Cf

INa

b

c

d

e

a

ARRAA

KHASNOC

1

caoHoest(p

Uf

W

0h

Applied Catalysis A: General 435– 436 (2012) 51– 60

Contents lists available at SciVerse ScienceDirect

Applied Catalysis A: General

j ourna l ho me page: www.elsev ier .com/ locate /apcata

omparison of alumina- and SBA-15-supported molybdenum nitride catalystsor hydrodeoxygenation of guaiacol

. Tyrone Ghampsona,b, Catherine Sepúlvedac, Rafael Garciac, J.L. García Fierrod,estor Escalonac,∗, William J. DeSistoa,e,∗∗

Department of Chemical and Biological Engineering, University of Maine, Orono, ME 04469, United StatesUnidad de Desarrollo Tecnológico, Universidad de Concepción, Casilla 4051, Concepción, ChileUniversidad de Concepción, Facultad de Ciencias Quimicas, Casilla 160c, Concepción, ChileInstituto de Catalisis y Petroquimica, CSIC, Cantoblanco, 28049 Madrid, SpainForest Bioproducts Research Institute, University of Maine, Orono, ME 04469, United States

r t i c l e i n f o

rticle history:eceived 10 April 2012eceived in revised form 23 May 2012ccepted 25 May 2012vailable online 4 June 2012

eywords:ydrodeoxygenationlumina

a b s t r a c t

The hydrodeoxygenation of guaiacol (2-methoxyphenol) has been studied in a batch reactor overalumina- and SBA-15 silica-supported molybdenum nitride catalysts at 300 ◦C and 5 MPa of hydrogenpressure. The catalysts were prepared by nitriding supported Mo oxide precursors with ammonia gasor nitrogen–hydrogen mixtures via temperature-programmed reaction. The alumina-supported cata-lysts had a higher activity relative to the SBA-15 silica-supported catalysts which was essentially dueto catechol production, an effect of the alumina support. The SBA-15 silica-supported catalysts trans-formed guaiacol directly to phenol by demethoxylation without noticeable catechol production. On bothsupports, nitridation by ammonolysis increased the activity by a factor of ∼1.1 relative to nitridation

BA-15itridexynitrideatalyst

by nitrogen–hydrogen. On SBA-15, ammonolysis preferentially produced the �-Mo2N phase while theN2/H2 mixture produced the �-Mo2N0.78 phase. The incorporation of Co led to a marginal improvementin exposed Mo species but generally had a diminishing effect on HDO activity. The lack of catechol pro-duction using the SBA-15 silica support is important in minimizing coking reactions and also opens uppossibilities for utilizing silica supports with highly controlled pore sizes to possibly influence productdistribution in HDO of more diverse feed streams derived from biomass conversion processes.

. Introduction

The removal of oxygen from biomass-derived oils throughatalytic hydrodeoxygenation (HDO) is receiving considerablettention because of its potential as feedstock for the productionf fuels and value-added chemicals. Most of the initial studies onDO reactions have been conducted over metal sulfides supportedn alumina [1,2]. However, recently there have been significantfforts on the development of catalysts, based on new or modifiedupports and new active phases, with minimal hydrogen consump-

ion and high selectivity toward direct oxygen removal [1]. Guaiacol2-methoxyphenol) has commonly been used as a model com-ound for the HDO studies because it is known to exist significantly

∗ Corresponding author. Tel.: +56 41 2207236; fax: +56 41 2245374.∗∗ Corresponding author at: Department of Chemical and Biological Engineering,niversity of Maine, Orono, ME 04469, United States. Tel.: +1 207 581 2291;

ax: +1 207 581 2323.E-mail addresses: nescalona@udec.cl (N. Escalona),

DeSisto@umche.maine.edu (W.J. DeSisto).

926-860X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.apcata.2012.05.039

© 2012 Elsevier B.V. All rights reserved.

in bio-oils, because of its propensity for coke formation, and alsobecause of its intransigence to deoxygenation [2]. Several of thestudies have reported that the HDO activity and selective trans-formation of guaiacol to phenol is distinctively influenced by thenature of the support [3,4]. Catalysts supported on alumina (Al2O3)displayed higher activity compared with alternative supports suchas silica and carbon owing to higher dispersion of the active phase[3]. However, alumina-supported catalysts suffer from coke for-mation which limits the lifetime of the catalyst [3]. The benefitsof using silica and carbon supports lie in the minimal coke forma-tion and greater selectivity toward phenol production over catechol[3].

Interests in exploring non-sulfided catalysts for HDO haveincreased due to issues related to contamination of the feed bythe sulfiding agent [5,6]. Novel active phases such as metal nitrideshave been shown to be an effective catalyst for HDS [7] and HDN [8]reactions. However, only a limited number of studies have reported

their performance for HDO reactions. Recently, Monnier et al. [9]reported activity on �-Al2O3-supported Mo, W, and V nitride cat-alysts for HDO of oleic acid and canola oil. In a recent study, wedemonstrated high activity and rapid demethoxylation of guaiacol

5 atalys

tpt

scdtesictteofSo

maiTamgas

2

2

poMotsmhtarwu

inApapwbM(lToMpp

2 I. Tyrone Ghampson et al. / Applied C

o phenol using unsupported Mo nitride catalysts [10]. It is the pur-ose of this paper to extend our investigations from bulk nitrideso alumina- and silica-supported nitrides for HDO of guaiacol.

Mesoporous silica materials have generated interests as catalystupports in heterogeneous catalysis owing to their very high spe-ific surface areas, controllable pore diameters, narrow pore sizeistributions and large pore volumes. These properties have madehem a more desirable alternative to conventional silica supports,nabling the control of catalyst particle size and influencing productelectivity through transport effects [11,12]. In particular, SBA-15s of general interest because of its high structure regularity, low-ost and non-toxicity [13]. Studies to extend our understanding ofhe SBA-15 pore structure and its relationship to transport proper-ies [14], and the hydrothermal stability of SBA-15 [15] have beenxamined to generate fundamental information relevant to the usef this material as a support in developments of catalysts for bio-uel production. We focus here on demonstrating the reactivity ofBA-15 silica supported molybdenum nitride catalysts for the HDOf a biomass-derived compound.

