6504
J. Phys. Chem. 1993,97,6504-6508
Anodic Growth and Interphasial Photoelectrochemistry of Cadmium Sulfide Thin Films As Probed by Laser Raman Spectroscopy N. R. de Tacconi'gt and K. Rajeshwar' Department of Chemistry and Biochemistry. The University of Texas at Arlington, Box 19065,Arlington, Texas 76019-0065 Received: February 19, I993 The anodic growth and interphasial photoelectrochemistry of CdS films were studied by cyclic voltammetry and laser Raman spectroscopy in aqueous sulfide electrolytes. The potential regimes and sulfide levels in the electrolyte were carefully chosen to avoid interference from oxide and hydroxide growth at the cadmium electrode surface. The use of the 488-nm Ar+ laser line was effective in generating Raman signals in the resonance scattering mode via absorption of the excitation light by the incipient CdS semiconductor layer. At 0.1 M sulfide, sulfur (predominantly SS)was detected via its Raman signature as a CdS photocorrosion product. The evolution of the Raman bands (attributable to these species and to CdS) as a function of time and potential was seen to reflect a complex interplay of several concurrent processes including photocorrosion, film regeneration, and desorption of the photogenerated sulfur from the CdS surface. On the other hand, an increase of the sulfide concentration to 0.5 M resulted in the absence of Raman signals due to sulfur at laser outputs ranging from 20 to 200 mW. The experiments described herein also serve to underline the utility of laser resonance Raman spectroscopy as an in situ tool for molecular-level tailoring of the variables in a photoelectrochemical system such that photogenerated carrier (electron or hole) transfer to an electrolyte species may be promoted at the expense of the electrode corrosion pathway.
Introduction
We thus show that a sulfur (corrosion) layer does not form at the CdS surface even at relatively high photon fluxes as long as Since the early studies by Miller and Hellerl and by Peter,2 the sulfide concentration in the electrolyte is 2-03 M. While the anodic electrosynthesisof CdS thin films has been extensively this is not a new finding, the in situ Raman monitor of sulfur in~estigated.~-IOA variety of electrochemical probes have been affords a molecular-level probe of the photocorrosion, as employed for the analysis of the Cd/CdS/aqueous sulfide demonstrated below for CdS, and contrasts with the (largely) interphaseincluding,for example, hydrodynamic~ o l t a m m e t r y , ~ ~ ~ ~ "chemistry insensitive" or ex situ techniques that have been cyclic/linear sweep ~oltammetry,4.~~9~.~.~~ analysis of charging employed thus far to assess the extent of this process. transients,*5 capacitancemeasurements,4q8chronoamperometry," and chronopotentiometry.8a The photoelectrochemicalproperties Experimental Section of these films (CdS is an n-type semiconductor with an optical bandgap of -2.4 eV) have also been profitably used to gain The working electrode was a polycrystalline cadmium disc mechanistic insights into the film growth and surface chem(Johnson-Matthey, 0.10 cm2). Before each experiment the istry.3~8b~9bJ0 However, as pointed out by us in a recent review," electrode was mechanically polished with alumina of decreasing measurement of charge, current, or potential alone only provides size down to 0.05 pm to yield mirrorlike surfaces. The electrode data with limited information content in terms of molecular details was then chemically polished by a 1:l mixture of glacial acetic concerning the electrochemicalsystem under study. On the other acid and 30% hydrogen peroxide for 5-10 s and washed with hand, spectroscopic probes in general, and Raman spectroelecwater. The CdS films were grown in the dark, either in 0.1 or trochemistry in particular, can be a useful complement in this 0.5 M sulfide solutions. The cell was shielded from ambient light regard." by a black box. Thus vibrational spectroscopies have been employed for the in The CdS single crystal (Cleveland Crystals) electrode was situ study of surface films on electrodesand for identifying surface polished with alumina (0.05 pm) etched for 5 s in 20% HCl and and bulk participant species in electrochemical systems.I2-l6 then rinsed with water. Photoelectrochemicalsystems have also been examined in situ by the Raman scattering probe.I7J8 In previous studies from this Anhydrous Na2S (Alfa Products) was used as received. The laboratory, we demonstrated the use of this technique for the sulfide solutions were prepared just before running each experstudy of organic (polypyrr~le)'~ and inorganic (cuprous thioiment by dissolving Na2S in deoxygenated