Extending the motif of the [FeFe]-hydrogenase active site models: Protonation of Fe₂(NR)₂(CO)_6–xL_x species

Abstract

Studies on diiron dithiolato complexes have proven fruitful for modeling the active site of the [FeFe]-hydrogenases. Here we present a departure from the classical Fe₂S₂ motif by examining the viability of Fe₂N₂ butterfly compounds as functional models for the diiron active site of [FeFe]-hydrogenases. Derivatization of Fe₂(BC)(CO)₆ (1, BC = benzo-[c]-cinnoline) with PMe₃ affords Fe₂(BC) (CO)₄(PMe₃)₂, which subsequently undergoes protonation at the Fe–Fe bond. The hydride [(μ-H)Fe₂(BC)(CO)₄(PMe₃)₂]PF₆ was characterized crystallographically as the C_2v isomer. It represents a rare example of a hydrido diiron complex that exists as observable isomers, depending on the location of the phosphine ligands – diapical and apical–basal. This hydride catalyzes the electrochemical reduction of protons.

In recent years, many compounds of the type Fe₂(S-R)₂(CO)₄L₂ have been prepared and examined as structural and functional models for the active site of [FeFe]-hydrogenases. Particular emphasis has been placed on variations of L, especially tertiary phosphines [1–3], but also isocyanides [4,5], N-heterocyclic carbenes [6–9], and cyanide [10–13]. While much effort has focused on functionalizing the dithiolate ligand [14–20], relatively little work has examined the possibility of replacing the thiolates with other bridging groups [21]. Best and coworkers demonstrated that the corresponding phosphides Fe₂(PR₂)₂(CO)₆ and [Fe₂(PR₂)₂(CO)₆H]⁻ are effective catalysts for hydrogen evolution [22].

Incentives for broadening the range of bridging ligands include changing the steric profile of the catalysts, altering the susceptibility of the bridge toward electrophilic attack (important since these catalysts function in acidic media), incorporating functionality, and manipulating the electronic character (basicity, E_1/2) of the metals. Naturally, a fundamental motivation for investigating new bridging ligands is to elucidate the elementary design rules that will lead to greater mechanistic insight and, ultimately, a deeper understanding of hydrogenogenesis.

Approximately 15 nitrogenous bridges are known to link pairs of iron centers, giving complexes of the general formula Fe₂(NR_1,2)₂(CO)₆ (Scheme 1). Many of the complexes were first reported in the 1960s and 1970s in an era where the interactions of nitrene-forming reagents and metal carbonyls were first systematized. We selected one representative of this series, Fe₂(BC)(CO)₆ [23–26] (1, where BC = benzo-[c]-cinnoline, N₂C₁₂H₈, Scheme 1) to examine as a candidate for a new model class for [FeFe]-hydrogenases. In a structural sense, 1 differs significantly from the well-studied Fe₂(SR)₂(CO)₆ motif. For this reason we felt that its properties would test most demandingly the scope of the hydrogenase behavior of the Fe₂(μ-X)₂L_n systems (L = CO, donor ligands). This report summarizes these results, which indeed demonstrate the prospect that many nitrogenous bridging ligands could be applied to the modeling of hydrogenases.

Scheme 1.

Generalized structures of azodiiron carbonyl complexes and the structure of 1.

Compound 1 was prepared using the original method, using Fe₃(CO)₁₂ in place of Fe(CO)₅ [27]. As for other donor ligands [26,28], PMe₃ displaced two of the CO ligands in 1 to afford the red-colored Fe₂(BC)(CO)₄(PMe₃)₂ (2) (Eq. (1)).¹ 1 is noticeably more electrophilic than analogous diiron dithiolates; disubstitution of the dithiolate system requires more demanding conditions (e.g. 80°C for 9 h vs. 25 °C for 1.5 h). The BC ligand is apparently robustly attached since it was not displaced by the highly nucleophilic PMe₃ ligand.

$$\underset{\mathbf{1}}{{\displaystyle {\text{Fe}}_2(\text{BC}){(\text{CO})}_6}}+\text{xs\hspace{0.17em}}{\text{PMe}}_3\underset{\text{RT}}{\overset{1.5\text{h}}{\to }}\underset{\mathbf{2}}{{\displaystyle {\text{Fe}}_2(\text{BC}){(\text{CO})}_4{({\text{PMe}}_3)}_2}}$$ (1)

