Characterization of the morphology of Pt clusters ... · Characterization of the morphology of Pt clusters incorporated in a KL zeolite by vapor phase and incipient wetness impregnation

Applied Catalysis A: General 188 (1999) 79–98

Characterization of the morphology of Pt clusters incorporated in a KLzeolite by vapor phase and incipient wetness impregnation. Influence of

Pt particle morphology on aromatization activity and deactivation

Gary Jacobs, Firoz Ghadiali, Adriana Pisanu,Armando Borgna1,Walter E. Alvarez2, Daniel E. Resasco∗School of Chemical Engineering, University of Oklahoma, Norman OK, 73019 USA

Received 19 February 1999; accepted 14 April 1999

Abstract

Two series of Pt/KL catalysts with varying metal loading were synthesized by the methods of incipient wetness impregnation(IWI) and vapor phase impregnation (VPI) to compare the effects of the different morphologies that result when the metalloading and, in particular, the preparation method are varied. Catalysts were characterized by a variety of techniques. TEMand DRIFTS studies indicated that on the low-loading samples the majority of particles were located inside the channels ofthe L-zeolite. In agreement with recent studies, the DRIFTS results evidenced the formation of Pt carbonyls, which furthersupport the presence of very small particles. EXAFS and TEM showed that the VPI catalysts resulted in smaller particlesthan the catalysts prepared by the IWI method. In addition, EXAFS demonstrated for this series a higher degree of interactionwith the L-zeolite framework oxygen atoms. Pulse testing of the methylcyclopentane ring opening showed that the very smallclusters produced by the VPI preparation did not result in collimation of the MCP molecule, implying that the reactants andproducts can easily diffuse over the Pt cluster. This is in contrast with the particles produced by the IWI method, whichclearly displayed a collimation effect. The characteristic morphology produced by the VPI method was found to improve theperformance of the catalyst under clean and sulfur-poisoned conditions, enhancing the catalyst’s resistance to the formationof coke and decreasing the particle agglomeration rate. ©1999 Elsevier Science B.V. All rights reserved.

Keywords:Pt/KL; Aromatization;n-hexane; EXAFS; DRIFTS; MCP; Vapor phase impregnation

∗ Corresponding author. Tel.: +1-405-325-4370; fax: +1-405-325-5813E-mail address:[email protected] (D.E. Resasco)

1 Permanent address: INCAPE, Santiago del Estero 2654(3000)Santa Fe, Argentina.

2 Permanent address: INTEMA, Juan B. Justo 4302, (7600) Mardel Plata, Argentina.

1. Introduction

The relationships between the structure and prop-erties of the Pt/KL zeolite catalysts have attracted theattention of researchers for almost two decades. Al-though their exceptionally high activity and selectiv-ity for the aromatization ofn-hexane to benzene arenowadays well established [1–5], the fundamental rea-son for these unique properties is not completely un-derstood [6,7]. Some authors have ascribed the unique

0926-860X/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved.PII: S0926-860X(99)00235-5

80 G. Jacobs et al. / Applied Catalysis A: General 188 (1999) 79–98

features of the Pt/KL catalysts to structural parametersof the L zeolite [8,9]; others have focused on its basic-ity [10], or a combination of both. Other authors haveproposed that the small Pt particles inside the chan-nels of the basic zeolite are electron-rich, and conse-quently exhibit unique catalytic properties [11]. Morerecent studies have proposed that the L zeolite struc-ture and its lack of acidity inhibit coke formation bybimolecular encounters [12,13].

Despite the different interpretations found in the lit-erature about the fundamental reason for the high ac-tivity and selectivity, most authors agree that an effec-tive Pt/KL catalyst should have as much Pt as possibleinside the channels of the zeolites [14,15]. However,even when most of the particles reside in the zeolite,different particle sizes and shapes could result. Ob-viously, if the Pt particles are located inside the zeo-lite, they must be smaller than the channel dimensions(i.e. lobe-shaped cages : 1.1 nm in diameter, 0.6 nmin length, apertures of 0.71 nm. [16]). However, it isnot known what morphology they may adopt and whatfraction of the channel may be blocked. The informa-tion obtained from EXAFS varies significantly withdifferent preparations and pre-treatments. For exam-ple, several studies have reported average coordina-tion numbers of about 4 [17,18]. The correspondencebetween coordination number and number of atomsin the particle strongly depends on the assumed mor-phology of the metallic particle. For spherical mor-phology in an FCC metal, a coordination number of 4would correspond to five atoms. However, this num-ber may represent as many as 12 atoms if the mor-phology is that of a (111) disk [19]. Other studieson Pt/KL catalysts have reported coordination num-bers between 5 and 6, which in spherical morphologywould correspond to 15–20 atoms [20]. One mightwonder whether these differences could significantlyaffect catalytic behavior. In other words, the questionthat remains unanswered is whether there is a particlesize effect, even when most of the particles are insidethe zeolite.

To address this question, we have investigated twoseries of Pt/KL catalysts, with varying Pt loading, pre-pared by two different methods. Our goal was to havethe majority of the Pt particles inside the zeolite, whilevarying their size and morphology. Therefore, we keptthe metal loading at or below 2.5 wt% and used twodifferent preparation methods. The first method was

the standard incipient wetness impregnation (IWI),which is the preparation preferred in most studies onPt/KL catalysts [21]. The second method was the va-por phase impregnation (VPI) [22], sometimes alsoreferred to as chemical vapor deposition (CVD) [23].

We anticipated that these two families of catalystswould only exhibit small variations in microstruc-ture. Therefore, in order to properly characterizethese variations, we have used a combination oftechniques, including EXAFS, DRIFTS, TEM, hy-drogen chemisorption, and probe reactions. We haveattempted to obtain a detailed description of the mi-crostructure of these catalyst series and analyze theconsequences of the subtle differences observed oncatalytic behavior (aromatization activity, selectivity,and deactivation).

2. Experimental section

2.1. Catalyst preparation

The K-LTL zeolite (series TSZ-500, BET area292 m2/g, SiO2/Al2O3 ratio = 6) was provided byTosoh Company. Before addition of the metal, theK-LTL zeolite was dried in an oven at 110◦C for 12 hand calcined at 400◦C for 5 h.

The incorporation of the platinum was realized bytwo different methods, IWI and VPI. In the first case,the dry support was impregnated with an aqueous so-lution of platinum salt tetraammineplatinum(II) nitratefrom Alfa. To achieve incipient wetness a liquid/solidratio of 0.69 cm3/g was used. After impregnation, thesamples were dried overnight at 110◦C. Subsequently,they were heated at a rate of 3◦C/min up to 350◦Cin an air flow of 100 cm3/min gcat. They were kept atthe same temperature and air flow for 2 h. The secondmethod was VPI [22]. In this case, the KL zeolite waspre-calcined for 5 h at 400◦C to remove chemisorbedwater. In an inert atmosphere, Pt(AcAc)2 was phys-ically well mixed with the KL, and the solid mix-ture was loaded into a reactor tube. The reactor tubewas sealed under vacuum and evacuated overnight to10−5 Torr. The catalyst was then slowly ramped to60◦C and held there for 1 h to remove traces of water,and then ramped again to 80◦C and held again for 1 hto remove water. After further ramping to 100◦C, thecatalyst was held for 1 h to sublime the Pt(AcAc)2,

G. Jacobs et al. / Applied Catalysis A: General 188 (1999) 79–98 81

at which temperature the pressure increased substan-tially. After sublimation, the catalyst was ramped to130◦C and held for 15 min to ensure that virtually allof the Pt(AcAc)2 was sublimed. The reactor tube wascooled to room temperature, and the sample was re-moved. At this point, the sample was light yellow incolor, indicating that the Pt(AcAc)2 did not decom-pose during the procedure. To decompose the platinumprecursor, the sample was ramped to 350◦C in flow ofair for 2 h and calcined at that temperature for 2 h.

