Structural and Kinetic Studies of Intermediates of a ... - ACS Publications

Jan 3, 2018 - Shihuai Wang, Alexander Aster, Mohammad Mirmohades, Reiner Lomoth, and Leif Hammarström*. Department of Chemistry-Ångström ...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/IC

Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX

Structural and Kinetic Studies of Intermediates of a Biomimetic Diiron Proton-Reduction Catalyst Shihuai Wang, Alexander Aster, Mohammad Mirmohades, Reiner Lomoth, and Leif Hammarström* Department of Chemistry-Ångström Laboratory, Uppsala University, Box 523, SE-751 20 Uppsala, Sweden S Supporting Information *

ABSTRACT: One-electron reduction and subsequent protonation of a biomimetic proton-reduction catalyst [FeFe(μ-pdt)(CO)6] (pdt = propanedithiolate), 1, were investigated by UV−vis and IR spectroscopy on a nano- to microsecond time scale. The study aimed to provide further insight into the proton-reduction cycle of this [FeFe]-hydrogenase model complex, which with its prototypical alkyldithiolatebridged diiron core is widely employed as a molecular, precious metal-free catalyst for sustainable H2 generation. The one-electron-reduced catalyst was obtained transiently by electron transfer from photogenerated [Ru(dmb)3]+ in the absence of proton sources or in the presence of acids (dichloro- or trichloroacetic acid or tosylic acid). The reduced catalyst and its protonation product were observed in real time by UV− vis and IR spectroscopy, leading to their structural characterization and providing kinetic data on the electron and proton transfer reactions. 1 features an intact (μ2,κ2pdt)(μ-H)Fe2 core in the reduced, 1−, and reduced-protonated states, 1H, in contrast to the Fe−S bond cleavage upon the reduction of [FeFe(bdt)(CO)6], 2, with a benzenedithiolate bridge. The driving-force dependence of the rate constants for the protonation of 1− (kpt = 7.0 × 105, 1.3 × 107, and 7.0 × 107 M−1 s−1 for the three acids used in this study) suggests a reorganization energy >1 eV and indicates that hydride complex 1H is formed by direct protonation of the Fe−Fe bond. The protonation of 1− is sufficiently fast even with the weaker acids, which excludes a rate-limiting role in light-driven H2 formation under typical conditions.



INTRODUCTION Hydrogen as a clean alternative to fossil fuels has attracted great attention over the past few decades. For the sustainable production of H2, biomimetic molecular catalysts related to the active site of the [FeFe]-hydrogenase enzyme have been widely considered as potential alternatives to catalysts based on precious metals. Although the catalytic activities of the model complexes in experiments of electrochemical and visible-lightdriven reduction of protons were readily demonstrated,1−7 they were generally less efficient than the enzyme, which catalyzes proton reduction at rates up to 9000 turnovers per second at a very modest overpotential.8 In order to improve the performance of the [FeFe]-hydrogenase mimics, better mechanistic understanding will be needed to identify bottlenecks in the catalytic cycle. Unfortunately, the characterization of catalytic intermediates9,10 with regard to their structures and reactivity has been largely limited because in the presence of proton sources, electrochemical or photochemical reduction results in rapid catalytic turnover. In particular, examples of direct spectroscopic observation of intermediates are very scarce. Only a few examples of mechanistic investigations were reported by Pickett et al. and other groups, which aimed to study the spectral characteristics of the electrocatalytic intermediates by stopped-flow spectroscopic and spectroelectrochemical techniques.11−13 The limited time resolution intrinsic to these techniques would, however, be insufficient to capture the spectroscopic signatures and kinetics of the © XXXX American Chemical Society

more reactive transients that can be anticipated to occur in the protonation of electron-rich species.12,14 In contrast, photochemically induced reduction in flash photolysis experiments allows one to follow even diffusion-controlled reactions of the flash-generated transients on nano- to millisecond time scales. We have recently reported on the spectroscopic and kinetic characterization of intermediates in the photocatalytic proton reduction of [FeFe(bdt)(CO)6] (bdt = benzenedithiolate) using a combination of flash-quench methods with timeresolved UV−vis and IR spectroscopies.15 The real-time observations of vibrational spectra with unprecedented time resolution facilitated the structural characterizations of key intermediates. In this study we employed time-resolved infrared and UV− vis spectroscopy to investigate the electron- and protontransfer reactions of the proton-reduction catalyst [FeFe(μpdt)(CO)6] (pdt = propanedithiolate), 1, on hitherto unexplored time scales. Our results indicate significant structural differences between the one-electron-reduced states of 1, with its prototypical alkyldithiolate bridging ligand, and those of its widely studied analogue [FeFe(bdt)(CO)6], 2, which has the aromatic bdt ligand. From direct comparison of our timeresolved UV−vis and IR spectral data of both complexes, Received: October 18, 2017

A

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry spectroscopic markers for the [(μ2,κ2-pdt)Fe2(CO)6]− and [(μ,κ2-bdt)Fe2(CO)6]− motifs could be identified. The reduced-protonated complex, 1H, features the same (μ2,κ2pdt)(μ-H)Fe2 core as previously found for 2H. Analysis of proton transfer kinetics yields an estimate of the reorganization energy and suggests that protonation of reduced metal centers follows a direct mechanism different from the previously reported protonation of neutral complexes with electron rich phosphine ligands.16



Scheme 1. (a) Reaction Scheme for the Photo-Induced Reduction of 1a and (b) Structural Indication of All Three States of [Fe2(pdt)(CO)6] (1) and [Fe2(bdt)(CO)6] (2)