Encouraged by the stability of the BC ligand towards nucleophiles, we examined cyanide derivatives. Treatment of 1 with two equiv. Et₄NCN indeed afforded the dicyanide complex [Fe₂(BC)(CO)₄(CN)₂]²⁻ as indicated by IR spectroscopy (MeOH: ν_CN = 2106; ν_CO = 2039, 2020, 1998, 1968, 1938, 1879, 1814 cm⁻¹). The appearance of numerous carbonyl bands is attributed to the presence of multiple isomers in solution, as also observed in phosphine derivatives discussed below. The dicyanide product proved to be unstable, even in the solid state. A stable monocyanide derivative was prepared via the reaction of 1 with KN(SiMe₃)₂ [11]. Similarly, treatment of 1 with Me₃NO gave the MeCN complex Fe₂(BC)(CO)₅(MeCN) (ν_CO = 2043, 2030, 1976, 1964, 1913 cm⁻¹) [29], which was found to react with Et₄NCN to afford Et₄N[Fe₂(BC) (CO)₅(CN)], as indicated by IR spectroscopy and ESI-MS.²

It was of interest to determine if the diiron azo system 2 could be protonated at the Fe–Fe bond and, if so, would the resulting hydride electrocatalyze proton reduction, analogous to the reactivity observed in Fe₂(SR)₂ systems [29–31]. Indeed, we found that HCl effected protonation at the Fe–Fe bond of 2, thereby forming a rare example of a (μ-hydrido)diiron azo system, [2H]⁺.³ The complex was isolated as its ${\text{PF}}_6^{-}$ salt from H₂O/MeOH solution [32]. Interestingly, protonation of 2 afforded two solution isomers (Scheme 2), as indicated by doublet-of-doublet (unsymmetrical isomer) and triplet (symmetrical isomer) ¹H NMR signals for the hydride, in a ratio of 1.0:1.8, respectively. In contrast, protonation of Fe₂(S₂C₂H₄)(CO)₄(PMe₃)₂ and Fe₂(S₂C₃H₆)(CO)₄(PMe₃)₂ gives only a trans-dibasal isomer, even at low temperatures [32]. Also, (μ-H)Fe₂(S₂C₃H₆)(CO)₄(CN)(PMe₃) exists as only one predominant isomer [1]. The only case where multiple isomers have been detected was in [(μ-H)Fe₂(S₂C₃H₆)(CO)₄(CN)₂]⁻, which was not isolated but characterized in solution [32].

Scheme 2.

Protonation reaction of 2 to form [2H]⁺.

The structure of the major (symmetrical) isomer was confirmed crystallographically (Fig. 1). Both PMe₃ ligands were found to be in the apical position, i.e. trans to the hydride. For comparison, diiron dithiolate bis(phosphine) complexes rarely adopt a diapical structure, e.g. Fe₂ (S₂C₃H₆)(CO)₄(PMe₂Ph)₂ [3], and examples are not known of protonated (i.e. μ-hydride) species exhibiting diapical phosphine ligands.

Fig. 1.

Two views of the structure of the cation in [(μ-H)Fe₂(N₂C₁₂H₈)(CO)₄(PMe₃)₂]PF₆ with thermal ellipsoids set at the 50% probability level.

Selected parameters derived from crystallographic measurements of [2H]PF₆ and Darensbourg’s [(μ-H)Fe₂ (S₂C₂H₄)(CO)₄(PMe₃)₂]PF₆ [2], featuring trans-dibasal phosphine ligands, are listed in Table 1. Comparing these data, significant differences are observed only for the Fe–E (where E = N or S) distances and angles. Similar to other diiron azo complexes, [2H]⁺ exhibits shorter distances to iron than do thiolate ligands. The most striking difference between the Fe₂N₂ and Fe₂S₂ butterfly cores is the E–Fe–E angle, wherein the N–Fe–N angle is ~40° more acute than the S–Fe–S angle. The short Fe–N distances and the small N–Fe–N angles are compensated by expanded Fe–N–Fe angles, thereby minimizing any effect on the Fe–Fe distance. Indeed, all Fe-ligand distances are virtually identical for the azo- and dithiolato-bridged systems. Thus, the structural similarities between the azo and dithiolate systems indicate the promise of a still wider range of bridging ligands in catalysts that effect hydrogenogenesis.

