palladium supported on structured and nonstructured carbon: a consideration of pd particle size and...

Journal of Colloid and Interface Science 322 (2008) 196–208www.elsevier.com/locate/jcis

Palladium supported on structured and nonstructured carbon:A consideration of Pd particle size and the nature of reactive hydrogen

Claudia Amorim a, Mark A. Keane b,∗

a Department of Chemical and Materials Engineering, University of Kentucky, Lexington, USAb Chemical Engineering, School of Engineering and Physical Sciences, Heriot-Watt University, Edinburgh EH14 4AS, Scotland

Received 3 September 2007; accepted 12 February 2008

Available online 19 February 2008

Abstract

This study sets out a comprehensive characterization of bulk Pd and Pd (ca. 8% w/w) supported on activated carbon (AC), graphite and graphiticnanofibers (GNF). Catalyst activation has been examined by temperature programmed reduction (TPR) analysis and the activated catalysts an-alyzed in terms of BET area, TEM, H2 chemisorption/TPD, and XRD measurements. While H2 chemisorption and TEM delivered the samesequence of increasing (surface area weighted) average Pd particle sizes, a significant difference (by up to a factor of 3) in the values obtainedfrom both techniques has been recorded and is attributed to an unwarranted (but widely adopted) assumption of an exclusive H2/Pd adsorptionstoichiometry = 1/2. It is demonstrated that TEM analysis provides a valid mean particle size once it is established that the associated standarddeviation is small and insensitive to additional particle counting. XRD line broadening yielded an essentially equivalent Pd size (20–25 nm)for each supported catalyst. The nature of the hydrogen associated with the supported catalysts has been probed and is shown to comprise ofchemisorbed (on Pd), spillover (on the carbon support), and hydride (associated with Pd) species. Physical mixtures of bulk Pd + support (AC,graphite, and GNF) were also considered in order to assess hydrogen spillover by H2 TPD analysis. Generation of spillover hydrogen at roomtemperature is established where temperatures in excess of 740 K are required for effective desorption from the supported Pd catalysts, i.e., 280 Khigher than that required for the desorption of chemisorbed hydrogen. Pd hydride formation (at room temperature) is shown to be reversible withdecomposition occurring at ca. 380 K. Taking the hydrodechlorination of chlorobenzene as a test reaction, the capability of Pd hydride to promotea hydrogen scission of C–Cl in the absence of an external supply of H2 is demonstrated with a consequent consumption of the hydride. Thiscatalytic response was entirely recoverable once the Pd hydride was replenished during a subsequent reactivation step.© 2008 Elsevier Inc. All rights reserved.

Keywords: Supported Pd; Bulk Pd; Palladium hydride; Spillover hydrogen; Activated carbon; Graphite; Carbon nanofibers; Hydrodechlorination

1. Introduction

Supported Pd catalysts are widely applied in hydrogen me-diated catalysis, e.g., hydrogenation [1–3] and hydrogenolysis[4,5]. Pd catalysts have been established as effective in promot-ing hydrodehalogenation [4,6–14] (hydrogenolytic C–X bondscission), which is used as a model reaction system in thisstudy. The hydrodechlorination (HDC) of chlorobenzene oversupported Pd has been shown to be influenced by such parame-ters as catalyst synthesis [7], nature of the support [4,7,15,16],Pd loading [7], Pd particle size/dispersion [4,7,8,12,16], acti-

* Corresponding author.E-mail address: [email protected] (M.A. Keane).

vation procedure [6,8,10,11,13], solvent [17,18], base addition[19–21], and the presence of a second metal [6,9–11,13,14,16].However, the surface requirements for optimum HDC perfor-mance are far from established and there is still a need for fun-damental catalyst characterization/catalysis data to correlate ac-tivity with catalyst structure. HDC has been considered a struc-ture sensitive reaction where dechlorination rates are enhancedat lower metal dispersions [7,22,23]. Hydrogen uptake/releaseis undoubtedly an important factor and is linked to metal parti-cle size (distribution) and support interactions, features that weexamine in this paper. Besides the surface interactions of hy-drogen with supported Pd (chemisorption and spillover to thesupport), Pd also has the ability to absorb hydrogen to form Pdhydride [24–26].

0021-9797/$ – see front matter © 2008 Elsevier Inc. All rights reserved.doi:10.1016/j.jcis.2008.02.021

http://www.elsevier.com/locate/jcis

mailto:[email protected]

http://dx.doi.org/10.1016/j.jcis.2008.02.021

C. Amorim, M.A. Keane / Journal of Colloid and Interface Science 322 (2008) 196–208 197

Metal dispersion is commonly obtained from gas chemisorp-tion analysis where a unique uptake stoichiometry is assumed;H2 [6–11,24,27–34], CO [4,24–29,35–38], and O2 [1,23,24,28]are the most widely employed adsorbates. Carbon monoxidecan adsorb on metals in linear and bridged forms [24] and an as-sessment of the literature has revealed an inconsistency in termsof the Pd/CO adsorption stoichiometry that has been employedwith values of 1 [4,25–29,36], 1.15 [37,38], and 2 [35]; H2 andO2 are typically assumed to adsorb dissociatively, i.e., Pd/H= 1 [1,9,24,27–33,39] and Pd/O = 1 [24,27,28]. Transmissionelectron microscopy (TEM), as an imaging technique, has alsobeen employed to determine metal particle size [1,26,28,32].However, such limitations as diameter measurement of irregu-larly shaped particles, the two dimensional imaging of a three-dimensional structure and poor contrast between metal and sup-port are sources of inaccuracy [40,41]. X-ray diffraction, alsoused to determine average particle diameter [28,29,35], has alower detection limit (typically 3 nm) which mitigates againstits use in the characterization of well dispersed metals [42].Although gas chemisorption is routinely used to characterizesupported Pd, complications due to adsorbate/support interac-tions and contributions due to spillover have been noted for Pdsupported on oxides [27] and carbonaceous [27–30,43] carriers.Taking an overview of the literature dealing with supported Pd,the majority of the studies have made use of only one techniqueto obtain Pd particle size/dispersion without recourse to a sec-ond measurement in order to check the accuracy/reproducibilityof the results; H2 or CO chemisorption analysis has beenthe predominant choice [7,8,31,33,34,37,38]. Where more thanone technique (different combinations of chemisorption and/orTEM and/or XRD) has been applied, both satisfactory [1,4,26,32,35,39] and unsatisfactory [12,24,27–30] data agreement hasresulted.

