Sterically Stabilized Terminal Hydride of a Diiron Dithiolate

Abstract

The kinetically robust hydride [t-HFe₂(Me₂pdt)(CO)₂(dppv)₂]⁺ ([t-H1]⁺) (Me₂pdt²⁻ = Me₂C-(CH₂S⁻)₂; dppv = cis-1,2-C₂H₂(PPh₂)₂) and related derivatives were prepared with ⁵⁷Fe enrichment for characterization by NMR, FT-IR, and NRVS. The experimental results were rationalized using DFT molecular modeling and spectral simulations. The spectroscopic analysis was aimed at supporting assignments of Fe–H vibrational spectra as they relate to recent measurements on [FeFe]-hydrogenase enzymes. The combination of bulky Me₂pdt²⁻ and dppv ligands stabilizes the terminal hydride with respect to its isomerization to the 5–16 kcal/mol more stable bridging hydride ([μ-H1]⁺) with t_1/2(313.3 K) = 19.3 min. In agreement with the nOe experiments, the calculations predict that one methyl group in [t-H1]⁺ interacts with the hydride with a computed CH⋯HFe distance of 1.7 Å. Although [t-H⁵⁷1]⁺ exhibits multiple NRVS features in the 720–800 cm⁻¹ region containing the bending Fe–H modes, the deuterated [t-D⁵⁷1]⁺ sample exhibits a unique Fe-D/CO band at ∼600 cm⁻¹. In contrast, the NRVS spectra for [μ-H⁵⁷1]⁺ exhibit weaker bands near 670–700 cm⁻¹ produced by the Fe–H–Fe wagging modes coupled to Me₂pdt²⁻ and dppv motions.

Graphical abstract

INTRODUCTION

Currently the hydrogenases (H₂ases) are receiving much interest because these enzymes exhibit unusual and highly efficient pathways for activating and producing H₂.^1–3 It has been long assumed that all hydrogenases operate via the intermediacy of metal hydrides,⁴ which are otherwise elusive in nature.^5–8 Such intermediates have long been directly detected for the [NiFe]-H₂ases.⁹ For the [FeFe]-H₂ases, hydrides were initially spectroscopically implicated by the effect of H/D exchange on ν_CO bands in the FT-IR spectra.^10,11 The terminal hydrides have only recently been detected directly, first in mutant proteins^12–14 and more recently in two native proteins (Scheme 1).¹⁵

Scheme 1. Proposed Catalytic Cycle of [FeFe]-H₂ase Enzymes Including All States That Have Been Identified in Wild-Type Enzymes and the Proposed H₂ Adduct^a,16.

^aTerminal CO/CN ligands are omitted for clarity.

For the investigation of iron hydrides, nuclear resonance vibrational spectroscopy (NRVS) is the premier technique^14,17 and, in combination with density functional theory (DFT), was central in our recent direct detection of Fe–H-containing intermediates in the [FeFe]-H₂ases.^14,15

Biomimetic terminal hydrides participate in two reactions that complicate their analysis. When the dithiolate is the azadithiolate (adt) cofactor (or its N-substituted derivatives), the hydride can be reversibly deprotonated by amine. This effect is illustrated by the case of [term-HFe₂(adt)-(CO)₂(dppv)₂]⁺ (adt = azadithiolate), where the tautomerization is affected by the nature and concentration of the counteranion.¹⁸ Regardless of the nature of the dithiolate bridge, terminal Fe–H hydrides tend to isomerize irreversibly to bridging Fe–H–Fe hydrides.¹ We sought to suppress this reaction, since these bridging hydrides are not relevant to the biological function of [FeFe]-H₂ases.¹⁹

Many studies have been reported on diiron dithiolates with “rotated structures”. The rotated motif is readily achieved with mixed-valence Fe(II)Fe(I) complexes, but Fe(I)Fe(I) compounds almost invariably adopt the C_2v (idealized) structure (Scheme 2). Fe(I)Fe(I) derivatives can also be stabilized in a rotated geometry using bulky dithiolates such as R₂C(CH₂S⁻)₂ (R ≠ H).^20–23 In this work, we extend that concept, showing that Me₂C(CH₂S⁻)₂ also stabilizes terminal hydride species, greatly slowing their isomerization to bridging hydrides. This approach worked so well that we proceeded to collect NRVS data on both the terminal and bridging hydrides. These results are relevant to benchmarking recent NRVS results on the enzyme.¹⁵

Scheme 2. Compounds of the Type Fe₂[(SCH₂)₂X]L₆ Normally Existing as Symmetrical Isomers, with Idealized C_2v Structures^a.

^aThey distort to rotated structures upon oxidation or, in some cases, with large ligands and bulky dithiolates even in the absence of redox.

A few comments on “flippamers” are required, since this kind of isomerism occurs in compounds discussed in this paper. Flippamers are conformers of the triatomic backbone of 1,3-propanedithiolate and related X(CH₂S⁻)₂ bridging ligands in complexes of the type μ-(μ-X(CH₂S)₂)[FeL₃]₂. The population of the two flippamers depends on the steric similarity of two FeL₃ modules. The flipping rate is fast on the NMR time scale but slow on the EPR and IR time scales.²⁴

RESULTS AND DISCUSSION

Characterization of Fe₂(Me₂pdt)(CO)₂(dppv)₂ Using FT-IR and NMR Spectroscopy

UV irradiation of a solution of Fe₂(Me₂pdt)(CO)₄(dppv)²⁰ and dppv efficiently gave Fe₂(Me₂pdt)(CO)₂(dppv)₂ ([1]⁰). This compound was obtained as a dark green solid that is mildly sensitive to air and exhibits good solubility in several organic solvents. The FT-IR spectrum revealed intense ν_CO bands at 1898 and 1865 cm⁻¹ that match well with those (1888, 1868 cm⁻¹) for Fe₂(pdt)-(CO)₂(dppv)₂ (Figure S2).²⁵

The ¹H NMR spectrum of [1]⁰ shows a methylene signal at δ 1.23, which also matches that for Fe₂(pdt)(CO)₂(dppv)₂. The methyl resonance, a broad singlet, is found at δ −0.15 (Figure S3). In contrast, the two Me resonances from the apical/basal and basal/basal isomers of Fe₂(Me₂pdt)(CO)₄(dppv) absorb at δ 0.98 and 1.08, and the Me resonances of Fe₂(Me₂pdt)(CO)₆ absorb at δ 0.98.²⁰ The upfield shift for the methyl ¹H NMR signal [1]⁰ is attributed to ring current effects resulting from interactions between the methyl groups and the phenyl groups of the ligand.

The room-temperature ³¹P NMR spectrum of [1]⁰ consists of a singlet at δ 87.0 (Figure S4). At lower temperatures, this signal splits into two singlets, and below −90 °C these singlets broaden further (Figure S6). However, the signal for the methyl groups remains coalesced even at −90 °C (Figure S5). The chemical shift difference for the ³¹P NMR signals in the slow exchange limit is δ 13.4, corresponding to 2707 Hz (202 MHz observing frequency). The estimated ¹H chemical shift difference between unique methyl signals is 2 ppm, corresponding to 1000 Hz (500 MHz). Therefore, the turnstile rotation of the two Fe(CO)(dppv) centers is slower than the flipping of the [Me₂pdt]²⁻ ligand.

Structure of Fe₂(Me₂pdt)(CO)₂(dppv)₂

The structure of [1]⁰, which crystallizes with a disordered molecule of CH₂Cl₂, was confirmed by X-ray crystallography. The overall structure was similar to that of many Fe₂(SR)₂L₆ compounds with an Fe–Fe bond distance of 2.594(1) Å (Figure 1).¹ The crystallographic analysis reveals that the coordination sphere of one Fe is highly distorted. In most compounds of the type Fe₂(SR)₂L₆, terminal ligands on the two Fe centers are eclipsed (i.e., torsion angle of 0°), whereas in [1]⁰ the P_apical–Fe–Fe–P_apical torsion angle is 66.43°.²⁶ In the related diiron complex Fe₂(Me₂pdt)(CO)₄(PMe₃)₂, the P_apical–Fe–Fe–P_apical torsion angle is 28.9°,²¹ the smaller angle reflecting the diminished bulk of PMe₃. Since the ν_CO values for [1]⁰ and Fe₂(pdt)-(CO)₂(dppv)₂ match well,²⁵ the distortion observed in the solid state appears not to predominate in solution.

