Characterization of the single-atom Ni catalystAn atomic Ni-modified carbon nanosheet array (Ni-SAC) electrode was synthesized via a confinement strategy30. As shown in Fig. 1a and Supplementary Fig. 1, a NiAl-LDH array with metanilic acid intercalation (denoted as NiAl-LDH(MA)) was vertically grown on hydrophilic carbon cloth via a hydrothermal method. Then, a pyrolysis process was performed to obtain the Ni nanoparticles and single-atom embedded carbon nanosheets (denoted as Ni-CNS), with the intercalated MA converted to carbon (Supplementary Fig. 2). This process was followed by further acid etching to obtain Ni-SAC. Carbon nanosheets (CNS) electrode with almost no Ni was also obtained by acid etching the Ni-CNS in 1 M HCl under heating conditions (60 °C) for 12 h (Supplementary Fig. 3 and Supplementary Table 1). CNS and Ni-CNS were used as control samples of Ni-SAC to verify the critical role of the single-atom Ni in the 2e− ORR. Figure 1b shows the X-ray diffraction (XRD) patterns of the as-prepared samples. The XRD pattern of NiAl-LDH(MA) shows a typical diffraction (003) peak at 5.82° (yellow curve in Fig. 1b), indicating an interlayer spacing of 1.52 nm, which correlates well with the MA-intercalated LDH31,32. For the Ni-CNS sample, the original characteristic peaks of LDH disappear after the pyrolysis process (blue curve in Fig. 1b) and are replaced by the peaks of metallic Ni (PDF#04-0850) at approximately 44.5° (corresponding to the (111) crystal planes) and 51.9° (corresponding to the (200) crystal planes), as well as the characteristic peak of graphitic carbon (002) at approximately 25°33. After acid etching treatment, the characteristic peaks of the metallic Ni disappear for the Ni-SAC sample, leaving the (002) and (100) peaks of graphitic carbon at approximately 25° and 44°, respectively (red curve in Fig. 1b)19,34,35. The scanning electron microscopy (SEM) image of the Ni-SAC sample shows a typical porous nanosheet array structure vertically grown on carbon cloth. The nanopores can be clearly observed on the carbon nanosheets and are formed by acid etching of the Ni nanoparticles (Fig. 1c and Supplementary Fig. 4). High-resolution transmission electron microscopy (HRTEM) further verifies the porous nanosheet structure of Ni-SAC (Fig. 1d and Supplementary Fig. 5). Moreover, no visible Ni nanoparticles are observed in the carbon nanosheets. The above results indicate that the acid etching process can effectively remove the aggregated and unstable Ni nanoparticles. The HRTEM images of the Ni-CNS and CNS are shown in Supplementary Figs. 6 and 7. The atomic-resolution high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images show that scattered bright spots are uniformly dispersed on the carbon nanosheets of the Ni-SAC sample (Fig. 1e and Supplementary Fig. 8). The energy-dispersive X-ray (EDX) mapping results also show the even distribution of Ni in the Ni-SAC catalyst (Supplementary Fig. 9). Notably, this strategy enables the easy synthesis of large-area electrocatalysts (Supplementary Fig. 10), demonstrating potential for further scale-up preparation of Ni-SAC.Fig. 1: Morphology and structural characterization of single-atom Ni.a Schematic diagram of the synthesis of Ni-SAC by the confinement synthesis strategy. b XRD patterns of NiAl-LDH(MA), Ni-CNS, Ni-SAC. c SEM, d HRTEM, and e HAADF-STEM images of Ni-SAC sample. f High-resolution XPS spectra of Ni 2p in Ni-SAC. g Ni K-edge XANES and the corresponding h Fourier transform EXAFS spectra of Ni-CNS and Ni-SAC.We then performed X-ray photoelectron spectroscopy (XPS) to elucidate the chemical composition and structure of Ni-SAC. As shown in Supplementary Fig. 11, Ni, Al, C, N, O and S were detected in the full XPS spectrum. Note that the proportion of Ni in the Ni-SAC sample is markedly lower than that in the Ni-CNS sample due to the acid etching of the Ni nanoparticles. This finding is consistent with the inductively coupled plasma optical emission spectroscopy (ICP-OES) results (Supplementary Table 2). The characteristic peak