Cobalt and nickel diimine-dioxime complexes as molecular electrocatalysts for hydrogen evolution with low overvoltages

Abstract

Hydrogen production through the reduction of water appears to be a convenient solution for the long-run storage of renewable energies. However, economically viable hydrogen production requests platinum-free catalysts, because this expensive and scarce (only 37 ppb in the Earth's crust) metal is not a sustainable resource [Gordon RB, Bertram M, Graedel TE (2006) Proc Natl Acad Sci USA 103:1209–1214]. Here, we report on a new family of cobalt and nickel diimine-dioxime complexes as efficient and stable electrocatalysts for hydrogen evolution from acidic nonaqueous solutions with slightly lower overvoltages and much larger stabilities towards hydrolysis as compared to previously reported cobaloxime catalysts. A mechanistic study allowed us to determine that hydrogen evolution likely proceeds through a bimetallic homolytic pathway. The presence of a proton-exchanging site in the ligand, furthermore, provides an exquisite mechanism for tuning the electrocatalytic potential for hydrogen evolution of these compounds in response to variations of the acidity of the solution, a feature only reported for native hydrogenase enzymes so far.

Hydrogen production, through the reduction of water, appears to be a solution to consider for the long-run storage of renewable energetic resources (1). The hydrogen evolution reaction (her) is apparently a very simple reaction but, as with most multielectronic processes, it is a slow process that should be catalyzed. Nickel-based electrodes can be used for that aim in strongly alkaline media, but this technology results in average energetic yields (≈50–70%), suffers from corrosion issues, and can hardly be miniaturized due to the lack of polymer exchange membrane (PEM) material that is stable under these conditions. Actually, technological research now focuses on PEM electrolytic cells for which carbon-supported platinum nanoparticles are generally used as catalysts. This scarce and expensive noble metal is, however, not a renewable resource. Thus, progress to a mature hydrogen economy depends on breakthroughs in finding new catalytic materials based on earth-abundant elements (2). To that aim, the combination of a common electrode material with a coordination complex of a first-row transition metal, able to catalyze the reaction at a potential close to the standard potential of the H⁺/H₂ couple, seems attractive. Macrocyclic cobalt complexes are very competitive in that respect (3). In the past four years, we and others reported on cobaloximes—cobalt complexes containing bridged bidentate oximato ligands (see Scheme 1)—(4–7) as a new class of such catalysts with good catalytic stability and ≈250–300 mV overvoltage—the difference between the observed catalytic potential and the thermodynamic H₂/2H⁺ + 2 e⁻ equilibrium. To date, these catalysts are among the most efficient ones and, thus, are currently exploited for the construction of H₂-evolving photocatalytic devices that actually rival previously reported systems using Pt nanoparticles as catalysts (8–12). However, despite the pseudomacrocyclic nature of their equatorial ligand, cobaloximes are prone to hydrolysis under acidic conditions: protonation at two oximato functions in the equatorial plane indeed results in the decomposition of the complex through the cleavage of the tetradentate macrocyclic ligand into two dioxime ligands and suppression of the chelate effect. To date, only a few of such catalysts have been reported (4–7), and there is an urgent need to expand the series to facilitate their grafting on surfaces for the construction of H₂-evolving electrodes or to allow their implementation into original supramolecular light-harvesting devices.

Scheme 1.

Molecular structures for the various Co and Ni diimine-dioxime complexes considered in this study. The structure of the cobaloxime complex [Co(dmgBF₂)₂L₂] is represented in the Upper Right.

Here, we report on a new family of nickel and cobalt H₂-evolving catalysts with stable tetradentate ligands providing a diimine-dioxime coordination sphere. Although the substitution of imine for oxime residues allows keeping of the remarkable electrocatalytic properties in terms of overvoltage and turnover frequencies (TOF), the tetradentate nature of the equatorial ligand warrants a high stability against hydrolysis. Insights into the catalytic mechanism are given with emphasis on the role of proton relays in the second coordination sphere of the metal centers.

