Fe–N₂/CO complexes that model a possible role for the interstitial C atom of FeMo-cofactor (FeMoco)

Significance

Biological nitrogen reduction is a fascinating transformation whose mechanism remains uncertain. Recently, an interstitial carbon has been identified within the FeMo-cofactor (FeMoco) of nitrogenase whose role is unknown and warrants model studies. In this report we disclose a series of five-coordinate Fe complexes bound to an ancillary ligand featuring a central C atom. This model system coordinates N₂ trans to the C atom, and displays unusual Fe–C bonding motifs that may shed light on a possible role of the interstitial carbon in FeMoco.

Abstract

We report here a series of four- and five-coordinate Fe model complexes that feature an axial tri(silyl)methyl ligand positioned trans to a substrate-binding site. This arrangement is used to crudely model a single-belt Fe site of the FeMo-cofactor that might bind N₂ at a position trans to the interstitial C atom. Reduction of a trigonal pyramidal Fe(I) complex leads to uptake of N₂ and subsequent functionalization furnishes an open-shell Fe–diazenido complex. A related series of five-coordinate Fe–CO complexes stable across three redox states is also described. Spectroscopic, crystallographic, and Density Functional Theory (DFT) studies of these complexes suggest that a decrease in the covalency of the Fe–C_alkyl interaction occurs upon reduction and substrate binding. This leads to unusually long Fe–C_alkyl bond distances that reflect an ionic Fe–C bond. The data presented are contextualized in support of a hypothesis wherein modulation of a belt Fe–C interaction in the FeMo-cofactor facilitates substrate binding and reduction.

MoFe-nitrogenase catalyzes the fascinating but poorly understood conversion of nitrogen to ammonia at its iron-molybdenum cofactor (FeMoco) (1, 2). The core of the FeMoco was originally thought to be vacant (3). Later work on Azotobacter vinelandii indicated the presence of a light interstitial atom coordinated to six central, so-called “belt Fe atoms” (4). Crystallographic and spectroscopic studies (5, 6), in addition to studies mapping the biosynthetic pathway of C-atom incorporation (7, 8), establish that carbon is the interstitial atom, as shown in Fig. 1.

Fig. 1.

A hypothetical N₂-binding event at a belt iron in FeMoco illustrating a proposed Fe–C elongation. The degree and positions of protonation are unknown under electron loading, but the inorganic sulfides are plausible candidate positions.

Although the site(s) of N₂ reduction remain(s) uncertain, a body of evidence that includes biochemical, spectroscopic, and computational studies on FeMoco point to a belt Fe center as a plausible candidate (2, 9–13). In a scenario in which N₂ binds terminally to one of the belt Fe centers, the N₂ ligand would initially be coordinated trans to the interstitial C atom (Fig. 1) (10). This hypothesis calls for model complexes that depict such an arrangement to explore factors that might govern substrate coordination and subsequent reduction. Structurally faithful models of the FeMoco that include an Fe₆C unit stabilized by sulfide or other sulfur-based ligands present a formidable synthetic challenge (14). Moreover, N₂ coordination to synthetic iron–sulfur clusters has yet to be established (15). Model complexes featuring a single Fe site with a C-atom anchor positioned trans to an N₂ binding site are unknown, but would provide a useful tool to evaluate how an Fe–C interaction might respond to N₂ binding and the Fe redox state. Model compounds of this type may facilitate the evaluation of theoretical (10) and spectroscopic studies (16) on FeMoco that suggest a single, flexible Fe–C interaction is observed under turnover conditions.

