Ab initio STO-3G molecular orbital methods have been used to calculate equilibrium geometries for the dimers, BeSi(OH)7 and H6BeSiO7, to model the BeOSi bonds in beryllosilicate minerals. The optimized geometry and calculated potential energy surface for the dimer BeSi(OH)7 are consistent with the Be–O and Si–O bridging bond length and BeOSi angle variations observed for these minerals. This indicates that the structures of beryllosilicates are determined in part by the same kind of short range forces that govern the geometry of the BeOSi bond in the dimer. Calculations for H6BeSiO7 indicate that the BeOSi angle is an intrinsic consequence of the electronic structure of the BeOSi bond and not merely a property of the coordination number of the bridging oxygen. The calculations also indicate that coupled substitutions of Be for either Si or Al will tend to occur in silicate structures that have TOT angles near the calculated equilibrium value (129°) for the BeOSi angle.
A refinement of the anisotropic temperature factors and Δρ maps of asbecasite, Ca3(Ti,Sn)As6Si2Be2O20, calculated with the published data of Cannillo et al. (1969) indicate that the BeOSi bond in the mineral may not be straight as reported. The potential energy surface for H6BeSiO7 appears to conform with the anisotropic charge density distribution observed for the BeOSi bond in asbecasite. When the short Be–O bond in the mineral is omitted from a regression analysis of experimental data, only two percent of the variation in d(Be–O) can be explained in terms of a linear dependence on the BeOSi angle.