11 ' , ‘1; 1:"{11' 111'. _'l ‘1 41i111t‘.“1 (.1111 11“") 111.10 .111! «111-, 11‘1“ 11111111111 1111111111111 11".11! '11 “H11 1-11 1 1 11111111111111.1111'111111 11111111 1 11'111111111 11 1 1110 .11‘ ‘11, 1 '~ 111111151 111811111 1121711: '11" 1111: 1 " [n 1111*“ ’1 11111 11 1,111 I! H. ' 3‘3 4“ v“ . ._ v". - Arr-w- - manta [QM' A ,, . -... . -' .r: .;'- wkaoel'st-W'. 11111111'1111111 ‘- . 1 :1“ " 11, 1 1421551111 1I11‘ 1" .1v1'1'!1"'1.111 1 [.1 113' |.1H >,1'1111f|1111111,’1111j 1 4 1 11' ‘.Ii 111,11-111‘11'1111?‘ i 1 11.11 :1. .11‘11'1‘1 11.11 :111 1L7“ 1111b - . 311.111.1111 ‘ 11'1 111' 11 111111 . 1' 11 '1 111111111141 11111111 11.111111 .'.'1..1 111111111111.1.1.1:.1:1111111111111141111111111111111'1‘111 111111111“" 1 111‘ THESI S This is to certify that the dissertation entitled Electron Transfer Kinetics at the Polycrystalline Silver-Aqueous Interface presented by Kendall Louis Guyer has been accepted towards fulfillment of the requirements for Ph . D . degree in Chemi stry jor pro WM 4 ”ll M Date (019!K‘ MS U i: an Affirmative Action/Equal Opportunity Institution 0- 12771 Arr—fi __ a“ “a “h. ..—~44 a MSU LlBRARlES ”- RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. ELECTRON TRANSFER KINETICS AT THE T’- ‘rm: IPOLYCRYSTALLINE SILVER-AQUEOUS INTERFACE BY Kendall Louis Guyer A DISSERTATION Submitted to Michigan State University $.11: partial fulfillment of the requirements . _ for the degree of DOCTOR OF PHILOSOPHY . ‘ l rDepartment of Chemistry A 1981 . ’V‘ o 0“ F ‘ u- >- 0“. h- I ...- - a u u- ‘ I- - u-u-vg. y.‘ . "O'l. F .. l ~-~-..... ‘ a. u- . "um... .. l _ ‘~ to. _‘ ‘- >n..... . “NH”: . 'n . o- n. _C. a .oI-._“- ~ g o. .. . .a- \n _I\.- ' u...- 5. ~- 9 ' t.. l." ‘\ ‘- h V \ o d I,‘ n c a O. i A ‘u‘ I :'~ ’b . a. \. d.. a l ‘I ~- .r . 3" a a , ‘ I - ‘l s ~~’ - ABSTRACT _ ELECTRON TRANSFER KINETICS AT THE POLYCRYSTALLINE SILVER-AQUEOUS INTERFACE BY Kendall Louis Guyer The behavior of polycrystalline silver electrodes in I'aumuedus electrolytes was investigated with a series of ...itransition metal complexes that undergo irreversible one- :Lelectron reductions and oxidations. The reactants were in a Hageneral Cr(III) and Co(III) ammine and aquo complexes that '\Eiserved as model systems for a study of electron transfer ,;i%kinetics. By altering the ligands bound to the metal ~teactant center, electrochemical reaction pathways could fl‘eéfie selected to be inner- or outer-sphere. In weakly adsorbing electrolytes the reactivity of 3. {fisgme cationic Co(III) ammine complexes that are reduced ”L yfia outer-sphere pathways is similar at mercury and '.‘\ , ,Jafilycrystalline silver electrodes. The addition of anions ltipn of these outer-sphere reactants at silver odes, and is a consequence of strong specific ...-'-: ._,..-. . 0: v!" “A-- c o..'-- VV' gnu-91'. . up a .0.- 0.10. d . u v---o-.— . . p n I."’ in... . -o:-Qn 0 c \ .O'mi‘. * . LL‘OvAo b "Hue“ . -"-~bu : n3: ' Ohv . ... '_‘,'6 s A. H . I-~ '.n‘.'.bs 0L. ' , x: “‘7'” ‘O..‘ “ .o' ' "s 3:55 ."" b. O A ’. .‘ 5 “In, ~-“l ‘ A - Q - 5 Q I .I _'-."n ., ..,... ’- Kendall Louis Guyer “395.5 inner-sphere (ligand—bridged) pathways. The apparent - reactivities of these complexes are generally larger on silver electrodes than on mercury electrodes at the same electrode potential. The increased reactivity of the iiigand-bridged Co(III) and Cr(III) reduction reactions on silver compared to mercury is in part due to larger reactant adsorption on the former surface. The adsorption xaf reactants that contain azido and isothiocyanato ‘ligands was quantified for silver electrodes and their electron transfer reactivities were measured in the adsorbed state by linear sweep voltammetry. The remaining differences in the electrocatalytic properties of mercury and silver electrodes (outside of variations in reactant adsorption) can be attributed to non-bridging ligand effects and a possible lowering of '"the intrinsic free energy barrier to electron transfer .,lfor reactions occurring at silver. ‘ Surface enhanced Raman spectroscOpY (SERS) was used iffio study some aromatic nitrogen heterocycles and Ru(II) -{:§entaamine complexes of these molecules adsorbed on :.uaroughened polycrystalline silver electrodes. A ‘4yséomparison of the SERS data for the adsorbed heterocycles ‘i‘aad their Ru(II) complexes was performed in order to To My Parents 41 I- H In ’ - ..‘...- .. - ' U.» ‘i 1‘.- o .. . Q “"1. .. ‘ - V ”1‘...” a .. ‘ P n: -.~~ -- - .““ ‘n. .c .. .- " '41- U‘ .‘\ ‘nk..-~ .- .s..‘ O t V ‘0‘“ . ~ _ n, J“ ‘- g a u .‘ -A b- a..‘ I z ”I“ ~ N "v‘\ :‘. ‘ha‘. c “‘ ."~“ . ‘b c t I“ ‘ L- .“ 1 -v A.‘ .; ‘ . I.-..~l'urv "v. .\‘ . . \ V v u U. . . I I 3'. ‘ . 1”, ~ I '~~ . \. ' ' '1'- l . ACKNOWLEDGEMENTS A few lines of prose are small reward for those ifluase friendship and guidance have seen me through my -ggraduate career. My list of thanks is far from complete. lnxt my gratitude extends to all who have aided me. Dr. Michael Weaver, my preceptor, has kept me on . jtrack in my pursuit of a doctorate. He has provided the . pscheduling for and connections to reward which are .';%exce11ed by few. V-za Those who have worked beside me in the laboratory .ifhaVe made my tasks all the more enjoyable. Paul Tyma q‘iahd Ed Yee have stood shoulder to shoulder in patiently éfihowing me the arts of the laboratory. Steve Barr and :fEu Ned Larkin have given me their invaluable advice j¥ifi6never asked. O’The chemistry department staff has provided many of 3; tools of my research. Marty Rabb, in particular, ‘ much time and expertise in instrument design. 7 dMy thanks to Margy Lynch for typing a manuscript If? handwritten copy which had the qualities of iglyphics. fig; Bruce Kowert, Dave Christmann, Scott Sandholm, ETieckelmann have all conspired to provide for Iieing outside of chemistry. ;,;1d also like to thank my family for their encouragement in all my adventures. iv 'p:00a' "‘4‘... o 1'.- a .10. A. v. .‘v—‘ .- .c‘ou- I. (I! h) a J- . (" TABLE OF CONTENTS ;:Chapter 7.3.181: OF TABLES. . . . . . . . . . . . 0'1‘ 'rrjgrsis'r OF FIGURES. . . . . . . . . . . . - I. INTRODUCTION. . . . . . . . . . . The Electrode Material. . . Transfer Studies. . . . . Pathways for Homogeneous and Interface. . . . . . . . . . EXPERIMENTAL METHODS. . . . . . Materials. . . . . . . . . . Water Purification. . . . . Oxygen Removal. . . . . . . Electrolytes. . . . . . . . Transition Metal Reactants. a. Co(III) Complexes. . . c. Ruthenium Complexes. . d. Miscellaneous Reactants , Electrodes. . . . . . . . . b. Chromium (III) Complexes. Reactants for Heterogeneous Charge Heterogeneous Electron Transfer. Surface Enhanced Raman Spectroscopy Applied to the Electrode-Electrolyte .26 .26 .28 .28 .29 .29 .32 .33 .33 . . a: V '. D I... 5" C (11 ("'1 D. Apparatus. . . . . . . . . . . . . . . . 1. Electrochemial Cells. . . . . . . . 2. Reference Electrodes. . . . . . . . 3. General Instrumentation. . . . . . . Electrochemical Techniques. . . . . . . . 1. Differential Capacitance. . . . . . 2. DC Polarography. . . . . . . . . . . 3. Rotating Disk Voltammetry. . . . . . 4. Normal Pulse Polarography. . . . . . 5. Cyclic Voltammetry. . . . . . . . . 6. Methods for the Direct Determination of Reactant Adsorption and Kinetics in the Adsorbed State. . . . . . . . Errors of Measurement. . . . . . . . . . THE SILVER-AQUEOUS ELECTROLYTE INTERFACE. . Aspects of Metal-Electrolyte Interfaces.. The Gouy-Chapman-Stern Model of the Double Layer. . . . . . . . . . . . . . . Double Layer Capacitance. . . . . . . . . Measurement of Anion Adsorption on Silver Electrodes. . . . . . . . . . . . Double Layer Properties of Silver. . . . 1. Experiments Performed with Polycrystalline Silver. . . . . . . 2. Estimation of the Surface Roughness Factor I I I I I I I I O I I I I I O 3. Measurements with Other Weakly Adsorbing Electrolytes. . . . . . . 4. Electrolytes Which Adsorb Strongly on Silver. . . . . . . . . . . . . . vi Page .44 .45 .48 .48 .49 . 49 .52 C 54 .57 .57 .58 .63 . .7 2 . 75 . 75 . 79 . 80 .85 '.""' ‘ - ',. .V. . A'-'l . ".... " ll, 3'. General Features of the Silver- Electrolyte Interface. . . . . . . . . . . 92 OUTER-SPHERE ELECTRODE REACTIONS. . . . . . . .95 It. The Precursor Model of Electrode Reactions. . . . . . . . . . . . . . 96 B. The Activation Process for Heterogeneous Charge Transfer. . . . . . . . . . . . . . 99 A Rate Formulation from Marcus Electron Transfer Theory. . . . . . . . . . . . . .104 l. The Effect of Potential on the Rate Constant for Electron Transfer. . . .106 2. The Calculation of a Rate Constant from a Measurement of Current. . . . 109 The Reactivity of Outer-Sphere Complexes at Polycrystalline Silver -- Results and Discussion. . . . . . . . . . . . . . 111 l. Co(III) Ammine Reactants in Weakly Adsorbing Electrolytes -- Results and Discussion. . . . . . . . . .112 2. Transition Metal Aquo Complexes Mn+I I I I I I I I I I I I I I I I I 125 aq 3. Results. . . . . . . . . . . . . . . 129 4. Discussion. . . . . . . . . . . . . .132 5. Kinetic Probes for the Specific Adsorption of Anions on Silver. . . .138 6. Results. . . . . . . . . . . . . . . 140 7. Discussion. . . . . . . . . . . . . .145 8. Perbromate Reduction as a Kinetic Probe. I I I I I I I I I I I I I I I 149 9. Results. . . . . . . . . . . . . . . 149 10. Discussion. . . . . . . . . . . . . .154 r.:ao:? I... M. a- . lvo" fl. _. I... '5‘ 1 FL- 50 Ba- 3 I I oc' I | ' J0 g~4 ‘ u- on- ‘. - 5 Oh \v. o ‘ ‘ '\'\‘-- 'ou .‘,“;: l I do 1 ‘ n a. I .~. No \" g" c.. A‘ t» . l‘ . . '~.. Chapter- Page V. THE DETERMINATION OF REACTANT ADSORPTION. . . .156 A- Chronocoulometry. . . . . . . . . . . . . 157 1. Results and Discussion. . . . . . . .160 B. Kinetic Probes for Complex Ion Zhdsorption. . . . . . . . . . . . . . . . 163 1. Results and Discussion. . . . . . . .164 C. Fast Linear Sweep Voltammetry. . .. . . .169 1. Results and Discussion. . . . . . . .174 VI. INNER-SPHERE ELECTRODE REACTIONS. . . . . . . 187 A" The Energetics of Inner- versus Outer- Sphere Reaction Pathways. . . . . . . . . 188 B. Results. . . . . . . . . . . . . . . . . .190 1. Monobridged Inner—Sphere Reactions of Cr(III) Complexes. . . . . . . . .190 Verification of Reaction Pathway. . i194 2. Multibridged Inner-Sphere Reactants..l98 The General Applicability of the Precursor Formulism. . . . . . . . . 200 C. Discussion. . . . . . . . . . . . . . . . 202 1. Metal Substrate Effects on the Reactivity of the Inner-Sphere Precursor Complex. . . . . . . . . . 202 2. Non-Bridging Ligand Effects. . . . . 205 3. Bridging Ligand Effects in Electron Transfer Reactions. . . . . . . . . .207 THE APPLICATION OF SURFACE ENHANCED RAMAN SPECTROSCOPY TO THE DETECTION OF ADSORBED METAL COMPLEXES. . . . . . . . . . . . 211 ReSUItS- I a o o o o o o a o o a o o o o 214 viii Hoot-1' v. ”n"' at.‘ tlobv -J I” r“) :l‘ (I) 9n “'0 (') (.1 o (I) u... \ o.~‘ " Y‘s-— / B. Discussion. . . . . . . . . . . . . . . . 225 1.. The Detection of an Adsorbed Metal Complex. . . . . . . . . . . . 225 2. Metal-Nitrogen Stretching Vibrations of M(II) Complexes. . . . 228 VIII . MISCELLANEOUS EXPERIMENTS . I I I I I I I I I231 A” Activation Parameter Measurements for _ the Electroreduction of Some Co(III) and ' Cr(III) Complexes on Polycrystalline ‘ Silver and Mercury Electrodes. . . . . . .231 l. The Calculation of the "Ideal" Entropy and Enthalpy of Activation. . . I I I I I I I I I I I232 2. Results. . . . . . . . . . . . . . . 233 3. Discussion. . . . . . . . . . . . . .237 B. Apparent Reactivities of Monobridged Co(III) Reactants at Polycrystalline 5.- Silver Electrodes. . . . . . . . . . . . .239 7':,:Ix. CONCLUSIONS AND SUGGESTIONS FOR FUTURE .239. RESEARCH. . . . . . . . . . . . . . . . . . . 243 - A. Conclusions . . . . . . . . . . . . . . . 243 B. Suggestions for Future Research. . . . . .248 .LIST OF REFERENCES. . . . . . . . . . . . . . . . . 252 ix LIST OF TABLES Page Kinetic Parameters for the Reduction of Cc>(III) Ammine Complexes at Hg and Ag Electrodes at 25°C. .. . . . .. . . . .. .113 I)ouble Layer Parameters from Co(III) Inmnine Electroreduction Kinetics on a Polycrystalline Silver Electrode. . . . . . .123 Rate Parameters for Water Exchange for Various Transition Metal Aquo Complexes at 25°C. . . . . 126 Rate Parameters for Ligand Substitution for Various Transition Metal Aquo Complexes at 25°C. . . . . . . . . . . . 127 Kinetic Parameters for the Electrooxidation of Transition Metal Aquo Complexes at Polycrystalline Silver and Mercury at 25°C. 0 n o o I o c o o n o o I I I I I I I I 130 Double Layer Corrected Rate Parameters ‘ for the Electrooxidation of Transition 7.. Metal Aquo Complexes at Polycrystalline Lijvsilver and Mercury at 25°C. . . . . . . . . .134 ‘ol': . .II" 1 ' 1 II! (T) |,. A, I”A.'-' I! A‘ ‘I up! I‘I angv a I. I.‘ 0“ ' '._1. 4" .- I Page Free Energy of Adsorption of Various Anions on Polycrystalline Silver from Kinetic Probe Measurements and Single Crystal Silver by Hurwitz—Parsons Analysis of Capacitance Data. . . . . . . . . . . . . 146 Cflnronocoulometric Adsorption Data for (3—Cr(OH2)4(NCS)2+ Reduction on Polycrystal- line Silver in 0.1 M NaClo4 at 23°C. . . . . 161 Kinetic Probe Data for Complex Ion Adsorption on Polycrystalline Silver at 4400 mV versus SCE in 0.1 M NaClO4 +5mMHClO4,T=23°C. . . . . . . . . . . .165 Adsorption Data for Complex Ion and Anion Adsorption on Polycrystalline Silver and Mercury from Linear Sweep Voltammetry and Kinetic Probe Measurements in 0.1 M NaClO4 + 2—5 mM Helo4 at 23°C. . . . . 175 Adsorption Data for some Cr(III) Isothiocyanate Complexes on Polycrystal- line Silver and Mercury Electrodes. . . . . .181 Apparent and Surface Rate Parameters for the Monobridged Inner-Sphere Reduction of Some Cr(III) Complexes at Polycrystalline A ' 7 Silver and Mercury Electrodes at 23°C. Electrode potential = -600 mV versus SCE. . .192 xi ‘ A ‘IL 0 nv;r- gong-u. V no» r.“ Ina Lb :Fv III ow. 'u. A 9' r V; .4. q a '1- ...‘ ‘V.'» I V! ‘1! If )I n . u“v-,. ‘a... w“ H“_ a I. ‘ll ‘0: a r L_. Page Kinetics Parameters for the Multibridged Inner—Sphere Reduction of Some Cr(III) and Co(III) Complexes on Polycrystalline Silver and Mercury at 23°C. . . . . . . . . .199 A Comparison of Bridging Ligand Effects for the Inner-Sphere Reduction of Some Cr(III) Complexes on Mercury and Polycrystalline Silver Electrodes. . . . . . 209 Fundamental Vibrational Frequencies of Liquid Pyridine, Pyridine in Aqueous Solution, and of Pyridine and RuII(NH3)5pyridine2+ Adsorbed on Polycrystalline Silver (SERS) at an Electrode Potential E(V) vs. SCE. . . .219 Fundamental Vibrational Frequencies of Liquid Pyrazine, Pyrazine in Aqueous Solution, and of Pyrazine and RuII(NH3)5pyrazine2+ Adsorbed on Polycrystalline Silver (SERS) at an Electrode Potential E(V) vs. SCE. . . .221 Fundamental Vibrational Frequencies of 4'4-dipyridyl and of 4'4-dipyridyl and 2+ RuII(NH 4,4'-dipyridyl Adsorbed on 3’5 Polycrystalline Silver (SERS) at an Electrode Potential E(V) vs. SCE. . . . . . .223 (II (1! I d ‘~.‘IO' I II'-. Cl I‘Ap ufiuot Page Frequencies for Some Ring Stretching Vibrational Modes of Various Nitrogen Heterocycles and Their RuII(NH3)5L2+ Complexes Adsorbed on Polycrystalline Silver Electrodes. . . . . . . . . . . . . . 227 Activation Parameters for the Electroreduc- tion of Some Inner- and Outer-Sphere Transition Metal Reactants at Polycrystalline Silver and Mercury Electrodes in 0.1 M NaClO4 + 2-5 mM HC104I I I I I I I I I I I I I I I .234 .fl 5, xiii LIST OF FIGURES Page Electronic States for Co(III)/(II) and Cr(III)/(Ii) Couples. . . . . . . . . . . . . 11 (A) Homogeneous Outer-Sphere Electron Transfer. Co(NH3)g+ + Ru(NH3)§+ 2+ + Ru(NH3)2+; Co(NH3)§+ 6 H 2+ + ESE?» Co(OH2)6 + 6NH4 , (B) Heterogeneous + Co(NH3) Outer-Sphere Electron Transfer. 3+ - 2+ 6 + e ———+ Co(NH3)6 , 2+ H 2+ + 3)6 m? CO(OH2)6 + 6NH4 o a o o a o 016 (C) Homogeneous Inner—Sphere Electron III 2+ 2+ Transfer. Co (NH3)5C1 + Craq Co(NH3) Co(NH ———+ COII(NH ClCrIII(OH2)g+; )4+ H+ 2 5 f3§€+ + + ,+ SNH4 ; (D) Heterogeneous Inner-Sphere Electron Transfer. 3’5 II III 2+ Co (NH3)5ClCr (OH Co(OH2)6 2 + Cr(OH2)5C1 2+ - + Co(NH3)5C1 + i ———+ CO(NH3)5C1 ; H 2+ + - 3)5 fast C°(OH2)6 + 5NH4 + Cl ....18 (A) Possible Inner-Sphere Pathway for C1 + Co(NH - ,Homogeneous Electron Transfer Involving a 2+ ‘ . 3+ V.Labile Reactant. Co(NH3)6 + Craq 2 II III 5+ ¢.fi———+ Co (NH3)6Cr (0H2)5 ; n_, 5:. 1:3, m- . V . v- -- ("no Av 09‘ -V‘-‘I~ “V. a s V v V'q' ‘5‘... kn . u... ‘- Page 5+ + 2+ 2’5 f5§€+ C°(OH2)6 + 5NH4+; (B) Homogeneous COII(NH3)6Cr:II(OH 3+ * 999325332 Outer-Sphere Electron Transfer Involving a Labile Reactant. Co(NH3)2+ + Cr2+ aq ———+ Co(NH3)§+ + Cr(OH2)2+; Co(NH3)§+ + H 2+ + WCO(OH2)6+6NH4............21 Insulated Shaft Rotating Disk Electrode. Materials list: (a) stainless steel; (b) Delrin (filled nylon); (c) copper spring; (d) Kel-F (bored 8-10 thousandths inch undersize for electrode seal); (e) brass; (f) silver (silver-soldered to brass holder). . . . . . . . . . . . . . . . .22 (A) Cyclic voltammogram for polycrystalline silver electrode (geometrical area = 0.049 cmz) at 0.1 V sec-1 in 0.1 M NaF following cyanide electropolishing; (B) Cyclic voltammogram as in Figure 2:2A but including treatment in 2 M HClO 37 4. . . . . ‘Arrangement of equipment for surface- enhanced Raman spectroscopy experiments. . . .42 (A) Model of electrode-electrolyte ‘ ‘rinterface for double layer capacitance iLmeasurements. R8 = solution resistance, ch1 = double layer capacitance; XV _ A ‘ . ‘l H a:*; ‘ gas-Av I A II— U..u y 0. I 3" .\ u” S ’ 9 ~- -~~l d 4 'I " O" 1 C) 6.. 2'5. II..“ [D I: 23=1 . W: Page (B) Capacitance bridge for double layer capacitance measurements. R1, R2, R, and C are variable elements of the bridge; Rs and Cd1 are as described in Figure 2:4A. . . . . . . . . . . . . . . . . . . . . 46 Log-log reaction order plot of current i(amps) versus reactant concentration just outside the diffuse layer cs (mol cm-3) for the reduction of Co(NH3)5F2+ (circles) and c-Co(en)2(NCS)2+ (stars) on a polycrystalline silver electrode in 0.1 M NaClO + 5 mM HClO 4 4 of plot yields reaction order. For at 23°C. Slope Co(NH3)5F2+ slope = 0.94 and for c-Co(en)2(NCS)2+ slope = 0.57. . . . . . . . .50 Model of an electrode—electrolyte interface. ¢m is the potential at the electrode surface and ¢s the potential in bulk solution. The solid line shows the decay Of potential from ¢m to ¢S. The label OHP stands for outer Helmholtz plane and is the average center of charge for ions at their plane of closest ,approach to the electrode. . . . . . . . . . .60 ;T‘€q xv1 Page Concentration dependence of the electrode double layer capacitance Cgic at the potential of zero charge. Also shown are the inner layer capacitances CEZC of silver and mercury at their respective potentials of zero charge. . . . . . . . . . .66 Illustration of anion specific adsorption at an electrode surface. Anions have penetrated the inner layer of solvent molecules and are in direct contact with the electrode surface. . . . . . . . . . . . .70 Potential dependence Of the capacitance of a polycrystalline silver electrode in 21 mM (squares) and 100 mM (circles) aqueous NaF solutions. T = 23 t 1°C. Capacitance values were calculated with the geometric electrode area. . . . . . . . . 77 Plot of electrode charge qm versus electrode potential E for a polycrystalline silver electrode in 0.1 M aqueous solutions of KPF6 (solid circles) and NaClO4 (Open ‘ Circles). T = 23 : 1°C. Values of qm V? were calculated by employing a roughness .v‘f‘tfactor rs= 1.2. . . . . . . . . . . . . . . .82 xvii '.r": ..I“' .lhw an“ -- “.mub A—v .w» ‘4” H . . . s . o a a w «4‘ .n u ‘ I b a. a. pm“ :4 a» o. a; 04" I run .u. 3. In 9“ al.- -LQQ R.( 4. ’hl Page Potential dependence of the capacitance of a polycrystalline silver electrode in 50 mM aqueous NaZSO4. T = 23 : 1°C. Capacitance values were calculated with the geometric electrode area. . . . . . . . . 86 Electrode Charge qm versus electrode poten- tial E for a polycrystalline silver electrode _ o 2804. T - 23 i 1 C. Values of qm were calculated by employing in 50 mM aqueous Na a roughness factor rS = 1.2. . . . . . . . . .88 Free energy versus reaction coordinate for an electron transfer reaction. AGAE—p is the activation barrier to electron transfer within the precursor complex. -RT 1n Kp and -RT 1n Ks represent the work required to form the precursor and successor states, respectively. . . . . . . . . . . . .102 Rate—potential data for the electroreduc- tion of Co(NH3)§+ (circles) and Co(NH3)5F2+ (triangles) at the mercury-aqueous (Figure 4:2) and polycrystalline silver- aqueous (Figure 4:3) interfaces; kapp is the apparent rate constant. Supporting electrolytes ([H+] = 5 mM): closed symbols, 0.1 M KPFG; open symbols, 0.1 M NaClO4. -The dashed and dotted-dashed straight lines xviii . - .Q-V‘ ..oo0' Pb. Au; :5 \" ,\ IV as V. .nd FL. 3. a“ _. .5 can A: .‘d "A Page are plots of the double layer corrected. rate constants loglo kcorr versus E for Co(NH3)5F2+ and Co(NH3)2+ respectively. These latter data were Obtained with mercury electrodes from values of kapp and double layer compositional data employed in Equation (4.13). . . . . . . . . 116 Ionic strength dependence of the electro- reduction of Co(NH3)5F2+ on polycrystalline silver in NaClO4 aqueous electrolytes ([H+] = 5 mM). Electrolyte concentrations: 0.025 M (circles); 0.1 M (triangles); 0.5 (squares). The dashed line is the double layer corrected rate parameter log10 kcorr versus E for Co(NH3)5F2+ reduction on mercury obtained with values of ka and double layer compositional PP data employed in Equation (4.13). . . . . . .118 Rate-potential data for the electroreduc- tion of Co(NH3)5F2+ in aqueous 0.1 M NaF (pH ; 4) at polycrystalline silver (triangles) and mercury (circles) electrodes. The dashed line is a plot of the double layer ”corrected rate parameter loglo kcorr for ',"Co(,NH3)5F2+ reduction on mercury electrodes ’F \ C '+ v'V’ I v ..¢.o . .1- on .‘e o. 1" p: U. Page obtained from values of kapp and double layer compositional data applied in Equation (4.13). . . . . . . . . . . . . . . 121 Rate-potential data for the electroreduc- tion of Co(NH3)5F2+ at a polycrystalline silver electrode in 0.1 M NaClO4 + x mM Cl- aqueous electrolytes ([H+] = 5 mM). Chloride ion concentrations: 1, 0; 2, 2 EM; 3, 5 uM; 4, 20 EM; 5, 50 uM; 6, 100 EM; 7, 1 mM; 8, 10 mM. . . . . . . . . . . . . . . . . . .141 Rate-potential data for the electroreduction of Co(NH3)5F2+ at a polycrystalline silver electrode in 0.1 M NaClO4 + x uM NCS- aqueous electrolytes ([H+] = 5 mM). Thiocyanate ion concentrations x (uM): 0.0 (open circles); 0.1 (open triangles); 0.5 (Open squares); 1.0 (solid circles). . . 143 Rate-potential data for the electroreduction 2+ Of Co(NH3)5F at the polycrystalline silver-aqueous interface. Supporting electrolytes ([11+] = 5 mM): 0.1 M Nac1o4 (circles); 0.1 M NaClO4 + 10 mM KCl ‘ (triangles); 33 mM Nazso4 (squares). . . . . 150 XX Page _/ Rate Potential data for the electroreduction of Br04- at the polycrystalline silver— aqueous interface. Supporting electrolytes ([on‘] = 2 mM): 0.1 M NaClo4 (circles); 0.1 M NaClO + 10 mM KCl (triangles); 4 33 mM NaZSO4 (squares). . . . . . . . . . . .152 Rate-potential data for the electroreduction of Co(NH (circles) and Co(NH3)2+ 2+ 3’5F (triangles) on polycrystalline silver in 0.1 M NaClO4 + 5 mM HClO4 (open symbols) and in the same electrolyte with 0.5 mM Cr (NIHI3)5NCS2+ present in solution (closed symbols). . . . .166 (A) Linear sweep voltammogram at 20 V 5-1 for a polycrystalline silver electrode in 0.1 M NaClO + 5 mM HClO4. T = 4°C, 4 initial potential = +150 mV versus SCE; (B) Linear sweep voltammogram as in Figure 5:2A but with 20 uM c—Co(en)2(NCS)2+ . in solution. . . . . . . . . . . . . . . . . 172 Surface concentration Po (mol cm_2) versus log bulk concentration cb (EM) for f-’Cr(OH2)3(NCS)3 adsorbed on polycrystalline J silver (closed symbols) and mercury (open VIQSymbols) electrodes. Electroltyes: 1 polycrystalline silver, 0.1 M NaClO4 mercury, l M NaClO4-+10 mM HC104. i?*.2 mM HC104; xxi Page Electrode potentials: -200 mV versus SCE (circles); -100 mV verSus SCE (triangles). The data for mercury electrodes were taken from reference 96. . . 184 Rate-potential data for the electroreduction of Co(NH3)5F2+ (open symbols) and Cr(NH3)5C12_+ (Closed symbols) on polycrystalline silver electrodes. The symbol kapp stands for the apparent rate constant. Electrolytes: acidified 0.1 M NaClO4 (circles); acidified 0.1 M NaClO + 1 mM Cl' (triangles). . . . . 196 4 SERS spectra of 4,4'—dipyridy1 and Ru(II)(NH3)5 4,4'-dipyridyl2+ adsorbed on a roughened polycrystalline silver electrode in 0.1 M KCl. The abscissa is labeled with the relative wavenumber (cm’l; 647.1 nm excitation source). (a) 4,4'-dipyridyl, -0.2 V vs. SCE; (b) 4,4'-dipyridy1, -0u4V vs. SCE; (c) 4,4'—dipyridyl, -0.6vs. SCE; (d) Ru(II)(NH3)5 4,4'-dipyridy12+, -0.6 vs. SCE. . . . . . . .215 Apparent rate constant kapp versus electrode potential E for the reduction of Co(III)(NH3)5xn+ complexes on mercury (open symbols) and polycrystalline silver (closed symbols) electrodes. Electrolytes: silver xxii W Page electrodes, 0.1 M NaClo4 + 5 mM HC104; ‘J'r mercury electrodes, 1 M KF. Reactants: l;?:_ 2+ . _ 2+ .- Co(NI-I3)5NO3 (Circles), Co(NH3)5N3 ‘(triangles); Co(NH3)SSOz (squares); Co(NH NCS2+ (diamonds). The data for . 3’5 :”~ mercury electrodes were taken from reference 14. . . . . . . . . . . . . . . . .240 Chapter I. Introduction Events which involve the transfer of an electron from one reactant to another mediate much of the chemistry in our environment. A segment of the science and technology encompassed within the heading "Electron Transfer Reactions" is that involving solid metal electrodes. Solid metals acts as electrochemical reagents by supply— ing or accepting electrons from a co-reactant. The applications of metal surfaces at which redox reactions (1’2) conventional (5,6) occur include fuel cell technology, (3'4) catalysts for Chemical reactions, (7) batteries, and electrosynthesis. The activity in the study of solid metal electrodes has resulted in an extensive library of information regarding particular aspects of their behavior. The bulk of this material is of a technological nature and has been derived from empirical studies of specific electrode systems. Projects in this area have tended to originate in industry and generally involve complex electrode surfaces and reactions. For this reason, the quantity of knowledge which concerns fundamental electrode properties is meager by comparison. Fundamental aspects 1 2 of electrodes include the detailed understanding of the electrode-electrolyte interface (the electrolyte being the medium the co—reactant is contained in), the manner in which electron transfer proceeds at an electrode surface (i.e. how does the electron travel between the electrode and the co—reactant?), the relationship between reactant structure and reactivity (the rate at which electron transfer proceeds), and the extent of catalysis at different electrode surfaces (i.e. why does a given reactant undergo electron transfer more rapidly with Some surfaces than others?). Much of what is understood about electron transfer rEractions has been derived from homogeneous redox chemis- tr3r. Significant progress has been made in understanding thug way in which reactant structure affects the pathway Erna rate of a homogeneous electron transfer reaction. (8-10) Corresponding information for hetero- g‘irleous charge transfer reactions, which includes those pr<>ceeding at metal electrodes, has been obtained for tJIEB most part from work with liquid mercury electrodes. (ll-l4) Solid metal electrodes have only lDEEean touched upon in many regards. Investigations of £21Eéctron transfer have generally concerned phenomena ‘Vllfilch make identification of the rate controlling step JLI‘ the overall reaction (which includes the electron Itll=éinsfer process) difficult. Also, the systems are tl§3\lally isolated cases, and it is therefore difficult to Identify particular reactant structural features with . .-o 0‘ . .‘ .“ ,.. J” I , .no .a .uovv. -o-.~ - b... - Ita‘I p III 11‘ . .~. - ‘-~- 'v... . ."V. ‘I- \ _’ "lb q Ia~ o " 'I... e.” \- ...v 'v._ I ‘i. I; 5‘; reactivity by intercomparison. Examples of the studies to date with solid metal electrodes are oxygen (15’16) oxide film formation(l7) and the reactions of organic molecules.(18) Information on the basic thermodynamic and electro- reduction, chemical properties of solid metal electrodes has only begun to accumulate relatively recently. Studies of the electrode-electrolyte interface have been performed for several different solid metals by measurements of electrode capacitance(19-23) and the optical behavior 0f the interphasial region.(24-26) There has also been Scnne study of relatively simple electron transfer reactions in which the exact composition Of the reactant at the electrode surface is known and the rate limiting step has been determined to be the electron transfer prt>cess.