Journal of Catalysis 297 (2013) 176–186

Contents lists available at SciVerse ScienceDirect

Journal of Catalysis

journal homepage: www.elsevier .com/locate / jcat

Bio-oil hydrodeoxygenation: Adsorption of phenolic compounds on sulfided(Co)Mo catalysts

Andrey Popov, Elena Kondratieva, Laurence Mariey, Jean Michel Goupil, Jaafar El Fallah,Jean-Pierre Gilson, Arnaud Travert, Françoise Maugé ⇑Laboratoire Catalyse et Spectrochimie, ENSICAEN, Université de Caen, CNRS, 6 Boulevard Maréchal Juin, 14050 Caen, France

a r t i c l e i n f o a b s t r a c t

Article history:Received 11 July 2012Revised 11 September 2012Accepted 5 October 2012Available online 16 November 2012

Keywords:HDOPhenolGuaiacolSulfide catalystIR spectroscopyAdsorption modeDeactivation

0021-9517/$ - see front matter � 2012 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.jcat.2012.10.005

⇑ Corresponding author.E-mail address: francoise.mauge@ensicaen.fr (F. M

This paper reports the interaction of aromatic compounds (as phenol, ethylphenols, and guaiacol) repre-sentative of oxygenated functions of pyrolysis bio-oils, with sulfided (Co)Mo/Al2O3 catalyst in order todetermine the origin of catalyst deactivation in hydrodeoxygenation (HDO) reaction.

Infrared spectroscopy shows that all the studied phenolic compounds anchor on the alumina supportas phenate-type species, whereas only the most basic ones (2(4)-ethylphenol and guaiacol) interact alsowith the sulfide phase. At 623 K (typical temperature of the HDO reactions), only phenate species on thesupport are formed that is confirmed by study of the catalyst tested in HDO reaction. Phenate-type spe-cies anchored on alumina hinder the accessibility of the sulfide edge sites. This shows that the nature ofthe oxygenated compound (basicity and nature of the substituent) as well as the surface properties of thecatalyst support are crucial in the mode and extent of HDO active sites poisoning.

� 2012 Elsevier Inc. All rights reserved.

1. Introduction

The increase in crude oil prices and environmental concerns inlimiting carbon dioxide emissions implies the substitution of con-ventional fuels by new products issued of renewable sources. InEuropean Union, 20% of the conventional fuels should be replacedby alternative fuels in the road transport sector by 2020 [1]. Sincethe development of bio-fuels should avoid competition with hu-man food production, it is widely admitted that a sustainable routeis the use of lignocellulosic biomass issued from agricultural andwood residues [2–6]. From the various thermochemical processesdeveloped to produce liquids from solid lignocellulosic materials,fast-pyrolysis seems the most promising one because it does notneed neither high pressure nor H2 supply [7,8]. However, lignocel-lulosic-derived bio-oils resulting from fast-pyrolysis contain veryimportant amounts of oxygenated compounds (up to 45 wt% O)[2,4,9]. Resasco [10] underlines that these bio-oils include threemain families of compounds: (i) small acids, aldehydes and ke-tones, (ii) sugar-derived compounds, and (iii) lignin-derived phen-olics. Consequently, the main challenges for production of bio-fuelfrom pyrolytic bio-oils are the elimination of oxygen, while retain-ing the carbon in the product, and with minimum hydrogen con-

ll rights reserved.

augé).

sumption. In this aim, activity, selectivity, and stability of theHDO catalyst have to be optimized.

As recently reviewed, hydrodeoxygenation (HDO) appears as avery promising route that could produce bio-fuel at reasonableprices compare to fossil fuel [11]. The number of papers dealingwith HDO have considerably increased in the last 10 years[3,5,12–14]. As emphasized by Hicks [7], the most important issueis the development of highly selective and durable catalysts.

Hydrodesulfurization/hydrodenitrogenation (HDS/HDN) cata-lysts based on mixed sulfides of (Co, Ni) and (Mo, W) dispersedon high surface area supports like c-Al2O3 are interesting candi-dates for HDO reaction. Some papers report their use in HDO ofpyrolysis bio-oils [15,16]. But, as important quantities of oxygen-ated aromatic compounds are present in pyrolytic bio-oils [17],and as these molecules are known to be the most refractory ones,most of the HDO studies concentrate on oxygenated aromatics likeanisole, phenol, 2-methoxyphenol (guaiacol), and other phenoliccompounds.

It is generally reported that deoxygenation reaction occurs viatwo main routes: HYDrogenation (HYD) and Direct DeOxygenation(DDO) [18,19], this latter route allowing deoxygenation while min-imizing H2 consumption. Within the most recent papers, Bui et al.[20] shows that the DDO route involved in 2-methoxyphenol con-version is strongly increased versus HYD route after Co addition toMoS2/Al2O3 catalyst. Romero et al. [18] emphasize the increase inthe HDO rate with the addition of Co or Ni to Mo/Al2O3 for

Table 1Characteristics of the support and catalysts.

Al2O3 Mo/Al2O3 CoMo/Al2O3

Mo (wt%) – 9.9 9.2Co (wt%) – – 4.2SBET (m2 g�1) 252 258 243Average slab length (nm) – 2.4 2.6

A. Popov et al. / Journal of Catalysis 297 (2013) 176–186 177

2-ethylphenol HDO and point out a highest DDO/HYD selectivityfor CoMo/Al2O3. Finally, Senol et al. [21] shows that CoMo catalystexhibits higher activity than NiMo for phenol HDO. Therefore, theCoMo catalyst seems promising. Furthermore, on CoMo/ZrO2, theDDO/HYD selectivity is even better, showing that support changeis an interesting way for H2 economy [22].

However, deactivation of HDO catalyst is a key problem, anddeactivation mechanisms are still unclear. Deactivation can be re-lated to water effect (that can modify active phase or support), sin-tering of the active phase, or coking [23]. In our previous work [24],we showed that water effect is not the main factor in the deactiva-tion of phenol HDO, and that Co addition to Mo/Al2O3 allows to re-duce this effect.

Therefore, deactivation of HDO catalyst is likely related to activephase sintering or poisoning by carbon deposition. Previous studieson HDS or HDN catalysts show that carbon is deposited under theform of polyaromatic species and can block the active sites of thecatalyst [25]. Carbon deposition strongly depends on the type offeed. Due to their stronger interaction with the catalytic surface, al-kenes and aromatics have the largest affinity for carbon formationrelative to saturated hydrocarbons [11]. Besides, for oxygen-containing compounds, those with more than one oxygen atomappear to have a higher affinity for carbon formation by polymeri-zation reactions [25]. Indeed, it has been shown that the interactionof phenolic compounds with sulfided catalysts produces thesetypes of poison species [26–28]. Note that coking increases withincreasing acidity of the catalyst, while acid sites are required inthe mechanism of HDO [11].

The positive effect of H2S against deactivation has been shownin various papers [19,21,29–32]. H2S allows reactivating the sulfid-ed phase. H2S could also prevent poison species formation. But be-sides, too much of H2S is known to inhibit HDO reaction [33].

To specify the role of the various components of the catalyst inthe deactivation process, in our first work [34], we focused on theadsorption mechanisms of phenolic model compounds on oxidesusually used as supports for hydrotreatment catalysts. Therefore,the adsorption of phenol, anisole and 2-methoxyphenol on alu-mina, silica, and silica–alumina was studied by IR spectroscopy.The main results show that at room temperature (RT), while phe-nolic-type compounds mainly interact via H-bonding with silica,chemisorption is the main adsorption mode on alumina, probablydue to interaction of phenol with the Lewis acid–base pairs of alu-mina. At typical temperatures of HDO operating conditions(�673 K), the phenate species cover 2/3 of the alumina surface.These species are likely responsible for the deactivation of the cat-alysts supported on alumina.

To validate this hypothesis, we investigate the influence of phe-nolic-type molecules (phenol, ethyl-phenol, anisole, 2-methoxy-phenol) on supported sulfided (Co)Mo/Al2O3 catalysts. In thisaim, this paper focuses on (i) the identification of the adsorptionmodes of the oxygenated molecules on catalysts, (ii) the measure-ment of the accessibility of catalyst sites, (iii) the analysis of poisonspecies on spent catalysts, and (iv) the validation of these resultsby a quantitative model.

2. Experimental

2.1. Materials

The c-Al2O3 and (Co)Mo/Al2O3 catalysts were supplied by TO-TAL. Cobalt composition was measured by X-ray fluorescence spec-troscopy (XRF) and the molybdenum content by InductivelyCoupled Plasma (ICP) method. The origin and the properties ofthe oxides studied are given in Table 1.

