Comptes Rendus Chimie

. Inthispaper,catalytichydro-deoxygenation(HDO)ofbio-oil’smodelmolecules(aceticacid, 4-ethylguaiacol, and furfural) using Ni 2 P/HZSM-5 catalyst was carried out to better identify the products and make the modeling work of HDO process more reliable. Results showed that low temperatures favored the formation of acetaldehyde and acetone during acetic acid HDO, but disfavored the formation of aromatic hydrocarbons. Acetone was produced via the self-ketonization reaction of acetic acid. In most cases of 4-ethylguaiacol HDO, phenol, cresol, and 2, 4-dimethylphenol were the primary products. For furfural HDO, the major furan and CO products proved that the direct de-carbonylation of furfural was the main reaction. Accordingly, the main pathways of acetic acid, 4-ethylguaiacol, and furfural HDO were proposed, which could provide signiﬁcant guidance for the up-grading of crude bio-oil.


Introduction
Bio-oil from fast pyrolysis of biomass contains abundantly soluble and oxygenated organics such as carboxylic acids, guaiacols and aldehydes etc. [1][2][3][4], which result in strong acidity, high viscosity, low nickel phosphide supported on acid solids (Al 2 O 3 , HZSM-5, etc.) as catalysts for HDO attracted researcher's attention since acid supports can promote synergic interactions between the metal active phase and their intrinsic acid sites [13][14][15]. Furthermore, the use of acid supports also favored sequential hydrogenation-dehydration-hydrogenation reactions in HDO of molecules [16,17].
Noteworthy, HDO studies of carboxylic acids, guaiacols and aldehydes are mainly focused on nickel catalysts with acid solids, but fewer works use nickel phosphide catalysts with acidic supports. For instance, Chen and Falconer [18] studied the HDO of formic acid via temperature-programmed reaction (TPR) using a Ni/Al 2 O 3 and found a total conversion of formic acid at 475°C and high exposure (520 µmol/g catalyst). Results showed that CO was the main product followed by CO 2 and a small amount of CH 4 . Similarly, Peng et al. [19] explored the HDO conversion of palmitic acid with a Ni/ZrO 2 catalyst. The deoxygenation mechanism indicated hexadecanal and 1-hexadecanol were the initial products and then undergo decarbonylation to produce n-pentadecane and CO.
The HDO of another important family of biooil products, the guaiacols, has also been studied. Broglia et al. [20] performed experiments on guaiacol HDO using Ni/alumina-silica catalysts with varying Ni and silica amount. High conversions of guaiacol were achieved (up to 84%) in a very short time scale (1 h) and, generally, methylguaiacol and phenol were the primary products at 300°C under H 2 (50 bar). Zhang et al. [11] studied the catalytic HDO of guaiacol using a series of high-loading nickel phosphide catalysts supported on SiO 2 -TiO 2 . Cyclohexane, cyclohexanol, and 2-methoxycyclohexanol are the main products over all of these catalysts from 200 to 260°C. In addition, Oyama et al. [21] investigated the guaiacol HDO using Ni 2 P/ASA, Ni 2 P/FCC, Ni 2 P/ZSM-5 catalysts. As a whole, the dominant products were cresol and phenol. Especially, the effect of contact time indicated that the main pathway on Ni 2 P/ASA was the conversion of guaiacol to the catechol as primary intermediate followed by dehydroxylation to phenol. Recently, Li et al. [22] studied the HDO of guaiacol on Ni/HZSM-5 catalysts and reported that a conversion rate of 74.8% and a selectivity of 87.1% for cyclohexane were achieved. Also, it was discovered that mesopores and strong metal-support interaction promoted the conversion of guaiacol and the hydrogenation of guaiacol. In another article, the HDO of guaiacol over Ni/γ-Al 2 O 3 was carried out by Tran et al. [23]. Results showed that 96.88% guaiacol conversion and 58.98% of 1,2dimethoxybenzene production were obtained over 10 wt% Ni/Al 2 O 3 catalyst calcined at 450°C. At the same time, it has been observed that guaiacols lead to easy coke formation [24], which results in obstacles in upgrading of bio-oil.
Besides, carboxylic acids and guaiacols, aldehydes (like furfural) are also one of the numerous families of pyrolysis oil components. Zhang et al. [25] investigated a one-pot hydrogenation/dehydration conversion of furfural using Ni/SiO 2 -Al 2 O 3 bifunctional catalysts in a batch reactor and found a high selectivity to pentane and a conversion of 62.99% of furfural. In the study of HDO of n-hexane-extracted pyrolysis oil, Zhao and Lercher [26] found that furfural underwent HDO/hydrolysis to form n-pentane (64%) and tetrahydropyran (36%) over a Ni/HZSM-5 catalyst at 250°C under 5 MPa H 2 for 2 h. In addition, Wang et al. [27] studied the in situ HDO of furfural in aqueous solution over Ni/Al 2 O 3 catalysts under 1 MPa N 2 . Results showed that high temperature and amount of Ni loading facilitated the conversion of furfural and a total yield above 85% for furan and 2-methylfuran was reached at 260°C and at the methanol-towater ratio of 2:1. More recently, Lan et al. [28] explored furfural HDO using SiO 2 -supported nickel phosphide catalysts. It was found that the improvement of P content on the catalysts weakens the furanring/catalyst interaction, which led to a decreasing activity of ring-opening and ring-hydrogenation reaction.
In this background, this work is dedicated to easily understanding the HDO mechanism of bio-oil by studying the HDO of various model molecules over Ni 2 P/HZSM-5 catalysts. Here, acetic acid, 4ethylguaiacol, and furfural were selected as model molecules of bio-oil due to their abundance in pyrolysis oil [29]. The catalysts were already selected for their performance demonstrated in our previous study of acetone HDO [30], and characterized through the N 2 adsorption/desorption (BET specific surface area and pore size calculation), Inductively Coupled Plasma-Optical Emission Spectrometry (ICP-OES) elemental analysis, X-ray Powder Diffraction (XRD), pyridine adsorption by Fourier Transform Infrared Spectroscopy (FT-IR), and H 2 -TPR using Differential Scanning Calorimetry (DSC). Then, various parameters were examined, such as amount of supported phase, temperature and pressure. An in-depth characterization of liquid and gas products of model molecules HDO was performed. Probable HDO reaction routes for the conversion of the aforementioned molecules were proposed, which provide significant guidance for the selective preparation of liquid fuels and high-value chemicals from bio-oil.

