Upgrading of nitrate to hydrazine through cascading electrocatalytic ammonia production with controllable N-N coupling

Abstract

Nitrogen oxides (NO_x) play important roles in the nitrogen cycle system and serve as renewable nitrogen sources for the synthesis of value-added chemicals driven by clean electricity. However, it is challenging to achieve selective conversion of NO_x to multi-nitrogen products (e.g., N₂H₄) via precise construction of a single N-N bond. Herein, we propose a strategy for NO_x-to-N₂H₄ under ambient conditions, involving electrochemical NO_x upgrading to NH₃, followed by ketone-mediated NH₃ to N₂H₄. It can achieve an impressive overall NO_x-to-N₂H₄ selectivity of 88.7%. We elucidate mechanistic insights into the ketone-mediated N-N coupling process. Diphenyl ketone (DPK) emerges as an optimal mediator, facilitating controlled N-N coupling, owing to its steric and conjugation effects. The acetonitrile solvent stabilizes and activates key imine intermediates through hydrogen bonding. Experimental results reveal that Ph₂CN* intermediates formed on WO₃ catalysts acted as pivotal monomers to drive controlled N-N coupling with high selectivity, facilitated by lattice-oxygen-mediated dehydrogenation. Additionally, both WO₃ catalysts and DPK mediators exhibit favorable reusability, offering promise for green N₂H₄ synthesis.

Nitrogen oxides are vital in the nitrogen cycle and renewable chemical synthesis, but selective conversion to multi-nitrogen products via precise N-N bond formation is challenging. Here, the authors report a two-step electrochemical process that achieves an impressive 88.7% selectivity for hydrazine production using a WO₃ catalyst with diphenyl ketone as a mediator.

Introduction

Manufacturing nitrogenous chemicals and fuels from renewable nitrogen sources using clean energy without producing harmful by-products could contribute to reduce CO₂ emissions and build our carbon-neutral and sustainable society^1–3. Hydrazine (N₂H₄) is widely used as a reductant, anti-corrosive agent, explosives, antioxidants, and fuel for satellites and rockets^4–6. Current industrial process for N₂H₄ production relies heavily on unsustainable chemical oxidants, including NaClO and H₂O₂ derived from pure O₂ and Cl₂ through harsh conditions (Fig. 1a). NH₃ substrates produced from energy-intensive Haber-Bosch process are used as the nitrogen source. In addition, many harmful by-products are also generated during traditional N₂H₄ production, raising notable environmental issues^7,8. There is a pressing demand for the exploration of innovative and sustainable methods for N₂H₄ synthesis.

Fig. 1. Synthesis of N₂H₄.

a Schematic illustration of the traditional N₂H₄ synthesis method. b Recently emerging NO_x-participating electrochemical reactions. c Upgrading of NO_x to N₂H₄ through electrocatalysis demonstrated in this work.

The excessive emission of nitrogen oxides (NO_x) poses environmental and human health concerns, but they also serve as renewable nitrogen sources for the synthesis of value-added chemicals driven by clean energy^9,10. For instance, NO_x could yield ammonia (NH₃) through electrochemical reduction (Fig. 1b)^11,12. NO_x electroreduction is a multi-electron transfer process involving various intermediates, most of which are unstable and fleeting^13–15. Consequently, only two thermodynamically stable products, N₂ and NH₃, are usually generated with satisfactory selectivity. Achieving selective conversion of NO_x to N₂H₄ remains a challenge, as only trace amount of N₂H₄ is detected after electrolysis due to excessive N-N coupling, leading to the formation of N₂. This excessive N-N coupling during direct NO_x reduction is attributed to the thermodynamically spontaneous nature of N-N coupling based on *NO intermediates, along with the reduction of multi-nitrogen intermediates like *NONH₂, *N₂O, and *H₂N₂O₂ to N₂^16,17. Therefore, it is important to develop advanced catalytic systems to precisely control the N-N coupling, leading to the enhancement of NO_x-to-N₂H₄ efficiency.

Designing an effective catalytic system requires comprehensive consideration of pathway, solvent, and catalyst^18–20. Various nitrogenous intermediates produced during NO_x electrolysis can either desorb directly from the catalyst surface to generate target products or couple with other intermediates^21,22. Insufficient N-N coupling activity results in generating products with single nitrogen atom, whereas excessive N-N coupling activity leads to the formation of N≡N bonds. Hence, designing reaction pathways with mediator molecules guided by functional group protection methods in organic synthesis may be feasible. It is noteworthy that some useable nitrogenous intermediates, such as imine, are unstable in water, so it is important to choose a suitable solvent that could stabilize intermediates for controllable N-N coupling. Moreover, developing efficient heterogeneous catalysts is crucial for adsorbing intermediates and transferring charge carriers, making high reactivity and selectivity possible for N₂H₄ formation.

Herein, we report the N₂H₄ production through NO_x upgrading with controllable N-N coupling, which involves the electrochemical reduction of NO_x (with NO₃⁻ as an example) to NH₃, followed by ketone-mediated NH₃ oxidation for selective single N-N coupling. It was discovered that N₂H₄ was successfully synthesized with a yield of 98.5% and a Faradic efficiency (FE) of 95.6% on WO₃ catalysts. In particular, an overall selectivity of 88.7% was achieved for the conversion of NO_x to N₂H₄. Through a combination of controlled experiments, in situ characterizations, and theoretical calculations, we investigate mechanistic insights into the ketone-mediated N-N coupling process. Diphenyl ketone (DPK) was found to be the mediator for moderating N-N coupling processes owing to its unique steric and conjugate effects, which could condense with NH₃ to form imines to enable single N-N coupling. Acetonitrile (CH₃CN) solvent was used to stabilize and activate key imine intermediates through hydrogen bonding. Furthermore, we have identified the Ph₂CN* intermediates, formed on WO₃ catalysts, as pivotal monomers driving controlled N-N coupling with high selectivity. Additionally, both the WO₃ catalyst and the DPK mediator could be reused.

