M.-H. Tsai

N vacancies in AlxGa1−xN

Predictions of deep levels associated with N vacancies in AlxGa1−xN as functions of alloy composition x explain both (i) the dramatic change from naturally n‐type to semi‐insulating behavior (for x=xc≂0.5) in terms of a shallow‐deep transition for the vacancy’s T2 level, and (ii) the major photoluminescence feature in terms of recombination from the vacancy’s A1 deep level. Extrinisic photoluminescence data for Zn‐doped AlxGa1−xN are interpreted in terms of a transition from the conduction band to a T2‐symmetric deep level in the lower part of the gap. This level is associated with antisite Zn on a N site, ZnN.

Dependence on ionicity of the (110) surface relaxations of zinc‐blende semiconductors

It is argued that the surface relaxation angle ω of zinc‐blende (110) surfaces should depend on ionicity or on longitudinal effective charge Z approximately as ω=ω0−ω1 Z2e2/eaL, with ω1≂6°/eV.

Relaxation of the ZnTe and CuCl (110) surfaces

Ab initio molecular dynamics simulations of the ZnTe and CuCl (110) surfaces are employed to study surface atomic relaxation. We believe that these are the first such computations for heteropolar semiconductors, and for their surfaces in particular. The molecular dynamics follows the Sankey–Niklewski method, and electrostatic interactions are incorporated using Ewald’s scheme for Gaussian atomic charge distributions. Hence the electrons are treated in the local‐density approximation, forces are computed using the Hellmann–Feynman method, and atoms move to equilibrium according to Newton’s laws. Using ‘‘dynamical quenching,’’ we allow the ‘‘ideal’’ surfaces to relax according to these laws of physics and then address a controversy concerning whether Coulomb forces can play a significant role in determining the (110) zinc blende surface relaxation: Coulomb effects are not negligible for ZnTe (110) and are as dominant as covalent effects for CuCl (110). They reduce the (almost rigid) bond rotation angle ω1 d...

A straight domain boundary of single-atom width on a Si(111)-(7 × 7) surface

Abstract Using scanning tunneling microscopy (STM), we have observed an antiphase domain boundary of single-atom width on a Si(111)-(7 × 7) surface. The extra row of adatoms forming the boundary lies on the unfaulted half of the 7 × 7 unit cell, in agreement with total energy calculations using the first-principles self-consistent pseudofunction method. The filled-state STM image shows a missing interior adatom on the unfaulted half, in agreement with calculations of the partial density of states of an adatom surrounded by three rest-atoms.

Ethereal “interstitials” on the (110) surface of InSb

Abstract Scanning tunneling microscopy (STM) images of the (110) surface of InSb reveal apparent surface interstitial Sb atoms when the sample bias voltage is small and negative, corresponding to tunneling of electrons from the top of the valence band of InSb. These surface “interstitials” disappear at other bias voltages. The observations of these “ethereal interstitials” are explained in terms of below-surface, second-layer Sb atoms, that appear to lie above the surface due to the re-hybridized character of the surface Sb wavefunctions near the top of the valence band.

Oxidation of the GaAs(110) surface

Abstract The principal absorption site of oxygen on the GaAs(110) surface has been deduced from scanning tunneling microscopy (STM) measurements (for low oxygen coverages of the surface) to be in the interchain trough and above the surface. In contrast, photoemission measurements of Ga and As core-level shifts and of oxygen-derived energy levels led to the conclusion that absorption occurs on or below the surface (for high oxygen coverages), with Ga and As atoms about equally oxidised. To understand these observations, we have performed total-energy calculations using the first-principles pseudofunction method for a model with a periodic half-monolayer of oxygen, adsorbed to a few selected sites in the surface unit cell. Our results, although computed for models with half-monolayer coverages of oxygen, strongly suggest that the oxygen image observed in STM measurements indeed is due to an oxygen atom in the interchain trough, but with the oxygen atom located slightly below the surface and bonded to a Ga atom. For high coverages, our results imply in-surface-plane multi-coordinated adsorption of oxygen, which accounts for the observed oxygen-derived spectral features about 5 eV below the valence band maximum.

Dimerization on GaN(001) Surfaces

The GaN(OOl) surface is polar with either N- or Ga-termination. We predict that the Ga-terminated surface does not dimerize, but instead the surface Ga atoms relax into the vacuum by about ≈0.38 A. The N-terminated surface is predicted to form a c(2×4) structure with N 2 dimers in rows of length two. This is to be contrasted with the (2×1) Si(001) surface, on which the dimer rows are infinitely long, and with the (2×4) GaAs(OO1) surface, on which the rows are three dimers long.