The 1% Pt/SiO2 catalyst used as reference materialwas prepared by IWI on a silica gel support (from W.R.Grace, grade 923, surface area 450 m2/g ), using anaqueous solution of H2PtCl6·xH2O from Aldrich anda liquid/solid ratio of 0.63 cm3/g. The sample was firstdried overnight at room temperature and then during8 h at 120◦C. Finally, it was calcined in flow of air at400◦C for 2 h and reduced at 500◦C for 2 h under flowof H2.

The Pt loadings present in each sample were mea-sured by ICP analysis after complete dissolution of thesamples in aqua regia/HF solution at Galbraith Labo-ratories. In all cases, the measured loading was prac-tically the same as the corresponding nominal load-ing. In the paper, the catalysts are identified accordingto the method of preparation and nominal Pt loading,e.g. IWI-1 represents the sample prepared by incipientwetness impregnation with 1 wt % Pt.

2.2. Catalyst characterization

Hydrogen chemisorption measurements were con-ducted on all samples in a static volumetric adsorp-tion Pyrex system, equipped with a high capacity, highvacuum station that provided vacuum of the order of10−9 Torr.

Transmission electron microscopy (TEM) imageswere obtained in a JEOL 200FX-TEM on all catalystsafter they were reduced in H2 at 500◦C and passivatedin He/air. It is known that exposure to the TEM beamcauses destruction of the zeolite structure and metalagglomeration [15]. Therefore, to avoid beam damagewe used low beam intensities and minimized the ex-posure times.

EXAFS data on in situ reduced samples wereobtained at the National Synchrotron Light Source(NSLS) at Brookhaven National Laboratory, Up-ton, NY, using beam line X-18b equipped with a Si

(111) crystal monochromator. The X-ray ring at theNSLS has an energy of 2.5 GeV and a ring currentof 80–220 mA. EXAFS data were taken near theLIII -edge of Pt (11 564 eV). The experiments wereconducted in a stainless steel sample cell, whichallowed in situ treatments at temperatures rangingfrom 500◦C to liquid nitrogen temperature. Beforeeach measurement, the catalysts, previously reducedat 500◦C, were re-reduced in situ at 300◦C (heat-ing ramp of 10◦C /min) for 30 min in flowing H2.The EXAFS spectra were recorded at liquid nitrogentemperatures under H2 flow. Six scans were recordedfor each sample. The average spectrum was obtainedby adding the six scans. The pre-edge backgroundwas subtracted by using power series curves. Subse-quently, the post-edge background was removed usinga cubic spline routine. The spectra were normalizedby dividing by the height of the absorption edge. Toavoid overemphasizing the low energy region [24],thex data werek3-weighted. The range ink-space uti-lized to do the analysis was 3.5–13.5 A−1. Theoreticalreferences for Pt–Pt, and Pt–O bonds were obtainedby using the FEFF program from the Universityof Washington [25–27]. The FEFFIT fitting routinewas employed to obtain the structural parameters ofthe Pt clusters after the various thermal treatmentson the different supports. The Debye–Waller factorsfor each bond-type (σ ), the edge energy difference(1Eo), the coordination numberN, and the differencein bond distances (1R) with respect to the theoreticalreference, were used as fitting parameters.

Infrared spectroscopy of adsorbed CO were ob-tained on a Bio-Rad FTS-40 spectrometer, equippedwith a MCT detector. The experiment was conductedin a diffuse reflectance cell from Harrick Scientific,type HVC-DR2 with ZnSe windows that allowed us toperform in situ reduction and oxidation pre-treatments.For each IR spectrum, taken at a resolution of 8 cm−1,128 scans were added. Prior to each spectrum, the cat-alyst was reduced in a flow of H2 and purged in Hefor 30 min. Then, the catalyst was exposed to a flowof 3% CO in He for 30 min at room temperature andpurged in He for 30 min.

2.3. Catalytic activity

The methylcyclopentane ring opening (MCP-RO)was employed as a test reaction in a pulse microre-


actor. In each run, 0.025 g of sample was placed ina Pyrex reactor, reduced in situ at 500◦C, and cooledto the reaction temperature, which varied from 260 to350◦C. At each reaction temperature, three consecu-tive 100ml pulses, containing 6.46× 10−7 moles ofMCP were sent over the catalyst. The products wereanalyzed in a Hewlett Packard 5890 chromatograph,equipped with a FID detector and a packed column30% DC200–Chromosorb 60/80 P AW from Alltech.

In all runs, the main products detected were 2MP,3MP, and hexane, with only small amounts of C1–C5products and benzene. On the same system, then-hexane aromatization reaction was also studied inthe pulse-mode opering at 450 and 500◦C, using inthis case 100ml pulses containing around 6.57× 10−7

moles ofn-hexane. As in the previous case, the carriergas was hydrogen.

At the same time, steady-state reaction tests wereconducted on continuous flow reactors. One reactorwas only used for cleann-hexane runs, while the otherreactor was only used for sulfur deactivation studies.To avoid contamination of the clean runs, all lines lead-ing to each reactor were kept segregated. Each reactorconsisted of a 0.5 inch stainless steel tube with an in-ternal thermocouple. The experiments were conductedusing 0.40 g of catalyst in each run. Each catalyst wassieved to give pellet sizes between 300 and 425mm(40/50 mesh granules). The catalyst bed was supportedon a bed of quartz glass wool. The reactor was operatedunder flowing hydrogen.n-Hexane (Aldrich, 99+%purity) was added by infusion with a syringe pump(Sage model M365) through a tee-junction prior to thereactor. In all experiments, the hydrogen/n-hexane ra-tio was kept at 6.0. Prior to reaction, the catalyst wasslowly ramped in flowing hydrogen at 100 cm3/(min gcat.) for 2 h to a temperature of 500◦C and reduced insitu at 500◦C in flowing hydrogen at 100 cm3/(min gcat.) for 1 h. All reactions were conducted at 500◦C ata WHSV of 10, so that the deactivation could be exam-ined after relatively short periods of time on stream.

A purge-valve was used to send samples to a gaschromatograph/mass spectrometer (Hewlett PackardG1800A GCD System) for analysis. The gas chro-matograph utilizes helium as the carrier gas andsends purged products of reaction through the column(HP-PLOT/Al2O3‘S’ deactivated) to achieve productseparation. Finally, products were ionized by the massspectrometer, which incorporates an electron ioniza-

tion detector (EID). A temperature ramp programprovided the means for adequate peak separation inthe GC column. To determine the signal/abundanceratio and quantify the concentration of each com-ponent in the products, normalization curves wereobtained using pure compounds.