(27_3l) What has not been performed is an in\Vestigation of a series of electrode reactions in which the reactant structure is varied in a controlled faSallion and the reaction pathway for simple electron t3|'-"ansfer is known in each case. This present work involves the study of a series <>ff onGE-electron reductions and oxidations at the polycrystal— Also, some of the thermodynamic simple transition metal reactants which undergo JLille silver electrode. alasI>ects of polycrystalline silver have been investigated 6‘3‘Ki measurements of simple anion and transition metal <2 ‘:’mnplex adsorption at this surface have been made. The t-‘V‘ A! ‘1 ..o. o I' I n..-no-- \ — unfit-vi o Ivan -A~ l..I-U‘ ' .. \ P . . ..0 Inuu —_ . \ ' no II" n... .-'~ v:._‘_ 9...: . o“ . ."‘t Io. 0' - y. 'n' "! n. ‘r-. .‘s. Q "t ‘~ study has been designed so that the resulting information bears directly on basic questions in electron transfer processes. Single crystal electrodes of silver were not studied because of the additional complications in producing and maintaining such surfaces. A. The Electrode Material Silver was selected as an electrode material for a careful investigation of electron transfer kinetics for a variety of reasons. In aqueous non-complexing electro- lytes such as neutral NaClO4 and KPF6 there is a wide range of potentials over which silver remains ideally POIarized.(21) If the electrolyte is acidified (pH==2—3) the If'+ H2 reduction overpotential is sufficiently large to allow measurements out to approximately -650 mV versus SCE (saturated calomel electrode) with steady state tecfrniques. The positive potential limit in non- conlplexing media is near +250 mV versus SCE. More positive potentials cause the formation of silver oxides on the electrode surface. Silver is one of the important metals from the pclilrt of View of industrial applications. It is useful file a component of fuel cells and battery systems because c> . . :8 its ability to enhance the rate of many electron 't . . relnsfer reactions. This electron transfer rate enhance- tn. . . . . £3ht, or eZectrocataZySls, is a phenomenon which is a S . <3llght-after quality of electrode materials in many disciplines involving heterogeneous electron transfer reactions. The ability of a particular electrode material to increase the rate of a given reaction significantly may make a previously unusable or uneconomic process viable. The factors responsible for electrocatalysis at a metal electrode are of immediate interest to the electrode kineticist. Understanding the individual steps that lead to the facile exchange of an electron between a metal surface and a reactant in solution is a major goal of basic studies in electrode kinetics. Since silver shows electrocatalytic properties and has a tractable surface chemistry in aqueous solution it A solid 15 a promising candidate for investigation. metalelectrode surface must meet certain requirements to 13e a useful investigative tool. The surface should be free of significant quantities of contaminants, reproducible in activity, and stable with time. For these reasons a suitable means of preparing, mechani- caJle, Chemically or electrochemically pretreating, a metalsurface is imperative. The pretreatment procedure should produce a surface that is as smooth as possible arua free of significant crystal dislocation and strain. Mechanical polishing may be a feasible method of prepar— ill‘SJ certain electrode materials. If the metal is Jreelatively soft, such as lead, and has a low tendency ‘:<3 adsorb contaminants mechanical polishing can produce a smooth, clean, and fairly stable surface. (32) For harder metals, with greater tendency to adsorb contaminants, silver, gold, and platinum being examples, mechanical polishing alone is generally unsatisfactory. The polishing procedure induces a significant change in the electrode material at its surface by producing an amorphous (reconstructed) layer of metal over the polycrystalline substrate. The overlayer may exhibit electrochemical properties at variance with the bulk material because of its different structure; there is also a large possibility that mechanical polishing will leave the surface poisoned by adsorbed species. Metals such as silver, gold, and platinum therefore require an additional pretreatment step which is generally chemical or electrochemical in nature. Methods for Chemically or electrochemically polishing silver which have been utilized in industry also serve in the prepara- tion of silver electrodes for kinetics studies.(33-35) Chemical or electrochemical polishing is a procedure in which the entire metal surface is etched at a uniform rate to yield a specularly reflective surface. For polycrystalline samples, such as the silver employed in this work, the different crystallites of the surface are not exposed by attack of the polishing agents at crystal grain boundaries. The latter more common phenomenon is Observed if one etches polycrystalline silver- iri dilute aqueous nitric acid. An electrode surface that is smooth on the order of the wavelength of visible light is useful in kinetics experiments and thermodynamic measurements of electrode capacitance. Surface roughness enters into the calculation of reaction rates since the real (not geometric) electrode area is contained in the relations yielding rate constants for electron transfer. It also affects the magnitude of electrode capacitance in thermodynamics studies. The overlayer of metal, generated in a prior mechanical polishing step, may contain significant crystal dislocationsanuistrain. Chemical or electrochemi- cal polishing removes this layer and exposes the clean bulk substrate. In the current work the polycrystalline silver electrode was electropolished in basic cyanide solutions containing Ag+.(36) Subsequent treatment, as described in the chapter on experimental procedures, removes CN- and Ag(CN);n-l) complexes which remain after this portion of the electrode preparation. Electrocatalytic activity in many cases results from the strong adsorption of the reactant on the electrode surface. Silver is known to adsorb many common anions and reactants incorporating these anions as ligands to a large extent.(22’37) It is also capable of strongly adsorbing many organic molecules from aqueous solution.(38) Determining the extent of adsorp- tion of these species (especially simple anions) is a current topic of investigation in several laboratories. That fact that polycrystalline silver does exhibit such significant adsorption of a variety of ions and molecules presents an experimental problem. Low levels of adsorbing contaminants in solution are capable of occlJniing a significant portion of the metal surface. Comnmnuly, trace levels of adsorbing anions are present in the commercial samples of metal salts employed as electrodytes. If these contaminants are present during the kinetics measurements for an electron transfer reaction the observed rate can be different from the rate measured at a "clean" silver surface. Even the composition of some ofthetransition metal reactants in this study can be altered if the contaminants are present in the inner coordination shell during the electron transfer step. Precautions in the selection and preparation of components of the electrochemical system are thus vital in order to obtain meaningful data. B. Reactants for Heterogeneous Charge Transfer Studies Having selected a suitable electrode material, the investigator has an equally important problem of choosing a set of reactants with which to probe the electrode characteristics. Since interest lies primarily in determining the rate of electron transfer for a given reactant the overall reaction rate should not contain Significant kinetic influences from other features of the reaction. Complex phenomena such as metal electro- dePOSition, which involves nucleation and crystal growth processes or oxide film formation in which the film propagation is dependent on ion transport through the oxide layer, are therefore excluded. Redox reactions of organic molecules are also avoided because of the prevalence of complex chemical reactions which are conumonly coupled with the electron transfer step. Numerous metal complexes are known to undergo single electron oxidation state changes without complications from other chemical processes.(8) Examples of these are Cr(III)/Cr(II), Fe(III)/Fe(II), Ru(III)/Ru(II), Co(III)/(II), Eu(III)/(II), and V(III)/(II). Many of these candidate probes have electron transfer rates which are too fast to measure by available electro— chemical techniques. In these cases the overall reaction rate is determined by diffusion of the reactant to the electrode or perhaps its rate of adsorption. This latter process will be discussed in a succeeding chapter. Complexes of Co(III) and Cr(III) undergo single electron reductions at measureable rates with a variety of inner coordination shell compositions. They are also reduced within a potential range that allow their investigation at the silver electrode surface. The 3a.}, '5 10 StatEB- 111 both metal ions the incoming electron is (8) An accepted into an eg antibonding orbital. illustration of the electronic energy levels for the III and II oxidation states of cobalt and chromium is given in Figure 1:1. For the III oxidations states the complexes are shown to be low spin, as is the case for all reactants studied. Of particular importance in the chemical behavior of truansition metal reactants for heterogeneous charge traxusfer studies is their substitutional lability. The Co(III) and Cr(III) reactants were discussed above as possibilities for study prior to Fe(III)/(II), Eu(III)/ (II), or V(III)/(II) because of their relative substitu- tional labilities. (The positive formal potential of Fe(III)/(II) (ca. 0.5 V versus SCE) precludes the study of many iron complexes at silver.) Co(III) and Cr(III) complexes containing aquo, ammine, and simple anionic ligands are substitutionally inert on the time scale of the electrochemical measurements employed in this work. The inner coordination shell of these reactants retains its identity regardless of changes in electrolyte composition or the presence of contaminants in solution or adsorbed on the electrode surface. With the structure of the reactant independent of its environment, the species undergoing electron transfer is known. Iron, europium, or vanadium in the +3 oxidation state have sufficiently 11 Figure 1:1. Electronic states for Co(III)/(II) and Cr(III)/(II) couples. 12 Co(m);d° _ — egla _e'_> +9 H H tzgm o= antibonding n= nonbonding Cr(IJIN3 _ _ egci Cr(II):d4 ”fife inner . fl... , . .“ggflIO'h* ‘ .~_-.a.'-\".‘ - '..-..un-AH ' ‘\§ ' ,.:vivvfig“ ' um; n;;v~ ~~' Ju- LV-vto U. . II'QIl-:. flfifl P Ioovoo-h vantvt O. - .‘0 ‘Fnar v-uh u..—d\,. Q . ., . 'vlq..;= "_ .oui.'.~ .. I. my“: A: .‘v“'. V. .-.' .1 a flag 1 Its...» eie‘ IA'."" (‘5' " “n! V ‘ o . I.- an~ e, ‘ _. huh: a“ e": - NA VI |~ u. a Y- ~~3J3 C . G A ‘~ .itt til" \ . Nu - In; A; ..-.u. V‘ 0‘ c .Ku.‘~ n,‘ ‘hly'. ‘ U ‘ J fi‘f 5v ' §~~~O ' LIQYL . l3 labile inner coordination shells to make their exact composition ill-defined at the moment of electron transfer. Discussion of the labilities of various complex ions (8’9) Under certain have been given by several authors. circumstances it is possible to scavenge the electrode surface of contaminants with an appropriate reactant which does not interact with the labile oxidant or reduc- tant under study.(39) An example of this will be given in the chapter on outer-sphere electron transfer. Upon reduction of Co(III) or Cr(III) the resulting products in the (II) oxidation state rapidly hydrolyze because of their high substitutional lability. In acidic electrolytes the hexaquo complex ions are formed. 3+/2+ 6 has an extremely positive formal potential (+1.65 V, Worthy of note is the fact that the Co(OHZ) couple versus SCE)(40) and does not undergo reoxidation under the conditions of the experiments. This allows measure- ment of only the forward reaction rate (reduction) without complications from a competing (oxidation) back-reaction. 3+/2+ 6 and can present problems with The formal potential for Cr(OH) (41) is much more negative (—0.655 V versus SCE) back-reactions when the reduction process is strongly catalyzed (moved in the positive direction with respect to electrode potential) at surfaces such as silver. Electrochemical rate measurements that employ large perturbation techniques can reduce or eliminate this problem. .u-vUOVH . ~ .A Q ‘3‘ fl- D \U o ‘ -.-~v‘ :LL ' . "fl: AYIHJFO: va'u vuo-‘H‘IVV — fl ..v-v hula- v 04:”.“F‘ O: . ol~~o‘f\fi RF." A . .t-vvovoo b¥u ‘n- .A .3:- fl: ' l h vnAiob)uvb L1- . “#05:... z.- .Iinv‘vcougdos, ‘ E."‘"“v A " bu...*:.s H .- r V‘F .A A ‘ ‘iu.-S-‘:‘ K .t. ”A ‘ l ."h ‘Pfih H! ..\,\'.. Ana ‘ bah .A ‘ ‘- . :n ~yu‘“l s?- ‘ Out I .A I .J“ G I- "-~.. 6 I V L a: . . v 5 ‘ s bue‘r C o 'v fi‘ . this ca ‘A '. t. 539 e“ a J. i~ ‘ l.\Y‘ ‘ “theA M a“ a? “ta." '1 Sv- \E e 14 Cobalt (III) has been intensively investigated by researchers studying homogeneous kinetics for many years.(8-lo) Complexes of Co(III) are relatively good oxidants and it is therefore easy to select a reductant of sufficient driving force to make a redox reaction occur at a conveniently measureable rate. Complexes of Cr(III), having more negative formal potentials, are less well characterized in homogeneous studies. C. Pathways for Homogeneous and Heterogeneous Electron Transfer Investigators have deduced that two mechanisms are primarily responsible for homogeneous electron transfer reactions between metal complexes in aqueous solution. The mechanisms have been termed "outer-sphere" and (8) "inner-sphere" electron transfer. In the former mechanism electron transfer occurs within the activated complex formed by the two co-reactants without disruption of their primary coordination shells. The equivalent process may occur for a heterogeneous reaction as well. In this case the layer of solvent molecules adjacent to the electrode surface ("coordinated" solvent) is not displaced by the reactant at its plane of closest approach. Interactions between the reactant and electrode surface are expected to be primarily electrostatic in nature Q -~“"' . IF ' v C“... . u on" . _1 A O 0 F luv :1 OFY Iain-'6' .b‘ a x‘ 1 Q ~‘nnn “hr.“ ”oh" VU-IHA o I A Q - ;‘.Vfllna ....'.UU5 .v‘.- .A *.:“: a? I ) v 0‘ (‘f ,. (D ’1 Ft. n~: :"I‘ u "n,”Y ~ ‘A « '* at t 1 u... l . ‘ ADI a s‘ ' no.» . u . 15 (the equivalent is anticipated for homogeneous reactions as well).(8'9) Inner-sphere electron transfer occurs when a ligand is simultaneously bound to both reactants in the activated complex for electron transfer. Here the ligand "bridge" which couples the two reactant centers together may allow a significant (mutual) perturbation of the energy levels of the participants. Charge tranfer via an inner-sphere pathway at an electrode surface is predicted to be sensitive to the chemical properties of the electrode metal. An illustration of the two reaction pathways for both homogeneous and heterogeneous charge transfer is given in Figure 1:2. It is evident that a knowledge of the labilities of the reactants in electron transfer would allow one to decide whether or not an inner-sphere pathway could occur. If the coordination shells of both reactants are substitutionally inert on the time scale of the experiment, electron transfer must occur by an outer- sphere process (except when coordination numbers are expanded). This is the case in Figure 1:2A. Both Co(III) and Ru(II) hexammine have slow ligand exchange; the redox mechanism is thus constrained to be outer- sphere. If Co(NH3)g+ or Ru(NH3)§+ were placed at an electrode surface the case for outer-sphere electron transfer would be less certain. The electrode surface is substitutionally labile, allowing either mode of electron .I v. ‘M My ,4 I. *IH‘Y. 16 Figure 1:2A. Homogeneous Outer-Sphere Electron Transfer. 3+ 6 3+ 2+ 2+ Co(NH3)6 + Ru(NH3)6 + Co(NH3)6 + Ru(NH3) 2+ H+ CO‘NH3)6 fast ’ 2+ + Co(OH2)6 + 6NH4 Figure 1:28. Heterogeneous Outer-Sphere Electron Transfer. 3+ - 2+ Co(NH3)6 + i ———+ Co(NH3)6 2+ H 2+ + 3)6 f3§E’*C°(OH2)6 + 6NH Co(NH 4 17 ® metal solvent .NH3 ion molecule ligand Co(l l I) metal 18 Figure 1:2C. Homogeneous Inner-Sphere Electron Transfer. COIII(NH3)5C12+ + Cr2+ ———+ COII(NH ) 4+ aq 3 5 III ClCr (0H2)5 III 4+ H+ 2+ 2+ ClCr (0H2)5 -—-———+Co(OH2)6 + Cr(OH2)5Cl II CO (NH fast 3)5 + + SNH4 Figure 1:2D. Heterogeneous Inner-Sphere Electron Transfer. 2+ - + Co(NH3)5Cl + e ———+ Co(NH3)5Cl Co(NH ) C1 + _§:_—+Co(OH )2+ + SNH + + Cl- 3 5 fast 2 6 4 l9 ® metal solvent 3 NH3 IOI‘I molecule ligand Co(lll) metal 20 transfer to take place. An analogous case in homogeneous redox reactions is the reduction of Co(NH3)2+ by Cr:;. Chromium (II) is very labile (kex (5—1) s 109)#9) yet it is known that the reaction proceeds by an outer-sphere process. When Cr(II) is oxidized to Cr(III) the ligands present in its coordination shell at the moment of electron transfer become "locked" in place since Cr(III) -l 5 is substitutionally inert (kex (s ) = 2 x 10- for Cr(OH2)2+).(9) The Co(II) product releases its ligands -l 6 rapidly (ke (s ) = 1 x 10‘ for Co(OH2)§+).(9) If x the homogeneous reaction were to proceed via an inner- sphere mechanism (Figure 1:3A) then an ammine ligand should be incorporated in the Cr(III) product. Instead, Cr(OH2)2+ is formed and the reaction appears to proceed as in Figure 1:3B. The point that the reactant configura- tion may actually be more like that of Figure 1:3A and the mechanism remain outer-sphere needs to be made. What is important in this particular case is the coordinative saturation of the ligands bound to Co(NH3)z+. Though Cr(II) oxidation has been found to proceed by an (8) inner-sphere mechanism whenever possible, the absence of coordination sites on Co(NH3)z+ forces the reaction to be outer-sphere. A similar argument can be developed for the reduction of Co(NH3)2+ at an electrode surface. The electrode surface is unable to bind to the reactant in the transition state and the reaction occurs by an outer-sphere route. It is likely that the layer of 21 Figure 1:3A. Possible Inner—Sphere Pathway for Homogeneous Electron Transfer Involving a Labile Reactant. 3+ 2+ II III 5+ Co(NH3)6 + Craq ———+ Co (NH3)6Cr (0H2)5 II III 5+ H+ 2+ 3+ Co (NH3)6Cr (0H2)5 fast—9 Co(OH2)6 + SE<9§2)§§§§ + + SNH4 Figure 1:3B. Homogeneous Outer-Sphere Electron Transfer Involving a Labile Reactant. 3+ 2+ 2+ 3+ Co(NH3)6 + Craq -——> Co(NH3)6 + Cr(OHZ)6 2+ H+ 2+ + Co(NH3)6 EEEE—+CO(OH2)6 + 6NH4 22 metal 0H2 e NH ® ion a ligand ligagd e Co(l l I) k \ Cr(l I) 23 solvent at the electrode surface is not displaced either. Work with the reduction of these and other metal complexes at mercury electrode lends credence to this idea. (13) Silver is anticipated to have even greater solvent- surface interactions which make the solvent-separated outer-sphere path probable (Figure 1:2B) . (42) From the preceding discussion a better set of constraints for an inner-sphere reaction can be defined. If one or both of the co-reactants are substitutionally labile (including an electrode surface) and a coordina- tively unsaturated ligand is available to bind the reactants in the activated complex for electron transfer, an inner-sphere pathway is possible. This is not to say the inner—sphere pathway is necessarily preferred energetically over the outer-sphere route, but only that it is available. Figures 1:2C and 1:2D illustrate respectively homogeneous and heterogeneous inner-sphere electron transfer. Note that in the homogeneous reaction the chloride ligand bridge (initially bound to Co(III)) be(Zomes incorporated in the Cr(III) product, proving the inner-sphere mechanism to be operative. The same 2+ electron transfer pathway for Co(NH3)5Cl reduction at an electrode surface must be inferred from the rate reSpouses to system changes such as the introduction of sDecifically adsorbed anions. (43) Criteria for distin- g\Ilishing reaction pathways at the silver electrode surface 24 will be given in the chapters on outer- and inner—sphere electron transfer . D. Surface Enhanced Raman Spectroscopy Applied to the Electrode-Electrolyte Interface The discovery of a Raman signal enhancement(44’45) for adsorbates on a roughened silver electrode surface has prompted interest in the chemical and physical PrOperties of the silver-electrolyte interface. (46-48) Surface enhanced Raman spectroscopy (SERS) has developed initially with emphasis on the phenomena responsible for the signal enhancement. (49) Investigators have been able to observe SERS signals for simple anions adsorbed on silver including Cl-, NCS-, and CN-.(50-53) Also, SERS has been applied to a cobalt phthalocyanine complex on the silver electrode surface. (54) Although the SERS Signals for the simple anions are weak compared to organic adsorbates such as pyridine, the information ac(quired on the metal-ligand bond is of special interest to electrode kineticists. The _i_n_ situ capability of SERS could allow the direct acquisition of information on metal-ligand interaction energies for the surface- ad sorbate and metal—ligand bond in adsorbed metal coltlplexes. The opportunity to follow the influence of a metal electrode on a reactive complex may clarify the factors contributing to inner-sphere reaction catalysis. 25 We have studied some simple heterocyclic compounds such as pyrizine and 4,4'-bipyridine and the ruthenium (II) ammine complexes of these ligands with SERS as a .first step in deciphering the particulars of electrode- audsorbate interactions. These compounds were chosen kxecause of their large SERS scattering cross-reactions (teasing the requirements on signal detection capabilities), tlieir tendency to adsorb on silver, and their importance iri electron transfer studies.(55) Chapter VII contains 'trie results of these experiments and discussion pertinent to the topic. CHAPTER II Experimental Methods The material within this chapter includes descrip- tjxons of the reagents and apparatus employed in the majority of the electrochemical measurements. Additional- 137, a discussion of the various electrochemical techniques and data analysis applied to the thesis research is; included. A brief review of measurement errors in tflaee electrochemical rate studies is given at the end of tfiis chapter. A - Materials 1. Water Puriffiation Water employed in experiments with mercury electrodes was obtained from laboratory distilled water which had been passed through a "Milli-Q" purification unit (Millipore Corporation). This consists of a set of ion exchange units and an activated charcoal filter. For experiments with solid metal electrodes a more thorough approach to water cleaning was required. The water from the Milli-Q system is in continuous contact Vvith plastic, and the feed water (house distilled) may 1lave steam-distillable anti-corrosion agents in it. As 26 27 noted in a paper by Bangham and Hill, contact of water (56) with most plastics leaches out organic materials. Organic chemicals may adsorb significantly at metal electrodes and affect the rates of electron transfer reactions at these surfaces. To reduce this source of contamination tap water was first distilled from alkaline (57) permanganate and then placed in a pyrodistillation unit. The unit recirculates steam and oxygen through a multi- pass silica tube heated to 750°C. This process oxidizes any carbonaceous materials which remain after distillation from Mn04- to C02(g) and water. Experiments with solid metals and mercury indicated that water prepared by distillation of tap water from alkaline permanganate and Subsequent distillation from an all-quartz non—boiling Still (Dida-Sciences Inc., Montreal, Quebec) was of egual or better quality. The non-boiling still has the advantage of reducing ionic impurities since no Spray entrainment of ions occurs as in boiling stills. The criteria applied in judging the purity of the Wat er included the reduction of protons on mercury electrodes and the time stability of electrode capaci- ta~I‘ice measurements. In the former experiment DC p0larography was performed in solutions of perchloric ac id, and the shape of the foot of the polarographic ways noted. In instances where solution contamination 0Qcurred, this region of the wave was distorted from the expected Tafel relationship. Electrode capacitance 28 measurements with hanging mercury drop and polycrystalline silver electrodes were very sensitive to solution cleanliness; the apparent capacitance drifted signifi— cantly with time when the water was not of requisite purity. Conway et al have also used H+ reduction and surface oxide from formation on platinum as indicators of water purity.(57) 2. Oxygen Removal Solutions were purged of dissolved oxygen by bubbling prepurified nitrogen through them. Residual traces of oxygen were removed from the nitrogen by either a trap Containing V(II) or a catalytic column. The latter method was preferred for work with solid metals since it does nOt aspirate contaminants into the electrolyte solutions as the V(II) traps might. The columns (5 x 28 cm, ~ 500 cm3) were packed with BASF R3-ll catalyst (Chemical DYllamics Corp., South Plainfield, NJ) and heated to "' 140°C. A trap containing purified water was placed a1"31:er the column to humidify the gas stream prior to entry into the electrochemical cell. 3. Electrolytes All electrolytes employed in the experiments were recrystallized in pyrodistilled water or water from the c3‘1artz non-boiling still. Sodium perchlorate was prepared fli‘om 60% perchloric acid and sodium carbonate. Stock solutions of HPF6 were made by adding concentrated HClO4 t-o a saturated solution of KPF6 (~0.5 M), cooling to 29 0°C and filtering off the Kc104 precipitate. The HPF6 solutions were kept frozen to retard acid-catalyzed hydrolysis of PF6 .(58) Hydrolysis of PFG- is also a problem during the recrystallization of KPF6 from warm The time the solution is held at saturated solutions . Sodium elevated temperatures should be kept to a minimum. fluoride employed in some of the capacitance measurements was calcined at 500°C in platinum crucibles. 4. Transition Metal Reactants a. Co(III) Complexes 3+ (59) 2+ (59) Co(NH3)5F , Co(NH3)50H2 , a d Co (NH3)SOSO3+ (60) were prepared from [Co (NH3)5CO3]NO3 ”SI-12069) starting material. The remainder of the C0 (III) complexes were synthesized by oxidation of COCIZ. These are Co(NH3)2+, (61) Co(NH3)5N032+, (62) 2+ (64) C33 (NH3)5NCSZ+,(63) Co(NH3)5N3 , cis- and tieezns-Co(en)2(N3)2+"65) and c-Co(en)2(NCS)2+.(66) The perchlorate salts were utilized in most experiments. IrIstances which required the fluoride salt were handled by passing a solution of the complex through an anion- e‘Kehange column (Bio-Rad AGl-XZ, 100-200 mesh) which was in the fluoride form. b. Chromium (III) Complexes Cr(III) (NH3)5Xn+ complexes were prepared from ElCIuopentammine-chromium(III) ammonium nitrate.(67) These are Cr(NH3)SBr2+,(67) Cr(NH3)5C12+,(67) Cr(NH3)5NCSZ+,(68) 2+ (69) NC82+,(70) Cr(OH2)5N32+,(7l) 30 and Cr(OH2)5NO§+ (72) were prepared as given in the references. The resulting solutions were separated on a cation exchange column (Bio-Rad AG SOW-XB) in the sodium form (n s 0.05 NaClO4 + 2-5 mM HC104). The column was cooled to 5°C for the Cr(OH2)5NO§+ complex to slow possible hydrolysis. After the reagent solutions were placed on the column the sample was washed with dilute solutions of acidified sodium perchlorate (ionic strength u z 0.05 NaClO4 + 2-5 mM HClO4) to remove counter anions such as NCS- and N3_ used in the synthesis. This was done until the free anion was no longer detectable in the eluent. The Column was next washed with 11 N 0.2 NaClO4 to elute higher complexes (CklOH2)6_nXé3-np)+). The desired pentaquo species were eeluted with acidified 1.3 M NaClO4 and frozen to retard 113fldrolysis. Cr(OH2)5NO§+ was kept in liquid nitrogen 1363cause of its sensitivity to hydrolysis. 