FUNDAMENTAL-STUDIES OF III-V SURFACES AND THE (III-V)-OXIDE INTERFACE

Work using photoemission with synchrotron radiation (8 eV < hv < 300 eV) to investigate chemisorption and the formation of thin oxide layers on GaAs is reviewed and preliminary correlations with thick oxides are made. An advantage of synchrotron radiation is that it allows examination of both core and valence electrons of the last few molecular layers of the solid. By studying both the core levels and the valence band as a function of oxygen exposure, insight can be obtained into the bonding of oxygen and into oxide formation on an atomic scale. It is found that oxygen can bind to the Group V elements on the (110) surface without the necessity of breaking III-V bonds. Thus, it is suggested that optimum bonding of oxides or other insulators to III-V surfaces will be through the Group V elements. The stability of the surface against the breaking of III-V bonds and the formation of bulk oxides is found to correlate very well with the heat of formation of the III-V compound. For example InP is more stable than GaAs. This appears to correlate with the ease of forming metal-oxide-semiconductor (MOS) or metal-insulator-semiconductor (MIS) type devices. Even when it is possible to bond oxygen to a surface without the necessity of breaking III-V bonds, Fermi level pinning is produced by submonolayer quantities of oxygen. This is associated with defects induced by the strain produced by oxygen chemisorption. The pinning position is near the conduction band minimum and near mid-gap for (110) InP and (110) GaAs surfaces respectively. A striking correlation is found between these pinning positions and those found for thick oxides on practical device structures. This suggests that the pinning mechanism may be the same for both types of interfaces. It is suggested that lattice defects play a key role in the pinning. A novel approach to forming the oxide or insulation layer for MOS or MIS structures is suggested. © 1979.