3. Results

3.1. Catalyst characterization

3.1.1. EXAFSX-ray absorption spectroscopy is one of the most

powerful techniques available to characterize a seriesof samples like the ones we are investigating, whichonly exhibit very subtle differences. The XANESspectra for the two catalyst series at the Pt LIII -edge,together with that of a Pt foil, are shown in Fig. 1.No significant shifts were found in the absolute edgepositions, so that all edges have been zeroed to theinflection point. It is clear that the structures of PtLIII -edges were very similar for all the Pt/KL cata-lysts, regardless of Pt loading or preparation method.The spectra have been displaced in the figure for thesake of clarity, because when plotted together theyalmost fall on top of each other. Moreover, only littledifferences were detected in XANES between thecatalysts and the Pt foil. A slight broadening of thePt white line, typical of highly dispersed supportedmetallic particles in the presence of H2 [28], wasobserved for all the Pt/KL catalysts in comparisonto the foil. Thus, the analysis of the XANES spectrasuggests that the electronic state of the Pt clusters isnot significantly different from one series to the other.

In order to obtain further insight into the structureof the Pt clusters, a detailed EXAFS analysis of bothcatalyst series was conducted. The magnitudes of theFourier transforms (FT) of thek3-weighted EXAFSspectra, filtered from 3.5 to 13.5 Å−1 are shown in Fig.2. From the inspection of this figure, it is clear that themorphology of the metallic particles may indeed varyfor the two series. While the FT of the catalysts with2 wt % Pt (IWI-2 and VPI-2) look very similar andare typical of metallic Pt, they significantly change asthe Pt loading decreases. In fact, the FT of IWI-2 andVPI-2 exhibit a main peak at approximately 2.60 Å


Fig. 1. XANES spectra of Pt LIII -edge. (a) Pt foil and IWI series, (b) VPI series obtained at liquid nitrogen temperature.

Fig. 2. Fourier Transforms corresponding to thek3-weighted Pt LIII -edge EXAFS spectra obtained at liquid nitrogen temperature on insitu reduced Pt/KL catalysts. (a) IWI series, (b) VPI series.


and a satellite peak at around 2.05 Å, which is nor-mally observed for bulk Pt when the FT has not beencorrected for the amplitude and phase shift of Pt [29].As metal loading decreased in the series, the relativeintensity of the satellite peak increased significantly.As shown in the figure, for the VPI-05 catalyst, theintensity of this satellite peak was even higher thanthat of the main peak. An increase in the relative in-tensity of the satellite peak seems to reflect that theenvironment of Pt is not only constituted by Pt atoms.This trend would indicate that the structure of the Ptclusters is strongly modified as Pt loading decreases.This modification is particularly strong for the lowestloadings and for the series prepared by vapor phaseimpregnation.

The structural parameters from the EXAFS datawere determined by a fitting procedure. An inverseFourier transform was applied over a restricted rangeof r, 1.3–3.7 Å, isolating the EXAFS contribution ofthe first coordination sphere of platinum. Accordingto the Nyquist Theorem [30] and the EXAFS specificmodifications described by Stern [31], the maximumnumber of parameters that can be used to describe theEXAFS function is given by the following equation:

nmax = 21k1r/π + 2

Therefore, the use of the seven parameters to fitour data is statistically justified, since for thek andrranges employed in this work, the Nyquist criterionwould allow up to 17 parameters. The filtered datawere fitted using the FEFFIT program [32]. In a firstapproach, the data were fitted with one Pt–Pt shell, us-ing theoretical standards generated by the FEFF soft-ware [33–36]. Although the fits obtained with onlyone coordination could be considered satisfactory forthe catalysts with 2% of Pt (IWI-2 and VPI-2), poorfits were obtained for the other samples with lowermetal loading. Therefore, new fittings including a sec-ond Pt–O shell were performed. Theoretical ampli-tudes and phase shifts corresponding to the Pt–O dis-tances were derived with FEFF, assuming Pt–O bondsat 2.56 Å. These ‘long Pt–O distances’ have been pre-viously reported by Koningsberger et al. [37,38] for Ptparticles inside the channels of the KL zeolite. The ad-dition of Pt–O distances in the first coordination sphereof Pt greatly enhanced the quality of all fits. However,for the samples with the lowest metal loadings the fitswere less satisfactory, indicating that the structure of

Table 1Structural parameters determined from EXAFS analysis

Catalyst 1E0(eV) Pt–Pt distance Pt–O distance

N R (Å) σ 2 (Å2) N R (Å) σ 2 (Å2)

IWI-2 5.2 4.5 2.73 0.0058 0.3 2.60 0.0041IWI-1 3.8 3.9 2.73 0.0058 0.4 2.57 0.0060IWI-05 −0.9 3.9 2.71 0.0049 0.6 2.56 0.0060VPI-2 6.0 4.0 2.73 0.0058 0.4 2.59 0.0033VPI-1 3.6 3.0 2.72 0.0058 0.6 2.58 0.0043VPI-05 1.7 2.6 2.72 0.0048 0.8 2.56 0.0055

the smallest Pt particles may be even more complex.The structural parameters resulting from the fits withtwo shells, Pt–Pt and Pt–O distances are summarizedin Table 1.

A slight contraction of the Pt–Pt distances com-pared to that of bulk Pt is observed for all samples.Such a distance contraction has been already reportedfor highly dispersed systems [15]. As previously pro-posed, the ‘long Pt–O distances’ give evidence for theinteraction between zero-valent Pt atoms and the sup-port, probably having the presence of H2 in the inter-facial layer between the Pt particles and the support[39].

The evolution of the calculated Pt–Pt and Pt–O coor-dination numbers as a function of metal loading is rep-resented in Fig. 3 for the two preparation series. Cleartrends are observed, indicating an increase in Pt–Ptcoordination and a decrease in Pt–O coordination asmetal loading increases. Two different morphologieshave been suggested in the literature for metal clustersinside the channels of the zeolite with coordinationnumbers as low as those reported here [40,41]:1. Small spherical clusters consisting of 6–10 Pt

atoms, with a Pt–Pt coordination number between4–5 and a low metal-support coordination, i.e.low Pt–O coordination.

2. Pt spread on the support, forming small Pt ‘rafts’.In this morphology the Pt–Pt coordination signif-icantly decreases and the Pt–O coordination in-creases.

From the above EXAFS results, it seems that, forfixed metal loading, the VPI method leads to a higherpopulation of Pt clusters small enough so that all theiratoms are in close proximity to the zeolite walls. Thisis reflected by decreased Pt–Pt coordination and in-creased Pt–O coordination for the VPI series com-pared to the IWI series. Within each series, the frac-


Fig. 3. Evolution of the Pt–Pt and Pt–O coordination numbersobtained from the EXAFS analysis as a function of the metalloading for IWI and VPI series.

tion of atoms interacting with the walls increases withdecreasing metal loading.

In agreement with the EXAFS data that reflect veryhigh metal dispersion for all the catalysts in the series,hydrogen uptakes measured by volumetric chemisorp-tion resulted in high H/Pt values, ranging from 1.2to 1.8. In previous studies, it was shown that calcina-tion of Pt/KL catalysts at temperatures above 400◦Cresulted in particle growth and channel blocking thatlead to large decreases in chemisorption capacity [42].Those large chemisorption losses were obtained whenthe catalysts were calcined at high temperatures priorto the reduction treatment. Here, we have comparedthe effect of re−calcining the IWI-1 and VPI-1 sam-ples at 500◦C after the initial low-temperature calci-nation and reduction. It was observed that, while thecatalyst prepared by incipient wetness impregnationexhibited a large loss in chemisorption capacity afterre-calcination (about 30%), the one prepared by vaporphase impregnation had a very modest loss (15%). Wewill identify these re-calcined catalysts as RC-IWI-1and RC-VPI-1, respectively.

