Introduction

Hydrogen peroxide (H2O2) is a versatile chemical with strong oxidizing properties, widely used in applications ranging from household cleaning to industrial processes, especially in chemical synthesis1,2,3,4. The traditional anthraquinone process, which relies on sequential hydrogenation and oxidation, demands substantial infrastructure, consumes considerable energy, and has notable environmental drawbacks, including a significant carbon footprint5,6. Electrochemical two-electron oxygen reduction (2e−-ORR) and water oxidation reaction (2e−-WOR) have emerged as promising alternatives for on-site and on-demand H2O2 production under mild conditions utilizing renewable electricity7,8. In the 2e−-ORR process, the low concentration of dissolved O2 in most solvents restricts the efficiency of H2O2 in-situ production. Consequently, achieving efficient H2O2 generation through aqueous 2e−-ORR is extremely challenging. In contrast, 2e−-WOR reaction without continuous O2 supply only takes water as a reactant for H2O2 production, neatly avoiding the low solubility of O2 in water as the condition in ORR9. Unfortunately, the 2e−-WOR process proceeds only at a high theoretical potential of 1.76 V and inevitably suffers from competition with 4e−-oxygen evolution reaction (OER) for O2 evolution (2H2O→O2 + 4H+ + 4e−, Eo = 1.23 V), thus resulting in low activity and selectivity for H2O2 electrosynthesis10,11.

Considering that different oxygen-containing intermediates (e.g., OH*, O*, and OOH*) are generated during relevant reaction types, the adsorption capacity between catalytically active sites and oxygen-containing intermediates is a prerequisite for the progression of either 2e−-WOR or 4e−-OER12,13,14. An optimal OH* adsorption capacity, along with a weak ability for the transition from OH* to O*, is crucial for suppressing O2 production and facilitating the direct generation of H2O215. Significant progress has been made in developing highly selective electrocatalysts for the ORR using O2 instead of water through various functionalization methods. However, catalysts capable of selectively and stably oxidizing water to generate H2O2 with stable performance remain scarce; besides, it is vital to elucidate the correlation between the intrinsic structure and selective 2e−-WOR transfer mechanisms in catalysts. To address this challenge, bismuth-based metal-organic frameworks (MOFs) emerge as promising candidates for both mechanism studies and performance optimization. Their unique electronic structure and high oxygen affinity make it become a promising catalyst, making them attractive for modulating adsorption energy for electrocatalytic reaction intermediates16,17,18,19,20. Nevertheless, the direct use of MOFs as efficient 2e−-WOR catalysts is in its infancy. Meanwhile, direct utilization of in-situ generated H2O2 in the sequential electrochemical synthesis remains.

Herein, we reported a bismuth benzene tricarboxylic acid MOF (Bi9(C9H3O6)9(H2O)9, denoted as CAU-17)21,22 as a platform, which features a complex structure composed of helical Bi-O rods cross-linked by benzene tricarboxylic acid (H3BTC) ligands23,24,25. By further partially doping hetero-linker isophthalic acid (H2IPA), which lacks carboxy groups compared to H3BTC, an unsaturated coordinative CAU-17, denoted as UC-CAU-17, was developed. The well-designed UC-CAU-17 exhibited an impressive alkaline H2O2 activity and selectivity via 2e−-WOR, achieving a current density of 10 mA cm−2 at a low overpotential of 1.04 V and a high Faradaic efficiency (FEH2O2) of 75.5%. Remarkably, when conductive fluorine-doped tin oxide (FTO) was used as a substrate, the overpotential further decreased to 0.98 V, as the FEH2O2 increased to 79.1%, which was the highest among all previously reported metal-based catalysts for H2O2 electrosynthesis (Supplementary Table 1). The accumulated ~6.17 wt.% H2O2 alkaline solution was directly applied to butanone ammoximation with ammonia and titanium silicalite-1, presenting a notable butanone conversion rate of 80.2%, with high selectivity of 81.1% towards butanone oxime formation. The experimental results confirmed that doping the H2IPA linker led to partial Bi atoms being uncoordinated with O atoms, contributing to the formation of unsaturated coordinative Bi sites. In-situ attenuated total reflection Fourier transform infrared (ATR-FTIR) analysis revealed that these unsaturated coordinative Bi sites optimized the adsorption energy of the OH* for H2O2 generation. Density functional theory (DFT) results demonstrated that such unsaturated coordinative Bi sites lowered the free energy of the OH* → H2O2 transition and inhibited the transformation of OH* to O* and OOH*, resulting in a significant enhancement in both H2O2 activity and selectivity.

