[2Fe–2S] Cluster Supported by Redox-Active o-Phenylenediamide Ligands and Its Application toward Dinitrogen Reduction

Abstract

As prevalent cofactors in living organisms, iron–sulfur clusters participate in not only the electron-transfer processes but also the biosynthesis of other cofactors. Many synthetic iron–sulfur clusters have been used in model studies, aiming to mimic their biological functions and to gain mechanistic insight into the related biological systems. The smallest [2Fe–2S] clusters are typically used for one-electron processes because of their limited capacity. Our group is interested in functionalizing small iron–sulfur clusters with redox-active ligands to enhance their electron storage capacity, because such functionalized clusters can potentially mediate multielectron chemical transformations. Herein we report the synthesis, structural characterization, and catalytic activity of a diferric [2Fe–2S] cluster functionalized with two o-phenylenediamide ligands. The electrochemical and chemical reductions of such a cluster revealed rich redox chemistry. The functionalized diferric cluster can store up to four electrons reversibly, where the first two reduction events are ligand-based and the remainder metal-based. The diferric [2Fe–2S] cluster displays catalytic activity toward silylation of dinitrogen, affording up to 88 equiv of the amine product per iron center.

Graphical Abstract

INTRODUCTION

Iron–sulfur clusters are ubiquitous and multipurpose cofactors in various living organisms.^1,2 As the simplest ones in the iron–sulfur cluster family, the [2Fe–2S] clusters have important biological functions. For example, the [2Fe–2S] clusters can combine to form [4Fe–4S] cubanes, which can couple further to generate clusters of even higher nuclearities.^3–5 Moreover, the [2Fe–2S] clusters, such as those in ferredoxin- and Rieske-type proteins, play a key role in one-electron-transfer processes,^1,2 in which the [2Fe–2S] clusters cycle between the oxidized diferric [2Fe–2S]²⁺ and mixed-valence [2Fe–2S]⁺ states.^6–9 The diferric [2Fe–2S]²⁺ cluster consists of two antiferromagnetically coupled iron centers, leading to an S = 0 ground state.^6–9 When the diferric cluster is reduced by one electron, the additional electron can be localized at one of the iron centers to give an S = ¹/₂ state or valence-delocalized to give an S = ¹/₂ or S = ⁹/₂ state.^6–9 The influence of the amino acid residues in the first and second coordination spheres on the redox potentials of these clusters has been studied using mutations.^10–14

Many biomimetic [2Fe–2S] clusters have been investigated in order to gain information on their electronic structures and electron-transfer processes.^15–34 A few biomimetic [2Fe–2S] clusters supported by N,N-bidentate ligands have emerged recently (Chart 1). Meyer and co-workers isolated and crystallographically characterized a series of [2Fe–2S]ⁿ⁺ clusters (n = 2, 1, 0) supported by two 2,2′-(phenylmethylene)bis(benzimidazolide) ligands.^25–27 Analogous [2Fe–2S] complexes have also shown proton-coupled electron-transfer reactivity.^28–31 Driess and Holland reported a series of [2Fe–2S]ⁿ⁺ (n = 2, 1, 0) clusters supported by β-diketiminate ligands.^32,33 Remarkably, the [2Fe–2S]⁺ clusters with diethylphenyl-substituted β-diketiminato ligands CH-[CMeN(2,6-Et₂C₆H₃)]₂ are extensively delocalized, as suggested by the ⁵⁷Fe Mössbauer and X-ray absorption spectroscopy/X-ray emission spectroscopy spectroscopic data and density functional theory calculations.³² Jones reported a diferric [2Fe–2S]²⁺ cluster supported by a guanidinato ligand.³⁴ No [2Fe–2S]ⁿ cluster supported by redox-active N,N-bidentate ligands has been reported to date.

Chart 1.

Examples of [2Fe–2S] Supported by N,N-Bidentate Ligands

Redox-active ligands have multiple energetically accessible oxidation states.^35–47 Their coordination to metal centers induces radical reactivity and electron reservoir behavior, which is often utilized to develop exciting new examples of catalysis.^35–54 The N,N-chelating o-phenylenediamide (pda) is a classic example of a redox-active ligand, whose coordination chemistry has been explored toward a large number of transition metals and main-group elements.^55–67 It has been clearly established that the pda ligand exists in three different redox states in coordination compounds: the closed-shell dianion, open-shell π-radical monoanion (S = ¹/₂), and closed-shell neutral form.^55–67 We have been investigating the coordination chemistry of the pda ligand toward transition metals and main-group elements.^64–67 Our group is interested in coordinating redox-active pda ligands to the diferric [2Fe–2S]²⁺ cluster. We envision that the resulting complex can store multiple electrons (i.e., reduction equivalents) and mediate multielectron redox processes. Herein we report the synthesis and rich redox chemistry of a diferric [2Fe–2S]²⁺ cluster supported by two monoanionic pda ligands. The combination of spectroscopic methods and X-ray crystallographic analysis was used to characterize the complexes in various oxidation states. The catalytic activity toward dinitrogen (N₂) reduction is also reported herein.

