Cobaltocenylidene: A Mesoionic Metalloceno Carbene, Stabilized in a Gold(III) Complex

Abstract Oxidative addition of cobaltoceniumdiazonium bis(hexafluoridophosphate) with (pseudo)halide aurates gave gold(III) complexes containing zwitterionic cobaltoceniumide as a ligand. Its selenium derivative, cobaltoceniumselenolate, was obtained by an electrophilic aromatic substitution reaction of iodocobaltocenium iodide with Na2Se. Spectroscopic and structural data in combination with DFT calculations showed that this cobaltocenylidene species is a mesoionic carbene quite different from common N‐heterocyclic carbenes. Its ligand properties (TEP, singlet‐triplet gap, nucleophilicity, π‐acidity, Brønsted basicity) are in part comparable to those of cyclic (amino)(alkyl/aryl)carbenes. Electrochemistry data showed that the mesoionic cobaltoceniumides are more electron‐rich than their parent ferrocenes. The reversible reduction of the tricyanido gold complex appears 50 mV negative of the cobaltocenium/cobaltocene couple, whereas that of the selenide derivative is shifted cathodically by 550 mV.


Electrochemical Section
Electrochemical measurements. Electrochemical experiments were executed in a home-built cylindrical vacuum-tight one-compartment cell. A spiral-shaped Pt wire and a Ag wire as the counter and pseudoreference electrodes are sealed into glass capillaries and fixed by Quickfit screws via standard joints. A platinum electrode is introduced as the working electrode through the top port via a Teflon screw cap with a suitable fitting. It is polished with first 1 μm and then 0.25 μm diamond paste before measurements. The cell can be attached to a conventional Schlenk line via a side arm equipped with a Teflon screw valve, allowing experiments to be performed under an argon atmosphere with approximately 5 mL of analyte solution. A 0.1 M solution of NBu4 + PF6in THF was used as the supporting electrolyte. Referencing was done with addition of an appropriate amount of ferrocene (Cp2Fe) as an internal standard to the analyte solution after all data of interest had been acquired. Representative sets of scans were repeated with the added standard. Electrochemical data were acquired with a computer controlled BASi CV50 potentiostat.
Spectroelectrochemistry (in THF, NBu4 + PF6 -, 0.1 M at r. t.) was performed with an optically transparent thinlayer electrochemical (OTTLE) cell was home-built and followed the design of Hartl et al. 5 It comprised a Pt working and counter electrode and a thin silver wire as a pseudoreference electrode sandwiched between two CaF2 windows of a conventional liquid IR cell. The working electrode is positioned in the center of the spectrometer beam.

Computational Section
Computational methodology. Structures of the free CcC ligand were derived from the X-ray crystal structure of (CcC)Au(CN)3 and subjected to structure optimization. Density functional theory calculations were performed with the program suite Turbomole. 6 Two different density functionals, the pure BP86 7,8 as well as the hybrid PBE0 9 functional with 25% Hartree-Fock exchange were used in combination with the Schäfer et al. triple-zeta basis set. 10 For Au, an effective core potential was provided for the inner electrons. 11 A hybrid and a non-hybrid density functional were chosen because it is well known that the singlet-triplet energy splitting depends on the amount of Hartree-Fock type exchange in the density functional. 12 BP86 was used in combination with the resolution-of-identity technique to reduce computational costs. 13 The effect of Grimme's empirical dispersion corrections of the Becke-Johnson type was also tested. 14 Solvation was treated implicitly by the conductor-likescreening model (COSMO) as implemented in Turbomole: 15 the solvent is described as a dielectric continuum and the solute is placed in a cavity in this continuum. To match experimental conditions, the dielectric constant of nitromethane is chosen, (ε = 35.9) as well as a dielectric constant of DMSO (ε = 46.7) for sake of comparison. If not noted otherwise, structures are fully optimized.
The molecular electrostatic potential (MEP) was calculated to evaluate the relative polarity and thus the reactivity of the molecule under study, here CcC. It is plotted on a density isovalue of 0.02 a.u., whereas red areas indicate a negative MEP and blue areas a positive MEP. 16 Proton affinities were calculated as the negative enthalpy -ΔH of the following reaction in gas phase: pKa values in solution were obtained with an implicit solvent description from the Gibbs free energy of the solvation process ΔG Rxn (solv) according to RTln (10) (2) with R being the universal gas constant and T denoting the temperature. ΔG Rxn (solv) cannot be calculated directly but has to be obtained via the thermodynamic cycle depicted in Scheme 1 by calculation of the gas phase reaction free energy ΔG Rxn (g) and the free energies of solvation for each species, ΔG(solv) (HA), ΔG(solv) (H + ), and ΔG(solv) (A -). Here, ΔG(solv) (HA) is ΔG(solv) (CcCH) and ΔG(solv) (A -) corresponds to ΔG(solv) (CcC).
The gas phase free energy of H + , G(g) (H + ) = -6.29 kcal/mol, was obtained from the Sackur-Tetrode equation and translational energy at 298 K 17 , while for the free energy of solvation, ΔG(solv) (H + ), the recommended literature value of -259.80 kcal/mol was used. 18 Tolman Electronic Parameters (TEP) are a simple means to experimentally characterize NHCs and related species in terms of their electronic properties as ligands in organometallic complexes. Quantum chemical TEP calculations of these complexes typically depend, amongst other, on the employed exchange-correlation functional and transferability of the results may not be straight forward. Fortunately, a parameterization scheme has been developed by Mathew and Sureh, which allows for TEP calculations of the ligand alone and which only requires the molecular electrostatic potential at the carbene (Vc). 19 For sake of comparison, calculation of TEPs in this study were performed with Gausian09 20 using the density functionals B3LYP, 21 BP86, 7,8 and M05 22 in combination with the Pople type basis set 6-311+G(d,p) basis set as implemented in Gaussian09.
Structures and MEPs were visualized with PyMol. 23 Scheme 1. Thermodynamic cycle used for the calculation of pKa values. While ΔG Rxn (solv) (depicted by the red arrow) is the property of interest, it has to be calculated via the gas phase reaction free energy and the solvation free energies of each species.

