ANALYST, MARCH 1986, VOL. 111 Photothermal Deflection Spectroscopy and Studies of P hotoelect roc hem ica I Processes Electrolyte Interfaces Robert E. Wagner, Victor K. T. Wong and Andreas Mandelis 299 Photoconductivity at (0001) n-CdS - Photoacoustic and Photothermal Sciences Laboratory, Department of Mechanical Engineering, University of Toronto, Toronto, Ontario M5S 1A4, Canada Photothermal deflection spectroscopy (PDS) was used to investigate the manner in which the degree of band bending in a photoelectrochemical cell (PEC), consisting of an illuminated n-CdS (0001) single crystal and a polysulphide electrolyte, affects the non-radiative recombination processes in the semiconductor. The photocurrent was also monitored simultaneously as a complementary energy conversion channel.The results show that PDS can be used successfully as an analytical tool for the understanding and interpretation of photoelectrochemical processes a t the photoelectrode - electrolyte interface. Keywords : Photothermal deflection spectroscopy; photoconductivity; photoelectrochemical cell; cadmium sulphide electrode; non-radiative process Several workers have studied cadmium sulphide and cadmium selenide based photoelectrochemical cells (PEC) in the last decade. The important consideration of stable CdS photoelec- trodes during a photoelectrochemical experiment has been successfully addressed by Ellis and co-workers in a series of publications.'-5 The authors have found that sulphide or polysulphides in aqueous solution quench the photoanodic dissolution of the CdS reaction: -A CdS --% Cd2+ (aq.) + S(s) + 2e- (1) (photoanode) The single crystalline CdS photoelectrode has been im- portant technologically, because when used in a PEC, it can form a simple device for sustained conversion of visible optical energy into electricity.The relatively large band gap energy of CdS compared with CdSe and other compound semiconduc- tors renders the former semiconductor more attractive for applications where large open-circuit photovoltages are desired. The analytical methods conventionally used to study photoelectrochemical effects at semiconductor electrode - electrolyte interfaces include photocurrent and photopoten- tial measurements'-5 as a function of the wavelength of the exciting radiation (i.e., photoaction spectra); differential capacitance measurements of the interface2; emission photo- luminescent spectra from the electrode surface following radiative deexcitation of carriers3-5; electroluminescent spec- tra6; voltammetric studies7; and/or combinations of these techniques.l-6 The recent development of photoacoustic spectroscopy (PAS) as an analytical electrochemical tech- niqueg19 has allowed the use of the photoacoustic effect10 in optical spectra acquisition of metal oxide and semiconductor interfaces. The main advantage of PAS over other conven- tional spectroscopies lies in its ability to measure the non- radiative pathway of the deexcitation manifold, i.e., carrier recombination processes and other heat-generating mechan- isms in electrochemical systems. The non-radiative de- excitation component is the main energy loss mechanism in such systems and is detrimental to their quantum and energy efficiency enhancement.11 From the experimental point of view, PAS cannot be used easily to monitor electrode - electrolyte processes in situ owing to the remote positioning of the transducer - detector system. For this reason, photother- ma1 deflection spectroscopy (PDS) has been recently applied to the investigation of electrochemical interface phenom- ena.12J3 PDS is a spectroscopic tool that utilises the mirage effect, the fact that the path of a laser beam, in a given medium, will be bent if the beam encounters a refractive index gradient in its propagation path. This refractive index gradient may be induced by temperature or concentration gradients.This technique therefore can yield solid electrode spectro- scopic information14 through probing the electrolyte portion close to the electrode surface, as well as information concern- ing chemical changes in the electrolyte owing to interfacial chemical reactions. 