Photophysical and photochemical aspects of coupled semiconductors

representative of the true situation, then each vanadium atom essentially creates a defect center within the aluminophosphate framework. At low vanadi...
0 downloads 0 Views 2MB Size
J . Phys. Chem. 1990, 94, 6435-6440

from H2-treated, 0,-calcined VAPO-5. Since the vanadium species are not exchanged from VAPO-5, they must be bonded to the framework rather than just being occluded material. Also, no occluded species are observed in the adsorption experiments. Since we have not counted the number of spins observed by ESR, the actual concentrations of V4+and V5+at any given conditions remain unknown to us. Finally, if this scenario is somewhat representative of the true situation, then each vanadium atom essentially creates a defect center within the aluminophosphate

6435

framework. At low vanadium concentrations, the framework apparently remains intact. However, higher vanadium incorporations may not be possible due to loss of framework structural integrity. Acknowledgment. C.M. and M.E.D. acknowledge the financial support of this work by the National Science Foundation and the Dow Chemical Company through the Presidential Young Investigator Award to M.E.D.

Photophysical and Photochemical Aspects of Coupled Semiconductors. Charge-Transfer Processes In Colloidal CdS-TiO, and CdS-Ag I Systemst K. R. Gopidas, Maria Bohorquez, and Prashant V. Kamat* Notre Dame Radiation Laboratory, University of Notre Dame, Notre Dame, Indiana 46556 (Received: October 13, 1989; In Final Form: January 30, 1990)

The mechanistic and kinetic details of the charge injection from excited CdS into a large bandgap semiconductor such as Agl and TiOz have been investigated by coupling the two semiconductor systems in the colloidal form. The interaction between the two colloids led to the quenching of CdS emission. The rate constants for the charge injection from excited CdS into the conduction band of Agl and TiOz colloids were determined to be 2.2 X lo7 and >5 X 1O'O s-I, respectively. Transmission electron microscopic analysis indicated the possibility of several CdS colloidal particles interacting with a single particle of Ti02and participating in the charge injection process. Primary photochemical events in the CdS-Ti02 system were investigated by picosecond laser flash photolysis. The charge injected into the TiO, colloid and trapped at the Ti4+site was characterized from its broad absorption in the region of 500-760 nm. The extended lifetime of these trapped charge carriers indicated an improved charge separation in the coupled semiconductor system.

Introduction One of the limiting factors that control the efficiency of photocatalysis is the quick recombination between photoinduced charge carriers in semiconductor particulate systems. (See for example refs 1-3.) Efforts have been made by several researchers to retard such a recombination of holes and electrons within the semiconductor particle and promote heterogeneous charge transfer at the semiconductor surface. Such efforts have included surface modification of the semiconductor with a noble metaI4q5or an ~ x i d eand ~ . ~simultaneous scavenging of holes and electrons by adsorbed redox species.' For example, platinization of TiO, and CdS particles led to the enhancement in the efficiency of photocatalytic reduction.s Surface modification of CdS semiconductor with diethyldithiocarbamate anion shifted the flat-band potentials to more negative values which in turn facilitated charge-transfer processes? Consequences of such modifications on the photophysical and charge-transfer processes of colloidal CdS have been reported by us recently.I0 Another interesting approach for achieving efficient charge separation involves the coupling of two semiconductor particles with different energy levels (Scheme I). For example, in a colloidal CdS-TiO, system a photogenerated electron can be transferred from cadmium sulfide into a TiO, particle while the holes remain in the CdS particle. The difference in energy levels of the two semiconductor systems plays an important role in achieving such a charge separation. These coupled semiconductor systems have been referred to in the past as "sandwich structure^".'^-'^ Enhanced yields of H2 from the reduction of H2S was observed in the CdS particulate system upon addition of TiO, powder.I4 The charge-transfer processes in mixed semiconductor colloids, CdS-Ti02, CdS-ZnO,I1 Cd3P,-TiOz and Cd3Pz-Zn0,l2 and Agl-Ag,S," have been reported. A IO-fold 'Part !6 of the series 'Photoelectrochemistry in Semiconductor Particulate Systems. *Address correspondence to this author.