Results

Theoretical predictions of unsaturated coordination

In the electrochemical WOR process (Fig. 1a), the adsorption of OH* intermediate is a critical factor in determining the reaction rate. Strong adsorption favors the 4e−-OER pathway, leading to the generation of O2, whereas weak adsorption promotes the 1e−-WOR pathway, forming hydroxyl radicals. Therefore, optimizing the adsorption of OH* is essential for steering the WOR pathway and enhancing the process efficiency. First, we conducted theoretical calculations to understand how varying the Bi-site coordination environment affects its intrinsic activity and selectivity towards H2O2 generation via the 2e−-WOR pathway. As displayed in Fig. 1b, Bi9(C9H3O6)9(H2O)9 is denoted as CAU-17 (CCDC No. 1426169), where Bi sites are directly linked to 1,3,5-benzenetricarboxylate (BTC3–) ligands. UC-CAU-17 is established by the partial doping of H2IPA with O atoms, contributing to unsaturated coordinative Bi sites (Fig. 1c). Clearly, for the 2e−-WOR (Fig. 1d), the Gibbs free energy change for OH* and HOOH* adsorption on UC-CAU-17 was determined to be ΔGOH* = 1.78 eV and ΔGHOOH* = 1.74 eV, respectively. This energetic profile identifies OH* adsorption as the rate-limiting step on UC-CAU-17 in the reaction mechanism. Notably, this energy barrier is appreciably higher than that observed for CAU-17 (ΔGOH* = 1.46 eV), indicating that UC-CAU-17 exhibits significantly weaker OH* binding strength. The result indicates the UC-CAU-17 structure is highly selective and active for H2O2 production via the 2e−-WOR pathway.

Fig. 1: Theoretical predictions for selective 2e−-WOR using different coordination constructions.
Fig. 1: Theoretical predictions for selective 2e−-WOR using different coordination constructions.The alternative text for this image may have been generated using AI.
Full size image

a Selective 2e−-WOR pathway by adjusting suitable OH* adsorption on catalytically active sites. Constructed models of b CAU-17 and c UC-CAU-17 structures (purple: Bi atom; yellow: O atom; green: C atom; pink: H atom). d The free energy diagrams of CAU-17 and UC-CAU-17 for 2e–-WOR process. Source data are provided as a Source Data file.

Catalyst preparation and characterization

The UC-CAU-17 was synthesized via a one-step solvothermal strategy using a certain proportion of bismuth-containing precursor of Bi(NO3)3·5H2O, H3BTC, and H2IPA dissolved in ethanol and deionized water, as illustrated in Supplementary Fig. 1. CAU-17 was obtained using the same protocol, except that the H2IPA was not added22. The X-ray diffraction (XRD) pattern of UC-CAU-17 was in good agreement with the typical diffraction peaks of CAU-17 (Fig. 2a and Supplementary Fig. 2), demonstrating the introduction of H2IPA linker into CAU-17 without changing the pristine crystal structure21,25. This was also verified by Raman spectroscopy (Supplementary Fig. 3). The high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image (Fig. 2b) and high-resolution TEM (HRTEM) image (Supplementary Fig. 4) showed a clear fringe with a 0.846 nm interval, corresponding to the lattice plane of CAU-17. Upon doping the H2IPA linker, this fringe expanded to 0.917 nm (Fig. 2c), which is ascribed to a hexagonal channel22,25, indicating a partial structural distortion occurred26,27. Energy-dispersive X-ray spectroscopy (EDX) elemental mapping images demonstrated that each element was evenly distributed (Fig. 2d) on UC-CAU-17.

Fig. 2: Morphological and fine structure analyses of UC-CAU-17.
Fig. 2: Morphological and fine structure analyses of UC-CAU-17.The alternative text for this image may have been generated using AI.
Full size image

a XRD patterns, HAADF-STEM images of CAU-17 b and UC-CAU-17 (c). d HRTEM image and corresponding EDX elemental mapping of UC-CAU-17. e, f Bi L3-edge XANES spectra and corresponding k3-weighted EXAFS spectra. WT-EXAFS spectra of g Bi powder, h CAU-17, and i UC-CAU-17. Source data are provided as a Source Data file.

To elucidate the electronic structure of catalyst UC-CAU-17, X-ray photoelectron spectroscopy (XPS) was employed. In the high-resolution Bi 4 f doublet, it is seen that the incorporation of the H2IPA linker resulted in a shift towards higher binding energy (Supplementary Figs. 5 and 6). In contrast, the peaks at 531.5 eV and 533.1 eV of high-resolution O 1 s spectra are ascribed to the Bi-O bond and the O in bismuth-oxo clusters22 of UC-CAU-17, respectively; the Bi-O peak redshifted to lower binding energies after doping H2IPA (Supplementary Fig. 7). This suggests a partial electron transfer occurred from the Bi atoms to H2IPA linker changes the valence-band structure of the entire Bi-MOF, and the repulsion between p-orbitals of Bi and bridging O facilitates the electron transfer28,29. Moreover, a negligible shift of the high-resolution C 1 s XPS spectra in UC-CAU-17 was observed compared with pristine CAU-17 (Supplementary Fig. 8), indicating that the introduction H2IPA linker modified the intrinsic electronic structure and efficiently adjusted the electron configuration of the center Bi atoms.