Commercial oxygen-containing compounds phenol, ethylphe-nols, and 2-methoxyphenol were used (Prolabo for phenol and eth-

ylphenols with purity of 99.5 wt%, Johnson Matthey Company for2-methoxyphenol with purity of 98 wt%).

2.2. IR spectroscopy measurements

The IR experiments were performed using a FTIR Nicolet MagnaAEM spectrometer equipped with a DTGS detector. Each spectrumcorresponds to the accumulation of 64 scans at 4 cm�1 of resolu-tion using one zero filling level. Most of the spectra displayed cor-respond to difference spectra, that is, spectrum after adsorption (ordesorption) minus the spectrum of the corresponding activatedsample. All spectra are normalized to a constant mass (10 mg ofdried catalysts for a disk of 2 cm2).

Each sample was pressed into a self-supported wafer (2 cm2,between 10 and 20 mg but precisely weighed), under a pressureof 107 Pa. Activation and adsorption/desorption steps of the vari-ous experiments were performed in situ in the infrared cell accord-ing to the procedures described in the following sections.

2.2.1. Sulfidation procedureThe wafer was flushed under argon gas flow for 1 h at 423 K to

remove adsorbed water and air traces and further cooled down toRT. Then, the sample was sulfided by heating from 298 K (RT) up to623 K (ramp 3 K/min) under H2S/H2 (10/90) flow at 30 mL/min.After 2 h sulfidation at 623 K, the temperature was decreased toroom temperature. Then, an activation procedure was followed at623 K (5 K/min) for 30 min to remove H2S.

2.2.2. Phenolic compounds adsorptionOxygenated compounds (20 Pa) were introduced on the pellet

at RT or at 623 K maintained at this temperature for 15 min andfurther evacuated the adsorption temperature for 15 min.

2.2.3. CO adsorption2.2.3.1. Experimental protocol. In order to identify the catalyst sitesinvolved in the interaction of oxygenated molecules, CO adsorptionwas applied before and after oxygenated compound introduction.Firstly, CO was introduced at about 100 K at equilibrium pressureof 133 Pa on the sulfided catalyst. After desorption of CO at298 K, oxygenated compound (pressure of 20 Pa) was introducedon the sample at room temperature, followed by evacuation for15 min at the adsorption temperature. Then, CO adsorption wasperformed in similar conditions than previously. In a second step,oxygenated compound was adsorbed at 623 K following the sameprocedure. CO was again adsorbed in the same conditions thanpreviously.

2.2.3.2. Calculation of the site concentration. On sulfided CoMo/Al2O3, the adsorption of CO at �100 K gives rises to several bandscharacteristic of various species: m(CO/Al3+) at 2185 cm�1, m(CO/OH) at �2154 cm�1, m(COphysisorbed) at �2141 cm�1, m(CO/MoS2)at �2109 cm�1, m(CO/CoMoS) at �2070 and �2050 cm�1. To quan-tify these species, spectral decomposition is required. The m(CO)zone was fitted using Gaussian/Lorentzian curves. Area of eachband was then obtained using the fitting procedure of the OMNIC

178 A. Popov et al. / Journal of Catalysis 297 (2013) 176–186

software. The concentration of MoS2 and CoMoS sites was calcu-lated using the Beer Lambert law

Ai ¼ eini

S

where Ai is the band area (cm�1); ei, the molar extinction coefficient(cm mol�1); ni, the amount of adsorption sites in the disk (mol) andSi is the surface of the disk (cm2); the values of ei corresponding toCO adsorbed on MoS2 and CoMoS sites were determined earlier inour laboratory [35].

2.3. Analysis of catalysts after HDO test

IR analysis of catalysts tested in ethylphenol HDO by Romeroet al. [18,19] (IC2MP – Poitiers, France) was performed. Three sam-ples – ‘‘CoMo tested without H2S’’ and ‘‘CoMo tested with H2S’’ and‘‘sulfided CoMo catalysts’’ were characterized.

Before the HDO test, the catalysts were sulfided in situ using amixture of 5.8 wt% dimethyldisulfide (DMDS) in toluene, at 623 Kand under a 4.0 MPa total pressure. After lowering the temperatureto the reaction temperature (613 K), the oxygenated model com-pound 2-ethylphenol (2-EtPh) was introduced. 2-EtPh was dilutedin toluene and DMDS (2.1 wt% S) was added to generate H2S in situduring the reaction. The partial pressure of 2-EtPh was maintainedat 49 kPa, and the pressure of hydrogen was 5.8 MPa. At the end ofthe test, the reactor was depressurized and then cooled from reac-tion temperature to RT under nitrogen flow. Unloading and storagewere performed under Ar atmosphere.

For characterizing the catalyst sites accessible after HDO run,the sulfide phase sites have to be completely preserved from aircontact. Thus, the samples were pelletized inside a glove boxalthough this preparation is rather tricky. Pelletizing was success-ful for the catalyst tested without H2S, but not for the sampletested with H2S, in spite of repeated trials. Catalyst disk was trans-ferred without air contact into the IR cell, treated under vacuum at623 K. Then, CO adsorption (133 Pa) was performed at about 100 K.

Nevertheless, to obtain information on the nature of the poisonspecies formed during the HDO test, the catalyst tested with H2Swas pelletized under air atmosphere, transferred into the IR cell,and treated under vacuum at 523 K.

2.4. HRTEM measurements

HRTEM analysis of the sulfided catalyst was performed using aJEOL-2010F transmission electron microscope fitted with CDDcamera and an X-ray (EDX) spectrometer. The field emission gunwas operated at 300 kV. Before HRTEM analysis, the catalysts weresulfided in the same conditions as for IR experiments or sulfidedand tested in LACCO (in the case of spent catalysts). After treat-ment, the sample was transferred under argon in a glove box andgently crushed under argon. A drop of a suspension of the solidsample in n-butanol was deposited on a 300 mesh (diameter3 mm) ‘‘holey carbon film’’ grid and dried at 298 K under argonflow before introduction in the high vacuum chamber of the micro-scope. For each sample, stacking degrees and lengths of �1000MoS2 crystallites were measured.

3. Results and discussion

3.1. Adsorption of phenolic compounds on (Co)Mo catalysts

3.1.1. Adsorption mechanism of phenolic compoundsIn the following part, the IR spectra of adsorbed phenolic mole-

cules in the low frequency region, that is, 1650–1200 cm�1 are con-sidered. This zone includes the bands related to stretchingvibrations of aromatic ring, bending vibrations of OH, possibly

bending vibrations of the CH2 and CH3 groups and stretchingCAO vibrations.

3.1.1.1. Phenol adsorption on oxidic catalyst. The IR spectra of phenoladsorbed on pure Al2O3 and on non-sulfided MoOx/Al2O3 at RT arecompared in Fig. 1A. Bands at 1597 and 1498/1492 cm�1 corre-sponding to aromatic ring vibrations m(CC) are present in the spec-tra of both samples. For MoOx/Al2O3 (spectrum b), all these bandintensities are lower than those for the pure support (spectruma). The intensity of the shoulder at �1470 cm�1, which character-izes d(OH) vibrations of phenol, is negligible for the both samples.This fact indicates the formation of phenates instead of molecularphenol. Major differences can be observed in the 1200–1300 cm�1

region that characterizes m(CO) vibrations. A multi-component at1295/1271 cm�1 is present in the spectrum of pure Al2O3 (spec-trum a). On MoOx/Al2O3, this broad component is only present asa minor shoulder and the main band is observed at 1217 cm�1.After adsorption and evacuation at 623 K (Fig. 1B), all the bands re-main present for MoOx/Al2O3, but the band at 1217 cm�1, moresensitive to evacuation at 623 K than that at 1295 cm�1, is shiftedto 1234 cm�1. It can be concluded that phenol leads to phenatespecies on both pure Al2O3 and MoOx/Al2O3 samples as indicatedby the absence of d(OH) band. The new band (at 1217 cm�1) de-tected on MoOx/Al2O3 in the m(CO) region corresponds to phenatespecies adsorbed on molybdenum oxide phase. The marked inten-sity decrease in m(CC) bands on MoOx/Al2O3 in comparison withalumina support can be explained by hindered access of phenolto alumina due to support covering by oxide phase.

