Our long-held interest in the resiliency of electrochemical functionalities upon surface immobilization has herded us from directly chemisorbed electroactive moieties [1, 2], to anchor group-leashed redox-active couples [3] and to surface-tethered enzyme-inspired molecular catalysts [4–6]. The latter represent the most intricate because the electrocatalytic activities involve mixed-valence states and may require certain entatic (fractionally rotated) configurations [6, 7]. In this regard, we recently investigated the proton-reduction electrocatalysis by hydrogenase-inspired di-iron complexes at polycrystalline and (111)-faceted Au electrodes [4–6].Footnote 1 One complex, (μ-S2C3H6)[Fe(CO)3][Fe(CO)2PPh3], was devoid of a surface anchor group and, hence, was present only in solution; the other complex, (μ-S2C3H6)[Fe(CO)3][Fe(CO)2(PPh2(CH2)2SH)] (I), was bound to the surface through its mercapto group that allowed the catalytic di-iron moiety to remain to be pendant. The electrochemistry experiments had to be performed in acetonitrile, with n-Bu4NBF4 as a supporting electrolyte, because the complexes are insoluble in aqueous solutions. No activity was displayed by the homogeneous compound; dramatic catalysis was shown by the heterogenized complex. Earlier computational studies had indicated that the iron hydride intermediate is formed at a terminal site on a partially rotated iron [7, 8]; evidently, such entatic process is not hindered even at a densely packed surface-sequestered catalyst. We have now expanded the work to explore the extensibility of the heterogenization protocol to highly acidic aqueous solutions (0.5 M H2SO4). The results are described in this letter.

figure a

The di-iron complex (I) was synthesized and purified according to literature procedures [9, 10]. All experiments were carried out with a flame-annealed 0.2-mm-thick Au foil of a geometric area of 1.0 cm2. Prior to each experiment, the Au electrode was cleaned in 1.0 M H2SO4 by sequential anodic oxidation [1.2 V versus Ag/AgCl (1.0 M NaCl) reference] and cathodic reduction (−0.25 V). The surface immobilization procedure was undertaken in acetonitrile due to the insolubility of the subject compound in water. It consisted simply of a 3-min immersion, at open circuit, of the Au electrode in a deoxygenated acetonitrile solution that contained 0.1 mM of the complex; no difference in coverage or catalytic activity was observed when the self-assembly was allowed to take place overnight. To exclude unadsorbed material, the electrode was rinsed sequentially with pure acetonitrile and acetone; it was then dried by a stream of Ar gas.

For the non-aqueous electrochemistry work, acetonitrile was the solvent employed with n-Bu4NBF4 (0.1 M) as the supporting electrolyte. The reference was a Ag/Ag+ electrode encased in a Vycor-fritted tube and prepared by anodization of a silver wire in an acetonitrile solution that was made up of 0.01 M AgNO3 and 0.1 M n-Bu4NBF4. A gold wire was used as the counter electrode. Changes in pH were accomplished by incremental addition, via a microsyringe, of glacial acetic acid. For electrocatalysis experiments in sulfuric acid, the Au electrode pre-coated with the complex in acetonitrile was rinsed with pure solvent and dried under Ar prior to transfer to an aqueous solution that contained 0.5 M H2SO4. The reference electrode was Ag/AgCl (1.0 M NaCl). Cyclic voltammograms were obtained with an SP-200 Bio-Logic potentiostat in a three-electrode configuration. The cell was a single-compartment glass vessel blanketed with Ar gas throughout the electrochemical experiments.

Figure 1 shows cyclic current-potential curves in the proton-reduction region in acetonitrile for the surface-immobilized complex in the absence and presence (ten equivalents) of glacial acetic acid; these results are identical to those reported previously that complex I, when (and only when) in the chemisorbed state, is able to catalyze the hydrogen evolution reaction [4, 6]. Information that bears more significance in the present context can be gleaned from the inset in Fig. 1 where the current density scale is expanded in the HER region. Such rendition helps clarify the relationship between the catalyzed proton-reduction reaction (solid curve) and the active site redox reactions (dashed curve) that enable it. In addition, the Faradaic charge associated with the electron transfer reactions of the di-iron center is a direct measure of the surface coverage (Γ) of the complex. The coverage, in turn, makes possible the determination of the catalyst turnover frequency (TOF).

Fig. 1
figure 1

Cyclic voltammogram of Au-attached (μ-S2C3H6)[Fe(CO)3][Fe(CO)2(PPh2(CH2)2SH)] in CH3CN–0.1 M n-Bu4NBF4 in the presence (solid curve) and absence (dashed curve) of acetic acid. Potential sweep rate is 200 mV s−1. Inset indicates expanded current density axis in the proton reduction region. The scan rate for the dashed curve was 800 mV s−1. Two equivalents of HOAc = 1.75 × 10−4 mol of HOAc. V total = 10.0 ml

Full size image

It is known [7, 8] that the two symmetric cathodic peaks in the dashed curve are attributable to the two one-electron reductions:

$$ \left[{\mathrm{Fe}}^{\mathrm{I}}-{\mathrm{Fe}}^{\mathrm{I}}\right]+{\mathrm{e}}^{-}=\left[{\mathrm{Fe}}^0-{\mathrm{Fe}}^{\mathrm{I}}\right] $$
(1)
$$ \left[{\mathrm{Fe}}^0-{\mathrm{Fe}}^{\mathrm{I}}\right]+{\mathrm{e}}^{-}=\left[{\mathrm{Fe}}^0-{\mathrm{Fe}}^0\right] $$
(2)

