J. Phys. Chem. 1983, 87,59-63
59
Time-Resolved Photoelectrochemistry. A Laser-Induced Coulostatic Flash Study of n-TiO, in Acetonitrile Prashant V. Kamat and Marye Anne Fox" Department of Chemistry, University of Texas at Austin, Austin, Texas 78712 (Received: April 27, 1982; I n Final Form: September 8, 1982)
A laser-induced photoelectrochemical study of single crystal n-Ti02semiconductor electrodes, in native and chemically modified form, immersed in conductive electrolyte solutions in acetonitrile, has been carried out by employing time-resolvedcoulostatic flash techniques. A two-componentresponse is seen for the rise in potential induced by flash excitation. The initial fast response, which occurs within about 10 ns, is attributed to electron-hole pair separation in the space charge layer. The slower response, initiated within 5 ws, involves redox-system responsive double layer effects and/or heterogeneous electron transfer at the electrode-electrolyte interface. In the presence of Ru(bpy)?, for example, the contribution of this second component to the overall photopotential was as high as 40%. The decay of the photopotential to dark equilibrium occurs through both e--h+ recombination and dark charge-transfer processes. Silanization facilitated the electron-hole recombination process.
Introduction Although much progress has been made in recent years in understanding the events which occur in photoelectrochemical cells, a need still exists to examine mechanistic aspects of interfacial charge-transfer processes a t semiconductor-electrolyte interfaces. Unusual effects occurring a t such interfaces have been explained by invoking Fermi level pinning or by implicating surface states between the band edges of the semiconductor,' but a clear demonstration of their relative importance is still lacking. Furthermore, most of the work to date has been carried out under steady-state irradiation conditions and very little is known about the dynamics of interfacial processes occurring in the time domain of nanoseconds or microseconds. We hoped to examine how intentional alteration of the local environment, through covalent modification or solvent effects, might affect the dynamics of processes occurring at the semiconductor-electrolyte interface. Various possibilities have been explored in the past few years to chemically modify semiconductor surfaces, either to improve their photostability or to enhance their photoactivity,2 but silanation has proved to be the technique of choice for effective, robust coverage. We have therefore directed our studies toward the goal of understanding the dynamics of charge generation at native and silanized TiO,, a typical n-type metal oxide semiconductor. Although all previous studies of transient coulostatic effects have been conducted with aqueous electrolytes, nonaqueous solvents are often more suitable for photoelectrochemical studies employing ranges of redox couples because such solvents typically exhibit large solvent windows (e.g., 5 V for acetonitrile compared to 1.5 V for water) and because of the enhanced stability of many organic systems in nonaqueous solvents. Furthermore, the rate of photodissolution of low band gap semiconductors like GaAs or Si is lower in nonaqueous media than in aqueous solvents. We felt it of interest therefore to conduct our studies in a typical nonaqueous solvent, acetonitrile. The initial attempts to undertake time-resolved photoelectrochemical studies were made over 10 years but a more refined approach has been described recently (1) (a) Nozik, A. J. Annu. Reu. Phys. Chem. 1978,29, 189. (b) Bard, A. J.; Bocarsly, A. B.; Fan, F. R. F.; Walton, E. G.; Wrighton, M. S. J.
Am. Chem. SOC.1980, 102, 3671. (2) Murray, R. W.; Acc. Chem. Res. 1980, 13, 135. (3) Myamlin, V. A.; Pleskov, Y. V. 'Electrochemistry Semiconductors"; Plenum Press: New York, 1967; p 111.
