Conversion of Fe-NH₂ to Fe-N₂ with release of NH₃

Abstract

Tris(phosphine)borane ligated Fe(I) centers featuring N₂H₄, NH₃, NH₂, and OH ligands are described. Conversion of Fe-NH₂ to Fe-NH₃⁺ by addition of acid, and subsequent reductive release of NH₃ to generate Fe-N₂, is demonstrated. This sequence models the final steps of proposed Fe-mediated nitrogen fixation pathways. The five-coordinate trigonal bipyramidal complexes described are unusual in that they adopt S = 3/2 ground states and are prepared from a four-coordinate, S = 3/2 trigonal pyramidal precursor.

Due the structural and mechanistic complexity of biological nitrogen fixation¹ a variety of mechanisms have been proposed that invoke either Mo or Fe as the likely active site for N₂ binding and reduction. Fe-NH₂ is an intermediate common to both limiting mechanisms (i.e., distal vs. alternating) being considered for Fe-mediated N₂ fixation scenarios at the FeMo-cofactor.^2,3 Such a species could form via reductive protonation of the nitride intermediate of a distal scheme (i.e., Fe(N) → Fe(NH) → Fe(NH₂) → Fe(NH₃)), or by reductive protonation of a hydrazine intermediate of an alternating scheme (i.e., Fe(NH₂-NH₂) → Fe(NH₂) + NH₃). In the latter context, detection of an EPR active Fe-NH₂ or possibly Fe-NH₃ common intermediate has been proposed under reducing conditions at the FeMo-cofactor from substrates including N₂, N₂H₄, and MeN=NH.^3a

One key to realizing a catalytic cycle in either limiting scenario concerns the regeneration of Fe-N₂ from Fe-NH₂ with concurrent release of NH₃.⁴ While there have been recent synthetic reports demonstrating NH₃ generation from Fe(N) nitride model complexes, these studies have not provided information about the plausible downstream Fe(NH_x) (X = 1, 2, 3) intermediates en route to NH₃ release, nor have these systems illustrated the feasibility of regeneration of Fe-N₂.⁵ Herein we describe a terminal, S = 3/2 Fe-NH₂ complex for which the stepwise conversion to Fe-NH₃, and then to Fe-N₂ along with concomitant release of NH₃, is demonstrated (eqns 1 and 2).

$$\text{Fe}-{\text{NH}}_2+{\mathrm{H}}^{+}\to \text{Fe}-{\text{NH}}_{3^{+}}$$ (1)

$$\text{Fe}-{\text{NH}}_{3^{+}}+{\mathrm{e}}^{-}+{\mathrm{N}}_2\to \text{Fe}-{\mathrm{N}}_2+{\text{NH}}_3$$ (2)

Addition of methyllithium to (TPB)FeBr⁶ affords the corresponding methyl complex (TPB)FeMe (1) in high yield (Scheme 1). Protonation of 1 by HBAr^F₄·2Et₂O (BAr^F_4⁻ = B(3,5-C₆H₃(CF₃)₂)_4⁻) in a cold ethereal solution releases methane to yield [(TPB)Fe][BAr^F₄] (2) which serves as a useful synthon with a vacant coordination site.

SCHEME 1.

XRD data were obtained for 1 and 2 (Figure 1). The geometry of 1 is pseudo trigonal bipyramidal about Fe with an Fe-C bond length of 2.083(10) Å and an Fe-B bond length of 2.522 (2) Å. In the solid state 2 possesses a four-coordinate distorted trigonal pyramidal geometry with no close contacts in the apical site trans to boron, making this complex coordinatively unsaturated. Additionally, there is one wide P-Fe-P angle of 136°. The origin of this large angle is not clear, but a possible explanation is increased back-bonding from a relatively electron rich Fe center into the phosphine ligands that would arise from this distortion.

Figure 1.

X-Ray Diffraction (XRD) structures of complexes 1 (A) and 2 (B) with hydrogen atoms and counterion (for B) omitted for clarity. See Table 1 for selected bond lengths and angles.

