Introduction

The finite nature of fossil fuels and the severe environmental pollution resulting from their extensive use have propelled the development and utilization of renewable energy sources to the forefront of scientific research1,2. Among these, hydrogen energy stands out as one of the most promising alternatives because of its high calorific value, pollution-free combustion products, and abundant sources3. Among the various hydrogen production methods, the electrochemical hydrogen evolution reaction (HER) has garnered widespread attention because of its mild reaction conditions and simple operation4. However, the slow kinetics of the anodic water oxidation reaction necessitate a high overpotential for hydrogen production via electrolytic water, considerably impeding the rate of hydrogen evolution and the overall energy conversion efficiency5. To address this, additives such as urea, methanol, and hydrazine have been introduced into the electrolyte. These small molecules offer more favorable thermodynamics for oxidation than water does, thus replacing water oxidation and reducing the anode reaction overpotential, leading to decreased overall energy consumption for hydrogen production. Urea, in particular, is widely available from agricultural and industrial wastewater as well as human and animal urine. The discharge of urea-containing wastewater poses substantial environmental hazards6, making the treatment of urea wastewater a crucial issue. The electrochemical electrolysis of urea wastewater not only addresses environmental pollution but also enables energy-efficient hydrogen production, representing a highly effective “two-for-one” solution.

Despite their low overpotential, noble metal catalysts are hindered by high costs, limiting their large-scale industrial applications7,8. In recent years, transition metals (e.g., Ni, Co, Mo, and Fe) and their oxides and hydroxides have emerged as promising electrocatalysts for urea oxidation due to their abundance and low cost9,10. However, current non-noble metal catalysts still exhibit high oxidation potentials and poor stability11,12, necessitating the development of non-noble metal electrocatalysts with low potentials, high catalytic efficiency, and enhanced stability for the urea oxidation reaction (UOR). Recent advances in atomic site engineering have revealed that single-atom catalysts maximize atomic utilization but struggle to activate complex UOR intermediates requiring multisite synergies13,14,15. Dual-atom catalysts partially address this through metal‒metal orbital interactions, yet their bimetallic configurations often exhibit insufficient electron delocalization and unstable coordination under high currents16,17,18. Emerging trimetallic systems with asymmetric coordination offer another structural and electronic advantages19. Geometrically, the asymmetric environment breaks local symmetry, creating an anisotropic electronic field essential for changing scaling relationships and independently optimizing adsorption energies for diverse UOR intermediates. Electronically, the intimate proximity of three distinct metals enables better multiorbital coupling, surpassing bimetallic systems20. This enables a highly tunable electronic structure, facilitating synergistic multisite activation. Furthermore, the heterotrimetallic configuration inherently possesses greater dynamic adaptability. Different metal preferences allow cooperative structural adjustments under operational conditions, increasing stability, especially at high current densities. Thus, asymmetric trimetallic sites address the concurrent needs for multisite synergy, optimized energetics, and operational robustness in complex reactions such as UOR21. However, precisely constructing trimetallic active sites and enabling them to exhibit a strong synergistic effect in electrocatalytic UOR reactions while maintaining stability during electrolysis at high current density remains a challenging problem.

In this study, we constructed a highly efficient and stable asymmetrically coordinated trimetallic atom catalyst by leveraging the structural characteristics of monolayer layered double hydroxides (LDHs). Monolayer LDHs possess inherent fully exposed edge-sharing [M(OH)6] octahedral units within their layers22. These units naturally facilitate the formation of multimetallic coordination centers, providing an ideal platform for constructing atomically dispersed trimetallic active sites23. Moreover, single-layer LDHs are self-adaptive, enabling them to dynamically adjust their surface structure and active sites during reactions; thus, they can adapt to diverse conditions and reaction intermediates, ultimately enhancing their stability and activity24,25. Specifically, we report the synthesis of a monolayer asymmetrically coordinated heterotrimetallic atom catalyst (A-NiCoMn-TAC/LDH) with varying manganese (Mn) contents using a one-step coprecipitation method. The electrochemical performance of these materials was evaluated, and their application in UOR-assisted energy-saving hydrogen production systems has been demonstrated26. Our electrocatalytic UOR experiments revealed that the construction of asymmetrically coordinated heterotrimetallic sites considerably enhanced the UOR activity. Specifically, A-NiCoMn-TAC/LDH with 0.2 wt% Mn exhibited the best UOR catalytic activity, requiring only 1.26 and 1.33 V vs. RHE to achieve current densities of 10 and 100 mA cm−2, respectively, outperforming other catalysts with other Mn contents. X-ray photoelectron spectroscopy (XPS) and ex situ X-ray absorption spectroscopy (XAS) analyzes revealed the presence of abundant oxygen vacancies and metal defects in monolayer A-NiCoMn-TAC/LDH. The reaction intermediates were revealed by attenuated total reflectance surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS) during UOR. Furthermore, in situ XAS investigations elucidated the evolution of the coordination structure of atoms and the critical role of defects. Density functional theory (DFT) calculations revealed that the asymmetrically coordinated NiCoMn heterotrimetallic sites surrounded by vacancies exhibited significantly enhanced d-p-d orbital couplings and electron transfer capacity, largely optimizing the adsorption of urea and key intermediates, thus promoting UOR. To investigate the potential for the utilization of clean light energy and industrial applications, we assembled a photovoltaic-electrolysis (PV-EC) system and an anion-exchange membrane (AEM) electrolyzer, comprising A-NiCoMn-TAC/LDH as the anode and its derived catalyst A-NiCoMn-TAC/LDA as the cathode, and it could generate high-performance urea-assisted energy-saving hydrogen.

Results and discussion

Synthesis, morphology and structure of A-NiCoMn-TAC/LDH

As shown in Fig. 1a, A-NiCoMn-TAC/LDH was synthesized through a coprecipitation process using the corresponding metal salts as precursors. A formamide solution was prepared in which formamide could effectively break the hydrogen bonds formed between LDH layers, thus contributing to peeling into an isolated lamellar structure. Through the addition of a Mn salt and changing its molar number in the reaction mixture, A-NiCoMn-TAC/LDH-x (x = 1, 2, 3, or 4) incorporated with various Mn amounts was obtained (Fig. S3). For convenience, A-NiCoMn-TAC/LDH-2 was also denoted as A-NiCoMn-TAC/LDH. X-ray diffraction (XRD) patterns were collected to study the crystalline structure of the synthesized samples (Figs. 1b and S1). The absence of (003) peaks in the XRD patterns provides strong evidence for the monolayer characteristic of A-NiCoMn-TAC/LDH-x, indicating the absence of lamellar stacking. The Fourier transform infrared (FT-IR) spectra (Figs. 1c and S2) show similar vibrations, among which the peak at 1384 cm−1 confirms the presence of NO3-. High-resolution transmission electron microscopy (HRTEM) was used to examine the morphology and crystal planes of the catalysts. As shown in Fig. 1d, the HRTEM image of A-NiCoMn-TAC/LDH shows a single-layer nanosheet morphology. The characteristic (018) and (015) lattice fringes serve as a hallmark of the LDH structure (Fig. 1e), which is consistent with NiCo-LDH (Fig. S6). The transmission electron microscopy (TEM) images of other samples (Fig. S5) and the detailed nanosheet morphology (Figs. S6–S11) demonstrate monolayer structures of NiCo-LDH, A-NiCoMn-TAC/LDH and A-NiCoMn-TAC/LDH-x. Furthermore, the solutions of NiCo-LDH and A-NiCoMn-TAC/LDH-x present the specific Tyndall effect characteristic of colloids, indicating their highly dispersed nature attributed to the monolayer structure (Fig. S5). The thickness of A-NiCoMn-TAC/LDH was further measured using atomic force microscope (AFM). The height profiles obtained (Fig. 1f) show that the thickness of A-NiCoMn-TAC/LDH is close to 1 nm, which is in good agreement with the TEM results and confirms the monolayer LDH structure. In a monolayer structure, the surface atoms are expected to exhibit unsaturated coordination. Additionally, the single-layer structure is enriched in edge sites, which are more susceptible to detachment and thus more likely to lead to vacancy formation. Moreover, these structures are prone to bending, and the resulting tensile or compressive strains in bent regions can promote the formation of vacancies and defects. The elemental composition ratio was determined using the inductively coupled plasma‒atomic emission spectrometry (ICP‒AES, Table S1), and the results revealed that the mass fraction of Mn in A-NiCoMn-TAC/LDH was ~0.2%.

