Tandem electrocatalytic N₂ fixation via proton-coupled electron transfer

Abstract

New electrochemical ammonia (NH₃) synthesis technologies are of interest as a complementary route to the Haber–Bosch process for distributed fertilizer generation, and towards exploiting ammonia as a zero-carbon fuel produced via renewably sourced electricity¹. Apropos of these goals is a surge of fundamental research targeting heterogeneous materials as electrocatalysts for the nitrogen reduction reaction (N₂RR)². These systems generally suffer from poor stability and NH₃ selectivity; the hydrogen evolution reaction (HER) outcompetes N₂RR³. Molecular catalyst systems can be exquisitely tuned and offer an alternative strategy⁴, but progress has been thwarted by the same selectivity issue; HER dominates. Here we describe a tandem catalysis strategy that offers a solution to this puzzle. A molecular complex that can mediate an N₂ reduction cycle is partnered with a co-catalyst that interfaces the electrode and an acid to mediate proton-coupled electron transfer steps, facilitating N–H bond formation at a favourable applied potential (−1.2 V versus Fc^+/0) and overall thermodynamic efficiency. Certain intermediates of the N₂RR cycle would be otherwise unreactive via uncoupled electron transfer or proton transfer steps. Structurally diverse complexes of several metals (W, Mo, Os, Fe) also mediate N₂RR electrocatalysis at the same potential in the presence of the mediator, pointing to the generality of this tandem approach.

Molecular catalysts offer a number of favourable characteristics compared with heterogeneous electrode materials for electrocatalytic nitrogen reduction reaction (N₂RR) studies. In particular, they can be carefully tuned to satisfy the electronic requirements of N₂ binding and activation. They can also afford access to insightful mechanistic studies at the level of critical bond-breaking and bond-making steps (Fig. 1a). Remarkable progress has been made over the past two decades in terms of chemically driven N₂RR catalysis and associated mechanistic understanding using molecular systems^4–7. Despite this, bona fide N₂RR electrocatalysis in this domain remains virtually unknown^8–10; only one such molecular electrocatalyst (a tris(phosphine)borane iron ((TPB)Fe) system from our laboratory; Fig. 1b) has been reliably demonstrated, but it requires low temperatures (−35 °C) to mitigate the background hydrogen evolution reaction (HER) and a highly reducing potential (−2.1 V versus ferrocenium/ferrocene, Fc^+/0; all potentials herein are reported versus Fc^+/0)¹¹. This state of affairs sharply contrasts the substantial progress that has been made applying molecular systems towards electrocatalytic HER, the carbon dioxide reduction reaction and the oxygen reduction reaction, among other transformations^12–14.

Fig. 1 |. Approaches to N₂RR electrocatalysis.

a, A representative distal cycle for catalytic N₂-to-NH₃ conversion. b, A previously reported molecular N₂RR electrocatalyst based on the tris(phosphine)borane Fe system, operating at −2.1 V on a glassy carbon electrode using a temperature of −35 °C. c, Early work by Pickett demonstrating electrosynthesis of NH₃ using the molecular complex W(N₂)₂ at an applied potential of −2.6 V on a Hg-pool electrode using TsOH. The protonation step had to be performed separately from the reduction step (0.21 equiv. NH₃ per W(N₂)₂ after one cycle; 0.73 equiv. total NH₃ per W(N₂)₂ after three cycles). d, Tandem catalysis described in this work based on coupling the PCET mediator, Co(II,NH)⁺, with molecular N₂RR catalysts to enable well-defined electrocatalysis at comparatively mild potentials (−1.2 V using TsOH).

Similar to candidate heterogeneous electrocatalysts, molecular systems typically mediate HER in preference to N₂RR, and/or operate at such reducing potentials that background HER at a working electrode dominates. Pioneering research from 1985 by Pickett and co-workers underscored this point (Fig. 1c). In a study involving a bis(diphenylphosphinoethane)tungsten system (abbreviated throughout as W), they showed that the hydrazido complex (TsO)W(NNH₂)⁺ (TsO, tosylate), generated via protonation of the bis-N₂ adduct W(N₂)₂ by tosic acid (TsOH), releases NH₃ (0.21 equiv. NH₃ per W) after application of a highly reducing potential (−2.6 V on a Hg-pool electrode), but only in the absence of the acid^15,16. Positively shifting the overpotential of the candidate electrocatalyst would attenuate the HER, but the linear free energy relationship between overpotential and catalytic rate requires an additional strategy for maintaining an appreciable N₂RR rate at a lower driving force¹⁷.