The Fe-B distance in 2 (2.217(2) Å) is markedly shorter than that in (TPB)FeBr (2.459(5) Å), which is noteworthy because one might expect the loss of an anionic σ-donor ligand to reduce the Lewis basicity of the metal and thus weaken the Fe-B bond. For example, the Au-B distance in (TPB)AuCl (2.318 Å) lengthens upon chloride abstraction to 2.448 Å in [(TPB)Au]⁺.⁷ To explain this difference we note that the boron center in four-coordinate 2 is less pyramidalized (Σ(C−B−C) = 347.3°) than that in five-coordinate (TPB)FeBr (Σ(C−B−C) = 341.2°), pointing to a weak interaction despite the short distance. These observations suggest that the geometry of 2 might be best understood as derived from a planar three-coordinate Fe(I) center distorted towards a T-shaped geometry,⁸ the unusually short Fe-B distance being due largely to the constraints imposed by the ligand cage structure. This interpretation is consistent with a computational model study: the DFT (B3LYP/6-31G(d)) optimized geometry of the hypothetical complex [(Me₂PhP)₃Fe]⁺ (see SI for details) exhibits a planar geometry with P-Fe-P angles of 134.8°, 113.1°, and 111.7°, very close to those measured for [(TPB)Fe]⁺ (137.5°, 113.2°, 109.1°).

When considering the bonding of the (Fe-B)⁷ subunit of 2 to estimate the best oxidation state and valence assignment, two limiting scenarios present themselves: Fe(III)/B(I) and Fe(I)/B(III). The structural data and computations for 2 are suggestive of a weak Fe-B interaction and indicate that this species is better regarded as Fe(I)/B(III) rather than Fe(III)/B(I). Calculations indicate that a small amount of spin density resides on the Batom of 2 (SI) and suggest some contribution from an Fe(II)/B(II) resonance form may also be relevant. The rest of the complexes 3–6 presented herein possess significantly longer, and presumably weaker, Fe-B interactions (vide infra) and are hence also better classified as Fe(I) species. Additional spectroscopic studies (e.g., XAS and Mössbauer) will help to better map the Fe-B bonding interaction across the variable Fe-B distances and also the spin states of the complexes. These studies would thereby help to determine the value and limitation of classically derived oxidation/valence assignments for boratranes of these types.⁹

Solutions of 2 are orange in Et₂O and pale yellow-green in THF. Titration of THF into an ethereal solution of 2 results in a distinct change in the UV-vis spectrum consistent with weak THF binding. Addition of an excess of N₂H₄ to an ethereal solution of 2 results in a slight lightening of the orange color of the solution to afford [(TPB)Fe(N₂H₄)][BAr^F₄] (3) in 89% yield. Complex 3 shows a paramagnetically shifted ¹H NMR spectrum indicative of an S = 3/2 Fe center that is corroborated by a room temperature solution magnetic moment, μ_eff, of 3.5 μ_B. Crystals of 3 were obtained and XRD analysis (Figure 2A) indicates a distorted trigonal bipyramidal geometry. The Fe-N distance of 2.205(2) Å is unusually long (2.14 Å is the average quaternary N-Fe distance in the Cambridge Structural Database) reflecting its unusual quartet spin state.

Figure 2.

XRD structures of the cores of complexes 3 (A), 4 (B), and 5 (C). See Table 1 for selected distances and angles.

Complex 3 is stable to vacuum, but solutions decompose cleanly at room temperature over hours to form the cationic ammonia complex [(TPB)Fe(NH₃)][BAr^F₄], 4, which was assigned by comparison of its ¹H NMR spectrum with an independently prepared sample formed by the addition of NH₃ to the cation 2. Analysis of additional degradation products shows only NH₃ and trace H₂ (SI). The assignment of 4 as an ammonia adduct was confirmed by XRD analysis (Figure 2B). Like 3, complex 4 shows a long Fe-N distance of 2.280(3) Å in the solid state. The complexes 2, 3 and 4 are unusual by virtue of their S = 3/2 spin states and underscore the utility of local 3-fold symmetry with respect to stabilizing high spin states at iron, even in the presence of strong-field phosphine ligands.

Addition of NaNH₂ to the cation 2 affords the terminal amide, (TPB)Fe-NH₂ (5) in ca. 85% non-isolated yield by ¹H NMR integration. The XRD structure of 5 (Figure 2C) shows an overall geometry similar to that observed in 1, 3, and 4. Of interest is the short Fe-N distance of 1.918(3) Å by comparison to 4 (2.280(3) Å). The amide hydrogens were located in the difference map and indicate a nearly planar geometry about N (with the sum of the angles around N being 355°).

While the XRD data set of 5 is of high quality, we were concerned about the difficulty in distinguishing an Fe-NH₂ group from a potentially disordered Fe-OH moiety. We therefore independently characterized the hydroxo complex, (TPB)Fe-OH (6) (Scheme 2), which possesses a geometry similar to that observed in 5 with an Fe-B distance of 2.4438(9) Å and an Fe-O distance of 1.8916(7) Å. Despite the structural similarity between 5 and 6, different spectral signatures in both their ¹H NMR and EPR (Figure 3) spectra allow for facile distinction between them. Like 2, 3, and 4, both 5 and 6 are S = 3/2.