Fig. 1: Schematic illustration of the synthesis, morphology, and structure characterizations of A-NiCoMn-TAC/LDH.
Fig. 1: Schematic illustration of the synthesis, morphology, and structure characterizations of A-NiCoMn-TAC/LDH.
Full size image

a Synthetic illustration of NiCo-LDH and A-NiCoMn-TAC/LDH. b XRD patterns of NiCo-LDH and A-NiCoMn-TAC/LDH. c FT-IR spectra of NiCo-LDH and A-NiCoMn-TAC/LDH. d HRTEM images and e measurement of lattice fringes (the embedded diffraction pattern is the corresponding fast Fourier transform diagram) of A-NiCoMn-TAC/LDH. f AFM image and g the height profile measurement of A-NiCoMn-TAC/LDH. h EDS mapping images of A-NiCoMn-TAC/LDH.

XPS revealed that the binding energies of Ni 2p and Co 2p increase after the introduction of Mn, suggesting that the valence states of Ni and Co are elevated (Figs. S12 and 13). X-ray absorption near-edge spectroscopy (XANES) reveals that after Mn incorporation, the absorption edges for Ni and Co shift toward higher binding energies in A-NiCoMn-TAC/LDH, indicating that the addition of Mn increases the valence of Ni and Co, which is consistent with the XPS results. Furthermore, linear oxidation state fitting based on the first-derivative XANES curves reveal that the valence of Ni in NiCo-LDH is +2.03, and it slightly increases in A-NiCoMn- TAC/LDH to +2.31 (Fig. S14). Similarly, Co has a higher oxidation state of +2.47 in A-NiCoMn-TAC/LDH than that of +2.41 in NiCo-LDH (Fig. S15). The K-edge spectrum for Mn compared to reference samples, including MnO, Mn2O3 and MnO2, shows an average valence state of approximately +3.29 for Mn in A-NiCoMn-TAC/LDH (Fig. S16). Moreover, a small amount of Mn (IV) is present in the lattice, which distorts the laminate structure of A-NiCoMn-TAC/LDH, and this distortion has been confirmed to facilitate the catalytic reaction26.

To further clarify the local coordination environment, we performed detailed EXAFS analysis. The Fourier transform (FT) k3-weighted EXAFS spectra (Figs. S17–S24) indicates high-quality data collection across all the measurements. EXAFS fitting of Ni (Fig. 2b) reveals that, compared with NiCo-LDH, the first-shell Ni‒O bond distance for A-NiCoMn-TAC/LDH is elongated to 1.99 ± 0.02 Å accompanied by a reduced coordination number (CN) of 4.7 ± 0.4. Similarly, Co‒O bonds in the Mn-doped sample show a longer distance (1.93 ± 0.02 Å) and a lower CN (4.3 ± 0.3 in Fig. 2e). The decreased CN of M‒O bonds indicate an increase in oxygen vacancy (Ov) and a distortion of octahedral coordination around Ni/Co, which is consistent with a more disordered local structure induced by Mn incorporation. For the second-shell M‒M coordination, Ni‒M bonds in A-NiCoMn-TAC/LDH show a longer distance (3.15 ± 0.02 Å) and a reduced CN of (4.2 ± 0.5). Similarly, Co‒M bonds exhibit an elongated distance (3.19 ± 0.02 Å) and lower CN (4.6 ± 0.4). The increased distance and decreased CN of M‒M bonds suggest that Mn ions are incorporated into the laminate lattice to replace partial Ni/Co sites, which leads to an increase in metal vacancy. The coexistence of abundant oxygen and metal vacancies, together with lattice distortion, enhances structural asymmetry and facilitates the formation of asymmetric Ni–Co–Mn trimetallic coordination sites.

Fig. 2: XAS analysis.
Fig. 2: XAS analysis.
Full size image

a Normalized Ni K-edge XANES oscillation functions k3χ(k) for NiCo-LDH, A-NiCoMn-TAC/LDH, Ni foil and Ni(OH)2. b FT-EXAFS fitting curves of A-NiCoMn-TAC/LDH in Ni R-space, e Co R-space, and h Mn R-space. WT-EXAFS spectra of A-NiCoMn-TAC/LDH for c Ni, f Co, and i Mn. d Normalized Co K-edge XANES oscillation functions k3χ(k) for NiCo-LDH, A-NiCoMn-TAC/LDH, Co foil and Co(OH)2. g Normalized Mn K-edge XANES oscillation functions k3χ(k) for Mn foil, MnO, Mn2O3, MnO2 and A-NiCoMn-TAC/LDH. j Schematic diagram of the trimetallic atom sites structure with oxygen vacancies and metal defects of A-NiCoMn-TAC/LDH. k Scheme of the electronic interactions among Ni, Co, Mn and O in NiCo-LDH and A-NiCoMn-TAC/LDH.

To further elucidate the local coordination environment and distinguish the nature of backscattering atoms, wavelet transform (WT) analysis was performed based on the Ni and Co K-edge EXAFS spectra. For the Ni K-edge (Fig. 2c), the WT contour plot reveals two prominent intensity maxima: the first located at R ≈ 1.5 Å, corresponding to backscattering from light O atoms (Ni–O bonds), and the second at R ≈ 2.7 Å, attributed to backscattering from heavier metal atoms (Ni–M bonds). Compared with NiCo-LDH (Fig. S26), the second maximum in A-NiCoMn-TAC/LDH exhibits a distinct shift toward a higher k value (10 Å−1 vs. 9 Å−1), characteristic of Ni–Mn interactions. This shift provides compelling evidence for the incorporation of Mn into the second coordination shell of Ni, in addition to Co. Similarly, for the Co K-edge (Fig. 2f), the WT map also displays two intensity maxima: the first at R ≈ 1.45 Å corresponding to Co–O coordination, and the second at R ≈ 2.6 Å associated with Co–M interactions. Compared to NiCo-LDH (Fig. S27), the second maximum in A-NiCoMn-TAC/LDH shifts to a higher k value (9.5 Å−1 vs. 8 Å−1), suggesting contributions from both Co–Ni and Co–Mn backscattering pathways. This shift arises from the different backscattering strengths of Mn and Ni at higher k values, indicating a more disordered and compositionally varied second coordination shell around Co, which supports the formation of an asymmetric Ni–Co–Mn trimetallic arrangement. Collectively, the EXAFS and WT-EXAFS results reveal that Mn incorporation enhances the asymmetric Ni-Co coordination in NiCo-LDH, introducing Ni–Mn and Co–Mn interactions. The coexistence of multiple metal–metal bonds with distinct bond lengths and disorder leads to asymmetric coordination environments around Ni and Co. Consequently, we propose an asymmetric coordination trimetallic atom sites model consisting of Ni‒Co‒Mn enriched with vacancies (i.e., oxide and metal vacancies) for A-NiCoMn-TAC/LDH (Fig. 2j).

