‘:.::.‘:I J b ‘. - u :23 .v" v! r- 7‘.- :35" 3‘ ' “:1 .'...'..:.:Lr:.r ‘ , I I (H I o-‘.a u o tdol n; . - _‘ nu V. a . ¢ 0 ¢. v' .u a u v . 1 fl- v. _.r. .l - . -3313... I. ‘ ‘0 ‘3 n o....' v' -. . -1 31:” (fl ' '1. ’- 61mg} ‘34 ' .7]-’.’ l'-. . ~ 7’ Wyn-‘27.: Q? L IVP~"- A" .- w- r- ...- .. {- -,_._, _. ....... .- . 3:?! ...... x, . . q...=(‘ 4— .a I . L.‘ .. .27.? w... —.. y-- 0-3. -m-oucwom ’M v var-7‘.“- v .o 31 .55 0% s. v- 4 a - ob I‘d'p‘ 0 -~.. .o—q..a'..-¢ < ‘7 " .- X W; IR EQ; ' m . - 4 . - I ,. o a.“ : E32“ . ‘3; 1 1 '11‘141-‘11'15 f 1111‘: ‘41 -3. ms— . n. ‘ v .3 -.. . .... . ‘!'"" .w .....Iz.m ' ...-_‘ . . § .. ml". w 3.“. H-I "a. . ' o JI‘ ‘ _... . 'Jo’I- u m w . _ 1.. 7" .3?! . I =;-‘ ! .:- 1“" 1n!!!” ' ‘ bl \ ' JI‘H II: ' _ " - , 3 1 91:»): ‘ :2? 5 t . , '. W1 ..:! " . 43-"- , O l ‘l I, ‘ ‘ J l V F) I . qP’IIp~-“ .1". I" itt ~12 i . i“. 1' 311G111 “I“? I; .11iI-I -mm'f’. {I ‘:. . ‘ 'i I". , , i 3%.; i' ! THESIS LIBRARY Michigan State U . . “'1 1”“ my This is to certify that the dissertation entitled ELECTRODE KINETICS AND DOUBLE LAYER STRUCTURE AT PLATINUM AND GOLD ELECTRODES presented by STEPHEN WAYNE BARR has been accepted towards fulfillment of the requirements for Ph.D. Chemistry 4,1va5 degree in D... 66%; Mm MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 OVERDUE FINES: 25¢ per cm per item W Nix .‘ Place in book return to remove “I", '4' charge fro-I circulation records I Izaak: I ELECTRODE KINETICS AND DOUBLE LAYER STRUCTURE AT PLATINUM AND GOLD ELECTRODES By Stephen Wayne Barr A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1981 ABSTRACT ELECTRODE KINETICS AND DOUBLE LAYER STRUCTURE AT PLATINUM AND GOLD ELECTRODES By Stephen Wayne Barr The interaction between the reactant and the elec- trode surface is explored for simple heterogeneous elec- tron transfer reactions at platinum and gold electrodes in aqueous solutions. The electroreductions of cobalt(III) ammine complexes were of primary interest. The ligands in the primary coordination sphere for these complexes could be varied and the effect of various ligands on re- action mechanisms or electrocatalysis could be determined. The reactivity of some complexes was employed to deduce the nature of the interphasial environment at the electrode surface. The heterogeneous electron transfer rates were determined by rotating disk voltammetry or cyclic voltam- metry at platinum and gold electrodes, and by d.c. polarography at mercury electrodes. Reactant adsorption was determined by fast cyclic voltammetry, by reduction Stephen Wayne Barr rate shifts in probe reactions sensitive to surface charge, and by measurements of the overall reaction order for electron transfer. Chronocoulometry was used for adsorp- tion measurements at mercury electrodes. Electron transfer rates and reactant adsorption were also found for iodide-treated platinum and gold electrodes. Adsorbed iodide was found to block reactant adsorption at these surfaces. The iodide layer was found to have only a small effect upon the double layer charge which indicates that the metal-iodide bond is fairly covalent. The iodide-treated electrode results were instrumental in determining the degree of catalysis caused by reactant adsorption. Several types of ligands were found to promote the ad- sorption of the cobalt(III) complexes. Among these are the chloro-, bromo-, isothiocyanato-, and pyridyl ligands. Significant catalysis of the electroreduction rates was noted for the chloro- and bromo- complexes. This catalysis was attributed to a lowering of the intrinsic barrier to electron transfer. It was also proposed that the intrinsic barrier to electron transfer was increased by certain ligands such as isothiocyanate. ACKNOWLEDGMENTS Several people deserve a share of the credit for the execution and completion of this research. Dr. Weaver's abundant suggestions and assistance kept the author from becoming hopelessly lost amidst a virtual quagmire of electrochemistry. The support and interest of the Weaver Group was greatly appreciated. Visiwanathan Srinivasan and Dr. David Larkin performed several mercury experiments included herein. Ken Guyer's work with silver electrodes offered many stimulating discussions when contrasted with platinum and gold, which often led to useful ideas for further work. Much of this research was supported by the Air Force Office of Scientific Research, which I gratefully acknowledge. The Chemistry Department may be credited with providing the atmosphere and financial support conducive to research, that made this work possible. I wish to thank Margy and Peri-Anne for their invaluable typing skills. Thanks are also due to my parents and family for their constant support and confidence which made all of this possible. ii Chapter LIST OF TABLE OF CONTENTS TABLES. LIST OF FIGURES INTRODUCTION. CHAPTER I - BACKGROUND AND HISTORICAL CHAPTER A. B. ASPECTS . . . . . Heterogeneous Electron Transfer Kinetics at Platinum and Gold Surfaces. Adsorption on Platinum and Gold in Aqueous Solution . . . Related Kinetics and Adsorption at Mercury Electrodes . . II - EXPERIMENTAL General Apparatus Solution Preparation. Preparation of Electroactive Complexes . . . . . . . . Electrochemical Techniques. . . . . . 1. Cyclic Voltammetry. 2 D.C. and Pulse Polarography 3. Fast-Cyclic Voltammetry A Rotating-Disk Voltammetry Solid Electrode Pretreatments 1. Platinum Electrodes 2. Gold Electrodes iii Page vi . viii IO IA l6 18 18 2O 23 2A 2A 25 25 26 27 29 3O Chapter Page 3. Iodide Treated Electrodes . . . . . . . 30 CHAPTER III - DIRECT MEASUREMENT OF REACTANT ADSORPTION . . . . . . . . . 32 A. Single-step Chronocoulometry. . . . . . . . 3A 1. Adsorption Isotherm for Cr(NCS)g- at Mercury. . . . . . . . . . 37 2. Single—step Chronocoulometry at Platinum and Gold. . . . . . . . . . A0 3. Staircase Chronocoulometry. . . . . . . AS A. Capacitance—related Errors in Single-step Chronocoulometry. . . . . . 51 B. Determination of Reactant Adsorption by Fast Cyclic Voltammetry. . . . . . . . . 57 1. Introduction. . . . . . . . . . . . . . 57 2. Adsorption of Transition Metal Complexes Determined by Fast Cyclic Voltammetry. . . . . . . . . . . 63 CHAPTER IV - KINETIC PROBES OF COMPLEX ION ADSORPTION . . . . . . . . . . . . 67 A. Introduction. . . . . . . . . . . . . . . . 67 B. Results . . . . . . . . . . . . . . . . . . 69 C. Discussion. . . . . . . . . . . . . . . . . 78 CHAPTER V - KINETICS OF ELECTRON TRANSFER AT PLATINUM AND GOLD ELECTRODES . . . . 85 A. Introduction. . . . . . . . . . . . . . . . 85 B. Reaction Orders . . . . . . . . . . . . . . 86 C. Outer-Sphere Kinetics of Co(III) Ammines . . . . . . . . . . . . . . . . . . 9A D. Unusual Aspects of Iodide- Treated Platinum and Gold Electrodes. . . . . . . . . . . . . . . . . 108 iv Chapter Transition-Metal Aquo Reactants . . . . . . Inner-Sphere Kinetics of Co(III) Ammines . . . . . . . . . . . . Heterogeneous-Homogeneous Rate Correlations. . . . . CHAPTER VI - ELECTRON TRANSFER WITH CARBOXYLIC OR AROMATIC LIGANDS AT PLATINUM AND GOLD ELECTRODES. . . . . . . Reduction Kinetics of Pentaammine- cobalt(III) Complexes Containing A,A'-Bipyridine and Related Ligands Reduction Kinetics of Pentaamine— cobalt(III) Complexes Containing Carboxylic Acid Ligands . . CHAPTER VII- MISCELLANEOUS EXPERIMENTS. A. The Standard Potential for Co(NH3)2+/2+ 1. Introduction. 2. Experimental. 3. Results A. Discussion. . Activation Parameters for Hetero- geneous Electron Transfer Reactions 1. Introduction. . . . . . . . . . . 2. Results 3. Discussion. Maxima in Pulse Voltammetry at an RDE O O O O O O O O O O O O O O O 0 CHAPTER VIII- CONCLUSIONS AND SUGGESTIONS FOR FURTHER WORK. GENERAL REFERENCES. REFERENCES. Page 111 117 129 137 137 150 158 158 158 161 162 173 175 175 176 182 183 183 192 193 Table II III IV VI VII VIII IX LIST OF TABLES Fast Cyclic Voltammetry Results at Platinum and Gold . . . . . . . Kinetic Probe Results at a DME Kinetic Probes of Adsorption at 3+ Platinum and Gold: Co(NH3)50H2 Reduction at 0 mV vs NaSCE. Apparent Change in the Reaction Plane of Co(NH3)g+. . . . . . . Reaction Orders . . . . . . . . . . ¢E in 0.1 M NaClOu at Various Elec- trode Potentials E. . . . . . . . . Reduction Rate Parameters for Co(NH3)2+ and Co(NH3)5F2+ at E=0. . . . . . . . Reduction Rate Parameters for Other Outer-sphere Cobalt(III) Complexes. Oxidation Rate Parameters of Various Aquo Complexes. . . . . . . . . Kinetics Data for Inner-Sphere Electron Transfer at Platinum Electrodes. . . . . . . vi Page 6“ 70 77 82 90 101 10A 105 112 119 Table XI XII XIII XIV Page Kinetics Data for Inner—Sphere Electron Transfer at Gold Elec- trodes. . . . . . . . . . . . . . . . . . 121 Electrochemical Rate Parameters for the Reduction of Various Co(III)- (NH3)5X Complexes at Mercury, Platinum and Gold Electrodes in Perchlorate Electrolytes. . . . . . . . . 139 Relative Heterogeneous and Homogeneous Rate Comparisons for Nitrogen Hetero— cycle-Containing Co(III)(NH3)5L Complexes . . . . . . . . . . . . . . . . 1A6 Reduction Rate Parameters for Car- boxylic Cobalt(III) Complexes . . . . . . 152 Cyclic Voltammetry Results for 5 mM CoCl2 in NH3/NHuCl Solution (u=2-5 M) at Gold Electrodes. . . . . . . . . . . . 166 Activation Plot Results at a Gold Electrode . . . . . . . . . . . . . . . . 181 vii Figure LIST OF FIGURES Page Free energy—reaction coordinate plot proposed for heterogeneous electron transfer. . . . . . . . . . . . . . . . . 6 Chronocoulometric data for various Cr(NCS)2_ concentrations (uM). The supporting electrolyte was 1M NaClOu. The potential was stepped from -300 to -1100 mV at an HMDE (Area=0.032 cm2). . . . . . . . . . . . . . 39 Adsorption Isotherms for Cr(NCS)g- from Chronocoulometric data at an HMDE in 1M NaClOu. Circles repre- sent data reported by Weaver and 91 Anson. The line and triangles are the isotherm from data like that in Figure 2 . . . . . . . . . . . . . . . . A2 Potential-time function for stair- i’ Ef’ and ES are the initial, final and case Chronocoulometry. E step potentials. tw’ tp, and td are the wait time, pulse time and viii Figure delay time respectively. Typical values are given in the text. . . . . . Comparison of capacitance curves from a.c. bridge and staircase Chronocoulometry at mercury electrodes. The supporting electrolyte was 0.115 M NaClOu. Triangles represent two trials of staircase Chronocoulometry with ES = +100 mV. The line and circles are the a.c. bridge result. Capacitance curve from staircase Chronocoulometry at a gold electrode in 0.5 M NaClOu . . . . . . . . Capacitance of a gold electrode as a function of NCS- concentration from single-step Chronocoulometric data. The potential was stepped from +200 to -300 mV in 0.5 M NaClOu. . . . . . . . . . . . . . Useful range of experimental param- eters for reactant adsorption measurements by fast cyclic voltam- metry. The numbers on the curves are the surface concentrations of reactant in mol cm-2. The useful ix Page “7 50 53 55 Figure 10 11 range lies between the curves Change in the rate constant for reduction of Co(NH3)50H3+ as a function of apparent surface charge density of giggCr(NH3)u(NCS); at a DME. Rate constants were found by d.c. polarography. Adsorbed charge density was determined by single-step Chronocoulometry at —200 mV . . . . . . . . . Reaction order plot for Co(NH3)5Cl2+ reduction on platinum in 0.1 M NaClOu at +A00 mV. Reactant bulk concentrations were 0.5, l, and 3 mM for circles, triangles and squares, respectively. Points are labeled with the rotation speed in rpm. . . . . . . . . . . . Rate-potential (Tafel) plots for 2+ Co(NH3)5F and Co(NH3)g+ reductions at platinum and mercury in 0.1 M NaClOu. Solid lines refer to platinum rate constants. Dashed lines are double layer corrected rate constants at mercury electrodes. Page 62 75 89 97 Figure 12 13 Page Heterogeneous-homogeneous correlation at platinum electrodes. The hetero- geneous data were extracted from Tables VII, VIII, X and XI. Circles and triangles denote outer- and inner-sphere mechanisms, respectively. The homogeneous data are for reduc- tion by Ru(NH3)é+ and are from Reference 160, and R. C. Patel and J. F. Endicott, J. Am. Chem. Soc., 90, 636A (1968). The solid symbols correspond to iodide—treated sur- faces. All of the ratios are relative to the reduction of Co- (NH3)5F2+. The points are labeled with the X ligand in Co(III)(NH3)5X. The symbols g—(OH2)2, and g and :- (en)2Cl2 refer to c__—Co(NH3)u(OH2)g+ and g and t—Co(en)2Cl: complexes, respectively. . . . . . . . . . . . . . . 132 Heterogeneous-homogeneous correla- tion at gold electrodes. The heterogeneous data were extracted from Tables VII, VIII, X and XI. Circles and triangles denote outer- xi Figure 1A 15 Page sphere mechanisms, respectively. The homogeneous data are for reduc- tion by Ru(NH3)g+ and are from Reference 160, and R. C. Patel and J. F. Endicott, J. Am. Chem. Soc., 90, 636A (1968). The solid symbols correspond to iodide-treated sur- faces. All of the ratios are rela- tive to the reduction of Co(NH3)5F2+. The points are labeled with the X ligand in Co(III)(NH3) X. The symbols g—(OH2)2, and g and t- (en)2C12 refer to g—Co(NH3)u(OH2)3+ and g and t—Co(en)2Cl: complexes, respectively. . . . . . . . . . . . . . . 134 Reduction rate constants for Co(III)— (NH3)5L at various electrodes. The supporting electrolytes were 0.5 M LiClOu for Hg or 0.1 M NaClOu for Pt and Au electrodes. The pH was adjusted to pH = 3 with HClOu . . . . . . 1AA Co(III)(NH3)5X electroreduction rates at 0 mV at various electrodes. The supporting electrolytes were 0.5 M LiClOu for Hg or 0.1 M NaClOu for solid electrodes. . . . . . . . . . . 15A xii Figure l6 17 18 19 20 Page Homogeneous and heterogeneous reduc- tion rate constants of Co(III)(NH3)5X as a function of the X ligand pKa . . . . 156 The possible reaction scheme for Co(NH3)2+/2+ in aqueous ammonia solutions . . . . . . . . . . . . . . . . 16A The cyclic voltammogram for CoCl2 in 1A.8 M NH + 1.5 M KCl + 1 M 3 NHACl at a gold electrode. The potential was scanned in the posi- tive direction first. This is an example of an anodic-cathodic scan. The potential sweep rate was 200 mV s‘l. pH = 10.50 . . . . . . . . . 169 The cyclic voltammograms for Co(NH3)6Cl3 in lu.8 M NH3 + 1 M NHuCl at a gold electrode. The anodic-cathodic scans for various electrolysis times (a—c) at -500 mV. The potential sweep rate was 200 mV s'l. pH = 10.50 . . . . . .‘. . . 171 Tafel plots from fast cyclic voltam- metry at various surface charge densities of Co(NH3)5Ncs2+ on a gold electrode at 25.0°C. The xiii Figure 21 Page supporting electrolyte was 0.1 M NaClOu. Triangles, squares, and circles represent sweep rates of 20, 10, and 5 V s’l, respectively . . . . 179 Pulse voltammogram for 0.5 mM 2+ Co(NH3)5NCS at a gold RDE in 0.1 M NaClOu + 2 mM H010“. The rotation speed was 5A0 rpm. The potential was scanned at 2 mV 3-1 . . . . 187 xiv INTRODUCTION A central and fundamental problem in electrochemical kinetics is understanding the means by which the inter— phasial environment at an electrode surface influences heterogeneous electron transfer. That such influences are important is shown by the many and varied uses found for electrodes in industrial syntheses, electroplating, analyses and in the production of electricity. However, the basic understanding of these processes is quite limited even among specialists. Indeed, the term "black magic" often arises in connection with these electro- chemical processes. The empirical formulae developed by such workers are often jealously guarded secrets. The empirical nature of much of this work is mainly the fault of the chemistry involved rather than the scientists. The mechanisms of many of the most useful applications are very complex, and those processes which exhibit the most simple and understandable mechanisms generally have no practical applications. It was my aim in the research presented here to examine an electrochemical system which had the potential for yield- ing fundamental information about heterogeneous electron transfer at electrode surfaces which have in the past been widely used in electrochemical catalyses. Platinum and gold were chosen as electrodes. Although mercury electrodes would be a more tractable choice, platinum and gold are more catalytic and hence a better choice to in- vestigate heterogeneous catalysis. The electrode is only one half of the electron transfer process. The choice of an appropriate reactant is equally important. For reasons which are discussed in some detail in Chapter I, cobalt (III) ammine complexes were chosen. The goal of the research was then to understand as well as possible the effect of the interphasial environ- ment at platinum and gold electrodes upon the electro- chemical reduction of cobalt (III) ammine complexes. In many cases much was revealed by comparison to their reduc- tion at mercury electrodes or to their homogeneous reduc- tion. This research can be roughly divided into two parts: the determination of reactant adsorption and the electron transfer kinetics. This latter part includes studies of reaction mechanisms and rates. The contributions of the reactant and the electrode surface to electrocatalysis are discussed. CHAPTER I BACKGROUND AND HISTORICAL ASPECTS A. Heterogeneous Electron Transfer It is useful to divide heterogeneous electron transfer mechanisms into inner— and outer-sphere categories, depending on whether the reactant or its coordinated ligands does or does not penetrate the layer of adsorbed 1’2 The inner- solvent molecules at the electrode surface. sphere transition metal complex undergoes electron trans- fer via an adsorbed intermediate where one or more of the coordinated ligands is bound to the electrode surface. The outer-sphere complex undergoes electron transfer at a more distant reaction plane, separated from the surface by at least one solvent layer. These definitions for heterogeneous reaction pathways are analogous to their homogeneous counterparts for electron transfer reactions.3 The electron transfer reaction, 0x + e' + Red (1.1) where Ox and Red denote some species in its oxidized and reduced forms, can be described by the so-called Dre-equilibrium model.u’5 According to this model the overall electron transfer reaction (Equation 1.1) can be divided into two discrete steps: + OX(bulk) Kp OX(reaction plane) + e- + Red (1.2) ket (reaction plane) The bulk reactant is in equilibrium with the reactant at a plane near or at the electrode surface depending on the type of mechanism. The reactant at this plane cor- responds to the "precursor" state for the elementary elec- tron transfer process. Some degree of reactant orbital overlap with surface metal orbitals is achieved in this state. Thermal activation of the precursor leads to the proper bond/orbital orientation for electron transfer in the activated complex at which point electron transfer occurs. The product thus formed represents the "successor" state. This product at the reaction plane is in equilibrium with the bulk product.”’5 The free energy surfaces are depicted in Figure l for this process for both inner— and outer-sphere pathways. The overall process can be treated more formally by a consideration of the free energy barriers between states. The bulk reactant-to-precursor and successor-to-bulk Figure 1. Free energy-reaction coordinate plot proposed for heterogeneous electron transfer. H opswfim tom 1% + x0. uusvoua ouuum xfism uowmooosm \II. 00’ \ 00 {IX .. oumum usauomou M nomusuoum «Sam 0 C \O.‘ DD- 000 0“ ’O.‘\\ a. .. 0 Q o. .oo. quuofionEfi .. a... N. on R 00 o o .... \ .04 ouozamluouso a. .\ HUM wCQH u :0..— UumHN product equilibria can be described by the equilibrium constants Kp and KS, respectively. The type of mechanism determines the nature of these equilibria. In an outer- sphere mechanism the reactant diffuses to the reaction plane through a potential gradient at a rate controlled by the magnitude of the potential, charge on the reactant ion, and the nature of the supporting electrolyte ions. The constant Kp is then given by6 Kp = r exp(—ZPF¢r/RT) (1.3) where r is the reactant radius, Zr its ionic charge num- ber, and ¢r the potential at the plane of reaction. The radius r is included as a statistical factor which converts a three-dimensional reactant concentration to an effective two-dimensional concentration on the reaction plane, 142;: the average surface concentration of reactant point charges in a layer centered at the reaction plane and extending %r in either direction. For an inner-Sphere mechanism Kp is equivalent to a more conventional adsorption equilibrium constant and is expressed as11 Kp = r/cb (1.u) the ratio of the surface concentration to the bulk con- centration. Equation 1.“ could be separated into two contributions, an electrostatic work term similar to Equation 1.3 and an adsorption term, but is most con- veniently given as the concentration ratio. The rate constant for electron transfer from the precursor state is shown as ket in Equation 1.2 and is related to the overall rate constant by11 k = K k (1.5) The measured, overall electrochemical rate constant for either pathway can also be given as6’7 2n kE = in K + in KV — (AG*)E/RT (1.6) app p p at a specified electrode potential E, where (AG*)E and vp are the reorganization energy and frequency factor for activation from the precursor state, and K is the prob- ability of the electron tunneling from one free energy surface to the other in the activated state (generally 8’9). The pre- taken to be unity for adiabatic mechanisms equilibrium represented by Kp corresponds to a thermo- dynamic contribution to the overall rate. The activation represented by (AG*)E can be more pre- cisely defined as8 (110*)E = (AG+)i _ 0.5 RT (in KS — 2n Kp) + 0.5 F(E — E°) (1.7) Here (AGI)i is the intrinsic free energy of reorganiza- tion and the remaining two terms on the right hand side of Equation 1.7 represent the thermodynamic contributions of relative precursor and successor state stabilities and the "driving force" from the potential difference from the standard potential. The term (AGI)i has been called the "intrinsic barrier to electron transfer".6’10-l2 The intrinsic barrier is equivalent to AGI when the driving force for the elementary electron transfer reaction (l;§;’ 6,11 precursor to successor) is equal to zero. One may expect KS 3 Kp for many simple reactions, which would eliminate the second term on the r.h.s. of Equation 1.7. The final term in that equation (a driving force term) is less easily taken into account. Values of E° for cobalt (III) complexes are usually not available. However, by comparing relative rates of similar complexes or the same complex reacting by different mechanisms, one can obtain estimates of A(AG+)i from A(AG*)E.6’10 The use of rela- tive rates accounts for any driving force differences. With adsorption or double layer data, values for Kp can be calculated and the intrinsic barrier to electron trans- fer evaluated from Equation 1.6 and relative apparent rate constants.11 The intrinsic barrier is expected to be of greatest importance for inner—sphere reactions which can exhibit significant catalysis. A determination of the intrinsic barrier for a variety of reactants could aid in 10 the understanding of the mechanisms and controlling factors in heterogeneous catalysis. B. Kinetics at Platinum and Gold Surfaces The first electrochemical studies were performed near the beginning of the twentieth century and featured primarily 13—15 Work proceeded with platinum and gold electrodes. electrolysis of metal ions and oxidation of anions at these surfaces until about 1930 when the advantages of working with dropping mercury electrodes became apparent.16 From that time on studies at solid electrodes ran parallel to those at mercury except for the "special" electrocatalytic area involving organic molecule oxidations at noble metal 3H surfaces. The generally less reproducible solid elec— trode studies at first did not lend themselves to theo- retical treatment as well as results obtained at mercury. However, the reliability of solid electrode results has greatly improved as better methods of assuring solution purity and surface pretreatments have been developed. Recent interest in platinum and gold electrodes can be divided into two major areas: kinetic studies of solu- tion redox systems, and fundamental investigations of electrode surface processes such as oxide formation. The latter area has always been important in studies of elec- trode kinetics as well as in the "activation" of elec- trode surfaces by electrochemical pretreatments.l7’20 11 Conway and coworkers have recently studied the oxide formation, reduction and hydrogen evolution processes in 17-19 Analogous studies for gold electrodes are reviewed by Woods.20 The potentials of some detail for platinum. interest in these experiments usually lie outside the potential range appropriate for solution redox kinetics experiments, where currents due to oxide or hydrogen processes only serve to complicate the analysis of kinetics parameters. In order that such interferences may be avoided, all kinetics studies must involve some determination of underlying surface phenomena for the appropriate experi- mental conditions of pH, electrolyte, potential and sur- face pretreatment. Kinetics studies of redox systems are much more preva- lent for mercury than for solid electrodes, mainly due to the effort applied to polarographic analysis and the ad- vantages of a continually regenerated clean electrode sur- face at a dropping mercury electrode.16 The large body of work involving platinum catalysts in organic molecule oxi- dation illustrates its ability to assist oxidation through chemical catalysis in addition to electrocatalysis.3Ll For the most part these reactions proceed by complicated multi-step mechanisms which are not well suited to a fundamental study. A summary of recent kinetics studies at platinum or gold electrodes follows. Transition metal ions have been investigated by 12 21 Oldham and Parry for Pb(II) at Pt and by Angell and 22 Dickinson for Fe(III)/(II) at Pt and Au. In these studies the inner-coordination sphere can be readily al- tered by complex formation with anionic contaminants in solution.33 Dissolved gases such as SO2 have been studied 23 at Au and their oxidations also involve multiple steps of more than one electron per molecule.23 Probably the most popular redox couple studied at these surfaces (apart from the Co(III) complexes used in this study), is the ferri-ferrocyanide (Fe(CN)2—/u-) couple. Kinetics of electron transfer for this couple have been determined by 22,2u,25 rotating disk voltammetry at Pt and Au. The in— fluence of adsorbed cations upon the kinetics at Au has been measured.26 Kuta and Yeager26 found Fe(CN)6-/Fe(CN)2- to exhibit more complicated charge transfer than previously suspected, and claim it is an inappropriate couple for checking the properties of new electrodes. In a similar 27 tions for the Fe(edta)—/_2 couple on Pt, which may in- study Mfiller and Prumke encountered mechanistic complica- dicate that iron (III)/(II) couples in general are not the 39 best Choice of redox systems. The other class of re- actants receiving some attention are the inorganic anions such as 1",28’29 320g' and 8A08-'27,30-32 The extremely strong bond formed between iodide and platinum29 or gold 10 is interesting and unusual. The reduction of iodide is also quite interesting, but hardly a model of typical l3 redox chemistry at these surfaces.29 The S2 g- and SHOE- reduction have been found to be extremely sensitive to the supporting electrolyte cations in much the same manner as Fe(CN)2-/u-.3l The suitability of Co(III) ammine complexes for kinetic studies has been amply demonstrated by Weaver 6’35'38 Although Cr(III) complexes were also well-suited for mercury experiments,l they almost always re- and coworkers. duce too slowly (i.e., at potentials too negative) to be used with either platinum or gold electrodes. The Co(III) complexes are recommended by several features: (i) Co(III) complexes are "substitutionally inert". They have stable, low spin tgg, configurations which allow the inner coordination sphere to be varied independently of the solution composition, and to remain intact until electron transfer occurs. (ii) They undergo one-electron reductions which are chemically irreversible in aqueous solution due to rapid aquation of the Co(II) reduction product to form Co(OH2)2+. This product is not oxidized until a very positive poten- 3+/2+ 2)6 is +1.65 V versus SCE).36 The irreversible nature of the tial is reached (the formal potential of Co(OH electron transfer reaction makes the determination of electrode kinetics straightforward since no back reaction 38 correction is necessary. 