Unprecedented Five-coordinate Iron(IV) Imides Generate Divergent Spin States Based on Imide R-groups

Abstract

Three five-coordinate iron(IV) imide complexes have been synthesized and characterized. These highly novel structures have disparate spin states on the iron as a function of the R-group attached to the imide with alkyls leading to low spin diamagnetic (S = 0) complexes and an aryl leading to an intermediate spin (S = 1) complex. The different spin states lead to significant differences in their bonding about the iron center as well as their spectroscopy. Mössbauer spectra confirm that all three imide complexes are in the iron(IV) oxidation state. The combination of diamagnetism and ¹⁵N labeling allowed for the first ¹⁵N NMR resonance recorded on an iron imide. Multi-reference calculations corroborate the experimental structural findings and suggest how the bonding is distinctly different on the imide ligand between the two spin states.

graphical abstract

The first five-coordinate monomeric iron imide complexes are reported. They display different spin states as a function of imide moiety.

Since the synthesis of the first monomeric iron imide, [PhBP₃]Fe(N-p-tolyl), by Peters in 2003,^[1] iron imides have been studied extensively due to their purported roles in oxidative group transfer catalysis, such as aziridination^[2–4] and amination,^[5–7] as well as for structural models of metalloenzymes.^[8] As such, a relatively varied set of iron imides have been reported to date with oxidation states on the iron ranging from iron(II) to iron(V) (Figure 1).^{[1, 9–28]} Despite the diversity of oxidation states isolated thus far, it is striking that these species are almost exclusively low coordinate (four or less) about the iron center.

Figure 1.

Examples of crystallographically characterized monomeric iron imide complexes in different coordination geometries. L’s are additional auxiliary ligands. Known oxidation states for iron shown in each geometry.

This outcome is particularly surprising because the imide’s isoelectronic ligand, the oxo, has wide application in heme and non-heme iron(IV) chemistry, where five and six coordinate iron oxo complexes are well known.^{[29, 30]} Yet, the analogous five- and six-coordinate iron imides have scarcely been reported. Indeed, no structurally characterized iron imides have been reported with porphyrins as the auxiliary ligand, which would be the imide’s heme analog.^[1] A single report from Que and coworkers showcases a six-coordinate iron imide complex with a very short half-life, but no structural characterization by X-ray diffraction was collected.^[31] In addition to Que’s example, a couple of other notable exceptions stand out in potential higher coordination environments. Thomas has described a dimeric iron imide complex with an iron-iron bond,^[32] while Peters has reported an iron imide with a hemi-labile borane trans to the imide, which he describes as pseudotetrahedral while in the imide conformation.^[33]

Given our continued exploration of iron catalyzed aziridination, we believed that an iron imide would be a key intermediate in the reaction cycle.^[34–36] We have recently reported the most “porphyrin-like” tetra-NHC macrocycle to date,^[37] and judged that its strong σ-donor strength^[38] combined with its rigidity compared to larger NHC macrocycles may be able to stabilize five-coordinate iron imide complexes in a square pyramidal geometry.

In this manuscript, we describe the synthesis of the first monomeric five-coordinate iron imide complexes. Notably, the complexes have disparate spin states as a function of the R-group attached to the imide moiety. Alkyl groups lead to low spin complexes while an aryl group leads to an intermediate spin complex. These differences in spin state are manifested in their respective X-ray structures, spectroscopy, and theoretical calculations.

The synthesis of the iron imide complexes proceeded facilely from the previously reported iron(II) complex [(^BMe₂,MeTC^H)Fe] (1).^[37] Addition of organic azide to 1 resulted in immediate bubbling and a color change (Scheme 1); addition of tert-butyl azide and adamantyl azide to 1 yielded dark red iron imide complexes, [(^BMe₂,MeTC^H)Fe(N^tBu)] (2) and [(^BMe₂,MeTC^H)Fe(NAd)] (3), while reaction with DiPP-azide resulted in a green product, [(^BMe₂,MeTC^H)Fe(NDiPP)] (4). All three imides can be isolated in high yield, are soluble in polar organic solvents, and are thermally stable at room temperature, but they all exhibit extreme water sensitivity. Specifically, addition of water leads to the formation of free amine plus the reported bridging oxo species [((^BMe₂,MeTC^H)Fe)₂O] (5) (See SI).^[37] Despite complexes similar to 1 demonstrating catalytic aziridination, 1 is ineffective for this reaction and, indeed, the imide ligands on 2-4 do not transfer to yield the aziridines upon addition of alkenes even up to 80 °C.^{[36, 38]}

Scheme 1.

Synthesis of Fe(IV) imide complexes.

