About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Theory and Design of Metallic Glasses
|
| Presentation Title |
Uncovering the Medium-Range Orders in Metallic Glass via Topology-Assisted Machine Learning |
| Author(s) |
Muchen Wang, Yue Fan |
| On-Site Speaker (Planned) |
Muchen Wang |
| Abstract Scope |
Understanding medium-range order (MRO) in metallic glasses is essential for unraveling their structure-property relationships, yet its structural identity and characteristic length scale remain poorly defined. In this work, we present a topology-assisted machine learning framework that integrates persistent homology with deep learning to uncover MRO signatures in ZrCu metallic glass. Our analysis reveals a consistent model focus on long-persistence topological features in thermodynamically stable samples, suggesting the emergence of skeleton-like structures that contribute to macroscopic stability. Among several structurally relevant length scales uncovered, the largest-approximately 5 Ĺ in radius-is attributed to MRO. To further probe their diversity, we implement an unsupervised clustering approach that classifies these motifs into distinct structural groups, ranging from compact single-cluster units to partitioned multi-cluster domains. This study offers direct, quantitative insight into the nature of MRO in metallic glasses and establishes a generalizable framework for dncoding hidden structural order in disordered materials. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Computational Materials Science & Engineering, |