About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Late News Poster Session
|
| Presentation Title |
H-37: PhaseForge: A Framework for Phase Diagram Predictions and Benchmarking Using Machine Learning Potentials |
| Author(s) |
Siya Zhu, Doğuhan Sarıtürk, Raymundo Arróyave |
| On-Site Speaker (Planned) |
Siya Zhu |
| Abstract Scope |
High-entropy alloys (HEAs) have attracted increasing attention due to their unique structural and functional properties. In studying HEAs, thermodynamic properties and phase stability play crucial roles, making phase diagram calculations particularly important. However, conventional CALPHAD assessments based on experimental or ab initio data are often computationally expensive. With the emergence of machine-learning interatomic potentials (MLIPs), we present PhaseForge, a program that integrates MLIPs into the Alloy Theoretic Automated Toolkit (ATAT) framework through our MLIP calculation library, MaterialsFramework, enabling efficient exploration of alloy phase diagrams. In addition, the workflow can also serve as a benchmarking platform for assessing the performance of different MLIPs from the perspective of phase-diagram prediction. We present several examples to illustrate the workflow and demonstrate the capability of PhaseForge. |
| Proceedings Inclusion? |
Undecided |
| Keywords |
Computational Materials Science & Engineering, High-Entropy Alloys, Machine Learning |