Electrocatalytic Nitrogen Reduction on a Molybdenum Complex Bearing a PNP Pincer Ligand

Abstract

Electrocatalytic nitrogen reduction (N₂R) mediated by well-defined molecular catalysts is poorly developed by comparison with other reductive electrocatalytic transformations. Herein, we explore the viability of electrocatalytic N₂R mediated by a molecular Mo-PNP complex. A careful choice of acid, electrode material, and electrolyte mitigates electrode-mediated HER under direct electrolysis and affords up to 11.7 equiv of NH₃ (Faradaic efficiency < 43%) at −1.89 V versus Fc⁺/Fc. The addition of a proton-coupled electron transfer (PCET) mediator has no effect. The data presented are rationalized by an initial electron transfer (ET) that sets the applied bias needed and further reveal an important impact of [Mo] concentration, thereby pointing to potential bimolecular steps (e.g., N₂ splitting) as previously proposed during chemically driven N₂R catalysis. Finally, facile reductive protonation of [Mo(N)Br(^HPNP)] with pyridinium acids is demonstrated.

Graphical Abstract

Because of the role of ammonia as chemical fertilizer feedstock and as a possible energy vector, the nitrogen reduction (N₂R) reaction offers promise in sustainable and decentralized energy storage and fertilizer synthesis where the energy to drive N₂R can be renewably sourced.^1–3 Toward this end, significant attention is turning toward achieving efficient N₂R via electrocatalysis.^4–7

From the perspective of fundamental studies, molecular N₂R electrocatalysts can offer distinct advantages in terms of synthetic tunability, selectivity, and suitability toward detailed mechanistic investigations.⁸ However, their systematic development has been hampered by the strong reduction potential required to activate nitrogen.^9–13 Under such conditions in acidic media, background hydrogen evolution reaction (HER) at the surface of the electrode usually dominates.¹⁴ For instance, the tris(phosphino)borane iron system, P₃^BFe, has been demonstrated to be a modest N₂R electrocatalyst when interfaced with a glassy carbon electrode and an ammonium acid but requires low temperature (−35 °C) to mitigate background HER.¹⁵ Relatedly, important progress has recently been made in electrochemical N₂ splitting by molecular complexes to generate metal nitrides (M–N),^16–19 but further development is warranted to incorporate a proton source toward successful electrocatalytic N₂-to-NH₃ conversion.

While chemical reductants can lead to similar selectivity issues, a careful choice of reagent cocktails and conditions can mitigate competing HER, and impressive selectivities have been demonstrated for N₂R catalysis via such approaches.^8,20–23 Analogous efforts to attenuate HER and, hence, enhance the relative rate of N₂R versus HER under electrochemical conditions are essential.^24,25

Very recently, our lab reported that a tandem electrocatalytic strategy based on a proton-coupled electron transfer (PCET) mediator (Figure 1A) affords a promising approach.⁷ Through the use of a cobaltocene-derived PCET mediator [Co (III,N)⁺; Figure 1],²⁶ H⁺ and e⁻ equivalents (net H atoms) are transferred to certain M-N_xH_y intermediates at milder potentials than would otherwise be required via stepwise electron transfer–proton transfer (ET−PT) steps. This strategy enables N₂R electrocatalysis at room temperature (rt) pinned to the potential of the PCET mediator (onset at −1.2 V vs Fc⁺/Fc; all potentials herein are referenced Fc^+/0) using a variety of molecular complexes (e.g., Fe, Os, Mo, W) with tosic acid as the H⁺ source; background HER is attenuated at this potential.

Figure 1.

(A) Tandem ePCET strategy for N₂R. (B) Chemically driven N₂R catalysis and electrochemical N₂ splitting by [Mo^IIIX₃(^HPNP)]. (C) Previously proposed mechanism for N₂ splitting by [Mo^IIIX₃(^HPNP)].

