Anion controlled selectivity in oxygen reduction catalysed by a dinuclear cobalt N,O-Schiff base complex

Abstract

A new dinuclear cobalt complex selectively catalyses 4e⁻ reduction of O₂ to water in methanol containing acetic acid. Its TOF of 0.031 s⁻¹ and overpotential of 690 mV outperform the few previous Co N,O-chelate based catalysts for the 4e⁻ ORR. Replacing acetic acid with NH₄PF₆ as proton source induces a complete and unprecedented switch to the H₂O₂ producing 2e⁻ pathway. Mechanistic studies suggest a peroxo intermediate for both pathways, with acetate coordination/decoordination determining the destination of a key electron transfer.

---

The oxygen reduction reaction (ORR) is important to biological respiration,¹ fuel cell technology,² metal–air batteries³ and aerobic oxidations.⁴ The ORR in acidic media can proceed via two pathways; the two-electron (2e⁻) pathway to generate hydrogen peroxide, or the four-electron (4e⁻) pathway to water (eqn (1) and eqn (2)). As it is more exergonic and does not form the corrosive and oxidising H₂O₂, the 4e⁻ pathway is generally preferred in the context of fuel cells. However, there is increasing interest in electrochemical synthesis of H₂O₂ via 2e⁻ ORR as an alternative to the current industrial anthraquinone process,⁵ and due to its importance to lithium–air batteries³ and potential as an energy carrier.^5c,6

O₂ + 4H⁺ + 4e⁻ → 2H₂O E^θ = 1.23 V (1)

O₂ + 2H⁺ + 2e⁻ → H₂O₂ E^θ = 0.68 V (2)

Molecular catalysis of the ORR in acidic organic media has been studied in detail for metalloporphyrins and phthalocyanines,⁷ and Schiff base and oxime complexes,⁸ but biasing the catalytic system to the 4e⁻ rather than the 2e⁻ pathway is challenging. Methods to increase selectivity for the 4e⁻ pathway include: using a proton–electron transport mediator (e.g. hydroquinone, quinol) to facilitate O–O bond breakage,^8b,9 adding proton relays/pendant bases to facilitate protonation of bound oxygen species,^8d,10 or using Brønsted scaling relationships to disfavour H₂O₂ production.¹¹ Using dinuclear complexes, such as “hangman” porphyrins, to favour the 4e⁻ pathway has also shown success.^12–16 Yet, for Co(II) salen-type complexes, which generally only catalyse the 2e⁻ pathway, dinuclear approaches to the ORR are to our knowledge unexplored.

Herein, we show that a novel, easily accessible dinuclear Co(II) complex with a salen-like N,O chelating unit (2, Fig. 1) catalyses the 4e⁻ ORR with high (>90%) selectivity, and better turnover frequency (TOF) and overpotential metrics than comparable mononuclear systems. Moreover, changing the anions in solution (from AcO⁻ to non-coordinating PF6⁻) enables a complete switch to the 2e⁻ pathway, independent of the pK_a effects that control selectivity in Co(TPP) catalysts.¹¹ While anion/co-ligand effects are known in catalysis,¹⁷ they are not previously reported as a way to control the pathway in the ORR.

Fig. 1 Structure of the dinuclear Schiff base ORR catalyst [Co₂(MeBSIP)(OAc)(H₂O)₂] (2).