Figure 1.

Solid-state structure of Fe₂(Me₂pdt)(CO)₂(dppv)₂ ([1]⁰). Disorder in the phenyl rings and solvent is removed for clarity.

Synthesis of Labeled ⁵⁷FeFe(Me₂pdt)(CO)₂(dppv)₂

Also prepared in this work were ⁵⁷Fe-labeled complexes, which are suitable for characterization by ⁵⁷Fe-NRVS (see below). To conserve on ⁵⁷Fe, we targeted derivatives of the [H1]⁺ hydride species where half of the iron was ⁵⁷Fe (Scheme 3). To prepare such “half-labeled” complexes, a ⁵⁷Fe(II) dithiolate precursor was condensed with an unlabeled Fe(0) reagent. The ferrous building block ⁵⁷Fe(Me₂pdt)(CO)₂(dppv) was prepared in good yield by the reaction of ⁵⁷FeCl₂, dppv, and Li₂Me₂pdt under an atmosphere of CO. Although Fe(Me₂pdt)(CO)₂(dppv) has not been reported, the closely related Ph₂PC₂H₄PPh₂-containing analogue is known.²⁷ The ³¹P NMR spectrum indicates a pair of isomers in a ratio of 17:1. The most abundant isomer has trans carbonyls, while the other isomer carbonyls are cis and the diphosphine exhibits two ³¹P NMR signals (Figure S33). The trans:cis ratio for the related but less bulky species Fe(pdt)(CO)₂(dppv) is 4:1.²⁷ The trans isomer is apparently favored by the bulkiness of the Me₂pdt²⁻ ligand. IR spectra for ⁵⁷Fe(Me₂pdt)(CO)₂(dppv) and Fe(pdt)-(CO)₂(dppv) are indistinguishable (Table S1).

Scheme 3. Synthesis of ⁵⁷FeFe(Me₂pdt)(CO)₂(dppv)₂ ([⁵⁷1]⁰) and Its Derivatives^a.

^aPh groups on phosphorus are omitted for clarity. The number 57 shown in gray implies 50% ⁵⁷Fe.

Treatment of ⁵⁷Fe(Me₂pdt)(CO)₂(dppv) with Fe(bda)-(CO)₃²⁸ gave (CO)₃Fe(Me₂pdt)⁵⁷Fe(CO)(dppv) (bda = benzylideneacetone). In this diiron compound, dppv remains exclusively coordinated to ⁵⁷Fe, as revealed by the observation of the doublet J(³¹P,⁵⁷Fe) = 50.9 Hz in the ³¹P NMR spectrum (Figure S36). This result indicates the absence of scrambling of the diphosphine. UV irradiation of a solution of this tetracarbonyl with dppv gave ⁵⁷FeFe(Me₂pdt)(CO)₂(dppv)₂ ([⁵⁷1]⁰). The ³¹P NMR spectrum of [⁵⁷1]⁰ is characterized by a broad singlet at δ 87.2, which is similar in line shape to that of unlabeled [1]⁰ (Figure S38).

Electrochemistry of [1]⁰

Cyclic voltammetry studies on CH₂Cl₂ solutions of [1]⁰ show a reversible oxidation at −1.04 V (all potentials are referenced vs Fc^+/0, Figure S8). With a redox couple at −0.83 V, Fe₂(pdt)(CO)₂(dppv)₂ is a poorer reductant than [1]⁰ by ∼210 mV. This trend is consistent with steric stabilization of the oxidized product (see below), whose rotated geometry accommodates one of the methyl groups of the Me₂pdt²⁻ ligand. In the case of [Fe₂(R₂pdt)-(CO)₄(PMe₃)₂]^0/+, the effect of pdt²⁻ vs Me₂pdt²⁻ is very similar: 240 mV.²⁹

Characterization of [Fe₂(Me₂pdt)(CO)₂(dppv)₂]⁺

Oxidation of [1]⁰ with FcBF₄ (Fc = [Fe(C₅H₅)₂]⁺) gave the mixed-valence salt [1]BF₄ (Figure 2). As for previously reported Fe(II)Fe(I) species, [1]⁺ is structurally related to the H_ox state of [FeFe]-H₂ase.^12,30–36 [1]BF₄ is analogous to [Fe₂(pdt)-(CO)₂(dppv)₂]⁺.³⁷. Crystallographic characterization of [1]BF₄ confirmed that [1]⁺ adopts the rotated structure characteristic of these mixed-valence species. One CO ligand is semi-bridging, with Fe_p–C and Fe_d–C distances of 2.461(2) and 1.790(6) Å, respectively. The dppv bound to the rotated iron occupies dibasal positions, whereas the dppv ligand on the second iron (Fe_p) occupies an apical-basal configuration, reminiscent of the location of the two donor ligands on the proximal Fe (Fe_p) in the enzyme. The Fe–P bonds are similar for Fe_p vs Fe_d (2.213(2)/2.243(2) Å vs 2.236(2)/2.254(1) Å, respectfully). The crystal structure also reveals an interaction between one of the protons from a methyl group and the Fe_d (d_C–Fe = 2.409 Å), consistent with an anagostic interaction.³⁸

Figure 2.

Crystal structure of [Fe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ ([1]BF₄) with thermal ellipsoids drawn at the 50% probability level. Hydrogen atoms and the counteranion are omitted.

EPR spectra of [1]BF₄ as a solution in toluene-THF were recorded at −150 and 30 °C (Figure 3). The low-temperature spectrum is characteristic of an axial symmetry with g = 2.105 and 2.013. For both D. desulfuricans (Dd) H₂ase and C. pasteurianum (Cp) H₂ase I, g₁ = 2.10 was reported, which is very similar to the g₁ value for [1]BF₄ of 2.105. However, the g₃ value of 2.013 for [1]BF₄ deviates from the g₃ values for Dd H₂ase and Cp H₂ase I: 1.999 and 2.00, respectively.³⁹ In comparison to other synthetic H_ox models, [1]BF₄ g values match well with those of [Fe₂(pdt)(CO)₃(dppv)(Pi-Pr₃)]-BF₄.³⁰ Hyperfine coupling, assigned to A(³¹P), is well resolved. In the isotropic spectrum, a triplet is observed at g_iso = 2.045 with A(³¹P) = 70 MHz. The hyperfine splitting is consistent with the spin residing on a single Fe(dppv) center. No hyperfine coupling to the dppv on the nonrotated Fe was resolved, consistent with little delocalization.

Figure 3.

X-band EPR spectrum of [1]BF₄ in 1/1 toluene/THF. The top spectrum was recorded at −150 °C and the bottom spectrum at 30 °C.

Characterization of [t-HFe₂(Me₂pdt)(CO)₂(dppv)₂]⁺

Like related diiron(I) complexes,¹⁸ [1]⁰ is protonated by HBF₄ to give a hydride complex (Scheme 3). The salt [t-HFe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ ([t-H1]BF₄) was isolated as a brown-green powder and characterized spectroscopically. Its room-temperature ³¹P NMR spectrum shows four resonances (Figure S15). Consistent with the low symmetry of the complex, the four resonances indicate apical-basal dppv on the proximal iron and a dibasal dppv on the FeH (Fe_d) center. The locations of the four phosphine ligands are similar for [1]BF₄ and [t-H1]BF₄, consistent with a biomimetic relationship between oxidized and protonated states. Solutions of [t-H1]BF₄ isomerize, albeit slowly, at room temperature (see the isomerization section). The slowness of this isomerization reflects the effect of steric congestion.⁴⁰ The nature of the steric congestion was probed by 1D nuclear Overhauser effect (nOe) NMR spectroscopy (Figure 4). Irradiation of the t-H signal at δ−2.23 revealed interactions between the hydride and methyl protons (δ 0.6) and between the hydride and one aryl proton (δ 7.8). Only the axial methyl interacts with the hydride. The DFT-optimized CH_Me⋯H_tFe_d distance is ∼1.7 Å. Additionally, the hydride is also close to a pair of aryl protons on the distal dppv; the closest DFT-optimized CH_A⋯H_tFe_d distance is ∼2.3 Å (see Figure 6a and further details on the DFT modeling below).