3.1.1.2. Phenol adsorption on sulfided catalyst. The IR spectra of phe-nol adsorbed at RT on Al2O3 and on sulfided CoMo/Al2O3 catalystare presented in Fig. 2A. The spectrum of phenol on the catalyst(spectrum b) is very similar to that obtained on pure alumina. For-mation of phenate species on sulfided catalyst is demonstrated bythe detection of bands at 1597, 1498/1492 cm�1 (m(CC)) and 1295/1271 cm�1 (vibrations with m(CO) component), as well as by theabsence of bands assignable to m(OH) vibration. Note that the phe-nate bands on CoMo catalyst have similar positions as those onalumina, but present lower intensities in comparison with the alu-mina support. The decrease in all the phenate bands that are obser-vable on sulfided catalyst in comparison with pure Al2O3 can beexplained by hindered access of phenol to alumina due to sulfidephase covering of the support. The same results were obtainedwith sulfided Mo/Al2O3 catalyst (spectra not shown).

Formation of phenate species on sulfided catalyst at RT givesrise to IR bands taking place at the same frequency as that on purealumina. It can be noticed that m(CC) and m(CO) phenate bands arequite sensitive to the nature of adsorbate as seen in the previoussection for MoOx, and in Xu et al. [36] for ZrO2. Indeed, this last pa-per shows that on ZrO2, phenate species have characteristic bandsat 1586, 1483 and 1282 cm�1 (corresponding to the bands ob-served on alumina at 1597, 1498/1492 and 1295/1250 cm�1). Thus,the similar frequencies for phenates anchored on alumina and onthe sulfide phase appears highly unlikely. This leads us to proposethat phenol adsorbed on the sulfided CoMo/Al2O3 at RT interactsonly with the alumina support and not with the sulfide phase.Moreover, we studied the effect of pre-adsorption of H2S on phenoladsorption (spectra not shown), since H2S can be dissociatively ad-sorbed on the Lewis sites of alumina (CUS Al3+, O) as well as of thesulfide phase (CUS Mo, S) [37,38]. Interaction of phenol with Lewispaired sites leads to the formation of phenate species [34]; thus,H2S – pre-adsorption should poison Lewis sites and change adsorp-tion of phenol on these phases. However, the spectral comparisonof phenol adsorbed on Al2O3 before and after H2S pre-adsorptiondoes not reveal any difference. Obviously, H2S is weakly adsorbedon Lewis sites of Al2O3 and can be easily replaced by O-donor mol-

Fig. 1. IR spectra of phenol adsorbed on Al2O3 (a), MoOx/Al2O3 (b) at RT (A) and at 623 K (B).

1470

1300 1400 1500 1600

0.1

1597 14981492

1597

1498

1295 1250

12951271

14501564

A B

1300 1400 1500 1600 (b)

(a)

cm -1

0.1

cm -1(b)

(a)

Fig. 2. IR spectra of phenol adsorbed on Al2O3 (a), sulfided CoMo/Al2O3 (b) at RT (A) and at 623 K (B).

A. Popov et al. / Journal of Catalysis 297 (2013) 176–186 179

ecules such as phenol. Regarding the CoMo/Al2O3, absence of effectof H2S pre-adsorption is also observed. In that case, the replace-ment of H2S in interaction with sulfide phase by O-donor mole-cules is rather unlikely due to the weaker interaction ofoxygenated molecules than sulfur containing molecule with thesulfide phase.

Spectra of phenol adsorbed on Al2O3 and sulfided CoMo/Al2O3

catalysts at 623 K are presented in Fig. 1.Fig. 2B shows that, at 623 K, in addition to the main m(CCring)

bands at 1597 and 1498 cm�1 and at 1295/1250 cm�1, some sup-plementary shoulders at 1564 and 1450 cm�1 appear on the sulfid-ed catalyst. Note that these bands are not detected on purealumina and on MoOx/Al2O3 (i.e., in the absence of sulfided phase)and are not formed on sulfided catalysts after phenol adsorption atRT. Apparently, these new species correspond to reaction of phenolwith sulfide phase and could correspond to the formation of carbo-naceous species, although the exact nature of these species is notyet established.

3.1.1.3. 2-Ethylphenol adsorption on sulfided catalyst. Spectra of 2-ethylphenol adsorbed at RT (Fig. 3A) on Al2O3 (spectrum a) and

A

Fig. 3. IR spectra of 2-ethylphenol adsorbed on Al2O3 (a) a

on sulfided CoMo/Al2O3 (spectrum b) present similar bands thatcan be assigned to: m(CC) bands at 1597 (with the shoulder at1578 cm�1) and 1492 cm�1, d(OH) + d(CH) band at 1454 cm�1,m(CO) bands at 1291 cm�1 (phenates) and 1263, 1236 cm�1

(molecular 2-ethylphenol). The intensity ratio between the1454 cm�1 band and the 1492 cm�1 band provides evidence for amore important fraction of molecular form in the case of sulfidedcatalyst than on pure support. As a supplementary indication ofthis observation, one can note the increased intensity ratio be-tween m(CO) bands at 1263, 1236 cm�1 (molecular form), and1291 cm�1 (phenates) on sulfided catalyst. Thus, at RT, 2-ethylphe-nol adsorption on CoMo/Al2O3 leads to the formation of 2-ethyl-phenate species as well as 2-ethylphenol in molecular form. Thelatter adsorbed species are present in larger quantities on CoMo/Al2O3 than on pure support. This fact can be attributed to weakinteraction of 2-ethylphenol with the sulfide phase through amolecular adsorption. Moreover, study of the H2S pre-adsorptionconfirms the interaction of 2-ethylphenol with the sulfide phase.Comparison of the spectra of 2-ethylphenol adsorbed on CoMo/Al2O3 before and after H2S pre-adsorption (spectra are not shown)indicates the decrease in the bands of molecularly adsorbed

B

nd sulfided CoMo/Al2O3 (b) at RT (A) and at 623 K (B).

180 A. Popov et al. / Journal of Catalysis 297 (2013) 176–186

2-ethylphenol but not of 2-ethylphenates. As reported [38], H2Scan be dissociatively adsorbed on sulfide phase by occupation ofits Lewis sites. Thus, H2S could poison Lewis sites of sulfide phaseand so decrease the adsorption of 2-ethylphenol on this phase.Hence, the decrease in the amount of molecularly adsorbed 2-eth-ylphenol after addition of H2S confirms that these species were ad-sorbed on sulfide phase. It should be mentioned that O-donormolecules less strongly interact with sulfide phase sites than withoxidic support ones (Al3+).

After adsorption of 2-ethylphenol at 623 K on CoMo/Al2O3

(Fig. 3B-b), only bands characterizing 2-ethylphenate species aredetected at 1597, 1492, 1450, 1297, and 1244 cm�1. Note thatthe new band at 1512 cm�1 present on Al2O3 at high temperature(spectrum a) is not formed on sulfided catalyst (spectrum b). Onsulfided catalyst, the broad shoulders at �1560 and �1460 cm�1

appear, pointing out the formation of new species in the presenceof sulfide phase. Similar shoulders were observed for phenol ad-sorbed at high temperature on sulfided catalysts (Figs. 1 and 2B).

3.1.1.4. Effect of the substituent on the adsorption of phenoliccompounds. The effect of the presence, nature, and position of sub-stituent in phenolic compounds on their adsorption on sulfidedCoMo/Al2O3 catalyst at RT and 623 K was studied by IR spectros-copy (spectra not shown). Table 2 summarizes the resultsobtained.

Regarding the adsorption at RT, IR spectroscopy points out thatphenol, guaiacol, and 3-ethylphenol interaction with sulfidedCoMo/Al2O3 leads to (methoxy, ethyl)-phenate formation withthe same frequencies as on pure Al2O3. The detection of similar fre-quencies for both phenates anchored on alumina and sulfide phaseappears highly unlikely. Therefore, we propose that in these condi-tions, phenol, guaiacol, and 3-ethylphenol only interact with thealumina support and not with the sulfide phase. Otherwise, theadsorption of 2-ethylphenol and 4-ethylphenol on support andon sulfided CoMo/Al2O3 at RT demonstrates ethylphenate forma-tion, as well as the presence of ethylphenol under molecular form.However, on the CoMo/Al2O3 catalyst, the fraction of molecularlyadsorbed species is greater than on the pure support where ethyl-phenates are predominant. This fact provides evidence for weakinteraction of (2-)4-ethylphenol with sulfide phase through molec-ular adsorption. H2S pre-adsorption decreases the amount ofmolecular adsorbed 2-ethylphenol on sulfided CoMo/Al2O3

whereas it does not lead to any change on Al2O3. This confirmsthe molecular interaction of 2-ethylphenol with the sulfide phase(Table 2). The difference between the adsorption modes of (2-)4-ethylphenol and the other oxygenates (phenol, guaiacol, 3-ethyl-phenol) might be explained by higher basicity of (2-)4-ethylphenol(pKa, Table 2). The donor effect of ethyl group in 2 and 4 positionsfacilitates the interaction of the oxygen atom of (2-)4-ethylphenolwith sulfide phase vacancies through donor–acceptor mechanism.

