ZrO2-enhanced Ni@C catalysts for the selective hydrogenation of maleic anhydride to γ-butyrolactone Authors: Shuowen Guo, Yanyang Yao, Wei Wei, Jiannan Lv, Dongjie Zhang, Weixun Wang, Yin Zhang, Haitao Li Published: 2026 ABSTRACT AbstractNi@C catalysts modified with ZrO2 (denoted as NiZr@C) were synthesized and evaluated for the liquid‑phase hydrogenation of maleic anhydride to γ‑butyrolactone. Characterization by XRD, TEM, XPS, H2‑TPD, ICP-OES, and in‑situ FTIR revealed that there is a strong metal–support interaction between Ni and ZrO2 with the introduction of ZrO2. During reduction procedure, with the reduction of Ni species, partial ZrO2 was reduced resulting in the formation of abundant surface oxygen vacancies on the catalyst surface. These vacancies act synergistically with activated hydrogen species, leading to a significant enhancement in both C = C hydrogenation and C = O hydrogenolysis activities. The NiZr@C catalyst with a Zr/Ni molar ratio of 2/10 exhibits optimal performance, achieving complete maleic anhydride conversion (100%) with 97.54% γ‑butyrolactone selectivity at 210 °C and 4.5 MPa H2, and maintains stable activity over 13 consecutive cycles without obvious deactivation, demonstrating great potential for industrial practical application.Subject terms: Catalyst synthesis, Heterogeneous catalysis Introduction As a pivotal chemical intermediate, γ-butyrolactone (GBL) has garnered significant and widespread utility in the domains of high-performance solvents, pharmaceutical precursors, and lithium-ion battery electrolytes. In recent years, driven by the burgeoning new energy vehicle sector, GBL, serving as an essential feedstock for the lithium-ion battery electrolyte N-methylpyrrolidone—has exhibited robust and expanding market demand1. Among the contemporary synthetic routes to GBL, the direct hydrogenation of maleic anhydride (MA) presents distinct advantages over both the Reppe process and esterification–hydrogenation pathways. These benefits notably encompass a shortened process flow, reduced equipment complexity, and enhanced operational simplicity2,3. As documented in the literature, catalysts employed for the direct hydrogenation of MA to GBL include both noble metal systems such as Pt/TiO2, PtPd/TiO2, Pd/BN, and Pd/AC4–8 and non-noble metal-based counterparts, including Cu/SBA-15, Cu-Ni/ZnO, Ni/Al2O3, and Ni/MMO, etc.9–18. Given the prohibitive cost and finite reserves associated with noble metals, which inherently constrain their viability for large-scale industrial deployment, research efforts have been predominantly directed toward the development of non-noble metal catalysts for MA hydrogenation to GBL. During the hydrogenation of MA, MA is first hydrogenated at the C=C bond to form succinic anhydride (SAH), which subsequently undergoes C=O hydrogenolysis to yield γ-butyrolactone (GBL). The C=C bond hydrogenation can proceed over metal catalysts under mild reaction conditions (50–100 °C, 3–5 MPa H2 pressure)5–7. In contrast, C=O hydrogenolysis requires higher temperature and H2 pressure (170–210 °C, 4–8 MPa) together with a catalyst possessing high C=O hydrogenolysis activity13–15. To enhance the C=O hydrogenolysis activity of the catalyst, a number of strategies have been developed, including: (i) Tuning the metal–support interaction by modifying the catalyst preparation method or varying the support. It has been found that catalysts with a specific strength of metal–support interaction exhibit excellent performance in MA hydrogenation to GBL7,19. For instance, For example, Regenhardt and co-workers compared the interactions of Ni with SiO2, SiO2–Al2O3, and H-BEA zeolite supports, and found that the Ni/SiO2–Al2O3 catalyst, which possessed moderate Ni²⁺–support interactions, delivered the highest GBL selectivity11. (ii) Introducing promoters to modify the electronic configuration of the active metal and thereby its hydrogenation activity1,20,21. For instance, Jung and co-workers proposed that Sn, through its interaction with Pd, alters the d-orbital electronic configuration of Pd/SiO2 catalysts; the resulting increase in d-orbital electron density lowers the MA adsorption energy, thereby enhancing the GBL selectivity from 23.7% to 57.4%20. and (iii) Employing reducible oxides (e.g., ZrO2, CeO2, or TiO2) as the support. During catalyst reduction, the reducible oxide at the metal–support interface can be partially reduced to generate oxygen vacancies and induce strong metal–support interactions. Through the synergy between the oxygen vacancies and the active metal, the catalyst exhibits outstanding C=O hydrogenolysis activity4,16,17. For example, Liao et al. prepared a Ni/CeO2−δ catalyst by wet impregnation. During MA hydrogenation, the spillover of metal-activated hydrogen onto the CeO2 surface promotes the formation of oxygen vacancies, reduces the energy barrier for C=O hydrogenolysis, and increases both the turnover frequency (TOF) of the SAH-to-GBL step and the GBL selectivity22. As an amphoteric oxide possessing both acid–base sites and redox properties, ZrO2 exhibits excellent catalytic performance as a catalyst support in numerous hydrogenation reactions23–28. In our previous work, Ni-based catalysts supported on ZrO2, SiO2, and Al2O3 were prepared and evaluated for maleic anhydride (MA) hydrogenation. It was found that the Ni/ZrO2 catalyst delivered a γ-butyrolactone (GBL) yield up to 79.2%, far exceeding