3.1.2. Transmission electron microscopyThe TEM observations agreed very well with the

EXAFS and hydrogen chemisorption data, which in-dicate a very high degree of metal dispersion. Onlyon the 2% Pt samples, Pt particles outside the zeo-lite were detected. Images of the IWI-05 and VPI-05catalysts are shown in Fig. 4. The zeolite channelsare very clearly seen, some of them containing verysmall particles. The sample prepared by IWI exhibitedsome particles that appear to partially fill the diame-ter of the channels. However, they were not presentin the VPI sample, in accordance with the EXAFSdata, which indicated a significantly higher ratio of(Pt–Pt)-to-(Pt–O) coordination for the IWI-05 catalystthan for the VPI-05. If the particles are very small andperhaps raft-like in the VPI-05, one may not expect tosee them by TEM.

3.1.3. Infrared spectroscopyFTIR of adsorbed CO has been widely employed to

characterize Pt/KL catalysts. Typically, the IR spec-trum obtained with Pt/KL catalysts is much more com-plex than that normally observed on Pt/SiO2. In mostFTIR studies of CO adsorption on Pt/KL a series ofbands between 2080 and 1950 cm−1 has been ob-served. The different bands have often been ascribed tovarious degrees of metal–zeolite interaction [43]. Ourresults were similar to those typically observed before.For example, the DRIFTS spectra of CO adsorbed atroom temperature over the IWI series are shown inFig. 5. As the loading of Pt increased in the series theintensity of the bands increased, but there were no sig-nificant differences in the position or relative intensi-ties. The VPI series (not shown) exhibit an equivalenttrend, without significant changes in band positions.Therefore, although the EXAFS data indicated a vary-ing degree of metal–zeolite interaction, and possibly,varying particle morphologies, the DRIFTS data didnot give evidence for such variations.

On the other hand, the evolution of the IR bands as afunction of time after introduction of CO presented in-teresting features that may explain why the final spec-tra all look alike, despite changes in Pt morphology.Such evolution of the bands as a function of exposuretime is illustrated in Fig. 6 for the VPI-1 catalyst. Itcan be seen that during the first few minutes a banddeveloped, centered at about 2068 cm−1. Only after


Fig. 4. TEM micrographs of Pt/KL catalysts calcined at 350◦C and reduced at 500◦C. (a) IWI-05, (b) VPI-05.

Fig. 5. DRIFTS spectra of CO adsorbed on Pt/KL IWI series. The reduced catalysts were exposed to a flow of 3% CO in He for 30 minat room temperature and purged in He for 30 min.


Fig. 6. Evolution of DRIFTS spectra of CO adsorbed on VPI-1 as a function of exposure time. The reduced catalyst was exposed to aflow of 3% CO in He at room temperature and purged in He for 30 min (dotted line).

10 min, new bands started emerging at about 2057 and2014 cm−1. As discussed below, it seems that duringthe exposure to CO, modifications of the sample oc-curred that were responsible for the appearance of thelower frequency bands. When the gas phase CO wasremoved by purging with He, the typical double bandscorresponding to gaseous CO quickly disappeared, andthe band at 1970 became more apparent.

Although DRIFTS did not show important differ-ences for the freshly reduced catalysts, it did exhibitsignificant changes after poisoning with sulfur. Figs.7a and b compared the IR spectra of adsorbed CO af-ter exposing the IWI-1 and VPI-1 catalysts to reactionconditions for 10 h with feeds containing 1 and 10 ppmsulfur. Interesting differences are observed. While thespectra for the IWI-1 catalyst showed a significant de-crease in the region of the 2014 and 1970 cm−1 bands,the spectra for the VPI-1 only showed a modest changein this region. Below we will discuss this lower sen-sitivity to sulfur when we discuss the higher stabilityof the VPI series under sulfur.

Another set of samples that displayed meaningfulvariations in the DRIFTS spectra were the RC-IWI-1and RC-VPI-1. As mentioned above, the re-calcinationtreatment caused a much less pronounced loss inhydrogen chemisorption on the catalyst prepared byVPI than on the one prepared by IWI. This differencecorrelates well with the IR spectra. As shown in Fig.8 for the RC-IWI-1, the high temperature treatment

caused the appearance of a clear shoulder at about2084 cm−1 that we attribute to Pt particles whichmigrated to the outer surface of the zeolite duringhigh-temperature calcination. This assignment is inagreement with the very low 3MP/2MP ratio observedfor these catalysts in the MCP-RO reaction. By con-trast, the RC-VPI-1 sample did not show this bandand is therefore more resistant to high temperaturetreatments. Similar resistance of catalysts preparedby VPI to oxidation/reduction thermal treatments hasbeen previously proposed [23].

The study of the time evolution of the absorptionbands on the RC-IWI-1 catalyst provides further in-sight into the nature of the species present during thisexperiment. As shown in Fig. 9, the features that im-mediately appeared were a band at 2084 cm−1 and an-other at 2068 cm−1. During the CO exposure, the for-mer did not shift while the latter shifted slightly. Asopposed to the dramatic modification exhibited by thestandard IWI-1 catalyst, the one re-calcined at 500◦C,RC-IWI-1, did not show the appearance of any newabsorption bands.

3.2. Catalytic activity measurements

3.2.1. Methylcyclopentane ring opening in a pulsereactor

The use of MCP ring opening to determine theexistence of collimating effects was first recognized


Fig. 7. DRIFTS spectra of CO adsorbed on Pt/KL. The reduced catalysts were exposed to a flow of 3% CO in He for 30 min at roomtemperature and purged in He for 30 min. (a) IWI-1 fresh and poisoned catalysts. (b) VPI-1 fresh and poisoned catalysts.

by Sachtler et al. [44–46] for the case of Pt on Yzeolite. The ring opening of MCP can occur on threedifferent C–C bond positions, two ina (yieldingn-hexane), two inb (yielding 2MP), and one ing(yielding 3MP). Therefore, the statistical distribu-tion is n-Hex : 2MP : 3MP = 40 : 40 : 20. Due to themicrogeometry of the system, it is expected that anincoming MCP molecule inside the pores of the ze-olite will be preferentially oriented with its longeraxis parallel to the direction of the pores. As a result,when a molecule encounters a metal particle insidethe pores, opening in theg position should be favoredin comparison to opening in the twob positions. If

that is the case, the resulting 3MP/2MP product ratioshould be higher than the statistical value of 0.5. Inprevious work on Pt/KL catalysts prepared by IWI wehave indeed found 3MP/2MP higher than on supportwithout microporosity. We also observed that thisratio decreased when the Pt/KL catalyst had a largefraction of Pt particles outside the channels of thezeolite [47,48].

In this contribution, we have used this test reac-tion again to compare the two catalyst series. As illus-trated in Fig. 10, the first difference exhibited by thetwo series is with respect to their hydrogenolysis ac-tivity. The IWI catalysts showed a significantly higher


Fig. 8. DRIFTS spectra of CO adsorbed on IWI-1 fresh and re-calcined at 500◦C (RC-IWI-1). The reduced catalysts were exposed to aflow of 3% CO in He for 30 min at room temperature and purged in He for 30 min.