Certain ET steps during N₂R may not be immediately followed by protonation; the utility of a PCET mediator would be limited in cases where discrete ET steps determine the applied potential needed to drive electrocatalysis. An illustrative example is the family of fascinating N₂R precatalysts described by Nishibayashi and co-workers on the basis of the [Mo^IIIX₃ (^H PNP)] platform (^HPNP = 2,6-bis(di-tert-butylphosphinomethyl)pyridine; X = Cl, Br, I; Figure 1B).^27,28 These complexes have been proposed to undergo bimolecular dinitrogen splitting upon accessing a [Mo^IX (^HPNP)] state to furnish a terminal nitride [Mo^IV(N)X-(^HPNP)] (Figure 1C). Further reductive protonation of the nitride intermediate releases NH₃. In this case, N₂ binding and activation is gated by a limiting ET step associated with the Mo^III/II redox couple (E° < −1.70 V).

Relatedly, an elegant mechanistic study by Miller and coworkers demonstrated electrochemical N₂ splitting using this same platform, where [Mo^IV(N)Br(^HPNP)] is generated from a net two-electron reduction of [Mo^IIIBr₃ (^HPNP)] at −1.89 V (Figure 1B).¹⁶ However, the use of [Mo^IIIX₃ (^HPNP)] as an electrocatalyst has remained elusive despite its remarkable activity for chemical N₂R. Here, we explore factors that enable N₂R electrocatalysis using [Mo^IIIBr₃ (^HPNP)] and find that the electrode, the acid, the electrolyte, and the solvent play essential roles in enhancing N₂-to-NH₃ conversion.

We first explored the impact of incorporating the Co (III,N)⁺ PCET-mediator (Figure 1A) to facilitate N−H bond-forming steps toward electrocatalytic N₂R with [Mo^IIIX₃ (^HPNP)] catalysts. Distinct from other catalysts canvassed,⁷ at −1.35 V, only trace NH₃ could be detected. This suggests a required applied potential of <−1.70 V for [Mo^IIIX₃(^HPNP)] in order to access an on-path N₂R intermediate via an ET step. We wondered whether a downstream PCET step (e.g., reductive protonation of a Mo−N intermediate) might still play an important role in enabling N₂R; hence, we evaluated the viability of this tandem PCET-mediated approach under more reducing conditions (Figure 2A).

Figure 2.

(A) Tandem attempts at electrocatalytic N₂R by [Mo^IIIX₃(^HPNP)] using a cobaltocene-derived PCET mediator. (B) Optimized conditions for electrocatalytic NH₃ generation using [Mo^IIIBr₃(^HPNP)]. (C) Cyclic voltammograms of 0.5 mM [Mo] (red trace), 50 mM [ColH][OTf] (blue trace), and 0.5 mM [Mo] with 50 mM [ColH][OTf] (purple trace) in a THF solution containing 100 mM [Li][NTf₂]. A BDD working electrode was employed at a 100 mV/s scan rate.

Controlled potential electrolysis (CPE) at −1.89 V yielded substoichiometric amounts of ammonia (0.7 equiv NH₃) using similar conditions as previously reported: 0.1 mM Co (III,N)⁺, 0.1 mM [Mo^IIIBr₃ (^HPNP)], 100 equiv of tosic acid (TsOH·H₂O), and 0.1 M [Li][NTf₂] electrolyte in THF solution with a boron-doped diamond (BDD) working electrode (see Supporting Information (SI) for electrochemical details and electrode sizes). An analogous CPE experiment in the absence of Co(III,N)⁺ produced similar results (0.8 equiv NH₃), which suggests that PCET mediation does not enhance the N₂R efficiency of this catalyst system.

We next explored other factors that might enable electro-catalysis with [Mo^IIIBr₃ (^HPNP)]. TsOH is well matched to the protonation of Co (III,N)⁺ but also leads to significant background HER (E_onset = −1.5 V). This HER likely outcompetes the alternative reduction of [Mo^IIIBr₃ (^HPNP)] and corresponding N₂-derived intermediates. Thus, conditions were explored to attenuate electrode-mediated HER. From the pyridinium acids typically used in chemical N₂R with Mo-based catalysts,²⁹ we found that collidinium triflate ([ColH]-[OTf]) provides a wide redox window for operation.

