Molecular prototypes for simple SiO ₂ potentials

A popular class of interionic pair potentials for simulation of the properties of condensed-phase SiO ₂ uses two calibrations: first-principles electronic structure results for a small cluster and experimental crystalline data. The clusters are argued to be broadly valid prototypes for the local structural and bonding behavior of condensed-phase SiO ₂. This approach introduces an issue: the extent to which there is prototypical computed behavior for small silicates irrespective of cluster size, symmetry constraints, methodological refinement, and accuracy of implementation (basis set). By extension, the issue of commonality vs diversity in small silicates arises. We address those issues by comparative study of 12 systems containing one or two Si atoms in combination with H, O, and bare protons. We use several different levels of theoretical refinement and various basis sets. Though some chemical trends are clear, there is no clear single prototype for condensed-phase SiO ₂ (including no clear preference to tetrahedral symmetry). We find a wide diversity of bond lengths, bond angles, and interaction energies and significant methodological consequences that go largely ignored in the potential-fitting literature.