978 94 011 0691 7

HICTROCHIMIHRY
ELECTROCHEMI~TRY
SECOND EOITION
PHILIP H. RIEGER
SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

© 1994 Springer Science+Business Media Dordrecht
Originally published by Chapman & HaU in 1994
Softcover reprint of the hardcover 2nd edition 1994
AII rights reserved. No part of this book may be reprinted Of reproduced or utilized
in any form Of by any electronic, mechanical or other means, now known or
hereafter invented, including photocopying and recording, or by an information
storage or retrieval system, without permission in writing from the publishers.
Library of Congress Cataloging-in-Publication Data

Reiger, Philip Henri, 1935-
E1ectrochemistry / Philip H. Reiger. - 2nd ed.
p. cm.
Includes bibliographical references and indexes.
ISBN 978-94-010-4298-7 ISBN 978-94-011-0691-7 (eBook)
DOI 10.1007/978-94-011-0691-7
1. Electrochemistry. 1. Title.
QD553.R53 1993
541.3'7 -dc20 93-25837
CIP
British Library Cataloguing in Publication Data available

Dedicated to the memory of those who
inspired my interest in electrochemistry:
Arthur F. Scott
William H. Reinmuth
Donald E. Smith
Contents
Preface ix
Chapter 1: Electrode Potentials 1
1.1 Introduction 1
1.2 Electrochemical Cell Thermodynamics 5
1.3 Some Uses of Standard Potentials 13
1.4 Measurement of Cell Potentials 27
1.5 Reference and Indicator Electrodes 31
1.6 Ion-Selective Electrodes 35
1.7 Chemical Analysis by Potentiometry 39
1.8 Batteries and Fuel Cells 44
References 54
Problems 55
Chapter 2: The Electrified Interface 59
2.1 The Electric Double Layer 59
2.2 Some Properties of Colloids 68
2.3 Electrokinetic Phenomena 73
2.4 Electrophoresis and Related Phenomena 81
2.5 Electrode Double-Layer Effects 85
2.6 Debye-Hiickel Theory 90
References 105
Problems 106
Chapter 3: Electrolytic Conductance 109
3.1 Conductivity 109
3.2 Conductance Applications 125
3.3 Diffusion 128
3.4 Membrane and Liquid Junction Potentials 136
References 146
Problems 147
Chapter 4: Voltammetry of Reversible Systems 151
4.1 Diffusion-Limited Current 152
4.2 Experimental Techniques 165
4.3 A Survey of Electroanalytical Methods 174
4.4 Cyclic Voltammetry 183
viii Contents
4.5 Polarography 194

4.6 Polarographic Variations 201
4.7 The Rotating-Disk Electrode 207
4.8 Microelectrodes 215
4.9 Applications 223
References 237
Problems 240
Chapter 5: Mechanisms of Electrode Processes 247
5.1 Introduction 247
5.2 Spectroelectrochemistry 257
5.3 Steady-State Voltammetry and Polarography 269
5.4 Chronoamperometry and Chronopotentiometry 287
5.5 Cyclic Voltammetry 296
References 308
Problems 310
Chapter 6: Electron-Transfer Kinetics 315
6.1 Kinetics of Electron Transfer 315
6.2 Current-Overpotential Curves 325
6.3 Electron-Transfer Rates from Voltammetry 338
6.4 Faradaic Impedance 351
References 367
Problems 368
Chapter 7: Electrolysis 371
7.1 Bulk Electrolysis 371
7.2 Analytical Applications of Electrolysis 376
7.3 Electrosynthesis 390
7.4 Industrial Electrolysis Processes 396
7.5 Corrosion 412
References 421
Problems 423
Appendices 427
1 Bibliography 427
2 Symbols and Units 433
3 Electrochemical Data 438
4 Laplace Transform Methods 448
5 Digital Simulation Methods 462
6 Answers to Selected Problems 467
Author Index 472
Subject Index 477
PREFACE
It has been fashionable to describe electrochemistry as a discipline

at the interface between the branches of chemistry and many other
sciences. A perusal of the table of contents will affirm that view.
Electrochemistry finds applications in all branches of chemistry as well
as in biology, biochemistry, and engineering; electrochemistry gives us
batteries and fuel cells, electroplating and electrosynthesis, and a host of
industrial and technological applications which are barely touched on in
this book. However, I will maintain that electrochemistry is really a
branch of physical chemistry. Electrochemistry grew out of the same
tradition which gave physics the study of electricity and magnetism.
The reputed founders of physical chemistry-Arrhenius, Ostwald, and
van't Hoff-made many of their contributions in areas which would now
be regarded as electrochemistry. With the post-World War II capture of
physical chemistry by chemical physicists, electrochemists have tended
to retreat into analytical chemistry, thus defining themselves out of a
great tradition. G. N. Lewis defined physical chemistry as "the study of
that which is interesting." I hope that the readers of this book will find
that electrochemistry qualifies.
While I have tried to touch on all the important areas of
electrochemistry, there are some which have had short shrift. For
example, there is nothing on the use of dedicated microcomputers in
electrochemical instrumentation, and there is rather little on ion-
selective electrodes and chemically modified electrodes. The selection of
topics has been far harder than I anticipated, a reflection of my
ignorance of some important areas when I started. On the other hand,
there may be a few topics which may appear to have received too much
attention. I confess that my interest in electrochemistry is primarily in
mechanistic studies, particularly with organometallic systems. This
orientation may be all too apparent for some readers.
Since this is a textbook with the aim of introducing electrochemistry
to the previously uninitiated, breadth has been sought at the expense of
depth. I have tried, however, to provide numerous entries into the
review literature so that a particular topic of interest can be followed up
with a minimum of effort. References in the text are of four types. Some
are primarily of historical interest; when I have traced ideas to their
origins, I have tried to give the original reference, fully aware that only a
science history buff is likely to read them but equally aware that such
references can be hard to find. A second class of references is to specific
results from the recent literature, and a third class leads to the review
literature. These references are collected at the end of each chapter. A
fourth class of references includes the books and monographs which are
collected in a classified Bibliography, Appendix 1.
ix
x
SI units have been employed throughout the book. References to
older units are given in footnotes where appropriate. In most cases, the
use of SI units eliminates unit conversion problems and greatly
simplifies numerical calculations. The major remaining source of
units ambiguity comes from concentrations. When a concentration is
used as an approximation to an activity, molar units (mol L-l) must be
used to conform to the customary standard state. But when a
concentration acts as a mechanical variable, e.g., in a diffusion problem,
the SI unit, mol m- 3 , should be used. The mol m- 3 concentration unit is
equivalent to mmol L-l and, in a sense, is a more practical concentration
scale since voltammetric experiments often employ substrate
concentrations in the millimolar range.
Several topics have been added or expanded in the second edition.
In particular, coverage of microelectrode voltammetry has been much
expanded, and previous discussions of steady-state voltammetry with
rotating-disk electrodes have been modified to include microelectrodes;
spectroelectrochemistry (electron spin resonance and infrared
spectroscopy) is now discussed as an aid to deducing mechanisms of
electrode processes; the discussion of cyclic voltammetry has been
expanded to include adsorption effects and derivative, semi-derivative
and semi-integral presentation; the discussion of organic
electrosynthesis has been considerably expanded; and many new
examples of work from the literature have been added to illustrate the
techniques discussed.
It has been said that no book is ever finished, it is just abandoned.
The truth of that aphorism is never more apparent than to an author
returning to a previously abandoned project. There has been more than
one instance when I have been appalled at the state in which I left the
first edition of this book. I have labored mightily to correct the errors of
commission and at least a few of the errors of omission, but the awful
truth is that the book must be abandoned again with topics which should
have been covered more completely or more clearly.
I am particularly grateful to my wife, Anne L. Rieger, for her
patience in listening to my problems and for her encouragement in
times of discouragement. My colleague, Dwight Sweigart, has been an
invaluable source of expertise and encouragement during the
preparation of the second edition. I am indebted to Petr Zuman for some
valuable suggestions after publication of the first edition and to Nancy
Lehnhoff for a stimulating discussion of microelectrodes which greatly
clarified the presentation. I am deeply grateful to Barbara Goldman of
Chapman and Hall for her thoughtful suggestions and timely support in
this project.
Thanks are still due to those who helped with the first edition: to
David Gosser, who listened to my ideas and offered many helpful
suggestions-the cyclic voltammogram simulations of Chapters 4 - 6 are
his work; to my colleagues at Brown who offered advice and
xi
encouragement, most particularly Joe Steim, John Edwards, and Ed
Mason; to Bill Geiger, who provided a stimulating atmosphere during
my 1985 sabbatical and gave some timely advice on electro analytical
chemistry; to James Anderson of the University of Georgia, Arthur Diaz
of IBM, San Jose, Harry Finklea of Virginia Polytechnic Institute, and
Franklin Schultz of Florida Atlantic University for their careful reading
of the first edition manuscript and numerous helpful suggestions.
The first edition was produced using the IBM Waterloo SCRIPT
word-processing system and a Xerox 9700 laser printer equipped with
Century Schoolbook roman, italic, bold, bold italic, greek, and
mathematics fonts. Seven years later, that system is hopelessly obsolete
and the present edition has been completely redone using Microsoft
Word on a Macintosh computer with equations formatted with
Expressionist. To maintain a semblence of continuity, the principal font
is again New Century Schoolbook. The figures all have been redone
using CA-Cricket Graph III, SuperPaint, and ChemDraw. Figures
from the literature were digitized with a scanner and edited with
SuperPaint.
Philip H. Rieger
May 1993
1
ELECTRODE
POTENTIALS
1.1 INTRODUCTION
Origins of Electrode Potentials
When a piece of metal is immersed in an electrolyte solution, an

electric potential difference is developed between the metal and the
solution. This phenomenon is not unique to a metal and electrolyte; in
general whenever two dissimilar conducting phases are brought into
contact, an electric potential is developed across the interface. In order
to understand this effect, consider first the related case of two dissimilar
metals in contact.
When individual atoms condense to form a solid, the various atomic
orbital energy levels broaden and merge, generally forming bands of
allowed energy levels. The band of levels corresponding to the bonding
molecular orbitals in a small molecule is called the valence band and
usually is completely filled. The band of levels corresponding to
nonbonding molecular orbitals is called the conduction band. This band
is partially filled in a metal and is responsible for the electrical
conductivity. As shown in Figure 1.1, electrons fill the conduction band
up to an energy called the Fermi level. The energy of the Fermi level,
relative to the zero defined by ionization, depends on the atomic orbital
energies of the metal and on the number of electrons occupying the band
and thus varies from one metal to another. When two dissimilar metals
are brought into contact, electrons flow from the metal with the
(a)
~ T (b) (c)
~k~%%Zy;:;~?/.~::;z~2$~Z0.Z2~-':~J.?~nZlJ bWff//ff/k;wuu~
Fermi level
Figure 1.1 The conduction bands of two dissimilar metals (a) when the
metals are not in contact; (b) at the instant of contact; and (c) at equilibrium.
1
2 Electrode Potentials
higher Fermi level into the metal with the lower Fermi level. This
electron transfer results in a separation of charge and an electric
potential difference across the phase boundary. The effect ofthe electric
potential difference is to raise the energy of the conduction band of the
second metal and to lower the energy of the conduction band of the first
until the Fermi levels are equal in energy; when the Fermi levels are
equal, no further electron transfer takes place. In other words, the
intrinsically lower energy of electrons in the conduction band of the
second metal is exactly compensated by the electrical work required to
move an electron from the first metal to the second against the electric
potential difference.
A very similar process occurs when a metal, say a piece of copper,
is placed in a solution of copper sulfate. Some of the copper ions may
deposit on the copper metal, accepting electrons from the metal
conduction band and leaving the metal with a small positive charge and
the solution with a small negative charge. With a more active metal, it
may be the other way around: a few atoms leave the metal surface as
ions, giving the metal a small negative charge and the solution a small
positive charge. The direction of charge transfer depends on the metal,
but in general charge separation occurs and an electric potential
difference is developed between the metal and the solution.
When two dissimilar electrolyte solutions are brought into contact,
there will be a charge separation at the phase boundary owing to the
different rates of diffusion of the ions. The resulting electric potential
difference, called a liquid junction potential, is discussed in §3.4.
In general, whenever two conducting phases are brought into
contact, an interphase electric potential difference will develop. The
exploitation of this phenomenon is one of the subjects of
electrochemistry.
Consider the electrochemical cell shown in Figure 1.2. A piece of
zinc metal is immersed in a solution of ZnS04 and a piece of copper
metal is immersed in a solution of CUS04. The two solutions make
contact with one another through a fritted glass disk (to prevent mixing),
and the two pieces of metal are attached to a voltmeter through copper
wires. The voltmeter tells us that a potential is developed, but what is its
origin? There are altogether four sources of potential: (1) the copper-
zinc junction where the voltmeter lead is attached to the zinc electrode;
(2) the zinc-solution interface; (3) the junction between the two solutions;
and (4) the solution-copper interface. The measured voltage is the sum of
all four interphase potentials.
In the discussion which follows, we shall neglect potentials which
arise from junctions between two dissimilar metals or two dissimilar
solutions. This is not to say that such junctions introduce negligible
potentials; however, our interest lies primarily in the metal-solution
interface and solid or liquid junction potentials make more or less
constant additive contributions to the measured potential of a cell. In
§ 1.1 Introduction 3
Zn Cu
glass frit
(salt bridge)
Figure 1.2 The Daniell Znso 4 CUS04

cell. solution solution
careful work, it is necessary to take explicit account of solid and liquid

junction potentials.
Origins of Electrochemistry
The electrochemical cell we have been discussing was invented in
1836 by John F. Daniell. It was one of many such cells developed to
supply electrical energy before electrical generators were available.
Such cells are called galvanic cells, remembering Luigi Galvani, who in
1791 accidentally discovered that static electricity could cause a
convulsion in a frog's leg; he then found that a static generator was
unnecessary for the effect, that two dissimilar metals (and an electrolyte
solution) could also result in the same kinds of muscle contractions.
Galvani thought of the frog's leg as an integral part of the experiment,
but in a series of experiments during the 1790's, Alessandro Volta
showed that the generation of electricity had nothing to do with the frog.
Volta's work culminated in the construction of a battery (the voltaic pile)
from alternating plates of silver and zinc separated by cloth soaked in
salt solution, an invention which he described in a letter to Sir Joseph
Banks, the President of the Royal Society of London, in the spring of 1800.
Banks published the letter in the Society's Philosophical Transactions
that summer, but months before publication, the voltaic pile was well
known among the scientific literati of London.
Among those who knew of Volta's discovery in advance of
publication were William Nicholson and Sir Anthony Carlisle, who
constructed a voltaic pile and noticed that bubbles of gas were evolved
from a drop of water which they used to improve the electrical contact of
the leads. They quickly showed that the gases were hydrogen and oxygen
Luigi Galvani (1737-1798) was a physiologist at the University of Bologna.

Beginning about 1780, Galvani became interested in "animal electricity"
and conducted all kinds of experiments looking for electrical effects in
living systems. Alessandro Giuseppe Antonio Anastasio Volta (1745-
1827) was Professor of Physics at the University of Pavia. Volta had
worked on problems in electrostatics, meteorology, and pneumatics before
Galvani's discovery attracted his attention.
William Nicholson (1753-1815) started his career as an East India

Company civil servant, was then a salesman for Wedgwood pottery in
Holland, an aspiring novelist, a teacher of mathematics, a physics
textbook writer and translator, a civil engineer, patent agent, and
inventor of scientific apparatus. He founded the Journal of Natural
Philosophy, Chemistry, and the Arts in 1797, which he published monthly
until 1813. Sir Anthony Carlisle (1768-1840) was a socially prominent
surgeon who dabbled in physics and chemistry on the side.
Sir Humphry Davy (1778-1829) was Professor of Chemistry at the Royal

Institution. Davy was an empiricist who never accepted Dalton's atomic
theory and spent most of his career looking for defects in Lavoisier's
theories, but in the process he made some very important discoveries in
chemistry. Michael Faraday (1791-1867) began his career as Davy's
assistant at the Royal Institution, but he soon made an independent
reputation for his important discoveries in organic chemistry, electricity
and magnetism, and in electrochemistry. Although his electrochemical
work was seemingly an extension of Davy's electrolysis experiments, in
fact Faraday was asking much more fundamental questions. Faraday is
responsible (with the classicist William Whewell) for many of the terms
still used in electrochemistry, such as electrode, cathode, anode,
electrolysis, anion, and cation. John F. Daniell (1790-1845) was Professor
of Chemistry at King's College, London. Daniell was a prolific inventor
of scientific apparatus but is best known for the electrochemical cell
which bears his name.
and that water was decomposed by electrolysis. The Nicholson-Carlisle

experiment, published in Nicholson's Journal only a few weeks after
Volta's letter, caused a sensation in scientific circles throughout Europe.
Volta's battery had provided for the first time an electric potential source
capable of supplying significant current, and this technical advance,
spurred by the discovery of water electrolysis, led in the next decade to
the real beginnings of the study of electricity and magnetism, both by
physicists and chemists. In the forefront among chemists was Sir
Humphry Davy, who used the voltaic pile as a source of electricity to
isolate metallic sodium and potassium in 1807, magnesium, calcium,
strontium and barium in 1808, and lithium in 1818. Davy's assistant,
Michael Faraday, went on in the next decades to lay the foundations of
the science of electrochemistry.!
! The early history of electrochemistry is brilliantly expounded in Ostwald's 1896

book, now available in English translation (C1).
§1.2 Electrochemical Cell Thermodynamics 5
1.2 ELECTROCHEMICAL CELL THERMODYNAMICS
Since the most obvious feature of a galvanic cell is its ability to

convert chemical energy to electrical energy, we begin our study by
investigating the thermodynamic role of electrical work. In § 1.3, we
discuss applications of data obtained from electrochemical cells. We
turn to some experimental details in §1.4-§ 1.6 and conclude this chapter
with introductions to analytical potentiometry in §1.7 and to batteries
and fuel cells in § 1.8.
Current also may be passed through a cell from an external source
to effect a chemical transformation as in the experiments of Nicholson,
Carlisle, and Davy; such cells are called electrolysis cells. We return to
that mode of operation, beginning in Chapter 4.
Electrical Work
The first law of thermodynamics may be stated as

!:J.U=q + w (1.1)
where !:J.U is the change in the internal energy of the system, q is the heat
absorbed by the system, and w is the work done on the system. In
elementary thermodynamics, we usually deal only with mechanical
work, for example, the work done when a gas is compressed under the
influence of pressure (dw = -PdV) or the expansion of a surface area
under the influence of surface tension (dw = ydA). However, other kinds
of work are possible and here we are especially interested in electrical
work, the work done when an electrical charge is moved through an
electric potential difference.
Consider a system which undergoes a reversible process at
constant temperature and pressure in which both mechanical (P- V)
work and electrical work are done, w = - p!:J. V + Welec. Since, for a
reversible process at constant temperature, q = T!:J.S, eq (1.1) becomes
!:J.UT,P = T!:J.S - p!:J.V + Welec (1.2)
At constant pressure, the system's enthalpy change is
Mip=!:J.Up+P!:J.V (1.3)
and at constant temperature, the Gibbs free energy change is
!:J.GT = MiT- T!:J.S (1.4)
Combining eqs (1.2)-(1.4), we have
!:J.GT,P = Wel ec (1.5)
Now let us see how electrical work is related to the experimentally

measurable parameters which characterize an electrochemical system.
Consider an electrochemical cell (the thermodynamic system) which
has two terminals across which there is an electric potential difference,
E.1 The two terminals are connected by wires to an external load (the
Figure 1.3 Electrochemical cell cell

doing work on an external
resistance.
surroundings), represented by a resistance R. When a charge Q is

moved through a potential difference E, the work done on the
surroundings is EQ. The charge passed in the circuit is the product of
the number of charge carriers and the charge per charge carrier. If we
assume that the charge carriers are electrons, then
Q = (number of electrons) x (charge/electron) = Ne
or
Q = (number of moles electrons) x (charge/mole) = nF
where F is the Faraday constant, the charge on one mole of electrons,
96,484.6 coulombs (C), and n is the number of moles of electrons
transferred. Thus the work done by the system on the resistor (the
resistor's energy is raised) is simply nFE. However, according to the
sign convention of eq (1.1), work done on the system is positive so that the
electrical work is negative if the system transfers energy to the
surroundings,
Welec =-nFE (1.6)
Substituting eq (1.6) into eq (1.5), we obtain the change in Gibbs free
energy of the system,
AGT,p=-nFE (1.7)
If E is measured in volts (V), Fin C mol-I, and n is the number of moles
of electrons per mole of reaction (mol mol-I), then AG will have the units
of joules per mole (J mol-I) since 1 J = 1 V-C. This quite remarkable
result immediately demonstrates the utility of electrochemical
measurements: We have a direct method for the determination of
1 In Chapter 2, where we will be dealing with electric potential in a slightly different

context, we will use the symbol cl> for potential. Here, we follow tradition and denote
the potential difference produced by an electrochemical cell by the symbol E, which
comes from the archaic term electromotive force. The electromotive force or emf is
synonymous with potential difference or voltage.
§ 1.2 Electrochemical Cell Thermodynamics 7
changes in the Gibbs free energy without recourse to measuring

equilibrium constants or enthalpy and entropy changes.
Electrochemical Cell Conventions

According to the second law of thermodynamics, a spontaneous
process at constant temperature and pressure results in a decrease in
Gibbs free energy. Thus a positive potential is expected when the cell
reaction is spontaneous. There is room for ambiguity here since the
sign of the potential depends in practice on how we clip on the voltmeter.
However, we recall the convention for the sign of !!.G for a chemical
reaction: if the chemical reaction is spontaneous, i.e., proceeds from left
to right as written, we say that!!.G is negative. We need a convention for
the sign of E which is consistent with that for !!.G.
In developing the required conventions, let us consider as a specific
example the Weston cell shown in Figure 1.4. 1 It is customary, in
discussing electrochemical cells, to use a shorthand notation to
represent the cell rather than drawing a picture of the experimental
apparatus. The shorthand representation uses vertical lines to
represent phase boundaries and starts from left to right, noting the
composition of each phase in the system. Thus, the Weston cell may be
represented as:
Cd( 12 .5% amalgam)ICdS04(S)ICdS04(aq,satd)IHg2S04(S)IHg(l)
We now agree by convention that, if the right-hand electrode is positive
with respect to the left-hand electrode, we will say that the cell potential
is positive.
The Weston cell was developed in 1893 by Edward Weston (1850-1936), an

inventor and manufacturer of precision electrical measuring
instruments.
Look now at the chemical processes going on at the two electrodes.

Consistent with the convention of reading from left to right, we say that
at the left-hand electrode, the process is
Cd(Hg) ~ Cd 2+(aq) + 2 e- (1.8)
and, at the right-hand electrode,
Hg2S04(S) + 2 e- ~ 2 Hg(l) + S042-(aq) (1.9)
The overall cell reaction then is the sum of these two half-cell reactions:
Cd(Hg) + Hg2S04(S) ~ Cd2+(aq) + S042-(aq) + 2 Hg(l) (1.10)
1 Because the potential of the Weston cell, 1.0180 V at 25°C, is very reproducible, it
has long been used as a standard potential source.
According to convention, the free energy change for the cell reaction is
negative if the reaction proceeds spontaneously to the right and,
according to eq (1.7), the cell potential should then be positive, i.e., the
right-hand electrode (Hg) should be positive with respect to the left-hand
electrode (Cd).
Figure 1.4 The Weston cell.
Let us see if this is consistent. If the Hg electrode is positive, then

conventional (positive) current should flow in the external circuit from +
to - (from Hg to Cd) and electron (negative) current in the opposite
direction. Thus electrons should enter the cell at the Hg electrode,
converting Hg2804 to Hg and 8042- [as in eq (1.9)], and leave the cell at
the Cd electrode, converting Cd to Cd2 + [as in eq (1.8)]; this is indeed
consistent with the overall cell reaction proceeding from left to right as
in eq (1.10).
The cell convention can be summarized as follows: For an
electrochemical cell as written, finding that the right-hand electrode is
positive, relative to the left-hand electrode, is equivalent to a negative !.'J.G
for the corresponding cell reaction. Conventional positive current flows
from right to left in the external circuit, from left to right in the cell.
Negative electron current flows from left to right in the external circuit,
from right to left in the cell. The left-hand (negative) electrode is called
the anode and the electrode process is an oxidation (removal of
electrons); the right-hand (positive) electrode is called the cathode and
the electrode process is a reduction (addition of electrons). 1
1 The identification of the cathode with the reduction process and the anode with the
oxidation process is common to both galvanic and electrolysis cells and is a better
Activities and Activity Coefficients

Consider a general chemical reaction
aA + f3B + 8D
~ yC (1.11)
According to chemical thermodynamics, the Gibbs free energy change
when the reaction proceeds to the right is
!J.G = !J.Go + RT In (ac)'Y(aD)O (1.12)

(aA)(l(aB)l3
where R is the gas constant, T the absolute temperature, and, for
example, ac. is the activity of species C. At equilibrium, !J.G = 0, and eq
(1.12) reduces to the familiar relation
!J.Go = -RT In Keq (1.13)
where
Keq = (ac)'Y{aD)O (1.14)
{aAf(aB)13
In the derivation of eq (1.12), the activities were introduced to account for
nonstandard states of the species. Thus for an ideal gas with standard
state partial pressure po = 1 bar, the activity is a = P / po; for a component
of an ideal solution with standard state concentration Co = 1 mol L-l (1
M), the activity is C / Co. Pure solids or liquids are already in standard
states, so that their activities are unity. The solvent in an ideal solution
is usually assumed to be essentially the pure liquid with unit activity.
In order to preserve the form of eqs (1.12), (1.13), and (1.14) for non-
ideal solutions or mixtures of nonideal gases, so-called activity
coefficients are introduced which account for the departure from
ideality. Thus for a solute in a real solution, we write
a = yC/Co (1.15)
where y is the unitless activity coefficient, C is the concentration, and Co
is the standard state concentration, 1 M.l Since Co = 1 M, activities are
numerically equal to yC and we will normally leave Co out of
expressions. We must remember, however, that activities, whether they
are approximated by molar concentrations or by partial pressures or
corrected for nonideality, are unitless. Thus equilibrium constants and
definition to remember than the electrode polarity, which is different in the two
kinds of cells.
1 We will use the 1 M standard state in this book, but another common choice is 1
molal, 1 mole solute per kilogram of solvent. Although activity coefficients are
unitless, they do depend on the choice of the standard state (see §2.6).
the arguments of logarithms in expressions such as eq (1.12) are also

unitless.
In this chapter, we will usually assume ideal behavior and ignore
activity coefficients. We will return to the problem of nonideality in §2.6.
The Nernst Equation
If we take eq (1.11) to be the overall reaction of an electrochemical

cell, then eqs (1.7) and (1.12) may be combined to give
-nFE = /)"Go + RT In (ac)'Y(aDyi (1.16)

(aA)U(aBf
If we introduce the standard potential, defined by
EO =-/)"Go/nF (1.17)
and rearrange eq (1.16) slightly, we have the equation first derived by
Nernst in 1889:
E =Eo_RT In (ac)'Y(aDr (1.18)
nF (aA)U(aBf
The N ernst equation relates the potential generated by an
electrochemical cell to the activities of the chemical species involved in
the cell reaction and to the standard potential, EO.
In using the Nernst equation or its predecessors, eqs (1.7) and (1.17),
we must remember the significance of the parameter n. We usually use
/)"G in units of energy per mole for a chemical equation as written. The
parameter n refers to the number of moles of electrons transferred
through the external circuit for the reaction as written, taking the
stoichiometric coefficients as the number of moles of reactants and
products. In order to determine n, the cell reaction must be broken down
into the processes going on at the electrodes (the half-cell reactions) as
was done in discussing the Weston cell.
If the Nernst equation is to be used for a cell operating at 25°C, it is
sometimes convenient to insert the values of the temperature and the
Faraday and gas constants and convert to common (base 10) logs. The
N ernst equation then is
E = EO _ 0.0592 log (ac)'Y(aD)O (1.19)
n (aA)U(aBf
Let us now apply the Nernst equation to the Weston cell. Identifying
the appropriate species for substitution into eq (1.18) we obtain (with n = 2
moles of electrons per mole of Cd)
§ 1.2 Electrochemical Cell Thermodynamics n
E = EO -MIn (aCd 2 +)(asol)(aHg)
2F (acd)(aHg2 SO.) (1.20)
Hg and Hg2804 are pure materials and so have unit activities. The
solution is saturated in Cd804, so that the Cd2+ and 8042- ion activities
are constant if the temperature remains constant. Finally, the Cd
activity is constant provided that the concentration of cadmium in the
amalgam (mercury solution) remains constant. If sufficient current is
drawn from the cell, enough of the cadmium could be oxidized to change
this concentration appreciably, but if only small currents are drawn, the
potential is seen to be relatively insensitive to changes such as
evaporation of the solvent. The potential of the Weston cell, 1.0180 V at
25°C, is easily reproducible and has a relatively small temperature
coefficient. The cell has been widely used as a standard potential source.
Hermann Walther Nernst (1864-1941) was Ostwald's assistant at the

University of Leipzig and later Professor of Physical Chemistry at the
Universities of Gottingen and Berlin. Nernst made many important
contributions to thermodynamics (he discovered the Third Law) and to
solution physical chemistry.
Half-Cell Potentials
As we have already seen, it is possible to think of the operation of a

cell in terms of the reactions taking place at the two electrodes
separately. Indeed we must know the half-cell reactions in order to
determine n, the number of moles of electrons transferred per mole of
reaction. There are advantages in discussing the properties of
individual half-cells since (1) each half-cell involves a separate reaction
which we would like to understand in isolation, and (2) in classifying
and tabulating results, there will be a great saving in time and space if
we can consider each half-cell individually rather than having to deal
with all the cells which can be constructed using every possible
combination of all the available half-cells.
Consider again the Weston cell, and think of the potential an
electron would see on a trip through the cell. The potential is constant
within the metallic phase of the electrodes and in the bulk of the
electrolyte solution and changes from one constant value to another over
a few molecular diameters at the phase boundaries. 1 The profile of the
potential then must look something like the sketch shown in Figure 1.5.
1 The details of this variation of potential with distance are considered in Chapter 2.
Hg
Figure 1.5 Hypothet-

t
E = 1.018 V
CdS04 solution
ical electric potential
profile in the Weston
cell. + Cd(Hg)
The Nernst equation for the Weston cell, eq (1.20), gives us the cell
potential as a function of the standard potential and of the activities of all
participants in the cell reaction. We would like to break eq (1.20) into two
parts which represent the electrode-solution potential differences shown
in Figure 1.5. It is customary in referring to these so-called half-cell
potentials to speak of the potential of the solution relative to the electrode.
This is equivalent to referring always to the electrode process as a
reduction. When, in the actual cell reaction, the electrode process is an
oxidation, the contribution to the cell potential will then be the negative of
the corresponding reduction potential. Thus in the case of the Weston
cell, we can write for the cell potential
Ecell = -ECd 2+/Cd + EHg2S04/Hg (1.21)
Breaking the standard cell potential into two components in the same
way, the half-cell potentials defined by eq (1.21) can be written in the form
of the Nernst equation such that substitution in eq (1.21) gives eq (1.20).
Thus
ECd2+/Cd = EOCd2+/Cd - RT In (acd)
2F (aCd 2+)
E Eo RI..1 (aHg)(asol)
Hg2S0JHg = Hg2S0JHg - 2F n ( )
aHg2so•
Unfortunately, there is no way of measuring directly the potential
difference between an electrode and a solution, so that single electrode
potentials cannot be uniquely defined. 1 However, since the quantity is
not measurable we are free to assign an arbitrary standard potential to
one half-cell which will then be used as a standard reference. In effect
this establishes a potential zero against which all other half-cell
potentials may be measured.
By universal agreement among chemists, the hydrogen electrode
was chosen as the standard reference half-cell for aqueous solution
1 We will see in §2.5 that some properties (the electrode-solution interfacial tension
and the electrode-solution capacitance) depend on the electrode-solution potential
difference and so could provide an indirect means of establishing the potential zero.
electrochemistry.1 The hydrogen electrode, shown in Figure 1.6, consists

of a platinum electrode coated with finely divided platinum (platinum
black) over which hydrogen gas is bubbled. The half-cell reaction
2 H+(aq) + 2 e· -t H2(g)
is assigned the standard potential of 0.000 V. Thus the potential of the
cell
PtIH2(g,a=I)IH+(a=I), Zn2+(a=I)IZn
is equal to the standard potential of the Zn2+/Zn half-cell. By measuring
the potentials of many such cells containing the hydrogen electrode
Pt
Figure 1.6 The hydrogen electrode.
chemists have built up extensive tables of half-cell potentials. A

selection of such data, taken from the recent compilation by Bard,
Parsons, and Jordan (HI3), is given in Appendix Table A.4.
1.3 SOME USES OF STANDARD POTENTIALS
With the data of Table A.4 or more extensive collections of half-cell

potentials (HI-HI4), we can predict the potentials of a large number of
electrochemical cells. Since the standard potential of a cell is related to
the standard Gibbs free energy change for the cell reaction by eq (1.17),
we can also use standard cell potential data to computel:lG 0, predict the
direction of spontaneity, or calculate the equilibrium constant (D6). In
this section we will work through several examples of such calculations.
1 See §4.2 for discussion of the problem of reference potentials for nonaqueous
solutions.
Many half-cell reactions in aqueous solutions involve H+, OH-, weak

acids, or weak bases, so that the half-cell potential is a function of pH.
While this dependence is predictable using the Nernst equation, it is
often inconvenient to take explicit account of pH effects and a variety of
techniques have been developed to simplify qualitative applications of
half-cell potential data. An understanding of these methods is
particularly important in biochemical applications. Several graphical
presentations of half-cell potential data have been developed in attempts
to make it easier to obtain qualitative predictions of spontaneity for redox
reactions.
Potentials, Free Energies, and Equilibrium Constants
Building a hypothetical cell from two half-cells is straightforward

on paper. (The construction of an actual working cell often can present
insurmountable problems.) Consider two half-cells
A+ n e- ~ B
C +m e- ~ D
The half-cell reactions are combined to give the cell reaction
mA+nD ~ mB+nC
where we have multiplied the first equation by m, the second by n, and
subtracted. To see that the cell potential is just
EO cell = EO NB - EOC/D (1.22)
we need a few lines of proof. When we add or subtract chemical
equations, we similarly add or subtract changes in thermodynamic state
functions such as U, H, S, or G. Thus in this case
~Gocell = m ~GoNB - n ~GoC/D
~Go cell = m (-nFEO NB) - n (-mFEoC/D)
~Gocell =-mnF(EONB -EoC/D) =-(mn)FEOcell
where mn is the number of electrons transferred per mole of reaction as
written.
Why can we combine half-cell potentials directly without taking
account of stoichiometric coefficients, whereas ~G's must be properly
adjusted before combination? There is a significant difference between
EO and ~GO: G is an extensive property of the system so that when we
change the number of moles we are discussing, we must adjust G; E, on
the other hand, is an intensive variable-it is independent of the size of
the system-related to ~G by the extensive quantity nF.
§ 1.3 Some Uses of Standard Potentials 15
Example 1.1 Compute EO and fl.Go for the cell

Pt(s)II-(aq),I2(aq)IIFe2 +(aq),Fe3 +(aq)IPt(s)
(the double vertical line refers to a salt bridge used to separate
the two solutions).
Referring to Table A.4, we find the following data:

Fe3 + + e- - t Fe2+ EOFe3+/Fe2+ = 0.771 V
12 + 2 e- -t 2 1- EOI21I- = 0.536 V
To obtain the overall cell reaction, we multiply the first
equation by 2 and subtract the second:
2 Fe3+ + 2 1- -t 2 Fe2+ + 12
EO cell =EOFe3+/Fe2+ - EOI2 /I - = 0.235 V
fl.Go = -- 2FEo
t-.Go = -45.3 kJ mol- 1
Thus the oxidation of iodide ion by ferric ion is spontaneous
under standard conditions. The same standard free energy
change should apply for the reaction under nonelectro-
chemical conditions. Thus we can use the standard free
energy change computed from the cell potential to calculate the
equilibrium constant for the reaction of Fe3 + with 1- to give Fe2 +
and 12.
In K = -t-.GoIRT
K =8.6 x 107 = [Fe2+J2[I2]

[Fe 3+]2[I-]2
Because K is large, it might be difficult to determine directly by
measurement of all the constituent concentrations, but we
were able to compute it relatively easily from electrochemical
data. Indeed, most of the very large or very small equilibrium
constants we encounter in aqueous solution chemistry have
their origins in electrochemical cell potential measurements.
Example 1.2 Given the half-cell potentials

Ag+(aq) + e- -t Ag(s) EO = 0.7991 V
AgBr(s) + e- -t Ag(s) + Bdaq) EO =0.0711 V
compute the solubility-product constant for silver bromide.
Subtraction of the first half-cell reaction from the second gives

the AgBr solubility equilibrium
AgBr(s) ~ Ag+(aq) + Bdaq)
The standard potential of the hypothetical cell with this
reaction is
Ecell = EO AgBr/Ag - EO Ag+/Ag = -0.7280 V
I'-.Go = -nFEo = 70.24 kJ mol- 1
K = 4.95 x 10- 13 = aAg+<lBr-
While potentials of half-cells are simply subtracted to compute the

potential of a cell, there are seemingly similar calculations where this
approach leads to the wrong answer. Consider the half-cell processes
A + n e- ~ B EO AlB
B + m e- ~ C EOB/C
A + (n+m) eo ~ C
Clearly the last reaction is the sum of the first two, but EO AlC is not the
sum of EO AlB and EOB/C. It must be true that
!:J.GoAlC = !:J.GoAlB + I'-.GoB/C
Converting to standard potentials, we have
-(n+m)FEoAiC = -nFEoAiB - mFEoB/C
or
(1.23)
so that when half-cell potentials are combined to produce a new half-cell,

the potentials are not additive. The best advice for calculations involving
half-cell potentials is: When in doubt, convert to free energies before
doing the calculation. Since the desired result is often a free energy
change or an equilibrium constant, this strategy usually involves no
more computations and is much less likely to lead to errors.
Example 1.3 Compute the standard half-cell potential for the

reduction of Fe3 + to Fe(s) given EOFe3+/Fe2+ and EOFe2+/Fe. The
half-cell potentials from Table A.4 are
Fe3 + + eo ~ Fe2+ EO = 0.771 V
2
Fe + + 2 e- ~ Fe(s) EO =-0.44 V
These half-cell potentials are combined using eq (1.23) to obtain
the half-cell potential for the three-electron reduction of Fe3 +:
Fe3+ + 3 e- ~ Fe(s)
EOFe3+/Fe = t<EoFe3+/Fe2+ + 2E°Fe2+/Fe)
1
EOFe3+/Fe = 3(0.771-0.88) =-0.04 V
Formal Potentials
Standard potentials refer to standard states, which for solution
species are the hypothetical 1 M ideal solutions. Very dilute solutions
can be assumed ideal and calculations using standard potentials are
then reasonably accurate without activity coefficient corrections. For
electrolyte concentrations less than about 0.01 M, activity coefficients can
be computed reasonably accurately using Debye-Huckel theory (§ 2.6), but
for more concentrated solutions, empirical activity coefficients are
required.
One way around the problem of activity coefficients is through so-
called formal potentials. A formal half-cell potential is defined as the
potential of the half-cell when the concentration quotient of the Nernst
equation equals 1. Consider the Fe(III)IFe(II) couple. The Nernst
equation gives
or
E = EOFe3+IFe2+ In YF
D'f1
_ll.L
2+ D'f1T [Fe 2+]
_e__ ll.L I n - -
F YFe3 + F [Fe3+]
Thus when the concentrations of Fe3 + and Fe 2+ are equal, the last term
on the right-hand side is zero and the formal half-cell potential is
RT YFe2+
EO' = EOFe 3+IFe 2+- - I n - -
F YFe3 +
As we will see in §2.6, activity coefficients depend primarily on the total
electrolyte concentration (ionic strength) of the solution, so that in a
solution where the ionic strength is determined mostly by a high
concentration of an inert electrolyte, the activity coefficients are nearly
constant. Molar concentrations can then be used, together with formal
potentials appropriate to the medium, in calculations with the Nernst
equation. A representative sample of formal potentials for 1 M HCI04, 1
M HCI, and 1 M H2S04 solutions is given in Table A.6.
The formal potential of the Fe(III)IFe(II) couple in 1 M HCI04 is EO'
= 0.732 V, significantly different from the standard potential of 0.771 V,
suggesting that the activity coefficient ratio is about 0.22 in this medium.
In a medium with coordinating anions, e.g., aqueous HCI, Fe(II), and
Fe(III) form a variety of complexes. In order to compute the half-cell
potential from the standard potential, we would have to know not only
the activity coefficients but the formation constants of all the complexes
present. The formal potential of the Fe(III)lFe(II) couple in 1 M HCI, EO'
= 0.700 V, thus differs from the standard potential both because of activity
coefficient effects and because of chloro complex formation. As long as
the medium is constant, however, the relative importance of the various
complexes is constant and the formal potential can be used as an
empirical parameter to compute the overall Fe(II)lFe(III) concentration
ratio.
Latimer Diagrams
When we are interested in the redox chemistry of an element, a
tabulation of half-cell potential data such as that given in Table A.4 can
be difficult to assimilate at a glance. A lot of information is given and it
is not organized to give a qualitative understanding of a redox system.
Nitrogen, for example, exists in compounds having nitrogen oxidation
states ranging from -3 (NH3) to +5 (N03') and all intermediate oxidation
states are represented. One way of dealing with complex systems like
this is to use a simplified diagram introduced by Latimer (HI) and
usually referred to with his name. An example of a Latimer diagram is
shown in Figure 1.7 for the aqueous nitrogen system. In the Latimer
diagram, any H+, OH-, or H20 required to balance the half-cell reaction
is omitted for clarity. Thus, if we wish to use the half-cell potential for
the N03-/N204 couple, for example, we must first balance the equation
N03-+2H++e- ~ ~N204+H20 EO=0.80V
Figure 1.7 also includes a Latimer diagram for nitrogen species in basic
aqueous solution. The Latimer diagram tells us that one-electron
reduction of nitrate ion again produces N204, but the half-cell potentials
differ by 1.66 V. The reason is clear when we write the half-cell reaction.
The difference between the half-cell reactions is
2H20 ~ 2H++20H-
the Gibbs free energy of which is -2RT In K w , 160 kJ mol-lor 1.66 eV.
N03- + H20 + e- ~ ~ N204 + 2 OH- EO = -0.86 V
Wendell M. Latimer (1893-1955) was a student of G. N. Lewis and later a

Professor of Physical Chemistry at the University of California
(Berkeley). Latimer's contributions were primarily in applications of
thermodynamics to chemistry.
§1.3 Some Uses of Standard Potentials 19
0.80 1.07 1.00 1.59 1.77 -1.87 1.41 1.28
N03· ~ N20 4 ~ HN02 ~ NO ~ Nl0 ~ N2 ~ NHrOH+ ~N2H5 + ~
1,--_~0.","",94_--,t I 1.29 t. -0.05 L 1.35
-0.86 0.87 -0.46 0.76 0.94 -3.04 0.73 0.1

N03· ~ N20 4 ~ N02· ~ NO ~ Nl0 ~ N2 ~ NHrH ~N2H4 ~ NH3
I 0.01 tI 0.15 t -1.05 t 0.42 t
Figure 1.7 Latimer diagram showing the half-cell potentials for the various
nitrogen redox couples in acidic and basic aqueous solutions. Hydrogen or
hydroxide ions or water required to balance the half-cell reactions have been
omitted for clarity.
Example 1.4 Compute the half-cell potential for the reduction

of N03- to N02- in basic solution given the potential for the
reduction in acid solution, 0.94 V, and the ionization constants
of nitrous acid and water, pKa =3.3, pKw = 14.00.
The desired half-cell reaction is the sum of the following:

N03- + 3 H+ + 2 e- ~ HN02 + H20 !lGo = - 2FEoNOa-IHN02
HN02 ~ H + + N02- !lGo = 2.303 RT ~
2 H20 ~ 2 H+ + 20H- !lGo = 2 x 2.303 RTpKw
The standard free energy change for the desired half-cell then
IS
!lGo = - 2F(0.94) + 2.303 RT (3.3 + 28.00)

!lGo = -2700 J moP
EO = -!lGo/2F = 0.01 V
Free Energy - Oxidation State Diagrams
While Latimer diagrams compress a great deal of information into

a relatively small space, they are expressed in potentials which, as we
have seen, are not simply additive in sequential processes. Some
simplification is possible if the potentials are converted to free energy
changes relative to a common reference point. If we use zero oxidation
state (the element itself) as the reference point, then half-cell potentials
can be converted to a kind of free energy of formation where the species
of interest is formed from the element in its most stable form at 25°C and
from electrons, H+ or OH- ions, water, or other solution species. Thus we
could form HN02 from N2(g) by the following half-cell process:
2"1 N2(g) + 2 H20 -1 HN02 + 3 H+ + 3 e-
The free energy change for this process can be obtained from half-cell
potential data such as those given in Figure 1.7. Since the reductions of
HN02 to NO, NO to N20, and N20 to N2 all involve one electron per
nitrogen atom, the potentials can be added directly to obtain the
(negative) overall free energy change for the three-electron reduction;
changing the sign gives t'1.Go for the oxidation:
t'1.Go =F(1.77 + 1.59 + 1.00) =4.36 F
t'1.Go is usually expressed in kJ mol-I, but for the present purpose it is
easier to think of t'1.GolF, which is equivalent to putting !1Go in units of
electron-volts (1 eV = 96.485 kJ mol-I).
Similar calculations for the free energies of the other nitrogen
oxidation states can be done using the data of Figure 1. 7. These free
energies are plotted us. nitrogen oxidation number in Figure 1.8. Free
energy - oxidation state diagrams were introduced by Frost (1) and often
are called Frost diagrams. These diagrams were popularized in Great
Britain by Ebsworth (2) and sometimes are referred to as Ebsworth
diagrams.
Arthur Atwater Frost (1909-) was a professor of chemistry at Northwestern

University. He is best known for his work in chemical kinetics and
molecular quantum mechanics.
A free energy - oxidation state diagram contains all the information

of a Latimer diagram but in a form which is more easily
comprehensible for qualitative purposes. The slope of a line segment
connecting any two points, t'1.Go/t'1.n, is just the potential for the reduction
half-cell connecting the two species. Thus we can see qualitatively, for
example, that EO is positive for the reduction of NOa- to NH4+ in acid
solution but is negative for the reduction of NOa- to NHa in basic solution.
Species corresponding to minima, e.g., N2(g), NH4+ (acid solution),
N02-, and NOa- (basic solution), are expected to be thermodynamically
stable since pathways (at least from nearby species) are energetically
downhill to these points. Conversely, points that lie at maxima are
expected to be unstable since disproportionation to higher and lower
oxidation states will lower the free energy of the system. Thus, for
example, the disproportionation of N204 to N02- and NOa- is highly
exoergic in basic solution. The point for N204 in acid solution is not a
maximum, but does lie above the line connecting HN02 and NOa-; thus
disproportionation is spontaneous in acid solution as well.
:>
Q)
4
0- NH:t>H
e" 3
<I
-1
-3 -2 -1 o 1 2 3 4 5
Oxidation State
Figure 1.8 Free energy - oxidation state diagram for the nitrogen oxidation
states in acid (solid line) and basic (dashed line) solutions.
The Biochemical Standard State
Oxidation-reduction reactions play important roles in biochemistry

and half-cell potential data are often used in thermodynamic
calculations. The usual standard reduction potentials tabulated in
chemistry books or used in Latimer diagrams or free energy - oxidation
state diagrams refer, of course, to 1 M standard states and thus to pH 0
or pH 14, depending on whether we choose to balance the half-cell
reactions with H+ or OH- ions. Because life rarely occurs in strongly
acidic or strongly basic solutions, however, neither of these choices is
convenient for biochemical purposes, and biochemists usually redefine
the standard states ofH+ and OH- as pH 7, i.e., [H+]O = [OH-]O = 10- 7 M.
Biochemical standard free energy changes, standard potentials,
and equilibrium constants are usually distinguished from the
corresponding chemical standard quantities by writing /!.G', E', or Kin
place of /!.Go, EO, or K. A collection of biochemical standard potentials for
some half-cell reactions of biological interest is given in Table A.5.
With the standard states ofH+ and OH- defined as 10- 7 M, we must
be careful to recognize that the activities of these species are no longer
even approximately equal to their molar concentrations. The activity of a
solute i is defined (neglecting nonideality) as
ai = Ci/Cio
so that when Cia = 1 M, ai = Ci. But when Cia = 10- 7 M, ai = 10 7 Ci.
Consider the Nernst equation for the hydrogen electrode:
E = EO _RT In (aH2f
F aw
With H2(g) at unit activity, this expression can be written
E =EO +RT In aw
F
Using the chemical standard state and EO = 0.00 V, we have
E =RIln[H+]
F
With the biochemical standard state, the corresponding expression is
E = E' + RT In (107[H+]) (1.24)
F
For equal [H+], the two equations must give the same potential;
subtracting one equation from the other thus gives
E' = -RIln 107
F
or E' = -0.414 V at 25°C. Similarly, for a half-cell reaction
A + m H+ + n e- ~ B
E' and EO are related by
E' = EO -l11B.X..ln 107
nF
or, at 25°C,
E' = EO - 0.414 min
Similar relations can be derived between· f.G' and f.Go and between
1(' and K (see Problems). For an equilibrium reaction
A+mH+ ~ B
we find that
K = 1O- 7m K (1.25)
f.G' = f.Go + mRT In 10 7 (1.26)
or, at 25°C,
f.G'IkJ mol- 1 = f.Go + 40.0 m
Conversion back and forth between chemical and biochemical
standard states is straightforward as long as the reactions involve H + or
OH- in clearly defined roles and the necessary pKa data are available.
§1.3 Some Uses of Standard Potentials 23
Example 1.5 Given the biochemical standard potential for the

conversion of acetate ion and C02 gas to pyruvate ion,
CH3C02-(aq) + C02(g) + 2 H+(aq) + 2 e- ~ CH3COC02-(aq) + H20(l)
E' = -0.699 V, compute the chemical standard potential for the
analogous process involving the neutral acids. The pKa's of
acetic and pyruvic acids are 4.76 and 2.49, respectively.
The biochemical standard free energy change for the half-cell

reaction is flO' = -nFE' = 134.9 kJ mol-I; converting to the
chemical standard state using eq (1.26), we have
CH3C02- + C02 + 2 H+ + 2 e- ~ CH3COC02- + H20 (i)
with flO° = +55 kJ moP. The acid ionization steps are
CH3COOH ~ CH3C02- + H+ (ii)
CH3COCOOH ~ CH3COC02- + H + (iii)
with flO° =-2.303 RT pKa =27.2 and 14.2 kJ moP, respectively.
Adding eqs (i) and (ii) and subtracting eq (iii), we have
CH3COOH + C02 + 2 H+ + 2 e- ~ CH3COCOOH + H20
and flO° = 55.0 + 27.2 - 14.2 = 68.0 kJ moP, E' =-D.352 V.
Unfortunately, it is not always clear exactly how many H+ or OH-

ions are involved in a half-cell reaction. The problem is that many
species of biochemical interest are polyelectrolytes having weak acid
functionalities with pKa's near 7. For example, the pKa ofH2P04- is 7.21,
so that at pH 7, [HP04 2-]I[H2P04-] = 0.62. When phosphate is involved in a
half-cell reaction, it is often bound (esterified) to give a species with a pKa
similar but not identical to that of H2P04-. Thus the exact number of
protons involved in the electrode process is a complex function of pH
involving several sometimes poorly known pKa's.
The problem is somewhat simplified by specifying the standard
state of inorganic phosphate as the pH 7 equilibrium mixture of H2P04-
and HP042- with a total concentration of 1 M. Phosphate esters and
other weak acids are given similar standard states. Calculations based
on biochemical standard potentials then give unambiguous results at
pH 7, but because the number of protons involved in a half-cell reaction
is ambiguous and pH-dependent, it is difficult, if not impossible, to
correct potentials to another pH.
Example 1.6 Calculate the biochemical standard free energy

change and equilibrium constant for the reaction of 3-
phosphoglyceraldehyde with nicotinamide adenine

dinucleotide (NAD+) and inorganic phosphate to give 1,3-
diphosphoglycerate. Estimate the chemical standard free
energy change and equilibrium constant.
The half-cell potentials from Table A.5 are

RC02P032- + 2 H+ + 2 e- -7 RCHO + HP042- E' = -0.286 V
NAD+ + H+ + 2 e- -7 NADH E' =-0.320 V
where R = CHOHCH20P032- and all species are understood to
be pH 7 equilibrium mixtures. The cell reaction is
RCHO + HP042- + NAD+ -7 RC02P032- + H+ + NADH
and the standard cell potential at pH 7 is
E' = (-0.320) - (-0.286) = - 0.034 V
The free energy change and equilibrium constant at pH 7 are
I1G' = -nFE'
110' = +6560 J mol- l (6.56 kJ mol-l)
K' = exp( -I1G'/RT)
so that
K' = 0.071 = aRCOJ>03 2 . aw aNADH
aRCHO aHP0 4 2- aNAD+
or
K' = [RC02P0 32-] (107[H+]}[NADH]
[RCHO] [HPoi-] [NAD+]
The cell reaction involves four different sets of phosphate
pKa's. If they are similar, the corrections will approximately
cancel; thus we use eqs (1.25) and (1.26) obtain estimates of K
and I1Go:
K", 7.1 x 10-9
I1Go '" 46.6 kJ mol- l
This result shows one reason for the use of a special standard
state for H+ and OH- in biochemistry. The reaction of NAD+
with 3-phosphoglyceraldehyde appears to be hopelessly
endoergic under chemical standard conditions. The values of
I1GO and K thus are misleading since at pH 7 the equilibrium
constant is not so very small and significant amounts of
product are expected, particularly if the phosphate
concentration is high compared with the other species.
Potential - pH Diagrams
Chemists interested in reactions at pH 0 or pH 14 or biochemists

willing to stay at or near pH 7 are well served by tables of half-cell
potentials, Latimer diagrams, or free energy - oxidation state diagrams.
For systems at other pH values, the Nernst equation gives us a way of
correcting potentials or free energies. Consider again the general half-
cell process
A + m H + + n e- -7 B
with potential
E=Eo_RTln aB
nF aA(aw)"'
When the A and B activities are equal, we have
E =Eo - 2.303 mRT pH
nF
Thus a half-cell potential is expected to be linear in pH with a slope of
-59.2 (min) mV per pH unit at 25°C. Electrode processes involving a
weak acid or weak base have potential - pH variations which show a
change in slope at pH = pKa. For example, the reduction of N(V) to
N(IlI) in acid solution is
N03- + 3 H + + 2 e- -7 HN02 + H20
so that the E us. pH slope is -89 mV pH-I. In neutral or basic solution,
the process is
or
N03- + H20 + 2 e- -7 N02- + 20H-
so that dEldpH = -59 mV pH-I. The N(V)/N(III) half-cell potential is
plotted us. pH in Figure 1.9. Plots of the half-cell potentials of the
N(III)/N(O) and N(O)/N(-III) couples are also shown in Figure 1.9,
together with the 02iH20 and H201H2 couples.
The significance of the 02/H20 and H20/H2 couples is that these
define the limits of thermodynamic stability of an aqueous solution. Any
couple with a half-cell potential greater than that of the 021H20 couple is
in principle capable of oxidizing water. Similarly, any couple with a
potential less than (more negative than) that of the H201H2 couple is in
principle capable of reducing water. As it happens, there are many
species which appear to be perfectly stable in aqueous solution which are

part of couples with half-cell potentials greater than that of the 021H20
couple or less than that of the H20/H2 couple. This is a reflection,
however, of the intrinsically slow 021H20 and H201H2 reactions rather
than a failure of thermodynamics. The reactions do not occur because
there is a very large activation barrier, not because there is no overall
driving force.
Predominance Area Diagrams
We see in Figure 1.9 that nitrous acid and the nitrite ion are in
principle capable of oxidizing water and are thus thermodynamically
unstable in aqueous solution. Although this instability is not manifested
because the reactions are slow, it raises the related question: What is the
thermodynamically most stable form of nitrogen at a given pH and
electrode potential? Referring to the free energy - oxidation state
diagram, Figure 1.8, we see that for both acidic and basic solutions the
points for N(-l) and N(-II) lie above the line connecting N(O) and N(-III);
these species thus are unstable with respect to disproportionation.
Similarly, the points for N(I), N(Il), N(lII), and N(IV) all lie above the
N(O) - N(V) line, so that these species are also unstable. The existence of
any of these species in aqueous solution therefore reflects kinetic
stability rather than thermodynamic stability. In fact, only N03-, NH3,
NH4+, and N2 are thermodynamically stable in aqueous solution.
Furthermore, according to thermodynamics, N03- is unstable in the
presence of NH4 + or NH3. One way of expressing the conclusions
regarding thermodynamic stability is by means of a predominance area
diagram, developed by Pourbaix (3,H5) and thus commonly called a
Pourbaix diagram.
Marcel Pourbaix (1904-) was a professor at the Free University of

Brussels and Director of Centre BeIge d'Etude de la Corrosion.
A Pourbaix diagram is a potential - pH plot, similar to that of

Figure 1.9, in which regions of thermodynamic stability are identified.
Lines separating regions represent the potential and pH at which two
species are in equilibrium at unit activity. Pourbaix diagrams usually
also include the region of water stability. The Pourbaix diagram for
nitrogen is shown in Figure 1.10. We see that in most of the potential -
pH area of water stability the most stable form of nitrogen is N2(g). At
low potential and low pH, NH4+ is most stable and at low potential and
high pH NH3 is most stable. Nitrate ion is stable in a narrow high-
potential region above pH 2. Below pH 2, N03- is in principle capable of
oxidizing water, though the reaction is so slow that we normally do not
worry about the decomposition of nitric acid solutions.
1.5 1.5
EIV
N03-(aq)
1.0 1.0
0.5 0.5
0.0 0.0
-0.5 -0.5
-1.0 -1.0
024 6 8 ill ~ M o 2 4 6 8 W ~ M
pH pH
Figure 1.9 Potential - pH diagram Figure 1.10 Predominance area
showing the variation in half-cell diagram (Pourbaix diagram) for
potentials with pH for the nitrogen in aqueous solution. The
N(III)/N(O), N(V)/N(III), and labeled regions correspond to
N(O)IN(-III) couples. Note changes areas of thermodynamic stability
in slope of the first two lines at pH for the indicated species. Water is
3.14 (the pKa of HN02) and of the stable between the dashed lines
third line at pH 9.24 (the pKa of showing the 02/H20 and H20/H2
NH4+). Also shown (dashed lines) couples.
are the potentials of the 02/H20
and H20/H2 couples.
Pourbaix diagrams for kinetically stable systems like the oxidation

states of nitrogen are not particularly useful. Most of the interesting
chemistry involves species which are thermodynamically unstable and
therefore do not appear on the diagram. For more labile redox systems,
on the other hand, where thermodynamic stability is more significant, a
Pourbaix diagram can be very useful in visualizing the possibilities for
aqueous chemistry; we will see an example of this kind in Chapter 7.
1.4 MEASUREMENT OF CELL POTENTIALS
In order to obtain cell potentials which have thermodynamic

significance, the potential must be measured under reversible
conditions. In thermodynamics, a reversible process is one which can
be reversed in direction by an infinitesimal change in the conditions of
the surroundings. For example, the direction of a reversible chemical
reaction might be changed by an infinitesimal increase in the

concentration of a reactant or product. Similarly, an electrochemical
cell is reversible if the direction of current flow in the external circuit
can be reversed by an infinitesimal change in one of the concentrations.
Alternatively, we can think in terms of an operational definition of
electrochemical cell reversibility: An electrochemical cell is regarded as
reversible if a small amount of current can be passed in either direction
without appreciably affecting the measured potential.
The words "small" and "appreciable" are ambiguous; the meaning
depends upon the context of the experiment. A cell which appears
reversible when measured with a device with a high input resistance
(and thus small current) may appear quite irreversible when measured
by a voltmeter with a low input resistance. There are a number of
causes of irreversible behavior in electrochemical cells which are
discussed in more detail in Chapter 5. Irreversible behavior usually
results from an electrode process having slow reaction kinetics, in the
electron transfer process itself, in a coupled chemical reaction, in the
delivery of reactants to the electrode, or the removal of products from the
electrode. For the purposes of discussion of cell potential
measurements, however, we can think of an electrochemical cell as
having an internal resistance which limits the current which can be
delivered. Strictly speaking, in any reasonable model the resistance
would be nonohmic, i.e., current would not be linear in potential. We
can, however, use a simple ohmic model to show one of the consequences
of irreversibility.
Consider an electrochemical cell with an internal resistance Rint
connected to a voltage-measuring device having an input resistance
Rmeter as shown in Figure 1.11. If we could draw zero current, the
voltage across the cell terminals would be the true cell potential Ecell. In
practice, however, we draw a current
i = Ecell
R int + Rmeter
and actually measure a voltage
E = iRmeter
Combining these expressions, we get
E = EcellRmeter
Rint + R meter
If the measured voltage is to be a good approximation to the cell
potential, we must have Rmeter »Rint. Our perception of the
reversibility of a cell thus depends on the measuring instrument.
§1.4 Measurement of Cell Potentials 29
r------,
R int :
Figure 1.11 Circuit showing: cell
the effects of cell resistance I
and leakage resistance on I
potential measurements. L ______ ...J
For cells having a high internal resistance (>10 6 Q) another

practical problem may arise even if a voltmeter is available with a high
input resistance. In very high resistance circuits, leakage of current
between the meter terminals becomes critically important. Dust, oily
films, or even a fingerprint on an insulator can provide a current path
having a resistance of 10 7 - 109 Q. This resistance in parallel with the
meter (as shown in Figure 1.11) results in an effective meter resistance
equal to the parallel combination of the true meter resistance and the
leakage resistance,
_1_=_1_+_1_
Reff R meter RJeak
which may be several orders of magnitude less than the true meter
resistance and even comparable with the cell internal resistance.
The problem of leakage resistance is generally encountered for any
cell which has a high internal resistance and thus tends toward
irreversibility. In particular the glass electrode commonly used for pH
measurements has a very high resistance and is prone to just such
leakage problems. Careful experimental technique with attention to
clean leads and contacts is essential to accuracy.
Potentiometers
The classical method of potential measurement makes use of a

null-detecting potentiometer. A potentiometer, shown in Figure 1.12,
involves a linear resistance slide wire R calibrated in volts, across which
is connected a battery E; the current through the slide wire is adjustable
with a rheostat R'. With a known potential source Es (often a Weston
standard cell) connected to the circuit, the slide wire is set at the
potential of the standard cell and the rheostat is adjusted to give zero-
current reading on the galvanometer G. The potential drop between A
and the slidewire tap is then equal to E s , and, if the slidewire is linear in
resistance, the voltage between A and any other point on the slidewire
can be determined. The switch is then thrown to the unknown cell, Ex,
and the slidewire adjusted to zero the galvanometer. The unknown
potential can then be read from the slidewire calibration.
...------IIII---..-'C"""
E
Figure 1.12 A potentiometer circuit.
Reversible cell potentials measured with a potentiometer are usually

accurate to ±0.1 mV and with care can be even better. The apparatus is
relatively inexpensive, but the method is slow, cumbersome, and
requires some technical skill. Furthermore, although in principle the
potential is read under zero current conditions, some current must be
drawn in order to find the zero. The current is small, typically on the
order of 10-8 A with a sensitive galvanometer, but may be large enough to
cause serious errors for a cell with high internal resistance.
Electrometers
With the development of vacuum tube circuitry in the 1930's, d.c.
amplifiers with very high input impedances became available. Such
devices, called electrometers, are particularly well suited to
measurement of the potentials of cells with high internal resistance.
A number of different designs have been used for electrometer
circuits, but one which is particularly common in electrochemical
instrumentation is the so-called voltage follower shown in Figure 1.13.
A voltage follower employs an operational amplifier, indicated by the
triangle in the figure. The output voltage of an operational amplifier is
proportional to the difference between the two input voltages with very
high gain (> 104 ). In a voltage follower, the output is connected to the
negative (inverting) input. Suppose that the output voltage is slightly
greater than that at the positive input. The difference between the inputs
will be amplified, driving the output voltage down. Conversely, if the
output voltage is low, it will be driven up. The output is stable, of course,
when the output and input are exactly equal. Since the input impedance
§ 1.4 Measurement of Cell Potentials 31
is high (> 10 10 0), a very small current is drawn from the cell. Since the
output impedance is low (ca. 10 0), a voltage-measuring device such as a
meter or digital voltmeter can be driven with negligible voltage drop
across the output impedance of the voltage follower. Direct-reading
meters typically are limited to an accuracy on the order of 0.1% of full
scale (±1 mV for a full-scale reading of 1 V), but the accuracy can be
improved considerably by using an electrometer in combination with a
potentiometer circuit.
Figure 1.13 An operational

amplifier voltage follower.
1.5 REFERENCE AND INDICATOR ELECTRODES
The standard half-cell potentials of all aqueous redox couples are

given with respect to the hydrogen electrode, the primary reference
electrode. Unfortunately, however, the hydrogen electrode is awkward
and inconvenient to use, requiring hydrogen gas and a specially
prepared platinum electrode. The platinum surface is easily poisoned
and other electrode processes compete with the H+/H2 couple in
determining the electrode potential. For these reasons, other electrodes
are more commonly used as secondary references (F1).
Reference Electrodes
A practical reference electrode should be easily and reproducibly
prepared and maintained, relatively inexpensive, stable over time, and
usable under a wide variety of conditions. Two electrodes-the calomel
and silver-silver chloride electrodes-are particularly common, meeting
these requirements quite well.
The calomeP electrode is shown in Figure 1.14 and represented in
shorthand notation as follows:
1 Calomel is an archaic name for mercurous chloride, Hg2C12.

(a) ___--lead wire

(b)
lead wire
filling hole
KCI solution
Hg/Hg 2CI 2 emulsion

Ag wire coated
with AgCI
junction fiber junction

Figure 1.14 Reference electrodes. (a) Saturated calomel electrode (b)
Silver-silver chloride electrode.
The half-cell reaction is

Hg2CI2(S) + 2 e- ~ 2 Cl-(aq) + 2 Hg(l) EO = +0.2682 V
The standard potential, of course, refers to unit activity for all species,
including chloride ion. In practice it is often convenient to use saturated
KCl solution as the electrolyte with a few crystals of solid KCI present in
the electrode to maintain saturation. In this way the chloride ion
concentration is held constant from day to day and it is not necessary to
prepare a solution of exactly known concentration. The potential of the
saturated calomel electrode (abbreviated s.c.e.) equipped with a saturated
KCI salt bridge is 0.244 V at 25°C. Because the solubility of KCI is
temperature dependent, there is a significant variation in potential with
temperature, -0.67 mV K-1. When this is a problem, 0.1 M KCI can be
used as the electrolyte; the potential and temperature coefficient then are
0.336 V and -0.08 mV K-l, respectively.
The silver-silver chloride electrode is also shown in Figure 1.14. Its
shorthand notation,
CI-(aq)IAgCl(s)IAg(s)
leads to the half-cell reaction
AgCI(s) + e- ~ Ag(s) + CI-(aq) EO = +0.2223 V
The Ag/AgCI electrode normally is used with 3.5 M KCI solution and has
a formal half-cell potential of 0.205 V and a temperature coefficient of
-0.73 mV K-l. The Ag/AgCI electrode is operationally similar to the
calomel electrode but is more rugged; AgCI adheres very well to metallic
§ 1.5 Reference and Indicator Electrodes 33
silver and there is no liquid mercury or Hg2Cl2 paste to deal with. The
Ag/AgCI electrode is easily miniaturized and is thus convenient for
many biological applications.
Since K+ and CI- have nearly the same ionic conductivities, liquid
junction potentials are minimized by the use of KCI as a salt bridge
electrolyte (see Sect. 3.4). Both the calomel and Ag/AgCI electrodes are
normally filled with KCI solution, but there are occasions when other
electrolytes are used. This matter is discussed further in Sect. 4.2.
Indicator Electrodes
If one of the electrodes of an electrochemical cell is a reference
electrode, the other is called a working or indicator electrode. The latter
designation implies that this electrode responds to (indicates) some
specific electrode half-reaction. At this point, it is appropriate to ask
what determines the potential of an electrode. Consider a solution
containing FeS04 and H2S04 in contact with an iron wire electrode. If
we make a cell by adding a reference electrode, there are at least three
electrode processes which might occur at the iron wire:
Fe 2+ + 2 e- --7 Fe(s) EO = -0.44 V
S042- + 4 H+ + 2 e- --7 S02(aq) + 2 H20 EO = 0.16 V
1
H+ + e- --7 2H2(g) EO = 0.00 V
What then is the actual potential of the iron wire electrode? We first note
that in order to have a finite reversible potential at an electrode, all the
participating species must be present in finite concentration so that
appreciable current can be drawn in either direction. Thus, as we have
defined the system with no sulfur dioxide or gaseous hydrogen present,
only the first electrode reaction qualifies and we might guess that the
potential should be determined by the Fe 2 +lFe couple. This is the right
answer but for the wrong reason. Suppose that we bubbled some
hydrogen gas over the iron wire electrode or added a little sodium
sulfite-what then? There is another way to look at this system:
according to the half-cell potentials, the iron wire should be oxidized
spontaneously by either H+ (t1Go = -85 kJ mol-I) or by S042- (t1Go = -116 kJ
mol-I). That neither reaction occurs to any appreciable extent is because
the reactions are very slow. Just as these homogeneous reactions are
very slow, the electrochemical reduction of S042- or H+ at an iron
electrode is slow. If one electrode reaction is much faster than other
possible reactions, only the fast reaction will contribute to the potential-
the slower electrode processes will appear irreversible under the
conditions where we can measure the potential due to the faster process.
Thus in the example, we expect the potential of the iron wire electrode
will be determined by the Fe2+lFe couple with
or, at 25°C,
EN = -0.440 - 0.0296 pFe
In other words, the iron wire electrode acts in this case as an indicator
ofthe Fe2+ activity.
Although it is difficult to predict the rates of electrode processes
without additional information, there are a few useful generalizations.
First, electrode processes which involve gases are usually very slow
unless a surface is present which catalyzes the reaction. Thus a surface
consisting of finely divided platinum (platinum black) catalyzes the
reduction of H+ to H2, but this process is slow on most other surfaces.
Second, as a general rule, simple electron transfers which do not involve
chemical bond breaking, e.g., Fe 3 + + e- ~ Fe 2+, are usually fast
compared with reactions which involve substantial reorganization of
molecular structure, e.g., the reduction of sulfate to sulfite.
What then should we expect if the components of two reversible
couples are present in the cell? Consider, for example, a solution
containing Fe3+, Fe 2+, 13-, and 1-. Both half-cell reactions,
Fe3 + + e- ~ Fe2+ EO = 0.771 V
EO = 0.536 V
are reversible at a Pt indicator electrode. If both couples are reversible,
the electrode potential must be given by both Nernst equations
E = EOF 3+1F 2+ -l1:I.ln aFe 2+
e e F aFe3+
E = EOI -11- _RT In (ad3

3 2F ali
The Fe3+lFe 2+ half-cell reaction thus proceeds to the right and the 13-/1-
half-cell reaction proceeds to the left until both half-cells have the same
potential. In other words, there is a constraint on the activities of the
four species
EOFe3+lFe2+ - E0Ia-II- = RT In (aI:iXaFe 2+)2
2F (ad(aFe3+)2
The argument of the logarithm is just the equilibrium constant
expression for the homogeneous reaction
2 Fe3+ + 3 1- ~ 2 Fe 2+ + 13-
and indeed in most cases, equilibration occurs via the homogeneous
reaction. Frequently this means that one of the components of the
mixture is consumed (reduced to a very low concentration). It is then
most convenient to compute the potential of the indicator electrode from
§ 1.5 Reference and Indicator Electrodes 35
the activities of the survivors, but the other couple remains at
equilibrium and the concentration of the minor component can be
calculated. All this, of course, is precisely what is happening in a
potentiometric titration and we will return to calculations of this kind in
§1.7.
1.6 ION-SELECTIVE ELECTRODES
The potential of a Ag/AgCI electrode depends on [CI-] and so we

might think of the Ag/AgCI electrode as an indicator electrode, the
potential of which is a measure ofln[CI-], or pCl. We could easily set up
an electrochemical cell to take advantage of this property. Suppose that
we have two Ag/AgCI electrodes, one a standard reference electrode with
KCI solution at unit activity, the other in contact with a test solution
having an unknown CI- activity. The cell can be represented
schematically by
AgIAgCl(s)IKCI(aq,aO)IICI-(aq)IAgCl(s)IAg
At the left-hand electrode, the half-cell reaction is
Ag(s) + CI-(aq,aO) ~ AgCI(s) + e-
and at the right-hand electrode, the process is
AgCl(s) + e- ~ Ag(s) + CI-(aq)
so that the overall cell reaction is
CI-(aq,aO) ~ CI-(aq)
In other words the cell "reaction" is simply the dilution of KCl. The
potential of the cell is given by the Nernst equation:
E = EO - RI. In .a...
F aO
or, since EO = 0 and aO = 1, the cell potential at 25°C is
EN = +0.0592 pCI
This arrangement would work well as a technique for determination of
unknown CI- activities. However, the Ag/AgCI electrode is not
particularly selective. It will respond, for example, to any anion which
forms an insoluble silver salt (e.g., Br or SCN-).
Glass Membrane Electrodes
It would be nice to have an indicator electrode which would respond

to only one specific ion. This could be accomplished if we had a
membrane permeable to only one species. A membrane with perfect

selectivity is yet to be found, but there are quite a number of devices
which come close. 1 There are several approaches to membrane design,
and we will not discuss them all. The 0ldest2 membrane electrode is the
glass electrode, which has been used to measure pH since 1919 (4) but
properly understood only relatively recently. The electrode, shown in
Figure 1.15, consists of a glass tube, the end of which is a glass
membrane about 0.1 mm thick (and therefore very fragile!). Inside the
tube is a Ag/AgCI electrode a:rid 1 M HCI solution. The glass electrode is
used by dipping it into a test solution and completing the electrochemical
cell with a reference electrode.
The glass used in the membrane is a mixture of sodium and
calcium silicates-Na2Si03 and CaSi03-and silicon dioxide, Si02. The
silicon atoms tend to be four-coordinate, so that the glass is an
extensively cross-linked polymer of Si04 units with electrostatically
bound Na+ and Ca 2 + ions The glass is weakly conductive, with the
charge carried primarily by the Na+ ions. (The Ca2+ ions are much less
mobile than Na+ and contribute little to the conductance.) The glass is
also quite hygroscopic and takes up a significant amount of water in a
surface layer perhaps as much as 0.1 11m deep. In the hydrated layers
(one on either side of the membrane) there is equilibrium between H+
and Na+ electrostatically bound to anionic sites in the glass and in
solution.
H+(aq) + Na+(gl) f:Z H+(gl) + Na+(aq) (1.27)
If the concentration of H+(aq) is low, this equilibrium shifts to the left;
N a + from the interior of the glass tends to migrate into the hydrated
region to maintain electrical neutrality. Hydrogen ions on the other side
ofthe glass penetrate a little deeper into the glass to replace the Na+ ions
that have migrated. This combination of ion migrations gives sufficient
electric current that the potential is measurable with a high-impedance
voltmeter. Since the H+ ions are intrinsically smaller and faster moving
than Na+, most of the current in the hydrated region is carried by H+
and the glass electrode behaves as if it were permeable to H+ and thus
acts as an indicator electrode sensitive to pH.
In a solution with low [H+] and high [Na+], NaOH solutions for
example, the Na+ concentration in the hydrated layer of the glass may be
much greater than the H+ concentration, and Na+ ions then carry a
1 See books by Koryta (All), Vesely, Weiss, and Stulik (DIl), and by Koryta and
Stulik (D 14) for further details.
2 The first membrane potential was discovered by nature eons ago when animals
first developed nervous systems. A nerve cell wall can be activated to pass Na+ ions
and so develop a membrane potential which triggers a response in an adjacent cell.
Chemists were slower in appreciating the potential of such a device, but we are
catching up fast.
§1.6 Ion-Selective Electrodes 37
(a) (b)
lead wire
internal
solution
dry glass layer

insulation ca. 0.1 mm,
conductance by
Na+ ions
Ag wire coated
with AgCI external
hydrated glass layers solution
Hel solution ca. 0.1 11m thick
conductance by
thin glass H+ and Na+ ions
membrane
Figure 1.15 (a) pH-sensitive glass electrode; (b) schematic view of the glass
membrane.
significant fraction of the current; the potential developed across the

membrane then is smaller than might have been expected. For this
reason, glass electrodes do not give a linear pH response at very high pH,
particularly when the alkali metal ion concentration is high Glass
electrodes respond with virtually perfect selectivity to hydrogen ions over
the pH range 0-11. Above pH 11, response to alkali ions becomes
important with some glasses and such electrodes become unusable
above pH 12.
We will see in Section 3.4 that the potential across a glass
membrane can be written
Ememlrane = constant + RJ In(aH + kH,N aaN a) (1.28)
where aH and aNa are the activities of H+(aq) and Na+(aq) and kH Na is
called the potentiometric selectivity coefficient. The selectivity coefficient
depends on the equilibrium constant for the reaction of eq (1.27) and on
the relative mobilities of the H+ and Na+ ions in the hydrated glass.
These properties depend on the composition and structure of the glass
and can be controlled to some extent. Thus some glass electrodes now
available make use of glasses where lithium and barium replace
sodium and calcium, giving a membrane which is much less sensitive
to sodium ions, allowing measurements up to pH 14. Corrections for
Na+ ions may still be required, however, above pH 12.
Because of the importance of the hydration of the glass surface,

glass electrodes must be conditioned before use by soaking in water or an
aqueous buffer solution. The glass surface is dehydrated on prolonged
exposure to nonaqueous solvents or to aqueous solutions of very high
ionic strength. Nonetheless, glass electrodes can be used (e.g., to follow
acid-base titrations) in alcohols or polar aprotic solvents such as
acetonitrile or dimethyl sulfoxide provided that the exposure is of
relatively short duration.
Glasses containing about 20% Al203 are significantly more
sensitive to alkali metal ions and can be used to produce electrodes
which are somewhat selective to specific ions. Thus a Na20 - Al203 -
Si02 glass can be fabricated which is selective for sodium ions (e.g.,
kNa K = 0.001). Although these electrodes retain their sensitivity to
hyd~ogen ion (kNa H = 100), they are useful in neutral or alkaline
solutions over the pNa range 0-6.
Other Solid Membrane Electrodes

Rapid progress has been made in recent years in the design of ion-
selective electrodes which employ an insoluble inorganic salt as a
membrane. For example, a lanthanum fluoride crystal, doped with a
little EuF2 to provide vacancies at anionic sites, behaves like a
membrane permeable to F- ions. The only significant interference is
from OR- ion. F- and OR- are almost exactly the same size, but other
anions are too large to fit into the F- sites in the crystal. The LaF3
crystal, together with a solution of KF and KCI and a Ag/AgCI electrode,
gives a fluoride ion-selective electrode usable in the pF range 0-6.
Other solid membrane electrodes make use of pressed pellets of
insoluble salts such as Ag2S. In this case, Ag+ ions are somewhat
mobile in the solid, so that a Ag2S membrane can be used in a Ag+-
selective electrode. Other metals which form insoluble sulfides could in
principle replace a Ag+ ion at the surface of the membrane, but in
practice only Rg2+ is a serious interference. Since Ag+ can move in
either direction through the membrane, either away from a source of
Ag+ or toward a sink-a source of S2--the Ag2S membrane electrode
can be used either for measurement of pAg or pS in the range 0 to 7.
Other solid membrane electrodes are commercially available for CI-, Br,
1-, CN-, SCN-, NH4+, Cu2+, Cd2+, and Pb 2+.
Liquid Membrane Electrodes

By replacing the glass or inorganic crystal with a thin layer of a
water-immiscible liquid ion exchanger, another type of ion-selective
electrode may be constructed. For example, by using the calcium salt of
an organophosphoric acid in an organic solvent as the liquid ion
exchanger and contacting the ion-exchange solution with the aqueous
§ 1.6 Ion-Selective Electrodes 39
test solution through a thin porous membrane, a membrane system

permeable to Ca 2+ ions (and to some degree to other divalent ions) is
obtained. On the inner side of the membrane is a Ag/AgCI electrode
with CaCl2 aqueous electrollte. Thus a membrane potential
proportional to In([Ca 2+]out/[Ca +]in) is developed. Liquid membrane
electrodes are commercially available for Ca 2+, K+, CI-, NOa-, CI04-, and
BF4-.
Other Types of Selective Indicator Electrodes

By surrounding the glass membrane of an ordinary glass electrode
with a dilute aqueous solution of NaHCOa which is separated from the
test solution by a membrane permeable to C02, an electrode responsive to
dissolved C02 is obtained. This technique, where an ion-selective
electrode is converted to respond to some other species by interposition of
a reaction system involving ordinary chemical reactions, is extendable to
a wide variety of applications. Electrodes are commercially available for
dissolved NHa and N02, as well as for C02.
By incorporating an enzyme in a membrane, this approach can be
extended to the detection of many other species. For example, a urea-
sensitive electrode can be constructed by immobilizing the enzyme
urease in a thin layer of polyacrylamide on a cation-sensing glass
electrode. The electrode responds to NH4 +, produced by the urease-
catalysed hydrolysis of urea. Similarly, by coating a glass electrode with
immobilized L-amino acid oxidase, an electrode is obtained which
responds to L-amino acids in solution. Clearly, the field of ion-selective
electrodes is large, and, more significantly, is growing rapidly. 1 There
are literally thousands of potential applications in chemistry,
biochemistry, and biology which await an interested investigator.
1. 7 CHEMICAL ANALYSIS BY POTENTIOMETRY
There are a number of analytical methods based on measurements

of electrochemical cell potentials (7,8,D2,D6,D7,DIO,Dll,D13,D14). For
convenience these can be divided into two groups: those which
determine concentration (or activity) directly from the measured
potential of an electrochemical cell; and those in which the potential of a
cell is used to determine the equivalence point in a titration. Both types
have some important advantages which will become apparent as we
discuss them.
1 For a good review, see Murray (5). The field is reviewed every two years in
Analytical Chemistry, e.g., by Janata in 1992 (6).
Direct Methods
The measurement of pH using a glass electrode is by far the most
common of all electroanalytical techniques (9,D7,DI0) and is also
characteristic of direct determinations using cell potentials. A typical
electrode arrangement for pH measurements is as follows:
Ag(s)IAgCI(s)IKCI( 1 M)IIH +(aq)lglassIHCl(1 M)IAgCl(s)IAg(s)
i.e., a solution in which a Ag/AgCl reference electrode and a glass
electrode are immersed. The half-cell reactions,
Ag(s) + Cl-(1 M) ~ AgCI(s) + e-
AgCI(s) + e- ~ Ag(s) + CI-(1 M)
lead to no net change so that the overall cell reaction is the nominal
transfer of H+ from the test solution to the 1 M solution inside the glass
membrane
H+(aq) ~ H+(aq, 1 M)
The free energy change for this process is
I'1G=RTln-1-
aH+
so that the cell potential is
E=RI..ln aw
F
or, at 25°C,
EN = -0.0592 pH
Thus the cell potential is directly proportional to the pH of the test
solution. In practice, this may not be exactly true because of a liquid
junction potential at the salt bridge linking the reference electrode to the
test solution and a small potential intrinsic to the glass electrode. Thus
buffers of known pH are used to calibrate the pH meter so that the pH is
determined relative to a standard rather than absolutely.
Most of the commercial ion-selective electrodes can be used directly
to determine pX = -log ax, just as the glass electrode is used to measure
pH. Two general procedures are used commonly.
(1) Calibration curve. The potential of a given cell can be
measured with standard solutions and a potential us. concentration
curve constructed. The concentration of an unknown may then be read
from the calibration curve given the potential of the unknown. When
this method is used, it is important that the unknown solution have the
same ionic strength as that of the standards so that the activity
§ 1. 7 Chemical Analysis by Potentiometry 41
coefficients remain constant. The concentrations of other species, such

as complexing agents, should also be identical.
(2) Standard addition method. By first measuring the potential
of an unknown solution and then adding a known amount of the
substance detected, the incremental response of the voltmeter in effect
calibrates the scale. The advantages of this procedure are that the
concentration is determined without the tedium of a calibration curve,
and more important, calibration is obtained on the same solution under
the same measurement conditions, possibly avoiding some systematic
errors. On the other hand, if the calibration curve was not linear, the
known addition method would introduce a large uncorrectable error.
As a rule, the calibration curve method is preferable if many
analyses are to be done. If only a few analyses are to be performed, the
known addition method is usually faster. In either case, the method
should have been carefully investigated to ferret out systematic errors.
Example 1.7 100 mL of a solution containing an unknown

concentration of fluoride was analyzed using a fluoride ion-
selective electrode and a Ag/AgCl reference electrode. The cell
potential was -97 mV at 25°C. After addition of 10 mL of 2.00 x
10-3 . M F- standard solution, the potential was -70 mY. What
was the original concentration of fluoride? If the potentials are
accurate to ±0.5 mY, what is the uncertainty in the
determination?
The Nernst equation gives the cell potential

E =Eo _RT In -.L-Eref
F aF"
so that
-0.0970 ± 0.0005 = ~ - 0.0592 pFO - Eref
-0.0700 ± 0.0005 = ~ - 0.0592 pFl - Eref
Subtracting, we get!
pFo - pFl = (0.0270 ± 0.0007)/0.0592 = 0.456 ± 0.012
pFO - pFl = log(al/ao)
Assuming that we can replace the activities by molar
concentrations, we have
[F-hI[F-]O =antilog(0.456) = 2.86 ± 0.08
The difference, [F-] - [F-]o, corresponds to the number of moles
ofF- added.
1 In addition or subtraction, Z =x ± y, the errors propagate as sz2 = sx2 + sy2"

[F-h - [F-]o = (2.00 x 10-3 mol L-1)(10 mL/110 mL)

[F-h - [F-]o = 0.181 x 10-3 mol L-1
Substituting [F-h = (2.86 ± 0.08)[F-]o, we obtain
[F-]o (2.86 ± 0.08 -1) = 0.181 x 10-3.
[F-]o = (9.7 ± 0.4) x 10-5 M
Notice that in the example above, the fluoride ion concentration was
determined to approximately 4% expected accuracy. Such accuracy is
characteristic of direct methods. For concentrations of chemical species
above 10- 4 M, there are usually methods (such as the titration methods
discussed below) which are capable of considerably greater accuracy.
But for very low concentrations, a direct electrochemical measurement
may well be the most accurate method and frequently is the only method
available.
Titration Methods
Potentiometric methods can be adapted to the detection of the

endpoint of a titration. If either the titrant or the substance titrated is
detected at an indicator electrode, a plot of the cell potential us. the
volume of titrant shows a sharp increase or decrease in potential at the
endpoint. pH titration curves of acids and bases are familiar examples
of this method.
Other ion-selective electrodes can be used to determine titration
curves. For example, organic thiols may be determined by titration with
silver nitrate standard solution where the silver ion concentration is
monitored with a Ag+-selective electrode. The titration reaction is
Ag+(aq) + RSH(aq) ~ H+(aq) + AgSR(s)
Typical solubility-product constants of silver thiolates are less than 10- 16 ;
thus the silver ion concentration prior to the endpoint is very low and
rises rapidly when the endpoint is passed, so that a sharp break in
potential is obtained.
In many cases, nonselective indicator electrodes can be used to
monitor a titration and to detect the endpoint.
Example 1.8 Compute the titration curve for the titration of 25

mL of 0.01 M FeS04 in 1 M H2S04 with 0.01 M Ce(S04)2 using a
1 In multiplication or division, Z =xy or Z =x/y, errors propagate as (sz/z)2 =(sx/x)2

+ Csy /y)2.
§ 1. 7 Chemical Analysis by Potentiometry 43
Pt indicator electrode and a saturated calomel reference

electrode.
Both the titrant and the sample form reversible couples at the
platinum electrode with formal potentials:
Fe3 + + e- --7 Fe2+ EO' = 0.68 V
Ce 4 + + e- --7 Ce3 + EO' = 1.44 V
The titration reaction is
Fe2+ + Ce4 + --7 Fe3 + + Ce3 +
We first compute the equilibrium constant for the reaction.
Consider the hypothetical cell (not the one being measured in
the experiment) in which the two half-cells are the Fe 3 +lFe 2 +
couple and the Ce 4 +/Ce 3 + couple; the formal cell potential is
1.44 - 0.68 = 0.76 V. Thus we have !:lGo = -FEo = 73 kJ mol- 1
which corresponds to an equilibrium constant K = exp(-
!:lGoIRT) = 7.0 x 10 12 . The reaction will go to completion if the
rate is fast enough.
Up to the endpoint, the potential of the Pt electrode will be
determined by the Fe 3 +lFe2 + couple. The potential of the cell is
E = EOF 3+/F 2+ _HI In [FeZt] -E

e e F [Fe3+-] see
[FeZtJ
E = 0.68 - 0.0592 log [--] - 0.244
Fe3+-
Consider the situation when 10 mL of titrant has been added.
Since we have added (0.01 L)(O.OI mol L-1) = 10- 4 mol ofCe 4 +,
there then must be 1 x 10- 4 mol Fe3 + and 1.5 x 10- 4 mol Fe2 +
must remain. The concentration ratio then is [Fe 2+]![Fe 3 +] =
1/1.5 and the cell potential is E = 0.45 V.
At the equivalence point, [Fe3 +] = [Ce3 +] '" 0.005 M and the
equilibrium constant expression gives [Fe 2+] = [Ce 4 +] = 1.9 x
10- 9 M. The Fe 2 +lFe 3 + concentration ratio is 3.8 x 10- 7 and the
cell potential is E = 0.82 V.
Beyond the endpoint, virtually all the Fe 2 + has been
oxidized; 2.5 x 10- 4 mol of Ce3 + has been produced by the
titration reaction and excess Ce 4 + is being added. The cell
potential is given by
E = EOee4+/c 3+ _ RT In [Ce3+-] - E
e F [Ce4+] see
1.25
EIV
1.00
0.75
0.50
Figure 1.16 Titration curve

for titration of 0.01 M FeS04 0.25
with 0.01 M Ce(S04)2 in 1 M o 5 10 15 20 25 30
H2S04. Vol. Ce 4 +/mL
[Ce3+]
E = 1.44 - 0.0592 log [--] - 0.244
Ce4+
When 30 mL of titrant has been added, there will be 0.5 x 10- 4
mol of Ce 4 + present in excess. The Ce 3 +/Ce 4 + concentration
ratio is 2.5/0.5 and the cell potential is E = 1.15 V.
Carrying out several such calculations, we can sketch the
titration curve shown in Figure 1.16.
1.8 BATIERIES AND FUEL CELLS
Throughout most of the nineteenth century, electrochemical cells

provided the only practical source of electrical power and a great deal of
effort was devoted to the development of inexpensive, efficient power cells
and storage batteries. l When steam-powered electrical generators came
into use in the 1880's and electrical power began to be widely distributed,
electrochemical power cells started into a long decline. In the recent
past, galvanic cells have been used mostly as small portable power
1 A battery is a collection of two or more cells connected in series so that the battery
potential is the sum of the individual cell potentials.
§ 1.8 Batteries and Fuel Cells 45
sources (e.g., cells used in flashlights, children's toys, etc.) or as portable
energy storage systems (e.g., automobile batteries).
Interest in electrochemical cells as power sources has revived
recently (10,11) as society has become more concerned for the
environment and as fossil fuels have become scarcer and more
expensive. Electric-powered automobiles offer the hope of substantially
reduced environmental pollution and direct conversion of fossil fuel
energy into electrical energy via an electrochemical cell could in
principle be achieved with much greater efficiency than is possible with
a heat engine. In this section, we will review briefly the operation of
some electrochemical cells which convert chemical energy into
electrical energy. For further information, see books by Mantell (G2),
Angrist (G4), Bagotzky and Skundin (G5), Pletcher (G6), or Ventatasetty
(Gg).
Electrochemical cells can be divided into two classes: primary cells,
which are intended to convert chemical energy into electrical energy,
and secondary cells, which are intended to store electrical energy as
chemical energy and then resupply electricity on demand. A flashlight
cell is used until all the chemical energy has been converted to electricity
and is then discarded. A fuel cell in a spacecraft converts hydrogen and
oxygen to water, extracting the energy as electricity. We usually do not
attempt to recharge a flashlight cell and the fuel cell would not normally
be run backwards to regenerate H2 and 02 gases. A storage battery, on
the other hand, is intended to store electrical energy. Discharging and
recharging are equally important parts of the operational cycle. The
difference in function leads to differences in design.
We first consider three common primary cells which have a family
resemblance, all having a zinc anode (12).
The "Dry Cell"
The so-called "dry cell" used to power flashlights is descended from

a cell invented by Georges Leclanche in 1868 and is sometimes called a
Leclanche cell. The cell, which can be represented schematically by
Zn(s)IZn2+(aq),NH4CI(aq)IMn02(s),Mn203(s)IC(s)
is shown in Figure 1.17. The carbon rod cathode is surrounded by a
thick layer of Mn02 (mixed with a little graphite to improve the
conductivity). A paper barrier separates the Mn02 from the aqueous
electrolyte (which is gelled with starch or agar so that the cell is "dry")
and the zinc anode which forms the cell container. A newly prepared
cell has an open circuit potential of about 1.55 V, but under load the cell
potential decreases to 1.3-1.5 V. The cell reaction
Zn(s) + 2 Mn02(S) + H20 ~ Zn(II) + Mn203 + 2 OH-(aq)
contact cap
insulation
t;=:l5~r-_air space
carbon rod cathode
manganese dioxide
1+--- paper spacer
gelled electrolyte
zinc anode
Figure 1.17 Construction of the
Zn/Mn02 dry cell. insulating disk
leads to an increase in pH which is buffered by the NH4Cl. The cell

reaction is only partially reversible, so that dry cells are not readily
recharged. The exact nature of Zn(II) depends on the pH and NH4CI
concentration. It may be precipitated as a hydroxide or oxychloride or
remain in solution as an ammine complex. On diffusion into the Mn02
phase, however, precipitation as ZnO-Mn20a occurs. There is some
evidence that formation of this phase is responsible for the irreversibility
of the cell. ZnJMn02 dry cells have a relatively short shelf life because of
diffusion processes which amount to an internal short circuit. Because
the cells are inexpensive, however, they have been widely used.
Because of the large market for this kind of cell, a considerable
effort has been expended on improvements of the Leclanche cell. When
NH4CI is replaced by KOH as the electrolyte, the cell reaction is
transformed to
Zn(s) + 2 Mn02(s) + H20 -+ Mn20a(S) + Zn(OH)2
Because the electrolyte is not gelled and is more corrosive than NH4CI,
there are some packaging problems, the solutions to which increase the
cost of the cell. However, the shelf life of an alkaline cell is much longer
than that of the Leclanche cell, and the alkaline ZnJMn02 cell turns out
to be well adapted to applications where a steady low-level drain of power
is required, a situation where polarization of the traditional dry cell
would lead to degradation of performance.
The Mercury Cell
One of the few commercially important cells which does not have
nineteenth century origins is the alkaline zinc/mercury cell, developed
for the U. S. Army during World War II. The cell,
Zn(Hg)IZn(OH)2(S)IKOH(aq),Zn(OH)42-(aq)IHgO(s)IHg(l)
has an open-circuit potential of 1.35 V and is very widely used to power
transistor radios, watches, hearing aids, etc. Unlike the ZnlMn02 dry
cell, the electrolyte is not consumed by the Zn/HgO cell reaction
Zn + HgO + H20 ~ Hg + Zn(OH)2
and as a result, the cell potential is much more constant during
discharge. Accordingly, mercury cells have often been used to provide a
voltage reference in electronic instrumentation.
The cell is constructed with a mercuric oxide cathode (mixed with a
little graphite to improve conductivity) pressed into the bottom of a
nickel-plated steel case. Upon discharge, this layer becomes largely
liquid mercury. The alkaline electrolyte is contained in a layer of
adsorbent material and the zinc amalgam anode is pressed into the cell
top. In addition to the relatively constant potential, mercury cells have a
long shelf life and a high energy-to-volume ratio. The major
disadvantages are that power output drops precipitously at
temperatures below about 10°C and that disposal presents
environmental problems.
The Silver Cell
A close relative of the mercury cell is the ZnlAg202 cell:

Zn(s)IZn(OHMs)IKOH(aq),Zn(OH)4 2-(aq)IAg20 2(S)IAg(s)
This cell has many of the advantages of the mercury cell although the
potential is less constant during discharge. The silver cell has a higher
voltage than the mercury cell, about 1.5 V, and thus has a somewhat
higher energy-to-weight ratio. It also operates successfully at
significantly lower temperatures. Applications have been largely in
situations where a very small, highly reliable cell is required, for
example, in hearing aids and watches and in the guidance systems of
rockets.
Fuel Cells
When hydrogen or a hydrocarbon fuel is burned in a heat engine,

the heat of the combustion reaction is partially converted to work. The
efficiency of the engine can be defined as the ratio of the work done on
the surroundings (-w) to the heat evolved in the chemical reaction (--q):
Efficiency = wlq
The maximum theoretical efficiency of the process is limited. For an
engine operating with a Carnot cycle (isothermal compression at TI,
adiabatic compression to T2, isothermal expansion at T2, and adiabatic
expansion back to T1), it can be shown that the maximum efficiency is

determined by the temperature limits between which the engine
operates:
(1.29)
In practice, efficiencies are considerably less than theoretical because of
friction and heat loss.
In contrast to a heat engine, the work obtained from an
electrochemical cell is equal to -/)"G for the cell reaction. Using the
definition above, the efficiency is
Efficiency = /)"G = 1- T/),,$ (1.30)
MI MI
When /)"H and /),,$ are of opposite sign, the "efficiency" of an
electrochemical power cell, based on the wlq criterion, is greater than 1.
When MI and /),,$ are both negative (the more common situation), the
efficiency decreases with increasing temperature but is usually much
greater than that of a heat engine. This is hardly a new idea. Ostwald
pointed out this difference in maximum efficiencies in 1894 (13) and
suggested that electrochemical fuel cells should be developed to replace
heat engines.
Example 1.9 Compute the maximum efficiency and the

maximum work when one mole of H2(g) reacts with 02(g) to
produce H20(g) (a) in a Carnot engine operating between 600 K
and 300 K, and (b) in an electrochemical cell operating under
standard conditions at 298 K.
The standard enthalpy of formation of H20(g) is -242 kJ moP.

According to eq (1.26), the maximum efficiency is 50%, so that
the maximum useful work, assuming that MI is temperature
independent, is 121 kJ mol-I.
The standard free energy of formation of H20(g) is -229 kJ
mol-I, and all this energy is theoretically convertible to
electrical work. The efficiency then is (229/242) x 100% = 94.6%.
Friederich Wilhelm Ostwald (1853-1932) was Professor of Physical

Chemistry at the University of Leipzig. Together with Arrhenius and
van't Hoff, Ostwald is regarded as one of the founders of physical
chemistry. His laboratory at Leipzig spawned a generation of physical
chemists. Sir William R. Grove (1811-1896) was a barrister by
profession, but he maintained an active scientific career on the side. He
is remembered for his work on galvanic cells and as a founder of The
Chemical Society of London.
§1.8 Batteries and Fuel Cells 49
The Hydrogen-Oxygen Fuel Cell

The hydrogen-oxygen fuel cell could hardly be simpler in concept.
The half-cell reactions
H2(g) ~ 2 H + + 2 e'
~ 02(g) + 2 H + + 2 e- ~ H20
or
H2(g) + 2 OH- ~ 2 H20 + 2 e-
~ 02(g) + H20 + 2 e- ~ 20H-
sum to the formation of water
H2(g) + ~ 02(g) ~ H20
and give a standard cell potential of 1.229 V. The H2/02 system was
studied very early in the history of electrochemistry (see Section 1.1) and
the hydrogen-oxygen fuel cell was first described by Grove in 1839 (14).
While it is relatively easy to construct a hydrogen-oxygen fuel cell in
the laboratory, it is very difficult to extract much power. The major
problem is that both half-cell reactions are very slow on most electrode
surfaces. The H2/H+ reaction is relatively rapid on finely divided
platinum or palladium surfaces (and fairly fast on other precious
metals), but platinum is not a practical option for large-scale
commercial use. The reduction of oxygen is an even more serious
problem, as this rate is quite slow even on precious metal electrodes.
The H2-02 fuel cell remained a gleam in the eye of electrochemists
(and an occasional laboratory curiosity) until the U.S. space program
developed a need for a lightweight, efficient, reliable, nonpolluting
energy source for use on spacecraft. With a need defined and a
customer willing to pay development costs, several companies began
work on the problem.
The operation of a H2-02 fuel cell is limited by the slow rate of the
electrode reactions and by ohmic heating of the electrolyte solution. The
rate problem can be attacked by increasing the electrode surface area
and by increasing the temperature. While the reactions do go faster at
higher temperatures, eq (1.30) predicts a decrease in theoretical
efficiency with increasing temperature (Ml and IlS are both negative),
so that a compromise is involved. Increasing the surface area is an
effective strategy provided that the increased surface can be contacted by
both the gas phase and the electrolyte solution.
The solution to the problem of the H2-02 fuel cell in general has
involved the use of electrode materials such as porous graphite with a
small amount of precious metal catalyst imbedded in the pores. The
interface between the H2 and 02 gas phases and the aqueous electrolyte
phase occurs within the body of the porous electrode. In order to reduce
ohmic heating of the electrolyte solution, most successful fuel cells use a
thin film of solution or in some cases a thin film of an ion exchanger
between the porous anode and cathode.
Cells Using Other Fuels

While the greatest development efforts thus far have been on the H2-
02 fuel cell, large-scale technological applications of fuel cells depends
on the use of other, more readily available fuels, such as carbon (coal),
hydrocarbons (petroleum or natural gas), or carbohydrates (plant
material). Research directed to the development of such fuel cells began
in Germany during World War II and has continued, mostly in Europe,
since then. While there has been significant progress and pilot-plant-
scale fuel cells have been built, the processes are not yet competitive with
traditional steam-powered turbines. For more details on fuel cells, see
Bockris and Srinivasan (G 1), Pletcher (G6), or reviews by Eisenberg (15)
or Cairns (16).
Storage Batteries
Energy storage cells (secondary cells) have somewhat different

requirements than primary energy conversion cells. A storage cell
must be capable of many charge/discharge cycles with high energy
efficiency; thus there must be no irreversible side reactions or processes
which convert components into unusable forms. Because storage cells
are mostly used in situations requiring a portable power source, the
energy available per unit weight should be as large as possible. A
common figure of merit applied to storage cells is the energy density,
expressible in units of J g-l or kWh kg-l (1 kWh = 3600 kJ).
By far the largest use of storage cells at present is in automobile
batteries. Here the major requirement is that the battery be capable of
delivering high power for a relatively short time (to start the engine).
Storage cells are also used to power many small appliances. A
potentially important application of storage cells is in powering electric
cars and trucks. Electrically powered vehicles were popular in the
period around World War I but did not successfully meet the competition
of the internal combustion engine. With society's recently acquired
awareness of the environmental damage of automobiles, there has been
a revival of interest in electric cars and in efficient light-weight batteries
to run them. Another potentially important application of storage cells
is as load-leveling devices at power generating stations.
The Lead-Add Storage Battery

The lead-acid cell (17), familiar because of its use in automobile
storage batteries, was invented by Plante in 1859. The cell can be
represented by
Pb(s)IPbS04(S)IH2S04(aq)IPbS04(S),Pb02(S)IPb(s)
The half-cell reactions are
Pb(s) + S042-(aq) -+ PbS04(S) + 2 e-
Pb02(S) + 4 H+(aq) + S042-(aq) + 2 e· -+ PbS04(s) + 2 H20
so that the overall cell reaction is
Pb(s) + Pb02(s) + 2 H2S04(aq) -+ 2 PbS04(s) + 2 H20
When the acid concentration is greater than about 2 M, the formal cell
potential is about 2 V; thus a 12-V battery requires six cells. Both the
lead sponge used as the anode material and the lead dioxide cathode
(mixed with an "expander" such as BaS04 to increase the surface area)
are packed into lead grids which provide the electrical contacts. When
the battery is discharged, nonconducting PbS04 is formed at both the
anode and cathode and the internal resistance of the cell increases.
Some recovery of cell efficiency occurs as the H2S04 electrolyte diffuses
into the PbS04 cake and reaches unreacted Pb or Pb02. The rate of
recovery is thus tied to the diffusion rate and is much slower at low
temperatures. Since the density of PbS04 is less than that of either lead
or Pb02, excessive discharge would rupture the lead grids. Thus a lead-
acid battery is normally operated so that only 20-30% of the theoretically
available energy is withdrawn before recharging. Since discharge of the
cell reduces the concentration of sulfuric acid, the density of the H2S04
electrolyte solution provides a convenient measure of the state of charge
of the battery.
Given the half-cell potentials, recharging a lead-acid battery should
result in generation of H2 (instead of reduction of PbS04 to Pb) and 02
(instead of oxidation of PbS04 to Pb02). Fortunately, these processes are
very slow at lead electrodes and thus are operationally irreversible.
However, when the battery is fully charged, H2(g) and 02(g) production
may occur if the charging potential is high enough. Since gas evolution
tends to dislodge PbS04 from the grids, this is undesirable and thus
regulation of the voltage of a battery charger is required.
Lead-acid storage batteries have undergone many generations of
engineering improvement and are now reasonably efficient and reliable.
Because of the low internal resistance, the lead-acid cell is capable of
impressive bursts of power (up to about 10 kW for a few seconds), and
can be charged and discharged for years. The principal disadvantage is
weight. Electric vehicles powered with lead-acid batteries thus must
devote a large fraction of their load-carrying capacity to batteries, a

restriction which has severely limited development.
Example 1.10 Compute the theoretical maximum energy

density for a lead-acid cell, taking into account the weight of
reactants in the cell reaction and neglecting the cell housing
and electrolyte solution.
The electrical energy delivered at 2 V is

Welec = nFE = 386 kJ mol- 1
The weight of one mole of the reactants (Pb, Pb02, and 2 H2S04)
is 642.5 g, so that
Energy density = (386 kJ mol- 1)/(642.5 g mol-I) = 601 J g-1
or 0.17 kWh kg-I. Because the cell is usually not more than 30%
discharged and the neglected weight is substantial, the
practical energy density is in the range 80-200 J g-l.
Gaston Plante (1834-1889) was Professor of Physics in Paris; he is best

known for his work on storage batteries. Thomas Alva Edison (1847-
1931) was a prolific inventor whose persistence made up for his lack of
formal scientific training. Although Edison acknowledged a debt to
Michael Faraday for his electrical and electrochemical inventions, his
approach was less one of pure reason than an exhaustive trial of every
possible solution to the problem at hand.
The Edison Cell

The alkaline iron-nickel cell can be represented as
Fe(s)INi02(S),Ni304(S)IKOH(aq)IFe304(s)IFe(s)
The half-cell reactions
3 Ni02(S) + 2 H20 + 4 e- ---t Ni304(S) + 4 OH-(aq)
3 Fe(s) + 8 OH-(aq) ---t Fe304(s) + 4 H20 + 8 e-
combine to give the overall cell reaction
6 Ni02(S) + Fe304(S) ---t 3 Fe(s) + 2 Ni304(S)
This cell was perfected by Edison in 1910 as a power source for electric
vehicles. The nominal cell potential is about 1.37 V but varies
considerably with the state of charge of the cell. Since the electrolyte is
not consumed in the cell reaction, a relatively small volume is required
and the cell can be more compact and lighter in weight than a lead-acid
cell. The theoretical maximum energy density is about 1300 J g-l (0.36
kWh kg- 1 ), which compares favorably with most other storage cells. The
Edison cell has several other advantages: the basic solution is less
corrosive than the concentrated acid used in a lead battery; the electrode
assemblies are more rugged and less susceptible to damage by complete
discharge or overcharging. However, the smaller cell potential requires
more cells per battery than a lead-acid cell and, because nickel is more
expensive than lead, the initial cost is greater. On the other hand, an
Edison cell has a much longer service life than a lead-acid cell and so
may bp. cheaper in the long run. Edison cells were used extensively in
the heyday of the electric car and are quite well suited to that purpose.
Because of its greater internal resistance, however, the Edison cell
cannot match the lead-acid cell for peak power production and thus is
not as satisfactory for turning starting motors on internal combustion
engines.
The Nickel-Cadmium Cell

This cell (18) is a first cousin to the Edison cell, with iron and FeS04
replaced by cadmium and Cd(OH)2. The cell, which has a nominal
potential of 1.25 V, has many of the advantages and disadvantages of the
Edison cell. It is intrinsically more expensive and has a lower energy
density, but it has found an important niche in small rechargeable
batteries used in transistor radios, tape recorders, pocket calculators,
etc.
The Sodium-Sulfur Cell
The traditional energy storage cells discussed above employ

moderately active metals and aqueous solutions. There have been many
attempts to develop cells based on the much more exoergic reactions of
the alkali metals. A successful cell based on the reaction of lithium or
sodium with fluorine or oxygen would combine high cell potential with
light weight, but there are some obvious technical problems in building
such a cell. A promising system with somewhat less severe problems is
based on the reaction of sodium and sulfur
2 Na + 8 -7 2 Na+ + 8 2-
The cell is operated at 300 a C with both sodium and sulfur in the liquid
state. These two phases are separated by a ~-alumina (Na20·11AI20S)
membrane which is very permeable to Na+ at high temperature. Liquid
sulfur supports the ionization of Na28. An inert metal such as
molybdenum is used to contact the conducting phases and the cell can be
represented as
Mo(s)INa(l)INa+(~-alumina)INa28(s),8(l)IMo(s)
The cell voltage is about 2 V, depending on the state of charge, and the
theoretical maximum energy density is about 5000 J g-I (1.4 kWh kg-I).
The major problem with the cell is the membrane. Na atom diffusion
into grain boundaries in the p-alumina membrane causes short circuits
and structural failure. The sodium-sulfur cell is probably not practical
for electric vehicles (heating the cell to 300°C before operation is one
problem), but as a power plant load-leveling device it has some promise.
REFERENCES
(Reference numbers preceded by a letter, e.g. (AI), refer to a book listed in

the Bibliography.)
1. A. A. Frost, J. Am. Chem. Soc. 1951, 73, 2680.

