Simulation of heme using DFT + U: a step toward accurate spin-state energetics.

We investigate the DFT + U approach as a viable solution to describe the low-lying states of ligated and unligated iron heme complexes. Besides their central role in organometallic chemistry, these compounds represent a paradigmatic case where LDA, GGA, and common hybrid functionals fail to reproduce the experimental magnetic splittings. In particular, the imidazole pentacoordinated heme is incorrectly described as a triplet by all usual DFT flavors. In this study, we show that a U parameter close to 4 eV leads to spin transitions and molecular geometries in quantitative agreement with experiments and that DFT + U represents an appealing tool in the description of iron porphyrin complexes, at a much reduced cost compared to correlated quantum-chemistry methods. The possibility of obtaining the U parameter from first principles is explored through a self-consistent linear-response formulation. We find that this approach, which proved to be successful in other iron systems, produces in this case some overestimation with respect to the optimal values of U.