Está en la página 1de 366

Scientific Fundamentals and Technological Applications


233 Spring Street, New York, New York 10013-1578

~. "


0-306-45736-9 90000

Electrochemical Supercapacitors
Scientific Fundamentals and Technological Applications
B. E. Conway
Fellow of the Royal Society of Canada University of Ottawa Ottawa, Ontario, Canada

Kluwer Academic / Plenum PubJish~

New York, Boston, Dordrecht, London, MOSC~

To my son, Dr. Adrian and his sons, Alexander and the "Little B"
Library of Congress Cataloging-in-Publication Data

Conway. B. E. Electrochemical supercapacitors scientific technological applications I B.E. Conway.




Includes bibl iographical references and index. ISBN 0-306-45736-9 1. Storage batteries. 2. Electrolytic capacitors. double layer. I. Title. TK2941.C66 1999 621.31·2424--dc21

3. Electric

98-48209 CIP

ISBN 0-306-45736-9 © 1999 Kluwer Academic / Plenum Publishers, New York 233 Spring Street, New York, N.Y. 10013 10987654321 A C.I.P. record for this book is available from the Library of Congress. All rights reserved No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording, or otherwise, without written permission from the Publisher Printed in the United States of America


The first model for the distribution of ions near the surface of a metal electrode was devised by Helmholtz in 1874. He envisaged two parallel sheets of charges of opposite sign located one on the metal surface and the other on the solution side, a few nanometers away, exactly as in the case of a parallel plate capacitor. The rigidity of such a model was allowed for by Gouy and Chapman independently, by considering that ions in solution are subject to thermal motion so that their distribution from the metal surface turns out diffuse. Stern recognized that ions in solution do not behave as point charges as in the Gouy-Chapman treatment, and let the center of the ion charges reside at some distance from the metal surface while the distribution was still governed by the Gouy-Chapman view. Finally, in 1947, D. C. Grahame transferred the knowledge of the structure of electrolyte solutions into the model of a metal/solution interface, by envisaging different planes of closest approach to the electrode surface depending on whether an ion is solvated or interacts directly with the solid wall. Thus, the Gouy-Chapman-Stern-Grahame model of the so-called electrical double layer was born, a model that is still qualitatively accepted, although theoreticians have introduced a number of new parameters of which people were not aware 50 years ago. Irrespective of the structural details, it has long been accepted that a double layer exists at the electrode/electrolyte solution boundary, which governs adsorption phenomena and influences charge transfer reaction rates, and where electrostatic energy is stored as in a capacitor a few molecular diameters thick. Nevertheless, the existence of a double layer has always been inferred from indirect observations of related properties and quantities, but never directly





probed, so much that it was compared to the Arab Phoenix: "Everybody says it exists, nobody knows where it is." This until recently, when it was realized that the energy stored per unit surface area of an electrode is noticeable per se and becomes technologically very interesting with the introduction of new materials with an exceptionally extended active surface: especially treated carbons, some transition metal oxides, electrosynthesized conducting polymers. The interfacial capacity is further increased if the purely capacitive charge is supplemented by a Faradaic charge related to bidimensional redox reactions or tridimensional intercalation processes. "Supercapacitors" are devices that store electrical energy on the basis of the above phenomena and that can be discharged at a much higher rate than conventional batteries. They have aroused interest for various applications, including electric vehicles, in particular cars as well as trains. I should say that in spite of our awareness of the principles, supercapacitors have appeared on the scientific scene rather suddenly, or at least this has been the impression of those who have realized that something was happening at the technological level. Of the many examples we can produce of innovations developed in technology first and then "discovered" from a fundamental point of view, supercapacitors furnish an authoritative example of the reverse: a technological innovation pushed by fundamental knowledge. The situation is now that fundamental researchers know everything of the electrical double layer but ignore its application to supercapacitors, while engineers know of supercapacitors but may ignore the fundamentals of their operation. This monograph comes at an opportune time to fill this gap, with a balanced presentation of fundamentals aimed at applications, and applications related to fundamental principles. B. E. Conway has worked for more than 50 years in almost all areas of electrochemistry, particularly interfacial electrochemistry. He is therefore a "veteran" in the field, being the first to realize the potentialities of some materials for their double layer energy. This volume offers what cannot be found in any other work for its comprehensiveness, exhaustiveness, and focus. For the first time a highly theoretical topic, the electrical double layer at electrodes, is shown to manifest itself in highly technological applications. It is with a real sense of pride that electrochemists in the near future will press the accelerator in their electric car knowing that certain performances are possible only thanks to the discharge of the "socalled" (but is it indeed there?) electrical double layer of which technologists have long maintained "electrochemistry can do without it." The content of this book is useful both for scientists working in fundamental research and technologists, in particular those interested in electrochemical energy conversion and chemistry and physics of electrified interfaces, as well as for engineers working in the field of electrochemical power sources and electrical energy storing devices. They will find the book an invaluable source of in-

formation and inspiration. For the way the topics are presented, people working in the area of materials chemistry and physics will find this book of great general interest in view of the typical dependence of the performance of supercapacitors on the structure of materials. Milan, Italy Sergio Trasatti


Systems for electrochemical energy production originated with Volta's discovery in 1800 of "voltaic electricity" and were developed in various forms during the nineteenth century. Toward the end of that period, reversibly chargeable batteries for electrical energy storage and utilization became a major development in applied electrochemistry and during the present century have been improved to a high state of the art. They also represent a large fraction of the economic activity in industrial electrochemistry. In relatively recent years, but originating with Becker's patent in 1957, a new type of electrochemically reversible energy storage system has been developed that uses the capacitance associated with charging and discharging of the double layer at electrode interfaces or, complementarily, the pseudocapacitance associated with electrosorption processes or surface redox reactions. In the first case, large interfacial capacities of many tens of farads per gram of active electrode material can be achieved at high-area carbon powders, fibers, or felts, while, in the second case, large pseudo capacitances can be developed at certain high-area oxides or conducting polymers where extents of Faradaic charge (Q) transfer are functionally related to the potential of the electrode (V), giving rise to a derivative corresponding to a capacitance dQ/dV. These large specific-value capacitors, especially of the double-layer type, are perceived as electrical energy storage systems that can offer high power-density in discharge and recharge, and cycle lives on the order of 105 to 106, many times those of conventional batteries. A variety of uses of such electrochemical or socalled "supercapacitors" are now recognized and a new direction of power-source development, complementary to that of batteries, is well established.





An important aspect of this monograph is that it gives a comprehensive account of the electrochemical science and technology of these capacitor systems. An attempt is made to present a self-contained and unified treatment of the field, including essential details of the background science (e.g. of double-layer capacitance and the origins of pseudocapacitance, the electrolyte solutions used in electrochemical capacitors) as well as basic concepts of electrode kinetics and interfacial electrochemistry, dielectric polarization theory, porous electrodes, and conducting polymer materials that give rise to large specific capacitances. In this way, understanding and study of the material presented in this volume will not require frequent reference to other textbooks of physical chemistry or electrochemistry. The text contains many illustrative diagrams and cross-references between chapters, and includes many literature references. For the convenience of the reader, three or four diagrams have been duplicated from one chapter or another to avoid the necessity of seeking earlier or later pages in the volume where cross-referenced material is cited. The author's work in this field originated with a research contract between Continental Group Inc. and the University of Ottawa's Electrochemistry Group. We would like to acknowledge here the work carried out by Drs. H. AngersteinKozlowska, V. Birss, J. Wojtowicz, and Visiting Professor S. Hadzi-Jordanov (University of Skopje) with Mr. Dwight Craig (electrical engineer) of Continental Group in the period 1975 to 1981. More recently, new work in this field is being carried out at the University of Ottawa and is supported by the Natural Sciences and Engineering Research Council of Canada. For this work, acknowledgment is made to Dr. W. J. Pell and Mr. T. C. Liu. Special thanks are due to Dr. B. V. Tilak of Occidental Chemical Corp., N.Y., for his critical reading of the manuscript before its submission for publication, and for his suggestions for additions and revisions. Appreciation is expressed to Dr. Tilak and Dr. S. Sarangapani (lCET Inc., Norwood, Mass.) for their detailed examination of Chapter 20 on technology development, and in particular for their suggestions for the best systematic organization of the manifold aspects of the subject treated in that chapter. Thanks are also due to Drs. S. Gottesfeld (Los Alamos National Laboratory) and J. Miller (J. M. Inc., Shaker Heights, Ohio) for reading the chapters on conducting-polymer capacitors and ac impedance, respectively. We are grateful to Dr. Miller for permission to reproduce some of his computer-generated graphs and data on ac impedance evaluation of capacitors. The author is also most grateful to Drs. S. P. Wolsky and N. Marincic for their permission to draw on various diagrams and tables from the proceedings of papers presented at the seminars on electrochemical capacitors held at Deerfield Beach and Boca Raton, Fla, over the period 1991 to 1997, under the auspices of Florida Educational Seminars Inc. (abbreviated as FES in the text).

Finally, special thanks are due to Denise Angel, who typed, with great efficiency and accuracy, all the chapters of this volume in several drafts, exercising literacy and care that would be difficult to match. Grateful thanks are also due to Eva Szabo for drafting most of the diagrams. Ottawa, Canada B. E. Conway



Chapter 1

Introduction and Historical Perspective 1.1. Historical Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Scope of the Monograph. . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Similarities and Differences between Supercapacitors and Batteries for Storing Electrical Energy 2.1. Introduction................................. 2.1.1. Energy Storage Systems. . . . . . . . . . . . . . . .. 2.1.2. Modes of Electrical Energy Storage by Capacitors and Batteries. . . . . . . . . . . . . . . .. 2.2. Faradaic and Non-Faradaic Processes. . . . . . . . . . . . . 2.2.1. Non-Faradaic......................... 2.2.2. Faradaic............................. 2.3. Types of Capacitors and Types of Batteries. . . . . . . . 2.3.1. Distinguishable Systems. . . . . . . . . . . . . . . . . 2.3.2. Cell Design and Equivalent Circuits. . . . . .. 2.4. Differences of Densities of Charge Storage in Capacitors and Batteries . . . . . . . . . . . . . . . . . . . . . . . 2.4.1. Electron Densities per Atom or Molecule. . . 2.4.2. Comparison of Energy Densities Attainable in Electrochemical Capacitors and Batteries. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2.5. Comparison of Capacitor and Battery Charging Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . .. xv

1 8 9

Chapter 2

11 11 12 13 14 14 15 15 17 18 18

19 20





2.6. Comparison of Charge and Discharge Behavior of Electrochemical Capacitors and Battery Cells Evaluated by Cyclic Voltammetry 2.7. Li Intercalation Electrodes-A Transition Behavior 2.8. Charging of a Nonideally Polarizable Capacitor Electrode 2.9. Comparative Summary of Properties of Electrochemical Capacitors and Batteries References ................................. General Reading References Chapter 3 Energetics and Elements of the Kinetics of Electrode Processes 3.1. Introduction 3.2. Energetics of Electrode Processes 3.3. Energy Factors in Relation to Electrode Potential 3.4. Kinetics of Electrode Reactions at Metals 3.4.1. Currents and Rate Equations 3.4.2. Linearization of the Butler-Volmer Equation for Near-Equilibrium Conditions
(low s)

4.2.1. . . . . . 22 25 28 29 4.3. 4.4. 4.5. 4.6. 4.7. 4.8. 4.9.

. 31

. . .. . .

33 34 37 41 41

45 46 48 4.10.

3.5. Graphical Representation of the Exchange Current Density, ia> and Behavior Near Equilibrium 3.6. Onset of Diffusion Control in the Kinetics of Electrode Processes ........................... 3.7. Kinetics when Steps Following an Initial Electron Transfer Are Rate Controlling 3.8. Double-Layer Effects in Electrode Kinetics 3.9. Electrical Response Functions Characterizing Capacitative Behavior of Electrodes 3.10. Instruments and Cells for Electrochemical Characterization of Capacitor Behavior 3.10.1. Cells and Reference Electrodes 3.10.2. Instruments ........................... 3.10.3. Two-Electrode Device Measurements References .................................. General Reading References Chapter 4

. . . . . . .

50 51 53 59 59 61 63 64 64 4.11. 4.12.

Coulomb's Law: Electric Potential and Field, and the Significance of the Dielectric Constant . Units . Dielectric Constant . Electrostatic Potential, Field, and Force . Potential e and Field E at an Ion .. Lines of Force and Field Intensity-A Theorem . Capacity of a Condenser or Capacitor. . Field Due to a Surface of Charges: Gauss's Relation Poisson's Equation: Charges in a 3-Dimensional Medium . The Energy of a Charge . Electric Tension in a Dielectric in a Field . Electric Polarization Responses at the Molecular Level . 4.9.1. Atoms and Molecules in Fields: Electronic Polarization . 4.9.2. Interaction of a Permanent Dipole with a Field Uniform Field . Nonuniform Field . Forces on a Quadrupole in a Field .. Atoms and Molecules in Fields: Dielectric Properties and Dielectric Polarization . 4.10.1. Dielectrics . 4.10.2. Polarization of Solvent Molecules in Double-Layer and Ion Fields . 4.10.3. Dipole Moments of Complex Molecules . Electric Polarization in Dielectrics . Energy and Entropy Stored by a Capacitor . References . General Reading References .

68 68 70 71

73 74 74 75 76 77 78 78 79 79 79 80 81 81 81 82 83 83 86 86

Chapter 5

Elements of Electrostatics Involved in Treatment of Double Layers and Ions at Capacitor Electrode Interphases 4.1. Introduction 4.2. Electrostatic P~i~~i~i~s' : : : : : : : : : : : : : : : : : : : : : : : :

67 68

Behavior of Dielectrics in Capacitors and Theories of Dielectric Polarization 5.1. Introduction 5.2. Definitions and Relation of Capacitance to Dielectric Constant of the Dielectric Medium 5.3. Electric Polarization of Dielectrics in a Field 5.4. Formal Electrostatic Theory of Dielectrics

. . . .

87 88 91 92





5.5. 5.6. 5.7. 5.8. 5.9.

Dielectric Behavior Due to Induced, Distortional Polarization ................................. Dielectric Polarization in a Simple Condensed Phase ...................................... Dielectric Polarization in a System of Noninteracting but Orientable Dipoles ......................... Dielectric Polarization of Strongly Interacting Dipoles (High Dielectric Constant Solvents) ....... Dielectric Behavior of the Solvent in the Double Layer ................................ References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..

98 98 99

102 104

Chapter 6

The Double Layer at Capacitor Electrode Interfaces: Its Structure and Capacitance 6.1. Introduction................................. 6.2. Models and Structures of the Double Layer. . . . . . .. 6.3. Two-Dimensional Density of Charges in the Double Layer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 6.4. Ionic Charge Density and Interionic Distances on the Solution Side of the Double Layer. . . . . . . . . . . . . .. 6.5. Electron-Density Variation: "Jellium" Model. 6.6. Electric Field across the Double Layer. . . . . . . . . . .. 6.7. Double-Layer Capacitance and the Ideally Polarizable Electrode 6.8. Equivalent Circuit Representation of Double-Layer Electrical Behavior. . . . . . . . . . . . . . . . . . . . . . . . . .. References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Theoretical Treatment and Modeling of the Double Layer at Electrode Interfaces 7.1. Early Models 7.2. Treatment of the Diffuse Layer. . . . . . . . . . . . . . . . .. 7.3. Capacitance of the Diffuse Part of the Double Layer 7.4. Ion Adsorption and the Treatment of the Compact or Helmholtz Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 7.4.1. Stern's Treatment. . . . . . . . . . . . . . . . . . . . .. 7.4.2. Quasi-Chemical Aspect of Anion Adsorption 7.5. The Solvent as Dielectric of the Double-Layer Capacitor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 7.5.1. General 7.5.2. Types of Solvents that Constitute the Double-Layer Interphase. . . . . . . . . . . . . . ..

105 108 114 116 117 119 121 123 124 Chapter 8 125 127 129 133 133 135 136 136 137 Chapter 9

Dielectric Constant in the Double-Layer Interphase ............................ 7.5.4. Electrostatic Polarization of Water as Solvent in the Double Layer. ............. 7.5.5. Molecular-Level Treatments of Solvent Dipole Orientation at Charged Interfaces .... Two-State Dipole Orientation Treatments ................... Cluster Models for Water Adsorption and Orientation ...... H-Bonded Lattice Models ............... 7.5.6. 7.5.7. Spontaneous Orientation of Water at Electrode Surfaces Due to Chemisorption ... 7.5.8. Solvent Adsorption Capacitance at Solid Metals ............................... 7.5.9. Recent Modeling Calculations ............ 7.6. The Metal Electron Contribution to Double-Layer Capacitance ................................. 7.6.1. Origin of the Metal Contribution .......... 7.6.2. Profile of Electron Density at Electrode Surfaces .............................. The Potential Profile across the Diffuse Layer ...... 7.7. 7.8. The Double Layer in Pores of a Porous Capacitor Electrode ................................... References .................................. General Reading References .................... 7.5.3. Behavior of the Double Layer in Nonaqueous Electrolytes and Nonaqueous Electrolyte Capacitors 8.1. Introduction......... . . . . . . . . . . . . . . . . . . . . . . .. 8.2. Fundamental Aspects of Double-Layer Capacitance Behavior in Nonaqueous Solvent Media. . . . . . . . . .. 8.3. Comparative Double-Layer Capacitance Behavior in Several Nonaqueous Solutions. . . . . . . . . . . . . . . . .. 8.4. General Outlook . . . . . . . . . . . . . .. References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..

138 139 141 141 143 148 149 151 152 156 156 157 160 161 165 168

Chapter 7

169 170 176 180 180

The Double Layer and Surface Functionalities at Carbon 9.1. Introduction ···· . 183 183 9.1.1. Historical ·········· 9.1.2. Carbon Materials for Electrochemical Capacitors . 185





9.2. Surface Properties and Functionalities of Carbon Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 9.3. Double-Layer Capacitance of Carbon Materials. . . .. 9.4. Oxidation of Carbon. . . . . . . . . . . . . . . . . . . . . . . . .. 9.5. Surface Specificity of Double-Layer Capacitance Behavior at Carbon and Metals. . . . . . . . . . . . . . . . .. 9.6. Double-Layer Capacitance at Edge and Basal Planes of Graphite. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 9.7. Materials Science Aspects of Carbon Materials for Conditioned Double-Layer Capacitors 9.7.1. Heat and Chemical Treatments of Carbon Materials for Capacitors. . . . . . . . . . . . . . . .. 9.7.2. Research Requirements for Carbon Materials in Electrochemical Capacitors. . . . . . . . . . .. 9.7.3. Electron Spin Resonance Characterization of Free Radicals at Carbon Surfaces. . . . . . . . .. 9.8. Interaction of Oxygen with Carbon Surfaces. . . . . .. 9.9. Electronic Work Function and Surface Potentials of Carbon Surfaces 9.10. Intercalation Effects. . . . . . . . . . . . . . . . . . . . . . . . . .. References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. General Reading References. . . . . . . . . . . . . . . . . . .. Chapter 10 Electrochemical Capacitors Based on Pseudocapacitance 10.1. Origins of Pseudocapacitance . . . . . . . . . . . . . . . . . .. 10.2. Theoretical Treatments of Pseudocapacitance (C¢) .. 10.2.1. Types of Treatment. . . . . . . . . . . . . . . . . . . .. 10.2.2. Electrosorption Isotherm Treatment of Pseudocapacitance: A Thermodynamic Approach . . . . . . . . .. 10.3. Kinetic Theory of Pseudocapacitance 10.3.1. Electrode Kinetics under Linearly Time- Variant Potential. . . . . . . . . . . . . . . . .. 10.3.2. Evaluation of Characteristic Peak Current and Peak Potential Quantities. . . . . . . . . . . .. 10.3.3. Transition between Reversibility and Irreversibility .. . . . . . . . . . . . . . . . . . . . . . .. 10.3.4. Relation to Behavior under de Charge and Discharge Conditions. . . . . . . . . . . . . . . . . .. 10.4. Potential Ranges of Significant Pseudocapacitances

186 193 196 198 199 203 203 208 209 212 213 217 219 220

10.5. Origin of Redox and Intercalation Pseudocapacitances . . . . . . . . . . . . . . . . . . . . . . . . . .. 248 10.6. Pseudocapacitance Associated with Specific Adsorption of Anions and the Phenomenon of Partial Charge Transfer. . . . . . . . . . . . . . . . . . . . . . .. 253 10.7. Pseudocapacitance Behavior at High-Area Carbon Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 255 10.8. Procedures for Distinguishing Pseudocapacitance (C¢) from Double-Layer Capacitance (Cd]) 255 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 256 General Reading References. . . . . . . . . . . . . . . . . . .. 257 Chapter 11 The Electrochemical Behavior of Ruthenium Oxide (Ru02) as a Material for Electrochemical Capacitors 11.1. Historical Aspects . . . . . . . . . . . . . . . . . . . . . . . . . .. 11.2. Introduction......................... . . . . . .. 11.3. Formation of Ru02 Films that Have Capacitative Properties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 11.4. The Transition from Monolayer to Multilayer Electrochemical Formation of Ru02. . . . . . . . . . . .. 11.5. States and Chemical Constitution of Electrochemically and Thermochemically Formed Ru02 for Capacitors . . . . . . . . . . . . . . .. 11.6. Mechanism of Charging and Discharging Ru02. . .. 11.7. Oxidation States Involved in Voltammetry of Ru02 and Ir02 Electrodes. . . . . . . . . . . . . . . . . . . .. 11.7.1. Oxidation States and Redox Mechanisms .. 11.7.2. Charging in Inner and Outer Surface Regions of Ru02 Films. . . . . . . . . . . . . . . .. 11.8. Conclusions on Mechanisms of Charging Ru02 Capacitor Materials. . . . . . . . . . . . . . . . . . . . . . . . .. 11.9. Weight Changes on Charge and Discharge. . . . . . .. 11.10. de and ac Response Behavior of Ru02 Electrochemical Capacitor Electrodes

259 264 265 267

270 276 277 277 279 282 284 285

221 224 224

224 236 236 239 241 243 246

11.11. Other Oxide Films Exhibiting Redox Pseudocapacitance Behavior. . . . . . . . . . . . . . . . . .. 286 11.12. Surface Analysis and Structure of RU02- TiOz Films 290 11.13. Impedance Behavior of Ru02- Ti02 Composite Electrodes . . . . . . . . .. 292 11.14. Use and Behavior of Irfrj 293





1l.15. Comparative Oxide Film Behavior at Transition Metal Electrodes. . . . . . . . . . . . . . . . . . . . . . . . . . .. References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. General Reading References. . . . . . . . . . . . . . . . . .. Chapter 12 Capacitance Behavior of Films of Conducting, Electrochemically Reactive Polymers 12.1. Introduction and General Electrochemical Behavior 12.2. Chemistry of the Polymerization Processes ....... 12.3. General Behavior in Relation to Pseudocapacitance ........................... 12.4. Forms of Cyclic Voltammograms for Conducting Polymers .................................. 12.5. Classification of Capacitor Systems Based on Conducting Polymer Active Materials ........... 12.6. Complementary Studies Using Other Procedures ... 12.7. Ellipsometric Studies of Conducting Polymer Film Growth and Redox Pseudocapacitative Behavior ... 12.8. Other Developments on Conducting Polymer Capacitors ................................. References ................................. General Reading References and Tabulations ..... The Electrolyte Factor in Supercapacitor Design and Performance: Conductivity, Ion Pairing and Solvation 13.1. Introduction ................................ 13.2. Factors Determining the Conductance of Electrolyte Solutions. . . . . . . . . . . . . . . . . . . . . . . .. 13.3. Electrolyte Conductance and Dissociation 13.4. Mobility of the Free (Dissociated) Ions . . . . . . . . .. 13.5. Role of the Dielectric Constant and Donicity of the Solvent in Dissociation and Ion Pairing. . . . . . . . .. 13.6. Favored Electrolyte-Solvent Systems ..... . . . . .. 13.6.1. Aqueous Media. . . . . . . . . . . . . . . . . . . . . .. 13.6.2. Nonaqueous Media 13.6.3. Molten Electrolytes . . . .. 13.7. Properties of Solvents and Solutions for Nonaqueous Electrochemical Capacitor Electrolytes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 13.8. Relation of Electrolyte Conductivity to Electrochemically Available Surface Area and Power Performance of Porous Electrode Supercapacitors

293 295 297

299 304 312 314 320 322 327 331 332 334

13.9. Separation of Cations and Anions on Charge and Its Effect on the Electrolyte'S Local Conductivity 13.10. The Ion Solvation Factor. 13.11. Compilations of Solution Properties. . . . . . . . . . . .. 13.12. Appendix: Selection of Experimental Data on Properties of Electrolyte Solutions in Nonaqueous Solvents and Their Mixtures. . . . . . . . . . . . . . . . . .. 13.12.1. Summary Tables 13.12.2. Some Graphically Represented Data from the Literature. . . . . . . . . . . . . . . . . . . . . . .. 13.12.3. Selected Tabulations , 13.12.4. Conductivities....................... References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. General Reading References. . . . . . . . . . . . . . . . . .. Chapter 14 Electrochemical Behavior at Porous Electrodes; Applications to Capacitors 14.1. Introduction........... . . . . . . . . . . . . . . . . . . . . .. 14.2. Charging and Frequency Response of RC Networks 14.3. General Theory of Electrochemical Behavior of Porous Electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . .. 14.3.1. System Requirements. . . . . . . . . . . . . . . . . .. 14.3.2. The de Levie Model and its Treatment. . . .. 14.3.3. Configuration of Double Layers in Porous Electrodes. . . . . . . . . . . . . . . . . . . . . . . . . . .. 14.4. Porous Electrode Interfaces as Fractal Surfaces . . . .. 14.5. Atom Densities in Surfaces and Bulk of Fine Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 14.6. Pore Size and Pore-Size Distribution 14.7. Real Area and Double-Layer Capacitance 14.8. Electro-osmotic Effects in Porous Electrodes References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Energy Density and Power Density of Electrical Energy Storage Devices 15.1. Ragone Plots of Power Density vs. Energy Density .. 15.2. Energy Density and Power Density, and Their Relationship 15.2.1. General Considerations 15.2.2. Power Density . . . . . . . . . . . . . .. 15.2.3. Relation to Energy Density 15.2.4. Power and Energy Density Relationships for Capacitors

361 362 365

366 366 366 366 373 374 375

377 380 383 383 383 403 405 406 408 411 415 416

Chapter 13

335 337 338 343 344 345 345 347 350

Chapter 15

417 421 421 425 427 433



-xxiv Contents Contents xxv

15.2.5. Power Density Rating of a Capacitor ....... 15.3. Power Limitation Due to Concentration Polarization 15.4. Relation between C-Rate Specification and Power Density ..................................... 15.4.1. Formal Definition ...................... 15.4.2. Significance of C-Rate in Battery and Capacitor Discharge .................... 15.5. Optimization of Energy Density and Power Density 15.5.1. Capacitor-Battery Hybrid Systems ....... 15.5.2. Condition for Maximum Power Delivery .. 15.5.3. Test Modes .......................... 15.5.4. Constant Power Discharge Regime for a Capacitor. ........................... 15.5.5. Effects of Temperature ................. 15.6. The Entropy Component of the Energy Held by a Charged Capacitor ........................... 15.7. Energy Density of Electrolytic Capacitors ........ 15.8. Some Application Aspects of Power-Density Factors .................................... 15.9. Energy Storage by Flywheel Systems ............ References ................................. Chapter 16 AC Impedance Behavior of Electrochemical Capacitors and Other Electrochemical Systems 16.1. Introduction.................... . . . . . . . . . . . .. 16.2. Elementary Introductory Principles Concerning Impedance Behavior. . . . . . . . . . . . . . . . . . . . . . . . .. 16.2.1. Alternating Current and Voltage Relationships ..... . . . . . . . . . . . . . . . . . . .. 16.2.2. Root-mean-square and Average Currents in ac Studies . . . . . . . . . . . . . . . . . . . . . . . . . . .. 16.3. Origin of the Semicircular Form of Complex-Plane Plots for Z" vs. Z' over a Range of Frequencies .... 16.3.1. Impedance Relationships as a Function of Frequency 16.3.2. Time Constant and Characteristic Frequency Wr . . . . . . . . . . . . . . . . . . . . . . .. 16.4. Significance of RC Time Constants . . . . . . . . . . . .. 16.4.1. Transient Currents and Voltages . . . . . . . .. 16.4.2. Formal Significance of the RC Product as a Time Constant. . . . . . . . . . . . . . . . . . . . . . .. 16.5. Measurement Techniques. . . . . . . . . . . . . . . . . . . ..

436 440 443 443 444 448 448 452 456 459 462 463 464 468 474 475

16.5.1. AC Bridges 16.5.2. Lissajous Figures 16.5.3. Phase-Sensitive Detection Using Lock-in Amplifiers. . . . . . . . . . . . . . . . . . . . . . . . . . .. 16.5.4. Digital Frequency-Response Analyzers (Solartron and Other Instruments) 16.6. Kinetic and Mechanistic Approach to Interpretation of Impedance Behavior of Electrochemical Systems 16.6.1. Procedures and Role of Diffusion Control. .. 16.6.2. Principles of the Kinetic Analysis Method .. 16.6.3. Example of the Kinetic Analysis of ac Behavior of the Cathodic H2 Evolution Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 16.6.4. Relation to Linear-Sweep Modulation and Cyclic V oltammetry . . . . . . . . . . . . . . . . . . .. Methodology ..... . . . . . . . . . . .. Response-Current Behavior Relation between Response Currents in Cyclic V oltammetry and Alternating Voltage Modulation. . . . . . . . . . . . . . . . . .. 16.6.5. Impedance of a Pseudocapacitance References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Chapter 17 Treatments of Impedance Behavior of Various Circuits and Modeling of Double-Layer Capacitor Frequency Response 17.1. Introduction and Types of Equivalent Circuits. . . . .. 17.2. Equivalent Series Resistance. . . . . . . . . . . . . . . . . . .. 17.2.1. Significance of esr . . . . . . . . . . . . . . . . . . . .. 17.2.2. Impedance Limits for Some Commercial Capacitors Due to esr 17.3. Impedance Behavior of Selected Equivalent Circuit Models 17.4. Discharge of a Capacitor with esr into a Load Resistance, RL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 17.5. Simulation of Porous Electrode Frequency Response by Multielement RC Equivalent Circuits. . . . . . . . .. 17.6. Impedance Behavior of a Redox Pseudocapacitance .. 17.7. Electrochemistry References at Porous Electrodes. . . . . . . . . . ..

502 503 504 505 506 506 509

510 513 513 513

515 518 524

479 486 486 489 491 491 496 497 497 501 502

525 528 528 530 532 538 547 549 555 556





Chapter 18

Self-Discharge of Electrochemical Capacitors in Relation to that at Batteries 18.1. Introduction........... . . . . . . . . . . . . . . . . . . . .. 18.2. Practical Phenomenology of Self-Discharge 18.3. Self-Discharge Mechanisms 18.4. Methodologies for Self-Discharge Measurements.. 18.5. Self-Discharge by Activation-Controlled Faradaic Processes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 18.6. Slope Parameters for Decline of Potential on Self-Discharge. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 18.7. Comparison with a Regular Capacitor Discharging through an Ohmic Leakage Resistance. . . . . . . . . .. 18.8. Self-Discharge under Diffusion Control 18.9. Charging of a Nonideally Polarizable Electrode 18.10. Self-Discharge of Double-Layer- Type Supercapacitor Devices 18.11. Time-Dependent Redistribution of Charge in Nonuniformly Charged Porous Electrodes 18.12. Temperature Effects on Self-Discharge 18.13. Self-Discharge of a Pseudocapacitance. . . . . . . . . .. 18.14. Examples of Experimental Measurements on Self-Discharge of Carbon Capacitors and Carbon Fiber Electrodes 18.14.1. Introduction 18.14.2. Potential Decay (Self- Discharge) and Recovery in Terms of a Faradaic Process 18.14.3. Self-Discharge Behavior of a Commercial Capacitor. 18.15. Self-Discharge and Potential Recovery Behavior at an Ru02 Electrode . 18.15.1. Background . 18.15.2. Potential Decay (Self-Discharge) and Recovery in Relation to Charge and Discharge Curves . 18.15.3. Model for Potential Recovery . 18.15.4. Quasi-Reversible Potentials of Ru02 after Self-Discharge . 18.16. Self-Discharge in a Stack . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..

Chapter 19 557 557 559 561 562 567 568 569 573 574 575 578 579

Practical Aspects of Preparation and Evaluation of Electrochemical Capacitors 19.1. Introduction 19.2. Preparation of Electrodes for Small Aqueous Carbon-Based Capacitors for Testing Materials 19.3. Preparation of Rut), Capacitor Electrodes 19.4. Preparation of RuOx Capacitors with a Polymer Electrolyte Membrane (U.S. Patent 5,136,477) 19.5. Assembly of Capacitors 19.6. Experimental Evaluation of Electrochemical Capacitors 19.6.1. Cyclic Voltammetry 19.6.2. Impedance Measurements 19.6.3. Constant Current Charge or Discharge 19.6.4. Constant Potential Charge or Discharge 19.6.5. Constant Power Charge or Discharge 19.6.6. Leakage Current and Self-Discharge Behavior 19.7. Other Test Procedures References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 597 . 598 . 599 . 600 . 600 . . . . . . 602 602 602 603 605 605

. 605 . 606 .. 606

Chapter 20

582 582

583 584 586 586

587 591 592 595 595

Technology Development 20.1. Introduction ········ 20.2. Development of the Technology of Electrochemical Capacitors . 20.2.1. Classes of Capacitors . 20.3. Summaries of Device Developments and Technology Advances . 20.4. Materials Requirements . 20.4.1. Electrodes ·· 20.4.2. Carbon Electrode Materials . 20.4.3. Activation Procedures for Carbon Particles and Fibers ................... 20.4.4. Oxide, Redox-Pseudocapacitance Systems 20.4.5. Conducting-Polymer Electrodes . ....... 20.4.6. Electrolyte Systems .................. 20.4.7. Practical Design Aspects .............. 20.4.8. Capacitor Stacking ................... 20.4.9. Bipolar Electrode Arrangements ........ 20.4.10. Current Distribution in Capacitor Devices 20.4.11. Scale-up Factors ..................... 20.5. State of the Art. .............................

609 610 610 612 613 613 615 615 618 618 618 620 620 622 623 625 627



20.6. 20.7. 20.8.

20.Y. 20.10. 20.11. 20.12. 20.13. 20.14.

20.15. 20.16. 20.17. 20.18.

20.5.1. Electrode Development. . . . . . . . . . . . . . .. 20.5.2. Ruthenium Oxide Materials. . . . . . . . . . . .. 20.5.3. Other Embodiments Self-Discharge: Phenomenological Aspects. . . . . .. Thermal Management Other Variables that Affect Capacitor Performance 20.8.1. Temperature Dependence of Capacitance and Capacitor Performance . . . . . . . . . . . .. 20.8.2. Constant Current versus Constant Potential Charging Modes. . . . . . . . . . . . . . . . . . . . .. 20.8.3. Rate Effects on Charge or Discharge Safety and Health Hazards in the Use of Electrochemical Capacitors. . . . . . . . . . . . . . . . . . .. Recent Advances in the Use of Materials. . . . . . . .. Usage Basis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Commercial Development and Testing. . . . . . . . . .. Capacitor-Battery Hybrid Application for Electric Vehicle Drive Systems. . . . . . . . . . . . . . . . . . . . . .. Market Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 20.14.1. Electrochemical Capacitors in the Capacitor Market . . . . . . . . . . . . . . . . . . .. 20.14.2. Market Status and Future Opportunities .. Technology Summary Based on Patent Literature .. Energy Storage by High-Voltage Electrostatic Capacitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Concluding Summary . . . . . . . . . . . . . . . . . . . . . . .. Appendix on Information Sources References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. General Reading References . . . . . . . . . . . . . . . . . . . . . ..

627 634 635 641 643 644 644 648 649 649 651 655 658 663 666 666 667 667 668 670 671 673 674 675 685

Electrochemical Supercapacitors
Scientific Fundamentals and Technological Applications

Chapter 21 Index

Patent Survey

Chapter 1

Introduction and Historical Perspective



The discovery of the possibility of storing an electrical charge on surfaces arose from phenomena associated with the rubbing of amber in ancient times. Of course, the origin of such effects was not understood until the mideighteenth century in the period when the physics of so-called "static electricity" was being investigated and various "electrical machines" were being developed, such as the Electrophorus and the Wimshurst machine. (Excellent examples of these, as well as Leyden jars, can be seen in the Museum of Science in Florence.) Understanding of electricity at the molecular electronic level did not begin until 140 years later, starting indirectly with the work of Michael Faraday and later with that of J. J. Thomson and of Millikan on the electron. In relation to such historic investigations, the development of the Leyden jar, and the discovery of the principle of charge separation and charge storage on the two surfaces of the Leyden jar, separated by a layer of glass, was of major significance for the physics of electricity and later for electrical technology, electronics, and electrochemical engineering. The discovery of the Leyden jar, referred to in early works and in technological applications up to the middle of this century as the "condenser," is variously attributed either to Dean Kleist at Leyden or almost simultaneously to Musschenbroek at Kamin, Pomerania. In later terminology, the device in various embodiments is referred to as a "capacitor" and its capability (in the extensive sense) for charge storage per volt, is referred to as its "capacitance," C. The word "capacity," used in battery terminology to indicate the extent of Faradaic charge storage (in units of coulombs or watt-hours) should not be confused with "capacitance" (given in units of farads), which applies to capacitors.

Chapter 1


and Historical Perspective

The original Leyden jar consisted of a glass phial containing an aqueous acidic electrolyte as a conductor which was contacted by an immersed electrode, with a metal foil coating on the outside surface of the glass phial, thus providing a dielectric material in between-the glass. Later improvements, leading to modern systems, used glass plates as dielectric materials with metal foil on each surface, or rigid metal plates separated by vacuum or air (air condensers), or by mica or polystyrene insulating films. Various forms of condensers, as illustrated in the Encyclopaedia Britannica, are shown in Fig. 1.1 and include an example of a Leyden jar. For many years there was much uncertainty about the nature of electricity after the separate discoveries of "animal electricity" by Galvani in 1776 and "voltaic electricity" by Volta in 1800, and it remained poorly understood for a long time, notwithstanding the work by Faraday (including his discovery of the chemical equivalence of electrical charge). It was not until the discoveries by J. J. Thomson! on the charge-to-mass ratio of the ubiquitously produced negative charge carriers arising in the ionization oflow-pressure gases (first investigated by Crookes/), and the work by Millikan3 and by Townsend" on the absolute value of the charges borne by such particles that a modern view of the nature of electricity could be proposed in terms of accumulation or deficiency of such charges, and the dynamics of their motion. In 1881 Johnstone Stoney ' coined the name "electron" (from the Greek dSKTOV for amber) for such negatively charged particles, which are the natural units of electricity, the electron charge being 4.80 x 10-10 electrostatic units (esu) of charge or 1.60 x 10-20 absolute electromagnetic units (emu) of charge. It is of historical interest that Faraday himself failed to reach the conclusion that his laws relating the extent of a charge's passage (current x time == coulombs) to quantitatively determinable extents of chemical change associated with the electrolysis of aqueous acid and metal-salt solutions, implied an atomic unit of electricity. However, the importance of his laws is in no way diminished by that difficulty because the laws clearly demonstrated in a quantitative way the equivalence of electrical charge to the extent of a generated chemical change that was dependent on the chemical identity of the element concerned and its equivalent weight or oxidation state in solution. It was only much later, in his Faraday Lecture of 1881, that von Helmholtz6 reached the key conclusion that Faraday's laws implied that a fundamental unit of electrical charge was universally involved. This paved the way for the development of the quantitative and more fundamental science of electrochemistry and for a quantitative science of the electrical nature of matter. This work was further elucidated through spectroscopy and the theoretical advances of Bohr, Sommerfeld, Schroedinger, and Heisenberg on electron energy states in atoms and molecules, and the significance of ionization and its relation to solvolysis in solution and the state of solid salts 7.