Cyclic voltammetry (CV) on a BDD working electrode at 100 mV/s in the presence of 50 mM [ColH][OTf] in 100 mM [Li][NTf₂] electrolyte THF solution revealed an increase in current associated with electrode-mediated HER starting at −1.80 V. CV of 0.5 mM [MoBr₃(^HPNP)] (Figure 2C) in the same electrolyte solution but in the absence of acid featured a reduction wave at −1.80 V associated with the Mo^III/II redox couple, which is consistent with previous observations. The addition of [ColH][OTf] (50 mM) resulted in a current increase at the Mo^III/II redox wave that is suggestive of N₂R electrocatalysis at potentials sufficiently anodic of the background HER. Accordingly, a controlled potential electrolysis (CPE) using 0.10 mM [Mo] and 100 equiv of [ColH][OTf] at −1.89 V furnished 9.4 ± 0.3 equiv of ammonia per Mo atom (FE = 34 ± 1%) over 10 h (Figure 2B). CPE experiments in the absence of [MoBr₃(^HPNP)] failed to produce significant amounts of NH₃. Control experiments corroborated the electrocatalytic nature of this system for N₂-to-NH₃ conversion (see SI).

Comparative analysis of other pyridinium acids revealed a strong impact on the basis of the acid choice during electrocatalytic N₂R, as inferred from CV data (SI) and measured NH₃ yields from CPE experiments (Table 1). For example, the use of more acidic lutidinium triflate ([LutH]-[OTf]) enhanced background HER and led to a corresponding drop in the FE to ~ 16% (4.5 equiv NH₃), despite the fact that [LutH][OTf] can perform well as the acid under chemically (as opposed to electrochemically) driven catalysis with Mo complexes.³⁰ Picolinium triflate ([PicH][OTf]) afforded the highest background HER current and very low NH₃ conversion (5% FE, 1.3 equiv NH₃). Therefore, the use of weaker acids that are still compatible with N₂R proves to be essential to optimize the FE of electrocatalytic NH₃ formation.

Table 1.

Summary of CPE Experiments after Passing 2.4 C of Charge, Corresponding to 83% Consumption of Acid

entry	deviation	NH3 (equiv/Mo)	Faradaic efficiency (%)
acid (↑ pKa, ↑ [NH3])
1	[LutH][OTf] instead of [ColH][OTf]	4.5	16
2	[PicH][OTf] instead of [ColH][OTf]	1.3	5
3	(TsOH)(H2O) instead of [ColH][OTf]	0.8	3
PCET mediator
4	(TsOH)(H2O) instead of [ColH][OTf] and 0.1 mM [Co(III,N)][OTf] added	0.7	3
electrode (↑ background HER, ↓ [NH3])
5	GC(−) instead of BDD(−)	5.2	19
catalyst concentration (↑ [Mo], ↑ [NH3])
6a	0.05 mM [Mo] instead of 0.1 mM [Mo]	4.7	17
7b	0.2 mM [Mo] instead of 0.1 mM [Mo]	11.7	43
catalyst precursor
8	[{(HPNP)Mo(N2)2}2(μ-N2)] instead of [MoBr3(HPNP)]	1.7	6
9	[Mo(N)Br(HPNP)] instead of [MoBr3(HPNP)]	4.6	17
Br− additive (↑ [Br−], ↑ [NH3])
10	[{(HPNP)Mo(N2)2}2(μ-N2)] instead of [MoBr3(HPNP)] and 0.3 mM [Li][Br] added	3.0	10
11	[Mo(N)Br(HPNP)] instead of [MoBr3(HPNP)] and 0.2 mM [Li][Br] added	6.0	22
solvent
12	DME instead of THF	1.1	4
13	2-MeTHF instead of THF	3.1	11
14	MeOH instead of THF	1.4	5

^a1.2 C of charge were passed instead of 2.4 C.

^b4.8 C of charge were passed instead of 2.4 C.

The electrode material also influences the kinetics of competitive HER. The use of a glassy carbon (GC) electrode instead of BDD in the presence of [ColH][OTf] led to a 2-fold increase in current at the catalytically relevant potential (−1.89 V). This trend was reproduced when [LutH][OTf] and [PicH][OTf] were employed, with an HER current response approximately double that obtained with BDD electrodes. Consistently, CPE experiments using [MoBr₃(^HPNP)] and a GC plate electrode (Table 1) resulted in a decreased NH₃ yield (5.2 equiv; 19%) relative to BDD.