Compound 2, [Co₂(MeBSIP)(OAc)(H₂O)₂] (H₃MeBSIP = 2,2′-[(2-hydroxy-1,3-propanediyl)bis(oxy-2-benzylidineamino)]di-[4-methylphenol], Fig. 1) was synthesised in two steps (see ESI† for full details) via dialdehyde precursor 2,2′-[(2-hydroxy-1,3-propanediyl)bis(oxy)]bis-benzaldehyde (1), which was reacted with cobalt(II) acetate and 2-amino-4-methylphenol to produce 2 by a templated imine condensation. Compound 2 has been characterised by ESI† mass spectrometry, CHN elemental analysis, and FTIR (Fig. S1, ESI†). CHN is consistent with the presence of two coordinated H₂O, IR indicates an imine stretch and also a bridging acetate coordination mode (ν = 1572 cm⁻¹), while MS detects a doubly charged ion at m/z = 625.066 consistent with loss of acetate and water ligands and formation of the dimer, [Co₄(MeBSIP)₄]²⁺. X-ray quality crystals of 2 could not be obtained, but two structures based on the {Co₂MeBSIP} unit have been solved (Fig. 2; Table S1 and Fig. S2, S3, ESI†). Trinuclear [Co₃(MeBSIP)(OAc)₃(MeOH)] (3) forms in the presence of excess Co(OAc)₂: the third Co occupies a coordination pocket defined by the two phenoxo groups, with its coordination sphere completed by acetato and methanol ligands. Defect cubane [Co₄(MeBSIP)₂(MeOH)₂](BF₄)₂ (4) was isolated by adding NaBF₄ to the 2 reaction mixture, and has alkoxo-bridged [Co₂(MeBSIP)] units linked by formation of μ₂ and μ₃ phenolato bridges. Compounds 3 and 4 support the proposed structure of 2, by showing {Co₂} units coordinated to MeBSIP through imine, ether and phenolate groups and a bridging alkoxo.

Fig. 2 X-ray crystal structures of 3, and the [Co₄(MeBSIP)₂(MeOH)₂]²⁺ complex cation in 4. Thermal ellipsoids are at the 50% probability level, C is grey; N, blue; O, red; Co, purple, H atoms are white spheres with arbitrary radii. Coordinate bonds are shown in yellow.

The strong bridging acetate IR signal of 2 supports (Fig. S3, ESI†) existence of 2 as a {Co₂} rather than {Co₄} structure, as the [Co₄(MeBSIP)₂]²⁺ unit in 4 cannot accommodate bridging acetate ligands. To further support the structure of 2, powder XPS measurements were obtained (Fig. S4, ESI†), which were consistent with a Co(II) complex containing MeBSIP and acetate ligands, moreover, distinct low temperature solid-state EPR spectra were obtained for the three compounds (Fig. S5, ESI†). Solution equilibria likely occur between the three species, but these must be slow, as assessment of the catalytic properties of 2 to 4 for O₂ reduction reveals stark differences. The three compounds were assayed for activity towards O₂ reduction in air saturated methanol, buffered by TBAOAc/HOAc, with decamethylferrocene (Fc*) as a stoichiometric reductant, with overall compositions such that [AcOH] > [Fc*] ≈ [O₂] ≫ [Catalyst]. Fc*, through the absorption of Fc*⁺ at 780 nm, also provides a spectrophotometric probe to monitor reaction progress.^13b,18 For 2, (Fig. 3 and Fig. S6–S8, ESI†) rapid emergence of absorption at 780 nm over 10 minutes indicates that 2 catalyses oxidation of Fc* to Fc*⁺, by reducing O₂. In the absence of 2, the increase in [Fc*] over the same time period is ca. 20× smaller (Fig. S11, ESI†). Iodometric titration with NaI (Fig. 3), which detects H₂O₂ through oxidation of I⁻ to I₃⁻, shows only a small spectral change consistent with ca. 10% of the O₂ being converted to H₂O₂ and thus a selectivity of around 90% for the 4e⁻ pathway to H₂O. Solubilities of 3 and 4 are poor in MeOH, but adequate concentrations (ca. 20 μM) could be achieved to show that 4 has almost no activity above the baseline, slow 2e⁻ reduction of O₂ to H₂O₂ by Fc*, while compound 3 shows around 1/3 of the activity of 2 (Fig. S9–S11, ESI†). This may be due to dissociation of the third Co centre, releasing active species 2. However, aging solutions of 3 and 4 in buffer does not increase activity.

Fig. 3 (a) Evolution of the 780 nm absorption from Fc*⁺ for a solution of 20 μM 2, 20 mM AcOH/[NBu₄][OAc] and 1 mM Fc* in air saturated MeOH. Inset: 780 nm absorbance vs. time. (b) Post-reaction iodometric titrations in MeCN for proton sources AcOH, (magenta) NH₄PF₆ (blue), NH₄OAc (green). 361 nm maxima are due to absorption by I₃⁻.