Results and Discussion

Scheme 1 shows the structures of the various compounds considered in this study. A first series is based on the diimine-dioxime ligand, N²,N²′-propanediylbis(2,3-butandione 2-imine 3-oxime), further abbreviated as (DOH)(DOH)pn (13, 14). A new ligand, ((MOH)(MOH)pn), where the methyl oxime groups in (DOH)(DOH)pn are replaced by hydrogen, has been prepared from E-pyruval-oxime and 1,3-diaminopropane according to the general procedure established by Marzilli et al. (14). Cobalt and nickel complexes of these two ligands with various axial ligands have been synthesized. We also prepared the BF₂-linked analogues, [Co((DO)₂BF₂)pnBr₂] and [Ni((DO)₂BF₂)pn](ClO₄), starting from Co(OAc)₂·4H₂O and Ni(OAc)₂·4H₂O, respectively, and carrying out the reaction in diethylether in the presence of excess BF₃·Et₂O. Alternatively, we observed that [Co((DO)₂BF₂)pnBr₂] forms spontaneously when [Co(DO)(DOH)pnBr₂] is placed in the presence of HBF₄·Et₂O in CH₃CN. The X-ray structures of [Co(MO)(MOH)pnCl₂] (Fig. 1) and [Co(DO)(DOH)pnBr₂] (Fig. S1) reveal an octahedral environment around the cobalt ion with two axial halide ligands located trans to each other. The same coordination sphere is assumed for [Co((DO)₂BF₂)pnBr₂], whereas the nickel ion in [Ni(DO)(DOH)pn](ClO₄) (Fig. S2) and [Ni((DO)₂BF₂)pn](ClO₄) is in a square-planar configuration. The nickel center in [Ni(MO)(MOH)pnCl] likely adopts a square-pyramidal configuration as reported for the [Ni((DO)₂BF₂)pnI] analogue (15).

Fig. 1.

X-ray structure of [Co(MO)(MOH)pnCl₂] (50% thermal ellipsoids). Selected bond lengths (Å) and angles (°): Co-N(4) 1.8902(19), Co-N(1) 1.8952(18), Co-N(3) 1.9140(17), Co-N(2) 1.9247(17), Co-Cl(1) 2.2373(6), Co-Cl(2) 2.2400(6), O(1)-N(1) 1.331(2), O(2)-N(4) 1.324(2), O(1)-H(1O) 1.03(3), O(2)-H(1O) 1.43(3), O(1)…O(2) 2.443(2), N(4)-Co-N(1) 96.79(8), N(4)-Co-N(3) 82.00(8), N(1)-Co-N(3) 178.77(8), N(4)-Co-N(2) 177.91(8), N(1)-Co-N(2) 82.04(7), N(3)-Co-N(2) 99.17(8), N(4)-Co-Cl(1) 90.78(6), N(1)-Co-Cl(1) 90.12(6), N(3)-Co-Cl(1) 89.65(6), N(2)-Co-Cl(1) 90.95(5), N(4)-Co-Cl(2) 88.64(6), N(1)-Co-Cl(2) 89.58(6), N(3)-Co-Cl(2) 90.64(6), N(2)-Co-Cl(2) 89.62(5), Cl(1)-Co-Cl(2) 179.32(3), O(1)-H(1O)…O(2) 169(3).

Table 1 summarizes the redox features of these diimine-dioxime compounds obtained from cyclic voltammetry measurements, together with those of two reported cobaloximes (4, 5). Cobalt complexes display two, successive, reversible, monoelectronic, electron-transfer steps assigned to Co^III/Co^II and Co^II/Co^I processes, respectively, as reported in ref. 14. The proton-linked diimine-dioxime cobalt derivatives are reduced to the Co^I state at potentials much more positive than that of H-linked cobaloximes, such as [Co(dmgH)₂pyCl] and [Co(dmgH)₂(H₂O)₂] (dmg²⁻ = dimethylglyoximato dianion). This shift indicates a lower electron density on the metal center. This density also depends on the nature (H vs. Me) of the oxime substituent, and [Co(MO)(MOH)pnCl₂] displays the highest potential values. The introduction of one difluoroboryl bridge between the oxime groups results in a further 200–250-mV positive shift for both Co^III/Co^II and Co^II/Co^I processes.

Table 1.