Results

Upgrading of NO₃⁻ to N₂H₄

As illustrated in Fig. 1c, we have conducted a proof-of-concept experiment aimed at converting NO_x to N₂H₄. This was achieved through a series of steps (Supplementary Fig. S1): first, electroreduction of NO_x to NH₃; second, ketone-mediated N-N coupling of NH₃; and finally, hydrolysis of N-N products with water to yield N₂H₄. The electrocatalytic reduction of NO_x to NH₃, utilizing aqueous NO₃⁻ as an example, was performed using an H-type cell, employing commercial oxide-derived copper (OD Cu) as catalyst (Supplementary Fig. S2). The NH₃ product obtained from electroreduction was detected using the well-established salicylate method, as demonstrated in Supplementary Fig. S3²³. It should be noted that this setup is also adaptable for the reduction of other NO_x species, including NO, NO₂⁻, and their mixtures^24–26. Subsequently, the NH₃ was transferred into CH₃CN by air stripping method for further applications. The analyses revealed a high efficiency in the conversion of NO_x to NH₃, establishing NO_x-derived NH₃ as a versatile platform molecule for the synthesis of various nitrogenous chemicals.

In the following, the electrooxidation of NH₃ was carried out in a single cell with NH₃ and DPK in CH₃CN. Methanol (5 vol%) was also added to the solution as sacrificial agent, which could produce H₂ on cathode to balance the whole electrolysis cell. DPK was used as the mediator to control single N-N coupling, yielding benzophenone azine (BPA) due to its unique structure. Herein, we used WO₃/carbon paper as the working electrode due to its good ability to dehydrogenation in electrooxidation of various molecules^27–29, and the WO₃ catalyst was synthesized via a straightforward hydrothermal method employing Na₂WO₄ as the W precursor³⁰. Various catalysts commonly employed in electrocatalytic oxidation reactions, such as Pt/C, Ru/C, Pd/C, MnO₂, Fe₂O₃, CoO, NiO, and CuO, were screened. Noble metal-based catalysts exhibited poor catalytic performance for the oxidation of NH₃ to BPA, with Ru/C showing no catalytic activity. In contrast, transition metal-based catalysts (MnO₂, Fe₂O₃, CoO, NiO, CuO) can catalyze the reaction, albeit with inferior performance compared to WO₃. WO₃ exhibited the highest conversion of NO_x-derived NH₃ to BPA after electrolysis, as illustrated in Supplementary Fig. S4. As a result, WO₃ was identified as the optimal catalyst for N₂H₄ synthesis because of its numerous active sites and suitable binding affinity for intermediates^31–33. The molecular weight of 360.1 determined by gas chromatography-mass spectrometry (GC-MS) showed that the product has the molecular formula of C₂₆H₂₀N₂ (Supplementary Fig. S5). In addition, the selectivity of DPK-mediated N-N coupling was as high as >99.9%, since no detectable by-product was observed in GC-MS. Moreover, all the peaks in ¹H NMR and ¹³C NMR spectra (Supplementary Fig. S5b, c) matched well with the standard samples, confirming the formation of the BPA product. Under the optimal potential of 2.1 V vs Ag/Ag⁺, the FE of BPA was as high as 95.6%, with a high yield of 98.5% and selectivity of >99.9%. No detectable H₂ was produced from the surface of the anode according to GC. By employing K¹⁵NO₃ as the reactant, we detected a molecular ion peak of the product at 362.1 (Supplementary Fig. S5e), precisely matching the calculated molecular weight of C₂₆H₂₀¹⁵N₂. This finding confirmed that the nitrogen source in BPA was derived from nitrate.

Afterwards, we investigated the hydrolysis of BPA with acid aqueous solution to produce N₂H₄ (Supplementary Fig. S6). The quantification of N₂H₄ was carried out through both the colorimetric technique and ¹H NMR using standard curves in Supplementary Fig. S7³⁴. According to the results from these two methods, the hydrolysis of BPA to N₂H₄ was also very efficient since the selectivity of N₂H₄ was up to 99.9%. After recrystallization of N₂H₄-based products, we could obtain N₂H₄·HCl and (N₂H₄)₂·H₂SO₄ products with a yield of >60%, which exhibited similar X-ray diffraction (XRD) patterns with commercial standard samples (Supplementary Fig. S8). As a result, this tandem process exhibits remarkable efficiency, and the overall selectivity of NO_x-to-N₂H₄ was 88.7% (Supplementary Fig. S9). In contrast, direct NO_x reduction and NH₃ oxidation without the DPK mediator yielded negligible amounts of N₂H₄ after electrolysis, underscoring the necessity of designing this tandem pathway for N₂H₄ synthesis. In addition, DPK-mediated NH₃ oxidation can directly convert NH₃ to N₂H₄, offering an electrocatalytic route for N₂H₄ production compared to the traditional NH₃ oxidation pathway.

