Sulfated ZrO2 supported CoMo sulfide catalyst by surface exsolution for enhanced hydrodeoxygenation of lignin-derived ethers to aromatics

Fuel xxx (xxxx) xxxx

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Sulfated ZrO2 supported CoMo sulfide catalyst by surface exsolution for enhanced hydrodeoxygenation of lignin-derived ethers to aromatics Wenjing Songa, Weikun Laib, , Yixin Lianb, Xingmao Jianga, , Weimin Yangc ⁎

⁎

a

School of Chemical Engineering & Pharmacy, Wuhan Institute of Technology, Wuhan 430073, PR China National Engineering Laboratory for Green Chemical Productions of Alcohols-ethers-esters, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005, PR China c SINOPEC Shanghai Research Institute of Petrochemical Technology, Shanghai 201208, PR China b

GRAPHICAL ABSTRACT

ARTICLE INFO

ABSTRACT

Keywords: In situ exsolution CoMo sulfide Sulfated ZrO2 Lignin-derived ethers Selective C–O cleavage Hydrodeoxygenation

Catalytic transformation of lignin is a sustainable way to provide aromatics, which depends on the structure and property of catalytic sites. A bifunctional sulfated ZrO2 supported CoMo sulfide catalyst has been synthesized by in situ exsolution approach. The exsolved CoMo-sZr(exs) catalyst significantly enhanced the initial C–O bond cleavage rate (0.058 mol·g−1·h−1) compared to the impregnated and physically mixed sulfated ZrO2 supported CoMo sulfide catalysts. Notably, the reusability tests also showed that the as-prepared catalyst presented total ethers conversion of 94% with aromatics yield of 86% after four consecutive cycles. This was related to the appropriate sulfide-support interaction, acid sites and highly-dispersed CoMo sulfide, in conformity with the catalyst characterization. Thereby it could facilitate reactant adsorption and effectively break down the C–O bonds in lignin derived fragments to form aromatics. This work points out a new avenue for the design of effective bifunctional catalysts for lignin depolymerization.

1. Introduction Lignin is a natural aromatic biopolymer that accounts for nearly 30% of the organic carbon on the Earth and is one of the few renewable sources of aromatic chemicals. Efficient cleavage of aromatic ether bonds in the lignin is the most important starting point for many lignin ⁎

valorization strategies, because it could generate valuable aromatic chemicals and provide a source of low-molecular-mass feedstocks suitable for downstream processing. However, selective cleavage of these aromatic ether bonds is difficult due to their high strength and stability. The cleavage of the representative β-O-4, α-O-4 and 4-O-5 linkages in lignin, as well as the β-5 linkage commonly found in the furan group,

Corresponding authors. E-mail addresses: [email protected] (W. Lai), [email protected] (X. Jiang).

https://doi.org/10.1016/j.fuel.2019.116705 Received 8 July 2019; Received in revised form 21 October 2019; Accepted 19 November 2019 0016-2361/ © 2019 Published by Elsevier Ltd.

Please cite this article as: Wenjing Song, et al., Fuel, https://doi.org/10.1016/j.fuel.2019.116705

Fuel xxx (xxxx) xxxx

W. Song, et al.

would thus be expected to determine the primary products when large clusters of lignin are depolymerized. The β-O-4 alkyl-aryl ether linkage (45–50%) between aromatic rings [1], is frequently selected as the starting material to investigate the mechanism for cleavage of the main ether linkages in lignin. Compared to homogeneous catalysts [2], heterogeneous catalysts are widely used due to their longer durability, recyclability and separability of reaction products, and can be tailor-made and implemented in desired shape [3]. Successful cleavage of aryl ether bonds is mostly achieved so far by hydrolysis with acid and base catalysts. In recent years, bifunctional metal-acid catalysts have been employed, showing high efficiency in the hydrocarbon fuel production from lignin conversion [4,5]. Increasing the acidity or basicity of the catalyst promotes the conversion of lignin-derived phenols (Table S1). Solid base catalyst has been used for efficient transformation of lignocellulose while increasing the yields of oxygenates [6]. However, the capacity for oxygen removal was limited [7]. Acid promoted C-O hydrogenolysis in conversion of raw lignocellulose into liquid alkanes is realized by employing a multi-functional Pd/sulfated zirconium [8] and Pt/NbOPO4 [9] catalysts. The promotion effect of Pt/NbOPO4 should be accompanied by strengthening acidity through interface interactions to achieve efficient C-O cleavage in the hydrodeoxygenation of birch wood. Furthermore, an increase in the acidity of Ni-based alumina-silica catalysts significantly promoted the HDO reaction during lignin upgrading [10]. However, the strategies explored to date using metalacid multi-functional catalysts are mostly associated with excessive hydrogenation and the achievable aromatics yields were limited [11]. As a contrast, sulfided CoMo- and NiMo-based catalysts are costeffective for C-O bond cleavage in lignin HDO conversion [12–14] and give relatively high aromatics selectivity under the same reaction conditions compared with metal-acid catalysts. The high HDO performance was attributed to the construction of CoMoS or NiMoS phase [15], where oxygenates molecule can easily be transformed to remove oxygen atom. Although over-hydrogenation of aromatic rings was retard, the capacity to break down the C-O bonds was limited. The acidpromoted HDO process provides a promising strategy to design a highly effective catalyst for selective hydrogenolysis of the C-O bonds. However, the property of acid support for NiMo sulfide was reported to largely influence the orientation of the active phase and/or the sulfurized process [16]. Zirconia promotes the sulfidation of Mo species to generate promoted “NiMoS” phase, which exhibits better catalytic activities than in alumina-based catalysts [17]. Similarly, the property of the support may be expected to determine the extent of C-O cleavage and the distribution of products [18]. The most often observed active phases in the MoS2/ZrO2 sample were the Mo species in the form of large aggregates. The sintering of MoS2 upon calcination and sulfidation treatment is well-expected considering the weak interaction between the Mo species and ZrO2 [19]. The metal-support interaction affects the structure and the distribution of active phase that decide greatly the catalytic activity [20]. Recently, supported metal catalysts prepared from exsolution of crystalline oxide precursors, such as hydrotalcite compounds [21], perovskite [22], and spinel [23], show enhanced metal dispersion and metal-support interaction. This exsolution from a composite oxide was reported to be promising to tune the active phase-support interaction strength and get better-distributed nanocatalysts. Sun et al. synthesized a sulfide nanodots implanted semiconductor parent matrix with high population, uniform dispersion, and strengthened metal-support interaction, thus, these features enhanced catalytic performance of WS2anchored-SrTiO3 in photocatalytic reaction [24]. Bringing this insight to multi-functional metal sulfide-solid acid catalysts, we successfully synthesized an exsolved metal sulfide from a composite oxide, that is, sulfated ZrO2-supported CoMo sulfide. The asprepared catalyst, as well as those prepared by impregnation and physical mixing, was explored for the HDO of diversely substituted lignin-derived mono- and dimeric aromatic ethers. Due to the

