordination.
110 CHAPTER 2

x Negative end of
water molecule
Fig. 2.51. Schematic diagram to show that
the valence electrons of the positive ion are
subject to the repulsion of the negative ends
(e) of the octahedrally coordinating water
molecules. The negative charge arises from
the presence of lone electron pairs on the
oxygen atoms of the water molecules.

influence of the water molecules of the primary solvation sheath, all the 3d
orbitals do not correspond to the same energy. They are differentiated into
two groups: The d, orbitals correspond to a higher energy and the dy or-
bitals to a lower energy. It will now be shown that this splitting of the 3d
orbitals into two groups (with differing energy levels) affects the heat of
hydration and hence makes it deviate from the values expected on the basis
of theory developed earlier in this chapter, which neglected interactions
of the water molecules with the electron orbitals in the ion.
Consider a free vanadium ion Y++ and a hydrated vanadium ion.

z z

y
y

Fig. 2.52. Schematic diagram that shows that the three p or-
bitals (directed along the axes) are equally affected by the re-
pulsive field of octahedrally coordinating water molecules.
 ION-SOLVENT INTERACTIONS 111

z z

y y

dE Orbitals
(a)
z z

(b) dy Orbitals

Fig. 2.53. Because the de orbitals [see (a)] are directed along the axes and toward
the negative ends of the water molecules, they correspond to a higher energy than
the d y orbitals [see (b)] which are directed between the axes.

In the case of the free ion, all the five 3d orbitals (the two de and the three
dy orbitals) are equally likely to be occupied by the three 3d electrons of
vanadium; the reason is that, in the free ion, all the five 3d orbitals corre-
spond to the same energy. In the hydrated Y++ ion, however, the dy orbitals,
corresponding to a lower energy, are more likely to be occupied than the
de orbitals. This implies that the mean energy of the ion is less when the
ion is subject to the electrical field of the solvent sheath than when it is free.
Thus, the change in the mean occupancy of the various 3d orbitals, arising
from the electrical field of the water molecules coordinating the ion, has
conferred an extra stabilization (lowering of energy) on the ion-water
system, and, to that extent, the heat of hydration is made more negative.
In the case of the hydrated divalent manganese ion, however, its five
112 CHAPTER 2

.,
(5
E
'8 525
:.;
c
'0 475
J:
..
t~ 425
:a;
~ 375~~~=-~~~~~~~~~

Fig. 2.54. The plot of the heat of hy-

dration of Ca++, Mn++, and Zn++ versus
atomic number.

3d electrons are distributed t among the five 3d orbitals, and the decrease
in energy of three electrons in the dy orbitals is exactly compensated for
by the increase in energy of the two electrons in the de orbitals. Thus, the
mean energy of the ion in the hydrated state is the same as that in the free
state, and there is no extra stabilization produced by the solvent sheath.
Similarly, for Ca++ with no 3d electrons and Zn++ with a completely filled
3d shell, the heat of hydration does not become more negative than would
be expected from the electrostatic theory of ion-solvent interactions develop-
ed in Section 2.3.10 and earlier. It can be concluded, therefore, that the
experimental heats of hydration of these three ions should vary in a mono-
tonic manner with atomic number, as, indeed, they do (Fig. 2.54).
All the other transition-metal ions, however, should have contribu-
tions to their heats of hydration from the energy stabilization produced
by the field of the water molecules. It is these contributions which pro-
duce the double-humped curve of Fig. 2.55. If, however, for each ion,
the energyt corresponding to the water-field stabilization is subtracted
from the experimental heat of hydration, then the resulting values should
lie on the same smooth curve yielded by plotting the heats of hydration
of Ca++, Mn++, and Zn++ versus atomic number. This reasoning is found
to be true (cf Fig. 2.55).
The argument has been presented here for divalent ions, but it is
equally valid (Fig. 2.56) for trivalent ions. Here, it is Sc+++, Fe+++, and
Ga+++ which are similar to manganese in that they do not acquire any

t The five electrons tend to occupy five different orbitals for the following reason: In
the absence of the energy required for electrons with opposite spins to pair up, elec-
trons with parallel spins tend to occupy different orbitals because, according to the
Pauli principle, two electrons with parallel spins cannot occupy the same orbital.
t This energy can be obtained spectroscopically.
 ION-SOLVENT INTERACTIONS 113

..~, 525
8
~ 475
o
-£425
+1
+:.:
~ 375

Fig. 2.55. The plot of the heat of hy-

dration of the divalent transition-metal ions
versus atomic number (0, experimental
values; ., values after subtracting water-
field stabilization energy).

stabilization energy from the field of the water molecules, i.e., water-field
stabilization energy.
In conclusion, therefore, it is the contribution of the water-field stabil-
ization energy to the heat of hydration which is the special feature distin-
guishing transition-metal ions from the alkali-metal, alkaline-earth-metal,
and halide ions in their interactions with the solvent.

2.3.13. Some Summarizing Remarks on the Energetics of lon-

Solvent Interactions

The first aspect of the ion-solvent interactions considered in this

chapter has been the energetics of these interactions, i.e., the theory of the
free energy and heat of solvation.
In an initial attempt to calculate these energies, a continuum approach
was adopted with the ion likened to a charged sphere and the structured

,
~
.. -1200~-----------'

8
~ -1100
.'E .0---0., /~~
o " ~~
x-IOOO /Cf.Y
tl
+
(Y/~./
I

:.: .l.
:t: -900!,:, I I I I I I I I
<:.:I Sc Ti V Cr Mn Fe Co Ni Go

Fig. 2.56. The same as Fig. 2.55 but for

trivalent ions.
114 CHAPTER 2

solvent to a continuum having the bulk dielectric constant of the solvent.

According to this first model of Born, the free energy of solvation is equal
to the electrostatic work of discharging the sphere (which represents the
ion) in vacuum, transferring the uncharged sphere into the solvent, and then
charging it up again.
To compare with experiment the free energies and heats of solvation
thus calculated, it was necessary to find a way of knowing the experimental
heats of solvation of individual ions. This provided a fair problem because
experimental measurements of the lattice energies and heats of dissolution
only gave the sum of the heats of solvation of both ions of a salt and
not those of individual ions. As a first attempt, therefore, it was decided
to split the heat of solvation of KF equally between K+ and F- ions,
which have almost equal radii. The rationale behind this step was the
Born model, which argued for ions of equal radii having equal heats of
solvation.
When these K F-derived experimental heats were compared with the
Born heats of solvation, it turned out that the latter were of the right order
of magnitude. This demonstrated the essentially electrostatic character of
ion-solvent interactions. The calculated values were however too high which
necessitates further theoretical considerations. Rather than be tempted into
tampering with the ionic radii and dielectric constant and making them
into adjustable parameters, it was decided to take detailed, structural view
of the process of ion-solvent interactions.
Confining oneself to water as the solvent, a sketch was presented of
the structure of water. Emphasis was laid on the fact that, while water is
an uncharged molecule, its centers of the negative and positive charges do
not coincide and thus confer a polar character on the molecule. The charge
distribution in the water molecule was, as a first approximation, taken as
equivalent to an electric dipole. Another feature of the structure of water
is the linking up of individual water molecules into tetrahedral networks
consisting of many water molecules in each unit.
When an ion enters water, the structure of the latter is disturbed to an
extent which depends on the distance from the ion. The ion is able to wrench
water molecules out of their continuous network and make them orient
around the ion by coulombic forces which arise from the polar nature of
the water molecule. Thus is formed the primary solvated ion. The water
structure just outside the primary hydration sheath is broken to some extent,
but, further away, the normal network structure obtains.
This picture of the structure of the solvent around ions was used as the
basis of the ion-dipole theory of solvation. The procedure was to view the
 ION-SOLVENT INTERACTIONS 115

total heat of solvation as consisting of three contributions. The first con-

tribution was considered to arise from the interactions between the ion
and the water molecules which are members of the primary solvent sheath;
and the calculation involved the heat of ion-dipole interactions. The second
contribution to the heat of solvation pertained to the interaction between
a primary solvated ion and the surrounding solvent, which, as a first step,
was reckoned to have the bulk structure of water. The calculation of this
second part of the total heat of solvation followed the procedure of the
Born model except that one was dealing with a primarily solvated ion rather
than a bare ion. Since the solvent in the immediate neighborhood of the
primary solvent ion does not have the bulk structure, a correction to the
second contribution was introduced to reckon with the structure breaking.
The change in heat content due to the structure breaking around a primary
solvated ion constituted the third contribution to the total heat of solvation
of ions.
When the values of the heat of solvation calculated by the ion-dipole
theory were compared with the so-called experimental KF-derived values,
it became clear that the structural picture of solvation was in far better
accord with the experimental than the Born model was.
The presentation would have been terminated at that point but for
an interesting check on the validity of the "experimental" values of ionic
solvation heats which were being used as the test of theory. Clear-cut experi-
mental evidence was produced to demonstrate that the difference in the
absolute heats of solvation of ions of opposite charge but equal radius was
neither zero nor independent of ionic radius. This difference should be zero
according to the so-called experimental KF-derived heats, and it should
be radius independent according to the ion-dipole theory. Thus arose the
need for a deeper probing into the structural picture of ion-solvent inter-
actions and the development of the ion-quadrupole theory of solvation.
The one important advance of the ion-quadrupole theory over the
ion-dipole theory lay in the more careful assessment of the interactions
between the ion and the water molecules of the primary solvation sheath.
It was realized that, while the charge distribution in the water molecule
may appear equivalent to an electrical dipole, to an ion situated far away,
the charge distribution is better represented as an electrical quadrupole,
from the point of view of an ion in contact with the water molecule. Thus,
what needed calculation was the heat of ion-quadrupole interactions in
the primary solvated ion.
A confidence in the ion-quadrupole approach was gained immediately
when the theory showed that ions of opposite charge but equal radii should
116 CHAPTER 2

have differing heats of solvation-as found experimentally. So, after also

taking into account interactions between the ions and dipoles induced in
the water molecules, the final expression for the heat of solvation accord-
ing to the ion-quadrupole theory was used to calculate values for the
various ions.
These calculated values obviously could not be compared with KF-
derived experimental values derived on the assumption that oppositely
charged ions of equal radii have equal heats of solvation. Fortunately,
it was possible to isolate experimentally unambiguous heats of hydration
of ions relative to the hydrogen ion and use these relative heats to yield
an absolute heat of hydration of the hydrogen ion and, thus, the absolute
heats of other ions. Comparison of these reliable experimental absolute
heats with those calculated by the ion-quadrapole theory showed close
correspondence between theory and experiment.
In the case of the transition-metal ions, a special factor had to be
considered. The electric field exerted by the solvent sheath alters the distri-
bution of the 3d electrons of the transition-metal ions among the various
3d orbitals and thus confers an extra stabilization energy upon the ion-
solvent system. The heats of hydration of these ions is made more negative
than it would be if no interactions with directed orbitals were considered.
This, then, is the basic theory of the energetics of ion-solvent inter-
actions. It is clear that the polar nature of the solvent molecule is of funda-
mental importance to the heat. of solvation. Viewing the water molecule as
a dipole gives reasonable heats of solvation; viewing it as a quadrupole
gives good results. One can, of course, go further. For instance, one can
consider lateral interactions between the water molecules of the primary
solvent sheath, or one can decide whether there are chemical as opposed
to primarily electrostatic interactions between the ion and the contiguous
water molecules. But these are matters of greater detail, which will be left
to the tomes.

Further Reading

1. J. D. Bernal and R. H. Fowler, J. Chern. Phys., 1: 515 (1933).

2. D. D. Eley and M. G. Evans, Trans. Faraday Soc., 34: 1093 (1938).
3. H. S. Frank and M. Evans, J. Chern. Phys., 13: 507 (1945).
4. L. E. Orgel, J. Chern. Soc., 1952: 4756 (1952).
5. H. S. Frank and W. Y. Wen, Discussions Faraday Soc., 24: 133 (1957).
6. A. D. Buckingham, Discussions Faraday Soc., 34: 151 (1957).
7. O. G. Holmes and D. S. McClure, J. Chern. Phys., 26: 1686 (1957).
8. H. S. Frank, Proc. Roy. Soc. (London), A247: 481 (1958).
 ION-SOLVENT INTERACTIONS 117

9. L. Pauling, The Nature of the Chemical Bond, 3rd ed., Cornell University
Press, Ithaca, N.Y., 1960.
10. H. S. Frank and A. S. Quist, J. Chem. Phys., 34: 604 (1961).
11. G. Nemethy and H. A. Scheraga, J. Chem. Phys., 36: 3882 and 3401 (1962).
12. G. R. Choppin and K. Buijs, J. Chem. Phys., 39: 2035 and 2042 (1963).
13. D. J. G. Ives, Some Reflections on Water, J. W. Ruddock, London, 1963.
14. H. F. Halliwell and S. C. Nyburg, Trans. Faraday Soc., 58: 1126 (1963).
15. J. P. Hunt, Metal Ions in Aqueous Solution, W. A. Benjamin, Inc., New
York, 1963.
16. J. Lee Kavanau, Water and Solute-Water Interactions, Holden-Day Inc.,
San Francisco, 1964.
17. R. P. Marchi and H. Eyring, J. Phys. Chem., 68: 221 (1964).
18. R. P. Feynman, R. B. Leighton, and M. Sands, The Feynman Lectures on
Physics, Addison-Wesley Publishing Company, Inc., Reading, Mass., 1964.
19. B. E. Conway, "Proton Solvation and Proton Transfer Processes in Solu-
tion," in: J. O'M. Bockris, ed., Modern Aspects of Electrochemistry, No.3,
Butterworth's Publications, Inc. London, 1964.
20. O. Ya. Samoilov, Structure of Aqueous Electrolyte Solutions and the Hydra-
tion of Ions, Consultants Bureau, New York, 1965.
21. S. Golden and C. Guttmann, J. Chem. Phys., 43: 1894 (1965).
22. D. R. Rosensteig, Chem. Rev., 65: 467 (1965).
23. B. E. Conway and M. Salomon, in: B. E. Conway and R. G. Barradas, eds.,
Chemical Physics of Ionic Solutions, John Wiley & Sons, Inc., New York, 1966.
24. V. I. Klassin and Yu. Zinovev, Kolloid. Zh. (English translation), 29 (5): 561
(1967).
25. B. V. Deryagin, Z. M. Zorin, and N. V., Churaev, Kolloid. Zh. (English
translation), 30 (2): 232 (1968).
26. A. K. Covington and P. Jones, Hydrogen-Bonded Solvent Systems, Taylor
and Francis Ltd., London, 1968.

2.4. THE SOLVATION NUMBER

2.4.1. How Many Water Molecules Are Involved in the Solva-

tion of an Ion?

Mathematically speaking, the electric force originating from an ion

becomes zero only at infinity. In effect, however, the force fades out to a
negligible value after quite a short distance (of the order of tens of ang-
stroms). Beyond this cutoff distance, solvent molecules may be regarded as
unaware of an ion's presence. There is therefore a certain effective volume
around the ion within which its influence operates. How many solvent
molecules are inside this volume and could therefore be said to be partici-
118 CHAPTER 2

TABLE 2.17
Hydration Number Ascribed to the Sodium Ion in According to Different
Experimental Methods

Ion Hydration numbers reported

Na+ 1, 2, 2.5, 4.5, 6-7, 16.9, 44.5, 71

pants in the solvation of the ion? This number may be termed the solvation
numbert (or hydration number when water is the solvent).
The question of the value of the solvation number is an interesting
one. It is no surprise, therefore, that a large number of different methods
have in the past been used to determine the solvation number (more about
these methods later). But, the alarming thing is that exceedingly discrepant
results are obtained by the various methods. For instance, widely varying
hydration numbers ranging from I to 71 (Table 2.17) have been ascribed
to the sodium ion. Are some of the methods wholly incorrect, or is there a
confusion as to what constitutes a hydration number?
The answer can be approached, if not attained precisely, by the
following considerations. What value of hydration number a particular
method gives depends on what types of ion-solvent interactions the method
senses. If it can pick up the interactions of an ion with water molecules
several molecular diameters away in the secondary region, it will report
that a large number of water molecules are involved in solvation, i.e., a
high hydration number. If, however, the method only detects how many
water molecules an ion takes along in the course of its thermal motions
through the solution (i.e., those tightly bound to it), then it will report a
small hydration number.
To avoid ambiguity, it is best to define a primary solvation number as
the number of solvent molecules which surrender their own translational
freedom and remain with the ion when it moves relative to the surrounding
solvent. Of course, a solvent molecule loses its independent translational
motions only when it is overwhelmed by the ionic force field into adopting

t This total effective number of solvent molecules involved in interactions should not
be confused with the number n used in the structural treatment of the energetics of
solvation. The latter number was meant to represent the number of solvent molecules
in contact with the ion and assumed to be aligned in its field.
 ION-SOLVENT INTERACTIONS 119

TABLE 2.18
Hydration Numbers

Number of independent methods

Ion Hydration number
on which result is based

Li+ 5± 1 5
Na+ 5± 1 5
K+ 4±2 4
Rb+ 3± 1 4
F- 4±1 3
Cl- 1± 1 3
1 ±1 3
1± 1 2

a mInImUm-energy orientation to the ion. Thus, the primary solvation

number can also be defined as the number of solvent molecules which are
aligned in the force field of the ion.
This definition provides a criterion for discussing the different methods
of determining solvation numbers. The primary solvation number should
be determined by only those methods which register the number of water
molecules which are associated with the ion in its travels through the
solution.
When, however, these methods are used (and they will be presented
in Section 2.4.5), it turns out (Table 2.18) that the number of water mole-
cules determined by some of them are less than what geometry says the
number of water molecules in contact with the ion should be. This latter
number is a coordination number, t i.e., the number of nearest-neighbor
water molecules which are in contact with or coordinate or surround an ion.
The question, therefore, arises: Why does not all the coordinated water join
the ion in its zig-zag motions through the solution? In fact, why is the
solvation number not always equal to the coordination number? Further,

t In the structural treatment of the heats of solvation, it was tacitly assumed that the
number n of primary solvent molecules aligned in the ionic field is equal to the coor-
dination number. In other words, the structural treatment slurs over the distinction
between the number that are oriented in the ionic field (i.e., move with the ion) and
the number in contact with the ion.
120 CHAPTER 2

what happens when coordinating solvent molecules desert their position~

in the coordination shell of an ion as soon as it begins its voyage through
the solvent, say, in response to an electric field? Do the missing solvent
molecules leave voids in the coordination shell of a moving ion? The
concept of solvation number will become clear only when such questions
are answered.

2.4.2. Static and Dynamic Pictures of the lon-Solvent Molecule

Interaction

Suppose that, in a thought experiment, a bare ion is made to stop

during its movements through the solution. At that instant, the hypo-
thetical stationary ion will be surrounded or coordinated by water mole-
cules still associated in a network structure (Fig. 2.57). What will happen?
The ionic force field will operate on the neighboring water dipoles. The
forces, which are essentially ion-dipole in nature, will cause some of the
water molecules to break away from the water network and attach them-
selves to the ion.
What is the consideration on the basis of which a particular water
molecule decides to embrace the ion by aligning into its field or to shun
it and remain in the water network? The consideration is simple: Is the
ion-dipole interaction energy greater in magnitude than the hydrogen-bond
energy keeping the particular water molecule in the network? If the ion-
dipole energy is greater in magnitude, the water molecule should link itself
with the ion and form part of the primary solvation sheath. If not, the
water molecule should remain in the water network.
The whole thought experiment described above is a static one. All
that has been done is to consider the energies in the initial state (a water

----~ ----~
~~~.::-:-:.;;;-:-:-:;:-~-~:"'-:::-::::-':"7 Coordination
: ~ water

,~

Fig. 2.57. A hypothetical stationary ion coor-

dinated by water molecules still associated into
a network structure.
 ION-SOLVENT INTERACTIONS 121

Fig. 2.58. Schematic diagram to show

that, of four water molecules which coor-
dinate an ion, two water molecules, A and
B must reorient from positions in which one
of their H atoms faces the ion to positions
in which the same H atoms are away from
the ion. (The required reorientation is shown
by an arrow.)

molecule in the water network and an ion nearby) and in the final state
(water bound to the ion by ion-dipole forces).
But ions can be kept stationary only in thought experiments. In reality,
they exist in a state of ceaseless motion (see Chapter 4). So time and move-
ment must come into the picture of ions interacting with water molecules.
One must abandon a static view for a dynamic view.
One can develop a dynamic view along the following lines (Samoilov).
Consider a water molecule bound by hydrogen bonds to the water network.
Suppose that, at a time taken as zero (t = 0), an ion suddenly appears next
to the water molecule. If the net force on the water molecule is in favor of
its association with the ion rather than with the water network, it will try
to get into an equilibrium position around the ion, i.e., the water molecule
will try to align into a minimum-energy orientation. This usually means
that the water molecule has to reorient (or jump through a small distance
or both) from the position it had in the water structure to the new positipn
of alignment in the ionic field (Fig. 2.58).
But these reorienting or jumping movements to be made by the water
molecule will require afinite time, the value of which depends on the critical
activation energy required for the reorientation or jumping process. Let
this time required for the orientation of a water molecule into the coordin-
ation sheath around an ion be <water orient (Fig. 2.59). This orientation time
will not have a unique value because it will depend on how far the ion is
situated from and on how the ion is located with respect to the water
122 CHAPTER 2

~ Final position

\~ Initial position
,,

e
Fig. 2.59. The time required for a water
molecule in contact with an ion to reorient
from an initial to a final position (shown in
figure) is Torient.

network holding the water molecule. So one is talking about an average

water-orientation time.
Now, instead of considering the ion suddenly placed next to the water
molecule at t = 0, one can visualize the ion resting or waiting near the
water molecule in between its hops from location to location in the solvent
(cf Section 4.2). Of course, if a water molecule belonging to the water
network is to orient toward the ion, it must do so when the ion is within
a certain small distance of the water molecule. But how long does the ion
stay within this jumping range? That depends on how long the ion pauses
next to the water molecule in the course of its jumps through the solvent.
The longer the ion waits near the water molecule, the longer is the time
available for the water molecule to break out of the water lattice and swing
into that intimate ion-dipole relationship with the ion which characterizes

Fig. 2.60. In the course of its hops through

the solution, the hopping ion can be con-
sidered to spend a time Tion wait in contact
with the neighboring water molecules. Will
one of these orient itself into a position of
minimum interaction energy with the ion
before the latter has jumped to a new po-
sition?
 ION-SOLVENT INTERACTIONS 123

a seat in the primary hydration sheath. The hopping ion spends a certain
time in "contact" with the particular water molecule under discussion.
Call this contact time Tion wait (Fig. 2.60).

2.4.3. The Meaning of Hydration Numbers

Now, an interesting qualitative conclusion becomes clear. If the time

an ion waits near a water molecule is long compared with the average time
a water molecule takes to orient into association with an ion, then the
probability of the water molecule's being captured by the ion is high. That
is, the probability of an ion's capturing a water molecule depends on the
ratio Tioll wait /Twater oriellt·
If Tioll wait/Twater oriellt is large, then the ion will be surrounded by the
full geometrically permitted complement of bonded water molecules during
all its zig-zag motions through the solution. Under these circumstances,
the hydration number (i.e., the number of water molecules which participate
in the translational motions of the ion) will be equal to the coordination
number.
If, however, Tion wait /Twater orient is of the order of unity, then the
situation is interesting. The time an ion spends in the neighborhood of a
water molecule is of the order of the water reorientation time, and, hence,
though the ion is not sure to capture a water molecule, there is a certain
probability, less than unity. At the same time, one must consider the opposite
process: An ion with a bound water molecule collides with a water molecule
belonging to the water network. There will be a certain probability that the
ion will lose its water to the water network. But there are plenty of water
molecules all around and the ion has a chance of making up its loss. Thus,
over a period of time which is long compared with the period of contact
between a moving ion and a specific water molecule, the ion has aligned
and trapped in its field a certain number of water molecules which is less
than the number of water molecules which geometrical close packing makes
possible, i.e., the coordination number.
The collisions between ions and water molecules linked to the water
network are analogous to any other collision process. Consider, for example,
the collisions between neutrons and U238 nuclei, in which slow neutrons
stand a better chance of being captured than fast neutrons. One says that
there is a large capture cross section for slow neutrons. It is as if a slow
moving neutron sees a bigger target than a fast moving one.
What happens if the ions wait for so short a time that, even before a
water molecule has had time to break out of the water structure and turn
124 CHAPTER 2

around, the ion has hopped away? Then, the probability of a water mole-
cule's being captured by the ion is zero, and, on a time average, the ion
will not have any aligned water molecules in contact with it, i.e., its primary
hydration sheath is empty. This does not mean that such ions are not
surrounded by interacting solvent molecules or that they would have no
coordination water. It only means that, because Tion wait/Twater orient ~ I,
the ion does not wait long enough at any particular site for the contiguous
water molecules to swing out of the water network into minimum-energy
orientation with the ion. Even if the ion does capture a water molecule,
it is bound to lose it soon. It also means that the moving ion exchanges
water molecules so easily with the surrounding solvent that, in effect,
the moving ion does not carry its sheath along with it. Its solvation number
is zero, though its coordination number is that dictated by geometry.
The picture of solvation numbers presented here is a dynamic one.
The solvation number refers to the number of water molecules which remain
aligned with the ion during its jumps through the medium. But it is not
necessary that the same individual water molecules serve in the solvation
sheath for an indefinitely long time. A given water molecule may serve the
ion for some time, but it is not imprisoned for life in its hydration shell.
A chance collision, and the particular water molecule may link up again
with the water network, get left behind by the hopping ion, and watch
another water molecule yield to the attraction of the ionic field and be
incorporated in the primary solvation sheath.

2.4.4. Why Is the Concept of Solvation Numbers Useful?

In all this dynamic exchange of solvent molecules between the coor-

dination region and the main bulk of solvent, has the concept of solvation
number any utility? Yes, the solvation number can be considered the
effective number of solvent molecules to be "permanently" bound to the
ion and to follow its motion from site to site. The kinetic entity is not the
bare ion but the ion plus the solvation number of water molecules.
The concept of solvation number permits one to suppress the dynamic
nature of the primary solvation sheath from many modelistic considerations
of ions in solution. This is important particularly in situations where one
would overcomplicate an analysis by considering the details of the constant
exchange of water molecules between the ionic primary hydration shell
and the solvent. The overall total action of the ion on the water may be
replaced conceptually by a strong binding between the ion and some effective
number (the solvation number) of solvent molecules; this effective number
 ION-SOLVENT INTERACTIONS 125

may well be almost zero in the case of large ions, e.g., iodide, cesium, and
tetraalkylammonium. The solvation number clearly diminishes with increase
of ionic radius because, with increasing ionic radius, the distance to the
coordinating water molecules increases and, thus, the ionic force field which
aligns the ion diminishes so that the water molecules have less inclination
to reorient away from their solvent-structure positions.
Of course, there may be situations where, quite independent of the
ratio Tion wait /Twater orient, there are thermodynamic restrictions against the
association of solvent molecules with the ion, e.g., the ion-solvent molecule
interaction energy may be less in magnitude than the solvent molecule-
solvent molecule energy. In such cases, the solvation number will be zero
on static considerations alone.

2.4.5. On the Determination of Solvation Numbers

All this discussion would be pointless if there were no agreement be-
tween the different methods of measuring primary solvation numbers.
Fortunately, it turns out that there is some degree of agreement between
the values reported by different methods so long as they are methods
which determine the primary solvation number (Section 2.4.1), as opposed
to the vague and asymptotic concept of total solvation number (see Table
2.19).
TABLE 2.19
Comparison between the Hydration Numbers Determined by Different
Methods

Ion Compressibility Mobility Entropy Theoretical calc.

Li+ 5-6 6 5 6
Na+ 6-7 2-4 4 5
Mg+ 16 14 13
Ca++ 7.5-10.5 10
Zn++ 10 -12.5 12
Cd++ 10 -12.5 11
Fe++ 10 -12.5 12
Cu++ 10.5-12.5 12
Pb++ 4 - 7.5 8
K+ 6-7 3 3
F- 2 5 5
Cl- 0.1 0.9 3 3
Br- 0 0.6 2 2
1- 0 0.2 o
126 CHAPTER 2

A detailed discussion of the various methods of determining solvation

numbers is not intended in this treatment. Nevertheless, it is illustrative
to present two examples.
Consider, for example, the compressibility method. The compressibility
{J is defined by the expression

{J = 1
__ (~) (2.62)
V 8p T

If a pure solvent is considered, then its compressibility may be written

thus

(2.63)

Now suppose that an ionic solution is considered. Will its compressibility

be the same as that of the pure solvent, i.e., {Jsolv? A physical picture of why
a solvent is compressible will provide a qualitative answer.
Let water be the solvent. It has been described (see Section 2.3) as
having quite an open framework structure with many holes in it. When a
pressure is applied, the water molecules can break out of the tetrahedral
framework and enter the interstitial spaces; the water molecules become
packed more closely (Fig. 2.61). Thus, the volume decreases.
This is not the only way of compressing water. When ions are introduced
into the water, they are capable of wrenching water molecules out of the
water framework so as to envelop themselves with solvent sheaths. Because
the molecules are oriented in the ionic field, the water is more compactly
packed in the primary solvation shell as compared to the packing if the

Fig. 2.61. Schematic diagram to show

that, when an external pressure is applied
to water, water molecules break out of the
networks and occupy interstitial spaces.
 ION-SOLVENT INTERACTIONS 127

solvated ion

Fig. 2.62. Schematic diagram to illustrate the principle

of electrostriction; owing to the ionic field, water mole-
cules are more compactly packed in the primary solvation
sheath than in the field.

ion were not there (Fig. 2.62). The water has become compressed by the
introduction of the ion. But what is the origin of the influence of the ion?
The origin is the electric field of the ion. Thus, electric fields cause com-
pression of the material medium upon which they exert their influence;
this phenomenon is known as electrostriction.
Since the introduction of ions into a solvent causes the solvent molecules
in the primary solvent shell to be highly compressed, these water molecules
may be supposed not to respond to any further pressure which may be
applied. Thus, the compressibility of an ionic solution is less than that of
the pure solvent because of the incompressibility of the primary solvation
sheath. t
It is easy to calculate the ratio of the compressibility of a solvent {Jsolv
to that of the solution {Jsoln' Suppose the primary hydration number is
nh' Then, ni moles of ions are solvated with njnh moles of incompressible
water. Now, if nw moles of water correspond to a total volume V of solution,
ninh moles of incompressible water would correspond to a volume VnhnJnw
of incompressible solution. Defining the symbol y thus

(2.64)

t Outside the primary solvent sheath, the water molecules are not oriented to the same
degree as those inside the primary solvation sheath because the orienting ionic field
is less. This means that the non primary water molecules are less electrostricted and
free to respond to pressure. One can, to good approximation, say that the water out-
side the primary solvation shell has the same compressibility as the pure solvent.
128 CHAPTER 2

the volume of the incompressible part of the solution is yV. This volume
must be excluded from the expression for the compressibility of the ionic

+[~
solution. Thus,

{lsoln = - (V - YV)]T (2.65)

{lsolv = - +[~ V L (2.63)

{lsoln = I _ Y (2.66)
{lSOlv
Hence, from (2.64),
(2.67)

_ -nw-
nh (1 _-{lsoln)
- (2.68)
ni {lsolv

This equation can be used to obtain the hydration number by deter-

mining the compressibility of the pure solvent and the ionic solution.
There are several methods available for studying compressibilities. The
ultrasonic method, for example, depends on the fact that sound travels
by a compression-rarefaction process, and, thus, the velocity of an ultra-
sonic wave can be used to determine the compressibilities of solvent and
solution, needed for Eq. (2.68) (Fig. 2.63).
The mobility method of measuring hydration numbers is based on the
following argument (cf also Section 4.4.8). Suppose an ion is made to

Ultrasonic wave the

~velocily of which
~ depends on compressibility

_Ionic solut ion

Ultrosonic source

Fig. 2.63. The ultrasonic method of determin·

ing hydration numbers is based on the fact that
the velocity of the ultrasonic wave depends, by
the compressibility of the solution, on the extent
of primary hydration in the ionic solution.
 ION-SOLVENT INTERACTIONS 129

drift by the application of an external electric field. The motion of the ion
is opposed by the viscous resistance of the solution. When a steady-state
velocity is reached, the electric force is equal to the hydrodynamic viscous
force (Fig. 2.64). The former is simply ZieOX, where X is the electric field
(or potential gradient) in the solution applied by two electrodes placed in
solution (X is often measured in volts per centimeter). The latter is ex-
pressed by a famous classical formula of hydrodynamics called Stokes'
law. This law, which describes the force experienced by a sphere moving
in a viscous medium, states that

Viscous force = 6nr1Jv (2.69)

where r is the radius of the moving ion and 'Y) is the viscosity of the medium.
Thus,
(2.70)
or
ZieO X
r=--
6nrJV

(2.71 )

where u( = vjX), i.e., the velocity under unit electric field, is a measurable
quantity and is often called the electrical mobility of the ion (cf Section
4.4.3).
Once the radius r of the solvated ion is obtained from Eq. (2.71),

Electric field X

Velocity : v

electric
1-----1 force :$i~

Primory solvoted iOIl

Fig. 2.64. The mobility method of determining hydration numbers is

based on finding out the radius of a primary solvated ion from the fact
that, when an ion in solution attains a steady-state velocity, the electric
force Zj e Xo is exactly balanced by opposing viscous force 6nt)rv.
130 CHAPTER 2

one can calculate the hydration number nh by a simple geometric argument

(see Fig. 2.65)

(2.72)

where rcryst is the crystallographic radius of the ion and rH 20 is the radius
of the water molecule, both of which are known from independent data.
Both the compressibility and the mobility methods of determining
primary hydration numbers are based on quite loose approximations. The
compressibility method assumes that the solvent inside the primary solvation
sheath is completely incompressible and the water outside has the same
compressibility as the pure solvent.
It will be recalled, however, that, in the secondary region (see Fig.
2.28 and Section 2.3.2) between the primary solvation sheath and the bulk
water, there is structure breaking and partial alignment of the water mole-
cules. Hence, instead of a sharp change of compressibilities at the boundary
of the primary solvation shell, it is likely that there will be a smooth variation
in compressibility from the ion out into the bulk solvent.
The mobility method, on the other hand, ignores the fact that, because
the secondary region does not have the structure of the bulk solvent, the
viscosity of the medium constituting the immediate neighborhood of the
moving primary solvated ion is not the viscosity of the bulk solvent. It
should be the local viscosity of the region surrounding the primary solvated
ion. Such local viscosities are uncertain in value. Another approximation
in the mobility method is that it neglects electrostrictional compression in
computing the volume occupied by the water molecules in the primary
hydration sheath (and, in the rudimentary version of the theory given
here, also free space between water molecules).

Primary solvation
sheath

Tcryst

Fig. 2.65. The calculation of the hydration

number from the radius of the primary solv-
ated ion.
 ION-SOLVENT INTERACTIONS 131

These approximations in the mobility method are offset by one big

advantage. The method gives directly the hydration number of one ionic
species, e.g., sodium ions, and not the sum of the hydration numbers of
positive and negative ions. In the compressibility method, however, one
only obtains the hydration number of the salt, and thus one has all the
problems of resolving the value for the salt into the individual ionic values
that were encountered in getting individual ionic heats of hydration from
heats of hydration of salts (cf Section 2.2.7). One has to depend on some
independent (and sometimes somewhat circular) argument, for instance,
that the relatively large iodide ion should have a hydration number of zero
wholly to the positive ion. Of course, once one is certain of the hydration
number of one ion, one can then get out those of other ions by taking the
appropriate salts.
There are in all about five experimental methods which yield primary
hydration numbers. The results show approximate agreement (± I). Each
method involves some doubts and approximations, and, in some cases, it
is difficult to estimate with even a tolerance of ±25% what effect the ap-
proximations would have on the hydration numbers. Nevertheless, when
one recalls the wild spread (cf Table 2.17) of the values of hydration num-
bers obtained by not distinguishing between methods which determine
primary and total hydration numbers, it must be accepted that the results
of Table 2.20 hang together at least very much better than those in which a
distinction between primary and other types of solvation is neglected. The
results permit one to conclude the basic correctness of the picture of an ion
influencing quite a bit of the surrounding solvent but actually succeeding

TABLE 2.20
Primary Hydration Numbers

From From From apparent From Most probable

Ion
compressibility entropies molal vol mobility integral value

Li+ 5-6 5 2.5 3.5-7 5± 1

Na+ 6-7 4 4.8 2.4 4±1
K+ 6-7 3 1.0 3±2
F- 2 5 4.3 4± 1
Cl- 0-1 3 0 2± 1
Br- 0 2 2± 1
1- 0 1 1± 1
132 CHAPTER 2

in trapping in its field only a certain number of water molecules which

become the baggage of the ion in its travels through the solution.
There is, however, one surprising thing in the extent of agreement
between the various methods. The mobility method is based on the non-
equilibrium process of conduction, and the compressibility method, for
example, is based on the system's being in equilibrium. Yet, the two methods
yield fairly concordant results. The point, however, is that, in considering
hydration numbers, one is not concerned with whether the whole system
(the assembly of ions and solvent particles) is in static equilibrium or
dynamic change. One is concerned with the state of the individual ions.
But these are in ceaseless motion irrespective of whether the whole assembly
is in equilibrium or not. Thus, even methods, such as the compressibility
method, which involve measurements of the solution at equilibrium,concern
in fact ions in a very dynamic state and should therefore give nearly the
same hydration and solvation numbers as one would expect when the ions
are drifting under nonequilibrium conditions, e.g., under an electric field.

Further Reading
1. H. Ulich, Z. Elektrochem., 36: 497 (1930).
2. H. Ulich, Z. Physik. Chem. (Leipzig), 168: 141 (1934).
3. A. Passynsky, Acta Physicochim. URSS, 8: 385 (1938).
4. J. O'M. Bockris, Quart. Rev. (London), 3: 173 (1949).
5. B. E. Conway and J. O'M. Bockris. "Ionic Solvation," in: J. O'M. Bockris,
ed., Modern Aspects 0/ Electrochemistry, Vol. I, Butterworth's Publications,
Inc., London. 1954.
6. J. Padova, J. Chem. Phys., 40: 391 (1964).
7. R. Zana and E. Yeager, J. Phys. Chem., 71: 521 (1967); 71: 4241 (1967).
8. J. F. Hinton and E. S. Amis, Chem. Rev., 67: 367 (1967).

2.5. THE DIELECTRIC CONSTANT OF WATER AND IONIC SO-

LUTIONS

2.5.1. An Externally Applied Electric Field Is Opposed by Coun-

terfields Developed within the Medium

The solvation of ions arises from the interactions between solvent

molecules and ions. These interactions result in the orientation of the,
e.g., water molecules toward the ions. It follows that, as the ionic con-
centration increases, the fraction of the water in a solution which is trap-
ped by ionic fields in the solvation sheaths also increases. Is this con-
 ION-SOLVENT INTERACTIONS 133

centration-dependent extent of ion-water interactions revealed in any

macroscopic property of the electrolyte? The dielectric constant of an
electrolyte is such a property, but, to understand the relationship between
dielectric constant and hydration and, thus, the dependence of dielectric
constants on ionic concentration, one has to have a picture of the atomistic
basis of the dielectric constant of a medium.
The dielectric constant always appears in the expression for the capacity
of an electrical condenser, so why not start to try to understand dielectric
constants in terms of the behavior of capacitors?
Consider a capacitor with plates which are A cm2 in area and d cm
apart (Fig. 2.66). Let there be a vacuum between the plates. To set up a
potential difference V between the plates, a charge of qvac has to be sup-
plied to the plates from an external source. The capacity C vac is defined
as the charge that the capacitor can store per unit of potential difference
between the plates, i.e.,
C = qvac (2.73)
vae V

Now suppose that one placed between the plates some material which
does not conduct electronically, i.e., a dielectric material such as paraffin.
(If the material is an electronic conductor, it will allow a current to flow
between the plates. This current would hamper the storage of charge and,
hence, would frustrate simple reasoning.) In the presence of a dielectric
material between the plates, how much charge must be supplied to the
plates to set up the same potential difference V across the capacitor? It is
found experimentally that a greater amount of charge is required in the
presence of the dielectric than without it. Let this charge be quic) (Fig.
2.67). Then,
C. ~ qdie! (2.74)
die! ~ V
Charge on plates =qvac Potential difference
/across plates
V

cms
Area A cm 2

Fig. 2.66. A capacitor with vacuum between

the plates.
134 CHAPTER 2

/ Potential difference
/ between plates
V

Area A cm 2

Charge on plates =q,.;., ........if/

Fig. 2.67. A capacitor with a dielectric material between

the plates.

The ratio of the capacity Cdie1 in the presence of the dielectric to the
vacuum value Cvac defines the dielectric constant e

(2.75)

But why does the presence of the dielectric material affect the charge
necessary to sustain a potential difference V between the capacitor plates?
It is here that dipoles enter the picture. The point is that the electric field
(due to the charge at the plates) orients the dipoles in the material (Fig.
2.68). Even if there are no permanent dipoles in the material (e.g., in paraffin)
the electric field has the ability to displace the centers of negative charge
(the electron clouds) from the centers of positive charge (the atomic nuclei)
in the molecules of the material (Fig. 2.69). This deformation of the atoms
and molecules then makes them behave as dipoles as long as the field is on.
These induced dipoles do not have to be oriented; they are born aligned
parallel to the field.

iEEfe
Capacitor plates
charged

8 --
(Dill
<S)8 ~ +
+86
(!)\G
Capacitor plates Permanent dipoles Dipoles
not charged randomly oriented oriented

Fig. 2.68. The charge on the capacitor plates orients the permanent
dipoles of a dielectric which consists of polar molecules.
 ION-SOLVENT INTERACTIONS 135

1
Capacitor
~~d;;ed
+ -
+88 -
: 88=
+88 -
+ \ -
Capacitor plates Induced dipoles
not charged are oriented

Fig. 2.69. The charge on the capacitor plates induces dipoles in

the molecules of a dielectric.

The effect of the externally applied field on the dielectric material can
now be qualitatively understood. In the absence of the field, the permanent
dipoles are all arranged higgledy-piggledy; the switching-on of the field
produces a net orientation of permanent dipoles and induces temporary
dipoles. These oriented dipoles (and the induced dipoles) generate an in-
ternal field which is directed counter to the external field, i.e., to the field
which is applied from an external source via the capacitor plates. Hence,
the net field between the capacitor plates is less than the external field to
the extent that an internal opposing field is produced in the dielectric
material (Fig. 2.70). It follows that a given charge on the plates produces
a smaller potential difference (field times distance between plates) compared
with the value in the absence of the dielectric, or, conversely, a larger charge
is required to produce unit potential difference. From Eq. (2.74), it is
clear that the capacity of the condenser has been increased by interposing
a dielectric material between the plates.

G
Chorgei+
'-+ External
field
• r-
+
+
:
1- ==
Interno
• field

rouk
~Oriented
dipoles

Fig. 2.70. The orientation of dipoles in

the dielectric sets up an internal field which
is directed counter to the external field prod-
uced by the charges on the plates.
 1
136 CHAPTER 2

@
:~
Charge /:~
density,q Oriented
dipole layer

Fig. 2.71. At the interface between the

capacitor plates and the dielectric, there are
charges on the plates and an oriented dipole
layer.

2.5.2. The Relation between the Dielectric Constant and In-

ternal Counterfields

Now a relation will be derived between dielectric constant (i.e., the

increase in the capacity over the vacuum value) and the internal counter-
field which depends on the susceptibility of the dielectric material to having
its dipoles oriented and its molecules distorted into dipoles.
Consider the interface between the metallic plates of the capacitor
and the dielectric material (Fig. 2.71). Let there be a charge density q on
the plates. The externally applied field has produced a dipole layer adjacent
to the capacitor plates. This dipole layer is equivalent to two sheets of
charge (Fig. 2.72), i.e., there will be a charge density Qdipole on the plane
going through one end of the oriented dipoles and an equal but opposite
charge density on the plane through the other end of the dipoles.
The next step in the argument is to compute the net field which is set
up in the dielectric material as a result of the external field and the internal
counterfield. For this purpose, use will be made of Gauss's law which states
that the electric field normal to the surface (called the Gaussian surface)
of any volume is equal to 4n times the charge enclosed in that volume.

~H¥
+ Dipole la yer is
C-+') electrically
+ equivalent to
+- two sheets of

+"
+_+ charge
I I
q -1
+%Ipole
-Qdlpole

Fig. 2.72. The oriented dipole layer at the

plate-dielectric interface is equivalent to two
sheets of charge.
 ION-SOLVENT INTERACTIONS 137

So, all that needs to be done is to choose a Gaussian surface which

will yield the net electric field between the capacitor plates and then use
Gauss's law.
For this purpose, a brick-shaped volume is considered (Fig. 2.73). Its
location and size are of importance. Two of its faces are considered of
unit area and parallel to the capacitor plates. The dimensions of the volume
and its location are such that it encloses the charge q on the plate and the
counter charge q"ipole on one end of the oriented dipoles-the end nearest
the plates. Thus, the net charge inside the brick surface is q - qdipolc'
Hence, by Gauss's law, the field X cxt directed from the capacitor
plate toward the dielectric is

(2.76)

The question is: What is q"ipole? In other words, how does the extent
to which dipoles are oriented and induced in the dielectric depend on the
magnitude of the applied electric field? In general, qdipolc can be expressed
by a power series
(2.77)

where AI, A 2 , • .• are constants independent of Xcxt, but, for small fields,
only the first term may be taken as significant. Thus,

(2.78)

The constant Al must depend on the number n of dipoles per cubic centi-
meter and a constant which expresses how susceptible the material is to

Brick - shaped
IIOlume

Field directed towards

t-t---o'ther capacitor plate =~xt=47r(q-qdiPole)
Charge -qdipole from one end
of dipole layer
Area = Icmt
Net charge in brick shaped
volume = q-qdiPole
Charge q from
capacitor plates Capacitor planes

Fig. 2.73. The brick-shaped Gaussian surface used to compute the field
that is directed from the capacitor plates into the dielectric.
138 CHAPTER 2

having its molecules deformed into dipoles and its polar molecules oriented
by the electric field. This electric susceptibility of the material is known as
the polarizability of a molecule and designated by the symbol a. Thus,

qdipole = naXext (2.79)

where n is the number of dipoles per square centimeter.

Introducing this expression for qdipolc into the Eq. (2.76) for the field,
one has
X ext = 4nq - 4nnaXext

or
4nq
(2.80)
I + nna
Xext = 4

But the electric field between the plates is the gradient of the potential,
so, since the potential varies linearly between the plates of a parallel plate
condenser,
(2.81 )

Hence, Eq. (2.80) becomes

q I + 4nna (2.82)
V 4nd

or, by multiplying both sides by A, the area of the plates

Aq (I + 4nna )A (2.83)
V 4nd

But Aq (area of the plates times charge density on the plates) is the total
charge on the plates. Hence Aq/ V is the capacity C of the parallel-plate
condenser, which is shown in elementary books on physics to be equal to
EA/4nd. Hence, from Eq. (2.83), one has

Aq = C = ~ = (I + 4nna)A
V 4:rrd 4nd
or
E- I = 4nna (2.84)

Thus, one has obtained the fundamental relation between the dielectric
constant and the polarizability of the stuff between the capacitor plates.
 ION-SOLVENT INTERACTIONS 139

o
\ Charge an capacitor plates
tends to or ient dipole and
thermal collisions tend to
disorient dipole

Fig. 2.74. The permanent dipoles of the

dielectric are subject to an electrical orient-
ing force and a thermal disorienting force.

2.5.3. The Average Dipole Moment of a Gas-Phase Dipole

Subject to Electrical and Thermal Forces

The picture so far has been macroscopic to the extent that the atomistic
basis of the dielectric constant is concealed in the polarizability a. But how
is a related to atomistic quantities?
As a prelude to tackling a condensed medium such as water, a simpler
system will first be analyzed. Consider polar molecules (permanent dipoles)
which are so far from each other that their mutual interactions can be con-
sidered negligible. For example, consider a dilute gas of dipoles. Now, the
electric field arising from the charge on the plates tends to line up the
dipoles with their positive heads oriented toward the negative plate of the
condenser, but, at the same time, thermal collisions between the dipoles
are trying to knock them out of alignment (Fig. 2.74). Hence, the
dipoles strike a compromise between the electrical orienting force and
the thermal disorienting force. The compromise is described, following
Debye, in the Boltzmann distribution law

ne = Re- WlkT (2.85)

where ne is the number of dipoles per unit solid angle at an angle e to the
applied field (Fig. 2.75), R is a proportionality constant, and W is the work

Fig. 2.75. An example of a dipole at an

angle e
to the external field.
140 CHAPTER 2

+
Electric
+ field,X ext
+
+
+
+
dn

Fig. 2.76. The infinitesimal solid angle dQ.

done by the molecule in aligning with the field, i.e.,

W = - p'xext cos () (2.86)

Hence,
no = Rep.xex , cos 8/kT (2.87)

or the number dna of dipoles in an infinitesimal solid angle dQ to the exter-

nal field (Fig. 2.76) is
dna = ReP cos 0 dQ (2.88)

where fJ has been written instead oLp'xextikT.

Consider the dipoles oriented at an angle ()1 (Fig. 2.77). The component
of their dipole moment in the direction of the field is fl cos ()1' Hence,
these dipoles are behaving as if they only have the moment fl cos (h in the
same direction as the field. Similarly, other dipoles oriented at an angle ()2
appear as if they have a moment fl cos ()2 in the direction of the field. Thus,
the average moment (fl) of a molecule under the combined influence of
electrical and thermal forces is (by the standard formula for finding an

J:
average quantity)
fl cos () ReP cos 0 dQ
(fl) = -------
J: ReP cos 0 dQ
(2.89)

External
field

Fig. 2.77. If a dipole is oriented at an

angle 0 to the external field, it has a moment
fJ, cos 0 in the direction of the field.
 ION-SOLVENT INTERACTIONS 141

Now, It can be seen from Fig. 2.78 that dQ = 2n sin e de = -2n d(cos e).
Hence,
I:,u cos e eilcosO d(cos e)

J:
<,u) = - - - - - - -
e!1cosO d(cos e)
(2.90)

or, by writing x = cos e (so that x = 1 when e = 0° and x = -1 when

e =n),
III Jl
-1
e{Jxx dx
<,u)=---- (2.91 )
J~1 e{JX dx

For small fields, (lx = ,uXextx/kT ~ 1. Hence, one can expand efJ x by
a Taylor's series
«(lx )2
e{Jx = I + (lx + - - + ... (2.92)
2!
and omit terms higher than the second

efJx = I + (lx (2.93)

Thus,
fl xdx + fl(lx dx 2

fl dx + fl (lx dx
[x2!2]~1 + [PX:J/3]~l
[X]~l + [px2!2]~1
fj
(2.94 )
3

dll Circumference of this

circle is 271"sin8

Rodius=sin8
-r----.h~h+~---.E

Fig. 2.78. Arriving at an expression for

the infinitesimal solid angle ~.n
 142 CHAPTER 2

~
ChOrge+4f

Chorge-e - + -d-
e :<)U>
Fig. 2.79. The average moment <f1> of a
gas dipole.

or
(2.95)

This is the average or effective moment which a gas dipole exhibits in the
direction of the weak t external field when it is subject to electrical orienting
and thermal randomizing forces.

2.5.4. The Debye Equation for the Dielectric Constant of a Gas

of Dipoles

The average moment (11) of a gas dipole in the direction of the field
may be considered to arise from charges +e and -e separated by a distance
£I (Fig. 2.79)
(11) = ed (2.96)

Suppose that one considers (Fig. 2.80) a brick-shaped volume in the

bulk of the dielectric. Let the volume be I cm2 in area and d cm in thick-
ness. If the gas contains n dipoles per cubic centimeter, then the brick-
shaped volume (£I cm3 ) can enclose nd dipoles. Further, the charge density
ql on a plane going through the dipole ends will be equal to the number
of dipoles per square centimeter (= nd) times the charge per dipole (= e);
i.e.,
(2.97)

and, inserting the expression (2.95) for (11), one has

(2.98)

An identical argument is valid at the interface between the capacitor

plate and the dielectric, which means that the orientation of dipoles
under electrical forces and thermal motions sets up a charge density ql

t The qualifying word weak is introduced here in the light of the approximation contained
in Eq. (2.93).
 ION-SOLVENT INTERACTIONS 143

Averaoe moment =<,.u>in"the

direction of field

~
dcms
Choroe on each end
of dipole =(6 )

Brick:shalled .volume
This plane ooes contams lid dipoles
through the I lema
char!MSan one -
end of dipoles
Charge density on this plane-
numl)er of dipole) z charge
per dipole=~=.lJdi
olume=dcm'

Fig. 2.BO. The net orientation of dipoles produces a

charge density q on a plane in the dielectric.

on a plane parallel to the capacitor plate. This charge q1 is opposite in

sign to that externally set up on the plates, i.e., to the charge which
causes q1'
It will be recalled however that the basic equation for the dielectric
constant [Eq. (2.84)] was derived on the basis that a charge density of
qdipole was produced on a plane adjacent to the capacitor plates. Is
q1 = qdiPole? No, because one must also consider the contribution from the
induced dipoles produced by the distortion of molecules subject to an
electrical field; the charge q1 only takes into account the effect of the per-
manent dipole moment of the molecules of the dielectric. The average
moment induced in a nonpolar molecule will be considered to contribute
a charge density q2 to the plane adjacent to the capacitor plates, so that

(2.99)
or, from Eq. (2.79),

(2.100)

One can substitute for q1 in this equation Eq. (2.98), obtaining

2
a=~+~ (2.101)
nXext 3kT

In this expression, the total polarizability a and the number of mole-

cules per cubic centimeter are constant for a given field; so is fJ2j(3kT) at
a fixed temperature. Hence, Q2j(nXext ) must also be a constant, and one
144 CHAPTER 2

can write
a = adeform + aorient (2.102)
fl2
= adeform + 3kT (2.103)

where adeform is the deformation polarizability (= Q2/nXcxt) and is a

measure of how the molecules deform under electric fields and become
induced dipoles. The term aoricnt is the orientation polarizability and reveals
how successful a given electric field is in producing a net dipole orientation.
The expressions (2.1 02) and (2.103) for the total polarizability a of
the dipole gas can be inserted into the fundamental relation (2.84) between
the polarizability and the dielectric constant. Thus,
c - 1= 4nnadeform + 4nnaorient (2.104)

(2.105)

Thus did Debye link the macroscopic property of dielectric constant

to a molecular property, namely, the dipole moment of the gas molecule
(cf Fig. 2.81).

/
0.004'- !
r
x
/'
I
0003- I
/
~ -I
/
/
/
0.002- /
/
/
/
I
0.001f- /
/
1/
°0~----~0~~0~01~--~0~.10=0=2--~0.003
lIT ('K' -I)

Fig. 2.81. The variation of the dielectric

constant of water vapor with temperature is
as predicted by the Oebye equation.
 ION-SOLVENT INTERACTIONS 145

There are two important features of this expression (2.105). Firstly,

temperature affects the dielectric constant by affecting the orientation of
dipoles; the higher the temperature, the less successful is the externally
applied field in opposing the thermal disorienting motions and producing
a net orientation of the dipoles. So it was a wise decision to treat the dielec-
tric constant as a function of temperature in differentiating the Born free
energy of solvation to obtain the entropy of solvation (see Section 2.2.5).
Secondly, as the temperature increases, the orientation polarization
4;rrnfl2/(3kT) and, thus, the dielectric constant decreases. Eventually, when
the temperature is large enough,

and
f - I """ 4nnadeform (2.106)

Things have become too hot for the dipoles to have any net orientation
at all; they become randomly oriented. But there still remains deformation
polarization due to induced dipoles, and this is what then decides the
dielectric constant.

2.5.5. How the Short-Range Interactions between Dipoles Af-

fect the Average Effective Moment of the Polar Entity
Which Responds to an External Field

The Debye theory for the dielectric constant of a gas of permanent

dipoles invites an obvious comment. It may be satisfactory for a gas of
water molecules t which are not involved in mutual interactions, but what
has it to do with the dielectric constant of liquid water with its network
structure and of ionic solutions containing solvated ions? One must review
some of the basic assumptions in the treatment of a gas of dipoles and see
to what extent they would have to be changed in discussion of the dielectric
constant of an associated liquid.
Firstly, it was assumed that there were no interactions between the
dipoles in the vapor. This assumption is obviously untenable in liquid
water. Water is a broken-down form of ice. It is quasi crystalline in the
sense that there are in liquid water large groups ("icebergs") of water
molecules associated by hydrogen bonding. In each of these large groups,

t For the consideration of the dielectric constant of water, it is not necessary to take
into account the quadrupole character of a water molecule; it is quite adequate to
consider a water molecule as it were an electrical dipole.
146 CHAPTER 2

Fig. 2.82. The tetrahedral unit of a central

water molecule linked to four neighboring
water molecules by hydrogen bonds.

an important structural unit may be distinguished. This structural unit or

subgroup consists of a central water molecule tetrahedrally linked to four t
other molecules by hydrogen bonds (Fig. 2.82).
How does such a subgroup respond to an electric field? When the
electric field tries to align the central molecule of a tetrahedral unit, it has
to reckon not only with randomizing thermal motions but also with the
coordinating water molecules which seek to maintain definite orientation
relationships with the central molecule, i.e., the orientations characteristic
of the tetrahedral subgroup with hydrogens facing oxygen lone pairs, etc.
So, if any aligning occurs, the whole subgroup has to alignt (Fig. 2.83).
What matters, therefore, is the moment flgroup of a subgroup of water
molecules, not the dipole moment fl of an isolated water molecule. The
effective moment of the group as a whole is equal to the dipole moment
of the central molecule plus the components of the dipole moments of the
four neighboring water molecules of the tetrahedral unit (these compo-
nents being taken along the moment of the central molecule). In other
words, the effective moment flgroup is the vector sum of the dipoles in the
group, i.e.,
flgroup = fl + g(fl cos y)

= fl(i + g cos y) (2.107)

where g is the number of nearest-neighbor water molecules linked with

the central molecule and cos y is the average of the cosines of the angles

t In ice, the number of coordinating water molecules is exactly four; in water, it is 4.4
to 4.6 (cf Table 2.11). For simplicity, this distinction will not be stressed.
t It has been suggested that the icebergs present in water contain (at 20°C) about 50
to 60 water molecules with hydrogen bonding effective throughout the group. Never-
theless, the subgroup of four neighboring water molecules plus a central water molecule
is considered in this discussion the orienting entity because the central molecule is
regarded as exerting short-range forces, apart from hydrogen bonding, only on its
nearest neighbors, i.e., on four water molecules.
 ION-SOLVENT INTERACTIONS 147

When this water

ma lecule wants
ta align with the
external field the
whole cluster must
align

Fig. 2.83. When the central molecule of

the tetrahedral group tries to align with the
external field, the whole group aligns.

between the dipole moment of the central water molecule and those of its
bonded neighbors.
For example, if the cluster consists of the tetrahedral group, then,
g = 4 and cos y = 1/3, in which case I + g cos y = 7/3. That is, owing
to the interlinking of the water dipoles, the effective dipole moment of the
aligning group is 7/3 times the dipole moment of a free water molecule.
Once the effective moment of a subgroup is computed, one can go
through the Oebye argument saying that thermal motions oppose the align-
ment of subgroups. After writing fJ = flgroIlPX/(kT), the average moment
of a dipole cluster is given by [cf. Eq. (2.95)]

<flgroup >= fl2( I + g COS Y )2 X

(2.108)
3kT

A simple test of this equation is to examine what happens when the

subgroups break up, say, by a temperature increase. The correlated align-
ments of the water molecules are diminished, and the dipoles become inde-
pendent. The quantity cos y tends to zero, t i.e., the various neighboring
dipoles contribute a zero average component along the particular central
dipole. One gets back the simple Oebye equation (2.95) for the average
dipole moment.

2.5.6. The Local Electric Field in a Condensed Polar Dielectric

From the effective moment of a group of dipoles which orient as one
unit (Eq. 2.108), one can obtain the orientation polariza bility thus

qI / ,u grollI') fl2( 1 + g COS)!)2 X

U· t = - - = - - - - =
nX Xcxt (2.109)
arlen ext 3kT

t The quantity cos ( is averaged over those neighboring molecules which hold an orien-
tation relationship with the central molecule. When thermal forces destroy the orien-
tation relationship, the average of cos ( is zero, i.e., the neighboring water molecules
are orienting independent of the central molecule.
148 CHAPTER 2

~Reference Reference
V dipole ddiPole

+
fontinuum

is equivalent to is equivolenllo

Reference
dipole Cavity Covity

Fig. 2.84. The approximately spherical cavity into which a reference dipole or group
of dipoles is caged.

where Xext is the externally applied field. But what is X, the electric field
operating on the central molecule and the group? Is it simply equal to the
external field Xext ?
These questions bring up the second main difference between a gas
of free, noninteracting dipoles and a liquid with dipoles associated into
subgroups which are part of networks. In the gas phase, all that a water
molecule feels is the external field emanating from the capacitor plates-the
other water molecules are too far away to exercise any influence. In the
liquid phase, however, any water molecule of group of molecules is in
"contact" with other water molecules. So, what matters to a given water
molecule (or group of water molecules) is the true local field X1oc , which
is the sum of the external field and the field of the surrounding particles on
the given molecule (or group). Thus, the electric field X in the expression
for the orientation polarizability is the local field and not the external field
and should therefore be replaced by X1oc '
Onsager showed an interesting way of calculating the local field X1oc '
The thinking proceeds thus: To a close approximation, a reference dipole
or a group of interlinked dipoles is caged by the surrounding particles in
a spherical hole (Fig. 2.84). So the local field operating on the reference
dipole (or group) is the field in the empty spherical cavity owing to the
externally applied field and the surrounding molecules which are partly
oriented by the field.
Now, to compute the field in the cavity, consider two regions (Fig.
2.85): one outside the cavity (region I) and one inside it (region II). Let
the center of the cavity be arbitrarily taken as the origin of a polar coor-
dinate system and as the point to which all potentials are referred, i.e.,
as the zero of potentials.
 ION-SOLVENT INTERACTIONS 149

E xte rna I field

Fig. 2.85. The computation of the field

in the spherical cavity.

The procedure will consist in obtaining expressions for the potential

in region I and region II. At the surface of the sphere, the spatial coordinates
are the same, and the potentials, as expressed for region I and region II,
must be the same and can hence be equated.
The potential !Pn at a point Z (with coordinates rII and e) inside the
e,
cavity, i.e., in region II, is simply equal to - Xlocrn cos where X loc is the
local field t
!Prr = - X10cr rr cos e (2.110)

The potential !PI at a point Y(rI, e) is a more interesting matter. It

is equal to the sum of the potentials due to the external field (this potential is
- Xextr/ cos fJ) and, what may appear surprising, that field due to the empty
spherical cavity.
Why should an empty cavity exert any influence on the surroundings?
Because, whenever a dielectric material is placed in an electric field, charges
appear on the boundary (or surface) of the material The cavity is an internal
surface, so there are charges on its surface. It is these charges which exert
an electrical influence on the surroundings.
Another point about these charges is that they are always set up so
as to oppose the externally applied field. The distribution of these charges

t The minus sign comes in because the potential decreases in the direction of an electric
field, i.e., "down" the field. Thus, any point has a potential which is negative relative
to that at a point farther "up" the field.
150 CHAPTER 2

External electr ic f e
i ld

is equivalent is equivalent
to to

Dipoles
Chorges Dielectric
on cavity Resultant
surface dipole

Fig. 2.86. The electrical effect of a cavity is equivalent to the effect of a single dipole.

looks like a series of dipoles all pointing in the direction of the field (Fig.
2.86). In fact, one can conceptually replace all these dipoles by one single
dipole of moment m located at the center of the cavity. In conclusion,
therefore, the electrical effect of an empty spherical cavity (in a dielectric
which is placed in an externally applied field) is equivalent to the effect of
a hypothetical dipole of moment m located at the center of the cavity.
Thus, the spherical cavity sets up a potential m(cos ()/rl (the usual ex-
pression for the dipole potential) at the point Y(rI' () in region I.
The total potential CPI at Y (in Fig. 2.85) is given by the sum of the
potential due to the external field, which is - XextrI cos (), plus that due to
the cavity, which at Y is m(cos ()/rI2 ,
m cos ()
CPI = - XextrI cos () + rI 2 (2.111 )

Now that expressions for CPr and CPu have been obtained, the situation
at the surface of the cavity can be considered. Here, rr = rII = a, the radius
of the sphere, and, thus, CPr = CPIh i.e.,

- Xexta cos () + m cos

a
()
2 = - X\oca cos ()

or
m
X\OC = X ext - -3 (2.112)
a

There is a further condition that must be satisfied at the cavity surface.

The electric field normal to the hole surface in the two opposite directions
 ION-SOLVENT INTERACTIONS 151

Field towards
cavity center= - r - _......

~
dr

Fig. 2.87. A Gaussian surface for obtaining the elec-

tric fields normal to the cavity surface.

(toward and away from the cavity center) must be equal. To get at these
fields, one invents an imaginary Gaussian surface shaped like a pillbox
and containing some charge q (Fig. 2.87). The field away from the cavity
is -dq;ddr, and, from Gauss's law, one has

-dq;r 4nq
~ e
or
(2.113)

The field toward the cavity center (Fig. 2.87) is

_ dq;" = 4nq (2.114)

dr

(e being unity because, inside the cavity, there is vacuum). Hence, at the
surface of the cavity,

(2.115)
152 CHAPTER 2

which, from Eqs. (2.110) and (2.111), is

X 10c = cXext + 2cm

-3-
a
(2.116)

From the simultaneous Eqs. (2.112) and (2.116), one has

3c
X10c = 2c + 1 Xext (2.117)

Hence, the local field in the cavity, which operates on the dipole cluster,
has been evaluated. Since c > 1 (i.e., 2c + 1 < 2c + c or 2c + 1 < 3c),
it is obvious that the local field in condensed media is always greater than
the externally applied field. This means, then, that a given dipole is subject
to a stronger electrical orienting force when it is inside a liquid than when
it is in a dilute gas.

2.5.7. The Dielectric Constant of Liquids Containing Associated

Dipoles

It is now easier to obtain an expression for the dielectric constant of a

liquid, such as water, in which the polar molecules are associated into sub-
groups and networks.
One starts from the fundamental expression

c - 1 = 4nna (2.84)

which, after substituting for a with

a = qdipole =~+~ (2.100)

nXext nXext nXext
becomes
c- 1 = 4nn ( --+--
ql
nXext
q2 )
nXext
(2.118)

The ql term which arises from orientation polarization is elaborated

as follows. Since [ef (2.97)]
(2.119)

one can use Eq. (2.108) to write

(2.120)
 ION-SOLVENT INTERACTIONS 153

or, introducing the expression (2.117) for the local field,

(2.121 )

Hence,
ql /12(1 + gCOSy)2 3E
(2.122)
nXext = 3kT 2E +I
The second term in Eq. (2.118) is connected with the deformation
polarization. Following the procedure of Section 2.5.4, one argues that q2
is proportional to the number n per unit volume of deformable molecules
and to the electric field operating on them (the proportionality constant
being adeform). In the gas phase, the field operating on the deformable
molecules is the external field Xext ; in condensed media, it will be the local
field X1oc . Thus,
(2.123)

Once again, the expression (2.117) for X10c can be introduced to give

(2.124)

or, by rearrangement,
q2 3E
--nx:-
ext
= adeform 28 +1 (2.125)

Now, the expressions for ql/nXext and Q2/nXext can be inserted into
Eq. (2.118) to give
3E [ /12 (l + g cos y)2 ]
E - 1 = 4nn 2E +1 adeform + 3kT (2.126)

It is conventional, however, to write this in a slightly altered form as

(E - 1)(28 + 1) _ 4nn [ /12 (1 + gcosy)2 ]

98 - -3- adeform + 3kT (2.127)

This is the Kirkwood equation for the dielectric constant of a con-

densed medium. It takes into account the short-range interactions between
polar molecules which lead to the formation of molecular groups orienting
as a unit under the influence of electric fields. It also considers the actual
local field, as distinct from the externally applied electric field, operating
on the orienting entities.
154 CHAPTER 2

Table 2.21
Dielectric Constant of Liquid Water at Various Temperatures

Temperature, Calc Obs

o 84.2 88.0
25 78.2 78.5
62 72.5 66.1
83 67.5 59.9

From the Kirkwood equation, it is clear that the association of dipoles

arising from short-range forces is a very important factor in determining the
dielectric constant of a liquid. The linking together of dipoles increases g,
the number of dipoles which are nearest neighbors to a reference dipole,
and thus increases the dielectric constant.
In using the Kirkwood equation (2.127), it is necessary to know g,
the number of dipoles which are nearest neighbors or coordinate in par-
ticular to a reference dipole. This quantity g can be obtained from X-ray
data, and, for liquid water, it is 4.4 to 4.6. Based on this value, there is
good agreement (Table 2.21) between the observed values of dielectric
constant and those calculated by the Kirkwood equation.
To illustrate the influence of dipole association on the dielectric con-
stant, consider liquid water and S02. The dipole moments of the S02 and
H 2 0 molecules are almost the same, but their dielectric constants are quite
different (Table 2.22). This difference arises because, in water, there are
groups of strongly interacting, hydrogen-bonded water molecules, whereas
such association of S02 molecules does not occur.
In the light of the above discussion, hydrogen-bonded liquids should
in general be expected to have high dielectric constants compared with
those of nonassociated liquids, and, indeed, this is the case (Fig. 2.88).

TABLE 2.22
The Dipole Moments and Dielectric Constants of Liquid Water and Sulfur
Dioxide

Dipole moment, Dielectric constant,

debyes at 25°C

H.O 1.85 78.5

SO. 1.67 12.35
 ION-SOLVENT INTERACTIONS 155

120

100

:2
~
e Hz02
.~
80 eHp
'0
C eHF
0
~0 60
u
u
c:;.,
~

40
Qj
0
20

AStiJ
0

Fig. 2.88. The dielectric constants of liq-

uids as a function of their dipole moments
(. 's represent unassociated liquids, and. 's
represent H-bonded liquids).

2.S.S. The Influence of Ionic Solvation on the Dielectric Constant

of Solutions

Some indication has now been given of how the association of dipoles
and the local field can increase the dielectric constant into the range of
several tens compared with values of 5 to 10 for unassociated liquids. An
elementary, quantitative picture of the dielectric constant of pure water
has also been presented. The next question is: What happens to the dielec-
tric constant when ions enter an associated liquid such as water? In other
words, how does the dielectric constant of an ionic solution differ from that
of pure water?
Recall the picture of the structure of water in the presence of ions
(Section 2.3.2). Far from an ion, the water structure is normal and the
usual bulk dielectric constant would obtain. In the secondary region, the
existence of structure breaking implies that there is a decrease of the para-
meter g of the Kirkwood equation (2.127), i.e., there is a decrease in the
number of water molecules linked to any particular water molecule. But,
as g decreases, the dielectric constant also decreases. This implies that the
156 CHAPTER 2

dielectric constant in the secondary region falls below the bulk value. It
will be, depending on the distance from the ion, somewhere between the
bulk value ('"'-'80) and the value in the solvation sheath, which turns out
(see below) to be about 6.
The situation in the primary solvation sheath is interesting. The solvent
dipoles here are oriented to the ion and are so firmly fixed in this orientation
that they are almost insensitive to the external field and to the field of the
surrounding water molecules. As far as the external and local fields are
concerned, they are so bound to the ions as to be unorientable. Thus, the
water molecules in the primary solvation sheath hardly contribute to the
orientation polarizability of the solution. Their main contribution is to the
deformation polarizability, and, hence, the dielectric constant of this
primary solvent water is very much lower than the bulk value. Measure-
ments of the dielectric constant of water at alternating field frequencies so
high that the dipoles are too sluggish to align with the alternating field
suggest that completely bound water has a dielectric constant of about 6.
So one has a picture of the dielectric constant of water being very
low ('"'-'6) in the primary solvation sheath around an ion and then rapidly
increasing till the bulk value ('"'-'80) is attained outside the structure-break-
ing region (Fig. 2.89). It will be seen later (Section 7.4.24b) that one should
reckon with a similar variation of dielectric constant near a charged elec-
trode adjacent to which there is a layer of oriented water.

80r---------------------~~---------.

70

60

50

40

30

20

10

r in 1.
Fig. 2.89. The variation of the dielectric constant around
an ion.
 ION-SOLVENT INTERACTIONS 157

TABLE 2.23
Dielectric Constants of Aqueous Solutions of Electrolytes t

Electrolyte Dielectric constant

LiCl 66 ±2
NaCl 69 ±2
KCl 70 ±2
RbCl 70 ±2
NaF 68 ±2
KF 67 ±2
NaI 65 ±2
KI 64 ±2
MgCl 2 50 ±2
BaCl 2 52 ±2
LaCl a 36 ±2
Na 2 SO. 58 ±2

t In all cases, the values pertain to solutions containing a mole of electrolyte.

A simple conclusion follows from the variation of dielectric constant

in the neighborhood of an ion. The dielectric constant of an ionic solution
must be determined partly by the dielectric constant of the water which is
unaffected in its structure by the presence of ions and partly by the dielectric
constant of the water in the primary and secondary solvation sheaths around
ions. But, the dielectric constant of the water both in the primary solvation
sheaths and in the structure-breaking secondary region is lower than the
value for the pure solvent. Hence, the dielectric constant of a solution must
be lower than that of the pure solvent (Table 2.23). Further, the depression
of the dielectric constant depends on the concentration of the ions-the
higher the ionic concentration, the lower the dielectric constant is (Fig. 2.90).

i:
.E
III
C
860
u
:E
j!
QI

o 40=-----~------~----~
o 6
Concentration (mole litre-')of NaCl

Fig. 2.90. The variation of the dielectric

constant of NaCI solutions with electrolyte
concentration.
 158 CHAPTER 2

Further Reading
1. L. Onsager, J. Am. Chem. Soc., 58: 1486 (1936).
2. J. G. Kirkwood, J. Chem. Phys., 7: 911 (1939).
3. G. Oster, and J. G. Kirkwood, J. Chem. Phys., 11: 175 (1954).
4. L. Pauling, The Nature of the Chemical Bond, 3rd ed., Chap. 12, Cornell
University Press, Ithaca, N.Y., 1960.
5. A. Prock and G. McConkey, Topics in Chemical Physics, Chap. 1, Elsevier
Publishing Co., New York, 1962.
6. R. H. Cole, J. Chem. Phys., 39: 2602 (1963).
7. N. E. Hill, Trans. Faraday Soc., 69: 344 (1963).
8. R. P. Feynmann, R. B. Leighton, and M. Sands, The Feynmann Lectures on
Physics, Vol. II, Chap. 10, Addison-Wesley Publishing Co., Inc., Reading,
Mass., 1964.
9. W. Dannhauser, and L. W. Bahe, J. Chem. Phys., 41: 2666 (1964).
10. B. E. Conway, and R. G. Barradas, Chemical Physics of Ionic Solutions,
John Wiley & Sons. Inc., New York, 1966.
11. C. P. Smyth, Ann. Rev. Phys. Chem., 17: 433 (1966).

2.6. ION-SOLVENT-NONELECTROLYTE INTERACTIONS

2.6.1. The Problem

The picture that has emerged in this chapter is of ions interacting with
the solvent and producing the interesting effects that go under the name
solvation. The quasi-lattice structure of the solvent is not the same after
the ions have entered it. Some of the water molecules are wrenched out of the
quasi lattice and appropriated by the ions as part of their primary solvation
sheaths. Even farther away in the secondary solvation sheaths, the ions
produce telltale effects of structure breaking.
What happens if, in addition to ions and water molecules, molecules
of nonelectrolyte are also present in the system? Or, what will occur if,
to a solution already containing nonelectrolyte molecules at saturation
concentration, ions are added?
One thing is certain: there will be less free water to dissolve the non-
electrolyte, because some of it will be removed from the solvent into the
solvation sheath of the ion. This means that the nonelectrolyte molecules
find themselves suddenly having much less water to associate with, and some
of them will hence shun the loneliness imposed by the water's preference
for ions, and reassociate themselves with their parent lattice, i.e., precipitate
out. This is the origin of a term from technological organic chemistry-
"salting out." Occasionally, however, the ions are deviants, and associate
 ION-SOLVENT INTERACTIONS 159

preferably with the nonelectrolyte solute, shunning the water. In the rarer
instances where these deviants appear, there is a rapid departure of the
nonelectrolyte from the parent lattice, and thus the solubility of the former
is enhanced. One speaks of salting in.
Two aspects of the theory of salting out-the normal case-are con-
sidered below. Firstly, the effects of the primary solvation sheath have to
be taken into account-how the requisition of water by the ions causes
the nonelectrolyte's solubility to decrease. Secondly, the effects of secondary
solvation (interactions outside the solvation sheath) are calculated.

2.6.2. The Change in Solubility of a Nonelectrolyte Due to

Primary Solvation
It is very easy to calculate this change of solubility, so long as there are
data available for the solvation numbers of the electrolyte concerned. Let
it be assumed that the normal case holds true, and the ions are solvated
entirely by water molecules. Then (cf Section 3.6.2) recalling that one liter
of water contains 55.55 moles, the number of water molecules left free
to dissolve nonelectrolyte after the addition of ions is

(2.128)

where Ci is the number of gram moles of the electrolyte [-I, and ns is the
solvation number of the electrolyte concerned.
Assuming at present that the solubility of the nonelectrolyte is simply
proportional to the number of water molecules outside the hydration sheath,
then
55.55 - Cins
(2.129)
55.55
or
s = S _ SOCins (2.130)
o 55.55

where So is the solubility of the nonelectrolyte before addition of electrolyte

and S that after it. Then, suppose ns = 6, Ci = 1M; then

(2.131 )

Correspondingly, S/So is 0.89. For comparison, with KCI0 4 as electrolyte,

and 2,4-dinitrophenol as nonelectrolyte, the corresponding experimental
value (extrapolated) is about 0.7. Some further effect must be taken into
account.
160 CHAPTER 2

2.6.3. The Change in Solubility Due to Secondary Solvation

It has been stressed in Section 2.4 that solvation is a far reaching phe-
nomenon, although only the primary solvation number can be determined
significantly (if by no means accurately, cf Table 2.19). Correspondingly,
there certainly are effects of ions on the properties of their solvent which
lie outside the radius of the primary hydration sheath. These effects must
now be accounted for, in so far as they relate to solubility of a nonelectro-
lyte. Let the problem be tackled as though no primary solvation had with-
drawn water from the solution. One can write

(2.132)

where nNE,r is the number per unit volume of nonelectrolyte molecules at

a distance r from the ion, nNE,b is the same number in the bulk, and the free-
energy change L1Go is the work Wr done to remove a mole of water molecules
and insert a mole of nonelectrolyte molecules at a distance r from the ion
outside the primary solvation sheath.
A naive calculation would run thus: The field Xr of the ion at a distance
r falls off with distance according to Coulomb's law t
x = ZieO (2.133)
r Er2

If, now, a dipole (aligned parallel to the ionic field) is moved from infinity
where the field Xr = 0 through a distance dr to a point where the corre-
sponding field is dX, the elementary work done is - f1 dX.
Thus, the work to bring a mole of nonelectrolyte molecules from infinity
to a distance r is - N,4 f;r f1NE dX, and the work done to remove a water
molecule to infinity is N,4 f;r f1w dX. The net work of replacing a water
molecule by a nonelectrolyte molecule would therefore be given by

Wr =
xr
N,4 (J ° f1w dX - ° f1 NE dX )
JXr (2.134)

Now, the reader will probably be able to see there is a flaw here.
Where? The error is easily recognized if one recalls the Debye argument
for the average moment of a gas dipole. For what is the guarantee that a

t Notwithstanding the considerations of Section 2.5.B, the use of the bulk dielectric
constant of water for dilute solutions of nonelectrolyte is not very inaccurate in the
region outside the primary solvation sheath. The point is that, in this region (i.e., at
distances > 5 to to A from the ion's center), there is negligible structure breaking
and, therefore, negligible decrease of dielectric constant from the bulk value.
 ION-SOLVENT INTERACTIONS 161

water dipole far from the ion is aligned parallel to the ionic field? What
about the thermal motions which tend to knock dipoles out of alignment?
So what matters is the average dipole moment of the molecules in the direc-
tion of the ionic field. Thus, one has to follow the same line of reasoning
as in the treatment of the dielectric constant of a polar liquid and think
in terms of the average moment <ft>
of the individual molecules, which
will depend in Debye treatment on the interplay of electrical and thermal
forces and in Kirkwood treatment also on possible short-range interactions
and association of dipoles. One has therefore

(2.135)

and, by making use of relations (2.79) and (2.97), which results in t

<ft> = aX (2.136)
one has
W1' = NA(J:'~r awX dX - f;r aNEX dX)
N,4 (aw - aNE)X/
(2.137)
2

This expression for the work of replacing a water molecule by a non-

electrolyte molecule at a distance r from an ion can now be introduced
into Eq. (2.132) to give (in number of nonelectrolyte molecules per unit
volume)
(aNE - aw)X/ ]
nNE" = nNE,b exp [ 2kT (2.138)

The exponent of Eq. (2.138) is easily shown to be less than unity at

25°C for most ions. Thus, for distances outside the primary hydration
shell of nearly all ions, one can expand the exponential and retain only
the first two terms, i.e.,
(aNE - aw)X/ ]
nNE" = nNE,b [ I + 2kT (2.139)

Thus, the excess number (not moles) per unit volume of nonelectrolyte
molecules at a distance r from the ion is (again, in number of molecules

t The polarizability a used here refers to the orientation polarizability (see Section 2.5.4).
Far away from the ion, the factor of deformation polarizability can be ignored.
162 CHAPTER 2

per unit volume)

(2.140)

But this is only the excess number of nonelectrolyte molecules per

unit volume at a distance r from the ion. What is required is the total excess
number per unit volume throughout the region outside the primary solva-
tion sheath, i.e., in region 2. One proceeds as follows: The excess number,
not per unit volume, but in a spherical shell of volume 4nr2 dr around the
ion is

and, therefore, the total excess number of nonelectrolyte molecule per ion
in region 2 is (with rh equal to the radius of the primary hydration sheath)

(2.141 )

If one sets Xr = zieO/(er2), then, the excess number of nonelectrolyte mole-

cules caused to be in solution per ion is

Hence, the excess number of nonelectrolyte molecules in a real solution

containing Ci moles liter- 1 of binary electrolyte is given by Eq. (2.142),
multiplied by the number of moles per cubic centimeter in the solution,
namely NAc;/lOOO. The expression is t

(2.143)

where nNE is the number of nonelectrolyte molecules per cubic centimeter

in the solution after addition of the electrolyte.

t A factor of 2 has been removed from the denominator of this equation, compared with
Eq. (2.142), because there are two ions in the binary electrolyte, each of which is assumed
to give the same effect on the solubility.
 ION-SOLVENT INTERACTIONS 163

Hence
nNE - nNE,b = NAc; [4n(Zi eo)2(aNE - aw ) ]
(2.144 )
nNE,b 1000 e2kTrh

The terms nNE and nNE,b are numbers of nonelectrolyte molecules

per cubic centimeter, Hence, NAnNE = Sand NAnNE,b = So, where these
terms have been defined in Section 2,6.2. It follows that

(2.145)

This equation shows up well the sign of the effect of the secondary
solvation. From (Section 2.5.4), and Eqs. (2.79) and (2.103) it is seen that
the orientation polarizabilities, aNE and aw, are largely dependent on the
square of the permanent dipole moments of the molecules. Water has, in
comparison with many nonelectrolytes, the highest dipole moment. When,
thus, aw > aNE, S is less than So and there is salting out. HCN is an ex-
ample of a substance the dipole moment of which is greater than that of
water. (It masquerades as a nonelectrolyte because it is little dissociated
in aqueous solution.) Appropriately, HCN is often salted in.

2.6.4. The Net Effect on Solubility of Influences from Primary

and Secondary Solvation

Equation (2.130) can be written as

(2.146)

Equation (2.145) can be written as

(2.147)

The treatment of the effect of the secondary solvation has assumed

that the primary solvational effects do not exist. In fact, the secondary
solvational effects work on the diminished concentration of nonelectrolyte
which arose because of the primary solvation. Hence,

(2.148)

(2.149)

(2.150)
164 CHAPTER 2

S as written in (2.148) has taken into account the primary and second-
ary solvation and can be identified with the solubility of the nonelectrolyte
after addition of ions to the solution. Hence,

(2.151 )

If one writes
(2.152)

the kl is called Setchenow's constant.

There is fair agreement between Eq. (2.152) and experiment. If the
nonelectrolyte has a dipole moment less than that of water, it salts out.
In the rare cases in which there is a dipole moment in the nonelectrolyte
greater than that of water, the nonelectrolyte salts in.
Salting out has practical implications. It is part of the electrochemistry
of everyday industrial life. One reclaims solvents such as ether from aqueous
solutions by salting them out with NaCl. Salting out enters into the produc-
tion of soaps and the manufacture of dyes. Detergents, emulsion poly-
merization (rubber), and the concentration of antibiotics and vitamins
from aqueous solutions, all depend in some part of their manufacture upon
salting in. But the salting in with which they are associated is not the rare
deviant phenomenon arising when the dipole moment of the nonelectrolyte
is greater than that of water. It possesses the characteristic of always being
associated with organic electrolytes in which the ions are big. Why do such
ions (the dipole moment of which is less than that of water) cause "anoma-
lous salting in"?

2.6.5. The Case of Anomalous Salting in

The picture given above seems satisfactory as a first approximation and
for dilute solutions, but it has this one disturbing feature, namely, there are
situations where theory predicts salting out, but experiment shows salting in.
In such cases, the theory seems to favor ions' being surrounded by water,
whereas, in fact, nonelectrolyte seems to accumulate near ions. Now, it
will be recalled that only ion-dipole forces have been reckoned with in
treating the interactions among the particles populating the primary sol-
vation shell. Thus, the ion-water and ion-nonelectrolyte forces have been
considered to be of the ion-dipole type of directional forces which orient
polar particles along the ionic field. Perhaps this restriction is too severe,
for there are also nondirectional forces, namely, dispersion forces.
 ION-SOLVENT INTERACTIONS 165

These dispersion forces can be seen classically as follows: The time-

average picture of an atom may show spherical symmetry because the
charge due to the electrons orbiting around the nucleus is smoothed out
in time. But an instantaneous picture of, say, a hydrogen atom would show
a proton "here" and an electron "there"-two charges separated by a
distance. Thus, every atom has an instantaneous dipole moment; of course,
the time average of all these oscillating dipole moments is zero.
Then, an instantaneous dipole in one atom will induce an instantane-
ous dipole in a contiguous atom, and an instantaneous dipole-dipole force
arises. When these forces are averaged over all instantaneous electron
configurations of the atoms and thus over time, it is found that the time-
averaged result of the interaction is finite, attractive, and nondirectional.
Forces between particles, which arise in this way, are called dispersion
forces.
The dispersion forces give rise to an interaction energy which has the
same distance dependence as that due to dipole-dipole interactions. That
is, the potential energy varies as ,-6 and may be written as }../r 6 , where A is
a constant independent of r. The rapid decrease of such forces with increase
of distance from the origin makes it unnecessary to consider dispersion
interactions outside the primary solvation shell; they are by then already
too feeble to warrant consideration. Inside the primary hydration sheath,
the dispersion interaction can be treated in the same way as the ion-dipole
interaction. That is, in the replacement of a water molecule by a non-
electrolyte molecule, one must take into account not only the difference
in ion-dipole interactions, but also the difference between the dispersion
interactions. Thus, one must now reconsider the situation in which the
ion interacts with water molecules. The picture of the last three sections was
that the water would usually be the entity to be predominantly attracted to
the ion, so that the amount of water available for the dissolution of the
nonelectrolyte would be reduced, and its solubility consequently would fall.
The only exceptions which were recognized for this were those unusual
cases in which the nonelectrolyte had a dipole moment greater than that of
water. As to the ion, only its radius and charge played a part in the matter:
it did not influence the situation in any more structural way, e.g., in terms
of its polarizability, etc.
Now, however, with the introduction of dispersive force interactions
in the competition of the water and the nonelectrolyte for ions, these
considerations may have to be modified. On what molecular features do
dispersion forces depend? What relative attractiveness has a given ion,
on the one hand for a water molecule, and on the other for a nonelectrolyte?
168 CHAPTER 2

Equations for the dispersive force interactions have been worked out
for interactions in the gas phase. A simple equation would be

u= hvad-ionad-mOl
2R6

where the v is the frequency of vibration of the electron in its lowest energy
state, the a's are the distortion polarizabilities of the entities indicated, and
R is their distance apart. It must be noted at once that the polarizability
indicated here, the distortion polarizability, differs from that which enters
into the equations for the dipole effects, which has simply been termed a.
This latter a, which influences the theory of the effects of the secondary
solvation upon salting out and salting in (Sections 2.6.3 and 2.6.4), is that
due to the orientation of dipoles against the applied field; distortion polar-
izability, ad, is that due to the stretching of the molecule under the influence
of a field. The former polarizability, a, is simply connected to the dipole
moment of the molecule according to a formula such as Eq.(2.103). But
the latter, ad, is more complexly connected to the size of the molecule.
There is a parallelism with the radius, and in spherical symmetry cases it
is found that ad approximately follows r 3, where r is the radius of the mole-
cule.
Now, if the size of the nonelectrolyte is greater than that of the water
molecule (and for organic nonelectrolytes this is often so), it is clear from
the above equation that the dispersive interaction of a given ion is going
to be greater with the nonelectrolyte than with the water. This is a reversal
of the behavior regarded as usual when only the permanent dipoles of the
water and the nonelectrolyte are taken into account (for the dipole moment
of water is higher than that of most nonelectrolytes). But it may be asked,
in view of the above situation, why is not salting in (that which happens
when the nonelectrolyte out-competes the water molecule in its attraction
to the ion) the normal case? In this section, it is the dispersive interactions
which have been the center of attention: one has suddenly, isolatedly con-
sidered them. But, one has to ask whether the dispersive ion-water (and
dispersive ion-nonelectrolyte) interactions will dominate over the ion-
dipole interactions which have been at the center of the stage in Sections
2.6.2-2.6.4. If the ion-dipole interactions dominate the ion-dispersive inter-
actions, the considerations of those earlier sections are applicable, and
salting out is the norm, with salting in a rare exception. When the dispersive
interactions predominate, it is the other way around-salting in becomes
the norm.
 ION-SOLVENT INTERACTIONS 167

What factors of structure tend to make the dispersive forces dominate

the situation? Clearly, they will be more likely to have the main influence
upon the situation if the nonelectrolyte is big (because then the distortion
polarizability is big); but there are many situations where quite big non-
electrolytes are still salted out. The dispersive interaction contains the pro-
duct of the polarizability of both the ion and the nonelectrolyte (or the
ion and water, depending upon which interaction one is considering), so
that it is when both the ion and the nonelectrolyte are big (hence, both the
a's large) that the dispersive situation has the likelihood of being that which
dominates the issue, rather than the ion-dipole interaction.
In accordance with this it is found that, for example, if one maintains
the nonelectrolyte constant t and varies the ion in size, though keeping it
of the same type, salting in begins to predominate when the ion size exceeds
a certain value. A good example is in the case of the ammonium ion, and
a series of the tetra-alkyl ammonium ions with increasing size, i.e., NR 4 +,
where R is CH 3 , C 2 H 5 , etc. Here, the salting in already begins with the
methyl ammonium ion (its ad is evidently big enough), the ammonium
ion alone giving salting out. The degree of salting in increases with increase
of the size of the tetra-alkyl ammonium cation. Thus, the observations
made at the end of Section 2.6.4 concerning the salting in of detergents,
emulsions, and antibiotics by organic ions is, in principle, rationalized.
An attempt has been made to make these considerations quantitative.
Within the rough demands which the approximate nature of the equation
for the dispersive energy allows, there is agreement between theory and
experiment.
These considerations of the effect of ions upon the solubility of non-
electrolytes are sufficiently complicated to merit a little summary. The
field is divided into two parts. The first part concerns systems in which the
dispersive interactions are negligible compared with the dipole interactions.
Such systems tend to contain relatively small ions and molecules. Here
salting out is the expected phenomenon, and salting in occurs only in the
rare case in which the nonelectrolyte dipole moment is greater than the
dipole moment of the solvent. In the other division of the phenomena of
solubility effects caused by ions, the substances concerned tend to be large,
and, because distortion polarizability increases with size, this makes the
dispersive energy interactions dominate. Then, salting in becomes the more
expected situation.

t For example, it may be benzoic acid, considered a nonelectrolyte because it dissociates

to a very small degree.
168 CHAPTER 2

Further Reading
1. P. Debye and J. McAulay, Z. Physik, 26: 22 (1927).
2. J. a'M. Bockris, J. Bowler-Reed, and J. A. Kitchener, Trans. Faraday Soc.,
47: 184 (1951).
3. R. McDevit and F. Long, Chem. Rev., 51: 119 (1952).
4. B. E. Conway and J. a'M. Bockris, "Solvation," in: J. a'M. Bockris, ed.,
Modern Aspects of Electrochemistry, No.1, Butterworth's Publications, Inc.,
London, 1954.
5. E. L. McBain and E. C. Hutchison, Solubilization and Related Phenomena,
Academic Press, New York, 1955.
6. R. M. Diamond, J. Phys. Chem., 67: 2513 (1963).
7. B. E. Conway, J. E. Desnoyers, and A. C. Smith, Phil. Trans. Roy. Soc. London,
A256: 389 (1964).
8. J. E. Desnoyers, C. Jolicoeur, and G. E. Pelletier, Can. J. Chem., 42: 3232 (1965).
9. W. W. Drost Hansen, Advanced Chem., Ser. No. 67, 70-120 (1967).
10. W. Drost-Hansen, Chem. Phys. Letters, 2 (8): 647 (1968).

Appendix 2.1. Free Energy Change and Work

The free-energy change LlG which a system undergoes in a process can be
written quite generally as
LlG = LIE + p LI V + V LIp - T LIS - SLIT (A2.1.1)

If the process occurs at constant pressure and temperature,

Llp=O=LlT (A2.1.2)
and, therefore,
LlG = LIE + p LI V - T LIS (A2.1.3)

If, further, the process is reversible, the heat Q put into the system is related to
the entropy change through
Q = T LIS (A2.1.4)

and, from the first law of thermodynamics,

LIE = Q - W
= TLIS - W (A2.1.5)

where W is the total work done by the system.

Substituting for T LIS from Eq. (A2.1.5) in Eq. (A2.1.3), one has
LI G = - (W - p LI V) (A2.1.6)

Since W is the total work (including mechanical work) and p LI V is the mechanical
work of volume expansion,
LlG = -(work other than mechanical work done by the system) (A2.l 7)
or
LI G = work other than mechanical work done on the system (A2.l.8)
 ION-SOLVENT INTERACTIONS 169

Appendix 2.2. The Interaction between an Ion and a Dipole

The problem is to calculate the interaction energy between a dipole and an
ion placed at a distance, from the dipole center, the dipole being oriented at an
angle 0 to the line joining the centers of the ion and dipole (Fig. A2.2.1). (By
convention, the direction of the dipole is taken to be the direction from the
negative end to the positive end of the dipole.)
The ion-dipole interaction energy U1- D is equal to the charge Zie. of the
ion times the potential 1J!r due to the dipole at the site P of the ion
(A2.2.1 )

Thus, the problem reduces to the calculation of the potential 1J!r due to the
dipole. According to the law of superposition of potentials, the potential due to
an assembly of charges is the sum of the potentials due to each charge. Thus,
the potential due to a dipole is the sum of the potentials +q/r, and -q/r2 due
to the charges +q and -q which constitute the dipole and are located at distances
r, and '2 from the point P. Thus,
q q
1J!r = -,--
r, r2

= q(_1 __1 ) (A2.2.2)

" r2
From Fig. A2.2.2, it is obvious that

(A2.2.3)
and, therefore,

= [( y2 + Z2) + d 2 + 2zd]-J
= (r2 + d + 2zd)-J
2

_ 1 [ (d)2 2dZ]-~
--1+-+~
, r ,2 (A2.2.4)

Fig. A2.2.1.
170 CHAPTER 2

~~~ ______- L_____ y

d Fig. A2.2.2.

At this stage, an important approximation is made, namely, that the distance

2d between the charges in the dipole is negligible compared with r. In other
words, the approximation is made that

d)2 2dz
1 + (- + -
r r2
'" 1 + -2dz
r 2
(A2.2.S)

It is clear that the validity of the approximation decreases the closer the ion
comes toward the dipole, i.e., as r decreases.
Making the above approximation, one has [see Eq. (A.2.2A)]

(A2.2.6)

which, by the binomial expansion taken to two terms, gives

(A2.2.7)

By similar reasoning,
_~
r2
= ~
r
(1 + dZ)
r 2
(A2.2.8)

By using Eqs. (A2.2.7) and (A2.2.8), Eq. (A2.2.2) becomes

2dq Z
'Pr = - ----;'2 ---,: (A2.1.9)

Since z/r = cos 0 and 2dq is the dipole moment It.

It cos 0
'Pr = - --r-2 - (A2.2.10)

or the ion-dipole interaction energy is given by

-ZieO/t cos 0
U1 _ D = -----,--- (A2.2.ll)
r2
 ION-SOLVENT INTERACTIONS 171

Appendix 2.3. The Interaction between an Ion and a Water

Quadrupole

Instead of presenting a sophisticated general treatment for ion-quadrupole

interactions, a particular case of these interactions will be worked out. The special
case to be worked out is that corresponding to the water molecule being oriented
with respect to a positive ion so that the interaction energy is a minimum.
In this orientation (see Fig. A2.3.1), the oxygen atom and a positive ion
are on the Yaxis which bisects the H-O-H angle. Further, the positive ion,
the oxygen atom, and the two hydrogen atoms are all considered in the XY
plane. The origin of the XY coordinate system is located at the point Q, which
is the center of the water molecule. The ion is at a distance r from the origin.
The ion-quadrupole interaction energy U1 _ Q is simply given by the charge
on the ion times the potential 'Pr at the site of the ion due to the charges of the
quadrupole,
(A2.3.1)

But the potential 'Pr is the sum of the potentials due to the four charges
ql' q2) q3) and q. in the quadrupole (I and 2 are the positive charges at the
hydrogen, and 3 and 4 are the negative charges at the oxygen). That is,

'Pr = 'PI + 'P2 + 'P3 + 'P. (A2.3.2)

Each one of these potentials is given by the usual coulombic expression for the
potential
(A2.3.3)

where the minus sign appears before the third and fourth terms because q3 and q.

Fig. A2.3.1.
172 CHAPTER 2

are negative charges. Further, the magnitudes of all the charges are equal

(A2.3.4)

and, because of symmetrical disposition of the water molecule,

and (A2.3.5)

Hence, from Eqs. (A2.3.3), (A2.3.4), and (A2.3.5),

1J!r = 2q(_I_ __1_) (A2.3.6)

r, r3

It is obvious (see Fig. A2.3.1) that

r,2 = (r + a)2 + X2 (A2.3.7)

a2 + X2 2a)
= r2 ( 1 + ---. + -- (A2.3.8)
r2 r
and
r3 = r - f3 (A2.3.9)

(A2.3.10)

Thus,

1+ a +
1 1 ( 2 X2 2a )-~
-=-
r, r r2
+-r
(A2.3.11)

and
_1r3 = ~
r
(1_1.-)-'
r
(A2.3.12)

One can now use the binomial expansion, i.e.,

n(n + 1)
(1 ± m)-n = 1 =t= nm + 2 m 2 =t= (A2.3.13)

and drop off all terms higher than the third. t Thus,

_1_",{~_~ a2 +x2 _~+~[(a2+x2)2 4a(a 2 +x 2)]} + 4a 2 +

r, r 2 r3 r2 8 r5 r4 r3
(A2.3.14)
and, omitting all terms with powers r greater than 3, one has
1 1 a 1
- ~ - - - 2 + --
3
(2a 2 - X2) (A2.3.15)
r, r r 2r

t It is at this stage that the treatment of ion-quadrupole interactions diverges from

that of ion-dipole interactions (el Appendix 2.2). In the latter, the binomial expansion
was terminated after the second term.
 ION-SOLVENT INTERACTIONS 173

Further,
_1
'3'
'" ~ (1 +i, ,+~)
2
1 fl fi2
=-+-+-
, ,2 ,3 (A2.3.16)

Subtracting Eq. (A2.3.l6) from Eq. (A2.3.15), one has

_1___1_ = _ (a + fl) + _t_ (2a2 _ X2 _ 2fl2) (A2.3.l7)

','3 ,2 2,3
and, therefore, by substitution of (A2.3.t7) in (A2.3.6),

1fT = -
2q(a + fl) + --
t
[2(2qa 2 - 2qfl2) - 2qx2] (A2.3.t8)
,2 2r 3

The first term on the right-hand side of (A2.3.t8) can be rearranged as

follows:
2q(a + fl) = 2qa + 2qfl
= ~ (2q)d where d = a or fl (A2.3.t9)

Thus, as a first approximation, the water molecule can be represented as a dipolar

charge distribution in which there is a positive charge of +2q (due to the H
atoms) at a distance a from the origin on the bisector of the H-O-H angle
and a charge of -2q (due to the lone electron pair) at a distance -fl from the
origin, it follows that

~ (2q)d = ~ magnitude of each charge of dipole

x distance of the charge from origin (A2.3.20)

The right-hand side of this expression is the general expression for the dipole
moment It, as is seen by considering the situation when a = fl, i.e., ~ (2q)d
= (2q)2d, where 2d is the distance between the charges of the dipole, in which
case one obtains the familiar expression for the dipole moment It,

It = 2q2d (A2.3.21)

Thus, the first term on the right-hand side of Eq. (A2.3.l8) is

2q(a + fl) (A2.3.22)

r2

The second term can be interpreted as follows: Consider (2qa 2 - 2qfl2). It

can be written thus

2qa2 - 2qfl2 = qa + qa2 + (_q)fl2 + (_q)fl2

2

= L: qd; where d y = a or fJ (A2.3.23)

174 CHAPTER 2

But the general definition of a quadrupole moment is the magnitude of each

charge of quadrupole times the square of the distance of the charge from the
origin. Thus, ~ qd. 2 is the y component P•• of the quadrupole moment for the
particular coordinate system which has been chosen. Similarly, ~ 2qx 2 is the x
component Pxx of the quadrupole moment. Thus,

2(2qa 2 - 2qfP) - 2qx2 = 2p .. - Px", (A2.3.24)

One can combine 2p •• - p"" into a single symbol and talk of the quadrupole
moment Pw of the water molecule in the particular orientation of Fig. A2.3. t.
Hence,
2(2qa 2 - 2qf32) - 2qx2 = pw (A2.3.25)

and, therefore, by substituting (A2.3.22) and (A2.3.25) in (A2.3.18),

Il Pw
'Pr =- -2 + - 23 (A2.3.26)
r r

The ion-quadrupole interaction energy [ef Eq. (A2.3.1)] thus becomes

(A2.3.27)

When a negative ion is considered, the water molecule turns around through
n, and one obtains by an argument similar to that for positive ions

(A2.3.28)
CHAPTER 3

ION-ION
INTERACTIONS

3.1. INTRODUCTION

A model has been given for the breaking-up of an ionic crystal into
free ions which stabilize themselves in solution with solvent sheaths. One
central theme guided the account, the interaction of an ion with its neigh-
boring water molecule.
But ion-solvent interactions are only part of the story relating an ion
to its environment. When an ion looks out upon its surroundings, it sees
not only solvent dipoles but also other ions. The mutual interaction between
these ions constitutes an essential part of the picture of an electrolytic
solution.
Why are ion-ion interactions of importance? Because, as will be shown,
they affect the equilibrium properties of ionic solutions, and also because
they interfere with the drift of ions, for instance, under an externally applied
electric field (Chapter 4).
Now, the degree to which these interactions affect the properties of
solutions will depend on the mean distance apart of the ions, i.e., on how
densely the solution is populated with ions, because the interionic fields are
distance dependent. This ionic population density will in turn depend on
the nature of the electrolyte, i.e., on the extent to which the electrolyte
gives rise to ions in solution.

175
176 CHAPTER 3

3.2. TRUE AND POTENTIAL ELECTROLYTES

3.2.1. Ionic Crystal Are True Electrolytes

An important point to recall regarding the dissolution of an ionic

crystal (Chapter 2) is that ionic lattices consist of ions even before they
come into contact with a solvent. In fact, all that a polar solvent does is
to use ion-dipole (or ion-quadrupole) forces to disengage the ions from
the lattice sites, solvate them, and disperse them in solution.
Such ionic crystals are known as true electrolytes or ionophores (the
Greek suffix phore means "bearer of"; thus, an ionophore is a "substance
which bears ions"). When a true electrolyte is melted, its ionic lattice is
dismantled and the pure liquid true electrolyte shows considerable ionic
conduction (Chapter 2). Thus, the characteristic of a true electrolyte is
that, in the pure liquid form, it is an ionic conductor. All salts belong to
this class. Sodium chloride, therefore, is a typical true electrolyte.

3.2.2. Potential Electrolytes: Nonionic Substances Which React

with the Solvent to Yield Ions

A large number of substances, e.g., organic acids, show little conduc-

tivity in the pure liquid state. Evidently, there must be some fundamental
difference in structure between organic acids and inorganic salts, and this
difference is responsible for the fact that one pure liquid (the true electrolyte)
is an ionic conductor and the other is not.
What is this difference between, say, sodium chloride and acetic acid?
Electron diffraction studies furnish an answer. They show that gaseous
acetic acid consists of separate, neutral molecules and the bonding of the
atoms inside these molecules is essentially nonionic. These neutral molecules
retain their identity and separate existence when the gas condenses to give
liquid acetic acid. Hence, there are hardly any ions in liquid acetic acid
and, therefore, little conductivity.
Now, the first requirement of an electrolyte is that it should give rise
to a conducting solution. From this point of view, it appears that acetic
acid will never answer the requirements of an electrolyte; it is nonionic.
When, however, acetic acid is dissolved in water, an interesting phenomenon
occurs: ions are produced, and, therefore, the solutions conduct electricity.
1 hus, acetic acid, too, is a type of electrolyte; it is not a true electrolyte,
but a potential one ("one which can, but has not yet, become"). Potential
electrolytes are also called ionogens, i.e., "ion producers."
How does acetic acid, which does not consist of ions in the pure liquid
 ION-ION INTERACTIONS 177

state, generate ions when dissolved in water? In short, how do potential

electrolytes work? Obviously, there must be some reaction between neutral
acetic acid molecules and water, and this reaction must lead to the splitting
of the acetic acid molecules into charged fragments, or ions.
A simple picture is as follows. Suppose that an acetic acid molecule
collides with a water molecule and, in the process, the H of the acetic acid
OH group is transferred from the oxygen atom of the OH to the oxygen
atom of the H 2 0. A proton has been transferred from CHaCOOH to H 2 0

H1 , /0 / H [ H1 , /0J- [ / HJ+
H-C-C + 0 ~ H-C-C + H-O

HI '"OH / '"H HI '"0 '"H

The result of the proton transfer is that two ions have been produced:
(1) an acetate ion and (2) a hydrated proton. Thus, potential electrolytes
(organic acids and most bases) dissociate into ions by ionogenic, or ion-
forming, chemical reactions with solvent molecules, in contrast to true
electrolytes, which give rise to ionic solutions by physical interactions be-
tween ions present in the ionic crystal and solvent molecules (Fig. 3.1).
The mechanism of the functioning of potential electrolytes will be
described in detail later (Chapter 5).

3.2.3. An Obsolete Classification: Strong and Weak Electrolytes

The classification into true and potential electrolytes is a modern one.
It is based on a knowledge of the structure of the electrolyte: whether, in
the pure form, it consists of an ionic lattice or neutral molecules (Fig. 3.2).
It is not based on the behavior of the solute in any particular solvent.
Historically, however, the classification of electrolytes was made on
the basis of their behavior in one particular solvent, i.e., water. Weak
electrolytes were those which yielded relatively poorly conducting solutions
when dissolved in water, and strong electrolytes were those which gave highly
conducting solutions when dissolved in water.
The disadvantage of this classification into strong and weak electrolytes
lies in the following fact: As soon as a different solvent, i.e., a nonaqueous
solvent, is chosen, what was a strong electrolyte in water may behave as
a weak electrolyte in the nonaqueous solvent. For example, sodium chloride
behaves like a strong electrolyte (i.e., yields highly conducting solutions)
in water; and acetic acid, like a weak electrolyte. In liquid ammonia, how-
ever, the conductance behavior of acetic acid is similar to that of sodium
178 CHAPTER 3

~ ~SOLVENT WATER MOLECULES

~ OXALIC ACID
PROTON

WM~;+~
[H3 O~+ OXALATE ION

~\.,,' ;'f--O :)(y(;EN

H" rHO-CO [H" ] + coo-

/0+ I --. /O-H + I
H HO-CO H COOH

OXALIC ACID CRYSTALS OXALATE ION

Fig. 3.1. Schematic diagram to illustrate the difference in the way

potential electrolytes and true electrolytes dissolve to give ionic solu-
tions: (a) Oxalic acid (a potential electrolyte) undergoes a proton-
transfer chemical reaction with water to give rise to hydrogen ions and
oxalate ions. (b) Sodium chloride (a true electrolyte) dissolves by the
solvation of the Na+ and CI+ ions in the crystal.

chloride in water, i.e., the solutions are highly conducting (Table 3.1).
This is an embarrassing situation. Can one say: Acetic acid is weak in
water and strong in liquid ammonia? What is wanted is a classification
 ION-ION INTERACTIONS 179

(a)

UNCHARC£D MOL£CUL£$ OF CRYSTALUSATION

Of" OXALIC ACIO

Noel CRYSTAl

Fig. 3.2. Electrolytes can be classified as (a) potential electro-

lytes (e.g., oxalic acid) which, in the pure state, consist of un-
charged molecules and (b) true electrolytes (e.g., sodium chloride)
which, in the pure state, consist of ions.

TABLE 3.1
Conductance Behavior of Substances in Different Media

Equivalent conductance

Water Liquid ammonia

NaC! 106.7 284.0

Acetic acid 4.7 216.6
180 CHAPTER 3

of electrolytes which is independent of the solvent concerned. The classifi-

cation into true and potential electrolytes is such a classification. It does
not depend on the solvent.

3.2.4. The Nature of the Electrolyte and the Relevance of lon-

Ion Interactions

Solutions of most potential electrolytes in water generally contain only

small concentrations of ions, and, therefore, ion-ion interactions in these
solutions are negligible; the ions are on the average too far apart. The
behavior of such solutions is governed predominantly by the position of
the equilibrium in the proton-transfer reaction between the potential elec-
trolyte and water (see Chapter 5).
In contrast, true electrolytes are completely dissociated into ions when
the parent salts are dissolved in water. The resulting solutions generally
consist only of solvated ions and solvent molecules. The dependence of
many of their properties on concentration (and, therefore, mean distance
apart of the ions in the solution) is determined, therefore, by the interac-
tions between ions. To understand these properties, one must understand
ion-ion interactions.

