Catalysis of dioxygen reduction at graphite electrodes by an adsorbed cobalt(ii) porphyrin

J. Electrounul. Chem., 134 (1982) 273-289 Elsevier Sequoia S.A., Lausanne-Printed

CATALYSIS OF DIOXYGEN AN ADSORBED COBALT(H)

RICHARD

R. DURAND

2 1st September

REDUCTION PORPHYRIN

Jr. and FRED

Arthur Amos Noyes L.uhorutories, Technology, Pusudenu, CA 91125 (Received

273 in The Netherlands

Division (U.S.A.)

1981; in revised

AT’GRAPHITE

ELECTRODES

BY

C. ANSON of Chemistry form

and Chemicul

3 1st October

Engineering,

Culiforniu

Institute

of

198 I )

ABSTRACT The in Fig. occurs simple peroxide. Oa/O,H catalyzed

electroreduction of dioxygen to hydrogen peroxide is catalyzed when the Co(H) porphyrin shown I is adsorbed on the surface of graphite electrodes. The electroreduction of the adsorbed catalyst at potentials well separated from those where the dioxygen is reduced showing that more than a redox catalysis is involved. The porphyrin also catalyzes the electro-oxidatoin of hydrogen In I M NaOH catalyst-coated electrodes yield reversible, almost nernstian responses to the couple. Cyclic and rotating disk voltammetry were used to examine the kinetics of the reactions and mechanistic schemes are proposed to account for the observed kinetics.

INTRODUCTION

Catalysis of the electroreduction of dioxygen by complexes of transition metals with macrocyclic ligands, including porphyrins and phthalocyanines, is an active area of current research in electrochemistry. Several reviews of the field are available [l-4] that discuss much of the current literature. One of the more intriguing new catalysts is a recently described cofacial dicobalt porphyrin [5,6] that effects the four-electron reduction of dioxygen to water at unusually positive potentials [6]. The catalyst is insoluble in aqueous media and catalytically active electrodes can be prepared by irreversible adsorption of the catalyst on the surface of graphite electrodes, [5,6]. Attempts to observe the cobalt-centered electrochemistry of the adsorbed catalyst in aqueous acid in the absence of dioxygen have been only partially successful because the voltammetric peaks are often broad and poorly defined. This has left some uncertainty regarding the detailed mechanism by which the cofacial dicobalt porphyrin catalyzes the reduction of dioxygen to water. Much better defined cobalt-centered electrochemical responses are obtained when the cobalt porphyrin shown in Fig. 1 is adsorbed on graphite electrodes. This complex, I, is one of the monomeric porphyrin precursors employed in the synthesis of the cofacial porphyrin catalyst [7]. We have therefore examined the electrochemistry of I as a function of pH and in the presence and absence of dioxygen and hydrogen peroxide in the hope that it might provide some insight into the behavior of the

Fig.

I. The cobalt

porphyrin

utalyst

cmpioycd

in this \tud\

cofacial dicobalt porphyrin that is the better catalyst. Among the novel properties exhibited by graphite electrodes coated with I is a notable potency toward catalytic oxidation of hydrogen peroxide at all pH values culminating in a nearly nernstian response to the O,/OzH ~-- couple at high pH. A clear separation in potential between the cobalt-centered redox electrochemistry and the catalytic dioxygen reduction was observed at all pH values. which suggests a different catalytic mechanism than that proposed for the cofacial dicobalt porphyrin [61. (I) EXPERIMENTAI.

The cobalt porphyrin, I (Fig. l), synthesized as described in ref. 6, was a gift from Professor J.P. Collman. All reagents employed were analytical grade and were used without further purification. Solutions were prepared from distilled water that was further purified by passage through a Barnstead Nanopure purification train. Buffer solutions had the following compositions: pH

0.3

PI-I 2 pH 3-- 5 pH 6-- 7 pH 8-10.8 pH 12-13 pH 14

0.5 M trifluoroacetic acid 0.05 MKCli-0.01 MHCl 0.05 M KH phthalate + HCl or NaOH 0.05 M KH,PO, + NaOH 0.02 M Na,B,O, + HCl or NaOH

