Light Enhanced Fe-Mediated Nitrogen Fixation: Mechanistic Insights Regarding H₂ Elimination, HER, and NH₃ Generation

Abstract

Despite their proposed accumulation at the Fe sites of the FeMo-cofactor of MoFe-nitrogenase, the presence of hydride ligands in molecular model systems capable of the nitrogen reduction reaction (N₂RR) appears to diminish the catalytic N₂-to-NH₃ conversion. We find that for an iron-based system bearing the trisphosphine ligand P₂P^Ph, a dramatic difference in yields is observed for N₂RR catalyzed by precatalysts with zero, one, or two hydride ligands; however, irradiating the three different catalysts with a mercury lamp results in similar yields. Although the efficacy for N₂RR versus the hydrogen evolution reaction (HER) is modest for this system by comparison to certain iron (and other metal) catalysts, the system provides an opportunity to study the role of hydrides in the selectivity for N₂RR versus HER, which is a central issue in catalyst design. Stochiometric reactions with hydride containing precatalysts reveal a hydrogen evolution cycle in which no nitrogen fixation occurs. Irradiation of the dihydride precatalysts, observed during turnover, results in H₂ elimination and formation of (P₂P^Ph)Fe(N₂)₂, which itself is unreactive with acids at low temperature. N₂ functionalization does occur with acids and silyl electrophiles for the reduced species [(P₂P^Ph)Fe(N₂)]⁻ and [(P₂P^Ph)Fe(N₂)]²⁻, which have been characterized independently. The requirement of accessing such low formal oxidation states explains the need for strong reductants. The low selectivity of the system for functionalization at N_β versus Fe creates off-path hydride species that participate in unproductive HER, helping to explain the low selectivity for N₂RR over HER. The data presented here hence lends further insight into the growing understanding of the selectivity, activity, and required driving force relevant to iron (and other) N₂RR catalysts.

SYNOPSIS TOC

Introduction

Substantial progress has been made in the development and understanding of molecular catalysts for N₂-to-NH₃ conversion, commonly referred to as the nitrogen reduction reaction (N₂RR).^1–3 The number of well-defined complexes capable of N₂RR is expanding rapidly, and significant improvements in turnover and efficiency have been made.^4,5 With the growing number of systems available, it becomes increasingly possible to uncover general design principles that will aid in further progress for the field. The selectivity of N₂RR versus the competing hydrogen evolution reaction (HER) is a central selectivity issue in need of model studies.⁶ Competing HER not only limits the efficiency of molecular catalyst systems but also limits nitrogenase enzymes.^7,8 Additionally, and relatedly, a deeper understanding as to why seemingly related synthetic catalysts often require very different reductant and acid combinations to be competent for N₂RR is needed.

HER can occur by the background reaction between the reductant and acid; synthetic N₂RR catalysts depend on limiting the rate of background HER relative to the catalytic N₂RR rate. A catalyzed HER process, presumably accessible and competitive for many N₂RR catalysts, can also limit the efficacy of N₂RR selectivity. Both scenarios can be at play.^4,9

For a complex that catalyzes both N₂RR and HER, numerous pathways for the latter process are possible. H₂ may evolve via protonation of a metal-bound hydride,^3,9 a commonly proposed pathway for synthetic HER catalysts. Accordingly, the build-up of M–H species has been observed both during and after catalytic N₂RR experiments.^2,9,10 The accumulation of M–H species is generally thought to attenuate N₂RR activity, and hydride precatalysts can give rise to diminished yields for N₂RR.^2,3,9 When hydride precursors serve as active precatalysts for N₂RR, it is presumed they react with acid and reductant to release H₂, thereby generating a species that is on-path for N₂RR.^3,9 As an example of this, for a tris(phosphine)borane iron catalyst system studied extensively by our lab, [(P₃^B)Fe(N₂)]⁻ (P = o-(PiPr₂)₂C₆H₄), a dihydride intermediate was observed as an off-path resting state of the system when KC₈ and HBAr^F₄(Et₂O) (BAr^F₄ = tetrakis-(3,5-bis(trifluoromethyl)phenyl)borate) were employed (Figure 1).⁹ This dihydride species can be converted to an on path intermediate by reductive protonation.⁹ A conceptually similar pathway has been described by Nishibayashi and coworkers for a (PNP)Fe system (Figure 1, left).³

Figure 1.

