Mild Redox Complementation Enables H₂ Activation by [FeFe]-Hydrogenase Models

Abstract

Mild oxidants such as [Fe(C₅Me₅)₂]⁺ accelerate the activation of H₂ by [Fe₂[(SCH₂)₂NBn](CO)₃(dppv)(PMe₃)]⁺ ([1]⁺). The reaction is first order in [1]⁺ and [H₂] but is independent of the E_1/2 and concentration of the oxidant. The analogous reaction occurs with D₂ and proceeds with an inverse isotope effect of 0.75(8). The activation of H₂ is further enhanced with the tetracarbonyl [Fe₂[(SCH₂)₂NBn](CO)₄(dppn)]⁺ ([2]⁺), the first crystallographically characterized H_ox model containing an amine cofactor. These studies point to rate-determining binding of H₂ followed by proton-coupled electron-transfer (PCET). In comparison with [1]⁺, the rate of H₂ activation by [2]⁺/Fc⁺ is enhanced by 10⁴ (25 °C).

The hydrogenases (H₂ases) are attractive targets for synthetic modeling because they catalyze the redox of H₂/H⁺, an important and topical reaction.¹ The [FeFe]- (and [NiFe]-) hydrogenases operate by the combined action of acid-base and electron-transfer. As has been previously shown by both biophysical studies² and synthetic modeling,³ the catalytic properties of the active site of the [FeFe] enzyme are enabled by the juxtaposition of functional groups dedicated to substrate binding, specifically the azadithiolate cofactor and the distal Fe center. This active site also features two redox-active components, the Fe₂(SR)₂ and the Fe₄S₄ subsites, each of which provides 1e⁻ required for the two-electron H₂/2H⁺ couple. In recent years, the advantageous cooperative reactivity of the amine cofactor and one Fe center has been demonstrated in models,⁴ which enables highly active proton-reduction catalysts. Unsolved in previous models is the ability of the same enzyme to activate H₂, an excellent substrate for the enzyme.^2b,2c

The activation of H₂ by diiron models requires that the Fe₂ center be (i) sufficiently electrophilic to attract H₂ but (ii) not so electrophilic to induce binding of the amine to Fe.⁵ For a variety of ligands, the mixed-valence complexes of the type [Fe₂[(SCH₂)₂NR](CO)_3-x(PR₃)_x]⁺ almost satisfy these criteria, but such models are very slow to activate H₂, requiring high pressures and many hours. In this report we show that rapid H₂ activation by these H_ox models can be achieved by the addition of a mild and fast oxidant.

Previously, we showed that the H_ox model [Fe₂[(SCH₂)₂NBn](CO)₃(dppv)(PMe₃)]⁺ ([1]⁺, dppv = cis-C₂H₂(PPh₂)₂, Figure 1) reacts with H₂ only slowly (>26 h, 25 °C, 1800 psi H₂).⁶ We have now discovered that the same complex in the presence of 1 equiv of the mild oxidant [Fe(C₅Me₅)₂]BAr^F₄ (Ar^F = 3,5-C₆H₃(CF₃)₂) reacts with 2 atm of H₂ quantitatively at 25 °C in hours to give the diferrous hydride product. The nearly isosteric complex that lacks the amine cofactor, [Fe₂(S₂C₃H₆)(CO)₃(dppv)(PMe₃)]⁺,⁵ is unreactive toward H₂ under the same conditions.

Figure 1.

Active site of the H_ox state of [FeFe]-hydrogenase (left) and the model [1]⁺ (right, R = CH₂Ph).

To simplify the analysis of the reaction by proton was trapped as [HP(o-tol)₃]⁺. Heterolytic cleavage in the presence of P(o-tol)₃ produced known hydride [1H]⁺ and [HP(o-tol)₃]⁺, which undergoes slow proton transfer on the NMR timescale and displays distinctive ¹H and ³¹P NMR signals. ²H NMR analysis of the same reaction in CH₂Cl₂ solution using D₂ showed equal deuterium incorporation into [1D]⁺ and coproduct [DP(o-tol)₃]⁺ (Supplementary Information). H₂ activation is proposed to initially produce the diferrous ammonium hydride complex [1HH]²⁺. Subsequent deprotonation and rearrangement of the incipient terminal hydride complex leads to the final product that contains a bridging hydride, [1H]⁺ (Scheme 1).

Scheme 1.

Activation of H₂ by [1]⁺ and Cp*₂Fe⁺ to form [1H]⁺.