Recent work from our laboratory introduced a strategy for attenuating the rate of (electro)catalytic HER. By using a proton-coupled electron transfer (PCET) mediator, consisting of cobaltocenium modified by a tethered Brønsted base (abbreviated herein as Co(III,N)⁺; Fig. 1d)¹⁸, the catalysed HER is prevented. This mediator design spatially and electronically separates the proton and electron relays, which is key to storing highly reactive H atom equivalents in Co(II,NH)⁺ at a potential that is sufficiently mild to also mitigate background HER at the electrode. Initial model studies using this mediator established that concerted proton–electron transfer (CPET) provides a means to reduce unsaturated organic substrates electrochemically at comparatively mild potentials in the presence of tosic acid^18,19.

Although these results point to the possibility of applying such a mediator towards electrocatalytic N₂RR, the mediator itself does not react with N₂, in contrast to some unsaturated organic substrates. Hence, we have pursued a tandem catalysis strategy (Fig. 1d)^20,21, pairing a candidate molecular catalyst that can bind N₂ (M−N₂) and facilitate its multistep reduction to NH₃ through various M−N_xH_y intermediates (for example, M−N=NH, M=NNH₂, M=NH)⁴, with a mediator that interfaces the electrode and the acid with the N₂ reduction cycle via PCET steps. Importantly, certain N₂RR intermediates are challenging to move through the cycle; they can be difficult to independently reduce or protonate (see below). In principle, a PCET step can circumvent this issue and favourably shift the overpotential needed to drive the net electrochemical N₂RR process. Here we show the feasibility of this tandem catalysis strategy at room temperature and atmospheric pressure.

As a model system to test our tandem approach we adopted the classic tungsten system studied by Pickett¹⁵. Using W(N₂)₂ and the same solvent (tetrahydrofuran; THF), electrolyte (0.2 M [TBA][BF₄]; TBA, tetra-N-butylammonium) and acid (100 equiv. TsOH), in the presence of the cobalt PCET mediator, Co(III,N)⁺, controlled potential coulometry (CPC) produced 4.7 ± 0.3 equiv. NH₃ at 18 ± 2% Faradaic efficiency (FE) over a period of 11 h using a glassy carbon (GC) electrode at −1.35 V (see Supplementary Section 4). Reloading the system with an additional 100 equiv. TsOH furnished a total of 7.6 equiv. NH₃. These initial results show that inclusion of the Co(III,N)⁺ mediator turns on electrocatalysis by W(N₂)₂, at a potential that is 1.25 V positive of Pickett’s original work (Fig. 1c)¹⁵. In the absence of the mediator, electrocatalysis is not observed; N₂RR is not kinetically competitive under these conditions, presumably because an uncoupled ET-PT pathway is not facile (see Supplementary Section 13).

Canvassing factors to improve electrocatalytic N₂RR by this tandem W(N₂)₂/Co(III,N)⁺ co-catalyst system (see Supplementary Section 5) led us to adopt a boron-doped diamond (BDD) working electrode, dimethoxyethane (DME) solvent and lithium bistriflimide ([Li][NTf₂]) electrolyte. Under these optimized conditions (BDD, 0.1 M [Li][NTf₂], DME, 5 mM TsOH, 0.05 mM W(N₂)₂/Co(III,N)⁺) N₂RR electrocatalysis notably improved, with 11.3 ± 0.5 equiv. NH₃ per W(N₂)₂/Co (44.5 ± 1.9% FE) being generated at −1.35 V over 5.5 h (Fig. 2a,b). Quantification of the H₂ in the headspace after an equivalent CPC experiment results in 39% FE for HER. A higher turnover number per W(N₂)₂/Co(III,N)⁺(up to 39.5 equiv.) was demonstrated by using a higher surface area GC foam electrode and lowering the catalyst concentration (Supplementary Table 1). Control experiments to demonstrate that both catalysts are required, to rule out the presence of catalytically active decomposition products and to show that N₂ is the source of the NH₃ generated (Fig. 2c) are provided in the Supplementary Information.