RESULTS AND DISCUSSION

The precursor [Fe(^Xylpda^•−)(η⁶-toluene)] [1; ^Xylpda = N,N′-bis(2,6-dimethylphenyl)-o-phenylenediamide] was prepared from N,N′-bis(2,6-dimethylphenyl)phenylenediamine⁶⁸ and Fe(HMDS)₂⁶⁹ (HMDS = hexamethyldisilazide) in toluene at room temperature. The spectroscopic data and metric parameters of 1 are similar to those of the previously reported Fe(^Dipppda^•−)(η⁶-toluene) [^Dipppda = N,N′-bis(2,6-diisopropylphenyl)-o-phenylenediamide].^66,67 The reaction of 1 with ¹/₈ equiv of S₈ in tetrahydrofuran (THF) at room temperature affords dark-purple complex 2 in 78% yield (Scheme 1). The ¹H NMR spectrum of complex 2 in THF-d₈ at 25 °C shows five paramagnetically broadened and shifted resonances between +14 and +4 ppm. The chemical shift span becomes narrower at lower temperatures (Figures S8 and S9). The molecular structure of 2 was established by X-ray diffraction, featuring a rhombic [Fe₂(μ-S)₂] core coordinated by two pda ligands (Figure 1). The metric parameters of the [Fe₂(μ-S)₂] core are similar to those of the synthetic and biological [2Fe–2S]²⁺ clusters in the literature.^15–34 The average C–N bond length of the ^Xylpda ligand is 1.341(4) Å (Table 1), between the typical C–N single and double bond lengths. The C_α–C_α and C_γ–C_γ bond lengths are 1.452(1) and 1.416(1) Å, respectively, while the other four bonds of the phenylene backbone feature the long C_α–C_β [avg. 1.418(2) Å] and short C_β–C_γ [avg. 1.358(7) Å] bonds. The pattern in these metric parameters suggests that the pda ligands are open-shell π-radical anions ^Xylpda^•− (S = ¹/₂). The intramolecular Fe–Fe distance [2.633(1) Å] is slightly longer than the sum of the covalent radii (2.50 Å).⁷⁰ The solution magnetic moment of 2 is 1.2 μ_B in THF-d₈ at 25 °C using the Evans method,⁷¹ close to an S_Total = 0 ground state. The observed magnetic moment is similar to those of the analogous literature compounds.^25,32 The zero-field Mössbauer spectrum of a solid sample of 2 recorded at 80 K shows one quadrupole doublet (Figure 3) with an isomer shift of δ = 0.23 mm s⁻¹ and a quadrupole splitting of |ΔE_Q| = 1.68 mm s⁻¹ (Table 2), confirming the presence of equivalent high-spin Fe³⁺ centers. These values are similar to those observed for other synthetic and biological diferric [2Fe–2S]²⁺ clusters.^15–34 The cyclic voltammetry experiment of 2 in THF at 23 °C revealed three reversible reduction events at −0.98, −1.39, and −2.42 V (with respect to Fc⁺/Fc), respectively (Figure 2). In addition, there is a reduction event with limited chemical reversibility at −3.31 V (with respect to Fc⁺/Fc), which suggests that the four-electron reduction of 2 leads to a species that has limited stability. The cyclic voltammogram of 2 shows that the cluster can store up to four electrons and that three of the reduced clusters might be stable enough for isolation.

Scheme 1.

Synthesis of 2

Figure 1.

X-ray structure of 2. The thermal ellipsoids are shown at 30% probability. Hydrogen atoms are omitted for clarity.

Table 1.