2. E. A. V. Ebsworth, Educ. Chem. 1964, I, 123.
3. P. Delahay, M. Pourbaix, and P. van Rysselberghe, J. Chem. Educ.
1950, 27, 683.
4. F. Haber and Z. Klemenciewicz, Z. phys. Chem. 1919, 67, 385.
5. R. W. Murray, ElectroanaZyticaZ Chemistry 1984,13, 191.
6. J. Janata,AnaZ. Chem. 1992, 64, 196R.
7. N. H. Furman in Treatise on Analytical Chemistry, I. M. Kolthoff
and P. J. Elving, eds, New York: Wiley, 1963. Part I, Vol. 4, p 2269.
8. S. Wawzonek in Physical Methods of Chemistry, Part ITA
(Techniques of Chemistry, Vol. I), A. Weissberger and B. W.
Rossiter, eds, New York: Wiley, 1971, p 1.
9. R. P. Buck in Physical Methods of Chemistry, Part ITA
(Techniques of Chemistry, Vol. I), A. Weissberger and B. W.
Rossiter, eds, New York: Wiley, 1971, p 61.
10. D. P. Gregory, Modern Aspects of Electrochemistry 1975, 10, 239.
11. K. V. Dordesch, Modern Aspects of Electrochemistry 1975, 10,339.
12. J. McBreen and E. J. Cairns, Adu. Electrochem. Electrochem.
Engin. 1978, 11, 273.
13. W. Ostwald, Z. Elektrochem. 1894, 1, 129.
14. W. R. Grove, Phil. Mag. 1839, 14, 127.
15. M. Eisenberg, Adu. Electrochem. Electrochem. Engin. 1962,2, 235.
16. E. J. Cairns, Adu. Electrochem. Electrochem. Engin. 1971,8, 337.
17. J. Burbank, A. C. Simon, and E. Willihnganz, Adu. Electrochem.
Electrochem. Engin. 1970,8, 157.
18. P. C. Milner and U. B. Thomas, Adu. Electrochem. Electrochem.
Engin. 1967,5, 1.
Problems 55
PROBLEMS
L1 An electrochemical cell is formed by placing platinum electrodes

in separate beakers, connected through a salt bridge. One beaker
contains a solution of FeS04, Fe2(S04)3, and H2S04 (all 0.01 M); the
other beaker contains KMn04, MnS04, and H2S04 (all 0.01 M).
(a) Using standard notation, write a representation of the cell.
(b) Write the half-cell reactions.
(c) Write the cell reaction.
(d) Compute the standard potential of the cell.
(e) Compute the actual potential of the cell (neglect activity
coefficients, but take account of the second ionization of H2S04,
}ll(a2 = 1.99).
(f) Compute the standard free energy change for the cell reaction
and the corresponding equilibrium constant.
1.2 Calculate the potential of each of the cells (ignore activity

coefficient corrections, but note that one of the half-cells is a
standard reference electrode):
(a) Zn(s)IZnS04(O.01 M)IIKCI(satd)IHg2CI2(S)IHg
(b) PtIFeCI3(0.01 M), FeCI2(0.002 M)IiKCI(3.5 M)IAgCl(s)IAg
(c) Pt1CrC13(O.1 M), K2Cr207(0.001 M), HCI(O.OOl M)II
KCI(3.5 M)IAgCl(s)IAg
1.3 Use the data of Table AA to construct a free energy-oxidation state

diagram for manganese in acidic solution. Given that the
solubility product constants for Mn(OHh and Mn(OH)a are,
respectively, 2 x 10- 13 and 4 x 10-43 , construct a free energy-oxidation
state diagram for manganese in basic solution.
1.4 Given that the standard potential of the half-cell

S042-(aq)IPbS04(S)IPb(s)
is -0.356 V, estimate the solubility-product constant of PbS04. You
will need more data; see Table AA.
1.5 If the standard potential for the reduction of Sb205(S) to SbO+ in

acid solution is +0.60 V and the standard potential for the reduction
of Sb205 to Sb203(S) in basic solution is -0.13 V, what is the solubility
product constant, Ksp = [SbO+][OH-]? What is the pH ofa saturated
solution ofSb203 at 298 K?
1.6 Devise general derivations of eqs (1.24), (1.25), and (1.26)
1.7 A silver electrode is immersed in 100 mL of 0.1 M KCI, and the

solution titrated with 0.2 M AgNOg solution. The potential is
determined us. a standard Ag/AgCI electrode (with 3.5 M KCI
electrolyte). Calculate the cell potential when 1, 10, 30, 45, 50, 55,
and 70 mL of the silver solution has been added. The solubility
product constant of AgCI is 2.3 x 10-10. Neglect activity coefficients.
Plot the calculated potential us. volume of titrant.
1.8 Overman (Anal. Chern. 1971,43,616) found that an iodide-selective

electrode can be used in the determination of Hg2+ by titration with
a standard NaI solution. When he titrated 50 mL of unknown Hg2+
solution with 0.0100 M NaI, he obtained the titration curve shown
in Figure 1.18. What was the concentration of mercuric ions in the
original solution assuming that the titration reaction produced
insoluble HgI2? This titration procedure worked for Hg2 +
concentrations down to about 5 x 10- 6 M, but at lower
concentrations, endpoints were found which corresponded to ratios
of iodide to mercury of more than 2:1. Why?
1.9 The cell

CdICd2+(o.00972 M)IICd 2+(0.00972 M),CN-(O.094 M)ICd
has a potential of -0.4127 V at 25°C. If the only significant reaction
between Cd 2+ and CN- is the formation of Cd(CN)4 2-, what is the
equilibrium constant for the formation of the complex ion?
1.10 The quinhydrone electrode is sometimes used to measure pH.

Quinhydrone is an easily prepared, slightly soluble, equimolar
mixture of benzoquinone and hydroquinone. The reduction of the
quinone is reversible on platinum:
EO = +0.700 V
In practice, an unknown solution is saturated in quinhydrone and
the potential measured with respect to a saturated calomel
electrode.
(a) Show that the potential of the quinhydrone electrode gives the
pH directly.
(b) Suppose that a cell potential of +0.160 V is measured us. s.c.e.
What is the pH of the solution?
(c) The quinhydrone electrode is not usable above pH 9. Why?
U1 The adult human brain operates at a power of about 25 W. Most of

the power is used to operate "sodium pumps" in the nerve cell
membranes, which maintain the internal ionic concentration at
Problems 57
0.1
End Point
1.272 mL
Fi~ure 1.18 Titration of 0.0
Hg + with NaI with potential
measured by an iodide-
sensitive electrode. Reprint-
ed with permission from R. -0.1 =--L-.---'-_'---L-.---L.---"'---'---L.____'---'
F. Overman, Anal. Chern. 1.0 1.1 1.2 1.3 1.4 1.5
1971,43, 616, copyright 1971
American Chemical Society. Vol. 0.0100 M NaIlmL
about 0.015 M Na+, while the external concentration is about 0.15 M

N a +. Assuming that all the power is used for such pumps and
that they have an overall efficiency of 50%, calculate the total flux of
Na+ out of the brain cells per second. Assuming that the brain has
10 8 cells and that each nerve impulse results in an uptake of 10- 11
mole ofNa+, estimate the firing rate of the brain cells.
1.12 In the Krebs citric acid cycle, nicotinamide adenine dinucleotide

(NAD+) acts as an electron acceptor in three separate steps: (1) in
the conversion of isocitric acid to a-ketoglutaric acid; (2) in the
conversion of a-ketoglutaric acid to succinic acid; and (3) in the
conversion of malic acid to oxaloacetic acid. From the data given in
Table A.6, compute (a) the cell potential of each of these processes;
(b) the standard free energy changes; and (c) the equilibrium
constants (all at 25°C, pH 7).
1.13 From data given in Table A.4, compute

(a) I1Go for the oxidation ofBr by H202.
(b) I1Go for the oxidation ofH202 by Br2.
(c) I1Go for the disproportionation of H202 to H20 and 02.
(d) Knowing that solutions of hydrogen peroxide are stable for long
periods of time, what do you conclude?
(e) Assuming that the reactions described in parts (a) and (b) above
are relatively fast, predict the effect of a trace of KBr on a hydrogen
peroxide solution.
1.14 From data given in Table AA, compute the standard free energy of
disproportionation for the following reactions:
2 Cu+ ~ Cu2+ + Cu
3 Fe2+ ~ 2 Fe3 + + Fe
5 Mn042- + 8H+ ~ 4 Mn04- + Mn2+ + 4 H20
Can you generalize from these results to obtain a criterion for the
stability of a species toward disproportionation based on half-cell
potentials?
1.15 The potential of a glass electrode was 0.0595 V us. s.c.e. when the
electrodes were immersed in a pH 7.00 buffer solution and 0.2598 V
when immersed in a solution of unknown pH.
(a) Calculate the pH of the unknown solution.
(b) If a possible junction potential introduces an uncertainty of
±0.0005 V in comparing the potentials of the known and unknown
solutions, what is the uncertainty in the pH of the unknown?
1.16 A sodium ion-selective electrode developed a potential of 0.2631 V

us. s.c.e. when immersed in 25.00 mL of an unknown solution.
Mter addition of 5.00 mL of a 0.0100 M Na+ standard solution, the
potential decreased to 0.1921 V.
(a) Calculate the sodium ion concentration in the unknown
solution.
(b) If the uncertainty in the potential measurements was ±0.0005
V, what was the uncertainty in the sodium concentration of the
unknown?
1.17 The potential of a glass electrode in contact with a solution with

[Na+] = 0.10 M is -0.450 V at pH 8.00, -0.568 V at pH 10.00, and -0.690
Vat pH 13.00.
(a) Compute the potentiometric selectivity coefficient, kH,Na.
(b) At what solution pH would the apparent pH (computed
assuming pH linear in potential without correction) be too low by
0.05 pH unit?
1.18 The silver-zinc cell,

Zn(s)IZnO(s)IKOH(aq, 40%),K2Zn02(aq)IAg20(s)IAg(s)
has had limited application in storage batteries.
(a) Write the half-cell reactions and the overall cell reaction.
(b) If the cell potential is 1.70 V, what is the maximum possible
energy density?
2
THE ELECTRIFIED
INTERFACE
In the last chapter, we focused on the equilibrium potential

developed by an electrochemical cell. We relied on thermodynamics and
paid no attention to the molecular details. We shall now look more
closely at the interface between an electrode and an electrolyte solution.
Virtually any surface in contact with an electrolyte solution acquires a
charge and therefore an electric potential different from that of the bulk
solution. There are four ways in which a surface may acquire a charge:
(1) imposition of a potential difference from an external potential source;
(2) adsorption of ions on a solid surface or on the surface of a colloidal
particle; (3) electron transfer between a metallic conductor and the
solution; and (4) for micelles, biological macromolecules and
membranes, ionization of functional groups such as carboxylate,
phosphate, or amino groups. Surface charge effects are particularly
important in biological systems. The surface-to-volume ratio of a
biological cell is large and most biochemical reactions occur at or near
the surface of an immobilized enzyme.
We begin in §2.1 with the development of a mathematical model for
a charged interface. We digress slightly in §2.2 - §2.4 to a discussion of
the effects of charged surfaces on the equilibrium and dynamic behavior
of interfacial systems. Returning to electrodes in §2.5, we consider the
phenomena of electrocapillarity and double-layer capacitance. The ideas
developed in our discussion of charged surfaces can be extended to the
interaction of small ions in solution, and we conclude this chapter with
a discussion ofthe Debye-Hiickel theory of ionic activity coefficients.
2.1 THE ELECTRIC DOUBLE LAYER
A charged surface in contact with an electrolyte solution is expected

to attract ions of opposite charge and to repel ions of like charge, thus
establishing an ion atmosphere in the immediate vicinity of the surface.
Two parallel layers of charge are formed-the charge on the surface
itself and the layer of oppositely charged ions near the surface. This
59
60 The Electrified Interface
Helmholtz Gouy Stern

model model model
+ + -- + + - -=-+
+ + -£+ + +-
+ + + --~+
+ + +
+ + .=-+ + - +
+ + +-+ + ~+
+ + + ---+
+ + -+ + +
+ + + + +
+ + +-+ - + --+
+ + - + +
Figure 2.1 Models

of the electric double
layer. r r r
structure is called the electric double layer. For reviews of double layer
theory, see Grahame (1), Parsons (2), Mohilner (3), or Reeves (4).
Helmholtz considered the problem of a charged surface in contact
with an electrolyte solution in 1871 (5). He assumed that a layer of
counter ions would be immobilized on the surface by electrostatic
attraction such that the surface charge is exactly neutralized. In the
Helmholtz model, the electric potential falls from its surface value, <Po, to
zero in the bulk solution over the thickness of the layer of counter ions.
This behavior is shown schematically in Figure 2.1.
Later, Gouy (6) and Chapman (7) pointed out that ions are subject to
random thermal motion and thus would not be immobilized on the
surface. They suggested that the ions which neutralize the surface
charge are spread out into solution, forming what is called a diffuse
double layer. According to the Gouy-Chapman model, the potential falls
more slowly to the bulk solution value, as shown in Figure 2.1.
In 1924, Stern (8) observed that neither the Helmholtz model nor the
Gouy-Chapman model adequately accounts for the properties of the
double layer and suggested that the truth lies in a combination of the two
models. Thus some ions are indeed immobilized on the surface (the
Helmholtz layer), but usually not enough to exactly neutralize the
charge; the remainder of the charge is neutralized by a diffuse layer (the
Gouy layer) extending out into the solution. This model is also pictured
schematically in Figure 2.1.
In the following sections, we shall work through some of the details
of the Gouy-Chapman theory, as modified by Stern, and discuss some of
the implications.
§2.1 The Electric Double Layer 61
Hermann Ludwig Ferdinand von Helmholtz 0821-1894) was trained as a

physician and held chairs in physiology and anatomy at the Universities
of Konigsberg and Heidelberg. His interest in physics led to an
appointment as Professor of Physics at the University of Berlin.
Helmholtz made pioneering contributions in thermodynamics,
physiological acoustics and optics, hydrodynamics, electrodynamics, and
epistemology. It should be recalled that Arrhenius' theory of electrolyte
solutions was put forward in 1887; in 1879 ions were thought to occur only
in rather special circumstances; the Helmholtz layer was such a special
case. Louis-Georges Gouy 0854-1926) was Professor of Physics at the
University of Lyons. His contribution to the theory of the electric double
layer was an extension of his interest in Brownian motion. David L.
Chapman 0869-1958) was a Fellow of Jesus College, Oxford. Most of his
scientific work was on gas-phase reaction kinetics. Otto Stern 0888-1969)
was Professor of Physics at the Universities of Frankfurt, Rostock, and
Hamburg and moved to the Carnegie Institute of Technology in 1933.
Stern is remembered primarily as the father of molecular beams
research.
Gouy·Chapman Theory
The immediate goal of a theoretical description of the electrified
interface is the derivation of an equation giving the electric potential <l>
as a function of the distance from the surface.
The electric potential at any point in the solution is related to the net
charge density at that point by a fundamental relationship derived from
electrostatics and known as Poisson's equation:
V2<l> = _ ~ (2.1)
10100
In eq (2.1), 10 is the dielectric constant of the medium and EO is a constant

called the permittivity of free space (EO = 8.854 x 10- 12 C2J-1m -1). The
operator V2 is, in Cartesian coordinates,
2 iii
V =-+-+-
ox 2 ()y2 oz2
and in spherical polar coordinates,
10( 0)
V 2 = - - r2- + 1 d(.
sm ~ - 0) + 1 -i
r2 or or r2 sin ~ o~ o~ r2 sin2~ 0$ 2
The space charge density p is related to the concentrations and charges
of the ions in solution and can be expressed as
p= L ZiFCi (2.2)
where p has units of C m- 2 , Zi is the charge on ion i, Ci is the

concentration l of ion i in mol m- 3 , and F is the Faraday constant.
Equations (2.1) and (2.2) could be combined to obtain a relation
between ionic concentration and the electric potential. However, since
neither of these quantities is known, we need yet another relation to
achieve our initial goal. To obtain the necessary link, we recall the
Boltzmann distribution law
Nl
No = exp (- El-Eo)
kT
which gives the equilibrium ratio of the number N 1 of particles having
energy El to the number No with energy Eo. In the equation, k is
Boltzmann's constant (the gas constant divided by Avogadro's number, k
=R / N A), and T is the absolute temperature. The Boltzmann equation is
applied here to calculate the relative concentrations of ions at positions
of different electrostatic potential energy. The electrostatic potential
energy of an ion with charge zie,2 located at a point where the electric
potential is <1>, is given by
Ei =zie<l>
Since the zero of electric potential is arbitrary, we will choose the
electric potential of the bulk solution (far from the surface) as zero; thus
the electrostatic potential energy of ions in the bulk solution is also zero.
The Boltzmann equation may then be written as
Ni = exp (_ Zie<l»
Nio kT
where Ni is the number of ions of type i (per unit volume) at the point in
the solution where the potential is <1>, and Nio is the number of these ions
per unit volume in the bulk solution. Converting the N's to molar
concentration units, and noting that elk = FIR, we can rewrite the
Boltzmann equation as
..flL = exp (_ ziF ) (2.3)
Cia RT
Notice that eq (2.3) predicts that a positive potential (due to a positive
surface charge) would result in lower concentrations of positive ions (Zi
> 0) near the surface than in the bulk solution and higher
concentrations of negative ions near the surface.
Combining eqs (2.2) and (2.3), we have
1 In this and subsequent chapters we will mostly use concentration units of mol m- 3
(mM). The major exceptions to this rule are those cases where the concentration is
an approximation to an activity referred to the 1 M standard state.
2 The electronic charge e is taken as a positive constant, e = 1.602 x 10- 19 C; zi then
includes both the magnitude (in units of e) and the sign of the charge on ion i.
p= ~ ZiFCi °exp (- z:; ) (2.4)

I
and finally, substitution of eq (2.4) into eq (2.1) gives the Poisson-

Boltzmann equation:
V2<J> = -!Eo ~ ZiCi

I
o exp (- Z:;) (2.5)
The solution to this differential equation is the potential as a

function of the position in the solution. In order to proceed further, we
must specify the geometry of the problem and decide upon the method of
solution. Integration of eq (2.5) is not a trivial exercise for any choice of
geometry because of the sum of exponential terms. There are two
methods which may be used in dealing with the summation: (1) a
general, but approximate, solution obtained by expanding the
exponential terms in a power series, and (2) an exact, but limited
solution obtained by making a specific assumption about the nature of
the electrolyte so that the summation can be eliminated.
Solutions of the Poisson-Boltzmann Equation

An exponential can be approximated by the power series
exp(-u) = 1- u + 1 u 2 + ...
2
Thus, for Zi<J> «RTIF, eq (2.5) can be written
V =_ L
2<J>
££0
L Zi Ci °[1- ZiF<J> + 1 (ZiF<J»2 + ...J
i RT 2 RT
The first term in the sum vanishes since the bulk solution is electrically
neutral. Retaining only the second term, we have
V 2<J> = F 2<J>
EEoRT i
L z·2Ci o
I
(2.6)
A convenient measure of the influence of electrolytes in problems like

this is the ionic strength, defined by
(2.7)
Thus eq (2.6) can be written as

V2<J> = F2I<J>
EEoRT
or, lumping all the constants together to define a new parameter
XA = ~ eeoRT (2.8)
F2[
we have the linearized Poisson-Boltzmann equation

V 2 <t> = ~ (2.9)
xl
Let us assume that we are dealing with a surface in the y-z plane; it
is reasonable to assume that the potential <l> is a function only of the
distance from the surface x, so that eq (2.9) becomes
d 2 <t> _ <l>(x)
dx 2 - xl
Differential equations of this form have the general solution
<l>(x) = A exp(-xIXA) + B exp(xlxA)
where A and B are constants of integration to be determined from the
boundary conditions. Since in our problem, <l> ~ 0 as x ~ 00, we see that
B must be zero.
The other constant A is the potential at the surface (x = 0) provided
that the theory is adequate right up to the surface. However, the distance
of closest approach to the surface must be at least as large as the ionic
radius. If there are ions at this closest approach distance, then their
position in space is more or less fixed and the assumption of random
thermal motion will fail at very short distances. If this first layer of ions
is essentially an immobilized Helmholtz layer, then we expect that the
potential will drop more or less linearly from the surface potential <l>s to
the value <l>a at the outer surface of the Helmholtz layer. We assume
then that eq (2.9) applies to the outer or diffuse region (the Gouy layer). If
a is the thickness of the immobilized Helmholtz layer, then the potential
will be given by
<l>(x) = <l>B (1-~) + <l>a (~) x<a (2.10a)
a-x
<l>(x) = <l>a exp-- x>a (2. lOb)
XA
The validity of eq (2.1Ob) depends on the correctness of eq (2.9) and the

assumption that z<l> «RTIF. Since RTIF = 25.7 mV at 25°C, <l>a
apparently must be less than about 0.0261z volts.
If, instead of a planar surface, we consider the surface of a sphere
(e.g., a macromolecule) and assume that the potential depends on the
radial distance r from the center of the sphere but not on the angles 1'} or
<\>, the solution to eq (2.9) is
<D(r) = <Da (ct.) exp a - r (2.11)

r XA
where <Da is the potential at the outer surface of the Helmholtz layer (r =
a). Notice that this result differs from eq (2.10b) by the factor (air).
If we restrict our attention to a symmetrical electrolyte such as
NaCI or MgS04, eq (2.5) can be integrated without approximations to
obtain (see Problems)
[exp(zF<D(x)lZRT} -1][exp(zF<D a /2RT} + IJ = exP[( a - x )',XA ] (2 12)
IJ .
;---.:..--------";-';c-----=----'C----~
[exp(zF<D(x)IZRT} + l][exp{zF<DaIZRT} -
Equation (2.12) reduces to eq (2.10b) in the limit where zF<Da «RT.
Now that we have solutions to the Poisson-Boltzmann equation, let
us examine some of the consequences. First, we look at the shape of the
solutions represented by eq (2.12). In Figure 2.2 are plotted <D(x)/<Da us. (x
- a)lxA for <Da = 0.01 and 0.1 V, assuming that z = 1 and T = 298 K. The
curve for <Da = 0.01 V is indistinguishable from the simple exponential
dependence predicted by the approximate solution, eq (2.10b). Indeed,
even for <Da = 0.1 V, deviations from the approximate solution are not as
great as we might have expected.
1.00
<D/<Da
0.75
0.50
0.25
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
(x - a)lxA (x - a)lxA
Figure 2.2 Electric potential Figure 2.3 Concentrations (relative

calculated from eq (2.12) as a to bulk solution concentration) as a
function of distance from the surface function of distance from the surface
for <l>a = 0.1 and 0.01 V. The points on for a 1:1 electrolyte and <l>a = 0.003 V
the <l>a = 0.01 V curve are calculated (dots), 0.01 V (dashes), and 0.03 V
from eq (2.10b). (solid line).
The concentration ratios C+(x)/CO and C_(x)/CO are plotted as

functions of (x - a)lxA for several values of <l>a in Figure 2.3, assuming a
1:1 electrolyte (such as NaCl) and 298 K. Notice that the concentrations
are quite different from the bulk values for large values of <l>a at small
distances.
Thickness of the Ion Atmosphere

Consider now the parameter XA, defined by eq (2.8). Values ofXA are
given in Table 2.1 for several concentrations of electrolytes of different
types, calculated assuming a temperature of 25°C and the dielectric
constant of water (E = 78.54). Judging from the plots of Figures 2.2 and
2.3, XA can be regarded as a measure of the thickness of the ion
atmosphere at the charged surface. 1 As we see in Table 2.1, this
thickness depends strongly on the concentration and charge type of the
electrolyte, varying from hundreds of water molecule diameters in dilute
solutions to a distance on the order of one water molecule diameter in
concentrated solutions of highly charged ions.
Table 2.1 Ion Atmosphere Thickness
xAlnm
C/mM 1:1 1:2 1:3 2:2 2:3
0.1 30.4 24.8 21.5 15.2 7.8
1.0 9.6 7.8 6.8 4.8 2.5
10.0 3.0 2.5 2.15 1.52 0.78
100.0 0.96 0.78 0.68 0.48 0.25
Relation of Surface Potential to Surface Charge Density

We turn now to the relationship between the potential a and the
net charge on the surface. If the system as a whole is electrically
neutral, then the net surface charge (i.e., charge on the surface adjusted
for the charge of the Helmholtz layer) must be exactly balanced by the
space charge in the Gouy layer. If a is the net surface charge density (in
C m- 2 ), then, for a planar surface,
a=- L~ p dx
1 In the context of Debye-Huckel theory (Section 2.6), XA is called the Debye length.
Substituting for p from eq (2.1), we can carry out the integration:
0=E£o1°O ~dx
dx 2
=EEo~t
dx
a
Since the derivative vanishes at the upper limit, we have

o = - EEO (d<l> )
dx x=a
Differentiation of eq (2.10b) and evaluation of d<l>/dx at x = a gives
0= EEo<l>a (2.13)
XA
A more complex expression results if we use eq (2.12) to evaluate d<l>/dx

(see Problems), but it reduces to eq (2.13) in the limit thatzFa «RT.
Example 2.1 Compute the potential at a charged surface in

contact with 0.01 M (10 mol m- 3 ) KCI solution if the charges, +e,
are spaced about 5 nm apart so that the average area per
charge is 25 nm 2 .
We can rearrange eq (2.13), putting 0 in units of electronic

charges per square nanometer and XA in units of nanometers.
Taking € = 78.5, we have
<l>a = 0.230 XAO
If 0 = 0.04 e nm- 2 and XA = 3.0 nm, then
<l>a = (0.230)(0.04)(3.0) = 0.028 V
Weaknesses of the Gouy·Chapman Theory

The theory as we have presented it can lead to absurd predictions.
Consider the case of aIM NaCI solution in contact with a surface with
<l>a = 0.1 V. According to eq (2.3), the concentration of chloride ions at the
surface (i.e., at the outer surface of the Helmholtz layer) is 49 M! This
corresponds to approximately 30 chloride ions per cubic nanometer-
very nearly a closest packed layer of chloride ions. The ion-ion repulsion
in such a structure would be far greater than the attraction to the
positive surface and thus the prediction is ridiculous. This absurdity
points up the major weakness in the theory: we take account of
electrostatic interaction between ions in solution and the charged
surface, but neglect interactions between ions. This is a reasonable
approximation in dilute solutions, but leads to gross errors in more

concentrated solutions.
There are several less serious problems. We have tacitly assumed
that the charge density p is a smoothly varying continuous function of
the distance from the surface. While this is not true on an
instantaneous microscopic level, this apparent weakness is not so bad if
we regard the results as representing a time average where, through
thermal motion of the ions, the charge is blurred more or less smoothly
over the solution. A more serious defect results from the use of the bulk
dielectric constant of the solvent in calculating electrical forces over
short distances; since the water dipoles are oriented around ions or at
the charged surface, the local properties are expected to be quite different
than those of the bulk liquid.
In summary, the Gouy-Chapman treatment of the diffuse double
layer should not be relied upon to give quantitatively accurate
predictions for concentrated electrolyte solutions or for high surface
charge densities or surface potentials. The theory is very useful,
however, as a means of gaining a qualitative understanding of
phenomena which are affected by the electrified interface. While the
Gouy-Chapman approach continues to be used as a starting point for
theoretical treatments of charged surfaces, some theorists have turned
to molecular dynamics calculations where the motion of ions near a
charged surface is followed in a computer simulation; potential and
concentration distributions are then obtained by averaging over time. As
it turns out, the qualitative results of the computer modeling studies are
in good agreement with the insights obtained from Gouy-Chapman
theory.
In the next few sections we will consider some important
applications of the ideas obtained from Gouy-Chapman theory. In light
of the weaknesses just discussed, we should not hope for high accuracy,
but will be content with qualitative or, at best, semi-quantitative
explanations of the phenomena discussed.
2.2 SOME PROPERTIES OF COLLOIDS
Particles in colloidal suspensions (e.g., gold or ferric oxide sols)

generally have electrically charged surfaces due to adsorption of ions.
Macromolecular colloids such as proteins also have charged surfaces
resulting from ionized functional groups. In general, therefore,
colloidal particles have ion atmospheres. In this section we shall
examine some of the characteristics of such particles which may be
understood in terms of their electrified interfaces with the liquid
solution.
§2.2 Some Properties of Colloids 69
The Stability of Colloids
We first ask why large particles such as proteins should be stable in
solution. Atoms or small molecules are subject to so-called London
forces, which give rise to a van der Waals potential energy of interaction.
London forces are short-range; the van der Waals energy varies with
with internuclear separation as r- 6 for interactions between atoms, but
the situation is more complicated for macromolecules. One must
consider the van der Waals attraction of each atom in one molecule for
every atom in the other molecule. Some complicated integrals arise and
we shall not go into the details of the calculation. 1 The result for
spherical particles of radius a is (10)
Ua(R) =_~[8a2(R + a) + 21n R(R + 4a)] (2.14)
12 R(R + 2a)2 (R + 2a)2
where R is the distance between the particle surfaces. The parameter A

is essentially empirical and usually is on the order of 10- 19 to 10- 20 J. The
multiple attraction of the many atoms in a macromolecule results in the
van der Waals energy falling off with distance much more slowly than
might have been expected. With relatively long-range attractive forces,
uncharged particles are expected to associate and eventually precipitate
from solution.
So what keeps them apart? The free energy change, !:J.G =!:J.H - T!:J.S,
governs the equilibrium behavior. We would expect that !:J.S for
coagulation would be negative, but for a relatively small number of large
particles, the entropy term should be negligible and we must look for an
explanation in the enthalpy term. Short-range repulsion sets in when
the electron clouds overlap, but by the time this interaction is important,
van der Waals attraction has already coagulated the particles.
If the two particles are electrically charged, then electrostatic
repulsion will give a positive contribution to the potential energy of
interaction, which rises exponentially as the particles approach one
another. Notice, however, that according to eq (2.11) the distance
parameter XA determines the range of the potential and thus of the
electrostatic repulsion. If XA is very small, the diffuse layer is very
compact and the particle's charge is shielded up to very short distances.
In other words, when the ionic strength is high, the particles can
approach quite closely before the electrostatic repulsion becomes
significant, and by that time the van der Waals attraction may be quite
large.
We expect therefore that when the ionic strength is increased, the
double-layer thickness will decrease and the colloid will precipitate.
This "salting-out" effect is commonly observed experimentally.
1 See Shaw (9) for further discussion of this problem.

The calculation of the electrostatic repulsion between two particles,

each of which has an ion atmosphere, is very difficult. All the
difficulties attending Gouy-Chapman theory remain, of course, and the
question of what happens when the two ion atmospheres begin to
interpenetrate adds an important additional complication (9). Several
attempts have been made and we will quote only one result. If it is
assumed that a »XA, that the surface charge density on the particles is
constant, and that the surface potential adjusts with the ion atmosphere
as the particles approach, the repulsive contribution to the potential
energy is (11)
Ur(r) =_ 21ta 3 (}"2 In [1- exp (-R/XA)] (2.15)
(1 + a/xA)2 EEO
The total potential energy is the sum of the attractive and repulsive
contributions given by eqs (2.14) and (2.15), respectively.
To get a feel for the nature of these equations, the total potential
energy is plotted as a function of distance between the particles in Figure
2.4, where we have assumed particles with radii of 50 nm, A = 5 x 10-20 J,
and E = 78.5. The energy is given in units of kT (T = 298 K). Since kT is a
measure of the amount of thermal energy available, such a scale
permits a rough judgment on the significance of a barrier to
coagulation.
Figure 2.4a suggests that particles with a surface charge density of
0.01 e nm- 2 should coagulate quickly when the electrolyte concentration
is sufficiently high that the ion atmosphere is only 5 nm thick. When the
thickness is 10 nm, there is a small barrier to coagulation (about 3.5 kT),
and coagulation would be expected to occur fairly rapidly (e- 3 .5 = 0.03).
When the ion atmosphere is 15 nm thick, the barrier is about 36 kT and
coagulation should be much slower (e-35 = 2 x 10- 16 ) and when the ion
atmosphere is 20 nm thick, the barrier is even greater. Figure 2.4b
shows that coagulation is very sensitive to surface charge density with
no barrier at all for ()" = 0, a small barrier for 0.01 e nm- 2 , and barriers of
33 kT and 96 kT for ()" = 0.015 and 0.02 e nm- 2 , respectively.
According to eqs (2.8), (2.14), and (2.15), the potential energy also
depends on the temperature and the dielectric constant of the medium
as well as the size of the particle. We expect that salting-out should
occur at lower ionic strength when either the dielectric constant or the
temperature is reduced. Since the surface charge of a protein, for
example, depends on the number of free carboxyl or amino groups
present and thus on the pH of the medium, protein solubility is strongly
pH dependent, going through a minimum at the isoelectric pH. Since
protein solubilities are such intricate functions of ionic strength,
temperature, dielectric constant, and pH, it is not surprising that protein
mixtures can frequently be separated by selective precipitation
techniques.
100
"
,'\ (a)
, ,, 5
'..,, (b)
o
'. ..•.••.••• 10 ,, 0.01
,
", _................. 15 " 0.015
50
, ,
'"'''''''' -------!2() ". . 0.02
'........... ,..... "',
........ " ...., ' ......
........ ! ......... . .
---...............
........ .. .................. ..
o ...............::::-..... _..... _..... _.- !
~~.:.:.:;:.:;.:==.........................'1!": t'· ................
---.. ..............."':::=:~,-
......... . . "-T_
:l
...
-50
o 10 20 30 o 10 20 40
Rlnm Rlnm
Figure 2.4 Potential energy of interaction (in units of kT) as a function
of separation for two spherical particles of radius 50 nm; (a) cr = 0.01 e
nm-2~z4 = 5,10,15, and 20 nm; (b) XA = 10 nm, cr = 0, 0.01, 0.015, and 0.02
e nm .
When salting-out occurs at electrolyte concentrations of 0.1 M or

less, the effect is largely independent of the nature of the electrolyte and
depends primarily on the ionic strength. Many proteins remain in
solution under these conditions but are precipitated if much higher salt
concentrations are used. Proteins are protected from coagulation in part
by a solvation sheath of water molecules. When high concentrations of
salt are used, the activity of water is reduced significantly by the
solvation of the added salt. Thus in effect the surface of the protein is
dehydrated and coagulation is facilitated. Under these conditions, the
salting-out effect depends strongly on the nature of the added salt since
some salts tie up more water than others. Protein and polymer chemists
sometimes refer to the lyotropic series of cations and anions, which show
decreasing salting-out effects:
Mg2+ > Ca2+ > Sr2+ > Ba 2+ > Li+ > Na+ > K+ > NH4+
S042- > CI- > N03- > r· > SCN·
Salts with low solubility such as CaS04 are oflittle use in salting-out, of
course; the most commonly used precipitants are very soluble salts with
a component high on the lyotropic series, such as ammonium sulfate or
magnesium chloride.
Surface pH
We have seen that ions are either concentrated in or excluded from
the double layer, depending on the relative charges. This conclusion
applies to hydrogen and hydroxide ions as well, so that the pH at the
surface of a charged particle will be different from that of the bulk
solution. According to eq (2.3), the concentration of H + ions at the
surface is given by
[H+]surface = [H+]o exp (-~~a) (2.16)
or, converting to pH and setting T = 298 K,

pHsurface = pHo + 16.9 <l>a (2.17)
Thus a surface potential <l>a = 59 mV would result in an increase in
surface pH by one pH unit.
When the pH is less than the isoelectric point of a protein, the
surface is positively charged (due to -NH3+ groups), and the surface pH
should be higher than that of the bulk solution; above the isoelectric pH,
the surface is negatively charged (because of -C02- groups), and the
surface pH should be less than the bulk pH. Thus, to a degree, the
surface pH on a protein tends to be self-buffered toward the isoelectric
pH. While singly charged ions such as H+ or OH- can be significantly
more (or less) concentrated near a charged surface than in bulk
solution, such effects are even more dramatic for multiply charged ions.
Thus eq (2.3) suggests that a surface potential <l>a = -59 mV would result
in a concentration magnification of 100 for Mg2+ and 1000 for AI3+.
Although we must be cautious about quantitative predictions from Gouy-
Chapman theory, the qualitative effect is real and has important
consequences, for example, in the action of metal ions as co-factors in
enzymatic reactions.
Example 2.2 A dodecylsulfate micelle contains about 100

dodecylsulfate ions and has a radius of about 2 nm. Compute
the surface charge density, the surface potential, <l>a, and the
difference between the surface pH and the pH of the bulk
solution. Assume an ionic strength of 100 mM.
The surface charge density is

(J =_ (100X1.6 X 10-19 ~) =_ 0.32 C m-2
(4n)(2 x 10-9 m)
The surface potential derived using the linearized Poisson-
Boltzmann equation (see Problems) is
<D -...Q.. a
a - &0 1 + a/xA
so that with XA = 0.96 run and a = 2 nm,
a =0.65run
1 + a/xA
and
<Pa = (- 0.32 C m- 1)(0.65 x 10-9 m) = _ 0.30 V
(78.5)(8.85 x 10-12 C2J-1 m -1)
which is rather too large to justify use of the linearized
Poisson-Boltzmann equation. Ignoring this problem, eq (2.17)
predicts
pHsurface - pHo = - 5.1
While this result is doubtless an overestimate, it is in
reasonable qualitative agreement with the following
experimental result. When lauric acid is added to a solution of
dodecylsulfate micelles, the long-chain carboxylic acid
molecules are included in the micelles. A pH titration of the
carboxylic acid yields an apparent pKa (the pH of the half-
neutralization point) of about 7 under these conditions. Since
the pKa of lauric acid is about 4.8, the surface pH is apparently
about 2.2 pH units lower than that of the bulk solution.
2.3 ELECTROKINETIC PHENOMENA
Four rather peculiar effects, known as the electrokinetic

phenomena, turn out to be relatively easily understood in terms of our
theory of the diffuse double layer (12,B8). These effects arise from the
motion of an electrolyte solution past a charged surface.
The Phenomena
Consider an electrolyte solution flowing through a capillary as

shown in Figure 2.5. Since in general there is a surface charge on the
capillary walls due to adsorbed ions, there will be a diffuse space charge
in the solution adjacent to the walls. As the solution flows through the
capillary, some of the space charge is swept along with it and, if
electrodes are provided at either end of the capillary, a current, called
the streaming current, can be measured between them.
(a)
pressure
head
~
capillary
current or
voltage meter
~------~~~------~
Figure 2.5 (a) Apparatus for measuring the streaming current or
streaming potential. (b) Ions moving in the capillary; with adsorbed
anions, positive space charge is swept along by solution flow.
If, on the other hand, we replace the ammeter with a voltmeter,

essentially no current flows and ions swept through the capillary
accumulate at one end, producing an electric potential difference across
the capillary, called the streaming potential. If the experiment starts
with a homogeneous solution, the streaming potential grows with time
until it is big enough that electric migration of ions upstream exactly
cancels the flow of ions downstream.
The two effects, streaming current and streaming potential, may be
jointly expressed by one phenomenological equation:
i = CJ.l/)JJ + CJ.2~ (2.18)
Thus, when current i is allowed to flow without impedance in the
external circuit, ~ = 0, and the current is proportional to the pressure
difference ~p which generates the liquid flow. However, when the
current is zero (e.g., because of high external impedance or slow
electrode kinetics), then a potential difference ~ is observed which is
proportional but opposed to the pressure difference.
A related effect may be observed if, instead of applying a pressure
difference, we impose a potential difference between the two electrodes.
Now in the bulk solution, positive ions move toward the negative
electrode and negative ions toward positive electrode. There is little drag
on the bulk solution, which has zero net charge. Near the surface,
however, the space charge is acted on by the field, tending to drag the
solution along with it. If the cylindrical shell of solution near the
capillary walls moves, then viscous drag will pull the solution in the
§2.3 Electrokinetic Phenomena 75
center of the capillary along as well. The effect of liquid flow under the
action of an electric potential difference is called electroosmosis.
If we carry out an electroosmosis experiment and allow flow to
continue such that a pressure difference is built up, the flow will
eventually slow and come to a halt as the pressure opposes and finally
cancels the effect of the field. The relation of this steady-state
phenomenon, called the electroosmotic pressure, to electroosmosis is
analogous to the relation of streaming potential to streaming current.
Electroosmosis and electroosmotic pressure may be expressed by the
phenomenological equation
} = a3MJ + a4~ (2.19)
where} is the flux ofliquid through the capillary (in m 3 s- I ). Thus, in the
absence of a pressure difference, the flux is proportional to the electric
potential difference ~. When the flux goes to zero, the pressure
difference is proportional but opposed to the electric potential difference.
The Zeta Potential
Electrokinetic effects arise because of motion of the diffuse layer of

ions in solution relative to the solid surface. We might guess that the
Helmholtz layer is stationary and that the Gouy layer moves with the
bulk solution, but this is an oversimplification. The behavior is as if
there were a slipping surface located in the diffuse part of the ion
atmosphere. The potential at the slipping surface is given the symbol ~
(zeta) and is called the electrokinetic potential (or simply the zeta
potential).
In a quantitative treatment of the electrokinetic effects, the
appropriate form of the \7 2 operator in the Poisson equation depends on
the geometry of the problem. Thus for the symmetry of a capillary tube,
cylindrical coordinates (r, l}, and z) should be used. If we can assume
that <l> depends only on the distance from the capillary wall, then we can
neglect terms in d<l>/dl} and d<l>/dz and eq (2.1) can be written as
per) =- ITo -li
r dr
(r d<l»
dr
(2.20)
If we assume that a, the radius of the capillary, is much larger than the
diffuse layer thickness XA, then we need not rederive an equation for the
potential but can use the result obtained for a planar surface,
<l>(r) = ~ exp[(r - a)/xA] (2.21)
where the coordinate r is measured from the center of the capillary. We
have assumed that a / XA is large enough that the potential in the center
of the capillary is nearly zero.
Calculation of the Coefficients

The coefficients in eqs (2.18) and (2.19) can be computed rather
easily using the Gouy-Chapman model. The coefficient <Xl is the current
per unit pressure difference when the potential difference is zero. Since
the current results from sweeping space charge of charge density per)
along with the flowing solution, which moves with velocity v(r), it must
be given by the integral
i= La vCr) per) 21tr dr (2.22)
Substituting cl>(r) from eq (2.21) into eq (2.20) and differentiating, we

obtain the charge density
per) = _ £Eel.. (1 + L) exp

rxA XA
(rXA- a) (2.23)
When a fluid of viscosity 11 flows throu~a cylindrical tube of radius a,

the velocity is given as a function of r"by Poiseuille 's equation,
a 2 -r2
v(r) = t:J> - - (2.24)
411L
where t,.p is the pressure difference over the length L of the tube.
Substituting eqs (2.23) and (2.24) into eq (2.22) and integrating, we get
i = - 1teEot;t:J>
211L
[2a 2 - 4xl (-'L-1) + (a 2_ 4xl) exp (_-'L)~
XA XA ~
But since a »
XA, the last two terms are negligible and
. 1ta 2eEot;
<Xl = _l_ = - -----=- (2.25)
t:J> 11L
The coefficient <X2 is the current per unit potential difference when there
is no pressure difference across the capillary. Under these conditions
the current and potential difference are simply related by Ohm's law, i =
t,.cl>/R. The resistance R of a cylinder of radius a, length L, and
resistivity p is R = pL/1ta 2 , so that the current is
i = 1ta2 t,.cl>/pL
and the coefficient is
. 1ta 2
<X2=_l_=_- (2.26)
t,.cl> PL
Similar calculations (see Problems) give
1ta 4
<Xa=-- (2.27)
811 L
and
(2.28)
This last result turns out to be more general than we might have
expected from our rather simple calculation. Indeed, the electrokinetic
effects can be treated by the methods of non equilibrium thermodynamics
with the completely general result that <Xl = <X4 independent of the size or
shape of the holes through which the solution flows. 1
Let us restate the four electrokinetic effects in terms of the
phenomenological equations
i = <X1.1P + <X2.1<1> (2.18)
j = <X3.1P + <X4.1<1> (2.19)
The streaming potential is the potential difference per unit pressure
difference at zero current, and thus may be expressed quantitatively as
Streaming potential = (.1<1» = _ <Xl = EEO~ P (2.29)

.1P i=O <X2 11
where ~ is the zeta potential, p is the resistivity, and 11 is the viscosity of
the solution. The streaming current is the electric current per unit flux
at zero potential difference:
.
Streammg current = ~)
~ = -<Xl = -8EEO~
-- (2.30)
~(I>= 0 aa a2
The electroosmotic pressure is the pressure difference per unit electric
potential difference at zero flux and may be expressed as
Electroosmotic pressure = (.1P) = _ <X4 = 8EE~ (2.31)

.1<1> j =0 <Xa a2
The electroosmotic flow is the flux per unit electric current at zero
pressure difference:
Electroosmotic flow = ({) = <X4 = _ EEO~ P (2.32)

l tJ' = 0 <X2 11
Comparing these results we see that more remarkable conclusions
have been obtained: the streaming potential, eq (2.29), is just the negative
of the electroosmotic flow, eq (2.32); the streaming current, eq (2.30), is the
I This result is an example of Onsager's reciprocal relations (13). For further

discussion, see Bockris and Reddy (B8).
negative of the electroosmotic pressure, eq (2.31).1 These equalities are

also obtained from nonequilibrium thermodynamics and are thus quite
general, independent of the nature of the holes through which the liquid
flows.
Notice that the streaming potential and electroosmotic flow are
predicted to be completely independent of the dimensions of the capillary
tube. Indeed, it can be shown that the results we have obtained for a
cylindrical capillary are independent not only of the dimensions, but also
of the shape of the holes through which the liquid flows, provided only
that (1) the flow is laminar (i.e., nonturbulent), and (2) that the radii of
curvature of the pores through which the liquid flows are much larger
than the thickness of the double layer.
The electroosmotic pressure and the streaming current, on the
other hand, depend on the radius of the capillary, and in general, on the
average pore size when flow is through a porous plug or membrane.
Example 2.3 Compute the streaming potential expected

when a solution having p = 1 Q-m, £ = 78.5, and 11 = 10.3 Pa-s
flows through a capillary tube with zeta potential S = -100 mV
under the influence of a pressure difference of 1 bar (10 5 Pa).
Substitution into eq (2.29) gives

S.P. = (78.5)(8.85 x 10. 12 C2J·1m2)(-O.1 V)(l Q-m)l(1O-3 Pa-s)
S.P. =-7.0 x 1O-8 VPa-1
Thus
M> = (S.P.)LlP = -7.0 x 10-3 V (-7.0 mY)
Example 2.4 Compute the streaming current expected when

a solution with £ = 78.5 flows through a bundle of capillaries of
radius 0.1 mm and zeta potential -100 mV with a total flow of
0.1 cm3 s- 1 .
Substitution into eq (2.28) gives

S.C. = -8(78.5)(8.85 x 10-12 C2J-1m2)(-O.1 V)/(10 4 m)2
S.C. = 5.6 x 10-2 C m-3
1 These results are the more surprising since it would appear that the four quantities
should have different units. However, the apparent units of the electroosmotic flow,
(m 3s- 1)/(C s-l), reduce to m 3C-1 and the apparent units of the streaming potential, Y
Pa-\ also reduce to m 3C·I. Simila;ly, the apparent units of the streamin! current,
(C s- )!(m 3s- I ), and the electroosmobc pressure, Pa y-l, both reduce to C m- .
or, with) = 10- 7 m 3 s- 1 ,

i = (S.C.V = 5.6 x 10-9 A (5.6 nA)
Zeta Potentials at Glass-Water Interfaces
Glass-water interfaces appear to acquire a double layer through

selective adsorption, usually of OR- ions. The surface potential therefore
is usually negative and S is typically on the order of -150 mV in dilute
solutions. The zeta potential is strongly dependent on the electrolyte
concentration, however, with typical results shown in Figure 2.6.
0.0
KOH
Figure 2.6 The zeta potential at SlV
a glass-aqueous solution inter- KN03
face as a function of electrolyte
concentration for (a) KOH, (b)
KN03, and (c) HN03 deter- .'
mined by measurement of the
electroosmotic flow through a
-0.1 ........... .,.,.,
....::::::::>..//
capillary tube. The lines
represent smooth curves drawn
through data points rather than
fits of data to a theoretical
equation. Reprinted with per- ._ .... -.- ." .....
..'
mission from A. J. Rutgers and -0.2
M. de Smet, Trans. Faraday -Q -4 -2
Soc. 1945, 41, 758, copyright 1945
Royal Society of Chemistry. log C
The dependence of S on the concentration of electrolyte is a

complicated combination of at least three effects. Equation (2.13) tells us
that the surface potential is proportional to the surface charge density o.
It is also proportional to XA and thus proportional to life. In addition, if
the slipping plane is located at a distance x into the diffuse layer, then S
is less than the surface potential by the concentration-dependent factor
exp(-x/XA):
s=<l>a exp (_..x..) = OXA exp (_..x..)

XA EEO XA
Thus we would expect S to depend on the concentration of electrolyte

according to
(2.33)
where ex and ~ are constants. If cr is independent of electrolyte

concentration, then ~ should be a monotonic decreasing function of C.
The situation is somewhat more complicated if the surface charge
density is affected by changes in electrolyte concentration. In Figure 2.6,
we see that the zeta potential is larger for KOH solutions than for the
other electrolytes, presumably because cr is more negative for these
solutions (hydroxide ions are preferentially adsorbed). Indeed, there is a
minimum in the ~ us. 10g(C) curves for KOH solutions which reflects a
balance between cr increasing with increasing concentration and XA
decreasing. This interpretation is supported by the behavior ofHN03, for
which the zeta potential falls more rapidly with increasing
concentration, presumably because of surface charge neutralization.
Sometimes more positive ions are adsorbed on the negative surface
than are required to neutralize the surface charge, in effect reversing
the sign of the zeta potential. The origin of the behavior is shown in
Figure 2.7, and some typical data are shown in Figure 2.8.
1.0
~/V ........ .............................
<1>/<1>0
/.~···········:rn03
0.1
0.5
0.0 0.0 l
!.: -----.....
##
Ca(NOs)2
.l /' .................... Al(NOn)
II
...
~S
,
-0.5
..
-0.1 ":/
-1.0
0 1 2 3 -7 -6 -5 -4
log C
Figure 2.7 Reversal of zeta potential Figure 2.8 Zeta potential at a glass-
by adsorption of ions. The upper aqueous solution interface v s.
curve shows the normal behavior concentration for (a) KNOs, (b)
with <l> falling exponentially through Ca(NOS)2, and (c) Al(NOS)S,
the ion atmosphere. The lower curve determined by electroosmotic flow
shows the behavior expected when through a glass capillary. Reprinted
oppositely charged ions are adsorbed with per-mission from A. J. Rutgers
on the surface forming a "triple and M. de Smet, Trans. Faraday
layer" and reversing the sign of the Soc. 1945,41, 758, copyright 1945
zeta potential. Royal Society of Chemistry.
§2.4 Electrophoresis and Related Phenomena 81
2.4 ELECTROPHORESIS AND RELATED PHENOMENA
The electrokinetic phenomena involve stationary solid phases, the

effects arising because of motion of the electrolyte solution past the
electrified interface. Several related phenomena arise when the
interface is also free to move.
1.0 r-----r---r----,r-----r----,
.$
.~ 0.9
'-'
en
:>
<l)
Figure 2.9 Relative viscos- >
ity of starch solutions as a ~
function of salt concentra- ~ 0.8
<l)
tion for 1: 1, 1:2, and 1:3 ~
electrolytes. Reprinted with
per-mission from H. G.
Bungenberg de Jong, Rec.
Trav. Chim. 1924,43 189, 0.7 1--_-'-_ _.L.-_-..L._ _.L.-_---J
copyright 1924 Royal Neth- o 1 2 3 4 5
erlands Chemical Society. Electrolyte ConcentrationlmM
The Electroviscous Effect
The viscosity of a colloidal solution is very sensitive to the ionic

strength of the medium. For example, the relative viscosity of a solution
of soluble starch, plotted in Figure 2.9, drops dramatically with
increasing electrolyte concentration (15). The effect has two causes: (1)
XA is large at low ionic strength; the thick ion atmosphere increases the
effective hydrodynamic size of the colloid and thus increases the viscous
drag when the molecule moves in solution. (2) Electrostatic repulsion of
charged functional groups results in an extended conformation of the
starch molecule. When the ionic strength is higher, intergroup
repulsion is shielded by the ion atmosphere and a more compact
conformation is adopted. Again the effective hydrodynamic volume is
reduced and the solution viscosity is lower. The first effect can be treated
theoretically but it is usually considerably smaller than the second.
Sedimentation Potential
When colloidal particles move through an electrolyte solution under
the influence of gravity (as, for example, in an ultracentrifuge), the
particles may leave some of their ion atmosphere behind, especially if
the diffuse layer is very thick (low ionic strength). Thus charge
separation may accompany sedimentation, leading to the development of
an electric potential gradient along the length of the sedimentation tube.
The phenomenon is thus analogous to the streaming potential discussed
above. A sedimentation potential is generally an unwanted complication
in an ultracentrifugation experiment. To reduce the magnitude of the
effect, ultracentrifugation is usually carried out in high ionic strength
media so that the diffuse layer is relatively compact and charge
separation is minimized.
Electrophoresis
If an electric field (or potential difference) is applied across a

solution containing charged particles, the particles are accelerated,
rapidly attaining a terminal velocity where the electrical force is exactly
balanced by the frictional retarding force as the particles move through
solution. The phenomenon of electrophoresis is characterized by the
electrophoretic mobility, u, defined as the velocity per unit electric field
strength:
U = l.!1 (2.34)
lEI
If v is expressed in m s-l and E is in V m- 1, U apparently must have
units of m 2 V- 1s-1. Typical mobilities of ordinary small ions in aqueous
solution are on the order of 5 x 10- 8 m 2V- 1s-1; proteins generally have
mobilities in the range (0.1-1) x 10-8 m2V- 1s- 1. Although the phenomenon
of electrophoresis can be understood qualitatively as the motion of
charged particles in an electric field, the details of the motion depend
upon the nature of the ion atmosphere surrounding the particles.
Consider first the motion of a small spherical particle of radius a
through a medium of viscosity 11. According to Stokes' law, the viscous
force opposing the motion is
F vise = - 67rJ'jav (2.35)
where v is the velocity of the particle. On the other hand, the electrical
force propelling the particle through the solution is
Felee = QE
where Q is the charge on the particle and E is the electric field strength.
If the surface charge density is cr, then the total charge on the particle is
Q = 41ta 2 cr. For a spherical particle moving through solution, the

effective charge density is related to the zeta potential,
cr
= eeo~
a (1 + .iL)
XA
Combining these expressions, we have the electrical force on the

particle:
FeJec = 41taeeo~E (1 +~)
When the particle reaches terminal velocity, the acceleration is zero, and
according to Newton's second law, the resultant of forces must be zero.
Thus we have
41taeeo~E (1 +~) - 61ta1']v = 0
The electrophoretic mobility then is
Ivl 2eeo~
u =jEf =311 (2.36)
where we have assumed that a «XA. Equation (2.36) thus represents a

limiting result for small spherical particles. When a/xA approaches
unity, the mobility should increase; however, the variation is more
complicated than we might expect from this approach.
We can obtain an estimate of the e1ectrophoretic mobility for large
values of a/xA by the following analysis. Suppose that the particle is very
large compared with the size of its ion atmosphere. We then regard the
surface as approximately planar and consider the relative motion of the
solution past the surface. The ca1culation, which is very similar to that
for the electrokinetic coefficient (4, gives
u = eeos (2.37)
1']
which should be the limiting value for a very large particle.

Charged particles of intermediate size distort the electrical lines of
force and a more sophisticated approach is required. More exact
calculations have shown that the electrophoretic mobility is (as we
might expect) a function of the size, shape, and orientation of the
particles. The mobility is, however, generally found to be proportional to
the dielectric constant and to the zeta potential, and inversely
proportional to the viscosity. The electrophoretic mobility is sometimes
written as
{eeos
u=-- (2.38)
1']
where f is a unitless quantity which depends upon the geometrical

factors. For spherical particles, f varies smoothly from 2/3 for a < XA to 1
for a > 100 XA.
Example 2.5 Estimate the electrophoretic mobility of a

particle of radius 3 nm and zeta potential 20 mV in a solution
with E = 78.5, 11 = 10-3 Pa-s, and ionic strength 1 mM.
For 1 mM ionic strength, XA = 9.6 nm, so that a/xA = 0.31 and f '"
2/3; eq (2.38) gives u = 9.3 x 10-9 m 2 s- 1V-l. This mobility is typical
for charged macromolecules (see §3.3).
Practical Electrophoresis
Although eq (2.38) does not explicitly include the thickness of the
diffuse layer, XA, the zeta potential depends on XA (or C) as shown by eq
(2.33) and the geometric factor f depends on alxA. Thus the
electrophoretic mobility is expected to decrease with increasing ionic
strength of the electrolyte solution. The mobility is also proportional, of
course, to the surface charge density on the moving colloidal particle
and, for a protein, is expected to be pH dependent, even at constant ionic
strength; indeed, the mobility is expected to go to zero at the isoelectric
pH. Because of the dependence of the electrophoretic mobility on ionic
strength and pH, a mechanism for control of protein mobilities is
available which is used in practical separation techniques.
In our discussion thus far, we have tacitly assumed that
electrophoresis experiments are done on a suspension of colloidal
particles in liquid solution, and, indeed, the early experiments were.
Thus, in classic experiments done in the 1930's, Tiselius found that
electrophoretic mobilities of protein molecules in solution could be
measured by a moving boundary method (16). Tiselius' apparatus was
essentially a U-tube, shown in very simplified form in Figure 2.10. The
U-tube was constructed so that the solution of the colloid could be placed
in the bottom of the tube and solvent (containing electrolyte) layered on
top on both sides. Electrodes were then mounted in the two ends of the
U-tube and a high voltage applied. When the protein molecules
migrated in the solution, the boundary between the two regions of the
solution remained intact and was seen to move with a velocity which
was easily related to the electrophoretic mobility of the protein. To
minimize convective mixing of the solutions and blurring of the
boundary, it was necessary to very carefully thermostat the U-tube, and,
since the rate of change of the density of water with temperature is
smallest near the density maximum (4°C), best results were obtained
when the system was thermostated at about 2-3°C. In practice, the
Figure 2.10 Moving boundary method

for measurement of electrophoretic
mobilities. (a) Initial condition with
protein solution layered at the bottom of
the U-tube. (b) After passage of current,
boundary between the colloid and
electrolyte solutions has moved. (a) (b)
boundary was observed by measurement of the refractive index. Fairly

complex mixtures can be separated by means of Tiselius'
electrophoresis techniques.
Arne W. K. Tiselius (1902-1971) was a Professor at the University of

Uppsala with research interests in biophysical chemistry. He received
the Nobel Prize in Chemistry in 1948 for his work on electrophoresis.
A great advance in the utility of electrophoresis came with the

marriage of electrophoretic migration with paper chromatography. By
using a solid support of filter paper, most of the problems associated
with convective mixing in solution can be eliminated and diffusive
blurring of the zone boundaries can be reduced as well. Even further
improvement can be obtained if a more homogeneous solid support such
as starch gel or cellulose is used in place of the paper.
The manifold of electrophoresis techniques now available,
including a host of procedures for detecting the protein molecules, goes
far beyond the space available here. Several rather long chapters would
be required to do the subject justice, and the reader is referred to a more
specialized text (17-19) for further details.
2.5 ELECTRODE DOUBLE-LAYER EFFECTS
Unlike a protein molecule where the surface potential and the ion
atmosphere are a response to the solution pH and ionic strength, the
potential of an electrode, and thus the nature of the double layer, can be
controlled by external circuitry. This control permits quantitative
experimental studies of double-layer effects. In this section, we will
discuss two effects which have important consequences In
electrochemical experiments to be described in subsequent chapters.
Electrocapillarity
The interfacial tension (surface tension) at an electrode-solution
interface turns out to be a function of the electrode-solution potential
difference. When the electrode is a rigid solid, the effects of interfacial
tension are difficult to discern, but when the electrode is nonrigid, a
mercury drop for example, the effects are more obvious. In a dropping
mercury electrode, mercury is allowed to flow through a fine-bore
capillary immersed in the electrolyte solution. A mercury drop forms at
the end of the capillary and grows in size until it becomes too heavy to be
supported by the interfacial tension. l Thus changes in interfacial
tension are reflected in changes in the maximum mercury drop size, or
if the flow through the capillary is constant, in the lifetime of a mercury
drop.
Consider a mercury drop attached to a mercury-filled capillary and
suspended in an electrolyte solution. If a second electrode is provided,
we can change the mercury-solution potential difference and thus
induce a positive or negative charge on the mercury drop by passing a
small current through the external circuit. Similarly, we can change
the drop size by allowing mercury to flow through the capillary.
Expansion of the drop involves an increase in surface area and thus
work is done against the interfacial tension, dw = ydA. Similarly,
electrical work is done when the drop is charged, dw = <l>dQ. Thus at
constant temperature, pressure, and composition, the change in Gibbs
free energy is
dG = y dA + dQ
This expression can be integrated
G=yA+Q
and redifferentiated
dG = ydA +A dy+ dQ + Q d<l>
Subtracting the two expressions for dG leaves a form of the Gibbs-
Duhem equation,
o=A dy+ Q d<l>
Introducing the surface charge density, cr = QIA, we have
(:;tR =-0
Substituting for (J from eq (2.13) gives
1 The dropping mercury electrode is discussed in more detail in Chapter 4.

§2.5 Electrode Double-Layer Effects 87
( if'( ) = _ EEo<l> (2.39)

a<l> T,P XA
Integration of eq (2.39) gives

2
Y = Yo- EEo<l> (2.40)
2xA
A plot of interfacial tension us. cell potential should be parabolic with a
maximum when the electrode-solution potential difference is zero.
If mercury flows through the capillary with constant rate u (mass
per unit time), then the mass of the drop after time t is m = ut. The
gravitational force (weight), corrected for the buoyancy of the solution, is
F = utg(1- dldHg)
where d and dHg are the densities of the solution and mercury,
respectively, and g is the gravitational acceleration. The surface tension
force holding the drop to the capillary is
F = 2rrrcY
where rc is the radius of the capillary bore. These two forces are in
balance the instant before the drop falls from the capillary (t = td). Thus
if td, u, d, and rc are measured, the interfacial tension can be computed
Y= ut<&{l- dldHg) (2.41)
21trc
Some typical data from the work of Grahame (1) are shown in Figure
2.11 for several aqueous electrolyte solutions of equal ionic strength.
The curves of Figure 2.11 show the qualitative behavior predicted by
eq (2.41). However, the simple theory predicts identical curves for all
electrolytes of equal ionic strength. The curves of Figure 2.11 coincide at
negative potentials but show rather different behavior near zero or at
positive potentials. When the mercury drop is positively charged, anions
are expected to be concentrated near the electrode surface, and the
different behavior suggests specific interactions of the various anions
with the surface. The "soft" bases like I- and SCN-- apparently interact
more strongly than the "hard" bases like OH-, CI-, and N03-. Since
mercury can be thought of as a soft acid, the interaction most likely
involves some degree of chemical bond formation and we say that the
anions are chemically adsorbed on the mercury surface. When the
potential is made sufficiently negative, anions are desorbed and all
David C. Grahame (1912-1958) was Professor of Chemistry at Amherst

College. He is remembered for his exceedingly careful and thorough
studies of the electric double layer.
0.45
yIN m- 1
0.40
,.,.,........................... .
,. /
0.35 I
./ ---
,.i ................... .
,
; NaCI
0.30 ! NaBr
------- KSCN
_..... _.... KI
0.25
0.50 0.00 -0.50 -1.00
EIV
Figure 2.11 Interfacial tension of mercury in contact with various
aqueous electrolyte solutions. The potential scale is adjusted so that zero
potential occurs at the electrocapillary maximum for a capillary-inactive
electrolyte. Reprinted with permission from D. C. Grahame, Chern. Rev.
1947,41,441, copyright 1947 American Chemical Society.
curves merge together, suggesting that cations (at least Na+, K+, and
Ca2 +) do not specifically interact with the mercury surface, even when it
is highly negatively charged.
It is commonly observed that the drop time of a dropping mercury
electrode depends on potential, usually varying by a factor of 2 or more
over the experimentally accessible range. More complex behavior is
sometimes observed when surfactants are adsorbed on the electrode
surface. Since adsorption is often potential dependent, very rapid
changes in interfacial tension (and thus drop time) with potential are
sometimes observed.
Double-Layer Capacitance
When the potential applied to an electrode immersed in an

electrolyte solution is decreased from zero, the surface charge becomes
negative and the net space charge of the double layer must increase to
maintain overall electrical neutrality. Similarly, an increase in
electrode potential must induce a net negative space charge. From the
point of view of the external electric circuit, the double layer thus
behaves as a capacitor, serving to store electric charge.
Capacitance is defined as the ratio of charge stored to voltage
applied, or, more appropriately in this case, as the derivative (the
§2.5 Electrode Double-Layer Effects 89
Figure 2.12 Capacity of a

mercury drop electrode
in contact with aqueous
NaF solutions, [NaFl = 0.32
0.001,0.01,0.1, and 1.0 M, Clf m-2
as a function of potential
measured vs. a normal 0.24
(1 M KC!) calomel
electrode. The potential
of minimum capacity
corresponds to the elect- 0.16
rocapillary maximum
(i.e., to maximum inter-
facial tension, see 0.08
Figure 2.11). Reprinted
with permission from D.
C. Grahame, Chem. Rev.
1947,41,441, copyright 0.8 0.0 -0.8 -1.6
1947 American Chemical
Society. EIV
differential capacitance)
C=dQ (2.42)
d<l>
If we take C to be the capacitance per unit area (the capacity), then we
can replace Q by the surface charge density cr. Consider first the
capacity of the Gouy layer, Co. Differentiating eq (2.13), we obtain
C - ££0 (2.43)
G - XA
A more exact calculation (see Problems) shows that Co is a function of

<l>a and that eq (2.43) represents the minimum capacity. Minima are
observed experimentally and provide a way to determine the potential of
zero charge for an electrode (20). The capacity us. potential curves for
some aqueous sodium fluoride solutions are shown in Figure 2.12.
Example 2.6 Compute the Gouy layer capacity for a solution

ofa 1:1 electrolyte with concentrations of 0.001, 0.01, and 0.1 M.
The double-layer thicknesses for these concentrations are

(Table 2.1) 9.6, 3.0, and 0.96 nm. Assuming the dielectric
constant of pure water (£ = 78.5), eq (2.43) gives
Co = 0.072, 0.23, and 0.72 F m- 2
for 0.001,0.01, and 0.1 M solutions, respectively.
The curve for 0.001 M NaF solution in Figure 2.12 goes through a
minimum of about 0.06 F m- 2 , in excellent agreement with the value
computed in Example 2.6 considering the simplicity of the theory.
However, the minimum capacity of the 0.01 M NaF solution is about 0.1 F
m- 2 and for 0.1 M NaF the observed capacity at the electrocapillary
maximum is only 0.2 F m- 2 , rather less than the computed values, 0.23
and 0.72 F m- 2 . Furthermore, the capacity minimum at = 0 nearly
disappears for concentrated solutions. The reason for these apparent
discrepancies is not hard to find: we have neglected the effect of the
immobilized ions and solvent dipoles on the surface which comprise the
Helmholtz layer. These also contribute to the capacity. From the point of
view of the external circuit, these two contributions-C H for the
Helmholtz layer and C G for the Gouy layer-behave like capacities in
series. Thus the total observed capacity is given by
1/C = 1/CH + 1/CG
or
C = CGCH (2.44)
CG+CH
For dilute solutions, the capacity of the diffuse layer C G is small
compared with the more or less constant Helmholtz layer contribution,
so that C'" CG. When the diffuse layer becomes more compact and its
capacity becomes large compared with that of the Helmholtz layer, then
the latter contribution should dominate. Indeed, it is found that at zero
potential the double-layer capacity initially increases with concentration
and then levels off. In the case of NaF solutions in contact with
mercury, the Helmholtz contribution to the double-layer capacity at = 0
appears to be approximately 0.29 F m- 2 .
2.6 DEBYE-HUCKEL THEORY
In 1923 Debye and Huckel (21) found a way to calculate ionic activity
coefficients for dilute electrolyte solutions. The basic idea of the theory is
that, just as in the case of the charged surfaces we have been
considering, individual ions possess an atmosphere of ions of opposite
charge. By calculating the electrostatic free energy of interaction of an
ion with its atmosphere, an estimate of the activity coefficient can be
obtained. Debye-Huckel theory starts with the Gouy-Chapman
description of the ion atmosphere derived from the linearized Poisson-
Boltzmann equation.
§2.6 Debye-Huckel Theory 91
Peter J. W. Debye (1884-1966) was one of the outstanding physical

chemists of this century. At one time or another, he held chairs at the
Universities of Zurich, Utrecht, Gottingen, Leipzig, and Berlin, finally
coming to rest at Cornell University during World War II. His
contributions to electrolyte theory were largely made during his tenure at
the Eidgenossische Technische Hochschule in Zurich in the 1920's. Erich
Huckel (1896-1980) was a student of Debye's; their collaboration continued
through most of the 1920's, whereupon Huckel went on to become a pioneer
in quantum chemistry.
Calculation of an Ionic Activity Coefficient

We begin by outlining the thermodynamic part of the argument.
The chemical potential of a species i is given by
(2.45)
where Ilt is the chemical potential of i in its standard state, and ai is its
activity. If the standard state is an ideal 1 M solution,! then in an ideal
solution, the activity is identical with the molar concentration.
Assuming that the standard state chemical potential is unaffected by
considerations of nonideality, the chemical potential in the hypothetical
ideal solution is
Ili(ideal) ::: Ilt + RT In Ci
If we now assume that the departure from ideality is due entirely to
electrostatic interactions of the ion with its surroundings, then the
electrical contribution to the chemical potential is
Ili(elec) ::: Ili - Ili(ideal)
Ili(elec) ::: RT In ai - RT In Ci
and, since the ratio of the activity to the concentration is the activity
coefficient,
(2.46)
we have
Ili(elec) ::: RT In Yi (2.47)
Thus if we can calculate the electrostatic contribution to the chemical
potential of an ion, we win have the activity coefficient.
Consider an ion of charge zie surrounded by its ion atmosphere and
suppose that the electric potential at the surface of the ion is <D. We now
1 The standard state in most treatments of activity coefficients is the 1 molal ideal
solution. Since most practical work employs the molar concentration scale, we shall
use the 1 M standard state here. The penalty for this choice is that most data
tabulations must be converted to molar concentration units.
consider the hypothetical process of reversibly charging an initially

uncharged ion up to its actual charge zie. The work done in this
charging process is the change in the Gibbs free energy, which in this
f'"
case is the electrical contribution to the chemical potential:
Ili(elec) = <l> dQ (2.48)
where <l> is the contribution to the potential from the ion atmosphere.
Notice that this chemical potential has units of energy per ion; we will
convert to a molar basis later.
For a spherical particle of radius a, the surface potential <l>a is
related to the surface charge density () by (see Problems)
<l> - aa
a - ££0(1 + a/xAl
where £ is the dielectric constant of the medium and XA is the thickness
of the ion atmosphere (called the Debye length in this context). If Q is the
total charge on the ion, the surface charge density is
a =--.SL
4n:a 2
We then have
<l>a =~ 1 (2.49)
4n:££0 a (1 + a/xAl
For the thermodynamic calculation, we need the potential at the ion site
due to the ion atmosphere. Equation (2.49) contains contributions from
both the ion atmosphere and the central or test ion. However, the latter
contribution is easily computed:
<l>central(a) = _Q_-
4n:££(j1
Subtracting this from eq (2.49), we have the ion atmosphere contribution
to the potential at the central ion:
<l>atm(a) =_~_1_
4n:££o a + XA
Substituting this expression in eq (2.48) and integrating, we have
Ili(elec) = _ _ 1 _ +1
_
4n:££0 a XA
r
Jo
e
Q dQ
Z·~2 1
Ili(elec) :;:; - - '- ---
8n:££0 a + XA
Finally, we substitute this expression into eq (2.47), solving for In 'Yi. In

doing so, we also multiply by Avogadro's numberl to convert to a molar
basis. The resulting equation is
Z·2e 2 1
In 'Yi = - -,------,-/----::-= (2.50)
87tEEokT a + XA
Mean Ionic Activity Coefficients

Unfortunately, there is no way of obtaining single ion activities or
activity coefficients from experimental measurements. Thus we must
pause for a moment to deduce the relationship between the single ion
activity coefficients calculated with the Debye-Huckel theory and the
activity coefficients which are actually measurable.
Consider an electrolyte, which, on dissolution of one mole, gives VI
moles of cations and V2 moles of anions. We define the mean chemical
potential of the electrolyte as
Il± = _V-,,-11l--=I'-.+_V-,2!l"'--=.2 (2.51)
VI + V2
where Il± is related to the mean activity a± by
Il± = Il±o + RT In a± (2.52)
Substituting the analogous equations for the single ion chemical
potentials, eq (2.45), into eq (2.51) gives
(2.53)
where V = VI + V2. Comparing eqs (2.52) and (2.53), we see that the mean
ionic activity is related to the single ion activities by
(2.54)
Since a± = C±'Y±, al = ClYI, and a2 = C2'Y2, eq (2.54) implies the following

relationships:
(2.55a)
C ±v -_ C IV1 C 2V2 (2.55b)
(2.55c)
But, since CI =VIC and C2 =V2C, C± is not the ordinary molar electrolyte
concentration C, but is
1 Multiplication by Avogadro's number is equivalent to conversion of the gas

constant in the denominator to Boltzmann's constant.
(2.56)
Example 2.7 Determine the relationship between the mean

ionic concentration C± and the ordinary molar concentration C
for binary electrolytes with ionic charges from 1 to 4. For a
symmetrical electrolyte (1:1, 2:2, etc.), vI = V2 = 1, v = 2. Thus
(VIVIV~2)lIv = ~ = 1
and C± = C. For a 1:2 electrolyte, V+ = 2, v_ = 1, v = 3 and
(VIvIV~2)lIv =~22x 1 =1.587
so that C± = 1.587 C. Similar calculations generate the
following table:
Electrolyte C±/C Electrolyte C±/C

1:1 1.000 2:3 2.551
1:2 1.587 2:4 3.175
1:3 2.280 3:3 1.000
1:4 3.031 3:4 3.536
2:2 1.000 4:4 1.000
We can now relate the mean ionic activity coefficient y± to the

computed ionic activity coefficients y+ and y_. Taking logs of eq (2.55c) we
have
v In(y±) = vlln(YI) + V2 In(Y2)
and, substituting eq (2.50) into this expression, we have
1 _ e2 1 VIZ} + V';ft22
n Y± - - 81t€€okT a + XA VI + V2
But, since IVIZII = IV2Z21 the third factor on the right reduces to IZlZ21 ,
and we are left with
In Y+ = _ e 21 z IZ21 1
- 81t€€okT a + XA
Recalling that XA is inversely proportional to the square root of the ionic
strength, eq (2.8), we can rewrite this expression in terms of the ionic
strength:
1 Al z lZ21 fl (2.57)
og y± = 1 + Ba fi
In eq (2.57), A and B are constants, I is the ionic strength in
concentration units of mol L-l, and we have converted to common logs.
Inserting the values of the constants and assuming a temperature of
25°C and the dielectric constant of water (78.54), we can write
1 0.5092I z 1Z21 fl (2.58)
ogy± =- 1 + 3.29afl
In eq (2.58), the ionic strength and IZ1Z21 are known for any given
electrolyte solution. The parameter a (in nm) entered the theory as the
ionic radius. However, in going from single ion activity coefficients to
mean ionic activity coefficients, it has lost its simple meaning, and
should to be regarded as a parameter adjustable to fit the theoretical
activity coefficients to experiment.
At low ionic strength, the second term in the denominator of eq
(2.58) can be neglected compared with unity and we have the limiting
form of Debye-Huckel theory:
log y± = - 0.50921 ZlZ21 fl (2.59)
Equation (2.59) contains no adjustable parameters and thus is an
absolute prediction of the activity coefficients.
Experimental Determination of Activity Coefficients

Thermodynamic quantities such as enthalpies or free energies of
formation, half-cell potentials, acid-base dissociation constants,
solubility product constants, complex formation constants, etc., are
generally determined as functions of concentration and extrapolated to
infinite dilution. Since activity coefficients approach unity at infinite
dilution, these tabulated quantities refer to ideal solutions. Once the free
energy change for an ideal solution process is known, comparison with
values of flGo determined for finite concentrations can yield the activity
coefficients.
The activity of the solvent or of a volatile solute can be determined
from vapor pressure measurements and the activity coefficients
computed from the departure of the measured pressure from ideal
behavior-Raoult's law or Henry's law. Solute activity coefficients can
also be determined from freezing point or osmotic pressure data. A text
on physical chemistry should be consulted for further details on these
methods.
In a cell potential measurement, the potential is given by the Nernst
equation, eq (1.18). With
the Nernst equation can be written in the form

E + RT InF(Ci) =Eo_RT InF(Yi)
nF nF
If the concentrations Ci and the cell potential E are known, the left-hand
side of this expression can be plotted, usually vs. the square root of ionic
strength. Extrapolation to I = 0 gives the standard cell potential EO and,
knowing this quantity, the activity coefficients can be determined.
Example 2.8 Consider the cell

PtIH2(g)IHCI(aq)IAgCl(s)IAg
with cell reaction
2 AgCI(s) + H2(g) ~ 2 Ag(s) + 2 H+(aq) + 2 CI-(aq)
The potential is given by
E =EO _RT In [H+]2[Cn2RT In (nd2(YCl-f
2F 2F
PH 2
where we have assumed ideal behavior for H2(g). Since
y±2 ='YH+ YCI-
the N ernst equation can be rearranged to
F(l) =E + MIn [H+]2[Cn2 =EO - 2RT..ln Y±
2F PH2 F
The left-hand side of this expression, F(I), is plotted vs. fl in
Figure 2.13a. Extrapolation to I = 0 yields EO = 0.2224 V.
Subtraction of EO from each value of F(I) and rearrangement
gives
F[EO -F(l)]
Y±= exp 2RT
The values of y± are plotted vs. fi in Figure 2.13b along with
values computed from eq (2.59).
Activity coefficients can also be determined from measurements of

the solubility of a slightly soluble salt as a function of ionic strength.
Consider the solubility equilibrium for a salt MmAn
MmAn(s) +Z m Mn+ + n Am-
The solubility equilibrium expression is
Ksp =aMm aAn =a±v
or
§2.6 Debye-Hiickel Theory 97
1.0
F(I)N (a)
y±
0.230
0.9
0.225
1oC..L..................................&....L.............................................I 0.8
0.0 0.1 0.2 0.0 0.1 0.2
[112 [1/2
Figure 2.13 (a) Corrected cell potential F(I) for the cell of Example 2.8 as
a function of the square root of ionic strength. The solid line corresponds
to a least-squares fit of the data to a parabola. (b) Mean activity
coefficients, y±, computed from the cell potentials. The solid line
corresponds to activity coefficients computed using eq (2.59). Data from
Harned and Ehlers (22).
Ksp = (C±y±)V
where v = m + n. If the solubility is measured at several ionic strengths,
adjusted with an electrolyte which has no ion in common with the
slightly soluble salt, Ksp can be determined by extrapolation to zero ionic
strength and, knowing K sp , the mean ionic activity coefficients y± can be
computed for each solution.
Some data from such a study (23) are shown in Figure 2.14 for the
1:1 electrolyte,
[Co(NH3M C204)] [CO(NH3)2(N02)2( C20 4)]
the 1:2 electrolyte,
and the 3:1 electrolyte,

[Co(NH3)6] [Co( NH3)2(N02)2(C204)h
For the data shown, extending up to ionic strengths of about 0.01 M, the
agreement with theory is seen to be quite good, probably within
experimental error.
At higher ionic strengths, the plots begin to curve. In most cases,

plots of activity coefficients vs. ionic strength go through minima and
the activity coefficients increase with concentration at high ionic
strength. Examples of this behavior are shown in Figure 2.15 for HCI,
KCI, CaCI2, and LaCI2. Even for the two 1:1 electrolytes, the behavior in
concentrated solutions is rather different.
Inclusion of the parameter in the denominator of eq (2.58) helps the
theory match experimental results to somewhat higher ionic strengths,
but it cannot ever produce a minimum in the activity coefficient - ionic
strength curve. Some experimental data for NaCI solutions are
compared with theoretical predictions from eqs (2.58) and (2.59) in
Figure 2.16. In fitting these data, the parameter a was taken to be 0.397
nm. With inclusion of the adjustable parameter, Debye-Huckel theory
activity coefficients fit the experimental data up to an ionic strength of
about 0.04 M. Beyond this the one parameter can no longer fit the data by
itself and some further development is necessary.
-0.00 2.0
log y± y+ KCI
-0.03 1.5 CaCl2
LaCIs
-0.05 HCI
1.0
-0.08
-0.10
0.0
0.00 0.04 0.08 0.12 0.0 0.5 1.0 1.5 2.0
[112 [112
Figure 2.14 Activity coefficients Figure 2.15 Mean ionic activity

computed from the solubilities of (a) coefficients for several electrolytes at
1:1, (b) 1:2, and (c) 3:1 electrolytes at 25°C. Data from Robinson and Stokes
15°C in pure water (crosses) and in (B9).
solutions with ionic strength adjusted
with NaCI (diamonds) and KNOs
(squares). The straight lines
correspond to eq (2.59) (with the
coefficient adjusted for the different
temperature). Data from Bronsted and
LaMer (23).
-0.00
logy±
-0.05
Figure 2.16 Mean ionic -0.10

activity coefficients of
NaCI solutions. Points are
experimental data (B9).
The solid line is calculated -0.15
from eq (2.59), the dashed
line from eq (2.58), and the
dotted line from eq (2.70).
The parameter a in eqs
(2.58) and (2.70) was taken -0.20
to be 0.397 nm and ns in eq 0.0 0.5 1.0 1.5
(2.70) was 3.5. [l/2
Conversion to Other Concentration Scales

Comparison of activity coefficients appropriate to the molar
concentration scale with literature values given for molal
concentrations or mole fractions can be confusing. As an aid to
conversion, we give here the necessary equations. The details of the
derivations are given by Robinson and Stokes (B9).
Molal concentrations, m, are defined as the number of moles of
solute per kilogram of solvent. Conversion to molar concentrations Cis
as follows:
c= md (2.60)
1 + mMBI1000
m= C (2.61)
d-CMw1000
where MB is the solute molecular weight and d is the solution density.
If an electrolyte of molality m ionizes to VI cations and V2 anions,
then the mole fractions Xi are
x. - vim (2.62a)
• - vm + 1000lMA
where v = VI + V2, and MA is the molecular weight of the solvent. In
terms of molar concentrations, the mole fractions are given by
x. - ViC (2.62b)
, - vC + 1000 dlMA - CMBIMA
If the molal concentration scale is used, then the mean ionic
activity is defined by eq (2.51) with
a± = m±y±m
The mean ionic activity coefficient y±m is related to the activity coefficient
for the molar concentration scale by
y±m = Cd y± (2.63)
m 0
where do is the density of the solvent. If the mole fraction scale is used,
the corresponding activity coefficients are
X d + 0.001 C(VMA - MB
(2.64)
~ = do ~
Solvent Activity
The standard state of the solvent in an electrolyte solution is usually
taken to be the pure solvent at 1 bar total pressure and 25°C. In simple
treatments of solution equilibrium, it is usually assumed that the solvent
activity is I, but a more exact value is sometimes required. If the
solution were ideal, the solvent activity would be equal to the mole
fraction, but in solutions where the mole fraction of solvent is
significantly less than unity, nonideality is usually sufficiently
important that the mole fraction is a poor estimate of the activity.
The most straightforward way to determine solvent activity is to
measure the partial pressure of solvent vapor. We define the activity of
the solvent in solution in terms of the chemical potential
IlA = IlA ° + RT In aA (2.65)
where IlA ° is the chemical potential of the pure solvent. When the pure
solvent is in equilibrium with its vapor,
IlAo = IlA(vapor) = IlAO(vapor) + RT In PAc
where PA ° is the equilibrium partial pressure. 1 When a solution is in
equilibrium with vapor, we have
IlA = IlA O(vapor) + RT In PA
Subtracting these expressions gives
1 In very careful work, account must be taken of nonideality of the vapor phase as
well. We ignore that complication here.
IlA -IlA °= RT In PA
PAo
so that
(2.66)
While the vapor pressure method is theoretically straightforward, it
is usually rather inconvenient experimentally. Another approach to
solvent activities is through solute activity coefficients. Consider the
fundamental equation for the Gibbs free energy:
dG = P dV - T dS + L Ili dni
i
Integrating and redifferentiating, we have
G = PV - TS + L J.li ni
i
dG = P dV + V dP - T dS - S dT + L (Ili dni + ni dll;)

i
Subtracting the two expressions for dG gives the Gibbs-Duhem equation
o = V dP - S dT + L ni dlli
i
which at constant temperature and pressure reduces to
L ni dlli = 0
i
If we have only two components, the solvent (A) and the solute (B), then
nA dllA = - nB dllB
Equation (2.65) gives IlA and a similar expression can be written for IlB.
Differentiating and substituting into the Gibbs-Duhem equation gives
nART d In aA = - nBRl' d In aB
But since
and
we have
which can be integrated to give

In aA =-f-
A 0
i y+c
CV d In(y±C) (2.67)
where we have assumed a constant number of moles of solvent. With

sufficient solute activity coefficient data, eq (2.67) can be integrated
numerically to obtain the solvent activity.
Beyond Debye-Huckel Theory

Debye-Huckel theory involves the calculation of the free energy
associated with the formation of an ion atmosphere about ions in an
electrolyte solution. There are details of the theory which leave much to
be desired, but perhaps the greatest limitation is the implicit assumption
that the formation of the ion atmosphere accounts entirely for deviations
of the activity coefficients from unity. Clearly, the assumption must be a
good one at high dilution because the limiting law form of Debye-Huckel
theory is essentially perfect. Yet we know that at higher concentrations,
the activity coefficients begin to increase again.
We can obtain a qualitative understanding of the behavior of
electrolyte activity coefficients at high concentrations (and indeed derive
a semiquantitative theory) from the following considerations. 1 When a
salt dissolves in water, the ions are solvated with several water
molecules held rather tightly to the ions. We should regard those water
molecules held in the solvation sheath of the ions as part of the ionic
species rather than as "free" solvent; thus the concentration of "free"
solvent molecules is lowered. This effect is very small in dilute solutions
but is quite significant at high electrolyte concentrations. For example,
in NaCI solutions, it appears from other experimental evidence that
about seven water molecules are involved in solvation of the two ions.
Thus in 0.01 M NaCI, about 0.07 mole of water per liter is removed from
the free solvent. However, since the concentration of water is about 55.5
M, this is quite negligible. In 1 M NaCI, on the other hand, 7 moles per
liter of water is immobilized, leaving 48.5 moles of free water, a more
significant change.
Consider an electrolyte solution containing nA moles of solvent per
mole of solute. The solute ionizes to give VI moles of cations with
chemical potential III and V2 moles of anions with chemical potential 112.
Apparently, the total free energy of the solution is
G = nAIlA + VIllI + V21l2
If we take into account the ns moles of solvent immobilized by the ions,
we can also express the total free energy by
G = (nA - nS)IlA + VIllI' + V21l2'
1 See Robinson and Stokes (B9) for a discussion of other approaches to this problem.
where Ill' and 112' are the chemical potentials of the solute ions corrected
for bound solvent. Both expressions for G should be correct, so that
equating them gives
nSIlA + V1(111 -Ill') + V2(112 -Ill') = 0 (2.68)
It is convenient to change (temporarily) to mole fractions as measures of
the activities with activity coefficients Y1 X and Y2X . The chemical
potentials then are
IlA == IlAo + RT In aA
III = 111° +RTlnXl +RTlnnX
Ill' == 111°' +RTlnX1' +RTlnnX

with similar equations for 112 and 112'. Substitution of these expressions
into eq (2.68) yields
nSIlA ° + Vl(1l1° -111° ') + Vi1l2° -112° ')
RT
XIy[C X2Y~
= - ns In aA - VI In - - - v2ln - - (2.69)
xl'YfC X2'YJC
The mole fractions are
X I' = __V-'I'---_
nA -ns +v
where v = VI + V2. Thus we have
Xl _ nA -ns+v _ X 2
Xl' - nA + v - X2'
When the solute concentration is very small, nA is very large (we have
assumed one mole of solute). Thus as the solute concentration
approaches zero, XliX I' ~ 1, aA ~ 1, all the activity coefficients
approach 1, and all the log terms on the right-hand side of eq (2.69) go to
zero; thus the sum of standard state terms on the left-hand side of the
equation must be zero, independent of the solute concentration.
Rewriting eq (2.69) with the new information included, and using eq
(2.55c) to convert the activity coefficient terms to mean ionic activity
coefficients, we have
nA -ns + v X X
ns In aA + v In + v In Y± = v In Y±
nA +v
We now convert back to the molar concentration scale. First we replace

nA, the number of moles of solvent per mole of solute, by
nA :::.1QQild._ MB
CMA MA
where MA and MB are the molecular weights of the solvent and solute,
respectively, d is the solution density, and C is the molar concentration.
Substituting eq (2.64) for In y±X and taking the Debye-Huckel result, eq
(2.57), for In y±X', we obtain (after some algebra)
I Al z lZ21U nSI I d-0.001qMB-MA (v-ns)] (2.70)
ogy±:::- l+BaU-V ogaA- og do
The parameters A and B are given in eq (2.58) and the solvent activity is
measurable, at least in principle. Thus a and ns are the only adjustable
parameters.
Equation (2.70) gives a very good fit to activity coefficients for
concentrated solutions. Some typical results are shown in Figure 2.16
for NaCI solutions. Best-fit values of the parameters a and ns for several
electrolytes are given in Table 2.2. The values of ns, although not grossly
out of line with solvation numbers obtained from other experimental
approaches, are generally somewhat small and certainly should not be
taken as reliable estimates of the degree of solvation.
Table 2.2 Empirical Values of ns and a
Electrolyte ns alnm
HCI 8.0 0.447
HBr 8.6 0.518
NaCI 3.5 0.397
NaBr 4.2 0.424
KCI 1.9 0.363
MgCl2 13.7 0.502
MgBr2 17.0 0.546
From Robinson and Stokes (B9).
References 105
REFERENCES
(Reference numbers preceded by a letter, e.g. (B2), refer to a book listed in

the Bibliography.)
1. D. C. Grahame, Chem. Rev. 1947,41,441.

