Identifying Packing Features of Atoms with Distinct Dynamic Behaviors in Metallic Glass by Machine-Learning Method

Characterizing and predicting atoms prone to rearrangements directly from atomic structure are longstanding challenges in disordered systems. Here we report a successful identification of liquid-like atoms susceptible to rearranging in a model Cu50Zr50 metallic glassy system by machine-learning method. Moreover, we observe that there exists a characteristic length rc within the first neighbor shell, the liquid-like atoms behave more tightly packed inside a spherical region of the radius r更多