Research Article

Molecular mechanisms of cobalt-catalyzed hydrogen evolution

See allHide authors and affiliations

PNAS September 18, 2012 109 (38) 15127-15131; https://doi.org/10.1073/pnas.1213442109

  1. Contributed by Harry B. Gray, August 6, 2012 (sent for review July 19, 2012)

Abstract

Several cobalt complexes catalyze the evolution of hydrogen from acidic solutions, both homogeneously and at electrodes. The detailed molecular mechanisms of these transformations remain unresolved, largely owing to the fact that key reactive intermediates have eluded detection. One method of stabilizing reactive intermediates involves minimizing the overall reaction free-energy change. Here, we report a new cobalt(I) complex that reacts with tosylic acid to evolve hydrogen with a driving force of just 30 meV/Co. Protonation of CoI produces a transient CoIII-H complex that was characterized by nuclear magnetic resonance spectroscopy. The CoIII-H intermediate decays by second-order kinetics with an inverse dependence on acid concentration. Analysis of the kinetics suggests that CoIII-H produces hydrogen by two competing pathways: a slower homolytic route involving two CoIII-H species and a dominant heterolytic channel in which a highly reactive CoII-H transient is generated by CoI reduction of CoIII-H.

Hydrogen (H2) is a clean and renewable fuel (1⇓⇓–4). Hydrogenase enzymes that contain iron and nickel cofactors evolve hydrogen catalytically from water near the thermodynamic potential with turnover frequencies as high as 9,000 s-1 at 30 °C (5, 6). However, the large size and relative instability of these enzymes under aerobic conditions has led to the search for well-defined molecular catalysts that can produce hydrogen from water in a nonbiological system. Platinum is an excellent catalyst for proton reduction and hydrogen oxidation, but its scarcity and high cost limit widespread use. These considerations have led to the development of molecular catalysts that employ earth-abundant metals (7⇓⇓⇓–11).

Synthetic complexes of nickel (12⇓⇓–15), cobalt (16⇓⇓⇓⇓⇓⇓⇓⇓⇓–26), iron (27⇓–29), and molybdenum (30⇓–32) have been developed recently as electrocatalysts for the production of hydrogen. Co-diglyoxime complexes have been shown to generate hydrogen from protic solutions at relatively modest overpotentials. Given the broad interest and potential application in artificial photosynthesis, the mechanism of hydrogen formation has been the subject of many experimental (16, 18, 33⇓⇓⇓⇓⇓⇓–40) and theoretical (41⇓–43) investigations. Several pathways have been proposed (Fig. 1), all beginning with protonation of a CoI complex to form CoIIIH. H2 evolution can occur via protonation of CoIIIH or upon bimolecular combination of two CoIIIH species. Alternatively, CoIIIH can be reduced by CoI to a mixture of CoIIH and CoII. The generated CoIIH can react via similar heterolytic or homolytic pathways to generate H2. The key intermediate in these transformations (CoIIIH) has, until now, not been observed directly. We report here the results of extensive kinetics studies that shed new light on the mechanism of proton reduction.

Fig. 1.
Fig. 1.

Reaction pathways for the evolution of H2 from the reaction of a CoI complex with acid (HA). Models based on the X-ray crystal structures of CoI (2) and CoII (3) triphos complexes appear in the lower portion of the figure. Thermal ellipsoids are drawn at 50% probability; hydrogen atoms, out-of-sphere anions, and solvent molecules are omitted for clarity. The 1H NMR spectrum and structural model of the CoIII-H (4) intermediate formed in the reaction of 2 with TsOH•H2O appear in the upper right.

The triphos ligand (1,1,1-tris(diphenylphosphinomethyl)ethane) allows for facile tuning of reduction potentials by introduction of electron-withdrawing or electron-donating aryl groups into the framework. Several Co(triphos) species have been synthesized (44⇓⇓–47): protonation of [(PP3)CoH] (48) (PP3 = P(CH2CH2PPh2)3) led to the formation of the dihydride complex, [(PP3)Co(H)2][PF6] (49⇓–51); and two [(triphos)CoIII(H)2]+ species also have been fully characterized (52, 53).

