About this Abstract |
| Meeting |
MS&T26: Materials Science & Technology
|
| Symposium
|
Progress in High Entropy Materials: Integrating Experiments, Computation, and Machine Learning
|
| Presentation Title |
Composition Design of Refractory High-Entropy Alloys with Machine Learning Models |
| Author(s) |
Haixuan Xu |
| On-Site Speaker (Planned) |
Haixuan Xu |
| Abstract Scope |
Refractory high-entropy alloys (RHEAs) are poised to transform high-temperature applications with their superior mechanical properties. Traditional exploration methods struggle with RHEAs' vast compositional space. This study employs the thermodynamic and first-principles methods for phase stability analysis, and machine learning (ML) for predicting mechanical properties at elevated temperatures, offering systematic rules for RHEA design. We assessed 466 multicomponent (ternary to novenary) systems and 43425 compositions with an incremental size of 10% in concentration. Additionally, ML models were trained using experimental datasets on temperature-dependent mechanical properties of BCC RHEAs. This approach predicts 7 new equiatomic alloys with high yield strengths at 1800 K. Furthermore, we identified 35 non-equiatomic systems surpassing 700 MPa in yield strength at elevated temperatures. |
