J. Phys. Chem. 1987, 91, 577-581 crossing is enhanced by quenching even when BN is used as a quencher. Since intersystem crossing is not enhanced by hydrogen bond formation but by exciplex formation,14 it is supposed that exciplex formation occurs simultaneously with hydrogen bond formation. Charge transfer from quencher to fluorescer is a prerequisite for both exciplex formation and hydrogen atom transfer. Therefore, such a supposition is consistent with the (12) Mataga, N.; Tanaka, F.; Kato, M. Acta Phys. Polon. 1968,34,733. (13) Yamamoto, S.; Kikuchi, K.; Kokubun, H. J. Photochem. 1976,5,469. (14) Hamai, S. Bull. Chem. SOC.Jpn. 1984, 57, 2700.
577
previous conclusion obtained for the two conjugate a-electronic systems
that hydrogen atom transfer and hydrogen bond formation are simultaneous proces~ess.~-~ Consequently, it may be concluded even for N and BN that photobleaching as well as enhanced intersystem crossing occur in.the exciplex. Registry No. CA, 1210-12-4;.N, 135-19-3; BN, 15231-91-1; M, 9304-9; BM, 51 11-65-9.
New Photocathode Materlals for Hydrogen Evolution: CaFe,O, and Sr,Fe,,O,, Yasumichi Matsumoto,* Masaru Omae, Kazuyoshi Sugiyama, and Ei-ichi Sato Department of Industrial Chemistry, Faculty of Engineering, Utsunomiya University, Ishi-icho 2753, Utsunomiya 321, Japan (Received: January 6, 1986; In Final Form: July 21, 1986)
The photoelectrochemicalproperties of p-type CaFe204and Sr7Fei0OZ2 are studied in Nz-saturated 0.25 M KzSO4 (pH 6.0). The differences between the Fermi level and the top of the valence band are determined to be 0.14 and 0.4 eV for Sr7Fei0022 and CaFe204,respectively, from the activation energies of the conductivities and the Seebeck coefficients. The cathodic photocurrents of the hydrogen evolution are observed in the potential region more positive than RHE by 0.7-0.8 V. The band gaps are 1.8 and 1.9 eV for Sr7Fe10022 and CaFe2O4,respectively. The flatband potential of Sr,Felo02, is 0.1 V vs. SCE, but that of CaFe204cannot be determined because of the Fermi level pinning. Pt deposited on the surfaces of both electrodes accelerates the electrochemical process of the hydrogen evolution reaction. CaFe204is more stable than Sr7Fe10022 in the long-term test. The short circuit photocurrent of 0.3-0.4 bA/cm2 is observed in a Pt-deposited CaFez04 (p type)/Znl,zFel,80,(n type) assembly under a xenon lamp illumination. The band structures of Sr7Fe,,02, and CaFezOp and their photoelectrochemical processes are also discussed.
Introduction The possibility of efficiently photoelectrolyzing water with illuminated semiconducting electrodes as a solar energy conversion technique has received considerable attention. Recent interest in this process has led to intensive studies of the photoelectrochemical properties of a variety of materials. The oxides having Fe in their lattice are promising as n-type photoanode materials because of their relatively narrow band gap (about 2.1 f 0.2 eV) and their stability in aqueous For the unbiased photoelectrolysis of water, a p-type semiconductor for photocathodic hydrogen evolution is necessary, because the n-type oxides have the flatband potentials or the onset potentials of anodic photocurrent of oxygen evolution reaction more positive than that of the reversible hydrogen electrode (RHE).'-I4 Recently, Turner (1) Hardee, K. L.; Bard, A. J. J . Electrochem. SOC.1976, 123, 1024. (2) Kung, H. H.; Tarett, H. S.;Sleight, A. W.; Fenetti, A. J. Appl. Phys. 1977, 48, 2463. (3) Butler, N. A.; Ginley, D. S.; Eibschutz, M. J . Appl. Phys. 1977, 48, 3070. (4) Ginley, D. S.; Butler, M. A. J. Appl. Phys. 1977, 48, 2019. ( 5 ) Butler, M. A.; Ginley, D. S.J. Electrochem. SOC.1978, 125, 228. (6) Scaife, D. E. Solar Energy 1980, 25, 41. (7) Kennedy, J. H.; Shinar, R.;Ziegler, J. P. J. Electrochem. SOC.1980, 127, 2307. (8) Koenitzer, J.; Khazai, B.; Hormadaly, J.; Kershaw, R.; Dwight, K.; Wold, A. J. Solid State Chem. 1980, 35, 128. (9) Hatanaka, N.; Kobayashi, T.; Yoneyama, H.; Tamura, H . Electrochim. Acta 1982, 27, 1129. (10) Pollert, E.; Hejtmaneck, J.; Doumerc, J. P.; Clawerid, J.; Hagenmulle, P. J. Phys. Chem. Solids 1983, 44, 237. (1 1) Kochev, K.; Tzvethova, K.; Gospodinov, M. Mater. Res. Bull. 1983, 8, 915. (12) Shami, A.; Wallace, W. E. Mater. Res. Bull. 1983, 18, 389. (13) Turner, J. E.; Hendewerk, M.; Parmeter, J.; Neiman, D.; Somorjai, G. A. J. Electrochem. SOC.1984, 131, 177. (14) Matsumoto, Y.; Omae, M.; Watanabe, I.; Sato, E. J . Electrochem. Soc. 1986, 133, 711.