Results and Discussion

Synthesis and Cyclic Voltammetry.

Upon treatment with triphos, cobalt(II) iodide undergoes spontaneous reduction to Co(triphos)(I) (1), a pink solid (SI Appendix, Fig. S1) (54). Crystal structure analysis of 1 reveals pseudotetrahedral geometry at CoI (SI Appendix, Fig. S2). The iodide can be abstracted with TlPF6 in 5∶1 tetrahydrofuran∶acetonitrile (THF∶CH3CN), to generate the cationic adduct [Co(triphos)(CH3CN)][PF6] (2) as a blue solid. The solid-state structure of 2 exhibits pseudotetrahedral geometry at the metal (Fig. 1), with the Graphic counter ion outside the coordination sphere.

Cyclic voltammograms of 2 at a glassy carbon electrode in a 0.1 M CH3CN solution of [nBu4N][PF6] feature a reversible wave at E1/2 = -0.68 V versus Fc+/0 (*) assigned to the CoII/I couple, and an irreversible oxidation at +0.66 V assigned to CoIII/II (Fig. 2 and SI Appendix). An irreversible reduction wave at -1.81 V is assigned to CoI/0, and the irreversible wave at -2.73 V is assigned to Co0/-1 (cyclic voltammograms of 1 are included in the SI Appendix). Voltammograms of 2 with varying amounts of p-toluenesulfonic acid monohydrate [TsOH•H2O, pKa = 8.7 in acetonitrile (55)] exhibit enhanced currents at potentials near that for the CoII/I couple; the currents increased markedly at more negative potentials (reduction of CoI) and at higher acid concentrations (Fig. 2). Cyclic voltammetry experiments performed in the absence of catalyst 2, or in the absence of the metal center (SI Appendix), produced no catalytic current, suggesting that 2 catalyzes H2 evolution. A maximum value of 9.5 for the ratio of the catalytic current to the peak current in the absence of acid (icat/ip,E = -1.8 V) was determined at an acid concentration of 14.7 mM, corresponding to an apparent turnover frequency of 1.8(2) × 101 s-1 at 25 °C (12, 56). Bulk electrolysis of a 0.3 mM solution of 2 in the presence of 6.0 mM TsOH•H2O in 65 mL 0.1 M acetonitrile solution of [nBu4N][PF6] at -1.0 V consumed 9.2 coulombs of charge after 2 h. Analysis of the gas mixture in the headspace of the electrolysis cell by gas chromatography confirmed production of H2 with a faradaic yield of 99 ± 10%.

Fig. 2.
Fig. 2.

Cyclic voltammograms of 2 (1.3 mM) in acetonitrile solution containing 0.1 M [nBu4N][PF6] in the presence of TsOH•H2O and ferrocene. Scan rate: 100 mV s-1; glassy carbon electrode.

Characterization of CoIII-H.

Addition of five equivalents of TsOH•H2O in CD3CN to a solution of 2 led to an immediate color change from blue to yellow/green. A singlet at δ 4.58 ppm was observed in the 1H NMR spectrum of the reaction mixture, suggesting formation of H2. The experiment was repeated in a 70 mL round bottom flask and allowed to stir at room temperature for 5 h. Analysis of the gas mixture in the headspace of the flask by gas chromatography confirmed production of 0.5 equivalents of H2. Broad, paramagnetically shifted peaks were detected in the 1H NMR spectrum of the reaction mixture, suggesting the formation of CoII. Recrystallization of the reaction mixture by vapor diffusion with diethyl ether led to the formation of yellow/green crystals. An X-ray structural study confirmed formation of a CoII complex containing a coordinated tosylate anion (3), with another tosylate outside the coordination sphere (Fig. 1). We conclude that protonation of the cationic CoI species leads to H2 evolution with consequent oxidation to CoII. Cyclic voltammetry results for 3 are included in the SI Appendix. The reverse reaction, the oxidation of H2 by CoII, was investigated as well. Addition of H2 (1 atm) to a 16.4 mM acetonitrile solution of 3 led slowly to formation of broad, paramagnetically shifted peaks at δ 15.40, -0.77, and -1.45 ppm (26% conversion after 1 d), consistent with formation of 2.