SCHEME I: Charge-Transfer Processes in a CdS-Ti02 Coupled-Semiconductor System

CdS

enhancement in the efficiency of methylviologen reduction has been observed upon increasing the colloidal TiO, content in a CdS-Ti02 system." ( I ) Kamat, P. V.; DimitrijeviE, N. M. Sol. Energy 1990, 44, 83. (2)Henglein, A. Top. Curr. Chem. 1988, 143, 115. (3) Kalyansundaram, K.; Grltzel, M.; Pelizzetti, E. Coord. Chem. Rev. 1986, 69, 57. (4) Kraeutler, B.; Bard, A. J. J. Am. Chem. Soc. 1978, 100, 5985. (5) Duonghong, D.;Borgarello, E.; Grltzel, M. J . Am. Chem. SOC.1981, 103,4685. (6) Darwent, J. R.;Porter, G. J. Chem. Soc., Chem. Commun. 1981, 145. (7) Kamat, P. V. Lungmuir 1985, 1, 608. (8) Kiwi, J.; Gratzel, M. J . Phys. Chem. 1984. 88, 1302. (9)Thackeray, J. W.; Natan, M. J.; Ng, P.; Wrighton, M. S. J. Am. Chem. SOC.1986, 108, 3570. (IO) Kamat, P. V.; DimitrijeviC, N. M. J . Phys. Chem. 1989, 93. 4259. (1 1) Spanhel, L.; Weller, H.; Henglein, A. J. Am. Chem. Soc. 1987,109,

6632. (12) Spanhel, L.; Henglein, A.; Weller, H. Ber. Bunsen-Ges. Phys. Chem. 1987, 91, 1359. (13) Henglein, A.; Gutitrrez, M.; Weller, H.; Fojtik, A.; Jirkovsky, J. Ber. Bunsen-Ges. Phys. Chem. 1989, 93, 593. (14)Serpone, N.; Borgarello, E.; Gratzel, M. J. Chem. Soc.. Chem. Commun. 1983, 342.

0022-3654/90/2094-6435$02.50/0 0 1990 American Chemical Society

6436 The Journal of Physical Chemistry, Vol. 94, No. 16, 1990

Gopidas et al.

Photosensitization of Ti02 semiconductor with various dyes has been discussed by us earlier.'s-18 Coupling with a short bandgap semiconductor would also enable sensitization of large bandgap semiconductors such as TiO2.I9 Since the coupled semiconductors are potentially useful in improving the photocatalytic conversion efficiency of photoelectrochemical systems, it is necessary to investigate their photophysical behavior and elucidate the kinetics and mechanistic details of charge-transfer processes. In order to address these important issues, we have now investigated CdSTiOz and CdS-Agl colloidal systems by transmission electron microscopy (TEM), emission measurements, and picosecond transient absorption spectroscopy. Experimental Section Preparation of Semiconductor Colloids. Colloidal T i 0 2 SUSpensions were prepared by the hydrolysis of titanium(1V) 2propoxide (Alfa) in acetonitrile (Aldrich, spectrophotometric grade). Colloidal CdS suspensions in acetonitrile were prepared by exposing 0.2 mM Cd(CIO& (Johnson Matthey) in acetonitrile with 0.02% Nafion (Aldrich) as stabilizer or 2 mM CdI, (J. T. Baker) in acetonitrile containing 1% water to H2S.% AgI colloids were prepared by the method of Vucemilovic and MiciE.Z1 AgI colloids (0.2 mM) in acetonitrile were prepared by rapid addition of K I solution to a solution of AgZSO4. (S042-exchanged polybrene (Sigma) was used as a stabilizer.) The water content in the Agl suspension was 1%. Coupled semiconductor colloids were prepared by mixing appropriate amounts of two colloidal suspensions. Care was also taken to keep the water content in the CdS suspension to a minimum which otherwise led to the precipitation upon mixing of two colloids. TEM Measurements. A Hitachi H600 transmission electron microscope was used to take micrographs of Ti02 and CdS semiconductor colloids. A drop of colloidal semiconductor suspension was applied to Formvar-coated copper grids, and excess solution was removed after 1 min. The grids were allowed to dry under ambient conditions. Absorptions and Emission Measurements. Absorption spectra were recorded with a Perkin-Elmer 3840 diode array spectrophotometer. Emission spectra were recorded with an SLM 800 photon-counting spectrometer. Fluorescence emission lifetimes were measured by the time-correlated single-photon-counting technique using an apparatus that has been described elsewhereZZ except that the excitation source was a mode-locked, Q-switched Quantronix 416 Nd:YAG laser which provided 80-ps pulses of 355-nm light with a frequency of 5 kHz and an integrated power of 10 mW. The emission lifetimes were calculated by iterative least-squares fitting with variable zero-time shift between the instrument response function and measured decay curve. Picosecond Laser Flash Photolysis. The excitation source was a mode-locked 355-nm laser pulse from Quantel YG-5OlDP Nd:YAG (output 4 mJ/pulse, pulse width I8 ps). The white continuum picosecond probe pulse was generated by passing the residual fundamental output through a D 2 0 / H 2 0solution. The output was fed to a spectrograph (HR-320, Instruments, Inc.) with fiber optic cables and was analyzed with a dual diode array detector (Princeton Instruments Inc.) interfaced with an IBM AT computer. The delay in the probe pulse was introduced by varying the length of the fiber optic cable. The details of the experimental setup and its operation are described e l s e ~ h e r e . ~The ~ * time ~~

