Advances in Molecular Dynamics Studies of Molten Salts Based on Machine Learning

Molten salt is a kind of molten material with important application value. However, the relationship between microstructure and macroscopic properties of molten salt has not been fully investigated, so it is of great significance to carry out extensive molecular dynamics studies. For high-temperature molten salts, molecular dynamics studies mainly relied on the development of force fields in classical molecular dynamics and first principles molecular dynamics previously. Due to the accelerated development of machine learning and neural networks, the potential for molten salts based on machine learning has been developed recently, and great progress has been made in exploring the coordination chemistry and accurately predicting physical properties. Herein, the latest research progress of molecular dynamics related to molten salt was firstly reviewed, especially the development status of machine learning potential. Secondly, the application progress of machine learning potentials in the research of molten salt was summarized. Finally, the research prospect of machine learning potentials for molten salt was discussed, and some suggestions were given.