Enhanced Fe-Centered Redox Flexibility in Fe–Ti Heterobimetallic Complexes

Previously, we reported the synthesis of Ti[N(o-(NCH2P(iPr)2)C6H4)3] and the Fe–Ti complex, FeTi[N(o-(NCH2P(iPr)2)C6H4)3], abbreviated as TiL (1), and FeTiL (2), respectively. Herein, we describe the synthesis and characterization of the complete redox families of the monometallic Ti and Fe–Ti compounds. Cyclic voltammetry studies on FeTiL reveal both reduction and oxidation processes at −2.16 and −1.36 V (versus Fc/Fc+), respectively. Two isostructural redox members, [FeTiL]+ and [FeTiL]− (2ox and 2red, respectively) were synthesized and characterized, along with BrFeTiL (2-Br) and the monometallic [TiL]+ complex (1ox). The solid-state structures of the [FeTiL]+/0/– series feature short metal–metal bonds, ranging from 1.94–2.38 Å, which are all shorter than the sum of the Ti and Fe single-bond metallic radii (cf. 2.49 Å). To elucidate the bonding and electronic structures, the complexes were characterized with a host of spectroscopic methods, including NMR, EPR, and 57Fe Mössbauer, as well as Ti and Fe K-edge X-ray absorption spectroscopy (XAS). These studies, along with hybrid density functional theory (DFT) and time-dependent DFT calculations, suggest that the redox processes in the isostructural [FeTiL]+,0,– series are primarily Fe-based and that the polarized Fe–Ti π-bonds play a role in delocalizing some of the additional electron density from Fe to Ti (net 13%).


SI
Supplementary X-ray Crystallographic Data Table S1. Full structure metrics for TiL and FeTiL complexes S12 Table S2. Ti-Fe bond lengths in other crystallographically characterized FeTi complexes S12 Figure S13. Plot of the average Ti-Neq bond distance versus total M-M d-count S13 Figure S14. Plot of the average Fe-P bond distance versus total M-M d-count S13 EPR Spectroscopy Figure S15. X-band EPR spectrum of FeTiL, 2 S14 Discussion of Physical origin of the EPR g-tensor S15 Figure S16. Qualitative depiction of spin-orbit coupling effects S15 Figure S17. X-band EPR spectrum of TiL, 1 S16 Figure S18. UV-Vis-NIR spectra of 1, 2-Br, 2 ox [BAr F ], 2, and 2 red S17 Table S3. UV-Vis-NIR electronic absorbance data for 1, 2-Br, 2 ox [BAr F ], 2, and 2 red S18 Molecular Orbital Analysis (DFT) Figure S19. Full molecular orbital compositional analyses for 2 ox , 2, and 2 red (B3LYP/def2-TZVP) S19 Table S4. Full molecular orbital compositional analyses for 2 ox , 2, and 2 red (B3LYP/def2-TZVP) S20 Figure S20. Full molecular orbital compositional analyses for 2 ox , 2, and 2 red (B3LYP/def2-TZVP(-f)-ZORA; pertaining to XAS) S21 Table S5. Full molecular orbital compositional analyses for 2 ox , 2, and 2 red (B3LYP/def2-TZVP(-f)-ZORA; pertaining to XAS) S22 Table S6. Comparison of X-ray and calculated geometrical parameters for FeTiL complexes S23 Discussion of Ti K-edge XAS of 1 and 1 ox reference complexes S32 Figure S31. Ti K-edge spectra of 1 and 1 ox S32 Figure S32. Ti K-edge spectra of 2, 2 red , 2-Br, and 2 ox overlaid with 1 and 1 ox S33 Figure S33. Ti K-edge spectra (TD-DFT calculated) of 2, 2 red , 2-Br, and 2 ox overlaid with 1 and 1 ox S33 Figure S34. Comparison of collected Ti K-edge spectrum to B3LYP calculated spectra for 1 S34 Figure S35. Comparison of collected Ti K-edge spectrum to B3LYP calculated spectra for 1 ox S34 Figure S36. Comparison of collected Ti K-edge spectrum to B3LYP calculated spectra for 2 ox S35 Figure S37. Comparison of collected Ti K-edge spectrum to B3LYP calculated spectra for 2 S35 Figure S38. Comparison of collected Ti K-edge spectrum to B3LYP calculated spectra for 2 red S36 Figure S39. Comparison of collected Ti K-edge spectrum to B3LYP calculated spectra for 2-Br S36 Figure S40. Comparison of collected Fe K-edge spectrum to B3LYP calculated spectra for 2 ox S37 Figure S41. Comparison of collected Fe K-edge spectrum to B3LYP calculated spectra for 2 S37 S4 Figure S42. Comparison of collected Fe K-edge spectrum to B3LYP calculated spectra for 2 red S38 Figure S43. Comparison of collected Fe K-edge spectrum to B3LYP calculated spectra for 2-Br S38 Table S9. Experimental Fe and Ti K-pre-edge peak energies and assignments for FeTiL complexes S39 Figure S44. B3LYP calculated transitions and accepting molecular orbitals for complex 2 ox in the Fe K-edge XAS in the XANES region S40 Figure S45. B3LYP calculated transitions and accepting molecular orbitals for complex 2 ox in the Ti K-edge XAS in the XANES region S40 Figure S46. B3LYP calculated transitions and accepting molecular orbitals for complex 2 in the Fe K-edge XAS in the XANES region S41 Figure S47. B3LYP calculated transitions and accepting molecular orbitals for complex 2 in the Ti K-edge XAS in the XANES region S41 Figure S48. B3LYP calculated transitions and accepting molecular orbitals for complex 2 red in the Fe K-edge XAS in the XANES region S42    shows the CV that is first scanned cathodically, while the right shows the CV that is first scanned anodically. Hence, the electrochemical event at Epc = −2.00 V is related to the oxidation of 2.