1“ (iii) The electroreduction proceeds relatively slowly as a consequence of the metal-ligand bond alterations necessary to accommodate the reducing electron in an eg orbital. (iv) The Co(III) complexes can be reduced by inner- or outer-sphere mechanisms depending on the ability of coordinated ligands to bond to the electrode surface.lo’39’L1O (v) Much homogeneous redox information has been ac- cumulated for these complexes, for both inner- and outer- sphere mechanisms.3’Lll Other advantages and disadvantages are more dependent upon the type of electrochemical technique employed. The electroreduction of some Co(III) complexes are observable on silver, mercury, platinum and gold electrodes. A com- parison with electrode kinetics at mercury surfaces is especially usefu1.lo’ll’39 C. Adsorption on Platinum and Gold in Aqueous Solution As previously noted, most of the adsorption studies at these surfaces have been concerned with O or H+ and 17,18,20 2’ H The adsorption/deposition of metal ions has 2 been extensively investigated by the Russian school with 60-63 radiotracer techniques. The adsorption of organic molecules and simple inorganic ions has been examined by 2 capacitance measurements with most success at gold 15 12—19 The adsorption of alcohols}12 pyri- dine:13 chloride,uu_u6 iodide,“6’”7 bromide,”5.u8 A9 electrodes. and per- chlorate on polycrystalline gold was determined from analyses of differential capacitance curves. The ad- sorption of bromide on gold has also been detected by 50’51 Isotherms for the adsorption specular reflectance. of chloride, bromide, iodide and H2PO; were obtained from their effects upon the reversible hydrogen adsorption/ 52 desorption cyclic voltammetry peaks. The adsorption of cesium and sodium cations on platinum was found by capa- 53 citance measurements. In nearly all of these studies of adsorption some difficulty in obtaining results in agree- ment with other investigators was noted. Most of these adsorbates are electroinactive at platinum and gold; not- able exceptions are iodide and some small molecular weight alcohols which can be electroactive at extreme positive potentials.29’”2 Electroactive complex ions have been considerably less studied at platinum and gold. Hubbard and coworkers have examined the adsorption and kinetics of Pt(II)/(IV) complexes at platinum electrodes by thin layer voltam- metry.5"'57 They found the reduction of Pt(IV) complexes to proceed by halide bridged mechanisms which were sensi- tive to other adsorbed species. The reduction of the octahedral Pt(IV) goes through an unstable Pt(III) inter- mediate to produce a square planar Pt(II) complex. The l6 two—electron reduction or oxidation was used to investi- gate various olefinic surfactants on platinum and their effect upon the double layer potential.57 D. Related Kinetics and Adsorption at Mercury Electrodes Several aspects of the kinetics of Co(III) ammine com- plexes have been detailed by Weaver and coworkers at mercury electrodes in aqueous solutions. The rates of reduction and the effect of anion adsorption upon these rates were measured and compared to contemporary double layer theory with excellent agreement.35’36 The outer- sphere (non-bridging) complexes such as Co(NH3)5F2+ and Co(NH3)g+ were found to be sensitive kinetic probes of the mercury-aqueous double layer structure, lLSL’ their electro— reduction rates were systematically altered as adsorbed anions changed the double layer potential at the reaction plane.35’36 The mechanism of electroreduction was found to be inner-sphere (ligand-surface bridging) for Co(III) 6,36 3 parison of the heterogeneous electrochemical reduction complexes containing N or NCS- ligands. The com- rates to the corresponding homogeneous reductions with known outer-sphere reducing agents as treated by the 8’58 was suggested by Marcus electron transfer theory, Weaver as a further means of identifying mechanisms for heterogeneous electron transfer reactions.36 The tem— perature dependence of the reduction rates for a few 17 Co(III) ammine complexes allowed the thermodynamic activa- tion parameters to be calculated for mercury electrodes.59 Adsorption measurements for the Co(III) complexes at the mercury-aqueous interface are not available, at least partly because the reduction potentials of adsorbed Co(III) complexes are positive enough to make measure- 10,11 ments difficult if not impossible. Cr(III) analogs have been used to estimate Co(III) complex adsorption.6 The mercury experiments demonstrated the utility and versatility of Co(III) complexes in electron transfer and double layer studies. The amount of double layer data available for mercury experiments exceeds that at solid electrodes. 0n the other hand the most interesting catalytic behavior occurs on solid electrodes rather than mercury. This compensates for the more difficult and less precise nature of solid electrode experiments. The success en- countered with the cobalt (III) ammines at mercury makes an extension of these studies to include platinum and gold electrodes desirable. Such studies can yield lfl.§lEE information about factors controlling catalysis and double layer structure at these electrode surfaces. CHAPTER II EXPERIMENTAL A. General Apparatus Electrochemical cells employed for most kinetic measure- ments were of the conventional two-compartment type. The working and reference compartments were separated by an "ultra—fine" glass frit (0.9 to l.” um pore size) specially ordered from Corning, Inc. This frit was found to allow good electrical conductivity while preventing the separate solutions from mixing over the timescale of a typical ex- periment (W2-A hrs). The cells had working compartments of two general shapes depending on whether they would be used for rotating-disk voltammetry. The regular working com- partment was cylindrical with a diameter of m1.5 cm, Joined to'a 2A/A0 female joint at the top, and a Teflon stop- cock at the bottom. Typical solution volumes were 5 ml. The working compartments for the rotating-disk cells were also cylindrical but had a larger diameter (m 3 cm) to ac— commodate a rotating—disk electrode (RDE). The compartment bottom was flat to insure the proper hydrodynamic solution flow. The cell top was constructed to fit a Teflon plug which closely fitted the RDE body. Typical solution volumes 18 19 for these cells were 7 to 10 ml. A special type of cell was made to allow bulk elec- trolysis by a mercury pool followed by rotating-disk volt- ammetry without transfer of the electrolyzed solution. The working compartment was similar to that of a RDE cell with a stopcock attached at the bottom to allow the mercury pool to be drained after use. Some of these cells included a platinum wire contact to the mercury pool through the cell wall. In addition to the reference compartment a frit separated the counter electrode from the working compart- ment. This frit was more porous to minimize cell resist- ance as much as possible without allowing free exchange of anodic and cathodic electrolysis products. Reference electrodes were aqueous saturated calomel reference electrodes (SCE). When perchlorate media were used the standard fill solution of saturated KCl was replaced by saturated NaCl to make a sodium calomel elec— trode (NaSCE). This was necessary to avoid the precipita- tion of KClOu in the reference electrode junction which would give rise to erroneous potentials. The NaSCE differs from the SCE by only -5 mV. Counter electrodes were simply lengths of platinum wire. Occasionally in electrolyses longer lengths of wire were submerged in the counter electrode compartment to give a larger electrode area and allow a higher current flow. The solid electrodes used (several of both platinum 20 and gold) were fabricated from high purity (99.999%) polycrystalline metal rods. These rods were enclosed in a cylindrical Teflon sheath and attached to a stainless steel shaft which fit into the electrode rotator. The sheath radius was about 0.6 cm and the exposed electrode disk radii were either 0.200 or 0.125 cm. A model ASR2 rotator (Pine Instruments Company) was employed for the RDE experiments. Rotation speeds up to 10,000 rpm were possible. B. Solution Preparation Solid electrodes have proven to be very sensitive to solution impurities, both organic and inorganic. These impurities have a strong tendency to be adsorbed at the surface. Sensitivity to impurities was especially notice- able in RDE experiments where their transport to the sur- face is assisted by the convection caused by the rotating electrode. Consequently, (and unfortunately) extreme measures were required to obtain pure water and support- ing electrolytes for solid electrode experiments. Water for solution preparation and electrode pre- treatments was produced from two sources. The first step in both purifications was distillation of tap water from alkaline KMnOu to oxidize most of any organic impuri- ties to O2 and CO While this procedure previously had 2. been adequate, many organic contaminants present in more 21 recent times are steam volatile and hence not removed in this distillation step.6u This permanganate feed water was then either pyrodistilled or fed into a quartz sub— boiling still.65 In pyrodistillation water vapor in an 02 stream is passed through a quartz column heated to 750°C to cause combustion of the remaining organic impurities. The quartz subboiling still employs infrared heaters to evaporate water which is condensed on a cold finger and collected. The subboiling still is most effective in the removal of inorganic impurities. The purity of the water is evidenced by the shape of the hydrogen evolution and adsorption peaks on platinum as well as by capacitance measurements on mercury. The supporting electrolytes (primarily NaClOu and KPF6) were obtained at the best purity commercially avail- able and recrystallized two or three times from pyrodis- tilled water. NaClOu was usually obtained as reagent grade from G. Frederick Smith Chemical Company. KPF6 came from Pfaltz and Bauer, Inc. as 96% purity. HClOu (also from G. F. Smith) was used from 70% ACS reagent grade without further purification. HPF6 was prepared from concentrated HClOu and KPF6 (taking advantage of the low solubility of KClOu) and stored frozen to prevent acid catalyzed hydrolysis to form fluoride. The concentration of HPF6 was determined by titration with standard NaOH. Although LiClOu is normally employed at mercury 22 electrodes (the conductance of Li+ being closest to H+), NaClOA was preferred for solid electrode experiments since it proved to be more easily purified. Concentrations of supporting electrolytes on solid electrodes were generally kept to about 0.1 M to reduce the effect of any residual impurities in these electrolytes. Solutions for mercury experiments, except for those involving Chronocoulometry, used tap distilled water passed through a Milli-Q Reagent-Grade Water System (Millipore Corp.), which proved adequate for the generally less stringent purity requirements at mercury. Chrono- coulometry experiments at mercury employed the same water as solid electrodes. All solutions required deaeration to remove dissolved oxygen before use in electrochemical measurements. This was achieved by bubbling the solutions with a stream of deoxygenated purified nitrogen for 15 minutes after which a stream of nitrogen was passed over the solution during the experiment. Oxygen and residual organic contaminants were removed from the prepurified nitrogen stream by a solid catalyst train. The catalyst (BASF, R3-ll copper based pellets from Chemical Dynamics Corp.) was packed in a glass cylinder (5 x 28 cm, volume m500 cm3) heated to W150°C. The gas stream was subsequently passed through a gas scrubbing bottle containing pyrodistilled water to humidify the nitrogen and prevent evaporation of the 23 working solution by a dry gas. Supporting electrolyte solutions were usually prepared before each experiment and not stored, in order to reduce the formation of hydrolysis products such as fluoride or chloride. C. Preparation of Electroactive Complexes Most of the transition metal complexes used were al- ready available and required only recrystallization as C10; salts. However, the following complexes were syn— thesized following literature procedures: [Co(NH3)uCO3]- 66 66,67 3H20,68 cis- and trans-[Co(en)2012]Clou,69 and cis-[Co(en)2- 7O K3[Fe(C2Ou)31° (NCS)2]ClOu. Several attempts to synthesize gig-[Co- (NH3)uF2]N03 by Bohm's method71 and a few modifications failed. Cr(OH2)§+ was synthesized by electrolysis of 3 mM Cr3+ in 0.5 M NaClOu + 20 mM HClOu over a mercury pool potentiostatted at -1100 mV versus NaSCE for 30 min. Solutions of Ru(OH2)g+ were prepared by electrolyzing a A mM solution of RuCl3 (Alpha Inorganics) in 0.18 M KPF6 + 0.02 M HPF6 over a stirred pool of mercury at -600 mV under an argon atmosphere. Then 20 mM AgPF6 prepared from Ag20 and HPF6 was added to precipitate the liberated chloride as AgCl. The solution was cooled, filtered and electrolyzed at 50 mV to electrodeposit the excess Ag+ 2+ 2+ 3+ and oxidize the Ru(OH2)6 to Ru(OH2)6 . Both Craq and 2N + Ru:q were stored in liquid nitrogen prior to use. V3; was prepared by electrolysis of V(V) from V205 over mercury 2+ . . . 3+ aq product then belng ox1dlzed to Vaq at -300 mV versus SCE. The synthesis of the cobalt (III) at -1100 mV; the V ammine complexes incorporating carboxylate and pyridine- related ligands by Visiwanathan Srinivasan is gratefully acknowledged. The solutions of cobalt (III) complexes were not stored for long periods since the complexes eventually are hy- drolyzed. The 010; salts were kept in a dessicator in the refrigerator to reduce the chances of photolysis and ex- plosion of dry C10; salts. D. Electrochemical Techniques 1. Cyclic Voltammetry A PAR (Princeton Applied Research) 17“ or 17AA po- tentiostat was usually employed to generate a triangular- wave potential sweep at a rate of 20 to 500 mV per second during which the current at a stationary electrode was monitored. The current—potential profile was recorded with a Hewlett-Packard HP70A5A x-y recorder. The RDE or small area flag electrodes prepared from the appropriate metal foil were used for platinum and gold. A hanging mercury drop electrode (HMDE) (Metrohm Model EA10, Brink- man Instrum.) was used for mercury. Cyclic voltammetry 25 was usually employed prior to rotating disk voltammetry to determine the potential range most suited for the rotat- ing disk experiment. However, the apparent heterogeneous rate constant, k can be obtained if necessary by app’ methods described by Nicholson and Shain72 (for quasi- reversible) or Galus73 (for totally irreversible) electron transfer reactions. 2. D.C. and Pulse Polarography The apparatus used for these techniques were usually the same as described for cyclic voltammetry. A dropping mercury electrode (DME) with a mercury flow rate of 1—2 mg s-1 and column height of 50 cm was employed. Drop times of 0.5, 1.0 or 2.0 s were obtained by means of a PAR 17U/70 mechanical drop timer.- The potential was usually scanned at a rate of 2 mV s-l. Kinetic parameters were 7A-76 obtained by Koutecky analysis (for d.c. polarography) AA and by equations derived by Oldham and Parry (for pulse techniques). 3. Fast-Cyclic Voltammetry For cyclic voltammetry at sweep rates above 500 mV s"1 (which is the fastest attainable on a PAR 17M), a combination of a fast-potential ramp generator (Michigan State University prototype built by Marty Rabb) and a 26 PAR 173 potentiostat as used to obtain sweep rates of up to 50 V 5.1. The cyclic voltammograms were recorded and photographed from a Tektronix 7623A storage oscilloscope. A. Rotating-Disk Voltammetry (Nualargenmjority of kinetic measurements at platinum and gold were performed using this technique. The reactants typically exhibited irreversible electron-transfer waves36 which considerably simplifies extraction of the kinetic parameters. The voltammetric waves were analyzed by determining the effective reactant concentration just out- side the double layer, Cr’ from or = CbEl - i/il] where Cb is the bulk reactant concentration, i the measured current in the kinetically-controlled portion of the reduction wave, t.78 and i1 is the limiting curren The apparent rate constant kgpp at a given electrode potential E was then calculated E from ka = i/FACS, where m is the reaction order, and A DD is the geometrical electrode area. The use of rotation speeds in the range of 100-2,000 rpm allowed values of A l -1 k in the range 10- to 10- cm s to be determined. app Tafel plots36 of log ka versus E were typically linear DD for potentials in the kinetically controlled region of the waves. In most cases the reaction order m was found to be unity by determining the slope of a plot of log 1 against log Cr at a given potential. Cr was varied by changing the rotation speed of the electrode as well as the bulk 27 concentration Cb' Diffusion coefficients for the re- actants could be determined from the limiting current using the Levich equation.79 Normal pulse polarography was also attempted at solid electrodes. A range of rotation speeds (200 to 1000 rpm) was found to provide enough replenish- ment of the reactant by convection to compensate for the depletion of reactant which occurs on a potential step into the reduction wave.21 This technique offered the possible advantage of having less current through the solid electrode- solution interface than in a conventional RDE experiment. The same potentiostat—recorder combination as used for cyclic voltammetry was employed for rotating-disk voltam- metry. Most kinetics parameters are an average of at least four determinations on different days. E. Solid Electrode Pretreatments Undoubtedly one of the primary advantages of the DME as an electrode is that surface effects relating to the history of the electrode surface can usually be ignored; each falling drop renews the electrode surface. This is, of course, not true for solid electrodes. Electrode history, pretreatment and contamination are serious and sometimes severely limiting considerations with solid electrodes. Since the electrode pretreatment, be it chemical, mechanical or a combination of these, is critical in any kinetics data obtained with these electrodes, a standard 28 operating procedure must be adopted. Reproducible kinetics results are almost always obtained provided the pretreat- ment is duplicated each time.80 However, the absolute values of any kinetics parameters are expected to depend on the electrode pretreatment. A rather extensive review oftflmeplatinum pretreat- 81 ments has been made by Gilman which is current to 1965. More recent pretreatments described by Hamelin82 and Woods83 are variations on some general principles described in the Gilman review.81 According to Gilman,81 pretreatments in- volve three steps: (1) mechanical polishing, (2) chemical or electrochemical oxidation of the surface, and (3) re- duction of the oxidized surface. Some criterion appropriate to the specific applica- tion must be found to judge the effectiveness of the pre- treatment. For platinum and gold in this work the hysteresis in the rate of reduction of Co(NH3)50Hg+ measured by rotat- ing—disk voltammetry served as this criterion. Rotating- disk voltammetry is a steady-state technique so any change (liii: hysteresis) in forward and reverse scans of a reduc- tion wave indicates a change in the state of the electrode surface. The reduction of Co(NH3)50Hg+, for poorly under— stood reasons, appears most sensitive. Such time-dependent "activation" of solid electrodes is common.81 The de— activation is usually attributed to either adsorption of solution impurities (which is especially rapid at the 29 highly convective RDE) or to formation of a surface oxide layer.80 The pretreatments of platinum and gold described below were found to give the least hysteresis for Co(NH3)5- 3+ OH2 . 1. Platinum Electrodes The electrode was first mechanically polished with successively finer grades of 1.0, 0.3 and 0.05 pm alpha alumina (Micropolish grades C, B and A, Buehler, Ltd.) on Buehler Microcloth. A two-speed polishing wheel (Buehler AA—1502-l60) was employed. Pyrodistilled water was used to wet the polishing cloth and to rinse the elec- trode. Mechanical polishing was followed by immersion of the electrode in hot (m80°C) 1:1 nitric acid to oxidize the surface and strip any organic film which may be present.81 The electrode was then transferred to a cell containing deoxygenated 0.1 M perchloric acid and potentially cycled between +900 mV and -100 mV for about 2,000 cycles. Fin- ally, the electrode was rinsed and transferred to the test solution where the potential was held at +100 mV to strip any remaining oxide. Cycling of the electrode potential was found to give a fairly stable, activated surface. A similar, but not identical, method has been described by 83 Woods. A more roughened platinum electrode was obtained by polishing with an abrasive disk (Carbimet, Buehler, 3O Ltd.) before pretreating. 2. Gold Electrodes The mechanical polishing was exactly as for platinum. This was followed by immersion in concentrated sulfuric acid for 30 s, rinsing and cycling the potential in 0.1 M perchloric acid between +1000 mV and -200 mV. The poten- tial was held at -100 mV in the test solution to strip residual oxide. An alternative pretreatment82 using cyanide to strip the oxide was found to offer no improve- ment over that described above. 