080; ([3) was separated from hydrated chromic Sulfate with ion exchange as described above. Cr(OH2)5 The eJl—tzlent was acidified 0.2 M NaClO4. Cr(OH2)5Cl2+ was £3Efrithesized by electrolyzing a solution of 50 mM Cr (0112):+ + 5 mM Hc104 + 100 mM NaClO4 (0.3 M total (:JLCD4-) under nitrogen at a stirred mercury pool electrode 6“: -1100 mV versus SCE. (The Cr(OH2)g+ was prepared by 1reducing CrO3 in perchloric acid with excess H202. The aCid concentration in the stock solutions was kept at aim-I. 31 0.5-1.0 M to retard the formation of Cr(III) polymers. After the CrO3 has been reduced the solution should be gently boiled for approximately two hours to destroy the 200 mM NaCl(s) was added and the remaining H202.) Next, Crzg solution oxidized at -200 mV versus SCE. was added after the reduction step since it can catalyze The resulting solution was The chloride proton reduction on mercury. separated by ion exchange with a cooled (5°C) column as described earlier. Cr(OH2)5Br2+ was prepared in an analogous manner. To generate specific isomers of chromium aquo isothiocyanato complexes an electrochemical method of synthesis was employed. (74) c-Cr(OH2)4(NCS)2+ and f--Cr(OH2)3(NCS)3 were made in this manner. Ligand txridged oxidation with NCS- adsorbed on the electrode surface causes the cis- and fac-complexes to be Siennerated preferentially to the corresponding trans- and mer- isomers. Approximately 50 mM Crag solution ‘Véiss generated at a mercury pool electrode and then Oxidized in the presence of 150 mM SCN-. The resulting (NCS)r(l3-n)+ products were separated by ion The column packing (Bio-Rad AG50W-X8) was The (:1: (032)6-n Eal'tczhange. generally 8-10 cm in length by 1.5-2 cm diameter. :L5311 exchange resin was initially in sodium form ((3104- counter ion) and the aqueous sodium perchlorate eluent at an ionic strength of 0.05. The concentrations .u-nl' i 32 of the solutions of chromium complexes were determined spectrophotometrically(75) by oxidation of Cr(III) to chromate by alkaline peroxide. Other complexes of Cr(III) were kindly donated by the laboratory of Dr. F.C. Anson. References to their synthesis follow; c-Cr(NH3)4(NCS);'(76) may be prepared ( ) 2+ from Cr(NH3)4 oxalate via Cr(NH3)4ClOH2 . c-Cr(en)2Cl; (78) serves as the precursor of c-Cr(en)2(NCS);.(76) Robinson and Bailar give a method (78) for t-Cr(en)2(NCS); synthesis. c. Ruthenium Complexes Ru(NH3)5NCS2+ was synthesized as detailed by Lim, (80) Ru(OH2)g+ solutions were made lxy'electrolyzing a 4 mM solution of RuCl3 (Alfa Barclay, and Anson. Idnorganics) in 0.18 M KPF6 + 0.02 M HPF6 at a stirred Huarcury pool at -600 mV versus SCE under argon (ruthenium ~i£3 capable of forming stable nitrogen complexes which necessitates the use of argon). Next, 20 mM AgPF6 Prepared from A920 and HPF6 were added to precipitate chloride ions from the electrolysis. The solution was cc><>1ed, filtered and electrolyzed at 50 mV versus SCE on E1 Dnercury pool electrode to electrodeposit Ag+ and i; to Ruzg. The resulting solution was stored 3‘ ‘1 liquid nitrogen . OX idize Ru Ru(II)(NH3)5 pyrazine2+,(81) Ru(II)(NH3)54,4'- +'(82) 'hhipyridine,(81) and c-Ru(III)(NH3)4Cl2 were already cInhand. Ru(II)(NH3)5 pyridine2+ was prepared according 33 (83) to Taube and co-workers. Ru(II)(NH trans-1,2- 3)5 bis(4-pyridy1)ethene was made according to reference 81. d. Miscellaneous Reactants Eu3+ was prepared by dissolving Eu203 in an excess aq of either HClO or HPF . Bug; could be generated by 4 6 g; at a stirred mercury pool electrode at ~1100 mV versus SCE. V2; was made by oxidizing Vi; at -300 mV versus SCE at a mercury pool electrode which reducing Eu had been prepared by the electroreduction of V(V) 2+ 2 U02(C104)-6H20 (G.F. Smith). KBrO4 (ROC/RIC, Belleville, (from V205) at -1100 mV. UO was obtainedtnrdissolving NJ) was employed as supplied. 5. Electrodes Polycrystalline silver electrodes were fabricated from zone-refined (99.999%) rods (Materials Research Corp.) as rotating disk electrodes. The disk radius was 0.20 cm surrounded by a sheath of Teflon or Kel-F with a 0.6 cm radius. The sheaths were bored 8-10 thousandths inch undersize to ensure a tight seal with the electrode. Assembly was preformed by heating the fluorocarbon plastic to 130-140°C and then pressing the sheath over the electrode and stainless steel shaft. The RDE design for studies involving temperature changes was modified to thermally insulate the electrode from the steel shaft. A tubular piece of Delrin was employed for this and a copper spring used to make contact between the shaft and 34 electrode mount. Electrically conductive silver-filled epoxy (Transene Company, Inc., Rowley, MA) was employed to bond the unit together. Figure 2:1 illustrates the materials and dimensions for the insulated electrode. The electrode pretreatment for kinetics and capacitance studies was initially similar to that (84) and later modified employed by Valette and Hamelin to improve the removal of adsorbed cyanide. The electrode was polished with a 0.3 pm alumina and water slurry on a polishing wheel (Buehler 44-1502-160). The electrode was then immersed in an argentocyanide electropolishing 1 NaCN, 44 g 1'1 AgNO 38 g l’1 K solution (41 g 1 3, 2 CO3) for two minutes while potentiostated at +0.25 V versus SCE. Following this the electrode was rinsed with water and immersed in 2 M HClO4 for 10-15 minutes. (This is the modification of the procedure given in reference 84 (x-l) which aids the removal of adsorbed CN- and Ag(CN)x complexes.) Finally the electrode was rinsed with water, placed in 0.1 M NaF while potentiostated at -0.7 V versus SCE, and then polarized at -1.7 V versus SCE for two one-minute periods with deaeration between cycles. The last step, which evolves H2(g) from the electrode surface, helps strip off remaining contaminants. Figures 2:2A and B show sensitive scale cyclic voltammograms in 0.1 M NaF for the pretreatment with and without immersion in 2 M HClO4 respectively. The peaks seen on the double- T Figure 2:1. 35 Insulated Shaft Rotating Disk Electrode. Materials list: a) b) C) d) e) f) stainless steel Delrin (filled nylon) copper spring Kel-F (bored 8-10 thousandths inch undersize for electrode seal) brass silver (silver-soldered to brass holder) :v-Iv V— 0 am e _ U 4L 1‘ <\\\\\\\\\\\\\\\\\\\\\\\\\\ \\\\\\\\ o—-) —-9|‘v-|v|(-— 36 37 Figure 2:2A. Cyclic voltammogram for polycrystalline silver electrode (geometrical area = 0.049 cm2) at 0.1 V sec.1 in 0.1 M NaF following cyanide electropolishing. Figure 2:2B. Cyclic voltammogram as in Figure 2:2A but including treatment in 2 M HC104. 38 0.... mdu odd .m> > . m mo- v.0- . NO- ll A3 0.. .. 0.0- 0.0 0.0+ O. _- 0.0: 0.0 0.0... 0.2. v" wanna v" wanna 39 layer charging envelope (Figure 2:2A) are thought to be due to adsorbed Ag(CN);x-l) complexes. Surface enhanced Raman spectroscopy (SERS) was conducted with a polycrystalline silver electrode which had been electrochemically roughened. After an initial polishing with 0.3 uM alumina the electrode was placed in 0.1 M KCl at -200 mV versus SCE. The potential was switched to +100 mV versus SCE (forming an overlayer of AgCl) until 50 mC/cm2 of anodic charge had passed and then returned to -200 mV to reduce the AgCl layer. This produces a reflective, beige colored surface. In many instances the reagent under scrutiny with SERS was present in the KCl solution during the roughening procedure. This yields generally enhanced SERS signals over the equivalent process with KCl alone. Experiments with mercury electrodes utilized either a dropping electrode (DME) or a hanging mercury drop electrode (HMDE, Metrohm Model E410, Brinkman Instru- ments) depending on the electrochemical potential-time perturbation. 8. Apparatus 1. Electrochemical Cells Electrochemical cells were of a conventional two- component design fabricated from Pyrex with a glass frit of "fine" or "very fine" grade separating the working and reference compartments. Frit material of this 4O porosity prevents mixing of the solutions on the time scale of most experiments (up to 3—4 hours) and yet provides a sufficiently conductive path for proper potentiostat operation at common electrolyte concentra- tions (11 = 0.1 - l) . The cell volumes were kept inten- tionally small so that only 10 ml of solution in the working compartment was sufficient for RDE experiments. The range of rotation speeds employed (ca. 200-2000 r.p.m.) low enough that a cell diameter of 3 cm gave limiting Cells was Currents in agreement with the Levich equation. uSed for preparative electrolysis had an additional <3(31'f1353artment for the counter electrode separated from the working compartment by two frits to insure isolation of products formed in the separate sections. The cell employed in double layer capacitance measure- ments had a working compartment of approximately 30 ml and a large platinized gauze counter electrode. (85) The colInter electrode was cleaned in l M HClO4 and stored in pyrodistilled water before any experiments were conducted. The counter electrode was of sufficient size relative to the working electrode that it did not contribute signifi- Q aIltly to the overall cell capacitance (CL = 51— + CL where c Q T W = working electrode capacitance '1:- = total capacitance, Cw e [16 Cc = counter electrode capacitance. For CC >> Cw' Q :1:- =cw). Special care was exercised in cleaning cells (and D 1:l'ler glassware) employed in the experiments. Prior to 41 each experiment glassware was placed in either chromic acid or concentrated nitric/sulfuric acid (1:1) for a day or two. The glassware was then removed, rinsed with water distilled from alkaline permagnate and left to soak in water for at least two days with an intermedi- ate change of water. Electrochemical cells were soaked in water from the quartz non-boiling still. Glassware was placed in a drying oven just before its use. The cell used for SERS work was designed to minimize Contamination problems. The body is of Pyrex and Teflon with a quartz window. The cell is assembled with an external frame and fluorocarbon polymer (Viton) o-ring Sea ls. Figure 2:3 illustrates the components of the e="=IE>eriment in the SERS investigation. 2. Reference Electrodes Reference electrodes for electrochemical kinetics rueasurements were procured commercially (Sargent-Welch) . These were potassium saturated calomel electrodes (KSCE) . For measurements in solutions containing perchlorate a“11:3.on the saturated KCl fill solutions of the reference electrode was replaced by saturated NaCl. Potassium berrchlorate is relatively insoluble and precipitates in the liquid junction of a KSCE reference electrode when placed in perchlorate electrolytes. Sodium perchlo- 1te‘te is soluble and thus a NaSCE avoids the problem. Since the ionic mobilities of Na+ and C1- are not equal this QJ~ectrode develops larger liquid junction potentials 42 Ejgure 2:3. Arrangement of equipment for surface-enhanced Raman spectroscopy experiments. 43 M p~4m uuz¢oumomhumam zecitance model is assumed for the electrode-electrolyte Interface as illustrated in Figure 2:4A. When the variable resistance and capacitance of the bridge (Figure 2:4B) are set equal to the solution resistance and electrode double layer capacitance (for R1 = R2) 3‘ condition of signal null (signal minimum with noise <2 . . . . OI“ponenu is achieved. Capac1tance and re51stance are then read directly. By altering the ratio Of R1 to R2 he condition of null is achieved Wlth a corresponding 46 Figure 2:4A. Model of electrode-electrolyte interface for double layer capacitance measurements. R8 = solution resistance, Cd1 = double layer capacitance. Figure 2:4B. Capacitance bridge for double layer capaci- tance measurements. R1, R2, R, and C are variable elements of the bridge; Rs and Cd1 are as described in Figure 2:4A. 47 R5 Cdl A. in“. B. detector 48 ratio of bridge resistance and capacitance to electro- chemical cell resistance and capacitance. This allows measurement of electrode double-layer capacitances which are larger than commonly available precision capacitors. 2. 2g Polarography Polarography of irreversible electrodes reaction at mercury was treated by Koutecky analysis. (88) 3. Rotating Disk Voltammetrg Most kinetic data for diffusing irreversible reac- tions at polycrystalline silver were obtained by rotating disk voltammetry. The voltammetric waves were analyzed by calculating the reactant concentration outside the double layer, cs, from c = cbl (i1 - iVill . where Ch s 13 the bulk reactant concentration, i the current in the rising part of the wave and il the diffusion limited current. (89) The apparent rate constant kapp at an e:Lecztrode potential E was calculated from kapp = i/I’IFACS w'luere n is the number of electrons transferred, F is the Faraday, and A the geometrical electrode area. Er“£310ying rotation speeds from 100-2000 rpm allowed 4 l méasurement of ka in the 10- - 10- cm 3-1 range. The PP ana- lysis assumes a reaction order of unity. Reactants expected to adsorb strongly, such as Co(NH3)5NCSZ+, emillibited reaction orders less than unity. Reactants were determined by varying C3 by adjusting Cb or more col'lveniently the rotating speed to) and noting the change i . . . 11 Current at a fixed potential E in the r131ng part of 49 the wave. A plot of loglo i versus loglo cS yields the reaction order from its slope. Figure 2:5 illustrates this for an outer-sphere reactant (Co(NH3)5F2+) and an inner-sphere reactant known to strongly adsorb (c-Co(en)2 (NCS);) . 4. Normal Pulse Polarography Normal pulse polarography(9o) was employed with both ‘4 a DME and a silver RDE. Rate analysis of irreversible Electrode reactions was performed with the Parry-Oldham analysis. (90) With a pulse width of 50 ms (P.A.R. 174) the reactant concentration at the RDE surface could be restored to its value in bulk solution (concentration polarization occurs from the preceding pulse(9l)) with Wait times of 2-5 s and rotation speeds in the range Au‘ 2 5 0—1000 rpm even with irreversible reactions. The S‘ta‘lzed rotation speeds do not cause sufficient convec- tion on the time scale of the pulse to significantly a. f feet the measured currents . (92) 5. Cyclic Voltammetry Reactant systems which exhibit irreversible electron (93) transfer were analyzed as outlined by Galus. The r O O 0 6-":es of quaSi-reverSible electrode reactions were (94) 1:3: eated in the manner developed by Nicholson. The JteCiuction of Cr(III) reactants was in some cases ac<=c>mpanied by a significant anodic back reaction from the ligand assisted oxidation of the Cr(II) product. Measurements by steady-state techniques such as rotating 50 Figure 2:5. Log-log reaction order plot of current i(amps) versus reactant concentration just outside the diffuse layer cs (mol cm ) for the reduction of Co(NH3)5F2+ (circles) and c-Co(en)2(NCS)2+ (stars) on a polycrystalline silver electrode in 0.1 M NaClO4 + 5 mM HClO4 at 23°C. Slope of plot yields reaction order. For 2+ Co(NH3)5F slope = 0.94 and for c-Co(en)2(NCS)2+ slope = 0.57. 51 -4.0 log“, ‘5 / "10' cm _ _ mQEO \ _ opac— "5.0 '7.0 111 - --.1.!1\/1} {11111111111111.1111 ,11 52 disk voltammetry are obviated by this occurrence. Ligand assisted oxidation of Cr(II) yields a distribution of Cr(III) products and therefore the overall electrode reaction is not microscopically reversible, a requirement for the back reaction corrections to have a simple form. Cyclic voltammetry is useful in these instances because it is a perturbation technique which leads to a greater separation of cathodic and anodic kinetically controlled processes. By employing cyclic voltammetry at high sweep rates (5-50 V 3.1) it was generally possible to obtain rates of reduction of Cr(III) complexes without contribu- tiOns from the back reaction. The current—potential data c011 1d then be treated as a simple irreversible electrode reaction. 6. Methods for the Direct Determination of Reactant Adsorption and Kinetics in the Adsorbed State (95) Single step Chronocoulometry is a useful method for determining both reactant adsorption and the reaction (96) rate of the adsorbed species on mercury. An attempt to perform similar experiments with silver electrodes was unsuccessful. A discussion of the problems with this 1:QCBI'mique when applied to solid electrodes is given in chapter V. An alternative method, fast linear sweep voltammetry, did prove to be a convenient and reasonably accurate theans of obtaining kinetic and thermodynamic information on Co(III) and Cr(III) complexes adsorbed on silver. 53 By utilizing bulk reactant concentratons < approx. 100 uM the current due to diffusing reactant is small relative to the component of current from a monolayer of a one-electron transfer reagent. Sweep rates of 310-100 V s-1 aid in separating current arising from diffusing and adsorbed reactant. For a diffusing Species the peak current is proportional to the square root of the sweep rate, whereas the peak current for an ad sorbed reactant is linearly related to the sweep rate. A discussion of the current from adsorbed and diffusing Species is given in Chapter V on reactant adsorption me a surements . The fast linear sweep voltammograms were analyzed in the following manner. The i-t profiles were integrated graphically to determine the extent of adsorption, P (moles cm- ), for a given bulk concentration and initial potential. The current due to double-layer charging was es‘timated roughly from the region of potential where the complex is adsorbed but not reduced. Large changes in electrode capacitance seen upon introduction of the transition metal complex to the electrolyte prevented a e . . . . eIlérarate determination and correction for non-Faradaic e urrent. P The surface reactant concentration, (mole cm-z) , was determined at a series of potentials and the surface rate constant, kit (5-1) calculated using 1; . . he instantaneous current at each potential, E. 54 E -1 _ i ket (8 ) - FKTT (2.1) where i — (C mole-l), A = geometric electrode area (cmz). Since the instantaneous current (A), F = Faraday The analysis assumes first order rate behavior. reaction order for the elementary electron transfer step was not determined in the measurements, comparisons of (3.1) values were done at the same surface concen- ket tration P' (mole cm-Z) whenever possible. D. Errors of Measurement The quoted rate parameters in the thesis have been :found to have a precision primarily determined by the eelectrode surface state rather than instrumental factors. Kitbltage errors associated with potentiostatic control of the working electrode potential in steady-state €33lle by studying the electrode-electrolyte interface under conditions in which no charge transfer occurs from metal to solution or vice versa. That is, the electrode areaInains ideally polarized under the conditions of the E353<19eriment. For an electrode to remain ideally polarized liiri- an electrolyte, reactions which could cause charge 't:3=’Einsfer across the phase boundary must be either IthZearmodynamically or kinetically inaccessible. An example <:’15 a thermodynamic limitation would be the oxidation of 57 58 silver in a l M solution of sodium chloride with the electrode potential maintained at -0.5 V relative to an Ag/AgCl reference electrode. One may calculate the Ag + Ag+ + e- [Ag+] e 10.18 M equilibrium concentration of Ag+ and find it to be infinitesimal. Effectively, no dissolution of silver (with an accompanying electron transfer) occurs. In contrast, a reaction which is thermodynamically favorable is the reduction of protons (hydronium ions) to form hydrogen. Even though the free energy change for this reaction is large at the chosen potential the large activation barrier for the process makes H2(g) production (exceedingly slow (relative to the rate of solvent dipole (Drientation and ion migration within the electrode- eelectrolyte interface). Within a defined range of gE>cmential, bound by metal oxidation at the positive end and proton or water reduction at the negative end, ESLilver remains ideally polarized. In neutral non- <=1r the potential gradient is possible in the Gouy- 60 Figure 3:1. Model of an electrode-electrolyte interface. ¢m is the potential at the electrode surface and ¢s the potential in bulk solution. The solid line shows the decay of potential from ¢m to ¢S. The label OHP stands for outer Helmholtz plane and is the average center of charge for ions at their plane of closest approach to the electrode. 61 (DriP 7/). \ ,e interface. and ¢S the @96 600 0669\mW@ @100” ¢m.3 93333339333 996 «New X———> _O—0E shows the OHP stands center of proach to 62 Chapman treatment since one may assume the electrode surface to be planar and avoid the need for Spherical coordinates required in Debye-Hfickel theory. The derivation of Gouy-Chapman theory is available in standard texts . (ll ’ 12 ’ 97) A component of the Gouy-Chapman (and Debye-Hfickel) theory is the Boltzmann distribution law given below. c(x) _ ZF (x) 7%— - exp '[—%"r—'] ”-1) The terms of the equation are c(x) the concentration of an ion at a distance x from an electrode surface, c the b ion concentration in bulk solution, Z the coefficient c>:ff charge for an ion, F the Faraday, R the gas constant, '1? the absolute temperature and ¢(x) the potential at £1 distance x from an electrode surface relative to the petential in bulk solution. The Boltzmann equation shows 3r1<>tential in solution. The charge qS in solution is counterbalanced by the electrode charge gm so that qm = -q8 (3.4) kui.th ¢OHP = 0, q = 0 and the capacitor formed by the ionic é1<:rub1e layer is uncharged. This potential is labeled the potential of zero charge (pzc). By differentiating I3(anation (3.3) with respect to ¢OHP one obtains the ionic (1(311ble layer capacitance per unit area, C d1 (commonly expressed in uF cm-z) . m 26 K c Ze¢ c =__d_a___ _9_€_b)%cosh(__<fli) (3,5) RT 2kBT The value of C c D 9 % d1 18 sen81tive to both (Ch) I :E and ¢OHP' ¢OHP = 0 the function cosh(u) = l and we have an eImpression for the ionic double layer capacitance at the pzc. l' 65 (3.6) Figure 3:2 shows the concentration dependence of Cgic and the component of total capacitance due to the solvent (dipole or "inner layer", C.. The value for Ci may be eastimated by assuming a large (0.1 to l V) linear poten- 1:ial drop from the electrode to the OHP (~65A). Because c>f dielectric saturation, KE for water falls to a value of 5-10 within the inner layer; thus -2 i x E 10 uF cm (3.7) ‘nrljere xH is the thickness of the inner layer. The two <:=eapacitances, Cd1 and Ci’ are effectively in series «Eilud therefore the total capacitance CT is _1__ d1 0 C_1_ + CL (3.8) T 1 13$? studying the concentration dependence of CT "Eilue of Ci may be obtained from a calculation of (2:11. a At this point we would like to make measurements ‘31? the electrode capacitance CT' integrate with respect ‘t&> electrode potential and arrive at the electrode c=harge per unit area qm as a function potential. This VnDuld allow the calculation of ¢ at the OHP via Equation (3.3) and its application to rate measurements. During 66 Figure 3:2. Concentration dependence of the electrode pzc d1 charge. Also shown are the inner layer capacitances c920 1 double layer capacitance C at the potential of zero of silver and mercury at their respective potentials of zero charge. electrode ial of zero citanCeS 've POtentials m C/uFC 67 200 100 1O 50 ch] mM 100 500 68 the integration however, an arbitrary constant appears with qm and we do not have immediate access to ¢OHP' For the liquid mercury electrode it is possible to . m . . obtain q directly from surface ten51on measurements as a function of electrode potential. The Lippmann equation gives the relationship (37%;) = - 9,1- = -qm (3.9) T,ui,.... where y is the surface tension, ¢OHP the potential drop, Qm the total charge on the electrode, and A the electrode area. The term (3y/3¢OHP)T 11' ,... is approximately I 1 equivalent to (3y/3E)T u_,... where E is the potential I 1 of the electrode with respect to a reference electrode. Changes in E produce corresponding changes in ¢0HP° A O O I m maximum in surface tensron occurs when q = 0 (the pzc) and repulsive interactions among charges at the surface are gone . With solid metal electrodes such as silver, the measurement of surface tension is experimentally intrac— table. An alternate route to the determination of qm is needed. It is found that in measurements of electrode c=apacitance a pronounced minimum in CT appears as the electrolyte concentration is made small (~10 mM). The I“inimum is due to the effect of Cdl on the total capacitance CT and occurs near the pzc for the particu- lar electrode material in question. Figure 3:2 shows 69 that at sufficiently low values of cb the value of Cd1 becomes comparable to or less than Ci' the inner layer capacitance. Since the two capacitances are in series the smaller of the two values dominates, as given in Equation (3.8). The minimum in CT disappears as the potential is altered from the pzc since C is potential dl dependent and becomes larger according to Equation (3.5) as ¢OHP increases. Once the pzc is known the capacitance data can be integrated to generate a qm versus E curve if the assumption qm = O at the pzc is made. This last step requires some qualification because the double layer model assumed earlier is incomplete. In common electro- lytes such as KCl the chloride ion, which is weakly solvated compared with K+, can penetrate the solvent layer at the electrode surface and come into direct contact with the electrode surface as shown in Figure 3:3. In this case the anion is said to be specifically adsorbed. If this occurs an additional charge Q' (uC cm-z), due to the specifically adsorbed anion, is present at the electrode surface. When the potential of minimum capacitance is found in such an electrolyte, the charge seen by the ionic double layer is q' + qm = 0, not qm = O. In order to apply capaci- tance data to the generation of qm versus E curves we need to determine the pzc in an electrolyte for which q' = 0; i.e. one that does not specifically adsorb. 70 Figure 3:3. Illustration of anion specific adsorption at an electrode surface. Anions have penetrated the inner layer of solvent molecules and are in direct contact with the electrode surface. 72 Electrolytes containing anions such as F-, ClO4-, and PF6- generally show little specific adsorption near the pzc. The fluoride ion is fairly well hydrated and does not interact strongly with metal electrodes. Both C104- and PF6- have small charge-to-radius ratios and are relatively non-polarizable; it is thus unlikely that these ions will penetrate the inner layer of solvent. Most alkali metal cations are strongly solvated and exhibit no specific effect on the inner layer composi- tion. A point about the orientation of water dipoles should be made here. Even when qm = 0 at the pzc there is some residual ordering of solvent dipoles due to chemical interactions with the electrode metal. Therefore the potential of the electrode ¢m generally differs from the potential at the OHP. All that can be said is that ¢OHP = ¢s where ¢s is the potential of bulk solution. D. Measurement of Anion Adsorption on Silver Electrodes A segment of the work in electrode kinetics has dealt with the effect of specific anion adsorption on the rates of electrode reactions in aqueous solution.(100- 103) These studies have been conducted with mercury electrodes and have utilized quantitative information on anion adsorption obtained from electrode double layer studies. The presence of adsorbed charge q' on an electrode surface alters the diffuse layer potential v at the ' A ZEICU use a: V. R! AJ- E O x H U fiafi‘ will 9X m‘b'n dvfl‘e r «3 ~ u a ‘II. IEiCt V! i a meats 73 at the plane of reaction and consequently the rate of reaction (as will be discussed in Chapter IV). Silver adsorbs anions extensively compared with mercury at the same electrode potential; for this reason some dramatic effects on electrode kinetics may be observed. In the present study information on the extent of anion adsorption is of value for reasons other than the interpretation of the effect of the double layer on electrode kinetics. Many common anions, such as Cl-, Br-, NCS-, and N3- are incorporated as ligands in the Co(III) and Cr(III) complexes under study. These ligands may cause extensive adsorption of the reactant and a subsequent inner-sphere electron transfer pathway as described in Chapter I. An understand- ing of the ordering and quantity of adsorption of various anions on silver would aid in determining which reactants with anionic ligands would be most likely to react via an adsorbed intermediate. Several avenues to adsorption measurements for silver (104) have been followed and include radiotracer, electro- (24) optical, and double layer capacitance measure- (22'37) Of these methods double layer capacitance ments. investigations are the most straightforward from an experimental standpoint. With proper application they are capable of yielding an accurate catalog of adsorption for most common anions. It is possible to gain some wv-lfi A” nnnnn 74 understanding of the adsorption of anions on silver (and other solid metal electrodes) by simply varying the concen- tration of a solution containing the anion and measuring the electrode capacitance as a function of potential at each concentration. Strongly adsorbed anions tend to increase the electrode capacitance at a given potential relative to an equal concentration of non-adsorbing anion. The measured capacitance C is given by T 1 a ' + —— [1 - (ii—)1 (3.10) Cd1 aqm L}. CT Ci where Ci is the inner layer capacitance, Cd1 the diffuse layer capacitance, and (aq'/3qm) a coefficient that accounts for the variation of the adsorbed charge q' as the charge on the electrode metal qm is altered. Since we are concerned with specific anion adsorption, the coefficient (Bq'/3qm) decreases from zero as the free energy of adsorption increases (i.e. more negative charge q' accumulates as qm is made more positive, increasing the affinity of the anion for the electrode surface). When specific adsorption occurs the value of CT increases and approaches Ci since the affect of Cd1 becomes prOportionally smaller. A quantitative analysis of the adsorbed charge q' is complicated by the fact that the diffuse layer and the inner layer contribution to the total capacitance are varying simultaneously. elECtr 75 (105) A method developed independently by Hurwitz and Parsons(106) allows a quantitative determination of anion adsorption at a solid metal electrode. By utilizing a mixed electrolyte composed of a non-adsorbing anion and the adsorbing anion of interest one may perform measurements at a constant ionic strength while varying the concentration of the adsorbed anion. The contribu- tion of the diffuse layer to the total capacitance is held essentially constant in this manner. From a series of differential capacitance curves (i.e.capacitance as a function of electrode potential) a set of Aqm versus E curves can be generated by integration of CT(E). Note that only a Aqm is available since the integration generates an arbitrary constant as mentioned earlier. This is satisfactory since the calculation of a surface concentration F' (mole cm_2) of adsorbed anions by the Hurwitz-Parsons method requires only changes in qm. The method has been applied successfully to mercury electrodes where alternative means of arriving at P' are possible.