Redox potentials of Co and Ni complexes (V vs. Fc^+/0) determined from cyclic voltammetry measurements (100 mV·s⁻¹) in CH₃CN at a GC electrode in the presence of nBu₄NBF₄ (0.1 mol·L⁻¹)

Complexe	E1/2 (MIII/MII)	E1/2 (MII/MI)
[Co(DO)(DOH)pnBr2]	−0.57	−1.11
[Co(MO)(MOH)pnCl2]	−0.65	−0.96
[Co((DO)2BF2)pnBr2]	−0.38	−0.84
[Co(dmgBF2)2(CH3CN)2] (4)*	−0.2 (irrev. cathodic)	−0.93
[Co(dmgH)2pyCl]	−1.05 (irrev. anodic)	−1.48
[Co(dmgH)2(OH2)2]	—	−1.48
[Ni(DO)(DOH)pn](ClO4)	—	−1.22
[Ni((DO)2BF2)pn](ClO4)	—	−0.97
[Ni(MO)(MOH)pnCl]	—	−1.04
[Co(DO)(DOH)pn(PPh3)]	−0.52	−0.82
[Co(DO)(DOH)pnBr(PPh3)]	−0.18	−0.94

*These values have been checked in the present study for correct reference toward the Fc^+/0 couple.

The nickel derivatives only display one reversible monoelectronic redox process assigned to Ni^II/Ni^I. Redox potentials are ≈100 mV more negative than the corresponding Co^II/Co^I derivatives, reflecting a higher electron density at the metal center. Here again, the introduction of a difluoroboryl bridge in [Ni((DO)₂BF₂)pn](ClO₄) is followed by a 250-mV positive shift of the Ni^II/Ni^I wave.

Fig. 2Top shows the cyclic voltammograms (CV) of [Co((DO)₂BF₂)pnBr₂] recorded in CH₃CN at a glassy carbon electrode in the presence of increasing amounts of p-cyanoanilinium tetrafluoroborate (pK_a = 7.6 in CH₃CN) (16) as the source of protons. A catalytic wave develops at potentials close to E°(Co^II/I) (Tables 1 and 2) and corresponds to electrocatalytic hydrogen evolution with an overvoltage of ≈230 mV—the overvoltage can be simply obtained by subtracting the standard potential for the H⁺/H₂ couple (E⁰(H⁺/H₂)) listed in Table 2 from the measured electrocatalytic potential (17). The same behavior, but with lower overvoltages (190 and 110 mV), is observed with weaker acids such as tosic acid monohydrate and anilinium tetrafluoroborate, respectively (see Table 2 and Figs. S3 and S4). In the case of trifluoroacetic acid (TFA) (Fig. S5), hydrogen apparently evolves with no overvoltage. Actually, the classical determination of the standard potential of the H⁺/H₂ couple (17) does not take into consideration the homoconjugation reaction (self-association of the acid with its conjugated base), leading to an underestimation of the overvoltage that can reach >100 mV under the conditions used. This mechanism indeed increases the effective acidity of these acids, in particular for TFA (18).* In terms of overvoltages, [Co((DO)₂BF₂)pnBr₂] is quite comparable to [Co(dmgBF₂)₂(CH₃CN)₂] (4, 19). The catalytic current enhancement is, however, slightly greater than that obtained with [Co(dmgBF₂)₂(CH₃CN)₂], for which an overall catalytic rate constant of 770 mol⁻¹·L·s⁻¹ has been estimated under the same conditions (19).

Fig. 2.

Cyclic voltammograms of [Co((DO)₂BF₂)pnBr₂] and [Co(DO)(DOH)pnBr₂]. (Top) Cyclic voltammograms of [Co((DO)₂BF₂)pnBr₂] (1 mM) in the presence of p-cyano-anilinium tetrafluoroborate (0, 1, 3, 5, 10 equiv). (Middle) Cyclic voltammograms of [Co(DO)(DOH)pnBr₂] (1 mM) in the presence of anilinium tetrafluoroborate. (Inset) Zoom on the Co^III/Co^II process. (Bottom) Same catalyst, but p-cyanoanilinium tetrafluoroborate; glassy carbon electrode–CH₃CN solution containing 0.1 mol·L⁻¹ [ⁿBu₄N][BF₄]. Scan rate: 100 mV·s⁻¹.

Table 2.

