Highly efficient hydrodeoxygenation of lignin-derived vanillin over CoN/C catalysts with isopropanol

As the richest renewable aromatic biomass in the world, lignin and its catalytic depolymerization products exhibit significant application potential and strategic value in the areas of energy, chemical industry, materials and environmental science. In this work, three catalysts (NiN/C, CoN/C, and CoNiN/C) were prepared using one-pot impregnation-pyrolysis and tested for the catalytic hydrodeoxygenation (HDO) of vanillin (VAN) to produce 2-methoxy-4-methylphenol (MMP), a valuable chemical intermediate for pharmaceutical or fuel applications. Among the three catalysts, CoN/C exhibited excellent catalytic performance and remarkable stability. Experimental results demonstrated that 30 mg of CoN/C achieved highly efficient conversion of VAN under the optimal reaction conditions (240 °C, 1.5 MPa H 2, 15 mL isopropanol, 4 h), affording a VAN conversion of 99.94% and an MMP selectivity of 96.15%. Characterization results further confirm that the CoN/C catalyst exhibited a large surface area, developed mesoporosity, and rich Lewis acid sites. These features work together to promote its catalytic activity and benefit the HDO of VAN. Further density functional theory (DFT) calculations reveal that the CoN/C catalyst exhibits a moderate adsorption energy of −1.04 eV. This work provides a meaningful route to simple, eco-friendly, and highly active metal catalysts, and offers a reference for lignin conversion into high-value chemicals.

As global fossil resources are being depleted, the development of renewable biomass to replace traditional fossil fuels has become a key strategic direction for advancing sustainable green energy and promoting energy structure transformation (Gaurav et al., 2017; Peng et al., 2024). Despite rapid growth in renewables like wind and solar, issues such as intermittency, high storage costs, and geographic limits prevent them from fully replacing fossil fuels (Boettcher, 2024). Lignin, a major biomass component, is a promising clean energy candidate because of its abundance and renewability. It has thus become a strategic focus in global energy transformation and an ideal raw material (Cheng et al., 2024; Guan et al., 2019; Garedew et al., 2021; Zhang et al., 2025). Vanillin is a core platform compound in lignin depolymerization, can be selectively transformed via HDO to fine chemical intermediates such as MMP. This compound holds broad application prospects in the pharmaceutical, flavor and fragrance, and polymer industries (Li et al., 2021; Yu et al., 2024). In the vanillin HDO to MMP experiment, researchers are primarily concerned with VAN conversion and MMP selectivity. These two parameters serve as key indicators of catalyst activity. Currently, noble metals are commonly used in VAN HDO because their unique d-orbitals enable them to form active hydrogen with H 2 or other hydrogen donors (Fan et al., 2018; Guo et al., 2025; Santos et al., 2018). Consequently, noble metal catalysts exhibit excellent hydrodeoxygenation performance for vanillin and high MMP selectivity. Despite their outstanding catalytic activity, they suffer from high cost, limited natural abundance, and poor stability, which severely restrict their widespread industrial implementation (Li Z. et al., 2022). Thus, developing efficient, environmentally friendly, and low-cost catalysts for lignin hydrodeoxygenation holds significant academic and industrial importance.

