Determining ground states of alloy by a symmetry-based classification

Abstract

Reducing the number of candidate structures is crucial to improve the efficiency of global optimization. Herein, we demonstrate that the generalized Hamiltonian can be described by the atom classification model (ACM) based on symmetry, generating competent candidates for the first-principles calculations to determine ground states of alloy directly. The candidates can be obtained in advance through solving the convex hull step by step, because the correlation functions of ACM can be divided into various subspace according to the defined index $l$. As an important inference, this index can be converted to the number of Wyckoff positions, revealing the dominant effect of geometry symmetry on structural stability. Taking Ni-Pt, Ag-Pd, Os-Ru, Ir-Ru and Mo-Ru as examples, we not only identify the stable structures in previous theoretical and experimental results, but also predict a dozen of configurations with lower formation energies, such as Ag$_{0.5}$Pd$_{0.5}$ ($Fd$-$3m$), Os$_{0.5}$Ru$_{0.5}$ ($Pnma$), Ir$_{1/3}$Ru$_{2/3}$ ($P6_{3}/mmc$), and Mo$_{0.25}$Ru$_{0.75}$ ($Cmcm$).