It is in this context that we have pursued mononuclear Fe complexes supported by tripodal, tetradentate ligands featuring three phosphine donor arms tethered to a tertiary alkyl anchor. Although a number of such ligands featuring a central C atom has been described (17–19) we reasoned that an alkyl ligand featuring only electropositive substituents adjacent to the C-atom anchor would provide a crude model of the interstitial carbide of the FeMoco and permit a high degree of ionic bonding to a single Fe–N₂ binding site. To achieve this goal, the C-atom anchor of the auxiliary ligand described is surrounded by three electropositive Si centers, in addition to the Fe site. This model system successfully coordinates N₂ trans to the C-atom anchor and shows how the local Fe geometry and the Fe–C interaction respond as a function of such binding. Using CO instead of N₂, this Fe system also presents an opportunity to study the Fe–C interaction as a function of the Fe redox state. As described below, unusually long Fe–C distances can be accessed that are consistent with a much higher degree of ionic character at the Fe–C interaction than would be anticipated for a prototypical Fe alkyl.

Results and Discussion

A suitable ligand framework for the present study, whose lithium salt was previously reported by Avent et al. (20), is (Ph₂PCH₂SiMe₂)₃CH. This ligand, denoted herein as (1) and depicted in Fig. 2, features a tri(silyl)methyl core tethered to three soft phosphines. Given that the interstitial carbide in FeMoco is surrounded by six electropositive Fe centers, we speculated that an Fe₅C subunit would stabilize partial negative charge on the C atom as the sixth Fe–C interaction elongates upon reduction/N₂ binding. Relative to methane, the three electropositive silyl substituents of the tri(silyl)methane subunit reduce the basicity of the central carbon (pK_a[(Me₃Si)₃CH]: 36.8 in THF) by ∼20 orders of magnitude (21, 22). We anticipated that the tempered basicity of the tri(silyl)methyl subunit would translate to a flexible Fe–C interaction.

Fig. 2.

Qualitative d-orbital splitting diagrams reflective of experimentally measured spin states for Fe–Cl complexes of and

Benzyl potassium was found to effect rapid and quantitative deprotonation of 1 as evidenced by ³¹P NMR spectroscopy. Subsequent salt metathesis with FeCl₂ (Fig. 3) afforded the Fe halide complex (2), as a pale yellow powder. Complex 2 exhibits broad ¹H NMR signatures from δ = −1 to 30 ppm and a solution magnetic moment of 4.9 μ_B (CD₂Cl₂, 21 °C), consistent with an S = 2 species. This high spin state is noteworthy given the strong ligand field strength that might be expected from a tris(phosphine)alkyl donor set and attests to the comparatively weak ligand field arising from the ligand. The structurally similar and isoelectronic five-coordinate Fe(II) chloride complexes supported by the tris(phosphine)silyl ligand (23, 24) display an intermediate S = 1 spin state as a consequence of the strong and highly covalent Fe–Si bond. Zero-field Mössbauer data (SI Appendix, Fig. S14) collected on 2 is consistent with its S = 2 assignment (25).

Fig. 3.

Synthesis of iron complexes supported by a tris(phosphino)alkyl tripod. .

The X-ray crystal structure of 2 exhibits a rigorously trigonal bipyramidal geometry about the Fe center as exemplified by a τ₅ value of 1.00 (SI Appendix), and confirms ligation of the tri(silyl)methyl carbon to Fe, albeit at an unusually long distance [2.263(2) Å]. Reported high-spin Fe(II)–C_alkyl distances range from 2.00 to 2.21 Å (26). The Fe–P distances are also quite long [2.5696(3) Å] as a consequence of the high-spin state. The Fe atom is displaced just 0.006 Å from the plane defined by the three crystallographically equivalent phosphine ligands. This structure contrasts the geometric parameters imposed by phenylene- and ethylene-linked tetradentate tripodal phosphine ligands that force the metal center to protrude out of the equatorial phosphine plane away from the apical ligand (27, 28). The incorporation of Si and P atoms in each five-membered chelate ring of likely provides the flexibility required for the metal complex to adopt such a symmetric geometry.

