Transition Metal Pyrithione Complexes (Ni, Mn, Fe, and Co) as Electrocatalysts for Proton Reduction of Acetic Acid

Abstract

A series of mononuclear first-row transition metal pyrithione M(pyr)_n complexes (M = Ni(II), Mn(II), n = 2; M = Co(III), Fe(III), n = 3) have been prepared from the reaction of the corresponding metal salt with the sodium salt of pyrithione. Using cyclic voltammetry, the complexes have been shown to behave as proton reduction electrocatalysts albeit with varying efficiencies in the presence of acetic acid as the proton source in acetonitrile. The nickel complex displays the optimal overall catalytic performance with an overpotential of 0.44 V. An ECEC mechanism is suggested for the nickel-catalyzed system based on the experimental data and supported by density functional theory calculations.

Introduction

Current energy resources are heavily reliant on nonrenewable fossil fuels which also pollute the environment.¹ An eco-friendly alternative is thus widely sought after to ensure sustainable development for the future. Hydrogen (H₂) is regarded as a viable option not only for its highest gravimetric energy density among all fuels but also for its environmental friendliness as its only combustion product is water.^2,3

One of the most promising strategies for the efficient and clean production of H₂ is electrochemical water splitting composed of two half-reactions: hydrogen evolution reaction (HER) and oxygen evolution reaction (OER).⁴⁻⁶ For the former process, the most efficient electrocatalysts are based on noble and expensive metals such as platinum.⁷⁻⁹ Earth-abundant transition-metal-based HER or proton reduction electrocatalysts have also been investigated as viable alternatives due to their economical practicality.^1,10−15 In particular, many synthetic metal complexes with various ligand structures modeled after the active sites in [Fe–Fe] and [Ni–Fe] hydrogenases have been studied and tested as catalysts to varying degrees of efficiency.^13,16−27 Thiolate ligands have been of particular interest as they mimic the ligation of cysteinyl residues in these hydrogenases.^28,29

In our laboratory, we are interested in working with or designing air-stable and economical metal complexes as proton reduction electrocatalysts modeled after on the active-site structure of the [Ni–Fe] hydrogenase. During a search for metal complexes bearing thiolate or sulfur-centered ligands, we came across first-row transition metal complexes where the mononuclear metal center is coordinated to two or three pyrithione (2-mercaptopyridine N-oxide) ligands consisting of both sulfur and oxygen donor atoms (Scheme 1).³⁰ To the best of our knowledge, this series of complexes has not been compared and contrasted as electrocatalysts for proton reduction. In this work, we are delighted to report that some of the complexes show promise as robust and efficient proton reduction catalysts in the presence of acetic acid as the proton source. The electrocatalytic performance was evaluated through the use of cyclic voltammetry (CV) while density functional theory (DFT) calculations were utilized to lend support to our proposed proton reduction mechanism catalyzed by the nickel complex.

Scheme 1. Four Transition-Metal Pyrithione Complexes Used in This Work.

Note that 1 and 3 may contain both cis and trans isomers while 2 and 4 may contain the mer- and fac-isomers.

Results and Discussion

The syntheses of complexes 1–4 were first carried out based on modified literature methods.^30,31 An aqueous solution containing the metal acetate or metal chloride was added dropwise to another aqueous solution containing the pyrithione ligand, giving rise to a colored precipitate, which was later isolated and purified. The products were characterized whenever possible, using electrospray ionization mass spectroscopy (ESI-MS) or electron ionization mass spectroscopy (EI-MS), ¹H nuclear magnetic resonance spectroscopy (NMR), UV–vis absorption spectroscopy as well as elemental analyses (EA) (see Supporting Information Figures S1–S11). However, NMR spectra of 3 and 4 were not obtained due to the paramagnetic nature of the complexes. The data obtained for all four complexes were in excellent agreement with the literature data. For 1 and 3, the metal centers are in the +2 oxidation state and coordinated by two pyrithiones, while in 2 and 4, the Co³⁺ and Fe³⁺ centers were coordinated by three pyrithiones, respectively (Scheme 1).

Unfortunately, the characterization data are unable to determine the isomeric forms of the complexes in an unambiguous manner. A previously determined X-ray crystal structure of the nickel complex shows a cis-configuration where the two sulfur atoms are adjacent to each other.³⁰ We have also assumed that the cis-isomer is dominant in our experiments although the presence of the trans-isomer could not be ruled out. Similarly, X-ray studies have shown 2 to be a fac-isomer, but both mer- and fac-isomers for 2 and 4 may still be present.³² We believe that even if the isomeric forms are present, they differ very little in energy and thus will most likely show very similar electrochemical properties.

