Effect of the carburization of MoO3-based catalysts on the activity for butane hydroisomerization

Effect of the carburization of MoO3-based catalysts on the activityfor butane hydroisomerization

A. Goguet a,*, S. Shekhtman a, F. Cavallaro a, C. Hardacre a, F.C. Meunier a,b,**a CenTACat, School of Chemistry, Queen’s University, Belfast BT9 5AG, Northern Ireland, United Kingdomb Laboratoire Catalyse & Spectrochimie, ENSICAEN, University of Caen, CNRS, 14050 Caen Cedex, France

Applied Catalysis A: General 344 (2008) 30–35

A R T I C L E I N F O

Article history:

Received 3 February 2008

Received in revised form 23 March 2008

Accepted 28 March 2008

Available online 7 April 2008

Keywords:

Molybdena

MoO3

Butane

Alkane

Isomerization

Oxycarbide

Temporal analysis of products (TAP)

A B S T R A C T

Catalysts based on molybdena (MoO3) reduced at mild temperatures are highly active and selective for

the hydroisomerization of alkanes; however, further catalyst development has been hampered by the

structural complexity of the material and the controversy regarding the nature of the active phase. The

present work is aimed at determining the relationship between the content of carbon present in an

oxycarbide phase and the activity for n-butane hydroisomerization. A series of temperature-

programmed oxidation (TPO) and temporal analysis of product (TAP) data showed that the oxycarbidic

carbon content is not related to the activity of the sample for the isomerization of n-butane to isobutane.

The formation of a carbon-containing phase is, therefore, not crucial to obtain an active catalyst. This

study also highlights the capability of the multi-pulse TAP technique to investigate structure–activity

relationships over materials with readily variable atomic composition.

� 2008 Elsevier B.V. All rights reserved.

Contents l is ts ava i lab le at ScienceDirec t

Applied Catalysis A: General

journal homepage: www.e lsev ier .com/ locate /apcata

1. Introduction

The need to tackle global warming has renewed the interest inn-alkane isomerization for the production of more efficient andcleaner fuels. Isoalkanes have intrinsically higher octane rates thanthe corresponding paraffins and can also be used for the formationof alkylates. Catalysts based on molybdena (MoO3) reduced at mildtemperatures, e.g. 350 8C, are active for the hydroisomerization ofC4–C7 alkanes [1–11]. These materials can be more selective athigh conversions than zeolite-supported platinum, are moreresistant to sulfur and nitrogen poisoning and do not catalysethe formation of significant levels of aromatics [3,4,12,13]. Furthercatalyst development has been hampered by the structuralcomplexity of the material, which is only obtained by mildreduction of MoO3 (which is a complex process, even simply usingH2 as reducing agent [14–16]), and by contradictory statementsregarding the related mechanism of reaction [9,17].

* Corresponding author.

** Corresponding author at: Laboratoire Catalyse et Spectrochimie, ENSICAEN,

University of Caen, CNRS, 6 Boulevard Marechak Juin, 14050 Caen Cedex, France.

Tel.: +33 231451359; fax: +33 231452822.

E-mail addresses: [email protected] (A. Goguet), [email protected]

(F.C. Meunier).

0926-860X/$ – see front matter � 2008 Elsevier B.V. All rights reserved.

doi:10.1016/j.apcata.2008.03.038

Some authors have proposed that the active phase is amolybdenum compound containing carbon as an oxycarbide[3,19,30,18], based on the fact that an induction period is neededwhen starting from MoO3 before any activity is observed and that aspecific carbon-containing phase has been observed [30]. Othershave proposed that the catalysis occurs on a MoOx-based phase [4–8,10], partly based on the fact that the MoO3 pre-reduced in H2 isimmediately active for isomerization upon alkane introduction.However, it must be recognized that carbonaceous deposits areformed almost instantaneously under reaction conditions overreforming catalysts and carbon can thereafter be readily incorpo-rated into the bulk of catalytic materials containing MoOx to forminterstitial compounds [19]. These observations stress that it isdifficult to answer whether or not carbon is needed in order toactivate the reduced MoOx catalyst.