Na/Hg amalgam reduction of 2 under an N₂ atmosphere affords the Fe(I) complex, (3) (Fig. 3), as a brick-red powder. The absence of IR absorptions in the range of 2,100 to 1,700 cm⁻¹ (KBr pellet) rules out terminal N₂ coordination to a mononuclear Fe center in the solid state. A solution magnetic moment of 3.8 μ_B (C₆D₆) is observed at room temperature, and EPR measurements of 3 recorded at 10 K display g values of 4.66, 3.77 and 2.02 arising from M_s = 1/2 transitions, in addition to a feature at g = 6.22 stemming from transitions within the M_s = 3/2 manifold, consistent with a quartet ground state (29). The solid-state structure of 3 (Fig. 4) reveals a four-coordinate Fe complex. The Fe center is positioned below the P₃ plane defined by the phosphine donors [d(Fe-P_plane): 0.322 Å], and displays a shortened Fe–C distance [2.153(2) Å; Fig. 3]. The local Fe geometry is intermediate between tetrahedral and trigonal monopyramidal (τ₄ = 0.68) (30). Its geometry is topologically similar, although more pyramidalized, to that of the belt Fe centers in the structurally characterized resting state of FeMoco (τ₄ = 0.46), where the Fe atoms are directed toward the interstitial C atom and resting below the S₃ plane of the surrounding sulfides (5).

Fig. 4.

X-ray crystal structures of 3, 4, 5, and 6 with thermal ellipsoids set at 50% probability. Hydrogen atoms and solvent molecules have been removed. The phenyl rings have been rendered transparent to aid in visualization of the inner coordination sphere of iron. The {K(benzo-15-crown- 5)₂} countercation of 4 has been removed for clarity. Only one independent molecule of 6 is shown.

In hopes of preparing an Fe–N₂ complex supported by this ligand scaffold, we investigated the chemical reduction of 3. Compound 3 cleanly reacts with KC₈, and encapsulation of the potassium countercation of the resulting species with two equivalents of benzo-15-crown-5 affords a dark purple powder displaying an intense IR ν(NN) absorption at 1,927 cm⁻¹ for the terminal Fe–N₂ complex {K(benzo-15-crown-5)₂} (4). A Toepler pump analysis (SI Appendix) of 4 analyzed for 0.9 molar equivalents of N₂ upon oxidation with ferrocenium trifluoromethanesulfonate in THF. Complex 4 is diamagnetic in THF solution (³¹P NMR, d₈-THF, δ = 43.6 ppm). Its solid-state crystal structure (Fig. 4) confirms axial coordination of dinitrogen trans to the Fe-C_alkyl donor and shows that, upon coordination, the Fe center slides into the P₃ plane of the phosphines [d(Fe-P_plane): 0.012 Å] and away from the apical carbon [d(Fe-C) = 2.254(5) Å].

Silylation of M–N₂ complexes has been used to gauge the reactivity of the bound N₂ ligand toward electrophiles (31–33) and we canvassed related reactions with the present system. Whereas trimethylsilyl trifluoromethanesulfonate reacts productively with to afford (24), Fe-N₂⁻ 4 required the use of a highly encumbered silylating agent, triisopropylsilyl trifluoromethanesulfonate, at −78 °C to furnish a tractable diazenido product, (5). Characterization data for 5 include a ν(NN) vibration at 1,719 cm⁻¹, a solution magnetic moment of 2.75 μ_B (C₆D₆, 23 °C), and a broad room temperature ¹H NMR spectrum. Upon cooling to −80 °C in d₈-toluene, the ¹H NMR signals of 5 resolve into 14 distinguishable resonances between δ = 32 and −10 ppm, suggestive of an unsymmetrical paramagnetic species. An X-ray Diffraction (XRD) study of 5 (Fig. 4) accounts for this asymmetry at low temperature by verifying silylation of the beta N-atom of the N₂ ligand and also establishing a dechelated phosphine donor arm. The Fe geometry of 5 in the solid state is thus four-coordinate with a short Fe–C_alkyl distance of 2.116(1) Å, an Fe–N(1) distance of 1.713(1) Å, and two Fe–P bonds with an Fe–P_avg of 2.332 Å. The diazenido ligand features an elongated N(1)–N(2) bond length of 1.203(2) Å, consistent with substantial activation of the N≡N triple bond in free N₂ (1.098 Å). The diazenido ligand is severely bent at N(2) with a contracted N(1)-N(2)–Si (4) angle of 135.11°. Similar bond angles found in (TPB)Fe(N₂SiMe₃) (33) [TPB = tris(phosphine)borane] and (24) of 166.64° and 165.55°, respectively, are far more obtuse. Dechelation of the phosphine donor arm in 5 can be attributed, at least in part, to the necessity of installing such a large silyl substituent.