15 In this work we have exploited the dependence of the PDS signal on the non-radiative quantum efficiency of the CdS photoelectrode deexcitation manifold to monitor in situ the non-radiative mechanism, simultaneously with the photo- generated current at the junction betwen a (0001)-oriented CdS electrode and a stable polysulphide electrolyte. In this fashion, PDS proved to be a valuable channel of information complementary to the conventional analytical methods, aiding in the establishment of a more complete picture of the electrode deexcitation process pathways at the electronic level. Experimental The material used in these experiments was a 1 cm x 1 cm x 0.2 cm low-resistivity (p = 20 ohm cm) n-CdS crystal from Eagle-Picher (Miami, OK), oriented with the optic axis perpendicular to the surface (0001) plane.The crystal was etched prior to mounting in the PEC in a solution consisting of 95% V/V of 3 M HCl and 5% V/V of a 30% hydrogen peroxide solution in water. The crystal was etched for 20 s and then rinsed in distilled water. Subsequently, the crystal had one of its faces coated with a 0.5 mm thick layer of liquid In - Ga amalgam and ohmic contact was assured by allowing diffusion of the In - Ga into the CdS for 3 h at 350 "C. Further, the sample was epoxied on to an acrylic backing, which was chosen for its durability and resistance to electrolyte penetra- tion.The completed working electrode (WE) consisted of the CdS crystal with one face bare and exposed to the electrolyte and the other face contacting a copper lead via the In - Ga amalgam. The metallised face of the WE was insulated from the electrolyte by the acrylic backing and epoxy. In between some of the experiments the WE was immersed in cyclohex- ane, in order to remove any trace amounts of sulphur that may have formed on the surface during experimentation116 Platinum foil was chosen for the counter electrode (CE), covered with Pt black in order to increase the effective surface area available for the cathodic reaction. The CE was300 ANALYST, MARCH 1986, VOL.111 positioned very close to the WE (ca. 5 mm) to decrease the solution resistance. A Fischer saturated calomel electrode (SCE) was employed as the reference electrode (RE). Mott - Schottky plots were obtained from n-CdS in 1 M each of NaOH, Na2S and S. This polysulphide solution was found to optimise the stability and reproducibility of the differential capacitance measurements required for the Mott - Schottky plots. Optical absorption spectra of the polysulphide solution were taken with a Cary 17D spectrophotometer. It was thus found that the 1 + 1 + 1 M solution would absorb most light with energies above the CdS band gap at ca. 510 nm.1717 In order to reduce this absorption a 1 M NaOH - 1 M Na2S - 0.05 M S electrolyte was used for the subsequent photoelectrochem- ical experiments and was found to give satisfactory results.The PEC was made of Teflon and had a screw-on lid, which supported a fused-silica UV grade window for entry of the exciting radiation into the cell. Two more windows made of Crown glass were located on opposite sides of the PEC to allow the passage of the 2-mW, 632.8-nm He - Ne laser probe beam used for PDS measurements. The laser - PEC assembly was mounted on stages, which allowed four degrees of freedom in the beam path movement, two translations and two rotations. The probe beam was further focused with its waist above the CdS electrode surface, using a 15 cm focal length lens. Fig. 1 shows an overview of the experimental apparatus. The source of UV - visible radiation was an Oriel Corp.Model 6141 1000-W Xe arc lamp in series with an Instruments S.A. H-20 monochromator with a concave holographic grating for wavelength selection. The He - Ne probe beam deflection was measured with a United Detector Technology (UDT) Model 431 Position Monitor connected to a UDT SC/25 light position detector. An optical filter with a 5% transmittance for wavelengths below 590 nm was placed over the detector to enhance the signal to noise ratio (SNR). The exciting radiation intensity was modulated by an AMKO OC 4000 mechanical chopper, which also referenced the EG & G Model 5204 lock-in amplifiers used as the PDS and photovoltage signal processors. External d.c. biases were required for Mott - Schottky plots and PDS - photocurrent measurements.A Stonehart and Associates Model BC 1200 potentiostat was used for the purpose of providing a regulated voltage (potentiostatic mode) between the WE and CE. The a.c. ripple