Fig. 2 |. Electrocatalytic N₂RR via tandem catalysis.

a, Electrocatalytic N₂RR on CPC at −1.35 V versus Fc^+/0 in 0.1 M [Li][NTf₂] DME solution containing 0.05 mM Co(III,N)⁺, 0.05 mM W(N₂)₂ and 5 mM TsOH, using a BDD plate working electrode. b, Current (j) profile for the CPC experiment described (black trace) and a similar CPC experiment in the absence of the Co(III,N)⁺ mediator. c, Quantification of NH₃ following CPC via ¹H NMR spectroscopy using either ¹⁴N₂, ¹⁵N₂ or an argon atmosphere. Average result for W(N₂)₂ and its associated error (standard deviation) corresponds to three electrocatalytic runs.

To assess the electrochemical behaviour of the W(N₂)₂/Co(III,N)⁺co-catalyst system, a series of cyclic voltammograms (CVs) were collected. Following previous studies, dissolution of W(N₂)₂ in THF with added TsOH quantitatively produces the doubly protonated hydrazido complex (TsO)W(NNH₂)⁺ (ref.¹⁶). CVs of (TsO)W(NNH₂)⁺ in a 0.1 M [Li] [NTf₂] THF solution on a BDD working electrode show two irreversible one-electron waves at low potential (less than −1.9 V; Fig. 3a). These waves are due to the generation of W(NNH₂)⁺ and W(NNH₂), respectively¹⁵, in which the strongly reducing potential reflects the challenge in reducing the 18-electron, closed-shell (TsO)W(NNH₂)⁺ complex. Although addition of excess TsOH (100 equiv.) to the solution containing (TsO)W(NNH₂)⁺ leads to an increase in current (irreversible) with an onset at −1.3 V, the same response is observed without added W and is due to background HER at the electrode (Fig. 3a). The independent CV of Co(III,N)⁺ in THF shows a reversible Co^III/II couple at −1.35 V (Fig. 3b), assigned to Co(III,N)⁺/Co(II,N)¹⁸. This couple shifts to −1.21 V when the mediator is protonated at the tethered dimethylaniline group (that is, Co(III,NH)²⁺/Co(II,NH)⁺) (Fig. 3a). Gratifyingly, CVs of (TsO)W(NNH₂)⁺ in the presence of Co(III,N)⁺ and TsOH result in an irreversible multi-electron wave at −1.2 V (Fig. 3a), consistent with electrocatalytic N₂RR. At 100 mV s⁻¹ an approximately threefold increase in the catalytic current is observed, compared with the current of the one-electron reduction wave of Co(III,NH)²⁺, which is indicative of relatively slow N₂RR electrocatalysis. When the scan rate is reduced to 25 mV s⁻¹, the increase is 6.5-fold. Therefore, on this timescale, NH₃ is expected to be produced not only at the surface of the electrode, as supported by rotating ring-disk electrode (RRDE) experiments (see Supplementary Information), but also in the bulk solution, as evidenced by chemical reactions between Co(II,NH)⁺ and (TsO)W(NNH₂)⁺ and (TsO)W(NH)⁺ (see below). The reversible CV response of the Co(III,NH)²⁺/Co(II,NH)⁺ couple at −1.21 V (in the absence of TsOH) is noticeably altered as (TsO)W(NNH₂)⁺ is added; if scanning at a slow rate (for example, 5–25 mV s⁻¹), the presence of the Co(III,N)⁺/Co(II,N) couple becomes clearly evident (Fig. 3b). The implication is that, as Co(II,NH)⁺ is generated in the presence of (TsO)W(NNH₂)⁺, a PCET step occurs that generates Co(III,N)⁺, the CV response of which becomes apparent at scan rates well matched to the kinetics of this chemical step in the absence of acid. On the basis of kinetic analysis viacyclic voltammetry (see Supplementary Section 13), a rate of approximately 0.5 s⁻¹ is estimated for this PCET reaction, which is consistent with the rate-contributing nature of this step and relatively slow catalysis overall.