combination of highly dispersed CoMo sulfide and acid properties of sulfated ZrO2, this catalyst could selectively and efficiently cleave C-O bonds in aromatic ethers such as benzyl phenyl ether (α-O-4 linkage), 2phenylethyl phenyl ether (β-O-4 linkage), diphenyl ether (4-O-5 linkage) and benzofuran (β-5 linkage) to form aromatics. Furthermore, the key individual steps involving intermediates conversion in the overall aromatic ethers HDO process are explored and compared in detail. 2. Experimental 2.1. Reagents All reagent-grade chemicals were employed without further purification. ZrOCl2·8H2O, Zr(NO3)4·5H2O, Co(NO3)2·6H2O, (NH4)6Mo7O24·4H2O, phenolic compounds, such as Diphenyl ether (DPE), Benzyl phenyl ether (BPE), Phenethoxybenzene (PEB), 2-Phenoxy-1-phenylethanol (PPE), 2,3Benzofuran (BF) were obtained from Shanghai Aladdin Reagent. NH3·H2O solution (25–28 wt%), decalin, and dodecane were purchased from Sinopharm Chemical Reagent. 2.2. Catalyst preparation The fabrication procedure of uniformly dispersed CoMo sulfide on sulfated ZrO2 (CoMo-sZr) by in situ exsolution approach was detailed as shown in Fig. 1A. CoMoZr composite oxides were synthesized via a modified hydrothermal reaction and followed by an impregnation and calcination process. Typically, 1.4 g ZrOCl2 was first dissolved in ethanol/H2O (v/v = 1/1) solution. Then 0.4 g MoO3 was dispersed in the above solution, followed by sonication for 30 min until a homogeneous solution was achieved. Then the solution mixture was heated at 180 °C in an autoclave for 6 h. After cooling to room temperature, the solid slurry was filtered, washed several times with deionized water and finally dried at 60 °C under vacuum. The obtained MoZr hydrothermal sample was further impregnated with a solution of Co(NO3)2·6H2O and 1 M (NH4)2SO4 (Co/Mo/Zr molar ratio = 0.25/0.5/1). The resulting sample was initially dried at room temperature for 12 h and afterwards at 100 °C for 24 h, followed by calcination at 550 °C for 5 h. Then, the calcined materials were subjected to a post-synthesis modification under 15% H2/H2S gas mixture and sulfurized at 400 °C for 2 h. The obtain catalyst was denoted as CoMo-sZr(exs) with “exs” referring to the exsolution method. For comparison, CoMo/sZr(imp) and CoMo + sZr(mix) with the same Co/Mo/Zr molar ratios (0.25/0.5/1) were prepared, where “imp” and “mix” were referring to the impregnation and physical mixing method using sulfated zirconia as supports. Typically, Zr(NO3)4·5H2O was dissolved in deionized water and ammonia solution (25–28 wt%) was added dropwise to form white precipitates. After washed and dried at 100 °C for 24 h, the obtained samples were immersed in a solution of Co(NO3)2·6H2O, (NH4)6Mo7O24·4H2O and (NH4)2SO4 (Co/Mo/Zr molar ratio = 0.25/0.5/1). After being dried at 100 °C, the powder was calcined at 550 °C for 5 h. The sulfated zirconia was obtained in the same procedure without addition of Co and Mo precursor. Hence, the impregnated and calcined sulfated ZrO2-supported CoMo oxide was sulfided under the above conditions. The finished catalyst was labeled as CoMo/sZr(imp). In addition, bulk CoMo sulfide sample was prepared by vaporizing aqueous solutions of 2.47 g Co(NO3)2·6H2O and 3 g (NH4)6Mo7O24·4H2O followed by drying at 100 °C overnight and calcination in air at 500 °C for 4 h, and then sulfurization as mentioned above. The CoMo + sZr(mix) catalyst was obtained by ball-milling with 0.67 g bulk CoMo sulfide and 1.0 g sulfated zirconia. 2.3. Catalyst characterization Powder X-ray diffraction (XRD) patterns were collected on a Bruker D8 ADVANCE instrument using Cu Kα radiation (λ = 1.54178 Å) as X2

Fuel xxx (xxxx) xxxx

W. Song, et al.

Fig. 1. (A) Schematic illustration of the formation strategy of a CoMo-sZr(exs) catalyst via in situ exsolution. (B) XRD and (C) Raman spectra of CoMo-sZr samples treated stepwise.

ray source. N2-physisorption isotherms of samples were measured at 77 K on a Quantachrome Autosorb-3B system after the samples were degassed at 200 °C for 4 h. The Brunauer-Emmett-Teller (BET) specific surface area was calculated using adsorption data in the relative pressure range from 0.05 to 0.35. Scanning electron microscopy (SEM) was performed using a field-emission scanning electron microscope (Sigma Zeiss). Transmission electron microscopy (TEM) images were taken on an FEI Tecnai 30 microscope at an acceleration voltage of 300 kV. Raman spectra were performed on a Renishaw Invia Raman spectrometer, equipped with a Spectra-Physics Excelsior CW solid state laser (λ = 532 nm) as excitation source. Temperature programmed reduction with hydrogen (H2-TPR) was conducted to understand the interaction between two oxides with a homemade gas chromatograph (GC) TPR apparatus. Prior to H2-TPR experiments, the samples (50 mg each) were pretreated in flowing Ar at 200 °C for 1 h and then cooled to 50 °C in flowing Ar (99.999%, 50 ml/ min). Next, 5% H2/Ar (50 ml/min) was introduced and the sample was heated to 900 °C at a rate of 10 °C/min. The H2 consumption was monitored by GC with a thermal conductivity detector (TCD). Temperature programmed desorption with NH3 (NH3-TPD) experiments of the catalysts were acquired with a Micromeritics ASAP 2920 apparatus. About 100 mg of the sample was pre-treated at 300 °C for 1 h under He flow (30 ml/min), cooled down to 110 °C and exposed to the adsorbing gas (10% NH3 in He) for 0.5 h at 110 °C. After purging with He for 1 h, the TPD data was recorded by an online MS (Hiden Analytical QIC-20 Quadrupole Mass Spectrometer) from 110 °C with a linear heating (10 °C/min) until 800 °C. In situ infrared spectra of pyridine adsorption (Py-IR) were recorded on a Bruker Vertex 70 infrared spectrometer. The samples were pressed into thin wafers and placed in a reaction chamber with two ZnSe windows and a vacuum system. Prior to adsorption, the samples were pretreated at 200 °C under vacuum, then cooled to 30 °C, and pyridine vapor was passed over the sample for 30 min, and the pyridine adsorption spectra were recorded after degassing at 250 °C for 1 h. X-ray photoelectron spectrum (XPS) was obtained with a PHI Quantum-2000 spectrometer. After referenced by C 1 s peak (284.6 eV), the Mo 3d, Zr 3d, S 2p and Co 2p peaks were fitted by a Gaussian/ Lorentzian functions with XPSPEAK software.

