Extensive Redox Non-Innocence in Iron Bipyridine-Diimine Complexes: a Combined Spectroscopic and Computational Study

Metal–ligand cooperation is an important aspect in earth-abundant metal catalysis. Utilizing ligands as electron reservoirs to supplement the redox chemistry of the metal has resulted in many new exciting discoveries. Here, we demonstrate that iron bipyridine-diimine (BDI) complexes exhibit an extensive electron-transfer series that spans a total of five oxidation states, ranging from the trication [Fe(BDI)]3+ to the monoanion [Fe(BDI]−1. Structural characterization by X-ray crystallography revealed the multifaceted redox noninnocence of the BDI ligand, while spectroscopic (e.g., 57Fe Mössbauer and EPR spectroscopy) and computational studies were employed to elucidate the electronic structure of the isolated complexes, which are further discussed in this report.


General Information
All reactions were performed at room temperature either by using standard Schlenk techniques or by using an N2-filled M. Braun Glovebox unless otherwise specified. Glassware was oven dried at 140 °C for at least 2h prior to use, and allowed to cool under vacuum. All reagents were used as received unless mentioned otherwise. 6,6'-Diacetyl-2,2' bipyridine were synthesized according to published procedures. S1-S3 Anhydrous iron triflate (Fe(OTf)2), iron chloride (FeCl3), and 2,3,6-trimethylaniline, were purchased from Strem Chemicals, Alfa Aesar and Sigma Aldrich respectively. Anhydrous unstabilized tetrahydrofuran (THF) and diethylether (Et2O) were purchased from Sigma Aldrich and used as received.

Physical Methods
Electrochemical measurements. CVs were recorded with a CHI 760E bipotentiostat using a standard three-electrode cell configuration that consisted of 1) a glassy-carbon (ø = 3.0 mm) working electrode, 2) a Pt wire as counter electrode, and 3) an Ag wire as reference electrode. All electrochemical measurements were performed at RT in an M. Braun N2-filled glovebox. Dry acetonitrile containing 0.1 M n Bu4NPF6 was used as the electrolyte solution. The ferrocene/ferrocenium (Fc/Fc + ) redox couple was used as an internal standard for all measurements.
Single-crystal X-ray diffraction. For compounds 2-7 low temperature (100K or 200K) diffraction data were collected using a Bruker APEX II diffractometer coupled to a APEX II CCD detector with graphite monochromated Mo Kα (λ = 0.71073 Å) radiation. All diffractometer manipulations, including data collection, integration, and scaling were carried out using the Bruker APEXII software. S4 Absorption corrections were applied using SADABS. S5 Structures were solved by direct methods using SHELXS S6 and refined against F 2 on all data by full-matrix least squares with SHELXL-2014 S7 using established refinement techniques. S8 All non-hydrogen atoms were refined anisotropically. All hydrogen atoms were included into the model at geometrically calculated positions and refined using a riding model. The isotropic displacement parameters of all hydrogen atoms were fixed to 1.2 times the U value of the atoms they are linked to (1.5 times for methyl groups). All air-and moisture-sensitive manipulations were carried out using Yorkshire, UK) under vacuum. EPR spectra were recorded on a Bruker ELEXSYS 580 spectrometer operating at X-band frequencies (9.5 GHz) equipped with an EN4118X-MD4 resonator at temperature of 10 K. The temperature was controlled by Bruker FlexLine cryogen free VT system ER4118HV-CF5-H.
Experimental conditions were, 750 points, with microwave power of 2 mW, 0.1 mT modulation amplitude and 100 kHz modulation frequency. Sweep range was 150mT. Field-sweep echo-detected (FS-ED) EPR spectra were recorded using the two-pulse echo sequence ( p /2 -tpt -echo) where the echo intensity is measured as a function of the magnetic field. The microwave pulse lengths, p /2 and p were 10 and 20 ns respectively and the time interval between the pulses,t was 220 ns.