1H NMR spectra recorded immediately after mixing a solution of 2 with 10 equivalents of TsOH•H2O revealed a doublet of triplets at δ -7.64 ppm (Fig. 1), and the coupling constants (Jcis-P-H = 65 Hz; Jtrans-P-H = 130 Hz) accord with a Co–H complex (57⇓⇓–60), [(triphos)CoH(CH3CN)2]2+ (4). The 31P{1H} NMR spectrum displays two peaks at δ 37.3 and -1.7 ppm in a 2∶1 ratio, confirming the assignment (SI Appendix, Fig. S17C). The equilibrium constant for the protonation reaction (Fig. 1: K1 = k1/k-1) was determined from measurements of the variation of the CoIII-H concentration with [TsOH•H2O] by integration of the Co–H 1H NMR signal relative to a dihydroanthracene internal standard (Table 1). At higher acid concentrations, nearly constant apparent K1 values (approximately 0.2) are observed (Table 1, entries 2–5), corresponding to a pKa value of 8.0(8) for CoIII-H in CH3CN.

View this table:
Table 1.

Results of kinetics studies of the reaction of 2 with TsOH•H2O

Kinetics.

1H NMR spectra provided a convenient probe of the kinetics of conversion of CoIII-H to 3. The CoIII-H concentration exhibits an inverse dependence on reaction time (25 °C; [2]0 = 16.4 mM; [TsOH•H2O]0 = 0.164 M; dihydroanthracene internal standard), consistent with a reaction that is second order in the concentration of CoIII-H (Fig. 3), and a factor of two decrease in reaction half-life upon doubling [2]0 supports this interpretation. Greater acid concentrations lead to slower reactions (Table 1), yet, over the 0.082–0.656 M TsOH•H2O concentration range, the kinetics remain second order in [CoIII-H]. Second-order rate constants were determined at four different temperatures (15, 25, 35, and 45 °C; [2]0 = 16.4 mM; [TsOH•H2O]0 = 0.164 M; SI Appendix, Table S1) and activation parameters were extracted from an Eyring plot (ΔH‡ = 25.0 ± 0.9 kcal mol-1; ΔS‡ = 17.9 ± 2.8 eu; SI Appendix, Fig. S19). The 31P{1H} nuclear magnetic resonance (NMR) measurements (using Ph4PBr in a sealed capillary as an internal standard) of the CoIII–D (prepared from 2 and TsOD•D2O) decay kinetics revealed that the isotope effect, kobsH/kobsD, varied with acid concentration up to a maximum value of 0.57 (Table 1).

Fig. 3.
Fig. 3.

Kinetics of CoIII-H decay measured using 1H NMR spectroscopy; [Co]0 = 16.4 mM; [acid]0 = 0.164 M; 0.7 mL MeCN-d3.

Cyclic voltammetric measurements indicate that both CoI and Co0 complexes will effect proton reduction. The thermodynamic potential for hydrogen evolution using TsOH•H2O is -0.65 V (55), corresponding to a 30-meV/Co driving force for proton reduction by CoI. These energetics are consistent with experiments in which H2 is evolved from acidic solutions of 2, and 2 is regenerated from 3 in the presence of excess H2.