double distilled water cyanate)20asemiconductor films and microstructures/bilayers (Coming Megapure). containing a combination of these electrode materials.20b A The Raman spectroelectrochemicalmeasurements employed preliminary study also showed the feasibility of the Raman the 488-nm line of an Ar+ laser and a Spex Ramalog instrument spectroelectrochemical probe as a useful monitor of anodic thin equipped with a Model 1680double monochromator and a cooled film growth.9a We develop this aspect in more detail and also photomultiplier tube (Model R928) operated in the photonexplore the postdeposition surface chemistry of CdS thin films counting mode. The experimental resolution was 5 cm-'. More in this paper. In particular, our use of suprabandgap laser details of the spectral measurements are given e l s e ~ h e r e . ' A ~.~~ excitation energy [the 488-nm (2.54-eV) Ar+ linesimultaneously holographic edge filter (Physical Optics Corporation) was added excites electron-hole pairs in the anodic CdS film] affords a for Rayleigh line rejection. dynamic in situ monitor of the interphasial photoelectrochemistry. All potentials quoted in this work are with respect to a Ag/ These experiments open a route to optimization of hole transfer AgC1/3 M KCl reference. to the solution at the expense of the film photocorrosion process. The laser source was operated at a nominal output of 20 mW in the experiments described below, except in instances where the f Visiting Scientist. Permanent address: INIFTA, Universidad Nacional de La Plata, C. C. 16, SUC.4 (1900) La Plata, Argentina. light intensity was a variable. 0022-3654/93/2097-6504$04.00/0 __
Q 1993 American Chemical Society
The Journal of Physical Chemistry, Vol. 97, No. 24, 1993 6505
Growth and Photoelectrochemistryof CdS Films
L
I
I
I
300
4 . 9 1 -1.4
V
1
1
-1.2
-1.0
1
,
I
I
4.8 4.6 4,4 4.2
W vs AgIAgCl
Results and Discussion Cd/O.l M SulfideInterphase in Dark. Cyclic voltammetry was first used to establish the potential regimes for the anodic growth of CdS films. Figure 1 contains a cyclic voltammogram (scan rate: 15 mV/s) for a freshly polished cadmium electrode in 0.1 M sulfide not containing any dissolved sulfur. (These solutions were also carefully deoxygenated in all cases.) The scan was reversed at systematicallymore positive potential limits with the cathodic excursion always limited to -1.3 V. The anodic peak corresponds to the electroformation of a CdS film on the cadmium surface:
-
+ SH- + OH-
UI
CdS
+ H,O + 2e-
(1) The implication of SH- ions in reaction 1 is consistent with the fact8sJ1that the equilibrium
S"
BM3
500
7w
Raman shift / cm-1
F'igure 1. Cyclic voltammogramsof a polycrystallinecadmium electrode in a deoxygenated 0.1 M NazS solution. Potential scan rate: 15 mV/s.
Cd
4w
+ H,O zsSH- + OH-
(2)
lies far to the right in a deoxygenated solution. (The pH of 0.1 M Sz- is -10.7.) Correspondingly, the characteristic S-H vibrations were detected at 2564 cm-l in these solutions.z2 The CdS films formed potentiodynamically under these conditions have a glassy yellow hue as long as the anodic limit is kept below -4.5 V. At higher potentials, the current was nearly independentof potential for a span of -0.8 V. This 'high field" domain has been considered by other authors4sand is beyond the scopeof this study. The cathodicpeak in Figure 1 corresponds to the electroreduction of the CdS film formed on the forward scan. This peak grows in magnitude as the anodic limit is increased. The above findings are in accord with those of previous authors.4s6 The component steps in reaction 1 have also been the topic of a recent study.8P Finally, it is important to note that, under the conditions employed for CdS film growth in this study, cadmium oxide or hydroxide formation7is not expected to play a significant role. Further, in aqueous sulfide solutions the equilibrium potential for CdS formation lies negative of that for Cd(OH)2 growth even at very small activities of sulfide?. Cd/CdS/O.l M Sulfide Interphase under High Intensity IUuminntion. Figure 2a contains an in situ Raman spectrum of a CdS film grown potentiodynamically in the dark (see above) by an anodic scan up to -0.05 V and then irradiated at constant potential for 15 min. The LO (longitudinal optical) phonon band at 305 cn-l and its overtone near 604cm-l are clearly diagnostic of CdSZ3 However, the spectrum is primarily composed of two
Figure 2. (a) In situ Raman spectrum of a cadmium electrode in 0.1 M NazS after a CdS film was grown in the dark and then irradiated at4.05 V by the 488.0-nm laser line (20-mW power) for 15 min. (b) Ex situ Raman spcctrum of a CdS single crystal electrode.