-

-

(IS) Kamat, P. V.; Fox, M. A. Chem. Phys. Lett. 1983, 102, 379. (16) Kamat, P.V.: Chauvet, J.-P.; Fessenden, R. W. J . Phys. Chem. 1986,

90. 1389. (17) Kamat, P. V.:Gopidas, K. R.; Weir, D. Chem. Phys. Lett. 1988, 149,

491. (18) Kamat, P. V. J. Phys. Chem. 1989, 93, 859. (19) Gerischer, H.; Liibke, M. J . Electroanal. Chem. 1986, 204, 225. (20) Kamat. P. V.: DimitrijeviC, N. M.; Fessenden, R. W. J . Phys. Chem. 1987, 91. 346. (21) Vucemilovic, M. 1.; MiciC, 0. Radiar. Phys. Chem. 1988, 32, 79. (22) Federici, J.; Helman, W. P.; Hug, G. L.; Kane, C.; Patterson, L. K. Comput. Chem. 1985, 9, 171. (23) Ebbesen, T. W. Rec. Sci. Instrum. 1988, 59, 1307.

0.0 250

300

350

400

450

500

550

600

Wavelength, nm Figure 1. Absorption spectra of colloidal semiconductors in acetonitrile: (a) 2 m M TiO,, (b) 0.2 m M Agl, and (c) 2 m M CdS.

zero in these experiments corresponds to the end of excitation pulse. The excitation beam, which was spread over an area of 10 mm X 1 mm, and the probe beam were at right angles to each other. The intensity of the 355-nm laser pulse was maintained around 0.75 mJ/pulse (2.25 X einstein), and only -10% of the excitation pulse was absorbed by the initial I-mm path of the sample. All the experiments were performed at room temperature (23 "C). Results and Discussion Absorption Characteristics. Absorption spectra of CdS, AgI, and Ti02 colloids are shown in Figure 1. Colloidal CdS in acetonitrile exhibits absorption below 500 nm, corresponding to a bandgap of -2.4 eV. Colloidal Ti02 in acetonitrile exhibits absorption at wavelengths less than 350 nm which corresponds to a bandgap of -3.2 eV. The details of the characterization of these semiconductor colloids can be found e l s e ~ h e r e . AgI ~~~~~ is an ionic semiconductor with a bandgap of 3.0 eV. The observed peak at 420 nm is due to the excitonic transition. Details of the absorption characteristics of AgI have been reported earlier.21 The flat-band potentials of semiconductor colloids are about 150-200 mV more negative than those for the compact electrodes.26 The energetics of these semiconductors thermodynamically favor electron injection from the conduction band of CdS into the lower lying conduction band of AgI or Ti02. Since the absorption of CdS extends well into the visible region, it can be excited selectively with the visible light. TEM Analysis. Direct association of two semiconductor colloids is necessary for achieving photoinduced charge transfer between these particles. In order to probe the physical interaction between the two semiconductor particles in a mixed semiconductor system, we have investigated the CdS-Ti02 system by transmission electron microscopy. The pictures recorded with 150000-250000 magnification of CdS, Ti02, and CdS-TiOz particles are shown in Figure 2. Both CdS and TiOz particles are spherical and diameters of -40 and 300-600 A, respectively. TEM pictures (Figures 2c,d) recorded upon mixing the two colloidal suspensions show the interaction between CdS and TiOz particles. Several CdS colloidal particles were found to interact with a single, larger Ti02 particle. Two distinct types of interactions were observed: (i) clustering of CdS colloids near the TiOz surface and (ii) well-spaced distribution of CdS colloids on the TiO, particle. The Ti02/CdS ratio influenced the type of interaction that was observed. For instance, at low Ti02 concentration the CdS particles tended to cluster, whereas at high Ti02 (24) Kamat, P. V.; Ebbesen, T. W.; DimitrijeviE. N. M.; Nozik, A. J. Chem. Phys. Lett. 1989, 157, 384. (25) Kamat, P. V.; Ford, W. E. Chem. Phys. Lett. 1987, 135, 421. (26) MiEiC, 0. 1.; Nenadovic, M.T.; Rajh, T.; DimitrijeviS, N. M.; Nozik, A . In Homogeneous and Heterogeneous Photocatalysis; Pelizzetti, E., Serpone. N., Eds.; D. Reidel: Dordrecht, 1986: pp 213-216.

