Photoelectrochemical reaction mechanisms The reduction of benzophenone and 4-halobenzophenones

JOURI,UlI. OF

ELSEVIER

Journal of Electroanalytical Chemistry 424 (1997) !65-171

Photoelectrochemical reaction mechanisms The reduction of benzophenone and 4-halobenzophenones Wayne M. Leslie a, Richard G. Compton ~'*, Toomas Silk b a Physical and Theoretical Chemistry Laboratory, Oxford Unit'ersity, South Parks Road, Oxfilrd OXI 3QZ, UK h Faculty of Physics and Chemistry, Tartu Unirersity. Jakobi 2, EE2400 Tartu. Estonia

Received 27 March 1996; revised 19 July 1996

Abstract The dual photo- and electro-chemical activation of benzophenone and the 4-halobenzophenones, 4-X-C6H4COOPh (X = F, CI) in acetonitrile solution using channel electrode methodology is reported. In the cases of benzophenone and 4-fluorobenzophenone one-electron re0uction leads to the formation of the corresponding radical anions which are stable in the absence of light. Irradiation using light of wavelength 325 nm induced significant photocurrents to pass; this was interpreted in terms of a light-induced disproportionation reaction, initially generating the parent material and the corresponding di-anion, with the latter ultimately leading to the formation of the corresponding alcohols, C6Hs-CH(OH)-Ph and F-C6H4-CH(OH)-Ph. The processes are shown to follow DISP2 mechanisms, in which there is both a significant fraction of back reaction (conproportionation) and also quenching of the photo-excited state of the radical anion by the parent material, in contrast, the reduction of 4-chlorobenzophenone at - 1.61 V (vs. Ag) was found to be accompanied by only tiny photo-currents on irradiation with light of wavelength 325 nm.

Keywords: Benzophenone;Halobenzophenones;Photoelectrochemistry;Reductionmechanism

1. Introduction It has been argued [1] that electrochemical transformations might be beneficially enhanced, or more usefully modified, through the introduction of light in the near vicinity of the electrode, so affording the possibility of the dual activation of selected species. Specifically, it may be speculated that the sequential use of electrochemical and photochemical activation of organic and organometallic molecules may lead to a wealth of hitherto unsuspected chemistry, including novel mechanistic pathways, the discovery of new compounds associated with unusual reactive intermediates, and ultimately new synthetic routes. In our previous studies we advocated the channel electrode approach (Fig. 1) for the kinetic and mechanistic investigation of photoeiectrochemical pathways [1,2] and reported results in a variety of systems, including aryl halides [3], cobalt dithiocarbamates [4], bis(cyclopentadienyl) metal dichlorides [5], iron phosphine complexes [6] and heteropoly anion clusters [7]. In particular, the varia-

" Corresponding author. 0022-0728/97/$17.00 © 1997 Elsevier Science S.A. All rights reserved. Pll S0022-072 8(96)04904-2

tion of the channel electrode photocurrent with mass transport (solution flow rate), coupled to spectroelectrochemical measurements, such as ESR or fluorescence [1,2], made

Metalfoil w~electrode

(a)cover

ii

~...°°°

hv

....." ...........

0

xo

channelunit

S-------:-'".'?----:-"-'"

flow

........

:>

Fig. I. (a) A practical channel flow cell for mechanistic photo-electrochemical studies. (b) A schematic diagram which defines the electrode geometry.

166

W.M. Leslie et al. / Journal of Electroanalytical Chemistry 424 (1997) 165-171

using the same cell have been shown to be particularly powerful in respect of the characterisation of photoelectrochemical mechanisms. In this paper we examine the photo-electroreduction of first benzophenone (BP)

BP =

o

and second, for comparison, the 4-halobenzophenones, XC6H4COPh (X = F, Ci)

