Relative energy of the high-(5T2g) and low-(1A1g) spin states of the ferrous complexes [Fe(L)(NHS4)]:: CASPT2 versus density functional theory

High-level ab initio calculations using multiconfigurational perturbation theory [complete active space with second-order perturbation theory (CASPT2)] were performed on the transition energy between the lowest high-spin (corresponding to (T-5(2g)) in O-h) and low-spin (corresponding to (1)A(1g) in O-h) states in the series of six-coordinated Fe(II) molecules [Fe(L)(NHS4)], where NHS4 is 2,2(')-bis(2-mercaptophenylthio)diethylamine dianion and L=NH3, N2H4, PMe3, CO, and NO+. The results are compared to (previous and presently obtained) results from density functional theory (DFT) calculations with four functionals, which were already shown previously by Casida and co-workers [Fouqueau , J. Chem. Phys. 120, 9473 (2004); Ganzenmuller , ibid. 122, 234321 (2005); Fouqueau , ibid. 122, 044110 (2005); Lawson Daku , ChemPhysChem 6, 1393 (2005)] to perform well for the spin-pairing problem in these and other Fe(II) complexes, i.e., OLYP, PBE0, B3LYP, and B3LYP(*). Very extended basis sets were used both for the DFT and CASPT2 calculations and were shown to be necessary to obtain quantitative results with both types of method. This work presents a sequel to a previous DFT/CASPT2 study of the same property in the complexes [Fe(H2O)(6)](2+), [Fe(NH3)(6)](2+), and [Fe(bpy)(3)](2+) [Pierloot , J. Chem. Phys. 125, 124303 (2006)]. The latter work was extended with new results obtained with larger basis sets and including the OLYP functional. For all considered complexes, the CASPT2 method predicts the correct ground state spin multiplicity. Since experimental data for the actual quintet-singlet (free) energy differences are not available, the performance of the different DFT functionals was judged based on the comparison between the DFT and CASPT2 results. From this, it was concluded that the generalized gradient OLYP functional performs remarkably well for the present series of ferrous compounds, whereas the success of the three hybrid functionals varies from case to case. (c) 2008 American Institute of Physics.