Our intended purpose in undertaking this work was to ex-amine and characterize the possible reactive hydrogen speciesassociated with supported Pd, i.e., Pd hydride, chemisorbed andspillover hydrogen. Since no definitive guideline has emergedwith regard to the technique that is most appropriate and accu-rate to obtain a meaningful particle size/metal surface area, wealso revisit the apparent discrepancy of Pd size/dispersion ob-tained from H2 chemisorption, TEM, and XRD and address themerits/limitations of each measurement in developing an effec-tive characterization strategy. In terms of catalytic phenomena,we have focused on the possible contribution of H2 from Pdhydride in HDC, providing for the first time evidence of Pdhydride participation in chloroarene conversion. Three carbona-ceous support materials (structured and amorphous) have beenused as Pd carriers with low, medium and high surface areas:graphite (Pd/graphite); high aspect ratio graphitic nanofibers(Pd/GNF); activated carbon (Pd/AC).

2. Experimental

2.1. Catalyst preparation

The activated carbon (G-60, 100 mesh, 905 m2 g−1) was ob-tained from NORIT (UK) and the graphite (synthetic 1–2 µm

powder, 14 m2 g−1) from Sigma–Aldrich. The GNF support(104 m2 g−1) was synthesized by the catalytic decompositionof C2H4 over bulk Ni, as described in detail elsewhere [44,45].The catalytically generated GNF was contacted (agitation at500 rpm) with 1 M HNO3 for 7 days to remove the Ni con-tent and avoid any possible contribution to HDC activity; thecommercial activated carbon and graphite samples were alsosubjected to the same demineralization. The carbon supportswere thoroughly washed with deionized water (until pH of thewash water approached 7) and oven-dried at 383 K for 12 h. ThePd loaded catalysts (7.8% w/w Pd/AC, 8.5% w/w Pd/graphite,and 8.7% w/w Pd/GNF) were prepared by standard impregna-tion where a 2-butanolic Pd(NO3)2 solution was added dropwise at 353 K to the substrate with constant agitation (500 rpm)and oven dried at 393 K for 16 h. Aqueous solutions were notemployed due to the hydrophobic nature of the carbon supportswhich can lead to difficulties with surface wetting that mayadversely affect the ultimate Pd dispersion. The catalyst precur-sors were sieved (ATM fine test sieves) into a batch of 75 µm av-erage particle diameter. The Pd loading (reproducible to within±4%) was determined by ICP-OES (Vista-PRO, Varian Inc.).Bulk PdO (99.998%) was obtained from Sigma–Aldrich.

2.2. Characterization

The three supports, the supported Pd catalysts, bulk PdO andsupport + PdO physical mixtures were subjected to a compre-hensive program of characterization. BJH pore volume analy-ses were performed on the supported catalysts using the com-mercial Micromeritics TriStar 3000 unit; N2 at 77 K servedas sorbate. The total pore volumes associated with Pd/AC,Pd/graphite and Pd/GNF are 0.47, 0.04, and 0.08 cm3 g−1,respectively. BET surface area, temperature programmed re-duction (TPR), H2 chemisorption, and H2 temperature pro-grammed desorption (TPD) measurements were conducted us-ing the commercial CHEMBET 3000 (Quantachrome Instru-ments), employing a thermal conductivity detector where aknown mass (�0.3 g) of sample was loaded into a U-tube(10 cm × 3.76 mm i.d.): data acquisition/manipulation em-ployed the TPR Win software. BET areas were recorded in a30% v/v N2/He flow; pure N2 (99.9%) served as the internalstandard. At least 2 cycles of nitrogen adsorption–desorptionwere employed to determine total surface area using the stan-dard single point method. TPR employed a reducing gas mix-ture of 5% v/v H2/N2 (mass flow controlled at 20 cm3 min−1)with a heating rate of 10 K min−1 to 523 K; the effluent gaswas directed through a liquid N2 trap. TPR was also moni-tored using a Seiko Instruments Inc. TG/DTA 320 Simulta-neous Thermo-Gravimetric/Differential Thermal Analyser cou-pled to a MICROMASS PC Residual Gas Analyser. A knownquantity of catalyst (ca. 18 mg) was placed in a Pt sample panand subjected to the treatments given above, monitoring theeffluent gas over the mass range 10–100. After TPR, the re-duced samples were swept with a flow of N2 for 1 h, cooledto room temperature and subjected to H2 chemisorption using apulse (50 µl) titration procedure. At this pulse volume, the max-imum H2 partial pressure in the sample cell (0.004 atm) is well

198 C. Amorim, M.A. Keane / Journal of Colloid and Interface Science 322 (2008) 196–208

below 0.013 atm, the pressure needed for Pd hydride forma-tion at room temperature [24,46]. Hydrogen pulse introductionwas repeated until the signal area was constant, indicating sur-face saturation. The samples were then thoroughly flushed withN2 for 30 min and a TPD was conducted in flowing N2 at50 K min−1 to 873 K. Alternatively, TPD (at 10 K min−1) to523 K was followed by a re-introduction of H2 and the samplestaken to room temperature where the resultant signal represents,in effect, a “reverse” TPR. A comprehensive examination of thetemperature induced Pd hydride decomposition/formation wasperformed with the three supported and bulk Pd samples, em-ploying the same reductant mixture with a forward and reverseramping (1–10 K min−1, 300–523 K). In order to evaluate thepossible occurrence of hydrogen spillover at room temperature,the samples (without prior TPR) were contacted with 5% v/vH2/N2 at room temperature for 6 h with a subsequent directTPD to 873 K at 50 K min−1. BET surface area and H2 uptakevalues were reproducible to within ±3%.

