An approach based on random sampling and density functional theory to identify highly stable structures of ABX3 compounds

[Display omitted] First-principle screening of the vast material space is one of the recently growing approaches to discover novel materials. However, without some sort of prior knowledge of materials’ structures, this can be computationally very expensive, complex, and inconclusive. Herein, we present an approach to predict the highly stable structures of ABX3 compounds using density functional theory (DFT) combined with proper random sampling and utilizing the precision library of Standard Solid-State Pseudopotentials (SSSP). The considered reduced material space is for 18 halide compounds where A = Cs+, K+, and Rb+; B = Pb2+ and Sn2+; and halogen X = Cl−, I−, and Br−. Initially, 40-atom supercells are assembled in the high symmetrical phase where the lattice parameters are determined from their ionic radii. Then, 10 random samples of each compound are constructed where atomic positions are randomly displaced from the high symmetrical phase. All these samples are then used as initial guesses to find stable phases. The phase with the lowest energy is considered the highly stable one. By comparing the results with some recent vigorous computational and experimental reports, the presented approach demonstrates good agreement. It illustrates adequate level of accuracy and computational efficiency to be considered for high-throughput calculations.