(Left) Protonation and reduction of off-path iron hydrides results in the formation of on-path species capable of N₂RR. (Right) Previously reported iron based catalysts and the conditions under which light enhanced nitrogen fixation by mercury lamp irradiation was observed.^1g

Another possible competing HER pathway present within N₂RR systems that has been considered by our lab involves a bimolecular, N_xH_y ligand-mediated step wherein two Fe(N_xH_y) intermediates that feature weak N–H bonds evolve H₂ (Eqs. 1 and 2).⁶

$$\text{2 Fe-NNH}\to {\text{2 Fe-N}}_{\text{2}}{\text{+ H}}_{\text{2}}$$ (Eq. 1)

$${\text{2 Fe-NNH}}_2\to {\text{2 Fe-NNH+ H}}_{\text{2}}$$ (Eq. 2)

Protonation at the metal versus at coordinated N₂ to form a metal hydride should be thermodynamically favored,⁶ though the kinetic site of protonation can involve the coordinated N₂ ligand. Even if protonation at N₂ is kinetically favored, this can be followed by intra- or intermolecular H-atom/hydride/proton transfer to form a metal hydride.¹¹ Initial protonation at a site on the auxiliary ligand can also be kinetically favored.¹² Protonation at the terminal nitrogen (N_β) is desired for selectivity towards nitrogen fixation. For the P₃^BFe-system, iron is by far the thermodynamically favored site for protonation. However, the steric profile of the complex and the acids used appear to render functionalization at N_β kinetically favorable.¹³

Recently, our group reported two related iron-based complexes bearing hydride ligands, (P₂P^Ph)Fe(H)]₂(µ–N₂) (1) and (P₂P^Ph)Fe(N₂)(H)₂ (2) (P₂P^Ph = bis(o-diisopropylphosphino-phenyl)-phenylphosphine), that are modestly active systems for N₂RR (Figure 1, right).² For complex 2, photo-induced H₂ elimination was proposed to yield a more activated Fe–N₂ species that could undergo subsequent reductive protonation steps to generate NH₃. An H₂-elimination step can also be observed for (P₃B)(µ-H)Fe(N₂)(H), resulting in the formation of (P₃B)Fe(N₂).^2,14

Although precatalysts 1 and 2 are significantly less efficient for N₂RR than [(P₃^B)Fe(N₂)]⁻ and certain other metal catalysts, they provide a fascinating model system for in-depth study because they have been shown to display substantial enhancement for N₂RR under irradiation.² Furthermore, these catalysts bear hydride ligands but are nonetheless active for N₂RR, affording an opportunity to investigate the role of the hydride ligands in N₂RR and competing HER. Finally, a better understanding of the electronic and structural factors that influence the required redox potential for N₂RR in this phosphine-iron catalyst system compared to other systems can aid in the development of selective catalysts that operate at a comparatively low net driving force.

Results and Discussion

In our prior communication we proposed that the product of H₂ elimination from well-characterized dihydride 2 might be “(P₂P^Ph)Fe(N₂)”.² Reasoning that this or a related species might be on-path for N₂RR, we targeted an independent synthesis. (P₂P^Ph)FeBr₂ (3) provided a logical starting point. Treatment of 3 with 1.05 equiv sodium mercury amalgam resulted in the formation of (P₂P^Ph)FeBr (4) in 65% yield (Scheme 1). 4 exhibits C_s symmetry in solution based on its ¹H NMR spectrum and a distorted tetrahedral geometry (τ₄ = 0.77)¹⁵ in the solid state (See Figure S1 and S62).

Scheme 1.

Preparation of dinitrogen adducts of (P₂P^Ph)Fe from the bromide precursors 3 and 4.