Discussion of Electrochemical Characteristics of 2.
The low intensity electrochemical event, which appears at ca. −2.0 V only after scanning anodically, is consistent with an ECE process. In this process, the electrochemically generated [FeTiL] + species undergoes a chemical transformation into another species, which is subsequently reduced at a more negative potential. The peak current ratio (ipc/ipa), however, remains greater than 0.90 at all scan rates examined (from 50−1000 mV/s), which suggests the kinetics for this chemical transformation are slow with respect to the time required to complete the voltage sweep in the CV experiment.         It should be noted that these simulations are not unique, and that the best description of this EPR spectrum probably involves perturbations of both the g matrix as well as the 31 P hyperfine coupling values.

S15
Additionally, the broad, unresolved signal at gz was simulated using both hyperfine coupling and line widths, and hence, Azz parameters are also not unique.

Physical origin of the EPR g-tensor
The competition of the Jahn-Teller distortion with the spin-orbit coupling effect is responsible for the easyaxis of magnetization found in the experimental EPR g-values of 2, g= (2.27, 2.05, 2.03). In addition, the lowest-energy excited state of 2 must be the one promoting the -electron from one of the doubly occupied component of the 2e set to the other, which introduces the largest positive g-shift along the Fe-Ti bond. As elaborated elsewhere 6 , an excitation from a doubly occupied orbital to a singly occupied orbital (SOMO) gives rise to a positive g-shift, whereas a negative g-shift is produced by an excitation from a SOMO to a virtual orbital. Furthermore, the magnitude of the g-shift is inversely proportional to the excitation energy.
A qualitative explanation of this phenomenon can be found in ligand-field theory. In an ideal C3 symmetry, the dxy and dx2-y2 orbitals couple through spin-orbit coupling to recover the orbital momentum of the d±2 orbitals. The population of three electrons in two of these orbitals leads to the apparition of a neat total orbital momentum along the C3 axis (ML =±2). The spin-orbit coupling effect splits the magnetic states into two Kramer's doublets, according to the relative orientation of their spin and orbital momenta, as shown in Figure S13. The large orbital momentum of the ground doublet adds to the spin momentum, hereby increasing the effective gz value up to a theoretical limit of 6 in the absence of covalence effect. The Jahn-Teller distortion competes with the spin-orbit coupling by lifting the degeneracy of the dxy and dx2-y2 orbitals, partially quenching the orbital momentum along the C3 axis. The slight shift of the gz value above the isotropic value is therefore due to the residual orbital momentum originating from the effective C3 symmetry.    Figure S19. DFT-predicted electronic structures of 2 ox (left), 2 (middle) and 2 red (right) at the B3LYP/def2-TZVP level of theory. The CP(PPP) basis set was used on Fe for all calculations. Note that these single-point calculations were performed on structures that were obtained from crystallographic coordinates where only the H atom positions were optimized.     Table S4. We note that these virtual MOs have more ligand character, but the %Ti/%Fe ratio is still similar.      Figure S27. Core orbitals taken in account in the calculation of isomer shift and quadrupole splitting contributions    Figure S34. Comparison of experimentally collected Ti K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 1 (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). Figure S35. Comparison of experimentally collected Ti K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 1 ox (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). S35 Figure S36. Comparison of experimentally collected Ti K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 2 ox (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). Figure S37. Comparison of experimentally collected Ti K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 2 (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). S36 Figure S38. Comparison of experimentally collected Ti K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 2 red (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). Figure S39. Comparison of experimentally collected Ti K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 2-Br (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). S37 Figure S40. Comparison of experimentally collected Fe K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 2 ox (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). Figure S41. Comparison of experimentally collected Fe K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 2 (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates).

S20
S38 Figure S42. Comparison of experimentally collected Fe K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 2 red (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). Figure S43. Comparison of experimentally collected Fe K-edge spectrum (solid black) to B3LYP calculated spectra (solid grey a and dotted black b ) for 2-Br (a=calculated using crystallographic coordinates; b=calculated using geometry optimized coordinates). Ti 1s → Fe-Ti π*, (Ti 3dx2-y2 + Ti 3dxy); Fe-Ti σ*; aromatic π* (ligand) a = the low-lying vacant molecular orbital that has Ti 3dz2 contribution is mainly dominated by Fe 4p character. Figure S44. B3LYP calculated transitions and accepting molecular orbitals for complex 2 ox in case of      a In case of complex 2 ox , the 1 st pre-edge peak in the Ti K-edge XAS arises mainly due to the transitions to 4e orbitals while for other two complexes, peak arises due to transitions to both 3e and 4e orbitals. (see figure S45, S47, S49)

S40
S45 Figure S52. Experimental and calculated metal 1s orbital energies as well as the transition energies (which correspond to intense metal K-pre-edge XAS) plotted against total d-count for 2 ox , 2 and 2 red .