About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Advances in Magnetism and Magnetic Materials
|
Presentation Title |
Physics-informed Machine Learning for Designing Novel Functional Magnetic Materials |
Author(s) |
Prashant Singh, Tyler Del Rose, Andriy Palasyuk, Yaroslav Mudryk |
On-Site Speaker (Planned) |
Prashant Singh |
Abstract Scope |
High performance permanent magnets with improved phase stability and higher Curie temperatures (T<sub>C</sub>) that contain less critical elements are integral to zero-carbon energy solutions. Machine-learning (ML) models, built over DFT calculated phase stability and experimentally measured T<sub>C</sub>, were developed for data-driven predictive approaches to designing novel functional magnetic materials. We chose two compositions from the pseudo-binary Zr-Ce-Fe system to experimentally validate the ML predicted phase stability and T<sub>C</sub>. We also discuss the utility of de Gennes scaling in T<sub>C</sub> prediction of rare earth compounds and show the breakdown of this rule in 4f-3d magnetic compounds when the 4f content is below a critical limit. Finally, our research reveals that physics-informed ML can be used for accurate design of new high-performance magnets with improved properties for environmentally sustainable applications. |
Proceedings Inclusion? |
Planned: |
Keywords |
Magnetic Materials, Machine Learning, Other |