Here, we report on the synthesis and characterization ofolybdenum nitride dispersed on SBA-15 mesoporous silica

nd �-alumina. The molybdenum nitride was synthesized bympregnation, oxidation and thermal conversion to the nitride.hermal conversion was achieved by two different procedures:mmonolysis and reduction/nitridation using hydrogen/nitrogenixtures. The resulting materials were characterized using nitro-

en adsorption–desorption (surface area, porosity), XRD, elementalnalysis and XPS, and their activity was compared with commercialulfided NiMo/Al2O3 catalysts for the HDO of guaiacol.

. Experimental

.1. Catalyst preparation

SBA-15 mesoporous silica was synthesized following a reportedrocedure [16]. In a typical synthesis, 6 g of Pluronic P123 ethylenexide/propylene oxide block copolymer block copolymer (BASF,avg = 5800, EO20PO70EO20) was dissolved with stirring in 45 g

f deionized water and 180 g of 2 mol L−1 HCl. Then, 12.75 g ofetraethyl orthosilicate (Aldrich, TEOS, 99%) was added to theolution with stirring at 40 ◦C for 24 h. The homogeneous sol–gelixture was then transferred into a polymer flask, sealed and

eated at 100 ◦C for 48 h. After filtration and washing with water,he white solid product recovered was air-dried at room temper-ture for 24 h. The sample was then calcined in air with a heatingate of 1 ◦C min−1 to 500 ◦C and held for 10 h. The alumina supportas obtained commercially (Alfa Aesar, �-Al2O3, 1/8′′ pellets) andsed without any further purification.

The supported molybdenum oxide precursors were prepared byncipient wetness impregnation using aqueous solutions of ammo-ium heptamolybdate (Fischer Scientific, (NH4)6Mo7O24·4H2O,.C.S. grade). The mixture was then left overnight at ambient tem-erature. After impregnation, the samples were dried for 12 ht 110 ◦C and 120 ◦C for SBA-15- and alumina-supported sam-les respectively, and subsequently calcined in a flow of dry airith a heating rate of 1 ◦C min−1 at 500 ◦C and held for 3 h. The

imetallic oxide precursors were prepared by impregnating theo oxide supported materials with an aqueous solution of cobalt

II) nitrate hexahydrate (Acros Organics, Co(NO3)2·6H2O, 99%), fol-owed by the same drying-calcination procedure described above.he supported oxides were prepared to obtain nominal loading

f 10 wt% Mo metal content for monometallic samples, 10 wt%o metal and 2.4 wt% Co metal content for the bimetallic sam-

les. All oxide precursors were sieved to obtain a 180–450 �marticle size. Supported molybdenum nitrides were prepared by

is A: General 435– 436 (2012) 51– 60

temperature-programmed reaction of the oxidic precursors withNH3 (Matheson, 99.99%), or N2/H2 mixtures (N2, BOC Gases, Grade5; H2, Matheson, 99.99%). The synthesis procedure involved flow-ing 300 mL min−1 of either NH3 or N2/H2 (N2/H2 = 5/1 (v/v)) overthe 2.5 g of the supported Mo oxide precursor while the temper-ature was increased from room temperature to 300 ◦C in 30 min,then from 300 ◦C to 500 ◦C within 5.6 h, and from 500 ◦C to 700 ◦Cwithin 1.7 h. The temperature was maintained at 700 ◦C for 2 h, andthen cooled to room temperature under NH3 flow for the nitridesprepared via ammonolysis or cooled under nitrogen flow for thenitrides prepared using the N2/H2 mixture. Finally, the sample waspassivated in 1% O2/N2 for 12 h at room temperature. For nota-tion, nitrides prepared using ammonia (method 1) have suffix “A”,while nitrides prepared using nitrogen–hydrogen mixture (method2) have suffix “NH”.

Molybdenum, cobalt and nitrogen contents in the catalysts wereperformed by the Analytical Laboratory of the Department of Plant,Soil and Environmental Sciences at the University of Maine, and byGalbraith Laboratory. ICP-AES was used for the metal content whilea combustion method was used for nitrogen.

2.2. Catalyst characterization

Nitrogen sorption isotherms were obtained at 77 K using aMicromeritics ASAP-2020 instrument to evaluate the BET spe-cific surface area (SBET), total pore volume (TPV) and averagepore diameter (dpore). Prior to the measurements, the sampleswere outgassed under vacuum following conditions common forthese materials: SBA-15-supported materials were outgassed at200 ◦C for 8 h, and alumina-supported materials were outgassedat 250 ◦C for 2 h. SBET was calculated using the adsorption branchin the range of 0.05 ≤ P/P0 ≤ 0.25 and the TPV was recorded atP/P0 = 0.995. The primary pore diameter was estimated from themaximum in the BJH pore size distribution. The micropore vol-ume was estimated from the ˛s-plot method using �-Al2O3 (FischerScientific, powder certified, SBET = 0.9 m2 g−1, �ref,0.4 = 0.28 cm3 g−1

STP) for alumina-supported nitrides and LiChrospher Si-1000 silicagel (SBET = 26.4 m2 g−1, �ref,0.4 = 9.12 cm3 g−1 STP) [17] for SBA-15-supported nitrides as reference adsorbents.

X-ray diffraction (XRD) patterns of powdered samples wererecorded on a PANalytical X’Pert PRO X-ray diffractometerequipped with a graphite monochrometer and Cu K� radiation(45 kV, 40 mA) in a parallel beam optical geometry. The standardscan parameters were 15–85◦ 2� with a step size of 0.02◦ and acounting time of 10 s per step. Identification of the phases wasachieved by reference to JCPDS diffraction file data.

X-ray photoelectron spectra of reduced catalysts were obtainedon a VG Escalab 200R electron spectrometer using a Mg K�(1253.6 eV) photon source. The passivated catalysts were activatedex situ with H2 at 450 ◦C for 6 h. After reduction, the samples werecooled to room temperature, flushed with nitrogen and stored inflasks containing isooctane (Merck, 99.8%), then transferred to thepre-treatment chamber of the spectrometer. The binding energies(BE) were referenced to the C 1s level of the carbon support at284.9 eV. Intensities of the peaks were calculated from the respec-tive peak areas after background subtraction and spectrum fittingby a combination of Gaussian/Lorentzian functions. Relative surfaceatomic ratios were determined from the corresponding peak areas,corrected with tabulated sensitivity factors [18], with a precisionof 7%.