those of the Ni/SiO2 catalyst (60.0%) and the Ni/Al2O3 catalyst (17.8%)29–31. To gain deeper insight into the role of ZrO2 in MA hydrogenation, a Ni@C catalyst with a simple preparation procedure was designed based on our earlier studies, and different amounts of ZrO2 were introduced into the catalyst to systematically investigate the influence of ZrO2 incorporation on its MA hydrogenation performance. The results demonstrate that carbothermal reduction treatment induces the migration of ZrO2, which then forms strong interactions with the active Ni species. The oxygen vacancies generated from ZrO2 during the reduction process construct an electron-rich Ni surface, which facilitates the heterolytic cleavage of H2 and the adsorption and activation of reactants, endowing the catalyst with outstanding overall catalytic performance in the direct hydrogenation of MA to GBL and holding promising potential for industrial application. Results Catalytic performance Figure 1 presents the MA hydrogenation evaluation results for all catalysts. As shown in Fig. 1a, after 1 h of reaction, the MA conversions over Ni@C, NiZr@C-1, NiZr@C-2, and NiZr@C-3 reached 97.2, 99.6, 99.8, and 98.5%, respectively. As the reaction progressed, the MA conversion over all catalysts gradually increased. After 2 h, the MA conversion over NiZr@C-2 reached 100%, whereas those over Ni@C, NiZr@C-1, and NiZr@C-3 reached 100% after 4, 3, and 3 h, respectively. NiZr@C-2 required the shortest time to achieve complete MA conversion, indicating its highest MA hydrogenation activity. Catalytic performance for MA hydrogenation. a MA conversion, b SAH selectivity, and c GBL selectivity. Reaction conditions: 0.5 g catalyst, 0.025 mol MA, 40 mL 1,4‑dioxane, 210 °C, 450 rpm, 4.5 MPa H2. Data were presented as the median values from at least three independent experiments, with experimental errors within ±0.5%. Figure 1b shows that at the initial stage of the reaction, SAH was the major product over all catalysts, suggesting that the C=C bond of MA was preferentially hydrogenated on the catalyst surfaces. As the reaction time increased, the selectivity to SAH gradually decreased, accompanied by an increase in GBL selectivity. No other MA hydrogenation products were detected, indicating that prolonging the reaction time primarily promoted the hydrogenolysis of the C=O bond in SAH to form GBL. Figure 1c reveals that under identical reaction conditions, the GBL selectivity initially increased and then decreased with increasing ZrO2 content. NiZr@C-2 exhibited the highest GBL selectivity among the catalysts, reaching 97.5% after 8 h of reaction. To better contextualize the catalytic performance of NiZr@C-2, we compared it with representative non-noble metal catalysts reported in the literature for the hydrogenation of maleic anhydride to γ-butyrolactone. As summarized in Table 1, most previously reported Ni-based catalysts achieve complete MA conversion under various reaction conditions; however, their GBL selectivity and stability vary considerably. NiZr@C-2 catalyst achieves a GBL selectivity of 97.5% under the reaction conditions employed and demonstrates exceptional stability over 13 consecutive cycles, outperforming most reported non-noble metal systems. This superior performance can be attributed to the synergistic interaction between metallic Ni and ZrO2, which enhances Ni dispersion, promotes C=O bond activation, and effectively suppresses catalyst deactivation. Table 1Comparison of catalytic performance of various catalysts for the hydrogenation of maleic anhydride to γ-butyrolactoneCatalystTemperature (°C)H2 pressure (MPa)Catalyst dosage (g)MA conversion (%)GBL selectivity (%)CyclesReferenceNi-Zr-MFI20050.310095.6532Ni/SiO2–Al2O3–PD,220-0.0510083-33Ni/γ-Al2O3190-1100--34PtPdCoNiCu/TiO226030.110091-35Ni/CeO221050.110037-36Ni5Co1/Clay1801110090-37Ni-PS20050.310070-38Ni/CeO2-δ21050.11008-22Ni/TiO226030.110026.9-35Pd-Mo-Ni/SiO225071010082-39Ni@C20030.0510099.89531Ni-NiO/45021030.0510087.7540NiZr@C-22104.50.510097.513This workDashes indicate data not available.MA maleic anhydride, GBL γ‑butyrolactone. Comparison of catalytic performance of various catalysts for the hydrogenation of maleic anhydride to γ-butyrolactone Dashes indicate data not available. MA maleic anhydride, GBL γ‑butyrolactone. These findings establish a clear two-step reaction sequence: MA is first hydrogenated to SAH via C=C bond activation, followed by C=O hydrogenolysis of SAH to yield GBL (Scheme 1). The incorporation of ZrO2 into the Ni@C system markedly promotes the intrinsic activity for both steps. The NiZr@C‑2 catalyst, with an optimal Zr/Ni ratio, delivers the highest activity, effectively accelerating the overall reaction cascade. Proposed surface reaction pathway for the hydrogenation of MA to GBL over the synthesized catalysts.The NiZr@C-2 catalyst, with an optimal Zr/Ni ratio, delivers the highest activity and accelerates the overall reaction cascade. SAH succinic anhydride. The operational stability of the optimal NiZr@C‑2 catalyst was evaluated over 13 consecutive reaction cycles (Fig. 2). Remarkably, complete MA conversion (100%) was maintained throughout, with GBL selectivity consistently ranging from 97.9 to 99.3%, confirming its exceptional stability and practical potential for repeated use. Recycling stability test of the NiZr@C‑2 catalyst for MA hydrogenation to GBL. Reaction conditions: 0.5 g