To discern how the Fe–C_alkyl interaction changes as a function of Fe redox state, we pursued structurally related carbonyl complexes of the scaffold. The advantage of the (n = +1, 0, −1) series of complexes is that a five-coordinate, approximately trigonal bipyramidal geometry is conserved across the three redox states with the strong-field CO ligand remaining bound at the axial position. This contrasts the situation when N₂ is the terminally bonded ligand, where probing the Fe–C_alkyl interaction across redox states is not possible due to a variety of other geometric and electronic changes, most notably the loss of the N₂ ligand that occurs upon oxidation of 4 to 3.

Preparation of this series (Fig. 3) begins with exposure of 3 to 1 atm of CO to afford the neutral and vacuum-stable carbonyl adduct (6). Carbonyl 6 features a solid-state IR absorption at 1,865 cm⁻¹ and is obtained in near-quantitative yield as a red-orange powder. Its room-temperature solution magnetic moment is 1.5 μ_B (C₆D₆) and it features a nearly axial signal in the frozen glass EPR spectrum (10 K, 2-MeTHF). An XRD study showed the presence of three independent molecules of 6 in the unit cell and confirms coordination of CO to a distorted (τ_{5, avg} = 0.68) trigonal bipyramidal Fe center (Fig. 4). The presence of slightly nonlinear C_alkyl–Fe–C_carbonyl angles (174.38°–178.13°) and marked asymmetry of the three P–Fe–P angles in the pseudoequatorial plane likely results from the orbitally-degenerate, Jahn–Teller active doublet ground state (SI Appendix, Table S2).

Reversible oxidation and reduction events at −0.62 and −1.85 V, respectively, are observed for 6 (SI Appendix, Fig. S12). Reduction of 6 with KC₈ in THF and subsequent encapsulation of the potassium ion with two equivalents of benzo-15-crown-5 furnishes, (7), as a dark purple powder. An IR absorption at 1,782 cm⁻¹ (KBr pellet) is indicative of considerable π-backbonding from the formally Fe(0) center into the C-O π* manifold, but to a lesser extent than in the isoelectronic complex {Na(12-crown-4)₂} in which isopropyl instead of phenyl substituents decorate the phosphines (34). Compound 7 is diamagnetic in solution and gives rise to a single resonance in the ³¹P NMR spectrum. Inspection of the XRD structure of 7 (SI Appendix, Fig. S15) indicates a substantial lengthening of the Fe–C_alkyl distance upon reduction (Table 1, Fig. 4) to greater than 2.30 Å. This fact highlights the effect of placing a flexible Fe-C subunit trans to a substrate-binding site.

Table 1.

Selected bond distances (in Å) for and compounds

Compound	d(Fe-Calkyl/Si)	d(Fe-CCO)	d(Fe-Pavg)
8	2.138(2)	1.786(2)	2.387
6	2.236†	1.734†	2.301†
7	2.303‡	1.738‡	2.177‡
11¶	2.3245(7)	1.842(3)	2.390
9¶	2.2942(4)	1.769(2)	2.276
10¶	2.2586(8)	1.732(3)	2.186

^†Average from three molecules in the unit cell.

^‡Average from two molecules in the unit cell.

^¶Ref. 34.