The Pd particle morphology and size distributions of thesupported catalysts were determined by transmission electronmicroscopy: JEOL 2000 TEM microscope operated at an ac-celerating voltage of 200 kV. The samples were dispersedin 1-butanol by ultrasonic vibration, deposited on a lacey-carbon/Cu grid (300 Mesh) and dried at 383 K for 12 h. Atleast 500 individual Pd particles were counted for each catalystand the mean Pd particle sizes are quoted in this paper as thenumber average diameter (d), surface area-weighted diameter(ds), and volume-weighted diameter (dv):

(1)d =∑

i

nidi

/∑i

ni,

(2)ds =∑

i

nid3i

/∑i

nid2i ,

(3)dv =∑

i

nid4i

/∑i

nid3i ,

where ni is the number of particles of diameter di and∑

i ni >

500. X-ray diffraction (XRD) analysis was also performed us-ing a Philips X’Pert instrument with Ni filtered CuKα radia-tion. The samples were mounted in a low background sampleholder and scanned at a rate of 0.02◦ step−1 over the 20◦ �2θ � 90◦ range with a scan time of 5 s step−1. The diffrac-tograms were compared with the JCPDS-ICDD [47] referencesfor identification purposes.

2.3. Catalytic procedure

The hydrodechlorination (HDC) of chlorobenzene (CB) wascarried out in the gas phase under atmospheric pressure in afixed bed glass reactor (i.d. = 15 mm) with a co-current flowof CB in pure He at 325 K in the presence or absence ofPd hydride. The CB feed (in n-hexane) was delivered, via aglass-teflon air-tight syringe and teflon line at a fixed calibratedflow rate, using a model 100 (kd Scientific) microprocessor-controlled infusion pump. The catalytic reactor and operatingconditions to ensure negligible heat/mass transport limitationshave been described in detail elsewhere [48,49]. A schematic

of the reaction conditions which identifies the prevailing con-ditions wherein Pd hydride is present or absent is given inFig. 1. The catalysts were activated (10 K min−1) at 523 Kfor 1 h in a mixture of 5% v/v H2/N2 (mass flow controlled at100 cm3 min−1), the reactor cooled to room temperature (RT)and held for 1 h to allow the formation of Pd hydride and thetemperature was then raised to 325 K (reaction temperature) at1 K min−1. After an isothermal (325 K) hold for 15 min, theH2 flow was discontinued and after two minutes, reaction wasperformed in He (95 cm3 min−1) for 50 min. The 2 min hold at325 K in a He flow ensured a thorough flushing of the lines; vol-ume of He introduced is ca. four times that of reactor + supplylines. Post-HDC, H2 (5% v/v H2/He) was reintroduced and thetemperature was decreased to room temperature for 1 h to allowthe reformation of Pd hydride. The catalyst was then “activated”for 1 h at 523 K (10 K min−1) and temperature was decreased to325 K (reaction temperature), held for 15 min, H2 supply wascut and, after 1 h in flowing He, reaction was performed. Thistwo reaction cycle was repeated three times. As blank tests, pas-sage of CB in a stream of H2 through the empty reactor or overthe carbon supports alone, i.e., in the absence of Pd, did notresult in any detectable conversion. HDC was conducted at aninlet Cl/Pd = 3.5 molCl mol−1

Pd h−1 and GHSV = 6 × 104. Thereactor effluent was collected in a liquid-nitrogen trap for off-line analysis using capillary GC (Perkin–Elmer Auto SystemXL, DB-1 50 m × 0.20 mm i.d., 0.33 µm column); detectionlimit corresponded to a feedstock conversion <0.4 mol% whereoverall analytic reproducibility was better than ±5%. The de-gree of hydrodechlorination (xCl) is given by

(4)xCl = [HCl]out

[Clorg]in,

where [Clorg] represents chlorine concentration associated withthe aromatic feed; in and out refer to the inlet and outlet reac-tor streams, respectively. Repeated catalytic runs with differentsamples from the same batch of catalyst delivered product com-positions that were reproducible to within ±6%. Chloroben-zene (99.9%) was obtained from Sigma–Aldrich and used with-out further purification; ultra high purity (99.999%) H2, He, andN2 were supplied by Scott-Gross Co. Inc.

3. Results and discussion

3.1. Catalyst characterization

3.1.1. Palladium hydrideThe TPR profiles generated for bulk PdO, physical mixtures

and supported systems are given in Fig. 2; repeated TPR runsare included to illustrate the level of reproducibility. All theTPR profiles are characterized by a single negative peak at 382± 6 K. A search through the literature has unearthed an array ofTPR profiles for carbon supported Pd that exhibit the sole ap-pearance of positive (H2 consumption) [11,25,26,33,34,36–38,50–53] or negative (H2 production) [6,25,26,34,35,38,53,54]peaks or a combination of both [35,36,38,55]. In a series ofblank runs, there was no observable H2 consumption or releaseby/from the carbon supports over the same temperature range.


Fig. 1. Schematic illustrating catalyst (Pd/AC, Pd/graphite, and Pd/GNF) treatment leading to CB HDC at 325 K in the presence and absence of Pd hydride: RT =room temperature.