Selected Bond Distances and Angles of Complexes 2–6

	2	3	4	5	6
Fe–N	1.973(4)	1.954(3)	1.955(5)	1.978(4)
	1.959(3)	1.944(4)	1.965(4)	1.992(4)	2.092(2)
	1.973(3)	1.994(3)	1.962(4)	2.049(5)	2.060(2)
	1.952(5)	1.977(3)	1.954(4)	2.033(4)
N–Cα	1.338(6)	1.385(5)	1.403(7)	1.380(7)
	1.344(7)	1.394(5)	1.385(8)	1.375(7)	1.387(2)
	1.336(7)	1.343(5)	1.396(6)	1.387(7)	1.378(3)
	1.344(5)	1.345(5)	1.392(6)	1.387(7)
Cα–Cα	1.453(6)	1.417(5)	1.429(7)	1.445(7)	1.432(3)
	1.451(6)	1.432(5)	1.423(8)	1.428(8)
Cα–Cβ	1.420(8)	1.400(6)	1.386(9)	1.384(7)
	1.420(6)	1.397(6)	1.390(8)	1.391(7)	1.397(3)
	1.417(5)	1.415(5)	1.387(7)	1.386(8)	1.400(3)
	1.414(7)	1.428(6)	1.396(7)	1.388(8)
Cβ–Cγ	1.353(6)	1.387(7)	1.408(8)	1.415(8)
	1.356(8)	1.403(6)	1.402(9)	1.395(8)	1.399(3)
	1.370(7)	1.372(5)	1.401(7)	1.388(8)	1.395(3)
	1.353(5)	1.357(6)	1.401(7)	1.395(8)
Cγ–Cγ	1.416(6)	1.370(7)	1.379(8)	1.378(8)	1.377(3)
	1.415(6)	1.406(5)	1.383(8)	1.374(9)
Fe–Fe	2.633(1)	2.6865(8)	2.764(1)	2.709(1)	2.7306(6)
			2.7486(9)
Fe–S	2.186(1)		2.251(2)	2.264(2)
	2.191(2)	2.243(1)	2.253(1)	2.207(2)	2.2804(7)
	2.184(2)	2.238(1)	2.254(2)	2.266(2)	2.2819(6)
	2.196(1)		2.245(2)	2.254(2)
∠S–Fe–S	105.98(5)	102.93(4)	104.30(6)	106.61(6)	106.47(2)
	105.91(5)		104.69(6)		104.94(6)

Figure 3.

Zero-field ⁵⁷Fe Mössbauer spectrum of 2–6 in the solid state at 80 K.

Table 2.

Solution Magnetic Moment at 298 K and Solid-State Zero-Field Mössbauer Data at 80 K

complex	μ eff a	S Total b	S Fe c	δ,c mm s−1	|ΔEQ|, mm s−1
1		0	1/2	0.38	0.78
2	1.2d	0	5/2	0.23	1.68
3	1.9	1/2	5/2	0.24	1.60
4	1.6d	0	5/2	0.26	1.68
5	2.3	1/2	5/2	0.33	1.84
			2	0.62	2.84
6	1.7d	0	2	0.70	2.61

^aSolution magnetic moment by the Evans method.

^bSpin state of the molecule.

^cIntrinsic spin state of the iron center.

^dThe residual magnetic moment data are similar to those of the analogous literature compounds;^25,32 the ¹H NMR chemical shift span becomes narrower at lower temperatures (Figures S8–S13).

Figure 2.

Cyclic voltammogram of 2 (2 mM in THF/0.2 M TBAPF₆) versus Fc⁺/Fc at a scan rate of 100 mV s⁻¹ (top) and the corresponding redox couples (bottom).

Reducing 2 using 1 equiv of KC₈ gives the new [2Fe–2S] cluster 3 as a dark-purple crystalline solid in 73% yield (Scheme 2). In the crystal lattice of 3, the complex anion [Fe₂(μ-S)₂(^Xylpda)₂]⁻ and [K(THF)₆]⁺ cation do not display any close contacts (Figure 4). One of the ^Xylpda ligands resembles those in complex 2 [C_α–N bond lengths, 1.343(5) and 1.345(5) Å; Table 1], while the other displays longer C_α–N distances [1.385(5) and 1.394(5) Å]. These metric parameters suggest that the former is closer to ^Xylpda^•− and the latter is closer to ^Xylpda²⁻; i.e., the reduction occurs at one of the ^Xylpda ligands in 2. The Fe–S distances [avg. 2.241(3) Å in 3 vs avg. 2.189(4) Å in 2] and Fe–Fe distances [2.6865(8) Å in 3 vs 2.633(1) Å in 2] are slightly longer than those of 2, respectively. The magnetic susceptibility measurements of 3 in THF-d₈ using the Evans method⁷¹ revealed a magnetic moment of 1.9 μ_B at 298 K (Table 2). The zero-field Mössbauer spectrum of 3 displays only one quadrupole doublet with an isomer shift of 0.24 mm s⁻¹ and a quadrupole splitting of 1.60 mm s⁻¹ (Figure 3), suggesting that the two iron centers are indistinguishable under the experimental conditions. The similarity between the Mössbauer parameters of 2 and 3 suggests that the oxidation states of the iron centers remain unchanged upon one-electron reduction of 2, consistent with the ligand-based reduction indicated by the metric parameters of the crystal structures.

Scheme 2.

