Technical lignin (40–200 mesh) was purchased from Shandong Longlive Biotechnology Co., Ltd., China. Ruthenium chloride hydrate (RuCl₃•xH₂O) with Ru content 35–42% was purchased from Aladdin Technical Corporation, China. Methanol (>99.5%) was purchased from Nanjing Chemical Reagent Co., Ltd., China. Thiourea (≥99.0%) was purchased from Sinopharm Chemical Reagent Co., Ltd. Deionized water was used throughout the experiment.

The biochar support was obtained from poplar pyrolysis under a N₂ atmosphere at 450°C for 3 h. Nitrogen and sulfur co-doped supports were prepared by carbonizing a mixture of char and thiourea at a 1:2 mass ratio under a N₂ or CO₂ atmosphere at 800°C for 1 h, which were noted as CNS-N₂ and CNS-CO₂ later. To understand the effect of nitrogen and sulfur co-doping, a sole biochar sample was also carbonized at 800°C under N₂ atmosphere to prepare a undoped support (marked as C). Then, the supports were volume-impregnated by a mixed solution of RuCl₃•xH₂O dissolved by ethanol and water (1:1 volume ratio) to load Ru, the amount of which was precalculated of 5 wt%. The dried samples were reduced at 400°C for 3 h with a mixed atmosphere flow of 5% H₂ and 95% Ar (total 1 L/min) to finally obtain the Ru@C, Ru@CNS-N₂ and Ru@CNS-CO₂ catalysts⁰.

For each hydrogenolysis reaction, 1 g of dried technical lignin, 0.2 g catalyst and 30 ml methanol were put into a 100 ml mechanical stirring autoclave. Residual air in the sealed autoclave was purged with nitrogen and hydrogen for five times successively, and then 1 MPa H₂ was pressurized. The reaction was maintained at 250°C for 2 h. After that, the reactor was cooled down to room temperature and the liquid products were collected for qualitative and quantitative analysis.

The obtained organic soluble oil was analyzed by gas chromatography/mass spectrometry (GC/MS) equipped with column Agilent J&W VF-1701ms (30 m × 250 μm × 0.5 μ m). The column was heated to 200°C at the rate of 5°C/min from the initial temperature of 40°C, and then raised to 280°C at the rate of 10°C/min and held for 3 min. The external standard sample concentrations of five concentrations were used for the standard curve, and then the concentration of the compound was calculated according to the peak area. The yield of phenolic monomer was calculated according to the following formula: Y m o n o m e r = C m o n o m e r × V 1000 × m l i g n i n (1) where Y _monomer (%) was the monomer yield based on the weight of technical lignin; C _monomer (g/L) represented the concentration calculated by the peak area; V (ml) was the total volume of liquid sample taken for GC/MS analysis; and m_lignin (g) was the mass of technical lignin.

Elemental analysis to determine the contents of C, S, N and H of ash-free base of the catalysts was conducted on a Vario ELIII elemental analyzer (Elementar Company, Gernany). The Brunauer–Emmett–Teller method (BET) with a nitrogen adsorption/desorption isotherm was conducted on BSD-PM4 analyzer from BSD INSTRUMENT (Beijing) to obtain the specific surface area. The degassing temperature was 300°C, and the degassing time was 510 min. The prepared catalysts were also analyzed by FTIR (Bruker Vertex 80V, Germany) using the potassium bromide pellet technique, in which a KBr:char ratio of 100:1 (w:w) was used. The FTIR spectral region was from 400 to 4,000 cm⁻¹. The carbon structures of the catalysts were measured by a Raman technique spectrometer (ThermoFisher DXR532). The excitation laser wavelength used in the equipment was 780 nm, while the laser power was 24 mW. Based on previous study (Zhang et al., 2011; Wang et al., 2017), the original Raman spectra were curve-fitted into 10 Gaussian bands, in which the ratio of D to (G_R + V_L + V_R) band areas would suggest the ratio of the large to small aromatic ring systems. Temperature-programmed desorption (TPD) was conducted on a PCA-1200 BUILDER chemical adsorption analyzer (China) to analyze the surface acid sites. The samples were first saturated using 10 vol% NH₃ (30 ml/min) at 50°C for 30 min. After purging by N₂ (30 ml/min) at 50°C for 1 h, the NH₃-TPD was conducted by raising temperature from 50 to 650°C at a rate of 10°C/min and keeping it at 650°C for 1 h. X-ray photoelectron spectroscopy (XPS) was measured to investigate the chemical states of elements, such as Ru, C, N and S, relying on AXIS UltraDLD (Shimadzu, Japan) instrument equipped with Al Kα radiation (150 W), whose wide pass energy was 160 eV and narrow pass energy was 40 eV. Before testing, pure silica was mixed up with the catalysts as an internal standard to calibrate the binding energies. High-resolution transmission electron microscopy (HRTEM) was carried out to analyze the Ru particle size and observe the dispersion, of which the average particle diameter (d) was calculated according to the following formula: d = ∑ n i d i 3 ∑ n i d i 2 (2) where n _i was the count of particles and d _i was each characteristic diameter of particles.