We also explored the viability of electrocatalysis using the analogous [Mo^IIIBr₃(^MePNP)] complex. The substituted MePNP ligand has been shown to improve turnover frequency and turnover number during chemically driven N₂R using related Mo⁰ complexes.³¹ The introduction of an electron-donating Me group shifts the reduction potential of [Mo^IIIBr₃(^MePNP)] cathodically by 60 mV with respect to [Mo^IIIBr₃(^HPNP)], as evidenced by cyclic voltammetry (see SI). In this case, the higher HER background expected to be associated with operating at a lower potential (−1.95 V) is potentially offset by the higher TOF of this catalyst, thereby generating 8.1 equiv of NH₃ (30% FE) after only 4 h of CPE. Additionally, a subsequent CPE experiment performed with a reloading of 100 equiv of [ColH][OTf] resulted in a total of 13.0 equiv of NH₃ per Mo atom (FE = 24% overall; Table S1).

Bimolecular steps have shown to play an important role in the mechanism of both chemical N₂R catalysis and electro-chemical N₂ splitting by the [Mo^IIIBr₃(^HPNP)] catalyst.^27,28 Specifically, reduction to [Mo^IBr(^HPNP)] leads to the formation of an N₂-bridged Mo complex, {[(^HPNP)- Mo^I(Br)]₂(μ-N₂)}, which undergoes exergonic N≡N bond cleavage to form [Mo(N)Br(^HPNP)]. We, thus, wondered whether concentration might play a role in enhancing the efficiency of electrocatalytic N₂R. CPE experiments at higher Mo concentrations (0.2 mM; Table 1, entry 7) maintaining the same acid ratio (100 equiv) led to a significant increase in the NH₃ production (11.7 equiv) and FE (43%). Conversely, a decrease of the concentration to 0.05 mM had the opposite effect, with the FE dropping to 17% (4.7 equiv NH₃). These results are consistent with (though do not require) a bimolecular N₂ splitting step being kinetically relevant under our electrocatalytic conditions. Importantly, further increase in the [Mo] concentration led to extensive THF polymerization attributable to higher anodic current passed during the CPE, which imposes practical limitations. A switch to solvents that provide convenient oxidation processes without suffering from polymerization resulted in significantly attenuated electro-catalytic performance (Table 1, entries 12−14), including 2-methyl tetrahydrofuran (2-MeTHF), 1,2-dimethoxyethane (DME), or methanol (MeOH).

Miller and co-workers have previously shown that electro-chemical reduction of [Mo^IIIBr₃(^HPNP)] at −1.89 V can also lead to formation of low valent {[(^HPNP)Mo(N₂)₂]₂(μ-N₂)}.¹⁶ Under acidic conditions, {[(^HPNP)Mo(N₂)₂]₂(μ-N₂)} is proposed to undergo N₂ fixation via a Chatt-type mechanism.^30,32 When we studied {[(^HPNP)Mo(N₂)₂]₂(μ-N₂)} during CPE experiments, we obtained only 1.7 equiv of NH₃ (6% FE; Table 1, entry 8); the addition of 3 equiv of LiBr under similar conditions doubled the yield (to 3 equiv NH₃). Previous observations have suggested that halide salts can generate [Mo(N)X(^HPNP)] from {[(^HPNP)Mo(N₂)₂]₂(μ-N₂)} under catalytic conditions, potentially involving a partial change to an N₂ splitting mechanism and associated improvement of NH₃ yields.²⁷

Within a N₂ splitting scenario, the nitride complex [Mo(N)Br(^HPNP)] should be a relevant intermediate, which prompted us to independently evaluate its competence as an electrocatalyst. CPE experiments with 0.1 mM [Mo(N)Br-(^HPNP)] under our standard conditions led to the catalytic formation of NH₃ (4.6 equiv) with a FE of 17% (Table 1, entry 9). While this yield is lower than those obtained with [Mo^IIIBr₃(^HPNP)] as the added catalyst, the addition of 2 equiv of [Li][Br] improved the FE of [Mo(N)Br(^HPNP)] to 22% (6.0 equiv NH₃). This observation might be attributable to an adverse anion exchange equilibrium with OTf⁻, which would be partially prevented at higher [Br⁻], or instead to a required generation of a higher valent [MoBrn(^HPNP)] (n = 2 or 3) species.

