Hydrogen Activation by Biomimetic Diiron Dithiolates

Abstract

Using the thermally stable salts of [Fe₂(SR)₂(CO)₃(PMe₃)(dppv)]BAr^F₄, we found that the azadithiolates [Fe₂(adtR)(CO)₃(PMe₃)(dppv)]⁺ react with high pressures of H₂ to give the hydride [(μ-H)Fe₂(adt)(CO)₃(PMe₃)(dppv)]BAr^F₄. The related oxadithiolate and propanedithiolate complexes are unreactive toward H₂. Molecular hydrogen is proposed to undergo heterolysis assisted by the amine followed by isomerization of an initially formed terminal hydride. Use of H₂ and D₂O gave the deuteride as well as the hydride, implicating protic intermediates.

Hydrogenases are enzymes that catalyze the interconversion of dihydrogen with protons and reducing equivalents.¹ Understanding the reactivity of these enzymes via active site models remains topical,² especially since these catalysts rely on inexpensive first-row transition metals.³ Significant progress has been made in [FeFe]-hydrogenases models,^4,5 but nearly all studies to date have focused on proton reduction.⁶ The opposite reaction, hydrogen oxidation, has proven elusive. This lack of reactivity is surprising because [FeFe]-hydrogenases are exceptionally active towards H₂ oxidation. The translation of models to applications in fuel cells requires progress on hydrogen oxidation. Herein we report the activation of dihydrogen by a model for the [FeFe]-hydrogenase, as well as the associated advances that have facilitated this progress.

Recently we reported [Fe₂(pdt)(CO)₃(PMe₃)(dppv)]BF₄ ([1]BF₄, pdt = S₂C₃H₆)^‡, a paramagnetic (S = 1/2) spin-localized species that represents a useful structural model for the H_ox state of the binuclear active site (dppv = cis-1,2-bis(diphenylphosphino)ethylene).^7,8 With a vacant coordination site on the Fe(I) center, H_ox and its models are poised to activate H₂. Although [1]⁺ binds CO, it exhibits no discernable reactivity toward H₂. The anticipated product of hydrogen activation, [Fe₂(μ-H)(pdt)(CO)₃(PMe₃)(dppv)]BF₄ ([1H]BF₄), was prepared independently by protonation of the corresponding Fe(I)Fe(I) precursor; a terminal hydride is initially formed which rapidly isomerizes to bridging hydrides.⁹

The thermal sensitivity of [1]BF₄ severely limits studies of its reactivity toward H₂, but we found that the corresponding salt [1]BAr^F₄ is stable in solution for days at room temperature (BAr^F₄ = B(C₆H₃-3,5-(CF₃)₂)₄⁻). The sensitivity of electrophilic Fe carbonyls towards fluorinated counterions is precedented.¹⁰ We confirmed that solutions of [1]BAr^F₄ rapidly decomposed upon treatment with [Bu₄N]BF₄. The robust salt [1]BAr^F₄, however, proved unreactive toward 1800 psi H₂, even upon addition of the bulky base 2,6-(^tBu)₂pyridine. Other more basic reagents such as 2,6-dimethylpyridine are not compatible with [1]BAr^F₄, causing it to decompose to unidentified products. Apparently H₂ activation by H_ox models is subject to a significant kinetic barrier.

The proposed azadithiolato cofactor has been shown to significantly affect the acid-base properties of diiron dithiolate complexes.¹¹ Furthermore, pendant amine bases dramatically affect the rate of H₂ uptake in Ni-based catalysts.¹² We therefore investigated azadithiolato analogues of [1]BAr^F₄. As described below, we present evidence that such mixed valence azadithiolato diiron complexes indeed activate H₂.