XBP = o

2. E x p e r i m e n t a l

All standard photo-voltammetry experiments were conducted using a platinum channel electrode made of optical quality synthetic silica to standard construction and dimensions [1,2]. Solution (volume) flow rates between 10 -4 and 10-m cm 3 s- m were employed. Working electrodes were fabricated from Pt foils (purity of 99.95%, thickness 0.025 mm) of approximate size 4 mm X 4 ram, supplied by Goodfellow Advanced Materials. Precise electrode dimensions were determined using a travelling microscope. A silver pseudo reference electrode (Ag) was positioned in the flow system upstream, and a platinum gauze counter electrode located downstream, of the channel electrode. Electrochemical measurements were made using an Oxford Electrodes potentiostat modified to boost the counter electrode voltage (up to 200V). Other methodological details were as described previously [1-3]. Irradiation was provided by an Omnichrome continuous wave 3112XM He-Cd source (Omnichrome, Chino, CA) which gave light of wavelength 325 nm at 20mW absolute power with a minimum beam diameter of 1.6 mm. The laser was used in conjunction with a beam expander (Optics for Research, Caldwell, NJ) which gave a 25-fold increase in beam area and a radiative power of 55 mWcm -2. Photocurrents were made by comparison of measurements of the current flowing in the dark and in the light at the same potential using a digital voltmeter to record the output from the currentto-voltage converter of the potentiostat. Simultaneous photo-electrochemical EPR experiments used a channel flow cell carefully positioned in the TEIo2

cavity of an X-band (9.0-10.0 GHz) Bruker ER200D spectrometer as previously described [1,2]. Experiments were performed using solutions of the electroactive substrate (ca. 10 -4 to 10 -3 M) in dried [8] acetonitrile (Fisons, dried, distilled) solution containing 0.1 M (recrystailised) tetrabutylammonium perchlorate (TBAP) (Kodak) as supporting electrolyte. Solutions were purged of oxygen by outgassing with pre-purified argon prior to electrolysis. p-Benzophenone, 4-fluorobenzophenone (FBP) and 4chlorobenzophenone (CBP) were used as received (greater than 99%) from Aldrich.

3. R e s u l t s a n d d i s c u s s i o n

We first consider experiments conducted on the reduction of FBP in acetonitrile solution containing 0. l M TBAP as supporting electrolyte. Channel flow cell voltammetry using platinum electrodes revealed a one-electron reduction at - 1 . 6 4 V (vs. Ag) and measurements of the transport-limited current as a function of solution flow rate gave a diffusion coefficient of 1.5 × 10-Scm 2 s -~. All these observations are in good agreement with literature reports [9,10] and additionally indicate the formation of the radical anion as a result of this electrode process FBP + e---* [FBP]'This was confirmed by in situ esr measurements, which revealed the spectrum shown in Fig. 2 attributed to [FBP]-. It has been shown [1 l] that for a kinetically stable radical formed at the electrode surface that the variation of signal intensity S with flow rate Vf is given by StX I V f 2/3

where I is the electrode current. A plot of log l0 ( S / I ) vs. loglo Vf had a slope of - 2 / 3 , which confirms the ~netic stability of [FBP]'- on the channel electrode time scale.

Fig. 2. in situ esr spectra obtained from the electrolysis of 1.0mM FBP in acetonitrile solution using a channel cell with negligible flow rates. The working electrode voltage was held at - 1.68 V (vs. Ag).

167

W.M. Leslie et al. / Journal of Electroanalytical Chemistry 424 (1997) 165-171

Experiments were next conducted using irradiation from an He-Cd laser. The transport-limited currents flowing under dark /dark and light lh,, conditions were recorded as a function of electrolyte flow rate Vf in the range 0.003 to 0.03 cm 3 s-l and values of the effective number of electrons transferred, deduced from the following equation: Neff = thv//dark

( 1)

Neff values in the range 1.04 to 1.09 were obtained from a solution containing 1.04 mM FBP. These data were analysed using an established protocol [1,2] in which Neff is related through an appropriate "working curve' to a dimensionless rate constant of the form (n-l) KMeeh = kMeeh[Alau, k (4h 4 x e2 d 2/ 9 V f 2O) ':3

Light absorption B + hv ~ C Disproportionation C + B -o A + D Product formation D ~ products De-excitation C --* B

(i) (ii) (iii) (iv) (v)

(3)

If, however, we add the following two steps Conproportionation A + D ~ B + B Quenching C + A --* B + A

Ot

a [A]

i~[A] 0y 2 - vx ~ O x + k°ii)[B][C] - kt"i)[A][D]

=O

,

(4) o[B]

where B = A - , C -- [ A - ] " and D -- Az-. De-excitation may ~ e u r , for example, through fluorescence or through radiationless internal conversion. In this case kMech "- kDlsP2 -"- k(ii)k(iii)/k(v)

a[A]