Recently, many studies on VAN HDO have focused on developing sustainable biomass-derived carbon-supported catalysts. For example, Duan et al. synthesized a novel Ni-Co-P/HAP catalyst by impregnation-chemical reduction and used for in-situ VAN HDO to MMP. Under optimized reaction conditions, VAN conversion reached 97.86% with 93.97% selectivity for MMP (Duan et al., 2021). Kumar et al. developed a highly active Pd-modified Bi 2 MoO 6 nanosheet catalyst capable of achieving efficient vanillin conversion under ambient temperature and mild hydrogen pressure, thereby resolving the technical challenge associated with conventional vanillin HDO processes that demand elevated temperature and high pressure (Kumar et al., 2023). The study revealed that NaBH 4 treatment during the deposition of Pd nanoparticles could enhance the surface oxygen vacancies of Bi 2 MoO 6 nanosheets, Consequently, the reactant is preferentially adsorbed onto the catalyst surface via the aldehyde group, offering an efficient catalytic system for vanillin HDO under mild conditions. Liu et al. introduced a phenylacetone-formaldehyde condensation product (RF) for the synthesis of a Pd/Nb 2 O 5 -T-RF catalyst rich in oxygen vacancies. Further performance tests indicated that electronic oxygen deficiency enhance vanillin/vanillyl alcohol adsorption/activation and hydrogen spillover, thereby boosting HDO performance (Liu et al., 2025). However, most catalysts currently employed for vanillin HDO are noble metal-loaded biomass-derived carbon materials or bimetallic catalysts, which increase costs and limit large-scale industrial application (Guan et al., 2025; Liu et al., 2016; Shu et al., 2019). Therefore, it is necessary to develop catalysts featuring simple and green preparation processes, high conversion efficiency, and high selectivity for hydrodeoxygenation reactions. Transition metal (Co, Ni, Fe)-based catalysts have attracted increasing research interest for replacing noble metals thanks to their low cost, abundant reserves, adjustable catalytic performance (Cheah et al., 2021; Huynh et al., 2016). In recent years, several non-noble metal catalysts have been explored for vanillin HDO. Tian et al. reported a Cu/N@ALC catalyst derived from alkaline lignin (AL) and dicyandiamide (DCA), in which the synergistic effect of surface Cu 0 and Cu + species enabled near-quantitative conversion of VAN to MMP under optimized HDO conditions (Tian et al., 2026). Separately, Xue et al. developed Co@C catalysts through a one-pot hydrothermal and carbonization route; the high VAN conversion and MMP selectivity achieved under mild, H 2 -free conditions were closely associated with their specific surface area (154.45 m 2 /g) and acid site density (4.67 mmol/g) (Xue et al., 2023). Cobalt-based catalysts are particularly promising for vanillin HDO due to their high C-O bond scission selectivity and excellent hydrogenation activity (Chen et al., 2024; Wu R. et al., 2024; Xue et al., 2026; Zhai et al., 2019). For instance, Song et al. developed a cobalt-nitrogen-doped carbon (Co-NC) catalyst capable of achieving redox-neutral depolymerization of lignin without the use of external oxidants or reducing agents, enabling efficient C-O bond cleavage via a cascade dehydrogenation-hydrogenolysis pathway (Song et al., 2024). Zeng et al. constructed a dual-Ni (bi-Ni) catalyst featuring both single-atom and nanoparticle sites, which achieved complete conversion of lignin model compounds at 200 °C without external H 2, leveraging the synergistic effect between Ni single-atom and nanoparticle sites (Zeng et al., 2025). Collectively, these studies underscore the great potential of non-noble metal catalysts for the sustainable valorization of lignin under mild conditions, which is in line with the design strategy of the present work. Nevertheless, conventional supported cobalt-based catalysts generally suffer from issues such as metal particle agglomeration, poor dispersion, and weak metal–support interactions. Moreover, the complexity of their preparation processes hinders industrial scale-up, resulting in insufficient catalytic activity and stability (Hong et al., 2020; Li L. et al., 2022). The one-pot impregnation-pyrolysis method offers significant advantages, including simple operation, low cost, and facile scalability. This method avoids complex multi-step synthesis procedures while ensuring a uniform distribution of metal species within the nitrogen-doped carbon matrix. Using kraft lignin as the carbon precursor, urea as the nitrogen source, and the corresponding metal nitrates as metal precursors, a series of nitrogen-doped carbon-supported transition metal catalysts were prepared via the one-pot impregnation-pyrolysis method. After calcination at 800 °C under a nitrogen atmosphere, a nitrogen-doped porous carbon matrix was formed, in which Co, Ni, or CoNi alloy nanoparticles were uniformly distributed, along with metal-nitrogen single-atom active sites. This unique architecture endows the catalysts with a high specific surface area, excellent electrical conductivity, and abundant catalytically active sites, making them highly promising non-noble metal catalysts for various applications (Zhao et al., 2025; Yu et al., 2023).

