Thiolate-bridged dinuclear iron(tris-carbonyl)–nickel complexes relevant to the active site of [NiFe] hydrogenase

Abstract

The reaction of NiBr₂(EtOH)₄ with a 1:2–3 mixture of FeBr₂(CO)₄ and Na(SPh) generated a linear trinuclear Fe–Ni–Fe cluster (CO)₃Fe(μ-SPh)₃Ni(μ-SPh)₃Fe(CO)₃, 1, whereas the analogous reaction system FeBr₂(CO)₄/Na(S^tBu)/NiBr₂(EtOH)₄ (1:2–3:1) gave rise to a linear tetranuclear Fe–Ni–Ni–Fe cluster [(CO)₃Fe(μ-S^tBu)₃Ni(μ-Br)]₂, 2. By using this tetranuclear cluster 2 as the precursor, we have developed a new synthetic route to a series of thiolate-bridged dinuclear Fe(CO)₃–Ni complexes, the structures of which mimic [NiFe] hydrogenase active sites. The reactions of 2 with SC(NMe₂)₂ (tmtu), Na{S(CH₂)₂SMe} and ortho-NaS(C₆H₄)SR (R = Me, ^tBu) led to isolation of (CO)₃Fe(μ-S^tBu)₃NiBr(tmtu), 3, (CO)₃Fe(S^tBu)(μ-S^tBu)₂Ni{S(CH₂)₂SMe}, 4, and (CO)₃Fe(S^tBu)(μ-S^tBu)₂Ni{S(C₆H₄)SR}, 5a (R = Me) and 5b (R = ^tBu), respectively. On the other hand, treatment of 2 with 2-methylthio-phenolate (ortho-O(C₆H₄)SMe) in methanol resulted in (CO)₃Fe(μ-S^tBu)₃Ni(MeOH){O(C₆H₄)SMe}, 6a. The methanol molecule bound to Ni is labile and is readily released under reduced pressure to afford (CO)₃Fe(S^tBu)(μ-S^tBu)₂Ni{O(C₆H₄)SMe}, 6b, and the coordination geometry of nickel changes from octahedral to square planar. Likewise, the reaction of 2 with NaOAc in methanol followed by crystallization from THF gave (CO)₃Fe(μ-S^tBu)₃Ni(THF)(OAc), 7. The dinuclear complexes, 3-7, are thermally unstable, and a key to their successful isolation is to carry out the reactions and manipulations at −40°C.

Hydrogenases catalyze reversible oxidation/reduction of molecular-hydrogen/protons in nature (1–4), and are classified into three groups on the basis of their metal contents, namely, the [NiFe] and [FeFe] hydrogenases (5–12), and the Fe–S cluster-free hydrogenase (Hmd) (13). Among them, the [FeFe] and [NiFe] hydrogenases have been studied relatively well, and have posed a challenge for inorganic chemists to synthesize structural models useful for understanding their functions. Although variations of [NiFe] hydrogenases have been reported/postulated, their active sites conserve a common structural feature in that four cysteines (or three cysteines and one selenocysteine) are bound to the dinuclear Fe–Ni core and the iron atom carries CO and CN ligands (14, 15). Two of the cysteines, or one cysteine and one selenocysteine, are coordinated to Ni, whereas sulfur atoms of the other two cysteines bridge Fe and Ni. In the oxidized forms, there is an additional atom X located at a bridging position, interpreted as O for Desulfovibrio gigas and Desulfovibrio fructosovorans and S for Desulfovibrio vulgaris Miyazaki F(6, 7, 10). The theoretical and spectroscopic studies for D. gigas, D. fructosovorans, Chromatium vinosum, and D. v. Miyazaki F suggested the “O”-bridge to be OH (16–19). On the other hand, the structure of the reduced form does not explicitly show an additional bridge, and the Fe–Ni distance is short (2.5–2.6 Å) relative to that of the oxidized forms (2.8–2.9 Å). There is speculation that a hydride may exist at a bridging position (20–22) (Fig. 1).

Fig. 1.