Azadithiolate-containing analogues of [1]⁺ have proven particularly sensitive: attempts to generate [Fe₂(adt)(CO)₃(PMe₃)(dppv)]BF₄ ([2]BF₄, adt = (SCH₂)₂NH) by oxidation of 2 with FcBF₄ resulted in complex mixtures above −78 °C. We found, however, that toluene solutions of [2]BAr^F₄, generated from treatment of 2 with FcBAr^F₄,¹³ are stable for days at room-temperature. We also prepared the related benzyl derivative [Fe₂[(SCH₂)₂NCH₂Ph](CO)₃(PMe₃)(dppv)]BAr^F₄ ([2′]BAr^F₄), which was used for the majority of our studies for reasons described below. The spectroscopy for [1]⁺, [2]⁺, and [2′]⁺ are very similar (see Supporting Information),⁷ indicating that one carbonyl is semi-bridging and the apical site on the Fe(dppv) center is vacant. The unsaturated character of [2′]⁺ was confirmed by its rapid and reversible carbonylation to the adduct [2′ (CO)]⁺.

Treatment of [2]BAr^F₄ with 1800 psi H₂ for 26 h resulted in 30% conversion to the hydride [2(μ-H)]BAr^F₄. The product [2(μ-H)]⁺ is spectroscopically identical to the hydride produced by protonation of 2 (see below). Analogous but more efficient reactivity was found for the tertiary amine [2′]BAr^F₄, which gave [2′ (μ-H)]BAr^F₄ in high yield. The mixed valence oxadithiolate [Fe₂(odt)(CO)₃(PMe₃)(dppv)]⁺ ([3]⁺, odt = (SCH₂)₂O) was also found to be thermally stable, being comparable to [2′]BAr^F₄. It was unreactive toward high pressures of H₂. Collectively, these results point to the participation of the amine in the activation of H₂ by the cationic diiron complex (Table 1). Under otherwise identical conditions but using argon in place of hydrogen, [2′]⁺ remain unchanged.

Table 1.

Reactivity of [Fe₂[(SCH₂)₂X](CO)₃(PMe₃)(dppv)]⁺ Derivatives Toward H₂. (10 mM toluene soln, 1800 psi H₂, 26 h).

complex (BArF4− salt)	yield of hydride
[Fe2(pdt)(CO)3(PMe3)(dppv)]+ ([1]+)	<5%
[Fe2(adt)(CO)3(PMe3)(dppv)]+ ([2]+)	~30%
[Fe2(adt-Bn)(CO)3(PMe3)(dppv)]+ ([2′]+)	~85%
[Fe2(odt)(CO)3(PMe3)(dppv)]+ ([3]+)	<5%

A number of controls were conducted to verify the significance of our findings. ¹H and ¹⁹F NMR analyses of reaction mixtures showed that BAr^F₄⁻ and ferrocene remained unaffected. The yield of the hydride was unaffected when the reaction was conducted in toluene-d₇. When toluene solutions of [2′]BAr^F₄ were stored in the absence of H₂ for several days, only trace amounts of [2′ (μ-H)]⁺ formed, probably from a reaction involving adventitious water. This interesting process was confirmed by treatment of [2′]⁺ with D₂O to give small amounts of [2′ (μ-D)]⁺. In the same way that the adt complexes [2]⁺ and [2′]⁺ exhibit greater reactivity toward H₂ than does [1]⁺, we found that [Fe₂(adt)(CO)₂(dppv)₂]⁺ is reactive towards H₂ whereas the analogous propanedithiolate [Fe₂(pdt)(CO)₂(dppv)₂]⁺ is not.

The formation of the μ-hydride [2′ (μ-H)]⁺ requires explanation since the mechanism for H₂ activation should produce a terminal hydride adjacent to the adt.^14,15 In an independent experiment, protonation of 2′ with H(OEt₂)BAr ₄^F was found to afford the ammonium salt [2′H]⁺, which was characterized by IR and ³¹P NMR spectroscopies.¹⁶ This ammonium compound was found to convert to [2′ (μ-H)]⁺ via a first-order process (k = 3.9 × 10⁻⁴ s⁻¹, 23 °C, CH₂Cl₂) (eq 1).

