Hydrodeoxygenation of Phenolic Compounds by Sulfided (Co)Mo/Al 2 O 3 Catalysts, a Combined Experimental and Theoretical Study Towards and

Résumé — Hydrodésoxygénation de composés phénoliques en présence de catalyseurs sulfurés (Co)Mo/Al 2 O 3 : une Abstract — Hydrodeoxygenation of Phenolic Compounds by Sulfided (Co)Mo/Al 2 O 3 Catalysts, a Combined Experimental and Theoretical Study — The hydrodeoxygenation of model phenol compounds (phenol and 2-ethylphenol) was carried over unpromoted Mo/Al 2 O 3 and promoted CoMo/Al 2 O 3 catalysts. Hydrodeoxygenation proceeds by two pathways:


INTRODUCTION
The development of new biofuels is an attractive alternative to decrease both the consumption of fossil fuels and the emission of carbon dioxide. The European Union is committed to use non-food cellulosic and lignocellulosic biomass as a sustainable alternative of bio-fuel production [1]. Liquids derived from pyrolysis of lignocellulosic biomass [2][3][4][5] may be used without competing with a growing global demand for agricultural commodities but they contain large amounts of oxygenates (25-45 wt%) having multiple functions like aldehydes, ketones, acids, esters and alcohols [2][3][4]. They lead to deleterious properties such as high viscosity, thermal and chemical instability [3,6] and poor engine efficiency. Hence, upgrading of bio-oils by reducing their oxygen content is required. Catalytic HydroDeOxygenation (HDO) to obtain oxygen-free molecules is one of processes that would allow to valorize these bio-oils.
Conventional sulfide catalysts used in hydrotreating processes such as (Co)Mo/Al 2 O 3 appear as good candidates for HDO processing [2,4,6,7]. The active phases of these catalysts consist of sulfide nano particles (MoS 2 or CoMoS) supported on high specific surface area alumina. These particles are made of MoS 2 sheets, the Co-promotor being localized on their (100) MoS 2 surface edges. The catalytic activity of these catalysts in hydrotreating reactions (HDS (hydrodesulfurization) or HDO) is usually related to the Coordinatively Unsaturated Sites (CUS sites) located on these edges, which accurate descriptions have been provided by a number of theoretical studies [8,9].
The aim of the present work is to determine the nature of the catalytic active sites and the mechanisms involved during the hydrodeoxygenation of two model phenolic compounds (phenol and 2-ethylphenol) over conventional (Co)Mo/Al 2 O 3 catalysts. To this aim, the reactivity of both compounds was evaluated at 613 K under 7 MPa, their interaction with the sulfide catalysts was monitored by IR spectroscopy and the reaction mechanisms were investigated using DFT modeling in order to rationalize experimental observations.

Activity Measurements
The particle size of the catalysts was in the range of 250-315 mm. The catalysts were diluted in carborundum to keep the volume of the catalyst bed constant. In order to obtain comparable conversions (close to 20%), 50 mg of the Co-promoted catalyst or 100 mg of the unpromoted catalyst were used.
All the catalysts were sulfided in situ into a high-pressure dynamic flow reactor (length: 40 cm; inner diameter: 1.25 cm) using a mixture containing 5.8 wt% of DMDS diluted in toluene. Under 4.0 MPa of total pressure, the sulfiding mixture was injected at a starting temperature of 423 K. After 1 hour, the temperature was raised to 623 K at a rate of 5 K.min -1 and it was maintained at this temperature for 14 hours. The temperature was then lowered to the reaction temperature (613 K).
The catalytic test was carried out under 7 MPa of total pressure using the oxygenated model compound (phenol or 2-ethylphenol), which was diluted in toluene. Their partial pressure was fixed to 49 kPa. DMDS was added to the feed to generate 30 kPa of H 2 S in order to preserve the sulfide state and to prevent deactivation of the catalysts [15]. The partial pressure of hydrogen was kept constant at 5.75 MPa.
The reactor effluents were condensed and liquid samples were periodically analyzed with a Varian 3300 chromatograph equipped with a DB1 capillary column (length: 30 m; inside diameter: 0.50 mm; film thickness: 0.25 mm) and a flame ionization detector. The oven temperature was first set during 5 min at 328 K, then raised to 448 K at a rate of 4 K.min -1 and kept at this temperature for 5 min. Considering a differential reactor (total conversion near 20%), the deoxygenation activity (HDO) was calculated using the following equation: where A HDO is the deoxygenation activity (in mmol.h -1 .g -1 ), X HDO the conversion into deoxygenated products (in mol%), F the molar flow of the oxygenated reactant (in mmol. h -1 ) and W the weight of sulfided catalyst (in g). The DDO, HYD and ACI (the ACI pathway involves the acidity of the catalyst) activities were calculated with the same formula. The conversion into deoxygenated products was replaced by the yield in aromatic products (benzene or ethylbenzene) for the determination of the DDO activity, by the yield in alkanes and alkenes compounds (ethylcyclohexane and ethylcyclohexenes for 2-ethylphenol, cyclohexane and cyclohexene for phenol) for the determination of the HYD activity and by the yield in oxygenated products for the determination of the ACI activity.