Table 2Species formed by adsorption of phenolic molecules on sulfided CoMo/Al2O3 catalyst.

Phenolic molecule pKa Tadsorption (K)

2-Ethylphenol 10.20 298623

4-Ethylphenol 10.00 298623

Phenol 9.95 298623

Guaiacol 9.93 298623

3-Ethylphenol 9.90 298623

Regarding the interaction at 623 K, the adsorption of all oxygen-ated molecules on sulfided CoMo/Al2O3 catalyst reveals (1) thepresence of (ethyl, methoxy-) phenate species on the support, (2)the absence of molecular forms, and (3) the formation of new spe-cies (X) characterized by the appearance of shoulders (�1550,1450 cm�1) assigned to the m(CC) bands that are not detected onpure alumina. These new species (X) correspond to reaction of oxy-genates with the sulfide phase or with species present on it, sincethey are only formed in the presence of sulfide phase. These spe-cies correspond to the formation of carbonaceous species, even ifthe exact nature of these species is difficult to establish.

The mechanism of interaction of the various oxygenates with sul-fided CoMo/Al2O3 catalyst at RT and 623 K is depicted in Scheme 1.Basicity of the oxygenated molecules and temperature of adsorptionappear as key parameters that determine the mechanism of interac-tion. Two series of oxygenated molecules can be distinguished at RT:(i) 3-ethylphenol, guaiacol, and phenol that only interact with thesupport forming (ethyl, methoxy-) phenates through dissociativemechanism (Scheme 1A), (ii) (2-)4-ethylphenol that interact bothwith the support and with the sulfide phase via dissociative mecha-nism or coordination (Scheme 1B). Interaction of all the molecules at623 K leads to the (ethyl, methoxy-) phenate formation on thesupport and some new species (X) formation, which correspond toreaction of oxygenates with the sulfide phase or with species presenton the sulfide phase (Scheme 1C).

3.1.2. Influence of phenolic compounds adsorption on the accessibilityof sulfided sites3.1.2.1. CO adsorption on sulfide catalyst. CO adsorption on sulfidedCoMo/Al2O3 (Fig. 4) gives rise to the m(CO) bands usually reportedin the literature [35,39–42]. The first doses of CO on the sulfidedCoMo/Al2O3 catalyst (spectra not shown) lead to the appearanceof two bands at 2110 and 2070 cm�1 together with a shoulder at2052 cm�1. Upon increasing the CO coverage, other bands at2186, 2154 cm�1 appear, whereas the intensity of the bandsat 2110, 2070, and 2052 cm�1 continues to increase. The bands at2186 and 2154 cm�1 are due to CO adsorption on the Al2O3 supportand correspond respectively to coordination on Al3+ Lewis acidsites (m(CO/Al3+)) and hydrogen bonding with the surface hydroxylgroups (m(CO/OH)). The three components at 2110, 2070, and2052 cm�1 correspond to CO adsorption on the sulfide phaseon either Mo edge sites of the sulfided MoS2 slab (2110 cm�1 –m(CO/MoS2)) or on promoted sites in various environments(2070, 2052 cm�1 – m(CO/CoMoS)) [42].

From the areas of CO/MoS2 and CO/CoMoS bands obtained byspectral decomposition (see experimental section) and taking intoaccount the previously determined molar absorption coefficientvalues for the bands of CO/MoS2 and CO/CoMoS species [35], theconcentration of unpromoted Mo sites and Co-promoted sites de-tected by CO adsorption can be calculated. Thus, on sulfided

Interaction with support Interaction with sulfide phase

2-Ethylphenates 2-Ethylphenol molecular2-Ethylphenates X = coke. . .

4-Ethylphenates 4-Ethylphenol molecular4-Ethylphenates X = coke. . .

Phenates –Phenates X = coke. . .

Methoxy-phenates –Methoxy-phenates X = coke. . .

3-Ethylphenates –3-Ethylphenates X = coke. . .

Scheme 1. Mechanisms of interaction of phenolic molecules with sulfided (Co)Mo/Al2O3 catalyst: (A) phenol at RT, (B) 2-ethylphenol at RT, (C) (ethyl)phenol at 623 K.

Fig. 4. IR spectra of CO (133 Pa at equilibrium) adsorbed on sulfided CoMo/Al2O3

before, after phenol adsorption at RT, after phenol adsorption at 623 K.

A. Popov et al. / Journal of Catalysis 297 (2013) 176–186 181

CoMo/Al2O3 catalyst, 116 lmol g�1 of sulfide sites is detected cor-responding to 59 lmol g�1 of Mo sites and 57 lmol g�1 of CoMoSsites.

3.1.2.2. Phenol and CO co-adsorption. After phenol contact at RTwith sulfided CoMo/Al2O3 catalyst (Fig. 4), comparison of CO spec-tra before and after oxygenates contact points out that phenolcompletely prevents the formation of the band at 2186 cm�1. Theintensity of the band at 2154 cm�1 is greatly decreased. Regardingthe sulfide phase, all the bands characteristics of CO uptake alsomarkedly diminish (60% of inaccessible sulfide sites – Table 3).The CoMoS band appears slightly shifted, but this effect can be ex-plained by the stronger diminution of the band at 2070 cm�1 thanthat of the component at 2052 cm�1. Taking into account the ab-sence of shift for Mo band at 2110 cm�1, this hypothesis is pro-posed instead of the shift of CoMoS bands. The effect of phenoltreatment at 623 K on the accessibility of the sites of sulfidedCoMo/Al2O3 is shown (Fig. 4). A decrease in the CO bands corre-sponding both to support (2186 and 2154 cm�1) and to sulfidedphase (region 2120–2000 cm�1) is observed. However, comparisonof the effect of phenol treatment at RT and 623 K shows a partial

Scheme 2. Mechanism for indirect poisoning of the sulfide sites by phenoladsorption on the support.

recovering of the accessibility of alumina support Lewis acidic sites(band at 2186 cm�1). In the same way, the accessibility of the sul-fided phase sites is slightly improved after phenol treatment at623 K (44% of the inaccessible sulfide sites at 623 K instead of60% at RT – Table 3). No selective poisoning of the different typesof sulfide sites is observed after high temperature treatment. Thesame results were obtained with sulfide Mo/Al2O3 catalyst (spectranot shown), with a similar decrease in accessible sulfide sites (40%)after phenol adsorption at 623 K.

After phenol contact at both RT and 623 K, CO uptake decreaseindicates a poisoning of the support (bands at 2186 cm�1 – m(CO/Al3+) and 2154 cm�1 m(CO/OH)) as well as of the sulfide phase(bands at 2110 cm�1 – m(CO/Mo) and 2069, 2052 cm�1 – m(CO/Co-MoS)). Regarding the support, the decrease in the accessibility ofAl3+ sites after phenate formation is in agreement with the disso-ciative adsorption of phenol on alumina that leads to the formationof AlAOAPh species (under mono or bidentate forms). The forma-tion of phenates on the support could also indirectly perturb theOH groups and consequently limit CO/OH interaction. As for thesulfide phase, a marked decrease in the accessibility of the sulfidesites is observed. This decrease directly depends on the amount ofphenate species as shown in Fig. 5 (when the amount of phenatesincreases, the accessibility of sulfided sites (MoS2 and CoMoS sites)decreases). This linear dependence indicates that phenate species(formed on the support) are involved in the poisoning of the sulfidesites. Our proposal is that an indirect poisoning of the sulfidedphase sites by phenate species anchored on alumina support oc-curs at RT (Scheme 2). The presence of phenate species on aluminain the vicinity of sulfide slabs would lead to a steric hindering ofthe sulfide sites (both Mo and CoMo sites). After phenol treatmentat 623 K, a partial recovering of the sulfide site accessibility is ob-served, which corresponds to evacuation of a fraction of phenatesfrom the alumina support.