2. R. Parsons, Modern Aspects of Electrochemistry 1954, 1, 103.
3. D. M. Mohilner, Electroanalytical Chemistry 1966,1,241.
4. R. M. Reeves, Modern Aspects of Electrochemistry 1974,9, 239.
5. H. von Helmholtz, Wied. Ann. 1879,7,337.
6. L.-G. Gouy, Compt. Rend. 1909,149,654; J. Phys. 1910,9,457.
7. D. L. Chapman, Phil. Mag. 1913, 25, 475.
8. O. Stern, Z. Elektrochem. 1924,30, 508.
9. D. J. Shaw, Introduction to Colloid and Surface Chemistry, 3rd ed,
London: Butterworths, 1980.
10. H. C. Hamaker, Physica, 1937,4, 1058.
11. R. Hogg, T. W. Healy, and D. W. Furstenau, Trans. Faraday Soc.
1966, 62, 1638; G. R. Wiese and T. W. Healy, ibid., 1970, 66, 490.
12. S. S. Dukhin and B. V. Deryagin, in Surface and Colloid Science,
Vol. 7, E. Matijevic, ed, New York: Wiley-Interscience, 1974.
13. L. On sager, Phys. Rev. 1931,37,405,38,2265.
14. A. J. Rutgers and M. de Smet, Trans. Faraday Soc. 1945,41,758.
15. H. G. Bungenberg de Jong, Rec. trav. chim. 1924,43, 189.
16. A. Tiselius, Kolloid-Z. 1938,85, 129.
17. M. Bier, ed, Electrophoresis, New York: Academic Press, 1959.
18. L. P. Cawley, Electrophoresis and Immunoelectrophoresis, Boston:
Little, Brown, 1969.
19. D. J. Shaw, Electrophoresis, New York: Academic Press, 1969.
20. R. S. Perkins and T. N. Anderson, Modern Aspects of
Electrochemistry 1969, 5, 203.
21. P. Debye and E. Huckel, Phys. Z. 1923,24, 185.
22. H. S. Harned and R. W. Ehlers, J. Am. Chem. Soc. 1932,54, 1350.
23. J. N. Bronsted and V. K. LaMer, J. Am. Chem. Soc. 1924,46, 555.
PROBLEMS
2.1 Derive eq (2.11). Hint: Use V2 in spherical polar coordinates and

make the substitution <P(r) = u(r) / r.
2.2 Derive eq (2.12). Hint: With u = zFc'PIRT, show that eq (2.5) can be
written
V 2 u = XA- 2 sinh(u)
This expression is then integrated to obtain
duldx = --{21xA) sinh(uI2)
A second integration, followed by rearrangement, yields eq (2.12).
2.3 Use the exact solution to the Poisson-Boltzmann equation for a

symmetrical electrolyte to show that the surface charge density, eq
(2.13), is the first term of a power series:
cr = eeo<Pa [1 + 1 (ZFc'P a )2 + ...J

XA 6 2RT
2.4 Use eq (2.11) to derive an expression for the surface charge density
cr at a spherical surface as a function of the surface potential <P a ,
the radius of the sphere a, and the ion atmosphere thickness XA.
2.5 Derive eq (2.27) for the electrokinetic coefficient U3 by integration of

Poiseuille's equation.
2.6 Derive eq (2.28) for the electrokinetic coefficient U4 by equating the

electrical and viscous forces on a cylindrical volume of radius r
and integrating to obtain vCr) and then integrating again to
determine the flux. Hint: The electrical force on a volume element
dV is dF =Ep(r)dV, where E is the field strength, E = -11c'PIL, and
per) is the space charge density. Integrate d<p to obtain the force on
the cylindrical volume. Use Newton's law of viscous flow to
compute the viscous force.
2.7 Use the result of Problem 2.3 to show that eq (2.43) is actually the
minimum on a C us. <Pa curve.
2.8 Consider an electrochemical cell with a planar electrode of area 1
cm 2 and a double-layer capacity of Cd = 0.20 F m- 2 in contact with
an electrolyte solution. A reference electrode is placed in the
Problems 107
solution at a point such that the solution resistance between the

electrodes is Rs = 1000 n. Suppose that the electrode-solution
interfacial potential is initially 0 V and that at time t = 0, the cell is
polarized so that the electrode is at +0.1 V relative to the bulk
solution. Neglecting charge transfer at the electrode and the
variation of Cd with potential, compute:
(a) the charge stored by the double layer at t = 00.
(b) the initial charging current io
(c) the time required for the charging current to fall to 0.01 io.
Hint: The potential across the cell can be written E = iRs + QICd,
where i = dQldt and Cd is the double-layer capacitance.
2.9 Butler and Huston (Anal. Chern., 1970, 42, 1308) used
electrochemical measurements to obtain activity coefficients in
mixed electrolyte solutions. Using solutions containing known
concentrations of NaCI and NaF, they measured potentials of the
following cells:
AgIAgCI(s)IN a+(aq),CI-(aq),F-(aq)IN a+ -selective glass electrode
AgIAgCI(s)INa+(aq),CI-(aq),F-(aq)IF--selective electrode
Na+-electrodeINa+(aq),CI-(aq),F-(aq)IF--selective electrode
What combinations of activity coefficients were obtained from each
of the three cells?
2.10 Using eq (2.67) and the Debye-Htickellimiting law, eq (2.59), derive

an expression for the solvent activity at low electrolyte
concentrations (assume constant volume).
2.11 Harned and Geary (J. Am. Chern. Soc. 1937,59,2032) measured the
potential of the cell
PtIH2(g)IHCl(O.01 M), BaCI2IAgCI(s)IAg(s)
at 25°C as a function of the BaCl2 concentration. The potential of
the cell is given by the Nernst equation, corrected for activity
coefficients:
E = EO _RT In [H+] [CI-] RT In it
F PH 2 F
Potentials, corrected to the standard state pressure of hydrogen,
are as follows:
[BaCI2J 0.0 0.00333 0.00667 0.01333 0.02000 0.03000

EIV 0.46411 0.45273 0.44529 0.43539 0.42864 0.42135
By a clever extrapolation procedure, the authors found that EO =

0.22234 V. Use these data to compute the mean ionic activity
coefficient for HCI in BaCl2 solutions for the six concentrations
given. Compute the ionic strengths of these solutions and use eqs
(2.59) and (2.58) to estimate the activity coefficients [assume that a =
0.447 om in eq (2.58)].
2.12 The mean ionic activity coefficient of 1.00 molal KOH is 0.735. The
solution density is 1.0435 g cm· 3 and the density of pure water is
0.9971 g cm-3. What is the molar concentration of KOH? What is the
mean ionic activity coefficient on the molar scale?
2.13 At the half-neutralization point in the titration of 0.100 M acetic

acid with 0.100 M sodium hydroxide, the concentrations of acetic
acid and sodium acetate, by definition, are equal, and it is
customary to say that the pH equals the pKa of the acid. If,
however, we take account of the activity coefficient of sodium
acetate, estimating y± with the Debye-Huckellimiting law, eq (2.59),
by how much does the pH actually differ from pKa at the half-
neutralization point?
2.14 Use the Debye-Huckel limiting law, eq (2.59), to estimate the

potential of the Fe3+IFe 2+ half-cell when [Fe(N03)2] = [Fe(N03)3] =
[HN03] = 0.002 M.
3
ELECTROLVTIC
CONDUCTANCE
With this chapter we begin consideration of experiments where

current passes through an electrochemical cell. We will focus first on
the measurement of the conductance of an electrolyte solution under
conditions where the electrode-solution interface can be ignored. The
interpretation of conductance data will lead to consideration of the
contributions of individual ions to solution conductance and to
conductance as a transport property related to diffusion. Finally we will
consider membrane and liquid junction potentials which arise because
of differences in ion transport rates.
3.1 CONDUCTIVITY
The resistance of any current-carrying device, whether it is a piece

of wire or an electrolyte solution, is defined by Ohm's law as the ratio of
the voltage across the device to the current flowing through it, R = 11(J>/i.
We are accustomed to think of resistance as a constant, dependent on
temperature perhaps, but independent of the applied voltage and current
flow; such a resistance is referred to as ohmic. Although some
resistances encountered in electrochemistry are nonohmic and depend
on voltage, it is usually a good approximation to consider the resistance
of an electrolyte solution as ohmic. Resistance is an extensive property
of the system since it depends on the size (and shape) of the current-
carrying device. For a device of uniform cross-sectional area A and
length L, we can define an intensive quantity called the resistivity, p:
p =RA/L
In discussing electrolyte solutions, it is more convenient to speak of
the conductance, the reciprocal of resistance, and of the conductivity, 1(,
the reciprocal of resistivity:
I( = LIRA (3.1)
109
110 Electrolytic Conductance
The unit of resistance is the ohm 0, so that resistivity has units of a-m.
The unit of conductance is the siemens (S), 1 S = 1 0-1, so that the units of
conductivity are S m- 1.
Example 3.1 Compute the conductivity of an electrolyte

solution if the cell has parallel electrodes 1 cm2 in area spaced
1 cm apart and the resistance is 1000 a.
Substituting numbers into eq (3.1), we have

K = 0.01 m =0.1 a-1m-1 =0.1 S m-l
(1000 0)(0.0001 m2)
Before proceeding with an analysis of conductance data, we must

give some attention to the measurement of the conductance (or
resistance) of an electrolyte solution. This is a nontrivial problem and
we must first ask how the resistance of a solution can be measured at all
without running afoul of the resistance and capacitance of the electrode-
solution interfaces which are present whenever we put electrodes into
the solution. Thus, we first consider the electrical equivalent circuit of
an electrochemical cell.
Electrochemical Cell Equivalent Circuit

There is a theorem in electrical circuit analysis which states in
effect that any two-terminal device may be represented (at least
approximately) by some combination of potential sources, resistances,
capacitances, and inductances. We saw in Chapter 1 that every
electrode-solution interface is a potential source, and, in Chapter 2, that
the electrode double layer behaves like a capacitance. The electrode-
solution interface may also be thought of as having an impedance which
represents the finite rate of mass transport and electron transfer at the
electrode surface. l If we add to these the series resistance due to the
solution itself, we can draw the electrochemical cell equivalent circuit
shown in Figure 3.1; ~<l>1 and ~<l>2 are the two electrode-solution potential
differences, the difference between which is the equilibrium cell
potential, E. Cl and C2 are the double-layer capacitances at the two
electrodes, ZFl and ZF2 are the faradaic impedances, and R is the
solution resistance. Our goal is the measurement of R without
interference from the other equivalent circuit parameters.
1 This so-called faradaic impedance has both a resistive and a capacitive

component; see §6.4.
§3.1 Conductivity ill
Figure 3.1 An electrochemical cell equivalent circuit.
If we were simply to attach an ammeter to the cell and measure the

direct current, i, we might use Ohm's law to find the cell resistance
Rcell = (.M>l - !J.(J)2)li
but we see from Figure 3.1 that the resistance thus obtained includes a
contribution from the faradaic impedances. Even if we were able to
separate the cell resistance into its components, there are other
problems. The resistive component of a faradaic impedance turns out to
be nonohmic, that is, it is not constant but depends on the potential;
furthermore, when a finite current is drawn, the cell potential does not
have its equilibrium value. In any case, it is most convenient to use
identical electrodes and a homogeneous solution so that !J.(J)1 = !J.(J)2 and
the net cell potential is zero. We could pass current using an external
potential source, but then electrolysis would change the solution
composition and there would still be problems with the faradaic
impedances.
The solution to these problems is to use alternating current in the
resistance measurement. If we apply an a.c. voltage across the cell and
measure only the a.c. component of the current, considerable
simplification is obtained. Since we are only concerned with the
alternating current which flows as a result of an applied sinusoidal
voltage, the d.c. potential sources, !J.(J) 1 and !J.(J)2, can be ignored.
Furthermore, the impedance of a capacitor toward flow of a sinusoidal
current is 1/00C, where 00 is the angular frequency (00 = 21t times the
frequency in Hz). If the capacitance of the electrode double layer is on
the order of magnitude of 10 /IF and the frequency of the a.c. signal is
1000 Hz, then the impedance ofthe capacitor is 1/(21t)(1000 s·l )(10. 5 F) = 16
Q. Faradaic resistances depend on the d.c. potential across the cell and
are often much larger than the impedance of the double-layer
capacitance. Referring to the equivalent, circuit of Figure 3.1, we see that
the path for a.c. current is primarily through the double-layer
capacitance. Thus to a good approximation, the equivalent circuit of the
cell may be reduced to a capacitor and resistor in series, as shown in
Figure 3.2. The equivalent circuit capacitance C is the series equivalent
of the two double-layer capacitances, C = CIC2/(CI + C2), and the

equivalent circuit resistance R is nearly exactly the solution resistance.
Figure 3.2 Equivalent circuit for large

electrodes and alternating current.
Measurement of Solution Resistance
The classical method for measurement of an unknown resistance

Rx involves the Wheatstone bridge (shown in Figure 3.3a). The bridge is
balanced by adjusting the variable resistance R s to zero the
galvanometer G. Analysis of the circuit then shows that the unknown
resistance Rx is given by
(a) (b)
v
ose
Figure 3.3 Wheatstone bridge circuits. (a) Direct-current bridge for

measurement of resistance. (b) Alternating-current bridge for measurement
of the resistance and capacitance of an electrochemical cell.
The d.c. Wheatstone bridge must be somewhat modified for

measurement of solution resistances as shown in Figure 3.3b. The
battery is replaced by an audio-frequency oscillator (usually operating at
about 1000 Hz) and the galvanometer is replaced by an a.c. detector,
perhaps an oscilloscope or simply a set of earphones. The bridge is
provided with an adjustable capacitor C s so that both the capacitive and
resistive parts of the impedance can be balanced.
From our discussion so far, it should be clear that the design of the
conductance cell is of considerable importance. It is necessary to have
relatively large electrodes in order that the double-layer capacitance be
as large as possible. Since the chemistry of the electrode process should
not interfere with the conductance measurement, an inert electrode
§3.1 Conductivity 113
material is used; the almost universal choice is platinum. A fine layer

of amorphous platinum (platinum black) is usually deposited on the
electrodes so that the effective surface area is much larger than the
apparent geometrical area and the double-layer capacitance is
correspondingly increased. The electrodes must be rigidly mounted so
that the distance between them is constant. Referring to eq (3.1), we see
that the conductivity depends not only on the measured resistance, but
also on the distance between the electrodes and the electrode area. Since
these geometrical factors are not easily measured, it is customary to
calibrate a conductance cell by first measuring the resistance when the
cell is filled with a solution of known conductivity. In this way the cell
constant, L / A, is obtained experimentally and the geometry need not be
measured accurately. Two common conductance cell designs are
shown in Figure 3.4. See reviews by Shedlovsky and Shedlovsky (1) or by
Holler and Enke (2) for further details.
Molar Conductivity
Development of the techniques for measuring solution

conductivities occupied the last half of the nineteenth century (with
further elaborations as electronic instrumentation evolved), but the first
observations of the phenomenon were made by Erman in 1801 (3). In
experiments on the electrolysis of water (discovered by Nicholson and
Carlisle in 1800), Erman found that the current increased with
Pt electrodes
(b)
Figure 3.4 Two types of conductance

cells. (a) Cell with electrodes
rigidly mounted at a fixed
separation; (b) an electrode assembly
which can be dipped into a solution insulating
contained in a beaker. spacers
increasing salt concentration. This observation was confirmed by later

workers and it became customary to define another quantity, called the
molar conductivity to approximately factor out the concentration
dependence. Molar conductivity is defined byl
A=KlC (3.2)
where C is the concentration in mol m- 3 , (1 mol m- 3 = 10- 3 mol L-I = 1
mM). The units of A are S m 2 moI-l.
Example 3.2 A 0.100 M solution of KCI, known to have a

conductivity of 1.2896 S m- I , was found to have a resistance of
89.3 n when placed in a conductance cell. What is the value of
L / A? A 0.200 M solution of NaCI, measured in the same cell,
gave Rx = 56.6 n. What is the conductivity of the solution?
What is the molar conductivity?
According to eq (3.1),
L/A = KR = (1.2896 S m- I )(89.3 n) = 115.2 m- I
so that for the NaCI solution,
1C =(115.2 m- I )/(56.6 Q) =2.035 S m- I
A = (2.035 S m- I )/(200 mol m- 3 ) = 0.0102 S m 2mol- I
The molar conductivities of several salts are plotted as functions of

concentration in Figure 3.5. The molar conductivity depends on
concentration; the magnitude of the dependence varies from one salt to
another, but in every case the variation is greatest at low concentrations.
Kohlrausch, who collected most of the early conductance data, tried
various empirical equations to fit this concentration dependence.
Finally in 1900 (4), he established that the limiting behavior for strong
(i.e., completely ionized) electrolytes is
A= N - s iC (3.3)
where A is the molar conductivity extrapolated to infinite dilution and
0
the slope s depends on the electrolyte and solvent.
I In the older literature K has units of Q-Icm- l (or the whimsical units, mho cml),
and A, with units of Q- 1cm 2mol- 1 , is called the equivalent conductivity and defined
as IOOOKlzC, where C is the molar concentration and z = n+z+ =n_1 z_1 is the number
of equivalents per mole. The new definition is rather more straightforward, but of
course leads to different numerical values; thus beware when using older
tabulations of data.
Paul Erman (1764-1851) was Professor of Physics at the University of

Berlin. His studies on electrical phenomena, although limited by the
crude instruments available at the time, foreshadowed much of the
nineteenth-century development of the physics of electricity and
magnetism. Friedrich Wilhelm Kohlrausch (1840-1910) was Professor of
Physics at Gottingen, Zurich, Darmstadt, Wiirzburg, Strasbourg, and
Berlin. He developed the a.c. methods for measurement of solution
conductance and spent his career on experimental and theoretical studies
of conductance.
Ionic Conductivities
If, at infinite dilution, the behavior of electrolytes is indeed ideal (i.e.,
we can neglect interionic interactions), and if electrolytes are completely
ionized, then the molar conductivity of an electrolyte should be the sum
of the molar conductivities of the individual ions. 1 If one mole of the salt
produces v+ moles of cations with molar conductivity A+ °and v_ moles of
anions with molar conductivity fLo, then
AO=v+~o+v~o (3.4)
The molar conductivity of an individual ion cannot be measured directly,
but if eq (3.4) is correct, then subtracting the molar conductivities of NaCI
0.05
6 NaCI
0.04
.....
,.:....
0
S
C'I
0.03
S
~ 0.02
Figure 3.5 Molar conductivi-

ties of several salts at 298 K 0.01
v s. the square root of
concentration. The dashed
lines represent the linear 0.00
extrapolation to infinite
dilution using eq (3.3). Data 0 10 ~ 30 40 ro
from (89). (e/mol m-3 )1/2
1 This idea, known as Kohlrausch's law of independent ionic migration, was first
proposed in 1876.
and KCI, for example, should give the difference between the molar
conductivities of the sodium and potassium ions.
AO(KCl) - N(NaCl) = (149.8 - 125.4) x 10-4 S m 2mol- 1
N(K+) - N(Na+) = 23.2 x 10-4 S m 2 mol- 1
Comparing KI and NaI, the difference is 23.8 x 10- 4 , and for KBr03 and
NaBr03, the difference is 23.3 x 10-4 . Although there is a little scatter, the
agreement is quite good.
Transference Numbers
When current passes through an electrolyte solution, it is carried in

part by cations moving toward the cathode and in part by anions moving
toward the anode. The fractions of the total current carried by positive
and negative ions are called the transference numbers
t+ = v+A-tlA (3.5a)
L =vJ\..JA (3.5b)
where, since
v+~+v~=A
the transference numbers must sum to 1:

t+ + L = 1
There are several experimental methods for the determination of
transference numbers (5,B9), the most straightforward being the
classical method introduced by Hittorf in 1853 (6). In Hittorfs
experiment, current is passed through a cell such as that shown in
Figure 3.6. In the two end compartments, electrodes are placed which
show uncomplicated behavior in the experimental solutions. For
example, if we wish to measure transference numbers in a solution of
silver nitrate, we would use silver electrodes. At the cathode, silver ions
are plated out when current flows; at the anode, silver ions go into
solution. Now suppose that one Faraday of charge is passed through the
cell. The current in the bulk solution is carried by Ag+ and N03- ions
with fractions t+ and L, respectively. Thus t+ moles of Ag+ must have
gone from the anode compartment to the central compartment, and the
same number from the central compartment to the cathode
compartment. Meanwhile L moles of N03- must migrate from the
cathode compartment to the central compartment and from the central
compartment to the anode compartment. Thus the anode compartment
loses t+ moles of Ag+ by transference but gains one mole of Ag+ by
dissolution of the anode, for a net increase of 1 - t+ moles. There is of
course a corresponding increase of L = 1 - t+ moles of N03-. The cathode
compartment has corresponding decreases in both Ag+ and N03-.
Johann W. Hittorf (1824-1914) was a professor at the University of

Munster. He is best known for his measurements of transference
numbers; his work played an important part in the formulation of
electrolyte theory by Arrhenius in the 1880's.
Figure 3.6 Hittorf cell

for the determination of
transference numbers.
If the solutions in each of the three compartments are analyzed

after passing a known charge through an initially homogeneous
solution, the number of moles of Ag+ gained in the anode compartment
and the number of moles of Ag+ lost in the cathode compartment can be
calculated. Dividing by the number of Faradays of charge passed gives
the transference numbers. The cell must be designed to minimize
convective mixing and the experiment should be of relatively short
duration to minimize diffusive mixing. Analysis of the central
compartment solution should show no change in the concentrations and
thus provides a check on the influence of convection and diffusion.
A moving boundary method, analogous to that used by Tiselius to
measure the electrophoretic mobilities of proteins (§2.4), has also been
used to measure transference numbers.
With transference numbers available, eqs (3.5) can be used to
separate molar conductivities into the contributions of the positive and
negative ions. If transference numbers are known over a range of
concentration, they can be extrapolated to infinite dilution to obtain the
infinite dilution ionic conductivities, A+ 0 and fLo.
A selection of molar ionic conductivities at infinite dilution is given

in Table A.B. These data can be used to compute the molar conductivities
of a wide variety of electrolytes at infinite dilution.
Example 3.3 Compute the molar conductivities of potassium

chloride, hydrochloric acid, and barium chloride solutions at
infinite dilution.
AO(KCl) = AO(K+) + N(CI-)
N = (73.5 + 76.3) x 10-4 = 149.8 x 10-4 S m 2mol- 1
N(HCl) = A°(H+) + AO(Cl-)
N =(349.B + 76.3) x 10-4 =426.1 x 10-4 S m 2mol-1
N(BaCI2) = AO(Ba 2+) + 2 N(CI-)
N = (127 + 2 x 76) x 10-4 = 279 x 10-4 S m 2mol- 1
With a table of molar ionic conductivities (B2,B9,H2,H3) such as

Table A.B, we can also compute transference numbers.
Example 3.4 Compute the infinite dilution transference

numbers in solutions of KCI, HCI, and BaCI2.
For KCI,
t+O = N(K+)IN(KCl)=73.51149.8 =0.491
LO =1-t+0 =0.509
for HCI,
t+O = AO(H+)IN(HCl) = 349.81426.1 = 0.821
LO =1-t+0 =0.179
and for BaCI2,
t+o = N(Ba2+)/ACBaCI2) = 127.7/280.3 = 0.456
LO =1-t+0 =0.554
Ionic Mobilities
It is sometimes useful to think of the transport of an ion in terms of
its mobility. (We encountered the electric mobility in our discussion of
electrophoresis in §2.4.) The mobility of an ion is defined as the velocity

per unit electric field strength,
(3.6)
To see how mobility is related to molar ionic conductivity, we start with

Ohm's law, written in the form i = 6.fl>/R. If the potential difference
across a conductance cell of length L is 6.fl>, the field strength is E = -
6.fl>/L. With eq (3.1) rearranged to R = L / AI(, we have
i=-lCAE (3.7)
Now consider 1 cubic meter of solution containing Ci moles of ions
carrying charge ziF coulombs mol- 1 and moving with an average
velocity Vi. The contribution to the current density across the end of the
cubic meter is the product of these three factors:
(i / A)i = CiZiFvi. (3.8)
Combining eqs (3.7) and (3.8), we have
Vi=-lCiE
--
CiZi F
or, with eq (3.6),
Ui=-~
CilzilF
Finally, with the definition of molar conductivity, eq (3.2), we have
Ui =-~ (3.9)
IZil F
Thus ionic mobilities at infinite dilution can be obtained from molar
ionic conductivities such as those given in Table A.8.
Example 3.5 Compute the mobilities of the sodium and

barium ions at infinite dilution.
The ionic mobility of the sodium ion (z = 1) is

uO(Na+) = AO(Na+)lF = 5.2 x 10-8 m2V- 1s- 1
The ionic mobility of the barium ion (z = 2) is
uO(Ba2+) = J\O(Ba2+)/2F = 6.6 x 10-8 m2V- 1s- 1
Frictional Coefficients
When an ion moves through solution, it is subject to a viscous drag
force proportional to the ion's velocity,
Fi =-fiVi
where the proportionality constant fi is called a frictional coefficient.
Under steady-state conditions, the ion moves at constant velocity so that
the viscous drag force exactly cancels the electrical driving force, zieE,
and the net force is zero:
or
(3.10)
or, using eq (3.9),

Ai =z?Fe (3.11)
fi
Thus the motion of an ion through solution can be described equivalently
in terms of molar ionic conductivities, ionic mobilities, or frictional
coefficients.
Stokes' Law Radii

Still another way of thinking about the rate at which ions move
through solution is by imagining the ion to be a hard sphere of effective
hydrodynamic radius ri and frictional coefficient given by Stokes' law:
fi = 61t1l r i (3.12)
Substituting eq (3.12) into eq (3.10), we solve for the Stokes' law radius ri:
r; = ---,-Iz--,-i!-IF_ (3.13)
61tl1 N A U i
where we have replaced e by F / N A.
Example 3.6 Compute the Stokes' law radii ofNa+ and Ba2 +.
The viscosity of water at 25°C is 0.890 x 10-3 kg m-Is- I ; inserting
the mobilities from Example 3.5, eq (3.13) gives
rCNa+) = 1.84 x 10- 10 m (184 pm)
r(Ba2+) = 2.89 x 10-10 m (289 pm)
§3.1 Conductivity 121.
Faster moving ions, of course, have smaller Stokes' law radii, and
slower ones have larger radii. While Stokes' law is not really valid for
small ions in ordinary solvents,! Stokes' law radii do give a rough
measure of the effective size of ionic species as they move through a
solution. Some Stokes' law radii, calculated as in Example 3.6, are
compared with ionic radii from crystal structure data in Table 3.I.
Several insights can be obtained from examination of these data. In the
first place, it is hard to imagine a hydrodynamic radius being smaller
than the crystal radius of the same ion. Comparison of the radii ofI-, for
example, suggests that the Stokes' law radii are probably systematically
underestimated, perhaps by about a factor of 2. However, within a
related series of ions, e.g., the alkali metal cations, the trends are
interesting. The crystal radii increase, as expected, in the series Li+ <
Na+ < K+, but the Stokes' law radii go in the opposite direction. This
must mean that the small, highly polarizing Li+ ion tightly binds a lot of
solvent molecules which must move with the ion as a unit. The same
effect is seen in comparison of the alkaline earth cations Mg2+, Ca2+, and
Ba2+.
Table 3.1 Stokes' Law and Crystal Radii of Some Ions a
Ion rs/pm rclpm Ion rs/pm rc/pm

Li+ 238 00 OH- 47 119
Na+ 184 116 F- 168 119
K+ 125 152 CI- 121 167
Mg2+ 348 86 Br- 118 182
Ca2+ 310 114 1- 120 206
Ba2+ 289 149 CI04" 136 226
A13+ 439 68 8042- 231 244
a Crystal radii from Shannon (7).
The three halide ions, CI-, Br-, and I- all have about the same Stokes'
law radii despite an increase in the crystal radii; this again is consistent
with larger, less polarizing ions binding fewer solvent molecules or
binding them less tightly so that fewer move as a unit. Comparing the
1 The law was derived in 1845 by Sir George Stokes assuming a rigid sphere moving
through a continuous medium; using Stokes' law for a charged deformable ion
moving through a solvent of discrete dipolar molecules of size comparable to the ion
must introduce some serious errors.
isoelectronic ions Na+, Mg2+, and A13+ we see an even more dramatic
effect which has the same qualitative explanation. Extending the
comparison to F-, which is isoelectronic with Na+ and has about the
same crystal radius, we see that the solvation sheath is apparently
greater for Na+ than for F-, consistent with the idea that cations are
rather more specifically solvated (formation of coordinate covalent
bonds) than are anions (orientation of water dipoles and/or hydrogen
bonding).
From crystal structure data, OH- and F- would be expected to be
nearly the same size, but the Stokes' law radius of OH- is 47 pm, much
less than the radius of any other anion. A similar discrepancy is found
for H+, which has a Stokes' law radius of 26 pm. Structural evidence
suggests that H+(aq) is better represented by H30+, which might be
expected to have a Stokes' law radius comparable to Li+ or Na+, say about
200 pm. The anomalously high conductivities of H+ and OH- in water
can be understood if we recall that these water-related ions are part of
an extensive hydrogen-bonded network involving solvent water
molecules. A small shift in hydrogen bonds thus can move the charge
from one oxygen center to another and several such shifts can move the
charge much faster than any of the nuclei travel. Thus the conductivity
is much higher than might have been expected from the size of the HaO+
or OH- ions.
Theoretical Treatment of Conductivity

Kohlrausch's discovery that molar conductivity varies as the square
root of concentration, eq (3.3), attracted the attention of several
theoreticians,l most notably Onsager, who derived a conductance
limiting law starting from the Debye-Huckel treatment of the ion
atmosphere (8). Onsager's result can be written as
A=N- Z lZ21F 2N I q (I Z ll+l z 2I)F 2 (3.14)
121t£EoXARTNA 1 + Vq 61tllNAXA
where XA is the ion atmosphere thickness, 11 is the solvent viscosity, Zl
and Z2 are the positive and negative ion charges, and
q= ZlZ2
(Zl-Z2)(Z2 t l-Zl t 2)
where t1 and t2 are the transference numbers. When Zl = -Z2 (a 1:1 or 2:2
electrolyte), q = 1/2 and is independent of the transference numbers. The
other parameters in eq (3.14) have their usual significance. Both
correction terms in eq (3.14) are proportional to lIxA, which in turn is
proportional to the square root of the ionic strength, eq (2.8), so that eq
(3.14) matchs the fC dependence found by Kohlrausch.
1 See Robinson and Stokes (B9) for a review of the various approaches.
Lars Onsager (1903-1976) taught at Brown University in the early 1930's,

later moving to Yale. He is best known for his work in non equilibrium
thermodynamics for which he received the Nobel Prize in 1968, but he
made important contributions in many other areas of theoretical
chemistry.
The first term in the Onsager limiting law arises from the so-called
ion atmosphere relaxation effect. When an ion is attracted by an electric
field, it is also subject to an opposite force exerted by its ion atmosphere,
which tends to restrain the ion and thus lowers its contribution to the
solution conductance. The effect increases with the density of ions in the
atmosphere, i.e., inversely proportion to XA. The second correction term
results from the electrophoretic effect. When the ion in question moves
through the solution, it tends to take its ion atmosphere with it, resulting
in a viscous drag force opposing the motion, thus the dependence on
viscosity.
Like Debye-Hiickel theory from which it is derived, eq (3.14) is a
limiting law valid only at very low concentrations, usually less than
O.OOlM.
Example 3.7 Compute the limiting law slope for a plot of the
molar conductivity ofKCl (AO = 149.8 x 10- 4 S m 2 mol- 1 ) us. -rc
assuming the dielectric constant and viscosity of pure water (10
=78.54, 'l1 =0.890 x 10-3 kg m- I s- l ), and compare with the slope of
the KClline in Figure 3.5.
Inserting the values of the physical constants and T = 298 K

into eq (3.14), we have
A = AD - (6.97 x 10- 11 )ADlxA - (1.84 x 10-12)!XA
Substitution of constants in eq (2.8) gives
-rc
XXI = (1.04 x 108 )
where C is in mol m- 3 . Substituting XA and AO, we have
)-rc
A = AD - (3.0 x 1O-4
which is in satisfactory agreement with the experimental
result:
)-rc
A = AD - (2.7 x 10-4
Effects of Frequency and Electric Field Strength

According to Onsager's description of ionic motion in an electrolyte
solution, the motion of an ion is restrained by the attraction of its
atmosphere. An ion atmosphere is not formed instantaneously since it
requires the arrangement of many ions in or near a minimum potential
energy configuration. The time required to configure an ion atmosphere
is on the order of 10-7/C seconds. Thus in a 1 mM solution, ionic motion
is subject to an ion atmosphere relaxation time on the order of 10- 7 s. If
the a.c. potential applied across the conductance cell is in the radio-
frequency range, above 10 MHz say, ions move faster than their
atmospheres can be rearranged. Under these conditions, ionic motion
escapes the retarding effect of the ion atmosphere and the measured
conductivity is increased. This effect was predicted by Debye and
Falkenhagen (9) and first observed by Sack (10) and is known as the
Debye-Falkenhagen effect or the Sack effect.
A related effect is observed when the a.c. potential across the
conductance cell is very large and the electrode spacing is small. With
an oscillating electric field on the order of 10 6 V m- I , an ion is
accelerated to a sufficiently high velocity that it effectively escapes its ion
atmosphere. That is, the electrical force on the ion greatly exceeds the
retarding force of the ion atmosphere. The ion may leave behind some of
the coordinating solvent molecules as well, so that its effective Stokes'
law radius is smaller. For both these reasons, the conductivity increases
for large electric field strengths. This effect was first observed by Wien
(11) in 1922 and is called the Wien effect.
Both the Debye-Falkenhagen and the Wien effects can be understood
quantitatively by extensions of On sager's theory of ionic motion (B5).
Ultrasonic Vibration Potential

When a high-frequency sound wave travels through an electrolyte
solution, solvent molecules are forced into an oscillatory motion. Ions in
the solution tend to move with the solvent but may lag a little behind. If
the lag in cation motion is different from the lag in anion motion
(because of different frictional coefficients), a small potential difference
will develop at the frequency of the ultrasonic wave. If a pair of
electrodes is placed in the solution with a spacing of half the wavelength
of the sound wave, this potential can be detected. This effect was
predicted by Debye in 1933 (12) but was not observed experimentally until
after World War II. Debye originally suggested this experiment as a
way of estimating the effective masses of the solvated ions. Subsequent
experimental work showed that it is really the ionic partial molar
volumes which are determinable, and there has been a substantial
volume of work directed to this end. Work on ultrasonic vibration
potentials has been reviewed recently by Zana and Yeager (13).
§3.2 Conductance Applications 125
3.2 CONDUCTANCE APPLICATIONS
The measurement and interpretation of conductance has given us

many important qualitative insights into the properties of electrolyte
solutions and has provided a means by which quantitative theories of
electrolytes can be tested. Many of the experimental methods are now
primarily of historical interest, at least for aqueous solutions, but some
applications remain useful in our arsenal of techniques for extracting
information from nature.
Measurement of Dissociation Constants

In the development of our understanding of electrolyte solutions in
the late nineteenth and early twentieth centuries, conductance
measurements provided much of the experimental evidence for ionized
salts in solution. Conductance measurements played a particularly
important role in classifying acids, bases, and salts into strong
electrolytes (i.e., completely ionized salts such as NaCI or strong acids
such as HCI) and weak electrolytes (i.e., weak acids such as acetic acid).
Thus strong electrolytes exhibit the linear dependence of A on fE which
we have seen in §3.1. Weak electrolytes, on the other hand, typically have
very small molar conductivities at moderate concentrations which
increase rapidly as the concentration is reduced.
In explaining the dependence of A on concentration, Arrhenius
assumed that, for a partially dissociated electrolyte at low concentration,
the fraction of the electrolyte which is dissociated is simply the ratio of
the molar conductivity to the molar conductivity extrapolated to infinite
dilution:
a=NAO (3.15)
Consider then the dissociation of acetic acid,
HOAc + H20 ~ H30+ + OAc·
with the corresponding equilibrium constant,
Ka = [H;P+] [OAc·]
[HOAc]
The individual species concentrations can be written in terms of the
degree of dissociation, a, and the total acetic acid concentration C:
[H30+] =[OAc·] = aC
[HOAc] = (1 - a)C
Substituting these into the equilibrium constant expression, we have

Ka = a2C
I-a
where C must have units of mol VI if Ka is to refer to the usual standard
states. Using eq (3.15) to eliminate a, we have
Ka JAlN)2C (3.16)
1-AlAo
This expression, which is known as Ostwald's dilution law, may be
rearranged to
CA = Ka(N)2 (l/A)-Ka N
so that a plot of CA vs. 1/A should give a straight line with the slope and
intercept yielding both the infinite dilution molar conductivity and the
equilibrium constant.
Svante A. Arrhenius (1859-1927) was a lecturer at the Technical High

School of Stockholm and later director of the physical chemistry
department of the Nobel Institute. Arrhenius is generally regarded as
one of the founders of the discipline of physical chemistry and is
remembered for his work on electrolyte solutions and on chemical
kinetics.
This method works well and many of the earlier acid-base

equilibrium constants were obtained in this way. Easier ways of
obtaining such data are now available, but it is sometimes useful to
remember that the conductance method exists.
Example 3.8 In 1894, Kohlrausch and Heydweiller (14)

measured the conductivity of extremely pure water, obtaining I(
= 6.2 x 10- 6 S m- 1 . Estimate the self-dissociation constant of
water from this measurement.
The sum of the molar ionic conductivities ofH+(aq) and OH-(aq)

at infinite dilution is
AO = N(H+) + N(OH-) = 0.0549 S m 2mol- 1
The ionic concentrations should be sufficiently low that we can
substitute this value, together with 1(, in eq (3.2), and solve for C.
C= KIA
C = (6.2 x 10-6 S m- 1)/(0.0549 S m-2 mol- 1)
C = 1.13 x 10-4 mol m-3
§3.2 Conductance Applications 127
[H+] =[ORo] =1.13 x 10-7 mol L-1
Keq = [R+][OH-] '" 1.3 x 10- 14
Determination of Electrolyte Charge Type
When new compounds are discovered and found to be electrolytes,

determination of the charges carried by the positive and negative ions
may be an important part of the characterization. If the molar
conductivity is determined at various concentrations, and a plot of A us.
fC constructed, the intercept, A0 , gives a hint of the charge type since the
conductivities of salts of a given type tend to fall within certain limits.
The limiting slope gives a better indication, however, since it depends
upon the charge type of the electrolyte; see eq (3.14).
Conductometric Titrations
One of the most generally applicable electroanalytical techniques

consists of following a titration reaction conductometrically (15).
Consider, for example, the titration of hydrochloric acid with sodium
hydroxide. Initially, the conductance of the HCI solution is due to H30+
and CI- ions. As the titrant is added, the H30+ ions are replaced by Na+,
and, since the molar ionic conductivity of Na+ is much less than that of
H30+ (see Table A.8), the conductance of the solution falls. Beyond the
equivalence point, on the other hand, excess NaOH is being added, and
Solution
------- H+
,, CI-
,, Na+
,, OH-
,,
,,
,,
,, ,
" ",'
'\, ;",.;
Figure 3.7 Conductometric

titration of HCI with NaOH;
......................
"
...... -- ................ -..".... --................;l'::.--..:-. . . .:::::
...... ......... -;;;
'\ ","
lines show the contributions
of H+ CI· Na+ and OH- to o 5 10 15 ~
the soiutio~ conductance. Volume NaOH/mL

since the OR- molar ionic conductivity is very large, the conductance of
the solution rises again. The conductance of the solution is due, of
course, to all the ions present. This sum is demonstrated graphically in
Figure 3.7. In practice, a conductometric titration is carried out by
adding aliquots of titrant, measuring the conductance and plotting the
measured conductance vs. the volume of titrant added. The points can
usually be fitted to two straight-line segments which intersect at the
equivalence point.
This technique can be extended to any titration reaction which leads
to a change in the slope of the conductance vs. concentration plot at the
equivalence point. Any ionic reaction which produces a precipitate, a
gas, or any other nonelectrolyte is well suited for use in a
conductometric titration. Although the technique is very general, there
are practical limitations: (1) the change in slope at the equivalence point
is often small; (2) the analyte and titrant concentrations usually must be
greater than 0.001 M; and (3) nonparticipating electrolytes should be
absent.
3.3 DIFFUSION
Ionic conductivity and electric mobility are measures of the rate of

charge transport through solution under the influence of an electric
potential gradient. Electrical conductivity is one of a class of physical
properties called transport properties. Other transport properties
include thermal conductivity (transport of heat energy under the
influence of a temperature gradient), viscosity (transport of momentum
under the influence of a velocity gradient), and diffusion (transport of
mass under the influence of a chemical potential or concentration
gradient). Because an understanding of diffusion will be particularly
important in Chapters 4 and 5 and because electric mobility and
diffusion can be related easily, we will digress for a few pages to a
discussion of diffusion.
Fick's Laws of Diffusion
Consider the motion of molecules or ions in a liquid solution. In the

absence of electric or gravitational fields, the motion is simply a random
thermal agitation where the particles respond to bumps they receive
from their neighbors to go shooting off in one direction or another.
Although classical mechanics would have us believe that the Newtonian
equations of motion at least in principle govern this motion, in practice it
is just as well to regard the motion as entirely random and
unpredictable. If we focus on an individual molecule, we will see it
§3.3 Diffusion 129
undergo a series of displacements in random directions and of random
length. However, we are interested in the average behavior of a large
number of molecules. Suppose that, in a time interval ot, a p~rt~cle
moves a distance Ox in the x-direction. The average displacement, \Ox}, is
zero because a molecule is as likely to go in the positive x-direction as in
the negative. On the other hand, if we square the disp)ace~ents before
averaging, the mean square displacement (in time ot) \(ox)2J is nonzero.
Now consider a section of the solution where there is a linear
concentration gradient in the x-direction, and focus on a small section of
cross-sectional area A and length (along the x-axis) 20x. Divide this
volume in half as shown in Figure 3.8, and suppose that the average
concentration of particles in the left-hand volume is N 1 and in the right-
hand volume N2 (particles per unit volume). Thus, since the volume of
each element is Aox, the number of particles in the two volumes are
NlAox and N~Ox, respectively. Now in time interval ot, on the average
half the particles will move to the right and half will move to the left.
Thus the number crossing the central dividing plane from left to right
will be %NlAox and the number crossing from right to left %N~Ox.
Thus the net number moving across the barrier (from left to right) is
1/2 (Nl - N2)A8x. Dividing by the time interval then gives the net rate of
transfer:
Rate = (N 1 -N2)A Ox (3.17)
2&
Recalling our original assumption that the concentration gradient is
linear, we can express the concentration difference, N 1 - N 2, in terms of
the concentration gradient
N 1 -N2=--Ox
aN
ax
Since N is apparently a function of both x and t, we have written the
gradient as a partial derivative. Substituting this expression into eq
(3.17), we have
Figure 3.8 Model for the

mathematical description
of diffusion. x
2
Rate = _ A(Bx) aN (3.18)
2 & ax
The rate in eq (3.18) is the number of particles passing through a
reference plane of cross-sectional area A per unit time. If we divide eq
(3.18) by A and by Avogadro's number, we will have the flux J in moles
per unit cross-sectional area per unit time
J=_D ac (3.19)
ax
where the diffusion coefficient D is given by
(Bx)2
D=- (3.20)
2&
with 81 units of m 2 s- 1 . This relationship was first established
empirically by Fick in 1855, and is known as Fick's first law of diffusion.
Notice that eq (3.19) tells us that the flux is positive (a net flow of
particles from left to right) only if the concentration gradient is negative
(there are fewer particles on the right to begin with). Thus this simple
equation contains the elements of the phenomenon of diffusion: flow of
particles into regions of lower concentration.
Adolf Eugen Fick (1829-1901) was Professor of Physiology at the

University of Wiirzburg. Fick's work on diffusion was part of a wide-
ranging effort to apply the principles of physics to physiology.
Example 3.9 An integral form of eq (3.20) is sometimes used

to estimate the distance a particle might diffuse in a certain
time:
(x~=2Dt
where (x~ is the mean-square displacement in the x-direction
in time t. Estimate the distance a particle with D = 10- 9 m 2 s- 1
will travel in 10-6 s.
(x~ =2(10-9 m 2s- 1)(1O-6 s) =2 x 10-15 m 2
Thus the root-mean-square displacement in the x-direction in 1
Ils is 4.5 x 10-8 m (45 nm). In an isotropic solution, we would
expect (x~ = (y~ ={z~ so that (r~ =3(x~ or
(x~ = 6Dt
Thus the total r.m.s. distance travelled in IllS is 135 nm.
§3.3 Diffusion 131
Fick also discovered a second empirical law, which was later found
to be derivable from the first law. Consider an infinitesimal volume
element of unit cross-sectional area and thickness dx. The flux into this
volume element is J(x), and the flux out of the volume element is J(x +
dx) = J(x) + (aJ/ax)dx. The difference between the flux in and the flux out
is the net increase in number of moles of particles in the volume element
per unit time. Dividing this by the volume of the element dx, we have the
time rate of change of the concentration ac/at.
ac = J(x) - J(x + dx) aJ
=--
at dx ax
Differentiating eq (3.19), we get
aJ =_Dic
ax ax 2
so we are left with
ac =_Dic (3.21)
at ax 2
which is Fick's second law. This result, of course, is a second-order
partial differential equation, which might not seem to be a particularly
simple way of expressing what we know about diffusion. However, it
turns out that the equation can be solved relatively easily for a wide
variety of boundary conditions.
Before we go on, it is instructive to write down the solution to eq
(3.21) for a particularly interesting case. Suppose that we bring two
solutions into contact at time t = O. In one solution the concentration of
the diffusing species is Co and in the other solution, the concentration is
zero; the initial interface is at x = O. We suppose that for t > 0, C ~ 0 as x
~ 00 and C ~ Co as x ~ - 0 0 . With these initial and boundary conditions,
the solution to eq (3.21) is (see Appendix 4)
C(x,t) = tco [1- erf{2illt}] (3.22)
The function erf(\jI) is called the error function and defined by the
integral
(3.23)
The error function is a transcendental function like a logarithm or

exponential function and is found in many mathematical tables. As
seen in Figure 3.9, the error function goes to zero as its argument goes to
zero and approaches unity as its argument increases. The error
function is an odd function of the argument, i.e., erf(-",) = -erf(",). The
derivative of an error function is an exponential function:
~ erf[",(x)] = ~ exp(-",2) d", (3.24)
dx fit dx
1.0
erfC",)
0.8
0.6
0.4
0.2
0.0
Figure 3.9 The error 0.0 0.5 1.0
function. 1.5 '" 2.0
The concentration profile, given by eq (3.22), and the concentration

gradient, computed using eq (3.24), are shown for various times in
Figure 3.10. The results are as we would expect intuitively: the
concentration gradient, which initially is a sharp spike at x = 0, broadens
and decreases in height as time passes, i.e., as the solute diffuses into
the right-hand solution. Since the flux of solute is proportional to the
concentration gradient through eq (3.19), Figure 3.10c is equivalent to a
plot of flux us. distance at successive times.
Example 3.10 Assuming D = 10- 9 m 2 s- 1 and Co = 0.1 M (100

mol m- 3 ), compute the concentration gradient and flux at the
position of the original boundary 1, 10, and 100 s after the
boundary is formed.
Differentiating eq (3.22), using eq (3.24), we obtain

aC(x,t) =~exp(-~)
ax 2"/ rtDt 4Dt
At x = 0, the exponential term reduces to 1. Substituting
numbers, we obtain
§3.3 Diffusion 133
- acco,t) = 8.92 x 105, 2.82 x 105, 8.92 x 105 mol m 4

ax
at t = 1, 10, and 100 s. Substituting the concentration gradient
into eq (3.19), we obtain the flux: J = 892, 282, 89.2 !lmol m-2s- 1
CO~~ __ ____r -__________

~
(a) C=C o
(b)
-x 0 +x 0------.1.---=----
-x o +x
(c)
Figure 3.10 (a) Initial distrib-
ution of solute in a diffusion
experiment. (b) Concentration
profiles at successive times after
beginning of diffusion. (c) Con-
centration gradient at successive
times after beginning of
diffusion. -x o +x
Relation of Diffusion and Mobility

We have seen that diffusion may be regarded as the net movement
of particles in solution due to a concentration gradient. Another more
general way of stating this is to say that particles will move in solution
with a velocity proportional to the chemical potential gradient. This may
be expressed mathematically as
1 a!li
Vi =----- (3.25)
fiNA ax
where Vi is the average velocity of particles of type i, fi is the frictional
coefficient, and NA is Avogadro's number. Suppose that the particles
are ions so that the chemical potential depends not only on their activity,
but also on the electric potential. Thus we have l
Ili = Ilio + RT In ai + ZiF(J) (3.26)
Differentiating eq (3.26) and substituting into (3.25), we have the Nernst-
Planck equation
v. =-~[kT d In ai + z.e~] (3.27)
'Ii dx 'dx
We now consider two limiting cases. First we assume that the
activity is uniform, but the electric potential nonuniform. Equation (3.27)
then becomes
Vi = + Zi e E
Ii
where we have recalled that the negative gradient of the electric
potential is the electric field strength, E = -d(J)/dx. Dividing through by E,
we have the mobility:
(3.28)
which is identical to eq (3.10).

For the second limiting case, we assume that the electric potential
is constant so that eq (3.27) contains only the first term on the right-hand
side. If the concentration of species i is Ci and the average velocity is Vi,
then the flux Ji is CiVi or
J,. -_ Ci kT d In ai
,- Ii dx
Noting that the activity can be written ai = Cm, we have
J,. =_kT[C' dIn Ci C. dIn Yi] (3.29)
'Ii'dx+'dx
and, if we assume that the activity coefficient is independent of x, either
because the solution is dilute and nearly ideal or because of the presence
of a homogeneous excess of some other electrolyte which mostly
determines the activity coefficient, then the second term may be dropped
and eq (3.29) becomes
which is Fick's first law, with

(3.30)
1 This generalized chemical potential is sometimes called the electrochemical

potential.
§3.3 Diffusion 135
Since the frictional coefficient occurs in both eqs (3.28) and (3.30), we can
eliminate it to obtain a relation between the diffusion coefficient and the
electric mobility,
Di = kT Ui = RT Ui (3.31)
IZi Ie IZi IF
Equation (3.31) is sometimes called the Einstein relation. Combining eqs
(3.9) and (3.31) gives the connection between the diffusion coefficient and
the ionic conductivity, the Nernst-Einstein equation,
Di = ~ 1\ (3.32)
z?e 2
Example 3.11 Compute the diffusion coefficients of Na+ and
Ba2+ at 25°C in dilute solutions.
At 25°C, the factor RTIF has the value 0.0257 V. Thus, using
the mobilities computed in Example 3.6, we have
DOCNa+) = (0.0257 V)(5.2 x 10-8 m2V- 1s- 1)
DO(Na+) = 1.34 x 10-9 m 2 s- 1
DO(Ba2+) = (0.0257 V)(6.6 x 10-8 m2V- 1s- 1)/2
DO(Ba2+) = 0.85 x 10-9 m 2s- 1
We now have three essentially equivalent measures of the transport

properties of an ion in solution: the molar ionic conductivity A, the ionic
mobility u, and the diffusion coefficient D. The three quantities are
compared for Na+, Mg2+, and AI3+ in Table 3.2. Notice that whereas AO
increases with ionic charge, UO is nearly independent of charge and DO
decreases with charge. This is a reflection of two effects; as the charge
increases, the size of the solvation shell increases; each of these ions has
a primary coordination of six water molecules, but with increasing
charge, outer-sphere water molecules are also oriented about the ion.
Thus a large aggregate such as Al3+ with its solvation sheath moves
through solution more slowly than Na+ and its smaller solvation sheath.
The dependence of DO on ionic charge is thus qualitatively reasonable.
When we look at the velocity of particles in a field, we expect the
retarding effect of the solvation sheath to be opposed by a larger electrical
force on ions with higher charge. Thus the two effects more or less
cancel in the case of the mobilities. Conductivities, on the other hand,
measure not only the velocity of ions in the field, but also the amount of
charge carried and so increase with increasing charge.
Table 3.2 Comparison of Ionic Conductivities, Mobilities,

and Diffusion Coefficients.
Ion AO/S m 2mol- 1 uO/m2V- 1s-1 DO/m2s- 1

Na+ 50 x 10-4 5.2 x 10-8 13.4 x 10-10
Mg2+ 106 x 10-4 5.5 x 10-8 7.1 x 10-10
AI3+ 189 x 10-4 6.5 x 10-8 5.6 x 10-10
3.4 MEMBRANE AND LIQUID JUNCTION POTENTIALS
As we noted in Chapter 1, in general there is a potential difference

across the boundary between any two dissimilar phases. In this section,
we will consider three cases of potentials arising across phase
boundaries.
Liquid Junction Potentials

Consider the junction of two electrolyte solutions. The two solutions
may be carefully layered or they may be separated by a fritted glass disk
or an asbestos fiber. The system as a whole is not at equilibrium and
mixing will eventually take place. The role of the glass frit or asbestos
fiber is simply to slow the rate of diffusive mixing. If the rate of mixing
is slow enough, we can regard the two solutions as essentially constant
in composition and can deal with a steady-state model of the boundary
between the two phases.
N ow consider what happens when the liquid junction is formed
between phases a and p and diffusion is allowed to proceed. Suppose for
the moment that a single electrolyte is present with concentrations Ca )
C~. Since the diffusion coefficients of the anions and cations in general
are not equal, one species is expected to diffuse more rapidly than the
other. If, for example, the cations diffuse from the a-phase into the p-
phase more rapidly than do the anions, then a charge separation results
and with it an electric potential difference which ultimately will oppose
the continued diffusion of the two species at unequal rates. The steady-
state condition then is diffusion of the two species at the same rate, but
with the positive ions slightly ahead of the negative ions; see Figure 3.11.
Consider now the more general case of a liquid junction between
two electrolyte solutions of different compositions and concentrations.
Suppose that the liquid junction has reached a steady-state condition,
and consider a charge 8Q which moves reversibly through a distance dx
in the phase boundary region. There will be contributions to 8Q from
§3.4 Membrane and Liquid Junction Potentials 137
cations moving in the +dx direction and from anions moving in the -dx
direction:
where Zi is the charge on ion i and oni is the number of moles of i

moving from x to x + dx. If i is a cation, Zi and oni will be positive, but if i
is an anion, both quantities will be negative. The fraction of the charge
carried by ion i (the transference number, ti ) then will be
. - FZi oni
t,-
oQ
c t=l t=3 t=5

.
- .. :, .......................... - ... '\".. ; ......................... ..
0 _ _ _, . __________ _
,, ,,
,
o ___________ L - -_ _
d
I
+
\:.)
--------+
Figure 3.11 Diffusion of a 1:1 electrolyte solution in a tube. (a)
Concentration profiles at several times; the solid lines represent the cations
and the dashed lines represent the anions. (b) Corresponding concentration
differences, C+ - C_.
oni can be written
oni =~oQ
Zi F
In moving from x to x + dx, ion i moves through a chemical potential
difference, given by eq (3.26):
dlli = RT d In ai + FZi d<l>
where d<l> is the electric potential difference across the distance dx. The
total change in free energy then is
oG = I. dlli oni
1
or
8G = [RFT "~ !L d In a'' k+ ."" t·, d<D] 8Q (3.33)

, Zi
,
However, since we assumed that charge transport through the phase
boundary is reversible, 8G must be zero; the bracketed expression must
then be zero, independent of 8Q. Furthermore, by definition
I ti = 1
i
so that we can rearrange the bracketed expression in eq (3.33) to obtain
d<D =_RT" EL d In a'
, Zi
F k." '
We then integrate over the phase boundary to obtain
~<D=_RT"
F k."
i~ ELdlna· Zi '
(3.34)
, CI.
Equation (3.34) is a general expression for the liquid junction potential,

but the integral is generally intractable. To proceed further, we must
build some simplifying assumptions into the model.
When only one electrolyte is present and the two phases differ only
in concentration, then it is reasonable to assume that the transference
numbers are constant. When these are factored out, the integrals of eq
(3.34) are easy and we obtain for a 1:1 electrolyte (ZI = 1, Z2 =-1)
~<D = - R1
-
F
'l afa~ a~]
t1 In _1 - t2 In --..2...
afJ
or if we further assume that af =afJ =aCi. and af =ag = a~,
~<D = RT (t1 - til In a CI. (3.35)
F a~
Thus we predict that when two solutions, differing only in the
concentration of a 1:1 electrolyte, are in contact the liquid junction
potential is proportional to the difference in transference numbers of the
positive and negative ions and to the log of the activity ratio.
Example 3.12 Compute the liquid junctions potentials for the

contact of 0.1 and 0.01 M KCI solutions and for the contact of 0.1
and 0.01 M HCI solutions neglecting activity coefficient
differences and assuming that the transference numbers are
independent of concentration.
Computing the transference numbers from ionic
conductivities (Example 3.5), we have, for KCI, t1 = 0.491 and t2 =
0.509, so that substitution in eq (3.35) gives (RTIF = 25.7 mY)
.1<1> =(25.7 mV)(-O.018) In(O.I) = 1.1 mV
For HCI, t1 = 0.821 and t2 = 0.179, so that
.1<1> = (25.7 mV)(0.642) In(O.I) = -38.0 mV
To deal with the more generally interesting case where mixtures of

several electrolytes are present, we must make an approximation to
allow evaluation of the transference numbers. One approach to this
problem, suggested by Henderson in 1907 (16), is to assume a linear
concentration gradient for each ionic species. Thus if we define a
dimensionless distance parameter x, scaled so that x ranges from 0 in
phase a to 1 in phase ~, the concentration of ion i can be written
Ci = clx + C? (I-x)
We will also assume that the activity coefficients are constant in the
phase boundary region, so that we can write
dIn ai dIn Ci
---=---
(l,X (l,X
Thus we have
d In ai =d In ai dx =l dCi dx
ax Ci ax
ca-c~
dIn ai = ' I dx
Ci
The transference number of ion i can be written in terms of the
concentrations and mobilities:
t. _ IZi ICiUi
, - L IZjlCjuj
j
where we assume that the mobilities are independent of concentration
and thus of x. Substituting the expressions for ti and d In(ai) into eq
(3.34), we obtain
where
a= LIZjlUPl
j
and
b=L IZjlUj (eJ - ell
j
The integral is readily evaluated:
t ~=ln(a
Jo a + bx
+ bx)1 1 =~lnCL.±..b..
bob a
Thus
(3.36)
Equation (3.36), sometimes called the Henderson equation, is a general

(albeit somewhat approximate) result for liquid junction potentials.
When a only a single 1:1 electrolyte is present, eq (3.36) reduces to eq
(3.35). Another relatively simple situation can be set up experimentally.
Suppose that only one cation is present and that it has the same
concentration in both phases (e.g., NaCI in one solution, NaBr in the
other). The prelogarithmic term in eq (3.36) then reduces to RTIF.
Alternatively, if a single common anion is present, the prelogarithmic
term reduced to -RTIF. Thus we have
L uJzil ep 1
~<l> = ± RT In
F
rL u,jz;i er
..":i:,,,-_ __ (3.37)
Example 3.13 Compute the liquid junction potential for the

contact of 0.1 M HCI and 0.1 M KCl solutions.
Equation (3.37) can be used in this case since we have a

common anion; since the concentrations are all 0.1 M, we have
~<l> = _RT In (U(H+) + u(Cn)
F u(K+) + u(Cn
Using mobilities calculated as in Example 3.6 (and assuming
these quantities to be independent of concentration), we have
i1<l> =- (25.7 mY) In (36.3 + 7.9) =-26.8 mV
7.7 + 7.9
Example 3.14 Compute the liquid junction potential for 0.1 M

HCI and 0.1 M KCI solutions contacted through a saturated
KCI (4.2 M) salt bridge.
In this case, we must deal separately with two liquid junctions,

one between the HCI solution and the salt bridge, the other
between the salt bridge and the KCI solution. In the first case,
we must use eq (3.36), but eq (3.35) will handle the second
junction. Starting with the simpler case, we can use the
method of Example 3.12 to obtain for the salt bridge/KCl
junction:
i1<l> = (25.7 mV)(-O.018) In(42) = -1.7 mV
Inserting the mobilities (computed as above from the infinite
dilution conductivities) and concentrations into eq (3.36), we
obtain for the HCl/salt bridge junction:
!\<l> = +4.2 mV
for a total liquid junction potcnti81 of +2.9 mV.
Comparing the results of Examples 3.13 and 3.14, we see that the
use of a saturated KCl salt bridge is expected to reduce the junction
potential for 0.1 M KCl and HCl solutions by nearly a factor of 10. The
secret, of course, is that the K+ and Cl- ions, which have very nearly
equal mobilities, carry most of the current at the salt bridge/solution
junctions. The result is generally applicable, however, and explains in
part the popularity of KCI as an electrolyte, particularly when liquid
junctions are unavoidable.
Donnan Membrane Equilibria
In a system containing a semipermeable membrane, an interphase

potential superficially like a liquid junction potential may develop.
Suppose that two solutions, one containing an ordinary electrolyte with
small ions and the other containing a solution of charged
macromolecules and small counterions, are separated by a membrane
which allows the free passage of solvent molecules and small ions but
which is impermeable to the macromolecules. At equilibrium, the
chemical potentials of the small ions and solvent! will be equal in the
two phases. However, since the condition of electrical neutrality
requires that the ionic concentrations be unequal in the two solutions, an
electric potential difference must develop across the membrane. Unlike
a liquid junction potential, however, the potential in this case does not
depend on the relative mobilities of the ions and the system is in a true
equilibrium state. The development of potentials across semipermeable
membranes was first studied by Donnan in 1911 (17) and the
phenomenon is usually referred to with his name.
Frederick G. Donnan (1870-1956) was Professor of Physical Chemistry at

the University of Liverpool and later at University College, London.
Donnan was a student of Ostwald and played an important role in
introducing the then new ideas of physical chemistry in Britain. He is
best remembered for his work on membrane equilibria.
Consider a 1:1 electrolyte in equilibrium across the membrane;

equilibrium requires that the ionic chemical potentials be equal:
I-IP = lJ.i~
or, since Ilt is the same for the ex and ~ phases,
RT In af = RTln af ± F~<l>
where the upper sign applies to cations, the lower sign to anions and ~<l>
= <l>~ - <l>u. If the cation (i = 1) and anion (i = 2) equations are added, we
get
RT In afa2u = RT In afaf
or
alu a2u = a 1~ a 2~
which may be regarded as the condition of Donnan membrane
equilibrium. If the activity coefficients in the ex- and ~-phases are equal,
we can also write
CfC~ = cfc! (3.38)
Electroneutrality of the ex phase requires:
Cf=C~
If the ~-phase also contains a macromolecule M with concentration CM
and charge ZM, then the electroneutrality condition is
cf +ZMCM = C!
1 The equalization of the solvent chemical potential leads to the phenomenon of
osmotic pressure.
In general there also will be conservation relations governing the total
number of moles of anions and cations. These relations, together with
the Donnan equilibrium expression, eq (3.38), can be used to compute the
individual concentrations. Since the activity of an ion is different in the
two phases, a potential difference develops:
t.<l> = RT In af = _RT In a~ (3.39)
F a~ F a~
1 2
Example 3.15 Compute the Donnan membrane potential for a

system initially containing 100 mL of 0.10 M KCI (phase a) and
100 mL of 10-3 M macromolecules with ZM = 100 with 0.10 M CI-
counterions (phase ~). Electroneutrality requires:
cf=cf Cg+0.1=CP
and conservation of potassium ions requires:
cf+cf=o.l
Substitution in eq (3.38) gives the quadratic equation
C 1P(0.1 + C1P) = (0.1- C 1P)2
the solution to which is
C1P = 0.033 M
Thus
CIa =C2a =0.067 M
C2P = 0.133 M
and
t.<l> = (25.7 mV) In(2) = 17.8 mV
Ion-Selective Membranes
Now that we have semiquantitative theories of the liquid junction

potential and the Donnan membrane potential, we can turn to the
operation of an ion-selective membrane such as the glass membrane of a
H+-sensitive glass electrode which WRS discussed qualitatively in Section
1.5.
We assume that the interior of the membrane is permeable only to
N a + ions, but the inner and outer surfaces of the membrane are
hydrated and can contain other cations, in particular H+. We can divide
the system into five regions which we imagine as separate phases:
a
outer electrolyte solution
~outer surface of the membrane
y interior of the membrane
~' inner surface of the membrane
a' inner electrolyte solution
The total membrane potential then has four contributions:
~<l> = ~<l>a~ + ~<l>~y + ~<l>~' + ~<l>Wa' (3.40)
We will assume that the membrane surfaces are in equilibrium
with the adjacent electrolyte solutions so that the arguments used in
discussing Donnan membrane equilibria can be used for the a-~ and W-
a'interfaces. If H+ is in equilibrium across the a-~ interface, then the
H + chemical potentials must be equal for the two phases:
!lHO(a) + RT In aHa + F<l>a = !lHO(~) + RT In aH~ + F<l>~ (3.41)
or
Similarly for the W-a' interface

~<l>a'W = !lHea') - !lH(W) + RT In aH'
F F a~
Since the a and a' phases are both aqueous solutions and the ~ and W
phases are both hydrated glass,
!lHO(a) = !lHO(a') and !lHO(~) = !lHO(W)
Thus the standard state terms cancel when ~<l>a~ and ~<l>I3'a' are added
to obtain the solution/membrane surface contributions to the membrane
potential:
aaa W
~<l>a~ + ~<l>a'W =RT In ~ (3.42)
. F aH'aA
On the other hand, the Na+ ions are also in equilibrium between the
outer membrane regions and the solutions. Thus for the a-~ interface,
!lNaO(a) + RT In aNa a + F<l>a = !lNaO(~) + RT In aNa~ + F<l>~ (3.43)
Subtracting eq (3.43) from eq (3.41) and rearranging, we have
a~aa
!lHO(a) + !lNaO(~) - !lHO(~) - !lNa O(a) = RT In H Na
F aaa~
H Na
but this is just the standard free energy change for the equilibrium
Na+(a) + H+(P) f:! Na+(p) + H+(a)
so that the argument of the logarithm term is just an equilibrium
constant expression
a aa P
K= H Na (3.44)
P a
aHaNa
By a similar argument, we can show that H+ and Na+ ions at the W-a'
interface are subject to the same equilibrium constant.
aa' W
K = HaNa (3.45)
W a'
aHa Na
We turn now to the interfaces between the surfaces and interior of
the membrane. The potential across the p-y (or y-W) interface results
from the difference in the mobilities of the H+ and Na+ ions and should
be given by the Henderson equation, eq (3.36). Since the mobilities of the
anions inside the membrane are assumed to be zero, the prelogarithmic
term simplifies, and we are left with
I P ~ l
~py = RT In lUHa H + UNaa NaJ
F UNaaka
where we have assumed that the activity of H + ions in the interior of the
membrane is zero and that the mobility of Na+ ions is the same in the
interior as at the surface of the membrane. Similarly,
~~. = RT In r UNaa~a lJ'

F LuHa~ + UNaa£·a
Thus the contribution to the membrane potential from the differential
mobilities of the H+ and Na+ ions inside the membrane is
~~y + ~~' = -RT I n faA + (UNa/UH) a£a]

F a~ + (UNa/UH) a£·a
Adding this potential difference to the contributions from the solution-
membrane surface interfaces, eq (3.42), we get the total membrane
potential,
A.rn
'-''-''membrane -
-JJ.::£1 n Ilafi+(UNa/UH)Ka~aJ
. . (3.46)
F aft + (UNa/UH) Ka~a
where we have used eqs (3.44) and (3.45) to eliminate the activities of the
H+ and Na+ ions in the p- and W-phases. We can define a parameter
called the potentiometric selectivity coefficient,
kR,Na = K(llNah~R) (3.47)

which is a function of properties of the membrane-the Donnan
equilibrium constant and the relative mobilities of Na+ and H+. If the
inner electrolyte solution (the a'-phase) is constant in composition, the
denominator of the log term in eq (3.46) makes a constant additive
contribution to the membrane potential. Incorporating kR Na to simplify
the expression further, eq (3.46) becomes '
.1<l>membrane = constant + RJ In [ail + kH,Naa~aJ (3.48)
If the potentiometric selectivity coefficient were zero, then the membrane

would respond exclusively to the H+ activity; the selectivity coefficient
thus is a measure of the interference ofNa+ in the operation of the glass
electrode. Although eq (3.48) was derived for the specific case of a glass
membrane, models of other membrane systems lead to the same general
t
result (Dll,DI4), which can be expressed by
.1<l>memlrane = constant + Rll n [ai + kijaj ] (3.49)
where species i is the principal diffusing species of interest and kij is the
selectivity coefficient for j relative to i.
REFERENCES
(Reference numbers preceded by a letter, e.g. (C3), refer to a book listed in

the Bibliography.)
1. T. Shedlovsky and L. Shedlovsky in Physical Methods of Chemistry,

Part IIA (Techniques of Chemistry, Vol. 1), A. Weisberger and B.
W. Rossiter, eds, New York: Wiley, 1971, p 163.
2. F. J. Holler and C. G. Enke in Laboratory Techniques in
Electroanalytical Chemistry, P. T. Kissinger and W. R. Heineman,
eds, New York: Marcel Dekker, 1984, p 235.
3. P. Erman,Ann. Phys.1801, 8,197; 1802, 10, 1.
4. F. Kohlrausch, Wiss. Abh. Phys. Tech. Reichsanstalt 1900,3, 155.
5. M. Spiro in Physical Methods of Chemistry, Part IIA (Techniques
of Chemistry, Vol. 1), A. Weissberger and B. W. Rossiter, eds, New
York: Wiley, 1971, p 205.
6. J. W. Hittorf, Ann. Phys. Chem. 1853,89, 177.
7. R. D. Shannon, Acta Crystallogr. 1976, A32, 751.
8. L. Onsager, Phys. Z. 1927, 28, 277; L. Onsager and R. M. Fuoss, J.
Phys. Chem. 1932, 36, 2689.
References 147
9. P. Debye and H. Falkenhagen, Phys. Z. 1928,29, 121, 401; Z.

Elektrochem. 1928,23,562.
10. H. Sack, Physik. Z. 1928,29,627.
11. ~.VVien,ilnn.Phys. 1927,388,327;1928,380,795;1929,393,400.
12. P. Debye, J. Chem. Phys. 1933,1,13.
13. R. Zana and E. B. Yeager, Modern Aspects of Electrochemistry 1982,
14,1.
14. F. Kohlrausch and A. Heydweiller, Ann. Phys. 1894,53,209.
15. J. VV. Loveland, in Treatise on Analytical Chemistry, 1. ~. Kolthoff
and P. J. Elving, eds, Part I, Vol. 4, New York: Interscience, 1963, p
2569.
16. P. Henderson, Z. phys. Chem. 1907,59,118.
17. F. G. Donnan, Z. Elektrochem. 1911, 17,572; F. G. Donnan and E. A.
Guggenheim, Z. phys. Chem. 1932,162,346.
PROBLEMS
3.1 (a) A conductance cell filled with 0.1 ~ KCl solution had a
resistance of 24.96 n at 25 C. Calculate the cell constant (LIA)
Q
given that the conductivity of 0.1 ~ KCl is known to be 1.1639 S m- I

at this temperature.
(b) VVhen the cell was filled with 0.01 ~ acetic acid, the resistance
was 1982 n. Compute the molar conductivity of acetic acid at this
concentration.
(c) Using the molar ionic conductivities of H+ and CH3C02- from
Table A. 7, compute A ° for acetic acid.
(d) Compute the degree of dissociation a for 0.01 ~ acetic acid and
estimate the acid dissociation equilibrium constant.
3.2 The molar conductivity of AgN03, measured at 25°C for various

concentrations, is as follows:
C/m~ 0.0276 0.0724 0.1071 0.3539 0.7538

lOONS m 2moI-l 1.329 1.326 1.325 1.316 1.308
Use these data to determine AO for AgN03.
3.3 Compute the Onsager limiting law slope for AgN03 and compare
it with the slope of the plot obLained in Problem 3.2.
3.4 Compute the Onsager limiting law slope for BaCh

3.5 What fraction of the total current is carried by K+ in a solution 0.10

Min KCI and 0.05 M in HCI?
3.6 The conductivity of a saturated silver iodate solution at 18°C is 13.0

x 10-4 S m- l when water with a conductivity of 1.1 x 10- 4 S m- l was
used to prepare the solution. The sum of the molar ionic
conductances is AO(Ag+) + A°(I03-) = 87.3 x 10- 4 S m 2moI-l at 18°C.
What is the solubility of silver iodate at this temperature? You can
assume that the species leading to the conductivity of the water do
not interfere with the dissolution of AgI03 and that the
conductivity contributions are additive.
3.7 The conductivity of a 0.0384 M solution of Cl2 was found to be 0.385

S m- l at O°C. Given that the molar conductivity of HCI is 0.0250 S
m 2mo l-l at O°C, compute the equilibrium constant for the
disproportionation of Ch
CI2(aq) + H20 ~ H+(aq) + CI-(aq) + HOCl(aq)
The ionization of HOCI is negligible under these conditions.
3.8 Compute the Stokes law radii, the ionic mobilities, frictional
coefficients, and diffusion coefficients of Fe(CN)63- and Fe(CN)6 4 -.
3.9 A potassium salt of an uncharacterized anion, An-, contains

12.50% potassium by weight. Conductance measurements on
some solutions of varying composition gave the following data
(aqueous solutions at 25°C):
Clg L-l 0.0250 0.0500 0.01000 0.200 0.500

KiS m- l 1.09 2.17 4.31 8.53 10.94
(a) Compute the "equivalent weight" (MWln) and use this value to
convert the concentrations to equivalents per cubic meter (nClmol
m -3) and the conductances to "equivalent conductivities" (A/n)
with units ofS m 2equiv- l .
(b) Plot Aln us. 'InC, determine the intercept and the slope. How
does this slope differ from the slope of a plot of A us. VC?
(c) Given the molar ionic conductivity of K+, 73.5 x 10-4 S m 2moI-l,
determine the transference numbers, t+ and L.
(d) Suppose that the salt is a 1:1 electrolyte (n = 1). Does the slope
obtained in part (b) agree with theory? Suppose it is a 2:1
electrolyte? Is the fit better? A 3:1 or 4:1 electrolyte?
Problems 149
3.10 Verify that eq (3.22) is a solution to Fick's second law, eq (3.21).
3.11 Show that the diffusion coefficient of a spherical particle of radius

r in a medium of viscosity 11 is D = kTI61tW.
3.12 A simple calculation can show that the development of a central

nervous system was essential to the evolution of large organisms.
Suppose that a primitive fish swimming in a primeval sea at 25°C
relied on the diffusion of information-carrying molecules to
control body movement. Imagine that the information-carrying
molecule was 10 nm in diameter and that diffusion was through
an essentially aqueous medium, 11 = 0.89 x 10.3 kg m·ls· l . Estimate
the diffusion coefficient (see Problem 3.11). Now suppose that a
predator bit the fish's tail. How long would it take for the
information to reach the fish's head if the fish was 10 cm long?
3.13 A solute diffuses through a long cylindrical tube of 2.00 mm

diameter such that during each second 0.025 nanomol crosses a
plane where the concentration gradient is 5.0 M m· l . Compute the
diffusion coefficient.
3.14 A solution of CUS04 was electrolyzed in a Hittorf apparatus with

Cu electrodes. The electrode processes were
anode: Cu(s) ~ Cu2 + + 2 e'
cathode: Cu2 + + 2 e- ~ Cu(s)
The initial concentration was 0.1473 M. After passage of 1.372 x
10.3 Faradays, the cathode compartment contained 30 mL of 0.1183
M CUS04. Calculate the transference number of Cu2 + in the
CUS04 solution.
3.15 Sketch conductometric titration curves for the following systems.

Ignore the contribution to the conductivity from weak electrolytes
or insoluble precipitates.
(a) 0.01 M ammonia titrated with 1 M hydrochloric acid.
(b) 0.01 M ammonia titrated with 1 M acetic acid.
(e) 0.01 M silver nitrate titrated with 1 M hydrochloric acid.
(d) 0.01 M silver nitrate titrated with 1 M potassium chloride.
3.16 150 m.L of a saturated solution of Ba(OH)2 was titrated with 1.00 M
H2S04 and the resistance of the solution measured with a dip-type
cell. The following data were obtained:
VollmL RID. VollmL RID.

0.00 53 6.00 1080
1.00 63 6.30 2500
2.00 84 6.50 1200
3.00 100 7.00 350
4.00 140 7.50 210
5.00 250 8.00 150
5.50 400 8.50 110
5.75 500 9.00 00
Given that the cell constant is LIA 104 m- I , compute the

conductivity K for each point and plot K us. volume of acid. Locate
the equivalence point and compute the concentration of Ba(OH)z in
the saturated solution. Knowing the concentration of Ba(OH)z and
the conductivity of the initial tlolution, determine the molar
conductivity of Ba(OH)z.
3.17 Show that the Donnan membrane equilibrium condition for an

electrolyte where the cations have charge ZI and the anions have
charge Z2 is
3.18 A 0.01 M solution of a co11oidal electrolyte NaX is placed on one

side of a membrane which is permeable to Na+, but not to X-.
Calculate the equilibrium distribution of ions if on the other side of
the membrane there is placed a 0.025 M solution of Na2S04.
Assume equal volumes in the two phases. Calculate the electric
potential difference across the membrane at 298 K.
4
VOL TAMMETRY OF
REVERSIBLE SYSTEMS
In Chapter 1, we considered cells in which chemical energy is

converted to electrical energy. Beginning with this chapter, our focus
shifts to electrolysis cells, where electricity is used to induce chemical
change.
In an electrolysis cell, reduction occur:; at the cathode and oxidation
at the anode just as in a galvanic cell. However, the cathode and anode
are negative and positive, respectively, in an electrolysis cell, seemingly
opposite to the convention adopten for galvanic cells. Consider the
Daniell cell (discussed in §1.1)
ZnIZn2+(aq )IICu 2+(aq)ICu
When operated as a galvanic cell, the cell reaction
Zn + Cu 2+ -7 Cu + Zn 2+
corresponds to oxidation of Zn and reduction of Cu 2+. The Zn electrode
then is the anode and the Cu electrode is the cathode. When we measure
the cell potential, we find that the Zn electrode is negative and the Cu
electrode is positive. When an external potential source is attached and
the direction of current flow is reversed, the relative polarities of the
electrodes are unchanged-Cu is still positive and Zn is negative.
However, the process taking place at the Zn electrode is now a reduction,
so that this electrode is the cathode; Cu is oxidized, so that the Cu
electrode is the anode. Strictly speaking, the designations anode and
cathode refer to the direction of current flow. In order to avoid
confusion, we will usually focus our attention on a single indicator or
working electrode and will describe its role by saying that the current is
anodic or cathodic, i.e., the electrode process at the indicator electrode is
an oxidation or a reduction. We adopt the following sign convention for
current at the indicator electrode: l cathodic current is positive, anodic
current is negative.
lOur sign convention for indicator electrode current is the usual one among
electroanalytical chemists, but many other electrochemists employ (and the
International Union of Pure and Applied Chemistry recommends) the opposite
convention with positive anodic current.
151
152 Voltammetry of Reversible Systems
When current flows through an electrochemical cell, it may be

limited by any of the three steps:
(1) conductance of the bulk solution;
(2) transport of reactants to the electrode or products away from the
electrode; or
(3) the rate of electron transfer and / or chemical reactions coupled to
the electron transfer step.
Examples can be found where anyone of these steps is rate determining.
In Chapter 3, we were concerned with cells designed so that step (1) was
rate limiting. In this chapter, we consider cases where the mass
transport step is slowest; we will assume that the solution conductivity is
high and that electron transfer is fast and reversible and is not
complicated by associated chemical reactions. We will consider cases
where the current is limited by the rates of coupled chemical reactions
or of electron transfer in Chapters 5 and 6. We will be concerned here
with processes where the total charge passed is small, so that the bulk
solution composition is essentially unchanged by electrolysis. In
Chapter 7, we will consider experiments in which the bulk
concentrations change during electrolysis.
We begin with a study of the general problem of diffusion-controlled
current. In §4.2, we digress to some practical matters such as
instrumentation and the choice of solvents. In §4.3 and §4.4, we survey
the host of electroanalytical methods which rely on diffusion control. In
§4.5 and §4.6, we discuss polarography, historically one of the most
important of the electroanalytical methods, and some of the many
variations on the polarographic method. In §4.7 and §4.8, we examine
the characteristics of rotating disk electrodes and microelectrodes, and
finally in §4.9, we will look at some applications of the various methods.
4.1 DIFFUSION-LIMITED CURRENT
There are several mechanisms for transport of ions or molecules to

an electrode surface:
(1) diffusion (transport through a chemical potential gradient);
(2) electric migration (transport through an electric potential gradient);
(3) convection (transport through mechanical motion of the solution).
Convection may occur either through gravity operating on a density
gradient (natural convection) or through stirring of the solution or
motion of the electrode (forced convection).
In this section we will assume that diffusion is the only significant
transport process and will restrict our attention to experiments in which
precautions are taken to eliminate or reduce other transport
§4.1 Diffusion-Limited Current 153
mechanisms. Thus the solution should not be stirred and the cell
should be free of vibration. In addition, we will require that the ratio of
electrode area to solution volume be small and that experiments be short
in duration so that relatively little electrolysis takes place. This
precaution will minimize local changes in concentration or local
heating which could lead to density gradients and convective mixing.
The solution should contain a large excess of an inert electrolyte (called
a supporting electrolyte) to ensure that the double layer is very compact
and that the electric potential is nearly constant throughout the solution,
thus minimizing electric migration effects.
Solution of the Diffusion Equation
Our goal is an equation giving the current as a function of electrode

potential for a relatively large planar electrode; the current will also be a
function of time, the concentrations of electroactive species, the diffusion
coefficients of these species, the electrode area, and the number of
electrons transferred per molecule oxidized or reduced. We assume that
the electrode process involves an oxidized species 0 and a reduced
species R, both in solution, and that the process consists simply of the
transfer of n electrons:
0+ n e- ~ R
The species 0 and/or R may be charged but we will omit the charges for
clarity. We will need to find the concentrations of 0 and R, Co(x,t) and
CR(X,t), as functions of time and distance from the electrode. We define
the problem by the following specifications:
(1) The electrode is planar with surface area A and is sufficiently large
that edge effects can be neglected. Only diffusion along the positive
x-axis perpendicular to the electrode surface need be considered.
(2) The solution is initially homogeneous, i.e., the initial concentrations
of 0 and Rare CO(x,O) = CO* and CR(X,O) = 0 for x ~ O.
(3) The electrolysis cell is large enough that the bulk concentrations of
o and R are unchanged during the time of an experiment, i.e.,
CO(x,t) ~ CO*, CR(X,t) ~ 0 as x ~ 00.
(4) For every 0 molecule consumed, an R molecule is formed; in other

words, the fluxes of 0 and R at the electrode surface are equal and
opposite in sign: JO(O,t) = - JR(O,t).
(5) The electron-transfer re3ction is very fast so that 0 and Rare
always in equilibrium at the electrode surface with the
concentration ratio given by the Nernst equation:
e = Co(O,t) = exp nF (E - EO) (4.1)
CR(O,t) RT
Since the reactant and product concentrations at the electrode

surface obey the Nernst equation, we sometimes say that an
electrochemically reversible system is nernstian. 1 Specification (1) tells
us that we need to consider diffusion of 0 and R along the x-axis only.
The diffusion problem is governed by .F'ick's second law, eq (3.21). Since
we have two diffusing species, we must solve two diffusion equations:
2
aco =Do a Co
at ax 2
2
aCR =DR a CR
at ax 2
Specification (2) represents the initial conditions imposed on the
solutions to the differential equations, and specifications (3), (4), and (5)
represent boundary conditions.
The solution of two coupled partial differential equations might
seem to be a formidable task, but it is really quite easy using the method
of Laplace transforms. 2 The solutions to the differential equations are
the concentrations of 0 and R as functions of x and t:
Co(x,t ) -- C0 * ~8 + erf(x/2V D ot} (4.2a)

1 + ~8
CR(x,t ) --C 0 *~[1-erf(x/2VDRt)l (4.2b)

1 + ~8
where ~ = VDolDR and erft\jf) is the error function, defined by eq (3.23)
and plotted us. \jf in Figure 3.9. Since erf(O) = 0, the concentrations at the
electrode surface (x = 0) are given by
~8
Co(O,t) = Co* - - (4.3a)
1+ S8
CR(O,t) = Co* -~- (4.3b)

1 + ~8
Notice that CO(O,t)/CR(O,t) = 8, as required by eq (4.1). The concentrations
CO(x,t) and CR(X,t) are plotted us. x in Figure 4.1.
1 Strictly speaking, eq (4.1) should include the activity coefficient ratio, YO/YR, but
we will ignore this complication; thus the standard potentials used here are really
formal potentials-see §1.3.
2 The use of Laplace transform methods for the solution of differential equations is
discussf'd, and eqs (4.2) are derived, in Appendix 4. See also Delahay (Bl), Bard
and Faulkner (B12), and MacDonald (F5).
1.0
CIC O*
0.8
0.6
0.4
Figure 4.1 Concentration

profiles for diffusion to a 0.2
planar electrode with 1; =
1 and e = 0.5. The x-axis
scale is in units of the 0.0
diffusion layer thickness, 0.0 0.5 1.0 1.5 2.0
xv = VrrDt. XIXD
Current-Potential Curves
The net current is proportional to the flux of 0 at the electrode:

i = - nFAJo(O,t) (4.4)
The flux is given by Fick's first law, eq (3.19),
Jo(x,t) = - Do ~CoCx,t)
ax
Using eq (3.24) to differentiate eq (4.2a), we obtain
VDohr.t -x2
JoCx,t) = - Co exp - -
I + ~e 4Dot
Setting x = 0, we have the flux at the electrode surface
JoCO,t) = - Co* fI5O!id (4.5)

1 + ~e
Substituting eq (4.5) into eq (4.4), we have
i = nFACo* VDohr._~ (4.6)
1 + ~e
Notice that the current is apparently infinite at zero time. This is an
artifact which results from the inconsistent assumptions that CR(O,O) = 0
and that CO(O,t)ICR(O,t) is finite. We will derive a more accurate
description of the current, including the finite rate of electron transfer,

in §6.3. In practice, eq (4.6) is accurate for reversible systems when t > O.
Since the current computed in eq (4.6) involves electron transfer and is
governed by Faraday's laws of electrolysis (§7.1), we sometimes refer to
faradaic current when we wRnt to distinguish it from other
contributions to the net current such as capacitive charging current.
When the electrode potential is large and negative, E «Eo, eq (4.1)
tells us that e and Co(O,t) are very small. In this case, every 0 molecule
that arrives at the electrode will be reduced. Making the potential more
negative cannot make the current any larger because it is limited by the
rate of diffusion of 0 to the electrode. Setting e = 0 in eq (4.6) then gives
the limiting diffusion current:
iD = nFACo*'iDo/rrt (4.7)
Equation (4.7) was first derived by Cottrell in 1903 (1) and is commonly
called the Cottrell equation.
Frederick G. Cottrell (1877-1948) is best known as the inventor of the

electrostatic precipitator fol' removing particles from flue gases and as
the founder of The Research Corporation. His contributions to
electrochemistry were made during his graduate work with Ostwald at
the University of Leipzig.
Example 4.1 Compute the current for a one-electron

reduction of a species with CO* = 1 mol m- 3 (1 mM) and Do =
10- 9 m 2 s- 1 at an electrode with A = 10- 6 m 2 (1 mm 2 ) at times 1,
10, and 100 s after application of the potential.
If we use SI units for all the quantities on the right of eq (4.7),

we will get the current in amperes (A). In particular, A must
be in m 2 , CO* must be in mol m- 3 , DO must be in m 2 s- 1 , and tin
s. Thus eq (4.7) gives
iD = 1. 72 x 10-6 t- 1I2 amperes
so that iD = 1.72, 0.54, and 0.17 IJA. at t = 1, 10, and 100 s.
The potential dependence of the current is most clearly seen if we

eliminate time and the other parameters of eq (4.6) by dividing by iD
i ~1~ (4.8)
iD 1+ Se
Rearranging and substituting for sand e, we obtain a form which bears
a superficial resemblance to the N ernst equation:
E=Eo_RTlnyDO +BI.lniD-i (4.9)

nF DR nF i
A plot of i/iD vs. E is shown in Figure 4.2. Because of the shape, current-
potential curves like this are referred to as waves. Common examples of
such curves are the polarographic waves discussed in §4.5, but similar
curves arise in other experimental contexts. The first two terms on the
right-hand side of eq (4.9) are usually lumped together to define a new
parameter, called the half-wave potential:
El/FEo_RTlnDo (4.10)
2nF DR
Equation (4.9) then can be written as
E=El/2+RTlnlD~l (4.11)
nF l
Equation (4.11) was derived by Heyrovsky and Ilkovic in 1935 (2) and is
called the HeyrovskY-Ilkouic equation.
Because of the term involving the diffusion coefficients, half-wave
potentials are not standard potentials (or formal potentials), even for the
electrochemically reversible processes we will be discussing. However,
even if the diffusion coefficients differ by a factor of 2, E 112 will differ
from EO by less than 10 mV (for n = 1, 25°C) and for many practical
purposes, half-wave potentials measured from current-potential curves
can be used as approximations to standard potentials. Some examples of
such applications are discussed in §4.9.
1.0
iliD
0.8
0.6
0.4
0.2
Figure 4.2 Current-potential -0.0

curve for reversible electron
transfer and diffusion- 0.2 0.1 0.0 -0.1 -0.2
limited current (n = 1).
(E - E 1I2 )N
JaroslavHeyrovskY (1890-1967) was Professor of Chemistry at the Charles

University in Prague. Heyrovsk.Y's interest in electrochemistry began
during his graduate work with Donnan at University College, London.
On returning to Prague after World War I, he invented polarography
and was responsible for many of the developments and elaborations of the
polarographic method; in recognition of this work, Heyrovsky was
awarded the Nobel Prize in Chemistry in 1959. Dionyz nkovic (1907-
1980) was a student and colleague of Heyrovsky in Prague and later was
Professor of Physics at the Komensky University in Bratislava. Ilkovic
is best known for his pioneering theoretical work in electrochemistry.
Example 4.2 Compute the electrode potential for i = 112 iD and

the potential difference corresponding to i = 1/4 iD and i = 3/4 iD .
When i = 112 iD ,
(iD -i)/i = 1
so that E = E1J2, hence the name half-wave potential. When i =
1/4iD and 3/4 iD ,
(iD - i)/i =3 and (iD - i)/i = 113
Thus
or, at 25°C,
E 114 - E3/4 = 56.51n m V
This property of a reversible wave was proposed in 1937 by
Tomes (3) as a quick indication of nernstian behavior and is
called the Tomes criterion of reversibility.
Diffusion Layer Thickness
It is sometimes convenient to think in terms of a layer near the

electrode surface in which the solution is depleted of electroactive
material. We define the thickness of this depleted layer, XD, as the
distance over which a linear concentration gradient would produce the
same flux at the electrode surface as calculated from an exact solution to
the diffusion equation. 1 The flux through a linear concentration
gradient is
1 This approach to diffusion problems is due originally to Nernst and we sometimes

refer to the Nernst diffusion layer.
JO(X,t) = -- Do aCo(x,t)
ax
J (0 t) = -D Co* - Co(O,t) (4.12)

o , 0 XD
Substituting for CO(O,t) from eq (4.3a), we have

JoCO,t) = - Co* DO/XD
1 + ~e
Comparing this expression with the flux obtained from the exact
solution, eq (4.5), we obtain
(4.13)
Note that the thickness of the diffusion layer increases as Vt for diffusion
to a planar electrode in unstirred solution.
The abscissa of the concentration us. distance plots of Figure 4.1 is
in units of XD. Notice that the concentrations are indeed linear in X/XD
for a short distance from the electrode surface. The hypothetical linear
concentration gradients (the dashed lines in Figure 4.1) correspond to
extrapolation of this linear region to x = XD. We will use the concept of
diffusion layer thickness in several problems discussed below. In effect,
we will then assume that the concentration us. distance gradients are
linear and that Co(x) = Co* and CR(X) = 0 for x> XD.
Mass-Transport Rate Constants
The diffusion of species 0 Lo Lhe electrode surface and its escape

from the surface may be thought of as first-order heterogeneous
processes governed by a mass transport rate constant kDO:
kDO
0* f2 00
kDO
where the net rate of delivery of 0 to the surface is
Rate = kDOCO* - kDOCO(O)
This rate has units of mol m- 2 s- 1 . If the concentrations have units of mol
m- 3 , the rate constant kDO must have units of m s-l, the SI unit for a
first-order heterogeneous rate constant. To put it another way, the rate
we are describing is just (the negative of) the flux of 0 at the surface,
JoCO) = - kDo[CO* - CoCO)]
Comparing this with eq (4.12), we get the relation between the mass
transport rate constant and the diffusion layer thickness for planar
diffusion
kDO = DolxDO (4.14)

It is sometimes convenient to write the diffusion current in terms of XDO
or kDO; with eqs (4.13) and (4.14), cq (4.7) becomes
iD = nFACo*DoIxDO (4. 15a)
iD = nFAkDoCO* (4. 15b)
Example 4.3 Compute XD and kDO for Do = 10-9 m 2 s- 1 at t = 1,

10, and 100 s.
Substituting Do into eq (4.13), we have

xv/m = 5.60 A 10-Pi Vt
so that XD = 0.056, 0.177, and 0.560 mm at t = 1, 10, and 100 s
Substituting these values in eq (4.14), we find kDO = 17.9,5.6,
and 1.8 11m s-l at the same times.
General Current-Potential Curve

The generality of eq (4.11) is restricted by our assumption that the
bulk concentration of R is zero. The general problem with both 0 and R
present is easily solved using Laplace transforms, but to illustrate the
use of mass transport rate constants, we will follow a less rigorous route
which nonetheless gives the correct result.
When the electrode potential is sufficiently positive that every R
molecule which arrives at the electrode surface is immediately oxidized,
the current is again diffusion-limited and, by analogy with eq (4.15b),
should be given by
(4.16)
where XDR is the diffusion layer thickness for R and the negative sign is
required to preserve the sign convention. In general, the current is
proportional to the fluxes Jo(O,t) or JR(O,t),
i =- nFAJo(O,tl = nFAJR(O,t)
and the fluxes can be written in terms of the hypothetical linear
concentration gradients:
Jo(O,t) = - kDO [CO* - CO(O,t)]
JR(O,t) = - kDR [CR* - CR(O,t)]
Solving for Co(O,t) and for CR(O,t), we have
Co(O,t) = Co* - i/nFAkDO

CR(O,t) = CR* + ilnFAkDR
Solving for Co* and CR* from eqs (4.15b) and (4.16) and substituting into
the expressions for the surface concentrations, we have
iDc-i
Co(O,t) = -=-=----
nFAkDo
i-iDa
CR(O,t) = 'Ak
nFn. DR
If the Nernst equation is obeyed at the electrode surface, we have
E = EO + RT In Co(O,t)
nF CR(O,t)
Substituting the expressions for the surface concentrations gives
E=Eo+RTlnkDR +RTln iDc- i
nF kDo nF i-iDa
or
E = EY2 + ~ In ~Dc ~ i (4.17)
nF l-lDa
where E1I2 is given by eq (4.10). The current-potential curve represented

by eq (4.17) is shown in Figure 4.3. Notice that this curve is identical in
shape to that derived from eq (4.11) and shown in Figure 4.2. Indeed, the
only difference is a displacement of the curve along the current axis.
When the solution contains two reducible species, 0 1 and 02, we can
usually assume that they diffuse independent1y and that the electron-
transfer processes do not interfere. 1 If the half-wave potentials are
separated by more than about 150/n mV, the composite current-potential
curve will show two resolved waves. Figure 4.4 shows some composite
voltammetric curves computed for C01* = Co 2*, CR1* = CR2* = 0, n1 = n2 =
1, and E1I2(2) - E1I2(1) = 0.1, 0.2, 0.3, and 0.4 mV.
There is one case which is not covered by the general result, eq
(4.16). When one member of the electrode couple is a solid coating the
electrode, its activity at the electrode surface will be independent of
potential; if the solid is pure, its activity is 1. The most obvious example
of this situation is the reduction of a metal ion at an electrode of the
same metal, e.g., Hg2+ reduced at a mercury cathode. In this case the
analog of eq (4.17) is
1 If one of the components of a couple is adsorbed or deposited on the electrode

surface, this may not be a good assumption, but ordinarily the contribution to the
current from each process can be computed from eq (4.17) independently and added
to get th e total curren t.
E = EO + RT In Co* + MIn iD - i (4.18)

nF nF iD
When i = 0, the last term ofeq (4.18) vanishes and the Nernst equation is
recovered. Since Co* in this equation is an approximation to the activity,
its units must be mol L-l. The current-potential curve represented by eq
(4.18) is shown in Figure 4.5. Notice that the anodic current increases
without limit as the potential becomes positive.
Diffusion to a Spherical Electrode
In our discussion so far, we have assumed diffusion to a planar

electrode big enough that "edge effects" could be ignored. We haven't
been very specific about what these edge effects are or what they might
do the our results. The planar diffusion approximation would be exact if
the electrode occupied the entire area of the end of a long tube, but in
practice, the radius of a planar electrode is usually small compared with
the radius of the cell. For a large electrode, most of the diffusion is
normal to the plane of the electrode, but there is also a component
O.B
2.0
0.6 if iD
1.5
~ 0.4
.8
I
<.> 0.2 1.0
.8
~ 0.0
0.5
iDa
-0.2
0.0
-0.4
0.2 0.1 0.0 -0.1 -0.2 0.2 0.1 0.0 -0.1 -0.2
(E -E 1/2 )N (E - E 1/2 )N
Figure 4.3 Voltammetric curve for Figure 4.4 Voltammetric curves

a one-electron process where for a two-component system
CO*ICR* = 7/3. showing the effect of half-wave
potential separation (t.El/2 = 0.1,
0.2, 0.3, 0.4 V).
1.0
iliD
0.5
0.0
-0.5
Figure 4.5 Current-potential
curve for reversible electron
transfer where transport of
o is diffusion limited but R -1.0
is a solid on the electrode
surface, n = 1, Co* = 1 mM. -0.1 -0.2 -0.3 -0.4
(E -EO)N
normal to the hemispherical surface around the edge. When we discuss

microelectrodes in §4.8, we will have to come to grips with that problem.
In the meantime, we can get some feel for the problem by considering
diffusion to a spherical electrode where the geometry is three-
dimensional, but we have only one radial coordinate to deal with.
Consider a spherical electrode of radius ro where the potential is
sufficiently negative that every 0 molecule arriving at the electrode
surface is reduced. In spherical symmetry, Fick's second law becomes
aC(r,t) = D (iC(r,t) + z. ac(r,o) (4.19)

at a r2 r ar
where r is the distance from the center of the sphere. Given the initial
and boundary conditions-C(r,O) = C*, r> ro; C(ro,t) = 0, t > O-the solution
to the differential equation (8<;;0 Appendix 4) is found to be
C(r,t) = C* [1- r; (1- erf ~ ~)] (4.20)
The equivalent expression for planar diffusion, derived from eq (4.2a) by

setting e = 0, is
C(X,t) = C* erf ~
2m
and we see that, for rolr '" 1, r - ro = x, and the two expressions are
identical. In other words, concentrations near the surface (r '" ro) are
nearly the same as computed assuming planar diffusion.
Differentiating eq (4.20) and setting r = ro, we get the flux at the electrode
surface
J(ro,t) =_D(aC(r,t)) =-C* (~ + fo)

ar r =1'0
so that the limiting diffusion current is

iD =nFADC* (_1_ + .l.) (4.21)
-{1tDt ro
The time-dependent term is identical to eq (4.7), derived assuming
planar diffusion, so that again we expect the current to decay as t- 1/2 .
For the sphere, however, we also have a time-independent term. Thus
we expect the current to decay to a steady-state value rather than to zero.
The relative importance of the two terms obviously depends on the
electrode radius. Figure 4.6 shows a log-log plot of diffusion current
density (in = in/A) us. time for spherical electrodes with radii varying
from 10 mm down to 1 Ilm. For the largest electrode, the plot is linear up
to 100 s (i.e., the current is proportional to t- 1I2 and the steady-state
contribution is negligible). For the 1 mm and 100 Ilm electrodes,
departure from linearity is noticeable after about 1 sand 10 ms,
respectively, but for the smaller electrodes, the steady-state regime is
entered very early in the experiment. Thus as the electrode radius
decreases toward zero (the electrode approaches a point sink) the time-
dependent current becomes negligible in comparison with the steady-
state term.
Spherical diffusion is apparently much more efficient than planar
diffusion in delivering an electroactive species to the electrode surface.
Although the details depend on electrode shape, the diffusion layer
thickness xn always increases with time as current flows. For a large
planar electrode, the area of the diffusion layer remains constant, so
that the surface flux diminishes with time; but for a spherical electrode
the diffusion layer area increases as (ro + xn)2; as it happens, the
influence of diffusion layer thickness and area exactly cancel in the
limit of zero electrode radius. For a large electrode or at short times, xn
« ro and diffusion is approximately planar. But when the electrode is
very small, spherical diffusion becomes important early in the
experiment. As inhabitanls of a spherical planet, we can readily
appreciate the situation: on the ground, or even in a jetliner at 10,000
meters, the earth's surface is pretty close to planar; but from a satellite,
the round shape is much more obvious.
For a planar electrode of finite size, the situation is analogous: so
long as xn « ro, planar diffusion dominates and the current decays to
near zero as t- 172 ; when the electrode is smaller, however, the
contribution of edge diffusion is more significant and the current decays
to a nonzero, steady-state value.
100
jlA m- 2
10
1
Figure 4.6 Diffusion-
limited current density
at spherical electrodes
with ro = 1 ~m - 10 mm; 0.1
note the decrease of the
steady-state contribution 0.001 0.01 0.1 1 10 100
with increasing radius. tis
4.2 EXPERIMENTAL TECHNIQUES
Measurement and Control of Potential

Before World War II, voltammetry experiments were done with two
electrodes, an indicator electrode, often the dropping mercury electrode,
and a reference electrode, often the saturated calomel electrode. In
Heyrovskfs polarography experiments (see §4.5), a d.c. ramp voltage
was applied across the cell from a motor-driven potentiometer which
was calibrated with a standard cell. The current was recorded as the
voltage drop across a measuring resistor. Heyrovsky recorded the
current with a damped galvanometer l which reflected a light beam onto
a piece of photographic paper mounted on a drum which was rotated by
the potentiometer motor. This device was later replaced by a strip-chart
recorder. The two-electrode system works reasonably well for aqueous
polarography where the solution resistance is low and the current is
small. When nonaqueous systems are used or the experiment generates
a larger current, iR drop in the solution and polarization of the
reference electrode become serious problems.
Beginning in the 1950's, most electrochemical experiments have
been done with three electrodes and instrumentation built around a
potentiostat. A common design based on an operational amplifier is
1 Because of the slow response time of the damped galvanometer, this instrument
recorded the average polarographic current. Thus many early treatments of
polarography discuss average currents rather than the current at drop fall.
shown in Figure 4.7. A true reference electrode is used to measure the

potential; a voltage follower (see § 1.4) is attached to the reference
electrode so that very little current is drawn in this part of the circuit.
An auxiliary electrode then is used to pass the bulk of the current. The
potentiostat works as follows: a control potential, equal and opposite to
the desired potential of the reference electrode, is added to the actual
reference electrode potential and the sum applied to the negative
(inverting) input of the operational amplifier. The positive input is at
ground potential (0 V) and, as we will see, the indicator electrode is also
at 0 V. Suppose that the indicator electrode is to be at -1.00 V relative to
the reference. Since the indicator electrode is at ground potential, the
reference electrode should be at +1.00 V; thus the control potential
should be -1.00 V. Ifthe actual reference electrode potential is 0.99 V, the
potential at the inverting input will then be -0.01 V and the output will be
large and positive. A large current is then passed through the cell,
increasing the potential of the solution relative to the indicator electrode
and thus increasing the reference electrode potential. As the reference
potential approaches 1.00 V, the input to the amplifier approaches zero
and the output becomes less positive. At equilibrium, the output voltage
applied to the auxiliary electrorle will produce enough current to
polarize the cell such that the reference electrode potential
Potential 0--_ _ _.......,

Control R
Inputs
R
R
Reference
Potential o-~--+
Monitor
r Current
Output
Figure 4.7 Potentiostat circuit for control of indicator electrode potential

in a three·electrode cell. Cell current is monitored with a current-
follower circuit, and the reference electrode is protected from excessive
current flow by an electrometer.
§4.2 Experimental Techniques 167
Figure 4.8 An operational <Dout

amplifier integrator circuit
provides a voltage ramp signal. r
exactly cancels the control potential. See Roe (4) for further details and
more advanced designs.
The current is measured by means of another operational amplifier
circuit attached to the indicator electrode lead. Since the input
impedance of the operational amplifier is high, the cell current must
flow through resistor R2. However, the high gain of the amplifier
produces an output voltage such that the potential at the inverting input
is 0 V. Thus the indicator electrode is held at virtual ground and the
output potential, -iR2, is proportional to the cell current.
If the reference electrode is to be held at constant potential, a simple
battery could be used to supply the control voltage. However many
experiments require a potential which changes linearly with time. Such
a linear ramp voltage is most commonly supplied by yet another
operational amplifier circuit, which acts as a voltage integrator. This
circuit, shown in Figure 4.8, again relies on a feedback loop which keeps
the potential at the inverting input at virtual ground. Current from the
battery flows through the resistor R. Because the amplifier input
impedance is high, the current, dQ/dt, charges the capacitor C. If the
input voltage, <Din, is positive, the output will be negative and
proportional to the charge on the capacitor:
<Dout = - QIC
Thus if Q = 0 at t = 0,
or, since i = <DinIR,

~ __ <l>jn t
out- RC
Thus for a positive input, the output voltage is a negative-going ramp. If,
for example, R = 10 MQ, C = 10 IlF, and <Din = 100 mY, then <DiniRC = 1
mV s·l. The output of the ramp generator is applied to one of the inputs
of the potentiostat along with a constant potential source which sets the
initial reference electrode potential.
Still other control potentials can he added to program the reference
electrode potential as required. Thus a triangular wave potential is used
for cyclic voltammetry (§4.4), a sinusoidal a.c. potential for a.c.
polarography (§6.4), and various kinds of voltage pulses for pulse
polarography (§4.6).
The three-electrode configuration avoids the problem of reference
electrode polarization and somewhat reduces the measured iR drop in
the solution. One way of further reducing solution iR drop is to locate
the reference electrode junction as close as possible to the indicator
electrode, often through use of a salt bridge with a long curved capillary
tip as shown in Figure 4.12b (a Luggin probe).l Even then, iR drop often
remains a problem and most commercial electrochemical instruments
contain a circuit which partially compensates for iR drop by electrically
subtracting an adjustable voltage, proportional to to the current, from
the measured cell potential.
Control of Current
In some experiments (e.g., chronopotentiometry; see §4.3), a
constant current is pa::;::;ed through the cell, and the potential of the
indicator electrode is monitored using a reference electrode. An
operational amplifier circuit which performs this function (a
galuanostat circuit) is shown in Figure 4.9. Since the operational
amplifier input impedance is high, the cell current must flow through
the resistor R. Feedback through the cell ensures that the potential at
the inverting input is at virtual ground. Thus the potential drop across
the resistor is equal to the battery potential and the cell current is
i = f,,¢IR
With the battery polarity as ::;hown in Figure 4.9, the indicator electrode
is the cathode; reversal of polarity would change the direction of current
flow. In this circuit, the ultimate source of current is the battery; the
function of the operational amplifier is to adjust the potential of the
auxiliary electrode to keep the current constant. In experiments where
the current is to be controlled as a function of time, the battery can be
replaced by a potential program circuit as described above.
Choice of Solvent
There is no single solvent which is ideal for all electrochemical

work. The choice is often dictated by the solubility and reactivity of the
materials to be studied. Reductions often produce a strongly basic anion
1 Named after the glassblower who constructed the prototype.