et al.13 reported the production of hydrogen from water by illumination of both electrodes in a Mg-doped (p type) and Si-doped (n type) iron oxide assembly. Thus, the investigation of p-type semiconducting materials for H2 evolution as well as n-type semiconducting materials will be very important for the unbiased photoelectrolysis of water by a p/n assembly. We have already discovered the p-type semiconducting oxides CaFez04 and Sr7FeloOz2with relatively narrow band gaps, that can act as the photocathode for H2evolution under visible light.l5 In this study, the details of the photoelectrochemical properties of the polycrystalline CaFe204and Sr7FeloOZ2 and the effect of surface deposited Pt on the H2evolution reaction are described. Experimental Section Solutions of metal nitrates were used as the starting materials. The solutions were stoichiometrically mixed so that the metal cations are distributed uniformly throughout the oxide. The mixed solutions were evaporated to dryness and then heated at 900 OC. The materials consisting of Fe and S r or Ca were ground in an agate mortar and then pressed into a pellet at 100 kg cm-z. The pellet samples were sintered at 1200 OC in air followed by oxidation in O2 at 1000 O C to make them sufficiently conductive. The sintered disk samples consisted of a single phase of Sr7Fe100z2 or CaFe204which was determined by X-ray analysis. The conductivities and Seebeck coefficients were measured at temperatures from room to about 600 O C in 02.All samples were waterproofed with polystyrene in order to obtain good reproducible results in the electrochemical tests. The whole electrode was sealed with epoxy except for the front surface. The solution was N2saturated 0.25 M K2S04 (pH 6.0) in order to inhibit the photo(15) Omae, M.; Matsumoto, Y.; Sato, E. Abstract of Papers, Electrochemical Society of Japan Meeting, Yamanashi, April 1985. Abstract A31 1.
0022-3654/87/2091-0577.$01.50/0 0 1987 American Chemical Society
Matsumoto et al. .40
-06.
-
Tg x
-
-x ?g
"7"D022
1-10
- 06
-03
0
Potentlal (Vvs SCE)
Figure 3. Current-potential curves of CaFe204and Sr7Felo022 electrodes in N2-saturated 0.25 M K2S04.
i
lo-;o
20
31)
T-? d o - 3 )
Figure 2. Conductivitiesof CaFe204and Sr7Felo022 as a function of TI.
current of the 0, reduction. A SCE was used as the reference electrode. Unless otherwise stated, the electrode potential was referenced to a SCE. A quartz H-type cell was used as the electrolytic cell. The polarization curves were determined by using a potentiostat (Hokuto, Ltd.), a function generator (Hokuto, Ltd.), and an X-Y recorder (Yokogawa, Ltd.). Monochromatic light of various wavelengths was obtained with a monochrometer (Ritsu, Ltd.). A 500-W xenon lamp was used as the light source. The photocurrents with the monochromatic light were converted into quantum efficiencies by using a thermopile (Epply, Ltd.). Impedance measurements of the cell oxide electrode/electrolyte/Pt system were made with a frequency response analyzer (S-5720C, N F Electronic Instruments) operating in a frequency range from 100 H z to 100 KHz. The capacitances of the space charge layer of the electrodes in an equivalent circuit to which the experimental data were fitted were calculated by using a computer. The impedance measurements were made several times for one sample. The H, gas evolved at the oxide electrodes under illumination was detected by a gas chromatograph (Shimazu, Ltd.). The short circuit photocurrent in the p/n oxide assembly (p-type was measured with a zero shunt CaFe,04/n-type Zn, 2Fel*04) ammeter (Nikko Keisoku, Ltd.).