3. Iodide Treated Electrodes To prepare the iodide treated electrodes the freshly pretreated platinum or gold electrodes were placed in a deoxygenated solution of 5 mM NaI in pyrodistilled water for one minute. The surface was then rinsed with support- ing electrolyte and transferred to the test solution. For some experiments, where the adsorbed iodide could possibly be displaced, 5 mM NaI was added to the test solution to maintain an adsorbed iodide layer. Such iodide treatment always improved the reproducibility of kinetics measurements. The iodide layer remained intact at po- tentials below +700 mV on gold and platinum. Oxidation of the adsorbed iodide to iodate occurs at more positive 31 potentials.8Ll The iodide was in fact so tenaciously ad- sorbed that it could only be completely removed by mechan- ical polishing. Cells used for iodide experiments were reserved for this function alone to avoid iodide contamina- tion of "clean" surfaces. The nature of the adsorbed iodide layer will be discussed in more detail in regard to its effect on electron-transfer rates in Chapter V. CHAPTER III DIRECT MEASUREMENT OF REACTANT ADSORPTION The adsorption of ions at electrode surfaces can be determined by various means. The nonfaradaic methods for measuring adsorption of electroinactive ions have received the most attention, since they seem most appropriate in tests of double layer theory at mercury electrodes.2’85 The adsorption of electroactive ions or neutral substan- ces (including transition metal complexes) can be determined by either nonfaradaic or faradaic methods, although the latter is usually more appropriate as the effect of ad- sorption upon the faradaic process is of fundamental value in electrocatalysis studies.6’10 The nonfaradaic methods include electrocapillary measurements based on a relation of surface tension to the surface excess of adsorbed ions, double layer capaci- A2-A9 radiotracer techniques,60'63 tance measurements, ellipsometry (which measures the change in refractive index of an adsorbed film by reflection of plane-polarized light), and surface tension measurements at solid electrode surfaces.87 Of these techniques double layer capacitance is probably the method of choice for routine adsorption 32 33 meaSUPGmGNtss2’6u although the thermodynamic analyses required to obtain the specifically adsorbed charge can be rather involved.88’89 Generally, it is fair to say that these techniques suffer additional complications at solid electrodes relative to mercury, as a result of surface roughness and the nonrenewable surface structure. RefleCtance and radiotracer techniques are especially limited by these factors. Uncertainty in the assignment of real electrode area affects all measurements at solid electrodes. -The application of double layer capacitance measurements to determine reactant adsorption is confined to potential regions where no faradaic prOcess occurs. This is partly because new species (the oxidation or reduc— tion products) are formed at the electrode surface and can then competitively adsorb with the bulk reactant. With solid electrodes the faradaic current can perturb the double layer by altering the metal surface structure in some ir- reversible fashion, and hence cause the capacitance to be changed in an uncertain and irreproducible way. Such a situation is proposed at platinum and gold electrodes.10’ll Of more immediate importance in this work is the de- termination of reactant adsorption at potentials just preceding reactant oxidation or reduction, lLEL’ the de- termination of the concentration of the reactant in the precursor state to electron transfer. Under appropriate conditions several faradaic techniques (chronopotentiometry, Chronocoulometry, voltammetry, etc.) can yield information 3A related to reactant adsorption. The faradaic techniques have the advantage of being sensitive only to electroactive adsorbates, presuming that a correction for the nonfara- daic effects of the supporting electrolyte can be made, which is sometimes an unreasonable presumption. In this study single-step Chronocoulometry and fast cyclic voltammetry were employed to determine reactant adsorption at platinum and gold surfaces. A novel method for the measurement of double layer capacitance relying on Chronocoulometry is also described and is used to il— lustrate some problems with single—step Chronocoulometry and an advantage of fast cyclic voltammetry at solid elec- trodes. The succeeding chapter will deal with an indirect method of detecting complex ion adsorption with outer- sphere kinetics probe reactions. A. Single-step Chronocoulometry As the name implies, in chronocoulometry the charge is monitored as a function of time after a potential step is applied to the electrode-solution interface. Usually the current is most easily recorded and then integrated numerically, or as in this case electronically, to obtain the charge. In single—step Chronocoulometry the potential is held initially at a point where the reactant is adsorbed but not reduced or oxidized (where no faradaic current is present). The potential is then stepped to a point at 35 which the faradaic reaction is entirely diffusion controlled. The charge in the presence of an adsorbed reactant is then given by90' q(t) = 2nFAC(Dt)l/2n-l/2 + qdl + nFAF' (3.1) where C and D denote reactant concentration and diffusion coefficient, qdl’ the charge consumed in charging the electrode-electrolyte solution double layer, A, the "real" electrode area, n, the number of electrons required in the faradaic reaction of one reactant molecule or ion, and P', the apparent reactant surface concentration in moles cm-z. The true surface concentration is related to P' by r = r' — (l/F)(Aqm)E (3.2) where Aqm is the change in the electronic charge density on the electrode produced by the adsorption of the reactant E.91,92 at a constant potential The quantity Aqm has been measured for a few Cr(III) complexes adsorbed at mercury 91,93,9L1 surfaces, but the techniques such as charge measure- ments at an expanding mercury drop are not applicable to solid electrodes. In the absence of Aqm values, F' can serve as an upper limit for the reactant surface concen- tration. Equation 3.1 requires a plot of q as a function of 36 t1/2 to be linear with a slope proportional to the reactant concentration and an intercept equal to the sum of the double layer charge and the charge due to the faradaic reaction of the adsorbate, nFAF'. The charges acquired at later times (>1 x 10‘3 s) are extrapolated back to tl/2 = 0. This is necessary to avoid interference by re- actant kinetics occurring at less than diffusion controlled rates. The extrapolation to zero time allows the adsorp- tion at the initial potential to be found. The correction for the double layer charge (nonfaradaic charge) can theo- retically be obtained as the q versus tl/2 intercept for the same potential step in a solution of supporting electrolyte alone. In practice the value of qd1 obtained in the absence of the adsorbed reactant provides a good estimate of the value obtained in the presence of the reactant only if the product of the faradaic process is not adsorbed at the final potential.ll An underlying assumption is that the electrode is ideally polarizable. As will be seen in the following discussion, this may not be the case at platinum and gold electrodes. An on—line Chronocoulometric data acquisition and analysis system based on an LSI-ll microcomputer was adapted from its previous incarnation as an oneline pulse polarograph.95 The major modification involved the inser- tion of a variable scale integrator between the potentio- stat current output and the analog-to-digital converter 37 (ADC) input to the computer.96 The charge points were acquired at 5 x 10—5 8 intervals for a total of 200 points. Electrodes and cells were the same as used in kinetics measurements. Since single-step Chronocoulometry had been 90,91,93,9L1 used quite successfully at mercury electrodes, this surface was chosen to characterize the on-line system. 1. Adsorption Isotherm for Cr(NCS)2- at Mercury U A determination of the adsorption isotherm for Cr(NCS)g- on mercury was deemed an appropriate test as this would test the sensitivity as well as the accuracy of our system. The supporting electrolyte was 1 M NaClOu, and the potential was stepped from -300 to -1100 mV. Concentrations of Cr(NCS)2- in the range 5 to 370 uM were obtained by addition from a concentrated Cr(NCS)2- stock solution. An HMDE with a drop area of 0.032 cm2 was used for the working electrode. Each new mercury drop was allowed ~30 sec to achieve adsorp- tion equilibrium before the potential step was applied. 1/2 data points for several Figure 2 depicts the q versus t bulk concentrations of Cr(NCS)g-. The adsorbed charge density, FF', was found by subtracting the blank (0 uM complex) intercept at t1/2 = 0 from the observed intercept for a given complex concentration and then dividing by the electrode area. When these charge densities were plotted against the logarithm of the bulk complex concen— tration the adsorption isotherm represented by triangles 38 Figure 2. Chronocoulometric data for various Cr(NCS)2- concentrations (uM). The supporting electro- lyte was 1M NaClOu. The potential was stepped from -300 to -1100 mV at an HMDE (Area=0.032 cm2)- 39 0.15 .. Charge/ Coulomb. ‘ XIo" 00,0 - ,. .......eo:oo .e”e~. , d.c.-o ”I "...‘,.-‘°W 0.0.0.0....000~.0..“°“ . . o _ . one o 0 o. e. e 0.00.00.0000. M~0I’wow:.‘.w L 1 AL I l l l 370 162 m Figure 2 0.1 1 AD in Figure 3 was obtained. The circles in Figure 3 represent the literature values for FF' reported by Weaver and Anson9l for the same initial potential. The agreement is fairly good, especially at the extremes of high and low bulk concentrations. The deviations at intermediate con— centrations may be a result of errors in measuring such small concentrations. In addition to this result several other Chronocoulometry experiments at mercury (related to kinetic probe work) were sufficiently in accord with expec- tations to inspire some confidence in the system. Un- fortunately, application of this technique to solid elec- trodes (platinum, gold and silver) proved to be consider- ably less useful than anticipated in the determination of reactant adsorption.ll’96 2. Single-step Chronocoulometry at Platinum and Gold Several of the Co(III) ammine complexes were suspected to be specifically adsorbed at these surfaces.10 Complexes containing adsorbing ligands such as chloride, bromide and isothiocyanate fall into this category. It was there- fore of interest to determine the extent of specific ad- sorption for these complexes by single-step Chronocoulometry. However, the values obtained for the apparent adsorbed charge density, FF', were always found to be very small (S 2 UC cm-z) or negative for these complexes at platinum and gold surfaces. Since their specific adsorption is A1 Figure 3. Adsorption Isotherms for Cr(NCS)g- from chrono— coulometric data at an HMDE in 1M NaClOu. Circles represent data reported by Weaver and 91 The line and triangles are the isotherm from data like that in Figure 2. Anson. 142 30- . ' d m'2 N O FIT/[LC c o - k A 1 l l n l A 1.5 I . A 2.0 2.5 log [Cb/[1M ] Figure 3 143 indicated by kinetics measurements10 and by mercury ex- periments,6 this result was clearly in error. Adsorption isotherm experiments for Co(en)2(NCS);, and Co(NH3)5Cl2+ 1/2 inter- showed large, negative changes in the q versus t cepts for the first small (ml uM) addition of complex, and further additions had relatively little effect. The addi- tion of thiocyanate anions had a similar effect, which indicated that the adsorption of reduction products (Elia: the ligands formerly bound to Co(III)) was interfering in the analysis. The strong adsorption of faradaic reaction products at the final potential invalidates the assumption that qdl in the absence of the adsorbed reactant is the same as qd1 in the presence of the adsorbed reactant. On mercury electrodes this problem can be avoided by choosing a final potential sufficiently negative of the potential of zero charge (PZC) to assure that anionic reaction products are not adsorbed.70’72 Negative surface charges of this magnitude are not accessible with gold, or especially plat- inum, since hydrogen evolution or proton reduction occurs at potentials only slightly more negative than the po- tentials of zero charge. The PZC of platinum and gold are 76 but are probably near the subject of some controversy —50 mV versus SCE for both surfaces in 0.1 M Na010u.119’97 The evolution of hydrogen from the surface can indeed cause anions to be desorbed, but it also makes a large contribution AA to the chronocoulometric charge. Charges for hydrogen evolution can exceed those for monolayer complex reduction by orders of magnitude.18“2O While it may seem possible to bypass the problem of unequal values for qd1 by adding an excess of the free adsorbing anions and thereby minimizing the effect of the anions released by the complex, this is impractical for two reasons. First, the anions will ad— sorb competitively and are capable of blocking complex adsorption.lO Secondly, the hydrogen evolution and proton reduction processes are highly sensitive to adsorbed anions which can cause substantial catalysis of proton reduc- tion.l7-19 Since proton reduction consumes so much more charge, it is likely that a very large excess of anions would be needed to make the contribution due to released anions negligible. The adsorption of released anions can be circumvented 2+ which is substi- by using a complex such as Ru(NH3)5NCS tutionally inert in both oxidation states.98’99 An ad- sorption isotherm was determined for this complex at a gold electrode. Limiting coverage was apparently reached by ”A uM where FF' was A uC cm-2. This corresponds to about 20% of monolayer coverage, when in all probability the actual limiting coverage is a monolayer. This result can also be explained by product adsorption, although in this case the product is Ru(NH3)5NCS+ rather than free anions. A5 The basis of the erroneous behavior can be found in electrode capacitance measurements. A description of the effect of anions on the double layer capacitance follows as a digression to explain the technique used to measure the capacitance. 3. Staircase Chronocoulometry The concept of obtaining the nonfaradaic charge from chronocoulometric intercepts was applied to the determina- tion of capacitance. Software modifications of the single- step chronocoulometry system allowed the electrode capa- citance over a one volt range to be found in one or two minutes as opposed to the 15 or 20 minutes needed to per— form the same measurement by conventional a.c. bridge tech- niques. The faster measurements are desirable since solid-metal surfaces are not easily renewable and are known to adsorb solution impurities readily. The technique ac- tually consists of several single-step chronocoulometric experiments performed in succession for small, fixed po- tential steps with the initial potential being incremented by the step size until the final potential is reached. A 1/2 provision for a poor data set (a poor q versus t linear regression fit) permitted steps to be repeated at the same initial potential until the designated regression test was met. Figure A is a plot of the potential—time func— tion for staircase chronocoulometry. The potential step Figure A. A6 Potential-time function for staircase chrono- coulometry. E. l, Ef, and ES are the initial, final and step potentials. t t and td w, p, are the wait time, pulse time and delay time respectively. Typical values are given in the text. \\ A7 \\ T 3 a 1 ;\ T 15" 1; Si} 1 1v :1 l. is T u- \\ LIJ leuuazod Time -—-> Figure A8 size ES was usually between 10 and 100 mV. The wait time, pulse time and delay time are denoted by tw, tp and td, respectively. The sum of these three intervals corresponds to one cycle of the program, 112;: one single-step chrono- coulometric determination. The wait time (typically, 5 s) allows the double layer to reach an equilibrium condition at some initial potential. The current is monitored, integrated and stored during the pulse time (typically, 1 ms). The delay time (typically, 1 5) allows time for the linear regression and the test for goodness of fit. An integral capacitance is calculated for each step by divid- ing the charge intercept qd1 by the potential step ES. Figure 5 shows the capacitance of a mercury electrode in 0.115 M NaClOu obtained by staircase chronocoulometry (triangles) and by a.c. bridge measurements (circles and 100 as a function of the electrode potential. The line) agreement is fairly close except for a slight accentuation of the extrema at more positive potentials. This may be the result of some perchlorate adsorption which is given less equilibration time in the chronocoulometric technique, or more likely it is a consequence of the larger perturba- tions inherent in staircase chronocoulometry. The two trials with the staircase technique show the capacitance 2 to be reproducibly measured to $0.3 uF cm' for different mercury drops. Figure 5. A9 Comparison of capacitance curves from a.c. bridge and staircase chronocoulometry at mercury electrodes. The supporting electrolyte was 0.115 M NaClOu. Triangles represent two trials of staircase chronocoulometry with ES = +100 mV. The line and circles are the a.c. bridge result. 5O I P 30- - O N.Eo ad. \ oocagoanao - o 1 -L5 -t0 -0.5 E / V vs. SCE 0.2 Figure 5 51 A. Capacitance-related Errors in Single-step Chronocoulometry The capacitance curve for a gold electrode in 0.5 M NaClOu by staircase chronocoulometry is given in Figure 6. The curve clearly has a minimum at —50 mV which is in agreement with published results found with a capacitance 142,29 bridge. The rapid rise in capacitance at potentials more negative than —250 mV is caused by the adsorption and A9 subsequent reduction of protons. In order to explain the behavior noted for single-step chronocoulometry the effect of adsorbing anions on the electrode capacitance of gold was investigated. The effect of added thiocyanate anions upon the gold electrode capacitance in 0.5 M NaClOu at a constant potential can be seen in Figure 7. This curve shows the apparent change in electrode capacitance at +200 mV for a -500 mV step (typical of single-step chrono— coulometry experiments) as a function of thiocyanate con- centration. This behavior, the rapid decline and partial recovery, of the capacitance was also noted for chloride and for adsorbing complexes such as Co(NH3)5NCS2+. Kinetic probe experiments have shown thiocyanate to be strongly adsorbed on platinum and gold surfaces (Chapter IV). The concentration required to reach limiting adsorption on gold was found to be WA uM. A similar decrease in capacitance 101 A2,A6 a has been noted by others at platinum, gold nd 52 Figure 6. Capacitance curve from staircase chrono— coulometry at a gold electrode in 0.5 M NaClOu. Capacitance HIP cm"2 53 20- —e U! I O I A O a a a a a n j A l +1.0 Figure 6 O E IV vs. SCE 5A Figure 7. Capacitance of a gold electrode as a func- tion of NOS. concentration from single-step chronocoulometric data. The potential was stepped from +2OO to -3OO mV in 0.5 M NaClOu. 55 15 ‘7‘ . E 010- U. 3. . \ d) (J C! (B 1‘: C) (B 5?. 0 5 o J j a a l A 1 1 J O 100 200 Cone. of NCS‘IpM Figure 7 56 96 silver. Capacity measurements are complicated by hydrogen and oxygen adsorption and show strong pH dependencies for platinum and gold.101,102 However, the decrease depicted in Figure 7 is at an essentially constant pH and surface roughness. This decrease in nonfaradaic charge is the major source of erroneous chronocoulometric adsorption measurements at platinum, gold and silver electrodes. Several explanations for this behavior are possible. Adsorbed anions may facilitate reorganization of surface atoms to produce a greater degree of orientation for sur- face atoms and water dipoles. Water dipoles at the sur— face are replaced by adsorbed anions which can form co- valent bonds to the surface thereby donating some electronic charge to the metal surface.91 Once monolayer adsorption has been reached further, more ionic layers might be formed which could explain the "recovery" of the capacitance beyond 20 uM thiocyanate in Figure 7. An alternative explanation is that uncharged substances are adsorbed on the supposedly clean surface and then displaced by the adsorbing anions. As much care as possible was used to assure the purity of electrolytes and cleanliness of the cells and electrodes (Chapter II). 57 B. Determination of Reactant Adsorption by Fast Cyclic Voltammetry 1. Introduction Linear sweep or cyclic voltammetry is a useful fara- daic means of determining reactant or product adsorption ,and has been extensively employed for this purpose at 103-107 The rate of electron 19,107 mercury and solid electrodes. transfer of an adsorbed reactant can also be deduced (see also Chapter V). Probably the most advantageous feature of cyclic voltammetry is the ease by which a cor- rection for nonfaradaic charge can be made, even in the presence of the reaction products. Nonfaradaic charge arising from the charging of the double layer is simply accounted for by extrapolating the i-E curve from in front of the faradaic peak to a point beyond the faradaic peak and integrating the current above this line to obtain the adsorbed faradaic charge. This "baseline" method is conventionally used to find peak 108,109 currents. The adsorbed faradaic charge nFAF i can then yield a value for the initial surface concentra- tion of the reactant Pi. Adsorption of the reactant causes the nonfaradaic current to decrease but also lowers the faradaic peak by the same amount. Hence, this method is much better suited than single-step chronocoulometry to accurately correct for the large changes in nonfaradaic 58 charge typically encountered at solid electrodes.11 The peak current for irreversible electron transfer for an adsorbed reactant is given by109 2 iad = nanaF AvI‘i (3 3) p 2.718 RT ' where a is the transfer coefficient, n the number of a) electrons in the rate determining step, v, the potential sweep rate in V 5.1; the other symbols have been previously defined. Here it is clear that ip depends on v and that the initial surface concentration could be calculated from 12d. The integration of the current to determine P is 1 preferred over the use of Equation 3.3 because the shape of the faradaic peak is assumed to be constant in the 109 derivation of Equation 3.3, when in fact the peak may change shape as a result of adsorbate-adsorbate inter- actions.105’107 The interference by reactant current as a result of reactant diffusion to the surface becomes more important as the bulk reactant concentration increases, especially at slower sweep rates. This becomes evident from the following analysis. The peak current due to diffusion of the reactant undergoing irreversible electron transfer is described by108 iglf = 3.01 x 10 5 1/2ADl/2C vl/2 (3.1) "a “L na] b 59 at 25°C, where C 3 b is the bulk reactant concentration in The current for the diffusing reactant is pro- 1/2 mol cm- portional to v ; so at faster potential sweep rates the current for the adsorbed reactant is favored. This has some practical limitations in that the current due to double layer charging is also proportional to the sweep rate and double layer capacitance121 1d1 = Cd1V (3'5) The nonfaradaic "background" current becomes an increasingly larger fraction of the measured current at faster sweep rates since the adsorbed, faradaic current is limited by Pi (Equation 3.3), which typically does not exceed the -10 value for monolayer coverage (m3 x 10 mol cm’2). The effect of uncompensated cell resistance becomes more ap— parent at faster sweep rates (for our cells at v > 50 V 3.1). Equation 3.A also indicates that the diffusion current is proportional to the bulk reactant concentration. This sug- gests that some optimum conditions of sweep rate and bulk concentration exist for accurate determinations of ad- sorbed reactant concentrations. If currents for adsorbed and diffusing reactants are compared directly, 142;: assuming they appear at the same electrode potential, where the interference of diffusion current in the integration of the faradaic peak is greatest, then the optimal ranges of Cb and v can be calculated from 60 Equations 3.3 and 3.A. If the interference is limited to 5%, then idif < 0.05 iad p — p (3.6) and from Equations 3.3 and 3.A with a = 0.5, T = 25°C, 6 2 -l n = na = l and D = 7 x 10- cm 8 Equation 3.6 becomes 1 V-l/2Sl/2 1/2 0b 5 (63.8 cm‘ )riv (3.7) A plot of log C against the sweep rate is shown in Figure 10 -11 b 8 for two surface concentrations of 3 x 10- and 5 x 10 mol cm-2 corresponding to monolayer coverage and the mini- mum detection limit, respectively. The area between these lines then indicates the conditions for adsorption de- termination by linear sweep or cyclic voltammetry. Figure 8 illustrates several interesting aspects. Sweep rates greater than about 2 V s—1 are necessary to observe an adsorption peak with Cb in the range A—25 uM. This range includes reactants with adsorption coefficients, Kp, greater than 2 X 10-3 cm, which can be obtained by very strongly adsorbing complexes at solid electrode surfaces.11 The use of even faster sweep rates up to about 30 V s-1 allows less strongly adsorbed reactants to be studied at 1 higher bulk concentrations. Beyond 30 V s— relatively little is gained in detection limit or maximum bulk Figure 8. 61 Useful range of experimental parameters for reactant adsorption measurements by fast cyclic voltammetry. The numbers on the curves are the surface concentrations of reactant in mol cm-z. The useful range lies between the curves . 62 1 40 x 10‘ s x 10'" l . 60 Sweep Rate/ Vsec‘1 Figure 8 80 63 concentration by increasing the sweep rate, while the contributions of double layer charging and uncompensated resistance become increasingly important.109 The practical ranges are summarized as 2 :vi30 (Vs-1) (3.8) A i Cb i 100 (HM) (3-9) for K 3 1.5 x 10‘1I (cm) (3.10) 2. Adsorption of Transition Metal Complexes Determined by Fast Cyclic Voltammetry The results of adsorption measurements by fast cyclic voltammetry are listed in Table I for those complexes for which a reduction peak was observed under the experimental conditions set by Equations 3.8 and 3.9. Ep is the reduc- tion peak potential for the adsorbed reactant. All of the complexes with the exception of Co(NH3)5pyrazine3+ were found to have reached maximum adsorption at bulk concen- trations of 100 uM or less. An uncertainty of about :1 x 10"11 mol cm'2 is present in the Pi values. This uncertainty is primarily a result of errors in the assign- ment of a "baseline" and in the reading of faradaic cur- rents from photographs of the oscilloscope display. The Co(NH3)5pyz3+ result is near the detection limit for this 6A Table I. Fast Cyclic Voltammetry Results at Platinum and Gold.a Cb v 11x10ll Ep Complex (pM) Electrode (V s‘l) (mol cm'2) (mV) g—Co(en)2(NCS); 10 Au 10 17.6 —305 10 20 20.6 -305 20 10 27.5 —285 50 20 23.6 -295 100 20 25.9 -300 Co(NH3)5NCS2+ 30 Au 10 18.8 -1u5 no 10 19.A —1u0 Co(NH3)5pyz3+ 100 Au 10 6.9 o 100 20 5.8 — A5 Co(NH3)5012+ 20 Pt 10 38.5 + 95 80 10 no.6 +1A5 + Co(NH3)SSOu 80 Au 5 22.6 -150 Cr(OH2)5Br2+ A8 Au 2 58.8 + 70 A8 10 65.5 - 30 72 10 77.1 - 50 100 10 78.8 - 50 aMeasurements were made in 0.1M NaClOu + 5 mM HClOu at 25°C. ben = ethylenediamine, pyz = pyrazine. 