Syntheses of 3–6

Figure 4.

Crystal structure of 3 with the thermal ellipsoids plotted at the 50% probability level. The [K(THF)₆]⁺ counterion and all hydrogen atoms are omitted for clarity.

Treating 2 with 2 equiv of KC₈ leads to the formation of cluster 4 as a dark-purple crystalline solid in 76% yield (Scheme 2). As shown in Figure 5, each molecule of 4 possesses a crystallographically imposed inversion center. The two potassium ions in each molecule are sandwiched between two xylyl groups of the pda ligands in a double η⁶ fashion and further coordinated by a bridging sulfide and a THF molecule. The ^Xylpda ligands display long C_α–N distances [1.385(8)–1.403(7) Å; Table 1] consistent with ^Xylpda²⁻, suggesting that the second reduction events for 2 are also ligand-based. The Fe–Fe distances in 4 [avg. of the two independent molecules: 2.756(8) Å] are slightly longer than those in 2 and 3 [2.6865(8) Å in 3 and 2.633(1) Å in 2], while the Fe–S distances are similar among all three compounds [avg. 2.251(3) Å in 4 vs avg. 2.241(3) Å in 3 and avg. 2.189(4) Å in 2]. The observed solution magnetic moment of 1.6 μ_B (Table 2) for 4 at 298 K in THF-d₈ suggests a strong antiferromagnetic coupling. The Mössbauer spectra of 4 display one quadrupole doublet (δ = 0.26 mm s⁻¹ and |ΔE_Q| = 1.68 mm s⁻¹; Figure 3), similar to those of 2 and 3. All results are consistent with ligand-based reduction, in which the metal oxidation states remain unchanged.

Figure 5.

Crystal structure of 4 with the thermal ellipsoids plotted at the 50% probability level. All hydrogen atoms are omitted for clarity.

Complex 4 can be further reduced by 1 equiv of KC₈, affording a dark-brown [2Fe–2S]⁺ cluster 5 in 81% yield (Scheme 2). In the crystal structure of 5 (Figure 6), the [Fe₂(μ-S)₂(^Xylpda)₂]³⁻ fragment is associated with three potassium cations to balance the charge. One of the three potassium ions (K3) is coordinated by a bridging sulfide (S2), three THF molecules, and the diamide moiety (N3–C23–C28–N4) of one ^Xylpda ligand in an η⁴ fashion. The other two potassium ions (K1 and K2) are bound to the other bridging sulfide (S1) with a K1–S1–K2 angle of 175.34(6)°. In addition, K1 is sandwiched between two xylyl groups of the pda ligands in a double η⁶ fashion and coordinated by an additional THF molecule, whereas K2 is sandwiched between an η⁶-xylyl group and an η²-xylyl group and further coordinated by a THF molecule. Furthermore, K2 and C3 in each molecule of 5 are bonded to two adjacent molecules of 5 through C3–K2 interactions, leading to the formation of a 1D coordination polymer in the crystal lattice (Figure S24). The C_α–N bond distances of the ^Xylpda ligands are similar to those of 4 (Table 1), consistent with iron-based reduction. The Fe–S distances in 5 [avg. 2.248(2) Å] are similar to those in 4 [avg. 2.251(2) Å], whereas the Fe–Fe distance shortens slightly upon going from 4 [avg. 2.756(8) Å] to 5 [avg. 2.709(1) Å]. Two quadrupole doublets are observed in the Mössbauer spectrum of 5 (δ = 0.33 mm s⁻¹ and |ΔE_Q| = 1.84 mm s⁻¹; δ = 0.62 mm s⁻¹ and |ΔE_Q| = 2.84 mm s⁻¹; Figure 3), which suggests a mixed-valence Fe³⁺–Fe²⁺ cluster with two different iron environments, i.e., no delocalization on the Mössbauer time scale.

Figure 6.

X-ray structure of one repeating unit of 5, with the thermal ellipsoids plotted at the 50% probability level. The minor component of the two-site disordering and all hydrogen atoms are omitted for clarity.