A turnover frequency (TOF) for 2 for O₂ production can be estimated at 0.031 s⁻¹, with initial concentrations of [2] = 20 μM, [AcOH/TBAOAc] = 16 mM, and [Fc*] = 1 mM, based on the initial rate of Fc* consumption adjusted for reaction selectivity (see ESI†). As ORR catalytic rates commonly show a strong dependence on overpotential, electrochemical measurements were used to evaluate the redox potentials of 2 (Fig. 4) and further underline (Fig. S12, ESI†) that the three species remain distinct in solution. The cyclic voltammogram of 2 in MeOH (50 mM TBAClO₄, Fig. 4), shows two pseudo-reversible waves at E_1/2s +0.27 V (ΔE = 300 mV) and 0.83 V (ΔE = 120 mV) vs. Fc*^0/+. Respectively, these are assigned to the Co^IICo^II/Co^IIICo^II ([2]^0/+) and Co^IIICo^II/Co^IIICo^III ([2]^+/2+) couples, the large peak separation of the [2]^0/+ process resulting from the reorganisation energy associated with a redox coupled spin crossover (HS–Co^IIHS–Co^II/LS–Co^IIIHS–Co^II). The smaller ΔE for [2]^+/2+ suggests that the HS state of the second Co is stabilised as previously proposed for dinuclear cobalt complexes.¹⁹ Adding AcOH/TBAOAc, to recreate the conditions used in the Fc* driven ORR (Fig. 4) substantially increases the peak currents and shifts the waves to E_1/2 +0.23 (ΔE = 260 mV) and 0.56 V (ΔE = 100 mV). Increased peak currents likely result from higher conductivity due to increased total electrolyte concentration – similar behaviour is observed for Fc*^0/+ – while the negative shift in the [2]^+/2+ redox potential implies an electrochemical anation reaction producing [2(OAc)] occurs on the first oxidation, the lowered charge facilitating the second process.

Fig. 4 Cyclic voltammetry of 2 (1 mM) before (blue) and after (red) addition of AcOH/TBAOAc (25 mM). MeOH, 50 mM TBAClO₄, glassy carbon electrode, 100 mV s⁻¹.

Like many ORR catalysts, the electrochemical response of 2 to air is too weak to distinguish an electrocatalytic wave, but the overpotential for molecular ORR catalysts (η_eff) can be estimated by the difference between the thermodynamic cell potential under the non-standard catalytic conditions, and the E_1/2 of the most negative catalyst redox event in the catalytic cycle: i.e. the [2]^0/+ couple. The E_O2/H2O in 25 mM AcOH/TBAOAc in methanol can be estimated at +0.92 V vs. Fc*^0/+,¹¹ giving an overpotential for 2 of 690 mV. For this overpotential, the TOF (0.031 s⁻¹) obtained is substantially faster than can generally be projected for Fe (<10⁻⁵)^8a porphyrins, or Co pyridyl derivatives (≪10⁻²) for the 4e⁻ process.^8d,e The literature lacks the data to enable detailed comparison with other dinuclear Co catalysts, but looking at metrics for molecular 4e⁻ ORR catalysts in general, 2 has a high selectivity and high TOF for its overpotential, although it should be noted that mononuclear Co complexes with pendant quinols have achieved better rate and overpotential combinations.⁹

Further chemically driven ORR experiments revealed a first-order dependence on [2] and [Fc*], but a zero-order dependence on [O₂] between 2 and 10 mM (Fig. S17–S20, ESI†). The latter is consistent with spectroscopic and electrochemical evidence for O₂ binding by 2 (Fig. S15, S16, S26–S30, ESI†), and [O₂] being ≥100× greater than [2] (20 μM), so nearly all of the 2 is O₂ bound. The former indicate a rate-determining step involving one equivalent of 2 and one electron, likely a slow electron transfer (ET) step that facilitates breakage of the strong O–O bond. Dependence on acid concentration is more complex: increasing [AcOH] in the absence of buffer decreases rate (Fig. S21, ESI†), but increasing buffer (AcOH/TBAOAc) concentration increases rate with a near first-order dependence (Fig. S22, ESI†). At constant buffer concentration, a second-order catalytic rate equation can be constructed (eqn (3)).

