Nitrate (NO₃⁻) pollution has become one of the major concerns in water management [1]. High-strength NO₃⁻ wastewater that was mainly induced by anthropogenic activities such as fertilizer and industrial applications poses a potential threat to human health [2,3]. Electrocatalytic NO₃⁻ reduction reaction (NO₃RR) highlights its advantages particularly for industrial wastewater (e.g., wastewater from explosives and photovoltaic industries), where the NO₃⁻ content is high and bacterial growth is unviable [4,5]. Instead of converting NO₃⁻ into N₂ for purification purposes, the acquisition of ammonia (NH₃) from NO₃⁻ provides a valuable pathway for the production of NH₃, an essential chemical and an alternative carbon-free energy carrier [6–8]. The separation of NH₃ can be efficiently achieved using extraction technology or regenerated resins, while the electricity required for this process can be generated from clean and renewable sources such as solar and wind power [9,10]. Therefore, NO₃RR may provide an opportunity for the NH₃ production industry and remediation of NO₃⁻-containing wastewater.

Cu-based catalysts have been widely explored for the NO₃RR due to its high electrocatalytic reduction kinetics of NO₃⁻ to nitrite (NO₂⁻), which is the rate-limiting step of the NO₃RR to NH₃ [11,12]. However, the intermediates NO₂⁻ will accumulate on the pure Cu electrocatalyst surface thereby leading to its rapid deactivation and increased energy consumption [13]. Substantial efforts such as alloying Cu with transition metals (e.g., Pt, Pd, and Ni) or formation of compounds with molecular solids have been made to overcome these limitations [14,15]. In particular, Cu-based oxide (CuO) was transformed into hybrids (Cu/Cu₂O) in situ in NO₃RR, which enhances NO₃⁻ removal and NH₃ generation [16,17]. However, the conversion of NO₃⁻ into NH₃ involves an eight-electrons transfer and multiple reaction intermediates. On the single metal oxide catalyst surface, optimizing the binding strength of one reactant would typically take the other steps away from their optima [18–20]. In this case, simultaneous acceleration of sequential reduction of oxygen-containing intermediates (e.g., ^*NO₂⁻ and ^*NO) and hydrogenation of nitrogen-containing intermediates (e.g., ^*NH and ^*NH₂) remains challenging.

A tandem electrocatalyst contains distinct active sites proceeding the individual reaction steps, which can couple and mediate the consecutive steps to effectively produce the desired products [21–23]. Transition metal oxides undergo potential-dependent phase formation during cathodic electrocatalysis, resulting in the coexistence of multiple species such as metallic, oxide, and hydroxide [24–26]. It presents an opportunity for optimizing the electrocatalytic performance if these freshly-generated species could be combined for tandem NO₃⁻ to NH₃ conversion. As mentioned above, Cu-based species suffer from the adsorption of NO₃⁻-reduction intermediates such as NO₂⁻ to block the electrode surface and slow down subsequent deoxidation and hydrogenation process [27,28]. Fortunately, Co-based oxides have been reported to exhibit satisfactory activity towards NO₃⁻ to NH₃ and specific NO₂⁻ reduction to NH₃ [29,30]. Therefore, we hypothesize that combining the advantages of Cu-based catalyst with high reduction kinetics of NO₃⁻ to NO₂⁻ and Co-based catalyst with favorable conversion towards NO₂⁻ to NH₃ would accelerate sequential NO₃⁻ to NH₃ with low overpotential.

In this work, a series of Cu-Co binary metal oxides (Cu-Co-O) with different Cu and Co content were prepared combining facile co-precipitation and calcination method. Interestingly, multiple species were formed in situ on the Cu-Co-O surface during the NO₃RR, which acted as tandem active species for NO₃⁻ to NH₃ conversion. Electrocatalytic measurements, X-ray photoelectron spectroscopy (XPS), in situ Raman spectra, and in situ differential electrochemical mass spectrometry (DEMS) revealed that Cu-based species are favorable for NO₃⁻ to NO₂⁻ conversion, while Co-based phases tend to catalyze NO₂⁻ reduction to NH₃. Consequently, Cu-Co-O catalysts demonstrated higher NO₃⁻ removal and NH₃ generation compared to single metal oxide (CuO or Co₃O₄) at low overpotentials. This study presents a general strategy for designing high-performance tandem catalysts for multi-step electrochemical reactions.