Although the cyclic voltammetry experiment suggests that the diferrous cluster from a further reduction of 5 may have limited chemical stability, we next attempted to prepare and isolate the diferrous cluster for structural characterization. Reducing 4 using 2 equiv of KC₈ gives the orange diferrous [2Fe–2S]⁰ cluster 6 in 62% yield (Scheme 2). As shown in Figure 7, X-ray crystallographic analysis shows that, unlike 5, compound 6 has a nonpolymeric structure and is formulated as K₄(THF)₆Fe₂(μ-S)₂(^Xylpda)₂ in the solid state. There is a crystallographically imposed 2-fold rotation axis going through the two bridging sulfur atoms. Each bridging sulfide of the [Fe₂(μ-S)₂(^Xylpda)₂]⁴⁻ unit is further bonded to a pair of symmetry related potassium cations. Each of the potassium ions bound to S1 is further coordinated by one THF molecule and the xylyl rings of the pda ligands in a double η⁶ fashion, whereas each of the potassium ions bound to S2 is further coordinated by two THF molecules and the N–C–C–N moiety of a pda ligand. The C_α–N bond lengths of the ^Xylpda ligands display negligible changes compared to those in 4 (Table 1), consistent with iron-based reduction. The Fe–S distance in 6 [avg. 2.2812(7) Å] is slightly longer than those in 4 and 5, while the Fe–Fe distance in 6 [2.7306(6) Å] is longer than that in 5 and shorter than that in 4. Compound 6 slowly decomposes in solution but is stable in the solid state for weeks at −35 °C. During the crystallization of 6, we observed a few crystals of [Fe(^Xylpda)]₂ (7; Figure S25), which might result from the decomposition of 6. The zero-field Mössbauer spectrum of 6 shows one quadrupole doublet with δ = 0.70 mm s⁻¹ and |ΔE_Q| = 2.61 mm s⁻¹ (Figure 3), indicating the presence of two equivalent high-spin Fe²⁺ sites. The Mössbauer parameters of 5 and 6 are close to those observed for the reduced and superreduced biological and synthetic [2Fe–2S] clusters.^15–34

Figure 7.

Crystal structure of 6 with the thermal ellipsoids plotted at the 50% probability level. All hydrogen atoms are omitted for clarity.

Prompted by compound 2’s capability of storing up to four electrons, we assessed the catalytic activity of compound 2 toward a multielectron process, namely, N₂ reduction. The results are summarized in Table 3. The reaction with 500 equiv of KC₈ and Me₃SiCl per iron center in THF generated 17 ± 2 equiv of N(SiMe₃)₃ per iron center in 24 h (entry 1). The reactions in dimethyl ether (DME) and dioxane under otherwise identical conditions gave 31 ± 1 and 32 ± 2 equiv of N(SiMe₃)₃ per iron center, respectively (entries 2 and 3). With 1000 equiv of KC₈ and Me₃SiCl per iron center, the reaction in DME gave 36 ± 3 equiv of N(SiMe₃)₃ per iron center (entry 4), whereas using dioxane as the solvent increased the amount of N(SiMe₃)₃ to 51 ± 2 equiv per iron center (entry 5). When the amounts of KC₈ and Me₃SiCl are increased to 1500 equiv per ion center, the reaction in dioxane produced 60 ± 2 and 88 ± 2 equiv of N(SiMe₃)₃ per iron center in 24 and 72 h, respectively (entries 6 and 7). Compared to the iron-based N₂ silylation catalysts in the literature,^72–83 the catalytic performance of 2 is moderate. Several iron–sulfur clusters have been used in N₂ reduction chemistry.⁸⁴ Our study reported herein, however, represents the first example using a [2Fe–2S] cluster in N₂ reduction. Last, we examined the catalytic performance of complex 1. The reaction with 1000 equiv of KC₈ and Me₃SiCl per iron center in dioxane generated 28 ± 2 equiv of N(SiMe₃)₃ per iron center in 24 h (entry 8). The lower catalytic activity of compound 1 (entry 8 vs entry 5) suggests that the sulfide ligands play an important role in the catalytic reaction. Although the exact role of the sulfide ligands is not well understood, we speculate that they may enhance the stability of the iron-containing species under the reaction conditions. To probe the nature (i.e., homogeneous vs heterogeneous) of the catalysis, we conducted a filtration test, whereby the soluble and insoluble fractions are separated by filtration and then independently tested for catalytic activity (see the Supporting Information for the detailed procedure). The filtrate displayed much higher catalytic activity than the insoluble fraction, suggesting that the active species is soluble. However, the formation of catalytically active small iron nanoparticles cannot be ruled out.

Table 3.

Catalytic N ₂ Silylation with Me ₃ SiCl and KC ₈ Using 2 ^a

entry	solvent	KC8 and Me3SiCl (equiv per iron center)	N(SiMe3)3 (equiv per iron center)
1	THF	500	17 ± 2
2	DME	500	31 ± 1
3	dioxane	500	32 ± 2
4	DME	1000	36 ± 3
5	dioxane	1000	51 ± 2
6	dioxane	1500	60 ± 2
7b	dioxane	1500	88 ± 2
8c	dioxane	1000	28 ± 2

^aN(SiMe₃)₃ was first converted to NH₄⁺ and then quantified via ¹H NMR experiments with 1,3,5-trimethoxybenzene as the internal standard.