EXPERIMENTAL SECTION

General Procedures. All chemicals, if not stated otherwise, were purchased from Sigma-Aldrich and used without further purification. FeFe(μ-pdt)(CO)6, 1, was synthesized according to published methods17 and confirmed to be pure by NMR and IR spectroscopy. Ru(dmb)3(PF6)2 was synthesized and purified in terms of the published procedures18 for Ru(dmb)3Cl2 and NH4PF6. All acids were dried under vacuum prior to use. All samples were prepared in the dark in a glovebox under an argon (Ar) atmosphere. Distilled and oxygen-removed acetonitrile (CH3CN) was used as the solvent in all the experiments. All measurements were performed at 22.5 ± 1 °C. Steady-State-Absorption Measurements. UV−vis absorption spectra were recorded on a Cary 50 UV−visible spectrophotometer. FT-IR spectra were collected with a Bruker IFS 66v/S FT-IR spectrophotometer. Transient-Absorption Measurements. A frequency-doubled Qswitched Nd:YAG laser (Quanta-Ray ProSeries, Spectra-Physics) was employed to obtain the pulsed 532 nm pump light with 10 mJ/pulse or 50 mJ/pulse in some cases and a fwhm of 10 ns. For the UV−vis measurements, the pump laser was coupled to a LP920 detection system (Edinburgh Instruments) equipped with a pulsed XBO 450 W xenon Arc Lamp (Osram) to provide the white light for probing. An iStar CCD camera (Andor Technology) and a LP920-K photomultiplier tube (PMT) detector connected to a Tektronix TDS 3052 500 MHz 5 GS/s oscilloscope were used for transient signal detection. Transient absorption data was acquired using L900 software (Edinburgh Instruments) and processed using Origin 2015 software. A fluorescence quartz cell cuvette (Starna) with a 1.0 cm path length was used for the measurements. For the IR measurements, two continuous-wave quantum-cascade (QC) IR lasers, tunable from 1300 to 1965 cm−1 for laser 1 and 1960−2150 cm−1 for laser 2 (Daylight solutions), were used to provide the infrared light for the probing. The IR probe light was overlapped with the laser beam in a quasi-colinear arrangement at 25° angle. A liquid-nitrogen-cooled mercury−cadmium−telluride (MCT) detector connected to an oscilloscope (Tektronix TDS 3052, 500 MHz, 5 GS/s) was employed for the transient signal detection. Transient absorption traces were acquired with the oscilloscope (Tektronix TDS 3052, 500 MHz, 5 GS/s) in connection with the L900 software (Edinburgh Instruments) and processed using Origin 2015 software. The samples were kept in a modified Omni cell (Specac) with O-ring-sealed CaF2 windows and a path length of 1 mm.

a

Ru = Ru(dmb)3 (dmb = 4,4′-dimethyl-2,2′-bipyridine), TTF = tetrathiofulvalene. (1) Photoexcitation, (2) excited state decay, (3) reductive quenching, (4) recombination of the photoreductant with the oxidized electron donor, (5) electron transfer to the catalyst, (6) recombination of the reduced catalyst with the oxidized electron donor, and (7) protonation of the reduced catalyst competing with process 6.

features of [Ru(dmb)3]+ (510 nm absorption and bleaching at 450 nm) and TTF+ (435 nm and 580 nm absorption, see Figure S1). The corresponding decay traces at 435 and 510 nm (Figure S2) show the recombination reaction between TTF+ and [Ru(dmb)3]+ (reaction 4), with a second-order rate constant of 6.6 × 109 M−1 s−1. This gives a recombination halflife of several tens of microseconds under the experimental conditions, during which time the reduced [Ru(dmb)3]+ may react with 1. Previous electrochemical investigations of 1 have showed that E0 is −1.58 V versus Fc+/Fc in acetonitrile.19 At this potential, [Ru(dmb)3]+ (E0 = −1.71 V vs Fc+/Fc in acetonitrile) is a good candidate for reducing 1, whereas the oxidation of 1 (Ep,a ≈ 1.00 V vs Fc+/Fc) by the weak oxidant TTF+ (E0 = −0.04 V) can be safely excluded. Because an excess of 1 is used, the reduction of 1 by the photogenerated [Ru(dmb)3]+ (Scheme 1, reaction 5) follows pseudo-firstorder kinetics with a bimolecular rate constant of 6.7 × 109 M−1 s−1 and hence competes efficiently with the [Ru(dmb)3]+/ TTF+ recombination. The reduction of 1 leads to the formation of a new absorption band peaking at 700 nm,



RESULTS One-Electron-Reduced Intermediate of 1. Timeresolved spectroscopy with the flash-quench-generated [Ru(dmb)3]+ reductant was used to reduce complex 1. As illustrated in Scheme 1 [Ru(dmb)3]+ was generated in the diffusion-controlled reductive quenching of the excited sensitizer *[Ru(dmb)3]2+ by tetrathiofulvalene (TTF). With an excess (1 mM) of TTF, the pseudo-first-order reaction is essentially complete after 100 ns and generates typical transient concentrations of [Ru(dmb)3]+ and TTF+ on the order of 10−5 and 10−6 M in IR (1 mm optical path length) and UV−vis (1 cm) light, respectively. The primary charge-separation reaction can be readily followed by the well-known transient absorption B

