A comparative study on the structure and properties in the Cu-Zr system based on the combination of machine learning force field and molecular dynamics

This paper systematically studies the structure and related properties of the Cu-Zr metallic glass system at the atomic scale using a combination of machine learning and molecular dynamics simulations, revealing the structural origins of the excellent glass-forming ability, mechanical properties, and thermal stability of Cu₅₀Zr₅₀ metallic glass. Initially, the most stable Cu_xZr_100-x (x = 20, 50, 65, 80) system is obtained by compositional screening and optimization. Subsequently, the underlying mechanism by which Cu concentration regulates the related properties was revealed by analyzing the variations in the radial distribution function g(r), the distribution of Voronoi polyhedral content, the distribution characteristics of interatomic bond angles, and the intrinsic correlation between electronic structure and mean squared displacement. To conclude, Cu₅₀Zr₅₀ has the structure with the optimal glass-forming ability in the binary Cu-Zr metallic glass system, and the paper provides a reference for the design and optimization of other metallic glasses.