^b72 h.

^c1 was used as the catalyst.

CONCLUSION

We have synthesized a diferric [2Fe–2S] cluster (2) supported by two redox-active ^Xylpda ligands from the oxidation of [Fe(^Xylpda)(toluene)] with S₈. The cyclic voltammetry studies of the diferric complex [Fe₂S₂(^Xylpda)₂] (2) show its rich redox chemistry. Chemical reductions were used to synthesize [Fe₂S₂(^Xylpda)₂]ⁿ in four additional oxidation states (n = 4–, 3–, 2–, 1–). The combination of crystallographic and spectroscopic studies revealed that the first two reduction events are ligand-based, whereas the third and fourth reduction events are metal-based. The third reduction afforded a mixed-valence Fe³⁺Fe²⁺ cluster, where the electron is localized. Finally, the fourth reduction event gave a superreduced diferrous cluster, which showed limited stability. We have also further demonstrated the catalytic activity of 2 toward the catalytic silylation of N₂ using Me₃SiCl and KC₈, producing up to 88 ± 2 equiv of N(SiMe₃)₃ per iron center. Unfortunately, we were unable to observe N₂ binding on any of the [2Fe–2S] species prepared herein. Investigations into a further reduction of the diferrous cluster and the N₂ reduction mechanism are underway in our laboratories.

EXPERIMENTAL SECTION

Materials and Methods.

All reactions were carried out in a N₂-filled glovebox or using standard Schlenk techniques under N₂. Glassware was dried in a 180 °C oven overnight. Diethyl ether, hexanes, pentane, and toluene solvents were dried by refluxing and distilling over sodium under N₂. The THF solvent was dried by refluxing and distilling over sodium benzophenone ketyl under N₂. C₆D₆ and THF-d₈ were degassed through three consecutive freeze–pump–thaw cycles. All solvents were stored over 3 Å molecular sieves prior to use. Unless otherwise noted, all NMR spectra were recorded on an Agilent DD2 600 MHz spectrometer at 25 °C. Chemical shifts are referenced to the solvent signals. Solution magnetic moments were measured at 25 °C by the method originally described by Evans⁷¹ with stock and experimental solutions containing a known amount of a cyclohexane standard. Elemental analyses were carried out by ANALEST at the University of Toronto. Cyclic voltammograms were recorded at 295 K with a 600E Series electrochemical analyzer/workstation. Unless otherwise noted, all chemicals were purchased from commercial sources and used as received. The ¹H NMR data for 3–6, where no coupling was resolved because of paramagnetic broadening, are presented in the following format: chemical shift (the peak width at half-height in hertz, integration value, and partial assignment based on integration). Although the crystal structures of 3–6 contain several coordinating and/or lattice THF molecules, the loss of THF molecules occurs under vacuum based on the ¹H NMR and elemental analysis results. Therefore, the number of moles and percentage yields of these compounds were calculated using the formula provided in the elemental analysis section.

⁵⁷Fe Mössbauer Spectroscopy.

All measurements for ⁵⁷Fe Mössbauer spectroscopy were performed using nonenriched solids of the as-isolated complexes. All samples were prepared in an inert-atmosphere glovebox equipped with a liquid-nitrogen fill port to enable sample freezing to 77 K within the glovebox. Each sample was loaded into a Delrin Mössbauer sample cup for measurements and loaded under liquid nitrogen. Low-temperature ⁵⁷Fe Mössbauer measurements were performed using a See Co. MS4 Mössbauer spectrometer integrated with a Janis SVT-400T He/N₂ cryostat for measurements at 80 K. Isomer shifts were determined relative to α-iron at 298 K. All Mössbauer spectra were fit using the program WMoss (See Co). Errors of the fit analyses were the following: δ ± 0.02 mm s⁻¹ and ΔE_Q ± 3%. For multicomponent fits, the quantitation errors were ±3% (e.g., 50 ± 3%).