Preparation and characterization of the catalysts
The Ni 2 P/HZSM-5 catalysts were prepared by the incipient wetness impregnation method as a first step and temperature-programmed reduction (TPR) as the second step [30]. Here, the support HZSM-5 was first ground and sieved to 0.35-0.6 mm particle size before loading the active phase. The specific preparation process and the characterization results of Ni 2 P/HZSM-5 catalysts by N 2 adsorption/desorption (BET surface), ICP-OES elemental analysis, and XRD can be found in our previous work [30]. In addition, the acidity of the catalysts was characterized by FT-IR (Spectrum BX, Perkin Elmer) at different pyridine desorption temperatures (100, 150, 275, and 400°C) after pyridine adsorption and could be found in the Supplementary materials ( Figure S1). The samples of the catalysts were saturated with a small amount of pyridine vapor. Prior to the pyridine adsorption, all the samples were dried at 200°C for 2 h. The H 2 -TPR of the Ni 2 P/HZSM-5 catalysts' precursors (not reduced) was performed using a DSC (SENSYS-EVO) instrument with a reduction temperature range from 100 to 600°C and a total gas flow rate of 30 mL/min (5 vol% of H 2 in N 2 ).

Activity test in setup
The HDO reaction of acetic acid, 4-ethylguaiacol and furfural were performed in a continuous fixedbed reactor. The experimental setup can be seen in Figure 1, and most of the operation conditions are the same as our previous study [30]. Here, the liquid reagent, acetic acid, 4-ethylguaiacol or furfural was vaporized in a preheater at a temperature of 150°C, 260°C or 200°C, respectively. The total pressure increased to the desired value of 0.5 MPa or 2 MPa for acetic acid HDO and 0.5 MPa or 3 MPa for 4ethylguaiacol or furfural HDO, and the temperature went to the designed value of 300-500°C once the catalysts' reduction process was completed at 450°C and 0.1 MPa of H 2 .