Furthermore, we used X-ray absorption spectroscopy (XAS) to examine the fine coordination structure of UC-CAU-17 at the atomic level. As illustrated in the Bi L3-edge X-ray absorption near-edge structure (XANES) spectra (Fig. 2e), UC-CAU-17 exhibits an adsorption edge position that is between the reference Bi power and Bi2O3 powder, revealing that the valence state of Bi in UC-CAU-17 is between 0 and +3. Additionally, the location of the absorption edge of UC-CAU-17 slightly surpasses that of CAU-17, implying a more positive valence of Bi species in UC-CAU-1728. The Fourier transform (FT) k3-weighted extended X-ray absorption fine structure (FT-EXAFS) spectra of UC-CAU-17 show one dominant peak at 1.6 Å in R-space, assigned to the Bi-O coordination-sphere along with oscillations around 3.6 Å associated with Bi-Bi coordination, which is almost identical to that of Bi2O3 powder8,30. For EXAFS fitting analysis, the average coordination number of O atoms around Bi atoms in the UC-CAU-17 and CAU-17 is determined as 3.5 ± 0.5 and 4.4 ± 0.8, respectively (Fig. 2f and Supplementary Table 2), which is caused by the structural difference between H2IPA and H3BTC ligands. To determine the linker vacancies per Bi node, 1H nuclear magnetic resonance (NMR) and thermogravimetric analysis (TGA) were conducted. As shown in Supplementary Figs. 9 and 10, the calculated ratio of the H3BTC: H2IPA linker is 1.38:0.36, which is close to the stoichiometric ratios (7:2) used in the synthesis. According to the above results, we calculated the linker vacancies per Bi node as 0.62, which is consistent with the structural parameters of UC-CAU-17 extracted from the EXAFS fitting. Moreover, the wavelet transform (WT) maximum for UC-CAU-17 and CAU-17 at the Bi L3-edge displays lower R-values than that for Bi powder (Fig. 2g), corresponding to a shorter Bi-Bi bond formation. Moreover, for the Bi-O path, the decreased K-values from UC-CAU-17 (Fig. 2i) to CAU-17 (Fig. 2h) match well with the lower Bi-O coordination numbers among UC-CAU-1731, indicating that an unsaturated coordinative Bi structure formed in UC-CAU-17 due to partial H2IPA linker replacement.

2e−-WOR performance for H2O2 electrosynthesis

The 2e−-WOR performance was evaluated in a 1.0 M Na2CO3 solution using a typical three-electrode configuration with the as-prepared catalyst sprayed onto carbon paper or FTO as the working electrode. The production of H2O2 in the electrolyte was determined using the KMnO4 titration method11 (see Methods for details). On both carbon paper (Fig. 3a, c) and FTO (Fig. 3b, d and Supplementary Figs. 11–14) substrates, UC-CAU-17 exhibited remarkable activity enhancements in both partial current density of H2O2 (jH2O2) and FE of H2O2 (FEH2O2) for 2e−-WOR catalysis compared with CAU-17, highlighting the significant role of unsaturated coordination Bi site in enhancing 2e−-WOR selectivity and activity. We further loaded the UC-CAU-17 catalyst on the surface of FTO substrate (UC-CAU-17/FTO), which demonstrated a commendable catalytic performance, featured a maximum FEH2O2 of 79.1% at 3.2 V versus a reversible hydrogen electrode (vs RHE) with a jH2O2 of 71.27 mA cm−2, while these values of CAU-17 supported on FTO substrate were only 43.8% and 32.15 mA cm−2, respectively.

Fig. 3: 2e−-WOR performance for H2O2 electrosynthesis.
Fig. 3: 2e−-WOR performance for H2O2 electrosynthesis.The alternative text for this image may have been generated using AI.
Full size image

a, b Partial current densities of H2O2 and c, d the corresponding FEH2O2 over CAU-17 and UC-CAU-17 catalysts supported on a, c carbon paper and b, d FTO, respectively. e Long-time stability (E and FEH2O2) of UC-CAU-17/FTO at 50 mA cm–2. f Comparison of jH2O2, FEH2O2, overpotential at 10 mA cm–2, and stability with other previously reported state-of-the-art 2e–-WOR electrocatalysts. Electrolyte: 1.0 M Na2CO3 solution (pH = 11.96 ± 0.08, 25 °C). Potentials reported without iR correction. The error bars represent the standard deviation of three independent measurements. Source data are provided as a Source Data file.