Table 1.

Selected distances (Å) and angles (deg) for [2H]PF₆ and [(μ-H)Fe₂(S₂C₂H₄)(CO)₄(PMe₃)₂]PF₆

[2H]PF6		[μ-H)Fe2(S2C2H4)(CO)4(PMe3)2]PF6
Fe–Fe	2.5599(8)	2.5742(13)	Fe–Fe
Fe–H	1.62(3)	1.69(6)	Fe–H
Fe–Papical	2.2362(12)	2.2392(16)	Fe–Pbasal
Fe–C	1.774(4)	1.770(6)	Fe–C
Fe–N	1.927(5)	2.2562(6)	Fe–S
N–N	1.399(4)	2.924	S–S
Fe–N–Fe	82.99(11)	69.57(5)	Fe–S–Fe
N–Fe–N	42.53(11)	80.78(6)	S–Fe–S
C–Fe–P	90.46(15)	94.8(4)	C–Fe–P

We confirmed that [2H]⁺ catalyzes hydrogen evolution from p-toluenesulfonic acid (HOTs) via experiments conducted electrochemically. The cathodic current at ~−1.1 V vs. Ag/AgCl grew proportionally with additional equivalents of HOTs until ca. 4 equiv., at which stage [2H]⁺ decomposes (Fig. 2). The overpotential of electrocatalysis is virtually identical to that observed for diiron dithiolate systems under the same reaction conditions.

Fig. 2.

Cyclic voltammagrams vs. Ag/AgCl of 2 + x HOTs, x = 0–3 equiv., in MeCN with Bu₄NPF₆.

The substitutional behavior of the diiron azo complexes bears both similarities and differences with the more commonly studied dithiolate complexes. Both the azo- and thiolato- bridged species undergo a maximum of disubstitution with conventional unidentate ligands. The Fe₂N₂ system differs significantly from the well-studied Fe₂(SR)₂ system in terms of the rates of substitution and the isomeric content. As shown by Ellgen, the activation enthalpies for substitution of Fe₂(BC)(CO)₆ are ca. 2 kcal/mol more favorable relative to the corresponding dithiolato complexes [26,33]. It is significant that [(μ-H)Fe₂(S₂C_nH_2n)(CO)₄ (PMe₃)₂]⁺ exists as only one isomer, but the BC analogue also features the diapical isomer. This difference highlights the significant influence of the steric properties of the bridging ligands on the apical coordination sites trans with respect to the Fe–Fe bond. This steric interaction is greater when the two iron centers are bridged by three ligands, as in complexes of the type [Fe₂(SR)₂(μ-X)L₆]^z (X = CO, H).

Taking advantage of the fact that [2H]PF₆ crystallizes as a single isomer, we dissolved crystals of this compound in CD₃CN at 0 °C to obtain the spectrum of this isomer. Indeed the solution exhibited one triplet in the hydride region of the ¹H NMR spectrum. Over a period of minutes, this symmetrical complex isomerized via a 1st order process to an equilibrium mixture of symmetrical and unsymmetrical isomers. The stability of multiple isomers is consistent with the minimal steric pressure exerted by the planar BC ligand on the ligands that occupy the apical sites. As we have discussed previously, protonation narrows the distance between apical ligands and the bridging ligand, which usually destabilizes large apical ligands [34]. Another case where multiple isomers of a substituted μ-hydride have been observed is [(μ-H)Fe₂(S₂C₃H₆)(CO)₄(CN)₂]⁻, which features the small cyanide ligands in the apical position.

In summary, a new class of [FeFe]-hydrogenase active site models, based on a Fe₂N₂ butterfly core, has been established. These diiron azo compounds catalyze the electrochemical reduction of protons to H₂. Although the catalysts based on BC are less robust than the dithiolate complexes, there is considerable scope for improvement. A wide range of azo-bridged diiron complexes is known [24,27,35–37], including ligands such as chloro-substituted BC [27], the sterically imposing 2,3-diazabicyclo[2.2.1]hept- 2-ene (DBH) [26,38], and PhN(X)NPh (where X = CO, SO₂) [39–41].

Supplementary Material

1. Appendix A. Supplementary data.

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jinorgbio.2007. 05.005.