Predicting proton transfer barriers with density functional methods

We study the ability of recently developed exchange-correlation functionals to model proton transfer in the H5O2+ model for water and in malonaldehyde. We are particularly interested in assessing the accuracy of the VSXC functional developed in our research group and compare it to other hybrid functionals and high-level wavefunction methods like coupled-cluster theory. We find that VSXC and the hybrid functionals all underestimate proton transfer barriers for H5O2+ but still give a good sense of the overall nature of the potential energy surface. In malonaldehyde, only VSXC produces a proton transfer activation barrier similar to those provided by high-level coupled-cluster methods.