The unified model for Schottky barrier formation and MOS interface states in 3–5 compounds

Abstract

The object of the work reported here has been to develop an understanding on an atomic basis, of the interaction between semiconductor and metals or oxygen overlayers which determine the electronic characteristics of the interface, e.g. the Schottky barrier height of metal-semiconductor interface or the density and the energy position of oxide-semiconductor interface states. The principal experimental tool used has been photoemission excited by monochromatized synchrotron radiation (10 < hv < 300 eV). Extreme surface sensitivity is obtained by tuning the synchrotron radiation so that the minimum escape depth is obtained for the excited electrons of interest. In this way only the last two or three atomic layers of the solid is sampled. By changing hv, core levels or the valence bands can be studied. Using a metallic reference, the Fermi level position at the surface, E Fs, can be directly determined. GaAs, InP and GaSb have been studied. On a proper cleaved surface, there are no surface states in the semiconductor band gap; thus, no pinning of E Fs. Pinning of E Fs can then be monitored as metals or oxygen are added to the surface — starting from sub-monolayer quantities. Two striking results are obtained: (1) pinning position is independent of adatom for oxygen and a wide range of metals, and (2) the pinning is completed by much less than a monolayer of adatoms. These results can not rationally be explained by the pinning being due to the levels produced directly by the adatoms. Rather, they suggest strongly that the adatoms disturb the semiconductors surface forming defect levels. This is supported by appearance of the semiconductor atoms in the metal and by the disordering of the semiconductor surface by sub-monolayer quantities of oxygen. Since these basic experiments have been reported previously they are only reviewed briefly here. When metals or oxygen are added under very gentle conditions, the following levels are formed (all energies relative to the conduction band minimum). III–V Acceptor (missing atom) Donor (missing atom) GaAs 0.65 eV (As) 0.85 eV (Ga) InP 0.45 eV (P) 0.1 eV (In) GaSb 0.5 eV (Sb) below VBM (Ga) These results explain why Schottky barrier gates will provide useful FET's on n-GaAs but not n-InP. Likewise they predict that MOS or MIS gates will be practical for n-InP but not n or p GaAs. Studies of the oxygen surface chemistry find the As oxides to be unstable and P oxides to be stable — reinforcing the prediction. Recent work of others is reviewed and alternate identification of the missing atoms in the defects is discussed. Some of the new process possibilities opened up by this work are considered.