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

In order to separate the spectrum of 1− from the superimposed TTF+ spectrum, analogous flash-photolysis experiments were performed in which TTF was replaced with the sacrificial electron donor triethylamine (TEA), which lacks any visible transient absorption. The transient spectrum (Figure S3) showed that besides the 700 nm band, two other bands peaking at 570 and 390 nm were also observed. A complete spectrum of 1− was thereby obtained by subtracting a properly scaled TTF+ spectrum from the transient spectra obtained with the reversible TTF donor. The result is in excellent agreement with the spectrum that was observed at an early time after the electrochemical reduction of 1 under a CO atmosphere by Pickett and co-workers and assigned to a oneelectron-reduced, structurally intact complex.11 Further reduction of 1− was not observed under our experimental conditions, in which reduced [Ru(dmb)3]+ reacted with a large excess of 1, which minimized the possibility of a double reduction caused by two subsequent encounters with [Ru(dmb)3]+. Moreover, disproportionation is disfavored by an excess of 1. Even in experiments with the sacrificial donor TEA, which prevents charge recombination, 1− persists for at least 20 ms with no formation of 12− by disproportionation of 1−. Also, no loss of CO was observed during that time even in the CO-free solutions. To further extract information on the structure of the reduced-state intermediate, analogous experiments were performed with transient mid-IR absorption. The carbonyl ligands are excellent IR reporter groups that are sensitive to the electron densities of metal centers as well as to geometry changes. The kinetic traces (Figure S4) were collected per 5 cm−1 interval by tuning the IR probe laser. The spectrum was generated point by point from these recorded single-wavenumber kinetic traces. The resulting IR spectrum (Figure 2c) reveals that the reaction of 1 with the reductant, [Ru(dmb)3]+, generated by a laser flash results in bleaching of the three absorption bands (2073, 2034, and 1995 cm−1) of 1 and the concomitant rise of product bands shifted by ca. 80 cm−1 toward lower wavenumbers (2006, 1942, and 1910 cm−1). The direction and overall magnitude of the shift are in agreement with the expected effect of a one-electron reduction delocalized over two iron centers. The IR spectrum of 1− is in good agreement with the reported spectrum of the electrochemically generated reduction product obtained under the CO atmosphere.11 This suggests that the 1− species generated on the 1 μs time scale in our study is also stable on the time scale of 1 s under electrochemical-reduction conditions and that the assignments of refs 11 and 20 are correct. The similar spectral envelopes for 1 and 1− also suggest that the geometry of the complex does not change substantially upon reduction (see Discussion). Reactivity of Reduced Catalyst with Protons. With the parent complex, no protonation was observed even with addition of 25 mM p-toluenesulfonic acid (TsOH) (pKa ∼8.5 in acetonitrile),21 as confirmed by the steady-state UV−vis and IR-absorption spectra. The reduced complex is a much stronger base, however, and the reaction of the flash-quench generated 1− with varying concentrations of dichloroacetic acid (DCA), trichloroacetic acid (TCA), and TsOH was followed by UV−vis spectroscopy in the flash-photolysis experiments. Figure 3 shows the transient absorption (TA) spectra for the reaction with 13 mM TCA (see Figure S5 for the spectra with DCA and TsOH).

which does not overlap with any of the transient absorption features of [Ru(dmb)3]+ and TTF+, as well as additional absorption below 600 nm, as shown in Figure 1a. The

Figure 1. (a) UV−vis transient absorption spectra obtained at microsecond time scales and (b) transient absorption (TA) kinetic traces collected at 700 nm after the nanosecond laser excitation. The 700 nm traces showing the recombination of 1− with TTF+ are fitted to second-order decay, and the inset indicates the formation of 1−, which fits to a single exponential (pseudo-first-order). Conditions: 36 μM [Ru(dmb)3]2+, 1 mM TTF, and 207 μM [FeFe(pdt)(CO)6] in ACN; excited at 532 nm; laser pulse of 10 ns.

corresponding rise and decay traces at 700 nm shown in Figure 1b provide kinetic insight into the 1− formation and subsequent recombination of 1− with TTF+. The traces can be fitted well to a (pseudo)first-order rise followed by secondorder decay kinetics, respectively, and both rate constants correspond to diffusion-controlled reactions. The fitting residuals are displayed in the Supporting Information (ESI). From our results, all the transient absorption features generated by the reaction between 1 and [Ru(dmb)3]+ rise simultaneously on the time scale of 1 μs and decay with the same second order kinetics as the TTF+ features, without leaving any residual transient absorption. These absorption features can therefore be assigned to the one-electron-reduced complex 1−, which does not undergo any transformations prior to reoxidation by TTF+. C

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 2. (a,b) UV−vis spectra of (a) [FeFe(pdt)(CO)6], 1, (black line) and its one-electron-reduced product, 1−, (red line) and (b) [FeFe(bdt)(CO)6], 2, (black line) and its one-electron-reduced product, 2−, (red line). (c,d) IR spectra of (c) [FeFe(pdt)(CO)6], 1, (black line) and its one-electron-reduced product, 1−, (red line) and (d) [FeFe(bdt)(CO)6], 2, (black line) and its one-electron-reduced product, 2−, (red line).