N-N coupling through lattice-oxygen-mediated dehydrogenation

The electrocatalytic N-N coupling performance using NO_x-derived NH₃ and DPK in CH₃CN solvent was evaluated through electrolysis conducted over WO₃ catalyst within a voltage range of 1.2 V to 2.2 V vs Ag/Ag⁺ for 10 h (Supplementary Fig. S10). Figure 2a illustrates that the BPA selectivity remained consistently ~100% across all applied potentials. At lower potentials, the efficiency of N-N coupling was low, resulting in lower NH₃ conversion. At 2.0 V, NH₃ achieved a conversion rate to BPA as high as 98.5%, and the overall cell voltage was found to be 4.1 V. As is presented in Fig. 2b, the FE of BPA across applied potentials ranging from 1.2 V to 2.0 V consistently stands at ~95%. However, when the voltages were higher than 2.0 V, the NH₃ conversion was not notably reduced, but the FE of BPA decreased to <75% due to organic solvent molecule oxidation at high potentials. Furthermore, the WO₃ catalyst demonstrated exceptional stability, being reusable for over five cycles without notable performance decline. After being employed five times consecutively at optimal 2.0 V, the WO₃ catalyst maintained a stable current during electrolysis, achieving an NH₃ to BPA conversion rate of 99.2% and a BPA FE of 94.7% (Supplementary Fig. S11).

Fig. 2. Electrochemical performance.

a Conversion and selectivity of NH₃ to BPA on WO₃ catalysts under different applied potentials. b FE of BPA product and total current density of WO₃ catalyst under different applied potentials. ATR-FTIR spectra conducted at the surface of WO₃ catalyst under different applied potentials during BPA formation in the range of c 1750–1610 cm⁻¹, d 1610–1550 cm⁻¹, and e 1500–1400 cm⁻¹. f ¹H NMR spectra of products obtained from DPK-mediated N₂H₄ synthesis. g N₂H₄ yield from BPA hydrolysis detected via UV-Vis and ¹H NMR spectroscopy. All the experiments were without experiment was without iR compensation. Values are means, and error bars indicate s.d. (n = 3 replicates).

The evolution of intermediates during BPA formation was further studied utilizing in situ ATR-FTIR spectroscopy. The peaks at 1674 cm⁻¹ (Fig. 2c), 1597 cm⁻¹ (Fig. 2d), 1570 cm⁻¹ (Fig. 2d), and 1450 cm⁻¹ (Fig. 2e) corresponded to the H₂O produced after C-N condensation between NH₃ and DPK, the active C-N bonds in Ph₂CN* intermediates, the DPK mediator consumed during the reaction, and the diphenyl ketone imine (DPK-I) intermediates respectively. Moreover, the observed signals of H₂O and Ph₂CN*/DPK-I (C=N) were intensified during electrolysis, indicating an increased concentration of these intermediates in the solution, while the signal corresponding to DPK (C=O) and DPK-I (N-H) demonstrated reverse peaks, confirming its consumption. Thereinto, as DPK-I underwent N-N coupling to form BPA, the N-H signal gradually diminished. The C=N structure co-existing in both DPK-I and BPA contributed to a gradual rise in the C=N signal. As a consequence, the dehydrogenation of DPK-I generated from the condensation between NH₃ and DPK to Ph₂CN*, followed by the dimerization of Ph₂CN*, was essential to BPA formation, and the catalysis system should boost these steps.

The kinetics of BPA formation on the WO₃ catalyst was subsequently examined. Supplementary Fig. S12a displays the recorded yields of BPA in various reaction times conducted at the optimal electrolysis conditions, demonstrating the near-complete conversion of NH₃ into BPA after electrolysis. The rate constant k₁ of the coupling of NH₃ with DPK to form DPK-I was notably higher than the k₂ of N-N coupling between DPK-I intermediates, indicating a more favorable C-N condensation step. Consequently, the formation of BPA could be effectively approximated as a pseudo first-order reaction (Supplementary Fig. S12b), with the N-N coupling step identified as the rate-determining step (RDS). BPA was also produced by direct electrooxidation of commercial DPK-I, which also indicated that BPA was formed through N-N coupling of DPK-I intermediates. Furthermore, the findings from the linear sweep voltammetry (LSV) investigation in Supplementary Fig. S13 suggested the sluggish nature of the DPK reaction in the absence of NH₃, further supporting the preference for N-N coupling based on DPK-I intermediates.

In order to hydrolyze BPA and regenerate the BPK mediator, we dissolved the BPA product in an equivalent acid solution to generate the final N₂H₄ product. The resulting N₂H₄ product displayed a consistent unimodal peak at 3.3 ppm in ¹H NMR, aligning with the N₂H₄ standard sample, as is presented in Fig. 2f. Furthermore, the formation of N₂H₄ products was confirmed through GC-MS exhibiting a m/z of 33.00, which is the same as N₂H₄ standard and ultraviolet-visible (UV-vis) absorbance spectra (Supplementary Fig. S14). Interestingly, neither the ¹H NMR signal of N₂H₄ products nor the color development were observed in UV-vis spectra in the electrolyte obtained from direct NO_x electrocatalytic reduction. This reiterated the necessity of BPA-mediated N-N coupling processes based on NH₃ as a direct nitrogen source during N₂H₄ generation. Furthermore, the electrochemical oxidation of DPK-I resulted in the efficient formation of Bisphenol A (BPA), thereby providing additional confirmation of the essential role played by DPK-I intermediates in BPA formation. Quantitative analysis of the N₂H₄ products, based on the integrated area of the ¹H NMR signal peak and UV-vis absorbance, revealed a very high NH₃ to N₂H₄ selectivity of >97.5%. Moreover, GC-MS, ¹H NMR, and ¹³C NMR analyses shown in Supplementary Fig. S15 demonstrated that over 99% of the DPK could be extracted and recycled for subsequent N₂H₄ synthesis. The initial DPK feed utilized in the electrochemical reactions retained its efficacy for over 5 cycles without a considerable decrease in BPA yield. The results underscored the essentiality, efficiency, and absence of by-products in the designed reaction pathways.

