Prediction of proton affinities and protonation sites using a multivariate linear correlation

Abstract

We have shown, for a large set of compounds, including oxygen, nitrogen, and carbon bases, that it is possible to express the gas-phase proton affinities (PA) as a linear function of 1s binding energies and first ionization potentials. This multivariate correlation holds for molecules of different homologous series and we conclude that it contains the two most important factors which contribute to the basicity of a given molecule. We have also shown that some species which deviate from single correlations between PAs and 1s binding energies, e.g. NN-dialkylanilines, naphthalene, etc. fit our multivariate correlation very well. These systems are stronger bases than expected from the 1s ionization energy of their basic centres, due to their special ability to accommodate a highly localized positive charge. We have shown that similar correlations can be obtained using orbital energies, calculated using an STO3G minimal basis set, instead of experimental ionization energies. Several cases are fully discussed.