Further Reading
1. G. Kortum and J. O'M. Bockris, Textbook of Electrochemistry, Vol. I, Elsevier,
Amsterdam, 1951.
2. R. M. Fuoss and F. Accascina, Electrolytic Conductance, Interscience Pub-
lishers, Inc., New York, 1959.

3.3. THE DEBYE-HUCKEL (OR ION-CLOUD) THEORY OF ION-

ION INTERACTIONS

3.3.1. A Strategy for a Quantitative Understanding of lon-Ion

Interactions

The first task in thinking in detail about ion-ion interactions is to

evolve a quantitative measure of these interactions. t
One approach is to follow a procedure similar to that used in the
discussion of ion-solvent interactions (cf Section 2.2.1). Thus, one can
consider an initial state in which ion-ion interactions do not exist (are

t The question of how one obtains an experimental measure of ion-ion interactions

is discussed in Section 3.4.
 ION-ION INTERACTIONS 181

INITIAL STATE FINAL STATE

WORK OF ION-ION
NO ION-ION ION-ION
INTERACTION ,4G1_1
INTERACTI~ INTERACTIONS

Fig. 3.3 The free energy flG,_1 of ion-ion in-

teractions is the free-energy change in going
from a hypothetical electrolytic solution, in which
ion-ion interactions do not operate, to a real so-
lution, in which these interactions do operate.

"switched off") and a final state in which the interactions are in play (are
"switched on"). Then, the free-energy change in going from the initial
state to the final state can be considered the free energy LJ GI -I' of ion-ion
interactions (Fig. 3.3).
The final state is obvious; it is ions in solution. The initial state is
not so straightforward; one cannot take ions in vacuum, because then there
will be ion-solvent interactions when these ions enter the solvent. The
following approach is therefore adopted. One conceives of a hypothetical
situation in which the ions are there in solution but are nevertheless not
interacting. Now, if ion-ion interactions are assumed to be electrostatic
in origin (a similar assumption was made with regard to ion-solvent inter-
actions, cf Section 2.2.2), then the imaginary initial state of noninteracting
ions implies an assembly of discharged ions.
Thus, the process of going from an initial state of noninteracting ions
to a final state of ion- ion interactions is equivalent to taking an assembly
of discharged ions, charging them up, and setting the electrostatic charging
work equal to the free energy LlG1 _ 1 of ion-ion interactions (Fig. 3.4).
One point about the above procedure should be borne in mind. Since,

o 0 0 CHARGING WORK
0 00 0
o
~ llG 1- 1

Fig. 3.4. The free energy flG,_1 of ion-ion interactions is

the electrostatic work of taking an imaginary assembly of dis-
charged ions and charging them up to obtain a solution of
charged ions.
182 CHAPTER 3

in the charging process, both the ppsitively charged and negatively charged
ionic species are charged up, one obtains a free-energy change which involves
all the ionic species constituting the electrolyte. Generally, however, the
desire is to isolate the contribution to the free energy of ion-ion inter-
actions arising from one ionic species i only. This partial free-energy change
is, by definition, the chemical-potential change iJfti-I arising from the inter-
actions of one ionic species with the ionic assembly.
To compute this chemical-potential change iJfti-I' rather than the free-
energy change iJ G1 - 1 , one must adopt an approach similar to that used
in the Born theory of solvation. One thinks of an ion of species i and imagines
that this reference ion alone of all the ions in solution is in a state of zero
charge (Fig. 3.5). If one computes the work of charging up the reference
ion (of radius ri) from a state of zero charge to its final charge of ZieO,
then the charging work W times the Avogadro number NA is equal to the
partial molar free energy of ion-ion interactions, i.e., to the chemical poten-
tial of ion-ion interactions
(3.1)

Further, one can consider a charged sphere (of radius ri and charge ZieO)
as a model for an ion (cf Section 2.2.2) and use the expression for the work
of charging the sphere from a state of zero charge to a charge of ZieO to
represent the work W of charging an ion, i.e.,

W = (Zi eO)2 (2.16)

2Ui

(3.2)

But ZieO/ui is the electrostatic potential "p at the surface of the ion, and,
therefore,
(3.3)

The essence of the task, therefore, in computing the chemical-potential

change due to the interactions of the ionic species i with the ionic solution,
is the calculation of the electrostatic potential produced at a reference ion
by the rest of the ions in solution. Theory must aim at this quantity.
If one knew the time-average spatial distribution of the ions, then one
could find out how all the other charges are distributed as a function of
distance from the reference ion. At that stage, one of the fundamental laws
of electrostatics could be used, namely, the law of the superposition of
 ION-ION INTERACTIONS 183

•
••
WORK OF CHARGING
A MOLE OF REFERENCE
•
ION. t. ILl-l

OTHER IONS
ARE CHARGED
DISCHARGED REFERENCE
ION OF SPECIES i REFERENCE ION
CHARGED

Fig. 3.5. The chemical potential !::,.fli-I arising from the inter-
actions of an ionic species i with the electrolytic solution is
equal to the Avogadro number times the electrostatic work of
taking an imaginary solution in which one reference ion alone is
discharged and charging this reference ion up to its normal charge.

potentials, according to which the potential at a point due to an assembly

of charges is the sum of the potentials due to each of the charges in the
assembly.
Thus, the problem of calculating the chemical-potential change LI,ui-l
due to the interactions between one ionic species and the assembly of all
the other ions has been reduced to the following problem: On a time
average, how are the ions distributed around any specified ion? If that
distribution becomes known, it would be easy to calculate the electrostatic
potential of the specified ion, due to the other ions and then, by Eq. (3.3),
the energy of that interaction. Thus, the task is to develop a model that
describes the equilibrium spatial distribution of ions inside an electrolytic
solution and then to describe the model mathematically.

3.3.2. A Prelude to the Ionic-Cloud Theory

A spectacular advance in the understanding of the distribution of

charges around an ion in solution was achieved in 1923 by Debye and
HUckel. It was as significant in the understanding of ionic solutions as the
Maxwell theory of the distribution of velocities in the understanding of
gases.
Before going into the details of their theory, a moment's reflection
on the magnitude of the problem would promote appreciation of their
achievement.
Consider, for example, a 10-3 mole liter-1 aqueous solution of sodium
chloride. There will be 10-6 X 6.023 X 1023 sodium ions per cubic centi-
meter of solution and the same number of chloride ions, together, of course,
184 CHAPTER 3

with the corresponding number of water molecules. Nature takes these

2 X 6.023 X 1017 ions cm~3 and arranges them so that there is a particular
time-average t spatial distribution of the ions. The number of particles
involved are enormous, and the situation appears far too complex for
mathematical treatment.
But there exist conceptual techniques for tackling complex situations.
One of them is model building. What is done is to conceive a model which
contains only the essential features of the real situation. All the thinking
and mathematical analysis is done on the (relatively simple) model, and,
then, the theoretical predictions are compared with the experimental be-
havior of the real system. A good model simulates nature. If the model
yields wrong answers, then one tries again by changing the imagined model
until one arrives at a model, the theoretical predictions of which agree
well with experimental observations.
The genius of Debye and HUckel lay in their formulation of a very
simple but powerful model for the time-average distribution of ions in very
dilute solutions of electrolytes. From this distribution, they were able to
get the electrostatic potential contributed by the surrounding ions to the
total electrostatic potential at the reference ion and, hence, the chemical-
potential change arising from ion-ion interactions [cf Eq.(3.3)]. Attention
will now be focused on their approach.
The electrolytic solution consists of solvated ions and water molecules.
The first step in the Debye-HUckel approach is to select arbitrarily any
one ion out of the assembly and call it a reference ion or central ion. Only
the reference ion is given the individuality of a discrete charge. What is
done with the water molecules and the remaining ions? The water mole-
cules are looked upon as a continuous dielectric medium. The remaining
ions of the solution (i.e., all ions except the central ion) are lapsed into
anonymity, their charges being "smeared out" into a continuous spatial
distribution of charge (Fig. 3.6). Whenever the concentration of ions of
one sign exceeds that of the opposite sign, there will arise a net or excess
charge in the particular region under consideration. Obviously, the total
charge in the atmosphere must be of opposite sign and exactly equal to
the charge on the reference ion.

t Using an imaginary camera (with exposure time of '" 10-12 sec), suppose that it were
possible to take snapshots of the ions in an electrolytic solution. Different snapshots
would show the ions distributed differently in the space containing the solution; but
the scrutiny of a large enough number of snapshots (say, '" 10'2 ) would permit one
to recognize a certain average distribution characterized by average positions of the
ions; this is the time-average spatial distribution of the ions.
 ION-ION INTERACTIONS 185

OTHER IONS

r.:--;:--~7?~;!:::::a.....- WATER MOLECULES

r.+-.~+l--_ REFERENCE ION

(a)

SHADING INDICATES
NET CHARGE DENSITY, P,
DUE TO OTHER IONS
REFERENCE ION

'---__ _ _ _---J'-... CONTINUOUS DIELECTRIC (E)

(b) IN PLACE OF WATER
MOLECULES

Fig. 3.6. A schematic comparison of (a) the assembly of

ions and solvent molecules which constitute a real electro-
lytic solution and (b) the Debye-Huckel picture in which
a reference ion is surrounded by net charge density (} due
to the surrounding ions and a dielectric continuum of the
same dielectric constant E as the bulk solvent.

Thus, the electrolytic solution is considered to consist of a central ion

standing alone in a continuum. Thanks to the water molecules, this con-
tinuum acquires a dielectric constant (taken to be the value for bulk water).
The charges of the discrete ions which populate the environment of the
central ion are thought of as smoothed out and contribute to the con-
tinuum dielectric a net charge density (excess charge per unit volume).
Thus, water enters the analysis in the guise of a dielectric constant E; and
the ions, except the specific one chosen as the central ion, in the form of
an excess charge density Q (Fig. 3.7).
Thus, the complicated problem of the time-average distribution of
ions inside an electrolytic solution reduces, in the Debye-Hiickel model,
to the mathematically simpler problem of finding out how the excess
charge density Q varies with distance r from the central ion.
An objection may be raised at this point. The electrolytic solution as
a whole is electroneutral, i.e., the net charge density Q is zero. Then, why
is not Q = 0 everywhere?
So as not to anticipate the detailed discussion, an intuitive answer
186 CHAPTER 3

THE SOLVENT MOLECULE

PROVIDE A

MEDIUM OF
DIELECTRIC CONSTANT, e

REFERENC~
EXCESS CHARGE
DENSITY,Pr

THE SURROUNDING IONS

GIVE RISE TO AN \
DEBYE - HuCKEL
SOLUTION MODEL

Fia. 3.7. The Debye-Huckel model is based upon selecting

one ion as a reference ion, replacing the solvent molecules by a
continuous medium of dielectric constant E and the remaining
ions by an excess charge density (!, (the shading used in this
book to represent the charge density is not indicated in this
figure).

will first be given. If the central ion is, for example, positive, it will exert an
attraction for negative ions; hence, there should be a greater aggregation
of negative ions than of positive ions in the neighborhood of the central
positive ion, i.e., (! =1= O. An analogous situation, but with a change in sign,
obtains near a central negative ion. At the same time, the thermal forces
are knocking the ions about in all directions and trying to establish elec-
troneutrality, i.e., the thermal motions try to smooth everything to (! = O.
Thus, the time average of the electrostatic forces of ordering and the thermal
forces of disordering is a local excess of negative charge near a positive ion
and an excess of positive charge near a negative ion. Of course, the excess
positive charge near a negative ion compensates the excess negative charge
near a positive ion, and the overall effect is electro neutrality, i.e., a (! of
zero for the whole solution.

3.3.3. How the Charge Density near the Central Ion Is Deter-
mined by Electrostatics: Poisson's Equation

Consider an infinitesimally small volume element dv situated at a

distance r from the arbitrarily selected central ion, upon which attention
is to be fixed during the discussion (Fig. 3.8), and let the net charge density
 ION-ION INTERACTIONS 187

ELECTROSTATIC
~ POTENTIAL, IfIr

~"""""'."'.d'
r CHARGE DENSITY, P

I
r

'" ~EFERENCE ION

Fig. 3.S. At a distance r from the refer-
ence ion, the excess charge density and
electrostatic potential. in an infinitesimal vol-
ume element dv, are (!, and 'Pr. respectively.

inside the volume element be (lr' Further, let the average t electrostatic
potential in the volume element be 1jJr' The question is: What is the relation
between the excess density (lr in the volume element and the time-average
electrostatic potential 1jJr?
One relation between (lr and 1jJr is given by Poisson's equation (Ap-
pendix 3.1). There is no reason to doubt that there is spherically sym-
metrical distribution of positive and negative charge and, therefore, excess
charge density around a given central ion. Hence, Poisson's equation can
be written as
(3.4)

where £ is the dielectric constant of the medium and is taken to be that of

bulk water.

3.3.4. How the Excess Charge Density near the Central Ion Is
Given by a Classical Law for the Distribution of Point
Charges in a Coulombic Field

The excess charge density in the volume element dv is equal to the

total ion density (total number of ions per unit volume) times the charge
on these ions. Let there be, per unit volume, nl ions of type I, each bearing
charge zleO' n2 of type 2 with charge Z2eO' and n, of type i with charge z,e o ,

t Actually, there are discrete charges in the neighborhood of the central ion and, therefore,
discontinuous variations in the potential. But, because in the Debye-Hiickel model
the charges are smoothed out, the potential is averaged out.
188 CHAPTER 3

where Zi is the valency of the ion and eo is the electronic charge. Then,
the excess charge density (]r in the volume element dv is given by

(3.5)

(3.6)

To proceed further, one must link up the unknown quantities n 1 ,

n 2 ,· .. , 11 i,' .. to known quantities. The link is made on the basis of the
Boltzmann distribution law of classical statistical mechanics. Thus, one
writes
(3.7)

where U can be described either as the change in potential energy of the i

particles when their concentration in the volume element dv is changed
from the bulk value nio to f1i or as the work that must be done by a hypo-
thetical external agency against the time average of the electrical and other
forces between ions in producing the above concentration change. Since
the potential energy U relates to the time average of the forces between
ions rather than to the actual forces for a given distribution, it is also
known as the potential of average force.
If there are no ion-ion interactional forces, U = 0; then, Ili = Ilio,
which means that the local concentration would be equal to the bulk
concentration. If the forces are attractive, then the potential change U is
negative (i.e., negative work is done by the hypothetical external agency)
and l1i > 11;°; there is a local accumulatiol1 of ions in excess of their bulk
concentrations. If the forces are repulsive, the potential-energy change is
positive (i.e., the work done by the external agency is positive) and l1i < l1io;
there is local depletiol1 of ions.
In the first instance, and as a first approximation, one may ignore all
types of ion-ion interactions except those deriving from simple coulombic t
forces. Thus, short-range interactions (e.g., dispersion interactions) are
excluded. This is a fundamental assumption of the Debye-Hlickel theory.
Thus, the potential of average force U simply becomes the coulombic
potential energy of an ion of charge Zieo in the volume element dv, i.e., to

t In this book, the term coulombic is restricted to forces (with r-' dependence on distance)
which are based directly on Coulomb's law. More complex forces, e.g., those which
vary as r- 4 or r7, may result as a net force from the resultant of several different
coulombic interactions. Nevertheless, such more complex results of the interplay of
several coulombic forces will be called noncoulombic.
 ION-ION INTERACTIONS 189

the charge Zieo on the ion times the electrostatic potential1J!r in the volume
element dv. That is,
(3.8)

The Boltzmann distribution law (3.7) thus assumes the form

(3.9)

Now that ni' the concentration of the ionic species i in the volume
element dv, has been related to its bulk concentration nio, the expression
(3.6) for the excess charge density in the volume element dv becomes

(3.10)

3.3.5. A Vital Step in the Debye-Huckel Theory of the Charge

Distribution around Ions: Linearization of the Boltzmann
Equation

At this point of the theory, Debye and HUckel made a move which
was not only mathematically expedient but also turned out to be wise.
They decided to carry out the analysis only for systems in which the average
electrostatic potential 1J!r would be small so that

or Zi eo1J!, ~ 1 (3.11 )
kT -<;:;

Based on this assumption, one can expand the exponential of Eq. (3.10)

+( r...
in a Taylor series, i.e.,

e-zieO~'T/(kT) = 1- Zi~o;r + Zi~o;r (3.12)

and neglect all except the first two terms. Thus, in (3.10),

(3.13 )

(3.14 )

The first term ~ npzieO gives the charge on the electrolytic solution
as a whole. But this is zero because the solution as a whole must be elec-
trically neutral. The local excess charge densities near ions cancel out
because the excess positive charge density near a negative ion is compen-
190 CHAPTER 3

sated for by an excess negative charge density near a positive ion. Hence,

(3.15)

and one is left with

(3.16)

3.3.6. The Linearized Poisson-Boltzmann Equation

The stage is now set for the calculation of the potential "Pr and the
charge density (!, in terms of known parameters of the solution.
Notice that one has obtained two expressions for the charge density
(!r in the volume element dv at a distance r from the central ion. One has
the Poisson equation [cf Eq. (3.4)]

(!,
e
= - 4n 7 [1 d
dr (2r dr
d"Pr)] (3.17)

and one has the "linearized" Boltzmann distribution

o 2 2
__ '" ni Zi eo "Pr (3.18)
(!r - ~, kT

where Li refers to the summation over all species of ions typified by i.

If one equates these two expressions one can obtain the linearized
Poisson-Boltzmann (P-B) expression

d (2 d"Pr ) _ ( 4n '" 0 2 2)
71 dr r dr - ekT ~ ni Zi eo "Pr (3.19)

The constants in the right-hand parentheses can all be lumped together

and called a new constant ,,2, i.e.,

(3.20)

At this point, the symbol " has come in only to reduce the tedium
of writing. It turns out later, however, that" is not only a shorthand symbol;
it contains information concerning several fundamental aspects of the
distribution of ions around an ion in solution. In Chapter 7, it will be shown
that it also contains information concerning the distribution of charges near
a metal surface in contact with an ionic solution. In terms of ", the linear-
 ION-ION INTERACTIONS 191

ized P-B expression (3.19) is

1 d (2 d'f/Jr)
---,:2 dr r fir =X'f/Jr
2 (3.21 )

3.3.7. The Solution of the Linearized P-B Equation

The rather messy-looking linearized P-B equation (3.21) can be tidied
up by a mathematical trick. Introducing a new variable p, defined by

P,
'f/J, =, (3.22)

one has

dr
= -drd -p,r= - -r2p, +r1-dp,
d'f/Jr
- -
dr
and, therefore,

(3.23)

Hence, the differential equation (3.21) becomes

1 d 2p, 2 P,
, dr2 =X , (3.24)

or
(3.25)

To solve this simple differential equation, it is recalled that the dif-

ferentiation of an exponential function results in the multiplication of that
function by the constant in the component. For example,

d
- e±xr = ±xe±x,
dr
and (3.26)
192 CHAPTER 3

Hence, if fl is an exponential function of " one will obtain a differential

equation of the form of Eq. (3.25). In other words, the "primitive" or
"origin" of the differential equation must have had an exponential in u,.
Two possible exponential functions, however, would lead to the same
final differential equation; one of them would have a positive exponent
and the other a negative one [ef Eq. (3.26)]. The general solution of the
linearized P-B equation can, therefore, be written as

fl = Ae- xr + Be+ xr (3.27)

where A and B are constants to be evaluated. Or, from Eq. (3.22),

e- xr e+ xr
1jJr = A -- , + B -,- (3.28)

The constant B is evaluated by using the boundary condition that,

far enough from a central ion situated at , = 0, the thermal forces com-
pletely dominate the coulombic forces which decrease as ,2, and there is
electroneutrality, i.e., the electrostatic potential 1jJr vanishes at distances
sufficiently far from such an ion, 1jJr ---+ 0 as , ---+ (X). This condition would
be satisfied only if B = o. Thus, if B had a finite value, Eq. (3.29) shows
that the electrostatic potential would shoot up to infinity, I.e., 1jJr ---+ (X)
as r ---+ (X), a physically unreasonable proposition. Hence,
e- xr
1jJr = A -,- (3.29)

To evaluate the integration constant A, a hypothetical condition will

be considered in which the solution is so dilute and, on the average, the
ions are so far apart that there is a negligible interionic field. Further, the
central ion is assumed to be a point charge, i.e., to have a radius negligible
compared with the distances otherwise to be considered. Hence, the po-
tential near the central ion is, in this special case, simply that due to an
isolated point charge of value Zieo,

(3.30)

At the same time, for this hypothetical solution in which the concen-
tration tends to zero, i.e., n iO ---+ 0, it is seen from Eq. (3.20) that 'X ---+ o.
Thus, in Eq. (3.29), e-><r ---+ 1, and one has

A
1jJr = -,- (3.31)
 ION-ION INTERACTIONS 193

POTENTIAL

I 2 3
DISTANCE FROM CENTRAL ION (IN r /~ UNITS)

Fig. 3.9. The variation of the electrostatic

potential 'P as a function of distance from
the central ion expressed in units of fix.

Hence, by combining Eqs. (3.30) and (3.31),

A = ZieO
(3.32)
s

By introducing this expression for A into Eq. (3.29), the result is

(3.33)

Here, then, is the appropriate solution of the linearized P-B equation (3.21).
It shows how the electrostatic potential varies with distance r from an
arbitrarily chosen reference ion (Fig. 3.9).

3.3.S. The Ionic Cloud around a Central Ion

In the Debye-HUckel model of a dilute electrolytic solution, a reference
ion sitting at the origin of the spherical coordinate system is surrounded
by the smoothed-out charge of the other ions. Further, because of the local
inequalities in the concentrations of the positive and negative ions, the
smoothed-out charge of one sign does not (locally) cancel out the smoothed-
out charge of the opposite sign; there is a local excess charge density of
one sign.
Now, as explained in Section 3.3.2, the principal objective of the Debye-
HUckel theory is to calculate the time-average spatial distribution of the
excess charge density around a reference ion. How is this objective attained?
The Poisson equation (3.4) relates the potential at r from the sample
ion to the charge density at r, i.e.,
4n
= - -s-er (3.4)
194 CHAPTER 3

Fig. 3.10. The variation of the excess

charge density 12 as a function of distance
from the central ion.

Further, one has the linearized P-B equation

d (2 d1jJr) 2
T21 dr r ---;r,:- = x 1jJ r (3.21)

From these two Eqs. (3.4) and (3.21), one has the linear relation between
excess charge density and potential, i.e.,

(3.34)

and by inserting the solution (3.33) for the linearized P-B equation, the
result is
(3.35)

Here then is the desired expression for the spatial distribution of the
charge density with distance r from the central ion (Fig. 3.10). Since the
excess charge density results from an unequal distribution of positive and
negative ions, Eq. (3.35) also describes the distribution of ions around a
reference or sample ion.
To understand this distribution of ions, however, one must be suf-
ficiently attuned to mathematical language to read the physical significance
of Eq. (3.35). The physical ideas implicit in the distribution will therefore
be stated in pictorial terms. One can say that the central reference ion is
surrounded by a "cloud," or "atmosphere," of excess charge (Fig. 3.11).
This ionic cloud extends into the solution (i.e., r increases), and the excess
charge density (2 decays with distance r in an exponential way. The
excess charge residing on the ion cloud is opposite in sign to that of the
 ION-ION INTERACTIONS 195

CENTRAL POSITIVE ION

_ SURROUNDED BY A
CLOUD OF NET CIiARGE
EOUAL AND OPPOSITE
TO TIiAT OF THE CENTRAL
ION

Fig. 3.11. The distribution of excess charge

density around a central ion can be pictured
as a cloud, or atmosphere, of net charge
around the central ion.

central ion. Thus, a positively charged reference ion has a negatively charged
ion atmosphere, and vice versa (Fig. 3.12).
Up to now, the charge density at a given distance has been discussed.
The total excess charge contained in the ionic atmosphere which surrounds
the central ion can, however, easily be computed. Consider a spherical shell
of thickness dr at a distance r from the origin, i.e., from the center of the
reference ion (Fig. 3.13). The charge dq in this thin shell is equal to the
charge density er times the volume 4nr2 dr of the shell, i.e.,

dq = eAnr 2dr (3.36)

The total charge qcloud contained in the ion atmosphere is obtained by

summing the charges dq contained in all the infinitesimally thick spherical
shells. In other words, the total excess charge surrounding the reference
ion is computed by integrating dq (which is a function of the distance r
from the central ion) from a lower limit corresponding to the distance from
the central ion at which the cloud is taken to commence to the point where
the cloud ends. Now, the ion atmosphere begins at the surface of the ion;
so the lower limit depends upon the model of the ion. The first model
chosen by Debye and HUckel was that of point-charge ions, in which case
the lower limit is r = o. The upper limit for the integration is r -- 00 be-

NEGATIVELY- CHARGED
IONIC CLOUD

Fig. 3.12. A positively charged ion has a negatively

charged ionic cloud, and vice versa.
196 CHAPTER 3

dr

CHARGE IN SHELL. dq
• P, 4n~r

Fig. 3.13. A spherical shell, of thickness dr, at a

distance r from the center of the reference ion.

cause the charge of the ionic cloud decays exponentially into the solution
and becomes zero only in the limit r -+ 00.
Thus,
r-+oo
f r~O fr-+oo
qcloud = dq = r~O eAnr 2 dr (3.37)

and, by substituting for er from Eq. (3.35), the result is

= - Z ieO fr oo
-+
r~O e- "r ("r) d("r) (3.38)

The integration can be done by parts (Appendix 3.2), leading to the result

(3.39)

which means that a central ion of charge +zieO is enveloped by a cloud

containing a total charge of - ZieO (Fig. 3.l4). Thus, the total charge on
the surrounding volume is just equal and opposite to that on the reference
ion. This is of course precisely how things should be so that there can be
electroneutrality for the ionic solution taken as a whole; a given ion, to-
gether with its cloud, has a zero net charge.
How is this equal and opposite charge of the ion atmosphere distrib-
uted in the space around the central ion? It is seen from Eqs. (3.35) and
(3.36) that the net charge in a spherical shell of thickness dr and at a distance
r from the origin is
(3.40)

Thus, the excess charge on a spherical shell varies with r and has a maximum
 ION-ION INTERACTIONS 197

THIS CHARGE(-Ze.,)GIVES THE

SAME EFFECT AS THE ION I C
CLOUD

(0)
NEGATIVELY-CHARGED
IONIC CLOUD
WITH CHARGE -Zi eo

Fig. 3.14. The total charge -zje o on the ionic cloud

is just equal and opposite to that +zjeo on the central
ion.

value for a value of r given by

0= dq
dr
d
= dr [-zie ox2(e-><rr)]
d
= -z·e X2_ (e-><rr)
! 0 dr
(3.41)

Since (ZjeoX2) is finite, Eq. (3.41) can be true only when

o= e-><r - rxe-><r
or
(3.42)

Hence, the maximum value of the charge contained in a spherical

shell (of infinitesimal thickness dr) is attained when the spherical shell is
at a distance r = X- I from the reference ion (Fig. 3.15). For this reason
(but see also Section 3.3.9), X-I is known as the thickness, or radius, of the
ionic cloud which surrounds a reference ion. An elementary dimensional
analysis [e.g., of Eq. (3.43)] will indeed reveal that X- I has the dimensions
of length. Also, X-I is sometimes referred to as the Debye-Hiickel reciprocal
length.
 CHAPTER 3

dq

CHARGE ENCLOSED
IN A dr-THICK
SPHERICAL SHELL
/ d q IS MAXIMUM AT r.~-I

..L
(I/~)
OISTANCE IN I{-I UNITS

Fig. 3.15. The distance variation (in ,,-1 units) of the

charge dq enclosed in a dr-thick spherical shell, showing
that dq is a maximum at r = ,,-1.

It may be recalled that X-I is given [from Eq. (3.20)] by

(3.43)

As the concentration tends toward zero, the cloud tends to spread out
increasingly (Fig. 3.16). Values of the thickness of the ion atmosphere
for various concentrations of the electrolyte are presented in Table 3.2.

320

280

240

200

160
If-I
0
(inA) 120 f-

80 r-
40 r-

0
0,04 0'12 0,20 0,28
C"2
Fig. 3.16. The variation in the thickness ,,-1 of the ionic
cloud as a function of electrolyte concentration.
 ION-ION INTERACTIONS 199

TABLE 3.2
Thickness of Ionic Atmosphere (in Angstroms) at Various Concentrations
and for Various Types of Salts

Type of salt
C,
moles Iiter- 1
1,1 1,2 2,2 1.3

10-4 304 176 152 124

10- 3 96 55.5 48.1 39.3
10-' 30.4 17.6 15.2 12.4
10- 1 9.6 5.5 4.8 3.9

3.3.9. How Much Does the Ionic Cloud Contribute to the Elec-
trostatic Potential 'Pr at a Distance r from the Central Ion ?
An improved feel for the effects of ionic clouds emerges from con-
sidering the following interesting problem.
Imagine, in a thought experiment, that the charge on the ionic cloud
does not exist. There is only one charge now, that on the central ion. What
is the potential at distance r from the central ion? It is simply given by
the familiar formula for the potential at a distance r from a single charge,
namely,
(3.44)

Then, let the charge on the cloud be switched on. The potential VJr
at the distance r from the central ion is no longer given by the central ion
only. It is given by the law of superposition of potentials (Fig. 3.17), i.e.,

1/<"",- I
r

(a) (c)
Fig. 3.17. The superposition of the potential 'Pion due to the ion
and that 'Pcloud due to the cloud yields the total potential at a dis-
tance , from the central ion.
200 CHAPTER 3

is the sum of the potential due to the central ion and that due to the ionic
'f/Jr
cloud
"Pr = 'f/Jion + 'f/Jcloud (3.45)

The contribution 'f/Jcloud can thus be easily found. One rearranges Eq.
(3.45) to read
'f/Jcloud = 'f/Jr - 'f/Jion (3.46)

and substitutes for 'f/Jion with Eq. (3.44) and for 'f/Jr with the Debye-Hilckel
expression [Eq. (3.33)]. Then,

z·e
= ~ (e- xr - I) (3.47)
tor

The value of x [cf Eq. (3.20)] is proportional to ~ niozi2eo2. In sufficiently

dilute solutions, ~ ntzleo2 can be taken as sufficiently small to make xr ~ I,

(3.48)

and, based on this approximation,

(3.49)

By introducing the expressions (3.44) and (3.49) into the expression (3.45)
for the total potential 'f/Jr at a distance r from the central ion, it follows that

(3.50)

The second term, which arises from the cloud, reduces the value of the
potential to a value less than that if there were no cloud. This is consistent
with the model; the cloud has a charge opposite to that on the central ion
and must, therefore, alter the potential in a sense opposite to that due to
the central ion.
The expression
ZieO
'f/Jcloud = - 8X-1 (3.49)

leads to another, and helpful, way of looking at the quantity X-I. It is

seen that "Pcloud is independent of r, and, therefore, the contribution of the
 ION-ION INTERACTIONS 201

CLOUO OF NET CHARGE

,...-_ _ _~.L--__. -Z ; eo

Fig. 3.18. The contribution 1I'c1oud of the ionic

cloud to the potential at the central ion is equiv-
alent to the potential due to a single charge,
equal and opposite to that of the central ion,
placed at a distance x- 1 from the central ion.

cloud to the potential at the site of the point-charge central ion can be
considered given by Eq. (3.49) above. But, if the entire charge of the ionic
atmosphere [which is - Zieo as required by electroneutrality-cf Eq. (3.39)]
were placed at a distance X-I from the central ion, then the potential prod-
uced at the reference ion would be -ZieO/(SX-I). It is seen therefore from
Eq. (3.49) that the effect of the ion cloud, namely, "Pclaud, is equivalent to
that of a single charge, equal in magnitude but opposite in sign to that of
the central ion, placed at a distance X-I from the reference ion (Fig. 3.18).
This is an added-and more important-reason that the quantity X-I is
termed the effective thickness or radius of the ion atmosphere surrounding
a central ion (cf Section 3.3.8).

3.3.10. The Ionic Cloud and the Chemical-Potential Change

Arising from lon-Ion Interactions

It will be recalled (see Section 3.3.1) that it was the potential at the
surface of the reference ion which needed to be known in order to calculate
the chemical-potential change L'l,ui-l arising from the interactions between
a particular ionic species i and the rest of the ions of the solution, i.e., one
needed to know "P in Eq. (3.3),

(3.3)

It was to obtain this potential "P that Debye and Hiickel conceived
their model of an ionic solution. The analysis threw up the picture of an
ion being enveloped in an ionic cloud. But what is the origin of the ionic
cloud? It is born of the interactions between the central ion and the ions
of the environment. If there were no interactions (e.g., coulombic forces
202 CHAPTER 3

between ions), thermal forces would prevail, distribute the ions randomly
(e = 0), and wash out the ionic atmosphere. It appears therefore that the
simple ionic cloud picture has not only led to success in describing the
distribution of ions but also given the electrostatic potential "Pcloud at the
surface of a reference ion due to the interactions between this reference ion
and the rest of the ions in the solution (the quantity required for reasons
declared in Section 3.3.1).
Thus, the expression (3.49) for "Pcloud can be substituted for "P in Eq.
(3.3) with the result that
LI - - NA (Zi eO)2 (3.51)
t-ti-I - -2- 8,,-1

The Debye-Hiickel ionic-cloud model for the distribution of ions in

an electrolytic solution has permitted the theoretical calculation of the
chemical-potential change arising from ion-ion interactions. But, how is
this theoretical expression to be checked, i.e., connected with a measured
quantity? It is to this testing of the Debye-Hiickel theory that attention will
now be turned.

Further Reading
1. P. Debye and E. Hiickel, Z. Physik, 24, 185 (1923).
2. H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solutions,
3rd ed., Reinhold Publishing Corp., New York, 1958.
3. R. A. Robinson and R. H. Stokes, Electrolyte Solutions, 2nd ed., Butter-
worth's Publications: Ltd., London, 1959.

3.4. ACTIVITY COEFFICIENTS AND ION-ION INTERACTIONS

3.4.1. The Evolution of the Concept of Activity Coefficient

The existence of ions in solution, of interactions between these ions,

and of a chemical-potential change Llt-ti-l arising from ion-ion interactions
have all been taken to be self-evident in the treatment hitherto presented
here. This, however, is a modern point of view. The thinking about elec-
trolytic solutions actually developed along a different path.
Ionic solutions were at first treated in the same way as nonelectrolytic
solutions, though the latter do not contain (interacting) charged species.
The starting point was the classical thermodynamic formula for the chemical
potential Iti of a nonelectrolyte solute

Iti = t-tP + RTln Xi (3.52)

 ION-ION INTERACTIONS 203

In this expression, Xi is the concentration t of the solute in mole fraction

units, and f1t is its chemical potential in the standard state, i.e., when Xi
assumes a standard or normalized value of unity

when (3.53)

Since the solute particles in a solution of a nonelectrolyte are un-

charged, they do not engage in long-range coulombic interactions. The
short-range interactions arising from dipole-dipole or dispersion forces
become significant only when the mean distance between the solute paiticles
is small, i.e., when the concentration of the solute is high. Thus, one can
to a good approximation say that there are no interactions between solute
particles in dilute nonelectrolyte solutions. Hence, if Eq. (3.52) for the
chemical potential of a solute in a nonelectrolyte solution (with noninter-
acting particles) is used for the chemical potential of a ionic species i in
an electrolytic solution, then it is tantamount to ignoring the long-range
coulombic interactions between ions. In an actual electrolytic solution,
however, ion-ion interactions operate whether one ignores them or not.
It is obvious, therefore, that measurements of the chemical potential f1i
of an ionic species-or, rather, measurements of any property which depends
on the chemical potential-would reveal the error in Eq. (3.52), which is
blind to ion-ion interactions. In other words, experiments show that, even
in dilute solutions,

In this context, a frankly empirical approach was adopted by earlier

workers, not yet blessed by Debye and Hlickel's light. Solutions that
obeyed Eq. (3.52) were characterized as ideal solutions since this equation
applies to systems of non interacting solute particles, i.e., ideal particles.
Electrolytic solutions which do not obey the equation were said to be non-
ideal. In order to use an equation of the form of (3.52) to treat nonideal
electrolytic solutions, an empirical correction factor Ii was introduced by
Lewis as a modifier of the concentration term t
(3.54)

t The value of x;" in the case of an electrolyte derives from the number of moles of ions
in species i actually present in solution. This number need not be equal to the number
of moles of i expected of dissolved electrolyte; if, for instance, the electrolyte is a
potential one, then only a fraction of the electrolyte may react with the solvent to form
ions, i.e., the electrolyte may be incompletely dissociated.
t The standard chemical potential It;" has the same significance here as in Eq. (3.52)
for ideal solutions. Thus, I' ,0 can be defined either as the chemical potential of an ideal
204 CHAPTER 3

It was argued that, in nonideal solutions, it was not just the analytical
concentration Xi of species i, but its effective concentration xiii which
determined the chemical-potential change fii - fit. This effective concen-
tration xdi was also known as the activity ai of the species i, i.e.,

(3.55)

and the correction factor fi, as the activity coefficient. For ideal solutions,
the activity coefficient is unity, and the activity ai becomes identical to the
concentration Xi, i.e.,
when fi = I (3.56)

Thus, the chemical-potential change in going from the standard state

to the final state can be written as

(3.57)

Equation (3.57) summarizes the empirical or formal treatment of the

behavior of electrolytic solutions. Such a treatment cannot furnish a theo-
retical expression for the activity coefficient fi. It merely recognizes that
expressions such as (3.52) must be modified if significant interaction forces
exist between solute particles.

3.4.2. The Physical Significance of Activity Coefficients

For a hypothetical system of ideal (noninteracting) particles, the

chemical potential has been stated to be given by

fii (ideal) = fiio + RTln Xi (3.52)

For a real system of interacting particles, the chemical potential has been
expressed in the form
fii (real) = fiio + RTln Xi + RTlnfi (3.57)

Hence, to analyze the physical significance of the activity coefficient

term in Eq. (3.57), it is necessary to compare this equation with Eq. (3.52).
It is obvious that, when Eq. (3.52) is subtracted from Eq. (3.57), the differ-

solution in its standard state of Xi = 1 or as the chemical potential of a solution in

its state of Xi = 1 and Ii = 1, i.e., Qi = 1. No real solution can have Ii = 1 when
Xi = 1; so, the standard state pertains to the same hypothetical solution as the standard

state of an ideal solution.

 ION-ION INTERACTIONS 205

ence, i.e., Pi (real) - Pi (ideal), is the chemical-potential change L1Pi-I

arising from interactions between the solute particles (ions in the case of
electrolyte solutions). That is,

Pi (real) - Pi (ideal) = L1Pi-I (3.58)

and, therefore,
(3.59)

Thus, the activity coefficient is a measure of the chemical-potential change

arising from ion-ion interactions. There are several well-established meth-
ods of experimentally determining activity coefficients, and these methods are
treated in adequate detail in standard treatises (ef Further Reading at the
end of this section).
Now, according to the Debye-HUckel theory, the chemical-potential
change L1Pi-I arising from ion-ion interactions has been shown to be
given by
(3.51)

Hence, by combining Eqs. (3.51) and (3.58), the result is

(3.60)

Thus, the Debye-HUckel ionic-cloud model for ion-ion interactions has

permitted a theoretical calculation of activity coefficients resulting in Eq.
(3.59).
The activity coefficient in Eq. (3.59) arises from the formula (3.57)
for the chemical potential, in which the concentration of the species i is
expressed in mole fraction Xi units. But one can also express the concen-
tration in moles per liter of solution (molarity) or in moles per kilogram
of solvent (molality). Thus, alternate formulas for the chemical potential
of a species i in an ideal solution read

Pi = pNe) + RTln ei (3.61 )

and
Pi = pNm) + RTln mi (3.62)

where ei and mi are the molarity and molality of the species i, respectively;
and pp(e) and pt(m), the corresponding standard chemical potentials.
When the concentration of the ionic species in a real solution is ex-
pressed as a molarity Ci or a molality mi, there are corresponding activity
206 CHAPTER 3

coefficients Yc and Ym and corresponding expressions for Pi

Pi = pP(c) + RTin Ci + RTln Yc (3.63)

and
(3.64)

3.4.3. The Activity Coefficient of a Single Ionic Species Cannot

Be Measured
Before the activity coefficients calculated on the basis of the Debye-
HUckel model can be compared with experiment, there arises a problem
similar to one faced in the discussion of ion-solvent interactions (Chap-
ter 2).
There, it was realized the heat of hydration of an individual ionic
species could not be measured because such a measurement would involve
the transfer of ions of only one species into a solvent instead of ions of two
species with equal and opposite charges. Even if such a transfer were
physically possible, it would result in a charged solution t and, therefore,
an extra, undesired interaction between the ions and the electrified solution.
The only way out was to transfer a neutral electrolyte (an equal number
of positive and negative ions) into the solvent, but this meant that one
could only measure the heat of interactions of a salt with the solvent and
this experimental quantity could not be separated into the individual ionic
heats of hydration.
Here, in the case of ion-ion interactions, the desired quantity is the
activity coefficient Ii' t which depends through Eq. (3.57) on Pi - pt This
means that one seeks the free-energy change of an ionic solution per mole
of ions of a single species i. To measure this quantity, one would have a
problem similar to that experienced with ion-solvent interactions, namely,
the measurement of the change of free energy of a solution, resulting from
a change in the concentration of one ionic species only.
This change in free energy associated with the addition of one ionic
species only would include an undesired work term representing the elec-
trical work of interaction between the ionic species being added and the
charged solution. t To avoid free-energy changes associated with interacting

t The solution may not be initially charged but will become so once an ionic species is
added to it.
The use of the symbol y for the activity coefficients when the concentration is expressed
in molarities and molalities should be noted. When the concentration is expressed as a
mole fraction, Ii has here been used. For dilute solutions, the numerical values of
activity coefficients for these different systems of units are almost the same.
 ION-ION INTERACTIONS 207

with a solution, it is necessary that, after changing the concentration of the

ionic species, the electrolytic solution should end up uncharged and elec-
troneutral. This aim is easily accomplished by adding an electroneutral
electrolyte containing the ionic species i. Thus, the concentration of sodium
ions can be altered by adding sodium chloride. The solvent, water, maintains
its electro neutrality when the uncharged ionic lattice (containing two ionic
species of opposite charge) is dissolved in it.
When ionic lattices, i.e., salts, are dissolved instead of individual ionic
species, one eliminates the problem of ending up with charged solutions
but another problem emerges. If one increases the concentration of sodium
ions by adding the salt sodium chloride, one has perforce to produce a
simultaneous increase of the concentration of chloride ions. This means,
however, that there are two contributions to the change in free energy
associated with a change in salt concentration: (1) the contribution of the
positive ions, and (2) the contribution of the negative ions.
Since neither the positive nor the negative ions can be added separately,
the individual contributions of the ionic species to the free energy of the
system cannot be determined. Thus, the activity coefficients of individual
ions, which depend by, e.g., (3.63) on the free-energy changes when the
particular individual species alone is added to the solution, are inaccessible
to experimental measurement. One can only measure the activity coefficient
of the net electrolyte, i.e., of at least two ionic species together. It is necessary,
therefore, to establish a conceptual link between the activity coefficient of
an electrolyte in solution (that quantity accessible to experiment) and that
of only one of its ionic species [not accessible to experiment, but calculable
theoretically from (3.60)].

3.4.4. The Mean Ionic Activity Coefficient

Consider a un i-univalent electrolyte MA (e.g., NaCi). The chemical

potential of the M+ ions is [cf (3.58)]

f-lM+ = f-l~1+ + RTln XM+ + RTlnfM+ (3.65)

and the chemical potential of the A-ions is

(3.66)

Adding the two expressions, one obtains

208 CHAPTER 3

What has been obtained here is the change in the free energy of the system
due to the addition of 2 moles of ions-l mole of M + ions and 1 mole of
A - ions-which are contained in I mole of electro neutral salt MA.
Now, suppose that one is only interested in the average contribution
to the free energy of the system from I mole of both M + and A-ions.
One has to divide Eq. (3.67) by 2

At this stage, one can define several new quantities

(3.69)

(3.70)

(3.71)
and
(3.72)

What is the significance of these quantities fl±' fl±o, x±, and f±? It is
obvious they are all average quantities-the mean chemical potential P,±'
the mean standard chemical potential fl ±o, the mean ionic mole fraction x ±,
and the mean ionic-activity coefficient f ±. In the case of fl ± and fl ±o, the
arithmetic mean (half the sum) is taken because free energies are additive;
but, in the case of x ± and f ±, the geometric mean (the square root of the
product) is taken because the effects of mole fraction and activity coefficient
on free energy are multiplicative.
In this notation, Eq. (3.68) for the average contribution of a mole of
ions to the free energy of the system becomes

(3.73)

since a mole of ions is produced by the dissolution of half a mole of salt.

In other words, fl ± is half the chemical potential flMA of the salt. t

(3.74)

t The symbol flMA should not be taken to mean that molecules of MA exist in the solu-
tion; flMA is the observed free-energy change of the system resulting from the disso-
lution of a mole of electrolyte.
 ION-ION INTERACTIONS 209

Thus, a clear connection has been set up between observed free-energy

changes flMA consequent upon the change from a state in which the two
ionic species of a salt are infinitely far apart to a state corresponding to the
given concentration, and its mean ionic-activity coefficient f ±. Hence the
value off± is experimentally measurable. This mean ionic-activity coefficient
cannot, however, be experimentally split into the individual ionic-activity
coefficients. All that can be obtained fromf± is the product of the individual
ionic-activity coefficients [Eq. (3.72)]. The theoretical approach must hence
be to calculate the activity coefficients f+ and f- for the positive and negative
ions [cf Eq. (3.60)] and combine them through Eq. (3.72) into a mean
ionic-activity coefficient f± which can be compared with the experimentally
derived mean ionic-activity coefficient.

3.4.5. The Conversion of Theoretical Activity-Coefficient Ex-

pressions into a Testable Form
Individual ionic-activity coefficients are experimentally inaccessible
(Section 3.4.3); hence, it is necessary to relate the theoretical individual
activity coefficient fi [Eq. (3.64)] to the experimentally accessible mean
ionic-activity coefficient f ± so that the Debye-Hlickel model can be tested.
The procedure is to make use of the relation (3.72)

(3.72)

of which the general form for an electrolyte which dissolves to give v+ z+-
valent positive ions and v_ L-valent negative ions can be shown to be (cf
Appendix 3.3)
(3.75)

where f+ and f- are the activity coefficients of the positive and negative
ions, and
(3.76)

By taking logarithms of both sides of Eq. (3.75), the result is

(3.77)

At this stage, the Debye-Hlickel expressions (3.60) for f+ and f- can

be introduced into Eq. (3.77) to give

(3.78)
210 CHAPTER 3

Since the solution as a whole is electroneutral, v+Z+ must be equal to

V_L, and, therefore,

V+Z+2 + V_Z_2 = V_Z_Z+ + V+Z+Z_

(3.79)

Using this relation in Eq. (3.78), one obtains

I
n
f ± -- _ NA (z+L)eo2
2eRT" (3.80)

Now, one can substitute for" from Eq. (3.43)

,,= ( ekT
4n '"
~
0 2eo2)'
ni Zi (3.43)

but, before this substitution is made, " can be expressed in a different form.
Since
(3.81 )

where c is the concentration in moles per liter, it follows that

(3.82)

Prior to the Debye-Hlickel theory, t L CiZi2 had been empirically introduc-

ed by Lewis as a quantity of importance in the treatment of ionic solutions.
Since it quantifies the charge in an electrolytic solution, it was known as
the ionic strength and given the symbol I

(3.83)

In terms of the ionic strength I, " can be written as [cf Eqs. (3.43), (3.82),

yn
and (3.83)]
= ( 8nNA eo2 (3.84)
" 1000ekT
or as
" = Bli (3.85)
where
B = ( 8nNA eo2
1000ekT
r (3.86)
 ION-ION INTERACTIONS 211

TABLE 3.3
Value of the Parameter 8 for Water at Various Temperatures

Temperature, °C 10- 8 8

0 0.3248
10 0.3264
20 0.3282
25 0.3291
30 0.3301
35 0.3312
40 0.3323
50 0.3346
60 0.3371
80 0.3426
100 0.3488

Values of B for water at various temperatures are given in Table 3.3.

On the basis of the expression (3.85) for ~, Eq. (3.80) becomes

lnf± = - NA(z+L)e o2 Bn (3.87)

2cRT
or
_ _ _1_ NAeo2 l
logf± - 2.303 2cRT B(Z+L)/ (3.88)

For greater compactness, one can define a constant A given by

(3.89)

and write Eq. (3.88) in the form

logf± = -A(Z+L)n (3.90)

For 1 : I-valent electrolytes, Z+ = L = 1 and 1= c, and, therefore,

logf± = -Ac~ (3.91 )

Values of the constant A for water at various temperatures are given in

Table 3.4.
In Eqs. (3.90) and (3.91), the theoretical mean ionic-activity coefficients
are in a form directly comparable with experiment. A quantitative com pari-
212 CHAPTER 3

TABLE 3.4
Values of Constant A for Water at Various Temperatures

Temperature, °C Values of constant A

0 0.4918
10 0.4989
20 0.5070
25 0.5115
30 0.5161
40 0.5262
50 0.5373
60 0.5494
80 0.5767
100 0.6086

son of the experimentally observed activity coefficients with those calculated

with the Debye-Huckel model can now be made.

Further Reading
1. H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solution,
3rd ed., Reinhold Publishing Corp., New York, 1958.
2. R. A. Robinson and R. H. Stokes, Electrolytic Solutions, Butterworth's
Publications, Ltd., London, 1959.
3. G. Kortlim, Treatise on Electrochemistry, Elsevier, Amsterdam, 1965.

3.5. THE TRIUMPHS AND LIMITATIONS OF THE DEBYE-

HUCKEL THEORY OF ACTIVITY COEFFICIENTS

3.5.1. How Well Does the Debye-Huckel Theoretical Expression

for Activity Coefficients Predict Experimental Values?

The approximate theoretical equation

(3.90)

indicates that the logarithm of the activity coefficient must decrease linearly
with the square root of the ionic strength or, in the case of I: I-valent
electrolytes, t with d. Further, the slope of the log!± versus l~ straight line
can be unambiguously evaluated from fundamental physical constants and

t That is, J = ! L: CiZ? For a 1:1 electrolyte, J = HCiP + c;P). As Ci = c; = c, I = c.

 ION-ION INTERACTIONS 213

TABLE 3.5
Experimental Values of Activity Coefficients of Various Electrolytes at
Different Concentrations at 25 0 C

1:1 electrolyte, HCI

Concentration, molal 0.0001 0.0002 0.0005 0.001 0.002
Mean activity coefficient 0.9891 0.9842 0.9752 0.9656 0.9521

2: 1 electrolyte, CaCI.
Concentration, (moles per liter) 0.0018 0.0061 0.0095
Mean activity coefficient 0.8588 0.7745 0.7361

2:2 electrolyte, CdS0 4

Concentration, molal 0.0005 0.001 0.005
Mean activity coefficient 0.774 0.697 0.476

from (z L). Finally, the slope does not depend on the particular electrolyte
j

(i.e., whether it is NaCl or KBr, etc.) but only on its valence type, i.e., on
the charges borne by the ions of the electrolyte, whether it is a 1 : I-valent
or 2 :2-valent electrolyte, etc. These are clear-cut predictions.
Even before any detailed comparison with experiment, one can use
an elementary spot check: At infinite dilution, where the interionic forces
are negligible, does the theory yield the activity coefficient which one
would expect from experiment, i.e., unity? At infinite dilution, e or
I ~ 0, which means that logf± ~ 0 or f ± ~ 1. The properties of an ex-
tremely dilute solution of ions should be the same as those of a solution
containing nonelectrolyte particles. Thus, the Debye-Hiickel theory comes
out successfully from the infinite dilution test.
Further, if one takes the experimental values of the activity coefficient
(Table 3.5) at extremely low electrolyte concentration and plots logf ±
versus n curves, it is seen that: (1) They are linear (el Fig. 3.19), and (2)
they are grouped according to the valence type of the electrolyte (Fig. 3.20).
Finally, when one compares the calculated and observed slopes, it becomes
clear that there is excellent agreement to an error of ±0.5% (Table 3.6 and
Fig. 3.21) between the results of experiment and the conclusions emerging
from an analysis of the ionic-cloud model of the distribution of ions in an
electrolyte. Since Eq. (3.90) has been found to be valid at limiting low
electrolyte concentrations, it is generally referred to as the Debye-Hiiekel
limiting law.
214 CHAPTER 3

(i) EXPERIMENTAL POINTS

-001

log l'
:!

-002

-003

15

Fig. 3.19. The logarithm of the experimental mean ac-

tivity coefficient of Hel varies linearly with the square
root of the ionic strength.

The success of the Debye-Hiickellimiting law is no mean achievement.

One has only to think of the complex nature of the real system, of the
presence of the solvent which has been recognized only through a dielectric

TABLE 3.6
Experimental and Calculated Values of the Slope of log f± - VT for Alcohol
VVater ~ixtures at 25 0 C

Slope
Solvent mole fraction water Dielectric constant
Observed Calculated

1:1 type of salt, Croceo tetranitro diamino cobaltiate

1.00 78.8 0.50 0.50
0.80 54.0 0.89 0.89

1:2 type of salt, Croceo sulfate

1.00 78.8 1.10 1.08
0.80 54.0 1.74 1.76

3:1 type of salt, Luteo iodate

1.00 78.8 1.52 1.51
 ION-ION INTERACTIONS 215

-002

1,1 Electrolyle(HCil
-004

-006

-0-08

-0 10

-0-14

-0·16

-0 ·18

-0·20
2,2 Eleclrolyle(ZnS<l.J

-0·22

-0·25

I 2 3 4 5 6 7 8 9 10 II 12 13 14
II XI02
Fig. 3.20. The experimental log f ± versus I! straight-
line plots for different electrolytes can be grouped ac-
cording to valence type.

constant, of the simplicity of the coulomb force law used, and, finally, of
the fact that the ions are not point charges, to realize (ef Table 3.6) that
the simple ion-cloud model has been brilliantly successful-almost unex-
pectedly so. It has grasped the essential truth about electrolytic solutions,
albeit about solutions of extreme dilution. The success of the model is so
remarkable and the implications so wide (see Section 3.5.6), that the Oebye-
Huckel approach is to be regarded as one of the most significant pieces of
theory in the ionics part of electrochemistry.
It is a theme of this book that model-oriented electrochemistry is to a
216 CHAPTER 3

_____ EXPEIUMENTAL POINTS

TMEORETICAL CURVE
-0'02

13

Fig. 3.21. The comparison of the experimentally

observed mean activity coefficients of Hel and those
that are calculated from the Debye-Huckel limiting
law.

great extent the result of the application of electrostatics to chemistry.

From this point of view, the Debye-Hiickel approach is an excellent example
of electrochemical theory. Electrostatics is introduced into the problem
in the form of Poisson's equation, and the chemistry is contained in the
Boltzmann distribution law and the concept of true electrolytes (Section
3.2). The union of the electrostatic and chemical modes of description to
give the linearized Poisson-Boltzmann equation illustrates therefore a
characteristic development of electrochemical thinking.
It is hence not surprising that the Poisson-Boltzmann approach has
been used frequently in computing interactions between charged entities.
Mention may be made of the Gouy theory (Fig. 3.22) of the interaction

DISTANCE FROM ELECTRODE

(0) (b)

Fig. 3.22. An electrode immersed in an ionic solution is

often enveloped by an ionic cloud [see Fig. 3.22 (a)) in
which the excess charge density varies with distance as
shown in Fig. 3.22 (b).
 ION-ION INTERACTIONS 217

ELECTROLYTE
EXCESS
CHARGE
DENSITY
~
DISTANCE IN TO SEMI-
SPACE-CHARGE OF EXCESS CONDUCTOR
ELECTRONS OR HOLES

(a) (b)

Fig. 3.23. (a) A space charge produced by excess electrons or

holes often exists inside the semiconductor. (b) The space charge
density varies with distance from the semiconductor-electrolyte
interface.

between a charged electrode and the ions in a solution (see Section 7.4).
Other examples are the distribution (Fig. 3.23) of electrons or holes inside
a semiconductor and in the vicinity of the semiqmductor electrolyte inter-
face (see Section 7.7), and the distribution (Fig. 3.24) of charges near a
polyelectrolyte molecule or a colloidal particle (see Section 7.8).
However, one must not overstress the triumphs of the Debye-Htickel
limiting law [Eq. (3.90)]. Models are always simplifications of reality.
They never treat all its complexities, and, thus, there can .never be a perfect
fit between experiment and the predictions based on a model.
What, then, are the inadequacies of the Debye-Hiickel limiting law?
One does not have to look far. If one examines the experimental logf ±
versus n curve, not just in the extreme dilution regions, but at higher
concentrations, it turns out that the simple Debye-Htickel limiting law
falters. The plot of logf± versus [il is a curve (Fig. 3.25 and Table 3.7)

EXCESS
CHARGE
DENSITY

DISTANCE FRDM SURFACE OF

COLLOIDAL PARTICLE

(a) (b)

Fig. 3.24. (a) A colloidal particle is surrounded by an ionic

cloud of excess charge density. (b) The excess charge density
in the cloud varies with distance from the surface of the col-
loidal particle.
218 CHAPTER 3

EXPERIMENTAL
CURVE
-0-2

logft. LIMITING LAW

EQUATION 5

-0'6

- 0 .8 '------L_-'-_....L_L---'~......J.._ _'
0·2 0-4 0-6 0-8 1·0 1·2 1·4
1111

Fig. 3.25. The experimental log f± versus I! curve is a

straight line only at extremely low concentrations.

and not a straight line as promised by Eq. (3.90). Further, the curves depend
not only on valence type (e.g., 1:1 or 2:2) but also (Fig. 3.26) on the part-
icular electrolyte (e.g., NaCl or KCI).
It appears that the Debye-Hlickel law is the law for the tangent to
the logf± versus /! curve at very low concentrations, say, up to O.OIN for
1:1 electrolytes in aqueous solutions. At higher concentrations, the model
must be improved. What refinements can be made?

,·0
\
\
0·9 \
\
\
l 0'8
±
0·7

0·6

O' 5 '---_.L..-_-'--_...L.._....L_ _'

o 0·5 1·0 1·5 2'0 2'5

Jm -
Fig. 3.26. Even though NaCI and KCI are
1 : 1 electrolytes, their activity coefficients
vary in different ways with concentration
directly one examines to higher concentra-
tions.
 ION-ION INTERACTIONS 219

TABLE 3.7
Comparison of Calculated [Eq. (3.90)] and Experimental Values of log f±
for NaCI at 25°C

Concentration, molal - logf± experimental - logf± calculated

0.001 0.0155 0.0162

0.002 0.0214 0.0229
0.005 0.0327 0.0361
0.01 0.0446 0.0510
0.02 0.0599 0.0722

3.5.2. Ions Are of Finite Size, Not Point Charges

One of the general procedures for refining a model which has been
successful in an extreme situation is to liberate the theory from its approx-
imations. So one has to recall what approximations have been used to
derive the Debye-Hilckel limiting law. The first one that comes to mind
is the point-charge approximation. t One now asks: Is it reasonable to
consider ions as point charges?
It has been shown (cf Section 3.3.8) that the mean thickness X-I of
the ionic cloud depends on the concentration. As the concentration of a
1:1 electrolyte increases from O.OOIN to O.OIN to O.IN, X-I decreases from
about 100 to 30 to about 10 A. This means that the relative dimensions of
the ion cloud and of the ion change with concentration. Whereas the radius
of the cloud is 100 times the radius of the ion at O.OOIN, it is only about 10
times the dimensions of an ion at O.lN. Obviously, under these latter cir-
cumstances, an ion cannot be considered a geometrical point charge in
comparison with a dimension only 10 times its size (Fig. 3.27). The more
concentrated the solution, i.e., the smaller the size X-I of the ion cloud
(Section 3.3.8), the less valid is the point-charge approximation.
If, therefore, one wants the theory to be applicable to 0.1 N solutions
or to solutions of even higher concentration, the finite size of the ions must
be introduced into the mathematical formulation.
To remove the assumption that ions can be treated as point charges,

t Another approximation in the Debye-Htickel model involves the use of Poisson's

equation, which is based on the smearing-out of the charges into a continuously varying
charge density. At high concentrations, the mean distance between charges is low, and
the ions see each other as discrete point charges, not as smoothed-out charges. Thus,
the use of Poisson's equation becomes less and less justified as the solution becomes
more and more concentrated.
220 CHAPTER 3

ION CLOUO

Fig. 3.27. At O.1N, the thickness of the

ion cloud is only 10 times the radius of the
central ion.

it is necessary, at first, to recall at what stage in the derivation of the theory

the assumption was invoked.
The linearized P-B equation involved neither the point-charge approx-
imation nor any considerations of the dimensions of the ions. Hence, the
basic differential equation
1 8
f2 Tr
(2 8'1fT ) 2
r dr = 'X '1fT (3.21 )

and its general solution, i.e.,

e- XT e+ XT
'If r = Ar - + Br- (3.28)

can be taken as the basis for the generalization of the theory for finite-sized
ions.
As before (cf Section 3.3.7), the integration constant B must be zero
because, otherwise, one cannot satisfy the requirement of physical sense
that, as r --+ 00, 'If --+ O. Hence, Eq. (3.28) reduces to
e- XT
1IJ
rT
= Ar - (3.29)

In evaluating the constant A, a procedure different from that used

after (3.29) is adopted. The charge dq in any particular spherical shell
(of thickness dr) situated at a distance r from the origin is, as argued earlier,

dq = eAnr 2 dr (3.36)

The charge density eT is obtained thus

eT -_ - 4n
B [18Tr (28'1f)]
~ r Tr -
__ B
4n
2
'X 'lfl' (3.34)

and, inserting the expression for '1fT from Eq. (3.29), one obtains

(3.92)
 ION-ION INTERACTIONS 221

Fig. 3.28. For a finite-sized ion, the ion atmos-

phere starts at a distance a from the center of the
reference ion.

Thus, by combining Eqs. (3.36) and (3.92),

(3.93)

The total charge in the ion cloud qcloud is, on the one hand, equal to - Zieo
[cf Eq. (3.41)] as required by the e1ectroneutrality condition and, on the
other hand, the result of integrating dq. Thus,

qcloud = -z·e
, 0 =
. I= dq dr = -AX I= e-XTr dr
?
2f
? (3.94)

What lower limit should be used for the integration? In the point-
charge model, one used a lower limit of zero, meaning that the ion cloud
commences from zero (i.e., from the surface of a zero-radius ion) and
extends to infinity. But now the ions are taken to be of finite size, and a
lower limit of zero is obviously wrong. The lower limit should be a distance
corresponding to the distance from the ion center at which the ionic atmos-
phere starts (Fig. 3.28).
As a first step, one can use for the lower limit of the integration a
distance parameter which is greater than zero. Then, one can go through
the mathematics and later worry about the physical implications of the
ion-size parameter. Let this procedure be adopted and symbol a be used
for the ion-size parameter.
222 CHAPTER 3

One has, then,

= - Ae f~ ",re-(xr) d",r (3.95)

As before (ef Appendix 3.2), one can integrate by parts, thus,

(3.96)

Hence, inserting Eq. (3.96) in Eq. (3.95), one obtains

(3.97)

from which
A=zieo~ (3.98)
e 1 ",a +
Using this value of A in Eq. (3.30), one obtains a new and less approx-
imate expression for the potential "Pr at a distance r from a finite-size central
ion,
_ZieO ~ e- xr
(3.99)
"Pr - e 1 + ",a r

3.5.3. The Theoretical Mean Ionic-Activity Coefficient in the

Case of Ionic Clouds with Finite-Sized Ions

Once again (ef Section 3.3.9), one can use the law of superposition
of potentials to obtain the ionic-atmosphere contribution "Pclollu to the
potential "Pr at a distance r from the central ion. From Eq. (3.46), i.e.,

"Pciolld = "Pr - "Pion (3.46)

it follows by substitution of the expression (3.99) for "Pr and Eq. (3.44)
for "Pion that
ZieO ex(a-r) ZieO
"Pciolld = er I + ",a - er
z.e [eX(a-r) ]
=~ ----1 (3.100)
er 1 + ",a
 ION-ION INTERACTIONS 223

It will be recalled, however, that, in order to calculate the activity

coefficient from the expressions

(3.59)
and
(3.3)

i.e., from
(3.101)

it is necessary to know 1p, which is the potential at the surface of the ion
due to the surrounding ions, i.e., due to the cloud. Since, in the finite-
ion-size model, the ion is taken to have a size a, it means that 1p is the value
of 1pcloud at r = a,
1p = 1pc1oud r=a (3.102)

The value of 1pcloud at r = a is got by setting r = a in Eq. (3.100). Hence,

(3.103)

By substitution of the expression (3.103) for 1p = 1pcloudlr=a) in Eq.

(3.101), one obtains

(3.104)

This individual ionic-activity coefficient can be transformed into a mean

ionic-activity coefficient by the same procedure as for the Debye-Hlickel
limiting law (ef Section 3.4.12). On going through the algebra, one finds
that the expression for logf± in the finite-ion-size model is

(3.105)

It will be recalled, however, that the thickness x- 1 of the ionic cloud can
be written as [Eq. (3.85)]
x = BIt (3.85)

Using this notation, one ends up with the final expression

1 f __ A(Z+L)Ji
og ± - 1 + Bali (3.106)
224 CHAPTER 3

If one compares Eq. (3.105) of the finite-ion-size model with Eq. (3.90)
of the point-charge approximation, it is clear that the only difference be-
tween the two expressions is that the former contains a term l/(l + %a)
in the denominator. Now, one of the tests of a more general version of a
theory is the correspondence principle, i.e., the general version of a theory
must reduce to the approximate version under the conditions of applicability
of the latter. Does the Eq. (3.105) from the finite-ion-size model reduce to
Eq. (3.90) from the point-charge model?
Rewrite Eq. (3.105) in the form

(3.107)

and consider the term a/%-I. As the solution becomes increasingly dilute,
the radius %-1 of the ionic cloud becomes increasingly large compared with
the ion size, and, simultaneously, a/[1 becomes increasingly small comp-
ared with unity, or

(3.108)

Thus, directly the solution is sufficiently dilute to make a ~ ;cl, i.e., to

make the ion size insignificant in comparison with the radius of the ion
atmosphere, the finite-ion-size model Eq. (3.105) reduces to the correspond-
ing Eq. (3.90) of the point-charge model because the extra term I (I - a!%-I)
tends to unity

(3.109 )

The physical significance of a.:[1 ~ I is that. at wry low concen-

trations. the ion atmosphere has such a large radius compared with that
of the ion that one need not consider the ion as having a finite size a.
Considering a%-I ~ I is tantamount to rewrting to the point-charge model.
One can now proceed rapidly to compare this theoretical expression
for logf± with experiment: but \\hat \alue of the ion-size parameter should
be used? The time has come to worry about the precise physical meaning
of the parameter a which was introduced to allow for the finite size of ions.

3.5.4. The lon-Size Parameter a

One can at first try to speculate on \\hat value of the ion-size parameter
is apprl)priate. A lower limit is the sum of the crl'sta//ograplzic radii of the
 ION-ION INTERACTIONS 225

qxf)
~SUM OF THE CRYSTALLOGRAPHIC
RADII OF THE IONS

(a)

WATER MOLECULE

- - - SUM OF THE RADII OF

THE SOLVATED IONS
(b)

SOLVATION SHELLS
CRUSHED

(e)

Fig. 3.29. The ion-size parameter cannot

be (a) less than the sum of the crystallo-
graphic radii of the ions or (b) more than the
sum of the radii of the solvated ions and is
most probably (c) less than the sum of the
radii of the solvated ions because the solva-
tion shells may be crushed.

posItive and negative ions present in solution; ions cannot come closer
than this distance [Fig. 3.29(a)]. But, in a solution, the ions are generally
solvated (ef Chapter 2). So perhaps the sum of the solvated radii should
be used [Fig. 3.29(b)]. However, when two solvated ions collide, is it not
likely [Fig. 3.29(c)] that their hydration shells are crushed to some extent?
This means that the ion-size parameter a should be greater than the sum
of the crystallographic radii and perhaps less than the sum of the solvated
radii. It should best be called the mean distance of closest approach, but,
beneath the apparent wisdom of this term, there lies a measure of ignorance.
For example, an attempted calculation of just how crushed together two
solvated ions are would involve many difficulties.
226 CHAPTER 3

-0.01,-----------------,

-0.02 )( Experimental POints

-003
......
'"
o -0.04
.J )(

-o.a;

-QOI

-0.02

...... -0.03

'"o
.J -0.04

-0.05

-0.0.6

o 2 16

Fig. 3.30. Procedure for recovering the ion-size pa-

rameter from experiment and then using it to produce a
theoretical log f± versus n
curve which can be compared
with an experimental curve.

To circumvent the uncertainty in the quantitative definition of a, it

is best to regard it as a parameter in Eq. (3.106), i.e., a quantity, the numer-
ical value of which is left to be calibrated or adjusted on the basis of exper-
iment. The procedure (Fig. 3.30) is to assume that the expression for logf±
[Eq. (3.106) 1is correct at one concentration, then to equate this theoretical
expression to the experimental value of log/" corresponding to that con-
centration. and to solve the resulting equation for a. Once the ion-size
parameter. or mean distance of closest approach. is thus obtained at one
concentration. the value can be used for the calculation of values of the
activity coefficient over a range of other and higher concentrations. Then,
the situation is regarded as satisfactory if the value of a obtained from
 ION-ION INTERACTIONS 227

experiments at one concentration can be used in Eq. (3.106) to reproduce

the results of experiments over a range of concentrations.

3.5.5. Comparison of the Finite-lon-Size Model with Experiment

After taking into account the fact that ions have finite dimensions
and cannot therefore be treated as point charges, the following expression
has been derived for the logarithm of the activity coefficient:

I j' __ A(Z+L)n
og ± - 1 + Ban (3.106)

How does the general form of this expression compare with the Debye-
Huckel limiting law as far as agreement with experiment is concerned? To
see what the extra term (1 +
Ban )-1 does to the shape of the logf± versus
Ii curve, one can expand it in the form of a binomial series

1
+ X)-l + -x2! -
2
(1 1- x (3.110)
-- =
1 +x =

and use only the first two terms. Thus,

1
1 + Ban = 1 - Ban (3.111)

and, therefore,
logf±,....., -A(Z+L)n(1 - Bali) (3.112)

,....., - A (Z+L)n + constant(Ii)2 (3.113)

This result is encouraging. It shows that the 10gf± versus n curve

give values of logf± higher than those given by the limiting law, the devia-
tion increasing with concentration. In fact, the general shape of the predicted
curve (Fig. 3.31) is very much on the right lines.
The values of the ion-size parameter, or closest distance of approach,
which are recovered from experiment are physically reasonable for many
electrolytes, They lie around 3 to 5 A, which is greater than the sum of the
crystallographic radii of the positive and negative ions and pertains more
to the solvated ion (Table 3.8).
By picking on a reasonable value of the ion-size parameter a, inde-
pendent of concentration, it is found that, in many cases, Eq. (3.112) gives
a very good fit with experiment, often for ionic strengths up toward 0.1.
For example, on the basis of a = 4.0 A, Eq. (3.112) gives an almost exact
228 CHAPTER 3

-002
log I:!;
-0,04

-0'06

I!
Fig. 3.31. Comparison of the experimental mean activity
coefficients with theory for Eq. (3.112).

agreement up to O.02M in the case of sodium chloride (Fig. 3.32 and

Table 3.9).
The ion-size parameter a has done part of the job of extending the
range of concentration in which the Debye-HUckel theory of ionic clouds
agrees with experiment. But has it done the whole job? One must there-
fore start looking for discrepancies between theory and fact and for the
less satisfactory features of the model.
The most obvious drawback of the finite-ion-size version of the Debye-
HUckel theory lies in the fact that a is an adjustable parameter. When
parameters which have to be taken from experiment enter a theory, they
imply that the physical situation has been incompletely comprehended or
is too complex to be mathematically analyzed. [n contrast, the constants
of the limiting law were calculated without recourse to experiment.
The best illustration of the fact that a has to be adjusted is its con-
centration dependence. As the concentration changes, the ion-size para-

TABLE 3.8
Values of lon-Size Parameter for a Few Electrolytes

Salt a, A

HCl 4.5
HBr 5.2
LiCI 4.3
NaCl 4.0
KCl 3.6
 ION-ION INTERACTIONS 229

-(}Ol

-(}02
10gIO l'
±
+ -(}o3
-0'04

-0,05

-0,06

Jlx 10 2

Fig. 3.32. Comparison of the experimental mean ar.tivity

coefficients for sodium chloride with the theoretical log f±
versus ,2 curve based on Eq. (3.112) with a == 4.0 A.

meter has to be modified (Fig. 3.33). Further, for some electrolytes at

higher concentrations, a has to assume quite impossible (i.e., large negative,
irregular) values to fit the theory to experiment (Table 3.10).
Evidently, there are factors at work in an electrolytic solution which
have not yet been reckoned with, and the ion-size parameter is being asked
to include the effects of all these factors simultaneously, even though these
other factors probably have little to do with the size of the ions and

TABLE 3.9
Experimental Mean Activity Coefficients and Those Calculated from Eq.
(3.112) with a = 4.0 A at 25 0 C at Various Concentrations of NaCI

Experimental mean activity coefficient

Molality Calculated
-Iogf±

0.001 0.0155 0.0155

0.002 0.0214 0.0216
0.005 0.0327 0.0330
0,0] 0.0446 0.0451
0.02 0.0599 0.0609
230 CHAPTER 3

16

14

12

0. in AO 10

Fig. 3.33. The variation of the ion-size pa-

rameter with concentration of NaCI.

may vary with concentration. If this were so, the ion-size parameter a,
calculated back from experiment, would indeed have to vary with con-
centration. The problem, therefore, is: What factors, forces, and interactions
were neglected in the Debye-Hiickel theory of ionic clouds?

3.5.6. The Oebye-Huckel Theory of Ionic Solutions: An As-

sessment
It is appropriate at this stage to register the achievement in the theory
of ionic solutions described thus far.
Starting with the point of view that ion-ion interactions are bound to
operate in an electrolytic solution, the chemical-potential change Ll,ui-l,
in going from a hypothetical state of noninteracting ions to a state in which
the ions of species i interact with the ionic solution, was considered a
quantitative measure of these interactions. As a first approximation, the

TABLE 3.10
Values of Parameter a at Higher Concentrations

Concentration, Concentration,
Value of a for HCl, A Value of a for LiCl, A
molality molality

13.8 2 41.3
1.4 24.5
1.8 85.0 2.5 -141.9
2 -411.2
2.5 27.9 3 -26.4
3 - 14.8
 ION-ION INTERACTIONS 231

ion-ion interactions were assumed to be purely coulombic in origin. Hence,

the chemical-potential change arising from the interactions of species i with
the electrolytic solution is given by the Avogadro number times the elec-
trostatic work W resulting from taking a discharged reference ion and
charging it up in the solution to its final charge. In other words, the charging
work is given by the same formula as that used in the Born theory of solv-
ation, i.e.,
(3.3)

where "P is the electrostatic potential at the surface of the reference ion,
contributed by the other ions in the ionic solution. The problem, therefore,
was to obtain a theoretical expression for the potential "P. This involved
an understanding of the distribution of ions around a given reference ion.
It was in tackling this apparently complicated task that appeal was
made to the Debye-Hilckel simplifying model for the distribution of ions
in an ionic solution. This model treats only one ion-the central ion-
as a discrete charge, the charge of the other ions being smoothed out to
give a continuous charge density. Because of the tendency of negative
charge to accumulate near a positive ion, and vice versa, the smoothed-out
positive and negative charge densities do not cancel out; rather, their
imbalance gives rise to an excess local charge density Or' which of course
dies away toward zero as the distance from the central ion is increased. Thus,
the calculation of the distribution of ions in an electrolytic solution reduces
to the calculation of the variation of excess charge density r!r with distance
r from the central ion.
The excess charge density r!r was taken to be given, on the one hand,
by Poisson's equation of electrostatics

(!r
BId
= - 4n f2 dr r
(2 (d"Pr)
jf (3.17)

and, on the other, by the linearized Boltzmann distribution law

(3.18)

The result of equating these two expressions for the excess charge density
is the fundamental partial differential equation of the Debye-Hilckel model,
the linearized P-B equation (cf Fig. 3.34)

f21 dr
d (2r (d"Pr
jf
)
= ~
2"Pr (3.21 )
232 CHAPTER 3

Poisson Equation Boltzmann Equation

o., = __
f [_I ~ ~I,±-j] c', = ~ n,o:,cu e"r~ -

I
4:r I' £II .II'

Linearized Boltzmann Electroneutrality

Equation

kT

~),. =
kT

I
!
Linearized Poisson-Boltzmann Equation

_1 ~ Ir'~l [
4:r v
--_rz~_,tll
u_ , , ,] .
~,.

r' '"~ \ '"~ ,'kT

r~
,/
./r
[r'
\
d~',
./r
I

Fig. 3.34. Steps in the derivation of the linearized Poisson-Boltzmann equation,

 ION-ION INTERACTIONS 233

where
4n '"
= __ n ,OZ ,2e
;,,2 2 (3.20)
ekT"-' • • 0

By assuming that ions can be regarded as point charges, the solution

of the linearized P-B equation turns out to be (cf Fig. 3.35)

(3.33)

Such a variation of potential with distance from a typical (central or ref-

erence) ion corresponded to a charge distribution which can be expressed
as a function of distance r from the central ion by

(3.35)

This variation of the excess charge density with distance around the
central or typical ion yielded a simple physical picture. A reference positive
ion can be thought of as being surrounded by a cloud of negative charge
of radius ;,,-1. The charge density in this ionic atmosphere, or ionic cloud,
decays in the manner indicated by Eq. (3.35). Thus, the interactions be-
tween a reference ion and the surrounding ions of the solution is equivalent
to the interactions between the reference ion and the ionic cloud which, in
the point-charge model, sets up at the central ion a potential 'f{'cloud given by

(3.49)

The magnitude of central ion-ionic-cloud interactions is given by intro-

ducing the expression for 'f{'cloud into the expression (3.3) for the work of
creating the ionic cloud, i.e., setting up the ionic interaction situation.
Thus, one obtains for the energy of such interactions

(3.51)

In order to test these predictions, attention was drawn to an empirical

treatment of ionic solutions. For solutions of noninteracting particles, the
chemical-potential change in going from a solution of unit concentration
to one of concentration Xi is described by the equation

(3.52)
234 CHAPTER 3

Linearized Poisson-
Boltzmann Equations

~~
r" dr
[r2 dr. ]
dr
= x4.'.

solution

Ar" Be"r
..... = - - - - - Boundary Condition
r r

•. -o·
.'t<
Boundary Condition "
...· r = - -
r
.\5 n,o ...... O. i.e.. x ...... 0

A
(1) ..... = -
r

=tt',)

=~e1) e-~'"
'r'r = - -
E r

Fig. 3.35. Steps in the solution of the linearized Poisson-Boltzmann equation for
point-charge ions.
 ION-ION INTERACTIONS 235

However, in the case of an electrolytic solution in which there are

ion-ion interactions, it is experimentally observed that

If one is unaware of the nature of these interactions, one can write an

empirical equation to compensate for one's ignorance

(3.58)

and say that solutions behave ideally if the so-called activity coefficient Ii
is unity, i.e., RTlnii = 0, and, in real solutions'!i *
1. It is clear that Ii
corresponds to a coefficient to account for the behavior of ionic solutions,
which differs from those in which there are no charges. Thus, fi accounts
for the interactions of the charges, so that
N A(zi eo)2
RT In .fi = L1fli-I = - 2 EX- 1 (3.60)

Thus arose the Debye-Hiickel expression for the experimentally in-

accessible individual ionic-activity coefficient. This expression could be
transformed into the Debye-Hiickel limiting law for the experimentally
measurable mean ionic-activity coefficient

(3.90)

which would indicate that the logarithm of the mean activity coefficient
falls linearly with the square root of the ionic strength /( = t ~ Cizl),
which is a measure of the total number of electric charges in the solution.
The agreement of the Debye-Hiickel limiting law with experiment
improved with decreasing electrolyte concentration and became excellent
for the limiting tangent to the logl± versus /! curve. With increasing con-
centration, however, experiment deviated more and more from theory, and,
at concentrations above I N, even showed an increase in I ± with increase
of concentration, whereas theory indicated a continued decrease.
An obvious improvement of the theory consisted in removing the
assumption of point-charge ions and taking into account their finite size.
With the use of an ion-size parameter a, the expression for the mean ionic-
activity coefficient became
10 f = _ A(z+=-)/i
g. ± I+ 'Xa
(3.105)

However, the value of the ion-size parameter a could not be the or-
236 CHAPTER 3

etically evaluated. Hence, an experimentally calibrated value was used.

With this calibrated value for a, the values of f± at other concentrations
[calculated from (3.105)] were compared with experiment.
The finite-ion-size model yielded agreement with experiment at con-
centrations up toward 0.1 N. It also introduced through the value of a,
the ion-size parameter, a specificity to the electrolyte (making NaCl dif-
ferent from KCl), whereas the point-charge model yielded activity coeffi-
cients which depended only upon the valency type of electrolyte. Thus,
while the limiting law sees only the charges on the ions, it is blind to the
specific characteristics which an ionic species may have, and this defect
is overcome by the finite-ion-size model.
Unfortunately, the value of a obtained from experiment by Eq. (3.105)
varies with concentration (as it should not if it represented simply the
collisional diameters), and, as the concentration increases beyond about
O.IM, a has sometimes to assume physically impossible (e.g., negative)
values. Evidently, these changes demanded by experiment do not only
reflect real changes in the sizes of ions, but they represent other effects
neglected in the simplifying Debye-Hiickel model. Hence, the basic postul-
ates of the Debye-Hiickel model must be scrutinized.
The basic postulates can be put down as follows: (l) The central ion
sees the surrounding ions in the form of a smoothed-out charge density
and not as discrete charges. (2) All the ions in the electrolytic solution are
free to contribute to the charge density, and there is, for instance, no pairing
up of positive and negative ions to form any electrically neutral couples.
(3) Only long-range coulombic forces are relevant to ion-ion interactions;
and short-range noncoulombic forces, such as dispersion forces, play a
negligible role. (4) The solution is sufficiently dilute to make 'lfr [which
depends on concentration through 'X-cf Eq. (3.20)] small enough to
warrant the linearization of the Boltzmann equation (3.10). (5) The only
role of the solvent is to provide a dielectric medium for the operation of
interionic forces, i.e., the removal of a number of ions from the solvent to
cling more or less permanently to ions other than the central ion is neglected.
It is because it is implicitly attempting to represent all these various
aspects of the real situation inside an ionic solution that the experimentally
calibrated ion-size parameter varies with concentration. Of course, a certain
amount of concentration variation of the ion-size parameter is understand-
able because the parameter depends upon the radius of solvated ions and
this time-average radius might be expected to decrease with an increase of
concentration. Hence, one must try to isolate that part of the changes in
the ion-size parameter which does not reflect real changes in the sizes of
 ION-ION INTERACTIONS 237

ions but represents the impact of, for instance, ionic solvation upon activity
coefficients.
This question of the influence of ion-solvent interactions (Chapter 2)
upon the ion-ion interactions will be considered in Section 3.6.

3.5.7. On the Parentage of the Theory of lon-Ion Interactions

Stress has been laid on the contribution of Oebye and HUckel (1923)
to the development of the theory of ion-ion interactions. It was Oebye
and H Uckel who ushered in the electrostatic theory of ionic solutions and
worked out its excellent predictions.
It is not often realized, however, that the credit due to Oebye and
HUckel as the parents of the theory of ionic solutions is the credit that is
quite justifiably accorded to foster parents. The true parents were Milner
and Gouy. These authors made important contributions very early in the
growth of the theory of ion-ion interactions.
Milner's contribution (1912) was direct. He attempted to find out the
virial t equation for a mixture of ions. However, Milner's statistical me-
chanical approach lacked the mathematical simplicity of the ionic-cloud
model of Oebye and HUckel and proved too unwieldy to yield a general
solution testable by experiment. Nevertheless, the contribution of Milner
was a seminal one in that, for the first time, the behavior of an ionic solution
had been linked mathematically to the interionic forces.
The contribution of Gouy (1910) was indirect. i Milner's treatment
was not sufficiently fruitful because he did not formulate a mathematically
treatable model. Gouy developed such a model in his treatment of the
distribution of the excess charge density in the solution near an electrode.
Whereas Milner sought to describe the interactions between series of discrete
ions, it was Gouy who suggested the smoothing-out of the ionic charges
into a continuous distribution of charge and took the vital step of using
Poisson's equation to relate the electrostatic potential and the charge density

t Virial is derived from the Latin word for force, and the vi rial equation of state is a
relationship between pressure, volume, and temperature of the form

PV K. Ka
-=1+-+-+···
RT V V'

where K., K a , . .. , the vi rial coefficients, represent interactions between constituent

particles.
t Chapman, in 1913, made an independent contribution on the same lines as that of
Gouy.
238 CHAPTER 3

in the continuum. Thus, Gouy was the first to evolve the ionic-atmosphere
model.
It was with an awareness of the work of Milner and Gouy that Debye
and Hlickel attacked the problem. Their contributions, however, were vital
ones. By choosing one ion out of the ionic solution and making an analogy
between this charged reference ion and the charged electrode of Gouy, by
using the Gouy type of approach to obtain the variation of charge density
and potential with distance from the central ion and thus to get the contribu-
tionto the potential arising from interionic forces, and, finally, by evolv-
ing a charging process to get the chemical-potential change due to ion-ion
interactions, they were able to link the chemical-potential change caused
by interionic forces to the experimentally measurable activity coefficient.
Without these essential contributions of Debye and Hlickel, a viable theory
of ionic solutions would not have emerged.

Further Reading
1. G. Gouy, J. Phys. 9, 457 (1910).
2. S. R. Milner, Phil. Mag. 6 (23): 551 (1912).
3. P. Debye and E. HUckel, Z. Physik, 24, 305 (1923).
4. R. A. Robinson and R. H. Stokes, Electrolytic Solutions, Butterworth's Pub-
lications, Ltd., London, 1959.
5. H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solutions,
3rd ed., Reinhold Publishing, Corps., New York, 1958.
6. H. L. Friedman, Ionic Solution Theory, Interscience Publishers, Inc., New
York, 1962.
7. M. H. Lietzke, R. W. Stroughton, and R. M. Fuoss, Proc. Nat. Acad. Sci.,
U. S., 59: 39 (1968).

3.6. ION-SOLVENT INTERACTIONS AND THE ACTIVITY COEF-

FICIENT

3.6.1. The Effect of Water Bound to Ions on the Theory of De-

viations from Ideality

The theory of behavior in ionic solutions arising from ion-ion inter-

actions has been seen (Section 3.5) to give rise to expressions in which, as
the ionic concentration increases, the activity coefficient decreases. In spite
of the excellent numerical agreement between the predictions of the inter-
ionic attraction theory and experimental values of activity coefficients at
sufficiently low concentrations (e.g., < 3 X 1O-3N), there is a most sharp
disagreement at concentrations above about IN, when the activity coef-
 ION-ION INTERACTIONS 239

I· I r---------...,

1·0

0·7

0·6

o· Sl..---::-'=-_---'-_ _.L-_...J
O'S 1·0 1·5 2·0
rm-
Fig. 3.36. The observed y± versus m V
curve for lithium chloride showing a minimum.

ficient begins to increase back toward the values it had in limitingly di-
lute solutions. In fact, at sufficiently high concentrations (one might have
argued, when the ionic interactions are greatest), the activity coefficient,
instead of continuing to decrease, begins to exceed the value of unity char-
acteristic of the reference state of noninteraction, i.e., of infinite dilution
(Fig. 3.36).
A qualitative picture can at once be given for these apparently anomal-
ous happenings. In Chapter 2, it has been argued that ions exist in solution
in various states of interaction with solvent particles. There is a consequence
which must therefore follow for the effectiveness of some of these water

Fig. 3.37. The distinction between free

water and hydration water which is locked
up in the solvent sheaths of ions.
240 CHAPTER 3

molecules in counting as part of the solvent. Those which are tightly bound
to certain ions cannot be effective in dissolving further ions added (Fig.
3.37). As the concentration of electrolyte increases, therefore, the amount
of effective solvent decreases. In this way, the apparently anomalous increase
in the activity coefficient occurs. For the activity coefficient is in effect that
factor which multiplies the simple, apparent ionic concentration and makes
it the effective concentration, i.e., the activity. If the hydration of the ions
reduces the amount of free solvent from that present for a given stoichio-
metric concentration, then the effective concentration increases and the
activity coefficient must increase so that its multiplying effect on the simple
stoichiometric concentration is such as to increase it to take into account
the reduction of the effective solvent. Experiment shows that sometimes
these increases more than compensate for the decrease due to interionic
forces, and it is thus not unreasonable that the activity coefficient should
rise above unity.
Some glimmering of the quantitative side of this can be seen by taking
the number of waters in the primary hydration sheath of the ions as those
which are "no longer effective solvent particles." For NaCi, for example,
Table 2.20 indicates that this number is about 7. If the salt concentration
is, e.g., 1O-2N, the number of moles of water per liter withdrawn from
effect as free solvent would be 0.07. As the number of moles of water per
liter is 1000/18 = 55.5, the number of moles of free water is 55.43 and the
effects arising from such a small change are not observable. But now consider
a IN solution of NaCI. The water withdrawn is 7 moles liter-I, and the
change in the number of moles of free water is from 55.5 in the infinitely
dilute situation to 48.5, a significant change. At 5N NaCl, more than half
of the water in the solution is associated with the ions, and a sharp increase
of activity coefficient-somewhat of a doubling, in fact-would be expected
to express the increase in effective concentration of the ions.
To what extent can this rough sketch be turned to a quantitative model?

3.6.2. Quantitative Theory of the Activity of an Electrolyte as

a Function of the Hydration Number

The basic thought here has to be similar to that which lay beneath
the theory of electrostatic interactions: to calculate the work done in going
from a state in which the ions are too far apart to feel any interionic attrac-
tion, to the state at a finite concentration c at which part of the ions be-
havior is due to this. This work was then [Eq. (3.59)] placed equal to RTlnJ,
where J is the activity coefficient. which was thereby calculated.
 ION-ION INTERACTIONS 241

When one realizes that one has to explain a reversal of the direction
in which the activity coefficient varies with concentration by taking into
account the removal of some of the solvent from effective partaking in
the ionic solution's activity, the philosophy of the calculation of what
effect this has on the activity coefficient becomes clear. One must calculate
the work done in the changes caused by solvent removal and add this to
the work done in building up the ionic atmosphere. What, then, is the
contribution due to these water-removing processes? [Note that ions are
hydrated at all times in which they are in the solution. One is not going to
calculate the heat of hydration; that was done in Chapter 2. Here, the task
is to calculate the work done as a consequence of the fact that, when water
molecules enter the solvation sheath, they are, so to write, hors de combat
as far as the solvent is concerned. It is to be an RT In(c r/c 2) type of cal-
culation. ]
Let it be assumed, as a device for the calculation, that the interionic
attraction is switched off. (Reasons for employing this artifice will duly be
given.) Thus, there are two kinds of work which must be taken into account.
1. The RTln(c1/c 2 ) kind of work done when the change of concentra-
tion of the free solvent water, caused by the introduction into the
solution of ions of a certain concentration, takes place.
2. The RTln(c1/c 2) kind of work done when the corresponding change
of concentration of the ions, due to the removal of the water to
their sheaths, occurs.

The work done for process 1 is easy to calculate. Before the ions have
been added, the concentration of the water is unaffected by anything; it is
the concentration of pure water, and its activity, the activity of a pure sub-
stance, can be regarded as unity. After the ions are there, the activity of
the water is, say, a w . Then, the work when the activity of the water goes
from 1 to a w is RT In(aw/l).
However, one wishes to know the change of activity caused to the
electrolyte by this change of activity of the water. Further, the calculation
must be reduced to that for I mole of electrolyte. Let the sum of the moles
of water in the primary sheath per liter of solution for both ions of the
imagined 1: 1 electrolyte be nh (for 1 molar solutions, this is the hydration
number). Then, if there are n moles of electrolyte in the water, the change
in free energy due to the removal of the water to the ions' sheaths is
- (n"fn )RT In a w per mole of electrolyte.
One now comes to the work 2, see above, and realizes why the calcul-
ation is best made as a thought process in which the interionic attraction
242 CHAPTER 3

is shut off while this calculation of work is done. For one wants to be able
to use the ideal-solution (no interaction) equation for the work done,
RTln(cdc 2 ), and not RTln(ada 2 ). Thus, to have to use the latter expression
would be awkward; it needs a knowledge of the activities themselves, and,
that, one is trying to calculate.
Now, the free-energy change due to the change of concentration of
the ions-consequent upon the removal of the effective solvent molecule-is

RT In X"fter water remov"l from free to solvated state

X before water removal from free to solvated state

where x is the mole fraction of the electrolyte in the solution.

Before the water is removed,
n
Xbefore =- --
nw+ n
(3.114)

where n is the number of moles of electrolyte present in nw of water. Then,

after the water is removed to the sheaths,

n
x ------
nw - nh + n
(3.115)
after -

The change in free energy is hence:

RT In __n--=w:....+_n__
nw + n - nh
Hence, the total free-energy change in the solution, calculated per mole
of the electrolyte present, is

- ~ RT In aw + RT In nw+n
n nw + n - nh
Now, one has to switch back on the coulombic interactions. If the
expression for the work done in building up an ionic atmosphere [e.g.,
Eq. (3.106)] were still valid in the region of relatively high concentrations
in which the effect of change of concentration is occurring, then, t

Ave nh n +n
RT logj±(exp)=---'----- 2.303RT -log a w + 2.303RT log---'W'---_
n nw+n-nh
I+BaVc
(3.l16)

tHere yc has been written instead of the l~ of Eq. (3.106). For 1:1 electrolytes, c and
I are identical.
 ION-ION INTERACTIONS 243

One sees at once that there is a possibility of a change of direction of

the change of logf± with increase of concentration in the solution. If the
last term predominates, RTlnf± may increase with concentration.
But the situation here does have a fairly large shadow on it because
of the use of the expression (3.106) in yC: It will be seen (Section 3.9)
that, at concentrations as high as I N, there are some fundamental difficulties
for the ionic cloud model on which this Vc expression of (3.106) was
based (the ionic atmosphere can no longer be considered a continuum of
smoothed-out charge). It is clear that, when the necessary mathematics
can be done, there will be an improvement on the ye expression, and
one will hope to get it more correct than it now is. Because of this shadow,
a comparison of (3.116) with experiment to test the validity of the model
for removing solvent molecules to the ions' sheath should be done a little
with tongue in cheek.

3.6.3. The Water Removal Theory of Activity Coefficients and

Its Apparent Consistency with Experiment at High Elec-
trolytic Concentrations

If one examines the ion-solvent terms in Eq. (3.116), one sees that,
since all' < 1 and in general nh > n (more than one hydration water per
ion), both the terms are positive. Hence, one can conclude that the Debye-
Hiickel treatment, which ignores the withdrawal .of solvent from solution,
gives values of activity coefficients which are smaller than those which take
into account these effects. Further, the difference arises from the ion-
solvent terms, i.e.,
nll' + n
nh
- 2.303RT-log all'
n
+ 2 303RT log nll'
---,----
+ n - nh
As the electrolyte concentration increases, all' decreases and nh increases;
hence both ion-solvent terms increase the value of logf Further, the
numerical evaluation shows that the above ion-solvent term can equal and
become larger than the Debye-Hiickel (ye) coulombic term. This means
that the log!± versus Jl curve can pass through a minimum and then start
rising, which is precisely what is observed (ef Fig. 3.36, where an activity
coefficient is plotted against the corresponding molality).
On the other hand, with increasing dilution, nll' + n ~ nh or
nll' + n - nh""'" nw + nand aw -+ 1, and, hence, the terms vanish, which
indicates that ion-solvent interactions (which are of short range) are
significant for the theory of activity coefficients only in concentrated so-
244 CHAPTER 3

TABLE 3.11
Water Activities in Sodium Chloride Solutions

M aw M aw

0.1 0.99665 3 0.8932

0.5 0.98355 3.4 0.8769
0.96686 3.8 0.8600
1.4 0.9532 4.2 0.8428
1.8 0.9389 4.6 0.8250
2.2 0.9242 5 0.8068
2.6 0.9089 5.4 0.7883

lutions. At extreme dilutions, only the ion-ion long-range coulombic in-

teractions are of importance.
In order to test Eq. (3.116), which is a quantitative statement of the
influence of ionic hydration on activity coefficients, it is necessary to know
the quantity Ilk and the activity aw of water, it being assumed that an ex-
perimentally calibrated value of the ion-size parameter is available. The
activity of water can be obtained from independent experiments (Table
3.11). The quantity I1h can be used as a parameter. If Eq. (3.116) is tested
as a two-parameter equation (Ilk and a being the two parameters, Table 3.12)
it is found that theory is in excellent accord with experiment. For instance,
in the case of NaC!, the calculated activity coefficient agrees with the ex-
perimental value for solutions as concentrated as 5M (Fig. 3.38).

TABLE 3.12
Values of nh of a One Molar Solution and a

Salt n" a, A

HCl 8.0 4.47

HBr 8.6 5.18
NaCi 3.5 3.97
NaBr 4.2 4.24
KCI 1.9 3.63
MgCI 2 13.7 5.02
MgBr2 17.0 5.46
 ION-ION INTERACTIONS 245

I·Or-

0'91-

t 0'81-

1+ ~ /
0·7r- .~NaCI

0·6r-

O' 5,---:-,1:----:,-::-1-:""::-1-::,-:-1----,
0·5 1·0 1·5 2'()
I"ffi"-
Fig. 3.38. Comparison of the activity co-
efficients of NaCI calculated from Eq. (3.116)
with a = 3.97 and nh = 3.5, with the exper-
imentally observed activity coefficients for
NaCI (the full line, theoretical curve; open
circles, experimental points).

Since the quantity nh is the number of moles of water used up in

solvating n = n+ + n_ moles of ions, it can be split up into two terms:
(nh)+ moles. required to hydrate n+ moles of cations, and (nh)- moles re-
quired to hydrate n_ moles of anions. It follows that [(nh)+ jn+] and [(nh)-jn_]
are the hydration numbers (see Chapter 2) of the positive and negative
ions and [(nh)+jl1+] + [(nh)-jn_] is the hydration number of the electrolyte.
It has been found that, in the case of several electrolytes, the values
of the hydration numbers obtained by fitting the theory [Eq. (3.116)] to
experiment are in reasonable agreement with hydration numbers deter-
mined by an independent method (Table 3.13). Alternatively, one can say
that, when independently obtained hydration numbers are substituted in
Eq. (3.116), the resulting values oflogJ± show fair agreement with experi-
ment.
In conclusion, therefore, it may be said that the treatment of the
influence of ion-solvent interactions on ion-ion interactions has extended
the range of concentration of an ionic solution which is accessible to theory.
Whereas the finite-ion-size version of the Oebye-Hlickel theory did not
permit theory to deal with solutions in a range of concentrations corre-
sponding to those of real life, Eq. (3.116) advances theory into the range of
practical concentrations. Apart from this numerical agreement with ex-
246 CHAPTER 3

TABLE 3.13
Nearest Integer Hydration Number of Electrolytes from Eq. (3.139) and the
Most Probable Value from Independent Experiments (cf. Table 2.20)

Hydration number from Eq. (3.116) Hydration number from

Salt
(nearest integer) other experimental methods

LiCI 7 7±1
LiBr 8 7± 1
NaCl 4 6±2
KCl 2 5±2
KI 3 4±2

periment, Eq. (3.116) unites two basic aspects of the situation inside an
electrolytic solution, namely ion-solvent interactions and ion-ion inter-
actions.

Further Reading
1. N. Bjerrum, Z. Anorg. AI/gem. Chern., 109: 175 (1920).
2. R. H. Stokes and R. A. Robinson, J. Amer. Chern. Soc., 70: 1870 (1948).
3. E. Glueckauf, Trans. Faraday Soc., 51: 1235 (1955).
4. R. H. Stokes and R. A. Robinson, Trans. Faraday Soc., 53: 301 (1957).
5. R. A. Robinson and R. H. Stokes, Electrolyte Solutions, 2nd ed., Butter-
worth's Publications, Ltd., London, 1959.
6. H. S. Frank, "Solvent Models and the Interpretation of Ionization and Solva-
tion Phenomena," in: B. E. Conway and R. G. Barradas, eds., John Wiley
& Sons, Inc., New York, 1966.

3.7. THE SO-CALLED "RIGOROUS" SOLUTIONS OF THE

POISSON-BOLTZMANN EQUATION

One approach to the understanding of the discrepancies between the

experimental values of the activity coefficient and the predictions of the
Debye-Htickel model has just been described (Section 3.6); it involved a
consideration of the influence of solvation.
An alternative approach is based on the view that the failure of the
Debye-HUckel theory at high concentrations stems from the fact that the
development of the theory involved the linearization of the Boltzmann
equation (cf Section 3.3.5). If such a view is taken, there is an obvious
 ION-ION INTERACTIONS 247

solution to the problem: instead of linearizing the Boltzmann equation,

one can take the higher terms. Thus, one obtains the unlinearized P-B
equation

~ ~ (r2 d lP r )
r2 dr dr

In the special case of a symmetrical electrolyte (z 1 = - L = z) with equal

concentrations of positive and negative ions, i.e., 111° = ILo = 11°, one gets

(3.118)
But
e+ X - e-X = 2 sinh x

and, therefore,
. zeolPr
n
r;:r
= - 2n ll ze 0 smh--
kT (3.119)

or
(3.120)

By utilizing a computer program, one could obtain from (3.120) so-

called "rigorous" solutions.
Before proceeding further, however, it is appropriate to stress a logical
inconsistency in working with the unlinearized P-B equation (3.117).
The unlinearized Boltzmann equation (3.10) implies a l1ol1lil1ear relation-
ship between charge density and potential. In contrast, the linearized Boltz-
mann equation (3.16) implies a linear relationship of er to lPr.
Now, a linear charge density-potential relation is consistent with the
law of superposition of potentials, which states that the electrostatic po-
tential at a point due to an assembly of charges is the sum of the potentials
due to the individual charges. Thus, when one uses an unlinearized P-B
equation, one is assuming the validity of the law of superposition of po-
tentials in the Poisson equation and its invalidity in the Boltzmann equa-
tion. This is a basic logical inconsistency which must reveal itself in the
predictions which emerge from the so-called rigorous solutions. This is
indeed the ease, as will be shown below.
Recall that, after obtaining the contribution of the ionic atmosphere
to the potential at the central ion, the coulombic interaction between the
central ions and the cloud was calculated by an imaginary charging process,
248 CHAPTER 3

generally known as the Guntelberg charging process in recognition of its

originator.
In the Guntelberg charging process, the central ion i is assumed to be
in a hypothetical condition of zero charge. The rest of the ions, fully charg-
ed, are in the positions that they would hypothetically have were the central
ion charged to its normal value ZieO; i.e., the other ions constitute an ionic
atmosphere enveloping the central ion (Fig. 3.39). The ionic cloud sets up
a potential 1J!clouu = - (ZieO/ EX-I) at the site of the central ion. Now, the
charge of the central ion is built up (Fig. 3.39) from zero to its final value
ZieO, and the work done in this process is calculated by the usual formula
for the electrostatic work of charging a sphere (see Section 3.3.1) I.e.,

(3.3)

Since, during the charging, only ions of the ith type are considered, the
Guntelberg charging process gives that part of the chemical potential due
to electrostatic interactions.
Now, the Guntelberg charging process was suggested several years
after Debye and HUckel made their theoretical calculation of the activity
coefficient. These authors carried out another charging process, the Debye
charging process. All the ions are assumed to be in their equilibrium, 01

IONIC CLOUD WITH

ITS FULL CHARGE

- CENTRAL ION IN A
HYPOTHETICAL STATE
OF ZERO CHARGE

CHARGE ON CENTRAL
JON BUILT UP FROM
ZERO TO +Zi eo

_ IONIC CLOUD WITH

ITS FULL CHARGE

~ CENTRAL ION WITH

IT S FULL CHARGE
Z, eo

Fig. 3.39. The Guntelberg charging process.

 ION-ION INTERACTIONS 249

CENTRAL ION IN
HYPOTHETICAL
IONIC CLOUD( i.e;ITS STATE OF ZERO
IONS) IN A CHARGE
HYPOTHETICAL STATE
OF ZERO CHARGE

CENTRAL ION CHARGED FROM 0 TO + ZI eo

ANO SIMULTANEOUSLY IONS OF CLOUD
CHARGED FROM 0 TO I ZI %1

IONIC CLOUD WITH

CHARGE - Z I eo

CENTRAL ION
WITH CHARGE + Zi eo

Fig . 3.40. The Debye charging process.

time-average, positions in the ionic atmosphere (Fig. 3.40), but the central
ion and the cloud ions are all considered in a hypothetical condition of
zero charge. All the ions of the assembly are then simultaneously brought
to their final values of charge by an imaginary charging process in which
there are small additions of charges to each. Since ions of all types (not
only of the ith type) are considered, the work done in this charging process
yields the free-energy change arising from the electrostatic interactions in
solution. Differentiation of the free energy with respect to the number of
moles of the ith species gives the chemical potential

aJG1 _ 1 _ J
an., - fli-l (3.121 )

From the rigorous solutions of the unlinearized P-B equation, one

gets the cloud contribution to the electrostatic potential at the central
ion; and, when this value of the electrostatic potential is used in the two
charging processes to get the chemical-potential change Jfli-l arising from
ion-ion interactions, it is found that the Guntelberg and Debye charging
processes give discordant results. As shown by On sager, this discrepancy
is not due to the invalidity of either of the two charging processes: it is
250 CHAPTER 3

a symptom of the logical inconsistency intrinsic in the un linearized P-B

equation.
This discussion of rigorous solutions has thus brought out an important
point: The agreement between the chemical-potential change LJfli-1 calculat-
ed by the Debye and Guntelberg charging processes serves as a fundamental
test of any theory of the electrostatic potential set up at a central ion by the
remaining ions of an ionic solution.

Further Reading
1. P. Debye and E. Hiickel, Z. Physik, 24: 305 (1923).
2. H. Milller, Z. Physik, 28: 324 (1927).
3. T. H. Gronwall, V. K. La Mer, and K. Sandved, Z. Physik, 29: 358 (1929).
4. R. H. Fowler, Statistical Mechanics, Cambridge University Press, New York
(1929).
5. L. Onsager, Chern. Rev., 13: 73 (1933).
6. J. G. Kirkwood, J. Chern. Phys., 2: 351 (1934).
7. J. G. Kirkwood, J. Chern. Phys., 2: 767 (1934).
8. J. G. Kirkwood, Chern. Rev., 19: 275 (1936).
9. W. G. McMillan, Jr., and J. E. Mayer, J. Chern. Phys., 13: 276 (1945).
10. J. E. Mayer, J. Chern. Phys., 18: 1426 (1950).
11. H. Falkenhagen and G. Kelbg, Ann. Physik VI, 11: 60 (1952).
12. H. Falkenhagen and G. Kelbg, Z. Elektrochem., 56: 834 (1952).
13. H. Falkenhagen, M. Leist, and H. Kelbg, Ann. Physik VI, 11: 51 (1952).
14. J. C. Poirier, J. Chern. Phys., 21: 965 and 972 (1953).
15. J. G. Kirkwood and J. C. Poirier, J. Phys. Chern., 58: 591 (1954).
16. H. S. Frank, Discussions Faraday Soc., 24: 66 (1957).
17. E. A. Guggenheim, Trans. Faraday Soc., 55: 1714 (1959).
18. E. A. Guggenheim, Trans. Faraday Soc., 56: 1152 (1960).
19. F. H. Stillinger, Jr., J. G. Kirkwood, and P. J. Wojtowicz, J. Chern. Phys.,
32: 1837 (1960).
20. F. H. Stillinger, Jr., and J. G. Kirkwood, J. Chern. Phys., 33: 1282 (1960).
21. F. H. Stillinger, Jr., J. Chern. Phys., 35: 1581 (1961).
22. E. A. Guggenheim, Trans. Faraday Soc., 58: 86 (1962).
23. H. L. Friedman, Ionic Solution Theory, Interscience Publishers, Inc., New
York, 1962.
24. J. C. Poirier, "Current Status of the Statistical Mechanical Theory of Ionic
Solutions," in: B. E. Conway and R. G. Barradas, eds., Chernical Physics of
Ionic Solutions, John Wiley & Sons, Inc., New York, 1966.
 ION-ION INTERACTIONS 251

3.8. TEMPORARY ION ASSOCIATION IN AN ELECTROLYTIC

SOLUTION: FORMATION OF PAIRS, TRIPLETS, ETC.

3.8.1. Positive and Negative Ions Can Stick Together: lon-Pair

Formation

The Debye-Hlickel model assumed the ions to be in almost random

thermal notions and therefore in almost random positions. The slight
deviation from randomness was pictured as giving rise to an ionic cloud
around a given ion, a positive ion (of charge +zeo) being surrounded by
a cloud of excess negative charge (- ze o). However, the possibility was
not considered that some negative ions in the cloud would get sufficiently
close to the central positive ion in the course of their quasi-random solution
movements so that their thermal translational energy would not be sufficient
for them to continue their independent movements in the solution. Bjerrum
suggested that a pair of oppositely charged ions may get trapped in each
other's coulombic field. An iOI1 pair may be formed.
The ions of the pair together form an ionic dipole on which the net
charge is zero. Within the ionic cloud, the locations of such uncharged ion
pairs are completely random, since, being uncharged, they are not acted
upon by the coulombic field of the central ion. Further, on the average,
a certain fraction of the ions in the electrolytic solution will be stuck to-
gether in the form of ion pairs. This fraction must now be evaluated.

