Relative energy of the high-(5T2g) and low-(1A1g) spin states of [Fe(H2O)6]2+, [Fe(NH3)6]2+, and [Fe(bpy)3]2+: CASPT2 versus density functional theory

High-level ab initio calculations using the CASPT2 method and extensive basis sets were performed on the energy differences of the high-[(5)T(2g):t(2g) (4)e(g) (2)] and low-[(1)A(1g):t(2g) (6)] spin states of the pseudo-octahedral Fe(II) complexes [Fe(H(2)O)(6)](2+), [Fe(NH(3))(6)](2+), and [Fe(bpy)(3)](2+). The results are compared to the results obtained from density functional theory calculations with the generalized gradient approximation functional BP86 and two hybrid functionals B3LYP and PBE0, and serve as a calibration for the latter methods. We find that large basis set CASPT2 calculations may provide results for the high-spin/low-spin splitting DeltaE(HL) that are accurate to within 1000 cm(-1), provided they are based on an adequately large CAS[10,12] reference wave function. The latter condition was found to be much more stringent for [Fe(bpy)(3)](2+) than for the other two complexes. Our "best" results for DeltaE(HL) (including a zero-point energy correction) are -17 690 cm(-1) for [Fe(H(2)O)(6)](2+), -8389 cm(-1) for [Fe(NH(3))(6)](2+), and 3820 cm(-1) for [Fe(bpy)(3)](2+).