IR Spectroscopy
For the InfraRed (IR) study, the catalysts were grounded and pressed into self-supported wafers. After introduction in a low temperature IR cell, sulfidation was carried out in situ under a flow of H 2 S/H 2 (10/90) at 623 K during 2 hours and followed by an evacuation during 1 hour at the same temperature.
Calibrated doses of phenol or 2-ethylphenol were introduced at Room Temperature (RT) up to a final equilibrium pressure of 20 Pa. Then, desorption was carried out at RT for 10 min evacuation prior to record the IR spectra.
The impact of phenol and 2-ethylphenol on the accessibility of the CUS sites of the sulfide phase was assessed using CO as a probe molecule. To this aim, CO was adsorbed on the freshly sulfided catalysts well as after phenol and 2-ethylphenol adsorption. After cooling at 100 K, calibrated doses of CO were introduced in the IR cell up to an equilibrium pressure of 133 Pa. For comparison, the spectra were normalized to 5 mg.cm -2 of the sulfided catalysts.

Computational Methods
The Density-Functional Theory (DFT, PW91 [26]) calculations were performed with the Vienna Ab initio Simulation Package (VASP) [27] using the PAW method [28]. The solution of the Kohn-Sham equations was improved selfconsistently until a difference lower than 10 -5 eV was obtained between successive iterations. The calculations were performed with a cut-off energy of 450 eV and a Methfessel-Paxton smearing with s = 0.1 eV.
Throughout this work, we used a large super cell (12.641 ¥ 12.294 ¥ 25.000 Å 3 ), which contains four elementary MoS 2 units in the x and z direction and two layers along the y-axis (Fig. 1). A k-point mesh (3,1,1) has been chosen to give an accurate sampling of the supercell's Brillouin zone. Previous studies [29][30][31] showed that this model is suitable to predict the electronic and structural properties of the MoS 2 surface. A vacuum layer of 15 Å is located above the MoS 2 slab in the z direction to avoid interactions between slabs.
Since the molecules are adsorbed only on one side of the slab, dipolar corrections have been added in the z direction.
The two upper rows were allowed to relax until forces acting on ions are smaller than 3.10 -2 eV.Å -1 . The two lower rows were kept fixed to simulate bulk constraints [29][30][31]. The promoted catalyst is modeled by substituting of the Mo atoms on the S edge and half of the atoms on the metallic edge by Co (Fig. 1), as proposed in literature [32]. The adsorption energy is computed following the next equation: Positive adsorption energy corresponds to an exothermic adsorption.