Electrocatalytic H₂ evolution peak potentials (V vs. Fc^+/0) in the presence of Co and Ni complexes measured at a GC electrode by cyclic voltammetry (100 mV·s⁻¹) in the presence of 3 equiv of various acids in CH₃CN (nBu₄NBF₄ 0.1 mol·L⁻¹)

Acid	p-Cyanoanilinium	Tosic acid	Anilinium	TFA	Et3NHCl
pKa	7.6	8.7	10.7	12.65	18.7
E0(H+/H2) in CH3CN*	−0.59	−0.66	−0.77	−0.89	−1.25
[Co((DO)2BF2)pnBr2]	−0.82	−0.85	−0.88	−0.89	—
[Co(DO)(DOH)pnBr2]	−0.78	−0.83	−0.97	−1.10	—
[Ni(DO)(DOH)pn](ClO4)	−0.95	−1.01	−1.09	−1.20	−1.54
[Ni((DO)2BF2)pn](ClO4)	−0.95	−0.94	−0.98	−1.00	—
[Ni(MO)(MOH)pnCl]	−0.97	†	−1.0	−1.00	—
[Co(dmgBF2)2(CH3CN)2]	−0.80	−0.93	−0.89	−0.90
[Co(dmgH)2pyCl]	—	—	—	—	−1.48

*These calculated values do not take into account the homoconjugation reaction (self-association of the the acid with its conjugated base) that may increase the effective acidity of these acids, in particular for tosic acid and TFA.

^†The species is unstable in the presence of high concentrations of acid.

Switching to [Co(DO)(DOH)pnBr₂] as the catalyst (Table 2 and Fig. 2 Middle and Bottom), hydrogen evolution occurs at a potential shifted anodically with regard to that of the Co^II/Co^I couple—the stronger the acid, the larger the shift (Table 2, Fig. 2 Middle and Bottom, and Figs. S6 and S7).

Regarding the nickel complexes, a similar anodic shift of the catalytic wave with regard to the Ni^II/I process is observed, going from BF₂-linked to H-linked nickel complexes (Table 2 and Figs. S8–S11). Whatever the acid used, [Ni((DO)₂BF₂)pn](ClO₄) catalyzes hydrogen evolution exactly at the potential of the Ni^II/Ni^I couple. Interestingly, [Ni(DO)(DOH)pn](ClO₄) catalyzes H₂ evolution from the weak acid, Et₃NH⁺ (pK_a = 18.7), but at potentials more negative than that of the Ni^II/I couple (Fig. S8).

Hydrogen evolution was confirmed by bulk electrolysis experiments performed in the presence of the various proton donors previously investigated. The results are reported in Table 3. Several 10ths of turnovers can be achieved with approximately quantitative faradaic yields in most cases using [Co(DO)(DOH)pnBr₂] and [Co((DO)₂BF₂)pnBr₂] as the catalysts, thus comparing well with the previously reported cobaloxime complexes (Table 3). The nickel compound [Ni(MO) (MOH)pnCl] appears less stable with a lower faradaic yield and an obvious decomposition after 3 h in the presence of p-cyanoanilinium tetrafluoroborate. [Ni(DO)(DOH)pn](ClO₄) decomposes more rapidly even in less-acidic solution. No hydrogen could be detected in the course of electrolysis experiments when [Co(MO)(MOH)pnCl₂] was used as the catalyst, although some modifications of its cyclic voltammograms could be observed in the presence of acid.

Table 3.

Bulk electrolysis experiments for H₂ evolution from various acids (50 mmol) catalyzed by Co and Ni complexes (0.5 mmol) carried out in CH₃CN (nBu₄NBF₄ 0.1 mol·L⁻¹) at a graphite rod electrode