Cyclic voltammograms (CV) of the four metal complexes were first performed in the absence of acid (Figure 1 and Table 1). As all of the recorded potentials at the glassy carbon (GC) working electrode were calibrated against the Fc⁺/Fc couple internal standard (set to 0.0 V) in acetonitrile, all potential values will henceforth be quoted with respect to this couple. A reversible molecular peak was observed in the CV scan at −1.68, −0.38, and −0.86 V for 1, 3, and 4, respectively, while a quasi-reversible molecular peak at −0.96 V_ave was recorded for 2 (Figure 1). Consistent with a previous literature report, the three reversible peaks were assigned to the Ni²⁺/Ni⁺, Mn²⁺/Mn⁺, and Fe³⁺/Fe²⁺ redox couples.³⁰ In contrast, the reduction peak of complex 2 at around −1.2 V appeared to be irreversible with an accompanying nearby oxidation peak at −0.7 V. As suggested by a previous study,³⁰ the two peaks could be assigned to the Co³⁺/Co²⁺ couple. For 1, 3, and 4, the reversibility of their peaks suggested that the structure of the generated anionic species remains intact and can be reoxidized easily to the starting complex. The reversibility is also indicative of the stability of the reductive species on our CV timescale.

Figure 1.

Molecular reduction peaks detected in the CV scan of (a) 1 (1 mM), (b) 2 (0.1 mM), (c) 3 (1 mM), and (d) 4 (0.1 mM) in acid-free acetonitrile with NBu₄PF₆ (0.1 M) and Fc (1 mM) scanned at 0.1 V s^–1.

Table 1. Electrochemical Parameters of the Synthesized Complexes for Proton Reduction Catalysis in Acetonitrile with Acetic Acid as the Proton Source^a.

complex	Ecat/2 (V)	overpotential (V)b (20 acid equiv)	molecular reduction peaks (V) and their redox assignmentsc
1	–1.84	0.44	E1/2 (+ΔE) = −1.68(0.07) (Ni2+/Ni+)
2	–1.95	0.55	Epc = −1.2 V; Epa = −0.7 V (Co3+/Co2+)
3	–2.37	0.97	E1/2 (+ΔE) = −0.38(0.07) (Mn2+/Mn+)
4	–2.06	0.66	E1/2 (+ΔE) = −0.86(0.03) (Fe3+/Fe2+)

^aAll reported potentials were calibrated with respect to the Fc⁺/Fc internal standard in acetonitrile.

^bCalculations made where E_HA° = −1.40 [pK_a (CH₃COOH) = 23.51 in acetonitrile].

^cΔE = peak-to-peak separation of the M⁽ⁿ⁺¹⁾⁺/Mⁿ⁺ couple except for complex 2; ΔE for Fc⁺/Fc = 0.04 V.

We have utilized the Randles–Sevcik equation, which can be applied to the reversible peak of 1, 3, or 4, to determine whether the metal complex in question is freely diffusing in solution or is adsorbed on the electrode surface (Figure S12). As the peak current turned out to vary linearly with the square root of the scan rate, this indicates the metal complex is a diffusing homogeneous species in solution.

As the solubilities of 2 and 4 in acetonitrile are poorer than those of 1 and 3, lower concentrations of 0.1 mM were used for 2 and 4 in the CV scans. As a result, their observed peaks appeared weaker. The difference in the diffusion coefficients of the four complexes may also play a minor factor in determining the peak intensities.

The CV scan of complex 1 was then recorded in the presence of acetic acid. A broad signal which emerged around −2 V for 1 and continued to grow with sequential addition of the acid has been identified as the proton reduction catalytic wave (Figure 2a). The appearance of a peak current indicates a competition between consumption of the acid substrate and diffusion of a new substrate to the electrode.³³

Figure 2.

CV scans performed in acetonitrile solution containing (a) 1 (1 mM), (b) 2 (0.1 mM), (c) 3 (1 mM), and (d) 4 (0.1 mM) with NBu₄PF₆ (0.1 M) and CH₃COOH (up to 250 equiv) at 100 mV s^–1.

