Journal of Colloid and Interface Science 299 (2006) 238–248www.elsevier.com/locate/jcis

Study of nickel nanoparticles supported on activated carbon prepared byaqueous hydrazine reduction

R. Wojcieszak, M. Zielinski, S. Monteverdi, M.M. Bettahar ∗

Catalyse Hétérogène, UMR 7565, Faculté des Sciences, Université Henri Poincaré, Nancy-I, BP 239, 54506 Vandoeuvre-lès-Nancy Cedex, France

Received 22 June 2005; accepted 31 January 2006

Available online 24 March 2006

Abstract

Nickel nanoparticles were obtained by the reduction in hydrazine aqueous media of nickel acetate as a precursor supported on activated carbonof high surface area. Classical catalysts using nickel acetate or nitrate were prepared for comparison. The catalysts were characterized by N2physisorption, H2-TPR, H2-adsorption, TPD, TEM, and XRD, and tested in the gas phase hydrogenation of benzene. Hydrazine catalysts werefound much more active in benzene hydrogenation than corresponding classically prepared catalysts. Remarkably, their reactivity is comparable(turn-over frequency of 0.2001–0.2539 s−1 at 393 K) to that of Pt classical catalysts supported on activated carbon in the same conditions.Evidence is given for the existence of the hydrogen spillover effect in benzene hydrogenation, not reported before in the literature. As a result ofthe hydrogen spillover effect, catalysts performances can be explained by a combination of surface metal atom reactivity, metal–support interactionstrength, and specific surface area extent. Maximum effect is observed with hydrazine preparation method, for 1% Ni content and nickel acetateas a precursor. Unexpectedly, it was also found that hydrazine preparation increases the specific area of the catalysts.© 2006 Elsevier Inc. All rights reserved.

Keywords: Nickel; Hydrazine; Catalysts; Activated carbon; Hydrogen chemisorption; Benzene hydrogenation

1. Introduction

The activity and selectivity of supported metal catalysts arestrongly influenced by the amount of metal, the size of dis-persed metal particles, the preparation method, and the sup-port composition [1,2]. To improve the catalyst activity andits durability, it is necessary to obtain a well-dispersed activephase in the catalyst [3,4]. The chemical species present on thecarrier surface interact, more or less, with the metal particlesdeposited thereon. The metal–support interaction (MSI) is usu-ally stronger when the metal particles are smaller [5]. For thecarriers employed in industrial catalysis the intensity of MSIdecreases from alumina to silica to active carbon [5]. In thecase of highly dispersed materials, the changes in the catalyticactivity were related to the combined effects of particle size,surface coverage with adsorbed species, and active site dimen-sion [6].

* Corresponding author.E-mail address: mohammed.bettahar@lcah.uhp-nancy.fr (M.M. Bettahar).

0021-9797/$ – see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.jcis.2006.01.067

Conventional supported metal catalysts are prepared by insitu reduction of metal salt. However, it is often difficult tocontrol the morphology of the final material. The control ofmetal particle nanosize is of the utmost importance for theperformance of any industrial catalyst based on supported met-als. The synthesis of the metal nanoparticles of uniform shapeand high purity has received considerable attention in the pasttwo decades due to their unusual properties compared to bulkmetal [7].

The metal nanoparticles are generally produced by the re-duction of metal compounds. The number of procedures isfairly large, for example photolytic reduction [8], radiolytic re-duction [9], solvent extraction reduction [10], microemulsiontechnique [11], alcohol reduction [12], chemical reduction inaqueous [13] and nonaqueous [14] media by hydrazine. Hy-drazine is a powerful strong reductant widely used in variouschemical operations. A series of striking results have been ob-tained where hydrazine is used as a reducing agent for the pro-duction of finely divided metals [15,16]. Degan and Macek [17]used hydrazine as a reducing agent to prepare nickel powdersin the submicrometer size range from nonaqueous solutions of

R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248 239

nickel salts. The rate and yield of the reaction were both en-hanced at higher reaction temperatures but were limited by therelatively low boiling point of water.

In the industrial and academic area a great attention is paidto the hydrogenation of benzene, the model aromatic feedstock[18–21]. This reaction has also been used as model reaction inheterogeneous catalysis by metals where metal–support interac-tions are involved [20–23]. The electronic and structural effectsof the MSI influence on the hydrogenation of benzene. This re-action is considered as a structure-sensitive reaction [24,25].

The use of activated carbon (AC) as a catalyst support isincreasing due to the advantages it offers as compared to tradi-tional oxides [26]. Cheap and relatively inert material, activatedcarbon exhibits stability in acidic or basic media, a high surfacearea and oxygen-containing functional groups. In metal–carbonsystems, intense interactions can be promoted and play an im-portant role conditioning both textural properties and reactivityof the catalyst [27]. Metal–carbon interactions attracted muchattention [26–30]. In this way, the thermal pre-treatment nature(neutral/reductive/oxidative) may change the surface area orporosity as well the activity or selectivity of the catalyst [27,28].Carbon atom dissolution in the metal [27,28] may cause the de-crease of the catalytic activity through site blockage [27].

Carbon is able to reduce the metal precursor introduced dur-ing the catalyst preparation [27,30].

We have undertaken a systematic study of nickel metalnanoparticles supported on silica of low surface area, preparedby reduction of nickel acetate by hydrazine in aqueous me-dia [13,31–33]. The reduction led to metal crystallites of var-ious size and shape according to the conditions of preparation.The shape influenced the surface and catalytic properties of themetal nanoparticles in the hydrogenation of unsaturated mole-cules.

In the present study we report results on the preparation,the characterization, and the surface and hydrogenating prop-erties of nonclassical catalysts supported on activated carbon.The catalysts were prepared by the chemical reduction withhydrazine in aqueous media using nickel acetate or nitrate asa precursor (1, 5, and 10 wt% Ni). Classical catalysts havealso been studied for comparison. The catalysts prepared werecharacterized by N2-physisorption, X-ray diffraction (XRD),transmission electron microscopy (TEM), H2-temperature pro-grammed reduction (TPR), and H2-adsorption and temperatureprogrammed desorption (TPD) techniques. Gas phase hydro-genation of benzene was chosen as a test-reaction. The hydro-gen spillover effect in this reaction was also examined.