XRD patterns of the WO₃ catalyst (Supplementary Fig. S16) confirmed the formation of WO₃ phase (PDF#32-1395). Scanning Electron Microscopy (SEM) (Supplementary Fig. S17) and transmission electron microscopy (TEM) (Supplementary Fig. S18) images illustrated the nanorod morphology of the obtained WO₃ electrocatalyst. High-resolution TEM (HR-TEM) images revealed consistent lattice fringes corresponding to (200) planes of WO₃ with a spacing of 0.36 nm (Fig. 3a). Energy-dispersive X-ray spectroscopy (EDS) elemental mapping (Supplementary Fig. S19) displayed a homogeneous distribution of W and O over the entire architectures. X-ray absorption near edge spectroscopy (XANES) and extended X-ray absorption fine structure (EXAFS) characterizations were carried out to get detailed structural information. The normalized W L₃-edge XANES spectra (Fig. 3b and Supplementary Fig. S20) indicated an overlap between the absorption edge of the as-prepared WO₃ catalyst and WO₃ standard sample, suggesting the presence of W⁶⁺ species^27,35. As shown in Fig. 3c, Fourier transforms (FT) of the W L₃-edge EXAFS spectra suggested the presence of W-O coordination in the WO₃ electrocatalysts, with a bond length of 1.38 Å, closely resembling the WO₃ standard (1.37 Å). These findings were corroborated by X-ray photoelectron spectroscopy (XPS) (Supplementary Fig. S21). These results signify the successful synthesis of the WO₃ catalyst. More positively, the morphology and crystal structure of WO₃ catalyst did not change significantly after stability test (Supplementary Figs. S21–25). The structural features and morphology of the NiO catalyst in the control group are depicted in Supplementary Fig. S26 in the supporting information.

Fig. 3. N-N coupling on WO₃ catalysts.

a HR-TEM image of the WO₃ catalyst. b The XANES spectra of XANES spectra and c W L₃-edge FT EXAFS spectra of the WO₃ catalyst, along with standard samples of W foil and WO₃, highlighting W-O and W-W coordination shells. d Conversion of NH₃ on NiO, WO₃, and W catalyst. Values are means, and error bars indicate s.d. (n = 3 replicates). e In situ electrochemical Raman spectra over WO₃ during N₂H₄ synthesis. f Ratio of the NH₃ conversion on NiO, WO₃, and W catalyst before and after the poisoning experiments using Li⁺ and SCN⁻. g Illustration of the catalytic active sites during the poisoning experiments. h Illustration of N-N coupling on WO₃ catalyst through lattice-oxygen-mediated dehydrogenation.

The coordination structure of W-O active sites in the WO₃ catalyst is crucial for overcoming the limitations of poor selectivity in the electrosynthesis of N₂H₄^34,36. WO₃ can be rapidly and reversibly doped with hydrogen to form H_xWO₃ phase, significantly influencing reaction barriers and pathways, thereby enhancing catalytic performance^37,38. A comparative analysis of the performance of NiO, WO₃, and W (derived from WO₃ through H₂/Ar treatment) catalysts under optimal conditions was undertaken. Despite maintaining consistent NH₃ to BPA selectivity, both the NiO and W catalysts demonstrated lower NH₃ conversion to BPA in comparison to WO₃ (Fig. 3d). This suggests that the presence of the W-O structure in the WO₃ catalyst is crucial for efficient BPA formation. The electrochemical behavior of the WO₃ catalyst was further examined through in situ Raman spectroscopy measurements. Figure 3e illustrates the characteristic peaks at 274.2 cm⁻¹, 327.9 cm⁻¹, 719.6 cm⁻¹, and 890.8 cm⁻¹, which can be associated with surface W-O bonds of WO₃^37–39. These peaks gradually diminished with an applied potential ranging from 1.0 V to 2.0 V. This phenomenon could be attributed to the interaction of N-H bond of DPK-I with the WO₃ surface, leading to the H intercalation into the WO₃ lattice and the formation of H_xWO₃ species on the catalyst surface^40–42. H_xWO₃, recognized as an active species present on the catalyst surface, could not drive bulk phase transitions in the WO₃ catalyst. Therefore, post-electrolysis tests verify that the structure of the used WO₃ remained comparable to its initial state. The generation of the active intermediate Ph₂CN* from DPK-I could supply subsequent N-N coupling. In contrast, the surface of the NiO catalyst showed no noticeable active species phase with increasing potential due to the reliance of highly active NiOOH species common in electrooxidation in aqueous solution (Supplementary Fig. S27). As a result, electrooxidation reaction on NiO was not favorable due to the limited availability of H₂O in CH₃CN solvents, hindering the generation of active NiOOH site from NiO and H₂O^43–45. DPK-I can only generate active species on the surface for N-N coupling, and no lattice species were involved. Therefore, the high electrochemical performance based on WO₃ was attributed to its ability to promote the dehydrogenation process of DPK-I under the mediation of lattice oxygen in coordination with CH₃CN solvent.

