| Abstract Scope |
We present three sequential contributions. First, a pan-metallic parameterization of the Embedded Atom Model (EAM) designed to facilitate parametric molecular dynamics (MD) alloy sweeps of metallic glass (MG) and melts. Second, a dataset of quenched MD binary MGs based on the new pan-metallic EAM parameterization. This dataset spans quench rate, interaction chemistry, atomic sizes, and compositions, and is offered to the community to lower the activation barrier for future atomistic studies of MG. Third, an experiment using Smooth Overlap of Atomic Positions (SOAP) and Principal Component Analysis (PCA) to learn structural features from the quenched glass dataset. This experiment shows that while volume alone is a sufficient internal state variable to predict mechanical behavior for some MGs, for other MGs mechanical behavior predictions are improved by including learned structural features incorporating chemical and geometric information. SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525. |