Tests of effective pair potentials for water: Predicted ice structures

As a test of the accuracy of the several proposed effective pair potentials for water, we have examined how well they reproduce the structures and properties of the proton-ordered ices II, VIII, and IX, and those of a hypothetical proton ordered form of ice I which is closely related to ice Ih. Despite leading to a consistent underestimate of all the ice densities by roughly 21%, the ab initio pair potential calculated by Matsuoka, Clementi, and Yoshimine succeeds better than any other in predicting the hydrogen bonding topology and the O–O–O angles characteristic of the various ice lattices. Some of the proposed water–water potentials predict the existence of ice polymorphs entirely different from known forms of ice, and/or give rise to structures with bifurcated hydrogen bonds. Using our new results, and others in the literature, we give a detailed discussion of the features of the MCY, ST2, and RSL2 potentials, each considered a representative of a class of potentials.