Liquid-Phase Hydrodeoxygenation of Guaiacol over Mo 2 C Supported on Commercial CNF. Effects of Operating Conditions on Conversion and Product Selectivity

: In this work, a Mo 2 C catalyst that was supported on commercial carbon nanoﬁbers (CNF) was synthetized and tested in the hydrodeoxygenation (HDO) of guaiacol. The effects of operating conditions (temperature and pressure) and reaction time (2 and 4 h) on the conversion of guaiacol and products selectivity were studied. The major reaction products were cresol and phenol, followed by xylenols and toluene. The use of more severe operating conditions during the HDO of guaiacol caused a diversiﬁcation in the reaction pathways


Introduction
Lignocellulosic bio-oils cannot be fractionally distilled for the separation of petrochemical cuts in the current oil industry infrastructure, because they repolymerize upon heating due to the chemical and thermal instability related with their high oxygen content [1,2].Consequently, bio-oils must be further processed (upgraded) before they can be used as liquid fuel or as a source of industrial chemicals.
Liquid-phase hydrodeoxygenation (HDO) is a catalytic upgrading process that is undertaken in a wide range of pressures (2-30 MPa) and temperatures (423-723 K), where hydrogen is used to exclude excess oxygen from lignocellulosic bio-oil, with the purpose of getting a product that is chemically similar to crude [2,3].High pressures are used to ensure adequate solubility of hydrogen in the reaction media, this way increasing its availability in the vicinity of the catalyst particles and consequently accelerating the reaction rate [2].Conventional hydrotreating and hydrogenation (HYD) catalysts have been tested, as well as novel ones that are based on alkaline, alkaline-earth, transition, and noble metals supported in carbon or metal oxides.However, all of the catalysts suffer from irreversible deactivation that is caused mainly by coking, active phase loss, and water poisoning [2][3][4][5].
Metal carbides, nitrides, and phosphides were also attempted.Of those, carbides have recently received increased attention, as their properties resemble those of noble metal catalysts.The introduction of carbon on the lattice of transition metals leads to an increase in the lattice parameter Catalysts 2018, 8, 127; doi:10.3390/catal8040127www.mdpi.com/journal/catalystsa, which leads to a contraction of the d-band [6][7][8][9].Through carburization the d-electron density of transition metals increases at the Fermi level, conferring to the parent metal an electron density similar to those of noble metals, and the higher the carbon content, the higher the noble metal-like behavior of the obtained materials [6][7][8][9].Molybdenum carbide (Mo 2 C) is an excellent catalyst for the hydrogenation and hydrogenolysis of hydrocarbons [8,10].There are three catalytically relevant molybdenum carbide phases, namely, cubic (δ-Mo 2 C), hexagonal (α-Mo 2 C), and orthorhombic (β-Mo 2 C), following the notation convention defined by the Joint Committee on Power Diffraction Standards.A density functional theory (DFT) study revealed that the β-Mo 2 C phase has the strongest metallic and ionic character, and is highly stable due to its small work function.Its (111) and (001) surfaces are polar, while the (011), (101), and (100) surfaces are nonpolar [11].It is particularly suitable as electrocatalyst in reducing processes, as well as heterogeneous catalyst in reactions where electron transfer is needed from the catalyst to adsorbed reactant, characteristic of bond breaking reactions [11].Another DFT study [12] revealed that dispersive forces predict an essentially spontaneous dissociation of hydrogen molecules on β-Mo 2 C (001), independently of the surface termination.These studies show that molybdenum carbides present an interesting potential as catalyst material for HDO [8].
Metal based catalysts have been usually supported on metal oxides and zeolites, which are sometimes pointed out as the cause for catalysts deactivation [2,3,6].Within the universe of carbon supports, carbon nanofibers (CNFs), and carbon nanotubes (CNTs) have been less exploited as catalyst supports in HDO of bio-oils [13].CNFs and CNTs present high mechanical resistance, high electrical conductivity, high thermal stability, high conductibility, high capacity for hydrogen adsorption even at high temperatures, and a morphology that allows for the rapid access of reactants and evacuation of products from catalytic active sites [14][15][16][17].Moreover, acid and alkali treatments can be used to enhance the hydrophilicity by creating oxygen functionalities over the carbonaceous materials, and facilitating the dispersion of metals over the CNFs with water solutions [14,15,18].These properties of CNFs and CNTs gained the interest of the catalysis community working on HDO [13,19,20].Supported molybdenum carbides can be synthetized by carbothermal reduction [21,22].By this process, the Mo 2 C phase is formed by using carbonaceous supports as carbon source under a hydrogen atmosphere.
Several model compounds of bio-oil that are produced by fast pyrolysis from lignocellulosic biomass have been used in literature on HDO [2][3][4]10].Among them, guaiacol (2-methoxyphenol) is a suitable model compound for the evaluation of HDO catalysts, since it incorporates two different oxygen groups that are commonly present in lignin derived compounds, namely, hydroxyl (-OH) and methoxyl (-OCH 3 ), which hamper the complete deoxygenation of lignocellulosic bio-oils [2,4,23].
In this work, catalysts based on Mo 2 C supported on commercial CNF pre-treated with nitric acid (HNO 3 ) where synthetized and tested in the HDO of guaiacol.The effects of operating conditions and the reaction time on the conversion of guaiacol and products selectivity were studied.