Detailed characterizations of the activated catalyst have beenreported, despite the fact that the use of characterizationtechniques is difficult because of the pyrophoricity of the sample.TEM, XRD, UPS and XPS studies were carried out over activatedsamples [8,18,20]. The surface and bulk composition of the catalystappears to be made of a mixture of two poorly crystallized phases,MoO2 and a face-centered cubic phase. The latter phase is aninterstitial compound, the composition of which appears to varysignificantly depending on the experimental conditions, leading tothe formation of an oxycarbide phase when carbon-containing

Fig. 1. n-Butane and isobutane fragmentation patterns for masses 29, 42 and 43.

Fig. 2. Activity of the reduced MoO3 as a function of time on stream. T = 350 8C,

feed = 10% n-butane in H2; total flowrate: 50 ml min�1; 100 mg of MoO3 was used.

A. Goguet et al. / Applied Catalysis A: General 344 (2008) 30–35 31

molecules are present in the feed [19]. A metallacyclic mechanismoccurring on the oxycarbide phase has been proposed as mainreaction mechanism [30]. In contrast, a traditional bifunctionalmechanism involving a metallic phase and an acidic site has alsobeen proposed [8,9]. The metallic function can arise from the MoO2

[8] and/or the fcc interstitial phase, while the acidity probablyoriginates from acidic hydroxyl groups present at the catalystsurface. These observations underline the complexity of the activematerial and we therefore aimed in the present paper at onlyinvestigating the effect of gradual increases of the carbon contentof the sample on the hydroisomerization activity.

Over the last few decades, the temporal analysis of products(TAP) technique, developed by Gleaves and co-workers [21,22], hasbeen successfully applied to transient kinetic characterization andinvestigation of reaction mechanisms of model and multicompo-nent industrial catalysts [23,24]. The TAP technique provides anopportunity to probe the catalyst at essentially fixed compositionduring a single pulse experiment because the amount of pulsedmolecules is several orders of magnitude smaller than the numbersurface active sites/species. In addition, the catalyst state may bechanged incrementally by exposing the catalyst to a large numberof pulses [25]. Such a multi-pulse TAP experiment is particularlyinformative when combined with a thin-zone TAP-reactor (TZTR)[26] which ensures uniformity in the catalyst zone [27].

The effect on the catalytic activity of increasing concentrationsof carbon in the MoO3-derived catalyst using both ambientpressure and vacuum pulse-response TAP techniques is reportedhere. The TAP data were particularly informative, since the minutereactant pulse size allowed measuring the sample activity withoutsignificantly changing the surface composition. This study shouldtherefore indicate whether an oxide form of the sample or itscarbon-modified counterpart hosts the active phase. Note that theaim of the present work was neither to determine the nature of theactive phase nor the reaction mechanism, but merely whethercarbon is needed to form this active phase.

2. Experimental

MoO3 from STREM (purity >99.5%) were used for the catalytictests. High purity gases were supplied by high-pressure cylindersfrom BOC. The BET surface area of the MoO3 was 2.7 m2 g�1. Gasflows were regulated by Aera mass flow controllers. For theambient pressure experiments, the catalytic activity was measuredat 350 8C. Before use, the samples were calcined at 450 8C for 2 h inair and then reduced at 350 8C for 24 h with 50 ml min�1 of H2. TheBET surface area of a H2-reduced sample was typically 150 m2 g�1

[19]. The catalytic tests were carried out using a feed made of 10%n-butane in H2 with a total flowrate of 50 ml min�1, using 100 mgof MoO3. A micro-pilot made of stainless steel and quartz was used,the entire length of the lines being kept at around 120 8C. Gassamples were analyzed by a gas chromatograph (PerkinElmerClarus 500) fitted with a Chromosorb 102 packed column and aflame ionization detector.