The aim of this work is to examine the effect of the methodof preparation, the nickel content and the nature of metal pre-cursor on the surface, and catalytic properties of Ni catalystssupported on a commercial activated carbon. The results ob-tained are an original contribution to the literature corpus dataon nickel–carbon catalysts and benzene hydrogenation. Themethod of preparation used is original (hydrazine reduction ofNi2+ ions in aqueous media at 353 K) and leads to materialswith much better surface and catalytic performances than clas-sical preparation. Moreover, to our best knowledge, the hydro-genation of benzene on nickel catalysts supported on activated

carbon has not been reported. The results obtained also give ev-idence for the hydrogen spillover occurrence in this reaction forthe first time for a Ni-activated carbon system.

2. Experimental

2.1. Materials

The active carbon material used was a commercial acid-washed (Merck, ash content �3%). A fraction of 40 mesh wasused as a support. Ni(CH3COO)2·4H2O and Ni(NO3)2·6H2Owere purchased from Fluka (�99.0%) and Hopkins and Wil-liams Ltd., respectively.

For classical catalysts, the supported precursor was preparedby pouring the appropriate quantity of the support over nickelacetate solution (or nickel nitrate solution) with the appropriateconcentration. After 15 min of rotation under vacuum, the mix-ture was heated and evaporated for 1 h. The obtained solid wasdried at 383 K for 1 h (temperature ramp 3 K min−1). The cat-alysts were denoted as xNi/AC-A(N) where x, AC, and A(N)refer to nickel content (in wt%), activated carbon support, andacetate (or nitrate) precursor, respectively.

The nonclassical acetate catalysts were prepared under Aratmosphere according to Ref. [32]. In this method (denoted asH-method), the above supported precursor is reduced in aque-ous media according to the following reaction:

2Ni2+ + N2H4 + 4OH− → 2Ni0 + N2 + 4H2O. (1)

The reduction of the supported precursors was performed un-der an argon flow in a 3 necked reaction flask of 110 cm3

plunged in an oil bath for heating. The reaction flask was fit-ted with a reflux condenser, a thermocouple for measurementof the reaction temperature. The exit gases pass through a bub-bler equipped with a gas chromatograph for nitrogen analysis.A suspension of the supported nickel precursor was stirred andslowly heated to 353 K in an excess hydrazine solution for 4 hat pH 10–12. After reduction the solid obtained was filtratedoff, washed with distilled water and dried. The catalysts weredenoted as xNi/AC-AH where H refers to the hydrazine prepa-ration method. The other symbols are the same as describedabove.

A set of nonclassical acetate catalysts was prepared by si-multaneous introduction of nickel acetate and hydrazine atroom temperature in a suspension of activated carbon thenreducing at 353 K with the same procedure as above. Thispreparation is denoted as PH-method. The final materials weredenoted xNi/AC-APH where PH denotes the procedure of hy-drazine reduction used. The other symbols are the same asabove.

The nitrate precursor was not reduced by hydrazine even at373 K for several hours: no gas evolved during the reactionweather it was supported or not.

The activated carbon was also treated in the hydrazine mediaand dried in the same conditions described above. It is denotedACH.

No nickel precursor residue was found in the washing so-lution of the final solid filtrated. This means that all nickelintroduced was incorporated into the carbon support.

240 R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248

2.2. Textural and structural studies

The textural parameters of the materials were determined byN2 sorption at 77 K, using an automatic Thermoquest Sorp-tomatic 1990 apparatus. Specific surface area and specific porevolume were determined according to BET method. Dubinin–Radushkevich method was used to determine the microporevolume.

XRD patterns I (θ) and TEM images were recorded at a clas-sical θ/2θ diffractometer using CuKα radiation and PhillipsCM20 STEM, respectively. X-ray microanalysis chemical com-positions were determined using energy dispersive X-ray spec-troscopy mounted on the Phillips microscope.

2.3. Chemisorption and thermal treatments

Chemisorption, thermal treatments and H2-TPR experi-ments were carried out in a pulse chromatographic microreac-tor. The reactor was equipped with the catharometric detector ofa microchromatograph (Agilent G2890A) fitted with molecularsieve columns and MTI software. The experiments were carriedout with 0.1 g samples. The gases were purchased from air liq-uid. Oxygen traces were eliminated from argon (99.995%) andhydrogen (99.995%) using a manganese oxytrap (Engelhardt)whereas helium (99.999%) and hydrogen (99.99%) diluted inargon were used as received.

The H2-TPR study was carried out using H2/Ar (1000 ppmH2) mixture as a reductant (flow rate = 90 cm3 min−1). Thesample was first flowed with argon at room temperature for 1 hthen heated at a rate of 5 K min−1 to 1123 K under the reduc-tant mixture. Hydrogen consumption was measured with theAgilent G2890A microchromatograph operated at 333 K.

H2 chemisorption and H2-TPD studies were carried out onthe reduced samples. The reduction of the catalyst was per-formed for 3 h in a flow of pure hydrogen (flow rate = 100cm3 min−1). 0.1 g of a catalyst was placed in a quartz reactorand heated at a rate of 1 K min−1 to 623 K. The reduced sam-ple was treated in a flow of argon at the reduction temperaturefor 1 h (flow rate = 100 cm3 min−1) and then cooled to roomtemperature under argon atmosphere. After purging in theseconditions, the hydrogen adsorption on the reduced catalyst wascarried out using a mixture of H2/Ar (100 ppm H2). The reac-tant gas was injected in the reactor and the amount of hydrogenchemisorbed was detected every 2 min. After saturation, thesample was purged under argon (1 h), then the subsequent hy-drogen TPD was conducted with the same gas (100 cm3 min−1)in programmed temperature mode up to 1123 K, using a rampof 5 K min−1. The adsorbed and desorbed amounts of hydro-gen were detected with a thermal conductivity detector in theAgilent G2890A microchromatograph.

2.4. Catalytic activity

The benzene hydrogenation was performed in a quartz re-actor equipped with a thermocouple. 0.1 g of a catalyst wasplaced in the reactor and it was reduced in a flow of pure hydro-gen (flow rate = 100 cm3 min−1). After the reduction, carried

out at 623 K after heating with a ramp of 1 K min−1, the sam-ple was cooled to the desirable temperature. Gas mixture con-taining benzene and H2 (1 vol% of benzene) was prepared bypassage of H2 stream through benzene (Merck, >99%) placedin a saturator maintained at 278.4 K. The reactant mixture waspassed through the reactor with total flow rate of 50 cm3 min−1.The catalytic test was carried out at various temperatures on thesame catalyst. The sample was heated or cooled at a rate of10 K min−1. Benzene and the reaction products were analyzedeach 30 min with 5730A Hewlett–Packard gas chromatograph,operated at the programmed temperature and with the flameionization detector.