bands peaking at 221 and 474 cm-l, respectively. For comparison, the spectrum of a CdS single crystal is also included (Figure 2b). The 604-cm-' band here is superimposed on the luminescent emission envelopefrom e--h+ recombination. (Note that, in these instances, we are observing the Raman features via the resonance scattering mode, vide supra). Interestingly enough, the signifkant attenuationofthis featurein theCdSfilmcase (Figure2a) signals quenching of the emission due to hole transfer at the interphase. The holes generated in the CdS photoanode by laser irradiation can either attack the semiconductor film itself
CdS
+ 2h+
03
+
(CdS)S CdZ+ (3) or oxidize a reducing agentu.zs such as S2-in the solution:
+
-
HS- + OH- 2h+
so+ xs2-
+
So HzO
s,+lz-
(4a)
(4b) The corrosion proms (reaction 3) has been represented in a modified form from that considered by previous authors2Q7 to underline the fact that the generated sulfur on the CdS surface stabilizes via formationof S, molecules (vide infra). The solution proms, on the other hand, yields polysulfides (reaction 4b). The important role played by the solubility of the corrosion product (sulfur in this case) in the (chalcogenide) electrolyte in the stabilization of a group 11-VI semiconductor such as CIS has been stressed by previous authors.21 We emphasize at this juncture that the additional features at 221 and 474 cm-l are not due to hydrated oxides on cadmium.91 Figure 3 contains a sequence of Raman spectra taken at increasingly positive potentials (c.f. Figure 1) at which the (illuminated) Cd/CdS/O.l M sulfide is poised. These spectra are shown for a wider frequency domain (140-700 cm-l) as compared to those in Figure 2. For these experiments, the anodic film was first held at a given potential for 2 min in the dark and for 10min under illumination prior to initiating the spectral scan. The f i m was then electroreducod at -1.4 V for 3 min prior to switching to the next (higher) potential in the sequence. The number of bands in these spectra (other than those assignable to CIS, c.f. Figure 2) excludes the possibility that the surface (corrosion) layer comprises single S atoms. Weassignthe bandsat 156,221,438,and474cm-l toprimarily SS species with contributions from shorter sulfur chains and polysulfides (vide infra). The intense line at 22 1cm-l corresponds to the totally symmetric bending of cyclic species, and the broad bands at 156 and 474 cm-l correspond to bending and torsional
de Tacconi and Rajeshwar
6506 The Journal of Physical Chemistry, Vol. 97, No. 24, 1993
1020
500
time Is
Figure 4. Time dependence of the Raman intensity at 474 cm-’of a Cd/CdS electrode under 488.0-nm laser light. The CIS film was previously grown in the dark by an anodic scan up to -0.1 V in 0.1 M Na2S.
I
I
200
300
500
400
600
Raman shift I cm-1 Figure 3. Potential-dependentsequence of Raman spcctra recorded on an electrochemically grown CdS in 0.1 M Na2S. Each spectrum was obtained after 10-min laser irradiation (as in Figure 2) at the indicated potentials. modes of S S . The ~ formation ~ ~ ~ of Sa is precedented, for example, during the electrooxidation of sulfide on gold electrode^.^^.^^ Evolution of Sulfur and CdS with Time and Potential at the Illuminated Cd/CdS/O.l M SuVide Interphase. The bands due to CdS increase initially with anodization potential and then decrease somewhat (Figure 3). We can discard the possibility that this decrease is due to inner oxide formation for reasons noted above. Instead we must consider the competition between the CdS photocorrosion process represented by reaction 3 and film regeneration mechanisms which include reaction 1 and chemical processes such as Cd2+
+ SH-+ OH-
us
CdS
+ H20
(5)
Thus in the initial stages, film growth dominates (Le. U I + us > u3, the D’S are the reaction rates) whereas, at potentials positive of -0.7 V, a net thinning of the CdS layer ensues from the unfavorable competition between film growth and photocorrosion (03
> D l + us).