Support Characterization
The results of the characterization of the as received (CNF o ) and pretreated (CNF) nanofibers by N 2 -physisorption and TPD are presented in Table 1.
When comparing the CNF o with the CNF, it can be seen from Table 1 that the HNO 3 pretreatment did not have any impact on the pore volume or surface area of the as received material, since the differences observed lies within the equipment experimental error.In the literature [24][25][26], the evolution of CO 2 during TPD (Temperature Programmed Desorption) has been assigned to the existence of acid groups over carbonaceous supports caused by the decomposition of carboxylic acids, carboxylic anhydrides, and lactones upon heating, while the evolution of CO has been assigned to the presence of basic and neutral groups over such support materials that is caused by the decomposition of phenols, ethers, quinones, and carbonyls.These oxygen groups can be created over carbonaceous supports by functionalization with acid or alkali pretreatments [24][25][26].The HNO 3 pretreatment substantially increased the number of oxygen groups over the as received CNF, as shown by the evolution of CO and CO 2 during the TPD experiments and by the (almost three-fold) increase in the support oxygen content, in agreement with previous works [27,28].The increase in the number of acid sites was two times higher than the increase in basic and neutral sites, in agreement with the results of other researchers [24].The evolution on CO and CO 2 during the TPD experiments indicates that the majority of the oxygen groups existent over the CNF o were mainly neutral and basic, whereas over the CNF an almost even distribution between acid and basic/neutral groups was observed.