2.4. Catalytic tests Catalytic HDO of lignin-derived aromatic ethers (Diphenyl ether, Benzyl phenyl ether, Phenethoxybenzene, 2-Phenoxy-1-phenylethanol, or 2,3-Benzofuran) were carried out in a 50 ml stainless steel autoclave. Typically, 1 mmol of lignin model compound of aryl ethers, 20 mg of catalyst, and 20 ml of decalin solvent were introduced into the reactor with 1 mmol dodecane as an internal standard. The air in the reactor was removed by repeated purging with N2. After the reactor was heated to 300 °C with a stirring speed of 1000 rpm, and finally it was pressurized up to 4.0 MPa with H2. H2 pressure remained stable throughout the whole reaction. A small volume of liquid product was collected at regular time intervals. Quantitative analysis of liquid products was carried out by gas chromatography (flame ionization detector; capillary column, HP-5). Catalytic result was calculated from response area of the GC-FID peaks and calibrated against dodecane as an internal standard. Conversion (Conv.), product selectivity (Sel.), and product yield were calculated as follows. The C-O cleavage rate was determined at low conversions (< 50%), defined as mole of consumed reactant per gram of catalyst per hour. The phenol accumulation rate was defined as mole of the generated phenol per gram of catalyst per hour at initial reaction stage. k and t represent the pseudo-first order rate constant (per hour) and reaction time (hour), respectively. The absence of diffusion limitation and mass transfer effect on the reaction rate were checked by varying the selected stirring speed and catalyst mass under the present reaction conditions [25].

Conv. (mol%) = 1

Sel. (A, mol%) =

moles of substrate in the product × 100% initial moles of substrate charged

moles of product (A) × 100% moles of all products

Yield (%) = Conv. × Sel. × 100%

ln(1

3

conv .) =

kt

Fuel xxx (xxxx) xxxx

W. Song, et al.

3. Results and discussion

process could produce finer and better-distributed nanocatalysts, which resulted to a higher dispersion of the active MoS2. In comparison, we prepared CoMo sulfide on sulfated zirconia by impregnation and physical mixing. As shown in Fig. 3, the XRD patterns illustrated that the mixed CoMo + sZr(mix) and impregnated CoMo/ sZr(imp) catalysts demonstrate intensive diffraction peaks, which were assigned to MoS2 and ZrO2, respectively. This is due to the larger MoS2 particles in these two samples. BET analysis revealed that the CoMo-sZr (exs) catalyst exhibited slightly higher surface area (50.8 m2 g−1) as compared to the CoMo/sZr(imp) and CoMo + sZr(mix) samples (Table S2). Additionally, the comparison of Raman spectra demonstrated both MoS2 and ZrO2 phases coexist at these samples, although these bands are much weaker in the CoMo-sZr(exs) sample than that in the other two samples, which is in agreement with XRD profiles. To gain more insights into the discrepancies in the interaction and reducibility among these samples, H2-TPR of corresponding oxide samples were measured and shown in Fig. 3C. As observed on the bulk MoO3, the peaks below 700 °C was ascribed to the reduction of Mo6+ species to Mo4+ species, and the last high temperature peak (> 750 °C) was corresponding to further reduction of Mo4+ species [27,28]. We note that the sulfated zirconia also exhibited a reduction peak at about 600 °C, corresponding to the reduction of SO42- [29] despite the fact that ZrO2 support did not show any appreciable H2 consumption [30]. As a result, according to the shape and intensity of the reduction patterns, the peak around 590 °C was attributed to the reduction of sulfated zirconia. It has been documented that the reduction of sulfated zirconia distinctly shifted to a lower temperature with increasing the metal loading [31]. Compared with the physically mixture of CoMo oxides with sulfated zirconia, the oxide of CoMo/sZr(imp) and CoMosZr(exs) samples obviously exhibited a different shape of reduction patterns. Actually, the interaction between Mo and Zr significantly changed the behavior of CoMoZr oxide toward H2 activation, in particular the reduction temperature of superficial MoO3 species shifts to a higher value over CoMo-sZr(exs) as compared with CoMo/sZr(imp). This was mostly due to the formation of Zr(MoO4)2 that made it harder to be reduced. However, its peak area was far larger than those of CoMo/sZr(imp) and CoMo + sZr(mix), indicating that the Mo species in CoMo-sZr(exs) was highly dispersed. These observations are in good agreement with XRD and Raman analysis and reinforce the presence of the synergy between CoMo sulfide and ZrO2 matrices. XPS was performed to investigate the valence states of surface atoms of the ZrO2-supported CoMo sulfide catalyst. As shown in Fig. 4A and Table S3, the binding energies of Mo6+, Mo5+, Mo4+ were found to be at 232.5–232.8 eV, 231.0–231.3 eV, and 228.7–229.0 eV, respectively [32]. After decomposition of Mo 3d spectra, the contribution of various Mo species has been calculated. 80% of molybdenum has been detected as Mo4+, the lattice fringes of MoS2 were well dispersed on the surface of ZrO2 as observed through TEM measurements. The surface fraction of Mo5+ and Mo6+ is caused by Mo oxide species that have not been completely sulfided [33]. Notably, the CoMo-sZr(exs) sample featured more Mo6+ species than the samples prepared via impregnation and mechanical mixing, which is consistent with TPR results that Mo6+ was more difficult to be reduced. However, despite that, there was no significant difference in the chemical composition of cobalt species over those investigated catalysts. The peaks of binding energy at 778.5 eV, 781.3 eV and 783.7/786.3 eV can be assigned to CoMoS, CoOx and their satellite peaks, respectively [34]. Around 70% of the cobalt was present in the CoMoS state with Co atoms located in MoS2 phase. Zr 3d in Fig. 4B shows well-known ZrO2 peaks near 184.8 and 182.4 eV for 3d3/2 and 3d5/2, respectively [35]. There was no change of the binding energy of Zr 3d5/2 core electron levels for those ZrO2-supported sulfide catalysts. Additionally, significant S 2p peaks are observed at 162.2 and 168.8 eV; these values are attributed to S2- and S6+ species, respectively [36]. The presence of S6+ can be due to SO42- groups in the catalysts, while S2- attributable to S atoms in MoS2. By contrast, S6+/S2- ratio of CoMo + sZr(mix) was significantly higher than that of exsolved and