Li⁺ and SCN⁻ poisoning experiments were conducted on WO₃, NiO, and W catalysts to elucidate the significance of the W-O lattice structure in BPA formation³⁸. As illustrated in Fig. 3f, after Li⁺ poisoning treatment, NiO and WO₃ catalysts experienced a substantial decrease in activity attributable to the strong coordination between Li⁺ and lattice oxygen in NiO and WO₃ (Fig. 3g). Conversely, the W catalyst, lacking lattice oxygen, exhibited negligible activity decline, underscoring the role of lattice oxygen as a crucial site for BPA formation. Subsequent poisoning of the catalysts with SCN⁻, known for its coordination with metal sites, revealed that WO₃ and W catalysts displayed significantly reduced performance, while NiO exhibited a comparable performance. This observation suggests the importance of W sites in the WO₃ catalyst. Consequently, the conversion from NH₃ to BPA mediated by DPK followed the NH₃→DPK-I→Ph₂CN*→BPA pathway, with the generation of Ph₂CN* significantly facilitated by lattice-oxygen-mediated dehydrogenation on WO₃ (Fig. 3h).

Ketone mediators to control N-N coupling

The fundamental mechanism by which ketones could facilitate single N-N coupling was illustrated in Fig. 4a, showing the stepwise transformation of NH₃ to imine and ultimately N₂H₄. NH₃ molecule contains three reactive N-H bonds, and direct oxidation of these bonds results in the undesired formation of thermodynamically stable N₂ and NO_x, presenting a significant challenge in achieving exclusive N-N coupling to generate N₂H₄. Therefore, mitigating the reactivity of NH₃ becomes crucial in the pursuit of N₂H₄ synthesis. Under optimal conditions, ketones could undergo condensation with NH₃ to yield imines. N atoms could form C=N bonds, while only one C-H bond remains active within the imine molecules. Consequently, ketones may act as mediators in the controlled formation of the N-N bond, utilizing imines as a key intermediate. More importantly, C=N bonds could be simply cleaved via hydrolysis⁴⁶, providing a viable pathway for the ultimate formation of N₂H₄ through the upcycling of NO_x.

Fig. 4. Ketone mediator.

a Illustration of the molecular mechanism of ketone-mediated single N-N coupling compared with direct NH₃ oxidation. b Reaction pathways and calculated free energy diagrams of the condensation of DMK, MPK, and DPK with NH₃ to DMK-I, MPK-I, and DPK-I, respectively. c Electrostatic potentials of DMK-I, MPK-I, and DPK-I.

After establishing the basic methodology to produce N₂H₄, we further conducted preliminary screening of ketone compounds that could facilitate the single N-N coupling process. To investigate the influence of ketone mediator molecules with varying electronic and steric properties (Fig. 4b and Supplementary Fig. S28), we examined the energy changes involved in the condensation of three ketones with NH₃, such as dimethyl ketone (DMK), methyl phenyl ketone (MPK) and DPK. This led to the formation of imines, namely dimethyl ketone imine (DMK-I), methyl phenyl ketone imine (MPK-I), and DPK-I, following the removal of water molecules through the corresponding intermediates (supplementary data 1). The reaction transition state energy barriers from (1) to (2) for the ketones of DMK, MPK, and DPK were approximately the same, 39.6, 39.1, and 39.4 kcal mol⁻¹, respectively. However, for the reaction process from (2) to (3), the transition state energy barriers decreased gradually to 52.0, 48.2, and 44.4 kcal mol⁻¹, respectively, indicating that the activity of the DMK-I, MPK-I, and DPK-I intermediates progressively increased. Hence, DPK exhibited a higher capacity for generating DPK-I due to the boosted dehydration process, which could serve as the mediator for single N-N coupling.

Furthermore, the electrostatic potentials (φ_max) on the N atom side exhibited significant disparities for imines substituted with different groups: φ_max, DPK < φ_max, MPK < φ_max, DMK < φ_max, NH₃ (Fig. 4c). This discrepancy could suggest that DPK-I has the capability to redistribute the charge density on the N atom of the imine group, thereby enhancing molecular stability. This effect could be attributed to the robust conjugation of phenyl groups (Ph), which surpasses that of methyl groups (Me). Consequently, among the three selected ketone candidates, DPK emerges as a viable mediator for moderating the N-N coupling reaction in NH₃ oxidation to N₂H₄, which was achieved through the formation of DPK-I intermediates, protecting two extraneous N-H sites with imine groups of optimal stability. The superior stability of DPK-I compared to DMK-I and MPK-I was further confirmed by GC-MS analysis (Supplementary Fig. S29). This analysis revealed that only DPK-I was detectable after mixing the corresponding ketone and NH₃.

The solvent effect for DPK-I intermediates

As illustrated in Fig. 5a, the condensation of DPK and NH₃ can produce DPK-I for further N-N coupling, which demands a suitable solvent to stabilize active DPK-I intermediates. We examined the stability of DPK-I intermediates in different types of solvents, taking H₂O and CH₃CN as examples. Commercial DPK-I was mixed with each solvent in a 1:1 volume ratio and stirred for 12 h. GC-MS in Fig. 5b and Supplementary Fig. S30 analysis showed that DPK-I could be stable in CH₃CN but underwent decomposition into DPK and NH₃ in an aqueous environment. CH₃CN is a typical polar aprotic solvent and could not break the imine bonds in DPK-I intermediates, while H₂O is a highly polar solvent and participates in the hydrolysis of DPK-I to DPK and NH₃. Consequently, DPK-I intermediates were stable in CH₃CN-based non-aqueous solvents, which favors the formation of N-N bonds.

Fig. 5. Solvent effect.

a Illustration of the behavior of DPK-I intermediates in H₂O, CH₃CN, and CH₂Cl₂ solvents. b GC-MS curves of DPK-I treated in H₂O and CH₃CN solvents. c ¹H NMR spectra and d ATR-FTIR of DPK-I in CH₂Cl₂ and CH₃CN solvents. e Electrochemical performance of NH₃ to BPA transformation using CH₃CN and CH₂Cl₂ as the solvent. All the experiments were without experiment was without iR compensation. Values are means, and error bars indicate s.d. (n = 3 replicates). f Free energy of various intermediates generated during N-N coupling with and without CH₃CN.

