Assessing the performance of density functional theory for the electronic structure of metal-salens: the d6-metals.

The energies and optimized geometries of the lowest lying singlet, triplet, and quintet states for the Fe(II)-, Co(III)-, Ni(IV)-, Ru(II)-, Rh(III)-, and Pd(IV)-salens have been computed with the B3LYP and BP86 density functional theory (DFT) methods, and the results are compared to more robust complete active-space self-consistent field (CASSCF) values. Density functional optimizations are performed using two different models of the salen ligand, and CASSCF relative energies at these DFT geometries show no appreciable difference whether the smaller or the larger model salen is considered. Unlike in our previous studies on the d0 and d2 metal-salens, DFT methods rarely predict the correct ordering of states compared to high-level complete active-space third-order perturbation theory (CASPT3) computations. The DFT energy gaps, moreover, are generally much smaller than those predicted by the CASPT3 method. Similarly to our previous studies, DFT optimized geometries closely match the CASSCF optimized geometries with errors mostly on the order of 0.1 A least root mean squared deviation. The electronic structure of the Co(III)- and Rh(III)-salens is particularly challenging, and significant differences between CASPT2 and CASPT3 relative energies were observed in these cases.