About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Modeling, AI Applications, and Method Development in Nuclear Materials
|
| Presentation Title |
Machine learning application in the design of irradiation-resistant nuclear materials |
| Author(s) |
Shijun Zhao |
| On-Site Speaker (Planned) |
Shijun Zhao |
| Abstract Scope |
Irradiation-resistant nuclear materials are critical to the advancement of next-generation fission and fusion reactors. However, their design is severely constrained by the vast compositional space and the intricate defect dynamics that span multiple time and length scales. Machine learning (ML), with its ability to capture nonlinear relationships among complex variables, is uniquely suited to uncovering the hidden correlations and mechanisms governing these materials. In this work, we propose three key ML-driven strategies for the study of irradiation-resistant nuclear materials: (1) direct prediction of irradiation performance using ML models; (2) multiscale bridging via ML to connect atomic-scale defect evolution with experimentally relevant scales; and (3) development of interatomic potentials with first-principles accuracy to enable precise atomistic simulations of irradiation damage. Using high-entropy alloys and high-entropy ceramics as representative case studies, we present our recent research progress and key findings across these three directions. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Nuclear Materials, ICME, Computational Materials Science & Engineering |