Catalyst	Proton source	Applied potential, V vs. Fc+/0	TON/3 h	Faradaic yield, %
[Co(DO)(DOH)pnBr2]	p-CN(C6H4)NH3(BF4)	−0.78	40	92
	Tosic acid	−0.83	4	50
	PhNH3(BF4)	−0.97	27	90
	TFA	−1.0	20	91
[Co((DO)2BF2)pnBr2]	p-CN(C6H4)NH3(BF4)	−0.82	20	100
	Tosic acid	−0.85	15	100
	TFA	−0.89	7	90
[Ni(DO)(DOH)pn](ClO4)	Et3NHCl	−1.54	7	30
[Ni((DO)2BF2)pn](ClO4)	p-CN(C6H4)NH3(BF4)	−0.95	4	60
[Ni(MO)(MOH)pnCl]	p-CN(C6H4)NH3(BF4)	−1.1	20	70
[Co(dmgBF2)2(CH3CN)2]	p-CN(C6H4)NH3(BF4)	−0.92	50	≈100
[Co(dmgH)2pyCl]	Et3NHCl	−1.32	50	>85

The amount of evolved hydrogen has been determined through GC analysis.

A striking difference of this series of diimine-dioxime cobalt complexes with the previously reported cobaloxime complexes resides in the large robustness of the former toward hydrolysis under acidic conditions. The stability of millimolar solutions of [Co(DO)(DOH)pnBr₂] and [Co(DO)₂BF₂)pnBr₂] in the presence of 30 equiv of p-cyanoanilinium tetrafluoroborate—the stronger acid in the above series—has been monitored spectroscopically in CH₃CN, and no evolution of either ¹H NMR (in CD₃CN) or UV-visible spectra could be noticed over two weeks. As a comparison, [Co(dmgBF₂)₂(CH₃CN)₂] was decomposed under the same conditions with a half-life time of 15h (4).

Three issues can be addressed regarding the mechanism for H₂ evolution catalyzed by these compounds: (i) the oxidation state of the active species initiating the catalytic cycle, (ii) the nature of the hydrogen evolution step (20), where two pathways, either heterolytic and monometallic or homolytic and bimetallic, may account for hydrogen evolution (Scheme 2), and (iii) the specific role of the bridging link, H vs. BF₂, as far as the electrocatalytic potential is concerned.

Scheme 2.

Possible mechanistic pathways for H₂ evolution.

In the case of [Co((DO)₂BF₂)pnBr₂], the observation of a catalytic wave developing very close to the Co^II/Co^I potential is an indication for hydrogen evolution from a Co^III-hydride intermediate formed by protonation of the electrochemically generated Co^I species (Scheme 2) (21). We calculated the equilibrium constants for both homolytic and heterolytic H₂ evolution routes from p-cyanoanilinium in CH₃CN using the methodology developed by Spiro et al. (22) (SI Text and Table S1). As a conclusion of this analysis, heterolytic hydrogen evolution (Δ_rG⁰_hetero = −2 kJ·mol⁻¹) is far less favorable and thus, less likely than the homolytic one (Δ_rG⁰_homo = −24 kJ·mol⁻¹). On that basis, the reactivity of this compound resembles that of [Co(dmgBF₂)₂(CH₃CN)₂] (Δ_rG⁰_hetero = 2.5 kJ·mol⁻¹, Δ_rG⁰_homo = −33 kJ·mol⁻¹).

In the case of the H-linked complex [Co(DO)(DOH)pnBr₂], the situation is quite different because the electrocatalytic wave is observed at potentials more positive than that of the Co^II/Co^I process. The thermodynamic constants determined in acetonitrile indicate that the heterolytic route is much more favorable (Δ_rG⁰_hetero = −24 kJ·mol⁻¹). Homolytic hydrogen evolution is also thermodynamically possible, so that both mechanisms may compete depending on the experimental conditions. This mechanistic issue was addressed by studying the reaction of a cobalt(I) derivative with protons. For that purpose, the cobalt(I) compound [Co(DO)(DOH)pn(PPh₃)] was prepared through the reduction of [Co(DO)(DOH)pnBr₂] by NaBH₄ in the presence of one equivalent of PPh₃ (23). The rotating-disk electrode steady-state voltammogram of this blue cobalt(I) compound (λ_max = 650 nm in benzene) displays an anodic wave at −0.87 V vs. Fc/Fc⁺ (E_pc − E_pa = 90 mV) in CH₃CN, corresponding to its oxidation to the Co^II state. Addition of an excess of p-cyanoanilinium bromide to a solution of this compound in CH₃CN or CH₂Cl₂ was shown to generate one equiv of H₂ per cobalt complex within 15 min (Scheme 3), thus supporting the hypothesis that the Co^I species is the active catalytic species. UV-vis, ¹H NMR, CV, and rotating-disk electrode steady-state voltammetry measurements show that the resulting solution contains the Co^III complex [Co(DO)(DOH)pnBr₂], together with free PPh₃.