3.8.2. The Probability of Finding Oppositely Charged Ions near

Each Other

Consider a spherical shell of thickness dr and of radius r from a refer-

ence positive ion (Fig. 3.41). The probability P r that a negative ion is in
the spherical shell is proportional, firstly, to the ratio of the volume 4nr2 dr
of the shell to the total volume V of the solution; secondly, to the total
number N_ of negative ions present; and, thirdly, to the Boltzmann factor
exp( - U/kT), where U is the potential energy of a negative ion at a distance
r from a cation, i.e.,

P = 4nr 2 dr N- e-U/kT (3.122)

r V

Since N _/ V is the concentration 11_° of negative ions in the solution

and

U= (3.123)
252 CHAPTER 3

dY

SPHERICAL SHELL OF
RADIUS \'" AND
THICKNESS cD'

Fig. 3.41. The probability P, of finding an

ion of charge Leo in a dr-thick spherical
shell of radius r around a reference ion of
charge z +e o.

it is clear that
(3.124)
or, writing
A= (3.125)

one has
(3.126)

A similar equation is valid for the probability of finding a posItIve

ion in a dr-thick shell at a radius r from a reference negative ion. Hence,
in general, one may write for the probability of finding an i type of ion in
a dr-thick spherical shell at a radius r from a reference ion k of opposite
charge
(3.127)
where
2
A = zizJ.;eO (3.128)
ekT

This probability of finding an ion of one type of charge near an ion

of the opposite charge varies in an interesting way with distance (Fig. 3.42).
For small values of r, the function P r is dominated by e AI ' rather than by r2,
and, under these conditions, P r increases with decreasing r; for large values
of r, eAlr ---+ 1 and P r increases with increasing r because the volume 4nr2 dr
of the spherical shell increases as r2. It follows from these considerations
 ION-ION INTERACTIONS 253

o r-
Fig. 3.42. The probability P, of finding an
ion of one type of charge as a function of
distance.

that P, goes through a minimum for a particular, critical value of r. This

conclusion may also be reached by computing the number of ions in a
series of shells, each of an arbitrarily selected thickness of 0.1 A (Table 3.14).

3.8.3. The Fraction of Ion Pairs, According to Bjerrum

If one integrates P, between a lower and an upper limit, one gets the
probability P, of finding a negative ion within a distance from the reference
positive ion, defined by the limits. Now, for two oppositely charged ions
to stick together to form an ion pair, it is necessary that they should be
close enough for the coulombic attractive energy to overcome the thermal
energy which scatters them apart. Let this "close-enough" distance be q.
Then, one can say that an ion pair will form when the distance r between a

TABLE 3.14
Number of Ions in Spherical Shells at Various Distances

Number of ions in shell x 1022

r, A
Of opposite charge Of like charge

2 1.7711; 0.00111;
2.5 1.3711; 0.00511;
3 1.2211; 0.0111;
3.57 1.1811; 0.0211;
4 1.2011; 0.0311;
5 1.3111; 0.0811 ;
254 CHAPTER 3

positive and negative ion becomes less than q. Thus, the probability of ion-
pair formation is given by the integral of Pr between a lower limit of a,
the closest distance of approach of ions, and an upper limit of q.
Now, the probability of any particular event is the number of times
that the particular event is expected to be observed divided by the total
number of observations. Hence, the probability of ion-pair formation is
the number of ions of species i which are associated into ion pairs divided
by the total number of i ions, i.e., the probability of ion-pair formation is
the fraction () of ions which are associated into ion pairs. Thus,

(3.129)

It is seen from Figure 3.43 that the integral in Eq. (3.129) is the area
under the curve between the limits r = a and r = q. But it is obvious that,
as r increases past the minimum, the integral becomes greater than unity.
Since, however, () is a fraction, this means that the integral diverges.
In this context, Bjerrum took the arbitrary step of cutting off the
integral at the value of r = q corresponding to the minimum of the P r
versus r curve. This minimum can easily be shown (Appendix 3.4) to
occur at
A
(3.130)
2

Bjerrum justified this step by arguing that it is only short-range

coulombic interactions that lead to ion-pair formation and, further, when
a pair of oppositely charged ions are situated at a distance apart of r > q,
it is more appropriate to consider them free ions.

r=o r=q

Fig. 3.43. The integral in Eq. (3.129) is

the area under the curve between the limits
, = a and, = q.
 ION-ION INTERACTIONS 255

Bjerrum concluded, therefore, that ion-pair formation occurs when

an ion of one type of charge, e.g., a negative ion, enters a sphere of radius q
drawn around a reference ion of the opposite charge, e.g., a positive ion.
But it is the ion-size parameter which defines the closest distance of ap-
proach of a pair of ions. The Bjerrum hypothesis can therefore be stated
as follows: If a < q, then ion-pair formation can occur; if a > q the ions
remain free (Fig. 3.44).
Now that the upper limit of the integral in Eq. (3.129) has been taken
to be q = ).(2, the fraction of ion pairs is given by carrying out the inte-
gration. It is

(3.131)

For mathematical convenience, a new variable y is defined as

A 2q
y=-=- (3.132)
r r

Ion-pair formation nat possible

/'

/ tJ>q
I

--- -
;'
"-
:::....----:::::...
/'
/'
/ tJ< q
/
I
I
I Ian-pair
I formation

\
possible
\
\
\
\
\( /
/
\ "-
/

Fig. 3.44. (a) lon-pair formation occurs if a :s; q;

(b) ion-pair formation does not occur if a > q.
256 CHAPTER 3

TABLE 3.15
Value of the Integral HeY y-4 dy

b He y-4 dy
Y b f~ eVy-4 dy b Jg eVy-4 dy

2.0 0 3 0.326 10 4.63

2.1 0.0440 3.5 0.442 15 93.0
2.2 0.0843 4 0.550
2.4 0.156 5 0.771
2.6 0.218 6 1.041
2.8 0.274

Hence, in terms of the new variable y, Eq. (3.131) becomes (c/. Appen-
dix 3.5)
() = 4nnt (
Z Z e fb2 eYy-4 dy +ck/2)3 (3.133)

where
(3.134)

S!
Bjerrum has tabulated the integral eYy-4 dy for various values of b (Table
3.15). This means that, by reading off the value of the corresponding

TABLE 3.16
Fraction of Association. (). of Univalent Ions in Water at 180 C

a X 108 cm: 2.82 2.35 1.76 1.01 0.70 0.47

q/a: 2.5 3 4 7 10 15
c t , moles liter- 1

0.0001 0.001 0.027

0.0002 0.002 0.049
0.0005 0.002 0.006 0.106
0.001 0.001 0.001 0.004 0.Dl1 0.177
0.002 0.002 0.002 0.003 0.007 0.021 0.274
0.005 0.002 0.004 0.007 0.016 0.048 0.418
0.01 0.005 0.008 0.012 0.030 0.030 0.529
0.02 0.008 0.013 0.022 0.053 0.137 0.632
0.05 0.017 0.028 0.046 0.105 0.240 0.741
0.1 0.029 0.048 0.072 0.163 0.336 0.804
0.2 0.048 0.079 0.121 0.240 0.437 0.854

t c in moles litec 1 ~ IOOOn;o/N.

 ION-ION INTERACTIONS 257

integral and substituting for the various other terms in (3.133), the degree
of association of an electrolyte may be computed if the ion sizes, the dielec-
tric constant, and the concentrations are known (Table 3.16).

3.8.4. The lon-Association Constant KA of Bjerrum

e
The quantity yields a clear idea of the fraction of ions which are
associated in ion pairs in a particular electrolytic solution, at a given con-
centration. It would, however, be advantageous if each electrolyte, e.g.,
NaCl, BaS0 4 , and La(NOah, were assigned a particular number which
would reveal, without going through the calculation of e,
the extent to
which the ions of that electrolyte associate in ion pairs. The quantitative
measure chosen to represent the tendency for ion-pair formation was
guided by historical considerations.
Arrhenius in 1887 had suggested that many properties of electrolytes
could be explained by a dissociation hypothesis: The neutral molecules AB
of the electrolyte dissociate to form ions A + and B-, and this dissociation
is governed by an equilibrium
AB~A+ + B- (3.135)

Applying the law of mass action to this equilibrium, one can define a
dissociation constant
K= (3.136)

By analogy, t one can define an association constant KA for ion-pair

formation. Thus, one can consider an equilibrium between free ions (the
positive M + ions and the negative A-ions) and the associated ion pairs
(symbolized IP)
(3.137)

The equilibrium sanctions the use of the law of mass action

(3.138)

t The analogy must not be carried too far because it is only a formal analogy. Arrhenius's
hypothesis can now be seen to be valid for ionogens (i.e., potential electrolytes), in
which case the neutral ionogenic molecules (e.g., acetic acid) consist of aggregates
of atoms held together by covalent bonds. What is under discussion here is ion as-
sociation, or ion-pair formation, of ionophores (i.e., true electrolytes). In these ion
pairs, the positive and negative ions retain their identity as ions and are held together
by electrostatic attraction.
258 CHAPTER 3

where the a's are the activities of the relevant species. From (3.138), it is
seen that KA is the reciprocal of the ion pair's dissociation constant.
Since () is the fraction of ions in the form of ion pairs, ()c is the con-
centration of ion pairs, and (I - ())c is the concentration of free ions. If
the activity coefficients of the positive and negative free ions are f+ and f- ,
respectively, and that of the ion pairs is fIP, one can write

()cfrp
KA = (l - ())cf+(1 - ())cf-
() I frp
(3.139)
(I - ())2 C f+f-

or, using the definition of the mean ionic-activity coefficient [cf Eq. (3.72)],
()
KA (I _ ())2 (3.140)
=
C f ±2
Some simplifications can now be introduced. The ion-pair activity
coefficient frp is assumed to be unity because deviations of activity coef-
ficients from unity are ascribed in the Debye-H tickel theory to electrostatic
interactions. But ion pairs are not involved in such interactions owing to
their zero charge, and, hence, they behave ideally like uncharged particles,
i.e., fip 1.=
Further, in very dilute solutions: (l) The ions rarely come close enough
together (i.e., to within a distance q) to form ion pairs, and one can consider
() ~ 1 or 1 - () "-' 1 ; (2) activity coefficients tend to unity, i.e., fi or f ± ---+ 1.
Hence, under these conditions of very dilute solutions, Eq. (3.140)
becomes
(3.141)

and, substituting for () from Eq. (3.133), one has

4nn·o z z e 2)3 fb e11y-4 dy

K4• = -C- ' ( + - 0 (3.142)
EkT 2

But
0_ cNA (3.143 )
ni - 1000

and, therefore,
( z +z - e02)3 fb e?ly-4 dy (3.144)
EkT 2
 ION-ION INTERACTIONS 259

TABLE 3.17
Ion Association Constant KA : Extent to Which lon-Pair Formation Occurs

Temperature,
Salt Solvent K
°C

KBr Acetic acid 30 6.20 9.09 x 106

KBr Ammonia -34 22 5.29 x 10'
CsCI Ethanol 25 24.30 1.51 x 10'
KI Acetone 25 20.70 1.25 x 10'
KI Pyridine 25 12.0 4.76 x 103

The value of the association constant provides an indication of whether

ion-pair formation is significant. The higher the value of K A , the more
extensive is the ion-pair formation (Table 3.17).
What are the factors which increase KA and therefore increase the
degree of ion-pair formation? From Eq. (3.144), it can be seen that the
factors which increase KA are (l) low dielectric constant E; (2) small ionic
radii, which lead to a small value of a and hence [ef Eq. (3.134)] to a
large value of the upper limit b of the integral in Eq. (3.144); and (3) large
z+ and L.
These ideas based on Bjerrum's picture of ion-pair formation have
received considerable experimental support. Thus, in Fig. 3.45, the associa-
tion constant is seen to increase markedly with decrease of dielectric con-

'"
.2
5

0·5 1·0 1·5

log ~

Fig. 3.45. Variation of the association con-

stant KA with dielectric constant for 1:1
salts.
 260 CHAPTER 3

0-012

0-010

CD
·0-008
z
o
j:
3o 0 -006
UI
UI

: 0-004
o
z
o
j: 0-002
()
<C
II:
IL

Fig. 3.46. Variation of (), the fraction of associated ions,

with a, the closest distance of approach (Le., the ion-size
parameter)_

stant. t The dependence of ion-pair formation on the distance of closest

approach is seen in Fig. 3.46.
When numerical calculations are carried out with these equations, the
essential conclusion which emerges is that, in aqueous media, ion association
in pairs scarcely occurs for 1: I-valent electrolytes but can be of import-
ance for 2:2-valent electrolytes. The reason is that KA. depends on Z+L
through Eq. (3.142). In nonaqueous solutions, most of which have dielectric
constants much less than that of water (e = 80), ion association is extremely
important.

3.8.5. Activity Coefficients, Bjerrum's Ion Pairs, and Debye's

Free Ions

What direct role do the ion pairs have in the Debye-Hiickel electro-
static theory of activity coefficients? The answer simply is: None. Since
ion pairs carry no net charge,t they are ineligible for membership in the

t But the critical dielectric constant above which there is no more ion-pair formation
(as indicated by Fig_ 3.45) is really a result of the arbitrary cutting off of ion-pair
formation at the distance q [see (3.8.6) J-
Remember that the equations for the Bjerrum theory, as presented here, are correct
only for electrolytes yielding ions of the same valency z, i.e., only for symmetrical
1 : 1- or 2 :2-valent electrolytes.
 ION-ION INTERACTIONS 261

ion cloud where the essential qualification is charge. Hence, ion pairs are
dismissed from a direct consideration in the Debye-Htickel theory.
This does not mean that the Debye-Htickel theory gives the right
answer when there is ion-pair formation. The extent of ion-pair formation
decides the value of the concentration to be used in the ionic-cloud model.
By removing a fraction 0 of the total number of ions, only a fraction I - 0
of the ions remain for the Debye-Htickel treatment which interests itself
only in the free charges. Thus, the Debye-Htickel expression for the activity
coefficient [Eq. (3.106)] is valid for the free ions with two important modi-
fications: (1) Instead of there being a concentration c of ions, there is only
(1 - O)c; the remainder Oc is not reckoned with owing to association.
(2) The closest distance of approach of free ions is q and not a. These
modifications yield
logf± = _ A(Z+L) V (I - O)c (3.145)
~1-+-'-B=-q·--'v7-(;=1=-~O=)c

This calculated mean activity coefficient is related to the measured

mean activity coefficient of the electrolyte (f ±)ObS by the relation (for the
derivation, see Appendix 3.6)

(3.146)
or
log (f±)obS = logf± + log (1 - 0)
A(Z+L) V (I - fJ)c
= - + I og (1 - 0) (3.147)
I + Bq V (1 - O)c

This equation indicates how the activity coefficient depends on the

extent of ion association. In fact, this equation constitutes the bridge be-
tween the treatment of solutions of true electrolytes and solutions of poten-
tial electrolytes. More will be said on this matter in the chapter on protons
in solution (Chapter 5), part of which deals with potential electrolytes.

3.8.6. The Fuoss Approach to lon-Pair Formation

Despite a considerable agreement with experiment, there are several

unsatisfactory features of the Bjerrum picture of ion-pair formation.
The first and most important defect of the Bjerrum picture is that it
identifies, as ion pairs, ions which are not in physical contact; the pair is
counted as an ion pair as long as r < q.
262 CHAPTER 3

A second defect is the arbitrary way in which the probability integral

[cf Eq. (3.129)] is terminated at q, i.e., the distance r at which P r is a
minimum. The physical reasons supporting this choice of a condition for
the maximum distance between ion centers at which pairing can occur are
not clear. In practice, however, the value of the ion-association constant
KA is not very sensitive to the actual numerical value chosen for the upper
limit of the integral.
It is these considerations that led Fuoss to present an alternative picture
of ion pairs and a derivation of the ion·association constant.
An ion pair is defined in a straighforward manner. For the period of
time (irrespective of its magnitude) that two oppositely charged ions are
in contact and, therefore, at a distance apart of r = a, the two ions function
as neutral dipole and can be defined as an ion pair.
To get an idea of the fraction of ion pairs in a solution, the following
thought experiment is a useful device. Let the motion of all the ions in a
solution be frozen and the number of oppositely charged pairs of ions in
contact be counted. If this thought experiment is repeated many times, then
one can determine NIP, the average number of ion pairs. The fraction of
ion' pairs is then obtained by dividing NIP by N i , the average number of
ions.
The calculation of the fraction () = N1P/N i is done as follows. Suppose
Z positive ions and an equal number of negative ions exist in a volume V
of solution. Let there be ZIP ion pairs; then there will be ZFl = Z - ZIP
free ions of each species. Now, suppose one adds oZ positive ions and a
similar number of anions. Since some of these will form ion pairs and some
will remain free,
(3.148)

The number OZFI of added negative ions that remain free is propor-
tional, firstly, to the number oZ of negative ions added to the solution and,
secondly, to the free volume V - v+Z not occupied by positive ions of
volume v+
(3.149)

The number OZIl' of negative ions that form pairs with positive ions is
proportional, firstly, to the number OZ of negative ions added; secondly,
to the volume V~ZFl occupied by free positive ions; and, finally, to the
Boltzmann factor e-U,kT, where U is the potential energy of a negative
 ION-ION INTERACTIONS 263

ion in contact with a positive ion t

(3.150)

By dividing (3.150) by (3.149), the result is

OZIF~__ 2 ZFle-UlkT oZ
(3.151)
OZFI - v+ (V - v+Z) oZ
or
(3.152)

If dilute solutions are considered, one can neglect the total volume v+Z
occupied by positive ions in comparison with the volume V of the solution,
and, therefore,
(3.153)

Now suppose that one adds increments oZ of positive and negative

ions until a total of N positive and N negative ions are present in the vol-
ume V; this process is equivalent to integrating Eq. (3.153). The result is

N~
N IF -- _F_I V e-UlkT
V + (3.154)
or
(3.155)

As a model of an ion pair, one can consider the positive ions to be

charged spheres of radius a and the negative ions point charges; this gives a
contact distance of a. Then, the volume of the cations is given by

(3.156)

With regard to U, the potential energy of a negative ion in contact

with a positive ion, the following argument is adopted. It has been shown
that the potential "Pr at a distance r from a central positive ion kf Eq.
(3.99)] is
e __
z :......Q.
- ----=' 1 _ eX (a-r)
"Pr - cr 1 xa + (3.157)

t The factor of 2 comes in because, when one adds 15Z negative ions, one must also add
15Z positive ions, which also form ion pairs.
264 CHAPTER 3

Hence, at the surface of the positive ion of radius r = a, the potential is

(3.158)

It follows therefore that

(3.159)

or
b
(3.l60)
1 + xa 1 + xa

where, as stated earlier,

(3.134)

By substituting for UjkT and v+ from (3.l60) and (3.l56) in (3.155), the
result is
NIP = ( NFl )2 ~ na 3 eb/ (1+ x a) (3.161)
V V 3
But
(3.l62)

where CIP and C are the concentration of ion pairs and electrolyte, respec-
tively, and
NFl (l - e)cNA
V- = nF1 = 1000 (3.163)

where CFI is the concentration of free ions. Hence, from (3.161), (3.l62),
and (3.163).
(3.164)

If the solution is considered dilute, 1 - e '"" 1, and, since X-I -+ 00 (i.e.,

x -+ 0),
(3.l65)

in which case, Eq. (3.164) becomes

e_ 4nNA 3 b
(3.166)
C- 3000 a e
 ION-ION INTERACTIONS 265

But, according to (3.141),

(3.141)

Hence, according to the Fuoss approach, the ion-association constant is

given by
_ 4nNA 3 b
KA - 3000 a e (3.167)

in contrast to the following expression (3.144) from the Bjerrum approach,

(3.144)

Now, the Bjerrum theory was tested in solutions with dielectric constants
such that b = zfz_e o2/wkT = 2q/a was significantly larger than 2. Under
these conditions, the following approximation to the integral in Bjerrum's
equation can be made

( z+z_eo2 3 eb
)3 Jb eYy -4 dy",-,a- (3.168)
EkT 2 b

in which case, the ion association constant of Bjerrum reduces to

(3.169)

As the dielectric constant of the solution is changed, b = z+z_e o2 jwkT

changes, but this change is overshadowed by the eb term. In other words,
the experimental results do not permit a distinction between the Fuoss
dependence of K.4 on eb [cf Eq. (3.167)] and the Bjerrum dependence of
KA on ebjb kf Eq. (3.169)]. However, the Fuoss approach is to be preferred
because it is on a simpler and less arbitrary conceptual basis.

3.8.7. From Ion Pairs to Triple Ions to Clusters of Ions ...

The coulombic attractive forces given by z+Le o2/cr 2 are large when
the dielectric constant is small. When nonaqueous solvents of low dielectric
constant are used, the values of dielectric constant are small. In such
solutions of electrolytes, therefore, it has already been stated that ion-pair
formation is favored.
Suppose that the electrostatic forces are sufficiently strong; then, it
may well happen that the ion-pair "dipoles" may attract ions and triple ions
266 CHAPTER 3

are formed thus

M+ + (A-M+)ion pair = [M+A-M+ltriplc ion (3.170)

or
(3.171)

From uncharged ion pairs, charged triple ions have been formed. These
charged triple ions playa role in determining activity coefficients. Triple-
ion formation has been suggested in solvents for which E < 15. The question
of triple-ion formation can be treated on the same lines as has been done
for ion-pair formation.
Further decrease of dielectric constant below a value of about 10 may
make possible the formation of still larger clusters of four, five, or more
ions. In fact, there is some evidence for the clustering of ions into grcups
containing four ions in solvents of low dielectric constant.

Further Reading
1. S. Arrhenius, Z. Phys. Chem., 1, 631 (1887).
2. N. Bjerrum, Kgl. Danske Videnskab. Selskab., 4: 26 (1906).
3. N. Bjerrum, Kgl. Danske Videnskab. Selskab., 7: 9 (1926).
4. R. M. Fuoss and C. A. Kraus, J. Am. Chem. Soc., 55: 476 (1933).
5. R. M. Fuoss and C. A. Kraus, J. Am. Chem. Soc., 55, 1019 (1933).
6. R. M. Fuoss, Z. Physik, 35: 59 (1934).
7. R. M. Fuoss, Trans. Faraday Soc., 30: 967 (1934).
8. R. M. Fuoss, Chem. Rev., 17: 27 (1935).
9. J. G. Kirkwood, J. Chem. Phys., 18: 380 (1950).
10. R. M. Fuoss and L. Onsager, Proc. Natl. Acad. Sci. U.S., 41: 274, 1010 (1955).
11. J. T. Denison and J. B. Ramsey, J. Am. Chem. Soc., 77: 2615 (1955).
12. W. R. Gilkerson, J. Chem. Phys., 25: 1199 (1956).
13. R. M. Fuoss and L. Onsager, J. Phys. Chem., 61: 668 (1957).
14. H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solution,
3rd ed., Reinhold Publishing Corp., New York, 1957.
15. R. M. Fuoss, J. Am. Chem. Soc., 80: 5059 (1958).
16. R. A. Robinson and R. H. Stokes, Electrolyte Solutions, 2nd ed., Butter-
worth's Publications, Ltd., London, 1959.
17. R. M. Fuoss and F. Accascina, Electrolytic Conductance, Interscience Pub-
lishers, Inc., New York, 1959.
18. C. B. Monk, Electrolytic Dissociation, Academic Press, London, 1961.
19. C. W. Davies, Ion Association, Butterworth's Publications, Ltd., London, 1962.
20. S. Levine and D. K. Rozenthal, "The Interaction Energy of an Ion Pair in
an Aqueous Medium," in: B. E. Conway and R. G. Barradas, eds., Chem-
ical Physics of Ionic Solutions, John Wiley & Sons, Inc., New York, 1966.
 ION-ION INTERACTIONS 267

21. J. E. Prue, "Ion Association and Solvation," in: B. E. Conway and R. G.

Barradas, eds., Chemical Physics of Ionic Solutions, John Wiley & Sons, Inc.,
New York, 1966.
22. G. H. Nancollas, "Thermodynamic and Kinetic Aspects of Ion Association
in Solutions of Electrolytes," in: B. E. Conway and R. G. Barradas, eds.,
Chemical Physics of Ionic Solutions, John Wiley & Sons, Inc., New York, 1966.
23. R. M. Fuoss, Proc. Nat. A cad. Sci., 57, 1550 (1967).

3.9. THE QUASI-LATTICE APPROACH TO CONCENTRATED

ELECTROLYTIC SOLUTIONS

3.9.1. At What Concentration Does the Ionic-Cloud Model

Break Down?

A powerful stimulus for a somewhat different model for concentrated

solutions has been an examination of the concentration dependence of the
validity of the Debye-Hiickel ionic-cloud model. A simple question is
posed: According to the ionic-cloud picture, what is the number of ions
responsible for any given fraction of the influence of the ionic cloud on
the central ion? In 10-8 mole liter-1 solutions of 1 : I-valent electrolytes,
calculations show that 474 ions produce about 50% of the effect of the
ionic cloud on the central ion. To achieve this effect, the 474 ions must be
smeared around the central ion over a spherical region of a radius of
21,000 A. Under these circumstances, it is quite legitimate to argue that
the central ion sees a smoothed-out cloud of charge around it and one
can use the Poisson equation with its implication of a continuous charge
distribution.
However, as the electrolyte concentration is increased, the situation
ceases to be so satisfactory. Thus, in 10-2 mole liter-1 solutions, only one
ion produces the 50% of the effect of the ionic atmosphere on the central
ion. To do this, the ion must be smeared around the central ion over a
spherical volume with a radius of only 25 A and yet yield a spherically
symmetrical continuous charge distribution. This is an impossible demand.
Hence, under these circumstances, the central ion experiences a discrete
charge, not a smoothed-out cloud of charge.
It has been concluded, therefore, that any specified fraction of the ion
cloud contains a number of ions which decreases as the concentration
increases. In other words, the cloud gets more and more coarse grained;
smoothness decreases, discreteness increases. The increase of coarse grain-
edness leads to large fluctuations with time of the electrostatic potential 'I/J.
Hence, the model of a continuous, smoothed-out charge density-the ionic
 268 CHAPTER 3

cloud-must break down at some concentration less than 10-2 mole liter-I.
But at what concentration?
The question at issue is how coarse grained the cloud is. The measure
of coarse grained ness shall be taken to be the dimension of the ion cloud
compared to the mean distance between ions. This decision has the follow-
ing rationale: The Deybe-Hiickel ionic-cloud model implies that neighbor-
ing ions in dilute solutions are so far apart that the short-range noncou-
lombic interaction of a neighboring ion on the central ion is less important
than the long-range coulombic interactions which set up the ionic cloud.
Now, the long-range interaction between an ion and the ionic cloud can
be taken as equivalent to the interaction between the central ion and the
electrical image (i.e., an ion of opposite charge) situated at a distance apart
of X-I. Thus, the relative values of X-I and I, the mean distance between
ions, provides an indication of the relative importance of long-range and
short-range nearest-neighbor interactions.
The mean distance 1 between ions is obtained thus: If c is the concen-
tration of the electrolyte in moles per liter, then 2cNA ions t are contained
in 1000 cm3 • The volume occupied by one ion is 1000/2NA c, and, therefore,
the average distance 1 apart of any two ions is

(3.172)

If X-I> I, then long-range interactions outweigh the nearest-neighbor

interactions, the ion cloud is fine grained, and the Debye-Hiickel theory
is in its element. If X-I < I, the cloud cannot be fine grained for the average
nearest ion is not within it. It can be shown that for 1 : I-valent electrolytes
dissolved in water at 25°C

or (3.173)

r
and

I= ( ;~~~ = 9.398c-! [A] (3.174)

Figure 3.47 has a plot of X-I and 1 versus ct. The two curves intersect at
c = CHm = 0.001 mole liter-I. At concentrations C < CHm, X-I> I, i.e., there
is fine grainedness of the ion cloud and the Debye-Hiickel approach is

t The factor 2 occurs because both positive and negative ions are counted.
 ION-ION INTERACTIONS 269

320

280

240
I/K
200
OC[
160
==0> 120
c
CD
...J
80

40

0
0 0.16 Q24 0.32
C~2
Fig. 3.47. The mean distance, 1 between ions and the
thickness of the ionic cloud, ,,-1, as a function of d.

fundamentally valid; at concentrations C > CHm, ,,-1 < I and the ion cloud
is too coarse grained to permit the type of smoothing of charge, etc., to
allow the application of the ionic-cloud model.
The concentration of 0.001 mole liter-1 can therefore be taken as a
limit for the validity of the simple ionic-cloud pictore. t

3.9.2. The Case for a Cube-Root Law for the Dependence of the
Activity Coefficient on Electrolyte Concentration
For concentrations above CHm, the mean distance apart of ions, i.e.,
I, would appear to be a fundamental quantity in a revised theory. But I is
obtained by taking the 2NA c ions and imagining them to be arranged in a
periodic manner to give a volume of 1000 cm3 . In such a regular arrange-
ment, or "lattice," the "lattice constant" is the mean distance apart of
the ions. Thus, I is a "lattice parameter," and the theoretical approach
to solutions more concentrated than CHm should be based on a "lattice
model."

t Of course, the calculation of the concentration at which coarse grainedness begins,

itself, depends through ,,-1 on the use of the linearized P-B equations. When ,,-1 < I,
the theory used for the calculation of the condition for coarse grainedness is not valid.
Hence, the above calculation of the concentration at which coarse grainedness sets
in is in fact on the borderline of validity and can only be considered to provide a rough
answer. However, in partial conformity with this answer are the data for Hel, which
show that the limiting law does begin to break down at 3 to 5 x to-aN.
270 CHAPTER 3

Since I ex c- 1, it would be expected that, if the parameter I (character-

izing a hypothetical lattice in solution) were of any importance, there
should be a dependence of the properties of an electrolytic solution (e.g.,
logf±) on d. Such an observation was indeed made before the advent of
the Debye-HUckel ionic-cloud theory. When log f± is plotted against d,
long linear regions can be noted (see Fig. 3.48) at concentrations above
about Co = 0.001 mole liter-I. This strengthens the argument for thinking
in terms of a lattice parameter, rather than an ionic cloud, for all but very
dilute solutions.
Some years before the Debye-HUckel theory, Ghosh had suggested
that, when an ionic crystal such as sodium chloride is dissolved in water,
it exists in the form of an expanded lattice. He obtained a dependence of
logf± on d by using the idea of lattice energy (cf Chapter 2). Thermal
motions, however, would surely disrupt and wash out the long-range order
so much a characteristic of a lattice. In fact, it was the realization of this
point that stimulated Debye and HUckel to develop the ionic-cloud model
according to which the excess charge density decays monotonically with
distance. Sufficiently far from the central ion, the excess charge density is
virtually equal to zero, i.e., indicates equal probabilities of finding negative
and positive ions. However, with increasing electrolyte concentration, the
ionic cloud becomes more and more coarse grained, which indicates short-
range order and a quasi-lattice structure. Under these conditions of elec-
trolyte concentration, a distance characteristic of the short-range (lattice-
like) order, rather than x-I, becomes important in calculating interactions
in solution. This distance would be related to I, the mean distance apart
of the ions.
-002

000 ,

0.02
........
0.04
'"
0
...J
0.06

o.oa
0.10

0.12
0.14
000 010 0.20 Y 0.30 0.40
c 3

Fig. 3.48. Mean activity coefficient f ± as a function of ci.

 ION-ION INTERACTIONS 271

These matters relevant to the possibility of a quasi lattice in con-

centrated aqueous solutions are closely connected to structural models for
pure liquid electrolytes (Chapter 6). Thus, one view of a molten electrolyte
is of the regular lattice of the parent ionic crystal from which about 10
to 30% of the ions are removed to give vacancies, the remaining ions being
placed in rapid motion. This pure liquid electrolyte would still show ap-
preciable vestiges (over local regions) of the lattice structure of the solid
crystal. Around any particular ion, the opposite charge would not be
smeared out into a density which falls away monotonically with radial
distance away from the ion. On the other hand, the charge density would
show periodic, damped maxima. These correspond to the probability of
finding alternate positive and negative ions, as in a lattice. The damp-
ing represents the decrease in the ordering of ions-hence, the term quasi
lattice.

3.9.3. The Beginnings of a Quasi-Lattice Theory for Concen-

trated Electrolytic Solutions

The basic characteristics which a quasi-lattice theory must possess in

order satisfactorily to account for the behavior of concentrated electrolytic
solutions are now considered.
The contribution of interactions to the chemical potential of an ionic
species must be proportional to the average interaction energy for a pair
of ions, i.e., to ZieOZjeO/El. Why is this an average interaction energy? Be-
cause I is the average ion-ion distance. The proportionality constant must
express the "latticeness" of the electrolytic solution and the "quasiness"
of the lattice.
The latticeness needs the use of a factor which describes the arrange-
ment of ions in its geometrical aspect. For ionic crystals, the geometry of
the particular arrangement of ions is taken into account in calculations of
lattice energy (cf Chapter 2) by the Madelung number. Hence, a Madelung-
like number M* can be inserted into the proportionality constant as an
ion-arrangement factor. One has, through the Madelung-like number,
some view of a greatly expanded and thermally smeared-out lattice. Over
local regions, there is an alternation of positive and negative ions, but
thermal forces wash out any long-range order. The exact value of the
Madelung-like number depends, for instance, on the average locations of
remote ions. Since periodicity is not simply analyzable in the liquid, the
values of the Madelung number will differ from those in the crystalline
state.
272 CHAPTER 3

To bring out the quasiness of the lattice, it is necessary, firstly, to insert

an inverse function of the temperature and, secondly, to include terms for
the short-range interactions which produce the ordering. The quasiness
factor will represent the disintegration of the lattice (i.e., the decrease in the
degree of order) as the temperature increases. By analogy with the Debye-
Hiickel dependence of interaction energy on T-~, the proportionality con-
stant may be considered to include a function of T-i, The functional rela-
tionship, yet unknown, is intended to represent short-range interactions
which produce local order.
By such physical arguments, Frank and Thompson have arrived at a
dimensionally acceptable guess in an expression for the activity coefficient.
Their expression has the form

- log!± = a - f3c! + yet (3.175)

in accordance with experiment.

It is important to emphasize that what has been accomplished is an
essay in interpretation rather than a theory. The matters discussed here
lie at the frontiers of research into the structure of concentrate solution.
The position is that no rigorous quasi-lattice theory exists today. The
problem has been attacked by several authors but has so far resisted un-
ravelment. The attempts do not survive simple criteria, for example, that
the Debye and Guntelberg charging processes must yield the same ex-
pression for the logarithm of the activity coefficient (Section 3.7).
The essential problem remains. It is the mathematical description of the
time-varying, thermally smeared partial order of ions in the electrolytic
solution, under conditions in which coarse-grained ness makes the ionic
cloud model invalid.

Further Reading

1. J. C. Ghosh, J. Chern. Soc., 113: 449, 707 (1918).

2. L. Onsager, Chern. Rev., 13: 73 (1933).
3. H. S. Frank, J. Chern. Phys., 13: 478 (1945).
4. R. H. Fowler and E. A. Guggenheim, Statistical Thermodynamics, Cambridge
University Press, New York, 1952.
5. R. A. Robinson and R. H. Stokes: Electrolyte Solutions, Butterworths Scientific
Publications Ltd., London, 1955.
6. H. S. Frank and P. T. Thompson, in: W. Hamer, ed., The Structure of Elec-
trolyte Solutions, John Wiley & Sons, New York, 1960.
 ION-ION INTERACTIONS 273

7. H. S. Frank, "Solvent Models and the Interpretation of Ionization and Solv-

ation Phenomena," in: B. E. Conway and R. G. Barradas, eds. Chemical
Physics 0/ Ionic Solution, John Wiley & Sons, Inc., New York, 1966.
8. S. R. Cohen, Trans. Faraday Soc., 63 (9): 2225 (1967).

3.10. THE STUDY OF THE CONSTITUTION OF ELECTROLYTIC

SOLUTIONS
3.10.1. The Temporary and Permanent Association of Ions
The effect of the temporary association of ions on the mean activity
coefficient has been considered in Section 3.8. The association of positive
and negative ions in ion pairs removed them from consideration in the
ionic cloud, and the Debye-Hiickel activity coefficient pertains only to the
free ions.
In the case of ion pairs, the forces are coulombic (electrostatic) in
origin. The ion pairs are formed when the thermal-dissociating forces are
exceeded by the coulombic-associating forces. But are the thermal forces
permanently exceeded?
For the following reasons, it is unlikely that ion pairs are permanent.
Pair formation involves solvated ions, and, hence, the center-to-center dis-
tance of the two ions is about 5 to 10 A. Coulombic forces are long range
in character. The electrostatic interaction between a pair of oppositely
charged ions may therefore be sufficient to overcome thermal forces tempor-
arily, but, at some later instant, one ion of the pair may be persuaded by a
momentum-imparting collision to break away from its partner. The as-
sociation of ions in ion pairs is therefore generally a temporary affair.
Ion pairs have a short lifetime.
Such electrostatic ion-pair formation does not exhaust the theoretical
possibilities or the experimental realities. When a cation and an anion
come closer than about 5 to 10 A, short-range chemical forces between
the ions are likely to predominate, particularly in the absence of intervening
solvent molecules. In such cases, where cations and anions are adjacent
and not separated by neutral solvent molecules, there is the possibility of
forming bonds stronger and more lasting than electrostatic interactions.
Such "chemical" interaction is distinguished from the coulombic variety
by stating that a complex-in contrast to an ion pair-is formed.
The formation of complexes would also lead to a deviation of the
Debye-Hiickel activity coefficient from the observed value. How, then,
to distinguish between ion-pair formation and complex formation? This
question increases the need for the study of the constitution of electrolytic
solutions.
 274 CHAPTER 3

3.10.2. Electromagnetic Radiation; a Tool for the Study of

Electrolytic Solutions

The understanding of the concentration dependence of activity coef-

ficients required the postulate of the concepts of ion-pair formation and
complex formation. Certain structural questions, however, could not be
answered unequivocally by these considerations alone. For instance, it was
not possible to decide whether pure coulombic or chemical forces were
involved in the process of ion association, i.e., whether the associated entities
were ion pairs or complexes. The approach has been to postulate one of
these types of association, then to work out the consequences of such an
association on the value of the activity coefficient, and finally to compare
the observed and calculated values. Proceeding on this basis, it is inevitable
that the postulate will always stand in need of confirmation because the
path from postulate to fact is indirect.
It is fortunately possible to gain direct information on the constitution
and concentrations of the various species in an electrolytic solution by
"seeing" the various species. These methods of seeing are based upon
shining "light"t into the electrolytic solution and studying the light which
is transmitted, scattered, or refracted. The light emerging from the elec-
trolytic solution, in the ways just mentioned, is altered or modulated as a
result of the interaction between the free ionic or associated species and
the incident light.
There are many types of interactions between electromagnetic radiation
and matter. The species as a whole can change its rotational state; the bonds
(if any) within the species can bend, stretch, or twist and thus alter its
vibrational state; the electrons in the species may undergo transitions be-
tween various energy states; and, finally, the atomic nuclei of the species
can absorb energy from the incident radiation by making transitions be-
tween the different orientations to an externally imposed magnetic field.
All these responses of a species to the stimulus of the incident electro-
magnetic radiation involve energy exchange. This energy exchange must,
according to the quantum laws, occur only in finite jumps of energy. Hence,
the light which has been modulated by these interactions contains inform-
ation about the energy that has been exchanged between the incident light
and the species present in the electrolytic solution. If the free, unassociated
ions and the associated aggregates (ion pairs, triplets, complex ions, etc.)

t The word light is used here to represent not only the visible range of the electro-
magnetic spectrum but all the other ranges for which analytical methods have been
developed.
 ION-ION INTERACTIONS 275

interact differently with the incident radiation-and they must at some

frequency (each species has a characteristic frequency)-, then information
concerning the structures will be observed in the radiation emerging from
the electrolytic solution.
These aspects of the interaction of radiation and matter are the concern
of absorption spectroscopy in the most general sense of the term. It is cus-
tomary, however, to use different terms depending on the wavelength of
the incident radiation used and the particular types of interaction involved.
Thus, one speaks of various kinds of spectroscopy-visible and ultraviolet
(UV) absorption spectroscopy, Raman and infrared (lR) spectroscopy,
nuclear magnetic resonance (NMR) spectroscopy, and electron-spin reso-
nance (ESR) spectroscopy. A brief description of the principles of these
techniques and their application to the study of electrolytic solutions will
now be given.

3.10.3. Visible and Ultraviolet Absorption Spectroscopy

Atoms, neutral molecules, and ions (simple and associated) can exist
in several possible electronic states; this is the basis of visible and UV
absorption spectroscopy. Transitions between these energy states occur by
the absorption of discrete energy quanta AE which are related to the fre-
quency v of the light absorbed by the well-known relation

AE = hv

where h is Planck's constant. When light is passed through the electrolytic

solution, there is absorption at the characteristic frequencies corresponding
to the electronic transitions of the species present in the solution. Owing
to the absorption, the intensity of transmitted light of the absorption fre-
quency is less than that of the incident light; and the fall-off in intensity at a
particular wavelength is given by an exponential relation known as the
Beer-Lambert law

where /0 and / are the incident and transmitted intensities, c is a character-

istic of the absorbing species and is known as molar absorptivity, I is the
length of the material (e.g., electrolytic solution) through which the light
passes, and c is the concentration of the absorbing molecules.
A historic use of the Beer-Lambert law was by Bjerrum, who studied
the absorption spectra of dilute copper sulphate solutions and found that
the molar absorptivity was independent of the concentration. Bjerrum
276 CHAPTER 3

concluded that the only species present in dilute copper sulphate solutions
are free, unassociated copper and sulphate ions and not, as was thought
at the time, undissociated copper sulphate molecules which dissociate into
ions to an extent which depends on the concentration. For, if any undis-
sociated molecules were present, then the molar absorptivity of the copper
sulphate solution would have been dependent on the concentration.
In recent years, it has been found that the molar absorptivity of con-
centrated copper sulphate solutions does show a slight concentration de-
pendence. This concentration dependence has been attributed to ion-pair
formation occurring through the operation of coulombic forces between
the copper and sulphate ions. This is perhaps ironic because Bjerrum's
concept of ion pairs is being used to contradict Bjerrum's conclusion that
there are only free ions in copper sulphate solutions. Nevertheless, there is
a fundamental difference between the erroneous idea that a copper sulphate
crystal dissolves to give copper sulphate molecules, which then dissociate
into free ions, and the modern point of view that the ions of an ionic crystal
pass into solution as free solvated ions which, under certain conditions,
associate into ion pairs.
The method of visible and UV absorption spectroscopy is at its best
when the absorption spectra of the free ions and the associated ions are
quite different and known. When the associated ions cannot be chemically
isolated and their spectra studied, the type of absorption by the associated
ions has to be attributed to electronic transitions known from other well-
studied systems. For example, there can be an electron transfer to the ion
from its immediate environment (charge-transfer spectra), i.e., from the
entities which are associated with the ion, or transitions between new elec-
tronic levels produced in the ion under the influence of the electrostatic
field of the species associated with the ion (crystal-field splitting). Thus,
there is an influence of the environment on the absorption characteristics
of a species, and this influence reduces the unambiguity with which spectra
are characteristic of species rather than of their environment. Herein lies
what may be considered a disadvantage of the technique of visible and UV
absorption spectroscopy.

3.10.4. Raman Spectroscopy

Visible and UV absorption spectroscopy are based on studying that

part of the incident light which is transmitted (after absorption) through
an electrolytic solution in the same direction as the original beam. There
is, however, a certain amount of light which is scattered in other directions.
 ION-ION INTERACTIONS 277

The scattered light in any direction consists mainly of light of the same
frequency as the incident radiation (in which case the phenomenon is
known as Rayleigh scattering) and gives no information concerning the
intramolecular structure of the molecules responsible for the scattering.
In addition, however, it is observed that the scattered beam includes light,
the frequency of which is shifted from the incident frequency by an amount
Llv either added or subtracted from the incident frequency; this phenomenon
is known as the Raman effect and the type of scattering, as Raman scattering.
The Raman shift ± Llv is due to energy exchanged ± LIE between the
incident radiation and the scattering species

LIE = h Llv
A simple view of the origin of the Raman effect is as follows: Rayleigh
scattering is produced because the electric field of the incident light induces
a dipole moment in the scattering species, and, since the incident field is
oscillating, the induced dipole moment also varies periodically. Such an
oscillating dipole acts as an antenna and radiates light of the same fre-
quency as the incident light.
It has already been stated (cf 2.5.4) that it is the deformation polar-
izability Udefurlll which determines the magnitude of dipole moment induced
by a particular field. If this polarizability changes from its time-average
value, then the induced-dipole antenna will be radiating at a new frequency
that is different from the incident frequency; in other words, there is a
Raman shift. Now the changes in polarizability of the scattering species
can be correlated with their rotations and vibrations and also their symmetry
characteristics. Hence, the Raman shifts are characteristic of the rotational
and vibrational energy levels of the scattering species and provide direct
information about these levels. Though the presence of electrostatic bond-
ing produces second-order perturbations in the Raman lines, it is the
species with covalent bonds to which Raman spectroscopy is sensitive
and not ion pairs.
Another feature which makes Raman spectra useful is the fact that
the integrated intensity of the Raman line is in general proportional to
the concentration of the scattering species.
The problem with Raman spectroscopy is the low intensity of the
Raman lines, which permits an easy detection of species only in concen-
trations of at least 10%. Thus, at low concentrations of the scattering species,
the method presents great difficulties in separating out the signal from
the noise, i.e., identifying the Raman lines from the ever-present general
background intensity. Fortunately, the availability of lasers, which are
278 CHAPTER 3

intense sources of monochromatic light, is stimulating further applica.

tions of this powerful technique which is noted for the lack of ambiguity
with which it can report on the species in an electrolytic solution. Devices
which distinguish a low-intensity signal from noise, recently available,
also help.

3.10.5. Infrared Spectroscopy

Infrared spectroscopy resembles Raman spectroscopy in that it provides

information on the vibrational and rotational energy levels of a species,
but it differs from the latter technique in that it is based on studying the
light transmitted through a medium after absorption, and not that scattered
by it.
The techniques of Raman and IR spectroscopy are generally considered
complementary in the gas and solid phases because some of the species
under study may reveal themselves in only one of the techniques. Never-
theless, it must be stressed that Raman scattering is not affected by the
aqueous medium, whereas strong absorption in the infrared shown by
water proves to be a troublesome interfering factor in the study of aqueous
solutions by the IR method.

3.10.6. Nuclear Magnetic Resonance Spectroscopy

The nuclei of atoms can be likened in some respects to elementary
magnets. In a strong magnetic field, the different orientations that the
elementary magnets assume correspond to different energies. Thus, trans-
itions of the nuclear magnets between these different energy levels correspond
to different frequencies of radiation in the short-wave, radio-frequency
range. Hence, if an electrolytic solution is placed in a strong magnetic
field and an oscillating electromagnetic field is applied, the nuclear magnets
exchange energy (exhibit resonant absorption) when the incident frequency
equals that for the transitions of nuclei between various levels.
Were this NMR to depend only on the nuclei of the species present in
the solution, the technique would be without point for the identification
of species in a solution. But the nuclei sense the applied field as modified
by the environment of the nuclei. The modification is almost exclusively
due to the nuclei and electrons in the neighborhood of the sensing nucleus,
i.e., due to the adjacent atoms and bonds. Thus, NMR studies can be
used to provide information on the type of association between an ion
and its environmental particles, e.g., on ion-solvent interactions or ion
association.
 ION-ION INTERACTIONS 279

Further Reading
I. J. E. Hinton and E. S. Amis, Chem. Rei!., 67: 367 (1967).
2. T. T. Wall and D. F. Hornig, J. Chon. Phys., 47: 784 (1967).
3. N. A. Matwiyoff, P. E. Darby, and W. G. Movius, II/org. Chem., 7: 2173
(1968).
4. E. R. Malinowski and P. S. Knapp, J. Chem. Phys., 48: 4989 (1968).

3.11. A PERSPECTIVE VIEW ON THE THEORY OF ION-ION

INTERACTIONS

A host of new ideas have been described in the present chapter. It

is time to lay them out in perspective.
The development was triggered by the intuitive idea that the presence
of free ions in an electrolytic solution must surely lead to some phenomenon
caused by the energy of interionic attraction. Hence, the analysis of ion-ion
interactions are as important to the understanding of ionic solutions as
are ion-solvent interactions.
The original Debye-Hlickel theory was a brilliant first step in the
theoretical treatment of ion-ion interactions. It was worked out by assum-
ing a very simple idealized model: (1) point-charge ions; (2) only coulombic
forces between ions; (3) ions smoothed out into a charged continuum;
(4) a small enough electrostatic potential near an ion, to permit the linear-
ization of the Boltzmann equation.
Many beautiful results fell out of the theory. For instance, it became
possible to present a mathematical description of the distribution of excess
charge density as a function of distance from any ion chosen as a reference
ion. This distribution function also yielded a physical picture of ions in an
electrolytic solution surrounded by clouds of charge, the opposite charges
predominating. The interactions between an ion and the rest of the electro-
lytic solution thus became equivalent to the interaction between an ion and
its ionic cloud. This interaction between the reference (or sample) ion and
the ionic cloud could be quantitatively treated in a simple way. All one
had to do was to calculate the potential lPl'hll II \ , set up at the surface of the
reference ion by the ionic cloud and use this 11'..\"".\ in the expression for
chargillg up the reference ion. Thus \\as obtained the chemical-potential
change arising from the interactions between an ionic species and the ionic
solution.
The linkup between theoretical calculations and experimental measure-
ments was made on the basis that the chemical-potential change arising
from ion-ion interactions was set equal to the empirical correction factor
280 CHAPTER 3

RTlnf± which had been introduced to account for the fact that ionic
solutions do not behave like so-called ideal solutions of noninteracting
solute particles. In this manner, the Debye-Hiickel ionic-cloud model could
be made to yield the limiting law-an expression for the concentration
variation of the logarithm of the activity coefficient f ±. The experimental
logf± versus n curves corresponded more and more closely to the limiting
law as the solutions were made more and more dilute. But the limiting
law argued for a constant slope at all concentrations, whereas the experi-
mental curves showed a slope which decreased with concentration. In fact,
the curves passed through a minimum and even acquired a positive slope.
The theory had to be improved.
The first step in the development of the theory for concentrated so-
lutions was to give the ions a finite size rather than to consider them point
charges.
An ion-size parameter a was defined, and the beginning of the decrease
of slope of the theoretical logf± versus I~ curve could be explained. It is
important to realize that the introduction of an ion-size parameter was
equivalent to an incorporation into the model of a short-range repulsive
force which would forbid the ions to approach closer than a particular
distance a. The parameter a, being different for different electrolytes of the
same valence type, introduced specific differences between these electrolytes
and gave to each of them a different logf± versus /! curve, as is experiment-
ally observed. These specific differences between two electrolytes of the
same valence type were unforeseen by the point-charge version of the theory,
which argued that all electrolytes of the same valence type would have the
same logf± versus n curve.
The actual values of the ion-size parameter, which had to be used to
make theory fit experiment, were larger than the sum of the crystallographic
radii of naked ions. Hence, the value of the ion-size parameter obviously
reflected the attachment of a sheath of water molecules to an ion, i.e.,
the primary hydration sheath. To the extent, therefore, that the simple
version of the Debye-Hiickel theory relegated the solvent to the role of a
ubiquitous dielectric and gave it a nonessential role, one had to introduce
the influence of solvation upon ion-ion interactions through the sizes of
ions.
The effect of the sizes of ions was only a subsidiary influence of the
phenomenon of solvation. Its main effect arose in a different way. As a
solution became more and more concentrated, a greater and greater frac-
tion of the solvent became trapped in the solvent sheaths of ions, became
hors de combat, which resulted in an increase of the effective concentration.
 ION-ION INTERACTIONS 281

It was this influence of solvation on the activity coefficient which was of

greater significance than the impact of solvation on the sizes of ions. When
the effect of solvation on the activity coefficient was added onto the Debye-
HUckel activity coefficient, one obtained an expression for logf± which
made the logf± versus /l curve pass through a minimum and then turn
upward in accordance with experiment.
At this stage of the presentation, a fundamentally different approach
was tried. Taking the view that the potential 1jJr may not be small enough
to justify the linearization of the Boltzmann equation, the obvious solution
was to work with an unlinearized Boltzmann equation. It turned out,
however, that attempts to use the unlinearized P-B equation have hitherto
led to a logical inconsistency most clearly seen in the fact that the GUntelberg
and the Debye procedures for arriving at the chemical potential of ion-ion
interactions lead to different results even though the two procedures ought
to be equivalent.
The whole theory, till that stage, had been worked out on the basis
that every ion from the parent crystal remained free in solution. But ions
tend not only to aggregate around each other in clouds but also to stick
together or associate to form neutral pairs. The Debye-HUckel theory was
held to remain valid for free ions, but the formation of uncharged ion pairs
reduced the concentration of free ions. The ion-association model led to
the concept of a minimum distance q, closer than which two ions approach
only at the risk of sticking together as an ion pair. By using a corrected
value for ionic concentration and the Debye-HUckel model only for the
free ions, one obtained an expression for the activity coefficient which was
valid when the solvation corrections were not important.
As the ideas of finite ion size, solvation effects, and ion association
were invoked, the model of an ionic solution slowly approached reality.
The experimentallogf± versus Jl curves were quasi-empirically "reproduc-
ed" by the theory even at a very high concentration, e.g., 5M.
The gathering triumphs and advances had been accompanied, of course,
by less satisfactory aspects.
Thus, the finite-ion-size model could not generate the value of the
ion-size parameter from theoretical arguments alone. The effecfil'e size of
solvated ions in collision was too complex a many-body problem for ac-
curate calculation. The ion size had to be incorporated as an adjustable
parameter. In contrast, the Debye-HUckel limiting law represented an
"absolute" theory with no semiempirical parameters.
Likewise, the influence of solvation on the activity coefficient, i.e., the
effect of taking water molecules from their positions in the water network
282 CHAPTER 3

into the solvent sheaths around ions, was difficult to calculate. There were
difficulties in counting the number of water molecules "removed" from
the bulk water. The primary water molecules must, of course, be counted,
but how much of the secondary solvation sheath? It turned out that, by
using the primary hydration numbers alone, quite good consistency with
experiment is observed. But these hydration numbers too are usually ob-
tained by calibration from experiment (cf Section 2.4). Finally, in the
Stokes-Robinson treatment, it is assumed that the Debye-HUckel activity
coefficient must be modified by a solvation correction, i.e., it is assumed
that the basis of the Debye-HUckel model is valid in concentrated so-
lutions.
When, however, one makes a fundamental analysis of the Debye-
HUckel model, it becomes clear that, even at concentrations as low as
about 10- 3 mole liter-I, the thickness X-I of the ionic atmosphere is less than
the mean distance I between ions. In other words, the image charge repre-
senting the ionic cloud is closer to the central ion than is the average nearest-
neighbor ion. This is ridiculous, for the ionic cloud consists of ions. Thus,
there is a fundamental invalidity of the Debye-HUckel model above any
but the lowest practical concentrations.
A clue to an alternative approach for higher concentrations lies in
observed experimental results that indicate a linear relationship between
log f ± and d. This points to the necessity of developing a quasi-lattice
model for concentrated solutions-a feat which has not yet been achieved.

Appendix 3.1. Poisson's Equation for a Spherically Symmet-

rical Charge Distribution

A starting point for the derivation of Poisson's equation is Gauss's law

which can be stated as follows: Consider a sphere of radius r. The electric field
Xr due to charge in the sphere will be normal to the surface of the sphere and
equal everywhere on this surface. The total field over the surface of the sphere,
i.e., the surface integral of the normal component of the field, will be equal to
Xr times the area of the surface, i.e., XAnr2. According to Gauss's law, this
surface integral of the normal component of the field is equal to 4n/ f' times the
total charge contained in the sphere. If Cr is the charge density at a distance r
from the center of the sphere, then 4nr 2cr dr is the charge contained in a dr-thick
shell distance r from the center and f: 4nr 2cr dr is the total charge contained in
the sphere. Thus, Gauss's law states that

XAnr2 =
4n
-
IT 4nr2eT dr
I> 0
 ION-ION INTERACTIONS 283

i.e.,

r'X, = -4:<
E
J' r'f) dr
0
.r (A3.1.1 )

Now, according (0 the definition of electrostatic potential V'" at a point 1',

'I', = - J~ X, dr (2.4 )

or
dll',
X,= (A3.1.2)
dr

Substituting (A3.1.2) in (A3.1.1), one gets

1"
dV'r = -
-
dr
-4:<
E
J' 0
r'f) dr
-,
(A3.1.3)

Differentiating both sides with respect to 1',

~
dr
(r'~)
dr
_ 4:< r'f)
E
-r

i.e.,

(A3.1.4 )

which is Poisson's equation for a spherically-symmetrical charge distribution.

Appendix 3.2. Evaluation of the Integral f::~ e-1xr)(;q) d(%r)

According to the rule for integration by parts,

f V du = uv - f /I dv
Thus,

r->OO(xr)e- '" d(%r) = [ - %re- Yo '

. '-0
- J (-1)
e- d(%r)]'->oo
Yo
'

'-0

= +1 (A3.2.1)
284 CHAPTER 3

Appendix 3.3. Derivation of the Result f± :::: (f~+ f~_)l/V

Suppose that, on dissolution, I mole of salt gives rise to v + moles of positive

ions and v_ moles of negative ions. Then, instead of Eqs. (3.65) and (3.66),
one has
(A.3.3.I)
and
(A3.3.2)

Upon. adding the two expressions (A3.3.l) and (A3.3.2) and dividing by
v = v+ + v_ to get the average contribution to the free-energy change of the
system per mole of both positive and negative ions, the result is

J.I± =

which may be written as

J.I± = J.I±o + RTln x± + RTlnf± (A3.3.4)

where
v+J.I+o
J.I±o=-----
v_J.l_o + (A3.3.5)
v

(A3.3.6)
and
(A3.3.7)

Appendix 3.4. To Show That the Minimum in the P, versus r

Curve Occurs at r:::: 11./2

From
(A3.4.l )
one finds the minImum in the Pr versus , curve by differentiating the expres-
sion for Pr with respect to , and setting the result equal to zero. That is,

dPr A
- - = 4nn ·e , r 2r - 4nn .or 2e-./
A/
_1
r- =0 (A3.4.2)
dr • • 2 r
i.e.,
2'min - A = 0
or
,1.
rmin = q = 2 (A3.4.3)
 ION-ION INTERACTIONS 285

Appendix 3.5. Transformation from the Variable r to the Var-

iable y = Air

From y = i.lr, it is obvious that

(A3.S.I)

i.
dr = - -dy (A3.S.2)
y'

(A3.S.3)

Further, when r = q = i./2, it is clear that

y=2 (A3.SA)

and, when r = G, it follows that

i.
y = - =b (A3.S.S)
G

By introducing the substitutions (A3.S.I), (A3.S.2), (A3.S.3), (A3.S.4), and

(A3.S.S) into Eq. (3.13.1), the result is

or

(A3.S.6)

Appendix 3.6. Relation between Calculated and Observed Ac-

tivity Coefficients

Consider a solution consisting of I kg of solvent in which is dissolved m

moles of a I: I valent salt.
Ignoring the question of ion-pair formation, there are 11/ moles of positive
ions and m moles of negative ions in the solution. Hence, the total free energy,
G, of the solution is given by

G= ( ~ooo )/IH 20 - 111/1_ ...:.. 111/1_ (A3.6.1 )

H20

where AfH20 is the molecular weight of the water and fiH20' IL, and /1- are the
chemical potentials of water, positive ions, and negative ions, respectively.
286 CHAPTER 3

Now, if a fraction 0 of the positive and negative ions form ion pairs des-
ignated (+ -), then the free energy of the solution assuming ion-pair forma-
tion is

where the primed chemical potentials are based on ion association.

Since, however, the free energy of the solution must be independent of
whether ion association is considered or not, Eqs. (A3.6.l) and (A3.6.2) can be
equated. Hence
(A3.6.3)

But, ions are in equilibrium with ion pairs; therefore

(A3.6.4)

Combining Eqs. (A3.6.4) and (A3.6.3), one has

(A3.6.5)
or

po,. + RTln y+m + 11"- + RTln y_m = ,If + RTln y~(l - O)m + p.'!.'
+ RTln In r'-(l - O)m (A3.6.6)

It is clear, however, that 1'". = 11f and 1'"- = I''!.' since, whether ion associa-
tion is considered or not, the standard chemical potentials of the positive and
negative ions are the chemical potentials corresponding to ideal solutions of unit
molality of these ions. Hence

(A3.6.7)

or
y± = (1 - O)y±' (A3.6.8)

But, Y ± is the observed or stoichiometric activity coefficient, and, therefore,

(A3.6.9)
Similarly,
(3.146)
CHAPTER 4

ION TRANSPORT
IN SOLUTIONS

4.1. INTRODUCTION

The interaction of an ion in solution with its environment of solvent

molecules and other ions has been the subject of the previous two chapters.
Now, attention will be focused on the motion of ions through their environ-
ment. The treatment will restrict itself to solutions of true electrolytes.
There are two aspects to these ionic motions. Firstly, there is the
individual aspect. This concerns the dynamical behavior of ions as individu-
als-the trajectories they trace out in the electrolyte, and the speeds with
which they dart around. These ionic movements are basically random in
direction and speed. Secondly, ionic motions have a group aspect which
is of particular significance when more ions move in certain directions
than in others and thus produce a drift, or flux, t of ions. This drift has
important consequences because an ion has a mass and bears a charge.
Consequently, the flux of ions in a preferred direction results in the transport
of matter and a flow of charge.
If the directional drift of ions did not occur, the interfaces between
the electrodes and electrolyte of an electrochemical system would run out

t The word flux occurs very frequently in the treatment of transport phenomena. The
flux of any species i is the number of moles of that species crossing a unit area of a
reference plane in 1 sec; hence, flux is the rate of transport.

287
288 CHAPTER 4

INCREASING CONCENTRATION
Of@IONS •
• DIFFUSION OF Cit> IONS

Fig. 4.1. The diffusion of positive ions

resulting from a concentration gradient of
these ions in an electrolytic solution. The
directions of increasing ionic concentration
and of ionic diffusion are shown below the
diagram.

of ions to fuel the charge-transfer reactions which occur at such interfaces.

Hence, the movements and drift of ions is of vital significance to the con-
tinued functioning of an electrochemical system.
A flux of ions can come about in three ways. If there is a difference
in the concentration of ions in different regions of the electrolyte, the result-
ing concentration gradient produces a flow of ions. This phenomenon is
termed diffusion (Fig. 4.1). If there are differences in electrostatic potential
at various points in the electrolyte, then the resulting electric field produces
a flow of charge in the direction of the field. This is termed migration or
conduction (Fig. 4.2). Finally, if a difference of pressure or density or tem-
perature exists in various parts of the electrolyte, then the liquid begins
to move as a whole or parts of it move relative to other parts. This is hydro-
dynamic flow.
It is intended to restrict the present discussion to the transport processes
of diffusion and conduction and their interconnection. (The laws of hydro-
dynamic flow will not be described mainly because they are not particular
to the flow of electrolytes; they are characteristic of the flow of all gases
and liquids, i.e., of fluids). The initial treatment of diffusion and conduction
will be in phenomenological terms; then, the molecular happenings under-
lying these transport processes will be explored.
Now, in looking at ion-solvent and ion-ion interactions, it has been
possible to present the phenomenological or non structural treatment in the
framework of equilibrium thermodynamics, which excludes time and there-
fore fluxes, from its analyses.
Such a straightforward application of thermodynamics cannot be made,
 ION TRANSPORT IN SOLUTIONS 289

SOURCE OF POTENTIAL DIFFERENCE

/

ELECTRODE

ELECTRIC FIELD IN
ELECTROLYTE

INCREASING ELECTROSTATIC
POTENTIAL
.
.MIGRATION OF $ IONS

MIGRATION OF e IONS.

Fig. 4.2. The migration of ions resulting from a gradi-

ent of electrostatic potential (i.e., an electric field) in an
electrolyte. The electric field is produced by the applica-
tion of a potential difference between two electrodes
immersed in the electrolyte. The directions of increasing
electrostatic potentials and of ionic migration are shown
below the diagram.

however, in the consideration of transport processes. The drift of ions

occurs precisely because the system is not at equilibrium; rather, the system
is seeking to attain equilibrium. In other words, the system undergoes change
(there cannot be transport without temporal change!) because the free
energy is not uniform and a minimum everywhere. It is the existence of
such gradients of free energy which set up the process of ionic drift and
make the system strive for the attainment of equilibrium by the dissipation
of free energy. An appropriate framework for the phenomenological or
gross view of ionic drift may be that of nonequilibrium thermodynamics,
a subject concerned with the rates at which systems near equilibrium move
toward it.

4.2. IONIC DRIFT UNDER A CHEMICAL-POTENTIAL GRA-

DIENT: DIFFUSION

4.2.1. The Driving Force for Diffusion

It has been remarked in the previous section that diffusion occurs when
a concentration gradient exists. The theoretical basis of this observation
will now be examined.
Consider that, in an electrolytic solution, the concentration of an ionic
290 CHAPTER 4

-+----t~x

z
Fig. 4.3. A schematic representation of a slab of
electrolytic solution in which the concentration of a
species i is constant on the shaded equiconcen-
tration surfaces parallel to the yz-plane.

species i varies in the x direction but is constant in the y and z directions.

If desired, one can map equiconcentration surfaces (they will be parallel
to the yz plane) (Fig. 4.3).
The situation pictured in Fig. 4.3 can also be considered in terms of
the partial molar free energy, or chemical potential, of the particular species
i. This is achieved through the use of the defining equation for the chemical
potential [Eq. (3.56)]

(The use of concentration, rather than activity, implies the solution is

assumed to behave ideally.) Since Ci is a function of x, the chemical potential
also is a function of x. Thus, the chemical potential varies along the x
coordinate, and, if desired, equi-/l surfaces can be drawn. Once again,
these surfaces will be parallel to the yz plane.
Now, if one transfers a mole of the species i from an initial concentra-
tion CI at XI to a final concentration CF at XF, then the change in free energy,
or chemical potential, of the system is (Fig. 4.4):

(4.1)

But the change in free energy is equal to the net work done on the system
in an isothermal, constant-pressure reversible process (cf Appendix 2.1).
 ION TRANSPORT IN SOLUTIONS 291

CONCENTRATION,C I ,
AfoiO CHEMICAL.
POTENTIAL., III
X

'* .
CHEMICAL. POTENTiAL.
CHANGE' /olF-/J-I-
WORK DONE-W

Fig. 4.4. A schematic representation of the work W done

in transporting a mole of species i from an equiconcentration
surface where its concentration and chemical potential are c,
and IJ, to a surface where its concentration and chemical po-
tential are CF and IJF'

Thus, the work done to transport a mole of species i from Xl to XF is

W = Ll.u (4.2)

Think of the analogous situation in mechanics. The work done to lift

a mass from an initial height Xl to a final height XF is equal to the difference
in gravitational potential (energies), Ll U, at the two positions (Fig. 4.5):

W=LlU (4.3)

One goes further and says that this work has to be done because a gravita-
tional force FG acts on the body and that t

(4.4)

In other words, the gravitational force can be defined thus

LlU
FG = --- (4.5)
Llx

t The minus sign arises from the following argument: The displacement XI - XF of the
mass is upward and the force Fa acts downward; hence, the product of the displacement
and force vectors is negative. If a minus sign is not introduced, the work done W will
turn out to be negative. But it is desirable to have W as a positive quantity because
of the convention that work done on a system is taken to be positive, hence, a minus
sign must be inserted (see also Section 2.2.3).
292 CHAPTER 4

I
r'.
STRING

FINAL POSITION. ISPLA-

XF.OF MASS iCEMENT

1
I

f~INITIAL POSITION. XI.

DIFFERENCE IN POTENTIAL OF MASS
ENERGY"4U

GRAVITATIONAL FORCE. FG

Fig. 4.5. Schematic diagram to illustrate that the

mechanical work W done in lifting a mass from an
initial height XI to a final height XF is W = !::.U =
-FG(xl - XF)·

The potential energy, however, may not vary linearly with distance, and,
thus, the ratio LI UjLlx may not be a constant. So it is better to consider
infinitesimal changes in energy and distance and write

F __ dU
G - dx (4.6)

Thus, the gravitational force is given by the gradient of the gravitational

potential energy, and the region of space in which it operates is said to be
a gravitational field.
A similar situation exists in electrostatics. The electrostatic work done
in moving a unit charge from x to x + dx defines the difference d1p in elec-
trostatic potential between the two points (Fig. 4.6)

W=d1p (4.7)

Electric field. X

______ Initial position of

/ y
~Iacement • dx 'i'/
-----~~j
unit charge

ol~~?~ ~~~::: ~ of Increase.d",.

electrostatic -
I
potential I

Fig. 4.6. Schematic diagram to illustrate the electrostatic

work W = dIP = -X dx done in moving a unit positive
charge through a distance dx against an electric field X.
 ION TRANSPORT IN SOLUTIONS 293

Further, the electrostatic work is the product of the electric field, or force
per unit charge X and the distance dx t

-Xdx = d'IjJ (4.8)

or
d'IjJ
X=-- (4.9)
dx

The electric force per unit charge is therefore given by the negative of the
gradient of the electrostatic potentials, and the region of space in which
the force operates is known as the electric field.
Since the negative of the gradient of gravitational potential energy
defines the gravitational force and the negative of the gradient of electro-
static potential defines the electric force, one would expect that the negative
of the gradient of the chemical potential would act formally like a force.
Further, just as the gravitational force results in the motion of a mass and
the electric force, in the motion of a charge, the chemical-potential gradient
results in the net motion, or transfer, of the species i from a region of high
chemical potential to a region of low chemical potential. This net flow of
the species i down the chemical-potential gradient is diffusion, and, there-
fore, the gradient of chemical potential may be looked upon+ as the dif-
fusional force FD' Thus, one can write

dfti
FIJ = --- (4.10)
dx

by analogy with the gravitational and electric forces [Eqs. (4.6) and (4.9)]
and consider that the diffusional force produces a diffusional flux J, the
number of moles of species i crossing per second per unit area of a plane
normal to the flow direction (Table 4.1).

4.2.2. The "Deduction" of an Empirical Law: Fick's First Law

of Steady-State Diffusion
Qualitatively speaking, the macroscopic description of the transport
process of diffusion is simple. The gradient of chemical potential resulting

t The origin of the minus sign has been explained in Section 2.2.3.
+ It will be shown later on (cf Section 4.2.6) that, for the phenomenon of diffusion to
occur, all that is necessary is an inequality of the number of diffusing particles in dif-
ferent regions; there is, in fact, no directed force on the individual particles. Thus,
-d/l/dx is only a pseudo force like the centrifugal force; it is formally equivalent to a
force.
294 CHAPTER 4

TABLE 4.1
Certain Forces in the Phenomenological Treatment of Transport

Force Acting on Results in

dU
Mass Movement of mass
dx

dip
Charge Movement of charge (current)
dx

dfl
Species i Movement of species I, I.e.,
dx diffusional flux of species i

from a nonuniform concentration is equivalent to a driving force for dif-

fusion and produces a diffusion flux (Fig. 4.7). But what is the quantitative
cause-and-effect relation between the driving force dflJdx and the flux J?
This question must now be considered.
Suppose that when diffusion is occurring, the driving force F D and
the flux J reach values which do not change with time. The system can be

CHEMICAL
POTENTIAL

L---------------__ x
(a)

-1 ,~ ~

--=-=--_r--r-
Po,ltive ehemicol potential 9radilllt

-i Negative gradient.-!&,
L - -_ _
dx

-1 Driving farce far diffusion. Fa

(b)
Fig. 4.7. Schematic diagram to show (a) the
distance variation of the chemical potential of a
species i; and (b) the relative directions of the
diffusion flux, driving force for diffusion, etc.
 ION TRANSPORT IN SOLUTIONS 296

I
I TRANSIT PLANE(SHADED)ACROSS
I WHICH THE DIFFUSION FLUX,", is
I RECKONED
~---
/
/

ICONCENTRATION ,CI ,
ADJACENT TO TRANSIT PLANE

Fig. 4.8. Diagram for the derivation of the linear relation

between the diffusion flux J i and the concentration gra-
dient dCi/dx.

said to have attained a steady state. Then, the as-yet-unknown relation

between the diffusion flux J and the diffusional force FD can be represented
quite generally by a power series

(4.11 )

where A, B, C, etc., are constants. If, however, FD is less than unity and
sufficiently small, t the terms containing the powers (of FD) greater than
unity can be neglected.
Thus, one is left with
(4.12)

But the constant A must be equal to zero; otherwise, it will mean that
one would have the impossible situation of having diffusion even though
there is no driving force for diffusion.
Hence, the assumption of a sufficiently small driving force leads to the
result
J=BFD (4.13 )

i.e., the flux is linearly related to the driving force. Now, the value of FD = 0
(zero driving force) corresponds to an equilibrium situation; therefore, the
assumption of a small value of FD required to ensure the linear relation

t Caution should be exercised in applying the criterion. The value of FD which will give
rise to unity will depend on the units chosen to express FD. Thus, the extent to which
F D is less than unity will depend on the units, but one can always restrict F D to an
appropriately small value.
296 CHAPTER 4

TABLE 4.2
Diffusion Coefficient D of Ions in Aqueous Solutions

Diffusion coefficient,
Ion
cm 2 sec- 1

Li+ 1.028 X 10-5

Na+ 1.334 X 10-5
K+ 1.569 X 10-5
CI- 2.032 x 10-5
Br- 2.080 X 10-5

(4.13) between flux and force is tantamount to saying that the system is
near equilibrium, but not at equilibrium.
The driving force on 1 mole of ions has been stated to be -df-lJdx
[Eq. (4.10)]. If, therefore, the concentration of the diffusing species adjacent
to the transit plane (Fig. 4.8), across which the flux is reckoned, is Ci moles
per unit volume, the driving force Fn at this plane is -c.j(df-lddx). Thus,
from relation (4.13), one obtains

Jj = - BCi ti (4.14)
Writing

which is tantamount to assuming ideal behavior, Eq. (4.14) becomes

J. = - Bc. RT dCi = _ BRT dci (4.15)

t I Ci dx dx

Thus, the steady-state diffusion flux has been theoretically shown to be

proportional to the gradient of concentration. That such a proportionality
existed has been known empirically since 1855 through the statement of
Fick's first law of steady-state diffusion, which reads

J. =- D dCi (4.16)
, dx

where D is termed the diffusion coefficient (Table 4.2).

4.2.3. On the Diffusion Coefficient D

It is important to stress that, in the empirical Fick's first law, the
concentration C is expressed in moles per cubic centimeter, and not in
moles per liter. The flux is expressed in moles of diffusing material crossing
 ION TRANSPORT IN SOLUTIONS 297

QONOENTRATION

DISTANCE

Fig. 4.9. Diagram to show that diffusion flow

J j is in a direction opposite to the direction of
positive concentration gradient dCjl dx. Matter
flows downhill, i.e., down the concentration gra-
dient.

unit area of a transit plane per unit of time. i.e., in moles per square cen-
timeter per second, and, therefore, the diffusion coefficient D has the
dimensions of centimeters squared per second. The negative sign is usually
inserted in the right-hand side of the empirical Fick's law for the following
reason: The flux J; and the concentration gradient dc;(dx are vectors, or
quantities which have both magnitude and direction. But the vector J
is in an opposite sense to the vector representing a positive gradient de Jdx.
Matter flows downhill (Fig. 4.9). Hence, if J i is taken as positive, dcJdx
must be negative, and, if there is no negative sign in Fick's first law, the
diffusion coefficient will appear as a negative quantity-perhaps an un-
desirable state of affairs. Hence, to make D come out a positive quantity,
a negative sign is added to the right-hand side of the equation which states
the empirical law of Fick.
Equating the coefficients of deJdx in the phenomenological equation
(4.15) with that in Fick's law [Eq. (4.16)], it is seen that

BRT=D (4.17)

Now, is the diffusion coefficient a concentration-independent constant?

A naive answer would run thus: B is a constant, and, therefore, it appears
298 CHAPTER 4

that D also is a constant. But the expression (4.17) was obtained only
because an ideal solution was considered, and activity coefficients Ii were
ignored in Eq. (3.61). Activity coefficients, however, are concentration
dependent. So, if the solution does not behave ideally, one has, starting
from Eq. (4.14), and using (3.63),

d"i
J.1 = - Bc·--
I dx

d
= - BCi dx ("iO + RT Infici)
RT d
= - BCj r.c. dx liCj
JI I

= _ BRT dCi _ BRTci dfi

dx Ii dx
= _ BRT dCi _ BRTci dfi dCj
dx Ii dc, dx

= _ BRT dCi
dx
(1 + 2.. dfi )
Ii dCi
= _ BRT dCi
dx
(1 + dIn
dlnfi)
Ci
(4.18)

and, therefore,
D = BRT(l + dlnfi) (4.19)
din Ci

Rigorously speaking, therefore, the diffusion coefficient is not a constant

(Table 4.3). If, however, the variation of the activity coefficient is not

TABLE 4.3
Variation of the Diffusion Coefficient D with Concentration

Diffusion coefficient D in units of 10-5 cm 2 sec-1 at

concentration (in molarity)
Electrolyte
0.05 0.1 0.2 0.5

HCl 3.07 3.05 3.06 3.18

LiCl 1.28 1.27 1.27 1.28
NaCl 1.51 1.48 1.48 1.47
 ION TRANSPORT IN SOLUTIONS 299

significant over the concentration difference which produces diffusion, then

(cJfi)(8jJ8ci) <{ I, and, for all practical purposes, D is a constant. t This
effective constancy of D with concentration will be assumed in most of the
discussions presented here.
The treatment so far has been phenomenological, and, therefore, the
dependence of the diffusion coefficient on factors such as temperature and
type of ion can be theoretically understood only by an atomi~tic analysis.
Thus, the quantity D can be understood in a fundamental way only by
probing into the ionic movements, the results of which show up in the
macroscopic world as the phenomenon of diffusion. What are these ionic
movements, and how do they produce diffusion? The answering of these
two questions will constitute the next topic.

4.2.4. Ionic Movements: A Case of the Random Walk

Long before the movements of ions in solution were analyzed, the
kinetic theory of gases was developed and it involved a consideration of
the movements of gas molecules. The overall pattern of ionic movements
is quite similar to that of gas molecules, and, therefore, the latter will be
recalled first.
Imagine a hypothetical situation in which all the gas molecules, except
one, are at rest. The moving molecule will, according to Newton's first
law of motion, travel with a uniform velocity until it collides with a station-
ary molecule. During the collision, there is a transfer of momentum (mass
m times velocity v). So, the moving molecule loses some speed in the c011li-
sion, but the stationary molecule is set in motion. Now both molecules are
moving, and they will undergo further collisions. The number of collisions
will increase with time, and, soon, all the molecules of the gas will be con-
tinually moving, colliding, and changing their directions of motion and
their velocities-a scene of hectic activity.
It would be of interest to have an idea of the path of such a gas molecule
in the course of time. One might think that the detailed paths of all the
particles could be predicted by applying Newton's laws to the motions of
molecules. The problem, however, is obviously too complex for a practical
solution. To use the laws of motion requires a knowledge of the position
and velocity of each particle, and, even in I mole, there are 6.023 x 1023
(the Avogadro number) particles.

t For example, in diffusion between solutions which have a large concentration difference
such as 0.1 to 0.01 mole liter-I, a rough calculation suggests that the activity-coefficient
correction is of the order of a few per cent.
300 CHAPTER 4

SCALE

_-f7>-_FIBRE SUPPORT

Fig. 4.10. Schematic representation of the

essential parts of a mirror galvanometer, used
for detecting Brownian motion.

One can however try another approach. Is the ceaseless jostling of

molecules manifested in any gross (macroscopic) phenomenon? Consider
a frictionless piston in mechanical equilibrium with a mass of gas enclosed
in a cylinder. Owing to its weight, the piston exerts a force on the gas.
But what force balances the piston's weight? One says that the gas exerts
a pressure (force per unit area) on the piston owing to the continual buffet-
ing which the piston receives from the gas molecules. Despite this fact,
the bombardment by the gas molecules does not produce any visible motion
of the piston. Evidently, the mass of the piston is so large compared to that
of the gas molecules that the movements of the piston are too small to be
detected.
Now, let the mass of the piston be reduced. Then, the jiggling of the
extremely light piston as a result of being struck by gas molecules should
make itself apparent to an observer. This is what happens if one tries to
make a mirror galvanometer more and more sensitive. The essential part
of this instrument is a thin quartz fiber which supports a light coil of wire
seated in a magnetic field (Fig. 4.10). The deflections of the coil are made
visible by fixing a mirror onto the quartz fiber and bouncing a beam of
light off the mirror onto a scale. To increase the sensitivity of the instrument,
one tries lighter coils, lighter mirrors, and thinner fibers. There comes a
stage, however, when the "kicks" which the fiber-mirror-coil assembly
receives from the air molecules are sufficient to make the assembly jiggle
about. The reflected light beam then jumps about on the scale (Fig. 4.11).
The excursions of the spot about a mean position on the scale represent
 ION TRANSPORT IN SOLUTIONS 301

scale
reading

time
Fig. 4.11. The time variation of the reading on
the scale of a mirror galvanometer.

noise. (It is as if each collision produced a sound, in which case, the irregular
bombardment of the mirror assembly would result in a nonstop noise.)
Signals (coil deflections) which are of this same order of magnitude ob-
viously cannot be separated from the noise.
Instead of mirrors, one could equally consider pistons of a microscopic
size, large enough to be seen with the aid of a microscope but small enough
to display motions due to collisions with molecules. Such small "pistons"
are present in nature. A colloidal particle in a liquid medium behaves as
such a piston if it is observed in a microscope. It shows a haphazard, zig-
zag motion as shown in Fig. 4.12. The irregular path of the particle must
be a slow-motion version of the random-walk motion of the molecules in
the liquid.
One has therefore a picture of the solvated ions (in an electrolytic
solution) in ceaseless motion, perpetually colliding, changing direction,
hopping hither and thither from site to site. This is the qualitative picture
of ionic movements.

4.2.5. The Mean Square Distance Traveled in a Time t by a Ran-

dom-Walking Particle
The movements executed by an ion in solution are a three-dimensional
affair because the ion has a three-dimensional space available for roaming

Fig. 4.12. The haphazard zig-zag motion

of a colloidal particle.
 302 CHAPTER 4

+8

.. +5
.5
'1'
..~ O~~~~~--~~~-r

-8~----~S----~I~O----~IS----~20

Number of ,tepI . N

Fig. 4.13. The distance x (from the origin)

traversed by the drunken sailor in two tries,
each of N = 18 steps.

around. So one must really call the movements a "random flight" because
one flies in three dimensions and walks in two dimensions. This fine differ-
ence, however, will be ignored and the term random walk will be retained.
The aim now is to seek a quantitative description of ionic random-
walk movements. There are many exotic ways of stating the random-walk
problem. It is said, for example, that a drunken sailor emerges from a bar.
He intends to get back to his ship, but he is in no state to control the direc-
tion in which he takes a step. In other words, the direction of each step is
completely random, all directions being equally likely.
The question is: On the average, how far does the drunken sailor
progress in a time t?
For the sake of simplicity, the special case of one-dimensional random
walk will be considered. The sailor starts off from x = 0 on the x axis.
He tosses a coin: Heads!-he moves forward in the positive x direction.
Tails!-he moves backward. Since, for an "honest" coin, heads are as
likely as tails, the sailor is equally as likely to take a forward step as a
backward step. Of course, each step is decided on a fresh toss and is un-
influenced by the results of the previous tosses. After allowing him N steps,
the distance x from the origin is noted t (Fig. 4.13). Then, the sailor is
brought back to the bar (x = 0) and started off on another try of N steps.
The process of starting the sailor off from x = 0, allowing him N

t It will be zero only if an equal number NI2 of heads and tails turns up. This will not
happen every time. So, after each small number of tosses, the sailor is not certain to
be back where he started (i.e., in the bar).
 ION TRANSPORT IN SOLUTIONS 303

steps, and measuring the distance x traversed by the sailor is repeated

many times, and it is found that the distances traversed from the origin
are x(I), x(2), x(3), ... , xU), where xU) is the distance from the origin
traversed in the ith try. The average distance <x), from the origin is

<) = 7xCi) = Sum of distances from origin

(4.20)
x ~i Number of tries

Since the distance x traversed by the sailor in N steps is as likely to be in

the plus-x direction as in the minus-x direction, it is obvious from the
canceling out of the positive and negative values of x that, for a sufficiently
large number of tries, the mean progress from the origin is given by

(x) =0 (4.21)

It is obviously not very fruitful to compute the mean distance <x)

traversed by the sailor in N steps. To avoid such an unenlightening result,
which arises because xCi) can take either positive or negative values, it is
best to consider the square of xCi), which is always a positive quantity wheth-
er xci) itself is negative or positive. Hence, if x(I), x(2), x(3), ... , xci) are
all squared and the mean of these quantities is taken, then one can obtain
the mean square distance <x2 ), i.e.,

~ x(i)2 .
<x2 ) = _i _ _ = Sum of dIstances x squared
(4.22)
~ i Number of tries

Since the square of x(i), i.e., [x(i)]2, is always a positive quantity,

the mean square distance traversed by the sailor is always a positive non-
zero quantity (Table 4.4). Further, it can easily be shown (Appendix 4. I)
that the magnitude of <x2 ) is proportional to N, the number of steps, and,
since N itself increases linearly with time, it follows that the mean square
distance traversed by the random-walking sailor is proportional to time

(4.23)

It is to be noted that it is the mean square distance-and not the mean

distance-that is proportional to time. If the mean distance were propor-
tional to time, then it means that the drunken sailor (or the ion) is proceed-
ing at a uniform velocity. This is not the case because the mean distance
<x) traveled is zero. The only type of progress which the ion is making
from the origin is such that the mean square distance is proportional to
time. This is the characteristic of random walk.
304 CHAPTER 4

TABLE 4.4
One- Dimensional Random Walk of the Sailort

Trial NFt NB* x** x2

11 19 - 8 64
2 16 14 + 2 4
3 17 13 + 4 16
4 15 15 0 0
5 17 13 + 4 16
6 16 14 + 2 4
7 19 11 + 8 64
8 18 12 + 6 36
9 15 15 0 0
10 13 17 - 4 14
11 11 19 - 8 64
12 17 13 +4 16
13 17 13 + 4 16
14 12 18 - 6 36
15 20 10 +10 100
16 23 7 +16 256
17 11 19 - 8 64
18 16 14 + 2 4
19 17 13 + 4 16
20 14 16 - 2 4
21 16 14 + 2 4
22 12 18 - 6 36
23 15 15 0 0
24 10 20 -10 100
25 7 23 -16 256
26 15 15 0 0
<x) = 0 <x2) = 47.67

Total number of steps in each trial = N = 30.

Number of forward steps = N F .
* Number of ba.:kwards steps = N n .
** Distance from the origin = x.

4.2.6. Random-Walking Ions and Diffusion: The Einstein-Smo-

luchowski Equation

Consider a situation in an electrolytic solution where the concentration

of the ionic species of interest is constant in the yz plane but varies in the x
direction. To analyze the diffusion of ions, imagine unit area of a reference
plane normal to the x direction. This reference plane will be termed the
 ION TRANSPORT IN SOLUTIONS 305

unit area

Fig. 4.14. Schematic diagram for the derivation of the

Einstein-Smoluchowski relation. showing the transit plane
<x
T in between and at a distance vi 2 ) from the left Land
right R planes. The concentrations in the left and right
compartments are cL and CR. respectively.

transit plane T (Fig. 4.14). There is a random walk of ions across this plane
both from left to right and from right to left. On either side of the transit
plane, one can imagine two planes Land R which are parallel to the transit
plane and situated at a distance v' <x
2 ) from it. In other words, the region

under consideration has been divided into left and right compartments in
which the concentrations of ions are different and designated by CL and CR,
respectively.
In a time of t sec, a random-walking ion covers a mean square distance
of (x 2 ), or a mean distance of~. Thus, by choosing the plane L to
be at a distance ~ from the transit plane, one has ensured that all
the ions in the left compartment will cross the transit plane in a time t
provided they are moving in the left-+right direction.
Now, the number of moles of ions in the left compartment is equal
to the volume V <x
2) of this compartment times the concentration CL of
ions. It follows that the number of moles of ions which make the L -+ T
crossing in t sec is ~CL times the fraction of ions making left to right
movements. Since the ions are random walking, right to left movements
are as likely as left to right movements, i.e., only half the ions in the left
compartment are moving away toward the right compartment. Thus, in t
the number of moles of ions making the L -+ T crossing is <X2)CL ,tV
and, therefore, the number of moles of ions making the L -+ T crossing in
I sec is Hv' <x2 )/t)CL' Similarly, the number of moles of ions making the
R -+ T crossing in I sec is iCV (x 2 )/t)CR .
306 CHAPTER 4

Hence, the diffusion flux of ions across the transit plane, i.e., the net
number of moles of ions crossing unit area of the transit plane per second
from left to right is given by

(4.24)

This equation reveals that all that is required to have diffusion is a

difference in the numbers per unit volume of particles in two regions. The
important point is that no special diffusive force acts on the particles in the
direction of the flux.
If no forces are pushing particles in the direction of the flow, then what
about the driving force for diffusion, i.e., the gradient of chemical potential
(cf Section 4.2.1)? The latter is only formally equivalent to a force in a
macroscopic treatment; it is a sort of pseudoforce like a centrifugal force.
The chemical-potential gradient is not a true force which acts on the in-
dividual diffusing particles and, from this point of view, is quite unlike,
for example, the coulombic force which acts on individual charges.
Now, the concentration gradient dc/dx in the left-to-right direction
can be written
dc
dx - V <x 2)

or
V(X2) dc
cL - cR = - <x2 ) -
dx
(4.25)

This result for CL - CR can be substituted in Eq. (4.24) to give

J= _ ~ <x2 ) ~ (4.26)
2 t dx

and, by equating the coefficients of this equation with that of Fick's first
law [Eq. (4.16)], one has

or
(4.27)

This is the Einstein-Smoluchowski equation; it provides a bridge between

the microscopic view of random-walking ions and the coefficient D of the
macroscopic Fick's law.
 ION TRANSPORT IN SOLUTIONS 307

The coefficient 2 is intimately connected with the approximate nature

of the derivation, i.e., one-dimensional random walk with ion's being
permitted to jump forward and backward. More rigorous argument may
yield other values of the numerical coefficient, e.g., 6.
The characteristic of random walk in the Einstein-Smoluchowski
equation is the appearance of the mean square distance (i.e., square centi-
meters), and, since this mean square distance is proportional to time
(seconds), -the proportionality constant D in Eq. (4.27) must have the di-
mensions of centimeters squared per second. It must not be taken to mean
that every ion which starts off on a random walk travels in a time t a mean
square distance <x 2 ) given by the Einstein-Smoluchowski relation (4.27).

If a certain number of ions are, in a thought experiment, suddenly introduced

on the yz plane at x = 0, then, in t sec, some ions would progress a distance
Xl; others, X 2 ; still others, X 3 ; etc. The Einstein-Smoluchowski relation

only says that [cf Eqs. (4.22) and (4.27)]

(4.28)

But how many ions travel a distance Xl; how many, X 2 ; etc.? In other
words, how are the ions spatially distributed after a time t, and how does
the spatial distribution vary with time? This spatial distribution of ions
shall be analyzed, but only after presenting a phenomenological treatment
of non-steady-state diffusion.

4.2.7. The Gross View of Non-Steady-State Diffusion

What has been done so far is to consider steady-state diffusion in which

neither the flux nor the concentration of diffusing particles in various regions
changes with time. In other words, the whole transport process is time
independent. But, what happens if a concentration gradient is suddenly
produced in an electrolyte initially in a time-invariant equilibrium condition?
Diffusion starts of course, but it will not immediately reach a steady state
which does not change with time. For example, the distance variation
of concentration, which is zero at equilibrium, will not instantaneously
hit the final steady-state pattern. How does the concentration vary with
time?
Consider a parallelepiped (Fig. 4. I 5) of unit area and length dx. Ions
are diffusing in through the left face of the parallelepiped and out through
the right face. Let the concentration of the diffusing ions be a continuous
function of x. If c is the concentration of ions at the left face, the concen-
308 CHAPTER 4

~.~----dx------~~

UNIT AREA
FACE L
ACE R
I
INWARD FLUX OF I
I
IONS. J.L =-D( de ) I
dK I
}----------
/
IONIC /
CONCENTRATION,C " - - - - -_ _ _ _ _Y IONIC CONCENTRATION
C·t{~)dK
dx
Fig. 4.15. The parallelepiped of electrolyte used in the derivation of Fick's
second law.

tration at the right face is

de
e+ dx dx

Fick's law [Eq. (4.16)] is used to express the flux into and out of the
parallelepiped. Thus the flux into the left face, JL , is

J
L
=-D~
dx
(4.29)

and the flux out of the right face is

(4.30)

The net outflow of material from the parallelepiped of volume dx is

d 2e
JL - JR = D dx 2 dx (4.31)

Hence, the net outflow of ions per unit volume per unit time is D (d 2e/dx 2 )
But this net outflow of ions per unit volume per unit time from the paral-
lepiped is in fact the sought-for variation of concentration with time, i.e.,
de/dt. One obtains partial differentials because the concentration depends
both on time and distance, but the subscripts x and t are generally omitted
because it is, for example, obvious that the time variation is in a fixed
region of space, i.e., constant x. Hence,

(4.32)
 ION TRANSPORT IN SOLUTIONS 309

This partial differential equation is known as Fick's second law. It is the

basis for the treatment of most time-dependent diffusion problems in
electrochemistry.
That Fick's second law is in the form of a differential equation implies
that it describes what is common to all diffusion problems and it has
"squeezed out" what is characteristic to any particular diffusion problem. t
Thus, one always has to calculate the precise functional relationship

c = I(x, t)
for a particular situation.
The process of calculating the functional relationship consists in solving
the partial differential equation, which is Fick's second law, i.e., Eq. (4.32).

4.2.8. An Often Used Device for Solving Electrochemical Dif-

fusion Problems: The Laplace Transformation
Partial differential equations, such as Fick's second law (in which the
concentration is a function of both time and space) are generally more
difficult to solve than total differential equations, where the dependent
variable is a function of only one independent variable. An example of a
total differential equation (of second order1:) is the Ii nearized Poisson-
Boltzmann equation

It has been shown (Section 3.3.7) that the solution of this equation (with "p
dependent on r only) was easily accomplished.
One may conclude, therefore, that the solution of Fick's second law
(a partial differential equation) would proceed smoothly if some mathem-
atical device could be utilized to convert it into the form of a total differential
equation. The Laplace transformation method is often used as such a device.
Since the method is based on the operation§ of Laplace transformation,
a brief digression' on the nature of this operation is given before using it
to solve the partial differential equation involved in non-steady-state elec-
trochemical diffusion problems, namely, Fick's second law.

t This point is dealt with at greater length in Section 4.2.9.

1: The order of a differential equation is the order of its highest derivatives, which in the
example quoted is a second-order derivative, d 2tp/dr 2•
§ A mathematical operation is a rule for converting one function into another.
• Many excellent treatments of the Laplace transformation technique are available, e.g.,
J. C. Jaeger, Introduction to Applied Mathematics, John Wiley & Sons, Inc., New York,
1952.
310 CHAPTER 4