Fresh Catalysts Characterization
As shown in Figure 1, the XRD patterns of as received and functionalized CNF showed a prominent reflexion at ca. 26 • and weak peaks at ca. 43-45 • assigned to graphitic carbon.No significative changes in reflexions assigned to C were observed after the functionalization treatment that was used.Mo 2 C/CNF catalyst also showed a major reflexion intensity at 26 • and a minor peak at 45 • , with both being assigned to the graphitic carbon phase.The same diffractograms also revealed seven small peaks assigned to the molybdenum carbide phase (Mo 2 C), with reflexions detected at 2θ angles of 34    , assigned to orthorhombic Mo 2 C (lattice parameters a = 0.4724 nm, b = 0.6002 nm, c = 0.5215 nm) [11,[29][30][31][32][33].The size of the crystallites (interpreted here as coherent diffraction domains) estimated by XRD is ca.11.2 nm.The molybdenum carbide crystallites in the Mo 2 C/CNF catalyst are of the same order of magnitude of those that are presented in other studies on similar catalysts, having in consideration that values as lower than 3 nm [34] and as high 11.3-12.6nm [19] have been reported.Moreover, other authors [29] synthetized Mo 2 C catalysts that are supported on activated carbon (AC) using similar preparation methodology and obtained particle sizes of comparable magnitude (9-12 nm) to those that were observed in this work.No other molybdenum phases were detected by XRD in the synthetized catalysts, indicating that the solid phase carburization of the molybdenum oxide used as metal precursor was fully accomplished.
Figure 2 shows the TEM micrographs of the CNF (a and b) and Mo 2 C/CNF (c and d).CNF present an average diameter of 100 nm with hollow core that was accounting almost the entire width.Mo 2 C/CNF catalyst evidenced the presence of Mo 2 C particles with a relatively narrow distribution between 5 and 12 nm, in agreement with XRD measurements.It can be seen that most of the catalytic particles were positioned at inner face of the CNF support.Apparently, some particles that were initially deposited at the external walls penetrated the CNF by etching channels from the surface inwards.This phenomenon has been earlier reported by other authors working on several metal-graphene composites [35][36][37][38][39][40][41][42].It was previously reported in the literature that iron oxide particles could gasify graphene supports, opening irregular channels throughout the graphene layers during a heat treatment at 773 K under a reductive hydrogen atmosphere for 2 h [36][37][38].When considering this literature information, the etching of the CNF support in the present study could have been caused by a similar phenomenon.The molybdenum oxides formed upon the thermal decomposition of the molybdate ions that were deposited on the surface of the CNF during the incipient wetness impregnation may have gasified the graphene layers during the carburization procedure, since this heat treatment was undertaken at a higher temperature (1023 K) and equal period of time (2 h).
Catalysts 2018, 8, x FOR PEER REVIEW 4 of 15 thermal decomposition of the molybdate ions that were deposited on the surface of the CNF during the incipient wetness impregnation may have gasified the graphene layers during the carburization procedure, since this heat treatment was undertaken at a higher temperature (1023 K) and equal period of time (2 h).thermal decomposition of the molybdate ions that were deposited on the surface of the CNF during the incipient wetness impregnation may have gasified the graphene layers during the carburization procedure, since this heat treatment was undertaken at a higher temperature (1023 K) and equal period of time (2 h).An alternative pathway for the etching of the CNF would be the catalytic hydrogenation of the graphene layers during the carbothermal reduction [38][39][40].By this route, graphene was etched by deposited metal particles at annealing temperatures above 873 K in a hydrogen atmosphere.This etching mechanism would involve the dissociation of molecular hydrogen on the metal surface, followed by the reaction of the so formed elemental hydrogen with the graphene carbon into methane [38][39][40].An alternative pathway for the etching of the CNF would be the catalytic hydrogenation of the graphene layers during the carbothermal reduction [38][39][40].By this route, graphene was etched by deposited metal particles at annealing temperatures above 873 K in a hydrogen atmosphere.This etching mechanism would involve the dissociation of molecular hydrogen on the metal surface, followed by the reaction of the so formed elemental hydrogen with the graphene carbon into methane [38][39][40].
In spite of the real mechanism involved, which is out of the scope of the contribution, to the best of our knowledge, the etching of carbonaceous materials involving molybdenum is reported in the present work for the first time.It should be noted that most published works use nitrogen gas during the carburization of molybdenum carbides supported on CNF, while in this contribution, hydrogen was used instead, and this may be an important cause for the observed etching.
Another fact that might have contributed for the appearance of the catalytic particles on the inner surface of the support could be related with the high average internal diameter of the CNF that was used in this work.Other authors [43] working on Fischer Tropsch synthesis prepared cobalt catalysts supported on CNTs and CNFs with internal diameters of 60 and 20 nm, respectively, by incipient wetness impregnation, in which more than 80% of the particles were located at the inner surface of the support materials.In another research [44], the authors impregnated CNT with internal diameters of up to 20 nm using aqueous solutions of Fe(NO 3 ) 3 and obtained catalysts in which the Fe 0 and Fe 2 C phases that were deposited inside the support material.In another research [45] on Fe/CNF catalysts for Fischer Tropsch synthesis similar results were obtained using CNT with internal diameters up to 25 nm.As so, a plausible explanation for the location of the catalytic particles at the inner surface of the CNF could be attributed to the higher average internal diameter (up to 100 nm) of the CNF that was used in this work (Figure 2).This way, the higher inner diameter would have allowed for the penetration of the impregnation solutions through the extremities of the fibers (by capillarity, allowing for the deposition of the precursors over the CNF inner surface, resulting in the formation of Mo 2 C particles that are smaller than those deposited in the outer surface.

Catalytic Activity
Table S1 (Supplementary Materials) shows the results that were obtained from the study of the effect of process conditions (time, t; temperature, T; pressure, P) on the guaiacol conversion and product selectivity over the synthetized catalysts, as well as of experiments with and without the CNF in order to determine the possible catalytic effects of the support material.