Products analysis
Liquid and gas compounds. The analysis of the liquid products was carried out using two different gas chromatographs equipped with the same medium polarity VF-1701-ms column (60 m × 0.25 mm × 0.25 µm film thickness) with two types of detector: mass spectrometer (GC-MS) (Clarus 580/SQ8S) system for identification and flame ionization detector (GC-FID) to quantify the amounts of compounds in liquid products. Isopropanol was used as the solvent to better analyze the liquid products in GC. Besides, the non-condensable products were gathered in a sampling bag and were then analyzed off-line using a Clarus 580 GC instrument from Perkin Elmer equipped with two detectors, a thermal conductivity (TCD) and a flame ionization detector (FID). The detailed analytical method of products was described in our previous work [30]. The level of experimental error was 2%, which was calculated by experiments repeated thrice.
In order to analyze the experimental results and to better understand the evolution of each model molecule, the conversion rate, selectivity, degree of deoxygenation (DOD), and mass yield were used. The expression of each term is given in the following: The feedstock conversion X (%) is defined as: n feed in : the amount of feedstock injected, mol. n feed out : the amount of feedstock in the liquid products, mol. The selectivity [31] Sel i (%) of the product chemicals was defined as: Carbon number i × n i Carbon number feed × n feed converted × 100% (2) Carbon number i : the number of carbon atoms in the molecular formula of product i, e.g. C 3 H 6 O, n i = 3. n i : the amount of product i, mol. n feed converted : the amount of feedstock converted, mol.
The DOD (%) was calculated according to the following equation: (3) mass of oxygen in liquid products: the total weight of oxygen in the liquid products (except for water), g. mass of oxygen in feed inlet : the total weight of oxygen in feedstock injected, g.
The yield Y i (%) of chemicals of the products (liquid, gas, and water) was determined using the formula: Mass of product i is the mass of product i, e.g. mass of

Catalysts' characterization
The nitrogen adsorption-desorption isotherms for the catalyst samples are illustrated in Figure 2(A). It can be seen that both the 5% and 10% Ni 2 P/HZSM-5 catalysts basically show a combination of IUPAC types I and IV isotherms with a typical type H4 of hysteresis loop. Hence, it can be said that these samples exhibited the co-existence of microporous and mesoporous structures: the extensive adsorption observed below relative pressures (P /P 0 < 0.2) is characteristic of microporous structures, while intermediate relative pressures are attributed to mesoporous structures. Type H4 hysteresis is generally associated with narrow slit pores [32].
The pore size distribution of Ni 2 P/HZSM-5 catalysts was shown in Figure 2(B). It can be seen that Ni 2 P/HZSM-5 catalysts possess a majority of mesoporous distribution from 3 nm to 18 nm and a pinch of the microporous distribution below 2 nm. A higher mesoporous contribution for 5% Ni 2 P/HZSM-5 than the case of 10% Ni 2 P/HZSM-5 was observed. This can be probably attributed to the smaller Ni 2 P phase particles and lower Ni 2 P content on the 5% Ni 2 P/HZSM-5 compared to the 10% Ni 2 P/HZSM-5 catalyst.

H 2 -TPR characterization using differential scanning calorimetry (DSC)
The H 2 -TPR (DSC) of the Ni 2 P/HZSM-5 catalysts' precursors with a reduction temperature range from 100°C to 600°C can be applied to study the formation process of the Ni 2 P active phase. As shown in Figure 3, a small peak was found between 250°C-300°C for 5% Ni 2 P/HZSM-5 precursor, indicating that some intermediates (for example, Ni x P y O z species) existed during this stage. No peak was observed during this stage in the case of 10% Ni 2 P/HZSM-5 precursor. This was probably because the higher Ni and P content formed bigger particles of active phase on the internal surface of pores and surface of 10% Ni 2 P/HZSM-5 than 5% Ni 2 P/HZSM-5 catalyst and thus prevented the internal Ni x P y O z intermediates from being transformed. The further transformation of intermediates to Ni 2 P via the reduction of Ni x O y and phosphorus species was associated with the presence of a broad peak at  an onset temperature around 400°C. This peak for 5% Ni 2 P/HZSM-5 was narrower than the case of 10% Ni 2 P/HZSM-5, which also indicated that the intermediates are easier to be reduced. In the literature, Oyama et al. also reported almost the same reduction peak of NiO species in the H 2 -TPR result of Ni 2 P/ASA precursor [21], and Chen et al. reported a higher onset reduction temperature around 500°C of oxidized nickel species and phosphorus species (such as PO 3− 4 , P 2 O 4− 7 , and (PO − 3 ) n ) to form Ni 2 P phase [33].

