Spin states, bonding and magnetism in mixed valence iron(0)–iron(II) complexes

"Xenophilic" complexes offer metal-metal bonds between disparate metal sites, but the nature of the metal-metal bonding is often unclear. Here, we describe two novel complexes with unsupported Fe-Fe bonds, LxFe-Fp (LX = b-aldiminate or b-diketim-inate; Fp = Fe(CO)2Cp), that offer insight into the Fe-Fe bonding. Mössbauer, magnetism and DFT analysis indicate that the most accurate electronic structure description is LFeII←Fe0(CO)2Cp, in which the Fe(CO)2Cp is low-spin iron(0) and acts as an X-type ligand toward the high spin iron(II) of the LFe fragment. This largely electro-static interaction has a bond order of only 0.5. The three-coordinate high-spin iron(II) site has large zero-field splitting, and in addition its Mössbauer parameters can be used to rank the Fp- "metalloligand" as a donor; it is nearly as strong a donor as phosphides and alkyls.


General considerations
All manipulations were performed in an argon-filled MBraun glovebox maintained below 1 ppm of O2 and H2O or under an N2 atmosphere using standard Schlenk techniques unless mentioned otherwise. Glassware was oven-dried at 150 °C for at least 12 h prior to use. Celite and molecular sieves were dried above 200 °C under vacuum for at least 12 h. Pentane, THF, hexanes, benzene, toluene, and diethyl ether were purified by passage through activated alumina and Q5 columns from Glass Contour Co, under argon. Benzene-d6 and THF-d8 were vacuum transferred from a solution of potassium benzophenone ketyl and was stored over 4 Å molecular sieves. KC8 was prepared in an N2 glovebox by manually grinding stoichiometric amounts of potassium metal and graphene at 90 °C until observation of a homogenous bronze-colored powder.
L b H, [1] [L a FeCl]2, [2] FeCl2(THF)1.43, [3] and K[CpFe(CO)2] [4] were synthesized according to literature procedures. NMR data were collected on Agilent 400 or 500 MHz spectrometers. Chemical shifts in 1 H NMR spectra are referenced to the residual protiated solvent peaks of C6D5H (δ 7.16 ppm) and THF-d8 (δ 3.58 ppm). Solution magnetic susceptibilities were determined by the Evans method. Elemental analyses were performed at the CENTC Elemental Analysis Facility at the University of Rochester. IR spectra were collected on an Alpha Platinum ATR IR Spectrometer. UV-vis spectra were recorded on a Cary 50 spectrometer using Schlenk-adapted quartz cuvettes with a 1, 2, or 10 mm path length.

Synthetic procedures Synthesis of [L b FeCl]2
1,3-bis(2,4,6-trimethylphenylimido)propane (L b H, 432 mg, 1.41 mmol) was dissolved in THF (5.5 mL). A solution of benzylpotassium (183 mg, 1.41 mmol) in THF (6.5 mL) was added dropwise to the L b H solution and this mixture was stirred for 10 minutes. FeCl2(THF)1.43 (324 mg, 1.41 mmol) was suspended in THF (5 mL) in a 100 mL bomb flask and stirred for 10 minutes. The solution of KL b was added dropwise to the suspension of FeCl2(THF)1.43 (N.B. the inverse addition produces significant amounts of L b 2Fe). The reaction mixture was stirred at room temperature for 30 minutes and then was heated to 85 °C for 1 hour.
The mixture was then concentrated under reduced pressure and the resulting red oil was heated at 85 °C under vacuum to give dark orange solids. The solids were extracted with toluene (12 mL), filtered, and concentrated to 8 mL. Pentane (11 mL) was layered on top and the solution was cooled to -40 °C overnight to yield orange crystalline solids (183 mg, 33%). The mother liquor was evaporated under reduced pressure and the solid was extracted with toluene (3 mL), filtered, and pentane (17 mL) was layered on top before cooling it to -40 °C overnight to yield a second crop of orange crystals (169 mg, total yield 63%). 36.0 (1H, α-CH). The resonance associated with the imine-CH was not found, likely due to its proximity to the paramagnetic Fe, as has previously reported for the analogous Co complex. [1]                 been modelled with a nested fit, while the right has been modelled using a staggered fit (See table). The circles represent the data, the solid colored lines are the fits, and the grey line represents the residual.      were performed under applied magnetic fields of 0.05 T and 0.10 T, and 0.5 T and corrected for the diamagnetism of each sample and eicosane, estimated using Pascal's constants. [9] Variable field, variable temperature magnetization measurements (reduced magnetization) were performed under applied magnetic fields of 1-7 T in 1 T increments, in the temperature range of 2-10 K. Dc magnetic susceptibility data and reduced magnetization data were simulated using the program MagProp in DAVE 2.0. [10] The spin