3. Results and discussion

All the catalysts were pre-treated under a hydrogen flow at623 K for 3 h before their study by XRD, TEM, H2-adsorption,or TPD or before testing in the gas phase hydrogenation of ben-zene.

3.1. Nickel ion reduction in hydrazine aqueous media

The nickel nitrate precursor did not react with hydrazinein the used conditions. In contrast, for the acetate precursor,gaseous nitrogen evolved and the solid obtained can be removedwith a magnet from the reduction flask. This is the result ofNi0 particle formation according to Eq. (1). Two modes of hy-drazine reduction were used (see Section 2): (i) the reductionof supported nickel acetate (H-method); (ii) the reduction ofdissolved nickel acetate in the presence of a suspension of acti-vated carbon (PH-method). In case of catalysts obtained by thePH-method, the reduction and growth of nickel particles takeplace in the solution. The next step is the deposition of metalspecies on the carbon surface. Contrary to that, the reductionof the impregnated catalysts (H-method) takes place directlyon the support surface. In this case the particle size and metaldispersion are determined by the previous impregnation anddrying steps. One can suppose that the impregnation permitsto obtain the smaller particles size. The degree of reduction bythe hydrazine may also be expected to be different accordingto the mode of reduction or nickel content. The structural andreducibility studies confirmed these predictions as reported be-low.

3.2. Textural and structural properties

The activated carbon used has a specific surface area of1073 m2 g−1 (Table 1) and a total pore volume of 0.596 cm3 g−1

comprising 66.4% of micropores. When the activated carbonwas treated in hydrazine media, the surface is not changed(1116 m2 g−1). The wet impregnation method causes thechanges in the texture of the final materials depending on themethod of preparation and nickel content (Table 1).

The chemical reduction with hydrazine causes a significantincrease of specific surface area, total pore volume and mi-croporosity which grows to 1694 m2 g−1, 0.743 cm3 g−1 and80.8%, respectively, for 1% Ni loading. The benefit is lesser

R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248 241

Table 1Characteristic of the catalysts

Catalyst BET surfacearea (m2 g−1

cat )% micropores(%)

H2ads at RT × 10−5

(mol g−1cat )

H2desa ×10−5

(mol g−1cat )

Metal surfacearea (m2 g−1

cat )Maximum degreeof reduction (%)

1Ni/AC-N 1071 65.6 0.17 32.6 0.13 –5Ni/AC-N 856 66.8 1.09 27.8 0.85 –10Ni/AC-N 524 63.9 1.98 28.4 1.55 51.2

1Ni/AC-A 1107 66.0 0.09 26.8 0.07 –5Ni/AC-A 705 62.3 0.17 17.8 0.13 81.410Ni/AC-A 487 65.5 0.34 21.4 0.27 72.1

1Ni/AC-AH 1694 80.8 0.18 37.6 0.14 –5Ni/AC-AH – – 0.21 49.2 0.16 76.910Ni/AC-AH 1056 65.5 0.31 40.2 0.24 22.9

1Ni/AC-APH 890 65.2 0.02 48.4 0.02 –5Ni/AC-APH – – 0.15 46.6 0.12 54.210Ni/AC-APH 971 65.7 0.26 49.4 0.20 35.8

AC 1073 66.4 0 8.71 – –ACH 1116 66.5 0 3.32 – –

a From the H2-TPD study.

with the Ni/AC-APH nonclassical catalysts: the specific areaof 10% Ni/AC-APH is only 971 m2 g−1 against 1056 for 10%Ni/AC-AH m2 g−1. For classical preparation, low nickel load-ing (1 wt%) almost does not change the specific surface area orpore volume. In contrast, larger Ni loadings (5 or 10 wt%) causea significantly reduction of the specific surface area which de-creases to 487 m2 g−1.

Several studies have shown that the specific area of carbonsis strongly decreased after impregnation with nickel in classi-cal supported catalysts [27,34]. The larger the nickel content,the larger the decrease of surface area [34]. For charcoal as asupport, the decrease in the specific surface area is attributedto pore blockage by nickel compounds in the supported precur-sor obtained [27]. This is probably the case of the Ni/AC-N orNi/AC-A catalysts (Table 1). Contrary to that, there is a strik-ing increase or almost stability of the specific area for hydrazinecatalysts (Table 1). The decrease of the specific area (10%) ob-served with the PH-method of preparation is small (10%) ascompared to classical materials. Thus, in hydrazine aqueousmedia, the surface area of the nickel precursors is enhanced orpractically restored. Such an effect has not been reported in theliterature. Complex surface reactions occur during the reductionprocess which need further work to be understood.

XRD study was undertaken for the estimation of phasespresent in the modified materials. XRD patterns for the cata-lysts with 1 and 5 wt% Ni did not show peaks which couldorigin from nickel in metallic form. However, one should re-member that XRD technique is not sensitive if the amount ofphase is too low (less than 5 wt%). Moreover, if the particlesize of nickel is very small, XRD patterns cannot detect them.The XRD pattern of the 10% Ni acetate catalysts showed twolarge reflection signals (as illustrated in Fig. 1) at 44.59 (111)and 51.90◦ (200) 2Θ characteristic of metallic nickel with a cfcstructure [35]. However, the hydrazine prepared catalyst signalsare less intense, suggesting a better metal dispersion (Fig. 1).

The TEM experiments also did not allow the observationof nickel nanoparticles with the lowest nickel contents. Metalnanoparticles were evidenced for 10% Ni content catalyst only.

Fig. 1. XRD patterns of 10Ni/AC-A and 10Ni/AC-AH catalysts.

In case of classical preparation, the nickel was found homoge-neously dispersed on the support surface (Fig. 2a). The shapeof the nickel particles is predominantly spherical. The particlespossess a mean size of 10–40 nm. In case of the 10% Ni cat-alyst prepared by the hydrazine chemical reduction, the TEMimage differs from that of the classical catalyst (Fig. 2b). Thenickel nanoparticles are practically not visible on this image,although the surface analysis confirmed the presence of nickelon the surface. Isolated aggregates are detected (Fig. 2b) whichgave rise to the XRD signal (Fig. 1). It could be supposed thatafter the reduction with hydrazine the nickel particles formedare very small (<5 nm).