To unravel the intrinsic activity of UC-CAU-17, the electrochemical impedance spectra (EIS) and electrochemically active surface area (ECSA) were performed. As evident from the Nyquist plots (Supplementary Fig. 15), UC-CAU-17 displayed smaller charge transfer impedances (Supplementary Table 3) relative to pristine CAU-17. Furthermore, UC-CAU-17 possessed a double-layer capacitance (Cdl) of 14.8 mF cm–2, which is 5.3 times larger than that of CAU-17 (2.8 mF cm–2) (Supplementary Figs. 16 and 17). These findings further highlighted that the unsaturated coordination structure of Bi sites enhanced the intrinsic 2e–-WOR performance within UC-CAU-17. The turnover frequency (TOF) value for UC-CAU-17 was determined to be 1.39 s–1, surpassing that of CAU-17 (0.37 s–1) (Supplementary Table 4 and Supplementary Fig. 18). This enhanced TOF corresponds with the significantly improved H2O2 production performance observed in UC-CAU-17 compared to CAU-17. Moreover, as presented in Supplementary Fig. 19, the UC-CAU-17 and CAU-17 catalysts exhibited specific surface areas of 6.18 and 5.00 m2 g–1, respectively, confirming their essentially non-porous nature. Given the negligible surface areas, the enhanced catalytic performance of UC-CAU-17 in the 2e–-WOR can be primarily attributed to its intrinsic catalytic activity rather than surface area effects. This conclusion is further supported by the ECSA measurements (Supplementary Fig. 17), which reinforce the role of active site efficiency in governing the observed catalytic behavior. Additionally, UC-CAU-17/FTO demonstrated a negligible loss in FEH2O2 and maintained a stable potential over 100 h of continuous electrolysis at 50 mA cm–2 (Fig. 3e). Notably, even after refreshing the electrolyte and reusing the UC-CAU-17/FTO electrode for subsequent 100-h electrolysis run, the catalyst still maintained a high FEH2O2 and showed no significant shift in the applied potential (Supplementary Fig. 20). Moreover, the UC-CAU-17/FTO catalysts well-preserved its original structure without transforming into corresponding oxides or hydroxides during electrocatalytic testing, as evidenced by comprehensive XRD patterns and XPS spectra analysis (Supplementary Figs. 21 and 22). These results collectively demonstrate the high electrochemical stability and structural durability of the UC-CAU-17 catalyst. Such a performance was superior to that of all previously reported 2e–-WOR electrocatalysts (Fig. 3f and Supplementary Table 1).

Insight into selective 2e–-WOR

To clarify the high activity and selectivity mechanism of UC-CAU-17 for 2e–-WOR catalysis, we first evaluated the intermediate OH* adsorption capacity of the UC-CAU-17 using methanol oxidation experiments. As illustrated in Supplementary Fig. 23, the UC-CAU-17 exhibited a 102 mV negative shift in onset potential, significantly more extensive than that of pristine CAU-17 (78 mV). This difference indicates the optimized OH* adsorption capability of UC-CAU-1732,33, emphasizing the crucial role of unsaturated coordination in optimizing MOF-based catalysts. Further, in-situ ATR-FTIR measurements were performed to probe the key intermediates during the 2e–-WOR process. It can be found that positive peaks located at approximately 1220 and 1055 cm–1 for UC-CAU-17 and CAU-17 are well assigned to the Si-O-Si stretching band and adsorbed OOH* species, which is beneficial for O2 generation. Especially, the peak at 3500 cm–1 is ascribed to OH* species34,35. As the applied potential increased, the intensity of the OH* peak on UC-CAU-17 became significantly stronger (Fig. 4b, e and Supplementary Fig. 24), which is different from that on CAU-17, indicating that more OH* species are adsorbed at the unsaturated coordinative Bi sites to generate H2O2. Compared with the contour image of CAU-17 (Fig. 4c), the OOH* species (red region) emerged at a potential of 1.20 V on CAU-17, and the intensity of the characteristic peak increased with increasing potential. In contrast, there is no obvious peak at around 1055 cm–1 on UC-CAU-17 (Fig. 4f) at the whole potential range, indicating that the OOH* is more difficult to generate on unsaturated coordinative Bi sites. Based on the above results, one can conclude that unsaturated coordinative Bi sites on UC-CAU-17 optimized the OH* adsorption energy on the catalyst surface (Fig. 4d), hindering further oxidation of adsorbed OH* to form O* and OOH* intermediates, restraining the competitive 4e–-OER pathway. In other words, all newly generated OH* interacted on the singlet potential surface to form H2O2 directly rather than ground-state O2 molecules36. Therefore, UC-CAU-17 possesses the unique capacity to trigger the 2e–-WOR for selective electrosynthesis of H2O2.

Fig. 4: In-situ ATR-FTIR measurements of 2e–-WOR and mechanism analysis.
Fig. 4: In-situ ATR-FTIR measurements of 2e–-WOR and mechanism analysis.The alternative text for this image may have been generated using AI.
Full size image

a Schematic illustration of in-situ ATR-FTIR tests. In-situ ATR-FTIR spectra and the corresponding contour images for b, c CAU-17 and e, f UC-CAU-17. Potentials reported without iR correction. d Schematics of different reaction pathways for water oxidation on CAU-17 and UC-CAU-17. g Free energy diagrams of UC-CAU-17 for 4e–-OER and 2e–-WOR processes. h Charge density differences of CAU-17 and UC-CAU-17, yellow and blue contours represent the isosurfaces of electronic charge accumulation and depletion, respectively. The isosurface level of UC-CAU-17 and CAU-7 is 0.000934052 e/bohr3. DOS results of the Bi p orbital in i UC-CAU-17 and j CAU-17 with the OH p orbital. Source data are provided as a Source Data file.

