About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Modeling, AI Applications, and Method Development in Nuclear Materials
|
| Presentation Title |
A Machine-Learning Interatomic Potential for Defect Migration and Cluster Evolution in Silicon Carbide |
| Author(s) |
Lihao Shi, Fei Gao |
| On-Site Speaker (Planned) |
Lihao Shi |
| Abstract Scope |
Silicon carbide is a leading structural material for advanced fission and fusion reactors due to its excellent thermal stability and radiation tolerance. Long-term dynamics simulations of radiation damage require interatomic potentials that accurately describe defect energetics, migration, and defect-cluster evolution. In this work, we develop a machine-learning interatomic potential for SiC using the Graphics Processing Units Molecular Dynamics (GPUMD) framework based on comprehensive density functional theory training database. The database includes bulk structures, elastic deformations, point defects, migration and rotation transition states, and small interstitial clusters. The resulting potential accurately reproduces defect formation energies, migration barriers, and transition-state energetics while retaining the computational efficiency of GPUMD for large-scale molecular dynamics simulations. By explicitly incorporating defect clusters into the training process, the potential significantly improves the description of cluster stability and migration, providing a reliable platform for investigating radiation-induced defect evolution and future long-term dynamics changes in SiC under irradiation conditions. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Machine Learning, Nuclear Materials |