In contrast to the signals from CdS, the S g Raman bands monotonically increase in magnitude with the anodization potential. As mentioned earlier, the electrode was cathodically treated between each successive spectral acquisition. The consequent roughening of the electrode surface (i.e. higher surface area) is believed to be a factor in the increase of the Ss signal and supports a model favoring adsorption of this photogenerated product (c.f. reaction 3). Figure 4 maps the temporal variation of the most prominent Sg signal at 474 cm-I at a fixed potential, namely, -0.10 V. For this experiment, the CdS film was initially potentiodynamically grown in the dark using a single scan at 20 mV/s from -1.3 to -0.1 V. The 474-cm-l band is seen to initially grow in intensity in a linear manner followed by a much slower rate of growth at times longer than 150 s (Figure 4). We presume that the Raman
-
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
E I V vs Ag IAgCl
Figure5. Simultaneous measurement of the Raman intensity vs potential (a, top) at 474 cm-1and the current-potential profile (b,bottom) during acathodicscanat 1mV/s. Thescanwasstartcdaftcr20minofirradiation
at 0.0 V. band intensity is reflecting the thickness of the photocorrosion product layer-the latter being proportional to the density of scatter centers in the laser light path.,* The tendency toward a plateau of the 474-cm-1 signal cannot be due to thinning of the CdS layer and theconsequent diminution of the sulfur generation rate (vide infra). Rather we attribute this trend to the chemical reaction of the sulfur layer with the electrolyte: (CdS)S
+ SH-+ OH-
v6
CdS
+ (S)S2- + H,O
(6) The initial growth of the superficial sulfur layer demands the following rate inequalities: u3 > 06 and u3 < u1+ us. on theother hand, in the Yplateau”region, the photocorrosion and the sulfur removal rates are probably of the same magnitude, i s . u3 = 06. Additionally, reactions 1 and 5 are facileenough to avoid depletion of CdS, thus maintaining an approximately steady-state level of sulfur at the CdS surface. Figure 5 illustrates thevariation of the 474-cm-I band intensity during a cathodic scan from 0 to --1.4 V. The electrode was pre-irradiated by the laser for 20 min at 0 V. The Raman signal diminishes monotonically down to a potential of -0.95 V (Figure
The Journal of Physical Chemistry, Vol. 97, No. 24, 1993 6501
Growth and Photoelectrochemistry of CdS Films
m
300
400
500
BOO
7W
Raman shift I cm -l
Figure 7. Raman spcctrum of cadmium electrode in 0.5 M Na2S after a CdS film was grown in the dark at -0.4 V and then irradiated by laser light for 10 min. The inset shows a cyclic voltammogram of cadmium in 0.5 M Na2S at 20 mV/s.
-0.03 -1.2
-1.0
0.6
-0.8
0.4
4.2
0
E I V vs AgIAgCl
Figure 6. Raman intensity-potential profile (a, top) at 474 cm-’ and the current-potential curve (b, bottom) as in Figure 5 but after the potential scan was reversed in the anodic direction.