protonation product, 1H. Its formation and slow recombination is, however, further evidenced by a much slower decay of part of the TTF+ transient absorption. This is observed in the transient spectra at longer times when 1− is completely consumed by recombination and protonation. The pseudofirst-order rate constants, kobs, of the protonation reactions were determined from a biexponential fit of the rise and decay at 700 nm (see Figure 4a−c). Plots of kobs versus acid concentration, shown in Figure 4d−f, were reasonably linear, resulting in second-order rate constants, given in Table 1, for the protonation of 1−. The nonzero intercept is attributed to the (second-order) recombination of 1− and TTF+, which becomes important at low acid concentrations. The good correlation of the yields of the protonation reactions, as estimated from the fraction of the long-lived TTF+, with the pseudo-first-order rate constants confirmed that the accelerated decay of 1− is predominantly due to protonation. Thus, any effects from TTF+ potentially accumulating in the acidified samples did not significantly affect the observed rate constants (see the protonation kinetics in the ESI). Transient IR experiments in the presence of acids gave the kinetics of 1− formation and decay (Figure 5), which agreed with the results at 700 nm (Figure 1b). No unique IR transients were detected from the 1H that formed. However, that can be expected, as the oxidative addition of a proton to 1− forming 1H shifts the νC−O band to the same extent but in the reverse direction as 1e− reduction. This might lead to the

Figure 3. TA spectra in the presence of TCA after the nanosecond laser excitation. Conditions: 40 μM [Ru(dmb)3]2+, 1 mM TTF, 200 μM [FeFe(pdt)(CO)6], and 13 mM TCA in ACN; excited at 532 nm; laser pulse of 10 ns.

At early times (1 μs) the absorption band of 1− can be observed at 700 nm together with the stronger bands of TTF+ at 585 and 435 nm and some residual absorption of [Ru(dmb)3]+ at 510 nm. Spectra at later times monitored the accelerated decay of 1− due to the protonation reaction, which did not lead to any distinct transient absorption of the D

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 4. (a−c) Normalized kinetic traces at 700 nm (original ΔOD ≈ 0.015) with selected concentrations of DCA (a), TCA (b), and TsOH (c) together with the biexponential (rise-and-decay) fits. The decay component gave the pseudo-first-order rate constant (kobs) for the protonation of 1−. (d−f) Plots of the pseudo-first-order rate constants, kobs, determined from the 700 nm traces vs the concentrations of DCA (d), TCA (e), and TsOH (f). Conditions: 40 μM [Ru(dmb)3]2+, 1 mM TTF, and 200 μM [FeFe(pdt)(CO)6] in ACN; excited at 532 nm; laser pulse of 10 ns.

workers23 suggest that the coordination geometry of the Fe ions in the DFT optimized species, 1−, is square pyramidal, similar to that in the parent complex, 1, and that the main structural change in the 1 + e− → 1− reduction reaction is a lengthening of the Fe−Fe distance by 0.28 Å. This is supported by our experimental results that the ν(CO)-band envelope of 1− has a similar profile to that of 1, only shifted by ca. 80 cm−1 toward lower wavenumbers. The pronounced absorption features of 1− in the UV−visible region suggests that oneelectron reduction did not give rise to the cleavage of the Fe− Fe bond. Upon protonation, the resulting 1H species shows IR spectra very similar to those of 1, and we were not able to clearly distinguish these transient IR spectra from each other. This suggests that protonation of the Fe−Fe bond leads to an

Table 1. Second-Order Rate Constants for the Protonation of 1− acid

pKa

kprot (M−1 s−1)

toluenesulfonic acid trichloroacetic acid dichloroacetic acid

8.5 10.7 13.3

7.0 × 107 1.3 × 107 7.0 × 105

carbonyl vibrational bands of 1 and 1H being virtually superimposable.22 We found a very similar result for 2 and 2H, whose IR-band maxima differed by only ∼2 cm−1.15



DISCUSSION Structures of the Reduced Intermediates. With regard to the structure of 1−, earlier calculations by De Gioia and coE

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

The protonation of 2− to form 2H leads to an IR spectrum that is very similar to that of 2, just like we observe here for 1 and 1H. This is consistent with the similar geometries of 2 and 2H, each of which has both S ligands in bridging modes and a bridging proton that essentially compensates for the increase in electron density upon reduction ([(μ 2 ,κ 2 -bdt)(μ-H)Fe2(CO)6]). We note that a recent study suggested that 2H has a twisted dithiolate ligand with only one of the sulphurs bridging.22 However, that experimental IR spectrum very closely matches both our experimental and calculated spectra for [(μ2,κ2-bdt)(μ-H)Fe2(CO)6], whereas the calculated spectrum for a ruptured Fe−S bond ([(μ,κ2-bdt)(μ-H)Fe2(CO)6]) shows worse agreement.15 Also, the calculated optical spectrum of (μ,κ2-bdt)(μ-H)Fe2(CO)6] shows a rather strong band around 660 nm,22 which was not observed experimentally for 2H in our previous study. Thus, we maintain our assignment of 2H to [(μ2,κ2-bdt)(μ-H)Fe2(CO)6] with the same coordination geometry as that of the parent complex. The changes in the Fe−Fe distance also illustrate the different behaviors of the pdt- and bdt-complexes. With pdt, the calculated Fe−Fe distances are 2.543 Å (1), 2.826 Å (1−), and 2.763 Å (1H),23 whereas with bdt, the corresponding values are 2.522 Å (2), 2.632 Å (2−), and 2.740 Å (2H).15 Thus, although the FeIFeI and FeIIFeI(μ-H) states show virtually identical distances for the two complexes, there is a substantial difference in the FeIFe0 state. The twisting of the bdt ligand and the breaking of one Fe−S bond in 2− minimizes weakening of the Fe−Fe bond upon reduction. This structural change is what causes the inversion of the 2 → 2− and 2− → 22− reduction potentials.27 In contrast, 1 maintains its coordination geometry upon reduction. Thus, the addition of an electron in an antibonding orbital weakens and elongates the Fe−Fe bond much more than it does in 2, and the subsequent protonation to form 1H decreases the Fe−Fe-bond distance again. The very different structures of 1− and 2− are important to consider in a comparison of protonation dynamics, as discussed below. Protonation Kinetics. The neutral complexes 1 and 2 are very poor bases that can only be protonated with superacids.22 For the reduced bdt complex, 2H, a pKa value can be estimated from the published data on the reduction potentials of the 2−/2− (−1.27 V; −1.33 V) and 2H0/− (−0.53 V) couples and the pKa value for 2H− (>23; 22.6);19,27 all the potentials are given versus Fc+/0. From the averages of these values and the use of a thermodynamic cycle, we can estimate a value of pKa ≈ 11 for 2H. For the reduced complex, 1H, in acetonitrile, a value of pKa = 9 has been estimated from simulations of voltammograms under electrocatalytic conditions with p-toluenesulfonic acid (TsOH, pKa = 8.5).11 This value appears incorrect, as it is lower than that for 2H, which has an electron-withdrawing bdt group. We also note that the rate constant for the protonation of 1− by TsOH from the simulations was given as 105 M−1 s−1, whereas here we measure this rate constant as 7.0 × 107 M−1 s−1. Simulations of electrocatalytic voltammograms include many reactions and parameters; many of them are often strongly interdependent, which can make the derived parameter values very uncertain. We can therefore propose that the pKa value of 1H is >11, higher than that for 2H. To put an upper limit on the pKa value, we use a thermodynamic cycle with reduction potentials for 10/− and 1H+/0 of −1.5528 and −0.69 V,22 respectively, and note that 1 is not protonated