Fig. 3 |. Mechanistic insights into tandem PCET N₂RR.

a, CV of 50 mM TsOH (dashed grey); 0.5 mM Co(III,N)⁺ with 50 mM TsOH (black); 0.5 mM (TsO)W(NNH₂)⁺ (solid red); 0.5 mM (TsO)W(NNH₂)⁺ with 50 mM TsOH (dashed red); 0.5 mM Co(III,N)⁺/(TsO)W(NNH₂)⁺ with 50 mM TsOH (purple). b, CV at 5 mV s⁻¹ of 0.5 mM Co(III,NH)²⁺/(TsO)W(NNH₂)⁺ (purple trace) compared to 0.5 mM Co(III,NH)²⁺ (red trace) and 0.5 mM Co(III,N)⁺ (blue trace). c, Chemical reaction of 0.5 mM (TsO)W(NNH₂)⁺ in THF with 2 equiv. Co(II,N) in THF in the presence of excess acid and the corresponding ³¹P NMR spectrum. d, CV of 0.5 mM (N₃)W(N) (black); 0.5 mM (N₃)W(N) and 1 equiv. TsOH (solid red); 0.5 mM (N₃)W(N) with 50 mM TsOH (dashed red); 0.5 mM Co(III,N)⁺/(N₃)W(N) with 50 mM TsOH (purple). e, Plot of the catalytic current (i_cat) versus the concentration of the different co-catalysts. f, CVs of several representative M(N₂) catalysts (see Fig. 4a for their chemical structures) studied under the standard electrocatalytic conditions in the presence of Co(III,N)⁺ and TsOH, showing a multi-electron catalytic wave for N₂RR in each case at the same applied bias as for W(N₂)₂. Note that all CVs in a–f were performed at 100 mV s⁻¹ (unless otherwise stated) in 0.1 M [Li][NTf₂] THF solution using a BDD disk as the working electrode, Pt disk as the counter electrode and Ag/AgOTf (5 mM) as the reference electrode. eN₂RR, electrocatalytic N₂RR.

To independently probe the PCET step, we generated (TsO)W(NNH₂)⁺ in THF with excess TsOH present, and added 2 equiv. Co(II,N) to the solution, conditions under which Co(II,NH)⁺ is instantly generated. Such a reaction liberates 0.39 equiv. NH₃ per (TsO)W(NNH₂)⁺ (Fig. 3c) over 4 h (note: this experiment was performed at 0 °C to attenuate the competing HER). Analysis of the reaction mixture by ³¹P NMR spectroscopy showed some remaining (TsO)W(NNH₂)⁺ starting material and also a new peak corresponding to the imido complex (TsO)W(NH)⁺. The identity of the latter species was confirmed by its independent generation via the protonation of the nitride precursor (N₃)W(N) with TsOH (Supplementary Fig. 56)^22–24. Although other processes are presumably operating in this reaction (for example, HER and proton transfer (PT) steps to other W−N_xH_y intermediates), the rate correlates well with the CV experiment noted above, suggesting that PCET to (TsO)W(NNH₂)⁺ occurs, presumably followed by N−N cleavage and NH₃ release (equation (1) below depicts one plausible scenario). Such a PCET step helps to explain why the system can be turned over at −1.2 V, whereas one-electron reduction of (TsO)W(NNH₂)⁺ requires a potential of −1.9 V.

$$\left(\text{TsO}\right)W{({\text{NNH}}_2)}^{+}+Co{\left(\text{II},\text{ NH}\right)}^{+}\to \left(\text{TsO}\right)W{\left(\text{N}\right)}^{+}+Co{\left(\text{III},\text{ N}\right)}^{+}+{\text{ NH}}_3$$ (1)

Related to this, CV of (TsO)W(NH)⁺ (Fig. 3d), a presumed downstream intermediate of the catalytic cycle generated by in situ protonation of the nitride complex (N₃)W(N), shows a similar electrocatalytic behaviour in the presence of Co(III,N)⁺. Additionally, the reaction of (N₃)W(N) with 4 equiv. Co(II,N) in the presence of excess TsOH afforded 0.70 equiv. NH₃ per W, along with (TsO)W(NNH₂)⁺ and (TsO)W(NH)⁺ detected by ³¹P NMR spectroscopy (Supplementary Fig. 59). Again, PCET from Co(II,NH)⁺ to (TsO)W(NH)⁺ is probably key, given the a one-electron reduction potential for a (TsO)W(NH)⁺ species of approximately −1.8 V (Fig. 3d).

