Neural Network Potential Molecular Dynamics Simulations of (La,Ce,Pr,Nd)_0.95(Mg,Zn,Pb,Cd,Ca,Sr,Ba)_0.05F_2.95

Tysonite structure fluorides doped with divalent cations, represented by Ce_0.95Ca_0.05F_2.95, are a class of good F^- ion conductors together with fluorite-structured compounds. Computational understanding of the F^- conduction process is difficult because of the complicated interactions between three symmetrically distinct F sites and the experimentally observed change in the F diffusion mechanism slightly above room temperature, effectively making first-principles molecular dynamics (FP-MD) simulations, which are often conducted well above the transition temperature, useless when analyzing behavior below the transition point. Neural network potential (NNP) MD simulations showed that the F diffusion coefficient is higher when the divalent dopant cation size is similar to the trivalent cation size. The diffusion behavior of F in different sites changes at roughly 500 K in Ce_0.95Ca_0.05F_2.95 because only the F1 site sublattice contributes to F diffusion below this temperature, but the remaining F2 and F3 sublattices become gradually active above this temperature. The paradox of higher diffusion coefficients in CeF₃-based compounds than similar LaF₃-based compounds even though the lattice parameters are larger in the latter may be caused by a shallower potential of Ce and F in CeF₃ compared to the LaF₃ counterparts.