The active site of [NiFe] hydrogenase

The dinuclear Fe–Ni complexes previously reported as models of the active site of [NiFe] hydrogenase are limited to those containing multidentate amine-thiolate ligands, phosphines, or nitric oxide at either the Ni or Fe site (23–29). Synthesis of better models is thus highly desired, to provide clues to understand the bonding properties and function of the active site. Our strategy of the synthesis of model Fe–Ni complexes have been based on the following two approaches. One approach is to use an Fe(II) complex carrying carbonyl and cyanide ligands, [Fe(CO)₃(CN)₂Br]⁻, as a precursor of the Fe site, and we have reported the isolation of [(CO)₂(CN)₂Fe(μ-pdt)Ni(S₂CNR₂)]⁻ (pdt = 1,3-propanedithiolate; S₂CNR₂ = dithiocarbamate) from the reaction of the iron complex with pdt²⁻ and (PPh₃)NiBr(S₂CNR₂) (30). The other approach is to use more accessible iron Tris-carbonyl complexes as building blocks, which is the focus of this article. The tetranuclear Fe–Ni–Ni–Fe cluster [(CO)₃Fe(μ-S^tBu)₃Ni(μ-Br)]₂, 2, derived from FeBr₂(CO)₄, NaS^tBu, and NiBr₂(EtOH)₄, was found to serve as a convenient entry into a series of thiolate-bridged Fe(CO)₃–Ni complexes. The resulting thiolate-bridged Fe(CO)₃–Ni complexes mimic well the core structure and ligand environments of the active site of [NiFe] hydrogense. Interestingly, the dinuclear complexes exhibit a variety of coordination geometries at Ni, ranging from square planar, to distorted square pyramidal, and to octahedral. Although the reactivity of these complexes remains to be elucidated, the geometrical flexibility at Ni implies that take-up/release of substrates such as molecular hydrogen may occur at the Ni site of [NiFe] hydrogenase.

Results and Discussion

Synthesis of Linear Fe–Ni–Fe and Fe–Ni–Ni–Fe Clusters.

It was reported that the reaction of FeBr₂(CO)₄ with 3 equiv of benzenethiolate (SPh⁻) produced a thermally unstable Fe(II) carbonyl/thiolate complex, formulated as either Fe(CO)₄(SPh)₂ or [Fe(CO)₃(SPh)₃]⁻ (31). Although [K(18-crown-6)][fac-Fe(CO)₃(SPh)₃] was isolated from the other route by using Fe₃(CO)₃(SPh)₆ and 3 equiv of NaSPh, the product can only be stored in solution at less than −10°C (32). We thus examined the reactions of NiBr₂(EtOH)₄ with Fe(II) carbonyl/thiolate complexes generated in situ from FeBr₂(CO)₄ and NaSR (R = Ph, ^tBu) at low temperature.

First, the reaction of NiBr₂(EtOH)₄ with 1 equiv of FeBr₂(CO)₄ and 2 equiv of NaSPh was carried out at −40°C. From the resulting brown solution, a thiolate-bridged trinuclear Fe–Ni–Fe cluster (CO)₃Fe(μ-SPh)₃Ni(μ-SPh)₃Fe(CO)₃, 1, was isolated as reddish brown crystals in 41% yield based on iron. When the amount of NaSPh was increased to 3 equiv, uncharacterizable byproducts precipitated, and the yield of 1 turned out to be lower. The Fe(II)/Ni(II) oxidation states of FeBr₂(CO)₄ and NiBr₂(EtOH)₄ are retained in 1, and the octahedral Ni(II) at the center makes the trinuclear complex paramagnetic, as was indicated by the broad ¹H NMR signals at δ 18.9, 10.2, and −2.4. The CO bands of 1 appear at 2,077 and 2,025 cm⁻¹ in the IR spectrum, and the observed ν(CO) values are similar to those reported for analogous clusters (CO)₃Fe(μ-SC₄H₂O-o-Me)₃Ni(μ-SC₄H₂O-o-Me)₃Fe(CO)₃ (33) (2,076 and 2,023 cm⁻¹) and (CO)₃Fe(μ-SePh)₃Ni(μ-SePh)₃Fe(CO)₃ (34) (2,070 and 2,019 cm⁻¹) (Scheme 1).

Scheme. 1.

Next, we examined the analogous reaction system FeBr₂(CO)₄/NaS^tBu/NiBr₂(EtOH)₄ (1:2:1), where 2 equiv of NaS^tBu was used instead of NaSPh. Interestingly, from this reaction, a tetranuclear Fe–Ni–Ni–Fe cluster [(CO)₃Fe(μ-S^tBu)₃Ni(μ-Br)]₂, 2, was obtained as brown crystals in 53% yield. Apparently, one of the bromides in NiBr₂(EtOH)₄ is retained, and it bridges the two Ni atoms of the tetranuclear Fe–Ni–Ni–Fe spine of 2. Again, increasing NaS^tBu to 3 equiv resulted in a lower yield of 2 because of precipitation of insoluble by-products. The ¹H NMR spectrum features broad signals appearing at δ 36.5, 30.7, and 26.3, suggesting the paramagnetic nature of 2, presumably owing to the five-coordinate Ni(II) centers.

