About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Phase Stability, Phase Transformations, and Reactive Phase Formation in Electronic Materials XXVI
|
| Presentation Title |
Enhanced Energy Storage Performance via Machine Learning Defined Phase Formation-Property Relationships in Multi-component Alloys |
| Author(s) |
Scott Broderick, Qinrui Liu, Md Tohidul Islam |
| On-Site Speaker (Planned) |
Scott Broderick |
| Abstract Scope |
In this presentation, statistical and machine learning tools are integrated with multi-scale property information for different classes of materials to enhance their capabilities for energy storage. The primary design consideration here is the modeling and prediction of phase formations and their evolution in multiple principal element alloys and ceramics. These materials have significant applications in fuel cells and catalysis, with the properties largely impacted by the number of phases present and their distribution. This work integrates multiple machine learning approaches, including manifold learning, regressions and evolutionary algorithms, into a novel design framework. In this analysis, the process controls and the overall material chemistry are simultaneously considered, and design guidelines for enhanced performance are defined. Through the careful consideration of the input data, not only is the design computationally accelerated, but the understanding of the underlying physics is enhanced. |
| Proceedings Inclusion? |
Planned: |