0.05 M KC1 i- NaOH 1 M NaOH

A conventional two-compartment electrochemical cell was used. All potentials were quoted vs. a standard saturated calomel reference electrode (SCE). P.A.R. Model 174, 175, 173 and 179 electrochemical instruments were used to obtain cyclic voltammograms that were recorded directly with an x-y recorder at scan rates up to

275

500 mV s-’ or after acquisition by a digital oscilloscope (Tektronix Model 5223) at higher scan rates. Rotating disk and ring-disk voltammograms were obtained with the apparatus and procedures previously described [6,8]. In calculating the Levich limiting currents at rotating disk electrodes, the following parameters were used: kinematic viscosity of water: 0.01 cm s- ‘; diffusion coefficient for dioxygen: 1.7 X 10e5 cm* s-‘; solubility of dioxygen in 1 M NaOH: 1.0 mM. The electrodes were edge plane pyrolytic graphite obtained as rods from Union Carbide Co. (Chicago, IL). The electrodes were attached to glass tubing with heat-shrinkable polyolefin tubing and were polished to a rough finish with No. 600 SIC paper (3 M Co., Minneapolis, MN). A smooth finish was obtained by further polishing with 0.3 pm alumina on a microcloth (Buehler, Ltd, Evanston, IL). The cobalt porphyrin was applied to the electrode by transferring a measured volume of a standard solution of the porphyrin in 1,Zdichloroethane to the electrode surface and allowing the solvent to evaporate. The concentration of the porphyrin in the dichloroethane solutoin was determined spectrophotometrically (c = 1.9 X lo5 M ~ ’ cm- ’ at A max = 393 nm). (II)

RESULTS

(II. I) Voltammetry of the adsorbed porphyrin in the absence of dioxygen The cobalt porphyrin, I, adsorbed on the surface of a pyrolytic graphite electrode exhibits prominent oxidation and reduction waves as shown in Fig. 2A for a series of supporting electrolytes having pH values form 0.3 to 14. Although a variety of anions were employed in the preparation of the (relatively dilute) buffer solutions we observed no evident dependence of the electrochemical responses on the composition of the buffers. The peak currents increase linearly with the potential scan rate and the anodic and cathodic peak potentials are very nearly equal as expected for reactants attached to electode surfaces [9]. The quantities of electroactive I present on the electrode were measured by measuring the areas under the cyclic voltammograms recorded in 1 M NaOH. As the concentration of porphyrin was increased in the aliquots of dichloroethane deposited on the electrode surface, the areas of the voltammograms increased but only until a coverage corresponding to about one monolayer (based on a calculated area of 1.4 nm* per adsorbed porphyrin molecule and an estimated electrode roughness factor of ca. 3) was reached. Further increases in the porphyrin concentration produced only small changes in the magnitudes of the electrochemical responses. This could have resulted from the physical loss of all but the first monolayer of the adsorbed porphyrin when the electrode was immersed in the aqueous solution of NaOH. Alternatively, only a single layer of multilayer deposits might retain its electroactivity. We were unable to distinguish between these two possibilities. In any case, all electrodes were coated by exposing them to more than monolayer quantities of I dissolved in dichloroethane. The quantity of electroactive I present in the resulting coating was 4 X 10 -lo mol cm-*. As is clear from Fig. 2A, the peak current of the voltammograms for the adsorbed

Fig. 2. Cyclic voltammograms various pH: (A) under argon.

of I adsorbed on pyrolytic graphite electrodes m .rupporting electrolges S= IO PA: (B) under dioxygen, S= 100 WA. Scan rate: X3 m\: 5 ’

ol

I are significantly smaller at pH values -=E6 than at pH 14. This does not result from stripping of the porphyrin from the surface by the acidic electrolytes because an electrode that exhibits the response shown for pH 2 will immediately yield the larger response shown for pH 14 when it is transferred to 1 M NaOH. The difference in the magnitude of the peak currents appears at about the same pH where the peak potentials become independent of pH (see below) and it might reflect a more sluggish electrode reaction rate for the aquo than for the hydroxy form of the Co(III) center. But, whatever their origin, the larger and sharper vohammograms obtained in 1 M NaOH were used to measure the quantities of I adsorbed on the electrodes. The peak potentials shift to more positive values as the pH decreases. A plot of