65 technique. As expected, Co(NH3)5oH3+, a known outer-sphere 39 complex, gave no peak under these experimental condi— tions. Other complexes which failed to exhibit adsorption peaks are as follows: for gold; Co(NH3)SCl2+, for platinum; 2+ Co(NH3)5NCS , gfco(en)2(NCS)+, Co(NH3)5pyz3+, and 2+ Cr(0H2)5Br ; for neither surface; p—Co(en)201+, ngo(NH3)u_ C15, Co(NH315pyridine3+, and Cr(OH2)5C12+. There is other evidence of adsorption for some of these complexes (Chapters IV, V and VI). Reactant adsorption has therefore been found to be a necessary, but not sufficient, condition for the appearance of an adsorption peak. A qualitative difference between the halide-bridged and thiocyanate-bridged complexes is noted in the much smaller values of limiting surface concentrations for the latter complexes. For a smooth electrode surface (rough- ness factor of 1.2, see Chapter II) there are approximately 2 x 10’9 meter.110 Apportioning surface atoms equally between ad- moles of surface metal atoms per square centi- sorbed complexes yields A, 5 and 9 surface atoms for each 2+ 2+ 2+ adsorbed ion, Cr(OH2)SBr , Co(NH3)5Cl and Co(NH3)5NCS respectively. Either the thiocyanate and sulfate complexes require more surface area or the types of surface sites needed for adsorption of these two groups of complexes are different. This latter conclusion seems more likely since the morphology of the complexes are not so different. However, the argument for different adsorption sites is 66 not a strong one from such a small number of results. The values for Pi in Table I are used later to determine the amount of adsorbed reactant present for inner-sphere electron transfer kinetics measurements (Chapters V and VI). The cyclic voltammograms were also analyzed in order to calculate the rate of electron transfer for the adsorbed reactant. This aspect is discussed in Chapter V. CHAPTER IV KINETIC PROBES OF COMPLEX ION ADSORPTION A. Introduction The most interesting catalytic behavior is expected for those reactants which are reduced (or oxidized) via inner-sphere mechanisms at solid electrodes. In hetero- geneous electron transfer involving transition metal com- plexes this requires that the reactant be bound to the electrode surface by one or more of its coordinated ligands when electron transfer occurs. Experimental detection of reactant adsorption (or equivalently, of an inner—sphere mechanism) is achieved by various means in this work. This chapter describes the use of outer—sphere "probe" reactions in the detection of Cr(III) complex adsorption at mercury, platinum and gold electrode surfaces. Outer—sphere reactants are known to exhibit kinetic shifts with changes in surface charge due to various amounts of adsorbed simple anions, such as chloride and thiocyanate, 35,36,112 on mercury electrodes. The kinetic effects of adsorbed tetra-alkylammonium ions have been studied for 113 vanadium(III) and europium(III)llu reductions. It has generally been found that small amounts of large ions such 67 68 as the tetra-alkylammonium ions retard cation reductions as anticipated from a simple electrostatic model.115 At higher adsorbed ion concentrations blocking effects and ion—ion association become more significant.113’115 For the ad— sorption of simple anions on mercury electrodes the ob- served kinetic shifts in the "probe" reduction rate were shown to be in accordance with the Gouy-Chapman-Stern- Frumkin (GCSF) model of the double layer.35’36 This work is an extension of that study to include complex ion ad- sorption at mercury, platinum and gold electrodes. The probes chosen were Co(NH3)2+, Co(NH3)50H3+, and CO(NH3)5F2+. These reactants are known to be reduced via 36 outer—sphere mechanisms at mercury and also at platinum and gold electrodes (Chapter V).39 They have been used successfully as probes of anion adsorption at mercury.35’36’112 The reduction products (Co(0H2)§+, F", H20, and NHZ) are not likely to be adsorbed and therefore should not inter- fere with the probe kinetics. A series of chromium(III) isothiocyanato complexes were chosen as adsorbates to characterize the technique at mercury electrodes. These complexes are substitutionally inert and are reduced at fairly negative potentials on mercury. These two factors are important since they allow these complexes to be ad- sorbed but not reduced at potentials where the probes are reduced. The adsorption of many chromium(III) isothio- cyanato complexes has previously been determined at 69 mercury by single-step chronocoulometry.91’93’9111'116 Chronocoulometry was also used in this study to determine adsorption under the appropriate experimental conditions. B. Results The kinetic probe results at a dropping mercury elec- trode (DME) are summarized in Table II. The reduction kinetics of Co(NH3)g+ and Co(NH3)5F2+ were monitored in the presence and absence of the indicated chromium complex by d.c. polarography. Values for the apparent heterogeneous rate constant, kapp’ were calculated from the conventional Koutecky analysis,7u’75 and tabulated as logarithms of the rate shift at constant electrode potential (in this case, -u00 mV against the saturated calomel electrode, SCE). All charge densities in Table II were also reported for E = -A00 mV. The supporting electrolyte was 0.1 M KPF6 in order to minimize supporting electrolyte ion adsorption.117 The change in the diffuse layer charge density as a consequence of chromium complex ion adsorption was found 91,92 from -Aqd = 2 FT + Aqm (A.1) where zCr refers to the net ionic charge of the chromium complex, and P to its surface concentration in mol cm-2 Cr The change in the electrode electronic charge density as 70 .HQ mocohomom Eopmo .mm oocogomom Eonmn .mom mm >s coat I m ..oow m n oeHc_ooso .oomm no page a H.o cHo 3 Ameoxo V* m.mx mm.m m.mx mm.c a.mm+ oo=.o a.ca o.om m.o IoAmozvso om.HI m:.m| oo.m| mo.:| F.HHI noa.o m.H N.HH m.o mAmonmAcovson oo.HI me.mu om.HI no.2- m.HHI omH.o s.H m.HH m.o mAmozvaAmmzvaoIm 03.0- ao.m- mo.ou mm.m- a.m- omH.o m.H :.m m.o mgmozvaAmmovaoum ma.on ma.o- OH.OI mo.HI o.m- ofi.oe m.o o.m m.o mfimozvmfisovnoum oa.o- mm.HI ma.ou mm.m- m.m- oH.oe m.o m.m m.o +mm02mgmmzvaO mm.o- me.n- ms.o- ma.m- m.m- omm.o m.m m.oa m.o manozomlmmovaoIM . m m o 0 ea m o +mA mszw dam capo dam oamo m Naev onoEoo x mo mo * < * * mo: m 0m 2 mob? o m 3 pg 1W3 53 so; .280 consong .osoos +mm A mzvoo .oaoos +mA mzvoo e m.mzo m pm muHSmom mochm afiuwcfix .HH mfinme 71 a result of the bonding of an adsorbate, Aqm, can be cal- culated from Ff r and (Aqm/FGPCr) coefficients reported 91,92 C by Weaver and Anson. The values of these parameters are also given in Table II. The adsorbed charge density due to the chromium complex, FFCr’ is given by m = ' — FI‘Cr FFCr Aq (A.2) where Fér is the apparent surface concentration as measured 91.92 by chronocoulometry. The apparent adsorbed charge density was determined at the same DME and at the same drop time that was used for the kinetics measurements. Substi- tution of Equation A.2 into Equation A.1 yields -Aqd = - 2C Fr' + (z - 1ndm (u.3) r Cr Cr Examination of this relationship reveals that Aqm need not be known for an adsorbate which has a net ionic charge of +1. This could be valuable for solid electrode experiments where Aqm is usually not known. At E = -A00 mV, qd(0.l M KPF6) z 0.118 Under these conditions Aqd = qd. A study by Frank, 23 21-119 indicated that multiligand- bridged isothiocyanato complex adsorption was slow, but that the monobridged complex adsorption rate was limited only by diffusion of the adsorbate to the electrode sur- face. The time required to reach limiting adsorption 72 under diffusion control can be calculated, and is found to be less than 5 x 10.6 5.120 As a consequence of this, the adsorption of mono-bridged isothiocyanato complexes is not expected to exhibit any time-dependent adsorption behavior in DME experiments. The much slower adsorption of the multi-bridged complexes could cause substantial deviations in adsorbed charge density measurements on the timescale of DME measurement times (20-50 msec). The ad— sorption and kinetics measurements were performed at the same drop times. The charge densities obtained at a DME are noted to be less than those reported by Frank119 at an HMDE where long equilibration times are possible. The change in the potential across the diffuse layer corresponding to the change in charge density can be cal— culated according to the Gouy-Chapman theory for a sym- metrical (+z:-z) electrolyte as121 Aod = £52 sinh_l(Aqd/(8kTee n)l/2) (A.A) zee 0 which for dilute aqueous solutions at 25°C simplifies to Aod = (l9.5ze)-lsinh-1(Aqd/11.701/2) (A.Aa) Here Z8 is the magnitude of the electrolyte ionic charge (unity for KPF6), Aqd is in uC cm-Z, and C is the concentra- 1 tion in moles liter- of the supporting electrolyte. From 73 these potential changes the theoretical changes in the electron transfer rate constants may be calculated employ- ing the Frumkin double layer correction:122 (an - Zr)F “05 kcalc = 2.303RT Ac1’s (14.5) where a is the intrinsic transfer coefficient (assumed to be 0.5 for the outer-sphere probeng), n is the number of electrons transferred and zr is the net ionic charge for the reactant. The experimental values of log ka (Table II) pp were found to be reproducible to within about 0.05 log units by repetitive measurements. The sensitivity of a kinetic probe reaction for a single adsorbate at various concentrations was determined with Co(NH3)50H3+ as a probe of pigeCr(NH3)u(NCS); adsorp_ tion on a DME. The results of this experiment are plotted in Figure 9 as the rate shifts versus adsorbed charge density from chronocoulometry at -200 mV against the NaSCE. An electrode potential of —200 mV was chosen rather than -A00 mV (as for the results in Table II) to compensate for the faster reduction kinetics of Co(NH3)50H3+ relative to the hexaammine and fluoro complexes. The supporting electrolyte was 98 mM NaClOu + 2 mM HClOu. The kinetic response is observed to be linear with respect to adsorbed charge density in the measured range corresponding to Figure 9. 7A Change in the rate constant for reduction of Co(NH3)50H3+ as a function of apparent sur— face charge density of cis— Cr(NH 3)u(NCS)2 at a DME. Rate constants were found by d. c. polarography. Adsorbed charge density was determined by single-step chronocoulometry at -200 mV. 75 Fr'xpc cm'2 Figure 9 76 bulk chromium complex concentrations from 8 to 2A0 UM. The line is given by -Alog kapp = 0.172F1‘ér + 0.05. Table III contains the rate shifts for Co(NH3)50H§+ reduction at 0 mV against the NaSCE for various chromium- (III) complexes at platinum and gold electrodes. The use Of Co(NH3)2+ OP CO(NH3)5F2+ as probes on these surfaces is precluded by the proximity of their reduction waves to the negative potential limit. The reduction rate constants were found from rotating-disk voltammetric data for the solid electrodes. The chromium complexes are not reduced on platinum or gold electrodes in the useful potential range, l;§;’ between the potentials for electrode surface oxidation and hydrogen evolution. Chronocoulometric data would therefore be unobtainable (apart from the difficul- ties described in Chapter III for chronocoulometry at solid electrodes). A Gouy-Chapman calculation (employing Equation A.Aa) was performed to determine whether the bulk concentration of nonadsorbed chromium complex would significantly con- tribute to the double layer potential and thereby affect the observed kinetic shift. For all bulk concentrations employed (< 1 mM) this contribution was found to be less than one percent of the potential drop across the diffuse layer, A¢d, and hence essentially negligible in Alog k calc determinations (Equation A.5). 77 Table III. Kinetic Probes of Adsorption at Platinum and Gold: a5Co(NH3) OH3+ Reduction at 0 mV vs NaSCE.a Conc. Alog k Alog k Adsorbed Complex (mM) Pt Au Cr(NH3)g+ 0.5 +0.10 -0.10 p—Cr(NH3)u(NCS): 0.5 -0.u0 —2.A0 Cr(NH3)5Ncs2+ 0.5 -1.00 -1.60 Cr(NH3)5012+ 0.5 -1.00 —0.80 2+ 0.5 m—3 m-3 Cr(NH3)5Br 3In 0.1 M NaClOu + 2 mM H010“ at 23:100, measured by rotat— ing disk voltammetry. 78 C. Discussion The linear response depicted in Figure 9 justifies the experimental method by which rate shifts and adsorbed charges were obtained, especially since pig-Cr(NH3)u(NCS); is probably di-bridged and may be expected to be sensitive 119 If the outer- to any time-dependent adsorption effects. sphere probes interact with their environment in a purely electrostatic fashion, rate decreases would be expected for cationic probes of cationic adsorbates. The magnitude _of the rate decrease is larger for higher cationic adsorp- tion. There is no indication of coverage-related effects other than that for adsorbed charge. Blocking of the re- actant by a large adsorbate to force a more distanc reac- tion plane would be one such effect. This would attenuate the kinetic double layer effects at higher adsorbate coverages.ll3"115 This situation has been noted by othersll3-115 for tetra-alkylammonium ion adsorption, but is not observed here for the chromium complexes. Blocking effects may, however, occur at higher adsorbate concentra- tions than were obtained in this experiment. The experimental values for the rate shifts reported in Table II are at least qualitatively correct in that cationic adsorbates cause rate decreases, and the anionic adsorbate Cr(NCS)2- causes a rate increase. The rate en- hancement by Cr(NCS)2- is large enough to shift the probe reductions into the mercury dissolution(positive potential) 79 limit, and could only be evaluated as a minimum value. The actual rate shift would probably not be much larger in any case, since ion pairing would act to decrease the enhancement. As expected from the double layer calcula- tions the presence of 0.5 mM Cr(NH3)g+, which is not ad- sorbed, does not measurably influence the probe reduction rates. The neutral adsorbate, fag-Cr(OH2)3(NCS)3 was ad— sorbed, and gayea a value of FPCr = 16.3 uC cm-2, but be— d cause 2C is zero,Aq = Aqm (from Equation A.3). r The experimental rate shifts seem to be only tenuously related to the adsorbate net ionic charge 2 This is Cr apparent in the large discrepancy between observed and cal- culated values for Alog k. In every case the calculated values for the rate shifts are found to be of much greater magnitude than those found experimentally. There may be several reasons for this discrepancy. It is probably naive to ignore the internal charge distributions of the chromium adsorbates, especially since the chromium complexes must extend about 6.8 A into the solution from the electrode surface.123 The diffuse layer itself is approximately 10 A thick in 0.1 M electrolyte. The reaction plane of the outer-sphere probes probably lies within or near the 6.8 A distance, and they would experience a very different electrostatic environment than a uniform layer of adsorbed charge. Such discrete charge effects are difficult to isolate or treat theoretically. Another possible cause 80 of the discrepancy is that the Gouy-Chapman theory may be inadequate when dealing with relatively large adsorbed charges.2’121 The plane of closest approach could be more distant in the presence of the adsorbed complex. The extra distance required to diminish the double layer effect to the point where experimental and calculated rate shifts agree can be found froml2u tanh(z f¢ /A) x = - fiL-1AI e app (1.6) D tanh(zef¢d/A) where KD’ the Debye-Huckel reciprocal length is KD = (8nz:FfC/e)l/2 = (3.29 x 107)zec1/2 (1.7) for dilute solutions at 25°C. The notations are: f = F/RT, ¢app and ¢d are the apparent and calculated (from Equation A.Aa) diffuse layer potentials, s is the bulk dielectric constant, and C is the bulk electrolyte concentration in 1 mol l- . A0 is obtained from Alog ka values and Equa- app Pp tlon A.5. A¢d and A¢app give 0d and ¢app respectively when the contribution to the potential due to the supporting electrolyte is zero. The distance x that the reaction plane would need to move for the various chromium com— plexes with Co(NH3)g+ as the outer-sphere probe are 81 listed in Table IV. A glance at these x values indicates that the discrepancy between theory and experiment cannot be entirely due to a more distant reaction plane, as the distances required would put the outer-sphere reactant an unreasonably long distance beyond the diffuse layer. The actual cause or causes are likely to be a good deal more complex than suggested here and actually are not the primary goal of this study. Several striking and useful results were nevertheless observed. It is clear that the kinetic probes actually can detect the adsorption of complex ions, and thereby aid in the identification of inner-sphere electron transfer mechanisms. For isothiocyanato complexes this method is most satisfactory for those complexes capable of forming multiple ligand- surface bonds, and marginally successful for the mono- bridged complexes. As indicated by Figure 9 the probe is sensitive to very small amounts (ml uC cm'2) of adsorbed charge for the di-bridged gig-Cr(NH3)u(NCS); complex. Even though the magnitudes of the probe kinetics shifts differ for different adsorbed complexes present at nearly the same amount of adsorbed charge; the shifts for each individual adsorbate are probably simply related to its adsorbed charge density. If this is the case the technique could be made more quantitative with calibration plots similar to Figure 9. Table III shows that the results at platinum and gold 82 Table IV. Apparent Change in the Reaction Plane of 3+ CO (511) l 0 1-0 ' To meet the requirement for the log expansion, the surface coverage is limited to the range 0 i O i 0.5. Substitution of Equation 5.1A into 5.13 yields the interesting result of m E l — O (0 5 0 g 0.5) (5.15) Thus in the absence of adsorbate interactions, under steady-state conditions,at a constant electrode potential, the overall reaction order is simply and directly related 9A to the surface coverage of the adsorbed intermediate by Equation 5.15. Adsorbate-adsorbate interactions are neglected. Such interaction can be either attractive 105’107 Since these interactions are or repulsive. not known at platinum or gold, their inclusion into the analysis would have little purpose. Equation 5.15 is not meant to show more than the fact that an adsorbed inter- mediate can lead to lower overall reaction orders. While reactant adsorption is certainly not the sole possible cause of depressed-reaction orders, it seems a very probable one for the cobalt(III) ammine complexes which offer few mechanistic tricks. Because they exhibit reaction orders of 0.5 or lower (Table V), the complexes Co(NH3)5X2+ (where X = Br', Cl’, Ncs'), tho(en)2C1: and p-Co(en)2(Ncs); are probably ad— sorbed to half coverage or greater on clean platinum and gold surfaces. By the same reasoning an inner-sphere mechanism appears to be blocked by iodide on the iodide- treated surfaces, where reaction orders near unity are ob- served. C. Outer-Sphere Kinetics of Co(III) Ammines In heterogeneous outer—sphere electron transfer the reactant is expected to interact only weakly with the 35,39,125 electrode surface. The plane of reaction is by definition at least one solvent molecule diameter plus a 95 reactant radius from the surface.35 This does not mean that the kinetics for electron transfer are insensitive to the electrode condition. On the contrary, outer-sphere reactants are quite responsive to electrode charge and 39,125 structure. The sensitivity of the outer—sphere reduction of cobalt(III) ammines to the charge on the electrode has been used to determine double layer potentials at mercury 35 electrodes. These potentials were found to compare favorably to those predicted on the basis of the Gouy- 35,121 Chapman-Stern—Frumkin theory. The observed (apparent) rate constant can then be corrected for the effect of the 36 ionic double layer. The corrected rate constant for a one-electron reaction is related to the apparent rate constant by36 log kE corr + (F/2-303 RT) (Zr i aI)¢: (5.16) E - log kapp where a is the intrinsic transfer coefficient, which is I probably close to 0.5 for these reactants.35’36 ¢E is the average potential at the reaction plane for the r given electrode potential E.126 (The plus/minus signs The term refer to electrooxidation and electroreduction reactions respectively.) The plots of log kcorr for the reduction of Co(NH3)g+ and Co(NH3)5F2+ at mercury are depicted in Figure 11 as dashed and dotted-dashed lines, respectively.36’39 Figure 11. 96 Rate-potential (Tafel) plots for Co(NH3)5F2+ and Co(NH3)g+ reductions at platinum and mercury in 0.1 M NaClOu. Solid lines refer to platinum rate constants. Dashed lines are double layer corrected rate constants at mercury electrodes. 97 I=°~ j 23” L E/mV ve.SCE Figure 11 98 The linear regression lines of log kapp for these two com- plexes at platinum electrodes are shown as the solid lines. These are linear regression lines representing averages of six to eight rate determinations at platinum. The slowest measurable RDE rate constant is about 10.5 cm 5'1. The rates in Figure 11 correspond to an extrapolation of the data to more positive potentials, where most of the rate data for the other cobalt complexes are measured.10’ll The comparison of sblid electrode rates to those at mercury can be used to calculate double layer potentials at the solid electrodes.10 Rate comparisons between dif- ferent types of electrode material must be made with care. There are a number of ways in which the nature of the electrode material can affect the rates and mechanisms of 10,39,125 electrode reactions. These effects may be rela- tively small if the reactions occur by an outer—sphere 39,125,127 mechanism on both surfaces. In such reactions, the variation of electrode material may lead to a change 125,127 in the ionic double layer effect, a change in the tunneling probability in the transition state (K in Equa- 128’129 or a different solvent structure at the 128,130 tion 1.6), All of these factors t.39 metal-solution interface. have some influence upon the apparent rate constan The variation of kapp with electrode material for a one- electron reaction at a given electrode potential can be expressed aslo’39 99 (Alog k = -(F/2.303RT)(zr i aI)A¢E + M (5.17) app)E where A0: is the change in the average potential at the reaction plane at a constant electrode potential. The first term is a form of the conventional Frumkin double 2,130 layer correction as it was applied in Equation 5.16. The M term contains any variation in the "specific" ef- fectsl28’l3o brought about by changes in the electrode material. These specific effects include those due to tunneling surface roughness and solvent interfacial struc- ture. By comparing values of (Alog k )E for two complexes app having different values Of Zr and similar ligand structure, we can eliminate the M term from Equation 5.17. It is assumed that outer-sphere reactants of similar structure, such as Co(NH3)2+ and Co(NH3)5F2+, would experience similar reaction environments, i.e., M for Co(NH3)g+ is the same as for Co(NH3)5F2+.35’39 The resulting value of E can then be used to calculate M from Equation 5.17. Values of M found by this method were near zero (10.1) for Pt, Pt(I) and Au(I) surfaces when compared to mercury. On gold M was found to be +0.3 relative to mercury. The assumption of M = 0 for these complexes seems reasonable for Pt, Pt(I) and Au(I) surfaces. Even for gold surfaces this assumption would only cause an error of about 10 mV in the value of A0: obtained (as calculated for Co(NH3)g+ at E i 100 mV). 100 Instead of comparing apparent rate constants at both surfaces, kEpp at the solid electrode can be compared to E k at mercury. Since k v -_ corr alues represent the condl COI’I’ tion of ¢E = 0, Equation 5.17 becomes (log ka - log ng E pp corr)E = -(F/2.303RT)(zr i aI)¢r (5,18) The value of ¢E from Equation 5.18 is the average reaction plane potential for the solid electrode. From data like those in Figure 11 for all four surfaces (Pt, Au, Pt(I), and Au(I)), $5 was calculated using Equation 5.18. These values are listed in Table VI for various electrode poten- tials. It can be seen that ¢r is more negative at these solid surfaces than at mercury. The differences between values of ¢r calculated from Co(NH3)g+ reduction and those 2+ are attributed obtained from the reduction of Co(NH3)5F to nonzero values for M in Equation 5.17 or to slightly different reaction planes for these two reactants at solid electrodes.39 To fit this latter explanation the reaction plane for Co(NH3)SF2+ would have to be closer to the surface than for CO(NH3)2+. The distance required is small, as can be calculated from Equation A.6. For example, at platinum for E = 0, the Co(NH3)5F2+ reaction site would need to lie 0.6 A closer than that for Co(NH3)2+. The absolute magnitudes of or on solid electrodes are considerably larger than on mercury electrodes at the same 101 Table VI: ¢E in 0.1 M NaClOu at Various Electrode Po— tentials E.a E (mV) Reactant Surface -100 0 +250C +u00° Co(NH3)2+ Pt -61 -55 -11 -32 Au -67 -57 -32 -18 Pt(I) -62 -56 -39 -29 Au(I) -69 —53 _12 +11 Hgb —25 _19 _ u Co(NH3)5F2+ Pt —63 —60 —5A -50 Au -78 -69 -A6 -33 Pt(I) -73 -65 —A6 -A0 Au(I) -81 —63 -19 + 9 Hgb -21 -18 — 3 aUnits of ¢E are in mV. bCalculated from data presented in References 35 and 39 using Equation 5.18. 