The TPO analyses were carried out in the same reactor as thatused for the butane hydroisomerization activity test, with the CO2

evolved being measured via an on-line FTIR analyzer. Theintegration of CO2 and CO was carried out over the wavenumberrange 2399–2348 cm�1 using a single point baseline at 2399 cm�1

for CO2 and 2240–2143 using a single point baseline at 2143 cm�1

for CO. Following the catalytic test, the reactor temperature wasdecreased to 30 8C under 30 ml min�1 of nitrogen. The TPO werethen carried out from 30 up to 500 8C using a temperature ramp of5 8C/min under 1% O2 in He, the total gas flow being 50 ml min�1.

The TAP reactor was packed with 30 mg of molybdenum oxidecatalyst in a thin-zone manner. After 6 h of catalyst reduction

under hydrogen at atmospheric pressure and 350 8C, the reactorwas evacuated and an Ar/H2/butane mixture (1/1/1) wasrepeatedly pulsed over the reduced catalyst at the sametemperature. To deconvolute butane and isobutane signals, themajor fragments of butane and isobutane were monitored in apreliminary experiment over inert particles. The different frag-mentation of these two molecules to 43, 42 and 29 AMU (Fig. 1)provided the basis for a straightforward deconvolution.

3. Results

3.1. Atmospheric pressure experiments

The MoO3 was oxidized in situ at 450 8C in a flow of synthetic airand then reduced in pure H2 for 24 h at 350 8C prior to measuringthe catalytic activity for n-butane hydroisomerization to isobutaneat the same temperature. Data reported elsewhere [18,28] haveshown that no molybdenum metal was observed by XRD after areduction of MoO3 with H2 at 350 8C over a period of 24 h. Theformation of Mo metal would actually lead to a sample with a lowisomerization activity, as shown by Matsuda et al. [29]. The only(poorly) crystalline phases observed after 24 h under H2 at thistemperature were MoO2 and the fcc MoOxHy phase [18,28].

The samples pre-reduced by H2 were then exposed to thebutane/H2 feed for three different durations (i.e. 10, 100 and600 min) before being purged in Ar and brought back to roomtemperature. The isobutane yield and selectivity obtained over thesample run for up to 600 min are shown in Fig. 2 as an example. Theselectivity to isobutane remained around 90% throughout theexperiment with the main by-products being propane, ethane andmethane. The value of the isobutane yield and the conversion (notshown) gradually declined with time on stream. The origin of thedeactivation is yet unclear, but could possibly be related to theformation of carbonaceous deposits.

Fig. 3. Temperature-programmed oxidation of the MoO3 reduced 24 h under H2 at

350 8C and subsequently exposed at the same temperature to 10-butane in H2 for

10 min (thick line), 100 min (dashed line) and 600 min (thin line).

Fig. 5. Isobutane yield measured over the reduced MoO3 catalyst as a function of the

uptake of carbon (normalized to the number of Mo atoms) determined by TPO.

T = 350 8C, feed = 10% n-butane in H2.

A. Goguet et al. / Applied Catalysis A: General 344 (2008) 30–3532

A temperature-programmed oxidation (TPO) was carried outafter each test to determine the amount of carbon present over thecatalyst. Both CO and CO2 were observed and the correspondingconcentrations could be quantified through a calibration of the IRgas-cell signal. At all temperatures CO2 was essentially the onlycarbon oxide formed, as the CO concentration remained negligible.The TPO revealed a sharp production of CO2 at around 365 8C foreach of the sample exposed to the butane-containing reactionstream irrespective of reaction time (Fig. 3). With increasing time,the amount of carbon on the sample was found to rise with smalladditional features appearing at 345, 400 8C and a broad feature attemperatures higher than 450 8C for the sample exposed for 100and 600 min.