Oxidation of 6 with [Cp₂Fe][B(3,5-(CF₃)₂-C₆H₃)₄] leads to a lightening of the solution and growth of an intense ν(CO) stretch at 1,937 cm⁻¹ arising from the cationic carbonyl complex {B(3,5-(CF₃)₂-C₆H₃)₄}, (8). The solution magnetic moment for 8 (2.79 μ_B, CD₂Cl₂, 20 °C) is consistent with the expected S = 1 spin state. Attempts to obtain combustion analysis data on 8 were frustrated by its instability to vacuum. An XRD study confirms its structure (SI Appendix, Fig. S15) and reveals a mildly distorted (τ₅ = 0.85) trigonal bipyramidal Fe center with an Fe–C_alkyl bond distance of 2.138(2) Å that is appreciably shortened by comparison with that in 6 and 7. Cationic 8 displays the shortest Fe–C_alkyl bond of all of the five-coordinate compounds detailed herein.

Mössbauer measurements on compounds 6–8 were undertaken and indicate a decrease in the isomer shift (δ) upon reduction from 8 to 6 to 7 (SI Appendix, Fig. S15). This behavior is consistent with increasing backdonation into unfilled ligand orbitals upon reduction, and thus a higher degree of overall metal-ligand covalency and an increase in the s electron density at Fe (35). The δ range observed for compounds 6–8 (0.13 mm/s) is small and suggests that reduction occurs in diffuse orbitals, minimizing the electronic impact at the Fe nucleus. The trend (or lack thereof) in quadrupole splittings (ΔE_Q) is not readily explained.

The substantial reductive elongation of the Fe–C_alkyl bond that is observed across the (n = +1, 0, −1) series from 8 to 6 to 7 differs markedly from the corresponding and isoelectronic series (n = +1, 0, −1), for which modest shortening of the Fe–Si distances is instead observed upon reduction (34). On reduction, the Fe–Si bond in {B(3,5-(CF₃)₂-C₆H₃)₄} (11) contracts from 2.3245(7) to 2.2942(4) Å in (9), to 2.2586(8) Å in —Na(THF)₃ (10) (Table 1).

In compounds 6–8, the Fe–C_carbonyl and Fe–P bonds generally shorten on reduction just as they do in 9–11. In contrast, the Fe–C_alkyl bond lengthens (Table 1), and in 7 the central C atom forms three short Si–C_alkyl bonds (Table 2). The stability of tri(silyl)methyl carbanions has been attributed to the electropositivity of Si and negative hyperconjugation into adjacent Si-C σ* orbitals (36, 37), resulting in shorter Si–C_alkyl bonds. Thus, the structural data suggests the formation of a partially dissociated tri(silyl)methyl carbanion subunit upon reduction.

Table 2.

Selected results from NBO analysis

Compound	d(Fe-Calkyl), Å	d(Si-Calkyl),† Å	Fe-Calkyl σ-bond occupancy	q(Calkyl)‡	q(Fe)‡	Mean q(Si)‡	¶
4	2.254(5)§	1.845(5)§	1.76 (22.7% Fe, 77.3% C)	−1.76	−1.39	1.79	102
7	2.305(2)||	1.840(3)||	1.75 (22.7% Fe, 77.3% C)	−1.76	−1.78	1.78	101
12	2.144**	Not relevant	1.78 (39.7% Fe, 60.3% C)	−0.74	−1.96	Not relevant	191
13	2.251**	1.866**	1.76 (31.6% Fe, 68.4% C)	−1.51	−1.87	0.96	114

^†Average, central C–Si bonds.

^‡Computed natural charge.

^¶Donor–acceptor interaction energies (in kcal/mol) from the second-order perturbation analysis, n_C denotes a lone pair centered on C_alkyl, denotes an unoccupied orbital centered on Fe.

^§Distance obtained from XRD structure.

^||Distance obtained from one molecule in the asymmetric unit of the XRD structure.

^**Distance obtained from geometry optimization.