Fig. 9. Evolution of DRIFTS spectra of CO adsorbed on RC-IWI-1 as a function of exposure time. The reduced catalyst was exposed toa flow of 3% CO in He at room temperature and purged in He for 30 min (dotted line).

activity than the VPI catalysts for this reaction. As aresult, different temperatures were needed in order towork at comparable conversions. We have chosen torepresent the catalytic activity in terms of the temper-ature needed to reach 10% MCP conversion, becausethis way of presenting the data does not require ex-trapolations in temperature. In any case, a similar ac-tivity trend could be obtained if the conversion datawere extrapolated to the same temperature. As illus-trated in the figure, it was observed that the activity of

the IWI-05 sample was comparable to that of the VPIseries. By contrast, the higher content IWI catalystsshowed activities comparable to that of the Pt/SiO2catalyst.

As mentioned above, our main motivation to studythe MCP ring opening on the two Pt/KL series wasto analyze the 3MP/2MP product ratio, and compareit to the 0.5 statistical value. It was observed that thisratio strongly depended on the catalyst, but varied lit-tle with temperature or conversion. Fig. 11 shows a


Fig. 10. Hydrogenolysis activity expressed in terms of the tem-perature needed to reach 10% MCP-RO conversion as a functionof the metal loading.

comparison of the 3MP/2MP ratios obtained on thetwo series at the same overall conversion (10%). Aclear trend is observed. The higher loading IWI-1 andIWI-2 catalysts exhibit significantly higher 3MP/2MPratios than the low loading IWI catalyst and the VPIseries, as well as the silica-supported catalyst. The in-terpretation that one can give to these results is thatonly the higher loading Pt/KL catalysts present a clearcollimation effect. This effect is absent on the Pt/SiO2catalyst, which is reasonable, but it is also absent onthe low-loading IWI catalyst and the VPI series. Thisresult is important and will be further discussed below.

The MCP-RO reaction was also conducted on theRC-IWI-1 catalyst. As mentioned above, this sample,which was re-calcined at high temperature and reducedafter the first calcination/reduction treatment, exhib-ited an absorption band at 2080 cm−1, not observed onthe standard IWI-1 sample. In this case, the MCP-ROactivity was much lower than on the standard cata-lysts and the 3MP/2MP ratio was only 0.3. In previ-ous work, we have indicated that this low 3MP/2MPratio is typical of Pt present in the form of large,

Fig. 11. MCP ring opening selectivity obtained at the same overallconversion (10%). 3MP/2MP ratio as a function of the metalloading.

well-ordered particles [48]. These results indicate thatthe high temperature re-calcination treatment causesthe agglomeration of particles outside the channels ofthe zeolite.

3.2.2. n-Hexane aromatization in a pulse reactorThe yields of benzene and hydrogenolysis products

obtained in the pulse reactor are depicted in Figs. 12aand b as a function of Pt loading. One must note herethat in the pulse mode, a small amount ofn-hexane ispassed over the catalyst under pure hydrogen. There-fore, particularly at lower temperatures, the resultsrepresent the behavior of a clean catalyst. It is inter-esting to note that the observed benzene yields wereonly a function of the metal loading and varied verylittle from one series to the other. Also, in agreementwith earlier reports, in the absence of de-activationby coke, Pt supported on silica exhibited a benzeneyield similar to that of the Pt/KL catalysts. When thepulse measurements were conducted at 500◦C and theoverall conversion increased, the silica-supported cat-alyst exhibited a lower benzene yield, which may be


Fig. 12.n-hexane aromatization (pulse mode) at 450◦C for Pt/KL and Pt/SiO2 catalysts. (a) Benzene yield as a function of Pt loading, (b)Hydrogenolysis yield as a function of Pt loading.

due to some degree of de-activation already presentunder these conditions. Although at 450◦C, the ben-zene yields obtained in the pulse reactor showed littledifferences among the series, the undesired side reac-tion hydrogenolysis did exhibit significant differencesamong VPI, IWI, and silica-supported catalysts. Thehydrogenolysis activity was highest for the 1% Pt/SiO2and lowest for the VPI series. Within each series, itincreased with Pt loading.

As described in previous work, in addition to the hy-drogenolysis (C1–C5) products, the main by-productsobserved during our flow reaction experiments werehexenes. Those experiments were conducted at 500◦Cand using a H2 : hydrocarbon ratio of 6 : 1. By con-trast, in the pulse experiments conducted at lower tem-peratures and in excess hydrogen, which was used asa carrier, the main by-products were MCP, 2MP, and3MP.

3.2.3. n-Hexane aromatization in a flow reactorThe two-catalyst series exhibited significant differ-

ences in their deactivation patterns, both in the absence

and in the presence of sulfur. Fig. 13a shows the vari-ation of benzene yield as a function of time on streamover the IWI-1, VPI-1, and Pt/SiO2 catalysts. It isclear that, in the flow experiments, the silica-supportedcatalyst de-activated so rapidly that, by the time thefirst experimental point was measured, it had alreadyde-activated. It is interesting to compare these resultswith those of the pulse experiments. In that case, thebenzene yield obtained on the Pt/SiO2 catalyst wasvery similar to that obtained on the Pt/KL catalysts. Inthe flow-mode, however, its activity was much lower.Perhaps, the most important result to draw from thisexperiment is the dramatic difference in de-activationpattern exhibited by the two 1% Pt/KL catalysts. Inparallel with the lower hydrogenolysis activity pre-sented in the pulse experiments, the catalyst preparedby VPI exhibited a much higher stability.

The differences exhibited in the benzene yieldsover the three catalysts correlated with similarly im-portant differences in the generation of methane andhexenes. The evolution of methane and hexenes withtime-on-stream are illustrated in Figs. 13b and c. Onthe Pt/SiO2 catalyst, the methane produced by hy-


Fig. 13. n-hexane aromatization (flow mode) at 500◦C for Pt/KL and Pt/SiO2 catalysts (WHSV: 10 h−1). (a) Benzene yield as a functionof time on stream, (b) Methane yield as a function of time on stream, (c) Hexene yield as a function of time on stream.


Fig. 14. n-hexane aromatization (flow mode) at 500◦C for IWIand VPI series (WHSV: 10 h−1). Benzene selectivity as a functionof the conversion. IWI series (open symbols) and VPI series(full symbols). Pt loading: 0.5% (circles), 1% (diamonds), 2.5%(squares).

drogenolysis was initially very high, which agreeswell with the pulse experiments. However, it rapidlydropped in the flow mode as the catalyst de-activated.Simultaneously, the production of hexenes on thiscatalyst was considerable. On the Pt/KL catalysts, themore selective VPI-1 exhibited a very low productionof either methane or hexenes.

In general, catalysts prepared by VPI resulted inhigher aromatization selectivities, lower formation ofhexenes, lower hydrogenolysis, and lower rates ofde-activation. An accepted way to compare the perfor-mance of Pt/KL catalysts is using benzene selectivityvs. conversion plots. Fig. 14 summarizes the perfor-mance of the catalysts in the two series. It is clearthat the VPI series exhibits better catalytic behaviorthan the IWI series and among the best ever reportedin the literature (For a comparison see ref. [15]).

Finally, as illustrated in Fig. 15a, the VPI catalystsdisplayed significantly higher aromatization activityin the presence of 1 ppm sulfur. When the sulfur con-tent was increased to 10 ppm, both catalysts rapidly

de-activated. It is interesting to note that although af-ter 9 h under 10 ppm sulfur the aromatization activitywas almost completely lost, the dehydrogenation ac-tivity was still relatively high. As shown in Fig. 15b,the yield of hexenes was higher for the IWI-1, butafter 9 h under 10 ppm S, the VPI-1 catalyst reachedabout the same yield of hexenes. This result correlateswell with the residual CO adsorption capacity of thetwo poisoned catalysts as shown in DRIFTS spectradepicted in Fig. 7a and b.