(100) E. Double Layer Properties of Silver 1. Experiments Performed with Polycrystalline Silver The initial double layer capacitance measurements of the polycrystalline silver electrode in the thesis work were performed in sodium fluoride solutions. The electrode surface was prepared in a manner analogous to .1“ A.» A V 76 (84) that employed by Valette and Hamelin. Data were obtained at ionic strengths u of 0.01, 0.021, 0.100, and 0.5 M NaF. At an ionic strength of u < 0.1 a distinct minimum in the CT - E curves could be observed near the pzc for the polycrystalline surface. Figure 3:4 shows some of the results of the measurements. The minimum is due to the effect of the diffuse double layer capacitance Cd1 on the total capacitance C Similar T' CT - E curves for polycrystalline silver in aqueous sodium fluoride have been observed in prior studies. 108) (107, Valette found the pzc of polycrystalline silver (-0.97 V versus SCE) to be essentially independent of ionic strength for NaF solutions and concluded that specific adsorption of F- does not occur near the pzc.(107) Invariance of the potential of minimum capacitance at the pzc is expected for measurements (11) In the conducted with non-adsorbing electrolytes. current study the value of the pzc for polycrystalline silver was found to be near -0.9 V versus SCE, but there was considerable scatter in the results. °This is thought to reflect some contaminant of the electrolyte capable of adsorbing on silver, a common problem in solid electrode studies. Despite this, the CT - E curves are similar in feature to the results of others and Show the surface to be reasonably clean. The greatest deviations of CT from the literature values occur at 77 Figure 3:4. Potential dependence of the capacitance of a polycrystalline silver electrode in 21 mM (squares) and 100 mM (circles) aqueous NaF solutions. T = 23 i 1°C. Capacitance values were calculated with the geometric electrode area. ‘S/Ad:cd72 78 50(— at m. «to “.30 OR -1.2 -o.3 E / V vs. SCE -O.4 '1 cur? writ 79 potentials positive of the pzc where the adsorption of trace levels of contaminants such as Cl- is most strong. 2. Estimation of the Surface Roughness Factor The surface roughness factor, rs, is defined as the ratio of real surface area to the apparent or geometric surface area. Though the silver electrodes are prepared as carefully as possible, small scale irregularities remain and cause a deviation from a perfectly smooth surface. When one measures an electrode capacitance the real surface area is required to calculate the speci- fic capacitance C (uF cm—z). Normally the electrode T capacitance is divided by the geometric electrode area to obtain an apparent capacitance C and a roughness A! factor is employed to arrive at C Knowledge of the T. specific capacitance is important, not only for compari- son with other work, but also in the generation of qm - E curves by integration. Using Equation (3.8) one may write 1 l l l —= =-—+—— (3.11) CT (CA rs) Ci Cd1 and l l l __.= + (3.12) CA rsci rstl For a homogeneous electrode surface (i.e. a liquid or single crystal specimen) in a non-adsorbing electrolyte a plot of CA-1 versus Cdl-l at the pzc yields rS from 80 (109) the inverse of the slope. Values of C at the pzc d1 for various electrolyte concentrations may be calculated from Equation (3.6). This method is not applicable to polycrystalline surfaces for which the pzc of different crystal faces varies. At the polycrystalline pzc the average surface charge qm = 0, but the values of qm for separate faces are not zero except by coincidence. This makes Cd larger than predicted by Equation (3.6). 1 An alternative approach is required. At potentials far negative of the pzc, the measured electrode capacitance is determined by Ci' the inner layer capacitance. At these potentials C is large and there d1 is little if any specific anion adsorption since qm is large and negative. By comparing the value of CA at negative potentials for polycrystalline silver with values obtained for single crystal surfaces of known roughness the value of rS for the polycrystalline surface (84) For the polycrystalline silver may be estimated. electrodes employed in the thesis studies rS E 1.2. 3. Measurements with Other Weakly Adsorbing Electrolytes The investigations of electrode kinetics at poly- crystalline silver in this work have generally been performed in acidified perchlorate or hexafluorophosphate electrolytes. Subsequent capacitance measurements of silver employed these (neutral) electrolytes to gain 81 some understanding of their tendency to adsorb on silver. Soviet investigators and other have demonstrated the pzc of silver to be unaffected by variations of pH.(110’111) The differential capacitance curves obtained with polycrystalline silver in these electrolytes are similar in feature to those observed in NaF solution. Assuming qm = 0 at the capacitance minimum in both electrolytes the capacitance curves can be integrated with respect to potential to generate the qm - E curves shown in Figure 3:5. The values of qm at potentials positive of the pzc in KPF (u = 0.1) are somewhat smaller than 6 the corresponding values of qm in NaClO4 (u = 0.1). This indicates, and is supported by kinetics data presen- 6 than C104- at positive electrode charge. Since ted in Chapter IV, that PF is less strongly adsorbed q' + qm is more negative for perchlorate than hexafluoro- phosphate electrolytes, the response of outer-sphere reactants is different in the two media. A recent study of the differential capacitance of a single crystal silver electrode in fluoride, perchlorate, and hexafluorophosphate electrolytes has shown that Specific anions adsorption occurs to a limited extent at the pzc and follows the order F- > c104' > PEG-.(Zl) This ordering is also confirmed for a polycrystalline silver electrode for fluoride and perchlorate solu- (110) tions. Evidence for specific anion adsorption at the pzc is taken from the shift of the potential of 82 Figure 3:5. Plot of electrode charge qm versus electrode potential E for a polycrystalline silver electrode in 0.1 M aqueous solutions of KPF6 (solid circles) and NaClO4 (open circles). T = 23 : 1°C. Values of qm were calculated by employing a roughness factor rs = 1.2. 83 so — 20 — 1o — N 'E 0 \ _. E o , ~10 _ ’20 _ 0 Q \\ \‘ \‘ \‘ pzc / lllllll 0.4 0.8 1.2 - E / V vs. SCE L... rt- ”1 fl) “‘ Gaga inte 84 minimum capacitance to more negative values as the electrolyte concentration is increased. This behavior (112) and results from is termed the Esin-Markov effect the compensatory change in electrode charge that accom- panies specific adsorption. In order to keep gm = 0 as the quantity of adsorbed anions increases (with increasing u), the electrode potential must be moved to (113) more negative values. Interestingly, the opposite ordering of perchlorate and fluoride adsorption on mercury is found.(ll4) Valette has argued that the physical structure of solid metal surfaces may play an important role in determining the extent of specific adsorption.(21) Defects such as steps on the surface are predicted to bear a positive charge with respect to the plane of atoms composing the rest of the surface. It is at these sites that specific adsorption first occurs. A model of the inner layer at the silver electrode surface shows that a smaller ion such as fluoride is more completely contained within the inner 6- or ClOéZ<21) With a greater portion of the ion charge residing in layer than an ion of the size of PF the inner layer the effect of specific adsorption is enhanced. This, in addition to specific chemical interactions, could account for the observed trends. Studies conducted with outer-sphere Co(III) reactants in perchlorate and hexafluorophosphate media have demonstrated that at positive electrode potentials the 85 values of q' and qm are nearly equal in magnitude (q' slightly larger than qm), indicating that adsorption occurs to "buffer" the diffuse layer potentials to low (39’43) The interactions between these anions values. and the electrode are felt to be largely electrostatic in nature and result in the observed balance of charges, q' and qm, on polycrystalline silver electrodes. 4. Electrolytes Which Adsorb Strongly on Silver 4 , and PF6- adsorb to a limited extent on silver, especially at positive values of qm, Although F", C10 other anions such as Cl-, Br-. I-) N3-, NCS-r and SO4= adsorb much more strongly. Figures 3:6 and 3:7 show the differential capacitance and qm - E curves for polycrystal— line silver obtained in 50 mM NaSO4 (u = 0.15). The values of qm at positive electrode potentials are much larger than in the examples for PF - or C104- electro- 6 lytes. This indicates that specific adsorption of SO 4 occurs to a significant extent at positive electrode potentials. It has been suggested that the adsorption of 804= near the pzc occurs primarily at defect sites(21) since measurements with carefully prepared silver single crystals show little variation of the capacitance minimum with concentration. (115) Quantitative data on the extent of adsorption of Cl- at polycrystalline(37:116) (22) and single crystal silver surfaces is available from capacitance measurements 86 Figure 3:6. Potential dependence of the capacitance of a polycrystalline silver electrode in 50 mM aqueous NaZSO4. T = 23 i 1°C. Capacitance values were calculated with the geometric electrode area. 87 \\ 4r (b \\ 0 xx ‘8 \\ 70- 60- 50— uuso “.30 40— 1.2 0.8 0.4 -E/Vvs.SCE 88 Figure 3:7. Electrode charge qm versus electrode potential E for a polycrystalline silver electrode in 50 mM aqueous NaZSO4. T = 23 : 1°C. Values of qm were calculated by employing a roughness factor rS = 1.2. qm/pC cm"2 36 24 12 89 l l l 0)— Q4 03 -E/VvsSCE 90 performed at constant ionic strength (mixed electrolyte) and analyzed by the Hurwitz—Parsons treatment. Larkin et al. have extended the investigation to include Br-, I , N3-, and NCS— adsorption on polycrystalline (37) silver. The Hurwitz-Parsons method has also been employed to determine the adsorption of CN- on polycry- stalline silver.(1l7'118) In general the adsorption of anions on silver has been found to exceed that observed at mercury surfaces at a given electrode potential. In some cases the structure of the silver surface may cause enhanced adsorption relative to mercury. For many ions such as Cl-, Br-, NCS-, and CN- though, specific chemical interactions are the most important factors in determin- ing the free energy of adsorption at an electrode surface. (119) Barclay and Caja have discussed the adsorption of ions at metals surfaces as coordinate covalent interactions and have applied Pearson's Principle of Soft and Hard Acids and Bases (SHAB)(120) to predict the ordering of adsorption for various anions on metals. It is anticipated that silver, With a higher work function and electron affinity than mercury, would adsorb ions more strongly than the latter surface. Also, the predicted order of specific adsorption for simple halides, I- > Br- > C1- > F-, is confirmed experimentally.(37’115) (37) Results from the study by Larkin, et a1. show the adsorption of chloride at the pzc's of polycrystalline 91 silver and mercury to be reasonably similar. Thus it appears the greater adsorption on silver at commonly employed electrode potentials (positive of the pzc) is due to the larger value of qm for silver relative to mercury. This suggests the interaction between chloride anion and either surface to be primarily electrostatic. Bromide does show greater adsorption on silver at its pzc than on mercury at the correspond- ing rational potential (the electrode potential relative to the pzc for the metal surface). This may be due to greater bond covalency on silver relative to mercury. Bodé has presented some interesting calculations on the free energy of adsorption of Br_, Cl-, F-, and OH- (121) on Hg, Ag, and Au surfaces. The model employed in the calculations is an extension of one developed by Anderson and Bockris.(122) An assumption of the model is that adsorption results from primarily electrostatic interactions, a step which is not wholly tenable for C1- and Br- adsorbed on gold and possibly silver. Of note is the conclusion that F- and OH- are adsorbed at the OHP rather than at the electrode surface as are Cl- and Br- on gold and silver. Specific adsorption involves contact with the electrode surface and in this sense the results for F- and OH- are incorrectly labeled as adsorption rather than a simple double layer effect. The work does show that the desolvation energies for F- and OH- Specific adsorption (including desolvation 92 of the electrode surface) are quite high and that the free energy of adsorption should be less negative than the other anions. It was mentioned earlier that in non-adsorbing electrolytes such as PF6- and C104- kinetic data suggested that q' tended to counterbalance qm through electrostatic interactions. The calculations for F-, indicating weak adsorption, lend some support to this conjecture. F. General Features of the Silver—Electrolyte Interface If the differential capacitance curves for silver in NaF solutions of various concentration are compared to the corresponding curves obtained with mercury, it is noted that the capacitance minimum near the pzc occurs at much higher bulk concentrations for the former surface. This behavior stems from the much larger inner layer capacitance of silver relative to mercury. Figure 3:2 shows a dashed line for CEZC (Ag), and from arguments already presented, the behavior noted above would be expected. The large values of Ci(Ag) may indicate that the silver surface is more hydrophilic than mercury.(123) This is of importance in a comparison of specific adsorption on silver and mercury surfaces. As discussed silver tends to adsorb anions more strongly than mercury. When one considers that the desolvation energy of silver exceeds mercury significantly, the energy stabilization afforded by metal-anion contact is yet more striking 93 on the former surface. Valette has proposed that the large inner layer capacitance of silver and other solid metal surfaces may be due to the specific structure of the surface rather than enhanced chemical interaction between the metal and solvent.(21) The pzc's of the various crystal faces of silver tend to be rather different. The values relative to a saturated calomel electrode are: -0.69 V (111), -0.91 V (100), -0.975 V (110), and -0.97 V (polycrystal- line sample).(21’84) The pzc of the polycrystalline surface tends to follow that of the most negative single crystal surface. This can be understood in part by noting the order of the single crystal pzc's to be related to the surface atom densities (sd110 < sd lution phase studies. For reactions involving charge tJransfer in solution, the reaction medium is itself altered during the activation process. A reaction tIIat proceeds at an electrode surface must also occur Etither in contact with or in the vicinity of the surface. c)I'lce the reactant is in the appropriate location, i.e. if! a precursor complex, the thermal activation processes “filich lead to electron transfer can ensue. A key feature cf? the precursor model is that the reactivity of the 97 precursor complex can be directly compared for both outer- and inner-sphere electron transfer. Rate differ- ences between the two pathways assessed in this manner are indicative of any catalysis afforded by an inner- sphere pathway at different electrode surfaces. The evolution of an electrode reaction can be outl ined as follows: k k 1 et bulk reactant a: precursor complex :# successor complex k_1 k—l et k2 =3 bulk product (4.1) k-z Tflne reactant in bulk solution must make its way to the efilectrode by diffusion to form the precursor complex. Since the reactant must traverse the electric field of tile double layer an energy change is usually associated With the formation of the precursor state. For an inner- sEDIhere reaction the reactant penetrates to the electrode Stlrface and can become adsorbed; an additional alteration CIE the precursor stability. Once in the precursor state tile reactant undergoes thermal activation and then Eilectron transfer at a rate described by the unimolecular rate constant ket (s-l) . Following electron transfer tine product diffuses into the bulk solution. It is cOlivenient to think of the precursor and successor state Stabilities in terms of equilibrium constants for their 98 formation. From Equation (4.1) the precursor complex k stability constant is Kp = EeL-and similarly for the k -l successor complex K8 = EjL. The rate at which the -2 precursor complex is formed is assumed to be fast relative to the rate of electron transfer and therefore equilibrium is maintained between the precursor and the bulk reactant. The precursor state phenomenology allows distinction between the free energy required to form the precursor (and successor) complex and that needed ‘to surmount the activation barrier to electron transfer. A value of RP for an outer-sphere reaction (Kg) can loe calculated from an extension of the Boltzmann Ciistribution law (Equation (3.1)). This is possible loecause of the double layer effects which determine tile effective concentration of reactant at the electrode Sllrface c . Thus rP c -ZF¢ KP = fl = 2r exp(—EP—) (4.2) O C RT b Where cb is the bulk reactant concentration, r is the twadius of the reactant, Z is the charge on the reactant, ¢I$’is the average diffuse layer potential at the IWaaction plane, R is the gas constant, and T the absolute temperature. The value 2r appears as a statistical iflactor when converting from a three dimensional reactant cCnncentration in bulk solution to a two dimensional c3011centration at the reaction plane. It is arbitrarily 99 assigned by assuming that any reactant within a distance 2r of the electrode surface undergoes an electron transfer reaction. In actuality a spectrum of possible reaction sites close to the electrode surface exists, and the appropriate term represents an average of these. For inner-sphere reactions the value of K? is given by the ratio of the surface reactant concentration to the bulk reactant concentration K? = Er— (4.3) b 13. The Activation Process for Heterogeneous Charge Transfer After the reactant has diffused to the reaction site, an ordered series of events precedes the electron transfer Ertep. The different time scales on which these processes Cuzcur separates them and makes the overall process of Eilectron transfer stepwise in nature. The following dxescription applies to the outer-sphere reaction of txransition metal complexes in particular, but many features of the discussion are pertinent to the entire Escope of heterogeneous charge transfer. Inner-sphere reactions on silver electrodes are treated in detail in Chapter VI. An analogy to the Franck-Condon restriction(125) may hue drawn to decide the proper ordering of changes which _" 100 precede electron transfer. First the ionic atmosphere near the reactant must achieve a configuration suitable for electron transfer. This takes place in approximately 10-3 seconds. Next, a reorientation of solvent dipoles 11 occurs near the reactant in roughly 10- seconds. Once the environment external to the reactant is configured (the outer-shell component) changes in the metal-ligand bond lengths (the inner-shell component) take place and Inatch the free energy of the reactant to the Fermi level (of the electron in the metal electrode. Inner-shell iseorganization requires about 10-13 seconds, on the order c>f molecular vibrational frequencies. At this point the ealectron can tunnel between the electrode and reactant 16 seconds. Since the transport of the electron in 10" 143 so much more rapid than any of the preceding events Eilectron transfer takes place with an essentially '5Erozen" nuclear geometry. This is the expectation flcom the Franck-Condon principle. The coupling of electronic orbitals between the reeactant and electrode surface is expected to be weak in all outer—sphere reaction because of the spatial separation Of'the reactant centers and the minimal interaction of ‘their coordinated ligands (the chemisorbed solvent layer in the case of the electrode). As mentioned in Cllapter I, the transfer of the electron is isoenergetic; IMD photon is emitted or absorbed as the activation ‘ fr'n 101 barrier is crossed. This type is termed "weak adiabatic" electron transfer and has been treated formally to develop a theory for the reactivity of outer-sphere (126-128) The solid line in homogeneous reactants. Figure 4:1 illustrates the shape of the free energy— reaction coordinate curve for an electron transfer process. The ordinate is labeled as free energy G rather than potential energy U since we wish to account for the many possible adiabatic paths over the activation barrier and their relative spacing. The spacing of the paths is the entropy of activation AS¢ ‘dhich is incorporated in the overall barrier. The iactivation barrier has a significantly different shape than that expected for the more familiar atom transfer tweaction; a consequence of the weak interaction between reactant centers. It has been claimed in the discussion of this and Eireceding chapters that the electron moves from one hxound state to another by tunneling through the activation barrier. This is a reasonable thought when iihe mass of the electron is considered; it is apparently IEeasible for tunneling to occur over the distances (5-10 A) between reactant centers which exist in outer- SPhere electron transfer. (129) If the electron tunnels through the barrier every time the system is in the proper (isoenergetic) configura— tfiion the probability of electron transfer K is unity, .474- 102 Figure 4:1. Free energy versus reaction coordinate for an electron transfer reaction. AG*-p is the activation barrier to electron transfer within the precursor complex. -RT 1n Kp and -RT ln KS represent the work required to form the precursor and successor states, respectively. 103 rxn. coord. 104 and the system is termed adiabatic. Instances of very poor electronic coupling between reactant centers lead to non-adiabatic electron transfer, and K falls below (9) unity. C. A Rate Formulation from Marcus Electron Transfer Theory (130) A variation of the treatment (126,127) originally presented by Marcus is useful for separating the various components that influence the energetics of an electrode reaction. The heterogeneous rate constant for electron transfer kE at an electrode potential E is given by -AG*"p T) ‘4-4’ kE = Kva exp( where K is the electron tunneling probability, v the frequency factor for activation, Kp the precursor stabili- ty constant, AG¢-p the difference in free energy between the precursor and activated states, R the gas constant, and T the absolute temperature. Notice that AG*-p is different than "normal" free energy barriers since the difference in energy between the bulk and precursor states has been separated and expressed as KP. The free energy barrier from the precursor state can be further divided for a one-electron reduction reaction as follows: 105 AG*'p = AG: - aRT(ln KS - ln KP) + dF(E - 3°) (4.5) The terms of the equation are AG: the intrinsic barrier to electron transfer, K8 the formation constant of the successor state following electron transfer, Kp the formation constant of the precursor state, E the electrode potential, E° the standard potential of the redox couple, and a the transfer coefficient. In Equation (4.5) the quantity aRT(1n Ks - ln KP) reflects the effect of the work in forming the successor and precursor states on the overall barrier, AG*-p. The driving force F(E - E°) for the reaction constitutes another thermodynamic portion of AG*-p. When these thermodynamic components are removed from AG*-p the remaining term AG: reflects the force constants of the bonds that are stretched or compressed during activation and also the solvent reorganization which takes place near the reactant. The transfer coefficient a expresses the relationship between the thermodynamic terms RT(ln KS - 1n Kp) and F(E - E°), which constitute the energetics of the elementary reaction, and the height of the activation barrier AG*-p. Figure 4:1 serves to illustrate the relationship of a to the shape of the free energy barrier. The dashed line represents the energy surface for a thermodynamically more stable precursor state. When the precursor stability is altered by an amount AG, a fraction a of this change 106 appears in the activation barrier. For a symmetric activation barrier the value of a is predicted to be close to 0.5, a value commonly derived from outer- sphere rate measurements. If the barrier to electron transfer is asymmetric the value of a is different than 0.5, but remains within the bounds 0 < a < l for commonly accessible driving forces F(E - E°). 1. The Effect of Potential on the Rate Constant for Electron Transfer If the values of RP and KS are assumed independent of electrode potential E, the differentiation of Equation (4.5) with respect to electrode potential yields 3AG*'P 3E = aF (4.6) Since AG;c F(E - E°), where AG;c is the free energy change for the overall one-electron reaction at potential E and E° the standard potential, we find that differentia- tion of this equation with respect to E yields aAGic/BE = F. One can see that *-p _ BAG 0, _ ,5____A (4.7) GO rc Similarly treating Equation (4.4), the expression for the rate constant, we determine that 340*"? ‘FSE" aln kE 1 =-—( W RT ’ (4'3) 11 107 and, RT aln kE _ aAG*'p _ 3AG*-p —— (——————) - (——————) — (———;——) (4.9) F 8E U FBE u BAGrC u where the subscript u refers to constant ionic strength. A plot of the log of the rate constant versus electrode potential shows the Tafel relation, a linear free energy relationship between AG*-p and AG;C. In general practice a plot of ln k: versus E is made, PP where kgpp is the apparent rate constant, and gives the apparent transfer coefficient dapp. The rate constant E O I k and transfer coeffic1ent a are a arent app app pp quantities because of electrode double layer effects which influence them. For an outer sphere reaction the effect of the electrostatic double layer can be accounted for by Equations (4.2), (4.4) and (4.5). Equation (4.4) can be rewritten as follows to express the apparent rate constant kgpp as a double layer corrected rate constant O E . kcorr plus a double layer correction term. *-p E = p _AG 1n kapp ln Kv + 1n K RT AGi-aRT(1n KS-ln Kp)+aF(E-E°) = 1n Kv + 1n Kp- 4- RT _ 'E A; s_ p p - ln kcorr + RT [aRT(1n K 1n K ) + RT ln K ] (4.10) The terms Ks and Kp may be computed from Equation (4.2) 108 and for a reactant of charge Z undergoing a one-electron reduction are -ZF¢ Kp = 2r exp(—ET-EE) (4.11) s -(Z-l)F¢rp K = 2r exp( RT ) (4.12) Incorporating these terms into Equation (4.10) and changing to base 10 logarithms, one arrives at (Z-a)F¢ E _ E _ rp log10 kapp _ log10 kcorr 2.3 RT (4'13) where the 2r factor in Equation (4.2) has been incorpor- ated in kgorr' If the reaction occurs at the OHP so that ¢rp - ¢OHP’ Equation (4.13) becomes the Frumkin equation for the effect of the double layer on outer-sphere (lll) electrode kinetics. There are a number of implicit assumptions in the foregoing derivation with regard to the activity coefficients of the reactant and activated state and also the potential the ion "sees" in the activated state (the (101) micropotential). The micropotential is different from the average potential at the plane of reaction ¢r (131) P because of ionic self-atmosphere effects. The remaining detail to consider is what modification of a occurs from double layer effects. Starting with Equation (4.9) we can say 109 ainkE a =..5$ (____2229 (4.14) app F‘ 3E and applying this to Equation (4.13) we arrive at E aln k 30 ___RT corr _ r OLapp — F‘ ( 8E )U + (Z acorr)( 5E ) u a¢rp = O‘corr + (Z - O‘corr)("§Em) (4'15) u The value of acorr is equal to a in Equation (4.5) if ionic self-atmosphere effects can be neglected and the reaction occurs at the OHP as assumed.