J. Phys. Chem. 1984, 88, 6096-6097
6096
A Mechanism of Electron Hole Pair Separation in Illuminated Semiconductor Particles H. Gerischer Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-I 000 Berlin 33, West Germany (Received: September 12, 1984)
The photocatalytic action of illuminated semiconductor particles in electrolytic media is based on electron hole pair separation followed by cathodic and anodic charge transfer reactions with components of the surrounding electrolyte. The conditions of efficient charge separation are discussed, and it is shown that the large increase of the photocatalytic efficiency found in some cases by locally contacting the particles with suitable metals can be caused by the formation of an accumulation layer at this contact.
The Problem Photocatalytic processes induced by illumination of a suspension of small particles of n-type semiconducting oxides (like TiO,, SrTi03, W03, or Fe203)and sulfides (like CdS and ZnS) have found wide attention in the past years.'-12 The generally accepted mechanism is that electron hole pairs are created in the particles which react separately on the surface. The particle acts like a microelectrolysis cell, with different parts of the surface forming the anode and the cathode. A particular advantage of these systems has been attributed to the fact that the photogenerated electrons and holes reach the surface very quickly and can be trapped there before they recombine. These trapped charge carriers can afterward carry out the anodic and cathodic reactions. Water splitting or sacrificial hydrogen production from water by parallel oxidation of a suitable compound is one of the most attractive goals. If this charge separation by trapping on the surface would be a merely statistical process, one can hardly imagine that it could reach a high efficiency. The anodic and cathodic reactions usually need more than one charge carrier of the same quality, and therefore a great deal of the electron hole pairs must be lost by recombination before it happens that a second carrier of equal charge will reach the same site on the surface. Electrostatic interaction between the trapped carriers and the free ones in the bulk will further favor the recombination relative to the accumulation of equal charge carriers at the same sites, even if the electrostatic forces may be shielded to a large extent by a counter charge from the surrounding electrolyte. Reported efficiencies on pure semiconductors are indeed very low. Reactions in which the cathodic process is hydrogen evolution can, however, be much enhanced by a partial coverage of these particles with metals of the platinum group.5-10 It appears that this treatment of the semiconductor particles not only catalyzes the hydrogen evolution reaction on the surface but also induces an electron hole pair (1) S.N. Frank and A. J. Bard, J . Phys. Chem., 81, 1484 (1977). (2) H. Yoneyama, Y. Yamashita, and H. Tamaru, Nature (London), 282, 817 (1979). (3) F. T. Wagner and G. Somorjai, J. Am. Chem. SOC.,102, 5494 (1980). (4) S.Sato and J. M. White, J . Phys. Chem., 85, 592 (1981). (5) T. Kawai and T. Sakata, Chem. Phys. Lett., 72, 87 (1980). (6) J. M. Lehn, J. P. Sauvage, and R. Ziessel, Nouu. J. Chim.,4, 623 (1980). (7) M. Gratzel, Ber. Bunsenges. Phys. Chem., 84, 981 (1980). (8) K. Kalyanasundaram, E. Borgarello, and M. Gratzel, Helu. Chim. Acfa, 64, 362 (1981). (9) D. Duonghong, J. Ramsden, and M. Gratzel, J. Am. Chem. Soc., 104, 2977 (1982). (10) J. M. Lehn, J. P. Sauvage, R. Ziessel, and L. Hilaire, Isr. J . Chem., 22,'168 (1982). (1 1) A. Henglein in "Photochemical Conversion and Storage of Solar Energy 1982", Part A, J. Rabani, Ed., The Weizmann Press of Israel, Jerusalem; 1982, p 115. (12) See also contributions of K. Kalyanasundaram and J. Kiwi or T. Sakata and T. Kawai in "Energy Resources through Photochemistry and Catalysis", M. Gratzel, Ed., Academic Press, New York, 1983.
0022-3654/84/2088-6096$01.50/0
separation mechanism which exceeds by far the statistical separation probability. Mechanism of Electron Hole Pair Separation The well-known electron hole pair separation mechanism in compact semiconductor electrodes operating in photoelectrochemical solar cells is based on the formation of a depletion layer at the illuminated semiconductor-electrolyte contact (Schottky barrier13). In these cases, the semiconductors must have a sufficiently high n-type conductivity in order to create electric fields in the space charge layer of a sufficient magnitude for electron hole pair separation and to prevent Ohmic losses in the bulk of the semiconductor during the photocurrent passage. In a small particle, however, with a size being smaller than the penetration depth of the light, such a doping level would have the opposite consequences for the efficiency of electron hole pair separation by the following reasons. Only the minority carriers can escape up to the surface while majority carriers accumulate in the bulk of the particle. The Schottky barrier is, however, decreased by the counter voltage due to this initial charge separation and now recombination will increase drastically. As long as the majority carriers cannot leave the bulk of the particle being surrounded by a Schottky barrier, there is no chance for the opposite redox reaction electrically compensating the reaction of the minority carriers. Different reaction rate constants of the anodic and cathodic process, if both carriers finally can reach the surface, must be compensated in the steady state by a different surface concentration of these carriers. The efficiency will remain small, even if we neglect the back-reactions of the products in the electrolyte with the opposite charge carriers. Impregnation of the surface with two different catalysts (e.g. platinum and ruthenium d i ~ x i d e ~can - ~ ) increase the rates of the two separate electrode reactions locally and therefore reduce the surface concentration of electrons and holes and herewith the probability of recombination to some extent. But one can hardly imagine that a high efficiency could be reached in this way. Two conclusions can immediately be drawn from these considerations. In order to obtain a high reaction rate by the two kinds of charge carriers on the same particle, recombination in the bulk must be kept low. This is only to be expected if the particle is nearly intrinsic. The other point is that one needs a mechanism for electron hole pair separation which discriminates different parts of the surface for the trapping of electrons or holes. Such a process seems to be induced by the impregnation of such particles with noble metals of the platinum group. By assuming that the surface is only partially covered with such metal islands, one can indeed derive such a mechanism from first principles. Aspnes and Heller14have recently demonstrated that the contact between metals of the platinum group and various n-type semiconductors becomes Ohmic, if these metals are exposed to hy(13) H. Gerischer, J. Electroanal. Chem., 58, 263 (1975). (14) D. E. Aspnes and A. Heller, J . Phys. Chem., 87, 4919 (1983).