Sa) corresponding to the cathodic peak in the voltammogram (Figure 5 b) . Further reduction of the film is manifestedas another peak at ca. -1.28 V accompanied by an arrest or even a slight increase in the 474-cm-l band intensity. The initial decrease of the 474-cm-l signal is accommodated by reduction and desorption of the SSadlayer along with the reduction of the CdS film:
--
+ 2e- Cd + S2SO + 2es2-
CdS
(74
(7b) On the other hand, shorter sulfur chains (than SS)and various polysulfides exhibits stretchingmodes in the 420480-cm-l range32 and could account for the leveling off of the 474-cm-l band intensity. Figure 6 shows the subsequent changes in the 474-cm-* band intensity as the scan is reversed in the anodic direction following theexperiment considered in Figure 5. Interestingly enough, the 474-cm-l band continues to fall (Figure 6a) through the CdSelectroformation wave (c.f. Figure 1 and Figure 6b) indicating the depletion of polysulfides and sulfur at the interphase in this potential domain.sa At potentials higher than -4.75 V, the 474-cm-’ band intensity starts to grow again as the compact CdS layer begins to photodegrade and generate SSagain via reaction 3. Thus the sequence of chemical events at the illuminated Cd/ CdS/sulfide interphase is rather complex, and admittedly reactions 3-7 as shown abovemay be oversimplified. Nonetheless, the above picture does reveal the complications induced by the competitionfrom many parallel processes, and in this senseperhaps augments the existing knowledge on the photoelectrochemistry at the CdS/sulfide interphase. In this regard, it is of interest to note that Sz-and polysulfides were claimed to suppress the photocorrosion of n-CdS single crystals even at 0.01 M Our results described above clearly indicate that such is not the case at least for the thin films employed in this study, and even at 0.1 M sulfide, corrosion prevails at high photon fluxes. If we accept that single crystals are even more prone to photocorrosion than thin (polycrystalline) films,33the scales are tilted even further against the conclusions of these early studies. Thus, we believe that the Raman monitor of the sulfur corrosion product provides a new and very sensitive in situ probe of electrode stability in
photoelectrochemical situations involving CdS or other sulfurcontaining semiconductors. We develop this issue next by considering the results for CdS at higher sulfide concentrations (0.5 M) in the electrolyte. Cd/CdS/O.5 M Sulfide Interphase under High Intensity Illumi~tiolzOver a limited potential regime, an increase of the sulfide concentration from 0.1 to 0.5 M had no substantive effect on the voltammetric behavior; for example, compare Figure 1 and Figure 7 (inset). However, at potentials positive of -0.2 V, the anodiccurrent due to the formation of solublecadmium species was found to increase relative to the 0.1 M case. The electrodissolution of cadmium is known to be assisted by the presence of high concentrations of Na2S and the consequent precipitation of CdS at the interphase.” Figure 7 contains the Raman spectrum of a CdS film grown in the dark at -0.4 V for 2 min, followed by laser illumination for 10 min. The spectrum shows a dramatic alteration from its counterpart for the 0.1 M sulfide case, c.f. Figure 2a. Notably only the bands attributable to CdS are to be seen and the Sa signals are absent. Figure 8 contains a set of spectra obtained at different bias potentials and with the laser source set at 30 mW. Again the pattern of good film stability against photocorrosion is persistent even at relatively positive potentials (compare with Figure 3) and at laser source output varying from 20 to 200 mW. A further contrast to the complex trends noted in Figure 3 is that the CdS signals monotonically increase with potential. As mentionedearlier, this trendcan beaccommodated by the growth of a “secondary” layer of CdS via precipitation of the elcctrogenerated Cd2+with the (rather high) concentration of sulfide (c.f. reaction 5 ) . The good stability of CdS at a concentration of 0.5 M of the sulfideis interpreted in terms of rapid hole transfer to theacceptor sulfide species according to reaction 4a. Interestingly enough, our failure to observe sulfur at the CIS surface even at very high photon fluxes argues in favor of a facile reaction 4b. On the other hand, an alternative mechanism invoking reaction 3 and a (very fast) reaction 6 is much less attractive, especially given that the adsorption-desorption equilibria involving sulfur would be unable to compete with the initial photooxidation step at high photon fluxes. Thus it appears safe to conclude that the photocorrosion of CdS (reaction 3) is not occurring under these conditions in contrast to the 0.1 M sulfide case. It is interesting to note that, with the exception of some early the vast majority of cases involving n-CdS (and
de Tacconi and Rajeshwar
6508 The Journal of Physical Chemistry, Vol. 97, No. 24, 1993
Acknowledgment. The authors thank the Texas Higher Education CoordinatingBoard for partial support of this research via a grant from the Energy Research and Applications Program. The laser Raman spectroscopy system was purchased with funds from a DARPA-URI project (contract monitored by the Office of Naval Research). References and Notes
1
200
300
400
500
600
R a m a n shift/ cm-1 Figure 8. Potential-dependent sequence of Raman spectra recorded on a Cd/CdS electrode in 0.5 M NaZS. Each spectrum was obtained after 10-min laser irradiation at the indicated potentials.