Although the experimental phenomenology of electrical charging of surfaces, including those of the Leyden jar condenser, was well understood in the mid- and later part of the eighteenth century, the full physical significance of charging or discharging processes at the plates of capacitors could not have been at all fully understood until the atomic nature of electricity, the electron, was characterized and its properties directly determined.i':" Similarly, the charging or discharging of capacitors by a flow of electrical charges in wires could not have been understood until the electron theory of metals had been developed and the mechanism of current flow in metal conductors was understood. Nor for that matter could the physical and chemical significance of the charging of amber be understood until satisfactory ideas about the ionization of molecules and macromolecules (in that case, through frictional or triboelectrical effects) had been formulated, partly through the results on spectroscopic ionization limits of electron energy states in molecules or atoms. Thus, the mechanisms of electrical charge storage in capacitors remained poorly understood at the atomic physical level until some 140 years after the development ofthe Leyden jar capacitor and related contemporary electrical machines. Nevertheless, it was Faraday who had some of the first (in principle) correct ideas about polarization in dielectrics and the significance of dielectric strain and lines of (electrical) force in the dielectric materials of charged condensers. At this point it should be stated that the charging of metallic plates of a capacitor involves production of an excess (negative plate) or deficiency (positive plate) of the density of the delocalized electron plasma of the metal over a short distance (ca. 0.1-0.2 nm, the Thomas-Fermi screening distance) from the formal outer surface of the metal plate. However, each plate has its own electric potential (an equipotential) throughout its material, except very near its surface. Hence, local charge density variation within the so-called "Thomas-Fermi screening distance" has to arise according to the Poisson relation that expresses the field gradient in terms of the local space charge density and the Gauss relation that expresses the field as a function of surface charge density (see Chapter 4). At an insulator such as amber, the excess charge density that arises upon charging has a different origin that is associated with localized molecular ionization (localized oxidation) of the insulator material at its surface, or in some cases is due to negative ionization by localized uptake of electrons at electron acceptor sites on the surface (localized reduction). These latter phenomena are the subject of "triboelectricity." The principle that electrical energy can be stored in a charged capacitor was known since 1745; at a voltage difference, V, established between the plates accommodating charges +q and -q, the stored energy, G, is 112 CV2 or 112 qV, G being a Gibbs (free) energy which increases as the square ofV.

Chapter 1


and Historical



Chapter 1


and Historical Perspective

The utilization of this principle to store electrical energy for practical purposes, as in a cell or battery of cells seems to have been first proposed and 8 claimed as an original development in the patent granted to Becker in 1957. The patent described electrical energy storage by means of the charge held in the interfacial double layer at a porous carbon material perfused with an aqueous electrolyte. The principle involved was charging of the capacitance, Cdl, of the double layer, which arises at all solid/electrolyte interfaces, such as metal, semiconductor, and colloid surfaces (and also at the phase boundary between two immiscible electrolyte solutions"), Carbon is an element almost uniquely suited for fabrication of electrochemical capacitors of the double-layer type. It exists in several, well-known allotropic forms-diamond, the fullerenes, and graphite; the latter and glassy carbon can be generated in the form of high-area fibers or felts. Amorphous carbons and carbon black are available as high specific-area powders. The fiber or felt materials are particularly convenient for formation of electrode structures having good mechanical integrity, while the high-area powders are more difficult to handle. However, glassy carbon, graphite, and carbon black materials are convenient for forming high-area electrode structures, often on a support matrix. From an electrochemical point of view, carbon is relatively, though not entirely, unreactive and thus has a potential voltage range of almost ideal polarizability (Chapters 6 and 7), approaching 1.0 V in aqueous solution and possibly up to 3.5 V in nonaqueous media. After Becker, the Sohio Corporation in Cleveland, Ohio, also utilized the double-layer capacitance of high-area carbon materials, but in a nonaqueous solvent containing a dissolved tetraalkylammonium salt electrolyte. Such systems provide higher operating voltages (3.4-4.0 V) owing to the greater decomposition voltage of nonaqueous electrolytes than those for aqueous ones. Thus they can accommodate higher charge densities and provide larger specific energy storage since the storable energy increases with the square of the voltage attainable on charge. A different principle, originating from ref. 14, was utilized and developed in 1975 onward to 1981 by Conway'? in Ottawa, under contract with Continental Group. This was based on the concept of D. Craig that was developed at Hooker Corp. Here, in one type of system, the large, so-called "pseudocapacitance," C¢, associated with the potential dependence of extents of electrochemical adsorption of H or monolayer levels of electrodeposition of some base metals (Pb, Bi, Cu) at Pt or Au was used!' as a basis for an energy-storing capacitor. In another type of system, the pseudocapacitance associated with solid oxide redox systems was used, especially that developed over some 1.4 V (practical range 1.2 V) in aqueous H2S04 at RU02 films.12-15 This system approaches almost ideal capacitative behavior, with a large degree of reversibility between charge and discharge, and multiple cyclability over some lOS cycles (see Chapter 11).

Work on the latter type of system has been continued by Pinnacle Research Corp., Cupertino (now at Los Gatos, California), in the former laboratories of Continental Group. Useful military applications have been developed, but the Ru materials required are too expensive for the development of a large-scale capacitor, e.g., for use in hybrid systems with batteries for electric-vehicle motive power. Pseudocapacitance arises whenever, for thermodynamic reasons, there is some continuous dependence of a charge, q, passed Faradaically in oxidation or reduction, upon the electrode potential, V. Thus, a derivative dq/dV can arise that corresponds to a pseudocapacitance that is directly measurable, or utilizable, as a capacitance (Chapter 10). The large capacitances (on the order of several or more farads per gram) that can be developed with the RU02 film system and also with the carbon double-layer-type capacitors led to the terms "supercapacitor" or "ultracapacitor" being coined, respectively, for these two types of high specific capacitance devices. Recently it has been suggested that the more general term "electrochemical capacitors" be used to refer to these systems. However, this name should not be confused with "electrolytic capacitor," the latter term referring to the well-known, moderately high-capacitance device (on the order of tens of millifarads) that is based on a thin- film oxide dielectric formed electrolytically with a gel electrolyte on such metals as Ta, Zr, Ti, or AI. The key practical factor that allows very high capacitances, on the order of farads or tens of farads rather than millifarads, to be achieved in a small volume, say 1 em", is the utilization of high-area materials such as activated porous carbons for which real areas are up to 1000 to 2000 m2 g-I. Similarly, with the RU02 pseudocapacitance system, the material, which is a hydrous oxide, has a quasi-3-dimensional, electronically conducting structure, giving accessibility to protons and electrons'< that are involved in two or three (Chapter 11) successive, reversible oxidation or reduction steps in charge or discharge, respectively. The continuous dependence of the extents of oxidation or reduction on electrode potential (over 1.4 V), with corresponding passage of charge, leads to the high specific redox pseudocapacitance of this material.l'' which is usually coupled with an appreciable double-layer capacitance component (see Chapter 11). The use of high-area carbon or oxide redox systems has led to the commercial production of practical high-capacitance electrochemical capacitor devices such as that developed (the Gold Capacitor) by Matsushita Electric Industrial Co. (Osaka, Japan) and by Pinnacle Research; the latter has been developed mainly for military applications. The commercial products are designed to provide standby power for random access memory (RAM) devices or telephone equipment, as power sources for operating activators, and as elements for long time-constant circuits, etc. An attractive technology employing RU02 in a thin film applied to a Nafion membrane, or a powder treated with Nafion, has been developed by Giner, Inc. (Waltham, Massachusetts) and gives high specific ca-


Chapter 1


and Historical Perspective

pacitance. The design avoids a liquid electrolyte and is analogous to membrane electrolyte fuel cell electrodes. Further details are described in Chapter 11. Recent opportunities for the use and development of larger scale capacitors arise from the possibility of using them in hybrid configurations with secondary batteries in electric vehicle power systems.


It is the aim of this volume to give a comprehensive and to a large extent self-contained account of the development of electrochemical capacitors, covering both the fundamental science (physics, chemistry and electrical engineering principles involved) and, at the end of the book, to provide an overview of recent technology development. Some "tutorial" aspects are also included (Chapters 3 and 5) to provide necessary background information and principles of electrochemistry, e.g., on topics such as the double layer, electrostatics, and electrode kinetics. Chapter 2 gives an overview of similarities and differences between electrochemical capacitor and battery systems for electrical energy storage, with stress on the differences between non-Faradaic and Faradaic electrochemical processes that are involved in these two types of devices. Considerable detailed attention is given to the essential topics of doublelayer capacitance and Faradaic pseudocapacitance. Chapters 6, 7, and 8 give the basic conceptual and theoretical background concerning the phenomenon of double-layer capacitance and its modeling. Chapter 10 gives an account of the complementary phenomenon of pseudocapacitance associated with electrochemical adsorption and redox processes involving charge transfer. Chapter 12 discusses the electrical behavior of conducting polymers as capacitor materials. The electrical response behavior of capacitors to ac and de, and pulse potential modulation signals is treated (Chapters 4, 16, and 17) since the results of such procedures provide the principal means of characterizing capacitor behavior, both fundamentally and in technological evaluation. Basic information on the principles of electrostatics (Chapter 4) and of ac impedance spectroscopy is covered. Material is also included on the characterization of various high-area carbon preparations (Chapters 9 and 14), and of Ru02 (Chapter 11) used for capacitor fabrication. General aspects of energy-density vs. power-density relations (Ragone plots) are treated (Chapter 15), including the results of computer-simulation evaluations of the effects of ohmic and Tafel (kinetic) polarization effects. Since the former effects are closely related to properties of the electrolyte solutions used in electrochemical capacitors of both the aqueous and

nonaqueous, aprotic solution type, Chapter l3 discusses electrolyte solutions and their properties. Chapter 18 covers the important practical problem of self-discharge on open-circuit and is followed by a summary of procedures for preparation and evaluation of electrochemical capacitors (Chapter 19). The volume concludes with an extensive survey of recent technology developments (Chapter 20) in the field. It is taken mainly from presentations made at seminars and conferences on electrochemical capacitors and batteries over the past 7 years. Finally, the principal patent literature is surveyed.

1. 1.1. Thomson, Phil. Mag., 5, 346 (1903); see also 1.1. Thomson, The Electron in Chemistry, Franklin Institute Lectures, Chapman and Hall, London (1923). 2. W. Crookes (1879), quoted by S. Glasstone in Textbook of Physical Chemistry, Van Nostrand, New York (1940). 3. R. A. Millikan, Phys. Rev., 2, 143 (1913). 4. 1. S. Townsend, Electricity in Gases (1879), quoted by Glasstone as in Ref. 2. 5. G. Johnstone Stoney, Phil. Mag., 11, 381 (1881); Sci. Trans. Roy. Soc. Dublin, 4, 583 (1891). 6. H. von Helmholtz, 1. Chern. Soc., Lond., 39, 277 (1881). 7. 8. 9. 10. 11. B. E. Conway, Ionic Hydration in Chemistry and Biophysics, Elsevier, Amsterdam (1981). H. E. Becker, U.S. patent 2,800,616 (to General Electric Co.) (1957). Z. Samec, 1. Electroanal. Chern., 103,1 (1979). B. E. Conway, 1. Electrochem. Soc., 138, 1539 (1991). B. E. Conway and H. A. Kozlowska, Acct. Chern. Res., 14,49 (1981). H. A. Kozlowska, M. Vukovic, and B. E. Conway, 1. Electrochem. B. E. Conway, and H. A. Kozlowska, 1. Electroanal. Soc.,

12. S. Hadzi-Jordanov, 125,1471 (1978). 13. S. Hadzi-Jordanov, (1975).

Chem., 60, 359

14. S. Trasatti and G. Buzzanca, 1. Electroanal. Chern., 29, App. I (1971). 15. R. Galizzioli, F. Tantardini, and S. Trasatti, 1. Appl. Electrochem., 4,57 (1974). 16. B. E. Conway, in Proc. Symp. on Electrochemical Capacitors, F. M. Delnick Tomkiewicz, eds., vol. 95-29,15, Electrochem. Society, Pennington, N.J. (1996).

and M.

Chapter 2

Similarities and Differences between Supercapacitors and Batteries for Storing Electrical Energy



Energy Storage Systems

A modern technological society demands the use and storage of energy on a major scale, employing large and small systems for that purpose. Energy stored as potential energy is involved in hydroelectric systems through the hydrostatic "head" of water behind dams; it is also stored in a potential sense in fuels (e.g., coal, oil, and cryogenic hydrogen) and becomes available, albeit with rather poor efficiency, through combustion utilizing steam-piston, steam-turbine, and internal combustion engines of various kinds as energy transduction devices. Energy may also be stored as rotational kinetic energy in flywheels. Electrical energy can be stored in two fundamentally different ways: (1) indirectly in batteries as potentially available chemical energy requiring Faradaic oxidation and reduction (see Sections 2.4.2 and 2.5) of the electrochemically active reagents to release charges that can perform electrical work when they flow between two electrodes having different electrode potentials (i.e., across the voltage difference between the poles of battery cells); and (2) directly, in an electrostatic way, as negative and positive electric charges on the plates of a capacitor, a process known as non-Faradaic electrical energy storage. The efficiency of these two modes of storing electrical energy is usually substantially


Chapter 2


and Batteries


larger than that of fuel combustion systems, which are limited by thermo~ynamic Carnot cycle considerations while electrochemical syste~s usual~y Involve more reversible processes, with direct conversion of potentially available chemical energy to free (or Gibbs) energy, I1G.
2.1.2. Modes of Electrical Energy Storage by Capacitors and Batteries

An important difference arises between the reversibility. of Faradaic and non-Faradaic systems [(1) and (2)]. In energy storage by capacitors, only an excess and a deficiency of electron charges on the capacitor plates have to be established on charge and the reverse on discharge; no chemical changes are involved. However, with storage of electrochemical energy in battery cells through Faradaic reactions, chemical interconversions of the anode and cathode materials must take place, usually with phase changes. Although the ~verall energy change can be conducted in a relatively reversible th~rmody~a~c v:a~, the charge and discharge processes in a battery often involve lITev.erslblhty In Interconversion of the chemical electrode reagents; thus the cycle hfe of battery cells is usually restricted to one thousand to several thousand discharge and r~charge cycles, depending on the type of battery. By contrast, a hardware capacitor ~as an almost unlimited cyclability since no chemical and phase changes are Involved in its charging and discharging. Ordinary capacitors have, however, a very small amount of charge storage unless they are large, i.e., they have a low energy density for electrical energy storage. However, charged electrode/solution interfaces contain double lay~rs 2 (Chapters 6 and 7) that have capacitances of about 16 to 50 ~F cm- : hen.ce, with the sufficiently large accessible electrode areas that are reahzable with high-area carbon powders, felts, and aerogels, very large double-layer c~pacitan~es ~n the order of 10 to 100 F per gram can be achieved. It is the practical realization of this possibility in recent years (but originating in early development some 35 years ago at Sohio) that has led to the relatively new field of electrochemical capacitors. These are now actively progressing as energy storage devices to complement batteries. Because the charging and discharging of such double-layer capacitors involves no chemical phase and composition changes which, in batteries, lead to materials irreversibility, such capacitors have a high degree of recyclability, on the order of 105_106 times. Only electrons need to be moved to and from the electrode surfaces through the external circuit, and cations and anions of the electrolyte transported within the solution to the charged interfaces. It is for these reasons that capacitor charging and discharging processes are highly reversible. . In the cyclic voltammetry of such systems (see Chapter 10), the chargl~g and discharging voltammograms are almost mirror images of one another, while for battery processes they are rarely of this kind (see Fig. 2.3 later in this chap-

ter). This is a major and characteristic difference between battery and capacitor electrical energy storage systems. It must be emphasized at the outset that there has never been an aim or projection of a possible substitution of batteries by supercapacitors; rather, opportunities arise for complementary operation of electrochemical capacitors that are electrically coupled in discharge and recharge with batteries, while other kinds of applications especially favor capacitor-type behavior, e.g., for backup power systems. Also, there are stand-alone opportunities for using multiply rechargeable electrochemical capacitors in a variety of independent functions, as briefly indicated in Chapter 1. General aspects of the electrochemistry and technology of batteries are to be found, for example, in Refs. 1 and 2. In the early stages of the development of electrochemical capacitor technology and related fundamental work, there was some confusion in the electrochemical engineering field and at symposia about the differences between the properties and operation of a battery and a supercapacitor, and what advantages one might have over the other. This confusion may have been assisted by some groups calling the capacitor devices "ultracapacitors" and others, "supercapacitors," the latter as originated by the Ottawa group in 1975. The present preferred name, proposed by Burke, as referred to in Chapter 1, is now more scientific and generic, namely "electrochemical capacitors." This chapter identifies and explains some of the similarities and differences between electrochemical capacitors and batteries in relation to the electrochemical processes that are involved in their discharge and recharge cycling, and in their potential uses as electrical energy storage devices. In particular, the fundamentally different mechanisms of charge storage that are normally involved will be emphasized, along with the consequent, usually different, relations between the extents of charge accommodated at the electrodes and the electric potential differences (cell voltage) between pairs of electrodes having conjugate, ±, polarities. One of the main and kinetically significant differences between capacitors and batteries is that the electrodes of the latter usually undergo substantial phase changes during discharge and recharge (minimally though for the intercalation systems), which lead to kinetic and thermodynamic irreversibility. On the other hand, capacitors of the double-layer type require only electrostatic charge accommodation with virtually no phase change, though some small but significant reversible electrostriction of the electrolyte can arise upon charging.



There is a general and fundamental difference between the mechanisms of operation of electrochemical capacitors and battery cells: for the double-layer


Chapter 2

Supercapacitors and Batteries


type of capacitor, the charge storage process is non-Faradaic, i.e., ideally no electron transfer takes place across the electrode interface and the storage of electric charge and energy is electrostatic. In battery-type processes, the essential process is Faradaic, i.e., electron transfer does take place across the double layers, with a consequent change of oxidation state, and hence the chemistry of the electroactive materials. Intermediate situations arise where Faradaic charge transfer occurs, but owing to special thermodynamic conditions that apply, the potential, V, of the electrode is some continuous function of the quantity of charge, q, passed so that a derivative, dq/dV, arises. This is equivalent to and measurable as a capacitance and is designated as a pseudocapacitance, as explained in Chapter 10. A somewhat different situation arises when chemisorption of ions or molecules takes place with partial charge transfer.v? e.g., in a process such as M + A- ~

+ Je(in M)


In a capacitor, actual electron charges, either in excess or deficiency, are accumulated on the electrode plates with lateral repulsion and no involvement of redox chemical changes. (However, in some cases of double-layer charging, some partial electron transfer does occur, giving rise to pseudocapacitance, e.g., when chemisorption of electron-donative anions such as Cl", Br", I", or CNS-, takes place as illustrated in Eq. (2.1). The electrons involved in double-layer charging are the delocalized conduction-band electrons of the metal or carbon electrode, while the electrons involved in Faradaic battery-type processes are transferred to or from valence-electron states (orbitals) of the redox cathode or anode reagent, although they may arrive in or depart from the conduction-band states of the electronically conducting support material. In certain cases, the Faradaically reactive battery material itself is metallically conducting (e.g., Pb02, some sulfides, Ru02), or else is a well-conducting semiconductor and a proton conductor, e.g., Ni·O·OH.

Such a reaction at the surface of an electrode M usually gives rise to a potentialdependent pseudocapacitance (see again Chapter 10) and the quantity Je is related to the so-called "electrosorption valence" treated in Refs. 3,4, and 5. To summarize, the important differences in the charge storage processes are as follows: 2.2.1. Non-Faradaic The charge accumulation is achieved electrostatically by positive and negative charges residing on two interfaces separated by a vacuum or a molecular dielectric (the double layer or, e.g., a film of mica, a space of air or an oxide film, as in electrolytic capacitors). 2.2.2. Faradaic The charge storage is achieved by an electron transfer that produces chemicalor oxidation state changes in the electroactive materials according to Faraday's laws (hence the term) related to electrode potential. Pseudocapacitance can arise in some cases. The energy storage is indirect and is analogous to that in a battery. In a battery cell, every electron charge is Faradaically neutralized by charge transfer, resulting in a change of oxidation stage of some redoxelectroactive reagent, e.g., (2.2) in the cathode of an Ni-Cd battery.
I,i 1,1

2.3. TYPES OF CAPACITORS AND TYPES OF BATTERIES 2.3.1. Distinguishable Systems Table 2.1 contains a summary of the types of capacitors and their mode of energy storage: electrostatic or Faradaic, the latter in the case when pseudocapacitance arises (Chapter 10). Types d and f, which are printed in bold, are the principal kinds treated in this volume. Normally, capacitors function as elements of electronic circuits or communications equipment, or as ballast for starting electric motors or electric discharge

TABLE 2.1. Types of Capacitors and Mode of Energy Storage

Type (a) (b) (c) (d) Vacuum Dielectric Oxide electrolytic (thin film) Double-layer Basis of charge or energy storage Electrostatic Electrostatic Electrostatic Electrostatic (charge separation at double-layer at electrode interface) Electrostatic (special double-layer system) Faradaic charge transfer (pseudocapacitance) Faradaic charge transfer (pseudocapacitance) Faradaic charge transfer (pseudocapacitance) Examples Mica, Mylar, paper Ta20s, Al203 C preparations, powders, fibers Undeveloped RuOx,

(e) Colloidal electrolyte (f) Redox oxide film (g) Redox polymer film (h) Soluble redox system



polyaniline, polythiophenes Fe(CN):--Fe(CN)~-, V2+N3+N02+




Chapter 2


and Batteries


tubes (fluorescent lights). As has been explained earlier, devices of very large capacitance are now available for storing electric energy in various applications. Table 2.2 summarizes the types of batteries currently extant.I,2 These are generally classified as primary (nonrechargeable) or secondary (multiply rechargeable). The discharge or recharge mechanism is mainly Faradaic, although all electrode interfaces exhibit a double-layer capacitance that is reversibly chargeable. For batteries the latter mechanism accommodates about 2-5% of the total charge accepted. In a different class from the battery systems listed in Table 2.2 are fuel cells in which the anode and cathode (02 or air) reactants are supplied on a continuous basis from external reservoirs, and the electrode surfaces provide an interface for either electrocatalytic oxidation or reduction of the reagents supplied. The primary metal-air cells are operated as semi-fuel cells, but the "fuel" is an easily oxidizable base metal and a gas-diffusion catalyzed air or O2 cathode is employed. Such cells using Al are not rechargeable except by mechanical replacement of the metal anodes. However, ifZn is used, electrochemical recharging is possible, but requires a bifunctional catalyzed counter electrode capable of evolving H2 on recharge or reducing O2 (air) on discharge.

TABLE Batteries Pb-PbClz

2.3. Materials

for Constructing



and Batteries



Pb-PbOz Pb-PbS04 Cd-Cd(OHh Ni(OHh-Ni·O·OH


MnOTMn(OHh Zn-ZnO-ZncYzAg-Ag20 or Ag-AgO Hg-HgO Li-SOClz Li-S02 Li-CFx

Carbon (double-layer capacitors), H on Pt (UPD) Underpotential-deposited Pb on Au RuOz lrOz Modified cheaper transition metal oxides, metal nitrides H in Pd, LaNis(?) Conducting polymer electrodes, e.g., polyanilines, polythiophenes

Intermediate transitional systems:

urts., Li-MoSz; Li-Mo02; Li-Mn02; Li-C002; Li-C


2.2. Types of Battery

and Mode of Energy


Basis of charge or energy storage Leclanche, zinc-Mn02 Alkaline, zinc-Mn02 Mg-AgCI Mg-PbClz Li-SOCI2 and other cathodes Li-CFx AI-air (catalyzed) Zn-air (catalyzed)



The general question of experimental and theoretical constraints in the choice of materials for electrochemical capacitors was examined in an article by Sarangapani, Tilak and Chen." Such factors as accessibility of the active surface, the shapes and reversibility of voltammograms, and the components of the redox capacitance at Ru02 (Chapter 11) needed to account for its almost constant value over a 1.4-V range are discussed, together with the question of the stability of materials in overcharging and extended cycling. Table 2.3 lists the principal materials used to construct batteries and electrochemical capacitors.

Lead acid, Pb-Pb02 Nickel-cadmium, Ni·O·OH-Cd Nickel-hydrogen, Ni·O·OH-metal hydride Nickel-zinc, Ni·O·OH-Zn Mercuric oxide-zinc, HgO-Zn Silver oxide(s)-zinc, AgO-Zn Zinc-air (catalyzed) Li-TiS2 Li-MoS2 Li-MnOz Li-CoOz Li-C-CoOz and other cathodes Li-iron sulfides Na-S


Cell Design and Equivalent Circuits

Faradaic (exhibiting intercalati ve pseudocapacitance)

As discussed and illustrated in Chapters 16 and 17, an electrochemical capacitor, like a battery cell, requires two electrodes, one of which is charged negatively with respect to the other, the charge storage and separation being electrostatic. At each of the two electrodes, double-layer electron and ion charge separations are established across the electrode interfaces. The macroequivalent circuit (in the absence of self-discharge processes) is represented by two capacitances linked in series with ohmic resistances representing the resistivities of the solution and the separator, as in a battery. In the usual case where the electrodes are high-area, porous matrices, a further microequivalent circuit is needed (see

18 Chapter 2 Supercapacitors and Batteries 19

Chapters 14 and 16) to represent the electrical behavior of the internal surfaces and the electrolyte-containing interstices. As with batteries, bipolar electrode configurations can be fabricated for higher voltage series combinations; these diminish the internal resistance of the device, but require that the edges of the electrodes be carefully sealed to the case in order to avoid shunt, i.e., leakage currents. Such systems optimize power density. The equivalent circuit for most battery-type charge storage systems involves simply a Faradaic resistance (RF) that represents the potential dependence of the reciprocal of the rate of the oxidation and reduction charge transfer process; it is in parallel with (an always significant) double-layer capacitance, Cdl. Under some high-rate discharge or recharge conditions, some diffusion control may arise, in which case RF is in series with a so-called Warburg C-R impedance element written as W. In addition, with some rechargeable battery systems (e.g., of the intercalation type), a pseudocapacitance element may also be required to represent the impedance behavior of the Faradaic process (Chapters 10 and 17). For both electrochemical capacitors and battery cells, a solution resistance element, Rs' in series with the Faradaic impedance, ZF' is usually necessary in order to fully represent the charging or discharging behavior. In fact, R, is usually very important in the evaluation and performance of capacitors and batteries for high-rate discharge applications, and is an important influence on the ac impedance spectrum (see Chapters 16 and 17) of the device. In the absence of self-discharge processes, or any parallel pseudocapacitance, the macroequivalent circuit of an electrochemical capacitor involves only a solution resistance (Rs) and a double-layer capacitance Cdl. However, R, and Cdl have distributed components in the microequivalent circuit in the case of high-area, porous electrode materials, as treated in Chapter 14. The same applies to battery electrodes fabricated in porous material configurations, a procedure and design that are commonly adopted in battery technology; then complex microequivalent circuits also apply, but with the inclusion of the essential RF components.

face, it is easily seen that the charge accommodated per atom will be 30/1015 /lC. This is equivalent to 30 x (10-6/105) x 6 x 1023/1015 electrons per atom where 105 and 6 X 1023 are taken approximately as the Faraday constant and Avogadro's number, respectively. The above quotient works out to be 0.18 electrons per atom stored as a double-layer charge at 1 V for the example chosen, but this is the average delocalized charge distribution associated with conduction-band electrons. By contrast, in most battery processes, redox reactions involving usually one or two valence electron charges per atom (sometimes three for Al or Bi) or molecule of electro active reactant are involved. Thus, electrochemical capacitor charge storage involves, per atom of active available surface area, only about 20 or 10% (respectively) of that involved (indirectly) or available with battery redox materials. Hence the energy densities available with capacitors are usually substantially smaller than those with batteries, This is a well-recognized limitation, but it is usually compensated by much larger cyclability (cycle life) and often better power density attainable from electrochemical capacitors. Of course, because the energy density of a capacitor increases with the square of the voltage on charge, a substantial improvement in energy density is attained by using nonaqueous electrolytes instead of aqueous ones, in which the decomposition voltage can be increased to 3 or 3.5 V or more per cell unit. Elsewhere in this volume (Chapter 14) it will be shown that with large-area carbon materials (e.g., 1000 m2 g-I), very large specific capacitances of 300 F g-I can (theoretically) arise. It is then of interest to compare energy densities attainable at double-layer capacitors with those realized in an Ni-Cd battery.



of Energy Densities Attainable in Electrochemical

Capacitors and Batteries




Electron Densities per Atom or Molecule

In the double layer at plane electrodes, charge densities of about 16 to 50 us: cm-2 are commonly realizable. Taking an average value of 30 /IF cm-2 or 30 /lC V-I, and an atom density of ca. lOIS cm-2 at a smooth electrode sur-

For a capacitor electrode of 1000 m2 operating at 1 V and having a spe2 cific double-layer capacitance of 30 /IF cm- (e.g., for carbon blacks), the total capacitance is 300 F g-t, as mentioned earlier. At 1 Y, the energy, G, stored is G = (112) CV2 = 112 x 300 x 12 = 150 Wsec g-I or J g-l or 150 kJ kg", theoretically. Equivalently, in watt-hours (1 Wh = 3600 J), the above energy density is about 42 Wh kg:". (In practice, it will usually be substantially less than this figure owing to the inaccessibility of the electrolyte solution to the fine pores of the porous electrode structure in the case of carbon powders or felts, and the weight of the packaging structure and electrolyte.) A comparison with an Ni-Cd battery can be made as follows: The molecular weight of Ni·O·OH is 92 and the equivalent weight of Cd is 11212, i.e., 56. The electron number is 2 for Cd (1 for the equivalent weight of 56) and approximately 1 for Ni·O·OH being reduced to Ni(OH)z. (The oxidation state of Ni in




Chapter 2

Supercapacitors and Batteries


charged nickel oxide electrodes can approach higher values at low temperatures and in strong aqueous KOH.) For a working voltage of 1.2, the ideal energy density (L'1G -zFE) will be = 1.2 X 1 X 1051148 J g-I == 8.1 X 105 J kg " = 810 kJ kg " where 105 is approximately F, the Faraday constant in coulombs mole ". With 1 Wh = 3600 J, the (theoretical) energy density for Ni-Cd (electroactive materials alone) is then 225 Wh kg". Thus an electrochemical double-layer capacitor electrode (charged to 1 V working potential) would have about 20% of the energy density of an Ni-Cd battery electrode, both figures being based on ideal, theoretical performance. A double-layer capacitor based on nonaqueous electrolyte technology would have a substantially larger energy density than that for I-V aqueous solution charging. In an actual two-electrode capacitor device, with one electrode worked against the other, the energy density will be about (112)2 smaller, namely, about 10 Wh kg " in aqueous electrolyte, since each electrode can be discharged down to only about half its initial voltage following ± charging.


... _-----

--- --_

-----1-------------------~----~, Recharge



-.- .. -..







b e,
... ,

z UJ



: State of : charge indication ,




FIGURE 2.1. Difference of discharge and recharge relationships for a capacitor and a battery: potential as a function of state of charge, Q. 2.5. COMPARISON OF CAPACITOR AND BATTERY CHARGING CURVES

As shown later in this chapter, the energy of the charging of a capacitor to a plate voltage difference of V is (1/2) CV2. It is an electrostatic free energy (Gibbs energy), G. Its entropy component S (= [H - G]IT, where H is the enthalpy of charging) arises from the temperature dependence of the permittivity of the dielectric, here the dielectric constant of the double layer. For a battery process, the maximum Gibbs energy is the product of charge Q and the difference of potential, M, between the Nernstian reversible potentials of the two electrodes, i.e., G = Q-M. For the capacitor case, for a charge Q accommodated, G is (1/2) QV. For a given electrode potential difference, M = V, in the two cases it is then evident that the energy stored by a two-electrode cell accommodating a given Faradaic charge Q at voltage M, = V, is twice that stored in a capacitor charged with the same Q to the same voltage. This difference can be understood in the following way: In the process of charging a pure double-layer capacitor, as explained in Chapter 4, every additional element of charge that is added has to do electrical work (Gibbs energy) against the charge density already accumulated on the plates, progressively increasing the interelectrode potential difference. In a battery cell being charged, a thermodynamic potential (ideally) exists independent of the extent of charge Q added, so long as the two components (reduced and oxidized forms) of the electroacti ve material remain coexisting. Thus the potential difference (electromotive force, emf) of the battery cell is ideally constant throughout the discharge or recharge half-cycles, so that G is Q·M

rather than Q·1/2 M (or 1/2V). This difference can be illustrated by the discharge curves shown schematically in Fig. 2.1 where the voltage on the capacitor declines linearly (for potential-independent Cdl) with the extent of charge, while that for an ideal battery remains constant* as long as two phases remain and are in equilibrium (upper curves in Fig. 2.1). The decline of the capacitor voltage arises formally and phenomenologically since C = QIVor V = QIC; therefore dVldQ = lie. The ideal battery cell voltages on discharge and recharge, as a function of state of charge, are shown as parallel lines of zero slope" in the upper parts of the diagram. These two lines differ as a result of any cathodic and anodic polarization (including so-called ohmic IR potential drop due to internal or solution resistance) arising in the discharge or recharge half-cycles. In the sloping discharge and recharge line for the capacitor (Fig. 2.1), there will also be significant IR drop, depending on the discharge or recharge rates, that is, the discharge line will actually be somewhat separated from the recharge line by a voltage difference equal to 2IR. The significance of the difference between the energy density for the capacitor and a battery cell charged to the same potential can be illustrated by ref-

*In practice, there is usually some decline of voltage in a battery with increasing extent of discharge. In certain battery systems e_g_,LirTiS2 , Li/Co02, and metal hydride anodes, this decline is larger and occurs for fundamental thermodynamic reasons connected with the form of the sorption isotherms.

22 Chapter 2 supercapacitors and Batteries


erence to the working diagram shown in Fig. 2.2. Here the voltage vs. charge relations for a capacitor and an ideal (constant voltage) battery are compared. In actuality, most batteries, with the exception of Li-SOCI2, do exhibit some decline of voltage with decreasing state of charge, but this is for reasons quite different from that in the case of capacitor discharge. The line VB is the voltage across the battery cell as a function of charge Q held during a constant rate charge, Vc is the plot of voltage across a capacitor being charged up to the same final voltage, VB' with accommodation of the same accumulated charge, Q. However, since the line Vc is for capacitor charging, it has a slope of lie (see Fig. 2.1). The integrals VQ under the two lines of the working diagram correspond to the energy of charging. Thus it is seen that VcdQ is half VB·dQ, i.e., the energy of charging the capacitor to a terminal voltage VB with charge Q is half that for charging the battery with the same charge also to VB' The superimposed hatched and slashed areas in Fig. 2.2 correspond to this difference of energy stored. As shown mathematically in Chapter 4, the factor of half in the charging energy of a capacitor arises for the same reason that it enters into the Born equation for the self-energy of charging of spherical ions in vacuo or in a dielectric (see Chapter 4).



Full charge







vc = q/C

; Ec = ~ av = ~ CV-

As shown in Chapter 3, a capacitor subjected to a cyclic regime of linear potential change with time at a rate dVldt s generates a response current ± 1= C(±s). This procedure provides a convenient and sensitive method for characterization of both double-layer and pseudocapacitance behavior of electrochemical capacitors. Under conditions where the response is reversible to positive and negative-moving sweeps (s having + and - values), the resulting voltammogram for one direction of potential sweep is the mirror image of that generated for the opposite direction of sweep, provided there is no involvement of diffusion control. This is a useful criterion for reversibility in capacitative or pseudocapacitative charge and discharge processes and is a fundamental characteristic of pure capacitative behavior. It corresponds, at all states of charge across the cyclic voltammogram or along the charging curve, to an equilibrium situation being maintained provided that (see Chapter 10) the voltammogram is not scanned at too high a rate or the discharge curve at too high a current density. This is a very important and basic matter to understand; it distinguishes capacitor and battery behavior in a fundamental way.


FIGURE 2.2. Working diagram illustrating storage of energy E = VdQ for a battery at ideally constant charging voltage V = VB, and for a capacitor with progressively changing voltage, V =

Vc, leading to EB = QVB or Ec = (t)QVc.

In contrast, battery-type processes are rarely reversible in the above sense and a substantially different range of potentials is required for oxidation of the active material compared with that for its reduction. The cyclic voltammogram is then asymmetric and no mirror-image appearance is manifested. An example of the contrast between a reversible, capacitative cyclic voltammogram (for the pseudocapacitance at Ru02) and an irreversible one (for the battery cell system Pb-PbClz) is shown in Figs. 2.3(a) and 2.3(b). For RuOz [Fig. 2.3(a)], the profile of positive charging currents over the potential range 0.05 to lAO V vs. the reversible H2 electrode potential (RHE) is almost the mirror im-


Chapter 2


and Batteries


-~ e


::J VI N


6.2 5-

~---~ /

a) RU02 film



<t E



c u 6.2 Sf-



'- 0 ::J J:





u 1Z.51-


1 1.4


o o o






b) Pb/PbCI2



IW 0:: 0:: U

::) 8

o o
Z <l



age of that for negative, discharging, currents and switching the direction of sweep during the potential scan produces an almost immediate reversal of direction of current. This behavior is also characteristic of a reversible, capacitative charge and discharge process (compare the theoretically calculated curves in Chapter 10, Fig. 10.1). Another characteristic feature of the behavior of this Ru02 example is that over a wide range of sweep rates, s, the current response, I, at any potential across the scanned range is linear in s. Since I = Cs, this means that C is independent of sweep rate and virtually pure capacitative behavior is exhibited; it is actually mainly pseudocapacitative (see Chapter lIon RU02) when the RU02 has been formed in a hydrous state, e.g., by an electrochemical procedure." Figure 2.3(b) shows, by contrast, the behavior of oxidation and reduction at Pb-PbCI2, a battery-type electrode (Pb + 2cr ~ PbCl2 + 2e). The voltammogram indicates a completely irreversible process: the potential range for formation of PbCl2 from Pb is widely different from that for the reverse process, reduction of PbCl2 back to Pb. Also, switching the direction of sweep during the voltammetric trace does not immediately reverse the sign of current at that switching potential, contrary to behavior observed with a capacitance . In this system, which is typical for a battery-type reaction (Pb-PbS04 is similar), overvoltages are required in both directions of the reaction in order to pass significant currents. Hence the current response profiles for each of the directions of reaction lie wide apart, on each side of the reversible potential, along the electrode potential axis of Fig. 2.3(b). The reversible potential corresponds approximately to the potential at which the current profile I(E) crosses the potential axis, i.e., where /(E) = o. Returning to the situation of a reversible process, it should be mentioned here that if such a reaction is diffusion controlled, then the cyclic voltamrnogram is again not of the mirror-image type, the potential for the anodic peak current being more positive than that for the cathodic one. This is because, even if the reaction is reversible, different potentials are required on the anodic sweep from those on the cathodic one to establish given concentration gradients in the boundary diffusion layer that lead to the same (equal but opposite) diffusion-controlled currents. If such a reversible reaction is conducted in a thin-layer cell where the solution thickness is less than the width of the boundary diffusion layer, then the voltammogram tends to acquire mirror-image symmetry like that for a reversible surface film process.

FIGURE 2.3. (a) Cyclic voltammogram for RU02 in I M aqueous H2S04 (298 K) showing mirror-image symmetry and responses to successive switching along the scanned potential range. (b) Cyclic voltammogram for Pb-PbCb battery electrode showing typical irreversibility arising with 3-dimensional materials undergoing chemical phase changes.





An important transitional behavior between electrochemical capacitors and batteries arises with processes involving Li+ intercalation into layer-lattice


Chapter 2


and Batteries


host cathode materials, e.g., MoS2, TiS2, V 6013, and C002.8,9 Normally these cathode materials, coupled with an Li anode or a Li-C anode, would be regarded as battery cathode materials. 1,2However, the forms of the charge and discharge curves and associated pseudocapacitance, and even the cyclic voltammetry profiles are similar to those for 2-dimensional electrosorption (underpotential dep~sition, UPD) (see Chapter 10). Hence this class of materials exhibits prop-




erties intermediate between those of bulk-phase battery reagents I and quasi-2dimensional pseudocapacitor electrodes. The thermodynamics of sorption ofLi into layer-lattice intercalation host materials have been worked out by Haering and Mackinnon" in terms of 3-dimensional lattice statistics which lead to 3-dimensional sorption isotherms in site fraction X, having forms similar to that of the relation for 2-dimensional adsorption into lattice arrays," Thus, such systems exemplify the transition between supercapacitor and battery behavior. Figure 2.4(a) shows an example of the sorption (discharging) isotherm for a 3-dimensional intercalation process, e.g., for Li into TiS2.8 Since a Faradaic charge is required for deposition ofLi into an intercalation host, X depends continuously on the charge passed. The relation of the potential V to X can be differentiated, so that formally a pseudocapacitance arises, as indicated by the differential quantity shown in Fig. 2.4(b) as a function of X. It is seen that it has a form like that of a cyclic voltammogram for a process exhibiting pseudocapacitance. This is because a cyclic voltammogram for an electrosorption process gives directly, as the current response, the differential coefficient of the charging curve or the sorption isotherm (see Chapter 10). The essential and common thermodynamic feature of systems that can give rise to pseudocapacitative behavior in electrochemical capacitors is thus the relation of free energy to the log of the configurational activity terms, i.e., 8/1-8 for 2-dimensional adsorption or Xll-X for 3-dimensional sorption and [Ox]/[Red]/(l-[Ox]/[Red]) for redox processes (see Table 2.4). While double-layer capacitance behavior on carbon materials has usually been distinguished from pseudocapacitance exhibited by RU02 and other systems, it has now become recognized that various high-area carbon preparations


~<I I






TABLE 2.4. Correlation of Types of Systems Giving Rise to Pseudocapacitance with Application to Development of Electrochemical Capacitors

System type (a) Redox system: Ox + ze ~ Red and 02- + H+ ~OHin oxide lattice (b) Intercalation system: Li+ into MA2

Essential relations E


= EO + (RTlzF) In 9\/(1 = [Ox]/([Ox] + [Red]);


- 9\) 9\/(1 - 9\)




E = EO + (RTlzF) In X/(1 - X) X = occupancy fraction of layer lattice sites (e.g., for

X IN Lix TiS2
FIGURE 2.4. (a) Discharge curve for Li+ intercalation into TiS2 as a function of extent of discharge, X, expressed as a stoichiometric factor in Li,TiS2 (solid line, calculated relation"). (b) Differential coefficient of the curve for discharge of Li+ into TiS2 as in Fig. 3(a). (From Haering and MacKinnon8.) (c) Underpotential deposition: Mz+ + S + ze ~ S.M (S surface lattice sites)

u' in TiS2)

E = EO + (RTlzF) In 81(1 - 8) 8

= 2-dimensional

site occupancy fraction

sites (I 0) or (I X).

