Molecular simulations of alkali metal halide hydrates

It has been known that classical molecular simulations of concentrated aqueous electrolyte solutions are limited by the ability of microscopic models (force fields) to predict the correct values of salt solubility, which also requires their ability to faithfully model crystalline salts. Previous simulation studies have often focused on the solubility of anhydrous crystalline salts, but virtually never on crystalline hydrates, except for hydrohalite, NaCl & BULL;2H2O, despite there are at least 23 experimentally known different hydrates that can precipitate from alkali-halide solutions. This work attempts to fill this gap in hydrate simulation studies by systematically investigating the ability of the best force fields selected to qualitatively capture the stability of the individual phases of various alkali-halide hydrates and to quantitatively predict their lattice parameters. First, we show that the nonpolarizable force fields studied often fail to model hydrates containing the Li+ cations, whereas the polarizable force fields recently refined by us are able to model all the hydrates except for LiCl & BULL;H2O. Second, we further refine our FFs for Li+ to yield stable LiCl & BULL;H2O. Third, our simulations clarify the positions of the Li+ cations in the ?????? phases of LiBr & BULL;H2O and LiI & BULL;H2O, whose distributions were previously described only as stochastic. As a byproduct, a simple and reliable simulation methodology suitable also for complex polarizable models and nonorthorhombic crystal lattices is proposed and tested, based on simulations of finite crystals floating in vacuum.