Consider a function y of the variable z, i.e., y = J(z), represented by

the plot of y against z. The familiar operation of differentiation performed
on the function y consists in finding the slope of the curve representing
y = f(z) for various values of z, i.e., the differentiation operation consists
in evaluating dyjdz. The operation of integration consists in finding the
area under the curve, i.e., it consists in evaluating f;~J(z) dz.
The operation of Laplace transformation performed on the function
y = J(z) consists of two steps:
1. Multiplying y = J(z) by e-Pz , where p is a positive quantity which
is independent of z
2. Integrating the resulting product ye-'PZ [= J(z)e- PZ ] with respect
to z between the limits z = 0 and z = 00
In short, the Laplace transform y = J(z) is

or f~ e-PZj(z) dz

Just as one often symbolizes the result of the differentiation of y by y',

the result of the operation of Laplace transformation performed on y is
often represented by a symbol y. Thus,

y = Laplace transform of y = I: e-PZy dz (4.33)

y =Sin Z

~---------------z

~----~~-------z

!O
HADED AREA. le-PZ Sin z dz
D

~~~~~------z

Fig. 4.16. Steps in the operation of the Laplace

transformation of (a) the function of y = sin z,
showing (b) e-Pz and (c) the product e- Pz sin z
integrated between the limits 0 and 00.
 ION TRANSPORT IN SOLUTIONS 311

TABLE 4.5
Laplace Transforms

Function Transform

!(t) j = f:' e-PIJ(t) dt

1
p

}, (a constant)
p

Vi

p-w

1
- sin wt
w p2 + w2
1
v;t exp
(k2
-41 )

2 10 exp(-~) _ kerfc (_k )

V---;; 4t 20
p
coswt

What happens during Laplace transformation can be easily visualized

by choosing a function, say, y = sin z, and representing the operation in a
figure (Fig. 4.16). It can be seen t that the operation consists in finding the
area under the curve, ye- PZ , between the limits Z = 0 and Z = =.
The Laplace transforms of some functions encountered in diffusion
problems are collected in Table 4.5.

t From Fig. 4.16, it can also be seen that, apart from having to make the integral converge,
the exact value of p is not of significance for, in any case, p disappears after the operation
of inverse transformation (el Section 4.2.11).
312 CHAPTER 4

4.2.9. Laplace Transformation Converts the Partial Differential

Equation Which Is Fick's Second Law into a Total Dif-
ferential Equation

It will now be shown that, by using the operation of Laplace trans-

formation, Fick's second law-a partial differential equation-is converted
into a total differential equation which can be readily solved.
Since whatever operation is carried out on the left-hand side of an
equation must be repeated on the right-hand side, both sides of Fick's
second law will be subject to the operation of Laplace transformation
(ef Eq. (4.33)]
ae = foo e-ptD--dt
foOO e-pt-dt ae 2
(4.34)
at ax2 0

which, by using the symbol for a Laplace-transformed function, can be

written

(4.35)

To proceed further, one must evaluate the integrals of Eq. (4.34).

Consider the Laplace transform

ae=
- fOO e - p t - ae dt (4.36)
at 0 at

The integral can be evaluated by the rule for integration by parts as follows

f~ e- pt :~ dt = e- pt f~ ae - f~ [J~ ae] de-pt

u dv u v v du

(4.37)

Since J: e-pte dt is in fact the Laplace transform of e [ef the defining

equation (4.33)] and, for conciseness, is represented by the symbol c and
since e- pt is zero when t ~ 00 and unity when t = 0, Eq. (4.36) reduces to

-
ae
at
=
fOO e- pt -ae dt
0 at
= - c[t = 0] + pc (4.38)

where e [t = 0] is the value of the concentration c at t = 0.

 ION TRANSPORT IN SOLUTIONS 313

Next, one must evaluate the integral on the right-hand side of Eq.
(4.34), i.e.,
f) 2c
D f)x 2 =
foo e-ptD
0
f) 2c
f)x 2 dt (4.39)

Since the integration is with respect to the variable t and the differentiation
is with respect to x, their order can be interchanged. Further, one can move
the constant D outside the integral sign. Hence, one can write

(4.40)

Once again, it is clear from Eq. (4.33) that f: e-ptc dt is the Laplace trans-
form of c, i.e., c and, therefore,

(4.41 )

From Eqs. (4.32), (4.38), and (4.41), it follows that, after Laplace
transformation, Fick's second law takes the form

d 2c
pc - c[t = 0] = D dx 2 (4.42)

This, however, is a total differential equation because it contains only the

variable x. Thus, by using the operation of Laplace transformation, Fick's
second law has been converted into a more easily solvable total differential
equation involving c, the Laplace transform of the concentration.

4.2.10. The Initial and Boundary Conditions for the Diffusion

Process Stimulated by a Constant Current (or Flux)

A differential equation can be arrived at by differentiating an original

equation, or primitive, as it is called. In the case of Fick's second law,
the primitive is the equation which gives the precise nature of the functional
dependence of concentration on space and time; i.e., the primitive is an
elaboration on
c = f(x, t)

Since, in the process of differentiation, constants are eliminated and

since three differentiations (two with respect to x and one with respect to
314 CHAPTER 4

time) are necessary to arrive at Fick's second law, three constants have been
eliminated in the process of going from the precise concentration dependence
which characterizes a particular problem to the general relation between
the time-and space-derivatives of concentration which describes any
nonstationary diffusion situation.
The three characteristics, or conditions, as they are called, of a particular
diffusion process cannot be rediscovered by mathematical argument applied
to the differential equation. To get at the three conditions, one has to resort
to a physical understanding of the diffusion process. Only then can one
proceed with the solution of the (now) total differential equation (4.42)
and get the precise functional relationship between concentration, distance,
and time.
Instead of attempting a general discussion of the three conditions
characterizing a particular diffusion problem, it is best to treat a typical
electrochemical diffusion problem.
Consider that, in an electrochemical system, a constant current is
switched on at a time which is arbitrarily designated t = 0 (Fig. 4.17).
The current is due to charge-transfer reactions at the electrode-solution
interfaces, and these reactions consume a species. Since the concentration
of this species at the interface falls below the bulk concentration, a con-
centration gradient for the species is set up and it diffuses toward the

SWITCH CLOSED AT t. 0

,..,.-
ELECTROOE

SINK SET UPAT T~ ELECTROLYTE

ELECTROOE- ELECTROLYTE
INTERFACE TO SUCK OUT THE IONIC
SPECIES J.

Fig. 4.17. Schematic representation of an electrochemical

system connected to a constant current supply which is
switched on at t = O. The current promotes a charge-transfer
reaction at the electrode-electrolyte interfaces, which results
in the diffusion flux of the species i toward the interface.
 ION TRANSPORT IN SOLUTIONS 315

interface. Thus, the externally controlled current sets up t within the solution
a diffusion flux.
The diffusion is described by Fick's second law

(4.32)

or, after Laplace transformation, by

d 2c
pc - c[t = 0] = D dx 2 (4.42)

To analyze the diffusion problem, one must solve the differential

equation, i.e., describe how the concentration of the diffusing species varies
with distance x from the electrode and with the time which has elapsed
since the constant current is switched on. But, first, one must think out the
three characteristics, or conditions, of the diffusion process described
above.
The nature of one of the conditions becomes clear from the term
c [t = 0] in the Laplace-transformed version [cf Eq. (4.42)] of Fick's
second law. The term c [t = 0] refers to the concentration before the
commencement of diffusion, i.e., it describes the initial condition of the
electrolytic solution in which diffusion is made to occur by the passage
of a constant current. Since, before the constant current is switched on
and diffusion starts, one has an unperturbed system, the concentration c
of the species which subsequently diffuses must be the same throughout
the system and equal to the bulk concentration CO. Thus, the initial condition
of the electrolytic solution is
c [t = 0] = CO (4.43)

The other two conditions pertain to the situation after the diffusion
begins, e.g., after the diffusion-causing current is switched on. Since these
two conditions often pertain to what is happening to the boundaries of the
system (in which diffusion is occurring), they are usually known as boundary
conditions.

t When the externally imposed current sets up charge-transfer reactions which provoke
the diffusion of ions, there is a very simple relation between the current density and
the diffusion flux. The diffusion flux is a mole flux (number of moles crossing I cm 2
in I sec), and the current density is a charge flux (cf Table 4.1.) Hence, the current
density i, or charge flux, is equal to the charge zF per mole of ions (z is the valence
of the diffusing ion; and F, the Faraday) times the diffusion flux J, i.e., j = zFJ.
316 CHAPTER 4

The first boundary condition is the expression of an obvious point,

namely, that, very far from the boundary at which the diffusion source or
sink is set up, the concentration of the diffusing species is unperturbed
and remains the same as in the initial condition

c [x ~ 00] = C [t = 0] = CO (4.44)

Thus, the concentration of the diffusing species has the same value cO
at any x at t = 0 or for any t > 0 at x ~ 00. This is true for almost all
electrochemical diffusion problems in which one switches on (at t = 0)
the appropriate current or potential difference across the interface and
thus sets up interfacial charge-transfer reactions which, by consuming or
producing a species, provoke a diffusion flux of that species.
What, however, is characteristic of one particular electrochemical dif-
fusion process and distinguishes it from all others is the nature of the
diffusion flux which is started off at t = O. Thus, the essential characteristic
of the diffusion problem under discussion is the switching-on of the constant
current, which means that the diffusing species is consumed at a constant
rate at the interface and the species diffuses across the interface at a constant
rate. In other words, the flux of the diffusing species at the x = 0 boundary
of the solution is a constant.
It is convenient from many points of view to assume that the constant
value of the flux is unity, i.e., 1 mole of the diffusing species crossing 1 cm2
of the electrode-solution interface per second. This unit flux corresponds
to a constant current density of 1 amp cm-2 • This normalization of the flux
scarcely affects the generality of the treatment because it will later be seen

FLUX

CONSTANT UNIT FLUX ;r

t .. o TIME

Fig. 4.18. When a constant unit flux of 1 mole

cm- 2 sec- 1 is switched on at t = 0, the variation
of flux with time resembles a step. (It is only an
ideal switch which makes the current and, there-
fore, the flux instantaneously rise from zero to
its constant value; this problem of technique is
ignored in the diagram).
 ION TRANSPORT IN SOLUTIONS 317

that the concentration response to an arbitrary flux can easily be obtained

from the concentration response to a unit flux.

°
If one looks at the time variation of current or the flux across the
solution boundary, it is seen that, for t < 0, J = and, for t > 0, there
is a constant flux J = 1 (Fig. 4.18) corresponding to the constant current
switched on at t = 0. In other words, the time variation of the flux is like
a step; that is why the flux produced in this setup is often known as a
step function (of time).
At any instant of time, the constant flux across the boundary is related
to the concentration gradient there through Fick's first law, i.e.,

Jx=O = 1= - D( ~c )
uX x~O
(4.45)

The above initial and boundary conditions can be summarized thus

crt = 0] = CO (4.43)

c[X -+ 00] = CO (4.44)

(4.45)

The three conditions just listed describe the special features of the constant
(unit) flux diffusion problem. They will now be used to solve Fick's second
law.

4.2.11. The Concentration Responseto a Constant Flux Switched

on at t = 0
It has been shown (Section 4.2.9) that, after Laplace transformation,
Fick's second law takes the form
d2c
pc - crt = 0] = D dX2 (4.42)

The solution of an equation of this type is facilitated if the second term

is zero. This objective can be attained by introducing a new variable C1
defined thus
(4.46)

The variable C1 can be recognized as the departure CO - c of the concen-

tration from its initial value co. In other words, C1 represents the perturbation
from the initial concentration (Fig. 4.19).
The partial differential equation [Eq. (4.32)] and the initial and bound-
318 CHAPTER 4

ACTUAL
CONCENTRATION
e(x)

(a)

~------------~-x

PERTURBATION
CI • c"-C(XI

( b)
~----~-------L~·X

Fig. 4.19. Schematic representation of (a) the

variation of concentration with distance x from
the electrode at t = 0 and t = t and (b) the var-
iation of the perturbation Cl = CO - c'" in con-
centrations.

ary conditions have now to be restated in terms of the new variable Cl'
This is easily done by using Eq. (4.46) in Eqs. (4.32), (4.43), (4.44), and
(4.45). One obtains
(4.47)

(4.48)

(4.49)

(4.50)

After Laplace transformation of Eq. (4.47), the differential equation

becomes
(4.51)

Since, however, Cl [I = 0] = 0 [cf Eq. (4.48)], it is clear that

(4.52)

This equation is identical in form to the linearized P-B equation [cf Eq.
 ION TRANSPORT IN SOLUTIONS 319

(3.21)] and, therefore, must have the same general solution, i.e.,

(4.53)

where A and B are the arbitrary integration constants to be evaluated by

the use of the boundary conditions. If the Laplace transformation method
had not been used, the solution of (4.47) would not have been so simple.
The constant B must be zero by virtue of the following argument.
From the boundary condition CI [x ---+ 00] = 0, i.e., Eq. (4.49), it is clear
that, after Laplace transformation,

CI [x ---+ 00] =0 (4.54)

Hence, as x ---+ 00, ci ---+ 0, but this will be true only if B = 0 as, otherwise,
ciin Eq. (4.53) will go to infinity instead of zero.
One is therefore left with

(4.55)

Differentiating this equation with respect to x, one obtains

(4.56)

which, at x = 0, leads to

( dCl lIP A
VD (4.57)
) __
dx x~o-

Another expression for (dcljdx)x=o can be obtained by applying the

operation of Laplace transformation to the constant flux boundary con-
dition (4.50). Laplace transformation on the left-hand side of the boundary
condition leads to (dcljdx)x=o; and the same operation performed on the
right-hand side, to -ljDp (cf Appendix 4.2). Thus, from the boundary
condition (4.50), one gets

( dCI ) 1 (4.58 )
dx x~o =- Dp

Hence, from Eqs. (4.57) and (4.58), it is found that

I
A = p!m (4.59)
320 CHAPTER 4

Taking logarithm of numbers

direct oddition of
multiplication logarithm

taking antilogarithms

(0 )

Laplace transformation

I
classical algebraic
solution initial and manipulation
boundory
conditions
I
I
Iexpression for ell Inverse transformation solution of total
differential
equation to get Co

( b)

Fig. 4.20. Comparison of the use of (a) logarithms and (b)

Laplace transformation.

Upon inserting this expression for A into Eq. (4.55), it follows that
_ I 1
C = -- e-(p!D)'lX (4.60)
1 D!p!

The ultimate aim, however, is not to get an expression for the c1 , the
Laplace transform of c1 , but to get an expression for C1 (or c) as a function
of distance x and time t. The expression c1 has been obtained by a Laplace
transformation of c1 , hence, to go from c1 to c1 , one must do an inverse
transformation. The situation is analogous to using logarithms to facilitate
the working-out of a problem (Fig. 4.20). In order to get C 1 from c1 , one
asks the question: What function C1 would, under Laplace transformation,
give the Laplace transform c1 of Eq. (4.60)? In other words, one has to
find C1 in the equation

(4.61 )
 ION TRANSPORT IN SOLUTIONS 321

't:
Q)

O~-----------+------------~2

Fig. 4.21. Variation of the error function erf (y)

with argument y.

A mathematician would find the function c1 , i.e., do the inverse trans-

formation, by making use of the theory of functions of variables which
are complex. Since, however, there are extensive tables of functions y and
their transforms y, it is only necessary to look up the tables in the column
of Laplace transforms (Table 4.5). It is seen that, corresponding to the
transform,

is the function
2n-lt! exp ( - -X2)
4Dt
2
- - xD-! erfc (-x-
4Dt
)!
where erfc is the error function complement defined thus

erfc (y) = 1 - erf (y) (4.62)

erf (y) being the error function given by (cf Fig. 4.21)

2 IY c
------=
erf(y)
vn du (4.63)
= U2

Hence, the expression for the concentration perturbation C1 in Eq. (4.61)

must be

C1 =
1 [2tl
Dl ---;r exp ( - 4Dt x2
X2) - xD-! erfc ( 4Dt )!] (4.64)

If one is interested in the true concentration c, rather than the deviation C1 ,

322 CHAPTER 4

"Z
0
;::
'"
II:
I-
Z
AT t=O
(initial condition)
(0) C·
"'"
Z
0
"
DISTANCE, X

~
z
0
5i
II:
AT t = tl

(b)
I-
z c·
"'"0z
"
DISTANCE, X

"
z
0
;::
'"
II:
I- AT t= te
(c) Z C·
"'"z
0
"
DISTANCE, x

Fig. 4.22. Graphical representation of the

variation of the concentration c with dis-
tance x from the electrode or diffusion sink.
(a) The initial condition at t = 0; (b) and
(c) the conditions at t1 and t., where t. >
t1 > t = 0. Note that, at t > 0, (dc/dx)",~o
is a constant, as it should be in the con-
stant flux-diffusion problem.

in the concentration from the initial value co, one must use the defining
equation for C1
(4.46)
The result is
2ti
[ -;J ( X2) X ( x2
exp - 4Dt - Di erfe 4Dt
)!] (4.65)

This, then, is the fundamental equation showing how the concen-

tration of the diffusing species varies with distance x from the eleetrode-
solution interface and with the time t which has elapsed since a constant
 ION TRANSPORT IN SOLUTIONS 323

unit flux was switched on. In other words, Eq. (4.65) describes the diffusional
response of an electrolytic solution to the stimulus of a flux which is in the
form of a step function of time. The nature of the response is best appreciated
by seeing how the concentration profile of the species diffusing varies as a
function of time [Fig. 4.22(a), (b), and (c)]. Equation (4.65) is also of
fundamental importance in describing the response of an electrochemical
system to a current density which varies as a step function, i.e., to a constant
current density switched on at t = O. It will be seen later (Chapter 8) that
this characteristic concentration response is the basis of an important
electro analytical technique.

4.2.12. How the Solution of the Constant-Flux Diffusion Prob-

lem Leads On to the Solution of Other Problems

The space and time variation of the concentration in response to the

switching-on of a constant flux has been analyzed. Suppose, however, that,
instead of a constant flux, one switches on a sinusoidally varying flux. t
What is the resultant space and time variation of the concentration of the
diffusing species?
One approach to this question is to set up the new diffusion problem
with the initial and boundary conditions characteristic of the sinusoidally
varying flux and to obtain a solution. There is, however, a simpler approach.
Using the property of Laplace transforms, one can use the solution (4.65)
of the constant flux-diffusion problem to generate solutions for other
problems.
An electronic device is connected to an electrochemical system so that
it switches on a current (Fig. 4.23) which is made to vary with time in a
controllable way. The current provokes charge-transfer reactions which
lead to a diffusion flux of the species involved in the charge-transfer reac-
tions. This diffusion flux varies with time in the same way as the current.
The time variation of the flux will be represented thus

J = get) (4.66)

The imposition of this time-varying flux stimulates the electrolytic

solution to respond with a space and time variation of the concentration c
or the perturbation in concentration, C1 . The response depends on the

t If one feels that current is a more familiar word than flux, one can substitute the word
current because these diffusion fluxes are often, but not always, provoked by control-
ling the current across an electrode-solution interface.
324 CHAPTER 4

CONTROLLED TIME VARIATION

OF CURRENT PRODUCED BY

ELECTROLYTE

Fig. 4.23. Use of an electronic device to vary

the current passing through an electrochemical
system in a controlled way.

stimulus, and the mathematical relationship between the cause J(t) and the
effect C1 can be represented (Fig. 4.24) quite generally thus t

(4.67)

where i\ and J are the Laplace transforms of the perturbation in concen-

tration and the flux, and y is the to-be-determined function which links
the cause and effect and is characteristic of the system.
The relationship (4.67) has been defined for a flux which has an arbitrary
variation with time; hence, it must also be true for the constant unit flux
described in Section 4.2.10. The Laplace transform of this constant unit
flux J = I is lip according to Appendix 4.2; and the Laplace transform
of the concentration response to the constant unit flux is given by Eq.
(4.60), i.e.,

(4.60)

Hence, by substituting lip for J and D-~p-ie-(PID>~X for c1 in Eq. (4.67),

t It will be seen further on that one uses a relationship between the Laplace transforms
of the concentration perturbation and the flux rather than the quantities c, and ],
themselves, because the treatment in Laplace transforms is not only elegant but fruitful.
 ION TRANSPORT IN SOLUTIONS 325

STIMULUS RESPONSE
SYSTEM
(CAUSE) • (EFFECT)
Y
J
·1 C.

Fig. 4.24. Schematic representation of the re-

sponse c1 of a system Y to a stimulus J

one has
(4.68)

On introducing this expression for y into the general relationship (4.67),

the result is

(4.69)

This is an important result: Through the evaluation of y, it contains

the concentration response to a constant unit flux switched on at t = O.
But, in addition, it shows how to get the concentration response to a flux
l(t) which is varying in a known way. All one has to do is to take the
Laplace transform J of this flux l(t) switched on at time t = 0, substitute
this J in Eq. (4.69), and get (1' If one inverse-transforms the resulting
expression for (1' one will obtain c1 , the perturbation in concentration as
a function of x and t.
Consider a few examples. Suppose that, instead of switching on a
constant unit flux at t = 0 (cf Section 4.2.10), one imposes a flux which
is a constant but now has a magnitude of A moles cm- 2 sec-I, i.e., 1 = A.
Since the transform of a constant is lip times the constant (cf Appendix
4.2), one obtains
- A
1=- (4.70)
p

which, when introduced into Eq. (4.69), gives

(4.71 )

The inverse transform of the right-hand side of (4.71) is identical to that

for the unit step function [cf Eq. (4.60)] except that it is multiplied by A.
That is,

C1 = D!
A [2t!
--;;r exp ( - 4Dt x2
X erfc ( 4Dt
X2) - I5f )!] (4.72)
326 CHAPTER 4

In other words, the concentration response of the system to a J = A. flux

is a magnified-A.-times version of the response to a constant unit flux.
One can also understand what happens if the flux, instead of sucking
ions out of the system, acts as a source and pumps in ions. This condition
can be brought about by changing the direction of the constant current
going through the interface and thus changing the direction of the charge-
transfer reactions so that the diffusing species is produced rather than
consumed. Thus, diffusion from the interface into the solution occurs.
Because the direction of the flux vector is reversed, one has

- A.
J= - A. or J=-- (4.73)
P
and, thus,

(4.74)

or, in view of Eq. (4.46),

c= CO
A. [2t~ exp (
+ Df nr - 4Dt X2) -
X
Df ( x 2 )1]
erfc 4Dt (4.75)

Jd~staf~:
STIMULUS RESPONSE TO A SINK

u
e
.~

Electrode
: ~l.
~C
Concentrotion falls below
initial value c'
(sink) .3 ~
Distance l

(a)

J
RESPONSE TO A SOURCE
STIMULUS

u Concentration rl..s above

flux J e initial value c'
0

Electrode
distance x
- -
~l.
Ole
(source)
~l
Distance l
(b)

Fig. 4.25. The difference in the concentration response to the

stimulus of (a) a sink and (b) a source at the electrode-electro-
lyte interface.
 ION TRANSPORT IN SOLUTIONS 327

FLUX J • J'max COS wt

J (tl

TIME. t

Fig. 4.26. A sinusoidally varying flux at an electrode-

electrolyte interface, produced by passing a sinusoidally
varying flux through the electrochemical system.

Notice the plus sign; it indicates that the concentration c rises above the
initial value eO (Fig. 4.25).
Consider now a more interesting type of stimulus involving a period-
ically varying flux (Fig. 4.26). After representing the imposed flux by a
cosine function
J = Jmaxcos wt (4.76)

its Laplace transform is (ef Table 4.5)

- p
J=J max 2+ 2 (4.77)
P w

When this is combined with Eq. (4.69), one gets

-; _
(1 -
p
p2 + w2
[IDip! e_(x!Di)p~] J Illax (4.78)

To simplify matters, the response of the system shall only be considered

at the boundary, i.e., at x = O. Hence, one can set x = 0 in Eq. (4.78),
in which case,

(4.79)

The inverse transform reads

e1 [x = 0] = J max (
(Dw)! cos wI -
:n;)
4"" (4.80)

which shows that, corresponding to a periodically varying flux (or current),

328 CHAPTER 4

TIME t
CONCENTRATION
PERTURBATION C.

Fig. 4.27. When a sinusoidally varying dif-

fusion flux is produced in an electrochem-
ical system by passing a sinusoidally varying
current through it, the perturbation in the
concentration of the diffusing species also
varies sinusoidally with time, but with a
phase difference of TI/4.

the concentration perturbation also varies periodically, but there is a n/4

phase difference between the flux and the concentration response (Fig. 4.27).
This is an extremely important result because an alternating flux can
be produced by an alternating current density at the electrode-electrolyte
interface, and, in the case of fast charge-transfer reactions, the concentration
at the boundary is related to the potential difference across the interface.
Thus, the current density and the potential difference both vary periodically
with time, and it turns out that the phase relationship between them provides
information on the rate of the charge-transfer reaction.

4.2.13. Diffusion Resulting from an Instantaneous Current Pulse

There is another important diffusion problem, the solution of which

can be generated from the concentration response to a constant current
(or a flux).
Consider that, in an electrochemical system, there is a plane electrode
at the boundary of the electrolyte. Now, suppose that, with the aid of an
electronic pulse generator, an extremely short time current pulse is sent
through the system (Fig. 4.28). The current is directed so as to dissolve
the metal of the electrode; hence, the effect of the pulse is to produce a
burst of metal dissolution in which a layer of metal ions is piled up at the
interface (Fig. 4.29).
Because the concentration of metal ions at the interface is far in excess
of that in the bulk of the solution, diffusion into the solution begins. Since
the source of the diffusing ions is an ion layer parallel to the plane electrode,
 ION TRANSPORT IN SOLUTIONS 329

ELECTRODE
WHICH DISSOLVES
DURING PULSE

Fig. 4.28. The use of an electronic pulse gen-

erator to send an extremely short time current
pulse through an electrochemical system so that
there is dissolution at one electrode during the pulse.

it is known as a plane source; and, since the diffusing ions an~ produced in
an instantaneous pulse, a fuller description of the source is contained in the
term instantaneous plane source.
Now, as the ions from the instantaneous plane source diffuse into the
solution, their concentration at various distances will change with time. The
problem is to calculate the distance and time variation of this concentration.
The starting point for this calculation is the general relation between
the Laplace transforms of the concentration perturbation C1 and the time-
varying flux J(t)

(4.69)

One has to substitute for J the Laplace transform of a flux which is an

instantaneous pulse. This is done with the help of the following interesting
observation.
If one takes any quantity which varies with time as a step, then the
differential of that quantity with respect to time varies with time as an
instantaneous pulse (Fig. 4.30). In other words, the time derivative of a
step function is an instantaneous pulse. Suppose, therefore, one considers
330 CHAPTER 4

ELECTRODE

ELECTRODE
WHICH DISSOLVES
IN PULSE

IONS PRESENT IN
SOLUTION
LAYER OF IONS
PRIOR TO PULSE
PRODUCED IN
PULSE OF ELECTRODE
DISSOLUTION

Fig. 4.29. The burst of electrode dissolution during

the current pulse produces a layer of ions adjacent
to the dissolving electrode (negative ions are not
shown in the diagram).

a constant flux (or current) switched on at t = 0 (i.e., the flux is a step

function of time and will be designated J step ), then the time derivative of
that constant flux is a pulse of flux (or current) at t = 0 referred to by
the symbol JpUlSC, i.e.,
d
J pulse = dt Jstep (4.81)

If, now, one takes Laplace transforms of both sides and uses Eq. (4.38)
to evaluate the right-hand side, one has

~1lI1sc = pfstcp - J step [t = 0] (4.82)

FLUX STEP-FUNCTION

L---..L--~"'---__ TIME
taO

FLUX

INSTANTANEOUS
/' PULSE
L-_ _..J::::...-_ _ _ _.... TIME
t=O ------
Fig. 4.30. The time derivative of a flux,
which is a step function of time, is an in-
stantaneous pulse of flux.
 ION TRANSPORT IN SOLUTIONS 331

But, at t = 0, the magnitude of a flux which is a step function of time is

zero. Hence,
(4.83)

If the pumping of diffusing particles into the system by the step-function

flux consists in switching on, at t = 0, a flux of - A moles cm- 2 sec-I, then

J step = - A (4.84)
and
-
J step = - pA (4.85)

Using this relation in Eq. (4.83), one has

Jpulse = - A (4.86)

and, by substitution in Eq. (4.69),

c1 = - AD-}p-i exp ( - ~x) (4.87)

By inverse transformation (ef Table 4.5),

e1 = - A(nDt)-~ exp( - 4~t) (4.88)

or by referring to the actual concentration e instead of the perturbation e1

in concentration, the result is

e = CO + A(nDt)-~ exp( - ~)
4Dt
(4.89)

If, prior to the current pulse, there is a zero concentration of the

species which is produced by metal dissolution, i.e.,

e [t = 0] = CO = ° (4.90)
then Eq. (4.89) reduces to

e = __ A_ e-X'/4Dt (4.91)
(nDt)t

It can be seen from Eq. (4.86) that A is the Laplace transform of the pulse
of flux. But a Laplace transform is an integral with respect to time. Hence,
A, which is a flux (of moles per square centimeter per second) in the constant-
flux problem (ef Section 4.2.12), is in fact the total concentration (moles
33:< CHAPTER 4

.
~
2
...
o
z
2
~
<>
«
...
0:

DISTANCE FROM ELECTRODE, X

Fig. 4.31. Plots of the fraction n/ntotal of

ions (produced in the pulse of electrode
dissolution) against the distance x from

°
the electrode. At t = 0, all the ions are on
the x = plane, and, at t > 0, they are
distributed in the solution as a result of
dissolution and diffusion. In the diagram,
t3 > t2 > tl > 0, and the distribution curve
becomes flatter and flatter.
per square centimeter) of the diffusing ions produced on the x = 0 plane
in the burst of metal dissolution. If, instead of dealing with concentrations,
one deals with numbers of ions, the result is

ntotal e-x'/4Dt
n= (4.92)
(nDt)~

where n is the number of ions at a distance x and a time t, and ntotal is the
number of ions set up on the x = 0 plane at t = 0, i.e., ntotal is the total
number of diffusing ions.
This is the solution to the instantaneous-plane source problem. When
n/ntotal is plotted against x for various times, one obtains curves (Fig. 4.31)
which show how the ions injected into the x = 0 plane at t = 0 (e.g., ions
produced at the elet::trode in an impulse of metal dissolution) are distributed
in space at various times. At any particular time t, a semi-bell-shaped
distribution curve is obtained which shows that the ions are mainly clustered
near the x = 0 plane, but there is a "spread." With increasing time, the
spread of ions increases. This is the result of diffusion, and, after an infinitely
long time, there are equal numbers of ions at any distance.

4.2.14. What Fraction of Ions Travels the Mean Square Distance

<x') in the Einstein-Smoluchowski Equation?
In the previous section, an experiment was described in which a pulse
of current dissolves out of the electrode a certain number ntotal of ions,
 ION TRANSPORT IN SOLUTIONS 333

RADIATION
CONTAINER WALL

:rAGGED IONS

RADIOACTIVE
ELECTRODE
\ ELECTRODE

WHICH DISSOLVES WINDOW

IN CURRENT
PULSE

GEIGER COUNTER
SYSTEM

Fig. 4.32. Schematic of experiment to note the time interval

between the pulse of electrode dissolution (at t = 0) and the
arrival of radioactive ions at the window where they are regis-
tered in the Geiger-counter system.

which then start diffusing into the solution. Now suppose that the electrode
material is made radioactive so that the ions produced by dissolution are
detectable by a counter (Fig. 4.32). The counter head is then placed near
a window in the cell at a distance of 1 cm from the dissolving electrode,
so that, as soon as the tagged ions pass the window, they are registered by
the counter. How long after the current pulse at t = 0 does the counter
note the arrival of the ions?
It is experimentally observed that the counter begins to register within
a few seconds of the termination of the instantaneous current pulse. Suppose,
however, that one attempted a theoretical calculation based on the Einstein-
Smoluchowski equation (4.27), i.e.,

(4.27)

using, for the diffusion coefficient of ions, the experimental value of 10-5
cm2 sec-I. Then, the estimate for the radioactive ions to reach the counter
is
t = mm
2 X 10-5 X 60
or
t,....., 103 min

This is several orders of magnitude larger than is indicated by experience.

<x
The dilemma may be resolved as follows. If 2 ) in the Einstein-
Smoluchowski relation pertains to the mean square distance traversed by
a majority of the radioactive particles and if Geiger counters can-as is
334 CHAPTER 4

Ci
ell
UI
z
Q
II.
o
Z _X2
o
....
t)
_1- e 401
c( 2 (lfDI)i12
II:
II.

t DISTANCE X

X-X rms- ~
Fig. 4.33. The Einstein-Smoluchowski equation,
<x 2 ) = 20t, pertains to the fraction (n/ntotal)ES,
of ions between x = 0 and x = x rms = <X2). This
fraction is found by integrating the distribution
curve between x = 0 and x = x rms '

the case-detect a very small number of particles, then one can qualitatively
see that there is no contradiction between the observed time and that
estimated from Eq. (4.27). The time of 103 min estimated by the Einstein-
Smoluchowski equation is far too large because it pertains to a number
of radioactive ions far greater than the number needed to register in the
counting apparatus. The way in which the diffusing particles spread out
with time, i.e., the distribution curve for the diffusing species (Fig. 4.31)
shows that, even after very short times, some particles have diffused to very
large distances, and these are the particles registered by the counter in a
time far less than that predicted by the Einstein-Smoluchowski equation.
The qualitative argument just presented can now be quantified. The
central question is: To what fraction n/ntotal of the ions (released at the
instantaneous-plane source) does <X2>
= 2Dt apply?t This question can
be answered easily by integrating the n versus x distribution curve (Fig.
4.33) between the lower limit x = 0 (the location of the plane source)
x
and the value of corresponding to the square root of This upper <X2>.
limit of V <x2>-the root-mean-square distance-is, for conciseness, re-
presented by the symbol X rms , i.e.,

(4.93)

Thus, the Einstein-Smoluchowski fraction (n/ntotal)ES is given by [cf Eq.

(4.92)]

t This fraction will be termed the Einstein-Smoluchowski fraction.

 ION TRANSPORT IN SOLUTIONS 335

(_'_1 _)
ntotal E5
= fIrms __1_ e-X'J4Dt
0 (nDt)~
dx

= V2 Jrrms e-x2J2(2DIl dx (4.94)

nj(2Dt)~ 0

According to the Einstein-Smoluchowski relation,

hence,
(2Dt)k = V (X2) = X rms (4.95)

By using this relation, Eq. (4.94) becomes

( -11-) -_ -V2
-- frrms e_l(xJx cms
2
)2 dX (4.96)
ntot~l ES nkXrms 0

To facilitate the integration, substitute

x • /-
--=V 2u (4.97)
X rms

in which case, several relations follow

( ~)2
X rms
= 2u 2 (4.98)

dx = V2 X rms du (4.99)

1
u=-- when x = X rms (4.100)
V2
and
u=O when X=o (4.101)

With the use of these relations, Eq. (4.96) becomes

(-n) 2 1/ v:!

n~ f
2
- =- e- U du (4.102)
Iltotal 1'5 0

The integral on the right-hand side is the error function of 1/V2 [ef
Eg. (4.63)].
336 CHAPTER 4

TABLE 4.6

The Value of the Integral ~ fY e- u2 du

'\ilT 0

y 0

0.00 0.00000
0,01 0.01128
0.02 0.02256
0.10 0.11246
0.20 0.22270
0.30 0.32863
0.40 0.42839
0.50 0.52050
0.60 0.60386
0.70 0.67780
0.80 0.74210
0.90 0.79691
1.00 0.84270
2.00 1.00000

Now, values of the error function have been tabulated in detail (Table
4.6). The value of the error function of ltv 2, i.e.,
-2 fllV2 e- u 2 du
nt 0

is 0.68. Hence,

( -n- ) =0.68 (4.103)

ntotai ES

and, therefore, about two-thirds (68%) of the diffusing species are within
the region from x = 0 to x = X rms = v'
2Dt. This means however that
the remaining fraction, namely, one-third, have crossed beyond this dis-
tance. Of course, the radioactive ions which are sensed by the counter almost
immediately after the pulse of metal dissolution belong to the one-third
group (Fig. 4.34).
In the above experiment, diffusion toward the x -+ - 0 0 direction is
prevented by the presence of a physical boundary (i.e., the electrode).
If no such boundary exists and diffusion in both the +x and - x directions
is possible, then the 68% of the particles will distribute themselves in the
region from x = -Xrms = -v' 2Dt to x = +x rms = +V2Dt. From
 ION TRANSPORT IN SOLUTIONS 337

THIS SHADED AREA

CONTAINS 68 'lit OF
THE IONS

......."+-.....;::-------<- X
DISTANCE
Xrmi.qor
Fig. 4.34. When diffusion occurs from an
instantaneous plane source (set up, e.g., by
a pulse of electrode dissolution), then 68% of
the ions produced in the pulse lie between
x = 0 and x = xrms ' after the time t.

symmetry considerations, one would expect 34% to be within x = 0 and

x = +xrms and an equal amount on the other side (Fig. 4.35).
From the above discussion, the advantages and limitations of using
the Einstein-Smoluchowski relation become clear. If one is considering
phenomena involving a few particles, then one can be misled by making
Einstein-Smoluchowski calculations. If, however, one wants to know about
the diffusion of a sizable fraction of the total number of particles, then the
relation provides easily obtained, although rough, answers without having
to go through the labor of obtaining the exact solution for the diffusion
problem (see, e.g., Section 4.6.8).

n
"rOTAL

N t OF DIFFUSING IONS "'1- OF DIFFUSING IONS

-x +x

Fig. 4.35. If it were possible for diffusion to occur

in the + x and - x directions from an instantaneous
plane source at x = 0, then one-third of the diffusing
species would lie between x = 0 and x = +xrmsanda
similar number between x = 0 and x = - x rms .
338 CHAPTER 4

4.2.15. How Can the Diffusion Coefficient Be Related to Mol-

eGular Quantities 7
The diffusion coefficient D has appeared in both the macroscopic
(Section 4.2.2) and the atomistic (Section 4.2.6) views of diffusion. But,
how does the diffusion coefficient depend on the structure of the medium
and the interatomic forces which operate? To answer this question, one
should have a deeper understanding of this coefficient than that provided
by the empirical first law of Fick, in which D appeared simply as the pro-
portionality constant relating the flux J and the concentration gradient
de/dx. Even the random-walk interpretation of the diffusion coefficient as
embodied in the Einstein-Smoluchowski equation (4.27) is not fundamental
enough because it is based on the mean square distance traversed by the
ion after N steps taken in a time t and does not probe into the laws govern-
ing each step taken by the random-walking ion.
This search for the atomistic basis of the diffusion coefficient will
commence from the picture of random-walking ions (ef Sections 4.2.4
to 4.2.6). It will be recalled that a net diffusive transport of ions occurs,
in spite of the completely random zig-zag dance of individual ions, because
of unequal numbers of ions in different regions.
Consider one of these random-walking ions. It can be proved (see
Appendix 4.1) that the mean square distance <x2 ) traveled by an ion depends
on the number N of jumps the ion takes and the mean jump distance I in
the following manner (ef Fig. 4.36):

(4.104)

It has further been shown (Section 4.2.6) that, in the case of a one-dimen-

Fig. 4.36. Schematic representation of

N = 11 steps (each of mean length /)
in the random walk of an ion. After 11
steps, the ion is at a distance x from the
starting point.
 ION TRANSPORT IN SOLUTIONS 339

sional random walk, <X2> depends on time according to the Einstein-

Smoluchowski equation
(4.27)

Hence, by combining Eqs. (4.104) and (4.27), one has the equation

NI2 = 2Dt (4.105)

which relates the number of jumps and the time. If, now, only one jump
of the ion is considered, i.e., N = 1, Eq. (4.105) reduces to

1 12
D=-- (4.106)
2 T

where T is the mean jump time to cover the mean jump distance I. This
mean jump time is the number of seconds per jump, t and, therefore, 1fT is
the jump frequency, i.e., the number of jumps per second. Putting

k=~ (4.107)
T

one can write Eq. (4.106) thus

D = t 12k (4.108)

Equation (4.108) shows that the diffusion depends on how far, on the
average, an ion jumps and how frequently these jumps occur.

4.2.16. The Mean Jump Distance I, a Structural Question

To go further than Eq. (4.108), one has to examine the factors which
govern the mean jump distance 1 and the jump frequency k. For this, the
picture of the liquid (in which diffusion is occurring) as a structureless
continuum is inadequate. In reality, the liquid has a structure-ions and
molecules in definite arrangements at anyone instant of time. This arrange-
ment in a liquid (unlike that in a solid) is local in extent, transitory in time,
and mobile in space. The details of the structure are not necessary to con-
tinue the present discussion. What counts is that ions zig-zag in a random
walk and, for any particular jump, the ion has to jump out of one site in

t This mean jump time will include any waiting time between two successive jumps.
340 CHAPTER 4

BEFORE JUMP AFTER JUMP

SITE

'~~ . ~
FILLED

B
\
FILLED SITE EMPTY SITE EMPTY SITE

Fig. 4.37. Diagram to illustrate one step in the

random walk of an ion.

the liquid structure into another site (Fig. 4.37). This jumping process
can be symbolically represented thus

(4.109)

where B is a site occupied by an ion and D is an empty accep-

tor site waiting to receive a jumping ion.
The mean jump distance I is seen to be the mean distance between
sites, and its numerical value depends upon the details of the structure of
the liquid, i.e., upon the instantaneous and local atomic arrangement.

4.2.17. The Jump Frequency, a Rate-Process Question

The process of diffusion always occurs at a finite rate; it is a rate

process.
Now, chemical reactions, e.g., three atom reactions of the type

AB +C -4- A + BC (4.110)

are also rate processes. Further, a three-atom reaction can be formally

described as the jump of the particle B from a site in A to a site in C (Fig.
4.38). With this description, it can be seen that the notation (4.109) used

BEFORE REACTION AFTER REACTION

~@
JUMP

Fig. 4.38. A three-atom chemical rection

AB + C -+ A + BC viewed as the jump of
atom B from atom A to atom C.
 ION TRANSPORT IN SOLUTIONS 341

LJump diSlane.-----l
Pre-jump
,ite It,
I L..
JumplnO ion
I POSI-JumpSiIe
/1< . .

(a) I ,
, / , j
,-' 'I-'
I
This slandard} Co _..-< I
fres eneroy rre..... ~. 0
1Thi, posilion a/
jumpino ion

.•
liI
c

(b) !
...Ii
...c
"
W

Posilion of jumping ion

Fig. 4.39. (a) The jump of an ion from a prejump

site to a postjump site through a jump distance I.
(b) Corresponding to each position of the jumping
ion, the system (sites + jumping ion) has a standard
free energy. Thus, the standard free-energy changes
corresponding to the ionic jump can be represented
by the passage of a point (representing the standard
free energy of the system) over the barrier by the
acquisition of the standard free energy of activation.

to represent the jump of an ion has in fact established an analogy between

the two rate processes, i.e., diffusion and chemical reaction. Thus, the basic
theory of rate processes (Appendix 4.3) should be applicable to both the
processes of diffusion and chemical reactions.
The basis of this theory is that the potential energy (and standard free
energy) of the system of particles involved in the rate processes varies as
the particles move to accomplish the process. Very often, the movements
crucial to the process are those of a single particle, as is the case with the
diffusive jump of an ion from site to site. If the free energy of the system
is plotted as a function of the position of the crucial particle, e.g., the
jumping ion, then the standard free energy of the system has to attain a
critical value (Fig. 4.39)-the activation free energy L1Go -for the process
to be accomplished. One says that the system has to cross an energy barrier
for the rate process to occur. The number of times per second, k, that the
342 CHAPTER 4

rate process occurs, i.e., the jump frequency in the case of diffusion, can be
shown to be given byt (cf Appendix 4.3)

(4.111)

4.2.18. The Rate-Process Expression for the Diffusion Coef-

ficient

To obtain the diffusion coefficient in terms of atomistic quantities,

one has to insert the expression for the jump frequency (4.111) into that
for the diffusion coefficient [Eq. (4.108)]. The result is

D=~f2l(
2

(4.112)

The numerical coefficient t has entered here only because the Einstein-
<x
Smoluchowski equation 2 ) = 2Dt for one-dimensional random walk was
considered. In general, it is related to the probability of the ion's jumping
in various directions, not just forward and backward. For convenience,
therefore, the coefficient will be taken to be unity, in which case

(4.113)

(4.114)

4.2.19. Diffusion: An Overall View

An electrochemical system runs on the basis of charge-transfer reactions

at the electrode-electrolyte interfaces. These reactions involve ions or
molecules which are constituents of the electrolyte. Thus, the transport of
particles to or away from the interface becomes an essential condition for
the continued electrochemical transformation of reactants atr- the interface.
One of the basic mechanisms of ionic transport is diffusion. This type
of transport occurs when the concentration of the diffusing species is dif-
ferent in different parts of the electrolyte. What makes diffusion occur?

t The k on the left-hand side is the jump frequency; the k in the term kT/h is the Boltz-
mann constant.
 ION TRANSPORT IN SOLUTIONS 343

This question can be answered on a macroscopic level and on a microscopic

level.
The macroscopic view is based on the fact that, when the concentration
varies with distance, so does the chemical potential. But a nonuniformity
of chemical potential implies that the free energy is not the same every-
where and, therefore, the system is not at equilibrium. How can the system
attain equilibrium? By equalizing the chemical potential, i.e., by transferring
the species from the high-concentration regions to the low-concentration
regions. Thus, the negative gradient of the chemical potential, -dt-t/dx,
behaves like a driving force (el Section 4.2.1) which produces a net flow,
or flux, of the species.
When the driving force is small, it may be taken to be linearly related
to the flux. On this basis, an equation can be derived for the rate of steady-
state diffusion, which is identical in form to the empirical first law of Fick,

J=-D~
dx

The microscopic view of diffusion starts with the movements of in-

dividual ions. Ions dart about haphazardly, executing a random walk.
By an analysis of one-dimensional random walk, a simple law can be
derived (el Section 4.2.6) for the mean square distance <x2) traversed by
an ion in a time t. This is the Einstein-Smoluchowski equation

(4.27)

It also turns out that the random walk of individual ions is able to
give rise to a flux, or flow, on the level of the group. Diffusion is simply
the result of there being more random-walking particles in one region than
in another (el Section 4.2.6). The gradient of chemical potential is therefore
only a pseudo force which can be regarded as operating on a society of
ions but not on individual ions.
The first law of Fick tells one how the concentration gradient is
related to the flux under steady-state conditions; it says nothing about
how the system goes from nonequilibrium to steady state when a diffusion
source or sink is set up inside or at the boundary of the system. Thus, it says
nothing about how the concentration changes with time at different dis-
tances from the source or sink. In other words, Fick's first law is inapplicable
to non-steady-state diffusion. For this, one has to go to Fick's second law

~ = D (Pe (4.32)
at ax 2
344 CHAPTER 4

which relates the time and space variations of the concentration during
diffusion.
Fick's second law is a partial differential equation. Thus, it describes
the general characteristics of all diffusion problems but not the details of
anyone particular diffusion process. Hence, the second law must be solved
with the aid of the initial and boundary conditions which characterize the
particular problem.
The solution of Fick's second law is facilitated by the use of Laplace
transforms which convert the partial differential equation into an easily
integrable total differential equation. By utilizing Laplace transforms, the
concentration of diffusing species as a function of time and of distance
from the diffusion sink when a constant normalized current, or flux, is
switched on at t = 0 was shown to be

c = CO -
I
D! nr exp ( -
[2t!
4Dt
X2) -
X (X2
Df erfc 4Dt
)!] (4.65)

With the solution of this problem (in which the flux varies as a unit-step
function with time), one can easily generate the solution of other problems
in which the current, or flux, varies with time in other ways, e.g., as a
periodic function or as a single pulse.
When the current, or flux, is a single impulse, an instantaneous-plane
source for diffusion is set up and the concentration variation is given by

c(x t) A
= --- e- x 2 /4 Dt (4.91)
, (nDt)!

in the presence of a boundary. From this expression, it turns out that, in

a time t, only a certain fraction (1) of the particles travel the distance given
by the Einstein-Smoluchowski equation. Actually, the spatial distribution
of the particles is given by a semi-bell-shaped distribution curve.
The final step involves the relation of the diffusion coefficient to the
structure of the medium and the forces operating there. It is all a matter
of the mean distance I through which ions jump during the course of their
random walk and of the mean jump frequency k. The latter can be expressed
in terms of the theory of rate processes, so that one ends up with an ex-
pression for the rate of diffusion which is in principle derivable from the
local structure of the medium.
This, then, is an elementary picture of diffusion. The next task is to
consider the phenomenon of conduction, i.e., the migration of ions in an
electric field.
 ION TRANSPORT IN SOLUTIONS 345

Further Reading

1. M. Smoluchowski, Ann. Phys. (Paris), 25: 205 (1908).

2. A. Einstein, Investigations on the Theory of Brownian Movement, Methuen
& Co., Ltd., London, 1926.
3. H. S. Carslaw, Introduction to the Theory of Fourier Series and Integrals,
Macmillan & Co., Ltd., London, 1930.
4. S. Glasstone, K. J. Laidler, and H. Eyring, The Theory of Rate Processes,
McGraw-Hill Book Company, New York, 1941.
5. S. Chandrasekhar, "Noise and Stochastic Processes," Rev. Mod. Phys., 15:
1 (1943).
6. J. C. Jaeger, An Introduction to the Laplace Transformation, John Wiley &
Sons, Inc., New York, 1951.
7. H. S. Carslaw and J. C. Jaeger, Conduction of Heat in Solids, Clarendon Press,
Oxford, 1959.
8. Ling Yang, and M. T. Simnad, "Measurement of Diffusivity in Liquid Sys-
tems," in: Physicochemical Measurements at High Temperatures, J. O'M.
Bockris. J. L. White, and J. W. Tomlinson, eds., Butterworth's Publicat-
ions, Ltd., London, 1959.
9. W. Jost, Diffusion in Solids, Liquids, Gases, Academic Press, Inc., New York,
1960.
10. P. Delahay, Chapter 5 in: P. Delahay and C. W. Tobias, eds., Advances in
Electrochemistry and Electrochemical Engineering, Vol. 1, Interscience Pub-
lishers, Inc., New York, 1961.
11. R. P. Feynman, Lectures on Physics, Vol. 1, Addison-Wesley Publishing Co.
Reading, Mass., 1964.

4.3. IONIC DRIFT UNDER AN ELECTRIC FIELD: CONDUCTION

4.3.1. The Creation of an Electric Field in an Electrolyte

Consider that two plane-parallel electrodes are introduced into an

electrolytic solution so as to cover the ends walls of the rectangular insulat-
ing container (Fig. 4.40). With the aid of an external source, let a potential
difference be applied across the electrodes. How does this applied potential
difference affect the ions in the solution?
The potential in the solution has to vary from the value at one electrode,
"PI, to that at the other electrode, "Pu. The major portion of this potential
drop "PI - "PH occurs across the two electrode-solution interfaces (see
Chapter 7), i.e., if the potentials on the solution side of the two interfaces
are "PI' and "P1l', then the interfacial potential differences are "PI - "P/
and "Pu' - "PI (Fig. 4.41). The remaining potential drop "PI' - "Pu', occurs
346 CHAPTER 4

- INSULATING
CONTAINER

METAL ELECTRODE
AT POTENTIAL. 'VII

METAL ELECTROOE ELECTROLYTIC

AT POTENTIAL. VI SOLUTION

Fig. 4.40. An electrochemical system consisting of

two plane-parallel electrodes immersed in an elec-
trolytic solution is connected to a source of poten-
tial difference.

in the electrolytic solution. The electrolytic solution is therefore a region

of space in which the potential at a point is a function of the distance of
that point from the electrodes.
Let the test ion in the solution be at the point Xl, where the potential
is "PI (Fig. 4.42). This potential is by definition the work done to bring a
unit of positive charge from infinity up to the particular point. (In the

Electrode I - electrolyte
interface

:2 ¥X ~--------,
..
1:
...
'0
.!!
~
e
..
u..
IU

Fig. 4.41. Diagram to illustrate how the total po-

tential difference 'PI - 'PII is distributed in the region
between the two electrodes.
 ION TRANSPORT IN SOLUTIONS 347

I infinity
I
I

ELECTRODE

ELECTROLYTE

Fig. 4.42. The work done to transport a

unit of positive charge from Xl to X. in the
solution is equal to the dilference ?PI - ?P.
in electrostatic potential at the two points.

course of this journey of the test charge from infinity to the particular
point, it may have to cross phase boundaries, for example, the electrolyte-air
boundary, and thereby do extra work (see Chapter 7). Such surface work
terms cancel out however in discussions of the potential differences between
two points in the same medium.) If another point X 2 is chosen on the normal
from x to the electrodes, then the potential at X 2 is different from that at XI
because of the variation of the potential along the distance coordinate
between the electrodes. Let the potential at X 2 be '!p2' Then, '!pI - '!p2, the
potential difference between the two points, is the work done to take a unit
of charge from Xl to X 2 .
When this work '!pI - '!p2 is divided by the distance over which the test
charge is transferred, i.e., Xl - X 2 , one obtains the force per unit charge,
or the electric field X:
X = _ '!pI - '!p2
(4.115)
XI - X2

where the minus sign indicates that the force acts on a positive charge in a
direction opposite to the direction of the positive gradient of potential.
In the particular case under discussion, i.e., parallel electrodes covering the
end walls of a rectangular container, the potential drop in the electrolyte is
linear (as in the case of a parallel-plate condenser), and one can write

X = _ Potential difference across the solution

(4.116)
Distance across solution
348 CHAPTER 4

POTENTIAL
'1/

DISTANCE

(0 )

Direction of
Increasing potential •

_*"
Direction of electric field
xa
Direction of motion of
positive charge

(b)

Fig. 4.43. (a) In the case of a nonlinear

potential variation in the solution, the elec-
tric field at a point is the negative gradient
of the electrostatic potential at that point.
(b) The relative directions of increasing po-
tential, field, and motion of a positive charge.

In general, however, it is best to be in a position to treat nonlinear potential

drops. This is done by writing (Fig. 4.43)

X= _ d1p (4.117)
dx

where now the electric field may be a function of x.

The imposition of a potential difference between two electrodes thus
makes an electrolytic solution the scene of operation of an electric field,
i.e., an electric force, acting upon the charges present. This field can be
mapped by drawing equipotential surfaces (all points associated with the
same potential lie on the same surface). The potential map yields a geo-
metrical representation of the field. In the case of plane-parallel electrodes
extending up to the walls of a rectangular cell, the equipotential surfaces
are parallel to the electrodes (Fig. 4.44).
 ION TRANSPORT IN SOLUTIONS 349

ELECTRODES

I
EQUIPOTENTIAL SURFACE

Fig. 4.44. A geometric representation of

the electric field in an electrochemical sys-
tem in which plane-parallel electrodes are
immersed in an electrolyte so that they ex-
tend up to the walls of rectangular insu-
lating container. The equipotential surfaces
are parallel to the electrodes.

4.3.2. How Do Ions Respond to the Electric Field?

In the absence of an electric field, the ions are in ceaseless random

motion. This random walk of ions has been shown to have an important
characteristic: The mean distance traversed by the ions as a whole is zero
because, while some are displaced in one direction, an equal number are
displaced in the opposite direction. From a phenomenological view, there-
fore, the random walk of ions can be ignored because it does not lead to

I~I- x:+p +

2 '+' _pV~~:: ct/

'IONSAT

+
REST

~-. RANDOM
WALK

(0) ATOMISTIC VIEW (b) MACROSCOPIC VIEW

Fig. 4.45. (a) Schematic representation of the random

walk of two ions showing that ion 1 is displaced a dis-
tance x = +p and ion 2, a distance x = -p and, hence,
the mean distance traversed is zero. (b) Since the mean
distance traversed by the ions is zero, there is no net flux
of the ions and they can legitimately be considered, from
a macroscopic point of view, at rest.
350 CHAPTER 4

(0) ATOMISTIC VIEW (b) MACROSCOPIC VIEW

Fig. 4.46. (a) Schematic representation of the move-

ments of four ions which random walk in the presence
of a field. Their displacements are +p, -p, +p, and +p,
i.e., the mean displacement is finite. (b) From a mac-
roscopic point of view, one can ignore the random walk
and consider that each ion drifts in the direction of the
field.

any net transport matter (so long as there is no difference in concentration

in various parts of the solution so that net diffusion down the concentration
gradient occurs). The net result is as if the ions were at rest (Fig. 4.45).
Under the influence of an electric field, however, the net result of the
zig-zag jumping of ions is not zero. Ions feel the electric field, i.e., they
experience a force directing them toward the electrode which is charged
oppositely to the charge on the ion. This directed force is equal to the charge
on the ion, Z ieo, times the field at the point where the ion is situated. The
driving force of the electric field produces in all ions of a particular species
a velocity component in the direction p of the potential gradient. Thus, the
establishment of a potential difference between the electrodes produces a
drift, or flux, of ions (Fig. 4.46). This drift is the migration (or conduction)
of ions in response to an electric field.
As in diffusion, the relationship between the steady-state flux J of
ions and the driving force of the electric field will be represented by the
expression
J = A + BX + CX2 + ... (4.118)

For small fields, the terms higher than BX will tend to zero. Further, the
constant A must be zero because the flux of ions must vanish when the
field is zero. Hence, for small fields, the flux of ions is proportional to the
field (cf Section 4.2.2)
J=BX (4.119)
 ION TRANSPORT IN SOLUTIONS 351

4.3.3. The Tendency for a Conflict between Electroneutrality

and Conduction

When a potential gradient, i.e., electric field, exists in an electrolytic

solution, the positive ions drift toward the negative electrode; and the
negative ions, in the opposite direction. What is the effect of this ionic
drift on the state of charge of an electrolytic solution?
Prior to the application of an external field, there is a time-average
electroneutrality in the electrolyte over a distance large compared with
X-I (cf Section 3.3.8), i.e., the net charge in any macroscopic volume of

solution is zero because the total charge due to the positive ions is equal
to the total charge due to the negative ions. Owing to the electric field,
however, ionic drift tends to produce a spatial separation of charge. Positive
ions will try to segregate near the negatively charged electrode; and negative
ions, near the positively charged electrode.
This tendency for gross charge separation has an important implication;
electroneutrality tends to be upset. Further, the separated charge causing
the lack of electroneutrality tends to set up its own field, which would run
counter to the externally applied field. If the two fields were to become equal
in magnitude, the net field in the solution would become zero. (Thus, the
driving force on an ion would vanish and ion migration stop.)
It appears from this argument that an electrolytic solution would sustain
only a transient migration of ions, and, then, the tendency to conform to the
principle of electroneutrality would result in a halt in the drift of ions after
a short time. A persistent flow of charge, an electric current, appears to
be impossible. In practice, however, an electrolytic solution can act as a
conductor of electricity and is able to pass a current, i.e., maintain a flow
of ions. Is there a paradox here?

4.3.4. The Resolution of the Electroneutrality-versus-Conduc-

tion Dilemma: Electron-Transfer Reactions

The solution to the dilemma just posed can be found by comparing

an electrolytic solution with a metallic conductor. In a metallic conductor,
there is a lattice of positive ions which hold their equilibrium positions
during the conduction process. In addition, there are the free conduction
electrons which assume responsibility for the transport of charge. Contact
is made to and from the metallic conductor by means of other metallic
conductors (Fig. 4.47a). Hence, electrons act as charge carriers throughout
the entire circuit.
In the case of an electrolytic conductor, however, it is necessary to
352 CHAPTER 4

METALUC
CONNECTING
WIRE

FREE ELECTRONS FLOATING

IN A LATTICE OF FIXED
POSITIVE IONS
MAGNIFIED
REPRESENTATION ----<--t;,:~~+1
OF VOLUME ELEMENT

(a)

Meta lIie
eonneClinQ
wire

electrons
transport charge
in metol
ions transport charge in electrolyte

Fig. 4.47. Comparison of electric circuits which consist of (a)

a metallic conductor only, and (b) an electrolytic conductor as
well as a metallic one.

make electrical contact to and from the electrolyte by metallic conductors

(wires). Thus, here one has the interesting situation in which electrons
transport charge in the external circuit and ions carry the charge in the
electrolytic solution (Fig. 4.47b). Obviously, one can maintain a steady
flow of charge (current) in the entire circuit only if there is a change of
charge carrier at the electrode-electrolyte interface. In other words, for a
current to flow in the circuit, ions have to hand over or take electrons from
the electrodes.
Such electron transfers between ions and electrodes result in chemical
changes (changes in the valence or oxidation state of the ions), i.e., in
electrodic reactions. When ions receive electrons from the electrode, they
are said to be "electronated," or to undergo reduction; when ions donate
electrons to the electrodes, they are said to be "deelectronated," or to
undergo oxidation.
 ION TRANSPORT IN SOLUTIONS 353

The occurrence of a reaction at each electrode is tantamount (oxida-

tion = deelectronation; reduction = electronation) to removal of equal
amounts of positive and negative charge from the solution. Hence, when
electron-transfer reactions occur at the electrodes, ionic drift does not lead
to segregation of charges and the building-up of an electroneutrality field
(opposite to the applied field). Thus, the flow of charge can continue, i.e.,
the solution conducts. It is an ionic conductor.

4.3.5. The Quantitative Link between Electron Flow in the Elec-

trodes and Ion Flow in the Electrolyte: Faraday's Law

Charge transfer is the essence of an electrodic reaction. It constitutes

the bridge between the current Ie of electrons in the electrode part of the
electrical circuit and the current Ii of ions in the electrolytic part of the
circuit (Fig. 4.48). When a steady-state current is passing through the
circuit, there must be a continuity in the currents at the electrode-electrolyte
interface, i.e.,
(4.120)

(This is in fact an example of Kirchhoff's law, which says that the algebraic
sum of the currents at any junction must be zero.) Further, if one multiplies
both sides of Eq. (4.120) by the time t, one obtains

(4.121 )

which means t that the quantity Qe of electricity carried by the electrons

is equal to that carried by the ions

(4.122)

Let the quantity of electricity due to electron flow be the charge borne
by an Avogadro number of electrons, i.e., Qe = NAe O = F. If the charge
on each ion participating in the e1ectrodic reaction is ZieO, it is easily seen
that the number of ions required to preserve equality of currents [Eq.
(4.120)] and equality of charge transported across the interface in time t
[Eq. (4.122)] is

(4.123)
= I g-eq

t The product of the current and time is the quantity of electricity.

354 CHAPTER 4

DE
ELECTROOE

ELECTRON
CURRENT
Ie

INTERFACES
{ THE SCENE OF CHARGE TRANSFER
REACTIONS}

Fig. 4.48. Diagram for the derivation of Faraday's

Laws. The electron current Ie in the metallic part of
the circuit must be equal to the ion current 'i.
in the
electrolytic part of the circuit.

Thus, the requirement of steady-state continuity of current at the in-

terface leads to the following law: The passage of 1 faraday (F) of charge
results in the electrodic reaction of one equivalent (ljzi moles) of ions
each of charge ZieO' This is Faraday's law.+ Conversely, if l/zi moles of
ions undergo charge transfer, then 1 F of electricity passes through the
circuit, or zF Faradays per mole of ions transformed.

4.3.6. The Proportionality Constant Relating the Electric Field

and the Current Density: The Specific Conductivity

In the case of small fields, the steady-state flux of ions can be con-
sidered proportional to the driving force of an electric field (cf Section
4.3.2), i.e.,
J=BX (4.119)

The quantity J is the number of moles of ions crossing a unit area per second.
When J is multiplied by the charge borne by 1 mole of ions, zF, one obtains
the current density i, or charge flux, i.e., the quantity of charge crossing unit
area per second. Hence,
i = JzF= zFBX (4.124)

t Alternatively, Faraday's law states that, if a current of I amp passes for a time t sec,
then It/zF moles of reactants in the electrodic reaction are produced or consumed.
 ION TRANSPORT IN SOLUTIONS 355

TABLE 4.7
Representative Values of Specific Conductivity

Specific conductivity,
Substance Type of conductor tOC
ohm- 1 cm- 1

Copper Metallic 5.8 x 105 20

Lead Metallic 4.9 x 105 0
Iron Metallic 1.1 x 105 0
4M H 2S0 4 Electrolytic 7.5 x 10-1 18
0.1 M KCl Electrolytic 1.3 x 10- 2 25
Xylene Nonelectrolyte 1 x 10-1• 25
Water Nonelectrolyte 4 x 10- 8 18

The constant zFB can be set equal to a new constant G, which is known
as the specific conductivity (Table 4.7). The relation between the current
density i and the electric field X is therefore
1 .
X=-l (4.125)
G

The electric field is very simply related (Fig. 4.49) to the potential difference
across the electrolyte, (1p/ - 1pn') [see Eq. (4.116)],

X = Ll1p (4.126)
1
where I is the distance across the electrolyte. Further, the total current I
is equal to the area A of the electrodes times the current density i

/ = iA (4.127)

Substituting these relations [Eqs. (4.126) and (4.127)] in the field-current-

density relation [Eq. (4.125)], one has
Ll1p /
1 G A
or
1
Ll1p = - / (4.128)
GA
The constants G, I, and A determine the resistance R of the solution

R =_1_ (4.129)
GA
356 CHAPTER 4

Electrolyte

Electrode :t Electrode :u:

- c
~.2
c- 1----
II':
~g L -_ _ _ _ _ _ _ _ _ _ _ _L-_~
Distance from electrode I

Fig. 4.49. Schematic representation of the var-

iation of the potential in the electrolytic conduc-
tor of length I.

and, therefore, one has the equation

Ll1p = RI (4.130)

which re-expresses in the conventional Ohm's law form the assumption of

(4.119) concerning flux and driving force.
Thus, an electrolytic conductor obeys Ohm's law for small fields, and,
under steady-state conditions, it can be represented in an electrical circuit
(in which there is only a dc source) by a resistor. (An analog must obey
the same equation as the system it represents or simulates.)

CONDUCTANCE OF THIS CUBE OF SOLUTION =

SPECIFIC CONDUCTIVITY rT

I
I
I
I
I
I
/
}-------
/

----I cm----
Fig. 4.50. Diagram to illustrate the meaning
of the specific conductivity of an electrolyte.
 ION TRANSPORT IN SOLUTIONS 357

As in the case of a resistor, the dc resistance of an electrolytic cell

increases with the length of the conductor (distance between the electrodes)
and decreases with the area [cf Eq. (4.129)]. It can also be seen, by
rearranging this equation into the form

1 I
(J =-- (4.131)
R A

that the specific conductivity (J is the conductance 1/ R of a cube of electrolytic

solution I cm long and I cm2 in area (Fig. 4.50).

4.3.7. Molar Conductivity and Equivalent Conductivity

In the case of metallic conductors, once the specific conductivity is
defined, the macroscopic description of the conductor is complete. In the
case of electrolytic conductors, further characterization is imperative be-
cause not only can the concentration of charge carriers vary but also the
charge per charge carrier.
Thus, even though two electrolytic conductors have the same geometry,
they need not necessarily have the same specific conductivity (Fig. 4.51
and Table 4.8); the number of charge carriers in that normalized geometry
may be different, in which case their fluxes under an applied electric field
will be different. Since the specific conductivity of an electrolytic solution
varies as the concentration, one can write

(J = fCc) (4.132)

where c is the concentration of the solution in gram-moles of solute dissolved

DIFFERENT CONCENTRATIONS
OF CHARGE CARRIERS

Fig. 4.51. A schematic explanation of the varia-

tion of the specific conductivity with electrolyte
concentration.
358 CHAPTER 4

TABLE 4.8
Specific Conductivity of KCI Solutions

Specific conductivities
KCI, g kg- l of solution
18°C 25 DC

1.0 0.065144 0.097790 0.11187

0.1 0.0071344 0.0111612 0.012896
0.01 0.00077326 0.00121992 0.001427

In I cm3 of solution. t The specific conductivities of two solutions can be

compared only if they contain the same concentration of ions. The con-
clusion is that, in order to compare the conductances of electrolytic con-
ductors, one has to normalize (set the variable quantities equal to unity)
not only the geometry but also the concentration of ions.
The normalization of the geometry (taking electrodes of I cm2 in area
and 1 cm apart) defines the specific conductivity; the additional normaliza-
tion of the concentration (taking 1 mole of'ions) defines a new quantity,
the molar conductivity (Table 4.9),

A m =~=(JV
c (4.133)

where V is the volume of solution containing I g-mole of solute (Fig. 4.52).

Defined thus, it can be seen that the molar conductivity is the specific
conductivity of a solution times the volume of that solution in which is
dissolved I g mole of solute; the molar conductivity is a kind of conductivity
per particle.
One can usefully compare the molar conductivities of two electrolytic
solutions only if the charges borne by the charge carriers in the two solutions
are the same. If there are singly charged ions in one electrolyte (e.g., NaCl)
and doubly charged ions in the other (e.g., BaS04 ), then the two solutions
will contain different amounts of charge even though the same quantity
of the two electrolytes is dissolved. In such a case, the specific conductivities
of the two solutions can be compared only if they contain equivalent amounts
of charge. This can be arranged by taking I mole of charge in each case, i.e.,

t As in the case of diffusion fluxes, the concentrations used in the definition of conduc-
tion currents (or fluxes) and conductances are not in the usual moles per liter but in
moles per cubic centimeter.
 ION TRANSPORT IN SOLUTIONS 359

TABLE 4.9
Molar Conductivities of Electrolytes

Molar conductivity,
Electrolyte
at 0.01 mole liter- 1

KCl 141.3
NaCi 118.5
MgCi. 229.2
Na.SO, 224.8

1 mole of ions divided by z, or 1 g-eq of the substance. Thus, the equivalent

conductivity A of a solution is the specific conductivity of a solution times
the volume V of that solution containing 1 g-eq of solute dissolved in it
(Fig. 4.53). Hence, the equivalent conductivity is given bl

A=~ (4.134)
cz

where cz is the number of gram-equivalents per cubic centimeter of solution.

There is a simple relation between the molar and equivalent conduc-
tivities. It is [cf Eqs. (4.133) and (4.134)]

(4.135)

The equivalent conductivities of some electrolytes are shown in Table 4.10.

CONDUCTANCE OF THIS CUBE OF SOLUTION _

MOLAR CONDUCTIVITY

v cm'AREA

I
I
)------
III mole of /
electrolyte "
Fig. 4.52. Diagram to illustrate the meaning of the
molar conductivity of an electrolyte.

t Since liz mole of ions is I g-eq, c moles is cz g-eq.

360 CHAPTER 4

CONDUCTANCE OF THIS CUBE OF SOLUTION. EQUIVALENT CONDUCTIVITY

,
/
ONE MOLE OF
CHARGES!
ONE GM.
EQUIVALENT OF
ELECTROLYTE

Fig. 4.53. Diagram to illustrate the meaning of the equivalent

conductivity of an electrolyte.

The equivalent conductivity A is in the region (±25%) of 100 ohm- 1 cm- 1

for most electrolytes.

4.3.8. The Equivalent Conductivity Varies with Concentration

At first sight, the title of this section may appear surprising. The
equivalent conductivity has been defined by normalizing the geometry of
the system and the charge of the ions; why, then, should it vary with con-
centration? Experiment, however, gives an unexpected answer. The equiv-
alent conductivity does vary significantly with the concentration of ions
(Table 4.11). The direction of the variation may also surprise some, for the
equivalent conductivity increases as the ionic concentration decreases
(Fig. 4.54).
Now, it would be awkward to have to refer to the concentration every
time one wished to state the value of the equivalent conductivity of an
electrolyte. One should be able to define some reference value for the

TABLE 4.10
Equivalent Conductivities of True Electrolytes in Dilute Aqueous Solutions
at 25 0 C

Equivalent conductivity,
Electrolyte
ohm- 1 cm 2 eq-l at 0.005 g-eq Iiter- l

KCl 143.55
NaOH 240.00
AgNO a 127.20
! BaCI. 128.02
! NiSO. 93.20
 ION TRANSPORT IN SOLUTIONS 361

TABLE 4.11
Equivalent Conductance Varies with Concentration

Concentration, eq Iiter- 1
(KCI solutions)

0.001 146.9
0.005 143.5
om 141.2
0.02 138.2
0.05 133.3
0.10 128.9

equivalent conductivity. Here, the facts of the experimental variation of

equivalent conductivity with concentration comes to one's aid; as the
electrolytic solution is made more dilute, the equivalent conductivity ap-
proaches a limiting value. This limiting value of equivalent conductivity
should form an excellent basis for the comparison of the conducting powers
of different electrolytes for it is the only value in which the effects of ionic
concentration are removed. The limiting value shall be called the equivalent
conductivity at infinite dilution, referred to by the symbol AO (Table 4.12).
Now, it may be argued: At infinite dilution, there are no ions of the
solute; then, how can the solution conduct? The procedure for determining
the equivalent conductivity of an electrolyte at infinite dilution will clarify
this problem. One takes solutions of a substance in various concentrations,
determines the (J, and then normalizes each to the equivalent conductivity
of particular solutions. If these values of A are then plotted against the

160

120

A 80

40

o L-_4:-'--_2~'--0~

log C
Fig. 4.54. The observed variation of the
equivalent conductivity of CaCI. with con-
centration.
362 CHAPTER 4

TABLE 4.12
Equivalent Conductivities at Infinite Dilution, AO, of Electrolytes in Aqueous
Solution at 25 0 C

Electrolyte AO, ohm- 1 cm" eq-l

HCI 426.0
NaOH 247.9
NaCI 126.4
KCI 149.8
K 4 Fe(CN)6 183.9
CaCI. 135.7

reciprocals of the concentration and this A versus log c curve is extrapolated,

it approaches a limiting value (Fig. 4.55). It is this extrapolated value at
zero concentration which is known as the equivalent conductivity at infinite
dilution.
Anticipating the atomistic treatment of conduction which follows, it
may be mentioned that, at very low ionic concentrations, the ions are too
far apart to exert appreciable interionic forces on each other. Only under
these conditions does one obtain the pristine version of equivalent conduc-
tivity, i.e., values unperturbed by ion-ion interactions, which have been
shown in Chapter 3 to be concentration dependent. The state of infinite
dilution, therefore, is not only the reference state for the study of equilibrium
properties (cf Chapter 3.3); it is also the reference state for the study of
the nonequilibrium (irreversible) process, which is called ionic conduction,
or migration (cf Section 4.1).

Equivalent conductivities
at infinite di lution

~.
.J\. 120
-:~
100
80

60~_4~_±5-_~2~_~1~o--
109 C

Fig. 4.55. The equivalent conductivity of

an electrolyte at infinite dilution is obtained
by extrapolating the A versus log c curve
to zero concentration.
 ION TRANSPORT IN SOLUTIONS 363

4.3.9. How the Equivalent Conductivity Changes with Concen-

tration: Kohlrausch's Law

The experimental relationship between the equivalent conductivity and

the concentration of an electrolytic solution is best brought out by plotting
A against d. When this is done (Fig. 4.56), it can be seen that, up to con-
centrations of about 0.0 I N, there is a linear relationship between A and ci ;
thus,
A =AO - Ad (4.136)

where the intercept is the equivalent conductivity at infinite dilution, AO,

and the slope of the straight line is a positive constant A.
This empirical relationship between the equivalent conductivity and
the square root of concentration is a law named after Kohlrausch. His
extremely careful measurements of the conductance of electrolytic solutions
can be considered to have played a leading role in the initiation of ionics,
the physical chemistry of ionic solutions. However, Kohlrausch's law [Eq.
(4.136)] had to remain nearly 40 years without a theoretical basis.
The justification of Kohlrausch's law on theoretical grounds cannot
be obtained within the framework of a macroscopic description of con-
duction. It requires an intimate view of ions in motion. A clue to the type
of theory required emerges from the empirical findings by Kohlrausch:
(1) the ci dependence, and (2) the intercepts AO and slopes A of the A
versus c~ curve depend not so much on the particular electrolyte (whether
it is KCl or NaCl) as on the type of electrolyte (whether it is a I: I or 2:2
electrolyte) (Fig. 4.57). All this is reminiscent of the dependence of the
activity coefficient on d (Chapter 3), to explain which, the subtleties of

100 L..--=2--'4=-~6~8-X 10-2

ct
Fig. 4.56. The experimental basis for
Kohlrausch's law. A versus cf plots. con-
sist of straight lines.
364 CHAPTER 4

fC-
Fig. 4.57. The experimental A versus c!
plots depend largely on the type of electro-
lyte.

ion-ion interactions had to be explored. Such interactions between positive

and negative ions would determine to what extent they would influence
each other when they move, and this would in turn bring about a fall of
conductivity.
Kohlrausch's law will therefore be left now with only the sanction of
experiment. Its incorporation into a theoretical scheme will be postponed
until the section on the atomistic view of conduction is reached (see Sec-
tion 4.6.12).

4.3.10. The Vectorial Character of Current: Kohlrau sch' 5 Law

of the Independent Migration of Ions

The driving force for ionic drift, i.e., the electric field X, not only has
a particular magnitude, it also acts in a particular direction. It is a vector.
Since the ionic current density i, i.e., the flow of electric charge, is propor-
tional to the electric field operating in the solution [see Eq. (4.125)],

i = aX (4.125)

the ionic current density must also be a vector. Vectorial quantities are often
designated by arrows placed over the quantities (unless their directed
 ION TRANSPORT IN SOLUTIONS 365

POSITIVE ELECTRODE

NEGATIVE
ELECTRODE

FLUX OF NEGATIVE IONS ~ -

AWAY FROM NEGATIVE ELECTRODE • .J_

FLUX OF POSITIVE IONS TOWARDS NEGATIVE ELECTRODE.

Fig. 4.58. Schematic representation of the direction of the

--
J+

drifts (and fluxes) of positive and negative ions acted on

by an electric field.

character is obvious). Hence, Eq. (4.125) can be written thus

-->- -->-
i = aX (4.137)

How is this current density constituted? What are its components?

What is the structure of this ionic current density?
The imposition of an electric field on the electrolyte (Fig. 4.58) makes
the positive ions drift toward the negative electrode and the negative ions
-->-
drift in the opposite direction. The flux of positive ions, J T ' gives rise to a
positive-ion current density i+; and the flux of negative ions in the opposite
+-
direction, J _, results in a negative-ion current density L. By convention
the direction of current flow is taken to be either the direction in which
positive charge flows or the direction opposite to that in which negative
-->-
charge flows. Hence, the positive-ion flux J t- corresponds to a current
-->- +-
toward the negative electrode, iT' and the negative-ion flux J _ also corre-
-->-
sponds to a current L in the same direction as that due to the positive ions.
-->-
It can be concluded therefore that the total current density i is made
up of two contributions, one due to a flux of positive ions and the other due
to a flux of negative ions. Further, assuming for the moment that the drift
of positive ions toward the negative electrode does not interfere with the
drift of negative ions in the opposite direction, it follows that the component
current densities are additive, i.e.,

(4.138)
366 CHAPTER 4

But do ions migrate independently? Is the drift of the positive ions in

one direction uninfluenced by the drift of the negative ions in the opposite
direction? This is so if, and only if, the force fields of the ions do not overlap
significantly, i.e., if there is negligible interaction or coupling between the
ions. Coulombic ion-ion interactions usually establish such coupling. The
only conditions under which the absence of ion-ion interactions can be
assumed occur when the ions are infinitely far apart. Strictly speaking,
therefore, ions migrate independently only at infinite dilution. Under these
conditions, one can proceed from Eq. (4.138) to write
-+ -+ -+
i i~ L
-+=-++-+ (4.139)
X X X
or [from (4.125)],
(4.140)

whence
(4.141)

This is Kohlrausch's law of the independent migration of ions: The equiv-

alent conductivity (at infinite dilution) of an electrolytic solution is the
sum of the equivalent conductivities (at infinite dilution) of the ions con-
stituting the electrolyte (Table 4. I 3).
At appreciable concentrations, the ions can be regarded as coupled or
interacting with each other (cf. the ion-atmosphere model of Chapter 3).
This results in the drift of positive ions toward the negative electrode, which
hinders the drift of negative ions toward the positive electrode, i.e., the
interionic interaction results in the positive ions' equivalent conductiv-
ity reducing the magnitude of the negative ions' equivalent conductivity to

TABLE 4.13
Equivalent Conductances of Individual Ions at Infinite Dilution
ohm _1 em' at 25 0 C

Cation },+o Anion }._ 0

H+ 349.82 OH- 198.5

K+ 73.52 Br- 78.4
Na+ 50.11 I- 76.8
Li+ 38.69 CI- 76.34
! BaH 63.64 CH 3 CO-. 40.9
 ION TRANSPORT IN SOLUTIONS 367

below the infinite dilution value, and vice versa. To make quantitative
estimates of these effects, however, one must make calculations of the
influence of ionic-cloud effects on the phenomenon of conduction, a task
which will be taken up further on.

Further Reading
1. S. Glasstone, An Introduction to Electrochemistry, D. van Nostrand Co., Inc.,
Princeton, N.J., 1949.
2. R. M. Fuoss and F. Accasina, Electrolytic Conductance, Interscience Publishers,
Inc., New York, 1959.

4.4. THE SIMPLE ATOMISTIC PICTURE OF IONIC MIGRATION

4.4.1. Ionic Movements under the Influence of an Applied Elec-

tric Field
In seeking an atomic view of the process of conduction, one approach
is to begin with the picture of ionic movements as described in the treat-
ment of diffusion (Section 4.2.4) and then to consider how these move-
ments are perturbed by an electric field. In the treatment of ionic move-
ments, it was stated that the ions in solution perform a random walk in
which all possible directions are equally likely for any particular step.
The analysis of such a random walk indicated that the mean displacement
of ions is zero (Section 4.2.4), diffusion being the result of the statistical
bias in the movement of ions, due to inequalities in their numbers in different
regions.
When, however, the ions are situated in an electric field, their move-
ments are affected by the fact that they are charged. Hence, the imposition
of an electric field singles out one direction in space (the direction parallel
to the field) for preferential ionic movement. Positively charged particles
will prefer to move toward the negative electrode; and negatively charged
particles, in the opposite direction. The walk is no longer quite random.
The ions drift.
Another way of looking at ionic drift is to consider the fate of any
particular ion under the field. The electric force field would impart to it
an acceleration according to Newton's second law. Were the ion completely
isolated (e.g., in vacuum), it would accelerate indefinitely until it collided
with the electrode. In an electrolytic solution, however, the ion very soon
collides with some other ion or solvent molecule which crosses its path.
This collision introduces a discontinuity in its speed and direction. The
motion of the ion is not smooth; it is as if the medium offers resistance
368 CHAPTER 4

to the motion of the ion. Thus, the ion stops and starts and zig-zags. But
the applied electric field imparts to the ion a direction (that of the oppositely
charged electrode), and the ion works its way, though erratically, in the
direction of this electrode. The ion drifts in a preferred direction.

4.4.2. What Is the Average Value of the Drift Velocity?

Any particular ion starts off after a collision with a velocity which may
be in any direction; this is the randomness in its walk. This initial velocity
can be ignored precisely because it can take place in any direction and,
therefore, does not contribute to the drift (preferred motiun) of the ion.
But the ion is all the time under the influence of the applied-force field. t
This force imparts a component to the velocity of the ion, an extra velocity
-+
component in the same direction as the force vector F. It is this additional
-+
velocity component due to the force F which is called the drift velocity Va.
What is its average value?
From Newton's second law, it is known that the force divided by the
mass of the particle is equal to the acceleration. Thus,
-+
F dv
(4.142)
m dr
Now, the time between collisions is a random quantity. Sometimes,
the collisions may occur in rapid succession; at others, there may be fairly
long intervals. It is possible however to talk of a mean time between colli-
sions, T. In Section 4.2.5, it has been shown that the number of collisions
(steps) is proportional to the time. If N collisions occur in a time t, then
the average time between collisions is tiN. Hence,
t
T=- (4.143)
N

The average value of that component of the velocity of an ion picked

up from the externally applied force is simply the product of the acceleration
due to this force and the average time between collisions. Hence, the drift
velocity Va is given by
dv
Va = dr T

=-T
F (4.144)
m

t The argument is developed in general for any force, not necessarily an electric force.
 ION TRANSPORT IN SOLUTIONS 369

This is an important relation. It opens up many vistas. For example,

through the mean time T, one can relate the drift velocity to the details
of ionic jumps between sites, as was done in the case of diffusion (Section
4.2.15).
Further, the relation (4.144) shows that the drift velocity is proportional
to the driving force of the electric field. The flux of ions will be shown
(Section 4.4.4) to be simply related to the drift velocityt in the following way

Flux = Concentration of ions x Drift velocity (4.145)

-+
Thus, if the Fin Eq. (4.144) is the electric force which stimulates con-
duction, then this equation is the molecular basis of the fundamental
relation used in the macroscopic view of conduction, i.e.,

Flux ex: Electric field (4.119)

In fact, the derivation of the basic relation (4.144) reveals the conditions
under which the proportionality between drift velocity (or flux) and electric
field breaks down. Thus, it was essential to the derivation that, in a collision,
an ion does not preserve any part of its extra velocity component arising
from the force field. If it did, then the actual drift velocity would be greater
than that calculated by Eq. (4.144) because there would be a cumulative
carry-over of the extra velocity from collision to collision. In other words,
every collision must wipe out all traces of the force-derived extra velocity,
and the ion must start afresh to acquire the additional velocity. This con-
dition can be satisfied only if the drift velocity and, therefore, the field are
small.

4.4.3. The Mobility of Ions

It has been shown that, when random-walking ions are subjected to
-+
a directed force F, they acquire a nonrandom, directed component of
velocity-the drift velocity vd. This drift velocity is in the direction of the
-+
force F and is proportional to it

T -+
Vd =-F (4.144)
m

t The dimensions of flux are in moles per square centimeter per second, and they are
equal to the product of the dimensions of concentration expressed in moles per cubic
centimeter and velocity expressed in centimeters per second.
370 CHAPTER 4

Since the proportionality constant -r/m is of considerable importance

in discussions of ionic transport, it is useful to refer to it with a special
name. It is called the absolute mobility because it is an index of how mobile
the ions are. The absolute mobility, referred to by the symbol fiabs , is a
measure of the drift velocity Vd acquired by an ionic species when it is
-+
subjected to a force F, i.e.,

(4.146)

which means that the absolute mobility is the drift velocity developed under
-+
unit applied force (F = I dyne) and has its dimensions in centimeters per
second per dyne.
For example, one might have an electric field X of 0.05 V cm-I in the
electrolyte solution and observe a drift velocity of 2 x 10-5 cm sec-I. The
-+
electric force F operating on the ion is equal to the electric force per unit
charge, i.e., the electric field X, times the charge Z.ieO on each ion

-+ I
F = ZieO X 300

= 4.8 X 10-10 X 0.05 X 3~O dynes (4.147)

for univalent ions. (The factor 1/300 comes in because the electronic charge
is in electrostatic units (esu) and the potential involved in the field is not
in electrostatic units of potential, but in volts. Hence, the potential also
must be expressed in electrostatic units, the conversion factor being 1/300,
i.e., 300 volts are equivalent to 1 esu of potential.) Hence, the absolute
mobility is
2 X 10-5 X 300
fiabs = 4.8 X 10-10 X 0.05

:::::: 108 cm sec-I dyne-I

In electrochemical literature, however, mobilities of ions are not usually

expressed in the absolute form defined in Eq. (4.146). Instead, they have
been defined as the drift velocities under the force exerted by unit electric
field (1 V cm-I) on the charge of the ion and will be referred to here as
conventional (electrochemical) mobilities with the symbol Uconv . Thus,

Vd [em sec-1 ]
(4.148)
Uconv = X Vern-I
 ION TRANSPORT IN SOLUTIONS 371

The relation between the absolute and conventional mobilities follows

thus from Eqs. (4.146) and (4.148)

I.e.,

(4.149)

Thus, the conventional and absolute mobilities are simply proportional to

each other, the proportionality constant being an integral multiple Zi of
the electronic charge divided by 300. In the example cited above,

2 X 10- 5
0.05
10-,1 cm sec- 1
I V cm- 1

Though the two types of mobilities are closely related, it must be stressed
that the concept of absolute mobility is more general because it can be used
for any force which determines the drift velocity of ions and not only the
electric force used in the definition of conventional mobilities.

4.4.4. The Current Density Associated with the Directed Move-

ment of Ions in Solution, in Terms of the Ionic Drift
Velocities

It is the aim now to show how the concept of drift velocity can be used
to obtain an expression for the ionic current density flowing through an
electrolyte in response to an externally applied electric field.
Consider a transit plane of unit area normal to the direction of drift
(Fig. 4.59). Both the positive and the negative ions will drift across this
plane. Consider the positive ions first, and let their drift velocity be (v d)""- ,
or simply v -c' Then, in I sec, all positive ions within a distance v _ cm of
the transit plane will cross it. The flux 1_ of positive ions, i.e., the number
of moles of these ions arriving in I sec at unit area of the plane is equal
to the number of moles of positive ions in a volume I cm2 in area and v ~ cm
 372 CHAPTER 4

TRANSIT OR
SLAB OF ELECTROLYTE . REI'ER~,NCE PLANE
POSITIVE IONS IN ntis
VOLUME, V+ CROSS REFERENCE
PLANE PER SECOND,I.e,
J+=V+C..

Fig. 4.59. Diagram for the derivation of a relation between the

current density and the drift velocity.

in length. Hence, J+ is equal to the volume v+ in cubic centimeters times

the concentration c+ expressed t in moles per cubic centimeter

(4.150)

The flow of charge across the plane due to this flux of positive ions, i.e.,
the current density i+ is obtained by mUltiplying the flux J+ by the charge
z-tE borne by I mole of ions
(4.151)

This, however, is only the contribution of the positive ions. Other

ionic species will make their own contributions of current density. In general,
therefore, the current density due to the jth species will be

(4.152)

The total current density due to the contribution of all the ionic species
will therefore be
i= ~ ij (4.153)
j

(4.154)

t Since one is concerned with the number of moles in a volume of v cm3 , one must not
express the concentration in moles per liter but in moles per cubic centimeter. The two
concentrations, however, are simply related as
c
c [moles cm-3] = - - [moles Iiter-1 ]
1000
 ION TRANSPORT IN SOLUTIONS 373

If a z:z-valent electrolyte is taken, then z+ = L = z and c = c+ = c and

one has
(4.155)

By recalling that the ionic drift velocities are related, through the force
operating on the ions, to the ionic mobilities [Eq. (4.148)], it will be realized
that Eq. (4.154) is the basic expression from which may be derived the
expressions for conductance, equivalent conductivity, specific conductivity,
etc.

4.4.5. The Specific and Equivalent Conductivities in Terms of

the Ionic Mobilities

Let the fundamental expression [£q. (4.148)] for the drift velocity of
ions be substituted in Eq. (4.154) for current density. One obtains

(4.156)
or, from (4.125),
(4.157)

which red uces in the special case of a z :z-valent electrolyte to

(4.158)

Several conclusions follow from this atomistic expression for specific

conductivity. Firstly, it is obvious from this equation that the specific
conductivity (J of an electrolyte cannot be a concentration-independent
constant (as it is in the case of metals). It will vary because the number of
moles of ions per unit volume, c, can be varied in an electrolytic solution.
Secondly, the specific conductivity can easily be related to the molar,
ilm, and equivalent, Lt, conductivities. Take the case of a z:z-valent elec-
trolyte. With Eqs. (4.158), (4.133), and (4.134), it is found that

(4.159)
and

(4.160)
374 CHAPTER 4

What does Eq. (4.160) reveal? It shows that the equivalent conductivity
will be a constant independent of concentration only if the electrical mobility
does not vary with concentration. It will be seen however that ion-ion
interactions (which have been shown in Section 3.3.8 to depend on con-
centration) prevent the electrical mobility from being a constant. Hence,
the equivalent conductivity must be a function of concentration.

4.4.6. The Einstein Relation between the Absolute Mobility and

the Diffusion Coefficient
The process of diffusion results from the random walk of ions; the
process of migration (i.e., conduction) results from the drift velocity acquir-
ed by ions when they experience a force. Now, the drift of ions does not
obviate their random walk; in fact, it is superimposed on their random walk.
Hence, the drift and the random walk must be intimately linked. Einstein
realized this and deduced a vital relation between the absolute mobility
Uabs, which is a quantitative characteristic of the drift, and the diffusion
coefficient D, which is a quantitative characteristic of the random walk.
Both diffusion and conduction are nonequilibrium (irreversible)
processes and are therefore not amenable to the methods of equilibrium
thermodynamics or equilibrium statistical mechanics. In these latter disci-
plines, the concepts of time and change are absent. It is possible however
to imagine a situation where the two processes oppose and balance each
other and a "pseudoequilibrium" obtains. This is done as follows (Fig. 4.60).
Consider a solution of an electrolyte MX to which a certain amount

POSITIVELY
CHARGED ELECTRODE

NEGATIVELY
CHARGED
ELECTRODE

DIFFUSION FLUX·
..
GONDUGTlON FLUX
•
1cond"C ua~

Fig. 4.60. An imaginary situation in which the ap-

plied field is adjusted so that the conduction flux of
tagged ions (the only ones shown in the diagram) is
exactly equal and opposite to the diffusion flux.
 ION TRANSPORT IN SOLUTIONS 375

of radioactive M+ ions are added in the form of the salt MX. Further,
suppose that the tracer ions are not dispersed uniformly throughout the
solution; instead, let there be a concentration gradient of the tagged species
such that its diffusion flux J D is given by Fick's first law t
dc
In = - D-r- (4.16)
(X

Now let an electric field be applied. Each tagged ion feels the field, and the
drift velocity is [cf Eq. (4.146)]
.....
I'd = i"iabs F (4.146)

This drift velocity produces a current density given byt [cf Eq. (4.151)]

i = ::FCt'd (4.161 )

i.e., a conduction flux Jc which is arrived at by dividing the conduction

current density i by the charge per mole of ions

(.t 162)

By introducing the expression (4.146) for the drift velocity into (4.162),
the conduction flux becomes
.....
Je = ciiahsF

Now, let the applied field be imagined to be adjusted so that the

conduction flux exactly compensates the diffusion flux. In other words,
if the tracer ions (which are positively charged) are diffusing toward the
positive electrode, then the magnitude of the applied field is such that the
positively charged electrode repels the positive tracer ions to an extent
that their net flux is zero. Thus,

JJ) + Jc = 0 (4.163)
or
dc .....
- D -,-
(X
+ CllabsF = 0
i.e.,
dc
(4.164)
dx

t All the c's in this derivation are in moles per cubic centimeter.
t In Eq. (4.151), one will find I' _: the reason is that. in Section 4.4.4. the drift \-elocit)'
of a positive ion, (I'd) .• had been concisely written as I' _.
376 CHAPTER 4

Under these "balanced" conditions, the situation may be regarded as

tantamount to equilibrium because there is no net flux or transport of ions.
Hence, the Boltzmann law can be used. The argument is that, since the
potential varies along the x direction, the concentrations of ions at any
distance x is given by
(4.165)

where U is the potential energy of an ion in the applied field and CO is the
concentration in a region where the potential energy is zero. Differentiating
this expression, one obtains

de _ _ 0 -U/kT I dU
dx - ee kT dx
e dU
(4.166)
kT dx
But, by the definition of force,
-+ dU
F=-- (4.167)
dx

Hence, from (4.166) and (4.167), one obtains

de e-+
-=-F (4.168)
dx kT

If, now, Eqs. (4.164) and (4.168) are compared, it is obvious that

or
(4.169)

This is the Einstein relation. It is probably the most important relation

in the theory of the movements and drift of ions, atoms, molecules, and
other submicroscopic particles. It has been derived here in an atomistic
way. But it will be recalled that, in the phenomenological treatment of the
diffusion coefficient (Section 4.2.3), it was shown that

D =BRT (4.17)

where B was an undetermined phenomenological coefficient. Now, if one

 ION TRANSPORT IN SOLUTIONS 377

combines (4.169) and (4.17), it is clear that

BRT = fiabskT
or

(4.170)

Thus, one has provided a fundamental basis for the phenomenological

coefficient B; it is the absolute mobility fiabs divided by the Avogadro
number.
The Einstein relation also permits experiments on diffusion to be linked
up with other phenomena involving the mobility of ions, i.e., phenomena
in which there are forces that produce drift velocities. Two such forces are
the force experienced by an ion when it overcomes the viscous drag of a
solution and the force arising from an applied electric field. Thus, the dif-
fusion coefficient may be linked up to the viscosity (the Stokes-Einstein
relation) and to the equivalent conductivity (the Nernst-Einstein relation).

4.4.7. What Is the Drag (or Viscous) Force Acting on an Ion

in Solution?
Striking advances in science sometimes arise from seeing the common
factors in two apparently dissimilar situations. One such advance was made
by Einstein when he intuitively asserted the similarity between a macro-
scopic sphere moving in an incompressible fluid and a particle (e.g., an
ion) moving in a solution (Fig. 4.61).

INCOMPRESSIBLE
M~OlUM OF VISCOSITY "r\
/
/
ION RAOIUS r

8 /
RADIUS r ELECTROLYTIC SOLUTION

REPRESEN TS r: (VISCOSITY. '1 I

Fig. 4.61. An ion drifting in an electrolytic solution

is likened to a sphere (of the same radius as the ion)
moving in an incompressible medium (of the same
viscosity as the electrolyte).
378 CHAPTER 4

The macroscopic sphere experiences a viscous, or drag, force which

opposes its motion. The value of the drag force depends on several factors
-the velocity v and diameter d of the sphere and the viscosity r; and density e
of the medium. These factors can all be combined together and used to
define a dimensionless quantity known as the Reynolds number Re defined
thus
Re = vd.2...- (4.171)
r;

When the hydrodynamic conditions are such that this Reynolds number
is much smaller than unity, Stokes showed that the drag force F opposing
the sphere is given by the following relation

F = 6nrr;v (4.172)

where v is the velocity of the macroscopic body. The relation is known as

Stokes' law. Its derivation is lengthy and awkward, essentially because the
most convenient coordinates to describe the sphere and its environment
are spherical coordinates and those to describe the flow are rectangular
coordinates.
The real question, of course, centers around the applicability of Stokes'
law to microscopic ions moving in a structured medium in which the
surrounding particles are roughly the same size as the ions. Initially, one
can easily check on whether the Reynolds number is smaller than unity
for ions drifting through an electrolyte. With the use of the values
dion ""'" 10-8 cm, Vion = Vd ,....., 10-4 cm sec-I, r;""'" 10-2 poise, and e ': : 1, it
turns out that the Reynolds number for an ion moving through an elec-
trolyte is about 10-1°. Thus, the hydrodynamic condition Re ~ I required
for the validity of Stokes' law is easily satisfied by an ion in solution.
But the hydrodynamic problem which Stokes solved to get F = 6nrr;v
pertains to a sphere moving in an incompressible continuum fluid. This is a
far cry indeed from the actuality of an ion drifting inside a discontinuous
electrolyte containing particles (solvent molecules, other ions, etc.) of
about the same size as the ion. Further, the ions considered may not be
spherical.
From this point of view, the use of Stokes' law for the viscous force
experienced by ions is a bold step. Several attempts have been made over
a long time to theorize about the viscous drag on angstrom-sized particles
in terms of a more realistic model than that used by Stokes. It has been
shown, for example, that, if the moving particle is cylindrical and not
spherical, the factor 6n should be replaced by 4n. While refraining from
 ION TRANSPORT IN SOLUTIONS 379

the none-too-easy analysis of the degree of applicability of Stokes' law to

ions in electrolytes, one point must be stressed. For sufficiently small ions,
Stokes' law does not have a numerical significance t of greater than about
±50%. Attempts to tackle the problem of the flow of ions in solution
without resorting to Stokes' law do not give much better results.

4.4.8. The Stokes-Einstein Relation

During the course of diffusion, the individual particles are executing
the complicated starts, accelerations, collisions, stops, and zig-zags that
have come to be known as random walk. When a particle is engaged in its
random walk, it is, of course, subject to the viscous-drag force exerted by
its environment. But the application of Stokes' law to these detailed random
motions is no easy matter because of the haphazard variation in the speed
and direction of the particles. Instead, one can apply Stokes' law to the
diffusional movements of ions by adopting the following artifice suggested
by Einstein.
When diffusion is occurring, it can be considered that there is a driving
force -dfl/dx operating on the particles. This driving force produces a
steady-state diffusion flux J, corresponding to which [ef Eq. (4.14)] one

t Stokes' law is often used in electrochemical problems, but its approximate nature is
not always brought out. Apart from the validity of extrapolating from the macroscopic-
sphere-<:ontinuum-fluid model of Stokes' to atomic near-spheres in a molecular liquid,
another reason for the limited validity of Stokes' law arises from questions concerning
the radii which should be substituted in any application of the law. These should not
be the crystallographic radii, and an appraisal of the correct value implies a rather
detailed knowledge of the structure of the solvation sheath (ef Section 2.4). Further,
the viscosity used in Stokes' law is the bulk average viscosity of the whole solution,
whereas it is the local viscosity in the neighborhood of the ion which should be taken.
The two may not be the same, because the ion's field may affect the solvent structure
and hence its viscosity.
In recent years, attention has been drawn by Fuoss, Boyd, and Zwanzig to the fact
that, while Stokes' law deals with uncharged spheres, the real situation involves charged
ions. The existence of a charge on the moving body has the following effect on a polar
solvent: The charge tends to produce an orientation of solvent dipoles in the vicinity
of the ion. Since, however, the charge is moving, the dipoles, once oriented, take some
finite relaxation time T to disorient. During this relaxation time, a relaxation force
operates on the ion; this relaxation force is equivalent to an additional frictional force
on the ion and results in an expression for the drag force of the form
s
F = 6:rliVr -'- 6:rliV-
E

where s is 4/9(T/6:TlI)eo'jr 3, and E is the dielectric constant of the medium.

380 CHAPTER 4

can imagine a drift velocity Vd for the diffusing particles. t Since this velocity
Vd is a steady-state velocity, the diffusional driving force -dfl,Jdx must be

opposed by an equal resistive force which shall be taken to be the Stokes

viscous force 6nr1JVd. Hence,

(4.173)

One can therefore define the absolute mobility Uabs for the diffusing
particles by dividing the drift velocity by either the diffusional driving force
or the equal and opposite Stokes viscous force
_ Vd Vd I
Uabs = _ dfl/dx = 6nr1Jvd = 6nr1J
(4.174)

Now, the fundamental expression (4.169) relating the diffusion coef-

ficient and the absolute mobility can be written thus
_ D
Uabs = kT (4.175)

By equating (4.174) and (4.175), the Stokes-Einstein relation is obtained

D=~ (4.176)
6nrYJ

It links the processes of diffusion and viscous flow.

The Stokes-Einstein relation proved extremely useful in the classical
work of Perrin. Using an ultramicroscope, he watched the random walk
of a colloidal particle, and, from the mean square distance <X2> traveled
in a time t, he obtained the diffusion coefficient D from the relation (4.27)

(4.27)

The weight of the colloidal particles and their density being known, their
radius r was then obtained. Then, the viscosity 1J of the medium can be
used to obtain the Boltzmann constant

k = 6nr1JD (4.177)
T

t The hypothetical nature of the argument lies in the fact that, in diffusion, there is no
actual force exerted on the particles. Consequently, there is not the actual force-derived
component of the velocity; i.e., there is no actual drift velocity (see Section 4.2.1).
Thus, the drift velocity enters the argument only as a device.
 ION TRANSPORT IN SOLUTIONS 381

But
R
k=- (4.178)
NA
or
R
N., . =k- (4.179)

and, thus, the Avogadro number can be determined.

The use of Stokes' law also permits the derivation of a very simple
relation between the viscosity of a medium and the conventional electro-
chemical mobility Ucom ' Starting from the earlier derived equation

ii,bs = -6--
:Tr'l)
(4.174)

one substitutes for the absolute mobility the expression from (4.149)

300uconv
(4.149)

and gets the result

(4.180)

This relation shows that, owing to the Stokes viscous force, the conven-
tional mobility of an ion depends on the charge and radius t of the solvated
ion and the viscosity of a medium. The mobility given by Eq. (4.180) is
often called the Stokes mobility. It will be seen later that the Stokes mobility
is a highly simplified expression for the mobility and ion-ion interaction
effects introduce a concentration dependence \vhich is not seen in Eq. (4.180).

4.4.9. The Nernst-Einstein Equation

Now the Einstein relation (4.169) will be used to connect the transport
processes of diffusion and conduction.
The starting point is the basic equation relating the equivalent conduc-
tivity of a z :z-valent electrolyte to the conventional mobilities of the ions,
i.e., to the drift velocities under a potential gradient of 1 V cm-I,

(4.160)

t Earlier, the radius dependence of the conventional mobility was used to obtain infor-
mation of the solvation number (el Section 2.4.5).
 382 CHAPTER 4

By using the relation between the conventional and absolute mobilities,

Eq. (4.160) can be written t

(4.181)

With the aid of the Einstein relation (4.169),

and (4.182)

one can transform Eq. (4.181) into the form

A = zeoF
kT
(D D)
+ + -
(4.183)

This is one form of the Nernst-Einstein equation; from a knowledge of

the diffusion coefficients of the individual ions, it permits a calculation of
the equivalent conductivity.
A more usual form of the Nernst-Einstein equation is obtained by
multiplying numerator and denominator by the Avogadro number, in which
case, it is obvious that

(4.184)

4.4.10. Some Limitations of the Nernst-Einstein Relation t

There were several aspects of the Stokes-Einstein relation which reduced

it to being only an approximate relation between the diffusion coefficient of
an ionic species and the viscosity of the medium. In addition, there were
fundamental questions regarding the extrapolation of a law derived for
macroscopic spheres moving in an incompressible medium to a situation
involving the movement of ions in an environment of solvent molecules and
other ions. In the Nernst-Einstein relation, the factors which limit its validity
are more subtle.
An implicit but principal requirement for the Nernst-Einstein equation
to hold is that the species involved in diffusion must also be the species
responsible for conduction. Suppose now that the species M exists not only
as ions MZ+ but also as ion pairs MH AZ- of the type described in Section 3.8.1.

t For simplicity, the conversion factor (1/300) of Eq. (4.149) has been omitted here.
It must be introduced, however, when one makes numerical calculations.
Further discussion of this topic is contained in Section 6.4.6.
 ION TRANSPORT IN SOLUTIONS 383

IONS AND ION PAIRS DIFFUSE

ION PAIRS 00 NOT CONDUCT

----.CI
-----v -;jlf/ ---I-- DOES NOT RESPOND
TO FIELD

FIELD CO"
I

Fig. 4.62. The difference in the behavior

of neutral ion pairs during diffusion and
conduction.

The diffusive transport of M proceeds both through ions and ion pairs.
In the conduction process, however, the situation is different (Fig. 4.62).
The applied electric field only exerts a driving force on the charged particles.
But an ion pair as a whole is electrically neutral; it does not feel the electric
field. Thus, ion pairs are not participants in the conduction process. This
point is of considerable importance in conduction in nonaqueous media
(see Section 4.7.9).
In systems where ion-pair formation is possible, the mobility calculated
from the diffusion coefficient, lIabs = D/kT, is not equal to the mobility
calculated from the equivalent conductivity, lIabs = uconv/z;eo = (A/z;eo)F,
and, therefore, the Nernst-Einstein equation, which is based on equating
these two mobilities, may not be completely valid. In practice, one finds
a degree of nonapplicability of up to about 20%.
Another important limitation on the Nernst-Einstein equation in
electrolytic solutions may be approached through the following consider-
ations. The diffusion coefficient is in general not a constant. This has been
pointed out in Section 4.2.3, where the following expression was derived,

D = BRT(I + dIn!)
dIn c
(4.19)

It is clear that BRT is the value of the diffusion coefficient when the solution
behaves ideally, i.e.,! = 1; this ideal value of the diffusion coefficient shall
384 CHAPTER 4

be called IJO. Hence,

D = IJO( 1 + d In f )
dIn c
(4.185)
= IJO( 1 + c d ~;f)
and, making use of the Debye-Hiickellimiting law for the activity coefficient
(cf Section 3.5),
logf= - Ac!
one has
D = IJO(1 - tAd) (4.186)

an expression which shows how the diffusion coefficient varies with con-
centration. In addition, there is Kohlrausch's law
A =Ao - Ac! (4.136)

where AO is the equivalent conductivity at infinite dilution, i.e., the ideal

value.
From Eqs. (4.186) and (4.136), it is obvious that the diffusion coeffi-
cient D and the equivalent conductivity A have different dependences on
concentration (Fig. 4.63). This experimentally observed fact has an impor-
tant implication as far as the applicability of the Nernst-Einstein equation
in electrolytic solutions is concerned. For, if the equation is true at one
concentration, it cannot be true at another because the diffusion coefficient
and the equivalent conductivity have varied to different extents in going
from one concentration to the other.
The above argument brings out an important point about the limita-
tions of the Nernst-Einstein equation. It does not matter whether the dif-

0-6

0-2 0-4 0-6 0-8

rc
Fig. 4.63. The variation of the diffusion
coefficient and the equivalent conductivity
with concentration.
 ION TRANSPORT IN SOLUTIONS 385

fusion coefficient and the equivalent conductivity vary with concentration;

to introduce deviations into the Nernst-Einstein equation, D and A must
have different concentration dependences. Now, the concentration depend-
ence of the diffusion coefficient has been shown to be due to nonideality
(f #- I), i.e., due to ion-ion interactions; and it will be shown later that
the concentration dependence of the equivalent conductivity is also due to
ion-ion interactions. But it is not the existence of interactions per se which
underlies deviations from the Nernst-Einstein equation; otherwise, molten
salts and ionic crystals, in which there are strong interionic forces, would
show far more than the observed few per cent deviation of experimental
data from values calculated by the Nernst-Einstein equation. The essential
point is that the interactions must affect the diffusion coefficient and the
equivalent conductivity by different mechanisms and, thus, to different
extents. How this comes about for diffusion and conduction in solution
will be seen later.
In solutions of electrolytes, the d terms in the expressions for D and A
tend to zero as the concentration of the electrolytic solution decreases,
and the differences in the concentration variation of D and A become more
and more negligible; in other words, the Nernst-Einstein equation becomes
increasingly valid for electrolytic solutions with increasing dilution.

4.4.11. A Very Approximate Relation between Equivalent Con-

ductivity and Viscosity: Walden's Rule
The Stokes-Einstein equation (4.175) connects the diffusion coefficient
and the viscosity of the medium; the Nernst-Einstein equation (4.184) relates
the diffusion coefficient to the equivalent conductivity. Hence, by eliminating
the diffusion coefficient in these two equations, it is possible to obtain a rela-
tion between the equivalent conductivity and the viscosity of the electrolyte.
The algebra is as follows:
D=~=~A
6nr1) zp
and, therefore,
A = zPk _1_ (4.187)
6nR r1)

Since F = NAe o and kjR = IjNA' one obtains

A = zeoF _1_
6n r1)
constant
A1) = - - - (4.188)
r
386 CHAPTER 4

TABLE 4.14
Tests of Walden's rule. The product A" for Potassium Iodide in Various
Solvents at 25 0 C

Solvent A TJ A"

Sulfur dioxide t 265 0.00394 1.044

Acetonitrile 198.2 0.00345 0.684
Acetone 185.5 0.00316 0.586
Nitromethane 124.0 0.00611 0.758
Methyl alcohol 114.8 0.00546 0.627
Pyridinet 71.3 0.00958 0.682
Ethyl alcohol 50.9 0.01096 0.560
Furfural 43.1 0.01490 0.642
Acetophenone 39.8 0.01620 0.644

t o °c.
t 200C.

Hence, if the radius of the moving (kinetic) entity in conduction, i.e.,

the solvated ion, can be considered the same in solvents of various viscosities,
the following relation is obtained
zeoF
A1J = constant = -6- (4.189)
nr
This means that the product of the equivalent conductivity and the viscosity
of the solvent for a particular electrolyte at a given temperature should be a
constant (at one temperature). This is indeed what the empirical Walden's
rule states.
Some experimental data on the product A7J are presented in Table 4.14
for solutions of potassium iodide in various solvents. Walden's rule has

TABLE 4.15
The Product A" for Sodium Chloride in Various Solvents at 25 0 C

Solvent A

Water 126.39 0.00895 1.l31

Methyl alcohol 96.9 0.00546 0.529
Ethyl alcohol 42.5 0.01096 0.466
 ION TRANSPORT IN SOLUTIONS 387

some rough applicability in organic solvents. When, however, the A1}

products for a solute dissolved, on the one hand, in water and, on the
other, in organic solvents are compared, it is found that there is consider-
able discrepancy (Table 4.15). This should hardly come as a surprise; one
should expect differences in the solvation of ions in water and in organic
solvents (Section 2.8) and the resulting differences in radii of the moving
ions.

4.4.12. The Rate-Process Approach to Ionic Migration

The fundamental equation for the current density (flux of charge) as

a function of the drift velocity has been shown to be

i = ZFCVd (4.161)

Hitherto, the drift velocity has been related to macroscopic forces (e.g.,
-+ -+
the Stokes viscous force F = 6nr1} or the electric force F = zeoX) through
the relation
-+
F
Vd=-T (4.144)
m

Another approach to the drift velocity is by molecular models. The

drift velocity Vd is considered the net velocity, i.e., the difference of the
v
velocity of ions in the direction of the force field and the velocity t of ions
in a direction opposite to the field (Fig. 4.64). In symbols, one writes

(4.190)

Any velocity is given by the distance traveled divided by the time taken

FORCE FIE'LO
..

DRIFT
VELOCITY. 'I'd•
WHICH IS GIVEN
BY
- -"
v

Fig. 4.64. The drift velocity vd' can be considered made

up of a velocity"t in the direction of the force field and a
velocity t' in a direction opposite to the force field.
388 CHAPTER 4

to travel that distance. In the present case, the distance is the jump distance I,
i.e., the mean distance that an ion jumps in hopping from site to site in the
course of its directed random walk, and the time is the mean time T between
successive jumps. This mean time includes the time the ion may wait in a
"cell" of surrounding particles as well as the actual time involved in jump-
ing. Thus,
-+ Mean jump distance I
(4.191)
v = Mean time between jumps T

The reciprocal of the mean time between jumps is the net jump frequency k,
which is the number of jumps per unit of time. Hence,

Velocity = Jump distance X Net jump frequency

or
-+ -+
v = [·k (4.192)
and
+- +-
v = [·k (4.193)

For diffusion, the net jump frequency k was related to molecular

quantities by viewing the ionic jumps as a rate process (Section 4.2.15).
In this view, for an ion to jump, it must possess a certain free energy of
activation to surmount the free-energy barrier. It was shown that the net
jump frequency is given by

(4.111)

To emphasize that this is the jump frequency for a pure diffusion process,
in which case the ions are not subjected to an externally applied field, a
subscript D will be appended to the net jump frequency and to the standard
free energy of activation, i.e.,

-+kn- -kT
- e-,1G oD*/RT (4.194)
h

Now, suppose an electric field is applied such that it hinders the move-
ment of a positive ion from right to left. Then, the work that is done on
the ion in moving it from the equilibrium position to the top of the barrier
(Fig. 4.65) is the product of the charge on the ion, z+eo , and the potential
difference between the equilibrium position and the activated state, i.e.,
the position at the top of the barrier. Let this potential difference be a
 ION TRANSPORT IN SOLUTIONS 389

...-
Electric field.X
..
~ .......",
..<> : : ~-1'{'''''·''·" .,__
]
(b)
e:
n.

I IPo.iIion of jumping ion

I I I
I I I

I
f--0
r- ~L::..J Activotion distance
(0 )

I' L~
I I
...
1

...
~ I 1

.
c
I
Diffusion barrier
(C) ~

"E
..,c0

0
Ui

Position of jumping ion

Fig. 4.65. (a) As the ion moves for the jump,

it has to climb (b) the potential gradient arising
from the electric field in the electrolyte, in addi-
tion to (cl the free-energy barrier for diffusion.
To be activated, the ion has to climb the fraction
{3IX of the total potential difference IX between
the initial and final positions for the jump.

fraction fJ of the total potential difference (i.e., the applied electric field X
times the distance I) between two equilibrium sites. Then, the electrical
work done on one positive ion in making it climb to the top of the barrier,
i.e., in activating it, is equal to the charge on the ion, z+e o, times the potential
difference {3XI through which it is transported. Thus, the electrical work is
z+eofJXI per ion, or z+F{3XI per mole of ions.
The electrical work of activation corresponds to a free-energy change.
It appears therefore that there is a contribution to the total free energy of
activation due to the electrical work done on the ion in making it climb
the barrier. This electrical contribution to the free energy of activation is

(4.195)
 390 CHAPTER 4

Hence, the total free energy of activation (for positive ions moving
from right to left) is
(4.196)

Thus, in the presence of the field, the frequency of right -+ left jump is

(4.197)

or
(4.198)
-+
By a similar argument, the left -+ right jump frequency k, or the number
of jumps per second from left to right, may be obtained. There are, how-
ever, two differences: When positive ions move from left to right, (1) they
are moving with the field and therefore are helped, not hindered, by the
field, and (2) they have to climb only through a fraction 1 - {3 of the
barrier. Hence, the electrical work of activation is - [z+F(1 - (3)XI], the
minus sign indicating that the field assists the ion. Thus,

(4.199)

If the factor {3 is assumed to be t. t then fJ = I - fJ = i and Eqs.

(4.198) and (4.199) can be written

(4.200)
and
(4.201)

where, for conciseness, p is written instead of z+Flj2RT. It follows from

+- -+- -+-t-
these equations that k < k n and k > kn, or k > k.