4. Discussion

As stated above, the aim of this work was to com-pare catalysts with most of the Pt particles in the inte-rior of the zeolite, but with different morphology, andto try to relate these differences to the catalytic prop-erties. Therefore, we will analyze first what we havelearned from our detailed characterization and then theimplications on the performance of the catalysts.

4.1. Catalyst morphology

The EXAFS analysis has demonstrated that the VPImethod results in a larger fraction of Pt clusters con-stituted by only a few atoms, and as a result theyare in very close interaction with the zeolite walls.The MCP-RO results can be combined with the EX-AFS and DRIFTS data to draw an interesting pic-ture. As mentioned above, only the higher loadingIWI catalysts presented high 3MP/2MP ratios. Thisratio was lower, as expected, for the Pt/SiO2catalyst,which has no microporosity that would lead to colli-mation of MCP. However, this ratio was also low forthe low-loading IWI and VPI catalysts. This low ra-tio is certainly not due to the massive presence of Ptparticles outside the channels of the zeolite. DRIFTSdata did not show for these catalysts larger absorptionbands in the characteristic region around 2070 cm−1.The fact that the ratio for the VPI catalysts is lowerthan that for the IWI-1 and IWI-2 is an indication that,for the IWI catalysts, the Pt particles may be largeenough to partially block the path of the incomingMCP molecule, forcing the C–C to occur at theg posi-tion. In contrast, on those catalysts with very small Ptparticles, having a very small Pt–Pt coordination andrelatively large Pt–O coordination, the collimation ef-


Fig. 15. n-hexane aromatization (flow mode) in the presence of 1 ppm (diamonds) and 10 ppm (squares) sulfur at 500◦C for IWI-1 (opensymbols) and VPI-1 (full symbols) catalysts (WHSV: 10 h−1). (a) Benzene yield as a function of time on stream, (b) Hexene yield as afunction of time on stream.

Fig. 16. Schematic representation of Pt particles inside the poresof the KL zeolite. In case (a) the Pt particle partially blocks thepath of the MCP molecule; in case (b) the flat Pt particle doesnot block the path of the MCP molecule.

fect does not result in preferential cleavage in theg po-sition since the path of the MCP molecule will not beimpeded. These two different morphologies and theireffect on MCP-RO are illustrated in the schematic rep-resentation shown in Fig. 16. The case of the VPI-2catalyst deserves some further discussion. This cata-lyst exhibits almost the same Pt–Pt and Pt–O coordi-nations as the IWI-1, but yet the VPI-2 catalyst showsno collimation while the IWI-1 does. To resolve thisapparent discrepancy one needs to bear in mind that

EXAFS only provides average coordination numbers.So, if the VPI-2 has part of the Pt forming small clus-ters constituted by only a few atoms and part as largerparticles outside the channels (as seen by TEM), theaverage coordination number may coincide with thatof the IWI-1, in which the clusters inside the zeoliteare larger. However, no collimation should occur onthe VPI-2 catalyst, while it will occur on the IWI-1.On the former, neither the small clusters inside or theparticles outside the zeolite promote C–C cleavage attheg position, on the latter, the larger particles insidethe zeolite do promote this cleavage and the preferen-tial formation of 3MP.

The analysis of the IR data may also help in elu-cidating the state of Pt particles in the two-catalystseries. Here, however, one needs to be careful sincethe interpretation of the FTIR data is not straightfor-ward. Besoukhanova et al. [10] were the first to iden-tify the appearance of several distinct bands in the re-gion 2070–1950 cm−1. Many authors [49] have sub-sequently observed this series of bands, but the originof these bands is still unclear.


The analysis of the IR bands could be dividedinto two regions, above and below 2050 cm−1. Thoseabove 2050 cm−1 are thought to be due to CO ad-sorbed on Pt particles located outside the zeolite andin the near-surface region [50]. According to the po-sitions of the higher frequency region observed onthe two series of catalysts investigated in this work(see Fig. 5), we can conclude that only a small frac-tion of particles reside outside. The main band at2060–2068 cm−1 may be associated with particlesnear the surface of the zeolite rather than outsidethe channels. In fact, only when the IWI catalystwas calcined at high temperature a shoulder arose at2078 cm−1.(see Fig. 8). That position is clearly asso-ciated with large Pt particles outside the channels ofthe zeolite and coincides with the position observedon Pt/SiO2.

The region below 2050 cm−1 is more complicatedto explain. It has been proposed that the bands appear-ing in the 1970–1920 cm−1 range correspond to COadsorbed on Pt particles inside the channels, with anelectronic structure strongly perturbed by the zeolite[50]. Lane et al. [51] have paid particular attention tothe band appearing around 1970 cm−1. They pointedout that the exact position of this band depended onthe basicity of the cation exchanged into the zeoliteand showed that a band at about the same positioncould be generated on a Pt/SiO2 catalyst when it isimpregnated with K+. Therefore, they interpreted thisband as a result of an electrostatic attraction betweenthe linearly bonded CO with the alkali cations. Theyfound no correlation between the appearance of thisband and the aromatization activity.

The time evolution of the bands that we have ob-served (see Fig. 5) is difficult to rationalize in termsof simple adsorption of CO. That is, unless there isa strong mass transfer limitation it is unreasonablethat the bands at 2060–2068 develop very quickly, butthose at lower frequencies take more than 10 min. Anidea recently advanced by Stakheev et al. [52] thatmay well explain our results, is that CO itself modifiesthe structure of the Pt clusters inside the KL zeolite.Consequently, CO adsorption does not probe the metalparticles in their original structure, but rather generatesnew molecular arrangements that are stabilized insidethe zeolite. These authors have proposed that duringCO adsorption, Pt carbonyls can be formed. They ob-served that the intensity of the bands greatly increased

when CO pressure was increased from 1 to 500 mbar,although in normal adsorption of CO, it is expectedto reach a saturation coverage at relatively low pres-sures. Based on this assumption, they matched the ob-served IR frequencies with known Pt carbonyls andindicated that although Pt carbonyls containing onlyCO ligands are very unstable, substitution with otherligands can greatly increase their stability. Therefore,they proposed that the zeolite could act as a ligandand affect the stabilization of these species. In sup-port of this idea, EXAFS data of Mojet and Konings-berger [53] have shown that the 5–6 atom metallicparticles present in the KL zeolite before the admis-sion of CO, completely disrupted and Pt carbonylsformed in the presence of CO. At the same time, theXANES spectrum exhibited a profound change indi-cating significant changes in the electronic structureof Pt. Menacherry and Haller [54] have also observeddrastic changes in the FTIR spectrum upon heatingunder low pressures of CO. They reported that, whenthe equilibration temperature under 1000 ppm CO wasincreased from room temperature to 200◦C, a band at1940 cm−1 greatly increased. It is possible that in thatcase, Pt carbonyls did not form under low CO pres-sures but they did form when the temperature was in-creased.

