The fractal correlation between relaxation dynamics and atomic-level structures observed in metallic glasses by computer simulation

Abstract

In the present study, atomistic simulation reveals that the microscopic mechanism of the relaxation dynamics in amorphous materials is governed by the activating atoms that jump more than half of the average nearest neighbor distance within a given time. Based on the unsupervised machine-learning algorithm, hierarchical clustering analysis shows that the activating atoms are excited in a cooperative and avalanche-like model to form activating units. Correlation analysis suggests that large free volumes facilitate the formation of activating atoms. Interestingly, a fractal correlation is found between the number and size of the activating units, indicating that when the maturation temperature, i.e. percolation threshold, is reached the activating units form a complicated and connected body in space.