Under the conditions of the experiment, the first reduction occurs at −1.58 V and the second at −1.92 V. The fact that the proton-reduction reaction occurs immediately after the onset of reaction 1 provides evidence that the catalysis follows not the electrochemical-electrochemical-chemical-chemical (EECC) pathway [4, 6] abided by a non-phosphine-substituted di-iron hexacarbonyl analogue [7, 8] but an ECCE or ECEC mechanism [7, 8]. An EECC process would have located the onset of the hydrogen evolution reaction [Eq. (2)] coincidentally with that of the second reduction peak. The ECCE route is given by Eqs. (3) to (6):

$$ \left[{\mathrm{Fe}}^{\mathrm{I}}-{\mathrm{Fe}}^{\mathrm{I}}\right]+{\mathrm{e}}^{-}=\left[{\mathrm{Fe}}^0-{\mathrm{Fe}}^{\mathrm{I}}\right] $$
(3)
$$ \left[{\mathrm{Fe}}^0-{\mathrm{Fe}}^{\mathrm{I}}\right]+{\mathrm{H}}^{+}={\left[\mathrm{H}-{\mathrm{Fe}}^{\mathrm{I}\mathrm{I}}---{\mathrm{Fe}}^{\mathrm{I}}\right]}^{-} $$
(4)
$$ {\left[\mathrm{H}-{\mathrm{Fe}}^{\mathrm{I}\mathrm{I}}---{\mathrm{Fe}}^{\mathrm{I}}\right]}^{-}+{\mathrm{H}}^{+}=\left[{\mathrm{H}}_2-{\mathrm{Fe}}^{\mathrm{I}}---{\mathrm{Fe}}^{\mathrm{I}}\right] $$
(5)
$$ \left[{\mathrm{H}}_2-{\mathrm{Fe}}^{\mathrm{I}\mathrm{I}}---{\mathrm{Fe}}^{\mathrm{I}}\right]+{\mathrm{e}}^{-}=\left[{\mathrm{Fe}}^{\mathrm{I}}-{\mathrm{Fe}}^{\mathrm{I}}\right]+{\mathrm{H}}_2 $$
(6)

The ECEC mechanism would reduce the intermediate [H–FeII---FeI]− in Eq. (4) to [H–FeI---FeI]− prior to the second protonation. The combined Faradaic charge of the two cathodic peaks in the inset of Fig. 1 was measured to be 15.4 μC, a value that corresponds to Γ of 0.080 nmol cm−2.Footnote 2

Figure 2 shows cyclic voltammograms of the di-Fe hydrogenase-based complex surface attached to Au in the hydrogen evolution region in 0.5 M H2SO4 for the 1st cycle and for the 120th cycle of uninterrupted operation; each cycle took 1 min to complete. The voltammetric curve for clean (complex-free) Au is also shown for reference. Two features in Fig. 2 are noteworthy: (i) the overvoltage at a current density (J) of 10 mA cm−2 is 300 mV (versus SHE); this is not an inauspicious value for a synthetic enzyme-based molecular complex [7, 8]. (ii) There is essentially no change in the overvoltage after 2 h of continuous proton-reduction catalysis; the robustness of the heterogenized molecular catalyst in highly acidic aqueous solutions is thus established [11].

Fig. 2
figure 2

Cyclic voltammogram of (μ-S2C3H6)[Fe(CO)3][Fe(CO)2(PPh2(CH2)2SH)], immobilized on smooth polycrystalline Au via the –SH group, in aqueous 0.5 M H2SO4. The dotted curve is for a clean Au electrode, the solid line is for the first cycle of the heterogenized complex, and the dashed plot is for the 120th cycle. Potential sweep rate is 10 mV s−1

Full size image

It will be mentioned that, while the voltammetry in n-Bu4NBF4-CH3CN (Fig. 1) was acquired prior to that in aqueous H2SO4 (Fig. 2), the results were essentially the same when the electrode was returned to the acetonitrile solution, and then back again in sulfuric acid, for additional voltammetric cycles. Likewise, no changes were noted when the (non-aqueous to aqueous) electrochemistry sequence was altered.

It is difficult to ignore the difference in the morphologies of the HER current-potential curves in acetonitrile and in sulfuric acid. The shape of voltammogram in the non-aqueous solvent is typical of molecular catalysts, whereas the form in aqueous acid is a reminiscent of metals. Investigations into the molecular integrity of the heterogenized complex, however, are well beyond the intent and scope of the present manuscript.

The turnover frequency, for a given current density, of the catalyst in molar acid can be obtained from Eq. (7):

$$ \mathrm{TOF}=\mathrm{J}/\left( nF\varGamma \right) $$
(7)

where n = 2 for the 2H+-to-H2 reaction, and F is the Faraday constant, 96,485 C equivalent−1. For J = 10 mA cm−2, the TOF was calculated to be 640 mol of H2 per mole of heterogenized catalyst per second.

In summary, (μ-S2C3H6)[Fe(CO)3][Fe(CO)2(PPh2(CH2)2SH)], a hydrogenase-based molecular complex that is insoluble in water, can be chemisorbed on Au, via its mercapto anchor group, in acetonitrile, and, after removal of non-aqueous solvent and unadsorbed complex, transferred to an aqueous solution of sulfuric acid for catalysis of the hydrogen evolution reaction. The heterogenized complex showed no change in activity when reran in the non-aqueous solution and then again in the aqueous acid. The packing density of the heterogenized complex was found, from its redox reaction in proton-free acetonitrile, to be 0.080 nmol cm−2. The overvoltage, at a current density of 10 mA cm−2 in 0.5 M H2SO4, was 300 mV. The turnover frequency, at 10 mA cm−2, was 640 mol of H2 per mole of immobilized complex per second. Remarkable operando robustness was evidenced by the absence of an increase in the overvoltage after 2 h of uninterrupted operation.