Given the supported intermediacy of a nitride [Mo(N)Br-(^HPNP)] complex, we further evaluated its electrochemical behavior (Figure 3A−C). CV of [Mo(N)Br(^HPNP)] in 100 mM [Li][NTf₂] THF solution showed a reversible Mo^V/IV oxidation wave at −1.2 V followed by a Mo^IV/III reduction peak at approximately −2.5 V (Figure 3A). The highly cathodic potential contrasts with the milder potential required for electrocatalysis (approximately −1.9 V). The addition of [ColH][OTf] to this solution resulted in the attenuation of the Mo^V/IV redox couple and the appearance of an electrocatalytic wave similar to that found with [Mo^IIIBr₃ (^HPNP)]. This result suggests a protonation equilibrium process via formation of the imido complex [Mo(NH)Br(^HPNP)]⁺, reduction of which takes place at around −1.8 V, as evidenced by an increase in the current, thereby enabling the system to enter the N₂R catalytic cycle. A more anodic scan reveals the appearance of a small redox feature at 0 V that we associate with the oxidation of [Mo(NH)Br(^HPNP)]⁺ to [Mo(NH)Br(^HPNP)]²⁺ (Figure 3B). This behavior follows a square scheme where the oxidation of [Mo(NH)Br(^HPNP)]⁺ promotes deprotonation to form [Mo(N)Br(^HPNP)]⁺ (Figure 3D). Conversely, the reduction of this cationic nitride to [Mo(N)Br(^HPNP)] is followed by protonation to produce [Mo(NH)Br(^HPNP)]⁺.

Figure 3.

(A) CV of [Mo(N)Br(^HPNP)] in the presence and absence of [ColH][OTf], and comparison with just [ColH][OTf] or [MoBr₃(^HPNP)] in the presence of [ColH][OTf]. (B) CV of [Mo(N)Br(^HPNP)] in the presence of either [ColH][OTf] or [LutH][OTf]. (C) Variable scan rate CV analysis of [Mo(N)Br(^HPNP)] in the presence of [ColH][OTf]. (D) Square scheme mechanism showing the protonation and reduction processes associated to [Mo(N)Br(^HPNP)]. Note: CVs were performed at 100 mV/s (unless otherwise noted) in a 0.1 [Li][NTf₂] THF solution using a BDD disk working electrode.

This mechanism correlates well with experiments employing the stronger acid [LutH][OTf]. Under such conditions, the protonation equilibrium is further shifted toward the formation of [Mo(NH)Br(^HPNP)]⁺, thereby resulting in a larger oxidation peak at 0 V (Figure 3B). In the subsequent cathodic scan, the corresponding reduction peak showed significantly lower current, and the [Mo(N)Br(^HPNP)]^+/0 reduction became clearer. Variable scan rate CV experiments allow for the calculation of the kinetic constant for the protonation of [Mo(N)Br(^HPNP)]: k_PT = ~3.3 M⁻¹ s⁻¹ with [ColH][OTf] (see Figure 3C and the SI). The rapid kinetics for the protonation of [Mo(N)Br(^HPNP)] and the facile reduction of [Mo(NH)Br(^HPNP)]⁺ at the applied potential (−1.89 V) suggest a minor influence of these steps in the overall kinetics and efficiency of N₂R.