Results The conductivities of CaFez04and Sr,FeloOz2at 25 OC were 7.0 X and 1.0 X Q-' cm-', respectively. Figure 1 and 2 show the temperature dependences of the Seebeck coefficients, Q,and conductivities, u, respectively. The sign of the Seebeck coefficients for all the oxides is positive, showing that the oxides are p-type semiconductors. The activation energies of Q (EQ)and CT (E,) were almost the same for each oxide, 0.40 f 0.03 eV for CaFez04 and 0.14 f 0.06 eV for Sr7FelOOz2.These values are equal to the energy difference of Ef - E, in the case of the p-type semiconductor,I6 where E, and E f denote the top of the valence (16) Bransky, I.; Tallen, N. M. Physics of Electronic Ceramics, Part A ; Hench, L. L., Dove, D. B., Eds.; Marcel Dekker: New York, 1971; Chapter 3, pp 67-98.
-
0.001 300
500
700
Wavelength(n m)
Figure 4. Dependences of the quantum efficiencies of the cathodic photocurrents on the wavelength at -0.6 V.
Figure 5. ( h ~ q ) vs. ~ ' hu ~ plots of CaFe204and Sr7Felo022 electrodes at -0.6 V.
band and the Fermi level in the bulk, respectively. Figure 3 shows the current-potential curves of CaFez04 and Sr7FeI0O2,electrodes in N,-saturated electrolyte. Their action spectra of the quantum efficiency, 7, of the photocurrent at -0.6 V are shown in Figure 4. The cathodic photocurrents were determined to be mainly based on H2 evolution as determined by gas chromatography analysis. The onset potentials of the photocurrents were about 0.1 V for both the CaFez04and Sr,FeloO,z electrodes as shown in Figure 3. It should be noted that the onset potentials of the photocurrents of H2 evolution were more positive than that of the RHE (-0.6 V at pH 0.6) by 0.7 V. To determine the band gap, E,, of the oxides, the following equation, which is based on the Gartner mode1,I7~'*was used hvq =
K(hv - E,)"12
(1)
where K is a constant and other symbols have their usual meanings. (17) Gartner, W. W. Phys. Rev. 1959, 116, 84. (18) Ginley, D. S.;Butler, M. A. J . Appl. Phys. 1977, 48, 2019.
New Photocathode Materials for Hydrogen Evolution RA
The Journal of Physical Chemistry, Vol. 91, No. 3, 1987 519
CA
RF Figure 6. Equivalent circuit of the semiconductor/electrolyte interface. The symbols are explained in the text. 0
400
0
-0.5
0
Potential ( V vs. SCE
Figure 8. Mott-Schottky plots for C,, and C, of Sr7Felo022 electrode in the equivalent circuit (Figure 6). light on
-3.0.
t
-5000
\-400
--
1.5.
9E
O
E
i
Figure 7. Impedance diagrams for CaFe204(a) and Sr7Felo022(b) electrodes. Solid lines are experimental results and circles denote the data calculated from the equivalent circuit (Figure 6 ) using a computer. The experimental and the calculated data are fitted to each other.