Considering the rate equation, alongside electrochemical measurements and studies of related catalysts from the literature, we propose a mechanism in which water ligands dissociate from 2, and O₂ is spontaneously and reversibly bound in an oxidative addition producing a peroxo-bridged Co^IIICo^III intermediate – evidenced by loss of paramagnetism in both NMR and EPR measurements (Fig. S26–S30, ESI†). Then, electron transfer (ET) to and protonation of this peroxo species results in an oxyl radical-hydroxo species which is quickly quenched by three protons and reducing equivalents to regenerate 2 (Scheme 1, right). The dependence on AcOH and AcO⁻ is complicated, because while protons are needed to complete the cycle, protonation of the catalyst appears to induce positive shifts in the [2]^0/+ redox potential (Fig. S13, ESI†), while coordination of acetate may block substrate access.

Scheme 1 Proposed mechanisms. Left: Two-electron reduction of O₂ to H₂O₂ by 2 in the absence of AcO⁻. Right: Four-electron reduction of O₂ to H₂O by 2 with AcO⁻ present.

The anion effect was further probed by changing proton source to NH₄PF₆ (pK_a in MeOH = 10.78, vs. 9.8 for AcOH). Remarkably, this flips the selectivity almost entirely to the 2e⁻ pathway, with iodometric titration revealing 93% of the theoretical [H₂O₂]. To test whether this change resulted from pK_a or coordination, catalysis was performed with NH₄OAc (pK_a = 10.78) – returning the system to the 4e⁻ pathway with 97% H₂O selectivity (Fig. 3). This result indicates that AcO⁻ binding is crucial to the 4e⁻ pathway. There are two possible explanations for this: (i) without excess AcO⁻ in solution, decoordination of the acetate ligand destabilises the dicobalt binding site for O₂, resulting in two, connected mono-Co salen-like catalysts that operate independently: mono-Co salens favour the 2e⁻ pathway via superoxo species; or (ii) the dicobalt O₂ binding site is retained, but loss of AcO⁻ makes the Co(III) centres more electron poor, so that electron transfer (ET) to Co and release of peroxide becomes more favourable than ET to the peroxo ligand and cleavage of the O–O bond.

Neither (i) nor (ii) can be definitively ruled out, but electrochemical, kinetic and spectroscopic evidence point towards (ii) (Scheme 1, left). Cyclic voltammetry in the presence of NH₄PF₆ (Fig. S14, ESI†) shows that the two redox processes of 2 are still present, indicating communication between the two Co centres (i.e. a dinuclear binding site), but shifted to more positive potential by ca. 80 mV vs. the acetate medium. A two-site catalyst could be expected to have a lower reaction order than Fc* or the proton source, but similar rate vs. concentration dependencies are seen for all of these species (Fig. S23–S25, ESI†). EPR under air at room temperature, 50 K and 10 K shows no evidence of superoxo species, but the Co^II signals disappear in the low temperature measurements, consistent with formation of a peroxo-bridged dicobalt(III) complex. This suggests a catalytic cycle of the type shown in Scheme 1, with the rate- determining step being a final PCET step to release H₂O₂. At the estimated overpotential of 80 mV, the TOF (0.125 s⁻¹) obtained for the 2e⁻ process is ca. 5x higher than expected for mononuclear Co^II salens,^7a which proceed via a Co^III superoxo intermediate, indicating an advantage for the dinuclear structure and consistent with a different mechanistic pathway.

In summary, we have designed and synthesised a novel dinuclear molecular catalyst with >90% selectivity for 4e⁻reduction of O₂ in acetate buffers, and >90% selectivity for the 2e⁻ pathway with non-coordinating PF₆⁻. This high selectivity is combined with highly competitive TOFs (0.031/0.125 s⁻¹) relative to overpotential (690/80 mV) for both processes. Both routes appear to involve a dicobalt(III) peroxo intermediate, with coordination/decoordination of acetate modulating the proton and electron transfer properties. Future studies will establish a η_eff vs. Log(TOF) trend between similar catalysts bearing electron withdrawing and donating substituents.