Note: (h) and (c) can he regarded as mixing of occupied (X or (J) sites with unoccupied Systems (a), (b), and (c) are dealt with in detail in Chapter 10.


Chapter 2

Supercapacitors and Batteries


also exhibit a small but significant pseudocapacitance. This appears to arise because of electrochemically active redox functionalities (Chapter 9) that exist on the surfaces of high-area carbon powders and felts, depending on the conditions of their pretreatment.




preciable toward the end of the charging half-cycle of a capacitor, there will be a corresponding appreciable Faradaic self-discharge current following termination of charging when i is cut to zero, until the generated overvoltage has decayed. In practice, self-discharge down to lower voltages often continues (see the discussion in Chapter 18). Once overcharge sets in, the efficiency of further charging to higher interelectrode potentials across the capacitor rapidly diminishes owing to the "Tafel" character of the potential dependence of the lr component (Chapter 3).

Battery electrodes often become overcharged either adventitiously or because of degradation and aging of their plates. Overcharge processes occur when the voltage between the plates exceeds the thermodynamic or, in practice, the kinetic limits for decomposition of the electrolyte solution. Also, at high imposed charging rates, some current component will often pass in parallel with the Faradaic charging process owing to decomposition of the battery's electrolyte, usually to form H2 and 02' This is caused by overvoltage effects. In the case of an electrochemical capacitor being charged, there is no particular thermodynamic cell voltage of the kind to which a particular battery should be charged. Thus, after the thermodynamic potential (1.23 V at 298 K) for decomposition of water (the normal "voltage window" for the solution) is reached, further charging to higher potential differences between the capacitor electrodes results in increasing rates of solvent decomposition, which is similar to that in overcharging of a battery. In both cases the double-layer capacitance continues to be increasingly charged in proportion to the prevailing capacitance at those potentials, but in addition, with carbons, some electrochemical oxidation of the active material can take place. The current is, however, divided into a charging current, i.; and a Faradaic leakage current, iF, that normally will increase exponentially with the potential beyond the solution's decomposition limit. The situation is thus: (2.3) or, correspondingly, in terms of respective current densities, i. The leakage current component ir increases as the electrode potential increases, usually exponentially, and at sufficiently high potential, where decomposition of the solution takes place, iF » ie and i -7 iF' Further details are presented in Chapter 3. This situation is the opposite of self-discharge, while for the ideally polarizable electrode or capacitor, i = i., For optimum efficiency in charging a capacitor, iF should be a minimum over the whole range of potentials corresponding to the charging process over the practical operating potential range of the capacitor: ca. 1.3 V for aqueous electrolytes and 3.5-4.0 V for nonaqueous, aprotic solvent solutions. If iF is ap-



In this section the behavior and properties of electrochemical capacitors and batteries are summarized. Table 2.5 lists the perceived advantages and disadvantages of electrochemical capacitors for storing electrical energy. Tables 2.6 to 2.8 compare the properties of the two systems. Some of this summarized information emphasizes thermodynamic differences between batteries and electrochemical capacitors, including those of the pseudocapacitance type. The properties and behavior of double-layer and pseudocapacity types of electrochemical capacitors are compared in Table 2.1 and in Table 10.1 in Chapter 10. As noted in that chapter, the mechanism of charging of the latter type of device is mainly Faradaic, like that of a battery. Table 2.8 gives an overall comparison of electrochemical capacitor and battery behavior under discharge and recharge conditions.

2.5. Perceived


and Disadvantages Energy Storage

of Electrochemical


Advantages Long cycle life, > 100,000 cycles; some systems up to 106 Good power density [under certain conditions, limited by IR or equivalent series resistance (esr) complexity of equivalent circuit] Simple principle and mode of construction (can employ battery construction technology) Cheap materials (for aqueous embodiments) Combines state-of-charge indication, Q = flV) Can be combined with rechargeable battery for hybrid applications (electric vehicles) Disadvantages Limited energy density Poor volume energy density Low working voltages (compared with electrolytics; satisfactory compared with batteries) Aq. voltage range 0 - 1.4 V; nonaq, to 4.5 V. In practice, 3.5 V. Nonaq. embodiments require pure, H20-free materials; more expensive. Requires stacking for high potential operation (electric vehicles) Hence, good matching of cell units is necessary

TABLE 2.6. Comparative Electrical Characteristics Behavior Electrochemical free energies of capacitor Capacitor 1. Has intrinsically sloping charge and discharge curve of Battery

Chapter 2
and Electrochemical

TABLE 2.8.

and Batteries
Overall Comparison of Electrochemical Capacitor and Battery


Capacitor Battery 1. Ideally has single-valued components

Characteristics Battery

2. emf is ideally constant with degree of charge and discharge, except for nonthermodynamic incidental effects, or phase changes during discharge 3. Behavior is not capacitative, except in very general sense 4. Irreversibility is usual behavior (materials irreversibility and kinetic irreversibility) 5. Response to linear modulation of potential gives irreversible i vs. V profile with nonconstant currents 6. Discharge at constant current arises at a more or less constant potential except for intercalation Li batteries

Has continuous variation of free energy with degree of conversion of materials or extent of charge held Potential is thermodynamically related to state of charge through log [XII - Xl factor, in a continuous manner for a pseudocapacitor, or directly to Q for a double-layer capacitor Behavior is capacitative High degree of reversibility is common (10-4_10-6 cycles with Ru02 or C double-layer capacitors) Response to linear modulation of potential gives more or less constant charging current profile but with some dependence on materials Discharge at constant current gives mainly linear decline of potential with time, which is characteristic of a capacitor

Ideally has constant (thermodynamic) discharge or recharge potential, except for Li intercalation systems Docs not have good intrinsic state-of-charge 2. Because of (1), has good intrinsic stage-of-charge indication indication except for Li intercalation systems 3. Has relatively poor energy density Has moderate or good energy density, depending on equivalent weights and electrode potentials of active materials Has relatively poorer power density, depending 4. Has good power density on kinetics 5. Has excellent cyclability or cycle life due to Has less cycle life by a factor of 1/100 111000 due to irreversibility of redox and simple addition or withdrawal of charges (in phase-change processes in three dimensions double-layer type) Has internal IR due to electrolyte and active 6. Has internal IR due to high-area matrix + electrolyte materials 7. Has little or no activation polarization but C Has significant Tvdependent activation polarization (Faradaic resistance) may be temperature-dependent Has poorer lifetime due to degradation or 8. Has long lifetime except for corrosion of reconstruction of active materials current collectors, etc. Electrolyte conductivity can decrease or 9. Electrolyte conductivity can diminish on increase on charging, depending on charging due to ion adsorption chemistry of cell reactions, e.g., with Pb-acid Can be constructed in bipolar configuration 10. Can be constructed in bipolar configuration

1. D. Linden, ed., Handbook of Batteries, 2nd ed., McGraw-Hili, New York (1995). 2. H. A. Kiehne, Battery Technology Handbook, Marcel Dekker, New York (1989). 3. W. Lorenz and G. Salit\ Zeit. Phys. Chern., N.F. 29,390,408 (1961). 4. 1. W. Schultze and F. D. Koppitz, Electrochim. Acta, 21,327,337 (1976). 5. B. E. Conway, H. A. Kozlowska, and W. B. A. Sharpe, Zeit. Phys. Chern., N.F., 98, 61 (1975). 6. S. Sarangapani, B. V. Tilak, and C.-P. Chen, 1. Electrochem. Soc., 143, 3791 (1996). 7. S. Hadzi-Jordanov, H. Angerstein-Kozlowska, and B. E. Conway, 1. Electrochem Soc., 125, 1473 (1978). 8. R. R. Haering and R. MacKinnon, in Modem Aspects of Electrochemistry, B. E. Conway, I.O'M. Bockris, and R. White, eds., vol. 15, Chapter 4, Plenum, New York (1981). 9. B. E. Conway. Electrochim. Acta, 38,1249 (1993).





of Therrnodynamic Capacitor

Behavior Materials




Electrochemical Battery During discharge or recharge has unique, single-valued free energies !!.G of the electroactive phases involved
!!.G ;: constant

Supercapacitor Has continuously changing free energy of electroactive material with extent of charge and discharge. G == 112 CV2 or !!.G == !!'Go + RTin [X/(1 - Xl] for pseudocapacitance Has corresponding continuous variation of potential during charge and discharge Recharge and discharge curves are mirror images of one another, i.e., in cyclic voltammetry. Very reversible


1. Electrochemical capacitors, in Proc. Symposium on Electrochemical Capacitors, F. M. Delnick and M. Tomkiewicz, eds., pp. 95-29. Electrochemical Society. Pennington. N.J. (1995). 2. Materials for electrochemical capacitors: theoretical and experimental constraints (review). S. Sarangapani, B. V. Tilak, and c.r. Chen, 1. Electrochem. Soc., 143, 3791 (1996). 3. B. E. Conway, 1. Electrochem. Soc., 138, 1539 (1991) (review). 4. NEC Capacitor Data Book, NEC Corp. (1995), NEC Electronics, Mountain View, Calif.

Has corresponding single-valued potential on discharge Usually not reversible, in the sense that recharge curve is not mirror image of discharge curve, e.g., in cyclic voltammetry

Chapter 3

Energetics and Elements of the Kinetics of Electrode Processes

3.1. INTRODUCTION On the basis of the assumption that this volume will be read not only by specialists in electrochemistry but also by electrochemical engineers and technologists, as well as materials scientists, it has been considered desirable to include a brief chapter on the basic aspects of charge transfer at electrode interfaces, especially their energetics and kinetics. Some notes on the requirements for experimental measurements on capacitor and battery electrodes are also included. Ideally, the behavior of capacitative charge or energy storage devices, unlike the situation with batteries, would not involve any excursions into, or applications of electrode kinetics. This statement is predicated on the assumption that the charging or discharging of a capacitor device involves only electrostatic processes and not any electrode processes of the Faradaic kind, the kinetics and energetics of which, if significant, would lead to deviations from pure capacitative behavior. However, electrode kinetic effects do come into play whenever a capacitor electrode ceases (practically) to be any longer ideally polarizable (Chapters 6 and 7). This can happen on overcharge when solution decomposition sets in and/or when Faradaic self-discharge processes occur owing to the presence of oxidizable or reducible impurities or reactive functional groups at carbon surfaces. Thus, in the case of electrochemical capacitors, some indirect or, in some cases, direct involvement of Faradaic electrode processes can occur as follows:

34 Chapter 3 Energetics and Elements of the Kinetics of Electrode Processes 35

1. in adventitious overcharge or overdischarge of double-layer-type capacitors when decomposition of the electrolyte occurs; 2. in charge or discharge of the small but significant pseudocapacitative component of the capacitance that occurs with most carbonbased double-layer capacitor devices; 3. in the self-discharge processes that most electrochemical capacitors undergo on open circuit, following charging; and 4. in the basic Faradaic mechanisms of charging or discharging pseudocapacitors of the oxide or conducting polymer redox type, or of the adsorption type (see Chapters 10, 11, and 12) where the kinetics and energetics of electrode processes are directly involved.

3.2. ENERGETICS OF ELECTRODE PROCESSES The electrode processes that are the subject of this chapter are of the socalled "Faradaic" type involving electrolytic charge transfer of electrons between electrode interfaces and ions or molecules in solution; hence the origin of the term, which is named after Michael Faraday's work, especially that on electrolytic decomposition of solutions. However, it should be mentioned that at the time (from 1834 onward) of this work, the details of such charge transfer processes involving electricity and chemistry'r' were far from being understood. It remained for Crookes and for J. 1. Thomson (1897) to demonstrate much later the common involvement of electrons in matter and their characterization in terms of their charge/mass ratio and charge (by Millikan in 1909 and 1913). As noted in Chapter 1, Faraday's laws implied an atomic unit of electricity, now called the electron. However, it still took many years for it to be understood how electrons became involved in electrochemical charge transfer processes. A fuller understanding had to await: (1) knowledge about the electronic states and their energy levels in atoms and molecules, a topic dependent on interpretation of spectroscopic and ionization phenomena by Bohr and Sommerfeld and (2) knowledge of the state(s) of free and bound electrons, their number density in metals and semiconductors (by Fermi and Bloch), and their thermal (Richardson) and photoemission (Einstein) characteristics in relation to the so-called "electronic work function" (cP) for emission of electrons from metals. The important conclusion from this work was that electrons are ubiquitously present in all metals, with relatively high concentrations being free, mobile, and delocalized-about one per atom; they are thus available for exchange with ions or molecules across the metal's interface. Later, the somewhat different states of electrons, (or deficiencies thereof, i.e., holes) at much lower concentrations in semiconductors became understood.

At the same time, the energetic states of ions (or molecules) in solutions in polar solvents had to be understood since the elementary act of an electrode process is the transfer of an electron to (cathodic) or from (anodic) a reactant in solution and from or to the electrode. A basic factor in such electron transfer processes began to be understood in a fundamental way in the 1930s and in more detail in the 1940s and 1950s; it was that electrochemical electron transfer processes occur in a so-called "radiationless" manner, i.e., with no photochemical absorption or emission oflight (radiation) quanta. This means that there must be selective adjustment of electron energy levels in the electrode, relative to those in a reducible or oxidizable reactant, so that the transfer of electrons takes place between levels that are equal in energy. The solvation energy of the reactant ions or molecules in such electron transfer processes is a major factor3-6 in determining the condition for radiationless electron transfer and the rates of such reactions. The general basis of this concept originated in papers by Gurney," Gurney and Fowler," and Butler." was followed in more detail by the important paper by Weiss,3 and later developed in a series of papers by Marcus'v' and Bockris and Parsons.6 The principal concepts are as follows: 1. The electrons involved in a Faradaic redox process are transferred from (in a cathodic process) or to (in an anodic process) the Fermi level of the delocalized conduction-band electrons of the metal electrode. The electrons in the metal obey the so-called "Fermi-Dirac statistics," and the upper level of halfoccupancy (the probability of the occupancy of that level being 0.5 at finite temperatures) is designated the "Fermi level." Electrons at this level differ in energy from those in a vacuum with zero kinetic energy by a quantity, cP, the electronic work function of the metal. cP varies from about 2 eV (ca. 46 kcal mor ' or 184 kJ mor') for the alkali metals to about 5 eV or more for Pt and other noble metals. The value of cP for a given metal also depends appreciably on the orientation of the crystal face at which the electron is exchanged. cP can be regarded as the negative of the electron affinity of the bulk metal through its surface and is thus related in a general way to the electronegativity of the metal forming the electrode material, and to its surface structure and ionization potential. The energy required to remove an electron in a cathodic process is compensated by the energy gained when it has been transferred to an electron acceptor (oxidant, "Ox," reactant) and vice versa for an anodic process, from a reductant, "Red." 2. The reactants that are electron acceptors, "Ox," or electron donors, "Red," in solution are "solvated" by the electrolyte solvent (hydrated in the case when the solvent is water). The transfer of an electron to or from such an Ox or Red ion in solution is associated with a major change in the solvation energy of the ion and also involves the energy of ionization 10 (see later discussion). When molecular reactants are involved, the electron transfer usually results in the for-

Chapter 3 Energetics and Elements of the Kinetics of Electrode Processes


mation of an ion, or an ion radical, or a radical, with a corresponding large solvation energy or with coupled generation or consumption of a proton in order to achieve an overall charge balance in the reaction. The overall energy changes in a simple electrode reaction can be usefully represented by means of an energy cycle (a Born-Haber cycle) based on the first law of thermodynamics, which is related to the principle of conservation of energy. For example, for electron transfer involved in the reduction of a ferric to a ferrous ion in aqueous solution at a metal electrode M (a single electrode reaction), we can write the energy cycle as:

1 1~
+ eg




I:3.G Fe2a~


e (M)

from which the overall AG for this half-cell electrode redox process is (3.1) Here G, quantities are the Gibbs energies of solvation (hydration) of the ions Fe3+ and Fe2+, and both are large negative energy quantities; I is the ionization energy of the Fe2+ ion in the gas phase to form Fe3+ in the gas phase, and cPM is the electronic work function of the electrode metal M. The AG quantities are preferably expressed as the standard-state values, AGo or GO, for a unit activity or fugacity of the species in solution and in the gas phase. The G, I, and cPM quantities in Eq. (3.1) are each relatively large, having magnitudes of ca. tens or a few hundreds of kilocalories or kilojoules per mole. However, their values with different signs add up algebraically to relatively small numbers of 0 to ± 100 kJ mol-I. A similar situation arises for the AGo values and corresponding acidity constant (pKa) values for acid ionization processes in aqueous solutions where solvation and bond energies determine the
2.3 RT(pKa) or AGo values.

is canceled out at the metal/metal contact somewhere in the external measuring circuit. This is an important but often poorly understood point in the energetics of electrode potentials (see General Reading Ref. 4). The foregoing discussion may suggest that the kinetics of a charge transfer process at an electrode should depend on the work function, cP, of the metal electrode. Indeed, empirical plots of log io (the exchange current density) for the H2 evolution reaction (HER) measured at a variety of metals vs. respective cP values do show some clear relations: low work-function metals tend to have small values oflog io. However, for the same reasons that experimentally measured cell emfs do not depend on the work functions of the two electrodes, the log t; values have no direct relation to cP values.i' The empirical relations actually observed are due to an indirect effect, namely, that because log io values depend for fundamental rcasons v' on the standard Gibbs energy of adsorption of the electrosorbed species deposited (in the earlier example, H), and that the energy of chemisorption is itself determined in part by -o (a measure of the electron affinity of the electrode metal),13 then there arises an apparent relationship between log io and cPo Only for a single-electrode process would there be a hypothetical direct relation between its log io value and the metal's work function, but an experimental indication of such a relation is conceptually and practically impossible.



In the example treated earlier it is seen that the work function of the metal M enters directly into the determination of the AG or AGo value for the half-cell process and hence into its hypothetical electrode potential E = -AG/zF. However, since a single half-cell potential can never be measured, only the AE (i.e., the potential difference between two single electrodes) is experimentally accessible. In this case the work functions of the respective two electrode metals do not enter into the experimentally measured emf of the cell since their difference

The energy conditions for radiationless electron transfer are considered as follows. Processes at electrodes are radiationless, i.e., no photoemission or photoexcitation arises at metal electrodes. Photoexcitation occurs only at semiconductors, where excitation of electrons from a low-lying valence electron band to a higher level conduction band takes place upon absorption of a quantum of radiation. Therefore, for significant rates of charge transfer to occur, energy levels at or near the Fermi level in the metal must be matched with suitable vacant (lowest unoccupied molecular orbital, LUMO) or occupied (highest occupied molecular orbital, HOMO) orbitals in the reactant, depending on the direction of charge transfer [Figs. 3. 1(a),(b)]. Normally, an applied or spontaneously generated potential, V, is required to modify the electron work function cP to some value cP ± eVto achieve the condition of balance [Figs. 3.1(a) and 3.1(b)] required for facile electron transfer to take place at the potential V. If I is the ionization potential of the reactant in the reduced form and I:3.G s . IS the change of its solvation energy upon electron transfer, then the energetic condition for the process to occur in the direction of donation of charge is 7-9:

cP - eV - I + I:3.Gs == 0


Energetics and Elements of the Kinetics of Electrode Processes Chapter 3 38



Redox reaction 0+ e(M)+ R (non - matching energy levels)


o + e(M)..- R at electrode potential

V (matching

energy levels)

if the charge transfer is to a cation [e.g., H30+ in the H2 evolution reaction or for underpotential deposition of H from H30+ or H20] where A is the chemisorption energy of H at the electrode metal in the H2 evolution reaction or in UPD below the H2 reversible potential. The latter case corresponds to the development of a potential-dependent coverage by adsorbed H, leading to an adsorption pseudocapacitance (see Chapter 10). For cathodic H2 evolution, the electrodeposited H is an overpotentially deposited (OPD) adsorbed intermediate in the H2 evolution reaction. Changes of A from one metal to another for a given process (e.g., the HER) provide the principal basis for dependence of the kinetics of the electrode process on the metal. They are recognized as the origin of the electrocatalysis associated with a reaction in which the first step is electron transfer with the formation of an adsorbed intermediate." In the case of the HER, this effect is manifested in the dependence of the log of the exchange current density, im on metal properties12,13 such as rP (Fig. 3.2). However, for reasons peculiar to electrochemistry, reaction rate constants cannot depend on rP under conditions of currents experimentally measured at controlled potentials (and referred to the potential of some reference electrode) since rP quantities cancel out around the interfaces of the measuring circuit, as explained earlier. Hence relations such as those in Fig. 3.2 must arise from some other factor, as discussed in Refs. 13 and

12 .Sn
~ ~.Mo AI






,To _



+ S ~O




.... .......




FIGURE 3.1. Conditions for energy mismatching (a) and energy matching (b) at the electrode
Fermi level for facile electron transfer to occur in a radiationless process (based on Ref. 3 and later representations by Gerischer).

C7' 0 I

_ ~






for the redox process Ox + e ~ Red; I1Gs in Eq. (3.2) is positive for a decrease of net charge. . . . When the result of electron transfer is the production of an intermediate chemisorbed with energy A (A being negative), condition (3.1) becomes"

0 3.5


4.5 Work function, Cb/eV



rP - eV -1-I1Gs

+ A == 0


FIGURE 3.2. Empirical plots of log of exchange current-density values, io, for the H2 evolution reaction at various cathode metals plotted against the electron work function, <1>, in electron volts. (From Conway and Bockris.13 Reprinted with permission from J. Chern. Phys., 26, 532. Copyright 1957 American Institute of Physics.)

-10 Chapter 3 Energetics and Elements of the Kinetics of Electrode Processes


a: w

~ ~

o ~

o ,.






M-H Bond Strength (kcal rnol')

FIGURE 3.3. Theoretical relation of Parsons for dependence oflog io for the HER on the standard Gibbs energy,l1G;ds.H, for chemisorption of H at the electrode metal. (From Parsons. 12 Copyright 1958 Royal Society of Chemistry.) FIGURE 3.4. Volcano plot for log io values for the HER at various cathode metals against heat of adsorption of H (MH bond energy) as given by Trasatti 15 based on critically selected data.

11; this must be the energy of adsorption (A H) of H at the metal. The apparent relation to 1> arises because AH usually depends on 1>,13 e.g., for the initial heats of chemisorption at low coverage. Parsons 12 derived a theoretical relation for the dependence of log io on the standard Gibbs energy (~G~) for chemisorption of H at the metal, and its form is as shown in Fig. 3.3. A maximum arises in log io when ~G~ is zero, because log io is dependent on the coverage product, ~(l - BH)l-a, which is determined by ~G~. This function has an obvious maximum when BH= 0.5 and a = 0.5, the transfer coefficient, and gives rise to the so-called "volcano relation" illustrated in Fig. 3.3. Thus, volcano relations can occur in electrocatalysis for reasons similar to those for regular heterogeneous reactions, as was discussed in classical catalytic work e.g., by Balandin.i" An experimental plot of log io for the HER vs. heat of adsorption of H based on critically selected data was given by Trasatti15 and is replotted in Fig. 3.4. Only for those metals that have a strong affinity for some electrosorbed species (e.g., H at Pt, Rh, Ru, Ir or Pb or Bi atoms at Au or Ag) does UPD occur or alternatively, in the OPD potential range, when a desorption step is rate determining, does appreciable potential-dependent surface coverage of the electrosorbed species occur. Then, for such conditions, adsorption pseudocapacitance (Chapter 10) arises. Thus, the energetics and kinetics of charge transfer processes that lead to an adsorbed, discharged (or partially discharged) species (see Ref. 16) are of major importance in determining which and what

type of systems develop pseudocapacitance that is practically significant and that might be utilized for supercapacitor devices. The cases of H on Pt or Pb on Au were the first systems to be kinetically investigated by the present author for the possibility of using their large pseudocapacitance for capacitor devices. The good reversibility of their charge and discharge kinetics is a prime factor in attaining favorable power densities, but the operable potential ranges are usually small, only 0.6 to 0.3 V, and the C values are potential dependent.

3.4. KINETICS OF ELECTRODE REACTIONS AT METALS 3.4.1. Currents and Rate Equations First it is useful to note that the rates of electrode processes are directly measurable as currents, /, or as current densities, i, expressed per square centimeter of available accessible area (A) of the electrode surface, which is important for electrode reactions at interfaces of high-area, porous electrode materials. Thus,

= zFv


Where v is the reaction velocity in mole S-1 cm-2 and z is the number offaradays passed per mole of reactants. The current density i is related to the current / passing at the available electrode area A by

Chapter 3 Energetics and Elements of the Kinetics of Electrode Processes 43

i = If A


Equation (3.4) applies to a given electrode process only if its curr~nt efficiency is 100%; otherwise a current e~f~ciency factor <1 ~~st be ap~~l~d. F?r example, in the cathodic electrodeposition of Zn from Zn or Zn02 ions In aqueous solution, significant co-deposition ofH, with H2 evolution, t~kes place, so Eq. (3.4) applies properly only to either the partia~ current den~lty for Zn deposition or to the partial current density for H2 evolution, The relative proportions must be analytically evaluated. The primary feature of the kinetics of electrode processes, distinguishing them from kinetics of regular chemical reactions, is their dependence on the potential, V, of the electrode, which is usually referred to that of some reference electrode in contact with the electrolyte solution. This dependence on V arises energetically, as explained earlier, becaus.e of the dependence of the electron work function cP on V: cPv = ~v=o ± e V. ~lnce the electron is a reactant in the electrode process, changes of Its energy m the metal relative to that in vacuum modify the activation energy for the charge transfer process and hence the rate constant of the electrode reaction. Thus if the energy of the initial state of the reactant ion or molecule, plus the electron in the metal, is changed by ±eV or ±VF per mole of electrons (96,500 coulombs) then the activation energy is normally changed by some fraction, /3, of ±VF. /3 is commonly around 0.5 ± 0.1 for many electron charge transfer processes, and is called the "barrier symmetry factor." It is analogous+' to the Brensted coefficient for a series of related homogeneous proton transfer processes which proportionates their Gibbs energies of activation to the standard overall Gibbs energies of the respective protonation reactions. The significance of the effect of change of VF in relation to /3 can be illustrated in the schematic energy diagram of Fig. 3.5, which shows the effect on the activation energy M+ (or ~G+ for the Gibbs energy) that is involved in the rate equation. It is found that (3.6) over an appreciable range of change of V. /3 = 0.5 when the energy curves for the course of the electrode reaction (e.g., discharge of H from an H30+ ion or from water) along its reaction coordinate, d (Fig. 3.5) cross one another symmetrically, i.e. (numerically) with the same slopes dGfdd at and near the crossing point corresponding to the transition state (activated state) of the electron transfer reaction. For the discharge of H30+ or solvated metal (M) ions, the immediate product of the electron transfer step is an adsorbed atom (H) or adion of M at the electrode surface.



\ \ \


\ I

, , , , , , ,

, , ,, ,


\ \


" VF




FIGURE 3.5. Significance of effect of change of electrode potential, as VF, on energy of activation, !ill+, of an electron transfer reaction in relation to the barrier symmetry factor, {3. (Equivalent to an electrochemical Brensted relation. 17)

The basic kinetic equation for such processes is as follows (Ref. 11 and General Reading refs.), taking the H deposition case as the example:
i = zFk

(1- 8H)exp -



where 8H is the coverage of the electrode surface by the product of discharge, H, and k is a rate constant for the process at a hypothetical V = o. C +s is the H local concentration of the reacting H+ ion at the electrode surface in the Helm?oltz compact-layer region of the double layer (see Chapter 6). Equation (3.7) III various related forms leads to the Tafel equation (Eq. 13). The single electrode potential V cannot be measured, but a practical measurable scale of potentials can be introduced if V is referred to the potential of a reference electrode in the same solution. Most conveniently, a reference electrode for the same reaction at reversible equilibrium (Vrev) can be employed; then the V scale becomes an overvoltage scale tT given by

Chapter 3 Energetics and Elements of the Kinetics of Electrode Processes


1/ = V - Vrev or V = 1/ + Vrev Then Eq. (3.7) becomes


where the two terms on the right-hand side (rhs) ofEq. (3.14) or (3.15) represent the forward and backward rates of the reaction, both of which are significant near (within 10-15 mV) the reversible potential. Equations (3.14) or (3.15) are called the "Butler-Volmer equation." When the forward direction is predominant at appreciable values of 1/,

(3.9) which can be written (3.10) where K = k exp -[{3 VrevFIRD, an electrochemical rate constant, denoted by the bar and applicable to the reversible potential of the studied process, Vrev· Equation (3.10) is the elementary basis of most expression~ for curre~t density as a function of potential, expressed here on an ?verpotenhal ~c~le. It IS seen that i increases exponentially with 1/, with a slope III terms of In 1 given by dIn ildn

= io exp[-{31/FIRT],


which is a Tafel relation in exponential form. Thus, from Eq. (3.16) 2.3 log i = 2.3 log io - {31/FIRT A comparison with Eq. (3.13), the empirical Tafel relation, reveals that 2.3RT I . 2.3RT I . 1/ = -og 1 - -og 1 {3F 0 {3F so that 23RT. a = ----;p:-log 23RT (3.18) (3.17)

= -{3FIRT

(3.11 )

The reciprocal ofthis derivative is called the "Tafel slope," and is denoted by b, so that b = -2.3 RTI{3F (3.12)


and b = -


(= -0.118 V at 298 K with

P = 0.5)


with In i expressed as 2.3 log i; then b is written empirically in the Tafel equation: 1/ = a + b log i (3.13)

Note that 1/ becomes more negative (for a cathodic process) as i or log i increases. The increase of 1/logarithmicaUy in i is characteristic of kinetic (in contrast to diffusion) or so-called "activation" control of an electrode reaction.

after Tafel's work (1905) on the HER, which arose originally from his research on the electrochemical reduction of strychnine. Various Tafel slope values (b) can arise for the HER, depending on the rate-controlling mechanism that obtains or the H coverage conditions 1 and potential dependent, or H ~ 1 and H« thus independent of 1/). More generally Eq. (3.7) can be written in the form


Linearization of the Butler-Volmer Conditions (low ~

Equation for Near-Equilibrium


Another aspect of the properties of Eq. (3.15) for near-equilibrium conditions (small 1/) is that the two exponential terms can be linearized in a Taylor series expansion to the approximation: i = io [1 -


= K cH:s(1


exp -[{31/FIRT] - K

exp[(1 - {3)1/ FIRT]



- 1 - (1 - P)1/ FI RT]


or equivalently in terms of the exchange current density, io' passing reversibly at the equilibrium potential (1/ = 0): i

= -io1/ FIRT Then, differentiating this latter relation, it follows that


= t; (exp

-[{31/FI RT] - exp[ (1 - {3)1/ RT]) FI



I I~


Chapter 3

Energetics and Elements of the Kinetics of Electrode Processes


dildn has the significance of a reciprocal resistance according to Ohm's law, E = iR. This resistance is called the Faradaic resistance, RF, of the electrode reaction and (3.23) Note that in Eq. (3.15) a negative sign is taken for the cathodic 17in the HER. This latter relation (3.23) is useful for characterizing the dynamics of the electrode reaction in terms of: 1. the so-called "microscopic polarization" (low 17)of the electrode reaction where i is linear in 17;and 2. the impedance of the reaction under alternating voltage modulation where the Faradaic resistance, RF, can be determined as well as the double-layer capacitance with which RF is in parallel (Randles' equivalent circuit) sometimes together with a diffusional, so-called Warburg impedance, W (see Chapters 16 and 17). The Faradaic resistance has special significance in work on the frequency response (impedance analysis) of electrochemical capacitors in cases where either some Faradaic self-discharge process takes place or when pseudocapacitance is involved in a primary way in the behavior of the device. In these cases, an RF component, which is usually dependent on the de-level potential to which the device is polarized, must be included in the equivalent circuit that represents the frequency response of the capacitor system. This matter is treated in detail in Chapters 16 and 17 on impedance spectroscopy.
Log I



iI a

Log I

Linear polarization region

(slope"" 1/RF)

small '1


.!I \""



FIGURE 3.6. The anodic and cathodic current components of the potential dependence of an electrode reaction rate near its reversible potential: graphical representation of the Butler-Volmer relation (Eq. 3.15) illustrating the dependence of the net current density. i = ia - i., on overpotential around the reversible potential (linear micropolarization relation) and the significance of the exchange current density. i.:




io. AND


The Butler-Volmer equation (3.15) can be represented graphically by plotting out the individual cathodic and anodic partial current densities, ic and ia (Eq. 3.14) as a function of overpotential, 17,expressed as negative or positive values around the reversible potential (17 = 0 or V = Vrev). This is shown in Fig. 3.6 in a schematic way. For fJ = 1 - fJ = 0.5 for the two terms of the Butler-Volmer equation, the shapes of the ic and ia partial current density curves are identical but they are oriented in different directions with respect to the axes of the current density vs. 17diagram (Fig. 3.6). The two curves become asymptotic to the overpotential axes as 17--? ± large values (± 00), i.e., ic --? 0 at large +17and ia --? 0 at large -17 (see Eq. 3.15).

The second significant point is that both curves intercept the axis of i values at +io for the anodic curve and -io for the cathodic curve. The distance between the intercepts in terms of current density is two times io' With increasing negative 11, the ic becomes progressively larger than ia until ic » ia and Tafel behavior applies, i.e., ic then increases exponentially with -17· On the other side of the curve, for positive polarization, ia behaves similarly with positively increasing 11. On logarithmic Tafel plots (i.e., for log ic or log ia plotted against appreciable negative or positive 11 values), the intercepts oflog ic or log t; on the 17axis give the log of the exchange current density. Thi s is one of the standard procedures for experimental determination of log io or io' If the net current density, i = ic - ia (see the Butler-Volmer relation in the form of Eq, 3.20), is plotted along with the individual component's partial ic and


Chapter 3

Energetics and Elements of the Kinetics of Electrode Processes

Increased stirring rate


ia values, then a linear polarization (the so-called "micropolarization relation") relation results, passing through zero 17when i = ic - i; = O.The latter condition corresponds to complete reversibility of the electrochemical reaction. It is from this linear relation between net i and 17for small ±17values that the equation of the reaction resistance arises as shown in Eq. (3.23). Its physical significance is represented by the slope, dnldi, of the linear region in Fig. 3.6 for small ±1J values corresponding to the condition for which the two exponential terms in Eq. (3.14) can be expanded to the linear terms of two Taylor series that, respectively, represent the exponential terms of the ic and i, components of i in Eq. (3.14). It is this derivative, dn/di, that is identified as the Faradaic reaction resistance, RF, derived in Section 3.4.2. In the representation of the equivalent circuit, RF is important in the time-dependent (transient and frequency-response) behavior of pseudocapacitors and of double-layer capacitors under certain conditions when overcharge or Faradaic leakage currents have to be considered. If fJ 0.5, then the slopes of the two partial-current polarization curves are different beyond the linear polarization region. This can lead to Faradaic rectification phenomena in alternating voltage modulation experiments. The exchange current density is represented by the equal and opposite current densities that are passing in opposite directions under conditions of electrochemical equilibrium (17 = 0 and V = Vrev)' This situation, which can be determined electrochemically, can be verified by means of isotopic exchange rate measurements, as has been done by Losev" for certain systems. The study of electrode reactions at equilibrium and the determination of the dynamic exchange rates are thus much easier experimentally than for regular chemical equilibria, where special isotopic labeling techniques are usually required.



F z

a.. a:: w

LOG [IimitingI

LOG [i 0]







~ a::



~ ~ u « w



FIGURE 3.7. Onset of diffusion control in a current density vs. potential relation for the kinetics of an electrode potential, at high current density or low reactant concentration. The effect of stirring or electrode rotation is also shown.

For so-called "activation-controlled" electrode processes, 17increases linearly with log i in the absence of any resistive (ohmic) potential drops in the solution. However, for high current densities and/or large io values, concentration polarization may set in owing to the kinetic limitations associated with finite rates of diffusion of the reactant ions or molecules to the electrode surface. Then, toward high i values, the experimental Tafel plot of 17vs. log i can curve upward with an approach toward a diffusion-controlled limiting current as illustrated in Fig. 3.7. Such diffusion-limited currents can, however, be increased by means of solution stirring or forced flow, or by rotation of the electrode. Then the diffusion-limited currents can be increased in proportion to the square root of the electrode rotation rate (Levich relation) in the case of an electrode having

the geometry of a disk with the rotation axis normal to its surface. Such experiments can be conducted quite easily with modern rotated-disk electrode equipment so that the role of diffusion limitation can be quantitatively characterized and, for appropriate conditions, the diffusion constant of the reactive species determined. However, this procedure is inapplicable for porous electrodes.


Chapter 3

Energetics and Elements of the Kinetics of Electrode Processes




Equations (3.7) or (3.14) have been written for a simple discharge reaction exemplified by the first step of the hydrogen evolution reaction (HER) or by the forward direction, or both directions of the R underpotential deposition (UPD) process. It should be mentioned that other steps in the HER may be rate determining, such as electrochemical atom-ion desorption of H forming H2, or the chemical desorption of two adsorbed H atoms by catalytic recombination to H2. Such steps are represented by electrochemical rate equations different from (3.7) or (3.14). However, the details are not relevant to the material treated in this volume but may be found worked out at length in the General Reading Refs. 3,4, and 5. The case actually examined here illustrates the general principles and energetics of electrode process kinetics involving electron transfer and overvoltage. When a step subsequent to that of the initial electron transfer in the cathodic direction of the HER, Tafel slope values (b) can be different from those usually observed, namely, 2.3RTlf3F with 13 =: 0.5. They then depend on (1) the type of desorption step involved, i.e., whether it is of the electrochemical kind (with a further electron transfer event) or of the chemical kind, e.g., discharged radical recombination or dissociation (as for H2 formation in the HER or involving CH3COO- and CH; in the Kolbe reaction that produces C2H6 from discharge of acetate); and (2) on the extent of coverage by the electro sorbed intermediate, i.e., whether the fractional coverage () is «1 and then potential dependent or whether at elevated overvoltages ()~ 1 and is then virtually independent of potential. The Tafel slopes that arise are then generally of the form 2.3RTI(n + f3)F or 2.3RTlmF where n or m are integers; the former arises when desorption of an adsorbed intermediate is rate controlling via a further electron transfer and the coverage by the intermediate is potential dependent according to some quasiequilibrium e1ectrosorption isotherm (Fig. 3.8); the latter arises for a nonelectrochemical desorption step involving, for example, recombination of adsorbed radicals (m = 2) (Fig. 3.8) or dissociation of a radical (m = 1), e.g., CH3COO·. Values of norm are usually the integers 1 or 2, but for more complex sequences (e.g., in anodic O2 evolution), n can be 2 or m 4.19 When the coverage, (), by an adsorbed intermediate (e.g., H) as a function of potential is known or can be assumed, the Tafel slope, b, takes the useful general form: (3.24)


> o

0: W

~ o >




FIGU~ 3.8. Schematic Tafel relations for alternative rate-controlling stages and chemisorption condltlon~ (BH« I or BH -: I) m the HER: (a) rate-controlling H+ discharge; (b) rate-controlling ~ de~orphon b~ a second H discharge; (c) rate-controlling H desorption by catalytic H + H recombination (Tafel s mechanism).

where the first term on the rhs ofEq. (3.24) is the logarithmic derivative of the adsorption isotherm, ()(V) (e.g., for H in the H2 evolution reaction), and the second term arises from the potential dependence of the electron charge transfer rate through the barrier symmetry factor (13 =: 0.5) or electrochemical Brensted factor." Special cases arise when d In ()ldV = FIRTor d In ()ldV = 0 (() ~ 1). In the case of under potential deposition (e.g. of H), no continuous Faradaic c~rrents pass (i.e., f3FIRT = 0 in Eq. 3.24), and adsorption pseudocapacitance anses because () (e.g., ()H)is a thermodynamic function of V over some defined potentiat range, about 0.05 to 0.35 V, depending on the substrate metal, which IS usually a noble one.