(132'l33) 2. The Calculation gfpg Rate Constant from a Measurement ofi Current The measured current i for a diffusing reactant, in terms of the rate constant and reactant concentration, . _ E i - nFA cskapp (4.16) where n is the number of electrons transferred, F the Faraday, A the electrode area, C8 the reactant concentra- tion just outside the diffuse layer, and kE the apparent aPP rate constant mentioned earlier. The dimensions of kgpp are length per unit time since one is relating a current (reactant flux) to a three-dimensional concentration. Thus the calculation of a rate constant is straightfor- ward once cS is known. Because the measurement of kgpp involves current, i.e. the conversion of reactant to product, a concentration gradient occurs in the vicinity "u'i__r 110 of the electrode and Ch # cs. For many electrode geometries, reactant transport conditions, and potential- time functions it is possible to correct for concentration polarization and determine Cs' For the methods employed in this study the corrections are detailed or referenced in Chapter II. An equivalent relationship for a reactant adsorbed on an electrode (inner-sphere electron transfer) is i = nFAP kE (4.17) et where F is the surface reactant concentration and kit a unimolecular rate constant for electron transfer with the dimension of reciprocal time. In certain instances it is possible to estimate or to determine P directly. Examples of measurements of F for transition metal reactants adsorbed on polycrystalline silver and mercury are given in the following chapter. We can see from Equation (4.3) that Equation (4.17) may be reexpressed as . = p E i nFA Ki ket (4.18) Cb to allow a comparison of the reactivity of an inner-sphere reactant that is measured either directly in the adsorbed state or by a technique such as rotating disk voltammetry which measures a diffusing reactant flux. The values of E et assumptions implicit in the precursor model are correct. k obtained in these two ways should be the same if the 111 Information is presented in Chapter VI to show that certain assumptions are improper for some inner-sphere reactions on silver. D. The Reactivity of Outer-Sphere Complexes at Polycrystalline Silver -- Results and Discussion At the beginning of this chapter it was stated that the kinetics of reduction of outer-sphere transition metal complexes on mercury are generally in accord with the conventional models of the diffuse layer. Similar tests with polycrystalline silver are also possible. From Equation (4.13) it can be seen that, for a given electrode potential, a change in the apparent rate constant for an outer-sphere reactant at two different surfaces would be due to changes in the average potential at the plane of reaction A¢rp' plus any specific metal effects M: E = corr app .12:§ RT (4’19) Aioglo k By varying the charge Z of the reactant an estimate of . . . E A¢rp can be obtained Since the magnitude of Alog10 kapp is proportional to (Z - a ). This presumes that M is corr close to zero. A deviation from the expected change suggests that effects that are particular to each metal- solution interface play a role in outer-sphere electron transfer kinetics. 112 1. Co(III) Ammine Reactants in Weakly Adsorbing Electrolytes -- Results and Discussion + The reduction of Co(NH3)5F2 , Co(NH3)2+, and Co(NH3)SOH3+ were studied with a polycrystalline silver electrode in acidified 0.1 M solutions of NaClO4 and KPF6 and buffered NaF solutions. At a pH‘< 4 in fluoride media the precipitation of Co(OI-I)2 on the electrode surface could be avoided. The selected Co(III) reactants have been demonstrated to react via outer-sphere pathways at mercury electrodes, and the observed rate responses to alterations in the double-layer structure are in (14) Outer- good accordance with the Frumkin equation. sphere pathways are also anticipated for silver electrodes. The kinetic data to be presented support this mechanism for electroreduction. The effect of specifically adsorbed anions on the rates of reduction at silver is also in accord with an outer-sphere mechanism.(4l) Perchlorate, hexafluorophosphate, and fluoride electrolytes were chosen because they tendtx>adsorb Specifically to only a small degree on mercury electrodes.(ll4'l35'136) Also, PEG. and C104- do not form ion-pairs significantly with the cationic (14,101) reactants. The investigation in F_ media was restricted to Co(NH3)5F2+, since the reactants of higher charge can form ion-pairs sufficiently to alter the electrode kinetics. Table 4:1 contains the pertinent - J 1.- , - - . . .vU .mN U C mupumvavzanuhvwv~ww UC. pcmv F: or- TCXUHnTEOrv fideEE/N Ah H HVCU k0 COw 303313 0390 .HOK. mhmeTUEChfunu UHUTUDHX H- V . . II - anHMNGDH. 113 .m.o 0n 0» pwumEHumm Huooc mucmwoflwmmoo Hommcmnu omponumo Umuomuuoo umwma mansoe map mum mononucmumm ow mwsHm> .Ava.ve coHumcvm suw3 UmcwEuoump Homeoemmmoo Hmmmcmuu oeponumo ucmnmmmmo .mmOe u Que mcflESmmm Ama.vv mcoflumsvm scum pm>wnmc Hafiucmuom 08mm may no woouuowaw mucouoe m MOM snoox wucmumcoo mums wouomuuoo nomad meadow may mum mononucmumm ca mmoam> .m Hmwucouom mcouuomam noumum um ucmumcoo much ucmummmma .mowmuom moouuomam way so mamovou mo cofiumufimflomum uco>wum ou «Oaum no when umcuflm mo SE m pwcwmucoo mmuhaouuowamc me 732 2:: when 2 To mm Am.ov Amnoa x ~.Hc OOH- om.o «nos x e.a OOH- meme 2 H.o om +mm0mxmm2coo ee.o m-oa x e com) mmz z H.o ms.o «nos x a com- eOwenz z H.o ss.o euoa x m.v com- the S H.o on Am.oi Amnoa x m.HV cos- Am.ov Aeuoa x m.ev com- ms.o eaoa x H.H com- mmz z H.o mm.o m-oe x m com- v08% 2 H.o me.o enoa x m.e com- meme 2 H.o om +~mmimmzeoo sm.o «lea x 5 com- eoHonz z H.o ew.o enoa x m com- when 2 H.o ma Am.oe Amuoa x o.av cos- Am.oe Avuoa x H.me com- se.o MIOH x ~.m com- eoHonz z H.o e mm.o sues x s.~ com- when 2 H.o mm +mxmmzcoo Ilmll A In Eve mum .m> Mom we MN a mamx A>EV m Nsmuwaouuomam mcouuomam ucmuommm .oomm um mmpouuomao we can mm um moxmamsoo unease AHHHVOU wo cofluooomm on» How muwumEmumm owuoaex In Hue mqm¢e the Ia‘ sil Ca; 114 kinetic parameters for the reduction reactions and lists the apparent rate constants at different electrode potentials E and the values of “app in each electro- lyte. The kinetic data for Co(NH3)5F2+ and Co(NH3)Z+ reduction in 0.1 M KPF and 0.1 M NaClO4 are shown 6 for mercury and Silver electrodes in Figures 4:2 and 4:3, respectively. The plots of loglo kapp versus E illustrate the linear Tafel relation. Though mercury and silver have significantly different properties, as discussed in Chapter III, the rates of reduction of the cobalt complexes are quite similar for the tmm>electrodes. 3+ ~ 6 , the value of acorr ( a) 2+ For Co(NH3)5F and Co(NH3) is expected to be close to 0.5 over the range of measurements.(l4'101) Equation (4.19) then suggests that the ratio of rate changes for the +3 and +2 reactants on the two surfaces is log kE (Co(NH )3+) 10 app 3 6 = (3-0.5) = (2-0.5) w|u1 E 2+ loglo kapp(Co(NH3)5F for M E 0. Inspection of Figures 4:2 and 4:3 indicates that the ratio is indeed close to 5 to 3. More information can be gained by comparing the rates of reduction for Co(NH3)2+ and Co(NH3)5F2+ on Silver to the double layer corrected rates, log10 kgorr' of these complexes. By using equilibrium data from capacitance measurements at the mercury-aqueous interface 115 Figures 4:2 and 4:3. Rate-potential data for the electroreduction of Co(NH3)2+ (circles) and Co(NH3)5F2+ (triangles) at the mercury-aqueous (Figure 4:2) and polycrystalline silver-aqueous (Figure 4:3) interfaces; kapp is the apparent rate constant. Supporting electro— lytes ([H+] = 5 mM): closed symbols, 0.1 M KPF6; open symbols, 0.1 M NaClO4. The dashed and dotted-dashed straight lines are plots of the double layer corrected 2+ rate constants log10 k versus E for Co(NH3)5F and corr Co(NH3)2+ respectively. These latter data were obtained with mercury electrodes from values of kapp and double layer compositional data employed in Equation (4°13L(101) 116 . m u v mHDDHm .o.o.m .m> .25 w) 00? n .m: 0. '9 '0. 9' (|_'S 'uJO/ «10,00! 50' m..- 00¢ . N u v mhgmflm 6.9m .m> .>E\w| 00m e _ QVI (..' swo/ddomo'fim 117 the values of ¢rp required in Equation (4.13) can be determined. A simple method of correction is to note that at the pzc in a nonadsorbing electrolyte there is E no double layer effect and therefore, loglo kapp _ E . _ . - loglo kcorr' Assuming acorr — 0.5 the apprOpriate Tafel line may be drawn as shown in Figures 4:2 and 4:3. The double layer corrected lines have proved to be consistent for a range of electrolytes at the mercury electrode.(101) At electrode potentials more negative than z-300mV (vs. SCE) , the apparent rate constant exceeds the corrected value for both reactants. This implies that ¢rp < 0 and from Equations (3.3) and (3.4) q. + gm < 0 since q + qm = -qs. The investigated potentials are far positive of the pzc for Silver (-0.97 V versus SCE) and the values of qm are positive. Accordingly Iq'l > Iqml and there is sufficient Specific adsorption to make the diffuse layer potentials negative. Another piece of evidence for weak specific adsorption of the electrolyte is the ionic strength 2+ I 0 reduction on Silver. dependence of the rate of Co(NH3)5F Figure 4:4 Shows the Tafel lines for sodium perchlorate electrolytes of u = 0.025, 0.1, and 0.5. As the ionic strength increases the rate of reduction also increases. If there were no specific adsorption the changes in rate as u was varied would be larger than observed. With 2 qm in the vicinity of 20 uC cm- the variation in rate 118 Figure 4:4: Ionic strength dependence of the electro- 2+ reduction of Co(NH3)5F on polycrystalline silver in NaClO4 aqueous electrolytes ([H+] = 5 mM). Electrolyte concentrations: 0.025 M (circles); 0.1 M (triangles); 0.5 M (squares). The dashed line is the double layer corrected rate parameter loglokcorr versus E for Co(NH3)5F2+ reduction on mercury obtained with values and double layer compositional data employed (101) fk ° app in Equation (4.13). 119 0.2 -E/Vvs.SCE 120 as the ionic strength is altered from 0.025 to 0.5 would be roughly two orders of magnitude (from Equations (3.3) and (4.19)). The effect of specific adsorption is evident for the reduction of Co(NH3)5F2+ on silver in fluoride in media illustrated in Figure 4:5. The log kapp - E line obtained in 0.1 M_NaF lies above the double layer corrected value over the entire range of measurement. In distinct contrast is the corresponding Tafel line for the mercury electrode, which indicates that ' lqml > Iq I in the same range of electrode potential. Comparison of the apparent rates of reduction of Co(NH3)5F2+ versus the double layer corrected values shows the order of Specific adsorption to be F- > ClO - > PF - on silver and C10 - > PF — > F- on 4 6 4 6 mercury. The results for silver are in agreement with (21) the conclusions reached by Valette as discussed in Chapter III. Table 4:2 presents some values of ¢rp and I q + qm at various potentials for Co(NH3)2+ and Co(NH3)5F2+ reduction on silver. Information derived from the reduction of Co(NH3)SOH§+ was not employed in the calculations of ¢r as the reproducibility of the P rate potential data was too poor. The pentammine aquo reactant appears to be especially sensitive to the electrode pretreatment and electrolyte purity. (Variations of the apparent rate constant between 121 Figure 4:5. Rate-potential data for the electroreduction of Co(NH3)5F2+ in aqueous 0.1 M NaF (pH 2 4) at poly- crystalline silver (triangles)anuimercury (circles) electrodes. The dashed line is a plot of the double layer corrected rate parameter loglokcorr for Co(NH3)5F2+ reduction on mercury electrodes obtained from values of kapp and double layer compositional data applied in Equation (4.13) (101) 122 _ 2 . — _ 3 . loom Eu \ 83. Que. -E/VvsSCE 123 .mmOe u Que mcwESmmm Av.mv pom Am.mv maceumcvm Scum concasoamo moouuomao co omumno kuoa .mum mucmfim Scum Ev mo mosaw> nuw3 woumaooamo omnmno ponHOmpm haamowmwommmm w .AmH.vv cofluwsvm ca o n S mcflfismmo AmH.vv coflumsvm Scum ecumesoawo cowuommu mo woman may um aweucouom ommuo>mo .>Ho>wuoommmu .Aaum Eov mucmumcoo oumn wouoouuoo Howma oHQDOG was ucoummmm on» mum Hmoox cam mmmx mEHou mafia .momwusm opouuooao may no mhmovou mo coauwuwmeooum uco>mum ou «Oaum no whom Honuwo mo 26 m cocwmucoo mouhaouuooamc H.s~ es.~- mm- mm.o ommu >.~N No.4- moo.o we- ~N.H ooe- onz z H.o m.m~ o o 0 ohm: ~.o~ He.au mH.o ma- me.o oov- eoHomz z H.o o.mH o o o mmm- m m s.sa .m.o- mH.o HH- m~.o cos: some 2 H.o +mo A mzvoo m.m~ o o o mum- H.o~ em.a- ma.o we- os.o ooe- v3on2 z H.o m.ma o o o mmmu o m m.sH mo.o- eH.o m- em.o cos- some 2 H.o +mx mzvoo ANIEO One ANIEO 01v SAIMblv >Ev Anuoox camoa mom .m> m .U new + .v anew 6 He RF: mmmx came: CE: m 69:30.30on ucmuomom .ooouuomam Ho>awm oceaamum>H0>Hom o co mowuocex cofluosoououuooam ocaEEfi AHHHVOU scum muouoEmumm “when menace II Nuv mqmde 124 replicate experiments on the order of 5-10 were observed). Also given in Table 4:2 are values of (Borp/BE)u calculated from Equation (4.15). In a given electrolyte the values remain sensibly constant as should ¢rp at potential E, when reactants of varying charge are employed. The plot of qm versus E (Figure 3:5) for 0.1 M KPF6 and NaClO shows the values of qm to be 15-25 uC cm-2 4 in the region the complexes are reduced. Since Iq'l > Iqml, it appears that q. 3 15-25 uC cm"2 (i.e. a fractional coverage 0 of ca 0.1-0.15). For the electrolytes chosen the interaction with the electrode is anticipated to be primarily electrostatic. When the diffuse layer charge is positive the specific adsorption of anions is enhanced; the converse applies to negative values of ¢rp° This effect contributes a "buffering" action which keeps the diffuse layer potentials relatively small.(l36) The adsorption of fluoride approaches a level that again causes one to question whether Co(NH3)5F2+ m aY react via an inner-sphere route, or more appropriately why it does not appear to do so given the propensity of F- to adsorb at Ag. Binding fluoride to Co(III) should leave most of the charge on F-, given its electronegativity. This should still allow some interaction with the electrode. It is probable that 125 the hydrated radius of Co(NH3)5F2+ and the overall charge make the desolvation of the electrode and reactant less energetically favorable for its specific adsorption relative to F-. Evidence of a similarity for the adsorp- tion of Cr(III) ammine and ethylenediamine complexes with halide ligands versus the free ligands is to be given in Chapter V. 2. Transition Metal Aquo Complexes, Mn+ —aq— The aquo complexes of certain transition metals are another class of suitable reactants for electrode kinetics studies. These include the Via/2+, Crgg/2+, 3+/2+ 3+/2+ 2+/+ . _ Ruaq , Euaq , and U02 aq couples. The substitu tional labilities of the metal ions vary significantly and depends on their oxidation state. Table 4:3 contains a compilation of the pseudo first order rate constants for aquo ligand exchange for some of the complexes, and Table 4:4 gives the rate constants for thiocyanate substitution. For most of the aquo complexes (with the exception of UO§+) the rate of ligand substitution is determined by the exchange rate of water, kex' Those species which have kex > ~lOS-1 can be considered labile on the time scale of electrochemical techniques such as cyclic voltammetry. In a non-complexing electrolyte these reactants would tend to react at the OHP since the aquo ligands would not bind effectively to the electrode. 126 TABLE 4:3 -- Rate Parameters for Water Exchange for Various Transitional Metal Aquo Complexes at 25°C. -1 % Complex kex (s ) T (sec) 1. V(OH2)2+ ~102 a ~w7x 10'3 b _ V(0H2)§+ 9 x 101 8 x 10 3 - b 2 Cr(OH2)2+ 2 x 10 5 4 x 104 b - Cr(onz)§+ N109 7 x 10 1° _. c 3. Ru(OH2)2+ 10 5 7 x 104 2+ -1 d Ru(OH2)6 10 7 4 Eu(OH2)3+ very fast27 <10"8 Eu(OH2)2+ very fast b <10"8 aH. Taube, "Electron Transfer ReactionscflfComplex Ions in Solution," Academic Press, New York, 1970, pg. 4. bR.G. Linck,"Survey of Progess in Chemistry," A.F. Scott, ed., Vol. 7, Academic Press, New York, 1976, pg. 94. cT.W. Kallen and J.E. Earley, Inorg. Chem., 19, 1149 (1971). d W. Bottcher, G.M. Brown and N. Sutin, Inorg. Chem., l8, 1447 (1979). 127 TABLE 4:4 -- Rate Parameters for Ligand Substitution for Various Transition Metal Aquo Complexes at 25°C. Reaction k (M"’1 sec‘l)‘1 1% (sec); 1 M NCS- V(OH2)2+ + NCS- 1.1 x 102 6.3 x 10‘3 2+ + V(OH2)5NCS Cr(OH2)g+ + ucs' 1.8 x 10‘ 3 9 x 105 2+ + Cr(OH2)5NCS Fe(OH2)2+ + NCS- 1.3 x 102 5 3 x 10-3 2+ + Fe(OH2)5NCS no:+ + NCS’ 2.9 x 102 2 4 x 10’3 + UOZNCS+ aR.G. Linck, "Homogeneous Catalysis," G.N. Schrauzer, ed., Marcel Dekker, New York, 1971. T = 20°C; substitution mechanism may be 5N2. b 128 If kex is sufficiently large relative to the rate of electron transfer, then, an inner-sphere pathway is possible. Trace levels of adsorbing anions on the electrode surface could replace one or more of the aquo ligands to form a suitable ligand bridge. Inner- sphere reactions would be anticipated to increase the rate of electron transfer by enhancing the surface reactant concentrations and/or through increased electronic coupling of the electrode and metal ion.(137'138) Solid metal electrodes such as silver, platinum, and gold are especially amenable to ligand catalysis because of their tendency to adsorb anions such as C1- or NCS- over a wide range of potentials. In the course of investigating the oxidation of Cr on silver electrodes it was discovered that the rate of oxidation was much slower than the anticipation (39,139) based on results with mercury electrodes. This measurement contrasted strongly with the relatively 2+/+ 3+/2+ fast electron transfer of U0 2 aq n and Eu(OHZ) which had been studied earlier with Silver electrodes. Both the reduction and oxidation processes of the latter two complexes could be conveniently followed by cyclic voltammetry. (The reduction of Cr(OH2)2+ on silver is not directly observable on silver because of the interfering proton reduction which occurs in the acidified electrolyte). The result with Cr:; 129 encouraged the inspection of other suitable aquo complexes to see if any would Show similarly slow oxidation kinetics. 3. Results Table 4:5 summarizes the rate parameters for several transition metal aquo complexes including the 3+/2+ and the actinide UO2+ lanthanide Eu(OH2)n 2 aq at silver and mercury electrode surfaces. The values of kgpp and aapp are listed for the different complexes and electrolyte compositions as well as the double layer corrected parameters. These latter data were derived in the case of mercury from equilibrium (139) measurements of 0 with Equation (4.13). For OHP silver the diffuse layer potentials were obtained from the rate of Co(NH3)2+ reduction compared to the double layer corrected values, as discussed earlier. In acidified perchlorate or hexafluorophosphate electro— lytes the value of kgorr on silver exceeds that for mercury for all systems except Crga oxidation (Table 4:5). 3+/2+ 3+/2+ 2+ 6 n and U02 aq only the standard rate constants could be obtained at silver For Ru(OHz) , Eu(OHz) because of the rapidity of these reactions. These (94) were analyzed by the method of Nicholson. The 3+/2+ and U02+ were also fast 6 2 aq on mercury. Since the double layer corrected rates reactions of Ru(OHz) for reactions on silver exceeded the corresponding 130 ommm mcfl3oHHom co muv magma How mouoz ANIOH x my NIOH X vofl Amloa x we mlOH x N mica x m Avloa x NV NIOH x H AmIOH x mv mIOH x h mica x m Am.mnv NIOH x m ANIOH x N29 MIOH X m Am.NV NIOH x m ANIOH x my «tea x h NIOH x m AHIOH x NVV me VIOHV mo eIOHv ANIOH X ©.vv AmICH x mv NIOH x o.N MIOH x H.N AHIoom Eve mam x .Oomm um annouoz can uo>aflm ocwHHmummuomaom um moxoamsou ma or on or on or mm mm on on mean mean eoaonz z H.o mean omamu whom z H.o mean eodomz z H.o mmou mmon some 2 m.o mum: com) meme 2 e.o com) some 2 H.o om: meme 2 ~.o om) om: whom s ~.c ooau coal when S m.o ooau ooau moon 2 e.o coal woos z H.o omm omm eoHomz z ~.o omm some 2 ~.o omm ooau ooau eoHomz z e.o ooau whom z v.o mmmlqmw oumaouuoon A>ev m No ooonuooam mmHI mmol oml omvl mmml mum .m> A>ev on N +OD Godumowxo codum on emsm . o. +~ coHum on qmom . o. +~ en a m axo :0.u o. +N> won +m c0wuommm c0eumowxo U coed Hmuoz cofiuflmcmue mo GOwumponouuoon on» how mumumfimnmm owumcwx In muv mamme l‘l‘Ch‘ 131 .ucmceEmucoo um on one >Hnfimmom .muwaouuooam Iomm cw o>fiummwc wouMflsw mm aneucouom Hmfiuomm .uconuzo owpocm manwuoouoo oz .mma mocmnmmou Scum “Hooo mucoeoflmmooo Hommcmuu owoocm omuoounoo mowed mansoo map mum mononucmumm ca mooam> .AvH.vv SOwumsqm Suez cocflEuouoo acoHOflmmmoo Hmmmcmuu owpocm ucoummmHem mcwaamumznowaom now muv manmm Scum Que mo mooam> pom museums MOM mmOe u use mcwasmmm Ama.ve cofluwsvm Eoum oo>flho© Hoox mpcmumcoo ouch wouoouuoo nomad cannon may ohm mononuconmm CH mosam> .m aneucouom ocouuooao pmumum um ucmumcoo ouch ucmnmmmmum ou v0aum no mmmm Honuwo mo SE m oocwmucoo mmuhaouuomamu t muv manna Hem mouoz 132 2+ aq oxidation), it appeared that a specific metal effect rates on mercury (with the possible exception of Cr was present; the value of M in Equation (4.19) would be non-zero. The coordination spheres of all the reactants except Ru(OH2)2+/2+ are labile, and reaction can proceed by inner-Sphere pathways if suitable ligands are at the electrode surface in reasonable quantity, a likely explanation with silver surfaces. This leaves the results for Crig oxidation and the 3+/2+ 6 The Cria ion is highly labile and its oxidation is standard rate constant of Ru(OHZ) unexplained. known to proceed via an inner-sphere pathway when (8) possible. The rate of oxidation should be fast if ligand catalysis is an operating effect in the aquo complex oxidations. 4. Discussion 3+/2+ aq couples in the relative substitutional labilities of its The Cr couple is very different than the other oxidations states. Table 4:3 Shows that Crig is very labile while Crag is extremely inert. The Via/2+, Eugg/2+, and UO§+é; couples are labile in both oxidation 3+/2+ states and Ru is essentially inert in either. aq If there are small quantities of adsorbed ligands on the electrode from contaminants in solution (such as traces of C1. in ClO4-) which allow Crig oxidation to proceed by an inner-sphere route, the ligand will be 133 bound into the inert inner coordination shell of the Cr(III) product and removed as a potential bridge. This process scavenges the electrode surface of bridging ligands, and the reaction is forced to proceed by an outer-sphere mechanism. The other couples do not retain the ligand bridge in the oxidized state and thus they allow it to remain on the electrode surface where it can catalyze the oxidation of another incoming reactant. When Cria is added to solutions containing the reduced form of a different metal aquo complex it serves as a "getter" and frees the electrode of adsorbed bridging ligands. This allows the monitoring of the outer-sphere oxidation rate of the reactant of interest. Since Crga is oxidized immeasurably Slowly on Silver, even at large positive electrode potentials, it does not interfere with other reactions. This experiment was done for the aquo oxidation reactions in Table 4:5 (with the exception of U0; which slowly disproportionates to 04+ and 003+).(l40) Table 4:6 Shows these results; the double layer corrected rates of oxidation for various metal ions on mercury and silver are now 2 relatively close. The addition of Cra+ produced no q effect on the oxidations of other aquo reactants at a mercury electrode nor did it affect Co(NH3)§+ reduction on silver. (The cross reaction between (0UP #0 Katha—[Lag TCG L$>~wm CCflHHTUm>HU>HOL U0 mOmelEOU OSU< H0002 COflUHMCMHE “HO COfiUQUflXCCLUUCHE mv£U MOM mMOUnvEFuuHQQ UUmwww UUUUQMMOU QQNTH UHQZOQ ’Illli' (I. W n V m‘HmANrH. mum mmmmaycmymm SH mmsam> um>HHm mcwaamym>y0>aom new Nuv maame Eoym pmcflmyao aye mo mmsam> cam Ama.vv GOeymsvm ayfis omymHsono m Hmwycmyom cmymym ym yucca mycmymcoo mymy pmyomyyoo “mamH mHasoo xomozHezoov .mycmymcoo mymy Apm>ymmaov ycmymmmm may .Amma mocmymmmyvmmOe aye mcflEdmmm wa mucoumE MOM pom .mycmyommya may mo mwm>H0H©>a ycm>mym oy v0aum Ho wmmm ymayflm mo SE m pmcwmycoo mmywaoyyomamc NIOH x 5 com) em I: Amnoa x H29 com: +~yo 2E H + when S «.0 ma on om.o «(OH x H com) whom z v.0 mm coflymoexo +msm mica x V cm) on Am.02¢ Avloa x my own +NHU SE a + meme 2 ~.o mm on u: muoy x m). cm) was s m e or +~=m Huoa x m.H). com we 4 II AVIOH 2v CON +NHU 2.5 H + u some 2 m.o ma m.H com em m~.o N-OH x m coy- some 2 «.0 on conunoyxo +~> u- Huoa x NV 0mm e082 2 ~.o “when 2 ~.o ma mica x w.v omm om.o muoy x m sea. eoHonzeee.o on “whom z ¢.o mm SOwymcwxo +~HU Huoo Aanomm Eve mum .m> mmyNHOHyomHm mooyyomam mom cowyommm o o .N unoox A>ec m x>ec om .Uomm ym xynoymz can ym>awm mcwaamymmyowaom ym mmmemEou osv< Hmymz cofiywmcmya mo GOwymowKOOHyomHm may “0m mymymsmymm mymm omyomyyou umhmq maaooo II one mam .Ava.ve sceymsvm ayflz cmymasoamo mycmyoemmmoo Hmmmcmyy Deccan ycmymmmm may mum mmmmaycmymm Ge .mma mocmymmmy scum yyooo mycmwoymmmoo ymmmcmuy Deccan omyomyyoo momma maasoao .Umacwycou mmyoz one mamas 136 Co(NH3)2+ and Cr:; is slow even though the driving force is large.)(14l) Remaining discrepancies probably arise from the assumptions made to perform the double layer corrections for Silver. The ionic strength of the electrolytes for the Co(III) reactions which provided estimates of or are lower than those employed P in the transition metal aquo studies. This leads to an overestimate of the diffuse layer potential and an apparent enhancement of corrected rates on silver versus mercury. The responses of outer-sphere Co(III) ammine and transition metal aquo complexes to variations of 0' are reasonably similar.(139) rP for the Crfia oxidation on silver can be assigned. Also, Only an upper limit the double layer corrected rate for V2; oxidation on Silver is likely to possess a significant error since the apparent rate of oxidation is very small at the most positive potentials employed in the measurement. It now appears that M in Equation (4.19) is close to zero. It is possible that the enhanced rates on silver versus mercury for complexes with moderate substitutional labilities arise solely from double layer effects. The greatest discrepancies in the double layer corrected rates for transition metal aquo complex oxidations at silver and mercury given in Table 4:5 occur at more positive electrode potentials where anion adsorption is 137 more favorable. The action of Cr:; would remove anionic species which enhance the reduction of the cationic reactants. For a reaction to proceed via an inner-sphere pathway under the conditions of the experi- ments the rate of substitution would have to be as fast as the unimolecular rate for electron transfer on the electrode surface. As an example one could consider an electrode with about ten percent of a 10 monolayer of contaminant anion (F E 10- mole cm—z) on its surface providing bridging sites for V(II) oxidation. With an apparent rate constant of 2 10- cm 8-1 the surface rate constant ke equals t p p . . kapp/Ko where K0 is calculated from Equation (4.2) 8 using 2r = 6 x 10‘ cm and ¢rp = -20 mV. Thus ket E 3 x 104 s.1 and the rate of oxidation is then _ -6 -2 -l a 2 —l -l ket P — 3 X 10 mole cm s . If kex — 10 M s (from Table 4:4), the rate of substitution would be kex[V(II)]P(103 cm3/1iter) a 10'11 moles cm'2 s'l; this is far below the observed oxidation rate. It appears that extremely labile reactants, such as Cr:; and Euig, might support an inner-sphere reaction with adsorbed ligand bridges, but this appears unlikely for the other reactants studied. Similar work performed by Steve Barr with platinum and gold electrodes indicates specific metal effects (compared with mercury) that lie outside the range of (142) double layer corrections. In these cases it is 138 surmised that the strong ordering of water at the electrode surface presents an unfavorable environment for the oxidation of reactants with aquo ligands presumably through hydrogen bonding effects.