In this work, three N-doped carbon-supported catalysts, namely CoN/C, NiN/C, and CoNiN/C, were synthesized and their catalytic performance was evaluated in the HDO of VAN to MMP. The effect of various catalysts on VAN HDO was studied, and VAN conversion and MMP selectivity were analyzed by gas chromatography (GC). Orthogonal experiments were designed based on the selection of appropriate catalysts to further explore the effects of catalyst loading, reaction solvent, gas atmosphere, reaction temperature, and reaction time on the VAN HDO reaction. The results demonstrate that the CoN/C shows outstanding catalytic activity. A series of characterization data reveal that the CoN/C catalyst possesses fine and well-dispersed grains, abundant carbon defect sites, and rich Lewis acid centers, with no obvious metal agglomeration. These multiple structural advantages synergistically promote the CoN/C catalyst to achieve good catalytic activity in the VAN HDO reaction. A composite hydrogenation system was established in this work, featuring molecular H 2 as the main hydrogen source for the HDO reaction and isopropanol as both an environmentally benign solvent and a sacrificial hydrogen donor (Li et al., 2024; Sun et al., 2021). Such a dual-hydrogen supply strategy was intended to realize high VAN conversion and high MMP selectivity under a mild H 2 pressure of 1.5 MPa. Further DFT calculations reveal that the CoN/C catalyst exhibits a moderate adsorption energy of −1.04 eV. The CoN/C catalyst showed good activity, achieving 99.94% VAN conversion and 96.15% MMP selectivity in isopropanol with 30 mg CoN/C at 240 °C under 1.5 MPa H 2 for 4 h. This work presents a new route to HDO catalysts with simple preparation, high selectivity, and high conversion.

Catalyst characterization

The X-ray diffraction (XRD) results of the catalysts’ phase structure, crystallite size, and crystallinity are shown in Figure 1c, the CoN/C shows peaks at 44.1°and 51.4°, corresponding to Co (111) and Co (200) (JCPDS No. 15-0806), confirming the presence of Co species. Upon introduction of Ni into the CoN/C catalyst, the XRD pattern shows notable changes. The diffraction peak corresponding to Co (111) becomes sharper, and an additional diffraction peak emerges at 2θ = 44.3°, which is assigned to the Ni (111) crystal plane (Wang et al., 2021; Wu X. et al., 2024; Xu et al., 2024; Xu et al., 2012). For the NiN/C catalyst, three distinct characteristic peaks are clearly observable in the XRD pattern at 2θ = 44.3°, 51.7°, and 76.1°, all exhibiting sharp peak shapes. This indicates that the Ni grains possess high crystallinity and large crystallite size, which is unfavorable for the HDO of VAN. In summary, CoN/C shows broader and weaker XRD peaks, indicating smaller Co crystallites and better dispersion, which favors VAN HDO.

Figure 1d shows the Raman spectra of the three catalysts. All three spectra exhibit characteristic carbon peaks at ∼1,340 cm -1 and ∼1,580 cm -1, attributed to the D band and the G band, respectively. The D band corresponds to defective carbon lattices (sp 3 carbon), and the G band represents sp 2 carbon in-plane vibration, indicating graphitization degree (Sadezky et al., 2005). Thus, the I D /I G ratio can be used to quantify the defect level of the carbon support. A higher I/Iɢ value indicates more carbon defect sites and an increased level of amorphous character. The I/Iɢ ratios of the three catalysts are 0.8423 for CoN/C, 0.8417 for NiN/C, and 0.8418 for CoNiN/C. The three catalysts show very similar I D /I G ratios (∼0.842), indicating that the nitrogen-doped carbon supports possess comparable degrees of graphitization and defect density (Chen et al., 2021). The abundant carbon defects serve a dual role: they provide efficient anchoring sites for metal nanoparticles, effectively suppressing particle agglomeration and thereby yielding smaller crystallite sizes which was consistent with the XRD results and they also act as active centers that enhance the activity and selectivity of the hydrodeoxygenation reaction.

In order to assess their adsorption capability as well as the diffusional accessibility of reactants, nitrogen physisorption (BET) analysis was carried out in this work in Figures 1a,b. Reveals that all catalysts show type IV isotherms with clear H4 hysteresis loops at P/P 0 = 0.8-1.0. This suggests that the catalysts predominantly feature mesoporous structures, accompanied by a certain amount of micropores or macropores. Regarding the adsorption capacity, CoNiN/C exhibited the highest adsorption volume, which is consistent with its largest total pore volume (0.27 cm 3 /g, Table 1). Despite its relatively lower BET specific surface area, the abundant meso-/macropores in CoNiN/C significantly enhanced its adsorption capacity at high relative pressures through capillary condensation. Among the three catalysts, CoN/C displayed the highest BET specific surface area of 444.63 m 2 /g, with a well-developed microporous structure. The average pore diameter of 2.14 nm and total pore volume of 0.24 cm 3 /g indicate a hierarchical porous architecture combining micropores and small mesopores. This structural configuration provides abundant anchoring sites for cobalt nanoparticles and facilitates efficient mass transfer, constituting a critical structural basis for the excellent catalytic performance of CoN/C in the HDO of VAN. It is worth noting that, although NiN/C exhibits a BET surface area comparable to that of CoN/C (424.95 vs. 444.63 m 2 /g), its catalytic performance is substantially inferior. This disparity can be rationalized by multiple characterization results. Specifically, XRD reveals larger Ni crystallites with poorer dispersion in NiN/C. XPS confirms that Ni predominantly exists as Ni 2+ and Ni 3+ rather than metallic Ni 0. And NH 3 -TPD demonstrates significantly fewer strong acid sites on NiN/C. These combined factors, rather than surface area alone, account for the superior catalytic activity of CoN/C. In summary, the CoN/C combines high surface area, high pore volume, and suitable mesopore size. This structural configuration maximizes the exposure of Co active centers and optimizes mass transport, constituting the key structural basis for its superior catalytic performance in the HDO of VAN.