Although the two reactions above gave rise to products of different nuclearity, a thiolate-bridged dinuclear Fe–Ni complex “Fe(CO)₃(μ-SR)₃NiBr” might be formed initially, in either case, from the reaction between NiBr₂(EtOH)₄ and preformed [Fe(CO)₃(SR)₃]⁻. Then the Fe–Ni intermediate complex may react with [Fe(CO)₃(SR)₃]⁻ to give an Fe–Ni–Fe trinuclear cluster, or two molecules of the Fe–Ni complex may be assembled through Ni–Br interactions to give a Fe–Ni–Ni–Fe tetranuclear cluster. The two reaction pathways are likely to be competitive, and the former pathway is favored for R = Ph. In contrast to the PhS analogues, the bulky ^tBu group appears to deter the reaction of Fe(CO)₃(μ-SR)₃NiBr with [Fe(CO)₃(SR)₃]⁻, and favors the formation of the Fe–Ni–Ni–Fe tetranuclear cluster 2.

Molecular Structure of 2.

In the tetranuclear structure of 2, the Fe and Ni atoms are bridged by three tert-butyl thiolates, and the two Ni atoms are linked by two bromides (Fig. 2). The Fe atoms are additionally bound to three CO ligands in a facial manner to complete an octahedral geometry. The Ni centers adopt distorted square-pyramidal geometries, each with two thiolate sulfurs and two bromo ligands at the basal sites, and S1 or S4 occupies the axial site in the molecule. Interestingly, the axial Ni–S bonds [2.2957 (15) ∼ 2.3184 (15) Å] are notably shorter than the basal Ni–S bonds [2.3560 (13) ∼ 2.3770 (13) Å].

Fig. 2.

Molecular structure of [(CO)₃Fe(μ-S^tBu)₃Ni(μ-Br)]₂, 2, with thermal ellipsoids at the 50% probability level. One of the crystallographically independent molecules is shown.

The Fe–Ni distances are longer than 3 Å, and the Ni–Ni separation is even longer [3.6321 (7) ∼ 3.6744 (9) Å]. The Ni₂Br₂ quadrilaterals deform from ideal rhombuses, where one Ni–Br bond is notably different from the other. Thus, cluster 2 may be considered as a weakly bound dimer of “(CO)₃Fe(μ-S^tBu)₃NiBr,” connected by Ni–Br interactions. The (CO)₃Fe(μ-S^tBu)₃NiBr fragment of 2 resembles the structures of the oxidized forms of [NiFe] hydrogenases, in that it consists of a square pyramidal nickel and an octahedral iron, which is coordinated by three diatomic terminal ligands and three bridging ligands.

Synthesis of Dinuclear Fe–Ni Complexes from 2.

Viewing 2 as a dimer of a dinuclear Fe–Ni complex, we anticipated that the tetranuclear structure could be split into two components by addition of a donor ligand. Indeed, treatment of 2 with SC(NMe₂)₂ (tmtu) at −40°C led to the formation of the dinuclear complex (CO)₃Fe(μ-S^tBu)₃NiBr(tmtu), 3. The IR spectrum of 3 in KBr exhibits strong ν(CO) bands at 2,073 and 2,006 cm⁻¹, a moderate band at 2,017 cm⁻¹, and a weak shoulder at 1,973 cm⁻¹. The IR spectrum in the solid state is similar to, but distinct from, that of 2. Interestingly, however, the IR spectra for 2 and 3 in cold THF (−40°C) are identical, showing bands at 2,066, and 2,005 cm⁻¹. This may indicate that the tmtu ligand at Ni of 3 is replaced by THF to generate (CO)₃Fe(μ-S^tBu)₃NiBr(THF), and that the tetranuclear cluster 2 dissociates into the same dinuclear complex in THF (Scheme 2).

Scheme 2.