The influence of phenol on the Mo/SiO2 catalyst has been inves-tigated (spectra are not shown), since the use of a less acidic sup-port can be a way to limit the formation of phenate speciesanchored on the support and thus preserve the accessibility ofthe active sites. Comparison of these results with phenol effecton Mo/Al2O3 shows that the accessibility of the sulfide phase is lessaffected by phenol adsorption on Mo/SiO2 than on Mo/Al2O3 cata-lyst (17% of inaccessible sulfide sites instead of 40%). This can berelated to the lower amount of phenates formed on Mo/SiO2

catalyst.

3.1.2.3. 2-Ethylphenol and CO co-adsorption. The spectrum of COafter 2-ethylphenol adsorption at RT shows hindered accessibilityof both the support and the sulfided phase sites (Fig. 6A, spectrumb). The bands of CO adsorbed on the support almost vanish.

Table 3Influence of phenolic molecule adsorption on the positions and areas of IR band of CO/sulfide sites on CoMo/Al2O3 catalyst.

Oxygenated molecule Tadsorption (K) Influence on the sulfide phase

Decrease in CO uptake (%) Shift of m(CO/MoS2) (cm�1)

2-Ethylphenol 298 �78 �5623 �46 �5

4-Ethylphenol 298 �77 �7623 �17 �7

Phenol 298 �60 0623 �44 0

3-Ethylphenol 298 �55 0623 �10 0

Guaiacol 298 �74 �10623 �63 �5

Fig. 5. Relationship between the areas of (m(CO/MoS2) + m(CO/CoMoS)) bands andphenol m(CC) band (at �1600 cm�1) on sulfided CoMo/Al2O3.

182 A. Popov et al. / Journal of Catalysis 297 (2013) 176–186

Regarding CO interaction with sulfide phase after 2-ethylphenoladsorption at RT, the intensities of the m(CO/MoS2) and m(CO/Co-MoS) bands are reduced by �78% and their maxima are shifteddownward (�5 cm�1) (Table 3). After 2-ethylphenol treatment at623 K (Fig. 6B, spectrum b), CO adsorption shows very weak acces-sibility of the support sites, whereas CO uptake on sulfide phase isimproved (46% of the inaccessible sulfide sites instead of 78% at RT(Table 3). Nevertheless, the shift of m(CO/MoS2) and m(CO/CoMoS)bands is still observed. Note that this shift is the same at RT and623 K.

Comparison of the spectra of CO after 2-ethylphenol and phenolcontact shows the same faction of poisoned sulfide sites at 623 K(about 45% of sulfide sites become inaccessible after treatment inboth ethylphenol and phenol), whereas at RT, the poisoning effectof 2-ethylphenol on sulfide phase is stronger than that of phenol(78% of the inaccessible sulfide sites after 2-ethylphenol contactand 60% after phenol contact – Table 3). 2-Ethylphenol adsorptionleads to the formation of 2-ethylphenates on the support, as wellas coordinated molecular 2-ethylphenol on the sulfide phase.Therefore, this direct adsorption on the sulfide phase, in additionto the indirect effect due to 2-ethylphenate species anchored on

2052

2110

2070

20582105

21862154

0.02

A

0

2000 2100 2200

(b)

(a)

Fig. 6. IR spectra of CO (133 Pa equilibrium) adsorbed on sulfided CoMo/Al2O3

the alumina support, can explain the stronger poisoning of the sul-fide sites by 2-ethylphenol than phenol. At 623 K, 2-ethylphenol isonly present under the form of 2-ethylphenate species on the sup-port. Hence, after adsorption of 2-ethylphenol and phenol at 623 K,the same amount of (2-ethyl)phenates is present and leads to thesame poisoning of the sulfide sites.

2-Ethylphenol interaction with CoMo/Al2O3 leads to some addi-tional modification of the sulfide phase. The m(CO/MoS2) and m(CO/CoMoS) bands are both shifted downward by 5 cm�1 for adsorptionperformed at RT and 623 K. This indicates that 2-ethylphenol leadsto a change of the electronic properties of the sulfide phase. Itshould be mentioned that no shift was observed in the case ofphenol.

3.1.2.4. Effect of the substituent in the phenolic compounds on COadsorption. Data issued from CO adsorption before and after con-tact of oxygenates with the sulfide catalyst provide evidence fortwo main effects: (1) decrease in the accessibility of the sulfidesites as shown by the intensity decrease in m(CO/MoS2) andm(CO/CoMoS) bands and (2) some change of the electronic proper-ties, as shown by the shift of the CO bands in some cases. As listedin Table 3, the extent of these effects can be related to the presence,nature, and position of the substituent of the phenolic compoundsas well as of the temperature of adsorption.

3.1.2.4.1. Accessibility of the sulfide phase after oxygenate adsorption.Phenol adsorption on sulfided CoMo/Al2O3 only gives rise to phe-nate formation on the support. The decrease in the accessibilityof sulfide sites linearly depends on the amount of phenate speciesanchored on the support (Fig. 5). Hence, an indirect poisoning ofthe sulfided sites by the presence of phenate species on the supportand in the vicinity of the sulfide slabs is proposed [43].

Like for phenol, 3-ethylphenol adsorption on sulfided CoMo/Al2O3 only forms phenate-type species on the support that inducea similar decrease in the accessibility of the sulfide sites (Table 3:

2186

2154

2052

2150

2110

2105

2070

2061

B

(b)(a)

.02

2000 2100 2200 cm-1

before (a), after (b) 2-ethylphenol adsorption at RT (A) and at 623 K (B).

A. Popov et al. / Journal of Catalysis 297 (2013) 176–186 183

at RT about 55% versus 60% of inaccessible sites). Note, that aftercontact with 3-ethylphenol at high temperature, a recovering of al-most all the sulfide sites is observed.

Adsorption of (2-)4-ethylphenol on sulfided CoMo/Al2O3 at RTforms both phenate-type species on the support and molecularform on the sulfide phase. (2-)4-Ethylphenol adsorption leads toa similar poisoning of the sulfide phase (Table 3: about 77% of inac-cessible sites). It should be mentioned that this poisoning is signif-icantly greater than that observed for phenol and 3-ethylphenol.Hence, this can be explained by the two forms of adsorption of(2-)4-ethylphenol at RT that lead to indirect poisoning (due to(ethyl)phenate species) on the support and also to direct poisoning(due to molecular form) on the sulfide phase.

Adsorption of 2-methoxyphenol at RT leads to a similar poison-ing of the sulfide phase than for (2-)4-ethylphenol (74% of inaccessi-ble sites). 2-Methoxyphenol interacts only with the support and notwith the sulfide phase, but the strong poisoning cannot be only ex-plained by an indirect mechanism due to the presence of phenolgroup. Hence, two features can account for the more marked poison-ing of the sulfide sites: (1) the stronger interaction of 2-methoxy-phenol with alumina at high temperature than (ethyl-)phenates[43] and (2) the presence of methoxy group on the aromatic ring thatcan perturb more strongly the sulfide sites accessibility than theethyl group. After treatment at 623 K, the strong poisoning effectof 2-methoxyphenol is hardly decreased, conversely to the (ethyl-)phenol molecules. This is in agreement with almost unchangedamount of methoxy-phenate-type species on the support.

As a general manner, the sulfide phase sites are more accessibleafter treatment at 623 K than after adsorption of oxygenates at RT.It should be noted that on sulfided CoMo/Al2O3 catalyst at 623 K,new species (X species) are formed. The question of the toxicityof the X species on the sulfide sites arises. As shown for phenolin Fig. 5, a linear dependency between the amount of phenatesand sulfide sites poisoning is evidenced. Thus, the contribution ofthe X species to the poisoning should be minor in the case of phe-nol adsorption at 623 K. Consequently, the location of these carbo-naceous X species on the edges of the sulfide slabs is very unlikely.Otherwise, the direct poisoning effect would be stronger than theindirect poisoning. So, we propose that these carbonaceous X spe-cies formed after high temperature treatment are located on thebasal planes of sulfide slabs or on the alumina support. Anyway,these carbonaceous species could not be the main reason of thecatalyst deactivation.