The three prepared catalysts were analyzed by various methods to understand the differences in their physicochemical structures caused by N,S co-doping and/or doping atmosphere. TEM analysis was conducted to measure the morphology and distribution of metal Ru particles on the three supports. It can be observed from Figure 1 that the Ru particles on the undoped char were agglomerated with an average size of 5.42 nm, while the Ru particles were well dispersed on the N,S-co-doped char with smaller sizes. Surprisingly, the particles size of Ru on CNS-CO₂ was as small as 1.41 nm. This is consistent with the results of the study from Li et al., in which they found that the strong chemical interaction between metal and the doped sulfur atoms can greatly suppress the aggregation of metal species (Li et al., 2020). It has also been reported that smaller metal particles showed better catalytic performance during reactions (Che et al., 1989; Isaifan et al., 2013). Thus, Ru@CNS-CO₂ was expected to have a good catalytic effect on lignin depolymerization.

CO₂ is a typical agent for activation of carbon material, and the specific surface area of the carbon material can be significantly improved. Figure 2A shows the nitrogen adsorption and desorption isotherms of the three catalysts, among which Ru@C belonged to a typical type IV isotherm, while Ru@CNS-N₂ and Ru@CNS-CO₂ were typical type I isotherm. The change of the isotherm type of the catalyst from type IV to type I indicated that the mesopores in the catalyst were consumed during the N,S-co-doping process, which was also proved by the pore size distribution shown in Figure 2B. Although the three catalysts were predominant in micropores, a small peak at around 5 nm for Ru@C could be observed. Compared with Ru@C, the specific surface area of Ru@CNS-N₂ slightly reduced from 481.3 to 442.3 m²/g, indicating that the doping of nitrogen and sulfur into char under N₂ atmosphere also consumed a portion of micropores. Conversely, the specific surface area of Ru@CNS-CO₂ significantly increased to 971.1 m²/g due to the reaction of CO₂ with carbon to generate micropores.

Table 1 lists the elemental composition of the three catalysts. The content of Ru was stabilized at 4.5–4.8 wt% for all prepared catalysts. Since the biochar support has experienced the temperature as high as 800°C, the carbon content in Ru@C reached 83.2 wt% and the oxygen content reduced to 11.3 wt%. After the biochar support was doped by nitrogen and sulfur, the carbon content in Ru@CNS-N₂ reduced to 74.2 wt%. However, the reduction in carbon content did not necessarily mean that carbon was successfully replaced by heteroatoms, which could also be present as surface functional groups. Reasonably, the contents of nitrogen and sulfur, which were negligible in Ru@C, respectively increased to 6.3 and 2.8 wt% in Ru@CNS-N₂. The doping process insignificantly affect the oxygen content. When the doping atmosphere changed from N₂ to CO₂, the carbon content obviously decreased to 58.5 wt%, mainly caused by the reaction between carbon and CO₂. It is well known that CO₂ atmosphere favors the formation of oxygen-containing functional groups on biochar surface under high temperature (Liu et al., 2020), which led to an oxygen content up to 25.0 wt% in Ru@CNS-CO₂. It has been reported that the carbon atoms close to oxygen-containing functional groups had unique electron cloud densities and thus exhibited special interactions with Ru nanoparticles and/or lignin molecules (Li et al., 2019).

