J. Phys. Chem. 1987, 91, 1173-1177
1173
Measurement of Electron Energy of Semiconductor Particles Dispersed in Aqueous Solution Takamasa Sagara,* Yoshihiro Aikawa, and Mitsunori Sukigara Institute of Industrial Science, University of Tokyo, Roppongi, Minato- ku, Tokyo 106, Japan (Received: June 26, 1986; In Final Form: October 15, 1986)
The electron energy of semiconductor particles under illumination was examined by measuring the electron injection rate from semiconductor particles dispersed in an aqueous solution into an electrode immersed in the solution. The electron injection current was formulated as a function of electrode potential, which agreed well with the observed photocurrent vs. potential curves for semiconductor particle dispersion systems, using metal or In203as a probe electrode. It was revealed that the illumination of Ti02, In2O3,or SnO, particles dispersed in a deaerated aqueous solution elevated their electron energy to a higher level compared to that under equilibrium in the dark. The resulting energy was high enough to reduce water to hydrogen in each case.
Introduction A bare or metal-deposited semiconductor particle has been known to show photocatalytic activity under illumination. One of the most remarkable reactions using such a catalyst is the reduction of water into hydrogen either in aqueous solution or in water vapor.]-' On a semiconductor particle, both of photoinduced oxidation and reduction occur at the same time. In such a case, photogenerated electrons react with oxidant and photogenerated holes with reductant, and hence, high reducing and oxidizing powers are respectively required for electrons and holes in order to achieve high efficiency of photocatalytic reactions. Consequently, a semiconductor with a deep valence band and shallow conduction band is advantageous for photocatalytic reactions, while a semiconductor with a small band gap is favorable for utilization of solar energy. In the case of the water-splitting reaction in an aqueous solution using semiconductor particles, an energetic condition is required so that electron energy of the conduction band edge is higher than the reduction energy of water (viz., t, > tH2H20) and also that the electron energy of valence band edge is iower than the oxidation energy of water (viz., tv < tH20i02).5-7 The positions of tc and e., vs. the redox levels of solution are fixed intrinsically under equilibrium in the dark under the given combination of semiconductor and s ~ l u t i o n . ~ *If~these ~ * levels are also fixed under illumination a t the same position, the reaction rate of electrons, u,, and that of holes, uh, will differ from each other. If u h is much larger than v,, electrons will accumulate on a Such accumulation of electrons on a small particle should make the electrical potential of the particle shift toward more negative because of the minute electrical capacity of the particle. This means that both tc and t, shift to higher energy with respect to the redox levels in the medium, and accordingly, u, will increase and u h will decrease. Thus, under continuous illumination, a nonequilibrium steady state where u, is equal to u h will be established. In this manner, it is expected that a semiconductor particle whose conduction band edge is even lower than tH1/H20in the dark might become able to reduce water under illumination. Since the (1) Sato, S.; White, J. J . Catal. 1981, 69, 128. (2) Kawai, T.; Sakata, T. Chem. Phys. Lett. 1980, 72, 87. (3) Dounghong, D.; Borgarello, E.; Gratzel, M. J . Am. Chem. SOC.1981, 103, 4685. (4) Kraeutner, B.; Bard, A. J. J. Am. Chem. SOC.1978, ZOO, 5985. (5) Morrison, S . R. Electrochemistry at Semiconductor and Oxidized Metal Electrodes; Plenum: New York, 1980. (6) Bolts, J. M.; Wrighton, M. S . J . Phys. Chem. 1976, 80, 2641. (7) Bard, A. J. Science 1980, 207, 139. (8) Butler, M. A,; Ginley, D. S . J . Electrochem. SOC.1978, 125, 228. (9) Aikawa, Y.; Takahashi, A.; Sukigara, M. J . Chem. SOC.Jpn. 1984, 292. (10) Aikawa, Y.; Takahashi, A,; Sukigara, M. J . Colloid Interface Sci. 1982, 89, 588. (1 1) Hauffe, V. K.; Volz, H. Ber. Bunsenges. Phys. Chem. 1973, 77, 967.