The Cu-Co-O catalysts were synthesized via a facile co-precipitation method coupled with subsequent calcination (Fig. 1a). By carefully adjusting the molar ratio of Cu to Co in the co-precipitation process, the Cu-Co-O with tunable element components could be acquired (Table S1 in Supporting information). Powder X-ray diffraction (XRD) patterns (Fig. 1b) show that main peaks located at 35.6° and 38.8° were assigned to the (111) and (111) crystal planes of CuO (JCPDS No. 89–5899), and the main peaks at 31.3°, 36.9° and 44.9° were well indexed to the (220), (311) and (400) of Co₃O₄ (JCPDS No. 74–1657). Comparing the diffraction peaks of these obtained samples, we found that the relative intensity of CuO signals were gradually weakened with the introduction of Co element. The observed peak shifts of the samples (Fig. 1c) synthesized using mixed metal sources, as compared to Co₃O₄ and CuO, suggested the doping of Co₃O₄ with Cu or CuO with Co that caused the contraction or expansion of the crystal lattice, respectively [31,32]. Furthermore, the diffraction signals of both CuO and Co₃O₄ were evident in Cu₃CoO_x, CuCoO_x, and CuCo₃O_x, indicating the composite nature of these catalysts. In addition, XPS analysis of the pristine catalyst revealed no nitrogen signals, suggesting that the use of metal precursor containing NO₃⁻ had no effect on the accurate quantification of NO₃⁻ in water (Fig. S2 in Supporting information).

The morphological characteristics of the prepared CuCoO_x were uncovered by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). As shown in Fig. 1d and Figs. S3a-d (Supporting information), CuCoO_x exhibited assembled nanosheets with a rough surface and the addition of Na₂CO₃ could modify the particle size and morphology of CuCoO_x (Fig. S3e in Supporting information). The TEM images, coupled with selected area electron diffraction (SAED), display that the nanosheets were mainly composed of nanocrystals and featured a polycrystalline structure (Fig. 1d). The corresponding energy dispersive X-Ray spectroscopy (EDX) mapping reveal strong and evenly distributed signals of Co, Cu, and O throughout the entire selected area (Fig. 1e).

Linear sweep voltammetry (LSV) and two-hour electrolysis were conducted to assess the performance of the obtained catalysts. Fig. 2a shows LSV curves in the presence of NO₃⁻. As observed, CuCoO_x catalyst displayed increased current density (j) compared with CuO and Co₃O₄. Moreover, CuCoO_x showed a clear increase of the reductive current after adding NO₃⁻, implying a catalytic response towards the NO₃RR (Fig. S4 in Supporting information). CuCoO_x demonstrated the highest NO₃⁻ removal over the entire potential range from −1.0 V to −1.3 V, with NO₃⁻-N removal percentage exceeding 90% at −1.1 V (Fig. 2b). Moreover, the catalytic current increased as the electrolysis potential became more negative, and CuCoO_x achieved a maximum FE of 81% for NH₃ at −1.1 V (Fig. 2c and Fig. S5 in Supporting information). At this applied potential, CuO exhibited low FE for NH₃, while Co₃O₄ showed weak NO₃RR activity. Notably, in some reported literature, single Cu- or Co- based catalysts required more negative potentials to achieve similar percentages of NO₃⁻ removal (Table S2 in Supporting information) [33,34], indicating that the synergy of Cu and Co sites in CuCoO_x may reduce the overpotential of NO₃RR.

Furthermore, we analyzed the product distribution catalyzed by these catalysts through continuous electrolysis measurement at a fixed potential. Over the potential range of −1.0 V to −1.3 V, the NO₃⁻-N removal gradually increased, while the FE of NH₃ exhibited a maximum value at −1.1 V for almost all of the prepared samples (Figs. 2b and c). Thus, we selected −1.1 V for potentiostatic electrolysis. As shown in Fig. 2d and Fig. S6 (Supporting information), NO₃⁻-N was removed while NH₄⁺-N kept increasing with the prolonging of reaction time. CuCoO_x exhibited the best NO₃⁻-N removal (90%) and NH₄⁺-N generation (87%). The values of v_NO3⁻ on CuCoO_x (0.027 mmol h⁻¹ cm⁻², 0.026 mmol h⁻¹ cm⁻²) at 120 min were higher than those on CuO (0.024 mmol h⁻¹ cm⁻², 0.021 mmol h⁻¹ cm⁻²) and Co₃O₄ (0.008 mmol h⁻¹ cm⁻², 0.007 mmol h⁻¹ cm⁻²) (Fig. 2e). Notably, above 6% of NO₂⁻-N was generated on CuO during the 120 min reaction, but there was no evident NO₂⁻-N accumulation after the introduction of Co in CuO (Fig. S7 in Supporting information). Since NO₂⁻ is an even more dangerous pollutant than NO₃⁻, it is desirable to keep its content as low as possible [35]. Additionally, NO₂⁻-N was detected in the case of Cu-Co-O electrodes only when the applied potential potentials were more positive than −1.1 V (Fig. 2f). These results imply that Cu-based species can convert NO₃⁻ to NO₂⁻, and Co-based species might further reduce NO₂⁻ to NH₃.