This was expected and is consistent with the literature [56,57].None of the Pd systems (bulk/physically mixed/chemically sup-ported) considered in this work generated a positive peak duringTPR in advance of H2 release. There is some consensus in theliterature that the reduction of supported [33,35,51,55,58,59]and unsupported PdO [58,60] occurs at room temperature. Itshould, however, be noted that Nag [25] generated a TPR pro-file that suggested reduction of PdO supported on carbon attwo different temperatures (318 and 345 K). We analyzed, byTG-MS, the effluent gas released during the room temperatureH2 contact of the supported catalysts and detected small weightchanges (<1%) accompanied by an exotherm which coincidedwith the evolution of H2O, N2, NO, and (trace) N2O. These re-sults are diagnostic of precursor decomposition/reduction dur-ing the room temperature H2 contact that preceded TPR. Thenegative peak that appears in each of our TPR profiles can beattributed to H2 release due to the decomposition of β-phasePd hydride [6,25,26,29,30,38]. It is known that Pd can absorbH2 at room temperature to form Pd hydride where H2 partialpressure exceeds 0.013 atm [24–46]. At lower partial pres-sures, H2 dissolves only sparingly in Pd to form α-phase Pdhydride [24]. The temperature corresponding to maximum H2

release/hydride decomposition for bulk Pd (ca. 386 K) is higherthan that (ca. 376 K) recorded for the three supported catalysts.Palladium hydride decomposition from supported Pd has been

reported in the literature [6,25,26,29,30,38,51,53] to occur overthe range 323–373 K. Moreover, this decomposition tempera-ture has been proposed to increase with increasing Pd particlesize [25] and H2 pressure [61]. A particle size dependence isconsistent with our observation that a higher temperature is re-quired for bulk Pd hydride decomposition compared with thesupported systems.

Arriving at a quantifiable measure of the hydrogen contentin Pd hydride, alternatively viewed by some authors [27,28,62]as hydrogen solubility in Pd, is problematic as was noted byBoudart and Hwang [62] in work dating from the 1970s. Hydro-gen solubility can be represented by the ratio Hab/Pd where Hab

is a measure (number of atoms) of absorbed hydrogen. How-ever, an explicit identification of the denominator term (Pd)does not emerge from the literature as it has been adopted ei-ther as the total number of Pd atoms in the catalyst (Pdt) [62]or the “number” of bulk Pd (Pdb) [27,28,62], which has beendefined as total number of Pd atoms less the number of sur-face (Pds, or exposed) Pd (Pdb = Pdt − Pds) or no particularassignment is provided [25]. Boudart and Hwang [62] con-cluded that a dispersed Pd phase (as opposed to bulk Pd) canabsorb less hydrogen and proposed that the ratio Hab/Pdt is alinear function of Pd dispersion, which approaches zero as dis-persion reaches 100%. Indeed, hydrogen solubility in Pd hasbeen shown to increase with increasing Pd size [26,32]. Ap-


Fig. 2. TPR profiles generated for: (a) bulk PdO (I), bulk PdO + AC (II) andPd/AC (III); (b) bulk PdO (I), bulk PdO + graphite (II), and Pd/graphite (III);(c) bulk PdO (I), bulk PdO + GNF (II), and Pd/GNF (III).

plying the ratio Hab/Pdb, rather than Hab/Pdt, implies that H2

solubility is independent of dispersion/Pd particle size for parti-cles larger than 1.5 nm as demonstrated in the literature [28,62]:Hab increases linearly with increasing Pdb keeping the ratioHab/Pdb constant but results in a discontinuity where dispersionapproaches unity. Since the measurement of surface/exposed Pdatoms (Pds) is difficult and subject to appreciable error (as dis-cussed later in this text), the use of the ratio Hab/Pdb (where Pdb

= Pdt − Pds) to measure hydrogen solubility is questionable:any error associated with Pds will only serve to compound thetotal error. As a result, it is proposed in the present study thatPd hydride composition should be quantified by the atomic ra-tio Hab/Pdt, which should have a maximum value of ca. 0.7[27,28,32] in the case of bulk Pd and a minimum value of zeroas dispersion approaches 100%. Bonivardi and Baltanas [32] re-ported Hab/Pdt values for Pd/SiO2 in the range 0.24–0.46 wherethe values were sensitive to heat treatment (in N2 and air) andreduction conditions. Nag [25] found that the value for Pd/Cvaried from 0.10 to 0.26 as a function of increasing Pd particlesize (from 2.4 to 10.6 nm). Krishnankutty et al. [27,28], expect-ing to obtain Hab/Pdb values for Pd/C and Pd/SiO2 around 0.7,recorded values over the range 0.005–0.77. The low hydrideratios were attributed to carbon contamination of the Pd parti-cles, i.e., incorporation of carbon atoms within the Pd lattice,which inhibited Pd hydride formation. The hydrogen solubil-ity (Hab/Pdt) generated in this study for the bulk and supportedPd systems are given in Tables 1 and 2, respectively, and fall

Table 1Hydrogen uptake and Pd hydride composition associated with bulk Pd andphysical mixtures with the carbon substrates

Sample H2 uptake(cm3 gPd

−1)Had/Pdt

a,b Hab/Pdtc,d Hab/Pdt

c,e

Bulk Pd 0.2 2.1 × 10−3 0.66 0.69Bulk Pd + AC 0.3 2.3 × 10−3 0.66 0.67Bulk Pd + graphite 0.2 2.0 × 10−3 0.67 0.70Bulk Pd + GNF 0.2 1.7 × 10−3 0.67 0.69

a Ratio of number of atoms of adsorbed hydrogen to number of atoms of Pdin the sample.

b Based on H2 chemisorption.c Ratio of number of atoms of absorbed hydrogen to total number of atoms

of Pd in the sample.d Hydride decomposition (TPR 298 → 523 K) (see Fig. 2).e Hydride formation (TPR 523 → 298 K) (see Fig. 3).

Table 2Hydrogen uptake, average Pd particles sizes, Pd hydride composition, and BET surface area of supported Pd

Sample H2 uptake(cm3 gPd

−1)ds

a

(nm)db

(nm)ds

b

(nm)dv

b

(nm)dv

c

(nm)σm

b,d Pd particle sizerangeb (nm)

Had/Pdta Hab/Pdt

e Hab/Pdtf BET surface

area (m2 g−1)

Pd/AC 1.0 131 10 49 80 20 0.20 1–125 0.9 × 10−2 0.57 0.59 875Pd/graphite 2.2 57 8 24 33 25 0.10 1–70 1.9 × 10−2 0.53 0.54 11Pd/GNF 1.7 75 13 37 55 23 0.01 2–135 1.5 × 10−2 0.56 0.55 86

a Based on H2 chemisorption.b Based on TEM (see Eqs. (1)–(3)).c Based on XRD line broadening.d Standard deviation of the mean.e Hydride decomposition (TPR 298 → 523 K) (see Fig. 2).f Hydride formation (TPR 523 → 298 K) (see Fig. 3).