Experimentally obtained results showed good correlation with this equation when n = 4,as shown in Figure 5 from which indirect phototransition is thought to prevail in these oxides. Equation 1 will hold, even if surface recombination occurs at the surface, because this process, in general, depends only on the applied p0tentia1.l~ The E, of CaFez04and Sr7Fe10022 were determined to be 1.9 and 1.8 eV, respectively. Various, but similar equivalent have been proposed for the semiconductor and liquid junction. The equivalent circuit shown in Figure 6,which is almost the same as that proposed by Haart et aLZ2for ferrite semiconductors, was adopted for the present semiconductor/electrolyte interface. In this circuit, R, is the resistance of the solution, RB and CBare the resistance and the capacitance of the semiconductor bulk, respectively, C, is the capacitance of the space charge layer, Rf is related to the faradaic impedance, and RA and CAare related to the surface state of the CaFez04 In the computer analysis, R, had a value of 30 9, while 2.1 X l o 3 and 71 9 were used as RB for the CaFe204and Sr7FeloOZ2electrodes, respectively. R, was measured by using a Pt black electrode situated at the same position as the semiconductor electrode in the test cell. RB’s were calculated from the electrode thicknesses by using their resistivities. The impedance spectra of the cell were measured in the potential region from -1 .O to 0.5 V. The typical spectra measured at 0 V are shown in Figure 7. The spectra for the Sr7Fe,0022 electrode depended on the potential, while the CaFe204electrode spectra was independent of applied potential. F cm-’, The impedances calculated by using CB= 1 . 1 X C,, = 1.0 X F cm-’, CA = 10-7-10-8 F cm-’, Rf = 104-105 9, and RA = 1.0 X 103-5.5 X lo3 9 for the CaFe204 electrode agreed with the experimental data as shown in Figure 7. Only RA and CAwere dependent on the frequency. C, obtained during the analysis was independent of potential and frequency as previously stated. This result suggests that the Fermi level pinning23 (1 9) Morrison, S. R. Electrochemistry at Semiconductor and Oxidized Metal Electrodes; Plenum: New York, 1980; Chapter 6 , pp 189-262. (20) Tomkiewicz, M. J . Electrochem. SOC.1979, 126, 2220. (21) ’t Lam, R. U. E.; Janssen, L. H. J. M.; Schoonman, J. Ber. Bunsenges. Phys. Chem. 1984, 88, 163. (22) de Haart, L. G. J.; Blasse, G. J . Electrochem. SOC.1985, 132, 2933. (23) Bard, A. J.; Bocarely, A. B.; Fan, F. R. F.; Walton, E. G.; Wrighton, M. S. J . A m . Chem. SOC.1980, 102, 3671.
i
CaFe204 , off,
3 7 “10022 -200\
lig t on
0. off 4
2001 0
10
21)
I
Tlme(hr)
Figure 9. Cathodic photocurrents as a function of the electrolyzing time at -0.35 V.
occurs by a surface state for the CaFe2O4electrode. The Fermi level pinning at the CaFez04electrode was proved by the following experiment. Au and Pt-Pd (1 5%) alloy films (1 50 A), whose work functions are 4.6 and 5.4eV, respectively, were attached to the electrode surfaces by the sputtering method (Ion Coater, Eiko Ltd.). A lead was then attached to a part of these films by a silver paste. The photovoltages at the interfaces of the Au/CaFe204 and the Pt-Pd/CaFe204 referenced to the CaFezOl were about -170 and -180 mV, respectively, when the light (300 mW/cm’) illuminated the interfaces. The light transmittances of both the Au and Pt-Pd films, which were measured by using metal (1 50 &/quartz, were about 20%. On the other hand, the photovoltages at the Au/Sr7FeloOzzand Pt-Pd/Sr7FeloOzz interfaces were -7.0 mV and nearly 0 mV, respectively, under the same conditions for the metal/CaFez04 electrodes. The large and the almost same photovoltages observed at the Au/CaFe,04 and the Pt-Pd/ CaFez04interfaces support the fact that the Fermi level pinning occurs at the CaFez04electrode. Therefore, it is concluded that, for the CaFez04 electrode, the rate-determining step for the photocathodic Hz evolution shown in Figure 3 is the electrochemical process in the Helmholtz layer. C,,, CA,and RA obtained for the Sr7Fe,oOzzelectrode depend on the potential; Rfwas found to be independent of the potential. RA and Rf were in the range of 175-320 9 and 1.0 X 103-7.0 X lo3 9, respectively, while CBwas nearly 0. The Mott-Schottky plots for C,, and CA are shown in Figure 8. It is obvious from this figure that both C, and CAobey the Mott-Schottky relation in the potential region from 0.1 to -0.5 or -0.3 V and show approximately the same flatband potential (0.1 & 0.1 V). Similar dependences of C, and CAwere also observed in 0.5 M KOH (pH 13.5), but the flatband potential was -0.4 V (Nernstian dependence of pH). This indicates that C,, and CAresult from charge accumulation at the semiconductor space charge layer and not from the surface state, as reported by Tomkiewicz who also saw the same behavior for the above capacitances at a TiO, electrode.” In Figure 8, C,-’ and CA-’ became constant in the potential region that was about -0.3 and -0.5 V more negative, respectively, at which the Fermi level pinning is expected to occur.