In light of these ideas, an interesting comparisoncan be made from Figs. 6 and 9, which depict the timeevolution of the DRIFTS spectra under CO exposure.It is seen that while the catalyst that was pre-calcinedat low temperature exhibited marked band shifts, theone pre-calcined at high temperature simply showeda monotonic increase in band intensity as a functionof time. This difference would indicate that on thecatalyst calcined at 500◦C, (RC-IWI-1) the formationof Pt carbonyls occurs to a much lesser degree than onthe other catalyst. Stakheev et al. [52] have pointed outthat the formation of Pt carbonyls can only occur whenthe Pt clusters are very small. They indicated that themetal–metal bond strength is much lower than that ofthe bulk metal only for particles with less than 15–20atoms. Therefore, one may not expect the Pt particlesoutside the zeolite in the RC-IWI-1 sample (i.e. bandat 2080 cm−1) to be converted to Pt carbonyls.

Similarly, the different changes observed in theDRIFTS spectra of the VPI and IWI samples af-ter exposure to sulfur may be rationalized in termsof formation of carbonyls. Upon exposure to sulfur


(see Fig. 7) the lower frequency bands (i.e. below2000 cm−1) did not decrease as markedly on the VPIcatalyst as on the IWI catalyst. One may conclude thatPt carbonyls can still be formed on the poisoned VPIcatalyst, but not on the poisoned IWI catalyst. Thisdifference may help to understand the mechanism ofsulfur poisoning. If sulfur poisoning is due to an ag-glomeration of Pt particles inside the channels of thezeolite, as suggested by McVicker et al. [55], it is pos-sible that this agglomeration results in a particle sizethat exceeds the maximum (i.e. 15–20 atoms) beyondwhich carbonyl formation is no longer favorable. Ifthat is the case, the VPI catalysts, which have a highconcentration of small particles in close contact withthe zeolite, could be more resistant to particle growth.Consequently, at least during the 9 h of the run undersulfur, they would not reach the size that preventsthe formation of carbonyls during exposure to CO.An alternative is that sulfur poisoning occurs on thepotassium cations, as suggested by Ponec [56]. If thatwere the case, then the lack of carbonyl formationwith IR bands below 2000 cm−1 on the poisoned IWIcatalysts could be attributed to a possible stabiliza-tion requirement of those carbonyls by the potassiumcations. However, this would not explain why the VPIcatalyst exposed to the same sulfur treatment can stillform those carbonyls. Therefore, it appears that sulfurhas a stronger influence on the Pt clusters on the IWIsample than on those of the VPI sample.

4.2. Catalytic properties

The pulse experiments allow us to compare thearomatization activity of different catalysts in the ab-sence of significant de-activation. Based on the ben-zene yields shown in Fig. 12a, and in agreement withIglesia et al. [12,13], one may conclude that Pt onsilica is intrinsically as active as Pt/KL. However, itis important to realize thatn-hexane aromatizationis an ensemble-sensitive reaction [57] and based onthat requirement alone we would not expect high cat-alytic activity on particles that are composed of a fewatoms. In fact, hydrogenolysis, which is known to beensemble-sensitive occurs at a much lower rate on thePt/KL catalysts than on the Pt/SiO2 catalyst. There-fore, the idea that the environment provided by the KLzeolite increases the C6-ring closure is not discred-

ited by our pulse experiments. On the other hand, thepulse experiments clearly show that the hydrogenoly-sis activity correlates with the propensity of the cata-lyst to de-activate in the flow mode. A catalyst suchas Pt/SiO2 or a poorly prepared Pt/KL catalyst with alarge fraction of Pt outside the channels of the zeolitewill exhibit a high hydrogenolysis activity and a rapidde-activation in the flow reaction.

In agreement with the conclusions drawn from theIR data, the reaction data have shown that the VPIcatalysts are not only more active and selective thanthe IWI catalysts. They are also more resistant tode-activation by calcination at high temperatures, bycoke, and by sulfur. It has been observed that in theabsence of sulfur, the VPI-1 catalyst shows a very lowdeactivation rate. In fact, the same catalyst has beenrun for 70 h and kept a benzene yield of 40% at aWHSV of 5 [14]. If the VPI preparation results in ex-tremely small Pt particles in close contact with the ze-olite walls, they can tolerate more coke deposits andgreater particle growth before plugging the pores andsubsequent de-activation. By contrast, if the catalystprepared by IWI result in Pt clusters that partially fillthe channel (see Fig. 16), then small amounts of cokeor minor particle growth could cause a much greaterdeactivation, as experimentally verified.

5. Conclusions:

The distribution of Pt clusters and their resultingcharacteristic morphology largely varied with loadingand the method of preparation. Over a wide range ofcatalysts studied, we observed three different types ofparticles that profoundly affected activity, selectivity,stability, and results of catalyst characterization. Thedifferent particles, as described elsewhere in the liter-ature, included large particles formed outside of thechannels, small particles located inside the channels,but large enough to block or partially block them, andclusters that were small enough to reside in the lobesof the L-zeolite channels.

TEM and EXAFS revealed the existence of Ptclusters having a very small size located inside theL-zeolite channels for both IWI and VPI preparationmethods. In addition, these results were confirmed byFTIR of adsorbed CO, which gave evidence that theparticles were small enough to allow the formation of


Pt carbonyls. Furthermore, the increase in coordina-tion of Pt with the oxygen of the L-zeolite frameworkand the decrease in Pt–Pt coordination observed forthe VPI series relative to the IWI series clearly in-dicated that the particles produced were smaller andof different characteristic morphology. In line withthis finding, lower collimation effects were observedfor VPI catalysts with respect to IWI catalysts dur-ing pulse testing of MCP-RO. This lower collimationstrongly suggests that the clusters produced by VPIwere small enough to allow for the diffusion of re-actants and products around the cluster during thearomatization ofn-hexane reaction.

The characteristic size and morphology of the par-ticles produced from the different preparation proce-dures had important consequences on conversion, se-lectivity, and stability of catalysts under reaction con-ditions. Under continuous flow conditions, VPI cata-lysts displayed higher aromatization activity and se-lectivity, lower hydrogenolysis activity, lower hexenesand a significantly lower deactivation rate.

This work has demonstrated that the characteristicmorphology produced by the VPI method substantiallyimproves the performance of the catalyst under cleanand sulfur-poisoned conditions. The VPI method leadsto the formation of very small particles in close inter-action with the zeolite walls. Consequently, the cata-lysts show less coke formation, higher activity undersulfur-poisoned conditions, and greater resistance toparticle agglomeration during calcination at high tem-perature.

Acknowledgements

This work was supported by the Oklahoma Cen-ter for the Advancement of Science and Technology(OCAST). We are grateful to Phillips Petroleum fora scholarship for one of us (FG). We acknowledgethe National Science Foundation for a GRT trainee-ship for one of us (GJ) and the International Divisionfor partial support. We also acknowledge the technicalsupport of Greg Strout from the OUSR Noble Elec-tron Microscopy Laboratory for TEM measurementsand the personnel at NSLS, Brookhaven National Lab,for the EXAFS experiments. One of us (AB) is es-pecially indebted to the Fundación Antorchas and theFulbright Commission for financial support.

References

[1] J.R. Bernard, in: L.V.C. Rees (Ed.), Proc. 5th Internat. ZeoliteConfer., Heyden, London, 1980, p.686.

[2] G. Lane, F.S. Modica, J.T. Miller, J. Catal. 129 (1991) 145.[3] T.R. Hughes, W.C. Buss, P.W. Tamm, R.L. Jacobson, Stud.

