Vanadocene de Novo: Spectroscopic and Computational Analysis of Bis(η5-cyclopentadienyl)vanadium(II)

The magnetic and electronic properties of the long-known organometallic complex vanadocene (VCp2), which has an S = 3/2 ground state, were investigated using conventional (X-band) electron paramagnetic resonance (EPR) and high-frequency and -field EPR (HFEPR), electronic absorption, and variable-temperature magnetic circular dichroism (VT-MCD) spectroscopies. Frozen toluene solution X-band EPR spectra were well resolved, yielding the V-51 hyperfine coupling constants, while HFEPR were also of outstanding quality and allowed ready determination of the rigorously axial zero-field splitting of the spin quartet ground state of VCp2: D = +2.836(2) cm(-1), g(perpendicular to) = 1.991(2), g(parallel to) = 2.001(2). Electronic absorption and VT-MCD studies on VCp2 support earlier assignments that the absorption signals at 17 000, 19 860, and 24 580 cm(-1) are due to ligand-field transitions from the (4)A(2g) ground state to the E-4(1g), E-4(2g), and E-4(1g) excited states, using symmetry labels from the D-5d point group (i.e., staggered VCp2). Contributions to the D parameter in VCp2 and further insights into electronic structure were obtained from both density functional theory (DFT) and multireference SORCI computations using X-ray diffraction structures and DFT-energy-minimized structures of VCp2. Accurate D values for all models considered were obtained from DFT calculations (D = 2.85-2.96 cm(-1)), which was initially surprising, because the orbitally degenerate excited states of VCp2 cannot be properly treated by DFT methods, as they require a multideterminant description. Therefore, D values were also computed using the SORCI (spectroscopically oriented configuration interaction) method, which provides multireference descriptions of ground and excited states. SORCI calculations gave accurate D values (2.86-2.90 cm(-1)), where the dominant (similar to 80%) contribution to D arises from spin-orbit coupling between ligand-field states, with the largest contribution from a low-lying (2)A(1g) state. In contrast, the D value obtained by the DFT method is achieved only fortuitously, through cancellation of errors. Furthermore, the SORCI calculations predict ligand-field excited-state energies within 1300 cm(-1) of the experimental values, whereas the corresponding time-dependent DFT calculations fail to reproduce the proper ordering of excited states. Moreover, classical ligand-field theory was validated and expanded in the present study. Thus older theory still has a place in the analysis of paramagnetic organometallic complexes, along with the latest ab initio methods.