To elucidate the fundamental origin of this catalytic behavior, we conducted comprehensive DFT calculations incorporating solvent effects to decode how the unsaturated coordinative environment at Bi active sites influences reaction energetics (Supplementary Figs. 25–27 and Supplementary Data 1). The free energy diagram for competing 2e−-WOR and 4e−-OER pathways on UC-CAU-17 (Fig. 4g) demonstrates that while the initial OH* adsorption energy remains identical for both processes, the subsequent mechanistic steps diverge significantly. The energy barrier for the OH* → O* transition (1.93 eV) substantially exceeds that for OH* → H2O2 formation (1.74 eV), thermodynamically favoring the peroxide generation pathway. Furthermore, the elevated energy requirement for the O* → OOH* conversion on UC-CAU-17 (1.22 eV) compared to CAU-17 (1.08 eV) (Supplementary Fig. 28) indicates that the unsaturated coordination environment effectively suppresses the conventional 4e−-OER process. The above results claim the OH*-OH* combination is the vital step to the 2e–-WOR process, disclosing that the OH* is more liable to generate H2O2 with unsaturated coordinative Bi sites selectively. The electron density difference (Fig. 4h) result reveals that the electron-deficient state around the Bi site is more prominent in UC-CAU-17, indicating unsaturated coordinative Bi site transfers out more electrons to form an electron-deficient Bi center. Specifically, the net charge loss at the Bi sites increases from +0.95 e in CAU-17 to +0.98 e in UC-CAU-17, and the change of valence state is in accordance with the XANES and XPS results (Fig. 2e and Supplementary Figs. 6 and 7). Although the difference in charge values is relatively small, the trend shows that the unsaturated coordinative environment modulates the generation of electron-deficient Bi sites. Given that Bi sites preferentially adsorb oxygen-containing intermediates, particularly OH*, through hybridization between O 2p and Bi 6p orbitals, we investigate the orbital interactions between OH* and Bi sites in different coordinative environments. Electronic structure analysis (Fig. 4i, j) reveals that the reduced coordination number at Bi sites causes a pronounced shift in the p-band center of UC-CAU-17 (−2.86 eV) away from the Fermi level relative to CAU-17 (−2.73 eV), providing a fundamental electronic basis for the observed alterations in adsorption behavior and reaction selectivity. Simultaneously, there is enhanced overlap between the Bi 6p and O 2p orbitals in the OH*, indicating optimized adsorption energy between the Bi active sites and OH* species37,38. Moreover, the density of states (DOS) analysis reveals a slight difference between the electronic structures of UC-CAU-17 (Supplementary Fig. 29) and CAU-17 (Supplementary Fig. 30), suggesting a negligible improvement of electrical conductivity due to the presence of unsaturated coordinative Bi sites. Moreover, the proposed 2e–-WOR mechanism is further unraveled by electron paramagnetic resonance (EPR) analysis and electrocatalytic performance measurement with Na2CO3 and NaHCO3 electrolytes. As shown in Supplementary Fig. 31, the solution obtained during electrolysis using the in-situ trapped method with 5,5-Dimethyl-1-Pyrroline-N-Oxide (DMPO) trap, exhibited a clear four-line 1:2:2:1 splitting pattern characteristic of the DMPO•-OH adduct, indicating that the carbonate radical was formed at the surface of the UC-CAU-17 electrode11. However, in stark contrast to previous reports39, the UC-CAU-17 catalyst exhibited much better 2e−-WOR performance in 1.0 M Na2CO3 electrolyte compared with that in the same concentration of NaHCO3 (Supplementary Figs. 32 and 33). No promotion effect could be observed using bicarbonate electrolyte, indicating the electrocatalytic formation of H2O2 using UC-CAU-17 catalysts follows the direct pathway rather than relying on carbonate radical or percarbonate mediator to promote H2O2 generation40. These findings demonstrate that the unsaturated coordination in UC-CAU-17 modifies intrinsic electronic structure, optimizing OH* adsorption and restraining the intermediate transition of OH* → O*, directly contributing to the superior 2e–-WOR performance.

