Dynamical analysis of low-energy electron-diffraction intensities from InAs(110)

Abstract

The measured intensities of 14 diffracted beams of low-energy ($30\ensuremath{\le}E\ensuremath{\le}240$ eV) electrons normally incident on InAs(110) are reported. The temperature of the InAs surface during the measurements was $T=110$ K. The surfaces were prepared by ion-bombardment and annealing cycles followed by a verification of surface stochiometry (to within 10%) via Auger electron spectroscopy. Two separate crystals were examined in two separate vacuum instruments in order to verify the reproducibility of the intensity data. The data were analyzed using a relativistic, Hara-exchange electron---ion-core potential and an x-ray $R$-factor structure-analysis methodology. This analysis leads to the best-fit structure of InAs(110) being a rotation of the uppermost layer with the As rotated outward and the In inward. The second layer also is reconstructed with the In being displaced upward by 0.07\ifmmode\pm\else\textpm\fi{}0.1 \AA{} relative to its position in the bulk and the As being displaced downward by an equal amount. The x-ray $R$ factor for the best-fit structure is ${R}_{\mathrm{x}}=0.23$. This structure gives a satisfactory visual description of the measured intensities. For bond-length-conserving top-layer rotations the angle between the plane of the surface In-As chains and the plane of truncated bulk surface is ${\ensuremath{\omega}}_{1}=31\ifmmode^\circ\else\textdegree\fi{}\ifmmode\pm\else\textpm\fi{}3\ifmmode^\circ\else\textdegree\fi{}$, in accordance with expectations based on correlations of covalent radii with prior zinc-blende-structure compound-semiconductor surfaces. The structure of InAs(110) provides the first test of these correlations. The best-fit $R$-factor structure corresponds to reduced relaxation parallel to the surface of the top-layer As relative to that for the bond-length-conserving structure. This best-fit structure is, however, equivalent to its bond-length-conserving counterpart (${\ensuremath{\omega}}_{1}=31\ifmmode^\circ\else\textdegree\fi{}$) to within the accuracy of the $R$-factor methodology.