of
0022-3654/83/2~87-0059$0 1.5010
by Richardson and P e r ~ n e . ~Their , ~ time-resolved coulostatic flash technique involves an experiment in which the electrode-electrolyte equilibrium is perturbed by an instantaneous pulse of charge induced by a laser pulse of band-gap photons. The subsequent time-resolved return of the electrode potential to equilibrium is observed a t open circuit. The overall photoelectrochemical effect observed is a combination of several processes: (1) formation of an electron-hole pair in the space charge layer, or deeper within the semiconductor, upon absorption of the photon; (2) electron-hole recombination; and (3) heterogeneous electron transfer a t the semiconductor-electrolyte interface. Since these processes are governed by different parameters (the nature of the semiconductor surface, the redox system in the electrolyte, etc.) it is necessary to analyze systematically each individual process contributing to the photoelectrochemical effect. Time-resolved coulostatic flash study can serve as a powerful tool in defining these different processes and could lead to a better understanding of the mechanistic features of photoelectrochemical effects at the semiconductor-electrolyte interface. In this paper we report our initial laser-induced timeresolved photoelectrochemical study of n-Ti02 in native and chemically modified forms in acetonitrile. Although Richardson and Perone have reported potential rise and decay kinetics for aqueous solutions in contact with single crystalline electrodes, no such time-resolved studies have been conducted on modified surfaces or in nonaqueous solvents. The studies reported here probe these experimental variables and represent the first investigations of modified semiconductor in contact with nonaqueous electrolytes in which the nanosecond to millisecond time scale has been examined. With this additional time resolution, we are able to resolve photoinduced changes within the irradiated semiconductor, within the electrode double layer, and within the electrolyte. Experimental Section
Materials. n-Ti02 single crystals (1mm thick, with the C axis perpendicular to major axis) of varying doping density (1016-1019/cm3)were employed as photoresponsive (4) Richardson, J. H.; Deutscher, S. B.; Maddux, A. S.; Harrar, J. E., Johnson, D. C.; Schmelizinger,W. L.; Perone, S. P. J.Electroanal. Chem. 1980, 109, 95. ( 5 ) Perone, S.P.; Richardson, J. H.; Deutscher, S. B.; Rosenthal, J.; Ziemer, J. W. J. Electrochem. SOC.1980, 127, 2580.
0 1983 American Chemical Society
Kamat and Fox
The Journal of Physical Chemistry, Vol. 87, No. 1, 1983
60
TABLE I: Characteristics of n-TiO, in Deaerated Acetonitrilea n o Redox couple type
Ti0, I
1016
TiO, I1
1019
TiO, HId (silanized)
1019
TiO, I TiO, I1 TiO, IIId (silanized) a
n D /cm3
10l6 1019 1019
T1,2rise,c ns A. I = 51 t 7.6 7.8
150 m J 10 i i
2
with Ru(bpy),’+
Av2IAvmax
(rO.O1)
T,,,rise,ns
0.02
5 6 5 10
0.03 0.03
2
B.I= 5mJ 6 0 i 10 21 t 5 21t 5
8.6 8.4
23
5 lli 5
0.12 X
i
2 2
61 * 10
0.06 0.13
All solutions were 0.2 M in LiClO,. Concentration of Ru(bpy),’+ was 2.7 Silanized by dipping in SiC1,ixylene f o r 10 min.
5
M.
i
av,1A v m ax (iO.01) 0.06
0.19 0.05 0.34 0.39 0.15
Half-life of the initial potential
rise.