Scheme 3.

Stoichiometric hydrogen formation from a reduced Co^I complex.

However, UV-visible spectroscopic monitoring of the reaction shows that a Co^II species (λ_max = 470 in CH₃CN) is quantitatively formed during the first 30 s of the reaction and further evolves to the final Co^III state. This observation thus supports a homolytic mechanism for H₂ evolution yielding an intermediate Co^II species. We independently prepared this Co^II compound by reacting (DOH)(DOH)pn, CoBr₂·H₂O, and PPh₃ in acetone. The resulting compound displays reversible Co^II/Co^I and quasi-reversible Co^II/Co^III processes in CH₃CN, both at (Table 1) distinct from those observed for [Co(DO)(DOH)pnBr₂] or [Co(DO)(DOH)pnPPh₃], consistent with a [Co(DO)(DOH) pnBr(PPh₃)] composition. This species is stable in CH₃CN, but readily reacts with p-cyanoanilinium, generating H₂ (half an equiv) and the cobalt(III) complex [Co(DO)(DOH)pnBr₂]. We thus propose that the slow disproportionation of the Co^II species generates the Co^III and Co^I compounds, with the latter compound further reacting with protons and evolving H₂ as described above.^†

Finally, the anodic shift of the electrocatalytic wave with regard to the Co^II/Co^I potential in the presence of strong acids is consistent with the formation of a protonated, and thus easier to reduce, catalytically active species. Protonation at the O–H…O bridge, yielding a species bearing two protonated oxime ligands is likely. This hypothesis is further supported by the following observations: (i) a careful examination of the cyclic voltammograms of [Co(DO)(DOH)pnBr₂] in Fig. 2 (see Inset) shows a slight shift of the Co^III/II system, when the acid concentration is increased. The shift of the Co^III/II couple is quite low relative to that of the electrocatalytic wave developing on the Co^II/I process. This difference indicates that the more reduced the species, the more displaced the protonation equilibrium, probably as a direct effect of the overall charge of the complex. (ii) The ¹H NMR signal assigned to the oxime-bound proton disappears in the presence of anilinium, indicative of a fast exchange process. (iii) The {BF₂}-linked cobalt complex cannot be protonated at the oxime functions, engaged in covalent bonds with the boron atom. Accordingly this complex catalyzes hydrogen evolution at the Co^II/Co^I potential strictly, with no effect of the strength of the acid used.

The presence of an H⁺-exchanging site in the catalyst provides the catalyst with a mechanism to adjust its electrocatalytic potential for hydrogen evolution to the acido-basic conditions of the solution and allows keeping the overvoltage for the reduction of acids within reasonable values over a wide range of pK_as. The protonation of amine residues in the second coordination sphere of some H₂-evolving catalysts has been used to lower the overvoltage (24–27) but, to the best of our knowledge, such a progressive shift of the electrocatalytic potential, as a function of the strength of the proton source used, has never been observed for a synthetic molecular catalyst. Similarly, hydrogenase enzymes (28) have this ability to adapt their electrocatalytic potential by modifying their surface charge through protonation and, thus, to catalyze H₂/H⁺ interconversion near the equilibrium over a wide range of pH values (29). Interestingly, such a H⁺-exchanging site in close proximity to the catalytic metal center opens the possibility for intramolecular proton transfer steps along the catalytic cycle, enhancing turnover frequency. Noteworthy, DuBois and colleagues have shown how the presence of a basic site in the second coordination sphere of a catalytic site, able to accommodate a hydride ligand, may enhance its catalytic activity for hydrogen evolution (24, 25). Whether such an open oxime-bridge is involved in the proton transfer steps along the catalytic cycle for hydrogen evolution (Scheme 4), promoting fast Co-H formation and/or accelerating its protonation via a proton–hydride interaction, is a fascinating hypothesis that needs to be further investigated.

Scheme 4.