(2)

where (Fig. 1) 2h is the channel depth, d the channel width and x e the electrode length. D is the diffusion coefficient of the substrate material A (assumed equal to that of any other mechanistically significant species) and kMech is the rate constant for the rate limiting step of the mechanism under consideration. The exponent n is also mechanism specific; for example, in the well worked cases of ECE and DISP1 mechanisms n = 0, whereas for a DISP2 process n = 1 [1,2]. The usual protocol for interpreting Neff-V f data is to convert the experimental Neff values into corresponding K values for selected mechanisms and then plot K against Vf2/3; the selection of a consistent mechanism is indicated by a straight line plot which passes through the origin [1,2]. The above procedure applied to the FBP photocurrent data indicated that the latter was inconsistent with either ECE, DISPI or EC' mechanisms. However, analysis using a DISP2 working curve gave the good straight line plots. However, on repeating this exercise for different substrate concentrations (0.965 < [FBP]a,tk < 3.12raM) the resulting values of K were found to have an inverse dependence on [FBP] rather than the direct relationship predicted by Eq. (2) for a DISP2 mechanism. This surprising behaviour was rationalised by considering first a general photo-DISP2 mechanism for the reduction of an arbitrary substrate A: Reduction A + e - ~ B

we have what is, in effect, a self-inhibiting disproportionation (SID) mechanism [12], in the sense that the steps in Eq. (vi) and Eq. (vii) remove species C and D respectively so inhibiting the overall transformation of A into D (or products). The extension from the familiar and well-worked E C E / D I S P systems should be apparent [13,14]. The convective-diffusion equations describing the distributions of A, B, C and D in time t and space ( x , y ) within a channel electrode flow cell are

(vi) (vii)

~

'

8t

o:[s] --" D

~

0y 2

o[s] vx Ox +

k,v,[cl + k,.i,[A][¢l

+ 2kt~i)[A][D] - koii)[a][c ] - k(ii)[B ]

o[c] o [c] • = o .... Ot

Oy 2

0[c] Ox

+

-- k0i0[B][C ] - k(vii)[A][C] O[D] Ot

82[D] = D - -

By 2

(5)

(6)

8[D] "x~ +k(iii)[B][C] 0x

- k(vi)[A][D] - k,i~)[D]

(7)

where D is the diffusion coefficient, k(,,) is the rate constant for the particular reaction in Eq. (i)-Eq. (vii), and the Cartesian coordinates x and y can be understood with reference to Fig. 1. v x is the solution velocity in the x-direction; the components in the y- and z-directions are zero. Given laminar flow conditions and that a sufficiently long lead-in length exists upstream of the electrodes, so as to allow the full development of Poiseuille flow, then vx is parabolic [15]:

.,=.o[I

(8)

where vo is the solution velocity at the centre of the channel. Eqs. (4)-(7) assume that axial diffusion effects may be neglected, which is valid provided the electrodes considered are other than microelectrodes [16]. Further simplifications are possible. The first corresponds to the assumption of steady-state experiments so that

O[L] Ot' = 0

(9)

where I, = A, B, C or D. Second, we apply the steady-state approximation to species C /(ii)[B] -~- k(v)[C ] @ k(iii)[B][C] q-/(vii)IA][C]

(10)

168

W.M. Leslie et al. / Journal of Electroanalytical Chemistry 424 (1997) 165-171

so that

k(.)[B]

[c]: k.) + k(..[B] + k.,,)[M

(11)

Similarly, for D

k(.,,[B][C] = k.i,[A][D] + k,,v)[D]

(12)

giving

k(iii)[B][C]

[D] =

(13)

+ k,.,[A]

k,v,

Thus the mass transport Eqs. (4)-(7) become 0[A]

02[A] =

Of

D

~

0y 2

0[A] Vx

0x k(,)k(,i)ko,,)[B ]2

+

(k, iv , + k(vi,[A]) (k`v, + k, iii,[B] + k, vii,[A])

(14)

a[B]

0 [8] -

at

D - -

ay 2

a[B] vx

ax

2 k(ii)k(iii)k(iv )[B]2 (k(iv) + k(vi)[A])(k(v) + k(iii)[B ] +

k, vii)[A])

(15)

0[c] at

=

a[D] at

D02[c]

0[c]

ay 2

ax

a [D] - D ~

ay 2

(16)

a[D] vx

ax

(17)

We consider the specific cases where k(vii,[A ] ::~ k, vii ), k, iii,[B ]