We have also ascertained that the contribution from the glassy carbon electrode amounted to only about 15% of the total current toward proton reduction (Figure S13).

The peak current is also found to be linearly proportional to the acid concentration. As the catalytic wave emerged at the vicinity of the Ni(II)/Ni(I) couple, it suggested that the 1^– anion species could have acted as the initiator in the catalytic cycle. The Ni(II)/Ni(I) couple also remained at the same potential upon acid addition, indicating that complex 1 was not protonated prior to electroreduction.

The overpotential (OP) for catalysis can be determined using the following equation^34,35

where E_cat/2 is the catalytic half-wave potential defined as the point at which the homogeneous catalytic wave reaches half of its maximum current and E_HA° = E_H⁺° – pK_a (HA).

Based on the values of E_H⁺° (−0.028 V) and pK_a of acetic acid in acetonitrile (23.51), the thermodynamic potential of E_HA° has been calculated to be −1.40 V.³⁶ As the overpotential is acid-dependent, we have chosen to determine the overpotentials of the complexes at a fixed value of 20 equiv of acid, which is also a typical value used for many previous electrocatalysis studies.¹⁰⁻¹⁵ At 20 equiv of acetic acid for which E_cat/2 is −1.84 V, the overpotential for 1 is 0.44 V. Homoconjugation effects due to appreciable acid concentrations could increase the overpotential by 0.1 V at 20 acid equivalent (refer to a sample calculation in the Supporting Information).³⁵

We have attempted to determine the k_obs for proton reduction catalysis for the metal complexes by varying the scan rate and substrate concentration to attain an S-shape or plateau current.^34,37 Although a near-plateau current was seen at scan rates higher than 1.0 V s^–1 for complex 1 (Figure S14), the same features could not be achieved for the other complexes. Hence, the extraction of k_obs values was not successful under our experimental conditions.

A rinse test was carried out to demonstrate the homogeneity of the catalytic process whereby the removal of the complexes from solution caused the respective catalytic waves to disappear (Figures S15–S18). However, we note that a weakly adsorbed active species can also account for this observation. A steady current recorded during a 5 h controlled-potential electrolysis (CPE) run at −2.0 V containing 100 equiv of acid provided support of the robustness of 1 as an electrocatalyst (Figure S19). Consistent bubbling observed at the tip of the working electrode was analyzed by mass spectrometry to comprise hydrogen gas. An average faradaic yield of 82 ± 15% was determined for the CPE experiment, which indicates that most of the electrons have been used to carry out hydrogen evolution.

Catalytic waves were observed at different E_cat/2 values for complexes 2, 3, and 4 (Figure 2b–d). Their respective overpotential has also been determined along the same lines as for 1. While complexes 2, 3, and 4 possess higher overpotentials compared to 1, the catalytic wave due to 3 appeared at the most negative potential range and very likely overlaps with the catalytic wave caused by the GC electrode (Figure S13).

Upon acid addition, the molecular peaks of 2–4 were also found to shift to more positive potentials (Figure S20). The reversible peak of complex 4 even broadened to become a quasi-reversible peak. Unlike 1, the catalytic waves for these three complexes did not occur immediately after their respective molecular peak, but at a reasonably large potential difference of 0.6 V for 2, 1.3 V for 3, and 0.8 V for 4 respectively. Thus, complexes 2–4 are likely to carry out proton reduction via a different mechanistic route from 1 by first undergoing protonation before reduction of the protonated species. The large potential gap also suggests that the reduced species is not sufficiently basic to undergo efficient protonation hence requiring further electroreduction or a PCET (proton-coupled electron transfer) process to take place. A significant overpotential is thus incurred due to these processes.

For the six-coordinated complexes 2 and 4, there could also be a loss in reactivity due to the unavailability of vacant sites. It is likely that for protonation to occur, at least one pyrithione ligand has to be detached. This would certainly incur some energy input and further increase the overpotential for proton reduction. Based on the electrochemical routes taken by the four complexes and their overpotential and k_obs parameters, complex 1 appears to show the most promise as an efficient electrocatalyst while also being more amenable to computational studies.

Complex 1 is considered to have SO^–-type bidentate ligands since pyrithione coordinates to the metal center via the S and adjacent O atoms. Many other varieties such as SN-, SP-, and SS (dithiolene)-type ligands have been used to prepare square-planar Ni(II) complexes as proton reduction electrocatalysts.³⁸ Most of these ligands are expected to exhibit redox non-innocent behavior, which may facilitate the electrochemical reduction of the complex or its protonation.