The Fe–C_alkyl distances in anions 4 and 7 are remarkably long, especially given that they are diamagnetic species. To aid in explaining this observation, we undertook natural bond orbital (NBO) analyses of these species as a means to compare the localized Fe–C bonding orbitals. NBO has been used previously to assess the differences in M–CH₃ and M–CF₃ bonding (38). Single-point calculations were performed at the B3LYP/6–31++G** level of theory using the crystallographically determined coordinates of the heavy atoms in 4 and 7. The calculations locate a highly polarized σ-interaction between the C_alkyl anchor of the ligand and the coordinated Fe atom (Fig. 5). The disparate contributions from Fe (22.7% 4, 22.7% 7) and C (77.3% 4, 77.3% 7) suggest a comparatively ionic σ bond (Fig. 5, Table 2) that is atypical of a mid-to-late transition metal alkyl (39). The dative Fe–P σ-bonds display an average of 22.9 and 25.7% Fe character in 4 and 7, respectively. The degree of ionicity in the Fe–C_alkyl interaction contrasts that calculated for the Fe–Si bond in the isoelectronic complex, which displays a far more covalent σ-bond between the Ar₃Si subunit and the Fe center (53.3% Fe and 46.7% Si character).

Fig. 5.

(A) Isocontour plot of the Fe–C_alkyl sigma bond of 7, and (B) the Fe–Si sigma bond of located from NBO analyses. Arrows indicate the location of highest electron density. Refer to methods for complete calculation details.

To better understand the unusually long bond lengths found in 4 and 7, we turned to computations on simple alkyl- and tri(silyl)alkyl-ligated five-coordinate Fe complexes. The structure of the S = 0, trigonal bipyramidal Fe alkyl complex [(Ph₃P)₂N][(CO)₄Fe-(n-propyl)] has been previously determined, and it features an n-propyl ligand in an axial site with a long Fe–C_alkyl distance of 2.20(2) Å. As such, it was of interest to us to compare the Fe–C_alkyl bonding of complexes of this framework to that in 4 and 7. To this end, NBO analyses were performed on the hypothetical model complexes , (12), and , (13). The optimized geometries of 12 and 13 also display axial coordination of the alkyl substituent to Fe. Notable structural parameters and NBO-derived data for 12 and 13 are tabulated in Table 2. Collectively, the data suggest that the introduction of electropositive Si substituents on carbon in 13 markedly increases the Fe–C_alkyl bond length relative to 12. The uncorrected energy of the Fe–C_alkyl interaction in 13 is substantially less than that of 12 (38). In addition, the Fe–C_alkyl σ-bond displays a decrease in covalency upon introduction of the electropositive Si substituents. As reflected in the natural charges (Table 2) and seen in the electrostatic potential maps of 12 and 13 (Fig. 6), the polarizable Si substituents in 13 display a comparatively positive charge, leading to a compensatory increase in the natural negative charge on the C_alkyl carbon relative to 12. As the Fe sites also display negative natural charges, an enhanced electrostatic repulsion between Fe and C_alkyl may explain the long Fe–C_alkyl distances in 13, as well as 4 and 7, relative to 12. Finally, an augmented electrostatic attraction between the electropositive Si atoms and the highly charged central carbon explains the short Si–C_alkyl bonds in 13, 4, and 7 [the Si–C bond length in Me₄Si is 1.875(2) Å] (40, 41).

Fig. 6.

(A) Model compounds 12 and 13 used as in silico models of an Fe(0) site interacting with an alkyl ligand. Partial charges on the C(SiH₃)₃ ligand of 13 are illustrated to emphasize the enhanced negative charge found on C as a result of three electropositive Si substituents. (B) Electrostatic potential maps of 12 (Left) and 13 (Right). Blue and red colors represent areas of negative and more positive potentials, respectively.

With the above structural, spectroscopic, and computational data in hand, it is of interest to return to the issue of the interstitial C atom of the cofactor and to consider whether the data presented herein can be of use as a model. In the resting state of FeMoco, the six belt Fe–C bonds are all relatively short [d(Fe-C) ∼2.0 Å] (5), imparting a pseudotetrahedral geometry to each Fe center. Such a geometry is observed for complex 3, albeit with an appreciably longer Fe–C distance [2.153(2) Å]. Our hypothesis, as suggested previously and depicted in Fig. 1 (11, 42) is that under electron-loading conditions the interaction between a belt Fe center and the interstitial carbon atom is weakened, allowing the Fe center to slide into a position closer to, or within, the plane of the three adjacent sulfide ligands (which may be protonated) concomitant with or before substrate coordination. This geometric change would serve to favor a terminal Fe–N₂ π-backbonding interaction, as observed for the transformation of 3 to 4.