catalyst, 0.025 mol MA, 40 mL 1,4‑dioxane, 210 °C, 450 rpm, 4.5 MPa H2. Data were presented as the median values from at least three independent experiments, with experimental errors within ±0.5%.Reaction conditions: 0.5 g catalyst, 0.025 mol MA, 40 mL 1,4‑dioxane, 210 °C, 450 rpm, 4.5 MPa H2. Data were presented as the median values from at least three independent experiments, with experimental errors within ±0.5%. While these near-complete conversion results demonstrate the excellent overall performance and industrial potential of the catalysts, a rigorous comparison of their intrinsic activities requires kinetic analysis under differential (low-conversion) conditions, as presented in section “Kinetic Studies of Ni@C and NiZr@C-2”. Kinetic studies of Ni@C and NiZr@C-2 To quantitatively compare the intrinsic hydrogenation activities under differential conditions (where conversion is kept low to reflect true kinetic regimes), kinetic experiments were conducted for both the C=C hydrogenation (60–100 °C) and C=O hydrogenolysis (180–210 °C) steps. Mass-transfer limitations were excluded under a stirring rate of 450 rpm (Fig. S1), and the reaction rate became independent of H2 pressure above 4.5 MPa (Fig. S2). Hence, all kinetic measurements were performed at 450 rpm and 4.5 MPa H2. The reaction pressure of 4.5 MPa was selected because the C=O hydrogenolysis step requires elevated H2 pressure. Lower pressures led to poor GBL selectivity (<60%), while higher pressures offered limited gains. This value is consistent with literature reports (4–8 MPa) and balances performance with practical feasibility. The hydrogenation of MA to SAH was studied between 60 °C and 100 °C, where SAH was the sole product (Figs. 3a, b). Assuming first-order kinetics with respect to MA concentration, the rate equation is: where CMA denotes the concentration of MA, kMA the apparent rate constant, and dt the reaction time. Introducing the fractional conversion XMA=C0,MA−CMAC0,MA the initial concentration), Eq.(1) can be rewritten as: MA hydrogenation kinetics over Ni@C and NiZr@C2 at different temperatures. a, b MA conversion to SAH versus time for Ni@C and NiZr@C-2 at different temperatures. c, d Corresponding first-order kinetic plots of −ln(1−XMA) versus reaction time. Conditions: 0.025 mol MA, 40 mL 1,4-dioxane, 4.5 MPa H2, 450 rpm, 0.5 g catalyst, 3 h. Data were presented as the median values from at least three independent experiments, with experimental errors within ±0.5%. 0.025 mol MA, 40 mL 1,4-dioxane, 4.5 MPa H2, 450 rpm, 0.5 g catalyst, 3 h. Data were presented as the median values from at least three independent experiments, with experimental errors within ±0.5%. Integration of Eq.(2) gives the linearized form: Thus, for a first-order reaction, a linear relationship is expected between ln(1−XMA) and time t. As shown in Fig. 3c, d, plots of ln(1−XMA) versus t display strong linear correlations (R2 > 0.99) in the time range 30–120 min, confirming appar ent first-order kinetics for MA hydrogenation to SAH over both catalysts. The slopes of these lines give the apparent rate constants kMA(in min−1) at each temperature, listed in Table 2. Table 2Apparent rate constants kMA for MA hydrogenation to SAHCatalystk(min−1)60 °C80 °C100 °CNi@C8.02 × 10−41.35 × 10−32.16 × 10−3NiZr@C-29.21 × 10−41.67 × 10−32.59 × 10−3Apparent first-order rate constants (min−1) with R2 > 0.99.MA maleic anhydride, SAH succinic anhydride. Apparent rate constants kMA for MA hydrogenation to SAH Apparent first-order rate constants (min−1) with R2 > 0.99. MA maleic anhydride, SAH succinic anhydride. The activation energies for C=C hydrogenation were obtained from the Arrhenius relationship: where k is the rate constant, R the gas constant, T the absolute temperature, and AA the pre-exponential factor. Figure 4 presents the corresponding Arrhenius plots (lnk versus 1T), which show good linearity for both catalysts. From the slopes, the apparent activation energies are determined as 39.6 kJ mol−1 for Ni@C and 21.1 kJ mol−1 for NiZr@C-2. The significantly lower value for NiZr@C-2 reflects its enhanced intrinsic activity toward C=C bond hydrogenation. Arrhenius plots for the hydrogenation of MA to SAH over Ni@C and NiZr@C-2. Arrhenius plots of lnk versus 1T with good linearity. k, apparent rate constant; T, absolute temperature. MA maleic anhydride, SAH succinic anhydride. Hydrogenolysis of SAH to GBL. The conversion of SAH to GBL was studied between 180 and 210 °C. Within this range, GBL was the sole product (Fig. 5a, b), indicating selective C=O bond cleavage. Plots of ln(1−XSAH) versus reaction time exhibited strong linear correlations (Fig. 5c, d; R2 > 0.99) over 30–120 min, confirming apparent first-order kinetics. The apparent rate constants (kSAH)derived from the slopes, are listed in Table 3. SAH hydrogenation kinetics over Ni@C and NiZr@C2 at different temperatures. a, b SAH conversion to GBL versus time over Ni@C and NiZr@C-2 at different temperatures. c, d Corresponding first-order kinetic plots. Conditions: 0.025 mol SAH, 40 mL 1,4-dioxane, 4.5 MPa H2, 450 rpm, 0.5 g catalyst. Data were presented as the median values from at least three independent experiments, with experimental errors within ±0.5%. 