Butanone conversion to butanone oxime via in-situ formed H2O2

To evaluate the practical applicability and feasibility of in-situ generated H2O2 over UC-CAU-17, we designed and constructed an integrated catalytic system for the electrosynthesis of butanone oxime from butanone with the assistance of ammonia (NH3) and H2O2 under mild conditions41,42. Employing electrochemically produced H2O2 acts as an environmentally benign oxidant, and titanium silicalite-1 (TS-1) acts as a highly selective catalyst, as illustrated in Fig. 5a. The TS-1 catalyst activates H2O2 to form peroxy-titanium species, which oxidize NH3 into hydroxylamine, then react with butanone to produce butanone oxime. After continuous electrolysis for 2 h at an applied potential of 3.2 V vs. RHE, the accumulated H2O2 concentration reached ~6.17 wt.% over UC-CAU-17/FTO, nearly 3.0 times higher than that produced by CAU-17/FTO (2.12 wt.%). This significant difference in H2O2 concentration was the primary factor influencing the catalytic performance in butanone ammoximation. As displayed in Fig. 5b, the in-situ produced H2O2 on UC-CAU-17/FTO delivered a major yield of butanone oxime after the ammoximation process, in which the conversion rate of butanone and selectivity towards butanone oxime on UC-CAU-17/FTO are 80.2% and 81.1%, respectively, calculated from 1H NMR results (Fig. 5c and Supplementary Fig. 34), much higher than those of CAU-17/FTO (61.4% and 72.4%). This on-site generated and sustainable H2O2 electrosynthesis system achieved direct butanone oxime synthesis, fundamentally figuring out the difficulty of separating H2O2 from the reaction solution.

Fig. 5: In-situ generated H2O2 for butanone ammoximation to butanone oxime synthesis.
Fig. 5: In-situ generated H2O2 for butanone ammoximation to butanone oxime synthesis.The alternative text for this image may have been generated using AI.
Full size image

a The schematic diagram for butanone ammoximation to form butanone oxime via in-situ electro-synthesized H2O2. b The butanone conversion rates and butanone oxime selectivity for butanone ammoximation reaction with in-situ obtained H2O2 on UC-CAU-17/FTO and CAU-17/FTO. c1H NMR measurements of the electrolyte after butanone ammoximation with in-situ obtained H2O2. The error bars represent the standard deviation of three independent measurements. Source data are provided as a Source Data file.

Discussion

In summary, we developed a hetero-linker doping strategy for synthesizing UC-CAU-17 with unsaturated coordinative Bi sites as a high-activity robust catalyst for 2e–-WOR toward H2O2 electrosynthesis, which exhibited a low overpotential of 0.98 V to reach 10 mA cm–2 and a high FEH2O2 of 79.1%. Further utilizing accumulated H2O2 for butanone ammoximation, displayed a butanone conversion rate of 80.2%, and high selectivity of 81.1% to form butanone oxime, enabling the efficient usage of in-situ generated H2O2. Combined with in-situ ATR-FTIR spectroscopy and DFT calculations, it revealed that the unsaturated coordinative Bi sites in the UC-CAU-17 facilitated OH* combination while inhibiting its conversion to O* and OOH*, strengthening Bi 6p and O 2p orbital hybridization and lowering the free energy for H2O2 formation, thereby accelerating 2e–-WOR reaction kinetics and improving H2O2 electrosynthesis selectivity. This work demonstrates the potential of this hetero-linker doping strategy for developing tailored MOF electrocatalysts for specific reaction pathways.

Methods

Chemicals

Benzene tricarboxylic acid (H3BTC, ≥97.0%) and isophthalic acid (H2IPA, ≥99.0%) were purchased from Macklin Biochemical Co., Ltd; bismuth (III) nitrate pentahydrate (Bi(NO3)3·5H2O, ≥99.5%), ammonia (NH3, ≥99.9%), sodium carbonate (Na2CO3, ≥99.5%), butanone (≥99.0%), butanone oxime (≥99.0%), N, N-dimethylformamide (DMF, ≥99.5%) and ethanol absolute (≥99.7%) were bought from Sinopharm Chemical Reagent Co., Ltd. The water used was deionized (DI) water. All the chemicals were used as obtained without further purification.

Synthesis of UC-CAU-17 (UC-CAU-17-7:2)

0.7 mmol of benzene tricarboxylic acid (H3BTC), 0.2 mmol of isophthalic acid (H2IPA), and 0.9 mmol of Bi(NO3)3·5H2O were dissolved in 60 mL of ethanol at room temperature (25 °C). The obtained mixture was transferred into a 100 mL Teflon-lined steel reactor after the powder was completely dissolved. Then the reactor was heated to 120 °C for 24 h. Subsequently, the precipitates were centrifuged with dimethyl formamide (DMF) and ethanol three times and dried in a vacuum oven at 60 °C to obtain UC-CAU-17. The sample synthesized at this specific ratio (7:2) demonstrated optimal catalytic performance and was therefore selected as the primary sample. Throughout this manuscript and in the Supplementary Information, ‘UC-CAU-17’ refers to the composition UC-CAU-17-7:2.

Synthesis of CAU-17

In a traditional synthesis, H3BTC (0.9 mmol) and Bi(NO3)3·5H2O (0.9 mmol) were mixed with 60 mL of ethanol at room temperature (25 °C). The obtained mixture was transferred into the 100 mL Teflon-lined steel reactor after the powder was completely dissolved. Then the reactor was heated to 120 °C for 24 h. Subsequently, the precipitates were centrifuged with dimethyl formamide (DMF) and ethanol three times and dried in a vacuum oven at 60 °C.