(1)

This tautomerization is assumed to occur via the terminal hydride ([2′ (t-H)]⁺), which is not observed. The related terminal hydride [1(t-H)]⁺, which is observable, isomerizes to [1(μ-H)]⁺ (see Supporting Info).¹⁴ The main point is that under the conditions of the hydrogenation (hours, 25 °C), [2′ (μ-H)]⁺ is the expected product.

The hydrogenation points to the stoichiometry shown in eq 2.

$$2{[{\text{Fe}}_2(\text{Radt}){(\text{CO})}_3({\text{PMe}}_3)(\text{dppv})]}^{+}+{\text{H}}_2\to 2{[{\text{HFe}}_2(\text{Radt}){(\text{CO})}_3({\text{PMe}}_3)(\text{dppv})]}^{+}$$ (2)

Although this reaction appears homolytic, the finding that it requires the azadithiolate indicates a heterolytic mechanism. A thermodynamic cycle highlights key steps – hydride binding, redox, and proton transfer (Figure 1). In the protein, the stoichiometry of course differs from eq 2, since the initial heterolysis is coupled to oxidation by an Fe-S cluster and one proton is released to the protein. In our system, the diiron complexes can interact, half of the [2′]⁺ serves sequentially as oxidant and then proton acceptor.

The proposed mechanism invokes the initial formation of the adduct [2′ (H₂)]⁺. Since exogenous CO binds only weakly to [2′]⁺,^8,17 the coordination of H₂ is expected¹⁸ to be unfavorable. Binding of H₂ initiates the heterolytic activation sequence as implied by the requirement for the amine-containing dithiolate. We have shown that the redox couple for CO adduct [Fe₂[(SCH₂)₂X](CO)₄(PMe₃)(dppv)]^+/2+ is ~250 mV more positive than the [Fe₂[(SCH₂)₂X](CO)₃(PMe₃)(dppv)]^0/+ couple.⁷ Thus, we anticipate that [2′ (H₂)]⁺ would be unable to reduce [2′]⁺. The hydride [2′ (t-H)] is, however, expected¹⁹ to be a powerful reductant (~−1.6 V vs Fc^0/+). Heterolysis of [2′ (H₂)]⁺ is apparently irreversible, since [2′]⁺ was observed not to catalyze the formation of HD from H₂ and D₂O under the conditions of the experiment. We thus propose that heterolysis induces rapid proton- and electron-transfers, giving [2(t-H)]⁺, H⁺, and 2′. Being highly basic, 2′ protonates, initially at nitrogen.

When a solution of [2′]⁺ was treated with H₂ (2000 psi) in the presence of D₂O, we obtained [2′ (μ-D)]⁺ and [2′ (μ-H)]⁺ in a ratio of ~4:1. Comparable results were obtained for the corresponding reaction involving D₂ and H₂O, which gives both [2′ (μ-D)]⁺ and [2′ (μ-H)]⁺. Control experiments confirmed that [2′ (μ-H)]⁺ does not exchange with D₂O.¹⁴ Proton transfer from an initial dihydrogen complex [2′ (H₂)]⁺ would account for 50% of the exchange. The high degree of exchange is consistent with tautomerization between [2′ (H)]⁺ and [2′ (μ–H)]⁺ (see eq 1). We have shown that terminal hydrides are also inert in the absence of the amine cofactor.¹¹

The results presented in this communication establish that biomimetic diiron dithiolato complexes can thermally activate hydrogen. The findings point to a mechanism that involves a coupling of the electron- and proton-transfers. Optimized models will require manipulation of these equilibria, as well as suppressing formation of the μ-hydride, which is a thermodynamic sink. The present results highlight the importance of the amine in the activation of dihydrogen by the diiron center.^12,20

Scheme 1.

Thermodynamic cycle for the reaction 2 [Fe₂(SR₂)(CO)₃(PR₃)₃]⁺ + H₂ → 2[HFe₂(SR₂)(CO)₃(PR₃)₃]⁺.

Scheme 2.

Proposed mechanism of H₂ activation by [Fe₂[(SCH₂)₂NR](CO)₃(PMe₃)(dppv)]⁺.