0.025 mol SAH, 40 mL 1,4-dioxane, 4.5 MPa H2, 450 rpm, 0.5 g catalyst. Data were presented as the median values from at least three independent experiments, with experimental errors within ±0.5%. Table 3Apparent rate constants kSAH for SAH hydrogenolysisCatalystk(min−1)180 °C190 °C200 °C210 °CNi@C2.15 × 10−43.63 × 10−48.52 × 10−41.01 × 10−3NiZr@C-25.34 × 10−48.47 × 10−41.29 × 10−32.86 × 10−3Apparent first-order rate constants (min−1) with R2 > 0.99.SAH succinic anhydride, GBL γ-butyrolactone. Apparent rate constants kSAH for SAH hydrogenolysis Apparent first-order rate constants (min−1) with R2 > 0.99. SAH succinic anhydride, GBL γ-butyrolactone. Activation energy for C=O hydrogenolysis. Arrhenius plots for this step are presented in Fig. 6. The apparent activation energies, calculated from the slopes, are 103.6 kJ mol−1 for Ni@C and 95.8 kJ mol−1 for NiZr@C-2. The lower value for NiZr@C-2 quantitatively reflects its enhanced intrinsic activity for C=O bond cleavage. Arrhenius plots for SAH hydrogenolysis over Ni@C and NiZr@C-2. Arrhenius plots of lnk versus 1T with good linearity. k apparent rate constant, T absolute temperature, SAH succinic anhydride, GBL γ‑butyrolactone. In summary, the kinetic analysis demonstrates that NiZr@C-2 exhibits lower apparent activation energies than Ni@C for both the C=C hydrogenation (39.6 vs. 21.1 kJ mol−1) and C=O hydrogenolysis (103.6 vs. 95.8 kJ mol−1) steps. This result provides quantitative evidence that the incorporation of ZrO2 enhances the intrinsic activity of the Ni@C catalyst across the entire reaction cascade from MA to GBL. Structural and textural characterization X-ray diffraction analysis The XRD patterns of the synthesized catalysts are presented in Fig. 7. All samples exhibit characteristic reflections at 44.5°, 51.8°, and 76.4°, corresponding to the (111), (200), and (220) planes of metallic nickel (PDF#04-0850)13. For the ZrO2-modified catalysts (NiZr@C-1, -2, -3), a weak but discernible peak at ~30.2° is observed, which can be indexed to the (101) plane of tetragonal ZrO2 (JCPDS 80-0965), confirming its presence in the composite41. The average Ni crystallite size, estimated from the Ni(111) reflection using the Debye–Scherrer equation, ranges from 6.2 to 7.6 nm across the series: Ni@C (6.7 ± 0.3 nm), NiZr@C-1 (6.4 ± 0.3 nm), NiZr@C-2 (6.2 ± 0.3 nm), and NiZr@C-3 (7.6 ± 0.4 nm). The comparable sizes indicate that the introduction of ZrO2 did not significantly alter the growth of Ni particles, suggesting that ZrO2 mainly acts as a promoter rather than affecting Ni dispersion. XRD patterns of the Ni@C and NiZr@C-X catalysts. XRD patterns of Ni@C and NiZr@C-X catalysts. Ni reflections at 44.5°, 51.8°, and 76.4° (PDF#04‑0850) are present in all samples. A weak peak at 30.2° (tetragonal ZrO2, JCPDS 80‑0965) is observed for ZrO2‑containing samples. Ni crystallite sizes were estimated via the Debye–Scherrer equation. Textural properties Nitrogen physisorption was employed to analyze the porosity of the catalysts, with key parameters summarized in Table 4. The incorporation of ZrO2 significantly enhances the textural properties. The specific surface area increases from 123 m² g⁻¹ for Ni@C to 285, 281, and 230 m² g⁻¹ for NiZr@C-1, -2, and -3, respectively. A similar trend is observed for the total pore volume. In contrast, the average pore diameter, derived from the adsorption branch, remains relatively constant at 3.1–3.2 nm for all catalysts, suggesting a preserved mesoporous framework. Table 4Textural parameters derived from N2 physisorptionCatalystSBETa (m2·g−1)Vporeb (cm3·g−1)Mesopore sizec (nm)Ni@C1230.073.2NiZr@C-12850.163.1NiZr@C-22810.163.1NiZr@C-32300.133.2aBET specific surface area.bTotal pore volume at P/P₀ = 0.99.cAverage pore diameter from BJH analysis of the adsorption branch. Textural parameters derived from N2 physisorption aBET specific surface area. bTotal pore volume at P/P₀ = 0.99. cAverage pore diameter from BJH analysis of the adsorption branch. The significant increase in specific surface area upon ZrO2 incorporation can be attributed to the spatial isolation effect of ZrO2 nanoparticles during carbonization. Dispersed ZrO2 particles physically separate Ni nanoparticles and prevent collapse of the carbon framework, thereby preserving more pore structures. Additionally, the partially reduced ZrO2 itself may contribute to the surface area41. The average pore diameter remains essentially unchanged (3.1–3.2 nm), indicating that the mesoporous structure is primarily governed by the carbon skeleton derived from epoxy resin carbonization, which is consistent across all catalysts. Morphology analysis (SEM) The morphological features of the catalysts were examined by SEM (Fig. 8). The Ni@C catalyst displays a dense bulk morphology with a surface covered by loosely aggregated, irregular particles (Fig. 8a, b). With the introduction of ZrO2, this loose surface texture becomes less prominent. The NiZr@C catalysts exhibit a more consolidated and uniformly wrinkled morphology (Fig. 8c–h), indicating a morphological evolution likely associated with the interaction between the NiO-ZrO2 precursor and the carbon matrix during pyrolysis. SEM images of the catalysts. a, b Ni@C; c, d NiZr@C-1; e, f NiZr@C-2; g, h NiZr@C-3. TEM/HAADF‑STEM/EDS analysis The morphology, crystal structure, and elemental distribution of Ni@C and NiZr@C‑2 were investigated using TEM, HAADF‑STEM, and EDS (Fig. 9). TEM images (Fig. 9b, e) show that Ni nanoparticles are well‑dispersed within the carbon matrix, exhibiting an elliptical morphology. The corresponding lattice fringes in HR‑TEM images (Fig. 