Figure 4.

¹H NMR spectrum and nOe spectrum of [t-H1]BF₄.

Figure 6.

DFT-optimized structures of the terminal (a, b) and bridging (c, d) hydride isomeric species including their d- (a and c) and p- (b and d) flippamer alternatives on the Me₂pdt²⁻ bridge conformation. Notations of the species are provided ([t_d-H1]⁺, [t_p-H1]⁺, [μ_d-H1]⁺, [μ_p-H1]⁺). Important atomic labels are provided in (a and c). The H_t/μ hydride site is highlighted as a blue sphere. The CH_X⋯H_t/μFe_d dihydrogen contacts within 3 Å are shown, with H_X highlighted as purple spheres. The relative energy ranges provided are based on the values given in Table S3.

Isomerization of [t-HFe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄

As for all other examples of [t-HFe₂(SR)₂(CO)_6−xL_x]⁺, [t-H1]⁺ isomerizes irreversibly to a bridging hydride.⁴¹ The decay rate follows first-order kinetics (k = 6.0 × 10⁻⁴ s⁻¹) with a half-life of 19.3 min at 40.3 °C (Figure 5). In contrast, the isomerization of the unmethylated analogue [t-HFe₂(pdt)(CO)₂(dppv)₂]⁺ proceeds with a half-life of minutes at room temperature (k =0.001 s⁻¹).⁴⁰ These disparate rates correspond to a difference in free energy of activation of about 1 kcal/mol.

Figure 5.

Kinetic trace of [t-H1]⁺ isomerization to [μ-H1]⁺ as followed by ³¹P NMR spectroscopy at 40.3 °C.

The final product of the isomerization of [t-H1]⁺ is the bridging hydride [μ-H1]⁺. Throughout the conversion a third hydride was observed, an intermediate in the isomerization. Its chemical shift at δ −3.63 indicates that this intermediate is a terminal hydride.^40,42 It is C_s-symmetric, as indicated by two ³¹P NMR signals (δ 79.9 and 72.7; see Figures S20 and 21). These chemical shifts are consistent with dibasal dppv ligands; in contrast, when dppv spans apical-basal sites, the apical phosphines typically absorb near δ 90^40,42). The intermediate is only observed during isomerization but does not accumulate. It thus appears that [t-H1]⁺ converts to a second terminal hydride ([t-H1′]⁺), which then converts to the μ-hydride (Scheme 4).

Scheme 4. Proposed Pathway for the Isomerization of [t-HFe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ to [μ-HFe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄^a.

^aPh groups are omitted for clarity.

Previous DFT analysis indicated that only the Fe with the hydride twists via a Ray–Dutt mechanism to form the μ-hydride.⁴³ Analysis of the isomerization of [t-HFe₂(pdt)-(CO)₂(dppv)₂]⁺ revealed that the symmetric terminal hydride is approximately 3 kcal/mol higher in energy than the asymmetric analogue.⁴² Thus, it is proposed that the isomerization starts with a twist at the Fe_p center to give a C_s-symmetric terminal hydride species. Subsequently, the Fe_d center rotates to give the μ-hydride. These measurements provide evidence for involvement of the proximal (non-hydride-containing) iron in the isomerization of the hydride. The rotation of the Fe_d(CO)(dppv)H center shifts the hydride ligand to the bridging site and the dppv_d to apical-basal positions.⁴³ By first forming [t-H1′]⁺, the cation is preorganized to generate the correct isomer of [μ-H1]⁺, wherein the hydride is trans to a CO ligand, a favored geometry.⁴³ This result implies that the flexibility of Fe_p influences the isomerization of terminal to bridging hydrides. A similar sequence of events was also seen in [t-HFe₂(adt)(CO)₂(PMe₃)₄]⁺, where the kinetically favored isomer akin to [t-H1]⁺ converts partially to an all-basal isomer structurally analogous to [t-H1′]⁺.⁴⁴

For the purpose of evaluating their vibrational spectra, [t-H⁵⁷1]BF₄, [t-D⁵⁷1]O₂CCF₃, [μ-H⁵⁷1]O₂CCF₃, and [μ-D⁵⁷1]-O₂CCF₃ were analyzed. CF₃CO₂D was selected as a convenient source of D⁺. Deuteration was incomplete but sufficient for unambiguous assignments.

DFT Modeling of the Structures and Energies of [t-H1]⁺ and [μ-H1]⁺

Density functional theory (DFT) was applied to elucidate structures and relative energies of the terminal and bridging hydride species. In the absence of their structural determinations, X-ray data on the similar complex [t-HFe₂(adt^NH2)(CO)₂(dppv)₂](BF₄)₂¹⁸ has been used to generate starting structures for the models, replacing the NH₂ bridgehead to Me₂pdt²⁻. The analysis included the two flippamers³⁰ for the Me₂pdt²⁻ bridgehead, denoted respectively as “d/p” depending on its orientation toward either Fe_d/p. The optimized structures of the 2 × 2 = 4 isomers are shown in Figure 6 and Figure S50 along with their notations used below in the text. Tables S2 and S3 provide the structural parameters of the diiron cores and relative energies.

Several computational schemes predict that the bridging hydride models [μ_d/p-H1]⁺ are 5–16 kcal/mol more stable in comparison to the terminal hydride models [t_d/p-H1]⁺, regardless of the flippamer (Figure 6 and Table S3). Earlier calculations on [FeFe]-H₂ase indicated 8–16 kcal/mol stabilization for the μ-H vs t-H species.^15,45,46 Notably, the [t_d/p-H1]⁺ models yield metal–ligand distances Fe_d–C_μ= 1.8 Å and Fe_p–C_μ = 2.4–2.6 Å, which reproduce the markedly asymmetric character of the semi-bridging CO binding observed in model complexes and [FeFe]-H₂ase active site.¹ Moderate Fe_d/p–H_μ distance asymmetries within ∼0.1 Å were obtained for the bridging hydride [μ_d/p-H1]⁺ models. For every isomer, the optimization produced two or three CH_X⋯H_t/μ < 3 Å interactions involving a hydride, where X implies alternatively alkyl protons of either −CH₃ (for [t_d-H1]⁺) or −CH₂− (for [t_p-H1]⁺) at Me₂pdt²⁻ and aryl protons (for all the hydride alternatives [t/μ_d/p-H1]⁺) at Ph as shown in Figure 6.

Computational analysis further revealed that the [t_d-H1]⁺ flippamer is stabilized by 3–5 kcal/mol in comparison to [t_p-H1]⁺. This stabilization may arise in part from the CH_Me·⋯ H_tFe_d ≈ 1.7 Å interaction (Figure 6a). Comparable interactions are invoked in isoelectronic [FeFe]-H₂ase models which feature NH⋯HFe distances of 1.4–2.0 Å.^1,15 The optimized Fe_p–C_μ distance is 0.2 Å shorter in [t_d-H1]⁺ vs in [t_p-H1]⁺ (Table S2). These calculations highlight the sensitivity of the Fe_p–C_μ interactions to subtle changes elsewhere in the molecule, as well as to the DFT methodology.¹⁴ In contrast to [t_d/p-H1]⁺, the flippamers [μ_d/p-H1]⁺ of the bridging hydride isomer are predicted to be essentially equal in energy with the calculated energy gaps within 1.6 kcal/mol.

