Describing transition metal homogeneous catalysis using the random phase approximation

The performance of the direct random phase approximation (RPA) method based on a Kohn-Sham reference for transition metal chemistry is studied by making comparison to (dispersion-corrected) density functional theory (DFT) and (spin-scaled) Moller-Plesset theory. The recently developed local-pair coupled-cluster method DLPNO-CCSD(T) is used as a benchmark. Emphasis is placed on the study of complete realistic mechanisms and reactions involving large systems. Electronic energies for the mechanism of C-H and C-C bond activation by rhodium fragments are presented as well as for ruthenium-catalyzed olefin metathesis. In addition, the WCCR10 test set, which comprises ten reactions, is revisited, and reaction energies for the reaction of a -chloride-bridged palladacyclic dimer with phosphane ligands are presented. RPA yields results that are on average within 2-3 kcal/mol of the theoretical benchmark with a maximum deviation of 5 kcal/mol. Of the methods studied, RPA behaves most systematically and is able to provide results of similar accuracy to dispersion-corrected functionals. RPA can thus serve as a complementary method to DFT to obtain insight into transition metal chemistry. Attention is paid to the basis set convergence behavior of RPA as well.