Synthesis of UC-CAU-17-8:1

0.8 mmol of H3BTC, 0.1 mmol of H2IPA, and 0.9 mmol of Bi(NO3)3·5H2O were dissolved in 60 mL of ethanol at room temperature (25 °C). The obtained mixture was transferred into the 100 mL Teflon-lined steel reactor after the powder was completely dissolved. Then the reactor was heated to 120 °C for 24 h. Subsequently, the precipitates were centrifuged with DMF and ethanol three times and dried in a vacuum oven at 60 °C to obtain UC-CAU-17-8:1.

Synthesis of UC-CAU-17-6:3

0.6 mmol of H3BTC, 0.3 mmol of H2IPA, and 0.9 mmol of Bi(NO3)3·5H2O were dissolved in 60 mL of ethanol at room temperature (25 °C). The obtained mixture was transferred into the 100 mL Teflon-lined steel reactor after the powder was completely dissolved. Then the reactor was heated to 120 °C for 24 h. Subsequently, the precipitates were centrifuged with DMF and ethanol three times and dried in a vacuum oven at 60 °C to obtain UC-CAU-17-6:3.

Preparation of CAU-17/FTO and UC-CAU-17/FTO

To prepare the catalyst-loaded FTO substrate, 20 mg of CAU-17 or UC-CAU-17 was suspended into 1.8 mL of ethanol and 0.2 mL of Nafion solution, and then the suspension was ultrasonicated for 8 h until the evenly dispersed catalyst ink was formed. The as-prepared catalysts’ ink was sprayed onto the FTO (1 × 1 cm2) substrate.

Characterizations

The morphologies of as-prepared samples were obtained by high-resolution transmission electron microscopy (HRTEM) (HT7700) with EDX (Oxford, X-max80) and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) (Titan Cubed Themis G2 300). The crystal structures of as-prepared samples were analyzed by X-ray powder diffraction (XRD) (Empyrean 200895) using Cu Kα radiation. The chemical structures of the as-prepared samples were measured by X-ray photoelectron spectroscopy (XPS, Escalab250Xi) with Al Kα radiation. Raman spectra of as-prepared samples were obtained by a LabRAM HR Evolution. X-ray absorption spectroscopy (XAS) measurements of as-prepared samples were conducted at the Beijing Synchrotron Radiation Facility and Shanghai Synchrotron Radiation Facility. The 1H nuclear magnetic resonance (NMR) spectroscopy was analyzed by a 400 MHz Bruker Avance AV1. The electron paramagnetic resonance (EPR) measurements were taken by a Bruker A300 spectrometer. The N2 adsorption-desorption curves of the as-prepared samples were tested by Brunauer-Emmett-Teller measurement (BET). The metal content in samples was analysed by inductively coupled plasma atomic emission spectrometry ICP-AES (Agilent 720). Thermogravimetric analysis (TGA) was performed on a Rigaku TG thermal gravimetric analyzer in the temperature range of 30–700 °C under a nitrogen atmosphere at a heating rate of 10 °C min−1.

In-situ ATR-FTIR tests

In-situ ATR-FTIR tests were carried out on an Invenio-R spectrometer equipped with a mercury cadmium telluride detector using liquid nitrogen to cool down. The Au-coated Si hemispherical prism was used as the conductive substrate for catalyst deposition and the IR reflection element. Saturated calomel electrode (SCE) and Pt electrode were used as reference electrodes and counter electrodes, respectively.

Electrochemical measurements

All electrochemical measurements were carried out at room temperature (25 °C) using an electrochemical workstation (CHI 760E). A customized gas-tight H-type glass cell with a Nafion 117 membrane (Fuel Cell Store, 180 μm, 2 × 2 cm2) was employed to separate cell compartments and ensure gas tightness during the experiments. In a typical three-electrode system, a platinum wire and an SCE (CH Instruments) were used as the counter and reference electrode, respectively. The working electrode was a 1  × 1 cm2 carbon paper or FTO with a catalyst loading amount of 1.0 mg cm−2. Before measurements, all samples were pre-stabilized at 1.2 V vs. RHE to achieve a stable current density in 1.0 M Na2CO3 electrolyte (pH = 11.96 ± 0.08), and the electrolyte was used immediately after preparation without storage. The electrolyte in the anodic compartment was stirred at a rate of 2,000 rpm during electrolysis. All potentials measured against SCE (ESCE) were converted to the RHE (ERHE) scale without iR correction in this work using ERHE = ESCE ×  0.244 V + 0.0591 × pH. No iR compensation was applied during data acquisition and all reported potential values represent the applied voltages without correction for solution resistance. Electrochemical impedance spectroscopy was tested at an open circuit voltage with an AC amplitude of 5 mV. Solution resistance in 1.0 M Na2CO3 is 0.037 ± 0.006 for catalysts.

After electrolysis with the passing of ~10 C, the generated H2O2 concentration was confirmed using the standard potassium permanganate (0.1 nmol KMnO4 solution) titration process11,43. To quantify the gas products during electrolysis, argon gas was delivered into the anodic compartment and vented into a gas chromatograph.

