A MECHANISM FOR THE FERROMAGNETIC COUPLING IN DECAMETHYLFERROCENIUM TETRACYANOETHENIDE

We discuss a model for the ferromagnetic coupling in a stack of alternating [Fe(C5Me5)2]+ and [TCNE]- ions. We present evidence against the proposed mechanism based on a model suggested by McConnell in which the ferromagnetism arises from configurational mixing of a virtual triplet excited state (D2+A2-) with the ground state D+A- (D = Fe(C5Me5)2, A = TCNE). Instead, we point to a different kind of configurational mixing that involves a singly excited configuration on ferrocenium. The excitation arises from a transition between orbitals of two e1g sets, which are the only ones with considerable covalent character. First, we set up the symmetry-adapted 2E2g ground- and excited-state wave functions for a single ferrocenium ion and discuss their mixing. It is shown that this mixing gives rise to an important spin correlation effect inducing negative spin densities on the Cp rings. Then the ferrocenium doublet wave functions are coupled to the TCNE doublet to form singlet and triplet wave functions for a donor-acceptor pair. The resulting singlet-triplet energy difference can be interpreted as an exchange effect between the negative spin density on the Cp rings and the radical electron density on TCNE in complete analogy to the Heitler-London model for the H-2 molecule. We obtain a term comprising a two-electron exchange integral and products of overlap and resonance integrals. Due to the negative sign of the spin density on the Cp rings, the signs of these contributions and, thus, the sign of the singlet-triplet energy splitting are inverted in our case as compared to the Heitler-London model, resulting in a triplet ground state for a donor-acceptor pair and, thus, ferromagnetic coupling. The exchange between the spin density on Fe and the one on TCNE can be neglected because the corresponding overlap densities are very small due to the large distance between Fe and TCNE (5 angstrom as compared to 3.5 angstrom for the distance Cp-TCNE). Our model corresponds to another coupling mechanism suggested by McConnell, which has nothing to do with the configurational mixing of donor-acceptor charge-transfer states.