Based on the XAFS results, it is speculated that strong interactions exist between metals and between metals and oxygen in A-NiCoMn-TAC/LDH. As is demonstrated in Fig. 2k. The π-symmetry (t2g) d-orbitals of Ni2+ are fully occupied, which induces strong e−‒e− repulsion between Ni2+ and bridging O2−. The configuration of the d-orbital of Co3+ is t2g5eg1, and its π-symmetry (t2g) d-orbital possesses one unpaired electron; thus, Co3+ can interact with bridging O2− via weak π-donation27. After incorporating Mn, Mn3+ can interact with the adjacent O2- via strong π-donation due to its π-symmetry (t2g) d-orbital with two unpaired electrons28. In such cases, the partial electrons of Ni2+ and Co3+ transfer to Mn3+ through the Ni‒O‒Mn‒O‒Co unit, causing an increase in the valence states of Ni2+ and Co3+ and a decrease in the Mn3+. Similarly, for Mn4+, its π-symmetry (t2g) d-orbitals have three unpaired electrons, and eg orbitals are empty; thus, it is also linked to bridging O2− through strong π-donation29, causing an increase in the Ni2+ and Co3+ oxidation states. Overall, the establishment of asymmetric coordination trimetallic atom sites trigger efficient electron transfer, forming higher-valence state metal sites, particularly Ni, which is favorable for the subsequent electrocatalytic reaction30,31,32.

Electrocatalytic performance

The electrocatalytic performance of the synthesized samples was evaluated through a three-electrode configuration (Fig. S29). Figure 3a shows the linear sweep voltammetry (LSV) curves of NiCo-LDH with different Mn doping contents in KOH with urea. All the Mn-doped samples exhibit lower potentials than NiCo-LDH at the same current density, suggesting that Mn incorporation can indeed enhance the electrochemical activity of the UOR. In particular, the A-NiCoMn-TAC/LDH electrode exhibits the highest activity among the A-NiCoMn-TAC/LDH-x samples. Specifically, for 10, 50, and 100 mA cm−2, the potentials needed for A-NiCoMn-TAC/LDH are 1.26 ± 0.01, 1.29 ± 0.01, and 1.33 ± 0.03 V, respectively, which are lower than the corresponding potentials of the other catalysts. LSV curves of A-NiCoMn-TAC/LDH for the OER and UOR were used for comparison (Fig. 3b). Without urea, the potentials delivering 10 and 50 mA cm−2 on A-NiCoMn-TAC/LDH are 1.43 ± 0.04 V and 1.83 ± 0.02 V, respectively, suggesting mediocre electrocatalytic OER activity33. An obvious negative shift in the potential of 533 ± 16 mV for the UOR is observed compared with that for the OER at a current density of 50 mA cm−2, indicating the feasibility of the UOR instead of the OER as an energy-saving electrolytic hydrogen production strategy34. Moreover, the overall performance of A-NiCoMn-TAC/LDH is among the UOR electrocatalysts, including noble and non-noble metal-based electrocatalysts (Table S8). The turnover frequency (TOF) was further calculated to evaluate the intrinsic activity of the catalysts35. The TOF value of A-NiCoMn-TAC/LDH is 0.038 ± 0.001 s−1 at 1.35 V, which is better than many reported catalysts (Table S10).

Fig. 3: Electrochemical performance of A-NiCoMn-TAC/LDH.
Fig. 3: Electrochemical performance of A-NiCoMn-TAC/LDH.
Full size image

a LSV curves of NiCo-LDH and A-NiCoMn-TAC/LDH-1, 2, 3, 4 in 1 M KOH and 0.33 M urea. b Comparison of OER and UOR performance of A-NiCoMn-TAC/LDH at 50 mA cm−2. c Radar chart for comparison of comprehensive catalyst performance of samples. d Stability test of A-NiCoMn-TAC/LDH at an industrial current density at 500 mA cm−2 for 600 h.

Tafel slope of A-NiCoMn-TAC/LDH is only 58.9 ± 2.3 mV dec−1, which is obviously smaller than those of other samples. Given that the Tafel slope reflects intrinsic reaction kinetics, the smaller slope of A-NiCoMn-TAC/LDH indicates accelerated kinetics and superior urea-oxidation activity relative to the other samples. This is in line with its largest electrochemically active surface area (estimated from double-layer capacitance, Fig. S30) and the lowest charge-transfer resistance. (Fig. S31a). As Fig. S32 showed, the electrochemical impedance spectroscopy (EIS) result show that the solution resistance (Rs) of A-NiCoMn-TAC/LDH is 1.49 ± 0.02 Ω. The stability of the electrocatalytic UOR of monolayer A-NiCoMn-TAC/LDH was subsequently tested at 10 mA cm−2, 200 mA cm−2 (Fig. S33) and an industrial-level current density of 500 mA cm−2 (Fig. 3d). A-NiCoMn-TAC/LDH exhibits reliable electrocatalytic reaction stability, maintaining performance for 600 h. This is the only UOR catalyst published to date that has operated at such a high current density over 600 h (Fig. S34 and Table S9). For further evaluation, we compared the potentials at current densities of 10 and 100 mA cm−2 for the A-NiCoMn-TAC/LDH synthesized in this study with those published in major high-quality journals in recent years (Fig. S35). A-NiCoMn-TAC/LDH clearly has a low onset potential, and its potential at 100 mA cm−2 is highly advantageous. A series of characterizations, including XRD, HRTEM, XPS and in situ Raman spectroscopy, were performed to explore the structural changes after UOR (Figs. S36–S40). The results demonstrate the formation of nickel (oxy)hydroxide with the incorporation of cobalt and manganese after UOR (the detailed analyzes are shown in the figure captions of Figs. S36–S40), which agrees with the previous literature8,36,37,38,39,40,41,42.

Chronoamperometry was employed to monitor the products of the UOR under applied potentials of 1.4 and 1.6 V vs. RHE. The gaseous products, including N2 and O2, were analyzed using gas chromatography (GC), whereas ion chromatography (IC) was used to detect potential anionic species in the electrolyte as liquid products43,44. For gaseous products, as shown in Fig. S41, GC analysis indicated that when the catalytic system was exposed to air, the volume ratio of O2 to N2 was approximately to 1: 4, corresponding to the background signal. At 1.4 V (Fig. S41a), a significantly increased N2 peak was clearly observed, suggesting the production of a large amount of N2, and only trace amounts of O2 were detected. At 1.6 V (Fig. S42b), a higher N2 peak was observed along with a slightly increased O2 signal, suggesting that the OER slightly occurs at this higher potential. Overall, the increase in the N2 peak area is markedly more pronounced than that of O2 compared with the background level, suggesting that the UOR process predominates under the applied potentials. The nitrogen-containing oxidation products include nitrogen (N2), nitrite (NO2−), nitrate (NO3−), and cyanate (CNO−), among which N2 is the major species. The IC results of the liquid anionic products are shown in Fig. S42a, the main nitrogen-containing products are NO2− with a minimal amount of NO3− at 1.4 V. The NO3− level increases at 1.6 V (Fig. S42b). The carbon-containing products are primarily CNO− and carbonate (CO32−). The formation of CNO- likely results from C‒N bond cleavage following partial dehydrogenation of urea. These liquid products are quantified by the standard curves (Fig. S42c–e) Since the oxidation states of C and N in these species are the same as those in urea (+4 and −3), they were not included in the calculation of Faraday efficiencies (FEs). Similarly, CO32− originates from the reaction of dissolved CO2 in KOH, where the oxidation state of C is unchanged at +4; thus, it was also excluded from the FE calculations. On this basis, the FEs for the UOR products were calculated. As shown in Fig. S43, the FEs (N2) reach as high as 94.2 and 92.1% at 1.4 and 1.6 V, respectively, indicating that the dominant product of N-containing UOR products is environmentally benign N2. The small FE (NO2−) values of 5.3 and 4.4% at 1.4 and 1.6 V, respectively, suggest that only a limited number of urea molecules undergo C‒N bond cleavage rather than dehydrogenation45. The negligible FE (NO3−) values of 0.45 and 1.2% indicate minimal overoxidation of NO2−. The FEs of NO3− and NO2− are significantly lower than those reported in the literature44,45,46,47, highlighting that this study simultaneously achieved treatment of urea wastewater and low-energy hydrogen production.