Figure 5. Comparison of the kinetic transients without acid (black) and in the presence of 20 mM TCA (red). The decay of absorption is at 1910 cm−1, and the recovery of bleaching is at 2034 cm−1.

FeIIFeI μ-H species with a very similar geometry to that of the parent complex. The increased electron density on the metal centers due to reduction is then neutralized by the bridging proton. Both the FeIIFeI μ-H structure and the close similarity of the IR bands of 1 and 1H are consistent with suggestions from calculated structures and spectra.22,23 The frontier orbitals of 1 have a predominantly metal character.24−26 The spin density in the reduced complex, 1−, is mainly located on the metals. Thus, the visible bands of 1−, at 580 nm (ε ≈ 7000 M−1 cm−1) and 700 nm (ε ≈ 5000 M−1 cm−1) can presumably be assigned to transitions involving predominantly metal-based orbitals. When this is protonated to form 1H, the direct Fe−Fe bond is replaced by an Fe(μH)Fe bond, which completely changes the electronic structure and the active orbitals. Thus, the visible absorption can be expected to decrease substantially upon protonation of 1−, as is also observed in our experiments. This interpretation is supported by the experimental spectrum of 1H, which has much weaker bands at 457 nm (ε ≈ 2500 M−1 cm−1) and 666 nm (ε ≈ 1500 M−1 cm−1)22 than that of 1−. On the other hand, it is interesting to compare the IR spectra of 1 and 1− with those of the analogous bdt-bridged complexes 2 and 2− (Figure 2). The neutral parent complexes feature very similar IR spectra, with a minor shift of all three bands in the bdt complex by 5−10 cm−1 to higher wavenumbers, consistent with the identical first-coordination spheres and the slightly diminished electron-donating ability caused by the aromatic dithiolate ligand. The one-electronreduced complexes, however, feature pronounced spectral dissimilarities that are indicative of structural differences between 1− and 2−. Although the spectrum of 1 shows a regular shift and maintains its envelope upon reduction (Figure 2c), 2− shows a different spectral shape with four resolved peaks instead of only three for the neutral form (Figure 2d). Thus, this points to an intact coordination sphere for 1−, whereas 2− undergoes structural changes upon its reduction, which can be attributed to the deligation of a thiolate from one of the iron centers, according to our previous computational results.15 Hence, the protonation behaviors of complexes 1− and 2− may be different, as discussed below. F

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry by HBF4,11 so that for 1H+, pKa < 1. Thus, we deduce that for 1H, pKa < 15.6. To conclude, we assign a value of 11 < pKa (1H) < 15.6. As another comparison, for the related complex [Fe2(edt)(CO)6(μ-H)] (edt = 1,2-ethanedithiolate), the calculated values (DFT) have been reported as 16.7−16.9.22,29 The protonation rate constants for 1− and 2− are virtually identical. This may seem surprising, as the pKa value of 2H with a benzyl dithiolate ligand can be expected to be lower than that for 1H. Also, the reorganization upon the protonation of 2− is larger because of the recoordination of one Fe−S bond. We propose that the more open structure of the bdt complex in the 2− state makes protonation sterically more facile and compensates for the other effects. Alternatively, protonation could occur via S-protonation, which may be facilitated in the reduced bdt complex in which one S is no longer bridging. There are relatively few studies of protonation kinetics at metal centers.16,30−32 One interesting example is by Pickett and co-workers who studied the protonation kinetics for a range of [Fe2(SRS)(CO)6−n(Me3P)n] complexes (SRS denoted different dithiolate ligands) by HBF4. Thanks to the electron rich phosphine ligands and a strong acid, these complexes could be protonated without prior reduction, and the reported rate constants (kH) varied from 3.7 × 10−3 to 8 × 105 M−1 s−1. The log kH was found to vary linearly with the 16 [Fe2]+/0 potential (E+/0 1/2 ): + /0 log kH = 0.20 − 11.7E1/2