NRVS and DFT Vibrational Characterization of [t-H1]⁺, [t-D1]⁺, [μ-H1]⁺, and [μ-D1]⁺

With thermally stable ⁵⁷Fe-labeled terminal diiron hydride and its bridging hydride isomer in hand in the forms of [t-H/D⁵⁷1]⁺ and [μ-H/D⁵⁷1]⁺, their ⁵⁷Fe–H/D vibrational signatures were characterized by ⁵⁷Fe NRVS.^47,48 NRVS, which is sensitive only to the motion of ⁵⁷Fe nuclei, is presented as ⁵⁷Fe partial vibrational density of states (PVDOS) spectra. NRVS data collected for the four samples [t-H⁵⁷1]BF₄, [t-D⁵⁷1]O₂CCF₃, [μ-H⁵⁷1]O₂CCF₃, and [μ-D⁵⁷1]-O₂CCF₃ (Figure 7a,d and Figure S51a–d) were interpreted using DFT-calculated ⁵⁷Fe-PVDOS spectra (Figure 7b,e and Figure S51a–d). The analysis is facilitated by plotting the contributions of hydride vs deuteride to the normal mode energies, provided in Figures 7c,f and Figure S52b as DFT-based H_t/μ/D_t/μ-PVDOS. Normal modes having significant H/D-PVDOS are marked in Figure 7c,f and shown in Figure 8 in arrow-style representations.

Figure 7.

⁵⁷Fe-PVDOS vibrational spectra for the H/D variants (correspondingly in blue/red) of the terminal [t-H/D⁵⁷1]⁺ (upper panel, a, b) and bridging [μ-H/D⁵⁷1]⁺ (lower panel, d, e) hydride species characterized by NRVS experiments (a, d) and DFT calculations (b, e). The ×10 intensity insets display the spectra >620 cm⁻¹. DFT-based stick-style spectra show PVDOS for the terminal H_t/D_t (c) and bridging H_μ/D_μ (f) hydrides and 2 × ⁵⁷Fe_d/p (d, e) iron nuclei. The dashed vertical line at ∼600 cm⁻¹ indicates the key difference between the [t-D⁵⁷1]⁺ and [μ-D⁵⁷1]⁺ spectra as explained in the text. The labels (cm⁻¹) indicate band positions (a, b, d, e) or individual normal-mode frequencies (c, f). Mode positions marked with thick dots (c, f) correspond to significant H⁻/D⁻ hydride motions; arrow-style molecular representation of these modes is available in Figure 8. See Figure S51 for alternative spectra arrangement and also DFT ⁵⁷Fe-PVDOS for the d/p-conformers of the Me₂pdt²⁻ bridge.

Figure 8.

Arrow-style molecular representation of selected normal modes having significant hydride motion character in the best-fit DFT models for the H/D variants of the terminal [t-H/D⁵⁷1]⁺ (a, b) and bridging [μ-H/D⁵⁷1]⁺ (c, d) hydride species. Red/blue coloring of labels and arrows corresponds to the H⁻/D⁻ hydride isotope alternatives. The computed mode frequencies (cm⁻¹) and the H/D_t/μ hydride nuclei displacement amplitudes (Å) are given. The same modes are marked in the H/D-PVDOS spectra in Figure 7c,f. For these and other vibrational modes animated, see the Supporting Information.

Weak ⁵⁷Fe motion and higher frequencies make Fe–H vibrational modes particularly difficult to observe. Additional challenging factors for NRVS presently included (i) only 50% enrichment with ⁵⁷Fe which effects in higher noise level and (ii) incomplete deuteration (see above). The DFT analysis considered the two Me₂pdt²⁻ d/p-flippamer alternatives for both [t-H⁵⁷1]⁺ and [μ-H⁵⁷1]⁺. Here, we restricted our attention to the ∼600–800 cm⁻¹ “active window”, mostly exclusive to the low-intensity NRVS bands based in the δ(⁵⁷Fe–H/D) bending (δ) vibrations.

As seen in NRVS studies of related complexes,^{12,14,15,17,49–54} the spectra exhibit intense bands associated with mixed ⁵⁷Fe_d/p-S/P/C modes <470 cm⁻¹ and ν/δ(Fe_d/p-CO) stretching (ν) and bending (δ) modes in the ∼500–600 cm⁻¹ region. These higher intensity NRVS bands <600 cm⁻¹, dominated by motions of the ⁵⁷Fe ligands heavier than hydrides, are discussed in the Supporting Information.

NRVS spectra for [t-H⁵⁷1]⁺ exhibit multiple features in the 700–800 cm⁻¹ region (Figure 7a), where our DFT calculations on [t_d-H⁵⁷1]⁺ predict two well-defined δ(Fe_d-H_t) modes at 749 and 789 cm⁻¹ (Figures 7b,c and 8a) and H_t moves in two perpendicular directions; these modes bear both similarity to and difference from the ⁵⁷Fe_d–H bends characterized recently in [FeFe]-H₂ases (see Figure S53 and supplementary discussions in the Supporting Information).^14,15,55 However, the relative [t-H⁵⁷1]⁺ vs [t-D⁵⁷1]⁺ intensity levels observed in this region are indicative of the isotope sensitivity, and regional maxima at 723 (NRVS) vs 749 cm⁻¹ (DFT) match their intensities relatively well. The situation is quite different for [t-D⁵⁷1]⁺, producing a band at 601 cm⁻¹ which is missing in the spectrum of [t-H⁵⁷1]⁺ (Figure 7a). The DFT-calculated ⁵⁷Fe-PVDOS for [t_d-D⁵⁷1]⁺ predicts a matching band at 604 cm⁻¹ produced by modes at 599 and 606 cm⁻¹, resulting from the coupled δ(Fe_d–D_t)/δ(Fe–CO) bending motions (Figures 7b,c and 8b). The δ(Fe–D)/δ(Fe–CO) coupling is responsible for significant amplification of otherwise pure δ(Fe–D) mode intensities and is well-known to enhance the diagnostic power of NRVS on iron hydrides.^14,15,54 In contrast with the 604 cm⁻¹ band produced by [t_d-D⁵⁷1]⁺, DFT modeling for another [t_p-D⁵⁷1]⁺ flippamer produces an additional band at 621 cm⁻¹ which does not fit the experiment (Figure S51b). Among the two flippamer alternatives, the [t_d-D⁵⁷1]⁺ modeling clearly provides a better fit to the observed NRVS data and therefore the d-flippamer spectra were used as representative in Figure 7b; importantly, this preference is consistent with the computed [t_d/p-H⁵⁷1]⁺ relative energies favoring [t_d-H⁵⁷1]⁺ and the nOe experimental results on [t-H⁵⁷1]⁺ (see the previous sections and Table S3). Interestingly, in spite of the tight CH_Me⋯H_tFe_d ≈ 1.7 Å dihydrogen contact achieved in [t_d-H⁵⁷1]⁺ (Figure 6a), this interaction does not yield any significant vibrational coupling between the H_Me and H_t nuclei (see Figures S52b,c and S53 and supplementary discussions in the Supporting Information).

For the bridging hydride complex, a doublet feature is observed at 694/702 cm⁻¹ in the spectrum of [μ-H⁵⁷1]⁺ (Figure 7d). A similar doublet feature is calculated at 670/687 cm⁻¹ for [μ_p-H⁵⁷1]⁺ (Figure 7e). Mixed ω(Fe_d–H_μ–Fe_t) out-of-plane bending, or wagging (ω), modes absorb at these energies: see e.g. the calculated modes at 670 and 692 cm⁻¹ indicated in Figure 7f and shown in Figure 8c. In contrast, [μ_d-H⁵⁷1]⁺ modeling produces a relatively pure ω(Fe_d–H_μ–Fe_t) wag at 724 cm⁻¹ (Figure S51c), which is not consistent with the observed spectra. The DFT spectra for the bridging hydride species in Figure 7e therefore assumed the presence of the p-flippamer absorbing in the ∼600–800 cm⁻¹ “active window” (see the Supporting Information for an extended discussion).