Encouraged by the isolation of 3, we next examined the reactions of 2 with thioether-thiolate hybrid ligands. Anionic bidentate ligands are capable of splitting the tetranuclear array by substituting the Br site with the thiolate S, and a further coordination of the thioether S at Ni would stabilize the resulting dinuclear Fe–Ni complexes because of the chelate effect. As a matter of fact, the Fe(II)–Ni(II) complexes, (CO)₃Fe(S^tBu)(μ-S^tBu)₂Ni{S(CH₂)₂SMe}, 4, and (CO)₃Fe(S^tBu)(μ-S^tBu)₂Ni{S(C₆H₄)SR}, 5a (R = Me) and 5b (R = ^tBu), were isolated in 60–77% yields, from the reactions of 2 with 2-methylthio-ethanethiolate NaS(CH₂)₂SMe and ortho-RS-benzenethiolates NaS(C₆H₄)SR (R = Me, ^tBu), respectively. The IR spectrum of 4 in KBr shows CO stretching vibrations as intense bands at 2,056 and 2,006 cm⁻¹ and a weak band at 1,959 cm⁻¹. Similar CO bands were observed for 5b, which include two strong (2,056 and 2,006 cm⁻¹) and one weak band (1,963 cm⁻¹). However, more CO signals appeared in the solid-state IR spectrum of 5a, namely, strong bands at 2,058, 2,009, 2,002, and 1,992 cm⁻¹ and a weak band at 1,961 cm⁻¹, whereas the solution IR spectrum in cold THF exhibits two strong (2,065 and 2,009 cm⁻¹) and one weak (1,977 cm⁻¹) band.

The molecular structures of the dinuclear Fe–Ni complexes, 3, 4, 5a, and 5b have been determined by x-ray crystallography. Representative drawings of 3 and 5a are shown in Figs. 3 and 4, respectively. The structure of 3 is similar to each Fe–Ni half of the tetranuclear complex 2. It is also relevant to the oxidized form of [NiFe] hydrogenases in that an octahedral iron and a distorted square pyramidal nickel are linked by three bridging ligands. In the distorted square-pyramidal Ni site, the base is formed from two bridging thiolate sulfurs, S2 and S3, the Br and the tmtu sulfur atoms, whereas the other bridging sulfur, S1, is axial. Like 2, the axial Ni–S1 bond is shorter than the basal Ni–S2 and Ni–S3 bonds in 3.

Fig. 3.

Molecular structure of (CO)₃Fe(μ-S^tBu)₃NiBr(tmtu), 3, with thermal ellipsoids at the 50% probability level.

Fig. 4.

Molecular structure of (CO)₃Fe(S^tBu)(μ-S^tBu)₂Ni{S(C₆H₄)SMe}, 5a, with thermal ellipsoids at the 50% probability level.

In contrast to the structure of 3, the Ni and Fe atoms are bridged by two thiolate ligands for 4, 5a, and 5b, and one thiolate is bound to the Fe site as a terminal ligand. Thus, coordination of the thioether-thiolate hybrid ligands at Ni alters the dinuclear Fe–Ni structure substantially. The tert-butyl groups of the two bridging thiolates orient differently, one with an upright conformation and the other with a sideways conformation. The Ni atom is coordinated by four sulfur donors in a somewhat distorted square planar geometry. Interestingly, the terminal thiolate is more weakly bound to Fe than the bridging thiolates in either 4, 5a, or 5b, and the Fe-S(terminal) bond length is ≈0.02 Å longer than the average Fe–μS length. The Fe–Ni distances are all longer than those of 2 and 3.

Dinuclear Fe–Ni Complexes with Octahedral Nickel Sites.

In analogy to the preparation of dinuclear Fe–Ni complexes, 4 and 5a∼b, we attempted to incorporate a bidentate phenolate-thioether ligand. Treatment of 2 with 2-methylthio-phenolate [ortho-O(C₆H₄)SMe] in methanol gave the dinuclear complex (CO)₃Fe(μ-S^tBu)₃Ni(MeOH){O(C₆H₄)SMe}, 6a. As shown in a detailed account of the molecular structure of 6a later in this section, the Ni atom is hexa-coordinate, ligated by a thioether sulfur, a phenolate oxygen, and a methanol oxygen in addition to three bridging thiolate sulfurs. The coordinated methanol in 6a is labile, and is readily removed under reduced pressure to afford (CO)₃Fe(S^tBu)(μ-S^tBu)₂Ni{O(C₆H₄)SMe}, 6b (Scheme 3). Interestingly, liberation of MeOH from 6a is accompanied by the cleavage of one of the Ni–μS bonds, leaving a square planar Ni center and a terminal tert-butyl thiolate on Fe. Conversion between 6a and 6b was found to be reversible, and thus the CO stretches for 6a [2,067(s), 2,013(s), 2,004(s), and 1,973(w) cm⁻¹] appeared on addition of MeOH to 6b, which exhibits intense CO bands at 2,058, 2,011, and 2,000 cm⁻¹ together with a weak band at 1,963 cm⁻¹. This reversible geometrical change at the Ni site is intriguing, because this may be a possible mechanism by which [NiFe] hydrogenase opens a binding site for an incoming H₂ molecule. The existence of both octahedral and square planar Ni(II) centers in 6a and 6b is indicative of a small energy difference between high-spin and low-spin electronic configurations that is attributed to the coordination of the phenolate O donor. The weak Ni-O(phenolate) bonding, relative to the Ni-S(thiolate) bonding, would lead to the square planar geometry of 6b being less stable, and being more prone to accommodate axial ligands, than the structurally very similar 5a. In fact, thermal decomposition of 6b gradually occurs in the solid state at room temperature to afford uncharacterizable insoluble materials, whereas 5a can be stored for months at ambient temperature in the crystalline form.