Bromide 4 is a useful synthon for several complexes of present interest. For example, treatment of 4 with NaHBEt₃ in toluene at −78 °C provides a more favorable route to the diiron complex 1 (Scheme 1), whose preparation was previously described by NaHBEt₃ reduction of 3. Furthermore, 4 can be reduced with sodium mercury amalgam in either benzene or THF to provide a new, maroon red complex (P₂P^Ph) Fe(N₂)₂ (5). 5 can be alternatively prepared by reduction of 3 with excess sodium mercury amalgam in benzene (Scheme 1).

18-electron 5 exhibits two intense bands in its IR spectrum (thin film; ν_symm = 2065 cm⁻¹, ν_asymm = 2005 cm⁻¹) and its solid-state crystal structure (See Figure S63), reveals a distorted trigonal bipyramidal geometry at iron (τ₅ = 0.54)¹⁶ with Fe–P distances ~0.15 Å shorter than in 3, reflecting its singlet ground state.² The structure and stretching frequencies of the five-coordinate N₂ complex 5 is closely related to recently reported (P^RP^Cy₂)Fe(N₂)₂ (P^RP^Cy₂ = RP(CH₂CH₂PCy₂)₂, R = Ph, tBu) complexes.¹⁷ In the latter N₂ complexes, facile N₂ dissociation hampers their isolation. Although 5 is also susceptible to N₂ dissociation (vide infra) it can be readily isolated by evaporation of the solvent in vacuo followed by extraction with pentane.

N₂-Binding Equilibria of 5.

A solution equilibrium exists between 5 and a dinuclear, mono-N₂-bridged complex 6 (Scheme 1). This is clearly gleaned from ¹H and ³¹P NMR spectroscopies. For example, the ¹H NMR signal intensities for 5 decay upon degassing the solution in a J-Young NMR tube and the signals corresponding to 6 grow in (See Figure S23). Addition of N₂ regenerates 5. The absence of an N₂ stretch in the IR spectrum of 6, and the release of 1.5 equiv of N₂ per Fe on conversion of 5 to 6, as measured by a Toepler pump experiment, are consistent with our formulation of 6 (Scheme 1; Eq. 3):

$$\text{2 }\left({\text{P}}_{\text{2}}{\text{P}}^{\text{Ph}}\right)\text{Fe}{\left({\text{N}}_{\text{2}}\right)}_{\text{2 }}\to {\left\{\left({\text{P}}_{\text{2}}{\text{P}}^{\text{Ph}}\right)\text{Fe}\right\}}_{\text{2}}\left({\text{µ–N}}_{\text{2}}\right){\text{ + 3/2 N}}_{\text{2}}$$ (Eq. 3)

Monitoring the ¹H and ³¹P chemical shifts of 6 over a 130 °C range under vacuum reveals deviation from Curie-behavior (Figure 2 and S31–S33). The singlet ground state of 6 likely arises from antiferromagnetic coupling of two S = 1 iron nuclei. The dramatic shifts in the NMR spectra are therefore attributed to partial population of triplet and quintet states separated by 2J and 6J from the ground state, respectively (as obtained for the Heisenberg-Dirac-VanVleck Hamiltonian in the notation H_HDVV = −2JS₁S₂). Fitting of the appropriate Boltzmann function to the experimental data yields J = −940 ± 9.4 cm⁻¹ (Equation S1).¹⁸ Antiferromagnetic coupling for an N₂–bridged diiron species has been observed previously.¹⁹

Figure 2.

The ¹H NMR chemical shifts of 6 plotted as a function of 1/T display a deviation from Curie-behavior, due to the population of a low-lying exited state. A fit of the data, indicated with black lines, gives J = −940 ± 9.4 cm⁻.

Increased Turnover with Non-Hydride Precatalysts and Identification of Off-Path Species.

Previous N₂RR studies using the (P₂P^Ph)Fe-system were performed with the hydride complexes 1 and 2 as (pre)catalysts. We wondered whether increased turnover numbers might be realized with (P₂P^Ph)Fe(N₂)₂, 5, devoid of hydride ligands. Indeed, catalysis under the same conditions with 5 resulted in significantly higher NH₃ yields than those afforded by 1 and 2. For example, at a loading of 150 equivalents acid and 180 equivalents reductant at −78 °C in Et₂O, in the absence of light, complexes 1 and 2 catalyzed only 3.6 ± 0.6 and 2.6 ± 0.01 equiv NH₃ per iron, respectively, whereas 5 catalyzed the generation of 6 ± 0.5 equiv. Interestingly, a comparable NH₃ yield (5.1 ± 0.02 equiv per iron center) could be realized with 5 using only 1/3 as much reductant and acid (50 equiv HBAr^F₄(Et₂O) and 60 equiv KC₈), which was not the case for either 1 or 2. A possible explanation for this difference is that HER catalysis from the hydrides, which are present in the highest concentration at the onset of runs with 1 and 2, outcompetes N₂RR.