In the presence of the field, therefore, the jumping frequency is aniso-
tropic, i.e., it varies with direction. The jumping frequency of an ion in
-+ +-
the direction of the field, k, is greater than that k against the field. When,
however, there is no field, the jump frequency k n is the same in all direc-
tions and, therefore, jumps in all directions are equally likely. This is the
characteristic of a random walk. The application of the field d~stroys the
equivalence of all directions. The walk is not quite random. The field makes
the ions more likely to more with it than against it. There is drift. In Eqs.

t This implies (el Fig. 4.65) that the energy barrier is symmetrical.
 ION TRANSPORT IN SOLUTIONS 391

(4.200) and (4.201), the kD is a random-walk term, the exponential factors

are the perturbations due to the field, and the result is a drift. The equations
are therefore a quantitative expression of the qualitative statement made in
Section 4.4.1.

Drift due = Random walk in X Perturbation

(4.202)
to field absence of field due to field

4.4.13. The Rate-Process Expression for Equivalent Conductivity

4 +-
Introducing the expressions (4.200) and (4.201) for k and k into the
equations for the component forward and backward velocities U and t,
[i.e., into Eqs. (4.192) and (4.193)] one obtains

(4.203)
and
(4.204)

where, as stated earlier,

The drift velocity Vd is obtained [cf Eq. (4.190)] by subtracting (4.203)

from (4.204), thus,
Vd = V - V = IkDe+ Px - IkDe- Px
= lk n(e+ Px - e- PX )

= 2lkD sinh pX (4.205)

The net charge transported per second across unit area, i.e., the current
density i is given by Eq. (4.161), i.e.,

(4.161)

Upon inserting the expression (4.205) for the drift velocity into Eq. (4.161),
it is clear that
i = z+Fc(2Ik D sinh pX) (4.206)

A picture of the hyperbolic sine relation between the ionic current density
and the electric field that would result from Eq. (4.206) is shown in Fig. 4.66.
The fundamental thinking used in the derivation of Eq. (4.206) has
wide applicability. Take the case of an oxide film which grows on an elec-
392 CHAPTER 4

CURRENT DENSITY

----"7'i~---- FIELD

Fig. 4.66. The hyperbolic sine relation

between the ionic current density and the
electric field according to (4.206).

tron-sink electrode (anode). All one has to do is to consider an ionic crystal

(the oxide) instead of an electrolytic solution, and all the arguments used
to derive the hyperbolic sine relation (4.206) become immediately applic-
able to the ionic current flowing through the oxide in response to the
potential gradient in the solid (Fig. 4.67). In fact, Eq. (4.206) is the basic
equation describing the field-induced migration of ions in any ionic con-
ductor.
Equation (4.206) is also formally similar to the expression for the
current density due to a charge-transfer electrodic reaction occurring under
the electric field present at an electrode-electrolyte interface.
In all these cases, two significant approximations can be made. One
is the high-field Tafel type (cf Section 8.2.10) of approximation, in which
the absolute magnitude of the exponents IpXI in Eq. (4.206), i.e., the
argument pX of the hyperbolic sine in Eq. (4.206), is much greater than

FIELD ACROSS ELECTROLYTE

Fig. 4.67. The similarity between the ionic current flowing (a)
through an oxide between an electrode and an electrolyte and
(b) through an electrolyte between two electrodes.
 ION TRANSPORT IN SOLUTIONS 393

CURRENT
DENSITY, .(

EXPONENTIAL
CURVE

FIELD X

Fig. 4.68. Under high-field conditions,

there is an exponential relation between
ionic current density and the field across
an oxide.

unity. Under this condition of pX)> I, one obtains sinh pX ""' eP'\"j2 because
one can neglect e- pT in comparison with ePT . Thus (Fig. 4.68),

(4.207)

i.e., the current density bears an exponential relation to the field. Such an
exponential dependence of current on field is commonly observed in oxide
growth, at electrode-electrolyte interfaces, but not in electrolyte solutions.
In electrolytic solutions, however, the conditions for the high-field
approximation are not often observed. The applied field X is generally
relatively small, III which case pX <{ I and the following approximation
can be used
(4.208)

and the current density in Eq. (4.206) is approximately given (Fig. 4.69) by

= ( 2£2 12kn) X (4.209)

Z+ C RT

All the quantities within the parentheses are constants in a particular elec-
trolyte, and, therefore,
i+ = constant X (4.210)

i.e., the ionic current density is proportional to the field. This is the low-
field approximation. It is, in fact, the rate-process version of Ohm's law.
An important point, however, has emerged; Ohm's law is valid only for small
394 CHAPTER 4

CURRENT
DENSITY

Fig. 4.69. Under low-field conditions.

there is a linear relation between the ionic
current density and the field in the elec-
trolyte.

fields. Of course, this was accounted for in the phenomenological treatment

of conduction where the general flux-force relation

J = A + BX + CX2 + ... (4.118)

reduced to the linear relation

J=BX (4.119)
only for small fields.

4.4.14. The Total Driving Force for Ionic Transport: The Gradient
of the Electrochemical Potential

In the rate-process view of conduction which has just been presented,

it has been assumed that the concentration is the same throughout the
electrolyte. Suppose, however, that there is a concentration gradient of a
particular ionic species, say, positively charged radiotracer ions. Further,
let the concentration vary continuously in the x direction (see Fig. 4.70),
so that, if the concentration of positive ions at x on the left of the barrier
is (c+)x. the number on the right (i.e., at x + 1) is given by

Concentration = Concentration + Rate of change of x Distance

on the right on the left c+ with distance

i.e.,
(4.211)

In this case, there will be diffusion of the tracer ions, and, therefore,
the current density i+ is not simply given by a conduction law, i.e., by
 ION TRANSPORT IN SOLUTIONS 395

FREE
ENERGY
DIFFUSION BARRIER

(0 )

I•
DISTANCE

CONCENTRATION

( b)

DISTANCE

FIELD
..
POTENTIAL

(c)

DISTANCE

Fig. 4.70. Measurement of ions under both a concen-

tration gradient and a potential gradient: (a) the free-
energy barrier for the diffusive jump of an ion, (b) the
concentration variation over the jump distance, and (c)
the potential variation over the jump distance.

Eq. (4.156), which governs the situation in the absence of a concentration

gradient. Instead, the expression for the current density has to be written
(the subscript x in (c+)x has been dropped for the sake of convenience)

(4.212)

But; and v have been evaluated as

(4.204)
396 CHAPTER 4

and
(4.203)

Under low-field conditions pX ~ I, the exponentials can be expanded

and linearized to give
(4.213)
and
(4.214)

Combining Eqs. (4.212), (4.213), and (4.214), one gets

= z+Fc)k n (1 + pX) - z+Fc+lkD(1 - pX)

- Fl2k (I - X) dc+
z+ D P dx

dc+
= 2z+Fc+1k DPX - Z+Fl2k n(l - pX) dx (4.215)

This expression can be simplified further, firstly, by applying the low-

field condition pX ~ 1. It becomes

i+ = 2z+Fc+lk DPX - z+Fl2k D ~; (4.216)

Secondly, by substituting z+Flj(2RT) for p, one has

(4.217)

and, finally, by replacing f2kD with D+ [cf. Eq. (4.113)], the result is

(4.218)

To go from the current density i+ to the flux J+ of positive tracer ions is

straightforward. Thus,

J =~= D+c+ (z FX) _ D dc+ (4.219)

+ z+F RT + + dx
 ION TRANSPORT IN SOLUTIONS 397

The second term on the right-hand side can be rewritten as

D dc+ _ Dc+ RT dc+

+ dx - RT 0 dx
Dc+ d(RT In c+)
RT dx
Dc+ d(p+o + RT In c+)
RT dx
Dc+ dp+
---- (4.220)
RT dx

since, according to the definition of the chemical potential for ideal solutions
[cf Eq. (3.54)],

In addition, from Eq. (4.9), the electric field X is equal to minus the
gradient of electrostatic potential, i.e.,

X= _ d1jJ (4.9)
dx
Hence,

(4.221)

This is an interesting result. The negative gradient of chemical potential,

dp+/dx, is known to be the driving force for pure diffusion; and - z+F(d1jJ/dx)
= z+FX, the driving force for pure conduction. But, when there is both a
chemical potential (or concentration) gradient and an electric field (-d1jJ/dx)
then the total driving force for ionic transport is the negative gradient of

zF1jJ + p+
Electrostatic Chemical
potential potential

This quantity zF1jJ + p+ could be called the electrostatic-chemical potential,

or simply the electrochemical potential, of the positive ions and is denoted
by the symbol p+. Thus,
p+ = p+ + zFcp (4.222)
398 CHAPTER 4

and the total driving force for the drift of ions is the gradient of the elec-
trochemical potential. Thus, one can write the flux 1+ of Eq. (4.221) in
the form
(4.223)

Or, by making use of the Einstein relation,

(4.169)

and the relation between absolute and conventional mobilities, i.e., t

(4.149)

one can rewrite Eq. (4.223) in the form

(uabs)+kT df-l+
RT c+ dx

(uconv )+ kT df-l+
z+eo RT c+ dx
(uconv )+ df-l +
----:::__ C -- (4.224)
z+F + dx

Expression (4.224) is known as the Nernst-Planck flux equation. It is an

important equation for the description of the flux or flow of a species under
the total driving force of an electrochemical potential. The Nernst-Planck
flux expression is useful in explaining, for example, the electrodeposition
of silver from silver cyanide ions. 1n this process, the negatively charged
[Ag(CNh]- ions travel to the negatively charged electron source or cathode,
a fact which cannot be explained by considering that the only driving force
on the [Ag(CNh]- ions is the electric field because the electric field drives
these ions away from the negatively charged electrode. If, however, the
concentration gradient of these ions in a direction normal to the electron
source is such that, in the expanded form of the Nernst-Planck equation,
I.e.,
Dc dc
1= --zFX - D - - (4.219)
RT dx

t In Eqs. (4.149) through (4.224), the 1/300 factor has been omitted. In making nu-
merical calculations, however, it must be taken into account. Thus, (4.224) is
 ION TRANSPORT IN SOLUTIONS 399

the second term is larger than the first, then the flux of the [Ag(CN)2]-
ions is opposite to the direction of the electric field, i.e., toward the nega-
tively charged electrode.

Further Reading
l. A. Einstein, Investigations on the Theory of the Brownian Movement, Methuen,
London, 1926.
2. S. Glasstone, K. J. Laidler, and H. Eyring, Theory of the Rate Processes,
McGraw-Hill Book Company, New York, 1941.
3. R. W. Gurney, Ionic Processes in Solution, Dover Publications, New York,
1953.
4. R. A. Robinson and R. H. Stokes, Electrolyte Solutions, Butterworth's Pub-
lications, Ltd., London, 1955.
5. R. H. Boyd, "Extension of Stokes Law for Ionic Motion to Include the Effect
of Dielectric Relaxation," J. Chem. Phys., 35: 19, 281 (1961).
6. R. W. Laity, J. Chem. Educ., 39: 56 (1962).
7. R. Zwanzig, '~Dielectric Friction on a Moving Ion," J. Chem. Phys., 38: 1603
(1963).
8. R. P. Feynman, Lectures on Physics, Vol. 1, Addison-Wesley Publishing Co.,
Inc., Reading, Mass, 1964.

4.5. THE INTERDEPENDENCE OF IONIC DRIFTS

4.5.1. The Drift of One Ionic Species May Influence the Drift
of Another
The processes of diffusion and conduction have been treated so far
with the simple assumption that each ionic species drifts independently of
every other one. In general, however, the assumption is not realistic for
electrolytic solutions because it presupposes the absence of ionic atmos-
pheres resulting from ion-ion interactions. One has been talking therefore
of ideal laws of ionic transport and expressed them in the Nernst-Planck
equation for the independent flux of a species i.

(4.225)

The time has come to free the treatment of ionic transport from the
assumption of the independence of the various ionic fluxes and to consider
some phenomena which depend on the fact that the drift of a species i is
affected by the flows of other species present in the solution. For it is the
whole society of ions that displays a transport process, and each individual
400 CHAPTER 4

ionic species takes into account what all the other species are doing. Ions
interact with each other through their coulombic fields, and, thus, it will
be seen that the law of electroneutrality which seeks zero excess charge in
any macroscopic volume element plays a fundamental role in phenomena
where ionic flows influence each other.
A stimulating approach to the problem of the interdependence of ionic
drifts can be developed as follows. Since different ions have different radii,
their Stokes mobilities, given by

(4.180)

must be different. What are the consequences that result from the fact that
different ionic species have unequal mobilities?

4.5.2. A Consequence of the Unequal Mobilities of Cations and

Anions, the Transport Numbers

The current density i; due to an ionic species i is related to the mobility

in the following manner [cf Eq. (4.156)]
UCOllV ,;

(4.156)

If, therefore, one considers a unit field X = I in an electrolyte solution

*
containing a z:z-valent electrolyte (i.e., z+ = L = z and c+, = c' = c),
then, since (uconv )+ (uconv )-, it follows that

(4.226)

This is a thought-provoking result. It shows that although all ions

feel the externally applied electric field to the extent of their charges, some
respond by migrating more than others. It also shows that, though the
burden of carrying the current through the electrolytic solution falls on the
whole community of ions, the burden is not shared equally among the
various species of ions. Even if there are equal numbers of the various
ions, those which have higher mobility contribute more to the communal
task of transporting the current through the electrolytic solution than the
ions handicapped by lower mobilities.
It is logical under these circumstances to seek a quantitative measure
of the extent to which each ionic species is taxed with the job of carrying
current. This quantitative measure, known as the transport number (Table
4.16), should obviously be defined by the/raction of the total current carried
 ION TRANSPORT IN SOLUTIONS 401

TABLE 4.16
Transport Number of Cations in Aqueous Solutions at 25 0 C
in 0.1 N solutions

Electrolyte HCI LiCI NaCl KCI KNOa AgNOa BaCl z

Transport number
of cation, t+ 0.83 0.32 0.39 0.49 0.5\ 0.47 0.43

by the particular ionic species, i.e.,

(4.227)

This definition requires that the sum of the transport numbers of all the
ionic species be unity for
(4.228)

Thus, the conduction current carried by the species i (e.g., Na+ ions in a
solution containing NaCI and KCI) depends upon the current transported
by all the other species. Here, then, is a clear and simple indication that
the drift of the ith species depends on the drift of the other species.
For example, consider a I : I-valent electrolyte, e.g., HCI, dissolved in
water. The transport numbers will be given byt

But ZH+ = ZCI- = 1 and CH+ = CCl- = c, and, therefore,

(4.229)

Similarly,
(4.230)

The mobilities of the H+ and Cl- ions in O.lN HCI at 25°C are

t To avoid cumbersome notation, the symbol Uconv for the conventional mobilities has
been contracted to u. The absence of a bar above the U stresses that it is not the absolute
mobility nabs.
402 CHAPTER 4

33.71 X 10-4 and 6.84 X 10-4 cm sec-IjV em-I, respectively, from which it
turns out the transport numbers of the H+ and Cl- ions are 0.83 and 0.17,
respectively. Thus, the positive ions carry a major fraction ("-' 83%) of the
current.
Now suppose that, to the HCl solution, an excess of KCl is added
so that the concentration of H+ is about 1O-3M in comparison with a K+
concentration of 1M. The transport numbers in the mixture of electrolytes
will be
CK+UK+/'L, C,U,
(4.231)
CWUH+/'L, C,Uj

(4.232)

The ratio CK+/CH+ is 103, and the ratio of mobilities is

UK + 6 X 10-4
---
UH+ 30 X 10-4 5
Hence,

which means that, although the H+ is about 5 times more mobile than the
K + ion, it carries 200 times less current. Thus, the addition of the excess
of KCL has reduced to a negligible value the fraction of the current carried
by the H + ions.
In fact, the transport number of the H+ ions under such circumstances
is virtually zero, as shown from the following approximate calculation.
(Note that the concentrations must be expressed in moles per cubic centi-
meter, not in moles per liter.)

(4.233)

Thus, the conduction current carried by an ion depends very much on

the concentration in which the other ions are present.

4.5.3. The Significance of a Transport Number of Zero

In the previous section, it has been shown that the addition of an
excess of KCl makes the fraction of the migration (i.e., conduction) current
 ION TRANSPORT IN SOLUTIONS 403

- +

Migration of
K+and cnons

Positive
electrode

NlQalive
electrode

Electronation of De - electronation of
H+ Ions CI- Ions

Fig. 4.71. A schematic diagram of the transport

processes in an electrolyte (of HCI - KCI, with an
excess of KCI) and of the reactions at the interfaces.

carried by the H+ ions tend to zero. What happens if this mixture of HCl
and KCI is placed between two electrodes and a potential difference applied
across the cell (Fig. 4.71)?
In response to the electric field developed in the electrolyte, a migration
of ions occurs and there is a conduction current in the solution. Since this
conduction current is almost completely borne by K + and Cl- ions (tII+ ~ 0),
there is a tendency for Cl- ions to accumulate near the positive electrode,
and the K + ions, near the negative electrode. If the excess negative charge
near the positive electrode and viee versa were to build up, then the resulting
field due to lack of electroneutrality (ef Section 4.3.3) would tend to bring
the conduction current to a halt. It has been argued (ef Section 4.3.4)
however that conduction (i.e., migration) currents are sustained in an elec-
trolyte because of charge-transfer reactions (at the electrode-electrolyte
interfaces), which remove the excess charge that tends to build up near the
electrodes.
In the case of the HCl + KCI electrolyte, the reaction at the positive
404 CHAPTER 4

electrode may be considered the deelectronation of the CI- ions. Further,

according to Faraday's law (ef Section 4.3.5), I g-eq of CI- ions must be
de-electronated at the positive electrode for the passage of 1 F of charge in
the external circuit. This means however that, at the other electrode, 1 g-eq
of positive ions must be involved in a reaction. Thus, either the K + or the
H + ions must react; but, by keeping the potential difference within certain
limits, one can ensure that only the H+ ions react.
There is no difficulty in effecting the reaction of the layer of H+ near
the negative electrode, but, to keep the reaction going, there must be a
flux of H+ ions from the bulk of the solution toward the negative electrode.
By what process does this flux occur? It cannot be by migration because
the presence of the excess of K + ions makes the transport number of H +
tend to zero. It is here that diffusion comes into the picture; the removal
of H+ ions by the charge-transfer reactions causes a depletion of these ions
near the electrode, and the resulting concentration gradient provokes a
diffusion of H+ ions toward the electrode.
To provide a quantitative expression for the diffusion flux JH +, one
cannot use the Nernst-Planck flux equation (4.224) because the latter
describes the independent flow of one ionic species and, in the case under
discussion, it has been shown that the migration current of the H+ ions is
profoundly affected by the concentration of the K+ ions. A simple mod-
ification of the Nernst-Planck equation can be argued as follows.
Since conduction (i.e., migration) and diffusion are the two possible t
modes of transport for an ionic species, the total flux J i must be the sum
of the conduction flux (JC)i and the diffusion flux (Jnk Thus,

(4.234)

The conduction flux is equal to l/ziF times the conduction current i, borne
by the particular species
(4.235)

and the conduction current carried by the species i is related to the total
conduction current iT = ~ ii through the transport number of the species i
[ef Eq. (4.227)]
(4.236)

t Throughout this chapter, another possible mode of transport, hydrodynamic flow, is

not considered.
 ION TRANSPORT IN SOLUTIONS 405

Hence,
(4.237)

Further, the diffusion flux (Jn)i is given byt

df-li
(I n )·I = - Bc·-
, dx (4.14)

or, approximately, by Fick's first law

(4.16)

so that the total flux of species is

(4.238)

or, approximately,
(4.239)

From these modified forms of the Nernst-Planck flux equation (4.224),

one can see that, even if ti - 4 0, it is still possible to have a flux of a species
provided there is a concentration gradient, which is often brought into
existence by interfacial charge-transfer reactions at the electrode-electrolyte
interfaces consuming or generating the species.
From the modified Nernst-Planck flux equation (4.238), one can give
a more precise definition of the transport number. If df-lj/dx = 0, in which
case dCj/dx = 0, then,
(.=
)
(ZFJ.)
_J_J
iT dPj/dx=o

(4.240)

It should be emphasized, therefore, that the transport number only pertains

to the conduction flux, i.e., to that portion of the flux produced by an
electric field, and any flux of an ionic species arising from a chemical po-
tential gradient, i.e., any diffusion flux, is not counted in its transport
number. From this definition, the transport number of a particular species

t The constant B has been shown in Section 4.4.6 to be equal to uabs/NA •

406 CHAPTER 4

can tend to zero, t i -+ 0, and, at the same time, its diffusion flux can be
finite.
This is an important point in electroanalytical chemistry where the
general procedure is to arrange for the ions which are being analyzed to
move to the electrode-electrolyte interface by diffusion only. Then, if the
experimental conditions correspond to clearly defined boundary conditions
(e.g., constant flux), the partial differential equation (Fick's second law)
can be solved exactly to give a theoretical expression for the bulk concen-
tration of the substance to be analyzed. In other words, the transport
number of the substance being analyzed must be made to tend to zero
if the solution of Fick's second law is to be applicable. This is ensured by
adding some other electrolyte in such great excess that it takes on virtually
the entire burden of the conduction current. The added electrolyte is known
as the indifferent electrolyte. It is indifferent only to the electrodic reaction
at the interface; it is far from indifferent to the conduction current.

4.5.4. The Diffusion Potential, Another Consequence of the

Unequal Mobilities of Ions
Consider that a solution of a z:z-valent electrolyte (of concentration
c moles liter-I) is instantaneously brought into contact with water at the
plane x = 0 (Fig. 4.72). A concentration gradient exists both for the
positive ions and for the negative ions. They therefore start diffusing into
the water.
Since, in general, t u+ #- u_, let it be assumed that u+ > u_. With the
use of the Einstein relation (4.169), it is clear that

and
or that

This means that the positive ions try to lead the negative ions in the
diffusion into the water. But, when an ionic species of one charge moves
faster than a species of the opposite charge, any unit volume in the water
phase will receive more ions of the faster-moving variety.
Now, compare two unit volumes (Fig. 4.73), one situated at X 2 and
the other at Xl' where X 2 > Xl' i.e., X 2 is farther than the plane of contact

t Though the subscnpt "abs" has been dropped, it is clear from the presence of a bar
over the u's that one is referring to absolute mobilities.
 ION TRANSPORT IN SOLUTIONS 407

SS TUBE

ELECTROLYTIC
SOLUTION

ELECTROLYTE
CONCENTRATION

Fig. 4.72. (a) An electrolytic solution is instanta-

neously brought into contact with water at a plane
x = 0; (b) the variation of the electrolyte concen-
tration in the container at the instant of contact.

(x = 0) of the two solutions. The positive ions are random walking faster
than the negative ions, and, therefore, the greater the value of x, the greater
is the ratio c + xl c 1"
All this is another way of saying that the center of the positive charge
tends to separate from the center of the negative charge (Fig. 4.73). Hence,
there is a tendency for the segregation of charge and the breakdown of the
law of electroneutrality.
But, when charges of opposite sign are spatially separated, a potential
difference develops. This potential difference between two unit volumes at
X 2 and Xl opposes the attempt at charge segregation. The faster-moving

positive ions face strong opposition from the electroneutrality field, and
they are slowed down. In contrast, the slower-moving negative ions are
assisted by the potential difference (arising from the incipient charge separa-
tion), and they are speeded up. When a steady state is reached, the accelera-
tion of the slow negative ions and the deceleration of the initially fast,
positive ions, resulting from the electroneutrality field that develops, exactly
compensate the inherent differences in mobilities. The electroneutrality
408 CHAPTER 4

(oj

(b)

eCl

Fig. 4.73. (a) At a time t > 0 after the electrolyte and water
are brought into contact, pure water and the electrolyte are sep-
arated by a region of mixing. In this mixing region, the c+/c_
ratio increases from right to left because of the higher mobility
of the positive ions. (b) and (c). The distance variations of the
concentrations of positive and negative ions.

field is the leveler of ionic mobilities, helping and retarding ions according
to their need so as to keep the situation as electro neutral as possible.
The conclusion that may be drawn from this analysis has quite profound
ramifications. The basic phenomenon is that, whenever solutions of differing
concentration are allowed to come into contact, diffusion occurs, there is a
tendency for charge separation due to differences between ionic mobilities,
 ION TRANSPORT IN SOLUTIONS 409

ELECTRODE

~r.T"("VT~ CONCENTRATION
DECREASES FROM LEFT TORIGKT

Fig. 4.74. A potential difference is registered

by a vacuum-tube voltmeter (VTVM) connected
to a concentration cell, i.e., to electrodes dipped
in an electrolyte, the concentration of which
varies from electrode to electrode.

and a potential difference develops across the interphase region in which

there is a transition from the concentration of one solution to the concen-
tration of the other.
This potential is known by the generic term diffusion potential. But,
the precise name given to the potential varies with the situation, i.e., with
the nature of the interphase region. If one ignores the interphase region
and simply sticks two electrodes, one into each solution in order to "tap"
the potential difference, then the whole assembly is known as a concentration
cell (Fig. 4.74). On the other hand, if one constrains or restricts the inter-
phase region by interposing a sintered-glass disk or any uncharged mem-
brane between the two solutions so that the concentrations of the two

Liquid Junction
potential
across membrone

\.
Electrode

Porous
0' hiQh
concentration
Solution membrone
low concentration

Fig. 4.75. A potential difference, the liq-

uid-junction potential, is developed across a
porous membrane introduced between two
solutions of differing concentration.
410 CHAPTER 4

solutions are uniform up to the porous material, then one has a liquid-
junction potential t (Fig. 4.75). A membrane potential is a more complicated
affair for two main reasons: (I) There may be pressure difference across the
membrane producing hydrodynamic flow of the solution; (2) the membrane
itself may consist of charged groups, some fixed and others exchangeable
with the electrolytic solution, a situation equivalent to having sources of
ions within the membrane.

4.5.5. Electroneutrality Coupling between the Drifts of Dif-

ferent Ionic Species

The picture of the development of the electroneutrality field raises a

general question concerning the flow or drift of ions in an electrolytic
solution. Is the flux of one ionic species dependent on the fluxes of the
other species? In the diffusion experiment just discussed, is the diffusion
of positive ions affected by the diffusion of negative ions? The answer to
both these questions is in the affirmative.
Without doubt, the ionic flows start off as if they were completely
independent, but it is this attempt to assert their freedom that leads to an
incipient charge separation and the generation of an electroneutrality field.
This field, which is dependent on the flows of all ionic species, curtails the
independence of anyone particular species. In this way, the flow of one
ionic species is "coupled" to the flows of the other species.
In the absence of any interaction or coupling between flows, i.e., when
the drift of any particular ionic species i is completely independent, the
flow of that species i (i.e., the number of moles of i crossing per square
centimeter per second) is described as follows.
The total driving force on an ionic species drifting, independent of
any other ionic species, is the gradient of the electrochemical potential,
dP,Jdx. In terms of this total driving force, the expression for the total
independent flux is given by the Nernst-Planck flux equation

(4.224)

When, however, the flux of the species i is affected by the flux of the

t Now that the origin of a liquid-junction potential is understood, the method of mini-
mizing it becomes clear. One simply chooses positive and negative ions with a negligible
difference in mobilities; K + and CI- ions are the usual pair. This is the basis of the so-
called" KCI salt bridge."
 ION TRANSPORT IN SOLUTIONS 411

species j through the electroneutrality field, then another modification

(cf Section 4.5.3) of the Nernst-Planck flux equations has to be made.
The modification that will now be described is a more thorough version
than that presented earlier.

4.5.6. How Does One Represent the Interaction between Ionic

Fluxes? The Onsager Phenomenological Equations

The development of an electroneutrality field introduces an interaction

between flows and makes the flux of one species dependent on the fluxes
of all the other species. To treat situations in which there is a coupling
between the drift of one species and that of another, a general formalism
will be developed. Hence, it is only when there is zero coupling or zero
interaction that one can write the Nernst-Planck flux equation

J. = - ~ c dfi; (4.224)
, ZiF' dx

Once the interaction (due to the electroneutrality field) develops, a correc-

tion term is required, i.e.,
u· dfi·
J i = - -F'
Zi
Ci -d
x
t + Coupling correction (4.241)

It is in the treatment of such interacting transport processes, or coupled

flows, that the methods of near-equilibrium thermodynamics yield clear
understanding of such phenomena, but only from a macroscopic or pheno-
menological point of view. These methods, as relevant to the present discus-
sion, can be summarized with the following series of statements:
1. As long as the system remains close to equilibrium and the fluxes
are independent, the fluxes are treated as proportional to the driving forces.
Experience (Table 4.17) commends this view for diffusion [Fick's law,
Eq. (4.16)], conduction [Ohm's law, Eq. (4.130)], and heat flow (Fourier's
law). Thus, the independent flux of an ionic species I given by the Nernst-
Planck equation (4.224) shall be written

(4.242)
~

where La is the proportionality or phenomenological constant and Fl is

the driving force.
2. When there is coupling, the flux of one species (e.g., J 1 ) is not
simply proportional to its own (or, as it is called, conjugate) driving force
412 CHAPTER 4

TABLE 4.17
Some Linear Flux- Force Laws

Phenomenon Flux of Driving force Law

Diffusion Matter J Concentration Fick

gradient dc/dx J = - Ddc/dx
Migration Charge i Potential gradient Ohm
(conduction) X= - d!p/dx i = aX
Heat conduction Heat Jheat Temperature Fourier
gradient Jheat = K dT/dx

-+
(i.e., F1 ), but it is contributed to by the driving forces on all the other
particles. In symbols,
-+
J1 = LnF1 + Coupling correction
-+
= LnF1 + Flux of 1 due + Flux of 1 due + etc.
to driving force to driving force (4.243)
on species 2 on species 3

3. The linearity or proportionality between fluxes and conjugate driving

force is also valid for the contributions to the flux of one species from the
forces on the other species. Hence, with this assumption, one can write
Eq. (4.243) in the form

(4.244)

4. Similar expressions are used for the fluxes of all the species in the
system. If the system consists of an electrolyte dissolved in water, one has
three species: positive ions, negative ions, and water. Hence, by using the
symbol + for the positive ions, - for the negative ions, and 0 for the water,
the fluxes are

(4.245)

These equations are known as the Onsager phenomenological equations.

They represent a complete macroscopic description of the interacting flows
 ION TRANSPORT IN SOLUTIONS 413

when the system is near equilibrium. t It is clear that all the "straight"
coefficients Lij' where the indices are equal, i = j, pertain to the inde-
pendent, uncoupled, fluxes. Thus LHX+ and L __ X_ are the fluxes of the
positive and negative ions, respectively, when there are no interactions.
All other cross terms represent interactions between fluxes, e.g., L+_X_
represents the contribution to the flux of positive ions from the driving
force on the negative ions.
5. What are the various coefficients Lij? Onsager put forward the
helpful reciprocity relation. According to this, all symmetrical coefficients
are equal, i.e.,
(4.246)

This principle has the same status in nonequilibrium thermodynamics as

the law of conservation of energy has in classical thermodynamics; it has
not been disproved by experience.

4.5.7. An Expression for the Diffusion Potential

The expression for the diffusion potential can be obtained in a straight-
forward, though hardly brief, manner by using the Onsager phenomeno-
logical equations to describe the interaction flows.
Consider an electrolytic solution consisting of the ionic species MH
and AZ- and the solvent. When a transport process involves the ions in the
system, there are two ionic fluxes J+ and J_. Since, however, the ions are
solvated, the solvent also participates in the motion of ions, and, hence,
there is also a solvent flux J o .
If, however, the solvent is considered fixed, i.e., the solvent is taken as
the coordinate system or the frame of reference, t then one can consider
ionic fluxes relative to the solvent. Under this condition, J o = 0, and one
has only two ionic fluxes. Thus, one can describe the interacting and inde-
pendent ionic drifts by the following equations

(4.247)

(4.248)

The straight coefficients L++ and L__ represent the independent flows;
and the cross coefficients L+_ and L_+, the coupling between the flows.

t If the system is not near equilibrium, the flows are no longer proportional to the driving
forces [ef Eq. (4.243)].
t Coordinate systems are chosen for convenience.
414 CHAPTER 4

/
j /
ri G£
: +;0..,
I ____ _
-
NEGATIVE CHARGE FLOWING
IN" Z_'3J_

/
POSITIVE CHARGE
FLOWING IN • Z.:1J.

Fig. 4.76. According to the principle of electroneu-

trality as applied to the fluxes. the flux of positive ions
into a volume element must be equal to the flux of
negative ions into the volume element. so that the total
negative charge is equal to the total positive charge.

The important step in the derivation of the diffusion potential is the

statement that, under conditions of steady state, the electroneutrality field
sees to it that the quantity of positive charge flowing into a volume element
is equal in magnitude but opposite in sign to the quantity of negative charge
flowing in (Fig. 4.76). That is,

(4.249)

For convenience, z+F and LF are written as q+ and q_, respectively.

Now the expressions (4.247) and (4.248) for J+ and J_ are substituted in
Eq. (4.249)

or
(4.250)

Using the symbols

(4.251)
and
(4.252)
one has
(4.253)
-+ -+
But what are the driving forces F+ and F_ for the independent flows
of the positive and negative ions? They are the gradients of electrochemical
potential (cf Section 4.4.14)

(4.254)

and
(4.255)
 ION TRANSPORT IN SOLUTIONS 415

With these expressions, Eq. (4.253) becomes

or
d1p p+ dfl+ + p_ dfl-
(4.256)
dx p+q+ + p_q_ dx p+q+ + p_q_ dx

It can be shown, however, that (ef Appendix 4.4)

(4.257)

and
(4.258)

where t+ and L .are the transport numbers of the positive and negative ions.
By making use of these relations, Eq. (4.256) becomes

-d1p
-=
dx
t+
-dfl+
z+F dx
-+ L
-dfl-
LF dx
-
= ~~ dfli (4.259)
ZiF dx

The negative sign before the electric field shows that it is opposite in direc-
tion to the chemical potential gradients of all the diffusing ions.
If one considers (Fig. 4.77) an infinitesimal length dx parallel to the
direction of the electric- and chemical-potential fields, one can obtain the
electric-potential difference d1p and the chemical-potential difference dfl

~POTENTIAL DIFFERENCE, d '" , ACROSS

I LAMINA OF THICKNESS,Clx

r- -- ELECTROLYTE OF
ACTIVITY, a.

ELECTROLYTE OF
~
INTERPHASE
ACTIVITY,OI REGION

Fig. 4.77. Schematic representation of a dx-

thick lamina in the interphase region between
two electrolytes of activities 81 and 811'
416 CHAPTER 4

across the length dx

_ ~ ti
_ -dtp dx-~---- dPi d x (4.260)
dx ZiF dx
or
1 t·
- dtp =- ~ - ' dp (4.261)
F Zi '

RT t·
=--~-' dIna· (4.262)
F Zi t

This is the basic equation for the diffusion potential. It has been derived
here on the basis of a realistic point of view, namely, that the diffusion
potential arises from the nonequilibrium process of diffusion.
There is however another method t of deriving the diffusion potential.
One takes note of the fact that, when a steady-state electro neutrality field
has developed, the system relevant to a study of the diffusion potential
hangs together in a fine and delicate balance. The diffusion flux is exactly
balanced by the electric flux; the concentrations and the electrostatic
potential throughout the interphase region do not vary with time. (Re-
member the derivation of the Einstein relation, cf Section 4.4.) In fact,
one may turn a blind eye to the drift and pretend that the whole system is
in equilibrium.
On this basis, one can simply equate to zero the sum of the electrical
and diffusional work of transporting ions across a lamina dx of the inter-
phase region (Fig. 4.78). If one equivalent of charge (both positive and
negative ions) is taken across this lamina, the electrical work is F dtp. But
this one equivalent of charge consists of t+lz+ moles of positive ions and
elL moles of negative ions. Hence, the diffusional work is dp+ per mole,
or (t+/L) dp+ per t+lz+ moles of positive ions and (elL) dp_ per elL
moles of negative ions. Thus,

(4.263)

or
1 t·
- dtp =- ~ - ' dPi (4.261)
F Zi

RT t·
= -- ~ -' dIn a· (4.262)
F Zi t

t This method is based on Thomson's hypothesis according to which it is legitimate

to apply equilibrium thermodynamics to the reversible parts of a steady-state, non-
equilibrium process.
 ION TRANSPORT IN SOLUTIONS 417

(a)

Fig. 4.78. The sum of (a) the electrical work

F dip and (b) the diffusional work (t+/z+)d{t+ +
(L/L) d{t_ of transporting one equivalent of
ions across a dx-thick lamina in the interphase
region, is equal to zero.

4.5.8. The Integration of the Differential Equation for Diffusion

Potentials: The Planck-Henderson Equation

An equation has been derived for the diffusion potential [cf Eq.
(4.261)], but it is a differential equation relating the infinitesimal potential
difference dtp developed across an infinitesimally thick lamina dx in the
interphase region. What one experimentally measures, however, is the total
potential difference L1tp = tp0 - tpl across a transition region extending from
x = 0 to x = I (Fig. 4.79). Hence, to theorize about the measured potential
differences, one has to integrate the differential equation (4.261); i.e.,

-L1tp = tp0 - tpl = -

1 ~
IX-I-' --'
t.. df1'
dx
F i x-o Zi dx

= RT ~ I-I ~ dlnai dx
F i x-o Zi dx

=-~
RT IX- 1 Ii
----dx
1 d(fiCi) (4.264)
F i x-o Zi J.ici dx

Here lies the problem. To carry out the integration, one must know:
1. How the concentrations of all the species vary in the transition
region.
418 CHAPTER 4

POTENTIAL DIFFERENOE" '-,. L.

DEVELOPED ACROSS INTERPHASE
REGION

ELEOTROLYTE OF CONOENTRATION
Ci (L)
ELECTROLYTE
OF CONCENTRATION
Ci(O)

Fig. 4.79. The measured quantity is the total potential differ-

ence tp0 - tpl across the whole interphase region between elec-
trolytes of differing concentration ci (0) and ci (I).

2. How the activity coefficients Ii vary with Ci'

3. How the transport number varies with Ci'

The general case is too difficult to solve analytically, but several special
cases can be solved. For example (Fig. 4.80), the activity coefficients can
be taken as unity, fi = I-ideal conditions; the transport numbers Ii can
be assumed constant; and a linear variation of concentrations with distance
can be assumed. The last assumption implies that the concentration Ci(X)
of the ith species at x is related to its concentration Ci(O) at x = 0 in the
following way
(4.265)
and
dc· c·(l) - c·(O)
__
t = Constant = k. = t , (4.266)
dx ' I

With the aid of these assumptions, the integration becomes simple. Thus,
with Ii -::FI(x), Ii = I, and Eqs. (4.265) and (4.266), one has in (4.264)

(4.267)
 ION TRANSPORT IN SOLUTIONS 419

CONCENTRATION

Fig. 4.80. In the derivation of the Planck-

Henderson equation, a linear variation of
concentration is assumed in the interphase
region which commences at x = 0 and ends
at x = I.

This is known as the Planck-Henderson equation for diffusion or Iiquid-

junction potentials.
In the special case of a z:z-valent electrolyte, z+ = L = z and
c+ = C = c, Eq. (4.267) reduces to

RT c(l)
-,11p = F (t+ - L) In c(O) (4.268)

and, since t+ +L = I,
RT c(l)
-,11p = (2/+ - I) F In c(O) (4.269)

In the highly simplified treatment of the diffusion potential which has

just been presented, several drastic assumptions have been made. The one
regarding the concentration variation within the transition region can be
avoided. One may choose a more realistic concentration versus distance
relationship either by thinking about it in more detail or by using experi-
mental knowledge on the matter. Similarly, instead of assuming the activity
coefficient to be unity, one can feed in the theoretical or experimental
concentration dependence of the activity coefficients. Of course, the in-
troduction of nonideality (Ii -#- I) makes the mathematics awkward; in
principle, however, the problem is understandable.
But what about the assumption of the constancy of the transport
number? Is this reasonable? In the case of a z:z-valent electrolyte, the
transport number depends on the mobilities

and

Thus, the constancy of the transport numbers with concentration depends

420 CHAPTER 4

on the degree to which the mobilities vary with concentration. That is

something to be dealt with in the model-oriented arguments of the next
section.

Further Reading
1. M. Planck, Ann. Physik, 40: 561 (1890).
2. P. Henderson, Z. Physik. Chem. (Leipzig), 59: 118 (1907); and 63: 325 (1908).
3. L. Onsager, Phys. Rev., 37: 405 (1931); and 38: 2265 (1931).
4. J. Meixner, Ann. Physik, 39: 333 (1941).
5. K. G. Denbigh, Thermodynamics of the Steady State, John Wiley & Sons,
Inc., New York, 1951.
6. R. A. Robinson and R. H. Stokes, Electrolyte Solutions, Butterworths' Pub-
lications, Ltd., London, 1955.
7. B. Baranowski, Bull. A cad. Polon. Sci., Ser. Sci. Chim., 8: 609 (1960).
8. A. Katchalsky and P. F. Curran, Non-Equilibrium Thermodynamics in Bio-
physics, Harvard University Press, Cambridge, Mass., 1965.
9. N. Lakshminarayanaiah, Chem. Rev., 65: 491 (1965).

4.6. THE INFLUENCE OF IONIC ATMOSPHERES ON IONIC

MIGRATION

4.6.1. The Concentration Dependence of the Mobility of Ions

In the phenomenological treatment of conduction (Section 4.2.12),
it has been stated that the equivalent conductivity A varies with the con-
centration c of the electrolyte according to the empirical law of Kohlrausch
[Eq. (4.136)]
A =Ao - Act (4.136)

where A is a constant, and AO is the pristine or ungarbled value of equivalent

conductivity, i.e., the value at infinite dilution.
The equivalent conductivity, however, has been related to the con-
ventional electrochemical mobilities t u+ and u_ of the current-carrying ions
by the following expression
(4.160)

from which it follows that

(4.270)

t To avoid cumbersome notation, the conventional mobilities are written in this section
without the subscript "conv," i.e., one writes U+ instead of (uconv)+"
 ION TRANSPORT IN SOLUTIONS 421

where the uO's are the conventional mobilities (i.e., drift velocities under a
field of I V cm-I ) at infinite dilution. Thus, Eq. (4.l60) can be written as

(4.271)

or it can be split up into two equations

(4.272)
and
(4.273)

What is the origin of this experimentally observed dependence of ionic

mobilities on concentration? Equations (4.272) and (4.273) indicate that,
the more ions there are per unit volume, the more they diminish each other's
mobility. In other words, at appreciable concentrations, the movement of
any particular ion does not seem to be independent of the existence and
motions of the other ions, and there appear to be forces of interaction
between ions. This coupling between the individual drifts of ions has already
been recognized, but now the discussion is intended to be on an atomistic,
rather than a phenomenological, level. The interactions between ions can
be succinctly expressed through the concept of the ionic cloud (Chapter 3).
It is thus necessary to analyze and incorporate ion-atmosphere effects into
the zero-approximation atomistic picture of conduction (Section 4.4) and,
in this way, understand how the mobilities of ions depend on the concentra-
tion of the electrolyte.
Attention should be drawn to the fact that there has been a degree
of inconsistency in the treatments of ionic clouds (Chapter 3) and the
elementary theory of ionic drift (Section 4.4.2). When the ion atmosphere
was described, the central ion was considered-from a time-average point
of view-at rest. To the extent that one seeks to interpret the equilibrium
properties of electrolytic solutions, this picture of a static central ion is
quite reasonable. This is because, in the absence of a spatially directed field
acting on the ions, the only ionic motion to be considered is random walk,
the essential characteristic of which is that the mean distance traveled by
an ion (not the mean square distance, see Section 4.2.5) is zero. The central
ion can therefore be considered to remain where it is, i.e., to be at rest.
When, however, the elementary picture of ionic drift (Section 4.4.2)
was sketched, the ionic cloud around the central ion was ignored. This
approximation is justified only when the ion atmosphere is so tenuous
that its effects on the movement of ions can be neglected. This condition
422 CHAPTER 4

of extreme tenuosity (in which there is a negligible coupling between ions)

obtains increasingly as the solution tends to infinite dilution. Hence, the
simple, unclouded picture of conduction (Section 4.4) is valid only at infinite
dilution.
To summarize the duality of the treatment so far: When the ion at-
mosphere was treated in Chapter 3, the motion of the central ion was ignor-
ed and, therefore, only equilibrium properties fell within the scope of
analysis; when the motion of the central ion under an applied electric
field was considered, the ionic cloud (which is a convenient description of
the interactions between an ion and its environment) was neglected and,
therefore, only the infinite dilution conduction could be analyzed. Thus,
a unified treatment of ionic atmospheres around moving ions is required.
The central problem is: How does the interaction between an ion and its
cloud affect the motion of the ion?

4.6.2. Ionic Clouds Attempt to Catch Up with Moving Ions

In the absence of a driving force (e.g., an externally applied electric

field), no direction in space from the central ion is privileged. The coulombic
field of the central ion has spherical symmetry, and, therefore, the probability
of finding, say, a negative ion at a distance r from the reference ion is the
same irrespective of the direction in which the point r lies. On this basis,
it was shown that the ionic cloud was spherically symmetrical (cf Section
3.8.2).
When, however, the ions are subject to a driving force (be it an electric
field, a velocity field due to the flow of an electrolyte, or a chemical-potential
field producing diffusion), one direction in space becomes privileged. The
distribution function (which is a measure of the probability of finding an
ion of a certain charge in a particular volume element) has to be lopsided,
or asymmetrical. The probability depends not only on the distance of the
volume element from the central ion but also on the direction in which the
volume element lies in relation to the direction of ionic motion. The pro-
cedure of Chapter 3 no longer applies. One cannot simply assume a Boltz-
mann distribution and, for the work done to bring an ion (of charge z;e o)
to the volume element under consideration, use the electrostatic work
Zieo'IjJ because the electrostatic potential 'IjJ was, in the context of Chapter 3,
a function of r only and one would then imply a symmetrical distribution
of function.
The rigorous but, unfortunately, mathematicaIly difficult approach to
the problem of ionic clouds around moving ions is to seek the asymmetrical-
424 CHAPTER 4

/ EGG-SHAPED IONIC
./ CLOUD

Fig. 4.81. The egg-shaped ionic cloud

around a moving central ion.

tween the ion and its cloud. The development of an electric force between
a moving ion and its lagging atmosphere means that the ion is then subject
to an electric field. Since this field arises from the continual relaxation (or
decay) of the cloud behind the ion and its buildup in front of the ion, it
is known as a relaxation field. Notice, however, that the centers of charge
of the ion and of the cloud lie on the path traced out by the moving ion
(Fig. 4.82); and, consequently, the relaxation field generated by this charge
separation acts in a direction precisely opposite to the direction of the
driving force on the ion (e.g., the externally applied field). Hence, a moving
ion, by having an egg-shaped ionic cloud, carries along its own "portable"
field of force, the relaxation field, which acts to retard the central ion and
decrease its mobility compared with that which it would have were it only
pulled on by the externally applied field and retarded by the Stokes force
[the zeroth-order theory of conductance, see Eq. (4.180)].

4.6.4. A Second Braking Effect of the Ionic Cloud on the Central

Ion: The Electrophoretic Effect
The externally applied electric field acts not only on the central ion
but also on its oppositely charged cloud. Consequently, the ion and its
atmosphere tend to move in opposite directions.
This poses an interesting problem. The ionic atmosphere can be con-
sidered a charged sphere of radius X-I (Fig. 4.83). Under the action of an
electric field, the charged sphere moves. The thickness of the ionic cloud

CEN T EFI OF CE R OF
POS ITIVE CHARGE NEGATIVE CHARGE

Fig. 4.82. The centers of charge of the ion and of

the cloud lie on the path of the drifting ion.
 ION TRANSPORT IN SOLUTIONS 423

distribution functions and then work out the implications of such functions
for the electric fields developed among moving ions.
A simpler approach will be followed here. This is the relaxation ap-
proach. The essence of a relaxation analysis is to consider a system in one
state, then perturb it slightly with a stimulus, and analyze the time depend-
ence of the response of the system to the stimulus. (It will be seen later
that relaxation techniques are much used in modern studies of the mech-
anism of electrode reactions.)
Consider therefore the spherically symmetrical ionic cloud around a
stationary central ion. Now, let the stimulus of a driving force displace the
reference ion in the x direction. The erstwhile spherical symmetry of the
ion atmosphere can be restored only if its contents (the ions and the solvent
molecules) immediately readjust to the new position of the central ion.
This is possible only if the movements involved in restoring spherical
symmetry are instantaneous, i.e., if no frictional resistances are experienced
in the course of these movements. But the readjustment of the ionic cloud
involves ionic movements which are rate processes. Hence, a finite time is
required to reestablish spherical symmetry.
Even if this time were available, spherical symmetry would obtain only
if the central ion did not move away still farther while the ionic cloud was
trying to readjust. But, under the influence of the externally applied field,
the central ion just keeps moving on, and its ionic atmosphere never quite
catches up. It is as though the part of the cloud behind the central ion is
"left standing." This is because its reason for existence (the field of the
central ion) has deserted it and thermal motions try to disperse this part
of the ionic cloud. In front of the central ion, the cloud is being continually
built up. When ions move, therefore, one has a picture of the ions' losing
the part of the cloud behind them and bulding up the cloud in front of them.

4.6.3. An Egg-Shaped Ionic Cloud and the "Portable" Field on

the Central Ion

The constant lead which the central ion has on its atmosphere means
that the center of charge of the central ion is displaced from the center of
charge of its cloud. The first implication of this argument is that the ionic
cloud is no longer spherically symmetrical around the moving central ion.
It is egg-shaped (see Fig. 4.81).
A more serious implication is that, since the center of charge on a
drifting central ion does not coincide with the center of charge of its op-
positely charged (egg-shaped) ionic cloud, an electrical force develops be-
 ION TRANSPORT IN SOLUTIONS 425

IONIC CLOUD CHASGED SPHERE

OF RADIUS ~-I

IS EQUIVALENT TO

Fig. 4.83. The ionic atmosphere can be considered

a charged sphere of radius le l .

in a millimolar solution of a 1: I-valent electrolyte is about 100 A (see

Table 3.2). One is concerned, therefore, with the migration of a fairly
large "particle" under the influence of the electrical field. The term electro-
phoresis is generally used to describe the migration of particles of colloidal
dimensions (10 to 10,000 A) in an electric field. It is appropriate, therefore,
to describe the 'migration of the ionic cloud as an electrophoretic effect.
The interesting point, however, is that, when the ionic cloud moves,
it tries to carry along its entire baggage, the ions and the solvent molecules
constituting the cloud plus the central ion. Thus, not only does the moving
central ion attract and try to keep its cloud (the relaxation effect), but the
moving cloud also attracts and tries to keep its central ion by means of a
-+
force which is then termed the electrophoretic force FE'

4.6.5. The Net Drift Velocity of an Ion Interacting with Its At-
mosphere
In the elementary treatment of the migration of ions, it was assumed
that the drift velocity of an ion was determined solely by the electric force
-+
F arising from the externally applied field.
When, however, the mutual interactions between an ion and its cloud
were considered, it turned out (Sections 4.6.3 and 4.6.4) that there were
two other forces operating on an ion. These extra forces consisted of: (1) the
-+
relaxation force FR resulting from the distortion of the cloud around a
-+
moving ion, and (2) the electrophoretic force FE arising from the fact that
the ion shares in the electrophoretic motion of its ionic cloud. Thus, in a
rigorous treatment of the migrational drift velocity of ions, one must
-+
consider a total force Ftotal, which is the resultant of that due to the applied
electric field together with the relaxation and electrophoretic forces (Fig.
4.84)
(4.274)
426 CHAPTER 4

Relaxation retarding
force

Electrophoretic
retarding force

Electric driving farce on Charged sphere

ion arising from externally- representing ioniC cloud
applied field

Fig. 4.84. The ion drift due to a net force

which is a resultant of the electric driving force
and two retarding forces, the relaxation and
electrophoretic forces.

The minus sign is used because both the electrophoretic and relaxation
forces act in a direction opposite to that of the externally applied field.
Since an ion is subject to a resultant, or net, force, its drift velocity,
too, must be a net drift velocity resolvable into components. Further, since
each component force should produce a component of the overall drift
velocity, there must be three components of the net drift velocity. The first
component, which shall be designated vO, is the direct result of the externally
applied field only and excludes the influence of interactions between the
ion and the ionic cloud; the second is the electrophoretic component VE
and arises from the participation of the ion in the electrophoretic motion
of its cloud; and, finally, the third component is the reaxation field com-
ponent VR originating from the relaxation force which retards the drift of
the ion. Since the electrophoretic and relaxation forces act in an opposite
sense to the externally applied electric field, it follows that the electrophor-
etic and relaxation components must diminish the overall drift velocity
(Fig. 4.85), i.e.,
(4.275)

The next step is to evaluate the electrophoretic and relaxation com-

ponents of the net drift velocity of an ion .

-..
VR _
vE
-v;-
Fig. 4.85. The components of the overall
drift velocity.
 ION TRANSPORT IN SOLUTIONS 427

4.6.6. The Electrophoretic Component of the Drift Velocity

The electrophoretic component VE of the drift velocity of an ion is equal
to the electrophoretic velocity of its ionic cloud because the central ion
shares in the motion of its cloud. If one ignores the asymmetry of the ionic
cloud, a simple calculation of the electrophoretic velocity VE can be made.
The ionic atmosphere is accelerated by the externally applied electric
force t zeoX/300, but it is retarded by a Stokes viscous force. When the cloud
attains a steady-state electrophoretic velocity VE, then the viscous force is
exactly equal and opposite to the electric force driving the. cloud
X
zeo 300 = Stokes' force on cloud (4.276)

The general formula for Stokes' viscous force is 6nrrjV, where r and v
are the radius and velocity of the moving sphere. In computing the viscous
force on the cloud, one can substitute ,,-1
for r and VB for V in Stokes'
formula. Thus,
(4.277)

from which it follows that

zeo X
VE = 6n,,-I'fj 300
(4.278)

This, therefore, is the expression for the electrophoretic contribution to the

drift velocity of an ion.

4.6.7. The Procedure for Calculating the Relaxation Compo-

nent of the Drift Velocity
From the familiar relation [cf Eq. (4.146)],

Velocity = Absolute mobility x Force

it is clear that the relaxation component VR of the drift velocity of an ion

-+
can be obtained by substituting for the relaxation force FR in

(4.279)

The problem, therefore, is to evaluate the relaxation force.

t The factor 1/300 is introduced to permit the field X to be expressed in volts per cen-
timeter rather than in electrostatic units of potential per centimeter.
428 CHAPTER 4

Since the latter arises from the distortion of the ionic cloud, one must
derive a relation between the relaxation force and a quantity characterizing
the distortion. It will be seen that the straightforward measure of the
asymmetry of the cloud is the distance d through which the center of charge
of the ion and the center of charge of the cloud are displaced.
But the distortion d of the cloud itself depends on a relaxation process
in which the part of the cloud in front of the moving ion is being bult up
and the part at the back is decaying. Hence, the distortion d and, therefore,
-+
the relaxation force F R must depend on the time taken by a cloud to relax,
or decay.
Thus, it is necessary firstly to calculate how long an atmosphere would
take to decay, then to compute the distortion parameter d, and finally to
-+
obtain an expression for the relaxation force FR' Once this force is evaluated,
it can be introduced into Eq. (4.279) for the relaxation component VEt of
the drift velocity.

4.6.8. How Long Does an Ion Atmosphere Take to Decay?

An idea of the time involved in the readjustment of the ionic cloud
around the moving central ion can be obtained by a thought experiment
suggested by Debye (Fig. 4.86). Consider a static central ion with an
equilibrium, spherical ionic cloud around it. Let the central ion suddenly
be discharged. This perturbation of the ion-ionic cloud system sets up a
relaxation process. The ionic cloud is now at the mercy of the thermal
forces which try to destroy the ordering effect previously maintained by
the central ion and responsible for the creation of the cloud.
The actual mechanism by which the ions constituting the ionic at-
mosphere are dispersed is none other than the random-walk process de-
scribed in Section 4.2. Hence, the time taken by the ionic cloud to relax or
disperse may be estimated by the use of the Einstein-Smoluchowski relation
(Section 4.2.6)
(4.27)

But what distance x is to be used? In other words, when can the ionic
cloud be declared to have dispersed or relaxed? These questions may be
answered by recalling the description of the ionic atmosphere where it was
stated that the charge density in a dr-thick spherical shell in the cloud
declines rapidly at distances greater than the Debye-Hiickel length ,,-1.
Hence, if the ions diffuse to a distance ,,-1, the central ion can be stated
to have lost its cloud, and the time taken for this diffusion provides an
 ION TRANSPORT IN SOLUTIONS 429

CHARGED CENTRAL
ION
(0 )
I ONIC CLOUD

CENTRAL ION
DISCHARGED AT t:O

( b)
IONIC CLOUD

IONIC CLOUD
DISPERSED AFTER
TIME 'tR

(C)
DISCHARGED
CENTRAL ION

Fig. 4.86. DeJ:>ye's thought experiment to calcu-

late the time for the ion atmosphere to relax: (a)
the ionic cloud around a central ion; (b) at t = 0,
the central ion is discharged; and (c) after time
TR , the ion atmosphere has relaxed or dispersed.

estimate of the relaxation time OR' One has, by substituting X-I III the
Einstein-Smoluchowski relation [Eq. (4.27)]
(x- 1)2
tR =---us- (4.280)

which, with the aid of the Einstein relation D = iiabskT [Eq. (4.169)], can
be transformed into the expression

(4.281)

4.6.9. The Quantitative Measure of the Asymmetry of the Ion-

ic Cloud around a Moving Ion
To know how asymmetric the ionic cloud has become owing to the
relaxation effect, one must calculate the distance d through which the
central ion has moved in the relaxation time t R' This is easily done by
430 CHAPTER 4

multiplying the relaxation time T R by the velocity pO which the central ion
acquires from the externally applied electric force, i.e.,

(4.282)

By substituting the expression (4.281) for the relaxation time T R, Eq. (4.282)
becomes
(4.283)

The center of charge of the relaxing ionic cloud coincides with the
original location of the central ion; in the meantime, however, the central
ion and its center of charge moves through a distance d. Hence, the centers
of charge of the central ion and its ionic cloud are displaced through the
distance d, which therefore is a quantitative measure of the egg-shapedness
of the ion atmosphere around a moving ion.

4.6.10. The Magnitude of the Relaxation Force and the Relax-

ation Component of the Drift Velocity
Consider first a static central ion. The ion may exert an electric force
on the cloud, and vice versa; but, at first, the net force is zero because
of the spherical symmetry of the cloud around the static central ion.
When the central ion moves, it can be considered at a distance d from
the center of its cloud. Now, the net force due to the asymmetry of the
cloud is nonzero. A rough calculation of the force can be made as follows.
The relaxation force is zero when the centers of charge of the ion and
its cloud coincide, and it is nonzero when they are separated. So let it be
assumed in this approximate treatment that the relaxation force is propor-
tional to d, i.e., proportional to the distance through which the ion has
moved from the original center of charge of the cloud. On this basis, the
-+
relaxation force FR will be given by the maximum total force of the at-
mosphere on the central ion, i.e., z 2eo2/[c(x- I )2], t multiplied by the fraction
of the radius of the cloud through which the central ion is displaced during
it motion under the external field, i.e., d/x- l . Hence, the relaxation force is

(4.284)

t This total force is obtained by considering the ionic cloud equivalent to an equal and
opposite charge placed at a distance X-I from the central ion. Then, Coulomb's law
for the force between these two charges gives the result z2eu'/[E(x- I )2].
 ION TRANSPORT IN SOLUTIONS 431

and, using Eq. (4.283) for d, one has

Z2eo2y" tfl
(4.285)
2c:kT a~bs

Since, however, the velocity tfl arises solely from the externally applied
field and excludes the influence of ion-ion interactions, the ratio tfl/agbs
is equal to the applied electric force

tfl _ F _ zeo X (4.286)

a~bs- - 300

On inserting this into Eq. (4.285), it turns out that

(4.287)

In the above treatment of the relaxation field, it has been assumed that
the only motion of the central ion destroying the spherical symmetry of
the ionic cloud is motion in the direction of the applied external field.
This latter directed motion is in fact a drift superimposed on a random
walk. But the random walk is, itself, a series of motions, and these motions
are random in direction. Thus, the central ion exercises an erratic, rather
than a consistent, leadership on its atmosphere.
Onsager considered the effect that this erratic character of the leader-
ship would have on the time-average shape of the ionic cloud and therefore
on the relaxation field. His final result differs from Eq. (4.287) in two
respects: (I) Instead of the numerical factor t. there is a factor t; and (2) a
correction factor w/2z 2 has to be introduced, the quantity w being given by

(4.288)

in which

(4.289)

where A+ and L are related to mobilities of cation and anion, respectively.

For symmetrical or z:z-valent electrolytes, the expression for q reduces to 1.
432 CHAPTER 4

TABLE 4.18
Value of w for Different Types of Electrolytes

Type of electrolyte w

1 :1 0.5859
2:2 2.3436

and that for w becomes (Table 4.18)

1
W=Z2----- (4.290)
1+ (1/VT)
Thus, a more rigorous expression for the relaxation force is
-+ Z3 eo3 x w X
FR = 3EkT 2Z2 300

zeo3xw X
----- (4.291)
6EkT 300

Substituting the expression (4.291) for the relaxation force in the

equation (4.279) for the relaxation component of the drift velocity, one gets

(4.292)

Further, from the definition (4.149) of the conventional mobility,

0_ 0 _;;() zeo
U - Uconv - Uabs 300

Eq. (4.292) becomes

(4.293)

4.6.11. The Net Drift Velocity and Mobility of an Ion Subject to

lon-Ion Interactions

Now that the electrophoretic and relaxation components in the drift

velocity of an ion have been evaluated, they can be introduced into Eq.
 ION TRANSPORT IN SOLUTIONS 433

(4.275) to give
(4.275)

If one divides throughout by X, then, according to the definition of the

conventional mobility Uconv = U = v/X, one has

(4.294)

An intelligent inspection of expression (4.294) shows that the mobility

U of ions is not a constant independent of concentration. Thus, it depends

on the Debye-Hiickel reciprocal length ". But this parameter" is a function

of concentration (see Eq. 3.84). Hence, Eq. (4.294) shows that the mobility
of ions is a function of concentration, as was suspected (Section 4.6.1)
on the basis of the empirical law of Kohlrausch.
As the concentration decreases, ,,-1 increases and " decreases, as can
be seen from Eq. (3.84). In the limit of infinite dilution (c -+ 0), ,,-1 -+ 00
or" -+ O. Under these conditions, the second and third terms in Eq. (4.294)
drop out, which leaves
Ulimit, c-+o = UO (4.295)

The quantity UO is therefore the mobility at infinite dilution and can be

TABLE 4.19
Transport Numbers of Cations in Aqueous Solutions at 25 0 C

Concentration HCI LiCI NaCI KCI

O.OIN 0.8251 0.3289 0.3918 0.4902

0.02 0.8266 0.3261 0.3902 0.4901
0.05 0.8292 0.3211 0.3876 0.4899
0.1 0.8314 0.3168 0.3854 0.4898
0.2 0.8337 0.3112 0.3821 0.4894
0.5 0.300 0.4888
1.0 0.287 0.4882
434 CHAPTER 4

considered given by the expression for the Stokes mobility (Section 4.4.8),
i.e.,
tfl =_1_ ~ (4.180)
300 6nT'TJ

To go back to the question which concluded the previous section (i.e.,

4.5), it is now clear that, since transport numbers depend on ionic mobilities,
which have now been shown to vary with concentration, the transport
number must itself be a concentration-dependent quantity (Table 4.19).
However, it is seen that this variation is a small one.

4.6.12. The Debye-Huckel-Onsager Equation

Now, the equivalent conductivity A of an electrolytic solution is simply

related to the mobilities of the constituent ions [Eq. (4.160))

Thus, to obtain the equivalent conductivity, one has only to write

down the expression for the mobilities of the positive and negative ions,
multiply both the expressions by the Faraday E, and then add up the two
expressions. The result is

(4.296)

For a symmetrical electrolyte, z+ = L = Z or z+ +L = 2z, and, there-

fore, Eq. (4.296) reduces to

(4.297)

However, according to Eq. (4.270),

(4.270)
Hence,
A = AO _ (ZeoE'X + eo w'X AO)
2
(4.298)
900n1] 6ekT
 ION TRANSPORT IN SOLUTIONS 435

Replacing x by the familiar expression (3.84), i.e.,

one has

A =
ze F (8 ~Z2e 2N
AO - [ 9000~'}') '" () A
)J + e02(,) (8~Z2e 2N)~
'" 0 A AO ] I (4.299)
"'{ IOOOl'kT- 61'kT IOOOl'kT C2

This is the well-known Debye-HUckel-Onsager equation for a sym-

metrical electrolyte. By defining the following constants

A ze0F_ (8nZ 2eO

= __ 2NA )~ B = e()2w (8nZeoNA )~
and
900n1] 1000l'kT 61'kT 1000l'kT

it can also be written thus

or
= AO - constant d (4.300)

Thus, the theory of ionic clouds has been able to give rise to an equation
which has the same form as the empirical law of Kohlrausch (Section 4.3.9).

4.6.13. The Theoretical Predictions of the Debye-Huckel-Onsa-

ger Equation versus the Observed Conductance Curves

The two constants

(4.301)

and
(4.302)

in the Debye-HUckel-Onsager equation are completely determined (Table

4.20) by the valence type of the electrolyte, z, the temperature T,
the dielectric constant f, and the viscosity 1] of the solution and by universal
constants.
The Debye-HUckel-Onsager equation has been tested against a large
body of accurate experimental data. The confrontation of theory and
experiment is shown in Fig. 4.87 and Table 4.21 for aqueous solutions of
true electrolytes, i.e., substances which consisted of ions in their crystal
lattices before they were dissolved in water.
436 CHAPTER 4

TABLE 4.20
Values of the Onsager Constants for Uni-Univalent Electrolytes at 25 0

Solvent A B

Water 60.20 0.229

Methyl alcohol 156.1 0.923
Ethyl alcohol 89.7 1.83
Acetone 32.8 1.63
Acetonitrile 22.9 0.716
Nitromethane 125.1 0.708
Nitrobenzene 44.2 0.776

H CI
426

422
KCI
150

146 Q)

-
u
AgN03 r::
g
U
:::J
"1:J
r::
Q) 131 o
u u
c:
g
U 129
:::J
"1:J
c:
8 127
C "'Q.o SOd:
.!! 125 ~ III",
g "'h....o -." CIt/o,..
.~ .>J~~
t1r 123
:::J

121 L-~L-~ __~__~__~__~~

~
0·02 0·04 0·06
No CI JConcentrotion
Fig. 4.87. Comparison of the equivalent
conductivities of HCI and some salts
predicted by the Oebye-HUckel-Onsager
equation (4.299) with those observed experi-
mentally.
 ION TRANSPORT IN SOLUTIONS 437

TABLE 4.21
Observed and Calculated On sager Slopes in Aqueous Solutions at 25 0 C

Electrolyte Observed slope Calculated slope

LiC! 81.1 72.7

NaNO a 82.4 74.3
KBr 87.9 80.2
KCNS 76.5 77.8
CsC! 76.0 80.5
MgC!z 144.1 145.6
Ba(NOa)2 160.7 150.5
K 2SO. 140.3 159.5

At very low concentrations « O.OOlN, say), the agreement between

theory and experiment is very good. There is no doubt that the theoretical
equation is a satisfactory expression for the limiting tangent to the ex-
perimentally obtained A versus d curves.
One cannot, however, expect the Debye-Hilckel-Onsager theory of
the nonequilibrium conduction properties of ionic solutions to fare better
at high concentration than the corresponding Debye-Hilckel theory of the

425

7
cr
Cll
N
E 421
u
'E
-8
<"

8
100 ../Concentrotion

Fig. 4.88. Deviation of the predicted equivalent conductivities from those

observed for He!.
438 CHAPTE R 4

equilibrium properties, e.g., activity coefficients, of electrolytic solutions;

both theories are based on the ionic-cloud concept.
In the case of the Debye-Hiickel-Onsager equation, it is seen from
Fig. 4.88 that, as the concentration increases (particularly above O.OOIN),
the disparity between the theoretical and experimental curves widens.

4.6.14. A Theoretical Basis for Some Modifications of the Oe-

bye-Huckel-Onsager Equation

A general procedure adopted by theorists for the improvement and

extension of any theory is to examine its basic assumptions critically.
The elementary theory of conduction in electrolytic solutions was
based on the assumption that the charge-carrying ions do not get in each
other's way, i.e., that the ionic cloud is so tenuous that it can be ignored.
The removal of this assumption required the incorporation of ionic-at-
mosphere effects, in particular, the development of an expression for the
relaxation field and the effects of the electrophoretic movement of the ionic
cloud. Thus arose the Debye-Hiickel-Onsager equation for equivalent
conductivity. Even this theory, however, is not free of assumptions and
approximations.
One obvious assumption in the Debye-Hiickel-Onsager theory centers
around the quantity c. The theory considers c the concentration of charge
carriers in the solution. Whether c will be equal to the stoichiometric con-
centration Cs depends on whether every ion from the ionic lattice of the true
electrolyte dissolves in the solution and remains free as a charge carrier. This
condition will be met only if ion association (see Section 3.8) does not occur.
A more serious assumption in the Debye-Hiickel-Onsager treatment
of conductance is that of point-charge ions. (This assumption is not self-
evident in the presentation adopted here, but it is explicit in the rigorous
derivations of the theory.) One way of considering finite, nonzero ion
sizes is to replace u in the Debye-Hiickel-Onsager equation by uf(l + ua),
where a is the closet distance of approach of a cloud ion to the central ion.
It may be recalled that, in the theory of activity coefficients, a similar change
was required to incorporate ion-size effects in the expression for the activity
coefficient (see Section 3.5).
The effect of considering that ions have a finite size is to introduce a
positive term proportional to concentration and make the theoretical
A versus cf curve concave from the limiting tangent up.
Finally, the relaxation and electrophoretic effects have been discussed
in the simple treatment as if they were independent of each other. In fact,
 ION TRANSPORT IN SOLUTIONS 439

both effects arise from the response of the same ionic cloud, populated by
the same ions and solvent molecules, to the changing field of a moving
central ion (relaxation effect) and to the external field (electrophoretic
effect). Since the same assembly of ions is responsible for the two effects,
they cannot be independent. For instance, the electrophoretic movement
of the ionic cloud must affect the relaxation field.
Fuoss and On sager, in a series of papers from 1955 onwards, have
considered the effects of finite-ion-size, ion-pairing, and electrophoretic
terms in the relaxation field on the conductance of electrolytes.
They have also allowed for two other factors. Firstly, since the ionic
cloud is egg shaped around a moving central ion, there will be more ions
of opposite sign behind the central ion than in front of it (i.e., in the direc-
tion in which it is drifting). These oppositely charged ions are attracted to
the central ion, which therefore gets more kicks from behind than from
the front. The central ions thus acquire a small velocity component in the
direction opposite to that of travel demanded by the external field. In other
words, there is in effect a small force in the same direction as the external
field. Secondly, due to the presence of an excess of oppositely charged ions
around the central ion, there is a small change in the local viscosity. This
change alters the mobility of the central ion, as can be seen from Stokes'
laws.
The Fuoss-Onsager extended theory appears to be as far as one can
go in the conductance problem while still retaining the basic concepts of
the ionic cloud. It may be recalled, however, that in concentrated solutions
when the mean ion-ion distance becomes of the order of ionic dimensions,
the ionic-atmosphere concept becomes rather untenable (see Section 3.9).
One must think of ion clusters and quasi-lattices. The problems of conduct-
ance in concentrated solutions and in the most concentrated solutions of
all-pure electrolytes devoid of water-demand an altogether different ap-
proach, to which some attention will be given in Chapter 6.

Further Reading
1. P. Debye and E. Hiickel, Z. Physik, 24: 185 (1923).
2. L. Onsager, Z. Physik, 28: 277 (1927).
3. H. Falkenhagen, Z. Physik, 32: 365 and 745 (1931).
4. R. M. Fuoss and F. Accasina, Electrolytic Conductance, Interscience Publish-
ers, Inc., New York, 1959.
5. H. L. Friedman, Ionic Solution Theory, Interscience Publishers, Inc., New
York, 1962.
6. R. M. Fuoss and L. Onsager, J. Phys. Chem., 66: 1722 (1962); 67: 621 (1963);
and 68: 1 (1964).
440 CHAPTER 4

7. R. M. Fuoss, L. Onsager, and J. F. Skinner, J. Phys. Chern., 69: 2581 (1965).

8. H. L. Friedman, "A New Theory of Conductance," in: B. E. Conway and
R. G. Barradas, eds., Chemical Physics of Ionic Solutions, John Wiley & Sons,
Inc., New York, 1966.

4.7. NONAQUEOUS SOLUTIONS: A NEW FRONTIER IN IONICS?

4.7.1. Water Is the Most Plentiful Solvent

Not only is water the most plentiful solvent, it is also a most successful
and useful solvent, and there are several good reasons which support this
description. Firstly, the dissolution of true electrolytes occurs by solvation
(Chapter 2) and, therefore, depends on the free energy of solvation. A
sizable fraction of this free energy depends on the Born contribution
(NAz i2eo2/2r)(1 - (1/ s». It follows that, the greater the dielectric constant
of the solvent is, the greater is its ability to dissolve true electrolytes. Since
water has a particularly high dielectric constant (Table 4.22), it is a success-
ful solvent for true electrolytes.
A second advantage of water is that, in addition to being able to dissolve
electrolytes by the physical forces involved in solvation, it is also able to
undergo chemical proton-transfer reactions with potential electrolytes and
thus produce ionic solutions (Chapter 5). Now, water is able both to donate

TABLE 4.22
Dielectric Constants of Some Solvents
(Temperature 25 0 C unless Otherwise Noted)

Solvent Dielectric constant

Water 78.30
Acetone 20.70
Acetonitrile 36.70
Ammonia (- 34°C) 22.00
Benzene 2.27
Dimethly acetamide 37.78
Dimethyl sulfoxide 46.70
Dioxan 2.21
Ethanol 24.30
Ethylene diamine 12.90
Hydrogen cyanide (16 0c) 118.30
Pyridine 12.00
Sulfuric acid 101.00
 ION TRANSPORT IN SOLUTIONS 441

TABLE 4.23
Boiling Points of Some Solvents

Solvent Boiling point, °C

Water 100.00
Acetone 56.20
Benzene 80.10
1 :1 Dichloroethane 57.00
Methanol 64.96

protons to, and to receive protons from, molecules of potential electrolytes.

Thus, water can function as both a source and a sink for protons and can
consequently enter into ion-forming reactions with a particularly large
range of substances. This is why potential electrolytes often react best with
water as a partner in the proton-transfer reactions.
Finally, water is stable both chemically and physically at ambient
temperature, unlike many organic solvents, which tend to evaporate (Table
4.23) or decompose slowly with time.
On the whole, therefore, ionics is best practiced in water. Nevertheless,
there are also good reasons why nonaqueous solutions of electrolytes are
often of interest.

4.7.2. Water Is Often Not an Ideal Solvent

If water were the ideal solvent, there would be no need in technology
to consider other solvents. But, in many situations, water is hardly the
ideal solvent. Take the electrolytic production of sodium metal, for example.
If an aqueous solution of a sodium salt is taken in an electrolytic cell and a
current passed between two electrodes, then all that will happen at the
cathode is the liberation of hydrogen gas; there will be no electrodeposition
of sodium (ef Chapter 8). Hence, sodium cannot be electrowon from
aqueous solutions. This is why the electrolytic extraction of sodium has
taken place from molten sodium hydroxide, i.e., from a medium free of
hydrogen. But this process requires the system to be kept molten ("-' 600°C),
and it, therefore, compels the use of high-temperature technology with its
associated materials problems. It would be a boon to industry if one could
use a low temperature conducting solution having the capacity to maintain
sodium ions in a nonaqueous solvent free of ionizing hydrogen. This
argument is valid for many other metals which are extracted today by
442 CHAPTER 4

electrodic reactions in fused salts at high temperatures with the attendant

difficulties of corrosion and heat losses.
Another vast field awaiting the development of nonaqueous electro-
chemistry is that of energy storage for automobiles. Many reasons, e.g.,
the growing danger of pollution from automobile exhausts, the increasing
concentration of CO 2 (with the consequences of a warmup of the atmos-
phere), the accelerating consumption of oil reserves, etc., make the search
for an alternative to the internal combustion engine a necessity. Nuclear
reactors with their attendant shielding problems will always be too heavy
for the relatively small power needed in road vehicles. Thus, there would
be attractive advantages to be obtained by the development of a fumeless,
vibration-free electric power source. However, the presently available cheap
electrochemical device-the lead-acid battery-is too heavy for the electric
energy which it stores to offer reasonable performance with convenient
distance range between recharging. Electrochemical-energy storers avail-
able today which do have a sufficiently high energy capacity per unit weight
offer difficulties because of their expense. The highest energy density theor-
etically conceivable is in a storage device which utilizes the dissolution of
lithium or beryllium. But aqueous electrolytes are debarred because, in
them, these metals corrode wastefully rather than dissolve with useful power
production. So one answer to the need for an electrochemically powered
transport system is the development of a nonaqueous electrochemical-en-
ergy storage system incorporating alkali or alkaline-earth metal electrodes.
Many other examples could be cited in which the presence of water
as a solvent is a nuisance. In all these cases, there may exist an important
future for applications using nonaqueous solutions.
However, a nonaqueous solution must be able to conduct electricity
if it is going to be useful. What, therefore, determines the conductivity of
a nonaqueous solution? The case of nonaqueous solutions in which the
charge carriers are produced by chemical reactions between molecules of
the solvent and the potential electrolyte will be dealt with in Chapter 5.
Here, the theoretical principles involved in the conductance behavior of
true electrolytes in nonaqueous solvents will be sketched.

4.7.3. The Debye-Huckel-Onsager Theory for Nonaqueous So-

lutions

An examination of the Debye-Hilckel-Onsager theory in Section 4.6

shows that it is in no way wedded to the particular solvent water. Does exper-
iment support the predicted A versus d curve in nonaqueous solutions, too?
 ION TRANSPORT IN SOLUTIONS 443

120 f-~b-+--+---+----1

A
110 I----i-"'-'Un\:. lo--="h-rr-+---I

100 I--~{-=---+-=-....:::>o.,""'-+---I

90 '--~":-:-_-'--_-'--_-L-_-'
0·05
.j Concentration

Fig. 4.89. Change in the equivalent con-

ductivity of some alkali cyanates with con-
centration in methyl alcohol.

Figure 4.89 shows the variation of the equivalent conductivity versus

concentration for a number of alkali sulphocyanates+ in a methanol solvent.
The agreement with the theoretical predictions demonstrates the applicability
of the Debye-Hlickel-Onsager equation.
When one switches from water to some nonaqueous solvent, the
magnitudes of several quantities in the Debye-Hlickel-Onsager equation
alter, sometimes drastically, even if one considers the same true electrolyte
in all these solvents. These quantities are the viscosity and the dielectric
constant of the medium and the distance of closest approach of the solvated
ions, i.e., the sum of the radii of the solvated ions. As a result, the mobilities
of the ions at infinite dilution, the slope of the A versus ck curve, and, lastly,
the concentration c of free ions cause the conductance behavior of an
electrolyte to vary when one goes over from water to nonaqueous solvent.
These effects will now be considered in detail.

4.7.4. The Solvent Effect on the Mobility at Infinite Dilution

At infinite dilution, neither the relaxation nor the electrophoretic effects

are operative on the drift of ions; both these effects depend for their exist-
ence on a finite-sized ionic cloud. Under these special conditions, therefore,

+ Abundant work exists on such electrolytes, rather than on, say, NaCI, because elec-
trolytes with ions larger than CI- ions are more soluble in organic solvents than are
the chlorides.
444 CHAPTER 4

the infinite-dilution mobility can be considered given by the Stokes mobilit)

o _ zeo
(4.180)
uconv - 6nr", 300

Hence, considering the same ionic species in several solvents, one has

ugonvr", = constant (4.303)

If the radius r of the solvated ion is independent of the solvent, then

one can approximate Eq. (4.303) tot

(4.304)

Hence, an increase in the viscosity of the medium leads to a decrease in

the infinite-dilution mobility, and vice versa.
Table 4.14 contains data on the equivalent conductivity, the viscosity,
and the Walden constant .110", for several electrolytes and several solvents.
It is seen that: (1) Eq. (4.304) is a fair approximation in many solvents and
(2) its validity is better for solvents other than water.
In fact, however, the radius of the kinetic entity may change in going
from one solvent to the other, essentially because of changes in the structure
of the solvation sheath. Sometimes, these solvation effects on the radius
may be as much as 100%. Hence, it is only to this rough degree that one
can use the approximate equation (4.304).
In some cases, the changes in the radii of the solvated ions are mainly
due to the changes in the sizes of the solvent molecules in the solvation
sheath. Thus, in the case of the solvents water, methanol, and ethanol,
the size of the three solvent molecules increases in the order

Water -- Methanol -- Ethanol

Since the radius of the solvated ions should also increase in the same order,
it follows from Eq. (4.303) that the mobility or equivalent conductivity at
infinite dilution should increase from ethanol to methanol to water. This
is indeed what is observed (cf Table 4.15).
One should be careful in using a simple Walden's rule .111] = constant,
which assumes that the radii of the moving ions are indepehdent of the

tActually, Eq. (4.304) containing the mobility is a form of Walden's rule [ef Eq. (4.189)]
which contains the equivalent conductivity. Since AO = F[(uconv)+o + (uconv)_O]-ef
Eq. (4.270)-, it is easy to transform (4.304) into the usual form of Walden's rule.
 ION TRANSPORT IN SOLUTIONS 445

solvent. Rather, one should use a generalized Walden's rule, namely,

(4.303)

where r is the radius of the ionic entity concerned in a given solvent.

4.7.5. The Slope of the A versus d Curve as a Function of the

Solvent

If one takes the generalized Walden's rule (4.303) and calculates

(from AO = FuO) the equivalent conductivity at infinite dilution for a number
of nonaqueous solutions, it turns out that the values of AO in such solutions
are relatively high. They are near those of water and are in some cases greater
than those of water.
One might naIvely conclude from this fact that, in using nonaqueous
solutions instead of aqueous solutions in an electrochemical system, the
conductivity presents no problem. Unfortunately, this is not the case. The
crucial quantity which often determines the feasibility of using nonaqueous
solutions in practical electrochemical systems is the specific conductivity
a at a finite concentration, not the equivalent conductivity AO at infinite
dilution. The point is that it is the specific conductivity which, in conjunction
with the electrode geometry, determines the electrolyte resistance in an
electrochemical system. This electrolyte resistance is an important factor
in the operation of an electrochemical system because the extent to which
useful power is diverted into the wasteful heating of the solution depends
on /2 R, where / is the current passing through the electrolyte; hence, R must
be reduced or the a increased.
Now, a is related to the equivalent conductivity A at the same con-
centration
a =Azc (4.305)

But A varies with concentration; this is what the Debye-HUckel-Onsager

equation was all about. Hence, to understand the specific conductivity a
at a concentration c, it is not adequate to know the equivalent conductivity
under a hypothetical condition of infinite dilution. One must be able to
calculate, t for the nonaqueous solution, the equivalent conductivity at finite
concentrations, utilizing the AO value and the theoretical slope of the A

t Of course, the validity of the calculation depends upon whether the theoretical ex-
pression for the equivalent conductivity (e.g., the Debye-Huckel-Onsager equation)
is valid in the given concentration range.
446 CHAPTER 4

versus c! curve. This will be possible if one knows the values of the constants
A and D in the Debye-Hiickel-Onsager equation

where

and

When one looks at the above expressions for A and D, it becomes

obvious that, as e decreases, A and D increase; the result is that A and
therefore (1 decrease. Physically, this corresponds to the stronger interionic
interaction's arising as the e is reduced.
So the question of the specific conductivity of nonaqueous solutions
vis-a-vis aqueous solutions hinges on whether the dielectric constant of
nonaqueous solvents is lower or higher than that of water. Table 4.22 shows
that many nonaqueous solvents have e's considerably lower than that of
water. There are some notable exceptions, namely, the hydrogen-bonded
liquids.
Thus, on the whole, the effect of increase of electrolyte concentration
on lowering the equivalent conductance is much greater in nonaqueous
than in aqueous solutions. The result is that the specific conductivity of
nonaqueous solutions containing practical electrolyte concentrations is
far less than the specific conductivity of aqueous solutions at the same
electrolyte concentration (Table 4.24 and Fig. 4.90).

TABLE 4.24
Specific Conductivity of Electrolytes in Aqueous and Nonaqueous Solvents
at the Same Concentration

Cor,centration, Specific conductivity, ohm-1 cm- 1 at 25°C in

Electrolyte
IT'( les liter- 1
Water Nonaqueous solvent

HCI 0.1 391.32 x 10-4 (methanol) 122.5 x 10-'

(ethanol) 35.43 x 10-4
(n-propanol) 8.80 x 10-4
NaCI 0.01 11.85 x 10-' (methanol) 7.671 x 10-4
KCI 0.01 14.127 x 10-' (methanol) 8.232 x 10-'
 ION TRANSPORT IN SOLUTIONS 447

2· 0 t::="::=I.=W=,atF·r===9=W"tI1=i~
E -78'6

<: 1·0 r---+---+----H

CI
o
0'0 J----"I~-__j--_'f4

-1,0 L - -_ _' - -_ _-'---_ _- ' - '

-4'5 -3·5 -2,5

log c

Fig. 4.90. Comparison of the concentra-

tion dependence of the equivalent con-
ductivity of tetraisoamylammonium nitrate
dissolved in water and in water-dioxane
mixtures.

In summary, it is the lower dielectric constants of the typical non-

aqueous solvent which causes a far greater decrease in equivalent con-
ductivity with increase of concentration than that which takes place in
typical aqueous solutions over a similar concentration range. Even if the
infinite-dilution value AO makes a nonaqueous electrochemical system look
hopeful, the practically important value of the specific conductivity is nearly
always much less than in the corresponding aqueous solution. That is the
sad aspect of nonaqueous solutions.

4.7.6. The Effect of the Solvent on the Concentration of Free

Ions: Ion Association

The concentration c which appears in the Debye-Hlickel-Onsager

equation pertains only to the free ions. This concentration becomes equal
to the analytical concentration, which shall be designated here as Ca , only
if every ion from the ionic lattice from which the electrolyte was produced
is stabilized in solution as an independent mobile charge carrier; i.e., c #- Ca
if there is ion-pair formation. Whether ion-pair formation occurs or not
depends on the relative values of a, the closest distance of approach of
oppositely charged ions and of the Bjerrum parameter q = (z+Le o2 /2kT)I/e.
When a < q, the condition for ion-pair formation is satisfied, and, when
a > q, the ions remain free.
From the expression for q, it is clear that the lower the dielectric con-
stant of the solvent is, the larger is the magnitude of q. Hence, when one
448 CHAPTER 4

replaces water by a nonaqueous solvent, the likelihood of ion-pair formation

increases because of the increasing q (assuming that a does not increase
in proportion to q).
It has already been emphasized that an ion pair, taken as a whole, is
electrically neutral and ceases to play its role in the ionic cloud (Section
3.8). For the same reason (i.e., that the ion pair is uncharged), the ion pair
does not respond to an externally applied electric field. Hence, ion pairs do
not participate in the conduction of current.
A quantitative analysis of the extent to which ion-pair formation affects
the conductivity of an electrolyte must now be considered.

4.7.7. The Effect of Ion Association upon Conductivity

In treating the thermodynamic consequences of ion-pair formation
(Section 3.8.4), it has been shown that the association constant KA for
the equilibrium between free ions and ion pairs is given by

() fip (3.172)

where () is the fraction of ions which are associated, C a is the analytical

concentration of the electrolyte, I ± is the mean activity coefficient, and
lIP is the activity coefficient for the ion pairs. Since neutral ion pairs are
not involved in the ion-ion interactions responsible for activity coefficients
deviating from unity, it is reasonable to assume that lIP"" 1, in which case,

(4.306)

A relation between () and the conductivity of the electrolyte will now

be developed. The specific conductivity has been shown [cf Eq. (4.307)]
to be related to the concentration of mobile charge carriers (i.e., of free
ions) in the following way

a = zF(u+ + U-)cfree ions (4.307)

One can rewrite this equation in the form

(4.308)

Since Cfree ions/ca is the fraction of ions which are not associated (i.e., are
free), it is equal to unity minus the fraction of ions which are associated.
 ION TRANSPORT IN SOLUTIONS 449

Hence,
Cfree ions = I _ () (4.309)
Ca

and, using this result in equation

(4.310)

or, from the definition of equivalent conductivity, i.e.,

(4.305)

one can write

(4.311)

If there is no ion association, i.e., () = 0, then one can define a quantity

11 0=0, which is given by [cf Eq. (4.311)]
(4.312)

By dividing Eq. (4.311) by (4.312), the result is

1-()=~ (4.313)
11 0=0
and
(4.314)

Introducing these expressions for () and 1 - () into Eq. (4.306), one

finds that

or
(4.315)

Though Eq. (4.315) relates the equivalent conductivity to the electrolyte

concentration, it contains the unevaluated quantity 11 0=0' By combining
Eqs. (4.310) and (4.312), one gets

(4.316)

from which it is clear that 11 0=0 is the equivalent conductivity of a solution

450 CHAPTER 4

in which there is no ion association but in which the concentration is

(1 - O)c a . Thus, for small concentrations (see Section 4.6.12), one can
express 11 0=0 by the Debye-Hiickel-Onsager equation (4.299), taking care
to use the concentration (1 - O)ca . Thus,

(4.317)

which can be written in the form (cf Appendix 4.5)

(4.318)

where Z is the continued fraction

Z = 1 - z {I - z [1 - z ( ... )-i]-i}-i (4.319)

with
(4.320)

Introducing expression (4.318) for 11 0=0 into Eq. (4.315), one has

1 1 KAf±2
A = I10Z + (I10)2Z2 ACa
or
(4.321)

This is an interesting result. It can be seen from (4.321) that the

association of ions into ion pairs has entirely changed the form of the
equivalent conductivity versus concentration curve. In the absence of sig-
nificant association, A was linearly dependent on ct , as empirically shown
by Kohlrausch. When, however, there is considerable ion-pair formation
(as would be the case in nonaqueous solvents of low dielectric constant),
instead of the Kohlrausch law, one finds that, when Z/A is plotted against
Af±2Ca/Z, a straight line is obtained with slope KA/(AO)2 and intercept 1/11°.
Figure 4.91 shows the experimental demonstration of this conductance
behavior.

4.7.8. Even Triple Ions Can Be Formed in Nonaqueous Solutions

When the dielectric constant of the nonaqueous solvent goes below
about 15, ions can associate not only in ion pairs but also in ion triplets.
This comes about by one of the ions (e.g., M+) of an ion pair M+· ·A-
 ION TRANSPORT IN SOLUTIONS 451

O·014,---,---r--.,.----,
HCl

O' 0 131-----j--7I'''9----+-------i
Flzl
A
O· 012 i""""---=--t--~_::c.-._t_---I

0'01l1-----j---+---+---i

0'00 0·04 0'08 0'12 0·16

Fig. 4.91. Plots of Eq. (4.321) for the

hydrogen halides in ethyl alcohol.

coulombically attracting a free ion A-strongly enough to overcome the

thermal forces of dissociation

A- + M+· ·A-:;;:=: A-· ·M+· ·A-

From the conductance point of view, ion pairs and triple ions behave
quite differently. The former, being uncharged, do not respond to an ex-
ternal field; the latter are charged and respond to the external field by
drifting and contributing to the conductance.

-1,0

V
-2'0
<: I
! i
, c
OJ
~ a 2'56
-3,0
~ ~
-4'0

-4,5 -H -2,5
log C
Fig. 4.92. Minimum in the curve for equiv-
alent conductivity versus concentration in
the case of tetraisoamylammonium nitrate in
a water-dioxane mixture of dielectric con-
stant of c = 2.56.
452 CHAPTER 4

The extent of ion-pair formation is governed by the equilibrium be-

tween free ions and ion pairs. In like fashion, the extent of triple-ion for-
mation depends on the eqilibrium between ion pairs and triple ions.

M+ + A- ~ M+·· ·A- ~
A-
A-· ·M+· ·A-

Thus, the greater the stoichiometric concentration, the greater is the ion-
pair formation and triple-ion formation.
With increasing concentration, therefore, ion-pair formation dominates
the equivalent conductivity, which decreases with increasing concentration
faster than had there been no formation. Then, at still higher concentrations,
when triple-ion formation starts becoming significant, the equivalent con-
ductivity starts increasing after passing through a minimum. This behavior
has been experimentally demonstrated (Fig. 4.92).

4.7.9. Some Conclusions about the Conductance of Nonaqueous

Solutions of True Electrolytes

The change from aqueous to nonaqueous solutions of true electrolytes

results in characteristic effects on the conductance. The order of magnitude
of the equivalent conductivity at infinite dilution is approximately the same
in both types of solutions. But the slope of the equivalent-conductivity
versus concentration curve is drastically more negative in nonaqueous
solutions than in the corresponding aqueous solutions. This means that
the actual specific conductivity G, which is the only significant quantity
as far as the conducting power of an actual solution is concerned, is much
lower for nonaqueous solutions. Ion-pair formation worsens the conduct-
ance situation; triple-ion formation is a slight help.
Thus, nonaqueous solutions of true electrolytes are not to be regarded
optimistically for applications in which there is a premium on high specific
conductivity and minimum power losses through resistance heating. If non-
aqueous solutions are going to be useful, one has to think of solutions of
potential electrolytes which interact chemically with the solvent. (Chapter 5).

Further Reading
1. N. Bjerrum, Kgl. Danske Videnskab. Selskab Mat-Fys. Medd., 7 (9) (1933).
2. R. M. Fuoss and C. A. Kraus, J. Am. Chern. Soc., 55: 476 (1933).
3. R. A. Robinson and R. H. Stokes, Electrolyte Solutions, Butterworths' Pub-
lications, Ltd., London, 1955.
4. H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolytic Solutions,
Reinhold Publishing Corp., New York, 1957.
 ION TRANSPORT IN SOLUTIONS 453

5. R. M. Fuoss and F. Accasina, Electrolytic Conductance, Interscience Publish-

ers, Inc., New York, 1959.
6. C. B. Monk, Electrolytic Dissociation, Academic Press, London, 1961.
7. C. W. Davies, Ion Association, Butterworths' Publications, Ltd., London,
1962.
8. J. E. Prue, "Ion Association and Solvation," in: B. E. Conway and R. G.
Barradas, eds., Chemical Physics of Ionic Solutions, John Wiley & Sons, Inc.,
New York, 1966.

Appendix 4.1. The Mean Square Distance Traveled by a Ran-

dom-Walking Particle
In the one-dimensional random-walk problem, the expression for <XZ) is
found by mathematical induction as follows. Consider that, after N - I steps,
the sailor has progressed a distance XN -1' If he takes one more step, the distance
XN from the origin will be either

XN = XN-I +I (A4.1.1 )
or
XN = XN-I - I (A4.1.2)

Squaring both sides of Eqs. (A4.1.1) and (A4.1.2), one obtains

(A4.1.3)
and
(A4.IA)

The average of these two possibilities must be

(A4.1.5)

This is the result for xn z when the distance traveled after N - I steps is exactly
XN -1' In general, however, one can only expect, for the value of the square of
the distance at the (N - 1)th step, an averaged value <XN'-,), in which case one
must write
(A4.1.6)

Now, at the start of the random walk, i.e., after zero steps, the progress is
given by
<X02) = 0 (A4.1.7)
After one step, it is
<XIZ) = liz = IZ (A4.1.8)

After two steps, from Aq. (A4.1.6), one has

<XZ2) = <x,") + [2 (A4.1.9)

454 CHAPTER 4

and, using Eq. (A4.1.8),

(A4.1.10)
Similarly,
(X3 2) = (X22) + [2
= 2[2 + [2
= 3[2 (A4.1.1l)
Hence, in general,
(XN2) = N[2 (A4.1.12)

This equation has been derived for one-dimensional random walk, but it
can be shown to be valid for three-dimensional random flights, too.
The mean square distance (X2) which a particle travels depends upon the
time of travel in the following manner. The number of steps, N, obviously in-
creases with time and is proportional to it, i.e.,
N=kt (A4. 1.1 3)

where k is the constant of proportionality. Hence, by combining (A4.1.12) and

(A4.1.13),
(X2) = kt[2 (A4.1.14)
which may be written
<X = 2) at (A4.1.15)

where a is a proportionality constant to be evaluated in the Einstein-Smolu-

chowski equation.

Appendix 4.2. The laplace Transform of a Constant

The Laplace transform a of a constant a is, by definition (4.33), given by

_ [ e- PZ ]00
a=a - -
-p 0

= 0- ( ~p)
a
p

Hence, the Laplace transform of a constant is equal to that same constant divided
by p.
 ION TRANSPORT IN SOLUTIONS 455

Appendix 4.3. A Few Elementary Ideas on the Theory of Rate

Processes
1. In any rate process, the initial state (condition) of the system is character-
ized by a certain configuration (arrangement) of the entities (ions, atoms, sites,
etc.) involved. Similarly, the final state is characterized by a certain (but dif-
ferent from the initial) configuration of the entities. Thus, in the case of an ionic
jump, the initial state of the system is

where the subscripts 1 and 2 serve to differentiate the sites, and the final state is

2. When the initial or final states are considered, the position of the ion is
described in "either-or" language: The ion is in either site 1 or site 2. During
the jump, however, the ion is in intermediate positions, and the jumping process
can be described by giving the two distances, d,- I and dI -2' which separate the
ion I from the two sites.
3. As the ion jumps from site 1 to site 2, the distance d,- I increases from
zero in the initial state to the mean jump distance I in the final state. While d,- I
is fairly small, one can reasonably [see Fig. A4.3.1 (a) and (b)] say that the system
is in the initial state (the jump has not yet occurred). But there is a limit to this
view. The quantitative change of the distances d,- I and d I - 2 leads to a qualitative
change in the configuration of the system from a configuration characterizing
the initial state to one characterizing the final state. There is a critical configur-
ation beyond which the configuration of the ion and the sites [Fig. A4.3.l(c)]
more closely resembles the system in its final state (after the jump), i.e., dI - 2
small, than in its initial state (before the jump), i.e., d,- I small. This critical
configuration is known as the activated complex (for reasons which will be explain-
ed). When the initial and final states are mirror images, then the activated complex,

(e)
456 CHAPTER 4

.---'_;- REACTION COORDINATE

d l- I
Fig. A4.l.2.

or critical configuration, must correspond to the halfway point; when these states
are not mirror symmetrical, one has to get the configuration of the activated
complex by considering the dependence of the energy of the system upon the
distances between the reacting entities.
4. Each configuration (during the process of jumping) corresponds to a
particular energy for the system. Thus, for every value of d1 - I and d I -2' there is
a particular value of the energy of the system. If these energies are plotted as a
function of the distance variables, then one obtains a three-dimensional potential-
energy surface (Fig. A4.3.2). Across this surface, there are an infinite number of
paths representing the passage of the system from the initial state or the final
state. The easiest of these paths is termed the reaction coordinate. If the surface
is cut along the reaction coordinate, then a potential-energy versus distance
diagram (Fig. A4.3.3) is obtained. It is necessary to be clear that any point on
this diagram represents the potential energy that the system has for the corres-
ponding configuration of the entities involved. Thus, in the case of an ionic jump
from one site to another, the jump can be described by a continuous change of
the coordinates of the ion in relation to the two sites. Hence, this jump is represent-
ed by the motion of a point along the potential-energy curve from the minimum
on the left-hand side to the minimum on the right-hand side.
5. It can be seen from the potential-energy diagram that, as the ion starts
moving away from one site (toward the other), the potential energy of the system
starts increasing and the point representing the energy of the system (on the

x
POTENTIAL
ENERGY

REACTION COORDINATE

Fig. A4.l.l.
 ION TRANSPORT IN SOLUTIONS 457

potential-energy-distance diagram) starts "climbing" a potential-energy hill

Near the halfway point, the potential energy attains a maximum and then starts
decreasing, and the representative point goes over the peak of the curve. The
peak corresponds to the configuration of the system known as the activated
complex, i.e., to the critical configuration which the system must attain before
the process (the diffusion jump, in this case) can be said to occur. This is why
it is said that the system has to climb a potential-energy barrier and has to be
activated with a critical activation energy iJH* for the process to occur, be it
diffusion or a chemical reaction. If the entropy change iJS* in going from the
initial to this activated state is also taken into account, then the rate process can
be considered to need a critical free energy of activation iJG* = iJH* - T iJS*.
6. The system in the activated state, i.e., the activated complex, has one
exceptional property: If one distance between the constituents increases (d1- I in
the diffusion-jump case), there is a breakdown of the activated complex into the
system corresponding to the final state, i.e., the system tumbles down the hillside
toward the final state. Why does d1- I change? One may consider the entities
D 1 and I in a sort of "bond" which vibrates. It is this vibration which leads

to decomposition of the activated complex.

7. The vibration leading to breakdown of the activated complex takes place
with a certain vibration frequency v which has associated with it, according to
Planck's law, an energy E given by
E= hv (A4.3.1)

where h is the Planck constant. But a vibration in the activated complex is a relative
displacement of the entities constituting the system. The displacement can also be
viewed as a translation of one of the entities in relation to the rest of the system
(displacement of the ion I in relation to site O. The translation has associated
with it a translational energy which, according to the kinetic theory, t is !kT per
entity or kT for a pair of entities (cl the calculation of the translational energy
in Bjerrum's ion-pair theory). Hence, when the vibration of the I-I bond is viewed
as a translation, the associated energy is E given by
E=kT (A4.3.2)

Since Eqs. (A4.3.1) and (A4.3.2) arise from two ways of looking at the
same process, the two expressions for the energy must be equal, i.e.,
hv = kT (A4.3.3)

The particular vibration frequency v of the activated complex leading to de-

composition (the production of the final state) is therefore given by the expression
kT
v=-- (A4.3.4)
h

t Note: The k here is the Boltzmann constant.

458 CHAPTER 4

8. At this stage, a central feature of the rate-process theory enters the picture.
The activated complexes are considered in equilibrium with the reactants
Reactants ~ Activated complexes
Hence, the law of mass action can be used to give

CAe = K = e-.1GO*IRT (A4.3.5)

CR

where CAe and CR are the concentration of activated complexes and reactants,
K is the equilibrium constant, and L1Go* is the standard free energy of activation.
Considering unit concentration of reactants, C R = 1, one has
(A4.3.6)

9. The number of times per second, k, that the rate process occurs is equal
to the concentration of activated complexes, CAe, times their vibration frequency v.
Hence,
(A4.3.7)

and, inserting the expressions for CAe and v from Eqs. (A4.3.6) and (A4.3.4),
one finds that

(A4.3.8)

This is the expression for the rate constant of a rate process, e.g., of ionic jumping.

Appendix 4.4. The Derivation of Equations (4.257) and (4.258)

According to notation [cf Eqs. (4.251) and (4.252)],
(A4.4.1)
and
(A4.4.2)

Hence, one can carry out the following expansions

q+(q+L++ + q_L+_) + q_(q+L+ + q_L_)

dq; dq;
q+L++ dx + q_L_+ dx
 ION TRANSPORT IN SOLUTIONS 459

Now, it has been stated [cf Eq. (4.247)] that

(A4.4.4)

and that [cf Eqs. (4.254) and (4.255)]

(A4.4.5)

and

(A4.4.6)

-+ -+
Hence, substituting for F+ and F_ in (A4.4.4) and setting dll/dx = 0, one has

(A4.4.7)

Similarly,

(A4.4.8)

In terms of these expressions, Eq. (A4.4.3) becomes

( q+J + )
(A4.4.9)
q+ q+J+ + q_J_ dllldx~o

But, by notation,
and (A4.4.IO)

and, therefore, Eq. (A4.4.8) can be rewritten as

(A4.4.11 )

Further, according to the relation between current density and flux,

and (A4.4.12)

Using these relations in Eq. (A4.4.1O), one has

- -p+ 1
----- ( - i+
- -) (A4.4.13)
p+q+ + p_q_ z+F i+ +L dl'ldx~O
460 CHAPTER 4

By definition, however [ef. Eq. (4.240)],

(A4.4.14)

Hence, by combining Eqs. (A4.4.12) and (A4.4.13), the result is

(A4.4.1S)

Similarly, it can be shown that

p_ L
(A4.4.16)
p+q_ + p_q_ z_F

Appendix 4.5. The Derivation of Equation (4.318)

One can rewrite Eq. (4.317), namely,

(A4.5.1)
in the form
(A4.S.2)

where, for conciseness, the symbol m is used instead of (A + BAO).

It has been shown, however, that [ef. Eq. (4.313)]
A
1-0=A -_- (A4.5.3)
oo

If this relation is used in Eq. (A4.5.2), one gets

A o_o = AO[ 1 _ me:~

A
(~)f]
A _o (A4.5.4)
o

and, substituting for A o_o in the right-hand side from Eq. (A4.5.2), the result is

(A4.5.5)

One has again been left with (1 - O)k on the right-hand side, and, thus, one again
substitutes from Eq. (A4.S.2). This process of substitution can be repeated ad
infinitum to give the result

(A4.5.6)
= AOZ (A4.S.7)
where
(A4.S.8)
CHAPTER 5

PROTONS IN SOLUTION

5.1. THE CASE OFTHE NONCONFORMING ION: THE PROTON

The proton is formed when a hydrogen atom is stripped of its electron.

Its electronic structure is singular: It has no electrons and, therefore, no
structure. It is a pure nucleus. Devoid of an electron shell, the center of
charge of a proton can approach far closer to a neighboring ion or atom
than can any other ion or atom. The proton, therefore, has few steric
restrictions in chemical reactions. Other ions bear the burden of electron
shells and are of angstrom dimensions ('"'-' 10-8 em); the proton is of fermi
dimensions ('"'-' 10-13 cm). It is the smallest ion, and also the lightest.
Indeed, it is an elementary particle-a structural building block of matter.
The proton is a nonconformist ion.
Such an abnormal ion must have abnormal properties. It has been
this knowledge which has dictated the exclusion of the proton from the
treatment of ion-solvent (Chapter 2) and ion-ion interactions (Chapter 3)
and the treatment of the movement of ions (Chapter 4). It will be noticed
that, while the solvation of metal ions and various anions has been discussed
in detail, little attention has been paid to the question of the solvation of
a proton. Similarly, in treating ion-ion interactions, the only ions considered
were those pulled out from an ionic crystal by ion-solvent forces. No at-
tention was paid to potential electrolytes, such as acetic acid, which yield
solutions containing hydrogen ions by a mechanism in which a molecule
of the acid reacts with water.

461
462 CHAPTER 5

It is intended in this chapter to give protons the special treatment they

deserve. The first step will be to consider the interaction between protons
and solvent molecules and to go into the details of proton solvation. Then,
the mechanism by which hydrated protons move in solution will be explored.
Here, there is a special feature, for protons migrate by a mechanism fun-
damentally different from that used by other ions. Thus, the latter retain
their identity and separateness as kinetic entities in their passage through
the solvent, whereas it will be seen that protons are handed on from water
molecule to water molecule and, when they are waiting to be passed on,
they are indistinguishable from the protons of the host solvent. Finally,
the interaction between hydrated protons and other ions will be treated,
with less emphasis on the long-range electrostatic effects characteristic
of the Debye-Hiickel approach than on the short-range chemical interac-
tions which are the basis of the behavior of acids and bases.

5.2. PROTON SOLVATION

5.2.1. What Is the Condition of the Proton in Solution?

The ionization energy required to remove an electron from a hydrogen

atom is much larger than the corresponding energy for the formation of a
univalent ion of any other element (Table 5.1). This fact reflects the extreme
attraction which the proton has for electrons. Thus, protons tend to form
covalent bonds by sharing electron pairs. They also tend to form hydrogen
bonds, which, as a first approximation, may be considered an electrostatic
attraction between a proton and an unshared pair of electrons.

TABLE 5.1
Ionization Energies of the First-Group Elements

Ionization energy,
Atomic number Element
kcai moie- 1

H 313
3 Li 124
11 Na 118
19 K 100
37 Rb 96
55 Cs 90
 PROTONS IN SOLUTION 463

a b

Fig. 5.1. The isomorphous nature of (a) perchloric acid monohydrate and
(b) ammonium perchlorate.

All this suggests that it is highly unlikely that free protons are stable
in solution. Then, in what form do they exist?
Suppose one considers substances like nitric acid or perchloric acid.
Normally, these are liquids, but it is possible to prepare solid forms of these
substances-the so-called "acid hydrates." The solids can then be subjected
to X-ray examination designed to reveal their crystalline structure.
It turns out that perchloric acid monohydrate HCl0 4 • H 2 0 has the
same structure as ammonium perchlorate NH 4Cl0 4 , i.e., the two substances
are isomorphous (Fig. 5.1). It is known, however, that ammonium per-
chlorate is an ionic crystal consisting of an assembly of NH4+ and ClO,-
ions. Hence, the implication of the fact that HCl0 4· H 2 0 is isomorphous
with NH 4 CI0 4 is that CI0 4 - ions must exist in the acid monohydrate and
the remaining proton and water molecules are associated in the way that
a proton is linked with an ammonia molecule to give NH4+' It is difficult,
therefore, to avoid the conclusion that the isomorphous character of the
HCl0 4 • H 20 and NH 4Cl0 4 is based on there being H30+ ions in the mono-
hydrate of perchloric acid corresponding to NH,+ ions in its perchlorate.
Further, since an NH4 + ion is known as an ammonium ion, it is appropriate
to call the H30+ ion a hydronium ion.
What does this X-ray evidence add up to? That protons are associated
with water molecules (or water molecules are protonated) to form H30+
ions. But the X-ray studies were performed on solids, and there is no gua-
rantee that H30+ ions exist in a solution of perchloric acid. This is true;
yet the solution froze into the solid, and it seems likely that the ions existing
in the liquid phase survived the phase transition without structural changes.
So, by an inferential argument, one is led to believe that H30+ ions exist
in solution.
464 CHAPTER 5

Water vapor

l
Water
vapor inlet
~

.. ...
,o.. m:,n
Filament .'.......~
emits electron. :~:••.•'
Neutral water
molecul .. .. .~ :;m:
::::: , .1lectron beam
_lIl!:I-Discharoe
tube
--, r- Jon lenl
-\
Colilmotino
.
_ ........:.~~!:e...
!~:.~
!.~.
••
~
Positive ions
slit.
ENLARGED SIDE VIEW OF
DISCHARGE TUBE
lMixed Ion beam
;
;
;
Maonetic field

.
into paper
Centrifuoa I force

'.
Z"
·. .. '.V
Electromaonetic Collectar
force '. '. slil
Main colleclor-plole
. Analyzed • / •
Ion beams

Fig. 5.2. Schematic representation of mass spectrometer to determine

ionic species in water vapor.

There are other indirect indications that the H 30+ ion is the mode of
existence of protons in solution. For example, consider the studies in the
gas phase. Water vapor is taken in an evacuated tube and a glow discharge
struck between two electrodes in the tube (Fig. 5.2). The electrons from
the negative electrode strike the water molecules and ionize them, and
these ions interact. To know what are the products of all these events, the
vapor can be routed for analysis to a mass spectrometer. The report on the
analysis shows clearly that the most abundant ionic species is HaO+. Of
course, there are other species, such as H50Z+' H70a+, H90~+' etc. (i.e.,
 PROTONS IN SOLUTION 465

MOLAR
REFRACTION

o '--~r:------;2:---'3~~4
NUMBER OF H ATOMS

Fig. 5.3. Molar refraction of isoelectronic series of

ions and molecules as a function of the number of hy-
drogen atoms in the ion or molecule. For the hydrogen
ion (Le., the proton in solution) to fall on the lowest
curve, the number of hydrogen atoms in the hydrogen
ion must be three.

HaO+ associated with 1,2, 3, etc., water molecules), but to smaller extents.
But, all this evidence for HaO+ ions in the solid and gas phases does not
prove that they exist in solution. Why not, therefore, use direct analytical
methods? The obvious tool to use is infrared spectroscopy, and the use of
this technique does permit the identification of HaO+ species in solution
(Falk and Giguere). A further clear piece of evidence comes from some
investigations of molar refraction. The refractive index of a solution is a
property which indicates the size and shape of the ions present. It is found
that, if a series of molar refractions are measured for an isoelectronic series
of ions and molecules (i.e., ions and molecules with the same number of
electrons), then the "proton in solution" only fits into the theory in the
right position if it is counted, not as a proton, but as an entity with three
hydrogen atoms, i.e., as an HaO+ ion (fig. 5.3).
To summarize: In solution, protons are unstable as isolatt:d entities
because their field is too powerful and their affinity for an electron pair
too strong. The result is that protons interact with water. Protons most
probably exist in solution as HP+ ions, but the evidence is not all that
certain. The fuller meaning of the statement, "exist as HaO+ ions," will
become clear when one considers the mechanism of proton drift in solution.
The shape and size of the HaO+ ion is fairly well established from NMR
data on solid acid hydrates. The HaO+ ion is a rather flat trigonal pyramidal
structure with the hydrogens at the corners of the pyramid and the oxygen
in the middle (Fig. 5.4). The O-H bond is found to be about 1.02 A in
length, the proton-proton distance is about 1.72 A, and the H-O-H
angle is 115°. The whole structure is similar to that of the ammonia molecule.
466 CHAPTER 5

~"~'1.O21
_--- I - ___ _

H~H
, \ 1'2-
H;::::-- .

Fig. 5.4. The trigonal-pyramid structure of the hydrogen ion HaO+.

5.2.2. Proton Affinity

The formation of an HaO+ ion is an expression, on the one hand, of
the instability of a free proton in solution and, on the other, of the affinity
of a water molecule for a proton. What is the energy change involved in
the protonation of water, i.e., in the reaction

To get at this energy, one has to resort to thermodynamic cycles. The

cycle suggested by Sherman is shown in Figure 5.5. Since all the quantities

Proton

~:f~:fl.~

-QIItP-1tDHz-iDx2
/
-IH-E x
~H20+H+X~

Fig. 5.5. Thermodynamic cycle for the calcula-

tion of the proton affinity PH,o of water. Here,
0HaO X is the heat of formation of HaOX (X is a
halide ion); U is the lattice energy of the halide,
0H2 0 is the heat of formation of water; DH2 and
DX2 are the dissociation energies of the hydrogen
and halogen molecules, respectively; 'H is the ion-
ization energy of hydrogen atoms; and Ex is the
electronegativity of the halogen.
 PROTONS IN SOLUTION 467

in the cycle, except the proton affinity, can be obtained directly from exp-
eriment or indirectly from other cycles, the one unknown, i.e., the proton
affinity of water, can easily be determined; it turns out to be about 170 kcal
mole-t.
It is this large energy which makes the existence of free protons in
solution unlikely. Their hypothetical concentration in aqueous solutions
at room temperature turns out to be about lO- t50 g of ions per liter, about
as zero as one can get.

5.2.3. The Overall Heat of Hydration of a Proton

The energy change associated with the interaction between a proton
and a single water molecule has been calculated as 170 kcal mole-t. Is this
the hydration energy of a proton? No. If one examines (Fig. 5.5) the
thermodynamic cycle used for the calculation, it will be seen that the H30+
ion formed by the interaction of H+ and H 20 has been left in the gas phase.
But, the heat of hydration of a proton (cf Chapter 2) is the heat evolved
when the proton is transferred from the gas phase into the solvent. Hence,
it is not enough to know the heat of interaction between a proton and a
water molecule; one must know the heat of interaction between the H30+
ion and the solvent.
It must now be recalled (cf Section 2.3.11) that there is experimental
support for Eq. (2.60)

L1HMi+[rel] - L1Hxi+[rel] = - 2L1HH+[abs] - 10 + (~~~Zitfs)3 (5.1)

where L1HMi+[rel] and L1Hxi -[rel] are the experimentally known relative
heats of hydration of a pair of oppositely charged ions Mi+ and X i - which
have the same radius ri; and L1HH+[abs] is the absolute heat of hydration
of the proton (i.e., the heat of interaction of the H + not only with one mole-
cule but with all of the surrounding water). Thus, by following Halliwell and
Nyburg and plotting the experimental values of HL1HM ,+[rel] - L1Hx-[rel))
,
versus (ri + 1.38)-3 and then extrapolating the resulting straight line to
infinite radius (cf Fig. 5.6), the intercept is equal to - L1HH+[abs] - 5.
By following this procedure, it was shown t that the absolute heat of hy-
dration of the proton ion, L1HH+[abs], is -266 kcal mole-t.

t It will be recaIled (el Section 2.3.11) that the value of - 266 kcal mole-I for the absolute
heat of hydration for the proton depends upon the approximation that the cavity-
formation and structure-breaking contributions to the absolute heat of hydration of
ions are independent of the radii of the ions.
468 CHAPTER 5

280

'.
"0
E
g
u
'"
.S
270
r:::;I
~
I
..
::a::
<I / Edrapalation
/
1
~ 260 --T-
\::a:: Intercept •
-AHpIObsl-1!
~
-IN

21!O 0 ()OOI ()O02 0·05 0'04

0-5
(rl + 1·38) in A

Fig. 5.6. Procedure for getting the abso-

lute heat of hydration of the hydrogen ion.
A plot of half the difference of the relative
heats of hydration of oppositely charged but
equiradii ions versus (ri + 1.38) -3 is extra-
polated to give an intercept which is equal
to - HH+(abs) - 5.

Now, if the proton affinity (Section 5.2.2) is subtracted from the

absolute (or total) hydration heat of a proton, one gets approximately
90 kcal mole-t. This, then, would be the heat of hydration of an H30+
ion. It seems to be roughly the same as the heat of hydration of a K + ion
(see Chapter 2, Table 2.15). Is this reasonable? Yes, because the radii of
the H30+ and K + ions are roughly the same.

5.2.4. The Coordination Number of a Proton

The fact that the overall hydration energy of a proton (266 kcal mole- t )