voltage required between the WE and CE for Mott - Schottky plots was supplied from a Krohn - Hite Model 500A generator coupled into the potentiostat. The PEC was operated in the conventional three-lead mode (Fig. 1). Data were collected using software programmed into a D.E.C. DPD-11/23 micro- computer via an A/D conversion board. Results Mott - Schottky Analysis In order to calculate the doping density and flat band potential of n-CdS in the 1 M OH- - 1 M S2- - 1 M S electrolyte, the space-charge layer capacitance was measured at the semi- conductor - electrolyte interface. Using the Mott - Schottky model for the junction, the space-charge layer capacitance C,, can be related to the flat band potential VF, by the equationls c-2= ' (V- v,, - kT/q) .. (2) sc q&&&A2 where q is the electronic charge, E is the dielectric constant of the space-charge layer (= 5.2),19 E, is the permittivity of vacuum (= 8.85 pF m-I), ND is the effective donor density, A is the semiconductor electrode area exposed to the electrolyte and V is the applied bias versus SCE. The key assumption to the validity of equation (2) is that the whole potential drop takes place in the space-charge region of the semiconductor. Tomkiewicz20 has shown that this assumption is generally valid at high bias modulation frequencies, where the equi- valent electrical circuit of the electrochemical interface can be represented as a single resistor and a capacitor connected in series.Fig. 2 shows the circuit diagram for the experimental apparatus for the Mott - Schottky plot determination. A d.c. bias was applied across the WE and CE, measured with respect to an SCE. A 5-mV r.m.s. a.c. voltage was super- imposed over V and the PEC reactance was plotted versus the modulation frequency of the a.c. voltage. Following Tom- kiewicz,20 the complex impedance of the cell can be written as 1 Z ( 0 ) = R,(0) + iX(0) = R,(0) - - C S C . . (3) where R,(o) is the resistance of the space-charge layer and X(w) is the cell reactance at angular frequency 03 = 2nf. A plot of log(2nX) versus logfis shown in Fig. 3(a). This graph is a straight line for bias voltages away from the flat band potential.By using the intercepts, s, of curves similar to Fig. 3 with the ordinate for a number of bias voltage values V, the space-charge capacitance was determined for each V from Csc(Vk) = lO-@k). Fig. 3(b) shows the combined experimen- tal and theoretical results for the equivalent resistor - capacitor electrical circuit for several capacitor values. A comparison of Fig. 3(a) and (b) is indicative of the validity of the simple resistor - capacitor representation of the semiconductor - electrolyte interface. Fig. 4 is the Mott - Schottky plot for the interface, from which the VFB value was found to be - 1.15 V versus SCE. Reference ' in - - Lock-in - = ~ N D converter amp I ifier and Position - = . In-phase sensor I - computer - Data storage plotter, Fig.1. Experimental apparatus for PDS and photocurrent measurements. See text for detailsANALYST, MARCH 1986, VOL. 111 + B * c3 f 301 )'A I REF 3 I Lock-in I Lock-in It CE CE REF 4 A B REF I WE 4 Potentiostat . Function A.C. generator 1 - Fig. 2. Experimental apparatus for Mott - Schottky measurements. See text for details '. 0.00 0.70 1.40 2.10 2.80 3.50 4.20 4.90 5.60 Log f Fig. 3. (a) Log - Io plot of the frequency dependence of photoelectrochemical ceflreactance. WE at -0.75 V vs. SCE. (b) Log - log plot of the frequency dependence of the capacitance (C) of a resistor - capacitor simulation circuit: A, C = 10 nF; B, C = 100 nF; and C, C = 1 pF The donor doping density was also calculated in terms of the slope of the Mott - Schottky plot: From this calculation, the doping density was found to be ND = 5.02 x 1014 cm-3.The value of VFB found in this work is in good agreement with the range of values -1.52 to -1.20 V versus SCE for n-CdS of comparable donor densities calcu- lated previously.2 PDS, Photovoltage and Photocurrent Spectra PDS spectra normalised by the Xe lamp spectrum were obtained in situ in the PEC with water and with a polysulphide elecrolyte. The PDS spectrum of CdS in water is shown in Fig. 5. Both the amplitude and the phase indicate an energy band gap at ca. 510 nm, in excellent agreement with previous / / / / / / / / / / / / A A A / lvp; , , n -1.3 -0.5 0.3 1.1 PotentialN vs. SCE Fig. 4. Mott - Schottky plot of n-CdS in 1 M OH- - 1 M S2- - 1 M S electrolyte .? 