0These ¢r values are from extrapolated Tafel plots of the reactant reduction rates. 102 electrode potential. This difference is primarily due to the different potentials of zero charge (PZC). The PZC 131 for mercury in 0.1 M NaClOu is -A57 mV. On solid elec- 97 trodes PZC values are less easily obtained. Values of PZC for polycrystalline platinum in 0.1 M NaClOLl have been 132 18,13A,135 -50,133 and 0 mV. For poly- 136 reported as -110, crystalline gold electrodes in the same electrolyte -250, -200,18 -80,87 —A0,u9 and 0 mV133 are given for the PZC in the literature. Despite the diversity of these values, it is clear that the PZC for mercury lies at least 300 to A00 mV more negative than for platinum and gold. The po- tentials at which ¢r is reported in Table VI for the solid surfaces are all near or positive of the PZC, while at the same potentials ¢r for mercury is several hundred milli- volts more positive of the PZC. It is known that ¢r for mercury becomes increasingly negative as the electrode potential approaches the PZC from the positive direc- tion.35’131 Compared at the same potential difference with respect to the PZC, the values of ¢r for platinum and gold would be much closer to the value of ¢r for mercury. The implication is that 010”“ adsorption on platinum and gold is similar to mercury. Adsorption studies support 18,A9,86 this conclusion. The adsorption of ClOu— from 0.2 M HClOu on gold and platinum has been reported to be too low to be detected by ellipsometry.86 Capacitance measurements in 50 mM_KClOu have shown no specific adsorption 103 of ClOu' on gold near the PZC (-A0 mV 1 10 mV).119 Per- chlorate is found to be weakly adsorbed on platinum and gold at potentials in the vicinity of the potential for oxide reduction (E z +A00 mV) by cyclic voltammetry.18 It therefore seems reasonable to assume weak or no C10“- adsorption at E = 0, with perhaps slightly more adsorption at more positive potentials from 0.1 M NaClOu. The cobalt ammine outer-sphere kinetics results are summarized in Tables VII and VIII for platinum and gold electrodes in 0.1 M acidified NaClOu. The rate parameters for each complex are given for the clean surface, iodide- treated surface and for mercury. The apparent rate constants were measured by rotating disk voltammetry at the listed electrode potential E. For these complexes, excepting CO(NH3)5NO§+, at least four rate determinations have been averaged to obtain the listed values. (The nitrato complex exhibited a large hysteresis in the "steady-state" kinetics at gold electrodes and is not reported. The results for this complex in Table VIII are averages of only two experi— ments.) The standard deviations for these averages are less than 20% of kapp’ except for those rate constants marked by an asterisk, which had a precision 550% of kapp‘ Tables VII and VIII also list values for the apparent trans- fer coefficient a defined by app’ aapp = -(2.303 RT/F)(310g kapp/BE)u (5.19) 10A Table VII. Reduction Rate Parameters for Co(NH3)g+ and 2+ Co NH = . ( 3)5F at E 0 c k kapp -l Kp etl Complex Electrode (cm sec ) “app (cm) (sec ) Co(NH3)g+ Pt 2.7x10'5 0.65 2.6x10’5 1.0 Pt(I) 2.8x10‘5 0.67 2.7x10‘5 " Au 3.2x10'5 0.75 3.2x10'5 " Au(I) 2.0x10'5 0.91 1.9x10'5 " Hgb 7.1x10'7 0.67 3.3x10‘7 " Co(NH3)5F2+ Pt 2.8x10'5 0.5M 11.0x10’6 7.0 Pt(I) 3.8xio‘5 0.62 5.9xlOT6 " ....‘X’ .. Au A.9x10 5 0.6A 8.Ax10 6 " Au(I) 3.3xlo"5 0.77 5.23.10"6 ." Hgb 2.2x10‘6 0.59 l.LIx10‘7 " aIn 0.1 M NaClOu + 5 mM H010“. bFrom Reference 39. CAn asterisk indicates a precision of 50%-20%, all other rate constants have precisions better than 20%. 105 mums socpo Ham .mom can» Lennon m20finfioosd o>mn mucmpmcoo .uomnom mo cofimfiooaa m mbudoapcfi xmfiaopmm c< .mm mocmhmhmm EOLEO w p .>Ho>fipooomop .+m no +m mm: owpmno pcmpomoa one Lennon: :0 wcfipcoaop .ae Lon modam> +mmmAmmzvoo so +mAmmzvoo Eopm popmefiummn Ocahmm pm .zoaom SE m + soHowz a H.o :Hm moax>.w :OHxH.H sloaxa.a mm.o *mloaxo.fi omm AHvsa moaxm.a moaxm.m mIonm.H m>.o mIOme.m omm s< :oaxm.m owe onoaxm.m mm.o emloax~.m omm AHVum :oaxm.a osa cloaxz.: mm.o zloax:.~ omm pm +mflmmovafimmzvoolm m: m: oIOme.m mm.o *zloaxm.m 0 Avam mu ms mnoaxo.: mm.o :Ioaxo.m 0 pm +m02mfimmzvoo com com suoexm.m om.o auoexo.fl 0 one cam cam m-oexm.e aw.o muoexm.o o AHvse omm omm mIOme.m 35.0 emuoaxa.w o 3< mm mm mloax~.m mm.o mloaxm.m o AHvum me me m-oaxp.m mm.o cmuoaxe.e o as +mmo mammzvoo AHImV AHImV AEoV moms AHIm Eov A>Ev opoppooam onano I pm pm Q Qua AOImV x x n x p x m .moonQEoo AHHHVpHmnoo oaonomlsopso Lonpo Lou msopoempmm opmm coaposoom .HHH> canoe 106 The partial derivative is the slope of the log kapp versus potential plot (as in Figure 11) at constant solution com- position. The equilibrium constant for precursor state formation Kp was described and defined (by Equation 1.3) for outer-sphere reactions in Chapter I. The 0: values in Table VI were used to calculate Kp from Equation 1.3 (with 8 r = 3.A x 10' cm).123 Complexes of the same charge 2 r were assumed to experience the same reaction plane poten- 35,127 tial ¢r' The rate constant for electron transfer from the precursor state ke was calculated from the ratio t of kapp to Kp (Equation 1.5). The final column lists the values of ket measured or extrapolated to an electrode po— tential of 0 mV. The rate constants are reported at po- tentials positive of, or into, the potential range for the kinetically controlled reduction of complexes. The constant Kp can be more accurately estimated when the reactant con- centration at the reaction plane is controlled by RDE hydrodynamics and kinetics rather than by diffusion and kinetics. Once ke is found, a t can be used to extrapolate t ket to some common potential for comparison. The slope 3+ (310g ket/BE) and Get were found from the Co(NH3)6 rate e constants at 0 and +250 mV. det was determined to be 0.A6, 0.A5, 0.A6 and 0.A2 for Pt, Au, Pt(I) and Au(I) surfaces, respectively. These values were used in the extrapolation of ket' The ke values for Co(NH3)g+ are all found to be quite 1 t close to 1.0 sec- At first glance this seems to be an 107 artifact of the calculation for ket‘ However, by combining Equations 1.3, 1.5 and 5.18 one can arrive at the following relationship for ket: ket = fitkgpp111'i)tk§§rrli (5.20) where A = Zr/(zr i cl), and kgpp is the rate constant measured at the solid metal electrode. From Equation 5.20 it is clear that ket maintains a dependence on the apparent rate constant. It can be noted that ket values are fairly similar for the same complex regardless of the electrode material. The relatively small differences can be at- tributed to uncertainties in ka The Co(NH3)50Hg+ complex 1313' shows the largest relative differences in ket as electrode material is varied, but it also exhibits the least repro- ducible values for kapp' The lack of any strong dependence on electrode material for these complexes reinforces the use of such complexes as probes of double layer structure. The different values of ket for different complexes are expected, if the complexes have different standard poten- tials. Equations 1.5 and 1.6 indicate that ke depends on t the reorganization energy (AG*)E, which in turn is dependent upon (E - E°) (from Equation 1.7). Driving force dif- ferences are more easily illustrated in the comparisons of heterogeneous and homogeneous reduction rates which are presented later in this chapter. 108 D. Unusual Aspects of Iodide—Treated Platinum and Gold Electrodes The interesting character of the iodide-treated elec- trodes is evident in Tables VII and VIII. It is noted that at E = 0, kapp(clean) a kapp(I). Nearly equivalent apparent rate constants lead to similar ¢r values from Equation 5.18. Under these conditions the adsorbed iodide does little to perturb the reaction plane potentials. Scrutiny of Table VI shows this remains the case for platinum over the potential range -100 g E 3 A00 mV. The diffuse layer 8A,137 charge for the iodide monolayer must be nearly the same as that due to any perchlorate adsorption in the ab- sence of iodide. This is also true for gold, but only for E i 0. Considerable electronic charge appears to be trans- ferred from the iodide monolayer to the platinum surface, resulting in a practically covalent Pt-I bond.811 Low energy electron diffraction (LEED) studies have confirmed the chemically irreversible adsorption of iodide on plat- inum in a ratio of one iodine atom per two platinum surface atoms.137 Bagotzky, gt a1. find a 1:1 ratio from 1 mM I-.52 138 Radiotracer6O and chronocoulometry experiments indicate a maximum iodide coverage of 1.5 and 1.6 x 10-9 g ion cm-2, respectively. (A smooth platinum surface has 2 2.1 x 10"9 g atom Pt cm- .)l37 The covalent "iodine atom" layer may explain the insensitivity of ¢E to E as observed 109 for Pt(I) surfaces. The nature of iodide on gold has been less thoroughly studied. The monolayer adsorption of iodide on gold has been proposed from differential capacitance data.L17 Specular reflectance studies have supported this conclu- sion over a wide (-0.A to 1.0 V) potential range.50 The qualitative behavior of the cobalt ammine complexes at Au(I) for E = 0 is not much different from Pt(I) where monolayer adsorption is present. However, there is not any evidence for the covalency of an Au-I bond. The capacitance data of Hamelin117 is consistent with conven- tional anion specific adsorption.2 The estimated values for 0: on Au(I) in Table IV clearly diverge from those on the clean Au surface as the electrode potential becomes more positive. Increasing iodide adsorption may be oc- curring. The fact that Au(I) appears to affect outer— sphere kinetics similarly to Pt(I) around E = 0 may be coincidental. Alternatively, a covalent layer of iodine atoms may be adsorbed followed by further ionic iodide adsorption at positive potentials. Multilayer adsorption of iodide proceeds by this mechanism at platinum,29 but at present there is not enough evidence to support this explanation at gold surfaces. The strength of the iodide bond to both clean platinum and gold surfaces is sufficiently great to block other adsorbing species, as is evident from the reaction orders 110 seen in Table V for normally ligand—bridged reactants. Evidence of a more qualitative nature was noted in the en- hanced reproducibility experienced in kinetics measure- ments at iodide-treated surfaces. The steady-state kinetics remained stable for at least an hour without any repre- treatment of the electrode. Typically, clean surfaces required a pretreatment after each RDE determination (about every 10 min) to "re-activate" the electrode surface (see Chapter II for a more detailed discussion of pretreatments). Such behavior may be ascribed to an iodide layer blocking the adsorption of solution impurities which deactivate the electrode surface. Iodide-treated platinum electrodes have perhaps been inadvertently used in some previous platinum studies. R. N. Adams8O describes a pretreatment used to form a "reduced platinum surface" which consists of immersing an oxidized platinum electrode in an iodide solution to strip off sur- face oxide. He calls such a surface "stripped Pt". This treatment is followed by a water rinse. In light of later 60,8A,l37,138 work, once the oxide is stripped off, iodide adsorbs as a monolayer and cannot be removed by simply 8A,137 rinsing with water. Adams gives no specific ref- erences to work employing "stripped Pt", but they undoubtedly exist.80 The unusual nature of the Pt(I) surface (i.e., covalent adsorption), and the lack of effect on double layer potentials tend to make the iodide adsorption nearly 111 undetectable by kinetics measurements over the usual plat- inum potential region (-100 < E < A00 mV). Without more sophisticated surface analysis techniques,60’137 Pt(I) could easily be mistaken for an unusually stable reduced Pt surface.80 E. Transition-Metal Aquo Reactants After finding that the electrode material had only a minor influence upon the electroreduction of the cobalt- (III) ammine complexes, it might be expected that this would prove to be true for other simple cationic redox couples as well. However, quite different behavior was obtained for the one-electron oxidations of Cr:;, Vid’ Rug; on platinum, gold and mercury surfaces.39 The rate data are summarized in Table IX for these surfaces. An 2+ , examination of kapp for Craq ox1dation reveals striking differences between platinum or gold and mercury. The 2+ aq was found to occur on these solid elec- oxidation of Cr trodes at very positive electrode potentials (+700 to +1000 mV). The rate constants listed were determined by both cyclic voltammetry and rotating disk voltammetry. In order to compare the rates at platinum and gold to mercury at the same electrode potential, long linear extrapolations of the log ka versus E (Tafel) plots were necessary. Even though pp some inaccuracy is probably introduced by this technique, it is clear that the rate constants at -100 mV are at least 112 Amnoaxmv 00 I 0 I. .m Iofixm 0m I came 20.0 mm 00 I am m +m 2.02 mu0fixm.m 00H- +mn0 as H+ H.0A 00m- aoeooz 20.0 0< 0.02 mI0me.H 00H- + to as H+ : m I H.0x 000- oaosz 2m.0 ca Amm.00 Amuoaxmv 00H- 0 .l m noses 00H- peas za.0 mm 00:- > 0 +0 oaloae OOHI 0.0a mn0fixm.a 000 aoeosz am.0 m.0e :I0Hxs 000 aofiooz ma.0 sa 0I0He 00H- 0.0a mI0Hx0 000 00802 20.0 0-0He 00H- 0.02 muoaxa 000 acaooz ma.0 ca AON.00 Amuoaxmv 00HI 0 mm.0 NI0HA0.N 00H- ones 23.0 mm mm0- +Mh0 ammo AHIm Eov A>EV. poumaospooam opoppoofim A>EV pcmpomom 0 one o 0 x m m 0 .moonQEoo osv< mSOHnw> mo neopoemawm opmm coapmpfixo .xH manna 113 mp pocaesopop mm .o Uopoossoo somma oaosop son mam mononpcosma Q0 monam> .ooooo some can .0aam so soaom as m tonnes oocnoocoo oosaoacooam U .m oocosomom song ma.m soanmsvm moans popoossoo nomma mansop oumofipcfi mononpcohma CH mosam>o so SE H popem poumoaop mm: pcosfisoa Ixo one page mocaoeoca +ms0 as H+ 2 .am oocomogom Soap Hwapcopoa xoooa HmEpomm . x I L E m: o8 mIOH m om +N 0 S H+ mI0meA 0m I made 2m.0 sa m . . an x I s NIOH mA om max 2m 0 pm +N m ammo AHIm Eov A>EV nopzaospooam opospooam A>Ev economom 000 m 0m o x w .ooscanoo .xH canoe llA several orders of magnitude slower than for mercury («:106 7 to 10 -fold). Such a large depression of the oxidation rate at platinum and gold cannot be explained by a simple ionic double layer effect.39 The rate differences are not only too large, but of the opposite sign to those observed for the cobalt(III) ammine reactants. Table VI shows the double layer potentials on platinum and gold are favorable 2+ for cationic reactants for -100 i E i +A00 mV, and yet Craq oxidation is not noted until E Z 700 mV. Oxidation of Cr:; was also found to be immeasurably slow on silver elec- trodes (no oxidation up to the positive potential limit at 39,95 +250 mV). The conclusion is that this reaction ex- hibits a large, specific metal effect in contrast to the 39 The electrooxidations of V2; and Rug; were also studied and their rate constants are also given in Table IX. The 3+/2+ aq reversible), which limits the Ru:; oxidation rate determina- cobalt(III) ammine reductions. kinetics of the Ru couple are more rapid (quasi- tions to cyclic voltammetry at the formal potential (Ef 72 For these two reactions the rates are found = -20 mV). to be faster on platinum and gold than on mercury (contrast- ing sharply with Cr:; oxidation). These three electro- oxidations occur at similar rates on mercury electrodes (Table IX). It seemed possible that trace anionic impurities might be catalyzing the oxidation of V2; through an inner-sphere, 115 ligand-bridged mechanism. Because both VS; and V2; are substitution-labile, a small amount of an anionic impurity (ppgp, chloride) could participate in the catalysis of a much larger quantity of Via, since the bridging ligand can be rapidly released after electron transfer.39 Such catalysis is not possible for Crig oxidation, since Cr(III) is substitutionally inert and would "trap" the trace amounts of impurity in the Cr(III) product.3 The Ru:;/3+ couple is substitutionally inert in both oxidation states on the 2+ time scale of cyclic voltammetry.39’98’99 The use of Craq to remove these hypothetically catalytic impurities suggests a means of testing the mechanism of electrooxidation. The Crgg will ligand—bridge with, and incorporate the impuri— ties, into the Cr(III) product; thereby drastically reduc- ing their participation in other catalytic processes.39 The electrooxidation rate for substitution-labile V2; should be decreased, while that for substitution-inert Rug; should remain unchanged. The addition of 1 mM Crig did not significantly affect electrooxidation rates of aquo couples at mercury, or the electroreduction rates of 3 96 Co(NH3)6+ at platinum, gold, silver or mercury.39 This indicates that 1 mM Org; does not significantly affect the ionic double layer (i.e., is not specifically adsorbed, and does not perturb the double layer potentials through some interaction with the supporting electrolyte). Conse- quently, double layer effects are not likely to complicate 116 this test of mechanism. The addition of Org; was found to 2+ decrease the rate of electrooxidation of Vaq on platinum and gold as predicted. Table IX shows the rate constant to be decreased by N103éfold. However, the rate constant for Rug; oxidation also decreases by about this same amount on gold. By this result, the ligand bridging hypothesis is invalidated. 2+ 2+ 2+ The Vaq and Ruaq rates in the presence of Craq were substantially slower on platinum and gold than on mercury, just as was Crig itself. The original rates (no Cr2+) at platinum and gold seemed more reasonable, because they were near or larger than the mercury rates, in accordance with the predicted double layer effect. Hence, it is the behavior of Cr:; that appears most unusual. The depres- 2+ sion of the Ruaq rate is particularly problematical. It 2+ aq and the solid metal surface is depressing the oxidation seems clear that some specific interaction between Cr rates, (and probably acts to suppress its own oxidation as well). One may speculate that active surface sites impor- tant for aquo reactants, but not for Co(NH3)g+, are ad- versely affected by Cr:;.39 The most likely difference between aquo and ammine reactants is in their relative abilities to form hydrogen bonds with the solvent, or adsorbed solvent on electrode surfaces.39 Reaction entrOpies for the aquo couples indi- cate that considerable hydrogen bonding is present between 117 aquo ligands and solvent water molecules.139 The reaction plane for reduction of Cr(III) aquo complexes was proposed to be more distant from the electrode than for Cr(III) ammine complexes, presumably due to an additional water 36 83,128,1u0,1u1 molecule diameter. The surfaces of platinum 83,1A2 and gold probably are covered by a layer of specific- ally adsorbed water molecules. (Trassatti has determined that water is not preferentially oriented at gold near the 128,1A0 If water molecules at these solid electrode PZC). surfaces are adsorbed "oxygen down",1113 then hydrogen bond- ing with aquo ligands is disfavored. The electrostatic effect of the cationic reactant adjacent to the inner-layer 39 water molecules is uncertain. In order to explain the observed reactivity, water molecules would need to be more "oxygen down" on the solid electrodes than on mercury. The evidence for this is inconclusive at present.39’128’luo’1L13 F. Inner-Sphere Kinetics of Co(III) Ammines The cobalt(III) ammine complexes containing chloride, bromide, or pseudo-halide ligands have been found to be reduced by ligand-bridged mechanisms at platinum and gold.10,11 An inner-sphere mechanism can be assigned on the basis of adsorption measurements (Chapters III and IV) or on the observation of substantially depressed reaction orders (Section B, this chapter).11 Because these complexes are capable of large catalyses relative to their 118 corresponding outer-sphere pathways, it is desirable to understand the cause of their enhanced reactivity on plat- inum and gold. Tables X and XI summarize the kinetics data for the reduction of the inner-sphere complexes at platinum and gold. The format of these tables are similar to that des- cribed for Tables VII and VIII for the outer-sphere com- plexes. A notable difference is the inclusion of values for Cr and P the reactant concentration near the surface p’ and surface concentration, respectively. These are needed to calculate the equilibrium constant for precursor forma- tion, Kp, from Equation l.A. Under the "Complex" column the kinetics parameters are reported first for the clean surface, then for the iodide-treated surface (marked as (1)), followed by a row giving the ratio of the clean/(I) parameters when appropriate (marked as A). Surface con- centrations were determined either by fast cyclic voltam- metry (Chapter III), or they were estimated from the re- action order. As before (Section C, this chapter), ket was found from the ratio of kapp to Kp. In a few cases ket could be directly determined from fast cyclic voltammetry. Chapter III described the use of this technique to determine Pp. The same data can also be used to calculate ket.ll Under the conditions for fast cyclic voltammetry given in Chapter III, the observed cur- rent is proportional to the adsorbed reactant concentration. 119 00.0 00000.0 00000.0 0 00000.0 000 N10000.0 00.0 0-00xs.0 000 0.0 A00 000 00 0-0000.0 00.0 0-0000.0 000 00 00.0 W0000002000Im I0000.0 00000.0 00000.0 0 00000.0 00 N30000.0 00.0 0-0000.0 000 0.0 A00 000 0.0 0-0000.0 00.0 0-0000.0 000 00 00.0 W000Acovooum .0000.0 0000.0 00 4 00000.0 000 0-0000.0 00.0 0-0000.0 000 0.0 A00 000 00 0-00xm.0 00.0 m-0000.0 000 00 00.0 W000Aco0ooum 00 00000.0 00000.0 0 00 00.0 0-00xm.0 00.0 0-0000.0 000 0.0 A00 00000000 00000.0 0-0000.0 00.0 0-0000.0 000 00 00.0 +0000A002000 000 00000.0 00000.0 0 0.0 00.0 0-00xm.0 00.0 0-0000.0 000 0.0 0Am0 00000.0 00 0-00xm.0 00.0 0-0000.0 000 00 00.0 +000 A mzvoo A Imv A Imv A53 new A In :80 A25 :5 A :8 08.00050 0 no 00.0 am 0 0900 m 05 005mlo: voumv 0 0 s o 0 A0 W0000 a0 0 0 0 .mopoppooam EzcfipmHm pm somwcmpe cospooam oponamlsoscH 000 mean weapocfix .x magma m 120 0000050000 0:0 0000EE0000> 000000 0000 an 0005000E 00000000 000 00000050000 :0 m0500> I00 02000000 8000 so .AH 00009V 0005000 0000E8000o> 000000 0000 8000 U000E0000 .o:.ou000 wC0E5mmm oum O0 00000o0000x0 003 0 x .m.0 2000050m 050 0> 00009 50 00500> we Eosg 00000050 0000000I000UO0 now .0.0 2000050m Eosm 0000050000 00000050 50000 000 pom m m .0m.m coH0050m E000 p Q v00 .m0.m :000050m 050 A> 000090 000000 5000mm 0 0 .000.0A00 as 00000 as 0 + 000002 a 0.0 000 was 0000 0000 0000 0:0000.0 00000.0 00.0 0 000 000 0-0000.0 00.0 0-0000.0 0 00.0 A00 00.0 00.0 0-0000.0 00.0 0.00000 0 00 00.0 mA00200Asovo0Im 0-0000.0 00 0.0 0 00 00 0+0000.0 00.0 0-0000.0 0 00.0 0A00 . 