The acid strength and acid site concentration of some selectedcatalysts were measured using a potentiometric method [19],

whereby a suspension of the material in acetonitrile was titratedwith n-butylamine. The variation in electric potential was regis-tered on a Denver Instrument UltraBasic pH/mV meter.

atalysis A: General 435– 436 (2012) 51– 60 53

2

3atflHtaMosasNttLtt(ddcgbw

3

3

codpmMM4eMrACipXtd1r

npftisp�c

[16,22]. According to IUPAC classifications, the N2 isotherms (Fig. 2aand b) belong to a type IV isotherm which is typical of mesoporousmaterials [23]. The isotherms for the SBA-15 materials in Fig. 2b

Table 1Chemical composition of passivated nitride catalysts.

Catalyst Mo (wt%) Co (wt%) N (wt%) N/Moatomicratio

MoN/Al2O3-A 8.39 – 1.76 1.44MoN/Al O -NH 8.74 – 0.49 0.38

I. Tyrone Ghampson et al. / Applied C

.3. Catalytic activity measurements

Guaiacol HDO activity measurements were carried out in a00 mL stainless steel batch autoclave (Parr Model 4841) at 300 ◦Cnd under a hydrogen pressure of 5 MPa. Prior to catalytic testing,he passivated samples were activated ex situ under H2 at 450 ◦Cor 6 h. As a basis for comparison, commercial Ni-Mo/Al2O3 cata-yst (Procatalyse HR 346) was pre-sulfided using a 10 vol% H2S in

2 mixture at 350 ◦C for 3 h. Approximately 0.25 g of freshly pre-reated catalyst (nitride catalysts or the commercial catalyst) wasdded to the liquid feed containing 2.53 mL guaiacol (0.232 mol L−1,erck, 99%), 80 mL decalin (Merck, 99.5%) as solvent, and 700 �L

f hexadecane (Merck, 99%). Hexadecane was used as an internaltandard for quantitative GC analysis. The sealed reactor was evacu-ted with nitrogen (AGA Chile, Grade 5) for 30 min with continuoustirring; after a leak check the reactor was heated to 300 ◦C under2 which was subsequently replaced with H2 and then pressurized

o 5 MPa. This pressure was maintained for the entire duration ofhe experiment by adding H2 to the reactor whenever necessary.iquid samples were periodically withdrawn during the course ofhe reaction after purging the sampling line with a small amount ofhe reactant mixture. The samples were analyzed by a Perkin ElmerClarus 400) gas chromatograph equipped with a flame ionizationetector (FID) and a CP-Sil 5 CB column (Agilent). The productistributions were identified by their column retention time inomparison with available standards. The initial concentration ofuaiacol was taken as 100% in order to ignore slight conversionefore isothermal condition was achieved. The catalytic activityas expressed by the initial reaction rate.

. Results and discussion

.1. Catalyst properties

Fig. 1 shows the X-ray diffraction profiles of the supports andatalysts used in this study. Examination of the XRD patternsf Mo and CoMo nitrides supported on alumina revealed onlyiffraction peaks associated with the supports. Broad diffractioneaks characteristic of �-Al2O3 (JSPDS ref no: 010-0425) may haveasked the nitride peaks. The XRD patterns of MoN/SBA-15-A andoN/SBA-15-NH catalysts showed broad peaks for crystalline �-o2N (2� = 37.13, 43.41, and 63.03) and �-Mo2N0.78 (2� = 37.51,

3.11, 62.89, and 75.45) respectively. The estimated average diam-ter of the Mo nitride crystals from the diffraction peaks of theoN/SBA-15-A and MoN/SBA-15-NH catalysts were 2 and 3 nm

espectively. The XRD pattern collected for the CoMoN/SBA-15- catalyst revealed the formation of Mo2N but no evidence ofo3Mo3N phase. The presence of Mo2N as the only phase observed

n this catalyst may suggest that the particle size of the Co3Mo3Nhase was small and below the detection limit. It is evident from theRD results that the crystal structure of supported Mo nitride par-

icles are closely related to their nitridation and purging treatmenturing the formation of the particles: catalysts prepared by method

resulted in �-Mo2N particles and those prepared by method 2esulted in �-Mo2N0.78.

Elemental analyses of passivated, supported Mo and CoMoitrides are listed in Table 1. The nitrogen contents for nitride sam-les from method 1 were higher compared to samples preparedrom method 2 in both supports. The lower nitrogen content forhe method 2 samples may be due to purging the samples in flow-ng nitrogen after nitridation which removed weakly bonded NHx

pecies [20]. The atomic N/Mo ratios of the nitride catalysts areresented in Table 1. While the theoretical N/Mo is 0.5 and 0.39 for-Mo2N and �-Mo2N0.78 respectively, it is plausible that excess Nould reside in interstitial sites and defects like grain boundaries.

Fig. 1. XRD patterns of (a) alumina-supported samples and (b) SBA-15-supportedsamples.

These findings, in addition to the absence of diffraction peaks forcrystalline Mo nitride phase on the alumina-supported catalyst,suggest that small crystallites of Mo nitrides below the detectionlimit were formed on the catalysts. This is consistent with otherreport in the literature for Mo2N/Al2O3 catalysts [21].

Nitrogen sorption analyses were performed to determine thedifference in support morphology between �-Al2O3 and SBA-15materials. The N2 adsorption–desorption isotherms and pore sizedistribution (PSD) of alumina and SBA-15 materials are shown inFig. 2. The isotherm and PSD for the as-prepared SBA-15 supportwere consistent with previously reported results for this material

2 3

CoMoN/Al2O3-A 8.33 1.87 1.66 1.54MoN/SBA15-A 7.22 – 2.26 2.15MoN/SBA15-NH 6.67 – 0.92 0.95CoMoN/SBA15-A 7.34 1.95 1.42 1.32

54 I. Tyrone Ghampson et al. / Applied Catalysis A: General 435– 436 (2012) 51– 60

F s of (ad .

s0bspoon

oamtbcraaemu˛s

ig. 2. Morphology of samples under study: N2 adsorption–desorption isothermistributions of (c) alumina-supported samples and (d) SBA-15-supported samples

how a sharp inflection in the relative pressure (P/P0) range from.7 to 0.8, indicative of the presence of uniform pore size distri-ution [24]. The isotherms of �-Al2O3-based materials revealed aharp inflection in the P/P0 range from 0.8 to 1.0, suggestive of wideore size distribution. Fig. 2a and b also shows that the quantityf nitrogen adsorbed and the P/P0 position of the inflection pointn the isotherm of the support decreased after impregnation anditridation.