211

Fig. 3. pH dependence

of the formal

potential

for the Co(III/II)

couple

of porphyrin

I.

the average of the cathodic and anodic peak potentials (i.e. the formal potential) vs. pH is linear between pH 6 and 14 with a slope of 65 mV/pH unit (Fig. 3). Below pH 6 the peak potentials become essentially independent of further changes in pH. This behavior is consistent with assignment of the surface wave to reactions (1) and (2) where L represents the porphyrin ligand: LCo”‘OH

+ H + + e = LCd’ OH

LCo”‘OH

2

+ e -- LCo”OH

2

2

0)

(2)

At pH values > 6, reaction (1) is predominant while reaction (2) represents the electrode reaction at lower pH values. This interpretation is supported by the response of the peak potentials to the addition to the supporting electrolyte of imidazole, a base known to coordinate to the axial positions of metalloporphyrins. Figure4 shows the surface waves for the adsorbed porphyrin in the presence and absence of imidazole. The shift of the peak to more negative potentials is in the direction expected for stronger binding of the imidazole by the Co(II1) porphyrin. Similar shifts in reduction potentials are seen when axial bases are coordinated to many metallporphyrins [lO,ll]. It is also noteworthy that the magnitude of the shift in peak potential caused by imidazole is independent of pH in the same range where the peak potentials exhibit a pH dependence in the absence of imidazole (Fig. 3). This is to be expected if the addition of imidazole converts reaction (1) into reaction (3): LCo”‘Im

+ e = LCo”Im

(Im stands for imidazole).

(3)

! + 0.7

I co.5

/ 40.3

E/Vvs

L +o.i

/ 0.1

0.3

S.C.E.

Fig. 4. Effect of imidazole on cyclic addition of imidazole: (-! after 9.2 buffer. Scan rate: 200 mV s ’

voltammogram~ of I adsorbed on graphite. t --------1. Bcfort solution was made I m M in imidazole. Supporting electrol! te: pli

(II.2) Voltammetry in the presence of dioxygen In Fig. 2B the catalyzed reduction of dioxygen at pyrolytic graphite electrodes coated with I is shown at a series of pH values. The corresponding responses at uncoated electrodes were negligible at the potentials where the prominent waves appear in Fig. 2B. At all pH values the catalyzed reducton of 0, proceeds at potentials well separated from the surface waves for the adsorbed catalyst so that these waves could be inspected (at higher current sensitivities) to determine if they are affected by exposure to dioxygen. In no case did the addition of dioxygen produce significant changes in the position or magnitude of the surface waves. At pH 14 the cyclic voltammogram for the reduction of dioxygen at the coated electrode corresponds to a totally reversible, nearly nernstian couple (Fig. 2A): essentially equal anodic and cathodic peak currents are obtained and the peak potentials are separated by 40 mV. The average of the anodic and cathodic peak potentials, 0.03 V vs. SCE, is very close to the standard potential for the O,/OzH couple in 1 h4 NaOH, 0.08 V [ 121. A peak splitting of 29 mV would be expected if the electrode reaction adhered to the Nemst equation. The slightly larger splitting observed may result from both uncompensated resistance in the cell and some small deviation from fully nemstian behavior. The most remarkable feature of the result is the ability of the adsorbed catalyst to transform the graphite surface into an electrode reversible to the 0, /O,H - couple. Mercury electrodes are known to respond reversibly to the 0, /O, H - couple at high pH [ 131 and Fig. 5 compares the responses of a mercury electrode with that of the graphite electrode coated with I. Their equivalence is apparent. As the pH of the supporting electrolyte is decreased, the separation of the cathodic and anodic peak potentials for the 0, /O,H - couple increases (Fig. 2B). The increasing separation results primarily from a shift of the anodic peak to more

279

A

010 I! 20pn

0.7

+0.5

+0.3

E/Vvs

+o.i

-0.1

-0.3

-Of

SC E.