3.1. Catalyst characterization The crystal structures of the CoMo-sZr samples treated stepwise were examined via XRD and Raman analysis. It depicted that the introduction of Zr cations in MoO3 caused some phase changes during the hydrothermal procedure. The diffraction peaks present in hydrothermal samples indicate the formation of ZrMo2O7(OH)2 compound (JCPDS card no. 27–0994). Then, the high-temperature calcination and sulfurization induced the exsolution and transformation of the Mo species to MoS2 phase. This process was accompanied with the phase transformation from ZrMo2O7(OH)2 to Zr(MoO4)2 after calcination, then further to ZrO2 after sulfurization. The characteristic peaks of ZrO2 and Zr(MoO4)2 species were observed in the diffraction pattern of the samples after calcination, while the MoS2 peaks appeared on ZrO2 after sulfurization. No characteristic peaks of MoOx were found after calcination and sulfurization. Hence, this exsolution method is effective to disperse CoMo sulfide on ZrO2 matrix. Raman spectroscopy was used to probe vibrational modes of Mo-O bonds of the sample after the hydrothermal, calcination and sulfurization procedures. In agreement with XRD, the materials after hydrothermal process exhibited weak intense bands corresponding to ZrMo2O7(OH)2. Raman bands at 748, 945 cm−1 indicate υsym (O-Mo-O) and υasym (O-Mo-O) vibrational modes of Zr(MoO4)2 over the calcined sample. After sulfurization, with disappearance of the bands belonging to ZrMo2O7(OH)2 and Zr(MoO4)2, the bands corresponding to MoS2 and ZrO2 appeared in the CoMo-sZr(exs) sample at a very low intensity, which indicated that the in situ exsolution-sulfurization method limited the aggregation of Mo species. Thus, uniformly dispersed CoMoS were successfully deposited on the sulfated zirconia surface (denoted as CoMo-sZr). It has also been documented in the literature that the MoS2 nanodots could be encapsulated and uniform dispersed in the perovskite oxide SrTiO3 matrix [24]. Hence, these exsolved sulfides are different from their bulk counterpart, serving as an effective catalyst for HDO. In order to observe the morphology differences and species distribution, SEM/TEM images of the ZrO2 supported CoMo sulfide catalysts prepared via in situ exsolution (exs), impregnation (imp) and physical mixing (mix) were shown in Fig. 2. Fig. 2A showed that CoMosZr(exs) consisted of ZrO2 particle groups with an average size of ~200 nm. In the catalyst typically used in our work, this method exquisitely leads to the uniformly distributed MoS2 (slab length ~5 nm) dispersed evenly on the ZrO2 nanoparticle surface to promote intimate interaction. High-resolution TEM images (Fig. 2C) further corroborated that the MoS2 slabs were distributed on ZrO2. The black thread-like fringes in Fig. 2B, C corresponded to the MoS2 crystallites (confirmed by XRD analysis) with an interplanar distance of 0.62 nm. Most of the Co atoms located on the side planes of MoS2 slabs [26]. In addition, the lattice spacing of ca. 0.29 and 0.31 nm were observed around the MoS2 slabs, corresponding to the distance of (0 1 1) and ( −1 1 1) crystal planes of tetragonal and monoclinic ZrO2 phase, respectively. Fig. 2B demonstrated that the well-stacked layered structure of MoS2 was uniformly distributed on the ZrO2 matrices consisting of 2–5 stacking layers with longitudinal length of 2–10 nm. Actually, the MoS2 slab lengths for impregnated and mixed samples are mainly 2–25 and 5–40 nm (Fig. 2D and Fig. 2E), respectively, far longer than that for the in situ exsolved sample. Slab length distributions and number of layers determined from corresponding TEM images are presented in Fig. 2F and Table S2. The statistical analysis was made based on at least 20 images including 500–600 slabs taken from different parts of each catalyst. The average slab length has been found to be 4.4 nm for CoMosZr(exs), 9.8 nm for CoMo/sZr(imp) and 20.5 nm for CoMo + sZr(mix), respectively. It can be seen visually that the impregnated and mixed catalysts showed a larger MoS2 slab length compared to the exsolved catalysts. As compared to conventional techniques, this exsolution 4

Fuel xxx (xxxx) xxxx

W. Song, et al.

Fig. 2. (A) SEM (B) TEM and (C) HRTEM images for CoMo-sZr(exs) catalyst; TEM images for (D) impregnation catalyst CoMo/sZr(imp) and (E) physical mixing catalyst CoMo + sZr(mix); (F) Corresponding MoS2 slab length distributions.

Fig. 3. (A) XRD and (B) Raman spectra of the CoMo-sZr(exs), CoMo/sZr(imp) and CoMo + sZr(mix) sulfide catalysts, (C) H2-TPR of the corresponding oxide samples.

Fig. 4. XPS spectra of (A) Co 2p3/2 and Mo 3d, (B) Zr 3d and S 2p for different sulfide catalysts. 5

Fuel xxx (xxxx) xxxx

W. Song, et al.

Fig. 5. NH3-TRD and Py-IR profiles of sulfide catalysts with different proximity.

impregnated samples, confirming more remaining SO42- groups on the mixed catalyst, which is consistent with NH3-TPD and Py-IR results. The acid strength and the different types of acidic sites on the synthesized catalysts were tested by NH3-TPD and Py-IR. The weak, medium, and strong acid sites were estimated from the desorption peaks in the temperature ranges of 30–275, 275–475, and 475–775 °C, respectively. As depicted in Fig. 5A, all samples gave a similar broad NH3 desorption peak between 150 and 475 °C, suggesting widely distributed weak and medium acid sites. Interestingly, the NH3-TPD profiles showed an slightly decrease in the peak at about 350 °C over CoMo/sZr(imp), compared to that of the CoMo + sZr(mix). This indicates that the amount of medium acid sites in CoMo/sZr(imp) slightly decreased, and CoMo-sZr(exs) possessed a relatively low acidity. Similarly, peaks at 1450 cm−1 and 1540 cm−1 assigned to Lewis and Brønsted acid sites (LAS and BAS) were also observed in the Py-IR spectra (Fig. 5B), and the band at 1490 cm−1 was ascribed to a combination of Lewis and Brønsted acid sites [37]. As expected, both LAS and BAS were present and showed great similarities in these samples. In particular, the impregnated and exsolved samples possessed more LAS relative to the mechanical mixture, which resulted in a decreases of BAS/LAS ratios in CoMo/sZr(imp) and CoMo-sZr(exs). Such materials featured with low BAS/LAS ratios have been reported as promising catalysts for the conversion of p-cresol and lignin [38].