The BJH pore size distributions (PSD), as derived from the des-rption branch of the N2 isotherm of the materials under study,re shown in Fig. 2c and d. As seen in Fig. 2c and d, the SBA-15esoporous silica support yielded a narrow pore size distribu-

ion centered at 8.3 nm, while the alumina support yielded aroad pore size distribution centered at 9.3 nm. In addition, PSDurves of metal nitrides supported on SBA-15 mesoporous silicaevealed a bimodal pore distribution system with peaks at 4.5 nmnd 7.2 nm, indicating the presence of small complementary poresnd ordered mesoporous pores, respectively, as observed by oth-rs [12,25]. The bimodal PSD and the broadening of the higher

esopore peak suggest that the metal nitride species were not

niformly distributed in the mesopores of the SBA-15 support. As-plot analysis revealed the presence of micropores for SBA-15upport (2.4% microporosity) and more than a 73% decrease in

) alumina-supported samples and (b) SBA-15-supported samples; BJH pore size

microporosity after impregnation and thermal treatment: this sug-gests micropore blocking by the metal nitride species, leading to thenear-elimination of micropores and the appearance of complemen-tary pores caused by inter-particle porosity. These complementarypores of size <4 nm have been discussed to provide connectivitybetween the primary mesopores of SBA-15 [26,27]. A closer look atthe PSD of the SBA-15-supported catalysts suggests greater com-plementary pores for the MoN/SBA-15-A catalyst as compared tothe MoN/SBA-15-NH catalyst despite the marginally higher Mocontent of the former. The XRD data of the two catalysts suggeststhat Mo nitride crystallites formed in the MoN/SBA-15-NH catalystwere larger than those formed in the MoN/SBA-15-A catalyst. Thus,it is possible that access for N2 adsorption/desorption could be morelimited by the relatively larger crystallites of the MoN/SBA-15-NHcatalyst, leading to lesser generation of the complementary pores.This is in good agreement with the difference in the N2 physisorp-tion data presented in Table 2.

In contrast to the pore size distribution results of the SBA-15-supported metal nitrides, the pore size distributions of the

metal nitride/Al2O3 catalysts were unimodal (which is similar tothe support’s PSD) and decreased slightly to smaller mesoporesin comparison to the PSD of the Al2O3 support. Furthermore, themicropore volumes of the catalysts estimated from the ˛s-plot

I. Tyrone Ghampson et al. / Applied Catalysis A: General 435– 436 (2012) 51– 60 55

Table 2Adsorption properties of supports and passivated supported nitride catalysts.

Sample SBET (m2 g−1) TPV (cm3 g−1) V� (cm3 g−1) dpore (nm)

�-Al2O3 207 0.62 0.03 9.3MoN/Al2O3-A 191 0.51 0.02 8.6MoN/Al2O3-NH 183 0.50 0.02 8.9CoMoN/Al2O3-A 182 0.49 0.02 8.9SBA-15 818 1.25 0.03 8.3MoN/SBA15-A 418 0.75 0.008 7.3MoN/SBA15-NH 397 0.70 0.006 7.2

.62

mimm

aM2pct2[i[itTfTcr(opaipmT

TX

CoMoN/SBA15-A 387 0

ethod showed a 33% decrease in microporosity. These resultsndicate a coating of the inner surfaces of the mesopores by the

etal nitride species, causing the pore entrance of a fraction oficropores to close.Table 3 lists a summary of XPS results of the surface composition

nd oxidation states of the components in the reduced, passivatedo nitride catalysts. The binding energies of the Mo 3d5/2, N 1s, Co

p3/2, Si 2p and Al 2p core levels and the surface atomic ratios areresented in Table 3. The Mo 3d and N 1s XPS spectra of the nitrideatalysts are illustrated in Fig. 3. The Mo 3d5/2 spectra presentedhree peaks centered at 229.0, 230.5 and 232.6 eV: peak position of29.0 ± 0.2 eV is typical of Moı+ (2 < ı < 4) assigned to Mo2N species28]; binding energies of 230.5 ± 0.1 eV and 232.6 ± 0.3 eV are typ-cal of Mo4+ and Mo6+ respectively in molybdenum oxynitrides29]. This result indicates that Mo2N and Mo oxynitrides coex-st on the surface of all the catalysts. The surface distribution ofhe Mo oxidation states in the Mo 3d (compiled in parentheses inable 3) shows that molybdenum oxynitride was the dominant sur-ace phase, while Mo2N particles were distributed from 9 to 30%.his result indicates that despite reduction of the passivated nitrideatalysts at 450 ◦C, the sample surfaces were mainly oxynitridesather than nitrides. Table 3 also shows that higher amount of Mo2NMoı+) were formed on the surface of the SBA-15 silica support thann the �-Al2O3 support, indicating that more nitrogen-deficientatches of Mo were on the surface of the former. This behavior isttributed to weaker metal-support interactions of the SBA-15 sil-ca support which leads to relatively easier reducibility of the MoO3

recursor. The XPS results also show that the Moı+ content on theethod 2 catalyst surfaces was higher than the method 1 catalyst.

his result is however unclear. The Co 2p3/2 binding energy for the

able 3PS binding energies (eV) and surface atomic ratios of reduced, passivated nitride catalys

Catalyst Mo3d5/2 N 1s Co 2p3/2

MoN/Al2O3-A228.9 (9) 394.7 (17)

–230.5 (22) 396.7 (36)232.7 (69) 398.7 (47)

MoN/Al2O3-NH229.0 (15) 394.5 (19)