anodes. Electrical contact was made by rubbing the back side of the crystal with an In/Ga alloy and connecting the shielded wire with Ag paint. Epoxy was then used to mount the crystal onto a glass tubing support, care being taken to expose only the flat crystal surface. The exposed area of such an electrode was typically about 0.2 cm2. Before each set of experiments, the electrodes were etched in a HN03:HF:CH3COOH mixture (3:3:1) containing a drop of Br2 and were then washed thoroughly with water and acetonitrile. The T i 0 2 crystals were silanized by dipping them into a dilute solution of SiClk6 Prior to their use in coulostatic flash experiments, the silanized electrodes were washed with water, ethanol, and acetonitrile. The dark cathode was platinum foil, cleaned thoroughly with acetonitrile. Acetonitrile was dried over CaH, and distilled. All other chemicals were reagent grade and were used without further purification. Unless otherwise specified, all the solutions contained 0.2 M LiC104 as supporting electrolyte and were deaerated by bubbling high-purity N2 for 15-20 min. Instrumentation. Our coulostatic flash instrumentation was essentially identical with that described by Richardson et al.’ except that the potentiostat control was omitted. The photoinduced potential variations a t the semiconductor-electrolyte interface were measured between the irradiated semiconductor and a dark Pt counterelectrode. The electrochemical cell, equipped with a quartz optically flat window, had a standard three-electrode configuration (n-Ti02working electrode, Pt counterelectrode, and saturated sodium ch1oride:calomel (SSCE) reference electrode). A large surface area Pt counterelectrode (10 cm2) was used in order to minimize the cell impedence which in turn minimized the response delay. The larger area of the platinum electrode allowed the working n-Ti02 electrode to dominate the various cell capacitances. The electrochemical cell was mounted in a Faraday cage and the electrical connections were made with BNC connectors. All leads were kept to a minimum to reduce the overall circuit capacitance. The excitation source was a 355-nm laser pulse (pulse width = 10 ns, third harmonic of a Q-switched Nd:YAG (Quantel YG 481) laser). The intensity of the full, unfiltered laser pulse at the electrode surface was 150 mJ. This intensity could be varied by introducing precalibrated gauze filters in the path of the excitation pulse. For the measurements of the rise in photopotential (in the submicrosecond time domain), the cell output was fed (6) (a) Fox, M. A.; Nabs, F. J.; Voynick, T. A. J. Am. Chem. Sac. 1980, 102,4036. (b) Dim, A. F. In “Siiylated Surfaces”, Leyden, D. E.; Collins, W. T., Ed.; Garden and Breach Scienctific: New York, 1980. (7) Richardson, J. H.: Perone. S. P.; Deutscher, S. B. J . Phys. Chem.
1981, 8s. 341
to a Tektronix 7A26 differential amplifier (200-MHz band width, 1-Mohm input), the output of which was connected to Tektronix R7912 digitizer. Measurements of the decay of the induced photopotential (in the microsecond-millisecond time domain) were carried out by connecting the cell output to Tektronix AM502 differential amplifier (1MHz band width, 1-kohm input), the output of which was fed to a Biomation 8100 controller (50-ohm input), a Tektronix 2A60 amplifier, and a Tektronix RM566 storage oscilloscope. Analysis of the transients were conducted by averaging three or more runs and computer fitting the resulting data to a simple first-order or to two simultaneous first-order kinetic processes. The capacitance of the cell is estimated to be less than 100 pF. The resistance of the cell was measured with a YSI Model 351 conductivity bridge as 200 ohm, giving an RC constant 20 ns. The capacitance measurements were done by employing a Wavetek VCG 114 (5-mV sinusoidal signal), PAR Model HR-8 lock-in amplifier, PAR Model 175 universal programmer, and a PAR Model 173 potentiostat/galvanostat.
Results and Discussion Electrode Characteristics. The doping density and band positions of the n-Ti02 photoanodes were characterized by Mott-Schottky plots in deaerated acetonitrile containing 0.2 M LiC104. The flat-band potential in all the cases was around -1 V vs. SSCE.8 Coulostatic Flash Study. The potential excursion (open circuit) after laser pulse excitation was monitored on different time scales to evaluate the functional dependence of the transient response on various parameters such as the nature of electrode surface, the intensity of excitation, and the influence of redox species in the electrolyte. More attention was devoted to the transient responses on the nanosecond-microsecond scale as they can be directly related to the photophysical and photoelectrochemical processes at semiconductor-electrolyte interface. Typical transient responses obtained on the various time scales are shown in Figure 1-3. Excitation with the laser pulse of a single crystal T i 0 2 electrode immersed in acetonitrile containing only inert electrolyte results in the generation of photopotential as shown by the potential excursion, Figure 1, A and B. After attaining the maximum photopotential, V,, the open circuit potential decays slowly to attain back the dark equilibrium (Figure IC-E). The effect of surface modification on the potential rise and decay profiles can be seen in Figures 2 and 3, respectively. The contrasting behavior between a native TiOp surface (Figure 2A) and a silanized surface (Figure (8) Kabir-ud-Din; Owen, R. C.; Fox, M. A. J . Phys. Chem. 1981,8S, 1679.