The interaction between CH₃CN and the imine functional groups in DPK-I potentially activated N-H bonds, thereby promoting N-N coupling. To assess these interactions, we conducted ¹H NMR spectroscopy to analyze the DPK-I in CH₃CN and dichloromethane (CH₂Cl₂) solutions (Fig. 5c). The chemical shift of protons in the N-H groups of DKP-I exhibited shifts of 9.27 ppm to 8.63 ppm, upon dissolution in CH₂Cl₂ and CH₃CN solvents, respectively. The ¹H NMR results align with the attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectra of DPK-I in diverse solvents (Fig. 5d), revealing notable shifts in peaks associated with imine groups in DPK-I. These shifts provide further evidence of hydrogen bonding formation between CH₃CN and the imine functional groups. In addition, under optimized conditions, electrochemical oxidation of NH₃ in the presence of DPK mediators was performed in CH₃CN. As depicted in Fig. 5e, the NH₃-to-BPA transformation exhibited significantly higher conversion and FE in CH₃CN than in CH₂Cl₂. Specifically, the NH₃ conversion rate was 16 times greater in CH₃CN than in CH₂Cl₂, accompanied by a corresponding 7-fold increase in FE upon transitioning from CH₂Cl₂ to CH₃CN as the solvent. As a result, while we acknowledge the intrinsic complexity of solvent systems, the existence of hydrogen bond interaction between CH₃CN and DPK-I intermediates enhances the activation of N-H bonds, facilitating the N-N coupling of DPK-I intermediates, thus promoting BPA formation.

We then investigated the reaction pathway from DPK-I to BPA with the assistance of CH₃CN molecules, utilizing a molecular model of R₂CNH (R = CH₃) on the surface of the WO₃(200) model. The construction of the catalyst model relied on the aforementioned results concerning the characterizations of WO₃ catalysts (supplementary data 2). To simplify the calculation process, we reduced the phenyl group in DPK-I to a methyl group. Density Functional Theory (DFT) analysis shown in Fig. 5f and Supplementary Fig. S30 revealed that in the presence of CH₃CN molecules, the energy required for R₂CNH* dehydrogenation to R₂CN* decreased by 0.59 eV compared to the absence of CH₃CN molecules, and the Gibbs free energy for the single N-N coupling step was reduced by 0.65 eV in the presence of CH₃CN. Furthermore, the energy barrier along the reaction path with CH₃CN was generally smoother than without CH₃CN. Hence, DFT calculations could elucidate the enhanced performance of DPK-I to BPA with the assistance of CH₃CN solvent.

Discussion

In summary, we have demonstrated a strategy for green N₂H₄ synthesis by using NO_x as a renewable nitrogen source through innovative electrochemical methodology. Our approach involved electrochemical NO_x conversion to NH₃, followed by DPK-mediated NH₃ to N₂H₄ conversion, achieving an exceptional N₂H₄ yield of 98.5% with a remarkable NO_x-to-N₂H₄ selectivity of 88.7%. Detailed characterization, control experiments, and theoretical calculation revealed insights into the intricacies of controlled N-N coupling. The efficacy of DPK as a mediator in regulating N-N coupling, complemented by CH₃CN solvents stabilizing pivotal DPK-I intermediates, underscored the significance of precise mediator and solvent selection. Furthermore, Ph₂CN* intermediates on WO₃ catalysts played a pivotal role in driving lattice-oxygen-mediated N-N coupling with high selectivity. Additionally, both WO₃ catalyst and DPK mediator demonstrated promising reusability, highlighting the potential for sustained, environmentally benign N₂H₄ synthesis. This work not only demonstrates a strategy for N₂H₄ production but also provides an in-depth understanding of the catalytic pathway. We believe that it may inspire the synthesis of other value-added chemicals that are difficult to produce by conventional routes.

Methods

Materials

Sodium tungstate dihydrate (Na₂WO₄·2H₂O, 99%), tungsten trioxide (WO₃, 99.9%), potassium nitrate (KNO₃, 99.9%), benzophenone (C₁₃H₁₀O, 99%), potassium hexafluorophosphate (KPF₆, 99.5%), sodium salicylate (C₇H₅NaO₃, 99.5%), sodium hypochlorite (NaClO, active chlorine>10%), sodium nitroferricyanide dihydrate (C₅FeN₆Na₂O·2H₂O, 99%), 4-(dimethylamino)benzaldehyde (C₉H₁₁NO, 99%), hydrazine monohydrochloride (N₂H₄·HCl, 98%), hydrazinium sulfate (N₂H₄·H₂SO₄, 99%), tetrabutylammonium perchlorate (TBAP, for electrochemical analysis) and nickel oxide (NiO, 50 nm) were purchased from Innochem (Beijing) Technology Co., Ltd. Copper oxide (CuO, 99.9%) was obtained from Aladdin. Ammonia (7.0 mol L⁻¹ in methanol), chloroform-d (CDCl₃, 99.8%), deuterium oxide (D₂O, 99.8%) benzophenone imine (C₁₃H₁₁N, 95%), acetophenone (C₈H₈O, 99%), benzophenone imine (C₁₃H₁₁N, 95%), sodium 3-(trimethylsilyl)propane-1-sulfonate (C₆H₁₅NaO₃SSi, DSS, 99.5%) were supplied by Energy Chemicals Inc. Ammonium chloride (NH₄Cl, 99.5%), sodium sulfate (Na₂SO₄, AR), potassium sulfate (K₂SO₄, AR), hydrochloric acid (HCl, 35%), sulfuric acid (H₂SO₄, 96%), sodium hydroxide (NaOH, 99%) and were provided by Sinopharm Chemical Reagent Co., Ltd. Acetate (C₄H₈O₂, 99.8%), acetone (C₃H₆O, 99.8%), methanol (CH₄O, 99.9%), ethanol (C₂H₆O, 99.5%), and dichloromethane (CH₂Cl₂, 99.9%) were provided by Concord Technology (Tianjin) Co., Ltd. Toray Carbon Paper TGP-H-060 was used to load electrocatalysts. Aqueous solutions were prepared with deionized (DI) water (Millipore 18.2 MΩ cm). All the chemicals were used as received without further purification. All electrodes were supplied by Gaossunion.