>-.......--0 Reference
Potential
Monitor
Figure 4.9 Galvanostat circuit for control of current through a cell.
or radical anion which is rapidly protonated by a solvent like water or an

alcohol. Oxidations often give an acidic cation or cation radical which
reacts with a basic or nucleophilic solvent. The ubiquity of water as an
impurity in other solvents makes the ease of solvent purification and
drying an important consideration. See Mann (5) or Fry and Britton (6)
for further details. Several common solvents are listed in Table A.B,
together with some other parameters which merit attention:
Liquid range. Generally, the greater the liquid range, the greater
the flexibility in experiments. It is often useful to examine
electrochemical behavior at low temperature if follow-up chemical
reactions are fast at room temperature or at higher temperature if
electron transfer is slow.
Vapor pressure. Since oxygen interferes with most
electrochemical experiments, the apparatus is usually flushed with
nitrogen or argon. The purge gas may be pre saturated with solvent, but
when the vapor pressure is high (e.g., CH2CI2) it is difficult to maintain
constant composition and constant temperature because of solvent
evaporation.
Dielectric constant. To obt:cin conducting electrolyte solutions,
the solvent dielectric constant should be large. With a dielectric constant
of only 2.2, p-dioxane is a poor electrochemical solvent and is usually
used with 5-25% water added. Dimethoxyethane, tetrahydrofuran, and
dichloromethane are only marginally acceptable; solution iR drop is
usually a serious problem in these solvents.
Solvent viscosity. Solvents of low viscosity will generally give
electrolyte solutions with greater conductivities. On the other hand,
because diffusion is faster in such cases, diffusion-controlled chemical
reactions will also be faster. Thus it is sometimes found that an
electrode process which is irreversible (because of a fast following
chemical reaction) in a nonviscous solvent becomes reversible in a more
viscous medium.
Supporting Electrolytes
Voltammetry and related electrochemical techniques require an
excess of inert electrolyte to make the solutions conducting and to reduce
the electrode double-layer thickness. In principle, any strong electrolyte
will satisfy these basic requirements, but there are other considerations
(6).
In aqueous solutions, KCI and HCI are common choices since the
liquid junction potentials with s.c.e. or Ag/AgCI reference electrodes can
be eliminated. Aqueous solutions are usually pH-buffered and the
components of the buffer system often act as the supporting electrolyte.
Clearly, the salt should not be easily oxidizable or reducible. The
accessible potential range at Pt and Hg electrodes is given in Table A.9
for several electrolyte/solvent combinations.
In low-dielectric-constant solvents, tetraalkylammonium salts are
more soluble and are less easily reduced than alkali metal salts.
Quaternary ammonium ions do not form tight ion pairs with anions.
Thus R4N+ salts are by far the most common choice, although
tetraphenylphosphonium salts and lithium salts are sometimes used in
organic solvents. When tetraalkylammonium ions are reduced,
however, they form surface-active polymers which coat electrodes, foul
dropping mercury electrode capillaries, and generally raise havoc with
experiments.
The choice of anion is less obvious. The simple halides, CI-, Br-, and
1-, are relatively easily oxidized and often form tight ion pairs, so that
they are frequently avoided. The most common choices-CI04-, BF4-,
PFs-, and BPh4--have delocalized charge, so that their salts are often
soluble in organic solvents. Ion pairing usually is not severe and these
anions are not easily oxidized or reduced at electrodes.
Reference Electrodes
In aqueous solutions, the calomel and Ag/AgCI electrodes are well
characterized and give reproducible potential readings with a minimum
of experimental difficulty. Unfortunately, there is no universally
accepted reference electrode for use with nonaqueous solvents. There
are two commonly used approaches.
One school of thought is to use an aqueous s.c.e. with a salt bridge.
The advantage is that the reference electrode is well understood and
generally reproducible. There are some major disadvantages: an
unknown liquid junction potential is introduced and water
contamination through the salt bridge is difficult to avoid completely.
Since the s.c.e. electrolyte is normally KCI, contamination by K+ and CI-
ions can sometimes lead to problems. For example, a KCI-filled salt
bridge in contact with a solution containing CI04- ions frequently leads
to precipitation of KCI04, which is quite insoluble in organic solvents
and not very soluble in water. The precipitation usually takes place at
the point of solution contact, often clogging the salt bridge and blocking
current flow. One way of avoiding this particular problem is to replace
the s.c.e. electrolyte with NaCl; NaCl04 is much more soluble.
A second common approach is to use the Ag/Ag+ couple as a
reference, dissolving a suitable silver salt in the same
solvent/supporting electrolyte system used in the experiment and
placing it in contact with a silver wire. The advantages of this approach
are that the liquid junction problems and solvent and electrolyte cross-
contamination problems are minimized. The disadvantage is that
solvent evaporation from the reference electrode or change in the silver
surface with time may lead Lo nonreproduciblc potentials.
With many different reference electrodes in use, potential
measurements from different laboratories are often difficult to compare.
It is becoming standard practice in electrochemical studies of
nonaqueous systems to use a standard reference couple of known (or at
least commonly accepted) potential against which the reference
electrode can be checked from time to time; potentials can then be
reported relative to the standard couple or corrected to a common scale.
Ferrocene, (C5H5)2Fe, is reversibly oxidized to the ferrocenium ion,
(C5H5hFe+, at +0.08 V us. Ag/Ag+ in acetonitrile (7) and is the most
commonly used potential standard. Ferrocene and ferrocenium salts
are soluble in most nonaqueous solvents; in addition, the
ferrocene/ferrocenium couple is relatively insensitive to solvation or ion-
pairing effects and so provides an approximation to an absolute
reference.
Indicator Electrodes
In §4.1, we assumed that the indicator electrode was planar and

sufficiently large that we could ignore edge effects in solving the
diffusion problem. Provided that the radius of curvature is large
compared with the diffusion layer thickness, qualitatively similar
results are obtained for cylindrical or spherical electrodes, indeed for
any stationary electrode in an unstirred solution. Common designs for
stationary electrodes include all three geometries: 1 planar (metal disk
electrodes), cylindrical (wires), and spherical (hanging mercury drop).
When the electrode is very small, edge effects dominate and steady-
state currents are often obtained. Since the behavior of microelectrodes
is qualitatively different, we will discuss them separately in §4.8.
Two other indicator electrodes are commonly used in
electroanalytical and mechanistic work. The dropping mercury
electrode (d.m.e.), shown in Figure 4.10, consists of a fine-bore capillary
1 For reviews on electrodes, see Adams (8,C3), Galus (9), Dryhurst and McAllister
(10), or Winograd (11).
contact
bushing
Teflon
electricalvA-...---_ insulation
contact
(1"electrode
capillary
':::::?'
Figure 4.10 Dropping Figure 4.11 Rotating-
mercury electrode. disk electrode.
through which mercury flows, forming a drop at the end of the capillary
which grows until its weight exceeds the force of surface tension holding
it to the capillary. Depending on the length and bore of the capillary, the
pressure of mercury above the capillary, and the Hg-solution interfacial
tension, the lifetime of a mercury drop can be anywhere from 1 to 10 s.
The factors governing the drop time were discussed in §2.S. The current
through a d.m.e. is time dependent, but because the solution is stirred
when a drop falls, each new drop starts the experiment anew. If the
current is measured just before the drop falls, experiments using a
d.m.e. are essentially at constant time. We will discuss the operation of
the d.m.e. in greater detail in §4.S.
The rotating-disk electrode (r.d.e.), shown in Figure 4.11, consists of
a flat disk, usually 1-3 mm in diameter, mounted at the end of an
insulating rod which is rotated rapidl.y in the solution. The rotational
motion stirs the solution so that the diffusion problem is reduced to
transport across a stagnant layer at the electrode surface. Since the
rotation speed is constant and the stirring effect is reproducible, an
experiment using the r.d.e. is carried out under steady-state (i.e., time
independent) conditions. We will discuss the operation of a r.d.e. in
greater detail in §4. 7.
Cell Design
The detailed design of electrochemical cells for voltammetric

experiments depends on the technique and on the requirements of the
(a)
+--N2
d.m.e.
waste Luggin
Hg probe auxiliary
electrode
Figure 4.12 Examples of cells for voltammetric experiments: (a) H-cell

for two·electrode polarography; (b) three-electrode polarography cell.
chemical system being studied (12). Here we describe two common

designs, shown in Figure 4.12, as representative examples.
The so-called H-cell, shown in Figure 4.12a, is a popular design for
aqueous polarography. A separate compartment for the reference
electrode prevents contamination of the test solution by KCl. Provision is
made for purging the test solution with nitrogen (or argon) gas before
the experiment and blanketing the solution with inert gas during the
experiment. The indicator electrode (here a dropping mercury
electrode) is mounted through a stopper with a small hole for escape of
the purge gas. This cell is easily adapted to a three-electrode
configuration by contacting the pool of waste mercury with a tungsten
wire and using this as the auxiliary electrode. However, because the
reference and indicator electrodes are widely separated, iR drop can be a
major problem with this cell.
A somewhat more flexible design is shown in Figure 4.12b. Here a
small glass cell body is clamped to a plastic top provided with ports for
the electrodes and gas purge tube. The reference electrode is mounted in
a salt-bridge tube which terminates in a Luggin probe situated as close
as possible to the indicator electrode. The auxiliary electrode is typically
a simple platinum wire and the purge tube is equipped with a two-way
stopcock to permit gas to flow through or over the solution. The tip of the
purge tube is sintered glass to disperse the gas into fine bubbles. Cells
like this are quite well suited to most voltammetric techniques and work
well, provided that the solvent is not too volatile. The cell body can be
fitted with a flow-through thermostat jacket for work away from room
temperature and the entire experiment can be mounted in a glove box in
the event that the analyte is very sensitive to air or moisture.
When solvent volatility is a problem or when the sample is
exceedingly sensitive to air or moisture, cells are often designed to be
filled on a vacuum line so that the purge gas is unnecessary.
4.3 A SURVEY OF ELECTROANALYTICAL METHODS
Because the limiting diffusion current in a voltammetric

experiment is proportional to the concentration of the electro active
material, such an experiment is potentially adaptable to chemical
analysis, and for many years electrochemists have been busy developing
instrumental methods for measuring concentrations of reducible or
oxidizable species in solution. There are so many methods now available
that the neophyte is often quite overwhelmed. Indeed, the very wealth of
electroanalytical techniques has acted sometimes as a barrier to their
use by other chemists. In this section we will attempt to reduce the
mystery of electroanalytical methods to manageable proportions.
If electron transfer is nernstian and the current is diffusion
controlled, the current and potential are related by eq (4.17) to three
parameters: the half-wave potential E 112 and the cathodic and anodic
diffusion currents iDe and iDa. The diffusion currents are related to the
bulk solution concentrations of the participants in the electrode process,
o and R, and to the mass-transport rate constants, kDO and kDR, by eqs
(4.15b) and (4.16). The mass-transport rate constants depend on the type
of indicator electrode used but in general are functions of time. Thus an
electrochemistry experiment has four variables which may be held
constant, varied, or measured: current, potential, time, and solution
composition.
Constant Time Experiments
We will consider first those electro analytical methods for which

time is not an explicit variable, either because the experiment is carried
out under steady-state conditions (using a microelectrode or an r.d.e.),
because measurements are always made at the same times in repetitive
experiments (using a d.m.e.), or because the current is zero. In these
cases, the kD'S are constant and it is convenient to lump the various
terms ofeqs (4.15) and (4.16) together and write
§4.3 A Survey of Electroanalytical Methods 175
Figure 4.13 Surface

showing the relationship
of current, potential, and
composition for a con-
stant time experiment.
iDe = OoCO*
iDa = - ORCR*
where
00 = nFAkvo
OR = nFAkvR
It is also convenient to let X be the mole fraction of the electroactive
material in the oxidized form and C* be to the total concentration so that
Co* =XC*
CR* = (l-X)C*
With these changes in parameters, eq (4.17) becomes
E =Ell'!. + RT ooXC* - i
~ In ---=----- (4.22)
nF i + oR(l-X)C*
Equation (4.22) describes the interrelationships of three variables:
potential, current, and composition. These interrelationships are most
easily visualized by thinking of a surface in three-dimensional space as
shown in Figure 4.13. 1 We can describe a number of electroanalytical
techniques as excursions on this surface.
1 This way of describing electroanalytical experiments is due to Reilley, Cooke, and

Furman (13).
Potentiometric Titrations
When the current is equal to zero, eq (4.22) reduces to the Nernst
equation, which describes a potentiometric titration (see §1.7) in which
the cell potential is measured as a function of added titrant, i.e., as a
function of solution composition. The potentiometric titration curve
corresponds to the zero current path across the surface of Figure 4.13 as
shown in Figure 4.14. Of course, in a real case, the titrant and its
reduced form (assuming that the titrant is an oxidizing agent) form
another electrode couple, and beyond the endpoint it is this couple which
determines the cell potential. Thus, to complete the titration curve, we
would have to graft another similar surface onto Figure 4.13, so that
when X approaches 1, the potential goes to some finite value rather than
to infinity as implied by the single surface.
Figure 4.14 The i = 0 path on the three-dimensional surface is the

familiar potentiometric titration curve.
Polarography and Steady-State Voltammetry
Current measurements using a dropping mercury electrode

(d.m.e.) are essentially constant time experiments. Since very little
current is passed, little electrolysis takes place, so that the solution
composition is constant. Thus a curve at constant X on the surface of
Figure 4.13 is identical to the current-potential curve of Figure 4.2.
When a d.m.e. is used, the voltammetric experiment is called
polarography (or d.c. polarography to distinguish the technique from the
variations discussed in §4.6 and §6.4) and the current-potential curve is
called a polarogram. The current on the diffusion-limited plateau, of
course, is related to the bulk solution concentration. Polarography is one
of the oldest electroanalytical techniques and, although it has been
displaced by other methods for many analytical applications, it
continues to be widely used. We will return to a more detailed
discussion of the method in §4.5 and will give some examples of
analytical applications in §4.9.
Currents through a microelectrode or a rotating-disk electrode may
be measured under steady-state conditions. A steady-state
voltammogram thus resembles a polarogram and again is essentially
the curve at constant X on the surface of Figure 4.13, i.e. the current-
potential curve shown in Figure 4.2.
Amperometric Titrations
Imagine an excursion on the surface of Figure 4.13 where the

current is measured as a function of composition, holding the potential
constant. We expect to get a straight line terminating at X=: 1. Such an
excursion would be obtained if the current were measured during a
titration of the electroactive material. In practice, such a titration would
be carried out using a d.m.e., a microelectrode or an r.d.e., so that again,
the current would be measured at constant time (i.e., the drop time) or
under steady-state conditions. The potential could be set so that either
the analyte or the titrant is reduced (or oxidized) at the electrode. If the
analyte is reduced, then the titration curve corresponds to a straight line
going to zero at the endpoint; if the titrant is reducible, then a linearly
increasing current is obtained, commencing at the endpoint.
Amperometric endpoint detection is applicable to a wide variety of
titration reactions (14). The titration does not have to involve a redox
reaction, so long as either the analyte or titrant is electroactive. Thus,
for example, we could monitor the Pb 2+ concentration in a titration of
Pb(N03)2 with Na2S04 to produce insoluble PbS04; the current would
decrease to zero at the endpoint.
Example 4.4 Show how the titration of Pb 2+ with Cr2072-

could be followed amperometrically.
The titration reaction is

2 Pb 2+ + Cr2072. + H20 --t 2 PbCr04(s) + 2 H+
The polarograms of Pb 2+ and Cr2072. are shown schematically
in Figure 4.15a. If we follow the titration with the potential set
at -0.7 V, on the diffusion plateau of the Pb 2+ polarogram, then
the current will fall linearly during the course of the titration
as the Pb 2+ is used up. At the endpoint, the current should be
near zero, but beyond the endpoint, the current should start to
rise again as the excess dichromate is reduced. The resulting

titration curve is shown in Figure 4.15b, An alternative
arrangement for the same titration would be to set the potential
at -0.2 V so that dichromate is reduced but Pb 2+ is not. The
resulting titration curve would then show nearly zero current
up to the endpoint with current increasing beyond.
1.0 Ita)
Cr<£)i2-
0.5 r-
I I
0.0
0.0 -0.2 -0.4 -0.6 o 4 8 ~ ffi
EIV Vol. K2Cr207 solutionlmL
Figure 4.15 (a) Polarograms of Cr2072- and Pb 2 + and (b) amperometric
titration curve for the titration of 25 mL of 0.001 M Pb 2+ with 0.0015 M Cr2072-
atE =-0.7 V.
Constant Composition Experiments
We now consider experiments involving a stationary electrode

where time is an explicit variable. We introduce the time dependence by
writing the diffusion currents as
iDc = KjVt
iDa =-KaIVt
where for a planar electrode
Kc = nFACo*vDolrc
Ka = nFACR*vDR/rc
Equation (4.17) then becomes
E = E 112 + RT 1 Kcffi-i
~ n -----''----
nF i + Ka1fi
or
RTI Kc-i'ft
E = E 112 + - - n ------==----- (4.23)
nF i'ft +Ka
The interrelationship of potential, current, and time (for fixed solution
composition) given by eq (4.23) can also be represented by three-
dimensional surface shown in Figure 4.16. A number of additional
techniques can be understood as excursions upon this surface. l
Figure 4.16 Surface

showing the relation-
ship of current, poten-
tial, and time for a
constant composition
experiment. Note that
the surface begins aL t
> O.
One technique which we found on the i - E - X surface-

polarography-is found here as the current-potential curve at constant
time.
Chronoamperometry
Perhaps the most obvious path to be followed on the surface is the

constant potential curve where the current decays as 11ft. Since current
is measured as a function of tim!:!, the experiment is called
chronoamperometry (16). In practice, an experiment is done by stepping
the potential applied to a cell from an initial value where negligible
current flows to a final value well beyond the half-wave potential. The
1 The description of electroanalytical techniques in terms of an i-E-t surface is due

to Reinmuth (15).
current-time response is then recorded. According to ~7), a plot of i

us. t- 1I2 should give a straight line with slope nFACO*VDohc. Thus if the
concentration and electrode area are known, the method provides a way
of measuring diffusion coefficients.
It is often found that such a plot deviates from linearity. There are
two causes. When the potential step is applied, the electrode double layer
capacitance must be charged. During the charging process, the change
in applied potential is the sum of Lhe potenLial across the double-layer
capacitance and the ohmic potential drop in the solution
i1<l> = Q + icR = Q + R
dQ
C C dt
The differential equation is easily solved to give
Q = C M) [1- GXp i?~]
i. = dQ = i1<l> exp - t (4.24)

C dt R RC
Thus the measured current is the sum of the faradaic current, given by
eq (4.7), and the capacitive charging current, given by eq (4.24). If the
solution resistance isn't too high, the time constant, RC, will be short
and the capacitive contribution to the current a short-lived transient
early in the experiment.
At long times, the diffusion layer may grow to a size comparable to
the electrode dimensions. When this happens, the assumption of planar
diffusion fails and the current has a time-independent component. For
microelectrodes, eq (4.7) fails at rather short times; we will discuss this
situation in more detail in §4.8.
An interesting variant on the technique is called double potential
step chronoamperometry. Suppose that at time 1: after the application of
the potential step, the potential is stepped back to the initial value. Any R
which remains near the electrode then will be oxidized back to 0 and an
anodic current should be observed which decays to zero as the R
molecules in the diffusion layer are consumed. The current during the
first stage of the experiment is given by the Cottrell equation,
i = nFA Co*Wolirt, t < 1: (4.25)
If we assume that the electron transfer is nernstian and that R is stable
during the time of the experiment, it can be shown (see Appendix 4) that
the current during the second stage is
i = nFACo*VDo/rr. (_l __ l)
vt-1: fi
(4.26)
Figure 4.17 shows the potentials and currents for such an experiment as
functions of time. Suppose that the current is measured at time t1 after
the first potential step and again at time t2 after the second step. The
ratio of the two currents will be
~= {f; - ~/ t2t~t
or iftl = t, t2 = 2t,
!1
il
= '1fT2 -1 =-0.293
V
The current after the second step is smaller in magnitude than that
after the first step since some R escapes into the bulk solution. This
measurement provides a good indication of an uncomplicated reversible
electron transfer. If the product R was consumed by a chemical
reaction, for example, the current after the second potential step would
be less than expected from eq (4.26), I i2/ill < 0.293 (see Example 5.7).
-----------
~ ---------.:..:.
if
-----------
Figure 4.17 Applied potential

and current response for a
double potential step chrono·
amperometry experiment. time
Chronopotentiometry
The curve followed on the surface of Figure 4.16 for constant

(nonzero) current is shown in Figure 4.18. This experiment, where the
potential is measured as a function of time at constant current, is called
chronopotentiometry (16,17). In practice, a planar electrode is used and a
constant current is switched on at zero time. The 0 molecules near the
electrode are quickly reduced and the diffusion layer grows. The O/R
ratio at the electrode surface decreases and the potential swings
negative. Eventually, diffusion can no longer supply enough 0 to provide
the required current and the potential heads toward - 0 0 . The time 't
required for this potential swing is called the transition time,
-Fe = Kc1i
In reality of course the potential doesn't go to since there is always
-00
something else in the solution reducible at a more negative potential

(solvent, supporting electrolyte, etc.), but in general there will be a sharp
change in potential at the transition time.
When the diffusion equations are solved using the proper boundary
conditions for the experiment (see Appendix 4), it is found that Kc differs
by a factor of rrJ2 from its value in a constant potential experiment, so
that the transition time is given by
-Fe = nFACo*VrcDo/2i (4.27)
This expression was first derived by Sand in 1900 (18) and is sometimes
called the Sand equation. The shape of the chronopotentiogram for a
nernstian process is given by eq (4.23), which can be rewritten as
RT
E = E't/4 + - I n - -
-Fe - fi (4.28)
nF fi
where E't/4 (= El/2) is the potential when t = 't/4. Since the concentration
of the electroactive species is proportional to -Fe, chronopotentiometry can
be used for chemical analysis.
A variation on this method, called current reversal chronopotentio-
metry, is analogous to double potential step chronoamperometry. A
-0.3
-0.2
.
~
-.j'
~
I
-0.1
ES 0.0
0.1
Figure 4.18 Chronopoten-

tiometry follows a 0.2
constant current path on 0.0 0.5 1.0
the i - t - E surface.
t / '"
cathodic current i is passed through the cell, reducing 0 to R. At time t1
(less than the cathodic transition time), the current is reversed and the
R molecules remaining near the electrode are reoxidized. The potential
then rises, as the OIR ratio increases at the electrode surface. When the
diffusion layer is depleted in R, the potential goes to +00. If R is stable in
solution, the anodic transition time is exactly t1/3. Thus during the
experiment, two thirds of the R molecules escape by diffusion and the
remaining one third are reoxidized at the electrode during the current
reversal phase If R is consumed by a chemical reaction, of course, the
second transition time will be shorter and in favorable cases the rate of
the chemical step can be determined (see Example 5.8)
4.4 CYCLIC VOLTAMMETRY
Another pair of voltammetric methods, linear potential sweep

voltammetry and its extension, cyclic voltammetry, can be described as
an excursion on the surface of Figure 4.16. These methods are very
commonly used and need more than a brief introduction.
Linear Potential Sweep Voltammetry

If the current is measured as the potential is changed linearly with
time, a current-potential curve is produced which corresponds to a
diagonal excursion on the surface of Figure 4.16. The resulting curve,
shown in Figure 4.19, is called a linear potential sweep voltammogram
or peak polarogram The origin of the peak is easily understood.
Consider a negative-going potential sweep; when the reduction potential
of species 0 is reached, the current rises sharply as 0 near the electrode
surface is reduced; the current then becomes diffusion-limited and
begins to fall. When the potential is well past the half-wave potential, the
current falls as lIVt as in a chronoamperometry experiment.
Solution of the diffusion equations for the time-dependent boundary
condition imposed by a rapid potential scan is complicated (19). For a
reversible process at a planar eJectrode, the following integral equation
is obtained (see Appendix 4):
nFA fitl50 Co* =

1 + ~e(t)
t~
)0 yt-t
dt (4.29)
where ~ = VDolDR and

e(t) = exp nF(Ei - vt - EO)
RT
where Ei is the initial potential and v is the potential scan rate in V s-l.
0.2 0.1 0.0 -0.1 -0.2

(E -E1/2)1V
Figure 4.19 Linear potellLial sweep voltammogram. (a) Path on i - E - t
surface; (b) current potential curve.
It is customary to define a dimensionless current function

X(crt) = i(crt)
nFACo*Yrr.Docr
where cr = nFu / RT. In terms of the dimensionless function, eq (4.29) is
1 _
1+~8(t) -
iat
0
X(z)
Ycrt-z
dz (4.30)
The current function can be computed by numerical integration of eq

(4.30), and Nicholson and Shain (19) provide tables of the results.
The peak current for a reversible process at a planar electrode is
found to be
ip = 0.446 nFACo* ~ nFuDo (4.31)
RT
where the parameters are in SI units. The current peak is somewhat
displaced from E1/2,
Ep::: E1/2 - 1.11 RTlnF (4.32)
so that at 25°C, the peak potential is 2S.51n mV past the half-wave
potential. (The half-peak potential comes 2S.01n mV before E1I2.) For a
reversible process, the shape of the curve and the peak potential are
independent of scan rate, but the peak current is proportional to ..fff.
§4.4 Cyclic Voltammetry 185
Since the peak current is proportional to concentration, a linear
potential sweep voltammogram can be used for analytical purposes.
However, there are experimental complications which reduce accuracy.
Because of the changing potential, the capacitive charging current may
be quite significant in a linear potential sweep experiment. The charge
stored in the double-layer capacitance is
Q = CrrAi\CP
where Cd is the double-layer capacity in f m- 2 ; thus the charging current
is
ic = dQldt = CdAv (4.33)
and this may be a significant fraction of the total current.
Example 4.5 Compute the faradaic and capacitive currents

for n = 1, CO* = 1 mol m- 3 (1 mM), DO = 10-9 m 2 s- 1, Cd = 0.5 F m 2 ,
A = 10-6 m 2 (1 mm 2 ), and T = 298 K for v =0.1 and 10 V s-l.
Substituting these parametcrs into ceq (4.31), we get

ip = 2.7 and 26.8 ~
for v = 0.1 and 10 V s-l, respectively. Substituting Cd, A, and v in

eq (4.33), we have
ic = 0.05 and 5.0 ~
Thus at the slower scan rate, the capacitive current accounts

for about 2% of the total current, but the fraction rises to about
16% at 10 V s·l. If the concentration were lowered to 0.1 mM,
the peak current drops by a factor of 10 but the capacitive
current is unchanged. The total current then is 16% and 65%
capacitive for v = 0.1 and 10 V s-l, respectively. For low
concentrations and/or high scan rates, the capacitive current
is a serious interference in linear potential sweep experiments.
Comparing the results of Examples 4.1 and 4.5, we see that linear
potential scan peak currents can be quite large. Because of this, ohmic
potential drop in the solution is often a problem. Since the cell potential
includes the solution iR drop and the iR drop varies as the peak is
traversed, the indicator electrode potential is not linear in the applied
potential. Thus the observed peak potential becomes a function of scan
rate. In practice, the revcrsible peak potential usually can be extracted
from experimental data by measuring the peak potential at several scan
rates and extrapolating to v == 0, but the absolute peak current is more
difficult to correct.
Cyclic Voltammetry
The analog of double potential step chronoamperometry and

current reversal chronopotentiometry is possible for linear potential
sweep voltammetry. A triangular-wave potential is applied to the cell so
that the indicator electrode potential is swept linearly through the
voltammetric wave and then back again as shown in Figure 4.20a. On
the forward scan, the current response is just the linear potential sweep
voltammogram as 0 is reduced to R. On the reverse scan, the R
molecules near the electrode are reoxidized to 0 and an anodic peak
results. The current-time curve, shown in Figure 4.20b, is usually folded
at the switching point (the vertical dashed line in the figure) so that, in
effect, current is plotted vs. potential. The resulting curve, shown in
Figure 4.20c, is called a triangular-wave cyclic voltammogram.
Detailed analysis of cyclic voltammetry also leads to integral
equations which must be solved numerically (19). However, we can get a
qualitative feeling for the experiment without a lot of mathematics. If
the initial scan is carried far beyond the cathodic peak so that the
diffusion layer is very thick and the cathodic current has decayed nearly
to zero, then the concentration of R at the electrode surface is equal to
Co* (within a factor of YDoIDR). Thus the amount of R available for
oxidation on the reverse scan is the same as the 0 available on the
forward scan and the current peak has the same shape and magnitude
as on the forward scan, but reflected at E = E 112 and changed in sign.
time potential
Figure 4.20 Cyclic voltamrnetry: (a) potential as a function of time; (b)
current as a function of time; (c) current as a function of potential.
:m 100 0 -100 -200 200 100 0 -100 -200

(E - E l/2)1mV (E - E l/2)1mV
Figure 4.21 Cyclic voltammogram Figure 4.22 Cyclic voltammogram
for a reversible one-electron process showing the effect of multiple scans.
showing the effect of the switching The first cycle is shown as a solid
potential. The bottom curve shows line, the second as a dashed line,
the expected anodic trace when the and the third as a dotted line.
potential is negative long enough to
completely polarize the electrode
The dashed (current-time) curves
show the response which would have
occurred had the potential scan
continued in the negative direction.
When the switching potential is less negative such that the cathodic
current is still significant at the switching point, the diffusion layer is
thinner, the R concentration falls to zero more rapidly with distance
from the electrode, and the resulting anodic peak is smaller. However, it
turns out that, if the anodic peak current is measured from a baseline
equal to the cathodic current which would have flowed at the time of the
anodic peak had the potential scan continued in the negative direction
(rather than from the zero current baseline), then the anodic-to-cathodic
peak current ratio is exactly 1 (for a reversible process uncomplicated by
capacitive current and iR drop). This result is shown schematically in
Figure 4.21 with cyclic voltammograms computed using the numerical
methods of Nicholson and Shain (19).
Provided that the process is reversible and is uncomplicated by
solution iR drop, Epc -El/2 = -28.5/n mV and Epa -E1I2 = 28.5/n mV (at
25
-0.0 -0.1 -0.2 -0.3 -0.4 0 5 10 15 25

EIV
~(c).. ~o~! •. •-:.:.'.~..............

~ ~........................ .. . . . .
:~.~.:.:.~.~.:~ ~.~~
'"
. . . . ....
o ~~~~~~~~~~~~
o 5 ill 15 ID 25 0 5 ill 15 ~ 25
o 5 10 15 25 0 5 10 15 25
xillm xillm
Figure 4.23 Concentration profiles for 0 (solid lines) and R (dashed lines) at
various points on a cyclic voltammogram. Concentrations computed using a
dig!tal simulation technique (see Appendix 5) for v = 1 V 5- 1 , DO = DR = 5 x
10- 10 m 2 s- 1.
25°C). Thus E1/2 = (Epc + Epa)/2 and Epa - Epc = 57.Dln mY. This peak
separation is often used as a criterion for a reversible process. l
1 Nernstian reversibility should not be confused with chemical reversibility

evidenced by the appearance of a peak on the reverse sweep.
Since the current direction is different for the two peaks, the sign of
iR is also different so that ohmic potential drop in the solution increases
the peak separation. Again, extrapolation to zero sweep rate allows
corrections to be made. Since the sign of the potential sweep changes at
the switching point, the sign of the capacitive current also changes.
Thus a current discontinuity is usually observed at the switching point.
The response on the second and subsequent cycles of a cyclic
voltammogram is qualitatively similar to that of the first cycle but with
current peaks somewhat reduced in amplitude. This effect is shown in
Figure 4.22. The reason for this decrease is that the concentration
profiles of 0 and R do not return to initial conditions at the completion of
a cycle. This is most easily seen by following these concentration us.
distance profiles over the course of the first cycle as shown in Figure
4.23. Initially, Co(x) = Co* and CR(X) = o. As the cathodic peak is
traversed, 0 is depleted at the electrode surface and the R which is
formed diffuses away from the electrode (curves a-c of Figure 4.23). On
the reverse scan, R is oxidized to 0 but there remains a region where Co
< Co* and CR > 0 (curves d-f of Figure 4.23). On subsequent cycles, the
response is as if the bulk concentration of 0 were reduced. On multiple
cycles, a steady state is eventually n::achcd with damped concentration
waves propagating out into the solution.
Adsorption Effects in Cyclic Voltammetry
It is fairly common to find that one or both members of an electrode

couple adsorb on the electrode surface. Although it can be an interesting
subject in its own right, adsorption is usually an unwanted
complication. Adsorption often can be reduced or eliminated by
changing the solvent or electrode material. Adsorption effects in cyclic
voltammetry are rather dramatic and are usually easy to recognize. We
consider here one simple case.
Suppose that 0 and R are both adsorbed on the electrode surface
with surface concentrations 10 and lR (with units mol m- 2 ) and that the
electrode process consists of the complete reduction of 0 to R on the
forward sweep and the oxidation of R to 0 on the reverse sweep. The
current then is not diffusion-controlled but is limited by the amount of
adsorbed material.
If the electron-transfer process is reversible, the surface
concentrations are related by the Nernst equation,
lo(t) = exp n F[E(t) - Es 0] = 8(t)
lR(t) RT
where Eso is the formal potential of the adsorbed couple and E(t) = Ei -
ut. If the total surface concentration is constant,
10(t) + lR(t) = 1*
the individual surface concentrations are time (and potential)
dependent,
ro(t) = e(t)r* rR(t) = r*
1 + 9(t) 1 + 9(t)
The rate of conversion of 0 to R is related to the current by
i aroCt) arR(t)
nFA ;:: - ----at = ----at
_i_ = _ r* a9 _ nFr*v 9
nFA (1 + 9)2 at - RT (1 + 9)2
Thus the current is given by
i = n 2F 2Ar*v 9 (4.34)
RT (1+9f
Equation (4.34) predicts a current peak at E = Eso (9 = 1) with peak
current proportional to the number of moles of 0 and R adsorbed (Ar*)
and to the scan rate v. The peak is symmetrical on the potential axis, as
shown in Figure 4.24, with a width at half-height, ~p = 3.53 RT/nF
(90.6In m V at 25°C). When the sweep is reversed, v changes sign and the
current is inverted so that the reverse sweep signal is the mirror image
of that obtained on the forward sweep.
In this extreme case, adsorption peaks are easily distinguished
from ordinary diffusion-controlled peaks by the symmetric shapes and
scan-rate dependence. The situation is more complicated when the
solution species also participate in the electrode process and/or when
only one member of the couple is adsorbed. In these cases adsorption
peaks may be seen along with the normal diffusion-controlled peaks.
When only the substrate 0 is strongly adsorbed, the adsorption peak
follows the diffusion-controlled peak (a postpeak); when R is adsorbed,
the adsorption peak precedes the diffusion-controlled peak (a prepeak).
See further discussion of adsorption effects in the context of
polarography in §4.5. Many of these cases were examined in detail by
Wop schall and Shain (20) and the interested reader is referred to their
work and to a more complete discussion by Bard and Faulkner (B12).
Derivative Cyclic Voltammetry
Cyclic voltammograms are particularly useful in qualitative

studies of electrode processes (we will see many examples of such uses
in Chapter 5), but the technique is less well suited for quantitative
measurement of potentials or currents. Because of the rounded peaks,
Figure 4.24 Computed cyclic

voltammogram for a
reversible, one-electron O/R
couple where both 0 and Rare
adsorbed on the electrode 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3
surface. EIV
peak potential meaurements are usually good at best to only ±5 mY.

Because of the contribution of capacitive charging current, the faradaic
contribution to a peak current is difficult to measure accurately. The
peak current for the reverse trace is particularly difficult to measure
because of uncertainty in the location of the baseline (see Figure 4.21).
One way around these problems is to record the first derivative of
current with respect to potential as shown in Figure 4.25. Since the
derivative curve crosses the baseline at the current peak, the peak
potential can be read considerably more accurately, to ±O.1 mV in
favorable cases. The contribution of capacitive charging current is more
or less constant over a trace so that it has little effect on diidE. The
derivative curve returns very nearly to the baseline after a peak. Thus
the faradaic contribution to (di/dE)peak, which is proportional to
concentration, is easily and accurately measureable. Because the
derivative curves return to the baseline and are somewhat sharper,
resolution of closely spaced peaks is better than in a conventional cyclic
voltammogram.
There are some drawbacks to this method: (i) the instrumentation
is necessarily more complex; (ii) distortions resulting from ohmic
potential drop are still present and may even be amplified by
differentiation; and (iii) the curves are somewhat more difficult to
interpret qualitatively. Nonetheless, for quantitative work, first
derivative presentation has significant advantages. See Parker (21) for
further details.
Although analog differentiation circuitry based on operational
amplifiers is available, the signal-to-noise ratio is often poor. Parker (21)
192 Vol tammetry of Reversible Systems
describes the use of a tuned amplifier to differentiate the current.

Another way of differentiating the current in a cyclic voltammetry
experiment is to modulate the applied potential with a small sinusoidal
component and to detect the a.c. component of the current. This method,
called a.c. cyclic voltammetry, is discussed in §6.4.
Figure 4.25 Reversible

cyclic voltammogrRm
(dashed curves) compared
with its first derivative 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3
(solid curves). (E -E1I2)N
Semi-derivatives and Semi-integrals
Although differentiation of a linear scan voltammogram or cyclic

voltammogram sharpens the signal shape and provides a means of
accurately measuring the peak potential, in a sense, differentiation has
gone too far and has produced a signal shape which is asymmetric in a
way opposite to that of the cyclic voltammogram. Another method of
signal-processing, called semi-differentiation, gives a much more
symmetrical signal shape with peaks corresponding to the half-wave
potential.
Consider a time-dependent function, f(t), and the convolution
integral, met)
met) =.l t tt't) d't (4.35)

Vi Jo yt-'t
The convolution of met) is equivalent Lo the simple integral of ttt),
_1
fit
it
0
met) dt
vt-t
=
it
0 ret) dt
In other words, m (t) is mid-way between the function f(t) and its
integral. Accordingly, m (t) is called the semi-integral of f, written
formally as
(4.36)
The semi-derivative of f then is

e(t) = dm(t) = d ll ?y'rt) (4.37)
dt dt 112
Thus, for example, if ret) = sin rot,
met) = IroC sine rot - rr/4)
e(t) = Vw sin(rot + rr/4)

d 1I2e(t) . df(t)
- - - = ro sm(rot + rr/2) = ro cos rot = - -
&~ &
Consider now eq (4.29), the integral equation leading to the current
in a linear potential sweep experiment. If i(t) is the (now known)
current, its semi-integral is given by the left-hand side of the equation.
But this function, plotted vs. time or potential, has the same shape as a
steady-state voltammetric wave, cq (4.8). Thus, if a steady-state wave is
the semi-integral of a linear sweep voltammogram, the latter is the
semi-derivative of a wave, and the semi-derivative of a linear sweep
voltammogram has the shape of the first ilerivative of a steady-state
wave, as shown in Figure 4.26.
The semi-derivative curve clearly has advantages: it is more
symmetrical than the cyclic voltammogram, and the peaks, which are
sharper, occur at E = E1I2. It is also less subject to interference from
capacitive charging currenL than the cyclic voltammogram, though the
removal of charging current effects is less complete than in a derivative
cyclic voltammogram. Distortions from ohmic potential drop remain
and, as for derivative cyclic voltammograms, are slightly exaggerated
compared with simple cyclic voltammograms.
Semi-differentiation of cyclic voltammograms was introduced by
Goto and Ishii (22) and further developed by Oldham and coworkers
(23,24). In practice, the current output of a potentiostat is processed
through an active filter based on operational amplifier circuits.
....
"'"' (a)
1...,
Q)
~
.';'
's
Q)
rn
0.2 0.0 -0.2 0.2 0.0 -0.2 0.2 0.0 -0.2

(E -E l / 2 )N (E -El/2 )N (E -E l / 2 )N
Figure 4.26 Reversible cyclic voltammogram (b) and its semi-integral

(a) and semi-derivative (c).
4.5 POLAROGRAPHY
The Dropping Mercury Electrode
Shortly after World War I, Heyrovsky discovered that the dropping

mercury electrode (d.m.e.) could be used to measure reproducible
current-potential curves. The technique, which he named polarography,
flowered in HeyrovskY's hands and become the source of many of the
subsequent developments in electroanalytical chemistry (25,Dl,D4,D5,
DI2).
In a classical d.m.e. such as that in Figure 4.10, mercury flows
through the capillary by gravity and a drop grows until its weight
exceeds the surface tension force holding it to the capillary. Depending
on the length and bore of the capillary, the height of the mercury head
above the capillary, and the mercury-solution interfacial tension (see
§2.5), drop lifetimes can range roughly from 1 to 10 seconds using this
arrangement. Modern dropping mercury electrodes are usually
furnished with a drop dislodging mechanism and a timer so that drop
times are constant and reproducible. A more recent innovation is the so-
called static mercury drop electrode, where the mercury flow rate is
controlled by a pump; this device will be discussed in §4.6.
When a drop falls off the capillary, it stirs the solution so that each
new drop begins life in contact with a nearly homogeneous solution. In
effect the experiment is repeated over and over again. If the current
through a d.m.e. is measured at the same time in the lives of successive
drops (at the time of drop fall, for instance), we can think of the
experiment being done at constant time. In ordinary polarography, the
Sec. 4.5 Polarography 195
electrode potential is scanned slowly with time so that the potential is
nearly constant during one drop life. Thus each successive drop
measures the current at a slightly different potential and the current-
potential curve, the polarogram, is generated. A major disadvantage of
the d.m.e. is that mercury is much more easily oxidized than electrode
materials like ~latinum or gold. The standard potential for the
reduction of Hg2 + to the metal is 0.8 V, but with even as weak a ligand
as CI-, the oxidation shifts to ca. 0.24 V. Thus polarography is largely
limited to the negative end of the potential scale and thus has been used
mostly to study reduction processes. 1
An exact solution of the diffusion problem for a d.m.e. is not easy. If
the electrode were simply a sphere, we could use the results we obtained
in §4.1 for spherical diffusion. However, the electrode expands into the
solution and this makes the problem considerably more complicated.
Furthermore, the mercury drop is really not a perfect sphere but is
slightly distorted and (more important) is shielded on one side by the
capillary. Finally, the stirring which occurs when a drop falls is not
complete and a rigorous model should take account of the resulting
nonuniform concentration distribution.
A simple approach to the problem of diffusion to the d.m.e. is to use
the results of the planar diffusion problem with two multiplicative
correction terms. The first correction accounts for the changing
electrode area, the second for the expansion of the electrode surface into
the solution. If we assume that the drop is a sphere of radius ret) and
that mercury flows through the capillary at constant rate u (mass per
unit time), then the mass of the drop at time t after the previous drop fell
is m = ut and the volume is
vet) = 41tr 3 = ut
3 d
where d is the density of mercury. The drop area then is
A(t) =41tr2 =41t ( 3ut )2'3 (4.38)
41td
The effect of the electrode expanding into the solution is to stretch the
diffusion layer over a larger area, in other words to make XD smaller.
The thinner diffusion layer results in a larger concentration gradient
and thus in a larger flux and current. The net effect is as if the diffusion
coefficient were increased by a factor of 7/3.
r
Substituting eq (4.38) in eq (4.7) and replacing Do by (7/3)Do gives
iD = nF [41t (!:~ 3
J Co* C~: (2
1 The convention that cathodic current is taken as positive results from this feature;
polarographic currents are usually cathodic and much of the theory of voltammetry
was developed with polarography in mind.
or, including numerical values for F and d (13.6 x 103 kg m- 3 ), we have

in = 706 nDo1/2Co*u2/3tl/6 (4.39)
in will have units of amperes (A) when Do is in m 2 s- 1 , Co* is in mol m- 3
(mM), u is in kg s-l, and t in s. Equation (4.39) was first derived by Ilkovic
in 1934 (26) and is usually called the Il1wuic equation. Ilkovic followed a
somewhat more rigorous route than we did but did not deal with all the
problems discussed above. Others have made serious attempts to solve
the problem correctly but the complexity of the resulting equations is not
wholly compensated by increased accuracy. The basic functional form of
eq (4.39) remains unchanged in the more complete treatments, so we
will not pursue the matter further. See, for example, Bard and Faulkner
(B12) for further details.
Since the current at a d.m.e. is related to that at a planar electrode
by the same scale factors for most electrode processes, expressions such
as eqs (4.10), (4.16), and (4.17), which were derived for linear diffusion to a
stationary planar electrode, are also applicable to the d.m.e.
One of the significant results of the Ilkovic equation is that, as
shown in Figure 4.27, the current increases with time rather than
decreasing as with a planar electrode. This means that the current at
drop fall is the maximum current during the drop life. Furthermore,
the time rate of change of the current, di/dt, decreases with increasing
time and is at a minimum at drop fall. These considerations make the
measurement of the current at or just before drop fall relatively easy and
accurate.
Figure 4.27 Faradaic current

(dashed lines), capacitivE'
.... \
.......................... \............................ ..........................
current (dotted lines), and total
current (solid lines) through a
d.m.e. for Co* = 10. 4 M. o 1 2 3
The favorable time dependence of the faradaic current for a d.m.e.

has a price. Since the electrode area changes with time, the double-layer
capacitance, which is proportional to the electrode area, is also time
dependent. The charge stored in the capacitance is
Q = CA~<l>
where C is the double-layer capacity (capacitance per unit area). Since Q
is time dependent, there must be a capacitive charging current
ic = dQldt = C (dAldt) ~<l>
Differentiating eq (4.38), we obtain

i = C 8rc
c
(JlliL)2'::l rJ/3 ~<l>
3 4rcd
ic = 0.00566 C u 2/3 t-1/3 ~<l> (4.40)
Although ~<l> also changes with time during the lifetime of a drop, the
scan rate is normally much slower than in cyclic voltammetry
experiments so that this contribution to capacitive charging current is
negligible. The capacitive current is shown as a function of time in
Figure 4.27, together with the total current.
Example 4.6 Compute the faradaic and capacitive currents

at t = 0.1 and 5 s assuming that u = 1 mg s-l, an electrode-
solution potential difference of 1 V, and a double layer capacity
of 0.1 F m- 2 . Assume a one-electron reduction with Do = 10- 9
m 2 s- 1 and Co* = 1 mM (1 mol m· 3 ).
Substitution into eqs (4.39) and (4.40) gives

iD = (2.23 x 10- 6 ) t1/6 amperes
ic = (4.66 x 10-8 ) t-1/3 amperes
Thus iD = 1.5 and 2.9 ).lA at t = 0.1 and 5 s, respectively. At the
same times, ic = 0.12 and 0.03 ).lA.
For millimolar solutions of electroactive materials, the maximum

current occurs at the end of the drop life,l and this current is mostly
faradaic. However, when the concentration of electroactive material is
reduced, the capacitive current remains the same and the ratio of
faradaic to capacitive current is smaller. The practical lower limit of
concentration in polarography is about 10 ).lM since at that level the
faradaic and capacitive currents are comparable in magnitude.
1 The frequency response of the current-measuring circuitry is usually too slow to

catch the short-lived transient in the capacitive current at drop fall.
Example 4.7 Suppose that diffusion-controlled current flows

at a dropping mercury electrode lu = 1 mg s-l, td = 5 s) for 10
minutes and that 0 is reduced by one electron to R. If the cell
contains 25 mL of 1 mM 0 (Do = 10-9 m 2 s- 1 ), what fraction of the
o has been reduced?
The charge passed during one drop can be determined by
integrating eq (4.39):
Q= E iD dt =~x 706 nYDo CO*u?J3t 7/6
Substituting the parameters, we have

Q = 1.25 x 10-5 C
In 10 minutes, 120 drops fall and so the total charge
transferred is 1.50 x 10-3 C. Dividing by the Faraday constant,
we have 1.56 x 10-8 mol, which is 0.062% of the total 0 originally
present. This result is Lypi.:al of voltammetric techniques. The
net amount of elecLl'olysis is usually extremely small.
Effects of Adsorption on the Mercury Drop
An important advantage of the d.m.e compared with solid

electrodes is that each successive drop presents a clean surface to the
solution. Thus products of the electrolysis or subsequent chemical steps
do not accumulate at the electrode surface beyond the amounts produced
during one drop life. The effects of adsorption are thus more
reproducible on a d.rn.e. than on solid electrodes and are usually easier
to study and to understand.!
Suppose that the product of the electrode process is adsorbed on the
mercury surface such that the first monolayer is strongly adsorbed but
subsequent layers are either weakly held or not adsorbed at all. The
electrode process then consists of two alternative reactions:
o + n e- f:! Rad
o + n e- f:! Rsol n
If the standard potential for the second reaction, where R remains in
solution, is E2° and the free energy of adsorption is ~Gado, then the
standard potential for the first reaction is
1 The original work on adsorption effects in polarography was done by Brdi&a (27)
in the 1940's.
(4.41)
Thus, if AGado < 0, the two reactions are expected to be distinguished by
different half-wave potentials. If the bulk concentration of 0 is small
enough that less than monolayer coverage by R is achieved during one
drop life, we expect a wave at E10 with a limiting current given by eq
(4.39). If the drop is completely covered before it falls, then we expect the
current to be limited by the available adsorption sites. Excess 0 then
remains unreduced at the electrode surface when the potential is in the
vicinity of Elo and a second wave is observed at E2°. The total diffusion
current of the two waves is limited by diffusion and is given by eq (4.39).
The second wave, of course, is the expected one for a process
uncomplicated by adsorption. The first wave is called an adsorption
prewave.
A very similar phenomenon occurs when the substrate 0 is
adsorbed on the mercury drop (we assume that R is not then adsorbed).
For low bulk concentrations of 0, the electrode will be partially covered
and the electrode process will be
Oad + n e- f:! R
When Co* is big enough that complete coverage IS achieved, direct
electron transfer may be possible:
Osoln + n e- f:! R
The potential of the first process will be more negative (if AGado < 0), but
the main wave will occur only at higher concentrations. The more
negative wave is called an adsorption post-wave. The appearance of
adsorption pre-waves and post-waves are shown in Figure 4.28.
Adsorption pre-waves and post-waves can be distinguished quite
easily from ordinary electron-transfer processes. In addition to seeing
the normal wave at higher substrate concentrations, heights of
adsorption waves have a unique dependence on drop size and the
variation of current during the drop life is different from the normal
behavior. For further details, a specialized book on polarography should
be consulted (D1,D4,D5).
Polarographic Maxima
Throughout this chapter, we have assumed that diffusion is the

only transport process involved in delivery of electroactive material to the
electrode. There are several circumstances, encountered most
frequently with the dropping mercury electrode, where convection also
makes a significant contribution to mass transport. As we will see,
convective mass transport is usually important in a limited potential
range, and, since convection necessarily increases the current, the effect
is a maximum in the current-potential curve.
2.0
(a)
iliD2
1.5
1.0
1.0
0.5
0.5
0.0
-0.3 -0.4 -0.5 -0.6 -0.7 -0.3 -0.4 -0.5 -0.6 -0.7
EIV EIV
Figure 4.28 Polarograms for various substrate concentrations, CO*, where
Ca) the product Rand Cb) the reactant 0 are adsorbed on the mercury drop.
The curves correspond to CO*/CO' = 0.5, 1.0, 1.5, and 2.0, where CO' is the
concentration leading to monolayer coverage at the time of drop fall.
Three kinds of polarographic maxima have been identified (28).

Type I maxima are the most conspicuous and thus most commonly
recognized. The polarographic current, instead of increasing smoothly
with potential through the wave, seems to go "haywire," often in the
rising portion of the wave. Such erratic currents usually subside, often
abruptly, on the diffusion-limited current plateau. Although the
behavior appears erratic, it is usually quite reproducible. Type I
maxima come about through variations in interfacial tension over the
drop surface. These variations in turn reflect a variation in potential
difference across the electrode-solution interface which arises through
nonuniform current density over the surface of the mercury drop. The
current density at the bottom of the drop is higher than at the "neck,"
where the drop is partially shielded by the capillary tube. Maxima of the
first kind can be reduced by increasing the solution conductance (higher
supporting electrolyte concentration), by decreasing the current (lower
concentration of electroactive species), or by stabilizing the electrode-
solution interface with a surfactant such as gelatin or Triton X-lOO
(polyethylene glycol p-isooctylphenyl ether).
A second kind of polarographic maximum can occur through
motion within the mercury drop. If mercury emerges from the capillary
fast enough, the new mercury will flow to the bottom of the drop and
then up the sides. The drop may be thought of as growing longitudinally
rather than radially. The circulation of mercury inside the drop will
tend to drag the surface layer along, thus leading to convective mixing of
the solution. Type II maxima are less dramatic in their effect; the
convective contribution to mass transport is only weakly potential
dependent, and the effect may pass unrecognized in many cases. Type
II maxima depend on the mercury mass flow rate u and are most easily
recognized by a departure from the dependence of current on flow rate
predicted by the llkovic equation.
Type III maxima arise from nonuniform adsorption of surfactants
which lead to variations in interfacial tension and thus to convective
flow of solution in the vicinity of the mercury drop. Like Type I maxima,
these effects can be reduced or eliminated by adding a competing
surfactant.
Polarographic maxima are more common in aqueous solution
polarography than when organic solvents are used, presumably because
the mercury surface is hydrophobic and therefore susceptible to
surfactants in aqueous media, whereas many organic solvents are
themselves somewhat adsorbed on the mercury surface.
4.6 POLAROGRAPHIC VARIATIONS
Since the introduction of polarography in the 1920's, many

instrumental modifications have been developed to improve the form of
the polarogram and to increase sensitivity and/or resolution. We will
discuss five of these improvements in this section.
Sampled D.C. Polarography
The simplest modification to polarography is called sampled·d.c. or

tast polarography. In this mpthoo, a voltage ramp is applied to the cell
as in ordinary polarography, but instead of measuring the current
continuously, it is measured fOl unly <l shull time late in the drop life. A
voltage proportional to the measured current is then applied to the
recorder and held until the next current sample is obtained. The d.m.e.
must be synchronized to the timing of the current-sampling circuitry so
that the sample occurs at the same relative time during each drop life.
Sampled d.c. polarography suppresses display of the current early in the
drop life, when there is a large capacitive component, and focuses on the
current late in the drop life, when the faradaic component is larger.
This stratagem does nothing for sensitivity but does slightly improve the
apparent signal-to-noise ratio, making polarograms somewhat easier to
analyze. A typical sampled d.c. polarogram is shown in Figure 4.29b.
Pulse Polarography
A weakness of both ordinary d.c. and sampled d.c. polarography is
that these techniques allow faradaic current to flow during a time in the
drop life when the capacitive current is large. A considerable
improvement is possible with a technique called pulse polarography,
introduced by Barker in 1960 (29). A series of voltage pulses is applied to
the cell as shown in Figure 4.30; the drop time is synchronized to the
pulses which increase in amplitude at a rate comparable to the ramp
voltage used in ordinary polarography. The total current response from
such a pulse train still has a large capacitive component, but this is
largely suppressed by sampling the current for a short time late in the
pulse life after the capacitive current has mostly decayed. The sampled
current is converted to a voltage which is held and applied to the
recorder until the next sample. The appearance of a pulse polarogram,
Figure 4.29c, thus is virtually indistinguishable from a sampled d.c.
polarogram.
As we saw in the derivation of the llkovic equation, the current
varies with time because of the time-dependent electrode area and
because of the growth of the diffusion layer with time. In pulse
polarography, essentially no current flows until late in the drop life, so
that the diffusion layer is much thinner at the time the current is
actually measured. In particular, we expect the measured current to
depend on the times t1 and t2 (see Figure 4.30):
i oc t1213t2-1/2
Time t1 is the age of the drop when the current is measured (t1 thus
determines the drop area) and t2 is the time between application of the
voltage pulse and current measurement (t2 thus determines the
diffusion layer thickness). Suppose that the drop time is 5 s, that the
pulse is applied in the last 50 ms of the drop life, and that the current is
measured during the last 10 ms of the pulse. Then t1 = 4.99 s, t2 = 0.04 s
and
t12/3 t2-1/2 = 14.6
In sampled d.c. polarography, t1 =t2 =4.99 s, so that
t12/3 t2-1/2 = 1.31
Thus the measured current in a pulse polarogram is about 10 times

bigger than in a sampled d.c. polarogram. The capacitive contribution to
the current is about the same in the two techniques so that the signal-to-
noise ratio is improved by about a factor of 10 and detection limits are
about 10 times lower.
§4.6 Polarographic Variations 203
(a) (b)
1.0 -
illlA
0.5
0.0 f---------------
I
(c) (d)
o -----------
-0.4 -0.6 -0.8 -0.4 -0.6 -0.8
EIV EIV
Figure 4.29 (a) D.c., (b) sampled d.c., (c) pulse, and (d) differential pulse
polarograms of 0.1 mM Cd(N03)2 in 0.10 M aqueous KCI. Potentials vs.
s.c.e., drop time 2 s, scan rate 2 mV s·l, differential pulse height 10 mV.
pulse train
Figure 4.30 Pulse train for pulse

polarography: t1 is the drop time
(0.5 . 5 s), t2 is the pulse width
(ca. 50 ms), and t3 is the current voltage
sampling time (ca. 10 ms). pulse
204 Vol tammetry of Reversible Systems
Geoffrey C. Barker (1915- ) was a member of the scientific staff of the

Atomic Energy Research Establishment, Harwell (England). He is best
known as an innovative inventor of electrochemical instrumental
techniques. He is credited with square wave polarography (which led to
the development of differential pulse polarography), normal pulse
polarography, anodic stripping voltammetry, and several other
techniques.
Pulse polarography gives current-potential curves which are

qualitatively similar to those from sampled d.c. polarography when the
reduced half of the couple is absent in bulk solution (as we have
assumed above). When CR* "* 0, however, and anodic current flows at
the initial potential, the situation is rather more complicated (30). The
complications can be turned to advantage, however, and Osteryoung and
Kirowa-Eisner (31) have described a technique, called reverse pulse
polarography, in which the pulse sequence is virtually a mirror image of
that shown in Figure 4.30, i.e., the "resting potential" is on the diffusion
plateau of the wave and a series of positive-going pulses of linearly
decreasing amplitude is applied.
Differential Pulse Polarography
Another pulse technique which is capable of good signal-to-noise

ratios employs the pulse train shown in Figure 4.31. Small-amplitude
voltage pulses are superimposed on a ramp voltage. Current is sampled
twice during a drop life, once just before the pulse and again late in the
pulse life. These currents are converted to analog voltages, subtracted,
and applied to the recorder until the next drop. With this technique,
called differential pulse polarography, a curve is produced which
resembles the first derivative of the d.c. polarographic wave. In the limit
that the pulse amplitude I1E goes to zero, the curve is exactly the first
derivative, but then the signal goes to zero as well. Pulse amplitudes of
5-50 mV are used in practice. If the differential current is plotted vs. the
ramp voltage, the peak potential is displaced by -DE12 from El/2; that is,
a negative-going pulse (!J.E < 0) shifts the peak to more positive
potentials, a positive-going pulse shifts the peak to more negative
potentials. A typical differential pulse polarogram is shown in Figure
4.29d.
The differential pulse peak current is proportional to the bulk
concentration of the electroactive species, just as the diffusion current in
d.c. polarography. However, detection limits are much lower, down to 0.1
11M or less in favorable cases. The improvement stems largely froni
suppression of the capacitive charging current.
pulse train
Figure 4.31 Pulse train, voltage

and current pulses for different- voltage
ial pulse polarography. pulse
Because of the first derivative-like presentation of the output, the

signal returns to the base line after a peak and successive current peaks
for a multi-component system are somewhat better resolved than are d.c.
polarographic waves. This effect is demonstrated in Figure 4.32, where
differential pulse polarograms are plotted for a two-component system
with C01* =C02*' CRl* =CR2* =0, nl =n2 = 1, and !lE112 = 0.1, 0.2, 0.3, and
0.4 V. Comparison with Figure 4.4 leaves little doubt that the analytical
accuracy of differential pulse polarography is considerably better than
that of d.c. polarography when several components give closely spaced
waves.
1.00
i/i p
0.75
0.50
Figure 4.32 Computed

differential pulse polaro· 0.25
grams for a two·component
system, showing the effect
of half-wave potential
separation, tJ.E1I2:: 0.1 0.00
(solid line), 0.2 (dotted -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
line), 0.3 (dashed line), and
0.4 V (dot·dash line). (E - E 1I2 avg )N
Square Wave Polarography
The time between pulses in differential pulse polarography (t1 in

Figure 4.31) is set equal to the d.m.e. drop time. Thus the pulse repetition
rate is on the order of 0.1-1 Hz. If the pulse repetition rate is increased to
30 Hz, many pulses can be applied during one drop life and a scan over
several hundred millivolts can be achieved during the lifetime of one
drop. This approach is adopted in square wave polarography, where the
positive- and negative-going pulses of Figure 4.31 are equal in width (t2 =
t 1/2). In effect, the poLential applied to the cell is a square wave-
modulated potential scan. If the square wave frequency is 30 Hz and
successive pulses advance the d.c. potential by 5 mV, the d.c. potential
scan rate is 150 mV s-l, comparable to typical scan rates in cyclic
voltammetry. The current measurement strategy is the same as in the
differential pulse technique: the current is measured at the end of
successive positive-going and negative-going pulses, subtracted, and
plotted as a function of the d.c. potential. The resulting square wave
polarogram thus closely resembles a differential pulse polarogram (32).
The instrumentation required for square wave polarography is
rather more sophisticated; the potentiostat must have very good
frequency response and the output must be stored in a computer or
recorded on an oscilloscope since an ordinary strip-chart recorder would
not be able to keep up with such rapid data collection. In practice, a
microprocessor is used to generate the wave form (the d.c. scan would
then be a "staircase" rather than a ramp) and to collect the current
signal. This arrangement allows the experiment to be repeated on
several successive drops, adding the output signals to improve the
signal-to-noise ratio.
Detection limits for square wave polarography are comparable to
those of differential pulse polarography, on the order of 0.1 11M in
favorable cases, but the data acquisition time is substantially shorter.
An advantage of the pulse techniques is that, for reversible couples,
they work almost as well for stationary electrodes as they do for the
d.m.e. In ordinary pulse polarography, the electrode spends most of its
time at an initial potential where the current is very small; thus the total
current passed is small and time is allowed for relaxation of the
diffusion layer between pulses. Square wave polarography treats the
d.m.e. as if it were a stationary electrode and so clearly will work just as
well on a real stationary electrode. A short rest period between scans
would, of course, be required to allow the diffusion layer to dissipate.
The Static Mercury Drop Electrode
An interesting recent technical advance in the design of dropping

mercury electrodes is the so-called static mercury drop electrode
(s.m.d.e.). In this device a larger bore capillary is used and mercury
flow is controlled with a pump. It can be used either to produce a
stationary hanging drop or the pump can be pulsed on and off to
simulate a d.m.e. Drop times can be as short as 0.1 s and as long as
desired. Because of the larger bore capillary, drop growth occurs in
about 50 ms, so that the drop is stationary and constant in area
thereafter. Thus when a s.m.d.e. is used in sampled d.c. polarography,
the current is measured on a stationary electrode, essentially
eliminating the capacitive charging current. Furthermore, the current
can be measured shortly after drop formation before the diffusion layer
has been depleted. Bond and Jones (33) found that for sampled d.c.
polarography using a s.m.d.e., detection limits are on the order of 0.1 11M
for Cd2 + and Ni 2 +, comparable to the performance of differential pulse
polarography with a normal d.m.e. The pulse techniques provide no
further enhancement in sensitivity with the s.m.d.e.
The static mercury drop electrode has significant advantages for
aqueous polarography. Unfortunately, however, it is often found to be
unworkable in organic solutions. Because organic solvents are better at
"wetting" mercury than water, there is a tendency for the solvent to
creep inside the large-bore capillary, taking solutes with it and leading
to contamination which is very difficult to remove.
4.7 THE ROTATING-DISK ELECTRODE
The dropping mercury electrode has many significant advantages

not the least of which is the time-independence of a polarogram. We also
have seen some limitations such as the relative ease of oxidation of
mercury. One way around this limitation without sacrificing time-
independence is to construct the electrode from an inert metal such as
platinum and to use the electrode to stir the solution in some
reproducible way such that the experiment is done under steady-state
conditions. Several electrode designs have been tried in attempts to
implement this strategy. For example, the so-called rotating platinum
electrode has a small Pt wire attached to the side of a glass rod; the rod is
rotated so that the electrode is swept through the solution. Reproducible
steady-state current-voltage curves are obtained (34), but the theoretical
hydrodynamic problem proves to be very difficult. Other similar
approaches have included vibrating wires, rotating loops, and a variety
of other devices.
In recent years, the strategy of a moving electrode has focused in
the rotating-disk electrode (r.d.e.). The r.d.e., which is shown in Figure
4.11, may be of virtually any material, but platinum, gold, and glassy
carbon are particularly common choices.
Hydrodynamics of the Rotating Disk

As the electrode rotates, adjacent solution is pulled along by viscous
drag and is thrown away from the axis of rotation by the centrifugal
force. The expelled solution is replaced by flow normal to the electrode
surface. The rotating disk thus acts as a pump which moves solution up
from below and then out away from the electrode as shown
schematically in Figure 4.33.
Figure 4.33 SchemaLic

representation of solul.ion flow at x
a rotating-disk electrode.
The hydrodynamic problem can be solved to obtain the solution

velocity. In cylindrical coordinates-r, cp, and x-the components of the
velocity are expressed as a power series in x:
Vr = cor [a (x/XH) -~ (x/XH )2_*b (x/XH)3 + ... J (4.42a)
V<p = ror[ 1 [- b (x/XH) + i a (x/XH)3+ ...J (4.42b)
Vx = - vrov[a(x/XH )2_*(x/XH)3_* b (x/XH)4 + ... J (4.42c)
where a = 0.510, b =-0.616, co is the angular speed in radians per second

(21t times the rotation frequency in hertz), and XH is the hydrodynamic
layer thickness parameter,
XH = 'Iv/co
where v = 11/d is the kinematic viscosity of the solution. Since the units of
the coefficient of viscosity 11 are kg m-1s- 1 and the units of density dare
kg m- 3 , v has units of m 2 s- 1 (the same units as a diffusion coefficient)
and XH is in meters. The x-component of the solution velocity is negative
§4.7 The Rotating Disk Electrode 209
(motion is toward the electrode) and increases in magnitude with

increasing X/XH, approaching the limiting value, vx(oo) = -0.884 YoN.
When x = 3.6 XH, Vx = 0.8 vx( 00) and this value of x is taken as a measure
of the thickness of the hydrodynamic layer, which tends to move with the
electrode. For an aqueous solution with v '" 10-6 m 2 s- 1 , XH = 0.063 mm
when co = 250 rad s-l. The hydrodynamic layer is about 0.23 mm thick
under these conditions. Notice that this thickness decreases with
increasing rotation speed-the solution can't keep up when the electrode
rotates very rapidly.
Equations (4.42) were obtained assuming that the radius of the disk
is large compared with the thickness of the hydrodynamic layer. In a
typical r.d.e. (see Figure 4.11), the electrode is 2 mm in diameter, but the
disk as a whole (including insulation) is about 2 cm in diameter. As a
rule of thumb, we can accept hydrodynamic layer thicknesses up to
about one tenth the disk radius, i.e., up to a millimeter or so. This
imposes a lower limit on the rotation speed of about 10 rad s-l (ca. 100
rpm). In practice there is an upper limit on the rotation speed as well.
The hydrodynamic problem was solved assuming laminar flow over the
electrode surface. For very rapid rotation speeds, solution turbulence
sets in and the equations fail. In practice turbulence usually becomes a
problem with co > 1000 rad s-l (10,000 rpm).
A surprising, but important, feature of the velocity is that the x-
component is independent of rand <po Thus the solution velocity
perpendicular to the electrode is uniform over the electrode surface. At
the surface only the <p-component is nonzero, reflecting the fact that
solution on the surface moves with the rotating disk at angular speed co.
Mass Transport to the Rotating Disk

In an experiment using a rotating-disk electrode, material is
transported to the electrode surface by a combination of diffusion and
forced convection. The mass transport equation is obtained from Fick's
second law by adding a term representing forced convection:
ae = Die _Vx ae (4.43)

at ax 2 ax
Because Vx is independent of r and we are only interested in transport to
the electrode (along the x-axis), we need only a one-dimensional
equation. One of the great advantages of the r.d.e. is that the experiment
is done under steady-state conditions. Thus we can set ae/at = 0 and.
solve the simpler differential equation
Die =Vx ae
ax 2 ax
We take Vx'" -afiiN (;(,/XH)2, an acceptable approximation for X/XH < 0.2,
and use the boundary condition C -+ C* as x -+ 00. Defining the
dimensionless variable
u = (a/D)113 0)112 y-116 x
the differential equation can be written as
£c.=_u 2 dC.
du 2 du
Setting C' = dC/du, we can integrate
(c ~=_
leo' C
r)0 u 2 du
In(C'/Co') = - u 3/3
or
dC.:::(dC) exp(-u3/3)
du du 0
(f) i
Integrating a second time, we have
U(X)
C(X) = C(O) + exp (-u S/3) du (4.44)

u 0 0
We will need the derivative (dC/du)o in order to compute the flux at the
electrode surface and the current. To evaluate it, we extend the integral
limits to infinity and apply the boundary condition C(x) -+ C* as x -+ 00.
The definite integral
3-lIs L~ exp (-u 3/3) du =r (4/3) =0.893

is called a gamma function. Thus
C* - C(O) = 311s r (413) (.dQ.)
du 0
or, returning to the physical variable, x
C* -C(O) = 1.288 (D/a)1I3 0)-112 y1l6 (f-)o
For small x, we can assume a linear concentration gradient
dx 0
(de)
= C* - C(O)
XD
where Xn is the diffusion layer thickness. Combining the last two
equations, we can solve for Xn:
xn = 1.288 (D fa )1130)-112 v1l6
or, inserting a = 0.510,
xn = 1.61 D1/3 0)-112 v 1/6 (4.45)
The ratio of xn to XH then is
Xn/XH = 1.61 (Dfv)1/3
For an aqueous solution (D", 10- 9 m 2 s- 1, v'" 10-6 m 2 s- 1) XnfXH '" 0.16,
consistent with the assumption that the linear form of eq (4.42c) could be
used in eq (4.43). This result affords a considerable simplification in
practical calculations on experiments using the r.d.e. The same result
could have been obtained had we neglected the forced convection term in
eq (4.43) and solved the simple equation
Dd~ =0
dx 2
We will have recourse to this simplification in Chapters 5 and 6.
Consider now the standard electrode process
o +n e- ~ R
The flux of 0 at the electrode surface is given by
Jo(O) =-Do(~)o
or
~ (0) =-D Co*-Co(O)
o 0 xn
As usual, the electrode current is
i = - nFAJo(O)
i == nFA(Do/xn)[Co* - CoCO)]
When the electrode is sufficiently negative that CoCO) = 0, we have the

limiting current,
iL = nFADoCo*fxn (4.46a)
or, with the mass-transport rate constant, kn = Dofxn,
iL = nFAknCo* (4.46b)
Substituting eq (4.45), we have
iL ::: 0.62 nFACo*Do2/3 v- 1I6 0)112 (4.47)
Benjamin G. Levich 0917-1987) was Chairman of the Theoretical

Department of the Institute of Electrochemistry of the Soviet Academy of
Sciences and a professor at the University of Moscow until 1972, when he
applied to emigrate. He was allowed to leave for Israel in 1977, but soon
moved to the United States, where he spent the next decade as Albert
Einstein Professor of Science at the City College of New York. His
contributions were mostly in the area of applications of hydrodynamics to
problems in physical chemistry.
Equation (4.47) was first derived by Levich in 1942 (35) and is usually
called the Levich equation. Notice that eqs (4.46) are identical to eqs
(4.15). The analogy with the planar diffusion problem is still more
complete, however, since we recall that we used eqs (4.15a) and (4.16),
together with the Nernst equRtion, to derive the Heyrovsky-Ilkovic
equation, eq (4.17), and this result is also valid for a reversible process at
the rotating-disk electrode. Since the flux ofR can be written
JR(O) = hDR CR(O)
the current can be expressed in terms of CR(O):
i = nFAhDR CR(O)
Thus we can write
iL-i _ h DO CoCO)
- - - hDR CR(O)
or, taking logs, using eq (1.45) to evaluate XDO and XDR, and assuming
nernstian behavior, we obtain the Heyrovsky-Ilkovic equation:
RT
E =Elf2+-ln-.-
iL-i
nF l
where
El/2 =Eo - 2RT In Do (4.48)
3nF DR
Current-potential curves for the r.d.e. thus look exactly like those for a
planar electrode (Figure 4.2) at const8nt time, but half-wave potentials
may be shifted slightly.
Concentration profiles for a r.d.e. can be computed by numerical
integration of eq (4.44) and are shown in Figure 4.34. The concentration
gradient actually is quite linear over some distance from the electrode.
Comparison with Figure 4.1 shows that the assumption of a linear
concentration gradient is rather closer to the truth for the r.d.e. than for
a stationary planar electrode. This of course means that results derived
from a simplified linear concentration gradient model will be more
accurate for the r.d.e. than for the d.m.e. or a planar electrode.
1.0
CICo
0.8
0.6
bulk
solution
0.4
0.2
Figure 4.34 Concentration

0.0
profiles for diffusion to a
rotating-disk electrode for 0.0 0.5 1.0 1.5 2.0
DO = DR, E = EO - 17.8 mV. XIXD
Further details on the rotating-disk electrode can be found in the

review by Riddiford (36) or in books by Albery (A6), Levich (C2), and
Pleskov and Filinovskii (C7).
Example 4.8 Estimate the diffusion layer thickness, mass-

transport rate constant and limiting current for a one-electron
reduction of 0 at a rotating-disk electrode for Do = 10- 9 m 2 s- 1 , v
= 10- 6 m 2 s- 1, as (0 increases from 10 to 1000 rad s-1.
Substituting in eq (4.45), we have the diffusion layer thickness
XD = (1.61 x 10- 4 m sl/2) (0-112
so that for as (0 increases from 10 to 1000 rad s-l, XD decreases

from 51 /lm to 5.1 /lm. The mass-transport rate constant, kD
then increases from 20 to 200 /l1l1 s-l. The limiting current for
an electrode of 1 mm diameter and Co* = 1 mM is, from eq
(4.47),
iL = (8.03 x 10- 10 A-m)lxD
Thus over the dynamic range of the r.d.e. (10 < (0 < 1000 rad s-l)
the limiting current ranges from 6 to 60 /lA.
We see from Examples 4.6 and 4.8 that currents expected for a 1-
mm r.d.e. are about a factor of ten larger than those expected for a d.m.e.
This has advantages and disadvantages. Everything else being equal,
the larger current implies a bigger signal-to-noise ratio. However, it

also implies a larger iRdrop in the solution. Rotating-disk voltammetry
generally is more susceptible to iR drop problems than is polarography.
The Rotating Ring-Disk Electrode

The hydrodynamics of the rotating-disk electrode can be turned to
advantage by adding another electrode as a ring surrounding the disk
electrode as shown in Figure 4.35. Solution which flows to the central
disk electrode is then flung outward by centrifugal force and flows past
the concentric ring electrode (see Figure 4.33). Thus the solution can be
sampled electrochemically at the ring shortly after undergoing an
electron-transfer process. Solution of the convective diffusion problem is
messy and we will not go into the details. The key parameter is the
collection efficiency, i.e., the fraction of disk electrode product which can
be sampled at the ring. This parameter depends on the electrode
geometry only, independent of rotation speed, diffusion coefficients, etc.,
and is typically on the order of 0.5. Thus if a species is reduced at the
disk, about half the product can be reoxidized at the ring. If the disk
product is involved in a chemical reaction, the collection efficiency will
be less than theoretical and will then depend on rotation speed,
approaching theory at faster speeds.
contact
bushings
Teflon
insulation
. electrodes
Figure 4.35
electrode.
The rotating ring.disk @f
':::?'
The method is comparable in some respects to cyclic voltammetry

(see §4.4) in that the products of electron-transfer reactions can be
studied, but, because the transport equations can be solved more exactly
(it is a steady-state experiment) and because the ring electrode potential
can be controlled independently of the disk, the method is capable of
greater flexibility. The main disadvantage of the method is economic:
ring-disk electrodes are difficult to make and thus are expensive. Two
indicator electrode potentials must be controlled using a so-called
bipotentiostat, which is about twice as complicated (and four times as
expensive) as an ordinary potentiostat. For further details, see the book
by Albery and Hitchman (F2).
4.8 MICROELECTRODES
Beginning in the late 1970's, a lot of effort has been devoted to the
development of microelectrodes, electrodes with dimensions on the order
of 0.1-25 /lm. This effort has been motivated by the need for very small
electrodes for in vivo biological studies, by the recurring problem of iR
drop in conventional voltammetric experiments, and by the need for new
methods capable of probing electrochemical processes on a very short
time scale. Progress has been made along these lines, but meanwhile
microelectrodes have been found to have other advantages, unlooked for
when they were first conceived, and they have become important tools in
the electrochemist's bag of tricks.
Microelectrodes have been constructed in a number of different
geometries, including inlaid microdisks, microcylinders, and
hemispheres, as well as inlaid bands and arrays. For any geometry, the
fabrication of a microelectrode is nontrivial, involving
micromanipulation techniques and/or some of the microlithographic
methods used in the fabrication of integrated circuits. Wightman and
Wipf (37) describe some of these methods and provide references to the
literature. A more recent paper by Baer, et al. (38), describes improved
methods for the construction of microdisk electrodes. By far the most
common geometry has been the inlaid disk, and we will confine our
attention to such electrodes. The construction of a typical microdisk
electrode is shown in Figure 4.36.
We begin with consideration of the diffusion problem in the context
of potential-step chronoamperometry and then discuss the use of
microelectrodes in cyclic voltammetry. Applications of microelectrodes
in studies of homogeneous and electron-transfer kinetics will be
discussed in §5.3 and §6.3. For further details, the reviews by Wightman
and Wipf(37) and by Bond, Oldham, and Zoski (39) are recommended.
lead
wire
epoxy
glass
tube
conducting
epoxy
Figure 4.36 Cross-sectional view of a
microdisk electrode constructed using 4-
mm glass tubing.
Diffusion to a Microdisk Electrode
Our approach to voltammetry has been through insights gained

from solutions to the diffusion equation assuming planar diffusion.
Thus we have assumed implicitly that electrode dimensions are large
compared with the diffusion layer thickness. In principle, this
assumption fails for large electrodes at long times when the diffusion
layer has grown very thick. Since convection usually limits experiment
times to a minute or so, at which point XD is on the order of 0.5 mm (see
Example 4.3), the failure is not serious in practice. When very small
electrodes are used, the assumption of planar diffusion fails at much
shorter times.
Figure 4.37 Schematic

representation of the growth of
the diffusion layer. For short
times, diffusion is mostly
normal to the electrode surface,
but at longer times, the radial
component becomes important
and eventually dominates.
§4.8 Microelectrodes 217
When we include both the coordinate normal to the electrode

surface, z, and the radial distance, r, measured from the center of the
electrode, the diffusion equation becomes
ac
at
=D(ic
ar2
+ 1 ac +
r ar
iC)
az2
The boundary conditions imposed by the electrode geometry are
C -+ C* as r -+ 00, Z -+ 00
(ac) = 0, r > ro; (ac) = _ j(r) ,r ~ ro

az z=o az z=o nFD
where ro is the electrode radius and j(r) is the current density at a
distance r from the electrode center. The problem is closely analogous to
the case of spherical diffusion whi.ch we examined in §4.1, but the
equations are more complex and a closed-form solution is not possible.
Solutions in the form of integral equations are possible, but not very
enlightening; see Oldham and Zoski for a summary of this approach
(40).
Based on our experience with spherical diffusion, we can make a
educated guess of how the diffusion layer will grow in a potential-step
chronoamperometry experiment, Figure 4.37. Diffusion starts out
normal to the electrode surface (tl), but as the diffusion layer thickness
approaches the dimension of the electrode, a radial component becomes
important (t2 - t4). Eventually the diffusion layer becomes hemispherical
in shape, and the surface flux-and thus the current-reaches a steady
state, i.e., hemispherical diffusion is able to keep up with the
consumption of the diffusing species at the electrode surface.
Soos and Lingane (41) showed that a modification to the Cottrell
equation, analogous to eq (4.21), provides satisfactory agreement with
experimental diffusion-controlled currents:
iD = nFADC* (_1_ + iL)
'/7tDt ro
(4.49)
In eq (4.48), r is the electrode radius and b is an empirical constant (b = 1

for a spherical electrode). For '/7tDt «ro (short times or large
electrodes), the correction term is negligible and the current drops as
lIVt;-the diffusion layer is thin compared with the electrode radius so
that diffusion is still nearly planar. At longer times, the correction term
becomes important and the current begins to decay less rapidly-the
radial component has begun to contribute and the diffusion layer has
spread beyond the edges of the electrode. When villt » ro, the
correction term dominates and the current becomes time-independent-
the diffusion layer has become hemispherical and the steady-state
regime is reached.
10
~ ~:-::~-=~-=~-:-~::-~~-~-~-_..J
Figure 4.38 Microelectrode o

current in a chronoampero- 0.00 0.05 0.10
metric experiment computed
by digital simulation as a
function of time. Inset: short
time behavior compared with 1
["1/2 current decay (dashed
line); data from reference o 5 10 15 20 25
(42). DtJr2
Heinze (42) used digital simulation techniques to test eq (4.49); he

found that b is not quite constant, ranging from 1 at short times to 41n at
long times. Thus the steady-state diffusion current is
iD =4nFAC*D = 4nFC*Dro (4.50)
nro
which is smaller than the steady-state current predicted by eq (4.21) for a
spherical electrode by a factor of n. Normalized currents computed by
Heinze are plotted us. Dtlr02 in Figure 4.38; the current approximates the
steady-state value for Dtlr02 > 10. The inset in Figure 4.38 shows the
response in the time range 0 < Dtlr02 < 0.1. For an electrode with ro = 1
mm, this corresponds to times up to about 100 s CD = 10-9 m 2 s- 1), the time
of a typical chronoamperometry experiment, and we see that the
computed response is very close to the t- 1I2 current decay predicted by the
Cottrell equation. For a microelectrode with ro = 111m, the inset time
scale extends only to 0.1 ms, so that, for an experiment lasting for more
than a second, most of the time scale is in the steady-state region.
Consider now the mass-transport rate constant kD, introduced in
§4.1. The mass-transport rate constant can be obtained from eqs (4.15b)
and (4.50):
kD = 4D1nro (4.51)
We will find this expression to be useful in our applications of
microelectrodes in Chapters 5 and 6, but meanwhile it provides a useful
insight. The diffusion layer thickness, XD, is less useful; we could obtain
XD by combining eqs (4.15a) and (4.50): XD = nro/4, but this is a misleading
result which we must immediately reject. The diffusion layer thickness
was introduced as the distance over which a linear concentration

gradient would deliver the flux corresponding to in. Since the
concentration gradient is far from linear here, the diffusion layer
concept is too far from the truth to be useful.
Example 4.9 Compute the steady-state diffusion currents,

current densities, and mass-transport rate constants for the
one-electron reduction of 1 mM 0 at electrodes with 11-lm and 1
mm radii, assuming D = 10- 9 m 2 s- 1. About how long does it
take for the current to reach steady-state after application of
the potential?
The electrode areas are n:r02 = 3.14 x 10- 12 and 3.14 x 10-6 m 2;
substituting into eq (4.50), we have
in = 0.096 and 96 nA
The current densities are
in/A = 123 and 0.123 A m- 2
Substituting into eq (4.51), we get the mass-transport rate
constants are
kn = 1.27 x 10-3 and 1.27 x 10-6 m s-1
Notice that the current is 1000 times larger at the 1-mm
electrode, but the current density and mass-transport rate
constant are 1000 times larger at the l-I-lm electrode!
Comparing this result with Example 4.8, we see that kn is 10-
100 times bigger for steady-state diffusion to the 1-l-lm electrode
than under forced convection conditions at a rotating-disk
electrode. The current is close to steady-state for t > 10 ro2/D.
Thus we have
tsteady-state = 0.01 s and 10 4 s
For the 1 mm electrode, more than two hours is required to
reach steady-state conditions, by which time convective mixing
will long since have perturbed the experiment.
The very small currents at a microelectrode have several

experimental implications. First, and most obviously, the small
currents require more sophisticated current-measuring equipment.
Electrical pick-up and noise are problems and cells need to be carefully
shielded. On the positive side, however, the reasons for using a three-
electrode cell controlled by a potentiostat largely disappear when
micro electrodes are used. Reference electrodes can easily handle
currents up to a few nanoamperes without significant polarization so
that most steady-state microelectrode experiments are done using a two-

electrode configuration, and, of course, with only two electrodes, a
potentiostat is unnecessary. Eliminating the potentiostat removes a
source of noise as well.
Solution iR drop is a less serious problem than in experiments with
conventional electrodes. Because the current is smaller, larger solution
resistivities can be tolerated; thus supporting electrolyte concentrations
can be reduced or nonpolar solvents like hexane can be used. When
current flows between a very small electrode and a large remote
electrode, the resistance is determined by the size of the small electrode
and can be shown to be R = p/4ro, where p is the solution resistivity (43).
For a microelectrode in the steady-state regime, iD is proportional to ro,
so that the ohmic potential drop reaches a minimum, independent of
electrode size,
iDR = nFC*Dp
Oldham (44) has shown that iDR is also independent of electrode shape.
For a one-electron process with C* = 1 mM, D = 10-9 m 2 s- 1, p = 1 Q-m, the
ohmic drop is about 0.1 mY.
Another important consequence of decreased electrode area is a
much smaller contribution of capacitive charging current. In
chronoamperometry, the capacitive charging current is a transient
response to the potential step,
ic = f...: exp R~d (4.24)
where f... is the potential step, Cd is the double-layer capacitance, and R

is the solution resistance. Since Cd cc A, the time constant for capacitive
current decay, RC, decreases with electrode size so that the charging
current transient is much shorter-lived for smaller electrodes. The
accessible time scale thus is much shorter for microelectrodes.
The attainment of a steady-state current, together with the removal
of interferences from ohmic potential drop and capacitive charging
current, make chronoamperometry a more accurate and more easily
applied analytical technique when microelectrodes are used.
Cyclic Voltammetry at a Microelectrode
Figure 4.39 shows cyclic voltammograms of ferrocene using a 6 )lm

radius electrode for scan rates ranging from 10 mV s-1 to 10 V s-l. Two
features of these voltammograms are particularly notable. First, we see
that, at the slowest scan rate, the reverse scan retraces the i-E curve of
the forward scan and that the cyclic voltammogram has the shape of a
steady-state voltammogram, i.e., like a sampled-d.c. polarogram or a
rotating-disk voltammogram. As the scan rate increases, the reverse
scan begins to deviate from the forward sweep, anodic and cathodic
J:_2_n_A. . .". -_
-:1~I:::2==nA=~= ~
Figure 4.39 Cyclic voltammograms

showing the oxidation of ferrocene in
acetonitrile at a Pt microelectrode (ro
= 6 11m) for v = (a) 0.01, (b) 0.1, (c) 1.0
and (d) 10 V s·l. Reprinted with
permission from J. Heinze and M.
Stiirzbach, Ber. Bunsenges. Phys 0.6 0.4 0.2 0.0 ·0.2
Chern. 1986, 90, 1043. EIV
peaks develop, and, at the fastest scan, the shape approaches that
expected for cyclic voltammograms using macroelectrodes. Second, we
notice that, for slow scans, the current is independent of scan rate,
approaching the usual ..v-dependence only when the peaks are well
developed.
This behavior is to be expected, given our results for
chronoamperometry. For a microelectrode with ro = 6 11m, we would
expect essentially time-independent current about 0.4 s after application
of a potential step (assuming D '" 10. 9 m 2 s· 1). In an experiment where
0.4 s is a long time (cyclic voltammetry at 10 V s·l), the steady state is not
reached and the voltammogram has a "normal" shape and the current
increases as..v. When 0.4 s is a short time (cyclic voltammetry at 10 mV
s·l), the steady state is reached early in the experiment and is
maintained. Under steady-state conditions, the current depends only on
the potential; it is independent of time and thus of scan rate and scan
direction. Notice the absence of cathodic current on the reverse scan at
0.01 and 0.1 V s·l. Apparently no ferrocenium ions remain near the
electrode to be reduced; in the steady-state regime, the electrode product
diffuses away rapidly and irretrievably. Thus steady-state cyclic
voltammograms do not provide information about the chemical stability
of a reduction or oxidation product.
Figure 4.40 Normalized peak

current density in a cyclic
voltammogram as a function of
scan rate v and Wlro = 0 - 10.
Reprinted with permission
from J. Heinze and M.
Stiirzbach, Ber. Bunsenges. -5 o 5
Phys Chern. 1986, 90, 1043. In u
As in chronoamperometry, the transition from steady-state to time-

dependence depends on the electrode area and the time scale. Heinze
and Storzbach (45) employed digital simulation methods to investigate
this transition in cyclic voltammetry. Their work is summarized in
Figure 4.40, which shows a plot of maximum current density (steady-
state or peak) in a linear potential-scan voltammogram as a function of
scan rate u and the parameter Wlro. Each curve in the figure shows a
steady-state limit where the current is independent of scan rate and a
planar-diffusion limit where the current increases as -Iff (the plot ofln i
us. In u has a slope of 1/2). From this correlation, Heinze and Storzbach
concluded that the steady-state regime is reached for u < RTDI9nFr02.
Thus for T= 298 K, D = 10- 9 m 2 s- 1, n = 1, steady-state current is expected
for u < 3 V s-l for a I-11m radius electrode. In principle, of course, steady-
state conditions are achievable for macroelectrodes, but the scan rate
would have to be very slow indeed: less than 3 I1V s-l (about 4 days for a
I-V scan!) for a 1-mm radius electrode.
Microelectrodes have significant experimental advantages for
cyclic voltammetry, but the details are somewhat different than for
chronoamperometry. Cyclic voltammograms at fast scan rates with
normal-sized electrodes are seriously distorted by ohmic potential drop;
the cathodic peak is shifted by -ipcR and the anodic peak by +ipaR, and
since ip = -Iff, the distortion becomes very large at fast scan rates,
imposing a practical upper limit of about 100 V s-l for millimeter-sized
electrodes. With microelectrodes, peak currents are two to three powers
of ten smaller and the interference of iR drop is small up to very fast
scan rates. The improvement is demonstrated by the microelectrode
cyclic voltammograms, obtained by Howell and Wightman (46) and
shown in Figure 4.41, for anthracene in acetonitrile solution at scan
rates ranging from 1000 to 10,000 V s-l.
Figure 4.41 Cyclic voltammo-

grams of 2.22 mM anthracene
in CH3CN solution, 0.6 M in
E t 4 N C I 0 4, at a gold
microdisk electrode, r = 6.5
11m for scan rates ranging
from 1000 to 10,000 V s-l;
potentials VS. Ag/AgCI04.
from J. O. Howell and R. M.
Wightman, Anal. Chern. 1984,
56, 524, copyright 1984 -1.6 -1.8 -2.0 -2.2 -2.4 -2.6 -2.8
American Chemical Society. EIV
The interference of capacitive charging current is quite different in

chronoamperometry and cyclic voltammetry. In cyclic voltammetry,
charging of the double-layer capacitance makes a constant contribution
to the current, given by eq (4.33), but the sign changes when the scan
reverses; thus the negative-going and positive-going traces are
displaced. The capacitive current is proportional to Cd and thus to
electrode area. In the steady-state regime, the faradaic current is
proportional to fA so that ie/if scales as fA, and capacitive charging
current is usually of negligible importance. For faster scan rates where
the current is time-dependent and scales as A, ie/if is independent of
area and capacitive current again leads to a displacement of the forward
and reverse scan curves, as seen in Figure 4.41.
4.9 APPLICATIONS
Applications of the voltammetric techniques discussed in this

chapter have been extremely varied. For the purposes of discussion,
however, we can roughly divide the applications into three groups:
(1) Quasi-thermodynamic applications, where a half-wave
potential, taken as an approximation to the standard potential, is
used in correlations with theory or with other kinds of experimental
data or in computation of thermodynamic quantities.
(2) Analytical applications. Historically, analytical applications
have been the most important and have provided the motivation for
much of the development of voltammetric techniques.
(3) Mechanistic studies, in which electrochemical data are used to

deduce the mechanism of the electrode process or the reactions
which occur following oxidation or reduction of a substrate.
In this section, we will discuss some examples of the first two kinds of
applications. Mechanistic applications will be discussed in Chapters 5
and 6.
''Standard'' Potential/Electronic Structure Correlations
The free energy change associated with the one-electron reduction

of 0 to R can be written
t:,.Go = t:,.GO(gas) + t:,.t:,.GO(solv)
where t:,.GO(gas) is the free energy change for the gas-phase electron
attachment process,
O(g) + e· ~ R(g)
and
t:,.t:,.GO(solv) = t:,.GRO(solv) - t:,.GoO(solv)
is the difference in solvation free energies of Rand 0 The free energy
change is related as usual to the standard reduction potential,
t:,.Go = -F(EO + const)
where the constant depends on the zero of the potential scale. If the gas-
phase process involves no entropy change, then t:,.GO(gas) = -A, the
electron affinity. The half-wave potential measured in a polarographic
experiment can be written, using eq (4.10):
EV2 = A t:,.t:,.GO(solv) RT In Do - const (4.52)
F F 2F DR
If half-wave potentials are measured for a series of compounds using
the same indicator and reference electrodes, the same solvent and
supporting electrolyte, and the compounds are sufficiently similar that
t:,.t:,.GO(solv) and Do/DR may be assumed to be nearly constant, we might
expect that El/2 will correlate well with changes in the electron affinity
of the compounds. The electron affinity is expected to be related to the
energy of the lowest unoccupied molecular orbital (the LUMO).
Following a similar line of reasoning, we expect the oxidation potential
to correlate with the energy of the highest occupied molecular orbital
(the HOMO).
The original work along these lines was done in the 1950's by
Hoijtink (47) who found that an excellent correlation is obtained between
polarographic half-wave reduction potentials for aromatic hydrocarbons
§4.9 Applications 225
G. Jan Hoijtink (1925-1978) was Professor of Physical Chemistry at the

Free University, Amsterdam and later at the University of Amsterdam
and the University of Sheffield. Hoijtink was a pioneer in mechanistic
organic electrochemistry and was one of the first to apply ideas from
molecular orbital theory to electrochemistry and spectroscopy.
-2.2 1.6
Ell2N
-2.0 1.4
-1.8 1.2
-1.6 1.0
-1.4 0.8
-1.2 0.6
-1.0 0.4
0.3 0.4 0.5 0.6 0.7 -0.3 -0.4 -0.5 -0.6 -0.7
(E + a)/~ (E + a)/~
Figure 4.42 Correlation of half-wave potentials with Huckel MO theory

energies for (a) polarographic reduction of aromatic hydrocarbons in 2-
methoxyethanol solution (potentials measured us. the mercury pool
anode) and (b) oxidation of aromatic hydrocarbons at a rotating platinum
electrode in acetonitrile solution (potentials measured us. s.c.e.). The
numbers correspond to: (1) tetracene, (2) 1,2-benzpyrene, (3) anthracene,
(4) pyrene, (5) 1,2-benzanthracene, (6) 1,2,5,6-dibenzanthracene, (7)
phenanthrene, (8) naphthalene, and (9) biphenyl. Data from Streitwieser
(48).
and LUMO energies obtained from Huckel molecular orbital theory.

Similar correlations are obtained between half-wave potentials for
oxidations of aromatic hydrocarbons, measured with a rotating
platinum electrode, and the HOMO energies obtained from Huckel
molecular orbital theory. Some examples of such correlations are
shown in Figure 4.42. (The energy scale of Figure 4.42 is in units of ~,
the resonance integral parameter of Huckel MO theory; the zero is
relative to a, the coulomb integral parameter.)
This approach has been used extensively in organic
electrochemistry not only for aromatic hydrocarbons but for compounds
with many different functional groups. In the first instance, the
motivation was to test molecular orbital theory calculations and the
insights obtained from the calculations, but once the correlation is

established, half-wave potentials for reductions or oxidations can be used
to predict other chemical or physical properties which have been found
to correlate with molecular orbital theory energies. A good review of this
kind of work is given by Streitwieser (48).
More recently, a similar approach has been used with
organometallic systems. For example, it is well known that metal
carbonyl complexes are stabilized by electron back donation from the
metal into empty 1£* orbitals on the CO ligand. Thus when CO is
replaced by a less strongly l£-acidic ligand, a phosphite for example, the
electron density on the metal is expected to increase and the molecule
should be easier to oxidize and harder to reduce. Since l£-back donation
to a CO ligand puts electron density in a CO 1£* orbital, the infrared C-O
stretching frequency is often taken as a measure of electron density on
the metal. Thus with two physical parameters allegedly measuring the
same thing, the obvious thing to do is to see if they correlate.
Example 4.10 de Montauzon and Poilblanc (49) have studied

the reduction potentials of a series of nickel carbonyl
complexes, Ni(CO)4_n[P(OMe)3]n, where n = 0, 1, 2 and 3, in
THF solution using a d.m.e. and a Ag/Ag+ reference electrode.
In this series, the reduction potential becomes more negative
with increasing n, as expected. Similarly, the C-O stretching
frequency decreases with increasing n, again as expected. The
reduction potential is plotted us. CO stretching frequency in
Figure 4.43a; a good (but somewhat nonlinear) correlation is
obtained.
de Montauzon and Poilblanc also studied the correlation

between El/2 and yeo for [Co(CObLh where L is a phosphine
or phosphite ligand. The results are shown in Figure 4.43b.
Taking either the phosphines alone or the phosphites alone,
there is a good correlation between EI/2 and Yeo, similar to that
shown in Figure 4.43a. However, the reduction potentials for L
= PPh3 and P(OMe)3 are identical, whereas the C-O stretching
frequencies differ by 18 cm- I . Furthermore, the mixed ligands,
Ph2P(OMe)3 and PhP(OMe}s, show equal values of El/2 but C-O
stretching frequencies intermediate between the PPh3 and
P(OMe)3 derivatives. The reason for this difference appears to
be that both PPh3 and P(OMe)3 are strong a-donors but P(OMe}s
is somewhat l£-acidic. Thus P(OMeh, which feeds considerable
charge density into the metal acceptor orbitals, also removes
charge density from the metal l£-donor orbitals. The C-O
stretching frequency is particularly sensitive to l£-electron
density, and C-O stretching frequency shifts probably reflect
-4.0 -2.0
EV2N 1
-3.8 -1.8
-3.6 -1.6
-3.4 -1.4
-3.2 -1.2
-3.0 -1.0
-2.8
-0.8
1950 2000 2050 2100 2150 1970 1980 1990 2000
vCO/cm- 1 v CO /cm- 1
Figure 4.43 Correlation of polarographic half-wave potentials with C-O

stretching frequencies for (a) Ni(CO)4_nLn" L = P(OMejs and n = 0-3,
and (b) [CO(CO)3L12 for L = (1) PMe3, (2) PEt3, (3) PMe2Ph, (4) PMePh2,
(5) PPh3, (6) P(OEt)3, (7) P(OMe)3, (8) P(OPh)3, (9) Ph2POMe, and (10)
PhP(OMe)2. Reprinted with permission from D. de Montauzon and R.
Poilblanc, J. Organornet. Chern. 1976, 104, 99, copyright 1976 Elsevier
Science Publishers.
changes in metal 7t-electron density rather than overall charge

density. Thus half-wave potentials may be a better measure of
gross charge density than C-O stretching frequencies.
Stability Constants of Coordination Complexes
The voltammetric behavior of metal coordination complexes is

determined in part by the kinetic stability or lability of the complex. For
substitution-inert species such as Co(ll!) or CrOll) complexes, electron
transfer is often followed by rapid chemical steps. In more labile
systems, it is often necessary to consider equilibria among several
different species prior to electron transfer, with only the most easily
reduced species participating in the electron-transfer step. We will
discuss the general case in §5.3. If the system is sufficiently labile that
the complexation steps can be assumed to be always at equilibrium, even
at the electrode surface, we can deal with the expected behavior using
the techniques we have developed in this chapter.
Consider an aquo metal ion, M(H20)m n+, which is reversibly

reduced to the metal
M(H20)m n+ + n e- ~ M(Hg) + m H20
and suppose that the metal ion forms a series of coordination complexes
with the electroinactive ligand L:
M(H20)m n+ +} L f::! M(H20)m-jLf+ +} H20
with overall formation constants
~. = [M(Hi))mjLjn+]
'J [M(Hi))':+] [LY
where ~o = 1. To keep the notation relatively compact, we will label the
species 0, 1, ... } ... m, so that Co is the concentration ofM(H20)mn+ and Cj
is the concentration of M(H20)m-jLf+. We suppose that these equilibria
are very fast, so that the relative concentrations at the electrode surface
are the same as in bulk solution Since the various complex species are
rapidly interconverted, the electrode current is related to the total flux of
all species at the electrode surface:
m
i = - nFA I .1;(0)
j=O
The fluxes of the various species are determined by the concentration
gradients. If knj is the mass transport rate constant for delivery of
species} to the electrode, we can write
JiO) = - knj [Cj* - CiO))
so that the current is
m
i = nFA I knj [Cj* -C;CO)]
j=O
easily reduced species is nearly complete, then all CiO) = and the
current is diffusion limited,
°
When the potential is sufficiently negative that reduction of the most
m
in = nFA I knj C/
j=O
Subtracting i from in, we have
m
in - i = nFA I knj C;(0)
j=O
If the complexation steps are fast enough that they are at equilibrium
even at the electrode surface, and if the free ligand concentration is
sufficiently high that CL is uniform throughout the solution, then we
have
Thus
m .
iD - i == nFACo(O) L kDj PjCr: (4.53)
j=O
Reduction of a metal complex at the d.m.e. produces metal atoms
which dissolve in the mercury and diffuse into the interior of the drop
with a flux
JM(O) == kDM [CM* - CM(O)]
where kDM is the rate constant for diffusion of M in the amalgam; we
assume that the concentration of metal in the interior of the drop, CM*,
is negligible. This flux is also related to the electrode current,
(4.54)
If we assume that the electron-transfer process is reversible, the
concentration ratio COCO)/CMCO) must be given by the Nernst equation
CoCO) nF(E - EO)
CMCO) == exp RT
Using eqs (4.53) and (4.54), this ratio can be written
CoCO) - - iD -=- i- - - xkDM
--
CMCO) L kDjpjCr:
Combining the two concentration ratio expressions, we have
E==Eo+RTln kDM +RTlniD-i (4.55)
nF L kDjPjCr! nF i
which is just the Heyrovsky-Ilkovic equation with

E1I2 == EO + RT In kDM - RT In ~ p·C j (4.56)
nF kD nF L. 'J L
where we have assumed that the complex species have nearly the same
diffusion coefficients so that kDj == kD. Since
kDM==,yDM
kD Do
the first two terms are just the half-wave potential in the absence of
ligand, eq (4.10). The last term in eq (4.56) then represents the shift in
half-wave potential as a function of ligand concentration. When the
complexation steps are sufficiently well separated that, for a particular
range of free ligand concentrations, only two species have significant

concentrations in the bulk solution, then one term, PjCr/, will dominate
in the sum and we can simplify eq (4.56) to
Lilll/2 = - RT Inp'Ci
nF 'J
or
RT
Lilll/2 =_RT In p. -J-In CL (4.57)
nF 'J nF
Equation (4.57) was first used by Lingane around 1940 and is sometimes
called the Lingane equation.
Example 4.11 Lingane (50) used eq (4.57) to analyze

polarographic half-wave potential data obtained for Pb(II) in
0.01-1.0 M NaOH solutions (of constant 1.0 M ionic strength).
Figure 4.44 shows a plot of E 112 us. pOH. The slope of the
straight line is 0.0799 ± 0.0015, which corresponds to j = 2.70 ±
0.05. Lingane interpreted this result as indicating that Pb(II)
in basic solution is HPb02- with the equilibrium
Pb 2+ + 3 OH- f:! HPb02- + H20
preceding electron transfer. The deviation of j from the
integral value of 3 was explained as arising from the use of
OH- concentrations rather than activities in the plot.
-0.60
E1/2N
-0.65
-0.70
-0.75
Figure 4.44 Polarograph-
ic half-wave potentials
(us. s.c.e.) for Pb(II) in
KN03iNaOH solutions at -0.80
25°C; data from reference 0.0 0.5 1.0 1.5 2.0
(50).
-log[OH-]
Only a few complexation equilibria can be characterized using

Lingane's approach. Even when equilibria are fast enough, it is often
impossible to focus on just one step. When several equilibria are
simultaneously important, the more complete expression, eq (4.56), must
be used. Several graphical methods have been developed to fit half-wave
potential data to an equilibrium model. One way of dealing with the
problem is to write eq (4.56) as
f..E]J2=_ RT l n ~ ~·ci
nF £. 'J
J
and define the function Fo(CL),
FOCCL) =L ~jCi = exp - nFf..ElJ2

. RT
J
Since f..E1I2 is measured, the function FO(CL) can be computed for each
experimental point. The function then is decomposed to
Fo(CL> -1 2
Fl(CL> = CL = ~l + ~~L + ~aCL + ...
A plot of FICCL) us. CL should give a curve with an intercept ~l at CL = O.

Similarly defining F2(CL),
FiCL> = F1(CL> - ~l = ~2 + ~sCL + ...

CL
we see that a plot of F2(CL) us. CL allows the determination of ~2 by
extrapolation to CL = O. Graphical methods such as this are inherently
inaccurate but have the advantage that the experimenter gets a good
"feel" for the data, making it less likely that the data will be fitted to a
model which they really do not support However, a nonlinear least-
squares refinement of the data should follow the rough graphical
analyses. See Crow (51) for further details and examples.
Analytical Applications
Polarography was the first partially automated instrumental
analytical technique and the first technique capable of routine analysis
at the submillimolar level; as such polarography had a tremendous
impact on analytical chemistry and was widely used from the 1930's.
With the development of competing instrumental methods with even
greater sensitivity and specificity (e.g., atomic absorption
spectrophotometry), analytical applications of polarography went into a
decline in the 1960's. Classical d.c. polarography is limited to analyte
concentrations in the 10 11M - 1 mM range, but the detection limits are
lowered considerably with the pulse techniques (or the a.c. methods to be
discussed in §6.4), which suppress the interference of the capacitive
charging current. The development of these techniques brought about a

renaissance of analytical voltammetry in the 1970's (52).
Electrochemical methods are now competitive with or preferable to
spectroscopic methods for many analyses. Developments currently
under way-e.g., the static mercury drop electrode and digital
techniques to harness the power of microcomputers to electrochemical
experiments-promise to keep analytical voltammetry competitive in the
years to come (53,D12).
Polarographic methods have been developed for most of the
common metal ions, including the alkali metals. Indeed, HeyrovskJ's
first paper on polarography (54) reported the reduction of the alkali
metal ions. Advantage is taken of the large hydrogen overpotential on
mercury (see §6.2) so that well-defined waves are seen for the alkali
metals in weakly acidic or basic aqueous solutions. Unfortunately,
however, Na+, K+, Rb+, and Cs+ all reduce at nearly the same potential so
that polarography is unable to distinguish among them. (Lithium is
sufficiently different that it can be determined in the presence of the
other alkali metals.)
Although the polarographic diffusion current is proportional to
concentration, the proportionality constant includes the diffusion
coefficient and the dropping mercury electrode parameters (mercury
flow rate and drop time). Polarographic methods reported in the
analytical literature usually include the so-called diffusion current
constant,
ID = iD/u2/3tV6C
so that a reported procedure in principle can be used without calibration.
As a practical matter, however, careful work should include the
preparation of a calibration curve determined under the actual
experimental conditions used in the analysis. Because of metal ion
complexation and variations in activity coefficients, the half-wave
potential and diffusion current constant are often strongly medium
dependent. These parameters are given for Cu(I!), Pb(II), Cd(II), and
Zn(II) in several aqueous electrolyte solutions in Table 4.1. Copper(II),
for example, gives polarographic waves ranging over about half a volt,
depending on the medium. In the presence of high chloride ion
concentrations, which stabilize Cu(!), two waves are observed which
correspond to the stepwise reduction of Cu(II).
Polarographic analytical methods are often transferable with better
sensitivity and/or better resolution to more modern techniques such as
steady-state voltammetry and the various pulse techniques.
One of the advantages of polarography and the other voltammetric
methods is the ability to distinguish among and analyze simultaneously
for several species. Resolution is particularly good with the differential
pulse technique (see Figure 4.32). Since half-wave potentials are
medium dependent, the solvent or aqueous buffer system used in a given
Table 4.1 Polarographic Data for Some Metal Ions
EIJ2 (lD)a,b
Medium Cu(II) Pb(II) Cd(II) Zn(II)
O.lMKCI +0.04(3.23) -0.40(3.80) -0.60(3.51) -1.00(3.42)

1MKCI +0.04, -0.22 -0.44(3.86) -0.64 -1.02
(3.39)C
1MHN03 -0.01(3.24) -0.40(3.67) -0.59(3.06) d
0.5 MH2S04 0.00(2.12) e -O.59(2.6)d d
1MNH3, -0.24, -0.50 -0.81(3.68) -1.33(3.82)
1 MNH4CI (3.75)C
1MNaOH -0.42(2.91) -0.76(3.39) e -1.53(3.14)
pH 4.5 bufferf -0.09(2.37) -0.48(2.37) -0.64(2.34) -1.23d
pH 9.0 bufferf -0.12(2.24) -0.50(2.30) -0.64(2.34) -1.15(2.30)
a From Kolthoff and Lingane (D1). Much more extensive tables

have been published by Meites (55).
b Diffusion current constant in units IlA mM-1(mg s-1)-2/3s-1/6.
Half-wave potential in volts us. s.c.e.
CTotal diffusion current for double wave.
d Wave masked by hydrogen reduction.
e Insoluble.
f 0.5 M sodium tartrate buffer.
analysis is chosen to optimize resolution of the expected waves. In

actual analytical work, it is usually necessary to separate the analytes
from a complex mixture, leaving behind interfering species. Ideally,
this initial separation step should leave the analytes in a form that will
give a well-resolved voltammogram. One general approach to this
problem is to extract metal ions as chelate complexes soluble in a polar
aprotic solvent.
Example 4.12 A weakness of differential pulse polarography

is that a single peak is obtained in analysis of solutions where
the analyte is present in both oxidized and reduced forms.
Thus, for example, a solution containing Fe(II) and Fe(III) as
aquo ions (or other complexes related by simple electron
transfer) gives a single peak with peak current proportional to
the sum of the FeCII) and Fe(III) concentrations. Leon and
Sawyer (56) extracted iron into propylene carbonate, FeCII) as
the tris-4, 7 -diphenyl-1, 10-phenanthroline complex and Fe(III)
as the tris-8-quinolinato complex. These complexing agents

are highly selective for Fe(II) and Fe(III), respectively. The
differential pulse polarogram then shows well-separated
reduction peaks for Fe(III) and Fe(II) at -0.55 and -1.25 V us.
s.c.e., respectively, with peak currents linear in concentration
in the 2 - 200 11M range, thus achieving the higher sensitivity of
the differential pulse method while at the same time resolving
Fe(II) and Fe(III) into separate peaks.
In the presence of ligands which form complexes with mercury,

e.g., halides and pseudohalides (CI-, SCN-, etc.), an anodic polarographic
wave is observed which corresponds to the oxidation of mercury:
Hg + p X- ----t HgXp 2-p + 2 e-
If the formation constant of the mercury complex is large enough, the
wave is shifted to a more negative potential and is resolved from the
normal mercury oxidation process. Since the process relies on the
diffusion of X- to the electrode surface, the limiting current is
proportional to the bulk concentration of X-. Using sensitive techniques
such as pulse polarography or sampled d.c. polarography on a static
mercury drop electrode, this wave provides a good way of determining a
wide range of anions and other complexing agents. In addition, the
wave shape and half-wave potential can be used to determine the
stoichiometry of the process and to estimate the formation constant of
the mercury complex (see Problems).
Example 4.13 Kirowa-Eisner and Osteryoung (57) have

reported a polarographic method for the determination of
small concentrations of hydroxide ion which takes advantage
of the mercury oxidation wave. The electrode process is
Hg + 20H- ----t Hg(OH)2 + 2 e-
where Hg(OH)2 is a solution species at the very low
concentrations formed in polarography. The half-wave
potential is linear in pOH, varying from 0.150 to 0.117 V (us.
s.c.e.) as [OH-] increases from 10 11M to 0.1 mM. With the static
mercury drop electrode, the technique affords a nearly unique
method for the determination of hydroxide ion in the
micromolar concentration range; the detection limit is 0.3 11M
and a linear response is obtained up to 0.4 mM. Other species
which form mercury complexes interfere, of course; thus CI-,
for example, must be at least five times lower in concentration
than OH-.
3.0
(a) i/flA (b)
o
2.5
0.25 flAI
2.0
1.5
1.0
0.29 mM BzSH
0.5
0.056 mM BuSH
0.0
-0.2 -0.4 -0.6 -0.8 -1.0 0 0.1 0.2 0.3 0.4 0.5
EIV C/mM
Figure 4.45 (a) Differential pulse polarograms of 0.056 mM butyl

mercaptan and 0.29 mM benzyl mercaptan in 0.2 M NaOH for 50 mV
pulses and a negative-going potential scan. (b) Differential pulse peak
currents for the -0.75 V peak (solid line) and for the -0.60 V peak (dashed
line) as functions of concentration for benzyl mercaptan. Reprinted with
permission from R. L. Birke and M. Mazorra, Anal. Chim. Acta 1980,
118, 257, copyright 1980, Elsevier Science Publishers.
Example 4.14 Polarograms of basic aqueous solutions of

thiols show anodic waves corresponding to the formation of
mercury thiolates,
Hg + 2 RS- -t Hg(SR)2 + 2 e-
Birke and Mazorra (58) have shown that a variety of thiols give
differential polarographic peak currents linear in thiol
concentration over the range 0.1-100 flM. Above 0.1 mM, two
peaks are observed (Figure 4.45a), with the total peak current
approximately continuing the calibration curve (Figure 4.45b).
This behavior is reminiscent of adsorption effects in d.c.
polarography (see §4.5) and Birke and Mazorra suggest that
insoluble Hg(SRh deposits on the electrode surface. At low
concentrations, less than monolayer coverage occurs, but above
about 0.1 mM, a multilayer structure is formed in two distinct
steps. The method shows good sensitivity when the medium is
0.1 M NaOH and the thiols are ionized. The sensitivity
decreases markedly when the pH is less than the thiol pKa.
Voltammetry is by no means confined to inorganic analysis.

Voltammetric methods have been developed for many organic functional
groups (59), for organometallics (60), for pharmaceuticals (61-63), and for
compounds of biological interest (64,65,D9). The advent of
microelectrodes has been particularly important for the development of
methods for in vivo determinations of electroactive species in biological
systems (66).
Example 4.15 Albery and coworkers (67) have performed

linear potential sweep voltammetric experiments with a
carbon paste microelectrode, a AgIAgCI reference electrode,
and a stainless steel auxiliary electrode implanted in the brain
of a rat. It was hoped that oxidation peaks corresponding to
catecholamines could be observed. (The catecholamines are
intimately involved in the function of nerve cells.) Several
oxidation peaks were observed as shown in Figure 4.46, and
one of the goals of this research was the identification of the
peaks. In vitro experiments showed that both catecholamines
and ascorbic acid are oxidized at about the potential of peak 1 in
the voltammogram. However, careful experiments in which
catecholamines were injected into the rat's brain showed
small shifts in the peak potential, whereas addition of ascorbic
acid simply increased the peak size; thus peak 1 was assigned
to ascorbic acid. Peak 2 was identified, after similar
experiments, as corresponding to the oxidation of indoles
(probably 5-hydroxyindoleacetic acid) and/or glutathione.
Peaks 3 and 4 were not identified with certainty. Comparison
Figure 4.46 Voltammogram

obtained with a carbon paste
electrode in the striatum of a rat.
The current peaks have been
sharpened by electronic semidiff-
erentiation. Reprinted with per-
mission from R. D. O'Neill, R. A.
Grunewald, M. Fillenz, and W. J.
Albery, Neuroscience 1982, 7, 1945, 0.0 0.2 0.4 0.6
copyright 1982 Pergamon Press. EIV
of the in vivo and in vitro results suggest an ascorbic acid

concentration on the order of 100 J.lM, and the resulting peak
thus obscures the oxidation of much smaller amounts of
catecholamines which might be present.
REFERENCES
(Reference numbers preceded by a letter, e.g. (D4), refer to a book listed in
the Bibliography.)
1. F. G. Cottrell, Z. phys. Chem. 1903,42, 385.

2. J. Heyrovsky and D. Ilkovic, ColI. Czech. Chem. Commun. 1935,7,
198.
3. J. Tomes, ColI. Czech. Chem. Commun. 1937,9,12,81,150.
4. D. K. Roe in Laboratory Techniques in Electroanalytical Chemistry,
P. T. Kissinger and W. R. Heineman, eds, New York: Marcel
Dekker, 1984, p 193.
5. C. K. Mann, Electroanalytical Chemistry 1969,3, 57.
6. A. J. Fry and W. E. Britton in Laboratory Techniques in
7. G. Gritzer, Inorg. Chim. Acta 1977,24, 5; G. Gritzer and J. Kuta,
Electrochim. Acta 1984,29,869.
8. R. N. Adams in Treatise on Analytical Chemistry, 1. M. Kolthoff and
P. J. Elving, eds, Part I, Vol. 4, New York: Interscience, 1963, p 2381.
9. Z. Galus in Laboratory Techniques in Electroanalytical Chemistry,
P. T. Kissinger and W. R. Heineman, eds, New York: Marcel
Dekker, 1984, p 267.
10. G. Dryhurst and D. L. McAllister in Laboratory Techniques in
11. N. Winograd in Laboratory Techniques in Electroanalytical
Chemistry, P. T. Kissinger and W. R. Heineman, eds, New York:
Marcel Dekker, 1984, p 321.
12. F. M. Hawkridge in Laboratory Techniques in Electroanalytical
13. C. N. Reilley, W. D. Cooke, and N. H. Furman, Anal. Chem. 1951,23,
1226.
14. J. T. Stock, Amperometric Titrations, New York: Interscience, 1965.
15. W. H. Reinmuth, Anal. Chem. 1960,32, 1509.
16. P. Delahay in Treatise on Analytical Chemistry, 1. M. Kolthoff and
Voltammetry of Reversible Systems
17. D. G. Davis, Electroanalytical Chemistry 1966,1,157.