Our observation of CoIII-H has opened the way for direct probing of the mechanisms proposed for Co-catalyzed H2 evolution (Fig. 1). The second-order kinetics suggest that H2 elimination in a bimolecular reaction between two CoIII-H complexes could be a dominant pathway (k4, Fig. 1). But, since the specific rate of this reaction should be independent of acid concentration, the observed inverse acid dependence demonstrates that there is a competing pathway. Direct protonation of CoIII-H to produce CoIII and H2 is inconsistent with both the second-order kinetics and the decreasing rate constant with increasing acid concentration.

The third pathway for H2 evolution from CoI in acid involves reduction of CoIII-H to a CoII-H complex. In our kinetics experiments, the only available reductant is CoI (2), so the position of the protonation equilibrium in the CoII-H pathway will have a substantial impact on the observed kinetics (the inverse acid dependence can be explained by the absence of CoI at higher acid concentrations). In the limit of rapid equilibrium between CoI and CoIII-H, and irreversible following reactions, we predict that the disappearance of CoIII-H will obey second-order kinetics with an observed rate constant given by Eq. 1. Using the data in Table 1, we can estimate that k4 = 2.5 × 10-3 M-1 s-1 and k2 = 4.4 × 10-2 M-1 s-1. Numerical simulations of the reaction kinetics defined by the model outlined in Fig. 1 indicate that 2.5 × 10-3 M-1 s-1 is an upper limit for k4 and k2 ∼ 5 ± 1 × 10-2 M-1 s-1 (SI Appendix) and that over the acid concentration that we examined, H2 evolution proceeds almost exclusively via the pathway involving CoII-H. This conclusion is bolstered by cyclic voltammetry, wherein large catalytic currents do not occur until potentials consistent with Co0 (or CoII-H in the presence of acid) formation are achieved. The large and positive entropy of activation is likely due to the solvent dissociation from [(triphos)CoH(CH3CN)2]2+ (4). The values observed for the Embedded Image[1]kinetic isotope effect are difficult to assess due to a combination of protonation and electron transfer processes.

Armed with rate and equilibrium constants for the homogeneous evolution of H2 from CoI, we undertook numerical simulations of the cyclic voltammetry data. These simulations can be particularly challenging, owing to the large number of parameters involved in modeling the kinetics and the limited number of available observables. Nevertheless, we have found a set of parameters that describes the cyclic voltammetry of 2 in the presence of TsOH•H2O that is also consistent with the simulations of the homogeneous kinetics (SI Appendix, Figs. S4, S7, and S8). Of particular interest is the large specific rate (107 M-1 s-1) estimated for the reaction of CoII-H with TsOH•H2O. The comparable value for the reaction of a CoII(H)-diglyoxime complex with 6-bromo-2-naphthol (pKa ∼ 26.1) is 4 × 104 M-1 s-1 (36).

Our homogeneous and electrochemical kinetics analyses demonstrate that CoII-H is the active species for hydrogen evolution (Fig. 4), in accord with recent experimental (61, 62) and theoretical work (41⇓–43). The relatively slow evolution of H2 in the reaction of 2 with TsOH•H2O is a consequence of the slow reduction of 4 by 2, owing to inadequate driving force coupled with large electrostatic and reorganization barriers associated with the CoIII/II-H couple (35). An analogous reaction pathway is operative in H2 evolution reactions driven by CoI-diglyoxime complexes. In this case, however, the rate constant for CoIII-H reduction by CoI (9 × 106 M-1 s-1) is about eight orders of magnitude greater than that of the corresponding Co-triphos reaction. We cannot eliminate CoIII-H bond homolysis, producing CoII and H2, as a viable reaction path, but our homogeneous kinetics measurements show that the specific rate of this reaction is at least a factor of 20 smaller than that of CoIII-H reduction by CoI.

Fig. 4.
Fig. 4.