The Journal of Physical Chemistry, Vol. 87, No. 1, 1983 61
Time-Resolved Photoelectrochemistry ......
I
”
,
,
,... ..................................
.......
.............. .......
....
. . . . .
......
........
, . . .
.......
......
~
...... 1 .......................
,
........
h ;& ;1 .....................
,
...
,
I.................... +.....
.....
1
...)........
>
;..................................
Flgure 3. Photopotentiai decay at (A) native and (B)silanized single crystalline Ti02. Electrolyte: 2.7 X lo-, M Ru (bpy),*+, 0.2 M LiCIO, in acetonitrile. Laser pulse intensity: 150 mJ.
Flgue 1. Typical transient potential response at different time intervals at single crystalline TiO, (n, = 10i9/cm3) in acetonitrile containing 0.2 M LICIO,. (A, E) The photopotentialrise observed immediately after laser pulse excitation: (A) 0-100 ns, (E) 0-1 ps (bl) high-laser intensity (- 150 mJ), (b2) low laser intensity (-5 mJ); (C) the initial decay in potential (100 ns-10 ps); (D, E) potential decay to the open circuit dark equilibrium. ...........
......
....
...
....
Figwe 2. Second component of the rise in photopotentialat (A) native and (E) silanized single crystalline Ti02. Electrolyte: 2.7 X lo-, M Ru(bpy),*+, 0.2 M LiCIO, in acetonitrile. Laser pulse intensity: 5 mJ.
2B) in the microsecond time range demonstrates the importance of the nature of the surface in determining the kinetics of double-layer effect^.^ Similarly, the slower decay observed on native TiOz (Figure 3A) than on silanized TiOz (Figure 3B) is suggestive that silanation may introduce surface states capable of enhancing electron exchange. Although the introduction of an insulating layer during silanation could explain the absence of a significant second component in the rise time (Figure 2), such a layer cannot be invoked to explain the fast decay of photopotential to the dark equilibrium value. (9)Since our experiment measures net change in capacitance, we are unable to distinguish unambiguously potential changes associated with interfacial charges transfer or rearrangement of the double layer leading to a capacitance change.
Richardson et ale5,’have given a detailed interpretation of these potential excursions in different time domains for aqueous electrolytes. The initial potential excursion, occurring within nanoseconds, is the result of h+-e- separation in the space charge layer. As the pulsed electrode attempts to attain the new equilibrium potential, h+-erecombination and dark charge-transfer become dominant, leading to potential decay. The latter process usually dominates the initial part of potential decay (in microsecond time domain). The final open circuit dark equilibrium is attained as the solution and semiconductor Fermi levels adjust independently. This slower process, occurring within a few milliseconds, can be attributed to bulk diffusion to the surface and to redistribution of charges in the depletion layer and double layer. The net photopotentials and their observed rise times obtained in laser coulostatic experiments with and without a redox couple in the electrolyte are summarized in Table I for several different TiOz electrodes. The time required to attain 50% of the maximum photopotential (TI/?””)is taken as the characteristic parameter of the potential excursion. As can be noted from parts A and B of Figure 1,the photopotential excursion comprises a two-component response. The initial fast rise in potential must be caused by the h+-e- separation in the space charge layer. The half-life of this rise in potential ( T I2rise) is of the order of 10 ns. Thus, the first component ojthe rise time is of the same order as the RC constant of the cell and is therefore governed by the limitations of measurements and instrumentation. In other words, our kinetic measurements represent lower limits for the actual rate of electron-hole separation. The half-life of the initial rise in potential was independent of the nature of the electrode surface and the nature of the redox couple in the electrolyte phase. For example, with silanized and unsilanized n-Ti02with the same prehistory, T1/Zrise remained constant (8 ns). However, the observed rise times did respond to changes in doping density, lower nD resulting in higher T1/Zrise. For a given value of the required potential energy drop, the length of the space charge layer varies with nD-1/2.8Hence lower nDwould extend the space charge layer deeper into the semiconductor. In the present case, though, quantitative data cannot be extracted because of the limitations of time resolution: the possibility of slower charge separation due to excitation deeper in the space charge layer cannot be ruled out. The slower, second component of the photogenerated potential excursion occurs within about 5 ps and is found
62
The Journal of Physical Chemistry, Vol. 87, No. 1, 1983
Scheme I: Photopotential Generation in a Laser-Induced Coulostatic Experiment*
Kamat and Fox 1.0
/
0.01 10-1
10-1 INTENSIT
PI
E
1
I
I on
RELATIVI U N ~ I ~
Flgure 4. Dependence of initial photopotential (AV,) on light intensity at single crystalline n-TiO, ( n , = lOi9/cm3) in deaerated CHJN.