Catalyst Performance at 573 K
A set of experiments (runs 1 to 3) was undertaken at 573 K over the bare CNF support (commercial CNF pre-treated with nitric acid) and over the Mo 2 C/CNF catalyst for 2 h, with the purpose of gaining some insight about the reaction pathways.The results of those experiments are presented in Figure 3 (and Table S1).A longer reaction test (4 h) with the Mo 2 C/CNF catalyst is also included in Figure 3.The fractions of products that were identified by gas chromatography (mass balance) in those experiments were higher than 92.5%.
The conversion attained using the Mo 2 C/CNF catalyst was relatively low (9.7%), although corresponds to a more than 2.5-fold increase with respect to that achieved over the bare support after 2 h of reaction.The product selectivities were also significantly different in both runs.In the reaction over the bare support cresols, (o-cresol and p-cresol) were the main reaction products, followed by phenol, xylenols (2,4-dimethyl-phenol and 2,6-dimethyl-phenol), and toluene.
On the other hand, over the Mo 2 C/CNF the selectivity for phenol and xylenols were much higher than over the bare support, while the selectivity for cresols was much lower.In comparison with the results that were attained over the bare support, these observations seem to indicate that over the catalyst the reaction pathway favored the conversion of guaiacol into phenol and cresols, the latter being converted into xylenols by the transalkylation of methyl (-CH 3 ) groups, while disfavoring the conversion of cresols to toluene.That is to say, the Mo 2 C/CNF catalyst that was used in this study was more active in disrupting aromatic ether bonds (A-O-CH 3 ) than in dehydroxylation reaction, since none of the reaction products presented methoxyl (-OCH 3 ) groups that were attached to the phenolic component.Other authors working on HDO over Co-Mo/Al 2 O 3 catalysts at 633 K and 7 MPa [46] have previously studied the methylation and demethylation of cresols into xylenols and phenol, respectively, and concluded that these reactions comprise electrophilic substitutions involving carbonations formed from methyl groups adsorbed on acid sites over the catalyst surface.In the case of methylation, cresols adsorb on catalyst acid sites at the hydroxyl groups, while the carbocations attack the aromatic rings in the nearest accessible positions [46].In the case of demethylation, cresols leave carbocations that are formed from the methyl groups on the catalyst surface, which can latter participate in methylation of other adsorbed molecules [46].
On the other hand, over the Mo2C/CNF the selectivity for phenol and xylenols were much higher than over the bare support, while the selectivity for cresols was much lower.In comparison with the results that were attained over the bare support, these observations seem to indicate that over the catalyst the reaction pathway favored the conversion of guaiacol into phenol and cresols, the latter being converted into xylenols by the transalkylation of methyl (-CH3) groups, while disfavoring the conversion of cresols to toluene.That is to say, the Mo2C/CNF catalyst that was used in this study was more active in disrupting aromatic ether bonds (A-O-CH3) than in dehydroxylation reaction, since none of the reaction products presented methoxyl (-OCH3) groups that were attached to the phenolic component.Other authors working on HDO over Co-Mo/Al2O3 catalysts at 633 K and 7 MPa [46] have previously studied the methylation and demethylation of cresols into xylenols and phenol, respectively, and concluded that these reactions comprise electrophilic substitutions involving carbonations formed from methyl groups adsorbed on acid sites over the catalyst surface.In the case of methylation, cresols adsorb on catalyst acid sites at the hydroxyl groups, while the carbocations attack the aromatic rings in the nearest accessible positions [46].In the case of demethylation, cresols leave carbocations that are formed from the methyl groups on the catalyst surface, which can latter participate in methylation of other adsorbed molecules [46].Aiming to get a better insight on those aspects, the reaction time was prolonged up to 4 h.The results show that the higher reaction time allowed for increasing the selectivity for phenol, while the selectivity to cresols dropped abruptly.These results indicated that the increase in phenol selectivity mainly occurred by the demethylation of cresols, and was favored over its methylation into xylenols since their selectivity decreased by increasing the reaction time from 2 to 4 h (Figure 3).On the other Aiming to get a better insight on those aspects, the reaction time was prolonged up to 4 h.The results show that the higher reaction time allowed for increasing the selectivity for phenol, while the selectivity to cresols dropped abruptly.These results indicated that the increase in phenol selectivity mainly occurred by the demethylation of cresols, and was favored over its methylation into xylenols since their selectivity decreased by increasing the reaction time from 2 to 4 h (Figure 3).On the other hand, the selectivity to xylenols slightly decreased, and catechols (catechol and 3-methyl-catechol) were formed by demethylation of guaiacol, which is in agreement with the results of other authors [19] while working on similar catalysts.