Our slope is somewhat smaller than that, which is in excellent agreement with a direct protonation of the Fe−Fe bond, with ΔG0/λ ≈ 1/4. As we have −ΔG0PT = 0.2−0.4 eV for the strongest acid, we estimate that λ > 1.0 eV. It is probably much larger; this is just a lower limit. However, we believe it is the first experimental attempt to estimate the reorganization energy for the protonation of a metal−metal bond. A value of λ > 1.0 eV does not seem unreasonable, given the large structural changes upon protonation. In contrast, Pickett’s correlation (eq 1) results in a slope of 27 eV−1, which is too large for a Marcus relationship according to eq 2, as the reactions are exergonic; this correlation assumes that each unit difference in a pKa value corresponds to a 59 meV difference in potential. In addition, protonation of the neutral complexes has been suggested to occur via a transient rotated structure by the initial protonation of a carbonyl.14,33 This may not be the case for the reduced complex. This supports our suggestion above that the protonation mechanisms for the neutral and reduced complexes are different and that protonation of the reduced complex is possibly a direct mechanism following a Marcus-type behavior.



CONCLUSIONS In summary, we have demonstrated that the application of flash-quench methods combined with time-resolved infrared and UV−vis absorption spectroscopy has permitted the spectroscopic characterization of key intermediates and their transformations in the proton-reduction cycle of complex 1 on hitherto unexplored time scales. The high time resolution is an advantage compared with those of electrochemical techniques. Thus, being able to exclude bimolecular transformations, such as disproportionation or dimerization, our results provide spectral signatures of the primary reduction product, 1−. On the basis of earlier DFT calculations and our experimental IR spectra, it could be concluded that the one-electron reduction of 1 is accompanied by a lengthening of the Fe−Fe bond. It does not lead to ligand rearrangements, such as the breaking of an Fe−S bond, as observed for the bdt complex, 2, or the formation of a bridging CO species. Upon the protonation of 1 − , the resulting 1H has an IR spectrum that was indistinguishable from that of the nonreduced complex, 1. This is consistent with the close similarity in structure between 1 and 1H, with a symmetric bridging hydride in the latter, as well as the closely similar electron densities on the metals. The free-energy dependence of the of Fe−Fe protonation kinetics for the reduced complex, 1−, is different from those of the protonations of phosphine-substituted analogues in their neutral forms. Specifically, the rate dependence on the acid pKa for protonation of 1− is smaller (Brønsted slope ≈ 0.4) and consistent with the Marcus theory of charge transfer, in contrast to the rate dependence on the Fe−Fe basicity of the neutral analogues (Brønsted slope = 0.68) . This suggests that the mechanism is direct proton transfer from the acid to the Fe−Fe bond of 1−, in contrast to the proposed indirect proton transfer via the CO ligands of the metal centers of the neutral analogues. The correlation suggests that the reorganization energy for the protonation of 1− is >1 eV, due to the significant reorganization of the complex structure. Nevertheless, the protonation rate constant is rather large, even with moderately weak acids (e.g., k ≈ 107 M−1 s−1 with trichloroacetic acid). Therefore, proton transfer to the one-electron-reduced complex, 1−, is likely to precede further reduction of the catalyst under many photochemical conditions, in which

(1)

which was rationalized by the fact that both oxidation and protonation concerned the metal−metal bond. The slope corresponded to a Brønsted coefficient β = 0.68 (i.e., Fβ/ 2.303RT = 11.7). It was proposed that this correlation could be used to estimate protonation rate constants that were outside the experimentally accessible regime if the [Fe2]+/0 potential is known. In the present study, we instead studied the protonation of the reduced complex, which is a different species. Even though both oxidation and protonation of 1− concern the metal−metal bond, reduction of 1 adds an electron to an antibonding orbital, as reflected in the significant increase in the Fe−Fe distance for 1 from 2.543 to 2.826 Å (see above). The resulting protonated species 1H has a much longer Fe−Fe bond (2.763 Å)23 than 1H+ (2.579 Å).22 Thus, protonation of 1 leads to a slight elongation of the bond by 0.036 Å, whereas protonation of 1− leads to a bond contraction by 0.063 Å. The direct effects of the electronic differences of the Fe−Fe bond may lead to the suspicion that the protonation kinetics of the reduced and oxidized complexes are different. The most useful way to compare them would be via the pKa values of 1− and the [FeFe]0 complexes of ref 16, but the latter are not available in the literature. When we compare the protonation kinetics for the three acids examined here, we see that log kH versus the pKa gives a slope of 0.4 (Brønsted coefficient = 0.4, see Figure S7). Each pKa unit difference of the acid means a 59 meV difference in the driving force for the protonation of the Fe−Fe bond. When plotting ln kH versus −ΔGPT0, the slope is instead 15.6 eV−1. The predicted slope from the Marcus theory for proton transfer at −ΔG0 = 0 is 20 eV−1 (eq 2):34,35 ∂(ln k) ΔG 0 1 = − + (1 ) λ 2RT ∂(ΔG 0)