18. H. T. S. Sand, Phil. Mag. 1900, 1,45; Z. phys. Chern. 1900,35,641.
19. R. S. Nicholson and I. Shain, Anal. Chern. 1964, 36, 706.
20. R. H. Wop schall and I. Shain, Anal. Chern. 1967,39, 1514.
21. V. D. Parker, Electroanalytical Chemistry 1986,14, 1.
22. M. Goto and D. Ishii, J. Electroanal. Chern. 1975, 61, 361.
23. P. Dalrymple-Alford, M. Goto, and K. B. Oldham, J. Electroanal.
Chern. 1977, 85, 1.
24. P. Dalrymple-Alford, M. Goto, and K. B. Oldham, Anal. Chern. 1977,
49,1390.
25. L. Meites in Treatise on Analytical Chemistry, I. M. Kolthoff and P.
J. Elving, eds, Part I, Vol. 4, New York: Interscience, 1963, p 2303.
26. D. llkovic, CoIl. Czech. Chern. Commun. 1934,6,498.
27. R. Brdicka, Z. Elektrochem. 1941,48,278,686; Coll. Czech. Chern.
Commun. 1947,12,522.
28. H. H. Bauer, Electroanalytical Chemistry 1975,8, 170.
29. G. C. Barker and A. W. Gardner, Z. anal. Chern. 1960, 173, 79.
30. J. R. Morris, Jr. and L. R. Faulkner, Anal. Chern. 1977,49,489.
31. J. Osteryoung and E. Kirowa-Eisner, Anal. Chern. 1980, 52, 62.
32. J. H. Christie, J. A. Turner, and R. A. Osteryoung, Anal. Chern.
1977,49, 1899; J. A. Turner, J. H. Christie, M. Vukovic, and R. A.
Osteryoung,Anal. Chern. 1977,49,1904.
33. A. M. Bond and R. D. Jones, Anal. Chim. Acta 1980,121,1.
34. H. A. Laitinen and I. M. Kolthoff, J. Phys. Chern. 1941, 45, 1079.
35. V. G. Levich,ActaPhysicochim. URSS 1942,17,257.
36. A. C. Riddiford, Adv. Electrochem. Electrochem. Engin. 1966,4,47.
37. R. M. Wightman and D. O. Wipf, Electroanal. Chern. 1989,15,267.
38. C. D. Baer, N. J. Stone, and D. A. Sweigart, Anal. Chern. 1988, 60,
188.
39. A. M. Bond, K. B. Oldham, and C. G. Zoski, Anal. Chim. Acta 1989,
216,177.
40. K. B. Oldham and C. G. Zoski, J. Electroanal. Chem. 1991,313) 17.
41. Z. G. Soos and P. J. Lingane, J. Phys. Chern. 1964, 68, 3821.
42. J. Heinze, J. Electroanal. Chem. 1981, 124, 73.
43. J. Newman, J. Electrochem. Soc. 1966,113,501.
44. K. B. Oldham, J. Electroanal. Chern. 1987,237,303.
45. J. Heinze and M. Srorzbach, Ber. Bunsenges. Phys. Chern. 1986, 90,
1043.
46. J. O. Howell and R. M. Wightman, Anal. Chern. 1984,56, 524.
47. G. J. Hoijtink, Rec. trav. chim. 1955, 74, 1525; Adv. Electrochem.
Electrochem. Engin. 1970,7,221.
48. A. Streitwieser, Molecular Orbital Theory for Organic Chemists,
New York: John Wiley, 1961.
49. D. de Montauzon and R. Poilblanc, J. Organometal. Chern. 1976, 104,
99.
50. J. J. Lingane, Chern. Rev. 1941,29, 1.
References 239
51. D. R. Crow, Polarography of Metal Complexes, New York:

Academic Press, 1969.
52. J. B. Flato, Anal. Chem. 1972, 44 (11), 75A.
53. A. M. Bond, J. Electroanal. Chem. 1981,118,381.
54. J. Heyrovsky, Chem. Listy 1922, 16, 256; Phil. Mag. 1923,45, 303.
55. L. Meites (and co-workers), Anal. Chim. Acta 1956, 14, 390, 482; 1959,
20,397;1978,98,163.
56. L. E. Leon and D. T. Sawyer, Anal. Chem. 1981,53, 706.
57. E. Kirowa-Eisner and J. Osteryoung, Anal. Chem. 1978,50, 1062.
58. R. L. Birke and M. Mazorra, Anal. Chim. Acta 1980,118,257.
59. P. Zuman, Talanta 1965, 12, 1337.
60. M. D. Harris, Electroanalytical Chemistry 1974, 7, 79.
61. G. J. Patriarche, M. Chateau-Gosselin, J. L Vandenbalck, and P.
Zuman, Electroanalytical Chemistry 1979,11,141.
62. H. Siegerman, Electroanalytical Chemistry 1979, 11, 291.
63. M. A. Brooks in Laboratory Techniques in Electroanalytical
64. P. Zuman, in Experimental Methods in Biophysical Chemistry, C.
Nicolau, ed, New York: John Wiley, 1973.
65. A. L. Underwood and R. W. Burnett, Electroanalytical Chemistry
1973,6, l.
66. J. Koryta, M. Bl-ezina, J. Pradec, and J. Pradecova,
Electroanalytical Chemistry 1979, 11, 85.
67. R. D. O'Neill, R. A. Gri.inewald, M. Fillenz, and W. J. Albery,
Neuroscience 1982, 7, 1945:
PROBLEMS
4.1 Show that eqs (4.2) are indeed solutions to the diffusion equations
and that they satisfy the boundary conditions.
4.2 Derive eq (4.18) assuming diffusion of 0 across a linear

concentration gradient of thickness XD.
4.3 Show that inclusion of the activity coefficient ratio in eq (4.1) leads
to another term in the expression for E 112
E 112 = FJO _.B:L In !2s! + MIn 'Yo
2nF DR nF 'YR
4.4 Use the standard potential for the Fe3 +lFe 2 + couple, diffusion
coefficients, estimated from the molar ionic conductivities of Fe3 +
and Fe 2 +, and activity coefficients, estimated using the Debye-
Huckel limiting law, to compute the half-wave potential for the
reduction of Fe3+ to Fe2+ in 0.1 M KCI04 at 25°C.
4.5 A triangular wave signal (needed for potential control in cyclic

voltammetry) can be generated using a voltage integrator such as
shown in Figure 4.8. What kind of input signal would be required?
4.6 An alternative design for a potentiostat connects the reference

electrode directly to the positive operational amplifier input,
eliminating the need for a voltage follower. The potential control
inputs are summed and applied to the negative input. Sketch the
circuit diagram. How would the indicator electrode potential
(relative to the reference) vary for a negative-going ramp as the
potential control input?
4.7 The analysis of the performance of the operational amplifier

circuits in Section 4.2 assumed infinite gain and infinite input
impedance. In real circuits, the gain and input impedance are
finite, typically 104-10 6 and 1-10 5 MO, respectively. Consider the
effect of finite gain and input impedance for each of the following
circuits:
(a) Current follower (the current measuring circuit of Figure 4.7).
Assuming that R2 = 10 kO, estimate the minimum amplifier gain
required to keep the indicator electrode within 0.1 mV of ground
potential when the cell current is 0.1 mAo
Problems 241
(b) Potentiostat (Figure 4.7). Assuming that R = 100 kil, that the
amplifier input impedance is 100 MO, the control potential is 1.00
V, and the auxiliary electrode potential must be -2.00 V, what gain
is required if the magnitude of the reference electrode potential
differs from the control potential by no more than 0.1 mY.
(c) Voltage ramp generator (Figure 4.8). Assuming that R = 10
Mil, C = 10 ~, cl>in = 1.0 V, an input impedance of 10 Mil and a
gain of 106 , estimate the nonlinearity of the ramp; i.e., by how
much does the potential scan rate differ when the output voltage is
o V and when it is -1 V?
(d) Galvanostat (Figure 4.9). Assuming the operational amplifier
parameters of part (c), a control voltage of -1.00 V, resistance R =
10 kil, and a maximum output voltage from the amplifier of 10 V,
what is the maximum effective cell resistance? By how much
would the current then differ from the nominal value?
4.8 Assuming that reduced species R forms a monolayer on the

electrode surface, what is the critical concentration below which
only a polarographic adsorption pre-wave is observed if td = 4 s, U =
1.78 mg sol, DO = 10- 9 m 2 s- l , and each adsorbed R molecule
occupies a surface area of 0.25 nm 2 ?
4.9 In the polarogram of 0, which is reduced to R, the main reduction

wave is preceded by an adsorption pre-wave. On the plateau
following the pre-wave, the current is limited by the available
surface area on the electrode. Assuming that the concentration of
o is large, that only a monolayer of R is adsorbed, and that the
potential is on the pre-wave plateau, prepare a plot of the current
variation with time during the life of one drop; compare with
Figure 4.27.
4.10 In a potential step chronoamperometry experiment, the observed

current includes a contribution due to charging of the double-
layer capacitance. Consider an experiment in which species 0 is
reduced by one electron at a disk electrode 2 mm in diameter with
a double-layer capacity Cd = 0.5 f m- 2 in contact with a solution of
effective resistance R = 1500 o. Assume CO* = 1 mM, DO = 10- 9
m 2 s- l and a potential step from E = E1I2 + 0.25 V to El/2 - 0.25 V.
(a) Compute and plot the faradaic current over the time range 0-
20ms.
(b) On the same graph plot the capacitive charging current and
the total current.
(c) In practice, how could the interference of capacitive charging
current be avoided?
4.11 The technique of chronoamperometry can be improved for

analytical purposes by electronically integrating the current and
recording charge vs. time curves. The experiment is then called
chronocoulometry.
(a) Show that the chronocoulometric response for a reversible
electrode process is
Q = 2nFACo*VD othc
(b) One advantage of chronocoulometry over chronoamperometry
is that the signal proportional to concentration grows with time
rathen than decaying. However, the current contribution from
double-layer charging is also integrated and thus contributes to
the signal long after the capacitive current has decayed to zero.
Use the data of Problem 4.10 to compute the faradaic and capacitve
charges and the total charge and plot these vs. t and Vi.
(c) In practice, it is not difficult to separate the faradaic and
capacitive contributions to Q. Describe how this would be done in
reference to the data plotted in part (b).
4.12 What current should be used to obtain a chronopotentiometric

transition time of 10 s if the concentration of electroactive material
is 0.1 mM, the electrode is a disk of 1 cm diameter, D = 5 x 10- 10
m 2 s- 1 and the electrode process is a one-electron reduction?
4.13 (a) Use eqs (4.29), (4.35) and (4.37) to derive an expression for
semiderivative of the current as a function of time in a linear
sweep voltammetry experiment.
(b) How does the scan rate dependence of the peak signals in
semi-integral and semi-derivative presentation compare with the
peak current in linear sweep voltammetry?
4.14 A series of rotating disk voltammograms at varying rotation

speeds gave the following half-wave potentials:
ro/rad s-1 ID 50 100 200 500

E1/2N -0.699 -0.703 -0.707 -0.713 -0.724
(a) Assuming that the electrode process is reversible and that the
shift in half-wave potential is due to solution iR drop, determine
the "true" value of E1I2.
(b) If the limiting current for ro = 500 rad s-1 is 42.0 1lA, what is the
effective solution resistance?
Problems 243
4.15 The assumption of linear diffusion breaks down when the

dimensions of the electrode are comparable to the thickness of the
diffusion layer. Suppose that a chronoamperometry experiment
lasts for one second, and that the electroactive material has a
diffusion coefficient DO = 10- 9 m 2 s- 1 . What is the minimum
diameter of the electrode if we require that the diameter be more
than ten times the maximum diffusion layer thickness?
4.16 The diffusion-controlled current at a microelectrode does not

decay to zero but reaches a time independent value. In a potential
step chronoamperometry experiment, what electrode radius is
required if the current is to fall to within 10% of its limiting value
in 0.1 s. Assume a diffusion coefficient of 10-9 m 2 s- 1 .
4.17 The steady-state voltammogram of 1.00 mM A at 230 K, using a

microdisk electrode with ro = 5.0 ~m, is shown in Figure 4.47.
Compute the number of electrons n, the half-wave potential, El/2,
and the diffusion coefficient D. Some points read from the
voltammogram are given in the table.
EN: -0.12 -0.16 -0.20 -0.24

ilnA 0.003 0.020 0.127 0.448
EN: -0.28 -0.32 -0.36 -0.40
ilnA 0.676 0.725 0.732 0.733
0.8
ilnA
0.6
0.4
0.2
Figure 4.47 Steady-state

voltammogram of A at 230 0.0
K with a microdisk -0.1 -0.2 -0.3 -0.4
electrode. EIV
4.18 (a) The resistivity of acetone containing 0.3 M Et4NPF6 varies

from 0.75 nom at 298 K to 5.8 nom at 198 K. Compute the ohmic
potential drop at the two temperatures for the steady-state
voltammetry of 1.0 mManalyte (n = 1, D =5 x 10- 10 m 2 s- 1) at a I-11m
radius microdisk electrode.
(b) Compute the ohmic potential drop for the same system at the
same electrode but for a rapid-scan voltammogram with v = 10 4 V
s·l.
4.19 Peak currents in microelectrode voltammetry are given by eq (4.31)

for fast scans and by eq (4.50) when the scan is slow enough that
steady-state conditions prevail. Use Figure 4.40 to estimate the
maximum scan rate v for which eq (4.31) is appropriate and the
minimum scan rate for which eq (4.50) can be used. Assume n =
1, D = 4 x 10- 10 m 2 s- 1 , T = 298 K, ro = 10 1lID.
4.20 Huckel molecular orbital theory gives energies in units of a

parameter p, the C-C bond resonance integral. Use the half-wave
potential correlations shown in Figure 4.42 to obtain an estimate
for Pin units of eV.
4.21 The pulse sequence for differential pulse polarography shown in

Figure 4.31 gives current-potential curves like those of Figure 4.32.
Suppose that the pulse sequence is rearranged so that positive-
going pulses are superimposed on a negative-going ramp. Sketch
the resulting current-potential curve. How is the peak potential
related to the half-wave potential?
4.22 An anodic wave is observed for solutions containing halide ions or

other species which coordinate Hg(II). The wave corresponds to
the process
Hg + p X- -7 HgXp2-p + 2 e-
Show that the wave has the diffusion current, half-wave potential,
and i - E shape given by
iD = - 2FAkDCX*lp
El/2 = EHgO - ~J In [pPJCx/2r--1]
E = EJJ2 - ~J In W~) (2(i~D- Of]

where Pp is the overall formation constant for HgXp 2-p, kD is the
mass transport rate constant (all diffusion coefficients are
assumed equal), and Cx* is the bulk concentration of X-. Hint:
Start with the Nernst equation for the Hg2+/Hg couple and
Problems 245
expressions for the current in terms of the flux of x- toward the

electrode and the complex away from the electrode.
4.23 Plot ius. E for mercury oxidation waves in the presence of a

coordinating ligand X- assuming that p = 2 and p = 4; compare
with the wave shape expected in the absence of ligands, Figure 4.5,
and with a normal reversible two-electron wave. What are the
differences?
4.24 Describe an experimental arrangement for an amperometric

acid-base titration of hydroxide ion in aqueous solution, taking
advantage of the anodic wave described in Example 4.13. What
titrant should be used? At what potential should the current be
monitored? Sketch the titration curve.
4.25 The diffusion current constants given in Table 4.1 are uniformly
smaller for polarograms in tartrate buffer solutions than for the
other media listed. Give a qualitative explanation for this
phenomenon.
4.26 Using the data of Figure 4.44, estimate the equilibrium constant
for
Pb 2+ + 3 OR- ~ HPb02 2- + H20
The half-wave potential for the polarographic reduction of Pb 2 + in
acid solution is -0.388 V us. s.c.e.
4.27 Kirowa-Eisner, Talmor, and Osteryoung [Anal. Chem. 1981,53,

581]. studied the polarographic oxidation of mercury in the
presence of cyanide ion using pulse and sampled d.c.
polarography on a static mercury drop electrode. A wave for 50.8
11M cyanide is shown in Figure 4.48. Potentials were measured
us. a Ag/AgCI (saturated KCl) reference electrode, E = 0.205 V, and
a few points are given in the following table:
EIV illlA EN i/llA

-0.299 -0.0112 -0.250 -0.0963
-0.282 -0.0321 -0.241 -0.1124
-0.273 -0.0481 -0.228 -0.1284
-0.266 -0.0642 -0.209 -0.1444
-0.255 -0.0802 -0.000 -0.1605
Figure 4.48 Sampled d.c.

polarogram at a 8.m.d.e. -0.4 -0.3 -0.2 -0.1 -0.0
for 50.8 ~M cyanide. EIV
(a) Using the equations given in Problem 4.22, determine p and

El/2 by plotting In(iDI2i)[2(iD - i)/iDJP us. E for several values of p.
(b) Use the half-wave potential found in part (a) to estimate the
formation constant for the mercury(II) cyanide complex.
(c) Polarograms for higher concentrations of cyanide had shapes
which suggested formation of complexes with larger values of p,
at least in the leading part of the wave. Measured diffusion
currents, however, were linear in the bulk cyanide concentration
over the range 0.003-10 mM. Explain why the diffusion current
would correspond to a constant value of p when higher complexes
are apparently formed in some instances.
5
MECHANISMS OF
ELECTRODE
PROCESSES
One of the most interesting applications of voltammetric techniques

has been in the study of reactions initiated by oxidation or reduction at
an electrode. In this chapter, we consider the mechanisms of electrode
processes influenced by coupled homogeneous chemical reactions. In
§5.1, we review some principles of chemical kinetics and discuss the
kinds of reactions which may be initiated by electron transfer. In §5.2,
we introduce some techniques of spectra electrochemistry which can be
used with voltammetric experiments to deduce mechanisms. In §5.3 -
§5.5, we consider applications of the voltammetric methods introduced in
Chapter 4 to several kinds of mechanistic problems.
We will assume that the electron-transfer processes, per se, are
nernstian. Irreversibility can be introduced using the methods of
Chapter 6, but the results would complicate our study unnecessarily.
For further details on these topics, see the monographs edited by Fry and
Britton (C9) and Lund and Baizer (E7) and reviews by de Montauzon,
Poilblanc, Lemoine, and Gross (1), Geiger (2) and Connelly and Geiger
(3) on organometallic electrochemistry, and Zuman (4) and Hawley (5)
on organic electrochemistry.
5.1 INTRODUCTION
Rate Laws and Mechanisms

In conventional kinetic studies, the rate of a reaction is expressed in
terms of a rate law, that is, an empirical equation which summarizes
the dependence of the experimentally measured rate on the
concentrations of reactants, products, catalysts and inhibitors. Thus, for
example, the reaction
2A+2B ~ D
might be found to obey the rate law
d[A] = _ hci:s [A] [B]2
dt
247
248 Mechanisms of Electrode Processes
The rate law tells us that the rate has been found to increase linearly
with the concentration of A but to increase as the square of the B
concentration. We refer to the exponents of [A] and [B] as the kinetic
orders of the reaction in A and in B. Thus the reaction is first-order in A
and second-order in B, third-order overall. Notice that the units of the
rate constant k depend on the overall order.
In electrochemical kinetic studies, rate information comes from the
measured current i (in A =C s·l). By the simple expedient of dividing by
i by nF, we have the rate in units of mol s·l. Thus unlike classical
kinetic studies where the approach to the rate law is often through
measured concentrations and integrated rate laws, electrochemical
rates can be measured directly. Otherwise, however, the concept of
empirical rate law, the methods used to determine kinetic orders, and
the interpretation of kinetic information are very similar in conventional
and electrochemical kinetics.
The mechanism of a chemical reaction is made up of a series of
elementary processes, i.e., uncomplicated single reaction steps in which
the reactant and product species are exactly those specified. For an
elementary process, the kinetic order in a chemical species is identical
to the molecularity, i.e., the stoichiometric coefficient of that species in
the step. Thus, given a mechanism, the rate law governing the rate of
formation or disappearance of a chemical species can be written by
inspection. For example, given the mechanism
k1
A+2B ~ C
k2
A+C ~ D
the predicted rate laws for each of the species can be written:
d[A] =-kdA][B]2_ k 2[A][C] d[B] =-2kdA][B]2
dt dt
d[D] =k2 [A] [C]
dt
Each occurence of the first step consumes one A and two B molecules;
thus the contribution to d[B]/dt is twice the contribution to d[A]/dt;
doubling the concentration of A should double the rate of the first step,
but doubling [B] should increase the rate by a factor of four. Notice that
the stoichiometry of the two steps requires
[A] + [C] + 2 [D] =[A]o [B] + 2 [C] + 2[D] = [B]o
so that we have a constraint on the rate equations
d[A] + d[C] + 2 d[D] = d[B] + 2 d[C] + 2 d[D] = 0
dt dt dt dt dt dt
§5.1 Introduction 249
Thus there are only two independent rate equations.

The empirical kinetic orders are related to the stoichiometric
coefficients of the overall reaction only when the reaction takes place in a
single elementary step. However, the kinetic orders do tell us the
composition of the activated complex, the highest energy point on the
reaction coordinate. Thus in this example, the empirical rate law tells
us that the activated complex contains one A and two B's and is
consistent with the mechanism we have written if the first step is rate-
limiting. Thus C is apparently a short-lived intermediate.
It is sometimes convenient to apply the steady-state approximation
to mechanisms involving transient intermediates. Since C never
increases to an appreciable concentration, its time derivative must be
small. If we set d[C]/dt = 0, the rate expression above becomes an
algebraic equation
k 1 [A][B]2 - k2[A][C] = 0
whence [C) = (k1/k2) [B]2. Substituting this into the d[D]/dt expression,
we have
d[D] =k2 [A] [C) =kl [A][B]2
dt
which matches the empirical rate law with kobs = k 1.
One of the advantages of electrochemical kinetic studies using
steady-state voltammetry is that all time derivatives can be set equal to
zero-the entire reacting system is in a steady state. However,
concentrations are functions of distance from the electrode so that the
problem is not entirely trivial.
Multielectron Processes
In electrode processes, we sometimes write a step such as
o +ne- ~ R
and treat it as if it were an elementary process. In fact, electrode
processes rarely, if ever, involve the concerted transfer of more than one
electron. Gas-phase electron attachment or ionization reactions always
proceed in discrete one-electron steps. Even if the same molecular
orbital is populated or ionized, electron repulsion will cause the two
steps to occur at well separated energies. For electron-transfer reactions
at an electrode-solution interface, solvation effects may bring the two
electron-transfer steps closer together in energy, but only in exceptional
cases would we expect the two steps to coincide.
Electrode processes involving two or more electrons are analogous
to rate laws for gas-phase reactions with overall kinetic orders of three
or more: they provide evidence for mechanisms of two or more
elementary steps. However, a process involving two one-electron steps
separated by a chemical step may be experimentally indistinguishable

from a concerted two-electron transfer if the chemical step is very fast.
In discussions of the mechanisms of electrode processes, we usually
write the simplest mechanism consistent with the experiment at hand.
If the process involves a chemical step which goes to completion in a
time short compared with the characteristic time of the experiment, it is
then reasonable to treat the process as if it involved a concerted transfer
of more than one electron. It should be remembered, however, that an
experiment with a shorter characteristic time may not be consistent
with the simplified mechanism.
Experimental Time Scales

Each of the experimental techniques of electrochemistry has a
characteristic time scale. In assessing the effect of a coupled chemical
reaction on the experimental response, we must ask how the
characteristic time of the reaction, say the half-life tl/2, compares with
the characteristic time of the experiment. If the reaction time is very
short compared with the experimental sampling time, the experiment
will "see" the reaction at equilibrium. On the other extreme, if the
reaction time is long on the experimental time scale, the reaction may
have no discernible effect on the experimental response. When the
reaction time lies within the experimental time scale, the experimental
response will be sensitive to the rate of the chemical step and, at least in
principle, the chemical rate constant will be determinable from the
experiment.
For techniques such as chronoamperometry, chrono-
potentiometry, and d.c. polarography, where the current or potential
is measured at a well-defined elapsed time after the start of the
experiment, the time scale is straightforward. The accessible time scale
for d.c. polarography is essentially the range of drop times, 1-10 s.
Chronopotentiometric and chronoamperometric experiments may last
as long as 100 s with potentials or currents measurable about 1 ms after
beginning the experiment. Thus the time scale for these techniques is
roughly 1 ms - 100 s. In practice, however, there are severe problems
with vibration or other departures from pure diffusion for experiments
that last more than 10 s or so.
In steady-state voltammetry using a microelectrode or rotating-
disk electrode, the time scale is set by the residence time of a diffusing
species at the electrode surface, on the order of (Ox)2/2D, where Ox = ro for
a microelectrode or XD (oc l/Vro'J for an r.d.e. Thus the time scale is set by
the diffusion coefficient and the electrode size or rotation speed. Times
range from 10-1000 ms for the r.d.e. down to 0.005-50 rna for
microelectrodes.
In cyclic voltammetry, the time scale is given by the range of the
parameter RTIFu, where u is the potential scan rate. The range of u is
roughly 0.02-100 V s-1 for ordinary size electrodes, so that the accessible
time scale is on the order of 0.2 - 1000 ms. With microelectrodes, scan
rates may be as fast as 105 V s-1, extending the time scale down to less
than a microsecond.
Another way of looking at the effect of a chemical reaction on an
electrochemical experiment is to compare the diffusion layer thickness,
XD, with the characteristic distance from the electrode through which a
reactant or product diffuses before the chemical reaction comes to
equilibrium. This so-called reaction layer thickness is typically on the
order of magnitude
XR = YDlk
where D is the diffusion coefficient and k is the first-order (or pseudo-
first-order) rate constant for the chemical step. If the reaction layer is
much thicker than the diffusion layer, XR» XD, the reaction is very slow
and the effects of the reaction rate on the experimental response may be
negligible. When XR «XD, the reaction comes to equilibrium very close
to the electrode surface and the experimental response to the
equilibrium mixture may again be independent of the rate of the
chemical step. When XR and XD are comparable in magnitude, the
experimental response is expected to depend critically on the rate of the
chemical step.
Chemical Reactions Initiated by Electron Transfer

Consider an electron-paired molecule in which main group atoms
obey the octet rule and transition metals (if any) obey the lS-electron rule.
The electron added in a one-electron reduction usually occupies an
antibonding molecular orbital.1 The resulting radical anion may be
stable on the time scale of the experiment, in which case the electrode
process is chemically reversible. If the electron-transfer step is fast, the
process will also be electrochemically reversible, i.e., nernstian. Often,
however, the initially formed radical anion undergoes a rapid chemical
transformation. There are three possibilities for the primary chemical
step:
(1) Cleavage of a chemical bond. The antibonding electron may
weaken a bond to the point that the activation barrier to bond cleavage
can be surmounted by thermally accessible energies. Bond cleavage will
then be rapid. For example, one-electron reduction of an alkyl or aryl
halide is followed by rapid loss of halide ion:
fast
RX + e- ~ RX7 ~ R + X-
1 Transition metal complexes often have empty essentially nonbonding orbitals

which can accommodate the added electron; somewhat different reactions then may
occur.
Similarly, organometallic molecules and coordination complexes often

lose a ligand on reduction:
fast
MLn + e- ~ MLn:" ~ MLn-l· + L-
In either case, an electron-deficient radical is formed which almost
always undergoes a secondary reaction. Common follow-up reactions
include further reduction, e.g.,
R + e- ~ R-
Generally speaking, further reduction occurs if the bond cleavage
reaction is fast enough that the radical species is formed at the electrode
surface. If the neutral radicals are formed less rapidly, but still fast
enough that radical-radical encounter is likely, we expect dimerization:
2R ~ R-R
If the radical is formed more slowly and farther from the electrode, it
may abstract a hydrogen atom from the solvent or supporting electrolyte,
R· + SH ~ RH + S·
or react with a nucleophile,
R + Nu- ~ RNu:"
Further reactions are expected from most of these secondary products.
The general scheme accommodates many other apparently disparate
electrode processes. Thus, for example, the two-electron reduction of
dicobalt octacarbonyl can be understood in terms of the mechanism
C02(CO)S + e- ~ C02(CO)8:" ~ CO(CO)4- + CO(CO)4·
Co(COk + e- ~ Co(COk
Here the metal-metal bond is broken in the primary step and the
resulting radical is reduced in the secondary step.
(2) Electrophilic attack. A radical anion generally is a strong
nucleophile, highly susceptible to electrophilic attack. In protic solvents,
the most readily available electrophile is often the hydrogen ion. Thus
the first reaction step of unsaturated radical anions often is protonation.
Anthracene, for example, undergoes an overall two-electron reduction to
9,lO-dihydroanthracene. The sequence of steps is
Cl~lO + e- ~ C14HlQ:"
Cl~lQ:" + H+ ~ C14HU·
Cl~ll· + e- ~ C14Hll-
Cl~ll- + H+ ~ C14H12
The neutral radical produced by protonation of the anthracene anion
radical is more easily reduced than anthracene itself, so that the second
electron-transfer step is fast at the potential of the first step. Similarly,
nitro compounds undergo a six-electron, six-proton reduction to amines:
Ar-N02 + 6 H+ + 6 e- -7 Ar-NH2 + 2 H20
Depending on the relative rates of the protonation steps, the nitroso (Ar-
NO) and hydroxylamino (Ar-NHOH) intermediates may be isolable.
(3) Rearrangement. A radical anion can sometimes undergo a
change in structure which reduces the energy of the molecule by better
distributing the excess charge. If the rearranged radical is stable on the
time scale of the experiment, reoxidation regenerates the starting
compound in a different conformation, usually at a different potential
than the reduction. Conformational rearrangement is probably a very
common primary process; however, if the rearranged radical then
undergoes bond cleavage or electrophilic attack, there may be no direct
evidence of the rearrangement step in the experimental results.
A similar analysis can be applied to electrode processes initiated by
an oxidation step. The resulting electron-deficient cation radical may be
stable, but more often will react by dimerization, by bond cleavage (this
time the bond is weakened by removal of a bonding electron), by
nucleophilic attack, or by rearrangement. See Alder (6) for an
interesting discussion of electrophilic and nucleophilic substitution
reactions initiated by electron transfer. We will discuss several other
examples in §7.3 in the context of organic electrosynthesis.
Example 5.1 Bond, Colton, and McCormick (7) studied the

electrochemical oxidation of Mn(CO)3(dppm)CI (dppm =
Ph2PCH2PPh2). Cyclic voltammograms of this species at a Pt
electrode in acetonitrile solution (0.1 M Et4NCI04, 500 mV s-l,
22°C and -35°C) are shown in Figure 5.1. At room
temperature, the one-electron oxidation is nearly reversible,
but a small reduction peak, matched by an oxidation on the
second anodic scan, suggests some decomposition of the
Mn(II) product. The cyclic voltammogram at -35°C shows
complete chemical reversibility of the oxidation with no trace of
the second electrode process. Half-wave potentials-i.e., E1I2 =
(Epa + E pc )/2-for the primary and secondary processes were
found to be +1.44 and +0.98 V, respectively (vs. Ag/AgCI in
CH3CN). Very similar results were obtained in acetone
solutions. Controlled potential electrolytic oxidation gave
confusing results (see below) but chemical oxidation by NOPF6
in CH2Cl2 gave a red compound which analyzed as
[Mn(CO)3(dppm)CI]PF6. When this compound was dissolved
in acetonitrile, the red color slowly faded to yellow and the

grams of Mn(CO)3(dppm)Cl
in acetonitrile solution at (a)
22°C and (b) -35°C. In (a), the
dotted curve represents the
first cycle and the solid curve
the second and subsequent
cycles. Reprinted with per-
mission from A. M. Bond, R.
Colton, and M. J. McCormick,
Inorg. Chern. 1977.16, 155,
copyrigh t 1977, American 1.6 1.2 0.8 0.4
Chemical Society.
EIV
solution showed a chemically reversible oxidation at +0.98 V

with no trace of the +1.44 V oxidation. In another experiment,
Mn(CO)s(dppm)CI was oxidized by NOPF6 in CH3CN at -35°C
to give a dark green solution. When this was warmed to room
temperature, the solution rapidly changed to red and then
more slowly to yellow.
Apparently, the red and green species are isomeric forms of

[Mn(CO)3(dppm)Cl]+. Only two isomers are possible for these
complexes, the facial (rae) and meridional (mer) isomers:
o 0
C CO C CI
CI:+CO OC:+CO
t-p t-p
rae- Mn( CO)f., d ppm)CI mer-Mn(CO)f.,dppm)CI
Since the starting material was known to be rae, the green
form of the MnOn complex, which is related to the Mn(!)
starting material by reversible electron transfer, is presumably
also rae. The red form must then be the mer isomer. Infrared
spectra of the red Mn(H) complex and its reduced form were
consistent with mer stereochemistry.
The results described above indicate that fa e-
[Mn(CO)3(dppm)Cl]+ is rapidly converted to the mer isomer at
room temperature but only slowly at -35°C. The mer-
[Mn(CO)3(dppm)CI]+, however, slowly oxidized acetonitrile (but
not CH2CI2) at room temperature to produce mer-
Mn(CO)3(dppm)Cl. (This reaction caused the controlled-
potential electrolysis experiments to lead only to confusion.)
The results can then be summarized by the following scheme:
-e
fae-Mn(CO)g(dppm)CI. .. fae-Mn(CO)idppm)CI+
+1.44 V Ifast
mer-Mn(CO)aCdppm)CI =--e t
mer-Mn(CO)aCdppm)CI
" +0.98 V /
~
CH 3CN
Several additional points of interest should be noted. Whereas
the fae+ ~ mer+ isomerization is fast at room temperature, the
interconversion of the isomeric forms of the d 6 Mn(1) species is
very slow. Given the well-known kinetic stability of the
isoelectronic Co(lII) complexes, this suggests that the Mn(l)
species are low spin. If we take the E 112 values to be
approximations to the standard reduction potentials, the
standard free energy change for the process
mer + rac+ ~ mer+ + rae
flGo = -0.46 For -44 kJ mol-I, corresponding to an equilibrium
constant of 6 x 10 7 . The mer isomer is clearly the
thermodynamically favored form of the Mn(II) species, and
since it is normally obtained in preparative work, the fae
isomer is probably the favored form of the Mn(1) complex.
(Rationalization of this result in terms of molecular orbital
theory arguments is left to the reader as an exercise.)
The spectroscopic results yield one further insight. High-spin

d 5 Mn(II) complexes are usually nearly colorless, whereas the
Mn(II) complexes in this case were highly colored (the red
mer+ and green fae+ isomers), indicating that the Mn(II)
species are also low spin.
Standard Mechanistic Schemes

In Example 5.1, the mechanism could be deduced from essentially
qualitative observations-the lack of chemical reversibility in a cyclic
voltammogram or the identification of secondary electrode processes.
Mechanistic work often proceeds on the basis of just such qualitative
arguments. Electrochemistry is capable of providing more quantitative
information about rates if the measured currents or potentials are
affected by the rates of the chemical steps. Although the details of the
chemistry may vary enormously from one system to another,
electrochemists have found that most electrode processes fall into one of
a rather small number of mechanistic schemes. In general an electrode
process consists of a series of electron-transfer steps, designated by the
symbol E, and chemical steps, designated by C. In §5.3 - §5.5, we will
discuss the effects of the following mechanistic schemes on
voltammetric experiments:
(1) Preceding Reaction (CE). The electroactive species 0 is in
equilibrium with an electroinactive precursor Y:
Yf:!O
0+ e- f:! R
(2) Following Reaction (EC). The electroactive product is
converted to an electroinactive species Z:
0+ e- f:! R
R~Z
(3) Catalytic Reaction (EC'). A special case of the EC

mechanism where the fol1owing reaction is the oxidation of R to 0 by an
electroinactive species P:
0+ e- f:! R
R+P~O+Q
In this scheme, the reduction of P must be thermodynamically more

favorable than that of 0, EOP/Q > EOO/R, but the electrochemical
reduction of P is too slow to be observable.
(4) Following Reaction with Electroactive Product (ECE).
The product of an electron-transfer step is converted to a species which
is electroactive. There are two possibilities:
(i) The product of the chemical step is reducible (the ECE
mechanism):
02 + e- ~ R2 (E2°)
If E is less than either E1° or E2° and the R1 ~ 02 conversion is fast
enough, the overall process may appear to involve two electrons. If E 1° >
E > E2°, the electrode process corresponds to the simple EC mechanism.
(ii) The product of the chemical step is oxidizable (the ECE
mechanism):
01 + e- ~ R1
R1 ~ R2
02+ e- ~ R2
When E is less than either E1° or E2° this mechanism corresponds to a
simple EC process, but with E1° > E > E2°, R2 is oxidized to 02. Thus if
the R1 ~ R2 conversion is fast enough, the current may approach zero.
In either of the ECE variations, homogeneous disproportionation
may be important:
R2+01 ~ 02+Rl
These four mechanisms, or combinations thereof, cover most
known electrode processes.
While voltammetric techniques provide the first line of attack in
studies of electrode processes, few successful investigations are confined
to electrochemical methods. Victory in the mechanistic game goes to the
scientist who is best able to piece together many bits of evidence from
spectroscopic, synthetic, structural and/or kinetic studies, as well as
from electrochemical work, to construct a mechanism capable of
explaining all the results. The kinds of thought required are very
similar to those involved in establishing mechanisms for ordinary
homogeneous reactions.
5.2 SPECTROELECTROCHEMISTRY
The application of spectroscopic methods to electrochemical

problems is rather broadly termed spectroelectrochemistry. A wide
variety of techniques have be applied to study electrode surfaces and the
electrode-solution interface, including X-ray surface scattering, X-ray
absorption, surface-enhanced Raman, photoemission, Mossbauer,
ultraviolet/visible and infrared reflectance spectroscopy. UV/visible and
IR transmission spectrophotometry and electron spin resonance
spectroscopy have played important roles in the identification of solution
species generated by electrolysis.
Space does not permit a discussion of all the above methods, and we
will confine our attention here to methods for the detection and
identification of electrode products and the use of that information in
deducing mechanisms. For further details on these and other methods,
two recent monographs, edited by Gale (C10) and Abruiia (Cll), are
recommended, together with general discussions by Bard and Faulkner
(B12) and Greef, Peat, Peter, Pletcher, and Robinson (B13).
Electron spin resonance

In electron spin resonance spectroscopy (ESR, also known as
electron paramagnetic resonance or EPR), the degeneracy of the electron
spin energy levels is lifted by application of a magnetic field. A constant
frequency radiation source, usually in the microwave region (v "" 9 GHz),
is used and photons are absorbed when the magnetic field has adjusted
the energy levels with a spacing hv, as shown in Figure 5.2a. Since only
paramagnetic species-free radicals, radical ions, and certain
transition metal complexes-have unpaired electrons, the observation of
an ESR signal in a spectroelectrochemistry experiment can be taken as
evidence for the formation of such a species in an electrode process. See
reviews by Goldberg and McKinney (8), Compton and Waller (9), and
Rieger (10) for further details and examples of the method.
Most ESR spectrometers operate with a small-amplitude field
modulation and detect absorption of photons by phase-sensitive detection
1st derivative
of absorption
Magnetic Field, B Magnetic Field, B

Figure 5.2 Ca) Energy levels for an unpaired electron in a magnetic
field, showing the ESR transition. Cb) Energy levels for an electron
interacting with a spin 1 14N nucleus showing the ESR transitions. (c)
and (d) ESR absorption lines and first-derivative signals corresponding
to the transitions in Ca) and (b), respectively.
Sec. 5.2 Spectroelectrochemistry 259
at the field-modulation frequency; thus ESR spectra are usually plotted
as first derivatives of absorption as shown in Figure 5.2c. ESR sensitivity
is generally quite good; as little as 10- 12 moles of a paramagnetic species
can be detected when the spectrum consists of a small number of sharp
lines.
Because nuclear spins--1H, 19F, 31p, 55Mn, 59Co, etc.-interact with
the electron spin to split the energy levels, ESR spectra often exhibit
more than one line, and the hyper(ine splitting pattern can sometimes be
used to identify the paramagnetic species. Thus for a radical species in
which the unpaired electron interacts with a 14N nucleus (nuclear spin
I = 1), the electron spin levels are split into three sublevels each,
corresponding to m[ = ±1 and o. Transitions are allowed which "flip"
electron spins or nuclear spins, but not both at once; thus three ESR
transitions are allowed and three lines are observed in the spectrum, as
shown in Figure 5.2b,d.
The first electrochemical application of ESR spectroscopy was in
1958 when Austen, Given, Ingram, and Peover (11) showed that free
radicals are produced in the one-electron reduction of anthracene,
benzophenone, and anthraquinone in DMF solution. Samples were
withdrawn from an electrolysis cell, frozen in liquid nitrogen and
transferred to an ESR spectrometer where spectra of the frozen solutions
were obtained. External electrolysis cells have been used frequently
since 1958, but it is somewhat more convenient to link the two
experiments more closely. The simplest approach is to mount an
electrochemical cell immediately above or below the ESR sample tube
(12). The solution is electrolyzed and allowed to flow through the sample
tube either continuously or in batches. A cell designed for reduction
processes is shown in Figure 5.3. This approach has several
advantages, but the ESR-active products must have lifetimes on the
order of minutes in order to be detected. Thus a more common approach
is to use an electrolysis cell with the working electrode in the ESR
sample tube (13). Continuous passage of current then builds up a
steady-state concentration of paramagnetic products; radicals with
lifetimes down to tens of seconds then become detectable.
In most ESR spectrometers, the sample is contained in a tube or
thin-layer cell mounted in a "cavity," a box with conducting walls in
which a microwave standing wave is maintained. The spectrometer
measures absorption of microwave power in the cavity. Molecular
dipoles absorb microwaves to excite rotational motion (as in a microwave
oven). In order to minimize this dielectric absorption of microwave
power, the sample is centered on an electric field node of the standing
wave and is generally less than 4 mm in diameter or thickness.
This geometry imposes severe restrictions on the design of an
electrochemical cell for use in an ESR spectroelectrochemistry
experiment. A simple design, shown in Figure 5.4, places the working
electrode in the cavity with the reference and counter electrodes in a
counter
electrode Luggin probe
to solution connecting
reservoir reference
electrode
t.
Hg pool
cathode
Figure 5.3 Extra muros electro- ESR
lysis cell for ESR spectro- sample
electrochemical experiment. tube
larger chamber above. Although such cells work for many applications,
there are some serious problems. Because of ohmic potential drop along
the tube, most of the current flows at the tip of the working electrode.
The current is usually small, on the order of a few IlA so that the steady-
state concentration of short-lived radicals is low. This problem can be
partially solved by replacing the wire by a Pt gauze electrode in a thin-
layer cell. However, in a long thin cell with the reference electrode
separated from the working electrode, there is very little control of the
working electrode potential; indeed many workers simply eliminate the
counter
electrode '/- Luggin probe
connecting
,.,.' reference
~, electrode
microwwave
~--~r---~ca~ty
Figure 5.4 Intra muros electrolysis

cell for ESR spectroelectrochemical working
experiment. electrode
reference electrode. When the electrode process is simple with only one
possible oxidation or reduction process, potential control is unimportant,
but in more complex cases, the relation of the species generated to the
electrode process may be tenuous.
A great improvement in both the ESR and electrochemical aspects
of the experiment is possible with a cell described by Allendoerfer (14)
and shown in Figure 5.5. A helical coil of wire (6 mm diameter) is used
as the working electrode; the reference electrode is connected through a
Luggin probe extending inside the helix. The working electrode also
serves as the inner cylinder in a coaxial cylindrical microwave cavity.
Microwaves do not penetrate inside the helix so that only radicals near
the electrode are observable. Because of the large electrode area (ca. 22
cm 3), a large current can pass, ESR sensitivity is good and the ESR
signal is immediately responsive to the electrode process.
(b)
ESR active
volume
.1.·
N
.).
wire
• •
l.~ '~::::
Figure 5.5 Allendoerfer ESR spectroelectrochemical cell. (a) Cutaway

drawing showing mounting of cell in a cylindrical cavity; the reference
and counter electrodes (not shown) are mounted above the cavity, with a
Luggin probe extending inside the helical working electrode. (b)
Expanded view of the helical working electrode showing the ESR active
volume.
Example 5.2 Rieger, Bernal, Reinmuth and Fraenkel (12)

studied the polarographic reduction of tetramethylammonium
1,I,2,3,3-pentacyanopropenide in DMF solution containing 0.1
M Pr3NCI04. Three one-electron waves were observed with
half-wave potentials -1.75, -2.25, and -2.75 V us. Ag/AgCI04 (in
DMF). Controlled potential electrolysis of the yellow solution at
the potential of the first wave resulted in a red solution which

gave the ESR spectrum shown in Figure 5.6. The spectrum
can be described as a 1:1:1 triplet of 1:4:10:16:19:16:10:4:1 nonets,
consistent with hyperfine coupling to a single 14N nucleus (1 =
1, aN = 2.64 G) and to four equivalent 14N nuclei (aN = 0.76 G).
Thus the reduction product is almost certainly the expected
radical dianion, [(NC)2CC(CN)C(CN)2]2-. When the reduction
was continued at the potential of the second polarographic
wave, the ESR spectrum decreased in intensity and eventually
disappeared. Continuing the reduction at the potential of the
third wave, however, resulted in the appearence of another
ESR spectrum, this time a 1: 1 doublet of nonets. The nonet
splitting was again 0.76 G, suggesting that the four terminal
nitrile groups remained. The doublet splitting, 12.87 G,
suggests that a hydrogen atom has replaced the unique nitrile
group. These results are consistent with the mechanism:
First wave: CN]- . e-.

[NchCN [CN]
NchCN 2r
CN CN CN CN
CN ] .2
Second wave: [NchCN
2- CN]
[NchCN 3-
CN CN CN CN
I -C~
+ H+
[NC}yCNj-
CN CN
Third wave:
[NC}yCNJ- [NC}yCN] 2-
CN CN CN CN
Thus the trianion loses cyanide ion to generate a highly basic
carbanion which abstracts a proton from the solvent. The
resulting 1,1,3,3-tetracyanopropenide anion is then reduced to a
dianion radical at the potential of the third polarographic
wave.
Figure 5.6 (a) ESR spectrum
a1 ,
obtained on reduction of 1 mM
Me4N [(NChCC(CN)C(CN)2J i!:.J •
in DMF at a Hg cathode. (b)
"Stick spectrum" showing the (a)
l !
I
.I .I .1
positions of the hyperfine T T1 I
components. Reprinted with I 1'
permission from P. H.
Rieger, I. Bernal, W. H.
Reinmuth, and G. K. (b)
Fraenkel, J. Am. Chern. Soc. I I I II II II II I I I
1963,85, 683, copyright 1963, Magnetic Field/Gauss
American Chemical Society.
Example 5.3 Connelly and co-workers (15) have recently

reported a cyclic voltammetrylESR study of the oxidation of a
series of alkyne complexes, ArCr(CO)2(RC2R'), Ar = Y)6-C6Me6,
R = R' = Ph, C02Et, C6H40Me, and R = Ph, R' = H. In all cases,
a reversible one-electron oxidation was observed at ca. -0.24 V
us s.c.e. in CH2Cl2 solution. ESR spectra of the oxidized
(a) j
I 20G I
I(
(b) )
If
Magnetic Field 0.2 0.0 -0.2 -0.4
EIV
Figure 5.7 ESR spectra of (a) Figure 5.8 Cyclic voltammogram of
[ArCr(CO)2(PhCCH)]+ and (b) ArCr(CO)2=C=C(SiMe3)2 in CH2CI2
[ArCr(CO)2(Me3SiCCSiMe3)1+ in solution, v = 100 mV s-l. Reprinted
CH2Cl2 solution at room temp- with permiss-ion from N. G.
erature. Reprinted with permiss- Connelly, et a!., J. Chern. Soc.,
ion from N. G. Connelly, et a!., J. Chern. Cornrnun.1992, 1293,
Chern. Soc., Chern. Cornrnun. 1992, copyright 1992 Royal Society of
1293, copyright 1992 Royal Society Chemistry.
of Chemistry.
solutions showed single-line spectra with four 5aCr satellites (l

= 3/2, 9.5% abundance) and, in some cases, laC satellites with
amplitudes consistent with the laC nucleus at one of two
equivalent positions. The only exception was the PhC=CH
cation which, as shown in Figure 5.7a, shows a 4.2 G doublet
due to coupling to the alkyne proton. The Cr(O) complexes were
prepared by the photochemical displacement of CO,
hv
ArCr(CO)a + RC2R' ~ ArCr(COMRC2R') + CO
but attempted synthesis of the R = R' = MeaSi derivative gave
not the expected alkyne complex but the vinylidene complex,
ArCr(CO)2=C=C(SiMea)2. The cyclic voltammogram of the
vinylidine complex, shown in Figure 5.8, exhibits a chemically
irreversible oxidation, followed on the reverse sweep by a
chemically reversible couple at about the potential expected for
the alkyne complex. The ESR spectrum of the oxidized
solution, shown in Figure 5.7b, is virtually identical to the
spectra of the cations of the other alkyne complexes.
Subsequent experiments with chemical oxidizing and
reducing agents showed that conversion of the neutral alkyne
complex to the vinylidene occurs only over the course of several
minutes. The system thus is an example of a "square scheme"
in which the oxidized and reduced forms of a couple have
* -,
different stable conformations:
V+
OC.. Cr", Oc.. Cr",
I .... ~ I"~
C ~C(SiMeg)2 C ~C(SiMeg)2
o o
tslow l fast
V,Sill.,_._..;,..+.;:;..e- V,S'M" +
OC_Cr..... C OC.. Cr....... C
I ......J" I .......11'
C C C C
o I o I
SiMeg SiMeg
In/TaredSpectroscopy
Absorption spectra in the infrared region, 650 - 4000 cm- I , arise

through excitation of vibrational modes. For a diatomic molecule the
vibrational frequency v is related to the bond stretching force constant k
and the reduced mass Jl = mIm2/(ml + m2) by
v=.l·[£
21t'V"it
Vibrational motion is quantized with energy levels,
Ev = (v + 1/2) hv
where h is Planck's constant and v is the vibrational quantum number
(0, 1, 2, ... ). The usual transition responsible for infrared absorption
takes the molecule from the ground state (v = 0) to the first excited state
(v = 1) and thus involves an energy difference hv.
Although the detailed vibrational analysis of a polyatomic molecule
can be very complicated, in many cases, it is possible to think of the
vibrations of particular bonds giving rise to infrared absorption signals.
A vibration corresponding to the stretching of a strong bond between
light atoms generally occurs at a higher frequency than the stretch of a
weak bond, a bond between heavy atoms, or a bending motion. Mixing of
vibrational modes-and the corresponding loss of identification with
specific bonds-tends to be more extensive for modes of similar
frequency. The higher frequency modes tend to be well separated and
are not extensively mixed and therefore usually are assignable to
specific bonds.
The intensity of absorption is proportional to the square of the
"transition dipole moment," which can be thought of as the
superposition of the ground and excited vibrational state dipole
moments. Vibrational modes with no change in dipole moment, e.g., the
stretching modes in homonuclear diatomic molecules, have zero
transition dipole moments and do not give rise to infrared absorption.
The vibration of a bond between two atoms of different electronegativity,
e.g., carbon and oxygen, generally has a large transition dipole moment
and a strong absorption in the infrared.
From this brief introduction, we see that the vibrational modes
which are most useful for identification of the products of electrode
processes are those which arise from the vibration of strong bonds
between dissimilar light atoms. This means in practice that CoO, CoN,
CoX, N-O, and N-X bonds (where X is a halogen) are the most useful
infrared chromophores. The stretching frequencies of these bonds is
often sensitive to changes elsewhere in the molecule so that a
spectroelectrochemistry experiment can follow several species watching
the same chromophore.
back plate
salt
o
D
1'efzel gaskets o 0
Figure 5.9 Expanded view of an optically-transparent thin-layer

electrolysis (OTTLE) cell; from reference (17).
Several designs for optically-tranparent thin-layer electrochemical

(OTTLE) cells have been reported (16), one of which (17) is shown in
Figure 5.9. In this design, based on a conventional infrared cell, the
working electrode is a gold microgrid, mounted between two Tefzel
gaskets and contacted through a piece of platinum mesh. The front
plate of the cell serves as the counter electrode and a fine platinum wire,
mounted in the filling hole, is used as a pseudo reference electrode.
Example 5.4 Zhang, Gosser, Rieger and Sweigart (18)

reported a cyclic voltammetry/IR spectroscopic study of the
oxidation of (mesitylene)W(CO)3 in CH3CN solution. Cyclic
voltammograms such as that shown in Figure 5.10 show that
the complex undergoes a chemically irreversible oxidation to
form a product which is irreversibly reduced. Further work
using a platinum microelectrode showed that the process
remains chemically irreversible at scan rates up to 104 V s-l
and at temperatures down to -45°C. Comparison of the cyclic

gram of (mesitylene)W(CO)3, 1
mM in CH3CN containing 0.1 M
BU4NPFS at 25°C, at a platinum
disk electrode, v = 0.4 V s·l.
Reprinted with permission from
Y. Zhang, D. K. Gosser, P. H.
Rieger, and D. A. Sweigart, J.
Am. Chern. Soc. 1991, 113, 4062,
copyright 1992 American 1.0 0.5 0.0 -0.5
Chemical Society. EIV
(a) (d)
2100 2000 1900 1800 2100 2000 1900 1800

v/cm- 1
Figure 5.11 IR spectra obtained with an OTTLE cell containing 1mM
(mesitylene)W(CO)3, 0.2 M BU4NPFS, in CH3CN at 238 K: (a) initial
spectrum; spectra obtained during (b)-(d) oxidation and (d)-(e) reduction.
The entire sequence required about 10 minutes. Reprinted with permiss-
ion from Y. Zhang, et aI., J. Am. Chern. Soc. 1991,113,4062, copyright 1992
voltammograms with those of the chromium analog, known to

undergo an irreversible two-electron oxidation, sugggested that
the tungsten complex is oxidized by two electrons. The
oxidation process was followed by monitoring the CO
stretching bands in the infrared using an OTTLE cell, as
shown in Figure 5.11. The bands at 1955 and 1870 cm- I , due to
(mesitylene)W(CO)3, decreased in intensity during oxidation
and were replaced by new bands at 2113 and 2044 cm- I . After
passage of 2 Faradays of charge per mole of substrate, only the
latter bands remained. The cell polarity was then reversed and
the product reduced; the original bands grew in as the product
bands decreased, and, after passage of 2 Faradays of reduction
current, only the bands of (mesitylene)W(CO)3 remained. The
complete reversibility shown suggests that neither mesitylene
nor CO ligands are lost in the oxidation. Similar results were
obtained at room temperature, but the product decayed with a
half-life of about 10 minutes to form (MeCN)3W(CO)3. These
results suggested that the oxidation product was the
acetonitrile adduct, [(mesitylene)W(CO)a(NCMe)]2+.
To explain these results, the following mechanism was

proposed:
W-e- ~ W+
kl
W++ S ~ WS+
k-l
WS+ ~ WS2+ E2° = 0.08 V
where W = (mesitylene)W(CO)3 and S = CH3CN. In other
words, one-electron oxidation of the substrate forms a 17-
electron intermediate which rapidly adds a solvent molecule to
form a 19-electron species which is more easily oxidized than
the starting material. On the reverse scan, reduction of WS2+
is followed by rapid ligand loss and spontaneous reduction of
W +. The unique feature of this system is that both the
association and dissociation steps are fast so that the oxidation
and reduction peaks both correspond to two electrons. If K =
k Ilk_1 were very different from unity, one of the peaks should
have shown a degree of chemical reversibility at fast scan
rates. Digital simulation studies of the system showed that a
good fit to the experimental cyclic voltammograms could be
obtained for kl '" k-I '" 10 5 s-l.
Sec. 5.3 Steady-State Voltammetry and Polarography 269
5.3. STEADy-STATE VOLTAMMETRY

AND POLAROGRAPHY
One of the advantages of steady-state voltammetry using a

microelectrode or a rotating-disk electrode is that the diffusion
equations, modified by the rates of coupled homogeneous reactions, can
be solved in closed form. Since we can derive equations for current-
potential curves, steady-state voltammetry provides a useful starting
point for our study of electrode process perturbed by homogeneous
reaction kinetics.
Although the corresponding theory of homogeneous reactions in
polarography is much less straightforward, the results are qualitatively
similar and we will be able to extend the insights gained from the more
exact theory of steady-state voltammetry to polarography.
We will start with the general problem of codiffusion of two species
interconverted by a homogeneous chemical reaction and derive some
expressions which will be useful in discussions of steady-state
voltammetry for the CE, EC, EC', and ECE mechanisms. The methods
are very similar for microelectrodes and rotating-disk electrodes, in the
end differing only in the interpretation of the parameters.
Codiffusion of Two Reacting Species

In general, when a species X diffuses to the electrode without the
complications of a homogeneous reaction, its flux at the electrode
surface can be expressed by
JxCO) = - kD (Cx* - CXS ) (5.1)
where kD is the mass-transport rate constant, Cx* is the bulk solution
concentration, and Cxs is the surface concentration. In this discussion,
we will assume identical diffusion coefficients for all diffusing species.
This assumption is for convenience rather than necessity. Expressions
can be derived with different values of kD for each species, but the
algebra becomes significantly more complex.
In each of the kinetic schemes, two species are coupled by a
chemical step,
kI
Y ~ Z
k.I
The steady-state diffusion equations for Y and Z are then modified to
DV2Cy-klCy+k.lCZ=O (5.2a)
(5.2b)
These equations can be solved by the introduction of new functions, C,
the total concentration of Y and Z, and C', which measures the
departure of the Y-Z reaction from equilibrium,
C = Cy+Cz (5.3a)
C' =KCy-Cz (5.3b)
where K = kI/k.I. Substituting eqs (5.3) into eqs (5.2), we obtain the
differential equations in C and C':
\72C=0 (5.4a)
\7 2 C' = C'/XR 2 (5.4b)
where XR, the reaction layer thickness, is given by
XR - •/ D
- 'y ki + k-I (5.5)
Since
Cy(X) = C(x) + C'(x) (5.6a)
l+K
Cz(X) = KC(x) - C'(x) (5.6b)

l+K
the surface fluxes of Y and Z are
Jy =_D(dCy(x)) =_~[(dC(X)} + (dC'(X)) 1 (5.7a)
dt x=o 1 +K dx x=o dx x=~
JZS =-D (dCz(x)) =_~[K(dC(X)) _(dC'(X)) 1 (5.7b)

dt x=o 1 +K dx x=o dx x=~
To proceed further, we must solve eqs (5.4) using the geometry and
boundary conditions appropriate to rotating-disk electrodes and
micro electrodes.
Rotating-Disk Electrode. We found in §4.7 that the
concentration gradient at a rotating-disk electrode is very nearly linear
through the diffusion layer. Indeed, neglecting the forced convection
term in eq (4.43) and setting \7 2 = d2/dx 2 in eqs (5.4), we find that the total
concentration C is
C(x) = Cs + (C* - Cs) (X/XD), 0 =::; x =::; XD
C(x) = C*, x > XD

where cs =C(O) =Cy(O) + Cz(O), C* =Cx* + Cy* and XD is the diffusion
layer thickness given by eq (4.45). The concentration gradient then is
(dC(X)} =C* - Cs (5.8)
dx %=0 XD
Beyond the diffusion layer, we assume that the solution is homogeneous
with concentrations equal to the initial values and that the Y/Z reaction
is at equilibrium so that C'(x) = 0 for x > XD. The solution to eq (5.4b)
consistent with this boundary conditions is
C'(x) =C tB exp[(xD -xYxR] - exp[-(xD -xYxR] , 0 ~X ~XD
exp(xDIXR) - exp{-xDIXR)
The first derivative, evaluated at the electrode surface, then is
(dC'dx(x») x=O
=_~
XR
(5.9)
where
XR
,
=XR t a nh XR
-XD = XR --'=--'-=--=:...--=--O-----=---=-=7
exp(xDIXR) - exp(-xDIXR)
exp{xDIXR) + exp(-xDIXR}
(5.10)
is an effective reaction layer thickness. Note that XR' ~ XR as XD/XR ~ 00
and XR' ~ XD as Xn/XR ~ o. Using eq (4.45) for XD, we can also write
~~ = 1.61(ijY'6 yk 1 ~ k-l
Substituting eqs (5.8) and (5.9) into eqs (5.7), we have the surface fluxes
J.,=_D(dCY(X») =_~[C*_CS~]
dt x=O 1 +K XD XR'
JZS =_D(dCY(x») =_~[KC* - C B + c.:.::]

dt x=O 1 +K XD XR'
or, introducing the mass-transport rate constant

kD =D/XD (5.11)
and the kinetic parameter
(5.12)
we have
J., =_-.lYL[C* - C S - AC's] (5. 13a)
l+K
JZB =-....!uL[K(C* - C S ) + AC's] (5. 13b)

l+K
Figure 5.12a shows the kinetic parameter A. as a function of the

r.d.e. rotation speed co for (kl + k-l) = 1 - 104 sol.
Microelectrodes. To solve eqs (5.4), we will assume spherical
diffusion so that the del-squared operator is
V2=~ .2.Ji
dr2 + r dr
With the boundary conditions, C(r) ~ C*, C'(r) ~ 0 as r ~ 00, the
concentration functions are found to be
C(r) = C* - ro (C* - C S ) (5. 14a)
r
C'(r) = C's ro exp ro-r (5. 14b)

r XR
where ro is the electrode radius. The surface gradients of these

functions are
(de.) _C* - C
dr r=ro - ro
S
(~)
dr r .. ro
= -C'·V +i'-}R
0
Substituting these in eqs (5.7), we obtain eqs (5.13), the same expressions
for the surface fluxes as found for r.d.e. electrodes, but with different
definitions for the mass-transport rate constant!
kD = Dlro (5.15)
and the kinetic parameter
A. = XR + ro (5.16)
XR
Figure 5.12b shows the kinetic parameter A. as a function of electrode

radius ro for(kl + k-l) = 10 - 107 sol.
Dropping Mercury Electrode. Most of the conclusions we will
draw regarding the effects of coupled homogeneous chemical reactions
on steady-state voltammograms will apply qualitatively to polarography.
Applying the correction for the expanding mercury drop to eq (4.13), we
have an approximate expression for the diffusion layer thickness at the
drop time
1 Equation (5.15) applies to a spherical or hemispherical electrode; for an inlaid disk

electrode, the mass-transport rate constant is kD = 4Dhtro.
The mass-transport rate constant and kinetic parameter then are
kD = 12 '" ,; 3D (5.17)
XD 77ttd
A. = EL (5.18)
XR'
where td is the drop time,

+ k.I)
XD '" " / 7rc td(k i
XRV 3
and XR' is given by eq (5.10). Figw-e 5.12c shows the kinetic parameter A.
as a function of drop time td for (k1 + k-1) = 0.1- 103 s·l.
100
10
100 10
o>/rad s·l
Figure 5.12 The steady· state kinetic parameter A for a range of rate
constants, ki + k.I. as a function of (a) r.d.e. rotation speed ro, (b)
microelectrode radius ro, and (c) d.m.e. drop time td.
Note that the dynamic range for the d.m.e. is rather small, that for
the r.d.e. is larger, and for microdisk electrodes very large indeed. This
is mainly a reflection of the fact that the relevant time scales are
proportional to td, I/o>, and 1/r02 for the d.m.e., r.d.e., and microelectrode,
respectively (see §5.1), and these parameters can be varied over about
factors of 10, 100, and 10 4 . The d.m.e. is sensitive to slower rate processes
than the r.d.e. or a microelectrode, again reflecting the difference in time
scales: up to 10 s for a d.m.e., 1 s for an r.d.e., and 50 ms for a
microelectrode.
We are now prepared to calculate steady-state current-potential

curves for a variety of homogeneous reaction schemes. 1
Preceding Chemical Reaction
When 0, the reactant in an electron-transfer step, is in equilibrium

with an electroinactive species Y, the resulting CE mechanism can be
written as
kl
Y ~ 0
k-I
O+e- ~ R
When E « EO, every 0 arriving at the electrode surface is reduced, but
the current will depend on the rate and equilibrium constant of the
preceding reaction step. If the reaction is very slow, the current i will be
proportional to the equilibrium concentration of 0, but for a very fast
reaction, i will be proportional to the total concentration of Y and O.
Between these limits, the current may afford a measure of the rate
constants. As we saw in §4.9, the half~wave potential can be used to
determine the equilibrium constant if the Y/O equilibrium is very fast,
but we will see that, for slower reactions, E1I2 depends on the rate as well
as the equilibrium constant.
Identifying Z with 0, we can apply the results of the general
codiffusion problem. Since Y is electroinactive, its surface flux is zero,
and eq (5.13a) gives:
C's:::: C* - C S
'A
Substituting this into eq (5.13b) gives
JOs :::: - kD [C* - CS] :::: - kD [C* - Cys - COS] (5.19)
which is consistent with the notion that the flux of 0 is related to the
overall concentration gradient. From the definition of C', we can also
write
C's:::: KCyS - Cos
Equating the two expressions for C's, and solving for Cys, we have
Cys:::: C* - Cos (1- 'A)
1 + AK
Substituting this into eq (5.19) gives
1 For further discussion of kinetic studies using the r.d.e. electrode, see Albery (A6);
essentially identical equations were derived for the microelectrode case by Oldham
(19)
Jos =_kn[AKC* - A.(l + KJ Ccf]

1 + AK
The current is then
i = FAkn [AKC* -)..(1 + KJ CcJ] (5.20)
1 + AK
When the electrode potential is sufficiently negative that COs = 0, we
have the limiting current
iL = FAknC* [---.UL]
1 + AK
(5.21)
The current can also be written in terms of the flux of R, using eq

(5.1). Assuming that CR* = 0, we have
i = FAknCRs (5.22)
Subtracting eq (5.20) from eq (5.21) and dividing by eq (5.22), we have
iL -:- i = [)"(1 + KJ] Co:
l 1 + AX CR
Evaluating the surface concentration ratio using the Nernst equation
and taking logs, we have the Heyrovsky-Ilkovic equation, eq (4.10)
E = El/2 _RJ In iL;i
with
ElJ2 = EO + RT In 1 + AX (5.23)
F ),,(1 +K)
The behavior of the CE reaction scheme can be characterized in

terms so-called kinetic zones corresponding to limiting cases of eqs (5.21)
and (5.23). When A.K » 1, either because the reaction is very fast or
because K is large, the current is limited by the rate of transport and is
proportional to kn. When K is also large, the half-wave potential is
unshifted from EO. In this case Y is really unimportant to the electrode
process; the kinetic zone of pure diffusion control (DP) thus corresponds
to the limit:
DP: K> 10, AK> 10
When K < 10 but A.K > 10, the reaction is fast and equilibrium is
achieved close to the electrode surface. The current is still diffusion
limited, but the half-wave potential then depends on K and is more
negative than EO:
Elf2=EO +RTln~ (5.24)

nF l+K
This modified diffusion-controlled zone (DM) is then characterized by
the limits:
DM: K < 10 < ')J(
In the DM zone, half-wave potential data can be used to determine
formation constants and formulas of complex ions as discussed in
Section 4.9; eq (5.24) is a simplified form of eq (4.56).
When K is small and the reaction is slow so that AK is also small,
the current is controlled by the rate of the A ~ 0 conversion,
iL = nFAkDC*AK
and the half-wave potential is shifted,
Ell2 = EO - BI.ln A.
nF
This is the pure kinetic zone (KP) with limits:
KP: ')J( < 1110, A. < 1110
In the kinetic zone intermediate between pure diffusion and pure
kinetic control, the intermediate kinetic zone (KI), the current and half-
wave potential are influenced both by the rate of transport and by the rate
of the reaction. The kinetic zones are summarized in Figure 5.13.
2
DP
log K Io.,----------~
-2
Figure 5.13 Kinetic zone -4

diagram for the CE o 1 2 3 4
mechanism.
log A.
Example 5.5 Albery and Bell (20) employed r.d.e. voltammetry
to measure the rate of proton transfer from acetic acid to water,
taking advantage of the fact that the free acid is electroinactive
and that the reduction of H+(aq) is quasi-reversible at a
platinum electrode. The electrode process is
kl
HOAc ~ H+(aq) + OAc-(aq)
k-l
H +(aq) + e- -+ ~ H2(g)
In order to partially compensate for reduction of H +(aq) from
the dissociation of water, the experiment used two cells, each
equipped with a rotating platinum disk electrode arranged in a
bridge circuit such that the applied potentials and currents
were identical when the bridge was balanced. One cell
contained 0.005 M HCI in 0.1 M KCI, the other 0.02 M acetic acid
in 0.1 M potassium acetate. The applied potential was adjusted
to bring the cell current onto the limiting plateau. The rotation
speed of the r.d.e. in the strong acid cell, COH, was adjusted to
balance the bridge, holding COHA constant. Since the currents
and electrode areas were identical, eq (5.17) gives
..i.lL = kHCH* = kHA CHA * ~
FA 1 +AX
Using eq (5.11) for the mass-transport rate constants and eq
(4.45) for the diffusion-layer thicknesses, we obtain
kHA = (DHA)2'3(COHA)1'2 = CH*
kH DH coH CHA *
+ (1 l)
AX
Since k1 «k_l[acetate], KYk 1 + k_1[acetate] "" YKk 1 . With eqs
(5.5), (5.10) and (5.12), we obtain
(COHA )1'2 ='~H )2'~ CH )[1 + f05HA(DHA/V)1'6]