XRD and XPS were conducted to identify the compositions of the Co₃O₄, CuO, and CuCoO_x catalysts. Fig. S8 (Supporting information) is the XRD profile of the three catalysts after 120 min of NO₃RR. The carbon signals appeared because the prepared catalyst powders were loaded on the carbon paper. After NO₃RR, CuO showed obvious Cu⁰ signals. It was difficult to observe the diffraction peak indexed to the metal hydroxides or reductive metal oxides due to the extremely low content or poor crystallization of the generated species in the samples. XPS was further carried out to analyze the species on the catalyst surface. High-resolution Co 2p and Cu 2p spectra displayed spin orbit splitting into 2p_1/2 and 2p_3/2 segments, and both segments qualitatively contained the same chemical valence information [36,37]. Therefore, only the higher-intensity Co 2p_3/2 and Cu 2p_3/2 bands were fitted in the obtained spectra. Cu 2p spectra of CuCoO_x showed increased signals associated with reductive phases (Cu(OH)_x and Cu⁺/Cu⁰ based species) after the NO₃RR (Fig. 3a). It should be noted that Cu 2p spectra of CuCoO_x after NO₃RR revealed the presence of observable satellite features at 943.0 eV, whereas the satellite signals on CuO after the reaction were almost undetectable (Fig. 3a and Fig. S10a in Supporting information). These results confirm that Cu species predominantly existed in a positive-valence state rather than a zero-valence state in CuCoO_x after reaction, and the opposite was true for CuO after the reaction [38,39]. Cu LMM spectra are more useful in determining chemical states of Cu. As shown in Fig. S9a (Supporting information), Cu LMM indicated the presence of multiple Cu-based phases, including Cu⁰ (566.8 eV), CuO (568.7 eV), Cu₂O (569.8 eV), and Cu(OH)_x (571.0 eV) in CuCoO_x after the NO₃RR, while the pristine CuCoO_x mainly contained CuO phase. Compared to CuO (Fig. S10b in Supporting information), CuCoO_x showed higher content of Cu(OH)_x phase, implying that the compositions in CuCoO_x were impacted by the coexistent Co-based phases. Moreover, the content of Co²⁺-dominated CoO (780.8 eV) and Co(OH)_x (782.1 eV) in CuCoO_x significantly increased after NO₃RR, which could be also observed in Co₃O₄ before and after reaction (Fig. 3b and Fig. S11a in Supporting information). The composition changes can be further confirmed by the O 1s spectra (Figs. S9b, S10c and S11b in Supporting information). Oxygen species with binding energies below 531 eV are characteristic metal-oxygen in metal oxides, while signals at 531–532 eV correspond to metal hydroxides. Furthermore, the peak at the higher values above 532 eV is associated with oxygen vacancies [40,41]. As expected, the content of metal hydroxides and oxygen vacancies increased markedly, while the metal-oxide phases decreased for all three catalysts after the NO₃RR.

To gain more insight into the phase evolution of the three catalysts at different applied potentials, we interrogated the catalyst surface using in situ Raman spectroscopy (Fig. 3c). The Co₃O₄ and CuO powder exhibited their characteristic Raman bands at 190/482/521/690 cm⁻¹ and 293/340/628 cm⁻¹ (Fig. S12 in Supporting information), respectively [25,42]. Note that the spectral response of CuCoO_x was not a direct combination of the spectra of Co₃O₄ and CuO, but displayed significant shoulder peaks, implying the interactions between Cu and Co element within the CuCoO_x catalyst. We then acquired Raman spectra on the prepared three electrodes (CuO, Co₃O₄, and CuCoO_x) at the applied potentials from the open-circuit potential (OCP) to −1.3 V vs. SCE (Fig. 3d and Fig. S13 in Supporting information). Under negative potentials, the peaks associated with CuO gradually disappeared, which could be attributed to the conversion of CuO to related reduced Cu-based species but difficult to distinguish from background noise across the entire series (Fig. S13a). On the Co₃O₄ electrode, the typical Raman peaks related to Co₃O₄ were attenuated with reducing potential while the signals about Co(OH)_x were enhanced after the potential of −1.0 V (Fig. S13b). In the case of CuCoO_x, the Raman signals associated with CuCoO_x gradually weakened, while bands corresponding to Cu(OH)_x and Co(OH)_x became increasingly evident at potentials below −0.8 V (Fig. 3d).