CoMoS phase and acid site. These features of CoMo-sZr(exs) catalyst enabled a decrease in phenol selectivity and an excellent HDO performance for aromatics production. The kinetics of PEB HDO was assessed to further examine the unique catalytic behavior of this CoMo-sZr(exs) catalyst. As expected, the exsolved CoMo-sZr(exs) catalyst enhanced the initial C-O bond cleavage rate from 0.008 mol·g−1·h−1 to 0.058 mol·g−1·h−1 compared to pure CoMo sulfide catalyst. The calculated C-O cleavage rate illuminated that CoMo-sZr(exs) are significantly more active for C-O bond cleavage, whereas both CoMo/sZr(imp) and CoMo + sZr(mix) performed poorly with slow cleavage rates of 0.011 and 0.022 mol·g−1·h−1, respectively. The closer proximity of the two components, along with a stronger metal sulfide and support interaction, suggested a more finely dispersed MoS2 phase and enabled more efficient adsorption and transfer of reactants and phenol intermediates on exposed reactive sites, thus favoring the formation of aromatics. To intuitively display the effect of support on the reaction, we carried out PEB HDO using the commercial ZrO2-supported CoMo sulfide catalyst without sulfated treatment as reference. As expected, the conversion was decreased to 30.1% at 4 h, and the yield for aromatics was very low under the same reaction conditions (Fig. S1).

3.2. Phenethoxybenzene HDO over the CoMo sulfide catalysts

Apart from Phenethoxybenzene (PEB), we extended the investigation to other lignin derived aryl ether over CoMo-sZr(exs) catalyst, including 2-Phenoxy-1-phenylethanol (PPE), Benzyl phenyl ether (BPE), Diphenyl ether (DPE) and 2,3-Benzofuran (BF) under the same operating conditions. These model compounds represented the β-O-4, α-O4, 4-O-5 and β-5 linkage in lignin. For the lignin models with distinct linkage, the product distribution as a function of reaction time was displayed in Figs. 7 and 8. All types of aryl ether linkages were intrinsically different in reactivity over CoMo-sZr(exs) catalyst. The C–O bond cleavage of α-O-4 is more favorable than that of β-O-4. In the case of BPE, 90% conversion was obtained after 1 h, and the yield for phenol intermediate reached 33%. Concurrently, the tandem conversion of phenol into benzene following hydrogenolysis of dimeric lignin models also took place. A similar change trend of the phenol yield was shown both in the conversion of PBE and PPE, where phenol first accumulated and then diminished as the reaction progressed. The plausible reaction route for the aromatic ether hydrogenolysis was depicted in Scheme 1. According to the product distribution versus reaction time (Fig. 7), the phenol yield during the BPE HDO process raised to about 33% after 60 min, higher than that for the transformation of PBE and PPE (18% and 11%, respectively). This phenomenon was caused by two plausible factors. As displayed in Fig. 7D, it was found that the differences of C-O bond

3.3. Different lignin-derived aromatic ethers HDO

The interaction between different active sites plays a pivotal role in bifunctional catalysis. The β-O-4 linkage is the most abundant connection unit (around 50%) in lignin, and thus, phenethoxybenzene (PEB) is selected to investigate the effect of the interaction between CoMo sulfide and sulfated zirconia on catalytic HDO. The increase in the interaction by changing the manner of integration from physical mixing to impregnation and further to in situ exsolution, employed in this work, significantly increased both the conversion of PEB and the selectivity of aromatics. Among the three sulfated ZrO2-supported CoMo sulfides, the CoMo-sZr(exs) catalyst exhibited the best activity to cleave C-O bonds in PEB for HDO, as revealed in Fig. 6A. All the supported CoMo sulfide and pure CoMo sulfide catalysts afforded benzene, ethylbenzene and phenol as major products. No further hydrogenation of the aromatic ring was observed. Significant differences in the relative amounts of phenol intermediates were observed over those catalysts. It could be seen from Fig. 6B that the pure CoMo sulfide preserved 21% phenol in the product at a reaction time of 4 h. The CoMo-sZr(exs) catalyst was able to hydrodeoxygenate the phenolic hydroxyl more efficiently than the CoMo + sZr(mix) and CoMo/sZr(imp) catalysts. The high dispersion of CoMo sulfide originated from the interaction between CoMo sulfide and sulfated ZrO2 led to synergic effects between 6

Fuel xxx (xxxx) xxxx

W. Song, et al.

Fig. 6. (A) Time course of phenethoxybenzene HDO, (B) initial C-O bond cleavage rates and product selectivity (4 h) over the ZrO2-supported CoMo sulfide catalyst prepared by exsolution, impregnation, physical mixing and pure CoMo sulfide. Reaction conditions: 300 °C, 4 MPa, 20 mg catalyst, 1 mmol phenethoxybenzene, 20 ml decalin.

cleavage rate between reactant BPE, PEB or PPE and phenol intermediates decreased in the order: ΔBPE > ΔPEB > ΔPPE. BPE exhibited 3 times larger C-O bond cleavage rate than that for phenol, resulting in more intermediates phenol accumulation during the initial stage of reaction. Accordingly, the BPE HDO process possessed higher phenol accumulation rate than the other two substrates PEB and PPE. Consequently, the as-formed phenol during β-O-4 hydrogenolysis underwent fast C-O cleavage to produce benzene, which clearly explained

our above results. Additionally, although PPE can be converted to phenylethanol and phenol through direct hydrogenolysis, phenylethanol has not been detected during PPE conversion. To further study the reaction pathway, the possible intermediates phenylethanol were used as reactants for HDO. Comparing the kinetic results, it is clear that the C-O cleavage rate of phenylethanol were 10 times larger than that of PPE. This result demonstrated that the deoxygenation of phenylethanol was a facile and