1. To a solid mixture of N,N′-bis(2,6-dimethylphenyl)-phenylenediamine (1.58 g, 3.00 mmol) and Fe(HMDS)₂ (1.13 g, 3.00 mmol) was added 15 mL of toluene. The reaction mixture was stirred at room temperature overnight, yielding a red-purple solution, and then filtered through Celite. The filtrate was concentrated to dryness to afford a metallic green crystalline solid of 1 (1.28 g, 92%). Crystals suitable for X-ray crystallography were obtained by cooling a concentrated diethyl ether solution at −35 °C. ¹H NMR (600 MHz, C₆D₆): δ 7.31 (m, 6H, overlapping, Xyl-H), 6.74 (dd, J = 6.5 and 3.2 Hz, 2H, Ar-H), 6.51 (dd, J = 6.3 and 3.3 Hz, 2H, Ar-H), 5.11 (td, J = 5.8 and 0.9 Hz, 1H, η⁶-tol-H), 4.86 (t, J = 6.0 Hz, 2H, η⁶-tol-H), 4.78 (d, J = 6.1 Hz, 2H, η⁶-tol-H), 2.17 (s, 12H, Xy-CH₃), 1.74 (s, 3H, η⁶-tol-CH₃). ¹³C NMR (151 MHz, C₆D₆): δ 156.18 (Xyl-C), 149.80 (Ar-C), 132.77 (Xyl-C), 128.50 (Xyl-C), 125.43 (Xyl-C), 119.73 (Ar-C), 111.95 (Ar-C), 95.31 (η⁶-tol-C), 83.17 (η⁶-tol-C), 83.10 (η⁶-tol-C), 82.04 (η⁶-tol-C), 19.33 (η⁶-tol-CH₃), 17.99 (Xy-CH₃). Anal. Calcd for C₂₉H₃₀N₂Fe: C, 75.33; H, 6.54; N, 6.06. Found: C, 74.33; H, 6.64; N, 6.20.

2. To a solution of 1 (924.8 mg, 2.00 mmol) in 10 mL of THF was added S₈ (64.1 mg, 0.25 mmol) as a solid. The reaction mixture was stirred at room temperature overnight and then filtered through Celite. Volatiles were removed under vacuum, leaving a black-purple solid. The solid was stirred with 5 mL of hexanes for 1 h and then collected on a fritted funnel, which was washed with hexanes (3 × 1 mL) and dried under a high vacuum (628.6 mg, 78%). Crystals suitable for X-ray crystallography were obtained by cooling a concentrated diethyl ether solution at −35 °C. ¹H NMR (600 MHz, THF-d₈): δ 13.99 (d, J = 6.4 Hz, 4H, Ar-H), 7.45 (d, J = 7.3 Hz, 8H, Xyl-H), 7.01 (m, 4H, Ar-H), 4.68 (s, 24H, CH₃), 4.00 (t, J = 7.2 Hz, 4H, Xyl-H). The Evans method (298 K, THF-d₈): μ_eff = 1.2 μ_B. Anal. Calcd for C₄₄H₄₄N₄S₂Fe₂: C, 65.68; H, 5.51; N, 6.96. Found: C, 65.14; H, 5.50; N, 6.86.

3. To a solution of 2 (201.2 mg, 0.25 mmol) in 5 mL of THF was added a suspension of KC₈ (33.8 mg, 0.25 mmol) in 3 mL of THF. The reaction mixture was stirred for 1 h at room temperature and then filtered through Celite. The filtrate was concentrated to ~2 mL, top-layered with 5 mL of pentane, and cooled to −35 °C overnight to yield dark-purple crystals that were suitable for X-ray crystallography. The supernatant was decanted off, and the solid was washed with diethyl ether (3 × 1 mL) and pentane (3 × 1 mL) and then dried under vacuum to afford 3 (193.1 mg, 73%). ¹H NMR (600 MHz, THF-d₈): δ 16.37 (Δ_1/2 = 18.71 Hz, 8H, Xyl-H), 7.61 (Δ_1/2 = 87.47 Hz, 24H, CH₃), −1.44 (Δ_1/2 = 19.56 Hz, 4H), −16.47 (Δ_1/2 = 62.60 Hz, 4H), −22.99 (Δ_1/2 = 90.67 Hz, 4H). The Evans method (298 K, THF-d₈): μ_eff = 1.9 μ_B. Anal. Calcd for C₄₄H₄₄N₄S₂Fe₂K·(C₄H₈O)₃: C, 63.45; H, 6.47; N, 5.29. Found: C, 63.39; H, 6.24; N, 5.23.

4. To a solution of 2 (402.4 mg, 0.50 mmol) in 5 mL of THF was added a suspension of KC₈ (135.2 mg, 1.00 mmol) in 3 mL of THF. The reaction mixture was stirred for 1 h at room temperature and then filtered through Celite. The filtrate was concentrated to ~2 mL, top-layered with 5 mL of pentane, and cooled to −35 °C overnight to yield dark-purple crystals that were suitable for X-ray crystallography. The supernatant was decanted off, and the solid was washed with diethyl ether (3 × 1 mL) and pentane (3 × 1 mL) and then dried under vacuum to afford 4 (363.2 mg, 76%). ¹H NMR (600 MHz, THF-d₈): δ 11.90 (Δ_1/2 = 18.23 Hz, 8H, Xyl-H), 7.25 (Δ_1/2 = 91.02 Hz, 24H, CH₃), 0.97 (Δ_1/2 = 59.72 Hz, 4H), −0.22 (Δ_1/2 = 19.42 Hz, 4H), −2.50 (Δ_1/2 = 59.59 Hz, 4H). The Evans method (298 K, THF-d₈): μ_eff = 1.6 μ_B. Anal. Calcd for C₄₄H₄₄N₄S₂Fe₂K₂·C₄H₈O: C, 60.37; H, 5.49; N, 5.87. Found: C, 60.05; H, 5.52; N, 5.84.

