| Abstract Scope |
A significant advancement of the past 20 years in computational materials science has been the ability to rapidly and accurately predict thermodynamic and kinetic parameters for arbitrary crystalline structures, which has enabled design of bespoke microstructures in metal alloys, functional ceramics, and many other material types. However, a fundamental difficulty arises when attempting similar design exercises for amorphous systems such as bulk metallic and oxide glasses, where atomistic simulations of 0K thermodynamics from DFT or phase diagrams are limited in applicability and fail to capture the inherently metastable and disordered nature of glassy structures. In this talk, the challenge of developing predictive models for glass formability based on machine learning, experimental databases, and complementary atomistic simulations will be discussed, with an eye towards their use in glass design. |