About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Materials and Chemistry for Molten Salt Systems
|
| Presentation Title |
Predicting Thermophysical and Structural Properties of NaCl–UCl3–PuCl3 Molten Salts with Machine-Learning Interatomic Potentials (MLIPs) |
| Author(s) |
Sudipta Paul, Siamak Attarian, David Andersson, Dane Morgan, Izabela Szlufarska |
| On-Site Speaker (Planned) |
Sudipta Paul |
| Abstract Scope |
Chloride salts containing uranium and plutonium are promising liquid fuels for fast-spectrum molten salt reactors (MSRs), but reliable thermophysical-property data remain limited. We developed a Machine-Learning Interatomic Potential (MLIP) based on an atomic cluster expansion (ACE) for the complete NaCl–UCl3–PuCl3 compositional space, including all unary end members, binary mixtures, and ternary compositions. We used molecular dynamics with the MLIP to predict density, heat capacity, viscosity, thermal conductivity, and mixing enthalpy over reactor-relevant temperatures with very good accuracy at substantially lower computational cost than full ab initio methods. Structural analyses revealed how actinide–chloride coordination and network formation evolve with composition and influence transport behavior. By providing a consistent description of structure–property relationships across the full composition triangle, this work generates broadly applicable data for chloride-fueled reactor systems. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Nuclear Materials, Modeling and Simulation, Computational Materials Science & Engineering |