The FEH2O2 is calculated using the following equations:

$${{{\rm{FE}}}}=\frac{{{{\rm{generated}}}}\;{H}_{2}{O}_{2}\left({{{\rm{mol}}}}\right)\times 2\times 96485}{{{{\rm{the}}}}\; {{{\rm{total}}}}\; {{{\rm{amount}}}}\; {{{\rm{of}}}}\; {{{\rm{charge}}}}\; {{{\rm{passed}}}}({{{\rm{C}}}})}\times 100$$
(1)

For the stability test, a continuous three-electrode H-type cell was employed to continuously produce H2O2 at a constant current.

To quantify the gas products during electrolysis, argon gas (Airgas, 99.995%) was delivered into the anodic compartment at a rate of 20 sccm and vented into a gas chromatograph. A thermal conductivity detector was mainly used to quantify gas product concentration. The partial current density for O2 (jO2) produced was calculated as follows43:

$${j}_{{O}_{2}}=x{{{\rm{i}}}}\times v\times \frac{ni{FP}^\circ }{R{{{\rm{T}}}}}/({{{\rm{electrode\; area}}}})$$
(2)

where xi is the volume fraction of the certain product determined by online gas chromatography referenced to calibration curves from the standard gas sample (Airgas), v is the flow rate of 20 sccm, ni is the number of electrons involved, P° = 101.3 kPa, F is the Faradaic constant, T = 298 K and R is the gas constant. The corresponding FE at each potential is calculated as jO2/j × 100.

The partial current density for the H2O2 (jH2O2) produced was calculated as follows:

$${j}_{{H}_{2}{O}_{2}}=j-{j}_{{O}_{2}}$$
(3)

Butanone ammoximation to produce butanone oxime experiments

The butanone ammoximation towards butanone oxime formation was studied by batch experiments41,44. In a typical experiment, 20 mL of alkaline H2O2 solution (1.0 M Na2CO3) in an H-type cell was cycled for 2 h at the potential of 3.2 V vs. RHE. Then, the obtained electrolyte after the 2e–-WOR reaction was transferred to a 50 mL glass reactor bottle. Afterward, 0.5 g titanium silicalite-1 as a catalyst and 100 mM of butanone were added to the above bottle and kept stirring. Then, 540 μL of ammonia was slowly added dropwise into the above solution and heated in the water bath at a controlled temperature of 80 °C. After 3 h of reaction, the obtained solution was collected to conduct 1H NMR spectroscopy analysis, in which 50 mM of maleic acid was selected as the internal standard and deuterium oxide as the lock solvent.

Butanone conversion and butanone oxime selectivity were calculated based on the initial amount of butanone, according to the following formulas. Where C0 is the initial concentration of the feed, C1 is the concentration of butanone extracted from the reactor after the ammoximation process, and CBO is the concentration of butanone oxime extracted from the reactor after the ammoximation process.

$${{{\rm{Conversion\; rate}}}}\;(\%)=\frac{{{{\rm{Co}}}}-{{{\rm{C}}}}1}{{{{\rm{Co}}}}}\times 100$$
(4)
$${{{\rm{Selectivity}}}}\;(\%)=\frac{{{{\rm{CBO}}}}}{{{{\rm{Co}}}}-{{{\rm{C}}}}1}\times 100$$
(5)

Computational details

All the calculations were performed based on the spin-polarized density functional theory (DFT), as implemented in the Vienna Ab initio Simulation Package (VASP) code45. The projector augmented wave (PAW) method46 was chosen to describe the core electrons’ interactions. The electron exchange and correlation energy were described using the generalized gradient approximation (GGA)47 proposed by the Perdew-Burke-Ernzerhof (PBE)48 parametrization. A plane-wave cutoff energy of 450 eV was used to describe valence electrons. All the geometry optimizations were set to 10−5 eV/atom in energy convergence and 0.05 eV/Å in force convergence, respectively. A sample with a gamma k-point was set in the Brillouin zone, and 15 Å of vacuum in all of the x, y, and z-directions was employed to avoid periodic interactions. Weak interaction was described by the DFT-D3 method49 using an empirical correction in Grimme’s scheme.

As part of the key active center of CAU-17, the CAU-17 model contained one bismuth atom in coordination with three trimesic acid molecules and one water molecule. The UC-CAU-17 model was formed by replacing a trimeric acid molecule with an isophthalic acid molecule in the CAU-17 model.

The equation for the Gibbs free energy calculations is as follows:

$$G={E}_{0}+{E}_{{{{\rm{ZPV}}}}}-{TS}$$
(6)

where E0 is the total energy obtained from DFT calculations, EZPV is the zero-point energy, T is the temperature (298.15 K), and S represents the entropy change. Zero-point energy and entropies of the adsorbed species were calculated from the vibrational frequencies. The solvent effect was considered with an implicit solvation model implemented in the DFT solvation model VASPsol with a dielectric constant ε = 80 for water50,51. For all calculations, the optimized structures are provided in Supplementary Data 1.