In situ characterization measurements and analysis

To elucidate the origin of the high UOR performance and reaction mechanism leading to the formation of asymmetric coordination trimetallic atom sites, we investigated the behaviors of the adsorbed species on the catalyst surface during the electrochemical process through in situ ATR-SEIRAS measurements (Fig. S44). Figures 4a and S45a show the spectra collected for NiCo-LDH and A-NiCoMn-TAC/LDH under various applied potentials. The band located at 1720 cm−1 corresponds to the vibration of C=O from urea and intermediate species containing the C=O group48. The peaks at 1497 and 1650 cm−1 are attributed to the reactant and intermediates with C–N and N–H49. Compared with those of NiCo-LDH, the above peaks are stronger under UOR conditions, indicating that urea oxidation is faster on A-NiCoMn-TAC/LDH, thus producing more intermediates. Moreover, the bands at 1380 cm−1 can be assigned to the absorbed CO32− species, which arises from the reaction between the absorbed CO2 and OH− in solution50. The peaks associated with CO32− for A-NiCoMn-TAC/LDH are less pronounced than those for NiCo-LDH, providing direct evidence that Mn incorporation facilitates the desorption of CO2, thus enhancing the catalytic performance51.

Fig. 4: In-situ ATR-SEIRAS spectra and operando analysis of the dynamic evolution of the relationship among surface species and electrolyte, electron transfer during OER and UOR processes for A-NiCoMn-TAC/LDH.
Fig. 4: In-situ ATR-SEIRAS spectra and operando analysis of the dynamic evolution of the relationship among surface species and electrolyte, electron transfer during OER and UOR processes for A-NiCoMn-TAC/LDH.
Full size image

a In the range of 1000 to 4000 cm−1 during UOR. b Regional in-situ ATR-FTIR spectra and c at different applied potentials during UOR. d Bode plots of A-NiCoMn-TAC/LDH in KOH. e The corresponding plots of A-NiCoMn-TAC/LDH in KOH with urea (The blue, red and yellow areas respectively represent the low, medium and high frequency regions).

To analyze the electron transfer related to the dynamic evolution of electrode surface species and different reactions, operando EIS, an effective method for identifying the reaction interface, was utilized. In the Bode plots of NiCo-LDH and A-NiCoMn-TAC/LDH in KOH (Figs. S45d and 4d), a transition phase peak in the low-frequency (10−2–100 Hz) region is usually related to the nonhomogeneous charge distribution caused by the OER at 1.45 V, which implies the existence of the OER and the activity of oxyhydroxide for water oxidation52. Compared with A-NiCoMn-TAC/LDH, NiCo-LDH undergoes a passivation reaction53 above 1.55 V, demonstrating that A-NiCoMn-TAC/LDH inhibits surface remodeling and has greater stability at high potentials. Additionally, the variation in the different phase peaks in the mid-frequency region (100–103 Hz) can be attributed to the inevitable self-oxidation reaction of the Ni species.

This is an indication of the conversion of low-conductivity hydroxide to high-conductivity oxyhydroxide54, and simultaneously, A-NiCoMn-TAC/LDH more easily forms reactive oxyhydroxides than NiCo-LDH does. In addition, in a urea solution with KOH, the Bode plot phase angle sharply decreases in the low-frequency region compared with that of the OER (Fig. 4d, e), revealing that the UOR is more likely to occur than the OER55. In addition, the phase angle of A-NiCoMn-TAC/LDH decreases more rapidly than that of NiCo-LDH does (Figs. 4e and S45e), indicating that A-NiCoMn-TAC/LDH has a faster electron transfer rate and reduces the accumulation of oxyhydroxide more than NiCo-LDH does. Additionally, a transition phase peak of the self-oxidation reaction of the Ni species is not observed in the mid-frequency region, providing robust evidence that electron transfer during the UOR triggered by Ni3+ is nearly barrier-free for A-NiCoMn-TAC/LDH44. Furthermore, a small semicircle is evident in the high-frequency region (105–106 Hz) of the Bode plots in KOH or urea, which may be due to the oxidation of Co species in both NiCo-LDH and A-NiCoMn-TAC/LDH16.

To probe the evolution of the local atomic and electronic structures of Ni, Co, and Mn in A-NiCoMn-TAC/LDH during UOR, in situ XAS was performed to track their valence states and coordination environments (Fig. S46)56. The in situ XANES spectra (Fig. 5a–c) display pronounced edge shifts for all three metals as the potential increases57,58. At open-circuit potential (OCP), the absorption edges are already positioned at higher energies compared to the ex situ spectra, reflecting the high surface exposure of the monolayer framework and its propensity for hydroxide-induced deprotonation. Upon polarization, the Ni K-edge exhibits a continuous positive shift, indicative of progressive oxidation to higher valence states. In contrast, Co and Mn edges initially shift negatively between 1.2 and 1.5 V before increasing again at higher potentials, a behavior consistent with Ov formation and transient electron back-donation to these centers. The first-derivative XANES curves (Fig. 5d–f) show that the oxidation states of Ni, Co, and Mn at different voltages are between those of the metal and oxide reference samples. Linear combination fitting reveals that Ni undergoes monotonic oxidation from OCP to 1.8 V. Co and Mn, in contrast, show a short-lived reduction at intermediate potentials, and then oxidize again at higher potentials. The in situ EXAFS spectra exhibit systematic amplitude and phase changes with potential, signaling local coordination rearrangements59. The Ni–O scattering path progressively shortens, mitigating lattice distortions caused by Ni oxidation and thereby enhancing structural stability. Co and Mn display similar trends, highlighting a concerted structural response within the trimetallic network (Fig. 5j–l). Quantitative EXAFS fitting (Figs. S47–S50 and Tables S5–S7) shows that the Ni–O coordination number increases from OCP to 1.2 V, consistent with urea adsorption. From 1.2 to 1.5 V, the Ni–O, Co–O, and Mn–O coordination numbers all decrease, indicating the generation of abundant Ov—particularly within Ni–O–Mn–O–Co bridging motifs—that act as transient electron reservoirs. The presence of these vacancies is further confirmed by room-temperature EPR spectra (Fig. S51), which show a significant increase in vacancy content after the 1.5 V test60. Electro-oxidative urea decomposition releases electrons, which are temporarily accommodated to maintain charge balance—most prominently within Ni–O–Mn–O–Co bridges—thereby generating oxygen vacancies around the metal centers.

