Abstract

A comprehensive survey of the total valence-band x-ray-photoemission spectra of 14 semiconductors is reported. The x-ray photoelectron spectra of cubic GaP, GaAs, GaSb, InP, InAs, InSb, ZnS, ZnSe, ZnTe, CdTe, and HgTe, and of hexagonal ZnO, CdS, and CdSe were obtained from freshly cleaved single crystals, in the 0-50-eV binding-energy range, using monochromatized $\mathrm{Al} K\ensuremath{\alpha}$ (1486.6 eV) radiation. The binding energies of the outermost $d$ shells are reported. They were determined relative both to the top of the valence bands (${E}_{B}^{V}$) and to the Fermi level of a thin layer of gold that was vapor deposited after each run (${E}_{B}^{F}$). These data also yielded accurate measures of sample charging. The Fermi level fell near the center of the gap for six samples, near the top for two, and near the bottom for three. Evidence for an apparent increase in core $d$-level spin-orbit splitting over free-atom values was interpreted as a possible spreading of a ${\ensuremath{\Gamma}}_{7}$ and a ${\ensuremath{\Gamma}}_{8}$ level from the upper $({d}_{\frac{3}{2}}) {\ensuremath{\Gamma}}_{8}$ level by a tetrahedral crystal field. The $s$, $p$ valence-band spectra showed three main peaks, with considerable structure on the "least-bound" peak. A discussion is given of the validity of comparing the valence-band (VB) spectrum ${I}^{\ensuremath{'}}(E)$ with the VB density of states, including cross-section modulation, final-state modulation, and relaxation effects. Characteristic binding energies of spectral features in ${I}^{\ensuremath{'}}(E)$ are tabulated. In addition, the energies of the characteristic symmetry points ${L}_{3}$, ${X}_{5}$, ${W}_{2}$, ${\ensuremath{\Sigma}}_{1}^{min}$, ${W}_{1}$, ${X}_{3}({L}_{1})$, ${X}_{1}$, ${L}_{1}$, and ${\ensuremath{\Gamma}}_{1}$ are given for the 11 cubic compounds. These are compared with UPS results where available and with theoretical band-structure results where available. The energies calculated using the relativistic-orthogonalized-plane-wave approach with ${X}_{\ensuremath{\alpha}\ensuremath{\beta}}$ exchange agree very well with experiment, on the whole. In particular, they predict the important "ionicity gap" ${X}_{3}\ensuremath{-}{X}_{1}$ quite accurately. The densities of states calculated using the empirical-psuedopotential method provided a useful basis for relating features in ${I}^{\ensuremath{'}}(E)$ to energies of the characteristic symmetry points. Band-structure calculations in combination with x-ray-photoemission spectra appear to provide a very powerful approach to establishing the total valence-band structure of semiconductors.