I

0.2v

Fig 5. Cyclic voltammograms of 1.0 mM H202 in 0.5 M NaOH recorded at: (A) 0.032 cm’ hanging mercury drop electrode; (B) 0.17 cm* graphite electrode coated with I. Scan rate: 100 mV s- ‘; initial potential: -0.5 V. Fig. 6. Effect of imidazole on cyclic voltammograms with I. (A) Dioxygen saturated solution before imidazole. Supporting electrolyte: pH 10.8 buffer;

for dioxygen reduction addition of imidazole; scan rate: 200 mV s-‘.

at graphite electrodes coated (B) after addition of I mM

positive values while the cathodic peak remains relatively fixed down to pH 7. The magnitude of the anodic peak current also diminishes at pH values below ca. 12. This is also observed with solutions of H,O, in the absence of dioxygen. It seems significant that the decrease in anodic current commences close to the pK, of H,O, (11.7) where the O,H- anions that presumably coordinate to the Co(I1) centers of the catalyst in order to be oxidized are converted into much less nucleophilic H,O, molecules. A similar diminution of the H202 oxidation current results from the addition of imidazole at fixed pH (Fig. 6). Note that the axial base also shifts the wave for the catalyzed reduction of dioxygen to more negative potentials. Both responses are in the direction expected if the imidazole were competing with H,O, and 0, for a coordination site on the cobalt center. It is also not surprising that the axial base is apparently more successful in competing with H,O, than with 0, for coordination to Co(I1).

(II.3)

Effects

of increasing

scan rates on the currerlr

respon.m

Figure 7 shows how the voltammograms for the adsorbed cataly ‘t and the OJO,H couple respond to large increases in the potential scan rate. The peak li that for current of the cobalt catalyst increases more rapidly with scan rate than does the dioxygen reduction, as expected for an attached reactant. However. the dioxygen reduction wave becomes a flat plateau at scan rates > 10 V 9 ‘. This is the behavior expected if the electroreduction step is preceded by a relatively rapid. potentialindependent chemical reaction. Further evidence of the intervention of such a reaction will be presented m Section 11.5, where the rotating disk studies are described. The small separation in peak potentials and the undistorted shapes of the waves for the cobalt catalyst at the highest scan rates (Fig. 7) show that it can be cycled between its two oxidation states at very high rates at potentials close to its formal potential.

Fig. 7. Cyclic voltammograms (-). Argon-saturated

couple xn I M NaOH at high xan for adsorbed I and the Oz / 0, H solutions: ( ___-1 dioxygen-saturated solutions. Initial potential:

rates. 0.4 V

281

(II.4) Anodic peaks during scans toward more negative potentials A revealing feature in the cyclic voltammograms for the 0, /O,H - couple at catalyst-coated electrodes can be observed in the vicinity of the surface wave for the adsorbed catalyst. Curve A in Fig. 8 shows the complete voltammogram recorded at a low current sensitivity so that the entire reversible 0, /O, H - response is observed. CurveB, recorded at a 10 times higher sensitivity, shows the response from the attached catalyst when the electrode potential is cycled repeatedly but restricted to values ahead of the dioxygen reduction so that no H,O, is formed. Curve C shows the response obtained when the potential is cycled repeatedly in an extended range where the reduction of dioxygen to H,O, proceeds. On the second and all subsequent scans to more negative potentials a prominent oxidation peak appears near the potential where the cobalt catalyst is converted from Co(II1) to Co(I1). Curve D shows how the anodic peak, once formed, decays away if the potential range is again restricted to values ahead of the dioxygen reduction. This anodic peak that appears during scans to more negative potentials seems clearly to originate from the oxidation of H,O, that is generated at the electrode while the electrode is at potentials on the dioxygen reduction wave. Much of the H,O, is reoxidized at potentials positive of the wave for the adsorbed catalyst. However, when the potential passes over the catalyst wave, converting the Co(I1) to its catalytically unreactive Co(II1) state, the oxidation of H,O, can no longer proceed. Thus, the H,O, that continues to diffuse to the electrode surface accumulates there until the

A

C

Fig. 8. Cyclic voltammograms of adsorbed I in dioxygen-saturated 0.5 M NaOH. (A) Low current sensitivity, potential scanned over the O2 /HO; wave; (B) high current sensitivity, potential scanned only to the foot of the Or/HO; wave; (C) high current sensitivity, potential scanned repeatedly over the O,/HO; wave; (D) high current sensitivity; potential scanned once over the O/HO; wave and then repeatedly to the foot of the wave. Scan rate: 100 mV s-‘. Current sensitivity: (A) S= 100 PA; (B, C, D) S= IO pA. Initial potential: 0.4 V.

