Empirical Tight-Binding Parameterizations for Accurate Heterostructure and Alloy Calculations

Empirical tight-binding is a valuable and reliable tool for the calculation of electronic and optical properties in semiconductor heterostructures and alloys. It has been applied in many contexts, ranging from inorganic semiconductors like arsenides, antimonides, nitrides and their alloys to 2D materials and hybrid perovskites. Here we present recent activities in development and application of empirical tight-binding, in particular regarding disordered alloys. We show some limitations of the most used schemes, and how they can be overcome by more recent parameterizations, and we introduce an alternative machine-learning based parameterization scheme.

I. Introduction

Empirical tight-binding (ETB) has proven to be extremely useful for the simulation of optoelectronic devices, when details of the atomistic structure matter, but ab-initio approaches like DFT are not feasible. ETB has been used in the last decades in particular for studying random alloy and disorder effects in III-nitrides, like InGaN quantum-well structures, and for III-V based alloys [1], [2].