3.3. H2-TPR profiles

The H2-TPR experiments of fresh AC carbon support orACH hydrazine treated carbon support do not give rise to hydro-gen consumption but to gasification, as expected according tothe literature data [36,37]. Indeed, the temperature profiles (notreported here) indicated the signals of hydrogen and methaneemission at 916 K. In the same conditions, gasification of thesupport as methane also occurs for the acetate catalysts but at a

242 R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248

(a)

(b)

Fig. 2. TEM images of 10Ni/AC-A (a) and 10Ni/AC-AH (b) catalysts.

lower temperature (603 K) and with a larger yield. The presenceof nickel upgrades the formation of methane. In contrast, for thenickel nitrate precursor, the carbon support is gasified to carbondioxide (large peak centered at 858 K and starting from 700 K)(profile not reported). Carbon dioxide would be due to the reac-tion between the support carbon atoms and the water moleculesproduced by the reduction of the nickel nitrate precursor [38].Gasification of the support was accompanied by hydrogen con-sumption for all the catalysts. The H2-TPR profiles obtained forthis consumption are reported in Fig. 3.

The profiles of the catalysts obtained from nickel nitrate(Fig. 3a) show two peaks of hydrogen consumption at ∼450and ∼700 K. The peak at 450 K is almost absent for 1% of

(a)

(b)

Fig. 3. H2-TPR profiles of 10Ni/AC-AH, 10Ni/AC-A, 10Ni/AC-N (a) and Ni/AC-APH (b) catalysts.

nickel content. The low temperature reduction peak is due to thereduction of NiO nickel oxide form, loosely attached to the sup-port. The high temperature reduction peak originates from thereduction of nickel species much more bonded to the active car-bon support or/and from the hydrogen spillover (see below). Ata low nickel loading, a great proportion of nickel atoms are inclose contact with the support, also the high temperature peakprevails. In contrast, at greater loadings, part of the precursornot directly or less attached to the support will probably moreeasily form NiO species which are reduced at lower tempera-tures. This explains increasing intensity of peak at 450 K at theexpense of that of peak at ∼700 K with increasing nickel con-tent. This also explains peak temperature decrease with increas-ing nickel loading. These reducibility properties are observedfor classical supported nickel catalysts [39–44].

The profiles of classical catalysts obtained from the nickelacetate (Fig. 3a) differ from that of nickel nitrate catalysts. Thepeaks are shifted to higher temperatures and the amounts ofhydrogen consumed are different. This is due to the differentnature of the nickel species involved. Indeed, for the nitrate pre-cursor, NiO species are intermediately formed by the heatingduring the reduction process [27] whereas, for the acetate pre-cursor, Ni2+ ion clusters are anchored at the support surface.The reducibility of these species is consequently different.

R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248 243

The profiles of hydrazine prepared catalysts are more com-plex. This is exemplified in Fig. 3b for the PH catalysts and inFig. 3a for 10Ni/AC-AH catalyst. The peaks correspond to thepresence of unreduced nickel species. These species may arisefrom an incomplete reduction of the supported Ni2+ ions bythe hydrazine [13,45] or from reoxidation of Ni0 species due tomoisture contamination. As a result, the presence of three formsof nickel species was observed. The very low temperature peakin the range 380–480 K may be due to isolated surface Ni2+ions. The two following peaks at higher temperatures (in theranges 520–630 and 650–800 K, respectively) are ascribed toreduction of nickel species more and more strongly bonded tothe support or/and also from the hydrogen spillover (see be-low). The temperature of these peaks decreases with increasingnickel content as a result of decreasing metal–support interac-tions. The profiles are quite similar for the H and PH-acetatecatalysts.

The amounts of hydrogen consumed (Table 1) were deter-mined from the surface area of the temperature peaks. Theseamounts could be due to metal reduction. However, remarkably,these amounts are greater (5–7 times) than the stoichiomet-ric amount of nickel in case of 1% Ni loaded catalyst. Thismeans that part of the hydrogen molecules are incorporated inthe carbon support through the spillover mechanism (see be-low) [46,47]. Spillover species are probably also produced incase of 5 or 10% nickel catalysts although the H/Ni ratio islower than 1. Indeed, experimental and theoretical thermody-namic calculations showed that a H/Ni of 0.2% at 298 K needsa pressure of hydrogen as high as 0.6 GPa.

Assuming that the amounts of hydrogen consumed are dueto metal reduction only, maximum nickel degree of reductionwas calculated (Table 1). It appears that maximum degree ofreduction decreases with increasing nickel content, althoughthe temperature of reduction decreases, as reported above (seeFig. 3a also). This would arise from the increase of the parti-cle size with increasing nickel content: large particles are moredifficult to reduce than smaller ones. The extent of the reduc-tion also changes with method of preparation of the catalystsand nature of the metal precursor.

3.4. H2-adsorption study

The accessible metal surface atoms of the catalysts weredetermined by H2-adsorption at room temperature after a hy-drogen treatment at 623 K for 3 h. Blank experiments showedthat the carbon support, treated or not with aqueous hydrazine,did not adsorb hydrogen. Assuming the stoichiometric adsorp-tion of hydrogen, i.e., that one molecule of hydrogen occupiestwo nickel atoms on surface, the total metal surface area (pergram of catalyst) was calculated. The results obtained are re-ported in Table 1. Several factors could explain the variationsobserved.

It can be seen that the total metal surface area increaseswith increasing nickel content. This may be ascribed to increas-ing number of nickel particles, roughly speaking. In case ofnonclassical preparation, the H-catalysts gave better metal sur-face area than the PH-catalysts. 1% Ni/AC-APH catalyst almost

does not adsorb hydrogen. In H-method, the metal dispersionis determined in the impregnation and drying steps giving riseto small metal particles. In contrast, in PH-method, it is deter-mined during the reduction step which rather takes place in thesolution and gives rise to bigger metal particles. For classicalcatalysts, nitrate precursor exhibits greater metal areas than ac-etate precursor. This may be due to a better reducibility of theformer, as showed the H2-TPR study.

