Concentration−Transferable Deep Potential Molecular Dynamics: Unraveling Component−Structure−Transport in Molten LiF−BeF2−EuF2

Abstract

Understanding how the soluble fission product EuF2 influences the structure and properties of the molten LiF−BeF2 (FLiBe) carrier salt is crucial for the safe operation and fuel cycle management of molten salt reactors. Here, a deep potential molecular dynamics (DPMD) approach that integrates first-principles calculations, machine learning, and molecular dynamics simulations is employed to systematically investigate the effects of EuF2 content on the local structure and transport properties of molten FLiBe at 823 K. The trained deep potential model not only reproduces ionic pair structures and densities within the training concentration range but also demonstrates robust predictive capability across a broader concentration window (0.50 to 6.25 mol%). As the EuF2 concentration increases, the shear viscosity rises nonlinearly, signaling a marked deterioration in collective transport performance. Concurrently, all ionic self-diffusion coefficients decline, with Be2+ and Eu2+ exhibiting the most pronounced and synchronous reductions (up to 70%). Structural analyses uncover a strong Eu−F−Be polarization interaction, with the proportion of the [BexEuyFz]2x+2y-z configurations escalating from 3% to 21% as the EuF2 content increases. Moreover, EuF2 accumulation severely disrupts the chain-like connectivity of Be−F tetrahedral units, driving a transformation toward to complex network structures. Overall, these results provide both quantitative and qualitative structure−property relationships for EuF2 in molten FLiBe and establish a transferable computational framework for evaluating fission product behavior in fluoride-based molten salt reactors.