. In the kinetic equation (Eq. 3.9) for i, the term for concentration of the re~ctmg species (for example, H+ discharge) was written as CH+ the concentratIo fh . " . . s' not e reacting proton (or some other IOllICreactive species). In terms of the structu~e of the double layer at the electrode interface (Chapter 6), the local concentratIOn of reacting cation (here aqueous H+) has to be expressed in terms of ~he bulk concentration of that species cH +b' From the theory of ion distribution In the double layer (Chapter 6), the local concentration at the outer Helmholtz plane is given by


Chapter 3


and Elements of the Kinetics of Electrode Processes


(3.25) where 'fI1 is the potential at the inner limit of the diffuse double layer (i.e., .at t~e Helmholtz compact layer) relative to the potential in the bulk of the sol~tlOn m the absence of net ionic currents that could lead to an ohmic iR drop. This ISthe potential that is derived from the theory of the diffuse double layer according to Gouy and to Chapman (Chapter 6, references 7 and 9, respect~vely). (In where the ions are significantly chemisorbed [usually amons at positively charged electrode surfaces or near the potential of zero charge, pz~, at negatively charged surfaces as well], the potential w, is appr~ciably modl~ed.) In the kinetic equation (Eq. 3.9), CH:s must be substituted accordmg to Eq. (3.25). In addition, in Eq. 3.7, when double-layer effects are to be prop~r~y included in the rate equation for i, account must be taken of the fact that It IS the local interfacial potential difference V - 'fI1' rather than V (or '7), that has to be written in the rate equation. When these two effects are included (in t~e a~sence of specific chemisorption of the reacting ions), the full electrode kinetic rate equation for the example of proton discharge becomes from Eq. (3.7):

layer region of the electrode/solution interphase. * A first attempt to include such specific adsorption effects in the theory of the double layer was, however, made in 1924 by Stern and later by Grahame, as described in Chapter 6. The original theories of Gouy and of Chapman were nonspecific in this regard since they treated only the diffuse part of the double layer. The specificity of ion adsorption effects, especially those involving anions, is especially manifested in the double-layer capacitance behavior at electrode interfaces as a function of electrode potential. Examples based on the results of Grahame+' will be shown in Chapter 6 and are important in determining electrochemical capacitor behavior in the case of the doublelayer types of capacitor device in aqueous or nonaqueous media, using higharea powdered or filamentous (fibrous) carbon preparations. It is to be understood that this is an elementary and brief account of the energetics and kinetics of electrode processes, such as may be useful as a basis for understanding the principles that may govern the involvement of Faradaic electrode processes in electrochemical capacitor behavior. This applies particularly for cases where overcharge, overdischarge, self-discharge, and Faradaic pseudo capacitance charging processes arise in this field. General source references on electrode kinetics are listed separately later.

. exp -[fJ(V -









In dilute solutions, the theory for ion distribution in the re~ion near and .up to the compact Helmholtz layer of the double layer formally g.Ives an ap_proxI~a~e solution for the ionic concentration dependence of 'fI1 for dilute solutions (ionic strength < ca. 0.1 M) as: (3.27)

Note that for a cathodic process (i.e., one involving discharge of cations), 'fI1 is a negative potential (relative to the bulk solution potential) that becomes numerically smaller with increasing ionic strength. The effective thickness of th.e diffuse double layer (the Debye length) becomes smaller in proportion, apprOXImately, to the log of the ionic strength at relatively low ionic concentrations. The basis of double-layer effects in electrode kinetics was thoroughly worked out by Frumkin and his school in the 1930s and by Gierst in more specialized ways in the 1950s and 1960s (see Refs. 11 and 20). For the case of specific (chemisorbed) ions in the double layer, usually polarizable and electron-donative anions, the double-layer effects cannot be treated in a generalized way, i.e., without individual evaluations of their adsorption behavior and their effects on the local potential difference across the compact Helmholtz

In the field of electrochemical capacitor development and testing, and on the fundamental side of the subject, it is usually necessary to be able to quantitatively evaluate capacitance and its dependence on various experimental variables by direct instrumental measurements. All electrode interfaces with solutions or solid electrolytes exhibit a double-layer capacitance, as treated in Chapters 6 and 7, and in addition some exhibit a pseudocapacitance as treated in Chapters 10 and II. Several simple and elementary criteria for characterizing the electrical response of an electroactive material as behaving capacitatively are as follows: 1. In an interfacial charging process at constant current density, i, the potential difference, Ll V, developed across the capacitor plates changes linearly with time as the charge supplied by i builds up across the interface, i.e.

c= LlqlLlV


"The term interphase is distinguished from interface in order to recognize the 3-dimensional nature of the former as a region of finite thickness residing on the 2-dimensional interface defined by the boundary of the electrode surface with the solution.

54 and !1q =

Chapter 3


and Elements of the Kinetics

of Electrode



J i-dt


= J i . dtl!1 V = i . !1tl!1V


ble layer continues to be increasingly charged with the rise of !1V, but the current i then becomes divided into two components, id1 and iF, where iF is the Faradaic current, which increases exponentially with !1V when !1V exceeds a value corresponding to a thermodynamic reversible potential for solution decomposition (e.g., H2 evolution) at that electrode. This increase of iF follows a Tafel relation in the overpotential, 1], involved, as discussed earlier and represented by Eq. (3.13). Then the charging situation can be written as i = C(dVldt) + iF (3.31)

over some time interval !1t. Equation (3.30), as written, applies if the capacitance, C, is constant with potential. Often experimentally it is not, so !1V deviates from a linear dependence on time at constant current. C is thus obtained as the reciprocal of the slope of the relation between !1V and !1t (Fig. 3.9), the time elapsed to some point on the !1V vs. !1t curve or differentially, at some point on it, as (d!1VI dtr1 when C is not constant with !1V. The measurable dependence of !1Von time at a constant current (i.e., on charge passed) is commonly referred to as the "charging curve." This applies to a so-called "ideally polarizable" electrode (see Chapter 6) where the current i simply passes charge into the interface without any Faradaic processes of transfer of charge across the double layer taking place, leading to chemical change at the electrode surface. As long as that condition obtains, the electrode interface remains ideally polarizable and i is purely a double-layer charging current, id1• However, in practice, the charging current may be maintained in a voltage range across the interface where Faradaic decomposition of the solution begins to take place at that electrode (and then also at the counter electrode). The dou/
/ /

where C(dVldl) = id1 and iF has the form iF= io exp [al1FIRTj, which is the Tafel equation in exponential form (Eq. 3.7), with a the transfer coefficient. The charging curve then has the form shown in Fig. 3.9 (for constant C) and as LlV increases beyond its value corresponding to solution decomposition, the LlV vs. Lll relation increasingly deviates from linearity as the iF component of i becomes progressively larger, i.e., a greater leakage current across the double layer passes until the Faradaic process is the main charge-carrying component, iF» id1. Then the electrode is behaving in a nonideally polarizable way. The situation of transition from double-layer charging to mainly Faradaic decomposition of the electrolyte as the potential is raised is illustrated in Fig. 3.9. Region ab is for pure double-layer charging; region be includes an increasing component of the passage of charge in a parallel Faradaic reaction. ab is the region of ideal polarizability of the interface, prior to decomposition of the solution over region be. Figure 3.10 shows schematically the relative values of the



b /.


!z w

/ ...---

Onset of non-idealpolarizability, iF>Q


onset of non-ideal polarizability, iF 0

ffi a:

a: :;,

Constant i = idl+ iF

FIGURE capacitance 3.9. C. Constant current charging curve for an electrode interface having a double-layer FIGURE 3.10. Components of the constant charging current of Fig. 3.9 at an electrode interface When a Faradaic partial current, ir, passes in parallel with the double-layer charging current, idl.




and Elements

of the Kinetics

of Electrode



components id1 and iF of the total (constant) i as a function of increasing potential as the interphase becomes increasingly nonideally polarizable, corresponding to decomposition of the solution. Quite generally, a charging current is C dVldt (since C dV == dq) where dVldt is the rate of change in potential with time in the absence of any time effects (e.g., deactivation) in the kinetics of a Faradaic reaction when iF becomes > O. In the first ascending region of Fig. 3.9, where iF = 0, dVldt at a constant current gives directly the reciprocal of C. In the region where iF becomes> 0, C dVldt still gives the non-Faradaic component of i, which can be evaluated if C has remained constant during the charging process with the increase of potential beyond the solution decomposition limit. Eventually, a steady state is reached where iF ~ i as dVldt ~ 0 and then id1 ~ 0, i.e., the charge transfer is entirely Faradaic at a sufficiently high steady potential or overvoltage for the Faradaic reaction. The double layer can also be charged potentiostatically in a potential step or potentiodynamically in a linear voltage sweep as described later. 2. In a sequence of potential steps, JV, the charge, oq, flows into the interface to an extent determined by CJV where C is the capacitance over the potential range <5 V. A current transient it is generated (Fig. 3.11) by the pulse J V over a small but finite response time and Jq is the integral of it over that time interval. The capacitance can be calculated from the recorded Jq response for the pulse JV, or the sum of Jqs over a sum of sequential JVs. In electroanalysis, this procedure is known as chronoamperometry. As in the first case of constant-current charging (chronopotentiometry), as potential is increased toward and through the solution decomposition limit or if a large-amplitude potential step, ~ V, is applied, the transient current response it contains transient components of doublelayer charging current and of Faradaic current. The response time, Jt, in which the transient current it passes, is ideally very short but in practice it can be tens of microseconds or some milliseconds, depending on the impedance characteristics of the cell and the measurement system. The time integral of the current-response transient, iII' in Fig. 3.11 gives the charge that has flowed into the electrode interface as a result of the application of the potential step, JV or ~ V. If the potential step covers some range of potential where a Faradaic current, iF, also flows (upper curve in Fig. 3.11), then the charge passed by this parallel current component must be allowed for in calculating the charge entering the double-layer capacitance. In some types of electrode-process measurements, a double-pulse procedure developed by Gerischer is employed, first to charge up the double layer quickly and then to observe the kinetic response of an electrode reaction to a second, subsequent pulse. 3. In a so-called linear potential-sweep (potentiodynamic) experiment, the applied potential (measured with respect to a reference electrode) is varied line-


W en z


en w a: a: a: ::;) o


Potential- step V

!z w

,, ,




FIGURE 3.11. Current-response

TIME (order of !lS or ms) transient for charging an electrode interface under the influence

of a step of potential from Vl to V2.

arly with time in a three-electrode cell and the resulting time-dependent response current (Fig. 3.12) is registered by an analog (X-Y recorder or oscilloscope) or digital (transient recorder, computer, or digital oscilloscope) recording instrument. In such a voltage sweep experiment at a sweep rate s = dV/dt (= constant), the capacitative charging current, i, is
i = C (dVldt) = Cs


or C= ils (3.33)

In a sweep-reversal experiment (cyclic voltammetry) at constant ±s, the current response profile is ideally a rectangle along the time-potential axis when C is constant (Fig. 3.12, line a) or if it is not, then the differential profile of C is generated, often with a peaked structure (line b in Fig. 3.12). 4. In a self-discharge experiment through a load resistance, R, the time dependence of potential is (see Chapter 18) Vet) = Vo exp [-tIRC] (3.34)

from which C can be evaluated; and 5. In an ac impedance experiment (Chapters 16 and 17), the (imaginary) component of impedance is Z'= l/jwC (3.35)


Chapter 3

Energetics and Elements of the Kinetics of Electrode Processes


" u

pedance behaves like that of a transmission line having a 45° phase angle over a wide frequency range instead of 90° for ideal capacitance. Finite or significant R limits power output performance as discussed in Chapter 15.

...: z








V, _ --_






0 0



Sweep - signaj"-

...... _A:'


b .......

_ --....





I~ , t~ >I


1% ·0


til " U 10




3.10.1. Cells and Reference Electrodes 22 Angerstein-Kozlowska has given a detailed account of various experimental procedures for electrochemical experimentation, including the design of cells and preparation of solutions, especially techniques for measurements in very pure solutions and on clean electrode surfaces. Almost all packaged electrochemical capacitors or batteries are sealed two-electrode (or a stacked series of two-electrode-) systems. However, electrochemical test and evaluation information is often required on individual electrodes (the positive or the negative one) of the device. In this case, a third, unpolarized, electrode is required as a reference probe. Provision for such measurements must then be made by using a three-electrode or three-compartment cell of the type shown in Fig. 3.13. Alternatively, a third (reference) electrode probe can be inserted between two working plate-type electrodes in a cell of the type shown in Fig. 3.14. Then, during tests, the potentials of each working electrode on charge, discharge, or open circuit (for self-discharge measurements) can be separately recorded against the potential of an appropriately compatible reference electrode. Some commonly available reference electrodes are as follows: The Hz-H+-H20 or Hz-OH--H20 electrode at platinized Pt. The PdH electrode at Pd containing sorbed H. The dynamic hydrogen electrode where H2 is generated in situ at low current density at a small platinized Pt electrode. Hg-Hg2Clz-CI-, reversible to CI- ion. Hg-Hg2S04-S0~-, reversible to SO~- ion. Hg-HgO-OH-, reversible to OH- ion at pH > 9. Ag-AgX-X-, reversible to halide, X-, ions. Ag-AgOH film-OH-, reversible to OW ion. Fe-FeF2 film or CU-CuF2 film for F- ion solution. Ag-Ag+-AgCl04 in nonaqueous solvents. Glass electrode, sensitive to pH in aq. solutions. Ion-selective electrode (Orion Co.) sensitive to various specific ions in solution.

FIGURE 3.12. Cyclic voltammogram for an interface having a potential-independent tance (C = its) (curve a) and potential-dependent capacitance (curve b).


and ideally the real component Z' is infinity for an ideally polarizable electrode. Most capacitor systems, especially those of the electrochemical type, have a complex R-C equivalent circuit of distributed Rand C due to porosity (see Chapter 14) and/or complex electrochemical reactions, and hence do not h~ve a simple impedance behavior as a function of frequency. In some cases, the im-


Chapter 3


and Elements of the Kinetics of Electrode Processes






:. I

Plate - type cell in plastic frame (Schematic)



FIGURE probe. 3.14. Plate-type test cell with arrangement allowing insertion of a reference electrode

When a reference electrode is used, potential measurements or other connections must be made into or through a high input impedance (ca. 106 ohms) recording device in order to minimize current drain from the reference electrode to below several microamperes. This ensures that the reference electrode maintains its proper reference potential to within a millivolt, i.e., it does not itself become polarized on account of its inclusion in the measurement circuit. A typical three-electrode measurement circuit is shown in Fig. 3.15.

3.10.2. Instruments The various electrochemical instrumental procedures available for studying the behavior of capacitor devices and batteries, and electrode processes in general, are well treated in the monographs by Delahay (General Reading Ref. 1) and by Bard and Faulkner (General Reading Ref. 6). Basically, for the evaluation of electrical response functions of electrochemical capacitors, the following equipment is required: 1. A potentiostat [e.g., EG and G (PAR), Hokuto or Wenking] that is capable of addressing and controlling cells up to 5 V. Some higher voltage options





cell for evaluation

of electrode


against a reference electrode.

Chapter 3 62

Energetics and Elements of the Kinetics of Electrode Processes




x- y



Operating electric circuit for potentiostatic system.

or potentiodynamic


with a function generator using a three-electrode

are available. Driving output voltages up to 50 V are sometimes required. For more ambitious testing of larger devices, higher power potentiostats are neces~y. .' . 2. A function generator capable of generatlllg single-step. slllgie-ramp, and repeated step and ramp potential forms is needed to drive the potentiostat for transient measurements and cyclic voltammetry. Most function generators also provide sinusoidal output signals at various frequencies and amplitudes. 3. A controlled current generator for dc charging and discharging experiments. Computer-controlled, multichannel systems are now available (e.g., from Mackor or Arbin) with the capability of repetitively recording the charge and discharge curves of potential with state of charge. 4. A digital oscilloscope with diskette recording, e.g., from Nicolet Corp. 5. Computer interfacing and appropriate software for data acquisition and processing. Electrochemical experiments provide ideal oPP?rtunities for cO.mputer control of the instruments themselves, and for recording and processing the resulting data. Potential and current are the conjugate variables of electrode processes, coupled with charge, the integral of the current passed during a controlled time interval (also digitally controllable down to the microsecond level or less). These variables can be precisely controlled and recorded in a digital

manner. P: variet.y ?f software is commercially available and "in lab" personal software IS not difficult to design and generate. . 6. A sy~tem for determining the frequency response of the capacitor deVIce over a WIde range of frequencies from 0.001 or 0.01 Hz up to 100 kHz or sometimes more. ~his enables the real (ohmic) and imaginary (capacitati~e) components of the Impedance of the system to be determined over the frequency range scanned (see Chapters 16 and 17). This is a widely used procedure and is essential for full evaluation of capacitor performance and the behavior of capacitor devices in the various circuits in which they are utilized. Excellent instruments are now available (e.g., from Schulumberger Solartron or EO and 0), together with the required software for complex-plane impedance and phase-angle plots. 7 .. A coulometer; this is useful for automatically recording the charge passed mt~ or ~ut from an electrochemical energy storage system over a given recorded time interval. Instruments giving digital or analog outputs are available. Coulometric measurements are specially relevant to evaluating the performance of electrochemical capacitors, coupled with recordings of cell voltages (see item 3). 8. A calorimeter or thermocouple recording system may also be a useful item of instrumentation. Especially for large capacitor devices, the problem of heat management becomes a significant factor in practical operation of capacitors at large current densities. This becomes of greater significance when scaleup of small devices to larger ones is under consideration. 9. Other equipment: various more specialized equipment is required for some of the more fundamental research operations in the field. For example, Gottesfeld (see Chapter 12) has profitably used the techniques of ellipsometry (~ee Chapter 12) and quartz microbalance gravimetry (to nanogram sensitivities) to examine the development and growth of conducting polymer films on substrates for electrochemical capacitor fabrication. Also, surface spectroscopy (e.g., at C) has been useful. . . 10. Of course, other regular minor equipment such as portable multirange digital ammeters and voltmeters is a daily requirement in an electrochemical testing laboratory. In the case of fabrication of nonaqueous double-layer capacitors, a good dry box (Vacuum Atmospheres Corp.) is a necessity, or better, on a larger scale, a dry room.

3.10.3. Two-Electrode Device Measurements Most packaged electrochemical capacitor devices are two-terminal systems,' so no reference electrode can be used as a third potential probe without making an opening in the device's case. Testing is therefore conducted mainly by recording charge and discharge curves at controlled currents at various tem-


Energetics and Elements of the Kinetics of Electrode Processes


Chapter 3


peratures and by potentiometric recording of open-circuit potentials or float currents (Chapter 18) when self-discharge rates are being evaluated. Under these conditions, of course, an overall evaluation of the two-electrode system is obtained so that information on the behavior of each electrode interface is not available from such measurements. However, in practice, useful performance information on the charge and discharge capacity, energy density, and power density of capacitor devices is easily obtainable by means of two-terminal measurements, but is less informative in a fundamental direction than data obtained at individual electrodes in three-electrode cells.

2. P. Delahay, Double-Layer and ~lectrode Kinetics, Interscience, New York 1965 3. K. J. Vetter, Electrochemical Kznetics, Springer-Verlag, Berlin (1965). ( ). ~: ~.g;~o~wa~ Theory and Principles of Electrode Kinetics, Ronald Press, New York (1964) . . oc s and A. K. N. Reddy, Modern Electrochemistry vols. 1 and 2 Pie . York (1970); Second edition, Plenum, New York (1998).' ,num, New 6. A. J. Bard and L. ~. Faulkner, Electrochemical Methods, Wiley, New York (1980). 7. ~.~. ~~~;ay, Io;~c Hydration in Chemistry and Biophysics, Elsevier, Amsterdam (1981) 8. N' 'Y am an . E. Myland, Fundamentals of Electrochemical Science Academic P . ew ork (1994). ' ress, 9. J . O'M . Bockris ,10 M 0d ern A spects of Electrochemistry, . 4, Butterworths, London (1954). J. O'M. Bockris ed v I 1 Ch ' ., 0., apter

1. L. Pearce Williams, Michael Faraday: A Biography, Chapman and Hall, London (1965). 2. B. E. Conway, in Electrochemistry, Past and Present, Chapter 11, ACS Symposium Series 390, American Chemical Society, Washington, D.C. (1989). 3. 1. Weiss, Proc. Roy. Soc., London A222, 128 (1954). 4. 5. 6. 7. R. R. R. R. A. Marcus, 1. Chern. Phys., 24, 966 (1956). A. Marcus, Ann. Rev. Phys. Chem., 15,15 (1964). Parsons, and 1. O'M. Bockris, Trans. Faraday Soc., 47, 914 (1951). W. Gurney, Proc. Roy. Soc., London, Al34, 137 (1931); A138, 378 (1932).

10. W. Schmickler, Interfacial Electrochemistry, Springer-Verlag Berlin (1993) 11. J . Koryta , J . Dvorak ,an. d J B 0 h ac k ova, Electrochemistry, Methuen, London (1972). ,. 12. L. Antropov, Theoretlcal Electrochemistry, Mir Pub!', Moscow (1975). 13. 14.

Gilea~i, E. ~irowa-Eisner, and J. Penciner, Interfacial Electrochemistry, an Experimental pproac ,Addison-Wesley, Reading, Mass. (1975). 15. Southampton Electrochemistry Or l H dC. oup, nstrumental Methods in Electrochemistry Ellis orwoou, hichcster, UK (1985). '



EhlectrNode Kinetics for Chemists, Chemical Engineers and Materials ers, ew York (1993).



8. F. P. Bowden, Trans. Faraday Soc., 28, 368 (1932). 9. J. A. V. Butler, Proc. Roy. Soc., London, A157, 423 (1936). 10. B. E. Conway, Ionic Hydration in Chemistry and Biophysics, Chapter 13, p. 260, Elsevier, Amsterdam (1981). 11. B. E. Conway, Theory and Principles of Electrode Processes, Ronald Press, New York (1964). 12. R. Parsons, Trans. Faraday Soc., 34 1053 (1958). 13. B. E. Conway and 1. O'M. Bockris, 1. Chern. Phys., 26, 532 (1957). 14. A. A. Balandin, Z Phys. Chern., B2, 289 (1929). 15. S. Trasatti, 1. Electroanal. Chem., 39, 183 (1977). 16. J. W. Schultze and F. D. Koppitz, Electrochim. Acta, 21, 327, 337 (1976). 17. B. E. Conway, in Progress in Reaction Kinetics, G. Porter, ed., vo!. 4, Chapter 10, Pergamon, Oxford (1967). 18. B. Losev, quoted by L. Antropov, Theoretical Electrochemistry, 19. 20. 21. 22. p. 395, Mir Pub!., Moscow,

1972. K. J. Vetter, Electrochemical Kinetics, Springer-Verlag, Berlin (1965). L. Gierst, in Electrochemical Society Symposium on Electrode Processes, E. Yeager and P. Delahay, eds., Electrochemical Society, Pennington, N.J. (1961). D. C. Grahame, Chem. Rev., 41, 441 (1947). H. Angerstein-Kozlowska, in Comprehensive Treatise of Electrochemistry, E. Yeager, J. O'M. Bockris, B. E. Conway, and S. Sarangapani, eds., vo!. 9, Chapter 2, Plenum, New York (1984).


1. P. Delahay, New Instrumental Methods in Electrochemistry, (1954). Wiley-Interscience, New York

Chapter 4

Elements of Electrostatics Involved in Treatment of Double Layers and Ions at Capacitor Electrode Interphases

4.1. INTRODUCTION The electrochemistry of double layers, and the ions and solvent molecules constituting them, involves the electrostatic energies and molecular or ionic distributions of these species in high interphasial fields. At charged electrode interphases in double layers, the electric fields can become as high as 107 V cm ", and similarly in the solvation shells of ions. The electrostatics of such interphases are concerned with: 1. the energies of individual charges and molecular electric dipoles, and of their interactions; 2. the motions of individual charges (ions) in fields; 3. the configuration of groups of charges on ionized complex molecules (e.g., ionic centers arising at conducting polymers) currently being developed as pseudocapacitors; 4. the orientational movement of electric dipoles in homogeneous fields; 5. the translational motion of dipoles in inhomogeneous fields; 6. the interaction of solvent dipoles with electric fields, and quadrupoles with field gradients; and 7. the effective local dielectric coefficient in the double-layer interphase.


Chapter 4


of Double Layers and Ions


Magnetic interactions are much weaker and are only important in spectroscopy, in some solids, and in transition metal ion properties. When a charged electrode interface is in contact with an ionic solution, there is an accumulation of ions of one sign or the other, forming a double layer (Chapter 6) and causing an orientation of dipoles of the solvent. Since ions in the double layer are normally situated in a liquid solvent in contact with the electrodes, the electric polarization of the medium will also be involved in determining the properties of the charged particles at or near an electrode interface. In order to provide a basis for discussing the details of such questions in the theory of the double layer and electrochemical capacitors, we present in this chapter some elementary principles of electrostatics.

a force of 1 dyne or equivalently in newtons. The value of the proportionality constant k in Eq. (4.1) in this system of units is simply 1. The electronic charge that arises in many calculations in the electrochemistry of double layers and ionic solutions has a value of 4.80 x 10-10 esu in this system of units, or the Faraday constant in coulombs per gram-equivalent divided by Avogadro's number in the electromagnetic units system, giving the results in coulombs as 1.6021 x

In the so-called "rationalized" mksa (meter, kilogram, second, ampere) system of units, k has the value 8.9876 x 109 N m2C-2, i.e., in newtons meter2 per coulomb/ where F, the force, is measured in newtons, r is in meters, and q1 and q2 are in coulombs. In this system of units, the constant k is written k







Law: Electric Potential and Field, and the Significance

of the Dielectric Constant

Perhaps the two most basic laws in physics and chemistry that determine the short-range properties of substances are Coulomb's law of electrostatic force, F, between two charges ql and q2, distant r from one another, and the corresponding law of interaction that determines the force between two magnetic dipoles. A third basic relation is the Pauli exclusion principle, which places restrictions on quantized energy states in assemblies of electrons in atoms or molecules and is hence fundamental in defining interatomic forces arising from electronic interactions in molecules. Analogous effects occur in metals, giving rise to their band structure. Coulomb's law has been experimentally verified as an inverse square law like that for gravitation. The general relation for the force F between two charges in a vacuum, is given by (4.1) where F is an attractive force when ql and q2 are of opposite sign and a repulsive one when they are of like sign. The constant k depends on the units in which the charges and distance are expressed and the force F is also dependent on the electrical properties of the medium in which the charges q1 and q2 are situated when it is not a vacuum. Units In the electrostatic system of units, the unit charge is defined by Coulomb's law as that charge which repels a like charge 1 em distant from it in vacuo with

where Ko is the permittivity of the vacuum. The rationalized system is introduced so that certain fundamental relations in electrostatics, such as Gauss's and Poisson's equations, can be written without the 471.' factor which otherwise enters into them. The value of KO is

= 8.85435

10-12 C2N-lm-2 or farads rrr '


In some texts the symbol eo is used for the permittivity of free space, but the symbol KO is preferred here in order to avoid confusion with the static (zero frequency) dielectric constant of a medium, also usually denoted by eo or eO. KO is also sometimes referred to as the diabattivity of free space. Electrostatic calculations in the rationalized units system thus depend on forces calculated from F; = QIQ2/4nKoe? or U, = QIQ2/4nKoBr from corresponding energies. The charges are written QI> Q2 here in coulombs to distinguish them from ql> q2 in Eq. (4.1), where the units are to be taken in esu. It is often, however, simpler to perform such calculations in the esu system with e then taken simply as the dielectric constant (see later discussion) of the medium; then forces appear as dynes and energies as ergs per particle. Conversion to joules mol"! is easily achieved by multiplying by NAil 07, where NA is Avogadro's number, noting that 107 ergs = 1joule. It is obvious that the two systems must give the same results with the appropriate interconversion of units. Thus a unit charge in coulombs is 3 x 109 times that in esu, or 1 C = 3 X 109 esu. The conversion factor 3 x 109 (actually more exactly 2.998 x J09) originates from the speed of light in a vacuum, 2.998 x 1010em S-I, and is 1110 of this figure, corresponding to the coulomb being a practical unit that is 1110 of the value of the absolute electromagnetic unit of charge, which is itself defined in terms of the magnetic field produced in a circular coil of one turn having a radius of 1 em, When a unit charge per second (current) is flowing.




of Double Layers and Ions


A comparison of calculations in the two systems of units for the energy of two like unit charges 1 em apart enables KO to be identified. In the esu system, for example, for ql = q2 = 1 esu in vacuo (s = 1): (4.4) In the emu system, using SI units, ql = q2 = (l(esu))/(lOc(ms-I)) = (1 erg)/(3 x 109) coulomb, where c is the velocity of light (2.998 x 108 m s-I). The energy Vr=lem must be the same in either system of units, but with r taken in meters and ql or q2 in coulombs in the Sl system. Hence

electrical properties of molecules of the medium is treated later. In terms of Eq. (4.1), it is seen that the quantity e characterizes the attenuation of the coulomb force of interaction between two (or more) charges by the material medium in which they are separated. Electrostatic Potential, Field, and Force For a charge Q, the electric potential ¢ at a distance d from the charge is defined as the work done to bring a unit charge from (charge-free) infinity to the point at distance d. At any distance x from the charge Q, the unit charge experiences an electrostatic force given by qxl



-F(x) =-2-



J (r = 10-2 m, ql = q2 = 113

109 coulombs)


Therefore, equating the result in joules from the two systems, it is seen that
Vr=1 em

If the unit charge is progressively moved from to d, increasing work is done against this force so that the work or energy VeX)expended is given by

= 10-7 J = 10-16/9

4nKo J

Vex) =


d F(x)dx =

s. f e

d~ = q/ed


10-16 = --7 /9


4n = 8.85435

10-12 farad rrr '


This is defined as the potential ¢ due to q at distance d; thus


as mentioned earlier in this chapter (Eq. 4.3). The significance of Ko also follows from the important definition of capacitance (see Section 4.4) since C = (As )/( 4nd) where d is the thickness of the dielectric medium (dielectric constant s) between two parallel plates of area A. In order for C to be correctly evaluated in terms of A and d in the emu system (C in farads or coulombs per volt), the relation must be written C = ASKO/d with KO = 8.85435 X 10-12 farads m-I. Dielectric Constant The dielectric constant, e. of a medium is measured by the capacitance C of a condenser containing the medium, relative to that of the same condenser in vacuo. As defined earlier, the capacity of the condenser in vacuo is sof4nd per unit area of the plates separated by a distance d. In a material medium, the capacitance is ASKo/d per unit area where Ko is the permittivity of the medium. The dielectric constant S is defined as the relative permittivity, K/KO, and hence has the value 1 for free space and is a pure number for any material medium. The physical significance of e in terms of the




The corresponding electric field E is defined as the gradient of the potential. For the above case,

= doidx = -q/sx2


Hence field varies as the inverse square of a distance x from a charge while potential varies as the inverse first power of this distance (Fig. 4.1). From the form of Coulomb's law, it is evident that the force on a charge q' in the field of q is given by F = - _!f._



i.e., force is the product of the charge multiplied by the field due to another (or other) chargers). This formula provides the basis for the definition of the unit charge given earlier.


Chapter 4

Treatment of Double Layers and Ions


<PA = <P, + <P2'" <P3

__g_+-q_ --q _
_ Ed, Ed2 Ed3

(0 )

Potential c/>

Field E




Relation between electric potential and electric field as functions of distance, x,

from an ionic charge (schematic).

Potential ¢ and Field E at an Ion

The field is a vector quantity since it depends on the direction of the gradient of the potential ¢>. The potential, on the other hand, is simply a scalar quantity having no direction. It can, however, have positive or negative magnitude, depending on the sign of q generating the potential ¢>. Potentials generated at a given point from a distribution of charges can be algebraically added [Fig. 4.2(a)], but fields at a given point originating from various charges must be veetorically summed, i.e., with due consideration of their directions or their components in given directions [Fig. 4.2(b)]. Strictly speaking, the dielectric constant 8 may also depend on the direction of the field if the medium concerned is electrically anisotropic, i.e., in the case of certain crystals or liquid-crystals, or organic molecules that are birefringent. The resultant field vector can usually be easily constructed for a 2-dimensional array of charges [Fig. 4.2(b)]. For the general case, 2 or 3-dimensional trigonometry is required. For orthogonal vectors, the results are easily calculated by means of Pythagoras's theorem. More generally, the field may be expressed in terms of the unit vector r where E = qrlex", and the result in terms of the vector sum or, within limits, the vector integral. The field calculation discussed here enters into the calculation of forces and fields caused by assemblies of oriented dipoles in liquids around ions or charge arrays on polyions or surfaces.

Field Resultant


= (,


FIGURE 4.2. (a) Scalar summation of potentials due to a distribution of charges at various distances d}, d2, d3 from point A. (b) Vectorial summation of fields to derive resultant field at a given point due to the distribution of charges shown in (a).



Field intensity gives a measure of the so-called "force" associated with the action of the field on a unit charge. One "line" per square centimeter is defined as existing in classical electrostatics for each dyne per esu of field strength. Lines of forces correspond to the tangential force experienced by a charge in the field at the particular point concerned. Density of lines of force corresponds to the flux and electrification. The concept of lines of force originated with Faraday. The force at r on a unit charge due to q is q/r2 (8 1), i.e., the field intensity. This field can be represented by drawing q/r2lines of force per square centime-


Chapter 4


of Double Layers and Ions


ter at r. The total of such lines at a distance is the line density multiplied by the total cross-section at r, i.e.,

i.e., a unit charge has 4

(4nr2) =4nq


lines afforce associated with it.


FIGURE 4.3. Model for discussion of the basis of Gauss's theorem in terms of normal induction across an element of sectional surface oS through a surface Be at distance r from a charge q.

The capacity C is defined as C = q/!3.¢ or, differentially, C(d) = dql d¢. The surface charge density, a, on the plates of the capacitor of area A can be defined as ±a = ±q/A and the field E is E = !3.¢/d where d is the distance between the plates, i.e., C = AalEd. Now, from the above theorem, the charge a is associated with 4n lines of force and this is the field E (= 4na) since a refers to 1 cm". Then Aa A C=--=--or-anad 4nd AB 4nd (4.14)


ds =4nq


with a dielectric present. In the rationalized system, C = ABKoId,as earlier. Equation (4.14) forms the basis for treating other problems (described later) and the experimental determination of dielectric constant. It is also used in determining the electric capacitance of a system of separated plus and minus charges in double layers at interfaces of colloids and electrodes in electrochemical capacitors.

Let q be the enclosed charge and the flux of induction will be In . ds, and this will be I cos ()·dsand be equal to the field (q/?) x (the shaded circle area, BM). However, the area BM/? measures the solid angle r5().Then the flux of induction across the area BC is qdil. Hence the total flux is

f qdii = q f d() = -q . 4n
However, this total induction is BE. Hence


E =-4nq/B



which is a form of Gauss's theorem. If the plate has two surfaces (i.e., it is not a bulk conductor on one side), the result is E = -Znqte.

When the charge distribution is a uniform one on a plane (charged plate or electrode), the resultant field takes a simple form. This is important in electrochemistry, for the field due to an electrode corresponds precisely to this case if the fluid adjacent to the plate is assumed to be a uniform dielectric. Similarly, the field due to a colloid interface may be treated in terms of this relation in limiting cases (neglecting discreteness of charge). This theorem relates the field due to charge density on a surface (e.g., a plane metal electrode) to the electric flux through the surface. Consider the normal induction In across an element of surface r5Sat a distance r from a charge q (Fig. 4.3). It was shown earlier that the total flux of electric induction across a closed surface, s, is 4n times the sum of the charges enclosed by that surface, i.e.




Gauss's equation is useful for relating the field caused by a charged surface to the charge density on that surface, as at an electrode. A related expression, Poisson's equation, expresses field gradients in terms of the space charge density, p, existing in a medium. Poisson's equation is of value in dealing with problems in ionic solutions or ion distributions near charged interfaces, e.g., at double layers at electrodes. Normally a net space charge cannot exist without a corresponding field directed to where the space charge is smaller or zero. In the language of vectors in electrostatics, Poisson's equation is expressed as the divergence (div) of the gradient (grad) of ¢, i.e., how the gradient of ¢, namely,

Chapter 4



of Double

Layers and Ions



the field components doldx, d¢/dy, or d¢/dz, increase or decrease in the various orthogonal directions of the Cartesian coordinate system. In terms of p and the dielectric constant 8 of the medium in which the space charge resides, Poisson's equation is written div (8 grad ¢)

W =f Aq·dq


= -4np


The integral is evaluated by regarding the fraction A as being progressively increased from 0 to 1, so that the integration is made with respect to A rather than q, i.e., because <5qcan be written as q·<5A. Then (4.22)

In terms of the x, y, z coordinate system, and for the case where 8 is not dependent on the field components dokb: d¢/dy or d¢/dz, the above equation is written in terms of the partial field derivatives (ptlo ()2t1o ()2t1o ~+~+~=-4np/s
ox2 oy2 ()Z2


Usually the series of partial second derivatives is represented by the symbol '\72, "del squared" (the Laplace operator), so that (4.20) The mathematical operations required to obtain the Poisson equation can be complex, depending on the degree of rigor involved in its derivation, but a simple treatment gives the required result by application of Gauss's theorem (Eq. 4.17) (General Ref. 1).

The same result may also be obtained by considering the charging energy (1I2)C(tJ.¢)2 of a spherical particle treated as a conductor having a capacitance C equal to its radius, a. If the charging is carried out in a medium of dielectric constant, e. (4.23) A similar concept of "self-energy" applies to a capacitor which has an energy of 1I2CV2 or 1I2C(tJ.¢f

4.8. ELECTRIC TENSION IN A DIELECTRIC IN A FIELD In terms of the electrical work of charging a capacitor [We = 1I2C(tJ.¢i], follows from the derivation of Eq. (4.14) that it

4.7. THE ENERGY OF A CHARGE The idea that a charge is associated with an energy is less familiar than the concept of field or potential due to a charge. As with all energy quantities, it is necessary to specify with respect to what reference state the energy is measured. In the case of a charge, the energy originates from the work associated with building up that charge (e.g., q) by adding elements of charge Jq brought from infinity. Hence the "energy of the charge" is the work done in the building up of that charge from an initial value of zero to its final value q, with increments of charge, oq, transported from reference distance 00 where the potential due to the charge built up at any stage in the charging process is zero. Let Aq be the charge established at a given stage of the charging of the initially uncharged particle, where A is a fraction 0 < A < 1. The electrical work done, JWe, in bringing up the next element of charge oq is (Aq·<5q)/a if a is the radius of the particle being charged. The total work done, W., for transfer of all the charges Jq (i.e., from A = 0 to A = 1) is given by


SE2) [s; Ad


where E is the field tJ.¢/d. Ad is the volume of dielectric in the capacitor so that sE2/Sn must be the electric pressure or really the internal tension associated with the field E, which follows from considering the units of the left- and righthand sides ofEq. (4.24). Equation (4.24) also provides a way of evaluating the self-energy of an ion by integrating the P·dV work experienced by the dielectric from the ion's periphery at a to 00. Thus =fsE2 W = -. 4nr2 . dr = q212sa



noting that Eat r for the charge q is q/s?



Treatment of Double Layers and Ions


As in the case of charging an ion, work is required to produce the induced polarization, given by: (4.27) This work, Wi, is done on the atom or molecule; hence the positive sign for Wi' The induced dipole, Jii, however, interacts with the field E that produced it with an energy -f.1.iE or -aeE2. Hence the total energy of induced polarization is -aeE212.

The electric polarization in a dielectric (i.e., the solvent in the interphasial double layer), originates at the molecular level in the dielectric material. We consider here the electric polarization in individual atoms and molecules and then the interaction of a permanent dipole in a field such as exists in a charged double layer.


Atoms and Molecules

in Fields: Electronic

Polarization 4.9.2. Interaction of a Permanent Dipole with a Field

From a sufficient distance, an atom presents an electrically homogeneous aspect to an applied electric field. However, the field will produce an attraction or repulsion of the electrons of the atom with respect to its positive nucleus and modify the initial, normal charge distribution. The resulting distribution corresponds to an induced electric dipole (Fig. 4.4) of moment Jii. The dipole moment of such an induced charge distribution is defined as e x g, i.e., the length g corresponding to the hypothetical displacement of a unit of electronic charge. The induced electric moment, Jii, is proportional to the field E:

Uniform Field

u, = aeE


where a, is called the "electronic polarizability." It depends on the number and type (0" or n) of electrons in the molecule and the presence or otherwise of conjugation. In fact a, has the units of volume, of magnitude 10-24 crrr'. In addition to displacing the electron charge, an electric field can modify the degeneracy of electronic orbitals and lead to spectroscopic line splitting called the Stark effect (linear and quadratic). Also, a vibrational Stark effect in molecules (e.g., CO) is observable in double layers at electrodes where CO is adsorbed.