Possible intramolecular proton transfer steps along the catalytic mechanism of H-linked dioxime H₂-evolving catalysts.

Conclusion

In summary, we have identified a previously undescribed family of molecular cobalt and nickel-based H₂-evolving catalysts that displays excellent catalytic properties (overvoltages, turnover frequencies) and greater stabilities than the previously reported cobaloximes. These catalysts are thus promising candidates for the molecular engineering of an H₂-evolving cathode material or the design of unique supramolecular H₂-evolving photocatalytic systems.

Materials and Methods

Methods and Instrumentation.

NMR spectra were recorded at room temperature in 5-mm tubes on a Bruker AC 300 spectrometer equipped with a QNP probehead, operating at 300.0 MHz for ¹H and 75.5 MHz for ¹³C. Solvent peaks are used as internal references relative to Me₄Si for ¹H and ¹³C chemical shifts. Elemental analyses were done at the service central d'analyse of the Centre National de la Recherche Scientifique, Vernaison, France.

Electrochemical Measurements.

Electrochemical analysis was done by using an EG&G potentiostat, model 273A.

The electrochemical experiments were carried out in a three-electrode electrochemical cell under nitrogen atmosphere, using glassy carbon working electrodes for cyclic voltammetry and a graphite rod for bulk electrolysis experiments. The auxiliary electrode was a platinum grid. The reference electrode was based on the Ag/AgCl/KCl 3M couples. All potentials given in this work are with respect to the ferricinium/ferrocene (Fc⁺/Fc) couple, whose potential has been measured after each experiment by adding authentic Fc in the cell. The experiments were conducted in acetonitrile with tetrabutylammonium tetrafluoroborate, [Bu₄N]BF₄, as the supporting electrolyte. Additions of acid were done by syringe from solutions of the acid in the same electrolyte.

For bulk electrolysis experiments, a tight cell was used, connected to a Perkin-Elmer Clarus 500 gas chromatograph equipped with a porapack Q 80/100 column (6′ 1/8″) thermostated at 40 °C and a TCD detector thermostated at 100 °C. The platinum-grid counter electrode was placed in a separate compartment connected by a glass-frit and filled with the electrolytic solution. During controlled-potential coulometry experiments, the cell was flushed with nitrogen (5 ml·min⁻¹), and the output gas was sampled (100 μl) every 2 min and analyzed. Calibration was done by using a commercial electrode loaded with platinum active layer (BASF LT140EWSI, platinum loading 0.5 mg·cm⁻²).

Synthesis.

(DOH)(DOH)pn, [Co(DO)(DOH)pnCl₂] and [Ni(DO)(DOH)pn](ClO₄) were prepared as described in refs. 13 and 30. [Co(DO)(DOH)pn(PPh₃)] was prepared in the glove box via a described procedure (23). Other chemicals were used as received. Commercial acetonitrile for electrochemistry was degassed by bubbling nitrogen through it. The supporting electrolyte (n-Bu₄N)BF₄ was prepared from (n-Bu₄N)HSO₄ (Aldrich) and NaBF₄ (Aldrich) and dried overnight at 80 °C under vacuum.

[Co(DO)(DOH)pnBr₂].

[Co(DO)(DOH)pnCl₂] was dissolved in water and NaBr (10 equiv) was added. After 5 min, the green precipitate was filtered and washed with Et₂O.

[Co((DO)₂BF₂)pnBr₂].

Co(OAc)₂·4H₂O (518 mg, 2.08 mmol) and BF₃·OEt₂ (2.63 ml, 20.8 mmol) were added to a suspension of (DOH)(DOH)pn (500 mg, 2.08 mmol) in Et₂O (30 ml). The mixture was stirred for 24 h at room temperature under argon. The brown precipitate was then collected by filtration, washed with Et₂O, and dissolved in water (20 ml). Addition of excess KBr gave a green precipitate that was collected by filtration, washed with H₂O and Et₂O, and dried under vacuum (772 mg, 1.53 mmol). Yield: 73%. ¹H NMR (300 MHz, acetone-d₆): δ 4.17 (m, 4H), 2.82 (s, 6H), 2.77 (s, 6H), 2.64 (m, 2H).

[Ni((DO)₂BF₂)pn](ClO₄).