2x2

electrode potential is returned to the value where the Co(III) is agam reduced to Co(I1). At that point the accumulated H,O, can again be oxidized and the anodic peak results. The position of the anodic peak shifts with pH in the same way that the peak potential of the wave for the cobalt catalyst shifts (Fig. 3). Solutions of H,OZ in the absence of dioxygen also exhibit the unusual anodic wave. An initial scan to more positive potentials produces a normal anodic current peak but the current in the “tail” of the wave decays abnormally in the vicinity of the cobalt catalyst wave. The subsequent scan to more negative potentials exhibits a prominent anodic peak at the same potential at which it appears in Fig. 8. The voltammetric behavior described here is somewhat unusual but the explanation proposed to account for it is not. Much higher activities of Co(I1) than of Co(II1) complexes have also been reported in the catalyzed disproportionation of hydrogen peroxide [14] and it is not surprising to see the same difference in reactivity reflected in the catalysis of the oxidation of H,O, by the cobalt porphyrin catalyst. (IIS)

Voltammetry at catalyst-coated

rotating disk electrode,7

Figure 9A shows a set of current-potential curves for the reduction of dioxygen at a rotating disk electrode coated with the cobalt catalyst in 1 M NaOH. The corresponding Levich plot (151 of limiting current vs. (rotation rate)‘/” and the line calculated from the Levich equation [IS] for the two-electron reduction of dioxygen are shown in Fig. 9B. The evident deviations from the Levich line point to control of the limiting currents by a chemical step that precedes the reduction Kinetic analysis

E/V vs S.C.E. Fig. 9. (A) Current-potential curves for the reduction of O2 in 1 M NaOH at a rotating graphite di& electrode coated with I. Electrode rotation rates in ‘pm are indicated on each curve. The electrode potential was scanned at 8.3 mV s ‘. (B) Levich. plot of the rotating disk limiting currents vs. (rotation rate)‘12. (- - - - -) The calculated response for the mass-transfer-limited, two-electron reduction of dioxygen.

283 l.co -

7-: E

0.75

-

/

/

/ I 0.025 P/(rpmY

O/ 0

Fig.

IO. Kouteck$-Levich

plot

I 0.05

of the current-rotation

rate data

from

Fig. 9; (- - - - - -) as in Fig. 9.

of such systems is facilitated by means of Koutecky-Levich plots of (limiting current) - ’ vs. (rotation rate) -‘I* [8,16]. A typical plot is shown in Fig. 10. These plots are interpreted on the basis of eqn. (4): l/i,i,

= l/i,,,

-t l/i,

(4)

where ilim is the measured limiting current. ilev, the Levich current, and i,, kinetic current, are defined by eqns. (5) and (6) respectively:

the

ilev = 0.62nFAD213 u-‘16 cb ail2

(5)

where D and cb are the diffusion coefficient and bulk concentration of dioxygen, respectively, Y is the kinematic viscosity of water, w is the electrode rotation rate and the other symbols have their usual significance. i, = 103nFAkI’cb

(6)

where k is the rate constant (M-’ SC’) governing the reaction of the catalyst with dioxygen and Qmol cm-*) is the quantity of catalyst on the electrode that participates in catalyzing the reaction. The slope of the solid line shown in Fig. 10 matches that of the dashed line that was calculated from eqn. (5) for the two-electron reduction of dioxygen showing that hydrogen peroxide is the product of the electrode reaction. The intercept of the solid line in Fig. 10 provides a measure of (kl?). Table 1 summarizes the values of (kI’) and k evaluated from Koutecky-Levich plots for the catalyzed reduction of dioxygen at a series of pH values. All of the electrodes employed to obtain the data in Table 1 were coated by the procedure

