When Hartree-Fock exchange admixture lowers DFT-predicted barrier heights: Natural bond orbital analyses and implications for catalysis

Abstract

The conventional wisdom in density functional theory (DFT) is that standard approximations systematically underestimate chemical reaction barrier heights and that exact (Hartree-Fock-like, HF) exchange admixture improves this. This conventional wisdom is inconsistent with the good performance of functionals without HF exchange for many reactions on metal catalyst surfaces. We have studied several "anomalous" gas-phase reactions where this conventional wisdom is upended, and a HF exchange admixture decreases or does not affect the predicted barrier heights [Mahler et al., J. Chem. Phys. 146, 234103 (2017)]. Here we show how natural bond orbital analyses can help identify and explain some factors that produce anomalous barriers. Applications to pnictogen inversion, standard benchmark reaction barrier datasets, and a model Grubbs catalyst illustrate the utility of this approach. This approach is expected to aid DFT users in choosing appropriate functionals, and aid DFT developers in devising DFT approximations generally applicable to catalysis.