Other factors could be involved in the determination of themetal surface area. The one could be nickel embedding by thecarbon support during the hydrogen pre-treatment at 623 K. Ineffect, it was shown that the H2 adsorption on nickel catalystsdeposited on charcoal was decreased after a pre-treatment underhydrogen at 773 K [27]. The authors attributed this to a poi-soning effect of the charcoal support. This action would be ac-complished by the carbon itself or/and additionally by mineralimpurities present in the support [27]. The second one couldbe the occurrence of the hydrogen spillover effect (see below).This effect results in the incorporation of hydrogen species inthe carbon frame, also the hydrogen would be partly fixed on thecarbon support and not on the metal phase only. Nickel embed-ding or hydrogen spillover effects both depend on the methodof preparation of the catalysts, the nature of the metal precursoror the nickel loading, and, indirectly, influences on the metalsurface area.

3.5. H2-TPD study

The H2-TPD study of activated carbon alone or catalyst wascarried out after a hydrogen treatment at 623 K for 3 h andhydrogen chemisorption at room temperature (see Section 2).Only hydrogen evolved at the outlet of the reactor. Hydrogenexhibits a main peak of desorption starting from 1000 K forall the materials and, in the presence of the metal phase only,a minor one at lower temperatures (∼760–800 K), as shownfor AC and 1Ni/AC-A materials in Fig. 4a. The temperature ofthe minor peak as well as the amounts of hydrogen desorbedvaried with the method of preparation, nature of the precursoror nickel content. This is reported in Fig. 4b for AH catalysts.The minor peak appears at almost the same temperature for theN- or A-classical catalysts (∼760 K) and at a roughly highertemperature for the H-catalysts (∼803 K for 1Ni/AC-AH). ThePH-method gave similar profiles as the H-method, but at highertemperatures. The amounts of hydrogen desorbed decrease withincreasing nickel content. Moreover, remarkably, the nonclas-sical catalysts desorbed greater amounts of hydrogen than theclassical ones (37.6–49.4 against 17.2–32.6 mol g−1

cat ). All theseresults may be attributed to the formation of different hydrogenadsorbed species, the structure of which depends on the kind ofthe nickel precursor, the nickel content or the method of prepa-ration.

The high temperature peak is due to the heat decomposi-tion of carbon functionalities without the help of the supportedmetal. The minor peak at 760–800 K is ascribed to hydrogenmolecules from the previous reduction treatment at 623 K re-tained at the metal–support interface [42,49] or incorporatedin the support. Indeed, the amounts desorbed are larger than

244 R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248

(a)

(b)

Fig. 4. H2-TPD profiles of classical Ni/AC-A (a) and Ni/AC-AH (b) catalysts.

that adsorbed at room temperature (Table 1). Moreover, a closeexamination of the amounts of hydrogen desorbed shows thatthe H/Ni ratio may exceed 1 (e.g., 3.8–4.7 for 1% Ni con-tent). This excess hydrogen suggests that a part of the reactantmolecules, if not all, is strongly retained on the active carbonsupport. This may also hold for 5 or 10% Ni content to a certainextent. In other words, hydrogen molecules from the reduc-tive atmosphere probably dissociated on the metal phase thenspilled over onto the carbon support. This is in good correla-tion with the H2-TPR study which showed that hydrogen wasincorporated in the carbon support during the reduction step.In case of Pt/active carbon catalysts reported in Ref. [48], animportant peak starting at about 1125 K, with a maximum des-orption rate at 1325 K, and a minor peak at 750 K were observedafter a heat treatment under hydrogen atmosphere. They wereascribed to heat decomposition of carbon functionalities and tohydrogen species strongly adsorbed on the carbon support, re-spectively [48]. Hydrogen spillover on activated carbons hasbeen evidenced by several authors [46,47].

3.6. Catalytic activity

After a H2 thermal treatment, the classical or nonclassicalcatalysts became active and selective in the gas phase hydro-

Fig. 5. Conversion of benzene for Ni/AC-N catalysts.

Fig. 6. Conversion of benzene for Ni/AC-A catalysts.

genation of benzene to cyclohexane. No activity was observedwithout this treatment. This is in good agreement with previ-ous studies in our group on Ni/SiO2 catalysts [13,31–33]. Fornonclassical catalysts the effect of the heat pre-treatment was toremove the acetate fragment embedding the nickel phase duringthe hydrazine reduction: it liberated the access to active sites forthe reactant molecules [13,31–33]. The activated carbon alone,previously treated or not in aqueous hydrazine media, was in-active. The results are reported in Figs. 5–8 and Tables 2 and 3.

The catalysts exhibited a maximum of activity as a func-tion of the reaction temperature. In addition, each maximumcould be traversed from either the high or low temperatureside without any loss of activity. This means that no deacti-vation occurred in the used conditions. Maximum of activitywith temperature in benzene hydrogenation has been frequentlyreported [13,19,20,47]. This is accounted for a positive influ-ence of the reaction kinetics at lower temperatures whereas,at higher temperatures, the thermodynamic effect of the reac-tion was prevailing and then conversion decreased [50]. Basedon benzene chemisorption studies, the maximum of activity isalso attributed to a decrease of the surface coverage by benzenemolecules at the higher temperatures [51,52]. Temperature ofthis maximum depends on the nickel loading, the nature of the

R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248 245

Fig. 7. Conversion of benzene for the nonclassical Ni/AC-AH catalysts.

Fig. 8. Conversion of benzene for 1% Ni acetate catalysts.

metal precursor and the method of preparation. The nonclas-sical H-catalysts exhibit a high activity with a maximum at alower temperature (420 K, whatever the nickel content is) thanthe classical ones (460 K). However, maximum activity shiftsto a higher temperature (460 K) also for the PH-nonclassicalcatalysts. In case of classical catalysts, maximum temperatureis lower (420 K) for the nitrate precursor than for the acetateprecursor (460 K) when the nickel content is 10%. It shifts tohigher temperatures (460 K) at lower loadings.