is larger than its energy of interaction with one water molecule (170 kcal
mole-t) indicates that the proton is engaged in interactions with more
than one water molecule, or, in other words, that an H30+ ion, itself, is
hydrated. But how many water molecules are engaged in this interaction
with an H30+ ion?
The number n of water molecules which interact with an H30+ ion can
be obtained by the following considerations. The H30+ ions have two main
effects on the water structure. Firstly, they cause some structure breaking,
 PROTONS IN SOLUTION 469

0 n=o
-I
T.
"0
-2
E -3 experimental
values
'"0E -4 00

i>
- -s
-6

o
Temperature. 'C

Fig. 5.7. Comparison between the experimental var-

iation of the molal volume of water with temperature
and the theoretical variation calculated on the basis
that the H30+ ion is associated with n = 0 and n = 3
molecules.

due to which there is a decrease in molar volume of the solvent (broken-up

water occupies less space than does the same water in a structured system).
Secondly, the hydration of H30+ ions influences the way in which an in-
crease in temperature affects the total volume. In the coordination shell of
an H30+ ion, the 11 water molecules are compressed owing to the electric
field of the ion (electrostriction, cf Section 2.4.5) and, thus, the molal
volume is decreased.
Thus, by measuring molal volume (or density) as a function of temper-
ature and comparing it with curves calculated on the basis that 11 = 0 or
11 = 3 molecules of water coordinating an H30t ion,. the coordination
number can be determined. It turns out that the experimental curves (Fig.
5.7) are best fitted by having 11 = 3, which means that the H30+ ion is
associated with three water molecules in an (H 9 0 4 )+ group. This group
has been detected in the gas phase. It involves a tetrahedral group of one
H30+ ion and three water molecules (see Fig. 5.8).
One can now summarize one's knowledge of the condition of a proton
in solution. (\) Free protons are highly unlikely in water and aqueous
solutions; (2) the association of a proton with a water molecule to form an
H30+ ion is accompanied by a large energy release-thus, the H30+ ion
appears to be the stable mode of existence of a proton in solution; and (3)
the H30+ ion itself interacts with three water molecules to form an (H 90 4 )+
cluster. t

t When the methods of determining primary hydration numbers (Section 2.4.1) are used
in solutions containing hydrogen ions, one obtains correspondingly a value of about
four.
470 CHAPTER 5

H 1
~ E.'ro
Ii
H2 0 mol.cul.
I
I

n _3 water molecu les associated

with H 3 0· ion to form H904+ group

Fig. 5.S. Schematic configuration of H 9 0 4 + group

shown with an extra H 2 0 molecule electrostatically
bound.

Further Reading
1. J. Sherman, Chem. Rev., 11: 98 (1932).
2. V. N. Kondratyev and N. D. Sokolov, Zh. Fiz. Khim., 29: 1265 (1955).
3. M. Falk and P. A. Giguere, Can. J. Chem., 35: 1195 (1957).
4. R. P. Bell, The Proton in Chemistry, Cornell University Press, Ithaca, N.Y.,
1959.
5. H. F. Halliwell and S. C. Nyburg, Trans. Faraday Soc., 56: 1126 (1963).
6. B. E. Conway, "Proton Solvation and Proton Transfer Processes in Solution,"
in: J. O'M. Bockris and B. E. Conway, eds., Vol. 3, Chap. 2, Modern Aspects
of Electrochemistry, Butterworth's Publications, Ltd., London, 1964.
7. L. L. Schaleger, P. Salomaa, and F. A. Long, in: Chemical Physics of Ionic
Solutions, John Wiley and Sons, Inc., New York, 1966.
8. P. Kebarle, J. Am. Chem. Soc. 89: 6393 (1967).

5.3. PROTON TRANSPORT

5.3.1. The Abnormal Mobility of a Proton

What one has been talking about so far are the equilibrium interactions
between a proton and an aqueous solvent. It is time now to think of a
nonequilibrium process such as proton conductance under an applied electric
field. One can begin the discussion with the question: When there is proton
 PROTONS IN SOLUTION 471

TABLE 5.2
Mobilities of Charge Carriers in Liquids and Solids

Mobility,
Particle
em 2 see- I V-I

Ions (e.g., K+) in water ~ 5 X 10- 4

Proton in water 3 X 10- 3
Ion (e.g., Li+) in ice ~ 10- 8
Proton in ice 10-'-1

transport, what is the entity which actually does the moving? Is it the proton
p, the hydrogen (or hydronium) ion H 3 0+, or the H 90 4 + cluster?
One way of resolving this question of the kinetic entity in proton drift
is to proceed by assuming that, whichever of the above species is the kinetic
entity, it drifts in the same way as other ions (e.g., K+) do (Chapter 4).
The resulting predictions can then be compared with the facts. The starting
point therefore is the picture of ions moving through the solvent at a velocity
which assumes a steady-state value when the electric driving force (zeoX)
is exactly balanced by the Stokes viscous force (6nr1]v). Thus, the ionic
mobility at concentrations so small that interionic forces are negligible is
given by the familiar expression based on Stokes' law

zeo
Uo = -:-::--:-::--- (4.180)
1800nr1]

But, what radius should one use? Free protons do not exist to any
significant extent in water and aqueous solutions, so it is logical to ignore
them in the first instance. Perhaps it is best to see what happens if the radius

TABLE 5.3
Transport Numbers of Cations in O.01N Chloride Solutions at 25 0 C

Ion Transport number

Hydrogen ion 0.8251

Li+ 0.3289
Na+ 0.3918
K+ 0.4902
Ba++ 0.440
La+++ 0.4625
472 CHAPTER 5

TABLE 5.4
Equivalent Conductivity at Infinite Dilution of HCI and LiCI at 20 0 C

Solvent

Water Methanol Ethanol n-Propanol

ARCI 426.2 192 84.3 22

ALiCI 115.0 90.9 38.0 18
ARCI - ALiCI 311.2 101.2 46.3 4

of the hydronium ion H30+ is used in the above expression for the limiting
mobility. Now, hydronium ions have roughly the same radii as potassium
ions, so one would expect that their mobilities would be approximately
the same, i.e., about 7.6 x 10-4 cm~ sec- l V-I.
The facts of proton transport, however, constitute a curious and inter-
esting story. Most important of all (see Table 5.2), the mobility of protons is
abnormally high. For most other ions in solution, the mobility is of the order
of 5 x 10-4 cm2 sec- l V-I; protons migrate with a limiting mobility of about
36 x 10-4 cm2 sec- l V-I. This abnormal mobility is reflected in the transport
number, too. Other positive ions in a I : I-valent electrolyte carry about 50%
of the total current; hydrogen ions, however, transport as much as 80%
of the total current (Table 5.3).
A number of other facts concerning proton mobility are as surprising.
Firstly, the abnormally high mobility is diminished to normal values if
water is replaced as solvent by some other substance. For example, in n-
propanol the conductance is quite normal (Table 5.4); and, in methanol-
water or ethanol-water mixtures, the abnormality in the conductance de-
creases with increasing alcohol content until, at about 80% alcohol, the
abnormal mobility is reduced almost to zero (Fig. 5.9). Secondly, the ratio
of the excess (or abnormal) mobilities between the hydrogen and deuterium
ions in water is 1.4 at 25°C (Table 5.5)-much more than might be expected

TABLE 5.5
The Mobilities at Infinite Dilution of Hydrogen and Deuterium Ions

Ion Infinite dilution mobility

Hydrogen ion in H 2 0 36.2 x 10- 4

Deuterium ion in 0 2 0 25.1 x 10- 4
 PROTONS IN SOLUTION 473

400

300
o
.t::
E AO
• 200 A O-OOI
<: -Au-Ol

100

Mole per cent methanol

Fig. 5.9. The equivalent conductivity of Hel in

methanol-water mixtures decreases with increasing
alochol content and reaches a minimum at about
80 mole % ot alcohol. At this concentration, the
abnormal mobility, i.e., the mobility of the H30+
ion compared with that of the K+ ion (which is of
similar size), is almost reduced to zero.

TABLE 5.6
Variation of Heat of Activation for Proton Transport in Aqueous Solution

Mean heat of activation

Temperature,
oK between temperatures shown.
kcal mole- 1

273
2.822
291
2.482
298
1.907
323
1.430
348
0.886
373
0.636
401
0.211
429
474 CHAPTER 5

on a Stokes' law basis since the two ions are virtually the same size. Thirdly,
the proton mobility shows a temperature coefficient which indicates that
the heat of activation decreases with temperature (Table 5.6); in the case
of most other ions, a single activation energy is good enough over an ap-
preciable range of temperature.

5.3.2. Protons Conduct by a Chain Mechanism

One fundamental conclusion can be drawn from this anomalous
behavior of migrating protons. Protons must conduct by a mechanism
which is radically different from that used by other ions. In particular, one
cannot use for protons the conventional Stokes' law approach with the
limiting mobility of ions' being decided by their radius and the viscosity
of the medium. This is because the predicted mobility is far too low if one
assumes migration of the stable H30+ ion; and, of course, if one assumes
bare protons drifting by a Stokes mechanism, the calculated mobility comes
out to be much too high because of the extremely small size of the proton.
Obviously, therefore, the atomistic picture of proton conduction has to
be sketched anew but not with the Stokes' law-oriented model used to
explain the drift of other ions.

(0) (b)

j
H3 0 + ION ACCEPTOR WATER
MOLECULE tttJ+ ION

~~ ~
BEFORE
JUMP

ELECTRON
PAIl! -.~,~

4
WATER MOLECULE H~+ ION

~
AFTER
JUMP
H ,-- r@'1d
WATER
MOLECULE ~O+ ION

Fig. 5.10. Two schematic views (a) and (b) of a water molecule
adjacent to an HaO+ ion. The free electron pair (orbital) of the 0 of
the water molecule is oriented along the 0 (of HaO+)-H+-O (of H 2 0)
line. The jump of the proton H+ from the HaO+ ion to the water mole-
cule converts the water molecule into an HaO+ ion and the HaO+ ion
into a water molecule.
 PROTONS IN SOLUTION 475

,
H H
I
I JjI
H
H+ transfer

H
I
I
I
I
H
I
,
H
1 I. I I
--H -O-H--o-H--O- H--O-H--O-H--O-H---
I I (O)
1 II 12 13 14 I
I I , I I I
~ H H H H H'1
~ 1
--H-O-
1 I. " I 1 1
H--O-H--O- H--Q-H--O-H--O-H-- (b)
I I' 12 13 14 I
I I I I I I
H H H H H H
1 1 1 I." 1
--H- O-H--<?-H--O-H--O-H--O-H---o-H--
1 (C)
! " 13 12 :4 :

Fig. 5.11. If there are a series of proton jumps down a line

of water molecules, the net result is equivalent to the migra-
tion of an H30+ ion (indicated by a charge on the oxygen)
along the line.

To approach an alternative view, one can start by considering the

environment of a hydrogen ion in solution. The chances are that there is
a water molecule next to the hydrogen ion. Assume also that the mutual
orientation of the H30+ ion and the H 2 0 molecule are as shown in Fig.
5.10. Notice that there is an attractive unattached orbital a short distance
away from one of the protons of the H30+ ion. Suppose, therefore, that the
proton jumps from the charged H30+ ion at the left to the neutral H 2 0
at the right. What is the net result of this jump? The erstwhile-neutral parti-
cle at the right has become a charged H30+ ion, and the erstwhile-charged
particle at the left has become a neutral water molecule. So the proton
jump has produced an effect which is equivalent to the translational move-
ment of an HaO+ ion and, therefore, to the movement of charge.
Now consider a whole row of water molecules, and initially let the
hydrogen ion be at the extreme left (Fig. 5.11). If the proton transfer is
from 0 1 to O 2 , O2 to 0 3, ... , right down the line, then the net effect is
equivalent to a H30+ ion's moving from left to right. Further, it is also
equivalent to the migration of charge from left to right.
If the proton transfers are in response to an electric field, a proton
conduction mechanism has been evolved. But this is an entirely different
kind of mechanism for charge transport in solution t from that discussed
previously. There is no movement of an ion as a whole. One water molecule

t It is interesting to note that Grotthus suggested, in 1806, the idea of the transfer of
charge down a chain of particles as a general mechanism of conduction in ionic solutions.
His view was later surpassed, but it becomes cogent again just for this special case of
proton migration.
476 CHAPTER 5

passes on a proton to the next water molecule, which passes on a proton to

the next molecule, and so on-a passing-the-proton game.
The difference between this mechanism of charge transport in solution
and the former one discussed shall be stressed. Thus, other ions have to move
bodily from one point in the solution to another; the HaO+ ion does not
have to move as a whole to transport charge. But the repetition of elementary
proton jumps not only results in charge transport but also gives the same
effect (if not the same rate) as HaO+ migration.
Another interesting point is that the proton jump mechanism does not
violate the idea that protons exist in solution as HP+ ions (coordinated by
other water molecules) and not as free protons. For it will be noticed that
the hopping protons have no independent existence except during the
actual jumps, and these will be shown (Section 5.3.11) to take up little time
in comparison with the time the protons remain attached to one or another
water molecule.

5.3.3. Classical Proton Jumps and Proton Mobility

An alternative mechanism of charge transport in solution, specially
made for protons, has been suggested. Now one must consider some quanti-
tative aspects of the picture to see whether they fit the facts well enough to
justify acceptance of the model. The crucial task, of course, is to obtain a
recognition of the factors which decide the mobility of the protons because
it is the value of this quantity which provides the major anomaly in compar-
ison with values expected from the previous model.
If the new model outlined in this chapter is correct, the mobility ob-
viously depends on the rate at which protons jump from water molecule
to water molecule. The basic theory for this rate of particle jump should
be the theory of rate processes (Appendix 4.3). It will be recalled that, for
a successful jump to occur, the proton has to be energized with a critical
activation energy before it can climb an energy barrier which represents
the change of energy of the system (two water molecules plus the jumping
proton) as the proton moves from the first water molecule to the second.
Once the activation energy is known, the mobility can be estimated (see
Appendix 4.3).
The rigorous way of obtaining the activation energy is to calculate
the energy of the system as the proton jumps from one water molecule to
the next and then plot a potential energy surface. This is a difficult, laborious
task. There is, however, an easier, if more approximate, approach. In the
initial state, the proton p is bonded to one particular water molecule (H 2 0).
 PROTONS IN SOLUTION 477

/ROTON
~
{HzOl1 H
o

I
_

I
I
-210 I
I
-220
POTENTIAl..
ENERGY -230
OF
H2O -H+
(Kcal mole-') -240

-250

-260

0·9 1·3 1·5

DISTANCE OF PROTOli FROM
OXYGEN ATOM (IN 1\ )

Fig. 5.12. Variation of the potential energy

of the (H.OlI-p system with the stretching
of the (H.O),-p bond.

and, as an approximation, the (H 20)[-p system may be treated as a pseudo-

diatomic system. As the (H 20)[-p bond is stretched, the energy of the
system increases. Since the stretching of diatomic molecules has been
studied in spectroscopy, one may as well borrow from these studies the
expression for the energy Ur of the diatomic system as a function of the
distance apart, r, of the two "atoms." This expression is known as the
Morse equation

where Do is the energy to dissociate the H 20-p bond, a is a constant, ~nd

re is the equilibrium distance between the p and H 20 in H30+. For Do,
the overall hydration energy of the proton is used, i.e., about 266 kcal mole-I;
and, for r e' the distance is 0.98 A.
Thus, one obtains (Fig. 5.12) the potential energy of the (H 20)cp
system as the proton moves away from the water molecule. The energy
versus distance plot is known as a Morse curve. Because of the symmetry
of the system, one has an identical curve for the approach of the proton
to the second water molecule.
The two Morse curves, one for the (H 20)cp and the other for p-(H 20)2'
are then put together. The problem of the relative positioning of the minima
of the two Morse curves is met by assuming a reasonable model based on
 478 CHAPTER 5

Fig. 5.13. Model for proton transfer between an

HaO+ ion and a favorably oriented H 2 0 molecule.

a knowledge of bond lengths, etc., such as those shown in Fig. 5.13. When
the two minima are thus fixed, it is seen that the two Morse curves intersect
(Fig. 5.14) and a potential-energy barrier has been synthesized. t
This then is the energy barrier for the jump of a proton from one water
molecule to another over a potential-energy barrier. From the barrier, one
can calculate the energy of activation and thus the rate of proton transfers.
It turns out that the mobility calculated according to this model has the
right order of magnitude in comparison with experiment.

5.3.4. Do Proton Jumps Obey Classical Laws?

Numerical agreement between the calculated and observed values of
a principal experimental quantity is an important step in giving credence
to a model. But it is not a sufficient one. In addition, the model has to be
in accord with all the other experimental observations. Thus, in the context
of the question of proton mobilities, the model which has been described
above must explain the observed facts on the ratio of the mobilities of the
H30+ and D30+ ions and on the temperature dependence of the mobility
of the proton.
When the comparison is made, it turns out that the model of simple
classical jumps of protons outlined above predicts that H30+ ions in H 20
should travel about 14 times faster than D30+ in D 2 0 in comparison with
the experimental ratio UHaO+/UDaO+ of about 1.4. Further, as the temperature
is raised from T2 to T1 , the distance between the minima in Fig. 5.14 is

t The procedure used involves a number of unstated approximations. Thus, systems

interact near the intersection (Fig. 5.14) so that there is not a sharp linear type of
intersection, as shown, but a rounding-off, and resultant reduction of the heat of
activation. Further, the zero-point energy levels are not shown in Fig. 5.14, and no
discussion has been given of the effect of the environment on the motions calculated.
 PROTONS IN SOLUTION 479

Til
~
...
Ci
u

E
.J!
-220
~
T
a. -230
L

-~..
o
'"
-240

.
I!' -250
c
:2 -260
C sta 1
~
&. 0·8 10 1·2 Iii
distance from 0 alom of
(H20), in 1

Fig. 5.14. Potential-energy barrier for proton

transfer from H30+ to H2 0.

increased, which leads to an increase in the energy of activation (minimum

to crossing point) for proton jumps (Fig. 5.15). Experimental observations
do indicate that the heat of activation changes with temperature; but it is
found that the heat of activation decreases with temperature, i.e., it changes
in a direction opposite to that predicted.
Thus, after the initial encouragement at finding a model which gives
the correct order of magnitude from the anomalous proton mobility, one
begins to have doubts and hence second thoughts.
The rate process theory presented in Chapter 4 and Appendix 4.3 was
developed for processes involving ions which are massive compared with
protons. Is it right to use this classical theory to treat the transfer of a
relatively light, elementary particle such as the proton? Thus, in the mobility
calculation of the previous section one insisted upon the proton's climbing

Fig. 5.15. Schematic diagram to show that

the activation energy for proton transfer
depends upon the distance between the
minima of the Morse curves.
480 CHAPTER 5

over the barrier for, in the usual treatment of rate processes, the only way
the moving particles can cross the energy barrier is by having the necessary
energy to surmount it.
Classical mechanics is an approximation valid for objects which are
large compared with their de Broglie wavelengths. Particles which are suf-
ficiently small and light do not behave according to classical laws. Their
motions are best described by the laws of quantum mechanics. One of
these laws is that the position of a particle cannot be defined with cer-
tainty; t one can only ascribe a certain probability to the particle's being in
a certain region of space. This means that, even if a particle has less than
the critical energy to surmount the barrier, there is a finite probability that
it crosses the barrier. It is customary to refer to this phenomenon as quan-
tum-mechanical tunneling; one says that the particle has leaked through the
barrier rather than climbed over it.
It can be shown that the velocity with which tunneling occurs increases
exponentially as the mass of the particle decreases. This means that tunnel-
ing is particularly important for light particles, in particular, therefore, for
electrons. (It will be seen in Chapter 8 that tunneling of electrons from an
electrode to particles in solutions constitutes the basis of charge transfer at
electrified interfaces.) Can tunneling be significant also for the proton
jumps which are suggested to be the basis of proton conduction in solutions?

5.3.5. Quantum-Mechanical Proton Jumps and Proton Mobility

The calculation of the rate of quantum-mechanical proton transfer

from hydrogen ions to water molecules was carried out in detail in a classical
paper by Bernal and Fowler and published in the same paper containing
their structural theory of ion-solvent interactions. Although the calculation
is of historic importance to the theory of proton conduction in aqueous
solution, the result proved to be rather disappointing-protons tunnel too
fast. In fact, the mobility of protons calculated on the basis of tunnel trans-
fers is about one hundred times faster than the observed value.
One can summarize the position at this stage. Migration of H30+ ions
by a conventional Stokes mechanism yields too Iowa mobility, and thus
stresses the need for an alternative model for the mechanism of proton
migration. Thus, a chain mechanism was conceived with an elementary
step of proton transfer between hydrogen ions and water molecules. It is
of importance, however, to understand what individual step in this migration

t The criterion of when to use quantum mechanics has been given in Chapter 1.
 PROTONS IN SOLUTION 481

mechanism controls the rate of proton transfer. Classical proton transfer

gives the right order of magnitude for the mobility of protons but disagrees
with other experimental facts. Fundamental considerations suggest the
probable presence of quantum-mechanical tunnel transfers. Tunnel transfer
through the barrier is possible, and calculations indicate that it is easier
than classical over-the-barrier transfer. Hence, tunneling should be con-
sidered the predominant mode of proton transfer from hydrogen ions to
water molecules. But calculation shows it is too fast to explain the observed
mobility. The analysis is apparently incomplete.

5.3.6. Water Reorientation, a Prerequisite for Proton Jumps

It has been assumed so far that the only factor in proton transfer is
the actual jump itself and that no preconditions must be satisfied. However,
from Fig. 5.16, it is clear that a proton from an H30+ ion cannot jump
indiscriminately to a water molecule. It can only jump when the oxygen
atom of the H 20 which receives the jumping proton is in a favorable orien-
tation. The point is that the oxygen of the water molecule can form bonds
with protons only along certain directions in space. These directions corre-
spond to the directions of the vacant orbitals. Thus, the proton jump can
occur only if the acceptor water molecules orient so that the oxygens,
the proton, and the unshared electron pair of the acceptor oxygen line up.
Then, the proton jumps along the pathway leading to the oxygen orbital.
It is as though the journey of the proton from the H30+ ion to the water
molecule starts with the tunnel through the barrier but is complete only
when, on emerging from the tunnel, the proton crosses a swing bridge.
But the latter obstacle in the journey can be overcome only if the swing
bridge is in line.
Now the analysis of proton mobility can be completed. The tunneling

(ol PROTON JUMP POSSIBLE (bl PROTON JUMP NOT POSSIBLE

Fig. 5.16. Schematic diagram to show that the transfer of a

proton from an H.O+ ion to a water molecule is possible in con-
figuration (a), but not in configuration (b).
482 CHAPTER 5

;JHOJ?
cffJ
THIS PROTON 3

CANNOT IUMPiJ I
'I)
~+
WATER MOLECULE !~ WATER MOLECULE
HAS TO ROlllTE - IN ORIENTATION
PRIOR TO PROTON FAVOURABLE FOR
TRANSFER I ..., ", PROTON TRANSFER
, ,....
,1"
.........
"-
I

(a) (b)

Fig. 5.17. In (a), the proton from the HaO+ ion cannot
jump to the water molecule because the latter is in an
unfavorable orientation. When the water molecule reorients
and is in a configuration corresponding to (b), the proton
jump can occur.

of protons will give too high a mobility only if the water molecules are
always oriented correctly. If, however, the orientation is "incorrect," then
the protons will have to wait till the water molecule turns around and, only
after this orientation process is complete, can the jump occur (Fig. 5.17).
So there are two processes which must cooperate to give the successful
proton transfers which are the basis of proton mobility. First, there is
water reorientation, and, then, the proton tunneling. Hence, the rate of
proton transfers will be limited by whichever of the two processes is slower.
One must therefore investigate whether water reorientation is the rate-
determining step in the process of proton transfer (for the tunneling through
the barrier has already been shown to be too fast to be consistent with the
observed mobility).

5.3.7. The Rate of Water Reorientation and Proton Mobility

The calculation of the specific rate of water reorientation is quite a

complex task. This is because one cannot consider the reorienting water
molecule as an isolated entity. If that were so, then one could work on the
basis of the rate of rotation of a gas molecule and calculate the rate. How-
ever, the water molecule is hydrogen bonded to other water molecules, and,
therefore, the reorientation involves the stretching and breaking of the
hydrogen bonds which bind the molecules together.
The calculation was nevertheless first performed by Conway et af. in
1956. At first, this calculation proved a disappointment because, if the
tunneling protons had to wait for the water molecules to turn around and
 PROTONS IN SOLUTION 483

get into the right position every time a proton arrived on the scene (i.e.,
turn around to present a free orbital to the next oncoming proton), then
the predicted mobility came out low compared with experiment. The analogy
has been made to a swing bridge which has to be in a position to receive
the proton tunneling through the barrier. There are two ways in which the
bridge could swing into position. It could do so spontaneously by random
motions; this would correspond to the acceptor water molecule's reorient-
ing by the thermal motions of the molecular world. It was this thermal

I IT m

¢, C;~[i'~~ ¢
(:t.
~or ~~I
,I,

(H20)6 hos
(H20F musl l'8-OI'ient re-orienled
to receive proton b

this prolon has

jumped

Fig. 5.18. A detailed schematic representation of the proton mobility model in

which proton tunneling has to be preceded by an induced reorientation of the ac-
ceptor water molecule. I. The proton d of the central HaO+ ion is transferred to (H 2 0)';
thus the HaO+ ion becomes (H 2 0)" and (H 2 0)' becomes an H30+ ion. II. The proton X
of an HaO+ ion (not shown in the diagram) arrives next to (H 2 0)' but cannot be
transferred to (H 2 0)' because the latter is in an unfavorable orientation. III. The
water molecule (H 2 0)' reorients so that the H atom b which was blocking the jump
of X is moved away to face H atom c of (H 2 0)". Another event is the jump of proton
Y of the HaO+ ion to an adjacent water molecule (not shown in diagram). IV. Proton X
jumps to (H 2 0)' which now becomes an HaO+ ion. Note that this newly created HaO+
ion cannot transfer proton b to the central water molecule (H 2 0)" because the last
is unfavorably oriented. V. The central water molecule (H 2 0)" reorients, and proton b
can now be transferred to it. VI. After the jump of proton b, the central water molecule
has again become an H30+ ion, and the H30+ ion, a water molecule (H 2 0)'; ct. step I.
484 CHAPTER 5

reorientation mechanism that Conway et al. found too slow to account for
the mobility. Alternatively, it could be automated to turn at the approach
of a vehicle: the approach of the proton can trigger a mechanism which
turns the water molecule around and makes it ready to meet the tunneling
proton. It is precisely this field-induced reorientation of the water molecules
that is necessary to explain the proton mobility. What happens is that,
as the proton approaches the dipole, its coulombic field attracts the water
dipole and drags it around into position (Fig. 5.18). Though the rate of
this field-induced water reorientation is faster than the rate of the spontane-
ous thermal rotation, it turns out to be much slower than the proton tunnel-
ing rate. Thus, it is the field-induced rotation of water that determines the
overall rate of proton transfer and the rate of proton migration through
aqueous solutions. The estimated value of the proton mobility obtained
according to the theory presented is 28 X 10-4 and that observed exper-
imentally as an abnormal contribution to the conductance is 36 X 10-4 cm2
sec-l V-l. It can be shown (see Section 5.3.9) that this model allows a
consistent series of interpretations for most of the facts of proton mobility.

5.3.S. A Picture of Proton Mobility in Aqueous Solutions

This chapter began with the view that the proton was an unorthodox
ion, and this is indeed true as regards the mechanism by which it goes
through solutions. The outstanding anomaly to explain was its sheer speed
under an applied electric field. It migrates about seven times faster than any
other ion. As one is used to ionic mobilities' being inversely proportional
to the radius, the first attempt at interpretation is in terms of drifting HaO+
ions, but the mobility turns, out to be too low. The temptation to explain
the mobility by thinking of drifting free protons cannot be yielded to, not
only because anything but a momentary nonattachment to water is surely
very unlikely for the proton, but also because, were it to be regarded in
this way, the model would predict too high a mobility compared with what
is observed.
So, the ordinary Stokes mechanism of ions drifting through the solution
does not tie up with experiment. One has to turn to other possibilities, and
an attractive one is a chain mechanism in which protons leap from the
oxygen of an HaO+ ion to the oxygen of an adjacent water molecule. The
model in which proton mobility is controlled by classical over-the-barrier
proton jumps between water molecules does not give results consistent
with all the experimental facts. Further, though the proton is much heavier
than the electron, it is still small and light enough (compared with other
 PROTONS IN SOLUTION 485

TABLE 5.7
Comparison of Different Models of Proton Mobility

Model and Absolute mobility, IIH+

rate-determining step cm' sec l V-I IID+

Classical proton transfer 1.24 x 10- 2 6.1

Proton tunneling 75.8 x 10- 2 13.9
Field-assisted reorientation of water
molecules followed by fast tunneling 0.28 x 10- 2 1.42

ions) to leak through the barrier in quantum-mechanical fashion. This

tunneling, with its noncommon sense laws, turns out to be a much faster
way of jumping than the classical path of going over the barrier. If, how-
ever, it is assumed that quantum-mechanical proton jumps determine proton
mobility, then the predicted overall mobility proves to be too high and one
must think of some precondition for proton transfer which limits the
tunneling rate. This other step is the rotation of a water molecule so that
its free orbital is in line with the proton which wishes to tunnel to it (particles
which tunnel must have a bond, or acceptor, state to tunnel to). Calculation
shows that this model does turn out to be in reasonable accord with the
facts-not too slow as is the Stokes transport and not too fast as is simple
tunnel transfer unfettered by consideration of the availability of acceptor
states. So the model which provides the best interpretation of the high
mobility of protons is that in which they tunnel from an H30+ ion to a
water molecule but the tunneling rate is limited by the rate at which the
acceptor water molecules reorient so that their free orbitals face the tunnel-
ing protons (Table 5.7).
One further remark must be added to this picture. For 96% of its life,
a proton exists attached to a water molecule, i.e., the H30+ ion has an
independent existence. Hence, in addition to the water-reorientation-fol-
lowed-by-proton-tunneling mechanism just described, there is also the usual
Stokes type of drift of H30+ ions. In fact, the Stokes transport of H30+
ions accounts for about 20% of the total mobility of protons.

5.3.9. The Rate-Determining Water-Rotation Model of Proton

Mobility and the Other Anomalous Facts
The numerical value of the high mobility of protons has been explained
reasonably well. What of the other anomalies listed in Section 5.3.1?
486 CHAPTER 5

The anomalous decrease of the heat of activation with increase of

temperature follows from the model because the mechanism depends on
the fact that there is available a hydrogen-bonded liquid with a consider-
able degree of ordering in its structure. Increase of temperature causes
increase of disorder in the water structure, and, consequently, there are,
on the average, fewer H bonds to break when water molecules reorient.
Since the water reorientation becomes easier as the temperature increases,
the heat of activation becomes smaller. An explanation can also be advanced
for the decrease of the anomalous mobility of the proton in the presence
of added alcohol solvents: The new bonding possibilities which arise when
alcohol is added to water make the alcohol reorientation more difficult
than that for water and thus cause a fall in the proton mobility by the
tunneling and solvent-orientation method.
It must be noted, however, that any proton-conduction mechanism in
which the rate-determining step involves the breaking of hydrogen bonds
would succeed equally well in explaining the above two anomalies; they
are therefore not stringent tests on the model. In contrast, the 1.4 ratio
of the mobilities of hydrogen and deuterium is a diagnostic criterion for
the various models of proton mobility, and, therefore, the correct calculation
of this ratio (as the model indeed makes possible) is strong evidence in
favor of a mechanism involving water reorientation as a precondition for
proton tunneling. There are, in addition, several other phenomena which
can be quantitatively interpreted in terms of the water-reorientation-tunnel-
transfer model. Mention may be made of the anomalous mobility of hy-
droxyl OH- ions compared with that of protons; and interpretation of the
mobility of NH4+ and NH 2- in NH a . These various interpretive successes
in related areas increase the plausibility of the water-rotation-controlled,
proton-tunneling mechanism in explaining the abnormal mobility of protons.

5.3.10. Proton Mobility in Ice

It is the matter of the proton mobility in ice which has been argued
by Conway et af. as providing the most striking evidence in favor of the
water-rotation, proton-tunneling model. The fact is: Ice exhibits a higher
proton mobility than water does. Does this mean that water molecules
turn faster in ice than in water? This appears most unlikely. Is the model
wrong, then? The following fact is the bridge to enlightenment on the matter.
The number of protons per cubic centimeter in ice is much less than that in
the usual systems in which measurements are made, namely, water or
aqueous solutions. This means that, when one measures the proton mobility
 PROTONS IN SOLUTION 487

in ice, the flux of protons across a given piece of the system is less than that
in water or in aqueous solutions. Now, recall what was said about swing
bridges which swing spontaneously but gradually back into place and the
modern ones which are automated to be ready, after a pause, when a vehicle
approaches them.
Suppose the number of vehicles approaching the bridge grew very
much smaller than had been the case when it was arranged for the bridge
to be back in place when a vehicle arrived. Then it would have time to
swing back by itself without help from some extra mechanism. There
would be no pause for the swingback accelerated by an approaching vehicle.
Thus, for protons in ice, there are so few protons that the thermal rotation
rate of the water molecules in ice is sufficient to bring the molecules back
into the position to receive the tunneling protons. They are always ready
in time in ice. If one may mix metaphors a little, for protons which so
infrequently come through the tunnels in the ice, the lights are always
green; the protons do not have to wait for the orbitals to turn around
and be ready to receive them, i.e., their mobility is no longer determined
by the rate of water rotation. Hence, the tunneling itself becomes the rate-
determining step; but this tunneling is a much faster process than that of
water rotation, so that it leads to higher mobilities of protons in ice than
in the water or aqueous solutions in which the water rotation is rate deter-
mining.

5.3.11. The Existence of the Hydronium Ion from the Point of

View of Proton Mobility

The rapid proton transfers from H30+ ions to water molecules clearly
imply that H30+ ions cannot be permanent entities in aqueous solution.
The moment an H30+ ion loses its proton to a neighboring water molecule,
it is transformed into a water molecule. However, the time taken in the
transfer process is about 10-14 sec, while the time an H30+ ion waits for
its adjacent water molecule to reorient is about 2.4 X 10-13 sec. Since, there-
fore, the proton spends some 96% of its time linked to a definite molecule
in an H30+ species rather than as a tunneling proton, it can be concluded
that H+ ions in aqueous solution have an effective existence 3S H30+.

5.3.12. Why Is the Mechanism of Proton Mobility So Important?

Protons are central to the electrochemistry of solutions; they are the

basis of the concepts of pH and of acids and bases. But that is not the only
reason why their state in solution is so important. Another reason stems
488 CHAPTER 5

from the central question: How are electric charges transported in solid
biological systems? A great many biological systems require concepts
connected with electrochemical charge transfer for an explanation of their
behavior. A difficulty is in explaining how electric current gets carried
within the macromolecules. Now, in the systems concerned, there is much
H bonding; recall the protein helix held together by hydrogen bonds. It is
a speculation, but perhaps one of interest, that the charge transport in
such systems is by protons, perhaps tunneling protons, not electrons. Hence,
any mechanism by which protons get carried along particularly quickly,
especially an H-bonded solid (e.g., ice), has important implications for a
large area of molecular biology.

Further Reading
I. J. D. Bernal and R. H. Fowler, J. Chern. Phys., 1: 515 (1933).
2. B. E. Conway, J. O'M. Bockris, and H. Linton, J. Chern. Phys., 24: 834 (1956).
3. M. Eigen and L. de Maeyer, Proc. Roy. Soc. (London), A247: 505 (1958).
4. B. E. Conway and J. O'M. Bockris, J. Chern. Phys., 31: 1133 (1959).
5. B. E. Conway, "Proton Solvation and Proton Transfer Processes in Solution,"
in: J. O'M. Bockris and B. E. Conway, eds., Modern Aspects of Electro-
chemistry, Vol. 3, Butterworth's Publications, Ltd., London, 1964.
6. B. E. Conway and M. Salomon, in: B. E. Conway and R. G. Barradas, eds.,
Chemical Physics of Ionic Solutions, John Wiley and Sons, Inc., New York,
1966.
7. R. G. Wawro and T. J. Swift, J. Am. Chern. Soc., 90: 2792 (1968).
8. T. A. Stephenson, T. J. Swift, and J. B. Spencer, J. Am. Chern. Soc., 90:
4291 (1968).

5.4. HOMOGENEOUS PROTON-TRANSFER REACTIONS AND

POTENTIAL ELECTROLYTES

5.4.1. Acids Produce Hydrogen Ions and Bases Produce Hy-

droxyl Ions: The Initial View

It is common knowledge that aqueous solutions of a certain class of

substance, called acid, corrode so-called "active" metals and turn blue litmus
paper red. In 1887, Arrhenius suggested that this characteristic behavior
of the class of substances called acids arises from the common characteristic
that they produce hydrogen ions in solution. Correspondingly, Arrhenius
suggested that hydroxyl ions are responsible for the well-known properties
of aqueous solutions of alkalis or bases, namely the ability to turn red
litmus paper blue and neutralize acids.
 PROTONS IN SOLUTION 489

According to this classical view, therefore, acids are substances which

dissolve in water to give hydrogen ions, and bases are substances which
dissolve to give hydroxyl ions. It follows that, in order to understand acids
and bases, one must understand the mechanism of the production of hy-
drogen ions and hydroxyl ions.

5.4.2. Acids Are Proton Donors, and Bases Are Proton Accep-
tors: The Bronsted View

A convenient starting point for understanding the mechanism of the

production of hydrogen and hydroxyl ions is the mechanism of proton
conduction in aqueous solutions. The elementary step in this mechanism
turned out to consist of proton transfers between H30+ ions and water
molecules. Notice, however, that the water molecule which accepts the
jumping proton is converted into a hydronium ion. One can now introduce
an obvious terminology. In the transfer of a proton from an H30+ ion to
a water molecule, the H30+ ion can be termed the proton donor, and the
H 2 0, the proton acceptor (Fig. 5.19).
The H30+ ion is not the only possible donor of a proton to a water
molecule; there are many other substances which can act as proton donors.
Consider, for example, acetic acid and sulphuric acid. Reactions such as
those shown in Fig. 5.20, in which there are proton transfers to water
molecules, result in the production of H30+ ions. But substances which
dissolve in water to produce H30+ ions are historically known as acids.
At this stage, therefore, one would define an acid as a donor of protons to
water.
Just as one wondered whether the H30+ ion was the only possible
proton donor, it is necessary to consider whether water molecules are the
only proton acceptors. Then, it becomes clear that, since free protons are un-
stable in solutions, one substance cannot donate protons unless there is an-

proton donor:

~'--~ n ,

'-.'
,
, ....:t;

proton acceptor

Fig. 5.19. The proton donor and proton ac-

ceptor in the proton transfer from an HaO+ ion
to a water molecule.
490 CHAPTER 5

acetate ion

Fig. 5.20. The production of HsO+ ions by the transfer of

protons from proton donors to water molecules.

other substance to accept them. Hence, the reactants in any protolHransfer

reaction must consist of one proton donor and one proton acceptor.
Consider the following reaction between ammonia and water

(5.3)

It will be noticed that, in this proton-transfer reaction, water molecules

are the proton donors and the ammonia molecules act as proton acceptors.
Further, the result of the reaction is the production of hydroxyl ions. This
is interesting because Arrhenius ascribed the characteristic behavior of
bases to hydroxyl ions. So one can at this point define a base as a substance
which accepts protons from water molecules in a proton-transfer reaction.
The classical (or Arrhenius) picture of acids and bases has been
presented above as proton-transfer reactions involving water: Acids donate
protons to water molecules to produce HaO+ ions, and bases accept protons
from water molecules to form OH- ions. t The trouble with this picture is
that it is linked too firmly with the solvent water. For instance, what is to
be done with the dissolution of acetic acid in liquid ammonia (Fig. 5.21)?
It is obvious that this reaction involves proton transfer between a proton
donor and a proton acceptor. In fact, it is the analogue in liquid ammonia
of what happens when acetic acid is dissolved in the particular solvent, water.

t Substances which can function either as proton donors (Le., acids) or as proton ac-
ceptors (Le., bases) are termed amphoteric (or amphiprotic). One of the reasons why
water is such a widely used solvent is that it is amphoteric and therefore enters into
proton-transfer reactions with either proton donors or proton acceptors.
 PROTONS IN SOLUTION 491

~ j~m,"i' acetate ion

'" t,j-<J
O-H N-H

H:sC-C, / H -

/ 0
+

[ /k]
acetic acid

+
H H H

I ..
ammonium Ion

Fig. 5.21. The proton transfer reaction between

acetic acid and liquid ammonia.

The fundamental similarities between proton-transfer reactions in

various solvents led Bronsted to suggest that the definition of acids and
bases be freed from the solvent water. He proposed that, irrespective of
the solvent in which the proton-transfer reaction takes place, the proton
donor be termed an acid and the proton acceptor, a base.

5.4.3. The Dissolution of Potential Electrolytes and Other Types

of Proton-Transfer Reactions

In the introduction to the treatment of ion-solvent interactions, it was

stressed that electrolytes dissolve in a solvent by two fundamentally dif-
ferent mechanisms: True electrolytes undergo physical interactions with
the solvent molecules and form solvated ions; potential electrolytes engage
in chemical reactions with the solvent molecules and give rise to ionic
solutions. It is now seen that the chemical reactions which potential elec-
trolytes engage in are proton-transfer reactions of a special type in which
an uncharged potential-electrolyte molecule reacts with an uncharged solvent
molecule to produce a pair of oppositely charged ions.
This is only one of the types of proton-transfer reactions. There are
others. For example, an ion and a neutral molecule can react to produce
another ion and another neutral molecule

(5.4)

Thus, the destruction of the electrical field of the NH4+ ion is compensated
for by the generation of the electrical field of the HaO+ ion. This is in con-
492 CHAPTER 5

TABLE 5.8
Some Types of Proton-Transfer Reactions

Reactants Products

1. Two uncharged molecules, e.g., A pair of oppositely charged ions, e.g.,

HCI and H 20 H30+ + CI-

2. An ion and an uncharged molecule, An ion and an uncharged particle, e.g.,

e.g., NH4 + + H 20 NH3 + H30 +
3. An ion and an uncharged molecule, A pair of oppositely charged ions, e.g.,
e.g., HC0 3 - +H02 CO a-- + H3 0 +

trast to the proton-transfer reaction involving a potential electrolyte where

the net result of the reaction is the creation of two species giving rise to
ionic fields.
Or an ion can react with a neutral molecule to produce two ions, e.g.,

(5.5)

In this type of proton-transfer reaction, two ionic field-giving species are

generated, while one disappears.
It is seen that the fundamental difference between the above mentioned
types of proton-transfer reactions arises from the electrostatic work involved
in the occurrence of the reaction, i.e., the work associated with generating
or destroying the electrostatic fields of the charged species (Table 5.8).
More will be said later on this electrostatic contribution to the free-energy
change in proton-transfer reactions.
Another interesting type of proton-transfer reaction is that involving
two molecules of an amphoteric solvent, i.e., a solvent which can function
either as an acid or as a base. The reaction consists in one solvent molecule's
being a proton donor and another solvent molecule's being a proton
acceptor. For example, one may have

(5.6)
or

It is because of such autodissociation reactions that even highly purified

amphoteric solvents display some electrical conductivity.
 PROTONS IN SOLUTION 493

5.4.4. An Important Consequence of the Bronsted View: Conju-

gate Acid-Base Pairs

An important consequence of the Bronsted definition of acids and

bases comes out by examining a particular proton-transfer reaction in detail.
Consider, for example, the dissolution of acetic acid in water

(5.8)

What is the result of the proton-transfer reaction? The acetic acid molecule
(Le., the proton donor, or acid) has been deprotonated, and the water
molecule (i.e., the proton acceptor, or base) has been protonated. In fact,
one can conceptually break up the overall proton-transfer reaction into one
deprotonation reaction

CHaCOOH ---->- CHaCOO- +p (5.9)

and one protonation reaction

(5.10)

where p stands for proton. It is clear that, according to these partial reac-
tions, the acetic acid has behaved as an acid (it is a proton donor) and the
water molecule, as a base (it is a proton acceptor).
If, however, a reaction can take place in one direction, it can also run
the opposite way (cf the principle of microscopic reversibility). Thus, when
the partial reactions proceed in the reverse direction, one has

(5.11 )

(5.12)

which add up to give the overall reaction

(5.13)

This is a proton-transfer reaction in which the acetate ion behaves as a

proton acceptor (base) and the HaO+ ion as a proton donor (acid).
Thus, the reaction between the acid CHaCOOH and the base H 20
results in the production of another acid HaO+ and another base CHaCOO-.
In general therefore, an acid Al and a base B2 react to give another acid A2
and another base BI
(5.14)
494 CHAPTER 5

This reaction can be broken up into two partial reactions

(5.15)
and
(5.16)

The species Al and Bl are the protonated and deprotonated forms of the
same substance and are called a conjugate acid-base pair. It is seen that,
in a proton-transfer reaction, two acid-base pairs are involved, i.e., Al and
Bl and A2 and B2.
The homogeneous proton-transfer (or acid-base) reactions described
above are similar to homogeneous electron-transfer reactions in that the
overall electron-transfer reaction, e.g.,

Fe+++ + Ce+++ ---+ Fe++ + Ce++++ (5.17)

can be decomposed into one electronation reaction

Fe+++ + e ---+ Fe++ (5.18)

and one deelectronation reaction

Ce+++ ---+ Ce++++ +e (5.19)

Further, each of these partial reactions involves the e1ectronated (reduced)

and de-e1ectronated (oxidized) forms of a substance, or what is called a
redox couple.

5.4.5. The Absolute Strength of an Acid or a Base

The attainment of equilibrium in proton-transfer reactions is very rapid,
and, therefore, it is valid to represent such reactions as at equilibrium:

(5.20)

By considering the conjugate acid-base pairs involved in the above reaction,

one can think of the partial equilibria

(5.21)
and
(5.22)

and the corresponding equilibrium constants

(5.23)
 PROTONS IN SOLUTION 495

and
(5.24)

It is obvious that K",A 1 is an absolute measure of the tendency of Al

to donate (or lose) a proton and K",B. is an absolute measure of the tendency
of B2 to accept a proton. Since the acidity of a substance is equivalent to
its proton-donating tendency, the quantity KII,A, may be termed the absolute
strength of the acid AI' and, since the basicity of a substance is equivalent
to its proton-accepting tendency, K",II. may be termed the absolute strength
of the base B2 •
Further, by considering the product K;,A 1 Kn,B., one has

(5.25)

which shows that the product K".A, Ka,ll. = Kr is, in fact, the equilibrium
constant for the reaction Al + B2 :;:=: BI + A2·
Thus, if one could determine the absolute acid strength Ka,A 1 of Al and
the absolute base strength K",ll2 of B2 , one could calculate Kr and thus decide
how far the equilibrium Al + B2 :;:=: BI + A2 is pushed to the right. It is
clear, for instance, that the stronger the acid Al (i.e., the larger the value
of Ka,A) and the stronger the base B2 (i.e., the larger the value of Ka,ll.),
the more will the position of the equilibrium be shifted to the right.
The attempt to determine Ka,A, and Ka,ll. would succeed if the partial
equilibriums (5.21) and (5.22) involving protons as reactants or products
are realizable in practice. It has already been stated, however, that the con-
centration of free protons in solution is far too small to permit them to be
recognized as reaction products or reactants. Hence, the partial equilibriums
(5.21) and (5.22) for conjugate acid-base pairs are experimentally unrealiz-
able as separate entities and are, therefore, purely hypothetical.
One must perforce conclude that the absolute strengths of acids and
bases cannot be measured and a scale of absolute acid or base strengths
has no operational significance.

5.4.6. The Relative Strengths of Acids and Bases

Since absolute acid (or base) strengths cannot be measured and since
it is only equilibriums of the type Al + B2 :;:=: B\ + A2 that are experiment-
ally realizable, an obvious step is to accept a particular acid-base pair as a
standard and then develop a scale of relative strengths.
496 CHAPTER 5

To define the relative strength of acids, it is customary to accept the

solvent as the standard proton acceptor. Thus, for aqueous solutions, the
water molecule is taken as the standard proton acceptor, in which case the
proton-transfer equilibrium must be written as

(5.26)

and the relative strength of the acid Ai is by definition

(5.27)

In very dilute solutions, activities can be replaced by concentrations,

and one can set aH 20 = 1. Then one obtains the acid dissociation constant
K C• A defined thus
(5.28)

The acid dissociation constant of A is a quantitative measure of the power

of the acid A to donate a proton to a water molecule.
Now, it is experimentally found that the dissociation constants of
acids vary over several powers of 10 (Table 5.9). Hence, instead of remem-
bering dissociation constants in the form K C• A = a X IO- p = IO- x , where
x is the exponent, it is convenient to represent dissociation constants on a
logarithmic scale thus
(5.29)

The x's of different acids are the powers of KC• A and are, therefore, fairly
logically referred to as pKA values, i.e.,

(5.30)

In terms of pKA' Eq. (5.28) becomes

(5.31)

Since hydrogen ion concentrations, too, vary over several powers of 10,
one can refer to the powers of hydrogen-ion concentration by the symbol
pH defined thus t (Table 5.10)

(5.32)

t Strictly speaking, one must define pH in terms of the activity of H30+ ions.
 PROTONS IN SOLUTION 497

TABLE 5.9
Dissociation Constants of Some Acids

Acid Base produced

1. Very weak acids, Kc,A < 10- 7

HCO a- CO a- 4.69 X 10-11
HIO 10- 1 X 10- 11

C6 H 5 0H C 6H 50 - 1.20 X 10-10
HBrO BrO- 2 x 10-9
H 2PO. HPO.- 6.22 X 10-"

2. Weak acids, Kr,A < 10- 2

HCIO CIO- 9.56 x 10- 7
H 2 CO a HCO a- 4.5 X 10- 7
C(CHa)a·COOH C(CHa)a' COO- 8.91 X 10- 6
CHaCOOH CHaCOO- 1.75 x 10-5
C GH 5 ·COOH C6H5'COO- 6.12 X 10-5
HCOOH HCOO- 1.77 x 10-'
CH 2Cl·COOH CH 2CI·COO- 1.34 x IO- a

3. Strong acids, Kc,A""" lOa

H 2SO. HSO. lOa
(HSO.- SO- 1.01 X 10- 2 )
HCI CI lOa
HCIO a CIO a- lOa

4. Very strong acids, Kc,A""" 10"

HCIO. CIO.- 108

TABLE 5.10
Hydrogen Ion Concentrations and pH

Hydrogen ion concentration pH Hydrogen ion concentration pH

10- 1 (O.IN HCI) 10-8 8

10- 2 2 10- 9 9
IO- a 3 10-10 10
10-4 4 10-11 11
10-· 5 10-12 12
10-6 6 10-13 13
10- 7 (pure water) 7 10-14 (IN NaOH) 14
498 CHAPTER 5

TABLE 5.11
The pK Values of Some Acids

Acid pK

HC0 3- 10.33
HIO II
CaHsOH 9.92
HBrO 9.92
H 2 PO.- 8.7
HCIO 6.02
H 2 C03 6.35
C(CH a)3' COOH 5.03
CHa·COOH 4.8
C6 H sCOOH 4.2
HCOOH 3.75
CH 2 CI·COOH 2.87
H 2SO. -3
HSO.- 2.0
HCI -3
HCIO a -3
HCIO. -8

and write Eq. (5.31) in the form

(5.33)

Since low pKA values indicate large values of the acid dissociation
constant Kc,A = CnCH3o+/CA, Eq. (5.33) shows that the lower the value of
pKA' the more is the conversion of the acid A to its base form B with
the concomitant production of H30+ ions. Strong acids therefore have small
pKA values (i.e., large KA values, Table 5.11).
The consideration of the strength of bases proceeds on lines similar
to that for acids. That is, the relative base strength is defined by taking
the solvent as the standard proton donor. In aqueous solution, this means
that the water molecule is the proton donor and the proton-transfer reac-
tion is
(5.34 )
 PROTONS IN SOLUTION 499

TABLE 5.12
Autodissociation Constants of Some Solvents

Reaction Temperature K

2NHa ~ NH. + + NH 2- -33°C 10- 2 •

2CH30H~ CH 30H.+ + CHaO- 25°C 10-17
2H 20 ~ HaO+ + OH- 25°C 10-1•
2CHaCOOH~ CHaCOOH.+ + CH3COO- 25°C 10-13
2H.SO. ~ HaO+ + HS.0 7- 10 °C 7 X 10-5
2HNO a ~ NO. + + NO a- + H.O 25°C 2 x 10- 2

The relative strength Kr,B of the base B is then defined as

K _ aOH-aA
r,B - a a (5.35)
B H 20

or, by replacing activities by concentrations and setting au 2 0 = I, one has

the analogue for bases of the acid dissociation constant, i.e.,

(5.36)

Now, the equilibrium constant for the autodissociation of water (Table

5.12), i.e.,

is termed the ionic product of water Kw and is given by

(5.37)

On combining Eqs. (5.36) and (5.37), the result is

(5.38)

or
pKB = pKw - pKA (5.39)
500 CHAPTER 5

TABLE 5.13
pKB Values of Some Bases and pKA Values of the Conjugate Acids

Base pKB Conjugate acid pKA

Ammonia 4.76 NH.+ 9.24

Methylamine 3.30 CH3NH3+ 10.70
Dimethylamine 3.13 (CH 3)zNH z+ 10.87
Trimethylamine 4.13 (CH3)3NH+ 9.87
Ethylamine 3.25 C zHsNH3+ 10.75
Diethylamine 2.90 (C 2 H s)zNH z+ 11.10
Triethylamine 3.20 (C ZH S)3NH+ 10.80
Aniline 9.39 C 6H SNH3+ 4.61
Benzylamine 4.63 (C6HsNHz)3NH+ 9.37
Diphenylamine 13.16 (C 6H s)zNH z+ 0.84
Pyridine 8.80 CsHsNH+ 5.2

where
(5.40)
and
pKw = - loglo Kw (5.41 )

Equation (5.39) shows that there is a simple relation between the pKn
value of the base B (Table 5.13) and the pKA value of the acid A in the
proton-transfer reaction
(5.34)

For example, consider the reaction

(5.42)

where NH3 and NH4+ are the conjugate base-acid pair. For this equilibrium
Eq. (5.39) becomes
(5.43)

which indicates that one can determine the pK of the acid NH4 + if one
knows the pK of NH 3 , or vice versa.

5.4.7. Proton Free-Energy Levels

Since proton-transfer reactions are the basis of the behavior of acids
and bases, the relative strengths of acids and bases must arise from the
 PROTONS IN SOLUTION 501

potential
energy

Distance
Fig. 5.22. Schematic potential-energy bar-
riers for proton transfers from an HaO+ ion
to a water molecule and from an acid A to
a water molecule.

differing energies of the reactants and products. This aspect will now be
discussed.
Consider the transfer of a proton from an H30+ ion to a water molecule.
This transfer process can be described by a potential-energy barrier syn-
thesized from two Morse curves (Fig. 5.14). One curve represents the energy
change as the proton leaves the proton donor H30+; and the other, the
energy change as the proton approaches the proton acceptor H 20.
Such curves can be drawn for the transfer of a proton from any proton
donor to a water molecule. t Suppose one compares the potential barrier
for two proton-transfer reactions, e.g., that from an H30+ ion to a water
molecule and that from some acid A to a water molecule (Fig. 5.22)

A proton, H 20

The Morse curve for the energy changes as a proton approaches the water
molecule (the proton acceptor) will be the same in both proton-transfer
reactions. But the curve for the removal of the proton from the H30+ ion
will not be the same as that for its removal from the acid A because it is

t Since one is considering the overall energy change between the initial and final states,
the type of mechanics to be used in considering the transfer-classical or quantal-
does not affect the issue. It is when proton-transfer rates are considered that the
mechanism of transfer becomes important.
502 CHAPTER 5

unlikely that the proton affinities of H 20 and B (the conjugate base of A)

will be equal.
This argument has an important consequence: The minima of the
deprotonation curves for HaO+ and A are not at the same level in relation
to the minimum of the curve for the protonation of a water molecule, but
the difference between the minima of the deprotonation (of HaO+ or A)
and the protonation (of H 20) curves defines the difference in energy be-
tween the initial state and the final state of the proton-transfer reaction.
If one also takes into account the entropy difference (usually small), one
will obtain the chemical or nonelectrostatic contribution L1GCh to the
standard free-energy change LI GO of the proton-transfer reaction. Since the
reactant and product species may be charged, one must also consider the
electrostatic free energy LI G~I of charging or discharging the ions. (The
method of calculating the electrostatic contribution to the total free-energy
change is shown in Section 5.4.9.)
The total free-energy change L1GO, which is therefore made up of two
contributions
(5.44)

represents the net work done to pull out a proton from the proton donor,
transfer it to a water molecule, and then separate the resulting species. One
could use the terminology suggested by Gurney and call the L1GO's "proton
free-energy levels" in the proton donors HaO+, AI' A2 , etc., defined in
relation to the water molecule as the standard proton acceptor. These
proton free-energy levels can also be represented on a diagram which
would reveal, on examination, whether the jump of a proton to a water
molecule would lead to a decrease of free energy and would thus tend to
occur spontaneously (Fig. 5.23).
At present, the statistical mechanical calculation of the L1Go's for proton-
transfer reactions involves so many approximations that it is preferable to
obtain them in the following way.
The standard free-energy change for the transfer of a proton from an
acid to a water molecule is simply related to the equilibrium constant for
the reaction
(5.45)

through the Van't Hoff reaction isotherm

(5.46)
 PROTONS IN SOLUTION 503

20--
1-0

0.9

SO.--
0.6 -F'O.-
Ct-t.CICOO-

HCOO -
0]
CoH.NH,
C CHfOO-

.'"
>
0.6
HCO.-
.!! '"
~
c HPO.-- .!!
0 0.5

..e
c
0
l5. .l:

il NH.
"
0 .4
'0. 80, -
3
l:l" NH 2CH2COO - .g
0 CO.--
0 .3

02
""09
0 .1

-1.0
OD OW
>11
C:
~=
w>
0>0

Fig. 5.23. Proton free-energy levels.

and, if conditions of infinite dilution are considered, then activities are

equal to concentrations and aH 2 0 '"" 1, i.e.,

(5.47)

Hence,
LlGO = - RT In K C• A
= (- loglO K c••''') 2.303RT (5.48)

and, from the definition of pKA [cf Eq. (5.30)],

LlGo = (2.303RT) pK (5.49)

504 CHAPTER 5

TABLE 5.14
Proton Free-Energy Levels of Some Acids

Acid Lleo, kcal mole- 1

Trimethylacetic 6.856
Acetic 6.486
Propionic 6.646
Hexoic 6.626
Isobutyric 6.606
Isohexoic 6.606
Valerie 6.606
Butyric 6.566
Isoullesic 6.516
Diethylacetic 6.456
Succinic 5.866
Lactic 5.256
Glycollic 5.216
Formic 5.106
Iodoacetic 4.326
Bromoacetic 3.956
Chloroacetic 3.906
Fluoroacetic 3.526
Cyanoacetic 3.366

i.e., the standard free energy for the transfer of a proton from the acid to
a water molecule is proportional to the pK of the acid.
Thus, the concept of proton free-energy levels, expressed in terms of
LlGo (Table 5.14), is simply related to the pK value and, therefore, to the
strength of an acid relative to a particular solvent functioning as a proton
acceptor.

5.4.8. The Primary Effect of the Solvent upon the Relative

Strength of an Acid

An obvious conclusion arises from a consideration of the potential-

energy curves for proton transfer from an acid to the solvent: the lower
the minimum of the solvent curve in relation to that for the acid, the
greater is the free-energy change for the proton-transfer reaction and, hence,
 PROTONS IN SOLUTION 505

the greater is the relative strength of the acid. This conclusion can also be
reached from Eq. (5.27), which can be written as

(5.50)
for the reaction
A + S~B + SH+ (5.51)

[t is clear that, the greater the absolute base strength Ka,s of the solvent
(i.e., the greater its proton-accepting tendency), the greater is the relative
acid strength Kr,A of the acid A. Hence, a particular acid may appear strong
in one solvent and weak in another.
For example, Hel is a strong acid in aqueous solutions because the
proton-transfer reaction
(5.51)

proceeds almost completely to the right (H 20 (lcting as a strong base);

but it is a weak acid in glacial acetic acid because the interaction

(5.52)

occurs only to a small extent. The reason, then, is that water is a stronger
base than acetic acid.
Suppose now one considers several acids AI' A2 , A3 , .•. , and suppose

Fig. 5.24. The potential-energy barriers for

proton transfer from strong acids to a weakly
acidic solvent.
506 CHAPTER 5

the minima of their proton-donating curves lie far above that of the proton-
accepting curve of the solvent (Fig. 5.24). Then these various acids will
react almost completely with the solvent, i.e., the position of the equilibrium

(5.53)

will occur so much to the right that, effectively, all the acid will be in its
base form B;. Under these circumstances, all the acids AI, A2 , ••• will
appear equally strong (Fig. 5.25). This is what happens with aqueous
solutions of the strong acids HCl0 4 , HBr, H 2S04 , HCl, and NH0 3. The
reactions such as
(5.54)

proceed almost completely to the right, and, in solutions less concentrated

than 2M, all these acids appear equally strong, i.e., the differences in their
relative acid strengths cannot be distinguished. This is because the behavior
of the aqueous solutions is determined by the H30+ ion, which is formed
virtually to the same extent in all these solutions. One says that, because

Normality of acid

Fig. 5.25. A strongly basic solvent levels

out the differences in the relative strengths
of strong acids. The relative strengths of the
acids are indicated in the diagram by their
activity in catalyzing a chemical reaction.
Below 2M concentration, the activities of
the various acids are almost the same.
 PROTONS IN SOLUTION 507

2·0

1·5

A
1·0

0·5

0{)2 0·04 0·06 0{)8

.rc
Fig. 5.26. A weakly basic (i.e., strongly
acidic) solvent brings out the differences
in the relative strengths of strong acids. In
the diagram, the relative strengths of the
various acids are indicated by their equiva-
lent conductivities in glacial acetic acid.

the aqueous solvent has a low absolute acid strength (a strong base strength),
it has leveled out-the so-called leveling effect-the differences in the relative
strengths of the various acids.
To bring out these differences, it is necessary to use a solvent which
has a greater absolute acid strength (or a lesser base strength). This is
equivalent to choosing a solvent which has a protonation curve, the mini-
mum of which lies close enough to the minima of the deprotonation curves
to make the difference in their minima significant. Thus, HCI0 4 , HBr,
H 2 S04 , HCI, and HN0 3 display quite different relative strengths (i.e., ionic
concentrations, as indicated by the equivalent conductivities) when dissolved
in glacial acetic acid (Fig. 5.26), a more acid, less basic solvent than water.

5.4.9. A Secondary (Electrostatic) Effect of the Solvent on the

Relative Strength of Acids

It has been demonstrated that the solvent plays an essential role in

determining the relative strength of an acid. The key characteristic of the
solvent is its absolute base strength K a •s , which determines the relative
508 CHAPTER 5

strength K r •A of an acid through the Eq. (5.50)

(5.50)

But, in addition to functioning as a proton acceptor in a proton-

transfer reaction, the solvent also serves as a dielectric medium, which
affects the electrostatic aspects of the reaction in the following way.
Suppose, for instance, the proton-transfer reaction is of the potential-
electrolyte type
HA + S:;;:= SH+ + A- (5.55)

in which an uncharged acid molecule HA interacts with a neutral solvent

molecule S to form a pair of oppositely charged ions. The dielectric constant
of the medium, which is effectively equal to the dielectric constant of the
solvent, determines the electrostatic attraction between the pair of ions;
the larger the dielectric constant, the less strongly do the ions SH+ and A-
attract each other and the less is the tendency of the backward reaction.
If, therefore, one considers two solvents which do not differ considerably
in their absolute acid strengths, one would expect the relative strength of
the acid to be greater in the solvent with a greater dielectric constant. This
is indeed the case with the uncharged carboxylic acids when one uses
alcohol instead of water as the solvent.
When, however, an ion and an uncharged molecule react to form an-
other ion and uncharged molecule, e.g.,

(5.56)

the proton-transfer reaction does not require the separation of oppositely

charged ions, and, therefore, the dielectric constant of the medium does not
have a significant effect on the relative strength of the acid.
To make these considerations quantitative, it is necessary to isolate
the dielectric-constant effect of a solvent from its proton-accepting prop-
erties. This can be done as follows.
If two acids Al and A2 are considered in the same solvent, then the
ratio of their relative strength is given by [cf (5.50)]

Kr.A, _ Ka.A,Ka,s
K r,A2 - Ka,A2 Ka,s

(5.57)
 PROTONS IN SOLUTION 509

Thus, if the absolute base strength Ka,s of a solvent is the only characteristic
of the solvent which affects the relative strengths of acids, then the ratio of
the relative strengths of a pair of acids should be independent of the solvent.
If, on the other hand, it is found that the ratio Kr,AJKr,A. varies with the
solvent, then the variation must be ascribed to the influence of the dielectric
constant of the medium.
It is easy to derive a relationship between the relative acid-strength
ratio Kr,}.)Kr,A. and the dielectric constant of the medium. The relative
acid-strength ratio can be elaborated upon thus

(S.S8)

from which it is clear that Kr,AJ Kr,Aa is the equilibrium constant for the
reaction
(S.S9)

This reaction may be written in the form

(S.60)

in order that the electrostatic effects may be seen more easily. The equi-
librium constant ratio KT,AJ KT,A. is given by

-RTln Kr,Al =,dGo (S.61)

Kr,A.

where ,dGo is the standard free energy change for reaction (S.60). This ,dGo
includes the electrostatic contribution ,dGgh arising from the disappearance
and appearance of charges in the reaction, as well as a nonelectrostatic, or
chemical, contribution ,dGgh , arising from the relative levels of the minima
in the proton-transfer barrier (cf Section S.4.7). Hence,

- RT In K
Kr,Al = ,dGoch + ,dGoel (S.62)
r,A.

To evaluate the electrostatic contribution to the free-energy change in

the proton-transfer reaction, one can assume that the ions A1 - and A 2-
are spherical, and set ~Ggl equal to the Born charging work for the singly
510 CHAPTER 5

charged A 1- ion minus the Born discharging work for the singly charged
A2 - ion, i.e.,

= NAeo2
2e
(_1___1_)
r A 1- r A 2-
(5.63)

where rA 1 - and rA 2 - are the radii of the A1- and A2 - ions and the other
terms have their usual connotation (ef Section 2.2.5).
Introducing this expression into Eq. (5.62), one has

(5.64)

Thus, the plot of the logarithm of the ratio of the relative strengths of
two acids versus the reciprocal of the dielectric constant should be a straight
line of slope

the sign of the slope being positive when r A2 - < rA 1- and negative when
r A 1- < rA 2 - ' Further, by extrapolating the In (Kr,A/ Kr,A.) versus 1/ e straight
line to infinite dielectric constant [i.e., (I/e) ~ 0], one should obtain the

TABLE 5.15
Ratios of the Relative Strengths of Substituted Benzoic Acids to the Relative
Strengths of Benzoic Acid in Various Solvents at 25 0 C

Value of log Kr relative to log Kr for benzoic acid

Derivative of
benzoic acid H 2 O, CH 3 OH, C 2 H 5 OH, n-C.HOH,
c = 78.5 c= 31.5 c = 24.2 c = 17.4

p-Hydroxy -0.36 -0.53 -0.55 -0.57

p-Methyl -0.17 -0.18 -0.18 -0.19
p-Chloro 0.22 0.34 0.42 0.39
m-Chloro 0.38 0.59 0.63 0.59
a-Chloro 1.28 1.21 1.12 1.08
a-Nitro 2.03 1.83 1.77 1.78
 PROTONS IN SOLUTION 511

. acid
. benzoIc
x !j _oinl\rO
0 2 1,"3,
0

'" u
.,
'¥.""
til
~
x
u
.:
.
)£
0 acetic acid
til
.!:!
H2 O CH30H
t
CzHs°H

0 0·01 0·02 0'03 0·04

liE

Fig. 5.27. The dielectric constant; variation of the

ratio of log K, for an acid to log K, for benzoic acid.

intrinsic acid-strength ratio (Kr,AJ K r,A 2 )r-+ oo which IS free of electrostatic

effects. Hence,

(5.65)

These predictions are borne out in practice, as may be seen from Fig.
5.27 and Table 5.15. It is seen, therefore, that, while the base strength of a
solvent is the primary factor in determining the relative strengths of a given
acid in various solvents, the solvents' dielectric constants do have a signifi-
cant secondary effect due to the changing solvation energies of the ions
in the various solvents.

Further Reading

1. W. F. K. Wynne-Jones, Proc. Roy. Soc. (London), A140: 440 (1933).

2. R. W. Gurney, Ionic Processes in Solution, McGraw-Hill Book Company,
New York, 1953.
3. R. P. Bell, The Proton in Chemistry, Cornell University Press, Ithaca, N.Y.,
1959.
512 CHAPTER 5

4. C. B. Monk, Electrolytic Dissociation, Academic Press, Inc., New York, 1961.

5. E. Grunwald and S. Meiboom, J. Am. Chern. Soc., 85: 204 (1963).
6. L. L. Schaieger, P. Saiomaa, and F. A. Long, B. E. Conway and R. G.
Barradas, eds., in: Chemical Physics of Ionic Solutions, John Wiley & Sons,
Inc., New York, 1966.
CHAPTER 6

IONIC LIQUIDS

6.1. INTRODUCTION

6.1.1. The Limiting Case of Zero Solvent: Pure Liquid Electro-

lytes

Modern electrochemistry is concerned not only with systems based on

aqueous solutions but also with water-free systems (see Section 4.7). Indeed,
it is in such systems that many important electrochemical processes are
carried out, e.g., the production of aluminum, sodium, and magnesium.
The rationale behind the use of (and the search for) other media shall
be restated (see also Section 4.7.2). In aqueous media, electrode reactions
involving water and its ionic constituents (hydrogen ions and hydroxyl
ions) sometimes altogether supplant the desired electrochemical process
(e.g., the electrowinning of magnesium). Further, in technologies based on
the conversion of electrical energy into chemical change and vice versa (e.g.,
the dissolution of a metal in an electrochemical energy-conversion device),
the desired rate of conversion may require media more conducting than
aqueous solutions.
Some of the difficulties associated with carrying out processes in
aqueous solutions can be sometimes overcome by using nonaqueous solvents
consisting usually of organic substances, e.g., acetonitrile, to which is added
some solute which dissociates in that solvent. But this often is not a good
approach because of the low specific conductances of such solutions (see
Section 4.7.9).

513
 514 CHAPTER 6

___ Ionic crystol

Solvent

/
~
'I:~
Solvated ions
I '-
.r
+ '"
I
,

Ionic liquid
Ionic solution

Fig. 6.1. An ionic crystal can be dismantled either by

the action of a solvent or by the action of heat.

So the question arises: Why have a solvent at all? This limiting case
of an aqueous or a nonaqueous ionic solution from which all the solvent
is removed is a pure liquid electrolyte. Conceptually, this definition is
accurate. Operationally, however, if one removes solvent molecules from
a solution, for example, by evaporation, one is left with ionic crystals, pure
solid electrolyte. A further step in conversion from the solid to the pure
liquid form is necessary.

6.1.2. The Thermal Dismantling of an Ionic Lattice

The process of dissolution of a true electrolyte was described in Chap-

ter 2. The basic picture is that the ions in an erstwhile-rigid ionic lattice
succumb to the strong attraction t of the solvent molecules and follow them
into solution, executing a random walk there as free, stable solvated ions.
The result is an ionic solution which has the ability to conduct electricity
by the drift of ions. The disassembly of the ionic lattice was achieved by
the solvent overcoming the coulombic cohesive forces holding together the
ions in the regular arrangement called a lattice (Fig. 6.1).

t In the case of aqueous solutions, the forces are essentially ion-dipole and ion-quad-
rupole in character.
 IONIC LIQUIDS 515

TABLE 6.1
Specific Conductivities of Solid and Molten NaCI

Specific conductivity,
ohm- 1 cm- 1

Solid NaCI X 10-3 at 800°C

Molten NaClt 3.9 at 900 °C

t Melting point of NaCl is 801 "C.

A solvent, however, is not the only agency which can dismantle an

ionic lattice. Heat energy, too, can overcome the cohesive forces and disrupt
the ordered arrangement of ions in a crystal (Fig. 6.1). This process of
melting results in the pure liquid electrolyte, a system having a conductance
several orders of magnitude larger than that of the corresponding solid
(Table 6.1).

6.1.3. Some Features of Ionic Liquids (Pure Liquid Electro-

lytes)

A common type of ionic lattice is that of a crystalline salt. One such

ionic lattice encountered in everyday life is sodium chloride. Molten sodium
chloride is a typical liquid electrolyte. It displays the characteristics of many
liquid electrolytes. t
An appreciation of the properties of liquid electrolytes can be gained
by a comparison between molten ice (water) and molten sodium chloride
(Table 6.2). Both liquids are clear and colorless. Their viscosities, thermal
conductivities, and surface tensions are not very different.
In fact, one can go further and make the following statement: Most
molten salts look like water and, near their melting points, have viscosities,
thermal conductivities, and surface tensions of the same orders of magnitude
as those of water. In general, however, fused salts are stable as liquids only
at relatively high temperatures (300 to 1250 dc) (Table 6.3).
One can quote exceptions to these generalizations. The tetraalkyl-
ammonium salts are liquid at much lower temperatures (Table 6.4).

t The terms pure liquid electrolyte, ionic liquid, fused salt, and molten salt are used syn-
onymously.
516 CHAPTER 6

TABLE 6.2
Comparison of Some Properties of Water and Molten NaCI

Water Molten NaCl

25 DC 850 DC

Viscosity, miIIipoise 8.95 12.5

Refractive index 1.332 1.408
Diffusion coefficient, cm2 sec-1 3 x 10-5 Na+ 1.53 x 10-4
CI- 0.83 X 10-4
Surface tension dynes cm-1 72 111.8
Density 1.00 1.539

TABLE 6.3
Melting Points of Some Inorganic Salts

Salt Melting point, DC Salt Melting point, DC

AgNO. 210 PbCl s 501

HgBrs 238 CdCl s 568
LiNO. 254 LiCl 610
ZnCl s 275 CaCl 646
HgCl s 277 NaI 651
NaNO. 310 MgCl s 714
KNO. 337 KCl 776
PbBrs 373 NaCl 808
AgBr 434 NasCO. 858

TABLE 6.4
Melting Points of Some Tetraalkylammonium Salts

Salt Melting point, DC

Tetramethylammonium bromide 230

Tetrabutylammonium iodide 144
Tetrapropylammonium iodide 280
 IONIC LIQUIDS 517

TABLE 6.5
Conductivities of Molten Salts and Water

Temperature, Specific conductivity,

Substance
°C ohm-1 cm- 1

H 20 18 4 x 10-8
LiCI melt 710 6.221
NaCI melt 908 3.903
KCI melt 872 2.407

6.1.4. Liquid Electrolytes Are Ionic Liquids

The crucial difference between the molten salts and molten ice lies in
the values of the specific conductivity (Table 6.5). Fused salts have about
108 times better specific conductivity than water.
The temptation to ascribe the high conductance of fused salts to
conduction by electrons must be rejected. Thus the conductivity of a
molten salt is high compared to that of water; but it is some ten thousand
times lower than that of a liquid metal, such as mercury (Table 6.6).
Fused salts conduct by the drift of ions. They are, in fact, ionic liquids.
Pure liquid electrolytes therefore are liquids containing only ions, the ions
being free or associated (see Section 6.5).
Another class of ionic liquids is the molten oxides. These are highly
conducting liquids formed by the addition of a metal oxide (e.g., Li 20)
to the oxide of a nonmetal (e.g., Si0 2 ). Some properties of the molten oxides
are shown in Table 6.7.
To develop a perspective on the properties of liquid electrolytes, a
tabulation has been made of some properties of water, liquid sodium,

TABLE 6.6
Conductivities of Molten NaCI and of Mercury

Temperature, Specific conductivity,

Substance
°C ohm- 1 cm- 1

Hg 20 1.1 X 104
NaCl melt 908 3.903
518 CHAPTER 6

TABLE 6.7
Some Properties of Molten Oxides near the Melting Point

Molten Temp., Density, Surface tension, Viscosity, Specific conductivity

oxide DC gcm- 3 dyne cm- 1 centipoise ohm- 1 cm~l

Li.O·SiO. 1250 2.07 354 2.88 x 10' 5.5 (1750 DC)

Li.O· 1!SiO. lIOO 2.13 331 5.02 x 103
Li,O·2Si0 2 lIOO 2.16 319 1.78 x 10' 2.5 (I750 DC)
Na 2O·Si0 2 1100 2.23 300 l.I9 x 103 4.8 (I 750 DC)
Na.O·2SiO. 900 2.28 289 3.33 x lOs 2.1 (1750 DC)
Na,O·3Si0 2 900 2.23 282 1.99 x 106
K.O·2Si0 2 lIOO 2.20 220 1.08 x lOs 1.5 (I750 DC)
CaO·SiO. 1550 400 2.73 x 10' 0.8 (1750 DC)

an aqueous solution of NaCl, crystalline NaCI, fused NaCI, and a mixture

of fused Na 2 0 and Si0 2 (Table 6.8).

6.1.5. The Fundamental Problems in Pure Liquid Electrolytes

In dealing with aqueous and nonaqueous solutions of electrolytes, the

procedure was, firstly, to seek a picture of the time-average structure of
the electrolytic solution and, secondly, to understand the basic laws of ionic
movements. The picture that emerged was of ions and solvent molecules
interacting together to form solvated ions; of ions interacting with each
other to form ionic clouds and associated ion pairs, or complexes; and of
all these entities executing an aimless random walk at equilibrium, which
becomes a directed drift under an external field.
The problems in pure liquid electrolytes are analogous, though perhaps
more difficult to treat mathematically.
The first problem can be defined as follows: What idealized model
could best approximate a solvent-free system of charged particles forming
a highly conducting ionic liquid?
In the case of the aqueous solution, it was easy to understand the drift
of ions at the behest of the applied electric field. Positive and negative
ions, separated by stretches of water, drift in opposite directions (Fig. 6.2).
In a pure ionic liquid, however, there is no water for the ions to float in;
the ions drift among themselves (Fig. 6.3). When an external field is switched
on, how is it that the ions are able to move past each other? Will not the
very large interionic forces make them stick together, somewhat in the
 TABLE 6.S
Comparison of the Properties of Various Liquids

Liquid sodium 1M sodium chloride Liquid sodium

Properties Water, 25°C Na,O·SiO,
at mp soln at 25°C chloride at mp

Melting point, °C 0.0 97.83 -3.37 801 1088

Vapor pressure, mm Hg 23.756 9.842 x 10- 8 16.8, 20°C 3045 X 10-1

30.8, 30°C

Molar volume, cm" 18.07 24.76 17.80 30047 55.36

Density, g cm-" 0.997 0.927 1.0369 1.5555 2.250, 1200 °C

Compressibility,
108 cm' atm- 1 46.3055 Isoth. 18.88 Isoth. 40.08 !sotho 29.08 Isoth. 59.579 adiabatic, 1200°C

Diffusion coefficient, 3.0 x 10-5 2.344 X 10-1 DNa+ = 1.25 X 10-5 DNa+ = 1.53 x 10-'
cm'sec- 1 1.584 x 10-1 DCI- = 1.77 X 10-5 DCI = 0.83 X 10-'
850°C

Surface tension dyne cm-1 71.97 192.2 74.3 113.3 294, 1200 °C

Viscosity, centipoise
oz
0.895 0.690 1.0582 1.67 980, 1100 °C
(')
,...
Specific electric conduc- is
tance, mho cm- J 4.0 x 10-8 1.04 x lOS 0.8576 mole cm-3 3.58 4.8, 1750°C c
c
en
Refractive index 1.333 0.04 1.3426 10408, 850°C 1.52, solid, room temp.
~
...
520 CHAPTER 6

Sol vent seporotes

drifting ions

Fig. 6.2. In an aqueous solution, the sol-

vent separates the drifti ng ions.

manner of the poorly conducting ionic lattices (of the solid state)? The
situation appears puzzling.
In fact, what is the essential difference between the solid form and
the liquid form of an ensemble of particles? This is a question which is
relevant to all processes of fusion, e.g., the process of solid argon's t melting
to form a liquid. In the case of ionic liquids, the problem is more acute.
One must explain the great fluidity and corresponding high conductivity
in a liquid containing charged particles in contact.
The second problem concerns an understanding of the sharing of
transport duties (e.g., the carrying of current) in pure liquid electrolytes.
In aqueous solutions (cf Section 4.5.2), it was possible to comprehend the
relative movements of ions in the sense that one ionic species could move
with greater agility and therefore transport more electricity than another
species until a concentration gradient is set up and the resulting diffusion
evens out the movements. In fused salts, this comprehension is less easy
to acquire. At first, it is even difficult to see how one can retain the concept
of transport numbers when there is no reference medium (such as the water
in aqueous solutions) in which ions can drift. The point is: Any movement
of one ion relative to the other affects the situation of the whole liquid.
Thus, a theory of transport numbers in liquid electrolytes must differ from
that for aqueous and nonaqueous solutions.
Thirdly, there exists the problem of complex ions. In aqueous and
nonaqueous solutions, it is possible to regard the entity: ion-ion atmosphere
as a type of incipient complex in which the mean distance between oppositely

t This is a relatively simple solid from the point of view of the forces between the un-
charged particles.
 IONIC LIQUIDS 521

Fig. 6.3. In an ionic liquid, there is no

solvent separati ng the drifti ng ions.

charged ions becomes smaller with increasing electrolyte concentration.

Eventually, the ions come sufficiently close to withstand the thermal forces
which tend to separate them. They remain coulombically stuck together
as ion pairs for appreciable times and sometimes bond together chemically
as complex ions (see Section 3.8).
For the ionic liquids, however, the situation is different for all the ions
are always in contact. By definition, i.e., zero solvent, nothing separates
the ions. This absence of solvent causes conceptual problems regarding the
existence of complex ions in ionic liquids.
Consider a particular ion associated with another to form a vibrating
complex. The ion is also in contact with, and jostled continually by, neigh-
boring ions which are exactly like its partner in the complex (Fig. 6.4).
Which is the partner and which the neighbor; which is the vibration and
which the collision? A distinction between these two types of contacts
constitutes one of the problems in this field.
In aqueous solutions, the situation is clarified by the solvent. This
solvent keeps the complex ions apart at mean distances, defines them as

Reference ion

One of these ""'--- -

vibrates with
respect to the One of these is a
reference ion; portner in a complex;
the other collides the other is 0 neillhbour
with it

Fig. 6.4. The problem of distinguishin a neigh-

boring ion colliding with the reference ion from
a ligand (Le., a partner in complex formation)
vibrating in relation to the reference ion.
522 CHAPTER 6

Complex ions separated

from other ions
by solvent

Fig. 6.5. In an aqueous solution, the

complex ion is spatially separated from the
other ions.

independent stable entities, and permits probing radiation (e.g., visible

light) to pick them out from the surroundings (Fig. 6.5).
The concept of complex ions is therefore more subtle in ionic liquids
than in aqueous solutions. It has even been asked: Is the concept valid
at all?

Further Reading
1. H. Bloom, in: J. O'M. Bockris, ed., Modern Aspects of Electrochemistry,
Chap. 3, Vol. 2, Butterworths' Scientific Publications, Ltd., London, 1959.
2. Iu. K. Delimarskii and B. F. Markoy, Electrochemistry of Fused Salts, The
Sigma Press, Publishers, Inc., Washington, D.C., 1961.
3. H. Bloom, Pure Appl. Chem., 7: 389 (1963).
4. B. R. Sundheim, ed. Fused Salts, McGraw-Hill Book Company, New York,
1964.
5. M. Blander, ed. Molten Salt Chemistry, Interscience Publishers, Inc., New
York, 1964.
6. E. A. Ukshe, Russ. Chem. Rev., 34: 141 (1965).
7. H. Bloom, The Chemistry of Molten Salts, W. A. Benjamin, Inc., New York,
1967.

6.2. MODELS OF SIMPLE IONIC LIQUIDS

6.2.1. The Origin of Liquid Electrolyte Models

The structure of a system can often be understood from the way that
system arose. What are the origins of a pure liquid electrolyte? It can be
conceived as being formed either by the fusion of a solid ionic lattice or by
condensation from a vapor of ions. Two types of models for pure liquid
electrolytes can therefore be developed: gas-oriented models or lattice-
oriented models (Fig. 6.6).
 IONIC LIQUIDS 523

gas lattice

0--

' ' '"'\ i'"liquid

Imti"

Fig. 6.6. The two types of models for pure

liquid electrolytes: gas-oriented and lattice-
oriented models.

6.2.2. lattice-Oriented Models

6.2.2a. The Experimental Basis for Model Building. One's first im-
pression of a liquid (its fluidity, conformity to the shape of the containing
vessel, etc.) would suggest that its structure has nothing to do with that
of the crystal from which it was obtained by melting.
If, however, a beam of monochromatic X rays is made incident on
the liquid electrolyte, the scattered beam has an interesting story to tell.
The ions are almost at the same internuclear distances in a fused salt as
in the ionic crystal (actually, at a slightly lesser distance) (Table 6.9).
This is the memory that a fused salt retains of the ionic lattice which gave
birth to it in the melting process. The X-ray patterns also indicate that, in
the liquid state, the local order extends over a very short distance (a few
angstroms). It is as if the fused salt forgets how to continue the ordered
arrangement of ions of the parent lattice (Fig. 6.7).
6.2.2b. The Need to Pour Empty Space into a Fused Salt. There is
another important fact about the melting process. When many ionic lattices
524 CHAPTER 6

TABLE 6.9
Internuclear Distances in an Ionic Crystal and the Corresponding Fused Salt

Distance between oppositely charged ions, A

Salt
Crystal, mp Molten salt

LiCI 2.66 2.47

LiHr 2.85 2.68
Lil 3.12 2.85
NaI 3.35 3.15
KCI 3.26 3.10
CsCI 3.57 3.53
CsHr 3.72 3.55
CsI 3.94 3.85

are melted, there is a 10 to 25% increase in the volume of the system

(Table 6.10).
This volume increase is of fundamental importance to one who wishes
to conceptualize models for ionic liquids because one is faced with an

Probability Ionic crystal

of finding
neC)Cltive ion
1.0 Fused salt

0.5

a 20 30
Distance from reference positive Ion

Fig. 6.7. Schematic diagram to show short-range and

long-range order in an ionic crystal as opposed to only
short-range order in a fused salt. In an ideal ionic crystal,
if one takes a reference positive ion, there is a certainty
of finding a negative ion at the lattice distance or a mul-
tiple of this distance; in a fused salt, there is a high prob-
ability of finding a negative ion one distance away; but
within two or three lattice distances away, the probability
becomes half. i.e., a negative ion is only half as likely as
a positive ion. Thus, in a fused salt. there is no long-
range order.
 IONIC LIQUIDS 525

TABLE 6.10
Volume Change on Fusion

Substance % Increase of volume on fusion

NaCl 25
NaF 24
NaI 19
KCl 17
KBr 17
KI 16
RbCl 14
CdCl 2 20
CdBr2 28
NaN0 3 11

apparent contradiction. From the increase in volume, one would think that
the mean distance apart of the ions in a liquid electrolyte should be greater
than in its parent crystal. On the other hand, from the fact that the ions in
a fused salt are slightly closer together than in the solid lattice, one would
think that there should be a small volume decrease upon fusion.
The clue to the resolution of this apparent contradiction lies in the
mobility of ions in simple molten salts, which is several orders of magnitude
('" 1()3 times) greater than that in the crystalline state (Table 6.1). It has
been seen (see Section 4.2.16) that ionic movements can be thought of as
occurring in elementary steps, each of which requires the jump of an ion
into an ionless region or site. Hence, the high mobility of ions in liquid
electrolytes implies that the number of ionless regions (or vacant sites) is
far greater in an ionic liquid than in an ionic crystal.
It can be concluded, therefore, that a liquid electrolyte can occupy
more volume than the corresponding ionic crystal and, at the same time,
preserve approximately the same short-range order, but only if empty
space is introduced into it. t

t In the case of some salts, the volume changes on fusion are much smaller than are
indicated in Table 6.10. Thus, calcium, strontium, and barium halides have volume
changes which are about a fifth of the volume changes for the corresponding alkali
halides. This is because such salts crystallize in a form which already contains plenty
of open space in the solid lattice. When these open-lattice salts are fused, or melted,
there is need for a smaller volume increase than is the case for the space-filled lattices
of the alkali halides.
 526 CHAPTER 6

New surface

Old surface

Fig. 6.S. In the vacancy model. empty

space is created in the system when ions
move from lattice sites to surface sites, i.e.,
when Schottky defects are produced.

How is this emptiness (injected into the liquid electrolyte) to be con-

ceptualized? The different models of fused salts involve different ways of
considering the empty space. Thus, the emptiness is described as vacancies,
holes, free volume, etc. The differences are not purely semantic. Some of
these models will now be described.
A model can only reproduce what, in the opinion of the designer of
the model, are considered to be the essential features of the system. There
should be no surprise, therefore, if a given model fails to reproduce all
aspects of the behavior of fused salts. But one has to determine whether
one of the various models is markedly more successful in its quantitative
prediction of phenomena than are the other competing models.

6.2.2c. The Vacancy Model: A Fused Salt Is an Ionic Lattice with

Numerous Vacancies. The simplest model of an ionic liquid is the quasi-
lattice or vacancy model originated by Frenkel and developed by Stillinger.
The picture is that of an ionic lattice, into which are injected vacancies of
a particular type known as Schottky defects. A Schottky defect is produced
in the following process. Ions are removed from lattice sites in the interior
to the surface of the crystal (Fig. 6.8). Thus, vacancies are produced inside
the system and, simultaneously, there is volume expansion through the
advance of the frontier (surface). t
As the temperature of the solid lattice is increased, the number of
Schottky vacancies increases until, at the melting point, there is a qualitative

t It will be seen later that the ion does not, in fact, jump from within the liquid to the
surface in order to make room for the hole. The actual process involves a train of ions
each of which moves outward by only a small amount.
 IONIC LIQUIDS 527

VACANCY MODEL
Ie 0
0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0
0 0 0 0
0
0 0 0 0
~ 0 0 0

Fig. 6.9. The vacancies occur at lattice

sites in the quasi-lattice model.

change in degree of order; the disturbance of the lattice is so great that long-
range order disappears. In this model, a sufficient number of Schottky
vacancies is introduced to account for the volume increase on fusion.
The location and size of the vacancies in a fused salt follow as a natural
consequence from the assumption that the empty space consists of Schottky
defects. The vacancies are about the same size as the positive and negative
ions of the ionic lattice. Further, the vacancies occur at lattice sites. This is
the characteristic feature of the quasi-lattice model (Fig. 6.9).

6.2.2d. The Hole Model: A Fused Salt Is Full of Holes like Swiss
Cheese. In the Schottky-vacancy model, the parent ionic lattice exerted a
dominating influence on the sizes and positions of the vacancies in the liquid
electrolyte. The hole model proposed by FUrth represents an emancipation
from that influence. The description of the empty space is not in lattice
language, i.e., in the language of a three-dimensional array of points.
In the hole model, it is considered that the sizes and spatial location
of the empty regions in the fused salt are random. These randomly located
and variable-sized vacancies are called holes (Fig. 6.10). Thus, the liberation
from lattice concepts leads to a fundamentally different model.
What is the process by which holes are produced? It is by a process
somewhat analogous to the formation of a vacancy in a crystal. The dis-
placement of an ion from a lattice site produces a vacancy at its former

Fig. 6.10. The hole model with randomly

located and variable-sized holes in the liquid.
528 CHAPTER 6

(0) Before hole is (b) After hole is

formed formed

$ WH~
Fig. 6.11. The formation of a hole in a
liquid by the relative displacement of ions
in contact.

site. In the case of the vacancy, however, the ion is removed so far from
the original site that the displaced ion can be forgotten altogether. Suppose
instead that, in the course of thermal motion, some of the ions constituting
a cluster are displaced relative to each other but only by small amounts.
Then, a hole is produced between them (see Fig. 6.11). Its size depends on
the extent of displacement, which must be random because thermal motions
are random. The hole size must therefore be a random quantity. Further,
since thermal motions occur everywhere in the liquid electrolyte, holes can
appear and disappear anywhere in this liquid.
An equivalent description (Fig. 6.12) is that holes occur by fluctuations
in local density, i.e., in njV, the number of ions per unit volume in a given
locality of the liquid. The volume V is constant; hence, the density can
change only through a change in n, i.e., by the ions' moving farther apart
(the hole size increases) or closer together (the hole size decreases).
The holes in a liquid electrolyte resemble the holes in Swiss cheese.
This is why, in the matter of their randomness of size and location, the hole
theory has been referred to in homely terms as the "Swiss cheese model."
What the cheese represents, however, is the time-average picture of the holes.
In the model, holes are continuously forming and disappearing, moving,
coalescing to form larger holes, and diminishing into smaller ones.
(0) Before hole is formed (b) After hole is formed

v",m.v--@
Fig. 6.12. The formation of a hole can also be looked
at in terms of the number of ions occupying a volume V.
In (a) seven particles occupy the volume V before the
hole is formed, and, in (b), six particles occupy the same
volume after hole formation.
 IONIC LIQUIDS 529

6.2.3. Gas-Oriented Models for Liquid Electrolytes

6.2.3a. The Cell- Theory Approach. A particle in a dilute gas has
available for its motion the whole volume occupied by the gas, i.e., the
entire volume of the container. As the pressure is increased, the spatial
domain accessible to each particle is decreased but, as long as the system
is a gas, every particle continues to possess the freedom to move in any
part of the available volume.
When, however, there is a phase transition from the gaseous to the
liquid state, the freedom (of motion) of a particle is largely curtailed. The
neighboring particles (of any particle) conspire to confine the central particle
in a "cell." The conspiracy is never completely successful; the central particle
does manage occasionally to break out. Nevertheless, it spends much of its
time imprisoned within its cell.
The cell theory of liquids (derived from a compressed-gas point of
view) proceeds on the basis that, if a liquid consists of N particles, then the
volume occupied by the liquid can be conceptually divided into N identical
cells. Now, cells are larger than their occupants; hence, the confined particle
is free to move within the cell. It has afree volume available for its motion.
This free volume VI is equal to the average volume available to each particle,
ii, minus the volume Vo of the particle considered an incompressible hard
sphere (Fig. 6.13). (The average volume per particle is simply equal to the
total volume V of the liquid divided by the number N of particles.) In
symbols,
Vf = ii - Vo (6.1)
V
=--Vo (6.2)
N
Cell {volume, ii l

Ion {volume, Vol

Ion
Fig. 6.13. The free volume VI available for
the motion of an ion is equal to the average
volume If per ion minus the hard-sphere
volume vo• of the ion.
530 CHAPTER 6

The cell theory accomplishes the task of picturing how the freedom
of a particle in the gas phase diminishes upon its confinement in the liquid
state. The decrease of freedom occurs through a restriction of the free
volume accessible for its motion. However, the cell model runs into three
basic difficulties.
Firstly, the cell model restricts the motion of a particle to the confines
of its cell. How then can the model explain transport properties? Transport
requires that particles be able to migrate from cell to cell. Secondly, by
insisting that a particular particle be sentenced to confinement in a particular
cell, the disorder (randomness) accompanying the exchange of particles
between cells is ruled out. The result is that the calculated entropy of
fusion-which is a measure of the increase in randomness predicted by the
model as its version of what really happens on melting-becomes in fact
much smaller than that observed.
More decisive in counting against the model than either of these dif-
ficulties, however, is the third one, which arises from the apparent con-
tradiction of volume expansion on melting, along with a small decrease in
mean internuclear distance (cf Section 6.2.2b). On the basis of a cell model
for a liquid, the only explanation for the positive volume increase on fusion
is to have all the ions move away from each other, i.e., an increase in average
cell volume. But this, of course, would mean an increase in the average
internuclear distance on fusion (as long as multiple occupancy of cells is
prohibited )-an increase which contradicts experiment (cf Section 6.2.2a).

6.2.3b. The Free Volume Belongs to the Liquid and Not to the Par-
ticles: The Liquid Free- Volume Model. In the cell model, the liquid was
divided into identical cells, each cell marking the boundaries of the mo-
tion of a particle and thus possessing a cell free volume. The difficulties
inherent in this model (see Section 6.2.3a) have been overcome in the
liquid free-volume model developed by Cohen and Turnbull.
This progress is achieved as follows. The liquid first appropriates the
available free volume; hence the term liquid free volume. The liquid then
distributes its free volume among the N particles. The distribution, however,
is not done equally to each molecule. There is a statistical distribution of free
volumes. In other words, Nl particles are assigned a free volume VI; N2 par-
ticles, a free volume V 2 ; etc. In general, Ni particles possess a free volume Vi-
In this model, the free volume of a particle is not its inalienable property.
Thermal forces are responsible for the statistical distribution of free volumes.
Hence, these same forces cause the free volume of a particle to fluctuate in
time.
 IONIC LIQUIDS 531

Cell A Cell B

(0)

( b)

Cell A Cell B
Fig. 6.14. In the liquid free-volume mod-
el. the motion of an ion toward another is
accompanied by an expansion of the cell of
the former and a contraction of the cell of
the latter. (a) The positions of a positive and
a negative ion and the sizes of their cells
before motion of the positive toward the
negative ion; (b) the situation after motion.

The movement of a particle from one position to another implies the

expansion of the cell of the moving particle and the contraction of the
neighboring cell (see Fig. 6.14). The expansion of the first cell (and the
associated energy increase) is just compensated for by the contraction of
the neighboring cell (and the associated energy decrease). Hence, there is
a zero net energy change involved in one cell's expanding and its neigh-
boring one's contracting as a result of the movement of a particle. In this
way, the liquid free-volume model can accommodate the movement of
particles, i.e., transport properties such as diffusion.
Once a movement of a particle is permitted and cell free volumes
fluctuate, there is sufficient randomness introduced into the model to explain
the experimental fusion entropies.
The volume increase that occurs on fusion implies an increase in the
total liquid free volume. This total free-volume increase does not however
imply any increase in mean internuclear distance. In all parts of the liquid,
except those occupied by free space, the particles have the same inter-
nuclear distances as in the crystalline state.
Thus, by setting the cell free volumes into thermal fluctuations, the
liquid free-volume model can overcome some of the difficulties (see Section
6.2.3a) which troubled the simple cell theory.
532 CHAPTER 6

6.2.4. A Summary of the Models for Liquid Electrolytes

A simple ionic liquid can be considered to arise either from the fusion
of an ionic lattice or from the condensation of a vapor of ions. Thus,
there are lattice-based and gas-based models for a fused salt. The central
fact which a model must attempt to explain is the volume increase upon
fusion, which is usually observed along with a retention or even diminution
of the mean interionic distances. This volume increase without a correspond-
ing increase of mean interionic distance suggests that, in the process of
fusion, empty space is introduced into the ionic liquid. The mode of des-
cription of this empty space is what differentiates one model from another.
The vacancy model treats a fused salt as an ionic crystal with Schottky
defects which arise when ions from lattice positions move to the crystal
surface. The movement to the surface accounts for the volume expansion,
and the creation of vacancies inside the system accounts for the empty
space. The vacancy model thus requires vacancies to be of the same size
as ions and to occur at lattice sites.
In the hole model, empty space is considered to arise from thermally
generated fluctuations in local density. The holes which constitute the empty
space are random in size and location. At anyone instant of time, the holes
would make a section through the liquid electrolyte appear like one through
Swiss cheese. It is important to realize that holes are constantly under-
going changes in size with new ones' being formed and old ones' destroyed.
Further, the holes can also move by the mechanism of having an ion jump
into one hole and thus create a new hole in the place it has just vacated.
The cell theory approach is based on considering the territory available
for the motion of a particle. The transition from the gaseous state to the
liquid state is accompanied by a drastic reduction in this territory. In the
gas phase, the entire volume of the container is accessible to a particle;
in the liquid state, the particle is confined to a cell. Within the cell, however,
the particle has a free volume available for its motion. The cell theory
prohibits multiple occupancy of cells and, therefore, cannot explain the
volume increase upon fusion.
The liquid free-volume theory argues against identical cells and equal
free volumes for the particles. It is the liquid as a whole which has a certain
free volume, and there is a statistical distribution of the free volumes in
the cells. Thus, the cells expand and contract, and the mean distance be-
tween particles need not increase to provide for the increase in the liquid
free volume which accounts for the volume increase upon fusion.
Now that the basic models for a simple ionic liquid have been quali-
 IONIC LIQUIDS 533

tatively described, the next task is to consider which model makes the
most successful predictions of experimental results. At first, however, one
should describe the more experiment-consistent models in quantitative
terms. Since, as will be seen later, the hole model appears to yield more
qualitative and quantitative agreement with experiment than do the other
models, the treatment from here on will be characterized by emphasis on
the hole model.

Further Reading
1. H. Eyring, J. Chem. Phys., 4: 283 (1936); 5: 896 (1937); J. Phys. Chem., 41,
249 (1937); J. Chem. Phys., 9: 393 (1941); Proc. Nat. Acad. Sci. U.S., 44:
683 (1958); 46: 333 (1960).
2. W. Altar, J. Chem. Phys., 5: 577 (1937).
3. R. Fiirth, Proc. Cambridge Phil. Soc., 252, 276, 281 (1941).
4. M. Born and H. S. Green, Proc. Roy. Soc. (London), A188: 10 (1946).
5. J. Frenkel, Kinetic Theory of Liquids, Oxford University Press, New York,
1946.
6. M. H. Cohen and D. Turnbull, J. Phys. Chem., 29: 1049 (1958); 31: 1164
(1959); 34: 120 (1961).
7. H. Reiss, H. L. Frisch, E. Helfand, and J. L. Lebowitz, J. Chem. Phys., 32:
119 (1960).
8. F. H. Stillinger, Chapter 1 in: M. Blander, ed., Molten Salt Chemistry, Inter-
science Publishers, Inc., New York, 1964.
9. H. Bloom, Chapter 1 in: B. R. Sundheim, ed., Fused Salts, McGraw-Hili
Book Company, New York, 1964.
10. K. D. Luks and H. T. Davis, Eng. Chem. Ind. Fundamentals, 6: 194 (1967).

6.3. QUANTIFICATION OF THE HOLE MODEL FOR LIQUID

ELECTROLYTES

6.3.1. An Expression for the Probability That a Hole Has a Radius

between rand r + dr
To quantify the hole model, it is necessary to calculate a distribution
function for hole sizes. As a first step toward this calculation, one can
consider a particular hole in a liquid electrolyte and ask: What are the
quantities (or variables) which are required to describe this hole? This
problem was resolved by a formulation published by FUrth in 1941.
Since a hole in a liquid can move about like an ion or other particle,
the dynamical state of a hole is specified in the same way that one describes
the dynamical state of a material particle. Thus, one must specify three
position and three momentum coordinates: x, y, Z, and Pr' PY ' P:. There
534 CHAPTER 6

G·
Fig. 6.15. The breathing motion of a hole
involves its radial expansion.

is, however, an extra feature of the motion of a hole, which is not possessed
by material particles. This feature concerns what is called the breathing
motion of a hole (Fig. 6.15), i.e., the contraction and expansion of the hole.
To characterize this breathing motion, it is sufficient to specify the hole
radius r and the radial momentum Pr corresponding to the breathing
motion. Hence, to characterize a hole completely, it is necessary to specify
eight quantities: x, )" Z, PI! PII ' Pz, r, and p" whereas the first six only
are adequate to describe the state of motion of a particle.
According to the usual equations of classical statistical mechanics
which are used to express velocities and momenta distributed in three
dimensions, the probability P that the location of a hole is between x and
x + dx, y and y + dy, z and z + dz, that its translational momenta lie
between Px and Px + dpx, Py and Py + dp y, pz and pz + dpz, that its breathing
momentum is between Pr and Pr -:- dPr, and, finally, that its radius is be-
tween rand r + dr, is proportional to the Boltzmann probability factor, i.e.,

P dx dy dz dpx dp~ dpz dr dPr ex e- ElkT • dx dy dz dpx dp~ dpz dr dPr (6.3)

where E is the total energy of the hole.

Since the desired distribution function only concerns the radii (or
sizes) of holes, it is sufficient to have the probability that the hole radius
is between rand r + dr irrespective of the location and the translational and
breathing momentum of the hole. This probability P dr of the hole radius'
being between rand r + dr is obtained from Eq. (6.3) by integrating over
all possible values of the location, and of the translational and breathing
momentum of the hole, i.e.,

P dr ex: (f f f f f f f e-ElkT • dx dy dz dpx dpv dpz dPr) dr (6.4)

But the total energy of the hole does not depend upon its position, i.e.,
E is independent of x, y, and z. Hence,

P dr ex: (f fff e- ElkT • dpx dpydpzdPr)(f ff dx dy dz) dr (6.5)

 IONIC LIQUIDS 535

Further,
fff dx dy dz = V (6.6)

the volume of the liquid. Thus, by incorporating this V into the propor-
tionality constant implicitly associated with Eq. (6.5), one has

P dr ex (J fff e-ElkT • dpx dpu dpz dPr) dr (6.7)

Now the total energy E consists of the potential energy W of the hole
(i.e., the work required to form the hole) plus its kinetic energy. This
kinetic energy is given by

(6.8)

where In[ is the apparent mass t of the hole in its translational motions,
and nJ 2 is the apparent mass t in its breathing motion. Hence,

(6.9)

Inserting this value of E into the expression (6.7) for the probability
of the hole's having a radius between rand r + dr, one has

(6.10)

From the standard integral

J oo
-00
e-a.rdx
,
= V+na
-- (6.11 )

it is clear that

t Any entity that moves displays the property of inertia, i.e., resistance to a change of
its state of rest or uniform motion. That is, the entity has a mass. If the entity is not
material (a hole is a region where, in fact, there is no material), one refers to an apparent
mass. Holes in semiconductors have apparent masses like holes in liquids. The inertia
of the hole arises as a result of the displacement of the liquid around the hole as it
moves, which gives rise to a dissipation of energy (Appendix 6.1.)
536 CHAPTER 6

and
(6.13)

By using these values of the integrals in Eq. (6.10), the result is

Pdf ex:: (2nkTYmli m2!e- W / kT dr (6.14)

It can be shown, however, that (ef Appendix 6.1)

(6.15 )
and
(6.16)

Hence, after taking all quantities which are radius independent into the
proportionality constant A, one has, by combining Eqs. (6.14), (6.15),
and (6.16),
(6.17)

The evaluation of the constant is achieved through the following

argument. The probability that a hole has some radius must be unity (a
probability of unity for an event corresponds to the certainty of its occur-
rence). Equation (6.17) expresses the probability of the radius of the hole
lying between rand r + dr. Similarly, one can write down the probabilities
of the radius' being between r l and r 1 + dr, between r 2 and r 2 + dr, etc.
If all these probabilies for r from zero to infinity are summed up (or in-
tegrated), then the sum must be unity, i.e.,

fcc
o
P dr = I = fX> Artie-lV,leT dr
• 0
(6.18 )

However, to carry out this integration, one must know whether W is a

function of r, i.e., one must understand what determines the work of for-
mation (or the potential energy) of a hole of radius r.

6.3.2. The Furth Approach to the Work of Hole Formation

An ingenious calculation of the work of hole formation was made by
FUrth, who treated holes in liquids, the sizes of which are thermally distri-
buted, in an article published in an erudite but little-read university journal,
an act which delayed recognition of the model. A hole in a fused salt is
considered to behave similarly to a bubble in a liquid (Fig. 6.16). There is
a net pressure acting on the surface of a bubble. The surrounding liquid
exerts a hydrostatic pressure PIon the bubble surface. Inside the bubble,
 IONIC LIQUIDS 537

(a) (b)

~I'i;-@
a liquid
Bubble
BOilinO
liquid

Fig. 6.16. The basis of the hole model of Furth is the analogy
between (a) a hole in a liquid and (b) a bubble in a liquid. An
inward pressure PI and an outward pressure Po act on the
bubble surface.

however, there exists included vapor, which exerts an outward pressure Po

on the surface. The net pressure is therefore PI - PO. Further, surface
tension operates in the direction of reducing the surface area and, therefore,
the surface energy of the bubble.
The total work required to increase the bubble size consists of two
parts, the volume work (PI - PO) V and the surface work yA, where
PI - Po, V, y, and A are the net pressure on the bubble surface, the in-
crease of volume, the surface tension, and the increase of surface area of
the bubble, respectively. Thus, the work done in making a bubble grow
to a size having radius r is

(6.19)

Simple numerical calculations show that the first term, i.e., the volume term,
is negligible compared to the second, or surface, term for bubbles of less
than about 10-5 cm in diameter. Hence, on neglecting the volume expansion
work, the work of bubble formation (i.e., the work of increasing its radius
from 0 to r) reduces to
(6.20)

This expression can also be obtained from the general equation (6.19) by
setting PI = Po. This equality between PI and Po represents the condition
that the liquid is boiling. The analogy between a hole and a bubble consists,
therefore, in assuming that the work of hole formation is given by the
expression for the work of bubble formation in a liquid.

6.3.3. The Distribution Function for the Sizes of the Holes in

a Liquid Electrolyte
Now that an expression for the work of hole formation has been obtain-
ed, it can be inserted into Eq. (6.18), which must be integrated to evaluate
538 CHAPTER 6

the constant A. One has

roo P dr = I (6.21 )
• 0

where
4;ry
a =--
kT

To carry out the integration, the following standard formula is used

I X)
r2l1e-ar,drI=
x 3 x 5·· ·(211 - I)
X
li+;r
-
O
21+la" a

where 11 is a positive integer.

The integral in Eq. (6.21) corresponds to the standard formula with
II = 3, and, therefore,

_ A 15;rk I
(6.22)
- 16 a72

Hence, the constant in Eq. (6.21) is given by

A = ~ a72 = ~ (4;rY)7/2 (6.23 )

15;ri 15;ri kT

and the expression (6.17) for the probability of the existence of a hole of a
radius between rand r + dr becomes

(6.24)

Pdr

Hole radius

Fig. 6.17. How the probability P dr that a hole has a radius

between rand r + dr varies with r.
 IONIC LIQUIDS 539

This is the basic distribution function (Fig. 6.17) from which the average
hole volume and radius can be obtained.

6.3.4. What Is the Average Size of a Hole?

The average radius <r) of a hole is obtained from Eq. (6.24) by

mUltiplying the probability (of the hole radius being between rand r dr) +
by the radius of the hole and integrating this product over all possible
values of r. This is, in fact, the general method of getting average values of
a quantity when the probability is given. Thus, the average hole radius is

<r) = J~ r P dr

= -16- a7/2 Joc r e- 7 ar 2 dr (6.25)

15n! 0

The integral in Eq. (6.25) can be evaluated by using the substitution t = ar2
t~ar2.dt/2a~r dr 1 Joc 3 -t
(6.26)
t31 a 3_
_r 6 • -2
a4 0 t e dt

which leads to

<r>-_ 8 1
15ni (if'
foc0 t 3 -t
e
d
t (6.27)

The integral f:t 3e- t dt is the gamma function r(3 + 1) (Appendix 6.2)
and is equal to 3! from

rex + 1) = f~ (Xe- t dt = x! (6.28)

Hence, Eq. (6.27) becomes

<r) = 15nt
8 (kT)!
y 1
(4n )! x 3 x 2 x 1

= ~
5n
(kT)!
y

= 0.51 (k:t (6.29)

540 CHAPTER 6

Since the average surface area of a hole calculated by this procedure gives

kT
4:7 ~r2) = 3.5 - (6.30)
Y

then, from Eqs. (6.29) and (6.30) one obtains

4:7: r 2 ) 3.5kT/y
[0.51(kTjy)1F = 13.5 (6.31 )
(r/

If all the holes were of the same size, this ratio would be 4:-z::-:: 12.6. The
value given by Eq. (6.31) differs from 4:7 by about 8%, which shows that
one obtains a fairly good approximation by taking the holes to be of the
same size.
What value of surface tension is to be used? The hole theory boldly
uses the macroscopic value with suitable correction for the curvature of
holes of atomic dimensions. Careful theoretical analysis appears to support
the use of the macroscopic value of surface tension in a molecular model.
What typical values of mean hole radius does Eq. (6.29) yield (Table
6.11)? By using the macroscopic surface-tension value, it is found from
Eq. (6.29) that the average radius of a hole in molten KCl at 900 :C is
2.1 A. The mean ionic radius, however, is 1.6 A.
A t)picai hole therefore is roughly the same size as an iOI2 and can
accommodate an ion. This result is all the more remarkable because of
the process by which it has been attained. One began by considering that

TABLE 6.11
Mean Hole Radius for Various Molten Salts at 900 0 C

Surface tension, Mean hole volume, Mean hole radius,

Molten salt
dyne cm- 1 Aa A

NaCl 107.1 32 1.7

NaBr 90.5 41.7 1.9
KCI 89.5 42.3 1.9
KBr 77.3 52.7 2.1
CsCl 72.7 51.8 2.2
NaI 66.4 66.2 2.3
KI 60.3 76.6 2.3
 IONIC LIQUIDS 541

a liquid electrolyte was a liquid continuum interspersed by holes of random

size and location. Holes in a fused salt could be treated somewhat like
bubbles in a boiling liquid. Thus, the work of hole formation was taken
to be equal to the work of expanding the surface area of a bubble in a boiling
liquid. With the use of this expression and simple probability arguments,
the average hole radius was calculated.
At a fixed temperature, the only parameter determining the mean hole
size is the surface tension. Though one is aiming at a microscopic (structural)
explanation of the behavior of ionic liquids, one goes ahead and uses the
macroscopic value of surface tension. Thereafter, the mean hole radius
turns out to have the same order of magnitude as the mean ionic radius.
The connection with an experimental fact is through the surface tension.
The provision from the theory of an indication of molecular-sized holes
supports the applicability of the approach. How can bubbles in liquids be
used as the basis of a calculation of liquid properties? The answer shall
be given by the degree of ability of such an approach to predict facts-for
example, the compressibility and coefficient of expansion. The fact that it
indicates just the size of holes needed for ions to jump into and diffuse is
an encouraging indication.

Further Reading

1. W. Altar, J. Chern. Phys., 5: 577 (1937).

2. R. Furth, Proc. Cambridge Phil. Soc., 252, 276, 281 (1941).
3. N. E. Richards, Proc. Roy. Soc. (London), A241: 44 (1957).
4. G. W. Hooper, Discussions Faraday Soc., 32: 318 (1962).
5. F. H. Stillinger, Chapter 1 in: M. Blander, ed., Molten Salt Chemistry, Inter-
science Publishers, Inc., New York, 1964.
6. K. D. Luks and H. T. Davis, Ind. Eng. Chern. Fundamentals, 6: 194 (1967).

6.4. TRANSPORT PHENOMENA IN LIQUID ELECTROLYTES

6.4.1. Some Simplifying Features of Transport in Fused Salts

An important characteristic of liquid ionic systems is that they lack

an inert solvent; they are pure electrolytes. Owing to this characteristic,
some aspects of transport phenomena in pure molten salts are simpler than
similar phenomena in aqueous solutions.
Thus, there is no concentration variable taken into account in the
consideration of transport phenomena in a pure liquid electrolyte. Hence,
there cannot be a concentration gradient in a pure fused salt, and, without
 542 CHAPTER 6

a concentration gradient, there cannot be net diffusion." In an aqueous

solution, on the other hand, it is possible to have a concentration gradient
for the solute and thus have diffusion.
Another consequence of the absence of a solvent is that the mean
ion-ion interaction field is constant (at constant temperature) in a pure
liquid electrolyte. In ionic solutions, however, the extent of ion-ion inter-
action is a variable quantity. It depends on the amount of solvent dissolving
a given quantity of solute, i.e., on the solute concentration.

6.4.2. Diffusion in Fused Salts

6.4.2a. Self-Diffusion in Pure Liquid Electrolytes: It May Be Re-
vealed by Introducing Isotopes. In the absence of a solvent, it is mean-
ingless to consider a pure liquid electrolyte, e.g., NaCl, as having dif-
ferent amounts of NaCl in different regions.
The possibility might be considered that the system could be made to
have more ions of one species, e.g., Na-, in one region than in another.
However, this is impossible because any attempt of a single ionic species
to accumulate in one region and decrease in another is promptly stopped
by the electric field which develops as a consequence of the separation of
charges. Overall electroneutrality must prevail, i.e., there can be no con-
gregation of an ionic species in one part of the liquid.
An electric field that results from incipient charge separation and
reduces the applied field is also set up in aqueous solutions. but. here. owing
to the much smaller number of ions per unit volume. the potential difference
arising from charge separation is spread over macroscopic distances, say,
microns, as, e.g., in the case of a liquid-junction potential (Section 4.5.4).
The restricting field, therefore, is much smaller in aqueous solutions than
in pure liquid electrolytes.
Fortunately, electroneutrality only requires that the total positive charge
in a certain region is equal to the total negative charge. Suppose therefore
that, in liquid sodium chloride electrolyte, a certain percentage of the Na-
ions is replaced by one of a radioactive isotope of sodium. There is no
difference between the Na 22 and Na23 as far as the principle of electro-
neutrality is concerned; it is only required that the number of Na- ions
plus the number of tagged Na*- ions are equal to the total number of Cl-

t The addition of tracer ions is not considered here because one can look upon a liquid
electrolyte containing tracer ions as a mixture of pure electrolytes-one pure electrolyte,
e.g., NaCl, without tracer ions, and the other, e.g., Na*CI, with tracer ions. In mixtures
of pure electrolytes there can be a concentration variable, e.g., of the tracer ions.
 IONIC LIQUIDS 543

@ T09ged positive ion

(±) Non- radioactive positive ion
e Non-radioacTive ne9allve ion

Fig. 6.18. The principle of electroneutrality

is satisfied if the number of tagged positive
ions plus the number of nonradiactive posi-
tive ions is equal to the total number of
negative ions.

ions (Fig. 6.18). But the labeled Na*+ ions and the nonradioactive Na+
are completely different entities from the point of view of a counter; only
the former produce the scintillations.
Herein lies a method of manifesting the diffusion of ions in pure ionic
liquids. One takes a pure liquid electrolyte, say, NaCl, and brings it into
contact with a melt containing the same salt but with a certain proportion
of radioactive ions, say, NaCl with radioactive Na* + ions. There is a neglig-
ible concentration gradient for Na+ ions, but a concentration gradient for
the tracer Na*+ ions has been created. Diffusion of the tracer commences
(Fig. 6.19).

Radioactive
positive ions

Concentration gradient
01 tracer

Diffusion of trocer

Fig. 6.19. The existence of a concentra-

tion gradient for tracer ions produces dif-
fusion of the tracer, i.e., tracer diffusion.
544 CHAPTER 6