3.1.2.4.2. Electronic properties after oxygenate adsorption. It isknown [37,44] that on metallic and on sulfide phases, the positionof the CO band mainly depends on the extent of backdonation of d-electrons of the adsorption center. This means that the electronicproperties of the sites account for m(CO) wavenumber. The down-ward shift of CO maxima after oxygenate adsorption indicates anincrease in the electronic density on the sulfide phase sites (Ta-ble 3). However, adsorption of phenol and 3-ethylphenol doesnot lead to any changes of the electronic properties of the sulfidephase. By contrast, adsorption of (2-)4-ethylphenol and 2-methoxyphenol on the sulfided CoMo/Al2O3 catalyst at RT as wellas at 623 K leads to a downward shift of m(CO/MoS2) and m(CO/Co-MoS) bands. Thus, the effect on the electronic properties of the sul-fide phase depends on the temperature of adsorption as well as onthe nature of the oxygenate. For adsorption at RT, it follows the or-der: 2-methoxyphenol (�10 cm�1) > 4-ethylphenol (�7 cm�1) > 2-ethylphenol (�5 cm�1)� 3-ethylphenol � phenol (0 cm�1). For(ethyl)phenol adsorption at 623 K, the variation of m(CO/MoS2)and m(CO/CoMoS) wavenumbers is equal to that measured at RT,whereas for 2-methoxyphenol adsorption, the shift of m(CO/MoS2) and m(CO/CoMoS) bands is smaller (from �10 cm�1 up to�5 cm�1).

Modifications of the electronic properties of the sulfide phaseby oxygenate adsorption could have different origins:

� Direct effect: After (2-)4-ethylphenol adsorption, a shift of m(CO/MoS2) and m(CO/CoMoS) bands is observed. Since the same shiftis observed both after interaction at RT and 623 K when theadsorbed species amount is different, the modification of theelectronic properties of the sulfide phase cannot be due to thedirect interaction of these oxygenates with the sulfide phase.� Indirect effect: In previous studies, using the IR spectroscopy of

CO adsorption on sulfide catalysts supported on b-zeolite [45],amorphous silica alumina [46] and alumina modified by boron[37], it was found that the CO wavenumber shifts upwardswhen the acidity of the support increases indicating theenhancement of the electronic deficiency properties of the sul-fide phase. In the same way, Mg addition [37], or K addition[37], leads to a downward shift of m(CO/MoS2) and m(CO/CoMoS) bands on sulfided CoMo/Al2O3 catalyst, while the acidstrength of Lewis and Brønsted acid sites decreases. The m(CO/MoS2) band is shifted down by 3 cm�1 in the case of Mg addi-tion and by 7 cm�1 for K modification. Thus, the electronic prop-erties of the sulfide phase vary with the acidity of the support.

In the present case, all the oxygenates form phenate-type spe-cies on the alumina support by interaction with Al3+ sites. Theirstrong interaction with Lewis acid sites that decreases the supportacidity could explain the change of electronic properties of the sul-fide phase. However, not all the oxygenates lead to a change ofelectronic properties of the sulfide phase. Thus, some additionalfactors, besides the decrease in the support acidity, have to beconsidered.

� Oxygenate basicity: Table 2 shows that adsorption of oxygenateswith ethyl group in ortho- and para-positions have a greaterimpact on the shift of m(CO/MoS2) and m(CO/CoMoS) bands thanthat observed for oxygenate with ethyl group in meta-positionor pure phenol. Consequently, there is a relation between thebasicity of (ethyl)phenol (2-ethylphenol (10.20) > 4-ethylphe-nol (10.00) > phenol (9.95) > 3-ethylphenol (9.90)) and theextent of the shift.� Nature of substituent: Since the intermediate basicity of 2-

methoxyphenol (pKa: 9.93) does not explain the marked shiftof the m(CO/MoS2) band (�10 cm�1), the nature of substituentalso influences the interaction of oxygenated with the sulfidephase. Indeed, the lone electron pair of oxygen atom of methoxygroup of 2-methoxyphenol adsorbed in the vicinity of sulfideparticles is likely to interact stronger with the sulfide phasethan the ethyl group of ethylphenol. This proposal is in goodagreement with the great m(CO/MoS2) shift, as well as the dim-inution of m(CO/MoS2) shift (�5 cm�1) observed for 2-methoxy-phenol contact at 623 K with respect to RT. Indeed, our previousstudy [34] of 2-methoxyphenol interaction with alumina showsthat at RT, methoxy phenate is formed, while at 623 K, a fractionof methoxy groups reacts with alumina leading to the formationof double anchored phenate species and methoxy grafted onalumina. Therefore, the lone electron pair of the oxygen atomof methoxy group cannot perturb any more the sulfide phase.

In summary, adsorption of (2-)4-ethylphenol and 2-methoxy-phenol on the sulfided CoMo/Al2O3 catalyst changes the electronicproperties of the sulfide sites, whereas adsorption of 3-ethylphenoland phenol does not lead to any changes. It should be mentionedthat the electronic properties are more sensitive to the presenceof phenate-type species on the support than to the direct interac-tion of the oxygenates with the sulfide phase. The effect of the oxy-genated molecules on the electronic properties of the sulfided

A1320

cm-11200 1300 1400 1500 1600

(RT)

(523 K)

(623 K)

1263

1597

1492

1460

12361578

0.1

B1260

1590

After HDO test

2186

cm-12050210021502200

21542110

20702052

Before HDO test

0.02

2070

21402154

Fig. 7. (A) IR spectra of sulfided CoMo/Al2O3 after HDO test without H2S: catalyst evacuated at RT, 523 K, and 623 K. (B) IR spectra of CO (133 Pa equilibrium) adsorbed onsulfided CoMo/Al2O3 before and after test without H2S.

Fig. 8. Comparison of IR spectra of sulfided CoMo/Al2O3 after HDO test without H2S(a) or with H2S (b). After HDO run, the catalysts are post-treated at 523 K in the IRcell.

184 A. Popov et al. / Journal of Catalysis 297 (2013) 176–186

phase depends on the basicity of the oxygenate that changes theacidity of the support, as well as the nature of the substituent.The methoxy group of 2-methoxyphenol adsorbed on alumina inthe vicinity of the sulfide phase should interact more strongly withthe sulfided sites than the ethyl group of ethylphenol.

3.2. CoMo catalysts after 2-ethylphenol HDO

Results of the previous section show that, for model phenolicmolecules, the species responsible of active site poisoning are phe-nate-type species anchored on the support. In order to confirm thenature of poison species, catalysts after HDO reaction wereanalyzed.

Since catalyst deactivation is influenced by the presence of H2Sduring the HDO run, surface species were analyzed on CoMo/Al2O3

catalysts tested with or without H2S addition. The activation andtest procedure as well as condition of storage and pelletizing be-fore the IR analysis are described in the experimental section.

3.2.1. Catalyst tested without H2S addition3.2.1.1. Characterization of spent catalyst. The catalyst tested in 2-ethylphenol HDO without added H2S (CoMo spent without H2S)is characterized by IR spectroscopy after evacuation at differenttemperatures. Stretching CH bands at 3100–3000 cm�1 and3000–2800 cm�1 specific of aromatic and aliphatic groups respec-tively are detected (spectra not shown). In the 1700–1200 cm�1 re-gion (Fig. 7A), spectrum (a) exhibits bands at 1597, 1578, 1492 and1460 cm�1, which are close to those observed for 2-ethylphenol(Fig. 3). The detection of bands at 1263 and 1236 cm�1 indicatesthe presence of molecular 2-ethylphenol as discussed for aluminain Section 3.1.1.3. Increase in the evacuation temperature up to623 K leads to changes in this zone. The bands at 1263 and1236 cm�1 are eliminated, indicating the disappearance of molec-ular 2-ethylphenol species. Even if the bands between 1300 and1240 cm�1 are too broad to be assigned for sure to ethylphenatespecies, variation of the intensity ratio of the m(CC) band at1492 cm�1 versus the [d(OH) + d(CH)] band at 1460 cm�1 with in-creased evacuation temperature confirms this point. Hence, IRanalysis points out the presence of ethylphenate species on ‘‘CoMospent without H2S’’ catalyst.

It could be surprising to observe the presence of molecular 2-ethylphenol on the catalyst surface after the HDO reaction. Thisis likely due to residual 2-ethylphenol in the catalytic reactor thatre-adsorbs on the catalyst surface at the end of the test during thecooling period.