other group 11-VI) photoanodeshaveemployedeven higher sulfide concentrations, nominally 1.O M NaB, 1.OM S,and 1.O M NaOH (e.g. refs 24, 35, 36). Given the results from this study, this would appear to be an optimal electrolyte recipe for ensuring good photoanode stability even at insolation levels much higher than would be normally encountered in terrestrial situations. Concluding Remarks
In a broader perspective and in conclusion,we have shown that resonance Raman spectroscopy is a useful in situ tool for optimizing the variables in a photoelectrochemical system such that “direct” hole transfer from the semiconductor to the electrolyte is promoted at the expense of the deleterious photocorrosion process. The latter generates (predominantly) SSspecies in the case of CdS at the interphase, whose Raman signals may be “tagged” for monitoring the efficacy of the optimization. In doing so,we havealsoattempted tounravel the kineticcomplexities in suboptimal situations (e.g. 0.1 M sulfide) wherein the film regeneration and corrosion processes are intertwined. The 0.5 M sulfide case is much ycleaner”and easier to interpret in this regard. It is important to note that the Raman technique (and indeed mast spectroscopic tools) allow for tailoring and optimization of the semiconductor/electrolyte interphase at a molecular level. Existing diagnostics of photocorrosion proctsses such as coulometry, voltammetry, rotating ring-disk techniques, and atomic absorption spectroscopy are either largely chemistry-insensitive or only applicable ex situ. Finally, we have used only the electrolyte composition (and particularly sulfide concentration) as a variable in this study. The optimization of other variables includingelectrodeorientation (in the case of single crystals), electrolyte pH, and cation composition using the Raman technique, and extension to other electrode and redox systems will be the topic of future studies in this laboratory.
(1) Miller, B.; Heller, A. Nature 1976, 262, 680. (2) Peter, L. M. Extended Abstract No. 227,27th International Society of Electrochemistry Meeting, Zurich, Sept 1976. (3) Miller, B.; Menezes, S.; Heller, A. J. Electroanul. Chem. 1978, 94, 85. (4) (a) Peter, L. M. Electrochim. Acta 1978,23, 165. (b) Peter, L. M. J. Electroanul. Chem. 1979,98,49. (c) DaSilva Pereira, M. I.; Peter, L. M. J. Electrounul. Chem. 1982, 140, 103. (5) (a)Yeh,L.S.R.;Hudson,P.G.;Damjanovic,A.J.Appl.Electrochem. 1982.12, 153. (b) Damjanovic, A.; Yeh, L. S. R.; Hudson, P. G. J. Appl. Electrochem. 1982, 12, 343. (6) Birss, V. I.; Kee, L. E. J. Electrochem. Soc. 1986, 133, 2097. (7) Saidmn, S. B.; Vilche, J. R.; Arvia, A. J. Electrochim. Acta 1987, 32, 1153. (8) (a) Krebs, M.; Sosa, M. I.; Heusler, K. E. J. ElectroamI. Chem. 1991,301, 101. (b) Krebs, M.; Heusler, K. E. Electrochim. Acta 1992,37, 1371. (9) (a) Ham, D.; Son,Y.; Mishra, K. K.; Rajeshwar, K. J. Electounul. Chem. 1991, 310, 417. (b) Ham, D.; Mishra, K. K.; Rajeshwar, K. J. Electrochem. Soc. 1991, 138, 100. (10) Peter, L. M. Electrochim. Acta 1978, 23, 1073. (1 1) Rajeshwar, K.; de Tacconi, N. R.; Lema, R. 0. Anul. Chem. 1992, 