at 853.1 eV in the Ni 2p3/2 spectrum of the Ni-CNS sample is assigned to zero-valent nickel (Supplementary Fig. 12)36. While the binding energy of Ni-SAC shows a positive shift of 0.8 eV (to 853.9 eV) relative to Ni-CNS, indicating that Ni in Ni-SAC is in an oxidation state (Fig. 1f). The deconvolution of the N 1s spectra for Ni-SAC and Ni-CNS reveal pyridinic‒N, Ni‒N, pyrrolic‒N, and graphitic‒N (Supplementary Fig. 13)35,37. The peaks located at 161.7 eV, 163.0 eV, 164.8 eV and 168.6 eV in the S 2p spectra of Ni-SAC are attributed to Ni‒S, C‒S‒C 2p3/2, C‒S‒C 2p1/2 and C‒SOx‒C, respectively (Supplementary Fig. 14)38. The above XPS results indicate the successful incorporation of N and S into the coordination environment, where single-atom Ni serves as the metal sites. Comprehensive information regarding the parameters and outcomes of XPS fitting can be found in Supplementary Table 3. Subsequently, we employed X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) analyses to investigate the coordination environment of the nickel atoms in Ni-SAC. The Ni K-edge reveals that the position of the white line peak for Ni-SAC is situated in a higher energy region than both the Ni-CNS and Ni foil, which is in accordance with the XPS results demonstrating that the Ni atoms in Ni-SAC are in the oxidation state (Fig. 1g). Fourier transform EXAFS (FT-EXAFS) shows the presence of a scattering path, which is attributed to the Ni-Ni bond (2.20 Å) in the Ni-CNS. However, only a strong peak at 1.75 Å for Ni-SAC is observed (Fig. 1h). Notably, unlike the widely reported Ni‒N4 (1.49 Å)39, the main peak of Ni-SAC is located in a higher R-space. We speculate that Ni-SAC is a structure in which Ni is co-coordinated with light heteroatoms (N/S) based on the XPS analysis. The wavelet-transform EXAFS (WT-EXAFS) results show that the maximum intensity of the WT contour plot for Ni-SAC occurs around 4.35 Å−1. This value is distinctly different from that of Ni foil (7.5 Å−1) and NiO (6.8 Å−1), confirming the atomic dispersion of Ni in Ni-SAC (Supplementary Fig. 15). Moreover, the WT contour maximum intensity of the Ni-CNS is located at 7.3 Å−1, akin to that of the Ni foil, echoing the presence of zero-valent Ni in the XPS analysis. The optimal fitting results of quantitative least-squares EXAFS curve fitting indicate that the Ni atom in Ni-SAC is coordinated with approximately 3.9 heteroatoms (N/S) in the first shell (Supplementary Fig. 16 and Supplementary Table 4).Electrochemical ORR to produce H2O2 over the Ni-SAC electrodeWe then evaluated the 2e− ORR performance of the Ni-SAC electrode to produce H2O2 using the Ni-CNS as a reference sample. The electrochemical tests were performed in a three-electrode configuration within a divided H-type cell using O2- or N2-saturated 0.1 M KOH solution as the electrolyte (Supplementary Fig. 17). Linear sweep voltammetry (LSV) curves show that the Ni-SAC and Ni-CNS samples exhibit negligible current responses in N2-saturated electrolyte (dash line in Fig. 2a). However, under O2-saturated conditions, the current densities of both the Ni-SAC and Ni-CNS electrodes are clearly enhanced, demonstrating the occurrence of the ORR (solid line in Fig. 2a). Specifically, the Ni-SAC electrode exhibits an onset potential (Eonset) of 0.857 V vs. RHE (defined as the potential at a current density of 0.1 mA cm−2) and the maximum current density of 37.40 mA cm−2, both superior to those of the Ni-CNS sample (0.828 V and 16.66 mA cm−2). The above results indicate that the Ni-SAC sample has better ORR performance. Note that the Eonset of Ni-SAC and Ni-CNS are slightly greater than the thermodynamic theoretical value required for the occurrence of the 2e− ORR under alkaline conditions (O2 + H2O + 2e− → HO2− + OH−, 0.75 V vs. RHE). The reason for this is potentially the shift in the Nernst potential caused by the low concentration of H2O2 in the electrolyte, which also may be related to its pH40. Moreover, Ni-SAC has a lower Tafel slope than Ni-CNS (79.5 mV dec−1 vs. 146.8 mV dec−1), revealing that Ni-SAC