HDO of acetic acid
To get a liquid product with a low oxygen content, high yield of liquid phase is important and interesting. First of all, HDO of acetic acid using both the 5% and 10% Ni 2 P/HZSM-5 catalysts was carried out. The specific results were presented in Supplementary materials (Table S1 and Figure S2). In summary, both the catalysts exhibited an excellent catalytic activity for HDO reaction, and their similar behavior was observed. Subsequently, the 5% Ni 2 P/HZSM-5 catalyst was selected for further study.

Effect of temperature and pressure
Conversion rate, DOD, and yield. Table 1 showed the effect of temperature and pressure on the conversion rate, DOD, and yield (liquid, gas, and H 2 O) of acetic acid HDO. The results exhibited a significantly higher conversion rate (97%) and DOD (96%) at 450°C than at 350°C, suggesting that temperature had a relevant effect on the catalytic activity of Ni 2 P/HZSM-5. The maximum yield of 50% for gas products was obtained at 450°C accompanied by the minimum yield of liquid products (28%, H 2 O free). This evolution can be attributed to a notable increase in the hydrogenolysis of C-C bonds for acetic acid followed by dehydration and hydrogenation reactions to give more gas products, especially, CO and CO 2 . In a previous study, Badari et al. [34] found similar results during acetic acid HDO using a Ni/SiO 2 catalyst, namely, that this catalyst favored the C-C bond hydrogenolysis to form lighter gaseous products. Concerning the water content, it first increased, and then decreased when the temperature increased from 350°C to 450°C, suggesting that the main reaction of acetic acid HDO at a medium temperature is dehydration.
Selectivity in the liquid phase. The selectivity of major products for acetic acid HDO by varying the  temperature and pressure is shown in Figure 4(A). Details of the chemicals identified in the liquid phase can be seen in Supplementary materials (Table S2).
The results exhibited a significantly increasing trend in the selectivity of BTXM (benzene, toluene, xylene, and mesitylene) from around 5% to 19% when the temperature increased from 350°C to 450°C. In another work, Bej and Thompson [35] reported that mesitylene was formed via aldol condensation between mesityl oxide (MO) and acetone during the investigation of acetone condensation over molybdenum nitride and carbide catalysts. Gamman et al. [36] also proposed the same reaction route during the synthesis of methyl isobutyl ketone (MIK) over Pd/MgO/SiO 2 . On the other hand, MO was not detected in our work, which can be explained by the fact that the reaction from MO to MIK is spontaneous (enthalpy: ∆H 400 K (kcal/mol) = −14.5; Gibbs free energy: ∆G 400 K (kcal/mol) = −12.1) [36,37]. Here, it can thus be speculated that the formation of BTXM can be caused by a further aldol condensation between mesityl oxide (MO) and acetone, followed by dehydration, hydrogenation, and isomerization.
Acetone was identified as the main product below 400°C due to the self-ketonization reaction of acetic acid. This reaction was also observed in the catalytic upgrading of carboxylic acids by Huo et al. [38]. The linear MIK was probably obtained at 350°C through an aldol condensation reaction of acetone [39]. These results indicate that high temperature promotes the condensation reaction to form dimers and trimers. Above 400°C, the selectivity of acetone decreased sharply from 11% to 1%, which can be attributed to the conversion of acetone to aromatic hydrocarbons [40]. In addition, acetaldehyde was observed at low temperatures (350°C and 400°C) owing to the reduction of acetic acid.
Selectivity in the gas phase. Figure 4(B) shows the selectivity for gas products over the 5% Ni 2 P/HZSM-5 catalyst at different temperatures and pressures. Gases obtained were CO, CH 4 , CO 2 , C 2 H 4 , C 2 H 6 , C 3 H 6 and C 3 H 8 . At 350°C, similar values were observed under the two different pressures, and the highest selectivities for CO (44%) and CH 4 (33%) were obtained under 2.0 MPa. The lower values of the selectivity of CO, CH 4 and C 2 H 6 were attained at 400°C and 450°C compared to those obtained at 350°C. At 400°C and 450°C, the results also demonstrated a higher selectivity of gaseous olefins (C 2 H 4 and C 3 H 6 ) compared to 350°C. C 2 H 4 and C 3 H 6 are derived from the direct deoxygenation and hydrogenation of acetaldehyde and acetone over the acid sites of the HZSM-5 support, and are favored by high temperature. Significant amounts of CO and CO 2 were released because of the cracking of either carbonyl (RR CO) or carboxyl (RCO 2 H) functional groups [41]. The further hydrogenation of C 2 H 4 and C 3 H 6 can lead to C 2 H 6 and C 3 H 8 .