Scheme 3.

In a manner similar to the preparation of 6a, the dinuclear Fe–Ni complex, (CO)₃Fe(μ-S^tBu)₃Ni(THF)(OAc), 7, was isolated as yellowish brown crystals, from the reaction of 2 with NaOAc followed by crystallization from THF (Scheme 4). Complex 7 again has a hexa-coordinate Ni, to which two acetate oxygen atoms, a THF oxygen, and three bridging thiolate sulfurs are bound. The infrared spectrum of 7 exhibits intense carbonyl bands at 2,069, 2,011, and 1,998 cm⁻¹ and a weak band at 1,969 cm⁻¹, the frequencies of which are similar to those of 6a. In contrast to 6a, the axial coordination of THF in 7 remains intact under reduced pressure. We reason the trend based on the different strength of ligand fields created by the O,O-chelate of acetate and the S,O-chalate of 2-methylthio-phenolate. The weak ligand field of acetate would make the high-spin octahedral arrangement of Ni stable for 7.

Scheme 4.

The crystal structures of 6a, 6b, and 7 were determined by x-ray analysis, and a perspective view of 6a is shown in Fig. 5. The slightly distorted octahedral geometry at Ni is completed by coordination of a bidentate thioether-phenolate and methanol for 6a, and a bidentate acetate and THF for 7. This structural feature is relevant to those proposed for the active site of the NAD-reducing soluble [NiFe] hydrogenase, in which the Ni site is proposed to be octahedral with O donors (35). The Ni–μS distances of 6a and 7 [2.3877 (8)–2.4657 (10) Å] are distinctly longer than those in 4, 5, and 6b [2.195 (3)–2.260 (1) Å], because the ionic radii of high-spin octahedral Ni(II) ions are known to be substantially larger than those of low-spin square planar Ni(II) ions. They are also slightly longer than those in 2 and 3 [2.308 (3)–2.4135 (18) Å], where the coordination geometry at Ni are square pyramidal. On the other hand, the Fe–μS bond lengths are similar among 2–7, falling in a narrow range of 2.326 (3)∼2.3547 (7) Å, because the (CO)₃Fe(μ-S^tBu)₃ coordination geometry is conserved. An interesting structural feature of 6a is the formation of intermolecular hydrogen bonds in crystals, with O–O short contacts being 2.645 (4)–2.747 (4) Å, linking the coordinated methanol, lattice methanol, and the phenolate oxygen atom of another molecule. This hydrogen bonding also indicates that the phenolate oxygen in 6a may act as a potential proton acceptor. This is interesting regarding the function of [NiFe] hydrogenases, because one of the important steps of the heterolytic activation of H₂ by the enzyme is the deprotonation of a presumably nickel-bound H₂ by a cysteine thiolate, leading to provide Ni–H and S–H bonds (36, 37).

Fig. 5.

Molecular structure of (CO)₃Fe(μ-S^tBu)₃Ni(MeOH){O(C₆H₄)SMe}, 6a, with thermal ellipsoids at the 50% probability level.

Relevance to [NiFe] Hydrogenase.

In the active site of [NiFe] hydrogenase, the Ni center is suggested to be a possible binding site for incoming H₂ (38, 39). Supporting experimental evidence has been provided by the structural study of D. V. Miyazaki F (40), which revealed that the enzyme was deactivated under a CO atmosphere because of coordination of CO to the Ni site. In this regard, reversible coordination of MeOH to the Ni atom of 6b, along with the variegated Ni coordination geometries found for 3, 4, 5a,b, 6a,b, and 7, raise the interesting possibility that the flexible coordination mode of Ni may be an important factor for H₂ activation by [NiFe] hydrogenases. Considering the range of structures found in this study for the Fe–Ni complexes, we propose that the coordination of H₂ could be promoted by the change of Ni geometry from square pyramidal to octahedral.

