Efficient Predictions of Formation Energies and Convex Hulls from Density Functional Tight Binding Calculations

Abstract

Defects in materials significantly alter their electronic and structural properties, which affect the performance of electronic devices, structural alloys, and functional materials. However, calculating all the possible defects in complex materials with conventional Density Functional Theory (DFT) can be computationally prohibitive. To enhance the efficiency of these calculations, we interfaced Density Functional Tight Binding (DFTB) with the Clusters Approach to Statistical Mechanics (CASM) software package for the first time. Using SiC and ZnO as representative examples, we show that DFTB gives accurate results and can be used as an efficient computational approach for calculating and pre-screening formation energies/convex hulls. Our DFTB+CASM implementation allows for an efficient exploration (up to an order of magnitude faster than DFT) of formation energies and convex hulls, which researchers can use to probe other complex systems.

Anshuman Kumar
Anshuman Kumar
Postdoctoral Scholar, SME

My research interests include machine learning, data science, computer vision, and computational sciences.