Proposed reaction pathways for acetic acid HDO
Based on the literature [38,39,42] and products formed, a set of reaction pathways for the HDO of acetic acid is proposed in Table 2. Reaction (1) is an acetic acid direct self-ketonization and decarboxylation reaction on a coordinatively unsaturated metal site of catalyst to a ketene (R 2 C=C=O) intermediate, and then to form acetone, CO 2 and H 2 O [43]. Acetone is subsequently hydrogenated and dehydrated to C 3 H 6 over the acidic function HZSM-5 in reaction (2) [44]. Reaction (3) corresponds to the reduction of acetic acid to acetaldehyde, together with the formation of H 2 O. Reaction (4) is a hydrogenation/dehydration reaction of acetaldehyde produced by the reaction (3), forming C 2 H 4 and H 2 O. Reaction (5) is an aldol condensation reaction of acetone molecule to produce diacetone alcohol and MO intermediates followed by dehydration to form MIK [35,39]. Mesitylene can be formed by further aldol condensation between MO and acetone molecules followed by dehydration in reaction (6). Mesitylene can be further converted to other aromatic hydrocarbons via demethylation and/or alkyl substitution, such as BTX (benzene, toluene and xylene) and ethylbenzene. Reaction (7) is the main hydrogenated reaction to form propane.