has better reaction kinetics (Supplementary Fig. 18). Compared to Ni-CNS, Ni-SAC has a larger electrochemical active surface area (ECSA), as evidenced by its higher double-layer capacitance (Cdl) (33.5 mF cm−2 vs. 21.8 mF cm−2). After normalizing the ECSA to the LSV curves, Ni-SAC still exhibits a significant current density, indicating high intrinsic activity (Supplementary Fig. 19).Fig. 2: O2 electroreduction performance of Ni-SAC for the 2e− ORR to produce H2O2.a LSV curves of Ni-SAC and Ni-CNS in 0.1 M O2/N2-saturated KOH. b Faradaic efficiency (FE) and c H2O2 yield rate of each sample in the voltage range of 0.2 V to 0.7 V. d Comparison of H2O2 yield rate (mmol gcat−1 h−1) with other reported literature under alkaline condition (0.1 M KOH) in H-cell. e Stability test of Ni-SAC at 0.3 V for 20 h. f Electron transfer number (n) and HO2− % of Ni-SAC, Ni-CNS, CNS. g LSV curves corresponding to H2O2RR in N2-saturated 0.1 M KOH with 10 mM H2O2 solution. h Corresponding ORR polarization curves before and after 1 mM KSCN poisoning in 0.1 M O2-saturated KOH for Ni-SAC. i Electron transfer number (n) and HO2− % after poisoning for Ni-SAC. The three-electrode setup performed in the H-cell, RRDE, and RDE have no iR-compensation. The error bars are defined as standard deviation, and the centre of each error bar represents the mean value of the corresponding three independent experiments.We then quantified the accumulated H2O2 in the cathode chamber during the ORR process using an ultraviolet-visible (UV-Vis) spectrophotometer to investigate the Faradaic efficiency (FE) and yield rate of H2O2. The standard curve for calculating the H2O2 concentration is displayed in Supplementary Fig. 20. Figure 2b demonstrates that the FE of the 2e− ORR over the Ni-SAC electrode achieves a maximum of 89.02 ± 0.84% and consistently remains an approximate value of 85% in a wide potential window from 0.3 V to 0.7 V. In contrast, the FE of the Ni-CNS electrode is substantially lower, with a maximum value not exceeding 45%. The yield rate of H2O2 over Ni-SAC progressively increases with increasing overpotential, reaching a maximum value of 0.29 ± 0.02 mmol h−1 cm−2 at 0.2 V vs. RHE, approximately 3.4-fold greater than that of Ni-CNS (Fig. 2c). The developed Ni-SAC electrode also exhibits an evident advantage in the yield rate of H2O2 (up to 0.73 mol gcat−1 h−1) compared to the currently reported work under alkaline conditions, as shown in Fig. 2d, Supplementary Fig. 21 and Supplementary Table 5. We further tested the long-term stability of Ni-SAC. As shown in Fig. 2e, the Ni-SAC electrode can operate stably for 20 h at 0.3 V vs. RHE. Results from SEM, HAADF-STEM, and XRD results after the stability test show that the structural integrity of the nanoarrays and the dispersion of the single atoms are maintained (Supplementary Figs. 22–24).To rationalize the role of single-atom Ni in enhancing the selectivity of the 2e− ORR, we employed rotating ring-disk electrode (RRDE) tests in O2-saturated 0.1 M KOH to access the 2e− ORR selectivity over different catalysts during the ORR. Owing to the unsuitability of integrated electrodes for RRDE studies, we scraped powder samples from these electrodes to gather evidence regarding the intrinsic activity of the catalyst. During the RRDE test, the H2O2 generated at the disk electrode diffuses to the ring electrode at a rotation speed of 1600 rpm and is oxidized at a ring voltage of 1.2 V vs. RHE. As shown in Fig. 2f and Supplementary Fig. 25, the Ni-SAC sample exhibits a low electron transfer number (n) value close to 2.5 in the potential window from 0.2 to 0.7 V vs. RHE with HO2− selectivity (%) up to 73%. Note that the HO2− % is lower than that in H-cell tests due to the limited mass transfer of the catalyst in the powder state relative to the integrated electrode. The RDE measurements align well with those of the RRDE, indicating high selectivity for the 2e− ORR over Ni-SAC (Supplementary Fig. 26). The n value of Ni-CNS is evidently greater than that of Ni-SAC (e.g., 3 vs. 2.5 at 0.8 V