580
The Journal of Physical Chemistry, Vol. 91, No. 3, 1987
I
Matsumoto et al.
I /
I
-a3
-02
-01
I
o
Potentlal(Vvs SCE)
Figure 10. Tafel relations in the photocurrent-potential curves of CaFe204and Sr7Felo02,electrodes. Open and closed symbols denote Ptdeposited and bare samples, respectively.
Figure 9 shows the cathodic photocurrents of H2 evolution as a function of electrolyzing time at -0.35 V. The cathodic photocurrents are very stable for the CaFe204electrodes, but not so for Sr7Felo022electrodes. No cations of Ca2+, Sr2+, and Fe3+, which constitute the present oxides, were detected in the electrolyte after electrolysis. In the case of Sr7Fe10022, the photocurrent consisted of the following hydrogen absorption reaction into the lattice as well as the hydrogen evolution reaction.
+
Sr7Felo022 6 H 2 0 + 6e-
-+
Sr7Felo022-6[H]+ 6 0 H -
(2)
[HI denotes the absorbed hydrogen in the oxide lattice. When the light was shut off, the anodic dark current which will be the reverse reaction of the eq 2 flowed at the beginning, as shown in Figure 9. The decrease of the photocurrent for the Sr7Fei0022, thus, will be mainly based on the formation of an absorbed hydrogen layer, which may accelerate the recombination at the surface and/or bulk and inhibit the charge-transfer process of H2 evolution. Figure 10 shows the Tafel relationship of photocurrent vs. potential for the CaFe204and Sr7Fe10022 electrodes in the potential region from about -0.3 to 0 V, in which no corrections of IR drop by the electrode itself are necessary because of very small photocurrent. In this figure, the results measured at the Pt-deposited electrodes are also shown. pt was deposited on the surface under cathodic constant currents of 50 MAcm-2 for CaFe204and 3 mA cm-2 for Sr7Felo022for 3 s in M platinic chloride solution. The presence of Pt on the surface was confirmed by XPS (Shimazu, Ltd.). The photocurrents at the Pt-deposited electrode are always larger than those at the nondeposited electrodes, indicating that Pt acts as an electrocatalyst for H2 evolution. Tafel behaviors are observed for CaFe204electrodes. Their slopes were about 0.15 V/decade, independent of the Pt deposition, while the Sr7FeI0Oz2electrodes show no Tafel-like behavior. The value of the Tafel slope for the former electrodes are almost identical with the 0.12 V/decade for the cathodic H2 evolution at metal elect r o d e ~ indicating ,~~ that the potential drop varies only in the Helmholtz layer with the electrode potential. On the other hand, the photocurrents at the Sr7FeIoOz2electrodes will be determined by the bulk surface process as well as the electrochemical process.
Discussion In the structures of CaFe20425and Sr7FeloOz2,26 Fe is octahedrally surrounded by oxygen anions. Therefore, these band structures will be similar to those of other oxides having Fe in the lattice, such as a-Fe203. Fe304, and FeO. There are two different interpretations for the band structure in the band-gap transition (about 2.0 eV). One is the charge-transfer transition, Le., 2Fe3+ Fe2+ + Fe4+. This model was used in the discussion concerning the photoelectrochemical properties of PbFeizOlglo and ZnFe204.22The other is the transition from the e,l band to
-
(24) Bockris, J, O'M.; Reddy, A. K. Modern Electrochemistry; Plenum: New York, 1970, Chapter 10, pp 1141-1264. (25) Decker, B. F.; Kasper, J. S. Acta Crystallogr. 1957, 10, 332. (26) Asti, G.; Carbucicchio, M.; Deriu, A,; Lucchini, E.; Sloccari, G.J . Magn. Magn. Mater. 1978, 8, 65.
e a y i j e v
-2.04
+io/
I
j
1
I Anodic
Pt
'Bar
iJB) i0(W Current (log I )
Figure 11. Band structures of CaFe204and Sr7Fe10022, and the electron-transfer process at CaFe20, electrode.
