Molecular dynamics simulations of CaCl2–NaCl molten salt based on the machine learning potentials

CaCl2–NaCl molten salt is a promising material for energy transport and storage. The microstructure evolution and properties of CaCl2–NaCl molten salt are significantly important for improving energy conversion efficiency. In this study, machine learning potentials were applied to explore the effect of composition and temperature on the structure and properties of CaCl2–NaCl molten salt. It showed great merits, such as considerable economic cost, high efficiency and accuracy, and a safe working environment. The structural information of CaCl2–NaCl molten salt was analyzed systematically, including partial radial distribution function, coordination number distribution, and angular distribution function. The distorted octahedral structure of Na–Cl and Ca–Cl ion pairs was observed in CaCl2–NaCl molten salt and the degree of distortion varied with the components. Besides, the influence of composition and temperature on properties were investigated, including density, ion self-diffusion coefficients, shear viscosity, electrical conductivity, thermal expansion coefficient, and specific heat capacity. In summary, the machine learning potential molecular dynamics simulations have a blazing application prospect in theoretical research of molten salt systems and can provide great fundamentals for practical application.