These findings indicate that new species, such as reductive metal oxides and metal hydroxides, were generated on the catalyst surface during NO₃RR, which may provide enriched active phases for enhanced NO₃RR. Based on the above results, CuCoO_x was a combination of CuO and Co₃O₄ in phase compositions. Therefore, it is reasonable to speculate that CuCoO_x catalyst exhibits tandem reactivity involving Cu-based and Co-based species. The aforementioned electrochemical results indicate that NO₂⁻ accumulated when CuO was used as catalyst, but not when Co₃O₄ was used. Accordingly, the sequential acceleration of the NO₃⁻ to NO₂⁻ on CuO and NO₂⁻ to NH₃ on Co₃O₄ would lead to fast NO₃⁻ to NH₃ conversion with high NH₃ selectivity at low overpotentials.

To confirm this hypothesis, we further evaluated NO₂⁻-N removal performance of CuO, CuCoO_x, and Co₃O₄ in a divided cell reactor. At −1.0 and −1.1 V, CuCoO_x exhibited the highest NO₂⁻-N removal compared to Co₃O₄ and CuO, and Co₃O₄ showed better NO₂⁻-N removal than CuO (Fig. 4a). Moreover, CuCoO_x inherited the advantages of Co₃O₄ and reached a maximum FE of 93.3% for NH₃ at −1.1 V, which was higher than that of Co₃O₄ (81%) and CuO (66.5%) (Fig. S14a in Supporting information). CuCoO_x also displayed higher NO₂⁻-N removal at −1.1 V at regular intervals (Fig. S14b in Supporting information). Furthermore, we evaluated the rate constants of the catalysts for NO₃⁻ and NO₂⁻ removal. As shown in Figs. 4b and c, CuO exhibited a larger rate constant (12.3 × 10⁻³ min⁻¹) for NO₃⁻ removal than Co₃O₄ (2.6 × 10⁻³ min⁻¹), but a smaller rate constant (14.1 × 10⁻³ min⁻¹) for NO₂⁻ removal than Co₃O₄ (17.3 × 10⁻³ min⁻¹). It suggests that CuO was more effective in reducing NO₃⁻ while Co₃O₄ reduced NO₂⁻ faster than CuO.

Furthermore, the molecular intermediate and product over CuO, CuCoO_x, and Co₃O₄ were observed by in situ DEMS (Fig. S15 in Supporting information) [43]. As shown in Figs. 4d and e, the signals of NO, N, NH, NH₂, NH₃, and H₂ were detected on the CuCoO_x and CuO. The strong m/z 17 of NH₃ signal and negligible m/z 28 of N₂ signal confirmed the high NH₃ selectivity of CuCoO_x. Notably, H₂ was not a preferred product on CuCoO_x while Co₃O₄ yielded more H₂ compared with CuCoO_x and CuO (Fig. S16 in Supporting information). In general, NH₃ as a NO₃⁻-reduction product is favored in a potential region close to the hydrogen evolution reaction (HER), where NH₃ would be formed by the reaction between adsorbed hydrogen (H^*) and adsorbed NO₃⁻ [44,45]. Therefore, it can be inferred that Cu-based species reduce NO₃⁻ to N-containing intermediates, which can then be utilized by Co-based species to generate more deeply hydrogenated intermediates (e.g., NH, NH₂ and NH₃) and suppress competitive HER as well.

Cu-Co-O catalysts effectively enhance catalytic activity through the following mechanism. The NO₃RR on Cu-Co-O catalysts to produce NH₃ (NO₃⁻ + 6H₂O + 8e⁻ → NH₃ + 9OH⁻) involves several elementary reactions: NO₃⁻ is initially adsorbed on the catalyst surface, followed by a series of deoxidation processes (^*NO₃⁻ → ^*NO₂⁻ → ^*NO → ^*N) and sequential hydrogenation of ^*N (^*N → ^*NH → ^*NH₂ → ^*NH₃). Under the reduction potential of the NO₃RR, electrochemically driven phase generation occurs on the Cu-Co-O surface, resulting in the formation of hybrid M(OH)_x/MO_x phases that act as cooperative active sites for the multi-electron NO₃RR. Specifically, Cu-based species as active sites preferentially adsorb NO₃⁻ and reduce it to NO₂⁻, which is then sequentially reduced to NH₃ on nearby Co-based sites. More importantly, the implementation of the tandem system makes it possible to lower the reaction overpotential while suppressing hydrogen evolution reaction, leading to highly efficient NO₃⁻ to NH₃ conversion (Fig. 4f).

