Molecular dynamics simulations of aqueous electrolytes generally rely on empirical force fields, combining dispersion interactions - described by a truncated Lennard-Jones (LJ) potential - and electrostatic interactions - described by a Coulomb potential computed with a long-range solver. Recently, force fields using rescaled ionic charges (electronic continuum correction, ECC), possibly complemented with rescaling of LJ parameters (electronic continuum correction rescaled, ECCR), have shown promising results in bulk, but their performance at interfaces has been less explored. Here we started by exploring the impact of the LJ potential truncation on the surface tension of a sodium chloride aqueous solution. We show a discrepancy between the numerical predictions for truncated LJ interactions with a large cutoff and for untruncated LJ interactions computed with a long-range solver, which can bias comparison of force field predictions with experiments. Using a long-range solver for LJ interactions, we then show that an ionic charge rescaling factor chosen to correct long-range electrostatic interactions in bulk also describes accurately image charge repulsion at the liquid-vapor interface, and that the rescaling of LJ parameters in ECCR models - aimed at capturing local ion-ion and ion-water interactions in bulk - also describes well the formation of an ionic double layer at the liquid-vapor interface. Overall, these results suggest that the molecular modeling of aqueous electrolytes at interfaces would benefit from using long-range solvers for dispersion forces, and from using ECCR models, where the charge rescaling factor should be chosen to correct long-range electrostatic interactions.
Download