0022-3654/87/2091-1173$01.50/0
electrical potential of a particle under illumination determines the reactivity of photogenerated electrons and holes, it is important to examine the electrical potential of a particle under illumination for the discussion of photocatalytic activity of semiconductor particles. For the purpose of elucidation of the electric property of semiconductor particles dispersed in solution, extensive studies have been reported. Bard et al.12913demonstrated that the photogenerated electric charges on TiOz particles can be collected at an electrode immersed in the powder suspension. They interpreted the observed anodic photocurrent at the electrode potential to be more negative than the flat-band potential of TiO, in terms of the negative shift of energy level of electrons in the powder. Kawai et aI.l4 reported a similar result that the onset potential of photocurrent was more negative than the flat-band potential in the dark when semiconductor particles held down on an Au electrode were illuminated. We demonstrated the measurement of photocurrent as a function of electrode potential at both semiconductor and metal electrodes immersed in a semiconductor particle dispersion.I5J6 In this paper, we describe the method of quantitative measurement of the electrical potential of semiconductor particles dispersed in a solution under illumination by means of a probe electrode and present the results of the measurement for three kinds of metal oxide semiconductor particles.
Experimental Section Materials. Semiconductor particles used in this work are listed in Table I. Centrifugal separation in water was applied to achieve homogeneity of particle size. The mean particle size and size distribution were confirmed by use of scanning electron microscopy (Akashi, Model ALPHA-25A). Water was deionized and then distilled. All other reagents were analytical grade and used as supplied. Preparation of Semiconductor Particle Dispersions. The semiconductor particle dispersions were prepared by dispersing 30 mg of semiconductor particle in 100 mL of medium containing 1 M HC104 or 1 M HCOOH. The pH of each dispersion was adjusted to be 0.4. Before each measurement, the semiconductor particle dispersion was stirred vigorously in an ultrasonic vibrator. The semiconductor particle dispersions used in this work are listed in Table I. Preparation ofBlectrodes. Each of four kinds of metal electrodes, Pt, Au, Ag, and W, was polished and then cyclic potential (12) Dunn, W. W.; Aikawa, Y.;Bard, A. J. J . Am. Chem. SOC.1981, 103, 3456. (13) Dum, W. W.; Aikawa, Y.; Bard, A. J. J . Electrochem. SOC.1981, 128, 222. (14) Fujii, M.; Kawai, T.; Kawai, S. Chem. Phys. Lett. 1984, 106, 517. (15) Aikawa, Y.; Sagara, T.; Sukigara, M. Denki Kagaku 1985,53,638. (16) Sagara, T.; Aikawa, Y.; Sukigara, M. Denki Kagaku 1986, 54, 177.
0 1987 American Chemical Society
1174
The Journal of Physical Chemistry, Vol. 91, No. 5, 1987
Sagara et al.
TABLE I: Metal Oxide Semiconductor Particle Dispersions Used in the Present Work
code no. DSP.T-1 DSP.T-2 DSPS DSP.1
semicond material TiO, (rutile) Ti02 (rutile) SnO, In203
“At 730 ‘C.
i
L
pretreatment vacuum reducedu vacuum reduced” separated centrifugally separated centrifugally
mean dia/rm
dispersing medium 1 M HC10, 1 M HCOOH 1 M HCOOH 1 M HCOOH
0.05 0.05 1.O
0.1
Torr for 1 h.
compartment B
I
supplied by Japan Aerozil (P25) Japan Aerozil (P25) Wako Wako
C
I
compartment
’j I
A
R N2
W
1
,Ag/AgCI
e- V.,
electrode
,
........
be electrode
salt b4dge
1 stirrer
1
&iconductor particle dispersion
Figure 1. Experimental cell used for electron energy measurement.
