Different Fermi-level pinning behavior on<i>n</i>- and<i>p</i>-type GaAs(001)

Abstract

Scanning-tunneling-microscopy studies of both the n- and p-type GaAs(001)-(2\ifmmode\times\else\texttimes\fi{}4)/c(2\ifmmode\times\else\texttimes\fi{}8) surfaces show important differences in the Fermi-level pinning behavior of n- and p-type material. It has been shown previously that Fermi-level pinning on the n-type GaAs(001) surface results from the formation of kinks in the dimer-vacancy rows of the (2\ifmmode\times\else\texttimes\fi{}4)/c(2\ifmmode\times\else\texttimes\fi{}8) surface reconstruction. These kinks form in the required number to pin the Fermi level close to midgap at all doping levels. On p-type GaAs(001) we now show that no similar surface donor state forms. As a result, at high p-doping levels (${10}^{19}$ ${\mathrm{cm}}^{\mathrm{\ensuremath{-}}3}$ Be), the Fermi level determined by tunneling spectroscopy is found to be within 150 meV of the valence-band maximum. At lower p-doping levels the Fermi level moves towards midgap as determined by the density of ``intrinsic'' surface defects such as step edges, and missing unit cells.