A permanent dipole interacts with a field E experiencing (I) an induction of the moment f.1.i determined by its ae; (2) an interaction with the field, f.1.E cos (), where () is the angle of orientation of the dipole to the field [Fig. 4.5(a)]; and (3) a torque if ()::f:. 0 [Fig. 4.5(b)]. If ()::f:. 0, f.1.i and f.1. must be added vectorially (Fig. 4.6) to give a resultant !1R along the field direction c!1R + also the torque tends to align the dipole eventually in the direction of E. The torque arises because the interaction of the charges ±e on the ends of the dipole with E produces a couple. Alternatively, the energy of a dipole, -f.1.E cos (), is minimized in a field as cos () ~ 1 «() ~ 0). The torque is responsible for the libratory motion of dipoles in a field, e.g., at an ion in its solvation shell. Alignment of a dipole in a field is normally incomplete at finite temperatures owing to thermal fluctuations in a gas or liquid. The average extent of alignment depends on the ratio of f.1.E cos () to kT. Only at high E or low kT is the alignment almost complete. At 298 K, E must be > 2 x 104 esu or 6 x 106 V em"! if f.1.E for H20 dipoles is to be > kT.

=!1i !1);

Atom in zero field
Induced dipOle ~--\---moment J..L

Nonuniform Field

Atom in field E

FIGURE 4.4. Electric polarization of an atom or molecular particle in a field leading to an induced dipole moment, tu.

In a nonuniform field, a dipole not only tends to become aligned with the field but to be translated down the field gradient to minimize its energy (i.e., to maximize the numerical value of f.1.E). The movement of the dipole continues as E increases (e.g., near an ion where the field varies as ± zelr.?, or near an electrode surface in its double layer) until the dipole experiences some sufficient Contact repulsive force. This is the reason for the internal compressional force near ions in a polar medium or in a double layer. This leads to "electrostriction" or a decrease in volume of the dielectric medium near ions.


Chapter 4

Treatment of Double Layers and Ions


Field direction

Field direction



Field E

Field direction ED

Field direction 6l

fL cos s


(o )







Vectorial addition of induced momcnt.zz., to permanent moment;«;

giving moment





Field E



Force +eE

FIGURE 4.5. Interaction of a permanent dipole of moment, /1, with an electric field E: (a) Interaction energy with the field, /1E cos I). (b) Torque resulting in orientation.

Insofar as the physical influence of electric fields on a chemical substance is concerned, the substance is referred to as a "dielectric." The material existing between the plates of a capacitor in a material medium is also called the "dielectric." The so-called "dielectric constant" of such a substance is a measure of the susceptibility of the substance to undergo electrical interaction with, and suffer electric "polarization" by, an applied field. More exact definitions and discussions of these quantities are given in Chapter 5 on dielectric polarization phenomena, including the relation of G to molecular and electronic properties of solvent molecules such as a, and fl.

Forces on a Quadrupole in a Field



of Solvent


in Double-Layer

and Ion Fields

Many poly atomic molecules (e.g., H20, CO2) have charge distributions that cannot be represented by only a single dipole. In these cases, more complex interactions with a field arise from the quadrupolar as well as the dipolar nature of such molecules. Quadrupole interactions are important in solvation and in inhomogeneous fields in the double layers at electrochemical capacitors. Note that a quadrupole experiences an energy only with an inhomogeneous field, i.e., with afield gradient.

Here it is sufficient to indicate that in an electric field (e.g., as in the double layers of an electrochemical capacitor) molecules of the dielectric of the double-layer interphase suffer electric polarization of three kinds: 1. electronic polarization, which is determined by the solvent molecule's electronic polarizability, i.e., displacement of its electron distribution under the influence of the electric field in which it finds itself situated;


Chapter 4


of Double Layers and Ions


2. cooperative or individual orientation of its dipole moment or that of a

conformationally mobile polar functional group within its molecular structure in the field; and 3. "atomic polarization," where there is some modification of interatom bond distances in polar molecules in a strong field. These three effects are significant in the very high fields obtaining in double layers and also in the solvation shells of ions populating double layers.
4.10.3. Dipole Moments of Complex Molecules

example, dioxane may exist in a boat (1) or chair (II) conformation. Structure I has the higher dipole moment and will be preferred in the high fields near ions or in double layers at electrode surfaces; it may also be preferred due to intermolecular H bonding by water.

In complex molecules, such as various nonaqueous solvents, the net dipole moment must be evaluated by resolution of the individual bond dipole moments, taking into account the geometry of the molecule. Chloroform and methyl chloride are interesting examples (see structures below).





Electric polarization in dielectrics, which relates molecular-level electric polarization to bulk-phase polarization in liquid dielectrics, is dealt with separately in Chapter 5. The liquid solution interphase in double layers has properties somewhat intermediate between those of individual solvent molecules and interacting liquid like assemblies, but in quasi-2-dimensional structures as treated in Chapter 7.

Ji-7/19~ce :


1 A central aspect of the electrical behavior of a capacitor is its ability to store electrical energy as well as to be an active element in electrical circuits. However, unlike a battery, most or all of the energy is stored electrostatically rather than electrochemically. The process of energy storage is associated with buildup on and separation between electrical charges accumulated on two conducting plates spaced some distance apart that comprise the capacitor (or in older terminology, the "condenser"). In a double layer, one plate is the electrode surface and the other the inner region of the ionic solution. The capacitor device is composed of two such double layers, as explained earlier. As charges of different sign are added to the surfaces of one plate and the other, more and more work (free energy) is required to add progressively further charges to the respective plates owing to the repulsion between like charges or to the accumulating "self-energy" of the charge distributions on each plate. This energy can be calculated as follows, in a way analogous to that for evaluation of the self-energy of a charge (Eq. 4.21). Let a fraction A «1) of the ultimately stored charge, q, be placed on the plates so that the potential is Aq/C. Let an infinitesimal Jq be added; then the element of energy of charging, dG, is

The overall dipole moments are:


= flCH

+ 3flcCICOS l/J



= flCCl

+ 3flcH cos l/J


The sp3 geometry of each molecule is such that 3flcCl cos l/J = flCCl and 3flcH cos l/J = flCH' Hence the molecules should have the same overall dipole moment. This is not the case: flCHC13 = 1.1 D; flCH Cl = 1.85 D [ID (debye) = 10-18 esu]. The discrepancy could arise because thetetrahedral angle is not exactly maintained in chloroform due to repulsion of the CI atoms and/or because there is some depolarization among the C-Cl bonds, making them have lower bond moments than the single C-Cl bond in CHCI3; the latter effect is the main one. In flexible polyatomic molecules, such as conducting polymers, the overall dipole moment will depend on the conformation the molecule may adopt. For


Chapter 4


of Double Layers and Ions


dG=-· but dq

Aq dq C


= qds,

Hence the energy G of charging (a Gibbs energy) is G=

ponent S == -dG/dT, the temperature derivative of that energy. Other things remaining constant, the only quantity involving temperature dependence is the dielectric constant of the dielectric medium determining C in the relation C = Auo/d (see Eq. 4.14 and the text following that equation). Hence the entropy of charging is given (for e > 1) by S=-dG/dT=-~q2.

f dG = f A~ . qdA







1 1 = - q2/ C == - CV2 2 2


(4.37) Since deld'I' is normally a negative quantity for most if not all dielectric materials, it is seen that the entropy of charging a capacitor having e > 1 is negative. This is due to the electric polarization of the dielectric that is established in the field between the plates of the capacitor. It is analogous to the entropy of magnetic polarization in magnetized paramagnetic or ferromagnetic metals. Analogously also, if one could switch off the electric polarization instantaneously, then an adiabatic dielectric depolarization cooling effect would arise which might have practical technological value. However, in a double-layer electrochemical type of capacitor, such an effect would tend to be reversed owing to the internal joule heating effect associated with the discharging current required to transport accumulated ionic charges away from the two electrode interfaces into solution when the two double layers were discharged, i.e., when Vand q ~ O. In a double-layer capacitor, this entropy is associated mainly with solvent dipole orientation in the double layer and coupled ion accumulation at the capacitor plate interfaces. The distribution of charge on conducting capacitor plates is another matter of interest: the distribution tends to be uniform except for edge effects. The electronic charge (±) on the conducting plates is a delocalized charge of conductionband electrons: an excess or deficiency of surface charge density compared with the situation at two uncharged plates. At the surfaces of an electrochemical double-layer type of capacitor, the situation can be somewhat different because the charges in the solution, complementary to the plus or minus electronic charge densities, are composed of discretely charged ions of finite sizes; also, the distance of closest approach of

recalling that always C = q/V. C contains the dielectric constant, e (> 1) of the dielectric medium between the plates; e = 1 for vacuum, hence C(e) > C(e = 1). This implies that a greater energy is stored in a capacitor constructed with a dielectric having e > 1 than in one in a vacuum by the factor e. The extra energy over that in a vacuum when e > 1 and geometrical dimensions of the capacitor are the same is stored within the dielectric medium as an energy of polarization of the dielectric, as explained in Section 4.4.7 concerning polarization of solvent by an ionic charge, and especially in Chapter 5. It is interesting that the calculation of electrical energy stored in a charged capacitor can be conducted (as above) in a way closely similar to the derivation of the "self-energy" of the charge on a chemical ion, i, due to Born 1 (1920). This self-energy (a Gibbs electrostatic energy) is given (Eq. 4.23) by G, = (Zje)212eri (in a vacuum) (4.33)

G, = (zje)212eri.e

(in a dielectric medium)


where rj is the radius of the ion in vacuum, rje is its effective radius in the dielectric solvent medium, and Zje is its charge. Note that these G, quantities depend on half the square of the ionic charge, ue, like the result in q2/2C for energy in the case of capacitor charging. Then, by analogy between these two cases, it is seen that C is equivalent to rj for a spherical ion. That is, the capacity of a conducting sphere is equal to its radius, a principle in classical electrostatics. Another aspect of electrochemical and physicochemical interest is that since the stored energy of a capacitor is a Gibbs energy, it has an entropy com-


Chapter 4

anions to the charged conductor surfaces is usually believed to be smaller than that for the co-cations of the electrolyte because the latter retain their solvation shells when they are attracted to the negatively charged surface, while anions can approach nearer (see Fig. 6.4 in Chapter 6) with some loss of solvation. The fact that hydrated or solvated ions in the double layer have a finite size has formed the basis of some theories of the double layer by Stern2 and Esin and Markov' that are improvements on the early treatments by Gouy and by von Helmholtz (see Chapter 6). In the vicinity of individual ions in the compact part of the double layer there is a reciprocal effect on the electron charge distribution that tends to respond to the discreteness of charge of the ionic charge distribution through an effect associated with the so-called "image charge distribution,,,4-7 a classical induction effect at surfaces containing a mobile distribution of charges.

Chapter 5

Behavior of Dielectrics in Capacitors and Theories of Dielectric Polarization

I. M. Born, Zeit. Physik, 1, 45 (1920). 2. O. Stern, Zeit. fur Elektrochem .. 20, 508 (1924). 3. Y. Esin and B. Markov, Acta Physicochim. U.S.S.R., 10, 353 (1939); see also Zhur. Fiz Khim., 17,236 (1943). 4. R. G. Sachs and D. C. Dexter, f. Appl. Phys., 21, 1304 (1950). 5. 1. Bardeen, Phys. Rev., 58, 727 (1940). 6. P. H. Cutler and 1. C. Davis, Surface Sci .• 1. 194 (1964). 7. A. A. Kornyshev, A. I. Rubinshtein, and M. A. Vorotyntshev, Phys. Stat. Solid., B 84, 125 (1977).



I. C. 1. F. Bottcher, Theory of Electric Polarization, Elsevier, Amsterdam (1952). 2. F. W. Sears and M. W. Zemansky, University Physics, 4 ed., Chapters Addison-Wesley, Reading, Mass. (1970).

24 to 27,

3. L. Pearce Williams, Michael Faraday: A Biography, Chapter 7, pp. 283-319, on dielectrics and lines of force, Chapman and Hall, London (1965). 4. E. A. Moelwyn Hughes, Physical Chemistry, 1st ed., Chapter 2, p. 92 et seq., Cambridge University Press, Cambridge, U.K. (1947). 5. H. Frohlich, Theory of Dielectrics, Oxford University Press, Oxford (1949). 6. J. B. Hasted, Aqueous Dielectrics, Chapman and Hall, London (1973). 7. C. P. Smyth, Dielectric Behavior and Structure, McGraw Hill, New York (1938).

In the case of electrochemical capacitors of the double-layer type, the behavior of the dielectric of the capacitor has a special significance since it is the solvent of the electrolyte solution that constitutes locally the dielectric of the double layer and provides the solvation shells of the ions in that medium. As shown in more detail in Chapter 6, the double layer at an electrode/solution interface consists of one real, electronically conducting plate (metal, semiconductor, oxide, or carbon surface) and a second virtual plate that is the inner interfacial limit of a conducting electrolyte solution phase. The double-layer distribution of charges is established across this interphasial region, which is composed of a compact layer having dimensions of about 0.5 to 0.6 nm, corresponding to the diameters of the solvent molecules and ions that occupy it, and a wider region of thermally distributed ions over I to 100 nm, depending on ionic concentration (Chapter 6). It is because of this very small thickness of the compact molecular interphasiallayer that a relatively large specific capacitance of 20 to 50 J.lF cm-2 can arise. A practical double-layer capacitor device must, however, be constituted from two such interphasial double layers, each exhibiting its own capacitance behavior. The dielectric ofthe capacitor device resides in both of these interphasial regions in the case of a double-layer type of supercapacitor. Unlike the situation with regular capacitors, in which the bulk properties of the dielectric medium determine (among other geometrical factors) the


Chapter 5

Behavior of Dielectrics in Capacitors and Theories of Dielectric Polarization


capacitance of the device, in electrochemical capacitors it is the microscopic properties of the dielectric (ionic solution) at the molecular level that determine the specific double-layer capacitance, as will be shown in Chapter 7. The dielectric of the double-layer capacitance is constituted locally of a thin layer of solvent molecules within 0.5 to 0.6 nm of the electrode interface (see diagrams in Chapter 6) that are interacting with the electrode and are also substantially oriented by the field in the double layer at potentials significantly displaced from the potential of zero charge. Some of the solvent molecules that find themselves in the interphasial region of the double layer are also already orientationally polarized by the cations or anions of the electrolyte that populate the compact Helmholtz region of the double layer [see Fig. 6.1(a) in Chapter 6]. The situation of the solvent dielectric which forms the interphase of the double layer, where the charge separation corresponding to the double-layer capacitance is developed, is thus structurally quite complex at the molecular level. In order to provide a background for discussion and treatment of the dielectric behavior of the interphasial region of double layers in double-layer types of electrochemical capacitors, we first give in some detail but not exhaustively, an account of (1) the formal treatment of capacitance and (2) the treatments of dielectric polarization that led to a molecular theory of the dielectric constant of solvent media. In addition, we give (3) a brief account of how concepts of polarization of solvent dipole orientation have been applied to the compact Helmholtz layer. However, the latter topic finds more detailed treatment in Chapter 7 on theories of the double layer.

or (5.4) where K is the permittivity of the dielectric medium. It is the ratio of the C of Eq. (5.4) to that of Eq. (5.2) that gives the dielectric constant of the medium, a quantity that Faraday' referred to as the "specific inductive capacity." The dielectric constant, B, can thus be defined as the relative permittivity, (5.5) using the symbol x for permittivity, so that for a material medium C = AKld or the specific capacitance is CIA = Kocld (5.7)







Formally the capacitance, C, of a plane capacitor having plates of equal area in parallel configuration, separated by a distance, d, in a vacuum (dielectric constant = 1) is C = AI4nd or C (in the conventional units system) (5.1)

= AKoId (in the so-called "rationalized units")


where KO is the permittivity of free space (see later discussion), sometimes symbolized by BO' In the presence of a dielectric medium between the plates, C is increased (cf. Eq. 4.14) to C=ABI4nd (5.3)

farads cm-2. Formally, B should be referred to as the dielectric coefficient since B is usually dependent on field at high fields, especially in polar media, or may depend on frequency in time-variant fields. Note that from the definition of capacitance as charge divided by potential (q/V or /1ql/1 V), measured in practical em units of farads, and from the geometrical definition C = AKoId, we find AKoId in farads for a vacuum. That is, BO or KO, defined as the permittivity of free space, has the units of farads per meter or farads per centimeter, depending on the choice of units for A and d in terms of meters or centimeters. The introduction of KO or BOensures that C works out, for given values of A and d, in practical units of farads, i.e., coulombs per volt. It does not if the simple equation (5.1) is used; it must be written as C = AI(9 x 109)4nd, for A in square meters and d in meters. Attention to the units is required with conversion from esu to practical emu (a factor of 3 x 109) and the esu of potential to volts (a factor of 3 x 102) to obtain C in farads (an em unit). KO or eo is then found as 1I4n(9 x 109) F m-I (see Eq. 4.7). A capacitance of 1 F arises if ±1 coulomb, placed on the plates of a capacitor, causes a difference of 1 V to be established between the plates. 1 C = 1110 emu of charge and 1 V = 11300 esu of electric potential; also, 1 esu of charge = 11109 practical emu of charge, i.e., 1 C. Hence, from the above relations for C in terms of 1 F = 1 CIl V and AKoId, we find equivalently KO = 11(3 x 109) x (3 X 102) x 4n, = 8.84 x 10-14 F cm-I or 8.84 x 10-12 F m ", the conversion factors being as defined earlier. (See also Chapter 4 and the tabulation in Ref. 2.) For a vacuum, B = 1 or for gases at normal temperature and pressure (NTP), e is a little over 1. For nonpolar liquids, B is between 1 and about 3 and is ap-


Chapter 5

Behavior of Dielectrics

in Capacitors

and Theories of Dielectric Polarization


proximately equal to the square of the refractive index, arising from electronic and atomic polarization. For polar but non-H-bonded liquids, it has values around 6 to 40, while for H-bonded liquids, it can attain much higher values, e.g., 78 for water or 113 for HCN at ordinary temperatures. Some oxide materials can have relatively large e values of several hundred. This is important for fabrication of so-called "electrolytic" capacitors in which a thin oxide film is the dielectric. It is seen that C scales with s, other conditions being constant. This is related to the involvement of the dielectric constant in Coulomb's law, where the repulsive or attractive force between two charges, q, and q2, in a vacuum (Chapter 4),

chemical double-layer capacitors utilizing various solvents. The answer is the electric polarization behavior of the dielectric material in the field E corresponding to the voltage difference between the plates, .1 V, divided by the separation distance d: E = .1V/d.




is attenuated by the factor lie if the charges are in a dielectric medium:

The force is vectored along the line between the charges. The corresponding energies of repulsion or attraction are Gibbs energies

The term "polarization" can be understood as describing the extent to which polarity is induced in a given unit volume of dielectric medium by the field E. Microscopically, in a related way, polarity or extra polarity is introduced by E in individual molecules or in assemblies of them. In advanced theories of dielectric polarization and the properties of dielectrics, the essential requirement has been the development of various models at different levels of approximation and detail for relating molecular electrical properties to bulk electric polarization of liquid (or gaseous) media, i.e., for assemblies of interacting polar molecules. This becomes a statistical-mechanical problem. The quantitative relations between this polarization and the field E (and thence K or c) for various types of dielectric media are the essence of the theory of dielectrics and the behavior of solvent molecules in the double layer at capacitor electrodes. Two kinds of approach have been made: 1. The first involves treatments in which the dielectric and its polarization are examined in terms of bulk properties. 2. The second involves treatments in which such bulk properties are evaluated by modeling at the molecular or molecular-assembly level. This latter treatment involves principles of statistical mechanics in order to relate averaged molecular properties in assemblies to bulk-phase behavior at finite temperatures. Since the capacitance of a capacitor of given geometrical proportions depends directly on the K or c of its dielectric medium, the interpretation of the dielectric constant is of main interest in this field of capacitor behavior, especially the dielectric behavior of the double-layer interphase at electrodes in solution. In the case of ordinary hardware capacitors, their C depends only on the geometrical parameters and K or e of the medium separating the plates. With electrolytic capacitors, the situation is essentially the same except that the dielectric is an anodically formed oxide film of an appropriate metal (e.g., Ta, Zr, Ti, AI) that can be formed under controlled electrolytic conditions (anodization), Its thickness (to which C is inversely related) is determined by the magnitUde of the anodic-forming voltage and the time, t, for which it is applied. (An

or (S.11) i.e., they are attenuated by the factor lis if they are situated in a dielectric medium having e > 1. Similarly, capacitance of a given geometrical configuration of plate electrodes is enhanced by the factor e (> 1) over that for a vacuum (s = 1) and correspondingly more (free) energy can be stored for a given applied voltage, .1V, when s > 1, the energy being 112 C(.1 V)2 (Chapter 2). For two capacitors having identical geometries, the difference of stored free energy for e > 1 (real dielectric) and e = 1 (vacuum) is (S.12)

The fundamental question regarding dielectrics is then how and in what state is this extra energy stored in a capacitor having a dielectric of constant e > 1 in relation to that for a vacuum (s = 1), for a given voltage difference between the plates? This question is fundamental to the materials dependence of the capacitance of capacitors utilizing various dielectric media, including electro-


Chapter 5

Behavior of Dielectrics in capacitors and Theories of Dielectric Polarization


inverse "log" law in time usually applies or a "parabolic" law for some conditions; see Ref. 3.) The situation with double-layer capacitors is different since at each of the capacitor ± electrodes a double layer exists across which there is a drop in potential (see Fig. 6.3, Chapter 6) between the electrode plate and the electrolyte solution; the dielectric of this double molecular capacitor is composed of only one or two molecular layers of solvent, including solvated anions or cations in that solvent layer, and it is polarized by the field associated with the differences in potential between the metal and solution, especially that part which falls across the first 0.3 to 0.6 nm (Helmholtz layer) of the solution near the electrode surface. Interfacial dielectric properties of this very thin film have to be formulated (see Chapter 7) to account for the magnitude of the double-layer capacitance (16 to 50 f1F cm=') and its dependence on electrode potential, the sign of the charge of the electrode plate, and the nature of the solvent and the electrolyte solute. Some of the principal aspects of the molecular and ionic modeling of the double layer and its effective average dielectric properties are given in Chapters 6 and 7. In order to provide the background for the discussion in those Chapters, we return now to the more general theory of bulk dielectrics in order to explain the concepts of polarization and polarizability more completely, and their relation to experimentally measurable dielectric constants and molar polarizabilities. These are fundamental matters that lie at the bases of capacitance behavior in electrostatics, and in the phenomenology of double-layer capacitance and electrical behavior of supercapacitors.

known, I despite his depth of scientific perception and his inventiveness Faraday never developed his concepts and interpretations of physical phenomena in mathematical terms. The electrostatic behavior of two identical configurations of capacitor plates is considered here. One capacitor [Fig. 5.l(a)] has a vacuum dielectric the other [Fig. 5.l(b)] a real dielectric medium having I: > 1. An external field Eo is applied and is operative between the plates generating or arising from plus and minus charge densities on the plates, equally and oppositely. For a given volt~g.e difference, LlV, between the plates or a corresponding field LlVld, charge densities 0"1 = ±qlIA or 0"2 = ±q21A appear, respectively, on the plates of the two capacitors. 0" I and 0"2 differ on account of the difference in I: for the two cases' 81 1; 82 > 1. This difference corresponds to an induced charge density which appears at the plates, changing the original 0" value [Fig. 5.1(b)]. When the dielectric medium is present [Fig. 5.I(b)], the field between the plates is reduced from the vacuum value, Eo, to a diminished value E.



Field Eo +
+ + + + +

Field Eo








Apart from the invention and use of the Leyden jar (Chapter 1), Faraday was the first to consider the state of dielectrics in charged capacitors. In fact it was in correspondence with the polymath, Whewell of Cambridge, that the term "dielectric" was suggested (December 1836) and it has remained in use ever since. Faraday! initiated ideas of "polarization" due to electric fields and recognized the induction of electric charges, with polarization being induced by one polarized or polar particle into or onto another. He also recognized concepts such as specific inductive capacity (1:), lines of inductive force, and the strain introduced in an electrified dielectric medium by an imposed field. Thus Faraday recognized some of the properties of dielectrics and the nature of electric inductive polarization before the middle of the nineteenth century. Of course, it was many years later before these ideas became more exactly and mathematically formulated by workers such as Mosotti, Clausius, Coulomb, Frohlich, Debye, and Kirkwood in terms of molecular polar models. Unfortunately, as is well



1 1

1 -'-..'-.1 1 Field E«Eo)

External field

--.. --


1""-- /'

1 I <, 1





F.IGURE 5.1. Fields E between charged capacitor plates: (a) in vacuum; (b) in the presence of a dlelectrically polarizable medium. Original charge density fa; induced charge density ±p.


Chapter 5

Behavior of Dielectrics in Capacitors and Theories of Dielectric Polarization


In terms of the dielectric polarization induced by the external applied field t1 V/d, = E, the dielectric medium in case (b) is regarded as composed of a density of induced dipole moments oriented in the direction of E. This results [see Fig. 5.2(a)] in opposing the respective charge densities on the two plates of the capacitor by the presence of the plus and minus ends of the bulk polarization set up by the field in the dielectric between the plates. This polarization can be expressed formally as an induced dipole moment per unit volume of dielectric, designated P [Fig. 5.2(b)]. Between the plates, the plus and minus ends of the induced dipole elements cancel out; only at the interfaces of the dielectric at the plates do the plus and minus ends of the induced polarity lead to surface charge densities induced at the boundaries of the dielectric. These determine the capacitance of the capacitor, increasing it over the induction of charge density by the field t1 V/d in a vacuum (8 = 1).

The origins of the polarization thus induced in the dielectric are fourfold: 1. The first is a redistribution of electronic charge density (Fig. 5.3), mainly of valence electrons, within the individual atoms or molecules of the dielectric material; it is characterized by the so-called "electronic polarizability," denoted by a, of the molecules. This a depends on the number of polarizable electrons and thus on the volume of the molecule (a has the units of crrr' mol ", on the order of 10-24) and on the type(s) of orbitals involved in the molecules, e.g., (J or tt. a for n orbitals is greater than a for (J orbitals; for example, with benzonitrile C6Hs . C=N, the polarizability is anisotropic, along rather than across the molecule. 2. There is a slight change in the atom positions or bond lengths of the molecule, and hence its polarity. This is called "atom polarization." 3. Third, for intrinsically polar molecules such as H20, HCI, HF, and CH3CN, a larger degree of polarization per unit volume is set up within the dielectric due to an orientational response (Fig. 5.4) of the already existing dipoles of the molecules of the dielectric medium. This is a statistical effect, evaluated in important work by Debye and by Langevin," and is the main source of dielectric polarization in fluids of intrinsically polar molecules. Since for ordinary fields of e.g., 10-100 V ern"! there is a small but significant net average orientation of dipoles, this has an influence on the induced electronic polarization in the molecules of the dielectric, depending in an average way on their momentary orientations in the field disturbed by thermal fluctuations in the bulk fluid. This effect does not of course arise with atomic fluids or 3-dimensionally symmetric molecules (e.g., CH4, CCI4) except to a small extent through dipole induction by the experimental field. For intrinsically polar molecules, polarization of the dipole orientation is normally substantially larger than the polarizations of types (1) and (2).

+ +

+ +


+ + +
1 cm


E o~___

+ --E--FIGURE 5.2. Schematic diagram of indue lion of polarity in a dielectric medium by an externally applied field. (a) Induced charges or orientation of polar molecules by an external field Eo in capacitor. (b) Resulting polarization vector, P, corresponding to induction of polarity, -, +, on the faces of a centimeter cube.


I'-i. ---'-0_+-

fLi = a E
FIGURE 5.3. Redistribution field (induced polarity). of electronic charge density in a polarizable molecule in an electric

Average momen'''j.I.


Behavior of Dielectrics in Capacitors and Theories of Dielectric Polarization


\ ,.."

,. --"

.. ---.
Field E








-- (b)

(Normal polarization experiment)

IJ.E» kT
( Dielectric saturalion •. Q. at ions or electrode interfaces)

FIGURE 5.4. (a) Orientational polarization of intrinsically polar molecules in a small electric field. (Effect is additional to the induced electronic polarization in Fig. 5.3.) (b) Approach to full orientation of solvent dipoles in the very strong field (JtFlkT» 1) of the double layer (dielectric saturation condition).

4. Finally, when there is substantial intermolecular electrostatic dipoledipole interaction between molecules of a dielectric material (e.g., CH3N02, (CH3hCO), and especially when the molecules are H-bonded as in liquid HF, H20, HCOOH, CH3·COOH, RCO·NH2, esters, propylene carbonate, etc., the process of orientational polarization becomes a cooperative one between molecules in small assemblies, giving rise to much larger polarizations per unit volume for a given field. These are the media that exhibit high dielectric constants in the range 50 to 113 at ordinary temperatures, e.g., CH3·OH, H20, liquid HCN, liquid HF, and HCOOH. From the point of view of choice of dielectric solvents for double-layer capacitors (see Chapters 8 and 13), the use of H-bonded solvents, though giving high e values, has the disadvantage of relatively low decomposition voltages in the range 1.23-1.5 V and rather high melting points. Higher operating voltages, per cell, for double-layer capacitors therefore require so-called "aprotic" nonaqueous solvents (Chapter 8) for their electrolytes (e.g., tetraalkylammonium salts) since such solvents, which do not have electrolytically active (i.e., electrolytically depositable) H or other atoms, have substantially higher decomposition voltages, e.g., 3.4 to 4.0 V. Useful practical examples are acetonitrile, benzonitrile, N,N-dimethylformamide, propylene carbonate, dimethoxyethane, and the tetrahydrofurans. Such solvents are also preferred media for construction of nonaqueous lithium battery systems operating at 33.5 V per cell. Here the requirement is of course that the very base metal, Li not discharge H2 from any "active" H atoms in functional groups of the solvent

molecule. "Active" H atoms in a solvent are usually those that are acidic and participate in intermolecular hydrogen bonding. However, as is shown in Table 8.1, Chapter 8, a number of aprotic solvents do have quite high dielectric constants (20-30), but since they are not H-bonded do not exhibit the strong cooperative polarization behavior that is found for such associated liquids as H20, CH30H, CH3·COOH, etc. Hence they have lower dielectric constants. In terms of the diagrams in Figs. 5.2 and 5.4, a vector quantity P is defined as the dielectric polarization, which is equivalent to a dipole moment per unit volume [Fig. 5.2(b)]. P is thus equivalent to a charge p (per square centimeter) induced at the plates of the capacitor multiplied by a unit length of 1 em, i.e., a dipole moment per unit (1 crrr') volume. p is sometimes written as a.; the induced a diminishing the original a value. In the absence of a polarizable medium (s = 1), the field due to a surface charge density a per square centimeter is [Fig. 5.1(a)] Eo = 4na (Gauss's relation), which is also referred to as the dielectric displacement, D. Owing to the presence of the dielectric, the field Eo, which in the absence of a dielectric is due to a, becoming reduced (see Coulomb's law, E = -qle?) by 4np [which is related to the induced charge density p, Fig. 5.l(b)], so that E=4n(a-p) or E=Eo-4np (5.13)

is the field inside the polarized dielectric medium [Fig. 5.1 (b)]. Then, introducing the dielectric displacement, D, and noting that DIE D=E+4np so that e = 1 + 4npiE or identically e = 1 + 4n PIE

= e,



In a vacuum, as before, note that e = 1 and D == E. Also, PIE is seen to be a polarization induced by field E in the medium so that PIE is a polarizability. In a material medium, D = eE. P is made up of an induced polarization due to the redistribution or distortion of charged species in the molecules of the dielectric and a contribution associated with the orientation of dipoles of the dielectric material in a preferred average direction along the field or having a component of orientation along it (Chapter 4). This is called the "orientation polarization" and it is dependent on the temperature. It is an important aspect of the behavior of solvent molecules in the interphase of double layers, and hence in electrochemical double-layer capacitors utilizing various solvents.


Chapter 5

Behavior of Dielectrics

in capacttors and Theories of Dielectric Polarization







The induced dipole moment, flj, is proportional to the field for low or moderate fields (i.e., flj = aE where a is the distortional polarizability), mainly owing to redistribution of electron density in the molecules. It has the dimensions of volume and magnitude on the order of 10-24, i.e., - (10-8)3 or angstroms cubed (nrrr' x 10-3). For a dilute gas, the effective polarizing field is the external field, Eo· If M is its molecular weight and p its density, MIp is the molar volume (here p is the density) or MINAP is the molecular volume (NA = Avogadro's or Loschmidt's number) and the induced polarization, P, is p.=

Then instead of P, (Eq. 5.17), a molar polarization, PM, is obtained as

(5.22) and this refers still to induced dipoles through flj. This last equation was derived by Mosottf and also independently (see Ref. 9, p. 199) by Clausius. It is now a classical relation in dielectric theory and is referred to as the Clausius-Mosotti equation. Actually, since a is a polarizability, PM also has the units of polarizability, i.e., of molar volume.

molecules dipole moment x unit volume molecule





(5.17) and the induced moment is


u, = aEo
Then from Eq. (5.18) defining flj, and using Eq. (5.15)
(8 - 1) Mlp


= 4nNA






For a dielectric medium of permanently polar molecules such as CH3CN and CH3N020 the polarization can attain much larger values beyond those due to charge redistribution (induction) on account of orientation of the permanent (intrinsically polarized) dipoles of the medium by the applied field. Permanent polarity in molecules arises in heteronuclear (asymmetric) molecules having two or more atoms of differing electronegativity. The field tends to cause orientation of permanent dipoles (fl) to an average extent that is determined by their energy flE in comparison with the average Boltzmann thermal energy, kT, i.e., by the quotient flElkT. A statistical-mechanical averaging calculation carried out by Debye" allowed this average dipole orientation polarization to be evaluated III terms of the average moment Ii along the field direction where
~ = coth (fl£)_

If the dielectric is not a gas but a condensed phase in which molecular interactions are considerable and short range, the influence of the molecular surroundings on a given dielectric molecule has to be taken into account: a given dipole in a hypothetical molecular-sized cavity in the medium induces charge density on the walls of that cavity which then generates an opposing (induced) local field, the reaction field. Then the effective field E, becomes related to E and P in a relatively complex way, leading to (5.20) and the induced moment becomes

J1 At low fields, flElkT«


kT flE


1 so that, by expanding the exps in coth, (5.24)

results. Then fl2/3kT is seen to be an average orientation polarizability for low fields. (Experimentally, fields of less than 100 V cm-I are used in 8 measuretnents.) This orientation polarization is now added to the induced polarization, giving a total molar polarization


Chapter 5

Behavior of Dielectrics

in CapaCitors and Theories of Dielectric Polarization



lower boiling point (19°C) of HF compared with that of H20 (lOO°C) at 1 atm pressure. . D~tail.ed treatments of cooperative polarization in strongly polar or associated liquids are complex and are largely outside the scope of this chapter. However, for completeness and not to leave this topic in "midair." w e gIVe th e . .I . , essentia eq~atIOns for such systems as follows. As stated earlier, various levels of complexity of the modeling, and hence the reality thereof, are involved. The. princi~al c~ncept utilized in treatment of the dielectric constant of strongly mt~ractmg dipolar systems is the introduction of a correlation factor g, representmg the cooperative orientation. The relation of e to g and 11 is ' (e - 1)(2e + 1) _ NA

a relation derived by Debye with the aid of the Langevin function for "jilll above (Eq.S.23).







In the case of dielectric liquids composed of strongly interacting dipoles (e.g., CH3N02, H20, CH30H, amides, and amino acids in zwitterion form), high dielectric constants are manifested. Water has an e value of 78 at 298 K and HCN 113 at that temperature; these are hydrogen-bonded or socalled "associated" liquids where the orientation of a given molecular dipole drags neighboring dipoles in the same direction, especially with hydrogenbonded solvents. Such solvents also have relatively high viscosities and long dielectric relaxation times, on the order of 1O-11s, in comparison with 10-12 - S X 10-13 s for weakly interacting solvent molecules. Similar effects occur with other strongly dipolar but not hydrogen-bonded solvents, but the resulting e values are rather lower, e.g., 3S.9 for CH3N020 and 38.0 for CH3CN and N,N-dimethylformamide. This cooperative orientation effect enhances the polarization P for a given field, giving rise to large e values. Some solid ceramic materials also have large e values, but for different reasons. Several attempts have been made, especially by Kirkwood and co-workers 7,8 to develop a theory for high dielectric constant solvents based on a treatment of cooperative dipole orientation polarization. The effect is crudely analogous to coupled orientation of iron filings when a magnet is placed below them when they are spread on a sheet of thin paper. However, such a model lacks the important factor of thermal disorientation, which applies to molecular interactions and orientations in any liquid at finite temperature [see the Debye treatment utilizing the Langevin function (Eq. S.23) referred to earlier]. The problem is again to express the polarization resulting from the applied field as a function of that field, but taking into account the cooperative nature of interactive orientations of dipoles. In the case of strongly hydrogen-bonded fluid" such as water or hydrogen fluoride, the )O-H- - - O( or F-H---F- hydrogen bonds remain strongly interactive for displacements of the hydrogen bond from linearity of up to ca. 3So. In the case of water, the interactions are 3-dimensional (fourfold) or, for HF, they are mainly quasi-linear (twofold); hence the substantially


- 3kT gil


given by Kirkwood? g = 1 when there is no more correlation between molecular dipole orientations than can be accounted for by means of a continuum treatment. . An approximate expression for Kirkwood's correlation factor can be denved when only nearest-neighbor interactions are considered; then

= 1 + Z < cos 8ij>


where ~ij depends only on the coupled orientations of the each WIth z nearest neighbors. Another relation is the Kirkwood-Frohlich equation:


r molecules,


where .e~ is a dielectric constant of a supposed continuum in which the permanent .dlpoles CIld) are embedded; Vm is the molar volume, M/p; and p denotes density, The case of water as a dielectric has, of course, special significance for many electrochemical systems, including the double layer at carbon and metaVwater electrode interfaces. For water it is found experimentally that the Kirk,,:oo~ g factor decreases with increasing temperature from a value of 2.7S at O°C ~lll liquid wat~r) to 2.49 at 8~0C. ~his follows the general properties of water, uch as VISCOSItynd X-ray diffraction structure, which both decrease with tema perature. Theoretical values were calculated in a paper by Kirkwood and Oster. 8 . Assuming free rotation about the H bond in water (not a realistic conditIon!), the previous equation (5.27) leads to
g = 1 + z cos28/2



Chapter 5

Behavior of Dielectrics in Capacitors and Theories of Dielectric Polarization


where z is the average number of neighbors in the first coordination shell and () is the HOH bond angle in water, about 105°. The value of z can be derived from X-ray diffraction data. Other general and more detailed aspects of dielectric polarization are treated in Refs. 9, 10, and 11, the monograph volumes by Frohlich, Smyth, and Bottcher. The correlation factor will also be important in the orientation polarization and its contribution to the local dielectric constant of the interphasial double layer where solvent dipole orientation occurs in response to that part of the metal-solution potential difference at an electrode/solution interphase across the compact Helmholtz region. The interaction energy factor in the treatment of interphasial solvent dipole orientation is of major importance in defining the approach to dielectric saturation in the double layer at electrodes as treated in a paper by Bockris, Devanathan, and Miillerl2 and in several recent papers by Marshall and Conwayy,14 In particular, some minimum strength of dipolar interaction is required to avoid a "catastrophe'v+'" in the values calculated for the contribution of dipole orientation to the double-layer capacitance arising from the reciprocal dependence of the orientation surface-potential contribution'r on the electrode surface charge density at a polarized electrode. This difficulty arises in the two-state orientation model (Ising-type model in magnetic polarization) first considered for the double layer. 12In more sophisticated three-state or cluster models for interphasial water polarization, these difficulties are avoided (see Chapter 7).

Using the treatment by Boothl ' (see Chapter 7), Conway '", Bockris and . " A mmar 17 , an d Gr ah arne 18 rna de the first calculations of the local dielectric coefficient of water in the diffuse and compact parts of the double layer at electrodes in aqueous media. Much more is now known about the state of the water dielectric in the double layer, e.g., see Chapter 7 and Ref. 19. The calculations indicate that the I: for water is lowered appreciably from its normal bulk (zerofield) value of78 at 298 K in the diffuse layer and more so in the compact Helmholtz layer, where its value can descend to about 10-20, depending on the electrode/solution potential difference. More recent estimates'? indicate a minimum value of around 6. The various treatments of high-field effects on the dielectric constant of water, including experimental attempts to detect a lowering at high fields (but much less than those in double layers), are summarized by Smyth in Ref. 9, Chapter 3, p. 88. The types of calculation referred to here are really based on bulk models of the water dielectric. These cannot, however, be expected to reliably represent the behavior of a lattice of water molecules only two or three molecular diameters wide at the electrode interface, not to mention the local solvational influence of the ions that also occupy an appreciable volume fraction of the compact layer comprising this thin film of interphasial solvent molecules. In fact, depending on the electrolyte and the electrode/solution potential difference, some 30-40% of the interphasial water molecules can be electrostatically influenced by the high radial fields of neighboring cations or anions.i" Under such conditions, the bulk dielectric saturation theories of high-field polarization must be




Because of the molecular dimensions of the double-layer interphase at electrode or electrode-particle surfaces (e.g., in double-layer capacitors) the bulk theory of dielectrics will not be expected to apply at all accurately. However, the principles of the relation between polarization and the field that causes it will still be pertinent, but in a molecular-level analysis (see Chapter 7). The solvent molecules in the double layer at electrochemically polarized electrodes are in a different state from those in the dielectric medium of ordinary capacitors. In the double layer at electrodes polarized to 1-2 V from a potential of zero charge, a very high field, on the order of 107 V em -I, exists across the interphase at the electrode surface. Under such conditions, the orientation polarization component [Fig. 5.4(a)] of the dielectric coefficient becomes saturated [Fig. 5.4(b)], i.e., this orientation polarization can no longer respond to and increase with the applied field E so the dielectric constant becomes lowered to the smaller value corresponding to the electronic and atomic polarizations (I: = 2-6) (Fig. 5.5).

w (3

Onset of dielectric saturation

o o o ~ ~

u: u,

---~;;st~ni.n2- - - - - - - - - - - - - - - - - - - - - - - - - -o
5 LOG [FIELDN ern"] 6

Optical dielectric

FIGURE 5.5. Variation of dielectric coefficient ofa solvent (e.g., water) (Refs. 15, 17) in an increasing electrostatic field (log scale), toward the dielectric saturation limit (schematic).