The TPO profile of the catalyst reported herein is similar to thatreported by Ledoux et al. for a molybdenum bronze H0.34MoO3 thatwas pretreated in a similar way [30]. One notable difference is thatthe main CO2 peak occurred at 395 8C compared with 365 8C for thesample presented in this study. The main CO2 peak has beenproposed to be associated with the oxidation of a metastable phaseof molybdenum represented by the formula MoOxCyHz [30], withvalues for x, y and z dependant on several experimental parameters[28]. The difference in temperature for the main peak observedbetween our data and those reported elsewhere [30] could besimply due to a difference in the gas flow-rate or sample mass used,which are not given in reference [30]. During oxidation, theoxycarbide is converted into MoO3 with an evolution of CO2 (andwater). The other minor ill-defined peaks (at 345, 400 8C andabove) can probably be assigned to the combustion of carbonac-eous deposits [10,30].

The amount of carbon (reported to that of molybdenum)associated with the oxycarbidic phase appeared to increasecontinuously as a function of time on stream, showing that the

Fig. 4. Mole ratio of carbon present in the phase associated with the TPO peak at

365 8C and Mo as a function of the reaction time.

carburization of the sample was gradual (Fig. 4). The isobutaneinstantaneous yield obtained at the end of each run was plottedagainst the corresponding C/Mo ratio (Fig. 5) and is found to beinvariant with carbon content of the MoOxCyHz phase. The presentflow setup was not appropriate to study the sample activity at amuch lower level of exposure to butane and, therefore, the TAPreactor was used to analyze the evolution of the sample activitywith an increasing number of minute butane pulses.

3.2. TAP experiments

The typical pulse intensity delivered by the valves in our multi-pulse TAP experiments is of ca. 1.7 � 1016 molecules. Forcomparison, the number of surface Mo atoms over the activatedcatalyst was about 4.7 � 1019 molecules or 7.8 � 10�5 moles.(Note: the oxycarbide has a fcc structure, with a cell parameter of0.42 nm [30], leading to a surface density of Mo of 12.5 Mo nm�2.30 mg of MoO3, that is ca. 28 mg of activated catalyst, lead to asurface area of 150 m2 g�1 [18,28].) Therefore, the number ofbutane molecules in a TAP pulse was lower than the number ofsurface Mo atoms by more than three orders of magnitude. Suchsmall pulses allow the reactivity of the carbon-free reduced oxidesprepared under pure H2 to be probed by examining the initialprofiles of the peaks at the start of the multi-pulse experimentwhen sending the first few hundred n-butane/H2 pulses. The wholemulti-pulse experiment that was carried out consisted of over20,000 pulses and resulted in some carbonization of the catalyst asdescribed below. Typical n-butane responses collected during themulti-pulse TAP experiment are shown in Fig. 6. The TAP responsesof isobutane were similar in shape and timescale as those of n-butane and therefore are not shown.

The normalized n-butane responses clearly became narrower asthe number of pulses passed over the sample increased. This

Fig. 6. n-Butane MS response during a TAP multi-pulse experiment.

Fig. 7. TAP data showing the evolution of the conversion of n-butane and the

isobutane yield as a function of the number of pulses. T = 350 8C.

A. Goguet et al. / Applied Catalysis A: General 344 (2008) 30–35 33

observation indicates that the adsorption properties of n-butanewere markedly modified upon increased exposure to the n-butane/H2/Ar pulses. The desorption properties may have also beenmodified; however, this is unlikely to be the cause as explained inthe response shape analysis discussed below. The single pulse n-butane conversion (i.e. the ratio of amount of n-butane consumedto amount of n-butane pulsed) and the isobutane yield (i.e. theratio of amount of isobutane produced to amount of n-butanepulsed) were calculated using Ar as internal standard and plottedversus the pulse number (Fig. 7). The single pulse n-butaneconversion decreased from 30% to 10% over the first 4000 pulsesand then remains at �10% thereafter. During the whole multi-pulse experiment the single pulse n-butane conversion was clearlyhigher than the yield of isobutane which was approximatelyconstant during the whole multi-pulse experiment (around 5%). Noother gas products were found in observable amount during thewhole multi-pulse experiment. This observation indicates thatmost of the missing carbon was being converted into carbonaceousdeposits and/or being incorporated into the sample to form theMoOxCyHz phase [19,30].