–230.5 (25) 396.3 (34)232.6 (60) 398.4 (47)

CoMoN/Al2O3-A228.9 (19) 394.5 (29)230.5 (30) 396.5 (35) 778.5 (18)

232.6 (51) 398.3 (36) 781.5 (82)

MoN/SBA15-A229.0 (25) 394.4 (26)

–230.5 (29) 396.5 (35)232.9 (46) 398.7 (39)

MoN/SBA15-NH228.9 (27) 394.4 (28)

–230.4 (31) 396.4 (38)232.6 (42) 398.5 (34)

CoMoN/SBA15-A228.9 (30) 394.4 (33)230.5 (28) 396.5 (38) 778.4 (19)

232.6 (42) 398.6 (29) 781.7 (81)

0.007 7.5

supported-CoMo nitrided samples of 778.4 ± 0.1 eV is within therange of the reported binding energy of Co0 for cobalt-nitridespecies [30], while the binding energy of 781.5 ± 0.2 eV compareswell with Co3+ cation of Co–Mo oxynitrides [30]. The amount ofCo3+ species in the bimetallic nitrides was higher than the amountof zero-valent cobalt. Thus, the most abundant surface metalspecies in the bimetallic nitrides were the oxynitrides. The N 1score-level spectra (shown in Fig. 3) made three contributions: thebinding energy = 394.4 ± 0.3 eV could be attributed to the Me O Nbond (Me:Mo, Co, etc.) [31]; the binding energy = 396.7 ± 0.3 eV isclose to Mo N bond [30]; the binding energy near 398.5 ± 0.3 eVhas been reported to be assigned to nitrogen atoms trapped in thegrain boundary of Mo nitrides [32]. Finally, the binding energiesof 103.4 eV obtained for SBA-15-supported catalysts and 74.5 eVobtained for alumina-supported catalysts were identical to the Si2p and Al 2p of the parent SiO2 [33] and Al2O3 respectively [34].This suggests that the alumina and SBA-15 silica supports were notnitrided under the synthesis condition.

The XPS Mo 3d/Al 2p (or Si 2p), N 1s/Al 2p (or Si 2p), and Co/Al2p (or Si 2p) atomic ratios for the reduced, passivated catalysts areshown in Table 3. The Mo/Al surface atomic ratio was greater thanthe Mo/Si atomic ratio in all the catalysts. This indicates that SBA-15-supported catalysts possess their molybdenum species locatedinside the inner silica porous structure while the Al2O3-supportedcatalysts had their molybdenum species located on the externalsupport surface of the catalyst. This is in good agreement withtheir N2 adsorption–desorption isotherms of the passivated cat-

alysts (shown in Fig. 1) which shows a more pronounced decreasein the quantity of N2 adsorbed and a shift of the hysteresis loop tolower P/P0 values of the SBA-15-supported catalysts, indicating the

ts.

Si 2p or Al 2p Mo/Si(Al) N/Si(Al) Co/Si(Al)

74.5 0.099 0.236 –

74.5 0.133 0.349 –

74.5 0.108 0.386 0.038

103.4 0.033 0.084 –

103.4 0.035 0.086 –

103.4 0.034 0.108 0.013

56 I. Tyrone Ghampson et al. / Applied Catalysis A: General 435– 436 (2012) 51– 60

ra for

pcpsbplssffw

3

3S

p

Fig. 3. XPS (a) Mo 3d5/2 and (b) N 1s spect

resence of particles in the porous structure. Table 3 also shows thatatalysts prepared by reduction/nitridation using a N2/H2 mixtureroduced more exposed Mo and N species on the surface of theupport than catalysts prepared by ammonolysis. However, thisehavior is inconsistent with the estimates of dispersion based onarticle sizes determined by XRD of SBA-15-silica supported cata-

ysts. These results suggest that the nitride species deposited on theurface were different from those deposited inside the pores of theupport. In other words, the SBA-15 silica support promoted theormation of nitrogen-deficient patches of Mo on the nitride sur-ace, while the formation of nitrides with higher amount of nitrogenas preferentially located inside the silica pore.

.2. Reactivity

.2.1. Activity of supported metal nitride catalysts: Al2O3 vs.BA-15

In the present study, we evaluated and compared the catalyticroperties of Mo nitrides supported on SBA-15 mesoporous silica

the reduced, passivated nitride catalysts.

and Mo nitrides supported on conventional alumina. Fig. 4 showsthe evolution of reactants and products during the HDO of guaiacolon supported Mo nitride catalysts: the product yields are reportedin g/100 g of guaiacol in the feed. Guaiacol transformation followedthe reaction scheme (Fig. 5) proposed by Bui et al. [35]. The schememainly involves two stages: the first stage involves methoxy groupremoval on guaiacol to form phenol by either demethylation (DME)and dehydroxylation, or by direct demethoxylation (DMO). Thesecond stage involves parallel pathways for phenol deoxygenationeither through direct hydrogenolysis (DDO) to produce benzene orthrough hydrogenation (HYD) of the benzene ring prior to oxy-gen removal to form cyclohexene. Methylation of the aromaticring forms heavy compounds such as di- and tetra-methyl phenolsand dimethyl catechols [35]. Fig. 3 shows that these heavy com-pounds were observed in significant quantity with both Al2O3

− and

SBA-15-supported catalysts. Fig. 4 also indicates that there wereclear differences between alumina- and SBA-15-supported cata-lysts in terms of the changes in products concentration with time.Over the alumina-supported catalysts, catechol was the major early

I. Tyrone Ghampson et al. / Applied Catalysis A: General 435– 436 (2012) 51– 60 57

F me forM

pTecotbsUmnosTohs

ig. 4. Variation of the transformation of guaiacol and the yield of products with tioN/SBA-15-NH, and (f) CoMoN/SBA-15-A catalysts.

roducts while phenol surpassed catechol at longer reaction time.he SBA-15-supported catalysts produced more phenol than cat-chol at both lower and higher conversions. These results areonsistent with previously published work by Centeno et al. [3]n metal sulfide catalysts. On the basis of the products evolu-ion in Fig. 3, alumina-supported catalysts proceeded throughoth the DME and DMO pathways, while the mesoporous silica-upported catalysts proceeded mainly through DMO pathway.nder batch reaction conditions the formation of methane andethanol, byproducts of DMO and DME respectively, could

ot be separated by the column used and hence were notbserved. Methylcatechol was observed only with the alumina-upported catalyst, consistent with other reported studies [4,36].