The [μ_p-D⁵⁷1]⁺ model produces a red-shifted ω(Fe_d–D_μ–Fe_p) wag at 590 cm⁻¹ (Figures 7f and 8d), having vibrational energy similar to that of the δ(Fe_d–D_t) bends from the [t_d-D⁵⁷1]⁺ model as discussed above. The ω(Fe_d–H_μ/D_μ–Fe_p) wags, however, contain much less H⁻/D⁻ vibrational energy (Figure 7c,f) and show smaller H⁻/D⁻ displacements (Figure 8) in comparison to the δ(Fe_d–H_t/D_t) bends, and therefore H_μ⁻ to D_μ⁻ exchange introduces only a minor perturbation of the δ(Fe–CO) bands for the bridging hydride isomer (Figure 7d,e). DFT rationalizes this behavior as respectively stronger vs weaker involvement of the ⁵⁷Fe nuclei in the δ(Fe_d–H_t/D_t) bending vs ω(Fe_d–H_μ/D_μ–Fe_p) wagging motion, consistent with ∼0.2 Å shorter Fe–H distances in [t-H1]⁺ vs [μ-H1]⁺ (Table S2). Calculations therefore rationalize the well-defined NRVS modification between the terminal and bridging species as exposed by the H⁻ to D⁻ isotope exchange, which is marked by a dashed red line at ∼600 cm⁻¹ in Figure 7.

Figure 7d reveals overlapping [μ-H⁵⁷1]⁺ and [μ-D⁵⁷1]⁺ NRVS features at 694/702 cm⁻¹, consistent with the [μ-D⁵⁷1]O₂CCF₃ sample contamination by [μ-H⁵⁷1]O₂CCF₃ as quantified by ¹H NMR (see above). The corresponding 670–690 cm⁻¹ doublet features from the DFT modeling of the pure species are however not completely eliminated by the H/D isotope exchanges regardless of the flippamers: see Figure 7b,e and Figure S51a–d. The calculations indicate that these bands, protruding into the δ(⁵⁷Fe–H) “active window”, are produced by many mixed modes of Me₂pdt²⁻, dppv, and aryl molecular fragments. These modes overlap with the hydride motion only in [μ_p-H⁵⁷1]⁺ (see the 670 and 692 cm⁻¹ modes indicated in Figure 7f and shown in Figure 8c and also the discussion on ω(Fe_d–H_μ–Fe_p) above).

CONCLUSIONS

In this work, a sterically congested diiron hydride complex was prepared as a model for a diferrous hydride state of the [FeFe]-hydrogenases. The steric effects were introduced by combining a pair of dppv ligands and 2,2-dimethylpropane-1,3-dithiolate. The preparative methods were adapted to allow introduction of ⁵⁷Fe for NRVS analysis. The considerable steric effects are manifested at many stages in the chemistry: ring current effects are seen in the ¹H NMR spectrum of [1]⁰, the 210 mV greater reducing power for [1]⁰, the enhanced stability of [1]⁺, the nOe interaction between the hydride, dithiolate, and diphosphine in [t-H1]⁺, the slow isomerization of [t-H1]⁺, and the detection of its second isomer [t-H1′]⁺. The last species served as a precursor to the bridging hydride [μ-H1]⁺ isomer, predicted by DFT to be 5–16 kcal/mol more stable than [t-H1]⁺.

Owing to their thermal stability, the salts of respectively [t-H⁵⁷1]⁺ and [μ-H⁵⁷1]⁺ are well suited for their comparative assessment by NRVS spectroscopy. The Fe–H/D NRVS bands in the ∼600–800 cm⁻¹ “active window”, in spite of their inherently low intensities, reveal important chemical information when they are rationalized by DFT modeling. The combined application of NRVS and DFT clearly discriminated between the terminal vs bridging hydride isomeric species that differ by the first-shell ligand arrangement in the diiron core. The 720–800 cm⁻¹ bands produced by two perpendicular bending δ(Fe–H) motions in [t-H⁵⁷1]⁺ are twice as intense as the 670–700 cm⁻¹ bands produced by wagging ω(Fe–H–Fe) motion in [μ-H⁵⁷1]⁺. For the deuteride samples, the higher NRVS intensity δ(Fe–CO) bands at ∼600 cm⁻¹ are noticeably perturbed in [t-D⁵⁷1]⁺ but not in [μ-D⁵⁷1]⁺. Analogous NRVS fingerprints supported our assignment of the catalytic H_hyd state in [FeFe]-H₂ase to the terminal Fe_d(II)–H hydride species (see ref 15; specifically, Figure S15). The δ(Fe–H) NRVS bands around 750 cm⁻¹ have been indicated in H_hyd as well by an independent DFT modeling.⁵⁵ Using techniques different from NRVS, others proposed bridging hydrides (Fe_p–H–Fe_d) for the H_sred and H_red states.^7,19

Analysis of the δ(Fe–H/D) and ω(Fe–H/D–Fe) bands for [t-H/D⁵⁷1]⁺ and [μ-H/D⁵⁷1]⁺ gave insights into the orientation of the Me₂pdt²⁻ bridge. The ⁵⁷Fe-PVDOS spectral signatures for the terminal vs bridging hydride isomers are distinctive because of the presence or absence of H_X⋯H_t/μ interactions. From the NRVS data, one can infer a considerable level of structural insight. Consistent with the NRVS and nOe experiments rationalized by the DFT calculations, a closed CH_Me⋯H_tFe_d ≈ 1.7 Å contact is stabilized in the best-fit conformer of [t-H1]⁺. The CH_Me⋯H_tFe_d interaction is reminiscent of the NH⋯HFe_d ≈ 2.0 Å dihydrogen interaction characterized recently for the H_hyd hydride state of [FeFe]-H₂ase.¹⁵ In contrast to [t-H1]⁺, the dihydrogen interaction in the catalytic H_hyd state additionally exhibits vibrational coupling between the two H nuclei.

EXPERIMENTAL SECTION

Reactions and manipulations were performed using standard Schlenk techniques or in a nitrogen atmosphere glovebox. HPLC-grade solvents were dried by filtration through activated alumina or distilled under nitrogen over an appropriate drying agent and then used. ESI-MS data for compounds were acquired using a Waters Micromass Quattro II spectrometer. ¹H NMR (500 MHz) and ²H NMR (600 MHz) spectra were referenced to residual solvent relative to TMS. ³¹P{¹H} NMR (202 and 242 MHz) spectra were referenced to external 85% H₃PO₄. Chemical shifts are reported for room temperature in the δ scale. FT-IR spectra were recorded on a PerkinElmer 100 FT-IR spectrometer. Crystallographic data for compounds [1]⁰ and [1]BF₄ were collected using a Siemens SMART diffractometer equipped with a Mo Kα source (λ = 0.71073 Å). cis-1,2-Bis(diphenylphosphino)ethylene (dppv) and HBF₄·Et₂O solution were purchased from Aldrich. The compounds Me₂pdtH₂,⁵⁶ Fe₂(Me₂pdt)(CO)₆, and Fe₂(Me₂pdt)(CO)₄(dppv) were prepared according to literature procedures.^20,29

Fe₂(Me₂pdt)(CO)₂(dppv)₂ ([1]⁰)

Method 1

A mixture of Fe₂(Me₂pdt)(CO)₄(dppv) (113 mg, 0.15 mmol) and dppv (60 mg, 0.15 mmol) was dissolved in 100 mL of toluene in a Pyrex Schlenk tube. The reaction mixture was irradiated with a 100 W UV immersion lamp (λ = 356 nm, Spectroline) until the reaction was complete (∼25 h) as indicated by FT-IR spectroscopy. After the solvent was removed under vacuum, the residue was extracted into 5 mL of CH₂Cl₂. The filtered extract was layered with 50 mL of pentane, and this mixture was stored at −35 °C overnight. The product was precipitated as a greenish black solid. Yield: 103 mg (63%). ¹H NMR (500 MHz, CD₂Cl₂): δ 7.71–7.27 (m, 44H, C₆H₅ and =CH), 1.33–1.25 (m, 4H, SCH₂), −0.19 (br s, 6H, CH₃). ³¹P{¹H} NMR (CD₂Cl₂): δ 87.2. FT-IR (CH₂Cl₂): ν_CO 1898 (vs), 1865 (vs) cm⁻¹. ESI-MS: m/z 1094.1 [M]⁺ (calcd 1094.1). Anal. Calcd for C₅₉H₅₄Fe₂O₂P₄S₂·2CH₂Cl₂ (found): C, 57.93 (58.08); H, 4.62 (4.64).