We thank Mr Tom Foster and Foster Contracting Ltd for funding a PhD studentship for CAJ at UEA. SPJ and RA thank the EPSRC for support through grant EP/X026876/1. X-ray data were obtained in facilities established by EPSRC grant EP/S005854/1. JF thanks Lancaster University (LU) for funding, and JES the RSC for Undergraduate Research Bursary U24-3344520590. AB acknowledges the EPSRC for support of the UK National Research Facility for EPR (EP/W014521/1). We thank Dr Nathan Halcovitch of LU for help finalising the X-ray structures.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data supporting this article are presented in the ESI† and are available at DOI:https://doi.org/10.17635/Lancaster/researchdata/731. Crystallographic data for 3 and 4 have been deposited at the CCDC, deposition numbers 2447929 and 2447930.†

Notes and references

1. (a) A. Sekretaryova, S. M. Jones and E. I. Solomon, J. Am. Chem. Soc., 2019, 141, 11304; (b) G. Gupta, V. Rajendran and P. Atanassov, Electroanalysis, 2004, 16, 1182; (c) E. C. M. Tse, D. Schilter, D. L. Gray, T. B. Rauchfuss and A. A. Gewirth, Inorg. Chem., 2014, 53, 8505.
2. (a) M. Chisaka, J. Mater. Chem. A, 2024, 12, 18636; (b) L. Zhang and Z. Xia, J. Phys. Chem. C, 2011, 115, 11170; (c) Z. H. Sheng, H. L. Gao, W. J. Bao, F. Bin Wang and X. H. Xia, J. Mater. Chem., 2012, 22, 390; (d) H. Zhong, L. Alberto Estudillo-Wong, Y. Gao, Y. Feng and N. Alonso-Vante, ACS Appl. Mater. Interfaces, 2020, 12, 21605.
3. (a) U. Latif, ACS Appl. Energy Mater., 2025, 8, 4838; (b) J. Zhang, X.-G. Zhang, J.-C. Dong, P. M. Radjenovic, D. J. Young, J.-L. Yao, Y.-X. Yuan, Z.-Q. Tian and J.-F. Li, J. Am. Chem. Soc., 2021, 143, 20049; (c) C. O. Laoire, S. Mukerjee, K. M. Abraham, E. J. Plichta and M. A. Hendrickson, J. Phys. Chem. C, 2009, 113, 20127–20134; (d) X. Ren, S. S. Zhang, D. T. Tran and J. Read, J. Mater. Chem., 2011, 21, 10118.
4. (a) K. V. N. Esguerra, Y. Fall, L. Petitjean and J.-P. Lumb, J. Am. Chem. Soc., 2014, 136, 7662; (b) A. Böttcher, M. W. Grinstaff, J. A. Labinger and H. B. Gray, J. Mol. Catal. A: Chem., 1996, 113, 191; (c) C. Wang and J. Xiao, Acc. Chem. Res., 2025, 58, 714.
5. (a) J. M. Campos-Martin, G. Blanco-Brieva and J. L. G. Fierro, Angew. Chem., Int. Ed., 2006, 45, 6962; (b) T. Nishimi, T. Kamachi, K. Kato, T. Kato and K. Yoshizawa, Eur. J. Org. Chem., 2011, 4113; (c) R. Zheng, Q. Meng, L. Zhang, J. Ge, C. Lui, W. Zing and M. Xiao, Chem. – Eur. J., 2023, 29, e202203180.
6. Y.-H. Wang, Z. K. Goldsmith, P. E. Schneider, C. W. Anson, J. B. Gerken, S. Ghosh, S. Hammes-Schiffer and S. S. Stahl, J. Am. Chem. Soc., 2018, 140, 10890.