Effect of temperature and pressure
Conversion rate, DOD, and yield. Table 3 illustrates the influence of temperature and pressure on the conversion rate of 4-ethylguaiacol, DOD, and yield of products for HDO of 4-ethylguaiacol. The evolution of the conversion rate of 4-ethylguaiacol and DOD showed a small improvement when the temperature increased from 300°C to 500°C. At 400°C, a significant increase of the conversion rate (84%) and DOD (51%) under 3.0 MPa was obtained compared to 0.5 MPa, demonstrating that pressure had a greater effect on the conversion rate and DOD than temperature.
The yield of gas products and the amount of H 2 O exhibited an increasing trend as the temperature increased. This can be attributed to an increase of the direct deethylation of 4-ethylguaiacol to form C 2 H 6 , C-O bond cleavage of the methoxy group for producing CH 4 , and further dehydroxylation of phenol intermediates to give H 2 O. The highest yield of 32% for gas products and of 14% for H 2 O was accompanied by the lowest yield of liquid products (50%, H 2 O free) at 500°C.
In addition, a notable amount of coke (4%) was formed at 500°C, although no coke formation was observed at lower temperatures (300°C and 400°C). It is clear that coke formation is favored at a high temperature, probably via the condensation of 4ethylguaiacol during the demethoxylation reaction. In another study, Gayubo et al. [45] found a notable coke formation from the upgrading of 2methoxyphenol over ZSM-5 at high temperature.  (Table S3). The results revealed that the Ni 2 P catalyzed the HDO of 4-ethylguaiacol to   MPa showed that the selectivity of MMMEB and DMP decreased when the temperature increased from 300°C to 500°C. This suggested that 4-ethylguaiacol was first converted to DMP via deethylation and methoxy removal, followed by alkylation/transalkylation, and the MMMEB was formed by dehydroxylation, alkylation/transalkylation and/or isomerization of 4-ethylguaiacol. These reaction pathways can be explained by the presence of a Lewis basic benzene ring of guaiacol molecules, which should be chemisorbed on Lewis acidic Ni sites of Ni 2 P phase at the outset of the reaction, and then be transformed into derivatives [46]. In partic-ular, the outset reactions needed active H species to cleave C aromatic -OCH 3 bonds since the PO-H groups on Ni 2 P phase allowed spillover of the H species to the chemisorbed guaiacol on the Ni sites of Ni 2 P [31], and then further transalkylation occurred on Brønsted acid sites [47].
Cresol selectivity. In particular, the selectivity of cresol first increased and then decreased as the temperature increased. In addition, the selectivity of phenol and BTX increased as the temperature rose. These results can be explained by the fact that the initial increase of cresol is due to the demethylation of DMP, and subsequent decrease to further demethylation and/or dehydroxylation to form phenol and aromatic hydrocarbons. High temperature favored C-C and C-O bond cleavage between the aromatic ring and its branched chains. The highest selectivity of cresol (16%) was obtained at 400°C and 3 MPa.
MMMEB selectivity. At 400°C and under 0.5 MPa, it was observed that the selectivity of MMMEB (3%) and DMP (14%) was higher than that obtained under 3 MPa, while the selectivity of cresol (8%), phenol (5%) and BTX (2%) decreased significantly. It is therefore clear that high pressure favored the further reaction of MMMEB and DMP to form cresol, phenol and BTX. Subsequently, cresol and phenol underwent dehydroxylation and/or demethylation to produce BTX. Other phenols identified in the liquid phase, such as 3-methyl-4-isopropylphenol and 2ethyl-6-methylphenol, are probably produced via the isomerization of reaction intermediates. Notably, the selectivity of 2-ethyl-6-methylphenol first decreased due to a further dealkylation and then increased due to coke formation. Probably, the coke affected the pore size of the catalysts, which suppressed further dealkylation.
Selectivity in the gas phase. Gas products selectivity. Figure 5(B) shows the evolution of the selectivity of gas products for 4-ethylguaiacol HDO over the 5% Ni 2 P/HZSM-5 catalyst at different temperatures and pressures. As can be seen, the selectivity of gas products increased with the increase of temperature. The main gas product was C 2 H 6 at 300°C and 400°C, followed by C 3 H 8 or CH 4 . At 500°C, CH 4 and C 2 H 6 were the major products in gas phase. C 2 H 6 can be produced via deethylation of 4-ethylguaiacol followed by hydrogenation. CH 4 can be formed via various reaction routes, mainly via methoxy removal from 4ethylguaiacol followed by further C-O bond to give some CO and, to a lesser extent, via direct demethylation of cresol intermediate.
The increase in selectivity of CH 4 and C 2 H 6 revealed that high temperature promotes C-O bond cracking of the methoxy group and multiple C-C hydrogenolysis of 4-ethylguaiacol and other intermediates. Chen et al. [48] found that methane was yielded as a main gas product in the hydrogenation of mcresol via C-C bond cleavage on silica supported Ni catalyst. The highest selectivity of C 2 H 6 (11%) and CH 4 (15%) was obtained at 500°C. Under 3 MPa, a significant increase in the selectivity of total gas products was observed and particularly C 2 H 6 and C 3 H 8 . This behavior proves that higher pressure improved the HDO of 4-ethylguaiacol to form gas products and especially favored C 2 H 6 and C 3 H 8 . This last was possibly formed by an addition reaction between CH 4 and C 2 H 4 .

Proposed reaction pathways of 4-ethylguaiacol HDO
Based on the products formed, and the discussion in Section 3.2.1, a set of reaction pathways for the HDO of 4-ethylguaiacol is proposed and reported in Table 4. Reaction (1) corresponds to the dealkylation of 4-ethylguaiacol to 2-methoxyphenol and demethoxylation to 4-ethylphenol on Lewis acidic Ni sites, with the formation of C 2 H 4 , CH 4 and H 2 O. The C 2 H 4 was then hydrogenated to C 2 H 6 . Reaction (2) is a direct demethoxylation and alkylation reaction of 4-ethylguaiacol forming DMP below 400°C. This is a key reaction because DMP is the most primary product at 300°C. Reaction (3) is a dihydroxylation and alkylation reaction of 4-ethylguaiacol to form MMMEB at 300°C and 400°C. Instead of the alkylation reaction favored by relatively low temperatures, the dealkylation reaction, namely, the C-C bond breakage between benzene ring and its side alkyl group, was promoted by high temperatures. The DMP underwent subsequent demethylation to form cresol (p-& o-) and phenol in reaction (4). Reaction (5) is a dealkylation reaction of cresol to produce phenol and CH 4 . Similarly, phenol and CH 4 were also identified as important products of guaiacol HDO [20,49]. Aromatic hydrocarbons can be formed by the further dehydra-tion reaction of alkylphenols, as shown in reactions (6) and (7). Other alkylphenols, (3-methyl-4isopropylphenol, for example), were probably produced via the isomerization of certain intermediates.

