Basis set dependence of S[double bond, length as m-dash]O stretching frequencies and its consequences for IR and VCD spectra predictions.

Benchmarking functionals and basis sets for the computational prediction of molecular properties is usually done on very small model systems. Larger organic molecules containing heavier second row atoms are not the typical model structures. We herein present the first survey of basis sets and functionals for the prediction of the IR and VCD spectra of chiral tosylates and sulfinates as we noted drastic deviations between computed harmonic frequencies obtained at B3LYP/6-311++G(2d,p) level of theory and those observed in experimental solution phase IR and VCD spectra. We show that the harmonic frequencies of the asymmetric and symmetric S[double bond, length as m-dash]O stretching modes of tosylates are predicted at significantly too low vibrational frequencies if the employed basis set does not provide higher order polarization functions. The results of our benchmarks show that at least the 6-311G(3df,2dp) basis (or equivalent Dunning and Ahlrichs variants) should be used.