All catalysts were characterized by Scanning electron microscopy (SEM) for surface morphology, pore structure, and particle morphology. Presented in Figure 2 are the SEM images of CoN/C (a), NiN/C (b), and CoNiN/C (c) (scale bar: 200 nm). The CoN/C catalyst exhibits a uniform, branched, interconnected porous network structure featuring well-distributed mesoporous channels, with Co metal particles dispersed on the carbon framework surface. In contrast, the NiN/C catalyst displays a distinct sheet-like structure, characterized by a more compact carbon framework, larger Ni metal particles, and pronounced agglomeration. The CoNiN/C catalyst forms a needle-like clustered nanoarray structure; although its pore diameter is larger, some active sites are blocked by the carbon framework, reducing vanillin HDO efficiency.

For a more in-depth exploration of the microcrystalline structure and internal morphology of the CoN/C catalyst, transmission electron microscopy (TEM) was conducted. The lattice fringe distance observed in Figure 3a measures 0.215 nm, matching the interplanar spacing of the (111) plane of metallic Co. This observation directly confirms the successful formation of metal active centers in the CoN/C, which agrees well with the XRD results (Li et al., 2024; Sun et al., 2021). Figure 3c displays the particle size distribution of the CoN/C. Statistical analysis of particle sizes followed by Gaussian fitting reveals that the Co nanoparticles have an average size of 11.44 nm. Compared with the bimetallic CoNiN/C catalyst, the CoN/C catalyst exhibits smaller metal particle sizes and a more concentrated particle size distribution, with no appearance of abnormally large particles. This indicates the high dispersion and uniformity of the Co nanoparticles, which is highly favorable for the vanillin hydrodeoxygenation (HDO) reaction. The elemental distribution maps of the CoN/C catalyst are shown in Figures 3b,d–f. The results show that Co element is uniformly dispersed in the form of nanoparticles on the carbon support, without obvious agglomeration or localized enrichment. This further corroborates the high dispersion of the Co nanoparticles, affording widespread exposed active centers for the VAN HDO reaction. Furthermore, the N and C elements exhibit substantial overlap and are uniformly distributed around the Co nanoparticles, with all three forming a Co-N-C coordination structure. This coordination structure boosts metal-support interaction and catalyst stability.

For comparison, TEM characterization was also performed on the CoNiN/C catalyst (Supplementary Figure S1, Supplementary Material). As shown in Supplementary Figure S1, CoNiN/C exhibits larger metal particles with a broader size distribution and more pronounced agglomeration compared to CoN/C. TEM imaging of NiN/C was not carried out in this study. Nevertheless, the much sharper and more intense diffraction peaks observed in the XRD pattern of NiN/C (Figure 1c) indicate the presence of larger Ni crystallites, and the SEM image (Figure 2b) reveals severe particle agglomeration on the catalyst surface. These combined observations strongly suggest that NiN/C suffers from poorer metal dispersion relative to CoN/C, which reasonably accounts for its significantly lower catalytic activity in VAN HDO.

