Electronic structure of the Si/GaP(110) interface

We present the results of an angle-resolved photoemission study of the Si/GaP(110) heterostructure. Band mapping of sputtered and annealed GaP(110) unambiguously determines the position of the valence-band maximum at Γ₁₅ to be 1.2 eV below the Fermi level. The Fermi-level pinning position for the sputtered and annealed surface is different from that of cleaved GaP(110) and suggests the introduction of defect levels by the cleaning process. However, deposition of one monolayer (ML) Si onto GaP(110) changes the pinning position despite the defect levels. The electronic structure of the interface evolves slowly as a function of the Si-overlayer thickness, but appears to be fully developed at 20 ML. We discuss the electronic structure of the heterojunction in light of the midgap energy rule for band alignments and try to clarify why it appears to fail for this particular system. Specifically, a comparison of theoretical and experimental band structures shows that a linear extrapolation of the leading edge of an angle-resolved valence-band spectrum does not determine the GaP Γ₁₅ valence-band maximum with the accuracy ({~}0.1 eV) desired for band-alignment measurements. This shortcoming conjures up questions about the validity of using the leading edge of angle-integrated spectra to determine the Γ₁₅ valence-band maximum in GaP and about the previously reported band alignment for the Si/GaP(110) heterostructure.