A comparison between the predictions of vibrational transition moments and force fields using various gaussian basis sets

Abstract

A comparison has been made between the predictions of molecular force fields and dipole expansions of methyl fluoride and carbonyl fluoride using wavefunctions built from various basis sets. The basis sets used were STO nG (n = 3, 4, 6) and an extended gaussian product set with split valence shell (5-31G). There is little difference between the predictions on increasing n. The 5-31G basis set is much superior in predicting vibrational properties. It predicts quadratic force fields within 20 per cent and stretching dipole gradients within 25 per cent. Bending dipole derivatives are found to be more variable. Where electron rehybridization is likely to play an important role the computed results are poor. The STO nG functions generally predict dipole properties with the correct sign, but with stretching derivatives which are much too low. On the other hand this short basis is remarkably good at predicting molecular geometries.