COH \D HA \CHA 1.61 YKk 1
Thus a plot ofYCOHA/COH us. -{CORA, such as shown in Figure 5.14,
gives a straight line. The ratio of the slope to the intercept is
slope (DHA/V)1'6
~~~- = --~~==
intercept 1.61 YKk 1
With separately determined values of DHA and v, Albery and
Bell foundKkl =17.0 s-l at 25°C. Thus withK =1.75 x 10-5 ,
kl = 9.7 x 105 sol, k-l = 5.5 x 10 10 M- 1s-l
These rate constants are in good agreement with similar
parameters determined by the temperature-jump method.
2.0
Figure 5.14 Plot of

frequency ratio v s.
frequency used in the
determination of the rate
........
of proton transfer from 1.0
acetic acid to water; from o 5 10 15 25
reference (20). (OHA 1I2
The effect of a preceding chemical step in d.c. polarography is

qualitatively similar to that found for steady-state voltammetry. When
k 1Ktd « I, the system is in the pure kinetic zone and the only time
dependence of the current is through the change in electrode area; thus
the current is expected to increase as td 2/3 . Since the time scale for d.c.
polarography is so long, however, only relatively slow reactions will meet
this requirement.
It was through these so-caned polarographic kinetic currents that
the voltammetric study of coupled chemical reactions began. The first
theoretical work was by Brdicka and coworkers (21) in the 1940's, using
both a reaction layer model (some aspects of which we have used above)
and a more direct approach to solution of the modified diffusion
equations (see §5.4).
RudolfBrdilka (1906-1970) was Heyrovsky's student in Prague and

continued as his assistant until 1939. After World War II, BrdiCka was
appointed Professor of Physical Chemistry at the Charles University and
later headed the Physical Chemistry Institute of the Czechoslovak
Academy of Sciences; he played a leading role in rebuilding Czech
science after the war. BrdiCka was responsible for many of the early
developments in voltammetric theory, including explanations of kinetic
and catalytic polarographic waves and adsorption effects.
Following Chemical Reaction

When R, the product of the electron-transfer step, is consumed by a
chemical reaction, the resulting EC mechanism can be written as
0+ e- +Z R
ki
R +Z Z
k-1
When E « EO, the current is limited by transport of 0 to the electrode
surface and should not be affected by the following reaction. The
limiting current then will be diffusion controlled and independent of the
rate of the chemical step. However, if R is consumed fast enough that its
concentration at the electrode surface is significantly reduced, we expect
a positive shift in the half-wave potential.
Using the general results described above, it is easy to show that the
half-wave potential is given by
E]/2 = EO + R...L In _1..(.:....1_+_10...;... (5.25)
F A+K
where K = k1/k_1 and the kinetic parameter A is given by eq (5.12), (5.16),
and (5.18) for r.d.e, microelectrode, and d.m.e. experiments, respectively.
Since A ~ 1, E1I2 is shifted to positive potentials by an amount which
depends on the homogeneous rate constant.
If the R --+ Z reaction is completely irreversible, 11K = 0 and eq (5.25)
becomes
(5.26)
At 25°C, we expect a shift of +59 mV for each 10-fold increase in A.

For rotating-disk electrodes, Figure 5.12a shows that A is
proportional to Vro for k greater than about 100 s-l so that a plot of E1I2 vs.
In (0 should be linear with a slope of RTI2F. If EO is known the rate
constant can be computed from the shift. Rate constants down to about 1
s-l can also be determined by fitting the curves of Figure 5.12a.
Similarly for microelectrodes (Figure 5.12b), rate constants greater
than about 10 s-l can be determined from shifts in the half-wave
potential. For k greater than ca. 1000 s-l, an independent means of
determining EO is required.
The polarographic diffusion current and wave shape are not
affected by a following reaction. According to (5.26), the half-wave
potential is shifted by (RTIF) In A, where A= Y(71f13) ktd. More detailed
calculations by Kern (22) show that A = V5.36 ktd (a difference of only 4
mV at 25°C from our approximate result).
Catalytic Reactions
When the product of an electrode couple reacts with another
solution species P to regenerate 0, the net effect is catalysis of the
reduction of P by the OIR couple, the so-called Ee' mechanism:
O+ne- ~ R
II
P+R ~ Q+O
If the reduction of P by R is spontaneous, then according to
thermodynamics, P is more easily reducible than O. The kinetic scheme
requires that the reduction of P at the electrode be so much slower than
the homogeneous redox reaction of Rand P that the reduction of P at the
electrode can be neglected. This might seem an unlikely situation, but it
is actually rather common. For example, the electrolytic reduction of
hydrogen peroxide is slow, but the homogeneous reaction of H202 with
Fe 2+ is fast. Thus the electrode process when Fe3 + is reduced in the
presence ofH202 is
Fe3+ + e- ~ Fe2+
!
Fe2+ + H202 + H+ ~ Fe3+ + H20
In effect, the Fe3 +/Fe 2+ couple catalyzes the electrolytic reduction of
H202. These so-called catalytic currents were discovered by Brdil:ka and
Tropp (23) in the 1930's; the theoretical model was developed (24) just
after World War II.
If Co* « Cp*, we can assume pseudo-first order kinetics and
identify 0 and R with Y and Z, respectively, in the general scheme
discussed above. We set k-l =k[P} and kl =K = 0; thus C' ="':'cR. Since
JOs =- JR s , we find that JOs = kDAC'S =-kDACR s and C* =Cs. The
current then is
i =nFAkDA (C* - COS)
iL =nFAkDACO* (5.27)
Combining these expressions, we again recover the Heyrovsky-llkovil:
equation withE1I2 =Eo.
For a slow reaction, A'" 1 and the limiting current is unaffected by
the coupled reaction.
When the reaction is fast enough, say A > 10, then A =rolxR (for a
microelectrode), A =XD/XR (for a r.d.e.), or A= Y(7rr13)ktd (for a d.m.e.),
=
and kDA YDk[P]. The limiting current density then is independent of
electrode size, rotation speed, or drop time (aside from the variation in
drop area with drop time) and gives a measure of the rate constant,
provided the electrode area and the diffusion coefficient are known.
Alternatively, the limiting current can be measured in the presence and
absence of P; the ratio of these limiting currents is A from which the rate
constant can be extracted.
For polarography with A» 1, we expect a current enhancement
factor of '/(71[/3) ktd. An analysis specific to the d.m.e. by Birke and
Marzluff (25) gives a rather smaller factor, V1.34 ktd, though the
dependence on rate constant and drop time are the same. Birke and
Marzluff also give equations to correct for departure from pseudo-first-
order conditions.
For a catalytic process with A » 1, the limiting current in a steady-
state voltammogram or polarogram is increased by a factor of A, i.e., by a
factor proportional to the square root of the pseudo-first order rate
constant k'. Since k' = k [P], a means is provided for the analysis of
electroinactive species such as H202 which react with electrode products
such as Fe 3 +. Numerous polarographic methods have been developed
which take advantage of catalytic processes, and most of these should be
extendable to steady-state voltammetry using r.d.e.'s or microelectrodes.
Example 5.6 Birke and Marzluff (25) measured

polarographic limiting currents for Cr(NH3)63+ in the presence
of chlorite ion in NH4+/NH3 aqueous buffer solution,
CrOll) + e- f=! Cr(Il)
The catalytic reaction,
4 Cr(Il) + C102- + 4 H+ -7 4 Cr(Il1) + CI- + 2 H20
produces four CrOll) for each CI02- consumed. Thus k[CI02-]
= 4k '. The experimental data and the second-order rate
constants derived therefrom are presented below (the drop time
was 4.85 s).
[Cr(III)]/mM [CI02-]/mM iLiiD k/10 4 M-1s-l
0.050 1.00 19.4 1.44

0.075 1.00 20.4 1.59
0.100 1.00 16.7 1.07
0.050 0.50 14.6 1.63
0.050 0.25 10.5 1.69
The EeE Mechanisms

When the product of a chemical step following electron transfer is
electroactive in the potential range of interest, the response of a
voltammetric experiment can be rather interesting. The nature of the
ECE mechanism depend!? on whether the product of the chemical step is
reducible or oxidizable and on whether the half-wave potential of the
second step is greater or less than that of the first step. The
mechanisms can be written (for one-electron steps):
ECE Mechanism: 01 + e- ~ R1 E1°
k
Rl ~ 02
ECE Mechanism: 01 + e- ~ R1
k
R1 ~ R2
02+ e- ~ R2
For an ECE process with E1° < E2° (case IA), 02 is reduced at
potentials where R1 is formed and, if the chemical step is fast enough,
the overall process will appear to involve two electrons. When E1° > E2°
(case IB), 02 is stable at potentials near E1°, and the process is
indistinguishable from an EC mechanism. However, 02 will be reduced
at more negative potentials, so that for E < E2°, case IB is identical to
case IA.
For an ECE process with E1° < E2° (case lIA), R2 is stable at
potentials where R1 is formed and the process is again indistinguishable
from the EC mechani~m. Case IIA can be distinguished from an EC
process by reversal techniques (double potential step
chronoamperometry, cyclic voltammetry, etc.) since the oxidation of R2
would then contribute to the reversal current. When E1° > E2° (case
lIB), 01 is converted to 02 via Rl and R2. The overall process then is
neither an oxidation nor a reduction and, if k is large enough, the
current may be nearly zero.
In all of these cases, the homogeneous electron-transfer
(disproportionation) reaction
01 + R2 ~ 02 + R1
may be important. The effect of this reaction depends on the case and on
the technique involved.
For the ECE mechanism, we identify R1 and 02 with Z and Y,
respectively, in the general scheme and assume that only 01 has a non-
zero bulk concentration so that C* = O. Using eqs (5.1) and (5.13), the
fluxes are
JOlS = - kD (C0 1* - C0 1 S)
JRl s == - kD AC's
J0 2s = kD(CS + AC'S)
JR2s = kDCR2s
Since JRls = -JO l s and JR2s = -J0 2s, and the surface concentrations of 01
and R1 and of 02 and R2 are related by the Nernst equations for the two
couples, we have four equations in the surface concentrations:
- AC's = (COl * - C01S)
Cs + AC's = - CR2S
COlS/CRlS = 81
C02s/CR2s = 82
Solving for the surface concentrations, we obtain
Go~= ~COI*
A + 91
C s- (A-I) C *
R2 - (1 + 82)(A + 81) 01
The current is i =- FA (JOls + J0 2 S) or

i = FAkD [COl * - GOIS + CR2S]
Substituting the surface concentrations, we have
i =FAkDCOI*[~ + A-I ] (5.28)

A + 91 (1 + 8~(A + 9 1)
The limiting current for 81, 92 « 1 is
i L =FAkDG01 *[2 - ~] (5.29)
When the chemical step is fast, A » 1, the process corresponds to a two-

electron wave. When the reaction is slow, A'" 1 and we expect a one-
electron wave.
For the ECE mechanism, we identify Rl and R2 with Z and Y,
respectively, and again assume that only 01 has a nonzero bulk
concentration so that C* = 0, K = O. The fluxes are the same as in the

first case except
JR2S = kD(CS + AC'S)
Jo 2s = kD C 0 2s
The surface concentrations are found to be identical to those computed
above, but the current is i = - FA (J0 1S + J0 2 S) or
i = FAkD [COl * - COl S - C0 2 S]
Substituting the surface concentrations, we have
i=FAh D C01 *[_A_-

1..+81
82(1..-1) J
(1+82)(1..+81)
(5.30)
When the electrode potential is sufficiently negative that 81 and 82 are

both near zero, the current is diffusion limited and corresponds to one
electron, independent of the rate of the chemical step.
We now consider each of the four cases and assess the effect of the
homogeneous disproportionation process.
Case IA. When E 1°< E2° for the ECE mechanism, 82 is very small
whenever 81 is small enough that significant current flows. Thus eq
(5.28) reduces to
i = FAhDC 01 * (21..-1)
1..+81
which corresponds to a wave of shape given by the Heyrovsky-Ilkovic
equation with iL given by eq (5.29) and E 1/2 by eq (5.26). Thus the
apparent number of electrons is (2 - III..) and the wave is shifted to a
more positive potential, as expected from the EC part of the mechanism.
The disproportionation reaction
01 + R2 f:! 02 + R1
has an equilibrium constant
K = exp F(E1°-E2°)
~--=-----=---'-
RT
and, for case lA, K «0. Consider the situation when E "" E1°. The effect
of the R1 ~ 02 reaction is to reduce the R1 concentration at the electrode
surface, resulting in a positive shift of the half-wave potential and an
increase in current (02 is almost entirely reduced to R2 at the electrode).
When disproportionation is allowed, 02 is scavenged by R1 to produce 01
and R2, but since there is very little 02 in solution, the effect on the
current is very small. Voltammograms computed from eq (5.28) are
shown in Figure 5.15a for A = 1, 2, and 20.
Case IE. When Elo > E2° for the BCE mechanism, two waves are
expected from eq (5.28). The first wave involves one electron and is
indistinguishable from that given by an EC mechanism; the second
wave has El/2 = E2° and n2 = 1- 111.. Although the disproportionation
equilibrium constant is very large, the Rl ~ 02 reaction is only a second-
order perturbation of the position of equilibrium, and the inclusion of
disproportionation has a negligible effect on the current. Computed
voltammograms are shown in Figure 5.15b.
Curves showing napp as a function of the rate constant of the
chemical step have been computed by Marcoux, Adams, and Feldberg
(26) for cases IA and IB for various values of the equilibrium constant K.
2 2
(b)
i/iD
1 1
o o
·0.2 ·0.4 ·0.6 0.2 -0.0 -0.2 -0.4 -0.6
1 1
i/iD Cd)
o o
-0.2 ·0.4 -0.6 0.2 -0.0 -0.2 -0.4 -0.6
ElY ElY
Figure 5.15 Computed steady-state voltammograms for the ECE
mechanisms for A = 1, 2, and 20. (a) Case IA; (b) case IB; (c) case IIA;
and (d) case lIB (the solid and dotted lines correspond, respectively, to
infinitely slow and infinitely fast disproportionation equilibria). For
cases IA and IIA, El ° = -0.4 V, E2° = 0.0 V, and for Cases IB and lIB, E10
= 0.0 V, E2° =-0.4 V.
Case llA. When E1° < E2° for the ECE mechanism, 81 » 82, and
the second term of eq (5.30) will be close to 0 whenever the first term is
significantly greater than zero. Thus case IIA is really just an EC
process as far as steady-state voltammetry is concerned. A one-electron
wave is expected, which shifts to more positive potentials as the rate of
the chemical step increases. The effects of disproportionation are
completely negligible. Voltammograms, computed using eq (5.30) are
shown in Figure 5.I5c for A. = 1, 2, and 20.
Case lIB. When E 1 ° > E2° for the ECE mechanism and the
potential is in the vicinity of E1°, we have 81 «82. We then expect a
wave with n1 = I/A. and half-wave potential shifted as expected for an EC
process. A second wave is expected with E1I2 = E2° and n2 = 1 - I/A. (the
total limiting current corresponds to one electron). When E1° > E > E2°,
the effects of disproportionation are very significant. Here 01 is reduced
irreversibly to R1, which reacts to form R2. At the electrode, R2 is
oxidized and the net current is diminished. Disproportionation
increases the efficiency of this process since any R2 which escapes from
the electrode is scavenged by 01, leading to a further decrease in the
current. Steady-state voltammograms, computed using eq (5.30) and by a
digital simulation method (assuming that the disproportionation
reaction is infinitely fast), are shown in Figure 5.I5d.
Figure 5.16 Computed current-

time curves for a preceding 0.01
chemical reaction with Kk1 =
0.01, 0.1, 1, and 10 s-l and for o
an ordinary t- 1/2 current o 1 2 3 4 5
decay. tis
In practice, disproportionation reactions are not instantaneous

(although homogeneous electron-transfer processes are often diffusion
controlled). Thus experimental results for a given value of A would be
expected to lie between the limits suggested by the curves of Figure 5.15d.
Kinetic information, obtained from half-wave potential shifts or limiting
current ratios, thus is at best semiquantitative for Case lIB.
For Cases IB and lIB, the range of rate constants which can be
estimated from limiting current ratios is quite wide. If we assume that
the limit of accurate measurement of wave height ratios corresponds to
0.1 < 111. < 0.9, (10 > A > 1.1), then Figure 5.12 suggests that rate constants
in the range 1 - 1000 s·l are accessible with an r.d.e, 10 - 10 6 s·l for
microelectrodes, and 0.1 - 10 s·l for the d.m.e.
5.4 CHRONOAMPEROMETRY AND

CHRONOPOTENTIOMETRY
Having examined the CE, EC, EC', and ECE mechanisms from the
point of view of steady-state voltammetry, we are now prepared to see
how electron-transfer processes influenced by homogeneous chemical
reactions behave when studied by chronoamperometry-current-time
curves at constant potential-and chronopotentiometry-potential-time
curves at constant current.

Solution of the planar diffusion problem, using the homogeneous
kinetics of the Y ~ 0 equilibrium as a boundary condition (see Appendix
4), assuming equal diffusion coefficients, K « 1,1 and a potential
sufficiently negative that CoCO) = 0, results in the following equation for
the current
i = nFAC* YDKk 1 exp (Kklt) [1- erfYKI?lt] (5.31)
This rather complex expression was first derived by Koutecky and
Brdicka in 1947 (21). As we will see in §6.3, the same current function is
obtained in the case of slow electron-transfer kinetics. Thus, the
electrode response to a slow homogeneous reaction preceding electron-
1 Since we assume the electron transfer process is nernstian and therefore infinitely
fast, the current should be infinite at zero time due to the small amount of 0 initially
present; for small K, this initial transient is of negligible importance after a few
milliseconds.
transfer is closely resembles slow electron transfer though we will see

that the two phenomena are easily distinguished. Current-time curves
for several values of Kk1 are shown in Figure 5.16. For small Kk1, the
current decays relatively slowly and extrapolation to zero time may be
possible,
i(O) = nFAC*YDKk l
allowing determination of the rate parameter Yk tK.
Reversal techniques such as double potential step
chronoamperometry and current reversal chronopotentiometry are
particularly appropriate to the study of chemical reactions which follow
electron transfer. Since these techniques sample the concentration of
the product of the electrode process a short time after it is formed, they
lead to a straightforward measurement of rate.
4
0 (b)
a ................ O&.
0.1 kt
.......... 0.2
------- 0.4 3
t =400 ms
0 t =300 ms
IJ t =200 ms
o 1 2 3 o ro 100 100 200
tlms
Figure 5.17 (a) Theoretical working curves for double potential step
chronoamperometry applied to the EC mechanism. The curves correspond to various
values of tit, where t is the time of the second potential step and t is the current
measurement time after each potential step. (b) Kinetic plot for the rearrangement of
hydrazobenzene followed by double potential step chronoamperometry: 1.0 mM
azobenzene in 1.98 M HCl04 in 50% aqueous ethanol, t = 200, 300, and 400 ms.
Reprinted with permission from W. M. Schwartz and I. Shain, J. Phys. Chern. 1965,
69, 30, copyright 1965 American Chemical Society.
Sec. 5.4 Chronoamperometry and Chronopotentiometry 289
In the case of double potential step chronoamperometry, the

potential is stepped to a point where every 0 molecule is reduced on
arrival at the electrode. At time 1:, the potential is stepped to a point
where every R is oxidized on arrival (see Figure 4.17). If the cathodic
current is measured at time t after the first potential step and the anodic
current is measured at time t after the second step (total time t + 1:), eqs
(4.25) and (4.26) can be combined to give the ratio ia1ic:
i.a = _ ¥..k1:,tl1:) + ~ t (5.32)
lc t + 1:
where <P(k't,th) = 1 when k = o. When a following chemical s~ consumes

R, the function <P(kt,tl't) becomes smaller, approaching ¥t/{t + 1:J as k
increases. Determination of the function requires solution of the linear
diffusion problem under the appropriate boundary conditions, a task
accomplished by Schwartz and Shain (27). The function turns out to be
an infinite series of hyperbolic Bessel functions which must be evaluated
numerically. Schwartz and Shain provide a set of working curves,
reproduced in Figure 5.17a, from which a value of k't can be read given
the anodic-to-cathodic current ratio measured for certain values of tho
The experiment works best for k't '" 1 and for tl1: less than 0.5.
Example 5.7 Schwartz and Shain (27) employed double

potential step chronoamperometry to study the rate of the
benzidine rearrangement of hydrazobenzene. Azobenzene
undergoes a fast ECEC reduction to hydrazobenzene,
Q-N=N-Q :~:/ Q-NH-NH-o

which undergoes acid-catalyzed rearrangement to benzidine
Q-NH-NH-o k H2N-O-ONH2
Hydrazobenzene is oxidized at about the same potential where
azobenzene is reduced, but benzidine is electroinactive in this
potential range. Thus the double potential step method could
be used to monitor the amount of hydrazobenzene remaining at
various times after formation. Anodic-to-cathodic current
ratios were measured for tl't = 0.1, 0.2, 0.3, 0.4, and 0.5 for
various values of t, the switching time. Values of k1: were read
off the working curves, multiplied by th to obtain kt, and kt
plotted us. t as shown in Figure 5.17b for data obtained for 1: =
200, 300 and 400 ms, [azobenzene] = 1 mM in 1.98 M HCI04 in
50% aqueous ethanol solutions. The slope of this plot is the rate
constant k = 23 s-l. Similar plots were obtained for other acid
concentrations with rate constants ranging from 0.6 to 90 s-l.
Current-reversal chronopotentiometry is also applicable to the

study of EC systems. In chronopotentiometry, a constant current is
applied to the electrochemical cell and, when the electroactive material
available near the electrode is no longer able to supply the required
current, the potential swings to that of a new couple (e.g., the solvent or
supporting electrolyte). This transition time 't is proportional to the
square of the concentration of the electroactive species, eq (4.27). In the
current reversal technique, the current is reversed before the cathodic
transition time is reached and an anodic transition is observed which
corresponds to the reoxidation of R. If R is consumed in a chemical step,
the anodic transition time will be correspondingly shorter.
Theoretical analysis of the EC reaction scheme for
chronopotentiometry is straightforward using Laplace transform
techniques, and this case is discussed in some detail by Bard and
Faulkner (B12). It can be shown that the reverse current transition time
't and the time of current reversal t are related by
erf rot+ 't) = 2 erf {fit (5.33)

Solutions of this equation are plotted as 'tIt us. kt in Figure 5.18. This plot
serves as a working curve whereby measured values of 'tIt can be used to
find kt, and knowing t, the rate constant k.
0.3
'tIt
0.2
0.1
Figure 5.18 Working curves for
determination of rate constants
from current-reversal chronopot-
entiometric measurements for
the EC reaction scheme. 1: is the 0.0
transition time after current o 1 2 3 4 5
reversal at time t. kt
Example 5.8 Testa and Reinmuth (28) used

chronopotentiometry to measure the rate of hydrolysis of p-
benzoquinone imine:
~
O~NH + H 20 - k 0 =0= _ 0 + NH3
Since this reaction is relatively rapid, the imine cannot be kept

in solution without rapid hydrolysis. To overcome this
problem, the imine was produced in situ by the electrolytic
oxidation of p-aminophenol:
HO-o-NH2 - O O N H + 2e-+2H+
A constant anodic current of 1.29 rnA was passed through a

cell with a platinum electrode (A = 12.6 cm 2) for times ranging
from 3 to 30 s; the current was then reversed and the amounts
of imine and quinone in solution near the electrode measured
chronopotentiometrically. Some typical experimental curves
are shown in Figure 5.19. Notice that the ratio of the quinone
transition time, 't2, to the imine transition time, 'tl, increases
with increasing time of electrolysis. The ratio 'tIlt measured
from a chronopotentiogram was used with the working curve
of Figure 5.18 to determine a value of kt. Data for various
electrolysis times gave a rate constant k = 0.103 ± 0.003 s-1 for
the hydrolysis reaction in 0.1 M H2S04 at 30°C. The transition
time corresponding to the quinone, 't2, provides an internal
check on the experiment since the sum 'tl + 't2, should be
exactly t13, independent of the rate of the chemical step.
40
tis 't 2
Figure 5.19 Current-reversal
3J !l
T
chronopotentiometric curves for the
oxidation of p-aminophenol (1 mM in
0.1 M H2S04 solution) and the ID
hydrolysis of p-benzoQuinone-imine; t
is the time of anodic electrolysis, 'tl t
1
and 't2 are the transition times 10
corresponding to reduction of the imine
and Quinone. Reprinted with permiss- 0
ion from A. C. Testa and W. H.
Reinmuth, Anal. Chern. 1960,32, 1512,
copyright 1960 American Chemical 0.8 0.6 0.4 0.2 0.0 -0.2
Society. EIV
Catalytic Reactions
For the EC' mechanism
o +ne- ~ R
k
R+P ~ O+Q
the current is enhanced by the feedback of R to 0 and
chronoamperometry provides a rather direct measure of the rate
constant k.
When the potential is stepped to a sufficiently negative value that
CoCO) = 0, the limiting current can be shown to be (29)
iL = nFACO* [YDohtt exp(-k't) + VDok' erfYk't] (5.34)
or
iLliD = exp(-k't) + Yd't erfYk't (5.35)
where k' = kCp and iD is the unperturbed diffusion-limited current, eq
(4.7). For short times, the error function is near zero, the exponential
close to one, and iL "" iD. For long times, k't » 1, the exponential
approaches 0, the error function goes to 1, and eq (5.34) becomes
4
k' = 10
i/iD
...,
W 3
'S 3
;::j
....>.
2
...,....CIl
..... 2
..//~/
..0
....
CIl
~
k' =1
1 1
k' = 0.1
k' =0 /
0 o ~..~.~~~~~~~~~~~
0 1 2 3 0.0 0.5 1.0 1.5 2.0 2.5
tis (kt)1J2
Figure 5.20 Chronoamperometric Figure 5.21 Chronoamperometric

curves for EC' process with k' = 0, response curve for catalytic
0.1, 1 and 10 s-l. electrode processes.
(5.36)
independent of time. The predicted behavior is shown in Figure 5.20 for
several values of k'. The current should decay at very long times since P
is consumed and pseudo-first-order conditions no longer prevail;
however, if Cp* » Co* and semi-infinite planar diffusion applies, the
current may be constant for quite a long time. Thus if the rate constant
is to be extracted, it is best to measure the current in the presence and
absence of P and to compute iLliD. In this way, k' = kCp can be
determined without havi~to know the diffusion coefficient or electrode
area. A plot ofiLiiD us. Yk't, shown in Figure 5.21, serves as a working
curve for such an experiment. Figure 5.21 also serves as a reaction zone
diagram with the pure diffusion (DP) zone defined by k't < 0.05, the pure
kinetic (KP) zone by k't > 1.5, and the intermediate kinetic (Kl) zone by
0.05 < k't < 1.5.
The ECE Mechanisms

For the ECE mechanism,
01 + e- ~ R1
the solution of the diffusion problem is relatively straightforward for a

large planar electrode if the homogeneous disproportionation step is
ignored. Alberts and Shain (30) showed that the current-time response
follows the Cottrell equation when the potential is stepped to E < E1°, E2°:
i = nappFAC*YDlrct
where
nappfn = 2 - exp(-kt) (5.37)
and n is the number of electrons involved in the individual electron-
transfer steps. Feldberg and co-workers (31,32) have examined the effect
of homogeneous disproportionation
01 + R2 ~ 02 + R1
on the current response in chronoamperometry using digital
simulation techniques. They found that eq (5.37) is accurate for kt < 0.2.
In case IA (E1° < E2°, K « 1), napp/n may be somewhat less that the
value predicted by eq (5.37), and in case IB (E 1 ° > E 2°), it may be
somewhat greater. In the later case, napp/n can be somewhat greater
than 2 when kt "" 3.
For the ECE mechanism,

01 + e- ~ R1
with E1° < E2° (case IIA), a potential step to E < E1°, E2° is expected to
involve only one electron regardless of the rate of the chemical step and
so should be independent of the existence of the homogeneous electron-
transfer process. In case IIB (E1° > E2°), on the other hand, the Alberts-
Shain theory predicts
napp = exp(-kt) (5.38)
but Feldberg and Jeftic (32) find that for kt > 1, napp is significantly
smaller than predicted by eq (5.38) when disproportionation is
considered and even becomes negative for kt > 2.5. Early in the
experiment, the current is due mostly to the reduction of 01 to R1. Later
when 01 is polarized at the electrode, the R2 oxidation dominates and the
current becomes negative if the conversion of R1 to R2 is fast enough.
Some computed working curves are shown in Figure 5.22.
napp
0.5
Figure 5.22 chronoamp-

erometric working curve
for Case IIA of the ECE
mechanism: nappln as a
function of log kt for (top 0.0
to bottom) kdispC*lk = 0,
0.1, 1, 10, and 0 0 .
from S. W. Feldberg and
L. Jeftic, J. Phys. Chem.
1972,76, 2439, copyright -0.5 I-....L---L..---I._'--..L---L..---I._L-....I..--l
1972 American Chemical -2 -1 0 1 2
Society. log10(kt)
Example 5.9 Alberts and Shain (30) used chronoampero-

metry to study the reduction of p-nitrosophenol in 20% aqueous
ethanol solutions buffered to pH 4.8. The electrode process
consists of the two-electron reduction of the nitroso group to a
hydroxylamine, dehydration to p-benzoquinoneimine and the
reduction of the imine to p-aminophenol:
HO-o-NO :~:: HO-o-NHOH
HO-Q-NHOH -H2 0 • O=O=NH
O=O=NH
-
~
+2H+
H0-o-/, NH2
\\ //
The corrected datal are plotted in Figure 5.23 as iIFAC*D1I2 us.
t- 1/2 , along with curves, computed using eq (5.37) for k = 0.4, 0.6,
and 0.8 s-l. The data fit the curve for k = 0.6 s-l reasonably well;
Alberts and Shain report k = 0.59 ± 0.07 s-l. The small
discrepancy between theory and experiment is probably due to
neglect of the homogeneous disproportionation reaction.
3
, ,,
,,,
,,
,
,,
"
,
,,
,,
Figure 5.23 Chronoampero- , ,,
metric data for the , ,,
reduction of p-nitroso- , ,,
phenol in 20% aqueous 1 ,,
,,
ethanol solution, pH 4.8.
, ,,
, ,,
from G. S. Alberts and I. ,,
Shain, Anal. Chern. 1963, ,,
35, 1859, copyright 1963 o
American Chemical 0.0 0.5 1.0 1.5 2.0
Society
(t/s)-1I2
1 The experiment employed a hanging mercury drop electrode; the theory for a
spherical electrode is considerably more complex than that for a planar electrode,
although the qualitative effect is similar to that described by eq (5.37). The data
plotted in Figure 5.24 have been corrected to correspond to a planar electrode.
5.5 CYCLIC VOLTAMMETRY
For qualitative studies of homogeneous reactions coupled to an

electron-transfer process, cyclic voltammetry is by far the most
commonly used technique. It has the advantages of steady-state
voltammetry in resolving a feature from each electrode couple as well as
the advantages of the reversal technques of chronoamperometry and
chronopotentiometry so that a single experiment often reveals the
nature of the system. For quantitative measurements the other methods
are often more accurate, but, with care, cyclic voltammetry can provide
useful kinetic data.

Based on the response of steady-state voltammetry and potential-
step chronoamperometry to the CE mechanism, we can predict the
qualitative appearance of a stationary electrode voltammogram or a
cyclic voltammogram. The relevant kinetic parameter is essentially the
ratio of the chemical reaction rate to the scan rate v:
A = lIT..
nF
(k +v k
I o I) (5.39)
For A > 10 and/or K > 10 (the DP zone), we expect a reversible cyclic

voltammogram with current peaks proport;ional to the equilibrium
concentration ofO. In the DM zone (K < 10, KYA > 10), the current peaks
will be proportional to the total concentration of Y and 0, but the peaks
will shift to more negative potentials, as in steady-state voltammetry, eq
(5.19). In the KI or KP zones, we expect an increase in peak current and
a negative shift of the peak potential with increasing A (decreasing scan
rate). This case has been analyzed in detail by Nicholson and Shain (3al,
who showed that the shape of cyclic voltammograms depends on KV A.
Curves of similar shape but different K will be shifted by amounts given
by eq (5.19). Nicholson and Shain provide working curves w1;rich relate
peak current ratios and peak potentials to the parameter KV A. Cyclic
voltammograms for K =0.2 and A = 25, 2.5, and 0.25 are shown in Figure
5.24.

Cyclic voltammograms of a system perturbed by a following
chemical reaction are expected to show smaller anodic current peaks
when the lifetime of R is comparable to the scan time; the cathodic peak
shifts to a more positive potential as the rate of the reaction increases.
1.5
~
rn
.§
~ 1.0
C
CIS 0.5
....,
I-<
:EI-<
~ 0.0
~
Figure 5.24 Cyclic
voltammograms at :S -0.5
constant scan rate for
the CE reaction scheme
for K = 0.2 and A = 0.25 -1.0
(solid line), 2.5 0.2 0.1 -0.0 -0.1
(dashes), and 25 (dots).
(E -E1/2)V
For an irreversible chemical step, the relevant kinetic parameter is

Ie = RTk (5.40)
nFu
Some computer-simulated cyclic voltammograms are shown in Figure
5.25. As expected, the anodic peak disappears for Ie > 0.1. In favorable
cases, the anodic-to-cathodic peak current ratio can be used to estimate
Figure 5.25 Simulated

cyclic voltammograms at
constant scan rate for a
chemical reaction
following a reversible one-
electron process with A =
0.01 (solid curve), 0.1
(dashed curve), 1.0 (dot-
dash curve), and 10 (dotted 0.2 0.1 0.0 -0.1
curve). (E - E1I2)N
0.6
0.4
0.2
-2.0 -1.5 -1.0 -0.5 0.0 0.5 -1 o 1 2
log k't log A.
Figure 5.26 (a) Ratio of anodic to cathodic peak currents for an EC
process as a function of kt. (b) Shift. of the cathodic peak potential as a
function of A for an EC process. Reproduced with permission from R. S.
Nicholson and I. Shain, Anal. Chern. 1964,36, 706, copyright 1964
the rate constant. A working curve, obtained from the work of Nicholson
and Shain (33), is shown in Figure 5.26a (the parameter 't is the time
required to scan the potential from the cathodic peak and the switching
point).
As we might have expected from the results for steady-state
voltammetry, the cathodic peak shifts to more positive potentials for very
fast following chemical steps. If the unperturbed peak potential is
known, the peak potential shift can be used to estimate the rate constant.
A working curve, obtained from the work of Nicholson and Shain (33), is
shown in Figure 5.26b.
When the following reaction is second-order and involves an
electroinactive species, an interesting experimental variation is possible.
Consider the reaction scheme
O+e- ~ R
k
R+X ~ Y
When k[X] is large, an anodic shift in the peak potential is expected.
However, if less than a stoichiometric quantity of X is available, the
reaction proceeds until X is exhausted, after which R remains
unreacted. If k is sufficiently large, a shifted peak will precede the
unperturbed current peak. The separation of this prepeak and the
reversible peak then gives an accurate measure of the rate constant. In
§5.5 Cyclic Yoltammetry 299
order to resolve the prepeak, llEp must be more than about 80 mY. The
working curve of Figure 5.26b then gives A. ~ 250 or k[X]/v ~ 104 y-1. If[X]
'" 1 mM and v = 0.1 Y s-l, we require k ~ 10 6 L mol-Is-I. The method thus
is restricted to very fast rates, but, because the reversible peak provides
an internal standard for the measurement of the kinetic shift, the
results can be quite precise. Although the working curve of Figure 5.26b
gives a qualitative indication of the shift, in practice digital simulation
using conditions close to the actual experiment should be used to refine
the determination.
This effect was first discussed by Jensen and Parker (34) in the
context of conventional cyclic voltammetry, but a considerable
improvement in accuracy is obtained by using first-derivative (or
semiderivative) presentation. Because the prepeak and reversible peak
are better resolved in a first-derivative voltammogram, peak separations
can be measured much more accurately.
Example 5.10 Parker and Tilset (35) have studied the reaction
of 9-phenylanthracene cation radicals (PA+) with 4-
methylpyridine (Nu) using derivative linear sweep
voltammetry. Sample results are shown in Figure 5.27 for
[Nu]I[PA] concentration ratios of 0, V4 and V2.
Although the overall stoichiometry of the reaction is

2 PA+ + 2 Nu ~ PA(Nu)2 2 + + PA
the rate-limiting step which influences the peak separation is
PA+ + Nu ~ PA(Nu)+
Figure 5.27 Derivative linear-scan ~

voltammograms for the oxidation ~
of 1.00 mM 9-phenylanthracene in ;a
the presence of 4-methylpyridine
(a, 0.0 mM; b, 0.25 mM; c, 0.50 mM)
in CHaCN containing 0.10 M
BU4NPFS, v = 0.1 V sol. Reprinted
with permission from V. D. Parker
and M. Tilset, J. Am. Chern. Soc.
1987, J09, 2521, copyright 1987 0.4 0.6 0.8 1.0 1.2
American Chemical Society. EIV
Comparison of the measured peak separation, 159 mY, with
working curves obtained from digital simulation, led to the rate
constant, k = 4.6 x 10 7 L moPs-I. The peak separation-and
thus the rate constant-were found to be virtually independent
of temperature over the range, 0 - 20°C, so that the activation
energy for the reaction is apparently near zero.
Catalytic Reactions
Linear scan voltammograms at a stationary electrode are closely
analogous to potential step chronoamperograms for catalytic processes.
The kinetic zone parameter is again
A=RTk' (5.41)
nFv
where k' is the pseudo-first order rate constant. In the pure kinetic
zone, defined by A > 1, the cyclic voltammogram reduces to a steady-state
wave with limiting current given by eq (5.35), independent of scan rate,
with E p/2 = E 112 equation. The reverse scan in cyclic experiments
virtually retraces the current-potential curve of the forWard scan. When
the homogeneous rate becomes slower (or the scan faster) cathodic and
anodic peaks develop and the half-peak potentials shift toward their
unperturbed values (E1I2 ± 28.5/n mV at 25°C). Some representative
curves are shown in Figure 5.28.
Figure 5.28 Simulated

cyclic voltammograms at o I----~.
constant scan rate for a
EC' process with A. = 0.01
(solid curve), 0.1 (dashed -1
curve), and 1.0 (dotted 0.2 0.1 -0.0 -0.1
curve). (E -E 1I2 )N
The ECE Mechanisms
Since the OI/RI and 02/R2 couples are observed separately, cyclic
voltammetry is a particularly powerful technique for the study of ECE
processes,
ECE Mechanism:
ECE Mechanism:
The relative size of the reduction (or oxidation) peaks thus usually can
give an indication of the magnitude of the rate constant for the chemical
step. Examples of cyclic voltammograms for several values of the rate
(a) (c)
0.2 -0.0 -0.2 -0.4 -0.6 0.2 -0.0 -0.2 -0.4 -0.6
EIV EIV
Figure 5.29 Cyclic voltammograms computed by digital simulation for
case IA of the ECE mechanism. The rate parameter A = 0 (a), 0.025 (b),
0.25 (c), and 2.5 (d); E10 =-0.4 V, E2° = 0.0 V.
parameter A., given by eq (5.40), are shown in Figures 5.29 - 5.32.

In case IA (the ECE mechanism with E1° < E2°, Figure 5.29), the
first negative-going scan shows only the 01 ~ R1 reduction. On the
reverse scan, however, there are several features worth noting. The R1
oxidation peak decreases in size and then disappears, as the rate of the
chemical step increases; an oxidation peak due to R2 grows with
increasing chemical rate. The current between these peaks is
particularly interesting. When the rate is slow (Figure 5.29b), R1
survives long enough to contribute a negative current in the region
between E1° and E2°. When the chemical step is faster, so that no R1
survives when E > E 10, but slow enough that most of the 02 is formed
some distance from the electrode, diffusion of 02 to the electrode gives a
net reduction current in the interpeak region (Figure 5.29c). In this
case, the current on the positive-going scan is actually greater than that
on the negative-going scan and the curves cross. When the chemical
step is so fast that most 02 is formed near the electrode, it is reduced
quickly, leading to a much larger cathodic peak and decreasing the
interpeak current to near zero (Figure 5.29d).
The disproportionation reaction
(c)
0.2 -0.0 -0.2 -0.4 -0.60.2 -0.0 -0.2 -0.4 -0.6

ElY EIV
case IB of the ECE mechanism. The rate parameter A = 0 (a), 0.025 (b),
0.25 (c), and 2.5 (d); E10 = 0.0 V, E2° = -0.4 V. The solid lines were
computed neglecting disproportionation. The dotted lines in (b) - (d) show
the effect of infinitely fast disproportionation.
01 + R2 +Z 02 + Rl
has K« 1 in case lA, so that 02 and Rl tend to scavenge one another.
The principal effect of disproportionation on case IA cyclic
voltammograms is to suppress the curve crossing seen for intermediate
rates (32,36). Amatore, Pinson, Saveant, and Thiebault (37) have pointed
out that disproportionation can have the opposite effect when the
chemical step is exceedingly fast. If K is not too small and Rl is very
short-lived, the disproportionation reaction may be pulled to the right
(uphill energetically) to supply more 02 and Rl (which is quickly
converted to 02). The effect is again to give a cathodic current in the
interpeak region.
In case IB (the ECE mechanism with El° > E2°, Figure 5.30), the
first negative-going scan shows cathodic peaks at both El° and E2°. The
height of the second peak approaches that of the first peak as the
reaction rate increases. The reverse scan shows an R2 oxidation peak
comparable in size to the 02 reduction; if the reaction is fast, little Rl
remains to be oxidized, so that the Rl oxidation peak may be missing.
Disproportionation has little effect on the first negative-going scan, but
the Rl oxidation peak is somewhat smaller on the reverse scan. When
the Rl ~ 02 conversion is not very fast (Figure 5.30b), the
(a)
0.2 -0.0 -0.2 -0.4 -0.6 0.2 -0.0 -0.2 -0.4 -C.6
EIV EIV
case I1A of the ECE mechanism. The rate parameters and standard
potentials are as described for Figure 5.29.
disproportionation reaction enhances the peaks due to the O2IR2 couple.

In case IIA (the ECE mechanism with E1° < E2°, Figure 5.31), the
01 reduction peak shifts toward positive potentials with increasing
chemical rate and the R1 oxidation peak disappears as expected, since
we have essentially an uncomplicated EC process for E < E2°. The R2
oxidation peak and (on the second scan) the 02 reduction peak grow with
increasing chemical rate. As we found for steady-state voltammetry and
chronoamperometry, disproportionation in unimportant for case IIA.
In case lIB (the ECE mechanism with E1° > E2°, Figure 5.32), the
first negative-going scan is significantly perturbed by the chemical step
and by disproportionation. The chemical step converts R1 to R2, but R2 is
oxidized when E > E2°. Thus the current drops more rapidly than
normal after the 01 reduction peak when the rate is fast. This effect is
enhanced by disproportionation as shown in Figure 5.32c and d (32). If
the 01 reduction peak is smaller than normal on the first scan, it will be
much smaller (perhaps not even detectable) on the second and
subsequent scans. The 021R2 reduction/oxidation peaks grow with
increasing rate (enhanced by the disproportionation reaction) and, for a
fast reaction, appear as an unperturbed reversible couple.
0.2 -0.0 -0.2 -0.4 -0.6 0.2 -0.0 -0.2 -0.4 -0.6
EIV EIV
case lIB of the ECE mechanism. The rate parameters and standard
potentials are as described for Figure 5.30. The solid lines were
computed neglecting disproportionation. The dotted lines in (b) - (d) show
the effect of infinitely fast disproportionation.
Example 5.11 Hershberger, Klingler, and Kochi (38) used
cyclic voltammetry to study the oxidation of (Tl 5 -
CHgC5H4)Mn(CO)2L (Mn-L). Cyclic voltammograms of the
derivatives with L = NCCHg and PPhg (traces a and d,
respectively, in Figure 5.33) are uncomplicated and apparently
reversible with peak current ratios of 1.0 and normal peak
potential separations. However, addition of PPhg to a solution
of Mn-NCCHg results in a dramatic change (traces band c of
Figure 5.33). On the first positive-going scan, the oxidation
peak due to the substrate decreases in size and an oxidation
peak due to the product, Mn-PPhg, appears. On the reverse
scan, the product cation is reduced but the cathodic peak
expected for the substrate cation is completely absent.
Comparison of traces b and c of Figure 5.33 with traces bod of
Figure 5.32 suggests that these results can be understood as an
ECE process-the analog of case IIB, making appropriate
allowance for the fact that the first step is an oxidation rather
than a reduction. The mechanism can be written as follows:
Mn-NCCHa =-e-
(Mn-NCCHa"t
~ V;;l PPha
~rCHaCN
Mn-PPha =-e [Mn-PPhal+
0.52 V
As we have seen in Figure 5.32, the shape of a cyclic
voltammogram for such a system is very sensitive to the rate
constant k 1 and somewhat dependent on k 2 as well.
Hershberger, et aI., carried out a systematic investigation
using digital simulation techniques and were able to establish
the rate constants: kl = (1.3 ± 0.2) x 104 M-1s-l, k2 > 104 M-1s-l.
Since the disproportionation step is exoergic and fast (!lGo = -32

kJ mol- 1), it represents the propagation step of a homogeneous
chain reaction initiated by oxidation of a few substrate
molecules. Indeed, Hershberger et ai. found that controlled
current oxidation of Mn-NCCHg in the presence of PPhg
(continued until the potential approached 0.5 V) resulted in
virtually quantitative yields of substitution product with the
passage of less than 0.001 Faraday of charge per mole of
substrate. Thus chain lengths in excess of 1000 are apparently
attained.
(a)

grams of Mn-L in acetonitrile
solution. (a-c) L = CHaCN with (a)
no added PPha, (b) 1 equivalent,
and (c) 9 equivalents of PPha
added (d) L = PPha. Reprinted with
perimission from J. W. I
Hershberger, R. J. Klingler, and J.
K. Kochi, J. Am. Chem. Soc. 1983,
105, 61, copyright 1983 American 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4
In the above example, the rate was found to be first order in

nucleophile, suggesting a displacement or associative mechanism, as
expected for nucleophilic substitution at a 17-electron center. In an ECE
substitution process initiated by reduction of a transition metal complex,
a formal 19-electron center is formed, and the substitution step might be
expected to be dissociative. This expectation was confirmed in work on
an analogous Mn(1) system, as described in the following example.
Example 5.16 Sweigart and coworkers (39) used cyclic

voltammetry to study the reduction of [(l15-CH3C5H4)-
Mn(NO)(CO)2]+ ([Mn-CO]+) in the presence of phosphines and
phosphites. At 25°C, Mn-CO+ shows a chemically irreversible
reduction at ca. -0.2 V (us. Ag/AgCl) as shown in Figure 5.34a.
However the addition of an excess of a phosphine or phosphite
(L), results in suppression of the primary cathodic current and
the appearence of a chemically reversible couple at ca. -0.7 V
(Figure 5.34b), identical to that observed for an authentic
sample of[Mn-L]+. At -65°C, the cyclic voltammogram of[Mn-
CO]+ is chemically reversible whether or not a nucleophile is
present, Figure 5.34d, and at intermediate temperatures, both
the [Mn-CO]+ and [Mn-L]+ couples are observed, Figure 5.34c.
Although the potential of the product couple varied slightly
!\...
I I
(a)
n:.J~
Figure 5.34 Cyclic voltammograms of

1 mM [(T]5_MeC5H4)Mn(NO)(CO)2J
PF6 in CH2CI2 containing 0.1 M
B u 4 N P F 6 and (a) no added
nucleophile, (b) . (d) 10 mM P(OEt)3.
The indicator electrode was a l·mm (d) ~
glassy carbon disk; v = 0.5 V sol.
Reprinted with permission from C. C.
Neto, S. Kim, Q. Meng, D. A.
~65°C
Sweigart, and Y. K. Chung, J. Am. I I
Chern. Soc. 1993, 115, 2077, copyright 0.0 -0.5 -1.0
1993 American Chemical Society. EIV
with nucleophile, at any given temperature, the relative

magnitudes of the cathodic peaks were independent of
nucleophile and of nucleophile concentration, suggesting that
the rate-limiting step in the reaction is loss of CO. When L was
a phosphine or phosphite, the cyclic voltammograms were also
independent of CO pressure, suggesting that CO does not
compete effectively with the stronger Lewis bases. When L was
the weaker nucleophile AsPh3, the voltammograms were
affected by CO pressure. Just as in the previous example,
controlled potential electrolysis required much less than one
Faraday per mole of substrate for complete conversion of [Mn-
CO]+ to [Mn-L]+ and the substitution reaction could be initiated
with a trace of chemical reducing agent. These results are
consistent with the ECE mechanism:
+ + e-
[Mn-CO] :;::::::: Mn-CO
~~CO
~lrL
+ + e-
[Mn-L] :;::::::: Mn-L
-0.7 V
Digital simulation studies provided estimates of the rate

constants, kl = 11 s-1 at -43°C, k2 > 104 L mol- 1K-l. The
activation parameters for the substitution step (MF = 72 ± 8 kJ
mol-I, ~S:j: = 90 ± 15 J mol- 1K-l) are consistent with the
dissociation of a ligand from a 19-electron complex:
Similar electrocatalytic processes, related in general to case lIB of

the ECE mechanism, have been found for a number of other organic (40)
and organometallic (41) systems.
REFERENCES
(Reference numbers preceded by a letter, e.g. (E5), refer to a book listed in

the Bibliography.)
1. D. de Montauzon, R. Poilblanc, P. Lemoine, and M. Gross,

Electrochem. Acta 1978,23,1247.
2. W. E. Geiger in Laboratory Techniques in Electroanalytical
3. N. G. Connelly and W. E. Geiger, Adv. Organomet. Chem. 1984,23, 1.
4. P. Zuman, The Elucidation of Organic Electrode Processes, New
York: Academic Press, 1969.
5. M. D. Hawley in Laboratory Techniques in Electroanalytical
6. R. Alder, J. Chem. Soc., Chem. Commun. 1980, 1184.
7. A. M. Bond, R. Colton, and M. J. McCormick, Inorg. Chem. 1977,16,
155.
8. I. B. Goldberg and T. M. McKinney in Laboratory Techniques in
9. R. G. Compton and A. G. Waller in Spectroelectrochemistry, Theory
and Practice, R. J. Gale, ed, New York: Plenum Press, 1988, p 349.
10. P. H. Rieger in Organometallic Radical Processes, W. T. Trogler,
ed, Amsterdam: Elsevier, 1990, p 270.
11. D. E. G. Austen, P. H. Given, D. J. E. Ingram, and M. E. Peover,
Nature 1958, 182, 1784.
12. P. H. Rieger, I. Bernal, W. H. Reinmuth, and G. K. Fraenkel, J. Am.
Chem. Soc. 1963, 85, 683.
13. A. H. Maki and D. H. Geske, J. Chem. Phys. 1959,30, 1356.
References 309
14. R. D. Allendoerfer, G. A. Martinchek, and S. Bruckenstein, Anal.
Chem., 1973,47,890; R. D. Allendoerfer and J. B. Carroll, J. Mag.
Reson.198O,37,497.
15. N. G. Connelly, A. G. Orpen, A. L. Rieger, P. H. Rieger, C. J. Scott,
and G. M.-Rosair, J. Chem. Soc., Chem. Commun. 1992,1293.
16. H. Tachikawa and L. R. Faulkner in Laboratory Techniques in
17. N. S. Lehnhoff, Ph. D. Thesis, Brown University, 1989.
18. Y. Zhang, D. K. Gosser, P. H. Rieger, and D. A. Sweigart, J. Am.
Chem. Soc. 1991,113,4062.
19. K. B. Oldham, J. Electroanal. Chem. 1991,313, 3.
20. W. J. Albery and R. P. Bell, Proc. Chem. Soc. 1963, 169.
21. R. Brdicka and K. Wiesner, Coll. Czech. Chem. Commun. 1947,12,
138; J. Koutecky and R. Brdicka, ibid. 1947, 12,337.
22. D. M. H. Kern, J. Am. Chem. Soc. 1954, 76, 1011.
23. R. Brdicka and C. Tropp, Biochem. Z. 1937,289,301.
24. R. Brdicka and K. Wiesner, Coll. Czech. Chem. Commun. 1947,12,
39.
25. R. L. Birke and W. F. Marzluff, Jr., J. Electroanal. Chem. 1968, 17, 1.
26. L. S. Marcoux, R. N. Adams, and S. W. Feldberg, J. Phys. Chem.
1969, 73,2611.
27. W. M. Schwartz and 1. Shain, J. Phys. Chem. 1965, 69, 30.
28. A. C. Testa and W. H. Reinmuth, Anal. Chem. 1960,32, 1512.
29. P. Delahay and G. L. Steihl, J. Am. Chem. Soc. 1952, 74, 3500.
30. G. S. Alberts and 1. Shain, Anal. Chem. 1963,35, 1859.
31. M. D. Hawley and S. W. Feldberg, J. Phys. Chem. 1966, 70,3459.
32. S. W. Feldberg and L. Jeftic, J. Phys. Chem. 1972,76,2439.
33. R. S. Nicholson and 1. Shain, Anal. Chem. 1964, 36, 706.
34. B. S. Jensen and V. D. Parker, Electrochim. Acta 1973, 18, 665.
35. V. D. Parker and M. Tilset, J. Am. Chem. Soc. 1987,109,2521.
36. S. W. Feldberg, J. Phys. Chem. 1971,75,2377.
37. C. Amatore, J. Pinson, J. M. Saveant, and A. Thiebault, J.
Electroanal. Chem. 1980, 107, 59.
38. J. W. Hershberger, R. J. Klingler, and J. K. Kochi, J. Am. Chem.
Soc. 1983, 105, 61.
39. C. C. Neto, S. Kim, Q. Meng, D. A. Sweigart, and Y. K. Chung, J. Am.
Chern. Soc. 1993,115,2077.
40. J. M. Saveant,Accts. Chem. Res. 1980, 13,323.
41. G. J. Bezems, P. H. Rieger, and S. J. Visco, J. Chem. Soc., Chem.
Commun. 1981,265.
PROBLEMS
5.1 A CE electrode process studied by potential step

chronoamperometry gave the current density - time curve shown
in Figure 5.35. If the concentration of the precursor species is 1
mM (1 mol m- 3 ), the diffusion coefficient is 10-9 m 2 s- 1 , and the
reduction is by two electrons, estimate k lK for the reaction
preceding electron transfer.
0.55
jtA m- 2
0.50
0.45
0.40
Figure 5.35 Potential-step

chronoamperogram for a CE o 2 4 6 8 10
process. tis
5.2 Potential step chronoamperometry is to be used to measure the

concentration of an electroinactive analyte P through the catalytic
intermediacy of a one-electron couple, OIR.
(a) Because of vibration and convective mixing, the current
cannot be trusted to remain diffusion-controlled for more than 10
s. If the rate constant for the reaction ofR with Pis 2 x 105 M- 1s-l,
what is the minimum P concentration accessible to the analysis?
Assume a diffusion coefficient D =5 x 10- 10 m 2 s- 1.
(b) What limiting current density would be expected at the
limiting P concentration? If the available current measuring
device can read down to 0.1 ~ with the accuracy needed for the
analysis, what is the minimum electrode area needed?
(c) In order to maintain pseudo-first-order conditions, it is
necessary that Co* be at least 10 times smaller than Cpo What
Problems 311
concentration of 0 should be used? What problems might be

expected from this result?
5.3 What requirements must be met in order that eq (5.23) reduces to

the Lingane equation, eq (4.57)?
5.4 Assuming that the reduction of PbOI) in basic aqueous solution

actually proceeds via the mechanism
HP02- + H20 ~ Pb 2+ + 3 OH-
Pb2+ + 2 e- f:! Pb(s)
as assumed in Example 4.10 and Problem 4.23, use the results of
Problem 5.3 to estimate a lower bound for the rate of the
homogeneous step. What conclusions can be drawn from the
result of this calculation?
5.5 (a) Show that the limiting current in a steady-state

voltammmogram is unaITected by the kinetics of an EC process.
(b) Using the methods of §5.3, derive eq (5.25) for the half-wave
potential of an steady-state voltammogram for an EC process
5.6 Show that the steady-state current for an EE process where both
steps are assumed to be nernstian, i.e.,
is given by
i = FACo*kD ( 2 + 62 )
1 + 62 + 6162
where 61 and 62 are the nernstian concentration ratios for the
first and second couples.
5.7 (a) Given the result of Problem 5.6, derive an expression for (iL -
i)/i.
(b) If E2° > E1°, a single wave is expected. Compute the half-wave
potential.
(c) Under what circumstances does the answer to part (a) reduce
to the Heyrovsky-Ilkovic equation for a two-electron wave?
(d) Compute E1I4 - E3/4 for E2° - E1° = 0, 50, 100, and 200 mY.
What value would be expected for a two-electron wave according to
the Tomes criterion for reversibility?
2.0
1.5
$ 1.0
'§ 0.5
Figure 5.36 Cyclic

voltammogram of an EC
i
........ 0.0
,.!:I
.~... -0.5
electrode process with
scan rate v = 400 mV s-l.
The dotted curve shows a
cyclic voltammogram of -1.0
the same system
uncomplicated by the -1.5
following chemical -0.2 -0.4 -0.6 -0.8
reaction.
EIV
5.8 The cyclic voltammogram shown in Figure 5.36 corresponds to an

EC electrode process. Estimate the rate constant k. You need not
correct for iR drop or for capacitive current.
5.9 Digital simulation of cyclic voltammograms for case IA of the
ECE mechanism leads to the working curves for Ep - EO and
iplip(O) shown in Figure 5.37, where ip(O) is the expected cathodic
peak current for k = O. Both electrode processes involve one
electron.
(a) Why does the peak current increase by more than a factor of
two? Give a qualitative explanation. Hint: What is the ratio
CO(O)/CR(O) at the electrode surface at the potential of the peak
current if the homogeneous rate is zero?
(b) Why is there a small negative shift in the peak potential before
the expected positive shift? (See also Figure 5.26b_)
5.10 Cyclic voltammograms for an ECE process are shown in Figure
5_38 for three different scan rates.
(a) Identify each peak in the cyclic voltammograms with the
reduction or oxidation of a species in the mechanism.
(b) Estimate the rate constant of the chemical step. You can
assume that the electron-transfer steps are nemstian and that
capacitive current and ohmic potential drop are negligible.
Problems 313
-40
--_ ... --' -- ,
, , 4
\
-20 2.0
;>
S 0 \
0'
'--'
....1:>..
~
rn
.....
+'
§
2
,
' .
::::.::. \
0 \ >.
~
\
, 1.5 I-<
CIS
20
\
,, I-<
,;
I ..... 0
+' ,
!S,I:>.. ,..c ........ .,-- .. -.............. :
-'..,
\ ,,~
,
\
I-<
,,
\
."
CIS ,
\
\ :;::; ,,
40 \
\
\
\
1.0 -2 '' '
,, ''
'
.
','
60
-2 -1 0 1 2 0.0 -0.5 -1.0 -1.5
log A EIV
Figure 5.37 Theoretical working Figure 5.38 Cyclic voltammo-
curves for cyclic voltammograms grams for an ECE process. The
for case IA of the ECE mechanism: potential scan rate v is 100 mV s-1
peak potential shift (dashed curve) (solid curves), 500 mV s-1 (dashed
and peak current (solid curve) vs. curves), and 2.5 V s-1 (dotted
the kinetic parameter I.. = RTkIFv. curves).
5.11 The concentration of an electrode-inactive oxidant P is to be

measured by the enhancement of the limiting current in a
voltammogram of 0, the reduced form of which reacts with P to
regenerate 0. If Co* =0.0100 mM, Cp* =1.00 mM, D = 10-9 m 2s- 1, k
= 1.0 x 105 M-1s -1, and the reaction stoichiometry is 1:1, what
fractional current enhancement would be expected for:
(a) a polarogram with td = 5 s?
(b) a steady-state voltammogram using a microelectrode with ro
= 11JID?
(c) a steady-state voltammogram using a rotating-disk electrode
with ro = 100 rad s-1?
5.12 Suppose that the product of a one-electron reduction underfoes an

irreversible reaction, R ~ P, with a rate constant k = 1000 s- .
(a) By how much is the half-wave potential shifted in a steady-
state voltammogram using a microelectrode with ro = 1 J.l.m? with
ro =10 IJID? Assume D =10-9 m 2s- 1, T =298 K
(b) Repeat the calculation of part (a) for a rotating-disk electrode
with ro = 10 rad 8- 1 and 1000 rad s-1.
5.13 Microelectrode steady-state voltammograms for an ECE process

gave the results shown in the table.
rolJ..Lm i1/nA i2/nA

0.50 0.030 0.116
1.0 0.034 0.232
2.0 0.037 0.463
5.0 0.039 1.158
10.0 0.039 2.316
where i1 and i2 are the limiting currents following E1I2(1) and

E 112(2), respectively (E 112( 1) > E 112(2)). Assuming that the
concentration of 01 was 1.00 mM and both steps of the ECE process
involve one electron, compute the diffusion coefficient D and the
rate constant for the R1 ~ R2 reaction.
6
ELECTRON-TRANSFER
KINETICS
In Chapters 4 and 5, we discussed voltammetric experiments in

which the current is limited by the mass-transport rate and/or the rates
of homogeneous reactions coupled to electron transfer. In this chapter,
we turn to experiments where the current is influenced by the electron-
transfer rate, beginning with a general discussion of electrode kinetics.
In §6.2, we consider experiments where transport is fast and electron
tranfer is rate-limiting. In §6.3, we consider voltammetric experiments
where the rates of transport and electron transfer are both important.
In §6.4, we turn to alternating-current experiments, which are best
understood in terms of the faradaic impedance.
6.1 KINETICS OF ELECTRON TRANSFER
In order to focus solely on the electron-transfer process, we will

assume that the solution is well stirred and that transport of electro-
active material to the electrode is fast. Initially we will also assume that
the electrolyte concentration is sufficiently high that the diffuse part of
the double layer is very thin and that the potential just outside the
Helmholtz layer is essentially zero, i.e., equal to the potential of the bulk
solution. This is not always a good assumption and we will have to
correct it later. To the extent that our assumptions are correct, the
concentrations of the electroactive species, 0 and R, will be essentially
the same at the outer edge of the Helmholtz layer as in the bulk solution.
Consider an electron-transfer process with a single elementary step
O+e- ~ R
Suppose that the rate of the forward reaction (the cathodic or reduction
rate) is first-order in 0:
cathodic rate =kcCO(O,t)
and that the reverse rate (anodic or oxidation rate) is first-order in R:
anodic rate =kaCR(O,t)
315
316 Electron-Transfer Kinetics
where ke and ka are heterogeneous rate constants with units of m s-l. If

the concentrations have units of mol m- 3 (mM), then the rates have units
of mol m- 2 s- 1 , the same units as flux. Indeed, the difference between the
cathodic and anodic rates is the net flux of 0 at the electrode surface and
thus is proportional to the net current:
(6.1)
We assume that the temperature dependence of the rate constants
ke and ka is given by
~G:j:
ke = Z exp - R T e (6.2a)
-~Ga:j:
ka =Zexp - - - (6.2b)
RT
where the ~Gt's are free energies of activation. The pre-exponential
factors Z can be estimated theoretically (1) but we will be content with
noting that the Z's in eqs (6.2a) and (6.2b) must be equal to satisfy the
principle of microscopic reversibility; otherwise, we will treat Z simply
as an empirical parameter with units of m s-l.
Consider now the special case of equilibrium at the electrode. The
net current must be zero and the surface and bulk concentrations
should be equal. Substituting eqs (6.2) into eq (6.1), taking logs and
rearranging, we have
In Co(O,t) = In ka,o = ~Ge:j: - ~Ga:j: (6.3)
CR(O,t) ke,o RT
where ke,o and ka,o are the rate constants at zero current. The ratio of
the reactant and product concentrations at equilibrium is given by the
Nernst equation,
In Co* = F(Ee - EO) (6.4)
CR* RT
where Ee and EO are the equilibrium and standard half-cell potentials,
respectively. Combining eqs (6.3) and (6.4), we have
~Get - ~Gat = F(Ee - EO)
This expression is a statement of a familiar idea from chemical
kinetics: as shown graphically in Figure 6.1, the difference between the
forward and reverse activation free energies is equal to the standard free
energy change for the reaction. The rates of the cathodic and anodic
processes depend on the electrode potential. By varying the potential, we
can change the free energies of 0 and R at the electrode surface. Thus
the activation free energies must depend on the potential. The details of
this dependence might be rather complicated, but for the moment we
§6.1 Kinetics of Electron Transfer 317
T
T
!J.Gc :t:
!J.Ga :t:
Figure 6.1 Free energy -

reaction coordinate diagram
for an electron-transfer
process. Reaction Coordinate
will take a phenomenological approach and assume a simple linear

relationship:
!J.Gc:t: : !J.GO:t: + aF(Ee - EO) (6.5a)
!J.Ga:t: : !J.Go:t: - ~F(Ee - EO) (6.5b)
where the parameters ex and ~ : 1- ex are, respectively, the cathodic and
anodic transfer coefficients, and !J.Go:t: is the activation free energy for
the process when Ee : EO. We will assume that !J.Go:t: is a constant at
constant temperature, characteristic of the electrode process.
Substitution into eqs (6.2) then gives
k k - aF(Ee - EO)
c,o: oexp RT (6.6a)
k k ~F(Ee - EO)
a,O: 0 exp '------R'---T-- (6.6b)
where
-!J.G:j:
ko: Z exp __0_ (6.7)
RT
When a net current flows through the electrochemical cell, the cell
is not at equilibrium. The deviation of the half-cell potential from the
equilibrium value is called the overpotential or overvoltage, 11.
E: Ee + 11 (6.8)
We assume that the form of eqs (6.5) is retained under nonequilibrium
conditions so that the cathodic activation free energy changes by an
amount aFTl and the anodic activation free energy by an amount -~FTl.
The rate constants then differ from the equilibrium values according to
the relations
(6.9a)
~FTl
ka = ka,o exp RT (6.9b)
Substituting these into eq (6.1) gives
i = FA [ke oCoCO,t) exp -

, RT'
~FTl]
aFTl - ka OCR(O,t) exp
RT
or
l. = lO - a.F'tl
. [CoCO,t) exp - ~FTl]
- - CR(O,t) exp - - (6.10)
Co* RT CR* RT
where io, the exchange current, is equal to the cathodic current (and to
the negative of the anodic current) at equilibrium:
F'AkJ' * - aF(Ee
.
lO .
= le,e = uuO exp RT- EO) (6
.11a )
. . F'AkJ" * ~F(Ee - EO)

lO =-la,e = uuR exp RT (6.11b)
If transport is really fast so that CO(O,t) = CO* and CR(O,t) = CR*, then eq
(6.10) reduces to
. . [exp~-exp
l=lO
- a.F'tl ~FTl
RT
] (6.12)
which is called the Butler-Volmer equation after the two electrochemists

who, in 1924 and 1930, respectively, contributed to its formulation (2) and
experimental test (3). The current-potential curve predicted by eq (6.12)
for a = 0.4 is shown in Figure 6.2.
John A. V. Butler 0899-1977) was a lecturer at the University of

Edinburgh during the 1920's and 1930's. His interests shifted to
biophysical chemistry after World War II.
Max Volmer (1885-1965), a student of Nernst, was a Professor at the
Technical High School in Berlin and a leading figure in German
electrochemistry before World War II.
iii 0
10
o
Figure 6.2 Current-potential
curve predicted by the
Butler-Volmer equation for
a = 0.4. The dashed lines -10
represent the anodic and
cathodic components of the
current. The dotted line
represents the linear -20
(ohmic) region for small 200 100 0 -100 -200
overpoten tials.
11/mV
The Transfer Coefficient
Before going further, let us consider the significance of the transfer

coefficients a and ~ (= 1 - a), which were introduced in order to separate
the potential contributions to t.G c+ and ~Ga+. These parameters have
taken an apparently significant role in eqs (6.9): a and ~ are seen to
reflect the dependence of kc and ka on the overvoltage 11. Thus if a = 0, ~ =
1, the cathodic free energy of activation (and thus the cathodic current)
would be independent of overpotential and all the variation of net current
with potential would be through the anodic component. This situation is
shown schematically in Figure 6.3a. In the opposite extreme, a = 1, ~ = 0,
shown in Figure 6.3b, the anodic activation free energy and anodic
current would be independent of overpotential. Ordinarily, the anodic
and cathodic transfer coefficients are equal, a = ~ = 112, and the anodic
and cathodic currents respond symmetrically to changes in
overpotential as shown in Figure 6.3c. 1
In order to get a physical feeling for the significance of the transfer
coefficients, let us consider an electron-transfer process at the molecular
level, outlining a theory developed by Marcus (1). An electron transfer is
like a charge-transfer transition in electronic spectroscopy. It occurs in
a very short time and the various nuclei can be thought of as fixed in
position during the transition (the Franck-Condon principle). Since 0
and R generally have somewhat different structures, there is a barrier to
1 Since a and ~ govern the symmetry of a current-potential curve, some authors refer
to transfer coefficients as symmetry factors.
Ca) (b)
reaction coordinate reaction coordinate

O+e ~ R O+e ~ R
Cc)
Figure 6.3 Hypothetical free

energy - reaction coordinate - reaction coordinate
potential surfaces for (a) ex = 1 (b)
ex = 0, and (c) ex = 0.5. O+e ~ R
electron transfer associated with changes in bond lengths and bond

angles. This is shown schematically in Figure 6.4, where the free
energies of 0 and R are plotted as functions of a generalized internal
coordinate x which represents a composite of bond lengths and bond
angles which change on going from 0 to R. The energy variation with x
is assumed to be harmonic, so that we have
1
00 = 00* + :2 k(x - xo)2
Figure 6.4 Free energy -

reaction coordinate diagram
for an electron-transfer
process, 0 + e- ~ R. Reaction Coordinate
GR = GR* + ~ k(x - xr)2

where k is a composite force constant and Go* and GR* are the free
energies of 0 and R in the conformations, represented by Xo and x r , just
prior to and just after electron transfer. The transition state
corresponds to the crossing of the GO and GR curves,
G* = Go* + ~ k(x* - xo)2 = GR* + ~ k(x* - xr)2
Rearranging and solving for x*, we have
GO* - GR* + ~ k(xo 2 - Xr2) = k(xo - Xr)x*
1 GO*-GR*
xt = 2(Xo + x r ) + k( _ )
Xo Xr
Substituting this value of x* back into the expression for G*, we have
Gt = 1 k(x o -xrf + 1(Go* _ GR*) + (Go* - GR*f (6.13)

8 2 2k(xo _ x r )2
The lead term is seen to be proportional to the energy required to

reorganize 0 to the conformation of R (or vice versa),
A = ~ k(x o -x r )2 (6.14)
If we take the free energy of 0 in bulk solution as zero, then GO* is
the work required to bring 0 from bulk solution to the electrode and, if
necessary, to affect any gross changes in conformation or coordination
such that the small distortion represented by Figure 6.4 will suffice for
electron transfer. Thus we have
Go* = Wo
Similarly, there will be work Wr required to bring R from bulk solution to
the electrode and affect any gross changes required before electron
transfer. With the zero of free energy defined as above, the bulk solution
free energy of R is F(E - EO). Thus
GR* = Wr + F(E _EO)
Substituting GO*, GR*, and 'A in eq (6.13), we have
~Gct = Gt - Go = Gt
~G:j:=l'A+IF(E-EO)+l(w -w j+[F(E-EO)-(w o -wr)]2 (6. 15a)

c 42 2 0 r 4'A
Thus when E = EO, the activation free energy is

~G:j:=l,,-+l(w _w)+(wo-wr'Y' (6.16)
o 4 2 0 r 4"-
According to eqs (6.5), the transfer coefficients are
a=a(~G}) = l+F(E-EO)-(wo-wr) (6.17a)
a(FE) 2 2'A
~=_ a(~Ga*) = l_F(E-EO)-(wo-Wr) (6.17b)

a(FE) 2 2"-
Thus the transfer coefficients, a. and ~, are expected to be exactly 1/2
when F(E - EO) and (w 0 - W r) are small compared with the
reorganization energy 'A. Notice, however, that the transfer coefficients
are not expected to be constant for large overpotentials. Several of the
general references listed in the Bibliography (A1,A6,B1,B6,B7,B8,B12)
give useful discussions of transfer coefficients and insights derived from
somewhat different perspectives.
Example 6.1 Cyclooctatetraene, CsHs, is a nonaromatic cyclic

polyene with a nonplanar "tub" conformation. The radical
anion, CSHs, however, is planar or nearly so. The rate of
electrochemical reduction in N,N-dimethylformamide solution
was studied by Allendoerfer and Rieger (4) and by Heubert and
Smith (5) using a.c. polarography (see §6.4).
It was found the cathodic transfer coefficient is ex = 0.40. Since

the transfer coefficient is measured for E "" EO in a.c.
polarography, eq (6.17a) suggests that wo »WR. This is
consistent with the notion that the electron is transferred only
after the ring is flattened and that the activation barrier is
mostly due to the conformation change. Indeed, the enthalpy of
activation for the electron-transfer process (32 kJ mol-I,
determined by a.c. polarographic measurements of the
electron-transfer rate as a function of temperature) is
comparable to the estimated enthalpy of activation for bond
isomerization, a process which goes through a planar
transition state.
Double-Layer Effects
We have assumed in the discussion above that the potential drop

between the electrode and solution occurs entirely within the immobile
Helmholtz layer and that a molecule or ion experiences no variation in
potential as it approaches the electrode. This ideal situation is rarely, if
ever, obtained and a correction is usually required.
We showed in §2.1 that the concentrations of 0 and R at the outer
surface of the Helmholtz layer (x = a) are
- zaFc'P
CoCa) = Co* exp RT a
- Z Fc'P
CR(a) = CR* exp ~T a
where Zo and Zr are the charges on 0 and Rand c'Pa is the potential at x =
a (relative to the bulk solution). If <l>a is significantly different from zero,
then we should use CoCa) and CR(a) in eqs (6.11) for the exchange
current rather than the bulk concentrations Co * and C R * .
Furthermore, the potential difference contributing to the activation free
energy needs to be corrected by subtracting c'P a from Ee. Thus eqs (6.11)
become
- zaFc'P - aF\E - EO - c'P )
io = FAkcf;o* exp RT a exp eRT a
.
lO =
F' "k J l
.n. uvR
* exp - zrF<l>a
RT exp
pF(Ee - EO - c'Pa}
RT
The apparent exchange current given by eqs (6.11) then is related to the
true exchange current by
.\
(lO/app= . (a - zo)FcI>a (6 18 )
loexp RT . a
. ) loexp
(lOapp= . - h~ +RT
zr}FcI>a
(6. 18b)
or, since Zo - Zr = 1 and a + ~ = 1,

(. ) . - (~zo + azr}FcI>a
lOapp= loexp RT
Thus the apparent electron-transfer rate constant determined using eqs
(6.11) should be corrected by
k (k) (~zo + azr}FcI>a (6.19)
0= oappexp RT
Equation (6.19) can also be obtained from eqs (6.15) by noting that the
work required to bring 0 and R up to that point in the double layer where
the potential is cI>a is
Wo + Wr = (zo + zr)FcI>a
Thus the correction to the activation free energy is
L\Gcorr:f: =~ (zo + zr)FcI>a
Since Zo - Zr = 1, the difference in the work terms due to the double-
layer effect is
Wo - Wr =FcI>a
so that there may be a contribution to the transfer coefficients. For E '"
EO, eq (6.17a) gives
a =21 - FcI>a/4'A. (6.20)
The double-layer contributions to the exchange current and
transfer coefficients depend on the sign and magnitude of the surface
potential cI>a. cI>a is expected to go through zero at the potential of zero
charge. Thus, for example, the potential of zero charge for mercury is
about -0.43 V (vs. s.c.e.) so that cI>a is positive for an electrode potential E
> -0.43 V. As we saw in §2.5, the potential of zero charge can be
determined, at least approximately, by the maximum in the
electrocapillarity curve (for liquid electrodes) or by the minimum in the
double-layer capacitance vs. potential curve.
The influence of the electric double layer on electron-transfer rates
was discovered and explained theoretically by Frumkin (6) and the effect
is usually referred to with his name. See also a review by Parsons (7).
Aleksandr N. Frumkin (1895-1976) was Director of the Institute of

Electrochemistry of the Soviet Academy of Sciences. Frumkin was a
leading figure in twentieth-century electrochemistry; his contributions to
the understanding of surface effects were particularly important.
6.2 CURRENT·OVERPOTENTIAL CURVES
The Butler-Volmer equation is often used in a strictly empirical

way to analyze electrode kinetic data, summarizing the information in
two parameters, the exchange current io (or exchange current density, Jo
= iotA) and the transfer coefficient a.. There are two ways of extracting
these parameters from experimental data on well-stirred solutions.
Both involve the analysis of current-overpotential curves.
The Tafel Equation

When the overpotential" is sufficiently large, I FrtIRT I » 1, one of
the exponential terms in eq (6.12) will be negligible compared with the
other. For example, when the overpotential is large and negative, the
anodic component of the current is negligible and we have
.. - a.Frt
£ = £oexp RT
1nL' =1n '£0- a.F"

RT (6.21)
or, at 25°C,
log i = log io - 16.90 001
Equation (6.21) is known as the Tafel equation. This logarithmic
current-potential relationship was discovered empirically by Tafel in
1905 (8), some years before the theory of electrode kinetics was developed.
The Tafel equation suggests the means by which the exchange
current and the transfer coefficient may be determined. If, for an
equilibrium mixture of 0 and R, the current is measured as a function
of overpotential and then plotted as log i vs.", a linear region should be
found. Extrapolation of the linear portions of the plot to zero
overpotential yields the log of the exchange current as the intercept; the
slopes should be -16.90 a. and 16.90 13. Such a plot is shown in Figure 6.5.
The Tafel equation and Figure 6.5 suggest that the current
increases exponentially with increasing overpotential. There must be a
point at which the current becomes limited by the rate of transport and
log ius." plots begin to flatten out. The point at which this happens
2.0
1.5
1.0
·2
........
0.5
-
'E0D
0.0
Figure 6.5 Tafel plot: -0.5

loge I i I /io) vs. 11 for a =
0.4. The slopes of the
dashed lines are -1.0
~F/2.3RT = 10.14 y-l 200 100 o -100 -200
and -aF/2.3RT = -6.76
y-l. TJ/mV
depends on the efficiency of stirring and on the diffusion coefficients of

the electroactive species but more critically on the electron-transfer rate.
For slow electron-transfer processes such as the reduction of H+(aq) at a
mercury cathode, the exchange current is so small that the rate of
transport is rarely a problem. For faster processes, however, the Tafel
plot approach may be impractical because, with increasing
overpotential, transport limitation sets in before the linear Tafel region
has been established.
Julius Tafel (1862-1918) was an organic chemist, a student of Emil

Fischer, and a professor at the University of Wiirzburg. Tafel was a
pioneer in the application of electrochemistry to organic synthesis.
The Charge-Transfer Resistance

When the electron-transfer rate is too fast to be measured by the
Tafel approach, all is not lost; the current-overpotential relationship for
small TJ yields the same information. Returning to the Butler-Volmer
equation, consider the limiting case of a small overpotential. If IFTJIRT I
« I, the exponential terms in eq (6.12) can be expanded in a power series
i=iO[I-a;; + ... -1-~;- ...]

Retaining only the first nonvanishing term, we have
§6.2 Current-Overpotential Curves 327
i;;:; - Fio 11 (6.22)

RT
Thus for small departures from equilibrium, the current is
approximately linear in 11 and the electrode behaves as an ohmic
resistance. The quantity
Ret;;:; RTIFio
is called the charge-transfer resistance. The charge-transfer resistance
is best determined by a.c. impedance measurements (see §6.4), but d.c.
measurements can also be used. Determination of the exchange current
from the charge-transfer resistance does not directly yield the transfer
coefficient; however, according to eqs (6.11a) or (6.11b), the exchange
current depends on one of the concentrations, Co or CR, and on the
equilibrium potential. Furthermore, it is easy to show, by differentiation
of eq (6.11a) or (6.11b), using eq (6.4), that the exchange current varies
with the equilibrium potential according to the partial derivatives:
(a In iO) ;;:;.L i3 (6.23a)

aEe CR RT
(a ;;:oL ; :;- IT U (6.23b)
Thus measurement of Ret (and thus io) for a series of solutions with
constant Cot and variable CRt (and thus variable Ee) allows the
evaluation of u.
Example 6.2 Vetter and Manecke (9) studied the reduction of

Mn 3 + to Mn 2 + at a platinum electrode in a well-stirred 7.5 M
sulfuric acid solution. A Tafel plot-log of current density, log j
(j;;:; iIA), us. E-is shown in Figure 6.6 for [Mn 2 +] ;;:; 0.01 and
0.001 M and [Mn3 +] ;;:; 0.01 and 0.001 M. At large overpotentials,
two curves are found corresponding to the different Mn3 +
concentrations but independent of [Mn 2 +]. The curves branch
at small overpotentials, E '" EO ;;:; 1.5 V; the equilibrium
potential depends on the Mn 3 +/Mn 2 + concentration ratio, of
course, so that the point of zero overpotential differs for the
different curves. At large overpotentials, the current levels out
to a transport-limited value. The linear regions of the two
curves are separated by exactly one base 10 log unit, as
expected if the rate of the cathodic process is first-order in
[Mn 3 +]. The slope of the linear regions corresponds to a
cathodic transfer coefficient u ;;:; 0.28.
Vetter and Manecke also determined exchange current
densities by measurement of the charge-transfer resistance
near equilibrium. A plot of their data, log jo vs. E e, is shown in
Figure 6.7. The two lines correspond to constant Mn3 +
(squares) and constant Mn2 + (diamonds) concentrations. The
slopes of the lines lead to independent estimates of the transfer
coefficient using eqs (6.23), obtaining (l = 0.29 and 0.23,
respectively, in satisfactory agreement with the results of the
Tafel plot method.
Multistep Mechanisms
Thus far we have assumed that the electrode process is simply the
addition of an electron to a single molecule of 0 to produce a single
molecule of R. Consider a process with the stoichiometric half-cell
reaction
Vo 0 + n e- fZ Vr R
The concerted addition of two or more electrons in a single step is highly
unlikely (see further discussion of this point in §5.1). In general, we
expect that any electrode process involving two or more electrons
necessarily involves two or more elementary steps. If one of these steps
is rate limiting, the rate laws should have the relatively simple forms:
anodic rate = kaCRnorCOno.
cathodic rate = kcCRn.,.CO ••
where the n's are the orders of the reactions. Equations (6.11) then are
i = nFAkcIJ n.rC nco exp - aF(Ee - EO) (6.24a)

ORO RT
io = nFAkcIJRnorcOoo exp ~F(~;EO) (6.24b)
The net current is given by eq (6.10). Since the equilibrium potential

depends on Co and CR through the Nernst equation
C v.
Ee =EO +RT ln~
nF cRVr
io is really a function of only two independent variables, the two
concentrations or Ee and one of the concentrations. Differentiating eq
(6.24a) or (6.24b), we have
-6 I I I -8.5
(a)
x
-7
~~
~ ~ -9.0
S S
~-8
::::>
XO~o ~
-
b:O ~
0
_ x _ ..s -9.5
t
-9
<>[Mn2+] = 0.01 M
x [Mn2+] = 0.001 M
-10
, , , ,<>
-10.0
0.6 0.8 1.0 1.2 1.4 1.6 1.40 1.45 1.50 1.55 1.60
EIV Ee /V
Figure 6.6 (a) Tafel plot (cathodic branch) for the Mn 3+/Mn 2+ couple in
7.5 M H2S04 solution at 25°C. The upper points correspond to [Mn 3+] =
0.01 M, the lower ~oints to [Mn 3+] =0.001 M. (b) Exchange current density
for the Mn 3 +/Mn + couple in 7.5 M H2S04 solution at 25°C. Reprinted
with permission from K. J. Vetter and G. Manecke, Z. phys. Chern. 1950,
195,270, copyright 1950 R. Oldenbourg Verlag.
(d In iO) = nco(d In Co) _oF

RT
dEe CR dEe CR
or
(d In iO) = nao(d In Co) + PF

dEe CR aEe CR RT
From the Nernst equation, we have
(aIn
aEe
co) CR
= ~
vJlT
Combining the partial derivatives, we obtain
(adE
In iO) = L (nnco _ a) (6.25a)
Ce R
RT VO
(d In iO) = L (nnao + p) (6.25b)

dEe CR RT Vo
Similarly, the partial derivatives for constant Co are:
(aaEe
In iO) =_L(nncr + <X)
Co RT Vr
(6.25c)
or
(aaE
In iO) =_L(nnar _ p) (6.25d)
e c 0
RT Vr
Comparing eqs (6.25a) and (6.25b), we obtain a relation between the

kinetic orders in 0 and the stoichiometric coefficients
nco - nao = vain (6.26a)
Equations (6.25c) and (6.25d) give a similar relation for the kinetic orders
in R:
nar - ncr = vrln (6.26b)
For a simple electron-transfer process with nco = nar = 1, ncr = nao = 0, Vo
= Vr = 1, and n = 1, eqs (6.25) reduce to eqs (6.23). The stoichiometric
coefficients are generally known. Thus if the exchange current is
determined as a function of the equilibrium potential, holding one of the
concentrations constant, the number of independent unknowns can be
further reduced from four to two and, with luck, there may be only one
set of parameters which fit the constraints and make chemical sense.
For further details and examples, see Vetter (B7).
Example 6.3 Vetter (10) studied the kinetics of the electrode

couple
la- + 2 e- ~ 3 1-
using an a.c. bridge technique to measure the faradaic
impedance. An a.c. potential (50-100 Hz, 10 mV peak-to-peak)
was applied to the cell, which contained a small platinum
indicator electrode (A = 0.3 cm2), a large Pt counter electrode (A
= 16 cm2), and a HgIHg2S04 reference electrode. The electrolyte
was 0.5 M H2S04 with variable amounts of iodine and
potassium iodide. The exchange current density was
computed from the measured faradaic impedance (see §6.4).
The equilibrium d.c. potential was a function of the iodide and
triiodide ion concentrations as given by the Nernst equation:
E = EO + RT In [IiJ
2F [I-P
0.16
~
Figure 6.7 Exchange

S
current density as a ~ 0.14
function of equilibrium
potential for the 13-11-
couple in 0.5 M H2S04
solution at 25°C. [1-] =
-g
~
t...
0.12
0.01 M, [13-] variable
(diamonds); [13 -] = f5
0.0086 M, [1-] variable
(squares). Reproduced 0.10
with permission from
K. J. Vetter, Z. phys.
Chern. 1952,199, 285, noo n~ nw nffi nw nM
copyright 1952 R.
Olden bourg Verlag. Eei V
Figure 6.7 shows a plot of CRT/F) In io vs. the equilibrium

potential for two sets of data, one with [1-] = 0.01 M and variable
13- concentration, the other for [I3-] = 0.0086 M and variable [1-].
The slopes of the two lines are +0.13 and +0.78 for the constant
13- and 1- concentrations, respectively. With n = 2, Vo = 1, and Vr
= 3, eqs (6.25) and (6.26) give
2
0.13 = - 3 ncr - a
3
nar - ncr = 2
0.78 = 2 nco - a
1
nco - nao = 2
There are five parameters and only four relations, but if we
insist that the orders be integral or half-integral (or even third-
integral), the only consistent set of parameters is
nar = 1
1
ncr = - 2
a=0.21
Thus the rates of the anodic and cathodic processes are:
anodic rate = ka[I-]
cathodic rate = k c[I3-]1I2[1-]-1/2
A mechanism consistent with these rate laws is
13- ~ 12 + 1-
12 ~ 21
2 (I + e- ~ I-)
where the last step is rate limiting. The small transfer
coefficient is not surprising; eq (6.17a) suggests that a. should
be less than 112 when a lot of work is required to get 0 ready for
electron transfer. The surface potential on platinum is
probably positive in this experiment, so that eq (6.20) would also
suggest a decrease in a. from 112.
Consider now an electrode process which consists of two electron-

transfer steps, one of which is slow and rate limiting, the other of which
is fast and can be assumed to be essentially at equilibrium:
O+e- ~ I Elo
1+ e- ~ R E2°
The standard potential for the overall two-electron process is
E12° = (E10 + E2°)/2
and the Nernst equation for the overall process gives the equilibrium
RIO concentration ratio
CR
-=exp -2F(Ee- E 12°) = ee
12
Co RT
where
If we assume that E 2° > E 1 0, so that the intermediate is never

thermodynamically stable, then two electrons will be transferred for
each occurrence of the rate-limiting step. Suppose that the first step is
rate limiting; the total current is twice the contribution of this step,
i = 2FA(kcCO - kaCr)
with
k k ~F(Ee - E 1°) k e -/3 ~F"

a = 0 exp RT = 0 1 exp RT
where a and ~ (= 1 - a) are the transfer coefficients and ko is the
standard electron-transfer rate constant. Since the second step is
assumed to be in equilibrium at the electrode potential E =Ee + 11, we
have
fR- -Frt
CI - 92 exp RT
Substituting CI, ke, and ka into the expression for the current, we have
i = 2FAkO(COSlll exp - aFrt _ CR exp I3F11 exp F11)

RT SfS2 RT RT
When 11 = 0, the net current is zero and the cathodic and anodic currents
are equal to the exchange current io:
io =2FAkoCOSla (6.27a)
iio =2FAko CR/sfs2 (6.27b)
For 11 *" 0, the net current is given by

. =£0. [exp -RT"
£
I3)F11]
aFrt - exp (1 +RT (6.28)
which has the form of the Butler-Volmer equation, but with aapp = a,
l3app = 1 + 13. We thus have the interesting result: For a two-electron
process where the first step is rate limiting, the apparent cathodic and
anodic transfer coefficients are expected to be approximately 112 and 3/2,
respectively.l Notice that the exchange current in this case is a
complicated function of the electron-transfer rate constant ko, the
transfer coefficient a, and the standard potentials for the two steps, E 1°
and E2°. The individual standard potentials are usually unknown, so
that ko cannot be determined directly.
The situation would be just reversed if the second step were rate
limiting; the apparent cathodic and anodic transfer coefficients would be
3/2 and 1/2, respectively.
Example 6.4 Bockris, Drazic, and Despic (11) studied the rate
of anodic dissolution of iron in 0.5 M FeS04, 0.5 M Na2S04.
Figure 6.8 shows Tafel plots for the process
Fe(s) -+ Fe2+ + 2 e-
at pH 2.0, 3.1, and 4.0. The slopes of the three lines are
approximately equal and give
1 Some authors use the term "transfer coefficient" to refer to these empirical
"apparent transfer coefficients," reserving the term "symmetry factor" for the
transfer coefficient of an elementary process.
~appF =25 ± 3 V-l

2.303KJ'
so that ~a'pJ2 = 1.48 ± 0.18. Furthermore, the displacement of the
lines at dltlerent pH values gives
a log iii ;;;; 1.01 ± 0.02
apR
suggesting that the process is first order in OH- (or -1 order in
H+). The cathodic branch Tafel plot is more difficult to come by
because of the interference of the reduction ofH+. However, the
data available suggested that a.app ", 0.5. These results are thus
consistent with the mechanism
Fe(s) + H20 ~ FeOH(surface) + H+(aq) + e-
FeOH(surface) FeOH+(aq) + e-
~
FeOH+(aq) + H+(aq) ~ Fe2+(aq) + H20

with the second step rate limiting. Since FeOH+ IS In
equilibrium with Fe2+, Co in eq (6.27a) must be replaced by
Co = [FeOH+] =Kh [Fe 2+]I[H+]
where Kh is the equilibrium constant for the hydrolysis of Fe 2+.
Thus the exchange current is
io =FAkoKh81 u [Fe2+]I[H+]
consistent with the observed pH dependence.
Kinetics of Hydrogen Evolution

Probably the most thoroughly studied problem in electrode kinetics
is the reduction of hydrogen ions, H+(aq), to form H2(g). Some
representative results are shown in Table 6.1. The mechanism has been
the subject of some controversy over the years and the details are still
imperfectly understood (B7,B8). Notice the enormous range of the
exchange current density for different electrode materials. H2 evolution
at a platinum electrode hasjo = 10 A m- 2, whereas at a lead electrode,jo =
2 x 10- 9 A m- 2. Furthermore, the apparent cathodic transfer coefficient
varies from less than 0.5 to about 2.0. These features can be understood,
at least qualitatively, in terms of a simple model.
Consider the electrode process
2 H +(aq) + 2 e- ~ H2(g)
Figure 6.8 Tafel plots for

the anodic dissolution of
iron at pH 2.0, 3.1, and
4.0. Reproduced with per-
mission from J. O'M.
Bockris, D. Drazic, and
A. R. Despic, Electro-
chim. Acta 1961, 4, 325, -4
copyright 1961 Pergamon 0.0 0.1 0.2 0.3
Press. Tl/V
Table 6.1 Kinetic Parameters for Hydrogen Evolution
Metal a log(jo/A m- 2 ) <Xapp
Pt (smooth) 1.0 1.5

Pd 0.3 2.0
Ni -1.4 0.5
Au -2.0 1.3
Fe -2.0 0.4
Mo -2.0 1.5
W -2.0 1.5
Cu -2.7 0.5
Cd -3.0 0.3
Sn -4.0 0.4
Hg -7.7 0.5
Ph -8.7 0.5
a Metal in contact with 1.0 M HCI at 20°C. Data from Conway
(H2).
The mechanism is likely to involve reduction of hydrogen ions to give

hydrogen atoms adsorbed on the electrode surface, step (i). The
hydrogen atoms then migrate on the surface and form hydrogen
molecules, step (iia). Alternatively, if the mobility of hydrogen atoms on
the surface is slow, H-H bond formation may be concerted with reduction
of a second hydrogen ion near an adsorbed atom, step (iib). Finally,
hydrogen molecules are desorbed from the surface, step (iii).
M + H+(aq) + e- ~ M-H (i)
2M-H ~ M.H2 (iia)
M-H + H+(aq) + e- ~ M·H2 (iib)
M·H2 ~ M + H2(g) (iii)
If step (i) is rate limiting, the mechanism is analogous to the
reduction of iodine with two parallel and identical rate-limiting electron-
transfer steps. In the absence of double-layer effects, ex '" P '" 112. The
data given in Table 6.1 are consistent (though they are hardly conclusive)
with step (i) being rate limiting for Pb, Hg, Sn, Cd, Cu, Fe, and Ni.
If step (iib) is rate limiting, the mechanism is analogous to the
oxidation of iron and we expect ex '" 3/2 and P'" 1/2. Step (iib) is probably
rate limiting for Pt, Au, Mo, and W.
This leaves exapp = 2 for Pd as an anomaly. The mobility of hydrogen
atoms in bulk Pd is known to be unusually high so it may be that the
reaction goes via step (iia) which is rate limiting. If this is the case, the
expected rate law is somewhat different. If ke is the forward rate
constant for step (iia) and CH is the surface concentration of adsorbed
hydrogen atoms, the net current for a cathodic overpotential should be
i = 2FAkeCH2
For TJ and CH small, the electron-transfer step is nearly at equilibrium,
so that the rates of the cathodic and anodic steps are nearly equal:
or
k [H+] - aF(E - EO) k C PF(E - EO)
o exp RT '" 0 H exp RT
Thus the surface concentration is
C [H+] F(E _EO)
H'" exp RT
and the net current is given by
. 2F'Ak [H+'2 -2F(E-EO)
t = .l'1 C j exp RT
or
.. - 2Frt
l=loexp RT
Thus the apparent cathodic transfer coefficient for this mechanism is
expected to be about 2, consistent with the results for Pd.
A strong M-H bond is expected to make step (i) more favorable but
hinder step (iia) or (iib). Conversely, a weak M-H bond would make step
(i) slow and (iia) or (iib) fast. Thus we might expect a correlation
between exchange current density and M-H bond strength which would
distinguish between the two kinds of mechanism. The exchange
current density for several metals is plotted as a function of the
estimated M-H bond strength in Figure 6.9. The exchange current
density increases with increasing bond strength in the series Tl - Co,
presumably because step (i) of the mechanism above is getting faster, but
is still rate limiting (i.e., slower than steps iia or iib). In the series Pt -
Ta, the exchange current density decreases with increasing bond
strength, presumably because step (iia) or (iib) is rate limiting.
0
C'I
S
Figure 6.9 Exchange
<
"0 -2
-
current density for H2 :.:>
b.O
evolution from aqueous 0
solutions for various
electrode materials as a -4
function of estimated M-
H bond strength. Repro-
duced with permission
from S. Trasatti, J. Elect-
roanal. Chern. 1972,39, 100 150 200 250 300 350
163, copyright 1972 Else-
vier Science Publishers. M-H bond enthalpy/kJ mol- 1
6.3 ELECTRON-TRANSFER RATES FROM VOLTAMMETRY
In §6.1, we considered the case where current through a cell is

limited only by the kinetics of the electron-transfer process and assumed
that transport to the electrode surface is fast. In §4.1, we considered the
opposite case, where electron transfer is fast and the current is diffusion
limited. The real world, of course, often contains intermediate cases and
we now consider the situation where the rates of mass transport and
electron transfer are of comparable magnitude.
The mathematical description of current-potential, current-time, or
potential-time curves can become quite involved when the rates of
transport and electron transfer are both important. For a rotating-disk
electrode or a microelectrode, however, the current can be measured
under steady-state conditions and the mathematics is relatively simple.
We will treat the cases of steady-state quasi-reversible and irreversible
voltammetric waves in some detail and, with this background, present
the results of more complex calculations for potential step
chronoamperometry, d.c. polarography, and cyclic voltammetry.
Steady-State Voltammetry
In §5.1, we found that the time scale for a rotating-disk electrode or

a microelectrode operating under steady-state conditions is governed by
the residence time of a diffusing species at the electrode surface. That
concept is particularly illuminating when we consider current
limitation by the rate of electron transfer. When a molecule spends very
little time near the electrode surface, the electron-transfer rate must be
fairly fast in order for current to flow. When the rate is slow, the
electroactive species may have come and gone without having lost or
gained an electron. For an r.d.e., the residence time decreases with
increasing rotation speed-the solution is being pumped by the electrode
faster. For a microelectrode, residence time decreases with decreasing
electrode radius-the electrode is a small target from which it is easy to
escape. Thus we expect the effects of slow electron transfer to be
magnified as we increase the rotation speed of an r.d.e. or decrease the
size of microelectrode used in a steady-state voltammogram.
Conversely, as we increase the residence time by slowing the r.d.e.
rotation speed or changing to a bigger electrode, the voltammetric wave
will appear to be more nearly reversible.
As we have seen in §6.1, electron-transfer rate constants are
functions of potential, so that when the electrode is made more negative,
the electron-transfer rate increases (for a reduction process) and
eventually becomes faster than the rate of transport. Thus at sufficiently
§6.3 Electron-Transfer Rates from Voltammetry 339
negative potentials, a diffusion-limited current is usually observed, even

for electron transfers which are intrinsically slow.
Since the rate process occurs entirely at the electrode surface, the
rate problem to be solved is considerably easier than that of coupled
homogeneous reactions dealt with in §5.3. We can focus on surface and
bulk concentrations which are related by the mass-transport rates and
connect the surface concentrations by the electron-transfer rates.
One-Electron Processes. Consider the one-electron reduction of
o to R, represented by
kDO ke
0* ~ Os ~
hDO ka
where kDO and kDR are the mass transport rate constants and ke and ka
are the electron-transfer rate constants given by eqs (6.9). The steady-
state rate equations for the surface concentrations, COs and CRs, are
d~tOS = kDOCO* + kaCR"-(kDO + k e) Cos = 0
d~:" =kDRCR* + keCos -(kDR + kal CRs = 0

Assuming that the bulk concentration of R is zero, we solve these
equations for COs and CR s :
Cos = Co* (1 + kalkDRl (6.29a)
1 + kjkDR + kJkDO
CR" = Co* (kJkDO) (6.29b)

1 + kalkDR + kelkDO
The net current is
i = ie + ia =FA[kcCOS - kaCRS]
Substituting for the surface concentrations, we have (for kDO = kDR = kD)
i = FA ( kDCO* ) (6.30)
1 + kalke + kDlke
When the electrode potential is sufficiently negative, the cathodic rate
constant ke will be large compared with ha or kD and the current will be
diffusion limited:
iD =FAkDCO*
Dividing iD by i and rearranging, we have
iD -i _ kD + ka
-- - ke
Substituting for ke and ka using eqs (6.6) and (6.9), we obtain

in-i
-- =
'YL exp oF (E _Eo) + exp F-'-----'-
(E _Eo)
i ko RT RT (6.31)
When kn1ko is small (fast electron transfer), the first term on the right-
hand side of eq (6.31) can be neglected and we get the Heyrovsky-Ilkovic
equation, eq (4.11), with EV2 = EO (since we assumed kno = knR). On the
other hand, when electron transfer is slow, so that kD » ko, the first
term dominates and we have a completely irreversible wave. Taking
logs and rearranging, we get
E = El/2 + RT In in -:- i (6.32)
a.F l
where the half-wave potential is given by

ElJ2 = EO + RT In ko (6.33)
a.F kn
In intermediate cases, 0.1 < kolkn < 10, the electrode process is
referred to as quasi-reversible and the wave shape must be calculated
using the full equation, eq (6.31). However, it is in this region that the
kinetic parameters are accessible from voltammetric measurements.
Figure 6.10 shows the transition of the current-potential curve from
reversible to quasi-reversible to irreversible as the electron-transfer rate
decreases. Figure 6.11 shows the shift in half-wave potential, EO - E 112,
and the width of the wave, E1/4 - E3/4, as functions of log kolkD and a..
1.0
ilin
0.8
0.6
0.4
Figure 6.10 Plot of relative 0.2

current in a steady· state
voltammogram us. E-
E 112(rev) for a reversible 0.0
wave (dashed curve) and 0.1 -0.0 -0.1 -0.2 -0.3 -0.4 -0.5
for knlko = 0.1, 1, 10, 100,
and 1000, a = 0.5. (E -E1/2)N
0.12
~ 0.08
~
~
Figure 6.11 Kinetic I
diagram showing the o
shift in E 112 as a ~ 0.04
function of the width
of the wave; solid
curves correspond to
constant IX and
variable log kolkn; -0.00
dashed curves to
constant log kolkn 0.05 0.10 0.15 0.20
and variable IX. (E 114 -E3/4 )V
Thus if the reversible half-wave potential is known and El/2 and El/4-
E3/4 are measured, the corresponding point on the diagram determines
the kinetic parameters.
For r.d.e. voltammograms, kn = Dlxn, where xn is a function of
rotation frequency and is given by eq (4.45); thus
kn = 0.62 D2/3y-1I6CJ)1I2
For D = 10-9 m 2 s- 1 , y = 106 m 2 s- 1 , and CJ) = 10 - 1000 rad s-l, kn ranges from
2.0 x 10- 5 to 2.0 x 10-4 m s-l. Thus the experimentally accessible quasi-
reversible region has 2 x 10-6 < ko < 2 x 10-3 m s-1.
For steady-state voltammograms using a microdisk electrode, kn is
given by eq (4.51)
kn = 4Dlnro
where ro is the radius of the microdisk. For D = 10-9 m 2 s- 1 and r ranging
from 0.25 11m to 25 Ilffi, we have 5 x 10-3 m s-l < kD < 5 x 10-5 m s-l. The
experimentally accessible quasi-reversible region then has 5 x 10-6 < ko <
5 x 10-2 m s-l. Note that the range extends to larger ko than with an r.d.e.
Furthermore, since experiments are much less troubled by ohmic
potential drop, microelectrode voltammetry is in principle the preferred
technique for measurement of electron-transfer kinetics. In practice,
however, electrode surface imperfections-poorly defined effective
electrode radius-lead to a potentially large systematic error.
In an experiment, a series of voltammograms is recorded using a
range of microelectrodes of varying sizes. If ko is near the upper end of
the range, the voltammogram will be essentially reversible for larger

electrodes, and EO (E 1/2rev) can be determined. The shift in half-wave
potential and increased width for smaller electrodes then can be used
with Figure 6.11 to determine the kinetic parameters. Oldham, et a1.
(13), discuss other methods for extraction of EO and the kinetic
parameters from microelectrode voltammograms and provide a more
extensive set of correlation diagrams analogous to Figure 6.11.
Example 6.5 Sweigart and coworkers (14) have used

microelectrode steady-state voltammetry to measure the
electron-transfer kinetic parameters for the reduction of Mn(p-
ClTPP)(DMSO)2+ (p-ClTPP is the dianion of tetra-p-
chlorophenylporphyrin) in DMSO solution.
Voltammograms for electrodes with diameters of 1, 2, 5, and 10

11m are shown in Figure 6.12. Note the shift in E 112 and
increase is E1I4 - E3/4 with decreasing electrode radius. EO
was determined from the average of E pc and Epa in a
conventional cyclic voltammogram. Analysis of these data
using the equivalent of Figure 6.11 gave ko = 6 x 10- 5 m s-l, a =
0.60 at 25°C.
Two-Electron Processes. Consider now the two-electron

reduction of 0 to R where we assume that, at least in principle, the
process occurs in two one-electron steps (an EE process):
0+ e- ~ I
1+ e- ~ R
Following the same procedure as above, we can compute the steady-state
surface concentrations of 0, I, and R as functions of the electron-
transfer rate constants kle, kIa, k2e, and k2a, and the mass transport rate
constant kD (assumed equal for all three species). Substitution into the
expression for the net current
i = FA[kleCOS + (k2e - kla)CI S - k2aCRS]
gives
(6.34)
where
Figure 6.12 Steady-state voltammo-

grams of Mn(p-CITPP)(DMSO)2, 2
mM in DMSO with 0.1 M BU4NC104
at 25°C. Indicator electrodes were 1,
2, 5, and 10 11m diameter platinum
microdisks, u = 20 mV s-l, potential
us. Ag/AgCI; The thin dashed lines
locate E 112. Reproduced with per-
mission from Y. Zhang, C. D. Baer,
C. Camaioni-Neto, P. O'Brien, and
D. A. Sweigart, Inorg. Chern. 1991,30,
1682, copyright 1991 American -0.4 -0.6
Equation (6.34) is a bit complicated; let us consider some limiting cases.

Case I. Assume that the first step is thermodynamically easier
than the second, E2° < E1°. When E "" E1°, we should have (92 + kDlk2e)
» 1, so that eq (6.34) simplifies to
i =
FAkDCO*
1 + 91 + kDlklc
which is identical to eq (6.30) and represents a one-electron wave which
may be reversible, quasi-reversible, or irreversible depending on the
magnitude of kDlk 1e.
When E = E2° < E1°, 81 will be very small and, if the first electron-
transfer step is fast because of the large overpotential, kDlk 1e will also be
small. Equation (6.34) then becomes
i = F Ak DC 0 * (1 + 1 + 82 +1kdk'2c )
which represents a one-electron wave in the vicinity of E2° which grows
out of the one-electron limiting current of the first wave. Thus in Case I,
we expect two essentially independent one-electron waves.
Case II. When E2° > E1°, the intermediate I is thermodynamically
unstable and, if the electron-transfer rates are fast enough, we expect to
see a two-electron wave. Since E2° > E1°, we have 82 « 81 whenever the
current is nonnegligible. If kD is small compared with k 1e and k2e (the
reversible case), eq (6.34) reduces to
i = 2FAknCo*
1 + 9192
or, with iD = 2FAkDCO*, we have
in-i
- . - = 99
12
l
which, on taking logs, is the Heyrovsky-Ilkovic equation with n = 2 and
EV2 = (E10 + E2°)/2.
Either of the two electron-transfer steps could be rate limiting.
Consider first the case where step one is slow and step two fast. For 92
« 1 and kdk2c « 1, eq (6.34) becomes
i = __ 2F_:A_"=n_C-",o,-*_
1 + 9192 + knlk1c
which is very similar to eq (6.30) except that the limiting current
corresponds to two electrons. There is an important difference, however.
In the reversible case (kD1klc « 1), the Tomes criterion gives
E V4 - E3/4 = 56.51n m V
or 28.2 mV for a two-electron wave at 25°C. But when the process is
completely irreversible, we have
E V4 - E3/4 = 56.5/a m V
or about 113 m V if a = 1/2. Thus the increase in width with decreasing
electron-transfer rate constant (or increasing mass transport rate) is
rather more dramatic than in the one-electron case. Computed waves
for E2° - E10 = 0.2 V and a = 112 are shown in Figure 6.13a for several
values of kDlko.
When step 2 is rate limiting, the wave shape is sensitive to the
difference in potentials, E2° - El ° and the electron-transfer rate. If the
separation is large compared with the kinetic shift, a single
symmetrical two-electron wave is expected with a width of
EV4 - E3/4 = 56.5/(1 + a) mV
This is analogous to the two-electron process discussed in §6.2, where we
found an apparent transfer coefficient of 3/2.
When the kinetic shift is larger than the separation of standard
potentials, a rather unsymmetrical wave results. If the second electron-
transfer step is intrinsically very slow, the current from the first
electron transfer may be nearly limiting before the overpotential is big
enough to provide a significant rate for the second step. The transition
from a reversible two-electron wave to this unsymmetrical situation is
shown in Figure 6.13b.
2.0
2ilin
1.5
1.0
0.5
0.0
0.0 -0.2 -0.4 0.0 -0.2 -0.4
(~ - ~1/2(rev)1V (~ - ~1/2(rev)1V
Figure 6.13 Plot of relative current in a steady-state voltammogram us. E

- E1I2(rev) for a two-electron wave with a = 0.5, El = -0.1 V, E2 = +0.1 V
for (a) step one rate limiting with kDlko = 0 (reversible, dashed line), 0.1,
1, 10, 100, and 1000, and (b) step two rate limiting with kD1kO = 0
(reversible, dashed line), 1, 10, 100, 1000, and 10000.
Criteria for Reversibility

Two methods are commonly used to test for reversibility. The first,
based on eq (4.11), uses a plot of ~ us. In[(in - i)li], which should be linear
with a slope of RTlnF if the process is reversible. Such a plot will not be
linear if the process is quasi-reversible. An essentially equivalent check
of reversibility is the Tomes criterion (see Example 4.2): for a reversible
process the difference in potentials aL which i = inl4 and i = 3inl4 is
~1I4 -~3/4 = (RTlnF) In 9 = 56.51n mV
at 25°C. These methods can lead to misinterpretation, however. If the
process is completely irreversible, a plot of ~ us. In[(in - i)li] also gives a
straight line but with slope RTIFa. app . An irreversible two-electron
process where the first step is rate limiting with a. = 0.5 is thus
indistinguishable from a irreversible one-electron process. This
ambiguity can be avoided, of course, if n is known from the measured
wave height or from a coulometric experiment (see §7.2).
A better method to check reversibility is to measure current-
potential curves for different r.d.e. rotation speeds or microelectrode
sizes (or different d.m.e. drop times-see below). Since the mass
transport rate constant varies with rotation speed or electrode size (or
drop time), the ratio kD1ko is affected. If the process is reversible, the
shape of the curve is unaffected and the apparent half-wave potential is
constant, but variations in El/2 will be found for quasi-reversible and
irreversible waves.
Potential Step Chronoamperometry

In order to obtain the current-time response in a potential step
experiment using a stationary planar electrode, the diffusion equations
must be solved with the boundary condition
(6.35)
where ke and ka are functions of the potential and are given by eqs (6.9).
The mathematics is fairly straightforward (using Laplace transforms,
see Appendix 4) and gives the surface concentrations
Co(O,t) = Co* - Co* (l_f().. Vt)) (6.36a)

1+ e AW
CR(O,t) = Co* (l_{(AVt)) (6.36b)

1+e AW
where we have assumed equal diffusion coefficients and CR* = o. As
usual, the parameter e is
e = ka = exp F(E _EO)
ke RT
and we have introduced the function
((AVt) = AW exp(A 2t)[1- erfl.'AVt)] (6.37)
where
(6.38)
The current is obtained by substituting the surface concentrations, eqs
(6.36), into eq (6.35), obtaining
i = FAkeCo* (((AVt)) (6.39)

AW
The functionJ(A Vt), shown in Figure 6.14, varies from 0 to 1 with
increasigg AVt. In the limit of fast electron transfer and long times, AVt
»l,{(AVt)/AW ~ 0 and eqs (6.36) reduce to eqs (4.3), which were derived
assuming a nernstian electrode process. The significance of e9- (6.39) is
clarified by considering two limiting cases. At short times, {(A Vt) "" AW,
and at t = 0, eq (6.39) becomes
1.0
F(A-.Jt)
0.8 ----
...,
Ul
3
·S;::l
0.6 >.
.......,e
I-<
2
0.4
1::til
'-'
1
0.2
....................... --_ ............. -- --- --- _.. ..
0.0
o 1 2 3 o 1 2 3 4 5
A-.Jt tIs
Figure 6.14 The function f('J..:Vt) Figure 6.15 Current-time curves
VS. 'J..Vt for potential step chronoamper-
ometry with a = 0.5, 11 = -0.1 V,
DO = 10. 9 m 2 s· 1 , and ko = 00
(solid line) and log (k o/m s-1) =

-5.0, -5.5, and -6.0.
i(O) = FAkeCO*
so that the current is initially limited by the rate of electron transfer.
When we make the overpotential sufficiently negative that ke »ka, A'"
kelW and eq (6.39) becomes
i = FACO*VDlrtt /tAVi) (6.40)
For a sufficiently negative overpotential or at long times, AVi » 1, /tA vi)
'" 1, and eq (6.40) reduces to the Cottrell equation, eq (4.7); the current is
then diffusion controlled.
Notice that eq (6.40) has the same form as eq (5.31) for the current
limited by a homogeneous chemical step preceding electron transfer. As
we remarked in §5.4, an electron transfer process preceded by a slow
reaction behaves very much as if the electron transfer step itself were
slow. However, we now see that the two phenomena are easily
distinguished: the current for a CE process is independent of
overpotential for EO - E greater than about 100 mV (where 0 is reduced
immediately on reaching the electrode), but when the current is limited
by slow electron-transfer kinetics, ke continues to increas with
increasing overpotential.
In an experiment, the potential is stepped to some negative value
and the current measured as a function of time. With increasing time,
AVi increases, and f(A Vi) -+ 1. Thus for long enough time, current decay
curves are expected to be independent of the electron-transfer rate

constant; of course, the current may be very small indeed when that
happens. Figure 6.15 shows the kinds of results expected for systems
differing only in electron-transfer rates. Notice that for very slow
electron-transfer rates, it is possible to extrapolate the current to zero
time and thus to determine the electron-transfer rate constant ke
directly if the electrode area and the concentration of 0 are known. Note,
however, that capacitive charging current may interfere at short times.
Irreversible Polarographic Waves
In §4.4, we corrected for the expanding mercury drop area by using

an effective diffusion coefficient, 7D/3, in the equations derived for a
planar electrode. That approach is less accurate for an irreversible
wave (although the right qualitative result is obtained). Meites and
Israel (15) have shown that the shape of an irreversible polarographic
wave is given approximately by
E = E1I2 + 0.916 RT In id -:- i (6.41)
aF l
E 112 = EO + R.J:.. In 1.35 koYt (6.42)

aF YDo
where t is the drop time. Since XD is proportional to YDot, eqs (6.41) and
(6.42) are functionally similar to eqs (6.32) and (6.33) for irreversible
steady-state voltammetric waves. Irreversible polarographic waves are
thus qualitatively similar to the curves shown in Figure 6.10. Again, if
EO is known, the shift in half-wave potential can be used to determine ko.
Cyclic Voltammetry
It is easy to predict the qualitative effect of slow electron transfer on

a cyclic voltammogram from our experience with chronoamperometry
and steady-state voltammetry. When ko decreases, a larger overpotential
is required so that cathodic peaks broaden and shift to more negative
potentials; similarly anodic peaks broaden and shift to more positive
potentials. This expectation is fulfilled by calculations based on the
mathematical model developed by Nicholson (16). The shape of a cyclic
voltammogram is found to depend on the transfer coefficient a and on
the dimensionless parameter
A = ko . / -.B:L
'V nDFu
where u is the potential scan rate. For A > 10, the voltammogram is
essentially reversible, independent of ko and a, but for smaller values of
A, peak shifts and broadening are expected as shown in Figure 6.16a.
When the transfer coefficient deviates from 0.5, cyclic voltammograms

become asymmetric, as shown in Figure 6.16b. Thus when a = 0.75, for
example, the cathodic peak is sharper than the anodic peak as expected
from the Butler-Volmer equation and Figure 6.2.
Just as plots of E vs. In[(iD - i)li] for voltammetric waves give
straight lines for both reversible and completely irreversible processes,
so plots of cyclic voltammetry peak currents vs. v 1l2 are linear for both
extremes. Such plots are not linear for quasi-reversible processes (A"" 1)
and so can be used (with care) for diagnostic purposes. The separation
of the anodic and cathodic peaks, ilEp , increases with decreasing values
of A as shown in Figure 6.17, and this parameter can be used to obtain a
rough measure of the electron-transfer rate constant; since !'!.Ep also
increases with increasing v because of solution iR drop, such a
measurement must be interpreted with caution.
The effects of iR drop are significantly reduced by the use of
microelectrodes. At fast scan rates, spherical diffusion negligible and
cyclic voltammogram peak separations can be interpreted using the
Nicholson approach. However, as we will see in Example 6.6,