Fig. 3. “Reverse” TPR profiles generated for: (a) bulk Pd (I), bulk Pd + AC (II),and Pd/AC (III); (b) bulk Pd (I), bulk Pd + graphite (II), and Pd/graphite (III);(c) bulk Pd (I), bulk Pd + GNF (II), and Pd/GNF (III).

within the range of values quoted in the literature [27,28,60]with a higher solubility value recorded for bulk Pd relative tosupported Pd. To evaluate if the formation of Pd hydride is areversible phenomenon, a “reverse” TPR was performed andthe results, which were reproducible (see replicate profiles), arepresented in Fig. 3. From a consideration of the Tmax associ-ated with the reverse TPR peaks, (re-)formation of Pd hydridetakes place at room temperature in the presence of an externalsupply of H2. Palladium hydride formation is entirely reversibleand the Hab/Pdt values obtained from the “reverse” TPR agreewell with those generated in the “forward” TPR; see entries inTables 1 and 2.

3.1.2. Pd particle size: H2 chemisorption, TEM, and XRDThe H2 uptake, BET surface area and average Pd particle

sizes (of supported Pd) derived from chemisorption, TEM, andXRD analyses are presented in Tables 1 and 2. Specific (pergram of Pd) H2 uptake was higher on the supported catalystswhen compared with bulk Pd, as expected, where the supportserves to disperse the metal, resulting in smaller metal parti-cle sizes and higher gas uptake. It is essential that H2 uptakemeasurements used to ascertain Pd dispersion are performedunder conditions where Pd hydride formation does not con-tribute to H2 consumption. While it is evident in certain studies[6,8–12,31] that H2 chemisorption was conducted under con-ditions where Pd hydride formation was circumvented, thereare a number of instances in the literature [63–65] where criti-cal experimental conditions, i.e., temperature and/or hydrogenpressure, are not given and so possible interferences due to Pdhydride formation can not be discounted. Representative TEMmicrographs and Pd particle size distributions of the three sup-ported catalysts are presented in Fig. 4. It is apparent that the Pdparticles supported on AC exhibit a pseudo-spherical or glob-ular geometry (Fig. 4a) whereas the metal phase on graphite(Fig. 4b) and, to a lesser extent, on GNF (Fig. 4c) can be char-acterized as faceted and relatively thin Pd particles. Faceting ofsupported metal particles is diagnostic of metal/support interac-tion [2], which seems to apply to the structured carbon carriers.Hydrogen TPD performed after H2 chemisorption generated theprofiles shown in Fig. 5; repeated TPD profiles are included todemonstrate reproducibility of these measurements. The TPDprofile generated for bulk Pd exhibited a broad desorption re-sponse with a Tmax at ca. 566 K; the volume of H2 releasedmatches that taken up in the chemisorption step which pre-ceded TPD, confirming that H2 uptake measurements did notinclude any contribution due to the formation of Pd hydride. Itshould be noted that potential contributions due to Pd hydrideformation to overall H2 uptake can be significant given that theamount of H2 associated with the hydride is up to ca. 60 timesgreater than that chemisorbed on Pd; see Had/Pdt and Hab/Pdtvalues in Table 2. The three Pd + support physical mixtures andthe three supported catalysts delivered a TPD response that wasquite distinct from bulk Pd in that two distinct desorption peaksare in evidence. The Tmax of the lower temperature peak falls inthe range 580–640 K where the associated H2 release is equiv-alent to that chemisorbed prior to TPD and can be linked tothe TPD response recorded for bulk Pd, i.e., surface H2 releasefrom Pd. The additional high temperature peak was generatedonly in the presence of the support, either as a physical mix-ture or as a chemical combination, suggestive of the involve-ment of hydrogen spillover on the carbon substrates. Hydrogenspillover involves the migration of atomic hydrogen to the sup-port after dissociation of molecular hydrogen on the metallicsurface [66]. Hydrogen TPD analysis represents a practical ap-proach to study spillover effects as hydrogen associated withthe metal and the support can be differentiated [43,57,67–70].It should be noted that the intensity of the higher temperaturepeak(s) is greater, in every instance, for the supported Pd cat-alyst when compared with the corresponding physical mix, aresponse that can be attributed to an enhanced (or extended)


(a) (b)

Fig. 4. Representative TEM images illustrating the nature of the supported Pd phase and Pd particle size distributions associated with (a) Pd/AC, (b) Pd/graphite,and (c) Pd/GNF.

metal/support interface when compared with physical contact.In addition, it should be noted that the amount of hydrogenspillover associated with Pd/AC is greater than that observedfor Pd/graphite or Pd/GNF, a result that can be ascribed to thegreater AC BET surface area. Evidence for hydrogen spilloveron supported metal catalysts [1,43,66,67,71,72] and physicalmixtures of supported catalyst + support [57,73] has been pre-sented in the literature. However, we could find no comparablepublished H2 TPD studies for bulk metal + support combi-nations. The Tmax associated with the “spillover” peak occursin the final isothermal hold (873 K) for the supported cata-lysts system and in the range 800–873 K for the physical mix-tures. Hydrogen desorption from the metal phase and from thesupport has been demonstrated to occur at different tempera-tures [43,57,67] where removal of spillover hydrogen requirestemperatures in excess of 503 K, regardless of the nature ofthe metal or support [43,57,67–70]. In their H2 TPD analy-sis of Rh/Al2O3 physically mixed with Al2O3, SiO2, activatedcarbon and zeolite, Benseradj et al. [57] observed that the des-orption peak associated with spillover hydrogen (Tmax in therange 503–873 K) increased with increasing Rh/Al2O3 dilutionfor each of the four supports. Ouchaib et al. [43] reported two

H2 desorption peaks from charcoal supported Pd at 373 and673 K and attributed the higher temperature peak to spilloverhydrogen. By applying TPD analysis to Pt/Al2O3, Ni/Al2O3,and the Al2O3 support following H2 chemisorption, Kramerand Andre [67] observed an additional high temperature peak(ca. 753 K) in the TPD profiles of both supported metal cata-lysts that was not present in the profile of Al2O3 or unsupportedPt. The occurrence of “spillover” hydrogen has been identifiedin the literature as a possible contribution to H2 uptake mea-surements [1,74] (which should result in an overestimate of H2

uptake) but is not the case in this study given the equivalencyof H2 uptake during pulse titration and the lower temperatureTPD peak.