(18)

attd

k(iv, ac k(vi)[A ]

state problems in the channel electrode geometry [17-19]. Refs. [17-19] include full details of the necessary computations, and since the adaptations required for the problem of interest are minimal the interested reader is directed to the literature for the appropriate detail. The relevant SIDQC working curve, deduced via simulations, is shown in Fig. 3, along with that for a pure DISP2 process [1,2] for the purposes of comparison. Note that the latter has been shifted appreciably along the x-axis for the purposes of comparison. The working curve, other than for values of Neff close to 2, has a similar shape to the working curve for the pure DISP2 mechanism. Thus measurements of Neff as a function of flow rate alone do not distinguish the two possibilities. However, since the definition of KDisr2 and Ksm.QC differ in their dependence on [A]eumk the variation of the concentration of the parent species permits this distinction; for DISP2 K depends directly on [A]eulk, whereas for SID-QC there is an inverse relationship. Thus the experimental variation of [A]e,lk in conjunction with working curve analysis permits the required mechanistic determination. Returning to the experimental Neu-Vf data we now see that observations reported above are consistent with the operation of a SID-QC mechanism for the reduction of FBP. This is emphasised by Figs. 4 and 5 which show the analysis as outlined above. The good straight lines in Fig. 4, together with the inverse dependence of K on [FBP]eulk, provide good evidence for the operation of a SID-QC mechanism. The slope of Fig. 5 permits the inference that the ratio of rate constants kSlD_QC ---- 5.05 X 10-12 mol cm-3 s- ' at the light conditions cited. We consider next BP as a substrate. Current-voltage curves measured at the platinum channel electrode showed a one-electron reduction at - 1 . 7 2 V (vs. Ag), and measurements of the transport-limited current as a function of flow rate permitted the inference of a value of 1.6 × 10 -5 cm 2 s -m for the diffusion coefficient. In situ esr measurements revealed the spectrum shown in Fig. 6. All

(19) 2.0

corresponding to quenching and conproportionation as the dominant fates of C and D respectively. Dimensionless analysis [1,2] then predicts that Neu depends on the parameter

@

1.8

KS,D.QC= (ksm.QC/[ A ]sulk)(4h4x2d2/9Vf2D) m/3 (20) Z

where

ks,o Qc =

A

1.6 A

1.4

(21)

in which the label SID-QC has been employed to describe the SID mechanism with both quenching (Q) and conpropomtionation (C). The relationship between Neff and KSlD.QC for the SID-QC mechanism can be deduced by solving Eqs. (14)(17) using simulations based on the backwards implicit method previously advocated for the solution of steady-

oA

1.2 1.0 -6

.^A', A -5

OA I -4

i -3

.. i -2

i -1

0

10glo(KsIDa,a-0C) Fig. 3. A workingcurve for the SID-QCmechanism(O). A pure DISP2 curve shiftedalongthe x-axis to exactlymatchthe SID-QCcurveat low K is shownfor comparison(A).

W.M. Leslie et at/Journal of Electroanalytical Chemistry 424 (1997) 165-171

16 Ia)

3.5

14 Od

12

3.0

oJ 2.5

E

ow

169

8

. . . . . i T . "j " •

v

6

o "~"

4-

vo

2.0-

o

1.5

(/)

o

1.0 = 0.5-

2C 0

I

I

I

10

20

30

I

I

40

so

0.0 0

60

I

I

I

I

I

2

4

6

8

10

(Flow rate/cmas 1)':'13

(b)

2

[FBP]'I/IO s cm3mol -t

,/~

Fig. 5. Plots of the slopes of the straight line fits in Fig. 3 against reciprocal

concentratio,~.

6 0

o 0

the formation of the radical anion. Measurements of Neff as a function of flow rate were made and found to be in the ran~ 1.04 < Neff <( 1.10 for 0.003 < Vf < 0.03. Analysis, as above, again indicated the operation of a SID-QC mechanism. Fig. 7 shows the dependence of K on (flow rate) -2/3 and Fig. 8 the inverse dependence of K on [BP]aulk. The slope of Fig. 7 gives kSlD.QC = 6.25 × 10-12 mol cm -3 s-~ in this case at the cited light conditions. The above suggests that both BP and FBP undergo photoeiectroreduction via a SID-QC mechanism generating the corresponding alcohols, Ph2CHOH and F C 6H4CHOHPh, after protonation of A2- [20]. We now turn to the reduction of CBP in acetonitrile. Under channel electrode conditions experiments on this showed a reduction wave with a half-wave potential of -1.61 V (vs. Ag) and measurements of the transport-