2x4 TABLE

I

Rate constants obtained from electrodes coated with I

the Intercepts

of KouteckCLcwch

plots

Ior the reduction

of 0, ,I! graphltc

described above to produce the limiting porphyrin coverage of 4 x 10 “’ mol cm-. The measured values of k are not affected by changes in pH except at pH 14 where a somewhat smaller value of k results. It is noteworthy that the values of k listed in Table 1 are of the same magnitude as the previously reported rate constants governing the kinetics of the coordination of dioxygen to a variety of Co(H) complexes in aqueous solution [ 171. Analogous experiments in which hydrogen perioxide is oxidized at the catalystcoated rotating disk electrode yielded the current potential curves shown in Fig. 1 I. The de&e in the limiting currents at potentials where the Co(I1) catalyst is oxidized to Co(II1) is evident. Intercepts of Koutecky-Levich plots for the oxidation H,Oz were used to evaluate the rate constants listed in Table 2. Measurements were limited to pH 12 or higher because the catalyst appeared to lose its potency to catalyze the

Fig. 1 I. Current-potential disk electrode coated with

curves for the oxidation of I mM I. Initial potential: - 0.5 V Other

0,H in I M NaOH at a graphite conditions as in Fig. 9.

rotating

285 TABLE

2

Rate constants obtained from graphite electrodes coated with PH

(kT)/cm

12 14

0.022 0.034

s-’

the intercepts I

of KouteckL-Levich

plots

for

the oxidation

of H,O,

at

k/M-Is-’ 6.0X 9.4x

IO4 IO4

oxidation at lower pH values. The values of rate constants in Table 2 are similar to those in Table 1 suggesting a common rate-limiting step for both dioxygen reduction and hydrogen peroxide oxidation. (II. 6) Voltammetry at rotating ring-disk electrode The rotating disk results already described were entirely consistent with the catalyzed reduction of dioxygen proceeding quantitatively to hydrogen peroxide. However, attempts to confirm this by means of a rotating graphite disk-platinum ring electrode were unsuccessful. Only 60-70% of the H,O, believed to be generated at the catalyst-coated disk was detected at the platinum ring electrode held at potentials where H202 is oxidized. This discrepancy was traced to a fairly rapid “poisoning” of the platinum ring electrode toward H,O, oxidation in most of the electrolytes employed. We have made similar observations in previous ring-disk experiments with the dioxygen-hydrogen peroxide system [18] where repeated cleaning and “activation” of the platinum ring electrodes were required to assure quantitative oxidation of hydrogen peroxide reaching the ring. Such remedies were only partially successful with the electrolytes employed in the present experiments and we abandoned the use of the ring-disk electrode to measure the quantities of H,O, generated at the disk electrode. The best evidence that the catalyzed reduction of dioxygen produces only hydrogen peroxide is the slope of the Koutecky-Levich plot (Fig. 10) which corresponds closely to the calculated value for n = 2. (III)

DISCUSSION

One of the most striking features in the behavior of the cobalt porphyrin catalyst is the significant separation between the potentials where the catalyst is oxidized and reduced and the potentials where it exhibits its catalytic activity. The catalyst is reduced at potentials well ahead of those where dioxygen is reduced, suggesting that the formation of a Co(II)-0, adduct is a necessary but not a sufficient condition for the electroreduction of dioxygen to ensue. The potential of the electrode must also be made sufficiently negative to reduce the Co(II)-0, adduct. At pH 14, where the catalyst endows the electrode with full reversibility toward the 0, /O, H - couple, a possible reaction sequence for the catalyzed reduction is reaction (1) followed by the

steps listed in Scheme 1 Scheme I LCo” + 0 2‘ = Lc’o” 0 I,/

(A )

LCo”0

i Bj

t 2e --1 -- 1 Co”O’ 2

2

LCol~OZ2 LCo”0

2

H

-I-H - we LC#O 2I-$

i C‘”)

