Chemical trends for native defects in III-V–compound semiconductors

We present energy levels calculated for both vacancies and antisite defects in nine different III-V compounds. It is shown that the chemical trends for the neutral, unrelaxed vacancy levels as obtained with more sophisticated linear combination of atomic orbital models can be reproduced and extrapolated by use of a simple, rescaled defect-molecule model. The only input parameters needed are the hybrid orbital energies of the nearest-neighbor atoms and the photothreshold energies of the host materials. We also use this model to calculate trends for the antisite-induced energy levels which are consistent with experiments on ${\mathrm{P}}_{\mathrm{Ga}}$ and ${\mathrm{As}}_{\mathrm{Ga}}$. Cation antisite defects are predicted to produce deep levels within the main energy gap for most of the compounds investigated. Furthermore, the trends for cation antisite defect levels are in striking quantitative agreement with the observed, characteristic Fermi-level pinning energies at Schottky barriers on both n- and p-type materials. Therefore, our calculations support the native-defect model proposed by Spicer et al., but with a single defect in different charge states mainly responsible for the observed pinning energies.