vs. RHE), emphasizing the important role of single Ni atoms in enhancing the 2e− ORR selectivity. RRDE measurement for the CNS shows an n value close to 4, indicating that the CNS substrate is not the reactive centre for the 2e− ORR to produce H2O2. The calculated turnover frequency (TOF) values of Ni-SAC are 0.65 s−1, 0.72 s−1, and 0.78 s−1, corresponding to 0.7 V, 0.65 V, and 0.6 V, respectively. Additionally, the mass activity of Ni-SAC at 0.65 V is 49.06 A g−1. These indices representing intrinsic activity are superior to most reported values. (Supplementary Fig. 27, Supplementary Tables 6 and 7, specific formulae are provided in Supplementary Note 1).We also tested the reaction activity of the H2O2RR in N2-saturated 0.1 M KOH with 10 mM H2O2 solution to compare the performance of the 2e− ORR over different catalysts. Figure 2g shows that the H2O2RR reactivity of Ni-SAC is significantly weaker than its 2e− ORR activity, indicating that H2O2 is relatively stable and can accumulate over Ni-SAC under reaction conditions. In contrast, the H2O2RR activity of the Ni-CNS and CNS catalysts is only slightly weaker than their 2e− ORR activity, leading to more pronounced 4e− ORR competition. To further demonstrate that the single Ni atom is the primary active site for the 2e− ORR to H2O2, we used 0.1 M KOH with the addition of 1 mM thiocyanate ion (SCN−) as the toxicant to block single Ni atoms. As shown in Fig. 2h and Supplementary Fig. 28, a significant current decay (e.g., 18.03 mA cm−2 at 0.4 V vs. RHE) is observed over Ni-SAC after adding SCN− to the electrolyte, and the current decay of the Ni-CNS decreases by only 4.51 mA cm−2 under the same conditions. Moreover, the n of the poisoned Ni-SAC was determined by the RRDE to be approximately 4, echoing the n of the CNS (Fig. 2i and Supplementary Fig. 29). The above results strongly indicate that the single-atom Ni is the principal active site for enhancing the 2e− ORR performance.Electrochemical ORR to produce H2O2 in a flow cellMotivated by the excellent 2e− ORR performance of the three-electrode system, we further evaluated the catalytic performance of the Ni-SAC electrosynthesis of H2O2 in a more practical scenario. Specifically, as displayed in Fig. 3a, we carried out two-electrode tests in a custom two-electrode flow cell using Ni-SAC as the cathode and NiFe-LDH grown on carbon cloth as the anode. NiFe-LDH has been shown to be one of the most efficient catalysts for the oxygen evolution reaction (OER)41. The flow cell was equipped with a Nafion 117 membrane, and the reaction was performed in 1 M O2-saturated KOH at room temperature. As shown in Fig. 3b, LSV tests were performed over a potential interval from −0.8 V to −2 V. The constructed system delivers a high current density of −261.73 ± 22.07 mA cm−2 at a cell voltage of −2 V. We then carried out potentiostatic measurements to evaluate the FE and productivity of H2O2 in the cathodic chamber at different cell voltages. As shown in Fig. 3c, the H2O2 yield rate gradually increases as the reaction potential becomes negative, rising from 0.28 ± 0.01 mmol h−1 cm−2 to 2.19 ± 0.02 mmol h−1 cm−2 when the cell voltage increases from −0.8 V to −1.7 V. The FE of H2O2 reaches up to 91.36 ± 1.61% and maintains approximately 80% in a wide potential window. A stability test was conducted for 35 h at a cell voltage of −1.7 V, during which the concentration of generated H2O2 remained essentially stable (Fig. 3d). We then performed a comprehensive performance comparison of reported electrocatalysts for the 2e− ORR considering factors such as FE, yield rate, stability, and jmax. As shown in Fig. 3e and Supplementary Table 8, the developed Ni-SAC has superior H2O2 productivity (up to 5.48 mol gcat−1 h−1) and can steadily operate even at high potentials. Furthermore, we calculated the electron consumption rate (Re) of the constructed systems42, and the maximum Re value achievable in the flow cell is approximately 7.56 times greater than that in the H-cell (28.52 e−s−1 vs. 3.76 e−s−1) (Supplementary Fig. 30).Fig. 