(39) 5. Kinetic Probes for the Specific Adsorption 2f Anions on Silver The results of the Co(III) ammine and transition metal aquo complex studies with polycrystalline silver in weakly adsorbing electrolytes presented above indicate that double layer effects constitute the major portion of outer-sphere reactivity differences between this surface and mercury. In view of this the outer- sphere Co(III) ammine complexes can be applied to electrolytes which interact more strongly with silver and produce larger values of ¢rp‘ The response of Co(III) ammine outer-sphere reactants to changes in the diffuse layer potential Amrp provides a method to estimate the quantity of specifically adsorbed anions on an electrode sur- (14'101) In conjunction with measurements of face. differential capacitance, this technique allows the investigation of adsorption over a wide range of adsorbing anion concentrations in solution. Kinetic probe methods are advantageous because they monitor an equilibrium quantity of adsorption and can be employed for very strongly adsorbing anions. 139 When an AC perturbation method is employed to determine electrode capacitance, the potential-dependent adsorption process must remain in equilibrium at all times; i.e. the adsorption-desorption process must "track" the AC signal. With strongly adsorbed anions the rates of adsorption and desorption may be too slow to maintain the required equilibrium. Since the rate of change of electrode potential in a kinetic probe experiment employing rotating disk voltammetry is slow, the condition of adsorption equilibrium is more easily maintained. Also, in the latter technique, forced convection aids the rate of surface concentration adjustment. Strongly adsorbing anions can also present difficul- ties when the Hurwitz—Parsons method is applied. Some anions induce changes in electrode capacitance at very low concentrations and the most negative accessible potentials. In these cases it is not possible to obtain a family of capacitance curves to use in the Hurwitz-Parsons analysis. Kinetic probes suffer an inherent disadvantage in that they require that the results follow a model that allows the correlation of electrode kinetics and adsorbed charge (the Gouy-Chapman-Stern-Frumkin analysis).(101) For the complex heterogeneous surface of a polycrystalline metal electrode such a model may have numerous foibles. 140 6. Results . E Figures 4.6 and 4.7 show Tafel plots (loglo kapp versus E) for the reduction of Co(NH3)5F2+ in perchlorate electrolytes with varying concentrations of added Cl- and NCS- respectively. Even at very small bulk anion concentrations (éllflluM) the change in rate at a constant B app’ from the rate in the base potential E, Alog10 k electrolyte is as large as three orders of magnitude. The sensitivity of kinetic probes makes them useful tools of inspection where perturbations to the system are kept small. Above a certain concentration of adsorbing anions the rate of reduction converges to limiting rate acceleration. This may signal adsorbed anion coverages approaching a monolayer. An alternative, and probably more likely, explanation for the rate saturation may be the very large cationic reactant concentrations in the diffuse layer (>1 M) induced by such high concentrations of adsorbed anions. At high concentrations coulombic interactions between the multicharged reactants could limit the reactant concentration in the diffuse layer. Equation (4.19) (with M = 0) can be used to calculate Amrp from the observed rate changes. With the Gouy-Chapman theory develOped in Chapter III it is possible to find A(qm + q.) at a given potential from the values of Am . We would like to determine q., the adsorbed rP charge. Therefore, we must consider what happens to 141 Figure 4:6. Rate-potential data for the electroreduction of Co(NH3)5F2+ at a polycrystalline silver electrode in 0.1 M NaClO4 + x mM Cl- aqueous electrolytes ([H+] = 5 mM). Chloride ion concentrations: 1, O; 2, 2 uM; 3, 5 uM; 4, 20 UN; 5, 50 uM; 6, 100 uM; 7, 1 mM; 8, 10 mM. 142 .m.o.m .m) > . m mdu NO- _.O.. IO I N I l.035 mo ‘ “W 0'60) 143 Figure 4:7. Rate-potential data for the electroreduction of Co(NH3)5F2+ at a polycrystalline silver electrode in 0.1 M NaClO4 + x uM NCS- aqueous electrolytes ([H+] = 5 mML Thiocyanate ion concentrations x (uM): 0.0 (Open circles); 0.1 (open triangles); 0.5 (open squares); 1.0 (solid circles). 144 l I _oas wo/ dd”). OL60' l l I F -- _<: 0 <1 0 y— d o _ <1 0 —— -92 <1 o 0 <1 o ~00 <1 _. co <1 0 N O o —o. I3 4 o — . 4....) I *- __": D o no _ D— l I l N (O sf - E/Vvs.SCE 145 the charge on the metal qm when anion adsorption occurs. The anion adsorption induces a compensating change in gm which is a fraction of the adsorbed charge. This fraction is termed the "electrosorption valency"(l43) and is in the vicinity of 0.25 (r0.l) for silver in the (37) l 0 probe experiments. Thus A(qm + q ) 5 Ag . Table 4:7 is a compilation of results for the anions Cl_, Br-, and NCS- investigated with Co(NH3)5F2+ and Co(NH3)2+ probes and includes values of Amrp, q', and AG; the free energy of adsorption. Also listed are literature values of AG; for C1. adsorption on silver single crystals. The free energy of adsorption can be estimated from the surface anion concentration I. (moles cm-Z) = FAq' and the bulk anion concentration cb. The ionic strength of the electrolyte remains essentially constant and allows the use of concentrations rather than activities. Choosing standard states of To = 1 molecule cm-2 and 1 co = 1 mole liter.1 the value of AG; (kJ mole-_) is written in the Henry's Law limit as o -l ' -2 AGa (kJ mole ) = -RT[ln I (molecules cm ) -ln cb (moles liter-1)] (4.20) 7. Discussion The data in Table 4:7 show the ordering of anion adsorption to be Cl- < Br- E‘NCS-. The results for C1- 146 .coHyoHOm sta may now HIumyHH mHOE H can mymayompm may you NIEU mHsomHOE H mum mmymym pymocmyma .HONOHO HN .mHH ..m.z .eoeo .nmeo .yyoN .mconumo .m can .aHHoemm .m .oyyon> .oe OOH O (Ho mmz z «0.0 NuHHHHO mm O -Ho omz z «0.0 sHOOHO OHH). ON OOH). OH mm O (Ho mnz 2 v0.0 oHOHHO mHH H.H mH ON OOmI -moz z: H.O eHH O.m Ne ON OOm- (um z: N.O NOH m.m me ON OOmu IHo 2: ON eoHonz z H.O ooHHHmymsuosHoo HHuoHoe no: HNIEo 03 was H .50 o3 mom .n> no? .o u? N so 35 m coda 3.3030on 00355 Q0 .mymo mocmyHommmu mo mHmmHmcm mGOmHmmINyH3ycm wa ym>HHm Hmymwyu mHmch com mycmEmHsmmmz maoum UHymcHM scum um>HHm mcHHHmymmyozHom so mcoHc< mooHym> mo coHyQHOmpd mo hmymcm mmym II huv mnmde 147 and Br- are in agreement with the calculations of Bodé.(121) Bodé's model accounts only for solvation and electrostatic interactions. Presumably the larger radius of Br- compared to Cl- makes desolvation of the Br- anion easier. The concept of coordinate covalent interactions between the adsorbate and the electrode and the application of Pearson's SHAB interaction guidelines (as given by Barlay and Caja)(119) also fit the pattern of adsorption free energies for C1- versus Br-. The value of AG; for C1- adsorption on polycrystalline silver is reasonably close to literature values of chloride adsorption on Silver Single crystals (Table 4:7). Notice that the value of AG; becomes more negative as qm increases. The kinetic probe results are available only at positive values of qm (because of the potentials at which the Co(III) reactants are reduced). The values of qm for polycrystalline silver at the stated potentials were taken from the qm versus E data in Figure 3:5 for 0.1 M NaClO4. The literature values of AG; for silver Single crystals were obtained from the Hurwitz-Parsons analysis of capacitance data and are given at the pzc (qm = 0) for the (110) surface and at an electrode charge corresponding to the kinetic probe work. For the cited literature data the base electrolyte was fluoride. Since this adsorbs somewhat more strongly than perchlorate electrolytes (see Chapter III) M 148 the two sets of data are not strictly comparable. The kinetic probe results for polycrystalline silver are not necessarily anticipated to be in good agreement with the thermodynamic measurements. If the extent of anion adsorption and consequently values of Amrp vary from crystallite to crystallite on the heterogeneous surface the reduction of the Co(III) reactant will proceed at different rates across the surface. The major portion of reduction current may appear from a particular portion of the surface (which includes step sites, dislocations, and other crystal defects) and would not, therefore, represent average behavior. Additional evidence of the complex double layer structure of polycrystalline silver has been (144) The obtained by researchers in the Soviet Union. reduction of 8208= (peroxydisulfate) was monitored at polycrystalline Pb, Bi, Sn, and Ag electrodes as well as Ag(lll), Ag(lOO), and Bi(lll) single crystals. It proved possible to obtain coinciding double layer E corrected rates loglo kcorr for 8208= reduction on all surfaces except polycrystalline silver by using values of ¢rp from capacitance data and GCSF theory. The results for the latter surface were thought to be aberrant because of the significantly different pzc's of the various Silver single crystal faces. 149 8. Perbromate Reduction as a Kinetic Probe The reduction of BrO4- on mercury has been shown to fit GCSF theory reasonably well in the presence of (145,146) anion specific adsorption. The reaction mechanism appears to be a two-electron outer-sphere reduction to bromate ion. BrO4 + 2e + H20 + BrO3 + 20H Similar experiments were carried out with a polycrystalline silver electrode to compare the response of an anionic probe to the results obtained with Co(III) ammine reactants. 9. Results The Tafel plots for Co(NH3)5F2+ reduction in 0.1 M NaClO 0.1 M NaClO + 10 mM KCl, and 4’ 4 33 mM NaZSO4 + 5 mM H2804 electrolytes are illustrated in Figure 4:8. Both the sulfate solution and the perchlorate/chloride mixed electrolyte Show the effects of anion specific adsorption on the rates of Co(NH3)5F2+ reduction. The adsorption of SO4= is anticipated on the basis of capacitance measurements (Chapter III). When essentially the same electrolytes are employed with the BrO4- probe, some interesting results are found. Figure 4:9 gives the Tafel plots for BrO4- reduction in 0.1 M NaClO4, 0.1 M NaClO4 + 10 mM KCL, and 33 mM NaZSO4 electrolytes made basic with 2 mM KOH. Base is added to stabilize the pH since OH- is released 150 Figure 4:8. Rate-potential data for the electroreduction of Co(NH3)5F2+ at the polycrystalline silver-aqueous interface. Supporting electrolytes ([H+] = 5 mM): 0.1 M NaClO4 (circles); 0.1 M NaClO + 10 mM KCl 4 (triangles); 33 mM NaZSO4 (squares). 151 _ _ 0.4 _ 2 3 _ _ Toma Eu \ 83. Sec. 0.2? - E /Vvs.SCE 152 Figure 4:9. Rate-potential data for the electroreduction of BrO4- at the polycrystalline silver-aqueous interface. Supporting electrolytes ([OH-] = 2 mM): 0.1 M NaClO4 (circles); 0.1 M NaClO4 + 10 mM KCl (triangles); 33 mM NaZSO4 (squares). 153 _ _ 2 oo _ com Eu \ ox Soc. uction 0.4 0.2 — E / V vs.SCE 154 upon reduction of BrO4 . The addition of Cl- depresses the rate of perbromate anion reduction as expected from GCSF theory. In distinct contrast is the rate of BrO4- reduction in the SO4= electrolyte; the rate parameters are nearly identical to the pure perchlorate electrolyte. 10. Discussion The degree to which the Co(III) ammine cations are hydrated Should be significantly greater than the perbromate anion. This may mean the plane of reaction of BrO4- lies closer to the electrode than it does for the Co(III) probes. Similar arguments have been presented for Cr(III) ammine and aquo outer—sphere (l3) reactants. A different reaction site could cause perbromate to respond to adsorbed SO4= and Cl- unequally, since Cl- would tend to interact more strongly with silver than 804:. The latter ion, like ClO4-, could be adsorbed primarily from electrostatic forces rather than any coordinate covalent binding to the surface. Since BrO4- would be expected to adsorb on silver similarly to ClO4- at positive electrode charges the interactions with strongly bound Cl- at the surface might exceed those for 804= despite the higher charge. This is tantamount to saying that Br04- reduction occurs via an inner-sphere pathway on Silver and that the observed rate of reduction in the presence of Cl- results from repulsive ion-ion interactions on the surface or the removal of possible reaction sites by adsorbed Cl-. 155 The observations made with perbromate reduction highlight the complexities of presumed outer-sphere reactions. Metals like silver, with more negative pzc's than mercury and generally larger tendencies to adsorb anions might change the predominant mechanism of the electron transfer reaction. Some additional uses of the Co(III) ammine outer-sphere reactants are given in Chapter V. CHAPTER V The Determination of Reactant Adsorption The discussion in Chapters III and IV described some of the methods available for quantifying the adsorption of simple anions on electrode surfaces. It was mentioned that many of these anions were incorporated as ligands in the transition metal complexes under study and could induce their adsorption as well. Knowing whether or not a complex is adsorbed and the value of the surface concentration is important in determining the thermodyna- mic contributions to the measured rate for an inner- sphere reaction. The effect of reactant adsorption on electrode kinetcs is developed in Chapter VI. The information in this chapter is arranged to illustrate some of the means of detecting an adsorbed reactant and also to provide discussion on the relative adsorption of metal complexes containing anionic ligands and the free ligands themselves on silver and mercury electrodes. The techniques utilized for adsorption measurements were Chronocoulometry, fast linear sweep voltammetry, and outer-sphere kinetic probes. With the exception of the kinetic probe technique these methods detect the Faradaic current from the reduction of 156 157 the adsorbed complex. The amount of Faradaic charge required to reduce all the surface bound reactant is a direct measure of its adsorption. Fast linear sweep voltammetry has proven to be the most useful method of the three mentioned when applied to solid electrodes such as silver. When the reactant is strongly adsorbed it is possible to arrange experimental conditions to detect (reduce) the adsorbed species with little contribution from the reactant in (95'147) has been demonstrated solution. Chronocoulometry to be a useful tool for measuring reactant adsorption on mercury electrodes, but it has proven much less applicable with silver electrodes. Kinetic probes, which are sensitive to small amounts of adsorbed charge, are useful when reactant adsorption is too small to be determined by Faradaic measurements. A. Chronocoulometry Rotating disk voltammetry measurements with Co(III) reactants containing NCS- and N3- ligands Showed the reaction order to be significantly less than unity, suggesting that the reactants are strongly adsorbed at the (142) This effect is (148) potentials where they are reduced. too large to be due to the electrode double layer since the reaction order for Co(NH3)5F2+ reduction is near unity in the same electrolyte and in the same region of potential (Figure 2:5). The discussion in Chapter II 158 (Experimental Techniques; Rotating Disk Voltammetry) gives the details of reaction order analysis for rotating disk voltammetry. When reactant adsorption occurs with an inner-sphere reactant, the value of cS computed as described on page 48 no longer represents the reactant concentration just outside the diffuse layer. Instead, the surface reactant concentration F is desired to compute the surface rate constant kit (3‘1) with Equation (4.17). The method of Chronocoulometry developed by Anson and co-workers(95’l47) has proven to be advantageous in determining F. In large-amplitude potential step Chronocoulometry the electrode potential is stepped from a region where the reactant is adsorbed but not reduced (E1) to a potential where reduction occurs at a diffusion limited rate (E2). (Only reduction processes will be discussed but the details are equivalent for oxidation reactions.) The electrode current is integrated as a function of time and the charge passed is given by(147) ZnFADlscbt15 Q = “a + le + nFAI‘O (5.1) where the first term on the right hand side of the equation is the Cottrell equation for the charge consumed by a diffusing reactant, with D the reactant diffusion coefficient and t the time following the potential step. The remaining symbols are defined in earlier chapters. 159 The second term is the charge consumed by double layer charging between potentials E1 (prior to) and E2 (following the step). The final term is the charge required to reduce the adsorbed reactant. A plot of 35 Q versus t gives the intercept le + nFAI‘O and effectively separates the charges from diffusing and adsorbed reactant. The value of le can be determined by performing the potential step in the absence of the reactant. In practice, reactant adsorption commonly alters le and its independent determination (i.e. without the reactant) is not possible. When only the reactant is adsorbed simple corrections for its effect (96) The on le are available with some reactions. required condition of no product adsorption can be achieved with mercury electrodes where potentials far negative of the pzc are accessible. If the products are adsorbed (thereby affecting le) a double potential (147) can be used to deter- step chronocoulometric method mine reactant adsorption if the reactants are regenerated quickly from the products upon returning the electrode potential to its initial value. Unfortunately, polycrystalline Silver electrodes suffer from a number of experimental complications which vitiate the application of Chronocoulometry or its variants to the study of Co(III) and Cr(III) reactants containing adsorbing anionic ligands. Reactant 160 adsorption on silver has a dramatic effect on le. Also, the affinity of silver for the anions released upon reduction of the reactant precludes selecting a potential at which the products are not adsorbed without complications from background reactions such as proton reduction. The oxidation of the Co(II) and Cr(II) reduction products to regenerate the reactants is not possible at silver for the former reactant and is not microscopically reversible for the latter, preventing the use of double potential step Chronocoulometry. Even rough attempts at determining Po by finding an approximate le in the electrolyte alone are difficult because silver tends to accumulate traces of adsorbing anions at common initial potentials. This makes the measured value of le time dependent unless extreme care is used in maintaining pure electrolytes. 1. Results and Discussion A demonstration of the problems in applying Chronocoulometry to polycrystalline silver electrodes in acidified electrolytes is Shown by the reduction of c-Cr(OH2)4(NCS);. This reactant is known to adsorb strongly on silver as determined by fast cyclic voltam- metry (following section). Table 5:1 gives the chronocoulometric intercepts (Q; t = 0) for the base electrolyte and with various concentrations of reactant. The potentials E1 and E2 are selected as described .mum msmym> > m.OI u HMHycmyom HmcHw “mum msmym> > m.oI u HMHycmyom HMHyHcHo 161 o.H y v.0 m.o y m.nm omb m.H m.mm com v.~ H.mm oov v.m H.ov com N.v m.ov com v.o y m.v m.o y m.ov ooH n.v v.Hv om m.m m.ov om o.~ m.mm oH o.o >.wm m m.o y m.mI v.Hm m III H.O H h.wm C Hoe you oMMoouoycH H7003 my H23 =pmyomyyoo= ymmoymyaH my .m> O .m> ANIEo one 0 coHymHycmocou ycmyommm .Uomm ym vOHUmz 2 H.O cH um>HHm mcHHHmym>Ho>Hom co coHyocomm .WHmuzeeammoquIo How mymo coHyQHOmpm owyymEoHcooocoyaU II Hum mamme 162 earlier. Upon adding the first aliquot of adsorbate the value of Q decreases because of the effect of the adsorbate on le yielding an apparent "negative" adsorption. Subsequent additions of reactant increase Q to values exceeding le in the electrolyte alone. This behavior makes it impossible to find To from Q - QDL' As the reactant concentration approaches 1 mM the error in the intercept becomes progressively larger 15 since the slope of the Q versus t plot is increasing (according to Equation (5.1)) as is the effect of reactant adsorption on le. The continued decrease of the intercept may be due in part to proton reduction which is an interfering reaction. Although Chronocoulometry has thus far not proved feasible for reactant adsorption measurements with solid electrodes, there are some applications that seem promising. Chronocoulometry does allow the investigation of adsorption at higher bulk reactant concentrations than fast linear sweep voltammetry; with mercury electrodes concentrations up to ca. 1 mM can be studied before the 3: lepe of the Q versus t plot becomes large enough to cause significant error in the intercept. Thus it should be possible to determine the bulk reactant concentration at which the surface coverage reaches a maximum on silver as long as c e 1 mM or less. There b also appears to be an Opportunity to perform electrode 163 capacitance measurements with chronocoulometric instrumentation.(l42) Small potential steps are employed to find the average value of CT within the increment since dQ = Cde. A series of these steps with varying initial potentials may be used to construct a CT versus E curve. B. Kinetic Probes for Complex Ion Adsorption The response of Co(III) ammine reactants to variations in the diffuse layer potential at the plane of reaction brought about by simple anions adsorption on mercury(101) and Silver (Chapter IV) have been noted to be in general accord with the prediction of the GCSF model. This led to the application of Co(III) ammine reactants to the detection of complex ion adsorption on mercury.(l30) The complex ions were cationic ammine complexes of Cr(III). The sensitivity of kinetic probes to changes in the diffuse layer potential and the opportunity to use higher values of c (ca. 1-10 mM) b than in direct Faradaic techniques makes it possible to detect smaller quantities of charged adsorbates. The mercury data Showed the response of the kinetic probes to be roughly three-fold too small according to the GCSF model when used with the known quantity of adsorbed charge for Cr(NH3)5NCSZ+.(96) Though simple double layer predictions did not fit the results, the 2+ 2+ response of Co(NH3)5F to Cr(NH3)5NCS adsorption 164 on mercury was used to "calibrate" its response to less strongly adsorbed Cr(III) complexes for which no other adsorption data was available. A similar investigation was pursued with polycrystalline silver electrodes. 1. Results and Discussion Table 5:2 is a list of results for Co(III) ammine probes of the adsorption of various Cr(III) ammine complexes on polycrystalline Silver. The first two entries in the table are for the response of Co(NH3)5F2+ and Co(NH3)2+ reduction kinetics to adsorbed Cr(NH3)5NCSZ+. Figure 5:1 shows the Tafel plots of + Co(NH3)5F2 and Co(NH3)g+ reduction on silver in 2+ 0.1 M NaClO4 and with 0.5 mM Cr(NH NCS added to the 3’5 electrolyte. The ratio of rate changes for the +3 to +2 probes is not in accord with the value of 5 to 3 predicted by the GCSF model (cf. Chapter IV). The response of Co(NH3)g+ is too small relative to Co(NH3)5F2+; this is also the case for adsorbed c-Cr(NH3)4(NCS);. When the quantity of adsorbed charge 3)SNCS2+ on Silver (following section) is 2+ for Cr(NH used to predict a rate change for Co(NH3)5F reduction the values of Alog kapp Should be 3-3.4, Significantly larger than observed. This behavior is equivalent to (130,142) that noted with mercury electrodes; the kinetic probe data lead to an underestimate of the amount of 165 .coHyoHom 0y coH memEoo OSHaHompm mo coHyHUpm com: mywHOHyomHm mmma CH coHyospmH macum you ycmymcoo mymu CH mmcmau . c m.O Nm Nm.HI m.O .mxmonNHeoOuouu +Nmmxmmzeoo O.H mN OO.HI m.O .mxmonOHmszuouo +mHmszoo m.m He mO.HI m.O 1meonOHmszuouo +NomHmszoo . . I . m m m m H N mN HN O N O +NHo H mzeuo +Nm H mzcoo . . I . m N O m N N mm Ow H O O +Nmoz H mzeho +mH mzeoo O.O OO ON.H- m.O +Nmo2mxmmzeuo +Nmmxmmzeoo Hm EU one A>Ee AMIomm Eoe HEEV .ocoo mymayompd maoym I a m .oO Hue..om0e1< o x OHooHO .e .Uomw u E OHUm SE m + vOHUmz 2 H.o GH mom msmym> >8 oovl ym Hm>HHm mcHHHmym>uo>Hom :0 coHymHOmoa coH memEoo you mymo maoym UHymch II mum mqm gem x.._Ol _ m._Ol_ .._O NO! MOI _.O Hm: .521 8n. 2: 8 as u a. i.OGoz 5:6 5 E2008 $108 3.3.82.9 ammzm tome: 38 3 .93 co cozosomt mamozenacmvou- o 174 of the LSV data for To and ket is described in the reactant adsorption section of Chapter II. 1. Results and Discussion The data from kinetic probe and LSV measurements of complex ion adsorption on polycrystalline silver are assembled in Table 5:3. Also included are corresponding data with mercury electrodes and the adsorption of the . free anions which serve as ligands in the complexes 1 studied. Values of the coefficient of adsorption Kg (cm) on mercury correspond to the Henry's Law region 4 (Po proportional to ob) for all data except the most strongly adsorbed complexes. For silver the data for free anions and certain complexes at bulk concentrations follow Henry's Law behavior. In other cases similar values of cb for mercury and silver have been given to allow a comparison of the extent of adsorption. Complexes containing NCS- and N3- ligands tend to adsorb on silver to a much greater extent than those complexes incorporating Cl- and Br- as ligands (Table 5:3). The values of K? for the latter species are upper limits from LSV measurements since no adsorption was detected at the highest usable value of cb. The adsorption of the free anions on silver signi- ficantly exceeds (ca. a factor of ten or more) the corresponding Cr(III) complexes even when several ligands are capable of binding to the electrode surface, as in 175 .mmDZHBZOU mum mqmfie smIOH H OOOI HmIOH x m OOmI or sH.O.I ON). NO0.0 OO¢I ma Ium sOIOH x mv H.O OOHI o4 +NummHNmoeuo OIOH x O OO¢I MmIOH x H OOMI or sOIOH x mv H.O OONI on +NummHmszuo IOH x e OOOI or an. smIOH xN.I .Hz mO0.0 OOVI on IHo MLIOH x mv H.O OOHI Os 1WHoNHsoOHOIs ANOIOHv OOmI or . I o m N MVTOH x mv H O OON a +NHo H moeyo wIOH XN.( oovl MOIOH xe.I OOOI or s OIOH x eIH Ow N.O OO¢I m m o HVIOH x mv H.O OOOI on +NHo H mzeuo Have HNIeo oHoe HHOHO Hzee mom .n> mommusm 0ynnyonp< mm 0H oo H>eO m .oomm ym voHum SE mIN + voHumz S H.O cH mycmEmyommmz maoum UHymcHM was anymEEmyHo> Qmm3m ymmcHH Eoyw wysoymz cam ym>HHm mcHHHmymhyowHom co COHymHOmoa coHca pom coH meQEOU you myma coHymyomp< II mum mam mommynm mymaHOmp< WM p O A>EV m QMDZHBZOU mum mqmfle 177 VIOH X m.m VIOH X 0.5 MIOH X H.H VIOH x H.m mIOH x H.H NIOH x N MIOH x m.H NIOH X N MIOH x N MIOH X v MIOH X m NIOH X H NIOH x H MIOH x N mIOH x H MIOH X m ONmIOH X m.N m VIOH X m H.O.l o . I B H 0 Have we m.VH h.mH NH m.m Int!) OH.( 0H.( eowoe AN! H OLHHOHV mo.o vo.o v0.0 mo.c vo.o mmoo.o mo.o moo.o mbo.o mmo.o N0.0 Ho.o moo.o mNo.o mo.o v0.0 Hoo.o Hoo.o HEEV oomI ooml oovI ooml OONI o OONI OOHI cowl OONI OONI OONI OONI o ooNI ooml oovl OONI oovl OONI mom .m> H25 m on on ma mm Om or on mommusm .mMDZHBZOU mum mqmde .mHmozvNHsmVHOIo imHmonmHNmoOHOIk inmoneHNmoOhoIo mUZ mymay0mp< QMDZHBZOU mum mqm<9 178 .a mocmymmmy CH mymp mcHymHommyyxm ma pmchyao mSHm>m .pmyyHanm .Hm>mm3 .b.z ocm .Hmhsm .H.M .cqumq .o scumNO .>yymEEmyHo> mmm3m HmmCHH ha Umyomymp yOZo .o.HIm.o n : .mymaHOmpm MOM cOHmmy 36H oy pcommmHyoo mycmEmysmmmz .Hmanv Nov .mH ..Ema0 .myocH .ym>mm3 .b.z Eoyma .mymEHymm maoym OHymcon I‘ll O.H H.NH OO.O OOOI O.H O.HH O0.0 OONI or x N OH OO.O OONI O x N s H0.0 OONI on ImHmonuo O.H O OO.O OOHI or 0.0 0.0 OH.O Os O.N 0.0 NO.O OOH e.O v.4 HO.O OOH O.m O.N NO.O ONH N N 0.0 0.0 H0.0 ONH Os .HHOon HsoOoOIs O.H O.OH OO.O OOtI O.H O.eH OO.O OOOI O.H H.OH OO.O OOOI N N O.H O.eH O0.0 OOHI om .+Hmon HsoOono HNIEo oHoe HHOHO HSEO mom .n> monousm monouomns 0H CH0 H>se m DWDZHBZQU mum mqm<9 179 f-Cr(OH2)3(NCS)3. For mercury the adsorption of free anions and monobridged complexes is reasonably similar. When multiple ligand bridging is possible the values of K? for thiocyanate containing complexes adsorbed on mercury exceed the values for the free ligands. The electrode potentials for which the values of K? are quoted are much more positive of the pzc for silver (0.97 V versus SCE)(84) (101) than for mercury (-0.435 V versus SCE). Larger positive values of electrode charge qm at silver can cause the surface-adsorbate bond to be somewhat more covalent by inducing greater charge transfer from the anionic ligand. The formation of a partially covalent surface bond could act to differentiate free ligand and complex ion adsorption since in the latter case partial charge transfer would have occurred in the formation of the metal ion-ligand bond. Ligand structure (size) can also be important in determining the adsorption of the free ligands versus complexes incorporating the ligands. Cr(NH3)5Br2+ adsorption on silver was not detected with LSV but Cr(NI-I3)5NCS2+ was measureably adsorbed. Both NCS- and Br- are strongly adsorbed on silver to comparable extents. With a monoatomic bridging ligand such as Br‘ the perturbation of charge density on the ligand by the metal ion-ligand bond may affect the surface- 1igand bond to a greater extent than for a polyatomic 180 species such as NCS-. Electrode surface-to-ligand backbonding may also play a role in aiding the adsorption of the thiocyanate complex.