In order to quantify the modification of the reactivity of thecatalyst due to the oxycarbide formation and/or the deposition ofcarbonaceous material the number of n-butane molecules con-sumed were compared with the amount of isobutane produced priorto each pulse. The difference between these two numbersdetermines the amount of carbon present on the catalyst. Fig. 8shows the single pulse n-butane conversion and the isobutane yieldplotted versus the amount of carbon (normalized to the number ofMo atoms) left on the sample. As observed in the case of the ambientpressure experiments (Fig. 5), the isobutane yield was essentiallyconstant over a wide range of C/Mo ratios. Importantly, the constantvalue of isobutane yield was present for the very first TAP pulses.

In summary, the TAP results lead us to the followingconclusions:

(a) d

Fig.isob

the

eposition of carbonaceous deposits or carburization of thecatalyst surface was taking place during the whole multi-pulse

8. TAP data showing the evolution of the conversion of n-butane and the

utane yield as a function of the proportion of C in the catalysts (normalized to

number of Mo atoms). T = 350 8C.

experiment, since the n-butane conversion was significantlygreater than the isobutane yield (and the yield of other gas-phase products was negligible) and

(b) t

3.3. Modeling of the TAP data

Single-pulse responses were analyzed using the momentformalism, with a view at better understanding the modificationof the n-butane/catalyst interaction with the number of pulses. Thetheory was presented in detail in [25]. According to this theory,each moment of observed response (M0, M1, M2) allowsdetermining only one independent kinetic parameter (R0, R1, R2)called primary kinetic coefficients. For the TZTR n-butane multi-pulse responses these coefficients are calculated via the followingequations:

The zeroth coefficient, R0, is calculated from the zeroth moment,M0, as

R0 ¼1�M0

tcatM0(1)

R0 has the dimension of reciprocal time (s) and represents anapparent rate constant of n-butane consumption.

The first coefficient, R1, is calculated from the first moment, M1,as

R1 ¼M1

tcatM20

� tin1

tcatþ R0

3

� �(2)

R1 is dimensionless and represents n-butane apparent ‘‘inter-mediate-gas’’ constant, which relates gaseous and adsorbed n-butane.

The second coefficient, R2, is calculated from the secondmoment, M2, as

jR2j ¼M2

2tcatM20

� M21

tcatM30

þ tin

6

tinR0

5þ 2R1 þ

tin

tcat

� �(3)

R2 has the dimension of time (s) and describes n-butane apparenttime delay on the catalyst surface, where tcat = LrLcat/2Din is theresidence time in the catalyst zone and tin ¼ einL2

r =2Din is theresidence time in inert zones. The explicit expressions (2)–(4) havebeen recently detailed by Shekhtman et al. [31].

From the fact that R1 calculated for n-butane is greater than zeroduring whole multi-pulse experiments, it was concluded that n-butane is held temporarily in the reactor bed (as suggested in Ref.[25]). The same conclusion follows from the evolution of the n-butane responses in Fig. 6. This indicates that the n-butane adsorbsreversibly during the whole multi-pulse experiment. The adsorbedn-butane can also transform into an intermediate that may remainon the surface or produce isobutane. This conclusions lead to thefollowing mechanism for n-butane adsorption/desorption andreaction:

n-butane�!ka

�kd

n� butaneads

n-butaneads�!ktrans

intermediate

(4)

The rate constants of these elementary steps, i.e. for n-butaneadsorption, desorption and surface reaction, were determinedfrom the primary kinetic coefficients (R0, R1, R2) using relationshipsreported for reversible adsorption + irreversible reaction described

Fig. 9. Apparent equilibrium constant K = kads/kdes and reaction constant ktrans

values as a function of the proportion of carbon in the catalysts (normalized to the

number of Mo atoms). T = 350 8C.

A. Goguet et al. / Applied Catalysis A: General 344 (2008) 30–3534

by Shekhtman et al. [25]:

1

ktrans¼ R1

R0þ R2

R1

�������� (5)

From the constantly small value of jR2/R1j = 1/(kd + ktrans), thedesorption rate constant kd was found to be approximatelyconstant and much greater than the transformation constant ktrans

(around 160 s�1). In this case, the coefficient R1 determines theapparent equilibrium adsorption constant Kads = ka/kd.