While the discussion of all aspects of such nickel catalysts is outside the scope of this work, we note a very interesting common electrochemical feature that many of these catalysts possess. For example, the proton reduction catalytic wave is observed immediately after a reversible one-electron reduction of bis-benzenedithiolato nickel(II) complexes with acetic acid as the proton source.³⁹ The potential at which the processes take place depends on the substituents on the ligand. Similarly, proton reduction takes place immediately after the bis-diphenylphosphinobenzenethiolato nickel(II) complex (SP-type) undergoes a one-electron reduction.⁴⁰ The same occurrence was also seen for square-planar nickel(II) complexes carrying thiosemicarbazone ligands (SN-type).⁴¹⁻⁴³ In this work, complex 1 also behaves in the same way albeit with the catalytic wave observed at relatively more positive potentials leading to a lower overpotential. Hence, there is still much promise in continuing to explore and tune the electrochemical properties of nickel(II) complex bearing SX-type ligands (X = heteroatom) to enhance the proton reduction electrocatalysis.

We have also used density functional calculations within the Gaussian 16 suite of programs to characterize the structures and free energies of the reaction intermediates generated from complex 1.⁴⁴⁻⁴⁶ Optimized structures and thermal free-energy corrections were obtained at the B3LYP/6-311+G* level using the SMD (solvation model based on density) model with acetonitrile as the solvent. The computed nickel complex structures and coordinates can be found in Table S1 in the Supporting Information under Computational Chemistry. The protonation energies were computed with respect to the CH₃COOH/CH₃COO^– couple. The standard redox potential E° was calculated using the relation E° = −ΔG°/nF – E°_ref where E°_ref is the absolute reduction potential of the ferrocene (Fc⁺/Fc) couple calculated at the same level of theory.

For benchmark purposes, the absolute ferrocene reduction potential has been determined to be 5.01 V in acetonitrile, which is in very good agreement with the experimental value of 4.98 V.⁴⁴ Similarly, a potential of −1.44 V calculated for the CH₃COOH + e → 1/2H₂ + CH₃COO^– redox reaction in acetonitrile is close to the experimental thermodynamic potential E_HA° of −1.40 V. More details of the calculations can also be found in the Supporting Information File under Computational Chemistry.

The structure of 1 was first optimized and compared to the previous experimental structure determined by single-crystal X-ray diffraction.³⁰ The calculated bond distances and angles agree very well (to within 0.1 Å for bond distance and 2° for bond angle) with the experimental values (see Table S2 in the Supporting Information).

In addition, we have also carried out similar calculations for the isomer of 1 (labeled as 1_iso) where the ligand O (or S) atoms are at trans position to each other. Complex 1 is indeed calculated to be slightly lower in energy by about 6 kJ mol^–1, which lends further support to 1 being the dominant species (see Tables 3, S1, and S3 in the Supporting Information). Henceforth, we will only describe the electrochemical behavior of 1 and the various species generated upon its reduction.

Table 3. Relative Gibbs Energy G Calculated for the Optimized Metal Complexes at the B3LYP/6-311+G* Level Incorporating the SMD Model (Acetonitrile)^a.

	metal center	multiplicity	rel. G (kJ mol–1)
1-cis (or 1)	Ni(II)	singlet	0
1-trans (or 1_iso)	Ni(II)	singlet	+6.0
2-mer	Co(III)	singlet	0
2-fac	Co(III)	singlet	–7.4
3-cis	Mn(II)	sextet	0
3-trans	Mn(II)	sextet	–6.4
4-mer	Fe(III)	sextet	0
4-fac	Fe(III)	sextet	–8.1

^aThe ground-state spin multiplicity is indicated for each complex and its isomer. Table S3 contains the computed Gibbs energy of each complex.

From the cyclic voltammetry data, the proton reduction catalytic wave is triggered upon electrochemical (E) reduction of 1 to 1^– at E_1/2 = −1.68 V. The formation of 1^– is thus regarded as the first step in the mechanism (Table 2 and Figure 3). Our calculations returned a value of −1.65 V for this reduction potential, which again is in very close agreement with the experimental value. The optimized structure of 1^– shows that the nickel complex retains its square-planar structure, which is also consistent with the reversibility nature of the electrochemical reduction step. The Mulliken spin density ρ of 0.917 at the Ni center of 1^– further supports the assignment of the reduction process to the Ni(II)/Ni(I) couple.