We have previously shown that the use of electrolytes containing Li⁺ as a Lewis acid enhances the efficiency of NH₃ production, presumably by activation of key N₂R intermediates.^7,33 We, thus, questioned whether [Li][NTf₂] plays a similar role in the present catalytic system. A substitution of Li⁺ with the commonly used cation tetrabutylammonium TBA⁺ in our electrocatalytic set up led to greater than stoichiometric amounts of NH₃ (2.8 equiv; Table 2, entry 1). The use of [K][NTf₂] as the electrolyte yielded catalytic, albeit lower, amounts of NH₃ (5.0 equiv; Table 2, entry 3) compared with [Li][NTf₂]; the difference between K⁺ and Li⁺ could correlate with their different Lewis acidities. [Na][NTf₂] performed more poorly and afforded only 1.9 equiv of NH₃ at 7% FE (Table 2, entry 2). Na⁺ might serve as a more efficient bromide abstracting agent to precipitate [Na][Br], thereby affecting the availability of Br⁻ during the catalysis; the latter can impact N₂R efficiency (vide supra).^34,35

Table 2.

Screening of electrolytes for CPE experiments

entry	electrolyte	NH3 (equiv/Mo)	Faradaic efficiency (%)
1	[TBA][NTf2]	2.8	10
2	[Na][NTf2]	1.9	7
3	[K][NTf2]	5.0	18

We next studied how the presence of Li⁺ might impact the efficiency of the electrochemical N₂ splitting reaction as a potential key step during electrocatalysis. Subjection of 0.1 mM [Mo^IIIBr₃ (^HPNP)] to a CPE at −1.89 V using a BDD plate electrode in a 100 mM [Li][NTf₂] THF solution afforded [Mo(N)Br(^HPNP)] in ~30% yield, as calculated via CV analysis after CPE (Figure 4 and SI); no {[(^HPNP)Mo-(N₂)₂]₂(μ-N₂)} was observed. When [TBA][NTf₂] was used, instead, at similar concentrations, neither of these reduced Mo species were detected. These results suggest the possibility that a Lewis acid such as Li⁺ may aid in lowering the barrier for the formation of key N₂-bridged species and/or stabilize the [Mo(N)Br(^HPNP)] product of the N₂ splitting reaction. Changes in the UV−vis spectrum of [Mo(N)Br(^HPNP)] upon addition of [Li][NTf₂] support this idea (see SI). Previous work has shown that [Mo(N)Br(^HPNP)] can also be produced with TBA electrolytes ([TBA][PF₆]) but at significantly higher [Mo^IIIBr₃ (^HPNP)] loadings (4 mM),¹⁶ presumably because of the influence of concentration on the bimolecular reaction. Although these conditions are not suitable for our electro-catalytic system, the results discussed herein show that the presence of Li⁺ facilitates this cleavage reaction at lower [Mo] concentrations.

Figure 4.

(A) Electrochemical reduction of [MoBr₃(^HPNP)] to generate [Mo(N)Br(^HPNP)] via N₂ splitting. (B) CVs after the electrochemical N₂ splitting reaction using the electrolytes [Li]-[NTf₂], [TBA][NTf₂], or [TBA][PF₆], and comparison with an authentic sample of [Mo(N)Br(^HPNP)].

To conclude, herein we have shown that Nishibayashi et al.’s [Mo^IIIBr₃(^HPNP)] N₂R catalyst system can be adapted to electrocatalysis on careful consideration of the reaction medium. Electrocatalytic N₂R by this system is gated by an initial ET step at −1.80 V that sets the needed applied bias for the observed catalysis. This is distinct from PCET-mediated N₂R electrocatalysis, where an anodically shifted potential can be used because of a rate-limiting PCET step that instead sets the required bias of the system. We have further shown how the nature of the working electrode, the acid, the electrolyte, and the catalyst concentration contribute to the observed N₂R electrocatalysis in the present system by enhancing the relative kinetics for N₂R versus background HER. In particular, the use of a BDD electrode, together with a relatively weak pyridinium acid ([ColH][OTf]), allows minimization of the influence of competitive HER. The use of an electrolyte with a hard Lewis acid, such as Li⁺, also increases NH₃ yields, potentially because of the stabilization and/or activation of key N₂R intermediates. These conditions contrast those often employed in small molecule electrocatalysis (e.g., employing GC electrodes and [TBA][X] electrolytes). We anticipate the findings disclosed here will aid in the future development of more efficient N₂R electrocatalysts using coordination complexes.