X-ray photoelectron spectroscopy was employed to characterize the elemental composition and chemical states of the catalyst. Figure 4a shows that the three catalysts all exhibit peaks at approximately 285.0, 400.08, and 532.08 eV, which are attributed to the elements C, N, and O. Meanwhile, the CoN/C and NiN/C catalysts display peaks at approximately 780.08 and 854.08 eV, assigned to Co and Ni, respectively, corroborating the presence of Co and Ni elements (Guo et al., 2022). The C 1s spectra in Figure 4b reveal three carbon species in the catalysts, with characteristic peaks corresponding to C–C (284.80 eV), C=N (285.78 eV), and C=O (289.28 eV), Notably, the CoN/C catalyst exhibits a higher relative content of C=N bonds, indicating a higher degree of carbon doping, which provides a favorable structural basis for active sites (Jo et al., 2019). Figure 4c displays seven peaks, revealing that the CoN/C catalyst contains a higher proportion of metallic Co 0 and Co-N coordination species. The abundant Co 0 active centers effectively promote the hydrodeoxygenation of vanillin, while the formation of Co–N bonds strengthen the metal–support interaction. After introducing Ni into the catalyst, the peak area corresponding to Co 0 decreases significantly, indicating that the presence of Ni influences the Co 0 species (Kang et al., 2021). As shown in the Ni 2p spectrum in Figure 4d, the Ni in the NiN/C and CoNiN/C catalysts predominantly exists in the Ni 2+ and Ni 3+ oxidation states, which results in relatively weak intrinsic hydrodeoxygenation activity. Furthermore, electron transfer between the bimetallic components alters the surface electronic structure of Co, rendering the catalyst unfavorable for the HDO of VAN. Moreover, Figures 4e,f displays the N 1s and O 1s spectrum of the catalyst. The spectrum exhibits four distinct binding energy peaks at 398.38, 400.98, 403.28, and 406.48 eV, correspond to pyridinic N, pyrrolic N, graphitic N, and oxidized N, respectively. According to the XPS results, the CoN/C catalyst possesses a high content of metallic Co active sites, an optimal nitrogen doping configuration, and abundant surface functional groups, endowing it with an excellent microelectronic structure and coordination environment. These characteristics render the catalyst highly favorable for the hydrodeoxygenation of vanillin.

The FT-IR spectra of the three catalysts are displayed in Figure 5. All samples exhibit nearly identical absorption features, which is unsurprising given that they were synthesized under identical pyrolysis conditions using the same carbon and nitrogen precursors. A broad absorption band centered at approximately 3,400 cm -1 is assignable to O-H stretching vibrations originating from adsorbed water or surface hydroxyl groups. The bands observed at around 1,600 cm -1 and 1,400 cm -1 correspond to C=C and C-N stretching vibrations, respectively, confirming the successful incorporation of nitrogen into the carbon framework. The peak at approximately 1,032 cm -1 is attributed to C-O stretching vibrations (Huang et al., 2022). Consequently, the marked differences in catalytic performance among the three catalysts should be more reasonably ascribed to variations in metal particle size, electronic state, meta-support interaction, and surface acidity, as substantiated by XRD, XPS, TEM, and NH 3 -TPD analyses.

H 2 -TPR was used to analyze the redox properties and metal-support interactions of the three catalysts. As can be seen in Figure 6b, all three catalysts show typical double reduction peaks. Two main reduction peaks are detected at 304 °C and 538 °C for the CoN/C catalyst, attributed to the reduction of Co 3+ and Co 2+ species (Zhang et al., 2023). It is worth noting that the samples were pyrolyzed solely under a nitrogen atmosphere without an additional hydrogen reduction treatment. Consequently, a portion of the cobalt species still exists in an oxidized or coordinated state, which is consistent with the detection of Co-N bonds in the XPS spectrum (Figure 4c). Relative to other catalysts, the low-temperature reduction peak of CoN/C is significantly shifted to a higher temperature relative to those of NiN/C (188 °C) and CoNiN/C (198 °C). Moreover, the CoN/C exhibits a higher proportion of the high-temperature reduction peak. In contrast, the NiN/C catalyst exhibits mainly low-temperature reduction peaks, suggesting a weak metal-support interaction and a higher propensity for agglomeration and deactivation. The smaller peak area of high-temperature reduction peaks for the CoNiN/C catalyst leads to decreased catalytic activity. Thus, the H 2 -TPR findings provide further evidence for the structural superiority of the CoN/C catalyst, characterized by a high dispersion of active metal species on the catalyst surface.