Kads and ktrans are plotted in Fig. 9. We did not observe any largechange of the desorption rate constant (because the n-butaneapparent time delay on the catalyst surface, R2/R1, remained small,vide supra). A marked decrease in the value of the adsorptionconstant solely contributed to a significant decrease of theapparent equilibrium adsorption constant Kads with the pulsenumber in Fig. 7. This could be related to the fact that more thanhalf of the consumed butane is used to carburize the surface. Thebutane transformation constant ktrans shows a minor variationwithin the experimental error while remaining close to a value of1.5 s�1. The variation of ktrans can be related to the production of arange of n-butane derived products, such as isobutane (the maingas-phase product) and irreversibly adsorbed by-products such ascarbonaceous deposits or the formation of the oxycarbidic phase.

3.4. Discussion

The TPO and TAP kinetic data are in agreement with respect tothe fact that an approximately constant rate of isobutaneformation was obtained as a function of time on stream despitethe fact that different levels of carbon are present in the sample.This observation coupled with the kinetic parameters for thereaction on carbon-free and carbon-modified catalyst surfaceswhich were probed systematically using the TAP techniquestrongly supports the view that the main phase active in theisomerization of alkanes over this type of catalyst is essentially acompound of molybdenum of the type MoOx [4,10], MoO2(Hx)ac [6]or MoOxHy [28].

The proportion of butane going to carbonaceous deposits orforming the oxycarbide (i.e. the difference between the butaneconversion line and that of isobutane formation) was gradually

decreasing with the number of pulses during the TAP experiments(Fig. 7). This observation can be simply rationalized by the fact thatthe sites where the carbonaceous deposits accumulated weregetting saturated and/or the rate of carburization was decreasingwith increasing carbon content, possibly because the surface wasless getting less reactive. In the case of the ambient pressure work,the carbon deposition curve appeared quite sharp (Fig. 4), of coursebecause the quantity of carbon used in these experiments wasseveral orders of magnitude higher than that used in the TAP work.

While carbon was gradually incorporated into the structure andformed an oxycarbide-type phase (converting the MoOxHy phaseinto a MoOaCbHc phase) alongside the MoO2 phase [18,19], it isimportant to note that the carburization was not essential toobserve the hydroisomerization activity. On the contrary, somedata would suggest that a high content of carbon in the oxycarbidicphase would lead to a catalyst with a high hydrogenolysis activity[19,28] as in the case of pure carbides [32] and, therefore, to a lowerselectivity in skeletal isomerization.

4. Conclusions

TPO and TAP data has been used to show that the oxycarbidiccarbon content of molybdenum oxide-based phases is not relatedto the activity of the sample for the skeletal isomerization of n-butane. While the TAP data indicated marked changes in theadsorption properties of the sample due to varying exposure of thepre-reduced MoO3 sample to the butane-containing feed, theisobutane yield remained constant throughout the experiment. Inaddition, the capability of the multi-pulse TAP technique success-fully used to unravel the structure–activity relationship, even inthe case of a ‘‘breathing’’ catalyst which readily modifies its bulkstructure depending on the feed composition, has been demon-strated herein.

Acknowledgement

The EPSRC (through the Carmac project) are acknowledged forfunding.

References

[1] R. Burch, J. Chem. Soc., Faraday I 74 (1978) 2982.[2] Y. Ono, Catal. Today 81 (2003) 3.[3] A.P.E. York, C. Pham-Huu, P. Delgallo, E.A. Blekkan, M.J. Ledoux, Ind. Eng. Chem.