Similar NH₃ yields were obtained for the three different precatalysts 1, 2, and 5 in catalytic experiments irradiated with a mercury lamp. We presume that dihydride 2 releases H₂ upon irradiation with light to yield 5, and that this transformation occurs rapidly under turn-over conditions as all (pre)catalysts give similar yields. The consumption of hydride species via photolysis reduces HER catalyzed by the hydrides, thus increasing overall efficiency for NH₃.

To determine whether catalyzed HER contributes to the low yields of NH₃ obtained with dihydride 2, hydrogen evolution was measured under catalytically relevant conditions. As shown in Figure 3, the initial rate of H₂ evolution at −78 °C, is significantly enhanced by the presence of either 2 or 5. These data suggests both 2 and 5 are comparatively competent catalysts for HER, whereas complex 5 is a more effective (pre)catalyst for N₂RR. Indeed, the fact that most of the acid is consumed within 30 minutes at −78 °C speaks to how rapidly 5 must catalyze NH₃ production for N₂RR to be kinetically competitive.

Figure 3.

Time profiles of the formation of H₂ from HBAr^F₄ and KC₈ in Et₂O at −78 °C. Data are presented for the background reaction of these reagents in the absence of catalyst (black diamonds), as well as in the presence of either 2 (red circles) or 5 (blue squares). Each time course was collected continuously from a single experiment. The final data points, recorded after 16 hours, are omitted from the graph.

Additional evidence for the active role of 2 in HER is obtained from Mossbauer studies of freeze-quenched samples. Freeze-quenching of a catalytic run using ⁵⁷Fe-labeled 2, 50 equiv acid, and 60 equiv reductant, shows its disappearance within 5 minutes (Figure 4, middle trace). A new broad feature, likely due to the overlap of several species, is observed. Freeze-quenching the reaction after 30 minutes provides a similarly broad signal (Figure 4, bottom trace), but one that also contains 2, with its characteristic small quadrupole splitting (constituting ~ 40% of the total). Experiments using 5 as the precatalyst provide an analogously broad signal after 5 and 30 minutes (See Figure S59–S60). Notably, dihydride 2 is always observed at the end of a catalytic experiment, once the sample has been warmed to room temperature. It is the major species present (typically ~ 90% by Mössbauer spectroscopy). Furthermore, IR and NMR spectra recorded after runs using 5 as the precatalyst show 2 as the only identifiable species upon warming. These data collectively suggest that the catalytic system converts to a Fe–H species (2), which is on path for HER (vide infra), as the major product.² This finding is similar to that of [(P₃^B)Fe(N₂)]^-, which also ends tied-up in an off-path hydride-borohydride state (Figure 1, left).⁹

Figure 4.

Frozen solution Mӧssbauer spectra collected at 80 K in the presence of a 50 mT parallel magnetic field. Spectra of a catalytic mixture, using 2 (blue) as precatalysts (top), quenched after 5 minutes (middle) and 30 minutes (bottom) of stirring. For parameters of individual components, see SI.

Oxidative addition and reductive elimination of H₂.

To investigate potential pathways by which hydride species form during catalysis, stoichiometric reactions were performed with dihydride 2, the bis–N₂ complex 5, and dinuclear 6. Addition of H₂ to 5 (or 6), followed by N₂, resulted in the quantitative formation of 2 (Scheme 2). However, addition of H₂ to 5 at −78 °C in a J-Young tube for one hour resulted in the appearance of a trace amount of 2. Full conversion was only observed upon warming to room temperature (see Figure S25). The latter result strongly suggests that the formation of 2 under the catalytic conditions at −78 °C does not occur by a reaction between 5 and H₂.