~~~~~~~~~_ _ Oiffusion of
Capillary tracer ions
initially _+~~~~ inta copi llary
contains
melt
without
tracer

Melt containin9
tracer ions

Fig. 6.20. A schematic of an experiment to study

tracer diffusion. A capillary containing inactive melt is
dipped into a reservoir of melt containing tracer ions.
Tracer ions diffuse into the capillary.

If, therefore, a capillary containing inactive melt is suddenly introduced

into a large reservoir of tracer-containing melt at t = 0, then diffusion of
the tracer into the capillary starts (Fig. 6.20). At time t, the experiment
can be terminated by withdrawing the capillary from the reservoir. The
total amount of tracer in the capillary can be measured by a detector of
the radioactivity. From the study of the diffusion problem and the experi-
mentally determined average tracer concentration in the capillary, the dif-
fusion coefficient of the Na*+ ions is then calculated.
Since the tracer ions (e.g., Na*+) diffuse among particles (e.g., Na+)
which are chemically just like themselves, one often refers to the phenom-
enon as self diffusion (tracer diffusion is a more explanatory term) and
to the diffusion coefficient thus determined a the self-diffusion coefficient.

6.4.2b. Results of Self-Diffusion Experiments. Self-diffusion coef-

ficient studies with fused salts really began to gather momentum after
radioisotopes became widely available, i.e., after about 1950. Some of the
available data are presented in Table 6.12.
It can be seen that the diffusion coefficients of these liquid electrolytes
(near their melting points) are of the same order of magnitude (,...., 10-5 cm2
sec-l) as for liquid inert gases, liquid metals, and normal room-temperature
liquids (Table 6.13). This fact suggests that the mechanism of diffusion is
the same in all simple liquids, i.e., liquids where the particles do not associate
into pairs, triplets, or network structures (cf Sections 6.6 and 6.7), etc.
The order of magnitude of the diffusion coefficient has evidently more to
do with the liquid state than with the chemical nature of the liquid for,
in the case of crystalline substances, the diffusion coefficient ranges (Table
6.14) over about four orders of magnitude (10- 7 to 10-11 cm2 sec-I).
 IONIC LIQUIDS 545

TABLE 6.12
Tracer-Diffusion Coefficients

Temperature, Tracer diffusion coefficient,

Molten salt Tracer ion DC em' sec-1

NaCl 22Na+ 840 9.6 x 10-5

36Cl- 840 6.7 x 10-5
RbCl 86Rb+ 740 4.7 x 10-5
36cl- 740 4.2 x 10-5
CsCI 134cs+ 670 3.5 x 10-5
36cl- 670 3.8 x 10-5
65zn++ 600 0.6 x 10-5
36cl- 600 0.4 X 10-5
140Ba++ 1000 1.84 X 10-5
36Cl- 1000 2.99 x 10-5

An expected feature of the results on tracer diffusion is that the diffu-

sion coefficient varies with temperature. The temperature dependence ob-
served experimentally can be expressed in the usual exponential way

(6.32)

where Do is found to depend little on substance and temperature, and ED

is the activation energy for self-diffusion (Fig. 6.21). A tabulation of some
preexponential factors and the corresponding energies of activation for
diffusion is given in Table 6.15.

TABLE 6.13
Self-Diffusion Coefficients of Various Types of Substance

Temperature Diffusion coefficient

Type of substance Example
°C em' sec- 1

Liquid inert gas Ar -173 3.70 x 10-5

Room temperature liquid CCl, 25 1.41 x 10-5
Liquid metal Zn 420 2.07 x 10-5
Molten salt NaCI 840 DNa+ 9.6 X 10-5
DCl- 6.7 X 10-5
546 CHAPTER 6

TABLE 6.14
Tracer Diffusion Coefficients of Crystalline Substances near the Melting
Point

Substance Tracer D, cm 2 sec- 1 Temperature, DC

Na Na 22 1.7 X 10- 7 97
Ag AgllO 2.8 X 10- 8 900
NaCl Na 22 4.0 x 10-8 727
PbS ThB 1.4 x 10-9 1043
Pb ThB 5.5 X 10-10 324
PbI. ThB 7.7 X 10-11 315

In some cases, a deviation (Fig. 6.22) occurs from the straight-line

log D versus liT plots expected on the basis of the empirical exponential
law for the diffusion coefficient [Eq. (6.32)]. An example of such a de-
viating liquid electrolyte is molten ZnCl z , but, in the case of this substance,

c
~

(II
'u
....
(II
o
u
c:
o
II)

....
:::3
....

5.50

9.3

Fig. 6.21. The straight-line plot of log D versus 1 IT ob-

served in the case of the diffusion of CS'34 (0) and CI"' (0)
in molten CsCI.
 IONIC LIQUIDS 547

TABLE 6.15
Energies of Activation and Preexponential Factors for Self-Diffusion in
Molten Group I and II Chlorides

Molten 103 x Do*, Ed*' Temp. range,

salt/tracer cm 2 sec- 1 kcal mole- 1 °C

NaCl/Na 22 2.1 7.14 ± 0.25 820-1020

NaCl/Cp6 1.9 7.43 ± 0.84 826-1035
KCI/K42 1.8 6.88 ± 0.51 798-983
KCl/ClaG 1.8 7.13 ± 0.49 794--987
CaCl./Ca 4 • 0.38 6.13 ± 0.66 783-1004
CaCl./Cl36 1.9 8.86 ± 0.96 787-1019
SrCl./Sr89 0.21 5.38 ± 0.97 921-1120
SrCl./CI36 0.77 6.88 ± 0.72 912-1157
BaCI 2/Ba 140 0.64 8.96 ± 1.23 994-1207
BaC1 2/C136 2.0 9.48 ± 1.02 993-1203
CdCI./Cd 1l5m 1.1 6.84 ± 0.62 607-806
CdC1 2 /Cl36 1.1 6.80 ± 1.01 607-802

marked structural changes have been noted with increasing temperature, a

possible explanation for the deviation from the straight-line log D versus
liT plot.
The activation energy for self-diffusion is usually a constant, inde-
pendent of temperature. It is, however, characteristic of the particular liquid
electrolyte. The dependence of the activation energy for self-diffusion on the
nature of the fused salt has been experimentally found to be expressible in
a simple form (Fig. 6.23). The following relation is approximately applicable

En = 3.7RTm (6.33 )

where Tm is the melting point.

It is important to emphasize here that this same relation is valid for
many liquids other than the pure nonassociated liquid electrolytes, including
the liquid inert gases and the liquid metals.

6.4.3. The Viscosity of Molten Salts

Only in recent years has attention been paid to the flow of pure liquid
electrolytes (Table 6.16). The still-incomplete examination shows that vis-
cosity varies with temperature in a way strictly analogous to that of self-
548 CHAPTER 6

5·00.---------------------------,

11·0 13·0 15·0 17·0 19·0

104
TDK
Fig. 6.22. An example of a log D versus 1 IT plot
which is not a straight line. The curve is for the dif-
fusion of Zn 65 (0) and CP6 (D) in molten ZnCI 2 •

lor--------------------------------~----------l
Molecular liquids 0
Molten soll.J·Cotions 4~y-----3·74 9 ....
9 /0
RTm
8 ieAnions • // 0
9 .,/
Liquid metals c /'~
ED Liquid inert gases A • 8...'7 9
9 9 9/.... •
Kcolmole'
.~t--' • "
4
~/ e
o /'"
~ [] []
2 O,P
/4' [] []
",fi
o
o 400 800 1200 1600

Fig. 6.23. The dependence of the experimental energy of activation

for self-diffusion on the melting point.
 IONIC LIQUIDS 549

TABLE 6.16
Viscosities of Fused Salts

Fused salt Temperature. DC Viscosity. centipoise

CdCl 2 600 2.31

CdBr2 600 2.61
PbBr 2 550 2.98
PbCl 2 600 2.75
AgCl 600 1.66
AgBr 600 2.27
NaCl 900 1.05
NaBr 755 1.43
KCl 773 1.51
KBr 730 1.57

diffusion. For simple, unassociated liquid electrolytes, the temperature

dependence is given by an empirical equation

(6.34)

where 'YJo is a constant analogous to Do, and E~ is the energy of activation

for viscous flow (Fig. 6.24 and Table 6.17).

0.4 -

//
0.3-

0.21-

~ 0.11-
/
g. Ol-/X:
-0.1-
I I
80 9.0 10.0
.L. xlO 4
T

Fig. 6.24. The straight-line plot of log 11 versus 1 IT

for viscosity of a molten salt.
550 CHAPTER 6

TABLE 6.17
Temperature Dependence of Viscosity for Simple Molten Electrolytes

E~, E~,
Molten salt Molten salt
kcal mole- 1 kcal mole- 1

LiCI 8.8 KOH 6.1

LiBr 6.0 AgCl 2.9
LiI 4.4 AgBr 3.1
LiNO. 4.2 AgI 5.8
NaCl 9.1 AgNO. 3.1
NaBr 8.0 TINO. 3.6
NaI 7.4 CuCl 4.2
NaNO. 4.0 MgCls 6.5
NaNO. 4.0 CaCI. Q.5
NaOH 5.5 BaCl s 9.0
NaCNS 5.8 PbCl. 6.7
KCl 7.8 PbBr. 6.2
KBr 7.5 CdCl s 4.0
KI 9.2 CdBr. 4.5
KNO. 3.7 NH 4NO. 4.6

This expression is formally analogous to Eq. (6.32) for the dependence

of the self-diffusion coefficient upon temperature. In fact, for simple liquid
electrolytes, the experimental activation energy for viscous flow is given
by an expression (Fig. 6.25) identical to that for self-diffusion, i.e.,

(6.35)

This implies that the basic factors determining viscous flow and self-
diffusion are the same.
Simple ionic liquids have viscosities in the range of I to 5 centipoises.
When, however, there is association of ions into aggregates, as, for example,
in ZnCl 2 near the melting point, the viscous force resisting flow of the melt
increases. Such complex ionic liquids are discussed later (Section 6.6).

6.4.4. What Is the Validity of the Stokes-Einstein Relation in

Ionic Liquids?
All transport processes (viscous flow, diffusion, conduction of elec-
tricity) involve ionic movements and ionic drift in preferred direction; they
must, therefore, be interrelated. A relationship between the phenomena of
 IONIC LIQUIDS 551

10

~
o
E 6

o
<..>
~

F"
w 4

o
o 200 400 600 800 1000 1200
TmoK
Fig. 6.25. The dependence of the experimental energy of activation for viscous
flow on the melting point.

diffusion and viscosity is contained in the Stokes-Einstein equation (4.176).

D=~ (6.36)
6:rrrrJ

The validity of this equation in ionic solutions has been discussed in

Section 4.4.8. It is therefore of interest to inquire about the applicability
of the relation in ionic liquids, i.e., fused salts. To make a test, the experi-
mental values of the self-diffusion coefficient D* and the viscosity 'Y} are
used in conjunction with the known crystal radii of the ions. The product
D*'Y}IT has been tabulated in Table 6.18, and the plot of D*'Y}IT versus l/r
is presented in Fig. 6.26, where the line of slope k16:rr corresponds to exact
agreement with the Stokes-Einstein relation. t

t The essential applicability of this phenomenological equation is well shown by using

the numerical comparison of DI)/T = k/6nr. The right-hand side is 0.7 x 109 for
r = 3 A, and the mean of the experimental values is 0.6 x 109 •
552 CHAPTER 6

TABLE 6.18
The DT]/T of Some Molten Salts for Testing the Stokes-Einstein Relation

D.",
Molten Temperature, Di x 105 , -'- x 100
Tracer OK
T '
salt em 2 see- 1
dyne deg- 1

NaCI Li 6 1180 13.2 1.09

NaCI Na 22 1180 10.2 0.85
NaCl K42 1180 9.7 0.81
NaCl Rb 86 1180 9.2 0.76
NaCl C S'34 1180 8.9 0.74
KCl K42 1150 8.9 0.63
NaI Na 22 1026 9.5 1.20
CaCI. Ca45 1154 2.6 0.79
SrCl 2 Sr 89 1260 2.4 0.70
BaCl 2 Ba140 1356 2.4 0.69
CdCl 2 Cd1l5m 925 2.7 0.58
PbCl. Pb 210 851 1.5 0.59
LiNO a Li 6 581 2.1 1.77
NaNO a Na 22 638 2.6 0.88
KNO a K42 667 2.1 0.68
AgNO a AgllO 534 1.5 0.80

It can be seen that there is a significant fit. The anions, particularly

those of the Group II halides, are not very consistent with the Stokes-
Einstein relation. But the poor fit is offset by the better Stokes-Einstein
behavior of the cations. The relatively good fit of the cations tempts one
to conclude that there is a particular reason, in the case of anions only,
which leads them into deviations. Some attempts have been made to eluci-
date this reason. For instance, it has been suggested that, since the anions
are larger than the cations, they require greater local rearrangements at a
site before they can jump into it, i.e., greater entropies of activation (Ap-
pendix 4.3).
The Stokes-Einstein relation is based on Stokes' law in hydrodynamics
according to which the viscous force experienced by a large sphere moving
in an incompressible continuum is 6nrfJv (cf Section 4.4.8). Hence, the
Stokes-Einstein relation depends on the view that an ion moving in an
electrolyte experiences a Stokes force 6nr'YJv, even though the ions do not
move in a continuum but among particles which are of approximately the
same dimensions as the ions themselves. In view of the "far-flungness"
 IONIC LIQUIDS 553

2.0r-------------------..

-
.,co
I

'0

~ 1.0-
'0 oNa+in NaI
oX OK + in KCI
o cat+in CaClz
~ o Srz+in SrClt
~
c5 6 Bat+i n Ba Clz
<> Cdt+in CdCl z
0.5 t-- Q PI" in Pb CI.

I I
1.0 2.0
Itr X 10-A cm-'
Fig. 6.26. When DIT is plotted against 1 If, a straight line
of slope kl6n should be obtained if the Stokes-Einstein rela-
tion is applicable to molten salts. The experimental points are
indicated in the figure to show the degree of applicability of
the Stokes-Einstein relation to molten salts.

of the similarity between an ion in a structured medium and a sphere in an

incompressible continuum, the rough applicability (in fused salts) of the
Stokes-Einstein equation is somewhat unexpected.

6.4.5. The Conductivity of Pure Liquid Electrolytes

The electrical conductance of molten salts is the easiest transport
property to measure. In addition, knowledge of the order of magnitude
of the equivalent conductivity of a pure substance was used as an important
criterion of the nature of the bonding present. For these reasons, the elec-
trical conductance of ionic liquids has been the subject of numerous studies.
The equivalent conductivities of some of the fused chlorides are given
554 CHAPTER 6

TABLE 6.19
Equivalent Conductivities of Molten Chlorides

HCI
~ 10- 6

LiCl BeCl! BCl 3 CCI.

166 0.086 0 0

NaCl MgCl. AlCl 3 SiCl. PCls

133.5 28.8 15 x 10- 6 0 0

KCI CaCI. ScCl 3 TiCI. YCI.

103.5 51.9 15 0 0

RbCl SrCl. YCl 3 ZrCl. NbCl s MoCi s

78.2 55.7 9.5 x = 2 X 10- 7 X= 1.8 X 10-·

CsCl BaCI, LaCI 3 HfCI. TaCi s WCl.

66.7 64.6 29.0 x = 3 X 10- 7 X= 2 X 10- 6

ThCl. VCl.
16 x= 0.34

in Table 6.19, where the substances have been arranged according to the
periodic table. The heavy line zigzagging across the table separates the ionic
from the covalent chlorides. This structural difference is shown up sharply
in the orders of magnitude of the equivalent conductivities.
Two further correlations emerge from Table 6.19. Firstly, the equivalent
conductivity decreases with increasing size of the cation (Table 6.20);

TABLE 6.20
Dependence of Equivalent Conductivity upon Cationic Radius

Molten salt Radius of cation, A Equivalent conductivity

LiCl 0.68 183

NaCi 0.94 150
KCI 1.33 120
RbCI 1.47 94
CsCI 1.67 86
 IONIC LIQUIDS 555

TABLE 6.21
Dependence of Equivalent Conductivity upon Valency of Cation

Molten salt Valency of cation Equivalent conductivity

NaCl 150
MgCl. 2 35
AICl, 3 15.1
SiCl, 4 <0.1

secondly, there is a decrease in equivalent conductivity in going from the

monovalent to the divalent and then to the trivalent chlorides (Table 6.21),
probably because of an increase in covalent character in this order.
As with the other transport properties, the specific (or equivalent)
conductivity of fused salts varies with temperature. t For most pure liquid
electrolytes, the experimental log A versus 1IT plots are essentially linear
(Fig. 6.27). This implies the usual exponential dependence of a transport
property upon temperature
(6.37)

For some substances, the plots are slightly curved. In these cases, structural
changes (e.g., the breaking-up of polymer networks) occur with change of
temperature.
When the activation energies for conduction are computed from the
log A versus liT plots, it is seen (Table 6.22) that they are lower than the
activation energies for viscous flow and self-diffusion, i.e.,

(6.38)

6.4.6. The Nernst-Einstein Relation in Ionic Liquids

6.4.6a. The Nernst-Einstein Relation: Its Degree of Applicability.

Just as the Stokes-Einstein equation gives the relation between the transport
of momentum (viscous flow) and the transport of matter (diffusion), the
connection between the transport processes of diffusion and conduction

t A convenient means of comparing different salts is to use "corresponding temperatures";

usually 1.05 or 1.10 times the value of the melting point in degrees Kelvin is used for
this purpose.
556 CHAPTER 6

1.5'--="=:----~:__--_:::"'=_--____::"'=_--____::_'_::_---'

IO·/TOK
Fig. 6.27. The straight-line plot of log A versus 1 IT.

leads to the Nernst-Einstein equation (cf Section 4.4.9), i.e.,

(6.39)

for 1:1 electrolytes. The more general expression is

P
A =-LZ·D· (6.40)
RT i ' ,
for asymmetrical electrolytes.

TABLE 6.22
Experimental Energies of Activation for Various Transport Processes

ED.
Molten salt EA
kcal mole- 1

LiCI 2.06 8.7

NaCI 2.92 7.3 9.1
NaNO. 3.12 5.0 4
KCl 3.36 7.0 7.8
KNO. 3.15 5.6 3.6
CdCl. 2.20 6.8 4
 IONIC LIQUIDS 557

TABLE 6.23
Test of the Nernst-Einstein Relation for Equivalent Conductivity of Molten
NaCI

Equivalent conductivity

1093 oK 1143 oK 1193 oK 1293 oK

Observed 138 147 155 171

Calculated from
Eq. (6.39) 159 177 198 240

The testing of the Nernst-Einstein relation can be carried out by using

the experimentally determined tracer-diffusion coefficients Di to calculate
the equivalent conductivity A and then comparing this theoretical value
with the experimentally observed A. It is found that the values of A calculat-
ed by Eq. (6.39) are distinctly greater (by,....., 10 to 50%) than the measured
values (see Table 6.23 and Fig. 6.28).

6.4.6b. The Gross View of Deviations from the Nernst-Einstein

Equation. It is important to understand the conditions under which there
are deviations from the Nernst-Einstein relations because such deviations
throw light on what is happening inside ionic liquids. The ultimate aim,
of course, is to acquire a molecular view of the deviations; but, in the first
instance, it is appropriate to define, in phenomenological terms, the general
basis for deviations from the Nernst-Einstein relations. This phenomeno-
logical "explanation" can be derived from the near-equilibrium thermo-
dynamic approach which has been described in Sections 4.5.6 and 4.5.7,
and has been applied to fused salts by Sundheim.
The fluxes (moles flowing per square ceritimeter per second) of the
positive and negative ions in a fused salt can be represented by [ef Eqs.
(4.247) and (4.248)]
(6.41 )

(6.42)
-+ -+
where F+ and F_ are the driving forces on the positive and negative ions,
respectively; L++ and L __ are the straight coefficients governing the inde-
pendent flows of these ions; and L+_ and L_+ are the cross coefficients
558 CHAPTER 6

en
ro
ui
c
.Q
0250
&
'"
~

o'"
"3
~200
.
8
o

.
~

~
'"
on
.a
o
< 150

1100 1250 1300

TOK
Fig. 6.28. Plot to show deviations from the Nernst-
Einstein equation; (0) observed equivalent conductivity
of molten NaCI and (U) calculated from Eq. (6.39).

-+
governing the influence of the driving force F _ on the flow J + of the positive
ions and vice versa.
What is the consequence of setting the cross coefficients equal to zero,
i.e.,
(6.43)

With the use of condition (6.43) in Eqs. (6.41) and (6.42), the result is

(6.44)
and
(6.45)

When L+_ = L_+ = 0, the positive and negative ions drift independently
-+
of each other, i.e., the driving force F_, acting on the negative ions, does
/lot affect the flux J t of positive ions, and vice versa. But it has been shown
that the total driving force on an ionic species is the gradient of electro-
 IONIC LIQUIDS 559

chemical potential [cf Eq. (4.222)]

= _ dll+ _ z+F dcp

dx dx
RT dc+
= - -- -- - z FX (6.46 )
c+ dx +

Hence, by combining Eqs. (6.46) and (6.44), one has

(6.47)

Now, the independent flow of an ionic species has been shown to be

given by the Nernst-Planck flux equation (4.224), i.e.,

(6.48)

or, since [cf Section 4.4.3 and Eqs. (4.149) and (4.169)]

D+ = ( ) kT = (ucom·)+kT (6.49)
U"bs +
z+c o
hence, from (6.49) and (6.48),

(6.50)

By equating the coefficients of dc+/dx and X from Eqs. (6.47) and (6.50)
the net result is
(Uconv)·cc+
(6.51 )
z+F
or
(6.52)

Similarly, it can be shown that

(6.53 )

By combining Eqs. (6.52) and (6.53) and utilizing the relation

i1
(u conv )+ + (uconv )- = F (6.54 )
560 CHAPTER 6

the final result is

(6.55)

p
= RT ~ ZiDi (6.56)
I

which is the Nernst-Einstein relation.

It is clear, therefore, that the Nernst-Einstein relation is obeyed when
L+_ = L_+ = 0, i.e., when the flow of each species is unaffected by the
driving forces on the other species.
Various terms are in use to describe the cross coefficients. For instance,
they have been called "coupling" coefficients or "interaction" coefficients,
though these terms mislead one into thinking that the presence of inter-
actions between particles inevitably leads to deviations from the Nernst-
Einstein relation. It must be emphasized, however, that such coefficients
do not have (and cannot have) literal mechanistic implications for there
may be abundant interactions between particles and yet the flows may be
independent, L+_ = L_+ = 0, i.e., the flows may conform to the Nernst-
Einstein relation. The only significance of the cross coefficients is that,
provided by Eqs. (6.41) and (6.42): L+_, for example, is a measure of the
extent to which the flux of positive ions is affected by the gradient of elec-
trochemical potential of the negative ions.
Since the cross coefficients do not contribute to a mechanistic (hence,
structure-indicating) interpretation of the deviations from the Nernst-
Einstein equation, it may be asked: Why use them? In the subject under
discussion, i.e., deviations from the Nernst-Einstein equation, it will be
shown in the next section that rationalization at a molecular level exists;
hence, phenomenological coefficients are, strictly, unnecessary. Nevertheless,
the nonequilibrium thermodynamic approach has been referred to here
because it represents a convenient shorthand way of expressing the condi-
tions under which the Nernst-Einstein relation is valid.

6.4.6c. Possible Molecular Mechanisms for Nernst-Einstein Devia-

tions. Since phenomenological treatments cannot throw light on the mo-
lecular mechanisms responsible for the deviation from the Nernst-Einstein
equation, one must consider the atomistic origin of the cross coefficient.
In other words, how does the drift of one ionic species induce the drift
of another species?
A possible answer emerges from the fact that the observed conductivity
is always less (Table 6.23) than that calculated from the sum of the diffusion
 IONIC LIQUIDS 561

coefficients, i.e., from the Nernst-Einstein relation [Eq. (6.39)]. Now,

conductive transport depends only on the charged species because it is
only charged particles that respond to an external field. If, therefore, two
species of opposite charge unite, either permanently or temporarily, to give
an uncharged entity, then they will not contribute to the conduction flux
(Fig. 6.29). They will, however, contribute to the diffusion flux. There will
be currentless diffusion, and the conductivity calculated from the sum of
the diffusion coefficients will always exceed the observed value. Currentless
diffusion, therefore, will lead to a deviation from the Nernst-Einstein
relation. Notice that, in currentless diffusion, the diffusion flow of i particles
induces a flow of j particles with which they (or a fraction of them) are
united. Thus, the mechanism of currentless diffusion is in accord with the
thermodynamic version of the deviation from the Nernst-Einstein relation.
What sort of situations lead to currentless diffusion? One situation is
that in which a stable, uncharged entity is formed. Thus, if the formation
of an MX* species occurs in a system consisting of M+, X-, and X* - ions,
then the diffusion flux of X*- consists of two components, the flux of MX*
and that of X*-. But the conduction flux of X* is due to the charged X*-
only. Now, the Nernst-Einstein equation is valid only when the same
charged particles are involved in diffusion and conduction. Hence, there
will be a deviation in the case of the formation of complex ions and stable
ion pairs.
It was suggested by Borucka et al. in 1957 that a permanent association
of positive and negative ions is not a necessary basis for a breakdown of

Ions mlQrate These entities

under electric donot respond
field to electric field

Fig. 6.29. An entity formed by the tem-

porary or permanent association of a pair of
oppositely charged ions is electrically neu-
tral and therefore does not migrate under an
electric field .
 -0-0-
562 CHAPTER 6

0-0-0 0-0-0 0-0-0

--- -0- -
-0-0- -0-0-
0o~ 0-0 0-0-0 0-0 0
-0/ -
0-0-0 0-0-0 0-0-0
_ No+

Fig. 6.30. Schematic diagrams to indicate how diffusive displacement can oc-
cur through a coordinated movement of a pair of ions into a paired vacancy.

the Nernst-Einstein equation. The only requirement is that diffusion should

occur partly through the displacement of entities which have (during jumps)
a zero net charge and thus do not contribute to conduction. The entity
may be, for instance, a pair of oppositely charged ions, in which case the
diffusive displacement occurs by a coordinated movement of such a pair
of ions into a paired vacancy (Fig. 6.30), i.e., a vacancy which is large
enough to accept a positive and a negative ion. The pair of oppositely
charged ions which jumps into a coupled vacancy is neutral as a whole,
and, therefore, such coordinated jumps do not playa part in the conduction
process, which is determined only by the separate and uncoordinated
movements of single ions.
Thus, the experimentally observed diffusive flux of either of the ionic
species is made up of two contributions, the diffusive flux occurring through
the independent jumps of ions and that occurring through paired jumps.
Thus, taking the example of the diffusion ofNa+ in an NaCl melt, one hast

(6.57)

where J Na+ is the experimentally observed flux of Na+ (primes refer here
to experimental quantities), JNa+,ind is the diffusion of Na+ by independent
jumps, and J NaCI is the flux due to coordinated jumps of Na+ and Cl- ions

t The subscript NaCI must not be taken to mean that there are entities in the melt which
might be considered "molecules" of sodium chloride. The NaCl does not refer to
Na+ and CI- ions which are bound together like an ion pair in aqueous solution;
rather, it refers to a pair of Na+ and Cl- ions which undergo a coordinated jump into
a paired vacancy during the short time for which they momentarily exist in contact.
They do not contribute to the conductance because their jumps are directed in not by
the externally applied field but by the distribution of the paired vacancies which
exist before the ions jump as a pair.
 IONIC LIQUIDS 563

into paired vacancies

' - D'
JNa+- dCNa+
(6.58)
Na+~

dc Nrt +
J Na+,ind = DNa+ ,ind -d--x (6.59)
and
(6.60)

Adding Eqs. (6.59) and (6.60), one has

(6.61)

But, from Eqs. (6.57) and (6.58),

(6.62)
Hence,
D Na+ = DNa+,ind + D NaCI (6.63)
Similarly,
Db- = DCI-,ind + D NaCi (6.64)

On adding, it is clear that

(DNa+,ind + DCI-,ind) = (DNa+ + Db-) - 2DNaCI (6.65)

The Na+ and Cl- ions which make coordinated jumps into paired
vacancies, i.e., the NaCl species, contribute to diffusion but not to conduc-
tion since such a coordinated pair is effectively neutral. Hence, the Nernst-
Einstein equation is only applicable to the ions which jump independently,
i.e.,
RT A'
DNa+,ind + DCl-,ind = ZF2 (6.66)

where A' is the experimentally observed equivalent conductivity of molten

NaCl. Making use of Eqs. (6.66) and (6.65), one has

RT A'
ZF2 = (DNa+
' + D 'Cl-) - 2DN",CI (6.67)

or
(6.68)

The first term on the right-hand side corresponds to the value of the
564 CHAPTER 6

equivalent conductivity which would be calculated on the basis of the ex-

perimentally observed diffusion coefficients. Using the symbol A calc for this
calculated value, i.e.,
zp ( , , )
A calc = RT DNa+ +D Cl- (6.69)

one has
(6.70)

which shows that the experimental value of the equivalent conductivity is

always less than that calculated from a Nernst-Einstein equation based on
experimental diffusion coefficients. This is what seems to be observed
(Table 6.23).

6.4.7. Trans,port Numbers in Pure Liquid Electrolytes

6.4.7a. Some Ideas about Transport Numbers in Fused Salts. The
concept and determination of transport numbers in pure liquid electrolytes
is one of the most interesting-and most confusing-aspects of the electro-
chemistry of fused salts.
The concept has been referred to in Section 4.5.2. The transport number
ti of an ionic species i is the quantitative answer to the question: What
fraction of the total current I [= ~ id passing through electrolyte is trans-
ported by the particular ionic species i? In symbols (ef Section 4.5.2):

(6.71 )

In the case of z:z-valent salts, the transport number is simply given by

Ui t
(.=-- (6.72)
! LUi

or, for a pure liquid electrolyte consisting of one cationic and one anionic
species,
and (6.73)

It was seen (Section 4.5.7) that, in aqueous solutions, the solvent could
not be relegated to the status of an unobtrusive background. The solvent
molecules, by entering into the solvation sheaths of ions, participated in

t The coordinate system with which these mobilities are measured is considered later on.
 IONIC LIQUIDS 565

M colhode
I equiy. of M + ions
dissolyed in 10 electrolyte

et positive charlie
lends 10 be produced

Fig. 6.31. Schematic U-tube setup with M electrodes

and MX electrolyte. When 1 F of electricity is passed
through the system, one equivalent of M+ ions is depos-
ited at M cathode and one equivalent of ions is produced
at M anode. Hence, negative charge tends to be produced
near cathode and positive charge near the anode.

their drift. Thus, in addition to the flows of the positive and negative ions,
there was a flux of the solvent.
This complication of solvent flux is absent in pure ionic liquids. There
is, however, a very interesting effect when a current is passed through a
fused salt.
Consider that a fused salt MX is taken as the ionic conductor in a
U-tube and two M electrodes are introduced into the system as shown in
Fig. 6.31. Let the consequences of the passage of one faraday of electricity
be analyzed. Near the cathode, one equivalent of M+ ions will be removed
from the system by deposition on the cathode; and, near the anode, one
equivalent of M+ ions will be "pumped" into the system. Since one equiv-
alent of M + has been added and another equivalent has been removed, the
total quantity of M + ions in the system is unchanged.
Is the system perturbed by the passage of a faraday of charge? Yes,
because, near the cathode, one equivalent of M + ions has been removed,
which has created a local excess of negative charge. This local unbalanee
of electroneutrality creates a local electric field. t A similar argument can
be used for the anode region.

t This unbalance of electroneutrality and creation of field should not be confused with
that arising from the presence of the electrode, which causes an anisotropy in the forces
on the particles in the electrode-electrolyte interphase region. That anisotropy also
produces an unbalance of electroneutrality and an electric double layer (Chapter 7)
with a field across the interface.
566 CHAPTER 6

"' 2:"",,/'"
(0)

Fig. 6.32. The tendency for elecfroneu-

trality to be upset near electrodes is avoided
in one of three ways. For example, near the
anode. where positive charge tends to be
produced, (a) positive ions can migrate away
from the anode, (b) negative ions can mi-
grate toward the anode. and (c) both (a) and
(b) processes can occur to various extents.

How do the ions of the liquid electrolyte respond to this perturbation,

i.e., this creation of local fields? The ions start drifting under the influence
of the fields so that the initial state of electro neutrality and zero field is
restored. How do the positive and negative ions share this responsibility
of moving to annul the unbalance of charges-more anions than cations
near the cathode and vice versa. It is to be noted (Fig. 6.32) that the original
electroneutral situation can be restored (1) by only cations moving in the
anode-to-cathode direction; (2) by only anions moving in the cathode-to-
anode direction; and (3) by both cations and anions moving in opposite
directions to different extents. But these possibilities represent different
values of the transport numbers which are the fractions of the total field-
induced ionic drift arising from the various species.

6.4.7 b. The Measurement of Transport Numbers in Liquid Electro-

lytes. Let t+ and L be the transport numbers of the M+ and X- ions
of the fused salt. The changes of the numbers of equivalents of M+ and X-
near the two electrodes are shown in Table 6.24 based on Fig. 6.31. The
analysis of the changes leads to a most interesting result. The passage of
1 F of charge is equivalent to transferring L eqUIValents of the fused salt
MX from the cathode region to the anode region (Fig. 6.33).
In the case of aqueous solutions, the ever-plentiful solvent could absorb
this L equivalent of MX and register the transfer as a concentration change
 IONIC LIQUIDS 567

TABLE 6.24
Changes in MX at Electrodes in Transport Experiment

Anode compartment, Cathode compartment,

Electrode material
Metal M Metal M

Electrode process per faraday

passed I g eq M+ dissolved I g eq M+ deposited
Move out of compartment eq M+
t+ g L g eq X-
Move into compartment Lg eq X- t+ g eq M+
Net change of M + l - t+ g eq gained = L I - t+ g eq. lost = L
Net change of MX L g eq gained L g eq lost
Mass change of MX in case LMMX g gained, where LMl\1x g lost, where
of molten salt MMX is eq. wt. of MX Ml\1x is eq. wt. of MX
Concentration change in case L/ VA g eq liter- 1 increase, L/ VC g eq Iiter- 1 decrease
of aqueous solution where VA is vol. of anode where Vc is vol. of
compartment (1.) cathode compart. (1.)

of magnitude, c /V equivalents per liter, where V is the volume of the

compartment. But, a pure molten salt has no concentration variable.
Hence, the transfer leads to a mass increase near the anode, as was first
suggested by Schwartz (1912) and proved mathematically by Sundheim
(1956).
In molten salts, therefore, it is the change in mass in a compartment
which reveals transport numbers; in aqueous solutions, it was the change

Net loss in - ( J-.... N •

cothode et goon In onode
region. Leq.of M region. Ceq . of MX

Tronsfer of '- eq.of MX

Fig. 6.33. As a result of the passage of 1 F of

electricity, L equivalents of MX electrolyte are trans-
ferred from the cathode region (where the electrolyte
level falls) to the anode region (where the level rises) .
568 CHAPTER 6

M anode

Gravi lollonol
bock flow

Fig. 6.34. The difference in electrolyte lev-

els produced by the passage of 1 F of elec-
tricity leads to a gravitational flow of the
electrolyte.

in concentration. However, the experiment, unless performed properly,

provides information only on the change in mass, not on the transport
property.
What future has this mass increase? Left alone, the mass increase is
short lived and the transport experiment fails. This is because gravitational
flow of the molten salt from the anode to cathode tends to equalize the
amounts of MX in the two tubes (Fig. 6.34). The liquid levels must equalize.
The first step, therefore, in determining transport numbers in pure
ionic liquids is to prevent the gravity flow from masking the transfer of
electrolyte. If the hydrodynamic backflow cannot be prevented, it must at
least be taken into account. The general procedure is to minimize the
gravitational flow by interposing a membrane between anode and cathode
(Fig. 6.35). However, serious objections have been raised to the use of a
membrane, owing to hydrodynamic interferences between this and the
moving liquid.
It is also possible to open out the U tube and make the whole liquid
"lie down" so that the movement of the fused salt occurs at one level and
not against gravity. The amount of salt entering the anode region is then
indicated, for example, by a sliver of molten metal pushed along by the
movement of the salt in the capillary (Fig. 6.36). This method is also sub-
ject to difficulties, for the movement of salts in capillary tubes is often not
smooth but jerky.
This experiment demonstrates directly that, when electricity is passed
through a fused salt, there is a movement of the salt as a whole. In other
words, the mass center of the liquid electrolyte moves. Now, the ions also
are drifting with certain mobilities, i.e., velocities under unit field. But
 IONIC LIQUIDS 569

Membrane to stop
gravi ty flow

Fig. 6.35. The gravitational flow can be

minimized by interposing a membrane be-
tween the anode and the cathode.

velocities with respect to what? One must define a coordinate system, or

frame of reference, in relation to which the velocities (distances traversed
in unit time) are reckoned. Though the laws of physics are independent of
the choice of the coordinate system-the principle of relativity-all coor-
dinate systems are not equally convenient. In fused salt it has been found
convenient to use the mass center of the moving liquid electrolyte as the
frame of reference, even though this choice, while providing a simple basis
for computations, suffers from difficulties.
Even the elementary presentation just given makes it clear that trans-
port-number measurements in fused salts are based on the transfer of the
fused salt from the anode to the cathode compartments. The quantities
measured are weight changes, the motion of indicator bubbles, the volume
changes, etc. Some basic experimental setups shown in Fig. 6.37 include the

Membrane
Electrolyte

Capi llary

Fig. 6.36. A simple arrangement by which gravitational flow is avoided by the

displacement of the electrolyte from the cathode to the anode region occurs at one
level. The change in position of the melt in the capillary indicates the amount of elec-
trolyte displaced.
570 CHAPTER 6

(0)

(d)

Aw

Fig. 6.37. Schematic diagrams of methods of determining transport numbers:

(a) Measure velocity of the bubble; (b) measure transfer of the tracer; (c) measure
the potential difference due to pressure difference; (d) measure the change in weight;
(e) measure the transport of liquid metal electrodes; (f) measure the steady-state
level; (g) measure the change in weight; (h) measure the moving boundary.

apparatus of Duke and Laity, Bloom, and other pioneers in this field.
The migration of the electrolyte from the anode to the cathode com-
partments can also be followed by using radioactive tracers and tracking
their drift. Since isotopic analysis methods are sensitive to trace concen-
trations, there is no need to wait for the electrolyte migration to be large
enough for visual detection.
 IONIC LIQUIDS 571

TABLE 6.25
Some Transport Number Results for Fused Salts

Molten salt Transport number of cation

LiNO a 0.84
NaNO a 0.71
AgNO a 0.72
NaCI 0.87
KCl 0.77
AgCl 0.54

The results of some transport-number measurements are given in

Table 6.25.

6.4.7 c. A Radiotracer Method of Calculating Transport Numbers in

Molten Salts. In the discussion of the applicability of the Nernst-Einstein
equation to fused salts, it was pointed out that the deviations could be
ascribed to the paired jump of ions resulting in a currentless diffusion.
With fused NaCl as an example, it has been shown that there is a
simple relation between the experimentally determined equivalent conduc-
tivity A' and the experimental diffusion coefficients of the Na+ and Cl-
as indicated by radio tracer Na+ and CI-. The relation is

(6.68)

From this expression, it is clear that one can determine D NaCI ' Knowing
D NaC1 , one can obtain the diffusion coefficients DNa+,ind and DCl-,ind, of
the independently jumping Na+ and Cl- ions from the relations (6.63)
and (6.64), i.e.,
DNa+,ind = DNa+ - DNaCI (6.74)
and
DC1-,ind = DC1- - DNaCI (6.75)

Further, by using the Einstein relation [Eq. (4.169)] and the relation between
absolute and conventional mobilities, one has
( ) ( ) Na+ DNa+.ind
Uconv Na+ = ZNa+eO iiabs = zNa+eO kT
z+F
= RT DNa+,ind (6.76)
572 CHAPTER 6

TABLE 6.26
Comparison of the Transport Number of Na+ in Molten NaCI Calculated
from Equation (6.78) with Measured Values

Transport number of Na+ in NaCl

Calculated from Eq. (6.78) 0.71

0.62
Measured values {
0.87

and similarly,
(6.77)

With these values of mobilities, the transport numbers can easily be calcu-
lated from the standard formulas

and (6.73)

which, in the case of NaCl (z+ = L = 1), reduces to

DNa+ ind D Cl- ind
= ' and t Cl - = ' (6.78)
tNa+
DNa+,ind + DCl-,ind DNa+,ind + DCl-,ind
The comparison between transport numbers calculated in this way and
those obtained by some of the experimental methods used is shown in
Table 6.26.
6.4.7d. A Stokes'-Law Approach to a Rough Estimate of Transport
Numbers. When an ion (of charge ZieO and radius ri) drifts in an elec-
trolyte of viscosity 'YJ, the Stokes mobility is given by [ef Eq. (4.180)]

Uconv = 1800n'YJr;

(6.79)

where
k = eo (6.80)
1800n'YJ

Upon introducing Stokes' mobilities into the expression (6.73) for transport
 IONIC LIQUIDS 573

TABLE 6.27
Testing of Stokes' Law on Transport Numbers

Transport number of Cl- in PbCl.

Calculated from Eq. (6.81) 0.25

Measured 0.37

numbers, the result is

(6.81)

Or, for I: I electrolytes,

(6.82)

For the radii of the ions, the crystallographic values can be used and,
thus, the transport numbers calculated approximately. Table 6.27 shows
the comparison between observed and estimated values. The reasonable
agreement indicates that this approach is useful to provide a first rough
approximation of the transport numbers (for it has been seen in Section 6.4
that Stokes' law has only partial applicability to molten electrolytes).

Further Reading
1. W. Klemm and W. Biltz, z. Anorg. AI/gem. Chem., 152: 255 and 267 (1926).
2. F. R. Duke and R. Laity, J. Am. Chem. Soc., 76: 4046 (1954); J. Phys. Chem.,
59: 549 (1955).
3. H. Bloom and N. J. Doull, J. Phys. Chem., 60: 620 (1956).
4. B. R. Sundheim, J. Phys. Chem., 60: 1381 (1956); 61: 485 (1957).
5. A. Z. Borucka and J. A. Kitchener, Proc. Roy. Soc. (London), A241: 554
(1957).
6. G. J. Janz, C. Solomons, and H. J. Gardner, Chem. Rev., 58, 461 (1958).
7. R. W. Laity, Ann. N.Y. A cad. Sci., 79: 997 (1960).
8. T. B. Reddy, Electrochem. Technol., 1: 325 (1963).
9. A. Klemm, in M. Blander, ed., Molten Salt Chemistry, lnterscience Publishers,
Inc., New York, 1964.
10. B. R. Sundheim, Chapter 3 in: B. R. Sundheim, ed., Fused Salts, McGraw-
Hill Book Company, New York, 1964.
11. W. K. Behl and J. J. Egan, J. Phys. Chem., 71: 1764 (1967).
574 CHAPTER 6

6.5. THE ATOMISTIC VIEW OF TRANSPORT PROCESSES IN

SIMPLE IONIC LIQUIDS

6.5.1. Holes and Transport Processes

Some facts about transport processes in molten salts have been mention-
ed (Section 6.4). The degree to which the hole model (Section 6.2) serves to
give an approximate rational basis for these must now be examined. How-
ever, at first, it is necessary to cast the hole model into a form suitable for
the prediction of transport properties.
The starting point for the derivation of expressions for transport prop-
erties on the basis of the hole model is the molecular-kinetic expression
(Appendix 6.3) for the viscosity 'Y) of a fluid, i.e.,

'Y) = 2nm<w»). (6.83)

where nand m are the number per unit volume and the mass of the particles
of the fluid, <w) is the mean velocity of the particles, and A is their mean
free path.
The quantity). is linked to the picture of the phenomenon of viscosity
in the kinetic theory of gases. According to this picture (Fig. 6.38), a fluid
in motion is considered to consist of layers (of fluid) lying parallel to the
direction of flow. (The slipping and sliding of these layers against each
other provides the macroscopic explanation of viscosity). When particles
jump between neighboring layers, there is momentum transfer between
these layers, the cause of viscous drag (Fig. 6.39). In this picture, <w) is
the component of the average velocity of the particles in a direction normal
to the layers.
Irrespective of whether the gas as a whole is in motion or not, the
particles constituting the gas are executing random motion. The particles
of a flowing gas have a drift superimposed upon this random walk. It is
by the random walk of the particles from one layer to another that the
momentum transfer between layers is carried on. This momentum transfer
manifests itself as the viscosity of the fluid.

..in Fluid
motion = _ - - - - - - } Moving layers
of fluid

Fig; 6.38. A fluid in motion is considered equivalent to moving

layers of fluid, the layers lying parallel to the flow direction.
 ·. ? r
IONIC LIQUIDS 575

~~-____ Moving layers

~ ,ffl,;'

~particle$ jumping
between layers

Fig. 6.39. Viscous forces are considered

to arise from the momentum transferred
between moving fluid layers when parti-
cles jump from one layer to another.

Holes also move. As argued earlier (cf. Section 6.3.1), anything which
moves at finite velocities must have an inertial resistance to motion, i.e.,
a mass (see Appendix 6.1). Holes, therefore, have masses and momenta.
According to the hole theory, holes play the role in pure ionic liquids
which.molecules play in gases. Thus, it is considered that the random walk
of holes between adjacent layers results in momentum transfer and, there-
fore, viscous drag in a moving fused salt (Fig. 6.40). The expression for the
viscosity of an ionic liquid, on the basis of this model, is

(6.84)

where nh and mh are the number per unit volume and the apparent mass
for translational motion of the holes.
Now, the velocity component <wh) is given by the ratio of Ah, the
mean distance between collisions (i.e., the mean free path), to r, the mean
time between collisions,
Ah
<Wh) = - or (6.85)
r

r~
'- L ,.
....-.-----11,_1
C Hales jumping
between fluid
layers

Fig. 6.40. According to the hole model,

viscous drag arises from the momentum
transferred between moving fluid layers
when holes jump from one layer to another.
576 CHAPTER 6

Hence, the viscosity can be written as follows

(6.86)

The theorem of the equipartition of energy can now be applied to the

one-dimensional motion referred to by <Wh>

(6.87)

and, using this equation, one has

(6.88)

6.5.2. What Is the Mean Lifetime of Holes in Fused Salts?

The parameter T now invites consideration. In the gas phase, T is

simply the mean time between collisions. What is the significance of T in
an ionic liquid?
In a liquid, on the present model, T would be the mean lifetime of a
hole, i.e., the average time between creation and destruction of a hole
through thermal fluctuation. To calculate this, one may use the formula
for the number of particles escaping from the surface of a body per unit
time per unit area into empty space, i.e.,

a = c (~)!e-A/RT (6.89)
2nm

where c is the number of particles per unit volume, m is the mass of a

particle, and A is the work necessary to remove a mole of particles from the
surface to an infinite distance. Here, A is also the work necessary for a
mole of particles lying on the surface of the hole to be released into its
interior.
In a time t, 4n<rh> at particles escape from the exterior into a spherical
hole of radius <rh>' The hole will be filled by these particles if this number
is equal to the number of particles in a sphere of radius <rh>3c, !n<rh>'
Then, the time for destruction of the hole is

(6.90)

Obviously, this is also the time for hole formation, and the lifetime is
consequently
(6.91)
 IONIC LIQUIDS 577

This is the mean lifetime of a hole. It is seen, therefore, that the hole
theory represents a Swiss-cheese sort of model of a liquid, with flick-
ering holes.
But now, before one leaves expression (6.91), it is well to note the
simplicity with which A has been treated. It is the heat term associated
with getting a hole "unmade," with collapsing the hole, the negative of
this work of forming the hole. But it has not yet been said how this will
be calculated, and what terms go into this. Such a calculation will be one
test which will be made of the hole theory here.

6.5.3. Expression for Viscosity in Terms of Holes

By inserting the expression [Eq. (6.91)] for the mean lifetime of a hole
into Eq. (6.88), one obtains the hole-theory expression for viscosity. Thus,

or
(6.92)

There are two quantities on the right-hand side of Eq. (6.92) which
need discussion. They are nh, the number of holes per unit volume of the
liquid, and A, which occurs in the Boltzmann factor exp( - AIRT), for the
probability of a successful filling of a hole.

6.5.4. The Diffusion Coefficient from the Hole Model

Now that the viscous flow properties of an ionic liquid have been
discussed, the next task is to derive an expression for the diffusion
coefficient.
The elementary act of a transport process consists of hole formation
followed by a particle jumping into the hole. The focus in this eiementary
act has hitherto been the center of the hole.
But what is the situation at the original site of the jumping ion? Alter-
natively stated: What has happened, as a consequence of the jump process,
at the point where the ion was before it jumped? At this prejump site,
there has been precisely that moving away of particles from a point which
corresponds to hole formation (Fig. 6.41).
Thus, when a particle jumps, it leaves behind a hole. So then, instead
of saying that a transport process occurs by particles hopping along, one
could equally well say that the transport processes occur by holes moving.
578 CHAPTER 6

~ar'ic/e jump

o:FX~'CD
uYif:D
\ Hale formed

Fig. 6.41. When a particle jumps into a

hole, there is the formation of a hole at the
prejump site of the particle.

The concept is commonplace in semiconductor theory, where the mov-

ement of electrons in the conduction band is parallel by a movement
of so-called "holes" in the valence band. It has already been assumed at
the commencement of the viscosity treatment (Section 6.5.1) that the
viscous flow of fused salts can be discussed in terms of the momentum
transferred between liquid layers by moving holes.
The upshot of this analysis is that, when diffusion of particles occurs,
there is a corresponding diffusion of holes. Hence, instead of treating
ionic diffusion, one can consider hole diffusion and write Stoke's law
[Eq. (4.176)] for the diffusion coefficient of holes

D= kT (6.93)
6n<rh)'Yj

But the hole-theory expression for viscosity is known. It is Eq. (6.92).

Let this be introduced into Stoke's law [Eq. (6.93)]. The result is, using
Eq. (6.29)
D = kT (2nmkT)-1!2 e- AIRT
4n<rh)2 nh

I L (2nmkT)-1! 2e-AIRT
4n(0.51)2 nh

= 0.31 L (2nmkT)-ie-AlkT (6.94)

nh (

The number of holes per unit volume can be expressed in terms of the
known volume expansion of the liquid at a temperature T over that of the
solid at the same T, L1 VT , divided by the mean hole volume, i.e. L1 VT/<Vh),
and reduced to the number per cubic centimeter, by dividing L1 VT/Vh by the
molar volume of the liquid at T, VT . Hence,

D = O.31y VhVT (2nmkT)-!e- A1R'l' (6.95)

L1 VT
 IONIC LIQUIDS 579

TABLE 6.28
Energies of Activation of Cation and Anion for Self-Diffusion in Simple
Molten Electrolytes

ED, kcal mole- 1

Molten salt
Cation Anion

NaCl 7.1 ± 0.3 7.4 ± 0.8

KCl 6.9 ± 0.5 7.1 ± 0.5
CaCl. 6.1 ± 0.7 8.9 ± 0.9
SrCl. 5.4 ± 0.9 6.9 ± 0.7
BaCl. 9.0 ± 1.2 9.5 ± 1.0
CdCl. 6.8 ± 0.6 6.8 ± 1.0

Utilizing the value of the hole volume derived by substituting for <rh)
from Eq. (6.29) in Vh = in<rh)3:

D = O.17kTVT e-A1RT (6.96)

(2nm)iLJ VTyf

The identity of the rate of hole diffusion and ionic diffusion is recalled.
Thus, the final expression for the diffusion coefficient of ions in a fused
salt is the same as that for holes, i.e., Eq. (6.96).
The first point to note about this expression, apart from the fact that
it is of the form of the experimentally observed variation of D with temper-
ature, i.e., D = Doe-EDIRT, is that the energy of activation for self-diffusion
of cations and anions should be the same [cf. Eq. (6.33)]. This is essentially
what is observed (Table 6.28).
Before going on to discuss the term A in terms of the structure of the
salt, one must note what is meant by the term "Arrhenius activation energy."
This term arose from the work of the early gas kineticists, who wrote
equations of the type
Rate = Ze- EiRT

and considered Z to have a negligible temperature dependence. Such an

assumption would, of course, give
8 In D
E = -R 8(lIT) (6.97)

and often it is simply this coefficient which is identified with the "energy
580 CHAPTER 6

of activation." Therefore, if one wants to calculate what a theory gives

for this, one has to take into account whatever temperature dependence
is possessed by the pre-exponential factor in the theory. Thus, if one knew
(as one hopes to know later on in this chapter) a theoretical expression
for A, of (6.96), one would have to calculate

a In D
-R a(1fT)

from (6.96) (including the effect of the temperature dependence of the

pre-exponential) and compare its theoretical value with the experimental
value of E calculated from Eq. (6.97). From (6.96),

a In D _ _ 2( I 1 dy _ I
_
R a(1fT) - A+RT RT iJVT -n
diJ V
T
+ 2Y dT VT
dVT
(iT
)

(6.98)
In the following sections, a value of A will be calculated. Utilized in the
right-hand side of (6.98), it gives there the theoretical prediction of the ex-
perimental energy of activation, i.e. the left-hand side of Eq. (6.98) (cf 6.97).

6.5.5. A Critical Test of a Model for Ionic Liquids is a Ration-

alization of the Heat of Activation of 3.7RTm for Trans-
port Processes

It has been pointed out (Sections 6.4.2b and 6.4.3) that the heats of
activation for viscous flow, and for self-diffusion, are given by the empirical
generalization E" = ED = 3.7 RTm , where T m is the melting point in degrees
Kelvin. Some of the data which support this statement are plotted in Figs.
6.23 and 6.25. The empirical law seems to be applicable for all nonassociated
liquids, i.e., not only ionic liquids. An empirical generalization which en-
compasses the rare gases, organic liquids, the molten salts, and the molten
metals is a challenge for theories of the liquid state, and hence for funda-
mental electrochemists interested in pure liquid electrolytes.
Several approaches to the theory of liquids can be distinguished (cf
Section 6.2.4). One may begin by expressing the properties of the liquid
in terms of a distribution function (an expression which indicates the
probability of finding particles at a distance r from a central reference ion).
Ideally, one should be able to calculate the distribution function itself
from a knowledge of the intermolecular forces between the particles.
However, this is in fact usually too difficult a task, and one falls back upon
an experimental determination of this function (X rays), or, to making
 IONIC LIQUIDS 581

educated guesses about the scenarios inside the liquid, i.e., one agrees
to assume temporarily some model of the structure, and then goes to
develop the mathematical consequences of the assumption. The results
of predictions from such developments for alternative models (cf Sections
6.2.2 and 6.2.3) are compared with experimental data and then one may
decide upon the most consistent model and use it as a working hypothesis
which gives a rough idea of the structure of the liquid until absolute calcula-
tions of this are possible (cf Section 6.2.4).
Tests of models have been referred to in the above-named sections.
In general, the present comparative status of the various viewpoints is that
some of them do give rise to reasonably close calculations of equilibrium
properties, e.g., compressibility and surface tension. What it has not been
possible to do up to recent times, however, is validly to rationalize
the data of irreversible phenomena, e.g., conductance, viscous flow, and
self-diffusion. For this reason, the discovery in 1963-1965 of the 3.7RTm
laws (Nanis, Richards) is of particular significance: it puts a simple, clear,
and very challenging target for testing models of liquids.

6.5.6. An Attempt to Rationalize ED = E,} = 3.7RTm

In order to attempt to find what the hole theory predicts for the heat of
activation in viscous flow, it is necessary to attempt to utilize this theory to
calculate the term A in, e.g., Eq. (6.96). The meaning of A has already
been defined in Section 6.5.2. It is the work done in transferring a mole of
particles from the surroundings of a hole into its interior.
An assumption will now be added to the model. It is that, near or at
the melting point, a hole is annihilated by the "evaporation" into it of one
particle, i.e., one particle just fills it. There is no violation of physical sense
in this assertion, for use of (6.29) shows that the size of the holes predicted
by the hole theory is near to that of the ions which are supposed, in the
model, to jump into them. Correspondingly, the work done to annihilate
a hole is numerically equal to the work done in forming a surface of radius
rT, namely 4n<rT )2 y , where y is the surface tension. Hence, if nT particles
must jump into a hole to annihilate it at temperature T

(6.99)

where AT is the term A for the temperature T. Hence, with the assumption
made (nm,p. = 1),
(6.100)
582 CHAPTER 6

What of nT at T's other than the melting point, at which temperature

it has been assumed to be unity? From (6.29), the hole volume (and hence
its surface area) should increase as T increases (and y decreases, as it does
with increase of T). Of course, ions surround the hole, and it seems reason-
able to suppose that, as the hole volume increases, the number of ions which
surround the hole increases, and the number which is needed to fill it
(thus causing the work 4.1l(rT )2 yT ) increases.
Let it be assumed that the difference (Ll VT ) between the volume of
the liquid salt and that of the corresponding solid is due only to holes. Then,
the number of holes, per mole of salt, is, at temperature T,

(6.101)

where (Vh.T) is obtained from (6.29).

Thus, the number of ions per hole at T is

(6.102)

Hence,
Ion per hole at T Ll Vm.p. (Vh.T)
(6.103)
Ions per hole at m.p. Ll VT (Vh.m.p)

One assumes that the number of ions which will be needed to fill the
hole would be proportional to the number of ions per hole. Thus,

nT _ Ll Vm.p,vh.T (6.104)
n m .p. - Ll VTVh.m.p.
But
(6.105)

(6.106)

with nm .p . = 1.
Thus [and with (6.29)]

(6.107)

One can obtain numerical values for the term on the right. It has
been calculated from experimental data for some fourteen simple molten
salts, and, if one restricts the range of experimental data used to about
 IONIC LIQUIDS 583

200°C above the melting point, it is found that

Tin. p . (~)i L1 VT .J'... 1 (6.108)

TTl Ym.p L1 Vm .p . -

Under such circumstances from (6.107) and (6.108)

(6.109)
But
(6.110)
From (6.29)
A m .p . = 4n(0.51)2kTm .p .
= 3.3kTm .p . per ion (6.111)

= 3.3RTm.p. per gram lOn (6.111a)

But, from (6.109),
AT = A m .p . = 3.3RTm.p. (6.112)

This result may be compared with the empirical heat of activation

by substituting it in (6.98). Thus, the terms on the left of (6.98), the experi-

TABLE 6.29
Comparison of Abilities of Three Models of Ionic Liquids to Predict Data

Quantity Hole Vacancy Liquid free volume

Molar vol. change on Good Not yet calc. Good

fusion
Molar entropy change Fair Very good Theory not yet
on fusion formulated
Compressibility Very good Not yet calc. Not yet calc.
Expansivity Very good
Do, ED Do good, Do good, Satisfactory, but
ED = 3.74RTm x ED qualitative predicts curvature
1
[1 + ~: (:~)] of logD vs-
T
excellent
E~ Excellent Qualitative
 584 CHAPTER 6

mental values, are found to be about 3.7 RTm.p.' Using A = 3.3RTm.p.'

for which the derivation (Emi) has been given here, one obtains agreement
between observed and calculated values of the heat of activation to within
about 5-10%. (The correction terms in RT and RT2 of (6.98) affect the
situation little because they work in opposite directions.)
The theory of holes is thus able to give some account of the heat
of activation in transport in ionic liquids. Nor is the major assumption of
the derivation (that only one ion suffices to fill a hole at T = Tm .p ., but a
larger number at higher T) a difficulty, for (so long as it is near to one) it
is the change of this number with temperature (rather than its absolute
value) which is of importance.
A brief comparison of three different important models of ionic liq-
uids is given in Table 6.29.

6.5.7. The Hole Model, the Most Consistent Present Model

for Liquid Electrolytes

The Swiss-cheese model approach is consistent with experimental

parameters and structural data (e.g., X-ray experiments, which show that
the distance apart of ions remains constant or decreases on melting) which
is better than that of other models for liquid electrolytes at this time.
The ability to reproduce experimental data numerically is well shown
in Table 6.30, which gives a comparison of experimental compressibilities

2.5

2.0

, 1.5
c:
0

E
1.0
..•
;,-

0
0.5

0 120
1014 uh (ml. hole-')

Fig. 6.42. Plot of the free volume per ion against the average hole volume; (.) LiCI,
(.6.) NaCI, (e) KCI, (.) CsCI, (+) NaBr, (V") KBr, «()} CsBr, (D) Nal, (O) KI.
 IONIC LIQUIDS 585

TABLE 6.30
Comparison of Calculated Isothermal Compressibilities and Expansivities
with the Experimental Values

Temp. 1012/1, calc, 1012/1, obs, 104a, calc, 104a, obs,

Salt °C-I °C-I
°C cm'dyn- l cm'dyn- l

LiCl 614 20.8 19.4

800 30.5 24.7 4.0 3.0
900 39.1 28.6
1000 47,9 33.0 4.0 3.3

NaCl 800 27.8 28.7 3.3 3.6

900 36,9 22.8
1000 49,6 40.0 3.6 3.9

KCI 772 27.6 36.2

800 30.2 38.4 2.8 3.9
900 42.3 45.7
1000 56.7 54.7 3.2 4.2

CsCl 642 20.2 38.0

800 39.0 51.8 4.2 4.1
900 57.0 62.6
1000 72.5 76.3 4.4 4.5

CdCl 2 600 31.0 29.8 2.7 2.4

700 37.3 33.1 2.7 2.4
800 46.9 36.9 2.9 2.5

NaBr 747 32.8 31.6

800 40.1 33.6 3.2 3.1
900 47.4 38.6
1000 59.5 44.9 4.0 3.4

KBr 735 28.0 39.8 2.5 3.7

800 34.5 43.8 2.6 3.8
900 51.3 52.1
1000 69.9 62.1 3.0 4.1

CsBr 636 56.3 49.1 4.0 4.4

800 76.0 67.1 4.2 4.6
900 97.5 82.7
1000 130.5 103.1 4.4 5.1
586 CHAPTER 6

with the values which the hole theory yields. An interesting aspect of the
evidence for this model is the relation of the free volume (ef Fig. 6.13)
to the volume of the expansion of melting. The free volume, in the sense
referred to here, is the space which is free to each atom on the average.
This relation is shown in Fig. 6.42. The continuous increase of the free
volume with the hole volume would be consistent with a model in which,
for approximately every fifth atom, there is a hole, so that, when a vibrat-
ing atom comes into contact with this space, its free volume is increased.
The free volume is thus related to the hole volume.
One must not give the reader the impression that all is calm in the
field of molten electrolytes. Firstly, the model which gives the greatest
degree of agreement with experiment is a very crude one indeed, and suffers
a fairly large objection because it is, in a sense, not a proper model and
attempts to deal in a curious-if perhaps ingenious-way by analogy with
the bubble in a near-boiling liquid. One might at first not take it seriously,
but one's opinion may change when one sees the ability of the model to
predict experimental data, without the calibration of any constants, and
in such superiority (with respect to transport phenomena) to other present
models, which do not yet yield an attempted theoretical value of the heat
of activation for transport, or of the volume change on melting.
Further consideration shows that there is a reasonable likelihood that
a liquid molten salt is really somewhat as the hole theory indicates. The
ions tend to cling together in clusters, and between these are many gaps
and cavities of varying sizes, in rapid change. It is difficult to avoid such
a model if one is to attain consistency with the considerable increase of
volume on melting and the fact that the internuclear distance decreases or
stays the same. No type of model is consistent with these facts except a
hole model, and, after that, it is largely a matter of how to describe it in
terms of physical chemistry. Of course, a much better mathematical treat-
ment than that above must· be given, but there are difficulties. In some
respects, the model bears resemblance to that containing molecular clusters
in gases.
In this chapter, an attempt has been made to give some account of the
structure of liquid electrolytes. One does not imply, necessarily, that the
hole theory may be invoked as a helpful description of all nonassociated
liquids. In ionic liquids, the lattice cement is stickier and hence there is
more tendency to cluster than, say, in liquid argon where the free space
demanded by expansion may be made up by increasing the gap between
the atoms.
 IONIC LIQUIDS 587

Further Reading

1. N. E. Richards, Proc. Roy. Soc. (London), A241: 44 (1957).

2. H. Eyring, J. Chem. Phys., 4: 249 (1937); Proc. Nat. A cad. Sci., 44: 683
(1958); 46: 333 (1960).
3. H. Reiss, H. L. Frisch, E. Helfand, and J. L. Lebowitz, J. Chem. Phys., 32:
119 (1960).
4. G. W. Hooper, Discussions Faraday Soc., 32: 318 (1962).
5. L. Nanis, J. Phys. Chem., 67: 2865 (1963).
6. F. H. Stillinger, Chapter 1 in: M. Blander, ed., Molten Salt Chemistry, Inter-
science Publishers, Inc., 1964.
7. H. Bloom, Chapter 1 in: B. R. Sundheim, ed., Fused Salts, McGraw-Hill
Book Company, New York, 1964.
8. S. R. Richards, J. Phys. Chem., 69: 1627 (1965).
9. J. D. Bernal, Discussions Faraday Soc., 43: 60 (1967).
10. K. D. Luks and H. T. David, Ind. Eng. Chem. Fundamentals, 6: 194 (1967).
11. R. Vi1cu and C. Misdolea, J. Chem. Phys., 46: 906 (1967).
12. H. T. Davis and J. McDonald, J. Chem. Phys., 48: 1644 (1968).

6.6. MIXTURES OF SIMPLE IONIC LIQUIDS-COMPLEX FOR-

MATION

6.6.1. Mixtures of Simple Ionic Liquids May Not Behave Ideally

A measure of understanding has been gained of the structure and

transport properties of simple ionic liquids. In practice, however, mixtures
of simple liquid electrolytes are more important. One reason for their impor-
tance is that mixtures have lower melting points. But what happens when
two ionic liquids, for example, CdCl 2 and KCI, are mixed together?
Consider, for instance, the electrical conductance of fused CdCl 2 and
KCl mixtures. If the equivalent conductivity of the mixtures (at a fixed
temperature) were given by a simple additivity relation, then a linear
variation of equivalent conductivity with mole fraction of KCl should be
observed (dotted line in Fig. 6.43). The straight line should run from the
equivalent conductivity of pure liquid CdCl 2 at a particular temperature
to that of pure liquid KCl at the same temperature. Some binary mixtures
of single ionic liquids do indeed exhibit the simple additivity implied by
the dotted line of Fig. 6.43.
There are, however, other systems in which deviations occur from a
simple additive law for conductance. The system of CdCl z and KCl is a
case in point (full line in Fig. 6.43). A minimum in the conductivity curve
is observed. What is the significance of this minimum?
588 CHAPTER 6

9·0r---------------------.

8'0

7·0

6·0

0·6 0·8 1'0

Mole fraction of KCI
Fig. 6.43. The variation of the observed equivalent conductivity
of CdCI.-KCI mixtures as a function of the mole fraction of KC!.
The dotted line corresponds to the variation which would be given
by the additivity behavior.

6.6.2. Interactions Lead to Nonideal Behavior

The situation is reminiscent of some happenings in aqueous solutions.

At infinite dilution of, say, a solution of KCI, the properties (e.g., equivalent
conductivity) due to the K+ and Cl- ions are additive (see Section 4.3.10).
This is ideal behavior (see Section 3.4.1). With increasing concentration,
however, there is a departure from ideality: the equivalent conductivities
are not simply additive.
The non ideality in aqueous solutions was ascribed to interactions be-
tween K + and CI- ions, and the ion-ion interaction theories were evolved
for aqueous solutions. Thus, a whole gamut of interactions was considered.
A pair of ions could experience long-distance coulombic forces. There could
be short-range forces between them. The electrostatic attraction between
a pair of oppositely charged ions could overwhelm thermal jostling and
result in the formation of ion pairs. And, finally, the ions could undergo
chemical binding and be joined in complex ion formation.
One can resort to similar explanations of departures from ideality in
mixtures of simple ionic liquids. But there are specific differences between
the situation in fused salts and that in aqueous solutions. In pure liquid
KCI, there is no concentration variable, and therefore, fused KCI has a
single value of equivalent conductivity (at a particular temperature). The
mean distance between the K+ and Cl- ions cannot be altered, as it can be
 IONIC LIQUIDS 589

in aqueous solutions, by interposing varying amounts of solvent because

there is no solvent. Hence, the equivalent conductivity of pure liquid KCI
embodies the effects of all the possible interactions for the temperature
concerned. The thermodynamics of mixtures of molten salts has been in-
tensively studied by Kleppa and by Bloom.
The interactions which one proposes to account for the deviations
from ideality in mixtures of ionic liquids are interactions between the ions
of one component of the mixture considered as a solvent and the ions of
the other component which is added. In the case of KCI added to pure
CdCI 2 , one can consider, for example, the interactions between Cd++ ions
and Cl- ions.

6.6.3. Can One Meaningfully Refer to Complex Ions in Fused

Salts?

It is intended here to discuss nonideality arIsmg from complex ion

formation. In the case of mixtures of pure liquid electrolytes, however, the
idea of complex ion formation raises some conceptual problems.
Consider complex ion formation in the CdCI 2 -KCl system, and let it
be assumed for the moment that a CdCl 3 - complex ion is formed. If such
complex ions were formed in an aqueous solution of CdCl 2 and KCl, they
would exist as little islands separated from other ions by large expanses of
water. In fused salts, there are no oceans of solvent separating the ions.
Thus, a Cd++ ion would constantly be coming into contact, on all sides,
with chloride ions, and yet one singles out three of these Cl- ions and says
that they are part of (or belong to) a CdCl 3 - complex ion (Fig. 6.44). It
appears that, in the absence of the separateness possible in aqueous solutions,
the concept of complex ions in molten salts is suspect. As will be argued
below, however, what is dubious turns out to be not the concept but the

(0) (b)

~-..
t
...-
+/
~
~
j

Shaded waler molecules Shaded ligonds

accompany randam- accompany
walking ion random- walking ion

Fig. 6.44. The similarity between (a) a solvated

ion in an aqueous solution and (b) a complexed
ion in a mixture of ionic liquids.
590 CHAPTER 6

TABLE 6.31
Lifetime of Complex Ion in Molten Salts

Complex ion Molten solvent .complex ion, sec

CdBr+ KN0 3-NaN0 3 0.3, at 263°C

comparison of complex formation in fused salts with complex formation

in aqueous solutions.
It is more fruitful to compare complex formation in ionic liquids with
the phenomenon of hydration of ions in aqueous solution (Chapter 2). It
will be recalled that, though an ion was seen as constantly nudged by the
water molecules of the surrounding medium, some of the water molecules
were considered part of the primary hydration sheath of the ion. The
criterion by which the two kinds of water were distinguished was that all
those water molecules which surrendered their translational degrees of
freedom and participated in the random walk of the ion constituted its
solvation sheath. Thus, a solvated ion is an entity in which a certain number
(the primary solvation number) of solvent molecules participate in the
random walk of the ion.
Similarly, a complexed ion is an entity in which a certain number of
ligand ions (e.g., three chloride ions in a CdCI 3 - complex) participate in
the random walk of the ion (i.e., the Cd++ ion in the CdCl3 - complex).
The other Cl- ions only undergo promiscuous contacts with the Cd++ ion
of the complex, not long-term affairs. The implication is that a complex ion
(i.e., the ion and its ligands) is an entity with a lifetime that is at least sev-
eral orders of magnitude longer than the time required for a single vibration.
Experiments on the lifetimes of complex ions confirm this view (Table 6.31).
From the standpoint of this comparison (Fig. 6.44), it is seen that the
concept of a complex ion in a molten salt is at least as tenable as that of
a hydrated ion in aqueous solutions. But what experimental evidence exists
for complex ions in fused salt mixtures? To answer this question, one must
discuss some results of investigating the structure of mixtures of simple
ionic liquids.

6.6.4. Raman Spectra, and Other Means of Detecting Complex

Ions
Some of the techniques for examining the constitution of electrolytes
have been described in Section 3.8. In particular, the usefulness of Raman
 IONIC LIQUIDS 591

spectroscopy for the detection of complex ions was stressed. The unique
advantage of this spectroscopic method is that analysis of its data may give
the formula of the complex entity present, together with detailed information
on its structure, for example, the angles between the constituent bonds in a
complex ion.
A simple comparative procedure in utilizing the Raman method is to
compare the spectra obtained in fused salts with those obtained from the
salts in other states of existence (aqueous solutions, solid state, or gas phase),
in which the spectra have been shown by independent methods to be due to
certain identified species.
An example of identification by this comparative method is given by
the study of liquid mercuric halides HgCl 2 and HgBr2 performed by Janz.
Pure liquid mercuric halides yield Raman spectra which show Raman fre-
quencies identical to those from the solid. In the crystalline state, however,
it is known that the chloride and the bromide have distinctly molecular
lattices with linear X-Hg-X (X = Cl or Br) triatomic molecules. It can
therefore be concluded that the essential elements of the solid structure
are retained in the liquid state, i.e., the mercuric halides from "molecular"
melts, not ionic liquids.
Upon the addition of KCI to a melt of HgCI 2, the frequencies corre-
sponding to the linear triatomic molecule persist but with diminishing
intensity. On the other hand, new frequencies appear that indicate the
formation of new species, i.e., complex ions (cf Section 3.8). These new
frequencies can be attributed to tetraatomic and pentatomic complex ions,
HgCl a- (planar) and HgCl!- (tetrahedral), respectively (Fig. 6.45).
The case of the CdCl 2-KCl system can be quoted as an example of
the absolute method of using Raman spectroscopy for the identification
of complex ions in fused salt mixtures (cf Chapter 3). The Raman spectra
of the CdCl 2-KCl system in the liquid state give four Raman peaks for
the 50% KCl system in the liquid state (Fig. 6.46). It was concluded from
these frequencies that a pyramidal CdCl 3 - is the predominant complex ion

Planar Tetrahedral
Hg CI3" HOCI;

Fig. 6.45. Planar HgCI2 and tetrahedral

complex ions.
592 CHAPTER 6

80r------,------,-------.------;

40~--~-+------~--_.~~----~

20~-----++_--~~~----~----~

300 2110 200 150 o

AVCrrr l

Fig. 6.46. The Raman spectrum of a

CdCI 2-KCI system (with 50% KCI) showing
four absorption peaks.

7·0

6·0

Igt
0
0 5·0
q

..
•
"0
E
Q
4·0

"
c
.~
~ 3·0
~
•c
u
0
u
c
.~ 2·0
•"
Ci.
E
0
u CdBr+
1'0

o 5 10 15 20 25 30
Ligand concentration x 10~ mole 1,000 g-I

Fig. 6.47. Dependence of the concentration of various complex

ions Cd(Brh upon the addition of KBr.
 IONIC LlOUIDS 593

in molten CdCI 2-KCl over the composition range examined, i.e., 33.3 to
66.6%.
Other methods of investigation (e.g., studies of the self-diffusion coef-
ficient of Cd in the mixtures) had suggested that the behavior of the CdCI;-
KCI system could be interpreted in terms of the formation of complex ions.
However, such methods only give a qualitative indication of the presence
of a complex ion and indicate its variation with composition. Raman
spectroscopy, on the other hand, has two distinct advantages. Firstly, the
spectra reveal in an unambiguous way the presence of complex ions in
molten salts; new frequencies (indicated by peaks in the intensity versus
wavelength curves) arising from the addition of the ligand must be attribut-
ed to the formation of new species. Secondly, the number, height, and
frequencies of the peaks can be compared with the corresponding quantities
which are indicated by the theory of Raman spectra for various assumed
ions. In this way, one can get comparatively detailed knowledge regarding
the nature of the complex ions present. For example, in CdC1 2-KCI mixtures,
it is possible to distinguish between the presence of tetrahedral and planar
forms of assumed complexes with the same molecular formula.
A misapprehension sometimes exists that a molten mixture of two salts
can give rise only to a single species of complex ion. This is not usually
the case, as may be seen by the results of an examination of the complex ions
formed by adding halide ligands to liquid cadmium nitrate dissolved in an
LiN0 3-NaN03 solution. Which of several possible ions is formed depends
largely on the amount of halide ions present per cadmium atom (Fig. 6.47).

Further Reading

1. W. Bues, Z. Anorg. Allgem. Chem., 279: 104 (1955).

2. D. Inman, J. Electroanal. Chem., 5: 476 (1963).
3. S. C. Wait, Jr., and G. J. Janz, Quart. Rev. (London), 17: 225 (1963); Chapter 5
in: H. A. Szymanski, ed., Raman Spectroscopy, Plenum Press, New York, 1967.
4. G. E. Walrafen and D. E. Irish, J. Chem. Phys., 40: 911 (1964).
5. K. Balasubramanyam, Electrochim. Acta, 8: 621 (1963); J. Chem. Phys., 40:
2657 (1964); 42: 676 (1965).
6. J. Lumsden, The Thermodynamics of Molten Salt Mixtures, Academic Press,
Inc., New York, 1966.
7. H. Bloom, The Chemistry of Molten Salts, W. A. Benjamin, Inc., New York,
1967.
8. J. H. R. Clark and C. Solomons, J. Chem. Phys., 47: 1823 (1967).
9. C. Solomons, J. H. R. Clark, and J. O'M. Bockris, J. Chem. Phys., 49: 445
(1968).
594 CHAPTER 6

6.7. MIXTURES OF LIQUID OXIDE ELECTROLYTES

6.7.1. The Liquid Oxides

Fused salts (and mixtures of fused salts) are not the only type ofliquid
electrolytes. Mention has already been made of fused oxides and, in par-
ticular, mixtures of fused oxides.
A typical fused oxide system is the result of mixing intimately a non-
metallic oxide (Si0 2, Ge0 2, B20 a , P 205' etc.) and a metallic oxide (Li 2 0,
Na 20, K 2 0, MgO, CaO, SrO, BaO, AI 20 a , etc.) and then melting the
mixture. The system can be represented by the general formula: MxOy-
RpOq, where M is the metallic element and R the nonmetallic element.
There is obviously a case for mentioning these fused oxide systems in
this brief presentation of pure liquid electrolytes, but why give them a
special consideration? Are not the concepts developed for the understand-
ing of molten salts adequate for the understanding of molten oxides? The
essential features of fused salts emerge from models of the liquid state and,
in particular, from the hole model based on density fluctuations. Is the hole
theory adequate as a basis for a rationalization of the behavior of the
fused oxides?

6.7.2. Pure Fused Nonmetallic Oxides Form Network Structures

Like Liquid Water
Some of the special features of molten oxides must now be described
for it is these features which do not permit the hole model of ionic liquids
to be applied to fused oxides in the way it is to molten salts-namely,
that, as far as transport properties are concerned, hole formation is the
principal feature of the model that is consistent with experiment.
The first interesting feature of molten silica is that its conductivity is
more like that of water (i.e., molten ice) than that of fused NaCl (Table
6.32).
TABLE 6.32
Specific Conductivity of Water and Molten Si0 2 and NaCI near the Melting
Point

Substance )t, ohm- l cm- l Temp.,oC

Si0 2 7.7 X 10- 4 1800

NaCI 3.6 801
H2O 4 x 10- 8 18
 IONIC LIQUIDS 595

The dissimilarity in the conductivities of liquid NaCl, on the one hand,

and liquid water and liquid silica on the other, is of fundamental import.
When NaCl is fused, the ionic lattice (the three-dimensional periodic arran-
gement of ions) is broken down (see Section 6.1.2) and one obtains an ionic
liquid. On the other hand, when ice is melted, the tetrahedrally directed
hydrogen bonding involved in the crystal structure of ice (Fig. 2.24) is
partially retained. Thus, water is not a collection of separate water molecules
but an association (based on hydrogen bonding) of water molecules in a
three-dimensional network. The network, however, does not extend inde-
finitely. There is a periodicity and only short-range order implying a certain
degree of bond breaking. It is this network structure that is responsible
for the small mole fraction of free ions (H+ and OH-) in water, in contrast
to the almost total absence of any ion association (into pairs, complexes,
etc.) in liquid NaCl. This great difference in the concentration of charge
carriers is responsible for the several-orders-of-magnitude difference in the
specific conductivities of liquid NaCl and liquid water.
Now, the specific conductivities of water and of fused silica are quite
similar. This suggests that the structures of crystalline water [Fig. 6.48(a)]
and crystalline silica [Fig. 6.48(b)] have much in common. Each oxygen
atom in ice is surrounded tetrahedrally by four other oxygens, the oxygen-

(a)

( b)

• 51 atom

00 atom
• H atom
Fig. 6.48. The similarity between the basic
building blocks of the ice and crystalline
silica structures.
596 CHAPTER 6

oxygen bonding occurring by a hydrogen bridge (the hydrogen bond). In

crystalline silica, there are Si04 tetrahedra occurring through an oxygen
bridge. The different forms of ice and the different forms of silica (Fig. 6.49)
correspond to different arrangements of the tetrahedra in space.
It is reasonable, therefore, to consider that fused silica resembles liquid
water. Just as liquid water retains, from the parent structure (ice), the three-
dimensional network but not the long-range periodicity of the network,
one would expect that liquid silica also retains the continuity of the tetra-
hedra, i.e., the space-network, but loses much of the periodicity and long-
range order which are the essence of the crystalline state. This model of

(a I Sinole chain of tetrohedra

(bl Double chain of tetrahedra

(c I Sheet of tetrahedra

(d I Three - dimensional net of tetrahedra

Fig. 6.49. Forms of silicates resulting from

different ways of linking up SiD. tetrahe-
dra: (a) single chain of tetrahedra; (b) dou-
ble chain of tetrahedra, as in asbestos; (c)
sheets of tetrahedra, as in clay, mica, and
talc; and (d) networks of tetrahedra, as in
ultramarine.
 IONIC LIQUIDS 597

fused silica, based on keeping the extension of the network but losing the
translational symmetry of crystalline silica, implies a low conceritration of
charge carriers in pure liquid silica and, therefore, the low conductivity
in comparison with a molten salt (cf Table 6.32).

6.7.3. Why Does Fused Silica Have a Much Higher Viscosity

Than Do Liquid Water and the Fused Salts?
It has just been argued that the conductivities of simple ionic liquids,
on the one hand, and liquid silica and water, on the other hand, are v'astly
different because a fused salt is an un associated liquid whereas both molten
silica and water are associated liquids with network structures. What is
the situation with regard to the viscosities of fused salts, water, and fused
silica? Experiments indicate that, whereas water and fused NaCl have
similar viscosities, fused silica is a highly viscous liquid (Table 6.33). Here
then is an interesting problem.
The theories of transport processes in liquids are based on elementary
acts, each act consisting of two steps: (I) holes are formed; (2) then particles
jump into these holes (cf Section 6.5.4). For fused salts and other non-
associated liquids, this theory based on holes was quite successful in ex-
plaining the movements and drift of particles. The mean volume of a hole
is determined by the surface tension as follows [Eq. (6.29)]

<Vh) ( kT)i
= 0.55 -y-

from which it turns out (Table 6.11) that, in fused salts, the size of the holes
is roughly equal to the size of the ions. Hence, holes can receive the jumping
ions of the fused salt. Further, in simple ionic liquids, the free energy of
activation for the jumping of ions into the holes is about an order of magni-

TABLE 6.33
Viscosities of Fused Silica. Water. and Fused NaCI

Viscosity,
Substance Temp., DC
poise

Liquid SiOs 1720 3 X 106

Water 25 9 X 10- 3
Fused NaCl 850 13 X 10-3
598 CHAPTER 6

TABLE 6.34
Surface Tension of Molten SiO. and NaCI near the Melting Point

Molten salt )" dyne cm- 1 Temp.,oC

SiO. 250 1570

NaCI 114 801

tude smaller than the free energy for forming the holes. Once the hole is
formed in a fused salt, the jump is easy.
The surface tension of fused silica is only about three times that of fused
sodium chloride (Table 6.34). Hence [see Eq. (6.31)], in fused silica too,
the range of sizes of holes present would be of atomic dimensions ("" I to
10 A), as for fused ionic liquids.
In simple ionic liquids (e.g., NaCl), the holes are atomic sized; so are
the jumping ions. That is why jumping is no problem. But what particles
can jump into the atomic-sized holes of fused silica? Obviously, whole
macro networks cannot jump into atomic-sized holes. How then can
jump-dependent transport processes occur? There is a way. Small segments
(one to a few atoms in size) can break off from the network, and these
pieces (segments) can do the jumping (Fig. 6.50).
The comparison between transport processes in simple fused salts and
in fused Si02 is interesting. In molten salts, the jump was relatively easy,
but holes had to form. The rate of the whole process was controlled pre-
dominantly by this rate of formation. But now, with molten silica, the

Empty hole

seom:~1
~ ,
(a) \
\
I
I
~/

(b)

Fig. 6.50. Schematic diagram to show a segment of an

Si0 4 chain breaking off and jumping into an empty hole;
(a) before jump; (b) after jump.
 IONIC LIQUIDS 599

TABLE 6.35
Comparison between Transport Processes in Simple Fused Salts and in
Liquid Silica

System

Fused salt Liquid silica

Essence of situation Particles waiting to jump Holes waiting for small

into holes enough segments of net-
works
Rate-determining step Hole formation Rupture of bonds between
segments and network
Energy of activation 11 Hhole formation L1 Hbonu rupture
for transport process
(approx.)

balance of influences is different. Holes are as easily formed as in the ionic

liquids because the rate of hole formation is controlled by the vibrations
of the atoms relative to each other, but it is difficult to produce the jumping
particles by rupturing strong Si-O-Si bonds holding the network together
(Table 6.35). Therefore, in the silicates, the rate-determining process is
the bond-rupture step. Thus, in simple ionic liquids, the experimental
activation energyt for the transport process, such as a viscous flow, is
determined by the enthalpy of hole formation; in associated liquids with
network structures, it is determined by the energy required to break the
bonds of the network.
But, then, why is it that, even though both fused silica and water have
network structures, their viscosities are quite different; in fact, the viscosity
of water is similar to that of molten NaCl, which has no network? The
difference in the viscosity behavior of water and of fused silica lies in the
ease with which segments can be broken off the two networks. The Si-O-Si
chemical bonds are much more difficult to rupture than the O-H-O
hydrogen bonds (cf Table 6.36). Thus, the small segments-probably H 20
molecules-are so easily produced in water that the holes do not have much

t It will be recalled that it was decided that the quantity obtained from the slope of the
log'f/ versus liT curve would be termed an energy of activation irrespective of whether
it pertains to constant pressure or constant volume conditions, though, under the former,
it is an enthalpy and, under the latter, it is an energy.
600 CHAPTER 6

TABLE 6.36
Heat of Dissociation of Si-O and O-H Bonds

Heat of dissociation,
Bond
kcal mole- 1

Si-O-Si 104
O-H-O, hydrogen bond 5

of a wait; an ease of flow, high fluidity or low viscosity, results. This is not
the case with fused silica because of the much higher bond-breaking energy,
and a high viscosity results.
Some support for the idea that the viscous-flow properties of associated
liquids such as liquid silica and water are determined by the step of bond
breaking rather than that of hole formation comes from the experimental
plots of log 1] versus liT. These plots suggest a slight trend away from
linearity (Fig. 6.51), which is not the case for fused salts (Fig. 6.21). For

5.2

5.1

5.0

4.9
Log 'I
4.8

4.7

4.6

4.5

4.4
o

4.3

Fig. 6.51. Two independent sets of data for the

difficultly determinable viscosity t) of liquid SiO •. Both
suggest a slight tendency to curve in the sense
that the energy of activation becomes higher at the
lower temperatures.
 IONIC LIQUIDS 601

water, too, the plot is curved slightly with the experimental energy of activa-
tion for viscous flow, E,/,p, decreasing with increasing temperature, The
explanation for this phenomenon is as follows, Because the energy of
activation for viscous flow depends upon the breaking of bonds and because,
according to the Boltzmann distribution, the fraction of broken bonds
increases with temperature, the fraction required to be broken by the
shearing force in viscous flow decreases with an increase in temperature
and, hence, there is a decreasing energy of activation with increasing tem-
perature.
To summarize: Unlike fused salts, mixture of fused oxides are associated
liquids, In fused oxides, hole formation is necessary but not the vital step
which determines the rate of transport processes. It is the rate of prod uction
of individual small jumping units which controls them. This conclusion
makes it essential to know what (possibly different) entities are present in
fused oxides and what are the kinetic entities. In fused salts, the jumping
particles are already present; the principal problem is the structure of the
empty space or free volume or holes and the properties of these holes. In
molten oxides, the main problem is to understand the structure of the
macrolattices or particle assemblies,

6.7.4. The Solvent Properties of Fused Nonmetallic Oxides

If fused silica is a three-dimensional, aperiodic network, all the atoms

are to some extent joined together, i.e., the liquid is a giant molecule.
Can ions dissolve in such a structure? Water, too, is a network structure
like Si0 2 , and ions dissolve in water, Hence, liquid SiO z may well be ex-
pected to have solvent properties leading to the production of ionic so-
lutions.
Water, it may be recalled (Chapters 2 and 5), has two modes of solvent
action depending on the nature of the added electrolyte. The water can
contact an ionic crystal (e,g" NaCI), detach the ions from the lattice through
the operation of ion-dipole (or ion-quadrupole) forces, and convert them
to hydrated ions (Chapter 2).
The water can also interact chemically with a potential electrolyte (e.g.,
CH 3COOH), The hydrogen atoms forming part of the hydroxyl group of
the organic acid do not differentiate the oxygen atoms of the water network
and those of the OH group. The hydrogen of the OH group detaches itself
from the organic acid. Two ions are thus formed, a hydrogen bonded to a
water molecule from the solvent, and an organic anion. This mode of
solvent action is a proton-transfer or acid-base reaction.
602 CHAPTER 6

:'~:~v~Z"~',,~:4
Si-~ ~
+~
Co _ Cat+
o
Fig. 6.52. The interaction between a metal oxide
and a silicon atom of the silica network.

The type of solvent action which fused nonmetallic oxides have on

metallic oxides may be likened to the second type of dissolution process,
i.e., proton-transfer reactions. The process may be pictured as follows. The
oxygens cannot discriminate between the metal ions M + (of the metallic
oxide), with which they have been associated in the lattice of a metal oxide
before dissolution, and the oxygen atoms of the Si04 tetrahedra contain-
ed in the solvent fused silica. The oxygens sometimes, therefore, leave the
metal ions and associate with those of the tetrahedra. Dissolution has
occurred with a type of oxygen-transfer reaction (see Fig. 6.52).
There is a further analogy between the solvent actions exercised by
water and by a fused nonmetallic oxide. Just as water dissolves an elec-
trolyte at the price of having its structure disturbed, so also the reaction
resulting from the addition of a metallic oxide to a fused nonmetallic oxide
like silica is equivalent to a bond rupture between the Si04 tetrahedrons
(Fig. 6.53).
Solvent action occurs therefore in fused oxide systems along with a
certain breakdown of the network structures present in the pure liquid
solvent (e.g., in pure liquid silica).

Fig. 6.53. The oxygen-transfer reaction leads to

a rupture between the SiD. tetrahedra.
 IONIC LIQUIDS 603

6.7.5. Ionic Additions to the Liquid-Silica Network: Glasses

An interesting aspect of molten oxide electrolytes may be mentioned

at this point. Some liquids can appear to be solids, i.e., some solids are
really liquids of such high viscosity that no significant flow occurs over tens
or hundreds of years. These solidlike liquids are called glasses. The structure
breaking which has just been described is an aspect of the basic mechanism
behind the formation of glasses, which might be regarded as "cold molten
silicates. "
When ions with a relatively large radius and therefore small peripheral
field are added to liquid silica, they produce structure breaking in the
network. This is shown in a one-dimensional way in Figs. 6.52 and 6.53.
With an increase in the number of ruptures in the network there is an
increase in the number of "free" or "dangling" ends of the ruptured net-
work. The network becomes, therefore, increasingly distorted with increase
in the mole fraction of the metallic oxide present.
If, now, the "broken-down network" is at a sufficiently high temper-
ature, it is known as a liquid silicate. Such a system results, for example,
from adding an alkali oxide (e.g., Na 2 0) in low concentration to Si0 2 •
The system can be, at sufficient temperature, distinctly a liquid, and the
viscosity near the melting point may be, for example, 427 poises (at 1550 °C).
When the temperature is dropped, the liquid silicate attempts to "freeze,"
but, to do this, the long-range order of the crystalline silicate has to be
reestablished. The establishment of order, however, requires rearrangement
of the structure, i.e., movements of the kinetic units of the broken-down
network to get into the sites corresponding to order. Were these particles,
or kinetic units, simple, they would be agile, i.e., their movements would
be easy, the viscosity would be low, and the restoration of crystalline order
would be accomplished almost immediately. A crystalline solid with sharp
melting point would result.
But this reorganization is precisely what is not quickly accomplished
by the entities in the broken-down networks in liquid silicates. The entities t
resulting from the rupturing of three-dimensioned networks when metal
oxides are added are big and sluggish. They cannot get into line in time;
the viscosity is too high. The loss of thermal energy during cooling catches
them still out of position as far as the regular three-dimensional arrange-
ment of the crystalline silicate is concerned. Then it is too late for, as the

t The nature of these big anionic entities which exist in glass-forming silicates is discussed
in Section 6.7.8.
604 CHAPTER 6

temperature drops still further, they are still less likely to be able to get
back into line. Thus, the loss of thermal energy freezes in the structure of
the liquid silicate-a glass is formed. It is a "frozen liquid," i.e., a liquid
which has been supercooled to such a high viscosity that it seems to have
the essential requirement of a solid, absence of flow. A beam of X rays,
however, would reveal an essential characteristic of the liquid state, namely,
the absence of long-range periodicity.
If, however, sufficient time is allowed (e.g., a few hundred years), a
sufficient number of the units of the broken-down network will get back
into line. Long-range order will be partly reestablished; the glass de vitrifies
or "deglassifies."
How is the liquid-silicate network affected by the addition of various
types of ions in the production of the peculiar and complicated kind of
pure electrolyte, the glass? It is the answer to this structural question which
provides the basis for the understanding of the glassy state.

6.7.6. The Extent of Structure Breaking of Three-Dim~nsional

Network Lattices and Its Dependence on the Concentra-
tion of Metal Ions

The extent of breakdown of the network structures present in the pure

liquid solvent can be treated in the following way. The Si0 4 tetrahedron is
accepted as the basic structural unit in a mixture of a metallic oxide and
fused silica. But are the tetrahedrons linked together at all, and, if so, how
are they linked together? What is the number of links per silicon? Thus,
water molecules are the basic structural unit in an aqueous solution, but
extensive linkage and intermolecular bonding occurs in pure water. How is
this affected by the presence of ions in solution (Section 6.1)?
In the fused-oxide system, the metal oxide is the structure breaker;
in aqueous solutions, the electrolyte is the structure breaker. Does the extent
of structural breakdown of the continuous Si-O network present in pure
silica before addition of MO depend on the concentration of MO? The
extent of breakdown depends on the concentration of the structure breaker;
then one would expect that properties which depend on the size and nature
of the structures present would also be concentration dependent.
In fused-oxide systems, a simple way of expressing the concentration
of metal oxide MxOy in the fused nonmetallic oxide RpOq is sometimes
used. This involves the so-called "O/R ratio." The O/R ratio is simply
related to the mole fraction of the metallic oxide. For example, an 0/Si
 IONIC LIQUIDS 605

ratio of 4 in a system of Li 20 and Si0 2 is obtained when the Li 20 has a

mole fraction of 66% (i.e., 2Li 20 +
Si0 2 has four O's to one Si).
One way of probing the sizes of structures presen~ in fused-oxide systems
and their variation with the mole fraction of MO added to the nonmetallic
oxide is through the variation of the ease of flow with composition. The
viscosity of the system must be keenly sensitive to the size and nature of
the kinetic entities present because it is these entities which must make the
jumps from site to site involved in viscous flow.
The experimental results on the variation of the energy of activation
for viscous flow, E", as a function of the mole per cent of the metal oxide,
are shown in Fig. 6.54.
The basic feature of the results appears to be a very high ("" 150 kcal
mole-I) energy of activation for viscous flow of the pure nonmetallic oxide
and then a rapid fall with addition of the metallic oxide, whereupon there
is a leveling off to a value of about 40 kcal mole-I, which remains relatively
unchanged between about 10 and 50 mole % of Na 20 in Si0 2 • This be-
havior can be used (Section 6.7.7) as a touchstone in deciding between
alternate models for the structural changes accompanying changes in
metal-ion concentration.

Pure silica -134 keol

-
I
.!!
o
E
o 40

UJ
...
20
NozO-Si0 2

o
Molar % No 2 0

Fig. 6.54. The variation of the energy of acti-

vation for viscous flow in a Na 2 0 + Si0 2 melt
as a function of the mole per cent of Na 2 0.
 606 CHAPTER 6

The structural theories presented will be in terms of the liquid silicates

-for most research in molten oxides has been done with them-but one
can extend the basic structural ideas to fused-oxide systems involving metal
oxides dissolved in B20 3 and P205 and probably to most liquid electrolytes
in which there are largely continuous network structures at very low con-
centrations of MxO y.

6.7.7. The Molecular and Network Models of Liquid Silicate

Structure

A naive view of the happenings consequent to the addition of metallic

oxides to molten silica is to think of different uncharged molecular species,
the species changing with the mole fraction of the metallic oxide. This
view-which was popular among metallurgists as late as 1950-has to be
given up with alacrity because conductance studies of mixtures of MxOy
and Si02 show that these systems are highly conducting and therefore rich
in charge carriers (Table 6.37). One has to suggest ionic, rather than mo-
lecular, structures.
A second attempt at the interpretation of the structure of liquid silicates
starts with a consideration of the curve of E'I versus mole % MxOy (Fig.
6.54). It is in terms of the gradual breakdown of the three-dimensional
network of fused silica. Just as there is thermal bond breaking on going
from crystalIine to fused silica, one can consider that, with the addition of,
say, Na 20, the additional 0 atoms cause Si-O-Si bonds in the originally
continuous network of Si0 2 to break. This gives structure breaking to
various composition-dependent degrees.
A mole fraction of 66% of M 20 implies, as already stated, an OjSi
ratio of 4 and must, therefore, be considered a composition at which simple
SiO~- ions exist. t
What model can be suggested for the corresponding structural changes
"inside" the fused-oxide system for the M 20 mole-fraction range from 0
to 66%?
From the fact that there is such a sharp fall in the energy of activation
for viscous flow between zero and about 10 mole % of M 20 (Fig. 6.54),
one must think of a radical change (over this composition range) in the
difficulty of causing a (possibly changing) kinetic entity to jump from site

t Correspondingly, for M.O > 66 molar %, there are oxygen atoms in excess of the ability of
the Si's present to coordinate them (OjSi ratio> 4). Hence, liquids with such compositions
probably contain SiO:- and 0- - entities in addition to the cations M+ present.
 IONIC LIQUIDS 607

TABLE 6.37
Electric Conductance in the Liquid Silicates

Composition
"1750 oC, LJH* ,
Cation XM",O-YSi0 2 , A 1750 ,c
ohm-1 cm-1 kcal g-l ion- 1
X:Y

K+ 1 :2 1.5 71.8 8.2

1:1 2.4 82.7 8.0
Na+ 1 :2 2.1 83.3 12.0
1:1 4.8 126.0 13.5
Li+ 1 :2 2.5 77.8 11.6
1:1 5.5 109.0 10.6
2 :1 23.2 332.0 9.6
Ba++ 1 :2 0.18 6.4 33.2
1:1 0.60 16.2 17.5
2:1 1.32 29.9 9.0
Sr++ 1 :2 0.21 7.7 36.0
1:1 0.63 15.7 26.7
2:1 1.4 26.8 17.0
Ca++ 1 :2 0.31 11.4 30.0
1:1 0.83 18.4 20.0
2: 1 1.15 18.8 20.0
Mn++ 1 :2 0.55 18.2 24.0
1:1 1.8 35.1 16.0
2 :1 6.3 85.5 12.0
Fe++ 1:1 1.82 44.0 15.0
Mg++ 1 :2 0.23 6.5 34.0
Mg++ 1:1 0.72 12.2 24.0
2:1 2.15 24.7 17.0
AI+++ 10 wt % 3.10-3 0.20 22.0
Ti++++ 10 wt % 6.10- 4 0.05 35.7

to site in the random walk which is the basis of diffusion (Section 6.5.4).
The model must of course contain the explanation of the generation of
more free ions to account for an increase of conductivity with an increasing
amount of MO. The anions postulated as predominant for a given OIR
ratio must meet some exacting requirements. Thus, (1) they must have
formulas consistent with the overall OIR ratio; (2) they must have a total
608 CHAPTER 6

charge which compensates the total charge of the cations and thereby
ensures overall electro neutrality ; (3) they must be shaped in a way consistent
with the bond angles, particularly the R-O-R angle, shown from X-ray
data in the corresponding solids.
The broad approach by Zachariasen on the network model of liquid
silicates is to break down the network present in the pure fused nonmetallic
oxide thus

Three-dimensional network with some M.O

Two-dimensional sheets
thermal bond breaking (e.g., Si0 2) added

More M 20 Still
added
I One-dimensional chains Simple SiOf- monomers

The details of the network theory of liquid-silicate structures, which

was first suggested to rationalize the glassy state, are presented in Table
6.38. The chief defect of this model is that it argues for very large changes
in the heat of activation for viscous flow in the composition range of 33 to
66% (but contrast the indication in Fig. 6.54). This is because, in the net-
work model, the size and shape of the kinetic unit-the jumping entity-
is supposed to undergo a radical change in the composition range of 33
to 66%. Thus (cf Table 6.38), sheets are being broken up into chains. The
kinetic unit of flow would therefore be expected, on this model, to change

TABLE 6.38
The Network Theory of Liquid Silicate Structure

Range of composition,
Silicate entities present
mole % M.O

o Continuous three-dimensional (3-D) network of Si0 4 tetra-

hedrons with small degree of thermal bond breaking
0-33 Essential 3-D network of Si0 4 tetrahedrons with number of
broken bonds equal to number of added 0 atoms from
M.O; end of 3-D boundary at 33%
33% "Infinite" 2-D sheets of Si0 4 tetrahedrons; M+ ions and 0-
ions between sheets
33-50% Region of sheets and some chains of tetrahedrons
50% Chains of infinite length
50-60% Chains of decreasing length
66% SiOt-
 IONIC LIQUIDS 609

60

50
.0
X 40
u
0
.,
0
E
..E
..... 30

~~
!
~Q. 6
--. 20

~'
jI
>I~
0000

10

0 II,
II,
,
II,

0 40 50 60
mole percent MfJ

Fig. 6.55. The sharp change in the expansivity of

M.O-SiO. melts around the 10 mole % of M.O compo-
sition; (L:-.) K.O + SiO., (0) Na.O + SiO., (0) Li.O +
SiO., ('V) SiO •.

radically in size over this composition range, and this diminution in size
of the flowing unit would be expected to make the heat of activation for
viscous flow strongly dependent on composition in this composition range
(33 to 66% M 20).
In reality, however, the E,} changes by only 25% over the composition
range of 10 to 50% MxOy (Fig. 6.54), whereas, from 0 to 10%, the change
in energy of activation is about 200% (Fig. 6.54).
Another defect of the network model concerns the implications which
it has for phenomena in the composition region around 10% MxO y. This
is an important composition region. Experimentally, whether one measures
the composition dependence of the heat of activation for viscous flow, of
expansivity, of compressibility, or of other properties, in all cases, there is
always a sharp and significant change (Fig. 6.55) around the 10% MxOy
composition, which indicates a radical structural change at this point. But
this composition has no special significance at all, according to the net-
work model. Thus, one must-although the network model served well in
610 CHAPTER 6

an earlier stage of the development of the theory of glasses--reconsider

the situation and develop a model which corresponds more, in its predic-
tions, to the experimental facts.

