Conductance-frequency measurements down to temperatures of 100 K have been performed on both p-type and n-type (100) silicon oxidized in dry oxygen at 900 "C. The metal electrode capacitors used were not given a postmetallization anneal in forming gas. This has allowed measurements of the intrinsic density of states, capture cross section. and surface potential fluctuations to within 0.06 eV of the band edges. Two peaks in the defect density at energies of 0.3 eV and 0.85 eV above t h e valence band are clearly visible above a n asymmetric background, which rises rapidly towards the conduction band edge. The capture cross Section is near constant at -10" cm2 across the gap and independent of temperature. The surface potential fluctuations reveal a peak value of -70 meV centred at 0.4 eV above the valence band superimposed on a constant background of -40 meV. W e attribute the peaks in the density of states to the amphoteric trivalent silicon P, centres. The probable causes of the asymmetric background are either the tail of a defect peak centred around the conduction band edge, or states descendina from t h e conduction band induced bv stress within the oxide.

Capacitance and conductance measurements are presented for dc-driven Au/ZnS:Mn/p-Si electroluminescence metal-insulator-semiconductor (MIS) devices, where the ZnS:Mn films are deposited by radio frequency sputtering. Stable dc operation is achieved by introducing oxygen into the film during deposition and subsequently annealing. The effect of the post-deposition annealing upon the density of states at the ZnS:Mn/p-Si interface is investigated. As deposited, the devices show unusual MIS C-V characteristics, that indicate a very high interface state density. Annealing at 700 °C, normal C-V characteristics are observed, indicating that the very high density of states is greatly reduced. For these films the conductance technique has been used to measure the density of states at the interface between the ZnS:Mn and p-Si. The statistical model is found to describe most accurately the interface state conductance response. The interface state density consists of a tail of states that varies between 3.7×1013 cm−2 eV−1 at the silicon Fermi level and 1.1×1013 cm−2 eV−1 at the silicon mid-gap. A small peak is superimposed upon this tail at (−0.16±0.01) eV below mid-gap. The tail of states is believed to be intrinsic to the ZnS:Mn/p-Si interface, but evidence suggests that the small peak is due to the presence of oxygen, which is shown by secondary-ion mass spectrometry analysis to accumulate at the interface after annealing at 700 °C. It seems likely that the very high density of interface states in as deposited devices is a consequence of a plasma damage to the silicon surface during growth, creating defects such as silicon dangling bonds. One possible explanation for the decrease in this density is that by annealing at 700 °C, oxygen in the bulk of the film diffuses to the interface, where it mops up these defects by forming compounds such as SiOx. A simpler model of interface recrystallization is also suggested. The doping density in the depletion region of the silicon is calculated as (7.5±0.5)×1014 cm−3, and the interface state capture cross section for holes is found to have mean value of approximately 10−15 cm−2.

No abstract

Conductance measurements have been performed on p-Si/SiO2 metal-oxide-semiconductor capacitors fabricated by thermal oxidation of the silicon in three different reactors under different conditions in three different commercial systems. In only one case (system 3) did we find the very broad conductance curves usually associated with p-Si/SiO2 junctions, the others from systems 1 and 2 exhibiting normal interface state response. However, in the sample from system 2, the response of bulk states inside the depletion region was found to distort the Gp/ω spectra, while the sample from system 1 showed no such response until annealing in H2/Ar gas at 500 °C reduced the interface state density to such a level that the bulk state responses were visible. We argue that the presence of bulk states probably explains the very broad Gp/ω curves observed by us, and further propose that some such mechanism accounts for the historical difficulties encountered with the conductance technique on p-Si/SiO2 capacitors.