GPU-Enhanced Density Functional Tight Binding for Probing Multi-Component Alloys

Abstract

DFT has been extensively employed in the search for stable thermodynamic compounds which is crucial for the design and synthesis of novel functional materials and the understanding of the behavior of materials under different conditions. In first-principles calculations, thermodynamically stable phases of materials are predicted by constructing a convex hull with respect to formation energy of the compounds. This work emphasizes an alternative approach to DFT wherein the computational efficiency of DFTB is harnessed to predict the convex hull of binary compounds.