Fig. 5: In-situ XAS analysis.
Fig. 5: In-situ XAS analysis.
Full size image

a–c Normalized Ni, Co and Mn K-edge XANES oscillation functions k3χ(k) of A-NiCoMn-TAC/LDH. d–f The first derivative XANES curves of A-NiCoMn-TAC/LDH. g–i The liner oxidation state fittings of A-NiCoMn-TAC/LDH at various applied potentials. j–l The corresponding magnitude of the FT of A-NiCoMn-TAC/LDH. m The fitting of coordination and oxidation state of Ni, Co and Mn for A-NiCoMn-TAC/LDH during UOR. n The structure evolution of A-NiCoMn-TAC/LDH with different applied potentials during UOR.

As shown in Fig. 5m, all their oxidation states overall increase from OCP to 1.8 V which indicates the electro-oxidation dehydrogenation of surface –OH61. The energy level of the unoccupied dx2−y2 orbital of Co (III) has been reported to be lower than that of Ni (III); thus, the excess electron prefers to occupy the lowest unoccupied orbital of Co (III), resulting in a decrease in the valence state of Co. When the applied potential from 1.2 to 1.5 V, initially Co (III) and Ni (III) atoms transition from the low-spin (LS) to intermediate-spin (IS) state. Likewise, Mn (III) with a 3d4 orbital of t2g3eg1 tends to gain excess electrons to fill the eg orbitals to form a half-full stable state of 3d5, thus causing a decrease in the Mn valence state62. As mentioned above, Co (III) and Mn (III) more easily obtain excessive electrons from Ov than Ni does, leading to valence reduction. Based on the above results, the evolution might be analyzed as follows. When a potential is applied to A-NiCoMn-TAC/LDH, dehydrogenation occurs, accompanied by the oxidation of metal active sites including Ni2+ to Ni3+, thereby initiating the UOR63. The CN of Ni–O path increases from OCP to 1.2 V, showing the adsorption of urea molecules on the catalyst surface. When the applied voltage further increases, urea releases electrons due to electro-oxidative decomposition, which then generates massive Ov around metal atoms to maintain charge balance64,65, especially Ni–O–Mn–O–Co bridge bonds for temporary storage of electrons (Fig. 5n). As the potential further increases, additional Ov are generated, leading to a slight decrease in the valence states of Co and Mn. Finally, with the further increase in potential, the valence states of Co and Mn also increase.

Theoretical study

To reveal the performance of urea oxidation on A-NiCoMn-TAC/LDH enriched with defects, we carried out theoretical calculations to demonstrate the electronic modulations among the heterotrimetallic atoms and the corresponding influences on the reaction trends. Currently, it is challenging to reproduce the practical catalytic environments in DFT calculations, and our calculations aim to supply qualitative supportive evidence to experimental results for understanding the UOR process on A-NiCoMn-TAC/LDH. Firstly, we demonstrate the bonding and antibonding orbital distributions near the Fermi level (EF). For defective NiCo-LDH, the bonding orbitals and antibonding orbitals are dominated by Ni sites and surface OH groups, whereas Co has a limited contribution to the electroactive electrons (Fig. 6a and Supplementary Data 1). With the introduction of Mn doping on the material surface, evident modulations of the surface electronic distributions occur (Fig. 6b and Supplementary Data 2). Although the bonding orbitals are still dominated by the Ni sites, the antibonding orbitals shift toward the Co and Ni sites near defects and their neighboring OH sites, promoting the orbital coupling of bonding and antibonding orbitals toward more efficient electron transfer. To explore the detailed electronic structures, we further examined the projected partial density of states (PDOS)for defective NiCo-LDH and A-NiCoMn-TAC/LDH (Fig. 6c).

Fig. 6: Theoretical insights into the UOR mechanism on A-NiCoMn-TAC/LDH.
Fig. 6: Theoretical insights into the UOR mechanism on A-NiCoMn-TAC/LDH.
Full size image

The electronic distributions of bonding and anti-bonding orbitals near the Fermi level on a defective NiCo-LDH, and b defective A-NiCoMn-TAC/LDH. Blue balls = Co, purple balls = Ni, brown balls = Mn, red balls = O, and white balls = H. Blue iso-surface = bonding orbitals, and green iso-surface = anti-bonding orbitals. PDOS of c defective NiCo-LDH, and d defective A-NiCoMn-TAC/LDH. Comparisons of PDOS for neighboring metal sites with e VCo, f VNi, and g Ov. h The d-band and p-band centre comparisons. i The PDOS of key intermediates of urea oxidation on the defective A-NiCoMn-TAC/LDH. j The formation energy of defects in NiCo-LDH and A-NiCoMn-TAC/LDH. k The adsorption energy of urea and OH* on defective NiCo-LDH and defective A-NiCoMn-TAC/LDH. l The reaction energy change of urea on defective NiCo-LDH and defective A-NiCoMn-TAC/LDH. Insets are the adsorption configurations of intermediates on A-NiCoMn-TAC/LDH. Blue balls = Co, purple balls = Ni, brown balls = Mn, grey balls = C, dark blue balls = N, red balls = O, and white balls = H.

The O-2p orbitals are located at deeper positions far from the EF, which overlap well with the 3d orbitals of both Co and Ni sites between an Ev of −4 and −6 eV. After introducing Mn into the structure, the Ni-3d orbitals display a slight downward trend, whereas the Co-3d orbitals exhibit the opposite trend, leading to stronger d-d orbital coupling with alleviated barriers for site-to-site electron transfer (Fig. 6d). The Mn-3d orbitals are located between the Co-3d and Ni-3d orbitals at an Ev of −2.19 eV, which promotes orbital coupling among different metal sites to supply highly electroactive surfaces66. The O-2p orbitals are not evidently affected, indicating that Mn doping does not affect the stability of the LDH structures. To elucidate the contributions of defects, we investigated the electronic modulations of metal sites near different defects in both NiCo-LDH and A-NiCoMn-TAC/LDH. For the Co vacancy (VCo), the defect induces the upshift of the Ni-3d orbitals and the downshift of the Co-3d orbitals near the Ev of −2.2 eV (Ev = 0 eV) in NiCo-LDH (Fig. 6e). In A-NiCoMn-TAC/LDH, the Ni-3d orbitals display a relatively weaker upshifting trend induced by Mn doping, whereas the Co-3d orbitals remain similar. With the formation of a Ni vacancy (VNi), a similar modulation trend is also observed in the Ni-3d orbitals, suggesting that the Ni sites are most sensitive to defects and doping (Fig. 6f). Moreover, the Co-3d orbitals remain like those of perfect NiCo-LDH, where a closer distance between the d-d orbitals is observed for the Co and Ni sites to support enhanced electron transport. Compared with VCo, VNi evidently upshifts the Mn-3d orbitals, which enhances the overlap with the Ni-3d and Co-3d orbitals. In contrast, the modulations induced by Ov are different (Fig. 6g). Notably, the formation of Ov also results in much stronger modulations in the Ni-3d orbital than in the Co-3d orbital. However, the further introduction of Mn incorporation with Ov has limited influence on the electronic structures of the metal active sites for both the Ni-3d and Co-3d orbitals. The d-band centers of metals and p-band centers of O were compared, and the opposite trend of d-band centers was observed for Ni and Co induced by defects (Fig. 6h). The d-band evolution of Ni sites is much stronger than that of Co sites, resulting in a higher sensitivity to nearby defects67.