3.2.1.2. CO adsorption on the tested catalyst. The purpose of theseexperiments was to characterize the accessibility of the sulfidephase of the ‘‘CoMo spent without H2S’’ catalyst. Pelletizing a com-plete disk inside the glove box was very difficult. Consequently, the

quality of the spectra is poor and their quantification is difficult.Fig. 7B compares spectra of CO adsorbed on the spent CoMo/Al2O3 without H2S and fresh sulfide catalyst. After CO contact, verylow CO uptake is detected on spent CoMo/Al2O3: only bands at2154 cm�1 (m(CO/OH)), 2140 cm�1 (physically adsorbed CO) anda low intense band at 2070 cm�1 (CO adsorbed on sulfided phase:CoMoS sites) are formed. Even if the characterization of the sulfidephase of spent catalyst is particularly tricky, the CO spectrum re-veals that the accessibility of both MoS2 and CoMoS sulfide sitesis strongly reduced after HDO reaction.

3.2.2. Catalyst tested with H2S additionIR spectrum of ‘‘CoMo spent with H2S’’ catalyst is presented in

Fig. 8b. It shows a very intense band at 1094 cm�1 that was not ob-served on the ‘‘CoMo spent without H2S’’ catalyst (spectrum a).This new band at 1094 cm�1 indicates sulfate species formation.However, according to the literature [47], a second band at1380 cm�1 characterizing (S@O) groups must be also present. Sincethe S@O species is very sensitive to water, some contaminationwith H2O during the unloading of the HDO reactor and/or pelletiz-ing could explain its absence. Indeed, after evacuation of the spent‘‘CoMo spent with H2S’’ catalyst for long duration (1 h) at 623 K,the two sulfate species bands (m(SAO) at 1090 cm�1 as well asm(S@O) at 1367 cm�1 (spectra not shown) are observed. Sincethese species are not observed on ‘‘CoMo spent without H2S’’ cata-lyst, the sulfate species are likely formed during pelletizing byreaction of sulfur residue with atmospheric oxygen. Consequently,probing the accessibility of catalyst by CO adsorption has not beendone since it would not be meaningful.

It should be noted that the comparison of the intensity of them(CC) band at 1492 cm�1 between spectra (a and b) provides evi-dence that the presence of carbonaceous species is higher (aboutfactor 3) on ‘‘CoMo spent without H2S’’ than on ‘‘CoMo spent withH2S’’. This is in line with the amount of carbon detected by elemen-tal analysis (respectively 9 and 3.3 wt%).

Table 4Coverage of alumina surface by the sulfide slabs (C) and sulfide phase interslab distance (d) calculated by methods 1 and 2.

Method MoS2 slabs CoMoS slabs

Alumina coverage (%) Interslab distance (nm) Alumina coverage (%) Interslab distance (nm)

Method 1 24 2.6 40 1.6Method 2 30 2.0 42 1.4

A. Popov et al. / Journal of Catalysis 297 (2013) 176–186 185

3.2.3. ConclusionOn the spent catalysts, IR analysis mainly detected three spe-

cies: molecular 2-ethylphenol (due to reactant re-adsorption afterreaction), sulfate species (formed during the pelletizing under air),and 2-ethylphenate species. The two first ones are contaminants;only the last one is responsible of catalyst deactivation. This con-firms the role of ethylphenate species as a poison that was sus-pected from the study in model conditions.

Fig. 9. Model for phenol adsorption in the interslab zone of sulfide catalyst.

3.3. Modeling of the effect of phenol adsorption on the sulfide sitesaccessibility

Experiments done with model oxygenates as well as after HDOrun supports the hypothesis of an indirect poisoning, that is, thatphenate species on alumina screen the edge sites of sulfide slabsleading to a poisoning of the catalyst. To validate this conclusion,a model is developed in order to quantify the screening effect ofphenate on support.

As previously shown for phenol adsorption (Fig. 5), the accessi-bility of sulfide sites is linearly dependant on the amount of phen-ates adsorbed on the support:

C(CO/sulfide phase) = K C(Ph/alumina) with C(CO/sulfidephase) = concentration of CO adsorbed on the sulfided phase,and C(Ph/alumina) = concentration of phenate species onalumina.

Considering the molar absorption coefficients of the phenateband at 1600 cm�1 (ecc = 5 lmol�1cm [34]) and that of CO in inter-action with sulfide sites (CO/MoS2 at 2115 cm�1, eco = 16 lmol�1cmand CO/CoMoS at 2065 cm�1, eco = 43 lmol�1cm [35]), and the rela-tionship presented Fig. 5, the K factor is calculated. K value is equalto 0.23. This value indicates that about 4 phenate species on alu-mina are necessary to poison one site of the sulfide phase. This ratioseems rather low. Probably, it is because only few surface phenatesare close enough to sulfide slabs for screening the active sites.

To clarify this point, we attempted to model the surface of thesulfide CoMo catalyst in order to quantify the fraction of phenateon alumina and their vicinity with sulfide slabs. Therefore, ourmodeling consists of three stages: (a) calculation of the aluminacoverage by sulfide slabs, (b) determination of the average distancebetween sulfide slabs, and (c) assessment of the amount of phenatelocated close enough from sulfide slabs to poison their accessibility.Two methods were used to determine the covering of alumina bysulfide phase: theoretical (method 1) and experimental (method 2).

– Method 1 is based on the calculation of the surface occupied bysulfide slabs determined from an hexagonal model of the sulfideslabs, considering a slab stacking of one and taking into accountof the amount of Mo and Co atoms determined by chemicalanalysis.

– Method 2 is based on the analysis of the IR spectra of phenoladsorbed on Al2O3 and on sulfided CoMo/Al2O3 and consideringmolar absorption coefficients determined previously [34,35].

The surface coverage of alumina by sulfide phase and the aver-age distances between sulfide slabs of the Mo and CoMo catalysts

are calculated according to the two methods (Table 4). The calcu-lations are detailed in Supplementary material.

Table 4 shows that, for each catalyst, there is a good agreementbetween the two methods. Note that these calculations clearlypoint out differences between CoMo and Mo catalysts since thecoverage of alumina by sulfide CoMo slabs is greater than byMoS2 slabs (41% instead of 27%), which induces a distance betweensulfide slabs smaller on CoMo than on Mo catalyst (1.5 nm insteadof 2.3 nm).

Taking into account the interslab distance calculated previouslyfor CoMo catalyst (1.4 nm in method 2), the size of phenate species(dPh = 0.64 nm) and that of CO molecule (0.44 nm), if a phenatespecies is located at equal distance of two sulfide slabs, it only re-mains 0.38 nm between the phenate species and each closestneighbor sulfide slabs (Fig. 9). Thus, in that case, one phenate pre-vents adsorption of two CO molecules on sulfide sites. The locationof phenate at a different position between the two slabs (at morethan 0.06 nm of the middle of the interslab) lets CO moleculeadsorption on one sulfide slab, whereas the opposite sulfide sitesslab is still screened. So, each phenate blocks one or two sulfideslab sites from CO adsorption. Accurate calculations based on slabsize distribution indicate that one phenate hinders the adsorptionof 1.35 molecules of CO (screening effect = 1.35).

Comparison between theoretical screening effect and thatdetermined experimentally (Fig. 5) shows strong difference. In-deed, experimental screening effect determines from phenol andCO adsorption (K = 0.23) is clearly smaller than the calculatedone (1.35).

The smaller experimental value can be explained by the lowamount of CO adsorbed on sulfide sites compared to the concentra-tion of edge sites that can be calculated from geometrical model,length of sulfide slab (determined from HREM), and amount ofMo and Co (determined by chemical analysis). Indeed, in the pres-ent study, CO adsorbed on the CoMo catalyst detected116 lmol g�1 sulfide sites of edge sites (59 lmol g�1 of Mo sitesand 57 lmol g�1 of CoMoS sites, Fig. 4a), whereas the amount ofedge sites calculated by equ. 1 is about 500 lmol g�1. Hence, COadsorption only probes 23% of the total sulfide sites. Taking into ac-count this limitation, the screening effect determined experimen-tally reaches 31%. Hence, the good agreement between

186 A. Popov et al. / Journal of Catalysis 297 (2013) 176–186

experimental and calculated screening effect shows that one phe-nate species can block at least 1.3 active sites of sulfide phase ofthe CoMo/Al2O3 catalyst. Consequently, this quantitative analysissupports the proposal of the indirect poisoning of active sites byphenate species adsorbed on the support.

4. Conclusion

The study of the interaction of oxygenated aromatic compounds(as phenol, ethylphenols, and guaiacol) with sulfided (Co)Mo/Al2O3

catalyst shows that basicity of the phenolic molecules as well asnature of the substituent are key parameters that determine theadsorption mechanisms.