SCHEME 2.

Figure 3.

X-Band EPR spectra for complexes 1–6. Conditions for 1: Toluene, 8 K; 2: 2:1 Toluene:Et₂O, 10 K 3: 2-MeTHF, 10 K; 4: 2-MeTHF, 10 K; 5: 2-MeTHF, 10 K; 6: Toluene, 10 K.

Low-temperature EPR data (Figure 3) have been obtained on complexes 1–6. All complexes show features shifted to large g-values consistent with quartet Fe species.¹⁰ This assignment is verified by the solution magnetic moments obtained for these complexes. Variable temperature solid-state SQUID magnetic data for complexes 2–5 (SI) also establish quartet spin state assignments and display no evidence for spin-crossover phenomena. These data show a drop in magnetic moment in the range 50–70 K for all compounds studied. We propose that this effect is due to a large zero-field splitting in these species, which is consistent with Fe centers in related geometries.¹¹ Simulations with zero-field splitting of 10–20 cm⁻¹ provide reasonable fits to the data.

Parent amide complexes of first row transition metals are rare.¹² Noteworthy precedent for related terminal M-NH₂ species includes two square planar nickel complexes^{12a, d} and one octahedral and diamagnetic iron complex, (dmpe)₂Fe(H)NH₂.^12e In addition to their different coordination numbers, geometries, and spin-states, (dmpe)₂Fe(H)(NH₂) and 5 show a distinct difference at the Fe-NH₂ subunit. Six-coordinate (dmpe)₂Fe(H)(NH₂) is an 18-electron species without π-donation from the amide ligand, which is pyramidalized as a result. By contrast, five-coordinate 5 accommodates π-bonding from the amide. This is borne out in its much shorter Fe-N distance (1.918(3) Å for 5 vs 2.068 Å for (dmpe)₂Fe(H)(NH₂)), and also its comparative planarity (the sum of the angles around N is 355° for 5 vs 325° for (dmpe)₂Fe(H)(NH₂)).

With the terminal amide 5 in hand we explored its suitability as a precursor to the previously reported N₂ complex (TPB)Fe(N₂) via release of NH₃ and hence explored reduction/protonation vs protonation/reduction sequences as a means of effecting overall H-atom transfer to the Fe-NH₂ unit. Attempts to carry out the one-electron reduction of 5 were not informative. For example, electrochemical studies of 5 in THF failed to show any reversible reduction waves, but the addition of harsh reductants (e.g., tBuLi) to 5 did show small amounts of (TPB)Fe(N₂) in the product profile. A more tractable conversion sequence utilized protonation followed by chemical reduction. Thus, the addition of HBAr^F₄·2Et₂O to 5 at low temperature (−35 °C) rapidly generates the cationic ammonia adduct 4. The conversion is quantitative as determined by ¹H NMR spectroscopy, and 4 can be isolated in ca. 90% yield from the solution. Subsequent exposure of 4 to one equiv of KC₈ under an atmosphere of N₂ releases NH₃ and generates the (TPB)FeN₂ complex in similarly high yield.

In summary, an unusual series of S = 3/2 iron complexes featuring terminally bonded N₂H₄, NH₃, NH₂, and OH functionalities has been thoroughly characterized. These complexes are supported by a tris(phosphine)borane ligand and are best described as Fe(I) species that feature weak Fe-B bonding, though other resonance contributions to the bonding scheme warrant additional consideration. The Fe-NH₂ species faithfully models the reductive replacement of the terminal NH₂ group by N₂ with concomitant release of NH₃, lending credence to such a pathway as mechanistically feasible in Fe-mediated N₂ reduction schemes. Because spectroscopic detection of a common Fe-NH₂ or Fe-NH₃ intermediate under reductive turnover of the FeMo-cofactor has been recently proposed,³ EPR active model complexes of the types described here should prove useful for comparative purposes.

Table 1.

Selected Metrics for Complexes 1–6

	Fe-X (Å)	Fe-B (Å)	Avg. Fe-P (Å)	Σ P-Fe-P	Σ C-B-C
1	2.083(10)	2.523(2)	2.40	339°	341°
2	-	2.217(2)	2.38	359°	347°
3	2.205(2)	2.392(2)	2.44	350°	339°
4	2.280(3)	2.433(3)	2.44	349°	341°
5	1.918(3)	2.449(4)	2.39	343°	339°
6	1.8916(7)	2.4438(9)	2.39	348°	337°