Shear deformation in CuZr metallic glass: A statistical and complex network analysis

Abstract

We have implemented a complex network description for metallic glasses, able to predict the elasto-plastic regime, the location of shear bands and the statistics that control the plastic events that originate in the material due to a deformation process. By means of molecular dynamics simulations, we perform a shear deformation test, obtaining the stress–strain curve for CuZr metallic glass samples. The atomic configurations of the metallic glass is mapped to a graph, where a node represents an atom whose stress/strain is above a certain threshold, and edges are connections between existing nodes at consecutive timesteps in the simulation. We made a statistical study of some physical descriptors such as shear stress, shear strain, volumetric strain and non-affine displacement to use them as construction tools for complex networks. We have calculated their probability density functions, skewness, kurtosis and Gini coefficient to analyze the inequality of the distributions. We study the evolution of the resulting complex network, by computing topological metrics such as degree, clustering coefficient, betweenness and closeness centrality as a function of the strain. We have obtained correlations between the physical phenomena produced by the deformation with the data recorded by these metrics. By means the visual representation of the networks, we have also found direct correlations between metrics and the local atomic shear strain, so that they are able to predict the location of shear bands, as well as the formation of highly connected and interacting communities, which we interpret as shear transformation zones. Our results suggest that the complex network approach has interesting capabilities for the description of mechanical properties of metallic glasses.