7.20 > g i m 0.801 7- c' c -200 O -240 2 I -260 a -220 $ E a 0.00 320 400 480 560 640 720 800 Wavelengthhm Fig.5. Photothermal deflection spectrum of n-CdS in water in the open-circuit configuration. Modulation frequency: 25 Hz spectroelectrochemical work.2 PDS signal profiles as a func- tion of beam offset distance from a black absorber surface and as a function of chopping frequency were found to be in general agreement with Murphy and Aamodt's work.21 Some differences in the signal profiles between this work and reference 21 were attributed to our optical versus their302 4 ANALYST, MARCH 1986, VOL. 111 1 0.06 1 I v) ul c c 50 0.05 - - 0.05 ‘5 40 - - Q E 0.04 - - 8 , c 2 0.02 - - 0.02 a Q : 2 , 0.01 v, 10 n a 0.00 0 -1.50 -1.15 -0.80 -0.45 -0.10 0.25 0.60 0.95 1.30 BiasN vs.SCE 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 Log[(lamp power) x 1.6 Wl light radiant flux (white light) S. Modulation frequency: 15 Hz i!r n-CDs in 1 M OH- - 1 M S2- - 0.05 M Fig. PDS and en-circuit photopotential vs. Xe lamp Fig. 9. PDS amplitude and photocurrent - voltage curves for n-CdS irradiated at 505 nm. Modulation frequency: 17 Hz Wavelengthhm Fig. 7. PDS amplitude and photovoltage spectra of n-CdS in 1 M OH- - 1 M S2- - 0.05 M S. Modulation frequency: 17 Hz. Signal drop-off above the band gap is due to electrolyte absorption Wavelengthhm Fig. 8. PDS amplitude and photoaction spectra of n-CdS under different biases vs. SCE. (a) V = -1.5 V; A, PDS; B, photocurrent. (b) V = -0.3 V; A, PDS; B, photocurrent. (c) V = +0.9 V; A, PDS; B, photocurrent.Both signal strengths are in volts measured by the lock-in amplifiers. Modulation frequency: 17 Hz c / ,,? Cd2+(aq.) Conduction band - - - - - - - I n-CdS Fig. 10. Energy level diagram of n-CdS - electrolyte interface showing various possible deexcitation pathways. ED, Activation energy for anodic decomposition; ERedox, energy of electrolyte redox couple; EG, band gap energy of semiconductor; EF, Fermi energy of semiconductor; A, band-to-band transition; B, radiative recombina- tion; C, non-radiative recombination; D, photoanodic decomposi- tion; E, anodic carrier separation; F, anodic electrode dissolution; G, electrolyte redox reaction; f i w , incident photon energy; fiw ’ , lumines- cent photon energy; fiQ, lattice phonon energy; and SCR, space- charge region resistive heating and to their tighter focusing of the probe beam.Fig. 6 shows the measured PDS signal amplitude and open-circuit photopotential for n-CdS in the 1 M OH- - 1 M S2- - 0.05 M S polysulphide electrolyte as a function of incident optical power as determined by an Oriel Model 7090-2 pyroelectric detector. The monochromator was employed with the grating removed from the lamp light beam path (white light) to enhance the SNR, especially of the PDS signal. Electrode illumination with monochromatic light of 510 f 4 nm resolution gave PDS and V,, curves similar to those in Fig. 6, but with a poorer SNR. The linear dependence of V,, on the logarithm of the light irradiance has been observed also by Ellis et aZ.2 and is in agreement with theoretical consider- ations.22 The PDS signal amplitude, however, is linear in the logarithm of the irradiance at low power levels and increases more rapidly at higher power levels.Spectra acquisition in the polysulphide electrolyte was severely hampered at h < 490 nm owing to the onset of strong absorption by the electrolyte. For reasons of SNR optimisa- tion CdS spectra taken in the presence of 1 M NaOH - 1 M Na2S - 0.05 M S electrolyte were not normalised with respect to the absorption spectrum of the electrolyte, as the latter is transparent in the band gap region of interest (A > 500 nm), and the semiconductor spectral features are flat below 500 nm (Fig. 5). PDS and photovoltage spectra were obtained simultaneously. Fig. 7 shows PDS and V,, spectra.The PDSANALYST, MARCH 1986, VOL. 111 303 signal is large even at wavelengths below the band gap and appears to be shifted to the right of the open-circuit photovoltage spectrum. The drop-off of both signals on the high energy side is due to absorption of the incident light by the electrolyte. The sub band gap strength of the PDS signal does not appear in Voc. This feature could either be excitonic