Deoxygenation of Phenol and 2-ethylphenol over Mo/Al 2 O 3 and CoMo/Al 2 O 3
The transformation of phenol and 2-ethylphenol was carried out at 673 K under 7 MPa of total pressure over the unpromoted sulfide Mo/Al 2 O 3 catalyst and the Co-promoted catalyst (CoMo/Al 2 O 3 ). Over both catalysts, these oxygenated reactants are deoxygenated according to two pathways which are depicted in Figures 2 and 3. The first deoxygenation route involves a direct C-O bond scission (DDO pathway) yielding aromatic compounds: benzene from phenol and ethylbenzene from 2-ethylphenol. The second deoxygenation route, called the HYD pathway, leads probably, after hydrogenation of the aromatic ring, to an alcohol (cyclohexanol and 2-ethylcyclohexanol from phenol and 2-ethylphenol, respectively). These compounds are never detected, probably due to their easy dehydration into cycloalkenes. The dehydration of cyclohexanol yields only cyclohexene, whereas two isomers (1-ethylcyclohexene and 3-ethycyclohexene) are formed by the dehydration of 2-ethylcyclohexanol, the former is observed in higher quantities than the latter due to thermodynamics [14]. By hydrogenation, the corresponding alkane (cyclohexane or ethylcyclohexane) appears as the main product of the HYD pathway at conversion measured (close to 20%), whatever the sulfide catalyst used.
In the case of 2-ethylphenol, in addition to the two deoxygenation routes, a third pathway, called the ACI pathway (isomerization and disproportionation of ethylphenol), is also observed which mainly involves the acid properties of the catalyst support. This route leads to the formation of phenol, 3-ethylphenol and diethylphenol isomers. These products can also be deoxygenated, yielding benzene, cyclohexane, cyclohexene, diethylbenzenes and diethylcycloalkanes (Fig. 3). Table 2 reports the activity of both sulfide catalysts measured after 8 hours on stream. These results clearly indicate an inhibiting effect of the presence of the alkyl group in ortho position on the reactivity of the phenolic compound. Indeed, over both catalysts, phenol is 1.3 times more reactive in deoxygenation than 2-ethylphenol. These results are in accordance with those reported by Massoth et al. [12], which show that phenol was 1.5 times more reactive than 2-methylphenol over a CoMo/Al 2 O 3 catalyst.
In addition, our results show that the difference of reactivity of these phenolic compounds depends on the deoxygenation pathway over both catalysts. Indeed, while the ethyl group in ortho position improves the DDO reactivity, 2-ethylphenol being more reactive than phenol, it is detrimental for the HYD DDO OH OH Figure 2 Reaction network for the deoxygenation of phenol over sulfide catalyst. HYD pathway since 2-ethylphenol is about twice less reactive than phenol. This latter result is in agreement with the results reported by Aubert et al. [33] over a NiMo/Al 2 O 3 catalyst. As a consequence, the DDO/HYD selectivity is much higher for 2-ethylphenol deoxygenation (~1) than for phenol deoxygenation (~0.34) over CoMo/Al 2 O 3 (Tab. 2).
Moreover, Table 2 indicates a promoting effect of cobalt on deoxygenation of both phenolic compounds, since the Copromoted catalyst is about 2.9 more active in deoxygenation than the unpromoted Mo/Al 2 O 3 catalyst. Such a promoting effect was already reported in literature [14,15] and was attributed to a decrease of the metal-sulfur bond allowing an easier formation of sulfur vacancies (CUS), which are proposed as active sites for such a reaction [14,15,[21][22][23], while maintaining an optimal metal-heteroatom bond energy [34]. Moreover, we can underline that the effect of Co depends on the involved deoxygenation route: the DDO pathway is promoted in a larger extent than the HYD pathway, whatever the phenolic reactant. It is worth noticing that the reactivity of analogous sulfur containing compound significantly differs from that of phenolic compounds. In particular, thiophenol only reacts through the DDS pathway [35], preventing any comparison of the influence of the Co promotor on DDS(O)/HYD selectivities to be made. The differences on the reactivity of the two phenolic compounds investigated and on the Co promoting effect between the two deoxygenation pathways can be explained by several factors: -the involvement of different active sites in the two deoxygenation pathways [22,33]; -differences in the adsorption mode of the phenolic compounds (e.g. h 1 vs h 5 adsorption [14, 15,23], dissociative vs non dissociative adsorption).
In order to get more insight in these phenomena, the adsorption of the phenolic compounds with both catalysts was studied by IR spectroscopy. Figure 4 shows the IR spectra of phenol and 2-ethylphenol in CCl 4 (spectra 1a and 2a, respectively) and the IR spectra obtained after adsorption on Al 2 O 3 (spectra 1b and 2b) and CoMo/Al 2 O 3 (spectra 1c and 2c). The spectra obtained on Mo/Al 2 O 3 were similar to the latter and are not reported here. The assignment of the main IR bands of the phenolic compounds in solution, based on previous works [36,37] are reported in Table 3.

Infrared Characterization of Phenol and 2-ethylphenol Adsorption
The spectra obtained after phenol adsorption on Al 2 O 3 (1b) and on CoMo/Al 2 O 3 (1c) are qualitatively similar. They evidence the formation of phenate species characterized by bands at 1 597 cm -1 , 1 498-1 492 cm -1 (nCC ring ) and 1 295-1 271 cm -1 (nCC ring + nCO). These bands have identical positions and relative intensities on Al 2 O 3 and CoMo/Al 2 O 3 Figure 3 Reaction network for the transformation of 2-ethylphenol over sulfide catalyst.