To check the possible occurrence of hydrogen spillover atroom temperature, a H2 TPD was performed on all the sam-ples after exposure to a flow of H2 at room temperature (forat least 6 h) without any other sample treatment. This roomtemperature H2 contact is sufficient to reduce the PdO contentas has been established in the TPR analysis; see Fig. 2. Theresultant TPD profiles are presented in Fig. 6. The TPD associ-ated with bulk Pd presents, again, only a single low temperaturepeak. In common with the results generated when the samples


(c)

Fig. 4. (continued)

were first reduced at 523 K, a second higher temperature TPDpeak arises due to the presence of the support, indicating thatH2 spillover does occur at room temperature. Moreover, theTmax associated with the “spillover” peaks match that gener-ated post-TPR/pulse H2 chemisorption. The low temperaturedesorption peak associated with bulk Pd containing sampleswas noticeably more intense than that observed post-TPR, a re-sponse that can be accounted for on the basis of metal sinteringduring TPR which also lowers the extent of spillover that canbe generated in the physical mix. These results demonstrate thathydrogen spillover can take place at room temperature as hasbeen reported for Au [75] and Pt [76–80] where carbon [80],Fe2O3 [75], TiO2 [75,79], WOx /ZrO2 [76,77], and Al2O3 [78]served as supports. However, we could find no evidence in theliterature of room temperature H2 spillover on Pd/C (or Pd + C)and consider that this is the first reported instance of this effect.

Considering the Pd particle size entries in Table 2 for thethree supported Pd catalysts, there is an obvious discrepancybetween the values obtained from H2 chemisorption, TEM, andXRD. The XRD diffractograms for bulk Pd and the three sup-ported Pd catalysts are given in Fig. 7 where the four peaksat 40.1◦, 46.7◦, 68.1◦, and 82.2◦ correspond, respectively, to(111), (200), (220), and (311) Pd planes and are consistentwith an exclusive cubic symmetry. In addition, the XRD pro-

Fig. 5. H2 TPD profiles post-TPR (see Fig. 2) associated with: (a) bulk Pd (I),bulk Pd + AC (II), and Pd/AC (III); (b) bulk Pd (I), bulk Pd + graphite (II),and Pd/graphite (III); (c) bulk Pd (I), bulk Pd + GNF (II), and Pd/GNF (III).

files for Pd/graphite and Pd/GNF include a peak at 26◦ that ischaracteristic of structured (graphitic) carbon [47]. While gaschemisorption provides a surface-area weighted diameter, theparticle size measured by XRD is based on the Scherrer for-mula which yields a volume weighted average [41]. For anygiven particle size distribution, the number average diameterwill be less than the surface-area weighted value which, inturn, is smaller than the volume weighted value; the devia-


Fig. 6. H2 TPD profiles after a room temperature H2 contact associated with:(a) bulk Pd (I), bulk Pd + AC (II), and Pd/AC (III); (b) bulk Pd (I), bulk Pd+ graphite (II), and Pd/graphite (III); (c) bulk Pd (I), bulk Pd + GNF (II), andPd/GNF (III).

tion in mean values is greater for a wide particle size distri-bution. In heterogeneous catalysis, the surface area weightedmean is more meaningful as it can be used to derive a spe-cific metal surface area, which is calculated from the ratio ofthe third moment to the second moment (d3/d2) of the aver-age size [41]. XRD line broadening analysis yielded mean Pd

Fig. 7. XRD patterns for the activated (a) bulk Pd, (b) Pd/AC, (c) Pd/GNF,and (d) Pd/graphite. Note: the solid lines indicate peak position (with relativeintensity) for cubic Pd.

particle sizes that were essentially equivalent (20–25 nm, seeTable 2) for the three supported catalyst, a result that is in directcontrast to the TEM/chemisorption measurements. Taking theTEM derived particle size distribution (see Fig. 4), number av-eraged, surface-area and volume weighted values are obtainedfrom the application of Eqs. (1)–(3) and recorded in Table 2.Agreement [26,32,35] and disagreement [4,28,29] of particlesize values calculated from gas chemisorption, TEM and XRDcan be found in the literature. It is worth flagging the workof Bonivardi and Baltanas [32], who reported comparable Pdsizes (on SiO2) using H2 chemisorption and TEM (surface areaweighted values) analysis. Pinna et al. [35], employing XRDand CO chemisorption, obtained conformity in terms of Pd sizeby assuming a Pd/CO stoichiometry = 2. Neri et al. [26] char-acterized eight activated carbon supported Pd samples, demon-strating good agreement of Pd size results from TEM (surface-area weighted) and CO chemisorption (Pd/CO = 1), with theexception of one catalyst. It was claimed that as this catalystexhibited a lower dispersion, a Pd/CO = 2 was more appropri-ate as CO tends to adopt a bridged orientation when interactingwith larger Pd particles. Bonarowska et al. [29], on the otherhand, reported comparable Pd dispersion values obtained fromH2 and CO chemisorption on Pd/C but these differed signif-icantly from that provided by XRD, a disagreement that wasattributed to Pd-C interactions. Benitez and Angel [4], charac-terizing Pd/C, Pd/Al2O3, Pd/SiO2, Pt/C, and Rh/C, found con-sistent and inconsistent particle sizes using H2 chemisorptionand TEM analysis; the nature of TEM particle size (number,surface-area or volume weighted) was not specified. Krish-nankutty and Vannice [28] recorded Pd particle sizes (on carbonblack) obtained from H2, CO, and O2 chemisorption that weremarkedly greater than those which resulted from TEM (surfacearea-weighted) and XRD analyses. Suppressed chemisorption(gas uptake lower than expected) was ascribed to carbon con-tamination of the Pd sites resulting from the decomposition ofthe precursor Pd(C5H7O2)2 and C migration from the highly


disordered carbon black surface during high temperature H2 re-duction. A suppression of H2 uptake has also been observed byWunder and co-workers [56] for graphite and carbon black sup-ported Pd.