0.0 0.0 0-00xm.m 00.0 0-0000.0 0 00 00.0 +0002 Ammzvoo A0Imv A0Imv Aflov 0000 A0Iw €00 A>EV AmIso AmI50 x00qsoo Aoumv0000 0000 0 x @000 Mm 008 00000 0oE 0000 o o 00 so .0oss0pco0 .x o0ose 121 .m m-0000.0 00000.0 0.0 0 0000.0 m0000.0 010000.0 00.0 0-00xm.0 0 00.0 000 000000 000000 m.00000 00.0 0-0000.m 0 00 00.0 +00020000zvo0 0-0000.m 00000.0 00000.0 00000.0 000 0-0000.0 00.0 000000 000 0.0 000 00 0.0000 00.0 0-00xm.0 000 00 00.0 000 00002000 0 00.0 00000.0 00000.0 0 0000.0 000 0-0000.0 00.0 0-0000.0 000 0.0 000 000 00 m-0000.0 00.0 0-00xm.0 000 00 00.0 W000000000nm 00 00000.0 00000.0 0 000 00.0 0-0000.0 m0.0 0-0000.0 000 0.0 000 00000.0 0.0 m.00000 00.0 0-0000.0 000 00 00.0 +00000mmzvo0 0.0 00000.0 00000.0 0 000 0.0 0-0000.0 00.0 0.0030 000 0.0 000 00000.0 0.0 m-0000.0 00.0 0-0000.0 000 00 00.0 +00000002000 A0I0V 00:00 0500 0000 00:0 800 A>EV A IE0 Amie 0000500 0 00 00 000x 0 .00 E0 000 005 000 000umv 0 x 0 E00 0 0 0 0 0 0 .0000000000 0000 00 00000009 00000000 000£0mu00000 000 0000 00000000 .Hx 0000B 122 .o:.oupmd wcHESmmm cum on Umpmaoawppxm mm: pmxm .Hm.m COHuwsvm Sega Umpdadoawo Ucm mhmeEmpHo> Ofiaomo pwmw mp UmLSmmmE manompfiv mam mmmmnpcmpma CH mmSHm>© .m.H COHumzvm Una H> magma CH mmSHm> we Eopm mmomMLSm UmummpplmUHUOH Low .:.H COHpmsvm Eopm UmpMadoaMo mmowmpzm cmmao map pom axo .mH.m coaumzdm cam A> mapwev wpmcpo cofipom Imp pampmaam Eop% no “AH manmev mpazwmp apmeEmpHo> oHHomo pmmm Eopm umumefiuwm app .oom.owzm pm :OHom EE m + :anm2 E H.o pom mam mumc mama Hfidm macaxm.w moaxz.p m.m owm Dom NIOHX©.: om.o :IOHXN.H O mm.o Am.:vm.a Aw.:vm.fi :uoax:.m no.0 queaxm.m o om mm.o mAmozvacmvooum AHImV Aalmv AMOV mama AHIMQMOV Amwv ANIEo AmIEo meQEoo onnmvpmx pmx o x x HoE HHOHV HoE moav U Q h p, g o .umscfipcoo .Hx magma 123 When the potential sweep is rapid (>2 V s'l) the change in the current as a function of time can be used to calculate k fromll et ket = i/nFAPi (5.21) Here F is the surface concentration that remains un— i reacted at the potential where i is measured. Values of Pi were obtained by partial integration of the cyclic peaks. This calculation assumes that ket is first-order with respect to the concentration of adsorbed reactant. The values of ke calculated from fast cyclic voltammetry t data are given in parentheses in Tables X and XI.. The directly measured values were in qualitative agreement with those calculated from steady-state measurements. The final column in these tables reports the values of Re when ex- t trapolated t0 E = 0 for an aet = O.UO. (onet was measured for several complexes with cyclic voltammetry and found to be 0.35 i aet : O.H5). Kp values for the iodide-treated surfaces (corresponding to "forced outer-sphere" mechanisms) were calculated from Equation 1.3 and ¢r values listed in Table VI. If one compares the apparent rate constants for these inner-sphere complexes on the clean electrode surface to that on the iodide-treated surface, striking effects are noted.10 For the chloro— and bromo-complexes the ratio 12“ of kapp (clean)/kapp(I) is as large as 106 for Co(NH3)SCl2+ on gold and 105 for CO(NH3)5Br2+ on platinum electrodes. The iodide treatment of these surfaces caused relatively minor rate changes for normally outer-sphere complexes.39 Reaction order measurements (Table V) support the idea that iodide blocks inner-sphere mechanisms for this group of complexes. The large rate decreases are therefore the probable result of a mechanistic change from inner-sphere on clean surfaces to outer—sphere on the iodide-treated surfaces. An inner-sphere pathway can enhance the rate of electron transfer (relative to the outer-sphere pathway) by several possible means. For convenience these have been grouped into thermodynamic and intrinsic effects as described in Chapter I. One obvious thermodynamic effect is the rela- tively higher reactant (more properly, "precursor") concen- tration present for an inner-sphere reaction. When a favorable adsorption equilibrium precedes the electron transfer step the concentration of reactant at the surface can be greatly enhanced relative to the bulk.11,1uu This is the case for the cobalt(III) complexes containing chloride, bromide or isothiocyanate anions as can be seen from Tables X and XI. Examination of the ratios for Kp(clean)/Kp(l) (given in the rows marked "A") shows the inner-sphere pathway enhances the formation of the precursor state. Because all of these complexes are at 125 or near monolayer adsorption from ml my bulk concentrations, the Kp ratios for inner-sphere versus forced outer-sphere are much greater than one. At electrode potentials where the electroreduction rates are kinetically controlled (eggfij those given in Tables X and XI), reactant adsorption can 5 cause 102 to 10 -fold increases in the overall electron transfer rates depending on the type of bridging ligand. The complexes exhibiting faster overall electron transfer 2+ 2+ 3 rates on the clean surfaces (Co(NH3)5Br and Co(NH3)SCl for instance) are reduced at electrode potentials where 35 diffuse layer adsorption is less favorable as a result of less negative values of ¢E (Table VI). A smaller value of Kp for the outer-sphere mechanism indicates that the con- centration of reactant at the reaction plane is lower rela- tive to the bulk concentration. From an electrostatic standpoint a less negative ¢E offers less attraction for cationic reactants. The larger ratios of Kp(clean)/Kp(I) for the faster reacting complexes are therefore mainly due to double layer effects upon Kp(I). An interesting point is illustrated when the effect of the double layer upon reactant adsorption is considered. The complexes which exhibit the largest extents of adsorp- tion (highest values for Kp(clean)) are adsorbed at elec- trode potentials where the double layer effect favors cationic adsorption the least. However, the adsorption of anions is increasingly favored as the electrode potential 126 becomes more positive. Adsorption of these complexes then appears to be dominated by the character of the ligands rathethhan the overall ionic character. That is, the ad— sorption can be said to be "ligand-induced", and might be anticipated to parallel the adsorptive behavior of the free 11’96 The bridging ligand character then is expected anion. to strongly influence both the thermodynamic and intrinsic behavior for an inner-sphere reaction.1 Ligands affect the outer-sphere reactions primarily through thermodynamic "driving force" differences (142;: differences in formal potentials). Under these assumptions a comparison of inner- sphere ket values to outer-sphere ke values for the same t complex at the same electrode potential should reflect primarily the intrinsic effect of the inner-sphere mechanism. When such a comparison of ket(clean)to ket(I) is made, + 2+ show a catalytic in- 2+ 2 only Co(NH3)5Br and Co(NH3)SCl trinsic effect (Chapter I). For Co(NH3)5Br on platinum the inner-sphere rate is 140 times faster than the outer- 2+ sphere rate. Although Co(NH3)501 is only 15 times faster on this surface by an inner—sphere route, this is still a substantial degree of catalysis. On gold electrodes the Co(NH3)5Cl2+ complex is catalyzed to a greater degree 2 than the Co(NH3)SBr + by an inner-sphere mechanism. The other chloro complexes have large values of ke but they t do not have higher ke values for the inner-sphere t mechanism relative to the outer-sphere. In fact, 127 the results indicate that for these complexes the outer- sphere pathway is favored for equal concentrations in the precursor state. The outer-sphere route is even more favorable for the isothiocyanate complexes. Intrinsic effects arise from specific influences upon the transition state for electron transfer.1 For inner- sphere heterogeneous electron transfer these effects are probably caused by the presence of the ligand-surface bond. For the monoatomic-bridged complexes, Co(NH3)5Cl2+ and 2+, a halide-surface bond can markedly influence the transition state.10 The cobalt center is connected Co(NH3)5Br through only a single ligand p orbital to the metal elec- trode orbitals, which may facilitate the transfer of the electron. The formation of the bond between the bridging ligand and the electrode surface weakens the cobalt-ligand bond. The lower force constant of this bond then could serve to diminish the reorganization energy necessary for electron transfer. The isothiocyanate complexes are intrinsically de- catalyzed for inner-sphere electron transfer. The iso- thiocyanate ligand differs from the halide ligands in two major features. The polyatomic bridge does not offer as intimate a connection between the cobalt center and the electrode surface. Secondly, the orbitals conducting the transferring electron are more likely w orbitals than 0 as for the halides. The sulfur-metal bond will have 128 relatively little effect upon the force constant of the more distant cobalt-nitrogen bond. A back-donation of w electron density from platinum or gold surface atoms may 1U5,1N6 strengthen the thiocyanate n system. The back- donation of n electrons is known to stabilize t Zs 1A5,1A7 And the back-donation of orbitals like those in cobalt(III). n electron density has been proposed for isothiocyanate transition metal complexes.lu5 This explanation is also consistent with the observation of more pronounced de— catalysis for the gig-Co(en)2(NCS); complex, which is expected to be di-bridged. In summary, the mono-bridged complexes have been found to be intrinsically catalyzed for chloride and bromide ligands, but decatalyzed for isothiocyanate ligands rela- tive to their outer-sphere reductions at platinum and gold electrodes. The proposed explanation has two points. First, that monoatomic bridges promote intrinsic catalysis through a more intimate connection between redox centers. Second, that platinum and gold surfaces may act as "sigma- withdrawing" and "pi-donating" reactants for the adsorbed complexes.1u6 The "sigma-withdrawing" property leads to strong adsorption and a weakening of the transition metal- ligand bond for chloro and bromo mono-bridged complexes. The same would be true for isothiocyanate but for this ligand the w character is more dominant. The cobalt(III) t2g orbitals are stabilized as a result of back-bonding 129 from the isothiocyanate ligand.1245 When adsorbed at the platinum or gold surfaces additional n electron density is acquired through filled t 2s gold surface atoms,1u6 which tends to further strengthen orbitals on platinum and the isothiocyanate-Co(III) n system. The adsorbed iso- thiocyanate complex then is stabilized and exhibits a larger force constant for the cobalt-nitrogen bond, causing the electron transfer to be intrinsically decatalyzed relative to the outer-sphere pathway. G. Heterogeneous-Homogeneous Rate Correlations The comparison of the observed heterogeneous reduc- tion rates with the rates of corresponding outer-sphere homogeneous reactions can provide additional evidence for the identification of ligand-bridged electrode reac- 10’36’38’lu8 According to the Marcus theory for tions. outer-sphere electron transfer,8 the ratios of the rate constants kh for homogeneous electron transfer of a series of reactants by a given reagent will be approximately independent of the reagent chosen. These rate constant ratios are also expected to be the same as the ratios of the rate constants for the corresponding heterogeneous electron transfer reactions at a fixed electrode poten- 8,10,58 tial. So the following relationship should be valid: 130 kfi kipp log (ER)A = log ( R )E (5.22) k h app where X and R superscripts denote the given reactant and the reference reactant. Subscripts A and E refer to constant homogeneous reducing (or oxidizing) agent and constant electrode potential, respectively. A plot of the two ratios in Equation 5.22 can be seen in Figures 12 and 13 for platinum and gold electrodes. The apparent rate constants were taken at an electrode potential of 0 mV. The homogeneous reducing agent was Ru(NH3)g+. The rate ratios are all relative to the Co(NH3)5F2+ rate constants for the apprOpriate reaction. The straight line in these two figures mark the location for a perfect correlation, lLEL’ unit slope through the origin. Circles mark the outer-sphere complexes and triangles mark inner-sphere com- plexes. The filled symbols represent the ratios for an iodide-treated surface. Figures 12 and 13 are similar in many respects. The outer-sphere complexes always appear above, but near, the correlation line for both clean and iodide—treated surfaces. The iodide-treated ratios are very near the clean surface ratios for the outer-sphere complexes, which is indicative of the similar reaction plane potentials (as noted earlier 2+ in Sections C and D in this Chapter). The Co(NH3)5NCS Figure 12. 131 Heterogeneous-homogeneous correlation at platinum electrodes. The heterogeneous data were extracted from Tables VII, VIII, X and XI. Circles and triangles denote outer- and inner-sphere mechanisms, respectively. The homogeneous data are for reduction by Ru(NH3)§+ and are from Reference 160, and R. C. Patel and J. F. Endicott, J. Am. Chem. Soc., 29, 636A (1968). The solid symbols correspond to iodide-treated surfaces. All of the ratios are relative to the reduction of Co(NH3)5F2+. The points are labeled with the X ligand in Co(III)(NH3)5X. The symbols 95(OH2)2, and g and t—(en)2Cl2 refer to g—Co(NH3)u(OH2)g+ and g and t—Co(en)2Cl; complexes, respectively. 132 I 2 3 . Iog[k;‘,/k;‘,],,u )- oL ‘h Figure 12 133 Figure 13. Heterogeneous-homogeneous correlation at gold electrodes. The heterogeneous data were extracted from Tables VII, VIII, X and XI. Circles and triangles denote outer— and inner-sphere mechanisms, respectively. The homogeneous data are for reduction by Ru(NH3)§+ and are from Reference 160, and R. C. Patel and J. F. Endicott, J. Am. Chem. Soc., 29, 6364 (1968). The solid symbols correspond to iodide—treated surfaces. All of the ratios are relative to the reduction of Co(NH3)5F2+. The points are labeled with the X ligand in Co(III)(NH3)5X. The symbols ge(OH2) g and 35(en)2C12 refer to _c_¢-Co(NH3)u(OH2 and g and EeC0(en)2Cl: complexes, respectively. , and 2 )3+ 2 1314 ’{D l 2 3 4 i '09 (kg, kg)» Figure 13 135 ion appears to mimic the outer—sphere behavior, even though it is known to be adsorbed strongly on clean platinum and gold.11 The remaining inner-sphere complexes all have heterogeneous rate ratios which lie above the correlation line on clean surfaces and below it on iodide-treated sur— 2+ ions have ratios faces. The Co(NH3)SBr2+ and Co(NH3)SCl substantially greater than predicted from Equation 5.22, which is reasonable in light of the probable intrinsic catalysis which they experience by an inner-sphere mechan— 10,11 ism The generally higher values for the heterogeneous ratios at gold electrodes may be a result of an erroneously 2+ at this Surface. (As low rate constant for Co(NH3)5F was noted in Table VII, kapp for this complex could only be determined to within m50% of the average value on gold electrodes.) The use of such correlations as diagnostic tools for 10’36’38 must be seen as a risky mechanism identification business at best. The assignment of an inner-sphere mechanism is only possible when substantial inner-sphere catalysis is present. In such a circumstance the rate response to adsorbed iodide seems at least as good a test. The assignment of an outer-sphere mechanism on the basis of a correlation with homogeneous rates can lead to errors, as it would for the Co(NH3)5NCS2+ complex on platinum. Even the rate response to adsorbed iodide gives the faulty impression that this complex reacts by an outer—sphere 136 route. Only the determination of reactant adsorption can unambiguously verify an inner-sphere pathway. The absence of such adsorption and a reaction order near unity are the best indications that the mechanism is outer—sphere. CHAPTER VI ELECTRON TRANSFER WITH CARBOXYLIC OR AROMATIC LIGANDS AT PLATINUM AND GOLD ELECTRODES The use of the cobalt(III) pentaammine complexes with simple anionic and neutral ligands has revealed much about the double layer structure and reactivity at platinum and 10,11,39,AO With the background of this work gold surfaces. and the techniques developed (to distinguish between dif- ferent electron transfer mechanisms, for instance), it would be interesting to examine a slightly more complicated group of reactants. To this end, cobalt(III) pentaammine complexes with various aromatic and carboxylic ligands were studied.1u9 A. Reduction Kinetics of Pentaamminecoba1t(III) Complexes Containing A,A'-Bipyridine and Related Ligands The electroreduction kinetics of Co(NH3)5L3+ complexes containing A,A'-bipyridine (BP), l,2-bis(A-pyridyl)ethane (BPA), Egags-l,2—bis(A-pyridyl)ethylene (BPE), pyridine (py), or pyrazine (pyZ) ligands were studied at mercury-, platinum-, and gold-aqueous interfaces in order to explore the ability 137 138 of such nitrogen heterocycles to mediate heterogeneous 1U9 These reactants have the electron transfer reactions. potential to be adsorbed at these surfaces and thereby to act as extended ligand bridges. The possibility of ligand V orbital overlap with surface orbitals is especially in- teresting in view of the previous results obtained for the isothiocyanate-bridged complexes in Chapter V. The electrochemical rate parameters for the reduction of the above mentioned Co(NH3)5L3+ complexes in perchlorate electrolytes at mercury, platinum and gold electrodes are summarized in Table XII as values of k3 at 0 mV and the pp Data for Co(NH3)5DMSO3+, apparent transfer coefficient “app which was the synthetic precursor to many of these complexes, are also tabulated. The rate parameters for the outer- sphere reactant Co(NH3)2+ are restated here for comparison purposes. (They also appear in Table VII.) The choice of 0 mV was made to minimize the extrapolation of Tafel 36 plots, which was sometimes necessary in order to compare rate constants at a constant electrode potential. The electrolyte solutions were acidified (to pH = 3, with HClOu) 0.5 M LiClOu and 0.1 M NaClOu for mercury and solid elec- trode measurements, respectively. All experiments were performed at 2H t 0.5 °C. The kinetics of most reactions were independent of pH in the range 1 3 pH i 3 at all sur— faces. The exception was Co(NH3)SBP3+ which exhibited de- creasing rates of reduction as the pH was lowered to 139 mp coEpompoQ who: mpcmeapmoxo owone mza .soaom as a mascaspsoo scaoaq a m.o :H assassspmo .CMmm>a:apm cwnpmcmzama> n .mUCflWHH UmUMEOPOcaQOCOE wCfiUCOwakeaOo m3». 0». .Hmrwwh mmmm USN am >5 c pm Aaim Eov oomx .mopzaoppomam mumpoanopmm Ca mocoppooam Uaoc com Escapmam .zLSOLmz pm moxmaQEoc xmammzv inaaavoo msoapm> mo coaposcom map pom mpmpoempmm mpmm amanmsoOLpooam .Hax oaome IMO .Auxmp mmmv meQEOO .HOQ GKQ £PH3 UGCHQECO moflmfingmU mQIw#ML EOLHM UmCHEeampmfi mmmlm ass ass +3 moaoon cmpmcouopa wcaccoomoppoo cam +mmmim mo :0apodcop mom a Una x mo modam>o .Hmz SE m mo coapaccm ozp Lopmw coax mo modam> c .zoacm as a msacaMQCOo :oacmz a a.c Ca cocasnopoao .ssssapsoo .Hax passe 141 pH 2 2, after which the rates become pH—independent. The pH dependence is probably due to the protonation of the uncoordinated end of the A,A'—bipyridine ligand to form Co(NH3)SBPHu+, which is then reduced at a slower rate than Co(NH3)5BP3+.lu9 The pKa for the protonated complex was determined by the titration of 1 my {Co(NH3)5BP}(ClOu)3 in 0.5 M LiClOu with HClOu. This pKa was found to be A.O i 0.2. The analogous pKa values for Co(NH3)5BPAHLl+ and Co(NH3)SBPEHu+ were too large (pKa > 5.0) to be accurately determined by this technique. Under the experimental conditions for kinetics measurements (pH = 3), these latter two complexes are almost entirely in their protonated forms. If it is assumed that the observed rate for the bipyridine complex is the result of the parallel reduc- tions of the protonated and unprotonated forms, the pKa can be used to calculate individual kinetics parameters for the two forms. These values are also listed in Table XII. Reactant adsorption measurements for these complexes yielded sparse results. The results indicated that rela- tively small amounts (2-3 uC cm-Z) of adsorbed complex were obtained from 1 mm solutions (143;, low values for Kp). Single-step chronocoulometry experiments for hanging mercury drop electrodes (HMDE) at +200 mV failed to detect the adsorption of any complexes, except Co(NH3)5BPHu+ and somewhat surprisingly Co(NH3)5py3+. From 1 mm bulk concentrations the apparent surface concentrations, T', 1A2 ll 10 + were 5 x 10- mol cm.2 and 2.0 x 10- mol cm.2 for the BPH and py complexes respectively. On gold electrodes Co(NH3)5— + pyz3 was the only complex detectably adsorbed. From a 100 UM bulk concentration, P' was m7 x 10-11 mol cm”2 as determined from fast cyclic voltammetry (Chapter III, Table I). The addition of 5 mm NaI caused this fast cyclic adsorption peak to disappear, as expected, if iodide blocks the specific adsorption of this complex.11 Adsorption measurements at platinum electrodes were not possible due to the proximity of hydrogen reduction current to the reduction waves (or cyclic peaks) for these complexes. Fast cyclic voltammetry on mercury electrodes confirmed the adsorption of the pyridine complex to at least a monolayer. A cursory examination of Table XII shows the rate constants for the heterocyclic complexes to be larger than the rate constant for CO(NH3)2+ on any electrode surface. The substitution of an ammonia ligand in the hexaammine by a nitrogen-ligating heterocycle yields substantial increases in some cases. For example, the rate constant for the A,A'-bipyridine complex is a factor of 105 faster than that for the hexaammine. The most striking result is that the rate constants for reduction of a given heterocyclic complex are relatively insensitive to the nature of the electrode surface. This is better illustrated by the plot of the rate constants in Figure 1A. The iodide—treated electrode rate constants are not drastically lower as was the case Figure 14. 1U3 Reduction rate constants for Co(III)(NH3)5L at various electrodes. The supporting elec- trolytes were 0.5 M LiClOu for Hg or 0.1 M NaClOu for Pt and Au electrodes. The pH was adjusted to pH = 3 with HClOu. 1AA _ _ . . ..\\ .:©i~:uluzui©i .:©i:ou: \ ,z- .128- $02. _L_igond Olll/ \ a E i Q- «$189.0- N i Iii/.....I/I/ :0 iii I 0121 . _ _ _ _ O O O. O O. .i 2 s ... Figure 1A 1A5 + 2+ for Co(NH3)SBr2 and Co(NH3)SCl (Chapter V). Evidently these heterocyclic ligands do not promote the kind of intrinsic catalysis noted for the two halide-bridged reactants. A correlation to homogeneous reductions (Table XIII) shows that the nitrogen heterocycle-containing complexes are catalyzed in their heterogeneous reductions at mercury, platinum, and gold surfaces.lu9 The outer-sphere homo- geneous reductions (for Ru(NH3)g+, for instance) have not been extensively investigated for this group of reactants. One outer-sphere reducing agent which has been studied - 153,15“ is Fe(CN)2 The reduction rate constants for this reducing agent refer to outer-sphere electron transfer with- 15A in a precursor (collision) complex. The homogeneous reduc- tion rate constants are used to calculate rate constant ratios relative to the outer-sphere reduction of Co(NH3)5- OH§+,10’39 by the procedure used in Chapter V for such conditions. The analogous heterogeneous rate constant ratios can be calculated from data in Table XII and those for Co(NH3)50H§+ given in Table VIII. Table XIII lists these ratios. The heterogeneous ratios are much larger (NSO-SOOO-fold) than the corresponding homogeneous ratios. In Chapter V, rate enhancements as much as 10A were seen 10’11 This same factor to result from reactant adsorption. could account for the rate enhancement here. This result together with the adsorption measurements indicates that 1146 Table XIII. Relative Heterogeneous and Homogeneous Rate Comparisons for Nitrogen Heterocycle-Containing Co(III)(NH3)5L Complexes. a. kapp/kggg kh/k:H2b Ligand L Hg Pt Au Fe(CN)2— Py 3.8 28 u.0 0.083 BP 600 640 280 0.13 BPAH+C 15 u.7 0.80 0.056 BPEH+0 165 22 3.0 0.055 aRate constants from Table XII relative to Co(NH3)50H3+ from Table IIX. bFrom References 153 and 15A, for homogeneous reducing agent Fe(CN)2_. [kh = 0.18s’1 for Co(NH3)50Hg+]. CDenotes protonated ligands. 1A7 the enhanced reactivity of the heterocyclic complexes relative to simple outer-sphere reactants is probably a result of reactant adsorption, iLQL, an enhanced reactant concentration at the electrode surface. The similarities in rate constants for the homogeneous reductions of the BPA and BPE complexes have been taken 155) as evidence that these ligands serve (in the literature to keep reacting centers in close proximity for outer- sphere electron transfer, and are not being utilized to mediate the electron transfer through orbital coupling.155 (EPA has a saturated linkage between pyridine rings, while BPE is conjugated.) The heterogeneous rate constant ratios indicate that the reduction of the BPE complex is favored over the EPA complex by at least a factor of 4. Perhaps the more extensive n orbital structure of the former complex leads to a greater extent of adsorption at electrode surfaces. In comparison to the previous kinetics studies,10’ll’39 these heterocycle-containing reactants exhibit behavior similar to the isothiocyanate complexes discussed in Chapter V. That is, rates on all surfaces are nearly the same, even though evidence for reactant adsorption exists. The reactant adsorption is not nearly as well documented for these complexes as it was for the isothiocyanate complexes. Only Co(NH3)5py3+ appears to be strongly adsorbed on mercury and probably platinum and gold. The assumption that the complex is also strongly adsorbed on the solid 148 electrodes is supported by evidence of strong adsorption of pyridine at these surfaces.u3’150 The A,A'-bipyridine and pyraZine complexes were also detectably adsorbed, but not to the same extent as the pyridine or isothiocyanate complexes. The adsorption of the reactant leads to a "surface thermodynamic" lowering of the overall free energy barrier to electron transfer as a result of the increase in the concentration of the precursor complex for electron 11,1A9 The close similarity between transfer (Chapter I). inner-sphere and forced outer-sphere (iodide-blocked) rate constants requires the inner-sphere rates to be intrinsically decatalyzed to compensate for the reactant concentration dif- ferences (Just as it did for the NCS- complexes in Chapter V). The other alternative is that these complexes are ad- sorbed on iodide-treated electrodes. The fast cyclic voltammetry results fortfimapyrazine complex on Au and Au(I) indicate that iodide is effective in blocking the adsorption of Co(NH3)5pyz3+ (Chapter III). Although similar to the isothiocyanate complexes in many respects, the heterocycle-containing complexes achieve the same pattern of reactivity by different means. Iso- thiocyanate complexes are undoubtedly bound to the electrode surface by a sulfur-electrode bond.ll’36’93’9u While most oftflmeheterocyclic ligands have an exposed nitrogen lone pair capable of forming a surface bond, the pyridine com— plex does not. Because the pyridine complex is adsorbed 1A9 to the greatest extent, the availability of a nitrogen lone pair is not the primary consideration for strong reactant- surface interaction. The heterocyclic ligand—surface inter— actions for the pyridine complex (and probably the other complexes as well) must almost certainly result from n orbital overlap or van der Waals forces.lu9 This result is somewhat surprising, since it does not seem likely that thepyridinecan.attain a high degree of n orbital overlap when bound to cobalt(III) pentaammine. The pyridine ligand extends only slightly (W1 A) beyond the ammine ligand "sphere" and consequently cannot lie flat on the electrode surface. Pyridine and BPE are known to be strongly ad- sorbed on mercury surfaces, either through the nitrogen lone pair or through the w system, iLEL, end-on or flat.105,151,152 As far as I know, this is the first time adsorption of pyridine through the "opposite end" (opposing the nitrogen) has been observed. The inner-sphere mechanism for these complexes is subtly different from that defined for more simple complexes (Chapter I).11 Rather than the formation of a bond to the surface, the interaction here is more delocalized, involving the ligand w orbitals or non-bonding van der Waals effects. The d-w back bonding explanation (Chapter V) is still applicable to the apparent intrinsic decatalysis observed for these complexes. The pyrazine and A,A'-bipyridine ligands appear to be most effective in increasing the rate 150 of electron transfer (Figure 14). These ligands can offer a nitrogen-electrode surface bond and a more conventional inner-sphere mechanism. B. Reduction Kinetics of Pentaamminecobalt(III) Complexes Containing Carboxylic Acid Ligands Homogeneous electron transfer studies by Fan and Gou1d156’157 have found that carboxylate anions tend to be effective mediators for electron transfer. They con- 2+ cluded that the Co(NH3)5X complexes are reduced mainly . 