To investigate the role of the oxidant, we carried out reactions with various ferrocinium-derivatives,⁷ [Me_nFc]BAr^F₄ ([Me_nFc]⁺ = [Fe(C₅Me₅)₂]⁺, [Fe(C₅Me₄H)₂]⁺, [Fe(C₅Me₅)(C₅H₅)]⁺, Table 1. Monitoring product formation by ¹H NMR spectroscopy, we found that the rate of reaction was independent of oxidant strength (for the BAr^F_4⁻ salts: −593 to −313 mV vs Fc/FcBAr^F₄)⁸. Furthermore the rate is unaffected by the concentration of the oxidant. These findings imply that electron transfer does not occur in or before the rate-determining step.

Table 1.

Observed Pseudo First-Order Rate Constants for the Conversion of 1 to [1]⁺ with Various Ferrocenium Oxidants. Conditions: 2 atm H₂, 0 °C, CD₂Cl₂ solution, [1]_o = [P(o-tol)₃] = 4.67 mM.

Oxidant	equiv MenFc+	E(MenFc+/0) vs E(Fc+/0)	kobs (s−1)× 105	Conversion
[Me10Fc]+	2	−593 mV	2.2(3)	100%
[Me10Fc]+	2*	−593 mV	2.2(4)	100%
[Me10Fc]+	4	−593 mV	2.7(3)	100%
[Me8Fc]+	2	−512 mV	4.2(6)	>75%
[Me8Fc]+	4	−512 mV	4.2(5)	>75%
[Me5Fc]+	2	−313 mV	3.3(8)	>50%

^*P(o-tol)₃ omitted from reaction.

Having observed that oxidant did not affect the rate of H₂ oxidation, we probed the effect of hydrogen pressure. When one equiv of oxidant was used, the appearance of product [1H]⁺ was strictly first order. Under these conditions, H₂ dissolves quickly and was present in large excess. As a result, [H₂] remains constant over the course of each experiment. A plot of k_obs vs [H₂] was linear, verifying a first order dependence on [H₂]. These observations imply a rate law that only includes terms in [H₂] and [Fe₂].

$$\mathrm{d}[\mathbf{1}{\mathrm{H}}^{+}]/\mathrm{d}t=k_{\text{obs}}[{\mathrm{H}}_2][{\mathbf{1}}^{+}]$$ (1)

The experimental rate law in eq 1 is consistent with two kinetic situations (Scheme 2). The first involves rate-determining binding of H₂ followed by rapid oxidation and/or heterolysis. In the second mechanism, fast H₂ binding is followed by rate-determining heterolysis to form the mixed-valence hydride, which is rapidly oxidized in a subsequent step. The acidity of transition metal dihydrogen complexes is well established.⁹ The favorability of H₂ cleavage is expected to depend on the hydride-acceptor ability of the Fe₂⁺ fragment. Also, numerous precedents show that the electrophilicity of metal centers correlates with their affinity for H₂.^9d Thus, more electrophilic diiron models should result in a faster reaction. An obvious choice of an electrophilic diiron center would be [Fe₂[(SCH₂)₂NBn](CO)₄(dppv)]⁺, a tetracarbonyl relative of [1]⁺.¹⁰ This mixed-valence compound was found to be unstable, probably owing to disproportionation caused by amine binding.⁵ We discovered however that a more rigid bulky diphosphine allowed isolation of the sought-for electrophilic, amine-containing H_ox model. Specifically [Fe₂[(SCH₂)₂NBn](CO)₄(dppn)]⁺ ([2]⁺) (dppn = 1,8-bis(diphenylphosphino)naphthalene) was accessed in two steps from Fe₂[(SCH₂)₂NBn](CO)₆. According to cyclic voltammetric measurements on CH₂Cl₂ solutions (Figure S4), the [2]^+/0 couple occurs at −254 mV, 390 mV more positive than the [1]^+/0 couple.

Scheme 2.

The rate law for oxidation of H₂ by [1]⁺ and ferrocenium is consistent with the rate determining step being H₂ addition (step 1) or redox/heterolysis of the H₂ complex (step 2).