in nature23 or surface defect-related.24 It has also been observed in photoacoustic spectra of n-CdS of the same lot obtained in this laboratory.25 PDS and photocurrent spectra were taken under several biases between -1.5 V and +0.9 V versus SCE at 0.3-V increments. The wavelength range was between 470 and 570 nm at 8-nm resolution. Fig. 8 shows results under extreme bias conditions and one intermediate value.A PDS spectral shift to longer wavelengths than the photocurrent spectrum, similar to that in Fig. 7, is apparent at all three bias levels and is characteristic of all other such spectra. For all biases shown, the PDS signal strength changes little compared with varia- tions in the photocurrent, which undergoes a dramatic increase when the applied bias becomes positive to the flat band potential. A substantial PDS signal is present at sub-band gap wavelengths, whereas the photocurrent signal is negligible. Photocurrent - voltage and PDS - voltage measurements were further performed, the WE being illuminated with 505-nm light. The resulting photocurrent and PDS signals were thus monitored as a variable d.c. bias was applied across the WE and CE.Fig. 9 shows the results: the photocurrent increases very rapidly at biases below and around the measured flat band potential, - 1.15 V versus SCE, and tends to saturation at more positive biases. This behaviour is typical of CdS photoresponse.2 The PDS signal amplitude anti- correlates with the photocurrent at biases negative to the flat band potential, exhibits a broad minimum around VFB and increases steadily at more positive biases without signs of saturation. Discussion The major electronic phenomena occuring at the n-type semi- conductor - electrolyte interface on irradiation with band gap or higher energy photons can be summarised as in Fig. 10. The band bending due to the Fermi level mismatch at the interface creates a depletion or space-charge layer, which separates the photocreated electron-hole pairs, thus preventing 100% recombination.For those carriers which eventually recombine (and do not contribute to the anodic photocurrent), both radiative and non-radiative deexcitation processes are the major recombination mechanisms competing for the carrier deactivation. The photoluminescence phenomena observed by Ellis and co-workersl-5 in n-CdS and other 11-VI semicon- ductors were found to be consistent with radiative deexcita- tion processes leading to electron-hole recombination at the interface. Non-radiative recombinations would tend to trans- fer the energy of the photoexcited carriers to lattice phonons resulting in localised heating at the interface. It is expected, therefore, that the PDS signal will be sensitive to the localised surface heating because of the minute variations the latter incurs in the refractive index of the liquid electrolyte adjacent to the heated surface.Assuming an electrochemically stable WE, our photocurrent and PDS results were interpreted as simultaneous monitors of carrier separation and non-radiative recombination mechanisms. The photoluminescent deexcita- tion pathway was not monitored in this work; however, comparisons were made with Ellis and co-workers' results .4 The donor-doping density calculated from the Mott - Schottky plot, ND, can be compared with the theoretical value for an n-type semiconductor: ND=(pePn)-' . . . . . . ( 5 ) 80 60 .- C 3 t E 40 -e Q 4- .- 20 0 Photocu went PDS -1.5 -1.0 -0.5 0 +0.5 +1.0 Applied biasiV vs.SCE Fig. 11. Semi-qualitative curves comparing relative contribution to PDS signal with photocurrent. Photocurrent, anodic carrier separa- tion; NR, interband non-radiative de-excitation; and T, other thermal processes in n-CdS in polysulphide electrolyte (e. g., space-charge layer carrier separation, electron injection into the working electrode and intraband non-radiative de-excitations). The NR curve is in qualitative agreement with Fig. 5a in Streckert et ~ 1 . ~ Any photode- composition of electrode was assumed to be negligible Using p = 20 ohm cm and kn = 300 cm* V-1 s-1,26 the theoretical doping density is found to be ND == 1 x 1015 cm-3, a value within a factor of two of the experimental value. The higher than linear dependence of the PDS signal amplitude on the logarithm of V,, at large radiant fluxes (Fig.6) is consistent with the