Catalyst Performance at 623 K
Aiming to study the evolution of product selectivities over the Mo 2 C/CNF catalyst at higher guaiacol conversions, the reaction temperature was increased to 623 K.This increase in temperature was accompanied by an increase in operating pressure since the autoclave reactor was initially charged with 2 MPa of pure hydrogen (corresponding to a maximum operating pressure of 4.8 MPa).
The contribution of supports and active metals to the overall activity may be decoupled by determining the activity of the support alone, and that of the catalyst containing active metals on the same support [53].The results of those experiments are presented in Figure 5 (and Table S1).When the bare CNF support was used, besides phenol, cresols, and xylenols, toluene was also produced with a selectivity of 5.8%, in agreement with the results that were obtained at 573 K (run 1).The HNO 3 treated commercial CNF catalyzed the formation of toluene and catechols, since that both compounds were not detected in a blank run (Table S1).These results hinted that the CNF exerted a certain catalytic effect on the hydrodeoxygenation of guaiacol, as evidenced by the change in product selectivities, which can be related to the oxygen sites that are created by the acid treatment applied to the commercial CNF (Table 1).
Aiming to study the evolution of product selectivities over the Mo2C/CNF catalyst at higher guaiacol conversions, the reaction temperature was increased to 623 K.This increase in temperature was accompanied by an increase in operating pressure since the autoclave reactor was initially charged with 2 MPa of pure hydrogen (corresponding to a maximum operating pressure of 4.8 MPa).
The contribution of supports and active metals to the overall activity may be decoupled by determining the activity of the support alone, and that of the catalyst containing active metals on the same support [53].The results of those experiments are presented in Figure 5 (and Table S1).When the bare CNF support was used, besides phenol, cresols, and xylenols, toluene was also produced with a selectivity of 5.8%, in agreement with the results that were obtained at 573 K (run 1).The HNO3 treated commercial CNF catalyzed the formation of toluene and catechols, since that both compounds were not detected in a blank run (Table S1).These results hinted that the CNF exerted a certain catalytic effect on the hydrodeoxygenation of guaiacol, as evidenced by the change in product selectivities, which can be related to the oxygen sites that are created by the acid treatment applied to the commercial CNF (Table 1).By comparing the experiments in which CNF support and the Mo2C/CNF catalyst were used (Table 1), the conversion increased from 39.7% up to 50.6%, whereas the selectivities for phenol, cresols, and xylenols were higher.On the other hand, the selectivities for hydrocarbons and catechols in the presence of Mo2C/CNF catalyst were lower.Moreover, anisole, which had not been detected the previous experiments (Table S1), formed by dehydroxylation of guaiacol in low By comparing the experiments in which CNF support and the Mo 2 C/CNF catalyst were used (Table 1), the conversion increased from 39.7% up to 50.6%, whereas the selectivities for phenol, cresols, and xylenols were higher.On the other hand, the selectivities for hydrocarbons and catechols in the presence of Mo 2 C/CNF catalyst were lower.Moreover, anisole, which had not been detected the previous experiments (Table S1), formed by dehydroxylation of guaiacol in low quantities at 623 K [49].No hydrogenation products were detected at 623 K, in accordance with the experiments performed under milder conditions.
Noticeably, catechols were produced at 623 K with selectivity of 2.7%, as a result of the demethylation of guaiacol, which had not been detected at 573 K. Other authors working on Mo 2 C/CNF catalysts reported that the formation of phenol during HDO of guaiacol did not involve its demethylation into catechol, but instead the formation of phenol occurred by direct demethoxylation of the guaiacol [20].However, the formation of phenol having catechol as an intermediate compound was previously reported over Mo 2 C/AC catalysts [54].Moreover, another research group [19] had previously reported that the formation of catechols in fact could occur over Mo 2 C/CNF catalysts by the demethylation of guaiacol, addressing the possibility of the coexistence of the two routes above mentioned, in accordance with the results of this work.The low selectivities of catechols that were obtained in this work seem to be supported by the results of other researchers working using transition metal (Ni, Co, Fe, W, Mo) phosphides supported on silica [51] and palladium catalyst supported on zeolites [49], who stated that catechol can only be detected between HDO products when the reaction rate is low, because its conversion is almost immediate.This may explain the different reaction pathways that are reported in the literature.