.

Colloidal CdS-Ti02 and CdS-Agl Systems

The Journal of Physical Chemistry. Yo/. 94, No. 16. I990 6431

=. (cl

. , ....l

,

’-

-

GOO

i4

c

Figure 2. Transmission elcclron micrographs of (a) CdS. (b) TiO,. and (c. d) CdS-TiOl colloidal systems. Note the clustered and isolated distribution of CdS particles on the TiO, particle in pictures c and d .

concentrations the CdS particles were likely to be well separated. These TEM pictures (Figure 2c.d) establish for the first time a direct interaction betuec-n the two semiconductor particles in a CdS-TiO, coupled semiconductor colloidal system. It is necessary that this association be strong for obscrving photoinduced charge transfer between these two semiconductors. Quenchinx 01 CdS Emission. CdS colloids prepared i n acetonitrile exhibit red emision at wavelengths g r a t e r than 550 nm. This red cmicsion. which arises as a result of sulfur vacancy. can be readily quenched b) electron acceptors such as violugensz’.’8 (27) Henglein. A. J. Phyr. Chem. 1982. R6.2291

and methylene blue.” In the present experiments CdS colloids exhibited emission maximum around 670 nm with a auantum neld of 0.02. Both TiO, and Agl colloids were found io que& the CdS emission. The emission spectra of CdS colloids recorded after the addition of various amounts of colloidal AgI are shown in Figure 3. Compared to the absorption of CdS, the absorption of Agl at the excitation wavelength was negligibly small even at the highest Agl concentration. Similar quenching of CdS emission was observed in the CdS-TiO, system. The quenching of CdS (28) Ramden. J. 1.; Webber, S. E.:Gritrcl. M.J. Php. Chrm. 1985.89. 2740.

6438 The Journal of Physical Chemistry. Vol. 94, No. 16. 1990

Gopidas et al.

0 Y

I

1

2 .o

3.0

I

0

I .o

4.0

I / [ Ag I] , IO5 M'' Figure 4. Dependence of [l/(boem- @,(obsd))] on I/[Agl] in the CdS-Agl system.

Wavelength, nm Figure 3. Quenching of CdS (2 m M in acetonitrile) with colloidal AgI. Concentrations of Agl were (a) 0, (b) 2.5 X IO", (c) 5.0 X IO", (d) 1.0 X and (e) 2.0 X M. Excitation wavelength was 370 nm.

emission in the coupled semiconductor system is attributed to the injection of an electron from CdS colloid into the conduction band of Agl or TiOz colloid (reactions 1 and 2). CdS

CdS (e-.