Treatment of a CH₂Cl₂ solution of 2 with 1 equiv of [Fc]BAr^F₄ gave the mixed-valence salt [2]BAr^F₄, solutions of which remained unchanged for up to 24 h at room temperature. The stability of this mixed-valence amine allowed us to obtain single crystals. Crystallographic analysis shows that the amine (proton acceptor) is poised over the electrophilic Fe center (hydride acceptor) only 3.2 Å away. The high stability of this organometallic frustrated Lewis pair¹¹ is attributed to the steric shielding provided by a pair of phenyl rings that project axially from the dppn ligand (Figure 2). IR spectra in the υ_CO region show that [2]⁺ is far more electrophilic than is [1]⁺:υ_CO = 1896, 2022, 2078 vs 1870, 1965, 2017 cm⁻¹ respectively (Figure 3).

Figure 2.

Structure of the cation in [Fe₂[(SCH₂)₂NBn](CO)₄(dppn)]BAr^F₄. Selected bond lengths: (Fe-N) = 3.234(3), (Fe-Fe) = 2.568(1), (Fe-P) = 2.231(1) Å.

Figure 3.

IR spectra of a CH₂Cl₂ solution of [2]BAr^F₄ and [Fc]BAr^F₄ before (black) and 5 (red) and 14 min. (blue) after introducing H (1atm).

The EPR spectrum of [2]BAr^F₄ features an axial pattern and exhibits triplets indicative of large hyperfine coupling to ³¹P. This suggests the assignment of [2]⁺ as formally [(dppn)(CO)Fe^I(μ-SR)₂Fe^II(CO)₃]⁺. The spectrum is similar to that reported for related cations such as [Fe₂(S₂C₃H₆)(CO)₄(dppv)]⁺. It is likely that, as previously reported for [1]⁺, the electrophilic site is the iron center formally assigned as Fe(I).¹²

Gratifyingly, in the presence of one equiv of [Fc]BAr^F₄, [2]⁺ was found to rapidly react with 1 atm H₂ (t_1/2 < 13 min at 20 °C). Monitoring the reaction by IR and ¹H NMR spectroscopies confirmed the formation of the same ammonium hydride produced by treatment of 2 with 2 equiv of [H(OEt₂)₂]BAr^F₄. Rate measurements showed H₂ activation by [2]⁺/[Fc]⁺ to be 10× faster than by [1]⁺/[Fc]⁺ and 10⁴× faster than [1]⁺ in the absence of a supplemental oxidant.

We measured the rates of reaction of a 1:1 mixture of [2]BAr^F₄ and [Fc]BAr^F₄ with H₂ and D₂ using UV-vis spectroscopy at 20 °C. The kinetic isotope effect was found to be k_H/k_D = 0.71(5). Reactions in which H₂ cleavage is known to be rate determining typically exhibit normal kinetic isotope effect (eg k_H/k_D ~ 2.0).^9d Although few reports describe kinetic isotope effects for H₂ binding, ¹³ an inverse kinetic isotope effect has been observed for H₂ binding to Ir(H)₂Cl(PBu^t₂Me)₂.¹⁴ The inverse isotope effect measured for our reaction is inconsistent with rate-determining heterolytic cleavage of H₂.

An enigmatic aspect of the present results is the observation that [1]⁺ and [2]⁺ do not serve as oxidants for the oxidation of H₂ by a second equiv of the same cations. This finding may be explicable if the activation of H₂ occurs via a concerted proton-coupled electron transfer (PCET), whereby intramolecular heterolysis of the H₂ ligand depends on the rate of the electron-transfer. ¹⁵ Proton-transfer associated with the intramolecular heterolysis of the H₂ adducts is expected to be extremely rapid,¹⁶ requiring a rapid oxidant. The rate of self-exchange for Fc^+/0 is indeed fast (5 × 10⁶ M⁻¹s⁻¹),¹⁷ but we propose that self-exchange for [1]^+/0 and [2]^+/0 are probably far slower owing to the substantial structural changes that accompany this redox process.¹⁸ Further work is required on these self-exchange rates.

We have shown H₂ activation by the organometallic radical [Fe₂[(SCH₂)₂NR](CO)_xL_6-x]⁺ requires the addition of an oxidant. It is intriguing that for heterolysis the oxidant must be both mild and fast. Kinetic measurements show that H₂ binding is rate-determining. The present results point to the important role of PCET for the heterolytic activation of dihydrogen in this class of enzyme mimics.

Figure 4.

X-Band EPR spectrum of [2]BAr^F₄ (110 K, 1:1 CH₂Cl₂:toluene frozen solution). Parameters: g_z = 2.1260 with A_z(³¹P) = 78 MHz, g_x ≈ g_y = 2.0165 with A_x(³¹P) ≈ A_v(³¹P) = 79 MHz.