Res. 35 (1996) 672.[4] T. Matsuda, H. Kodama, H. Sakagami, N. Takahashi, Appl. Catal. A: Gen. 248 (2003)

269.[5] T. Matsuda, K. Watanabe, H. Sakagami, N. Takahashi, Appl. Catal. A: Gen. 242

(2003) 267.[6] P. Wehrer, L. Hilaire, E. Petit, Appl. Catal. A: Gen. 273 (2004) 249.[7] H. Al-Kandari, F. Al-Khorafi, H. Belatel, A. Katrib, Catal. Commun. 5 (2004) 225.[8] A. Bennada, A. Katrib, A. Barama, Appl. Catal. A: Gen. 251 (2003) 93.[9] F.C. Meunier, Chem. Commun. (2003) 1954.

[10] E. Torres-Garcia, G. Rodriguez-Gattorno, J.A. Ascencio, L.O. Aleman-Vazquez, J.L.Cano-Dominguez, A. Martinez-Hernandez, P. Santiago-Jacinto, J. Phys. Chem. B109 (2005) 17518.

[11] G. Boskovic, P. Putanov, K. Foettinger, H. Vinek, Appl. Catal. A: Gen. 317 (2007)175.

[12] G.A. Tsigdinos, W.W. Swanson, Ind. Eng. Chem. Res. 17 (1978) 208.[13] P. Delgallo, C. Pham-Huu, A.P.E. York, M.J. Ledoux, Ind. Eng. Chem. Res. 35 (1996)

3302.[14] P. Arnoldy, J.C.M. de Jonge, J.A. Moulijn, J. Phys. Chem. 89 (1985) 4517.[15] J.J. Sloczynski, Solid State Chem. 118 (1995) 84.[16] T. Ressler, R.E. Jentoft, J. Wienold, M.M. Gu1nter, O. Timpe, J. Phys. Chem. B 104

(2000) 6360.[17] F.C. Meunier, F. Cavallaro, T. Le Goaziou, A. Goguet, C. Rioche, Catal. Today 112

(2006) 64.[18] P. Delporte, F.C. Meunier, C. Pham-Huu, P. Vennegues, M.J. Ledoux, J. Guille, Catal.

Today 23 (1995) 251.[19] M.J. Ledoux, F.C. Meunier, A.O.I. Krause, B. Heinrich, C. Pham-Huu, E. Harling, Appl.

Catal. A: Gen. 181 (1999) 157.

A. Goguet et al. / Applied Catalysis A: General 344 (2008) 30–35 35

[20] T. Matsuda, H. Kodama, H. Sakagami, N. Takahashi, Micropor. Mesopor. Mater. 51(2002) 155.

[21] J.T. Gleaves, J.R. Ebner, T.C. Kuechler, Catal. Rev. Sci. Eng. 30 (1988) 49.[22] J.T. Gleaves, G.S. Yablonskii, P. Phanawadee, Y. Schuurman, Appl. Catal. A 160

(1997) 55.[23] J. Perez-Ramırez, E.V. Kondratenko, J. Catal. 250 (2007) 240.[24] J. Perez-Ramırez, E.V. Kondratenko, V.A. Kondratenko, M. Baerns, J. Catal. 229

(2005) 303.[25] S.O. Shekhtman, G.S. Yablonsky, J.T. Gleaves, R. Fushimi, Chem. Eng. Sci. 58 (2003)

4843.

[26] S.O. Shekhtman, G.S. Yablonsky, J.T. Gleaves, S. Chen, Chem. Eng. Sci. 54 (1999) 4371.[27] S.O. Shekhtman, G.S. Yablonsky, Ind. Eng. Chem. Res. 44 (2005) 6518.[28] F. Meunier, Ph.D. Thesis, Strasbourg, 1995.[29] T. Matsuda, Y. Hirata, H. Sakagami, N. Takahashi, Micropor. Mesopor. Mater. 42

(2001) 345.[30] C. Bouchy, C. Pham-Huu, B. Heinrich, C. Chaumont, M.J. Ledoux, J. Catal. 190

(2000) 92.[31] S.O. Shekhtman, N. Maguire, A. Goguet, R. Burch, C. Hardacre, Catal. Today 121

(2007) 255.[32] F.H. Ribeiro, M. Boudarta, R.A. Dalla Betta, E. Iglesia, J. Catal. 130 (1991) 498.