(119’154) Another factor which should be considered when interpreting the LSV data is the rate of desorption of the complex ions. If the potential dependence of adsorption for a complex is such that it is only weakly adsorbed or desorbed at the potentials where it is reduced a significant portion may desorb during the 2+ adsorbed on silver potential sweep. For Cr(NH3)5Br this seems unlikely since Br- remains strongly adsorbed at potentials where Cr(NH3)5Br- is reduced (cf. reference 37 and Chapter VI). The rates of adsorption and desorp- tion of complexes which bind strongly to mercury have been found to be quite slow.(155) This is probable for silver as well Since complex ion adsorption is generally larger on silver than on mercury at comparable potentials (Table 5:3). Also, the extent of complex ion adsorption on silver as measured by LSV shows no significant sweep rate dependence (within the range where the diffusing reactant current remains small), suggesting that the kinetics of surface bond rupture are not introducing detectable measurement errors. Some of the data for the adsorption of Cr(III) isothiocyanato complexes presented in Table 5:3 have been reformatted in Table 5:4. Values of the coefficient 181 .mymayompm you conmy 3mg m.>ycmm 0y UGOQmmHHOU mycmEmHDmmmz .HmmmHV Nov .mH ..Emau .mHOCH .ym>mm3 .h.2 Ecumo sHVIOH x m OONI H.O). HOO.O OONI Imoz OIOH x O.H OO.O OONI mIOH x N H0.0 OONI IOHOonuo OIOH x O.H OO.O OONI H.IOH x O OO.O OONI .mHOonNHsoOHOIs IOH x O O.O OONI OH x N OO.O OONI OonHOrzOno o m ml +N mIOH x O.H O0.0 OONI mIOH x O NO.O OONI NHOonmHNmoOHOIk mIOH x H OO.O OONI mIOH x O «O.O OONI .wHOonOHNmoOuoIn . I I . I O N n sOIOH xO H OON OIOH x O NO O OOH +Nmoz H moi 6 Have HSEV mom .m> HEov HZEV mum .m> mymay0m©a or awe a0 H>EO m on “we no H>eO m .mmpoyyomHm xuooymz can ym>HHm mcHHHmymayowHom co mmmemEoo mymcmmooHayOmH HHHHVHU meow you mymn coHyQHOmpd II vum mqm<9 182 of adsorption K? which correspond to the Henry's Law region were chosen when possible. If the column of data for K? on mercury is inSpected it is apparent that having more than one bridging ligand available for binding to the electrode has a dramatic effect on Kg. When monobridged and polybridged adsorbates are compared there is a 50-100 fold increase in K? for the latter species. An equivalent inspection of the column of K? values for polycrystalline Silver Shows the coefficients of adsorption for mono- and polybridged adsorbates to be much more similar than for mercury electrodes. The major difference between mercury and silver for the adsorption of the Cr(III) isothiocyanato species is in the monobridged complexes. As noted, the mono- and polybridged complexes tend to adsorb to a Similar extent on silver whereas on mercury there is a distinct difference. The origin of this dissimilarity is not yet understood and its explanation will require additional adsorption data over a range of electrode charges qm for the two surfaces. The values of K? in Table 5:4 for silver are given for an electrode charge of ~25 uC cm"2 and for mercury gm 2 5 uC cm"2 at the same electrode potential. The much larger values of qm on silver could cause some leveling of the adsorption of mono- versus polybridged adsorbates. 183 The dependence of the extent of adsorption of f-Cr(OH2)3(NCS)3 on potential (electrode charge) at mercury and polycrystalline silver electrode surfaces is illustrated in Figure 5:3. The data are values of surface concentration To versus bulk concentration cb at a given electrode potential (charge). If the elec- trode potential of mercury is made 100 mV more positive than silver the values of To for the two surfaces are very similar. When electrode charge is again considered it would appear that mercury exhibits a greater tendency to adsorb than does silver at the same electrode charge. A comparison of adsorption on two different metals at a constant electrode charge is more reasonable than a comparison at a constant electrode potential because of possible difference in the p.z.c.'S of the two metals. Competitive electrolyte anion adsorption or greater surface-solvent interactions at silver may be in part responsible for this behavior (cf. Chapter III). The potential dependence of adsorption might also be more important for some complexes than others when one attempts to understand relative values of K? among different complexes and at different surfaces. The adsorption of c-Cr(en)2(NCS); on silver is lower than anticipated when compared with the c-Cr(OH2)4(NCS); complex. It appears that non-bridging ligand effects are important in determining the free energy of 184 Figure 5:3. Surface concentration To (mol cm—z) versus log bulk concentration cb (mM) for f-Cr(OH2)3(NCS)3 adsorbed on polycrystalline silver (closed symbols) and mercury (open symbols) electrodes. Electrolytes: polycrystalline silver, 0.1 M NaClO4 + 2 mM HClO4; mercury, 1 M NaClO4 + 10 mM HClO4. Electrode potentials: -200 mV versus SCE (circles); -100 mV versus SCE (triangles). The data for mercury electrodes were taken from reference 96. 185 (<1 l H“: <1 N 4 __°. 0 N <1 ’ o o o o *‘D' .q 1- d —( o N d —v-: 0 4 0— _Q <1 00 o l l l l .9. :9 .9 In 0 z_uI:) Iow / OJ x no) '0910 Cb /'£M 186 adsorption. On mercury the adsorption of c-Cr(en)2(NCS): increases on going to more positive electrode potentials until approximately -400 mV versus SCE. A further positive excursion of electrode potential causes the adsorption to decline.(96) In contrast, the adsorption of the Cr(III) aquo isothiocyanato complexes to increase monotonically upon going to more positive potentials (i.e. more positive electrode charge). Thus the exact value of qm at which values of K? are compared influence the relative ordering of adsorption for these complexes. The large positive values of qm on silver near -200 mV versus SCE may exert a similar influence on the adsorp- tion of the ethylenediamine complex versus the other Cr(III) adsorbates. The data on reactant adsorption at silver have been applied to the evaluation of the surface rate constant k: (5.1) for electron transfer in Chapter VI. t The topics of bridging and nonbridging ligands effects will be extended to discussions of the factors which determine the height of the activation barrier for elementary electron transfer reactions. CHAPTER VI Inner-Sphere Electrode Reactions The information derived from the study of outer- sphere electron transfer reactions (Chapter IV) and reactant adsorption (Chapter V) provides a means of interpreting the data for the rates of inner-sphere f. electrode reactions at silver surfaces. Once it is known how the electrostatic interactions in the interphasial region and reactant adsorption affect the stability of the precursor state it is possible to extract values of k the rate constant for the et’ elementary electron transfer step. The kinetics of heterogeneous inner-sphere electron transfer reactions should reflect the chemical properties of the electrode material, since the transition state is directly bound to the surface as illustrated in Figure 1:2D. Because of this interaction between reactant centers it is possible for some metals to exert an influence on the stability of the activated complex which the thermodynamic parameters Kp and KS, the stability constants for the precursor and successor states reSpectively, do not account for. That is, the magnitude of AGL'p in Equation (4.5) is also 187 188 perturbed by changes in AGi, the intrinsic free energy barrier to electron transfer. This may be visualized as a rounding of the activation barrier to electron transfer and is illustrated in Figure 4:1. The rate formalism develOped in Chapter IV allows a separation of the factors which perturb the precursor and successor state stabilities from those associated solely with the stability of the activated complex. A. The Energetics g£_lnner- versus Outer-Sphere Reaction Pathways Homogeneous inner-sphere redox reactions of transition metal complexes are expected to be preferred energetically over their outer-sphere analogues because of coordinated (8) nuclear motion in the former pathways. The activation of two reactant centers in an outer-sphere reaction is not strongly coupled. The probability of both reactants Simultaneously possessing ligand configurations suitable for electron transfer is part of the limit imposed on the rate of transfer. Contrastingly, ligand motion during inner-sphere electron transfer is coupled by the bridging ligand and may act to increase the probability that inner shell structures are properly configured for the transition state. When electron transfer occurs at an electrode surface, only one reactant undergoes activation. The 189 electrode surface does not, and merely acts to accept or supply the electron. Transfer takes place between the orbitals of the redox center and the Fermi level of the electrons in the electrode metal. Although the coupling of ligand motion for an inner-sphere reaction at an electrode surface is not equivalent to the homogeneous counterpart, bond stretching which occurs for the reduction of a Co(III) or Cr(III) complex could be aided by adsorption of the reactant and the subsequent weakening of the metal-ligand bond. The degree to which the metal-ligand bond is weakened may be influenced by the type of ligand bound to the electrode surface. Monoatomic bridging ligands are likely to exhibit a greater perturbation of the metal ion-ligand bond upon adsorption than polyatomic ligands.(9’43) The structure of the bridging ligand also mediates the degree of overlap between the reactant centers in the activated complex. Sufficiently strong interaction of the reactants can cause a "rounding of" the activation barrier, as mentioned earlier.(8) The transmission coefficient K for electron tunneling can also be affected by the electronic properties of the ligand bridge. H. Taube and co-workers at Stanford have studied intramolecular electron transfer between transition metals bound together by a bifunctional (55) bridging group. The immediate coordination spheres 190 of the metal ions are maintained constant while the structure of the interposed bridge is altered. The molecules have been designed to allow a study of K by a measurement of the entropy of activation AS*. This can be understood from an examination of Equation (4.4). The temperature dependence of kE (k for homogeneous reactions) yields the enthalpy of activation AH* (maintained constant by design) and AS* with the latter term reflecting variations in the probability for electron tunneling. Redox systems adsorbed on electrode surfaces provide the heterogeneous analogue of intramolecular electron transfer. Chapter VIII provides the results of some activation parameter measurements for adsorbed reactants on polycrystalline silver and mercury electrodes. The information within this chapter is applied to questions on the influence of the bridging ligand during electron transfer, the effect of non—bridging ligands for this same process, the general applicability of the precursor formalism developed in Chapter IV, and the effect of the metal substrate on the intrinsic barrier to electron transfer. B. Results 1. Monobridged Inner-Sphere Reactions of Cr(III) Complexes 191 Table 6:1 is a list of the apparent rate constants kapp and apparent cathodic transfer coefficients “app for a set of Cr(III) pentaammine and pentaaquo complexes, each containing a single bridging ligand. The rate parameters were measured by either rotating disk voltammetry, normal pulse polarography, or cyclic voltammetry at -600 mV versus SCE with a polycrystalline silver electrode in 0.1 M NaClO . The table includes 4 literature data for mercury electrodes at the same potential(13’4l) (chosen to minimize extrapolation of the data). The studies with mercury were performed at a higher ionic strength (1 M uni-univalent electrolytes) than the work with silver, which causes a difference of ~50% in the apparent rate constants compared to those obtained in 0.1 M (1:1) electrolytes. An ionic strength (u) of 0.1 was chosen for polycrystalline silver studies to minimize surface contamination from impurities in the electrolytes. Also listed are values of the effective reactant concentration cS (the concentration just outside the diffuse layer, Chapter II), the surface reactant concentration To (from Chapter V), the surface rate constant ke and the apparent cathodic transfer t! coefficient Get for a adsorbed reactant. For strongly adsorbed reactants such as Cr(NH3)5NCS2+ it was possible to obtain kinetics data for the reactant 192 .mmmm mcH3oHH0w :0 Hum mHame How mmwmm mv.o m.NI o.H mm m N O.OA ON.O on +Num H movuo HNOH x Nev ov.o mIOH x O N.O). O.H mm o o ‘6 2 m m m .H mm o H o .No a +Nym H mzv o HOOH x O.HIO N0.0 OIOH x O.H H.O). O.H or O N HOOH x O.HIO m.o). O). O.H on +NHo H movuo HOHIQ vm.o mIOH x o H H.O). O.H mm m m HON); N0.0 NIOH x m H m). mN.o ma +NHU H mzvuo HOOHV Nv.o mIOH x o N No.o o.H mm HOHIV NN.o mIOH x m ON). mm.o mm m m N mv.o OH OH H.O ma +Nz H moeyu Ho.Hv N0.0 H.IoH x m mo.o O.H mm m m m OIOHN. OHw. O.H ma +Nz H mzvuo HomHv mm.o vIOH x m.N m.o O.H mm va Ov.o mIOH x N.H ON). mm.o m< m N ON.O ON ON N0.0 o4 +Nmoz H movyo H0.0V me.o mIOH x O.H N O.H mm HNV mm.o eIOH x O ON). om.o mm m m ov.o m.oH mH OH.O ma +Nmuz H mzvyo ym ym man In so So HOE OH HZEV mommysm ycmyommm m o NO x O H mam NI 0 HH m mpoyyomHm o a H a o U .muw msmnm> >8 oomI wysoymz was ym>HHm mCHHHmymmuowHom ym mmmemEou HHHHVHU mEom mo coHyoopmm n HMHycmyom mpouyomHm .UoMN ym mmpoyyomHm mymammIHmccH ommpHyaocoz may now mymymEmumm mymm mommysm new ycmymmmd II Hum mHmde 193 .mum momym> >5 oomI ym pmyHEHH conSMMHU mH mymy coHyommmk .ycmyommy pmaHOmUm may mo aoHyospmHOHyomHm may you yamHOHmmmoo Hmmmamyy UHanyMUm m .HOH.vV can Hm.vv mcoHymsvm ayH3 mama mo mmsHm> Eoym amymHSOHmo HywmxeymxmfiummymaHymm mum mmmmayamymm CH mmsHm> mas .HmHycmyom pmymym may ym yamyommy UmaHOmpm may mo coHyospmHoyyomHm may How .yma .yamymcoo mymy momwysmNO .HmHyamyom pmymym ym coHyoopmuouyomHm How .mmmo .ycmHonmmoo Hmmmcmyy 0Hpoaymo ycmymmmm cam . mma .yamymcoo mymy yamymmmao .HH Hmymmau aH UmaHHommp mm omyMHsono mym3 mycmyommy mCHmSMMHU How mo mo mmch> maB .ao u no .ymx mo mmsHm> pmycmmmfi NHyomHHc Mom .HmhmH mmSMMHp may mpHmyso ymsm ycmyommy mo coHymyycmoaoo m>Hyommmma .OMH 020 MH mwufiwhwmmh EOHH Gmxmu QH03 mGUOHUUwHQ EDUHGE HON 0960 @398 ”HuO mHame oy mmyoz 194 in the adsorbed state by employing linear sweep voltammetry (LSV; cf. Chapters II and V). Values of cs in this case are equal to the bulk reactant concentra- est et (ket not adsorbed to a detectable extent with LSV were tions. Estimates of k ) for reactants which were obtained from values of ka measured by steady-state PP techniques in conjunction with Equations (4.3) and (4.16). The reactants listed in Table 6:1 are known to reduce via inner—Sphere pathways at mercury elec- L (13,41) trodes. The kapp values in the table show that in all cases (with the possible exception of 2+) the rate of reduction of a given Cr(III) Cr(OH2)5Br complex on silver exceeds that for the mercury surface. Since the double layer properties of the two metals are similar in weakly adsorbing electrolytes (cf. Chapter IV), it is likely that the reaction mechanism for these complexes is inner-Sphere on Silver as well, rather than an accelerated outer-sphere reaction. Verification gf Reaction Pathway The detection of the reduction of metal complexes in the adsorbed state certifies an inner-sphere electron transfer mechanism. In cases in which the investigator is unable to monitor the reaction of the adsorbed complex directly, alternate proof of an inner-sphere reaction pathway is needed. The rate differences noted 195 in the preceding paragraphs are part of this proof. The use of Simple outer-sphere probes of the electrode double layer can provide corroborative evidence of reactant adsorption. The data in Chapter V Showed that even fairly weak complex ion adsorption may be detected in this fashion. Knowing that a reactant is weakly adsorbed on silver, and is reduced at least as fast as the proven inner-sphere reaction on mercury at the same electrode potential, suggests that an inner- sphere pathway is Operative at the former surface. The rate responses of inner-sphere reactants are qualitatively distinct from outer-sphere reactants when adsorbing anions are added to the electrolyte. This phenomenon was pointed out in work by Weaver and Anson (41) with mercury electrodes and later at platinum and gold surfaces by Guyer, Barr, Cave, and Weaver.(43'l42) Upon the introduction of strongly adsorbing anions the rates of reduction of the inner—sphere cationic reactants generally decreased. The expectation based on changes in the diffuse layer potentials requires a rate enhancement for reactants bearing a positive charge in the transition state. An example of the behavior of outer- and inner- sphere cationic reactants in the presence of an adsorbing anion is given in Figure 6:1. The Tafel plots (loglokapp 196 Figure 6:1. Rate-potential data for the electroreduction of Co(NH3)5F2+ (open symbols) and Cr(NH3)5C12+ (closed symbols) on polycrystalline silver electrodes. The symbol ka stands for the apparent rate constant. PP Electrolytes: acidified 0.1 M NaClO4 (circles); acidified 0.1 M NaCio4 + 1 mM 01‘ (triangles). ICthI‘I difie: 197 I I ' ' l I -2- _ E QO. ace-3 - _ Q 3 -4 - _ ‘200 400 I -600 E/mvvseoe 198 versus E) are shown for Co(NH3)5F2+ (open symbols) and Cr(NH3)5Cl2+ (closed symbols) reduction on Silver in acidified 0.1 M NaClO4 (circles) and the same electro- lyte with 1 mM Cl- added (triangles). A large rate 2+ increase for Co(NH3)5F reduction occurs when Cl- is present, as expected (cf. Chapter IV), but the rate of reduction of Cr(NH3)5Cl2+ is decreased. Rate decreases could occur for reactants containing anionic ligands which are adsorbed to the electrode during electron transfer. In this case the electrostatic interactions between the adsorbing anions from the electrolyte and the anionic ligands of the reactant are repulsive. The observed rate change would depend in part on the relative magnitudes of the electrostatic interactions in the diffuse and inner layer regions. Adsorbed anions may also exclude inner-Sphere reactants from catalytic sites on the electrode surface. Additional data for the apparent inner-Sphere reactivities of some monobridged Co(III) complexes is given in Chapter VIII on Miscellaneous Experiments. The data are not included here since it was not generally possible to obtain reliable estimates of ket' the surface rate parameter, for these reactants. 2. Multibridged Inner-Sphere Reactants Listed in Table 6:2 are some kinetics parameters for the reduction of various multibridging Cr(III) and .ycmyommy pmaHOmpm may mo coHyOO©mHoyyomHm may MOM ycmHOHmmmoo Hmmmamuy OHpoaymONO .HOH.eV cam Hm.ve mcoHymoUm ayH3 amyommy maHOSMMHo m MOM ycmymaoo mymy yamymmmm may mo mmon> Eoym pmymHSOHmo Hymwav yma mo mmymEHymm mum mmmmaycmnmm aH mmsHm> .ycmyommy pmay0m©m may mo coHyosomHoyyomHm .may you yamymcoo mymy mesmysm yamymmmao .HH Hmymmao cH pmaHHommo mm pmymHDOHmo mIHmz myamyommy maHmSMMHU now no mo mmon> mam. .mymmo >mH may HON H> .Hmymmao .mov OHMHycmyom mpoyyomHm m>HyHmom mHOE 0y Hmmmy Op mmmum>oo mommysm HmHyHCH may was no HmmmH mOOMMHp may mpHmyso ymcm yamyommy mo :oHymHycmoaoo m>Hyommmm maea .myamEmH5mmmE H>mHv myymEEmyHo> mmm3m HomaHH ma UmaHmyao mym3 mmommysm ayoa you mymp yamyommy cmaHOOUm mam.o O.OI. OH.O O OO.O OOHI or HNNO HN ON.O OOHI on N N % O.O.I OO OH H.O OOHI on +HOon HsoOoOIo 1 OO.O N OH OO.O OOOI or N N O0.0 OO O O0.0 OOOI on +Hmozv HsoOHOIo e.OI. OH HH OO.O OOOI or O O.OI. OOH OH O0.0 OOOI OH IOHOOzOno O0.0 ON OH ONO.O OOOI or O O N O.OI. OOH N ONO.O OOOI as HOOzO H moOHOIN NO.O ON a O0.0 OOOI or N a N O.OI. OO HH OO.O OOOI on +HOon H OOOHOIs o0 H Inc H I50 Hos OHO Hzav mom .n> monuusm schooner s H N no HH n n 8 our AH O O o H>aO m 6.0 MN ym Nycoymz cam ym>HHm mcHHHmyONHONHom co mmmemEoo HHHHVOU paw HHHHwHU meow mo coHyoopmm mymammIHmacH ommoHyaHyHoz may now mymymfimymm moHymaHM II NOO mamas 200 Co(III) reactants adsorbed on Silver and mercury electrodes. The surface rate constants ket were obtained from linear sweep voltammetric measurements that were analyzed as described in Chapter II. Also given are the apparent cathodic transfer coefficients Get for the adsorbed reactant. Comparable values of the initial surface reactant concentration To were selected for both mercury and silver electrodes. The values of Get were sufficiently similar on either surface for a given reactant (except for c-Co(en)2(NCS)2+) to make the rate differences between mercury and silver reasonably insensitive to the potential selected for comparison. The data for c-Co(en)2(NCS)2+ reduction on mercury were extrapolated to more positive potentials for comparison with Silver since rate measurements with the former surface are more reproducible. est Also given in Table 6:2 is a value of ket c-Co(en)2(NCS)2+ reduction on Silver obtained from the for apparent rate constant kapp obtained by rotating disk voltammetry. Measurement of kapp at silver for the other complexes in Table 6:2 were not successful because of anodic back-reactions and interference from proton reduction. The General Applicabilitygf the Precursor Formulism est The values of ket for a given reactant in Tables 6:1 and 6:2 computed from measurements of kapp tend to 201 be 2-5 fold smaller than values of ket determined directly in the adsorbed state. A potential contribution to the steady-state constant kzit is the rate at which the reactant is able to adsorb to the electrode and form the precursor state for inner-Sphere electron transfer. The desorption rate of the reaction products also influences the overall rate in the precursor complex model outlined in Equation (4.1). Slow adsorption- desorption kinetics can effectively make a part of the electrode surface inaccessible during a steady-state measurement such as rotating disk voltammetry, rendering invalid the assumption of equilibrium for these steps in the precursor model. As mentioned in Chapter V, complexes which bind strongly to mercury have been demonstrated to have slow rates of adsorption- desorption.(155) Similar behavior is anticipated for silver. It is also conceivable that the activity of the average reaction site for a steady-state measurement differs from that of the equilibrated reactant in the LSV determination. Sites on the electrode which exhibit more favorable adsorption-desorption rates may provide the major contribution to the overall electron transfer rate. Gerenser and Baetzold, using ultraviolet photoelectron Spectrosc0py, found the initial rates of halogen adsorption on silver surfaces with step 202 sites [3(111) X (100) surface] to be twice as large as a low-index (lll) surface.(156) C. Discussion 1. Metal Substrate Effects 23 the Reactivity 9f the Inner-Sphere Precursor Complex The values of kapp for a given reactant in Table 6:1 have been noted to be generally larger at silver than at mercury electrodes. When the differences in the stability constants Kp for the formation of the precursor complex are accounted for at each surface, via Equation est et are much more alike (Table 6:1). Thus a portion of the (4.18), the resulting corresponding values of k differences in reduction rates for these complexes at silver and mercury surfaces arise from surface reactant concentration changes, a thermodynamic effect. The rates of adsorption on mercury of the monobridged complexes listed in Table 6:1 are thought to be diffusion (155) est controlled or nearly so, et making k z ke for t this surface. A question which has been posed in this and earlier chapters is whether or not the electrode metal can influence AGi, the intrinsic activation barrier to electron transfer. Equation (4.4) allows one to compute #- . . . values of AG p, the activation barrier to electron transfer within the precursor complex, from the ket 203 and RESt et data given in Table 6:1. The term AGdt-p does reflect variations in AG: as Shown in Equation (4.5), but still contains influences of a thermodynamic nature. Though values of E0 are not available for the reactants under consideration, the contribution of the driving force term F(E - E°) can be held constant for a given reactant at different surfaces by selecting a constant electrode potential E, as was done for the data in Table 6:1. Any changes in the relative values of AGit-p are then due to changes in AG: and aRT ln(KS/Kp). The latter component of AG‘Z-p is zero if the stabilities of the precursor and successor states are equal. This is thoughttx>be approximately the case for these Cr(III) systems reacting at mercury,(130) The adsorption of the complexes in Table 6:1 and the free anions incorpora- ted in these complexes are similarauz—600 mV versus SCE on mercury. For this reason, the adsorption of the Cr(II) successor complex is anticipated to be nearly equal to the Cr(III) precursor. Jahn-Teller distortion and the lower metal ion charge in the Cr(II) complexes make the Cr(II)-X interaction, where X is the bridging ligand, smaller than that for Cr(III)-X species.(157) The arguments presented regarding the relative magnitudes of-Kp and K8 for mercury are less satisfactory in certain instances for silver. Table 5:3 shows that the adsorption of Br- and NCS- is 102-103 greater at 204 the stated potentials than the corresponding CrIIIL5X2+ species where L = OH2 or NH3. The largest K8 to Kp ratio occurs for the ammine complexes since they have the weakest adsorption. The relative potential dependence of free anion and complex ion adsorption will (37,130) but Ks make this difference less at -600 mV, could still exceed Kp. The complexes containing Cl- and N3- are expected to have Similar successor and precursor state stabilities (of. Table 5:3). If KS exceeds KP, then the relative values of AG: are larger than those reflected by ket or AG*-p est data for silver alone. The inspection of ke and ket t and mercury respectively (Table 6:1), indicates that Cr(III) ammine reactants are reduced 10-20 times faster on silver than mercury, and that roughly the opposite is true for these species with aquo ligands. The relative values of AG: for the reduction of ammine complexes on silver and mercury are likely to be quite close when the effects of the aRT(KS/Kp) term on AGjtmp are considered. At the least, the rate data for the precursor complex reactivities show that silver does not have a pronounced effect on AG: relative to mercury, despite the larger electronic charge qm for silver (cf. Chapters III and IV). Also, it does not appear that the energetics for the inner-Sphere reduction of Co(III) ammine or Cr(III) ammine or aquo complexes are more favorable than the outer-sphere 205 route when the estimates of AG: for these two pathways are compared.(130) Apparently, neither Silver nor mercury perturb the metal-ligand bond in a manner that significantly aids the reorganization processes unique to the activated complex. Platinum and gold electrodes do appear to alter AG: for complexes containing Cl- and Br- ligands.(43'l42) The ligand-surface bonds for halide anion adsorption at these surfaces are much more covalent than the bonds (37,158) to silver or mercury. 2. Non-Bridging Ligand Effects The Cr(III) aquo complexes in Table 6:1 react more slowly at Silver relative to the ammine complexes than is expected from a consideration of thermodynamic terms. Aquo ligands are lower in the spectrochemical series (159) and therefore stabilize the than ammine ligands Cr(III) ion to a lesser extent. At the same electrode potential, the driving force (E - E°) is larger for the aquo reactants than the corresponding ammine reactants since E is more negative. This assumes, reasonably, that the shift in E° is determined primarily by ligand interaction with the Cr(III) ion and not Cr(II). The stronger bonds formed by ammine ligands should also cause AG: to be larger than for aquo ligands because of bond stretching which occurs in the activa- tion process. The homogeneous inner-sphere reductions of 2+ 2+ (160) Cr(OH2)5X2+ and Cr(NH3)5X complexes by Cr and 206 Yb2+ (161) have shown the aquo reactants to be 50-500 times more reactive than their ammine counterparts. These considerations make it likely that non-bridging ligand effects lower the reactivity of Cr(III) aquo reactants at the silver-electrolyte interface. The values of ke for the multibridged complexes t listed in Table 6:2 also indicate that as aquo ligands are replaced in the Cr(III) inner coordination shell by NH3 or NCS-, the rate constant for reduction increases on silver relative to mercury. Part of the activation process for electron transfer, outlined in Chapter IV, involves the reorientation of solvent dipoles about the reacting species. A reactant containing ligands which can interact strongly with the solvent, such as aquo ligands which form hydrogen bonds to surrounding water molecules, should be especially sensitive to solvent dipole reorganization energies. Silver is apparently more hydrophilic than mercury,(123) as discussed in Chapter III. Stronger surface-solvent bonds would cause the solvent environment at the silver-aqueous interface to be more ordered than that for the mercury-aqueous system. Thus more work would be required to orient the solvent for the transition state at Silver than mercury. This Should differentiate the reactivities of ammine and aquo Cr(III) complexes, since the latter will have stronger solvent—ligand bonds. 