While the studies cited above have demonstrated that dif-ferent characterization techniques can deliver varying particlesizes for a given catalyst, no meaningful guideline has emergedwith regard to the methodology that is most appropriate and ac-curate to obtain an average particle size/surface area. It shouldbe stated that the studies which considered alternative measure-ment techniques are few when compared with the innumerablereports in the literature that have made use of only one charac-terization technique (typically gas chemisorption) to quote Pdparticle size/dispersion without recourse to a second techniquein order to check accuracy/reproducibility. Our H2 chemisorp-tion results delivered markedly higher Pd particle sizes whencompared with the TEM surface weighted values. As the sup-ported catalysts were prepared from an inorganic precursor andgraphite and GNF have ordered structures and small surfaceareas, carbon contamination of the Pd particles is not a likelycontributing factor in our system. A closer examination of thetwo assumption [7,27–31,33] inherent in particle size estima-tion from H2 chemisorption values is called for: (a) Pd crys-tals possess a spherical morphology; (b) an exclusive dissocia-tive adsorption on Pd, i.e., H2/Pd uptake stoichiometry = 1/2.The representative TEM images reveal that the geometricalmorphologies adopted by Pd on graphite (Fig. 3b) and GNF(Fig. 3c) and the pseudo-spherical shape adopted by Pd onAC (Fig. 3a) are not consistent with the first assumption. Inaddition, the suggestion that the adsorption stoichiometry isindependent of support and dispersion is at best a convenientapproximation and must contribute to a divergence in particlesize values. Indeed, Matyi et al. [41] have considered the assign-ment of a particular adsorption stoichiometry as the principalimpediment to obtaining reliable particle size values from gaschemisorption. In studying H2 uptake on Ir/Al2O3 and Ir/SiO2,Guil and co-workers [81] found the adsorption stoichiometryto vary from 1.1 to 2.4, depending on support and particlesize/shape. Moreover, Reut and Kamalov [82] observed the ad-sorption stoichiometry for H2, O2, and CO on Pd to increasenon-linearly with increasing dispersion. Another strong indi-cation that the adsorption stoichiometry is sensitive to metalparticle size comes from the work of Chou and Vannice [39]who demonstrated that the heat of adsorption of H2 on Pd sup-ported on a range of oxides was dependent on metal dispersion.

TEM analysis, as an imaging technique, provides a more di-rect measure of particle size than that which results from gaschemisorption. However, such limitations as diameter measure-ment of particles with irregular shapes, the two-dimensionalimaging of a three-dimensional structure and poor contrast be-tween metal and support are possible sources of inaccuracy[40,41]. It is readily evident from the TEM images presentedin Fig. 3 that metal/support image contrast is not a constraint inthis study whereas the morphological diversity of the supportedPd particles does present problems in terms of consistent diam-eter measurement. Particle size of nonspherical shaped particlescan be obtained by taking an average between the longest and

Fig. 8. Standard deviation of the mean (σm) as a function of the number of Pdparticles counted (n) from representative TEM micrographs of Pd/AC.

shortest segments that intersect the center of the particle. It is,nevertheless, worth noting (see Table 2) that the order of in-creasing H2 uptake, Pd/AC < Pd/GNF < Pd/graphite, followsthe sequence of decreasing surface-weighted particle size fromTEM analysis. One of the greatest difficulties in TEM analysis,due to the small quantity of material that is examined, is to esti-mate accurately the degree to which the observed particle pop-ulation is representative of the entire ensemble of crystallites.In order to determine how closely the observed distribution ap-proaches the true population, the standard deviation of the mean(σm) must be determined [41],

(5)σm =∑

d2i − nd2

n2,

where di is the diameter of the ith particle, n is the numberof particles counted, and d is the number average diameter.The standard deviation (σm) decreases as n increases and, ingeneral, a broad particle size distribution necessitates a highercount when compared with narrow distributions. In addition,it is essential that both the very small and very large particlesare included in the count. Therefore, whenever TEM analysis isapplied to obtain average particle size, it is imperative that anevaluation is made of the standard deviation of the mean in or-der to establish the accuracy (or reliability) of the measurement.None of the studies that we consulted in the literature whereinan average Pd particle size was obtained from TEM analysis in-cluded a value for σm. The validity of the average particle sizevalue can be established from an examination of σm sensitivityto the total particle count (n) where σm invariance with increas-ing n ensures the attainment of a truly representative particlesize. The latter response is illustrated in Fig. 8, taking Pd/ACas a representative case. The σm values recorded in Table 2confirm that the observed size distribution from TEM analy-sis captures the true population and is a valid representation ofPd dispersion.

From the foregoing, we can state that H2 chemisorption doesnot deliver an accurate or true measure of particle size and, byassociation, dispersion/surface area of Pd on carbon supports.TEM analysis generates more reliable/valid Pd size informationwhere image contrast between the metal and the support is not


a constraint. Provided that a consistent measurement is appliedto take account of particle morphological heterogeneity, the sta-tistical rigor to which the measurement is subjected remainsthe critical factor. A direct measurement of H2 uptake capac-ity is, nonetheless, an important measurement for catalysts usedto promote H2 mediated reactions and should be related to themetal characteristics obtained from TEM analysis. Hydrogenuptake alone represents an incomplete picture but in tandemwith strategic thermal desorption measurements can facilitate apartitioning of the hydrogen into that associated with Pd (dis-sociatively chemisorbed H atoms and absorbed as Pd hydride)and spillover species on the support.

