Spatiotemporal characterization of water diffusion anomalies in saline solutions using machine learning force field.

Abstract

Understanding water behavior in salt solutions remains a notable challenge in computational chemistry. Conventional force fields have shown limitations in accurately representing water's properties across different salt types (chaotropes and kosmotropes) and concentrations, demonstrating the need for better methods. Machine learning force field applications in computational chemistry, especially through deep potential molecular dynamics (DPMD), offer a promising alternative that closely aligns with the accuracy of first-principles methods. Our research used DPMD to study how salts affect water by comparing its results with ab initio molecular dynamics, SPC/Fw, AMOEBA, and MB-Pol models. We studied water's behavior in salt solutions by examining its spatiotemporally correlated movement. Our findings showed that each model's accuracy in depicting water's behavior in salt solutions is strongly connected to spatiotemporal correlation. This study demonstrates both DPMD's advanced abilities in studying water-salt interactions and contributes to our understanding of the basic mechanisms that control these interactions.