Table 2. Proposed 1-Catalyzed Proton Reduction Pathways Using Acetic Acid as the Proton Source^a.

	reaction pathways	ΔG (kJ mol–1)
1	1 + e ⇌ 1–	–324.7 (−1.65 V)
2	1– + (H+ + e) → 1H– (triplet)	–292.0
3	1H– + H+ → 1 + H2	–72.4
4	1– + (H+ + e) → 1H– (singlet)	–265.9
5	1– + H+ → 1H_Ni	+95.3
6	1– + H+ → 1H_S	+99.7
7	1– + H+ → 1H_O	+79.5
	PCET step 2 is preferred over PCET step 4

^aAll molecular structures including 1 and the resulting intermediates were optimized, and their Gibbs energy G and the Gibbs energy change ΔG for each step were calculated at the B3LYP/6-311+g* level in Gaussian 16 incorporating the SMD model (acetonitrile). The computed structures and coordinates can be found in the Supporting Information.

Figure 3.

Computed proton reduction mechanism (ΔG, kJ mol^–1 in parentheses) catalyzed by 1. These optimized structures can also be found in the Supporting Information.

The subsequent step taken by 1^– is likely to involve protonation or a proton-coupled electron transfer (PCET) process. An optimized square-pyramidal geometry of 1H_Ni, the protonated form of 1^– was obtained by placing the proton on the nickel center. The Mulliken spin density associated with the nickel remains close to 1 (ρNi = 0.967). However, the computed protonation step turns out to be endergonic (ΔG = +95.3 kJ mol^–1), rendering the process unlikely to take place at room temperature.

We noted that the protonation could also take place at one of the sulfur or oxygen atoms of the ligand. Hence, optimization was also carried out for these two possible isomers of 1H_Ni, which were labeled as 1H_S and 1H_O. The optimized structures for all three isomers turned out to be similar, whereby the H atom is preferentially located in a perpendicular direction to the molecular plane. From the calculations, the Gibbs energy of 1H_O is about 15.8 kJ mol^–1 lower than that of 1H_Ni, which in turn is 4.4 kJ mol^–1 lower than the 1H_S energy (see Table 2). However, the protonation of 1^– to 1H_O is still too endergonic (ΔG = +79.5 kJ mol^–1) to proceed efficiently.

PCET is then considered as the alternative path in which the structure of 1H^– was considered either as a singlet or triplet state. The optimized triplet state structure of 1H^– reveals the H atom located at the axial position above the Ni center forming a square-pyramidal geometry. The geometry of the two pyrithione ligands is affected in the same way and hence remains very similar to each other. In contrast, the singlet state however retains the square-planar structure with the H atom replacing the sulfur atom of one of the pyrithione ligands.

Contrary to the protonation-only step, the PCET process to form either state is found to be exergonic with the triplet state being favored (ΔG = −292 kJ mol^–1 ≡ −1.99 V) over the singlet state (ΔG = −265.9 kJ mol^–1 ≡ −2.28 V). The gap between the triplet-state PCET process and the electroreduction of 1 to 1^– step is estimated to be about 0.34 V (1.99 – 1.65 V), which is larger than the corresponding experimental difference of 0.16 V (|E_cat/2 – E_1/2| = 1.84 – 1.68 V). Here, we have drawn a parallel between the PCET potential and E_cat/2 assuming that the PCET is rate-determining. The discrepancy between the two values (0.34 vs 0.16 V) is probably due to the difference between the bulk H⁺ concentration (provided by the pK_a value) used in our calculations and the local H⁺ concentration at the cathode layer where electrochemical processes such as PCET would take place.

In the final step, hydrogen is released when an incoming proton combines with the hydride of 1H^– and regenerates 1. Indeed, this chemical (C) step is determined to be exergonic (ΔG = −72.4 kJ mol^–1).

Overall, an ECEC proton reduction mechanism is believed to be in operation for complex 1 where a PCET process comprising the middle CE steps drives the proton reduction forward. The major contribution to the rate-limiting step most likely comes from the PCET step as it occurs at a more negative potential than the first electrochemical reduction step.

