Kenneth J. Miller
Northeast Louisiana University Monroe, Louisiana 71201
An Introduction to Semiconductor Surfaces as Catalysts
Because most of the research effort in semiconductors has been directed towards device technology and theory, the role of semiconductors as catalysts has been largely overlooked. Chemistry students have t,ypically not had an opportunity in course work to become familiar with semiconductor surface chemistry, and chemists, in general, have not, therefore, been in a position to utilize new solid state surface theory and material development. Hannay's recent (1967) text (1) presents a brief introduction to the subject of semiconductor catalysis and Vol'kenshtein ( 2 ) , Law (S), and Zettlemoyer and Iyengar (4) present more detailed reviews. The general subject of semiconductor surfaces has been discussed by a number of authors (5-10). This present paper will review from a pedagogical point of view some of the fundamentals of semiconductors as t,hey relate to catalysis on semiconductor surfaces. Students who have had little familiarity with solid state chemistry may find this paper a stimulus for furt,her study of greater depth, and it may also provide a small bridge across the mental gap separating chemistry and physics. Semiconductors provide ideal surfaces for catalytic study because they have simple conduction mechanisms to study charge t,ransfer during chemisorption, their crystalline-interior can be purified typically to impurity concentrations of parts per billion, single crystals are readily available, and because electrical measurements can be easily made on both their crystalline bulk and on their surfaces. The semiconductor surface, therefore, provides the best type of solid surface as compared with eit,her conductors or insulators for studies of chemisorption which leads to reactions which either do not take place in the gaseous state or which take place with less efficiency in the gaseous state. Fundamental to the theory of semiconductor catalysis is the degree to which hulk semiconductor properties may be related to surface properties. Theory must also question the nature of catalytic "act.ive centers" (11, 12) or "active sites" on solids as first proposed by Taylor (IS) and whether they are indeed synonomous with "surface states," a term applied by Bardeen (14) in 1917 to semiconductor surfaces, following theoretical treatment by Tamm (16) and Shockley (16). The t,heory of surface states gave birth to what can be termed modern semiconductor surface chemistry. Surface states may be defined as localized electronic energy levels a t the semiconductor surface. Presented st the Southead-Southwest Combined Regional American Chemical Society Meeting, New Orleans, Louisiana, December 1970. I t is a pleasure to acknowledge the support of this study by a Dupont Small Grants Award, awarded through the Division af Chemical Education of the American Chemical Society.
582 / Journol o f Chemical Education
It is on the electronic theory of surface states which concepts of semiconductor catalysis must he based and perhaps much of metal catalysis since the real surfaces of metals may be composed of films of metal oxide defect semiconductors (17). Discussion in this paper primarily relates to heterogeneous catalysis of gases on surfaces of the elemental semiconductors germanium and silicon because the properties of these semiconductors have become particularly well-characterized and provide good examples for discussion. However, compound semiconductors such as NiO, CuzO, and ZnO, whose semicoaducting properties depend on a nonstoichiometric defect structure, can also be described by sim~lartheory. Electron energy hand statistics (18,19) is a necessary tool to use to understand the theory of semiconductor catalysis. Figure 1 shows the well-known FerrniDirac statistical electron distribution function P ( E ) = (1
+ es-W"T)-1
as related to pure (intrinsic) semiconductors, as a function of energy, E, for several temperatures. For
Figure 1. The Fermi-Diroc distribution function probability of filling electron energy levels or a function of energy of levels for on intrinsic remiconductor.
semiconductors at room temperat,ure approximately fikT E, SOkT, where kT is approximately 0.026 electron volt, and E , represents the magnitude of the energy band separation. The lower and upper limits of E, given approximately differentiate metals and insulators, respectively, from semiconductors. In Figure 1, EP is the reference energy level about which the probability of filling an electron level is symmetrical, and may also he defined as the electrochemical potential, P = ( t W / d N ) T , Pwhere , N is the total number of electrons and P is the free energy. It is convenient to consider such electrochemical potentials for adsorbate and semiconductor surfaces during chemisorption leading to catalysis. In order for electronic equilibrium to be reached in chemisorption, t,he electrochemical potentials (Fermi levels) of the adsorbate and the semiconductor surface must he equal. Electronic equilibrium is achieved by electrons transferring from states of higher to lover potential energy in the adsorbate-semiconductor systems. The shaded areas in Figure 1 represent the room tenlperature concentrations of electrons in the conduction band, whose lover edge is E,, and holes in the valence band, whose upper edge is E,, respectively. In an intrinsic semiconductor these carrier concentrations can be approximated from E , using the relation given by Shive (20) % = = n ,. = - 4.84 X 10~6TB1ae-%mT particles c m P