Method 2

A solution of Fe₂(Me₂pdt)(CO)₆ (83 mg, 0.20 mmol) and dppv (158 mg, 0.40 mmol) in 100 mL of dry toluene was irradiated with an LED lamp (λ = 356 nm). The reaction was complete in ∼7 h, as indicated by FT-IR spectroscopy. Solvent was removed under vacuum, and the residue was extracted into 5 mL of CH₂Cl₂. The filtered extract was layered with 100 mL of hexane, and the mixture was allowed to stand at −35 °C overnight. The product precipitated as a greenish black solid. Yield: 170 mg (78%).

[Fe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ ([1]BF₄)

A dark green solution of Fe₂(Me₂pdt)(CO)₂(dppv)₂ (33 mg, 0.030 mmol) in 2 mL of CH₂Cl₂ was treated at −40 °C with FcBF₄ (9 mg, 0.033 mmol) in 3 mL of CH₂Cl₂. The reaction mixture was stirred at −40 °C until the reaction was complete (5 min) as indicated by FT-IR spectroscopy. The resulting solution was then layered with 25 mL of hexane, which was allowed to stand at −35 °C overnight. The product precipitated as blue-black needlelike crystals. Yield: 27 mg (77%). FT-IR (CH₂Cl₂): ν_CO 1952 (vs), 1854 (s) cm⁻¹. ¹H NMR (500 MHz, CD₂Cl₂): δ 9.40–6.80 (m, 55H, C₆H₅, Ph₂PCH=CHPPh₂), 1.26 (m, 4H, 2 × SCH₂), −2.07 (br s, 6H, 2 × CH₃). Anal. Calcd for C₅₉H₅₄BF₄Fe₂O₂P₄S₂·CH₂Cl₂ (found): C, 56.9 (57.22); H, 4.46 (4.43).

[(t-H)Fe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ ([t-H1]BF₄)

A dark green solution of Fe₂(Me₂pdt)(CO)₂(dppv)₂ (33 mg, 0.030 mmol) in 5 mL of CH₂Cl₂ was treated at 0 °C with HBF₄·Et₂O (6.3 μL, 0.045 mmol). This species was not formed when THF was used as the solvent. The reaction mixture was stirred at 0 °C until the reaction was complete (10 min) as indicated by FT-IR spectroscopy. The resulting solution was then layered with 50 mL of hexane, which was allowed to stand at −35 °C overnight. The product, a greenish black solid, was stored at −35 °C in the glovebox. Yield: 15 mg (42%). ¹H NMR (500 MHz, CD₂Cl₂): δ 8.22–6.73 (m, 44H, C₆H₅ and =CH), 1.59–1.27 (m, 4H, SCH₂), 0.50, 0.03 (s, s, 6H, CH₃), −2.23 (t, J_PH = 81 Hz, 1H, (t-H)Fe). ³¹P{¹H} NMR (CD₂Cl₂, 20): δ 95.3 (s), 87.2 (s), 80.4 (s), 68.7 (s). FT-IR (CH₂Cl₂): ν_CO 1964 (vs), 1889 (s) cm⁻¹. ESI-MS: m/z 1095.1 [M − BF₄]⁺ (calcd 1095.1). Anal. Calcd for C₅₉H₅₅BF₄Fe₂O₂P₄S₂·2.5CH₂Cl₂ (found): C, 52.95 (53.01); H, 4.34 (4.38).

[(μ-H)Fe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ ([μ-H1]BF₄)

A dark green solution of Fe₂(Me₂pdt)(CO)₂(dppv)₂ (44 mg, 0.040 mmol) in 5 mL of CH₂Cl₂ was treated at room temperature with HBF₄·Et₂O (8.4 μL, 0.060 mmol). The reaction mixture was stirred at room temperature until the reaction was complete (∼41 h) as indicated by FT-IR spectroscopy. After the solvent was removed, the residue was washed with 2 × 10 mL of Et₂O and dried under vacuum. The product was obtained as a brown-red solid. Yield: 35 mg (72%). ¹H NMR (500 MHz, CD₂Cl₂): δ 8.08–6.67 (m, 44H, C₆H₅ and =CH), 2.57–1.37 (m, 4H, SCH₂), 1.15, 0.20 (s, s, 6H, CH₃), −14.70 (m, 1H, Fe(μ-H)Fe). ³¹P{¹H} NMR (CD₂Cl₂, 20 °C): δ 86.8 (s), 81.3 (s), 78.5 (s), 72.5 (s). FT-IR (CH₂Cl₂): ν_CO 1974 (broad), 1952 (vs) cm⁻¹. ESI-MS: m/z 1095.3 [M − BF₄]⁺ (calcd 1095.1). Anal. Calcd for C₅₉H₅₅BF₄Fe₂O₂P₄S₂·1.5CH₂Cl₂ (found): C, 55.47 (55.73); H, 4.46 (4.85). The isomerization of [t-H1]⁺ to [μ-H1]⁺ was followed at 40.3 °C using ³¹P NMR spectroscopy. Tri-p-tolylphosphine was used as an integration standard.

[(t-D)Fe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ ([t-D1]BF₄)

A dark green solution of Fe₂(Me₂pdt)(CO)₂(dppv)₂ (55 mg, 0.050 mmol) in 5 mL of CH₂Cl₂ was treated at 0 °C with D₂O (9.0 μL, 0.050 mmol) followed by HBF₄·Et₂O (11.0 μL, 0.075 mmol). The reaction mixture was stirred at 0 °C until the reaction was complete (10 min) as indicated by FT-IR spectroscopy. The resulting solution was then layered with 50 mL of Et₂O, which was allowed to stand at −35 °C overnight. The product precipitated as a greenish black solid and was stored at −35 °C in the glovebox. Yield: 35 mg (59%). The deuteration was incomplete, as indicated by a small t-H signal at δ −2.23. ¹H NMR (500 MHz, CD₂Cl₂): δ 8.22–6.73 (m, 44H, C₆H₅ and =CH), 1.57–1.27 (m, 4H, SCH₂), 0.49, 0.02 (s, s, 6H, CH₃), −2.23 (t, J_PH = 80 Hz, 0.3H, (t-H)Fe). ²D NMR (600 MHz, CH₂Cl₂): δ −2.14 (t, J_PD = 12.4 Hz, (t-D)Fe). ³¹P{¹H} NMR (CD₂Cl₂): δ 95.2 (s), 87.3 (2 × br d), 80.6 (2 × br d), 68.7 (s). FT-IR (CH₂Cl₂): ν_CO 1964 (vs), 1890 (s) cm⁻¹. ESI-MS: m/z 1096.2 [M − BF₄]⁺ (calcd 1096.2).