As mentioned earlier, we were unable to determine experimentally which geometric isomers of 2, 3, and 4 have been synthesized. Hence, their molecular structures were optimized, and then the Gibbs energies were computed to locate the lower-energy isomers (see Tables 3 and S3 in the Supporting Information). Similar to the Ni(II) system, two possibilities, cis or trans, exist for Mn(II) complex 3 in their sextet ground state. The mer- and fac-isomers were the two choices for the singlet Co(III) and sextet Fe(III) complexes.³⁰

From the computational data in Table 3, the energy difference between the isomers ranges from 6 to 8.1 kJ mol^–1. Apart from the cis-Ni(II) complex, the lower-energy isomers for the other three systems turn out to be trans-Mn(II), fac-Co(III), and fac-Fe(III) complexes, respectively. A fac-Co(III) isomeric complex was previously found to crystallize out of acetonitrile solution.³² While it is tempting to directly assign these lower-energy isomers as the complexes that were investigated here, we note that some ambiguity would still exist because the energy difference between the isomers is small. While our calculations and experiments were conducted in acetonitrile, it is also possible for the higher-energy isomer to become dominant in other solvent systems. However, the small energy difference suggests that very similar electrochemical properties will be exhibited for either isomer.

Conclusions

Four mononuclear first-row transition metal pyrithione (mercaptopyridine N-oxide) complexes comprising Ni(II), Mn(II), Fe(III), or Co(III) center have been prepared from the reaction of their metal salt with the sodium salt of pyrithione. All four complexes have been found to act as proton reduction electrocatalysts when acetic acid was used as the proton source in acetonitrile. The nickel pyrithione complex possesses the best overall performance with an overpotential of 0.44 V. In contrast, the manganese pyrithione complex is a poor catalyst with an overpotential close to 1 V. Density functional calculations carried out on the nickel-catalyzed system also supported the experimental data where an ECEC mechanism is in operation whereby proton reduction catalysis was triggered by the electrochemical reduction of nickel pyrithione followed by a PCET step and completed with another protonation step to regenerate 1.

Experimental Section

Nickel(II) acetate tetrahydrate (98%), cobalt(II) acetate tetrahydrate (99%), copper(II) dichloride dihydrate (99%), manganese(II) acetate tetrahydrate (99%), iron(II) chloride tetrahydrate (99%), and 2-mercaptopyridine N-oxide sodium salt (96%) were all purchased from Sigma-Aldrich and used as received.

Acetonitrile (CH₃CN; Fulltime Chemical, HPLC/Spectro grade, >99%) was dried with 4 Å molecular sieves (Alfa Aesar) before usage. Tetrabutylammonium hexafluorophosphate (NBu₄PF₆; Tokyo Chemical Industry Co., Ltd., >98.0%) and glacial acetic acid (CH₃COOH; Merck, 100%) were used as received.

Electrospray ionization (ESI) mass spectrometry was done using a Finnigan MAT LCQ spectrometer while high-resolution mass spectroscopy (HRMS) was done using a Bruker micrOTOF-QII spectrometer. Elemental analyses (EAL) of carbon, hydrogen, nitrogen, and nickel were done using an ElemenVario Micro Cube. NMR spectra were performed on Bruker AV III 400HD (BBFO probe) spectrometer. ¹H NMR spectra chemical shifts were reported in δ ppm relative to DMSO-d₆ (δ = 2.05 ppm). Multiplicities were recorded as: d (doublet) and t (triplet) respectively. The number of protons (n) for a given resonance was indicated by nH while coupling constants were reported as J value in Hertz (Hz).

Synthesis of (C₅H₄NOS)₂Ni (1)

Nickel(II) acetate tetrahydrate (1 mmol, 248.8 mg) was slowly added to 2-mercaptopyridine N-oxide sodium salt (2 mmol, 298.3 mg) in deionized water (10 mL) and left to stir for an hour. The brown precipitate was then collected via vacuum filtration before being dried under high vacuum. Yield: 77.2%. ESI-MS: expected m/z = (+)309.94; found = (+)309.9 (M⁺). ¹H NMR (400 MHz, DMSO-d₆): δ 8.75 (d, J = 6.4 Hz, 2H), 7.61 (d, J = 8.4 Hz, 2H), 7.43 (t, J = 7.6 Hz, 2H), 7.17 (t, J = 6.4 Hz, 2H). UV–vis in DMF, λ, nm (ε, M^–1 cm^–1): 322 (13 000), 470 (95), and 523 (60). EA calcd: C, 38.62; H, 2.59. Found: C, 39.22; H, 2.74.

