How accurate is DFT for iridium-mediated chemistry?

Abstract

Iridium chemistry is versatile and widespread, with superior performance for reaction types such as enantioselective hydrogenation and C−H activation. In order to gain insight into the mechanistic details of such systems, density functional theory (DFT) studies are often employed. But how accurate is DFT for modeling iridium-mediated transformations in solution? We have evaluated how well DFT reproduces the energies and reactivities of 11 iridium-mediated transformations, which were carefully chosen to correspond to elementary steps typically encountered in iridium-catalyzed chemistry (bond formation, isomerization, ligand substitution, and ligand association). Five DFT functionals, B3LYP, PBE, PBE0, M06L, and M11L, were evaluated as-is or in combination with an empirical dispersion correction (D2, D3, or D3BJ), leading to 13 combinations. Different solvent models (IEFPCM and SMD) were evaluated, alongside various correction terms such as big basis set effects, counterpoise corrections, frequency scaling, and different entropy modifications. PBE-D type functionals are clearly superior, with PBE-D2,IEFPCM providing average absolute errors for uncorrected Gibbs free energies of 0.9 kcal/mol for the nine reactions with a constant number of moles (1.2 kcal/mol for all 11 reactions). This provides a straightforward and accurate computational protocol for computing free energies of iridium-mediated transformations in solution. However, because the good results may originate from favorable error cancellations of larger and oppositely signed enthalpy and entropy errors, this protocol is recommended for free energies only.