Effect of temperature and pressure
Conversion rate, DOD, and yield. Table 5 presents the effect of temperature and pressure on the conversion rate, DOD, and yield of products for furfural HDO. Results under 3 MPa pressure exhibited a significant increase in conversion rate (from 54% to 100%) and DOD (from 39% to 92%) when the temperature increased from 300°C to 500°C. The highest yields of gas products (51%) and H 2 O (10%) were obtained at 500°C and 3 MPa and were accompanied by the lowest yield of liquid products (39%, H 2 O free). At 400°C and 3 MPa, the results showed a slight increase of the conversion rate and a decrease of liquid products and H 2 O compared to the results under 0.5 MPa. These results indicated that high pressure promoted the C-C hydrogenolysis between the furan ring and its branched chains and reduced the direct dehydration reaction to some extent. Selectivity in the liquid phase. The selectivity of major products of furfural HDO is illustrated in Figure 6(A). The detailed chemical composition of products can be found in the Supplementary (Table S4). Generally, high selectivities of furan, 2-methylfuran (2-MF), butanal, BTXM, and 1methylindan were observed in these conditions. Furan selectivity. Furan was observed as the most primary product of furfural HDO, with the selectivity of 17% (300°C and 3 MPa), 22% (400°C and 3 MPa) and 17% (400°C and 0.5 MPa). It can be deduced that direct aldehyde removal was the main reaction of furfural HDO in these conditions, and this removal was promoted by the pressure increase. Furthermore, deoxygenation of the aldehyde group of furfural under 3 MPa was also a significant reaction, since high selectivities of 2-MF (17% and 10%) were achieved at 300°C and 400°C. Accordingly, these results proved that the high pressure favored the aldehyde removal and deoxygenation of furfural but less furan ring-opening on the acid sites of 5% Ni 2 P/HZSM-5

2-vinylfuran selectivity.
A small amount of 2vinylfuran was detected, which might be linked to a route of direct substitution reaction between furan and C 2 H 4 . Although, 2-vinylfuran was observed as a secondary product in the case of hydrotreatment of the furfural-acetone condensation reaction by Ulfa et al. [51] and the catalytic reduction of furfural-methanol by Grazia et al. [52], a few studies reported that 2-vinylfuran was a product of furfural hydrotreatment without other co-reagents. The 2-vinylfuran can potentially be obtained from the methylenation reaction of furfural involving several consolidating approaches. The most frequently used methods for furfural conversion reaction are the Wittig [53] and Horner-Wadsworth-Emmons reactions [54], which have the common feature of employing the co-reagent of phosphorus-substituted carbanions. One can then suggest that the active Ni 2 P phase of the Ni 2 P/HZSM-5 catalyst favored the olefin substitution reaction of furan via the interaction between Ni 2 P and ethylene.
Other compounds. At low temperature, the formation of butanal can be attributed to the ring-opening reaction and further hydrogenation of furan. The acetone was possibly produced via the ring-opening reaction of 2-MF followed by the further hydrogenolysis reaction of C-C bond together with the release of C 2 H 4 .
Selectivity in the gas phase. Figure 6(B) shows the analysis of various gas products during the furfural HDO reaction. The results showed a very low gas production at 300°C due to the low conversion rate of furfural at low temperatures. The main gas product was CO with a selectivity of 17%. High selectivity of other gases (CH 4 , CO 2 , C 2 H 6 , C 3 H 6 and C 3 H 8 ) at 400°C and 3 MPa was also observed. CO was mainly formed by the direct decarbonylation of furfural and aldehyde intermediates (like butanal). CO 2 was possibly produced by the 2-MF ring-opening reaction followed by C-C bond cracking at high temperatures. Among these gaseous products, light alkanes were formed by the hydrogenation reaction of intermediates of furfural HDO since the ring-opening reaction produced olefins. Notably, the selectivity of CH 4 (13%) at 500°C is significantly higher than at 400°C, but the selectivity of CO was lower compared to 400°C. CH 4 can be formed mainly by the demethylation of 2-MF and probably by hydrogenation of CO and CO 2 , as reported in the literature in the case of metal Ni under high pressure of H 2 [55]. These results suggested that the selectivity of alkanes depends on the relative rates of the C-C bond cleavage, dehydration and hydrogenation reactions of furfural and its intermediates. Similarly, Davda et al. also stated that the selectivity of various alkanes is strongly linked to the relative rates of the C-C bond cleavage, dehydration and hydrogenation reactions during the aqueous-phase processing of sorbitol [55]. At 400°C, the gas product selectivity under 0.5 MPa showed a relatively narrow distribution compared to 3 MPa, and CO was the most important product, with 25% of selectivity. Thus, it can be deduced that pressure had a significant effect on the distribution of gas products of furfural HDO and that high pressure facilitated the ring-opening reactions.   Table 6 summarizes a series of reaction pathways for the HDO of furfural according to the main products obtained. Reaction (1) corresponds to the hydrodehydration of furfural to 2-MF, together with the formation of H 2 O. Reaction (2) is a direct decarbonylation reaction of furfural forming furan and CO [56]. This reaction is the primary route of furan formation due to the highest selectivity of CO among other gas products ( Figure 6B). Reaction (3) can be considered a minor reaction for furan production, associated with the formation of CH 4 . Acetone was produced by the ring-opening reaction of 2-methylfuran accompanying the release of C 2 H 4 in reaction (4). C 3 H 6 and C 3 H 8 can be produced mainly from the dehydrationhydrogenation of acetone in reaction (7), and C 2 H 4 can be easily converted to C 2 H 6 by hydrogenation.