Consistent with these ideas, the magnitude of AV2 was found to vary with the type of redox couple employed (e.g., 02,ferrocene, Ru(bpy)p). The observed potential excursions can therefore be explained by the sequence of charges shown in Scheme I, i.e., as sequential alterations within the semiconductor at the double layer and within the electrolyte. The influence of the Ru(bpy),2+ couple on AV,/AV,,, ratio is recorded in Table I. With unsilanized n-Ti02,the D C presence of this redox couple increases the ratio of AV2/ AV,,. In our experiment, both ground- and excita A. Dark equilibrium with significant band bending. Fermi level of TiO, in equilibrium with E R in solution. ~ ~ ~ ed-state ~ R ~ ( b p y ) exist ~ ~ +a t the interface since this molB. Immediately after band gap photoexcitation. Initial ecule absorbs considerably at the wavelength of excitation electron-hole separation. C. Response of t h e Fermi level (355 nm). The redox level of triplet Ru(bpy)gZ+lies close to photogenerated electrons in the conduction band. D. to the conduction band of n-Ti02 (-1.3 V in CH3CN).l2 In Interfacial electron transfer from the solution phase redox this situation, electron transfer is feasible either from the couple (ERedox t o E'Redox). excited state to the conduction band13 or from the ground state to the photogenerated hole in the valence band. to vary with the type of redox couple employed in the However, since no additional photopotential could be obelectrolyte. Since the redox couples chosen here were served in this time and sensitivity range when the rutheemployed because they have been demonstrated unnium complex was intentionally excited with subband gap equivocally to react through heterogeneous electron wavelengths (at 532 nm) and since the lifetime of Rutransfer,lOJ1 this slower response may thus involve heter(bpy)gZ+ is short (0.6 ws),14 we assume that band-gap irogeneous electron transfer a t the electrode-electrolyte radiation is the more significant primary process under our interface, but is probably restricted to redox partners experimental conditions. found initially within the double layer. The response time With silanized n-Ti02, Ru(bpy)gZ+had very little effect of this component is a t least two orders of magnitude on the slower component of photopotential excursion. higher than the RC constant and hence is not affected by Apparently, the silane layer without any electro/phothe limitations of measurements. toactive group acts as a barrier to any heterogeneous In Table I, the fraction of the photopotential due to this electron transfer a t the interface, inhibiting electron excomponent of heterogeneous electron transfer is recorded change with the electrolyte. as the ratio of AV2 (the potential difference due to second The Effect of Light Intensity. The dependence of AV, component) to the maximum photopotential, AV,,. A on the laser intensity at the surface of the irradiated similar two-component response has been reported for semiconductor is shown in Figure 4. At lower light inn-CdS in Na2S/NaOH