9b1–b2, e1–e2) are indexed to the (111) and (200) planes of metallic Ni, with particle sizes of ~6.75 nm (Ni@C) and ~6.25 nm (NiZr@C‑2), consistent with XRD results. Microstructural and elemental characterization of Ni@C and NiZr@C‑2. a TEM image and particle‑size distribution for Ni@C. b1, b2 HR‑TEM images of Ni@C. c1–c4 HAADF‑STEM image and EDS elemental maps (Ni, C, O) of Ni@C. d TEM image and particle‑size distribution for NiZr@C‑2. e1, e2 HR‑TEM images of NiZr@C‑2. f1–f5 HAADF‑STEM image and EDS elemental maps (Ni, Zr, C, O) of NiZr@C‑2. a TEM image and particle‑size distribution for Ni@C. b1, b2 HR‑TEM images of Ni@C. c1–c4 HAADF‑STEM image and EDS elemental maps (Ni, C, O) of Ni@C. d TEM image and particle‑size distribution for NiZr@C‑2. e1, e2 HR‑TEM images of NiZr@C‑2. f1–f5 HAADF‑STEM image and EDS elemental maps (Ni, Zr, C, O) of NiZr@C‑2. HAADF‑STEM imaging coupled with EDS elemental mapping provides direct evidence of the spatial distribution. For NiZr@C‑2, the signals for Ni and Zr show clear overlap (Fig. 9f1, f3, f5), confirming the intimate association and uniform dispersion of ZrO2 species with the Ni nanoparticles throughout the support. XPS and ICP-OES analysis Further X-ray photoelectron spectroscopy (XPS) analysis was performed to investigate the surface chemical states of the active Ni species. The deconvolution of the Ni 2p3/2 region (Fig. 10a) revealed characteristic peaks at ~852.3, 852.2, 852.0 eV, and 852.2 eV for the Ni@C, NiZr@C-1, NiZr@C-2, and NiZr@C-3 catalysts, respectively. These peaks are ascribed to metallic Ni0. An additional feature in the 853.4–854.6 eV range, attributable to Ni2+ species, suggests partial surface oxidation during sample handling or measurement42. The metallic Ni0 species detected by XPS originate from the in situ reduction during the carbonization step. During pyrolysis of the epoxy resin under N2 at 500 °C, the generated carbon and reducing gases (CO, H2) reduce NiO to metallic Ni, as confirmed by the XRD patterns (Fig. 7). Although minor surface oxidation may occur upon brief exposure to air during sample transfer, the dominant Ni⁰ signal indicates that the metallic state is well preserved on the catalyst surface. XPS spectra of the catalysts. a Ni 2p; b Zr 3d; c O 1s. The corresponding Zr 3d spectra (Fig. 10b) were fitted with two components, assigned to Zr³⁺ and Zr4+ oxidation states. The presence of Zr3+ confirms the partial reduction of Zr4+ during catalyst preparation. The binding energies for the Zr3+ 3d3/2 and Zr4+ 3d3/2 states are detailed in the manuscript43. Notably, the introduction of ZrO2 induced a shift in the binding energies for Ni0, Zr3+, and Zr4+, with an initial decrease followed by an increase. The NiZr@C-2 catalyst (Zr/Ni = 2/10) exhibited the lowest binding energies for all three species. Quantitative analysis of the Zr 3d3/2 peaks yielded surface Zr³⁺/Zr4+ molar ratios of 1.92, 2.64, and 1.73 for NiZr@C-1, NiZr@C-2, and NiZr@C-3, respectively. The highest ratio in NiZr@C-2 indicates its most pronounced surface reduction degree. The O 1s spectra (Fig. 10c) were resolved into three contributions: lattice oxygen in ZrO2 (~529.0–529.2 eV), low-coordination oxygen adjacent to vacancies (~530.8–531.1 eV), and surface-adsorbed water (~532.5–532.8 eV)44,45. As summarized in Table 5, the relative concentration of the low-coordination oxygen species (indicative of oxygen vacancies, OVs) and the NiZr@C-2 catalyst possessed the highest proportion (39.58%) of these species, demonstrating it hosts the most abundant surface oxygen vacancies among the series42. Table 5Peak area percentages of the fitted O 1s orbital peaks for NiZr@C-X catalystsCatalystProportion of the peak area of the O 1 s orbitalOLOVsOH2ONiZr@C-110.53%29.22%60.25%NiZr@C-216.05%39.58%44.37%NiZr@C-318.00%30.26%51.74%Peak area percentages of the three O1 s components: lattice oxygen (OL), oxygen vacancies (OVs, corresponding to low-coordination oxygen adjacent to vacancies), and surface‑adsorbed water (H2O). The sum of the three components is 100% for each catalyst. Peak area percentages of the fitted O 1s orbital peaks for NiZr@C-X catalysts Peak area percentages of the three O1 s components: lattice oxygen (OL), oxygen vacancies (OVs, corresponding to low-coordination oxygen adjacent to vacancies), and surface‑adsorbed water (H2O). The sum of the three components is 100% for each catalyst. The bulk and surface compositions of the catalysts were analyzed using X-ray fluorescence (ICP-OES) and X-ray photoelectron spectroscopy (XPS), respectively, with results summarized in Table 6. The bulk ZrO2/NiO ratios determined by ICP-OES correspond well to the nominal formulations. A key finding is that the surface Zr/Ni molar ratios obtained from XPS are substantially higher than the bulk values from ICP-OES across all catalysts. This evidence strongly suggests the surface enrichment of Zr species, attributable to the migration of ZrO2 toward the Ni active sites during thermal treatment, thereby fostering a pronounced metal–support interaction44. Table 6Bulk and surface composition analysis of the NiZr@C catalystsCatalystAtomic ratio of Ni, Zr, O, C(%)aSurface Zr/Ni atomic ratioAtomic ratio of Ni, Zr(wt.