Our proposed mechanism for hydrogen binding and heterolysis is schematically shown in Scheme 5, where the bridging ligand X is assumed to be OH. The bridging OH ligand has been indicated to occupy the basal position of nickel in [NiFe] hydrogenases (6, 10, 16–19). It has been suggested that heterolysis of H₂ is a nonredox process (41–44), involving proton transfer to a base from the Ni-bound H₂. In the mechanism suggested in Scheme 5, the μ-OH ligand serves as such a base, giving rise to a water molecule and a Fe–Ni complex with a μ-hydride ligand. In fact, we have recently demonstrated the reversible heterolysis of H₂ mediated by the bridging OH group of a Ge–(μ-OH)–Ru complex, generating H₂O and a dinuclear Ge–Ru complex with a μ-hydride ligand. The resulting Ge–(μ-H)–Ru complex reacts with water, and the Ge–(μ-OH)–Ru complex and H₂ are regenerated (45). An analogous H₂ activation has also been reported by using a dinuclear Ni–Ru aqua complex to give a proton and a Ni–(μ-H)–Ru complex (46). The relevant Fe–(μ-H)–Ni form has been postulated as one of the states of [NiFe] hydrogenase (20–22), and the reversible conversion between Fe–(μ-H)–Ni and Fe–(μ-OH)–Ni forms may account for existence of the various Ni-SI states such as Ni-SU, Ni-SIr, and Ni-SIa.

Scheme 5.

A possible mode of hydrogen binding and heterolysis in [NiFe] hydrogenase.

Alternatively, the cysteine sulfur [Cys-530 in D. gigas (6)] may act as a proton acceptor (Scheme 5). In contrast to the liberation of water via protonation of the μ-OH ligand, proton transfer to the Cys-530 sulfur would produce a thiol moiety at the vicinity of the Ni site. If the thiol proton is readily transferred back to Ni, a heterolytic H₂ activation occurs reversibly. Such reversible H–H heterolysis induced by a metal-thiolate group has been proposed because of the observation of facile H/D exchange reactions of various transition metal thiolate complexes with H₂/D₂ in the presence of water or alcohols (47–49).

Concluding Remarks

A series of thiolate-bridged, dinuclear (CO)₃Fe(II)–Ni(II) complexes, 3–7, have been prepared as structural models of the active site of [NiFe] hydrogenase. The dinuclear complexes are all thermally unstable in solution, and yet they can be synthesized and manipulated at −40°C. This is in contrast to our (CO)₂(CN)₂Fe(II)–Ni(II) models reported previously (30), which can be handled in solution at 0°C for several hours without degradation. One possible interpretation for the different thermal stability would be the number of CO ligands on Fe(II). Fewer Fe(II)-bound CO ligands results in stronger π-back donation per CO ligand from the Fe(II) center. In this context, thermal stabilization of the low-spin (CO)(CN)₂Fe center may be a reason behind the fact that the Fe site of [NiFe] hydrogenases carry both CN and CO ligands.

Our study also reveals that the coordination mode of the Ni(II) center can vary from square planar, to distorted square pyramidal, and to octahedral geometries, dependent on the nature of the ligands, whereas the (CO)₃Fe(II) site is octahedral in all cases. Such flexible coordination structures at Ni could be responsible for the interconversion between H₂ and (2H⁺ + 2e⁻) as the unique function of the [NiFe] hydrogenase. The Fe–Ni complexes reported in this article exhibit important structural features of the active sites, and they present the possibility of reaction chemistry relating to the function of the [NiFe] hydrogenase.

Materials and Methods

All reactions and the manipulations were performed under nitrogen or argon atmospheres by using standard Schlenk techniques. Solvents were dried, degassed, and distilled from CaH₂ (CH₂Cl₂), or from Mg turnings (methanol) under nitrogen. Hexane, ether, toluene, THF, and CH₃CN were purified by columns of activated alumina and a supported copper catalyst supplied by Hansen & Co. Ltd.

All experimental conditions and procedures, spectroscopic data, and details of characterization of complexes are given in the supporting information (SI) Materials and Methods. Molecular structures of 1, 4, 5b, 6b, and 7 are given in Figs. S1–S6, respectively, IR spectra of 1–7 are shown in Figs. S7 and S8, and selected bond distances and angles for 1–7 are listed in Tables S1–S5. Crystal data are summarized in Table S6.