0 1984 American Chemical Society
The Journal of Physical Chemistry, Vol. 88, No. 25, 1984 6097
Letters
n
Figure 2. Qualitative picture of the electric potential distribution in spherical particles of nearly intrinsic semiconductors with electron injecting metal islands on their surface. The lines inside the spheres represent the intersection of equipotential planes with a plane through the center of the sphere.
electrolyte 1
O x y e-* eRed
e
d
*
X
Figure 1. Energy diagram for a thin metal-coated semiconductor mem-
brane between two electrolytes where the metal creates an accumulation layer in the semiconductor by electron injection. drogen. The interaction with hydrogen reduces the work function of these metals, obviously by the formation of a dipole layer between the chemisorbed hydrogen atoms and the surface or interface of these metals. In this way the work function can approach the electron affinity of the conduction band of several semiconductors. This prevents the formation of a Schottky barrier which otherwise would be created at such contacts due to their high work function in the pure state.14 If the same experiment would be done with an intrinsic semiconductor or a semiconductor with a very low doping level, the hydrogen-saturated metal would have the smaller work function and consequently electrons will be injected into the semiconductor. At the metal-semiconductor contact, an accumulation layer of electrons will be formed in which a potential gradient exists which is very steep next to the interface. This potential gradient will drive electrons to the metal and repel holes from this part of the interface. The Model for Charge Separation The extension of such an accumulation layer at a contact with a compact semiconductor is controlled by the Debye length, LD, of the semiconductor, which for a very lowly doped semiconductor will be comparable with or larger than the dimensions of such small particles.
where no is the electron concentration in the bulk, t is the relative dielectric constant, to is the permittivity of free space, and eo is the elementary charge. The situation is qualitatively depicted in Figure 1 for a thin plate of infinite extension in the yz direction separating two electrolytes. The metal is supposed to be on the one side in equilibrium with a redox electrolyte and on the other side in electronic equilibrium with the semiconductor while the semiconductor contact with the second electrolyte shall be impermeable for electric charge carriers. In order to enable the application of the potential distributions which have been derived for accumulation layers of semiconductors with surface state^,'^*'^ it is assumed that the thickness may be larger than the Debye length. If light is absorbed in the semiconductor, the electric field in the accumulation layer will drive the electrons to the metal and the holes to the opposite side of the plate as indicated in Figure 1. This separation mechanism will be very efficient if the diffusion length of both charge carriers is larger than the thickness of the plate. The charge separation process will continue until the counter voltage created by the accumulation of the minority carriers at (15) R. H. Kingston and S . F. Neustadter, J . Appf. Phys., 26,718 (1955). (16) C. E.Young, J . Appl. Phys., 32, 329 (1961).
the semiconductor back contact has decreased the electric field strength to such an extent that recombination compensates further electron hole pair generation. This situation corresponds to an open circuit photovoltage between the metal and the backside of the semiconductor. Experiments are on the way to demonstrate this consequence. The field strength at the contact between the metal and the semiconductor has been calculated by Kingston and Ne~stadter.'~ It can for our case be expressed in the modified form of
(g)x=o
= 21/2-F(A9) kT
eOLD
= -F(A+) 0 036 LD
(2)
with (3) where A 9 is the difference in work function between the semiconductor and the metal before contact has been made. With €/no= and A+ = 4 kT, one obtains from eq 1 and 2 a field strength of 23.4 V cm-' at the interface; for A@ = 8 kT, the result is 1273 V cm-'. These figures demonstrate that such an accumulation layer can exert a very efficient force for electron hole pair separation. The electric field distribution in a spherical particle with a small metal contact on its surface is geometrically much more complicated, but an analogous driving force for electron hole pair separation will be created if the metal injects electrons into this particle. In the absence of surface charges on the areas in contact with the electrolyte, the distribution of the electric potential in a spherical particle of T i 0 2 or similar materials should look like shown qualitatively in Figure 2 for a particle with one or two metal islands on its surface. In reality, the injected electrons may be trapped in acceptor states inside the particle. This can only result in a higher charge density and larger field strength next to the metal contact and should make the electron hole pair separation even more efficient. The role of the electrochemical reactions on the platinum and on the uncovered T i 0 2 surface with electrons and holes, respectively, is to prevent the accumulation of these charge carriers there and to keep the recombination rate small. While hydrogen evolution on the Pt islands has a large enough rate constant, oxygen formation on the oxide surface is slow or the reaction of holes with water may result primarily in adsorbed intermediates. Impregnation of the TiOz surface with RuO, can increase the rate of O2 formation and in this way the efficiency of the net photochemical If this model is correct and the electron hole pair separation can be as efficient as it indicates, the low efficiency of the overall process of water splitting by particulate systems is due to the electrochemical back-reaction of the oxidized product (0, or H202)at the metal islands while the back-reaction of the hydrogen on the oxide surface should be negligible. An increase of the efficiency depends therefore on finding means to prevent the back-reaction on the metal islands.