%)bBulk Zr/Ni molar ratiobNiZrOCNiZrNiZr@C-12.670.4813.4183.430.1829.358.280.11NiZr@C-22.320.7114.1482.830.3129.0510.390.23NiZr@C-32.031.1614.9381.880.5728.7615.640.35a The atomic ratio of Ni, Zr, O, C were showed by XPS results.b The atomic ratio of Ni, Zr and the bulk Ni/Zr molar ratio were showed and calculated by ICP-OES results. Bulk and surface composition analysis of the NiZr@C catalysts a The atomic ratio of Ni, Zr, O, C were showed by XPS results. b The atomic ratio of Ni, Zr and the bulk Ni/Zr molar ratio were showed and calculated by ICP-OES results. Elucidating the structure–activity relationship The hydrogen activation behavior of the catalysts was characterized by H2-temperature-programmed desorption (H2-TPD). As shown in Fig. 11, all profiles exhibit two overlapping desorption features centered at ~85 °C (Hα) and 125 °C (Hβ). The low-temperature Hα peak is ascribed to hydrogen species associated with surface oxygen vacancies, while the high-temperature Hβ peak corresponds to hydrogen chemisorbed on metallic Ni sites41,45–47. H2-TPD profiles of the Ni@C and NiZr@C-2 catalysts. Both Ni@C and NiZr@C‑2 exhibit two hydrogen desorption features: Hα (~85 °C) assigned to hydrogen associated with surface oxygen vacancies, and Hβ (~125 °C) assigned to hydrogen chemisorbed on metallic Ni sites. Quantitative analysis of the desorption profiles (Table 7) reveals a distinct evolution with increasing ZrO2 content. Both the total hydrogen desorption capacity and the intensity of the Hα peak show a marked increase, whereas the Hβ peak intensity progressively declines. A pivotal change is observed at a Zr/Ni ratio of 2/10 (NiZr@C-2), where the Hα contribution rises dramatically. Further increasing the ratio to 3/10 results in a saturation of the Hα signal, accompanied by a sharp decrease in Hβ desorption. Table 7Quantitative analysis of H2-TPD profiles for the synthesized catalystsCatalystH2 desorption peak fitting valuesTotal H2 (μmol/g)Hα(μmol/g)Hβ(μmol/g)Ni@C57.0412.4944.55NiZr@C-147.5315.3732.16NiZr@C-2106.1288.5227.60NiZr@C-3106.5888.6317.95Hα hydrogen on oxygen vacancies, Hβ hydrogen on metallic Ni. Quantitative analysis of H2-TPD profiles for the synthesized catalysts Hα hydrogen on oxygen vacancies, Hβ hydrogen on metallic Ni. Integrating these findings with the surface and bulk compositional data (XPS/ICP-OES), we propose the following mechanistic insight. The thermal treatment induces the migration of ZrO2 to the Ni interface. During this process, the concomitant reduction of surface ZrO2 creates a strong Ni–ZrO2 interaction and generates abundant oxygen vacancies, while partially covering the Ni surface46,48. This structural evolution facilitates the spillover of hydrogen atoms from the metallic Ni sites (Hβ origin) to the neighboring oxygen vacancies (Hα origin), rationalizing the observed inverse trend in the two desorption features40. The intrinsic activities of the Ni@C and NiZr@C-2 catalysts for C=C hydrogenation and C=O hydrogenolysis were compared by calculating their turnover frequencies (TOF). The TOF values were derived from hydrogenation kinetic data (60 °C-100 °C and 180–210 °C for C=C and C=O reactions, respectively) and normalized using the quantity of surface Ni sites estimated from Hβ-TPD measurements. As presented in Fig. 12, the NiZr@C-2 catalyst exhibits significantly higher TOF values than Ni@C for both reactions across the studied conditions, demonstrating its superior intrinsic catalytic activity. Turnover frequencies of Ni@C and NiZr@C‑2 for C=C hydrogenation and C=O hydrogenolysis. Turnover frequency (TOF) for a C=C hydrogenation and b C=O hydrogenolysis over the Ni@C and NiZr@C-2 catalysts. The TOF was calculated using the formula: TOF = NumberofconvertedreactantmoleculesoverthereactiontimeNumberofNiatomsonthecatalystsurface×reactiontime(min−1), Number of Ni atoms on the catalyst surface = catalyst mass × Hβ desorption amount × 2. Mechanistic interpretation The collective findings confirm that introducing ZrO2 into the Ni@C catalyst substantially improves its activity for both C=C hydrogenation and C=O hydrogenolysis. Among the series, NiZr@C-2 (Zr/Ni = 2/10) demonstrates optimal performance in the hydrogenation of maleic anhydride (MA) to γ-butyrolactone (GBL). Theoretical studies indicate that MA preferentially adsorbs parallel to the Ni surface, where sequential hydrogenation of the C=C bond yields succinic anhydride (SAH)18,40,49. The presence of surface oxygen vacancies promotes MA adsorption at neighboring Ni sites. These vacancies, formed at the Niδ-–Ovs–Zr3+ interface during the development of a strong metal–support interaction, also facilitate H2 adsorption and activation30,43,50. Partial electron transfer from the vacancies to Ni creates a negatively charged surface (Niδ-), which favors the heterolytic cleavage of H251–54. The resulting hydrogen atoms migrate to and stabilize at the oxygen vacancies, generating a reservoir of active hydrogen species that explains the enhanced C=C hydrogenation activity upon ZrO2 incorporation. The role of oxygen vacancies in C=O bond activation was probed using in situ FTIR with cyclohexanone (Fig. 13). The C=O stretching frequency for the adsorbed species shifts from 1705 cm⁻¹ on Ni@C (identical to free cyclohexanone) to lower wavenumbers on the NiZr@C catalysts (1702, 1699, and 1701 cm⁻¹ for NiZr@C-1, -2, and -3, respectively). This red-shift signifies bond weakening and activation upon interaction with the catalyst surface30,55,56. NiZr@C-2 shows the most pronounced shift which is consistent with its highest C=O activation capability. The vacancies effectively adsorb and polarize