HYD
DDO ACI OH OH OH OH OH but present much lower intensities on CoMo/Al 2 O 3 indicating that phenol mostly dissociates on the support but does not interact strongly with the sulfide phase. The very weak intensity of the band at ~1 470 cm -1 , assigned to nCC ring + dOH vibrations of phenol (Tab. 3) indicates that in these conditions (adsorption followed by evacuation at room temperature), most of phenol species are dissociated on the Al 2 O 3 support [38]. Spectra of 2-ethylphenol adsorbed on Al 2 O 3 (2b) and CoMo/Al 2 O 3 (2c) present bands at 1 597-1 578 cm -1 and 1 492 cm -1 (nCC ring ), 1 454 cm -1 (dCH + dOH) and 1 291, 1 263, 1 236 cm -1 (nCC ring + nCO). In contrast with phenol adsorption, the large intensity of the band at 1 454 cm -1 indicates that a significant fraction of 2-ethylphenol is molecularly adsorbed on both surfaces. Moreover, the increase in the intensity ratio between this band and the nCC ring band at 1 492 cm -1 on CoMo/Al 2 O 3 provides evidence for a higher fraction of molecularly adsorbed 2-ethylphenol on CoMo/Al 2 O 3 than on the pure Al 2 O 3 support. As a supplementary indication of this observation, one can note the increased intensity ratio between n(CO) bands at 1 263 cm -1 , 1 236 cm -1 (molecularly adsorbed form) and 1 291 cm -1 (2-ethylphenate) on CoMo/Al 2 O 3 . Figure 5 shows the IR spectrum of CO adsorbed on the freshly sulfided CoMo/Al 2 O 3 catalyst (spectrum a). It leads to the appearance of bands at 2 186 and 2 154 cm -1 corresponding to CO interacting with Lewis acid sites (Al 3+ ) and acidic AlO-H groups of the uncovered alumina support, a band at 2 110 cm -1 characterizing non promoted CUS Mo edge sites [39] and two bands at 2 070 and 2 052 cm -1 corresponding to promoted CUS sites in various environments [40].
CO adsorption on CoMo/Al 2 O 3 after phenol adsorption (Fig. 5, spectrum b) shows the complete disappearance of the band at 2 186 cm -1 (Al 3+ sites) and a large decrease of the band at 2 154 cm -1 (AlO-H hydroxyl groups). The nCO bands characterizing the sulfide phase are also markedly diminished, by nearly 60%. Similar but amplified effects are observed   (Fig. 5, spectrum c) with a decrease of the intensity of the bands characterizing the sulfide phase of about 80% with a downward shift to 2 105 and 2 058 cm -1 which is not observed on the case of phenol adsorption (Fig. 5, spectrum b). This slight shift specifically observed in the case of 2 ethyl phenol could result from lateral interactions between CO and 2-ethylphenol adsorbed on the sulfide phase. IR analysis of phenol and 2-ethylphenol adsorption on CoMo/Al 2 O 3 thus shows that while phenol mostly adsorbs dissociatively, a significant fraction of 2-ethylphenol adsorbs molecularly on the Al 2 O 3 support. The higher fraction of nondissociated 2-ethylphenol observed on the CoMo/Al 2 O 3 catalyst vs pure Al 2 O 3 also suggests a non-dissociative adsorption on the sulfide phase in our conditions. This difference in adsorption modes is in line with the lower acidity of 2-ethylphenol (pKa = 10.20) as compared to phenol (pKa = 9.95), explaining why the former dissociates in a lower extent. While this infrared study does not provide direct evidence for phenol adsorption on the sulfide phase in our conditions, one may suggest that phenol adsorption on the sulfide phase, if any, is also dissociative.
This study also shows a decrease in the accessibility of the sulfide phase: both bands characteristic of CO adsorption on unpromoted and promoted sulfide sites decrease, which can be explained by direct or indirect blockage of these sites. In the case of phenol, the analysis of the spectra (Fig. 4) shows the absence of molecularly adsorbed phenol, indicating that the decrease of the unpromoted and promoted sites of the sulfide phase (Fig. 5) might result from the dissociative adsorption of phenol on these sites. The larger decrease of the amount of remaining CUS sites detected by CO after adsorption of 2-ethylphenol (Fig. 5) parallels the increase in the fraction of molecularly adsorbed species (Fig. 4), suggesting that a part of this decrease is due to molecular adsorption of 2-ethylphenol on the CUS sites of the sulfide phase.
Finally, the decrease of the amount of CUS sites after adsorption of both phenolic compounds might also result from the presence of vicinal phenolate species on the support which might prevent the access of CO to the sulfide phase sites [38].
In order to get more insight on the adsorption behavior and reaction mechanisms of both phenolic compounds on the sulfide phase, DFT calculations have been carried out on non-promoted and promoted model surfaces.