4

tD

o

2

t

0

.I

I

10

0

I

I

20

I

I

I

30

40

(Flow rate/cmasl) 2/a

(c)

.j-

q

_a ¢/)

°_o 0.0 0

i

i

t

~

i

10

20

30

40

50

60

(Flow rate/cm3s "1)'~'/3 Fig. 4. Plots of dimensionless rate constant K (as deduced from the SID-QC working curve) against Vf 2/3 for concentrations of FBP of (a) 1.06 ~ l , (b) 1.98 ~ and (c) 3.12 mM.

these observations are in good agreement with earlier reports [9,10] and consistent with the generation of the blue radical anion: BP + e---* [BP]'On irradation of the channel electrode with 325 nm light photocurrents were observed at voltages corresponding to

5G

t

Fig. 6. In situ esr spectra obtained from the electrolysis of 1.0mM BP in acetonitrile solution using a channel cell with negligible flow rates. The working electrode voltage was held at - 1.75 V (vs. Ag).

W.M. Leslie et aL / Journal of Electroanalytical Chemistry424 (1997) 165-171

170

4.0

(a)

o 12 -

oJ A "7

~

o,

E

0

v

3.0 2.5 2.0

¢D

e~ o 1.5

8

v

//,//

3.5

16

"

¢D

• e

~

¢/)

o

r,, O

1.0

T--

0.5 I

0 0

I

I

10

I

20

I

I

I

I

30

40

z

0.0 50

0

I

I

I

I

2

4

6

8

Fig. 8. Plots of the slopes of the straight line fits in Fig. 6 against reciprocal concentration.

b)

8

itself. In contrast, effectively no current (less than 0.25 p,A) was seen to flow under irradiation at potentials corresponding to potentials at which dark currents were observed at any flow rate examined (0.002 < Vf < 0.03 cm 3 s - ] ). If the n = 2 process produces B P - , as might seem likely, then the clue to the apparent anomaly may lie in the respective reduction potentials of BP and CBP, which can be seen to imply that the equilibrium

~,

-i 6

~"

to O ,t-.-

4--

t

-

,~

2 0

0

I

10

[ B P ] I / 1 0 S c m 3 m o l "1

(Flow rate/cm3s "1)'2/3 10

,,,

10

20

30

40

50

(Flow rate/cm3s "1)'2/3 Fig. 7. Plots of dimensionlessrate constant K against Vf 2/3 for concentrations of BP of (a) 0.99mM and (b) 2.09mM.

limited current as a function of flow rate showed good agreement with the Levich equation [13] /lim = 0 . 9 2 5 n F [ C B P ] B u I k x 2 / 3 w D 2 / 3 ( V f / h 2 d ) I/3

B P - + CBP ~ BP + C B P lies to the right hand side (with an equilibrium constant of approximately 70). This has the effect that B P ' - leaving the electrode, under transport-limited conditions where the electrode potential is more negative than - 1.72 V, brings about the homogeneous reduction of CBP, since if the electron transfer occurs as above then the formation of C B P ' - will be followed by rapid chloride loss. The subsequent discharge of BP at the electrode will sustain the current at the full two-electron value. Consistent with this interpretation the reduction wave is split into two distinct

(22)

where F is the Faraday constant and D the diffusion coefficient of the electroactive species CBP, for n = 2 given a value of D = 1.5 X 1 0 - S c m 2 s - t . Fig. 9 shows the corresponding 'Levich plot'. The observation of n = 2 is in conflict with other reports [9,22,23] which suggest that under transient voltammetric conditions a single electron transfer is expected. However, the flow rates shown in Fig. 9 are very slow; therefore, the steady-state channel electrode conditions, which would be sensitive to follow-up chemistry with first-order rate constants of less than ! s - t [1,2], may, in this case, more closely reflect ahnost preparative conditions where follow-up reactions are suggested [9]. Finally, experiments were conducted in which the reduction of CBP was carded out in the presence of light of wavelength 325 nm and at potentials which were shown to produce appreciable photocurrents in the reduction of BP

250

<=L 2OO o

•-

I~ "%

150 OO

to

OO

._ lOO I:: -J

5o0 0

i 0.05

i 0.10

i 0.15

i 0.20

i 0.25

• 0.30

(Flow rate/cm3s -1) 113 Fig. 9. A Levich plot showing that the reduction o f CBP ( l . 9 6 m M in acetonitrile solution) is a full two-electron process under the conditions

employed.