Figure 3 shows the FTIR spectra of the three catalysts. Compared with Ru@C, the band at around 3,435 cm⁻¹ from Ru@CNS-CO₂ and Ru@CNS-N₂, which is assigned to O-H and/or N-H, became significantly higher because of the presence of N-H groups derived from nitrogen doping. The band at 1,620 cm⁻¹ is assigned to C=O group. It is interesting to find that the doping of nitrogen and sulfur under both N₂ and CO₂ atmospheres favored the formation of C=O group. More importantly, the CO₂ atmosphere was obviously beneficial to the doping of sulfur into the char carbon skeleton, since the signals of C=S and C-S groups from Ru@CNS-CO₂ were much stronger than those from the others (Taheri-Torbati et al., 2017; Cui et al., 2018). Similar to oxygen doping, since sulfur has a larger electronegativity than carbon, the successful doping of sulfur into carbon skeleton can also cause changes in the electron cloud density of the char, which is expected to improve the catalytic performance of Ru@CNS-CO₂.

The chemical states of Ru, C, N, and S elements in the three catalysts were studied by XPS measurements. The XPS spectra of Ru 3p 3/2 can be deconvoluted into Ru⁰ and Ruⁿ⁺ species, in which the Ru⁰ species had been recognized as the active sites for hydrogenolysis of lignin (Wu et al., 2021; Ding et al., 2022). As shown in Figure 4A, the content of Ru⁰ species slightly increased from 75.45% (Ru@C) to 78.14% (Ru@CNS-N₂), suggesting that the co-doping of nitrogen and sulfur could facilitate the formation of Ru⁰ species. When the doping atmosphere changed from N₂ to CO₂, the content of Ru⁰ species in Ru@CNS-CO₂ significantly increased to 87.51%. This can be explained by the better doping of sulfur into the carbon skeleton under CO₂ atmosphere and the greater electronegativity of sulfur in comparison to carbon, thus promoting the electron transfer from CNS-CO₂ support to Ru particles. The interactions between the doped chars and Ru particles were also demonstrated by the shift of the binding energy of metallic Ru (462.9 vs 461.4 eV). Figure 4B showed the XPS spectra of C 1s. Except for the overlapping peak of Ru 3d 3/2, the C 1s XPS spectra can be deconvoluted into C-C/C=C at 284.8 eV, C-O at 285.9 eV, C-O-C at 286.9 eV, and π-π bond at 289.4 eV. The content of C-C/C=C reduced from 46.9% (Ru@C) to 42.3% (Ru@CNS-N₂), indicating that the doping of nitrogen and sulfur in N₂ reduced the graphitization degree of the char to a certain degree. However, that number increased to 58.5% in the case of Ru@CNS-CO₂, resulting from the selective consumption of small aromatic rings in carbon structure by CO₂. This conclusion was also proved by the Raman analysis of the catalysts. The value of (G_R + V_L + V_R)/D, which represents the ratio of the small to large aromatic ring systems in carbon materials (Huang et al., 2020), increased from 0.81 (Ru@C) to 1.08 (Ru@CNS-N₂) and then decreased to 0.93 (Ru@CNS-CO₂), as listed in Table 1. The N 1s XPS spectra, as exhibited in Figure 4C, were deconvoluted into pyridinic-N (398.6 eV), pyrrolic-N (400.5 eV), graphitic-N (402.0 eV), and nitrogen oxide (405 eV). The high contents of pyridinic-N and pyrrolic-N suggested the doping of nitrogen into carbon skeleton of char. Since the CO₂ atmosphere would introduce oxygen-containing functional groups to the char surface, the content of the O-N species in Ru@CNS-CO₂ was significantly higher than that in Ru@CNS-N₂. Figure 4D shows the S 2p XPS spectra, which were deconvoluted into C-S-C species of S 2p 1/2 and S 2p 3/2 as well as C-SO_x. The higher contents of C-S-C species in Ru@CNS-CO₂ also indicated the successfully doping of sulfur into the carbon skeleton of char, which was in good agreement with the FTIR analysis.