- -_LCO” + 0 2H

(I-),

All of the steps in this sequence are reversible but at high rates of reaction, the binding of the substrate to the catalyst. i.e. the forward direction of step (A) or the reverse direction of step (D), become rate-limiting. The insensitivity of the peak potentials of the reversible Co(III)/(II) couple of the adsorbed catalyst to the addition of 0, or H,O, (Figs. 2, 6) is strong evidence that the equilibrium constants governing the formation of the Co(I1) adducts of dioxygen and 0,H are not large, probably < 1 M ‘. [The alternative possibility, that Co(U) and Co(II1) bind 0, (and 0,H - ) equally strongly, seems much less likely.] Thus, even if the proposed cobalt-dioxygen adduct were best regarded as a complex of Co(II1) and superoxide, most of the cobalt centers in the adsorbed catalyst remain as Co(I1) throughout the catalytic cycle. The rate constants listed in Tables 1 and 2 for pH 14 give the specific rate of the forward direction of step (A) and the reverse direction of step (D) in Schemel. respectively. It is unlikely that the rate could be limited by the decomposition of the products of the catalyzed reactions, i.e. by step (D) or the reverse of step (A), because, as has already been argued, the equilibrium constants for steps (A) and (D) must be rather small and rather large, respectively. The nearly equal values of the rate constants argue for a common rate-determining step such as the loss of the ligand, probably H,O, occupying the site of binding of the incoming 0, or 0,H ~- molecule to the cobalt porphyrin. The same argument has been used to account for the similarity in the rate of binding of dioxygen to a variety of different Co(I1) complexes [ 17,191. At pH values between 7 and 1 I, the electrode process is less reversible because the oxidation of H,Oz moves to more positive potentials while the reduction of 0, continues to proceed at virtually the same potential (Fig. 2B). The dioxygen reduction probably continues to follow the same sequence (steps A-D) in the forward direction in this pH range while the oxidation of hydrogen peroxide must follow a somewhat different course such as that shown in Scheme II: Scheme II I-i*02 G=H - +O,H~ HO,LCo”0,

+ LCo” ;=rLCon02H H

-+LCo”0,

LCO” 0 2 = LCO” -t 0,

+ H i -t 2e

287 TABLE Slopes

3 of plots

of log[ i/( i,i,

PH

Slope/mV

0.3 3.0 7.0 10.8 14

122 150 115 108 30

-i)]

decade

vs. E for the reduction

of 0,

at graphite

disk electrodes

coated

with

I

- ’

This sequence accounts for the shift in the oxidation wave to more positive potentials as the pH decreases, because the concentration of the proposed catalytically active complex, LCd’O, H - , is linked directly to the pH through steps (E) and (F) (the pK, of H,O, is 11.7). At pH values below ca. 7, the dioxygen reduction peak begins to exhibit a pH dependence (Figs. 2 and 3). This can be explained if steps (B) and (C) in Scheme I are replaced by the following two steps: LCo”0, LCo”02H+

+ H + = LCo”O,H +2e-

+

--) LCo”O,H-

(J) WI

The oxidation of hydrogen peroxide at pH values < 7 could continue to proceed according to Scheme II. On the rising portions of current-potential curves for dioxygen reduction at rotating disk electrodes, step (C) rather than step (A) in Scheme1 would be rate-limiting. Tafel plots of log[i/(i,i, - i)] vs. E [20] prepared from the rotating disk data recorded at the foot of the waves (E > E,,,) were linear with slopes near 120 mV/decade at pH < 11 (Table 3) indicating that the addition of the first electron limits the rate of reduction of the Co(II)-dioxygen adduct. At pH 14, the slope decreased to 30 mV/decade as expected when the 0, /O,H - couple approached nernstian equilibrium. Behavior of related catalysts Coatings of both cobalt octaethylporphyrin and cobalt tetraphenylporphyrin on graphite electrodes behave similarly to the cobalt porphyrin described in this report. In particular, the cobalt redox waves appear well ahead of the catalyzed dioxygen reduction and hydrogen peroxide oxidation [21]. Yeager and co-workers [22] have described the catalyzed reduction of dioxygen to hydrogen peroxide by a cobalt phthalocyaninetetrasulfonate adsorbed on graphite electrodes. The behavior closely resembied that described in this report, except that the separation between the Co(III)/(II) couple and the catalyzed reduction was much larger, almost 1 V. Osa and co-workers [23], who examined the catalysis of dioxygen reduction by cobalt

tetra-a-aminophenylporphyrin in acidic electrolytes, also observed a separation between the catalyst wave and the dioxygen reduction wave. Recently. we described the catalysis of the reduction of dioxygen by a soluble cobalt macrocyclic complex [IS]. in which there was also a significant separation between the cobalt-centered electrochemistry and the catalyzed reduction of dioxygen. The only reported exceptions to this pattern that we are aware of are the catalysis of dioxygen reduction by cobalt tetrapyridyl porphyrin [24] and the cofacial dicobalt porphyrin [5,6] for both of which the catalytic mechanisms that have been proposed imply a close connection between the cobalt-centered and dioxygen reduction waves. In the case of the cofacial dicobalt porphyrin, surface waves for the adsorbed catalyst are very difficult to observe in the highly acid electrolytes in which the catalyst has been examined up to now. Experiments with this catalyst in electrolytes at higher pH are now under way in the hope of establishing more clearly the mode of its catalytic activity. ACKNOWLEDGEMENTS