64,429A. (12) Comgan,D. S.;Foley, J. K.;Gao, P.;Pons,S.;Weaver, M. J. hngmuir 1985, I, 616. (13) Hamilton, J. C.; Farmer, J. C.; Anderson, R. J. J . Electrochem. Soc. 1986, 133, 739. (14) Melendres, C. A. In Spectroscopic and Diffraction Techniques in Interfucial Electrochemistry; Gutierrez, C., Melendres, C. A., Eds.;Kluwer: Dordrecht, The Netherlands, 1990; p 181. (15) Mayer, S. T.; Mfiller, R. H. J. Electrochem. Soc. 1992, 139, 426. (16) AbruAa, H. D., Ed. Electrochemical Interfaces; VCH: New York, 1991. (17) Schwartz, D. T.; Miiller, R. H. Surf. Sci. 1991, 248, 349. (18) (a) Van Duyne, R. P.; Haushalter, J. P . J . Phys. Chem. 1983, 87, 2999. (b) Van Duyne, R. P.; Haushalter, J. P.; Janik-Czachar, M.; Levinger, N. J. Phys. Chem. 1985,89, 4055. (19) Son,Y.; Rajeshwar, K. J. Chem. Soc., Faraday Trans. 1992,88,605. (20) (a) Son, Y.; de Tacconi, N. R.; Rajwhwar, K. J . Electround. Chem. 1993,345,135. (b) deTacconi, N. R.; Son,Y.; Rajeshwar, K. J. Phys. Chem. 1993, 97, 1042. (21) Hodes, G.; Miller, B. J. Electrochem. SOC.1986, 133, 2177. (22) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordinution Compounds, 4th ed.;Wiley: New York, 1986. (23) Tell, T.; Damen, T. C.; Porto, S . P. S. Phys. Reu. 1966, 144, 771. (24) Ellis, A. B.; Kaiser, S.W.; Wrighton, M. S . J. Am. Chem. Soc. 1976, 98, 1635. (25) (a) Fujishima, A.; Inoue, T.; Watanabe, T.; Honda, K. Chem. Lett. 1978, 357. (h) Inoue, T.; Watanabe, T.; Fujishima, A,; Honda, K.; Kohayakawa, K. J. Electrochem. Soc. 1977, 124, 719. (26) Bard, A. J.; Wrighton, M. S. J. Electrochem. Soc. 1977,124, 1706. (27) (a) Gerischer,H. J . Elecrrwnal. Chem. 1977,82,133. (b) Gerischer, H.; Gobrecht, J. Ber. Bunsen-Ges. Phys. Chem. 1978,82, 520. (28) Nimon, L. A.; Neff, V. D.; Cantley, R. E.; Buttlar, R. 0. J . Mol. Spectrosc. 1967, 22, 105. (29) Lenain, P.; Picquenard, E.; Corset, J.; Jensen, D.; Steudel, R. Ber. Bunsen-Ges. Phys. Chem. 1988, 92, 859. (30) Lema, R. 0.;de Tacconi, N. R.; Arvia, A. J. J. Electroanul. Chem. 1990,283,319. (31) Gao, X.; Zhang, Y.; Weaver, M. J. Longmuir 1992, 8, 668. (32) (a) Janz, G. J.; Coutts, J. W.; Downey, J. W . ;Roduner, E.Inorg. Chem. 1976, 15, 1755. (b) Dubois, P.; Lelieur, J. P.; Lepontre, G. Inorg. Chem. 1988, 27, 1883. (33) Cahen, D.; Manassen, J.; Hodes, G.; Tenne, R. ACS Symp. Ser. 1981, 369. (34) Tsou, C. C.; Cleveland, J. R. J. Appl. Phys. 1980, 51, 455. (35) (a) Cahen, D.; Hodes, G.; Manassen, J. J. Electrochem. Soc. 1978, 125,1623. (b) Lando, D.; Manassen, J.; Hodes, G.;Cahen, D. J. Am. Chem. Soc. 1979,101,3969. (c) Hodes,G.; Manassen, J.;Cahen, D. J. Electrochem. Soc. 1980, 127, 544. (d) Cahen, D.; Vainas, B.; Vandenberg, J. M. J. Electrochem. Soc. 1981,128, 1484. (e) Hodes, G.; Manassen, J.; Cahen, D. J. Electrochem. Soc. 1981, 128,2325. (f) Licht, S.; Tenne, R.; Dagan, G.; Hodcs, G.; Manassen, J.; Cahen, D.; Triboulet, R.; Rioux, J.; Levy-Clement, C. Appl. Phys. Lett. 1985, 46, 608. (g) Licht, S. J. Phys. Chem. 1986.90, 1096. (36) (a) Miller, B.; Heller, A.; Robbins, M.; Menezes, S.; Chang, K. C.; Thomson, J., Jr. J. Electrochem. Soc. 1977, 124, 1019. (b) Heller, A.; Schwartz, G. P.; Vadimsky, R. G.; Menezes, S.; Miller, B. J. Electrochem. Soc. 1978, 125, 1156.