NH 3 -TPD was used to study how the number of acid sites affects catalyst activity. The desorption profiles are shown in Figure 6a. Acid sites were divided into weak, medium strong, and strong based on desorption temperature (Guo et al., 2023; Han et al., 2019). All three catalysts exhibit desorption peaks to varying extents within specific acidic site ranges. Notably, the CoN/C catalyst displays a sharper and broader peak in the strong acid region, indicating a higher strong acid content. The ammonia desorption quantities are listed in Table 2. The order based on acid content is:CoN/C > CoNiN/C > NiN/C. Among them, the CoN/C catalyst exhibits the highest total acid content, suggesting a more abundant population of acidic active centers on its surface. In terms of acid site distribution, the surface of CoN/C is dominated by strong acid sites. By contrast, NiN/C and CoNiN/C exhibit an acidic characteristic dominated by moderately strong acids and the proportion of strong acid sites is relatively low. For the VAN HDO reaction, both the quantity and strength of acid sites play a critical role in the polarization, scission, and hydrogenolysis of the C-O bond. The NH 3 -TPD results confirm that the CoN/C catalyst possesses the highest total acid content and the most abundant strong acid sites. This superior surface acidic structure constitutes the key acidic feature underlying its optimal catalytic activity, high conversion, and high selectivity in VAN HDO.

In summary, the excellent catalytic performance of CoN/C arises from the synergistic combination of its high surface area and mesoporosity, well-dispersed Co nanoparticles with high Co 0 content, strong Co-N interaction, abundant strong acid sites, and favorable electronic structure. Among these, the high Co 0 content and strong acid sites play the most critical role in promoting both the hydrogenation and hydrogenolysis steps of VAN HDO.

Catalyst activity experiment

Vanillin, as a typical model of lignin compounds, is usually used in the research of Hydrodeoxygenation (HDO). In this experiment, the catalytic performance of three catalysts (CoN/C, CoNiN/C, and NiN/C) was evaluated using vanillin as the substrate. As shown in Figure 7a, using isopropanol as the solvent, specific amounts of vanillin and catalyst were added to the autoclave for the reaction. After filtering the reaction products into GC vials and analyzed by gas chromatography (GC). Notably, CoN/C demonstrated good catalytic activity, achieving a VAN conversion of 94.24% and an MMP selectivity of 96.32%. However, the bimetallic catalyst CoNiN/C achieved only 71.76% conversion of VAN and 89.71% selectivity for MMP in the HDO of VAN. Therefore, this study selected CoN/C as the best catalyst and explored its optimal reaction conditions through a series of reactions.

This experiment investigated the most economical catalyst dosage for the HDO method of preparing MMP. From Figure 7b, the lowest VAN conversion was observed in the absence of a catalyst, indicating that spontaneous hydrogenation of the aldehyde group in vanillin can occur to some extent under elevated temperature and hydrogen pressure, yielding VAL and subsequently MMP. However, GC analysis revealed that VAL, as the hydrogenation intermediate, accounted for only 2.58% of the products, while the majority consisted of other aldol condensation products and various byproducts. This result demonstrates that, although thermal reduction of the aldehyde group in vanillin is feasible to a limited degree in the absence of a catalyst, the subsequent C-O hydrogenolysis of VAL to MMP is extremely inefficient under such conditions. It was found that the introduction of the CoN/C catalyst significantly enhanced both the VAN conversion and MMP selectivity, with the optimal performance achieved at a catalyst loading of 30 mg. The results show that CoN/C can activate vanillin and reduce the activation energy necessary for the conversion of MMP. As CoN/C was added, selectivity and conversion first increased and then plateaued, indicating that more catalyst (more active sites) speeds up the reaction. Therefore, excessive catalysts cannot break the reaction equilibrium. However, if the amount of catalyst is further increased, it not only cannot effectively utilize the catalyst but also causes waste of cost and resources.

The reaction temperature has an extreme influence on the catalytic HDO of VAN. As shown in Figure 8a, after reacting at 1 MPa H 2 pressure for 4 h, With increasing reaction temperature, the conversion rate of vanillin rose from 74.61% (180 °C) to 99.25% (240 °C). The selectivity increased from 61.65% (180 °C) to 98.56% (240 °C). When the temperature reached 260 °C, the conversion and the selectivity slightly decreased, possibly resulting from partial catalyst deactivation induced by the elevated temperature. Thus, an excessively high temperature was actually unfavorable for the HDO of VAN. The increase in conversion rate can be explained by the Arrhenius equation. Under the condition that the activation energy for converting vanillin to MMP (vanillin alcohol and other substances) remains constant, a rise in temperature would enhance the reaction rate to produce more target products. Regarding the increase in selectivity, according to existing research reports, the HDO of VAN proceeds via initial hydrogenation of the aldehyde group to form vanillyl alcohol (VAL), followed by hydrogenolysis of the C-O bond in VAL to yield MMP. The increase in temperature may have accelerated the hydrogenation and dehydrogenation rates, thus resulting in a certain degree of improvement in the conversion and the selectivity. As for Others, the literature reports that they are mainly ethers formed by vanillin and solvents. The percentage of Others decreased with the increase in temperature. In conclusion, we chose 240 °C as the optimal reaction temperature for CoN/C to catalyze the hydrogenation dehydrogenation of vanillin.

