Density functional theory (DFT) has emerged as a powerful means of practical predictions in computational quantum chemistry, providing often reliable relative free energies for reaction barriers, isomerizations, stereoselectivities, etc. While the two main sources of errors in such predictions (choice of basis set and exchange-correlation functional) are well recognized, there is a third axis of error that often goes undiscussed: choice of integration grid. We will present sobering data concerning the sensitivity of DFT-computed free energies to molecular orientation relative to the lab-fixed Cartesian frame. In particular, we will show that DFT-computed free energies for many realistic chemical processes can vary by as much as 3 kcal/mol by simply changing molecular orientations. These variations arise from the sensitivity of low-frequency vibrational modes to the relative orientation of atom-centered integration grids, whose effects are amplified in the entropic contribution to computed free energies. This is a major cause for concern, as many chemical problems to which DFT is applied (e.g. stereoselectivity, etc.) hinge on predicted relative free energies of this same order of magnitude.

Other axis of error: DFT integration grids and relative free energies of organic molecules

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.