By independently varying the concentration of the W(N₂)₂ and Co(III,N)⁺ we could determine a positive order for both co-catalysts in the electrocatalytic response (Fig. 3e). Interestingly, a positive order in acid was also evident (Supplementary Fig. 73), as was a primary kinetic isotope effect (2.5) when comparing TsOH versus TsOD (Supplementary Fig. 79). These electrochemical and chemical data are consistent with one or perhaps two rate-contributing PCET steps, involving (TsO)W(NNH₂)⁺ and possibly also (TsO)W(NH)⁺, and a rate-contributing protonation step (such as initial protonation of W(N₂)₂; see Supplementary Section 9). Thermodynamic arguments using associated rate estimates (see Supplementary Section 13)^25,26 point to a probable concerted PCET step as being turnover limiting in this catalysis, akin to simpler reactions using this mediator with unsaturated organic substrates that we have previously attributed to CPET^18,19.

To explore our strategy more broadly, we turned our attention to a series of complexes known to mediate catalytic N₂RR in the presence of various reductant/acid reagents (Figs. 3f and 4a). We opted to test them under the standard conditions (Fig. 2a), reasoning that some degree of electrocatalysis might turn on at −1.2 V if PCET steps from Co(II,NH)⁺ can similarly circumvent the need for challenging ET steps requiring more negative potentials.

Fig. 4 |. Electrocatalytic N₂RR using reported molecular catalysts.

a, Molecular N₂RR catalysts explored in combination with the PCET mediator under electrocatalytic conditions. b, Results of the electrocatalytic experiments for each molecular catalyst on CPC at −1.35 V versus Fc^+/0 in 0.1 M [Li][NTf₂] DME solution containing 0.05 mM Co(III,N)⁺, 0.05 mM of N₂RR catalyst and 5 mM TsOH, using a BDD plate working electrode. Average result for W(N₂)₂ and its associated error (standard deviation) corresponds to three electrocatalytic runs. *A GC foam was used as the working electrode instead and the concentration was 0.01 mM for both co-catalysts and 1 mM TsOH. c, Estimated overpotential (ΔΔG_f) for N₂RR, including the tandem PCET strategy reported here, a nitrogenase enzyme, Li-mediated N₂RR using EtOH as the H⁺ source and various reductant and acid partners used in chemically driven N₂RR.

To probe this, we examined the group VIII complexes Fe(N₂) and Os(N₂), where Fe and Os feature tris(phosphine)borane and silane ligands, respectively (Fig. 4a). Each mediates chemical N₂RR at −78 °C, but requires a comparatively strong reductant (Cp*₂Co at −2 V) owing to an M−N₂^0/− couple that is key to moving through their respective N₂RR cycles via uncoupled electron transfer-proton transfer steps (equations (2) and (3))^27,28. Strikingly, both Fe and Os display an electrocatalytic wave at −1.2 V (Fig. 3f), akin to W(N₂)₂, and CPC at −1.35 V produced 5.6 and 4.5 equiv. NH₃, respectively (Fig. 4b). Despite their relatively lower selectivity for NH₃ generation compared to W(N₂)₂ under these conditions, the electrocatalysis observed represents a remarkable shift in thermodynamic efficiency for the overall N₂RR cycle relative to previously reported conditions (see below). To explain this, we posit that the neutral M−N₂ adduct species are converted directly to M−N=NH intermediates via PCET from Co(II,NH)⁺ (equation (4)), circumventing the M−N₂^0/− couple in the cycle. For Fe specifically, generation of the on-path Fe(N₂) species requires reduction of the Fe⁺ pre-catalyst used here, which occurs at about −1.4 V (see Supplementary Section 14). Thus, applying slightly more bias in the CPC (−1.45 V instead of −1.35 V) results in improved NH₃ yield (9.3 equiv. NH₃ per Fe). A related tris(phosphine)silyl iron–N₂ complex, (SiP₃)Fe(N₂), is instead electrocatalytically inactive, presumably due to the generation of an undesired (SiP₃)Fe(H)(N₂) state⁷

$$M-{\text{N}}_2+{\text{e}}^{-}\to M-{\text{N}}_2^{-}(-2\text{V})$$ (2)

$$M-{\text{N}}_2^{-}+{\text{H}}^{+}\to M-{\text{N}}_2\text{H}$$ (3)

$$M-{\text{N}}_2+Co{(\text{II},\text{NH})}^{+}\to M-{\text{N}}_2\text{H}+Co{(\text{III},\text{N})}^{+}$$ (4)

