CX-020780: Learning-based Computational Study of the Thermodynamic, Structural, and Dynamic Properties of Molten Salts at the Atomic and Electronic Scale and Experimental Validations – University of Illinois at Urbana-Champaign and Argonne National Lab

The University of Illinois at Urbana-Champaign (UIUC), in collaboration with Argonne National Laboratory (ANL), proposes to obtain the thermophysical, thermochemical, and transport properties, construct the phase diagrams, and build empirical physical models of molten salts using simulations driven by machine-learned high-dimensional statistical learning modules combined with experimental validations. The tasks associated with this project are (1) Use machine learning (ML) to generate neural network potentials (NNPs) for molten salts from first-principles calculations; (2) Perform large-scale and longtime Molecular Dynamics (MD) simulations with the machine-learned NNPs, compute thermodynamic, structural, and dynamic properties and construct phase diagrams and empirical models of the thermophysical and thermochemical properties of the molten salts; and (3) Perform neutron/X-ray scattering and thermodynamic experiments to validate the empirical models. Existing laboratory facilities will be used.