On the other hand, the results that were obtained at 573 and 623 K (Figures 3 and 5) indicate that the more severe operating conditions caused a diversification of the HDO reaction pathways [52], expressed by the formation of catechol via demethylation of guaiacol as well as by the formation of anisole by its dehydroxylation.In the case of the reactions over the Mo 2 C/CNF catalysts, the use of harsher operating conditions also allowed for the formation of toluene by the dehydroxylation of cresols.Alternatively, toluene may have been produced by the deoxygenation of anisole, since the last was present in the products poll with a very low selectivity (0.4%), indicating its low formation rate or its rapid conversion.Therefore, based on the literature [2,10,13,46-52], the possible reaction pathways that support the experimental results of this work are illustrated in Figure 4 (solid and dashed lines), which includes all of the compounds identified by gas chromatography (between 69.9% and 79%).
By expanding the reaction time from 2 to 4 h, the conversion increased from 50.7% to 77.4%.By far, the major products of the guaiacol HDO were phenol and cresols, as shown in Figure 5.In spite that the selectivities of all the other identified compounds have increased, the selectivity of toluene decreased and the selectivity for catechols did not change when the reaction time was increased.
Apparently, with the increase in the reaction temperature, the Mo 2 C/CNF catalyst had better activity in the deoxygenation of the methoxy groups than in the removal of hydroxyl groups, while having a low hydrogenation activity.Supporting this reasoning was the fact that the selectivities for cresols and phenol were both high, while the selectivity for toluene was much higher than the summed selectivities for benzene and cyclohexane.However, it should be noted that the compounds identified by gas chromatography accounted for 79.1% and 69.9% of the HDO products in runs carried out at 2 and 4 h, respectively.This clearly indicates that a substantial part of the products pool was constituted by heavier condensation compounds that were impossible to identify by gas chromatography, in agreement with the results reported by other authors [20], being this fact more striking at largest reaction times.
For the sake of completeness, and aiming to have an estimation of the effect of pressure in the reaction with the Mo 2 C/CNF catalyst, a catalytic test was performed using an initial pressure of 3.0 MPa, while the temperature was kept at 623 K and the reaction time at 2 h (Figure 6).These operating conditions caused an increase in the maximum operating pressure of 1.6 MPa, which was close to the pressure increase caused by raising the temperature from 573 K to 623 K (from 3.6 to 4.8 MPa).
The results show that the selectivities for phenol (from 32.5% to 62.6%), cresols (from 15.4% to 20.6%), and xylenols (from 3.85% to 8.60%) increased with the when the pressure was increased, while minor variations were observed on the selectivities of the other products that were identified by gas chromatography (Table S1).The results can be explained by the fact that high pressures are used in HDO to ensure adequate solubility of hydrogen in the reaction media, this way increasing its availability in the vicinity as well as on the surface of the catalyst, which in turn accelerates the reaction rates while hindering the formation of soluble higher molecular weight products than cannot be detected by GC [2].In fact, when the pressure was increased in this work, the conversion of guaiacol increased from 50.7% to 59.1% and the identification of the compounds in the reaction pool increased from 79.1% to 88.6%, indicating that an increase in H 2 pressure inhibited the formation of heavier condensation products, in agreement with the literature [55].
chromatography, in agreement with the results reported by other authors [20], being this fact more striking at largest reaction times.
For the sake of completeness, and aiming to have an estimation of the effect of pressure in the reaction with the Mo2C/CNF catalyst, a catalytic test was performed using an initial pressure of 3.0 MPa, while the temperature was kept at 623 K and the reaction time at 2 h (Figure 6).These operating conditions caused an increase in the maximum operating pressure of 1.6 MPa, which was close to the pressure increase caused by raising the temperature from 573 K to 623 K (from 3.6 to 4.8 MPa).The results show that the selectivities for phenol (from 32.5% to 62.6%), cresols (from 15.4% to 20.6%), and xylenols (from 3.85% to 8.60%) increased with the when the pressure was increased, while minor variations were observed on the selectivities of the other products that were identified by gas chromatography (Table S1).The results can be explained by the fact that high pressures are used in HDO to ensure adequate solubility of hydrogen in the reaction media, this way increasing its availability in the vicinity as well as on the surface of the catalyst, which in turn accelerates the reaction rates while hindering the formation of soluble higher molecular weight products than cannot be detected by GC [2].In fact, when the pressure was increased in this work, the conversion of guaiacol increased from 50.7% to 59.1% and the identification of the compounds in the reaction pool increased from 79.1% to 88.6%, indicating that an increase in H2 pressure inhibited the formation of heavier condensation products, in agreement with the literature [55].