the C=O group, which then cooperates with surface hydrogen species to drive hydrogenolysis—consistent with our prior work40. Specifically, SAH adsorbs parallel to the surface with its carbonyl oxygen interacting directly with a vacancy40,43,46. This configuration significantly elongates the C=O bond and orients the molecule such that both the carbon and oxygen atoms become more susceptible to attack by active hydrogen, significantly lowering the energy barrier for the dehydroxylation step that forms GBL. In situ FTIR spectra of cyclohexanone adsorption on Ni@C and NiZr@C-2 catalysts. Spectra of cyclohexanone adsorbed on Ni@C (no shift, 1705 cm⁻¹) and NiZr@C‑2. The red‑shift signifies bond weakening, consistent with enhanced C=O hydrogenolysis activity. In summary, thermal treatment drives ZrO2 migration to the Ni surface, where subsequent reduction creates oxygen vacancies at the Ni–ZrO2 interface. These vacancies serve a dual function: (i) they modulate the electronic state of Ni, promoting H2 heterolysis and the generation of mobile active hydrogen species, and (ii) they provide specific sites for the adsorption and activation of reactant molecules (MA/SAH). The synergy between these activated reactants and the abundant hydrogen species on NiZr@C-2 culminates in its exceptional activity for both C=C and C=O bond transformations. Conclusions The introduction of ZrO2 into the Ni@C catalyst framework significantly improves its catalytic performance for both C=C hydrogenation and C=O hydrogenolysis. Kinetic analysis under low-conversion conditions revealed that NiZr@C-2 exhibits lower apparent activation energies for both C=C hydrogenation (21.1 kJ mol−1) and C=O hydrogenolysis (95.8 kJ mol−1) compared to Ni@C (39.6 and 103.6 kJ mol−1, respectively), confirming the intrinsically enhanced activity conferred by ZrO2 incorporation. It is worth noting that the H2 pressure employed in this study (4.5 MPa) is within the typical range reported for non-noble metal catalysts in MA hydrogenation (4–8 MPa). A combination of catalyst characterization and mechanistic studies reveals that thermal treatment induces ZrO2 migration to form a strong interaction with the active Ni species. During reduction, due to the partial reduction of ZrO2, abundant oxygen vacancies are generated which create an electron-rich Ni surface, facilitating the heterolytic cleavage of H2. The resultant active hydrogen species migrate to these vacancies, creating a highly active hydrogen reservoir that boosts hydrogenation efficiency. Furthermore, the oxygen vacancies at the Ni–O–Zr interface promote the adsorption and activation of the C=C bond and facilitate the production of SAH. The synergistic effect between these oxygen vacancies and the abundant active hydrogen species enhances the adsorption,activation and hydrogenlysis of C=O bond, and ultimately leads to improved GBL selectivity. Moreover, the exceptional stability of NiZr@C-2 over 13 cycles minimizes catalyst replacement frequency, and the high reaction temperature (210 °C) enables rapid conversion (2 h to 100% MA conversion), maximizing space-time yield. These combined factors contribute to the overall economic viability of this catalytic system for potential industrial application.This work provides fundamental insights for the rational design of high-performance catalysts for MA hydrogenation. Methods Materials All chemicals were used as received. Anhydrous sodium carbonate (Na2CO3, ≥99.8%), nickel(II) chloride hexahydrate (NiCl2·6H2O, ≥98%), and absolute ethanol (EtOH, ≥99.7%) were purchased from Tianjin Dama Chemical Reagent Co., Ltd. 1,4-Dioxane (≥99%) was supplied by Shandong Keyuan Biochemical Co., Ltd. Maleic anhydride (MA, ≥99.5%) was obtained from Sinopharm Chemical Reagent Co., Ltd. Epoxy resin was sourced from Henan Borun Casting Materials Co., Ltd. Zirconyl nitrate hydrate (ZrO(NO3)2, ≥99.5%) was acquired from Shanghai Aladdin Biochemical Technology Co., Ltd. Catalyst synthesis The synthesis procedure is illustrated in Scheme 2. An aqueous solution containing 2.12 M Ni2+ and 0.5 M Zr4+ (Solution A) was prepared from NiCl2·6H2O and ZrO(NO3)2·2H2O. A separate 1.7 M Na2CO3 solution (Solution B) was used as the precipitant. Both solutions were added dropwise simultaneously into a stirred round‑bottom flask at 45 °C, and the pH was maintained at 7–8. The resulting suspension was aged for 2 h, then filtered, washed thoroughly with deionized water, and dried at 333 K for 12 h. The solid was calcined in static air at 200 °C for 2 h to obtain a NiO–ZrO2 mixed oxide precursor. Schematic of the synthesis procedure for the NiZr@C catalysts.A Ni–Zr carbonate precursor was precipitated at 45 °C and pH 7–8, then calcined to NiO–ZrO2. The mixed oxide was ground with epoxy resin (1:2.15) and carbonized at 500 °C under N2, yielding NiZr@C. The Zr/Ni molar ratio was varied to obtain NiZr@C‑1, ‑2, ‑3, and a Zr‑free Ni@C reference. The precursor was mixed mechanically with epoxy resin at a mass ratio of 1:2.15. The mixture was placed in a tubular furnace, dried at 130 °C for 2 h under flowing N2 (40 mL min−1), and subsequently carbonized at 500 °C for 2 h under the same N2 atmosphere to yield the final NiZr@C catalysts. Samples with nominal Zr/Ni molar ratios of 0.1, 0.2, and 0.3 are denoted as NiZr@C‑1, NiZr@C‑2, and NiZr@C‑3, respectively. A reference Ni@C catalyst was prepared identically without the addition of the zirconium