Modeling Next Generation Sensor Chips: Towards Predictive Band Structure Models for Quarternary III-V Semiconductor Alloys

Abstract

In this work, an ab initio scheme based on the DFT+U method is presented to model the electronic structure of ternary/quaternary semiconductor alloys. Within this method, optimal sets of Hubbard-U parameters are found for each binary compound of the alloy using a Bayesian optimization scheme. We introduce two interpolation schemes for the U parameters to describe the electronic structure in alloy supercells. By applying them on the InGaAsSb semiconductor alloy, we show that the predicted bandgap bowing is much closer to experimental data compared to results based on the virtual crystal approximation. The two schemes are shown to predict almost identical electronic bandgaps and provide a computationally efficient alternative to other approaches like the use of hybrid functionals.