We also explored a series of Mo complexes, including two tetrakis (phosphine) systems that are structurally related to W(N₂)₂ (compounds 1 and 2 in Fig. 4a), and highly active pincer-type bis(phosphine)pyridine complexes (3 and 4 in Fig. 4a) pioneered more recently^6,29. Among the reductants that have proved to be effective for these systems, SmI₂/H₂O has led to the most impressive results in chemically driven catalysis³⁰. We find that the Mo(N₂)₂ complexes 1 and 2 are both effective co-electrocatalysts with favourable selectivities, furnishing 13 and 14 equiv. NH₃ (51 and 55% FE for NH₃), respectively. The dinuclear Mo catalyst system (3) also displays electrocatalysis under these conditions (8.7 equiv. NH₃ per Mo)⁶. By contrast, the mononuclear triiodide complex (4), which has been demonstrated to be highly active for N₂RR³¹, is electrocatalytically inactive under these conditions (<0.1 equiv. NH₃ detected). The latter observation is readily explained; the strong reduction potential (−1.8 V) required to access an on-path N₂RR intermediate by iodide loss is not accessible at −1.35 V.

The free energy for the electrocatalytic N₂RR processes described here compares quite favourably to estimates for other systems that mediate catalytic and electrocatalytic N₂RR. This can be readily quantified by ΔΔG_f(NH₃), a term that compares the energetic input for N₂RR relative to a reaction that derives the needed protons and electrons from H₂ (equation (5))²⁶. Using the bond dissociation free energy (BDFE) for H₂ (102.5 kcal mol⁻¹)³² and that of the PCET mediator Co(II,NH)⁺ (38.9 kcal mol⁻¹)¹⁸, the ΔΔG_f(NH₃) is 36.5 kcal mol⁻¹ for the electrocatalysis observed at −1.2 V by our CV studies. This net driving force is at least 50 kcal mol⁻¹ lower than has been reported for most other reductant/acid cocktails used with synthetic N₂RR catalyst systems (Fig. 4c): SmI₂/H₂O (75 kcal mol⁻¹), Cp*₂Co/[Ph₂NH₂]⁺ (77 kcal mol⁻¹), KC₈/HBAr^F₄ (196 kcal mol⁻¹) (BAr^F₄, B(3,5-(CF₃)₂-C₆H₃)^7,27,30. A crude comparison with the biological nitrogenases (approximately 117 kcal mol⁻¹ accounting for ATP) is also favourable³². Likewise, heterogeneous systems based on Li⁺/Li (E°(Li^+/0) < −3.7 V), which commonly utilize ethanol as the acid, operate at an estimated ΔΔG_f(NH₃) = 133 kcal mol⁻¹ (ref.³³). Interestingly, one combination of reductant and acid, Cp₂Co and lutidinium, first studied in the Schrock system and later applied towards N₂RR catalysis with the Nishibayashi bis(phosphine)pyridine molybdenum catalyst studied herein (complex 3 in Fig. 4)^5,6, is thermally favourable by comparison (ΔΔG_f(NH₃) = 26 kcal mol⁻¹). This suggests that alternative acids and mediator designs may yet improve the efficiency achievable by tandem electrocatalysis.

$$\Delta \Delta G_{\text{f}}\left({\text{NH}}_3\right)=3\left[\text{BDFE}\left({\text{H}}_2\right)/2-{\text{BDFE}}_{\text{eff}}\right]$$ (5)

In closing, it is widely appreciated that PCET steps can offer thermodynamic advantages relative to distinct ET-PT or PT-ET pathways in enzyme catalysis, in which multi-electron redox reactions must be driven at biologically accessible potentials²⁶, and also in synthetic catalyst systems⁴. The tandem catalytic approach to N₂RR via electrochemical PCET described herein provides a vivid example of the latter, in which a PCET step that we suggest is largely concerted turns on catalysis that is otherwise inaccessible at the applied potential. A comparison with nitrogenase enzymes is illustrative here. It has been posited that the active-site cofactors of nitrogenases store up proton and electron equivalents via H atoms bound at or near to the active-site cluster, to be able to mediate N₂ reduction at a single redox potential (set by the potential of the Fe protein)³⁴. Our two-component tandem catalyst system functions in a conceptually similar manner, in which the mediator independently stores an H atom equivalent at its own redox potential for delivery to a synthetic M−N₂ active site.

Data availability

Details on the procedures and the corresponding datasets generated during and/or analysed during the current study are available within the paper and its Supplementary Information files, and from the corresponding author on reasonable request.