Surf. Sci. Catal. 28 (1986) 725.[4] P.W. Tamm, D.H. Mohr, C.R. Wilson, Stud. Surf. Sci.Catal.

38 (1988) 335.[5] D. Rotman, Chemical Week 8 (1992)[6] R.J. Davis, Heterog. Chem. Rev. 1 (1994) 41.[7] P. Meriaudeau, C. Naccache, Chem. Rev. Sci. Eng. 39 (1997)

5.[8] S.J. Tauster, J.J. Steger, J. Catal. 125 (1990) 387.[9] E.G. Derouane, D.J. Vanderveken, Appl. Catal. 45 (1988)

L15.[10] C. Besoukhanova, J. Guidot, D. Barthomeuf, J. Chem. Soc.

Faraday Trans. I. 77 (1981) 1595.[11] G. Larsen, G.L. Haller, Catal. Lett. 3 (1989) 103.[12] E. Iglesia, J.E. Baumgartner, in: L. Guczi et al., (Eds.), Proc.

X Internat. Congr. Catal, Elsevier, 1993, p.993.[13] E. Iglesia, J. E. Baumgartner, Proc. IX Internat. Zeolite Conf.,

Montreal, 1992.[14] G. Jacobs, C.L. Padro, D.E. Resasco, J. Catal. 179 (1998) 43.[15] R.E. Jentoft, M. Tsapatsis, M.E. Davis, B.C. Gates, J. Catal.

179 (1998) 565.[16] D.W. Breck, Zeolite Molecular Sieves Structure, Chemistry

and Use; Wiley; New York, 1974, p. 636.[17] M. Vaarkamp, J.V. Grondelle, J.T. Miller, D.J. Sajkowski,

F.S. Modica, G.S. Lane, D.C. Koningsberger, Catal. Lett. 6(1990) 369.

[18] M. Vaarkamp, B.L. Mojet, M.J. Kappers, J.T. Miller, D.C.Koningsberger, J. Phys. Chem. 99 (1995) 16067.

[19] P.V. Menacherry, M. Fernandez-Garcia, G.L. Haller, J. Catal.166 (1997) 75.

[20] E. Mielczarski, S.B. Hong, R.J. Davis, M.E. Davis, J. Catal.134 (1992) 359.

[21] D.J. Ostgard, L. Kustov, K.R. Poeppelmeier, W.M.H. Sachtler,J. Catal. 133 (1992) 342.

[22] S.B. Hong, E. Mielczarski, M.E. Davis, J. Catal. 134 (1992)349.

[23] M. Bellatreccia, R. Zanoni, C. Dossi, R. Psaro, S. Recchia,G. Vlaic, J. Chem. Soc. Farad. Trans. 91 (1995) 2045.

[24] D.E. Sayers, B.A. Bunker, in: D. Koningsberger, R.Prins, (Eds.), X-ray Absorption: Principles, Applications,Techniques of EXAFS, SEXAFS, and XANES, Wiley, 1988,p.211.

[25] J.J. Rehr, S.I. Zabinsky, R.C. Albers, Phys. Rev. Lett. 69(1992) 3397.

[26] J.J. Rehr, J. Mustre de Leon, S.I. Zabinsky, R.C. Albers, J.Amer. Chem. Soc. 113 (1991) 5135.

[27] J. Mustre de Leon, J.J. Rehr, S.I. Zabinsky, R.C. Albers,Phys. Rev. B 44 (1991) 4146.

[28] N. Ichikuni, Y. Iwasawa, Catal. Lett. 20 (1993) 87.[29] E.C. Marques, D.R. Sandstroem, F.W. Lytle, R.B.J. Greegor,

J. Phys. Chem. 77 (1982) 1027.[30] E.O. Brigham, The Fast Fourier Transform, Prentice-Hall,

Englewood Cliffs, NJ, 1974.


[31] E.A. Stern, Phys. Rev. B 48 (1993) 9825.[32] M. Newville, B. Ravel, D. Haskel, E.A.M. Sterm, Y. Yacoby,

Physica B vol. 208/209 (1995), 154–156[33] J.J. Rehr, R.C. Albers, S.I. Zabinsky, Phys. Rev. Lett. 69

(1992) 3397.[34] J.J. Rehr, R.C. Albers, S. I. Zabinsky, Phys. Rev. Lett., (1991)

4146[35] J.J. Rehr, J. Mustre de Leon, S.I. Zabinsky, R.C. Albers, J.

Amer. Chem. Soc. 113 (1991) 5135.[36] J. Mustre de Leon, J.J. Rehr, S.I. Zabinsky, R.C. Albers,

Phys. Rev. B. 44 (1991) 4146.[37] D.C. Koningsberger, B.C. Gates, Catal. Lett. 14 (1992) 271.[38] M. Vaarkamp, J.T. Miller, F.S. Modica, G.S. Lane, D.C.

Koningsberger, J. Catal. 138 (1992) 675.[39] M. Vaarkamp, F.S. Modica, J.T. Miller, D.C. Koningsberger,

J. Catal. 144 (1993) 611.[40] S.J. Cho, W-S Ahn, S.B. Hong, R. Ryoo, J. Phys. Chem.

100 (1996) 4996.[41] P.V. Menacherry, M. Fernandez-Garcia, G.L. Haller, J. Catal.

166 (1997) 75.[42] G. Larsen, G.L. Haller, Catal. Today 15 (1992) 431.[43] W.-J. Han, A.B. Kooh, R.F. Hicks, Catal. Lett. 18 (1993) 193.

[44] G. Moretti, W.M.H. Sachtler, J. Catal. 116 (1989) 361.[45] X. Bai, W.M.H. Sachtler, J. Catal. 132 (1991) 165.[46] X. Bai, Z. Zhang, W.M.H. Sachtler, Appl. Catal. 72 (1991)

165.[47] W.E. Alvarez, D.E. Resasco, J. Catal. 164 (1996) 467.[48] W.E. Alvarez, D.E. Resasco, Catal. Lett. 8 (1991) 53.[49] L.M. Kustov, D.J. Ostgard, W.M.H. Sachtler, Catal. Lett. 9

(1991) 121.[50] A.Y. Stakheev, E.S. Shpiro, N.I. Jaeger, G. Schulz-Ekloff,

Catal. Lett. 32 (1995) 147.[51] G.S. Lane, J.T. Miller, F.S. Modica, M.K. Barr, J. Catal. 141

(1993) 465.[52] A.Y. Stakheev, E.S. Shpiro, N.I. Jaeger, G. Schulz-Ekloff,

Catal. Lett. 34 (1995) 293.[53] B.L. Mojet, D.C. Koningsberger, Catal. Lett. 39 (1996) 191.[54] P.V. Menacherry, G.L. Haller, Catal. Lett. 44 (1997) 135.[55] G.B. McVicker, J.L. Kao, J.J. Ziemak, W.E. Gates, J.L.

Robbins, M.M.J. Treacy, S.B. Rice, T.H. Vanderspurt, V.R.Cross, A.K. Ghosh, J. Catal. 139 (1993) 48.

[56] T. Fukunaga, V. Ponec, J. Catal. 157 (1995) 550.[57] S.M. Davis, F. Zaera, G.A. Somorjai, J. Catal. 85 (1984) 206.

Documents

Characterization of the morphology of Pt clusters ... · Characterization of the morphology of Pt clusters incorporated in a KL zeolite by vapor phase and incipient wetness impregnation