his product was formed probably through methyl-substitutionf catechol [4] or through transalkylation of guaiacol [36]. Itas recently been reported that catechol formation and its sub-equent methylation can be attributed to the acidic properties

(a) MoN/Al2O3-A, (b) MoN/Al2O3-NH, (c) CoMoN/Al2O3-A, (d) MoN/SBA-15-A, (e)

of the support [4]: an �-Al2O3 support possessing more acidsites than TiO2 and ZrO2 supports were found to favor DMEand methylation routes. The acid strength of MoN/Al2O3-A andMoN/SBA-15-A catalysts, estimated from potentiometric titrationof the catalyst in acetonitrile with n-butylamine [19], were iden-tical with the initial electrode potential of 120 mV and 115 mV,respectively; however, the MoN/Al2O3-A catalyst possessed fourtimes a higher density of acid sites (8.2 meq m−2) than theMoN/SBA-15-A catalyst (2.1 meq m−2). This suggests that thepropensity for DME and methylation routes of the alumina-supported catalysts were influenced by their higher total acidity.Another possible reason could be due to the nature of sitesat the surface of the respective catalysts. The presence of

Lewis acid sites may be responsible for alumina promotingcatechol formation. Furthermore, the active sites on aluminaare different to those on SBA-15, as suggested by XRD andXPS. Trace amounts of deoxygenated products such as benzene,

58 I. Tyrone Ghampson et al. / Applied Catalysis A: General 435– 436 (2012) 51– 60

pathways proposed by Bui et al. [35].

cii[

wIriFabtmTosbthLfiTdpwc

mlcchtssr

Fig. 5. Guaiacol transformation

yclohexene and cyclohexane (collectively denoted HDO productsn Fig. 4) were also observed. Other possible products from gua-acol conversion including anisole, o- and m-cresol, veratrole, etc.36] were not detected.

The activities of the catalysts expressed as the reaction rates, asell as the calculated intrinsic activities are presented in Table 4.

n addition, the total guaiacol conversion after 4 h of reaction iseported in Table 4. The reaction rates were calculated from thenitial reaction during the transformation of guaiacol shown inig. 4. Blank reactions with the Al2O3 and SBA-15 silica supportslone showed an appreciable conversion of guaiacol to catecholy the former while no significant conversion was observed forhe latter. The reaction rates of the Al2O3-supported catalysts were

ore than two times higher than the SBA-15-supported catalysts.he differences in catalytic activity between Mo nitride supportedn alumina and SBA-15 could be related to the density of acidites of the supports. Thus, the higher reaction rates displayedy the alumina-supported nitride catalysts could be explained byheir faster conversion of guaiacol to catechol, methylcatechols andeavy compounds as a result of their higher total acidity [5]. Theewis acidic nature of the alumina support have previously beenound to be prone to substantial coke formation through strongnteraction with guaiacol, forming doubly anchored phenates [37].he intrinsic activities, calculated using experimental reaction rateata normalized by the molybdenum content of the catalyst, areresented in Table 4. The trend in the initial intrinsic activitiesas similar to that of the initial reaction rates since the catalysts

ontained similar Mo contents.Although the Mo nitride catalysts supported on SBA-15

esoporous silica showed lower activities, they have particu-ar advantages over the alumina-supported nitride and sulfideatalysts in terms of the higher specificity for phenol over cate-hol. This has implications with regards to lower consumption ofydrogen and less formation of coke in HDO applications. In addi-

ion, the ability to finely tune pore sizes in ordered mesoporousilica supports offers other opportunities to influence activity andelectivity by controlling catalyst dispersion and the diffusion ofeactants and products from the active catalyst site.

Fig. 6. Phenol and catechol yields calculated at 10% guaiacol conversion.

The reaction rates obtained with the supported catalysts werehigher than the reported activities of unsupported molybdenumnitride catalysts [10] (except for the CoMoN/SBA-15-A). This resultsuggests that although the surface area of supported nitride cat-alysts consists of ∼10% of the active nitride phase in comparisonto unsupported nitride catalysts whose surface area are made upof only the active nitride phase, the use of a support in this reac-tion enhanced activity through greater accessibility of active sites toreactants as a result of the support’s porosity. In addition, improvedcatalyst dispersion and possible electronic effects of supported cat-alysts may have played a role in activity enhancement.

The phenol/catechol selectivity (g/100 g guaiacol converted) cal-culated at 10% guaiacol conversion is presented in Table 4 and

illustrated in Fig. 6: SBA-15-supported catalysts produced signif-icantly more phenol than catechol, while the alumina-supportedcatalysts produced more catechol than phenol. These results indi-cate a higher selectivity of the nitrides supported on alumina

I. Tyrone Ghampson et al. / Applied Catalysis A: General 435– 436 (2012) 51– 60 59

Table 4Total conversion, reaction rates and phenol/catechol selectivity of catalysts under study.

Catalyst Conversiona (%) Reaction rate(×106 mol g−1

catalysts−1)

Intrinsic activity(×104 mol Mo at−1 s−1)

Product distributionb (%)

Phenol Catechol

MoN/Al2O3-A 66 14.0 160.6 1 26MoN/Al2O3-NH 62 12.7 139.2 2 26CoMoN/Al2O3-A 58 10.8 124.7 5 22NiMoS/Al2O3 63 7.4 76.3 1 11MoN/SBA15-A 44 6.4 84.7 26 6MoN/SBA15-NH 40 5.2 74.7 22 9CoMoN/SBA15-A 24 0.5 6.8 12 6

tscpoCtttbo

3m

uctmscTwt�baatsf

1p1cttpcms

dwtwaChn

a Total conversion after 4 h of reaction.b Phenol and catechol selectivity calculated at 10% guaiacol conversion.

oward the demethylation pathway compared to that of nitridesupported on SBA-15 which had a higher capacity for aromaticarbon–oxygen hydrogenolysis. The results confirm that both sup-ort modified the active sites of the nitrides. Hydrogenolysisf the methyl-oxygen bond, as well as hydrogenolysis of thearomatic OCH3 bond, has been suggested to take place on bothhe support surface and on the nitride species [38]. However, sincehe bare SBA-15 silica support was catalytically inert for HDO reac-ion the higher phenol production displayed by these catalysts maye attributed primarily to the DMO sites on the Mo nitride andxynitride.