5. To a solution of 4 (238.7 mg, 0.25 mmol) in 5 mL of THF was added a suspension of KC₈ (33.8 mg, 0.25 mmol) in 3 mL of THF. The reaction mixture was stirred for 1 h at room temperature, yielding a brown solution, and then filtered through Celite. The filtrate was concentrated to ~2 mL, top-layered with 5 mL of pentane, and cooled to −35 °C overnight to yield brown crystals that were suitable for X-ray crystallography. The supernatant was decanted off, and the solid was washed with diethyl ether (3 × 1 mL) and pentane (3 × 1 mL) and then dried under vacuum to afford 5·THF (201.6 mg, 81%). ¹H NMR (600 MHz, THF-d₈): δ 12.24 (Δ_1/2 = 111.5 Hz, 8H), 8.10 (Δ_1/2 = 1229.87 Hz, 24H, CH₃), −0.10 (Δ_1/2 = 164.90 Hz, 4H), −1.31 (Δ_1/2 = 92.42 Hz, 4H). The Evans method (298 K, THF-d₈): μ_eff = 2.3 μ_B. Anal. Calcd for C₄₄H₄₄N₄S₂Fe₂K₃·C₄H₈O: C, 58.00; H, 5.27; N, 5.64. Found: C, 57.90; H, 5.02; N, 5.75.

6. To a solution of 4 (238.7 mg, 0.25 mmol) in 5 mL of THF was added a suspension of KC₈ (67.6 mg, 0.50 mmol) in 3 mL of THF. The reaction mixture was stirred for 1 h at room temperature, yielding an orange–brown solution, and then filtered through Celite. The filtrate was concentrated to ~2 mL, top-layered with 5 mL of pentane, and cooled to −35 °C overnight to yield orange crystals that were suitable for X-ray crystallography. The supernatant was decanted off, and the solid was washed with diethyl ether (3 × 1 mL) and pentane (3 × 1 mL) and then dried under vacuum to afford 6 (205.8 mg, 62%). ¹H NMR (600 MHz, THF-d₈): δ 7.87 (Δ_1/2 = 51.34 Hz, 4H), 7.25 (Δ_1/2 = 131.14 Hz, 12H, CH₃), 6.77 (Δ_1/2 = 20.82 Hz, 4H), 6.43 (Δ_1/2 = 7.47 Hz, 4H), 6.20 (Δ_1/2 = 92.92 Hz, 4H), 1.98 (Δ_1/2 = 183.01 Hz, 12H, CH₃), 0.50 (Δ_1/2 = 15.10 Hz, 4H). The Evans method (298 K, THF-d₈): μ_eff = 1.7 μ_B. Because of the limited stability and high air sensitivity of complex 6, we were unable to obtain satisfactory elemental analysis results. The best elemental analysis results obtained so far are provided below. Anal. Calcd for C₄₄H₄₄N₄S₂Fe₂K₄·(C₄H₈O)₅: C, 58.16; H, 6.41; N, 4.24. Found: C, 56.55; H, 6.37; N, 3.87.

Standard Procedure for the Catalytic Conversion of N₂ to N(TMS)₃ Using 2.

KC₈ (90.0 mg, 666 μmol) was suspended in dioxane (1.8 mL) in a 2-dram vial equipped with a stir bar. Me₃SiCl (85.0 μL, 666 μmol) was added, followed by a solution of 2 in dioxane (200 μL, 3.3 mM, 0.66 μmol). The reaction mixture was stirred at room temperature for 24 h and then filtered through a plug of Celite. To the filtrate was added a solution of HCl in Et₂O (2 M, 5 mL). The resulting mixture was stirred for 1 h before removal of the solvent under reduced pressure. The resulting white solids were dissolved in DMSO-d₆ with 1,3,5-trimethoxybenzene as an internal standard. The ammonium was quantified using ¹H NMR spectroscopy. The experiments were performed in triplicate. The control experiment was performed in the absence of 2 under otherwise identical conditions. The control experiment gave only 3 equiv of N(TMS)₃ with respect to the 1 equiv of 2 that is absent (i.e., 1.5 equiv per iron center that is absent).