3: Properties of the Ni-SAC catalyst in a two-electrode flow cell.a Schematic diagram of the flow cell. b ORR polarization curves in the flow cell. c FE and yield rate of H2O2 in the voltage range of −0.8 V to −1.7 V. d Stability test of Ni-SAC at −1.7 V for 35 h. e Comparison of 2e− ORR performances with reported electrocatalysts in the flow cell. The two-electrode system constructed in the flow cell has no iR-compensation. The error bars are defined as standard deviation, and the centre of each error bar represents the mean value of the corresponding three independent experiments.Mechanistic studiesThe 2e− ORR consists of two proton-coupled electron transfer processes in which *OOH serves as the key reactive intermediate. In situ Fourier transform infrared spectrometry (FTIR) was carried out to monitor the adsorbed *OOH intermediate over different catalysts (including Ni-SAS, Ni-CNS, and CNS) during the ORR process to understand the reaction mechanism. As shown in Fig. 4a and Supplementary Fig. 31, the FTIR spectra display a characteristic peak at 1100 cm−1, representing the stretching of O−O bonds in the adsorbed *OOH on the electrode surface43,44. The Ni-SAC exhibits stronger *OOH adsorption than the Ni-CNS and CNS, indicating a greater 2e− ORR preference. The peak intensity of *OOH over Ni-SAC gradually increases from 0.8 to 0.2 V vs. RHE due to the accumulation of the *OOH intermediate, thus attaining higher H2O2 productivity (consistency with the results in Fig. 3c). The intensity of *OOH adsorption in the Ni-CNS is weak, with evident peak vibrations occurring only at higher overpotentials. The CNS shows almost no *OOH vibration at the test potentials, demonstrating poor H2O2 production ability. Focusing on the steps of *OOH generation (i: O2 + * → *O2; ii: *O2 + H2O + e−→ *OOH + OH−), *OOH formation is related not only to suitable adsorption between the electrocatalyst and O2 but also to the continuous and efficient supply of active hydrogen (H*) from water splitting.Fig. 4: Active hydrogen and oxygen adsorption jointly promote the 2e− ORR.a In situ FTIR spectra of Ni-SAC, Ni-CNS, CNS at potential of 0.2 V, 0.5 V, 0.8 V in 0.1 M O2-saturated KOH. b In situ FTIR spectra of Ni-SAC, Ni-CNS, CNS at potential of 0 V in 0.1 M O2-saturated KOH for 10 min. c ESR spectra of Ni-SAC, Ni-CNS, CNS. d KIE of H/D and current densities over Ni-SAC, Ni-CNS and CNS at 0 V. e O2-TPD results for the Ni-SAC, Ni-CNS, CNS. f Mechanism diagram. The error bars are defined as standard deviation, and the centre of each error bar represents the mean value of the corresponding three independent experiments.We employed in situ FTIR as well as electron spin resonance (ESR) to determine the capacity for H* production, followed by the kinetic isotope effect (KIE) to assess the proton transfer capacity. According to previous studies, the vibrations appearing in the 3000−3700 cm−1 band can be attributed to the stretching of O−H in H2O45. As shown in Fig. 4b, the peak intensity of the O−H vibration over the CNS is stronger than that over the Ni-SAC and Ni-CNS under the same electrolytic conditions, indicating a greater water splitting ability for the CNS sample. We then conducted ESR to monitor the amount of H* during the ORR process using 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) as an H* capturing agent. The classical nonuple peak intensity ratio (1:1:2:1:2:1:2:1:1) and the hyperfine coupling constants values (AN = 16.5 G and AHβ = 22.5 G) both indicate the presence of DMPO-H, confirming H* aggregation in the three electrocatalysts(Fig. 4c and Supplementary Fig. 32; for detailed discussion, see Supplementary Note 2)46,47. The highest peak intensity observed in CNS can be attributed to its strong water splitting capacity, thereby maximizing the accumulation of H*. Based on this phenomenon, we determined the KIE of H/D (H2O/D2O) to rationalize the kinetic significance of reactive hydrogen (H*) production from the water splitting. The KIE value, representing the electron transfer rate, is determined by comparing