Scheme 2.

Light induced reductive elimination of H₂ from 2 leads to a transient unobserved four coordinate species, which binds N₂ to form 5. The H₂ elimination is reversible as 5 reacts back to 2 in the presence of H₂.

Irradiating solutions of 2 with a 100 W mercury lamp at −78 °C or room temperature results in darkening of the solution and the formation of 5 (Scheme 2). Complete disappearance of 2 is not observed, suggesting the reaction is reversible (Figure S27). A possible 16-electron intermediate, such as “(P₂P^Ph)Fe(N₂)”,² could not be identified by NMR, IR, or Mössbauer spectroscopy.

H₂ elimination from 2 to 5 leads to a significant decrease in ν(NN) stretching frequencies due to increased backbonding upon H₂ elimination. A similar effect was observed previously for (P₃^B)(µ-H)Fe(N₂)(H).² In this context, these systems crudely model a proposed N₂ binding/activation via H₂ elimination at the E₄ state of the iron-molybdenum cofactor (Figure 5).²⁰ Clearly, increased N₂ activation upon H₂ elimination observed for this P₂P^PhFe-system would be even more pronounced for the unobserved, but perhaps catalytically relevant mono-N₂ adduct “P₂P^PhFe(N₂)” (vide infra).

Figure 5.

(Top) H₂ elimination from the E₄ state resulting in a more electron rich center is proposed from Mo-nitrogenase, and Fe-nitrogenase. Light induced reductive elimination of H₂ from 2 leads to increased back-bonding due to the formal reduction from Fe^II to Fe⁰ (Bottom).

Stoichiometric reactivity and hydrogen evolution.

To further probe HER catalysis by the present system, HBAr^F₄ was added to (P₂P^Ph)Fe(N₂)₂ 5 at −78 °C, causing a color change from maroon to dark yellow upon warming. The product of protonation at iron was identified as [(P₂P^Ph)Fe(N₂)₂(H)][BAr^F₄] (7) (Scheme 3), featuring a diagnostic ¹H NMR hydride resonance (−17 ppm) and bands at 2069, 2194 and 2264 cm⁻¹, corresponding to the ν(Fe–H) and ν(NN) IR stretches. Its solid-state structure was also determined (see Figure S64). This complex can also be obtained by oxidation of {(P₂P^Ph)Fe(H)}₂(µ–N₂) 1 with either FcBAr^F₄ (Fc = bis(η₅-cyclopentadienyl)iron) or HBAr^F₄ (Scheme 3). Of primary interest, protonation of 2 with HBAr^F₄ likewise generates 7 with concomitant H₂ release, possibly via an “[Fe(H₂)(H)]⁺” adduct.²¹

Scheme 3.

Pathways towards 7 and 8. Preparative and NMR scale reactions were performed at −78 °C.

Monohydride 7 can be cleanly reduced to dinuclear 1 using either Cp*₂Co or stoichiometric KC₈. Reduction of 7 with an excess of KC₈ by contrast generates a different diamagnetic species which, following addition of 18-crown-6, could be isolated in pure form as [P₂P^PhFe(N₂)(H)][K(18-crown-6)] (8) (Scheme 3). Complex 8 features a diagnostic hydride resonance in its ¹H NMR spectrum (δ = −9.69 ppm, dt), and its solid-state structure displays a short Fe–N (1.774(1) Å) and an elongated N–N (1.139(2) Å) bond. A high degree of activation of N₂ is reflected by its ν(NN) (1924 cm⁻¹).

Stoichiometric mixing of cation 7 and anion 8 resulted in comproportionation to 1 (> 90% yield). Proton transfer (PT) from 7 to 8 might have alternatively resulted in the formation of 2 and 5 (Scheme 4), but this was not observed. This may be rationalized by low acidity of the hydride ligand in 7, which is not deprotonated by NaOtBu. Relatedly, 8 is weakly basic and is not protonated by MeOH at −78 °C. The absence of proton transfer between 7 and 8 makes this an unlikely step for (re)generating 2 and 5 under turnover conditions. Dihydride 2 can, however, be obtained readily by protonation of 8 with HBAr^F₄ at low temperature (Scheme 4).