Spent Catalysts Characterization
Catalysts after HDO test were analyzed by XRD, and the diffractograms and the calculated Mo2C crystallite size is presented in Figure 7 and Table 2, respectively.

Spent Catalysts Characterization
Catalysts after HDO test were analyzed by XRD, and the diffractograms and the calculated Mo 2 C crystallite size is presented in Figure 7 and Table 2, respectively.No noticeable changes were observed after the catalytic tests, given that the only phases that were detected were Mo 2 C, and no MoO x were observed.Besides this, Mo 2 C crystallite size presented slightly lower values than in the fresh catalysts, more accentuated at higher temperature, which evidenced that no sintering takes place during the reaction.

Catalyst Preparation and Characterization
The as received commercial nanofibers (CNF) were functionalized, aiming to create oxygen groups and facilitating the metal dispersion on the CNF surface, as detailed in previous works [27,28].Succinctly, the CNF were treated for 2 h in diluted boiling nitric acid (HNO 3 ) using the following proportions: 2 g CNF/50 mL HNO 3 /150 mL distilled water.The so treated nanofibers were separated by vacuum filtration from the diluted acid using a cellulose filter (0.2 µm porosity and 8 cm diameter, Whatman ® , Maidstone, UK), rinsed with distilled water until the filtrate has reached pH = 7, and oven dried for 12 h.
The catalyst was prepared by incipient wetness impregnation, followed by carbothermal reduction.The dried CNF were impregnated with ammonium heptamolybdate hexahydrate (Sigma Aldrich ® ) in one step, and then oven dried for 12 h, followed by carburization in flowing hydrogen (100 mL/min) using a heating rate of 10 K/min up to 1023 K, maintained for two additional hours at this set point temperature.Then, the catalyst was cooled down to room temperature under pure nitrogen flow and passivated in 1% O 2 in N 2 (Air Liquid) for 2 h.
Phase composition and structure of supports and fresh catalyst were determined by X-ray Diffraction (XRD), in a D8 Advance Series 2 diffractometer (Bruker, Billerica, MA, USA) equipped with a Ni-filtered Cu Kα radiation and a secondary graphite monochromator, using a θ-2θ configuration.Bulk morphology was studied by Transmission Electron Microscopy (TEM) using a Jeol 2011 microscope (JEOL, Tokyo, Japan) with a LaB 6 gun operating at 200 KV, and coupled to Energy-Dispersive X-ray Spectroscopy (EDX).Crystallite size was determined by XRD and TEM.The textural properties were measured by N 2 adsorption at 77 K in a Micromeritics Tristar apparatus (Micromeritics Instrument Corp., Norcross, GA, USA).Specific surface area and pore volume were calculated by applying the BET method to the N 2 adsorption isotherm and the pore size distribution was calculated by the BJH method based on the desorption branch of the isotherm.Temperature Programmed Desorption (TPD) was performed to evaluate the CO and CO 2 eluted from pretreated (CNF) in an AutoChem II 2920 apparatus (Micromeritics Instrument Corp., Norcross, GA, USA) under constant Argon flow (50 mL/min) using a heating rate of 10 K/min up to 1273 K.The eluted gas was analyzed by mass spectroscopy.The total amount of CO and CO 2 released was calculated by integrating the area under the concentration versus volume curve.Oxygen content was determined from the amount of CO and CO 2 released.