Figure 8b illustrates the impact of H 2 pressure on VAN HDO. With CoN/C as the catalyst, the effect of H 2 pressure was investigated over a range of 0.5–2.0 MPa, while keeping other conditions constant. The results demonstrate that the ability of CoN/C to catalyze the HDO of VAN progressively improves with rising H 2 pressure. As the hydrogen pressure was raised from 0.5 MPa to 2.5 MPa, both the conversion of VAN and the selectivity toward MMP were significantly enhanced. This is because under the catalytic reaction of CoN/C, H 2 first diffuses from the gas phase to the liquid phase, then from the liquid phase to the catalyst surface. As the hydrogen pressure increases, it helps to increase the diffusion flux of hydrogen in the liquid film, allowing the active sites to continuously obtain sufficient hydrogen. Observations show that at a hydrogen pressure of 1.5 MPa, the conversion of VAN and the selectivity toward MMP reached their optimal values. If the hydrogen pressure continues to increase, the catalytic ability does not show significant changes. Therefore, considering energy consumption and cost factors, 1.5 MPa was ultimately selected as the best hydrogen pressure for catalyzing the HDO of VAN.

Figure 8c shows the effect of reaction time on VAN HDO. From 1 h to 4 h, VAN conversion and target selectivity steadily increased. However, extending the reaction time caused a slight decrease in catalytic performance. This is because, with increasing reaction time in the initial stage, the catalyst is gradually activated, and vanillyl alcohol is continuously hydrogenolyzed to form MMP. Therefore, both the VAN conversion rate and the MMP selectivity increase. In the later stage, the catalyst gradually loses its activity due to carbon deposition, oxidation of active sites, and structural changes. At the same time, MMP undergoes side reactions resulting in a decrease in the VAN conversion rate and a synchronous decrease in the MMP selectivity. In this experiment, when the reaction time is 4 h, the catalyst exhibits the best catalytic performance. At this point, the conversion rate of VAN is 99.94% and the selectivity of MMP is 96.15%. Therefore, the optimal reaction time of 4 h is selected for the catalytic reaction.

Solvent polarity significantly affects the HDO reaction. During the HDO process, the solvent acts not only as a reaction medium but also as a hydrogen donor. Therefore, exploring the solvent effect is advantageous for enhancing both VAN conversion and MMP selectivity. In this study, methanol, ethanol, isopropanol, n-propanol, and 1,4-dioxane were selected to explore the effects of different solvents on the HDO process catalyzed by CoN/C (Gao et al., 2022; Mamun et al., 2019). As shown in Figure 8d, when methanol and 1,4-dioxane were used as solvents, the catalyst exhibits the lowest catalytic activity, with the VAN conversion rates being 39.14% and 61.61% respectively, and the MMP selectivity being 79.80% and 82.17% respectively. However, when ethanol, n-propanol, and isopropanol were used as solvents, the catalytic activity of CoN/C gradually increased, and when isopropanol was used as the solvent, its catalytic efficiency was the best. In conclusion, the best conditions for HDO of VAN catalyzed by CoN/C are: 30 mg CoN/C as the catalyst, H 2 pressure of 1.5 Mpa, temperature of 240 °C, time of 4 h, and reaction solvent of isopropanol.

Catalyst recycling

Catalytic cycle experiments are essential for evaluating the stability of a catalyst under continuous operation and its retained activity after multiple uses. Accordingly, the CoN/C catalyst was subjected to five consecutive cycles in the VAN HDO reaction under the following conditions: 240 °C, 1.5 MPa H 2, and a reaction time of 4 h. The results are presented in the Figure 9. In the first cycle, the CoN/C catalyst exhibited good catalytic activity. With increasing cycle number, the catalytic activity showed a slow and controllable decline. After five consecutive cycles, the vanillin conversion remained at 87.68%, and the MMP selectivity was 85.77%, representing only a slight decrease compared to the first cycle. These results clearly demonstrate the excellent structural stability and catalytic durability of the CoN/C catalyst. While the recycling experiments confirm the good reusability of CoN/C, the gradual decline in conversion and selectivity over five cycles is consistent with the deactivation behavior commonly observed for similar N-doped carbon-supported metal catalysts, which is often ascribed to metal nanoparticle sintering, carbon deposition, or surface oxidation of active species during prolonged reaction (Zhou et al., 2020). Although the catalytic performance of CoN/C declined to some extent after five cycles, the VAN conversion and MMP selectivity remained at relatively high levels. This confirms the good reusability and industrial application potential of the CoN/C catalyst, providing important performance support for its practical application in the field of biomass hydrodeoxygenation.