The conversion increases with increasing catalyst nickelcontent (Figs. 5–8). The hydrazine-prepared catalysts exhibitbetter conversion than classical catalysts. This is notably thecase for 1% Ni content: the conversion is 30% from 344 K with1Ni/AC-AH and only 2% with 1Ni/AC-A (Fig. 8). Mediumconversions are obtained with the APH-catalysts: 10% for1Ni/AC-APH at the same temperature (Fig. 8). As to conven-tional catalysts, they were more active (∼100% benzene con-version from 440 K) for the nickel nitrate precursor (Fig. 5)than that for the nickel acetate precursor (maximum benzeneconversion of 30% at 348 K) (Fig. 6).

The intrinsic rate (mol per min per gram of nickel) is a goodindication of the influence of the nickel content on the reaction

Table 2Catalytic activity in benzene hydrogenation at 373 K

Catalyst Benzeneconversion (%)

Specific rate ×10−3

(mol min g−1Ni )

TOF(molec Bz site−1 s−1)

1Ni/AC-N 3.35 0.610 0.03015Ni/AC-N 9.38 0.332 0.012710Ni/AC-N 37.70 0.659 0.0277

1Ni/AC-A 1.20 0.229 0.02175Ni/AC-A 4.84 0.173 0.042410Ni/AC-A 3.62 0.064 0.0157

1Ni/AC-AH 30.88 5.485 0.25395Ni/AC-AH 32.74 1.115 0.221210Ni/AC-AH 42.18 0.516 0.2001

1Ni/AC-APH 10.56 1.885 –5Ni/AC-APH 20.23 0.688 0.191110Ni/AC-APH 11.01 0.188 0.0618

AC 0 – –ACH 0 – –

Table 3Rate of benzene hydrogenation. Effect of dilution of the catalyst by the carbonsupport

Catalyst Specific rate ×10−3 (mol min g−1Ni )

348 K 373 K 398 K 423 K 448 K

10Ni/AC-N 0.135 0.659 1.377 1.722 1.772AC/10Ni/AC-N (1:1) 0.674 2.535 3.280 3.358 3.358

10Ni/AC-A 0.018 0.064 0.173 0.339 0.356AC/10 Ni/AC-A (1:1) 0.093 0.140 0.209 0.232 0.550

kinetics. The results obtained at 373 K show (Table 2) a muchhigher rate for the hydrazine than the classical acetate catalysts(up to 25 times for 1% Ni content). The H-method of prepa-ration is better than the PH-method. For classical catalysts thebest activity is obtained with the nitrate than the acetate pre-cursor, up to 10 times greater. On the other hand, for the acetatecatalysts, the intrinsic rate increases with decreasing nickel con-tent, most probably as a result of decreasing metal particle size(see above). However, this is not the case for nitrate catalystswhich activity passes through a minimum. Other factors actedso as to determine the catalytic activity.

The use of the turn-over frequencies (TOFs) allows one tomake a valuable comparison of the intrinsic activity of thenickel surface atoms. The results obtained at 373 K are reportedin Table 2. Remarkably, the highest TOFs are obtained with thehydrazine-prepared catalysts. It is worth noting that the TOF ofbenzene hydrogenation at 373 K on 5% Pt/AC is 0.494 s−1, avalue quite comparable to that observed for 5Ni/AC-AH whichis 0.2212 s−1. Also, one may conclude that the hydrazine prepa-ration allows obtaining performing nickel catalysts for benzenehydrogenation.

The TOFs increased with decreasing nickel content for hy-drazine catalysts. This could be correlated to the decrease ofboth the particle size and degree of reduction. This is in goodaccord with the results reported for benzene hydrogenation onnickel–silica catalysts [49,53]. The explanation proposed by theauthors is that unreduced nickel decreases the size of the nickelparticles required for the adsorption of benzene in a planar

246 R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248

mode. Unfortunately, this correlation does not hold for classi-cal catalysts: the TOFs pass through a maximum or a minimumaccording to the nature of the metal precursor (Table 2). Otherfactors acted so as to determine the catalytic activity.

The discussion on the chemisorption study showed that thesurface nickel atoms reactivity toward hydrogen is strongly af-fected by the nickel–carbon interactions, the degree of reduc-tion of nickel or the hydrogen spillover effect. These parame-ters also may influence the hydrogenation chemical processes.This was reported for acetone hydrogenation on nickel cata-lysts impregnated on charcoal: the activity was found to sharplydecrease after activation under a hydrogen flow [27]. For ben-zene hydrogenation on nickel–silica catalysts, it was reportedthat the activity of the surface nickel atoms might be modifiedwhen hydrogen spillover is involved [49,53]. In such a case,an additional activity is observed due to benzene hydrogena-tion on support by protons produced from spillover hydrogen.This was shown for the same reaction over Pt/activated carboncatalyst [46]. Hydrogen spillover effect would also be involvedhere, as reported below.

3.7. Spillover hydrogen study

The surface diffusion of active species plays an importantrole in reactions on metal catalysts. Especially, migration of hy-drogen atoms from a metal, which is active in the dissociativeadsorption of hydrogen, to an oxide or carbon surface, whichitself has no activity for dissociative hydrogen adsorption, isvery important [54,55]. It has been claimed that the hydrogenspillover plays a very important role in aromatic hydrogena-tion on metal supported catalysts. For example, Teichner et al.[56] found that alumina itself is active for benzene hydrogena-tion due to the spillover hydrogen introduced by the platinumcatalyst supported on alumina. Actually, there is a lot of exper-imental evidence that hydrogen, which has been activated onmetallic sites, is able to spill over onto another phase whichplays a role of hydrogen acceptor. Various materials can act asthis hydrogen acceptor, for example: SiO2, Al2O3, or activatedcarbon. To our best knowledge, hydrogen spillover effect onactivated carbon supported nickel catalysts has not been inves-tigated for benzene hydrogenation. This prompted us to studythis reaction over the catalysts we prepared.

A spillover phenomenon is postulated when the dilution ofcatalyst increases its activity [57–65]. We have investigated thehydrogen spillover in benzene hydrogenation using the me-chanical mixtures of 10Ni/AC-N and 10Ni/AC-A catalysts withactivated carbon. The results of the dilution ratio (weight of cat-alyst/weight of catalyst+activated carbon) effect on the activityare given in Fig. 9 and Table 3.

