Nickel–Iron Dithiolato Hydrides Relevant to the [NiFe]-Hydrogenase Active Site

Several themes motivate the synthetic modeling of the active sites of the hydrogenase enzymes: interest in complexes of dihydrogen and hydride ligands,¹ novel biochemistry implied by structurally unusual active sites,^2–4 and the potential for utilizing hydrogen as an energy carrier.⁵ Through recent intense efforts, synthetic models for the active site of the [FeFe]-H₂ases are yielding useful mechanistic insights.^6,7 Modeling the more prevalent [NiFe]-H₂ases, however, has proven more challenging.⁸ The functional core of this enzyme features a nickel tetrathiolate wherein two thiolates are bridging to an Fe(CN)₂(CO) center. Several states of the [NiFe]-H₂ases have been detected, although only two oxidation states are catalytically significant, Fe(II)Ni(III) (Ni–C, with a hydride) and Fe(II)Ni(II) (Ni–R, with a hydride, and Ni–SI, without; see Figure 1).⁹ Despite intensive efforts, compounds featuring the NiFeS₂(μ-H) core have resisted synthetic modeling, although progress with related RuNiS₂(μ-H) has recently been described.¹⁰ We now report successful and potentially generalizable routes to the long-sought nickel–iron dithiolato hydrides.

Figure 1.

Structure of the active site in [NiFe]-hydrogenase (left, PDB #1WUL) and its key conversions (right).⁹

In our initial approach to biomimetic models, we treated the iron dithiolato complex Fe(pdt)(CO)₂(dppe)¹¹ (1) with NiCl₂(dppe) in hot acetone solution to give the μ-chloro bimetallic cation [(CO)-(dppe)Fe(pdt)(μ-Cl)Ni(dppe)]⁺ ([2Cl]⁺), isolated as its BF₄⁻ salt (pdt= 1,3-propanedithiolate; dppe = 1,2-C₂H₄(PPh₂)₂; OTs⁻ = CH₃C₆H₄-4-SO₃⁻) (Scheme 1). The brown-red colored complex is diamagnetic, consistent with low spin Fe(II) and Ni(II) centers. Attempts to replace the chloride ligand with hydrides proved unfruitful. Nonetheless, with three bridging ligands, the structure of the bimetallic complex verified that the ligand set is geometrically compatible with the corresponding targeted μ-hydrido complexes.

Scheme 1. Synthetic Approaches to Nickel–Iron Dithiolato Hydrides.

Seeking chloride-free precursors, we next examined the related condensation of 1 with “Ni(OTs)₂(dppe),” generated from [Ni(H₂O)₆](OTs)₂¹² and dppe. Treatment of a CH₂Cl₂ solution of Ni(OTs)₂(dppe) with 1 rapidly gave the salt of the dicationic dicarbonyl [(CO)₂(dppe)Fe(pdt)Ni(dppe)]²⁺ ([2(CO)]²⁺), obtained as a mixture of symmetric and unsymmetric stereoisomers. A pair of isomers (in a similar ratio) are also seen in the iron precursor. Thus, it appears that the Ni(dppe)²⁺ module binds to the dithiolate without perturbing the coordination sphere of the Fe(II) center. The dicationic nickel–iron complex [2(CO)]²⁺ is observed as an intermediate in the reaction of 1 and NiCl₂(dppe). Solutions of [2(CO)]²⁺ in CH₂Cl₂ were found to react with NBu₄BH₄ to give a complex mixture. Analysis of this mixture revealed the presence of the targeted cationic hydride [(CO)(dppe)Fe(pdt)(μ-H)Ni(dppe)]⁺ [2H]⁺: a molecular ion was observed in the ESI-MS, and the ¹H NMR spectrum revealed two high-field triplets consistent with two isomers depending on the stereochemistry of the Fe(dppe) center. Solutions of crude [2H]⁺ are more robust than the dicationic precursor consistent with a complex that has three bridging ligands (like the μ-chloride [2Cl]⁺).