(2) G

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

(10) Hunt, N. T.; Wright, J. A.; Pickett, C. Detection of Transient Intermediates Generated from Subsite Analogues of [FeFe] Hydrogenases. Inorg. Chem. 2016, 55, 399−410. (11) Borg, S. J.; Behrsing, T.; Best, S. P.; Razavet, M.; Liu, X.; Pickett, C. J. Electron Transfer at a Dithiolate-Bridged Diiron Assembly: Electrocatalytic Hydrogen Evolution. J. Am. Chem. Soc. 2004, 126, 16988−16999. (12) Wright, J. A.; Webster, L.; Jablonskyte, A.; Woi, P. M.; Ibrahim, S. K.; Pickett, C. J. Protonation of [FeFe]-hydrogenase sub-site analogues: revealing mechanism using FTIR stopped-flow techniques. Faraday Discuss. 2011, 148, 359−371. (13) Jablonskytė, A.; Wright, J. A.; Fairhurst, S. A.; Webster, L. R.; Pickett, C. J. [FeFe] Hydrogenase: Protonation of {2Fe3S} Systems and Formation of Super-reduced Hydride States. Angew. Chem., Int. Ed. 2014, 53, 10143−10146. (14) Jablonskyte, A.; Wright, J. A.; Pickett, C. J. Mechanistic aspects of the protonation of [FeFe]-hydrogenase subsite analogues. Dalton Trans. 2010, 39, 3026−3034. (15) Mirmohades, M.; Pullen, S.; Stein, M.; Maji, S.; Ott, S.; Hammarström, L.; Lomoth, R. Direct Observation of Key Catalytic Intermediates in a Photoinduced Proton Reduction Cycle with a Diiron Carbonyl Complex. J. Am. Chem. Soc. 2014, 136, 17366− 17369. (16) Jablonskytė, A.; Webster, L. R.; Simmons, T. R.; Wright, J. A.; Pickett, C. J. Electronic Control of the Protonation Rates of Fe−Fe Bonds. J. Am. Chem. Soc. 2014, 136, 13038−13044. (17) Lyon, E. J.; Georgakaki, I. P.; Reibenspies, J. H.; Darensbourg, M. Y. Coordination Sphere Flexibility of Active-Site Models for FeOnly Hydrogenase: Studies in Intra- and Intermolecular Diatomic Ligand Exchange. J. Am. Chem. Soc. 2001, 123, 3268−3278. (18) Thummel, R. P.; Lefoulon, F. Polyaza cavity shaped molecules. 11. Ruthenium complexes of annelated 2,2′-biquinoline and 2,2′-bi1,8-naphthyridine. Inorg. Chem. 1987, 26, 675−680. (19) Capon, J.-F.; Ezzaher, S.; Gloaguen, F.; Petillon, F. Y.; Schollhammer, P.; Talarmin, J.; Davin, T. J.; McGrady, J. E.; Muir, K. W. Electrochemical and theoretical investigations of the reduction of [Fe2(CO)5L{[small mu ]-SCH2XCH2S}] complexes related to [FeFe] hydrogenase. New J. Chem. 2007, 31, 2052−2064. (20) Tye, J. W.; Darensbourg, M. Y.; Hall, M. B. Correlation between computed gas-phase and experimentally determined solution-phase infrared spectra: Models of the iron−iron hydrogenase enzyme active site. J. Comput. Chem. 2006, 27, 1454−1462. (21) Eckert, F.; Leito, I.; Kaljurand, I.; Kütt, A.; Klamt, A.; Diedenhofen, M. Prediction of acidity in acetonitrile solution with COSMO-RS. J. Comput. Chem. 2009, 30, 799−810. (22) Liu, Y.-C.; Chu, K.-T.; Huang, Y.-L.; Hsu, C.-H.; Lee, G.-H.; Tseng, M.-C.; Chiang, M.-H. Protonation/Reduction of CarbonylRich Diiron Complexes and the Direct Observation of Triprotonated Species: Insights into the Electrocatalytic Mechanism of Hydrogen Formation. ACS Catal. 2016, 6, 2559−2576. (23) Greco, C.; Zampella, G.; Bertini, L.; Bruschi, M.; Fantucci, P.; De Gioia, L. Insights into the Mechanism of Electrocatalytic Hydrogen Evolution Mediated by Fe2(S2C3H6)(CO)6: The Simplest Functional Model of the Fe-Hydrogenase Active Site. Inorg. Chem. 2007, 46, 108−116. (24) Fiedler, A. T.; Brunold, T. C. Computational Studies of the HCluster of Fe-Only Hydrogenases: Geometric, Electronic, and Magnetic Properties and Their Dependence on the [Fe4S4] Cubane. Inorg. Chem. 2005, 44, 9322−9334. (25) Marhenke, J.; Pierri, A. E.; Lomotan, M.; Damon, P. L.; Ford, P. C.; Works, C. Flash and Continuous Photolysis Kinetic Studies of the Iron−Iron Hydrogenase Model (μ-pdt)[Fe(CO)3]2 in Different Solvents. Inorg. Chem. 2011, 50, 11850−11852. (26) Galinato, M. G. I.; Whaley, C. M.; Lehnert, N. Vibrational Analysis of the Model Complex (μ-edt)[Fe(CO)3]2 and Comparison to Iron-Only Hydrogenase: The Activation Scale of Hydrogenase Model Systems. Inorg. Chem. 2010, 49, 3201−3215. (27) Felton, G. A. N.; Vannucci, A. K.; Chen, J.; Lockett, L. T.; Okumura, N.; Petro, B. J.; Zakai, U. I.; Evans, D. H.; Glass, R. S.;

catalytic turnover is typically limited by the photogeneration of reducing equivalents.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.7b02687.