2+ 2+ 156 2+ 157 by inner-sphere mechanisms by Craq, Vaq’ and Euaq' Through homogeneous rate correlations with known outer- sphere reactants, they were able to determine the rate constants for outer-sphere homogeneous reduction by Ru(NH3)§+.156 The electron supply by the ligand was more important for the outer—sphere reactivity (the Ru(NH3)§+ rates), while non—bonding, steric effects governed 2+ 2+ 2+ , 156 aq’ Vaq and Craq) reduction rates. The inner-sphere mechanism proceeds throughtflmecarbonyl the inner-sphere (Eu functionalgroup.157 This section describes the hetero- geneous kinetics for electron transfer of these carboxylate- containing complexes. The heterogeneous kinetics were studied for the reduc— tions of various Co(NH3)5X2+ complexes, where the X ligand was an anion of a carboxylic acid. These ligands are bound to the cobalt center by a carboxylate oxygen. The reduction 151 rate parameters are given in Table XIV for mercury, platinum, and gold electrodes at 0 mV in 0.5 M LiClOu for mercury, and 0.1 M NaClOu for solid electrode experiments. For the rate constants reported here, the pH was adjusted to pH = 3 with HClOu. Rate constants were determined by d.c. polarography or rotating-disk voltammetry. The mercury results were obtained by V. Srinivasan. A plot of the rate constants as a function of the ligand X is depicted in Figure 15. This figure shows that these reactants are relatively insensi— tive to the type of electrode material. Unlike the pyridine complexes, reactant adsorption was not detected for any of the carboxylate-containing complexes by fast cyclic voltam- metry at mercury or gold electrodes. These complexes are reduced at rates near or slower than the reduction rate for Co(NH OH3+, a simple outer-sphere reactant.39 (The aquo 3)5 complex is used for comparison since it also features an oxygen-ligating ligand.) These observations would be consistent with an outer-sphere mechanism for reduction.lo’39 The effect of the electron supply to the carbonyl group from the attached R group (ELEL, CH3 for acetic) may be expected to be reflected in the pKa values of the appropriate carboxylic acids. Electron-withdrawing R groups should stabilize the acid anion, giving a lower value for the pK If the heterogeneous reaction is indeed outer—sphere, a' then a correlation between the ligand pKa and the reduction rate constant might be anticipated. If, on the other hand, 152 mm. m:. mo. :icaxm.: :icaxa.a micax:.a micaxc.m micaxm.a oumomconim mm. cu. cm. micaxa.: micaxc.a :icaxm.a micaxa.m micaxm.a opmam>agim mm. ma. mc. micaxm.: micaxm.a micaxc.m :icax:.m :icaxm.w mumpoom macim mm. mm. mm. micaxm.m :icaxm.a micaxc.a :icaxc.m micaxm.m mumpoom mmim co. m:. cm. :icaxm.m micaxa.m :icax:.a :icaxa.a cicaxm.m mumpoomim mm. mm. mm. :icax>.m micaxm.m :icaxm.a :icaxw.a micaxa.a omeLOMim s< on ma Aavsa Aavom sa us we goaosoo ammo A>e o co aim Soc ounx .moxoaQEoo Aaaavpamnoo oaaaxonhmc pom whopoempmm opmm coaposcom .>Hx canoe 153 Figure 15. Co(III)(NH3)5X electroreduction rates at 0 mV at various electrodes. The supporting electrolytes were 0.5 M LiClOu for Hg or 0.1 M NaClOu for solid electrodes. 15A oPt DAu 0Pt(l) UAufl) NIOI Ligand X Figure 15 155 Figure 16. Homogeneous and heterogeneous reduction rate constants of Co(III)(NH3)5X as a function of the X ligand pKa' 156 4 pKa of X Ligand Figure 16 157 the reaction were to proceed via an inner-sphere mechanism, then this correlation should be disrupted by steric ef- fects of the carboxylate R groups. Figure 16 shows the results of such a correlation for the reductions by Ru(NH3)g+,156 and platinum and gold electrodes (Table XIV). The rate constants are observed to correlate reasonably well with the pKa of the ligand, with slightly stronger depen- dence for the heterogeneous reductions. Steric effects do not appear to affect the correlation. This result also sup- ports the assignment of outer-sphere mechanisms to the hetero- geneous reductions. The generally higher rates at gold than at platinum probably result from the double layer po- tential differences that favor outer—sphere cation reduc- tions at the former surface (Table VI). The contrast between homogeneous and heterogeneous reductions for the carboxylate-containing complexes is unusual. The mechanisms for the simple complexes (Chapter V) and the pyridyl complexes for reduction by Eu2+ a aq 2+ Vaq are usually the same as for heterogeneous reductions at platinum and gold. Perhaps these metal electrodes have nd less affinity for oxygen-bridging than for nitrogen or halogen-bridging. The rates of outer-sphere reductions in Figure 15 do not exhibit any steric effects that would suggest a more distant reaction plane for the larger ligands. The reactant may be oriented with the carboxylate ligand pointing away from the electrode surface. CHAPTER VII MISCELLANEOUS EXPERIMENTS A. The Standard Potential for Co(NH3)g+/2+ 1. Introduction The electron transfer reaction represented by Equa- tion 7.1 has been of fundamental interest in theories of 58,159,160,162 homogeneous electron transfer. The very slow rate constant usually assigned 3+ 2+ 3+ 2+ Co(NH3)6 + CO*(NH3)6 2 Co(NH3)6 + Co*(NH3)6 (7.1) for the electron exchange rate, kex’ has provoked con- siderable commentary58’159-166 and some revision of electron transfer theory.159’l62’168 Stranks161 determined an 10 -l -1 upper limit of kex < 8 x 10. M s at 6A.5°C. This limit is undoubtedly lower at 25°C,58’l60’167 and was estimated to be m3 x 10'12 M-ls-l at this temperature.160 This contrasts sharply with kex for Ru(NH3)g+/2+, which -1 169,170 has been measured to be about 103 M718 at 25°C. The large difference in electron exchange rate constants for these two couples has been attributed to high internal 158 159 reorganization energies for Co(III/II)—N bond length 161,164 63 159 changes, quantum effects,1 electron tunneling effects, and the work required to achieve the neces- sary reactant orbital overlap when donor and acceptor e* 168 3+/2+ g' orbitals are both The fact is, kex for Co(NH3)6 is such an anomalous value that it requires an extraor- dinary explanation. Up until now most investigators have chosen to trust the experimental result reported by 161,16u Stranks. Actually Stranks' value and the consequent discussion generated by it, can be traced to measurements of the stability constants for cobalt ammines first per- 171 formed by Lamb and Larson followed by more extensive measurements by Bjerrum.172 The electron exchange rate for Equation 7.1 is not a directly measurable quantity. The standard rate constant for heterogeneous electron transfer, ks, is presumably (see later discussion) measurable by some form of voltam- metry. Once kS is known, kex can be calculated from the 58 Marcus relation, (kex/Z)l/2 = kS/ze (7.2) 1 for outer-sphere electron transfer, where Z and Z6 are l the collision frequencies of the homogeneous and hetero- geneous processes. In Marcus' treatment Z and Zel are taken to be 1011 and 10”, when kex and kS have the units 160 of M-ls-l and cm 5'1, respectively. 58 The standard rate constant is the rate constant for heterogeneous electron transfer that is measured at the standard potential E0 for the particular redox couple. Hence a determination of kex from Equation 7.2 depends on accurate measurements of electron transfer rates and a determination of E°. It is this latter value that is cast into doubt by the present work. 172 The stability of Co(NH as determined by Bjerrum )2+ 3 6 indicates that high concentrations of ammonia (6 N) are needed for this complex to be favored (>50%) over other cobalt(II) mixed aquo, ammine complexes. Measurements of k and E° have consequently been performed under conditions 3 or high pH and high ammonia concentrations.16l’l6l'l'166’172 + In all of these measurements the Co(NH3)§ concentration is presumed to be that given by Bjerrum's stability con- 165 stants. Bartlet and Landazury have reported platinum RDE data for the Co(NH3)g+/2+ redox couple in 2 N NHuC1 and concentrated NH3, that seem to be in agreement with Bjerrum's estimate of the Co(NH3)E+ concentration for these conditions (m70% of all Co(II) species). They also find the same value (:2 mV) for the standard potential, 165,172 E° = -l82 mV versus SCE. On the other hand, Laitinen 177 performed platinum RDE experiments with and Kivalo Co(NH3)g+ in 12.7 M NH3 and 1 M NHuNO3 and found the anodic current to be about 1% of the value expected on the basis 161 of Bjerrum's stability data for Co(NH3)§+. This stability constant has also been employed in determinations of the Co(OH2)2+/2+ electron exchange rate,173 studies of Co(II) complexes}?!4 and in radiotracer exchange experiments for CO(NH3)E+,175:176 in spectroscopic The importance and widespread use of Bjerrum's result prompted us to take another look at the chemistry and electrochemistry for the Co(NH3)g+/2+ redox system. We chose to examine this system with cyclic and rotating disk voltammetry at gold electrodes in alkaline aqueous solu- tions. Cyclic voltammetry offers the opportunity of obtain- ing both E° and kS values. It also is especially useful in monitoring preceding or succeeding chemical steps to electron transfer. Rotating disk voltammetry was employed in an attempt to duplicate the results of Bartlet166 at gold electrodes instead of platinum. The formation of Co(NH3)§+ in solution and its electrooxidation to Co(NH3)2+ were of primary interest in our work. The value of kex is probably less certain than formerly believed,l6l’l6u as we have observed considerable reason to doubt previously 165,166,172 reported values for the standard potential and the stability constant for Co(NH3)§+.172 2. Experimental The following data were obtained with a gold RDE or a gold "flag" electrode constructed from a small portion of 162 gold foil. These electrodes were pretreated in the normal fashion (Chapter II) with one exception; the gold flag electrode was not mechanically polished. Electrolyte solu- tions were deareated for two hours with prepurified nitrogen passed through a gas scrubbing bottle containing the same concentration of NH3 to reduce the loss of NH3. Measure- ments of solution pH with an Orion glass electrode (Model 91-02) were in close agreement with the calculated pH, indicating that loss of NH3 was minimal during the dearea— tion process. The electrolyte solutions were stored in sealed glass vessels under a nitrogen atmosphere until needed. The appropriate cobalt complex was added to a por- tion of this electrolyte and dissolved by the bubbling of nitrogen through the electrochemical cell. Unless other-- wise noted, the electrochemical measurements were made as soon as possible (as soon as the cobalt complex had com- pletely dissolved). The UV-Visible spectra were obtained with a Beckman DB-GT SpectrOphotometer. The effective electrode areas were calculated from cyclic voltammetry peak heights for the reversible Ru(NH3)2+/2+ couple as 0.127 and 0.210 cm2 for the RDE and flag electrodes, respectively. 3. Results Figure 17 illustrates the possible electrochemical and chemical processes that may be observed in an 163 Figure 17. The possible reaction scheme for Co(NH3)g+/2+ in aqueous ammonia solutions. 1611 Na madman .0 iv M319 eflezzvoo I ezzw w... .0 .mfifzvoo .m m .. firefaxzeo .1 i. ..w :58 < 165 investigation of the redox behavior of Co(NH3)2+/2+. This figure represents the electron transfer reactions possible for Co(NH3)g+/2+ and its partially solvent-substituted (n>O) congeners. Knowledge of processes B and C is es- sential in a determination of k8 , while processes A and D x constitute possible interferences. High concentrations of NH3 presumably favor the hexaammine species in processes A and B. It is known that Co(NH3)g+ is substitution inert over a reasonable time period (a few hours),10 so process A heavily favors the hexaammine species. Cobalt(II) com- plexes are substitution labile, so exchange of aquo and ammine ligands occurs rapidly, and process B may be expected to be sensitive to the NH3 concentration. The solution pH becomes important for longer time scale experiments; at pH > 10 the precipitation of Co(OH)2 on the electrode surface can interfere with electrochemical measurements.176 The latter point is a more serious complication in electrol- yses than it is for cyclic voltammetry experiments. For the typical pH in this study (N10), process D more likely involves hydroxyl-substituted complexes (112;: aquo ligands are deprotonated). Table XV lists the cyclic voltammetry results at gold electrodes for 5 mM CoCl2 in 6 and 1A.8 M NH3, where the solution pH was varied at constant ionic strength in the mixed electrolyte (2.5 — x)M KCl + xM NHuCl. The results are separated with respect to the direction of potential 166 Table XV. Cyclic Voltammetry Results for 5 mM CoCl2 in NH3/NHuCl Solution (u = 2.5 M) at Gold Elec- trodes.a i b Conc. NH3 Ef Ep,a p,a_2 Ca (M) pH (mV) (mV) (mA cm ) (mM) Anodic—Cathodic 6 10.80 -275 -200 1.30 1.10 10.25 -325 -275 1.10 0.93 - 10.02 -330 -267 0.98 0.83 1A.8 ll.A0 -388 -3A5 0.66 0.56 10.50 -38A -35A 0.88 0.7M 10.03 -391 -357 0.72 0.61 Cathodic-Anodic 6 10.80 -291 —l97 1.02 0.86 10.25 -30A -23A 0.63 0.53 10.02 -302 -229 0.A3 0.37 1A.8 ll.A0 -392 -335 0.67 0.56 10.50 -380 -337 0.71 0.60 10.03 -375 —325 0.6“ 0.5M a200 mVs'l at 25.000. b oxidation, calculated from Equation 3.“. The equivalent concentration of i paa assuming one electron 167 scan (143;, "anodic-cathodic" indicates the potential was scanned positive first then negative, causing oxidation followed by reduction). The apparent formal potential Ef was the median potential between the anodic and cathodic peak potentials, E and Ep c’ respectively. The anodic 9 p33 peak current density ip a was the background corrected our- 3 rent density at Ep a’ All of these results refer to a 9 potential sweep rate of 200 mV 5.1 and 25.0°C. The con- centration of oxidizable species Ca was calculated from ip a 3 from Equation 3.A, and Ca was always observed to be sub- stantially smaller than 5 mM (the bulk concentration of CoCl2 added). A typical cyclic voltammogram is shown in Figure 18. This cyclic was for the anodic-cathodic po- tential scan in 1A.8 M NH3 at pH = 10.50. Cyclic and rotating-disk voltammetric measurements were also made for solutions of Co(NH3)6C13 in 1 M NHuCl and 1A.8 M NH at platinum and gold electrodes. The 3 rotating disk results were similar to those of Laitinen and Kivalo}77 i.e., the anodic limiting current was between 1 and 10% of the cathodic limiting current. (Bjerrum's 2+ should comprise greater than 172 data indicates that Co(NH3) 90% of all Co(II) species under these conditions.) Cyclic voltammetry of Co(NH3)6Cl3 in the same solution at gold electrodes yielded some unusual results, some of which are shown in Figure 19. The formal potential Ef appeared to be around —210 i 5 mV which is near the -l82 mV value Figure 18. 168 The cyclic voltammogram for CoCl2 in 1A.8 M NH3 + 1.5 M KCl + 1 M NHuCl at a gold electrode. The potential was scanned in the positive direc- tion first. This is an example of an anodic- cathodic scan. The potential sweep rate was 200 mV s'l. pH = 10 50. 37"? 169 QH @cHSMWHrm Figure 19. 170 The cyclic voltammograms for Co(NH3)6C13 in 1A.8 M NH3 + 1M NHuC1 at a gold electrode. The anodic—cathodic scans for various elec— trolysis times (a-c) at —500 mV. The poten- tial sweep rate was 200 mV s-l. pH = 10.50. 171 wmrum .. ...—acne .- cmsflva an OONVI ma enemas 172 172 165,166 reported by Bjerrum and Bartelt , but is on the average 170 mV more positive than Ef for the CoCl2 solutions listed in Table XV and shown in Figure 18. A time-sweep rate dependence was noted in these Co(NH3)6Cl3 experiments. The cathodic peak for reduction of Co(NH3)2+ was not affected, but the anodic peak current was found to depend on the time that the potential was held at -500 mV (where Co(NH3)g+ is reduced at a duffusion- limited rate). At longer electrolysis times at -500 mV the anodic peak current increased. From 20 s to 6 min the current was seen to increase by about 60%; (this is illustrated by cyclics a to c in Figure 19). The same type of behavior was noted in cathodic-anodic scans, but to a lesser degree. In these scans the anodic peak was pro- portionally larger (relative to the cathodic peak) as the sweep rate decreased from 500 to 20 mV s-l. This is op- posite to what one might expect, if Co(NH3)§+ is the source of anodic current. If Co(NH3)§+ undergoes a fairly slow aquation, more anodic current for Co(NH3)2+ should be present at faster sweep rates. The reduction peak seemed to be for Co(NH3)g+, as it was noted to increase when more Co(NH3)6Cl3 was added. The reductions of Co(NH3)50H2+ and gig-Co(NH3)u(0H): occurred at distinctly different (>50 mV) potentials than Co(NH3)g+. The bulk electrolysis of CoCl2 in 1A.8 M + l M NHuCl at a gold electrode (area = 10 cm2) potentiostatted at 173 -100 mV for 9 hours at 10°C produced a yellow solution with an absorbance maximum at “78 i A pm. A solution of Co(NH3)6Cl3 in the same electrolyte gave an identical ab- sorbance maximum. The wavelengths of maximum absorbance 2+ for Co(NH3)50H and cis-Co(NH3)u(OH); solutions were 505 and 525 pm respectively. A. Discussion The cyclic voltammetry results have been found to be distinctly different for solutions of CoCl and Co(NH3)6Cl3 2 in 1 M NHuCl and 14.8 M NH3 (Figures 18 and 19). For the latter reactant one can be reasonably certain that the ob- served reduction peak is for the hexaammine species. The subsequent oxidation of the reduction products produced a much smaller current than was expected (from Bjerrum's data)172 for Co(NH3)§+. The larger anodic peaks at slower sweep rates are indicative of at least one chemical.step following the reduction of Co(NH3)g+. This process is probably similar to B in Figure 17, although other processes such as formation of hydroxyl-bridged surface species are possible. In any case, it seems that some electroactive species is formed rapidly from Co(NH3)g+, which can then be oxidized to a Co(III) complex. This Co(III) complex, which is most likely not Co(NH3)g+ (only process C, Figure 17 could account for direct oxidation to the hexaammine), is not observable as a separate peak on a subsequent 17A cathodic scan. The formal potential is therefore probably not the E° for Co(NH3)2+/2+, but is a mixed potential for Co(NH3)g+ and some Co(II) complex other than Co(NH3)2+. The amount of this Co(II) complex is much less ( 1%) than was expected on the basis of Bjerrum's stability constants. The experiments with CoCl2 approach the problem from the opposite end. When Co(II) is added to a solution of concentrated aqueous ammonia, it may be reasonably assumed that it rapidly achieves an equilibrium between the various possible Co(II) aquo, hydroxyl, and ammine complexes. The observation of an apparent formal potential for this solution that is 170 mV more negative than for the Co(NH3)g solution is a strong indication that Co(NH3)2+ is not being formed. The redox behavior may then represent something similar to process D in Figure 17. Eventually the CoCl2 solution forms CO(NH3)2+ by electrolysis, as was determined from the spectroscopic measurements. However, this only occurs after several hours. It may be that a very small amount of Co(NH3)‘:+ is continuously electrolyzed to the thermodynamically stable Co(NH3)g+. This could explain the long time required for electrolysis. The results in Table XV show the apparent formal po— tential to be fairly pH-independent in 1A.8 M NH3. The differences between Ef values at 6 and 1A.8 M NH3 can be attributed to medium effects (e.g., junction potential dif— ferences). These factors argue against a mixed potential 172 + 175 for this system, although pH-independence of a process like D in Figure 17 is difficult to rationalize. As a consequence of this work, the stability of Co(NH3)E+ appears to be much lower than previously report- d.165,166,172 165 e The experimental result of Bartelt on a platinum RDE could not be duplicated. The E° for Co(NH3)§+/2+ was, in fact, not observed. A mixed potential, however, was determined to be near previously reported values of the standard potential. It is our opinion that this mixed potential could easily be mistaken for the standard potential in potentiometric measurements, such as those 172 The dubious nature of the actual done by Bjerrum. value for the standard potential would seem to preclude extensive speculation regarding the anomalous value of kex for Co(NH3)g+/2+ electron exchange. B. Activation Parameters for Heterogeneous Electron Transfer Reactions 1. Introduction The measurement of the rate constant for electron transfer in the adsorbed state, k by fast cyclic voltam- et’ metry (Chapter V) allows a determination of the thermo- dynamic activation parameters for an adsorbed reactant. If Equations 1.5 and 1.6 are combined the following rela- tionship is obtained, 176 _ E in ket _ 2n Kvp — (00*) /RT (7.3) Substituting for (AG*)E gives in ket = 2n KVp + (AS*)E/R = (AH*)E/RT (7.u) It is clear from Equation 7.“ that the temperature de— pendence of k8 can be used to find the enthalpy of activa- t tion from the precursor state (AH*)E from the $10pe of a plot of fin ke versus T-l. If one knows or assumes a t value for Kv the activation entropy for the precursor p, state (AS*)E can be determined from the intercept of such a plot. For a rigorous treatment the activation parameters should be evaluated at a fixed coverage (or surface con- centration) of the adsorbed reactant to eliminate the effect of adsorbate-adsorbate interactiOns. Consequently, ket must be determined for a fixed adsorbate coverage and electrode potential as a function of temperature. 2. Results S2+ and cis-Co(en)2(NCS)+ The complexes Co(NH3)5NC were chosen for this study. They exhibited reproducible fast cyclic peaks with gold electrodes that occurred at sufficiently positive electrode potentials to minimize 177 the interference of proton reduction current. The rate constant for electron transfer could be evaluated at dif- ferent coverages from a single fast cyclic peak. Equation 5.21 gives the dependence of ket on the surface concentra- tion of the reactant at various points on the cyclic voltam- metric peak. These different coverages were necessarily at different electrode potentials. (Different points on the cyclic peak correspond to different electrode potentials as well as different adsorbate concentrations.) The procedure used to obtain values of ket at constant potential and constant surface concentration was as follows: (i) Calculate ke at several electrode potentials t from Equation 5.21 for several temperatures. (ii) Interpolate values for Zn ket and E for a fixed surface concentration of the reactant. (iii) From knowledge of the Tafel slope (3th ket/aE)P extrapolate the tn ket values to a single elec- trode potential. The necessary Tafel slope was determined from plots of log (i)FF against EFF’ where FF signifies the reactant surface charge density. Variation of (1)FF with EFF was achieved by changing the potential sweep rate from 2 to 1 50 V s- Figure 20 shows an example of this type of plot for 80 MM Co(NH3)5NCS2+ on gold for various surface charge densities at 25.0°C. Because (1)FF is proportional 178 Figure 20. Tafel plots from fast cyclic voltammetry at various surface charge densities of Co(NH3)5- NCS2+ on a gold electrode at 25.0°C. The supporting electrolyte was 0.1 M NaClOu. Triangles, squares, and circles represent sweep rates of 20, 10, and 5 V 5.1, respec- tively. 179 Figure 20 180 to (ket)FP (from Equation 5.21), the slopes of these lines give the Tafel slopes that were required. Tafel lepes for various coverages were found for each temperature. 2+ The average Tafel slope for Co(NH3)5NCS was 1U.3 i 0.1 V-l. These slopes seemed to be fairly independent of coverage and temperature. The results for the activation plots of £naam um venom opoz *mc no mosam> o>apmwoc owpma magmaaEamo .pam poommoo m mom a n ma acoammopwoe Lmocaa mo ucoaoammooo ones .cocc on c Soon wcawcmp moLSpmpooEop c com aim> ca com >8 c on copmSam>o pox pom .aiB unsamwm no x Ga mom who: mpoam .zoaom SE m + :anm2 a a.c mm: opmaoppooao wcapLOQQSm 039m cmai a.cm+ mmm.c m.ca cmzmi ca osa- :.ma+ :ma.o ms.a osmmi on em mamozvmfleovooim aai w.mc+ :mc.c m.cm cmmwi m mm: c.mm+ cmm.c m.cm cmmsi ca mmi m.nm+ cmc.c m.cm cmcci ma cm +mchmAmszoo A mop aoEcv aoEcxv p paoo omoam A 80 02V Aazv xoaQEoc ai ai i om mi 9 o *m< *m< ipopca mm c m.oooscooam oaoo o co ncasnom ooaa soacm>aco< .H>x canoe 182 concentrations which differ by as much as a factor of 2. A striking difference is observed between activation parameters for these two complexes. 