Since no square pyramidal iron imides have been reported, the structural characterization of 2-4 is of particular significance. The X-ray crystal structure of 2 revealed a highly distorted square pyramidal structure where the imide nitrogen is tilted towards one side of the macrocycle (Figure 2). The Fe-N9 bond distance of 1.6552(12) Å and the N9-C19 distance of 1.4621(19) Å are within the typical range observed for iron(IV) imides, ^{[12, 15, 18, 21, 22, 26, 28]} yet the imide moiety is quite bent, 150.07(10)°, compared with the majority of iron imides.^{[9, 26]} The combination of short Fe-N bond distance with a bent Fe-N-C angle is most similar to a cis-divacant octahedral imide reported by Caulton.^[18] The desymmetrization of the macrocycle in regards to the imide ligand is observed in both of the Fe-C bond distances, where the Fe-C1 and Fe-C2 bond distances are shorter than the Fe-C3 and Fe-C4 lengths (~0.05 Å), as well as in the C-Fe-N bond angles, where the C3-Fe-N9 and C4-Fe-N9 are approximately 20° wider than the C1-Fe-N9 and C2-Fe-N9 angles. The X-ray crystal structure of 3 is very similar to 2 (See SI).

Figure 2.

X-ray crystal structure of [(^BMe₂,MeTC^H)Fe(N^tBu)], 2, (top) and [(^BMe₂,MeTC^H)Fe(NDiPP)], 4, (bottom). Green, blue, gray, and olive ellipsoids (50% probability) represent Fe, N, C, and B atoms, respectively. Solvent molecules and H atoms are omitted for clarity.

In contrast to 2 and 3, the X-ray structure of 4 shows a highly symmetric imide ligand situated between the four carbene carbons. The Fe-N9 bond distance of 1.7300(12) Å is significantly longer than that observed in 2 or 3, as is expected with the increase in spin state, and this is one of the longest reported iron imide bond distances. Indeed, the only known examples of longer Fe-N imide bonds are by the Betley group and show significant imidyl character.^{[14, 15]} The Fe-N9-C19 bond angle is slightly bent at 163.04(11)° and is within the typical range for iron(IV) imides.^{[18, 21, 22, 28]} Unlike 2 and 3, the imide ligand for 4 sits exactly between the four carbene ligands.

Spectroscopic characterization demonstrated that there are clear differences between the red (2–3) and green (4) imide complexes. ¹H NMR for 2 shows a diamagnetic complex with diastereotopic splitting of the methylene protons (6.11 and 5.08 ppm) and the methyl protons attached to the boron atoms (0.50 and −0.73 ppm), which is consistent with complexes with similar NHC macrocycles (Figure S5).^[38] The ¹³C NMR resonances for the NHC carbons are observed at 176.20 ppm, which shows fluxionality in solution (i.e. C_2v symmetry), while the tert-butyl carbons are at 66.91 and 25.54 ppm (Figure S6). The NMR spectra of 3 are in close agreement with 2 (Figures S10–11). In contrast, complex 4 shows a paramagnetic ¹H NMR spectrum with peaks ranging from 18.5 to −30.2 ppm that is consistent with a C_2v symmetric species. Finally, Evans method measurements are consistent with a spin state of S = 1 for 4.

Given that NMR spectra for 2, 3, and 4 showed different spin states, we collected zero-field ⁵⁷Fe Mössbauer spectra to ascertain the oxidation states of the iron centers. Complex 2 showed a quadrupole doublet with an isomer shift, δ, of -0.18 mm/s and a quadrupole splitting, ΔE_Q, value of 0.97 mm/s (Figure 3, left, blue line). Despite repeated efforts, we could not prevent the formation of the bridging oxo, 5 (green line), during measurements, which occurs when 3 reacts with trace water (See SI). The isomer shift is consistent with the other measured iron(IV) imides^[22] and is particularly close to the value (δ = −0.09) reported by Caulton.^[18] The Mössbauer spectrum for 3 was nearly identical to 2 (Figure S19). Complex 4 gave a similar isomer shift, δ = −0.11 mm/s, but had a notably larger quadrupole splitting of ΔE_Q = 2.67 mm/s (Figure 3, right). The isomer shifts of these complexes were observed to be more negative than the isomer shift observed for an isostructural iron(III) complex [(^BMe₂,MeTC^H)FeBr], (δ = 0.07 mm/s, Figure S22), consistent with more oxidized iron centers present in the imide complexes.^[37] These measurements are consistent with assigning the oxidation states of iron in complexes 2, 3, and 4 as iron(IV) centers.

Figure 3.

Zero-field ⁵⁷Fe Mössbauer spectra [δ, |ΔE_Q|, Γ (mm/s)] of compounds 2 and 4 measured at 90 K in frozen benzene.