by Perone et al.7 Such behavior tensities, the initial component of the observed photopowas seen only with the excitation at the band edge and was tential AV1 varied linearly with log I , but at higher inattributed to the fast process of h+-e- charge separation tensities it attained the limit of saturation. This behavior followed by heterogeneous electron transfer at the interagrees with that observed in steady-state illumination face. However, these same workers failed to see a similar experiment^.'^ The half-life of the initial potential rise two-component response with n-Ti02 in aqueous medium.5 varied little over this intensity range. In the present study with n-Ti02 in acetonitrile, we have As can be seen from Table I, the ratio of AV2/AV,, was succeeded in resolving the photopotential excursion into higher at lower light intensity excitation. As with AVl, the a two-component response. fraction of photopotential due to heterogeneous electron The contribution from heterogeneous electron transfer transfer (AV2/AVmJ was linear with log I for both ferat the interface to the overall photopotential is found to rocene and R ~ ( b p y ) , ~redox + couples (Figure 5). The be rather small. This is probably caused by two factors: lower slope observed with ferrocene can be attributed to the n-Ti02 semiconductor electrode is not held a t a conthe larger energy difference between the oxidation postant anodic potential in our experiments and the comtential of ferrocene and the valence band edge of TiO,. At peting process of h+-e- recombination would be significant in the present case because of lower band bending. (10) Memming, R. In "Electroanalytical Chemistry"; Bard, A. J., Ed.; Marcel Dekker: New York, 1979; pp 1-84. (11)Wrighton, M. S.; Bolts, J. M.; Bocarsly, A. B.; Palazotto, M. C.; Walton, E. G.J . Vac. Sci. Technol. 1978, 15, 1429.
(12) Miyashita, T.; Matsuda, M. J . Phys. Chem. 1981, 85, 3122. (13) Clark, W. D. K.; Sutin, N. J. Am. Chem. SOC.1977, 99, 4676. (14) Slama-Schwok, A.; Feitelson, Y.; Rabani, J. J. Phys. Chem. 1981, 85, 2222. (15) Ellis, A. B.; Bolts, J. M.; Kaiser, S. W.; Wrighton, M. S. J . A m . Chem. SOC.1977, 99, 2848.
The Journal of Physical Chemistry, Vol. 87,No. 1, 1983 63
Time-Resolved Photoelectrochemistry
TABLE 11: Photopotentials and Rise Times at an Irradiated n-TiO, Single Crystal Electrode TiO, IIc E(Pt)b
Tinrise,
Av,/ Avmax
0.303 0.167 0.675
7.6 7.1 7.5 8.6
0.03 0.09 0.13 0.19
vs. SSCE
redox couple"
0, ferrocene Ru( bpy ), z + ' 3 +
nsi 2
t0.01
TiO, IIIc (silanized) (AV,, ~Vioops~)/ Avm, i 0.1
0.02 0.04 0.08 0.12
Ti,,rise,
Av,/ AVmax
+0.01
(Avmax ~Viooks)/ Avm, i 0.1
7.8 8.0 7.0 8.4
0.03 0.07 0.03 0.05
0.22 0.15 0.22 0.21
nst 2
a All redox couples were in deaerated acetonitrile (0.2 M LiCIO,), by bubbling N, except with 0, where oxygen was bubbled for 2 0 min. Dark potential as measured a t P t electrode with SSCE as reference electrode. Electrode was etched separately f o r every set of experiment. Silanized TiO, was washed thoroughly with water, acetone, and acetonitrile.
