| About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
Microstructural, Mechanical, and Chemical Behavior of Solid Nuclear Fuel and Fuel-Cladding Interface II
|
| Presentation Title |
An Irreducible Derivatives Approach to Improve Empirical Interatomic Potentials for Thermophysical Property Predictions: the Case of ThO2 and UO2 |
| Author(s) |
Shuxiang Zhou, Chao Jiang, Enda Xiao, Sasaank Bandi, Michael Cooper, Miaomiao Jin, David Hurley, Marat Khafizov, Chris Marianetti |
| On-Site Speaker (Planned) |
Shuxiang Zhou |
| Abstract Scope |
The prediction accuracy using classical molecular dynamics is determined by the empirical interatomic potentials (EIPs). Here we introduce a training approach based on the second- and third-order interatomic force constants (IFCs) computed by density functional theory. This work’s unique aspect is the utilization of irreducible derivatives of the total energy, which leverage on the symmetry of the crystal and provide a minimal representation of the IFCs. We demonstrate this approach as an efficient and robust manner to train EIPs for predicting phonon-related properties, by optimizing parameters of an embedded-atom method potential with core-shell interactions for both ThO2 and UO2. Our newly trained EIPs provide thermophysical properties in good agreement with DFT and outperforms widely utilized EIPs in phonon dispersion and thermal conductivity predictions. The usage of irreducible derivatives in training potentials and improving phonon predictions is not limited to EIPs; it can also be applied in advanced machine-learning models. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
High-Entropy Alloys, Nuclear Materials, |