7. (a) Y. H. Wang, B. Mondal and S. S. Stahl, ACS Catal., 2020, 10, 12031; (b) A. C. Brezny, S. I. Johnson, S. Raugei and J. M. Mayer, J. Am. Chem. Soc., 2020, 142, 4108; (c) A. Bettelheim and T. Kuwana, Anal. Chem., 1979, 51, 2257; (d) T. Kuwana, M. Fujihira, K. Sunakawa and T. Osa, J. Electroanal. Chem., 1978, 88, 299; (e) J. Zagal, M. Páez, A. A. Tanaka, J. R. dos Santos and C. A. Linkous, J. Electroanal. Chem., 1992, 339, 13; (f) T. Marshall-Roth, L. Liu, V. Mannava, D. M. Harraz, B. J. Cook, R. M. Bullock and Y. Surendranath, ACS Catal., 2024, 14, 18590.
8. (a) Y. H. Wang, M. L. Pegis, J. M. Mayer and S. S. Stahl, J. Am. Chem. Soc., 2017, 139, 16458; (b) C. W. Anson and S. S. Stahl, J. Am. Chem. Soc., 2017, 139, 18472; (c) M. A. Kamyabi, F. Soleymani-Bonoti, F. Alirezaei, R. Bikas, N. Noshiranzadeh, M. Emami, M. S. Krawczyk and T. Lis, Appl. Organomet. Chem., 2019, 33, e5214; (d) A. Das, A. Santra, A. Kumari, D. Ghosh and S. Paria, J. Am. Chem. Soc., 2025, 147, 6549; (e) A. W. Nichols, J. S. Kuehner, B. L. Huffman, P. R. Miedaner, D. A. Dickie and C. W. Machan, Chem. Commun., 2021, 57, 516.
9. S. V. Obisesan, C. Rose, B. H. Farnum and C. R. Goldsmith, J. Am. Chem. Soc., 2022, 144, 22826.
10. (a) S. N. Chowdhury, S. Biswas, P. Das, S. Paul and A. N. Biswas, Inorg. Chem., 2020, 59, 14012; (b) R. McGuire, D. K. Dogutan, T. S. Teets, J. Suntivich, Y. Shao-Horn and D. G. Nocera, Chem. Sci., 2010, 1, 411; (c) E. N. Cook, D. A. Dickie and C. W. Machan, J. Am. Chem. Soc., 2021, 143, 16411.
11. Y. H. Wang, P. E. Schneider, Z. K. Goldsmith, B. Mondal, S. Hammes-Schiffer and S. S. Stahl, ACS Cent. Sci., 2019, 5, 1024.
12. H. Y. Liu, M. J. Weaver, C. B. Wang and C. K. Chang, J. Electroanal. Chem. Interfacial Electrochem., 1983, 145, 439.
13. (a) Y. Liu, G. Zhou, Z. Zhang, H. Lei, Z. Yao, J. Li, J. Lin and R. Cao, Chem. Sci., 2019, 11, 87; (b) A. M. J. Devoille and J. B. Love, Dalton Trans., 2011, 41, 65.
14. G. Passard, A. M. Ullman, C. N. Brodsky and D. G. Nocera, J. Am. Chem. Soc., 2016, 138, 2925.
15. (a) H. Arima, M. Wada, T. Nakazono and T. Wada, Inorg. Chem., 2021, 60, 9402; (b) S. Fukuzumi, S. Mandal, K. Mase, K. Ohkubo, H. Park, J. Benet-Buchholz, W. Nam and A. Llobet, J. Am. Chem. Soc., 2012, 134, 9906.
16. J. Han, H. Tan, K. Guo, H. Lv, X. Peng, W. Zhang, H. Lin, U.-P. Apfel and R. Cao, Angew. Chem., Int. Ed., 2024, 63, e202409793.
17. J. Xie, Q. Zhou, C. Li, W. Wang, Y. Hou, B. Zhang and X. Wang, Chem. Commun., 2014, 50, 6520.
18. S. Fukuzumi, K. Okamoto, C. P. Gros and R. Guilard, J. Am. Chem. Soc., 2004, 126, 10441.
19. T. LeBlond and P. H. Dinolfo, Inorg. Chem., 2020, 12, 50.

---

Footnote

† Electronic supplementary information (ESI) available: Synthetic and other experimental details, CIF files. CCDC 2447929 and 2447930. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d5cc02476e

---