Complementary time-resolved UV−vis and IR spectroscopic data, more detailed description of the protonation kinetics, and dependence of the protonation rate constant on the pKa of the acid (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Leif Hammarström: 0000-0002-9933-9084 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Sascha Ott and Sonja Pullen (Uppsala University) for contributing to the synthesis of the diiron biomimetic catalyst and for discussions. This work was supported by the Swedish Research Council (grant 2016-04271) and the Olle Engkvist Byggmästare Foundation (grant 2016/3). S.W. gratefully acknowledges a grant from the Chinese Scholarship Council.



REFERENCES

(1) Gloaguen, F.; Lawrence, J. D.; Rauchfuss, T. B. Biomimetic Hydrogen Evolution Catalyzed by an Iron Carbonyl Thiolate. J. Am. Chem. Soc. 2001, 123, 9476−9477. (2) Na, Y.; Pan, J.; Wang, M.; Sun, L. Intermolecular Electron Transfer from Photogenerated Ru(bpy)3+ to [2Fe2S] Model Complexes of the Iron-Only Hydrogenase Active Site. Inorg. Chem. 2007, 46, 3813−3815. (3) Nann, T.; Ibrahim, S. K.; Woi, P.-M.; Xu, S.; Ziegler, J.; Pickett, C. J. Water Splitting by Visible Light: A Nanophotocathode for Hydrogen Production. Angew. Chem., Int. Ed. 2010, 49, 1574−1577. (4) Streich, D.; Astuti, Y.; Orlandi, M.; Schwartz, L.; Lomoth, R.; Hammarström, L.; Ott, S. High-Turnover Photochemical Hydrogen Production Catalyzed by a Model Complex of the [FeFe]-Hydrogenase Active Site. Chem. - Eur. J. 2010, 16, 60−63. (5) Samuel, A. P. S.; Co, D. T.; Stern, C. L.; Wasielewski, M. R. Ultrafast Photodriven Intramolecular Electron Transfer from a Zinc Porphyrin to a Readily Reduced Diiron Hydrogenase Model Complex. J. Am. Chem. Soc. 2010, 132, 8813−8815. (6) Jian, J.-X.; Ye, C.; Wang, X.-Z.; Wen, M.; Li, Z.-J.; Li, X.-B.; Chen, B.; Tung, C.-H.; Wu, L.-Z. Comparison of H2 photogeneration by [FeFe]-hydrogenase mimics with CdSe QDs and Ru(bpy)3Cl2 in aqueous solution. Energy Environ. Sci. 2016, 9, 2083−2089. (7) Wang, F.; Wang, W.-G.; Wang, X.-J.; Wang, H.-Y.; Tung, C.-H.; Wu, L.-Z. A Highly Efficient Photocatalytic System for Hydrogen Production by a Robust Hydrogenase Mimic in an Aqueous Solution. Angew. Chem., Int. Ed. 2011, 50, 3193−3197. (8) Adams, M. W. W. The structure and mechanism of ironhydrogenases. Biochim. Biophys. Acta, Bioenerg. 1990, 1020, 115−145. (9) Tschierlei, S.; Ott, S.; Lomoth, R. Spectroscopically characterized intermediates of catalytic H2 formation by [FeFe] hydrogenase models. Energy Environ. Sci. 2011, 4, 2340−2352. H

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Lichtenberger, D. L. Hydrogen Generation from Weak Acids: Electrochemical and Computational Studies of a Diiron Hydrogenase Mimic. J. Am. Chem. Soc. 2007, 129, 12521−12530. (28) Morris, R. H. Brønsted−Lowry Acid Strength of Metal Hydride and Dihydrogen Complexes. Chem. Rev. 2016, 116, 8588−8654. (29) Surawatanawong, P.; Tye, J. W.; Darensbourg, M. Y.; Hall, M. B. Mechanism of electrocatalytic hydrogen production by a di-iron model of iron-iron hydrogenase: A density functional theory study of proton dissociation constants and electrode reduction potentials. Dalton Trans. 2010, 39, 3093−3104. (30) Kramarz, K. W.; Norton, J. R. Slow Proton-Transfer Reactions in Organometallic and Bioinorganic Chemistry. In Progress in Inorganic Chemistry; Karlin, K. D., Ed.; Wiley: New York, 1994; Vol. 42, pp 1−65. (31) Henderson, R. A.; Ibrahim, S. K.; Oglieve, K. E.; Pickett, C. J. Selective release of dihydrogen upon deuteriation of polyhydrido complexes: studies on [WH3(OCMeO)(Ph2PCH2CH2PPh2)2]. J. Chem. Soc., Chem. Commun. 1995, 1571−1572. (32) Henderson, R. A.; Oglieve, K. E. The mechanisms of protonation of [M([small eta]5-C5H5)2H2](M = Mo or W). J. Chem. Soc., Dalton Trans. 1993, 3431−3439. (33) Liu, C.; Peck, J. N. T.; Wright, J. A.; Pickett, C. J.; Hall, M. B. Density Functional Calculations on Protonation of the [FeFe]Hydrogenase Model Complex Fe2(μ-pdt)(CO)4(PMe3)2 and Subsequent Isomerization Pathways. Eur. J. Inorg. Chem. 2011, 2011, 1080−1093. (34) Marcus, R. A. Theoretical relations among rate constants, barriers, and Broensted slopes of chemical reactions. J. Phys. Chem. 1968, 72, 891−899. (35) Marcus, R. A.; Sutin, N. Electron transfers in chemistry and biology. Biochim. Biophys. Acta, Rev. Bioenerg. 1985, 811, 265−322.

I

DOI: 10.1021/acs.inorgchem.7b02687 Inorg. Chem. XXXX, XXX, XXX−XXX