Synthesis of (C₅H₄NOS)₃Co (2)

Cobalt(II) acetate tetrahydrate (1 mmol, 177.0 mg) was slowly added to 2-mercaptopyridine N-oxide sodium salt (3 mmol, 447.5 mg) in deionized water (10 mL) and left to stir for an hour. The dark green precipitate was then collected via vacuum filtration before being dried under high vacuum. Yield: 81.9%. ESI-MS: expected m/z = (+)436.94; found = (+)436.77 (M⁺). ¹H NMR (400 MHz, DMSO-d₆): δ 8.52 (d, J = 7.2 Hz, 2H), 7.48 (d, J = 7.6 Hz, 2H), 7.29 (t, J = 7.0 Hz, 2H), 7.07 (t, J = 6.2 Hz, 2H). UV–vis in DMF, λ, nm (ε, M^–1 cm^–1): 418 (890) and 645 (410). EA calcd: C, 41.19; H, 2.77. EA calcd: C, 41.19; H, 2.77. Found: C, 40.80; H, 2.41.

Synthesis of (C₅H₄NOS)₂Mn (3)

Complex 3 was synthesized and purified in a similar fashion as complex 1 except manganese(II) acetate tetrahydrate (1 mmol, 245.1 mg) was used instead of nickel(II) acetate tetrahydrate. The final product is a green solid. Yield: 86.9%. ESI-MS: expected m/z = (+)306.94; found = (+)306.96 (M⁺). UV–vis in DMF, λ, nm (ε, M^–1 cm^–1): 291 (16 000) and 337 (4800). EA calcd: C, 39.09; H, 2.62. Found: C, 38.60; H, 3.06.

Synthesis of (C₅H₄NOS)₃Fe (4)

Complex 4 was synthesized and purified in a similar fashion as complex 2 except iron(III) chloride hexahydrate (1 mmol, 198.8 mg) was used instead of cobalt(II) acetate tetrahydrate. The final product is a black solid. Yield: 83.7%. ESI-MS: expected m/z = (+)456.93; found = (+)456.79 ([M + Na]⁺). UV–vis in DMF, λ, nm (ε, M^–1 cm^–1): 318 (23 000), 515 (6300), and 570 (5900). EA calcd: C, 41.48; H, 2.79. Found: C, 41.02; H, 2.56.

Electrochemical Methods

• Electrochemical studies were conducted with the Princeton Applied Research Potentiostat Model 263A using a three-electrode system; 3-mm-diameter planar glassy carbon (GC) disks were used as both the working and counter electrodes, alongside with a silver wire as the pseudo-reference electrode, immersed in a cell fitted with a porous glass Vycor tip containing Bu₄NPF₆ in acetonitrile; before every scan, the 0.1 M electrolyte Bu₄NPF₆ solution with 1 mM (or 0.1 mM) electrocatalyst in acetonitrile (7 mL) was purged with nitrogen gas for 5 min continuously; GC electrodes were also polished using a diamond polishing pad (PK-3) manufactured by ALC Co., Ltd.; all scans were calibrated with regard to the standard ferrocene Fc⁺/Fc couple (set to 0.00 V) and scan rate of 0.1 V s^–1;

• for controlled-potential electrolysis, a 1 mM solution of the nickel complex with 100 mM CH₃COOH in 0.1 M Bu₄NPF₆ in acetonitrile run at the respective catalytic peak potential (vs Fc⁺/Fc) in a sealed inert electrochemical cell; gaseous content within the cell was removed using a syringe of 1.0 cm³ volume before being injected into a mass spectrometer tuned to m/z = 2, corresponding to the detection of H₂; signal intensity is calibrated with respect to the signal produced by the injection of a known pressure of H₂ gas obtained from a H₂ cylinder (Soxal, 99.99%); and Faraday yield (83 ± 10%) was then obtained from the [H₂ detected]/[electrons used for electrolysis] ratio.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.3c00412.

• NMR, ESI, and UV spectra for 1–4; electrochemical studies (Randles–Sevcik, scan rate, and shape of catalytic wave, GC electrode reduction, rinse test, bulk electrolysis, peak shifts upon protonation, homoconjugation); and computational chemistry (thermodynamic parameters, molecular structures, and coordinates) (PDF)

The authors declare no competing financial interest.