Scheme 4.

Electron and proton transfer from 7 and 8

Apparent differences in reactivity of the hydrides with respect to N₂RR can be rationalized by the availability, or lack of, kinetically competent pathways for the hydrides to be converted to on-path Fe–N₂ species at −78 °C. For the P₂P^Ph system, no pathway has been identified via which hydrides convert back to on-path Fe–N₂ species. Instead, in stoichiometric reactions, the different hydrides interconvert in an HER cycle (Figure 6). However, the observation of NH₃ production during catalytic experiments with 1 or 2 as precatalysts indicates that there must be some pathway to a species active for N₂RR, even if comparatively inefficient.

Figure 6.

Summary of the stoichiometric reactivity observed with [P₂P^PhFe(N₂)_x(H)_y]^+/0/−. Grey arrows indicate reactions that do not occur at −78 °C. Black arrows indicate reactions that occur at −78 °C

Reduction of (P₂P^Ph)Fe(N₂)₂.

The fact that there is no reactivity between 5 and HBAr^F₄ at −78 °C indicates that the N₂ ligands are not sufficiently activated to be protonated, in accord with comparatively high N₂ stretching frequencies for 5 (2065 and 2009 cm⁻¹). To explore whether further reduction might generate a more reactive and hence on-path species, 5 was stirred with 1 equiv potassium naphthalide followed by the addition of 18-crown-6. This produced the anionic, 4-coordinate S = ½ (µ_eff = 1.80) complex [(P₂P^Ph)Fe(N₂)][K(18-crown-6)] (9) (Scheme 5). 9 features a single and highly activated N₂ ligand (1872 cm⁻¹). Its solid-state crystal structure shows a disordered tetrahedral iron center (τ₄ = 0.75), and CV measurements show a quasi-reversible Fe^0/− redox event centered at −2.5 V vs Fc/Fc⁺ (Figure 7). At more negative potential, an irreversible, presumably Fe^−2/−1 redox event is observed.

Scheme 5.

The reduction of 5 results in the formation of 9 or 10 depending on the equivalents of reductant used.

Figure 7.

(Left) Cyclic voltammetry data of 9 scanning cathodically, (Middle) Mӧssbauer spectrum of 10, (Right) Asymmetric unit of the XRD structure of 10. In the dimeric structure, the iron centers are related by an inversion center. Hydrogen atoms and disorder in one of the i-Pr moieties are omitted for clarity.

Reduction of 5 with an excess of KC₈ produced a diamagnetic species (³¹P NMR: δ = 113.13, doublet; 95.15 ppm, triplet) identified as [(P₂P^Ph)Fe(N₂)][K₂(THF)₃] (10) (Scheme 5). Complex 10 is an unusual iron species in that it is isoelectronic with [Fe(CO)₄]^2- (vide infra).^22,23 A ¹⁵N-labeled analogue was synthesized by reduction of (P₂P^Ph)FeBr₂ under ¹⁵N_2; its ¹⁵N NMR spectrum (2.36 and −26.23 ppm) rules out the possibility of a dinuclear structure [{(P₂P^Ph)Fe}₂(μ-N₂)]K₂.²⁴ Consistent with our assignment of dianion 10, its IR spectrum displays a ν(NN) at 1677 cm⁻¹ (1591 cm⁻¹ for 10–¹⁵N₂) that is broadened due to ion-pairing, consistent with a very strongly activated N₂ ligand. Addition of 18-crown-6 resulted in intractable decomposition, suggesting tight ion-pairing is important to its stability.²⁵

The structure of 10 in the solid-state (Figure 7) presents two distorted tetrahedral iron centers (τ₄ = 0.71) that are related by an inversion center within a dimeric unit. Tight ion pairing is evident from the close proximity of each iron center to the potassium cations (Fe–K = 3.442 and 3.567 Å); each N₂ ligand interacts with three potassium ions. The Fe–N bond is remarkably short (1.728(2) Å), ~ 0.1 Å shorter than the Fe–N bonds in 5, reflective of very strong backbonding. Relatedly, significant N–N elongation is also observed (1.189(3) Å). The Fe–P bond distances are also highly contracted at 2.1494(6) Å, in line with the very strong covalency expected of a d¹⁰ tetrahedral iron center. Prior to this study, tetrahedral Fe^−II species have been limited to complexes with very strong pi-acceptor ligands, such as CO,^22,23 PF₃,²⁶ (C₂H₄),^27,28 (COD) (COD = cyclooctadiene),²⁷ and CNAr.^25,29 Additional species bearing phosphorine³⁰ and nitrosyl³¹ ligands have also been reported, however the assignment of their oxidation state is ambiguous.