the 4S(Fe) band which was proposed by Balberg and Pinch.27 The e,l band is formed by the crystal field splitting of iron s, d orbitals. This band structure explains the various spectroscopic results in spin polarized electron, cathodoluminescence, and soft X-ray ~tudies.~'Moreover, the photoelectrochemical properties of the Zn-Ti-Fe spinel oxidesI4 are also explained very well by this model, especially for the existence of the egcfand a],? hopping bands which are associated with the trigonal distortion of the lattice. Therefore, we used this model for the studied oxides, as shown in Figure 11. The E, values of the CaFez04 and Sr7FeloOz2(1 2-1.9 eV) 4S(Fe) as stated above. correspond to the transition of the e,l The e,,f and/or al,f hopping band will also be formed in the band gaps of both oxides as shown in Figure 11, where the flatband state at 0.1 V for the Sr7FeloOz2and the Fermi level pinning state at 0.2 V for the CaF%04 are illustrated. One can roughly estimate the band bending in the state of the Fermi level pinning for the CaFe,04, if a surface state contributing to the electron transfer does not exist in the energy position between the Ef and the E, (conduction band edge); that is, the photoproduced electron directly transfers from the E, level to the RHE level. It was assumed for this estimation that the overvoltage in the Helmholtz layer for the Hz evolution is 0.1 V, and that the E, is therefore the same as the R H E level (-0.6 V) at about 0.2 V, because the onset potential of the photocurrent during H2 evolution is about 0.1 V as shown in Figure 3. This assumption seems to be reasonable, since the overvoltage for the H2 evolution is not as large in the Helmholtz layer. Finally, it was estimated that the energy difference between the 4S(Fe) band edge and the Ef at the oxide surface is about 0.8 eV (that is, the energy difference between the top of the e,J valence band and the Ef is 1.1 eV because of E, = 1.9 eV) and the potential drop in the band bending is about 0.7 V, as illustrated in Figure 11. However, 0.7 V will be the maximum in the band bending, because we cannot prove the absence of a surface state in the energy position between the E, and the Ep It should be noted that the Ef is overlapping with the alBt hopping band for the CaFe204because of a 1.O-eV energy difference between the al,f band and the e& band.27 That is, the Fermi level pinning of CaFe204 is thought to be based on the overlapping of the Efand the a],? band at the surface. The Fermi level pinning of Sr7FeloOz2at -0.5 V may also be explained by this overlapping, if the energy difference between the a,,? band and the egl band is assumed to be in the range of 0.7-0.8 eV for Sr7FeI0Oz2.Thus, the al,f hopping band will always affect the photoelectrochemical properties of the Fe-containing oxides. If the CaFe204electrode potential is less than 0.2 V (more negative than -0.6 V for the 4S(Fe) conduction band edge), the variation in the potential drop occurs only in the Helmholtz layer and therefore the Tafel relationship (0.12 V/decade) in the photocurrent vs. potential is analogous to a metal electrode,24as follows
-
i, = n,k exp(AE,Fa/RT) (27) Balberg, I.; Pinch, H. L. J . Magn. Magn. Mater. 1978, 7, 12.
(3)
J. Phys. Chem. 1987, 91, 581-586 r
-
4WnmCui-tYier
lo0
0
10 Time (hr)
20
Figure 12. Photoelectrolysis of water by a Pt-deposited CaFe204(p type)/Znl,2Fe,,804(n type) assembly: (a) photocurrent-potential curves (n type) electrodes; of Pt deposited CaFe204(p type) and of Znl,2Fel,804 (b) photocurrents in short circuit of the p/n assembly as a function of light intensity; (c) photocurrents in short circuit as a function of electrolyzing time.