The introduction of oxygen-, nitrogen-, and/or sulfur-containing functional groups into carbon materials significantly affects their acidity, NH₃-TPD analysis was thus conducted to determine the acidic sites of the prepared catalysts. As shown in Figure 5, a small peak at 143.4°C with an area of 1,558 was found for Ru@C, suggesting the presence of very few weakly acidic sites. When the nitrogen and sulfur were doped into the char, the peak was slightly shifted to 140.7°C with an area of 2,605 (Ru@CNS-N₂) and 128.7°C with an area of 3,488 (Ru@CNS-CO₂), indicating that the acidity of the catalyst became weaker but the acid sites were more abundant, particularly under the CO₂ atmosphere.

On the basis of understanding the properties of the three catalysts, a technical lignin derived from corncob was selected to test their catalytic performances on the production of phenolic monomers. The liquid products from the lignin depolymerization were qualitatively and quantitatively analyzed by GC/MS, and the total ion chromatograms (TICs) of the products as well as the yield of each compound are shown in Figure 6 and Table 2, respectively. For comparison, a run without catalyst was also conducted, in which the total yield of all identified phenolic monomers was only 10.25 wt% (entry 16, Table 2). The most abundant compounds from the blank run, compounds 14 (methyl ferulate, 2.38 wt%) and 15 (methyl coumarate, 2.74 wt%), were not the products from the typical β-O-4 subunits. When the non-doped catalyst (Ru@C) was employed, the richest product shifted to compound 12 (methyl 3-(4-hydroxy-3-methoxyphenyl)propanoate, 5.14 wt%), followed by compounds 5 (4-ethyl-2-methoxyphenol, 4.47 wt%) and 4 (4-ethylphenol, 4.03 wt%). It should be noted that compound 12 was a hydrogenation product of compound 14. Based on our previous study, the saturation of the double bond (Cβ = C_γ) in compounds 14 and 15 reduced the bond dissociation energies of C_β-C_γ, thus contributing to the production of compounds five and four by decarboxylation, respectively (Wu et al., 2021). With the help of Ru@C, the total yield of phenolic monomers of corncob lignin depolymerization increased to 20.47 wt%. Repolymerization of lignin-derived intermediates to macromolecules has been recognized as common reactions during lignin conversion, which led to the low yield of total phenolic monomers from the blank run (Deuss et al., 2015). Since the compounds 12, 4, and five could be derived from the products 14 and 15, the improved yield of total phenolic monomers from the Ru@C run meant that the catalyst not only facilitated the depolymerization of lignin to compounds 14 and 15 but also hindered their repolymerization. Unfortunately, the yield of total phenolic monomers and the main composition of the lignin oil from the run over Ru@CNS-N₂ were very close to those from the Ru@C run. This could be explained by the similar properties between Ru@C and Ru@CNS-N₂, such as Ru particle sizes, specific surface areas, and contents of Ru⁰ species, also reflected the inapplicability of N₂ atmosphere in doping of heteroatoms into carbon materials to prepare catalyst. On the other hand, as discussed in Section 3.1, the CO₂ atmosphere not only obviously improved the specific surface area but also helped the doping of sulfur into the carbon skeleton of char, thus leading to finer Ru particle size, more Ru⁰ content, and more weak acid sites. Unsurprisingly, the yield of total phenolic monomers significantly increased to 36.41 wt% in the case of Ru@CNS-CO₂ (see Table 2), demonstrating the importance of CO₂ for the doping process. This yield was even higher than the data from our previous study by a factor of 1.34, in which Ru supported on aminated carbon nanotubes was employed as the catalyst (Wu et al., 2021). The most abundant compound from the Ru@CNS-CO₂ run was still compound 12 with a yield as high as 13.29 wt%, suggesting a promising hydrogenation ability of Ru@CNS-CO₂. Different with the runs over Ru@C and Ru@CNS-N₂, the second richest compound here was compound 10 (2,6-dimethoxy-4-propylphenol) rather than compounds four or 5, implying that Ru@CNS-CO₂ also had high activity in fragmentation of typical β-O-4 bonds (Wang et al., 2018; Wang et al., 2019). Based on the experimental results, a schematic diagram of lignin depolymerization to the main compounds can be proposed in Figure 7.