Because of the inefficiency of the hydride routes but encouraged by the stability of crude samples of [2H]⁺, we investigated an alternative route to [2H]⁺ starting from more reduced reagents. The Fe(I) complex (CO)₃Fe(pdt)Ni(dppe) (3)¹³ readily protonated with HBF₄·Et₂O in CH₂Cl₂ solution to give the corresponding hydride [3H]⁺ (Scheme 1). When monitored by FT-IR spectroscopy, protonation induces a shift in ν_CO of 54 cm⁻¹. The retention of the ν_CO pattern suggests that the protonation occurs at the Ni–Fe bond. Consistent with this assignment, the hydride region of the ¹H NMR spectrum for[3H]BF₄ displays a triplet (δ –3.53, J_PH ∼6 Hz), and the ³¹P NMR spectrum shows a singlet. Excess acid is not deleterious; i.e. [3H]⁺ does not protonate further, even with a large excess of HBF₄·Et₂O. Solutions of [3H]BF₄ are stable in air for days.

Representing the first example of a nickel–iron thiolato hydride,¹⁴ [3H]BF₄ was characterized crystallographically (Figure 2). Including the bridging hydride ligand, the Fe center is quasi-octahedral and Ni is approximately square pyramidal. The hydride connectivity, the metal coordination numbers, and the presence of three diatomic terminal ligands on Fe match features of the active site of the enzyme in the Ni–R state.^2,9 Completing the synthetic cycle outlined in the scheme, [3H]⁺ reacts with dppe under photochemical conditions to give the hydride [2H]⁺, which exhibits spectroscopic properties seen in the crude samples (high field ¹H NMR, ESI/MS, FT-IR). This substitution method should allow the synthesis of related complexes with a range of ligands on Fe.

Figure 2.

Structure of the cation in [(dppe)NiFe(pdt)(H)(CO)₃]BF₄ ([3H]BF₄). Key distances (Å): Ni–Fe, 2.6131(14); Ni–S(1), 2.210(2); Ni–S(2), 2.219(2); Fe–S(1), 2.321(2); Fe–S(2), 2.322(2); Ni–H, 1.64(6); Fe–H, 1.46(6); S(⋯)S, 3.101. The Ni–Fe distance in [(CO)(dppe)Fe(pdt)(μ-Cl)Ni(dppe)]BF₄ is 3.076 Å.

The complex [3H]⁺ resembles the 34e diiron dithiolato hydrides [Fe₂(pdt)(μ-H)(CO)_6−x(PR₃)_x]⁺, which are considered to be useful models for the [FeFe]-hydrogenases.⁶ Deprotonation of methylene chloride solutions of [3H]⁺ with NEt₃ ([HNEt₃]BF₄, pK_a = 18) were found to rapidly give 3. Complex [3H]⁺ is an active catalyst for the reduction of protons to H₂, as indicated by electrochemical measurements. For a CH₂Cl₂ solution of CF₃CO₂H (pK_a in CH₃CN = 12.65; E° ∼ −0.90 V vs Fc/Fc⁺),¹⁵ a catalytic current is observed near −1.37 V vs Fc/Fc⁺, which is milder than that seen for the diiron complexes (Figure 3). The proposed catalytic cycle begins with protonation of 3 and probably follows the mechanistic pattern described for the substituted diiron catalysts, i.e., a cycle that involves reduction of the hydride followed by protonation.^6,16

Figure 3.

Cyclic voltammetry of [3H]BF₄ in the presence of varying equivalents of CF₃CO₂H. Conditions: 0.001 M [3H]BF₄ in CH₂Cl₂ solution, 0.1 M NBu₄PF₆, scan rate of 100 mV/s.

Complex 3 is also easily oxidized to the monocation, E_1/2∼ −520 V (vs Fc/Fc⁺). Using ferrocenium as the oxidant (FcBF₄), we generated the salt [3]BF₄. This cation exists in the same oxidation state (FeNi)³⁺ assigned to the Ni–L state of the enzyme.⁴ The IR spectrum shows that oxidation shifts ν_CO by ∼29 cm⁻¹. Although the argument is qualitative, the values for ν_CO indicate that one-electron oxidation affects the Fe(CO)₃ center less than protonation, which formally corresponds to a Fe(II) ion. A frozen 8:2 CH₂Cl₂/THF solution of [3]BF₄ exhibits an intense nearly axial EPR spectrum lacking phosphine hyperfine coupling. The EPR and FT-IR properties are similar to those for square pyramidal Fe(I) species.¹⁷ It is likely that replacement of CO ligands in 3 will enable access to higher oxidation states of the NiFe(SR)₂ system beyond the (FeNi)³⁺ state.

The methods described above provide access to functional models for the [NiFe]-hydrogenases. The models are amenable to extensive modifications by variations of the ligands. Our results should encourage the development of a host of biomimetic hydrides, leading to new mechanistic insights relevant to Nature's most pervasive catalysts for processing hydrogen.