Since very few low spin iron(IV) imides have been synthesized in any geometry,^{[18, 22]} we were particularly interested in the effect of isotopic labeling on the imide nitrogen. To this end, we prepared ^tBu¹⁵N₃ which has a single ¹⁵N label either at the α or γ position (in a 1:1 ratio).^[11] Reaction of ^tBu¹⁵N₃ with 1 gave 50% ¹⁵N enriched imide complex, [(^BMe₂,MeTC^H)Fe(¹⁵N^tBu)] (2-¹⁵N) in 83% yield. ¹H-¹⁵N HMBC 2D NMR of 2-¹⁵N allowed us to assign the ¹⁵N resonance at −413 ppm (versus neat nitromethane at 0 ppm) to the imide nitrogen. To our knowledge, there have been no previous measurements of ¹⁵N NMR for iron imides, but the ¹⁵N resonance of a pseudotetrahedral iron(IV) nitride by Peters was reported at 572 ppm (versus a nitromethane reference).^[39] In addition, the coupling of the ¹⁵N imide can be observed in the ¹³C NMR in both the NHC carbons (J = 2.6 Hz) and quaternary carbon of the t-butyl group (J = 7.2 Hz).

Theoretical calculations for these unique iron(IV) imides are particularly valuable since different R-groups on the imide ligand lead to different spin states on the iron^[20] and the imide ligands on 2 and 3 are highly tilted. Multi-configurational calculations with the complete active space self-consistent field (CASSCF) method were performed on 2-4 optimized at the density functional theory level (DFT) with potential spin states of S = 0, S = 1 and S = 2. The singlet state was found to be the lowest energy state for complexes 2 and 3, for which the calculated bond distances are in good agreement with the experimentally determined values (See SI). Analyses of the molecular orbitals for 2 and 3 show that the d_xy and d_xz are both fully populated non-bonding orbitals and the d_x²-_y² forms σ-bonds to the NHCs, which leaves the d_z² and d_yz orbitals to form a pair of Fe-N bonds where the former case has an unusual overlap due to the canting of the imide away from the z-axis (Figure 4, top). Taken together, there are almost two fully populated Fe-N bonds with a net effective Fe-N bond order of 1.72, which is consistent with the experimentally observed Fe-N bond distance and bent Fe-N-C angle.

Figure 4.

Top left: Fe-N bonding molecular orbitals in the CASSCF(20,13) for 2 showing Fe-N bonding with occupations of 1.88 and 1.85, respectively. Top right (in box): Spin density of 4 (triplet) from DFT shows almost all unpaired spin present on iron atom. Bottom: Fe-N bonding molecular orbitals in the CASSCF(20,13) for 4 showing Fe-N with occupations of 1.99, 1.80, and 1.76 respectively.

Calculations for 4 showed that the triplet was the lowest energy state; however, 4 has significant multi-configuration character (<S²> = 2.46, rather than 2.00). Analysis of the molecular orbitals showed that the d_xy and d_x²-_y² are similar to 2. For 4, the d_z², d_xz, and d_yz all exhibit bonding character to the imide nitrogen. The π-bonds are more symmetric as expected for square pyramidal complexes with metal-ligand multiple bonds. Notably, the anti-bonding orbitals also have high occupancy, which accounts for the more linear Fe-N bond angle of 4 relative to 2 (Figure S29). From the state-average CASSCF, a bond order of 1.54 was computed for 4, which is consistent with the highly elongated Fe-N bond observed in the solid state structure. Finally, the calculated spin density for 4 reveals almost all the unpaired spin is present on the iron atom, giving very little imidyl radical character.^{[15, 28]}

In conclusion, we have synthesized the first monomeric five-coordinate iron(IV) imide complexes. These three complexes exhibit two disparate spins on iron as a function of the imide ligand R-group, where alkyl groups lead to low spin complexes while an aryl DiPP group results in intermediate spin. The different spin states lead to numerous distinctions between the complexes in their X-ray structures, ¹H NMR resonances, and Mössbauer spectra. In addition, isotopic labeling allowed for the determination of the first ¹⁵N NMR resonance for an iron imide moiety. Multi-configurational wave function theory calculations are consistent with experimental findings by predicting the same spin state for each imide complex as well as the relative Fe-N bond order. The tipping of the imide ligand from the axial position appears to increase the bond order and allow for two non-bonding d-orbitals to be fully populated. The increased steric bulk of the DiPP ligand may preclude this unusual tipping, and thus lead to a more symmetric imide complex of intermediate spin. These unique five-coordinate iron imides provide insight into how subtle changes in electronic structure and steric interactions of the imide moiety transform the bonding between the metal and ligand, which may assist in developing the next generation of iron-based catalysts for oxidative NR group transfer reactions.