sweep was performed repeatedly in a deaerated 1 M HC104 aqueous solution before use. An n-type indium oxide (In203)thin film on glass substrate was supplied by Stanley Electric Co., and its surface conductance was lW3 S/o donor density 4 X 10” An ohmic contact was established through indium solder. The surface area of each electrode was 1 cm2. For the purpose of the measurement of flat-band potentials, V, the electrodes of TiO, (vacuum-reduced rutile single crystal), SnOz(thin film), and In203 (thin film) were also used. Apparatus. The cell was composed of probe electrode compartment (compartment A), counter electrode compartment (compartment B), and a salt bridge (agar with saturated NaC104) by which the two compartments were electrically connected, as shown in Figure 1. A potentiostat (Hokuto Denko, Model HA301) was used to measure current-potential characteristics. The shape of compartment A was cylindrical with a diameter of 5 cm and the probe electrode was set 2.0 cm apart from the center of the compartment. The semiconductor particle dispersion was always stirred at 180 rpm around the center of the compartment. Before and during the measurement, pure N, gas was bubbled in compartment A continuously. The semiconductor particle dispersion was illuminated with the light from a Xe lamp (Ushio, Model UI-SO/c) through an IR cut filter (Toshiba, IRA-25s). For the measurement of photocurrent under monochromatic illumination, a monochromator (Jobin Yvon, Model H-20) was employed.
Theory We are concerned with a system where an electrode is immersed in an aqueous solution in which semiconductor particles are dispersed. We assume that photogenerated holes in a particle possess high enough reactivity to oxidize reductants in the solution rapidly, while photogenerated electrons accumulate on the particle because of their lower reactivity. When these particles collide with the electrode, electrons accumulated on the particles may transfer to the electrode when the electrode potential is sufficiently positive. This electron-transfer process can be observed as an anodic photocurrent. We define tp as the highest energy of electrons which can be supplied from the semiconductor particle to oxidants in medium with respect to the equilibrium electron energy of the reference electrode in the same medium. As illustrated in Figure 2, the potential of the particle in the steady state is Vp = cp/e-, where e- is the electron charge with minus sign. When the electrode potential, V,, is set more positive than Vp,electrons will be injected from the particle into the electrode, and an anodic photocurrent will be observed. On the other hand, when V, is more negative than V,, electrons cannot be injected to the probe electrode. Thus,
Figure 2. Schematic electron energy diagram for the interface between
the probe electrode and the semiconductor particle dispersion under illumination.
i
arriving at surface
-e-Vp( t = 0 1
...k.e-v
(t = t
\1 e- VCct = t ) -
e- V
R
:j
..........
’$1
leaving from surface
electrode dispersion Figure 3. Model for the electron-transfer process between the probe electrode and the colliding particle. one can obtain Vp in principle by measurement of the onset potential of anodic p h ~ t o c u r r e n t . ~ ~But, J ~ J ~in practice, it is always difficult to determine the onset potential of photocurrent without arbitrariness because the dark current is much larger than the photocurrent near the onset potential, particularly when one uses a metal electrode. We should, therefore, formulate a method to get the actual photocurrent as a function of electrode potential so as to obtain Vp more precisely. We assume that each of the particles colliding with the electrode has the steady-state electron energy under illumination, tp = e-Vp, and following the collision it stays on the electrode for tc. As shown in Figure 3, a particle with e-Vp arrives at the surface of the probe electrode at t = 0, and loses its excess negative charge until it leaves the electrode a t t = t , when the electron energy of the particle is e-V,. If the rate of electron injection is governed by the electron-transfer step from particle to the electrode, the electron injection current is written as i = -dQ = C - dV dt dt
where Q is the charge on a particle, C the capacitance of the particle, and Vthe electric potential of the particle at t. Assuming that the above current will also be described in terms of a Butler-Volmer rate equation, the current is written as
where 9 = V, - Vis the overpotential at t , cy the symmetric factor (17)
Chojnowski, F.; Clechet, P.; Martin, J.-R.; Herrmann, J.-M.; Pichat,
P. Chem. Phys. Lett. 1981, 84, 5 5 5 .
The Journal of Physical Chemistry, Vol. 91, No. 5, 1987
Electron Energy of Semiconductor Particles of electron transfer, io the exchange current, k the Boltzmann constant, and T the temperature. We will consider three cases: a = 0,0.5, and 1. Introducing y = exp(eg/kT), and integrating eq 2 from t = 0 to t = t,, one gets
a. a = 1
0
(3) where y, = exp(e(VR- V , ) / k T ) and y , = exp(e(VR - V , ) / k T J . In the case of a = 1, eq 3 is rewritten as -y{’) = 1-
(1 -J’;’)/(l
K
(4)
When I(V, - V,)/VRl