Proposed reaction pathways of furfural HDO
Furan can be converted by hydro-opening-ring and hydrogenation to form butanal in reaction (5). Subsequently, butanal can be hydrogenated, and then dehydrated forming C 4 H 8 in reaction (6) and/or directly decarbonylated to form CO and C 3 H 6 . C 3 H 8 is then formed by hydrogenation of C 3 H 6 in reaction (7). Reaction (8) is the probable pathway allowing the formation of 2-vinylfuran from furan and C 2 H 4 . Reaction (9) is a condensation reaction between furfural and furan molecules to synthesize the 2,2-methylenebisfuran. Reaction (10) is a probable route of aromatic hydrocarbons via the addition reaction of olefins on Ni 2 P and subsequent cy-clization on HZSM-5 acidic sites. As stated in the literature, aromatic hydrocarbons can be formed by cyclization-dehydration of dienes, which comes from the oligomerization of mono-olefins (C 2 H 4 , C 3 H 6 and C 4 H 8 ) [50] or the Diels-Alder cycloaddition of 2,5-dimethylfuran with ethylene and subsequent dehydration to form p-xylene, followed by alkylation to other aromatics [57,58]. Thus, it can be deduced that 1-methylindan was also formed by these routes.

Conclusion
The HDO of acetic acid, 4-ethylguaiacol, and furfural as model molecules of bio-oil with the use of prepared Ni 2 P/HZSM-5 catalysts was studied, which could provide significant guidance for the upgrading of crude bio-oil. The results showed that the reaction temperature has a pronounced effect on the conversion rate and DOD of acetic acid and furfural using 5% Ni 2 P/HZSM-5 catalyst. However, pressure had a greater effect on the conversion rate and DOD of 4ethylguaiacol HDO than temperature.
For acetic acid HDO, the temperature and pressure mainly affected the decarboxylation, hydrogenation, and further decarbonylation associated with the release of CO, CH 4 , and CO 2 . Aromatic hydrocarbons were obtained via a further aldol condensation between MIK and acetone molecules followed by dehydration and hydrogenation.
The results of 4-ethylguaiacol HDO illustrated that 2,4-dimethylphenol, cresol, and 2-ethyl-6methylphenol were the most important reaction intermediates. The final products, such as phenol and BTX (benzene, toluene, and xylene), can be produced from those intermediates via dealkylation, dehydroxylation and isomerization. Notably, the coke formation had a slight effect on the conversion rate of 4-ethylguaiacol, but significantly affected the further dealkylation of the 2-ethyl-6-methylphenol intermediate.
Moreover, it was proved that the principal reaction of furfural HDO was the direct decarbonylation with furan and CO formation. Higher temperatures and pressures promoted the ring-opening reaction and C-C hydrogenolysis of furfural HDO. Aromatic hydrocarbons from furfural HDO were probably formed by the addition reaction and subsequent cyclization-dehydration of dienes, which come from the oligomerization of mono-olefins during the ringopening and C-C bond cleavage process.

Conflicts of interest
There are no conflicts of interests to declare.