Mechanism of VAN HDO over CoN/C catalyst

Under the action of the CoN/C catalyst, VAN undergoes HDO to form various intermediates, and the final product, MMP, is generated through the reaction pathways illustrated in Figure 10. Firstly, CoN/C facilitates the cleavage of isopropanol and molecular H 2, producing active hydrogen species that drive the HDO of VAN. Hydrogen molecules adsorb onto the active centers of Co nanoparticles and dissociate into active hydrogen atoms (Ran et al., 2022). Meanwhile, under the activation of the Co 0 active sites, vanillin molecules are adsorbed onto the CoN/C catalyst and rapidly hydrogenated to form the intermediate vanillyl alcohol (VAL). Subsequently, mainly driven by acid sites, C-O bond cleavage leads to VAL hydrogenolysis to MMP. Concurrently, the hydroxyl group of VAL participates in the hydrogenation reaction, resulting in partial reduction to the by-product HMA. This reaction may proceed via hydrogen transfer on the surface of the CoN/C catalyst. Subsequently, HMA undergoes further deoxygenation, during which the hydroxyl group is hydrogenated and removed, yielding the methyl-substituted aromatic compound MMP. In addition, a small fraction of MMP undergoes C-C bond breaking to form GUA as a by-product. The proposed transformation mechanism indicates that vanillin HDO primarily depends on the Lewis acid sites and their catalytic efficiency. Under varying reaction conditions, including different temperatures and pressures, the target product MMP is obtained. Gas chromatography (GC) analysis of the reaction mixture under various conditions confirmed VAL as a key intermediate in this reaction. As the reaction proceeded, the VAL content gradually decreased, while the concentration of the target product MMP, continuously accumulated over time. These results support the proposed stepwise reaction pathway involving aldehyde hydrogenation followed by hydroxyl hydrogenolysis.

Density functional theory (DFT) calculations and charge transfer quantitative analyses for the CoN/C and CoNiN/C are shown in Figure 11. The CoN/C exhibits an adsorption energy of −1.04 eV, whereas the CoNiN/C catalyst is −1.05 eV. Both values fall within the optimal range for catalytic reactions (−0.6 to −1.2 eV), indicating that both catalysts can effectively activate the substrate. Although the adsorption energies of CoN/C and CoNiN/C differ by only 0.01 eV, the superior catalytic performance of CoN/C is convincingly supported by comprehensive characterization results, including its smaller and better-dispersed Co crystallites (XRD/TEM), higher Co 0 content (XPS), and more abundant strong acid sites (NH 3 -TPD). Furthermore, Bader charge analysis reveals the electron transfer mechanism during the adsorption process, as shown in Figure 11e. For the CoN/C catalyst, the Co atom exhibits electron accumulation (charge value: −0.2390 eV), whereas the O atom (with a charge of +1.0988) is electron-deficient, suggesting electron transfer from O to the Co sites. In addition, details of the DFT procedure could be found in Supplementary Material.

In this work, three catalysts were successfully prepared using activated carbon as the support, employing one-pot impregnation-pyrolysis technology to enhance their catalytic activity. Experimental results demonstrated that the CoN/C exhibited the best catalytic activity. Under the optimal reaction conditions (240 °C, 4 h, 1.5 MPa H 2 and 15 mL isopropanol), the CoN/C catalyst achieved highly efficient conversion of VAN, with a conversion of 99.94% and an MMP selectivity of 96.15%. Further cycling experiments revealed that the catalyst maintained excellent performance after five consecutive reuses. Detailed characterization results showed that the Co nanoparticles in the CoN/C catalyst were highly dispersed, forming strong interactions with the nitrogen-doped carbon support, and possessed a high surface area along with plentiful strong acid sites. In addition, the presence of Co 0 and Co-N bonds provided more active centers for the reaction, significantly enhancing the overall catalytic activity. Furthermore, a possible reaction mechanism and product distribution are proposed based on the experimental findings.