the current densities obtained from LSV curves scanned in 0.1 M O2-saturated KOH/KOD solutions. A KIE value close to 1 indicates an acceleration of the water splitting process during hydrogenation48,49. Figure 4d and Supplementary Fig. 33 display the varying degrees of attenuation in catalyst reactivity following the replacement of H2O and KOH with D2O and KOD. The KIE values of Ni-SAC, Ni-CNS, and CNS are 1.98, 2.97, and 1.28, respectively. Among them, CNS has the highest proton transfer rate, followed by Ni-SAC and Ni-CNS. On the basis of the above experimental results, CNS is found to be the main site of water splitting for H* production. However, the 2e− ORR performance does not positively correlate with the H* production capacity, indicating that it is not the only factor contributing to the performance improvement.In addition to H* generation, the adsorption of O2 on the catalytic surface (configuration, capacity) is another issue that needs to be considered. The geometrical configuration of the catalyst affects its adsorption configuration with respect to O2. Atomically dispersed metal sites tend to absorb O2 via an end-on configuration, which is not conducive to O−O breakage in *OOH; however, continuous metal nanoparticles prefer to absorb O2 via a side-on configuration, thus favouring O−O breakage23,50,51. Therefore, ensuring the efficient conversion of *OOH through the construction of dispersed sites is crucial for enhancing the performance of the 2e− ORR relative to that at continuous sites. We then conducted O2 temperature-programmed desorption (O2-TPD) to further investigate the difference in adsorption capacity between the three electrocatalysts and O2. The peaks shown in Fig. 4e indicate the release of chemisorbed O2 from the different catalysts. The desorption peaks for Ni-CNS, Ni-SAC, and CNS are located at 146.54 ± 3.89 °C, 139.41 ± 2.70 °C, and 136.52 ± 2.79 °C, respectively. A higher desorption temperature corresponds to an increase in the adsorption strength. Notably, Ni-CNS exhibits the highest adsorption strength for O2, while Ni-SAC closely resembles CNS. Similar to previous studies indicating that Ni metal possesses excellent adsorption capacity for O2, thereby promoting the oxygen dissociation processes (*OOH dissociation to *OH and *O)33,52,53. Although the desorption temperatures of Ni-SAC and CNS are similar, the desorption peak intensity of CNS is significantly lower compared to that of Ni-SAC (normalized by Brunauer-Emmett-Teller (BET) results to exclude the effect of pore size; Supplementary Fig. 34). The results reflect the variations in the quantity of O2 desorbed over different electrocatalysts. Consequently, CNS is more difficult to adsorb O2 compared to Ni-SAC.In situ FTIR, ESR, and KIE experiments confirmed that the CNS is the main site at which water splitting occurs to produce H*. Moreover, previous studies have confirmed that dispersed atomic sites are able to adsorb O2 via an end-on configuration, thus converging to the 2-electron path. The O2-TPD results also show that Ni-SAC has a suitable adsorption capacity for O2. Based on the above findings, the successful integration of the excellent H* generation ability for the CNS and the appropriate adsorption of O2 on a single Ni atom can effectively enhance the activity of the 2e− ORR for Ni-SAC (Fig. 4f).2e− ORR coupled with ethylene glycol oxidationThe cathodic ORR is typically coupled with the anodic OER, which is kinetically sluggish and generates low-market-value O254. To further reduce the overall energy consumption of the system, we employed a thermodynamically favourable small organic molecule oxidation reaction to replace the original OER at the anode (Fig. 5a). Note that the onset potential of ethylene glycol oxidation (EOR) is < 1 V, which is lower than that of the OER (1.23 V)29. Thus, coupling the ORR with the EOR (denoted as ORR | | EOR) is more easily driven relative to the ORR | | OER system, which contributes to lowering the cell voltage (Fig. 5b). Moreover, given the high annual consumption