Further reduction of the substrate-bound cluster (and possibly substrate protonation) could additionally weaken and thereby elongate the Fe–C_interstitial interaction. Such elongation is modeled by the carbonyl complexes 6 and 7, where additional Fe–C_alkyl lengthening occurs on reduction. Upon reduction, the buildup of negative charge on C_alkyl is compensated by shorter Si–C_alkyl bonds observed in 4, 7, and hypothetical 13 (Table 2). As the Pauling electronegativities of Fe (χ_Fe = 1.83) and Si (χ_Si =1.90) are quite similar (43), the behavior of the Si atoms in may crudely model that proposed for the remaining five Fe atoms within FeMoco. These electropositive Fe atoms may stabilize an interstitial carbide bearing an increasing degree of negative charge as it moves away from the substrate-bound Fe site, while maintaining overall structural integrity of the cofactor. This hypothesis finds some theoretical support in computational studies performed by Dance (44), Nørskov and coworker (10), and Noodleman and coworkers (45) on the FeMoco. Their studies, performed before (10, 45) and after (44) the assignment of the light interstitial atom (X) as C, suggest that substrate coordination to Fe at a site trans to X is favorable, and induces an elongation of the Fe–X interaction. Likewise, these studies suggest that an increase in charge density at X, caused by the singular elongated Fe–X interaction, is compensated by shorter Fe–X bonds to the five additional belt Fe atoms in what has been described as coordinative allosterism (44). It is interesting to note that George et al. (16) have recently communicated Nuclear Resonance Vibrational Spectroscopy (NRVS) and Extended X-ray Absorption Fine Structure (EXAFS) results that suggest elongation of a single Fe–C bond in the FeMoco under catalytic conditions in the presence of propargyl alcohol.

Conclusions

In conclusion, we have prepared a series of Fe coordination complexes designed to explore a possible role of the interstitial C atom recently assigned for the FeMoco. We have shown that the ligand can be installed on Fe to afford complexes featuring an unusually flexible Fe–C_alkyl interaction owing to a high degree of ionicity in the Fe–C_alkyl bond made possible by the three silyl substituents bonded to the C atom. A trigonal pyramidal complex can be prepared that, upon reduction, exhibits Fe–C_alkyl bond lengthening concomitant with N₂ binding trans to the C atom. The resulting five-coordinate trigonal bipyramidal anion thus features an unusually long Fe–C_alkyl bond distance and, according to an NBO analysis, an electron pair between Fe and C_alkyl that is polarized toward the carbon atom (i.e., carbanion character). Silylation of the coordinated N₂ ligand at the beta N-atom furnishes a silyldiazenido product , wherein one phosphine arm has dissociated from Fe in the solid state. To further explore how the Fe–C_alkyl interaction responds solely as a function of the formal Fe redox state, the series of geometrically similar trigonal bipyramidal carbonyl complexes (n = +1, 0, −1) was characterized. Combined Mössbauer, structural, and DFT data collectively suggest a decreasing degree of covalency in the Fe–C_alkyl bond and a corresponding increase in the covalency of the Fe–P and Fe–C_CO bonds as the formal oxidation state at Fe is decreased from Fe(II) to Fe(I) to Fe(0). The presence of the three electropositive and polarizable silyl substituents allows the C_alkyl carbanion to partially dissociate from the Fe center as electrons are added to the complex. The collection of data described has been considered as an inorganic model to explore the hypothesis that a belt Fe–C_interstitial bond in FeMoco might be modulated as a means of facilitating N₂ binding and reduction at a single Fe site.

Materials and Methods

General Considerations.

