A theoretical study of zero-field splitting in Fe(IV)S ₆ (S=1) and Fe(III)S ₆ (S=1/2) core complexes, [Fe ^IV(Et ₂dtc) _3-n(mnt) _n] ^(n-1)- and [Fe ^III(Et ₂dtc) _3-n(mnt) _n] ^n- (n=0, 1, 2, 3): The origin of the magnetic anisotropy

Multireference ab initio calculations and ligand field analysis of a series of complexes with Fe(IV)S ₆ (S=1) [Fe ^IV(Et ₂dtc) _3-n(mnt) _n] ^(n-1)- and Fe(III)S ₆ (S=1/2) [Fe ^III(Et ₂dtc) _3-n(mnt) _n] ^n- cores ((mnt) ^2-=maleonitriledithiolate (2-), (Et ₂dtc) ^1-=diethyldithiocarbamato (1-) ligands, n=0, 1, 2, 3) are reported and used to understand their magnetic and spectroscopic (ESR) properties. These systems feature large and variable values of D for the S=1 complexes of Fe(IV) and strongly anisotropic g-tensors for the S=1/2 complexes of Fe(III). The calculations are in good to excellent agreement with experiment. We utilize a historic concept put forward by Orgel as early as 1961 [39] in order to analyze the computational data. The non-additive contributions to ligand field due to the π-conjugated systems of the chelate ligands mnt ^2- and Et ₂dtc ^- are responsible for the large magnetic anisotropy. These contributions are even more important than geometric distortions imposed by the rigid ligand cores. The correlations have been demonstrated and quantified using an extended ligand field (LF) model with parameters adjusted to complete active space self-consistent field (CASSCF) calculations corrected for dynamic correlation with the second order N-electron valence perturbation theory (NEVPT2). According to this analysis, the topology of the intrinsic π-electron system of the mnt ^2- and Et ₂dtc ^- ligands causes a splitting of the octahedral t _2g orbitals of different sign for mnt ^2- (e>a ₁, in-phase coupling) and Et ₂dtc ^1- (a ₁>e, out-of-phase coupling). When combined with the π-donor ability of the mnt ^2- and Et ₂dtc ^- shown by theory and experiment to be much stronger in mnt ^2- compared to Et ₂dtc ^1- this leads to large orbital contributions to the magnetic moment and to a negative D for [Fe(mnt)(dtc) ₂] with an easy axis of magnetization bisecting the SFeS(mnt) bite angle. Using this ab initio based renewed concept, field dependent isothermal magnetizations reported previously (Milsmann et al., 2010 [25]) have been re-interpreted. We show that the orthorhombic anisotropy for [Fe ^IV(Et ₂dtc)(mnt) ₂] ^1- (2 ^ox) and [Fe ^IV(Et ₂dtc) ₂(mnt)] ⁰ (3 ^ox), that has never been discussed before, leads to large zero-field splitting parameter E. At the same time it is pointed out, that the D and E spin-Hamiltonian parameters cannot be uniquely extracted from a fit to the magnetic susceptibility data, unless combined with other sophisticated spectroscopic experiments. Applying the same anisotropic π-bonding model, orbital contributions leading to strongly anisotropic g-tensors reported from simulation of ESR data of the Fe(III)S ₆ (S=1/2) cores in complexes [Fe ^III(Et ₂dtc) _3-n(mnt) _n] ^n- (n=0, 1, 2, 3) have been rationalized.