Fig. 7. HDO kinetic curves of the lignin models with (A) α-O-4 and (B, C) β-O-4 linkages, (D) corresponding initial C-O bond cleavage rate and phenol accumulation rate at the initial reaction stage over the CoMo-sZr(exs) catalyst. 7

Fuel xxx (xxxx) xxxx

W. Song, et al.

Fig. 8. HDO kinetic curves of the lignin models with (A) 4-O-5 and (B) β-5 linkage over the CoMo-sZr(exs) catalyst.

fast process, suggesting that the C-O bond in phenylethanol was cleaved rapidly and sequentially deoxygenated to ethylbenzene once phenylethanol was produced. It explained well that the initial reaction products detected during HDO of PPE contained ethylbenzene, instead of phenylethanol. Hence, we concluded that the aromatics yield depends on the C-O cleavage rate of both lignin and phenolic intermediates. It was worth mentioning that no further hydrogenation of the aromatic ring was observed. Benzene, toluene and ethylbenzene were the major products in the reaction as shown in Scheme 1. Compared with the hydrogenolysis of BPE (α-O-4), PBE and PPE (βO-4), the hydrogenolysis conversion of DPE (4-O-5) and 2,3-Benzofuran (β-5) was much more challenging. The HDO kinetic curves of DPE and 2,3-Benzofuran were shown in Fig. 8, it was clear that the CoMo-sZr (exs) catalyst can efficiently cleave the 4-O-5, even employed for the β5 linkage. Over 80% conversion of both the 4-O-5 model DPE and β-5 linkage 2,3-Benzofuran were achieved. For DPE HDO, CoMo-sZr(exs) catalyst afforded benzene and phenol as major products, while ethylbenzene, dihydrobenzofuran (DHBF) and ethylphenol were generated

during 2,3-Benzofuran HDO. Although the extent of deoxygenation was low relative to the α-O-4, β-O-4 C-O linkage cleavage, the hydrogenation of the aromatic rings was not observed, which was consistent with the proposed reaction pathway in Scheme 1. Overall, the reaction activity decreased in the order: α-O-4 > β-O-4 > 4-O-5 > β-5. Several recent reports describe that hydrogenolysis of the aryl-aryl ether bond requires severe conditions [39]. Similarly, Zhao et al. demonstrated that the apparent activation energies, Ea(α-O-4) < Ea(β-O-4) < Ea(4O-5), vary proportionally with the bond dissociation energies [40]. This correlates well with our results. 3.4. Role of acid and CoMo sulfide sites The acidity of catalyst, especially the Lewis acid sites, can enhance HDO activity by facilitating oxy-compounds adsorption and reaction. The CoMoS sites can activate and dissociate molecular hydrogen to generate S-H groups on the MoS2 surface [41], while oxy-compounds adsorbed on the sulfide sites [42], as well as acid sites on support

Scheme 1. Reaction pathway for the cleavage of five lignin-derived ethers over the CoMo-sZr(exs) catalyst. 8

Fuel xxx (xxxx) xxxx

W. Song, et al.

mainly aromatic compounds and minor phenols. Deep hydrogenation of aromatic ring did not occur under the reaction conditions. These results suggest that the ZrO2-supported CoMo sulfide catalyst prepared by in situ exsolution approach was stable during the reaction process without obvious loss of its activity. 4. Conclusion In summary, we designed a supported metal sulfide catalyst consisting of CoMo sulfide dispersed in sulfated ZrO2 matrix. Both the sulfated ZrO2-CoMo sulfide interaction and their influence on the cleavage of aryl ether linkages were studied. Compared with the impregnated and physical mixed sulfated ZrO2 supported CoMo sulfide (0.011 and 0.022 mol·g−1·h−1), the C-O ether bonds cleavage rate over the catalyst prepared by in situ exsolution is remarkably enhanced to 0.058 mol·g−1·h−1, due to the synergistic effect between CoMo sulfide and acid sites with close proximity. With an increase in the interaction and the enhanced MoS2 dispersion, the activity toward oxygen removal distinctly increased and aromatics yield above 90% was obtained for βO-4 linkage cleavage after 5 h reaction. Kinetic results demonstrated that the aromatics yield depends on the C-O cleavage rate of both aryl ether and phenol intermediates. Meanwhile, a high proportion of aromatics (86%) could be obtained after 4 consecutive HDO cycles with the aryl ether mixture. The catalyst was stable during the reaction process without obvious loss of its activity and selectivity.

Fig. 9. Recyclability test for HDO of aryl ether mixture on the CoMo-sZr(exs) catalyst after 8 h.

surface [43]. Active hydrogen atoms on sulfide surface react with the adsorbed oxy-compound intermediates [44]. Gevert et al. [45] reported that a σ-bonding adsorption through the oxygen atom would give C-O bond hydrogenolysis in CoMoS catalysts. Hence, the introduction of sulfated ZrO2 could improve the dispersion of CoMo sulfide, which facilitated the C-O hydrogenolysis of catalyst. On the other hand, it has been postulated that oxy-compound molecule would mostly be adsorbed through O atom in an inclined adsorption mode due to the presence of the incompletely coordinated Zr4+ cations (Lewis acid sites) [46]. However, more condensation products were formed on alumina compared to ZrO2 due to the presence of strong interaction between oxy-compounds with Lewis acid site of the alumina support. Benzene as a main product was obtained over ZrO2-supported CoMo catalysts [47]. In our case, sulfated ZrO2 acted as Lewis acid site could adsorb oxy-compounds during the reaction (Scheme S1) and selectively favored the production of aromatics, in accordance with previous study [48]. Likewise, Wang et al. demonstrated that Ga doping increased the Lewis acidity, which could effectively chemisorb and convert vanillin and acetovanillone into their deoxygenated products [49]. Accordingly, in present work, lignin-derived ethers were absorbed on the CoMo sulfide and Lewis acid site of sulfated ZrO2 through O atoms. Hydrogenolysis of ether linkages was facilitated by the generated S-H on CoMo sulfide phase, and then the deoxygenation reaction was subsequently performed over CoMo sulfide surface under the same reaction conditions. Benzene, toluene and ethylbenzene were obtained as major products without further hydrogenation.

Declaration of Competing Interest 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. Acknowledgements This work was funded by National Natural Science Foundation of China (21703179, 21878237, U1463210), the Fundamental Research Funds for the Central Universities of China (20720170103) and the project (K201913) from Wuhan Institute of Technology. The authors gratefully acknowledge funding through Wuhan Application Foundation and Frontier Project (2018010401011291). Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.fuel.2019.116705.