The Mössbauer spectrum of a perfectly tetrahedral d¹⁰ iron complex, such as Na₂[Fe(CO)₄], should show a singlet instead of a quadrupole doublet due to the spherical electric field gradient at the iron nucleus;^22,32 any quadrupole splitting in Na₂[Fe(CO)₄] is barely discernable.³² Similarly, the Mössbauer spectrum of 10 shows an apparent singlet (Figure 7) which can be fit by a small quadrupole splitting (δ = 0.27 mm/s, ΔE_Q = 0.26 mm/s). The very small quadrupole splitting in 10, which is required to have at least a modest electric field gradient owing to the presence of three unique types of donor ligands, indicates its classification as a d¹⁰ tetrahedral structure is appropriate, at least to the extent this description is apt for Na₂[Fe(CO)₄] given the significant covalency in both species.

Functionalization of formal Fe^−I and Fe^−II species.

Current examples of Fe-mediated N₂RR are thought to proceed through Fe–N₂ intermediates with ν(NN) stretching frequencies below 1970 cm⁻¹.^1d,3,33 The N₂ ligand of [(P₂P^Ph)Fe(N₂)][K(18-crown-6)] (9) has a stretching frequency of 1872 cm⁻¹ and in this context should be activated enough to be functionalized. Attempts to protonate 9 with stoichiometric HBAr^F₄ unfortunately resulted in complex product mixtures. Silylium ions (R₃Si⁺) have been used as surrogate electrophiles for protons to model unstable protonated Fe-N_xH_y species.^1c,34–38 Reacting 10, generated in situ, with one equivalent of Me₃SiCl at −78 °C, results in an immediate color change from dark purple to dark orange. After work-up, diamagnetic [(P₂P^Ph)Fe(NNSiMe₃)]K ([11-NNSiMe₃]⁻) was isolated as a dark brown solid in 50% yield (Scheme 6). As for 10, a tight ion-pair seems to be important for its stability; addition of 18-crown-6 results in its decomposition. The solid-state structure of [11-NNSiMe₃]⁻ reveals a four-coordinate iron center with a distorted tetrahedral geometry (τ₄ = 0.76). The Fe–N bond length of [11-NNSiMe₃]⁻ is even shorter than that in 10 (1.664(7) Å vs. 1.728(2) Å respectively), and the N–N bond length is much longer (1.270(9) Å vs. 1.189(3) Å) in 10.

Scheme 6.

Synthesis of Fe–silyldiazenido complexes [11NNSiMe₃]K, [11-NNSiⁱPr₃]K and 12-NNSiⁱPr₃.

Attempts to oxidize [11-NNSiMe₃]⁻ at −78 °C with cobaltocenium to generate the neutral diazenido species (P₂P^Ph)Fe-NNSiMe₃ resulted in a mixture of species, presumably complicated by the loss of Me₃Si·. Based on low temperature EPR data (Figure 8), we assign the major product of oxidation to be the iron-silyl complex (P₂P^Ph)Fe(SiMe₃)(N₂) (its EPR signature is highly similar to that of (P₂P^Ph)Fe(N₂)(H); ² see Figure 8 ). There is also a minor component in the EPR trace that can be tentatively assigned as the expected diazenido (P₂P^Ph)Fe(NNSiMe₃). Use of ⁱPr₃SiOTf instead leads to the analogous [11-NNSiⁱPr₃]⁻ complex, but in this case its oxidation affords a clean EPR spectrum consistent with the diazenido species 12-NNSiⁱPr₃ (Figure 8). Addition of ⁱPr₃SiOTf to [(P₂P^Ph)Fe(N₂)]⁻ 9 generates the same species as is evident by IR and EPR spectroscopy (Figure 8). The IR spectrum of 12-NNSiⁱPr₃ displays an intense band corresponding to νNN at 1660 cm⁻¹, characteristic of iron diazenido species.^34,37,39 In contrast with the -SiMe₃ derivative, the -SiⁱPr₃ species is stable for days. We suspect that for the less bulky -SiMe₃ derivative, kinetically competitive N-to-Fe silyl migration is operative.