enhancement of the PDS signal with respect to the open-circuit photovoltage of Fig. 7, at sub-band gap wavelengths, as all spectra were obtained using the highest power rating of the Xe lamp. The shift of the PDS absorption peak to the right in Fig. 7 is most likely due to surface recombination processes, intra-band gap defects and/or sur- face states that would provide efficient non-radiative deexcita- tion pathways22 detectable by the PDS probe. A systematic study of the PDS signal as a function of controlled surface conditioning of the n-CdS crystal will be necessary to elucidate the particular mechanism(s). For energies around and above 510 nm in Fig.7 an anti-correlation is apparent between the V,, and PDS signals. At present, we propose that this is due to an increase in the numbers of efficiently separated electron - hole pairs across the space-charge region and therefore an increased value of VOc, while the number of non-radiatively recombining carriers has accordingly decreased with a sub- sequent decrease in the PDS signal. The maximum value of V,, is seen to occur at ca. 503 nm, in agreement with previously reported results1 within the resolution of our monochromator (8 nm). In Fig. 8, the PDS spectral shifts to longer wavelengths than the photoaction spectra are consistent with the energy balance mechanism proposed above: at super-band gap energies the efficiently photoseparated electron - hole pairs contribute the electron to the anodic current, while the minority carrier participates in a redox reaction at the semiconductor - electrolyte interface2 (process G in Fig 10).This mechanism would tend to pull electrons away from the interface towards the CdS bulk, thus decreasing the probability of non-radiative recombination and would be responsible for the anti- correlation of signal observed in Fig. 8 for photon energies above ca. 510 nm. Another interesting feature of Fig. 8 is the effect of the bias. At applied bias negative to the flat band potential [Fig. 8(a)], band bending is significantly reduced. The decreased electric field across the space-charge layer all but inhibits electron - hole separation, as seen from the greatly decreased photocurrent signal, while the non-radiative deexci- tation pathway remains efficient.As the space-charge layer increases with increasing bias positive to the flat bands, the photoseparation mechanism becomes more efficient and the photocurrent signal increases [Fig 8(b) and (c)]. It is interest- ing that the PDS signal, associated with the non-radiative component, remains approximately as strong in Fig. 8(b) and (c) as in Fig. 8(a). It must be deduced, therefore, that there is a304 ANALYST, MARCH 1986, VOL. 111 significant decrease in the carrier numbers deexciting via pathways other than non-radiative with increasing bias and re-channeling of the excess of carriers to the external circuit. This mechanism is in agreement with the inhibition of the photoluminescent emission observed by Streckert et al.4 at positive biases. An analvsis of Fig. 9 indicates that as the degree of band bending in the CdS space-charge layer is changed, via the applied bias, the predominant heat-generating mechanisms in the WE are altered. For biases at or below -1.2 V, the major source of heat is non-radiative recombination; note that the PDS signal increases and the photocurrent decreases as the voltage decreases from -1.2 V. However, for biases greater than - 1.2 V, non-radiative recombination becomes less important as the carrier-separation efficiency increases. At these positive voltages, where a significant photocurrent is present, the dominant heat sources are: (a) carrier separation in the space-charge layer; (b) electron injection and subse- quent deexcitation into the valence band from the solution redox couple; and (c) non-radiative transitions of hot elec- trons in the conduction band.A semi-qualitative indication as to the relative percentage lcontribution to the PDS signal from the inter-band non-radiative recombination compared with other current - flow related heating processes, such as those discussed above, is shown in Fig. 11. The non-radiative component anti-correlates with the photocurrent, while the other thermal components correlate with the photocurrent, as expected.11 The non-radiative component versus voltage