of poly(ethylene terephthalate) (PET) plastics and their negligible recycling, ethylene glycol (EG) derived from PET depolymerization is an excellent candidate as a key monomer for small organic molecules55,56. Glycolic acid (GA) prepared by the selective oxidation of ethylene glycol is also notable for its high value-added properties. We adopted Au/Ni(OH)2, which has been reported by our group, as the electrocatalyst to evaluate the EOR performance in an electrolyte configuration of 1 M KOH + 0.3 M EG57. The morphology and structure of Au/Ni(OH)2 are described in detail in Supplementary Fig. 35.Fig. 5: ORR | | EOR coupling system performance and economic feasibility analysis.a Coupling system construction diagram. b Comparison diagram of cell voltage for different coupling system. c Polarization curves in ORR | | EOR, ORR | | OER and HER | | EOR system in the flow cell. d FE and yield rate of H2O2 and GA in ORR | | EOR system. e Mass of H2O2 produced per kWh of electricity at −1.7 V in ORR | | EOR and ORR | | OER systems. f XRD pattern of SPB product. g Techno-economic evaluation of ORR | | EOR and ORR | | OER systems (left column is ORR | | OER, right column is ORR | | EOR).We first investigated the EOR performance of Au/Ni(OH)2 in a three-electrode system, achieving a FE and yield rate of GA up to 92% and 1.85 mmol h−1 cm−2, respectively (Supplementary Figs. 36 and 37). We then evaluated the coupled system performance in a custom two-electrode flow cell with Ni-SAC as the cathode and Au/Ni(OH)2 as the anode. As shown in Fig. 5c, the ORR | | EOR shows the lowest cell voltage and the highest current density compared to the ORR | | OER and HER | | EOR systems. Quantitative analysis of the cathode and anode products over a wide potential range of −0.3 V to −1.7 V shows that the FEs of both products primarily maintain above 90%. The FE and yield rate for H2O2 reach up to 99.83% and 2.92 mmol h−1 cm−2 (7.3 mol gcat−1 h−1), respectively, while the FE and yield rate for GA reach up to 97.80% and 2.09 mmol h−1 cm−2, respectively (Fig. 5d). The stability tests of the coupling system were conducted for 10 h at a cell voltage of −1.7 V (Supplementary Fig. 38). Moreover, compared with that of the ORR | | OER (NiFe-LDH as the anode), the H2O2 productivity of the ORR | | EOR is also improved (Supplementary Fig. 39). The ORR | | EOR system can produce more H2O2 per 1 kWh of electricity at −1.7 V relative to ORR | | OER systems (0.240 g vs. 0.226 g; Fig. 5e). All above results show that coupling enhance the performance of the system.Then, we converted the alkaline electrolyte containing H2O2 directly into the downstream product, sodium perborate (SPB), to reduce the separation cost. SPB is also an oxidant that can be used for water treatment. The specific process is shown in Supplementary Fig. 40. Sodium metaborate was reacted with the electrolyte containing H2O2 in an ice-water bath for 1 h, followed by filtration and drying to obtain SPB. The XRD patterns and FTIR spectra confirm the successful preparation of the SPB products (Fig. 5f and Supplementary Fig. 41). In addition, techno-economic evaluations of the ORR | | EOR and ORR | | OER systems were carried out to further validate the potential of the constructed systems for industrial application as well as the superiority of the coupled systems. We evaluated the input costs in three broad categories: input chemicals, capital, and operations. The product revenues were then considered to determine the gross profits from the above two systems (the specific calculation process is detailed in Supplementary Note 3). Notably, the ORR | | EOR system generates higher revenue from H2O2 production relative to the ORR | | OER (7.06*106 $ vs. 5.6*106 $), consistent with the quantitative results. Additionally, the corresponding anode product, glycolic acid, has an extraordinarily higher value added relative to oxygen (2.83 $/kg vs. 0.09 $/kg), resulting in a substantial increase in the overall profit (15.65*106 $ vs. 2.7*106 $) (Fig. 5g and Supplementary Table 9).