considerable care is required to get accurate kinetic parameters by this
method.
0.4
0.3
~
en
.....,
0.2
'2
::l
0.1
t'
ell
.....,
1-0
0.0
.....
..c
1-0
-0.1
~
-0.2
-0.3
200 100 0 -100 -200 200 100 0 -100 -200

(E -E 1I2 )mV
Figure 6.16 Computed cyclic voltammograms for (a) A = 10, 1, 0.1, and
0.01, a. = 0.5, and (b) A = 0.3 and C1. = 0.25 (dashed curve), 0.5 (solid curve),
and 0.75 (dot-dash curve).
Figure 6.17 Separation of

cyclic voltammogram
peaks as a function of the
parameter A (ex = 0.5).
Reproduced with
permission from R. S.
Nicholson, Anal Chern.
1965,37, 1351, copyright
1965 American Chemical -1.0 -0.5 0.0 0.5 1.0
Society. log A.
Example 6.6 Safford and Weaver (17) have reported a careful

study of the oxidation offerrocene, 3 mM in acetone (with 0.3 M
Et4NPF6), propionitrile or butyronitrile (with 0.3 M BU4NPF6)
using cyclic voltammetry at a 5-llm gold microdisk.
Temperatures ranged from 198 to 298 K. The kinetic
parameter A. varies with temperature through both ko and D,
MI*(A.!"ff) = Iili*(ko) - MI*W)/2
so that the temperature dependence of D must be known in
order to interpret cyclic voltammetry data. Safford and Weaver
determined the diffusion coefficient of ferrocene from the
steady-state current in a microelectrode voltammogram-see
eq (4.50). The diffusion coefficient ranged from 2.0 x 10-9 to 2.8 x
10- 10 m 2 s· 1 in acetone, from 1.8 x 10- 9 to 2.1 x 10-lO in
propionitrile and from 8.7 x 10-10 to 9 x 10- 11 in butyronitrile as
the temperature changed from 298 to 198 K. The separation of
the anodic and cathodic peaks is increased by iR drop. Solution
resistivities were found to vary with solvent and temperature:
0.75 to 5.8 nom in acetone, 1.0 to 9.2 in propionitrile, and 1.3 to 11
in butyronitrile from 298 to 198 K. Voltammograms were
computer-simulated using experimental conditions of
temperature and resistance, and the computed peak
separation was subtracted from the experimental values to
obtain the kinetic contribution. A plot of this excess width is
shown as a function of scan rate in Figure 6.18 for data
obtained at 198 and 298 K in all three solvents.
200
6
Figure 6.18 Excess
(kinetic) separations of
the anodic and cathodic
>150 f- 6
peaks in cyclic voltam- ~
<.>
mograms of ferrocene Q,
in acetone (D), propio-
nitrile (0), and butyro-
~ 100
I
f-
6
6
g
D
0 -
nitrile (6) at 298 K (open <:!
symbols) and 198 K ~Q, 6 g
'-'
(filled
Reproduced
symbols).
with
50 I- D 6 -
D
permission from L. K. 6
!
0
Safford and M. J. 0
Weaver, J. Electroanal. 0 .B. Q D.
Chern. 1992 ,331, 857,
copyright 1992 Elsevier 0 25 50 75 100
Science Publishers. (uN s·1)112
Analysis of these data using curves similar to that in Figure

6.17 gave the ko = 0.055, 0.030, and 0.0095 m s-l at 298 K for
acetone, propionitrile, and butyronitrile solutions, respectively.
The temperature dependence led to activation enthalpies, /),}{:t:
= 19 kJ mol- 1 for all three solvents.
6.4 FARADAIC IMPEDANCE
Some of the best methods for studying the rates of electron-transfer

reactions involve a.c. measurement of the faradaic impedance, the
equivalent circuit parameter representing the barrier to current flow
between electrode and solution (see §3.1). The faradaic impedance
includes both the finite rate of electron transfer and the rate of transport
of electroactive material to the electrode surface. Since (as we will see
below) the current is expected to lag a bit behind the a.c. potential, the
faradaic impedance is not purely resistive but has a capacitive
component. The faradaic impedance acts in parallel with the double-
layer capacitance Cd, and this combination is in series with the solution
resistance R, as shown in Figure 3.1. In measurements of solution
conductance, the cell is designed with large electrodes to maximize the
double-layer capacitance and to minimize the faradaic impedance. Here
we are concerned with the faradaic component and the opposite strategy
is adopted. Low currents minimize the solution iR drop and small
electrodes tend to amplify the faradaic contribution. In practice,

attention must be paid to the separation of the effects of solution
resistance and double-layer capacitance from the measured impedance,
but we will assume in this section that the faradaic impedance is
measured directly.
Resistive and Capacitive Components

Suppose that the faradaic impedance is equivalent to a resistance
Rs and capacitance C s in series and that a sinusoidal current flows
through the circuit:
I(t) = 10 sin rot (6.43)
where ro is the angular frequency (21t times the frequency in hertz). The
potential drop is
<1> = IRs + Q/Cs
so that
or
dd<l> = Rsloro cos rot + 10 sin rot (6.44)
t Cs
The solution to this differential equation can be written
<l>(t) = ~<l> sin(rot + <p) + <l>dc
so that
~ = ro~<l> cos (rot + <p) = ro~<l> (cos rot cos cos sin = -JRs2 + lI(roCs )2
10 (6.46)
§6.4 Faradaic Impedance 353
Thus the magnitude of the faradaic impedance is given by the ratio of
the amplitudes of the a.c. potential and a.c. current and the resistive and
capacitive components can be separated if the phase angle is known.
Solution of the Diffusion Problem

For the electrode process
O+e- ~ R
the diffusion problem (see §4.1) can be solved with the periodic boundary
condition
10 sin rot == -FAJO(O,t) == FAJR(O,t)
to obtain the following expressions for the surface concentrations (see
Appendix 4):
Co(O,t) = Co* _ lo{sin rot - cos rot) (6.47a)
FAY 2roD o
(6.47b)
The surface concentrations undergo sinusoidal variations with

amplitudes proportional to the current amplitude 10 and inversely
proportional to VroD. Thus the amplitudes of the concentration
variations decrease as the frequency or the rate of diffusion increases.
Notice that the sinusoidal variation of CR(O,t) is 180° out of phase with
CO(O,t); since
sin rot - cos rot == f2 sin(rot - 1t/4)
CR(O,t) lags 45° behind the current.
Calculation of the Faradaic Impedance
Driving these concentration variations leads to both dissipative

(resistive) and reactive (capacitive) terms in the faradaic impedance. To
see how these develop, we need to derive an expression for d<l>/dt for
comparison with eq (6.44). Since the electrochemical system is assumed
to be near equilibrium, 11 is very small and CO(O,t) "" CO*, CR(O,t) "" CR*.
Thus eq (6.10) can be expanded, retaining only first-order terms, to obtain
/(t) == io(CO(o,t) CR(O,t) _ F11(t»)
Co* CR* RT
or
Tl(t) == _ RX.(Co(O,t) CR(O,t) + l(t»)

F Co* CR* io
The time derivative ofTl can then be written:
dTl =R M+ A dCo + A dCR (6.48)
dt et dt pO dt pR dt
where Ret is the charge-transfer resistance
Ret = OTt
al
=-B:!
FLO
(6.49)
and the other parameters are
/30 = OTt =_ RT (6.50a)

aco FCo*
/3R = OTt =+ RT (6.50b)

aCR FCR*
The time derivatives of I, CO, and CR can be computed from eqs (6.43)
and (6.47):
~ = loro cos rot (6.51)
dCo _ loro(cos rot + sin rot) (6.52a)

(it-- FAV 2Doro
dCR loro (cos rot + sin rot)

- - = + -"---'-'-~-======----'-'- (6.52b)
dt FAV2DRro
Combining eqs (6.48), (6.51) and (6.52), we have
~ = (Ret + ~) 10ro cos rot + 10aVID sin rot (6.53)
where
a---L(~-~)
- FA V2DR V2Do
or, using eqs (6.50),
a-~( 1 + 1 ) (6.54)
- F2A CR*V2DR Co*V2Do
We can now identify the equivalent circuit parameters Rs and Cs by
comparison of eqs (6.44) and (6.53):
§6A Faradaic Impedance 355
Rs = Ret + cr/iffi (6.55)

Cs = 1/('5'{(j) (6.56)
or
IZfl =-IRc?+ 2cr2/w (6.57)
cot <p = 1 + Rct'fw/cr (6.58)
The faradaic impedance is seen to have two components: the charge-
transfer resistance, which is inversely proportional to the exchange
current ~nd thus is a measure of the rate of electron transfer; and the
term crY 2/w, which relates to mass transport and contributes equally to
the resistive and reactive parts of the impedance. This term is
sometimes called the transport impedance or Warburg impedance to
distinguish it from the kinetic contribution to the faradaic impedance.
For a reversible charge-transfer process, the transport impedance
dominates, crY 2/w » Ret, so that cot <p = 1. Thus in a faradaic impedance
experiment, <p = 45° is a criterion for reversibility. An immediate
consequence of slow electron-transfer is a reduction of the phase angle.
If the magnitude of the faradaic impedance, IZA, and the phase
angle <p are measured, eqs (6.57) and (6.58) can be used to separate the
contributions from mass transport and charge transfer. Thus faradaic
impedance measurements have been rather commonly used for
determination of electron-transfer rate constants.
Example 6.7 A faradaic impedance experiment at 1 kHz (w =

21t x 103 rad s-I), a disk indicator electrode with r = 1 mm, and a
solution with CO* = CR* = 1 mM, DO =DR = 10-9 m 2 s- 1 , T = 298
K, gave <p = 35°. Compute the components of the transport
impedance, the exchange current io, and the electron-transfer
rate constant ko. Assume that the d.c. potential is very near the
equilibrium value. What phase angle would be expected if the
frequency were increased to 10 kHz?
Substituting the concentrations, diffusion coefficients, electrode

area and temperature into eq (6.54), we have
cr = 3.79 x 103 Q s-1I2
cr/Vw = 478 Q
Cs = 1/('5'{(j) =3.33 JlF
With <p = 35°, cot <p = 1.428, we have
RetVw/cr = 0.428
Ret = 205 Q
Using eq (6.49), we obtain

io = 1.26 x 10-4 A
and eq (6.11) with E = Ee gives
ko = 4.1 x 10-4 m s-1
Increasing the frequency by a factor of 10 decreases the
transport impedance, making the faradaic impedance more
resistive and decreasing the phase angle. Since Ret and (J are
frequency-independent, eq (6.58) gives
cot cp =1 + 0.428 -riO =2.353
Direct measurement of the faradaic impedance for a system at

equilibrium works well when both components of the system are
chemically stable. There are many cases, however, where one of the
components of the electrode couple is of limited stability so that a
solution containing known amounts of 0 and R cannot be prepared. The
general ideas of the faradaic impedance method can still be used;
however, we start with a solution containing only 0 (or R) and apply a
d.c. potential to adjust the OIR ratio at the electrode surface. This
marriage of d.c. voltammetry and faradaic impedance methods can be
applied using a variety of indicator electrodes, but the most common
application has been with the dropping mercury electrode, a technique
called a.c. polarography.
A.C. Polarography
In a.c. polarography, a small sinusoidal a.c. voltage is
superimposed on the d.c. ramp voltage and the a.c. component of the cell
current is measured using a lock-in detector so that the in-phase and
out-of-phase current components can be determined separately.! In
other words, the cell potential is modulated by an amount l:!.E and the
resulting current modulation !::.i is measured. For small values of l:!.E,
the a.c. polarogram should approximate the first derivative of the d.c.
polarogram (see Problems). The a.c. modulation voltage is typically in
the audio-frequency range, ca. 10 Hz - 10 kHz, with an amplitude of 5-10
mY. In practice, the a.c. current oscillates with drop growth and
dislodgement; the current is usually measured at maximum drop size,
i.e., just before drop fall.
! Alternatively, the magnitude of the a.c. current and the phase angle can be
determined. Fourier transform methods can also be applied to a.c. polarography.
For further details on the theory and practice of a.c. polarography, see Smith (18).
We can derive the a.c. current - d.c. potential curve for a reversible
process from the results of our discussion of the faradaic impedance. If
the process is reversible, the charge transfer resistance is negligible
and, according to eq (6.57), the faradaic impedance is
I Zrl = crY 2/ro
In general the surface concentrations CoCO,t) and CR(O,t) will be very
different from their bulk solution values. However, the sinusoidal
variations are about mean values determined by the d.c. potential. Thus
we can replace Co* and CR* in eq (6.54) by the mean values given by eqs
(4.3), which we can rewrite in slightly more compact form in terms of
the parameter
e' = exp F(E - EJ/2)
--'--=-=---=
RT
The mean surface concentrations (for CR* = 0) are
Co(O,t) = Co* ~
1+ e'
CR(O,t) = Co* YDolDR
1+ e'
Substituting these expressions in eq (6.54) in place of Co* and CR*, we
have
If the potential is
E(t) = Edc + till sin(rot)
the a.c. component of the current will be
I(t) = 10 sin(wt + 7U'4)
since cot <p = 1 for Ret = O. The current amplitude is
10=1~
or
10 = F 2ACo*YJ5O(.;) till e' (6.59)
RT (1 + e'f
Although the problem can be formulated to account for the expanding
spherical drop of the d.m.e., it is usually more convenient to make the
required corrections empirically by determining the ratio of the a.c. to
d.c. currents, measured for the same drop time on the same solution.
Using the Cottrell equation for iD, we obtain
10 FfiW)[!ill 8'
(6.60)
iD = RT(l + 8,)2
The current at E =El/2 (8' = 1) is

I _ F2ACO*fl5()O) !ill
p- 4RT
The shape of the current-potential curve can be seen more easily
considering the ratio IlIp:
I 48'
Ip =(1+8,)2
The ratio can be inverted and written as
Ip(1 + 8,)2
T= 48'
or
Ip-I
--='----'-
(1- 8,)2
I 48'
Taking the square root and subtracting these equations, we have
fi_yIp;I=W
Taking logs, we obtain
E=E1J2±2~Tln[{!f-- yIp;I] (6.61)
This equation describes the shape of an a.c. polarogram for a reversible

system, shown in Figure 6.19. Implicit in the derivation of eqs (6.59) and
(6.61) has been the assumption that !ill is small (5-10 mV in practice). At
larger modulation amplitudes, the proportionality between peak current
and !ill breaks down and the peak potential is no longer equal to E 1/2.
At low concentrations of electroactive material, the contribution of
the double-layer capacitance is no longer negligible. However, because
the resistive and capacitive components of the a.c. current are 90 0 out of
phase, phase-sensitive detection of the a.c. signal can select the resistive
component and almost wholly reject the contribution of the double-layer
capacitive charging current. The sensitivity of a.c. polarography is
therefore very good, comparable with differential pulse polarography.
Even better selectivity for the resistive component, and correspondingly
better analytical sensitivity, can be achieved with second-harmonic a.c.
1.0
0.8
Q
:-s 0.6
r..
0
~Q.. 0.4
.....
0.2
Figure 6.19 Comparison (0.0)

of the shapes of
reversible a.c. and d.c. 0.2 0.1 0.0 -0.1 -0.2
polarographic waves. (E -E I /2)N
polarography where the a.c. current at twice the modulation frequency is

detected. Since the faradaic impedance is nonlinear, the second
harmonic signal is quite large; the nearly linear double-layer capacitive
contribution gives very little second harmonic current. The second
harmonic a.c. polarographic curve resembles the second derivative of
the d.c. polarographic wave. Just as the resolution of an a.c. or
differential pulse polarogram is better than that of a d.c. polarogram, a
further improvement in resolution is achieved in second-harmonic a.c.
polarography.
Example 6.8 Compute the ratio of the peak current

amplitude for an a.c. polarogram to the d.c. diffusion-limited
current. Assume a modulation amplitude of 5 mY, a
modulation frequency of 1 kHz, and a drop time of 5 s.
At the peak potential, e' = 1, so that eq (6.60) becomes

Ip = F t,E Vrrwt
iD 4RT
Substitution of the numbers gives
IoIiD = 15.3
Since the a.c. signal can be amplified by a tuned amplifier, the
signal-to-noise ratio for the a.c. polarographic peak current is
expected to be very high.
So far we have assumed nernstian behavior in this discussion of

a.c. polarography, but one of the most interesting applications of the
technique is the measurement of electron-transfer rate constants. An
a.c. polarogram will appear reversible when YwDolko «1. When the
rate of electron transfer is slower or the frequency higher, the height of
the a.c. polarographic peak is decreased and the peak broadens and
becomes somewhat unsymmetrical. The current can then be expressed
by
I(rot) = 10' sin(rot + <p)
where
(6.62)
(6.63)
a. and ~ are the cathodic and anodic transfer coefficients, 10 is given by eq

(6.59), Ais given by eq (6.38),
A = ~(a'-(J. + 9'~)
W
and we have again assumed equal diffusion coefficients for 0 and R. If
we use eqs (6.36) for the surface concentrations, the bracketed term in eq
(6.62) becomes
1 + (al9' _ ~)F(Avt)
AW
where F(Afi) is given by eq (6.37) and t is the d.m.e. drop time. The phase
angle <p is given by
<p = cot- 1[1 + f20)/A] (6.64)
These equations are sufficiently complex that pictures are more useful
in gaining a feeling for the behavior of the system. Thus a.c.
polarographic currents computed using eqs (6.62) and (6.63) for several
values of ko and a. are shown as functions of Edc in Figure 6.20.
When the electron-transfer process is fast enough that the d.c.
polarographic wave appears reversible, F(Afi)/AW" 0 and the term in
brackets in eq (6.62) reduces to 1. The function G(ro) varies from 1 when
vro/A « 1 to A/vro for high frequencies or slow rate constants. The
predicted frequency dependence of the peak current for a. = 0.5 is shown
in Figure 6.21. Since 10 is proportional to vro, the a.c. current becomes
frequency independent at high frequencies. Although this condition
may not be experimentally attainable, it is often possible to extrapolate
experimental data to infinite frequency where the a.c. current at E = E1I2
(the peak current for ex = 0.5), scaled by the d.c. diffusion current, is given
by
('D
~) =~ vrttlD t1E ko
2RT
Cl)=oo
(6.65)
The electron-transfer rate constant can also be extracted from the
phase angle. According to eq (6.64), cot <p is linear in "f'iO)/'A.. If a. = 0.5, the
peak current corresponds to e' = 1 so that eq (6.38) reduces to A. = 2kc!W.
Thus a plot of cot <p us. Yo) should have a zero-frequency intercept of 1 and
a slope of VDl2lko.
Figure 6.20 Ac. current

vs. d.c. potential for kO =
00 (dotted curve) and for
(solid curves, top to
bottom) k 0 = 10- 2 , 10- 3 , 15
and 10- 4 m s-l, ex = 0.5.
The dashed curves
correspond to kO = 10- 3 10
and 10-4 m s-l, but with ex
= 0.25 (peaks shifted to
positive potentials) or 5
0.75 (peaks shifted to
negative potentials).
Other parameters are D
= lO- g m2 s- 1 , A = 3.5 o ~~~~~~~~~DB~~
mm 2 , CO· = 1 mM, 00 = 200 100 0 -100 -200

2500 s-l, and t.E = 5 mY. (Edc -E 1/2)mV
Example 6.9 Dietz and Peover (19) employed a.c.

polarography to study the rate of reduction of trans-stilbene,
2,4,6-trimethyl-trans-stilbene, and 2,4,6,2',4' ,6'-hexamethyl-
trans-stilbene in N,N-dimethylformamide solutions. The
phase angle <p, measured using an impedance bridge, is plotted
as cot <p us. (01/2 in Figure 6.22. Given diffusion coefficients,
measured under the same conditions, the slopes of these plots
were used with eq (6.64) to compute the electron-transfer rate
constants, obtaining:
trans-stilbene ko = 1.2 x 10-2 m s·l
trimethyl-trans-stilbene ko = 5.2 x 10.3 m s·l
hexamethyl-trans-stilbene ko = 1.8 x 10-3 m s·l
50 I
2.5
I
[pillA I
I
I cot cp
I
40 I
I
I
2.0
I
I
I
I
30 I
I
I
1.5
""
ID
1.0
10
0 0.5
0 50 100 150 o 50 100 150
(OY'S-l )1/2 (OY'S-l )1/2
Figure 6.21 Frequency dependence Figure 6.22 Plots of cot cp us. 00 112
of the a.c. polarographic peak for a.c. polarographic currents
current for kO = (dashed line) 00 measured for trans-stilbene (0),
and ko = 0.01, 0.001, and 0.0001 m trimethyl-trans-stilbene (Il), and
s-l. Other parameters are as in hexamethyl-trans-stilbene (<». Re-
Figure 6.20. produced with permission from R.
Dietz and M. E. Peover, Disc. Far-
aday Soc. 1968,45, 154, copyright
1968 Royal Society of Chemistry.
Since the activation energies obtained by comparison of the

above data with values determined at -20°C were about the
same for trans-stilbene and the methyl derivatives (ca. 20 kJ
mol-I), Dietz and Peover suggested that the reduction in ko with
increasing methyl substitution is due to steric requirements on
the orientation of the molecule at the electrode prior to electron
transfer, i.e., to changes in the entropy of activation. Electron
transfer to trans-stilbene is apparently possible for more
different orientations than for the more hindered methyl
derivatives.
Coupled Chemical Reactions

In principle it is possible to use a.c. polarography to measure the
rates of many types of chemical reactions coupled to electron-transfer
processes. The technique has not been as widely used as the reversal
methods we have just discussed primarily because a.c. polarography is
not readily used in a qualitative way to characterize a system;

quantitative measurements often grow out of qualitative studies using
the less precise methods. Nonetheless, for accurate rate constant
measurements on a well-characterized system, particularly for very fast
rates, a.c. polarography is the method of choice.
In general, theoretical expressions for the a.c. polarographic
current are extremely complex, involving integrals which must be
evaluated numerically. Although this can be done, in practice it is
usually much easier to measure the phase angle, which can be
expressed as an analytical function of rate constants, potential and
frequency. Furthermore, the phase angle is dimensionless and
independent of electrode area and geometry. For an irreversible
chemical step following electron transfer, it can be shown that
cot (j):= V/U (6.66)
where
v = iiiii. + 1 + G+(k/w) e' (6.67a)
A 1+ e'
U:= 1 + G_ (k/w) e' (6.67b)
1+ e'
with
Gik/w) = (-..! 1 + (k/W)2 ± k/w )V2 (6.68)

1 + (k/W)2
and A is given by eq (6.38). The effect of the added terms in G±(k/w) is to
produce a hump in the cot (j) us. Vw plot at w/k '" 0.39. If the electron-
transfer rate is fast enough, this hump may be resolved and the rate
constant of the chemical step can be determined quite simply. When the
electron-transfer rate is slow enough to contribute a significant slope in
the absence of the chemical step, however, the hump may be manifested
only by a change in slope.
Example 6.8 Moraczewski and Geiger (20) studied the

reduction of (TJ 4- cycloocta tetraene)-( TJ 5- cyclopen tadienyl)cobal t,
(COT)CoCp, using the techniques of d.c. and a.c. polarography
and cyclic voltammetry.
Cyclic voltammograms of (COT)CoCp in DMF solution, shown

in Figure 6.23, have two cathodic peaks (El/2 := -1.82 and -2.05 V
us. s.c.e.). These reductions are assigned to the isomeric forms
in which the cyclooctatetraene ligand is bound through the 1,3
and 1,5 double bonds, respectively. Since the peak currents are
proportional to concentration, the peak current ratio gives the

equilibrium constant
K)1,5]
[1,3]
Thus K = 3.3 at 298 K and 1.46 at 343 K (MIo =-13 kJ mol-I). On
the reverse scans of the cyclic voltammograms, a single anodic
peak is observed, apparently corresponding to the oxidation of
the radical anion of the 1,3-isomer. Furthermore, on repetitive
cycles, the cathodic peak due to the 1,3-isomer increases at the
expense of the peak due to the 1,5-isomer until the cyclic
voltammogram shows an apparently reversible pattern,
centered at -1.82 V, with only a trace of the more negative peak
remaining. Isomerization of neutral (COT)CoCp is quite slow
but (COT)CoCp- seems not to exist at all as the 1,5-isomer.
These results could be accommodated by either of the two
schemes:
e-
(1,5-COT)CoCp P (1,5-COT)CoCp- ( 1,5-COT)CoCp
!Tslow e- !fast !Tslow

(1,3-COT)CoCp p (1,3-COT)CoCp- 1l,3-COT)CoCp P (1,3-COT)CoCp-
Scheme I Scheme II

grams at a hanging mercury drop
electrode (u '" 19 V s-l) of
(COT)CoCp in DMF solution at
various temperatures. Reproduced
with permission from J.
Moraczewski and W. E. Geiger, J.
Am. Chemi. Soc. 1981, 103, 4779, 328K 343K
copyright 1981 American Chemical
Society.
In scheme I, electron transfer to (1,5-COT)CoCp is presumed to

be reversible but perturbed by very rapid isomerization of the
radical anion to the more stable 1,3-isomer. In scheme II,
isomerization is presumed to be concerted with electron
transfer. If scheme II is correct, the electron-transfer rate
constant might be expected to be slower than normal. Indeed,
a.c. polarograms show that the peak due to reduction of the 1,5-
isomer is much smaller than that due to the 1,3-isomer. Plots
of cot q> us. Vw were linear for both peaks for frequencies up to
about 200 Hz (Vw < 35 s1l2). Poor signal-to-noise ratios
precluded work at higher frequencies. The slopes of the cot <1>
us. Vw plots suggested that electron transfer to the 1,5-isomer is
about 10 times slower than to the l,3-isomer.
It was later realized that, because of the limited frequency

range, the a.c. polarographic result might be consistent with
scheme 1. Accordingly, the experiment was repeated using a
fast Fourier-transform technique which permitted
measurements at considerably higher frequencies (21). This
work gave the cot q> us. Vw plots shown in Figure 6.24. The data
for the 1,3-isomer give a good straight line with slope
corresponding to an electron-transfer rate constant ko = 2.8 x
10- 3 m s·l. The 1,5-isomer data, on the other hand, show
considerable kinetic control at low frequencies; cot <1>
asymptotically approaches the linear dependence on Vw at high
Figure 6.24 Plots of cot q> cot q>

v s. Yro for a.c. polaro·
graphic currents measured
for (COT)CoCp in DMF 3
solution. The squares and
circles correspond to th<.;
1,3· and 1,5 isomers,
respectively. The dashed
2
line corres~onds to ko = 6 x
10. 4 m s·. Reproduced
with permission from M.
Grzeszczuk, D. E. Smith,
and W. E. Geiger, J. Am. 1
Chern. Soc. 1983, 105, 1772,
copyright 1983 American
o 40 80 120
Chemical Society. (wls·1 )112
frequencies. The data of Figure 6.24 can be interpreted in

terms of an electron-transfer rate constant ko =6 x 10-4 m s-l,
somewhat slower than the value for the 1,3-isomer, but with a
following chemical step with rate constant k =2000 s-l.
These results suggest that electron-transfer rate constants in
general must be viewed with caution. There is always the
chance that a rate constant appears to be slow not because the
electron-transfer step is slow, but because a very fast chemical
transformation is coupled to electron transfer.
A.C. Cyclic Voltammetry

A relatively recent technical development is the combination of the
methods of a.c. polarography and cyclic voltammetry. A low-frequency
triangular-wave potential, modulated by a higher-frequency, small-
amplitude, sinusoidal signal, is applied to the cell and the a.c. current at
the frequency of the modulation signal is measured. The method has
most of the advantages of a.c. polarography but is not restricted to the
dropping mercury electrode. Since stationary planar electrodes can be
used, the method is more general and the detailed theory somewhat
easier. Since double-layer capacitive charging current is less important
than in conventional cyclic voltammetry, quantitative measurements
are considerably easier. Bond, o'Halloran, Ruzic, and Smith (22) have
shown that the theory of a.c. polarography can be used without
significant modification provided that the triangular-wave frequency is
much less than the sinusoidal frequency. Thus eqs (6.62) - (6.64) are
valid for a.c. cyclic voltammetry. Unless the d.c. scan rate is very slow,
however, eqs (6.36) cannot be used for the surface concentrations
required in eq (6.62). Bond and coworkers used a digital simulation
technique (see Appendix 5) to obtain these parameters, but an approach
based on Nicholson's methods (16) would probably also work.
For electron-transfer processes which are reversible or quasi-
reversible on the cyclic voltammetry time scale, a.c. cyclic
voltammograms show current peaks on both the cathodic and anodic
scans. The magnitudes and potentials of the peak currents are sensitive
to the electron-transfer rate constant ko and to the transfer coefficient a..
The latter parameter is particularly easy to measure by this technique.
Since the current returns to zero following the peak, the zero of current
is well defined and accurate peak current measurements are much
easier than in conventional cyclic voltammetry.
A.c. cyclic voltammetry is particularly well suited to the study of EC
processes (23). The ratio of the a.c. current peaks on the forward and
reverse scans can be accurately measured and this ratio is very sensitive
to the rate constant of the following chemical reaction.
REFERENCES
(Reference numbers preceded by a letter, e.g. (E5), refer to a book listed in

the Bibliography.)
1. R. A. Marcus, J. Chern. Phys. 1965,43,679; Ann. Rev. Phys. Chern.
1964, 15, 155.
2. J. A. V. Butler, Trans. Faraday Soc. 1924, 19, 734.
3. T. Erdey-Gruz and M. Volmer, Z. phys. Chern. 1930, 150A, 203.
4. R. D. Allendoerfer and P. H. Rieger, J. Am. Chern. Soc. 1965,87,
2336.
5. B. J. Heubert and D. E. Smith, J. Electroanal. Chern. 1971,31,333.
6. A. N. Frumkin, Z. phys. Chern. 1933, 164A, 121.
7. R. Parsons, Adv. Electrochem. Electrochem. Engin. 1961, 1, 1.
8. J. Tafel, Z. phys. Chern. 1905,50,641.
9. K. J. Vetter and G. Manecke, Z. phys. Chern. 1950,195,270.
10. K. J. Vetter, Z. phys. Chem. 1952, 199, 285.
11. J. O'M. Bockris, D. Drazic, and A. R. Despic, Electrochim. Acta 1961,
4,325.
12. S. Trasatti, J. Electroanal. Chem. 1972, 39, 163.
13. K. B. Oldham, J. C. Myland, C. G. Zoski, and A. M. Bond, J.
Electroanal. Chern. 1989,270, 79.
14. Y. Zhang, C. D. Baer, C. Camaioni-Neto, P. O'Brien, and D. A.
Sweigart, Inorg. Chern. 1991,30, 1682.
15. L. Meites and Y. Israel, J. Electroanal. Chern. 1964,8, 99.
16. R. S. Nicholson, Anal. Chern. 1965,37, 1351.
17. L. K. Safford and M. J. Weaver, J. Electroanal. Chern. 1992,331,857.
18. D. E. Smith, Electroanalytical Chemistry 1966,1, 1; Crit. Rev. Anal.
Chern. 1971,2,247.
19. R. Dietz and M. E. Peover, Disc. Faraday Soc. 1968,45,154.
20. J. Moraczewski and W. E. Geiger, J. Am. Chern. Soc. 1981,103,4779.
21. M. Grzeszczuk, D. E. Smith, and W. E. Geiger, J. Am. Chern. Soc.
1983,105, 1772.
22. A. M. Bond, R. J. O'Halloran, I. Ruzic, and D. E. Smith, Anal. Chern.
1976, 48, 872.
23. A. M. Bond, R. J. O'Halloran, 1. Ruzic, and D. E. Smith, J.
Electroanal. Chern. 1982,132,39.
PROBLEMS
6.1 Show that either eq (6.lla) or (6.llb) can be rearranged (with the
help ofthe Nernst equation) to
io = FAkoCOPCRP
6.2 Derive an equation for the current, analogous to eq (6.30), for the
case that Co* and CR* are both nonzero.
6.3 Derive eqs (6.23) from eqs (6.4) and (6.11).
6.4 Show that the data discussed in Example 6.2 are consistent with
the simple mechanism
Mn3 + + e- ~ Mn 2+
when subjected to the more complete analysis of Example 6.3.
6.5 Figure 6.25 shows a Tafel plot for the reduction ofH+(aq) in a 0.1 M
solution of HCI at a nickel electrode in the presence of 1 bar H2(g) at
20°C.
(a) Determine the exchange current density jo and the apparent
transfer coefficient.
(b) Compute the electron-transfer rate constant ko.
1
c:1
S
~o
b.O ,
o
, ,,
,,
- -1
,
"",
-2
Figure 6.25. Tafel plot for the

-0.0 -0.1 -0.2 -0.3 -0.4
reduction of H+ at a nickel
cathode. 1l /V
Problems 369
6.6 Derive eqs (6.26) by an alternative route in which the derivatives
(ddl~n~:LR and (ddl~n~:)Co

are first computed.
6.7 Bockris, Drazic, and Despic (see Example 6.4) found that the
exchange current density for the oxidation of iron depends on the
concentration of Fe2 + according to
dIn IJol -08
d In [Fe2+] - .
Show that the mechanism discussed in Example 6.4 leads to
d In liol = 1- a/2
dInGo
and thus is consistent with the experimental result.
6.8 The exchange current for an EE reduction with E2° »E1° and the
first electron-transfer step rate-limiting is given by eqs (6.27) and
the net current is given by the Butler-Volmer equation with J3app = 1
+ J3 where J3 is the anodic transfer coefficient for the rate-limIting
step. Derive analogous expressions for the case where the second
electron-transfer step is rate-limiting.
6.9 Equation (6.33) gives the half-wave potential of an irreversible wave

for ko/kD «1. For quasi-reversible waves, El/2 must be computed
from eq (6.31).
(a) Compute El/2 - EO for kD = ko, a = 0.5.
(b) What is the maximum value of kolkD for which eq (6.33) gives
E1/2 - EO correct to within 1 mY?
6.10 The Tome~ criterion for a reversible one-electron wave is !!.E = E1/4
-E3/4 = 56.51n mV at 25°C.
(a) Compute !!.E for several values of kD1ko in the range 0.1 to 2
using eq (6.31) (a = 0.5) and plot!!.E us. log(ko/kD).
(b) A r.d.e. voltammogram gave a value of !!.E = 75 mV after
correction for solution iR drop. Ifv = 10- 6 m 2 s- 1, D = 10-9 m 2 s- 1, and
ro = 500 s-1, what is the approximate value of ko?
6.11 (a) Show that the chronocoulometric charge for a potential step
experiment (see Problem 4.11) involving an electron-transfer step of
finite rate is
Q = FACO*kcA.-2(exp(A,2t)[I- erflNI"t)] + 2A,M -I}
where A, is given by eq (6.38). Hint: Integrate eq (6.39) by parts.
(b) For sufficiently long times, the first term in the bracketted
expression goes to zero and Q becomes linear in Vt. The linear
region of a plot of Q vs. Vt can be extrapolated to t = 0 or to Q = O.
Obtain expressions forQ extrapolated to t = 0 and for Vt
extrapolated to Q = O. Describe how such data could be used to
obtain the electron-transfer rate constant ko.
6.12 The following data were obtained for a potential step
chronocoulometry experiment. The solution contained 1 mM 0
which is reduced by one electron, and the potential step was to E =
Em - 50 mY. Assume that DO =10-9 m2 s- 1, (l =0.5,A =3.1 x 10-6 m 2
'tIs 1 3 5 7 9
Q/mC 1.13 2.10 2.78 3.33 3.81
Plot Q vs. -ft, extrapolate to Q = 0 and estimate the electron-transfer

rate constant ko using the equations derived in Problem 6.11.
6.13 Derive expressions for the resistive and capacitive components of
the faradaic impedance in terms of I Zfl and <po
6.14 (a) Show that eq (6.58) can be derived (within a multiplicative
factor) by differentiation of the Heyrovsky-Ilkovic equation to obtain
dildE.
(b) What is the width at half height of a reversible one-electron a.c.
polarographic wave?
6.15 (a) Find the shape of a second-harmonic a.c. polarogram by

computing d 2i1dE 2 ; prepare a plot of this function vs. Edc.
(b) What is the width (between current extrema) of a reversible
one-electron second-harmonic a.c. wave?
6.16 Given the time scales accessible to the various experimental
techniques (§5.1), and assuming D = 10-9 m 2 s- 1 , v = 10-0 m 2 s· 1, (l =
0.5, what is the fastest electron-transfer rate constant ko which
could be measured for a simple one-electron reduction using:
(a) Potential step chronoamperometry, (E _EO) = -0.2 V?
(b) r.d.e. voltammetry?
(c) a.c. polarography?
6.17 Derive eq (6.34) for the current in an irreversible EE process.

7
ELECTROL YSIS
In this chapter we will be concerned with electrolysis experiments

in which a significant fraction of the electroactive material is consumed.
We begin in §7.1 with a discussion of some general aspects of electrolysis
and proceed to analytical applications in §7.2. In §7.3 and §7.4, we
explore applications of electrolysis to chemical synthesis, including
examples from industrial chemistry. A short-circuited galvanic cell is a
self-powered electrolysis cell. This notion will give us a useful model for
a discussion of corrosion in §7.5.
7.1 BULK ELECTROLYSIS
Volta's invention of the first practical source of electrical power in

1800 stimulated a flurry of activity among scientists. Indeed, the ideas
for many bulk electrolysis experiments can be traced to experiments
done in the first decade of the nineteenth century. We have already
mentioned the electrolysis of water by Nicholson and Carlisle. This
work led to Erman's discovery that the conductivity of salt solutions
depends on concentration. In another series of experiments following
the Nicholson-Carlisle discovery, Cruikshank found that when a
solution of a salt such as copper sulfate is electrolyzed, metallic copper is
plated onto the negative electrode and no hydrogen is evolved (1). Many
other important discoveries were made in the following years, and in
1833 Faraday (2) summarized the quantitative aspects of electrolysis
experiments in two statements which have come to be known as
Faraday's laws of electrolysis:
The weight of metal platen on the cathode during passage of
current through a solution of the metal salt is proportional to:
(1) the charge passed through the solution; and (2) the
equivalent weight (atomic weight/oxidation number) of the
metal.
371
372 Electrolysis
William Cruikshank was Lecturer in Chemistry at the Royal Military

Academy, Woolwich, from 1788 until 1804. Little is known of
Cruikshank's life, but he made important contributions to early
electrochemistry and to the chemistry of simple gases (he is credited with
the discovery of carbon monoxide).
In modem terms, we would say that one mole of electrons (one Faraday)
will reduce one mole of Na+, one-half mole of Ca2 +, or one-third mole of
AI3+. Faraday's work not only put electrolysis on a quantitative basis but
laid the groundwork for the more complete understanding of the role of
electrons in chemistry which began to emerge 50 years later.
Exhaustive Electrolysis
When an electrolyte solution is subjected to electrolysis, either to
separate a component of the solution or to oxidize or reduce a substrate,
rather large amounts of electrical charge must be passed through the
cell. As we saw in Chapter 4 (Example 4.7), the net electrolysis at a
small electrode (e.g., a d.m.e.) in an unstirred solution is negligible, even
after several hours. Thus our first concern is to see how the current can
be increased to obtain significant net electrolysis in a reasonable time.
One obvious way to increase the current through an electrolysis cell
is to stir the solution, thus increasing the rate of mass transport. In a
stirred solution, the diffusion layer thickness can be reduced to as little
as tens of microns. Suppose that the electrode process is
O+n e- ~ R
and that a linear concentration gradient in 0 is set up across a diffusion
layer of constant thickness. If the potential of the working electrode is
sufficiently negative that the concentration of 0 is zero at the electrode
surface, then the current is given by eq (4.14b),
i = nFAkDCo* (4. 14b)
where kD = DlxD is the mass transport rate constant. The number of
moles of 0 reduced is related to the charge transferred by
dno=- dQ
nF
Since dQ =idt and dno =VdCo* (V is the solution volume), the change in
bulk concentration on passage of current for a time interval dt is
dCo*=-~dt
nFV
Substituting eq (4.14b) for i, we have
dCo* = - AkDCO* dt
V
§7.1 Bulk Electrolysis 373
Integration of this expression gives
CO*(t) = CO*(O) exp(-kt) (7.1)
where
k =knAfV (7.2)
Thus we expect the bulk concentration C* to fall exponentially with a
rate proportional to the rate of mass transport and to the ratio of the
electrode area to the solution volume. Thus if the process is to be
completed in minimum time, the general strategy for exhaustive
electrolysis should include efficient stirring and a high area-to-volume
ratio. Since the stirring must be particularly efficient at the working
electrode surface, careful attention to electrode geometry is required for
optimum performance.
Example 7.1 Suppose that 250 cm3 of 0.1 M CUS04 is
electrolyzed using a cathode of area 250 cm2 and an initial
current of 5 A. How long will it take to remove 99% of the
copper from solution? 99.99%? Assume that the cathode
process is rate limiting. From eq (4.14b), we obtain the
diffusional rate constant
kD = il2FACO*
Substituting the electrode area and the initial current and
concentration, we have
kD = 1.04 x 10-5 m s-l
With the area-to-volume ratio of 100 m- 1, eq (7.2) gives
k = 1.04 x 10-3 s-l
To reduce the copper concentration from 0.1 M to 0.001 M thus
requires a time
t =- [In(O.Ol)Vk =4440 s (74 minutes)
To reduce the concentration to 10- 5 M amounts to another
factor of 100 decrease in concentration and thus will require
another 74 minutes or 148 minutes in all.
Electroseparation
Electrolysis is often used to separate the components of a mixture.
For example, a desired metallic component might be plated out at the
cathode or a nonmetal could be oxidized to a gas or deposited as an
374 Electrolysis
insoluble oxide at the anode. Alternatively, we might try to selectively

remove an impurity from solution by oxidation or reduction.
The factors which govern the efficiency of electro separations are
important in both the analytical and synthetic applications of
electrolysis. If another electrode process competes with the one of
interest, then nonproductive current flows. Electrode side reactions lead
to departures from a simple application of Faraday's laws and thus
invalidate analytical procedures based on the proportionality of analyte
consumed to charge passed. Side reactions also introduce unwanted
contaminations in electro synthetic methods and constitute an important
waste of electrical energy in industrial applications. A useful measure
of the significance of competing electrode processes is the current
efficiency, defined by
n x moles of product formed
~=--------~---------- (7.3)
Faradays of charge passed
where n is the numbers of Faradays consumed per mole of product
formed.
As a simple model for separation efficiency, we consider two
couples,
01 + n1 e- -t Rl
02 + n2 e- -t R2
both involving solution species only. We assume that both are reversible.
We then ask what separation between standard (formal) potentials is
required in order for one species, say 01, to be 99.9% reduced to R1
without reducing 02 by more than 0.1%. The Nernst equation can be
written for the first couple as
E = Et + RT In COl
nlF CRI
Defining Xl as the fraction in the reduced form
Xl = CRI
CRI + COl
the N ernst equation can be reformulated:
E =Elo + RT In I-Xl (7.4a)
nJR' Xl
and similarly for species 2,
E = E20 + B:l:...ln 1 - X2 (7.4b)
n<j? X2
§7.1 Bulk Electrolysis 375
1.0
"C
0.8
~
~
"C
Q) 0.6
~
d
.....
....0 0.4
~
e
c:.l
0.2
Figure 7.1 Optimum
separation is achieved in
electroseparations when 0.0
the potential is set 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3
midway between half-
wave potentials. (E - E 1I2 avg )N
The fractions Xl and X2 are plotted us. potential in Figure 7.1 for a
difference in standard potential, E1° - E2° =0.20 V and n1 =n2 = 1. It is
fairly easy to prove (see Problems) that, when n1 = n2, the optimum
separation is achieved, (i.e., Xl -X2 is maximized) when
E = (E1° + E2°)/2 (7.5)
When this condition is met,X2 = 1-X1 and
E I -E20 = RT In Xl (7.6)
2nF X2
or, at 25°C,
EI-E20 = QJl8,llog X I
n X2
Thus for X1/X2 = 999 (i.e., Xl = 0.999, X2 = 0.001), the difference in
standard potentials must be 0.355/n V.
In practice the situation is usually more complicated. Processes
are rarely strictly nernstian when large currents are flowing, and if the
interfering couple has very slow electrode kinetics, the potential
separation can be considerably smaller than predicted from eq (7.6). If
one or both of the couples involves deposition of a metal on the cathode,
the activity of the plated metal initially may be proportional to the
fraction reduced, but as a monolayer of atoms is formed on the cathode,
the activity approaches unity, i.e., the electrode surface approaches the
376 Electrolysis
property of the pure plated metal. In the limit of aR = 1, the electrode
potential is
E =Eo +RI.ln Co
nF
so that the electrode potential for small Co may be rather more negative
than expected from eq (7.4a).
Despite these complications, the simple model is useful in
predicting electroseparation efficiencies when both couples have similar
properties. Thus if both couples involve the deposition of a metal and
they have similar rates, eq (7.6) gives a semiquantitative guide to the
required difference in standard potentials.
7.2 ANALYTICAL APPLICATIONS OF ELECTROLYSIS
The electroanalytical techniques discussed in Chapters 1, 3, and 4

may be divided into two general classes: (1) titration methods where the
endpoint is detected electrochemically, e.g., potentiometric,
conductometric, and amperometric titrations; and (2) methods in which
little or no chemical transformation occurs and the electrochemical
measurement gives a direct measure of concentration, e.g., direct
potentiometry, polarography, chronopotentiometry, etc. Most analytical
methods based on electrolysis are similar to titration methods in that the
analyte undergoes a stoichiometric chemical transformation, but here
the role of electrochemistry is to affect the transformation. The
equivalence point of the transformation mayor may not be determined
electrochemically. In some methods, the reaction is judged to be
complete when the electrolysis current falls to zero, while other methods
use an independent endpoint detection method (visual,
spectrophotometric, potentiometric, etc.).
We will discuss four general methods. Electrograuimetry involves
the controlled potential reduction of a solution of a metal salt which is
terminated when the current falls to near zero; the analysis is based on
the increase in weight of the metal-coated cathode. Thus electrolysis is
used solely as a means of chemical transformation. Controlled potential
coulometry also detects the endpoint by the decay of current to zero, but
here the analysis is based on the net charge transferred. Constant
current coulometry detects the endpoint by some method independent of
the electrolysis process and bases the analysis on the net charge
transferred. Finally, a technique known as stripping uoltammetry uses
electrolysis to deposit the analyte on an electrode surface (or in a
mercury drop) and then determines the surface (or amalgam)
composition by a voltammetric technique.
§7.2 Analytical Applications of Electrolysis 377
Electrogravimetry
Although the first electroplating experiments were reported by
Cruikshank in 1801, it was not unti11864 that Wolcott Gibbs adapted the
techniques of electroplating to quantitative chemical analysis (3).
Nonetheless, electrogravimetry is the oldest of electroanalytical
techniques, having become well established by the end of the nineteenth
century. In electrogravimetric analysis, a solution of a metal salt is
exhaustively electrolyzed, plating the metal onto a previously weighed
cathode. The increase in weight of the cathode affords a direct
determination of the metal content (4,D2).
Oliver Wolcott Gibbs (1822-1908) taught at City College of New York from
1849 until his appointment as Rumford Professor at Harvard in 1863.
The following decade saw many contributions to the analytical chemistry
of the platinum metals as well as the development of electrogravimetry.
A curricular change at Harvard in 1871 effectively ended Gibbs' career
as a research scientist.
As an example of the technique, consider the electrogravimetric

determination of copper. An acidic solution of a copper salt is
electrolyzed using platinum electrodes. Metallic copper plates out on the
cathode and we suppose that oxygen is produced at the anode. Thus the
cell reaction is
t
Cu2+(aq) + H20 ~ Cu(s) + 02(g) + 2 H+(aq)
The reversible potential of the cell is given by the Nernst equation
E = E" -RT..l n [H+]2 fl5O;
2F [Cu2f.]
where EO = -0.889 V. When [H+] = 1 M, P(02) = 0.2 bar and [Cu2+] = 0.1 M,
E = -0.91 V. For every tenfold reduction in the copper concentration, the
equilibrium cell potential becomes more negative by about 30 mY. Thus
when the Cu2+ concentration is reduced to 10-4 M, E = -1.00 V. Because
the evolution of 02 is inherently slow, an activation overpotential of about
0.4 V is required to give reasonable current; iR drop in the electrolyte
solution is typically about 0.1 V so that an applied potential of (-1.0 - 0.4-
0.1) = -1.5 V is required to deposit the copper. With this applied potential,
we might expect hydrogen evolution since the standard potential for the
decomposition of water is -1.23 V. Some hydrogen evolution may occur
early in the experiment, but the exchange current density for reduction
of H+ is about 200 times slower on a copper surface than on platinum.
Thus hydrogen evolution is expected to be negligible once the cathode
surface is coated with copper.
378 Electrolysis
If the electrodeposition is carried out at constant cell potential (say -

1.5 V), the cathode potential will become more negative late in the
experiment when the current density decreases. This effect results from
the decrease in magnitude of the solution iR drop and anode
overpotential with decreasing current. Thus some H+ may then be
reduced. If copper is being plated out from a solution containing other
reducible metal ions, the more negative cathode potential may lead to
plating of other metals, such as lead. A positive gravimetric error would
result. For this reason, electrogravimetry is normally carried out in a
three-electrode cell (see §4.3) with the cathode potential controlled such
that only the metal of interest is plated out.
The nature of the metal deposit on the cathode in electrodeposition
turns out to be critically dependent on the identity of the metal species in
solution. It has been found that addition of complexing agents such as
tartrate ion (for Cu2+) or cyanide ion (for Ag+) results in dense, lustrous
deposits, relatively free of inclusions, which adhere well to the platinum
cathode.
Hydrogen evolution at the cathode is generally undesirable since
hydrogen atoms tend to be included in the deposit, leading directly to a
gravimetric error, but more important, the deposit is often flaky or
spongy and thus more prone to solvent or electrolyte inclusions or to loss
of metal particles during washing. Hydrogen reduction can be
suppressed either by buffering the solution to pH 5-6 or by controlling the
cathode potential.
If no more easily oxidizable species is present, water will be
oxidized at the anode. In solutions containing halide ions, halogens will
be produced at the anode. Complexing agents, added to improve the
characteristics of the cathode deposit, are often oxidizable and may also
contribute to the anode process. To one degree or another, all these
processes are undesirable, either because of the production of large
amounts of corrosive gases or the consumption of complexing agents or
because the anode potential must be inconveniently high. Furthermore,
if the anode products are reduced at the cathode, the current efficiency of
the electrolysis is reduced. Thus it is common practice in
electrogravimetry to add a species (called an anodic depolarizer) which
is easily oxidized to an innocuous product. A common choice is
hydrazine, added as the hydrochloride salt, which is oxidized to nitrogen
gas:
N2H5+ ~ N2(g) + 5 H+ + 4 e-
Hydroxylamine hydrochloride is also used, but the anode products then
are a mixture of nitrous oxide, nitrous acid, and nitrate ion, some or all
of which may be reducible at the cathode.
Since electrogravimetry depends only on the completeness of
electrolysis and on the increase in cathode weight for its success,
current efficiencies of less than 100% can be tolerated. Cathode side
§ 7.2 Analytical Applications of Electrolysis 379
reactions have no effect on the results as long as they do not result in

inclusions in the deposit or cause loss of metal on washing. Thus as
long as the anode process does not produce a reducible metal, the two
electrodes need not be isolated; indeed, the resistance of a membrane
separating the anolyte from the catholyte would decrease the current
and increase the analysis time. Cells for electrogravimetry then can be
designed for maximum current with large electrodes and efficient
stirring. A typical arrangement is shown in Figure 7.2.
salt
bridge
Figure 7.2 Electrolysis cell

stirring
for electrogravimetric deter- bar
mination of metals.
Electrogravimetry has been used most frequently in the analysis of

copper-based alloys. By careful control of the cathode potential, the pH
and concentration of complexing agents, Cu, Bi, Pb, and Sn can be
successively plated from the same solution. For example, electrolysis at
-0.30 V (us. s.c.e.) of a solution, buffered to pH 5 and containing 0.25 M
tartrate, deposits copper. Further reduction of the same solution at -0.40
V deposits bismuth, and at -0.60 V, lead. If the solution is then acidified
with HCl and electrolysis continued at -0.65 V, tin is deposited. If the
cathode is removed, washed, dried, and weighed at each step, the
amount of each of the four metals in the original sample is determined.
Like any gravimetric method, electro gravimetry works well only
when accurately weighable amounts of the analyte are present in the
sample. The techniques lacks the sensitivity required for trace metal
analysis, but it may still be useful as the first step in the analysis of an
alloy. After removal of the bulk constituents by electrolysis, the trace
constituents can be analyzed by more sensitive electroanalytical
methods.
380 Electrolysis
Constant Potential Coulometric Analysis

Electrogravimetric analysis uses the gain in weight of the cathode
as the analytical sensor in determining the amount of metal ion in a
solution. While the method works well for copper and certain other
metals, it is limited to the analysis of metals which can be plated out and
weighed. An alternative approach is to base the analysis on Faraday's
laws of electrolysis, i.e., to measure the charge passed through the
electrolysis cell. Thus exhaustive electrolysis of no moles of species 0,
O+n e- ~ R
should consume
Q = nonF coulombs
If Q is measured and n is known, no can be determined. The major
limitation on this strategy is that the current efficiency must be very
near 100%. That is, all the current passed through the cell must go to
the reduction of 0; no electrode side reactions can be tolerated. Although
the basis for coulometric analysis has been around since 1833, it was not
until 1938 that Szebelledy and Somogyi (5) suggested coulometry as a
general analytical technique. Extensive work on perfection of analytical
methods really began only after World War II.
The requirement of 100% current efficiency generally means that
the anode and cathode must be separated. If the product of the cathode
reaction is a solution species, it must not be reoxidized at the anode and
anode products must not be reduced at the cathode. Thus, with a few
exceptions, coulometric methods employ cells in which the anode and
cathode are separated by a salt bridge, membrane, diaphragm or some
such barrier to prevent mixing of anolyte and catholyte. A typical cell
design is shown in Figure 7.3.
Until relatively recently, an electrochemical coulometer was used
for the measurement of total charge passed in an electrolysis
experiment. A coulometer is simply an electrolysis cell designed to
operate at 100% current efficiency to give a weighable cathode deposit, a
titratable solution product, or a gas, the volume of which can be
measured using a gas buret. The ultimate device was (and for very
precise work, still is) the silver coulometer. In this device, silver is
deposited on the cathode which can be removed, washed, dried, and
weighed. Other practical coulometers employed the electrolysis of water,
2 H20 ~ 2 H2(g) + 02(g)
to produce a mixture of hydrogen and oxygen gases, or the electrolysis of
an aqueous acidic solution of hydrazine,
N2li5+ ~ H+ + 2 H2(g) + N2(g)
to produce a mixture of hydrogen and nitrogen. In both the H2/02 and
reference
electrode
Figure 7.3 Cell for

controlled potential
coulometric analysis.
the H2/N2 coulometers, 0.75 mole of gas is produced per mole of

electrons, so that, at 298 K, 1 bar pressure, 0.193 mL of gas is obtained per
coulomb of charge passed through the coulometer. If gas volumes can
be measured with a precision of ±0.01 mL, a measurement of charge
with ±D. 1% precision would require passage of at least 50 coulombs.
Nowadays, electronic instrumentation allows the measurement of
much smaller charges with comparable precision and considerably less
bother. There are two common approaches to the problem. The most
straightforward arrangement is to record a current-time curve using a
strip-chart recorder. Graphical integration of the current-time curve
then gives the charge passed through the cell. This procedure works
well but is tedious in routine applications. If such measurements are to
be made frequently, it is more convenient to use an electronic coulometer
built around a voltage integrator circuit similar to that used to generate
a voltage ramp (see Figure 4.6). In the coulometer circuit shown in
Figure 7.4, the cell current is divided so that
icell = i 1 + i2
Because of the capacitive feedback loop, the potential at the inverting
input of the operational amplifier is a virtual ground, 0 V. Thus the
potentials across resistors Rl and R2 must be equal:
ilRl = i~2
With this relationship, we obtain
. . Rl + R2
lcell = l2-'--"'='-~
Rl
382 Electrolysis
cell
power
r
supply
Figure 7.4 Electronic coulometer based on an

operational amplifier integrator circuit.
Since the input impedance of the operational amplifier is very large, the
current i2 must produce a charge Q2 on the capacitor C:
. dQ2
L2=di
But the potential at the output is proportional to this charge
Q2 = C <Pout
Thus
Rearranging, we have
and integration gives
or
Qcell = - R 1 + R2 C<Prut
Rl
By selecting appropriate values of RI, R2, and C, charges ranging from
less than 1 IlC up to about 1 C are easily measured. The method requires
accurately known resistances R I and R2 and capacitance C as well as
an accurate potential measurement.
§ 7.2 Analytical Applications of Electrolysis 383
Controlled potential coulometric analysis (6,7) can be used in much

the same way as electrogravimetry to determine reducible metal ions.
Controlled potential coulometry at a mercury cathode is a particularly
attractive variation. The advantages are that competition from hydrogen
reduction is negligible (because of the large overpotential) and that most
metals dissolve in the mercury to form amalgams.
The coulometric method has considerably greater scope than
electrogravimetry. In particular, it is not limited to reduction of metal
ions but can be used for any electrode process which can be operated at
near 100% current efficiency. For example, controlled potential
coulometry has been used to determine trichloroacetic acid. The
reduction
Cl3CCOOH + H+ + 2 e· ~ Cl2CHCOOH + CI-
is sufficiently well separated from the reduction of di- and
monochloroacetic acid that a mixture of these acids can be accurately
analyzed, a feat impossible to contemplate with a pH titration.
Coulometry is not limited to reductions-the analytical reaction can
just as well occur at the anode. Thus controlled potential coulometry at
a silver anode is a convenient method for the determination of halide
ions or other species forming insoluble silver salts. In effect, the Ag+
generated coulometrically acts as a titrant for halides just as in the
classical Volhard titration and its variants.
Controlled potential coulometry is commonly used in mechanistic
electrochemical studies to determine the number of electrons involved in
an oxidation or reduction. Determination of n from a measured
voltammetric diffusion current depends on a reasonably accurate value
for the diffusion coefficient. When n might be either 1 or 2, D need be
known only to ±10% or so to get a satisfactorily unambiguous result.
When n is larger or when slow following chemical steps lead to other
electro active species, the results from voltammetry may be less
satisfactory. A coulometric measurement gives a value for n which
includes the effects of follow-up chemical reactions but it does not
depend on the diffusion coefficient, electrode area, or other unknowns
(other than the current efficiency),
Example 7.2 An extreme case of voltammetric uncertainty

arose in the polarographic reduction of picric acid. The
diffusion current constant is
ID = 27 ~ mM-1(mg s-1)-213 s -1/6
in 0.1 M aqueous HCI. If we assure that D is about the same as
for the 2+ metal ions of Table 4.1, comparison of the values of In
suggests that n is about 18 (but obviously with a large
uncertainty). Since the overall reduction is expected to yield
2,4,6-triaminophenol,
384 Electrolysis
N02 NH2
o.N-Q-OH + 18 H· + 18 .. - H21'-Q-OH + 6 H,O

N~ N~
n = 18 is a reasonable result. Lingane (8) determined n

coulometrically in 1945 and found n = 17.07. This result was
confirmed by others but Meites (9) later showed that the value
of n depends on the initial concentration of picric acid and on
the HCI concentration. When the initial concentration was
less than 0.1 mM and/or the HCI concentration is 3 M or more,
coulometry gave the expected value, n = 18. For higher picric
acid concentrations and/or lower HCI concentrations, smaller
values of n were obtained. Apparently, side reactions compete
with protonation of radical intermediates. The dependence on
total picric acid concentration suggests that a radical coupling
reaction is important and Lingane suggested that the eventual
product is a substituted hydrazobenzene,
N02 H2N OH NH2
2 02N-Q-0H+34H++34e- - <:>-NH-NHQOH
N02 H2N + 24 H20 NH2
but other reactions must also be important since Meites found

values of n less than 17 under some conditions.
Constant Current Coulometry

There are a number of coulometric methods which employ an
electrolysis cell operated at constant current rather than constant
working electrode potential. One might think that such an arrangement
could not work. Surely, the analyte would be quickly exhausted near the
electrode and the potential would swing to begin consuming the solvent
or supporting electrolyte. The result would be analogous to
chronopotentiometry (§4.3), where a constant current is passed through
an unstirred solution. Perhaps (we might reason) the method could be
made to work if the current were so small that mass transport could
keep up with consumption at the electrode, even when the analyte is
nearly exhausted in the bulk solution. However, to make this approach
work, the current would have to be small and the time required for
electrolysis would then be impossibly long.
There is a trick involved. Constant current coulometry invariably
involves another electrode couple, the product of which reacts rapidly
and stoichiometrically with the analyte. For example, suppose that we
want to determine Fe (II) coulometrically. A large excess of Ce(lII) is

added to the cell and a constant current is passed. Initially, the Fe(l!)
concentration is high enough that the oxidation
Fe 2+ ~ Fe3+ + e-
is the major contributor to the current. However, as the Fe(I1)
concentration begins to drop, the potential swings positive and Ce(I1!)
begins to be oxidized:
Ce3+ ~ Ce 4+ + e-
However, Ce(IV) reacts rapidly with Fe(l!),
Ce4+ + Fe2+ ~ Ce3+ + Fe3 +
so the effect is as if Fe(II) were oxidized directly. As the Fe(l!)
concentration drops, more and more of the current is carried by the
Ce(I1I) oxidation process. The net effect, nonetheless, is the oxidation of
Fe(l!), at least up to the point that the Fe(l!) is entirely consumed. After
the equivalence point, current continues to flow, of course, and the
coulometer continues to run, clocking up the charge passed. Clearly,
some means is required to note the time at which the last Fe(I1) was
consumed.
A moment's reflection will show that the situation is exactly
analogous to a titration in homogeneous solution of Fe(I1) with Ce(IV)
solution added from a buret. What is needed is a means of detecting the
endpoint of the titration. In this example, the obvious solution is to add
another pair of electrodes-an indicator electrode and a reference
electrode-to the anode compartment of the electrolysis cell. If the
potential of this pair is monitored, it should show a sharp positive swing
when the last of the Fe(l!) is consumed and the Ce(IV) concentration
starts to increase.
A constant current coulometry experiment is often referred to as a
coulometric titration to emphasize the similarity to ordinary volumetric
analysis. There is another analogy which should be pointed out. The
overall process can be written (for a net reduction)
O+n e- ~ R
R+A ~O+B
where A is the analyte and R is the internally generated titrant. This

electrode process is identical to the EC' (catalytic) process discussed in
§5.8. In the voltammetric context, we supposed that A was not reduced
at the electrode because of slow electron transfer kinetics. Here the rate-
limiting step is mass transport, but in either case, coupling of the OIR
and AlB redox couples results in a catalytic current enhancement.
386 Electrolysis
cell
power
supply
Figure 7.5 A simple circuit for
constant current electrolysis.
Instrumentation for constant current coulometry can be quite

simple. A satisfactorily constant current can be produced by connecting
a high-voltage d.c. power supply across the cell in series with a
resistance R as shown in Figure 7.5. The current through the cell is
i= A
R + Reell
If R is much larger than Rceu, the current will be nearly independent of
the cell resistance and thus nearly constant. For example, if A = 500 V
and R =500 kO, then i =1 mA when the cell is out of the circuit. When
Ecell =-1 V and Rcell = 1 ka, i =0.996 mA, only 0.4% different. Variations
of Ecell and Rcell during electrolysis will cause similarly small changes
in i.
The charge passed through the cell generally is determined by
measuring the time between the application of current and the
appearence of the titration endpoint, Q =it. If 1% analytical accuracy is
acceptable, a current variation on the order of a few tenths of a percent is
tolerable. If more constant current is required, an operational
amplified-based constant current source (a galvanostat) is the next step
up in sophistication. A galvanostat circuit is shown in Figure 4.6, but
this circuit is not well adapted to passage of large currents since the
reference source supplies the current. A better high-current circuit is
shown in Figure 7.6. Here the current is supplied entirely by the
operational amplifier. As usual with operational amplifier feedback
circuits, the potential at the inverting input is at virtual ground (0 V).
Thus the potential across the resistor R must be equal to the battery
potential A, and the current through R must be i =AIR. But since the
input impedance of the operational amplifier is high, virtually all the
current through R flows through the cell, independent of the cell's
internal resistance and potential. Thus provided that Rcell is much less
than the input impedance of the operational amplifier and i(Rcell + R) is
within the range of the operational amplifier output voltage, the circuit
will supply a constant current. With the addition of a reference
electrode, the working electrode potential can be monitored if desired.
+
Reference
Monitor
Figure 7.6 Constant Current

current source based Output
on an operational
amplifier.
Once we realize that constant current coulometry is equivalent to a

titration, the possibilities open up in all directions. We have mentioned
the electrochemical generation of Ce(IV), but many other oxidants can
be produced. Thus Mn(IIl) or Ag(ll) can be obtained by oxidizing
Mn(ll) or AgCl) solutions. Chlorine, bromine, or iodine can be generated
from CI-, Br-, or 1- solutions. The halogens are excellent oxidizing
agents, but, because of their volatility, are relatively difficult to use in
volumetric analysis. Similarly, reductants such as Cr(ll) or Ti(Ill) can
be produced from CrOll) or Ti(IV), respectively. Coulometry is not even
limited to generation of oxidants or reductants. Hydroxide or hydrogen
ions can be produced by reducing or oxidizing water at a Pt electrode, so
that acid-base titrations are possible. Precipitation titrations using Ag(l)
or Hg(l) can be done coulometrically as well.
Coulometric titrations have some significant advantages over
volumetric methods. We, have mentioned the use of titrants which
would be difficult to use under ordinary circumstances, but there are a
number of other considerations. Since the titrant is generated internally
and only in the required amounts, there is some savings in expensive
reagents; furthermore, no standardization of titrant is necessary.
Coulometric methods are much more easily automated than volumetric
procedures. Since charges on the order of microcoulombs are easily
measured with precision, much smaller samples can be used in a
coulometric titration than in the analogous volumetric method. See a
review by Curran (10) for further details.
Example 7.3 The standard analytical method for the

determination of water in organic solvents employs a titration
using the Karl Fischer reagent-a pyridine/methanol solution
of iodine and sulfur dioxide. The reagent contains the
388 Electrolysis
monomethyl sulfite ion (11), formed by the reaction of methanol
with 802, pyridine acting as a proton acceptor:
CHaOH + 802 + C5HsN ~ CHa0802- + C5H5NH+
This species is oxidized by iodine in the titration reaction,
which requires one mole of water per mole of iodine:
CH30802- + 12 + H20 + 2 C5H5N -7 CH30803- + 2 1- + 2 C5HsNH+
Needless to say, Karl Fischer reagent is moisture sensitive and
is difficult to use in normal volumetric analysis. Coulometric
generation of iodine (or 13-) from KI affords a substantial
improvement in the method. The endpoint is normally
detected potentiometrically; the platinum indicator electrode
responds to the 13-1I- couple.
While potentiometric endpoint detection is probably most common

in coulometric titrations, other electrochemical methods are often
employed. One particularly simple method is called biamperometric
endpoint detection. In this method, a small potential (10-100 mY) is
imposed between two identical platinum electrodes, and the current
through this circuit is measured. If both members of a reversible
couple, 0 and R, are present in solution, a small current (typically less
than 100 1lA) will flow as 0 is reduced at the negative electrode and R is
oxidized at the positive electrode. When either 0 or R is absent, no
current flows. In the example of the Karl Fischer titration, the 13-/1-
couple is reversible, so that current flows when both 13- and 1- are
present. The CH30803-/CH30802- couple, on the other hand, is
irreversible, so that the current is negligibly small up to the endpoint
and increases more or less linearly with time beyond the endpoint. This
approach is applicable when the titrant is added from a buret; the
technique is then called a biamperometric titration. The name recalls
the relation to amperometric titrations discussed in §4.3, and the
presence of two polarized electrodes. The method is applicable whenever
the titrant forms a reversible couple and the analyte an irreversible
couple (or vice versa).!
lOne of the earliest applications of the biamperometric method was in the titration of
iodine solutions with standard sodium thiosulfate solution from a buret. The 82032-
/S406 2 - couple is irreversible, so that a steadily decreasing current is observed up to
the endpoint, with essentially zero current beyond. Such a titration was called a
dead-stop titration, but this name is inappropriate when the titrant furnishes the
reversible couple.
Stripping Voltammetry
When a solution containing metal ions such as Pb 2 +, Cu2 +, Cd2 +,

Zn 2 +, etc. is subjected to polarographic analysis, the reduced metal
dissolves in the mercury drop as an amalgam. The process is reversible
and if the potential is made anodic, the metal redissolves in the aqueous
solution. In anodic stripping voltammetry (12-14), a mercury electrode
is made the cathode of an electrolysis cell. Current is passed for a few
minutes through a stirred solution and metals are deposited in the
electrode. The potential is then scanned and the anodic current
measured, usually using the differential pulse technique. As the
potential reaches the dissolution potential of each metal present, a peak
is observed in the current-potential curve. The technique achieves very
high sensitivity because, in the electrolysis step, metal ions are
concentrated in the mercury electrode. In the most common form of
anodic stripping voltammetry, a hanging mercury drop electrode is
used. If the solution volume is 25 mL and the electrode volume is 5 ilL,
exhaustive electrolysis would increase the concentrations by a factor of
5000. In practice, electrolysis is far from exhaustive (it would take too
long), but a large concentration enhancement is achieved. Since
ordinary applications of differential pulse polarography (see Sections 4.5
and 4.8) have sensitivities down to less than 1 IlM, a concentration
enhancement on the order of lOS pushs the sensitivity limit down into
the nanomolar range. Clearly the concentration enhancement could be
even greater if the volume of the mercury electrode were smaller and
electrolysis would be both faster and more complete if the electrode area
were larger or if transport were more efficient. A rotating disk
electrode, in which in which a thin film of mercury is deposited on a
platinum or glassy carbon base, achieves this goal, pushing the
sensitivity of the technique down to concentrations of 10- 11 M!
Since controlled potential electrolysis in the concentration step is
not exhaustive, and differential pulse peak heights depend on a number
of experimental parameters, anodic stripping voltammograms require
calibration and careful reproduction of calibration conditions in the
analysis. With attention to detail, anodic stripping voltammetry can be
the method of choice for trace metal analysis. Sensitivity is usually
competitive with atomic absorption spectrophotometry, and, because
several metals can be determined in one differential puse voltammetric
scan, the analysis can be quicker and easier.
Even with care, however, anodic stripping voltammetry is not
without some pitfalls. Unanticipated electrode processes sometimes
occur during the electrolysis phase which lead to apparently anomalous
peaks in the voltammetric scan. Ostapczuk and Kublik (15) traced one
such peak, observed when solutions containing AgNOs were subjected to
anodic stripping analysis, to hydroxide ion formed in the Ag+-catalyzed
reduction of nitrate ion:
390 Electrolysis
NOa-+H20+2e- ~ N02-+20H-
The local increase in pH led to the precipitation of various metal
hydroxides on the electrode. On the anodic scan, these gave rise to a
mercury oxidation peak:
HgO) + M(OH)2 ~ Hg(OH)2 + M2+ + 2 e-
Another source of trouble is the formation of intermetallic
compounds in the mercury phase. Thus reduction of solutions
containing gold and cadmium or zinc leads to the intermetallic
compounds AuCd, AuaCd, and AuZn. These species oxidize at a
different potential than cadmium or zinc amalgam, so that the normal
cadmium and zinc peaks in thevoltammetric scan are smaller than
expected and extra peaks are observed due to oxidation of the
intermetallic compounds.
The stripping voltammetry strategy has been most commonly used
to concentrate metals in a mercury drop or thin-layer electrode, followed
by anodic stripping. The same strategy can often be applied in reverse,
depositing the analyte on the electrode surface in an oxidation step; the
analysis is then based on a cathodic stripping voltammetric scan.
Example 7.4 Shimizu and Osteryoung (16) used a rotating

silver disk electrode to determine traces of sulfide by a cathodic
stripping technique. Anodic current was passed, holding the
electrode at -0.4 V (vs. s.c.e.) for 100-1000 s (co = 260 rad s-1). A
cathodic scan using differential pulse voltammtry then gave a
well-defined peak corresponding to the reduction of Ag2S. The
peak current was linear in the initial sulfide concentration
over the range 0.01 - 10 J,J.M.
7.3 ELECTROSYNTHESIS
There are enormous numbers of applications of electrolysis

methods to chemical synthesis, both in inorganic and organic
chemistry. Unfortunately, academic synthetic chemists have tended to
ignore electrosynthetic methods, and the field has been developed mostly
by specialists. This is partly due to a lack of equipment for large-scale
electrosynthesis in most synthetic laboratories, but it is more often a
result of unfamiliarity with electrochemistry on the part of synthetic
chemists. Industrial applications have been relatively more important
and it is no accident that research in electrosynthesis has been more
common in industrial laboratories than in academia. In this section, we
will discuss four representative organic electro synthetic methods; in
§7.3 Electrosynthesis 391
§7.4, we will consider some specific examples of industrial
electrosynthetic processes. An extensive secondary literature in
electro synthetic methods has developed and the reader is referred to one
of the general references (E1-E8) for further details.
Reductive Elimination Reactions

Reduction of an organic molecule with a good leaving group
generally leads to a neutral radical:
+e +e· +BH
RX - R· - R" - RH
-X" -B-
Leaving groups include the halide ions (F-, CI-, Br, 1-) and pseudohalides
(CN-, SCN-, etc.) as well as ORo, SR-, NR2-, RS02-, etc. When R is an alkyl
group, the ionization step may be concerted with electron transfer, i.e.,
the CoX bond is somewhat stretched and polarized before electron
transfer and ionization is quickly completed after electron transfer. In
these cases, the neutral radical is formed at the electrode surface and
further reduction to the carbanion nearly always occurs immediately.
The strongly basic carbanion usually abstracts a proton from any
available source (solvent or supporting electrolyte). Electrolysis thus
provides a means for replacement of halogens or other groups by
hydrogen. Yields are usually quite good; for example, reduction of
benzyl bromide in acetonitrile (Rg cathode, -2.1 V vs. Ag/AgCI04, Et4NBr
electrolyte) gives an essentially quantitative yield of toluene.
When the initially formed radical anion is resonance-stabilized, the
ionization step may be slow enough that the radical anion diffuses away
from the electrode. Radical dimerization may then result. For example,
reduction of 4-nitrobenzyl bromide in an aprotic solvent gives mostly 4,4'-
dinitrobibenzyl (17):
2 02N-o-~ CH2Br ~-_ 02N-o~ CH2""CH2o-~ N02

+ 2
- 2Br
Geminal dihalides, RCRX2, usually reduce in two distinct steps to
give RCR2X and then RCR3, presumably via the same mechanism as
with alkyl halides. Vicinal dihalides reduce to olefins:
-2X-
The reduction is stereospecific. Thus, for example, reduction of d,l-3,4-
dibromobutane in DMF (Rg cathode, -1.1 V vs. s.c.e., BU4NBF 4
electrolyte) gives 100% cis-2-butene. Reduction of the meso isomer under
the same conditions gives trans-2-butene (18).
With 1,3- and 1,4-dihalides, cyclization competes with proton
abstraction. The detailed mechanism is unclear. It may be that two-
electron reduction of one end of the molecule leads to a carbanion,
392 Electrolysis
which then displaces X- from the other end, forming a ring. However,
there is some evidence to suggest that C-C bond formation is concerted
with ionization of the halide ions and electron transfer. The product
distribution depends on the reduction potential. Thus, for example,
reduction of 1,4-dibromobutane at -1.75 V (us. s.c.e.) in DMF solution
gives 26% cyclobutane and 74% butane; under these conditions, step-wise
reduction of the two C-Br bonds seems to occur with protonation
competing successfully with cyclization (19). However, if the reduction is
carried out at -2.3 V, the products are 90% cyclobutane and only 10%
butane, suggesting that the concerted mechanism may be operative at
the more negative potential. Electrochemical reductive elimination
processes are particularly useful for formation of strained ring systems.
Thus reduction of 1,3-dibromo-2,2-di(bromomethyl)propane gives first
1,1-di(bromomethyDcyclopropane and then, on further reduction,
spiropentane (19):
CH2Br CH B
I +2e- 2 r +2e-
BrCHr-C-CH2Br - ~ - ~
I - 2 Br- (../\ - 2 Br- V""-J
CH 2Br CH2Br
Reduction of Aromatic Compounds

Most aromatic compounds, and especially those with electron-
withdrawing substituents, form stable radical anions in polar aprotic
media. However, in the presence of a proton source, the radical anion is
protonated to form a neutral radical which is usually more easily
reduced than the substrate. Thus reductions in protic media usually
involve at least two electrons and two protons, In now-classical work,
Wawzonek and Laitinen (20) showed that anthracene is reduced in
aqueous dioxane to 9,1O-dihydroanthracene via the ECEC mechanism:
HH HH
~~c¢o~
~+H+ ~.
I~~~
~ +H+ vyv
H HH
If the potential is controlled, reduction stops at this point since the
isolated benzene rings are much more difficult to reduce than
anthracene. Other aromatic hydrocarbons behave similarly; thus,
naphthalene is reduced to 1,4-dihydronaphthalene and phenanthrene to
9,10-dihydrophenanthrene. Pyrene can be reduced to 4,5-dihydropyrene
and, at more negative potentials, to 4,5,9,10-tetrahydropyrene:
When an electron-withdrawing substituent is present, the radical

anion charge is largely concentrated on the substituent so that it is
usually the site of protonation. Thus, for example, in aqueous media
above pH 5, nitrobenzene is reduced by four electrons overall. The first
steps form nitrosobenzene
+ e- + H+. + e-
Ph-N02 - - [PhN02r - PhN(O)OH - - PhN=O
-OH-
but this species is more easily reduced than nitrobenzene so that further
reduction occurs rapidly, yielding phenyl hydroxylamine in weakly
acidic to weakly alkaline solutions:
+e + H+ . +e-
PhN=O - - [PhN=Or - PhNOH -- PhNHOH
~ + H+
[ Ph_N_%_Ph]Z. -OH-
:..!: Ph-N=N-Ph
o
I I
0
When the solution is alkaline, protonation of the nitrosobenzene radical
anion apparently is slow and this species dimerizes to give, after
protonation and loss ofOH-, azoxybenzene, which can be isolated in high
yield. Further reduction at low pH yields hydrazobenzene which
rearranges to benzidine in strong acid:
+4e- H+ ~
Ph-N=~-Ph ~ Ph-NH-NH-Ph - H2N~NH2
-H20
At higher temperatures or in more strongly acidic media, PhNHOH
rearranges to p-aminophenol. At low pH, and at a more negative
potential, it can be reduced to aniline:
HO -0-
~ ~
H+
NH2 - - PhNHOH
high
+2e-
- - PhNH a+
+ 3 H+
temperature
In general, the reduction of nitro compounds thus can lead to a wide
variety of products, and by judicious control of solution polarity, acidity,
and temperature, high specificities can be achieved.
The reduction of carbonyl compounds proceeds analogously. For
example, benzophenone is reduced in neutral or alkaline media to
benzhydrol
394 Electrolysis
- + e-
At lower pH, benzophenone is protonated; the one-electron reduction

product is reduced only at more negative potentials. Thus under
controlled potential electrolysis, the neutral radical dimerizes to
benzpinacol which undergoes acid-catalyzed rearrangement to the
pinacolone:
- HOI
OH
oII
Ph2
Arylalkyl ketones and aryl aldehydes behave similarly to diaryl ketones,

C-9Ph2 - PhC-CPh3
affording good yields of the pinacol (or pinacolone) on reduction in

solutions of moderate acidity.
Reactions of Olefin Radical Cations

When a unsaturated system is oxidized, the initially formed product
is usually the radical cation when then may undergo proton loss,
nucleophilic attack, rearrangement, or dimerization. Oxidation of an
olefin in the presence of a nucleophile, for example, might lead to
addition or dimerization:
,
I
C=C -
\
1 - e- [,
I
I] +
c=c\
+ Nu-
-
,
. C-C-Nu
I , \
1
- -e
+ Nu-
, 1
NupC-C-Nu
I \
\ I I I
Nu-C-C-C-C-Nu
I I I \
or to substitution, via proton loss, oxidation and nucleophilic addition to
the allyl cation:
I I
-C-C=C
1-
-e- , I -e- [,
. c-c=c
I 1- + + Nu- I I
c-=-=-c-=c
I] - -c-c=c
1
I \ - H+ I \ I \ Nu \
Examples of all these routes can be found. Thus electrolysis of
cyclohexene in acetic acid with Et4NOTs supporting electrolyte led to
55% 3-acetoxycyclohexene and 12% trans-1,2-diacetoxycyclohexane (21).
Oxidation of styrene in methanol solution containing sodium methoxide
and NaCI04 supporting electrolyte gave 64% 1,4-dimethoxy-1,4-
diphenylbutane (22).
Oxidation of conjugated olefins leads primarily to 1,2- or 1,4-addition
via allylic radical/cation intermediates:
c-e
I I .
C
.... C~ ........ C~ ... -
::>:
I I
,]+
I ,-e'
-C-C-C=C -
I . '
Nu
I I [,
-c-c--=c:.:.;c
I
Nu
I I
\ -
+Nu' I I I I
Nu
I I
-C-c=C-C- + -C-C-C=C
I
Nu
I I
Nu
I I
I
"-
Nu
Thus, for example, cyclopentadiene in methanol (with NaOMe) or acetic

acid gives ca 50% 1,4-addition (1:1 mixture of cis and trans isomers) and
6% 1,2-addition (primarily trans) (23).
Oxidation of aromatic hydrocarbons in the presence of a
nucleophile often leads to nuclear substitution:
ArH -
-e.
[ArH]
+ + Nu-
-
~H
A\. -
-e- [ "H
A\
1+ --H +
ArNu
Nu Nu
but the substitution product is often more easily oxidized than the
starting hydrocarbon, leading to reduced yields. Better results are
obtained when the product is less easily oxidized. Thus, for example,
oxidation of 1,4-dimethoxybenzene in methanol with KOH leads to a 75%
yield of quinone bisacetal (24):
CHgO OCHg
OCHg
CHgOH/KOH
•
Q
CHgO OCHs
Kolbe Hydrocarbon Synthesis

The first important organic electrosynthetic reaction was
investigated by Kolbe in 1843 (25). The reaction consists of the oxidation
at a Pt anode of partially neutralized carboxylic acids in DMF or
methanol solution. One-electron oxidation presumably gives an acyloxy
,
radical intermediate, which rapidly loses C02 and dimerizes to the
hydrocarbon product:
-e-
RCO£ - RC0 2" - R" + C02
R-R
Yields vary from 50 to 90%, quite a bit higher than might be expected
considering the reactions open to an alkyl radical; indeed, there is good
396 Electrolysis
reason to believe that the radicals are adsorbed on the electrode surface
and protected to some extent from the solvent (from which hydrogen
abstraction might be expected). Three kinds of side reactions compete
with radical dimerization: (1) In cases where radical rearrangement
can occur, a variety of coupling products is obtained. Thus, for example,
a f3;y-unsaturated acid leads to an allylic radical intermediate,
R-CH=CH-CH 2 • R-CH-CH=CH2
and three coupling products are obtained:
R-CH=CH-CH2-CH2-CH=CH-R
CH2=CH -CHR-CHR-CH=CH2
R-CH=CH2-CHR-CH=CH2
(2) When the radical intermediate is secondary or tertiary, or substituted
such that the carbonium ion is stabilized, further oxidation takes place:
-e-
R·-R+
and the eventual products are derived from carbonium ion reactions. (3)
In cases where the acyloxy radical is stabilized by conjugation (aromatic
or a.,f3-unsaturated acids), Kolbe products are not obtained, most likely
because the acyloxy radical survives long enough to escape from the
electrode surface. Thus benzoic acid gives benzene under Kolbe
conditions, presumably through loss of C02 and hydrogen abstraction
from the solvent by the resulting phenyl radical.
Adolf Wilhelm Hermann Kolbe (1818-1884), Professor of Chemistry at

Marburg and Leipzig, made very important contributions in organic
chemistry and to the development of the concept of radicals. However, he
never accepted the idea of structural formulas and wrote some of the most
vitriolic polemics in the history of science attacking KekuIe, van't Hoff,
and Le Bel.
7.4 INDUSTRIAL ELECTROLYSIS PROCESSES
Electrolysis has found many important applications in the

chemical, metallurgical, and metal finishing industries. In this section,
we will examine representative examples from four classes of industrial
processes:
(1) Electroplating and other metal finishing techniques;
(2) Reduction of ores and purification of metals;
(3) Production of inorganic chemicals; and
(4) Electrosynthesis of organic chemicals.
§7.4 Industrial Electrolysis Processes 397
In an industrial process, the object is to produce a chemical which
has an economic value greater than the sum of the costs of raw
materials, energy, labor, and capital-in short, to make a profit. An
inefficient process might be acceptable if the product is produced a few
grams at a time but would be economically ruinous if thousands of tons
are needed. Thus several figures of merit are important in industrial
electrochemistry.
As in ordinary synthetic chemistry, percent yield measures the
conversion of the limiting reagent to the desired product:
m. • Id
-10 yIe =molesmoles
of starting material
. converted to product x 100m. YO
of startmg material consumed
When a mixture of products is obtained, it is sometimes more useful to
express the yield in terms of the product selectivity:
m. 1 ti··t
YO se ec Vly=
moles of desired product X 100m.YO
total moles of all products
In some processes, the overall yield may be high but only a fraction of
starting material is consumed during one pass through the
electrochemical cell. The unreacted starting material must then be
separated from the product and recycled. The percent conversion per
pass is an important consideration since separation and recycling add
to the cost of the process.
In some electrochemical processes, electrode side reactions such as
the production of hydrogen at the cathode or oxygen at the anode may not
consume starting material or cause separation problems. Such side
reactions draw current, however, and thus consititute a nonproductive
use of electrical energy. Thus the current efficiency, cp, eq (7.1), is also an
important figure of merit.
Some industrial electrochemical processes use huge amounts of
electrical energy, so that the energy consumption per ton of product is an
extremely important consideration. If the potential applied to the cell is
E, n Faradays are required to produce one mole of product, and the
current efficiency is cp, then the energy consumption per mole is
Energy consumption =~ J mol-l
cp
If M is the molecular weight (g moP) and we express electrical energy
in kilowatt-hours (1 kWh = 3600 kJ), then the energy consumption per
ton (1 metric ton = 1000 kg) is
Energy consumption = 0.278 nEE. kWh ton-l
cpM
Energy consumption may increase either because of low current
efficiency (electrode side reactions) or because of energy losses in the
electrochemical process (larger applied cell potential). The applied cell
398 Electrolysis
potential includes the difference of the reversible half-cell potentials, Ec

and E a , the cathodic and anodic overpotentials, TIc and Tla, and the iR
drop in the electrodes, leads, and electrolyte solution:
E = Ec - Ea - I TIc I - I Tla I - i(Rsoln + Rcircuit)
Both the overpotentials and the iR drop increase in magnitude with
increasing current. Thus the energy cost per ton of product must
increase with the rate of production. Materials costs may be more or
less independent of the rate, but the costs of labor and capital per ton of
product almost always decrease with increasing rate. Thus the
optimum current density in an electrochemical process is determined by
a trade-off between energy costs on the one hand and labor and capital
costs on the other.
Electroplating
An important application of electrodeposition is the plating of a
layer of metal on an object to improve the appearance or to impart
hardness or corrosion resistance. Metals commonly used in
electroplating include Cr, Ni, Zn, Cd, Cu, Ag, Au, Sn, and Pb. Some
alloys such as brass (Cu/Zn) and bronze (Cu/Sn) also can be
electroplated.
The basic theory of electroplating is extremely simple. For example,
to plate copper on a steel substrate, the object to be plated is made the
cathode in an electrolysis cell where a piece of copper is used as the
anode. There are many subtleties, however, which must be considered
in practice. Thus, in plating copper on steel, for example, we
immediately recognize that the reaction
Cu2 +(aq) + Fe(s) ~ Cu(s) + Fe 2 +(aq)
is spontaneous. To prevent the dissolution of iron from the substrate, the
activity of Cu2 +(aq) must be reduced by the addition of a complexing
agent which coordinates strongly to Cu(Il) but much less so than Fe(II).
In electroplating applications, it is usually desirable to deposit a
layer of uniform thickness. This requirement is not difficult to meet if
the substrate has a simple geometry. If there are holes or recesses,
however, a uniform layer can be quite difficult to achieve. When a
potential is applied across an electrolysis cell, the potential drop is the
sum of several contributions:
(1) the equilibrium anode-solution and solution-cathode potentials (i.e.,
the equilibrium cell potential);
(2) the activation and polarization overpotentials at the anode and
cathode; and
(3) the solution iR drop.
Since the iR potential drop is proportional to the length of the

current path, it will be smallest at that part of the cathode which is
closest to the anode. Since the total cell potential is constant, this means
that the cathodic overpotential will be greatest at that closest part of the
cathode. The current density-and the deposited layer thickness-will
be greatest, therefore, at exposed parts of the cathode.
Electroplaters refer to the ability of a plating bath to deposit metal at
hard to reach spots in terms of the throwing power of the bath. l The
throwing power of a plating bath can be controlled (to some extent) in
three ways: (1) by adding a large concentration of an inert electrolyte,
the solution resistance is lowered and differences in iR drop to various
points on the cathode surface are reduced; (2) by operating the bath
under conditions where H2(g) evolution occurs at exposed points where
the overpotential is large, the iR drop is increased locally, thus reducing
the overpotential and slowing the rate of plating at these exposed points;
(3) by adding a complexing agent, the electrode process is converted to
MLxn+ + n e- ~ M+xL
It is generally found that complexing agents decrease the slope of a Tafel
plot (log i vs. ,,), so that current variations with overpotential are
smaller. In addition, throwing power is usually found to be a function of
temperature through the temperature dependences of the various rate
processes involved. Plating baths with good throwing power usually
have optimized all these parameters, most commonly through trial-and-
error investigations.
The nature of an electroplated metal deposit can be modified by the
addition of organic additives to perform one or more of the following
functions: (1) Wetting agents facilitate the release of bubbles of H2(g)
from the surface, preventing the occlusion of hydrogen in the deposit
and enhancing current density control by hydrogen evolution. (2)
Levelers are preferentially adsorbed at surface dislocations and sharp
corners and thus inhibit current flow at the points at which the current
density would otherwise be highest. (3) Structure modifiers and
brighteners change the nature of the deposited layer, perhaps changing
the crystal growth pattern or reducing the crystallite size in the
deposited layer.
In most commercial electroplating operations, the current
efficiency (i.e., the fraction of the current which produces the desired
metal deposit) is high, 90% or better. The most notable exception is in
chromium plating, where chromium is added to the bath as Cr03 and
the plating process is based on the six-electron reduction of Cr(VI). Here
Cr(III) is an unwanted by-product, unreducible under normal
1 The concept of throwing power can be made quantitative by defining a standard

test cell geometry; see Pletcher (G6) for further details.
400 Electrolysis
experimental conditions. 1 The detailed electrode process is not well

understood, but there are apparently several current-consuming side
reactions.
Organic polymer coatings can be applied to metal surfaces by a
technique called electrophoretic painting, which is closely related to
electroplating. The polymer to be deposited is solubilized with charged
functional groups. In anodic electropainting, negatively charged
groups, usually -C02-, cause the polymer molecules to move toward the
anode, where water is oxidized to 02 and H+. The local decrease in pH
leads to neutralization of the carboxylate groups and the polymer
precipitates. In cathodic electropainting, positively charged polymers
with -NH3+ groups are attracted to the cathode, where water is reduced
to H2 and OH-, the ammonium groups are neutralized, and the polymer
deposited on the surface. Electropainted polymer coatings adhere
exceptionally well, apparently because of coordination of surface metal
atoms by the polar functional groups. The technique is commonly used
to impart corrosion resistance to automobile parts.
For further details on electroplating and related techniques, see
Pletcher (G6) or Lowenheim (G3).
Anodization
The electrochemical formation of a protective oxide layer on a metal
surface is called anodization. The process is most commonly applied to
aluminum but sometimes also to copper, titanium, and steel. The metal
object to be anodized is made the anode of an electrochemical cell and
current is passed through an acidic electrolyte solution, so that the
electrode process (for aluminum) is
2 Al + 3 H20 --+ Al203(S) + 6 H +(aq) + 6 e-
The process is much like electroplating, although the cell polarity is
reversed.
The formation of the oxide layer involves a complicated electrode
process which tends to be slow, requiring a large overpotential,
particularly after a thin oxide layer is formed. For this reason, throwing
power is not a problem in anodization; the oxide layer grows most
rapidly where it is thinnest.
Aluminum is anodized using sulfuric acid, chromic acid, or oxalic
acid baths, depending on the nature of the surface desired. Anodized
aluminum surfaces can be colored by adsorbing an organic dye on the
surface. Alternatively, transition metals can be deposited in the pores of
the oxide layer; colors then arise from interference effects.
1 This is obviously a kinetic effect, not a result of thermodynamics.

Electrolytic Aluminum Production
By far the most important electrometallurgical process is the
production of aluminum from bauxite ore-hydrated aluminum oxide
with silicates, iron oxides, and other impurities. Worldwide production
of aluminum amounts to about 2 x 107 tons per year. Since the energy
consumption is about 15,000 kWh ton- 1 (5.4 x 10 10 J ton- 1 ), aluminum
production is the largest single industrial consumer of electrical energy.
Since aluminum is quite an active metal, the traditional smelting
technique used for iron,
Fe203 + 3 C -7 2 Fe + 3 CO
will not work and electrolysis is the only practical method_ Reduction of
Al(III) from aqueous solution is also impossible since hydrogen would
be evolved first even from strongly basic solutions. The solution to these
restrictions was discovered in 1886 independently by Hall (in the United
States) and Heroult (in France). The Hall-Heroult process makes use of
the solubility of alumina in molten cryolite, Na3AIFs, to give a
conducting solution from which aluminum metal can be obtained at the
cathode on electrolysis.
The bauxite ore is first treated with concentrated sodium hydroxide
solution (caustic soda to the industrial chemist) under pressure to
dissolve the alumina as NaAI(OHk After removal of the iron and
silicate salts by filtration, hydrated A1203 is reprecipitated, filtered off,
washed, and dried at 1200°C.
Carbon Anode
Molten Cryolite
& Alumina
Molten
Aluminum
Carbon Cathode
Figure 7.7 Electrochemical cell used in the production of aluminum by

the Hall-Heroult process.
402 Electrolysis
The purified alumina is then added to a bath of molten cryolite at

about 1000°C up to about 15 weight percent. The electrolysis bath, shown
schematically in Figure 7.7, is constructed of steel with alumina
insulation and uses carbon anodes. The cathode is molten aluminum
(m.p. 660°C) contacted through steel-reinforced carbon at the bottom of
the cell. Aluminum is removed from the cell from time to time. Small
amounts of AIF3 and CaF2 are normally added to the cryolite to reduce
the melting point and increase the conductivity. The exact nature of the
species in the cryolite solution is unknown but it is likely that most of the
aluminum is in the form of anionic oxyfluorides. The stoichiometry of
the cathode process is
Al203 + 6 e- ~ 2 AI(l) + 3 0 2-
At the anode, carbon is oxidized to C02:
C(s) + 2 0 2 - ~ C02(g) + 4 e-
Since 0.75 mole of carbon is oxidized per mole of aluminum produced,
the preparation of carbon anodes from coal is an important subsidiary
activity of an aluminum smelter. Since the anodes are consumed
during electrolysis, they are constructed so that they can be gradually
lowered into the cryolite melt to maintain an anode-cathode separation
of about 5 cm. The oxidation of the carbon anodes provides about half the
free energy required for the reduction of AI203.
The reversible cell potential in the Hall-Heroult process is about -1.2
V and since three Faradays of charge are required per mole of the
aluminum, we would expect an electric power consumption of about 350
kJ mol- 1 (1.3 x 1010 J ton- 1 ). This is only part of the power consumption.
The anode process is slow and requires an overpotential of about 0.5 V to
achieve a reasonable rate; iR drop in the electrodes and electrolyte adds
another 2.5 V, so that the actual working cell potential is nearer -4.2 V.
Added to this is the energy cost of heating the cell to 1000°C, purifying
and drying the bauxite, and preparing the carbon anodes. As we noted
at the outset, aluminum production is a very energy-intensive process.
Charles M. Hall (1863-1914) set to work on development of an aluminum

smelting process after graduation from Oberlin College. After his
success in 1886, he founded a company which became Aluminum
Corporation of America (Alcoa). Paul L. T. Heroult (1863-1914) went to
work on the aluminum smelting problem on completion of his studies at
Ecole des Mines in Paris (where he had studied with Henri Le Chlitelier).
He later developed electric furnace techniques for the production of steel.
Other Electrometallurgical Processes

The production of lithium, sodium, and magnesium resembles the
Hall-Heroult process in that electrolysis of a molten salt solution is
employed. In these cases, chloride salts are electrolyzed and chlorine
gas in an important by-product. Sodium, for example, is prepared by
electrolysis of a NaCl/CaCI2 eutectic mixture at 600°C. Chlorine is
liberated at a graphite anode and sodium at a steel cathode. Sodium is a
liquid at 600°C and is considerably less dense that the molten salt
mixture, so that it floats to the top of the cell and is taken off into a
reservoir. Some calcium is also produced, but this sinks to the bottom of
the cell, where it is consumed via the equilibrium
Ca+2NaCI ~ CaCI2+2Na
Less active metals can be produced by electrolysis of aqueous
solutions. Thus about 10% of world copper production and half the zinc
is via hydro metallurgical processes. Small amounts of chromium,
manganese, cobalt, nickel, cadmium, silver, gold, gallium, thallium, and
indium are also obtained from electrolysis of aqueous solutions. In
general, ores are roasted to convert sulfides to oxides,
2MS+302 ~ 2MO+2S02
and leached with sulfuric acid to separate the desired metal from
silicates and aluminosilicates. The resulting metal sulfate solution is
then purified, removing traces of less active metals, and then
electrolyzed to plate out the desired metal on an aluminum or titanium
cathode. The anode is most commonly lead or Pb02, with a little silver
added to catalyze the generation of oxygen.
Some metals obtained by conventional smelting techniques are
purified electrolytically. Thus cobalt, nickel, copper, tin, and lead are
frequently refined by anodic dissolution and deposition at the cathode of
a cell. In the case of copper, which accounts for the largest aqueous
electrorefining volume, common impurities are Fe, Co, Ni, Pt, Ag, Au,
Zn, Pb, As, Sb, and Bi. The cell potential is controlled such that Pt, Ag,
and Au do not dissolve at the anode and these metals eventually fall to
the bottom of the cell and are found in the "anode slime". Sn, Pb, Sb, and
Bi dissolve at the anode but form insoluble oxides or sulfates and so are
also found in the anode slime. Fe, Co, Ni, Zn, and As dissolve and
remain in solution but are not reduced at the cathode at the potential
provided. The anode slime is collected from time to time and refined to
separate the traces of precious metals.
404 Electrolysis
Production of Chlorine and Sodium Hydroxide

Electrolysis of an aqueous sodium chloride solution produces
hydrogen at the cathode, chlorine at the anode, and leaves a solution of
sodium hydroxide; variations on this process are responsible for most of
the world's production of chlorine and sodium hydroxide (caustic soda)
and contribute significantly to the production of hydrogen. 1 In tonnage,
this process is the largest electrochemical industry though aluminum
production consumes more electric energy.
Although the chlor-alkali industry has been in operation since the
1890's, there have been some significant technical modifications to the
process. These have consisted both in the development of better
electrodes to reduce overpotentials and in changes in the overall strategy
of the process. Three distinct approaches to the electrolysis of NaCI
brine have been developed and all three are in current use.
50% N aCl---i~ 1--~1>17% NaCl
<E--Hg
50% NaOH
Figure 7.S Schematic diagram of mercury cell used in the production of

chlorine, sodium hydroxide, and hydrogen.
The oldest technology divides the overall process into two parts by
using a cell with a mercury cathode as shown in Figure 7.8. Since the
1 Hydrogen is also produced from coal or natural gas as "synthesis gas" (a mixture
of H2 and CO) or, when very high purity is required, from the electrolysis of water.
overpotential for H2(g) production on Hg is so high, the cathode process
is the formation of sodium amalgam,
Na+(aq) + e- -+ Na(Hg)
The anode is titanium, coated with RU02, C0203, or other transition
metal oxides to catalyze the evolution of chlorine gas,
2 CI- -+ CI2(g) + 2 e-
Since the evolution of 02 is thermodynamically more favorable than that
of CI2(g), the anode must have a high overpotential for water oxidation so
as to prevent contamination of the Cl2 product. The reaction of sodium
amalgam with water is thermodynamically favorable but is very slow,
owing to the large overpotential for H2 production. Sodium amalgam is
withdrawn from the cell at about 0.5% Na content, washed with water to
remove NaCI and passed with water through a column packed with
graphite impregnated with Fe, Ni, or some other catalyst. Hydrogen is
then evolved and collected at the top of the column. Mercury and 50%
NaOH solution emerge at the bottom of the column in quite pure form.
The NaCI feedstock is typically 50% brine which has been treated with
base to precipitate group II and transition metal hydroxides and
reacidified to pH 4. The cell normally operates at about 60°C.
The mercury cell technology produces high-purity products with
little additional treatment required. However, there is a potentially
serious problem of mercury contamination in the plant environment
and in the wastewater effluent. Because Na+ is reduced in the
electrolysis step rather than H+, the cell potential-and thus energy
consumption-is significantly higher than should be necessary (in
principle) for the electrolysis of NaCI to form Ch(g) and H2(g).
Beginning in the 1950's, environmental concerns prompted the
development of another approach, the so-called diaphragm cell. In this
method, the cathode material is steel gauze coated with a catalyst to
reduce the H2 overpotential. An asbestos-based diaphragm is used to
separate the anode and cathode compartments to prevent mixing of the
hydrogen and chlorine gases and to somewhat reduce mixing of the
N aCI feedstock on the anode side with the N aOH solution of the cathode
side. The arrangement is shown schematically in Figure 7.9. While this
eliminates the mercury problem, the cell potential is still much larger
than theoretical because of an iR drop of about 1 V across the
diaphragm. This is not the only problem: (1) OH- diffusing into the
anolyte reacts with Cl2 to form hypochlorite ion:
Cl2 + 2 OH- -+ H20 + CIO- + CI-
reducing the yield of C12; (2) when the anolyte pH goes up, the potential
for formation of 02 at the anode becomes more favorable and
contamination of the CI2(g) results; (3) to reduce the magnitude of
problems (1) and (2), the catholyte cannot be allowed to exceed 10% in
406 Electrolysis
NaOH concentration; (4) chloride ion diffusion from the anolyte to the
catholyte results in a very considerable contamination of the NaOH
solution; (5) to deal with problems (3) and (4), evaporation of the caustic
H2 Cl 2 Cl 2
t t t
~
0
CIS
--
Q
Z
CIS
Q
Z
CIS
Z
~ ~ ~
0 0
0 C',) C',)
,...;
Figure 7.9 Schematic representation metal asbestos steel

of electrodes in a diaphragm cell used gauze diaphram gauze
in the chlor-alkali industry. anode cathode
soda solution is required to bring the concentration up to 50% (the usual

commercial form of caustic soda), at which point much of the NaCI
precipitates; and (6) precipitation of Ca(OH)2 and Mg(OH)2 in the pores of
the diaphragm increases the iR drop with time; to minimize this
problem, more rigorous brine purification is required than for the
mercury cell.
The upshot of all this is that the purity of the NaOH and Cl2
products from a diaphragm cell is significantly lower and, because of iR
drop in the diaphragm and the requirement of an evaporation step, the
energy requirement is nearly the same as in a mercury cell.
Furthermore, the use of asbestos introduces another potential
environmental problem.
The most recent technical development has been the introduction of
chlor-alkali cells which incorporate an ion-exchange membrane in
place of the asbestos diaphragm. With a cation-permeable membrane,
diffusion of OH- into the anolyte can be drastically reduced, improving
the yield and purity of C12. Electrolysis can be continued to produce
somewhat more concentrated NaOH solutions, up to about 40%, so that
much less evaporation is required to get to 50% commercial caustic soda;
because very little CI- gets through the membrane, CI- contamination of
the NaOH is much less serious, and finally, iR drop across the
membrane is somewhat less than in a diaphragm cell (but still much
greater than in a barrier-free mercury cell).
§ 7.4 Industrial Electrolysis Processes 407
The major drawback of the membrane cell is that (at present) it is
limited in size by the ability to produce and handle large sheets of cation-
exchange membrane. To obtain the same product volume, more cells
are required and therefore more plumbing and higher capital costs to
the manufacturer.
Other Inorganic Electrosyntheses

Although no other electrolytic process approachs the volume of the
chlor-alkali process, many other inorganic chemicals are produced
electrochemically.
Fluorine. Electrolysis provides the only source of elemental
fluorine. An anhydrous mixture of KF and HF (mole ratio 1:2, m.p. 82°C)
is used as the electrolyte (K+ and HF2- ions) with carbon anodes and mild
steel cathodes. The cell process
2 HF ~ H2(g) + F2(g)
gives gaseous fluorine and hydrogen. Both gases are contaminated with
HF which is removed and recycled by adsorption by solid KF. Fluorine is
used in the nuclear industry to produce UF6 (used in gaseous diffusion
isotope separation plants), for the production of SF6 (a nonflammable,
nontoxic gas of low dielectric constant used to prevent voltage breakdown
around high-voltage electrical equipment), and for the production of
fluorinated hydrocarbons.
Chlorates, bromates, and perchlorates. Sodium chlorate is
produced via electrolysis of sodium chloride solutions, taking advantage
of an undesirable side reaction of the chlor-alkali process. The
electrolysis steps are the same,
2 CI- ~ Cl2 + 2 e-
2H++2e- ~ H2
In neutral or basic solution (pH> 6), Clz disproportionates:
Cl2 + 2 OH- ~ CI- + CIO- + H20
Hypochlorite ion also disproportionates:
3 CIO- ~ ClOa- + 2 CI-
The last reaction is catalyzed by H+ and is slow in basic solution. When
the pH is in the range 6-7, both disproportionation reactions are
relatively slow. However, since the anode process consumes two OH-
ions and the cathode process consumes two H+ ions, there is a
substantial pH gradient across the cell and the rates of the various
processes are strongly dependent on stirring and on flow patterns in the
electrolyte solution. If hypochlorite ion is formed near the anode, it can
408 Electrolysis
be oxidized in a complicated process which generates oxygen with the
overall stoichiometry:
!
6 CIO- + 6 OH- ~ 2 CI03- + 4 CI- + 02 + 3 H20 + 6 e-
In effect, CIO- catalyzes the anodic oxidation of water, reducing the
current efficiency. This process is thus undesirable and must be
suppressed. Efficient stirring of the solution will maintain pH control
and thus reduce the rate of hypochlorite formation. In practice, the
solution is cycled through a holding tank where the disproportionation
reactions can proceed without the complications of the unwanted
electrode processes. Other chlorate and bromate salts are produced by
similar electrolytic methods.
Electrolysis of sodium chlorate solutions at pH 0-1 results in
oxidation to perchlorate:
CI03- + H20 ~ CI04- + 2 H+ + 2 e-
Hydrogen is evolved at the cathode.
Manganese dioxide. Purified pyrolusite provides Mn02 of
adequate quality for most chemical purposes. When Mn02 of higher
purity is required (for Leclanche cells, for example, or for use as an
oxidizing agent in industrial organic synthesis), it can be produced by
the electrolytic oxidation of Mn(II) in aqueous sulfuric acid solution at a
graphite anode. Hydrogen evolution occurs at the cathode.
Potassium permanganate. Pyrolusite (impure Mn02) is used as
a starting material in the production of KMn04. The ore is ground,
slurried in 50% KOH, and air oxidized to manganate ion:
2 Mn02 + 02 + 4 OR- ~ 2 H20 + 2 Mn042-
Impurities are filtered off and the manganate ion is oxidized at a nickel
anode:
Mn042- ~ Mn04- + eo
The cathode process again is hydrogen evolution, although some Mn04-
is lost to cathodic reduction.
Potassium dichromate. Potassium dichromate is produced by
electrolytic oxidation of Cr(III) in aqueous sulfuric acid solution, using a
Pb02 anode; oxygen evolution is an important competing anodic process.