Catalyst synthesis

The hydrothermal method was used to synthesize WO₃³⁰. At ambient temperature, 14 mmol of Na₂WO₄·2H₂O, 3.5 mmol of Na₂SO₄, and 3.5 mmol of K₂SO₄ were dissolved in 80 mL of DI water. Subsequently, the pH was adjusted to 2.0 using a 10 mol L⁻¹ HCl aqueous solution until the appearance of yellow precipitates. After continuous stirring for 12 h, the solution was transferred to a Teflon-lined stainless-steel autoclave of 100 mL and subjected to heating at 180 °C for 24 h in an oven. The resulting precipitates were harvested through centrifugation, underwent repeated washing with DI water, ethanol, and acetate and were then dried under vacuum conditions at 80 °C overnight.

Electrochemical reactions

To prepare the working electrode, 10 mg of the synthesized catalyst was mixed with 1 mL of isopropanol and 20 μL of 5 wt% Nafion D-521 dispersion. After sonication for 15 min, the catalyst ink was evenly applied to 10 pieces of carbon paper (0.5 × 2 cm² each) using a micropipette, achieving a loading of approximately 0.5 mg cm⁻².

Electrochemical data were acquired using a CHI 660E electrochemical analyzer. The reduction of NO₃⁻ was conducted in a 1.0 mol L⁻¹ KOH aqueous solution with 0.1 mol L⁻¹ KNO₃ employing a typical H-cell (50 mL) separated by a Nafion 117 membrane at room temperature²¹. The Nafion 117 membrane was protonated sequentially with boiled water, 5% H₂O₂ solution, and 0.5 mol L⁻¹ H₂SO₄ solution. Electrochemical reactions were performed using a three-electrode configuration, which included CuO electrocatalyst loaded on carbon paper (0.5 mg cm⁻²) as the working electrode, an Ag/AgCl reference electrode (3 mol L⁻¹ KCl solution), and a Pt film counter electrode (1 cm²). To collect NH₃ product in CH₃CN solution for further use, NH₃ was removed from the aqueous solution to CH₃CN solution through gas stripping using Ar gas flow of 50 sccm.

Electrosynthesis of BPA was carried out in a single cell (30 mL) with a WO₃/carbon paper anode, a surface area of ~0.5 cm² was immersed in the solution, a platinum gauze cathode (~1.0 cm²), and Ag/Ag⁺ (0.01 mol L⁻¹ AgNO₃ and 0.1 mol L⁻¹ TBAP) reference electrode⁴⁷. Then, 15 mL of 0.1 mol L⁻¹ DPK, 0.1 mol L⁻¹ NH₃ in methanol, and 0.1 mol L⁻¹ KPF₆ acetonitrile solution was electrolyzed for 10 h with magnetic stirring (500 rpm). The total amount of methanol was 5 vol% of the electrolyte. Ohmic resistance (R) of the cell was 2.4 Ω, which was measured by electrochemical workstation. All the electrochemical performance tests were without experiment was without iR compensation.

In this study, the Ag/Ag⁺ electrode was calibrated using ferrocene/ferrocenium (Fc/Fc⁺) as the redox standard in the same solvent/electrolyte mixture as in the experiments. The Ag/AgCl electrode potential was determined using [Fe(CN)₆]³⁻/[Fe(CN)₆]⁴⁻ in a KCl supporting electrolyte. All the calibrations were carried out at room temperature.

Product analysis

BPA and DPK were identified using gas chromatography-mass spectrometry (GC-MS). The analysis was performed on an Agilent 5977A instrument equipped with an HP-5MS capillary column, which has a 0.25 mm internal diameter and is 30 m long. For high-resolution GC-MS, a Thermo Fisher Scientific Exactive GC system was utilized. ¹H and carbon ¹³C nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Fourier 300 spectrometer with CDCl₃ as the solvent.

The FE of BPA was calculated using the equation:

$$FE=\frac{nFcV}{Q}$$ 1

where n represents the number of electrons transferred per mole during BPA formation, F is the Faraday constant, ccc is the concentration of the product, V is the volume of the electrolyte, and Q is the total charge passed during the reaction.

To measure the concentration of NH₃ in the catholyte, modified colorimetric techniques were employed²³. Specifically, the salicylate method was adapted, which involved converting NH₃ to indophenol blue via a reaction with salicylate and hypochlorite. The preparation of the necessary solutions was as follows: Solution A was a mixture of 0.32 mol L⁻¹ NaOH and 0.4 mol L⁻¹ sodium salicylate. Solution B consisted of 0.75 mol L⁻¹ NaOH mixed with sodium hypochlorite (~4.5% active chlorine). Solution C was a 10 mg mL⁻¹ solution of sodium nitroprusside dihydrate. For the analysis, 3 mL of the aqueous sample was combined with 500 μL of Solution A, 50 μL of Solution B, and 50 μL of Solution C in a sample tube. A fresh electrolyte served as the blank for UV-vis measurements. A calibration curve was created by preparing standard solutions with known NH₄⁺ concentrations and measuring their absorbance at 675 nm. The NH₃ concentration in the samples was determined based on this calibration curve using a Perkin Elmer Lambda 1050 + UV-vis spectrophotometer.