The activity of 10Ni/AC-N catalyst increases with the dilu-tion ratio, in good accord with the hydrogen spillover effect.The maximum of benzene conversion at 448 K is observed forthe mixture with 0.5 dilution ratio (Fig. 9). The activity of thehybrid catalyst is twice higher than that of undiluted whichcould be ascribed to the hydrogen migration process which iscontrolled by a distance factor [55]. Dilution is believed to in-crease the number of H species stabilized by the active carbon

Fig. 9. Effect of the catalyst dilution by the support on the rate of benzenehydrogenation at 448 K.

by increasing the available surface area and, in the meantime,lowers the recombination of these species in molecular hydro-gen. At a certain point, the additional surface area of the diluentbrings no further benefit to the hydrogen uptake. At higher di-lution ratio, the number of H spillover acceptor sites stabilizedby the support increases but the generation of the H spilloverdecreases as the number of Ni atoms decreases [54].

The effect of the reaction temperature on the catalyst dilutionwas investigated. The results obtained (Table 3, 10Ni/AC-N)show that the spillover effect decreases with increasing reactiontemperature: the ratio between the activity of diluted catalystand the activity of undiluted catalyst decreases from 5.0 to 2.0when the reaction temperature is increased from 348 to 448 K.Since aromatic hydrocarbon hydrogenation on metal supportedcatalysts is believed to occur on the support, at least partly[46,51,52,55], this suggests that the reaction rather occurs onthe metal phase when the reaction temperature is increased.This is consistent with chemisorption studies showing that aro-matic hydrocarbons are only weakly adsorbed on the supportbut much more strongly on the metal atoms [51,52]. However,hydrogen spillover effect could depend on other factors such assurface coverages, bonding energies of molecules to the surfaceor surface contaminants [55] which are also influenced by thereaction temperature.

The support dilution effect was also verified with the leastactive 10Ni/AC-A catalyst (Table 3): the rate of benzene con-version is multiplied by 5.2 at 348 K. This effect similarlydecreases with increasing reaction temperature. It sharply di-minishes to 0.68 at 423 K.

These results are in good accord with the H2-TPR and H2-TPD experiments reported above which clearly showed that Hspillover species could be incorporated in the carbon frame forboth classical and nonclassical catalysts. We infer that hydrogen

R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248 247

spillover effect also holds for the hydrazine catalysts in benzenehydrogenation. They exhibit highest activity in this reaction.

Recall that the hydrogen spillover consists of three steps: thedissociation of hydrogen on metal atoms, then the migration ofthe H atoms formed onto the support, and finally their spreadingover the support. One may conclude that the intensity of this ef-fect is certainly a function of the reactivity of metal atoms, thecarbon–metal contact and the support specific surface area [47]which, in turn, depend on the method of preparation, the na-ture of the metal phase precursor and the metal content. Nickelphase in 1% Ni/AC-AH exhibited very small metal particles(attested by the XRD and TEM studies) and very high specificsurface area (Table 1) and strongly interacted with the support(attested by the H2-TPR experiments). This explains the high-est performances observed for the catalyst. No direct correlationwas found between catalytic activity and metal surface area, asreported above, because metal–carbon interactions and extentof specific surface area have also to be accounted for.

4. Conclusions

Hydrazine catalysts were found much more dispersed andmore active in benzene hydrogenation than corresponding clas-sically prepared catalysts. Better performances are obtainedwhen hydrazine reduced supported Ni2+ ions than unsupportedNi2+ ions in a suspension of the support. Strikingly, in ben-zene hydrogenation, the reactivity of the metal surface atoms ofthe nonconventional catalysts is comparable to that of Pt classi-cal catalysts supported on activated carbon: e.g., 5% Ni/AC-AHexhibited a turnover frequency of 0.2212 s−1 at 393 K, a valuequite comparable to that of 5% Pt/AC (0.4694 s−1). High dis-persion and catalytic performances are favored by a low nickelcontent and by nickel nitrate rather than by nickel acetate pre-cursor. Diluting the catalyst by the support led to increasedactivity in benzene hydrogenation as a result of the existencethe hydrogen spillover effect. This was not reported before inthe literature for nickel catalysts supported on activated carbon.Hydrogen spillover species were also found to be incorporatedin the carbon frame during the catalyst H2/623 K pre-treatment.The discussion of the results obtained confirmed that hydro-gen spillover effect is the driving force in the surface reactivityof nickel catalysts supported on activated carbon. Its intensityis the optimum combination of metal atoms reactivity, metal–support interaction strength and specific surface area extent.

Finally, it has to be underlined that hydrazine catalysts ex-hibit a specific surface area 1.5–2 times larger than that ofclassical catalysts. This interesting result needs further work tobe understood.

References

[1] S.J. Tautster, S.C. Fung, J. Catal. 5 (1978) 29.[2] C.H. Bartholomew, Mustard, J. Catal. 67 (1981) 186.[3] G. Schmid, Chem. Rev. 92 (1992) 1709.[4] F.M. Bautista, J.M. Campelo, A. Garcia, R. Guardeno, D. Luna, J. Mari-

nas, J. Catal. 125 (1990) 171.[5] N. Pernicone, Cattech 7 (6) (2003) 196.[6] G.A. Martin, J.A. Dalmon, C. Mirodatos, in: Proceedings of the Eighth In-

ternational Congress on Catalysis, vol. IV, VCH, Weinheim, 1984, p. 371.

[7] D. Poondi, J. Singh, J. Mater. Sci. 35 (2000) 2467.[8] S. Remita, M. Mostafavi, M.O. Delcourt, Radiat. Phys. Chem. B 47 (1996)

275.[9] J.H. Hodak, A. Henglein, M. Giersig, G.V. Hartland, J. Phys. Chem. 104

(2000) 11708.[10] M. Brust, M. Walker, D. Bethell, D.J. Schiffrin, R. Whyman, Chem. Com-

mun. 801 (1994).[11] K.O. Asare, F.J. Arriagada, Ceram. Trans. 12 (1990) 3.[12] H.H. Huang, X.P. Ni, G.L. Loy, C.H. Chew, K.L. Tan, F.C. Loh, J.F. Deng,

G.Q. Xu, Langmuir 12 (1996) 909.[13] A.G. Boudjahem, S. Monteverdi, M. Mercy, M.M. Bettahar, J. Catal. 221

(2004) 325.[14] S.H. Wu, D.H. Chen, J. Colloid Interface Sci. 259 (2003) 28,216.[15] J.P. Chen, L.L. Lim, Chemosphere 49 (2002) 363.[16] U. Nickel, A.Z. Castell, P. Karin, S.A. Schneider, Langmuir 16 (2000)