DFT Modeling
DFT calculations were carried out on the metallic edges of non promoted MoS 2 and promoted CoMoS surfaces (50% Co substitution). Under HDO conditions, the presence of H 2 , H 2 S and H 2 O will modify the stoechiometry and the nature of the atoms on the edges of the catalyst active phase. For the reaction conditions used in the experimental studies, the most MoS 2 stable surface [21] is presented in Figure 1. This figure shows that the molybdenum edge is covered with sulfur atoms, yielding sixfold-coordinated Mo edge atoms [29,41]. Adsorption of large molecules will not be possible on this stable surface. The removal of one sulfur atom from the edge will lead to the formation of the active site. For the unpromoted catalyst, this departure is endothermic and the number of vacancy will be low. For the promoted catalyst, formally, one Mo-S group is substituted by one Co atom, creating stable vacancies on the edge. In this study, we consider a 50% substituted particle edge with two vacancies and two Mo = S groups per unit cell (Fig. 6) [32]. The reaction pathway of both molecules on these surfaces has been calculated for both DDO and HYD mechanisms.

Adsorption on Non-Promoted and Promoted Surfaces
Both non-dissociative and dissociative h 1 adsortptions of phenol and 2-ethylphenol were considered. The adsorption energies on the unpromoted phase are reported in Table 4, showing that the values for the non dissociative adsorption are similar for both molecules. The adsorption energies are slightly larger for 2-ethylphenol due to the inductive effect of the ethyl group. This increase demonstrates that the steric constraints are very small for an h 1 adsorption mode. The dissociation of the O-H group is possible on the unpromoted catalyst but the reaction is endothermic (17 kJ. mol -1 for phenol and 48 kJ.mol -1 for 2-ethylphenol). According to these values, equilibrium will be possible on the surface. The dissociation will be larger for the phenol in agreement with the pKa of both compounds and with the IR data reported above. DFT calculations show that the adsorption by the aromatic ring is also possible on the metallic edge with one vacancy. However, due to the steric constraint the adsorption energy are negative for both h 2 (E ads = -110 kJ.mol -1 ; Fig. 5c) and h 6 (E ads = -70 kJ.mol -1 ; Fig. 7d) phenol adsorption modes. These values show that such adsorption modes are unlikely and this suggests that h 1 adsorption will be involved in both DDO and HYD mechanisms on these edges.
In the case of promoted surfaces, we have considered the 50% promoted M-edge (Fig. 7) that exhibits stable CUS sites under typical hydrotreating conditions [39]. The h 1 adsorption energies are similar to the ones observed for the unpromoted catalyst. They are slightly greater on the CoMoS promoted edge (E ads = 37 kJ.mol -1 for phenol and E ads = 42 kJ.mol -1 for 2-ethylphenol), and the difference between both molecules is smaller than over the unpromoted system. Table 4 also shows that the dissociative adsorption is more difficult on the promoted catalyst but the M=S group is less basic than the Mo-S-Mo group. The h 6 adsorption mode is very unlikely on the CoMo catalyst since the adsorption energy is negative (E ads = -310 kJ.mol -1 ), the surface deformation due to steric repulsions being responsible for this very large negative value (Fig. 7).
The adsorption energies of phenol and 2-ethylpenol are significantly smaller than those previously reported of CO adsorption on similar surfaces [21,35,42]. This might explain why only few phenolic compounds adsorbed on the sulfide phase have been evidenced by IR spectroscopy. This also explains the inhibiting effect of CO on the HDO of phenolic compounds [13].