Figure 8.

Collected EPR data; (Black) Spectrum of [(P₂P^Ph)Fe(N₂)][18-crown-6)] (9). (Maroon) Spectrum observed upon addition of ⁱPr₃SiOTf to 9, showing conversion to 12-NNSiⁱPr₃. (Red) Spectrum of 12-NNSiⁱPr₃ obtained by oxidation of [11-NNSiⁱPr₃]⁻ with [Cp*₂Co][PF₆]. (Orange) Spectrum of the oxidation of [11-NNSiMe₃]⁻ with [Cp*₂Co][PF₆] showing the formation of (P₂P^Ph)Fe(SiMe₃)(N₂) and (P₂P^Ph)Fe(NNSiMe₃). (Yellow) Spectrum of (P₂P^Ph)Fe(H)(N₂).²

We intuit that 9, or its further reduced state [(P₂P^Ph)Fe(N₂)]^2- 10, must be reached before nitrogen functionalization occurs via protonation or silylation. The iron centers in 9 and 10 are exposed, and are therefore susceptible to direct protonation at iron, or to facile migration from N-to-Fe. An N-protonated form of 9 (or 10) can presumably react further under the catalytic conditions to produce NH₃, when both excess acid and reductant are present. Such reactivity must be kinetically competitive with a step that produces an off-path hydride.

The need to access an anionic state of the system (either 9 or 10) before functionalization at N₂ can occur sets the requirement of a potent reductant for N₂RR in the P₂P^PhFe-system. The Fe^−I/0 couple of 9 is −2.47 V vs. Fc/Fc⁺, which is ~ 0.30 V more negative than the corresponding Fe^−I/0 couple for [(P₃^B)Fe(N₂)]^0/−; N₂RR can be driven rather efficiently with the latter system using Cp*₂Co paired with anilinium acids, which are ineffective with this precatalyst.^4,5

Conclusions

The present study highlights the detrimental effect of hydride ligands on an iron–catalyzed N₂RR model system whose efficiency is enhanced by irradiation. Stoichiometric reactivity as well as freeze-quench Mossbauer studies reveal that off-path (P₂P^Ph)Fe(N₂)_x(H)_y species are formed but are not inert resting states. On the contrary, they rapidly produce hydrogen in an HER cycle operating parallel to the desired N₂RR cycle. In the absence of light, these pathways compete with one another but operate along different cycles. Irradiation of (P₂P^Ph)Fe(N₂)(H)₂ 2 results in photoinduced H₂ elimination and the formation of (P₂P^Ph)Fe(N₂)₂ (5), which is significantly more competent for N₂RR. Thus, photolysis shifts the speciation from favoring an unproductive HER cycle to one where N₂RR becomes kinetically competitive.

A deeper understanding of the required driving force for N₂ functionalization is obtained by stoichiometric reactions with Fe–N₂ species. No protonation reactivity is observed with a strong acid for (P₂P^Ph)Fe(N₂)₂ at −78° C; further reduction is required before functionalization can take place. Protonation experiments with the Fe^−I and Fe^−II species 9 and 10 provide complex mixtures, but silylation experiments are informative.

The need to access an anionic or a dianionic state of the system before productive functionalization at N₂ occurs sets the low reduction potential required for N₂RR by this P₂P^PhFe-system and explains why comparatively milder reductants such as Cp*₂Co, which are effective for a related (P₃^B)Fe-catalyst system, are ineffective in the present case. Future catalyst designs for iron systems should focus on anodically shifting the needed redox couple to generate an Fe–N₂ species while maintaining a strongly activated N₂ ligand.