where n, is the concentration of the electron in the conduction band at the surface, k is the rate constant, A& is the potential drop in the Helmholtz layer, a is the transfer coefficient (a = 0 = 0.5, if the rate-determining step is not chemical desorption or electrodic de~orption~~), while the other symbols have their usual meanings. The energy position of the conduction band in the above potential region is illustrated by a dotted line in Figure 11. The
581
steps in the charge transfer at the bare and the Pt-deposited CaFe204electrodes in connection with the Tafel relationships with the same surfaces are also illustrated in the right side of this figure. In this case, the exchange current, io, for the H2evolution on the Pt-deposited surface is larger than that on the bare surface (kpt > kbarcin eq 3). Thus, the deposited Pt electrocatalytically accelerates the electrochemical process of the H2 evolution and eventually inhibits surface recombination. Finally, we measured the open circuit voltage, and the short circuit photocurrent as a function of the light intensity and as a function of the electrolyzing time in a Pt-deposited CaFe204(p type)/Znl,2Fel,804(n type)14 assembly. This is the photoelectrolysis of water without bias. The results are shown in Figure 12. The open circuit voltage under a 245 mW/cm2 light was about 0.4 V, which almost agrees with the difference between the onset potentials of the CaFe204 electrode (0.1 V) and of the (-0.3 V)I4 under the same light intensity. The short circuit photocurrent showed a linear dependence on light intensity (Figure 12b). The photocurrent decreased slightly for about 1 h and then stabilized as shown in Figure 12c. The same results were observed when light with a 400-nm cutoff filter was used. The efficiency of the energy conversion from the light to H2 and O2was roughly calculated to be in the case of the xenon lamp illumination. Some improvements in the CaFez04and Sr7Fel0OZ2 electrodes are necessary for higher energy conversion efficiencies during the photoelectrolysis of water. The main reason for the low efficiency is based on the small n, in eq 3 for the CaFe204 electrode. Therefore, the bulk recombination must be inhibited, although a small increase in this efficiency is brought about by the inhibition of the surface recombination by Pt. In the case of the Sr7FeloOz2 electrode, the surface reaction represented by eq 2 must be inhibited for long-term stability of the photocurrent. Registry No. CaFe204, 12013-33-1;Sr7Fe10022r 12346-10-0; Fe9Zn6OZ0,105500-77-4;Pt, 7440-06-4; H2, 1333-74-0;H20, 7732-18-5; K2S04, 7778-80-5.
Photochemlstry of Colloidal Cadmlum Sulfide at Dlhexadecyl Phosphate Vesicle Interfaces: Electron Transfer to Methylviologen and Colloidal Rhodium Hyeong-C. Youn,l* Yves-M. Tricot,lb and Janos H. Fendler*la Department of Chemistry and Institute of Colloid and Surface Science, Clarkson University, Potsdam, New York I3676 (Received: February 24, 1986; I n Final Form: August 7 , 1986)
Dihexadecyl phosphate (DHP)vesicles were used to in situ generate and stabilize CdS colloids and to adsorb methylviologen (MV2+). Formation of reduced methylviologen radical cation (MV") by CdS conduction band electrons (ece-) was followed by laser flash photolysis and steady-state irradiation. The quantum yield of MV" formation in optimal pH and CdS and MV2+concentration conditions was ca. 1.6% in the absence of hole scavenging. Efficient hole scavenging by benzyl alcohol under low-intensity (steady-state) irradiation increased the quantum yield up to ca. 50%. Benzyl alcohol was a much more efficient hole scavenger for outer- rather than inner-vesicle-surfaceCdS colloids, due to rate-limiting transmembrane diffusion of benzyl alcohol. Rhodium, an active H2-generation catalyst, was found to be 18 times more efficient at ece- capture than MV2+at or near the surfaces of colloidal CdS. Combination of laser flash and steady-state photolysis led to the estimation of the CdS-adsorbed MV2+concentration and to an apparent rate constant of approximately lo9 s-' for MV'+ formation. No transmembrane diffusion of MV2+ was observed unless the vesicles were heated above 60 " C .
Introduction In view of their utilization in solar energy conversion and photocatalysis, semiconductor colloids have received considerable attention in recent Cadmium sulfide is an n-type direct (1) (a) Present address: Department of Chemistry, Syracuse University, Syracuse, NY 13244-1200. (b) Present address: Department of Materials Research, The Weizmann Institute of Science, Rehovot 76100, Israel. (2) Fox, M. A., Ed. Organic Phototransformations In Nonhomogeneous Media; American Chemical Society: Washington, DC, 1985; ACS Symp. Ser.
278.
band gap semiconductor which has been widely because of its chemical simplicity and insensitivity to impurities. Its ~
~
~
(3) Gratzel, M., Ed. Energy Resources Through Photochemistry and Catalysis; Academic: New York, 1983. (4) Williams, F.; Nozik, A. J. Nature (London) 1984, 312, 21. (5) Fendler, J. H. J . Phys. Chem. 1985, 89, 2730. (6) Bard, A. J. J . Phys. Chem. 1982, 86, 172. (7) Moser, J.; Griitzel, M. J . Am. Chem. SOC.1984, 106, 6557. (8) Dimitrijevic, N. M.;Savic, D.; Micic, 0. I.; Nozik, A. J. J . Phys. Chem. 1984, 88, 4278.
0022-3654/87/2091-0581$01.50/00 1987 American Chemical Society