“…They associated this effect with incorporation of residual As in the Si interface layer. 7 A possible explanation of the above discrepancy was provided by our recent photoemission and transport results from Al/Si/GaAs͑001͒ diodes, 8,9 which showed that the As and Al fluxes required during Si deposition to produce an effect are much larger than those commonly employed for Si doping purposes, i.e., comparable with, or higher than the Si flux, and presumably higher than those employed by Koyanagi et al 7 Most empirical 6,10 and theoretical models 7,11,12 proposed to explain subsets of the above experimental results consider the electronic structure of the Si interfacial layer as identical to that of bulk Si, require the presence of an As-doped ͑Ga-doped͒ n ϩ (p ϩ ) degenerate Si layer of sufficient thickness at the interface, e.g., to justify tunneling, 6 compensation, 7,10 or metallic screening of the interface states, 12 and assume that the band alignment across the Si-GaAs heterojunction in the Al/Si/GaAs structure is independent of the Si layer thickness. Conversely, no substantial change of the Schottky barrier relative to the Al/n-GaAs͑001͒ case was reportedly observed in Al/Si/n-GaAs͑001͒ structures in which Si layers 10-40 Å thick had been grown under Ga flux.…”