207 3. Bridging Ligand Effects i1 Electron Transfer Reactions A comparison of the inner-sphere reactivity of various transition metal complexes at different metal surfaces is one method of ascertaining the presence of any Special substrate effects on AG:, the intrinsic reorganization barrier to electron transfer. An alternate route is to select a series of transition metal complexes in which only the bridging ligand is varied. A ligand which "transmits" to the metal ion-ligand bond a greater portion of the electronic perturbation induced by surface bond formation may cause smaller values of AG: than other ligands. Monoatomic and polyatomic bridging anionic ligands are useful model systems, for the reasons given in a preceding section. Some suitable reactant pairs for the proposed test would be Cr(OH2)5NCS2+ versus Cr(OH2)5Cl2+ 2+ NOS2+ and Cr(NH3)5 versus Cr(NH3)5Cl . To extract AG: from AG*-p, values S, and Kp are required (Equation (4.5)) assuming for E, E°, K a 3 0.5. The chloro complexes are likely to have nearly equal values of KS and Kp on Silver and mercury, and for the isothiocyanato complexes KS should exceed Kp on silver and be approximately equal to Kp on mercury. The formal potential Ef is more appropriate in Equation (4.5) than E° under the conditions of the 208 experiments. Unfortunately, values of Ef are not known for the selected chromium complexes because of the lability of Cr(II). An estimate of the maximum differ- ence in the formal potentials for the chloro and isothiocyanato complexes may be obtained from the stability constants for the Cr(III) species, since they would be expected to account for the greater portion of the difference in Ef for the pairs of complexes. Thus, III F(E2 - El) ~ RT 1 El__ (6 l) f f - n III . K 2 Where Kill and Kill are the stability constants for the Cr(III) reactant with ligands X1 and x2,(130) If the differences in AGi-p for the reactant pairs containing Cl- and NCS- ligands significantly exceed S the differences caused by changes in Ef, K , and Kp, it would indicate an alteration of AG:. Table 6:3 gives estimates of the maximum difference in the driving force term aF(E - E?) and the differences in AG:_p for the substitution of ligand X in a CrIIILSX2+ complex. The reactivity difference (AG*—p - AGzip) in the NCS 2+ 2+ Cr(OH NCS , Cr(OH2)5Cl reactant pair reduced on 2)5 silver is the only entry which is significantly greater than the corresponding difference in the driving force term. The relative magnitudes of KS and Kp for 2()9 m m .memEoo +Nx HHHH no a pH IHU Na Imuz uo aoHynyHymaSO may new &I yuc :H mocmumuuHom .mom mamum> >5 OOOI ym meQEoo Moonsomym may :Hasz ymmmamyy GouyomHm 0y umHyyma :oHym>Hyo¢% m mmmeano NON x IF\H" 1" “I.” m mHmHycmyom Hmsyoa .nymy mmm. myamymcoo :oHymEu0u +Nxma HHHU Eouu omymEHymm mum: x pcmmHH mchHmyaoo .HH.OV :oHymzvu Scum mEumy mo On mcH>Huo cH mocmymmuHc Esames mo mymEHymm v.OH N.H H.NH m.MHIMH N.HIm.H m.vH H HOE Hmox. HHIHOE Hmoxv HomwoaImwmoSI .H. oIaoOI IIIIIIIIII NyaoyszIIIIIIIII 3 .vOH moamymumu scumHO .NOH mocmymumy Eoumo .NOH moamumumu sauna .memEoo UmymHH may mo :oHymEMOH may now ycmymcoo ESHuaHHHsvmo OHIO.OH OH x N.O HOOHmszuo HI +N O.HIO.H N.H n 0.0H NOH x O.N +NmonHm=zOuo . I n . OOHNrocuo O OH OH sHIOH x H H +NH O.NIO.N N.N 0.0H o NOH x N.H +NmonHN=oeuo H Hoe Hours H Hos Hnox. H Hoe Hana. H as onosoo H Hoe ..mozoOOI HI OOI H «mm. Omen O.OI HI e O eIa on k aIa 8 Ho moz u HHH IIIym>HHm mcHHHmymayohHomIII annoymz :o mmmemEou HHHHVHU meow no coHyoavmm mymammlymccn may new .mmvoyyomHm um>HHm maHHHmymayoxHom 0cm myommum vcmmHH maHmcHym no :OmHummEou a II mum mamas 210 Cr(OH2)5NCS2+ at silver will make the difference in AG: values even larger for the reactant pair. Why the aquo reactants alone might show a bridging ligand effect is not clear. The Cr(III) aquo reactants have been noted to be reduced unusually slowly on silver. Considering this, the use of the Cr(III) aquo systems as well-defined examples of bridging ligand effects is not warranted. CHAPTER VII The Application of Surface Enhanced Raman Spectroscopy to the Detection of Adsorbed Metal Complexes The investigations in the thesis work with surface enhanced Raman spectroscopy (SERS) have had two primary facets; the detection of an adsorbed transition metal complex containing an aromatic nitrogen heterocyclic ligand by comparison with spectra of the adsorbed ligand, and the identification of the transition metal ion- ligand stretching frequencies if the complex is observed with SERS. A description of the general procedures for the experiments and an illustration of the equipment that was used are given in Chapter II. The studies were initiated with nitrogen hetero- cycles because of the large SERS signals that had been found for these molecules adsorbed on roughened polycry- (44,45,52) stalline silver electrodes. The molecules are presumed to adsorb to silver via the nitrogen lone pair.(45) A roughened electrode surface yields an additional signal enhancement of 2102 in the SERS spectra.(48) 211 212 It was stated in Chapter I that a possible applica- tion of SERS would be the detection of metal ion—ligand vibrational modes in adsorbed metal complexes. Knowledge of the extent to which a metal surface perturbs the metal ion-ligand bonds of an adsorbed metal complex could help clarify the factors which determine the size of the free energy barrier to inner-Sphere electron transfer (cf. Chapters I and VI). The Ru(II) pentammine complexes of the nitrogen (83) (83) heterocycles pyridine, pyrazine, and 4,4'-dipyridyl(81) were chosen because of the substitu- tional inertness of Ru(II) ammine species.(l65'l66) Alternate metal ions, such as the Co(III) and Cr(III) pentammine complexes of these ligands, are reduced either (167) than those accessible at potentials more positive with silver electrodes (Co(III) complexes) or have not been synthesized (Cr(III) complexes). Unfortunately, the formal potential of the Ru(III)/(II) redox couple is too positive to allow the Ru(III) oxidation state of most complexes to be examined with silver electrodes. Changing the oxidation state of an adsorbed metal complex while obtaining SERS Spectra could help elicit the desired metal ion-ligand vibrational modes. The Ru(II) pentammine complexes of nitrogen hetero- cycles are moderately sensitive to photoaquation. Spectrosc0pic quantum yields ¢aq of 0.045 213 (Ru(NH3)5pyridine2+, pH = 3), 0.001 (Ru(NH3)pyrazine2+, 4 pH = 7) and l X 10- mol einstein-l (Ru(NH3)pyrazineH3+, pH = l) are obtained when the complexes are irradiated at the wavelength of maximum absorption for their metal- (168) to-ligand charge transfer bands. For this reason, the experiments with SERS were conducted with an excita- tion wavelength in the red region of the visible spectrum (647.1 nm line from a krypton ion laser), well outside of the electronic absorption bands for these complexes in aqueous solution. However, the adsorbed complexes may have significantly different absorption (169) spectra. Red light is less likely to induce a resonance Raman signal from either the free ligands or their Ru(II) complexes that are in solution or adsorbed on the silver electrode. The half-life of a monolayer of a Ru(II) complex adsorbed on an electrode surface that is irradiated with light can be calculated from the irradiance E (510W cm-2 R in the thesis work) and an estimate of 4a g at the excitation wavelength for the SERS experiments. If 4a is on the order of l X 10.6 mol einstein-l at q 647.1 nm the rate of photolysis per unit area is 1 2 1 _2 ER(J s cm _ )Oa (mol einstein-l) R'(mol s cm ) = q; th(J einstein-l) 11 (7.1) 5.7 x 10' 214 where L is Avogadro's number, h is Planck's constant, and v is the frequency of the radiation. The rate of photolysis can also be expressed as , -1 -2 _ , -l -2 R (mol 5 cm ) — kaq(s )Fmax(mol cm ) (7.2) where kéq is the first order rate constant for decomposi- tion and Pm is the concentration of a monolayer of ax adsorbate. Thus the first order rate constant k' is aq kéq = 5.7 x 10'11(mo1 s'l cm‘2)/e3x io'lomoi cm‘z) with rmax e 3 x 10'l°(moi cm'z). The half-life r is T = 1n2/k‘;q E 4 s. The estimate of T is yegy approximate, but it does serve to indicate that an adsorbed complex might not be long lived under intense laser irradiation. Roughly one-half hour is required to obtain a SERS spectrum in the thesis work with a scanning monochromator (2000 cm—1 scan). However, it is possible that the metal surface would partially quench the excited state responsible for the photoaquation of an adsorbed complex. A. Results Figure 7:1 is an illustration of SERS spectra for 4,4'-dipyridyl adsorbed on a roughened Silver electrode. The electrode was roughened in aqueous 0.1 M KCl + 10 uM 4,4'-dipyridyl, as described in Chapter II. Roughening 215 Figure 7:1. SERS spectra of 4,4'—dipyridyl and Ru(II)(NH3)5 4,4'-dipyridy12+ adsorbed on a roughened polycrystalline silver electrode in 0.1 M KCl. The abscissa is labeled with the relative wavenumber (cm-1; 647.1 nm excitation source). (a) 4,4'- dipyridyl, -0.2 V vs. SCE; (b) 4,4'-dipyridyl, —0.4 V vs. SCE; (c) 4,4'dipyridyl, -0.6 vs. SCE; (d) Ru(II)(NH3)5 4,4'—dipyridy12+, -0.6 V vs. SCE. and I1 A, 0119. 6.161. The ber CE; SCE. 216 +10 ...______ , é 0'9 In" ' 'Irruoat 8:9 881 CI.“ [.88 99m 963 “St 1.6” 069$ (c) (b) (a) Ausuaiul 200 400 1000 -1 cm 1200 1400 1000 217 the electrode with the ligand or the Ru(II) complex present was generally necessary to obtain intense SERS spectra. The excitation (647.1 nm from a krypton ion laser) was line focused with a cylindrical lens on to the electrode surface with approximately 30 mW of incident power. The laser beam was oriented at 60° to the normal from the surface as shown in Figure 2:3, and the monochromator bandpass was 523cm-1 for all experiments. Lines which appear at 480 cm-1 and 1448 cm"1 (AU(cm-l) relative to the 647.1 nm(15,453.6 on?“ laser line) in the Spectra shown in Figure 7:1 are from non-lasing emissions of the krypton ion laser plasma and have been used to align the individual spectra. The dependence of the intensity of different vibra- tional modes in Figure 7:1 on the electrode potential as well as the low concentration of 4,4'-dipyridyl in solution are evidence that these are SERS spectra and not simple Raman spectra of 4,4'-dipyridyl in the (45) aqueous phase. Of note is the absence of "cathedral" peaks in the 1500 cm-1 region that have been attributed to graphitic carbon on the electrode surface.(170) In order to ascertain the presence of a SERS signal for an adsorbed Ru(II)(NH3)5L2+ complex where L = pyridine, pyrazine, and 4,4'-dipyridyl, the spectra of both the complexes and free ligands were obtained at a common electrode potential. Also, the Raman spectra of polycrystalline samples of the Ru(NH3)5 pyridine and 218 pyrazine complexes in a KBr matrix were taken by using a spinning sample holder. The Raman Spectra of the highly colored Ru(II) complexes are difficult to obtain because of thermal and photolytic decomposition that occurs in the laser beam. The data for these samples have relatively low signal-to-noise ratios because of the low light levels employed for the measurements. Tables 7:1 through 7:3 contain the results of these experiments. The tables give the fundamental vibrational I" - . ( 4" *-__. frequencies of the ligands for the pure liquid or polycrystalline samples and also for aqueous solutions of pyridine and pyrazine. The Raman data and assignments for liquid phase pyridine and pyrazine were taken from reference 171. The equivalent data for 4,4'-dipyridyl are from reference 172. Tables 7:1 through 7:3 also contain the SERS data for the ligands and their Ru(II) complexes. The SERS data for pyridine, and pyra- zine at -400 mV versus SCE, are from reference 52. Very low bulk concentrations (< 100 (M) of either the free ligands or their Ru(II) pentammine complexes are required to obtain SERS spectra such as those shown in Figure 7:1. It was not possible to detect SERS signals when the electrode was roughened in 0.1 M KCl without the adsorbate present in solution. This suggests that special sites on the electrode are responsible for the signal enhancement, and are generated most readily 219 u> NHN .5 .2 .5 :N 3 2n a N3 .5 HOO 5 ONO .. NOO .5 OON : OON .5 OH. .5 OON N .5 2.. N .5 OON OON ‘0 go 0 :5 NON 3> mes Ha :IU mamflwouonsw HH coHyQIHONOv 35 man man No ocvuocmHnIuonao moH .. HOO .5 OOO .. NOO 532530.. a. ucHu ocuHaIcH :oHymIHONOO .- 3: .5 OOHH .5 OOHH .5 :3 no a .70 8-31: a ONNH N 5 $2 .5 =2 sfiHwfimW“ n .. ONOH a ONOH .. NNOH Na :39:- as: a. 2 HNO .5 OHO .. OOO SfiHwfimwuw «O u: o .5 OOO . NOS .5 OOOH 5 OOOH 5 NO. mHnuHfiNoumMHauwu H - OHOH 5 NNOH 5 OHS osmwswuunmuum NH 2. OOOH .. OOOH .5 ONOH .5 OOOH :mwawssuumwnw 2: . OHNH a HNNH .. OHNH Onwawsuummmnw no 3 ON: .5 NOOH .5 HO: 3 NO: sous.»- osHu 03 z OOOH .. HOOH . NOOH .. OOOH ta HOOH H. :38». or? a. umHommw coHymuaHD umammm ahyoll>w no :oHymuaH> H Ila. >ucm=vmuh >N :oHyaHuu-ma :OIHHS H o umymIHx0uen¢ xHuymI no: > v.0: > O.OI mchHuaaI mcHoHMNm cH mHnlm- mcHHHmy-NMUNHOA .meum. .Hmzum. mucosa: OHSUHH N O O n O n a a H 68.75353 H Esta... b+N3H3uE H zzOCszmosHBuE .nom .5 H3» 33:30.. 0.6.503» a. an Harem. :53» 0523-3030.. so woe—0.3. NEHOHNEOHNHNEHHB. van mchHuhm no can .coHyaHom maoosu< :H mcHoHyam .mchHusm vHavHa no umHucmsv0uh choHymuaH> Hmycmlmvcah II HHN man‘s 220 Hamm3 .3 HEDHUmE .E Hmaoyym .aymcmHm>m3 coHymyHoxm Ea m.¢Hm .Nm moamummmy Eoym .m Hmcoyym mym> .ammB Nym> .3> .m> "mHanmm myHmcmyaH m>HymHmmNO .ayoamHm>m3 coHymyHoxm Ea H.Nvmo a .HNH mocmymmmy Somme Hun mHame oy mmyoz 221 III .momm ucHonHOu :0 Nun mHHIWIOy Imyoz 2 man aI ovN I OONI. I OON aI can I NO. N ua.I ONO I van I New 5 NNN :5 NMN : :oHyIIHONIc N 5 ONO .5 NON 55 OHO II .8 On .76 2.393550 HH IoHyIIHONIv I can 55 woe I nor nan ucHu IIIHquonao v I .5 OON I :N .5 OON OON 3.. OIo osuflwflumwuw 3H .5 ONOH II OH: 3.. r38»- OsHu aOH so no I I ONOH 5 OOOH .- OOOH NOOH OIonuuHomsHH. 2: N 5 ONOH .. OOOH II OOOH 5H,. :33». use. IOH an :oHyIIuOumo 3 Ono Hvo a ucHu mcmHmI:H an I OHOH I OHOH I5 OHOH I OHOH I5 NHOH OHOH aoymuyI maHy H IoHyIIHONmp I ONNH I5 vNNH I NvNH I5 NNNH I H.NH NNNH qu mcmHnIcH me I OOOH I vomH I5 camH I5 HOOH H.I canH comH um anymuyI ucHu II ImHommm IoHnyaH> umaeuz anymIlhm no :oHyIuaH> H IIO. Noamsvmyh coHynHHOImn :OIHH: H o I myIIonunH-d xHuyII aux > O.OI > O.OI > O.OI IcHIIuNm meHImuam 5 «Ho-:3 SHHHBOCKHK House. 2932 Human. Ounces... 8 3.5: N O m n m n I u _ mA.—maHumuha H 35:3: IONS—Hanna H Esau—Eh mcHIIqumcHumuR .mom .I> H>vn HIHyImyoa mucuyomHn um um meumv.um>HHm IIHHHmyIauuaHom so vIaHOIod +chHumuaannazvnnsz cam maynmuhm no cam .aoHySHom Inom=v< IH mcHuIyam .mcHnmuam cHSUHH no ImHuamaumuh HIIOHyIuaH> Hmycmlmunuh II Nah nnmaa 222 Hamm3 .3 HESHUOE .E HmCOHym .m .aymamHm>m3 aoHymyHOXm Ea m.va .Nm mocmymmmy Scum .Umoya .Ha HympHdoam .am Hammz Nym> .3> Hmcoyym Xym> .m> umHoaewm NyHmcmyaH m>HymHmmNO .aymamHm>m3 coHymyHoxm Ea H.Nvmo a .HNH mocmymmmy Eoymo NuN oHona 0y noyoz 223 55 NON 55 NNN 55 OON 5 ON. 5 OOO 55 HO. 5 OON 5 OH. 5 ONHH I 55 NNOH 5 555 HO. 555 55 555 HH. NO. N 5 NNOH H5. OOOH 5 OHHH HO. NHHH 5N5 5 HO. HO. HO. 5 ONN H5. OON N 55 ONN N 55 ON. 5 555 HO. OON 5H5 5 ONO HN. ONO ONO N 5 555 HO. HON N 55 55a HH. N55 55 NOOH HO. ONOH 5 OONH HH. OONH 5HI 5 NHN HN. NHN N 5 NNOH 5 OOOH H5. HHOH amuuwuumuuanuuwu H N 5 ONNH N 5 NONH HN. OHNH Imwmunuuuwnu 55 5 HOOH 5 OONH HOH. OONH 5555555 qu 55 OHOH 5 NOOH HO. NOOH 5555555 55H. 5OH 55 HOOH 5 OHOH 5H5. NHOH 5H5 5555555 55H5 55 an i 5i: but. no Ion-53> HHII. 555555555 5N5 585355555 555535 IIIIIIIIIIIIII. ..5555H55555¢ 5 O.OI 5 O.OI HHNHNHuHOI.O.5 .HIIIOO .HOIIIO 555:5 cHaoHH o5- ONHNSNNOHY. O.OO HO.O... :55 235537... 5 O 552555550535 .85 .55 H2- H5353..- 5oouao5HI 55 55 .35.: 555:. 5523555535 55 3550534 NHOIHIEHO I.5.5OHN-:.H555 cs5 HNoHnNNHOI.5. 5 .5 O55 HNIHNNNHOI.O. 5 Ho 55H55555555 H555H5555H5 H555555O555 II N.O 55555 224 H .CHmHyo mmCHEymymmCC Ho yCmCHEmyCoo m oy mom mH mCm EomCmH ym myyommm mmmm may CH mmymmmmm mma HIEo ommH).mo mama am .xmm3 hym> n 3> 3 .ECHUmE u E .mCOHym n m .mcoyym Xym> u m> "mum mHoafiam ymayo HmCma "mHoafihm hwaCmyCH m>HymHmmo .NhH mOCmymwmy Eoym .xmmB u IEO mmNH may oy m>HymHmH mum mmmmayCmymm CH mmCHm> .mmmoe HmCOHymyaH> may Ho ymHH HmHyymm a .HNH mocmnmmmy .mCHmHHNQ How myCmECmHmmm Somme mun mHamB 0y mmyoz 225 when the adsorbate is at the electrode surface during the anodization step (cf. Chapter II). B. Discussion 1. The Detection 2f g3 Adsorbed Metal Complex The data for the SERS spectra of pyridine and Ru(II)(NH3)5pyridine2+ at -600 mV versus SCE exhibit a Significant number of differences. The strong 1215 cm"1 and 1035 cm-1 bands of pyridine are absent in the spectrum of its Ru(II) complex (Table 7:1). Also, a number of low frequency (<600 cm-l) modes are apparent in the spectrum of the Ru(II) complex alone. For some bands which appear in the spectra of both the ligand and the complex, such as the 9b (Wilson vibration number) mode at 1155 cm-1 1 for pyridine (1140 cm- in the Ru(II) complex), there are both significant frequency shifts and changes in the intensities of these bands relative to others in the spectra. This suggests that the intact Ru(II) complex is adsorbed to the electrode. An examination of Tables 7:2 and 7:3 for the pyrazine and 4pT-dipyridyl systems,respectively, also leads one to Similar conclusions. Figure 7:1 illustrates the SERS Spectra of 4,4'-dipyridyl and Ru(II)(NH3)54,4'-dipyridyl2+ at -600 mV versus SCE. It is difficult to predict the changes in the numerous vibrational modes of the heterocyclic ligands 226 upon complexation to Ru(II), especially for an adsorbed species, but certain modes might exhibit behavior that can be anticipated. The Ru(II) ammine complexes of aromatic nitrogen heterocycles are known to have (83) Electron extensive metal-to-ligand n back—bonding. density from the 4d t2g orbitals of Ru(II) is transferred to theIrantibonding orbitals of the aromatic ligand. This causes the bond order of the n system in the aromatic heterocycle to become smaller, and thus it is expected that the vibrational modes associated with ring stretch- ing would shift to lower frequencies. Table 7:4 contains the frequencies of some ring stretching vibrational modes for pyridine, pyrazine, 4,4'-dipyridyl and their Ru(II) pentammine complexes from the SERS data in Tables 7:1 through 7:3. The data for the free and complexed ligands in Table 7:4 do not show a consistent decrease in the ring stretch frequencies. A complication in the w back-bonding argument is the influence of the metal electrode surface on the n electron density of the ligand. For this reason, it is perhaps unreasonable to make such a simple assessment of the effects of complexation to Ru(II). One other factor must be weighed when interpreting these and other published SERS spectra. The exact features of the Spectra are very sensitive to the manner in which the electrode is roughened and the “IF—"T" .xmm3 Num5 I 35 .xmm3 u 3 .EsHmmE n e .mcoyym u I .oaony Nym5 I I5 .ayOCmHm5m3 CoHymymem EC 0.0Hm HIHOAENI NyHICmyCH m5HymHmC .Nm moamumumu Ioum 0 v: .mCHnmyan u no HHNOHCNCHCI.Q.O u NmHm.v.O .mCHmHHOC I yam HImHoaooymyma ammouyHC New IHOAENma or .CUm I=Iym5 5E OOOI HmHycmyoa mmoyyomHm .aymcmHm5m3 CoHymyHoxm EC H.NOO HIyCmEmyswmmE mama Iona mmCHmyao mymoc 2 III I. IIIIII .I IIIII- N3 ONOH II 35 mHmH E NOOH 3 OOvH 35 NOOH IOH I vomH I5 HOOH I5 HOOH E OHOH 3 HOOH I hOmH mO I OHOH I OHOH N3 NNOH 3 OOOH I NOOH I5 OOOH H 6 amasaz IIIII IIIIIIII III I II COHymyaH> HHIEo. Nocmsvmuh COIHH3 mmlm m N“ . mlm NM 5 lNlemllIl JNI MININII.HIIHHz.HHHIHICIIIIIINIIIIII +N HO .v v 35:3— HH... v +Ny H =2:st P y HOImCOHyomHm ym5HHm mCHHHmymhyoaHOm Co mmauompa mmmeQEoo +NHmHn=ZVHHCC yHmae cam mmHozooumymm cmOOHsz IsoHym> no ammo: HmCOHymyaH> OCHaOymyym OCHC mEom new ImHocmawmuh II «On man‘s 228 constitution of the solution for the roughening (anodiza- tion) step. The data for pyrazine in Table 7:2 include SERS spectra from the thesis research and from reference 52. The SERS Spectra from the latter source Show strong Raman forbidden bands at 1485 cm_1 and 1420 cm-1(173) while the SERS spectra for the thesis work do not. The potentials at which the anodization-roughening cycle was performed and the concentration of pyrazine (50 mM in the literature spectra) are quite different in the two experiments. (The potentials for the SERS data obtained with pyrazine in Table 7:2 are different by 200 mV, but moving the electrode potential to -400 mV versus SCE in the thesis experiments did not cause the 1 and 1420 cm.1 bands to appear.) It is 1485 cm- conceivable that the differences in the SERS spectra of the free ligands and their Ru(II) complexes is inherent in the roughening process when different adsorbates are present in solution. If the SERS experiments could be conducted by roughening the electrode and Eben introducing the adsorbate to solution the attempted comparisons would be much more valid. 2. Metal-Nitrogen Stretching Vibrations 9f M(II) Complexes The metal-nitrogen stretching frequencies v of MN the M(II) complexes of aromatic nitrogen heterocycles l (174) generally lie within the 200-300 cm- range. The 229 spectra for polycrystalline (Ru(II)(NH3)Spyridine) (C104)2 and (Ru(II)(NH3)5pyra2ine)(PF6)2, in Tables 7:1 and 7:2 respectively, do Show vibration modes at 1 ~233 cm- (both complexes) and at ~244 cm-l (pyridine complex) which might be the VMN band. The SERS spectrum of Ru(II)(NH3)5pyridine2+ exhibits a nearby band at 1 1 212 cm- and also a band at 333 cm-1. The 212 cm- mode is much more intense relative to other modes in the SERS spectrum of the Ru(II) complex than are the 1 233 cm- and 244 cm.1 bands relative to the vibrational modes in the Raman spectrum of the polycrystalline sample. Also, a 207 cm.1 mode of large relative intensity is observed with SERS spectra of 4,4'-dipyridyl. These 1 considerations make it unlikely that the 212 cm- band of Ru(11)(NH3)5pyridine2+ is the v mode, though MN . . (175) the feature may result from Ag-N Vibrations. An increase in the VMN frequency to 333 cm.1 seems implausible, and it is not observed in the Ru(II) (NH3)5pyrazine2+ spectrum. The strong 296 cm_1 band present in the SERS spectrum of the Ru(II) pyrazine complex also appears in the spectrum of the free ligand. The results of the SERS investigation in the thesis project are far from clear-cut, but they do point in reasonable directions for future studies. An intriguing experiment which could help identify the v frequencies MN of metal complexes would employ a system which adsorbs 230 E29 has an accessible oxidation state change. Osmium(III) /(II) ammine complexes hold promise in this respect for SERS experiments with silver electrodes, since the formal potentials of these complexes are much more negative than their ruthenium analogues.(l76) AS an alternate approach to the observation of a metal ion-ligand band, a gold electrode could be utilized to obtain SERS spectra for Ru(III)/(II) complexes. The SERS spectra of adsorbates on gold d,(177) have been observe and gold is sufficiently stable with respect to oxidation to allow access to the Ru(III) oxidation state. Chapter IX contains some more thoughts on these matters. CHAPTER VI I I Miscellaneous Experiments A. Activation Parameter Measurements for the Electrore- duction 9: Some Co(III) and Cr(III) Complexes 9g Polycrystalline Silver and Mercury Electrodes It was noted in Chapter VI that the behavior of Cr(III) aquo and ammine inner-sphere reactants reduced on Silver electrodes was significantly different from their reduction on mercury. The reactivity of inner- Sphere Cr(III) aquo systems on silver appears to be influenced in part by non-bridging ligand effects associated with the solvent structure at the Silver- aqueous interface. The measurement of the entrOpy of activation for heterogeneous electron transfer reactions at different electrode metals can provide additional information about the influence of the electrode surface on the reactant environment.(178) A set of inner- and outer-sphere Co(III) and Cr(III) complexes was chosen for this reason and a cursory study of their activation parameters was performed. Some of the complexes were strongly adsorbed on both silver and mercury (of. Chapter V), allowing 231 232 the activation parameters to be determined for the reduction of the adsorbed reactant. For the chemically irreversible Cr(III) and Co(III) systems which were investigated, the "real" activation parameters (179) derived from the temperature dependence of the standard electrochemical rate constant are not accessible. One does have access to the "ideal" enthalpy and entropy of activation(179) though, which are determined from the temperature dependence of the electrochemical rate constant at a constant metal-solution Galvani (180) potential difference mm. The "ideal" activation parameters correspond to the activation of a single reacting species at a given electrode potential.(180) It is possible to maintain mm essentially constant by utilizing a non-isothermal electrochemical cell where the temperature of the electrode-solution interface is varied while the reference electrode is kept at a constant (180,181) temperature. 1. The Calculation gf the "Ideal" Entropy and Enthalpy 2f Activation The rate equation based on the precursor model (Equation (4.4)) may be rewritten as A *"p AH*’p ) exp(- —— RT ) (8.1) ket = Kv exp( R where AS!”p and AHL'p are the entrOpy and enthalpy of activation for electron transfer within the precursor 233 complex. Other terms are as described in Chapter IV. The ideal activation parameters may be obtained from aln k An? 9 - -R( et) (8.2) ideal 3(1) ¢ T m AS*-p ' R 1 k - R 1 + (AH*-p /T) (8 3 ideal _ n et n Kv ideal ° ) It can be shown that other contributions to the variation of mm with temperature are small with proper (178) design of the experiment. These contributions are the thermal liquid junction potential difference ¢tlj between "hot" and "cold" portions of the cell and the "thermocouple" potential difference developed between the "hot" and "cold" sections of the working electrode. It is estimated that these quantities cause a maximum 1 mol-l) for measurements at the (178) error of :1 cu (cal K- mercury-aqueous interface. 2. Results The data in Table 8:1 are the ideal activation parameters for the electroreduction of some Co(III) and Cr(III) ammine, ethylenediamine, and aquo complexes on silver and mercury electrode at the stated electrode potentials. The reaction mechanism for the reduction of each complex (inner- or outer-Sphere) is also given. Ionic double layer effects were not accounted for in the calculation of the activation parameters. Values of ke for outer-Sphere reactants were computed by t .QWDZHBZOU Hum mqm<8 OB mMBOZ .EmHCmaomE mymammlymCCH mmyocmm .m.H HEchmaomE mymaCmIHmyCo mmyOCmO .m.o HymmmCmyy CoyyomHm COM EmHCmaomE COHyommmo . Im OH H > yoyomw Noamswmym .HIHOE x5 Hmo NH ConHomym yCmEmyCmmmz .CoHym5Hyom mo NMOHymw HmmmHa .HIHOE Hmoa Hy ConHomHQ yCmEmyCmmmz .CoHym5Hyom mo NQHmaycm HmmmHNO .n.H O OH OOOI Or .n.H OI OH OOOI .Nom O O N .n.H OHI HH OOOI ON .. HOOzO H OOOHOIN .n.H OH OH OOOI Om N O N 4 .n.H OI NH OOOI Om Hmon H moOuOIs n . .. .n.H ON ON ONNI Om .mofl HNI m mNNI O< N N .n.H OHI NH ONNI O< Hmoze HsoeoOIo . k m + .n.H NHI O OOOI ma +NHOOHOszno . . O O m o I m 0 HH OH OOO a H.+OH sz o .m.o N HH OOOI ma +NmmHmmzvoo EmHCmaomz HHIHOE MIxo Hmov HHIHOE Hmoxv mom .m5 mommymm yCmyommm . .1 s. a... . as... as .. .OOHUm SE mIN + OoHomz z H.O CH mmmoyyomHm Nysoymz mam Hm>HHm mCHHHmymayozHom ym myCmyommm Hmymz CoHyHmCmye mymammemyCo mcm IymCCH mEom mo CoHyOCmmyoyyomHm may NOH mymymEmymm CoHym5Hy04 II H.O mamae 235 .NIEO HOE OHIOH x O.H u C NHCOHmE Co COHyUCOmy mHmUzvaNmovaIk How Cam N EU HOE HH OH x O u C ym>HHm Co CoHyUCUmy MHmUZVNHCmVOUIo Com .C CoHymyyCmOCoo yCmyommy mommysm ycmymCoo m ym mmCHEymymp mymymEmymm CoHym5Hyoak .NyymEEmyH05 Cmm3m HmmCHH ayH3 ycmyommy mmaHOIOm Com mmCHECmymm mymymEmymm COHym5Hyo