UV/vis spectra of metal complexes
Hamiltonians employed accounted for g-anisotropy and axial and transverse zero-field splitting, D and E, respectively. Magnetic data for 1a and 1b were modelled according to the Van Vleck model using the following spin Hamiltonian: Ĥ = Ŝ # $ + (Ŝ ) $ -Ŝ + $ ) + . µ 0 , (i = x, y, z). In this Hamiltonian, D and E are the axial and transverse zero-field splitting parameters, respectively, and g is the electron g-value. No satisfactory simulation of the reduced magnetization data for 1b was achieved due to the presence of iron impurities, as evident by the 11.9% impurity determined from Mössbauer spectroscopy. However, simulation of the dc magnetic susceptibility data for 2 afforded a preliminary value for D, while E was fixed to be zero.   Table 1 in the manuscript.  Table 1 in the main manuscript.   Table 1 in the manuscript.

Crystallographic data
Low-temperature diffraction data (ω-scans) were collected on a Rigaku MicroMax-007HF diffractometer coupled to a Saturn994+ CCD detector with Cu Kα (λ = 1.54178 Å) for the structure of [L b FeCl]2. Similar low-temperature diffraction data (ω-scans) were collected on a Rigaku MicroMax-007HF diffractometer coupled to a Dectris Pilatus3R detector with Mo Kα (λ = 0.71073 Å) for the structure of 1a and 1b. All diffraction images were processed and scaled using Rigaku Oxford Diffraction software (CrysAlisPro; Rigaku OD: The Woodlands, TX, 2015). The data for [L b FeCl]2 was refined as a 2-component twin. The fractional volume contribution of the minor twin component was freely refined to a converged value of 0.3502 (13). The structure was solved with SHELXT and was refined against F 2 on all data by full-matrix least squares with SHELXL. [11] All non-hydrogen atoms were refined anisotropically. Hydrogen atoms were included in the model at geometrically calculated positions and refined using a riding model. The

Computational details
Geometry optimizations, frequency calculations, and calculations of Mössbauer parameters were performed using the ORCA program package (Version 3.0.3). [12] Optimized geometries were calculated using the B3LYP functional and the scalar relativistically recontracted version of the Aldrichs triple-z basis set (def2-TZVP). The scalar relativistic zero-order approximation (ZORA) was used to correct for relativistic effects.
The atom-pairwise dispersion correction with the Becke-Johnson damping scheme (D3BJ) was used on all atoms. The calculated structures were confirmed to be minima based on the absence of imaginary frequencies from frequency calculations on the optimized geometries. Mössbauer parameters were calculated using the correlations that were described previously. [13] Unrestricted corresponding orbitals (UCOs) were calculated using the ORCA program package (Version 4.1.2) with the UCO keyword. Orbitals are plotted with an isosurface value of 0.05. The UCOs from the B3LYP/def2-TZVP calculation are shown in Figure 5 in the main text. Intrinsic atomic orbitals and intrinsic bond orbitals (IAOIBO) were calculated with B3LYP functional and def2-TZVP/D3BJ/ZORA basis set, as described above. [14] The Fe-Fe bonding orbitals were identified by the contributions from Fe1 and Fe2, with 61α and 54β being the only orbitals with contributions from both metal centers.