.2.2. Activity of differently synthesized Mo nitride catalysts:ethod 1 vs. method 2

It can be observed in Table 4 that the nitride catalysts preparedsing method 1 displayed higher reaction rates than the nitrideatalysts prepared using method 2. In general, it was concludedhat the higher bulk N/Mo ratio (shown in Table 1) and the for-

ation of �-Mo2N phase (deduced from XRD) by the method 1ynthesis procedure led to a more favorable conversion of guaia-ol than the formation of �-Mo2N0.78 by the method 2 procedure.his is consistent with the results of unsupported Mo nitrides,hich revealed higher guaiacol conversion for catalysts that con-

ain predominantly �-Mo2N in comparison to catalysts that contain-Mo2N0.78 particles [10]. There is no observable relationshipetween the activity and the distribution of surface Mo nitridend oxynitride species, suggesting that there might be multiplective phases with differing activities for guaiacol conversion inhe molybdenum–oxygen–nitrogen system. A study on the atomiccale knowledge of the active phase of Mo nitride and oxynitrideor HDO catalysis is warranted.

Fig. 6 shows differences in phenol/catechol yields for the SBA-5-supported nitride catalysts prepared using different nitridationrocedures. The higher phenol yield obtained for the MoN/SBA-5-A catalyst could also be due the presence of the �-Mo2Nrystalline phase which has a greater capacity to directly cleavehe Caromatic OCH3 bond [10]. Also from Fig. 6, it can be observedhat MoN/Al2O3-A and MoN/Al2O3-NH catalysts displayed similarhenol/catechol production. This behavior is not clear. However, itould be due to the dominant effect of the Lewis acidity of the alu-ina support (forming catechol) which slightly modified the active

ites.The addition of Co to the supported Mo nitrides surprisingly

id not enhance guaiacol conversion rate. In fact, the reaction rateas 1.1 times lower for the CoMoN/Al2O3-A catalyst as compared

o the MoN/Al2O3-A catalyst, while the CoMoN/SBA-15-A catalystas significantly less active as compared to the MoN/SBA-15-A cat-

lysts. This could be explained by the incomplete formation of theo3Mo3N phase (from XRD and XPS results) which typically has aigher C-X (X:S or N) hydrogenolysis rate than monometallic Moitride catalyst [39,40].

Fig. 7. Variation of the transformation of guaiacol and the yield of products withtime for NiMoS/Al2O3 catalyst.

3.2.3. Comparison of nitrides to commercial sulfided NiMo/Al2O3catalyst

The activity of a commercial sulfided NiMo/Al2O3 catalyst wasalso tested for the HDO of guaiacol and illustrated in Fig. 7: the prod-uct distribution as a function of time was similar to those obtainedwith the alumina-supported nitride catalysts. This confirms thestrong influence of the acidic properties of the alumina support.The activity per gram of catalyst given in Table 4 obtained with thecommercial reference catalyst was two times lower as compared tothe Mo nitrides supported on alumina. Furthermore, the alumina-supported nitrides showed more than two times higher activity perMo atoms than the sulfided NiMo/Al2O3 catalyst. This preliminarycomparison of results is encouraging for the application of nitridesfor HDO catalysis.

The phenol/catechol ratio as shown in Fig. 6 with the sul-fided NiMo/Al2O3 catalyst was comparable to the Mo nitridessupported on alumina. This result further proves the overall capac-ity of alumina-supported catalysts for demethylation of guaiacol,regardless of the active phase [41]. This is in good agreement withother reported findings which indicated that Lewis acid sites ofthe alumina support were mainly responsible for the conversion ofguaiacol to catechol [5,41]. Further information regarding leachingand post-catalyst evaluation will be required for a more completeevaluation of catalyst stability.

4. Conclusion

We have prepared alumina- and SBA-15-supported Mo nitridecatalysts and showed their reactivity for the HDO of guaiacol in

terms of reaction rates and phenol/catechol yields. Catalysts weresynthesized by nitridation using two procedures: thermal conver-sion in ammonia and thermal conversion in nitrogen/hydrogenmixtures. Nitridation using ammonia resulted in higher activities

6 atalys

iapN�acrpeapavaaTpi

A

GCg0aH

R

[

[

[

[[

[

[

[[

[[[

[[

[

[

[[

[[[[

[[

[[

[

[

[38] G. de la Puente, A. Gil, J.J. Pis, P. Grange, Langmuir 15 (1999) 5800–5806.

0 I. Tyrone Ghampson et al. / Applied C

n both supports and correlated with the formation of �-Mo2Nnd a higher N/Mo ratio in the catalyst, in good agreement withreviously published results with unsupported nitride catalysts.itridation using nitrogen/hydrogen resulted in the formation of-Mo2N0.78. Consistent with findings for MoS2-based catalysts onlumina, the alumina supported nitrides resulted in significantonversion of guaiacol to catechol. The silica supported catalystsesulted in minimal catechol production, and maximum phenolroduction. For the catalysts studied here, the major factor influ-ncing activity was the active phase whereas the major factorffecting phenol production over catechol production was the sup-ort which modified the nature of the active sites of nitrides. Thelumina-supported nitride catalysts were more active than a con-entional sulfided NiMo/Al2O3 catalyst and were generally morective than previously reported unsupported nitride catalysts. Theddition of cobalt did not have a promoting effect on HDO activity.hese results are encouraging for the application of ordered meso-orous silicas as supports for molybdenum nitride based catalysts

n HDO applications.

cknowledgments

The authors acknowledge the financial support of DOE Epscorrant #DE-FG02-07ER46373 and financial support from CONICYThile, projects PFB-27, PIA-ACT-130 and FONDECYT No. 1100512rants. I. Tyrone Ghampson is indebted to NSF Career Award547103 for sponsoring a trip to the University of Concepción. Theuthors also gratefully acknowledge the technical assistance of Nickill and Manuel Veliz.

eferences

[1] J. Zakzeski, P.C.A. Bruijnincx, A.L. Jongerius, B.M. Weckhuysen, Chem. Rev. 110(2010) 3552–3599.