This work was supported by the National Science Foundation. We are grateful to Professor J.P. Collman and members of his research group for synthesizing the cobalt porphyrin. REFERENCES 1 E. Yeager in A.D. Franklin (Ed.), Electrocatalysta on Non-Metallic Surfaces, N.B.S. Special Publication No. 455. (1976) p. 203. 2 E. Yeager. J. Zagal, B.Z. Nikolic and R.R. i\dzic, in Proceedmgs of the Third Symposium on Electrode Processes, The Electrochemical Society, lY79. p. 436. 3 H. Behret. H. Binder. W. Clauberg and G. Sandstede. Proceedings of Symposium on Electrode Materials and Processes for Energy Conversion and Storage, The Electrochemical Society, 1977. p 519. 4 F. van den Brink. E. Barendrecht and W. Visscher, J. R. Neth. Chem. Sot.. 99 ( 1980) 253. 5 J.P. Collman. M. Marrocco. P. Denisevich. C. Koval and F.C. Anson. J. Electroanal. Chem., 101 (1979) 117. 6 J.P. Collman. P. Denisevich, Y. Konai, M. Marrocco. C Koval and F c‘ Anson. J Am. Chem. Sot . 103 (1980) 6027. 7 J.P. Coilman, F.C. Anson, S. Benscomc, A. Chang. T. Collins, P. Denisevich, E. Evett. T. Geiger, J Ibers, G. Jameson. Y. Konai. C, Koval, K. Meier. R. Oakley. R. Reitman, E. Schmittou. J. Sessler, in B.M. Trost and C.R. Hutchinson (Eds.), Organic Synthesis Todav and Tomorrow. Pergamon Press. New York, 1981. 8 N. Oyama and F.C. Anson. Anal. Chem., 52 ( 1Y80) I 192. 9 E. Laviron, Bull. Sot. Chem. Fr., (1967) 3717: A.P. Brown and F.C. Anson, Anal. Chem.. 49 (1977) 1589. IO J. Manassen, Isr. J. Chem., 12 (1974) 1059. I I L.A. Truxillo and D.G. Davis. Anal. Chem., 47 (1975) 2260. 12 W.M. Latimer. Oxidation Potentials, Prentice-Hall, New York, 1952. 13 V.S. Bagotzkyand I.E. Jablokowa. J. Phys. Chem. U.S.S.R.. 27 (1953) 1665: J. Koryta, Collect. C‘zech. Chem. Commun., 18 (1953) 206. 14 P. Waldmeir and H. Sigel, Chimia, 24 (1970) 195; Inorg. Chim. Acta. 5 (197 I) 659. 15 V.G. Levich, Physicochemical Hydrodynamics, Prentice-Hall, Engiewood Cliffs, New York. 1962.

289 16 17 I8 19 20 2I 22 23 24

J. Koutecky and V.C. Levich, Zh. Fiz. Khim., 32 (1956) 1565. R.D. Jones, D.A. Summerville and F. Basolo, Chem. Rev., 79 (1979) 139. T. Geiger and F.C. Anson, J. Am. Chem. Sot., 103 (198 I) 7489. R.G. Wilkins,Adv. Chem. Ser., lOO(l971) Ill. D. Jahn and W. Vielstich, J. Electrochem. Sot., 109 (1962) 849. R.R. Durand, unpublished results. J. Zagal, R.K. Sen and E. Yeager, J. Electroanal. Chem., 83 (1977) 207. N. Kobayashi, T. Matsue, M. Fujihira and T. Osa, J. Electroanal. Chem., 103 (1979) A. Bettleheim. R.J.H. Chan and T. Kuwana, J. Electroanal. Chem., 99 (1979) 391.

427.