Quantum mechanical development of potential energy parameters for molecular dynamics simulations of the clay/solution interface

Since geochemical processes ultimately occur at the molecular level, the overall goal of this research is to help create a new research tool that will allow environmental chemists to routinely simulate the basic molecular interactions at the clay mineral/aqueous solution interface. Both thermodynamic and spectroscopic studies of adsorption would benefit from such a tool, but existent molecular dynamics programs lack the potential energy parameters necessary to model clays. To begin compiling detailed information about the potential energy surfaces for clay minerals, we examined the effects of basis set size and electron correlation corrections on the properties of three increasingly complex molecules often used as models for silicates, disiloxane ({dollar}rm Hsb3SiOSiHsb3),{dollar} orthosilicic acid (Si(OH){dollar}sb4),{dollar} and disilicic acid ({dollar}rm Hsb6Sisb2Osb7).{dollar} The extra effort needed to include electron correlation effects, which we calculated at the MP2 level, would seem to be worthwhile in light of the importance of accurate local electronic structure and geometry. It is argued on the basis of electrostatic potential maps, as well as ab initio partial charges, that {dollar}rm Hsb3SiOSiHsb3{dollar} and its analogues provide poor quantitative models for silicate functional groups. However, structural features and partial charges on Si(OH){dollar}sb4{dollar} and {dollar}rm Hsb6Sisb2Osb7{dollar} proved to be very similar at a given level of theory. Thus, effective ab initio parameter development for silicates should focus on molecules comprised of silicon that is fully coordinated by oxygen.; The quantum mechanical results, along with available experimental data, were used to derive a self-consistent set of potential energy terms (the "force field") for simulating purely siliceous minerals. The force field reproduced the structures and vibrational spectra of the silicate molecules as well as those of {dollar}alpha{dollar}-quartz. While not thoroughly tested, the force field developed in this manner seems to possess several inherent advantages. The atomic partial charges are based on physically meaningful molecular electrostatic potentials and a measure of electron correlation is inherent in all the nonbonded terms. Such a force field seems especially well suited for use in simulating nonbonded interactions between adsorbate molecules and mineral adsorbents.