Computational details.
In the course of this work, we performed several different types of calculations. To allow for utmost clarity and reproducibility, we detail here the types of calculations performed, essential details about the calculation method and the software used for each, and the purpose. Calculations of types 1-4 were performed with ORCA S9 5.0.0 at the PBE0 S10 /def2-TZVP S1 D3 S12 BJ S13 level of theory. Calculations of type 5 were performed with ORCA 5.0.0 at the TPSSh S14 /TZVP S15 + CP(PPP) S16 level of theory, and isomer shifts were calculated using the parameters reported by Römelt. S17 Calculations of type 6 were performed with MultiWFN S18 3.7. Input template for each type of calculation, as well as for the visualization with PyMOL, are given below. In this section, we provide template input files for the calculations performed in this work. The keywords related to resource management are omitted for conciseness.
1. All computations were carried out with ORCA 5.0.0 All input files (except for calculations of Mössbauer parameters, which require an additional section following this line) ended with: *xyzfile 0 <M> <XYZ> Where <M> is the placeholder for spin and multiplicity and <XYZ> is the placeholder for the file path of the Cartesian coordinates in XMOL format. This line is omitted in the following input templates for conciseness.
2. All broken-symmetry calculations included a section as follows: %scf BrokenSym <m,n> end Where <m,n> are placeholders for the expected number of unpaired electrons on the two fragments (m > n).
3. Single-point calculations with the PBE0 functional, the def2-TZVP basis set, and Grimme's D3 correction with Becke-Johnson damping were calculated to using the following input lines: ! UKS PBE0 D3BJ def2-TZVP def2/J DefGrid2 4. Constrained geometry optimizations were performed with the following input lines, reading in the wavefunction file from the appropriate single-point calculation: ! UKS PBE0 D3BJ def2-TZVP def2/J DefGrid2 TightOpt MOread %moinp "<GBW>" %geom optimizeHydrogens true end where <GBW> is the placeholder for the wavefunction file obtained from the single-point calculation for the appropriate solution.
5. Full optimization of the structures was performed with the following input, reading in the wavefunction file from the constrained-optimization calculation: ! UKS PBE0 D3BJ def2-TZVP def2/J DefGrid2 TightOpt MOread %moinp "<GBW>" 6. Mössbauer parameters were calculated for the constrained-optimized structures, reading in the wavefunction file from the constrained-optimization calculation. The TPSSh functional and the TZVP basis set were used, with the following input:

Multiwfn calculations
The Fe oxidation state was calculated with the localized orbital bonding analysis approach (LOBA). S20 The localized orbitals were read in as a .molden file generated with orca_2mkl from the ORCA .loc wavefunction file. The localized orbitals were then internally relocalized by Multiwfn using the Pipek-Mezey localizationS 21 with Löwdin populations. S22 The threshold for LOBA was set to 50%. This calculation was performed with the following sequence of Multiwfn commands:

Visualization
The cube files containing the volumetric orbital and spin-density data were generated with orca_plot. The molecular orbitals were visualized with Pymol version 1.74 S23 using the following input file: Where NAME is a placeholder for the name of the complex and NAME_ORB is a placeholder for the specific molecular orbital.
The spin-density was visualized with the following input: # Lighting set ambient, 0.2 set ray_shadows, 1 set spec_reflect, 0.6 set spec_power, 600 set spec_count, 3 set shininess, 70 set reflect, 0.5 # General visuals set ray_trace_mode, 1 set ray_texture, 2 set antialias, 3 set fog, 1 set fog_start, 0.4 # Orientation turn x, -60 turn y, 5 turn x, 5 turn z, 5 turn y, 5 zoom center, 6 S13 Note: We note that the bond metrics calculated via broken symmetry DFT calculations are in good agreement with bond metrics calculated with our previously used methodology (prior to revision) in which geometry optimizations were carried out using the Gaussian 16 program suite. S24 The PBE0 S10 density functional theory (DFT) exchange correlation functional was used in conjunction with the SDD basis setrelativistic effective core potential (RECP) combination. PBE0 has been found to be more accurate for equilibrium properties of transition metal complexes. S25 The SDD basis set is a combination of the Stuttgart-Dresden basis set-RECP combination on Co and the Huzinaga-Dunning double-zeta basis set on all other elements. S26 Equilibrium geometries were verified to have all real harmonic frequencies. The bond metrics obtained from these calculations can be found in Table S8.

Synthesis of [Fe(BDI)(OTf)] (3).
In the glovebox, to a solution of 7 (150 mg, 0.17 mmol) in MeCN (3 mL) was added freshly prepared 1% w/w Na(Hg) amalgam (475.0 mg, 0.20 mmol Na). The reaction mixture turned to green immediately and was stirred for another hour at room temperature. Then the solution was decanted to remove the remaining mercury and the solvent was evaporated under reduced pressure. The          S31 Tables   Table S1. Crystal and refinement data for iron complexes 2-4     Table S5. Outcomes of single-point calculations of alternative electronic structure solutions attempted for complexes 2-5. The outcomes were identified by summing up the α and β Löwdin spin densities, respectively (see Table S6). 3,2 2,1 0.00 --2 4,3 2,1 1.08 --3 2,1 2,1 0.00 0.00 4 doublet 2,1 0.00 --5 quartet 3,0 7.97 7.53 a This optimization of the structure with this electronic configuration resulted in dissociation of the ACN group (see xyz file), indicating that the solution is unstable for complex 4. Therefore, this energy should not be construed as representative of complex 4.