Computer Simulation of Water and Aqueous Solutions

Abstract

During the last five years or so a forceful attempt (in fact a brute force attempt) has been made to obtain a structural view of liquid water and aqueous solutions by the application of a “first principles” approach starting from an initial statement of the total intermolecular potential energy of an assembly of N water molecules, or N water molecules and n solute molecules. The calculations that have been performed have been truly massive computer simulations of either (a) the time evolution of an isolated assembly of molecules or (b) the direct numerical evaluation of the equilibrium thermodynamic functions and spatial correlations using a Monte Carlo sampling of points in classical configuration space. Neither (a) or (b) constitutes a true molecular theory because in both methods the linkage between the microscopic description and the derived equilibrium and non-equilibrium properties vanishes in the black box complexity of the computation. The ambition of any true theory of a many-body system must be to relate the microscopic details of atomic and molecular interactions to the behavior of the observed thermodynamic functions. Statistical mechanics is the means by which this objective can in principle be realized; however, such a relation is seldom, if ever, accomplished to our satisfaction. Even in cases where an excessive modeling of the particle interactions yields an exact mathematical treatment, the key to such relationships is frequently obscured owing to the removal of all but the very coarse grained features of the Hamiltonian function in the modeling process.