For the determination of N₂H₄ content, the Watt and Chrisp method was used⁴⁸. The color reagent consisted of 5.99 g of C₉H₁₁NO, 30 mL of concentrated hydrochloric acid, and 300 mL of ethanol. The calibration curve was developed by preparing reference solutions with varying concentrations of hydrazine hydrate in a 0.1 mol L⁻¹ hydrochloric acid solution, which was then diluted to 5 mL with deionized water. Following this, 5 mL of the color reagent was added, and the mixture was stirred for 10 min at room temperature. The absorbance was measured at 457 nm, with the hydrazine yield determined using the calibration curve established with known N₂H₄·HCl standards.

Recrystallization of N₂H₄ salts

The aqueous N₂H₄ solution was placed in a distillation flask, and the distillation apparatus was assembled. The solution was gradually heated to distill off the water, maintaining the temperature below the boiling point of hydrazine (114 °C) to prevent its loss. The concentrated hydrazine solution was then collected for further use. In a well-ventilated fume hood, concentrated HCl was added dropwise to the concentrated hydrazine solution with continuous stirring. The addition rate was carefully controlled to manage the exothermic reaction. HCl was added until the solution reached a pH of <1. The reaction mixture was cooled in an ice bath to promote the crystallization. The mixture is left in the ice bath for 2 h to ensure complete crystallization. The hydrazine hydrochloride crystals were then isolated by vacuum filtration. The crystals were washed with a small volume of cold ethanol or acetone to remove impurities. Finally, the washed crystals were transferred to a drying oven set at 60 °C.

In situ Raman measurements

During in situ Raman measurements, a flow cell with a quartz window from GaossUnion (Tianjin) Photoelectric Technology Company was employed, utilizing the Horiba LabRAM HR Evolution Raman microscope. A 785 nm laser was utilized for excitation, and signals were captured with a 30-s integration time, averaging three scans for enhanced accuracy.

ATR-FTIR measurements

The ATR-FTIR spectra of DPK-I in different solvents were collected on a Bruker VERTEX 70 v FT-IR Spectrometer. In situ ATR-FTIR measurements were carried out in a customized electrochemical cell integrated into a Nicolet 6700 FTIR spectrometer with an MCT detector cooled by liquid nitrogen. The catalyst ink was deposited onto a germanium ATR crystal coated with a gold film. Each spectrum was acquired through 32 repetitions, maintaining a resolution of 4 cm⁻¹.

DFT calculations

All computational calculations were performed using the Gaussian16 software package⁴⁹. The M06-2X hybrid functional was employed throughout the study⁵⁰. Geometry optimizations were conducted with the 6-31G(d,p) basis set, and analytical frequency calculations were carried out at the same theoretical level to confirm the nature of each stationary point, identifying whether it corresponded to a minimum (characterized by no imaginary frequencies) or a transition state (indicated by a single imaginary frequency). These frequency calculations also provided Gibbs free energy corrections at 298.15 K. For the optimized structures, the final energies were refined using the larger 6-311+G(d,p) basis set.

In the case of WO₃, the computed lattice parameters were 7.309 × 7.522 × 7.678 Å, with a space group of P-1 (2). Density functional theory (DFT) calculations were carried out using the Vienna Ab initio Simulation Package (VASP). The generalized gradient approximation (GGA) was applied with the Perdew-Burke-Ernzerhof (PBE) functional. Projected augmented wave (PAW) potentials were utilized to describe the ionic cores, while the valence electrons were represented with a plane wave basis set, using a kinetic energy cutoff of 450 eV. Geometry optimizations were performed with a force convergence criterion of less than 0.05 eV Å⁻¹. The original bulk structures of WO₃ were optimized before surface construction, using a Monkhorst-Pack k-point grid of 5 × 5 × 5. For the WO₃(200) surface, the lattice parameter and the supercell size were set at 15.211 × 15.211 Å, with a vacuum space of 15 Å to prevent interactions between the periodic slabs. A Monkhorst-Pack k-point grid of 2 × 2 × 1 was employed for these calculations. To address the limitations of GGA, a GGA+U correction was applied, with U-J = 6.2 eV for tungsten atoms.

The Gibbs free energy change (ΔG) for each electrochemical process was calculated using the equation:

$$\Delta G=\Delta E+{\Delta E}_{\mathrm{ZPE}}-T\Delta S$$ 2

where ΔE represents the change in DFT energy, ΔE_ZPE is the zero-point energy correction, and ΔS is the entropy change at 298.15 K.

Author contributions

S.H.J., X.F.S., and B.X.H. proposed the project, designed the experiments, and wrote the manuscript; S.H.J. performed the whole experiments; L.B.Z., H.L.L., R.H.W., X.Y.J., L.M.W., X.N.S., X.X.T., X.D.M., and J.Q.F. conducted a part of characterizations. Q.G.Z., X.C.K., and Q.L.Q. participated in discussions. X.F.S. and B.X.H. supervised the whole project.

Peer review

Peer review information

Nature Communications thanks Tengfei Li and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data that support the findings of this study are available from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-024-52825-1.