Combining the defects and Mn incorporation, the d-band center difference between Co and Ni is largely alleviated, leading to much stronger d-d orbital overlap68. Notably, the formation of defects also results in the downshifting of Mn-3d orbitals to increase the electron transfer efficiency. For surface O sites, the existence of defects has induced significant upshifting with increased electroactivity to benefit urea oxidation processes. Moreover, the corresponding PDOSs of key intermediates of urea oxidation were determined (Fig. 6i). With the gradual oxidation of urea molecules, the corresponding σ orbitals of intermediates tend to linearly upshift, which supports efficient oxidation processes with small energy barriers69. From the energetic perspective, the energy costs for different levels of formed defects have been calculated using the formation energy, where values that are more negative indicate easier formation of defects (Fig. 6j). Apparently, all the vacancy defects become easier to form after Mn incorporation, which agrees well with the increasing defect concentrations in the experiments. In particular, the formation of multiple defects becomes considerably favored with Mn incorporation, which further improves the surface electroactivity for the UOR. Defective A-NiCoMn-TAC/LDH also displays a stronger affinity for urea, which is highly beneficial for the activation of urea molecules during electrocatalysis (Fig. 6k). In comparison, defective NiCo-LDH shows much stronger OH* adsorption70, which is even stronger than that of urea, leading to blockage of active sites and affecting the performance of the UOR. The adsorption configurations of the intermediates are presented in Figs. S52 and S53. On the basis of the reaction energy trend, conversion to CONH2NH*, CON2*, and COOH* meets the energy barriers, among which the formation of CON2* has the largest energy barrier as the rate-determining step (RDS) (Fig. 6l). Compared with that of the defective NiCo-LDH (1.14 eV), the energy barrier of the RDS for the defective A-NiCoMn-TAC/LDH (0.55 eV) is considerably lower, supporting the largely promoted electrocatalysis of the defective A-NiCoMn-TAC/LDH toward urea oxidation.

Practical device performance for UOR-assisted HER

While LDHs commonly exhibit limited HER performance, a key advantage is their facile convertibility to desired materials71. Motivated by this, we subjected A-NiCoMn-TAC/LDH to hydrogen calcination across multiple temperatures (300, 400, 500, 600 and 700 °C, denoted as A-NiCoMn-TAC/LDA-x, where x represents the calcination temperature) to prepare high-activity HER electrocatalysts for two-electrode electrolysis using non-noble components. The XRD patterns (Fig. 7a) reveal that A-NiCoMn-TAC alloy phases formed at temperatures above 400 °C72, and the morphologies of the nanoparticles gradually increased in size with increasing temperature (Fig. S54). Then, their HER activities were tested (Fig. 7b). The pristine A-NiCoMn-TAC/LDH has poor HER activity, whereas after calcination, the samples showed significantly improved activities. In particular, A-NiCoMn-TAC/LDA-700 has the lowest overpotentials (e.g., 0.19 ± 0.01 V at 50 mA cm−2 and 0.21 ± 0.01 V at 100 mA cm−2), which are better than the corresponding values of the other samples (0.22 ± 0.01 and 0.26 ± 0.02 V for A-NiCoMn-TAC/LDA-300, 0.24 ± 0.01 and 0.29 ± 0.02 V for A-NiCoMn-TAC/LDA-400, 0.21 ± 0.01 and 0.25 ± 0.01 V for A-NiCoMn-TAC/LDA-500, and 0.20 ± 0.01 and 0.22 ± 0.01 V for A-NiCoMn-TAC/LDA-600). Additionally, the electrochemical impedance spectra of A-NiCoMn-TAC/LDA-x calcined at different temperatures show that A-NiCoMn-TAC/LDA-700 has the smallest semicircle, indicating that it has the fastest charge transfer ability73 among the A-NiCoMn-TAC/LDA-x samples (Fig. S55). And based on three independent tests, the Rs is 1.95 ± 0.01 Ω (Fig. S56). Given the decent performance of A-NiCoMn-TAC/LDH and A-NiCoMn-TAC/LDA-700 for the UOR and HER, respectively, a two-electrode system was constructed by employing A-NiCoMn-TAC/LDH as the anode and A-NiCoMn-TAC/LDA-700 as the cathode. The LSV curves (Fig. 7c) show that a cell voltage of 1.86 ± 0.02 V is needed to reach 25 mA cm−2 in KOH, whereas it decreases to 1.56 ± 0.01 V after urea is added to KOH, which again presents the advantage of UOR replacement.

Fig. 7: Electrochemical measurement of PV-EC device and AEM electrolyzer.
Fig. 7: Electrochemical measurement of PV-EC device and AEM electrolyzer.
Full size image

a XRD patterns of calcinated products of A-NiCoMn-TAC/LDA. b HER performance LSV curves of NiCo-LDH and A-NiCoMn-TAC/LDA-x (x = 300, 400, 500, 600 and 700). c LSV curves of A-NiCoMn-TAC/LDH||A-NiCoMn-TAC/LDA-700 in KOH with urea. d Schematic diagram of the PV-EC system. e The current density-time curves of the constructed PV-EC system under chopped illumination. f LSV curves measured by AEM electrolyzer of A-NiCoMn-TAC/LDH||A-NiCoMn-TAC/LDA-700 in KOH and KOH with urea. g Schematic illustration of AEM electrolyzer. h Durability of the urea assisted AEM electrolyzer at 1000 mA cm−2 for 1500 h in 1 M KOH with 0.33 M urea.

To utilize solar power, the system employed a III–V GaInP2/GaAs/Ge triple-junction cell to drive the two-electrode configuration (Fig. 7d). Illumination immediately produced an anodic photocurrent, which was greater in KOH containing urea than in KOH (Fig. 7e). The PV–EC demonstration (Fig. S57) indicates that substituting the anodic OER with UOR can reduce energy input in solar-driven hydrogen production74,75 and supports future application to urea-wastewater electrolysis and environmental protection76. Furthermore, to demonstrate the potential of this catalyst for practical industrial applications77, an AEM flow electrolyzer was assembled (Figs. S58 and 7g). In this setup, A-NiCoMn-TAC/LDH was employed as the anodic UOR catalyst, whereas A-NiCoMn-TAC/LDA served as the cathodic HER catalyst. With the aid of electrolyte flow to enhance mass transfer, the urea-assisted water electrolyzer demands only a cell voltage of 1.516 ± 0.003 V to reach a current density of 500 mA cm−2, which is 291 ± 5 mV lower than that required by the water-splitting system (1.807 ± 0.002 V), leading to a savings of 16.1 ± 0.28% (Fig. 7f). Moreover, the electricity expense of each electrolysis system for H2 production was calculated based on the cell voltage. The results indicate that A-NiCoMn-TAC/LDH||A-NiCoMn-TAC/LDA has the lowest energy cost (3.62 ± 0.01 kWh Nm−3), which reveals a prominent economic benefit over water electrolysis and commercial RuO2/Pt urea electrolysis (4.14 k ± 0.01 Wh Nm−3)78. Moreover, this electrolyzer can achieve a high current density of 1000 mA cm−2 at a cell voltage of 1.79 ± 0.01 V, which is substantially higher than the 578 mA cm−2 attained by the RuO2||Pt-C assembly under identical conditions. The AEM electrolyzer of A-NiCoMn-TAC/LDH||A-NiCoMn-TAC/LDA exhibits reliable stability, with almost no performance degradation observed at an industrial current density of 1000 mA cm−2 over a period of 1500 h (Fig. 7h), which is significantly better than those of other systems reported thus far (Fig. S59). The above results demonstrate the promising potential of asymmetrically coordinated trimetallic atom site catalysts in the utilization of solar energy and industrial urea-assisted energy-saving hydrogen generation applications.

