About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
Bulk Metallic Glasses XX
|
Presentation Title |
J-85: Using Machine Learning to Find Correlations of Structure Motifs with Metallic Glass States and Mechanical Properties |
Author(s) |
Suyue Yuan, Paulo Branicio |
On-Site Speaker (Planned) |
Suyue Yuan |
Abstract Scope |
Vacancies are ubiquitous defects in crystalline materials that are related to their electronic, diffusion, and mechanical properties. We use machine learning to study the local atomic configuration in metallic glasses (MGs) and discover two unique topological footprints, T5 and Q7, which resemble vacancies in crystals and contribute significantly to the short-range structural disorder in MGs. The T5 and Q7 refer to atomic Voronoi polyhedra with five triangular faces and seven quadrangular faces respectively. Their concentrations in MGs follow an Arrhenius relationship with temperature before melting, as accurate indicators of the glass transition. They also show strong correlations with the yield and failure of the MG during deformation. Furthermore, atoms centered in T5/Q7 polyhedra display larger local entropy, atomic volume, and smaller activation energy. The finding of T5/Q7 motifs provides missing insights to understand the local disorder and their intrinsic relationships with thermal as well as mechanical behavior of MGs. |
Proceedings Inclusion? |
Planned: |
Keywords |
Computational Materials Science & Engineering, Mechanical Properties, Machine Learning |