Characterization of Si-doped GaAs cross-sectional surfaces via ab initio simulations

Si-doped (110) GaAs cross-sectional surfaces are investigated using first principles calculations with the aim of identifying simple configurations compatible with experimentally detected scanning tunneling microscopy (STM) images characterized by bright signals at negative bias and strongly attenuated when the bias is reversed. Since Si-donor is the most common defect in Si-doped GaAs, we study mainly several Si-donor based configurations, from the isolated impurity to extended defect configurations, up to the limit of an entire embedded donor interlayer. We consider also donor-acceptor configurations, stimulated by the experimental evidence of self-compensation effects in particular in highly doped samples. The systematic study of simulated cross-sectional STM images for all these and other configurations allows not only to identify the self-compensated donor-acceptor configurations as the simplest ones which are compatible with those experimental observations, but also to exclude most of the other candidates.