Schematic energy landscape for H2 evolution from CoI (2) and TsOH•H2O (HA). The vertical dimension in the plot corresponds to free energy; the lateral dimensions represent different routes through configuration space. The lowest energy pathway involves nearly isoenergetic protonation of {CoI}+, producing the common intermediate {CoIII-H}2+ (4). Endergonic reduction by {CoI}+ generates a highly reactive hydride, {CoII-H}+, that produces H2 upon reaction with a second equivalent of HA. Overprotonation of {CoI}+ traps the metal complex as {CoIII-H}2+, from which only a high-energy bimolecular pathway is available for the production of {CoII}2+ (3) and H2. Protonation of {CoIII-H}2+ is an unfavorable pathway, owing to formation of a high-energy {CoIII}3+ intermediate.

Conclusions

The failure of CoIII-H to evolve H2 heterolytically in the presence of acid is attributable to unfavorable thermodynamics associated with formation of CoIII. The CoIII/II reduction potentials in the glyoxime and triphos complexes are more than 0.5 V positive than the CoII/I couples, which are within 100 mV of the HA/H2 reduction potentials. The reduction of HA by CoI with concomitant formation of CoIII and H2 is energetically highly unfavorable and can be driven only by subsequent reduction of CoIII to CoII.

A sufficiently negative reduction potential is not the sole criterion for H2 evolution from Co complexes. Prior investigations of a PP3 triphos complex revealed that exposure of [(PP3)CoI(H)] to strong acid leads to a CoIII dihydride complex rather than H2 (48⇓–50). This behavior can be explained by the extreme proton basicity of [(PP3)Co-1]: double protonation to form [(PP3)CoIII(H)2] has a greater thermodynamic driving force than reduction of HA to H2. This behavior illustrates a key requirement of hydrogen evolution chemistry: The reduced metal complex must be sufficiently basic to deprotonate the acid, but if it is too basic the metal hydride intermediate will be too stable, inhibiting hydrogen formation.

A single-metal catalyst for the reduction of protons to hydrogen must utilize two redox couples. For hydrogen formation to be spontaneous, the average of the reduction potentials for the two couples must be more negative than the HA/H2 couple (63). To minimize the overpotential for catalysis, the difference between the two metal-based potentials should be as small as possible. The latter can be accomplished through the sequence reduction-protonation-reduction, wherein the protonation step helps moderate the potential for delivery of the second electron. In the case of Co triphos and glyoxime complexes, our research indicates that the CoII/I and CoIII-H/CoII-H couples are extremely well suited for catalysis of hydrogen evolution from acid (41).

Materials and Methods

General.

All manipulations of air- and moisture-sensitive materials were conducted under a nitrogen atmosphere in a Vacuum Atmospheres glovebox or on a dual-manifold Schlenk line. The glassware, including NMR tubes, were oven dried prior to use. Diethylether, tetrahydrofuran, dichloromethane, and acetonitrile were degassed and passed through activated alumina columns and stored over 4 Å Linde-type molecular sieves. The deuterated solvents were purchased from Cambridge Isotope Laboratories, Inc., and were dried over 4 Å Linde-type molecular sieves prior to use. The 1H spectra were acquired at room temperature unless otherwise noted through the use of Varian spectrometers and referenced to the residual 1H resonances of the deuterated solvent (1H: CD2Cl2, δ 5.32; CD3CN, δ 1.94) and are reported as parts per million relative to tetramethylsilane. The 31P{1H} NMR spectra were referenced to external phosphoric acid at δ 0 ppm. The 19F{1H} NMR spectra were referenced to external fluorobenzene at δ -113.15 ppm. Triphos and CoI2 were purchased from Sigma-Aldrich and used without further purification. The elemental analyses were performed by Midwest Microlabs.

Cyclic Voltammetry.

Electrochemical measurements were recorded in a nitrogen glovebox at 25 °C with a Pine Instruments WaveNow potentiostat. Electrochemical analyses were carried out in a three-electrode cell, consisting of a glassy carbon working electrode (surface area = 0.07 cm2), a platinum wire counter electrode, and a silver wire reference electrode. For all electrochemical measurements, the electrolyte solution was 0.1 M [nBu4N][PF6]. At the end of each experiment, ferrocene was added to the electrolyte solution and the ferrocenium/ferrocene couple was used to calibrate the reference electrode. In the case of acid titration cyclic voltammetry experiments, a frit separated silver wire reference electrode was used, and the ferrocene was added to the electrolyte solution at the beginning of the experiment.