Discussion

In conclusion, we prepared an asymmetrically coordinated trimetallic atom sites ternary mono-layer A-NiCoMn-TAC/LDH with high UOR performance (1.26 V at 10 mA cm−2) and reliable stability (up to 600 h at a current density of 500 mA cm−2) by the coprecipitation method. The monolayer and trimetallic atom site structures were analyzed by XRD, AFM, HRTEM, XPS, XAFS and other methods. The structural changes and performance enhancement mechanism after Mn incorporation during the UOR process were further investigated by in situ XAS, operando EIS and EPR tests. The results show that the introduction of Mn increased the number of oxygen vacancies and metal vacancies and that Mn3+ with a 3d4 orbital promoted the electron transfer of the Ni and Co sites. ATR-FTIR confirmed that trimetallic sites promoted CO2 desorption, thus accelerating the regeneration of active sites and enhancing the catalytic performance. In situ XAS monitoring revealed changes in the coordination environment of the Ni, Co and Mn atoms during the UOR. In DFT calculations, we systematically explored the electronic modulations and various shifts of the d-band center to EF by Mn incorporation as well as defects, revealing strong d-p-d orbital couplings between Mn, Ni, Co and O. By integrating different vacancies and Mn incorporation, the electroactivity of surface-active sites was optimized due to the decreased energy barriers for electron transfer, which further enhanced the adsorption preferences of urea and lowered the RDS barriers to achieve efficient electrocatalysis. Moreover, a extended long-term durability test of 1000 mA cm−2 over 1500 h in an AEM electrolyzer demonstrated the feasibility of industrial electrochemical treatment of urea wastewater coupled with hydrogen production. This work provides an insight into the effects of trimetallic atom sites synergy and defects in the UOR, offering an optimal scheme for the design of nonprecious metal-based UOR electrocatalysts with high performance and stability.

Methods

Chemicals

Nickel nitrate hexahydrate (99%, Aladdin), cobalt nitrate hexahydrate (99%, Aladdin), manganese nitrate tetrahydrate (98%, Aladdin), formamide (99%, Aladdin), sodium hydroxide (96%, Aladdin), potassium hydroxide (95%, Macklin), urea (95%, Aladdin), absolute alcohol (95%, Aladdin), Nafion 117 solution (5 wt%, Dupont). Deionized (DI) water (18.2 MΩ) by ultra-pure water purifier. The deionized water used in all experiments was obtained through ion-exchange and filtration. The working electrode is loaded with 1 mg catalyst, the counter electrode is Pt foil, and the reference electrode is Hg/HgO (1 M KOH). All of electrodes are made in Gaossunion (Tian Jing, China).

Preparation of NiCo-LDH by a step co-precipitation strategy

Firstly, all the water involved in the synthesis steps was decarbonized water obtained by boiling deionized water for 1.5 h. The 0.67 mmol Ni(NO3)2·6H2O and 0.33 mmol Co(NO3)2·6H2O were dissolved in 20 mL water to obtain a mixed salt solution. Under continuous magnetic stirring at 80 °C, the solution was dropwise introduced into 20 mL of an aqueous formamide solution (23 vol%) over a period of 10 min. At the same time, 0.25 mol/L sodium hydroxide solution was dropped to maintain a pH of 10. The sample was centrifuged and dried to obtain NiCo-LDH.

Preparation of A-NiCoMn-TAC/LDH by a step co-precipitation strategy

The synthesis of A-NiCoMn-TAC/LDH is like that of NiCo-LDH. The mixed salt solution consisting of the 0.67 mmol Ni(NO3)2·6H2O and 0.33 mmol Co(NO3)2·6H2O and different molar amounts (0.083, 0.167, 0.25 and 0.332 mmol) of Mn(NO3)2 4H2O was dissolved in 20 mL water. The next steps are the same as the synthesis steps of NiCo-LDH. According to the amount of Mn(NO3)2 from less to more, the synthesized materials were labeled as A-NiCoMn-TAC/LDH-1, A-NiCoMn-TAC/LDH, A-NiCoMn-TAC/LDH-3 and A-NiCoMn-TAC/LDH-4.

Characterization

XRD patterns of materials were recorded by the Rigaku XRD-6000 diffractometer, Cu-Kα (λ = 1.5405 Å). Morphology structures were obtained by transmission electron microscope (TEM, JEOL, JEM-2100) and scanning electron microscopy (SEM, Carl Zeiss AG, Supra 55). Atomic force microscope (AFM) was analyzed using a Multimode Nanoscope Ⅲa. The elemental composition and chemical valence states of materials were characterized by X-ray photoelectron spectroscopy (XPS, THERMO VG, Escalab 250). The inductively coupled plasma-atomic emission spectrometer (ICP-AES, SHIMAZU, 2030) was used to determine the content of each element in A-NiCoMn-TAC/LDH. The  X-ray absorption near-edge spectroscopy (XANES) (Ni K-edge, Co K-edge and Mn K-edge) were measured at beamline 4B7B of BSRF. The detail for the ex-situ XAFS tests method was displayed in Supplementary Note 1. The XAFS data processing was displayed in Supplementary Note 2. The computer-controlled electrochemical analyzer was used to conduct electrochemical measurements79,80,81,82. The detail for the in-situ XAFS measurements is described in Supplementary Note 3. The in-situ ATR-SEIRAS spectra were recorded by a Nicolet iS50 FT-IR spectrometer, equipped with a MCT detector cooled with liquid nitrogen, and a PIKE VeeMAX III variable angle ATR sampling accessory. The detail of in-situ ATR-SEIRAS measurements was exhibited in Supplementary Note 4.

Electrochemical measurement in single cell

All the electrochemical data in this experiment were measured at room temperature by CHI 760e multifunctional electrochemical workstation (Shanghai Chenhua, China). The single electrochemical cell is made in Gaossunion (Tianjing, China). The test adopts a three-electrode system, including working electrode (preparation), reference electrode (Hg/HgO electrode), and counter electrode (Pt foil). The working electrode was prepared by the following method: 4 mg catalyst, 15 μL Nafion (5 wt%) and 1.5 mg acetylene black were dispersed uniformly in 1 mL of ethanol by sonication. Then, the obtained suspension of 250 μL was drop-casted to carbon paper (1 × 1 cm2). The reference electrode (Hg/HgO) was changed 1 M KOH solution every week. Before the test, Ag/AgCl was used as the counter electrode and reference electrode, and Hg/HgO was used as the working electrode. The standard electrode potential of the Hg/HgO electrode was corrected by open-circuit potential test in 1 M KOH electrolyte. After corrected, the standard potential of the Hg/HgO electrode is 0.098 V and the standard.

The electrolyte is 1 M KOH or 1 M KOH with 0.33 M urea. Generally, the electrochemical test electrolyte should be prepared as needed. Prepare 1 M KOH or 1 M KOH with 0.33 M urea solution in a 100 mL volumetric flask. After cooling, store it in a cool and dark place and maintain the usage period should not exceed 3 days. The pH values of the electrolyte were measured by a pH meter to be 13.89, 13.91, and 13.86 respectively, and the error bar was calculated to be 0.025 based on the standard deviation.

The applied potentials were converted with respect to reversible hydrogen electrode (ERHE). E(Hg/HgO) is the potential tested on electrochemical station with a Hg/HgO electrode as reference electrode. The iR is the compensation of potential due to the internal resistance between electrodes in an electrolyte solution. All the potentials were calibrated with 0.9 iR compensation unless noted otherwise.

$${E}_{{RHE}}={E}_{\left({Hg}/{HgO}\right)}+0.098+0.0592\times {pH}-0.9\times {iR}$$
(1)

LSV curves were tested at scan rate of 10 mV s−1, the potential (E) window 0–1 V vs. Hg/HgO. The EIS was measured from 100,000 Hz to 0.01 Hz in 1 M KOH with 0.33 M urea at a potential of 1.34 V vs. RHE. The Rs obtained by EIS are based on the results of three independent tests. Tafel slopes were obtained by plotting potential against log (J) from LSV curves.