Chapter 5

replaced by some more microscopic molecular treatment of the polarization of solvent in the double layer, especially the orientational component; see, for example, Refs. 19,21, and 22. This problem has attracted a lot of attention since ca. 1960 and several useful molecular-level treatments have appeared (see Chapter 7). These are especially relevant to the behavior of double-layer electrochemical capacitors where large-area, thin-layer solvent/particle interfacial geometries are involved in pores, with minimum volumes of free bulk solution electrolyte being present. The treatments at the molecular level involve models of solvent orientational polarization of the following kinds: (1) individual molecular orientations, with pairwise interaction, in the double-layer field; (2) clusters of interacting solvent dipoles and their orientation in the double-layer field; and (3) restructuring in an H-bonded lattice of solvent molecules (water or alcohols) in the interphase at the electrode.19•21,22 These models are compared in more detail in Chapter 7, especially with regard to their capability of accounting for the principal features of double-layer capacitance behavior, e.g., the dependence of the capacitance on electrode surface charge and potential, and on solvent (see Chapters 6 and 8).

Chapter 6

The Double Layer at Capacitor Electrode Interfaces: Its Structure and Capacitance

I. L. Pearce Williams, Michael Faraday: A Biography, Chapman and Hall, London (1965). 2. B. E. Conway, Electrochernical Data, Elsevier, Amsterdam (1952). 3. N. F. Mott and B. Cabrera, Rept. Prog. Physics, 12, 163 (1949). 4. P. Debye, Phys. Zeit., 13,97 (1912); see also P. Langevin, 1. Phys., 4 (4), 678 (1905). 5. O. F. Mosotti, Mern. di Mathern. e Fisica, Modena, 24 (II) 49 (1850). 6. R. Clausius, Die mechanische Wiirme theorie fl, p. 62, Braunschweig (1879). 7. 1. G. Kirkwood, 1. Chern. Phys .. 7. 911 (1936). 8. J. G. Kirkwood and G. Oster, 1. Chern. Phys., 11, 175 (1943). 9.


c. P. Smyth,

Dielectric Behavior and Structure, McGraw-Hill,

New York (1938).

10. H. Frohlich, Theory of Dielectrics, Oxford University Press, Oxford (1949). 11. C. J. F. Bottcher, The Theory of Electric Polarization. Elsevier, Amsterdam (1952). 12. J. O'M. Bockris, M. A. V. Devanathan, (1963). and K. Miiller, Proc. Roy. Soc., Lond., A274, 55

13. S. Marshall and B. E. Conway, 1. Electroanal. Chern., 337, 1 (1992). 14. S. Marshall and B. E. Conway, 1. Electroanal. Chem., 337, 19 (1992). 15. F. Booth, 1. Chern. Phys., 19, 391, 327, 1451 (1951). 16. B. E. Conway, Ph.D. thesis, University of London (1949). 17. B. E. Conway, J. O'M. Bockris, and 1. A. Ammar, Trans. Faraday Soc., 47, 756 (1951). 18. D. C. Grahame, 1. Chern. Phys., 18, 903 (1950). 19. Symposium on the Structure of the Electrified Interface, D. J. Henderson and O. R. Melroy, eds., Electrochirn. Acta, 36, 1659-1894 (1991). 20. B. E. Conway, 1. Electroanal. Chem., 123, 81 (1981). 21. G. M. Torrie and G. N. Patey, Electrochim. Acta, 36, 1677 (1991). 22. A. M. Brodsky, M. Watanabe, and W. P. Reinhardt, Electrochirn. Acta, 36, 1695 (1991).

As indicated in Chapter 1, electrochemical capacitors are principally based on two types of capacitative behavior: (1) one associated with the so-called double layer at electrode interfaces and (2) another associated with the pseudocapacitance that is developed in certain kinds of electrode processes where the extents of charge passed (q) (up to some limit) are some function of potential (V) so that a derivative dq/dV arises that is electrically equivalent to a capacitance. This is commonly termed a "pseudocapacitance," C¢, (Chapter 10) and is also experimentally measurable as a capacitance. As will be described further in Chapter 14, the principle of the double-layer electrochemical capacitor is the use of the large capacitance developed at high specific area carbon powder or porous carbon materials, on the order of 1000 to 2000 m2 g-I. With a nominal specific capacitance of, say, 25 flF cm", an overall capacitance of 1000 X 104 (cm2) X 25 flF cm-2 (i.e., 250 F g-I), is theoretically realizable. At 1 V operating potential, such a capacitance can theoretically store 250 J g-1 or 250 kJ kg". Practically, however, about 20% or less of these figures can be realized. Comparable energies per cubic centimeter can be stored with Ru02 metal oxide-type capacitors (see Chapter 11). It is essential to explain at the outset that a practical electrochemical capacitor, utilizing the double-layer capacitance of an electrode/solution interface, rnusr be constructed with two such interfaces (as emphasized in Chapter 5), Which are worked against each other (i.e., one is charged positively and the other


Chapter 6

The Double Layer at Capacitor Electrode Interlaces a Electrolyte


negatively) with respect to the electrolyte solution, with a separator (as in a battery cell) usually being arranged between the two electrodes. The two-electrode, two-interface system in a single capacitor cell is illustrated in Fig. 6.1(a). While the capacitative behavior of individual electrode interfaces can be studied experimentally, the electrical circuitry must always include a second (counter) electrode and also preferably a separate reference electrode with respect to which the potential measurements can be independently scaled and controlled. Similarly, a double-layer capacitor device must always involve (at least) two electrodes. This is an unavoidable requirement of all electrode interface measurements and all constructions of electrochemical devices, e.g., capacitors and battery cells. If the electrochemical behavior of each of the pair of electrodes is to be examined, then a third, reference, electrode is required. In the charged condition there are two interphasial drops in potential across the capacitor cell, one across each double layer [Fig. 6.1(a)]. Upon discharge, there is also a current-dependent, ohmic IR potential drop within the solution, and the opposite on recharge [Fig. 6. 1(b)]. Since one of the main technological directions of electrochemical capacitor development is in the field of the double-layer type of capacitor, this Chapter is devoted to a somewhat detailed description of the structure of the double layer and the origin of the capacitance associated with it. A detailed understanding of the properties and structures of double layers is essential for a full understanding of the operation of double-layer capacitors and the capacitance values that are achievable per square centimeter, or per cubic centimeter of active porous material in various systems. This will also require the mathematical representation of the properties of the double layer related to the nature of the electrolyte and solvent on the solution side of the electrode-material/solution interface and to the properties of the electrode itself. The theoretical treatment and modeling of the double layer are discussed in Chapter 7. Determinations of the double-layer capacitance and its potential dependence provide detailed information for examining the behavior and structure of electrified interfaces. This is mainly because the dependence of the charge, q, held on each side of the electrode interface (electron density on the metal and ion density on the solution side) on the measured electrode potential, E (which is equivalent to the Fermi level of electron energy states in the metal) is directly determinable as a differential quantity, Cdh dq/dE; such types of quantities always give more resolved information than corresponding integral relations, e.g., total charge q plotted vs. the electrode potential E at which it has been accumulated, or q divided by E at a given electrode potential. That is why cyclic voltammetry and ac impedance measurements are especially valuable and preferred in studies of the double layer and the resulting electrochemical capacitor devices; they both give differential information. A typical relation between

- z+ +-+-~+--+- ~+- +-+ - ~++ +-+ -.%+- +-+ -~++ +-+ - ~ + -+ - +- ~ + -+ - +- ~ + -+ - +Porous

-+- -+-+-++-+-+-+-+-+----



-+- Porous


Configuration of an electrochemical capacitor requiring two electrodes and thus two double-layers



---- ':,---

+ + + + + + + +


Potential profile across an electrochemical capacitor on discharge

FIGURE 6.1. Diagrams of electric potential profiles in an electrochemical capacitor comprising a double layer at each of two electrodes: (a) charged capacitor at open circuit, (b) capacitor passing current on discharge with IR drops.



The Double Layer at Capacitor Electrode Interfaces


f( y)



"-s' (19+)
+ + +

+ (11-)

+ + + +


-+ ++

\ \
\ \ \

<7".. '







FIGURE 6.2. The relation between arbitrary integral signal curve x,y, (e.g., for a double-layer charge as a function of potential) and its differential coefficient, dyldx (e.g., for differential capacitance).

FIGURE 6.3. Models of the double layer: (a) Helmholtz model, (b) Gouy point-charge model (specific charges a per unit area as indicated for anions and cations as an example), (c) Stem model for finite ion size with thermal distribution, combining Helmholtz and Gouy models.

an arbitrary integral curve and its differential, illustrating this point, is shown schematically in Fig. 6.2. It is appropriate first to describe the structure of the double layer, i.e., of the interphasial region of the metal/solution boundary. The term interphasial region 1 is deliberately chosen over the word "interfacial" (see Chapter 3, section 3.8) since the boundary region is really a 3-dimensional one, albeit very thin, some 0.3 to 0.5 nm in thickness, rather than the 2-dimensional one implied if the term "interfacial" were used. This means that the concepts of 3-dimensional phases can be used to describe the double layer, such as surface concentration or surface excesses, e.g., of ions of the electrolyte and solvent molecules. 1,2

6.2. MODELS AND STRUCTURES OF THE DOUBLE LAYER It is logical to describe these models in the order in which they were historically proposed in the literature, which is also the order of their progressive approach toward a correct description of the structure of the interphase at electrode surfaces. The concept of a double layer corresponds to a model consisting of two array layers of opposite charges, separated by a small distance having atomic dimensions, and facing each other, as on the plates of a two-plate hardware capacitor. This model was adopted by von Helmholtz3 to describe his perception of the distribution of opposite charges, quasi-2-dimensionally, first at the interface of colloidal particles. It is illustrated in Fig. 6.3(a), which shows its compact structure, and is referred to as the Helmholtz double-layer modelr':"

In the original model for colloid interfaces, the charges on the surface side of the double layer arise either from acid-base ionization, as with proteins or polyelectrolytes, or on account of the adsorption of ions, as at lyophobic colloids. On the solution side of the double layer, counterions of opposite sign of charge accumulate to balance the charge on the colloid, forming a double-layer array of positive and negative charges. The Helmholtz model was later adapted to the case of electrode interfaces where, on the metal side, a controllable surface density of excess negative or positive charge can arise that corresponds to an excess or deficiency of electron charges of the delocalized electron plasma of the metal. Owing to the high free electron (e) density in the metal (approximately I e per atom), any net charge density of electrons at the surface is strongly screened, so the gradient of electron density at a charged metal interface is highly localized over a distance of only 0.05 to 0.2 nm, the so-called Thomas-Fermi screening distance (Fig. 6.4). Because the wave function amplitudes of the conduction-band electrons retain significant but diminishing magnitudes outside the formal electrode surface plane, there is significant spillover of electron density into the double layer on the solution side of the interface+? and the effect is potential dependent. In the case of p- or n-semiconductors, the charge-carrier (hole or electron) densities are, however, very much smaller than in metals by a factor varying from about 10-4 to 10-15• As a consequence, there is a distribution of the charge carriers away from the interface (Fig. 6.5) but extending into the bulk of the semiconductor over a relatively large distance that is inversely related to the charge-carrier density (cf. the situation for ion distribution in an electrolyte in the solution side of the interface, Section 6.3). This distribution of charge carriers within semiconductors, near their interfaces, follows mathematically exactly


Chapter 6

The Double Layer at Capacitor Electrode Interfaces


Approx. 1e per atom in bulk


Thomas - Fermi screening distance



...... ....


Free electron density distribution


SOLUTION Electron overspill

Locus of atomic nuclei


.... -----

to zero surface plane

L_ Nominal

FIGURE 6.4. Electron overspill profiles at an electrode surface illustrating Fermi-Thomas screening length in electron space charge over narrow region at a metal surface (schematic).





n space - charge / region 0



FIGURE 6.5. A space charge distribution of charge carriers within an n-type semiconductor its surface that is like ion distribution in the diffuse layer in solution.

the same form as for ion distribution in the diffuse layer on the solution side (Gouy-Chapman theory for dilute ionic solutions). Some time after von Helmholtz's model was proposed, it became realized that ions on the solution side of the double layer would not remain static in a compact array as in Fig. 6.3(a) but would be subject to the effects of thermal fluctuation? according to the Boltzmann principle.f This latter effect would depend on the extent to which the electrostatic energy U; (together with any chemisorption energy Vc) of the ions' interactions with the charged metal surface exceeded, or were exceeded by, the average thermal energy, kT, at temperature, T, K, i.e., the ratio (U, + Vc)/kT. Gouy ' introduced this thermal fluctuation factor into a modified representation of the double layer in which the counterions conjugate to the metal surface's electron charge were envisaged as a 3-dimensional diffusely distributed population of cations and anions [Fig. 6.3(b)] of the electrolyte having a net charge density equal and opposite to the virtually 2-dimensional electron excess or deficit charge on the metal surface. In this model, the ions were assumed to be point charges. Historically, this was an important restriction since it led to a failure of Gouy' s model on account of (I) an incorrect potential profile and local field near the electrode surface and (2) consequently a too-large capacitance being predicted, that quantity being defined as the rate of change of net ionic charge on the solution side with the change of metal-solution potential difference across the interphase. The interphasial capacitance associated with this model is commonly referred to as the "diffuse" double-layer capacitance. A full mathematical treatment of the Gouy diffuse-layer model was given in some detail by Chapman in 19139 (see Section 6.3), based on the combined application of Boltzmann's energy distribution equation'' and Poisson's equation'" for the relation between ionic space charge density in the interphasial region to the second derivative of electric potential, 'fI, with respect to distance from the electrode surface. It is interesting to note that the mathematics and principles used by Chapman anticipated the approach taken by Debye and Hiickel in 192311 in determining ion distribution in three dimensions around a given ion in their treatment of activity coefficients and conductance of electrolytes. Later it was used by Onsager.I? in an improved treatment of the conductivity of electrolytes. In both Chapman's and Debye and Huckel' s treatments of ionic charge distribution, the key equation resulting from the combination of Boltzmann's energy distribution function and Poisson's electrostatic equation'? has been referred to as the "Poisson-Boltzmann" equation. It is also utilized in the treatment of band profiles and space charge effects in semiconductors.P The serious problem with the Gouy-Chapman treatment, overestimation of the double-layer capacitance, was overcome by Stern in 192414 in the next


Chapter 6

The Double Layer at Capacitor Electrode Interfaces


stage of development of the theory of double layers. In his model and calculations it was recognized that the inner region of the ion distribution could be treated in terms of an adsorption process according to Langmuir's adsorption isotherm, and the region beyond this inner layer, into the solution, could be validly treated in terms of a diffuse region of distributed ionic charge [Fig. 6.3(b)] as treated by Gouy" and by Chapman." In addition, if the ions were recognized as having finite size, including the annular thickness of their hydration shells (their so-called Gurney cosphere radii 1\ it was easy to define a geometrical limit to the compact region of adsorption of ions at the electrode surface [Fig. 6.3(c)]. This is taken to correspond to a Helmholtz type of compact double layer having a capacitance CH, while the remaining ionic charge density beyond this compact ion array is referred to as the "diffuse" region of the double layer, having a capacitance Cdiff. Cdiff and CH are conjugate components of the overall double-layer capacitance, Cdh related by the equation


1 Cdl

1 CH

1 Cdiff


corresponding to a series relation between CH and Cdiffaccording to an equivalent circuit:

-----II I

I t-I --

On account of the reciprocal form of the terms of Eq. (6.1), it will be seen that the Cdl will be determined by the smaller of the two components, CH and Cdiff.This is of considerable importance in determining the properties of the double layer and its capacitance as a function of electrode potential and ionic concentration of the solution. The original paper by Sternl4 in Zeitschrift fur Elektrochemie (1924) is somewhat obscurely written, but Parsons, in an important article in 1954,4 presented a much clearer version of this treatment in which the limit for a distinction between the Helmholtz compact layer and the diffuse layer beyond it [Fig. 6.3(c)] can be understood in terms of the distance of closest approach of counter anions or counter cations to the metal electrode surface. By introducing a distance of closest approach of finite-sized ions and thus geometrically defining a compact Helmhc1tz inner region of the double layer, the problem of a far too high capacitance that arises in the Gouy-Chapman treatment is automatically avoided. This difficulty arises since the capacitance of two separated arrays of charges increases inversely as their separation distance, so very large capacitance values would arise in the limit of infinitesimally small (point charge) ions very closely approaching the electrode surface.

The Stern theory of the double layer remained a good basis for general interpretations of electrode interface phenomena, including double-layer effects in electrode kinetics!" until the detailed work of Grahame in the 1940s on the double layer capacitance at the mercury electrode in aqueous electrolyte solutions reported in various papers, particularly in the seminal source review paper in Chemical Reviews in 1947.17 Its semicentenary was celebrated at a special anniversary Electrochemical Society symposium in May 1997. Grahame's work emphasized the great significance of the specificity of double-layer capacitance behavior at Hg to the nature of the cations and anions of the electrolyte, particularly the size, polarizability, and electron-pair donor properties of the anions of the electrolyte. This led Grahame to make an important distinction between an inner and an outer Helmholtz layer in the interphase which correspond to the different distances of closest approach that can arise for anions vis a vis cations at the electrode surface. This difference of distance of closest approach is mainly caused by the fact that most common cations are smaller than common anions (Table 6.1) and retain solvation shells due to strong ion-solvent dipole interaction.IO,IS,23 Thus, the Grahame model (Fig. 6.6) consists of three distinguishable regions: the inner Helmholtz layer, the outer Helmholtz layer, and always a diffuse ion distribution region. At extremes of polarization (i.e., for high positive or high negative charge densities), one or the other of the Helmholtz layer regions dominates, with a population of anions or cations, corresponding to such polarizations. Because anion distances of closest approach are usually smaller than hydrated cation distances of closest approach, the inner layer capacitance at positively charged electrode surfaces is usually about twice that at a corresponding negatively charged surface, (16-25 flF cm "), though this depends on the metal and the ions of the electrolyte, and the solvent. These aspects of double-layer capacitance behavior are of great significance for understanding the properties of double-layer supercapacitors and the magnitude of capacitance that can be achieved per square centimeter over various ranges of potential and at various electrode materials. TABLE 6.1. Pauling Crystal Ionic Radii of Alkali and Halide Ions (nm)

uNat K+ Rb+

0.060 0.095 0.133 0.148 0.169 0.139 0.138

F Cl-

0.136 0.181 0.195 0.216


Notes: SeeRef.10forother cales fionic s o radii. Source:Reprinted withpermission 1. Am. Chern. Soc., 49,771 from Society.


Copyright American Chemical


Chapter 6

The Double Layer at Capacitor Electrode Interfaces


and thus to + 30



i.n x 1015




i.n x io



10--{i 6.04 x


ILl 1 X 10

x 96,500

= +0.17 e atom-lor -0.11 e atom"!





f /'



FIGURE 6.6. General representation of the structure of the double layer showing different regions for adsorption of hydrated cations and less hydrated anions (Grahame model 17), together with solvent molecules and an adsorbed neutral molecule.



The range of values of accumulated electronic charges on the metal side of interfaces of double layers at electrodes in aqueous solutions extends from about + 30 f.1C cm-2 at potentials positive to the potential-of-zero charge (pzc), to about -20 f.1C cm-2 at corresponding negative potentials. This range depends on the electrode material and the electrolyte ions in the solution and the solvent, as well as the temperature and pressure (in the case of high applied pressures). For the above figures, it is of interest to calculate the range of excess charge per atom of an assumed planar interface. We can deduce that the number n of atoms having a diameter of 3 A (0.3 nm) in, say, a square array [i.e., in a (100) lattice configuration of particles in the surface of the electrode], is n = 1/(3 X 10-8)2 = 1.1 X 1015. Per atom, the excess charge densities in C cm-2 correspond then to

These figures are shown just to give an idea of the extent to which electron deficiency or electron excess can arise on the electrode surface of the double layer. These charge densities depend on the electrode potential, E, and hence can be controlled or modulated by direct, alternating, or pulsed voltage. At capacitor plates, as also at the interfaces of two-electrode double-layer electrochemical capacitors, the surface charge densities (plus and minus charges) arise on account of the application of a potential difference, till, between the two electrodes. In response to the applied till, electrons from one surface are driven through the external circuit containing the polarizing device (a power supply, a battery, or a regenerative braking dynamo in an electric vehicle hybrid system) to the other surface, establishing a difference of sign of charge density between the two plates. In an electrochemical capacitor, the respective plus and minus charge densities on the two plates are matched by net equal and opposite accumulations of respective negative (anion) and positive (cation) charge densities in the interphasial regions of the solution [Fig. 6.3(c)] over distances from ca. 1 to 100 nm into the bulk solutior., depending on total ionic concentration. Note also that when an interphasial potential difference is generated through a Nernstian thermodynamic equilibrium (as for the electrode potentials of a pair of battery cell electrodes on open circuit), a double layer is also spontaneously set up at each electrode interface but is not generated by charge flow from or to an external source of electric charge; the electrode equilibration processes generate their own double layers. Upon charge or discharge of the battery, the two double-layers will become more, or less, charged, depending on the direction of current flow and the changes (if any) of electrode potentials resulting from charging or discharging. Their dependence on the electrode potential E corresponds to the development and manifestation of the double-layer capacitance, Cdl = dq/dE, or integrally Cdl = q/ E or f1q/ till. It is of interest chemically that the above extents of controllable variation of surface electron excess or deficiency are comparable with the local changes of electronic charge density that can arise chemically on various conjugated aromatic ring structures such as benzene or naphthalene, Owing to the presence of substituents such as -OH, -S03H-, -CH3, -NH2, and -COOH, where charge density changes arise on account of electronic inductive and resonance effects.


Chapter 6 ON THE

The Double Layer at Capacitor Electrode Interfaces


A complementary variation of charge density on the solution side of the electrode interface arises through adjustment of the distribution of cations and anions of the electrolyte in response to changes of electron density, qM, on the metal or carbon surface, as was illustrated in Figs. 6.3(a), 6.3(b), 6.3(c) and 6.6. This adjustment arises in the inner or outer Helmholtz layer regions ofthe structure of the electrode/solution interphase [Figs. 6.3(a) and 6.6] and in a coupled way in the diffuse-layer region [Figs. 6.3(b) and 6.6]. The average interionic distances in the Helmholtz region of the double layer can be easily evaluated and correspond to the numbers of charges per square centimeter or the e/atom numbers. For the two data derived above, the time-average, square-lattice, interionic spacings would be 0.73 and 0.89 nm, respectively. Of course the Helmholtz layer is continually in a state of thermal fluctuation, as noted by Gouy ' and by Chapman," and some of the ionic charge (depending on ionic strength) will be distributed in the diffuse part of the double layer. At a hexagonally close-packed atomic surface [i.e., in a (111) lattice configuration], the number of atoms cm-2 is 2/-{3 times larger than on a (100) surface of atoms having the same diameter. On the basal plane of graphite crystals, the carbon atoms are arranged in such a hexagonal lattice, with interatomic distances of about 0.1 I nm. The interatomic distance normal to the hexagonal lattice (i.e., between the lattice planes) is substantially larger. Similarly, the electronic properties (e.g., work function) of the basal plane are very different from those of edge sections in which intercalation processes can also occur, e.g., accommodation of Li, K, and F2. The average 2-dimensional interionic distance in the Helmholtz layer (a square-lattice approximation, which is sufficient for the purposes of this discussion) will be inversely proportional to the square root of the total cation + anion surface excess of ions, F= F-,i + F+,i' However, according to Grahame, 17 the location planes (distances of closest approach) of adsorbed anions and cations in the Helmholtz compact region of the double layer will be different on account of the differences of closest approach of anions and cations to the electrode surface owing to specific electronic effects in the chemisorption of anions and to the usually more strongly developed coordination structure of hydrated cations than of comparable anions.1O,17 Taking into account the hydration of adsorbed ions in the double layer [e.g., the hydrated ion radii of K+ and F- are about 0.133 (rc,F- or rc,K+) nm] plus the 0.276-nm diameter of hydrating water molecules, it is seen that there is rather little free space (or free interionic distance) between the ions in the Helmholtz layer when their hydration radii are taken into account. In the case of the

relatively concentrated aqueous acid (H2S04) or base (KOH) solutions that are used as electrolytes for a number of electrochemical capacitor embodiments, it is clear that there must be very little free water in the interphase that is not influenced by the ions either in the bulk electrolyte or in the Helmholtz layer.'" This is an important but little realized aspect of double-layer modeling and has significant practical consequences, in that there may be local changes of solvent activity and balance during charging or discharging processes.





It is to be emphasized that charging the double layer involves, on the metal (or carbon) side of the electrode/solution interface, only changes in the density and distribution of the delocalized electrons of the metal-electron plasma (approximately one delocalized electron per atom). This is treated in recent works according to the so-called "jellium" model in which the lattice of "ions" of the metal containing the free conduction electrons is regarded as a structureless jellylike distribution. (In practice, at some carbon electrode materials, a change in potential can also cause modification of surface functionalities owing to a Faradaic partial current that corresponds to some pseudocapacitance charging.) The surface of a conductor can be regarded in two ways: (1) as the relatively fixed (apart from intralattice vibrations) atoms of the material or (2) electrically, as the moveable plane of emergent electron density'" at the boundary of the atomic lattice (Fig. 6.4), depending on electrode potential (Fig. 6.7). The wave theory of electrons leads to a concept and model of electron distribution at metal or carbon surfaces. The wave functions of the delocalized conduction-band electrons spill over at the discontinuity of the lattice at its nominal surface (see, e.g., Fig. 2 in Chapter 2 of Ref. 20), leading to a significant but exponentially decreasing probability of finding electron density beyond the nominal atomic surface of the metal. This spillover effect is enhanced by negative polarization of the electrode, corresponding to the situation in Fig. 6.7. In a vacuum, under the influence of an electric field of an appropriate polarity, cold (field) emissiorr'" of electrons into the vacuum can in fact be promoted. This well-known phenomenon is a good example of the spillover or spill out effect. The redistribution of this electron density relative to the "jellium edge" at the surface of the metal with changes of potential was illustrated in Fig. 6.7 (cf. Fig. 6.4). At relatively positive potentials to the potential-of-zero charge, the electron density boundary retreats inward at the electrode surface relative to the plane of the centers of surface metal atoms. At potentials relatively negative to the pzc, the electron density becomes pushed outward toward the inner region of the solution boundary, where it will interact more easily with cations and sol-

118 Double layer

Chapter 6

The Double Layer at Capacitor Electrode Interfaces




Bulk electron density








LNominal L__ surface

metal plane

LOCuS of centers of metal surface atoms

FIGURE 6.8. Overall construction of the electrode/solution interphase illustrating its surfacechemical and double-layer aspects. XM ami Xd are the intrinsic surface potentials of the metal and adsorbed dipoles, respectively.

FIGURE 6.7. Schematic profiles of emergent electron density at a metal at three potentials: one positive (1), one negative (3) to the potential-of-zero charge, and one (2) at the potential-of-zero charge. (After Lang and Kohn ref. 5).

vent molecules in the compact region of the double layer. This model has been envisaged and treated by Lang and Kohn.' and more recently by Amokrane and Badiali." More details are discussed in Chapter 7. The variation of the locus of the electron density overspill with changing electrode potential is important for modeling such processes as solvent-dipole adsorption and orientation, and the chemisorption of ions, especially anions, each of which factors determines, among others, the capacitance of conductor/solution interfaces as in a double-layer supercapacitor. In energy terms, this modification of surface-region electron density by a changing potential corresponds to effecti ve changes of the electronic work function, cP, of the metal (when it is an electrode) or its electron affinity (-cP), which is one of the bases of the effects of changing potential on the kinetics of electrode processesl'' (Chapter 3) and on the chemisorption of ions. The overall structure of the electrode/solution interphase is seen to be quite complex. Its construction in terms of surface-chemical, ion hydration, and double-layer aspects is illustrated schematically in Fig. 6.8.


Electric fields arise whenever there is a separation of electric charges. This is a matter related to Coulomb's law and Poisson's equation.l" and is a fundamental property of electrical behavior of the universe or physical systems. At a plane metal electrode interface with a potential difference of, say, 1 V across an ideally polarizable electrode double layer, the field E will be approximately E

= 1.0/3.8

10-8 V cm-1

(6.2) corresponds Na", The accrystal ionic or solvation

for a double-layer thickness of 3.8 A (0.38 nm). The latter figure to the distance of closest approach of simple hydrated cations, e.g., tual value of the thickness of the double layer will depend on the radius of the ion and the thickness of its time-average hydration shell.



The Double Layer at Capacitor Electrode Interfaces


From this it is seen that E has a very high value-some 2.9 x 107 V cm'". Sometimes nonelectrochemists are puzzled as to how such enormous fields can exist without electrical breakdown. Thus, in an ordinary capacitor, e.g., with a polystyrene dielectric, fields cannot be sustained beyond ca. 5000 V cm ". However, in the double layer, the behavior is quite different since there is no bulk dielectric in the normal sense associated with dielectrics in regular capacitors. Only the water of hydration of the ions and the monolayer film of adsorbed solvent water at the electrode interface (Fig. 6.6) constitute the dielectric of the double-layer capacitance. The situation is analogous to that on the atomic scale in, e.g., a molecular dipole such as HCl or HzO where the internal, interatomic field is also on the order of 107 V cm ". Similarly, the interion (±) local fields in an ionic crystal are of about the same magnitude. No charge transfer breakdown by passage of charges between the ions can occur since they are very stable in their regular ionic states, Na+ and cr. e.g., in a rock-salt crystal. In the double layer, leakage currents across it can arise only when thermodynamically and kinetically, electron charge transfer processes are allowed beyond certain critical potentials and corresponding interphasial fields, as outlined earlier. The interphasial field, E, across the double layer can also be calculated in a different way by employing the electrostatic equation of Gauss for the field generated normal to a metal plate charged to a density of q charges per square centimeter.

course has to be consistent with the first approximate way of estimating the field. The agreement or disagreement obviously depends on the value chosen for e in Eq. (6.4). More sophisticated calculations of E would take into account the microscopic distribution of charges of ions and associated solvent dipoles, and the electron overspill from the metal toward the solution side of the interface in the whole interphasial region (Fig. 6.8) constituting the double layer.




E =-4nq!e


where e is the dielectric constant (coefficient) of the medium in which the field is established by the charge density q. In the double layer, e is probably on the order 6 but varies with the field. Earlier we showed that the maximum charge density sustainable at an Hg electrode in aqueous electrolyte is about 0.17 e/atom of the surface. The electron charge is 4.8 x 10-10 electrostatic units (esu, see Chapter 4). Then we find

4n x 4.8 x 10-10 x. 0 17 x 3 x lOiS esu em-I 6


where the factor 3 x 1015is approximately the number of metal atoms per square centimeter in a close-packed Hg surface. Thus E::::: 5 x 105esu cm ". In practical units of V em -I, E is then found to be 5 x 105 X 300 V em -I, the factor 300 being from the definition I V = I esu/300. The resulting figure for E, depending on the value to be assigned for e, is seen to be of the same magnitude as that derived above by dividing a typical potential difference value across the double layer by an appropriate thickness value d. A very high value for E again results and of

The significance of double-layer capacitance at electrode interfaces has to be understood in terms of the conditions under which charge separation takes place between the electrode metal surface and the solvated ions in the solution, near the metal interface; in particular in relation to Grahame's concept of an "ideally polarizable" metal interface.l ' This concept of the ideally polarizable electrode was implicit in the early work of Bowden and Rideal,21 based on the observation of the change in potential of a mercury electrode with time in response to a constant charging current (galvanostatic charging). It was more clearly and definitively described in the work by Grahame'": an ideally polarizable electrode is one where changes of potential due to flow of charge to or from the electrode cause only changes of charge density on the metal and conjugately of ion density on the solution side of the electrode interface, leading to charging of the resulting double layer. The essential aspect of the ideally polarizable electrode is that with changes of potential, charges flow from the external circuit and within the solution only to charge the double layer (-+), with no charges passing across the double-layer interphase, i.e., through some Faradaic reaction. Such a charged interface is at electrostatic equilibrium at a given potential rather than Faradaically at a Nernstian thermodynamic equilibrium, though one of the important contributions of Grahame's paperl7 was to show how double-layer properties could be treated thermodynamically, for example, how surface charge densities in relation to surface excesses of cations or anions could be derived as a function of electrode potential from differential capacitance measurements. The Hg electrode in aqueous electrolyte solutions, ideally NaF, NaOH, or NaZS04, comes close to the above requirements for an ideally polarizable interphase, i.e., one with zero Faradaic current passing across the double layer. Hg is almost ideally polarizable over the potential range +0.23 V to -0.9 V (RHE). At potentials more negative than -0.9 V, significant Faradaic charge transfer currents pass owing to the reaction of solvent water decomposition, giving rise to Hz:



The Double Layer at Capacitor Electrode Interlaces



At potentials approaching 0.23 V or higher Hg becomes electrochemically oxidized according to the reaction (e.g., in KCI solution) 2Hg + 2CI- ~ Hg2Clz + 2e or in alkaline solution at a different potential Hg + 20H- ~ HgO + HzO + 2e (6.7) (6.6)

Thermodynamically, cathodic decomposition of HzO in the standard-state H+ ion concentration (activity) can commence at, or negative to, 0.0 V standard hydrogen electrode (SHE), but in practice, electrocatalysis for Hz evolution at Hg via the consecutive steps: HzO + Hg + e ~ HgH + OW followed by HgH + HzO + e ~ Hg + Hz + OH(6.9) (6.8)

a Faradaic solution decomposition component, iF, i.e., i = idl + iF, as discussed in Chapter 3, Section 3.8. In the presence of other solutes or impurities that have thermodynamic oxidation and reduction potentials that lie between the potential limits for water (or other solvent) decomposition, Faradaic currents can also pass in parallel with the double-layer charging current. Such Faradaic currents can obey the Tafel equation with respect to variation of the electrode potential, E (or overpotential, 11,i.e., log iF is proportional to IJ or E); or they may be diffusion controlled, depending on how their io values compare with their diffusion-limited maximum current densities, e.g., when some impurities are present at low concentrations. With modern impedance spectroscopy equipment, now there is usually no problem in distinguishing double-layer charging processes from Faradaic ones, including the onset of diffusion control in the latter.




. is very poor (exchange current density, io = 10- 13 A em Z ) so th at sigruifIc.antcurrents for Faradaic charge transfer across the double layer (correspondmg to a leakage current across the interphase) do not begin to pass until potentials more negative (vs. RHE or SHE) than ca. -0.9 to -1.0 V are attained. Then between +0.23 V and ca. -0.9 V, only double-layer charging currents effectively pass so that within that range the interphase approaches ideal polarizability with an almost ideal nonleaky double-layer capacitance being then manifested which can be experimentally characterized by means of transient charging curves or ac ~mpedance. The electrode interphase is then capacitor like in its electrical behavior, so that charging energy can be stored. Gold is another metal that exhibits almost ideal polarizability over a certain range of potentials: in this case between -0.2 V and + 1.30 V in aqueous HzS04 solution. Beyond +1.30 V, surface oxidation begins, eventually leading to Oz evolution on a gold-oxide film while below ca. -0.2 V, Hz evolution leakage currents become significant (> 10-5 A cm-z). Within the above limits, the electrode interphase behaves like an almost ideal capacitor but with significantly po4 tential-dependent capacitance, as is also the case at Hg. ,I7 It should be noted that with excursions of potential, positively or negatively, beyond the potential limits for solvent-electrolyte decomposition, further double-layer charging (or discharging) still occurs so that the total current density passing is then the sum of a double-layer charging component, idb and

The electrode interphases referred to here can usefully be represented by equivalent circuits (Chapter 17); that for an ideally polarized electrode is simply a capacitance (a in the diagram) which may, however, have a potential-dependent value. In the case where a Faradaic process may also pass a current that is parallel with the double-layer charging current, the equivalent circuit b in the diagram applies, with an equivalent Faradaic leakage resistance RF (Eq. 23 in Chapter 3). RF is usually exponentially dependent on electrode potential, E, but, for small excursions, LlV, of potential (see Chapter 16), it is approximately linear in LlV or overpotential, IJ. Its variation with electrode potential can be indicated by so-called "micropolarization" experiments or by observing changes in the diameters of Z" vs. Z' plots in the complex-plane representation of impedance measurements at various constant electrode potentials (Chapters 16 and 17).

The behaviors of equivalent circuits (a) and (b) are easily distinguished by the ~ifference between their impedance spectra with variation of frequency; thus (a) IS.purely capacitative while (b) has a maximum capacitative impedance for a gIVen value of Cdl and RF (for a particular potential) at a certain frequency.


Chapter 6

The Faradaic leakage resistance, RF, in circuit b is very important as the basis of self-discharge (see Chapter 18) in electrochemical capacitors and in battery cells. Its role will be analyzed in Chapter 18. In the case of electrodes that are base metals, nonideal polarizability usually occurs because of the possibility of anodic corrosion or oxide film formation at potentials already near the H2 reversible potential; this leads to Faradaic leakage currents in parallel with double-layer charging. In addition, such metals usually have larger exchange current density for H2 evolution from water, so they cannot be polarized very far cathodically to the RHE or SHE potentials without appreciable H2 evolution currents arising.

Chapter 7

REFERENCES J. E. A. Guggenheim, J. Chem. Phys., 4, 689 (1936). 2. N. K. Adam, The Physics and Chemistry of Surfaces, 3rd ed., Chapter 3, p. 107, Oxford
University Press, Oxford (1941). 3. H. von Helmholtz, Ann. Phys. (Leipzig), 89, 211 (1853). 4. R. Parsons, Chapter 4 in Modem Aspects of Electrochemistry, Conway, eds., vol. I, Chapter 4, Butterworths, London (1954). 1. O'M. Bockris, B. E.

Theoretical Treatment and Modeling of the Double Layer at Electrode Interfaces

J. O'M. Bockris and B. E. 7.1. EARLY MODELS

5. N. D. Lang and W. Kohn, Phys. Rev., Bl, 4555 (1970); B3, 1215 (1971). 6. S. Amokrane and J. P. Badiali, in Modem Aspects of Electrochemistry, 7. G. Gouy, Ann. Phys., Paris, 7, 129 (1917); 1. de Phys., 9, 457 (1910). 8. R. H. Fowler and E. A. Guggenheim, Statistical Thermodynamics, p. 77, Cambridge University Press (1939). 9. D. L. Chapman, Phil. Mag., 25, 475 (1913). 10. B. E. Conway, Ionic Hydration in Chemistry and Biophysics, Chapter 10, Elsevier, Amsterdam (1981). 11. P. Debye and E. Hiickel, Phys. Zeit., 24, 185 (1923). 12. L. Onsager, Phys. Zeit., 27, 388 (1926); 28, 277 (1928). 13. M. Green, in Modem Aspects of Electrochemistry, 343, Butterworths, London (1961). 30, 508 (1924). USSR, 6, 502 (1937). 14. O. Stern, Zeit. Elektrochem., J. O'M. Bockris, ed., vol. 2, Chapter 5, p. Conway, and R. White, eds., vol. 22, Chapter I, Plenum. New York (1992).