[(μ-D)Fe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ ([μ-D1]BF₄)

A dark green solution of Fe₂(Me₂pdt)(CO)₂(dppv)₂ (33 mg, 0.030 mmol) in 5 mL of CH₂Cl₂ was treated at room temperature with D₂O (5.4 μL, 0.300 mmol) followed by HBF₄·Et₂O (12.6 μL, 0.090 mmol). The mixture was stirred at room temperature until the reaction was complete (∼52 h) as indicated by FT-IR spectroscopy. The resulting solution was then layered with 25 mL of hexane, which was allowed to stand at −35 °C overnight. The product was obtained as a brown-red solid. Yield: 20 mg (56%). The deuteration was not fully complete, as indicated by a small μ-H signal at δ −14.8. ¹H NMR (500 MHz, CD₂Cl₂): δ 8.08–6.68 (m, 44H, C₆H₅ and =CH), 260–1.37 (m, 4H, SCH₂), 1.15, 0.21 (s, s, 6H, CH₃), −14.74 (m, 0.3H, Fe(μ-H)Fe). ²D NMR (600 MHz, CH₂Cl₂): δ −14.8 (s, Fe(μ-D)Fe). ³¹P{¹H} NMR (CD₂Cl₂): δ 86.8 (s), 81.2 (s), 78.4 (2 × br s), 72.5 (s). FT-IR (CH₂Cl₂): ν_CO 1974 (broad), 1952 (vs) cm⁻¹. ESI-MS: m/z 1096.1 [M − -BF₄]⁺ (calcd 1096.1).

⁵⁷FeCl₂

To a solution of ⁵⁷Fe powder (250 mg, 4.39 × 10⁻³ mol) in MeOH (1.6 mL) was added concentrated HCl (12 M, 1.6 mL, 1.92 × 10⁻² mol). The reaction mixture was stirred under nitrogen at room temperature for 2 h. The solvent was then removed under vacuum. The residual white powder was dried further by heating under vacuum overnight. The product was used without purification. Yield: 544 mg (97%).

⁵⁷Fe(Me₂pdt)(dppv)(CO)₂

A 100 mL Schlenk flask, equipped with a magnetic stir bar, was charged with ⁵⁷FeCl₂ (544 mg, 4.25 × 10⁻³), dppv (1.69 g, 4.25 × 10⁻³), and THF (30 mL). The reaction mixture was placed under an atmosphere of CO, and the reaction mixture was stirred at room temperature for 2 h. A solution of Me₂C(CH₂SLi)₂ (4.25 × 10⁻³ mol) in THF (10 mL) was generated by treating Me₂C(CH₂SH)₂ (576 mg, 4.25 × 10⁻³ mol) with 4.25 mL of a 1 M solution of BuLi in hexane. This mixture was diluted with MeOH (10 mL) before cannula-transferring to the ⁵⁷FeCl₂/dppv mixture. The mixture was stirred at room temperature for 12 h. A dark red, heterogeneous solution formed. The solvents were removed under vacuum, and the residual solid was purified by column chromatography. Elution with CH₂Cl₂ gave a yellow band that was discarded. Subsequent elution with a 5/1 mixture of CH₂Cl₂ and Et₂O gave a red band containing the product. Yield: 719 mg (26%). IR (CH₂Cl₂): ν_CO 2014, 1975 cm⁻¹. ³¹P{¹H} NMR (CD₂Cl₂): δ 86.1 (dd, J = 35.8, 21.6 Hz), 79.7 (d, J(⁵⁷Fe) = 38.5 Hz) 59.8, (dd, J = 35.4, 21.6 Hz); ratio 1:17:1. ¹H NMR (500 MHz, CD₂Cl₂): δ 7.96 (d, J = 11.3 Hz, 1H), 7.87 (d, J = 10.7 Hz, 1H), 7.72–7.63 (m, 6H), 7.44 (q, J = 7.7 Hz, 10H), 2.19 (m, 4H), 0.98 (m, 6H).

(dppv)(CO)⁵⁷Fe(μ-Me₂pdt)Fe(CO)₃

A toluene (25 mL) solution of ⁵⁷Fe(Me₂pdt)(dppv)(CO)₂ (719 mg, 1.12 × 10⁻³ mol) was treated with a toluene solution of Fe(bda)(CO)₃²⁸ (322 mg, 1.12 × 10⁻³ mol). The reaction mixture was stirred at room temperature for 16 h, at which point IR spectroscopy indicated that the reaction was complete. The solution was filtered through Celite, and the solvent was removed under vacuum. The residual solid was rinsed with pentane (3 × 10 mL) and dried under vacuum. Yield: 706 mg (87%). IR spectrum (CH₂Cl₂): ν_CO 2022, 1953, 1913 cm⁻¹. ³¹P{¹H} NMR (CD₂Cl₂): δ 92.83 (d, J(⁵⁷Fe) = 50.9 Hz), 79.92 (m); ratio 1:0.6. ¹H NMR (500 MHz, CD₂Cl₂): δ 8.16–7.88 (overlapping signals, 4H), 7.78 (m, 1H), 7.67 (m, 1H), 7.57–7.10 (overlapping signals, 16H), 1.72 (m, 2H), 1.29 (m, 2H), 1.08 (s, 3H), 0.09 (s, 3H).

(dppv)(CO)⁵⁷Fe(μ-Me₂pdt)Fe(CO)(dppv) ([⁵⁷1]⁰)

A mixture of (dppv)(CO)⁵⁷Fe(μ-Me₂pdt)Fe(CO)₃ (120 mg, 1.65 × 10⁻⁴ mol) and dppv (66 mg, 1.65 × 10⁻⁴ mol) in toluene (100 mL) was irradiated with 356 nm light (LED, 350 mA/m²). Reaction progress was monitored by IR spectroscopy. After 6 h, irradiation was stopped, and the solvent was removed under vacuum. The crude product was dissolved in a minimum amount of CH₂Cl₂. This extract was layered with pentane, and the mixture was stored at −35 °C overnight. Dark green crystals formed. Yield: 146 mg (83%). FT-IR (CH₂Cl₂): ν_CO 1974, 1952, 1897, 1865 cm⁻¹. ³¹P{¹H} NMR (CD₂Cl₂): δ 87.21 (br s). ¹H NMR (500 MHz, CD₂Cl₂): δ 8.13–7.56 (br s, 10H), 7.56–6.72 (broad, overlapping signals, 34H), 1.29 (m, 4H), −0.15 (br s, 6H).

⁵⁷Fe-Labeled [(t-H)Fe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄ ([t-H⁵⁷1]BF₄)

A CH₂Cl₂ solution of (dppv)(CO)⁵⁷Fe(μ-Me₂pdt)Fe-(CO)(dppv) (63.4 mg, 5.78 × 10⁻⁵ mol) was treated with HBF₄·Et₂O (7.9 μL, 9.4 mg, 5.78 × 10⁻⁵ mol) at room temperature. A color change from green to green-brown was observed. Solvent was removed under vacuum, and the residual solid was rinsed with THF and Et₂O. Yield: 30 mg (44%). ³¹P{¹H} NMR (CD₂Cl₂): δ 95.3 (br s, 1P), 87.3 (br s, 1P), 80.5 (br s), 68.8 (br s, 1P). ¹H NMR (500 MHz, CD₂Cl₂): δ 8.68–6.37 (overlapping signals, 44H), 1.59 (m, 2H), 1.30 (m, 2H), 0.53 (s, 3H), 0.05 (s, 3H), −2.20 (t, J = 81 Hz, 1H). ¹H NMR analysis of the sample for NRVS revealed a small signal at δ −14.7 corresponding to the presence of ∼13% of [μ-H⁵⁷1]BF₄ in this sample (Figure S43).

⁵⁷Fe-Labeled [(t-D)Fe₂(Me₂pdt)(CO)₂(dppv)₂]O₂CCF₃ ([t-D⁵⁷1]O₂CCF₃)

A CH₂Cl₂ solution of (dppv)(CO)⁵⁷Fe(μ-Me₂pdt)-Fe(CO)(dppv) (63.4 mg, 5.78 × 10⁻⁵ mol) was treated with CF₃CO₂D (4.7 μL, 6.6 mg, 5.78 × 10⁻⁵ mol). A color change from green to green-brown was observed. Solvent was removed under vacuum, and the residual solid was washed with THF and Et₂O. Yield: 40 mg (58%). ³¹P{¹H} NMR (CD₂Cl₂): δ 95.3 (br s, 1P), 87.3 (br s, 1P), 80.5 (br s), 68.8 (br s, 1P). ¹H NMR (500 MHz, CD₂Cl₂): δ 8.68–6.37 (overlapping signals, 44H), 1.59 (m, 2H), 1.30 (m, 2H), 0.53 (s, 3H), 0.05 (s, 3H).