[2] E. Furimsky, Appl. Catal. A 199 (2000) 147–190.[3] A. Centeno, E. Laurent, B. Delmon, J. Catal. 154 (1995) 288–298.[4] V.N. Bui, D. Laurenti, P. Delichère, C. Geantet, Appl. Catal. B 101 (2011) 246–255.[5] E. Laurent, B. Delmon, Appl. Catal. A 109 (1994) 97–115.

[6] M. Ferrari, R. Maggi, B. Delmon, P. Grange, J. Catal. 198 (2001) 47–55.[7] G.M. Dolce, P.E. Savage, L.T. Thompson, Energy Fuels 11 (1997) 668–675.[8] M. Nagai, Y. Goto, A. Irisawa, S. Omi, J. Catal. 191 (2000) 128–137.[9] J. Monnier, H. Sulimma, A. Dalai, G. Caravaggio, Appl. Catal. A 382 (2010)

176–180.

[

[[

is A: General 435– 436 (2012) 51– 60

10] I.T. Ghampson, C. Sepulveda, R. Garcia, B.G. Frederick, M.C. Wheeler, N. Escalona,W.J. DeSisto, Appl. Catal. A 413–414 (2012) 78–84.

11] I.T. Ghampson, C. Newman, L. Kong, E. Pier, K.D. Hurley, R.A. Pollock, B.R. Walsh,B. Goundie, J. Wright, M.C. Wheeler, R.W. Meulenberg, W.J. DeSisto, B.G. Fred-erick, R.N. Austin, Appl. Catal. A 388 (2010) 57–67.

12] R. Nava, B. Pawelec, P. Castano, M.C. Álvarez-Galván, C.V. Loricera, J.L.G. Fierro,Appl. Catal. B 92 (2009) 154–167.

13] K. Soni, K.C. Mouli, A.K. Dalai, J. Adjaye, Catal. Lett. 136 (2010) 116–125.14] R.A. Pollock, B.R. Walsh, J. Fry, I.T. Ghampson, Y.B. Melnichenko, H. Kaiser,

R. Pynn, W.J. DeSisto, M.C. Wheeler, B.G. Frederick, Chem. Mater. 23 (2011)3828–3840.

15] A. Galarneau, M. Nader, F. Guenneau, F. Di Renzo, A. Gedeon, J. Phys. Chem. C111 (2007) 8268–8277.

16] Z.H. Luan, M. Hartmann, D.Y. Zhao, W.Z. Zhou, L. Kevan, Chem. Mater. 11 (1999)1621–1627.

17] M. Jaroniec, M. Kruk, J.P. Olivier, Langmuir 15 (1999) 5410–5413.18] C.D. Wagner, L.E. Davis, M.V. Zeller, J.A. Taylor, R.H. Raymond, L.H. Gale, Surf.

Interface Anal. 3 (1981) 211–225.19] R. Cid, G. Pecchi, Appl. Catal. 14 (1985) 15–21.20] M. Nagai, Y. Goto, O. Uchino, S. Omi, Catal. Today 45 (1998) 335–340.21] H.K. Park, J.K. Lee, J.K. Yoo, E.S. Ko, D.S. Kim, K.L. Kim, Appl. Catal. A 150 (1997)

21–35.22] D.Y. Zhao, J.Y. Sun, Q.Z. Li, G.D. Stucky, Chem. Mater. 12 (2000) 275–279.23] S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, Academic Press,

London, 1982.24] O. González, H. Pérez, P. Navarro, L.C. Almeida, J.G. Pacheco, M. Montes, Catal.

Today 148 (2009) 140–147.25] R. Nava, B. Pawelec, J. Morales, R.A. Ortega, J.L.G. Fierro, Micropor. Mesopor.

Mater. 118 (2009) 189–201.26] M. Kruk, M. Jaroniec, C.H. Ko, R. Ryoo, Chem. Mater. 12 (2000) 1961–1968.27] A. Galarneau, H. Cambon, F. Di Renzo, R. Ryoo, M. Choir, F. Fibula, New J. Chem.

27 (2003) 73–79.28] C. Shi, A.M. Zhu, X.F. Yang, C.T. Au, Appl. Catal. A 276 (2004) 223–230.29] S.W. Yang, C. Li, J. Xu, Q. In, J. Phys. Chem. B 102 (1998) 6986–6993.30] K. Hade, M. Nagai, S. Omi, J. Phys. Chem. B 105 (2001) 4084–4093.31] T. Nagata, G. Koblmuller, O. Brewage, C.S. Gallant, J.S. Speck, Appl. Phys. Lett.

95 (2009).32] M. Nagai, J. Takada, S. Omi, J. Phys. Chem. B 103 (1999) 10180–10188.33] J.S. Girard on, E. Quintet, A. Removal-Constant, P.A. Chernavskii, L. Gingerbread,

A.Y. Kodak, J. Catal. 248 (2007) 143–157.34] K. Hade, M. Nagai, S. Omi, J. Phys. Chem. B 104 (2000) 2090–2098.35] V.N. Bui, G. Tousling, D. Laurenti, C. Microdots, C. Geantet, Catal. Today 143

(2009) 172–178.36] T. Nimmanwudipong, R.C. Runabout, D.E. Block, B.C. Gates, Energy Fuels 25

(2011) 3417–3427.37] A. Popover, E. Kondratieva, J.M. Gospel, L. Marie, P. Basin, J.-P. Gibson, A. Trivet,

F. Maguey., J. Phys. Chem. C 114 (2010) 15661–15670.

39] J.W. Logan, J.L. Heiser, K.R. McCrea, B.D. Gates, M.E. Bussell, Catal. Lett. 56 (1998)165–171.

40] Y. Li, Y. Zhang, R. Raval, C. Li, R. Zhai, Q. In, Catal. Lett. 48 (1997) 239–245.41] E. Laurent, B. Delmon, Appl. Catal. A 109 (1994) 77–96.