Ni(OAc)·4H₂O (207 mg, 0.83 mmol) and BF₃·OEt₂ (1.05 ml, 8.03 mmol) were added to a suspension of (DOH)(DOH)pn (200 mg, 0.83 mmol) in Et₂O (20 ml). The mixture was stirred for 24 h at room temperature under argon. The precipitate was then collected by filtration, washed with Et₂O, and dissolved in water (20 ml). Addition of excess NaClO₄ gave an orange precipitate that was collected by filtration, washed with H₂O (10 ml) and Et₂O, and dried under vacuum (185 mg, 0.42 mmol). Yield: 50%. ¹H NMR (300 MHz, acetone-d₆): δ 3.54 (t, 4H, J = 5.1), 2, 50 (s, 6H), 2.30 (s, 6H), 2.22 (m, 2H).

(MOH)(MOH)pn.

Anti-pyruvic aldehyde-1-oxime (960 mg, 11 mmol) and 1,3-diaminopropane (460 μl, 5.5 mmol) were stirred in CH₂Cl₂ (30 ml) at 40 °C in the presence of molecular sieves (4 Å). After 18 h, the molecular sieves were removed by filtration. The desired product (720 mg, 3.4 mmol, 43%) was then precipitated by addition of Et₂O (50 ml), yielding a white powder collected by filtration and dried under vacuum. ¹H NMR (300 MHz, DMSO-d₆): δ = 11.57 (broad, s, 2H), 7.54 (s, 2H), 3.44 (t, 4H), 1.96 (s, 6H), 1.92 ppm (m, 2H). ¹³C NMR (75 MHz, DMSO-d₆): δ = 163.61, 152.24, 49.57, 31.67, 13.58 ppm.

[Co(MO)(MOH)pnCl₂].

CoCl₂·6H₂O (224 mg, 0.94 mmol) was added to a suspension of (MOH)(MOH)pn (200 mg, 0.94 mmol) in acetone (20 ml). The mixture was stirred for 3 h at room temperature and bubbled with air for 1 h. The green precipitate was then collected by filtration, washed with Et₂O, and dried over vacuum, which gave 160 mg (0.47 mmol). Yield: 50%. ¹H NMR (300 MHz, DMSO-d₆): δ 18.85 (s, 1H, H_ox), 8.12 (s, 2H), 3.95 (t, 4H), 2, 61 (s, 6H), 2.35 (m, 2H). ¹³C NMR (75 MHz, DMSO-d₆): δ 173.8, 148.8, 49.5, 27.2, 18.4. IR (KBr, cm⁻¹): 3441(broad), 1614, 1520.

[Ni(MO)(MOH)pnCl].

NiCl₂·6H₂O (112 mg, 0.47 mmol) was added to a suspension of (MOH)(MOH)pn (100 mg, 0.47 mmol) in EtOH (25 ml). The mixture was stirred for 3 h at room temperature. The brown precipitate was then collected by filtration, washed with Et₂O, and dried over vacuum (145 mg, 0.43 mmol). Yield: 91%. ¹H NMR (300 MHz, D₂O): δ 8.11 (broad, s, 2H), 4.01 (bs, 4H), 2.00 (s, 6H), 1.83 (bs, 2H). IR (KBr, cm⁻¹): 3509, 3412, 2986, 1504, 1438.

Mechanistic Study.

The blue [Co(DO)(DOH)pn(PPh₃)] complex was dissolved in CH₃CN (0.0123 mol·L⁻¹) under argon. An excess of p-cyanoanilinium bromide (10 equiv) was added and the reaction monitored using UV-visible spectroscopy.

The cobalt(II) complex formed within 30 s was synthesized independently in the glove box by reacting (DOH)(DOH)pn (200 mg, 0.88 mmol), CoBr₂·H₂O (288 mg, 0.88 mmol), and PPh₃ (232 mg, 0.88 mmol) in acetone (20 ml). The solution was evaporated, and CH₃CN was added. The resulting solution displays an absorption maximum at 470 nm. An excess of p-cyanoanilinium bromide (10 mmol) was added, and the reaction was monitored using UV-visible spectroscopy.

Crystallographic data (Table S2) and additional electrocatalytic data are given in the Supporting Information.