Abstract
Copper is ubiquitous and its one-electron redox chemistry is central to many catalytic processes. Modeling such chemistry requires electronic structure methods capable of the accurate prediction of ionization energies (IEs) for compounds including copper in different oxidation states and supported by various ligands. Herein, we estimate IEs for 12 mononuclear Cu species previously reported in the literature by using 21 modern density functionals and the DLPNO-CCSD(T) wave function theory model; we consider extrapolated values of the latter to provide reference values of acceptable accuracy. Our results reveal a considerable diversity in functional performance. Although there is nearly always at least one functional that performs well for any given species, there are none that do so for every member of the test set, and certain cases are particularly pathological. Over the entire test set, the SOGGA11-X functional performs best with a mean unsigned error (MUE) of 0.22eV. PBE0, B97X-D, CAM-B3LYP, M11-L, B3LYP, and M11 exhibit MUEs ranging between 0.23 and 0.34eV. When including relativistic effects with the zero-order regular approximation, B97X-D, CAM-B3LYP, and PBE0 are found to provide the best accuracy.