L
10-3
I
,
10-2
10-1
I NT E N 5 I T Y
IPELATIVE UNITS1
Flgure 5. Dependence of the fraction of redox system responsive potential rise on light intensity; 2.7 X lo4 M Ru(bpy),*+ (*) and 2 X M ferrocene (X) in deaerated CH3CN.
higher light intensity (150 mJ) where the photopotential is saturated, more h+-e- pairs are created in the space charge layer and the charge separation becomes less efficient as the bands shift closer to the flat-band position. In such a case, heterogeneous electron transfer a t the interface is expected to be less efficient. In case of lower intensity (5 mJ), the photopotential is still in the region in which AV- increases with laser intensity. Under such a situation the bands are still bent and heterogeneous electron transfer a t the interface is facilitated. Thus, our work has presented direct kinetic evidence for heterogeneous electron transfer in several ways: (a) the second component of the rise in photopotential depends directly on the identity of the redox couple present in solution; (b) the variation of the rise in potential with light intensity requires previously demonstrated effects on band bending which in turn imply kinetic effects consistent with heterogeneous electron transfer; and (c) invariance of the observed kinetics with electrolyte redox couple with surface-blocked semiconductors provides a useful contrast to that observed when heterogeneous electron transfer is possible. The Influence of Redox Systems on the Decay of Photopotential. The response of the photopotential to the presence of different redox systems is summarized in Table 11. An interesting observation is seen in the initial part of potential decay. The fraction of decay of the photopotential occurring within 100 ps is taken as the measure of the competing processes, viz. holeelectron recombi'nation and (dark) charge transfer at the interface. For silanized n-Ti02,this ratio remains almost invariant with a change in the identity of the redox couple and is higher than that observed a t the unsilanized n-Ti02 (e.g., in deaerated
acetonitrile the ratio is higher by a t least an order of magnitude). This observation serves as a good explanation for the reduced photoactivity observed with silanized nTiOz electrodes. Silanization of n-Ti02 has been shown to introduce surface states between the band edges thereby facilitating the hole-electron recombination processes.16 According to Tomkiewicz, when light is absorbed by the semiconductor, holes in the valence band will be driven to the surface. The electron will then either be driven to the interior of the semiconductor or a t potentials which are not too positive than EFB(as in the present case) they will tunnel through the barrier to the surface states. Heller" has presented a chemical model for the surface recombination process. The surface recombination can be decreased by reacting the appropriate surface species with a strongly chemisorbed species which would displace the surface state beyond the band edges. With unsilanized n-Ti02, this ratio varied with the type of redox systems employed. This shows that these redox systems play a key role in the dark reactions a t the semiconductor-electrolyte interface. The silanized surface studied here has no electroactive group. Such an attachment of silane seems to inhibit any electron transfer at the semiconductor-electrolyte interface. The observed conwith silanized n-Ti02 stant ratio of (AV- - AV,,J:AV,, supports this view. This study poses several important questions regarding the role of an electroactive or photoactive group if linked to the silane and the effect of silanization vs. stabilization on small band gap semiconductors. Currently we are attempting to answer these and many other questions which could lead to a better understanding of the photophysical and photoelectrochemical processes a t the semiconductor-electrolyte interface. Acknowledgment. This work was generously supported by the U S . Department of Energy, Funamental Interaction Branch, Office of Chemical Sciences. Coulostatic flash experiments were carried out in Austin a t the Center for Fast Kinetics Research which is supported by the National Institutes of Health and by the University of Texas at Austin. We are grateful to Dr. M. A. J. Rodgers for assistance in construction of the flash coulostatic apparatus and to Professor Micha Tomkiewicz (Brooklyn Polytechnic) and Dr. Ronald Wilson (General Electric Research and Development) for gifts of doped single crystal Ti02used here. M.A.F. gratefully acknowledges support as an Alfred P. Sloan Research Fellow (1980-2) and as a Camille ane Henry Dreyfus Teacher-Scholar (1980-5). Registry No. TiOz, 13463-67-7; Ru(bpy)3Z+,15158-62-0; 02, 7782-44-7; Sic&, 10026-04-7; acetonitrile, 75-05-8; ferrocene, 102-54-5.
(16) Tomkiewicz, M. J. Electrochem. SOC. 1980, 127, 1518-25. (17) Heller, A. In 'Photoeffects at Semiconductor-Electrolyte Interfaces"; Nozik, A. J., Ed.; American Chemical Society: Washington, DC; 1981 p 57. ACS Symp. Ser., No. 146.