Water at Interfaces Investigated by Neural Network-Based Molecular Dynamics Simulation

Interfacial water plays a crucial role in various biological, chemical, physical, and material processes. Sum-frequency vibrational spectroscopy (SFVS) is a powerful method for studying molecular orientation, structure, and dynamics at interfaces. By employing molecular dynamics simulations based on artificial neural network potentials, we explore water at interfaces, taking into account non-Condon effects from the electric fields of surrounding water molecules and neglecting a rigorous treatment of quantum nuclear motion. Our simulations are not able to replicate the shoulder of the free OH peak. The influence of the electric dipole dominates SFVS, while that of the electric quadrupole is relatively small but still important and nonignorable. This influence contributes to the absence of the low-frequency positive peak of OH stretch vibrations. This research represents a significant advancement in the understanding of the properties of water at interfaces and the mechanics of SFVS.