CdS (e--.h+) CdS (e--.h+)

h')

&

CdS

+ hv'

(la)

CdS

+ heal

(lb)

+ AgI -% CdS (h+) + AgI (e-) + Ti02-k!+ CdS (h+) + Ti02 (e-)

(2a) (2b)

The participation of Agl (or Ti02) colloids in the quenching of CdS emission can be analyzed by considering an equilibrium between associated and unassociated forms of the two colloids with an apparent association constant of Kapp(eq 3). CdS

+ AgI & [CdS-AgI]

(3) As shown earlier,'6v'8 the observed quantum yield, &,,(obsd), of the CdS-Agl (or CdS-Ti02) system can be related to the emission yields of unassociated ($Oem) and associated ($'em) CdS colloids by the equation $em(obsd) = ( 1 - a)$',, + adim (4) where a is the degree of association between CdS and Ti02. Equation 4 could then be simplified16 to 1 $'em

- $em(obsd)

-

I

$'em

- $'em

colloids was necessary to observe 90% quenching of CdS emission. The differnence in size of the particles, preparative conditions, and the nature of interaction between the semiconductor colloids could be responsible for this difference. It is evident from the TEM analysis (Figure 2c,d) that several CdS colloidal particles can interact with a single Ti02 particle and participate in the charge injection process. Strong interaction and favorable energetics are essential factors in observing an efficient charge injection into the large bandgap semiconductor. Rate Constants ofthe Charge Injection Process. If the decrease observed in the CdS emission yield is due to a charge injection process, it is possible to determine the rate constant for the charge injection from excited CdS into AgI (or Ti02) colloid. In the absence of a quencher the deactivation of excited CdS occurs by radiative (reaction 1a) and nonradiative (reaction 1b) processes. The fluorescence yield and the emission lifetime of CdS can be equated to expressions 6 and 7, respectively @'em = kr 1

7=-

(7) + knr When CdS is associated with AgI (or Ti02), reaction 2a (or 2b) also contributes to the emission decay. If we consider that the decrease observed in emission yield is entirely due to the charge injection process, the emission yield (4") and the emission lifetime (7') of the coupled semiconductor colloids can be expressed by the equations kr

and

1

+ Kapp(4Oem

- d~'em)[AgIl (5)

The linear dependence between 1 / ( $ O m - $,(obsd)) and l/[AgI] shown in Figure 4 supports the above analysis. The apparent association constant and the emission quantum yield of the [CdS-Agl] associated complex, as derived from the intercept [ l / ( 4 ° e m - $ ' d l and the slope [ ~ / ( ~ a p p ( $ o e m - ~ ~ m )ofthe l plot in Figure 4, were 2.1 X 1 Os M-'and 0.0005, respectively. The similarly determined value of Kappfor the associated complex [CdS-Ti02] was 8800 M-I, and nearly zero The large value of Kappin both these cases indicates a strong interaction between the two colloids. Hence, only a small amount of AgI or Ti02 colloid is sufficient to quench more than 90% of the CdS emission. This observation is rather different from the one reported earlier in aqueous solution." In this study, a 6:1 ratio of Ti02/CdS

If we assume that the radiative and nonradiative decay processes of the coupled semiconductor occur at the same rate as in CdS colloid, one could correlate the two quantum yields as

or

Hence

The Journal of Physical Chemistry, Vol. 94, No. 16, 1990 6439

Colloidal CdS-Ti02 and CdS-AgI Systems

103

103

u)

v)

c

c

5

C

3 102 y 0

102

0

0

0

IO'

IO'

i

i .I

0

400

I

I

800

I200

T i m e (ns) Figure 5. Emission decay profiles of colloidal CdS (2 mM in acetonitrile) containing colloidal Agl. Concentrations of Agl were (a) 0, (b) 5 X IOd, and (c) 1 .O X M. Emission was monitored at 680 nm. (See Table I for the lifetimes.)

Since, the charge injection efficiency, (4',,,, - 4Lm),and the emission yield of coupled semiconductors, can be determined from the plot in Figure 4, it is possible to calculate k,,. For example, do,,,, and 4:,, for CdS and CdS-AgI colloids are 0.01 and 0.0005,respectively. If we substitute these values and the value of average emission lifetime of CdS in acetonitrile (( T ) = 943 ns, see next section) as T in eq 12, we obtained the value of k, as 2.2 X IO7 s-I. This rate constant for heterogeneous electron transfer between CdS and AgI colloids is smaller compared to the rate constants obtained in the dye sensitization experiments. Rate constants of the order of 108-1010s-l have been reported earler for the charge injection from excited dye into the conduction band of Ti02.16,2e33 In the case of CdS-Ti02, 4imwas found to be very small (