Direct Deoxygenation Mechanism
As DDO of phenolic compounds requires the activation of the C-O bond, the first step of this mechanism is the h 1 adsorption on CUS sites through the oxygen atom of these molecules. Our calculations show that the most efficient activation of the C-O bond is obtained when theses reactants are molecularly adsorbed on the CUS sites. The dissociation of the O-H group increases the C-O bond strength, as demonstrated by the evolution of the distance i.e. 1.39 Å for phenate and 1.42 Å for phenol. The same trend is observed on the CoMo catalyst (1.32 and 1.39 Å for phenate and phenol species, respectively).
The following step is the dissociative adsorption of H 2 on the surface, the adsorption of which being endothermic whatever the adsorption mode (Fig. 8). Indeed, the displacement of atomic hydrogen from one Mo atom to a S atom is an athermic process [41].
The rate determining step is the addition of the proton on the carbon atom bearing the OH group, leading to a Wheland complex. On the non-promoted surface, the activation energy required for this step is close to 95 kJ.mol -1 for 2-ethylphenol and 110 kJ.mol -1 for phenol. Such a difference can be explained by a better stabilization of the carbocation due to the inductive effect of the ethyl group in the case of 2-ethylphenol. The computed activation energies on the promoted catalyst are slightly reduced but rank similarly (90 kJ.mol -1 for 2-ethylphenol and 100 kJ.mol -1 for phenol). The effect of cobalt is slightly larger for phenol (-10 kJ.mol -1 ) than for 2-ethylphenol (-5 kJ.mol -1 ). Besides, these results are in line with the higher DDO reactivity of 2-ethylphenol observed experimentally on both catalysts (Tab. 2).  Finally, C-O bond scission occurs, which leads to the formation of the aromatic product and an OH group adsorbed on the surface, the last step of the catalytic cycle being the departure of the water from the surface. This step is endothermic since the adsorption of water on the vacancy is exothermic [23]. However, because of the low concentration of water in the gas phase, the water desorption should be thermodynamically favored (Fig. 8) [21].
The present mechanism differs from the mechanism proposed by Dupond et al. [43] for the HDO of aliphatic alcohols. In the later case, the rate determining step is a C-O/C-S bond exchange leading to the formation of a thiol.  However, the same exchange is unlikely for phenol molecules, for which the C-O bond is strengthened by the electron delocalization. The activation energy will thus be much larger than either the activation energy computed for aliphatic alcohol (143 kJ.mol -1 [43]) and or than the activation energy calculated for the phenyl ring protonation (110 kJ.mol -1 , Fig. 8).

Hydrogenation (HYD) Mechanism
As no h 6 adsorption is possible on the sulfur edge, a hydrogenation mechanism involving an h 1 adsorption through the oxygen atom is considered in the present work. In such a configuration, the first hydrogen addition should occur on the carbon atoms in b position with respect to the oxygen atom, because: -addition on the carbon in a position would result in the direct CO bond cleavage (DDO pathway); -the other carbon atoms are located at distances too large from the surface for such an attack. In this case, the first hydrogen addition consists in a proton attack from a sulfhydryl group in the vicinity of the b carbon atom of the aromatic ring (Fig. 9). Although a hydride addition cannot be completely ruled out, such a situation is less likely because of the larger distance between hydrides bonded to a metallic atom and the b carbon. This attack in b position results in the formation of a Wheland complex which energy was found to be higher for an attack on the tertiary carbon atom bearing the ethyl group of 2-ethylphenol (Fig. 9b) than on the secondary carbon atoms of either phenol or 2-ethylphenol (Fig. 9a). The difference between these two reaction energies is about 25 kJ.mol -1 . Hence, phenol should be statistically almost twice more reactive than 2-ethylphenol. This result is in very good agreement with our experimental results showing that the HYD reactivity of 2-ethylphenol is about twice lower than that of phenol (Tab. 2).
Finally, our calculations also show that the formation of the Wheland complex involved on the HYD route is always favoured (by more than 30 kJ.mol -1 ) when phenolic compounds are dissociatively adsorbed on the surface. This leads to an increased negative charge on the carbon atoms in b position, which favors the protonation of the aromatic ring as compared to the non-dissociative adsorption. As shown by both IR characterization (Fig. 5) and DFT calculations, the dissociative adsorption of phenol is favoured as compared to that of 2-ethylphenol, which also explains the higher HYD reactivity of phenol over both catalysts (Tab. 2).

CONCLUSIONS
The main findings of the present study may be summarized as follows: -2-ethylphenol is more reactive than phenol in the DDO pathway which involves the direct cleavage of the C-O bond but less reactive in the HYD pathway involving prior hydrogenation of the aromatic ring; -Co promotes both pathways in different extents, the DDO pathway being much more promoted that the HYD pathway; -IR spectroscopy shows that while phenol easily dissociates on these catalysts, adsorption of 2-ethylphenol leads to a larger amount of molecularly adsorbed species; -DFT calculations show that both pathways can occur on CUS sites, after h 1 adsorption and protonation on either the carbon atom bearing the OH group (DDO pathway) or the carbon atoms on b position with respect to the oxygen atom (HYD pathway), which allows us to explain our main experimental findings.