The long-term stability is another key parameter for electrocatalyst evaluations. Comparing obvious attenuation of the NO₃⁻-N removal on CuO catalyst, CuCoO_x showed insignificant variations for NO₃⁻-N removal after 10 cycles (Fig. 5a and Fig. S17a in Supporting information). Moreover, the NH₄⁺-N selectivity of CuCoO_x remains ~90% after 10 cycles. In addition, the pristine assembled nanosheets morphology of CuCoO_x is maintained after the cycling test, which further corroborated its favorable durability (Figs. S18a and b in Supporting information). The energy dispersive spectroscopy (EDS) mapping images of the CuCoO_x after cycling test indicated the uniform distribution of Cu, Co, and O elements (Fig. S18c in Supporting information). During NO₃RR, the pH of the electrolyte gradually became alkaline [46]. Previous studies have reported that single Cu catalyst gradually deactivated due to the formation of amorphous Cu resulting from the decomposition of CuH or CuH₂, which inhibited electrochemical performance in alkaline solution [35]. This also explains the reason for the inferior stability of CuO, as above mentioned. Furthermore, we investigated the role of pH played on CuCoO_x catalyst for the NO₃⁻-N removal and NH₄⁺-N selectivity. As shown in Fig. 5b and Fig. S17b (Supporting information), after 120 min treatment, NO₃⁻-N removal and NH₄⁺-N selectivity reached 90% or higher at different pH levels, implying that the CuCoO_x catalyst prepared in this study had a broad working pH range. It is speculated that the Cu-Co compound could prevent the formation of amorphous Cu and promoted the presence of Cu²⁺/Cu⁺-based species, as demonstrated in XPS results. These Cu²⁺/Cu⁺-based species in CuCoO_x served as active sites for achieving long-term reaction towards NO₃RR. Furthermore, we investigated the impact of Cl⁻ and CO₃²⁻ on the NO₃RR performance. The concentration of the impurities was set based on the NO₃⁻ concentration of 3.57 mmol/L. Our results demonstrated that the presence of Cl⁻ and CO₃²⁻ had no significant effect on the NO₃⁻-N removal or the selectivity of NH₄⁺-N (Fig. S17c in Supporting information).

In order to evaluate the potential practical applications of CuCoO_x, we used a two-electrode system with CuCoO_x cathode and Ir-Ru/Ti anode to investigate the performance of NO₃RR. The polarization curve (Fig. 5c) revealed an increased j at the applied potential of 2.0 V. Potentiostatic electrolysis showed that the NO₃⁻-N removal and NH₄⁺-N selectivity reached > 90% at a cell voltage of 2.4 V after 120 min treatment (Fig. 5d and Fig. S19 in Supporting information). Accordingly, the energy consumption was calculated to be 0.69 kWh/mol, which was comparable to the FeNi/graphitized mesoporous carbon/Ni foam [47] and Pd-based TiO₂ cathode [48].

In summary, we demonstrated that Cu-Co oxides exhibited highly tandem reactivity towards NO₃⁻ reduction. During NO₃RR, electrochemical reduction induced phase reconstruction of Cu-Co oxides to generate more Cu- and Co-based phases as catalytic active species. In this tandem system, Cu-based species preferred to reducing NO₃⁻ to NO₂⁻, while the generated NO₂⁻ intermediates would further be converted to NH₃ on Co-based species. Benefiting from the favorable Cu- and Co-based partners, Cu-Co oxide electrode reached over 90% of nitrate removal and ammonia selectivity, which was superior than those of single metal oxides (CuO and Co₃O₄). Moreover, the good stability and low energy consumption (0.69 kWh/mol) of Cu-Co oxides were comparable to most of the NO₃RR catalysts. These findings indicate a universal strategy for enhancing NO₃⁻ conversion into value-added NH₃ with the efficient tandem electrocatalysts, which can also be extended to other multi-step electrocatalytic reaction.

Acknowledgments

This work was supported by National Natural Science Foundation of China (Nos. 52131003 and 42007180), Special Research Assistant Program of Chinese Academy of Science, Natural Science Foundation of Chongqing (No. cstc2020jcyj-msxmX0775), Scientific Research Instrument Development Project of Chinese Academy of Sciences (No. YJKYYQ20200044), Outstanding Scientist of Chongqing Talent Program (No. CQYC20210101288).

Supplementary material

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2023.108864