W.M. Leslie et al. / Journal of Electroanalytical Chemistry 424 (1997) 165-171

waves, with half-wave potentials differing by 110mV, at slow flow rates (Vf < 0.005 cm 3 s - m).

4. Conclusions The photo-electrochemical reductions of BP and FBP appear to proceed via a disproportionation route in a mechanistically different manner from that of CBP. In the quantitative interpretation of the DISP mechanism, both quenching and conproportionation steps need to be included for a full mechanistic picture.

Acknowledgements We thank the Royal Society for supporting a Joint Project between the Universities of Oxford and Tartu.

References [1] R.G. Compton and R.A.W. Dryfe, Prog. React. Kinet., 20 (1995) 245. [2] R.G. Compton, R.A.W. Dryfe and J.C. Eklund, Res. Chem. Kinet., ! (I 993) 260. [3] R.G. Compton, R.A.W. Dryfe and A.C. Fisher, J. Electroanal. Chem., 361 (1993) 275. R.G. Compton, R.A.W. Dryfe and A.C. Fisher, J. Chem. Soc. Perkin Trans. 2:, (1994) 1581. [4] R.G. Compton, J.C. Eklund, A. Hallik, S. Kumbhat, L. Nei, A.M. Bond, R. Colton and Y.A. Mah, J. Chem. Soc. Dalton Trans., (1995) 1917.

171

[5] R.G. Compton, J. Booth and J.C. Eklund, J. Chem. Soc. Dalton Trans., (1994) 1771. [6] R.G. Compton, R. Barghout, J.C. Eklund, A.C. Fisher, S.G. Davies and M.R. Metzler, J. Chem. Soc. Perkin Trans. 2:, (1993) 39. S.G. Davies, M.R. Metzler, W.C. Watkins, R.G. Compton, J. Booth and J.C. Eklund, J. Chem. Soc. Perkin Trans. 2:, (1993) 1603. [7] A.M. Bond, D.M. Way, A.G. Wedd, R.G. Compton, J. Booth and J.C. Eklund, lnorg. Chem., 34 (1995) 3378. [8] J.F. Coetzee, Recommended Methods for Purification of Solvents, IUPAC, Pergamon, 1982. [9] L. Nadjo and J.M. Sav6ant, J. Electroanal. Chem., 30 (1971) 41. [10] K.R. Walczyk, G.S. Poprikov and R.N. Schindler, Ber. Bunsenges. Phys. Chem., 99 (1995) 1028. [11] R.G. Compton and B.A. Coles, J. Electroanal. Chem., 144 (1983) 87. [12] W.M. Leslie, R.G. Compton and T. Silk, submitted to J. Phys. Chem. [13] C. Amatore and J.M. Sav6ant, J. Electroanal. Chem., 85 (1977) 27. [14] C. Amatore, M. Gareil and J.M. Sav6ant, J. Electroanal. Chem., 147 (1983) 1. [15] V.G. Levich, Physicochemical Hydrodynamics, Prentice-Hall, New Jersey, 1962. [16] R.G. Compton, A.C. Fisher, R.G. Wellington, P.J. Dobson and P.J. Leigh, J. Phys. Chem., 95 (1991) 7538. [17] S. Moldoveanu and J.L. Anderson, J. Electroanal. Chem., 175 (1984) 67. [18] J.L. Anderson and S. Moldoveanu, J. Electroanal. Chem., 179 (1984) 107; 119. [19] R.G. Compton, M.B.G. Pilkington and G.M. Steam, J. Chem. Soc. Faraday Trans. 1, 84 (1988) 2155. [20] R.F. Michielli and P.J. Elving, J. Am. Chem. Soc., 90 Q ~,68) 1989. [21] N.V. Rees, R.A.W. Dryfe, J.A. Cooper, B.A. Coles, B .G. Compton, S.G. Davies and T.D. McCarthy, J. Phys. Chem., 99 t1995) 7096. [22] C.P. Andrieux, J. Pinson and J.M. Sav6ant, Nouv. J. Chim., 8 (1984) 107. [23] J.M. Sav6ant and A. Thiebauit, J. Electroanal. Chem., 89 (1978) 335.