6.7.8. Liquid Silicates Contain Large Discrete Polyanions

Consider the situation as one decreases the 0/R ratio, i.e., decreases
the mole per cent of the metallic oxide MxOy. Between 100 and 66%,
there is little need for special modelistic thinking because the quadrivalency
of silicon and the requirements of stoichiometry demand that the ionic
species present are monomers of SiOi- tetrahedrons.
It is in the composition range of 66 to 10% MxO/, that the network
model stumbles (cf Section 6.7.7) in the face of facts.
What are the requirements of a satisfactory structural model for ionic
liquids in this composition range? First, since transport number determ-
inations show that the conduction is essentially cationic, the anions must
be large in size compared with the cations (see Sections 4.4.8 and 4.5).
Secondly, the marked change in properties (e.g., expansivity) occurring at
10% MxOy indicates radical structural changes in the liquid in the region
of this composition. From the sharp rise of the heat of activation for the
flow process to such high values (toward 100 kcal mole- 1 at compositions
below 10 mole % MxOy) with decreasing mole per cent of MxOy in this
composition region, one suspects that the structural change which is the
origin of all the sudden changes near 10 mole % M 2 0 must involve sudden
aggregation of the Si-containing structural units into networks. The difficulty
of bond breaking to get a flowing entity out of the network and into another
site gives a rationalization to the very sluggish character of the liquid at 10 or
less mole % of MxOy. Finally, from 66 to 10% there must be no major
changes in the type of structure, except some increase in the size of the
entities, because there is only a small increase in the heat of activation for
viscous flow over this composition region. This relative constancy of the
heat of activation for flow over this composition region means that the

sm"'~.""
Fig. 6.56. The dimer ion Si20~-.
 IONIC LIQUIDS 611

various structural units present can become the kinetic entities of flow
over this region without great change of the energy involved, i.e., the
anions present must have similar structures.
The construction of a model can therefore start from the SiOf- ions
present at the orthosilicate composition (66% MxOJl)' With a decrease in
the molar fraction of MxO~, the size of the anions must increase to maintain
the stoichiometry. One can consider that there is a joining-up, or polymer-
ization, of the tetrahedral SiOl- monomers. For example, the dimer Si20~­
(Fig. 6.56) could be obtained thus

This polymerization into larger polymerized anions (or polyanions) is the

essential concept of the discrete-polyanion model suggested by Lowe, Mac-
kenzie, et al. for the structure of mixtures of liquid oxides corresponding
to composition greater than I 0% MxO~ and less than 66%.
At the outset, it seems no easy task to derive the structure of the poly-
anions predominant at each composition of the liquid oxides. However,
several negative criteria limit the choice of possible polyanions. Thus,
electroneutrality must be maintained at all compositions, i.e., the total
charge on the polyanion group per mole must equal the total cationic
charge per mole for a given composition. Since the cationic charge per
mole must decrease with decreasing MxOy mole per cent, the negative
charge on the polyanions per mole equivalent of silica must also decrease.
Thus, the size of the polymerized anions must increase as the molar fraction
of MxOy decreases.
After a dimer, i.e., S20~-, is formed, the next likely anionic entity to
appear as the M x Oy jSi0 2 ratio falls might be expected to be the trimer

0-
I
3-03Si-0-SiO~- + SiOl- ---+ 3-03Si-0-Si-0-SiO~-
I
0-

Following the trimer, a polymeric anion with four units may be in-
voked to satisfy the requirements of electroneutrality and stoichiometry,
etc., the general formula being SinO&~~+i2)-. On this basis, however, when a
composition of 50% MxOy is reached, i.e., when the mole fraction of MxOy
is equal to that of Si0 2, the SijO ratio is 1 :3. However, from the general
formula for the chain anion, i.e., Si n O&;,"ti2)-, it is clear that OjSi = 3 when
612 CHAPTER 6

(3n + 1)/n = 3, i.e., when n -+ 00. Hence, near to 50% MzO y, an attempt
to satisfy the electro neutrality and stoichiometry by assuming that linear
chain anions, extensions of the dimers and trimers above discussed, would
imply a large increase in the energy of activation for viscous flow in the
composition range (say 55 to 50 molar % of MzOy) because the linear
polymer would here rapidly approach a great length. But no such sharp
increase in the heat of activation for viscous flow is observed experimentally
in the range of 55 to 50 molar % of MzO y • Hence, the composition range
in the liquid oxides in which linear anions can be made consistent with the
flow data is relatively small, between 66 and somewhat greater than 50
molar % of MzO y . The linear anion must be given up as a predominant
anionic constituent before the metal oxide composition has dropped to
50 molar %.
An alternative anionic structure near the 50% composition which
satisfies stoichiometric and electroneutrality considerations is provided by
ring formation. If, in the composition range of 55 to 50% MzO y, the linear
anionic chains (assumed to exist at compositions between some 50 and 60
mole per cent MxOy) link up their ends to form rings such as Si30~- or
Si 40r2" (Fig. 6.57), then such ring anions satisfy the criteria of the O/Si
ratio, electro neutrality, and also the Si-O-Si valence angle which X-ray
data make expectable. Thus, the Si a03- anion corresponds to an O/Si ratio
of 3, and, if one is considering a 50% CaO system, the charge per ring
anion is 6-, and the charge on the three calcium ions required to give
3CaO/3Si0 2 is 6+. Further, the Si30~- anion is not very much larger than
the Si20~- dimer, and, hence, there would not be any large increase in the
heat of activation for viscous flow. With regard to the Si-O-Si bond angle,
in the Si30~- and Si 4 0r2" ions, it is near that observed for the corresponding

Fig. 6.57. The proposed ring anions: (a)

SiaO:- and (b) Si,O~;.
 IONIC LIQUIDS 613

Oxygen

Silicon

(0) (b)

Fig. 6.58. The proposed large ring anions:

(a) Si60~5 and (b) Si80~o.

solids, i.e., the minerals wollastonite and poryphyrite, respectively, which

are known to contain Si 60g- and Si4 0ri.
Further structural changes between 50 and 30% MxO" are based on
the Si 30g- and Si 40r2 ring systems. At the 33% compositions, polymers
(Si 60 15 )6- and (Si s0 2o )S- (Fig. 6.58) can be postulated as arising from
dimerization of the ring anions Si 30g- and Si 40i2 (Fig. 6.57). As the MxOy
concentration is continuously reduced, further polymerization of the rings
can be speculatively assumed. For example, at M x O,,/3Si0 2 when MxOy
is 25%, the six-membered ring would have the formula Si90~1 and would
consist of three rings polymerized together (Fig. 6.59).
Ring stability might be expected to lessen with increase of size with
increasing proportion of Si0 2 because the silicate polyanions which corre-
spond to compositions approaching 10 molar % of MxOy would be very
long. The critical 10% composition at which there is a radical change in
many properties may be rationalized as that in the region of which a discrete
polymerized anion type of structure becomes unstable (because of size) and
rearrangement to the random three-dimensional network of silica occurs.
That is, a changeover in structural type occurs to what is fundamentally
the Si0 2 network with some bond rupture due to the metal cations. The
very large increase in the heat of activation for flow which takes place at

Oxygen

Silicon

Fig. 6.59. The proposed six-membered ring

anion Si.O:~.
614 CHAPTER 6

this composition (Fig. 6.54) would be consistent with this suggestion, as

would also the sudden fall in expansivity.
These ideas of the discrete-polyanion model for liquid-silicate structure
are summarized in Table 6.39. The description is, as in most models, highly
idealized. Thus, all the silicate anions may not have the Si-O-Si angle
of the crystalline state; only the mean angle may have the crystalline value.
Further, the discrete anion suggested for a particular composition is only
intended to be that which is there dominant, not exclusive. Mixtures of
polymerized anions may be present at a given composition, the proportions
of which vary with composition.
The discrete-polyanion model is a speculative one because no direct
proof of the existence of, e.g., its ring-polymeric anions is available. It
provides a much more consistent qualitative account of facts concerning
the behavior of liquid silicates than does the network model. Thus, it
predicts the observed marked changes in properties near 10% M 2 0 (Fig.
6.55), the relatively small variation in E" over the concentration range of
10 to 50% (Fig. 6.54), etc. The structural suggestions for the nature of the
anions receive some indirect support from solid-state structural .analyses
of certain mineral silicates.

TABLE 6.39
The Discrete Polyanion Model of Liquid Silicates

Range of composition,
Type of silicate entities present
mole % M.O

o Continuous 3-dimensional networks of SiO. tetrahedrons

with some thermal bond breaking and a fraction of SiO.
molecules

°
0-10 Essentially SiO. network with number of broken bonds
approximately equal to number of added atoms from
M.O, having fraction of SiO. molecules and radicals con-
taining M+
10-33 Discrete silicate polyanions based upon a six-membered ring
(SisO!;)
33-55 Mixture of discrete polyanions based on Si30~- and SisOYi'"
or Si40~. and SisOI.
~55-66 Chains of general form SinO~~n:121- , e.g., Si.O;-
66-100 SiO:- + 0'- ions
 IONIC LIQUIDS 615

6.7.9. The "Iceberg" Model

There are some facts, however, which cannot easily be rationalized
in terms of the discrete-ion model.
First, the partial molar volume of Si0 2 , related to the size of the
Si0 2-containing entities, is relatively constant from 0 to 33 mole % M 20
(Fig. 6.60). On the basis of the discrete-ion model, the critical change at
10% MxOy involving the breakdown of three-dimensional networks and
the formation of discrete-polyanions would require a decrease in molar
volume of Si0 2 at about 10% for the following reason. The Si0 2 structure
is a particularly open one and hence has a large molar volume; in contrast,
a structure with discrete polyanions would involve a closing-up of some
of the open Si0 2 volume and thus a decrease in partial molar volume
compared with the Si0 2 networks. Secondly, it is a fact that, in certain
ranges of composition, e.g., 12 to 33% M 20, M 20 and Si0 2 are not com-
pletely miscible. The two liquids consist of an Si0 2 phase and a metal-rich
phase. The discrete-anion model cannot accommodate this phenomenon.
It is suggested by White et al., therefore, that, in the composition range
of 12 to 33% M 20, two structures are present, one similar to that which
exists at 33% in the discrete-anion model. The other structure is a structure
corresponding to glassy, or vitreous, silica, i.e., fused silica with the random-
ized three-dimensional networks frozen in. The vitreous silica is in the form

30
I I
..e-ure SiO z"",,",,,
fm Innnm
0
(\I
26
SiOz
~
-
III

o~
Ol-
E I.. 22
'"
'0 '0
> E
.. If)
., E
'0 u
E .5 18

-.,..
~~

Q.

14

10 30 50 70 90
Mole percent

Fig. 6.60. The negligible change in the partial molar

volume of an Li 2 0-Si0 2 melt over the range of 0 to 33
mole % of Li.O.
616 CHAPTER 6

of "islets" or "icebergs." Hence, the name, the iceberg model of liquid-

silicate structure. These icebergs are similar to the clusters that occur in
liquid water (Section 6.7.2). The submicroscopic networks may be pictured
as continually breaking down and reforming. Microphase regions of
Mx Oy ·2Si0 2 (the 33% structure) occur in the form of thin films separating
the Si0 2-rich icebergs-hence, the possibility of a separation of the liquid
into two phases, one rich in Si0 2 and the other in MxOy. Since most of the
Si0 2 is present in the icebergs, the almost constant partial molar volume
of Si0 2 in the icebergs has the same Si-O-Si angle as vitreous silica.
An estimate of the order of magnitude of the iceberg size can be made.
For 12% M 2 0, the radius of an (assumed spherical) iceberg is about 19 A,
and, at 33% M 2 0, the iceberg of the iceberg model becomes essentially
identical with the discrete ion of the discrete ion model, which has a radius
of about 6 A.
In the iceberg model, the structure of the medium on a micro scale is
heterogeneous. Flow processes would involve slip between the icebergs and
the film. No Si-O-Si bonds need be broken.
At present, it seems that both the discrete-polyanion model and the
iceberg model probably contribute to the structure of liquid silicates. In a
sense, the iceberg model is the most complete model for it involves the
discrete polyanions as well as the Si0 2 entities called icebergs.

6.7.10. Fused-Oxide Systems in Metallurgy: Slags

Knowledge of what goes on inside fused-oxide systems is of importance

not only as a basis for future advances in glass technology (Section 6.6.4)
but also to metallurgical processes.
Consider, for example, one of the most basic processes in industry,
the manufacture of iron in the blast furnace (Fig. 6.61). Iron ore, coke,
and flux (essentially limestone and dolomite) are fed into the top of the
furnace. Compressed air fed in through openings in the bottom of the
furnace converts the carbon in coke to carbon monoxide, which reduces
the iron oxide to iron. Molten iron collects at the bottom. But, on top of
the molten metal is a layer of molten material called slag.
What is slag? A typical chemical analysis (Table 6.40) shows that it
consists mainly of silica Si02 , alumina A1 20 3 , lime CaO, and magnesia
MgO-in fact, precisely the kind of substances, i.e., nonmetallic oxides
(e.g., Si0 2 ) and metallic oxides (e.g., CaO), the structure of which was
under discussion here. (Some slags, in fact, might be regarded as molten
glasses.) The constituents of the slags are present in the ores and in coke.
 IONIC LIQUIDS 617

Throat

Air blast
"W..JI{1.~-Sla 9 ho Ie
Top hole-.......''''--':-~H

Fig. 6.61. Schematic of a blast furnace.

Successful operation of the furnace and production of the iron with

the desired composition (and, hence, metallurgical properties) depends so
much on making the slag with the right composition by controlling the raw
materials fed into the furnace that it is sometimes said: "You don't make
iron in the blast furnace, you make slag!"
Can one rationalize this importance of the slag? Measurements of
conductance as a function of temperature and of transport number indi-
cate that the slag is an ionic conductor (liquid electrolyte). But in the
metal-slag interface, one has the classical situation (Fig. 6.62) of a metal
(Le., iron) in contact with an electrolyte (Le., the molten oxide electrolyte,
slag), with all the attendant possibilities of corrosion of the metal. Corrosion
of metals is usually a wasteful process, but, here, the sequence of current-
balancing partial electrodic reactions which make up a corrosion situation

TABLE 6.40
Analysis of a Typical Slag

Constituents Fe Mn Si0 2 CaO MgO S

Percentage 0.34 1.18 34.67 14.58 44.78 3.21 1.36

618 CHAPTER 6

Eleclroly Ie (jonie conduclor)

Molten Iron Metal (electronic conductor)

Fig. 6.62. Molten metal in contact with slag is electrochem-

ically equivalent to a metal in contact with an electrolyte.

in fact are the factors which control the equilibrium of various components
(e.g., S--) between slag and metal and, hence, the properties of the metal,
which depend so much on its trace impurities. For example,

S[in metal] + 2e --+ S--[in slag]

Fe --+ Fe++ + 2e

The quality of the metal in a blast furnace is thus determined largely by

electrochemical reactions at the slag-metal interface.

Further Reading

1. W. Zachariasen, Z. Krist., 74: 139 (1930); J. Am. Chern. Soc., 54: 3841 (1932).
2. K. Endell and J. Hellbriigge, Naturwiss., 30: 421 (1942).
3. A. E. Davies, J. Chern. Phys., 19, 225 (1951).
4. J. O'M. Bockris and D. C. Lowe, Proc. Roy. Soc. (London), A226: 423 (1954).
5. J. D. Mackenzie, Trans. Faraday Soc., 51: 1734 (1955).
6. J. D. Mackenzie and J. A. Kitchener, Chern. Rev., 56: 455 (1956).
7. J. W. Tomlinson and J. L. White, Trans. Faraday Soc., 52, 299 (1956).
8. J. W. Tomlinson and M. S. R. Heynes, Trans. Faraday Soc., 54, 1822 (1958).
9. H. Bloom, Chapter 3 in: J. O'M. Bockris, ed., Modern Aspects of Electro-
chemistry, Vol. 2, Butterworths' Publications, Ltd., London, 1959.
10. E. Kojonen, J. Am. Chern. Soc., 82: 4493 (1960).
11. J. D. Mackenzie, Chapter 8 in: J. D. Mackenzie, ed., Modern Aspects of the
Vitreous State, Vol. 1, Butterworths' Publications, Ltd., Washington, D.C.,
1960.
12. H. Bloom, Chapter 1 in: B. R. Sundheim, ed., Fused Salts, McGraw-Hill
Book Company, New York, 1964.
13. J. L. Barton, Compt. Rend., 264: 1139 (1967).
14. L. I. Sperry and J. D. Mackenzie, Phys. Chern. Glasses, 9: 91 (1968).
 IONIC LIQUIDS 619

Appendix 6.1. The Effective Mass of a Hole

The pressure gradient in a fluid in the direction x, as a result of an

instantaneous velocity u in that direction, can be expressed as

Similar equations exist for the pressure gradients along the other two mutually
perpendicular axes y and z. In these equations, p is pressure; I] is density of fluid,
and 11 is its viscosity; u, v, and ware the instantaneous fluid velocities at the points
x, y, and z in the directions of the three coordinate axes; and X is the component
of the accelerating force in the x direction.
Stokes has shown that, in cases where the motion is small, the fluid is in-
compressible and homogeneous, etc., these equations can be simplified to a set
of three equations of the form

_dp = rJ (d 2U + d 2u + d 2U) _ I] du (A6.1.2)

dx dx 2 dy2 dz 2 dt

plus an equation of continuity

du dv dw
-+-+~=O (A6.1.3)
dx dy dz

Solution of Eqs. (A6.1.2) and (A6.1.3) in spherical coordinates leads to a

general expression of the form

Force = 2na I: (-P r cos 0 + To sin O)a sin 0 x dO (A6.1.4)

for the force of the fluid acting on the surface (r = a) of a hollow sphere oscillat-
ing in it with a velocity V given by ce iwt , where P r and To are the instantaneous
normal and tagential pressures at the points rand 0, c is the velocity at time
t = 0, and w is the frequency of oscillation. The term in To can be ignored for
present purposes since it corresponds to a viscous force acting on the sphere
owing to its motion through the liquid.
Inserting the appropriate expressions for P r' ignoring the terms arising from
the viscosity of the liquid since there can be no viscous slip between a liquid and
a hole, and proceeding through a number of algebraic stages yields
Force = -(ina 3 e)cwie iwt (A6.1.5)

where e is the density of the fluid. Writing now

m' = !na 3 1J (A6.1.6)
620 CHAPTER 6

for the mass of fluid displaced by the sphere gives

Force = - ( ~' ) cwie iwt (A6.1.7)

But, from the equation

dv .. .. t
- = X = Clwe'w (A6.1.8)
dt

and on remembering that, by Newton's law of motion, action and reaction are
equal and opposite, it follows that

Effective force due to spherical hole = ( ~' )x (A6.1.9)

This force thus corresponds to the force (m'/2)x which would be produced by a
solid body of mass m'/2 operating under conditions where the fluid is absent;
it is thus produced by a hole of effective mass m'/2 or ina 3 /2liq.

Appendix 6.2. Some Properties of the Gamma Function

The gamma function, r(n) is defined thus

r(n) = f~ e-tt n - 1 dt

Some of its properties are as follows:

1. When n = 1,
r(l) = f: e-tdt = 1

2. When n = !,

Put t = x', in which case

dt = 2x dx
ddt = 2dx
and
r(!) = 2 f~ e- x ' dx

V-;
Using Eq. (6.11), i.e.,

foo
o
e-ax2 dx = -1
2
-
a
one has
ru) = yn

3.
 IONIC LIQUIDS 621

Integrating by parts,

= nr(n)
Hence,
r(n + 1) = nr(n)

Appendix 6.3. The Kinetic Theory Expression for the Viscosity

of a Fluid
Consider three parallel layers of fluid, T, M, B (Fig. 6.63), moving with
velocities v + (iJvjiJz))., v, and v - (iJvjiJz))., respectively, where z is the direction
normal to the planes and ). is the mean free path of the particles populating
the layers, i.e., the mean distance traveled by the particles without undergoing
collisions. In the direction of motion of the moving layers, the momenta of
the particles traveling in the T, M, and B layers is m[v + (iJvjiJz)}.], mv, and
m[v - (iJvjiJz)).], respectively.
When a particle jumps from the T to M layers, the net momentum gained
by the M layer is mv - m(v + (iJvjiJz)).) = - m(iJvjiJz)).. If <w) is the mean
velocity of particles in the direction normal to the layers, then, in 1 sec, all particles
within a distance <w) will reach the M plane. If one considers that there are n
particles per cubic centimeter of the fluid and the area of the M layer is A cm 2,
then n<w)A particles make the T -+ M crossing per second, transporting a mo-
mentum per second of - [n<w)Am(iJvjiJz)).] in the downward direction.
When a particle jumps from the B to M layers, the net momentum gained
by the M layer is mv - m[v - (iJvjiJz)).] = + m(iJvjiJz)A, i.e., the momentum
transported per particle in the downward direction is -m(iJvjiJz)).. Hence, the

Fig. 6.63. Viscous forces arise from the

transfer of momentum between adjacent
layers in fluid.
622 CHAPTER 6

momentum transferred per second in the downward direction owing to B -)- M

jumps is -[n<w)Am(8vJ8z)}.].
Adding the momentum transferred owing to T ~ M and B ~ M jumps, it
is clear that the momentum transferred per second in the downward direction,
i.e., the rate of change of momentum, is -2n<w)mlc[A(8vJ8z)]. This rate of
change of momentum is equal to a force (Newton's second law of motion).
Thus, the viscous force p'J is given by
8v
p'J = -(2n<w)mlc)A 7h (A6.3.1)

But, according to Newton's law of viscosity, the viscous force is proportional

to the area of the layers and to the velocity gradient, and the proportionality
constant is the viscosity, i.e.,
8v
p'J = -l/A -.- (A6.3.2)
uz
From Eqs. (A6.3.l) and (A6.3.2), it is clear that
1] = 2nm<w)lc (A6.3.3)
INDEX
absolute mobility, and Einstein relation, 374 activity coefficient(s) (cont.)
absolute potential, impossibility of and ionic strength, 215
measurement, 645 mean, 207, 269
absolute potential difference model for effect of solvent molecule, 240
attempts to measure, 644 of single ionic species, 206
structured, 660 and solvation, 241
absolute strength of acid or base, 494 theory and experiment, 209, 213, 216, 228
absorption spectroscopy, in study of solutions, theory of solvent effects on, 242
275 ada toms, 1177
acceptor particles, need for in tunneling, 977 additives, organic, 1222
acid(s) adiabatic change, 974, 975
Bronsted's view, 489 adion(s)
dissociation constant of, 496 and adatoms, 1177
relative strength of, 495 charge on, 1177
acid-base strength(s), 501, 509, 511 concentration, determination of, 1197
acid strength, 506, 510 concentration profile of, 1200
acid strength theory, 505 concentration as function of time, 1189
action potential, and nerve impulse, 940 and electric field, 1181
activation, of electro catalysis, 1170 existence of, 1177
activation barrier, and rate-determining step, formation of hydrated ions from, 1187
1003 function in electrode position, 1177
activation energy, 457 lattices from, and spiral growth, 1203
for diffusion, related to melting point, 547 surface, deposition from, 1187
electrical contribution to, 872 tabulated, 1198
for multistep reaction, 1002 adsorbed intermediates, 1016
for viscous flow in fused salts, related to adsorption
melting point, 550 of atomic hydrogen, and coverage, 1246
activation-transport control, of electrode of cations and anions, at interfaces, 680
reaction, 1054 contact (or specific), 748
activity coefficient (s) definition of, 682
concept, 202 of desolvated ions, 742
cube-root law, 269 free energy change in, 742
Debye-Htickel model, 223 variation with potential, 638
Debye-Htickel, parameters of, 211, 212 in electrochemistry and chemistry, 683
further reading on, 246 organic, 792, 793
at high concentration, 245 and free energy of flip-flop water
and hydration number, 240 molecules, 799
and ideal solutions, 205 superequivalent, 638, 748
and ion pairs, 260 and surface excess, distinction, 684

xxxiii
xxxiv INDEX

adsorption energy, effect on tunneling, 965 barrier (s) (cont.J

adsorption intermediates, further reading on, width of, and probability of tunneling, 954
1036 base, 494, 495
adsorption isotherm, for ions, theory and batteries
experiment, 773, 774 in ancient times, 1265
adsorption step, in hydrogen evolution, classical, 1413
1153, 1233 and dendrites, 1221
aeration, differential, 1302 mission time in which useful, 1432
aggregation of colloidal particles, theory, 838 benzene, 1098, 1099
aims, of this book, vi biological cells, and charge transfer processes,
air 841, 1266
containing CO 2 , effect on metals, 1268 biological membranes, 937, 981
electrochemically burned, 13 82 biological processes, 4, 840
air electrode, in storage, 1427 biological reactions, quantum nature of, 21
alkali metals, and storage, 1423 biological situation, 941
alloy(s), 1224, 1225 biology, and electrochemistry, 29, 42, 43
American convention, 1118 Bjerrum approach, to ion pairs, 253, 257, 260
American space program, contributions to Bjerrum's integral, 256
fuel cells, 1388 blister, bursting, 1334
analogy (ies) blood, clotting of, and electrochemistry, 840
mass transfer and heat transfer, 1041 bond, stretching of, and electronation
semiconductors and electrolytic solutions, reaction, 971
812 bond strength, effect on electrocatalysis, 1157
angel, Gibbs', 679 Born charging process, 84
anode(s), 1311, 1312 Born cycle, 50, 86
anodic protection, 1318 Born equation, 57, 69
artificial organs, and electrochemistry, 43 and strength of acids, 510
association, of ions, temporary and Born model, 49, 50,70, 71
permanent, 273 Bridgman technique, for single crystals, 1218
atom(s) Bronsted's view of acids, 489
of hydrogen, recombination of, in void, bronzes, 1169
1334 Brownian motion, and movement of piston,
labeled, and surface coverage, 1036 300
atom-atom step, in hydrogen evolution, 1234 bulk properties, and interphase properties, 641
atomic energy bunching, 1209
and electricity storage, 1397
butanol
and electrochemical power sources, 1395
adsorption of, on mercury, 795, 798
and electrochemistry, 43
atomic power, coming era of, and Butler-Volmer equation, 862, 984, 1054
electrochemistry, 43 and biological situation, 941
autodissociation, proton transfer reactions in, and catalysis, 1141
492 and contact adsorption, 911
autodissociation constants, 499 deduction, 880
automobiles effect of water coverage?, 1015
necessary electric energy density for driving, and effective potential in catalysis, 1142
1420 and electric car, 928
exhaust products from, 1357 final form for multistep reactions, 1000
auxiliary electrode, and cathodic protection, further details, 910
1312 further reading on, 929
auxiliary electrode, use of in determining in galvanostatic transients, 1190
overpotential, 890 high-field approximation, 885, 888, 889,
averaging process, Debye, 141 1001
for iron dissolution and deposition, 1084
axon
as a cell, 937 low-field approximation, 882. 892
potential across, 938 for multistep reaction, 998, 1090
of squid, 940 in terms of stoichiometric number, 1006
Balkans, map of, 1430 and order of reaction, 1009
bands, bending of, near surface of in quantum mechanics. 980
semiconductor, 816 and rate-determining step, 1139
band picture, 804, 807, 810, 820 and structure of interface, 911
barrier(s) a summary, 928
for consecutive steps, 1002 and surface coverage, 1014, 1015
electron leak through, 945, 952, 962 and zeta potential, deduction, 913
transfer of charge across, further reading cadmium-nickel battery, 1401
on,946 . calomel electrode, 654
 INDEX xxxv

capacitance cel1 potential (cont.)

constant, equation for, 760 maximum, 1137
and contact adsorption, 751 minimum, 1137
dipole, 798 in self-driving cel1, 1131
of diffuse layer, 731, 732 cel1ulose, oxidation of, 1169
electrical, determination at interface, 703 central ion, excess charge density as function
and electrocapillary curves, 721 of,194
and Gouy-Chapman theory, 731 charge
hump, 754, 761 at boundary, 627
at interface, as function of dipole of individual ions, in diffuse layer, 747
orientation, 788 storage of, 703
of semiconductor-solution interface, 817 charge density
of whole electrode, 790 in double layer, components of, 712
capacitance-potential curve, lateral on electrode, determination of, 702
adsorption model, 778 excess, and ionic atmosphere, 636
capacitance hump, equation for, 761 charge transfer
capacitor(s) across barriers, further reading on, 946
and dielectric, 134 and blockage of electrode surface, 1014
and dipoles, 135 chemical and electrical implications, 846
in series, 735 and Fermi level, 977
capacity, constant region of, 753 and formation of intermediates, 1027
capillary electrometer, 689 and instability of surfaces, 1268
cars and metal-slag equilibrium, 618
and batteries, 1419 in perspective, 974
and fuel cel1s, 1158 quantification of quantum-mechanical
and pol1ution, 1419 picture, 977
carbon dioxide and rate-determining step, 1185
in atmosphere, as function of time, 1355 charge transfer catalysts, 10
and possible rise in sea level, 13 56 charge transfer theory, summary, 893
carbon monoxide-air cel1s, 1393 charged sphere, and Born model, 50, 52
Carnot limitation, 1358, 1360 charging process, effect of double layer on,
and electrochemistry, 1364 1190
and internal combustion, 1358 cheap heat, and electrochemistry, 43
physical interpretation, 1367 chemical desorption step, in hydrogen
catalysis evolution, 1234
Butler-Volmer equations in, 1142 chemical and electrochemical reactions,
and electrocatalysis, 987 further reading on, 989
for oxygen on doped tungsten bronzes, 1422 chemical energy, conversion to electrical, and
tabulated, 1377 symmetry factor, 1138
catalyst(s) chemical potential
charge transfer, 10 change arising from ionic cloud, 201
chemical and electrocatalysts, 1141 and computation at interfaces, 672
distribution of, and porous electrode, 1384 and flux of ions, 397
distribution of, in porous electrodes, 1172 standard, 203
catalytic activity and thought experiment, 694
for the oxidation of ethylene, 1161 chemical reaction
oxide-free and oxide surfaces, 1258 and electrical energy, 15
cathodic protection, 1309, 1312, 1313 and electrode reactions, 896
cavity, 150, 1335 chemistry, and electrochemistry, 28, 38, 869
cell chi potential, 667
driven, 1128 Christmas trees, mini, 1221
electric, potentials in, 649 chronopotentiometry, 1051
electrochemical, discussion of dependence circuitry, and electrochemistry, 43
on current, 1128, 1129 circuits, 33, 1316
entire, and Nernst equation, 904, 1114 circuits in the body, and electrochemistry, 43
local, in corrosion, 1270 civilization, and surfaces, 1267
and metal-metal potentials in cliff, and symmetry factor, 937
electrocatalysis, 1148 closest approach, 225, 741
for the observation of transients, 1184 clotting, of blood, and zeta potential, 841
relations in, further reading on, 1137 cloud, near central ion, 193
short-circuited, and stability of metals, 1269 clusters of ions, 82, 265
cel1 model, for liquid electrolytes, 529 coagulation, 839
cel1 potential codeposition, of hydrogen, with metal, 1227
and current, in electrochemical energy cold combustion, 1369
conversion, 1371 cold emission, of electrons, 944
xxxvi INDEX

colloids, 835, 838 conservation of momentum, and radiationless

colloid chemistry, further readings on, 841 transition, 950
collodial particle (s) constant-flux diffusion problem, and solution
energy of interaction for coagulation, 840 of other problems, 323
and potential distance relation, 729 constitution of proton, further reading on, 470
space charge near, 217 contact adsorption, 748
colloidal nature of biological processes, 4 and Butler-Volmer equation, 911
combustion and capacitance, 751
cold,1369 and capacitance of interface, 749
products of, other than carbon dioxide, definition of, 682
1353 of de solvated ions, 742
competition, between adsorbed water and determination of, 743
hydrogen, 1015 free energy change in, 742
complex formation, and mixtures of ionic as function of charge, 748, 763
liquids, 587 and image charge, 767
complex ion(s) and ionic radius, 744
concentration as a function of added lateral repulsion model, 766
ligands, 592 measurement of, 745
lifetime, 590 of negative ions, 637
in molten salts, further reading on, 594 and stability of colloids, 840
tests for, 589, 593,594, and surface state, 821
and Raman spectra, in molten salts, 590 tests for isotherm, 769
compressibility contact adsorption model, tests for, 775
calculated from hole model, for molten contact potential difference, 647
salts, 585 convection, 1051-1057
and hydration number, 127 convention
concentrated solutions, skepticism on theory American, 1118
of,281 international, 1118
concentration gradient, 1059 conversion, of chemical energy to electrocity,
and chemical potential, 291 14,42,1266,1358
and diffusion flux, 295, 1040 converters, photogalvanic, and
ion diffusion in, 288 electrochemistry, 43
linear, and Planck-Henderson equation, converter-storers, 1430
419 coordination number, of proton, 468
in tracer diffusion, 543 copper, deposition of, summary of
concentration overpotential, 1052, 1053 mechanism, 1202
concentration perturbation, Laplace correspondence principle, 20, 224
transform of, 329 corrosion, 1266, 1267
condenser as affected by solid phases, 1283
parallel-plate, and double layer, 634 and agitation of solution, 1300
and storage of charge, 703 basic kinetic conditions for, 1274, 1286
conductance bird's eye view of, 1347
in migration, further reading on, 367 common examples, 1301
in nonaqueous solutions, further readings cost of, 1346
on,452 effect of equilibrium potential, 1294
theory, further reading on, 439 effect of transport difficulties, 1294
theory of, in terms of Debye-Hiickel- effect of purity, 1273
Onsager equations, 435 effect of Tafel slope, 1294
conductance of true electrolyte, in and electrodics, 1272, 1275, 1285
nonaqueous solution, 452 embrittlement in, 1347
conduction, 345, 351 electrochemical mechanism, 1268
and flip-flop water dipole molecule, 790
conduction band, 809 and future of fuel cells, 1388
conductivity and hydrogen evolution, 1232
equivalent, 358 inhibition of, and electrodics, 1306
and ion association, 448 local cell theory of, 1270, 1273
molar, 357 Nernst equation and potential-pH
of molten salts, 517 diagrams in, 1279
of pure liquid electrolytes, 553 and oxygen electronation, 1252
specific, and current density, 354 potential of, 1285
conductor (s), 806 rate of , 1276, 1284, 1285
electronic, and passivation, 1320 rate-determining step, 1296
ionic, and effect on corrosion, 1273 and reversible potential, 1274
configuration, of water molecules around in sand, 1304
proton, 470 spontaneous energetics, 1278
 INDEX xxxvii

corrosion (cont.) current-potential relation (cont.)

summary of mechanisms, 1345 in terms of cell potentials, 1135
thought experiments in, 1272 current efficiency, 1229, 1232
through oxygen starvation, 1303 current transients, and Fick's law, 316
through paint, 1302 cybernetic organism, 1267
of ultrapure metals, 1273 cyborg, 1267
and undevice, 860 cytochrome
and wastage, 861 quantum mechanical tunneling to, 981
at water line, 1303 tunneling to, and enzymes as electrodes,
yield-assisted, 1337 1253
corrosion inhibition, and film-forming Daniell cell, 858
inhibitors, 1309 sign of voltage of, 1125
corrosion and passivity, further reading on, d-band character of metals, and oxygen
1349 adsorption on, 1163
costs, reduction of, in fuel cells, 1385 de Broglie wavelength, 20, 21, 948, 949
coupled reactions, 1235 Debye charging process, 248, 249
coverage Debye equation, for dielectric constant, 142
determination of, 1029 Debye theory of dielectric constant of gas, 140
of electrode, 1235, 1245 Debye-Hiickel activity coefficient, parameters
with inhibitors, tabulated, 1309 of, 211, 212
and mechanism determination, 1097 Debye-Hiickel constant, 190, 197
with hydrogen, determination of, 1245 Debye-Hiickellength
of organic molecules, on electrodes, 796 and diffiuse charge, 730
of surface with hydrogen, variation with in semiconductors, 816
potential, 1246 Debye-Hiickel model
crack(s), 1335-1341 approximations of, 219
crack initiation, testing of, 1343 breakdown of, 266
cracking Debye-Hiickel radius, of ionic atmosphere,
of hydrocarbons, 1158 197
stress corrosion, 1335, 1338 Debye-Hiickel theory, 180, 189
crevices, corrosion associated with, 1302 an assessment, 230
cross coefficient, 828 basis of, further reading on, 212
cryogenics, and storage of hydrogen, 1421 comparison with experiment, 212
crystal, growth of, and fast-growing face, further reading on, 238
1216 parentage, 237
crystal faces, rates of deposition on, theory postulates, 236
of,1216 summary of derivation, 233
crystal facets, 1212, 1213, 1214 triumphs and limitations, 212
crystal growth, faster at projection under Debye-Hiickel thickness, thickness of ionic
electric field, 1217 cloud in, 220
crystal growth, morphology of, for copper, Debye-Hiickel-Onsager equations, 434, 438
1222 comparison with experiment, 436
crystal lattice, 65, 1205 for nonaqueous solutions, 442
crystal lattice plane, kink site on, 1178 decay,4,861
crystal plane, 1179, 1213, 1216 decoration, surface, 1266
crystallization, 1174, 1202, 1204, 1218 de-electronation, 352, 847
crystalline solids, band theory of, 804 definition of, 853
cube root law, for activity coefficients, 269 and desorption of hydrogen, 1246
current-centric view, 16 effect of field, 875
current-distance relation, along meniscus in and quantum mechanics, 973
single pore, 1385 dehydration, and electrodeposition, 1176,
current-potential curve, and internal 1177,1180
resistance of fuel cells, 1373 delay time, and electronics, 33
current-potential diagram, for passivation, demon, Maxwell's, 679
characteristic, 1317 dendrites, 1220-1221
current-potential laws, 930, 1113 deposition
current-potential relation of alloys, 1223
and alloy composition, in oxygen reduction, of alloys, equations for, 1224
1259 and crystallization, 1202
and alloy deposition, 1259 metal, advance of growth step in, 1204
in cells, 1133 metal, and screw dislocation, 1206
characteristic shape in passivation, 1317 metal, steps in, 1203
for Gemini fuel cells, 13 89 metal, and nucleation, 1204
and mass transfer limitation, 1373 random walk process in, 1186
at Il-P junction, 936 on single crystal, 1218
xxxviii INDEX

deposition overpotential, and exchange diffusion (cont.)

current density, 1215 surface, deduction of equations for, 1188
desorption step, 1153, 1233, 1234 surface, and lattice formation, 1195
deuterons, mobility of protons and, 472 time in, by Einstein-8moluchowski
device equation, 333
electrochemical systems as, 851 diffusion coefficient, 296
electronic, 324 and critical concentration for hydrogen,
diagnostic coefficients, and propane 1328
oxidation, 1107 and Einstein relation, 374
diagnostic criteria, for de-electronation and Einstein-8moluchowski relation, 339
of ethylene on platinum, 1162 and holes, in molten salts, 577
dielectric constant, 55, 152 and mobility, 374, 376
of aqueous solutions, 157 and molecular quantities, 338
in bulk near ion, 71 rate process expression for, 342
Debye equation for, 142 and structural properties, 344
and deformation polarization, 145 of substances near melting point, 546
and dipoles, 139 tracer, 545
of electrolytic solutions, 157 diffusion control, 1046
gas, Debye theory of, 140 diffusion equation, and Nernst-Planck
and ionic solutions, 132 flux equation, 411
and ionic solvation, 155 diffusion flux
and oriented dipole layer, 136 and Kirchhoff's laws, 1039
and polarizability, 138 produced by sinusoidal variation of current,
and relative strength of acids, 507 328
and solvation sheath, 156
diffusion layer, 1055-1058
of solvent and solution, further reading
on, 158 diffusion layer thickness, and microrough
theory of, for water, 146 surface, 1219
in water, alignment of group, 146 diffusion potential
of water, calculation for, at various as a function of transport number, 418
temperatures, 154 further reading on, 420
of water, in double layer, 756 and transport number, 406
of water, theory of, 153 diffusion process, boundary conditions, 313
diesel oil, burned electrochemically, 1391 digestion, biochemical, 40
differential equation, for diffusion, dimerization, in liquid silicates, 610, 612
integrated,417 dipole
diffuse charge, and Debye-Hiickellength, 730 difference of contribution to potential
diffuse layer of flip and flop positions, 787
charge of individual ions in, 747 and electric field, interaction, 784
effect on Tafel relation, 915 interaction with electrode, 784
and streaming current, 830 orientation of, and capacitor, 135, 143
diffusion, 27 water, flip-flop model, 790
of adion to kink site, 1182 of water, in interior of electrolyte, 624
at constant current, quantitative, 1044 dipole-covered phase, potential difference
and convection, 1051 through, 667
driving force for, 290 dipole-dipole interaction potential, at
further reading on, 345 electrodes, 786
in fused salts, 542 dipole orientation, net, at interphase region,
of hydrogen by interstitial and 627
intergranular paths, 1330 dipole potential
of hydrogen into metals, 1328 and electrocatalysis, 1145
of hydrogen, into regions of stress, 1330 at interface, 667
linear, independent of time, 1077 at electrodes, 783
in molten salts, at constant temperature, direct energy conversion, 1358, 1359
548 and Carnot limitation, 1360
of neutral ion pairs, 383 by electrochemical means, 1360, 1361
nonsteady state, gross view, 307 further reading on, 1400
an overall view, 342 summary, 1398
to peak of crystal, 1219 discharge of ions, and dependence on
as pseudoforce, 306 lattice site, 1178
spherical, 1002, 1220 discrete polyanions, 610, 614
in solution, and electrocrystalIization dislocation, 1201, 1205, 1206
forms, 1219 dispersion forces, 166, 167
on surface, contribution to overpotential, dissociation constant, of acid, 496, 497, 498,
1199 500
 INDEX xxxix

dissolution edge vacancy, and electrodeposition, 1179

field-assisted, 13 37 effective mass, of hole, 619
of iron, intermediates in, 1087 efficiency, 1358-1371
of iron, mechanism for, 1085 e-i junction, law for, 936
of iron, prediction of various mechanisms, Einstein relation, and absolute mobility, 374
and experiment, 1092 Einstein-Smoluchowski equation, 333, 334
of kinky surface at bottom of crack, 1337 how many ions diffuse?, 333
of metals under stress, and Miller index, electric car
1335 and Butler-Volmer equation, 928
dissolution-precipitation mechanism, of and electrocatalysis, 1155
passivation, 1321, 1324 electric conduction, and liquid silicates, 606
distribution electric energy storage, needed for
of electrons, among energy levels in metal, automobiles, 442
956 electric field
time average, spatial, 182 effect upon electrocatalysis, 1168
distribution function, for holes in liquid effect on rate, 869
electrolyte, 537 influence on random walk, 350
distribution law local, in polar dielectric, 147
in Einstein-Smolchowski diffusion, 335 electric reactions, and electrochemistry, 14
for size of hole, 534 electrical energy
Doddario Committee, and John Malone's and free energy, 15
baleful prediction, 1357 production by thermal and electrochemical
donor, 499 means, 40
doping agent, 819 electricity
doping of bronzes, and electrocatalysis of from chemical energy, by electrochemical
oxygen, 1169 means, 1362
double layer conversion of energy to, and
charge density components, 712 electrochemistry, 42
charging of, 1027 from heat, 1360
concentration of reactants in, and work electricity storage
function, 1150 in alkali metals, 1423
constant capacitance of, model for, 758 and atomic energy, 1397
"constant" value for capacitance, 753 in hydrogen, 1420
effect upon electrocatalysis, 1012 important quantities in, 1404
electrical, becomes trouble layer, 750 electricity storage density, 1404, 1406
and Gauss's law, 727 electricity storer(s), 1402, 1412, 1413
interaction of, and stability of colloid, 837 future ones, 1420
ionic cloud at, 722 electrification, 6, 7, 623
isotherm for ions in, 764 electrified interface (s)
and parallel plate condenser, 634 absolute potential difference at, 675
and Poisson equation, 724 further reading on, 717
potential difference at, 635 importance of, in practical situations, 642
special position of mercury in studies of, mobile, electrokinetic properties, 826
687 retrospect and prospect, 715
structure of, further reading on, 790 structure of, 718
thickness of, and colloids, 835 thermodynamics of, 688, 698
double-layer charging, in galvanostatic electrocapillary curves
transients, 1190 basic equation for, 698
double-layer structure, further reading on, and capacitance, 721
717 differentiation of, 704
double-layer studies at solids, further facts on, 690
reading on, 803 as perfect parabola, 705
double-layer theories, review of, 752 and surface excess, 710
double-layer treatments, history of, 724 electrocapillary equation, final general form,
drift, ionic, to interface, 391, 845 701
drift velocity electrocapillary maximum, 691
calculation of relaxation components, 427 electrocapillary thermodynamics, 713, 714
electrophoretic components of, 427 electrocatalysis, 1141
and interacting ions, 425 activation in, 1170
relaxation component of, 430 and cancellation of thermionic work
driven cell, 1128, 1131 function, 1147
driving force, 343,412 and cars, 1155
droplets, corrosion under, 1305 and chemical catalysis, 987
economics, and social importance of of copper deposition, on various surfaces,
overpotential, 16 1202
xl INDEX

electrocatalysis (coni.) electrochemical era, 43

dependence on electronic properties, 1147 electrochemical generators, 1386
difference from chemical catalysis, 1143 examples of, 1385
and dipole potential, 1145 electrochemical methods, for surface
effect of double layer on, 1012 coverage, 1035
effect of metal-metal potentials, 1148, 1155 electrochemical model, for slag-metal
and exchange current density, 1146 equilibrium, 618
and Galvani potential difference at electrochemical potential (s)
nonpolarizable electrodes, 1149 digression on, 693
heat of activation in, of nonbonding equality of, in different phases, rationalized,
reactions, 1149 864
of hydrocarbons, 1156, 1158 gradient of, 395
and hydrocarbon oxidation, 1391 and work of bringing charge particles into
of the hydrogen evolution reaction, material phase, 695
1155,1232 electrochemical producer, 1361
irrelevance to, of experiments with porous electrochemical reaction (s), 7
electrodes, 1376 and chemical reactions, 8, 987
lack of effect of work function on, 1148 always quantal, 985
of oxygen, and doping of bronzes, 1169 electrochemical reactor(s), 9,1405,1406
potential of, comparison, 1145 electrochemical system (s)
and potential of zero charge, 1142 as devices, 851
rate equations for, compared with those and metal-metal potential, 1113
for catalysis, 1143 series of potential drops in, 1112
in reactions involving adsorbed species, electrochemical vista, 43
1153 electrochemist (s)
reactivity at low temperatures in, 1169 frustrations of, 1042
of redox reactions, 1149 training of, 44
reference potential for, 1143, 1144 electrochemistry
secondary effects due to double layer, 1151 advances expected, 42
secondary effect of work function on, 1151 and artificial limbs, 43
and simple redox reactions, 1146 and atomic reactors, 43
in space vehicles, 1158 awakening, 24
special position of platinum in, theory of, as basic science for advances in
1166 postindustrial era, 42
tunneling condition for hydrogen evolution, and biology, 29
1154 brilliant beginning, 14
volcano relations in, 1165 and cheap heat, 43
electrocatalyst, determination of adsorbed and chemistry, 28, 869
entities at, 1030 and circuits in the body, 43
electrochemical cells, 1114, 1132 and coming era of atomic power, 43
electrochemical converter (s) and conversion of energy to electricity, 42
Carnot efficiency limitation avoided in, conversion of, to charge transfer
1364 orientation, 17
efficiency, 1364, 1366 delay in development of, 18
electrochemical desorption, in hydrogen and development of circuitry, 43
evolution, 1234 and developments in molecular biology,
electrochemical device (s) 42,43
as energy producer, 855 and direct energy conversion, 1361
as substance producer, 851 disciplines in, 27
electrochemical electricity storers, 1412 and electric reactions, 14
electrochemical energy conversion and electronics, 18,27,33
and atomic energy, 1395 and electron transfer, 25
its central problem, 1369 future role, 41
dominating role of electrocatalysis, 1372 and geology, 29
and Tafel relation, 1370 an interdisciplinary area, 1,25,29, 31
electrochemical energy converters and interfaces, 22, 23
cost of, and porous electrodes, 1385 interfacial, degree of ubiquity, 39
deduction of real efficiency, 1366 involvement of, in many sciences, 28
and power-rate relation, 1378 and machining, 42, 43
electrochemical energy producer, and power and medical developments, 43
density, 859 need for books, vi
electrochemical energy storage and new towns, 43
feasible goals, 1431 and other fields, 25
summaryof,1430 perspective from afar, 23
electrochemical engine, 1157 perspective from a medium distance, 24
 INDEX xli

electrochemistry (cont.) electrodics (cont.)

perspective in time, 39 elementary, further reading in, 909
and photogalvanic converters, 43 elementary, summary, 908, 983
place in science, 26, 41 and inhibition of corrosion, 1306
and polluted liquids, 43 and quantum mechanics, 30
and possible fuel cell heart, 43 transient techniques in, 34
and powering of ships, 43 transport aspects of, summary, 1076
and powering of vehicles, 43 in the west before 1950, 1380
quantum nature of reactions in, 20, 21 e1ectrodissolution, burst of, and Laplace
as separate discipline, 38 transformation, 330
and sewage disposal, 43 electrogrowth,1215
sign convention in, 1115 basic aspects of, 1173
and stabilization of materials, 42 and kink sites, 1203
and storage of energy, 42 topographical features, 1184
and time, 38 electrokinetic properties, 826
and tools, 43 further reading on, 835
and transportation, 43
electrolyte(s)
and urban living, 42
forces at boundary of, 623
and water purification, 42
wider significance?, 38 glasses as, 603
electrocrystallization, 1129, 1173, 1174, 1219 potential and true, conductance in
electrode(s) different media, 179
pure liquid, 513
as catalyst, 34, 1139
true and potential, 176
de Broglie wavelength at, 20
porous, 1171 electrolytic solutions
activity near tip of meniscus, 1384 electromagnetic radiation in study of, 274
diffusion of reactant in, 1384 and infrared spectroscopy, 278
and distribution of catalyst, 1384 and nuclear magnetic resonance
vital importantance in fuel cells, 1172 spectroscopy, 278
sick, 1231 and Raman spectra, 277
electrode kinetics and semiconductors, 811
and double-layer effects, 916 electromagnetic methods for investigating
and organic reactants, 916 solutions, further reading in, 279
transfer coefficient as center of, 918 electromagnetic radiation, in study of
and zeta potential, 912 electrolytic solutions, 274
electrode processes, quantum-mechanical electromagnetic theory of light, 35
approach to, 947 electron(s)
electrode reactions cold emission of, 944
and heterogeneous reactions, 989 collision with impurity atom, 956
history of quantum-mechanical distribution among energy levels in metal,
developments, 983 956
and tunneling, 955 in holes, and the double layer, 825
electrode surfaces, 36 image energy of, as function of distance
electrode-electrolyte potential difference, from metal, 945
analysis of, 659 leak through barrier, 945
electrode posits, organic, 1222 their mechanics, 947
electrodeposition near interfaces, potential of, 943
consecutive stages in, 1180 number of which strike surface of metal,
and dehydration, 1180 990
electronation in, 1176 penetration into forbidden region, 950
function of ad ions in, 1177 in space region, and wave function, 669
and hole vacancy, 1179 and tunneling to solution, 959
influence of potential of zero charge, 1180 in vacuum, and probability of passage
of metals, and tunneling, 1177 through,952
rotation of a spiral in, 1205
which are free to tunnel, 959
stepwise dehydration in, 1177
and surface adions in random walk, 1181 electron overlap potential difference, 670
and surfaces which change with time, 1182 electron sink, 853
in terms of consecutive reactions, electrode, 1126
schematic, 1183 electron source electrode 853, 1126
elecrodeposition rate, as a function of crystal electron transfer
plane, 1216 and electrochemistry, 25
electrodics, 19, 846 probability for tunneling, expression for,
and corrosion, 1285 978
and electronics, 30, 31, 32 type of, in electrochemistry, 20
 xlii INDEX

electron transfer reactions, 494 energy converter(s) (cont.)

and electroneutrality difficulties in power and efficiency in, 1379
conduction, 351 thermionic, 1359
the 1950's, 17 thermoelectric, 1359
electron transfer theory, beginnings, 18 energy density, 1407, 1408
electron tunneling, 946 idealized maxima, tabulated, 1408
condition for, 972 and rate of working of cell, 1410
electronation, 352, 847 of stores, feasible values, 1418
in corrosion, 1275 versus power density, 1411
of oxygen, 1251 energy gap, 810, 931
and enzymes, 1253 energy levels, 805, 807, 956
electronation reaction energy producer, 855
and bond stretching, 971
energy-producing device, current and
effect of field, 874
potential in, 1131
of hydrogen on platinum, energy terms in,
968 energy sources, 1350
electroneutrality distribution of, 13 51
conflict with conduction, 351 energy states
and coupling between ionic species, 410 discrete, 806
in fused salts, 543, 566 those accounting for free electrons, 959
principle of, 414 energy storage, 1266
electronics, 18,27, 30, 31, 32, 33, 43 electrochemical, summary, 1412, 1430
electro-osmosis, 826 terminology, 1402
theory of, 827 energy storage density
electro-osmotic motion, of phases relative to and non-aqueous electrolyte, 1427
each other, 831 for some realized cells, 1409
electrophoresis, 832, 834 energy storers, 1420, 1426, 1428
electrophoretic effect, 424, 425 energy waster, 859
electrostatic potential enthalpy
and charged sphere, 52 by Born, and ionic radii, 60, 69
and field strength, 347 and electrochemical energy conversion,
near ion, as function of distance, 193 1368
variation of, near interface, 664 and entropy, of ion-solvent interaction, 59
electrostriction, 126 entropy, and electrochemical energy
ellipsometric spectroscopy, 37 conversion, 1364, 1368
ellipsometry, 37,1319 enzymes, 1253
embrittlement equilibrium
in corrosion, 1347 difficulty of observing rates near to, 1263
by hydrogen, 1314, 1338, 1339, 1344 and electrochemical potential, 696
emission at interface, 876
cold,944,952,955 and Nernst equation, 898
hot, of electrons, 941 and steady state, 1018
thermionic, 953 equilibrium cell potentials, useful ?, 1124
empty space, in fused salts, 523 equilibrium potential, 876
energetics, of certain corrosion reactions, 1278 and activity in solution, 905
energy (ies) limitation in usefulness, 876
of activation, for self-diffusion, 579 equivalent circuit(s)
of crack, 1341 for galvanostatic transients, 1026
electrical, 11 for interface involving ideally polarizable
free and electrical, 15 electrode, 654
as function of repulsion between water and pseudocapacitance, 1029
molecules, 970 equivalent conductivity, 358
of interaction between colloid particles, 838 and concentration, 360
of strain, 1340 and ionic mobility, 372
energy barrier, and rate theory, 341 significance of, 360
energy consumption, electrochemical, 1350 era, electrochemical, 43
energy conversion, 1266, 1358, 1361 error function, 321, 1065
to electricity, and electrochemistry, 42 Esaki tunnel diode, and tunneling, 956
as an interdisciplinary field, 31 ethylene
and storage, terminology of, 1402 adsorption of, on platinum, 1160
energy conversion efficiency, maximum of, rate-determining step in the oxidation of,
1376 1159,1160
energy converter (s) ethylene oxidation
electrochemical, efficiency of, 1374 and catalytic activity, 1161
 INDEX xliii

ethylene oxidation (cont.) film(s) (cont.)

diagnostic criteria for de-electronation on passive, formation of, at the bottom of
platinum, 1162 pits, 1335
negative pressure effect in, 1160 precursor, 1319
radiotracer measurements of coverage, film-covered surfaces, determination of
1160 properties by ellipsometry, 37
rate-determining step, 1160, 1161, 1164 Flade potential, 1316
volcano relation in, 1164 flash photolysis, 34
European convention, 1118 flip dipoles, at electrodes, lateral
evolution, of gases, 1102, 1104 interaction between, 785
excess charge density, as function of distance flip-flop model, 790, 797
from central ion, 194 flip-flop water on dipoles, 779
exchange current density, 876, 877 flocculation, 839
catalytic effects due to double layer flux
properties, 1151 and forces, 894
and deposition overpotential, 1215 as a function of field strength, 350
determination of, for hydrogen evolution, of ions, and chemical potential, 397
1238 of ions, and electrostatic potential, 397
and electrocatalysis, 1146 at low field gradient, 295
and heat of activation, 1150 sinusoidally varying, near electrode-
for hydrogen evolution, tabulated, 1238 solution interface, 327
for metal deposition, 1202 time derivitive of, in step function, 330
on noble metals, and platinum-rhodium flux equality condition, 1038
alloys, 1260 flux equations, and Onsager equations, 411
and polarizability, 895 forbidden region, electron penetration into,
and rate-determining step, 997 950
and reaction order, 1012 forces
schematic diagram of, 878 anisotropic, 626
small, difficulty of measurements with, at boundary of electrolyte, 623
1260 and fluxes, 894
for various crystal faces, 1215 in organic adsorption, at electrodes, 792
and work function, 1149 fossile fuels
exclusion principle, 963 available for centuries if used mainly for
expansivity food and textiles, 1350
calculation from hole model, 585 converted to food and textiles, 1358
of liquid silicate models, 609 lack of rationality of burning up, 1352
faces, fast growing, 1217 Fourier's law, 894
facets, 1213, 1214, 1217 fraction, of ions, produced in pulse, near
faradaic rectification, 885 electrode, 332
faradaic resistance, 996 free electrons, energy states for, 959
fatigue of metals, 1267 free energy
Fermi-Dirac distribution, deduced, 958 change of, when ion goes from OHP to
Fermi energy, 959, 962 IHP,763
Fermi level, 957, 964, 977 determination by chemical and
Fick's first law, 293,315,316,343, 1056, electrochemical means, 40
1065 and electrical energy, 15
Fick's second law, 308, 344, 1040 of ion-ion interactions, 181
field of organic adsorption, at electrodes,
current produced, 883 792, 796
in double layer, 630 free energy of activation, as function of
excess, 882 potential, 390
induced reorientation, 484 free energy change
at interphase, 630 in contact adsorption, 742
nonlinear, 348 in ion-solvent interactions, 49, 50
producing current, 882 free energy level, of proton, 500
in semiconductors, 814 frog, electrical movement of its nerve, 11
field strength Frumkin, effect of leadership in Russia on
effect on current, 881 electrochemistry throughout world, 18
and electrostatic potential, 347 fuel cell
and flux, 350 catalyst, 1383
film(s) first one, 1380
electronic conductivity of, and future of, and corrosion, 1388
passivation, 1320 history of, 1386, 1387
xliv INDEX

fuel cell (coni.) Gouy-Chapman model, and potential

immediate uses of, 1396 dependence of capacitance, 732
mission time in which useful, 1432 Gouy-Chapman region, and colloidal
practical applications of, 1396 stability, 839
and production of water, 1420 Gouy-Chapman theory, and capacitance, 731
as source of drinking water, 1389 grains, 1218
vital importance of porous electrodes in, greenhouse effect, 1356
1172 Griffith crack, 1340
fuel cell heart, possible electrochemical group dipole, 147
development, 43 growth step, 1210, 1211
fuel cell research, funding of, and pollution, Guntelberg charging process, 248
1386 happenings, thermal and electrochemical,
fuels, and electrocatalysis, 1387 alternative versions, 40
Fuoss approach, to ion pairs, 261 heart, artificial, possible electrochemical
Furth approach to work of hole formation, development, 43
536 heat of activation
fused oxides, as slags, 616 change of, with electrode potential, 924
fused salts (see also molten salts) in electrocatalysis, of nonbonding
and activiation energy for diffusion, 547 reactions, 1149
and activation energy for viscous flow, 550 and exchange current density, 1150
atomistic theory of transport, 574 for flow in silicate melt, 605
diffusion in, 542 for proton transport, in aqueous
empty space in, 523 solutions, 473
holes and diffusion coefficients in, 577 and temperature, in proton mobility, 486
heat of activation for viscous flow, 551 for viscous flow, and melting point, 551
internuclear distances in, 524 heat to electricity conversion, 1360
radiotracer method for transport number heat engine, essence of working of, 1352
determination, 571 heat of hydration
and self-diffusion, 542 ofhydrogenion,105,467,468
Stokes-Einstein relation in, 552 by quadrupole model, 101
transport, and holes, 574 relative, as a function of radius, 97
transport numbers, Stokes' law approach of transition metal ions, as a function of
to, 572 atomic number, 112
viscous forces and momentum in, 575 of transition metal ions, and water
Galvani potential, thought experiment stabilization energy, 113
synthesis of, 672 heat of solution, 65, 67
Galvani potential difference, 670 heat of solvation, 63, 66, 88, 96
at non polarizable electrodes and Helmholtz and Gouy capacities, in series, 736
electrocatalysis, 1149 Helmholtz-Perrin model, 718
galvanostatis rise time, 1193 Helmholtz-Perrin and Gouy-Chapman,
galvanostatic transient, 1021, 1185 relative contributions, 737
and Butler-Volmer equation, 1190 Henderson-Planck equation, solution for,
and double-layer charging, 1190 459
equations for, 1024 heterogeneity, surface, on solid electrodes,
galvanostatic transient method, for surface 803
coverage, 1030 Hiatus, the Great Nernstian, 16
Garrett-Brattain space charge region, 812 history of double-layer treatments, 724
Gaussian box, 728
hole
Gaussian surface, 137, 151
average size of, 539
gels, 839 concept of formation of, 528
Gemini fuel cells, used in space, 1388 and diffusion coefficients, in fused salts, 577
generators, electrochemical, 1385, 1386 effective mass of, 619
geology, and electrochemistry, 29 formation of, in valency band, 810
Gibbs, his angel, 679 lifetime of, in fused salts, 576
Gibbs' surface excess, 680, 683 in liquid electrolyte, size of, 537
glasses, 603 and transport in fused salts, 574
electrolytic structures of, 603 viscosity in terms of, 577
as ionic liquids, 603 hole current, and electron current, 933
liquid silicate as, 603 hole mobility, values of, 931
goals, in storers, 1420 hole model, 527
Gouy-Chapman and Helmholdtz-Perrin, and compressibility, 584
relative contributions, 737, 822 and expansivity, 585
 INDEX xlv

hole model (cont.) crystallization, 1129

for liquid electrolytes, further reading on, and change of mechanical properties, 1129
541 diffusion into
most consistent model at present, 584 metals, 1328
normalizing conditions, 536 regions of stress, 13 30
probability of finding hole of radius r, 535 diffusion by interstitial and intergranular
and rationalization of relation of heat of paths, 1330
activation to melting point, 582 effects, and passivation, 1349
hole motion, and electron motion, 811 electricity storage in, 1420
hole vacany, and electrodeposition, 1179 initiation of cracks by, 1335
Hooke's law, 1331, 1334 and instability of metals, 1338
hot emission of electrons, 941 kinetics of discharge of, favorable
hump effects on storage of hydrogen, 1421
of capacitance, 754 partial molar volume of, 1331, 1332
experimental, in capacity-charge curve, penetration into bulk of metal, 1329
762 pressure of in metals, 1333
storage of, cryogenics, 1421
hydrated ions hydrogen-air battery, 1421
distance of closet approach to electrodes, hydrogen-oxygen cell, 1388, 1390
741 hydrogen adsorption, atomic, on electrodes,
formation from adions, 1187 and coverage, 1246
hydration, 80 hydrogen bond (s)
calculations involving quadrupoles, 104 and proton mobility, 468
effect of ligands, 109 in solvation process, 90
and orbitals, 110 and clusters of water molecules, 88
of transition metal ions, 106 hydrogen coverage, 1245
of transition metal ions, and stabilization of surface, variation with potential, 1245,
of field, III 1246
hydration number(s) and various mechanisms, 1247
activity coefficient as function of, 240 hydrogen codeposition, 1227, 1229
of alkali and halide ions, by independent hydrogen desorption, and de-electronation,
methods, 118 1246
and compressibility, 127 hydrogen embrittlement, 1314, 1338
primary, table of, 131 hydrogen evolution
and radius, 130 adsorption step in, 1153, 1233
by various methods, 118 atom-atom step in, 1234
hydration of proton, heat of, 467 chemical desorption step in, 1234
hydrazine-oxygen cells, performance and corrosion, 1232
tabulated, 1390 and current efficiency, 1232
hydrazine-oxygen fuel cells, 1389 deduction of values for transfer
coefficient, 1241
hydrocarbons
catalysis of and cars, 1158 desorption in, 1153, 1234
cracking of hydrocarbons, reforming of determination of path and rate-determining,
hydrocarbons, 1158 step, 1237
electrocatalysis of, 1158 determination of transfer coefficient by
reformed by steam, 1390 various means, 1238
reforming of, 1391 and electrocatalysis, 1232
saturated, 1391 equations for various mechanisms, 1240
mechanism determination of, 1107 exchange current density for, tabulated,
mechanism of oxidation of, 1107 1238
rate-determining step in the de- further readings on, 1250
electronation of, 1110, 1158 general, 1231
unsaturated, electrochemical data history of, 1231
concerning, 1392 ion-atom recombination, 1234
hydrocarbon-air cells, 1391 mechanisms, 1235
hydrocarbon fuels, how used?, 1351 on metals, equations for, 1228
hydrocarbon oxidation, 1158-1159 paths, 1233
hydrodynamic flow, and convection, 1050 reaction paths, 1235
hydrogen recombination mechanism in, 1234
accumulated, in cracks inside metal, 1339 and separation factor, values for, 1249
accumulation of, at regions of stress, 1333 tabulated summary of probable
adsorbed, and stability of metals, 1328 mechanisms, 1250
adsorption of, and change the path of and transfer coefficient, 1102
 xlvi INDEX

hydrogen evolution (cont.) interface(s)

tunneling conditions for, and electro- accumulation of substances at, 679
catalysis, 1154 adsorption of ions at, 738
hydrogen ion, trigonal pyramid structure of, affected by image forces, 662
466 bird's eye view of structure, 824
hydrogen overpotential, experimental creation by thought ':'l(periment, 5
characteristics, 1232 current-potential laws at, 930
hydronium ion, existence of, 487 dipole potential at, 667
proton mobility in, 486 electrified
and water, structure of, 76 importance of, 642
iceberg model, for liquid silicates, 615 retrospect and prospect, 715
I1kovic's equation, 1068 review, 716
image charge, and contact adsorption, 767 structure of, 718, 632
image energy thermodynamics of, 688
of electron, as function of distance from electron potential near, 943
metal,945 examination in terms of transients, 1026
its place in charge transfer kinetics, exchange of electrons through moisture
960 films, 6
andtunneling,961 ionic drift to, 845
image forces, 661-662 at metals, other than mercury, 801
and quantum mechanical tunneling, 960 metal-solution, and Volta potential, 665
image interactions, with charged electrodes, nonpolarizable, 697, 701, 894
660 polarizable, 653, 700, 894
impedance, 896 potential differences at
high, and measurement of overpotential, further readings on, 687
897 and surface tension, 688
and measuring circuit, 896 profile of concentration at, in adsorption,
impurity atoms, collision of electron with, 681
956 semiconductor-electrolyte, 803
impurity conduction, in silicon and structure of, and Butler-Volmer equation,
germanium, 930 911
inclusions, microscopic, of copper, effect on thermodynamic deduction of surface
reaction rate, 1272 excess equations, 700
indifferent electrolyte, 1060, 1069 two-way electron traffic across, 873
induction time, in passivation, 1322 under transient conditions, 1017
infrared spectroscopy, and electrolytic water molecules in oriented layer at, 633
solutions, 278 interfacial tension
inhibition, of corrosion, and electrodics, and applied potential, 701
1306 measurement of, 688
inhibitor(s) intermediates
coverage, tabulated, 1309 adsorbed, 1016, 1029
adsorption, 1310 determination of, in benzene oxidation,
practical examples of, tabulated, 1311 1099
initiation, of cracks, 1335 in dissolution of iron, 1087
inner potential, 673 and potential-time transients, 1026
inner potential difference formed by charge transfer, 1027
between dissimilar phases, 677 and propane oxidation, 1110
between two identical phases, internal combustion engine, 1358
measurability of, 679 wasting fuel as heat, 13 52
instability internal resistance of fuel cells, effect upon
of metals, 1338, 1343 current-potential curve, 1373
of surface, and internal decay, 1335 internuclear distance, in solid and liquid
instantaneous pulse, 329 fused salts, 524
instruments, high impedance, need for, 896 interphase, 2, 630
instrumentation, for potentiostatic . surface excess at, 683
transients, 1033 interphase properties, and bulk properties,
interaction(s) 641
ion-ion compared with ion-electrode, 723 interphase region, 626, 627
metal-water, 740 ion(s)
minimal, between dipoles at electrodes, 785 adsorbed, on kink sites, and passivity, 1325
interaction energy, and orientation of association of, temporary and permanent,
dipole near ion, 81 273
interatomic spacing, 807, 808 clusters of, 265
 INDEX xlvii

hydrated, distance of closest approach to ionic cloud (cont.)

electrodes, 741 and Debye-Hlickel constant, 200
in sheath of water molecules, 77 further reading on, 202
ion-atom combination, in hydrogen evolution, relaxation of, 423
1234 shape of, 431
ion-cloud theory, 180 ionic drift
ion-dipole interaction(s), 83 interdependence of, 399
equations deduced, 169 to interface, 845
ion-dipole model, of solvent interaction, 80 ionic fluxes, and Onsager equations, 411
ion-dipole theory, of solvation, evaluation, ionic groups, in silicates, 596
93 ionic liquids, 513
ion-electrode interactions, 723 further reading on general aspects of, 522
ion-electrode and ion-ion interactions, 723 and glasses, 603
ion-ion interaction(s), 175 lattice-oriented models for, 522
and activity coefficients, 202 and liquid silicates, 603
and Debye-Hlickel theory, 725 mixture of, and complex formation, 587
free energy of, 181 and Nernst-Einstein relation, 555
parentage of theory of, 273 slags, 617
in perspective, 279 ionic migration, 367, 387, 420
ion-quadrupole theory, of solvation, 103 ionic mobility, 401
ion-size parameter, 224, 225, 230 ionic movement, as function of random walk,
ion-solvent interaction, 49 299
and cavities, 81 ionic product, 499
effect on activity coefficient calculation, and semiconductors, 819
238 ionic radius, and contact adsorption, 744
and effect of quadrupole theory, 115 ionic solutions, and dielectric constant, 132,
equations for, 59 155
experiment and theory, 94 ionic solvation, 155
free energy change in, 49, 50, 57, 58 ionic species, single
further reading on, 116 activity coefficient of, 206
heat of, and thermodynamic cycle, 66 ambiguity of measurement of properties,
improvement, by quadrupole model, 100 64
of individual ions, 114 ionic strength, definition, 210
quadrupole model of, 99 ionic transport, and electrochemical potential,
summarizing remarks, 113 395
thought experiments in, 80 ionics, 19, 846
ion-solvent-nonelectrolyte interactions, 158 analogy to behavior of electrons in holes,
ion-water interactions, quadrupole theory, 32
171 and charge transfer, 1036
ion association, 447 definition of, 3
and conductivity, 448 rise and fall, 24
ion association constant, 257, 259 ionization potential, and energy released
ion migration, as function of electrostatic from electronation of proton, 967
potential gradient, 288, 289 iridium, 1253, 1255
ion pairs iron
and activity coefficients, 260 mechanisms of dissolution of, tabulated,
Bjerrum approach, 260, 263, 264, 265 1091
Fuoss approach, 261 potential of, as affected by solid phase,
further reading on, 266 in corrosion, 1283
and triple ions, 265 potential-pH diagram for, 1280
ion pair formation, 251, 255 irradiation, ultrasonic, and electrocatalysis,
ion pair fraction, 258 1170
ion size parameter, skepticism, 280 isotherm
ionic association, 251 for adsorption of oxygen on platinum, 1257
ionic atmosphere, 428, 636 contact adsorption, tests of the value of,
and ionic migration, 420 770
radius of, Debye-Hlickel, 197, 199 deduction of, for contact adsorption, 768
and variation with potential, 636 for ions in the double layer, 764
ionic cloud lateral repulsion model, discussion, 777
asymmetric, 422 and organic adsorption, 797
catching up with moving ions, 422 tests for, 771, 772
chemical potential change arising from, isotopes
201 radioactive, self-diffusion, 542
xlviii INDEX

isotopes (cont.) liquid electrolytes

substitute, dependence of reaction rate on, cell model for, 529
1106 distribution function for holes in, 537
IUPAC sign convention, in detail, 1119 further reading on, 533
journals, international, in electrochemistry, v gas-oriented models for, 529
kilowatt hours per kilogram, 1407, 1410 hole model for, further reading, 541
kinetics, of hydrogen discharge, and storage, hole size in, 537
1421 liquid free volume model for, 530
kinetic theory, for viscosity, 621 summary of models for, 532
kink sites transport numbers in, 566
adsorbed ions on, and passivity, 1325 liquid free volume model
diffusion of adion to, 1182 for pure liquid electrolytes, 530, 531
and electrogrowth, 1203 liquid oxides, 594
Kirchhoff's law, and diffusion flux, 1039 liquid silicates
Laplace transform dimers in, 610, 612
for concentration perturbation, 329 and electric conduction, 606, 609
of constants, 454 as glasses, 603
in diffusion problems, 1041 model of discrete anions, 610
and Fick's second law, 344 model of icebergs, 615
use in polarography, 1063 network model of, and table, 606
of pulse of flux, 331 and polymerization, 611
Laplace transformation structure breaking in, and concentration
and burst of electro dissolution, 330 of metal ions, 604
definition of, 310 ring anions in, 613
explanation of, 309 ring formation in, 612
and Fick's law, 315 as slags, 616
initial and boundary conditions for local cell, diagrammatic, 1271
diffusion process stimulated by local field, 147
constant current, 313 calculation of, 148
use in Fick's second law, 312 logistics, 1036
and transients, 1041 Los Angeles, and the smoggy state, 1354
lattice approach, to concentrated solutions, Luggin capillary, 891
266 machining, and electrochemistry, 42, 43, 1267
lattice dislocations, and spiral, 1206 macrostep(s), 1207, 1208
lattice energy, and heat of solution, 67 and bunching, 1209
lattice formation, after surface diffusion, 1195 and irregular edge, 1209
layer growth, 1208, 1222 magnetohydrodynamic conversion, 1360
leveling, 1222 map of the Balkans, 1430
life, and oxygen electronation, 1252 mass of hole, effective, 619
lifetime materials
of complex ions, in molten salts, 590 decay of, 4
of proton in solution, 485 dependence of properties on surface, 3
ligands, effects on hydration heat, 109 materials science, 3, 5
light definition of, 5
absorption of, and passivation, 1319 as an interdisciplinary field, 31
polarized, application to electrode maximum, electrocapillary, 691
surfaces, 36 maximum cell potential, 1137
reflected, and passivation, 1320 maximum efficiency, of electrochemical
theory of, electromagnetic, 35 converter, 1364
visible, application to electrode surfaces, maximum energy density, for electrode
36 couples, idealized, 1408
limiting current Maxwell, his demon, 679
and hydrogen evolution, 1229 mean jump distance, 340
practical importance of, 1059 mean square distance, time, 453
typical experimental values of, 1060 measurement
limiting current density, and electric of overpotential, and high impedance, 897
migration, 1075 of Volta potential difference, 841, 842
limiting law, breakdown of, 267 mechanical properties
linear diffusion, time-independent, 1077 change of, due to hydrogen, 1129
Lippman equation, 702 and hydrogen codeposition, 1229
liquids, comparison of properties of various mechanisms
kinds, 519 biological, 1266
 INDEX xlix

mechanisms (cont.) metallurgy, 27, 1266

of ethylene oxidation, on various metals, and electrochemistry, 28
1160 methanol-water mixtures, abnormal
of hydrogen evolution, 1235 conductivity in, 473
affected by coverage of hydrogen, 1235 microrough surface, and diffusion layer
and coverage, 1247 thickness, 1219
of the oxidation of saturated hydrocarbons, microscopy, 37, 38
1107 microspiral, 1207
of oxygen electronation, evaluation, 1253 microsteps, 1207, 1208, 1209
of passivation, 1319 migration
of porous electrode, 1384 conductance in, further reading on, 367
probable, of hydrogen evolution, 1250 and diffusion, further reading on, 399
of reaction of ions, as function of electrostatic
and Faraday's law, 1096 potential gradient, 289
and surface coverage, 1096 ionic, 367
on surface which change with time, 1182 and radiotracers, 570
mechanism determination, 1080 Miller index, of metal surface, and cracking
and coverage of electrodes, 1097 under stress, 1335
for ethylene, 1094 mixed potential, 1286
further reading on, 1093, 1110 determination of, 1105
and mixed potentials, 1105 as a function of concentration, 1104
and separation factor, 1106 and mechanism determination, 1105
on surfaces which change with time, 1182 mobile electrified interfaces, 826
techniques for, .1099
mobility
mechanistic studies, summarizing remarks,
absolute, 370
1090 of charge carriers, in liquids and solids, 471
medical electronics, and electrochemistry, 43
conventional,371
membrane and diffusion coefficient, 374, 376
effect on gravitational flow, 569
of ions, 369
Tafel relation at, 942
of proton, 471, 472, 476
membrane potential, 410, 941
solvent effect on, at infinite dilutions, 443
meniscus, 1383, 1384, 1388
mercury mobility method
and double layer studies, 687 in determining solvation number, 125
and hydrogen evolution, 1232 for solvation number, 130
metal(s) for ionic liquids
band picture of, with interatomic spacing, comparison of ability to predict, 583
808 facts, 522
break-up when strained, 4 further reading in, 587
of great strength, and electrochemistry, 43 gas-oriented, 529
stretched, dissolution of, 1337 lattice-oriented, 522
sudden failure of, 4 summary, 532
metal deposition comparison of predictions, 485
consecutive steps in, 1183 model-oriented approach, 1
further reading on, 1230 molecular biology, developments in, and
with hydrogen, 1227 electrochemistry, 42, 43
and rise time, 1185 molten oxides, properties near melting point,
transients in, 1183 518
metal-metal potential difference, and molten salts
reaction rates, 1147 average size of hole in, 539
metal oxide and silicon atom, interaction in compressibility calculated from hole
silica network, 602 model,585
metal-slag equilibrium, and charge transfer, detection of ions by Raman spectra, 590
618 deviations from Nernst-Einstein
metal-slag system, as electrode-solution relation, 557
system, 618 diffusion in, at constant temperature, 548
metal-solution interface, and Volta potential, distribution law for hole size, 534
665 electroneutrality in, near electrode, 566
metal-solution junction, and semiconductor, future of model, 586
850 lifetime of complex ions in, 590
metal-water interactions, 740 liquid free volume model, diagrammatic,
metallurgical extraction, and 531
electrochemistry, 43 models in , further reading, 587
 INDEX

molten salts (cont.J network model

nonideal behavior due to interactions in, for liquid silicates, 606
588 defects of, 609
quasi-lattice model, 527 table describing, 606
Schottky defects in, 527 neutral ion pairs, diffusion of, 383
tests for complex ions in, 594 new towns, and electrochemistry, 43
vacancy model for, 526 Nobel Prize, given to electrochemist, 1060
volume change on fusion, 525 nonaqueous solutions, 400
Morse curve and electricity storers, 1424
and acid-base strength, 501 standard electrochemical potential in,
and symmetry factor, 922, 926 1425
Morse equation nuclear magnetic resonance spectroscopy,
a gross approximation, 981 and electrolytic solutions, 278
in proton conductance calculations, 477 nuclear power, direct conversion to
moving ion, asymmetric cloud around, 429 electricity, 1396
multistep reactions, 991 nucleation
activation energy for, 1002 conditions for, 1204
and Butler-Volmer equation in final form, and metal deposition, 1204
1000 ohmic resistance, 997
further reading on, 1017 Onsager equations, and ionic fluxes, 411,
involving stoichiometric number, 1005 828
n-p junction(s), 930 optical method, of examining passivation,
diffusion of holes and electrons across, 1319,1326
932 optimization, for energy conversion, 1374
exponential laws for, 936 orbitals, and hydration, 11 0
potential difference at, 932 order, of electrodic reaction, 1008
Il-P product, 819 order of reaction in electrodic reactions, 1009
n-type semiconductors, 818 compared with order of chemical
band picture, 820 reactions, lOll
natural convection, 1050 organics, effect on smoothing, 1222
natural gas, electrochemical burning, organic adsorption, 791
1393 1394 and electrode charge, 795
negative potential, superposition of, and flip-flop model for, 797
stabilization,1314 forces in, 792
Nernst diffusion layer, 1056 further reading on, 800
Nernst-Einstein equation, limitations, 382 maximum, and potential, 798, 799
Nernst-Einstein relation, 377, 381 organic electrolytes, use in electrochemical
and diffusion of ion pairs, 583, 586 electricity storers, 1413
gross view of deviations in molten salts, 557 organic reactants, and electrode kinetics, 916
and heats of activation, 556 organic substances, and morphology, 1222
and ionic liquids, 555 outer Helmholtz plane, 635
irreversible thermodynamic view of and Butler-Volmer equation, 911
deviations, 558 locus or reaction, 917
and paired vacancy theory of diffusion, 562 position of, 757
Nernstian Hiatus, 16,23,24 outer potential, 673
and smog, 16 definition, 663
and pollution, 17 diagrammed, 665
Nernst-Planck equation, and electrolytic and double-layer studies, 665
transport, 405 overall rate, and rate-determining step, 1002
Nernst-Planck flux equation, 398 overpotential, 883
and diffusion equation, 411 activation, 1053
Nernst equation attitude toward, in pre-electrodic days,
dilemmas associated with, 907 1131
discussion of deduction, 899 classical picture, 1231
kinetically deduced, 898 concentration, 1052
physical significance associated with, 908 consequences of lack of understanding,
and potential-pH diagrams, in corrosion, 1381
1279 definition of, 880
sphere of relevance, 906 deposition, and exchange current density,
as zero-current special case, 898 1215
Nernst's relation, and alloys, 1223 and electron queue, 993
nerve cell, diagram of, 937 and electronation reaction, 1133
network, and associated water, 77 and exchange current density, 896
 INDEX Ii

overpotential (cont.) passivation (cont.)

measurement of, and high impedance, 897 competition in models, 1324
and multistep reaction, a near-equilibrium dissolution-precipitation, mechanism of,
relation, 994 1321
and pollution, 1420 electrochemical model, 1325
and self-driving cell, 1136 and electronic conductivity of films, 1320
social importance of knowledge of, 16 ellipsometry in, 1319
for various steps in multistep reaction, 996 induction time in, 1322
overvoltage of hydrogen, 1231 mechanism of, 1319
oxidation of ethylene, dependence of rate monolayer model, 1325
upon substrate, 1161 optical method of examining, 1319
oxide film and protection from corrosion, 1348
and electrocatalysis, 1258 and reflected light, 1320
electronic conduction of, and solid state model, 1325
passivation, 1320 passivation potential, 1316, 1317
and oxygen catalysis, 1258 passive films, formation at the bottom of pits,
oxide path, in oxygen evolution, 1104 1335
oxygen passivity
adsorption on noble metals, related to and adsorbed ions on kink sites, 1325
d-band character, 1163 competing theories of, 1325
catalysis, and doped tungsten bronzes, electrodic model of, 1326
1169, 1422 criteria for distinction between models,
electronation 1326
and corrosion, 1252 electrodic model of, 1326
evaluation of mechanism, 1253 and oxide formation, 1327
and life, 1252 and potential-pH diagram, 1327
evolution, oxide path in, 1104 and thickness of oxide in critical region,
oxygen pressure, and corrosion rate, 1276 1325
oxygen reactions, further reading on, 1263 passivity and corrosion, further reading on,
oxygen reduction 1349
chemical step in, 1255 paths of hydrogen evolution, 1233
rate-determining step in, 1255 Pauling equation, use in electrocatalytic
and stoichiometric number, 1254 theory, 1164
oxygen reduction reaction penetration of hydrogen, into bulk of metal,
catalysis of, 1256 1329, 1334
observation of equilibrium potential, 1263 Pennsylvania, Electrochemistry Laboratory
oxygen starvation, and corrosion, 1303 in [University of], vii
ozone, connected with poIlution, 1353 permeation
pH, and current efficiency, 1229 affected by embrittlement, 1344
pH-potential diagrams, 1120,1278 as function of stress, 13 31
for iron, 1280 permeation currents, and evidence for
for lead, 1284 cracking, 1343
and Nernst equation, 1279 perturbation methods, 1019
and solution concentration, 1280 phases, moving, and the double layer, 826
uses and abuses, 1281, 1284 phase boundary, 624
paint, corrosion through, 1302 double layer at, 630
paired vacancies photochemical reactions, and electrical, 11
diffusion of, 1124 photoelectric effect, 674
and Nernst-Einstein relation, 562 photogalvanic converters, and electro-
parabola chemistry, 43
in electrocapillary curves, 705 photolysis, flash, 34
in surface tension-potential relation, 691 pipes, corrosion of, 1304, 1345
parallel plate condenser model, surface, Planck's relation, 948
tension-potential relation for, 719 Planck-Henderson equation, 417, 419
partial molar volume platinum
of hydrogen, in metal, 1331 hydrogen e1ectronation reaction on, 968
of hydrogen in metals, and distortion of special position in electro catalysis,
lattices, 1334 reasons for, 1166
particles and formation of holes, 578 platinum electrocatalyst, difficulties, 1158
passivation, 1315 platinum-gold, catalysis on, 1259
and absorption of light, 1319 poisons, 1262
characteristic current-potential course, Poisson equation
1317 applied to diffuse charge region inside
Iii INDEX

Poisson equation (cont.) potential(s) (cont.)

semiconductor, 812 of metals, in equilibrium with 1M solution
deduction of, 282 of ions, 1282
solution of, for double-layer situation, 726 metal-solution, measurement of changes in,
Poisson-Boltzmann equation 650
linearized form of, 190, 234 mixed, as a function of concentration, 1104
so-called rigorous solution, 247 of oxide formation, 1282
tests for validity of solution, further of passivation, tabulated, 1317
reading on, 250 produced by electrochemical energy
polarizability, 895 converter, and current, 1370
degree of, at various electrode surfaces, 802 reversible, of oxygen, determined, 1261
effect in salting out calculations, 161 standard, of certain electrode reactions,
and exchange current density, 895 tabulated, 1116
polarizable interface and surface tension at interface, 689
and interfacial thermodynamics, 700 variation of film thickness with, and
and lack of equilibrium, 697 passivity, 1325
polarographic wave, 1066 variation of, and ionic atmosphere, 636
polarography, 1060, 1062, 1065, 1069 Volta, and potential of zero charge, 707
pollution of zero charge, 691
of atmosphere, with products of internal potential-pH diagram, 1278
combustion, 1352 potential-distance relation, for colloid
and cars, 1419 particles, 729
and electrochemistry, 43 potential difference
and funding of fuel cell research, 1386 abs91ute
and Nernstian Hiatus, 17 attempts to measure, 644
and overpotential, 1420 at electrode-electrolyte interface, 670
prediction of, 1357 across single interface, 648
polyanion, discrete, 610, 614 structured, 660
pore across barrier, 871
thin, high limiting current at, 1383 across electrochemical system, 1112
wetted, and contact angle, 1383 across interphase, generality, of 631
porous eIectrode(s) between two electrodes, 346
activity near tip of meniscus, 1384 contact, 647
and cost of electrochemical energy of dipole, at electrode-electrolyte
converters, 1385 interface, 670
distribution of catalyst in, 1172 due to electron overlap at interface, 670
and three-phase boundary, 1383 in electro-endosmotic motion, 831
use of, 1171 Galvani, 670
vital importance in fuel cells, 1172 inside semiconductors, 804
possible recharge cycles, as function of at interface, 688, 912
depth discharge, 1417 at interfaces, further reading on, 687
measurability of outer and inner, tabulated,
potential(s)
absolute, impossibility of measurement, 677
and n-p junctions, 933
645
periodically varying, 1020
anode,1l26
Volta, measurement of, 841, 842
of average force, 183
potential drop, in diffiuse layer, 728
between metals, in cell, 652
potential electrolytes, 176, 488
changes during increase of current, as
function of type of cell, 1131 dissolution of, 491
and proton transfer reactions, 491
of charged sphere, 52
and true electrolytes, difference between,
chemical, 672
thought experiment in, 694 178
potential energy
of comparison, in electrocatalysis, 1145
and Morse curves, 920
of corrosion, 1285
of water system, during rotation, 477
in diffuse layer, as function of distance, 729
potential energy-distance profile, 866
ele.::trochemical, 693, 694
of electrochemical energy converter, its for successive motions of ions, 868
regions, 1372 potential energy-distance relations, and
at electrodes, due to dipoles, 783 tunneling through barrier, 962
electrostatic, variation of near interface, potential energy-distance theory, and
664 Morse equation, 919
Flade, 1316 potential energy barrier, for proton transfer,
at membrane, 410 479
 INDEX liii

potential energy barrier (cont.) probability (cont.)

in acid strength theory, 505 that hole has radius r, 538
potential energy curves, 921, 925, 971 of tunneling, on both sides of the barrier,
and effect of increased M-H bond 952
strength, 1156 probability amplitude, and electron passage,
in proton transfer, 969 947
and stretching of bonds, 971 projection, 1266
vertical shift under potential change, 924 and concentrated electric field at, 1217
potential pulse, 1020 propane oxidation
potential step, 1020 diagnostic coefficients, 1107
potential sweep method, 1033 and intermediates, 111 0
potential of zero charge, 706 possible rate determining steps, 11 09
and electrocatalysis, 1142 protection
and electrodeposition, 1180 anodic, diagrammatic, 1318
and heat of activation, for metal cathodic, 1309, 1312
deposition, 1181 theory of, 1309
and organic adsorption, 798 surface, 1266
and potential of maximum adsorption, protein, quantum mechanical tunneling to,
tabulated, 800 981
and surface potential, 707 proton(s), 461
tabulated, 864 affinity, 466
and Tafel slope, 915 conductance and Morse equation, 477
in terms of Volta potential, 707 chain mechanism, 474
potential-time transients constitution of, further reading on, 470
effect of intermediates on, 1026 free energy level, 500, 503
under diffusion control, 1046 heat of hydration of, absolute, 105
potentiodynamic method, 1033, 1034 hydrated, and electron tunneling to, 962
potentiostatic circuit, for examination of jumping, from water, 475
passivation, 1316 lifetime in solution, 485
potentiostatic transient(s), 1032 mobility of
and adsorption of benzene on platinum, abnormally high, 472, 485
1033 in aqueous solutions, model, 484
instrumentation for, 1033 further reading on, 488
Pourbaix diagrams, 1121, 1123, 1281 in ice, 486
and spreading of cracks, 1343 rate-determining step in, 485
power, 1410 and water rotation, detailed scheme, 483
and efficiency, in energy converters, 1379 as nonconforming ion, 461
and energy density, 1429 in solution, existence of, 465
power density solvation, 462
feasible goals for, 1431 proton jumps
as function of energy density, for and classical laws?, 478
silver-zinc cell, 1411 quantum-mechanical,480
and lithium-chlorine cell, 1423 water re-orientation necessary for, 481
power density versus energy density, 1411 proton transfer
power output, 1171, 1376 in autodissociation, 492
power stations, and fuel cells, 1396 conditions for, 968
powering of vehicles, and electrochemistry, 43 between hydroxonium ions and favorably
precipitation, of materials, near electrode, oriented water molecules, 478
1322 and potential electrolytes, 491
precursor film, 1319 potential energy barriers for, 505,479
pre-electrolysis, purification by, 1261 and tunneling conditions, 966
prepassive film, 1319 proton transport, 470, 473
pressure pseudocapacitance, equivalent circuit for,
critical, for spreading of crack, 1341 1028, 1029
of hydrogen in metals, 13 33 pseudo-equilibrium, 1236
high,1333 pseudoforce, and diffusion, 306
primary hydration number pulse
of proton, 469 double, and staircase pulsing, 1019
table of, 131 square wave, 1019
primary solvation, 79, 160 with step reversal, 1019
probability pulse generator, 329
of electron transfer, 978 pulsing, and double pulse, 1019
of finding one ion near another, 253 pure liquid electrolytes, 513, 518
liv INDEX

purification, 1260, 1261, 1263 rate-determining step (cont.J

purity, effect on corrosion, 1273 and charge transfer, 1185
quadrupole model, 99, 100, 101, 104 with chemical step, 1002
quantum electrochemistry, in 1960's, 19 concept of, 997
quantum-mechanical charge transfer, in context, 1138
relation to proton transfer, 974 in corrosion processes, 1296
quantum-mechanical proton jumps, 480 deduction of in terms of barrier height,
quantum-mechanical theory, or radiationless 1003
tunneling, 960 and energy barrier for multistep reaction,
quantum-mechanical transfer, basic condition 1002
for, 945 in ethylene oxidation, 1160
quantum mechanics, 27, 41 and exchange current density of partial
deduction of Butler-Volmer equation in, reactions, 997
980 and highest standard free energy, 1003
desirable refinements in electrochemical and hydrogen evolution, 1237
application to, 981 in hydrocarbon oxidation, 1159
of electrode processes, 947 and iron deposition and dissolution, 1084
of electrode processes, further reading in some metal depositions, 1195
on,985 in oxygen reduction, 1255
of electrode reactions, history of and path, 1139
developments, 983 in propane oxidation, 1109
and electrodics, 30 in proton motion, 485
symmetry factor theory in, 974 rate constant
tunneling conditions for, 972 for adion diffusion, as function of
of tunneling to proteins, 981 dislocation density, 1201
quantum nature of biological reactions, 21 determined from transient measurements,
quantum nature of electrochemical reactions, for surface diffusion, 1196
21 of non-rate-determining reaction, first
quasi-lattice theory, 271, 272 determination of, 1195
queues, 992 rate equations, for electrocatalysis, compared
radicals with catalysis, 1143
examination of, 1034 rate processes, 340, 341, 342, 387, 391,
intermediate detection of, 34 455,867
on surface, nature of, and mechanism, 1105 ratio of currents, in alloy deposition, 1226
radiotracer method rationality, of burning fossile fuels, lack
and electrochemical method, 1035 of,1352
for surface concentration, 1030 Rayleigh scattering, 277
radiotracers reaction (s )
and determination of transport numbers chemical and electrochemical, 8, 40
in fused salts, 570, 571 coupled, 1235
and electrochemical methods, comple- on electrodes, several simultaneous, 1214
mentary nature of, 1098 electron transfer, in 1950's, 17
Raman scattering, 277 multistep, 991
Raman spectra photo- and electro-, 11
and detection of complex ions in molten thermal, and electrochemical, 14
salts, 590, 591 reaction coordinate, 456
and electrolytic solutions, 277 reaction mechanism
ramp, various types, 1019 determination of, 1080
random walk elucidation of its stages, 1095
in deposition, 1186 and solution entities, 1095
distances moved and time, 306 and surface coverage, 1096
influence of electric field on, 350 reaction order, 1012, 1013, 1099, 1241
ionic movement as a function of, 299 reaction paths, for oxygen electronation,
sum of distance from origin, squared, 303 1255
and surface adions, in electrodeposition, reaction rate
1181 its dependence on isotope substitution, 11 05
rate(s) electrochemical, and relation to
of deposition, on different planes, theory metal-metal p.d., 1147
of,1215 reactivity, at low temperatures, in
as a function of substrate, for oxidation electrocatalysis, 1169
of ethylene, 1161 reciprocity relation. 413
rate-determining step, 1138 recovery of materials, and electrochemistry,
and activation barrier, 1003 43
 INDEX Iv

rectification, faradaic, 885 semiconductors (conl.J

redox reactions, 982, 1046, 1146 intrinsic, 910
reference potential, for electrocatalysis, and metal-solution junction, 850
1143 and tunneling, 956
reforming, of hydrocarbons, 1158, 1391 values of mobilities in, 931
relations in cells, further reading on, 1137 semiconductor-electrolyte interfaces, 803
relative strength of acid, 495 semiconductor electrochemistry
relative strength of base, 495 further reading on, 823
relative potential differences, 655, 656 and nonmetals, 823
relaxation, of ionic cloud, 423 semiconductor junctions, 930
relaxation methods, and various types of separation, of isotopes, and hydrogen
stimuli, 1019 evolution, 1246
resistance, 996, 997 separation factor, 1106, 1231, 1248, 1249
response separators, and dendrites, 1221
in concentration, to switch-on of current servicing center, 992
flux, 317 sewage disposal and electrochemistry, 43
of dielectric medium to field, 145 SHE and IUPAC convention, 1117
of system to outside stimulus, 325 signs, of electrode potentials, 1120
reversibility, microscopic, 874 sign convention
ring anions, in liquid silicates, 613 in electrochemistry, 1115
rise time, 1185-1194 IUPAC, in detail, 1119
rotating disk electrode, 1058 silica, fused, and liquid water, 596
development of, 1070 silicates
rotation liquid, and transport processes in simple
potential energy of water system during, fused salts, 609
477 structure of ions of,' 596
of screw dislocation, schematic, 1206 liquid network model of, 606
of a spiral, in electrode position, 1205 silicate melt, heat of activation for flow in,
of water and proton mobility: detailed 605
scheme, 483 silver-zinc cell, 1411, 1416, 1417
sacrificial anode, 1312 single crystals
salting in, 159 Bridgman technique for, 1218
anomalous theory, 164 deposition on, 1218
normal,163 single ions, 61, 62
salting, out, 158 sink, for electrons, 853
further reading on, 168 sinusoidal stimulus, rectified, 1019
saturated hydrocarbons, 1107, 1391 slags, 616-618
scattering, Rayleigh and Raman, 277 slow discharge, 1153, 1236
scavenger electrolysis, theory of, 1262 smog
Schottky defects, 526 its formation, 1354
Schrodinger equation, 947 and Nernstian Hiatus, 16
science, advances in, and electrochemistry, smoothing, 1222
39,41 sodium-sulfur cell, 1417, 1418
screw-thread, and spiral, analogy, 1207 sols, 839
screw dislocations, 1203, 1206, 1210, solid metals, and mercury, in double-layer
1211, 1212 studies, 802
secondary solvation, 79 solubility, 160, 163
effect on solubility, 161 solution (s)
self-diffusion agitation of, and corrosion, 1300
coefficients, of various substances, 545 entities in, and reaction mechanism, 1095
energies of activation for, 579 ideal, and activity coefficient, 205
experiments, results of, 544 ionic, quasi-lattice approach to, 266
experiments with radioactive isotopes, 542 ionic, standard state in, 204
self-driven cell, 1128 visible and ultraviolet absorption
self-driving cell, 1131, 1136 spectroscopy in study of, 275
semiconductors, 806 solvated ions, at interface, 633
analogy to electrolytes in solution, 32 solvation, 80
conduction of, 809 effect on activity coefficient, calculation of
and Debye-Hiickellength, 816 241
doped,819 evaluation of ion-dipole of, 93
and electrolytic solutions, 811 as function of orientation time of water
analogies tabulated, 812 molecule, 122
impurity additions to, 818 heats of, for pairs of ions, 63
Ivi INDEX

solvation (cont.) square wave pulse, 1019

ion-quadrupole theory of, 103 stability
primary, 79,160 electrodic principles of, 1348
of protons, 462, 469, 470 of interior of metal, and surface hydrogen,
secondary, 79 1328
as time-dependent phenomenon, 121 of metals, 1268
of transition metal ions, 106 and adsorbed hydrogen, 1328
solvation effects, on activity coefficients, electrodic approach to increase, 1306
consistency of theory?, 243 of surface and charge transfer, 1268
solvation number, stabilization of materials, and
and compressibility, 125, 126 electrochemistry, 42,1314,1318,1323
dynamic model, 124 standard electrochemical potentials, in
further reading on, 132 nonaqueous solutions, 1425
and hopping, 122 standard electrode potentials, table of,
and mobility method of determination, 125 problems, 1115
primary, explanation of, 119 standard free energy of activation, for
and theory of activity, 244 multistep reactions, 1002
ultrasonic method of determination of, 25 standard free energy of intermediate states,
usefulness of concept, 124 and water coverage, 1016
solvation theory, electrostatic, further standard hydrogen electrode, 1011
reading on, 72 steady state
solvent, its effect upon the strength of an and concentration of intermediates, 1038
acid,504 and convection, 1052
solvent effect(s) equality of velocity for all steps in, 1038
on activity coefficient, theory of, 242 and equilibrium, 1018
on mobility, at infinite dilutions, 443 and Fick's first law, 343
solvent levels, and strength of acids, 506 and galvanostatic transients, 1025
solvent properties, fused nonmetallic oxides, how long to establish?, 1018
602 steam hydrocarbon process, 1390, 1391
solvent sheath, structure of water in, 78 steel, and cathodic protection, 1312
source, for electrons, 853 step function flux, as function on time, 331
space charge, 813 step site, surface diffusion to, 1182
capacitance associated with, 816 Stern model, 733, 734, 735
inside semiconductor, 815 stimulation, of interface, to show variation,
in ionic solution, 217 1019
species, in solution, and mechanism stimuli, 1019, 1020, 1021
determination stoichiometric number, 1004
specific adsorption, 748 and Butler-Volmer equation, 1006
and constant adsorption, 638 determination of, 1101
specific conductivity and hydrogen evolution, 1241
of aqueous and nonaqueous solutions, 446 and multistep reactions, 1005
and ionic mobility, 372 and oxygen reduction, 1254
spectra, Raman, and electrolytic solUtions, and symmetry factor, 1007
277 tabulated, for hydrogen evolution, 1245
spectroscopy, 27 and transfer coefficient, 1007
ellipsometric, 37 Stokes-Einstein relation, 377, 379, 380,
infrared, and electrolytic solutions, 278 551,552
nuclear magnetic resonance, and Stokes' law
electrolytic solutions, 278 and proton mobility, 471
Raman, and study of solutions, 276 and transport numbers in fused salts, tests,
in study of electrolytic solutions, 272 573
visible and ultraviolet absorption, in study Stokes' law approach, to transport numbers
of solutions, 275 in fused salts, 572
sphere, 50, 52, 55 Stokes mobility, 381
spherical cavity storage
field in, 148 of charge, 703
with reference dipole, 148 of electricity, 40, 1266, 1421
spherical diffusion, 1062 strain, in metals, caused by hydrogen, 1333
and growth at surface, 1219 streaming current
spiral, formation of, from screw dislocation, and diffuse layer, 830
1206 relative motion of phases in, 827, 829
spiral growth, 1203 streaming current density, equation relating
spiral tip, concentration at, 1220 to zeta potential, 831
 INDEX Ivii

streaming potential, and specific conductivity, surface diffusion parameter, in metal

828 deposition, 1202
strength surface excess
of acid, absolute, 494 and amount adsorbed, 684
of acid, and solvent effect, 504 determination, 710
of base, 494 and distribution of species in the interface
stress, 1330, 1331 region, 683
stress corrosion, examples, tabulated, 1338 electrocapillary equations for, 710
stress corrosion cracking, 1335, 1336, 1337, of individual species, and surface tension,
1349 711
stretching, of bond, in ions at electrodes, 920 as macroscopic concept, 684
structure breaking in liquid silicates, and and radiotracer measurements, 686
concentration of metal ions, 604 surface potential, 667, 673
structure of charged interfaces, birds'-eye across interface, 677
view, 824 between two wires in cell, 675
substance producer, 854, 1127 measurement of change of, 674
as electrochemical device, 851 origin of, 669
summarizing remarks, on mechanistic and potential of zero charge, 707
studies, 1090 surface state
summary and contact adsorption, 821
charge transfer theory, 893 and Gouy-Chapman diffuse charge, 822
of corrosion mechanisms, 1345 in semiconductors, and contact-adsorbed
of criteria in good electrochemical ions, 822
energy storage, 1412 surface tension
of direct energy conversion, 1398 and capacitance, 705
of electrochemical electricity storers, 1412 at interface, and potential, 689
of electrochemical energy storage, 1430 and potential difference at interface, 688
superequivalent adsorption, 638,748 and surface excess, 711
superimposition of negative potential, and and variation with electrolyte
stabilization, 1314 concentration, 693
superposition, of potential of ion and cloud, surface tension-potential relation, from
199 parallel plate model, 719
surface(s) sweep, triangular, 1019
and civilization, 1267 Swiss cheese, and hole model, 527
decoration of, 1266 symmetry, dependence on hill-shaped
electrification of, consequences, 7 potential barrier, 936
electrode, state of, 36
symmetry factor, 923
of metals
and cracks by hydrogen, 1333 basic condition for the presence of, 977
and biological situations, 941
and instability, 1328
deduction in terms of potential energy
protection of, 1266
curves, 925
stability of, and charge transfer, 1268
dependence on potential, 926
and time change, in respect to
electrode position, 1182 elementary theory, 923
first model, 871
surface adion(s)
deposition from, 1187 and flow of chemical to electrical energy,
1138
in random walk, 1181
surface coverage and Morse curves, 922, 923, 926
potential dependence as a function of
during benzene oxidation, 1098
of exchange current density, 927
and Butler-Volmer equation, 1014
in quantum mechanics, third model, 674,
dnd charge transfer reaction, 1014
by galvanostatic transients, 1030 977
second model, 922
and mechanism of reaction, 1096
and stoichiometric number, 1007
surface diffusion
and transfer coefficient, 1007
of adions, in electrode position, 1177
contribution to overpotential, synthesis, 40
in steady state, 1199 Tafel's law, 15
dependence upon time, 1194 Tafel lines
equations for, 1188 for alloy deposition, 1225
to kink site, 1182 for oxygen reduction on doped bronzes,
and lattice formation, 1195 1422
to a step site, 1182 Tafel relation
and stepwise dehydration, 1177 effect of diffuse layer on, 915
Iviii INDEX

Tafel relation (cont.) transient techniques, and prospects for

and electrochemical energy conversion, e1ectrodic research, 34
1370 thrombosis, electrochemical mechanism of,
at membranes, 942 840
Tafel slope transfer
effect on corrosion, 1294 of hydrated ion, to hole site, 1180
and potential of zero charge, 915 mass transfer and heat transfer, analogies
and reaction order, 1090 between, tabulated, 1041
tanks, stabilization of, 1318 representation of, for ion to kink or edge
teflon, use in porous electrodes, 1391 vacancy, 1180
temperature transfer coefficient
of atmosphere, as function of time, 1356 and Butler-Volmer equation, 1007
and dielectric constant, 60 for cathodic and anodic reactions, 1007
terminology, of electrodes, historical, 1126 as center of electrode kinetics, 918
terminology, in energy conversion and determination of, 11 00
storage, 1402 and hydrogen evolution mechanisms,
test charge, and definition of Galvani tabulated, 1235
potential, 672 and stoichiometric number, 1007
textbooks, electrochemical in English, and symmetry factor, 1007
absence of, v tabulated summary of values, 1241
thermal combustion, and waste of energy, and Tafel plots, 11 03
1357 for various mechanisms, 1007
thermionic emission, 953 transition(s)
thermionic energy converter, 1359 electronic, adiabadic, 974, 977
thermionic work function, and radiationless, and conservation of
electrocatalysis, 1147, 1150 momentum, 950
thermoelectric energy converter, 1359
transition metal ions
thickness of atmosphere, variation of, with heat of hydration and atomic number, 108
concentration, 198 hydration of, and stabilization of field,
Thompson's hypothesis, 416
106, 111
thought experiments
interaction with water, 106
at charged interface, and surface potential,
transition time, 1043
668 and concentration, 1048
and definition of chemical potential, 694 concept of, 1047
with image forces, 661 and convection, 1052
three-electrode system, 891 relation to potential, 1048
three-phase boundary, at porous electrode, and charge transfer, 1036
1383 to electrode, effect on kinetics, 916
importance of, 1384 control, further reading on, 1079
transients, 32, 1020 forces in phenomenological treatment of,
basic equations for, 1027 294
cell for measurement, 1184 in fused salts, atomistic theory, 574
and equivalent circuits, 1026 in fused salts, further reading on, 573
galvanostatic transients, 1021, 1185
of proton, 470
arrangement for, 1022 in solution, work done in, 292
and Butler-Volmer equation, 1190
and determination for surface radical transport aspects of electrodics, summary,
concentration, 1197 1076
and double-layer charging, 1190 transport numbers, 400, 40 I, 402, 403, 406,
solution of equations for, 1191 415,418,565,566,568,571,572
and steady state, 1025 transport phenomena, 541
at interfaces, 1017 transport theory, and history, 1042
and Laplace transformation, 1041 transport, and electrochemistry, 43
in metal deposition, 1183 triangular sweep, 1019
potential-time triple ions, 265
under diffusion control, 1046 and nonaqueous solutions, 449
effect of intermediates on, 1026 triple layer, 750
potentiostatic, 1032 tritium
instrumentation for, 1033 use in adsorption measurements, 1030
transient behavior, and equivalent circuit, use in mechanism determination, 1106
1026 tungsten bronzes, doped, and oxygen
transient methods, 1019 catalysis, 1422
transient state, and Fick's second law, 344 tunneling, 952
 INDEX lix

tunneling (cont.) waiting lines, 992

conditions for, effect of adsorption water
energy on, 965 adsorption of, and Butler-Volmer
condition for, in hydrogen evolution, 1154 equation, 1015
to cytochromes, and enzymes as electrodes, bound near to ion, and deviations from
1253 ideality, 238
and electrode reactions, 955 constitution of, in vapor, 464
and electrodeposition of metals, 1177 de-electronation reaction, and ethylene
of electron, 946 oxidation, 1160
conditions for, 972 de-ionization of, thermal and
and de-electronation reaction, 973 electrochemical, 40
through barrier, probability of, 952 as dielectric, 132
and Esaki tunnel diode, 956 dielectric constant, and association of
and Fermi level, 964 dipoles, 154
forbidden conditions for, 972 dielectric unit in, 146
to hydrated proton, 962 on electrodes, flip-flop, 779
and image energy, 961 immobilized near ion, 77
and Morse equation, 981 interaction, with transition metal ions, 106
need for acceptor particles in, 977 a new form, 76
probability of, and width of barrier, 954 orbitals around, 73
of protons, 1155 potable, from fuel cells, 13 89
in ice, 487 primary region of solvation, 79
in water, 487 production of, and fuel cells, 1420
quantum-mechanical, and biological as quadrupole, 98
membranes, 981 in secondary region of solvation, 79
radiationless, 961 and silica, 595
and semiconductors, 956 as solvent, 440
simultaneous, of two electrons, structure of, 72, 80
unlikelihood of, 1082 theory of its dielectric constant, 1461
work function in, 963 water coverage, possible effect on
turbulence, 1055 intermediate concentration, 1016
ultrasonics, and hydration number, 128 water dipoles
ultrasonic irradiation, and electrocatalysis, at electrodes, and capacitance, 788
1170 flip-flop model, 741, 780
water molecules, adsorption on electrodes,
ultraviolet and visible adsorption
spectroscopy, in study of solutions, 275 739
condensation of, in solvation calculation,
undevice, 859, 860 89
unsaturated hydrocarbons, electrochemical orientation of, at electrodes, 741
data concerning, 1392 water purification, and electrochemistry, 42
urban living, and electrochemistry, 42 water reorientation, 482
vacancies, types of, 1179 and proton mobility, 482
vacancy model for fused salts, 526 wave form, sinusoidal, 1052
vacuum, work function in, 963 wave function, of electron, near interface,
virial, definition of, 237 670
viscosity wavelength, de Broglie, 949
kinetic theory expression for, 621 work
of molten salts, 547 chemical and electrical, separation of, 695
visible and ultraviolet absorption of transfer, of ion from vacuum to
spectroscopy, in study of solutions, solvent, 50
275 work done
volcano relation, in electro-organic chemistry, in transport in solution, 292
1164 in transport of unit charge from infinity
volcano relation, interpretation of, in to interior of phase, 694
electrocatalysis, 1165 work function
Volta potential cancellation of, in electrocatalysis, 1147
measurability of, 666 and electrocatalysis, 1146
and metal-solution interface, 665 and exchange current density, 1149
Volta potential difference, theory of influence of concentration of reactants
measurement, 841 in double layer, 1150
volume change on fusion, molten salts, 525 influence upon rate of nonbonding
Wagner-Traud mechanism for corrosion, electrochemical reaction, 1149
1274 of metals, tabulated, 944
Ix INDEX

work function (cont.) zero charge situation, 863

in tunneling, 963 zero current, and Nemst's law, 897
and zero charge potential, 1151 zeta potential
water discharge, in ethylene oxidation, 1160 and clotting of blood, 841
water line, corrosion at, 1303 and dependence on concentration, 913
X-rays, and constitution of protons in and electrode kinetics, 912
solution, 462 relation to concentration, 914
Young's modulus, 1340 and streaming current, 831
zero charge, potential of, 691, 706 zinc, and air electrode, storage, 1427
and electrocatalysis, 1142 zinc-air cell, and storage, 1428
zero charge potential, relation to work zinc-containing cells, and dendritic growth,
function on, 1151 1428