Electrochemical specific surface areas (ECSAs) were measured by cyclic voltammetry (CV) at the potential window 0.18 V–0.23 V vs. Hg/HgO, with different scan rates of 50, 60, 70, 80 and 100 mV s−1. By plotting the current density difference (ΔJ) at 0.205 V vs. Hg/HgO against the scan rate, the linear slope that was twice of the double layer capacitance (Cdl) was used to represent ECSA. ECSA can be determined by dividing Cdl by the specific capacitance:

$${{{\rm{ECSA}}}}={C}_{{dl}}/{C}_{s}$$
(2)

Where, the Cs is 0.04 mF cm−2 based on values reported in alkaline electrolyte.

To test the stability of catalysis, the multi-step chrono-potential tests were carried out at current densities of 10, 200 and 500 mA cm−2. The TOF was calculated using the following equation35:

$${{{\rm{TOF}}}}=\frac{J*A}{{nzF}}\,$$
(3)
$$n=\frac{{m}_{{cat}}*{\omega }_{{Ni}}}{{M}_{{Ni}}}$$
(4)

Here, J represents the current density measured at 1.35 V (vs. RHE) and normalized by the geometric area; A is the geometric surface area; z is the number of electrons involved in the urea electro-oxidation to nitrogen; F is the Faraday constant. n (in this context) denotes the mole number of Ni atoms present on the electrode, mcat is the mass of catalyst on the electrode, ωNi is Ni loading in the catalyst based on ICP-MS results and MNi is the atomic mass of Ni.

For electrochemical test results (half-cell and AEM), all of data are based on the three independent tests, the error bars are calculated from the standard deviation (SD). Such as LSV, EIS, ΔE, Tafel, overpotential, energy efficiency and various potentials under different current densities. For the products analysis, FEs are calculated from the standard deviation during two independent tests based on GC and IC tests.

Pretreatment of anion exchange membrane (AEM)

AEM that has not been pre-treated is usually in a dry state or in the form of counterions (such as Cl⁻), with very low ionic conductivity and unstable size. The AEM (AMI-7001S, made in America) is ~0.45 mm thick and has an area of 1 × 1 cm2. Firstly, immerse AEM in a 5 wt% NaCl solution for 24 h (change the solution several times) to allow its interior to fully hydrate and expand. Then immerse the membrane in 0.5 M KOH solution for 24 h (change the solution several times). Finally, rinse it several times with deionized water.

Electrochemical measurement in the urea-assisted anion exchange membrane (AEM) electrolyzer

To demonstrate the potential of the catalyst for practical industrial applications, an AEM flow electrolyzer (MTQN-1 cm-60) integrating the UOR with the HER was assembled. The A-NiCoMn-TAC/LDH catalyst was utilized as the anode catalyst (working electrode), and the A-NiCoMn-TAC/LDA catalyst was used as the cathode catalyst (counter electrode and reference electrode). The flow chamber with two compartments separated was circulated with a solution consisting of 0.33 M urea in 1 M KOH at a flow rate of 5 mL min−1 at 60 °C. The catalyst ink was prepared by mixing 10 mg catalyst, 5 mg carbon black, 1 mL ethanol and 30 μL Nafion (5 wt%) solution. The catalysts were then air-brushed onto Ni foam (1 × 1 cm2) at a mass loading of 1.0 mg cm−2. The duration test was conducted under a constant industrial current of 1 A. All potentials without compensation.

The specific electricity consumption Q for producing 1 Nm−3 H2 was calculated as83:

$$Q=\frac{W}{V}\,\left({kWh}*{{Nm}}^{-3}\right)$$
(5)

where W refers to the total electrical energy input, and V is the corresponding volume of generated H2. The total energy input W was obtained from:

$$W=I*\int U \, {dt}$$
(6)

with I and U denoting the applied current and potential, respectively. The hydrogen production volume V was determined as:

$$V=\frac{22.4*I*t}{e*F}$$
(7)

where e the number of electrons required to produce one molecule of H2 (e = 2), and F is the Faradaic constant.

Computational method

To investigate the urea oxidation performances of A-NiCoMn-TAC/LDH, the DFT calculations were performed by the CASTEP package embedded in the Materials Studio84. We selected the generalized gradient approximation and Perdew–Burke–Ernzerhof functionals to properly describe the exchange-correlation interactions in all the catalyst systems85,86,87. To realize the geometry optimizations, we applied the ultrasoft pseudopotentials with 380 eV cutoff energy based on the ultrafine quality. The spin polarizations had been included in all the calculations. To balance the calculation efficiency and accuracy, the k-point was set to coarse quality with the application of the Broyden–Fletcher–Goldfarb–Shannon algorithm for all the energy minimizations88. For PDOS calculations, a denser k-point of fine quality was applied, which adopted the 4 × 4 × 1 Monkhorst–Pack and a separation of 0.04 Å−1. For NiCo-LDH, we constructed a 6 × 6 × 1 supercell with 180 atoms. For A-NiCoMn-TAC/LDH, Mn doping was introduced for both Ni and Co sites. For all the geometry optimizations, the following convergence criteria were applied: (1) Hellmann-Feynman forces should not exceed 0.001 eV/Å, and (2) the total energy difference should be converged to smaller than 5×10−5 eV/atom.

The formation energy (\(\Delta {E}_{{form}}\)) represented the energy costs of vacancy formation in both NiCo-LDH and A-NiCoMn-TAC/LDH, which were calculated as follows:

$$\Delta {E}_{{form}}={E}_{{perf}}-{E}_{{def}}-{\sum}_{i}{n}_{i}{\mu }_{i}$$
(8)

In the above equation, \({E}_{{perf}}\) and \({E}_{{def}}\) represented the total energy of the perfect structure and defective structure, respectively. The coefficient \({n}_{i}\) represented the number of atoms of element i forming the defect. \({n}_{i}\) should be negative for vacancy and positive for interstitial defects. The chemical potential \({\mu }_{i}\) was the reference energy for the added/removed atoms. The more negative \(\Delta {E}_{{form}}\) indicated that the defect formation was easier in the structures.

The calculations of adsorption energies were based on the following equation.

$$\Delta {E}_{{ads}}={E}_{{total}}-{E}_{{LDH}}-{E}_{{adsorbate}}$$
(9)

In the above equation, \({E}_{{total}}\) represented the total energy of adsorbates (e.g., urea or H2O), \({E}_{{LDH}}\) was the energy of the defective NiCo-LDH or defective A-NiCoMn-TAC/LDH, and \({E}_{{adsorbate}}\) was the energy of adsorbate molecules in the gas phase. For the calculations of the adsorption energies, the solvation effect had been considered through the embedded implicit solvation model in CASTEP, which treated the solvent as a continuous dielectric medium rather than explicit solvent molecules. The dielectric constant of water had been applied to describe the solvent effect.

For the energy change of each reaction step, the zero-point energy difference (∆ZPE) had been considered as follows.

$$\Delta G=\Delta E-\Delta {ZPE}-T\Delta S$$
(10)

From the above equation, ∆E was the total energy change calculated by DFT, ∆ZPE was the change in zero-point energies of the system, and T∆S was the total entropy change.