Electrolytic production of Cr(V!) is mostly used in recycling Cr(III)
residues from chromate oxidations in industrial organic synthesis or in
conditioning chromate plating baths. As such, the scale is usually quite
small.
Electrosynthesis of Adiponitrile
The largest volume organic electro synthetic process presently in
use is the production of adiponitrile, a precursor to nylon, from
acrylonitrile. The stoichiometric half-cell reactions are
2 CH2=CHCN + 2 H20 + 2 e- ~ NC-(CH2)6-CN + 2 OH-
2 H20 ~ 02(g) + 4 H + + 4 e-
which sum to the overall reaction stoichiometry
4 CH2=CHCN + 2 H20 ~ 02(g) + 2 NC-(CH2)6-CN
The actual mechanism of the reaction is considerably more complex and
probably involves several parallel pathways. Some of the possible
reaction pathways are shown in Figure 7.10. The major side reactions
which must be minimized are the production of propionitrile (the
monomeric reduction product) and trimers and higher polymers.
,
CH2CHCW
.
l
CH2CHCN
+e-
«H2 CHCN- +H+

CH2CHCW
trimer,
- etc .
.CH2CH2CN
+SH
-S t+ e-
CH2CH2CN"" +SH
t + H+
-S
CH3CH2CN
Figure 7.10 Possible reactions occurring in the reduction of acrylonitrile.
Monomeric product can be avoided if the initially formed radical

anion either dimerizes or reacts with substrate acrylonitrile before it is
protonated. This can be achieved if the pH (at least near the electrode
surface) is kept quite high. Radical anion dimerization is most desirable
410 Electrolysis
since this avoids the neutral radical, which is the most likely route to
higher polymers. On the other hand, the pH cannot be allowed to get too
high since acrylonitrile reacts with hydroxide ion to form ~
hydroxypropionitrile and bis-(2-cyanoethyl)ether. These considerations
led to a definite strategy for the process:
(1) The concentration of acrylonitrile should be as high as possible to
promote dimerization; in practice an aqueous emulsion is used so
that the aqueous phase is saturated, about 7% by weight. The
organic phase is a mixture of acrylonitrile and adiponitrile.
(2) The aqueous phase must be buffered to prevent reaction of
acrylonitrile with base; in practice the emulsion contains about 15%
Na2HP04.
(3) The electrode surface pH should be high to prevent protonation of
the initially formed radical anion. This can be achieved by
adsorbing positive ions on the electrode surface (see §2.2); the
industrial process uses the diquaternary ammonium ion, BU2EtN-
(CH2)6-NEtBu22+.
(4) Hydrogen evolution at the cathode must be suppressed so that the
cathode material should have a large overpotential for H+
reduction; in practice, cadmium-plated steel is used.
Oxygen is produced at a steel anode, spaced 2 mm from the cathode l
to reduce solution iR drop. Corrosion of the anode is a problem since
deposition of iron on the cathode would promote hydrogen reduction and
a decrease in current efficiency. Additives are used to suppress anode
dissolution and EDTA is added to scavenge transition metals before they
plate out on the cathode.
The two-phase mixture is cycled rapidly between the electrolysis
cell and a reservoir. Solution is continually withdrawn from the
reservoir for product isolation, removal of by-products and metal ions,
and recycling of unreacted acrylonitrile. The overall process has about
90% selectivity toward adiponitrile. The reversible cell potential is about
-2.5 V; the overpotentials and solution iR drop are relatively small at the
current density used (about 2000 A m- 2) so that the working cell potential
is about -3.8 V.
Production of Lead Tetraalkyls

Another major commercial electro synthetic process is the
production of tetraethyl- and tetramethyllead by the oxidation of a
1 About 100 anode-cathode pairs are used in the cell to allow for greater output.
Grignard reagent at a lead anode in a mixture of ether solvents. l The

anode reaction stoichiometry is
4 RMgCI + Pb ~ R4Pb + 4 MgCI+ + 4 e-
At the steel cathode, Mg(II) is reduced:
MgCI+ + 2 e- ~ Mg(s) + CI-
and the magnesium is used to regenerate the Grignard reagent with
excess alkyl chloride,
RCI + Mg ~ RMgCI
The electrolysis reactor contains many cell assemblies, arranged as
concentric vertical tubes as shown in Figure 7.11; lead shot which acts
as the anode is fed by gravity down the center tube.
Lead Rod
Electrode
Lead Shot
(anode)
tIl--+- Polypropylene
diaphram
Figure 7.11 Cross-section Solution

of electrode assembly used
in the electrochemical
production of lead tetra- Steel tube
alkyls. cathode
Careful control of reaction conditions is required since a

hydrocarbon is produced if the alkyl chloride concentration is too high:
RMgCl + RCI ~ R-R + MgCl2
If the alkyl chloride concentration is too low, magnesium metal will
accumulate on the cathode, eventually bridging the anode-cathode gap
and shorting the cell.
Solution flows from a reactor where the Grignard reagent is
prepared into the electrolysis cell assemblies; the flow rate must be
adjusted to ensure complete reaction of the Grignard reagent by the time
the solution exits at the bottom of the cell. Excess alkyl halide is then
1 With the phase-out of leaded gasoline in the United States and Europe, this process
has decreased in importance; tetraethyllead is still used as a motor fuel additive in
much of the rest of the world.
412 Electrolysis
stripped from the solution and recycled to the Grignard reactor, the
MgCl2 is extracted with water, and the ether solvent is separated from
the lead tetraalkyl, purified by azeotropic distillation and recycled to the
Grignard reactor.
Other Organic Electrosyntheses

Although no other organic processes use electro synthetic
techniques at the scale of the adiponitrile or lead tetraalkyl processes,
several other processes are in use. They fall into two categories:
(1) Reduction of aromatic compounds in aqueous acidic solutions at a
lead cathode. The initially formed radical anion is rapidly
protonated; the resulting neutral radical is then reduced to a
carbanion which adds another proton. For example, phthalic
anhydride is reduced to 1,2-dihydrophthalic acid, pyridine is
reduced to piperidine, and N-methylindole is reduced to N-methyl-
2,3-dihydroindole.
(2) Oxidations of organics with inorganic oxidants such as chromic
acid, periodate, bromine, or hypobromite. These homogeneous
reactions are coupled with a step where the oxidant is regenerated
electrolytically. For example, anthroquinone is obtained from the
chromic acid oxidation of anthracene and the CrOll) is cycled
through an electrolysis cell to regenerate the oxidant.
7.5 CORROSION
Corrosion is an important problem in all practical uses of metals-

in automobiles, bridges, buildings, machinery, pipelines, ships, indeed
in most of the works of man. Corrosion consists of the spontaneous
oxidation of metals, usually by atmospheric oxygen but also by water or
atmospheric pollutants such as 802. In simplest terms, we may think of
corrosion in terms of the metal oxidation half-cell reaction,
M -0 Mn+ + n e-
coupled with the reduction of oxygen
02 + 2 H20 + 4 e- -0 4 OH-
or the reduction of water
2 H20 + 2 e- ~ H2 + 2 OH-
The direction of spontaneous change and the thermodynamically
most stable products will depend on the standard reduction potentials of
Mn+, 02, and water, and on the pH, 02 partial pressure, and
§7.5 Corrosion 413
temperature. The equilibrium thermodynamics of a corrosion problem
is most easily visualized in terms of a Pourbaix diagram (see §1.3). In
the Pourbaix diagram for iron, Figure 7.12, we see that iron is
susceptible to oxidation by water or 02 at all pH values.
Thermodynamics thus tells us that corrosion of iron is inevitable in the
wet, oxygen-rich environment in which we place our buildings and
machinery. The questions then are: (1) How fast does the process occur?
and (2) what can we do to make it slower? These questions lead us to a
more detailed look at the kinetics and mechanism of corrosion
processes. For further discussion of corrosion and the prevention
thereof, see Uhlig and Revie (G8).
1.5
:: ~
E/v·:S.
J.; ••••••
...........~:;;:;.............
Fe 2+(aq)
0.0 ••••••••••••••••••••••
-0.5
Figure 7.12 Predominance Fe(s)
area diagram for iron. The
dashed lines correspond to -1.0
the H+(aq)/H2(g) and o 5
pH
10 15
02(g)/H20 couples.
Rate of Dissolution ofa Metal in Water

Consider the reaction of a metal with aqueous acid,
M(s) + n H+(aq) -7 Mn+(aq) + (n/2) H2(g)
We could study the rate of this reaction by classical techniques, e.g., by
measuring the volume of H2(g) as a function of time. However, the half-
reactions
M(s) -7 Mn+(aq) + n e-
2 H+(aq) + 2 e- ~ H2(g)
can be studied independently using the electrochemical techniques
described in §5.2. If a piece of the metal were used as an indicator
414 Electrolysis
electrode in a well-stirred solution and the current density measured as

a function of potential, a current density vs. potential curve would be
obtained similar to that of Figure 7.13.
Cathodic current at negative potentials corresponds to the reduction
ofH+(aq) and, for E« Ee H , is given by the Tafel equation, eq (5.21),
1 ·H -1 . H aHF(E -EeH) (7.7)
og} - og}o - 2.303RT
where a H is the apparent cathodic transfer coefficient. Anodic current
2
o
~
S
S0
C'l
------
,~.-----=-"-::-~-
<
-
< ~-2
:;:;:'-2 o
-0.2 -0.3 -0.4 -0.5 -0.6 -0.7 -0.1 -0.2 -0.3 -0.4 -0.5 -0.6 -0.7
ElY ElY
Figure 7.13 Current density us. Figure 7.14 Tafel plot corresponding
potential curve for iron in contact to the current density us. potential
with an aqueous acidic solution (pH curve of Figure 7.13.
2, [Fe 2+] = 1 mM) saturated with H2
(1 bar partial pressure).
at more positive potentials (E »Ee M ) corresponds to oxidation of the

metal l and is given by
log IjMI = loglj~1 + ~MF(E -EeM) (7.8)

2.303RT
where ~M is the apparent anodic transfer coefficient. Consider now the
point on thej-E curve wherej = O. Zero current corresponds to the point
where the anodic and cathodic components of the current are equal. At
1 We assume that the experiment is done in an oxygen-free solution and that the
anion is not electroactive.
§7.5 Corrosion 415
the potential of zero current, the rate of reduction of H+(aq) is exactly

equal to the rate of metal oxidation. This, of course, is just the situation
which pertains when we immerse the metal in the acid solution and
allow the reaction to proceed. The potential at zero current then
corresponds to the metal-solution potential in the absence of
electrochemical control circuitry and is called the corrosion potential,
Ec·
The corrosion potential is easily obtained by equating jH and IjM I ,
_ (aHEe H + J3MEeM) + 2.30j RT (logjoH -log li~D
Ee------------------A~~------------- (7.9)
a H + 13 M
Substituting eq (7.9) into either eq (7.7) or eq (7.8) gives the corrosion
current density at zero net current:
I . _aHlogljoMI+~MlogjoH aH~MF(EeH-EeM)
ogle - M + ( M) (7.10)
a H + 13 2.303 RT a H + 13
Thus electrochemical kinetic methods can be applied to a study of
the rate of corrosion reactions. Furthermore, given exchange current
densities and standard potentials for the half-cell reactions of interest,
we should be able to compute the corrosion potential and the rate of the
corrosion reaction.
Example 7.5 Compute the corrosion rate for metallic iron in

contact with an aqueous acid solution, pH 2.00, P(H2) = 1 bar, with
[Fe2 +] = 0.001 M using the results of Bockris, Drazic, and Despic
(see Example 6.4 and Problem 6.7) for the anodic dissolution of
iron and the data of Table 6.1 for the cathodic reduction ofH+(aq).
We first use the Nernst equation to compute the equilibrium half-
cell potentials
Ee H = 0.000 + (0.0592) 10g[H+] = -0.118 V
Ee M =-0.440 + (0.059212) 10g[Fe2+] =-0.529 V
According to Figure 6.9, log IjoM I = -2.2 at pH 3.1, 0.5 M Fe2+, with
13M = 1.5. Extrapolation to [Fe2+] =0.001 M at pH 3.1 gives 10gljoM I
= -4.4. Correcting this value to pH 2.0 with the observed pH
dependence, we have log IjoM I = -5.5. The reduction ofH+(aq) at an
iron electrode has 10gjoH = -2.0, a H = 0.4 for pH O. CorrectingjoH to
pH 2 using eq (5.25a), we have
10 'H=_2.0+(1-aH)FLille -3.2
g}o 2.303RT
Thus the Tafel lines corresponding to the reduction of H+(aq) and
the oxidation of iron under the specified conditions are
416 Electrolysis
logjH = -3.2 - 6.76(E + 0.118)

log IjM I = -5.5 + 25.4(E + 0.529)
These are plotted in Figure 7.14 and the total current density,jtot =
jH - IjM I, is plotted us. potential in Figure 7.13. The corrosion
potential found from eq (7.9) is Ec = -0.371 V and the corrosion
current density from eq (7.10) is jc = 0.033 A m- 2. The rate of the
corrosion reaction thus is
Rate =jc/nF = 1.7 x 10-7 mol m- 2s- 1
Such a rate seems very slow, but corrosion reactions need not be
very fast to do their work; this corrosion rate corresponds to 300 g
of metal dissolving per square meter per year.
Reaction of a Metal with Air-Saturated Water

In most corrosion problems of practical interest, the aqueous
solution (which might be just a thin film) is in contact with air and the
reaction of the metal with oxygen is generally much more important
that the reaction with H+(aq) from the water.
In principle, we should be able to use the model developed above to
measure or predict oxygen corrosion rates. Since oxygen is a stronger
oxidant than water by 1.23 V, the cathodic current density due to oxygen
reduction will almost always be larger than that due to water reduction.
The oxygen reduction Tafel line will intersect the line for metal
oxidation at a more positive potential and higher current density, thus
giving a greater corrosion rate. The actual situation may be
considerably less straightforward.
There is relatively little electrochemical kinetic data available for
the reduction of oxygen and the mechanism of the process is, for the
most part, not understood. Even when reliable data are available for the
electrochemical reduction of oxygen, they may not be relvant to the
actual corrosion process. For example, in low pH solutions, the reaction
of dissolved oxygen with Fe(II) is very much faster than the reaction
with solid iron. The Fe(III) product, however, undergoes electron
transfer at the metal surface quite rapidly. The overall reaction then is
the sum of the homogeneous and heterogeneous steps:
4 Fe2+ + 02 + 4 H + ~ 4 Fe3+ + 2 H20
2 Fe3 + + Fe(s) ~ 3 Fe2+
2 Fe(s) + 02 + 4 H+ ~ 2 Fe2+ + 4 H20
Thus the stoichiometry is identical to that of the direct reaction,
although the mechanism is more complicated.
§7.5 Corrosion 417
Passivation
Some metals-most notably aluminum, chromium, and nickel-
are essentially unreactive in air-saturated water despite a large
thermodynamic driving force. Aluminum, for example, can be used in
cookware despite an expected free energy of oxidation of -717 kJ mol-I.
Metals are said to be passive when they corrode anomalously slowly.
Many other metals, including iron, show passive behavior under some
circumstances and "normal behavior" under others. It has been known
since the eighteenth century that iron is attacked rapidly by dilute nitric
acid but is essentially immune to attack by concentrated nitric acid.
Treatment of iron with concentrated nitric acid imparts temporary
corrosion resistance when the iron is immersed in dilute acid. Iron can
also be temporarily passivated by electrochemical anodization. Whereas
iron loses passivity in a matter of minutes, chromium and nickel show
long-term passive behavior.
Figure 7.15 Anodic current-

potential curve showing the
onset of passivation at the Flade
potential, EF. Potential
The passivation of a metal can be demonstrated by a current-

potential curve such as that shown in Figure 7.15. The experiment
starts with a clean metal surface in contact with a stirred solution
saturated in oxygen with the potential of the metal held sufficiently
negative that no metal oxidation occurs. A positive-going potential scan
then gives an anodic current as the metal begins to oxidize. Initially, the
current increases exponentially as expected. Instead of a continued
increase, however, the current levels off and then drops precipitously,
perhaps by several powers of 10; the current remains very small until
418 Electrolysis
the potential becomes sufficiently positive to oxidize water. The potential

of onset of passivation is called the Flade potential. The details of the
mechanism of passivation-in particular, the sudden onset of the
passive region at the Flade potential-are imperfectly understood. The
Flade potential depends on the pH, usually according to
EF = Epo - 0.0592 pH
suggesting the formation of an oxide film by the process
M+H20 ~ M·O+2H++2e-
The Flade potentials (at pH 0) for iron, nickel, and chromium are,
respectively, +0.6, +0.2, and -0.2 V, and these values correlate well with
the apparent kinetic stability of the protective oxide film. The Flade
potential of iron-chromium alloys decreases rapidly with increasing
chromium content; E FO is about 0 V at 20% chromium. Thus the
corrosion resistance of "stainless" steels is brought about through
stabilization of the passivating oxide film.
In the case of main-group metals such as magnesium, aluminum,
or zinc, the oxide film apparently grows into a layer of the metal oxide
and can be regarded as a separate phase. The passivity of such metals
then depends on the slow rate of diffusion of water or oxygen through the
oxide layer. Transition metals, on the other hand, show no evidence of a
surface oxide phase and there is reason to believe that the passivating
film is a layer of oxygen atoms chemisorbed to the metal surface, M-O.
Coulometric experiments suggest that the passivating layer on iron is 4-
5 oxygen atoms thick, but, because of surface roughness, this is an upper
limit to the film thickness. There is some reason to suspect that the film
contains both oxygen atoms and oxygen molecules. Uhlig (26) has used
other thermodynamic data to estimate a potential of 0.57 V for the
process
Fe--O-02 + 6 H+ + 6 e- ~ Fe + 3 H20
This is sufficiently close to the experimental value, EFo = 0.63 V to
encourage belief in the model. If the passivating film is a layer of
chemisorbed oxygen atoms and molecules, its success in protecting the
metal from further oxidation must mean that the oxide film is lower in
free energy than either the bare metal or a thin layer of the metal oxide
phase_ Thus, although thermodynamics says that complete conversion
of the metal to metal oxide is spontaneous, the chemisorbed oxide layer
represents a pronounced local free energy minimum on the pathway
from metal to oxide.
§7.5 Corrosion 419
Differential Aeration
Although ionic conduction across a passivating metal oxide film is
very slow, electronic conduction may be much faster. Thus even at a
passivated metal surface, reduction of oxygen may occur provided that a
source of electrons is available. Consider the experimental
arrangement shown in Figure 7.16. Two initially identical pieces of
metal are dipped into identical electrolyte solutions. Nitrogen is bubbled
over one electrode and oxygen over the other and we suppose that the
metal exposed to oxygen is passivated. If we connect the two electrodes
by closing the switch, we have set up an electrochemical cell in which, at
the left-hand electrode, metal is oxidized,
M(s) ~ Mn+ + n e-
and at the right-hand electrode, oxygen is reduced,
02(g)+4H++4e- ~ 2 H20
This overall cell reaction will almost always be spontaneous and the rate
may be quite significant, despite the fact that the right-hand electrode is
passivated.
+-°2
Figure 7.16 Experiment to

demonstrate corrosion
through differential aera-
tion.
This experiment gives an insight into one of the most troublesome

sources of corrosion, differential aeration. If a metal object is exposed to
well-aerated water, it will often passivate, so that the rate of corrosion is
negligible. If, however, part of the object is exposed to oxygen-deficient
water, an electrochemical cell exactly like that of Figure 7.16 is set up.
One can think of many situations where this might occur. For example,
a metal piling in a river is exposed to well-aerated water near the
surface but anaerobic mud at the bottom. The stagnant water trapped in
a crevice in a metal object may have an oxygen concentration below the
point required for passivation. Notice that in both these examples, the
420 Electrolysis
surface area of metal exposed to aerated water is large, so that oxygen

reduction may be quite rapid overall; the rate of oxygen reduction must
be balanced by metal oxidaLi.on, orlen in a relatively small area. Thus
corrosion by differential aeration is usually characterized by rapid
dissolution of metal in very localized areas. Cracks in a metal structure
are thus enlarged, pits deepened, imperfect welds destroyed, etc.
Passive metal surfaces remain susceptible to corrosion in other
ways as well. Although a passivating oxide layer behaves as if it were
impermeable to oxygen or metal ions, it generally is either permeable to
or is disrupted by halide ions. Thus metal passivity is usually lost when
exposed to a solution containing chloride ions. The action of chloride
ions tends to be localized, resulting in the formation of corrosion pits.
Apparently chloride attacks thin or otherwise weak spots in the oxide
layer, creating small anodic areas of acti ve metal surrounded by large
cathodic areas of passive metal. The mechanism thus is analogous to
different aeration. Once the process starts, corrosion occurs primarily
at the anodic sites and a corrosion pit forms and grows.
Prevention of Corrosion
Perhaps the most common approach to the prevention of corrosion
is to cover the metal surface with a protective coating, e.g., paint, plastic,
ceramic, or an electroplated layer of a passive metal such as chromium,
nickel, or tin. A mechanical barrier is effective, of course, only as long
as it remains a barrier. Bare metal exposed by a scratch is susceptible to
corrosion. Once corrosion begins at a breach in the protective layer, it
often continues under the layer through the differential aeration
mechanism, lifting the coating and eventually leading to massive
damage. Differential aeration is particularly effective with damaged
metal-plated surfaces since oxygen reduction can occur over the entire
undamaged surface to balance the base metal oxidation at the point of
damage.
Coatings can be made more effective by the addition of corrosion
inhibitors. A corrosion inhibitor works by decreasing the rate of metal
oxidation, by decreasing the rate of oxygen (or water) reduction, or by
shifting the surface potential into the pastsive region. Thus rust-
inhibiting paints often contain chromate salts which passivate the metal
surface by making the surface potential positive. Other common paint
additives are aliphatic or aromatic amines which appear to adsorb on
the metal, slowing the metal oxidntion half-reaction. Since inhibitors
impart a resistance to corrosion over and above the effect of the physical
barrier, small scratches do not usually lead to massive damage.
Another way to protect a metal object from corrosion is to attach a
sacrificial anode made of zinc, magnesium, or an aluminum alloy. In
effect a short-circuited galvanic cell is produced where the protected
object is the cathode (oxygen is reduced) and the more active metal is
§7.5 Corrosion 421
oxidized. Consider, for example, the corrosion potential of zinc. Since

the reduction potential of zinc(II) is -0.76 V, then (assuming similar
kinetics) its corrosion potential will be about 0.3 V more negative than
that of iron. Thus if the two metals are in electrical contact, the surface
potential of iron will be more negative. According to Figure 7.14, a 0.3-V
negative shift should reduce the rate of metal oxidation by about five
powers of 10.
As the name implies, sacrificial anodes do corrode and may have to
be replaced from time to time. The ideal anode material has some
passivity so that the rate of corrosion is low, but a completely passive
anode would impart no protection at all; thus aluminum must be alloyed
(with magnesium, for example) to increase its corrosion rate. Sacrificial
anodes are commonly used to protect such objects as ship hulls,
pipelines, and oil-field drilling rigs from environmental corrosion.
Surface treatment with chromate imparts corrosion protection by
making the metal surface anodic and hopefully passive. Contact with a
sacrificial anode makes the protected surface cathodic and thus slows
the rate of metal dissolution. Either of these approaches could be used if
we add an inert electrode and control the surface potential with a
potentiostat. This sounds like a strictly laboratory approach to the
problem, but in fact potentiostatic control of corrosion is used in certain
kinds of applications, e.g., in the protection of ship hulls.
REFERENCES
(Reference numbers preceded by a letter, e.g., (F6), refer to a book listed in

the Bibliography.)
1. W. Cruikshank, Ann. Phys. 1801, 7, 105.

2. M. Faraday, Experimental Researches in Electricity, London:
Bernard Quaritch, 1839 [facsimile reprint, London: Taylor &
Francis, 1882; included in Vol. 45 of Great Books of the Western
World, R. M. Hutchins, ed, Chicago: Encyclopedia Brittanica, 1952],
Vol. 1, Series V.
3. O. W. Gibbs, Z. anal. Chern. 1864,3,327.
4. N. Tanaka in Treatise on Analytical Chemistry, 1. M. Kolthoff and
5. L. Szebelledy and Z. Somogyi, Z. anal. Chern. 1938, 112, 313.
6. D. D. DeFord and J. W. Miller in Treatise on Analytical Chemistry,
1. M. Kolthoff and P. J. Elving, eds, Part I, Vol. 4, New York:
Interscience, 1963, p 2475.
7. J. E. Harrar, Electroanalytical Chemistry 1975, 8, 1.
8. J. J. Lingane, J. Am. Chern. Soc. 1945,67,1916.
9. L. Meites and J. Meites, Anal. Chern. 1956,28, 103.
422 Electrolysis
10. D. J. Curran in Laboratory Techniques in Electroanalytical

11. J. C. Verhoef and E. Barendrecht, J. Electroanal. Chern. 1976, 71,
305; 1977, 75, 705.
12. I. Shain in Treatise on Analytical Chemistry, I. M. Kolthoff and P. J.
Elving, eds, Part I, Vol. 4, New York: Interscience, 1963, p 2533.
13. E. Barendrecht, Electroanalytical Chemistry 1967,2,53.
14. W. R. Heineman, H. B. Mark, Jr., J. A. Wise, and D. A. Roston,
Laboratory Techniques in Electroanalytical Chemistry, P. T.
Kissinger and W. A. Heineman, eds, New York: Marcel Dekker,
1984,p499.
15. P. Ostapczuk and Z. Kublik, J. Electroanal. Chern. 1976, 68, 193;
1977,83,1.
16. K. Shimizu and R. A. Osteryoung, Anal. Chern. 1981,53,584.
17. M. Mohammad, J. Hajdu, and E. M. Kosower, J. Am. Chern. Soc.
1971, 93, 1792.
18. J. Casanova and J. R. Rogers, J. Org. Chern. 1974,39, 2408.
19. M. R. Rifi, J. Am. Chern. Soc. 1967,89,4442; Tetrahedron Let. 1969,
1043; Coll. Czech. Chern. Commun. 1971,36, 932.
20. S. Wawzonek and H. A. Laitinen, J. Am. Chern. Soc. 1942, 64, 2365.
21. T. Shono and T. Kosaka, Tetrahedron Lett. 1968,6207.
22. R. Engels, R. Schafer, and E. Steckhan, Ann. Chern. 1977, 204.
23. T. Shono, I. Nishiguchi, and M. Ohkawa, Chern. Lett. 1976, 573.
24. N. L. Weinberg and B. Belleau, Tetrahedron 1973,29, 279.
25. H. Kolbe, Ann. 1843,69,257.
26. H. H. Uhlig, Z. Elektrochem. 1958, 62, 626.
Problems 423
PROBLEMS
7.1 Show that the charge passed through an electrolysis cell of

volume Vis
Q(t) = nFVCo*(O) [1- exp(-kt)]
where k is given by eq (7.2), CO*(O) is the initial bulk concentration
of 0 and the mass transport rate constant kD is assumed constant.
7.2 Repeat the calculation of Example 7.1 assuming that the solution
is unstirred and that the current remains diffusion-controlled
indefinitely. The diffusion coefficient of Cu2+ is ca. 7 x 10- 10 m 2 s- 1.
7.3 If the electrode surface area to solution volume ratio is large,

density gradients usually lead to convective mixing so that the
mass transport rate is often much greater than might be expected
from diffusion alone and may remain more or less constant
during an electrolysis. For example, radical anions are prepared
for electron spin resonance study by electrolysis in a 0.2 mL cell
equipped with a cylindrical working electrode 5 mm long and 1
mm in diameter. With a substrate concentration of 1 mM, a
convection-limited initial current of about 10 ~ is obtained.
(a) How long will it take to reduce halfthe substrate?
(b) Suppose that the radical anions produced decay by first-order
kinetics with a half-life of 5 minutes. What is the radical anion
concentration after electrolysis for 5 minutes? 10 minutes? 25
minutes?
7.4 Show that the optimum conditions for an electro separation

process are achieved when the electrode potential is halfway
between the standard potentials of the two couples (assuming that
nl = n2).
7.5 Derive eq (7.6) given eqs (7.4) and (7.5).
7.6 What is the minimum charge passed through a silver coulometer

if the charge is to be measured with a precision of ±0.1% if the
cathode is weighed before and after electrolysis to ±0.1 mg?
7.7 An electronic coulometer is used in the controlled-potential

coulometric determination of trichloroacetic acid.
424 Electrolysis
(a) With Rl = 1 kQ, R2 = 1 MQ, C = 50 JlF, the final output potential

of the coulometer was 4.679 V. What charge was passed through
the electrolysis cell?
(b) If the volume of the Cl3CCOOH solution was 25 mL, what was
the molar concentration?
7.8 The operational amplifier galvanostat circuit of Figure 7.6 is used

to supply a constant current of 1 rnA through an electrolysis cell.
(a) If the control battery supplies a potential of 1.00 V, what size
resistor should be used in the circuit?
(b) If the maximum output voltage of the operational amplifier is
±10 V, what is the maximum allowable cell resistance?
7.9 A constant current of 13.64 mA was used in a coulometric Karl

Fischer titration of water in 5 mL of acetonitrile. The endpoint
was obtained after 248 s. What was the water concentration?
7.10 Suppose that you have a constant current source which can
deliver 1, 2, 5, 10, 20, 50, or 100 rnA (±O.25%), an endpoint detection
system and timer which is reliable to ±1 s, and a collection of
pipets (1,2,5,10,25, and 50 mL), each of which is accurate to ±0.02
mL. You wish to do a coulometric titration of a solution where the
analyte concentration is approximately 1 mM. What volume and
current should be chosen if the analytical uncertainty must be
less than 1.0% and it is desirable to use minimum sample size
and do the titration in minimum time? Assume a one-electron
oxidation is involved.
7.11 Unsaturated organic compounds can be determined by

coulometric titration with bromine
RCH=CH2 + Br2 ~ RCHBr-CH2Br
where bromine is generated by oxidation of bromide ion and the
endpoint is determined biamperometrically with two small
platinum electrodes polarized by 100 mV. 10 mL ofa solution of an
unsaturated acid was placed in a coulometric cell, sulfuric acid
and potassium bromide was added and a constant current of 18.5
rnA passed. The current between the detector electrodes was
measured as a function of time with the following results:
tis 250 200 270 280 290 300

ilJlA o 6 15 24 3i 43
Problems 425
What was the molar concentration of the unsaturated acid?
Assume mono-unsaturation.
7.12 Suppose that copper is deposited on a platinum electrode at a

cathode potential of -0.30 V (us. s.c.e.) and an initial current
density of j = 100 A m- 2. Early in the experiment, the electrode
surface is mostly platinum. If the exchange current density for
the reduction ofH+(aq) on Pt is 10 A m- 2, (lapp = 1.5, at pH 0, what is
the initial current density contribution from hydrogen reduction if
the pH is 1.0? What is the current efficiency of the copper
reduction process? When the current density has dropped to 10 A
m- 2, the cathode is covered with copper. If the exchange current
density for H+ reduction on copper is 2 rnA m-3, (lapp = 0.5, at pH 1,
what is the hydrogen reduction current? What is the current
efficiency for copper reduction?
7.13 An industrial electrolysis process has raw material costs of $1.00

per kilogram of product and capital and labor costs of $5.00 per
electrolysis cell per day (independent of whether any product is
made). The cell potential (for i > 1 A) is
E = Eo + Plog i + iR
where 13 = 1.0 V, R = 0.10 n, and Eo = 1.50 V. The cost of electric
power is $0.10 per kilowatt-hour. What is the optimum current
per cell for minimum cost per kilogram of product? Assume that
50 g of product is produced per Faraday of charge passed, that the
current efficiency is 100%, and that the plant runs 24 hours a day.
What is the production per cell per day? What are the costs per
kilogram for raw materials, capital and labor, and electrical
energy? If the cost per kilowatt-hour doubled, what would be the
optimum current? What would the total cost of product be?
7.14 Two common ways of protecting steel from corrosion are tin
plating and zinc coating. Typical examples of this approach are
"tin cans" and galvanized buckets. Discuss qualitatively the
chemical reactions which would occur if pieces of tin-plated steel
and zinc-coated steel were scratched to expose the steel and placed
in a wet oxygen-rich environment. Indicate clearly where
(relative to the scratch) each reaction would take place.
7.15 Consider the reaction of zinc with 1 M hydrochloric acid if[Zn2+] =

1 M. The exchange current densities are
jQZn = 0.2 A m- 2 , J3Zn = 1.5
joH = 1 x 10-6 A m-2, (lH = 0.5
426 Electrolysis
for [Zn2+] = 1 M, pH 0, and 25°C. Assume that
alogljoZn lfa log[Zn2+] = 0.75.
(a) Compute the corrosion potential, E e , the zinc corrosion rate in
g m 2s· I and the rate of hydrogen gas evolution in mL m- 2s- I
(b) Corrosion chemists often express corrosion rates in units of
mm year-I. Given that the density of zinc is 7.14 g cm- 3 , convert
the corrosion rates to units of mm yearl.
1
)(
cz BIBLIOGRAPHY
w
a..
a..
<C
A. Introductory Electrochemistry Texts

Al C. N. Reilley and R. W. Murray, Electroanalytical Principles, New
York: Interscience, 1963.
A2 E. H. Lyons, Jr., Introduction to Electrochemistry, Boston: D. C.
Heath, 1967.
A3 B. B. Damaskin, The Principles of Current Methods for the Study of
Electrochemical Reactions, New York: McGraw-Hill, 1967.
A4 J. B. Headridge, Electrochemical Techniques for Inorganic
Chemists, London: Academic Press, 1969. .
A5 J. Robbins, Ions in Solution (2): An Introduction to Electrochemistry,
London: Oxford University Press, 1972.
A6 W. J. Albery, Electrode Kinetics, London: Oxford University Press,
lW5.
A7 N. J. Selley, Experimental Approach to Electrochemistry, New York:
John Wiley, 1976.
A8 D. R. Crow, Principles and Applications of Electrochemistry, 2nd ed,
London: Chapman and Hall, 1979.
A9 D. Hibbert and A. M. James, Dictionary of Electrochemistry, 2nd ed,
AlO T. Riley and C. Tomlinson, Principles of Electroanalytical Methods,
All J. Koryta, Ions, Electrodes and Membranes, 2nd ed, New York:
John Wiley, 1991.
B. More Advanced Texts

Bl P. Delahay, New Instrumental Methods in Electrochemistry, New
York: Interscience, 1954.
B2 H. S. Hamed and B. B. Owen, Physical Chemistry of Electrolytic
Solutions, New York: Reinhold,1958.
B'3 G. Charlot, J. Badoz-Lambling, and B. Tremillon, Electrochemical
Reactions, Amsterdam: Elsevier, 1962.
427
428 Appendix 1
B4 D. A. MacInnes, The Principles of Electrochemistry, New York:
Dover, 1966 (corrected reprint of 1947 edition).
B5 G. Kortum, Treatise on Electrochemistry, 2nd ed, Amsterdam:
Elsevier, 1965.
136 B. E. Conway, Theory and Principles of Electrode Processes, New
York: Ronald Press, 1965.
B7 K. J. Vetter, Electrochemical Kinetics, New York: Academic Press,
1967.
B8 J. O'M. Bockris and A. Reddy, Modern Electrochemistry, New York:
Plenum Press, 1970.
139 R. A. Robinson and R. H. Stokes, Electrolyte Solutions, 2nd ed (rev),
London: Butterworths, 1970.
B10 J. S. Newman, Electrochemical Systems, Englewood Cliffs, NJ:
Prentice-Hall,1972.
Bll L. I. Antropov, Theoretical Electrochemistry, Moscow: Mir, 1972.
B12 A. J. Bard and L. R. Faulkner, Electrochemical Methods, New
York: John Wiley, 1980.
B13 R. Greef, R. Peat, L. M. Peter, D. Pletcher, and J. Robinson,
Instrumental Methods in Electrochemistry, Chicester: Ellis
Horwood, 1985.
B14 J. Koryta and J. Dvorak, Principles of Electrochemistry, New York:
John Wiley, 1987.
c. Specialized Books and Monographs
Cl W. Ostwald, Electrochemistry, History and Theory, Leipzig: Veit,
1896. Republished in English translation for the Smithsonian
Institution, New Delhi: Amerind Publishing, 1980.
C2 V. G. Levich, Physiochemical Hydrodynamics, Englewood Cliffs,
NJ: Prentice-Hall, 1962.
C3 R. N. Adams, Electrochemistry at Solid Electrodes, New York:
Marcel Dekker, 1969.
C4 C. K. Mann and K. K. Barnes, Electrochemical Reactions in
Nonaqueous Systems, New York: Marcel Dekker, 1970.
C5 A. Weissberger and B. W. Rossiter, eds, Physical Methods of
Chemistry, Vol. 1 (Techniques of Chemistry), Parts IIA and lIB
(Electrochemical Methods), New York: John Wiley, 1971.
C6 J. S. Mattson, H. B. Mark, Jr., and H. C. MacDonald, Jr.,
Electrochemistry; Calculations, Simulation, and Instrumentation
(Computers in Chemistry and Instrumentation, Vol. 2), New York:
Marcel Dekker, 1972.
C7 Yu. V. Pleskov and V. Yu. Filinovskii, The Rotating Disc Electrode,
New York: Consultants Bureau, 1976.
CB J. O'M. Bockris and S. U. M. Khan, Quantum Electrochemistry,
New York: Plenum Press, 1979.
Bibliography 429
C9 A. J. Fry and W. E. Britton, eds, Topics in Organic Electrochemistry,

New York: Plenum Press, 1986.
C10 R. J. Gale, ed, Spectroelectrochemistry, Theory and Practice, New
York: Plenum Press, 1988.
Cll H. D. Abrwia, ed, Electrochemical Interfaces: Modern Techniques
for In-Situ Interface Characterization, New York: VCH Publishers,
1991.
C12 J. O'M. Bockris and S. U. M. Khan, Surface Electrochemistry, New
D. Electroanalytical Methods
D1 1. M. Kolthoff and J. J. Lingane, Polarography, 2nd ed, New York:
Interscience, 1952.
D2 J. J. Lingane, Electroanalytical Chemistry, 2nd ed, New York:
Interscience, 1958.
D3 W. C. Purdy, Electroanalytical Methods in Biochemistry, New
York: McGraw-Hill,1965.
D4 L. Meites, Polarographic Techniques, 2nd ed, New York: John
Wiley, 1965.
D5 J. Heyrovsky and J. Kuta, Principles of Polarography, New York:
D6 H. Rossotti, Chemical Applications of Potentiometry, Princeton, NJ:
Van Nostrand, 1969.
D7 R. G. Bates, Determination of pH: Theory and Practice, 2nd ed, New
DB Z. Galus, Fundamentals of Electrochemical Analysis, Chichester:
Ellis Harwood, 1976.
D9 G. Dryhurst, Electrochemistry of Biological Molecules, New York:
D10 C. C. Westcott, pH Measurements, New York: Academic Press,
1978.
Dll J. Vesely, D. Weiss, and K. Stulik, Analysis with Ion-Selective
Electrodes, Chichester: Ellis Horwood, 1978.
D12 A. M. Bond, Modern Polarographic Methods in Analytical
Chemistry, New York: Marcel Dekker, 1980.
D13 J. A. Plambeck, Electroanalytical Chemistry, New York: John
Wiley, 1982.
D14 J. Koryta and K. Stulik, Ion-Selective Electrodes, 2nd ed, London:
Cambridge University Press, 1983.
E. Organic Electrosynthesis
El M. R. Rifi and F. H. Covitz, Introduction to Organic
Electrochemistry, New York: Marcel Dekker, 1974.
430 Appendix 1
E2 N. L. Weinberg, ed, Technique of Electroorganic Synthesis

(Technique of Chemistry, Vol. V), New York: John Wiley, 1974 (Part
1), 1975 (Part Il).
E3 D. K. Kyriacou, Basics of Electroorganic Synthesis, New York: John
Wiley, 1981.
FA K. Yoshida, Electrooxidation in Organic Chemistry, New York:
John Wiley, 1984.
E5 T. Shono, Electroorganic Chemistry as a New Tool in Organic
Synthesis, Berlin: Springer-Verlag, 1984.
E6 A. J. Fry, Synthetic Organic Electrochemistry, 2nd ed, New York:
John Wiley, 1989.
E7 H. Lund and M. M. Baizer, eds, Organic Electrochemistry, 3rd ed,
New York: Marcel Dekker, 1991.
E8 T. Shono, Electroorganic Synthesis, San Diego: Academic Press,
1991.
F. Experimental Methods
F1 D. J. G. Ives and G. J. Janz, eds, Reference Electrodes, Theory and
Practice, New York: Academic Press, 1961.
F2 W. J. Albery and M. L. Hitchman, Ring-Disc Electrodes, Oxford:
Clarendon Press, 1971.
F3 D. T. Sawyer and J. L. Roberts, Jr., Experimental Electrochemistry
for Chemists, New York: John Wiley, 1974.
F4 E. Gileadi, E. Kirowa-Eisner, and J. Penciner, Interfacial
Electrochemistry--An Experimental Approach, Reading, MA:
Addison-Wesley, 1975.
F5 D. D. MacDonald, Transient Techniques in Electrochemistry, New
F6 P. T. Kissinger and W. R. Heineman, eds, Laboratory Techniques in
Electroanalytical Chemistry, New York: Marcel Dekker, 1984.
G. Technological Applications of Electrochemistry
G1 J. O'M. Bockris and S. Srinivasan, Fuel Cells: Their
Electrochemistry, New York: McGraw-Hill,1969.
G2 C. L. Mantell, Batteries and Energy Systems, New York: McGraw-
Hill,1970.
G3 F. A. Lowenheim, ed, Modern Electroplating, New York: John
Wiley, 1974.
G4 S. W. Angrist, Direct Energy Conversion, Boston: Allyn and Bacon,
1976.
G5 V. S. Bagotzky and A. M. Skundin, Chemical Power Sources, New
York: Academic Press, 1980.
G6 D. Pletcher, Industrial Electrochemistry, London: Chapman and
Hall,1982.
Bibliography 431
G7 N. L. Weinberg and B. V. Tilak, eds, Technique of Electroorganic
Synthesis (Technique of Chemistry, Vol. V), Part III, New York:
John Wiley, 1982.
G8 H. H. Uhlig and R. W. Revie, Corrosion and Corrosion Control, New
G9 H. V. Ventatasetty, ed, Lithium Battery Technology, New York:
John Wiley, 1984.
G10 R. E. White, ed, Electrochemical Cell Design, New York: Plenum
Press, 1984.
Gll Z. Nagy, Electrochemical Synthesis of Inorganic Compounds, New
H. Electrochemical Data
HI W. M. Latimer, Oxidation Potentials, 2nd ed, Englewood Cliffs, NJ:
Prentice-Hall, 1952.
H2 B. E. Conway, Electrochemical Data, Amsterdam: Elsevier, 1952.
H3 R. Parsons, Handbook of Electrochemical Data, London:
Butterworths, 1959.
H4 A. J. de Bethune and N. A. S. Loud, Standard Aqueous Electrode
Potentials and Temperature Coefficients at 25°C, Skokie, IL:
Hampel, 1964.
H5 M. Pourbaix, Atlas of Electrochemical Equilibria, New York:
Pergamon Press, 1966.
H6 G. J. Janz and R. P. T. Tomkins, Nonaqueous Electrolytes
Handbook, New York: Academic Press, 1972.
H7 A. J. Bard and H. Lund, eds, The Encyclopedia of the
Electrochemistry of the Elements, New York: Marcel Dekker, 1973.
H8 L. Meites and P. Zuman, Electrochemical Data. Part I. Organic,
Organometallic, and Biochemical Systems, New York: John Wiley,
1974.
H9 D. Dobos, Electrochemical Data, Amsterdam: Elsevier, 1975.
H10 G. Milazzo and S. Caroli, Tables of Standard Electrode Potentials,
Hll L. Meites and P. Zuman, eds, CRC Handbook Series in Organic
Electrochemistry, Boca Raton, FL: CRC Press, 1977-.
H12 L. Meites, P. Zuman, E. B. Rupp and A. Narayanan, eds, CRC
Handbook Series in Inorganic Electrochemistry, Boca Raton, FL:
CRC Press, 1980-.
H13 A. J. Bard, R. Parsons, and J. Jordan, eds, Standard Potentials in
Aqueous Solution, New York: Marcel Dekker, 1985.
H14 A. J. Bard, R. Parsons, and J. Jordan, eds, Oxidation-Reduction
Potentials in Aqueous Solution, Oxford: Blackwell, 1986.
432 Appendix 1
1. Review Series
11 J. O'M. Bockris and B. E. Conway, eds, Modern Aspects of
Electrochemistry, New York: Plenum Press, from 1954.
12 P. Delahay (Vols. 1-9), C. W. Tobias (Vols. 1- ), and H. Gerischer
(Vols. 10- ), Advances in Electrochemistry and Electrochemical
Engineering, New York: John Wiley, from 1961.
13 A. J. Bard, ed, Electroanalytical Chemistry, New York: Marcel
Dekker, from 1966.
14 E. B. Yeager and A. J. Salkind, eds, Techniques of Electrochemistry,
New York: John Wiley, from 1972.
15 G. J. Hills (Vols. 1-3) and H. R. Thirsk (Vol. 4- ), Senior Reporters,
Electrochemistry, A Specialist Periodical Report, London: Royal
Society of Chemistry, from 1971.
16 Analytical Chemistry, Fundamental Annual Reviews (April issue
of even-numbered years), Washington: American Chemical
Society.
2
-zc>< SYMBOLS AND UNITS
w
a..
a..
til(
Table A.t The International System (SI) of Units
Physical Quantity Unit Symbol

Fundamental units:
Length meter m
mass kilogram kg
time second s
electric current ampere A
temperature kelvin K
amount of substance mole mol
luminous intensity candela cd
Derived units:
force newton N (kg m s-2)
energy joule J (Nm)
power watt W (J s-l)
pressure pascal Pa (N m- 2)
electric charge coulomb C (A s)
electric potential volt V (J C-l)
electric resistaunce ohm n (V A-l)
electric conductance siemens S (A V-I)
electric capacitance farad F (C V-I)
frequency hertz Hz (s-l)
433
434 Appendix 2
Table A.2 Values of Physical Constants
Constant Symbol Value

pennittivity of free space €() 8.8541878 x 10-12 C2J-1m-1
electronic charge e 1.602189 x 10- 19 C
Avogadro's number NA 6.02204 x 1023 mol- 1
Faraday constant F 96484.6 C mol- l
gas constant R 8.3144 J moI-lK-1
Boltzmann constant k 1.38066 x 10-23 J K-l
Planck constant h 6.62618 x 10-34 J s
gravitational acceleration g 9.80665 m s-2
Table A.3 List of Symbols
Symbol Name Units

a activity none
a radius m
A area m2
Ci molar concentration, species i mol L-1 (M)
mol m- 3 (mM)
C differential capacity F m-2
d density kgm- 3
D·I diffusion coefficient, species i m 2s- 1
E energy J
E cell potential V
E1I2 half-wave potential V
E electric field strength V m- l
Ii frictional coefficient, species i kg s-1
F force N
G Gibbs free energy J mol- l
H enthalpy J mol- l
electric current A
Symbols and Units 435
Symbol Name Units

iD diffusion-limited current A
iL limiting current A
1 ionic strength mol L-1
1 a.c. current amplitude A
ID diffusion current constant rnA mM-1
(mg s-1)-213s -1I6
j volume flux m 3s- 1
j current density A m- 2
J molar flux density mol m- 2 s- 1
k rate constant variable
kD mass-transport rate constant m s-l
kij potentiometric selectivity coefficient none
K equilibrium constant none
L length m
m mass kg
mi molal concentration, species i mol kg-I
M molecular weight g mol- 1
n number of moles mol
ni kinetic order, species i none
N number of molecules none
p pressure bar (105 Pa)
q heat J
Q electric charge C
r,R radial distance m
ro microelectrode radius m
R resistance n
S entropy J mol-1K-l
t time s
ti transference number, species i none
T temperature K
436 Appendix 2
Table A.3 List of Symbols (continued)
Symbol Name Units

Ui mobility, species i m2V- 1s- 1
U mass flow rate kg s-l
U internal energy J mol- 1
v velocity m s-l
V volume m3
v potential scan rate V s-l
w work J
x distance m
XA ion atmosphere thickness m
(Debye length)
XD diffusion layer thickness m
XH hydrodynamic distance parameter m
XR reaction layer thickness m
Xi mole fraction, species i none
Zi charge, species i none
Z impedance n
(l cathodic transfer coefficient none
(l degree of dissociation none
C1.i electrokinetic coefficient variable
(l,~,y,o stoichiometric coefficients none
(l,~ phase labels none
~ anodic transfer coefficient none
y surface tension Nm- 1
'W activity coefficient, species i none
(molar scale)
Yim activity coefficient, species i none
(molal scale)
y,.x activity coefficient, species i none
(mole fraction scale)
Y± mean ionic activity coefficient none
Symbols and Units 437
Symbol Name Units

£ dielectric constant none
Tl coefficient of viscosity kg m-1s- 1 (Pa-s)
Tl overpotential V
l'} polar angle rad
e exp[F(E - EO)IRTJ none
1C conductivity Sm- 1
A reaction zone parameter none
A molar conductivity S m 2mol- 1
Il chemical potential J mol- 1
v kinematic viscosity m 2 s- 1
Vi moles of ion i per mole of salt none
Vi stoichiometric coefficient, species i none
X (DoIDR) 112 none
p resistivity Om
p space charge density C m- 3
a surface charge density C m- 2
't characteristic time s
4> current efficiency none
q> azimuthal angle, phase angle rad
<l> electric potential V
CJ) angular frequency rad s-1
~ zeta potential V
ELECTROCHEMICAL
DATA
Table A.4 Standard Reduction Potentials at 25°C.
Main group elements:

Half-Cell Reaction
2 H + + 2 e- -+ H2(g) 0.0000
Li+ + e- -+ Li(s) -8.045
Na+ + e- -+ Na(s) -2.714
K+ + e- -+ K(s) -2.925
Rb+ + e- -+ Rb(s) -2.925
Cs+ + e- -+ Cs(s) -2.923
Be2+ + 2 e- -+ Be(s) -1.97
Mg2+ + 2e- -+ Mg(s) -2.356
Ca2+ + 2 e- -+ Ca(s) -2.84
Ba2+ + 2 e- -+ Ba(s) -2.92
Al3+ + 3 e- -+ Al(s) -1.67
C02(g) + 2 H+ + 2 e- -+ CO(g) + H20 ~.106
C02(g) + 2 H+ + 2 e- -+ HCOOH ~.199
2 C02<g) + 2 H+ + 2 e- -+ H2C204 ~.475
Pb2+ + 2 e- -+ Pb(s) ~.1251
Pb02(S) + 4 H+ + 2 e- -+ Pb2+ + 2 H20 1.468
N03-+3H++2e- -+ HN02+H20 0.94
N03- + 4 H+ + 3 e- -+ NO(g) + 2 H20 0.96
N03- + 10 H+ + 8 e- -+ NH4+ + 3 H20 0.875
PO(OH)3 + 2 H+ + 2 e- -+ HPO(OHh + H20 ~.276
HPO(OH)2 + 2 H+ + 2 e- -+ H2PO(OH) + H20 ~.499
438
Electrochemical Data 439
Half-Cell Reaction EON

HPO(OH)2 + 3 H+ + 3 e- ~ P(s) + 3 H2O -0.454
P(s)+3H++3e' ~ PH3 -0.111
AsO(OH)s + 2 H+ + 2 e- ~ As(OH)a + H2O 0.560
As(OH)a + 3 H+ + 3 e- ~ As(s) + 3 H2O 0.240
As(s) + 3 H + + 3 e- ~ AsHS(g) -0.225
02(g) + H+ + e- ~ H02 -0.125
02(g) + 2 H+ + 2 e- ~ H202 0.695
H202 + H+ + e- ~ HO· + H2O 0.714
H202 + 2 H+ + 2 e- ~ 2 H2O 1.763
&.208 2- + 2 e- ~ 2 8042- 1.96
8042- + H20 + 2 e- ~ 8Os2- + 2 OH- -0.94
28042- +4H++ 2 e- ~ 82062- + 2 H20 -0.25
2 802(aq) + 2 H+ + 4 e- ~ 820S2- + H2O 0.40
802(aq) + 4 H+ + 4 e- ~ 8(s) + 2 H2O 0.50
84062- + 2 e- ~ 2 &.20a2- 0.08
8(s) + 2 H+ + 2 e- ~ H2S(aq) 0.14
F2(g) + 2 e- ~ 2 F- 2.866
CI04- + 2 H+ + 2 e- ~ CIOs- + H2O 1.201
CIOs- + 3 H+ + 2 e- ~ HCI02 + H2O 1.181
CIOs- + 2 H+ + e- ~ CI02 + H2O 1.175
HCI02+2H++2e- ~ HOCI+H20 1.701
2 HOCI + 2 H+ + 2 e- ~ CI2(g) + 2 H2O 1.630
CI2(g) + 2 e- ~ 2 CI- 1.35828
CI2(aq) + 2 e- ~ 2 CI- 1.396
Br04- + 2 H+ + 2 e- ~ BrOs- + H2O 1.853
2 BrOs- + 12 H+ + 10 e- ~ Br2(l) + 6 H2O 1.478
2 HOBr + 2 H+ + 2 e- ~ Br2(l) + 2 H2O 1.004
Br2(l) + 2 e- ~ 2 Br 1.0652
Br2(aq) + 2 e- ~ 2 Br 1.0874
10(OH)5 + H+ + e- ~ 103- + 3 H2O 1.60
2 lOs- + 12 H+ + 10 e- ~ 12(S) + 6 H2O 1.20
440 Appendix 3
Table A.4 Standard Reduction Potentials at 25°C (continued)
Main group elements:

Half-Cell Reaction
2HOI+2H++2e- ~ lis)+2H20 1.44
lis) + 2 e- ~ 2 l- 0.5355
Is- + 2 e- ~ 31- 0.536
liaq)+2e- ~ 21- 0.621
Transition and post·transition elements:

Half-Cell Reaction
V02+ + 2 H+ + e- ~ V02+ + H20 1.000
V02+ + 2 H+ + e- ~ \13+ + H20 0.337
\13+ + e- ~ V2+ -0.255
V2+ + 2 e- ~ V(s) -1.13
Cl"2072- + 14 H+ + 6 e- ~ 2 Cr3+ + 7 H~ 1.38
Cr3+ + e- ~ Cr2+ -0.424
Cr2+ + 2 e- ~ Cr(s) -0.90
Mn04- + e- ~ Mn042- 0.56
Mn04- + 8 H+ + 5 e- ~ Mn2+ + 4 H20 1.51
Mn02(S) + 4 H+ + 2 e- ~ Mn2+ + 2 H20 1.23
Mn3+ + e- ~ Mn2+ 1.5
Mn2+ + 2 e- ~ Mn(s) -1.18
Fe3+ + e- ~ Fe2+ 0.711
Fe(phen)3+ + e- ~ Fe(phen)2+ 1.13
Fe(CN)63- + e- ~ Fe(CN)64- 0.361
Fe(CN)64- + 2 e- ~ Fe(s) + 6 CN- -1.16
Fe2+ + 2e- ~ Fe(s) -0.44
Co3+ + e- ~ Co2+ 1.92
CO(NH3)63+ + e- ~ CO(NH3)62+ 0.058
Co(phen)33+ + e· ~ Co(phen)3 2 + 0.327
Co(C204)33- + e- ~ CO(C204)34- 0.57
Half-Cell Reaction
C02+ + 2 e- --+ Co(s) ....().2:l7
NiOis) + 4 H+ + 2 e- --+ Ni2+ + 2 H20 1.593
Ni2+ + 2 e- --+ Ni(s) ....().257
Ni(OH>2(s) + 2 e- --+ Ni(s) + 2 OH- ....().72
Cu2+ + e- --+ Cu+ 0.159
CuCI(s) + e- --+ Cu(s) + CI- 0.121
Cu2+ + 2 e- --+ Cu(s) 0.340
Cu(NHa)42+ + 2 e- --+ Cu(s) + 4 NHa ....().oo
Ag2+ + e- --+ Ag+ 1.980
Ag+ + e- --+ Ag(s) 0.7991
AgCI(s) + e- --+ Ag(s) + CI- 02223
Zn2+ + 2 e- --+ Zn(s) ....().7626
Zn(OH)42- + 2 e- --+ Zn(s) + 4 OH- -1.285
Cd 2+ + 2 e- --+ Cd(s) ""().4025
2 Hg2+ + 2 e- --+ Hg22+ 0.9110
Hg22+ + 2 e- --+ 2 Hg(l) 0.7960
Hg2CI2(S) + 2 e- --+ 2 Rg(l) + 2 CI- 0.26816
Data from Bard, Parsons, and Jordan (RI3).

442 Appendix 3
Table A.S Biochemical Reduction Potentials
Half-Cell Reaction E'/V

Reduction of a carboxyl group to an aldehyde:
1,3-diphosphoglycerate + 2 e- -0.286
-+ 3-phosphoglyceraldehyde + HP04 2-
acetyl-CoA + 2 H+ + 2 e- ~ acetaldehyde + coenzyme A -0.412
oxalate + 3 H+ + 2 e- ~ glyoxal ate -0.462
gluconate + 3 H+ + 2 e- ~ glucose -0.47
acetate + 3 H + + 2 e- ~ acetaldehyde -0.598
Reduction of a carbonyl group to an alcohol:
dehydroascorbic acid + H+ + 2 e- ~ ascorbate 0.077
glyoxylate + 2 H + + 2 e- ~ glycolate -0.090
hydroxypyruvate + 2 H + + 2 e- ~ glycerate -0.158
oxaloacetate + 2 H + + 2 e- ~ malate -0.166
pyruvate + 2 H+ + 2 e- ~ lactate -0.190
acetaldehyde + 2 H + + 2 e- ~ ethanol -0.197
acetoacetate + 2 H+ 2 e- ~ ~-hydroxybutyrate -0.349
Carboxylation:
pyruvate + C02 (g) + H+ + 2 e- ~ malate -0.330
a-ketoglutarate + C02(g) + H+ + 2 e- ~ iso-citrate -0.363
succinate + C02(g) + 2 H+ + 2 e- ~ a-ketoglutarate + H20 -0.673
acetate + C02(g) + 2 H+ + 2 e- ~ pyruvate + H20 -0.699
Reduction of a carbonyl group with formation of an amino group:
oxaloacetate + NH4+ + 2 H+ + 2 e- ~ aspartate + H20 -0.107
pyruvate + Nl4+ + 2 H+ + 2 e- ~ alanine + H20 -0.132
a-ketoglutarate + NH4+ + 2 H+ + 2 e- ~ glutamate + H20 -0.133
Reduction of a carbon-carbon double bond:
crotonyl-CoA + 2 H+ + 2 e- ~ butyryl-CoA 0.187
fumarate + 2 H+ + 2 e- ~ succinate 0.031
Half-Cell Reaction E'/V

Reduction of disulfide:
cystine + 2 H+ + 2 e- ~ 2 cysteine -D.340
glutathione climer + 2 H+ + 2 e- ~ 2 glutathione -D.340
Other reductions of biochemical interest:
02(g)+4H++4e- ~ 2 H20 0.816
cytochrome c (Fe3 +) + e- ~ cytochrome c (Fe 2+) 0.25
FAD+ + H+ + 2 e- ~ FADH -D.20
NAD++H++2e- ~ NADH -D.320
2 H + + 2 e- ~ H2(g) -D.414
Reduction potentials at 25°C, pH 7 standard state; data from H. A. Krebs,

H. L. Kornberg, and K. Burton, Erg. Physiol. 1957,49,212.
444 Appendix 3
Table A.6 Some Formal Reduction Potentials
EO'/V
Couple 1MHCI04 1MHCl
Ag(l)lAg( 0) 0.792 0.228 0.77
As(V)1As(III) 0.577 0.577
Ce(IV)/Ce(III) 1.70 1.28 1.44
Fe(IIl)lFe(II) 0.732 0.700 0.68
Ag(l)lAg( 0) 0.792 0.228 0.77
As(V)1As(III) 0.577 0.577
Ce(IV)/Ce(III) 1.70 1.28 1.44
Fe(IIl)lFe(II) 0.732 0.700 0.68
H(I)IH(O) -0.005 -0.005
Hg(II)IHg(1) 0.907
Hg(l)IHg(O) 0.776 0.274 0.674
Mn(IV)/Mn(II) 1.24
Pb(II)lPb(O) -0.14 -0.29
Sn(lV)/Sn(II) 0.14
Sn(Il)/Sn( 0) -0.16
Data from E. H. Swift and E. A. Butler, Quantitative

Measurements and Chemical Equilibria, San Francisco:
Freeman, 1972.
Table A. 7 Reference Electrode Potentials
Electrode EO'IV (dE/dT)/mV K-l

Calomel (0.1 M KCD 0.336 -0.08
Calomel (1.0 M KCD 0.283 -0.29
Calomel (satd. KCl) 0.244 -0.67
Ag/AgCl (3.5 M KCl) 0.205 -0.73
Ag/AgCI (satd. KCl) 0.199 -1.01
Table A.8 Molar Ionic Conductivities
Ion AD Ion AD Ion AD

H+ 349.8 Pb2+ 139.0 103- 40.5
Li+ 38.7 Mn 2+ 107. 104- 54.6
Na+ 50.1 Fe2+ 107. Mn04- 62.8
K+ 73.5 Co2+ 110. HC03- 44.5
Cs+ 77.3 Ni 2+ lOB. H2P04- 36.
Nli4+ 73.6 Al3+ 189. HC02- 54.6
(CH3)4N+ 44.9 Cr3+ ~1. CH3C02- 40.9
(C2HS)4N+ 32_7 Fe3+ ~. C2HSC02- 35.8
(C3H 7)4N+ .23.4 OH- 199.2 C6HSC0 2- 32.4
Ag+ 61.9 F- 55.4 C032- 138.6
Mg2+ 106.1 Cl- 76.3 8042- 160.0
Ca2+ 119.0 Br 78.1 &P32- 174.B
Sr2+ 118.9 I- 76.B Cr042- 170.
Ba2+ 127.3 CN- 7B. HP042- 114.
Cu2+ 107.2 N02- 72. C20 42- 148.3
Zn 2+ 105.6 N03- 71.5 P30s3- 25O.B
Cd2+ 108.0 ClO3- 64.6 Fe(CN)63- 302.7
Hg2+ 127.2 ClO4- 67.4 Fe(CN)64- 442.
Conductivities from Robinson and Stokes (B9) and Dobos (H9) in units of
10- 4 S m 2mol- 1, aqueous solutions at infinite dilution, 25 DC.
446 Appendix 3
Table A.9 Solvent Properties
Liquid Vapor Dielectric

Solvent Range/oC Pressure b Constant Viscosity
water 0 3.2 78.4 0.89
propylene carbonate (PC) -49 to 242 0.0 64.4 2.5
dimethylsulfoxide 19 to 189 0.1 46.7 2.00
(DMSO)
N ,N-dimethylformamide -60 to 153 0.5 36.7 0.80
(DMF)
acetonitrile -44 to 82 11.8 37.5 0.34
nitromethane -29 to 101 4.9 35.9 0.61
methanol -98 to 65 16.7 32.7 0.54
hexamethylphosphor- 7to233 0.01 30.d 3.47d
amide (HMPA)
ethanol -114 to 78 8.0 24.6 1.08
acetone -95 to 56 24.2 20.7 0.30
dichloromethane -95 to 40 58.1 8.9 0.41
trifluoroacetic acid -15 to 72 14.4 8.6 0.86
tetrahydrofuran (THF) -108 to 66 26.3 7.6 0.46
1,2-dimethoxyethane -58 to 93 10.0 72 0.46
(glyme, DME)
acetic acid 17 to 118 2.0 6.2d 1.13
p-dioxane 12 to 101 4.9 22 1.2
a Data at 25°C from J. A. Riddick and W. B. Bunger, Organic Solvents,

3rd ed, New York: Wiley-Interscience, 1970.
b Vapor pressure in units of kPa.
C Viscosity in units of 10-3 kg m-ls-l.
d 20°C.
Table A.1O Potential Range for Some Solutions
Potential RangeN (us. s.c.e.)

Solvent Electrolyte Pt Hg
Propylene carbonate E4NCI04 1.7 to-1.9 0.5to-2.5
Dimethylsulfoxide NaC104 0.7 to-1.8 0.6to-2.9
E4NCI04 0.7 to-1.8 0.2to-2.8
BU4NI -0.4to-2.8
Dimethylformamide NaCI04 1.6to-1.6 0.5to-2.0
E4NCI4 1.6to-2.1 0.5to-3.0
E4NBF4 -to-2.7
BU4CI04 1.5 to-2.5 0.5to-3.0
BU41 -0.4 to-3.0
Acetonitrile NaCI04 1.8to-1.5 0.6to-1.7
E4NCI04 0.6to-2.8
E4NBF4 2.3 to- -to-2.7
BU4NI -O.6to-2.8
BU4PF 6 3.4to-2.9
Acetone NaCI04 1.6 to-
E4NCI04 -to-2.4
Dichloromethane BU4NCI04 1.8 to-1.7 0.8to-1.9
BU4NI 0.2to-1.7 -O.5to-1.7
1,2-Dimethoxyethane BU4NCI04 0.6to-2.9
Data from C. K. Mann, Electroanal. Chem. 1969,3, 57.

4
-cz
><
LAPLACE TRANSFORM
W
Q. METHODS
Q.
cc
The method of Laplace transforms provides a powerful aid to the

solution of differential equations. l The method is particularly useful in
solving the coupled partial differential equations which are encountered
in electrochemical diffusion problems. Here we will introduce the
technique and demonstrate the method by deriving a few of the results
quoted in the text.
Laplace Transformations
The Laplace transform of a function F(t) is defined by
f(s) = L- F(t) exp(-st) dt (A. I)
Not all functions possess a Laplace transform. Clearly, F(t) must be

finite for finite t and F(t) exp(-st) must go to zero as t ~ The Laplace
00.
transformation can be thought of as an operation in linear algebra:

f(s) = L[F(t)]
which is reversible by the inverse operation
F(t) = L-llfls)]
The Laplace transformation is a linear operation, that is, sums or
differences of functions are transformed as
L[F(t) + G(t)] = f(s) + g(s) (A.2)
Multiplicative constants are unaffected by Laplace transformation:
L[a F(t)] = a f(s) (A.3)
1 F. E. Nixon, Handbook of Laplace Transformations: Tables and Examples,

Englewood Cliffs, NJ: Prentice-Hall, 1960; R. V. Churchill, Modern Operational
Mathematics in Engineering, 2nd ed, New York: McGraw-Hill, 1963; P. A.
McCollum and B. F. Brown, Laplace Transform Tables and Theorems, New York:
Holt, Rinehart, and Winston, 1965; M. G. Smith, Laplace Transform Theory,
London, D. Van Nostrand, 1966.
448
Laplace Transform Methods 449
The Laplace tranform of a constant is
L(a) = als
Functions of variables other than t behave as constants in the
transformation
L[H(x)] = H(x)ls
The utility of Lapace transforms in the solution of differential equations
is that the transform of a derivative is a simple function
L[dF(t)/dt] = s f{s) - F(O) (AAa)
L[d 2F(t)/dt 2 ] = s2f{s) - s F(D) - (dFldt)o (A4b)
A short selection of Laplace transforms are found in Table All.
Two properties of the Laplace transformation are sometimes useful
in finding the inverse transform. The shift theorem allows the zero of s
to be displaced by a constant:
L-llfts + a)) = F(t) exp(~t) (A5)
The convolution theorem is useful when the inverse transformation of
f(s)cannot be found, but f(s) can be written as the product of two
functions, f{s) = g(s)h(s), the inverse transforms of which can be found.
If
G(t) = L-lfg(s)]
H(t) = L-l[h(s)]
then
(A.6)
The specific solution to a differential equation depends on the initial

and boundary conditions on the problem. The solution to a differential
equation using Laplace transform methods in general follows the steps:
(1) Transform the differential equation to remove derivatives with
respect to one of the variables. An ordinary differential equation
will then be an algebraic equation and a partial differential
equation with two independent variables will become an ordinary
differential equation.
(2) Transform the initial and boundary conditions.
(3) Solve the resulting system of algebraic equations or ordinary
differential equations, using the transformed boundary conditions
to evaluate constants.
(4) Take the inverse transform to obtain the solution to the original
differential equation.
450 Appendix 4
Table A.ll Some Laplace Transforms
F(t) f{s) F(t) f{s)
a (a constant) als sin at a

a 2 + s2
t s-2 cos at s.
a 2 + s2
t n- 1 _a_
s-n sinh at s2-a 2
(n -I)!
~
l/'fii s-1I2 cosh at s2-a 2
_I_
2fiTiC s-3/2 exp at s-a
F(t) f{s)
exp(at) - exp(bt) 1
a-b (s-a)(s-b)
ifI
exp(at) erfW -IS(s - a)
1
exp(at)[l- erfYat]
1- exp(at)[l- erfYat] s (-IS + ifI)

~ exp(-l/4at) exp (-Vs/a)
1tat 3
.b
,1tt
exp(-l/4at) (l/-IS) exp (-Vs/a)
1- erf(lJ2W) (1/s) exp (-V sial
Consider as an example the ordinary differential equation

d2F(t) = a 2F(t)
dx 2
with boundary conditions F(O) = 0, (dF/dx)o = a. Taking the Laplace
transform, using eqs (A.2) - (A.4), we have
s2f{s) - a + a 2f(s) = 0
from which we obtain
(ts) =
a
a 2 + s2
The inverse transform from Table A,11 gives
F(t) = sin at
We really didn't need a fancy method to solve this problem, but other
cases arise which are not quite so simple.
Solutions of the Diffusion Equation

Now let us apply Laplace transform methods to the solution of the
one-dimensional diffusion equation
aC(X,t) = D a2C(X,t) (A.7)
at ax2
We can do step (1) of the solution procedure in general. Writing the

Laplace transform of C(x,t) as c(x,s), the transformed diffusion equation
is
2
s C(X,s) _ C(X,O) = D a c(x,s) (A,B)
ax2
We need initial and boundary conditions to solve eq (A,B), and these differ
from one problem to another.
Derivation of eq (3.22). Let us start with the problem posed in
§3.3. We considered a solution layered on pure solvent so that the initial
condition was C =C* for x < 0, C = 0 for x > O. The boundary condition is
C -7 C* as x -7 - 0 0 , C -7 0 as x -7 +00. We will divide the problem into two
regimes, - 0 0 < X < 0 and 0 < x < +00, with the requirement that C(x,t) and
J(x,t) be continuous at x = 0 for t > O. Thus for x > 0, we have
D a2c(x,S) () 0
sex,s =
ax 2
the general solution to which is
c(x,s) =A(s) exp(-YslD) + B(s) exp(+YslDx) (A.9a)
where A(s) and B(s) are to be determined from the boundary conditions.
One of the boundary conditions requires c(x,s) -7 0 as x -7 00 so that B(s) =
O.
For x < 0, eq (A,B) gives
D ic'(x,s)
dX 2
s
,
C (x,s) +
C*
= °
452 Appendix 4
The general solution to this differential equation is

c'(x,s) = C*ls +A'(s) exp(-VsIDx) +B'(s) exp(+VsIDx) (A.9b)
The boundary condition requires c'(x,s) ~ C*ls as x ~ --00 so thatA'(s) = 0.
We now apply the continuity restraints to determine A(s) and B'(s). If
C(O,t) = C'(O,t), then C(O,S) = c'(O,s). Thus we have
C(O,S) =A(s)
c'(O,s) = C*ls + B'(s)
so that
A(s) - B'(s) = C*ls
The equal fluxes at x = °means that
dC(O,s) dC'(O,s)
---=
dX dX
Differentiating eqs (A.9) and setting x = 0, we have
-VsIDA(s) = +VsIDB'(s)
Thus
A(s) =-B'(s) =C*/2s
Equations (A.9) then become
C(x,s) = ~ exp (- fk Ixl) x>o
c'(x,s) = c;- - ~; eXP(-fklxl) x<o

Taking the inverse transforms, we have
C(x,t) =~[1-erf~] x>o
2 2Wt
C'(x,t) =C* - ~[1-

2
erf---L-]
2Wt
x<o
Since the error function is an odd function of the argument, i.e., erf(-'I') =
-erf('I'), we see that these two functions are in fact identical:
C(x,t) =C*2 [1- erf 2,Dt
.~]
(3.22)
Derivation of eqs.(4.2). In typical electrochemical applications of

the diffusion equation, the concentrations of the diffusing species are
uniform at the beginning of the experiment, C(x,O) = C* and, at later
times, approach the initial concentration at sufficient distance from the
electrode, C(x,t) ~ C* as x ~ 00. With the electrode as x = 0, we need not
consider negative values for x. Thus eq (AB) is
2
D a c(x,s) ( ) C* - 0
s C x,s + -
ax 2
The solution consistent with the boundary condition (at x ~ 00) is
c(x,s) = 9- + A(s) exp (- VIi x) (A10)
where A(s) must be determined by the boundary condition at x = O. Ifwe

have two species, 0 and R, which are involved in an electrode process,
each transformed concentration will have the form of eq (A.10). If the
initial concentrations are Co(x,O) = Co*, CR(O,t) = 0, then eq (A.10) gives
Co(x,s) = Cf +A(s)exp(-ffox) (Alla)
CR(X,S) = B(s) exp (- ~ DR x) (Al1b)
For a reversible electrode process, the surface boundary conditions are:

(1) the concentration ratio atx = 0, governed by the Nernst equation
Co(O,t) = e = exp nF(E - EO) (A.12a)
CR(O,t) RT
and (2) the continuity restriction
Jo(O,t) = - JR(O,t)
or
-Do aCo(O,t) = DR aCR(O,t) (A12b)
ax ax
The boundary conditions transform to
coCO,s) = e CR(O,S) (A.13a)
and
-Do o<:o(O,s) =DR aCR(O,S) (A13b)
ax ax
Differentiating co(x,s) and CR(X,S) with respect to x and setting x = 0, we
have on substitution in eq (A13b),
YsDo A(s) = - YSDR B(s)
454 Appendix 4
or
B(s) =- ~A(s) (AI4)
where
~= "IDo/DR
Setting x = 0 in eqs (All) and substituting in eq (A.I3a) gives
Co*/s + A(s) = eB(s) = - ~e A(s)
so that
A(s) =_ Co*
+ ~e)
s(1
Thus the transformed concentrations are
Co*
( ) = -s-
cox,s [1 exP{-Ys7i50x}]
1 + ~e
CR(X,S) = ~~o* exp{-Vs7l5'Rx}

1 + ~e
Taking the inverse transform, we have
( ) -C
Cox,t *~e+erf(x/2YDot)
- 0 (4.2a)
1 + ~e
CR(x,t ) -- C0 * ~ [1- erf(x/2Y DRt }] (4.2b)

1 + ~e
Derivation of eq (4.20). The Laplace transform of eq (4.19) is
( ) - C* = D [d C(r,S) + .2.
2
scr,s rdC(r,S)]
--
dr 2 dr
This differential equation can be converted to a more familiar form by
the substitution
vCr,s) = r c(r,s)
2
s vCr,s) _ rC* = D d vCr,s)
dr 2
Remembering that the range of r is ro to 00, the solution analogous to eq
(AIO) is
Reverting to the transformed concentration function, we have
c(r,s) = ~ + A~) exp [-VI (r-rol]
This expression is consistent with the boundary condition
C(r,t) -+ C*, c(r,s) -+ C*/s as r -+ 00
The surface boundary condition,

C(ro,t) = 0, t > 0
transforms to c(ro,s) = O. Thus we find A(s) = -roC*/s and have
c(r,s) = ~ II-~exp [-VI (r-rol])
Taking the reverse Laplace transform, we have
C(r,t) = C* [I-E.Q(I- erf r- ro)~ (4.20)
r 2Wt~
Derivation of eq (4.26). In double potential step
chronoamperometry, the electrode is polarized for a time 't at a
sufficiently negative potential that C o(O,t) = 0; the potential is then
stepped to a positive potential so that CR(O, t - 't) = O. This problem is
easily solved using Laplace transforms by noting that eq (4.2b) gives the
initial concentration distribution of R for the second potential step. Thus
substituting eq (4.2b) with t = 't and e = 0 into eq (A.IO), we have
~C *
CR(X,S) = +[I-e1~)] +A(S)ex~-ff.x)
where the transform variable s corresponds to (t - 't). The boundary
condition CR(O,t) =0 for t > 't implies that CR(O,S) = O. Thus
(cCo*ls) + A(s) = 0
so that
A(s) = - ~Co*ls
We need the flux of R at the electrode in order to calculate the current.
Thus we compute the derivative with respect to x
CkR(O,S) ~Co* ~Co*
---=- +--
ax s "1tDR't YDRS
and take the inverse transform to obtain
aCR(O,t) ~Co* ~Co*
-a-x- = - "1tDR't + J"=1tD~R(;::;:t=-='t=<=)
456 Appendix 4
Since the current is
we get
i = nFACo*fl5Qiii(-L-..l..) t >t (4.26)
vt-t Yt
Derivation of eq (4.27). In a constant current experiment such
as chronopotentiometry, the flux of 0 at the electrode surface is constant
up to the transition time when Co(O,t) -+ 0 and the potential swings
negative. Thus the boundary condition is
Do OCo(O,t) ---.L
ax nFA
Taking the Laplace transform, we have
Do aco(O,s) =~
ax nFAs
=0, we have
Differentiating eq (A.10) and setting x
-Doh/Do A(s) =nJAs
Solving for A(s) and substituting in eq (A. 10) with x =0:
cO<O,s) =C0 * .
I
S nFAsYsDo
Taking the inverse transform, we get
CO<O,t) =Co* _ 2ifi
nFAVreDo
Apparently, Co(O,t) goes to zero when
nFA VreDo Co* =2ifi
so that the transition time is given by
fC = nFA VreDo Co*/2 i (4.27)
Derivation of eq (4.29). For a linear potential scan experiment,
the boundary conditions are similar to those used in deriving eq (4.2)
except that the potential is time-dependent. Thus we can proceed as
before up through eq (A.14), but A(s) cannot be so simply evaluated. We
can write the transformed current as
i(s) = nFADo (aco(x,s))

ax %=0
Substituting eq (A.lla) for CO(x,s), we have
i(s) = - nFAV sDo A(s)
Solving for A(s), substituting into eqs (A.ll), and setting x = 0, we have
Co· i(s)
co(0)
,s = - - - - - = =
s nFA VsDo
i(s)
CR(0 ,s )
= ---'--===-
nFA VSDR
We now use the convolution theorm, eq (A.6), with h(s) = i(s), g(s) = s-1/2 to
obtain
Co(O,t) = Co. - 1
nFA Y7r.Do
t i(t) dt
10 yt-t
For a nernstian process, the ratio of the surface concentrations is

Co(O,t) = 9(t) = exp nF(Ei - vt _EO)
CR(O,t) RT
Substituting the surface concentrations into this expression and
rearranging, we have
nFA fit1JO Co· = t ~dt (4.29)

1 + ~9(t) 10 yt-t
where ~ = VDo/DR.
Derivation of eq (5.31). We are concerned here with the flux of 0
at the electrode surface when Co(O,t) = 0 and 0 is formed from Y in an
equilibrium prior to electron transfer. We start with diffusion equations
in Y and 0, similar to eqs (5.2) but including the time derivatives. With
the new functions of eqs (5.3), we obtain the differential equations
aC(x,t) = D a2C(X,t)
at ax 2
458 Appendix 4
dC'(X,t) = D iC'(x,t) _ (k i + k_l ) C'(X,t)

dt dx 2
We will assume that the equilibrium strongly favors Y so that K = kl/k_1
« 1 . Taking Laplace transforms, we obtain simple differential
equations which are readily solved to give
c(x,s) = C*/s + A(s) exp -Ysil5 x
c'(x,s) = A'(s) exp -yes + k_I)ID x
With eq (5.6b), the boundary condition, CO(O,t) = 0, gives
CO(O,s) = 6[-9-+ A (S)] -l;K A '(s)=O
Since Y is not electroactive, Jy(O,t) = 0 and eq (5.6a) gives
( dCO) =-A(s)~rr _A'(S).yS + k-I =0

dXx=o VD D
Solving the simultaneous equations, we have
A(s) = C* = _ A'(s) K 2s + Kki
S(l + • /
A /
S ) K·V s
'V K% + Kki
Neglecting K compared with 1 and assuming that K 2 s «Kkl,l the
transformed flux of 0 is
;o(O,s) = -C* w s +fKki.
YKklS
Taking the inverse Laplace transform and converting to current, we
have
(5.31)
Derivation of eqs (6.36). Allowing for a finite electron transfer

rate, the surface boundary conditions are
D o dCo(O,t) -- - D R dCR(O,t) -- k c C0 (0)

,t - k a CR(0 ,t)
dx dx
Transforming the boundary conditions, we again have eq (A.13b), but
instead of eq CA.13a), we get
1 This approximation is equivalent to neglecting the very small amount of 0

initially present near the electrode; thus the result if only approximate at short times.
Do oco(O,s) =kceo(O,s) - kaCR(O,S)

ax
Thus, with co(x,s) and CR(X,S) given by eqs (A.ll), and B(s) = -x A(s), we
have
- VsDo A(s) = kc[Co*ls +A(s)] + ka~A(s)
or
A(s) = _ keCo*
VDos(A +-IS)
where
A = kc!V Do + ka/V DR
Substituting A(s) into eqs (A.ll) with x = 0, we have
coCO,s) = C~* [1- Wo ke
Do (A +-IS)
]
CR(O,s) = keCo*
VDR seA + -IS)
Taking the inverse transforms, we have
Co(O,t) = Co* - kcCo* (1- exp(A2t}[1- erf(A Vt)))
'A.VDo
CR(O,t) = keCo* !1-eXP('A.2t}[1-erf('A.Vt)J)

'A.VDR
Defining
(('A.Vt) = 'A.fi[i exp('A.2t)[1- erf('A.Vt)]
we have
Co(O,t) = Co* _~[l_t{'A.Vt)] (6.36a)
1 + ~9 'A.fi[i
CR(O,t) = ~Co* [l_t{'A.Vt)] (6.36b)

1+~9 Afi[i
where we have used the relation kalke = 9.
Derivation of eqs (6.47). When the boundary condition is
determined by a sinusoidal current
let) = 10 sin rot

460 Appendix 4
the flux of 0 at the electrode surface is

- Do dCO(O,t) =- ~ sin rot
dX FA
Taking the Laplace transform, we have
Do dCO(O,S) = 10 0)
dX FA s2+ 0)2
Substituting the first derivative of eq (A.ll), evaluated at x = 0, we get
_ VsDoAs = 10 0)
FA s2+ 0)2
Solving for A(s) and substituting in eq (A.ll) with x = 0, we have
co(O,s) = Co* _ 10 [ 0) ]
s FAVDo VS(s2 + 0)2)
The inverse transform of this function cannot be found in tables, so we
have recourse to the convolution theorem, eq (A.6), taking
g(s) = _ 0 ) _ => G(t) = sin rot
s2+c02
h(s) = l => H(t) = _1_
VS f1ti
Thus
L-1 [ 0) ] = t_1- sinroCt-'t)d't

VS(s 2 + co2) Jo iii
U sing the trigonometric identity
sin roCt - 't) = sin rot cos CJ)'t - cos rot sin CJ)'t
the integral becomes
sin rot
fit
t ~ cos
Jo ft
CJ)'t d't - ~ t ~ sin
fit Jo ft
CJ)'t d't
The factor of 11ft in the integrand represents a transient response to the

application of the sinusoidal current which dies off to give a steady-state
sinusoidal variation in the concentrations. Since we are interested only
in the steady state, the limits on the integrals can be extended to infinity,
obtaining
Thus we have
CO<O,t) = Co. _ 10 {sin rot - cos rot)
FAY2roDo (6.47a)
Equation (6.47b) for CR(O,t) results from a similar development.
5
-c>< DIGITAL SIMULATION
z
W
Il.
METHODS
Il.
<
The theoretical description of an electrochemical experiment

usually requires the solution of a set of coupled partial differential
equations based on the diffusion equation. In experiments which
include forced convection, such as r.d.e. voltammetry, a driving term is
added to each equation as in eq (4.43). When a diffusing species is
involved in a chemical reaction, reaction rate terms must be added to the
equation describing its concentration. The set of equations often has
time-dependent boundary conditions and can be devilishly difficult to
solve. In some fortunate cases, such as those treated in Appendix 4, the
use of Laplace transforms leads to closed-form analytical solutions, but
more often solutions are obtained in terms of infinite series or
intractable integrals which must be evaluated numerically. In the
1950's and 1960's, a great deal of effort was expended by theorists in
obtaining mathematical descriptions of electrochemical experiments.
While the results provide an invaluable aid to understanding (we have
quoted many of these results), one often finds that the theoretical results
available in the literature do not quite cover the experimental case at
hand. If the problem seems to be of sufficient generality and interest, it
may be worthwhile attempting an analytical approach. More often,
however, electrochemists have turned to the digital computer to
simulate experiments. The details of digital simulation are beyond the
scope of this text, but a brief outline of the strategy is in order. For
further details see reviews by Feldberg,l Maloy,2 or Britz.3
In general the problem to be solved involves equations of the form
2
ae a C- + k·InetIc. and/or d··
- =D - nvmg terms
at ax 2
together with a set of initial and boundary conditions. The
concentrations are functions of time and distance from the electrode and
1 s. W. Feldberg, Electroanalytical Chemistry 1969, 3, 199.

2 J. T. Maloy in Laboratory Techniques in Electroanalytical Chemistry, P. T.
Kissinger and W. R. Heineman, eds, Kew York: Marcel Dekker, 1984.
3 D. Britz, Digital Simulation in Electrochemistry, Lecture Notes in Chemistry, Vol.
23, Heidelberg: Springer·Verlag, 1981.
462
Digital Simulation Methods 463
the general strategy in a digital simulation is to divide the time and

distance axes into discrete elements of size ot and Ox, respectively. The
first step in the development of a simulation program is to convert the
differential equations into finite difference equations.
Thus Fick's first and second laws,
J(x,t) = -D aC(x,t)
ax
aC(x,t) aJ(x,t)
---=----
at ax
which describe the diffusional part of the problem, can be written in
approximate form in terms of the finite differences, Ox and ot:
J(x,t) '" -D C(x + 0xI2,t) - C(X - 0xI2,t)
Ox
C(X,t + ot) - C(x,t) J(X + 0xI2,t) - J(x - &X/2,t)
ot Ox
Combining the two expressions, we have
C(x,t + Ot) = C(x,t) + .D.aL[C(x + Ox,t) - 2C(x,t) + C(X - Ox,t)] (A.15)
(axf
If the spatial boxes, of equal width Ox, are labeled 1,2,3 .. .j. .. and the time
boxes of width ot, are labeled 1,2,3 ... k ... we can rewrite eq (A.15) as
C(j,k+l) = C(j,k) + D[C(j+I,k) - 2C(j,k) + C(j-l,k)] (A. 16)
where D = Dotl(ox)2 is a dimensionless diffusion coefficient. Thus eq
(A.16) models the diffusion process as follows: during the time interval
ot, an amount DC(j,k) moves from boxj to each of the adjacent boxes. But
meanwhile DC(j-I,k) and DC(j+l,k) move from the adjacent boxes into
box j; the change in concentration in box j is the sum of these
contributions. Clearly, we can't take more out of a box than was there to
begin with, so that we require D s; 0.5.
The net flux at the electrode (related to the current) is determined by
the changes in concentrations in the j = 1 box (adjacent to the electrode)
which are required to satisfy the surface boundary condition. For
example, for a reversible electrode process, the boundary condition
corresponds to the surface concentration ratio CoiCR = e, specified by the
Nernst equation. The equilibrium surface concentration of 0 can be
written as
Co(eq) =_e_[Co(I,k) + CR(I,k)]
1+8
464 Appendix 5
The net flux then corresponds to the difference between Co(1,k) and the
equilibrium surface concentration
SCo(k) =Co(1,k) - 9CR(1,k) =- SGR(k)
1+9
The surface concentrations then are corrected to
Co(1,k + 1) =Co(1,k) - SGo(k) (A.17a)
CR(1,k + 1) =CR(l,k) + SCo(k) (A. 17b)
Since the current is given by
i =nFAJo(O)
the simulation current is proportional to SCo(k).
In an experiment where the electrode potential is time-dependent
(e.g., cyclic voltammetry), the potential (and thus the concentration ratio
9) will be different for each time increment. If the simulated experiment
involves two or more electron-transfer processes, these are handled
independently and the current contributions added to get the total
current. Experiments with slow electron-transfer kinetics can be
simulated by including Butler-Volmer electron-transfer rate equations
in place of the above expression which assumes that equilibrium is
attained instantaneously.
If species A is consumed in a first-order chemical reaction,
A~B
the contribution to the rate equations is

aCB(X,t) = _ aCA(X,t) = kCA(X,t)
at at
In terms of the finite differences, these expression are
CB(X,t + at) - CB(X,t) =_ CA(X,t + at) - CA(X,t) = kCA(x,t)
at at
or, using the indicesj and k,
CAV,k+1) = CAV,k) - kCAV,k) (A.1Ba)
CBV,k+1) =CBV,k) + kCAV,k) (A.1Bb)
where k = kSt is the dimensionless rate constant. In practice, k must be
small (~ 0.1) in order to accurately model the system; for a given value of
k, this places a restriction on 8t. This restriction can be somewhat
relaxed (k ~ 1) by using an analytical solution for the extent of reaction
during a time increment, e.g., for a first-order process
(A.19a)
Digital Simulation Methods 465
CB(j,k+l) = CB(j,k) + CA(j,k) [1- exp (-k)] (A. 19b)
Analytical expressions can be used for more complex reaction schemes,

or, for still more complex schemes, the extent of reaction can be
estimated using the modified Euler method. 1
For a cyclic voltammetry simulation, the time increment is
determined either by the scan rate with v ot '" 1 mV, or, when coupled
chemical reactions are considered, by the rate constant. The number of
time increments to be used in the simulation is determined by
nt = tlot (A.20)
where t is the total time of the experiment to be modeled. The size of the
spatial increment is determined by
&cD- o-tlD
= yr- - (A.21)
If the solution is isotropic at the beginninuf the experiment, the

diffusion layer grows to a thickness of about 6Y Dt during the time of the
experiment. Thus the number of spatial boxes required in the
simulation is determined by
nx&c = erlDnt Ot
or
nx = 6V Dnt &X&cf
nx = 6VDnt (A.22)
or nx '" 4 fflt ifD = 0.45.
Even with the use of analytical expressions such as eq (A.19) to
model the effects of chemical reactions, nt can be very large for schemes
with fast reactions. Since computer execution time increases as the
product of nt and nx, there is a practical upper limit to rate constants.
One solution to this problem is to use variable-width increments in the
simulation. For very fast reactions, the reaction layer, XR, is thin and, in
the simulation, most of the action takes place in the first few spatial
boxes adjacent to the electrode. Thus some saving in execution time can
be realized if the size of the spatial boxes is allowed to increase with
increasing distance from the electrode. 2 This approach leads to a
different D for each box and a somewhat more complex diffusion
algorithm. A still more efficient approach is to expand the time grid as
well for boxes far from the electrode. 3 Thus boxes in which not much is
happening are sampled less frequently.
The overall structure of the simulation program then is as follows:
1 D. K. Gosser and P. H. Rieger, Anal. Chern. 1988,60, 1159.

2 T. Joslin and D. Pletcher, J. Electroanal. Chern. 1974,49, 171.
3 R. Seeber and S. Stefani, Anal. Chern. 1981,53, 1011.
466 Appendix 5
(1) Set concentrations to initial values.
(2) Correct each concentration in each box for the results of the
chemical reactions using eqs (A.1B) or (A.19).
(3) Correct each concentration in each box for the results of diffusion
using eq (A.16).
(4) Change the concentrations in box 1 to satisfy the surface boundary
condition, e.g., using eqs (A.17) for a nernstian process; the changes
correspond to the flux and thus to the current.
(5) Go to the next time increment and adjust time-dependent
parameters such as the electrode potential.
(6) Repeat steps (2) - (5) for the required number of time increments.
6
-zc>< ANSWERS TO
W
Il.
SELECTED PROBLEMS
Il.
<C
U (a) PtIFe 2+, Fe3+) I I Mn04-, Mn2+, H+IPt

(b) Fe3+ + e- ~ Fe2+
Mn04- + 8 H+ + 5 e- ~ Mn2+ +4 H20
(c) 5 Fe 2+ + Mn04- + 8 H+ ~ 5 Fe3+ + Mn2+ + 4 H20
(d) EO = 1.51- 0.771 = 0.74 V
(e)E =0.56 V
(f) !lGo =-357 kJ mol-I, K =3 x 1062
1.2 (a) 1.0662 V

(b)-O.607V
(c)-O.76V
1.4 Ksp =1.6 x 10-8

1.5 Ksp =4.5 x 10-18, pH =7.00
1.7 E =0.077, 0.085, 0.107, 0.145, 0.302,0.458,0.491 V

1.8 1.27 x 10-4 M
1.9 K =9.8 x 1018

LlO (b) pH 5.0
lJ.1 flux = 2.2 x 10-3 mole Na+ s-l; firing rate =2.2 s-l
L12 (a) E' =+0.043, +0.353, -0.154 V
(b) !lG' ;;; -8.3, -68.1, +29.7 kJ moP
(c) K =28, 8.6 x 1011, 6.2 x 10-8
L13 (a) !lGo ;;; -130 kJ mol- 1

(b) !lGo =-76 kJ mol- 1
(c) !lGo =-206 kJ mol- 1
Ll4 (a) !lGo =-35 kJ mol- 1

467
468 Appendix 6
(b) !l.GO = +234 kJ mol- 1

(e) !l.Go = -458 kJ mol- 1
1.15 pH = 10.386 ± 0.008
1.16 [Na+] = (1.12 ± 0.03) x 10-4 M
1.17 (a) kH Na = 7.8 x 10- 12

(b) pH 11.19 solution would give apparent pH 11.14
1.18 (a) Zn(s) + 2 OH- ~ ZnO(s) + H20 + 2 e-

Ag20(S) + H20 + 2 e- ~ 2 Ag(s) + 20H-
Ag20(S) + Zn(s) ~ 2 Ag(s) + ZnO(s)
(b) 1100 J g-1
2.8 (a) Qoo = 2.0 IlC

(b) io = 1.0 rnA
(e) t = 0.092 s
2.9 (a) y±(NaCl)

(b) y±(NaF)fy±(NaCl)
(e) y±(NaF)
2.11 Y±(exptl) = 0.905, 0.875, 0.854, 0.826, 0.807, 0.786

Y±(eq 57) = 0.889, 0.847, 0.816, 0.769, 0.733, 0.690
Y±(eq 56) = 0.903, 0.872, 0.851, 0.821,0.800,0.776
2.12 [KOH] = 0.988 M, y± = 0.742
2.14 E = 0.767 V
3.1 (a) L fA = 29.05 m- 1

(b) A = 14.66 x 10-4 S m 2mol- 1
(e) A °= 390.7 x 10-4 S m 2mol- 1
= =
(d) ex 0.0375, K 1.5 x 10-5
3.2 N = 133.4 x 10-4 S m 2mol- 1

3.3 s(calc) = 2.88 x 10-4 S m1l2mol-3/2
s(expt) = 3.03 x 10-4
3.4 s = 1.94 x 10-3 S m1l2mol-3/2
3.5 0.203
Answers to Selected Problems 469
3.6 1.4 x 10-4 M
3.7 K = 1.6 x 10-4

3.8 u = 1.046 x 10-7, 1.145 x 10-7 m2V- 1s-1
r = 274, 334 pm
f = 4.60 x 10-12, 5.60 x 10- 12 kg s-l
D = 8.96 x 10- 10, 7.36 x 10- 10 m2s- 1
3.12 D = 2.45 x 10- 11 m 2s- 1, t '" 6.5 years
3.13 D = 1.59 x 10-9 m 2s-1
3.14 teu = 0.366
3.16 [Ba(OH)2] = 0.0422 M, A = 411 x 10-4 S m 2mol- 1
3.18 [Na+]a = 0.0319 M, [Na+]~ = 0.0281 M, ~<l> = 3.3 mV
4.4 E1I2 = 0.764 V
4.5 A square wave signal is required.
4.7 (a) Gain> 10 4

(b) Gain> 4 x 104
(c) Scan rate is 2 ppm smaller when output voltage is -1 V.
(d) i = 100 j.4A, RcelJ < 10 kn, nominal current correct to 0.001%
4.8 Co = 0.14 mM
4.12 4.75 j.4A
4.14 (a) El/2 = -0.693 V

(b)R = 1480 n
4.15 Diameter greater than 0.5 mm
4.16 Electrode radius less than about 2.3 ~
4.20 13 '" 2.4 eV (Figure 4.42a), 13 '" 2.1 eV (Figure 4.42b)
4.26 13 '" 4 x 10 12
4.27 (a) p =2, El/2 =-0.258 V

(b) 13 "" 1035
470 Appendix 6
5.1 k1K = 0.008 s-l
5.2 (a) CA> 10-6 M

(b) A = 1 cm2
5.7 (a) (iL - i)/i = 82(1 + 28 1)/(2 + 82)

(b) E1I2 = (E 1° + E2°)/2
(c) The Heyrovsky-Ilkovic equation is obtained when E2° »E1°
(d) E3/4 -E1/4 = 42.9,33.8,30.3,28.5 mV for E2° -E1° = 0, 50,100,
and 200 mY. The Tomes criterion for a reversible two-electron
wave gives 28.2 mY.
5.8 k = 0.8 s-l
5.10 (b) k = 1.0 s-l
5.11 (a) 60
(b) 1.3
(c) 5.1
5.12 (a) DEl/2 = 17.8,61.6 mV

(b) DEl/2 = 101.9,41.8 mV
6.2 i - nFA ( kcCo* - kaCR* )

- 1 + kalkn + kclkn
6.5 Jo = 0.0079 A m-2 , (Xapp = 0.58, ko = 2.2 x 10-10 m s-l
6.9 (a) E1I2 _EO = -24.7 mV

(b) kolkn = 0.141
6.10 (b) ko = 2.1 x 10-4 m s-l
6.12 ko = 5.8 x 10-5 m s-l
6.13 Resistive component: Rs = IZrl cos cp

Capacitive component: 1/roCs = IZrl sin cp
6.14 (b) Width at half height = 90.6 mY.
6.15 (b) Width between extrema = 67.6 mY.

6.16 (a) ko s 10-5 m s-l
(b) ko s 0.0002 m s-l
(c) ko s 0.002 m s-l
Answers to Selected Problems 471
7.2 t = 27.5 days
7.3 (a) 22 min

0.10, 0.14, and 0.12 mM at t = 5, 10, and 25 min
(b) GB(t) =
7.6 126C
7.7 (a)0.234C
(b)48.5 J.!M
7.8 (a) R = 1000 Q

(b) Rcell < 9 kQ
7.9 3.51 roM
7.10 5-mL sample, 2 rnA current give t = 241 s, uncertainty is ±0.6%
7.12 Hydrogen reduction current: j = 0.27 A m- 2 on Pt (cp = 0.997),j = 1.9

rnA m- 2 on Cu (cp = 0.9992)
7.13 Optimum current = 133 A

Cost = (raw materials + capital and labor + energy)
Cost = (1.00 + 0.84 + 0.91) = $2.75 per kilogram
With doubled energy costs, optimum current = 91 A
Cost = $3.57 per kilogram
7.15 (a) Ec = -0.73 V, Corrosion rate = 0.5 mg Zn m- 2 s- l , Hydrogen

evolution rate = 0.18 mL m- 2 s- l
(b) Corrosion rate = 2.2 mm yearl
AUTHOR INDEX
An author's work is mentioned on page numbers given in roman type,

biographical sketches on page numbers given in bold, and literature
citations on page numbers in italics.
Abruiia, H. D. 429 Britz, D. 462

Adams, R. N. 171,237,286,309, 428 Brooks, M. A 236, 239
Alberts, G. S. 293, 309 Brown, B. F. 448
Albery, VV.J.236,239,277,309,42~430 Bronsted, J. N. 98, 105
Alder, R. 253,308 Bruckenstein S. 261,309
Allendoerfer, R. D. 261,309,322,367 Buck, R. P. 40, 54
Amatore, C. 303, 309 Bungenberg de Jong, H. G. 81, 105
Anderson, T. N. 89, 105 Bunger, VV. B. 446
Angrist, S. VV. 430 Burbank, J. 51,54
Antropov, L. I. 428 Burnett, R. VV. 236, 239
Arrhenius, S. A 126 Burton, K 443
Austen, D. E. G. 259, 308 Butler, E. A 444
Badoz-Lambling, J. 427 Butler, J. A V. 318, 367
Baer, C. D. 215,238,342,367 Butler, J. N. 107
Bagotzky, V. S. 430 Cairns, E. J. 45, 50, 54
Baizer, M. M. 430 Camaioni-Neto, C. 306,309, 342,367
Bard, A J. 428, 431, 432, 441 Carlisle, A 4, 113, 371
Barendrecht, E. 388, 389,422 Caroli, S. 431
Barker, G. C. 202, 2M, 238 Carroll, J. B. 261,309
Barnes, K K 428 Casanova,J. 391, 422
Bates, R. G. 429 Cawley, L. P. 85, 105
Bauer, H. H. 200,238 Chapman, D. L. 60, 61, 105
Bell, R. P. 277, 309 Charlot, G. 427
Belleau, B. 395, 422 Chateau-Gosselin, M. 236, 239
Bernal, I. 259, 261,308 Christie, J. H. 206, 238
Bezems, G. J. 308, 309 Chung, Y. K 306, 309
Bier, M. 85, 105 Churchill, R. V. 448
Birke, R. L. 235,239, 281,309 Colton, R. 253, 308
Bockris, J. O'M. 333, 367, 369, 415, 428, Compton, R. G. 258,308
430,431 Connelly, N. G. 247, 263,308,309
Bond,A. M. 207,215,232,238,239,253, Conway,B.E.42~431
308,341,366,367,429 Cooke, W. D. 175,237
Brdi&a, R. 198,238,278,280,309,287 Cottrell, F. G. 156, 237
Brezina, M. 236, 239 Covitz, F. H. 429
Britton, vv. E. 169,237,428 Crow, D. R. 232,239,427
472
Author Index 473
Cruikshank, W. 371, 372, 377, 421 Gerischer, H. 432
Curran, D. J. 387, 422 Geske, D. H. 259,308
Dalrymple-Alford, 193, 238 Gibbs, O. W. 377,421
Damaskin, B. B. 427 Gileadi, E. 430
Daniell, J. F. 4 Given, P. H. 259,308
Davis, D. G. 181,238 Goldberg, I. B. 258, 308
Davy, H. 4 Gosser, D. K 266,309,465
de Bethune, A J. 431 Goto, M. 193,238
de Montauzon, D. 226,238,247,308 Gouy, L.-G. 60,61, 105
de Smet, M. 79, 80, 105 Grahame, D. C. 60,87,88,89,105
Debye,P.J. W.90,91,105, 124,147 Greef, R. 428
DeFord, D. D. 383,421 Gregory, D. P. 45,54
Delahay, P. 26, 54, 179, 181,237,292, Gritzer, G. 171,237
309, 427, 432 Gross, M. 247,308
Deryagin, B. V. 73, 105 Grove, W. R. 48, 49, 54
Despic, A R. 333,367, 369,415 Grunwald, R. A 236, 239
Dietz, R. 361, 367 Grzeszczuk, M. 365, 367
Dobos, D. 431, 445 Guggenheim, E. A 142, 147
Donnan, F. G. 142, 147, 158 Haber, F. 36, 54
Dordesch, K V. 45, 54 Hajdu, J. 391,422
Drazic, D. 333,367,369,415 Hall, C. M. 402
Dryhurst, G. 171, 237, 429 Hamaker, H. C. 69, 105
Dukhin, S. S. 73, 105 Hamed, H. S. 97, 105, lO7,427
Dvorak, J. 428 Harrar, J. E. 383, 421
Ebsworth, E. A V. 20, 54 Harris, M. D. 236, 239
Edison, T. A. 52 Hawkridge, F. M. 173, 237
Ehlers, R. W. 97, 105 Hawley, M. D. 247, 293,308,309
Eisenberg, M. 50, 54 Headridge,J.B.427
Engles, R. 394, 422 Healy, T. W. 70, 105
Enke, C. G. 113, 146 Heineman, W. R. 389,422,430
Erdey-Gruz, T. 318,367 Heinze,J.218,220,222,238
Erman,P.113,115,146,371 Helmholtz, H. von 60, 61, 105
Falkenhagen, H. 124, 147 Henderson, P. 139, 147
Faraday, M. 4,52,371,421 Hershberger, J. W. 305, 309
Faulkner,L. R. 204,238,266,309,428 Heubert, B. J. 322,367
Feldberg,S. W.286,293,294,303,304, Heydweiller, A 126, 147
309,462 Heyrovsky, J. 157,158, 165, 194,232,
Fick, A E. 130 237,239,429
Filinovskii, V. Yu. 428 Heroult, P. L. T. 402
Fillenz, M. 236, 239 Hibbert, D. 427
Flato, J. D. 232,239 Hills, G. J. 432
Fraenkel, G. K 259, 261, 308 Hitchman, M. L. 430
Frost, A. A 20, 54 Hittorf, J. W. 116, 117, 146
Frumkin, A N. 324, 367 Hogg, R. 70, 105
Fry, A J. 169,237,428,430 Hoijtink, G. J. 225,238
Fuoss, R. M. 122, 146 Holler, F. J. 113, 146
Furman, N. H. 39,54,175,237 Howell, J. O. 222,238
Furstenau, D. W. 70, 105 Huston, R. 107
Gale, R. J. 429 Huckel, E. 90, 91, 105
Galus, Z. 171, 237, 429 Ilkovic, D. 157, 158, 196,237,238
Galvani, L. 4 Ingram, D. J. E. 259,308
Gardner, A. W. 202, 238 Ishii, D. 238
Geary, C. G. 107 Israel, Y. 348,367
Geiger, W. E. 247,308,363,365,367 Ives, D. J. G. 430
474 Author Index
James, A. M. 427 Manecke, G. 327, 367
Janata, J. 39,54 Mann, C. K 169,237, 428, 447
Janz,G.J.43~431 Mantell, C. L. 430
Jefti~L.293,294,303,309 Marcoux, L. S. 286, 309
Jensen, B. S.299,309 Marcus, R. A. 316, 319, 367
Jones, R. D. 207,238 Mark, H. B., Jr. 389, 422, 428
Jordan,J.431,441 Martinchek, G. A. 261,309
Joslin, T. 465 Marzlutf, W. F., Jr. 281, 309
KekuIe, A. 396 Mattson, J. S. 428
Kern, D. M. H. 279,309 Mazorra,M.235,239
Khan, S. U. M. 428 McAllister, D. L. 171,237
Kim, S. 306, 309 McBreen, J. 45, 54
Kirowa-Eisner, E. 204, 234, 238, 239, McCollum, P. A. 448
245,430 McCormick, M. J. 253, 308
Kissinger, P. T. 430 McKinney, T. M. 258,308
Klemenciewicz, Z. 36, 54 Meites, J. 384, 421
Klingler, R. J. 305, 309 Meites, L. 194, 233, 238, 239, 348, 367,
Kochi, J. K 305,309 384,421,429,431
Kohlrausch, F. 114, 115, 126, 146, 147 Meng, Q. 306, 309
Kolbe, H. 395, 396,422 Milazzo, G. 431
Kolthotf, I. M. 207, 234,238, 429 Miller, J. W. 383, 421
Kornberg, H. L. 443 Milner, P. C. 53, 54
Kortum, G. 428 Mohammad, M. 391, 422
Koryta, J. 236,239, 427, 428, 429 Mohilner, D. M. 60, 105
Kosaka, T. 394, 422 Moraczewski, J. 363,367
Kosower, E. M. 391, 422 Morris J. R. 204,238
Koutecky, J. 278,309 Murray, R. W. 39, 54, 427
Krebs, H. A. 443 Myland, J. C. 341,367
Kublik, Z. 389, 422 Nagy, Z. 431
Kuta, J. 171,237, 429 Narayanan, A. 431
Kyriacou, D. K 430 Nernst, W. 11, 158,318
Laitinen, H. A. 207,238,392,422 Neto, C. C. see Camaioni-Neto, C.
LaMer, V. K 98, 105 Newman,J.220,238
Latimer, W. M. 18,431 Newman, J. S. 428
Le Bel, J. A. 396 Nicholson, R. S. 183, 187, 238, 296, 298,
Le Chatelier, H. 401 309, 348, 366, 367
Leclanche, G. 45 Nicholson, W. 4, 113, 371
Lehnhotf,N.S.215,238,266,309 Nishiguchi, I. 395, 422
Lemoine, P. 247,308 Nixon, F. E. 448
Leon, L. E. 233,239 O'Brien, P. 342, 367
Levich, V. G. 212, 238, 428 O'Halloran, R. J. 366, 367
Lewis, G. N. 18 O'Neill, R. D. 236, 239
Lingane,J.J.230,234,238,384,421,429 Ohkawa, M. 395, 422
Lingane,P.J.217,238 Oldham, K B. 193,215,217,238,274,
Loud, N. A. S. 431 309,341,367
Loveland, J. W. 128, 147 Onsager,L. 77,105,122, 123, 146
Lowenheim, F. A. 430 Orpen, A. G. 263, 309
Lund, H. 430, 431 Ostapczuk, P. 389, 422
Lyons, E. H., Jr. 427 Osteryoung, J. 204,234,238,239,245
MacDonald, D. D. 430 Osteryoung, R. A. 206,238, 390, 422
MacDonald, H. C., Jr. 428 Ostwald, F. W. 5,48,54, 156,428
MacInnes, D. A. 427 Overman, R. F. 56
Maki, A. H. 259, 308 Owen, B. B.427
Maloy, J. T. 462 Parker, V. D. 191,238,299,309
Author Index 475
Parsons,R.60,105,324,367,431,441 Selley, N. J. 427
Patriarche, G. J. 236,239 Shain, I. 183, 187, 190,238,289,293,296,
Peat, R. 428 298,309,389,422
Penciner, J. 430 Shannon, R. D. 121, 146
Peover,~.E.259,308,361,367 Shaw, D. J. 69,70,85,105
Perkins, R. S. 89, 105 Shedlovsky, L. 113, 146
Peter, L.~. 428 Shedlovsky, T. 113, 146
Pinson,J.303,309 Shimizu, K 390, 422
Plambeck, J. A. 429 Shono,T. 394, 395,422,430
Plante, G. 52 Siegerman, H. 236, 239
Pleskov, Yu. V. 428 Simon, A. C. 51, 54
Pletcher, D. 428,430,465 Skundin, A. ~. 430
Poilblanc, R. 226,238,247,308 Smith, D. E. 322, 356, 365, 366, 367
Pourbaix,~.26,54,431 Smith, ~. G. 448
Prada!; J. 236, 239 Somo~,Z.380,421
Prade~va,J.236,239 Soos, Z. G. 217,238
Purdy, W. C. 429 Spiro, ~. 116, 146
Reddy, A. 428 Srinivasan, S. 430
Reeves, R. ~. 60, 105 Steckhan, E. 394, 422
Reilley, C. N. 175,237,427 Stefani, S. 465
Reinmuth, W. H. 179,237, 259, 261, 290, Steihl, G. L. 292, 309
308 Stem, O. 60,61, 105
Revie, R. W. 431 Stock, J. T. 177,237
Riddick, J. A. 446 Stokes, G. G. 121
Riddiford, A. C. 213,238 Stokes, R. H. 428, 445
Rieger, A. L. 263, 309 Stone, N. J., see N. S. Lehnhoff
Rieger,P. H. 258,259, 261,263,266,308, SWrzbach, ~. 220, 222, 238
308,309,322,367,465 Streitwieser, A. 226, 238
Rifi, ~. R. 392, 422, 429 Stulik, K. 429
Riley, T. 427 Sweigart, D. A. 215, 238, 266,306,309,
Robbins, J. 427 342,367
Roberts, J. L., Jr. 430 Swift, E. H. 444
Robinson, J. 428 Szebelledy, L.380,421
Robinson, R. A. 428, 445 Tachikawa, H. 266, 309
Roe, D. K. 167,238 Tafel, J. 325, 326,367
Rogers, J. R. 391, 422 Talmor, D. 244
Rosair, G. ~. 263, 309 Tanaka, N. 377,421
Rossiter, B. W. 428 Testa, A. C. 290, 309
Rossotti, H. 429 Thiebault, A. 303, 309
Roston, D. A. 389, 422 Thirsk, H. R. 432
Rupp, E. B. 431 Thomas, U. B. 53,54
Rutgers, A. J. 79, 80, 105 Tilak, B. V. 430
Ruzic, I. 366, 367 Tilset, ~. 299, 309
Rysselberghe, P. van 26, 54 Tiselius, A. 84, 105, 117
Sack, H. 124,147 Tobias, C. W. 432
Safford, L. K. 350, 367 Tome'§, J. 158,237
Salkind, A. J. 432 Tomkins, R. P. T. 431
Sand,H.T.S.182,238 Tomlinson, C. 427
Saveant,J.~.303,308,309 Trasatti, S. 337, 367
Sawyer,D.T.233,239,430 Tremillon, B. 427
Schafer, R. 394, 422 Tropp, C. 280,309
Schwartz, W. M. 289, 309 Turner, J. A. 206,238
Scott, C. J. 263,309 Uhlig, H. H. 418,422,431
Seeber, R. 465 Underwood, A. L. 237,239
476 Author Index
van't Hoff, J. H. 396
Vandenbalck, J. L 236,239
Verhoef, J. C. 388,422
Vesely, J. 429
Vetter, K J. 327, 330, 367, 428
Visco, S. J. 308,309
Volmer, M. 318,367
Volta, A 4, 371
Vukovic, M. 206, 238
Waller, A G. 258, 308
Wawzonek,S.39,54,392,422
Weaver,M.J.350,367
Weinberg, N. L. 395, 422, 429,430
Weiss, D. 429
Weissberger, A. 428
Westcott, C. C. 429
Weston, E. 7
Whewell, W. 4
White, R. E. 431
Wien, M. 124, 147
Wiese, G. R. 70,105
Wiesner, K 278, 280, 309
Wightman, R. M. 215, 223,238
Willihnganz, E. 51,54
Winograd, N. 171, 237
Wipf, D. O. 215,238
Wise, J. A. 389, 422
Wopschall, R. H. 190, 238
Yeager, E. B. 124, 147,432
Yoshida, K 430
Zana, R. 124, 147
Zhan~ Y.266,309,342,367
Zoski, C. G. 215, 217,238,341,367
Zuman, P. 236,239,247,308,431
SUBJECT INDEX
a.c. cyclic voltammetry 366 Karl Fischer titration 387

a.c. polarography 322, 356-366 polarography 231-236, 281
EC mechanism 362-366 potentiometric titration 42-44
irreversible process 360-362 stripping voltammetry 389-390
second-harmonic 358 of cyanide ion 245
activation free energy 316 of hydroxide ion 234
activity 9 of metal ions 223
of solvent 100-102 ofthiols 235
activity coefficients 9, 17 chemical kinetics 247-249
experimental determination 96-98 chemical potential 91
from Debye-Hiickel theory 91-95 electrochemical potential 134
mean ionic activity coefficients 93 chlor-alkaIi process 404-407
molal & mole fraction scales 99-100 chronoamperometry 179-181
adsorption CE mechanism 287
effect on cyclic voltammetry 189-190 double potential step 180, 289
effect on polarography 198-199 EC mechanism 288-289
aluminum production 401-2 EC' mechanism 292-293
anodic stripping voltammetry 389-390 ECE mechanisms 293, 294, 295
anodization 400 capacitive charging current 180, 220
band theory of solids 1 irreversible process 346-348
biamperometric titrations 388 time scale 250
biochemical half-cell potentials chronocoulometry 241, 369
table 442 chronopotentiometry 181-183
Butler-Volmer equation 318 current reversal 182, 290-291
calomel electrode 31, 170 EC mechanism 290-291
catalytic current 280 time scale 250
cathodic stripping voltammetry 390 colloids 68-73, 81-85
CE mechanism 256, 274-278, 287, 296 electrophoresis 82-85
charge-transfer resistance 326-327, 354 electroviscous effect 81
chemical analysis sedimentation potential 82
a.c. polarography 358 stability of 69-71
amperometric titration 177-178 surface pH 72
biamperometric titration 388 surface potential 64, 72
conductometric titration 127-128 conductance 109
constant current coulometry 384 measurement 110-113
constant potential coulometry 380 relation to conductivity 109
coulometric titration 385 relation to resistance 109
dead-stop titration 388 conductivity 109-124
electrogravimetry 377-379 electrolyte charge type from 127
electrolysis methods 376-390 equilibrium constants from 125
ion-selective electrodes 40-42 at high electric field strength 124
477
478 Subject Index
at high frequency 124 diffusion 128-136
ionic conductivity 115 Fick's first law 130
of pure water 126 Fick's second law 131
relation to diffusion 134 flux 130
relation to mobility 119 random walk model 128-130
theory of 122-124 to a microdisk electrode 216-219
units 109 to a planar electrode 153-162
conductometric titrations 128 to a spherical electrode 162-165
convection 152 diffusion coefficient 130
r.d.e. forced convection 208 relation to conductivity 135
conventions relation to frictional coefficient 134
potential scale zero 12 relation to mobility 133-135
sign of current 151, 195 diffusion layer thickness 158, 195, 211
sign of potential 7 diffusion-limited current 152-164, 196
sign of work 5 digital simulation 462-466
writing half-cell reactions 11 Donnan membrane potential 141-143
corrosion 412-421 double layer
corrosion inhibitors 420 effect on
differential aeration 419 colloid stability 68
passivation 417-418, 421 electron-transfer rate 323-324
prevention of corrosion 420-421 interfacial tension 86-88
reaction of metal with air 416 Gouy-Chapman theory 59-68
reaction of metal with water 413-416 surface pH 72, 410
sacrificial anodes 420 thickness of 66, 92
corrosion potential 414 double layer capacitance 88-90, 110,351
Cottrell equation 156 effect on
coulometric titrations 387 chronoamperometry 180, 220
coulometry 380-388 cyclic voltammetry 185, 223
constant potential coulometry 380 polarography 197, 202-206
constant current coulometry 384 dropping Hg electrode 172, 194-201
current-potential curve 155-162 effect of potential on drop time 86
cyclic voltammetry 183-194 homogeneous kinetics at 272
adsorption effects 189-190 see polarography
capacitive charging current 185, 223 dry cell 45
CE mechanism 296 Ebsworth diagrams 20
EC mechanism 296-300 EC mechanism 256, 279, 288-291, 296-
EC' mechanism 300 300, 362-366
ECE mechanisms 301-308 EC' mechanism 256, 280-281, 292-293,
derivative presentation 190-192 300
irreversible,quasi-reversible 348 ECE mechanisms 256, 282-287,293-295,
ohmic potential drop 185 301-308
semiderivative presentation 192-194 Edison cell 52
time scale 250 EE mechanism 332-334, 342-345
Daniell cell 2 efficiency of fuel cells 47
dead-stop titration 388 electric migration 152
Debye length 66, 92 electrical circuits
Debye-Falkenhagen effect 124 a.c. bridge 112
Debye-Huckel theory 90-98 cell equivalent circuit 110, 351
comparison with experiment 95-98 current source 168
extensions from 102-104 electrometer 30
limiting law 95 constant current source 386
relation to conductivity 122 for measuring a cell potential 28, 29
dielectric constant 61, 169 potentiometer 29
table 446 operational amplifier 30, 166
Subject Index 479
current follower 165 electrolysis 391
ramp generator 167 analytical applications 376-390
voltage follower 165 current efficiency 374
coulometer 382 electroseparation 374-375
potentiostat 165 electrosynthesis 390-396
galvanostat 386 industrial processes 396-412
voltage integrator 167, 382 electrometer 30
Wheatstone bridge 112 electromotive force 6
electrocapillarity 86-88 electron spin resonance 258-264
electrochemical cell potentials electron-transfer rate 316-319, 321-324,
measurement of 27 360,366
electrochemical cells Frumkin effect 324
conductance cells 113 Marcus theory 319-322
for coulometry 380 electroosmosis 75
Daniell cell 2, 151 electroosmotic pressure 75
Edison cell 52 electrophoresis 82-85
for electrogravimetry 379 electrophoretic painting 400
electrolysis cells 5, 151 electroplating 398-400
equivalent circuit 110,351 electrorefining 403
fuel cells 47-50 electrosynthesis 390-396
galvanic cells 2 Kolbe hydrocarbon synthesis 395-396
lead-acid cell 51 oxidation of olefins 394-395
Leclanche cell (dry cell) 45 reduction of aromatics 392
mercury cell 46 reductive elimination reactions 391
nickel-cadmium cell 53 equation
production of aluminum 402 Boltzmann distribution law 62
production of Cl2 & NaOH 405 Butler-Volmer equation 318
production of lead tetraalkyls 411 Cottrell equation 156
silver cell 47 Debye-Huckel limiting law 95
silver-zinc cell 58 Einstein relation 135
sodium-sulfur cell 53 Fick's first law 130
storage batteries 50-54 Fick's second law 131
thermodynamics 5 Gibbs-Duhem equation 86, 101
three-electrode configuration 165 Henderson equation 140
for voltammetry 173 Heyrovsky-Ilcovic equation 157
Weston cell 7-12, 29 Ilkovic equation 196
electrochemical potential 134 Kohlrausch equation 114
electrodes Levich equation 212
anode 4 Lingane equation 230
auxiliary electrode 165 Nernst equation 10-13
cathode 4 Nernst-Einstein equation 135
in conductance cell 113 Nernst-Planck equation 134
for stripping analysis 389 Ohm's law 76
see indicator electrodes Ostwald's dilution law 126
see reference electrodes Poiseuille's equation 76
electrogravimetric analysis 377-379 Poisson equation 61
electrokinetic phenomena 73-80 Poisson-Boltzmann equation 63
electroosmosis 75 Sand equation 182
electroosmotic pressure 75 Stokes law 82
streaming current 73 Tafel equation 325
streaming potential 74 equilibrium constants
theory of 76-79 from cell potential data 14
zeta potential 75, 79-80, 83 from conductance data 125
from polarographic data 227-231
480 Subject Index
error function 131-132 rotating platinum electrode 207
exchange current 318 rotating ring-disk electrode 214
faradaic impedance 110, 351-356 static Hg drop electrode 206-207
Faraday's laws of electrolysis 371, 380 industrial processes 396-412
Fermi level 1 anodization 400
ferrocene electrophoretic painting 400
as a potential reference 171 electroplating 398-399
rate of oxidation 350 electrorefining 403
Fick's laws of diffusion 128 hydrometallurgical processes 403
Flade potential 417 organic syntheses 412
formal potentials production of
table 444 adiponitrile 409-410
Franck-Condon principle 319 alkalies & alkaline earths 403
free energy-oxidation state diagrams aluminum 401-402
19-21 chlorates and bromates 407
frictional coefficient 120 Cl2 and NaOH 404-407
relation to diffusion coefficient 134 fluorine 407
Frost diagrams 19-21 lead tetraalkyls 410-412
Frumkin effect 324 manganese dioxide 408
fuel cells 47-50 perchlorates 407
galvanic cells 2 potassium dichromate 408
Gibbs free energy potassium permanganate 408
of activation 316-317 infrared spectroscopy 265-268
from cell potential data 14, 224-227 ion-selective electrode 35-39, 143-146
relation to electrical work 5 ionic conductivity
Gouy layer 60 table 445
Gouy-Chapman theory 59-68 ionic radii
half-cell potentials crystal radii 121
table 438 Stokes law radii 120-121
half-cell reactions 8 ionic strength 63
half-wave potential 157, 212 kinetic current 274
correlation with IR frequencies 226 kinetic zones
correlation with MO theory 224 CE mechanism 275-276
stability constants from 227-231 EC' mechanism 292-293
table 233 kinetics of electron transfer 315-324
Hall-Heroult process 401 Kohlrausch law of independent ionic
Helmholtz layer 60 migration 115
heterogeneous rate constants 316 Kolbe hydrocarbon synthesis 395
Heyrovsky-IlcovU: equation 157 Laplace transforms 448-461
history of electrochemistry table 450
double layer theory 60 Latimer diagrams 18
conductivity 114-115 lead-acid cell 51
electrolysis & Faraday's laws 371 Lec1anche cell 45
origins 3-4 liquid junction potentials 2, 136-141
polarography 194 London force 69
hydrodynamic layer thickness 209 lyotropic series 71
hydrogen evolution kinetics 334-337 Marcus theory 319-322
indicator electrodes 171-172 mass transport rate constants 159, 211,
dropping Hg electrode 172, 194-201 218
glass electrode 35 maximum suppressor 200-201
ion-selective electrode 35-39, 143-146 mechanisms 247
microelectrodes 215-223 CE 256, 274-278, 287, 296
quinhydrone electrode 56 EC 256, 279, 288-291, 296-300, 362-366
rotating-disk electrode 172, 207-215 Ee' 256, 280-281, 292-293, 300
Subject Index 481
ECE 256, 282-287, 293-295, 301-308 overpotential (overvoltage) 317
EE 332-334, 342-345 oxidation state diagrams 19
bond cleavage 251 peak polarogram 183
electrophilic attack 252 pH measurements 40
multi-electron processes 249 phase angle 352
multi-step processes 328-334 planar diffusion 153-162
reactions of olefin radical cations polarography 176, 194-201,231-234
394 a.c. polarography 322, 356-366
rearrangement 253 adsorption effects 198-199
reductive elimination reactions 391 analytical applications 231-234
reduction of aromatic anodic waves 235
hydrocarbons, nitro and capacitive charging current 197,
carbonyl compounds 293 ID2-200
mechanistic data on CE mechanism 274
hydrogen evolution 334-337 criteria for reversibility 345
oxidation of current maxima 201
p-aminophenol291 differential pulse 204
ArCr(CO)2(alkyne) 263 EC mechanism 279
iron 333 EC' mechanism 281
(mesitylene)W(CO)3 266 ECE mechanisms 282
Mn(CO)3(dppm)CI 253 instrumentation 165-168
CpMn(CO>2L 305 irreversible waves 348
9-phenylanthracene 299 maximum suppressors 200
reduction of peak polarography 183
azobenzene 289 polarographic wave 157, 161
(COT)CoCp 363 pulse polargraphy 202
cyclooctatetraene 322 resolution 162, 205
triiodide ion 330 reverse pulse polarography 204
Mn3 + 327 sampled d.c. (tast) polarography 201
CpMn(NO)(CO)2 306 sensitivity 197,201-206
p-nitrosophenol 295 square wave 206
1,1,2,3,3-pentacyanopropenide 261 stability constants from 227-231
mercury cell 46 static Hg drop electrode 206
microelectrode voltammetry time scale 250
see steady-state voltammetry, cyclic total electrolysis in 198
voltammetry potential
microelectrodes 215-223 corrosion potential 414
electron transfer kinetics 338, 350 between dissimilar conductors 1
homogeneous reactions at 272 Donnan membrane potential 141-
mobility 143
electrophoretic mobility 82-84 electrochemical cell potential 6
ionic mobility 118-119 ferrocene as a potential standard
relation to conductivity 119 171
relation to diffusion coefficient 133 Flade potential 417
molecular orbital theory 224 formal potentials 17, 154
N ernst diffusion layer 158 half-cell potentials 11-16
Nernst equation 10-13 half-wave potential 157,212
nickel-cadmium cell 53 ion-selective membrane 37, 143·146
ohmic potential drop Latimer diagrams 18
in cyclic voltammetry 185, 191, 193 liquid junction potential 2, 136-141
in steady· state voltammetry 213, measurement 27
220,222 potential of zero charge 87-90, 324
Onsager reciprocal relations 77 sedimentation potential 82
osmotic pressure 142 standard reference half-cell 12
482 Subject Index
surface potential 64, 66 spherical diffusion 162-165,216-219
ultrasonic vibration potential 124 square scheme 263, 305, 307, 364
zeta potential 75, 79-80, 83 standard states 9,17,21,91
potential range for solvents 447 static Hg drop electrode 206
potential-pH diagrams 25-26 steady-state voltammetry 176, 212, 220
potentiometric titration 42-44 CE mechanism 274-278
potentiostat 165-166 coupled homogeneous reactions 269
predominance area diagrams criteria for reversibility 345
(Pourbaix diagrams) 26-27, 413 EC mechanism 279
r.d.e. voltammetry EC' mechanism 280-281
see steady-state voltammetry ECE mechanisms 282-287
stripping analysis 389 EE process 342-345
rate laws 247 irreversible/quasi-reversible 338
rate of electron transfer in time scale 250
oxidation of ferrocene 351 Stem model for double layer 60
reduction of Stokes' law 120
(COT)CoCp 363 Stokes' law radii 120-121
cyclooctatetraene 322 storage batteries 50-54
Mn porphyrin complex 342 streaming current 73
substituted stilbenes 361 streaming potential 74
reaction layer thickness· 251, 270-273 stripping voltammetry 389-390
reference electrodes 31, 170-171 supporting electrolyte 153, 170
Ag/AgCI electrode 32, 170 surface charge density 66, 86, 89
calomel electrode 31, 170 surface tension
hydrogen electrode 12 effect of potential on 86-88
Luggin probe 168 symmetry factor 319, 333
table of potentials 444 table
residence time 250, 338 biochemical half-cells 442-443
resistance 109 conductivities, mobilities and
resistivity 109 diffusion coefficients 136
reversibility formal potentials 444
operational definition 28 half-cell potentials 438-441
reversibility, criteria for ion atmosphere thickness 66
a.c. polarography 360 ionic conductivity 445
cyclic voltammetry 188, 348 Laplace transforms 450
steady-state voltammetry 158, 345 list of symbols 434-437
rotating platinum electrode 207 parameters for extended Debye-
rotating ring-disk electrode 214 Huckel theory 104
rotating-disk electrode 172, 207-215 physical constants 434
homogeneous reactions at 270 polarographic data 234
Sack effect 124 potential range 447
salting-out effect 69-71 reference electrode potentials 444
sedimentation potential 82 SI units 433
silver cell 47 solvent properties 446
silver-zinc cell 58 Stokes law and crystal radii 121
silver/silver chloride electrode 32, 170 Tafel plot 325-326
sodium-sulfur cell 53 tast polarography 201
solvents thermodynamics 5
choice of 168-169 thickness
properties 169 ion atmosphere thickness 66
table of properties 446,447 diffusion layer thickness 158, 195,
spectroelectrochemistry 257 210
electron spin resonance 258-264 hydrodynamic layer 209
infrared spectroscopy 265-268 reaction layer thickness 251, 270
Subject Index 483
time scales for experiments 250 diffusion coefficient 130
titration for variables used in text 434-437
amperometric titration 177 heterogeneous rate constants 159,
biamperometric titration 388 316
conductometric titration 128 kinematic viscosity 208
coulometric titration 385 resistivity 109
dead-stop titration 388 SI units 433
Karl Fischer titration 387 van der Waals attraction 69
potentiometric 42, 176 vapor pressure 169
Tomes:riteria for reversibility 158, 345 table 446
transfer coefficient 317-323, 324, 333 viscosity 169
transference numbers 116-118 table 446
Hifforf method 116-117 voltammetry
moving boundary method 117 differential pulse 204
transition state theory 316 linear potential sweep 183
transition time 182 square wave 206
transport impedance 355 see steady-state voltammetry, cyclic
transport processes 128 voltammetry
ultrasonic vibration potential 124 Warburg impedance 355
units Wheatstone bridge 112
cell potential and free energy 7 Wien effect 124
concentration 62 work
conductivity 109 electrical work 5-6
surface tension - area work 86
zeta potential 75, 79-80, 83

978 94 011 0691 7

Uploaded by

Copyright:

Available Formats

978 94 011 0691 7

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

978 94 011 0691 7

Uploaded by

Copyright:

Available Formats

HICTROCHIMIHRY

SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.

Library of Congress Cataloging-in-Publication Data

British Library Cataloguing in Publication Data available

4.5 Polarography 194

It has been fashionable to describe electrochemistry as a discipline

Origins of Electrode Potentials

When a piece of metal is immersed in an electrolyte solution, an

Figure 1.2 The Daniell Znso 4 CUS04

careful work, it is necessary to take explicit account of solid and liquid

Luigi Galvani (1737-1798) was a physiologist at the University of Bologna.

William Nicholson (1753-1815) started his career as an East India

Sir Humphry Davy (1778-1829) was Professor of Chemistry at the Royal

and that water was decomposed by electrolysis. The Nicholson-Carlisle

! The early history of electrochemistry is brilliantly expounded in Ostwald's 1896

1.2 ELECTROCHEMICAL CELL THERMODYNAMICS

Since the most obvious feature of a galvanic cell is its ability to

The first law of thermodynamics may be stated as

Now let us see how electrical work is related to the experimentally

Figure 1.3 Electrochemical cell cell

surroundings), represented by a resistance R. When a charge Q is

1 In Chapter 2, where we will be dealing with electric potential in a slightly different

changes in the Gibbs free energy without recourse to measuring

Electrochemical Cell Conventions

The Weston cell was developed in 1893 by Edward Weston (1850-1936), an

Look now at the chemical processes going on at the two electrodes.

Figure 1.4 The Weston cell.

Let us see if this is consistent. If the Hg electrode is positive, then

Activities and Activity Coefficients

!J.G = !J.Go + RT In (ac)'Y(aD)O (1.12)

the arguments of logarithms in expressions such as eq (1.12) are also

The Nernst Equation

If we take eq (1.11) to be the overall reaction of an electrochemical

-nFE = /)"Go + RT In (ac)'Y(aDyi (1.16)

Hermann Walther Nernst (1864-1941) was Ostwald's assistant at the

As we have already seen, it is possible to think of the operation of a

Figure 1.5 Hypothet-

electrochemistry.1 The hydrogen electrode, shown in Figure 1.6, consists

Figure 1.6 The hydrogen electrode.

chemists have built up extensive tables of half-cell potentials. A

1.3 SOME USES OF STANDARD POTENTIALS

With the data of Table A.4 or more extensive collections of half-cell

Many half-cell reactions in aqueous solutions involve H+, OH-, weak

Potentials, Free Energies, and Equilibrium Constants

Building a hypothetical cell from two half-cells is straightforward

Example 1.1 Compute EO and fl.Go for the cell

Referring to Table A.4, we find the following data:

K =8.6 x 107 = [Fe2+J2[I2]

Example 1.2 Given the half-cell potentials

Subtraction of the first half-cell reaction from the second gives

While potentials of half-cells are simply subtracted to compute the

so that when half-cell potentials are combined to produce a new half-cell,

Example 1.3 Compute the standard half-cell potential for the

Wendell M. Latimer (1893-1955) was a student of G. N. Lewis and later a

-0.86 0.87 -0.46 0.76 0.94 -3.04 0.73 0.1

Example 1.4 Compute the half-cell potential for the reduction

The desired half-cell reaction is the sum of the following:

!lGo = - 2F(0.94) + 2.303 RT (3.3 + 28.00)

Free Energy - Oxidation State Diagrams