All manipulations were carried out using standard Schlenk or glovebox techniques under an N₂ atmosphere. Unless otherwise noted, solvents were deoxygenated and dried by thoroughly sparging with Ar gas followed by passage through an activated alumina column in the solvent purification system by SG Water. Nonhalogenated solvents were tested with a standard purple solution of sodium benzophenone ketyl in tetrahydrofuran to confirm effective oxygen and moisture removal. All reagents were purchased from commercial vendors and used without further purification unless otherwise stated. (1), benzyl potassium, KC₈, and [Cp₂Fe][B(3,5-(CF₃)₂-C₆H₃)₄] were synthesized following literature procedures. Elemental analyses were performed by Midwest Microlab. Deuterated solvents were purchased from Cambridge Isotope Laboratories, degassed, and dried over activated 3-Å molecular sieves before use. Deuterated THF was dried over NaK alloy before use. The ¹H chemical shifts are reported in parts per million relative to tetramethylsilane, using residual solvent proton resonances as internal standards; ³¹P chemical shifts are reported in parts per million relative to 85% aqueous H₃PO₄. Solution phase magnetic measurements were performed by NMR line-shift analysis. IR measurements were obtained on samples prepared as KBr pellets on a Bio-Rad Excalibur FTS 3000 spectrometer. The X-band EPR spectra were obtained on a Bruker EMX spectrometer on 5-mM solutions prepared as frozen glasses in 2-MeTHF. Optical spectroscopy measurements were taken on a Cary 50 UV-Vis spectrophotometer using a 1-cm two-window quartz cell.

X-Ray Crystallography.

XRD studies were carried out at the Beckman Institute Crystallography Facility on a Brüker Kappa Apex II diffractometer and Brüker Smart 1000 CCD diffractometer (Mo Kα radiation). Structures were solved using SHELXS and refined against F² on all data by full-matrix least squares with SHELXL. The crystals were mounted on a glass fiber with Paratone N oil.

Electrochemistry.

Electrochemical measurements were carried out in a glovebox under an N₂ atmosphere in a one-component cell using a CD instruments 600B electrochemical analyzer. A glassy carbon electrode was used as the working electrode and platinum wire was used as the auxiliary electrode. All reported potentials were referenced to the ferrocene couple Cp₂Fe⁺/Cp₂Fe. Solutions (THF) of electrolyte (0.3 M tetra-n-butylammonium hexafluorophosphate) and analyte were also prepared under an inert atmosphere.

Mössbauer Spectroscopy.

Spectra were recorded on a spectrometer from SEE Co. operating in the constant acceleration mode in a transmission geometry. Spectra were recorded with the temperature of the sample maintained at 80 K. The sample was kept in an SVT-400 Dewar from Janis, at zero field. Application of a magnetic field of 54 mT parallel to the γ-beam did not cause detectable changes in the spectra recorded at 80 K. The quoted isomer shifts are relative to the centroid of the spectrum of a metallic foil of α-Fe at room temperature. Samples were prepared by grinding polycrystalline material into a fine powder and then mounted in a cup fitted with a screw cap as a boron nitride pellet. Data analysis was performed using the program WMOSS (www.wmoss.org) and quadrupole doublets were fit to Lorentzian lineshapes.

DFT Calculations.

Single-point calculations and NBO analyses were performed using the Gaussian03 suite of programs with the RB3LYP level of theory and a 6–31++G** basis set for all atoms. The geometries of 12 and 13 were optimized with RB3LYP/6–31++G**. The geometries of 4 and 7 were obtained from the XRD coordinates. To obtain donor–acceptor interaction energies from the second-order perturbation analysis portion of NBO for the Fe–C_alkyl sigma bonds, the occupancy threshold was manually adjusted. The total electron density described by these alternative Lewis structures is only slightly lower than that from the default options (percent electron densities for default/alternative Lewis structures: 4, 97.7/94.8; 7, 97.5/94.7; 12, 97.0 /95.9; 13, 97.7/97.1).