3.5. Catalyst reusability for aryl ether mixture HDO

References

Finally, we tested the catalyst stability of the CoMo-sZr(exs) catalyst in typical aryl ether mixtures derived from lignin with successive runs. As shown in Fig. 9, benzene (49.8% yield), toluene (11% yield), ethylbenzene (32.2% yield) and phenols (4% yield) were detected as major products after 8 h. On the basis of time course of the product distribution analysis, at the initial reaction stage, only lignin models and dodecane were observed in GC spectra (Fig. S2). When the reaction duration was increased to 1 h, aromatic compounds and phenol gradually appeared at the preferential expense of BPE and PPE. With the reaction time prolonged to 3 h, products consisting of dihydrobenzofuran and ethylphenol were obtained with the consumption of 2,3-benzofuran, and the benzene and ethylbenzene yields significantly increased. After 5 h, full conversion of the aryl ether mixture was achieved except for the diphenyl ether. Nevertheless, the total conversion of aryl ether mixture reached 97%, and the total yield of benzene, toluene and ethylbenzene increased to 93% after 8 h. Moreover, we explored the reusability of this CoMo-sZr(exs) catalyst for the HDO of aryl ether mixture. Overall, the catalytic performance remained stable during all recycle runs, still presenting a total conversion of 94% after 4 consecutive cycles. The compositions of reaction products are

[1] Chakar FS, Ragauskas AJ. Review of current and future softwood kraft lignin process chemistry. Ind Crops Prod 2004;20:131–41. [2] Gao J, Zhu S, Dai Y, Xiong C, Li C, Yang W, et al. Performance and mechanism for extractive desulfurization of fuel oil using modified polyethylene glycol. Fuels 2018;233:704–13. [3] Endalew AK, Kiros Y, Zanzi R. Inorganic heterogeneous catalysts for biodiesel production from vegetable oils. Biomass Bioenergy 2011;35:3787–809. [4] Zhang W, Chen J, Liu R, Wang S, Chen L, Li K. Hydrodeoxygenation of ligninderived phenolic monomers and dimers to alkane fuels over bifunctional zeolitesupported metal catalysts. ACS Sustainable Chem Eng 2014;2:683–91. [5] Yao G, Wu G, Dai W, Guan N, Li L. Hydrodeoxygenation of lignin-derived phenolic compounds over bi-functional Ru/H-Beta under mild conditions. Fuel 2015;150:175–83. [6] Alper K, Tekin K, Karagöz S. Hydrothermal liquefaction of lignocellulosic biomass using potassium fluoride-doped alumina. Energy Fuels 2019;33:3248–56. [7] Nakagawa Y, Ishikawa M, Tamura M, Tomishige K. Selective production of cyclohexanol and methanol from guaiacol over Ru catalyst combined with MgO. Green Chem 2014;16:2197–203. [8] Luo Z, Zhao C. Mechanistic insights into selective hydrodeoxygenation of ligninderived β-O-4 linkage to aromatic hydrocarbons in water. Catal Sci Technol 2016;6:3476–84. [9] Xia Q, Chen Z, Shao Y, Gong X, Wang H, Liu X, et al. Direct hydrodeoxygenation of raw woody biomass into liquid alkanes. Nat Commun 2016;7:11162.

9

Fuel xxx (xxxx) xxxx

W. Song, et al. [10] Broglia F, Rimoldi L, Meroni D, De Vecchi S, Morbidelli M, Ardizzone S. Guaiacol hydrodeoxygenation as a model for lignin upgrading. Role of the support surface features on Ni-based alumina-silica catalysts. Fuel 2019;243:501–8. [11] Zhao C, Lercher JA. Upgrading pyrolysis oil over Ni/HZSM-5 by cascade reactions. Angew Chem 2012;51:5935–40. [12] Song W, Zhou S, Hu S, Lai W, Lian Y, Wang J, et al. Surface engineering of CoMoS nanosulfide for hydrodeoxygenation of lignin-derived phenols to arenes. ACS Catal 2018;9:259–68. [13] Jongerius AL, Jastrzebski R, Bruijnincx PCA, Weckhuysen BM. CoMo sulfide-catalyzed hydrodeoxygenation of lignin model compounds: An extended reaction network for the conversion of monomeric and dimeric substrates. J Catal 2012;285:315–23. [14] Araya PE, Droguett SE, Neuburg HJ, Badilla-Ohlabum R. Catalytic wood liquefaction using a hydrogen donor solvent. Can J Chem Eng 1986;64:775–80. [15] Topsøe H, Clausen BS, Massoth FE. Hydrotreating catalysis. Anderson JR, Boudart M, editors. Ctalysis science and technology, 11. Berlin, Germany: Springer Verlag; 1996. p. 34. [16] Ninh TKT, Laurenti D, Leclerc E, Vrinat M. Support effect for CoMoS and CoNiMoS hydrodesulfurization catalysts prepared by controlled method. Appl Catal A 2014;487:210–8. [17] Maugé F, Duchet JC, Lavalley JC, Houssenbay S, Payen E, Grimblot J, et al. The sulphided state of nickel molybdenum catalysts supported on zirconia and aluminates. Catal Today 1991;10:561–77. [18] Kumar CR, Anand N, Kloekhorst A, Cannilla C, Bonura G, Frusteri F, et al. Solvent free depolymerization of Kraft lignin to alkyl-phenolics using supported NiMo and CoMo catalysts. Green Chem 2015;17:4921–30. [19] Breysse M, Afanasiev P, Geantet C, Vrinat M. Overview of support effects in hydrotreating catalysts. Catal Today 2003;86:5–16. [20] Lai W, Song W, Pang L, Wu Z, Zheng N, Li J, et al. The effect of starch addition on combustion synthesis of NiMo-Al2O3 catalysts for hydrodesulfurization. J Catal 2013;303:80–91. [21] Wu G, Zhang C, Li S, Huang Z, Yan S, Wang S, et al. Sorption enhanced steam reforming of ethanol on Ni-CaO-Al2O3 multifunctional catalysts derived from hydrotalcite-like compounds. Energy Environ Sci 2012;5:8942–9. [22] Neagu D, Tsekouras G, Miller DN, Ménard H, Irvine JTS. In situ growth of nanoparticles through control of non-stoichiometry. Nat Chem 2013;5:916–23. [23] Muroyama H, Nakase R, Matsui T, Eguchi K. Ethanol steam reforming over Ni-based spinel oxide. Int J Hydrogen Energy 2010;35:1575–81. [24] Sun YF, Yang YL, Chen J, Li M, Zhang YQ, Li JH, et al. Toward a rational photocatalyst design: a new formation strategy of co-catalyst/semiconductor heterostructures via in situ exsolution. Chem Commun (Camb) 2018;54:1505–8. [25] Madon RJ, Boudart M. Experimental criterion for the absence of artifacts in the measurement of rates of heterogeneous catalytic reactions. Ind Eng Chem Fundam 1982;21:438–47. [26] Wu Z, Guo J, Wang J, Liu R, Xiao W, Xuan C, et al. Hierarchically porous electrocatalyst with vertically aligned defect-rich CoMoS nanosheets for the hydrogen evolution reaction in an alkaline medium. ACS Appl Mater Interfaces 2017;9:5288–94. [27] Bhaskar T, Reddy KR, Kumar CP, Murthy MRVS, Chary KVR. Characterization and reactivity of molybdenum oxide catalysts supported on zirconia. Appl Catal A 2001;211:189–201. [28] Chary KVR, Reddy KR, Kishan G, Niemantsverdriet JW, Mestl G. Structure and catalytic properties of molybdenum oxide catalysts supported on zirconia. J Catal 2004;226:283–91. [29] Yang K, Li H, Zhao S, Lai S, Lai W, Lian Y, et al. Improvement of activity and stability of CuGa promoted sulfated zirconia catalyst for n-butane isomerization. Ind Eng Chem Res 2018;57:3855–65. [30] Li Z, Tian Y, He J, Wang B, Ma X. High CO methanation activity on zirconia-