⁵⁷Fe-Labeled [(μ-H)Fe₂(Me₂pdt)(CO)₂(dppv)₂]BF₄/O₂CCF₃ ([μ-H⁵⁷1]O₂CCF₃) and [(μ-D)Fe₂(Me₂pdt)(CO)₂(dppv)₂]O₂CCF₃ ([μ-H⁵⁷1]O₂CCF₃)

Solutions of [t-H⁵⁷1]O₂CCF₃ and [t-D⁵⁷1]O₂CCF₃ in 2–4 mL of CH₂Cl₂ were stirred until the IR spectrum showed complete conversion to [μ-H⁵⁷1]⁺ and [μ-D⁵⁷1]⁺, respectively. The appearance of an IR band at 1855 cm⁻¹ in the spectra of [μ-H⁵⁷1]O₂CCF₃ and [μ-D⁵⁷1]O₂CCF₃ indicates the presence of a small amount of [⁵⁷1]⁺ in these samples (Figures S44 and S47). Finally, the appearance of a signal at δ −14.7 in the ¹H NMR spectrum of [μ-D⁵⁷1]O₂CCF₃ (Figure S49) indicates only 65% deuteration, the remainder being [μ-H⁵⁷1]O₂CCF₃. ¹H NMR (500 MHz, CD₂Cl₂): δ 8.08–6.67 (m, 44H, C₆H₅ and =CH), 2.57–1.37 (m, 4H, SCH₂), 1.17, 0.23 (s, s, 6H, CH₃), −14.72 (m, 1H, Fe(μ-H)Fe). ³¹P{¹H} NMR (CD₂Cl₂): δ 87.1 (t), 81.5 (t), 78.8 (t), 72.8 (t). FT-IR (CH₂Cl₂): ν_CO 1974 (broad), 1952 (vs) cm⁻¹. [μ-D⁵⁷1]O₂CCF₃. ¹H NMR (500 MHz, CD₂Cl₂): δ 8.08–6.67 (m, 44H, C₆H₅ and =CH), 2.57–1.37 (m, 4H, SCH₂), 1.17, 0.23 (s, s, 6H, CH₃). ³¹P{¹H} NMR (CD₂Cl₂): δ 87.34 (t), 81.8 (t), 78.9 (t), 72.9 (t). FT-IR (CH₂Cl₂): ν_CO 1974 (broad), 1952 (vs) cm⁻¹.

Cyclic Voltammetry

Electrochemical experiments were carried out in a nitrogen-filled glovebox fitted with adapters to connect to a CH Instruments Model 600D series electrochemical analyzer. These experiments were conducted in a 3 mL glass cell fitted with a Teflon top. A glassy-carbon electrode (d = 3 mm) was used as the working electrode. A silver wire was used as a pseudoreference electrode, and ferrocene was added to the cell so the collected data could be referenced to the Fc^0/+ couple (0.00 V). The counter electrode for these experiments was a platinum wire. Due to these experiments being collected in a glovebox, sparging was not necessary. iR compensation was applied to each voltammogram. These experiments were conducted at room temperature with 1 mM of the complex and 0.125 M of supporting electrolyte in dried and deoxygenated solvents.

NRVS

Nuclear resonance vibrational spectroscopy (NRVS) data of [(μ-H/D)⁵⁷1]BF₄/O₂CCF₃ were collected at the Advanced Photon Source (APS) sector 3-ID, Argonne National Laboratory, with the storage ring operating in the standard operating mode (24 electron bunches spaced by 153 ns). A water-cooled high heat-load monochromator (HHLM) consisting of two diamond single crystals with (1,1,1) as the reflection plane and a four-bounce 2Si(4,0,0) × 2Si(10,6,4) high-resolution monochromator (HRM) provided the 14.4 keV X-rays with 1 meV energy resolution. The flux was about 2.5 × 10⁹ photons/s positioned on our sample at the 3ID-D station. The operation temperature was set to 10 K, while the real sample temperature was ∼58 K for [(μ-H)⁵⁷1]BF₄ and 62 K for [(μ-D)⁵⁷1]O₂CCF₃. The nuclear scattering induced ⁵⁷Fe nuclear fluorescence at 14.4 keV, and the converted Fe Kα fluorescence at 6.4 keV was recorded with an Avalanche Photo Diode (APD) detector, with an active area of 10 mm by 10 mm in size and 0.1 mm in thickness.

The NRVS spectra of [(t-H/D)⁵⁷1]BF₄/O₂CCF₃ were collected at SPring-8 BL09XU and BL19LXU in the “C” bunch mode (29 electron bunches spaced by 145.5 ns).⁵⁷ The samples were placed in a helium cold finger maintained at 10 K, and the true temperature ranged from 65 to 75 K. Using a double Si(1,1,1) HHLM and a [Ge(4,2,2),2×Si-(9,7,5)] HRM, the nuclear resonance energy at 14.4 keV with a 0.8 meV resolution was achieved (flux = 6 × 10⁹ cps at BL19XU and 2.0 × 10⁹ cps at BL09XU).¹⁷ An APD array was used to detect the ⁵⁷Fe nuclear fluorescence and Fe Kα fluorescence. [t-D⁵⁷1]O₂CCF₃ was measured using sectional scans: −100 to 550 cm⁻¹ at 1 s/pt, 550–800 cm⁻¹ at 3 s/pt, 800–1250 cm⁻¹ at 1 s/pt, and 1250–1500 cm⁻¹ at 20 s/pt with respect to ⁵⁷Fe nuclear resonance. All NRVS spectra were processed with PHOENIX executed through the web application spectra.tools to yield ⁵⁷Fe partial vibrational density of states (PVDOS).⁵⁸

DFT Calculations

Structural optimization and normal-mode analysis was done using GAUSSIAN 09,⁵⁹ on the basis of the densities exported from single-point calculations using JAGUAR 9.4.⁶⁰ The BP86^61,62 (for most of the results) and B3LYP⁶¹ (for single-point energy calculations only) functionals together with the LACV3P** basis set were employed. For the first- and second-row elements, LACV3P** implies 6-311G** triple-ζ basis sets including polarization functions. For the Fe atoms, LACV3P** consists of a triple-ζ quality basis set for the outermost core and valence orbitals and the quasi-relativistic Los Alamos effective core potential (ECP) for the innermost electrons.^63,64 The model environment was considered using a self-consistent reaction field (SCRF) polarizable continuum model and integral equation formalism (IEF-PCM) as implemented in GAUSSIAN 09, with the static dielectric constant set to ε = 4.0 and the remaining IEF-PCM parameters at their default values for water. The calculations (i) including the two-body D3 dispersion corrections by Grimme et al.^65,66 as implemented in GAUSSIAN 09 and (ii) excluding the D3 correction have been done to generate the normal modes in the (i) <470 cm⁻¹ and (ii) >470 cm⁻¹ areas, respectively; this combined approach ultimately produced DFT-based ⁵⁷Fe-PVDOS spectra in better agreement with the experiment. The optimized structures shown in the figures are based on scheme ii, if not otherwise mentioned. On the basis of the normal mode outputs, an in-house Q-SPECTOR program successfully applied previously^15,51–54 was utilized to generate the ⁵⁷Fe-PVDOS from the normal mode composition factors. An empirical scaling of calculated frequencies has not been applied. The resolution of the observed NRVS spectra was accounted for by convolution of the computed PVDOS intensities with a full width at half-maximum (fwhm) equal to 14 cm⁻¹ Lorentzian.