Catalytic Testing and Product Characterization
The effects of operating conditions (temperature and pressure) and reaction time (2, 4 and 6 h) on the conversion of guaiacol and products selectivity were studied.
The catalytic tests were performed in a 100 mL high pressure batch reactor (Parker Autoclave Engineers, Erie, PA, USA).Prior each run, the autoclave reactor was loaded with 1.2 mL (3 wt %) of guaiacol in 40 mL of decane (97 wt %), and 0.2 g of catalyst.The reactor was purged at an ambient temperature first with pure nitrogen, and then with pure hydrogen.Afterwards, the reactor was loaded with 2 MPa of hydrogen (or 3 MPa in run 9), and heated to the reaction temperature while stirring at 300 rpm (in order to minimize reaction during heating).Once the reaction temperature has been achieved, the stirring was increased to 1000 rpm and the reaction time counting was initiated.The reaction was stopped by decreasing the stirring to 300 rpm and cooling the reactor with compressed air until ambient temperature has been achieved.Blank experiments both with and without CNF were performed in order to distinguish eventual catalytic effect of the bare CNF support from the effect of reaction temperature on conversion of guaiacol and selectivity to HDO products.
Conversion of guaiacol (X) and product selectivity (S) were calculated using Equations ( 1) and (2) as follows.
where C i is the initial concentration guaiacol (mmol/mL), C f is the final concentration of guaiacol in the reaction products (mmol/mL), and C pi is the concentration of product i (mmol/mL).The operating conditions used during the HDO experiments with the three synthetized catalysts are presented in Table S1.
After each experiment, the catalyst was separated by gravity filtration from the reaction products using a cellulose filter (0.2 µm porosity and 4 cm diameter, Whatman ® ).The so recovered catalyst was then rinsed with 50 mL pure ethanol, and the obtained filtrate was chemically analyzed aiming the detection of intermediate reaction compounds.The characterization of reaction products was performed by GC-FID (Perkin Elmer ® , Waltham, MA, USA) and GC-MS (Perkin Elmer ® ).For that, aliquots of the catalyst free liquid were collected using a syringe assembled with a filter (Whatman ® ) and injected directly in the chromatographic analyzers.

Conclusions
A Mo 2 C catalyst supported on commercial CNF was tested in the HDO of guaiacol, the major reaction products were cresol and phenol, followed by xylenols and toluene.The increase in the reaction temperature up to 623 K use of more severe operating conditions in the HDO of guaiacol caused a diversification in the reaction pathways, and consequently, in the selectivity to products, allowing for demethoxylation reactions being almost absent at 573 K.
The etching of carbonaceous materials involving molybdenum is reported in the present work for the first time, and is possibly related with the formation of molybdenum carbides by carburization of in hydrogen atmosphere.

Figure 3 .
Figure 3. Conversions and product selectivities of the main reaction products during the hydrodeoxygenation of guaiacol at 573 K over the bare CNF support and the Molybdenum carbide (Mo2C)/CNF catalysts after 2 h and 4 h of reaction.

Figure 3 .
Figure 3. Conversions and product selectivities of the main reaction products during the hydrodeoxygenation of guaiacol at 573 K over the bare CNF support and the Molybdenum carbide (Mo 2 C)/CNF catalysts after 2 h and 4 h of reaction.

Figure 5 .
Figure 5. Conversions and product selectivities of the main reaction products during the hydrodeoxygenation of guaiacol at 623 K over the bare CNF support the Mo 2 C/CNF catalyst after 2 band 4 h of reaction.

Figure 6 .
Figure 6.Conversion and product selectivities of the main reaction products during the hydrodeoxygenation of guaiacol over the Mo2C/CNF catalyst at 623 K, 2 h reaction time and 2 and 3 Mpa H2 pressure.

Figure 6 .
Figure 6.Conversion and product selectivities of the main reaction products during the hydrodeoxygenation of guaiacol over the Mo 2 C/CNF catalyst at 623 K, 2 h reaction time and 2 and 3 Mpa H 2 pressure.

Figure 7 .
Figure 7. X-ray diffraction patterns of fresh and spent catalysts after 4 h reaction tests at 573 and 623 K.

Figure 7 .
Figure 7. X-ray diffraction patterns of fresh and spent catalysts after 4 h reaction tests at 573 and 623 K.

Table 2 .
Mo2C crystallite size of the fresh and spent catalysts determined by X-ray Diffraction (XRD).

Table 2 .
Mo 2 C crystallite size of the fresh and spent catalysts determined by X-ray Diffraction (XRD).