In order to provide a fundamental basis for understanding the properties and behavior of double-layer types of capacitor devices, this chapter gives a broad account of the theoretical treatments of the structure and capacitance of the double layer at electrode interfaces. This topic has been one of major activity and interest in electrochemistry for about a hundred years, and has now found substantial technological applications. In 1997, the Electrochemical Society Sponsored a major symposium on the double-layer to recognize the 50th anniversary of Grahame's seminal paper I in Chemical Reviews (1947). Electrostatic and thermodynamic treatments of the double layer are based on a model in which the interphasial region between an electrode and an ionic solution is ideally polarizable (Chapter 6), i.e., a potential difference can be established between a metal electrode surface and the inner boundary of an elec~rolyte solution without Faradaic charge transfer processes taking place. The Interphase is then ideally capacitative and the electrode is referred to ' as an ideally polarizable one. The interphase then has a pure double-layer capacitance that ideally is frequency independent in an ac evaluation of that capacitance. This is the ideal requirement for an electrochemical capacitor device. In practice, some frequency dependence is commonly observed, i.e., the phase angle for the double-layer capacitance may not have the ideal value of 90° at all frequencies, and potential-dependent de leakage can also occur (Chapter 18). Deviations from ideal capacitative behavior can arise when there is some

15. R. W. Gurney, Ionic Processes in Solution, Dover, New York (1940). 16. A. N. Frumkin, Zeit. Phys. Chem., AI64, 121 (1933); Acta Physicochim., 17. D. C. Grahame, Chem. Rev., 41, 441 (1947). 18. J. A. V. Butler, Proc. Roy. Soc. Lond., A 157, 423 (1936). 19. B. E. Conway, 1. Electroanal. Chem., 123, 81 (1981) 20. B. E. Conway, Theory and Principles of Electrode Processes, Ronald Press, New York (1964). 21. F. P. Bowden and E. K. Rideal, Proc. Roy. Soc., Lond., AI07, 486 (1925); A114, 103 (1927); A119, 680; 686 (1928). 22. J. A. V. Butler, Electrocapillarity, Methuen, London (1940). 23. J. D. Bernal and R. H. Fowler, J. Chem. Phys., I, 515 (1933).


Chapter 7

The Double Layer at Electrode Interfaces


dielectric loss associated with the solvent orientation polarization in the doublelayer dielectric (at very high frequencies) and/or when there are some slow anionic chemisorption processes that lead to lower frequency losses. In either case, there is energy dissipation in the charging and discharging cycles at an incompletely polarizable electrode interface. Another source of nonideal behavior is distribution of the double-layer capacitance over a porous electrode surface (Chapter 14). In a practical electrochemical capacitor device, frequency dependence of the overall capacitance is generally observed and is due, in addition to the "porous electrode" effect, to coupling with other equivalent series resistance (esr) components. As emphasized by Grahame,' the double layer at an ideally polarizable electrode is in a state of electrostatic equilibrium, in contrast to other types of electrode interphases across which free charge transfer takes place (nonpolarizable interphases). This corresponds to an electrochemical thermodynamic equilibrium, ideally of a Nernstian kind. Theoretical treatments of the structure (Chapter 6) and properties of the double layer can be conveniently considered in a hierarchy of four sections: 1. The first is the diffuse part of the double layer [Fig. 6.3(b), Chapter 6], beyond (toward the solution) the inner contact layer of ions (solvated or otherwise) that defines the so-called "Helmholtz layer" [Fig. 6.3(a), Chapter 6]. Historically, this was the first theoretical examination of double-layer properties in work by Gouy and by Chapman.v' 2. The second is the compact or Helmholtz layer," which became treated in terms of adsorption of ions as part of Stern's combination.' of the Helmholtz and Gouy models, taking into account the finite sizes of the ions of the electrolyte, which determine their distances of closest approach to the electrode surface [Fig. 6.3(c), Chapter 6]. Later, Grahame! treated the compact region in terms of two layers, one for anions located at their distances of closest approach to the electrode metal, and the other for hydrated cations located at their somewhat longer distances of closest approach, which are determined by their hydration (or solvation) radii, as shown in Chapter 6. The two regions distinguished by Grahame are referred to as the "inner Helmholtz layer" (or plane) and the "outer Helmholtz layer," (or plane), respectively. 3. Work on the two classically distinguished regions of the double layer in (1) and (2), was followed by important treatments of the role of adsorbed solvent dipoles and their potential (field)-dependent orientation in the compact regions of the double layer6-9 in determining the capacitance. 4. Finally, in relatively recent years, attention has been paid to the potential-dependent spillover of electron density from the formal surface of

the metal toward (or away from) the solution and compact layers, depending on the sign of charge of the electrode metal, and its influence on the interphasial capacitance of the electrode. 10-12 The above factors in double-layer behavior are interactive and determine the overall capacitance of the electrode interphase and its dependence on electrode potential, the types of ions in the electrolyte solute, and the solvent in which they are dissolved. Such an interphasial system is obviously complex and has been modeled according to various levels of recognition of this complexity. These are treated in the following sections.



For completeness and for historical reasons, in discussing the models of the double layer, we first outline the electrostatic statistical treatment of the diffuse part of the double layer. In practice, however, the behavior of the diffuse-layer capacitance is of less significance for the properties of electrochemical capacitors than is that of the inner layer since the latter dominates the overall interfacial properties at the high concentrations of electrolyte commonly employed in capacitors to minimize equivalent series resistance or internal pore resistance. However, treatments of the diffuse part of the double layer have formed a cornerstone in our understanding of double-layer properties, and an accurate theory of diffuse-layer capacitance is an essential requirement for evaluating compact layer capacitance. This latter mainly determines the capacitance of doublelayer-type devices, through application of the reciprocal sum relationship (Eqs. 7.11 and 7.12) to experimental data on overall capacity. Direct theoretical treatment of the inner layer itself, although included in Stern's paper.i' has been slow to develop in reliable form, owing to the complexity of the modeling involved (see later discussion). The treatment of the diffuse-layer ion distribution [Fig. 6.3(b), Chapter 6] and its dependence on the surface charge density, qM, on the metal, which is related to the latter's electrode potential relative to the potential at which the electrode bears a zero charge (the potential-of-zero charge), proceeds in the following way: Poisson's electrostatic equation is used in conjunction with the Boltzmann distribution law" and applied to the electrostatic energy experienced by both cations and anions owing to the interaction of their charges with the potential that results from the electrode surface charge or, more particularly, to the gradient of the electric field near the electrode. If 1jI, is the potential at a distance r from the electrode surface, the local concentration of cations and anions, c, and c., at r will be4


Chapter 7

The Double Layer at Electrode Interfaces


(7.1) where cO terms refer to the bulk average stoichiometric concentrations of cations and anions related by c+z+ = lc.z.l, and z, and z., are the charge numbers of the cations and anions, respectively. The space charge density p at r is then (7.2) if only electrostatic interactions determine the ion distribution. As in the Debye-Hiickel theory, a further relation is available between !fir an.d p, by use of Poisson's equation relating field gradient to space charge density, p, but expressed for a one-dimensional electrostatic distribution of ions in the form

At this point in the treatment of the diffuse double layer it is assumed that Gauss's relation (7.7) can be applied near the electrode surface bearing the charge density, qM, at the distance of closest approach, a, of the ions to the electrode surface. ea is the local value of the dielectric constant, e, in that region. ea is taken as a mean value though ea may depend on potential (field':') and will probably have a different mean value on each side of the pzc. Also, solvent dipole orientation (cf, dielectric saturation effects) at the electrode surface'<" leads to a lower value of e in the region r = a than further away from the electrode. From Eqs. (7.6) and (7.7), qM is obtained with the above assumptions as


~ (er alf/r)

ar l ar

= -4nPr


qM- _ [kTe


I- c+ {exp [ze(lf/a kT If/s)±] - I}JII2 +


since e may be a function of r in the double layer as first treated by Conway, Bockri;, and Ammary,14 Substitution of p; from Eq. (7.2) into Eq. (7.3) gives (7.4)

For a symmetrical electrolyte, when Iz±1= z, = Iz_l (= z, say) and c± = c+ = c: (= c, say),

qM Using the identity (7.5)

= (2kTciJ1/2 n

inh [Ze(lf/a - If/s)]




in Eq. (7.4), and integrating with er = a mean constant value of the dielectric constant s in the double layer (since e is not very sensitive to variation of r except very close to the electrode surface), gives


noting that when qM is positive !fIa- If/swill also be positive. It is seen that Eq. (7.9) is a relation between the qM and the potential difference between the compact layer plane (If/a) and the solution (If/s). Therefore differentiation of qM with respect to that difference will give a capacitance quantity, the capacity of the diffuse part of the double layer.



a:,' ~±[

8:kT ~

cI {ex







If the ions are point charges as considered by Gouy, If/ becomes identical a with ¢JM, the potential of the metal with respect to the solution, and distribution of the ions continues right up to the metal surface, a = O.By Eq. (7.9), a quantity i:)qMli:)(lf/a - !fIs) can be evaluated as aqM a(lfIa - IfIs)

for the boundary conditions alf/,/ar ~ 0, r ~ 00; and If/,~ If/s, the potential in the bulk of the solution where p r = O. The term -1 in the braces in Eq. (7.6) arises from the integration constant, i.e., when r ~ 00, exp[(!fIr - If/s)/kT] = 1 for alf/r1ar = O.

= (z2e2ce)112

cosh [ze(lf/a-If/s)]




The Double Layer at Electrode Interfaces


which may be identified with the differential double-layer capacity, C, that is associated with the ionic atmosphere charge distribution dependent on iJlf/ /iJr and hence on qM' This capacitance will be referred to as Cdiff· A comparison of values of the capacity calculated from Eq. (7.10) with those measured experimentally from the dependence of surface charge (qM), at the metal on electrode potential E, indicates, however, a major discrepancy with regard to the rate of change of C with tPM - If/sand the absolute magnitude of C at values of tPM - If/s> ca. 0.1 V. Only near the potential at which qM = 0 (the pzc) is the experimental behavior numerically comparable with that predicted by Eq. (7.10). The principal reason for the failure of this relation lies in the assumption that the ions are point charges so that the ionic charge distribution can be continuous up to the electrode surface. This allows a large space charge to arise very near the electrode, and hence an anomalously large capacitance is calculated. If, however, the ions have a finite size (e.g., as determined by their crystallographic or, more appropriately, their primary hydration radii, a,), the continuous charge distribution must be cut off at r = a, and the potential-distance relation near the electrode is then discontinuous as in Fig. 6.1(c) in Chapter 6. The importance of the distance of closest approach a for ions of finite size, referred to earlier, was discussed by Grahame! but was recognized by Stern5 and was mentioned by Gouy in regard to an effective thickness of the double layer. The overall potential difference, tPM - If/s, between the metal and the solution can then be regarded as made up of two parts tPM - If/I and If/l - If/swhere If/Iis the mean potential at r = a, i.e., If/a == If/I' If/l is referred to as a mean potential since the potential in the double layer will fluctuate laterally across the interphasial region because of the discreteness of the ionic charges involved. The symbol e, is used to maintain conformity with the earlier Russian literature, but tP2 has been used as the symbol for this quantity in some later publications. Both these contributions can be dependent on qM so that

Cz is the diffuse-Ia~er cap~city, r~fer:red to earlier as CH and ediff, respectively. The Cz term associated WIth the rome atmosphere distribution is given b E (7.10) with If/a = If/I,so that taking reciprocals ofEq. (7.12) and introducin~ E~' (7.10) yields . C = C1 (zZe2cel2nkT) Cl + (zZezcel2nkT) I 12 cosh[ze(lf/I - If/s)2kT]
I 12


- If/s)2kT]


For small deviations of potential around the pzc, this function for C is mainly determined by Cz and increases approximately exponentially with If/I_ If/son each side of the pzc. It is seen that when CI »Cz, C = Cz (from Eq. 7.10 with If/a = If/I),or when Cz» Cl> C = CI; the first inequality is valid when If/I- If/sis quite small and the second when it is large. It is important to note that it is the value of the smaller capacity contribution that mainly determines the overall value of C since the two contributions are in a series relationship (Eq. 7.12), as discussed in Chapter 6 (Eq. 6.1). . Apart from giving a quite accurate representation of the diffuse-layer capacitance, the Gouy-Chapman theory is important because it provides the means of evaluating the inner-layer capacitance, Cl>which is of greater interest, from t~e overall C measured, through Eq. (7.12). In fact, at the rather high concentrations of electrolytes employed in aqueous double-layer capacitors, it is the compact or Helmholtz layer capacitance that mainly determines the capacitance of a double-layer electrochemical capacitor device. . Equation (7.13) represents the data for capacitance at the mercury/solution Interface (e.g., for a KCl solution I), if CI = 18 f-lF cm-z is assumed when cations pre~;rentially pop.ulate the .inter~ace region (tPM - If/snegative) and CI = 38 f-lF em when the amons dominate III the double layer (tPM - If/spositive). Typical capacity-potential profiles for a mercury electrode are shown in Fig. 7.1 (from Ref. 1). The capacity contribution CI is referred to as the capacity of the Helmh?l~z layer and Cz is that of the diffuse layer or ionic atmosphere region. Further dIVisions of CI into contributions from the inner or the outer Helmholtz layers, I referred to earlier, will be discussed later. . In Eq. (7.13) it will not generally be validl.15 to assume that the same potential e-. applies to the position of closest approach for anions and for cations Owing to the more polarizable nature of anions and their specific affinity as adsorbed surface ligands for some metal surfaces, anions may approach more closely than hydrated cations at a given potential, so that If/Iwill not have quite the ~ame significance for anions as for cations. It is convenient to call this potentI~llf/z for the region of closer approach and specific adsorption, i.e., chernisorptIon associated with anions. This matter is now examined in more detail according to the theory developed by Stern ' and reviewed by Parsons. 15

(7.11) The first term is the reciprocal of the overall capacity C mentioned earlier, and the other terms have a similar nature, so that Eq. (7.11) can be written


1 C

1 CI

1 C2


where CI and Cz are obviously defined by comparison with the two terms on the right-hand side ofEq. (7.11). Equation (7.12) also implies that CI and C2 are additive in a series combination. CI is the Helmholtz compact layer capacity and


Chapter 7

The Double Layer at Electrode Interfaces





0.1 M 28 38 20


7.4.1. Stern's Treatment We referred earlier to the fact that the diffuse layer, as a model for the double layer, is applicable only to dilute electrolyte solutions over a small potential range near the pzc. Stern? recognized that a satisfactory theory of the double layer must take into account both the finite size of the ions adsorbed and any specific chemisorption interactions they may experience with the metal surface. The charge qs on the solution side of the double layer may thus be regarded as being made up of two contributions, q2 and q" arising, respectively, from the space charge associated with the ionic atmosphere and from the charge associated with specifically adsorbed ions. Then, since qM = -qs, (7.14)

12 22

Influence of /Cdiff.






Z <l:






-1.6 -20



0:: W

O.lM 80 1M





_J I


34 60

The charge q2 will be given by Eq. (7.8) with '1'2 as the relevant potential, i.e., 'l'a = '1'2' The charge ql was obtained by Stern in terms of site-fraction statistics for solution and surface phases by applying a form of the Langmuir isotherm with an electrochemical Gibbs energy of adsorption determining the extent of adsorption as a function of solution concentration of the ions. Stern's derivation for the quantity '1'2 was unnecessarily involved, and Parsons15 has given a useful clarification of the calculation which need not be repeated here. The result of applying the Langmuir isotherm is


CD 0

:::> 0


q, = «a


+ (1/Xs) :P[SGi:lkT]

+ -1-+-(-l/-X-s)-:-:-P-[L'i-=V::-~-lk-T-]


18 20 10 -0.4 -0.8 10 0.8 0.4 0 -0.4 -0.8 -1.2 -'.6





TO p.z.c./V

FIGURE 7.1. Differential capacitance of the Hg/aqueous solution interface at 298 K in four electrolytes: NaF, NaCl, NaI, and NaNG3. Concentrations indicated. Reprinted with permission from Chern. Rev., 41, 441 (1947). Copyright American Chemical Society.

where L'iG~ the standard electrochemical Gibbs energies of adsorption of the are indicated cations or anions, i.e., the L'iG~terms normally contain both electrical and chemical Gibbs energy contributions, the latter defined for some standard state concentration or mole fraction Xs of salt in the bulk solution. The term Q is defined as the number of sites per square centimeter available for adsorption at the electrode interface. It is assumed that this number is the same for cations and anions, and that the adsorbed ion can freely replace solvent molecules, which are also necessarily adsorbed at the surface, with a relative fractional coverage depending on coverage by chemisorbed anions. Reference has been made earlier to the fact that the L'iG~ terms in Eq. (7.15) are electrochemical Gibbs energies of adsorption; following Stern's assumptions, they may be written for convenience in the form


Chapter 7

The Double Layer at Electrode Interfaces


(7.16) where flG~terms are the chemical standard Gibbs energies of adsorption of the indicated ions (±), and IfIz is the mean potential in the plane of specifically adsorbed ions. Experimentally, however, it will not usually be possible to separate the /:}'G~terms into their component chemical and electrical contributions. A~ explicit mathematical expression for the inner-layer capacitance, Cl>in Eq. (7.12) corresponding to oqMlo(¢M - '1'1) in Eq. (7.11) or dqlld(¢M - '1'1) in Eq. (7.15), is not easy to obtain directly, though the electric potential is implicit in the flG~terms. Hence, in practice, experimental data for the overall C are split, reciprocally, into CI and C2 components by calculating the theoretical C2 diffuse-layer capacitance for a given concentration and electrode potential, and combining it reciprocally with the experimentally measured C: lICI = lIC In most cases only one type of ion is specifically adsorbed (the anion, except in cases of organic electrolytes such as tetraalkylammonium salts), so that Eq. (7.15) could be written for anions as (7.17) lIC2·

form of the isotherm written as Eq. (7.18) may in fact be physically the more reasonable one. Stern expressed the integral capacity KI of the double layer as (7.19) and assumed that '1'2 = '1'1, which enables values of IfII to be calculated as a function of ¢JM, the potential of the metal, for known KI and solution concentrations in the absence of specific adsorption. (Usually this evaluation is most conveniently performed numerically by choosing '1'1 values and obtaining ¢JM as the dependent variable, since IfII enters the arguments of exponentials in q2') The term KI is the integral capacity of the region between the metal surface and the plane of centers of the adsorbed ions. The assumptions 1f12 = IfII and flG; = 0 are equivalent to taking ql = 0, i.e., when specific adsorption is absent. Stern did not distinguish, however, between '1'2 and '1'1 for the case of specific adsorption, and this distinction constituted an important element of Grahame's treatment I referred to earlier. As mentioned several times before, the inner-layer capacitance cannot be properly interpreted, especially with regard to its substantially differing values on the positive and negative sides of the pzc at Hg (Fig. 7.1) or Au, without recognizing different values of the closest approach distance, a, for positively and negatively charged electrode surfaces. This situation corresponds to different effective radii of anions and cations, owing to their usually differing hydration and crystallographic radii, and because anions tend to be more strongly chemisorbed (owing to their greater lone-pair electron donicities) than corresponding isoelectronic cations, e.g., Na+-F-, K+-Cl-, Rb+-Br-. Other factors (such as hydrophobicity) are involved in the adsorption of large organic cations (e.g., Et4N+, n-Bu4N+), which have major effects on the interphasial capacitance of electrodes. These ions are important in nonaqueous solution capacitors, where they often form the electrolyte with co-anions such as PF6, BF4, Cl04, and are adsorbed at carbon powder or fiber surfaces. The recognition of the importance of different distances of closest approach for cations and anions, and the corresponding distinction between inner Helmholtz and outer Helmholtz planes for the description of the compact layer, was one of Grahame's major contributions, I complementing his detailed treatment of the thermodynamics of the interphase at the ideal polarized electrode.

since usually for dilute solutions, (1/Xs) expflG~/kT» 1. It is clear that q-lz:e is the number nl of ions in the adsorbed layer and that nl/Q is the relative cov. 15 . erage e I 0 f these IOns, r.e., (7.18) which is a limiting form of the Langmuir isotherm for low coverage (1 - el ~ 1) or a two-dimensional form of Henry's law (the concentration in the surface phase, l, proportional to the mole fraction in the bulk phase is analogous to the gas solubility relation where the mole fraction of dissolved gas in solution is proportional to its partial pressure in the gas phase). It should be noted that the form of the adsorption Eq. (7.18) also follows rigorously for mobile adsorption of species at a surface, whereas the Langmuir isotherm applies strictly only to fixed-site adsorption on a lattice of identical sites, with no lateral interactions between the adsorbed species. Neither of these assumptions is likely to be physically realistic at a mercury electrode, where the adsorption will certainly be mobile, and in general, two-dimensional lateral interactions between adsorbed ions themselves and with their image charges "in" the metal will not be negligible. Apart from interaction effects, the approximate

7.4.2. Quasi-Chemical

Aspect of Anion Adsorption

In the case of anions, their adsorption at electrodes can be usefully regarded as a quasi-chemical process involving partial charge transfer. The extent, 0, of this charge transfer is related to the so-called "electrosorption valence." It


Chapter 7

The Double Layer at Electrode Interfaces


is determined by the polarizability of the anion, its electron-pair donicity, and the electronic structure of the electrode metal. In addition, the extent and energy of solvation is an important factor. The chemisorption process can be represented by the reaction equation:

+ M ~ M·A(l-J)-


+ Je(in M) + (m

- n)S

where m and n are the inner coordination solvation (S) numbers of anions A-in solution and in the adsorbed state, respectively. A-adsorption is usually enhanced when the metal surfaces, M, is positively charged, i.e., when its Lewis acidity is increased. The conjugate relation between A-and M shown here is usually also dependent on the electronic properties of M. This is an important factor that determines the dependence of double-layer capacitance in various electrolytes on the substrate electrode material.

In parallel with this theoretical work were new (in the 1960s) experimental studies by Hills and Payne20-23 on the properties of the double layer at Hg in nonaqueous solvents, the electric polarization properties of which are substantially different from those of liquid water. 24 Also to be noted is the fact that ions (cations or anions, depending on the e~ectrode potential relative to the pzc) populating the compact (Helmholtz) region of the double layer are usually substantially hydrated (especially cations24) and hence locally influence the orientation and state of polarization of the water molecules that also always populate the double layer.6-8 Thus the structure of the double layer at the molecular level is quite complex. .The energies ~f solvation of ions in nonaqueous solvents are usually substantially less than III water (hence the lower solubilities of many salts in such solvents), so that the tendency for ions, especially anions, to become chemisorbed at electrodes is greater in aprotic media than in water. This leads to an indirect effect of the solvent type on double-layer capacitance.


7.5.2. Types of Solvents that Constitute the Double-Layer




In the field of double-layer electrochemical capacitors, major interest has developed in the use of nonaqueous electrolyte solutions, rather than aqueous ones, owing to the substantially higher operating potentials that can be achieved, as has been noted earlier. In addition, the capacitance behavior as a function of electrode potential usually differs in a major way from that for water and has therefore attracted interest for theoretical and computational modeling. Until about 1961, much of the classical work on the capacitance of the double layer had ignored the fact that solvent molecules largely populate the double layer. However, several important exceptions led the way toward more complete recognition of the role of the solvent in double-layer properties and molecular-model treatments. First, the papers of the present author13,14 and of Grahame'" considered the field dependence (i.e., dielectric saturation) of the effective dielectric permittivity of the solvent medium constituting the dielectric of the double layer in its inner region. Later Macdonald 17 took into account both the field dependence of the dielectric polarization and the associated self-comp:ession of the solvent dielectric in an inhomogeneous field, referred to as electrostriction. Also, implicitly, the properties of solvent water at electrode interfaces were taken into account in simple dielectric polarization models of the potential dependence of adsorption of organic molecules at electrode surfaces in well-known papers by Frumkin" and Butler.l?

Solvents are of three main types: (1) associated ones such as water and D20, methanol and other alcohols, hydrogen fluoride, formamide, and carboxylic acids in which strong interactions and correlated orientations of the molecules (Chapter 5) occur through hydrogen bonding; (2) polar ones where there are still quite strong electrostatic dipole-dipole interactions but without the directional specificity and strength of H bonding; examples are acetonitrile, nitromethane, acetone, dimethyl sulfoxide, dimethoxyethane, propylene carbonate, ethyl and methyl carbonates, dimethylformamide, methylene dichloride, and esters; and (3) finally nonpolar ones such as hydrocarbons and fluorocarbons. The use of nonaqueous aprotic solvents for electrolytes in carbon capacitors has become very attractive because substantially higher voltages on charge (up to 3.5-4.0 V) can be obtained owing to the higher solution decomposition voltages that apply (Chapter 2) with such systems. Hence, higher energy and power densities can be achieved. This preferred use of nonaqueous solutions for ~lectrochemical capacitors in order to maximize energy density has enhanced interest in the specific dependence of double-layer structure and capacitance on the type of solvent in recent years. Thus the fundamental treatments of the role ?f solvent polarity and structure in determining double-layer capacitance behavtor have come into new focus in recent years. This matter is sufficiently important for a separate chapter (8) to be devoted to it, in relation to the development of nonaqueous electrochemical capacitors. . There are several indirect aspects of the use of nonaqueous solvents for hIgher voltage capacitors. (1) They are usually more difficult to purify than


Chapter 7

The Double Layer at Electrode Interfaces


water and hence can contain small but significant quantities of impurities that can be either electrochemically reduced or oxidized, leading to self-discharge. (2) Because of their usually smaller solvating power for ions, ion adsorption at electrode interfaces can be stronger, modifying the double-layer capacitance behavior. (3) Because their dielectric constants are usually lower than that of water, specific double-layer capacitances tend to be lower than in aqueous solutions. Also, (4) because of these lower dielectric constants, the specific resistivities of their solutions of electrolytes are usually less. This tends to raise equivalent series resistance values in nonaqueous solution capacitors, leading to diminished power densities (see Chapters 15 and 17). For aprotic solvents, the polarity is often measured by their bulk dielectric constants. However, in evaluating local intermolecular interactions, and interactions with charged surfaces and ions, the individual dipole moment and/or the donicity of unshared electron pairs are usually found to be the more significant quantities. In fact, as Table 13.1 in Chapter 13 shows, the dielectric constant is little related to the lone-pair donicity numbers (based on heats of complexation of the solvent to a Lewis acid such as SbF5). The behavior of the metal/solution double layer depends very much on these electrical polarity properties of the solvent of the electrolyte solution in contact with the electrode metal surface.

of the double-layer dielectric as d == sA (0.5 nm) (== radius of electrostatically adsorbed Na+, plus the diameter of a water molecule), G is found to be about 9.6. A smaller value of d == 0.3 nm would give G == 6. Obviously, there are uncertainties about what value to assign to d on a molecular basis, but the above thumbnail calculation indicates that for Hg at potentials appreciably negative to the pzc, G is well below its normal value (ca. 78 for water at 298 K) for bulk water. Hence there is appreciable saturation of the dipole orientation polarization contribution in G for water in a polarized double . .. 1ayer at H g.'1314Th'IS IS a cone I'usIOn t h at IS Important for the state of water in the double layer at polarized electrode interfaces. Similar conclusions apply to other polarizable electrodes, e.g., gold and carbon. On the positive side of the pzc at Hg or Au, C has values between 35 and 45, about twice those for the negative side. This is probably mainly due to d being smaller than on the negative side owing to the closer approach of anions (because they are in a less hydrated condition than cations of the same charge) than of cations to the electrode surface, as argued by Grahame.' Since it is difficult to evaluate G and d in the quotient eld separately from an experimental value of C, some of the increased value of C could be due to a value of G in the double layer that is larger for positive than for negative values of potentials relative to the pzc. This could be due to the different orientation and hydration interactions found near anions in the double layer from those found near cations.


Dielectric Constant in the Double-Layer

Interphase 7.5.4. Electrostatic Polarization of Water as Solvent in the Double Layer

The electrical state of the solvent in interphasial double layers can be usefully determined from evaluations of the effective dielectric constant, G, of the solvent medium in the double layer, which is derived from capacitance measurements. Under certain conditions, complementary information can also be obtained by applying Fourier transform infrared spectroscopy to thin solution layers at electrode surfaces. The latter procedure provides information on the intra- and intermolecular vibrations and hindered rotations (librations) of the solvent molecules. The former procedure provides information on the overall electrical polarizability of the solvent molecular system in the narrow interphasial region of the electrode metal boundary with the solution where molecular orientation'<" can be influenced by the electrical field across the double layer. The effective dielectric constant, G, of the double layer at Hg at potentials negative to the pzc in an electrolyte of aqueous NaF, the ions of which are not chemisorbed (specifically adsorbed), can be calculated from the relation referred to earlier, namely (in the system of rationalized electrostatic units,)


where KO = 8.85 X 10-12 F m-I is the permittivity of free space. C for the above system has a value of about 17 .uF cm-2 negative to the pzc. Taking the thickness

Solvent polarization at charged electrode interfaces was recognized in early theories of the adsorption of ions and organic substances in the double layer, e.g., in the theories of Butler!" and Frumkin.l" The treatments were, however, oversimplified for a complex solvent like water,24,25where dielectric saturation and orientation correlation effects occur, as discussed in Chapter 5. A central aspect of the inner-layer capacitance of the Hg/aqueous solution interphase is the hump in the compact layer C that is observed! at qM values displaced a little positively to qM = 0 (the pzc) (Fig. 7.1). This phenomenon, which is also anion dependent, has attracted a lot of attention in relation to the role of solvent polarization in determining the behavior and properties of the inner region of the double layer. Various models have been treated in attempts to represent this experimental behavior.i" The problem of solvent dielectric polarization in the double layer was taken up by Conway= and by Grahame.l" who gave a useful empirical expression for the dependence of G on field strength according to a treatment by Booth27 on the dielectric saturation of water that is based on the KirkwoodOnsager treatment (see general reference 1 in Chapter 4) of the dielectric constant of polar liquids.


Chapter 7

The Double Layer at Electrode Interfaces


Macdonald'" provided calculations on the double-layer capacity at Hg electrodes in terms of (1) the dielectric constant in the charge- free layer of water at the interface and (2) the electrostriction in this layer associated with the e1ectrostrictive tension eE2/8n due to the field E. By means of three or four adjustable constants, he obtained a good fit to the experimental differential capacitance-potential relation for 0.01 N aq. NaF at 298 K. The main parameters that determine the fit of the theoretical treatment to the experimental data (of Grahame) are the constants a and [3, which are defined by a

include the effects of electrostriction (see Ref. 17), while lateral interaction effects in the inner layer of oriented adsorbed water molecules are only crudely treated. For example, the angular dependence of H bonding in water-water interactions25,27,28 is not considered and cooperative effects in H bonding have received little attention in treatments of water in the double layer, except qualitatively in terms of semiempirical cluster models, and in lattice treatments by Parsons and Reeves29 and Guidelli." referred to in Section 7.5.6.

= (do/d

- 1)/8np

= [3/8n





of Solvent Dipole Orientation at

where p is the electrostrictive pressure eE2/8n and dO/dis a reduced relative thickness of the double layer; b is the coefficient of field squared in the BoothGrahamel6,27 relation (7.22) giving the differential dielectric constant, e, as a function of field E. The constant a determines the compression in the double layer and b in Eq. (7.22) the dielectric saturation (see Refs. 13 and 16). In relation to other treatments (see later discussion) of the double-layer capacity at Hg, based on properties of the water solvent in the inner layer, it should be noted that the theories that treat the problem in terms of a field-dependent dielectric constant are equivalent in some ways to the molecular model theories that allow for field-dependent orientation of H20 dipoles or clusters of dipoles (see later discussion). In the field-dependent dielectric theories of Conway, 13,14 Macdonald.!" and Grahame.!" the orientation aspect of water molecule polarization in the double layer is implicit in the variation of e with E, while interaction effects are implicit in the correlation factor g for orientation in the Kirkwood-Onsager theory of dielectric polarization. This latter is the basis of Booth's treatment of dielectric saturation employed by Conway.P Grahame.'? and Macdonald. 17 A frequent criticism of these dielectric theories is that the dielectric saturation function (Chapter 5) may not be applicable to a lamina of solvent in the inner layer that is only one or two water molecules in thickness. The question then resolves itself into whether such dielectric theories are any less realistic than the somewhat arbitrary molecular model treatments that have been given more recently6-8,26 (see later discussion) and that will now be discussed. The answer is not an unequivocal "yes," as will be seen in the following section where it will be shown that molecular model treatments do not give unified unambiguous conclusions regarding the role of solvent adsorption and orientation in double-layer properties. The molecular model treatments do not, for example,

Charged Interfaces

Two-State Dipole Orientation Treatments

A useful molecular model treatment (a so-called "primitive model") of solvent orientation polarization at an electrode interface was given by WattsTobin and Mott.6,7 Two orientation states of the solvent dipoles of moment 11., up i and down -L, aligned with the electrode field E arising from net surface charge density qM were envisaged. The orientation polarization in the interphase at the electrode surface could thus be calculated in terms of the relative populations Ni/(Ni + N-L) and N-L/(Ni + N-L) of the two states of orientation. Ni and N-L are determined (1) by the field, (2) by temperature, and (3) by any lateral interactive forces8,27 between the oriented and unoriented dipoles. Interaction effects were not, however, taken into account in the original treatment. The Boltzmann distribution function is employed to calculate the relative populations xt and N-L;hence the local orientation polarization per square centimeter may be calculated. Thus

(7.23) where U is the net energy of the oriented dipole in the field. When lateral interactions are significant.'' U has the form
U = II.E

Wn N11N - Wn NilN


where N Ni + Nl, W is the pairwise lateral interaction energy between oriented dipoles and 2n is the coordination number in the interphase. The direction of II. in relation to that of the field vector E determines the sign of II.E and hence that of the energy U. Since


Chapter 7

The Double Layer at Electrode Interfaces



Nt Nt



an orientation distribution function '1( can be defined as

9\ =

Nt -Nt t I
N +N-¥

of oriented dipoles and a here is the electronic polarizability of the dipoles determining the magnitude of induced moments. The relation between LlV, d, and the other quantities shown can be used to obtain an empirical~xpre~ion for the effective dielectric constant ce by writing LlV/d = mean field E and E= -4nqMlce by means of Gauss's law. Then Ce is seen to be given" by (7.30)

= tanh UlkT,


9\ = tanh [flE - 9\WnlkT ] kT


9\ is a measure of the orientation polarization per unit area in the interphase at the charged surface since Nt + Nt is equal to the total number of orientable dipoles per square centimeter, assuming that electrostriction in the double layer does not change this number. The variation of 9\ toward a saturation orientation value of ± 1, i.e., when Nt == Nt + Nt or Nt == Nt + Nt, occurs over a range ofE dependent on the magnitude of nW. Here xt or Nt == Nt + Nt, i.e., ot or = 1, corresponds to orientational dielectric saturation in the double layer. E can be related to the surface charge qM by E = 4nqM/ that 9\ asj(qM) can be obtained. A similar treatment by Levine, Bell, and Smith9 took into account (1) the contribution of the induced dipole moment, fli, due to neighboring dipoles of permanent moment flp; (2) the effect of neighboring dipoles on the field at a given site, namely,


(7.28) where ne is an effective coordination number for dipoles at distance d from each other; ne is taken as 11 according to a relation due to Topping;" and (3) the possibility that one orientation is intrinsically preferred to another owing to chemisorption or image forces. Levine et a1.9 derived the total potential drop LlV across the oriented water layer as fl(9\ - J) - 4naqM LlV=4nqMd+4nN·--------~-1 + anJd3

and its variation with field toward a saturation value depends on the variation of the dipole orientation function 9\ with field; 9\0 is any residual (specific) orientation when qM = 0 at the potential-of-zero charge referred to earlier; it is indicated experimentally. The main weakness of the treatments by Watts-Tobin and Mott," and by Levine, Bell, and Smith9 is the neglect ofH-bond structural effects25,27-30in the interphase, i.e., correlated orientation effects. Such effects are not adequately taken into account by the lateral interaction term wn9t, since this has no general angular dependence on orientation,25,27 only extreme orientations being con sidered.8,9 The limiting two-state situation is undoubtedly a serious oversimplification since molecules of the water dielectric in the interphase will be oriented at various angles and some average polarization orientation angle 0 will be a function of field and Wne, and will also be determined by the angular dependence of H-bond energy between neighboring molecules.P:" The problem then becomes a 2-dimensional version of the Kirkwood theory for associated dielectrics so that the simple, elegant, but perhaps oversimplified features of the Watts-Tobin and Mott modee,8 would become lost. The problem of some overestimate= of the dipole orientation effect in the two-state models was solved by Pawceu+' by introducing the idea of a three-state orientation distribution, with the "third" state lying flat along the electrode surface as in Reeve's model.i"

Cluster Models for Water Adsorption and Orientation


where J = tan20/2 for a mean angle 0 of preferential orientation of the H20 dipole toward the electrode, e.g., due to asymmetric dipole image interactions and/or to specific chemical affinity of one end of the dipole, such as H20 for the electrode metal. fl(9\ - £5) is the mean moment per site on the surface in the array

Experimental studies at Hg indicate that some specific direction of orientation of water dipoles tends to occur34-36 at Hg, Ga, and other metals in the absence of an opposing Coulombic field due to excess charge. Hence it is necessary to take into account both orientation effects due to chemisorption and those due to changing electrode charge. This was recognized by Levine, Bell, and Smith," while Bockris, Devanathan, and Muller'' noted an equivalent preferential orientation effect which they considered would occur because of interactions of the noncentral dipole in water with its image charges in the metal (Hg)


Chapter 7

The Double Layer at Electrode Interfaces


surface. This is, of course, a different effect from that associated with specific interaction of the lone-pair orbitals on 0 in H20 with the atoms in the Hg surface, but both are connected with the electron charge distribution in the water molecule and its donor and acceptor properties at metal interfaces. At molecules containing an S atom (e.g., thiourea), the specific affinity of the S-end of the molecular dipole for Hg is very much larger. Frumkin and Damaskin37 assumed that the contribution of the surface dipole potential (Xd) to the total surface potential could be written (7.31) At electrode interfaces, tJ.Xd is a function of potential and hence qM; then the dependence of tJ.Xd on qM can be written as a differential coefficient d(tJ.Xd)ldqM, which is to be identified as a reciprocal differential capacitance, lICd, due to dipole orientation. It combines in a series relationship with other capacitance contributions of the overall double layer due to free charge accumulation. Therefore, only if Cd is sufficiently small will its effect on the overall measured capacitance be important. Thus its correct evaluation is a critical matter in modern theories of the double layer which properly take into account the solvent layer.26,27 This tJ.Xd dependence of qM or electrode potential is the basic origin of a solvent polarization contribution to the overall inner-layer capacitance. It was first treated in the two-state dipole orientation model of Bockris, Devanathan, and Muller referred to earlier. The components 1 and 2 in Eq. (7.31) arise from freely oriented dipoles in the electric field due to excess charge and from chemisorbed H20 dipoles, respectively. The nonchemisorbed dipoles were regarded as being H-bonded in groups (clusters) which themselves only weakly interact with one another (cf. some theories of water structure/"). This conclusion about association was deduced " from the shape of the adsorption isotherm for various organic substances at Hg. However, it must be stated that the general shape of the electrostatic isotherm in terms of the dependence of the Gibbs energy of adsorption for a given coverage (isosteric condition) on electrode surface charge (e.g., the width of the bell-shaped curve at half its height39,4o) can be reproduced by a lateral repulsive interaction effect term of various forms, so it seems unlikely that the shape of the isotherm can be uniquely indicative of the role of clusters. For one thing, the value assigned to the size of the adsorbate relative to that of the water molecu1es39 or clusters37.38 is critical in evaluating the proper Gibbs energy of adsorption from experimental data as a function of charge at a given coverage. Quantitative modeling calculations were made, allowing this approach to be compared with experiment (Fig. 7.2). The parameters were two capacitance contributions, K, and K2.

N 1


~ o




-5 -2




'1".1 ~C'cm

Inner-layer capacitance as a function of surface charge qM calculated according to Damaskin and Frumkin,37 for the parameter values indicated. Points: data for Hg.