9087.[17] A. Degan, J. Macek, Nanostruct. Mater. 12 (1999) 225.[18] A. Stanislaus, B.H. Cooper, Catal. Rev. Sci. Eng. 36 (1994) 75.[19] M.A. Keane, J. Catal. 166 (1997) 347.[20] R. Molina, G. Poncelet, J. Catal. 199 (2001) 162.[21] B. Coughlan, M.A. Keane, Zeolites 11 (1991) 12.[22] L. Daza, B. Pawelec, J.A. Anderson, J.L.G. Fierro, Appl. Catal. 87 (1992)

145.[23] R. Burch, R. Flambard, J. Catal. 85 (1984) 16.[24] G.A. Martin, J.A. Dalmon, J. Catal. 75 (1982) 233.[25] P. Marecot, E. Paraiso, J.M. Dumas, J. Barbier, Appl. Catal. 74 (1991)

261.[26] L.R. Radovic, F. Rodriguez-Reinoso, in: P.A. Thrower (Ed.), Chemistry

and Physics of Carbon, vol. 25, Dekker, New York, 1997, p. 243.[27] L.M. Gandia, M. Montes, J. Catal. 145 (1994) 276.[28] C. Prado-Burguette, A. Linares-Solano, F. Rodriguez-Reinoso, C. Salinas-

Martinez de Lecc, J. Catal. 128 (1991) 397.[29] L.C. Isett, J.M. Blakeley, Surf. Sci. 58 (1976) 397.[30] I.F. Silva, F.M.B. Fernandes, L.S. Lobo, in: Actas del XIII Simposio

Iberoamericano de Catalisis, vol. 2, Segovia, 1992, p. 841.[31] A.G. Boudjahem, S. Monteverdi, M. Mercy, J. Ghanbaja, M.M. Bettahar,

Catal. Lett. 97 (3–4) (2004) 177.[32] A.G. Boudjahem, S. Monteverdi, M. Mercy, M.M. Bettahar, Langmuir 20

(2004) 208.[33] A.G. Boudjahem, S. Monteverdi, M. Mercy, M.M. Bettahar, Appl. Catal.

A 250 (2003) 49.[34] D. Mehandjiev, E. Bekyarova, M. Khristova, J. Colloid. Interface Sci. 192

(1997) 440.[35] 1999 JCPDS 01-1260.[36] V. Siva Kumar, B.M. Nagaraja, V. Shashikala, A.H. Padmasri, S. Shakun-

tala Madhavendra, B. David Raju, K.S. Rama Rao, J. Molec. Catal. A 223(2004) 313.

[37] J. Krishna Murthy, S. Chandra Shekar, V. Siva Kumar, K.S. Rama Rao,Catal. Commun. 3 (2002) 145.

[38] A. Bagreev, J. Lahaye, V. Strelko, PICS 119, Materiaux Carbones Adsor-bants, Zakopane, 1994, p. 195, in: D. Mordenti, D. Brodzki, G. Djega-Mariadassou (Eds.), J. Solid State Chem. 141 (1998) 114.

[39] C.H. Bartholomew, R.B. Pannell, R.W. Fowler, J. Catal. 79 (1983) 34.[40] J.T. Richardson, M. Lei, B. Turk, K. Forster, M.V. Twigg, Appl. Catal. 110

(1994).[41] R. Molina, G. Poncelet, J. Catal. 173 (1998) 257.[42] J.J. Sholten, A.P. Pijpers, A.M.L. Husting, Catal. Rev. Sci. Eng. 27 (1985)

151.[43] C. Li, Y.-W. Chen, Thermochim. Acta 256 (1995) 457.[44] J. Zielinski, J. Catal. 76 (1982) 157.[45] Y.D. Li, L.Q. Li, H.W. Liao, H.R. Wang, J. Mater. Chem. 9 (1999) 2675.[46] S.T. Srinivas, P.K. Rao, J. Catal. 148 (1994) 470.[47] A.D. Lueking, R.T. Yang, Appl. Catal. A General 265 (2004) 259.[48] F. Coloma, A. Sepuvelda-Escribano, J.L.G. Fierro, F. Rodriguez-Reinoso,

Appl. Catal. 150 (1997) 165.[49] A. Aguinaga, J.C. De la Cal, J.M. Asua, M. Montes, Appl. Catal. 51 (1)

(1989) 1.[50] P. Antonucci, N. Van Truong, N. Giordano, R. Maggiore, J. Catal. 75

(1982) 140.

248 R. Wojcieszak et al. / Journal of Colloid and Interface Science 299 (2006) 238–248

[51] S.D. Lin, M.A. Vannice, J. Catal. 143 (1993) 563.[52] I. Ionnanides, X.E. Verykios, J. Catal. 143 (1993) 175.[53] J.A. Anderson, L. Daza, J.L. Fierro, T. Rodrigo, J. Chem. Soc. Faraday

Trans. 89 (1993) 3651.[54] F. Benseradj, F. Sadi, M. Chater, Appl. Catal. A 228 (2002) 135.[55] W.C. Conner, J.L. Falconer, Chem. Rev. 95 (1995) 759.[56] S.T. Teichner, A.R. Mazabrard, G. Pajonk, G.E.E. Gardes, C. Hoang-Van,

J. Colloid Interface Sci. 58 (1977) 71.[57] A.M. Sica, E.M. Valles, C.E. Giola, J. Catal. 51 (1978) 115.

[58] V. Rahaman, M.A. Vannice, J. Catal. 143 (1991) 554.[59] S. Ceckiewicz, B. Delmon, J. Catal. 108 (1987) 294.[60] J. Wang, L. Huang, Q. Li, Appl. Catal. A 175 (1998) 191.[61] R. Wojcieszak, S. Monteverdi, M. Mercy, I. Nowak, M. Ziolek, M.M. Bet-

tahar, Appl. Catal. A 268 (2004) 241.[62] M.G. Yang, I. Nakanamura, K. Fujimoto, Appl. Catal. A 127 (1995) 115.[63] B. Sen, J.L. Falconner, J. Catal. 117 (1984) 404.[64] S.D. Lin, M.A. Vannice, J. Catal. 143 (1993) 539.[65] S.D. Lin, M.A. Vannice, J. Catal. 143 (1993) 554.