Singlet-Triplet Splitting
Singlet-triplet relative energy splittings of fully optimized NHC and CcC were calculated. While we found that CcC shows a significantly smaller singlet-triplet gap as NHC as may be seen from Table S19, individual values depend on the basis set, and more importantly, on density functional employed. The singlet-triplet splitting decreases if a density functional with Hartree-Fock type exchange as the PBE0 density functional with 25% HF exchange is used. Empirical dispersion corrections (denoted as BJ) exert only a minor influence on the energy splittings implying structural effects to be small.

Structural Parameters
Structural parameters of the ligand CcC and the complex (CcC)Au(CN)3 (both depicted in Figure S27) are listed in Table S20 and S21. These include Co-C and C-C bond lengths, selected bond angles and dihedrals. In the case of CcC, the angle between the two Cp-planes (denoted as ΘCp-Cp) as well as the distance of the Co-Cp axis to the center of mass of the upper Cp (dout-of-center) were also evaluated.
Structural parameters agree well with experimentally determined values for (CcC)Au(CN)3 and small differences may be attributed to packing effects. Coordination of CcC to Au(CN)3 has only a small impact on its structural parameters and Co-C and C-C bond lengths are very similar (compare Table S20 and Table S21).
Concerning CcC, ΘCp-Cp and the distance of the Co-Cp axis to the center of mass of the upper Cp (dout-of-center) are very small, the two Cp rings are almost perfectly stacked on top of each other, even though the two Cp planes are slightly rotated in (CcC)Au(CN)3.  ΘCp-Cp denotes the angle between the two Cp-planes, and dout-of-center denotes the distance of the Co-Cp axis to the center of mass of the upper Cp. Angles are given in °, bond lengths and distances in Å. CcC

Tolman Electronic Parameter
Tolman Electronic Parameter were calculated according to a parametrization by Mathew and Suresh on the PBE0/BJ/def2-TZVPP optimized structures. As may be seen from calculated NHC TEP values (Table S22), they are little functional dependent and within an error range of around 2 cm -1 to the experimental value of 2051.2 cm -1 . While for CcC no experimental data is available, all calculated TEP are around 2038 cm -1 and significantly smaller than those of NHC. To separate structural from electronic effects, a TEP calculation with B3LYP/6-311++G(d,p) was performed on the BP86/def2-TZVPP optimized but a very similar value was obtained and electronic effects account for less than 1 cm -1 .

Proton Affinities and pKa Values
Proton affinities and pKa values are a further characteristic of NHC and related ligands and are also investigated by DFT (Table S23). Proton affinities were calculated according to Eq. (1), whereas pKa values according to the thermodynamic cycle depicted in Scheme 1. The solvent was described implicitly by a continuum solvation model (see Computational Methodology for details). While for NHC an experimental pKa of approx. 24 was found 19 , no experimental value is available for CcC.
Although individual pKa values depend on the density functional employed and -in the case of NHC -differ somewhat from the experimental values, CcC shows significantly higher values between 38 and 40 than NHC. Thus, it is a much stronger base in comparison.