source. Characterization and catalytic evaluation Characterization X‑ray diffraction (XRD) patterns were recorded on a Bruker D2 PHASER diffractometer using Cu Kα radiation (λ = 1.5418 Å) with a scanning range of 10–80°(2θ). Textural properties were determined from N2 adsorption‑desorption isotherms at 77 K using a Micromeritics ASAP 2020 analyzer. Morphology was examined by scanning electron microscopy (SEM, TESCAN MIRA LMS). Microstructural and compositional analyses were performed using transmission electron microscopy (TEM), high‑angle annular dark‑field scanning TEM (HAADF‑STEM), and energy‑dispersive X‑ray spectroscopy (EDS) on an FEI Talos F200X system. X‑ray photoelectron spectroscopy (XPS) measurements were conducted on a Thermo Scientific ESCALAB 250Xi spectrometer with an Al Kα source. Bulk elemental composition was analyzed by inductively coupled plasma optical emission spectrometry (ICP-OES, Thermo Fisher iCAP PRO). 60 mg of the catalyst was digested in aqua regia at 80 °C, and the resulting metal solution was diluted 10:1 with deionized water. H2-temperature‑programmed desorption (H2‑TPD) was carried out on a Micromeritics Autochem II 2920 instrument equipped with a thermal conductivity detector. Prior to analysis, the sample (ca. 100 mg) was reduced in a 10% H2/Ar flow (50 mL·min⁻¹) at 50 °C for 1 h, purged with Ar, and then heated to 800 °C at 10 °C min−¹ under Ar flow. In situ Fourier‑transform infrared (FTIR) spectra were collected on a Nicolet iS50 spectrometer using cyclohexanone as a probe molecule to study carbonyl adsorption. Notably, the fresh catalysts already contain metallic Ni0 as a result of the carbon-thermal reduction that occurs during the carbonization step (section “Catalyst synthesis”). Fresh (as-synthesized, post-carbonization) catalysts were used for each measurements. Catalytic testing The hydrogenation of maleic anhydride (MA) was performed in a 100‑mL stainless‑steel batch reactor. In a typical run, the reactor was charged with 40 mL of 1,4‑dioxane, 2.45 g of MA, and 0.5 g of catalyst. Air was removed by purging with N2 (five cycles), followed by H2 (ten cycles). The reactor was pressurized to 3.8 MPa with H2 and then heated to 210 °C, resulting in a final pressure of ~4.5 MPa. Stirring was then initiated at 450 rpm to start the reaction. Liquid products were analyzed by gas chromatography (Shimadzu 2010) using an HP‑5 column and a flame ionization detector (FID); quantification was performed with an internal standard method31. Conversion of MA and selectivity to GBL were calculated as: Where nMA0 and nMA are the initial and final molar amounts of MA, respectively, and nGBL is the molar amount of GBL produced. All catalytic measurements were repeated at least three times, with the reported values representing the average. The estimated analytical precision is ~±0.5% for both conversion and selectivity, based on repeated injections of standard samples. Supplementary information Transparent Peer Review file support information Acknowledgement Declaration of generative AI and AI-assisted technologies in the manuscript preparation process. During the preparation of this work, the author(s) did not use any generative AI or AI-assisted technologies. The authors take full responsibility for the content and originality of the manuscript. Author contributions Shuowen Guo: Conceptualization, methodology, validation, investigation, data curation, writing—original draft, and visualization. Yanyang Yao: Methodology and resources. Wei Wei: Writing—review & editing. Jiannan Lv and Dongjie Zhang: Methodology. Weixun Wang: Resources. Yin Zhang and Haitao Li: Supervision, funding acquisition, writing—review & editing, and project administration. Peer review Peer review information Communications Chemistry thanks Yingquan Wu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding This work was supported by the National Natural Science Foundation of China [grant numbers: 22278252 and 22075167] and Joint Fund of the Shanxi Provincial Basic Research Program [grant number: 202403011241003]. Data availability The data supporting the findings of this study consist of catalytic performance metrics (MA conversion, GBL and SAH selectivity) and associated characterization results. All data were available within the article and its Supplementary Information, or from the corresponding author upon reasonable request. Requests should be directed to the corresponding author Pro. Zhang, who will respond within a reasonable timeframe. No datasets with accession codes or DOIs were generated or analyzed in this work. Competing interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Yin Zhang, Email: sxuzhy@sxu.edu.cn. Haitao Li, Email: htli@sxu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s42004-026-02111-6. References Associated Data Supplementary Materials Transparent Peer Review file support information Data Availability Statement The data supporting the findings of this study consist of catalytic performance metrics (MA conversion, GBL and SAH selectivity) and associated characterization results. All data were available within the article and its Supplementary Information, or from the corresponding author upon reasonable request. Requests should be directed to the corresponding author Pro. Zhang, who will respond within a reasonable timeframe. No datasets with accession codes or DOIs were generated or analyzed in this work.