From the data required to empirically fit the experimental results, an estimate of the area of the clusters of adsorbed water was given as 0.24 nnr', The mean dipole moment of the clusters was found to be ca. 0.75 D. The effective value of s was also derived asj(qM)' It should be noted that a capacity hump arises in curves 2 and 3 of Fig. 7.2, but displaced from the pzc. This is an important feature of most of the experimental capacitance results at the Hg/aqueous solution interface, which has been the subject of much previous discussion.v"! including the idea that H20 dipole reorientation was involved. The hump is, however, usually sensitive to the type of anion of the electrolyte, which led Bockris, Devanathan, and Muller8 to offer an explanation in terms of changing anion adsorption and repulsion in the double layer with increasing positive qM' Probably both oriented chemisorbed water


Chapter 7

The Double Layer at Electrode Interfaces


and specifically adsorbed anions" playa role since the state of orientation of chemisorbed water will almost certainly depend on the presence of anions in the compact layer. Curve 4 in Fig. 7.2, for K2 = 50 flF cm ", corresponds to the experimental behavior of Ga where the capacity rises sharply at the pzc with no hump. Stronger chemisorption of water dipoles corresponds to a lowering of K2 (from 200 to 50). Following Damaskin and Frumkin's paper,37 Parsons=' presented a more detailed treatment of the state of a solvent at an electrode interface in terms of a primitive four-state model. A similar multi state treatment was subsequently given by Bockris and Habib,43 but apparently without comparison with the previous42 work. Parsons42 took account more quantitatively of the number of molecules per square centimeter in the double layer and avoided some of the arbitrary assignments of parameters, discussed earlier, that were involved in the DamaskinFrumkin theory.37 Solvent molecules were assumed to exist in two states of aggregation at the surface of the electrode: free molecules and small clusters. The free molecules could adopt either of two extreme "up" or "down" orientations, as in the Watts-Tobin model,6.7 with a dipole moment fl. The clusters could also adopt one of two extreme orientations, depending on the orientation of their resultant perpendicular dipole moment, fle, with relation to the metal surface. fle was expressed per molecule ofthe cluster and a parameter p was defined as the ratio flei fl to characterize the state of polarization of the cluster. The solvent polarization was expressed in terms of Boltzmann orientation distribution functions, A+ and A_, for clusters and other terms for free dipoles. Inner-layer capacities were calculated as a function of a reduced charge parameter S and are shown in Figs. 7.3(a) and 7 .3(b) for the two values of the reduced dipole capacity K with A_ taken as 10-8. A series of curves was given for various A+ values, varying from 10-2 to 10-6 or 10-7. With A+ taken as 10-2 and A_ as 10-4, the general features of the experimental capacitance curve for aq. NaF at 0°C41 are quite closely reproduced. The calculated capacities C which are plotted in Figs. 7.3(a) and 7.3(b) are expressed in terms of the ratio C = C;lKo where C, is the inner-layer capacitance derivable from analysis of experimental results and Ko is an inner-layer capacity for a fixed orientation of dipoles. The physical significance of A+ must be mentioned because it determines the general shape of the capacitance vs. S curves [Figs. 7.3(a) and 7 .3(b)], e.g., in changing from a form similar to that for Hg [A+ = 10-4 - 10-5 in Fig. 7.3(a) and 7.3(b)] to that for Ga (A+ = 10-2). A+ is of course analogous to Damaskin and Frumkin's parameter K2; here it measures the energy associated (among other things42) with binding of the dipole to the metal. A large value of flb,+ (0end of H20 dipole adsorbed at the metal) displaces the sharply rising part of the







(0 )

-6 -4 -3-2






Calculated behavior of inner-layer capacitance as a function of reduced charge S 8 (after calculations and model ofParsons42): (a) with K = 0.02, P = 0.1, A_ 10- ; (b) with K = 0.08, 8 P = 0.1, A_ = 10- . The values of the log A+ parameter (see text) are indicated on the curves.


capacity curve to more positive S values where the field effect on orientation becomes dominant. In this work, the inner-layer capacity behavior is quite well accounted for by H20 dipole and cluster orientation. With solvents other than water, different types of inner-layer capacitance behavior can be distinguished, as shown schematically in Fig. 7.4. It is possible that type (ii) (Fig. 7.4) would manifest behavior like that of water if a wider range of positive qM could be explored. Type (iii) is, however, different and corresponds to weak association and binding to the surface. The transition from type (i) to type (iii) is qualitatively reproduced by the theoretical calculations as larger values of log A+ in Figs. 7.3(a) and 7.3(b) are taken. In certain solvents, two humps appear, one on either side of the pzc. In such cases, it seems that ion interaction in the double layer must be invoked" to explain at least one of the humps. Also, in water, the development of the hump depends very much on the nature of the electrolyte anion (Fig. 7.1), so the capacitance behavior cannot be treated only in terms of water dipole or cluster orientation without some allowance for specific solvent orientation effects caused by the adsorbed anions in addition to that caused by excess surface Charge, qM' Indeed, from the knowrr" specific influence of ions on the local state of bulk water, it would be very surprising if similar effects were not of major importance in the solvent layer in the interphase.


Chapter 7

The Double Layer at Electrode Interfaces


l i)

l i i)

( iii)


oqM +


FIGURE 7.4. Schematic representation of three principal types of inner-layer capacitance profiles as a function of qu for aqueous and nonaqueous solvents: (i) water type; (ii) ring-compound type, e.g., propylene carbonate, sulfolane; and (iii) humpless type, e.g., CH30H, HCOOH, NH3. (After Parsons.42)

An opposite position is taken in a paper by Bockris and Habib,32 who argue that the hump cannot arise from H20 dipole orientation effects. However, based on results by Hills and Payne'" on the temperature dependence of Hg/aq. solution interfacial tension, Reeves" showed that the maximum in the surface excess entropy (related to the hump) as a function of the double-layer charge qM due to the water molecule contribution occurs at potentials negative to the pzc. A similar conclusion was reached by Harrison, Randles, and Schiffrin.44 The results of Valette41 on single-crystal Ag (see Fig. 7.7 later) seem, however, to lend strong support to the original idea that the hump is mainly due to solvent dipole reorientation near or at the pzc.

evaluated25,28; also, the intermolecular stretching behavior can be calculated and has been characterized spectroscopically in bulk liquid water. 47The H20 dipole orientation in a field due to excess charge at a metal boundary interface would be deduced, in such a treatment, in terms of the bending and breaking of H bonds in a quasi-2-dimensionallattice rather than in terms of artificial limiting it, J,J, or i J, orientations and dimer interactions. Some indication that his approach may provide a more realistic treatment of solvent orientation in the inner layer at charged metal interfaces is afforded by the fact that Pople's analogous calculationsf for bulk water gave a good account of the bulk dielectric properties of water and its structural radial distribution functions at several temperatures. A related and more extensive treatment using this approach was that given by Guidelli_3°,45It seems to be the right direction for further advances in understanding the double-layer dielectric at electrode interphases. Two attempts (apart from the cluster approaches) have been made to take into account the H-bonded network structure of water in the double-layer interid I ' ; incid phase: one by Parsons and Reeves 29 and the other by G Ul e 11'3045 COInCIentally they appeared in the same issue of the journal. There has also been work on some related detail by Marshall and Conway." The types of laterally associated dipole arrays considered by Parsons and Reeves'? are shown in Fig. 7.5 while an element of the interphasial H-bonded water lattice envisaged by Guidelli30 is illustrated in Fig. 7.6. 7.5.7. Spontaneous Orientation of Water at Electrode Surfaces Due to Chemisorption Electrocapillary curves at Ga in aqueous solution show much more asymmetry than at Hg. This is, however, not due to stronger anion adsorption. Anii ons, such as 1-, are in fact less strongly adsorbed. 49 Thus, the extent 0 f Increase of capacitance on Ga as potential changes to more positive values (Fig. 7.2) cannot be accounted for quantitatively by increasing adsorption of anions. Stronger specific adsorption and orientation of water dipoles than at Hg is the explanation . ., h H 37I of the different properties of the double layer at G a In companson WIt g. t is clear that if the specific adsorption of water at metals depends on the chemical identity of the metal; then image interactions alone are not the only factor in 50 preferential orientation of water at Hg and other metals. Kemball has studied the related question of entropies of ad layers of water and methanol at the vaporlHg interface and given a detailed statistical-mechanical interpretation ofthe results in terms of mobilities of the adsorbate molecules. The general difference of adsorbability and specific orientation of H20 di36 poles in aqueous solution at various metals was noted by Trasatti34- in relation to the dependence of the potentials-of-zero charge of the metals on their elec-

7.5.S. H-Bonded Lattice Models The state of water in the double layer at electrode interfaces is best represented by a network of hydrogen-bonded molecules,29.30,45.46 local average the structure of which is modified by the potential-dependent interphasial field, leading to hydrogen-bond bending25,28 and partial orientation of the water dipoles. The two-state models cannot include such structural effects. Some of the artificial aspects of two-state orientation models, monomerdimer treatments, and cluster models would be eliminated if the orientation polarization in a two-dimensional interphase of water could be treated. In such a model, a lattice ofH-bonded water molecules would be considered and their orientation in the double-layer field would be calculated in terms of bending and breaking of the H bonds between H20 molecules laterally in the interphase, as well as those connecting such water molecules to the bulk. The potential energy functions for bending of H bonds between H20 molecules have been previously


Chapter 7

The Double Layer at Electrode Interfaces


tronic work functions. His conclusions provide further evidence of a different kind that water adsorption at metals is specific for the metal or families of metals.

7.S.S. Solvent Adsorption Capacitance at Solid Metals

The contribution of potential-dependent solvent orientation to the interfacial capacitance of solid metal electrodes is well demonstrated from work on well-ordered, single-crystal surfaces. Unlike polycrystalline solid metal interfaces, single-crystal surfaces exhibit a well-defined single pzc and hence welldefinable surface charge densities and interphasial fields as a function of displacement of electrode potentials from the pzc. A good example of double-layer studies on single-crystal metal surfaces is that by Valette" on Ag(llO) in aqueous NaF, NaCI04, and KPF6 electrolytes for which the order of relative strengths of specific adsorption is P- > CI04" > BF4 == PF6 ~ O. This order is approximately the inverse of that at Hg electrodes and is attributed to local adsorption on surface defects. Figure 7.7 shows the curve for the inner-layer differential capacitance for Ag(110) in 0.04 M aqueous KPF6. Data for three other concentrations down to 0.005 M were recorded at five frequencies. Minima in the curves of the overall
FIGURE 7.5. Laterally associated H20 dipole arrays in the solvent dipole layer at an electrode surface. (After Parsons and Reeves.i") 100

<.> 50 u,







qM If'C cm-



~ ___

FIGURE 7.6. As in Fig. 7.5, for an element of the H- bonded interphasial H20 lattice at an elec30 trode surface. (After Guidelli. )

FIGURE 7.7. Double-layer capacitance behavior of single-crystal Ag(1lO) as a function of potential around the pzc in aqueous 0.04 mol dm -3 KPF6 at 298 K. Reprinted from G. Valette, J. Electroana! Chern., 122, 285 (1981), with permission from Elsevier Science.



The Double Layer at Electrode Interfaces TABLE 7.1. Inner-Layer Differential Related Metal Au Hg TI(Ga)" Bi Sn Pb Cd In(Ga)" Zn Ga
Source: Reprinted from

153 Capacity Parameters


capacitance arise from the (reciprocal) contribution of the diffuse-layer capacitance in the usual way (Eq. 6.1 in Chapter 6). The PF(; anion is minimally adsorbed. When the diffuse-layer capacitance is factored out from the total series interphasial capacitance combination, the Helmholtz, compact-layer capacitance can be evaluated in the usual way and gives rise to the interesting curve of Fig. 7.7. The inner layer capacity, thus decoupled from the diffuse-layer capacity, shows a very well-defined maximum that is almost independent of (KPF6) concentration and is located close to the zero charge (qM = 0) condition (Fig. 7.7). At substantial positive or negative surface charge densities, the capacity tends to values of about 25 f1F cm ", which is rather larger than that for Hg, namely, about 16 f1F cm-2, which corresponds to an interphasial 10 of the water layer of about 5.7. Valette considered the structure of the Ag(110) surface in terms of the channeled arrangement of surface-Ag atoms in relation to possible structured arrangements of adsorbed water molecules. His analysis gave a good account of the observed capacity values at substantial (±) values of the surface charge density in terms of the model of Ref. 42. Appreciable self-orientation of water dipoles caused by the discontinuity of the water structure at the electrode wall and to the role of dipole image interactions with the metal is possible; in fact, some statistical-mechanical calculations'" do indicate a degree of icelike structuring of water molecules at a wall interface+' with the solvent water. The whole topic of states of gases and liquids in confined systems, such as fine pores, is attracting much current interest (e.g., see Refs. 53-60) and is relevant to double-layer behavior in high-area porous supercapacitor electrodes. This section is concluded with a summary (Table 7.1) of the inner-layer differential capacitance, Ci, of eight solid metals and two liquid alloys at their respective potentials-of-zero charge. Values of the respective inner-layer relative permittivities, e.; are given in the third column and are based on an assumed effective thickness of the inner layer of 0.31 nm. It is seen that both C, and e, are appreciably dependent on the chemical nature of the metal. This reflects the specificity of solvent adsorption at the metal coupled with the metal electron componentlO-12 (see Section 7.6) of the double-layer capacitance and the free electron density in the metal. II

for MetallW

ater Interfaces


C, I flF em 23 28 34 36 39 40 52 60 100 135


8.0 9.8 11.6 12.6

14.0 18.2 21.0 35.0 47.3
from Elsevier


Trasatti, 1. Electroanal

Chem., 123, 121 (1981), with permission

Science. "Inner-layer differential capacity at the potential-of-zero charge. bInner-layer permittivity derived from C, with the inner-layer thickness taken as 0.31 nm. 'Liquid alloys.

Hg-aqueous solutions, albeit on a semiempirical basis. Some more recent papers have approached this problem by means of advanced statistical-mechanical and computational simulation procedures. Several of these papers are summarized below.

" n


Recent Modeling Calculations

We described in earlier sections of this chapter how relatively modern treatments of the double-layer capacitance of electrode interfaces have rather successfully established the role of solvent-dipole orientation and interphasial solvent structure in determining the charge-dependent capacitance. Relatively simple models have given good representations of the capacitance behavior of

FIGURE 7.8. lcelike ordered H-bonded water structure near an electrode surface. The two inclinations of the dipoles in the contact layer with respect to the surface normal fI are distinguished by the horizontal and vertical shadings. For an uncharged surface they are energetically equivalent. (From Torrie and Patey.51)


Chapter 7

The Double Layer at Electrode Interfaces





180 ·0




p (ell' r)

0 N



Probability density p(8,r) for molecular orientations as a function of distance r from contact with a neutral surface for a fluid of water molecules with tetrahedral symmetry. All distances are in units of the solvent diameter. ds = 2.8 A. (a) OOH, OH-bond orientation; (b) Op., dipolar orientation. (From Torrie and Patey51)

FIGURE 7.10.

As in Fig. 7.9 but for a surface charge density of -13 j.lC cm-2. Note the more intense orientations. (From Torrie and Patey.51)


Chapter 7

The Double Layer at Electrode Interfaces


In an interesting paper by Brodsky, Watanabe, and Reinhardt/" molecular dynamics simulation calculations of the state of water structure at and near electrode surfaces indicated that liquid-crystal structures arose between the second and third water molecule layers, and the first and fourth layers. Laterally ord~red structures across the electrode surface, with periodicities of ca. 0.27 nm (i.e., near the molecular diameter of H20), were also found. The ordering normal to the electrode surface corresponds qualitatively to the experimental results recently obtained from interfacial X-ray scattering studies. . Feldman and Partenskii62 considered the reorientation of molecular dipoles coupled with relaxation of the gap in the interphasial double-layer capacitor model as responses to electric polarization (a kind of electrostriction effect'") in their study of an elastically bonded molecular capacitor. A relationship to experimentally known capacitance behaviors, however, was not clearly demonstrated. Another paper by Holovko, Pizio, and Halytch'" also considered the dipole orientation distribution and the effects of ions upon it. . Torrie and Patey" applied an ambitious series of statistical-mecham~a1 calculations, based on the so-called "reference hypernetted-chain" (RHNC) mtegral equation approximation, to derive a whol~y mole~ular theory ~f t~e solution side of the double layer in aqueous solutions usmg modern liquid-state theory. The influence of the electrode surface, as a wall, on the ~ocal structure of water was determined, together with the influence of the resultmg restructuring of the solvent on ion distribution; the effect of net surface charg~ o~ the solvent restructuring effect was also considered. As in other work, an Icelike local structure of water at the uncharged interface was predicted (Fig. 7.8). Probability densities for OH bond direction and the dipole orientation of water molecules at an uncharged wall surface were calculated, as exemplified in Fig. 7.9. Comparative results were obtained for a charged interface bearing a ~har~e density of -1311C cm-2 (Fig. 7.10). Significant differences can be seen m Figs. ~.9 and 7.10. Related calc~lations f~r an ion + dipo~e mixture at \~ard wall With specific dipole adsorption were given by Outhwaite and Molero. Reference 64 is a general symposium covering a range of topics involving these and other subjects.



metal (approximately one per atom) allow a finite probability of finding electron density outside the metal to an extent that diminishes exponentially outward from the metal surface. Qualitatively, the electron spillover effect was recognized a long time ago (e.g., Fig. 2 in Ref. 65) in relation to the intrinsic surface potential, X, of metals. In the case of an electrode, this spillover effect depends on the electrochemical polarization of the metal so that at potentials positive to the pzc, this spillover electron density is drawn in toward the metal while at negative potentials it is expelled outward to a corresponding extent. The variation may be over 1-2 A (0.1-0.2 nm) or so. This region over which there is an interfacial gradient of electron density is referred to as the Thomas-Fermi screening distance (see Fig. 4 in Chapter 6). Development of this concept of a spillover electron density was initially due to Lang and Kohn,IO,66but some earlier treatment is attributed to Rice.67 More recently, the spillover effect has been examined in more detail by Badiali and co_workersI2.68,69 and by Schmickler.U'"" especially as it may influence double-layer properties and the capacitance. The treatments employ the jellium model for the behavior of the delocalized electron plasma in or near the metal surface, taking into account the solid atom nuclei array as a diffuse positive background, a "jellium." As there is a variation of spillover electron density with electrode potential around the pzc, a corresponding capacitance contribution, CM' formally arises, which is referred to as the "metal contribution" to the double-layer capacitance. It adds as a series contribution to the capacitance, being reciprocally additive to the other contributions (compact layer, diffuse layer, and solvent polarization). Hence, only when it is small does it lead to a significant contribution to the overall C. Insofar as this metal contribution was not included in the classical conventional treatments of the double layer, taking account of it will introduce some modification to the evaluation of the other components. For example, conventionally, the compact (Helmholtz) layer capacitance contribution, CH, is usually evaluated from the overall Cdt value (at a given potential) (actually as IICdD by subtracting a calculated value of the reciprocal diffuse-layer capacitance, since the diffuse-layer capacitance theory is regarded as reliable at least for dilute solutions up to about 0.1 M. If an additional IIC M term should be included, obviously then some changes in IICH will result, depending how small CM is. In fact CM is really part of CH' This complication has been pointed out and emphasized by Badiali.t"


Origin of the Metal Contribution 7.6.2. Profile of Electron Density at Electrode Surfaces

It has been realized for some time that at the metal/vacuum interface there is a spillover of electron density just outside the nominal atomic surface of a metal. This arises because the wave functions of the delocalized electrons of the

Following the work by Lang and Kohn,10,66we show schematically in Fig. 7.11 the profiles of electron density at an electrode interface at three potentials.


Chapter 7

The Double Layer at Electrode



Bulk electron density

'- -to


H20 dipoles (or quadrupoles). For example, on the negative side, electron spillover can influence shielding of the protons of oriented H20 molecules at the metal surface. Both these effects can have an influence on the interphasial capacitance in a rather complex, interactive way, and can also affect the specific adsorption of ions, especially for anions, as noted earlier. In the treatment of the surface region of metals from the point of view of the electron-density distribution, the so-called "jellium model" is commonly used. In this model, the discrete nature of the semi-infinite array of ion cores of the crystal structure of the metal near its interface is ignored and is replaced by a uniform, structureless background of positive charge, ending abruptly at the metal's nominal surface (Fig. 7.12). This charge is equal and opposite to that of the free-electron gas (approximately one valence electron per atom). Owing to the structureless nature of the jellium, the properties of the electron-jellium sys-



l_Nominal L____ surface

metal plane


of centers of meta I surface atoms



FIGURE 7.11. Profiles of electron density at the surface of an electrode at three potentials corresponding to: (I) positive surface charge density, (2) zero charge density at the pzc, and (3) negative charge density. (Schematic; repeated from Chapter 6.) 0.5

It is seen that at potentials negative to the pzc there is a greater penetration of electron spillover toward the solution and adsorbed ions in the Helmholtz layer region. This can influence (1) chemisorption of these ions, especially anions; and (2) the orientation, and orientation distribution, of solvent molecules, e.g.,



drop 41TO"D+Xm(O)

Ideal metal potential drop 47TOD

eopoeifo nee 1 Ie = 47TD


lie: 47T(D-Xo)





-5 z/ou FIGURE 7.13. Calculated electron-density profiles n(z)/n for (a) rs = 6 and (b) r, = 2 in the jellium model. (-) from Smith,71 (---) from Lang and KohnlO In (a), the dotted line shows the effect of second gradient terms. (After Amorkrane and Badiali.69)

FIGURE 7.12. lellium model of charge distribution at a metal interface: left, microscopic view of the metal interface; right, representation of the metal interface by a charged plane located at Zo. D is the position of an ideal charged plane. (After Amorkrane and Badiali.69)


Chapter 7

The Double Layer at Electrode Interfaces


tem are invariant parallel to the metal interface and vary only normal to it, making a substantial mathematical and modeling simplification. A treatment of this model in an approximate way was first given by Smith7l in 1969 and a more exact one was published in a now well-known paper by Lang and Kohnlo in 1970. In Fig. 7.13, we show the electron-density profile for the jellium model n(z)ln (along the normal coordinate z) for rs 6 and for rs = 2 according to the calculations of Smith 71 and Lang and Kohn.1O The parameter rs is defined by rs = (3/4nn)l/3 where n is the bulk free-electron density, which varies from metal to metal; in fact, rs is the radius of the sphere containing on the average one electron. The significance of the surface-region variation of electron density for the structure of the electrode/solution interphase, in particular that component associated with the surface region of the metal itself, has been explained in the referenced papers and in works by Amokrane and Badiali, as reviewed in Ref. 69.

responding to the inner limit of the diffuse layer for the case of finite-sized ions (Stern-Grahame model'-"). The overall result of a rather complex calculation 73is that 'fI as a function of distance r into the solution and of concentration is given by the complex expression: cosh[u'fIl12]1



cosh[u'fII2]-1 cosh[u'fIl2]

cosh[u'fIll2] + 1



7.7. THE POTENTIAL PROFILE ACROSS THE DIFFUSE LAVER The usual mathematical treatment'r' of the double layer aims to derive a function relating the electrode surface charge density qM to the electrode potential, hence enabling the diffuse-layer capacitance to be calculated (Eq. 7.13). In another direction, however, it is of interest to derive how the potential profile, 'fin in the solution varies along the direction away from the electrode/solution boundary toward the solution. This calculation is relevant to derivation of, e.g., the profile of potential distribution normal to the surface of a pore72 in a porous carbon capacitor electrode. This variation of 'fin declining outward in the bulk solution (eventually becoming equal to the bulk solution potential v., i.e., 'fIr'fIs ~ 0 as r ~ 00, serves to define an effective thickness of the diffuse layer. Of course, formally, the diffuse layer has an infinite thickness since 'fir - 'fIs does not really decline to zero until r ~ 00. However, a practical definition can be made for this thickness taking a limit, say, of x for 'fir - 'fIs:::: 10 mY, for dilute solutions. The calculation of the profile of 'fir from the Poisson-Boltzmann equation is more complicated than that for evaluation of Cdiff, but a solution has been obtained by Eversole and Lahr73 in a significant but little-quoted paper, and was numerically evaluated for various conditions by Conway.!" It was noted earlier that the profile of the fall of potential normal to a charged electrode surface in the diffuse layer gives an idea of the extent of penetration of the diffuse layer into the bulk of the electrolyte solution. This extent of penetration depends on (1) the ionic strength or the extent of charge screening by the ions of the electrolyte and (2) the potential of the electrode itself relative to the solution at infinity and thus to the 'fIl potential at the Helmholtz plane cor-

where u = ze/kT and 'fIl is again the potential of the inner limit of the diffuse layer at the Helmholtz plane. The interesting point about this relation is that the final function for 'fI(r), which appears as the variable 'fI/2, is equal to the product 2(Kr), i.e., K and r scale together in determining 'fI(r), so that values of r (proportional to the square root of concentration) are larger for smaller values of K (nl/2) for a given value of 'fir relative to 'fl. The first term on the right-hand side of Eq. (7.32) for 2Kr is a constant for given electrode polarization and solution conditions. The above calculation also allows calculation of the zeta potential (~) associated with the distance over which the diffuse layer is mobile with respect to the Helmholtz plane in the case of a moving electrode surface or a colloidal particle subject to electrophoresis. The result is also significant for porous doublelayer capacitor electrodes in relation to electro-osmotic flow of electrolyte upon discharge or recharge. Several years after Eversole and Lahr's paper, Grahame, in his review, 1 gave an approximate expression for the 'fI potential as a function of distance across the double layer, a result that is useful for restricted conditions.


The double-layer type of electrochemical capacitor is constructed from high specific-area materials, usually carbon. With 1000 to 2000 m2g-l carbons, the pore structure is very fine so that the interstitial regions of the contained liquid electrolyte extend only over very short distances and much of the electrolyte resides interphasially near (within 0.5-1.0 nm) the surfaces of such pores (Chapter 14). Under these conditions, the state of the double layer, especially its diffuse region,72 becomes of particular interest. It was explained earlier that the diffuse layer is to be regarded as the "ionic atmosphere" conjugate to the charge-density difference between the electrode surface (qM) and the compact layer (q,). This ionic atmosphere extends out into the bulk solution to distances determined by


Chapter 7

The Double Layer at Electrode Interfaces


the electrolyte concentration or ionic strength (cf. Ref. 73 and Section 7.7). In the cases of ions in the electrolyte itself, the ionic atmosphere charge distribution, conjugate to a central reference ion's charge, extends effectively to a 3-dimensional distance that is reciprocal in the square root of the ionic strength according to the Debye-Hiickel theory.74,75This relation applies to dilute solutions, but for strong solutions (ca. 5 M or more), such as those employed in capacitors, the concept of an ionic atmosphere breaks down, as it does also for the diffuse-layer ion distribution at electrode interfaces. Nevertheless, a rough guide is that the diffuse layer at electrodes effectively extends about 100 nm into the solution, i.e., for the If! potential to fall to within a few millivolts of the solution potential, ¢Js, for a 0.001 M solution of a uni-univalent electrolyte, about 10 nm for 0.01 M and I nm approximately for a 0.1 M solution. For stronger solutions (> 1 M), a rather different model consisting of a quasi-lattice of anions, cations, and solvent molecules near the electrode interface over distances ::;1 nm has to be adopted 15and a diffuse layer, in the original sense, ceases to exist. At an electrode interface distributed within a microporous structure, the configuration of the distribution of electrode potential into the solution phase is much more complex than at an infinite planar electrode.F The problem arises on account of the thickness.i'' i.e., the Debye length, (K-I) of the diffuse part of the double layer in relation to the width or space available for the ion distribution to be set up within the pores of the electrode. In the original Debyc-Huckel theory,74.75 the Poisson-Boltzmann equation relating the second-derivative operator, V21f!, to the ionic space charge density became transformed to a second-order differential equation, V21f! = K21f!, where K2 involves the ionic strength I of the solution.i" i.e., K is square-root in I. It was also shown that K-I had the dimensions of distance, namely, the effective radius of the ionic atmosphere distribution around any given ion. A similar significance applies to the effective thickness of the diffuse-layer ionic atmosphere in double-layer theory, which is inversely related to a function of the ionic strength of the electrolyte. From this it can be appreciated that in very fine pores of a porous electrode matrix, say 2-10 nm in extent (diameter), the diffuse region of the double layer in dilute electrolyte solutions could extend all the way across the pore diameter and overlap'? with the diffuse layer ion distribution at the other side of the pore (Fig. 7.14). This situation then becomes similar to that envisaged and mathematically treated by Verwey and Overbeeck in their monograph (see general reading reference 1) on the theory of the stability oflyophobic colloids, and the fall of potential with distance in the diffuse layer73 between the inner boundaries of the double layer is then less than for free, noninterfering interfaces. The ion distributions then overlap, so that the charge density and potential equations become much more complexf than those given in the Gouy-Chapman theory.

_J <[ t-

Z l.LI

b o,

FIGURE 7.14. Overlap of diffuse double-layer potential profiles (and corresponding butions) in an element of a pore in a porous electrode matrix (schematic).

ion distri-

In a real electrochemical capacitor, however, operating with acidic or alkaline electrolytes at several molar or higher concentrations, this situation of overlapping diffuse layers will rarely arise. The net charge density on the pore surfaces (e.g., of carbon particles) will be already almost fully screened by the compact layer counterion charges and residually by a substantially collapsed diffuse layer only -0.5 nm in thickness. Then the overlap situation that would arise in dilute electrolytes will no longer apply. Another situation of a different kind may arise, however, on charging of high-area double-layer capacitors: depending on the electrolyte concentration and its bulk amount relative to the weight and area of the carbon powder matrix electrode, significant differential depletion of the electrolyte concentration can occur because of the migration and electrostatic accumulation of cations at the interface of the negatively charged electrode and of anions at the conjugate positively charged electrode. Then the bulk cation plus anion electrolyte concentration can be appreciably lowered; for sufficiently diminished bulk region concentration, the diffuse layer will expand its penetration into the bulk and some overlap effect might then become established between the sides of pores. Similarly, the bulk conductivity of the electrolyte could become significantly diminished on charge, leading to enhancement of the internal resistance of the capacitor device, a matter of practical significance and concern for the maintenance of optimum operating power. Such problems can be of greater significance in the case of nonaqueous electrolyte solutions in higher voltage capacitors, and have been experimentally observed. The situation in a double-layer capacitor is rather different from that arising at porous electrocatalytic electrodes (e.g., Raney Ni); there, in fine pores the


Chapter 7

The Double Layer at Electrode Interfaces


electrochemical evolution of gas (e.g., H2) on polarization displaces electrolyte out of some fraction of the pores so the active area accessible for electrocatalytic processes is made appreciably smaller; Wendt77 recently estimated that in fact only 10 percent is available. In a porous electrochemical capacitor structure, this effect does not occur unless the device is adventitiously overcharged beyond the electrochemical decomposition potential difference for the electrolyte, giving rise to production of gases in the pores. This effect may be irreversible, i.e., any gas produced in pores will not easily disappear.78-8o A different but related kind of problem arises when electrolyte is admitted to a porous carbon capacitor material: unless the electrolyte can invade a previously evacuated structure in the porous matrix the finest pore fraction will not become wetted with electrolyte. Since it is the finest pore fraction that accounts for much of the real area of the carbon matrix, capacitance may be lost unless effective invasion of the fine-pore fraction of the pore distribution is achieved. However, as noted in Chapter 14, such fine pores may be electrochemically poorly accessible owing to cumulative electrolyte resistance. At the level of the fine-pore fraction of the real-area distribution, solvent occupancy of pores may be achieved by distillative sorption into the finest pores. This aspect of sorption of liquidlike films into pores has been considered by EveretC8,79 and by Everett and Nordo.i" Sorption occurs in a quasi-2-dimensional film along the surfaces of pores until they become filled; desorption takes place from the meniscus of the resulting liquidlike phase within the pores, resulting in hysteresis between the sorption and desorption processes into and out of the pores, respectively. These authors described an ingenious magneticswitch model for the origin of hysteresis between sorption and desorption processes. It represented the essential feature of hysteresis in a process, i.e., its forward direction is not microscopically the reverse of the backward direction of the process. The pore-size distribution has been measured for a variety of porous carbon powder or aerogel carbon materials and is a property important for optimization of double-layer, carbon-type capacitor materials. Horovitz'" plotted the cumulative areas for pores with effective diameters greater than 1.5 nm against pore width for a series of proprietary high-area carbon materials (Fig. 7.15). Generally, after some sharply rising specific area (m2g-I), a plateau is reached with increasing pore width. The corresponding double-layer capacitances in F g-l of the indicated materials are shown in the annotations in Fig. 7.15. For these very finely porous materials, the classical electrocapillary effects of colloid science82 tend to occur. In conclusion, readers are referred to Rangarajan's comprehensive review (dated 1980) on properties and modeling of the double layer.83It critically and analytically covers a number of the matters treated in this chapter, but in greater

A-222S BETl786, 36 (144)F/g



<l) I-<

e::- 200
A-2200 BET 1556, 37(148)F/g


~ 150
<l) I-<


::s U 50


PF-36 BET 551, 58 (232) FIg

o~~~~~~~~~~~~~~ o 10 20 30 40 50 60



90 100

Pore Width C~)

FIGURE 7.15. Cumulative pore surface area in a porous electrode matrix as a function of pore width. Annotations refer to proprietory porous carbon preparations. (After Horovitz.t')

depth and detail than is required here to provide the basis for understanding the origins of double-layer capacitance of supercapacitor devices.

I. D. C. Grahame, Chern. Rev., 41, 441 (1947).

2. G. Gouy, 1. Phys., 9, 457 (1910).


Chapter 7

The Double Layer at Electrode Interfaces


3. D. Chapman, Phil. Mag., 25, 475 (1913). 4. H. von Helmholtz, Monats. Preuss. Acad. Sci., Nov., p. 431 (1881); see also Wied. Ann., 7, 337 (1879); Ann. Phys. (Leipzig), 89, 211 (1853). 5. O. Stem, Zeit. Elektrochem., 30,508 (1924). 6. N. F. Mott and R. J. Watts-Tobin, Electrochim. Acta, 4, 79 (1961). 7. R. 1. Watts-Tobin, Phil. Mag., 6,133 (1961); 8, 333 (1963). 8. J. O'M. Bockris, M. A. V. Devanathan, and K. Miiller. Proc. Roy. Soc .. Lond., A274, 55 (1963). 9. S. Levine, G. M. Bell, and A. L. Smith, 1. Phys. Chem., 73, 3534 (1969). 10. N. D. Lang and W. Kohn, Phys. Rev., Bl, 4555 (1970); B3, 1215 (1971). 11. W. Schmickler, J. Electroanal. Chem., 176, 383 (1984). 12. S. Amorkrane and J. P. Badiali, 1. Electroanal. Chem., 266, 21 (1989). 13. B. E. Conway, J. O'M. Bockris, and I. R. Ammar, Trans. Faraday Soc., 47, 756 (1951). 14. B. E. Conway, Ph.D. thesis, University of London (1949). 15. R. Parsons, in Modem Aspects of Electrochemistry, J. O'M. Bockris and B. E. Conway, eds., vol. 1, Chapter 3, Butterworths, London (1954). 16. D. C. Grahame, J. Chem. Phys., 18, 903 (1950). 17. 18. 19. 20. J. R. Macdonald, J. Chem. Phys., 22,1857 (1954). A. N. Frumkin, Zeit. Phys. Chem., 103, 55 (1923). 1. A. V. Butler, Proc. Roy. Soc., Lond., A122, 399 (1929). G. 1. Hills and R. Payne, Trans. Faraday Soc., 61, 316 (1965).

47. B. E. Conway, Ionic Hydration in Chemistry and Biophysics, Elsevier, Amsterdam (1981). 48. S. Marshall and B. E. Conway, J. Electroanal. Chem. J. Chem. Phys., 82, 923 (1984). 49. A. N. Frumkin and I. Bagtozkaya, Electrochim. Acta, 10,793 (1965). 50. C. Kemball, Proc. Roy. Soc., Lond., A188, 117 (1947). 51. G. M. Torrie and G. N. Patey, Electrochim. Acta, 36, 1677 (1991). 52. C. W. Outhwaite and R. Molero, Electrochim. Acta, 36,1685 (1991). 53. 1. Klafter and 1. M. Drake, Molecular Dynamics in Restricted Geometries, Wiley, New York (1989). 54. D. Richter, A. 1. Dianoux, W. Petry, and 1. Teixeira, in Dynamics in Disordered Materials, Springer Proceedings in Physics, Vol. 38, Springer-Verlag, Berlin (1989). 55. K. L. Ngai and G. B. Wright, Relaxation in Complex Systems, North-Holland, (1991). Amsterdam

56. J. M. Drake, 1. Klafter, R. Kopelman, and D. D. Awschalom, eds., Dynamics in Small Confining Systems I, Mat. Res. Soc. Symp. Proc., p. 290 (1993), publ. Material Res. Soc., Warrendale, PA. 57. 1. M. Drake, 1. Klafier, R. Kopelman, and S. M. Troian, eds., Dynamics in Small Confining Systems II, Mat. Res. Soc. Symp. Proc., p. 366 (1995), publ. Materials Res. Soc., Warrendale, PA. 58. S. Stapf, R. Kimmich, and R. O. Seitter, Phys. Rev. Lett., 75,2855 59. G. Liu, Y. Li, and J. Jonas, J. Chem. Phys., 95, 6892 (1991). 60. G. P. Crawford, R. Stannarius, and 1. W. Doane, Phys. Rev., A44, 2558 (1991). 61. A. M. Brodsky, M. Watanabe, and W. P. Reinhardt, Electrochim. Acta, 36, 1695 (1991). 62. V. Feldman and M. Partenskii, Electrochim. Acta, 36,1703 (1991). 36,1715 (1991). Interface, 63. M. F. Holovko, O. O. Pizio, and Z. B. Halytch, ElectrochimActa, (1995).

21. R. Payne, J. Electroanal. Chem., 15,95 (1967). 22. R. Payne, J. Phys. Chem., 70, 204 (1966). 23. R. Payne, in Advances in Electrochemistry and Electrochemical Engineering, P. Delahay and C. W. Tobias, eds., Chapter 2, Intcrscicnce, New York (1970). 24. B. E. Conway, Ionic Hydration in Chemistry and Biophysics, Elsevier, New York (1981). 25. 1. A. Pople, Proc. Roy. Soc., Lond., A205, 163 (1951). 26. R. W. Reeves, in Modern Aspects of Electrochemistry, J. O'M. Bockris and B. E. Conway, eds., vol. 9, Chapter 4, Plenum, New York (1974). 27. 28. 29. 30. F. B. R. R. Booth, J. Chem. Phys., 19, 391; 327 and 1451 (1951). E. Conway, Can. J. Chem., 37, 613 (1959). Parsons and R. M. Reeves, 1. Electroanal. Chem., 123,141 (1981). Guidelli, J. Electroanal. Chem., 123, 59 (1981).

64. D. 1. Henderson and O. R. Melroy, eds., The Structure of the Electrified Electrochim. Acta, special symposium volume, 36, pp. 1657-1889 (1991).

65. B. E. Conway, in Theory and Principles of Electrode Processes, Chapter 2, Ronald Press, New York (1964). 66. N. D. Lang and W. Kohn, Phys. Rev., B8, 6010 (1973); B7, 3541 (1973). 67. O. K. Rice, Phys. Rev., 31,1051 (1928). 68. J. P. Badiali, Electrochim. Acta, 31, 149 (1986). 69. A. Amokrane and J. P. Badiali, in Modern Aspects of Electrochemistry, J. O'M. Bockris, B. E. Conway, and R. E. White, eds., vol. 22, Chapter I, Plenum, New York (1992). 70. W. Schmickler, J. Electroanal. Chem., 100, 533 (1979). 71. J. R. Smith, Phys. Rev., 181, 522 (1969). 72. R. Farina and K. Oldham, J. Electroanal. Chem., 81, 21 (1977). 73. B. Eversole and G. Lahr, J. Chem. Phys., 9, 798 (1941). 74. P. Debye and E. Hiickel, Phys. Zeit., 24,185 (1923). 75. P. Debye and E. Huckel, Phys. Zeit., 24, 305 (1923). 76. R. A. Robinson and R. H. Stokes, in Electrolyte Solutions, Chapter 4, Butterworths, (1955). 77. H. Wendt, private communication (1997). (1955). and Similar Boca Raton, 78. D. H. Everett, Trans. Faraday Soc., 50, 187 (1954). 79. D. H. Everett, Trans. Faraday Soc., 51,1551 80. D. H. Everett and P. Nordon, Proc. Roy. Soc., Lond., A259, 351 (1960). 81. Z. Horovitz, in Proc. Fifth Inti. Symposium on Double-layer Capacitors Devices, S. P. Wolsky and N. Marincic, eds., Florida Educational Seminars, Fla. (1995). London

31. J. Topping, Proc. Roy. Soc., Land., A114, 67 (1927). 32. J. O'M. Bockris and M. Habib, J. Electrochem. Soc., 24,123 (1976). 33. R. W. Fawcett, J. Phys. Chem., 82, 1385 (1978). 34. S. Trasatti, J. Chem. Soc., Faraday Trans., 1, 68 (1972). 35. S. Trasatti, J. Electroanal. Chem., 39,163 (1972); 44, 367 (1973). 36. S. Trasatti, Surface Sci., 32, 735 (1972). 37. A. N. Frumkin and B. B. Darnaskin, Electrochim. Acta, 19,173 (1974). 38. B. B. Darnaskin, Elektrokhimiya, 1, 63 (1965). 39. B. E. Conway, H. P. Dhar, and K. M. Joshi, Electrochim. Acta, 18,789 (1973); see also B. E. Conway and H. P. Dhar, Croatica Chem. Acta, 45,173 (1973). 40. B. E. Conway, E. Gileadi, and M. Dzieciuch, Electrochim. Acta, 8, 143 (1963). 41. G. Valette, J. Electroanal, Chem., 122, 285 (1981). 42. R. Parsons, J. Electroanal. Chem., 59, 229 (1975). 43. 1. O'M. Bockris and M. Habib, J. Electroanal. Chem., 65, 473 (1975). 44. 1. A. Harrison, 1. E. B. Randles, and D. 1. Schiffrin, J. Electroanal. 45. R. Guidelli, J. Electroanal. Chem., 197,77; 103 (1986). 46. 1. D. Bernal and R. H. Fowler, J. Chem. Phys., 1, 515 (1933). Chem., 48, 359 (1973).

82. N. K. Adam, The Physics and Chemistry of Surfaces, 3rd ed., Oxford University Press, Oxford (1941).