3. Discussion Both of these complexes are strongly adsorbed at gold electrodes (Chapter III), and exhibit nearly zero-order overall reaction orders for reduction (Chapter V).11 Kinetics probe results for analogous chromium complexes indicated that the gig-diisothiocyanate complexes were more strongly adsorbed than the monoisothiocyanate com- plexes (Chapter IV). The difference between mono- and di- bridged complexes is readily apparent from the activation parameter results in Table XVI. Although these results may be in error by as much as 20—30% because of the great deal of derivation employed; clearly, the two complexes are different both enthalpically and entropically. This result is somewhat unexpected for activation from the precursor state. The activation from the precursor state might be ex- pected to have only a small entrOpic contribution. The structure of (and around) the adsorbed precursor should be similar to the adsorbed transition state. Metal—ligand bonds must stretch, but no solvent should be displaced or significantly reorganized in the activation process. The large, negative entropies (especially for the cis complex) 183 are therefore surprising. It is possible that the orienta- tion of the adsorbed reactant is substantially different from that of the activated complex. The isothiocyanate ligands could prefer to be adsorbed at an angle to the 1A6 electrode surface, and then reorient perpendicular to the electrode surface in the activated complex. For the gig complex this could even entail the transformation from a di—bridged precursor to a mono—bridged activated complex. However, from such a small amount of data these speculations are not well-supported lthough measurements at silver 96 and mercury surfaces had similar results. C. Maxima in Pulse Voltammetry at an RDE As was mentioned in Chapter II, pulse voltammetry at an RDE was occasionally used to determine the electron transfer rate constants. This necessarily required some characterization of the technique under various conditions. At slow electrode rotation speeds (<200 rpm) maxima were invariably obtained in the voltammetric waves. The apparent limiting currents were also observed to be lower than expected on the basis of the Cottrell Equation.79 These effects were initially attributed to a reactant depletion effect, which results from the inability of reactant and product diffusion to keep pace with the electrochemical reaction. The reactant depletion can also result from the 178,179 adsorption of the reactant. The maximum arises 184 for nernstian reactions because the rate of reduction of an adsorbed reactant is limited by the diffusion of the 178 product away from the surface. As a result, the addi- tional current for the reduction of the adsorbed reactant is present when the current is sampled in normal pulse 179 Maxima for outer—sphere complexes such as + voltammetry. Co(NH3)2+, Co(NH3)5F2 , Co(NH3)50H3+, and cis-Co(NH3)u- (0H2);+ were typically smaller than for inner-sphere com- plexes such as Co(NH3)5NCS2+ and gig-Co(en)2(NCS);. The depletion effect was thought to result from different causes for outer- and inner-sphere reactants. At rotation speeds in the range 300 - 1000 rpm the depletion effect was overcome for the outer-sphere com- plexes. No maxima were observed at platinum or gold electrodes, and the limiting current was as expected for pulse voltammetry. The time after the application of the potential pulse at which the current was sampled was 48 ms for these experiments (with a PAR 174 potentiostat). On- line normal pulse experiments indicated that faster rota- tion speeds over a smaller range were required to obtain correct kinetics parameters with shorter pulse measure- ment times. For example, with Co(NH3)g+ the appropriate rotation speed range was W500 - 750 rpm for a pulse measure- ment time of 20 ms. The upper limit was determined by the point at which the limiting corrent is no longer correct 79 for pulse voltammetry. 185 Because some inner—sphere complexes at platinum and gold exhibited persistent maxima at any rotation speed, it was thought that this behavior could be used as another criterion for an inner—sphere mechanism. Figure 21 shows a typical result for the pulse voltammetric wave of 0.5 mM Co(NH3)5NCS2+ in 0.1 M NaClOu + 2 mM HClOu at a gold RDE. The rotation speed was 540 rpm, and the potential was scanned at 2 mV s-l. This maximum was still observed at a rotation speed of 1000 rpm. With digital simula- 178 tion and curve fitting, perhaps the extent of adsorp- tion could be found from the shape of the maximum. By using such simulations the kinetics and adsorption might be determined from a single experiment, which would be a great advantage for solid electrode experiments. 186 Figure 21. Pulse voltammogram for 0.5 mM Co(NH3)5NCSZ+ at a gold RDE in 0.1 M NaClOu + 2 mM HClOLl. The rotation speed was 540 rpm. The potential was scanned at 2 mV s_l. 187 am opsmam >\.a=:0aon —.i o —. .4 ‘ I A! ‘ To >E m . En. coo. .58.. Eat. :< b .l 3 M g a sawed/we: m3 CHAPTER VIII CONCLUSIONS AND SUGGESTIONS FOR FURTHER WORK The cobalt(III) ammines proved to be a versatile group of reactants for the investigation of kinetics and double layer structure at platinum and gold electrode surfaces. Their reactivity was found, for the most part, to depend on the types and arrangement of the coordinated ligands. Complex adsorption and inner-sphere catalysis were strongly dependent upon the nature of the bridging ligand. Some generalizations with regard to ligand types and reactivity at platinum and gold are possible. Fluoro, ammine, and aquo ligands do not induce adsorption or inner-sphere mechanisms.l0’39 Chloro and bromo ligands induce reactant adsorption and intrinsic catalysis of the electron transfer reaction.ll’uO Mono-bridged halide complexes are catalyzed the most of all complexes investigated, showing an overall catalysis (inner- versus outer-sphere) of up to 107-fold and an intrinsic catalysis as high as 200-fold (Chapter V). Isothiocyanato ligands induce strong adsorption of the reactant, but cause the electron transfer rate to be in- trinsically decatalyzed by an inner—sphere mechanism.ll’LIO Pyridine and pyridyl ligands induce reactant adsorption 188 189 through n orbitals or van der Waals interactions, but are not effective as mediators of the electron transfer.lu9 Adsorbed iodide was very effective in blocking inner- 10,11,40 The covalent sphere electron transfer mechanisms. nature of the iodide layer allowed outer-sphere rates of reduction to be easily determined for normally inner- Sphere complexes. This greatly simplifies the task of assessing inner-sphere catalysis. Iodide-treated electrodes were also 1ess_susceptible to deactivation by solution impurities. The adsorbed iodide apparently renders platinum and gold surfaces inert with respect to further adsorption. The measurement of reactant adsorption was more dif- ficult than anticipated. Single-step chronocoulometry gave erroneous results for platinum and gold electrodes. This technique could not account for substantial changes in the electrode capacitance caused by reactant and product ad- sorption.ll The apparent overall reaction order for the electron transfer reaction was used to roughly quantify 10 The use of outer-sphere the amount of reactant adsorbed. reactions to "probe" for adsorbed nonreacting species of- fered a means to determine complex ion adsorption without interference from adsorbing anions produced by reduction of the adsorbed complex. By far the most useful and ac- curate technique was fast cyclic voltammetry.11 The capaci- tance changes that gave erroneous chronocoulometry results were easily accounted for by integrating fast cyclic 190 peaks over a baseline corresponding to the nonfaradaic capacitance.ll The electron transfer rate from the pre- cursor state could also be calculated from these peaks. However, not all adsorbed complexes produced fast cyclic peaks. Intrinsic catalysis at platinum and gold surfaces seems to be limited to mono—atomic, mono—bridged complexes, such 2+, Other complexes that are as strongly as Co(NH3)5C1 adsorbed, but involve multi-bridged intermediates or poly- atomic bridges, do not exhibit a lowering of the intrinsic barrier to electron transfer. In fact these latter com- plexes appear to be intrinsically decatalyzed as a result of their adsorption. These results indicate that the catal- ysis is to a large degree determined by the vibrational freedom of the cobalt-ligand bond. The short mono-atomic chloro— and bromo—bridged complexes weaken the cobalt- ligand bond through donation of o electron density to the electrode surface. The weaker force constant causes a decrease in the reorganization energy 00* for electron transfer. The weakening of the cobalt—ligand bond is attenuated for poly-atomic bridging ligands. It was proposed that for some ligands with substantial n orbitals the cobalt-ligand bond may actually be strengthened through additional w orbital electron density acquired from the electrode surface, which would also stabilize the adsorbed reactant by d—n back-bonding (Chapter V). 191 Platinum and gold have been observed to be generally more catalytic than mercury electrodes. Much, but not all, of this catalysis is a result of a greater extent of reac- tant adsorption on the solid metal surfaces. The catalysis of outer-sphere complexes is probably caused by more anion adsorption on platinum and gold than on mercury. The solid electrode results are less reproducible and more difficult 39 to obtain than similar results at mercury electrodes. Suggestions for Further Work The intramolecular electron transfer reactions for cobalt-ligand-platinum complexes could be used to test many of the above conclusions, especially since vibrational spectroscopy might be used. Surface enhanced Raman spec- trOSCOpy (SERS) could possibly be effective in determining the same kind of information. The reproducibility and unusual characteristics of iodide-treated platinum and gold electrodes make them suitable for use in other (perhaps, analytical) applications. The reactivity of complexes containing sulfer heterocyclic ligands might prove interesting in comparison to the nitrOgen heterocycle results of Chapter VI. REFERENCES GENERAL REFERENCES J. E. Huheey, "Inorganic Chemistry: Principles of Structure and Reactivity", Harper and Row, N.Y., 1972. A. J. Bard and L. R. Faulkner, "Electrochemical Methods: Fundamentals and Applications", J. Wiley and Sons, N.Y., 1980. P. Delahay, "Double Layer and Electrode Kinetics", Interscience, N.Y., 1965. "Encyclopedia of Electrochemistry of the Elements", A. J. Bard (ed.), M. Dekker, Vol. IV and VI, N.Y., 1976. 192 10. 11. 12. 13. 14. REFERENCES M. J. Weaver and F. C. Anson, Inorg. Chem., M5, 1871 (1976). P. Delahay, "Double Layer and Electrode Kinetics", Interscience, N.Y., 1965, Ch. 4. H. Taube, "Electron Transfer Reactions of Complex Ions in Solution", Academic Press, N.Y., 1970, Ch. I and II. G. M. Brown and N. Sutin, J. Am. Chem. Soc., 101, 883 (1979). N. Sutin, in "Tunneling in Biological Systems", B. Chance, D. C. Devault, H. Frauenfelder, R. A. Marcus, J. R. Schrieffer and N. Sutin (eds.), Academic Press, N.Y., 1979, p. 201. M. J. Weaver, Inorg. Chem., 18, 402 (1979). N. Sutin, in "Inorganic Biochemistry", G. L. Eick- horn (ed.), Elsevier, N.Y., 1973, Ch. 19. R. A. Marcus, J. Chem. Phys., 43, 679 (1965). W. L. Reynolds and R. W. Lumry, "Mechanisms of Electron Transfer", Ronald Press, N.Y., 1966, Ch. 6. K. L. Guyer, S. W. Barr, R. J. Cave and M. J. Weaver in "Proc. Third Symposium on Electrode Processes 1979", S. Bruckenstein, J. D. E. McIntyre, B. Miller and E. Yeager (eds.), Electrochem. Soc. (pub.), Princeton, N.J., 1980, p. 390. K. L. Guyer, S. W. Barr and M. J. Weaver, submitted to the 159th Meeting of the Electrochem. Soc. 1981, Electrochem. Soc. (pub.), Princeton, N.J. N. Sutin, Acc. Chem. Res., 1, 225 (1968). W. A. Caspari, Z. Physik. Chem., 32, 89 (1899). F. Haber, Z. Physik. Chem., 32, 193, 271 (1900). 193 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 194 R. Russ, Z. Physik. Chem., 55, 641 (1903). J. Heyrovsky, in "Nobel Lectures in Chemistry for 1942-1962", Elsevier, Amsterdam, 1963, p. 564. H. Angerstein—Kozlowska, B. E. Conway and W. B. A. Sharp, J. Electroanal. Chem., 53, 9 (1973). H. Angerstein-Kozlowska, B. E. Conway, B. Barnett and J. Mozota, J. Electroanal. Chem., 100, 417 (1979). H. Angerstein—Kozlowska and B. E. Conway, J. Electro- anal. Chem., 23, l (1979). R. Woods, in "Electroanalytical Chemistry", A. J. Bard (ed.), M. Dekker, N.Y., Vol. 9, 1976, p. 119. K. B. Oldham and E. P. Parry, Anal. Chem., 33, 867 (1966). D. H. Angell and T. Dickinson, J. Electroanal. Chem., 35, 55 (1972). Z Samec and J. Weber, Electrochimica Acta., 39, 403, 413 (1975). D. J. Meyers, R. A. Osteryoung and J. Osteryoung, Anal. Chem., £6, 2089 (1974). R. P. Buck and H. E. Keller, Anal. Chem., _3, 400 (1963). J. Kuta and E. Yeager, J. Electroanal. Chem., _2, 110 (1975). L. Mfiller and H. Prumke, Elektrokhimiya, 33, 1083 (1975). C. Toren, Jr. and C. P. Driscoll, Anal. Chem., 38, 872 (1966). D. C. Johnson, J. Electrochem. Soc., 119, 331 (1972). N. V. Fedorovich, A. N. Frumkin and H. E. Case, Coll. Czech. Chem. Commun., 36, 722 (1971). A. N. Frumkin, N. V. Fedorovich, N. P. Berezina and K. E. Keis, J. Electroanal. Chem., 33, 189 (1975). N. V. Fedorovich, M. D. Levi and S. I. Kulakovskaya, Elektrokhimiya, 33, 766 (1976). 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 195 D. C. Johnson and E. W. Resnick, Anal. Chem., 22, 1918 (1977). This topic encompasses many industrial applications; for representative studies see G. C. Bond, "Catalysis by Metals", Academic Press, N.Y., 1962; or J. O'M. Bockris and S. Srinivasan, "Fuel Cells: Their Electro- chemistry", McGraw-Hill, N.Y., 1969. M. J. Weaver, J. Electroanal. Chem., 93, 231 (1978). T. L. Satterberg and M. J. Weaver, J. Phys. Chem., 83. 178A (1978). M. J. Weaver, P. D. Tyma and S. M. Nettles, J. Electro- anal. Chem., 114, 53 (1980). M. J. Weaver, J. Phys. Chem., 22, 568 (1980). S. W. Barr, K. L. Guyer and M. J. Weaver, J. Electro- anal. Chem., 111, 41 (1980). S. W. Barr and M. J. Weaver, to be published. For example, see Reference (3), or for more recent reviews, R. G. Linck, MTP Int. Rev. Sci., Inorg. Chem., Ser. One, 1972, 2, Ch. 8 (1972); and R. G. Linck, Surv. Prog. Chem., 1, 89 (1976). G. M. Schmid and N. Hackerman, J. Electrochem. Soc., 110, uuo (1963). M. Petit, N. Van Huong and J. Clavilier, c. R. Acad. Sci., Ser. 0, 266, 300 (1968). S. D. Abdulov, A. G. Zelinskii and R. Yu. Bek, Elektro- khimiya, 16, 655 (1980). J. Clavilier and N. Van Huong, C. R. Acad. Sci., Ser. C, 270, 982 (1970). J. Clavilier and N. Van Huong, J. Electroanal. Chem., A. Hamelin and P. Dechy, C. R. Acad. Sci., Ser. C., 275, 33 (1973)- A. Hamelin and P. Dechy, C. R. Acad. Sci., Ser. C., 272, 1450 (1971). J. Clavilier and C. N. Van Huong, J. Electroanal. Chem., 29, 101 (1977). 50. 51. 52. 53. 54. 55. 56. 57. 58. 59- 60. 61.. 62. 63. 64. 65. 66. 67. 196 T. Takamura, K. Takamura and E. Yeager, J. Electro- anal. Chem., 22, 279 (1971). R. Adzic, E. Yeager and B. D. Cahan, J. Electroanal. Chem., 33, 267 (1977). V. S. Bagotzky, Yu. B. Vassilyev, J. Weber and J. N. Pirtskhalava, J. Electroanal. Chem., 21, 31 (1970). V. Kazarinov and O. A. Petry, J. Electroanal. Chem., 21. Appl- (1970). J. P. Cushing and A. T. Hubbard, J. Electroanal. Chem., 23, 183 (1969). A. L. Y. Lau and A. T. Hubbard, J. Electroanal. Chem., 33, 77 (1971). C.-N. Lai and A. T. Hubbard, Inorg. Chem., 11, 2081 (1972). ‘— R. F. Lane and A. T. Hubbard, J. Phys. Chem., 77, 1411 (1973). “ R. A. Marcus, J. Phys. Chem., 31, 853 (1963). M. J. Weaver, J. Phys. Chem., 33, 1748 (1979). V. E. Kazarinov and N. A. Balashova, Dokl. Akad. Nauk 88823 $35, 864 (1960); 132, 6M1 (1961);.251, 117m (19 ). V. E. Kazarinov and N. A. Balashova, Coll. Czech. Chem. Commun., 32, 4184 (1965). V. E. Kazarinov and N. A. Balashova, Elektrokhimiya, 1. 512 (1965). N. A. Balashova and V. E. Kazarinov, in "Electro- analytical Chemistry", M. Dekker, N.Y., Vol. 3, 1969, p. 135- B. E. Conway, H. Angerstein-Kozlowska, W. B. A. Sharp and E. E. Criddle, Anal. Chem., 23, 1331 (1973). D. T. Sawyer and J. L. Roberts, Jr., "Experimental Electrochemistry for Chemists", J. Wiley and Sons, N.Y., 1974, p. 200. G. Schlessinger, Inorg. Syntheses, 3, 173 (1960). J. D. White and J. W. Newton, J. Phys. Chem., 13, 2117 (1971). 197 68. . Bailor, Jr. and E. M. Jones, Inorg. Syntheses, J c 1, 35 (1939). 69. A. Werner and F. Braunlich, Z. Anorg. Ch., 22, 125, 141 (1900). 70. A. Werner and F. Braunlich, Z. Anorg. Ch., 22, 95 (1900). 71. E. Bohm, Z. Anorg. Ch., 23, 338 (1905). 72. R. S. Nicholson and I. Shain, Anal. Chem., 36, 706 (1964). —_ 73. Z. Galus, "Fundamentals of Electrochemical Analysis", E. Horwood, N.Y., 1976, p. 236. 74. J. Koutecky, Chem. Listy, 21, 323 (1953). 75. J. Koutecky, Coll. Czech. Chem. Commun., 18, 597 76. J. Weber and J. Koutecky, Coll. Czech. Chem. Commun., .29, 980 (1955)- 77. K. B. Oldham and E. P. Parry, Anal. Chem., 40, 65 78. J. E. B. Randles, Can. J. Chem., 31, 238 (1959). 79. A. J. Bard and L. R. Faulkner, "Electrochemical Methods: Fundamentals and Applications", J. Wiley and Sons, N.Y., 1980, p. 288. 80. R. N. Adams, "Electrochemistry at Solid Electrodes", M. Dekker, N.Y., 1969, Ch. 7. 81. S. Gilman, in "Electroanalytical Chemistry", A. J. Bard (ed.), M. Dekker, N.Y., 1967, pp. 112—192. 82. J. Clavilier, A. Hamelin and G. Valette, C. R. Acad. Sci., Ser. C, 265, 221 (1967). 83. J. R. Gardner and R. Woods, J. Electroanal. Chem., - Q1. 285 (1977). 84. R. F. Lane and A. T. Hubbard, J. Phys. Chem., 12, 808 (1975). 85. M. A. Habib and J. O'M. Bockris, in "Comprehensive Treatise of Electrochemistry", J. O'M. Bockris, B. E. Conway and E. Yeager (eds.), Plenum Press, N.Y., 1980, Ch. 4. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102; 103. 104. 198 Y.-Chech Chiu and M. A. Genshaw, J. Phys. Chem., 13, 3571 (1969); 13, 4325 (1968). J. O'M. Bockris, S. D. Argade and E. Gileadi, Electrochim. Acta, 33, 1259 (1969). H. R. Hurwitz, J. Electroanal. Chem., 32, 35 (1965). E. Dutkiewitz and R. Parsons, J. Electroanal. Chem., 2;, 100 (1966). F. C. Anson, Anal. Chem., 33, 54 (1966). M. J. Weaver and F. C. Anson, J. Electroanal. Chem., 60, 19 (1975). K. J. Vetter and J. W. Schultze, Ber. Bunsenges Phys. Chem., 13, 920 (1972). M. J. Weaver and F. C. Anson, J. Electroanal. Chem., S. N. Frank and F. C. Anson, J. Electroanal. Chem., P.8D. Tyma, Ph.D. Thesis, Michigan State University, 19 0. K.8L. Guyer, Ph.D. Thesis, Michigan State University, 19 1. G. M. Schmid and M. E. Curley-Fiorino, in "Encyclo- pedia of Electrochemistry of the Elements", A. J. Bard (ed.), M. Dekker, N.Y., 1975, p. 87. T. W. Kallen and J. E. Early, Inorg. Chem., 19’ 1149 (1971). W. Bottcher, G. M. Brown and N. Sutin, Inorg. Chem., lg. 1447 (1979). Dr. David Larkin performed the capacitance bridge measurements. M. Brieter, Electrochim. Acta, 3, 925 (1963). M. C. Banta and M. Hackerman, J. Electrochem. Soc., 111, 114 (1964). R. H. Wopschall and I. Shain, Anal. Chem., 32) 1514, 1527, 1535 (1967)- M. Sluyters-Rehbach and J. H. Sluyters, J. Electro— anal. Chem., 33, 831 (1975). 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 199 E. Laviron, J. Electroanal. Chem., 32, 355 (1974). A. P. Brown and F. C. Anson, Anal. Chem., 32, 1589 (1977). H. Angerstein—Kozlowska, J. Klinger and B. E. Conway, J. Electroanal. Chem., 13, 45 (1977). D. T. Sawyer and J. L. Roberts, Jr., "Experimental Electrochemistry for Chemists", J. Wiley and Sons, N.Y., 197“, Pp- 338, 339. A. J. Bard and L. R. Faulkner, "Electrochemical Methods: Fundamentals and Applications", J. Wiley and Sons, N.Y., 1980, pp. 229, 525. S. Gilman, in "Electroanalytical Chemistry", A. J. Bard (ed.), M. Dekker, N.Y., 1967, p.117. J. H. Cristie, G. Lauer and R. A. Osteryoung, J. Electroanal. Chem., 1, 60 (1964). M. J. Weaver, H. Y. Liu and Y. Kim, Can. J. Chem., 59, 1944 (1981). K. M. Joshi, M. Mehl and R. Parsons, in "Transac- tions of the Symposium on Electrode Processes", E. Yeager (ed.), J. Wiley and Sons, N.Y., 1961, p. 249. L. Gierst, J. Tondeur, R. Cornelissen and F. Lamy, paper presented at Moscow C.I.T.C.E. meeting, 1963. For a review of quartenary ammonium ion effects see, A. N. Frumkin in, "Advances in Electrochemistry and Electrochemical Engineering", P. Delahay (ed.), Integscience, N.Y., Vol. 1, 1961, p. 65; Vol. 3, 1963, p. 2 3. S. N. Frank, Ph.D. Thesis, California Inst. of Tech., 1974. L. M. Baugh and R. Parsons, J. Electroanal. Chem., .22, 407 (1972). P. D. Tyma and M. J. Weaver, J. Electroanal. Chem., 111, 195 (1980). S. N. Frank, M. J. Weaver and F. C. Anson, J. Electro- anal. Chem., 32, 387 (1974). J. Koryta, Coll. Czech. Chem. Commun., 8, 206 (1953). 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 200 A. J. Bard and L. R. Faulkner, "Electrochemical Methods: Fundamentals and Applications", J. Wiley and Sons, N.Y., 1980, p. 501. B. E. Conway, "Theory and Principles of Electrode Processes", Ronald Press, N.Y., 1965, Ch. 4 and 5. The complex diameter is for a cis _—dithiocyanate with two opposing ammonia ligands and bisects the cis- 1igand angles. Radii were obtained from, J. E. Huheey, "Inorganic Chemistry: Principles of Structure and Reactivity", Harper and Row, N.Y., 1972, pp. 73-77. P. Delahay, "Double Layer and Electrode Kinetics", Interscience, N.Y., 1965, p. 43. R. Parsons, Surface Sci., 2, 418 (1964). R. Parsons, Croat. Chem. Acta, 22, 281 (1970). A. N. Frumkin, J. Electroanal. Chem., 2, 173 (1965). S. Trassatti, in "Advances in Electrochemistry and Electrochemical Engineering", H. Gerischer and C. W. Tobias (eds. ), J. Wiley and Sons, N.Y., 1977, p. 258. R. A. Marcus, Can. J. Chem., 31, 155 (1959). A. N. Frumkin, B. Damaskin, N. Grigoryev and I. Bagotskaya, E1ectrochim.Acta, 19, 69 (1974). R. Parsons and R. Payne, Z. Phys. Chem., 9§, 9 (1975). R. Kh. Burshtein, A. G. Pschenichnikov and L. A. Sherchark, Elektrokhimiya, 3, 332 (1969). G. M. Schmid and N. Hackerman, J. Electrochem. Soc., 332, 243 (1962). A. N. Frumkin and O. A. Petrii, Electrochem. Acta, 29, 347 (1975). S. Ya. Vasina and O. A. Petrii, Sov. Electrochem., g, 231 (1970). D. T. Larson and D. L. Cash, J. Phys. Chem., 13, 2814 (1969). T. E. Felter and A. T. Hubbard, J. Electroanal. Chem., 100, 473 (1979). 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 201 R. A. Osteryoung, G. Lauer and F. C. Anson, J. Electro- chem. Soc., 110, 926 (1963). E. Yee, R. J. Cave, K. L. Guyer, P. D. Tyma and M. J. Weaver, J. Am. Chem. Soc., 101, 1131 (1979). S. Trassatti, J. Electroanal. Chem., 33, 351 (1971). G. F. Voronina, L. A. Larin and T. V. Kalish, Elek- trokhimiya, 23, 172 (1980). E. Momot, M. Bonnemay and G. Bronoel, C. R. Acad. Sci., Ser. C, 270, 2108 (1970). R. M. Reeves, in "Modern Aspects of Electrochemistry, Vol. 9", B. E. Conway and J. O'M. Bockris (eds.), Plenum Press, N.Y., 1974, Ch. 4. The actual reactant concentrations being compared are: for the clean surface, the amount of reactant bound to the metal surface; for the iodide-covered surface, the amount of reactant in the layer of thickness r adjacenttx>the iodide layer. J. E. Huheey, "Inorganic Chemistry", Harper & Row, N.Y., 1972, pp. 350, 606. D. J. Barclay and J. Caja, Croat. Chem. Acta, 33, 221 (1971). This exploration often arises in attempts to explain the trans effect in platinum complexes. Examples are: J. Chatt, L. A. Duncanson and L. M. Venanzi, J. Chem. Soc., 4456 (1955); and L. E. Orgel, J. Inorg. Nucl. Chem., 2, 137 (1956). M. J. Weaver, Israel J. Chem., 23, 35 (1979). V. Srinivasan, S. W. Barr and M. J. Weaver, submitted to Inorg. Chem. J. F. Llopis and F. Colom, in "Encyclopedia of Electro- chemistry of the Elements", A. J. Bard (ed.), M. Dekker, Vol. VI, N.Y., 1976. R. G. Barradas and B. E. Conway, Electrochim Acta, 2. 319. 349 (1961). B. B. Damaskin, A. A. Survila, S. Ya. Vasina and A. I. Fedorova, Sov. Electrochem., 3, 729 (1967). D. Gaswick and A. Haim, J. Am. Chem. Soc., 23, 7347 (1971). 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 202 A. J. Miralles, R. E. Armstrong and A. Haim, J. Am. Chem. Soc., 22, 1416 (1977). J.-J. on, P. L. Gaus and A. Haim, J. Am. Chem. Soc., 101, 6189 (1979). F. R. F. Fan and E. S. Gould, Inorg. Chem., 33, 2647 (1974)- F. R. F. Fan and E. S. Gould, Inorg. Chem., 33, 2639 (1974)- K. L. Scott and A. G. Sykes, J. Chem. Soc., Dalton Trans., 1832 (1972). P. Siders and R. A. Marcus, submitted to J. Am. Chem. Soc. J. F. Endicott and H. Taube, J. Am. Chem. Soc., 86, 1686 (1964). _— D. Stranks, Disc. Far. Soc., 22, 116 (1960). E. Buhks, M. Bixon, J. Jortner and G. Navon, Inorg. Chem., 23, 2014 (1979). H. Stynes and J. Ibers, Inorg. Chem., 23, 2304 (1971). N. S. Biradar, D. R. Stranks and M. S. Vaidya, Trans. Far. Soc., 33, 2421 (1962). H. Bartelt and S. Landazury, J. Electroanal. Chem., _2_2, 105 (1969). ' H. Bartelt, Electrochimica Acta, 23, 307 (1971). . Yamaoka, Z. Phys. Chem., Leipzig, 259, 2 (1978). H J. F. Endicott, R. R. Schroeder, D. H. Chidester and D. R. Ferrier, J. Phys. Chem., 11, 2579 (1973). T G ( . Meyer and H. Taube, Inorg. Chem., 22, 2304 (1971). . Brown and N. Sutin, J. Am. Chem. Soc., 101, 883 1979)- A. B. Lamb and A. T. Larson, J. Am. Chem. Soc., 32, 2024 (1920). J. Bjerrum, "Metal Ammine Formation in Aqueous Solu- tion", COpenhagen, 1941, reprinted 1957, p. 251, 295. 173. 174. 175. 176. 177. 178. 179. 203 E. Appelman, M. Anbar and H. Taube, J. Phys. Chem., 93. 126 (1959). C. J. Ballhausen and C. K. Jorgensen, Acta Chem. Scand . 2. 397 (1955). R. Murray, S. F. Lincoln, H. H. Glaeser, H. W. Dodgen and J. P. Hunt, Inorg. Chem., 3, 554 (1969). K. Ogura and Y. Fukushima, Electrochim. Acta, 23, 93 (1981)- H. A. Laitinen and P. Kivalo, J. Am. Chem. Soc., 13, 2198 (1953)- J. B. Flanagan, K. Takahashi and F. C. Anson, J. Electroanal. Chem., 257, 257 (1977). G. C. Barker and J. A. Bolzan, Z. Anal. Chem., 216, 215 (1966).