curve in Fig. 11 agrees well with photoluminescent emission intensity versus voltage data from a similar photoelectrochemical experiment with CdS [reference 4, Fig. 5(a)]. This correlation between radiative and non-radiative processes is to be expected as they constitute complementary carrier deexcitation pathways to the current-producing electron - hole separation.Further evidence for assuming that the heat-generation mode in the WE changes nature close to the flat band potential is found when one considers the phase data of the PDS signal for various applied biases. A phase shift of 20” was observed in the region corresponding to the “knee” of the PDS signal versus voltage curve in Fig. 9. This phase shift would suggest a fundamental change in the heat generation processes. The various heat generation processes discussed above are expected to take place at different locations relative to the WE - electrolyte interface; a PDS phase shift would indicate a shift in the thermal source location within the WE.In conclusion, the PDS technique coupled with photoaction spectra has been shown to be capable of monitoring non-radiative recombination and other heat-generating processes at the semiconductor - electrolyte interface and to measure directly the so far little studied non-radiative efficiency of PECs. These observations could be significant in the calculation of PEC efficiency losses and their minimisation through the physical understanding of the interfacial loss mechanisms, for which PDS appears to be a very promising probe. The authors acknowledge the support of the National Sciences and Engineering Research Council of Canada (NSERC) throughout the duration of this project. They are also grateful to the Institute for Hydrogen Systems (IHS), Mississauga, Ontario, for contributing the PDS apparatus towards the completion of this work.Useful initial discussions on some experimental aspects with Drs. S.-M. Park (Department of Chemistry, University of New Mexico) and M. Weber (Department of Chemistry, University of Toronto) are gratefully acknowledged. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. References Ellis, A. B . , Kaiser, S. W., and Wrighton, M. S., J. Am. Chem. SOC., 1976, 98, 6855. Ellis, A. B., Kaiser, S. W., Bolts, J. M., and Wrighton, M. S., J . Am. Chem. SOC., 1977,99,2839. Karas, B. R., and Ellis, A. B., J . Am. Chem. SOC., 1980, 102, 968. Streckert, H. H., Tong, J.-R., Carpenter, M. K., and Ellis, A. B., J . Electrochem. SOC., 1982, 129,772. Streckert, H. H., and Ellis, A. B., J . Phys. Chem., 1982, 86, 4921. Smiley, P. M., Biagioni, R. N., and Ellis, A. B., J. Electro- chem. SOC., 1984, 131, 1068. Wilson, J. R., and Park, S.-M., J. Electrochem. SOC., 1982, 129, 149. Masuda, H., Fujishima, A., and Honda, K., Bull. Chem. SOC. Jpn., 1982, 55, 672. Dohrmann, J. K., and Sander, U . , J. Phys, (Paris), 1983, C6, 281. Bell, A. G., Am. J . Sci., 1880, 20, 305. Fujishima, A., Maeda, Y., Honda, K., Brilmyer, G. H., and Bard, A. J., J . Electrochem. SOC., 1980, 127, 840. Mendoza-Alvarez, J. G., Royce, B. S. H., Sanchez-Sinencio, F., Zelaya-Angel, O., Menezes, C., and Triboulet, R., Thin Solid Films, 1983, 102, 259. Roger, J. P., Fournier, D., and Boccara, A. C., J . Phys. (Paris), 1983, C6, 313. Mandelis, A., J . Appl. Phys., 1983, 54, 3404. Mandelis, A,, and Royce, B. S. H., Appl. Opt., 1984,23,2892. Williams, R., J . Chem. Phys., 1959, 32, 1505. Dutton, D., Phys. Rev., 1958, 112,785. Salvador, P., J . Appl. Phys., 1984, 55, 2977. Van Vechten, J. A., Phys. Rev., 1969, 182, 899. Tomkiewicz, M., J . Electrochem. SOC., 1979, 126, 2220. Murphy, J. C., and Aamodt, L. C., J . Appl. Phys., 1980, 51, 4580. Gerisher, H., in Eyring, H., Henderson, D., and Jost, W., Editors, “Physical Chemistry: An Advanced Treatise,” Volume 9A, Academic Press, New York, 1970, Chapter 5. Thomas, D. G., Hopfield, J. J., and Power, M., Phys. Rev., 1960, 119, 570. Wasa, K., Tsubouchi, K., and Mikoshiba, N., Jpn. J . Appl. Phys., 1980, 19, L475. Siu, E., and Mandelis, A., “Fourth International Topical Meeting on Photoacoustic, Thermal and Related Sciences, Montreal, Canada, 1985,” Technical Digest, WD 11.1, &ole Polytechnique, Montreal. Sze, S. M., “Physics of Semiconductor Devices,” Wiley, New York, 1969, p. 21. Paper A51235 Received July lst, 1985 Accepted November loth, 1985