NMR Measurements.

A 0.7 mL MeCN-d3 stock solution of 2 (16.4 mM) and dihydroanthacene (4.1 mM) was added to a 1 dram vial, which was charged with the corresponding amount of p-toluenesulfonic acid monohydrate. The reaction mixture was then transferred to a J. Young tube. Reaction progress was monitored by 1H NMR spectroscopy at the corresponding temperature, where integral areas of CoIII-H (δ -7.64 ppm, dt, 1H, JPcis-H = 65 Hz; JPtrans-H = 130 Hz) relative to the aliphatic protons of dihydroanthracene internal standard (δ 3.9 ppm, s, 4H) were obtained at intervals of 2 min (SI Appendix, Fig. S18).

A 0.7 mL MeCN stock solution of 2 (16.4 mM) was added to a 1 dram vial, which was charged with the corresponding amount of p-toluenesulfonic acid monohydrate. The reaction mixture was then transferred to a J. Young tube, charged with a sealed capillary tube with Ph4PBr (8.2 mM). Reaction progress was monitored by 31P{1H} NMR spectroscopy at 25 °C, where integral areas of the phosphorous peaks in [(triphos)CoIII-H]2+ (δ 37.3 ppm, 2P, Co-Pcis) relative to the Ph4PBr internal standard (δ 24.4 ppm, s, 1P) were obtained at intervals of 4 min.

Acknowledgments

We thank Lawrence M. Henling and the late Dr. Michael W. Day for crystallographic assistance, and Dr. Jay A. Labinger for insightful comments. Our work is supported by the National Science Foundation Center for Chemical Innovation in Solar Fuels (CHE-0802907); Center for Chemical Innovation postdoctoral fellowship to S.C.M. We thank Chevron-Phillips for additional support. The Bruker KAPPA APEX II X-ray diffractometer was purchased via a National Science Foundation Chemistry Research Instrumentation and Facilities: Departmental Multi-User Instrumentation (CRIF:MU) award to the California Institute of Technology, CHE-0639094.

Footnotes

  • ↵1To whom correspondence may be addressed. E-mail: winklerj{at}caltech.edu or hbgray{at}caltech.edu.

  • Author contributions: S.C.M. designed research; S.C.M. performed research; S.C.M. contributed new reagents/analytic tools; S.C.M., J.R.W., and H.B.G. analyzed data; and S.C.M., J.R.W., and H.B.G. wrote the paper.

  • The authors declare no conflict of interest.

  • Data deposition: The atomic coordinates have been deposited in the Cambridge Structural Database, Cambridge Crystallographic Data Centre, Cambridge CB2 1EZ, United Kingdom http://www.ccdc.cam.ac.uk, [CSD reference numbers 838815 (1), 844589 (2), and 846384 (3)].

  • This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1213442109/-/DCSupplemental.

  • ↵*All electrode potentials are reported relative to the ferricenium/ferrocene (Fc+/0) couple.

Back to top
Article Alerts
Email Article
Citation Tools
Request Permissions
Mechanisms of cobalt catalyzed hydrogen evolution
Smaranda C. Marinescu, Jay R. Winkler, Harry B. Gray
Proceedings of the National Academy of Sciences Sep 2012, 109 (38) 15127-15131; DOI: 10.1073/pnas.1213442109
del.icio.us logo Digg logo Reddit logo Twitter logo CiteULike logo Facebook logo Google logo Mendeley logo

Article Classifications

Proceedings of the National Academy of Sciences: 109 (38)
Table of Contents

Submit

Jump to section

You May Also be Interested in

Similar Articles