All the phenolic compounds anchor on the alumina support asphenate-type species, whereas only the most basic phenolic mole-cules (as 2 and 4 ethylphenol) and guaiacol interact also with thesulfide phase. At temperature typical of the HDO operating condi-tions (�673 K), only phenate species on the support are detected.

As expected, oxygenated compounds adsorbed on the sulfidephase poison the sulfide sites. But, phenate-type species anchoron the alumina also hinder the accessibility to the sulfide edgesites. Modeling of the catalyst surface quantitatively confirms thescreening effect of these phenate species. Characterization of cata-lysts tested in ethylphenol HDO confirms the role of poison of theethylphenate species that was suspected from the model conditionstudy. Hence, this paper points out that the nature of the oxygen-ated aromatic compound as well as the surface properties of thecatalyst support are determining in the mode and extent of HDOactive sites poisoning.

Acknowledgments

Thanks to Prof. Xavier Portier from CIMAP (ENSICAEN, France)for performing the HREM observations.

Thanks to Dr. Yilda Romero, Dr. Frederic Richard, and Dr. Syl-vette Brunet from IC2MP (University of Poitiers, France) for provid-ing the spent catalysts as well as for all the stimulating discussions.

This work has been performed within ECOHDOC, a joint projectbetween CNRS, Universities of Caen, Lille and Poitiers and Totalfunded by the ‘‘Agence Nationale de la Recherche’’ (France).

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jcat.2012.10.005.

References

[1] 25 February 2003. The Promotion of the Use of Biofuels or Related OtherRenewable Fuels for Transport. EU Directive.

[2] A.V. Bridgwater, G.V.C. Peacocke, Renew. Sust. Energy Rev. 4 (2000) 1–73.[3] C.P. Briens, Jan, Berruti, Franco Int. J. Chem. React. Eng. 6 (2008) R2.[4] G.W. Huber, S. Iborra, A. Corma, Chem. Rev. 106 (2006) 4044–4098.[5] J.C. Serrano-Ruiz, J.A. Dumesic, Energy Environ. Sci. 4 (2011) 83–99.

[6] M. Stöcker, Angew. Chem. Int. Ed. 47 (2008) 9200–9211.[7] J.C. Hicks, J. Phys. Chem. Lett. 2 (2011) 2280–2287.[8] D. Mohan, C.U. Pittman, P.H. Steele, Energy Fuels 20 (2006) 848–889.[9] Q. Zhang, J. Chang, T. Wang, Y. Xu, Energy Convers. Manage. 48 (2007) 87–92.

[10] D.E. Resasco, J. Phys. Chem. Lett. 2 (2011) 2294–2295.[11] P.M. Mortensen, J.D. Grunwaldt, P.A. Jensen, K.G. Knudsen, A.D. Jensen, Appl.

Catal. A: Gen. 407 (2011) 1–19.[12] D.C. Elliott, Energy Fuels 21 (2007) 1792–1815.[13] E. Furimsky, Appl. Catal. A: Gen. 199 (2000) 147–190.[14] J. Zakzeski, P.C.A. Bruijnincx, A.L. Jongerius, B.M. Weckhuysen, Chem. Rev. 110

(2010) 3552–3599.[15] J. Wildschut, F.H. Mahfud, R.H. Venderbosch, H.J. Heeres, Ind. Eng. Chem. Res.

48 (2009) 10324–10334.[16] S. Zhang, Y. Yan, T. Li, Z. Ren, Bioresour. Technol. 96 (2005) 545–550.[17] D. Ballerini, Les Biocarburants, Paris, France, 2006.[18] Y. Romero, F. Richard, S. Brunet, Appl. Catal. B: Environ. 98 (2010) 213–223.[19] Y. Romero, F. Richard, Y. Renème, S. Brunet, Appl. Catal. A: Gen. 353 (2009) 46–

53.[20] V.N. Bui, D. Laurenti, P. Afanasiev, C. Geantet, Appl. Catal. B: Environ. 101

(2011) 239–245.[21] O.I. Senol, E.M. Ryymin, T.R. Viljava, A.O.I. Krause, J. Mol. Catal. A: Chem. 277

(2007) 107–112.[22] V.N. Bui, D. Laurenti, P. Delichère, C. Geantet, Appl. Catal. B: Environ. 101

(2011) 246–255.[23] J. Wildschut, PhD thesis, Pyrolisis Oil Upgrading to Transportation Fuels by

Catalytic Hydrotreatment, Rijksuniversiteit Groningen, 2009.[24] M. Badawi, J.F. Paul, S. Cristol, E. Payen, Y. Romero, F. Richard, S. Brunet, D.

Lambert, X. Portier, A. Popov, E. Kondratieva, J.M. Goupil, J. El Fallah, J.P. Gilson,L. Mariey, A. Travert, F. Maugé, J. Catal. 282 (2011) 155–164.

[25] E. Furimsky, F.E. Massoth, Catal. Today 52 (1999) 381–495.[26] M. Badawi, S. Cristol, J.-F. Paul, E. Payen, C. R. Chim. 12 (2009) 754–761.[27] M. Badawi, J.-F. Paul, S. Cristol, E. Payen, Catal. Commun. 12 (2011) 901–905.[28] E. Laurent, A. Centeno, B. Delmon, Stud. Surf. Sci. Catal. 88 (1994) 573–578.[29] C. Bouvier, Y. Romero, F. Richard, S. Brunet, Green Chem. 13 (2011) 2441–

2451.[30] A.Y. Bunch, X. Wang, U.S. Ozkan, J. Mol. Catal. A: Chem. 270 (2007) 264–272.[31] I. Gandarias, V.L. Barrio, J. Requies, P.L. Arias, J.F. Cambra, M.B. Güemez, Int. J.

Hydrogen Energy 33 (2008) 3485–3488.[32] T.R. Viljava, R.S. Komulainen, A.O.I. Krause, Catal. Today 60 (2000) 83–92.[33] J. Leglise, G. Van, C.J. Duchet, Promotion and Inhibition by Hydrogen Sulfide of

Thiophene Hydrodesulfurisation Over a Sulfide Catalyst, Royal Society ofChemistry, Cambridge, ROYAUME-UNI, 1994.

[34] A. Popov, E. Kondratieva, J.M. Goupil, L. Mariey, P. Bazin, J.P. Gilson, A. Travert,F. Maugé, J. Phys. Chem. C 114 (2010) 15661–15670.

[35] F. Maugé, J.C. Lavalley, J. Catal. 137 (1992) 69–76.[36] B.Q. Xu, T. Yamaguchi, K. Tanabe, Mater. Chem. Phys. 19 (1988) 291–297.[37] W. Chen, Effect of B and Mg Addition on Structure and Selectivity of Sulfide

Catalyst, Ph.D. Thesis, Université de Caen Basse Normandie, Beijing (Chine),2009.

[38] A. Travert, F. Maugé, in: G.F. Froment, B. Delmon, P. Grange (Eds.), StudiesSurface Sciences and Catalysis, vol. 127, Elsevier, 1999, pp. 269–277.

[39] C. Dujardin, M.A. Lelias, J. van Gestel, A. Travert, J.C. Duchet, F. Maugé, Appl.Catal. A: Gen. 322 (2007) 46–57.

[40] F. Maugé, A. Vallet, J. Bachelier, J.C. Duchet, J.C. Lavalley, J. Catal. 162 (1996)88–95.

[41] A. Travert, C. Dujardin, F. Maugé, S. Cristol, J.F. Paul, E. Payen, D. Bougeard,Catal. Today 70 (2001) 255–269.

[42] A. Travert, C. Dujardin, F. Maugé, E. Veilly, S. Cristol, J.F. Paul, E. Payen, J. Phys.Chem. B 110 (2006) 1261–1270.

[43] A. Popov, E. Kondratieva, J.-P. Gilson, L. Mariey, A. Travert, F. Maugé, Catal.Today 172 (2011) 132–135.

[44] D. Mey, S. Brunet, C. Canaff, F. Maugé, C. Bouchy, E. Diehl, J. Catal. 227 (2004)436–447.

[45] C.E. Hedoire, C. Louis, A. Davidson, M. Breysse, F. Maugé, M. Vrinat, J. Catal. 220(2003) 433–441.

[46] G. Crepeau, V. Montouillout, A. Vimont, L. Mariey, T. Cseri, F. Maugé, J. Phys.Chem. B 110 (2006) 15172–15185.

[47] O. Saur, M. Bensitel, A.B.M. Saad, J.C. Lavalley, C.P. Tripp, B.A. Morrow, J. Catal.99 (1986) 104–110.