supported molybdenum sulfide catalyst. J Energy Chem 2014;23:625–32. [31] Córdoba LF, Sachtler WMH, Montes de Correa C. NO reduction by CH4 over Pd/Cosulfated zirconia catalysts. Appl Catal B 2005;56:269–77. [32] Lai W, Pang L, Zheng J, Li J, Wu Z, Yi X, et al. Efficient one pot synthesis of mesoporous NiMo-Al2O3 catalysts for dibenzothiophene hydrodesulfurization. Fuel Process Technol 2013;110:8–16. [33] Song W, Lai W, Chen Z, Cao J, Wang H, Lian Y, et al. Fabrication of 3D porous hierarchical NiMoS flowerlike architectures for hydrodesulfurization applications. ACS Appl Nano Mater 2018;1:442–54. [34] Dai X, Du K, Li Z, Liu M, Ma Y, Sun H, et al. Co-doped MoS(2) nanosheets with the dominant CoMoS phase coated on carbon as an excellent electrocatalyst for hydrogen evolution. ACS Appl Mater Interfaces 2015;7:27242–53. [35] Damyanova S, Petrov L, Centeno MA, Grange P. Characterization of molybdenum hydrodesulfurization catalysts supported on ZrO2-Al2O3 and ZrO2-SiO2 carriers. Appl Catal A 2002;224:271–84. [36] Manoli JM, Potvin C, Muhler M, Wild U, Resofszki G, Buchholz T, et al. Evolution of the catalytic activity in Pt/sulfated zirconia catalysts: structure, composition, and catalytic properties of the catalyst precursor and the calcined catalyst. J Catal 1998;178:338–51. [37] Wu H, Duan A, Zhao Z, Li T, Prins R, Zhou X. Synthesis of NiMo hydrodesulfurization catalyst supported on a composite of nano-sized ZSM-5 zeolite enwrapped with mesoporous KIT-6 material and its high isomerization selectivity. J Catal 2014;317:303–17. [38] Ma D, Lu S, Liu X, Guo Y, Wang Y. Depolymerization and hydrodeoxygenation of lignin to aromatic hydrocarbons with a Ru catalyst on a variety of Nb-based supports. Chin J Catal 2019;40:609–17. [39] Bulut S, Siankevich S, Muyden APV, Alexander DTL, Savoglidis G, Zhang J, et al. Efficient cleavage of aryl ether C-O linkages by Rh-Ni and Ru-Ni nanoscale catalysts operating in water. Chem Sci 2018;9:5530–5. [40] He J, Zhao C, Lercher JA. Ni-catalyzed cleavage of aryl ethers in the aqueous phase. J Am Chem Soc 2012;134:20768–75. [41] Ding S, Jiang S, Zhou Y, Wei Q, Zhou W. Catalytic characteristics of active corner sites in Co-Mo-S nanostructure hydrodesulfurization-a mechanism study based on DFT calculations. J Catal 2017;345:24–38. [42] Qian W, Ishihara A, Okoshi Y, Nakakami W, Godo M, Kabe T. Elucidation of molybdenum-based catalysts using a radioisotope tracer method Part 2 Promotion effect of cobalt on molybdena/alumina catalyst. J Chem Soc, Faraday Trans 1997;43:4395–400. [43] Zhang C, Lu J, Zhang X, MacArthur K, Heggen M, Li H, et al. Cleavage of the lignin β-O-4 ether bond via a dehydroxylation-hydrogenation strategy over a NiMo sulfide catalyst. Green Chem 2016;18:6545–55. [44] Jiang L, Guo H, Li C, Zhou P, Zhang Z. Selective cleavage of lignin and lignin model compounds without external hydrogen, catalyzed by heterogeneous nickel catalysts. Chem Sci 2019;10:4458–68. [45] Gevert BS, Otterstedt JE, Massoth FE. Kinetics of the HDO of methyl-substituted phenols. Appl Catal 1987;31:119–31. [46] de Souza PM, Rabelo-Neto RC, Borges LEP, Jacobs G, Davis BH, Sooknoi T, et al. Role of keto intermediates in the hydrodeoxygenation of phenol over Pd on oxophilic supports. ACS Catal 2015;5:1318–29. [47] Bui VN, Laurenti D, Delichère P, Geantet C. Hydrodeoxygenation of guaiacol Part II: support effect for CoMoS catalysts on HDO activity and selectivity. Appl Catal B 2011;101:246–55. [48] de Souza PM, Rabelo-Neto RC, Borges LEP, Jacobs G, Davis BH, Resasco DE, et al. Hydrodeoxygenation of phenol over Pd catalysts. Effect of support on reaction mechanism and catalyst deactivation. ACS Catal 2017;7:2058–73. [49] Verma D, Insyani R, Cahyadi HS, Park J, Kim SM, Cho JM, et al. Ga-doped Cu/Hnanozeolite-Y catalyst for selective hydrogenation and hydrodeoxygenation of lignin-derived chemicals. Green Chem 2018;20:3253–70.

10