Kernel-Based Modeling of Electron-Density Polarization at Metal-Liquid Interfaces

Accurate modeling of metal polarization is crucial for understanding molecular interactions at metal-liquid interfaces. In this paper, we present a novel computational method for incorporating the polarization of metallic electrons into classical molecular dynamics simulations. Our approach employs a kernel-based polarization model to describe the real-time polarization of the metal electron density on a three-dimensional grid, with parameters fitted to quantum mechanical calculations. We applied this model to investigate the water-Au(111) interface, analyzing the effects of varying levels of metal polarization: (1) no polarization, (2) full polarization, and (3) time-averaged polarization. The results showed that metal electron polarization enhanced the orientational fluctuations of water molecules, stabilized the O-down configuration near the metal surface, and increased the population of nondonor hydrogen-bond configurations. The time-averaged approximation reproduces some trends observed with full polarization but introduces a bias toward lay-down configurations, leading to an overestimation of double-donor configurations. Our grid-based polarization method offers a computational approach for simulating metal polarization effects, providing new methods to investigate the electrostatics and dynamics of metal-liquid interfaces.