3.2. Role of Pd hydride in CB HDC

In a previous publication [83], we examined the possible roleof chemisorbed and spillover hydrogen in determining HDC be-havior and established that both can contribute to the catalyticactivity of supported Pd. We consider here the possible partic-ipation of Pd hydride in the HDC of CB where the reactionwas performed (at the same temperature) in the presence andabsence of Pd hydride. A series of comprehensive pre-reactionscreening tests were undertaken (see Fig. 1 and experimentalsection for details) to ascertain the precise temperature require-ments where the hydride is stable or undergoes decomposition.These results are presented in Fig. 9 where it can be seen that,depending on the temperature ramping sequence, Pd hydrideis present (Fig. 9a) or absent (Fig. 9b) at 325 K. The profilespresented in Fig. 9a establish that once Pd hydride is formedat room temperature, it does not decompose when the tem-perature is increased to 325 K where the following should benoted: the absence of a negative peak (hydride decomposition)over the temperature range 298–325 K and in the isothermalregion (325 K maintained for 3 h); the appearance of a neg-ative peak (hydride decomposition) once the temperature wasraised from 325 to 523 K. The entries in Fig. 9b demonstratethat when the temperature is decreased from 523 to 325 K (re-action temperature) Pd hydride does not form, an assertion thatis based on: absence of a positive peak (hydride formation) inthis temperature range; absence of a negative peak (due to hy-dride decomposition) when the temperature was subsequentlyincreased from 325 to 523 K; the appearance of the positivepeak (hydride formation) when the temperature was decreasedto room temperature. The manipulation of temperature shownin Fig. 9 matches the activation conditions employed beforecatalysis to ensure the presence or absence of Pd hydride. Thetemporal variation of CB fractional conversion (xCl), where thetwo reaction cycle shown in Fig. 1 were repeated three times forthe three supported catalysts is shown in Fig. 10. In the absenceof Pd hydride, there was no detectable CB conversion over eachcatalyst, which means that any chemisorbed or spillover hydro-gen does not contribute to HDC under the reaction conditionsapplied. Involvement of chemisorbed and spillover hydrogenobserved in our earlier study [83] occurred at a quite differentreaction temperature, Cl/Pd mole ratio and H2 partial pressure.It must be stressed that, in the present study, HDC was per-formed in the absence of an external supply of H2 in order

(a)

(b)

Fig. 9. Pd hydride formation/decomposition as a function of temperature for (I)bulk Pd, (II) Pd/AC, (III) Pd/graphite, and (IV) Pd/GNF; (a) establishes condi-tions where Pd hydride is present at 325 K; (b) establishes conditions where Pdhydride is absent at 325 K.

to unequivocally demonstrate the participation of Pd hydridein HDC. In the presence of Pd hydride, CB HDC generatedbenzene as the only product with no detectable cyclohexane inthe effluent stream. The HDC activity delivered by each cata-lyst declined with time to a negligible conversion after ca. 40min on-stream due to consumption of the available hydride.HDC activity was fully recoverable as can be seen from theentries in Fig. 10 for the repeat runs once Pd hydride was re-plenished during the reactivation step; bulk Pd exhibited com-parable behavior. We can explicitly assign the observed HDCactivity to the action of the Pd hydride. In terms of CB activa-tion, it has been reported that haloarenes can adsorb from thegas phase on both the dispersed metal [5,12] and the carbonsupport [84,85]. The slight variations in fractional dechlorina-tion with time over the three catalysts may be accounted for interms of differences in CB dynamic interactions with the sup-port which may be sensitive to carbon structure/area/electroniccharacteristics.


Fig. 10. Fractional CB HDC (xCl) as a function of time-on-stream where Pdhydride is initially present (see Fig. 9a) in (a) Pd/AC, (b) Pd/graphite and,(c) Pd/GNF; 1st run (solid bars), 2nd run (open bars), and 3rd run (hatchedbars).

The role of Pd hydride in catalysis remains contentious withboth positive [86,87] and negative [46] effects ascribed to Pdhydride in hydrogenation applications. With regard to catalyticHDC, Rupprechter and Somorjai [88] have proposed a partici-pation of Pd hydride in the HDC of CCl2F-CF3 over Pd(111) asthey observed that catalytic activity was maintained for a cer-tain period of time after a supply of H2 was removed from thesystem. Palczewska, in an older review article [46] dealing withthe reactivity of Pd and Ni hydride phases, actually drew atten-tion to studies where it was claimed that Pd hydride formationrepresents a “self-poisoning” effect where Pd loses its high ac-

tivity when transformed into a hydride. The latter effect wasrationalized on the basis that with the formation of Pd hydride,the hydrogen 1s electron fills the empty energy levels in thePd d band, consequently transforming the initial d transition Pdinto an s–p metal with a resultant loss of chemisorption capa-bility. On the other hand, Rennard and Kokes [86], studyingthe effect of Pd hydride on the hydrogenation of ethylene overPd (powder), reported a sixty-fold increase in activity when theH content in the hydride was raised to 0.50. Such apparent in-consistencies prompted Karpinski [89] to draw attention to theneed for fundamental studies that can irrefutably establish Pdhydride involvement in heterogeneous catalysis. In this study,we have clearly demonstrated that the hydrogen component inPd hydride can participate in catalytic HDC.

4. Conclusions

We have subjected bulk Pd and Pd supported on AC, graphiteand GNF to a range of complementary characterization analy-ses and have considered the possible participation of Pd hydridein catalytic HDC. Hydrogen associated with the activated sup-ported catalysts is present as Pd hydride, chemisorbed (on sur-face Pd) and spillover (on the support) species. In terms of Pdhydride, the results obtained demonstrate that: (i) hydride pres-ence or absence at 325 K (reaction temperature) can be tunedthrough alternative temperature ramping sequences where hy-dride formation and decomposition take place at room tem-perature and ca. 380 K, respectively; (ii) hydride formation isreversible. H2 TPD analysis of supported Pd and bulk Pd + sup-port physical mixtures clearly demonstrates the presence of hy-drogen spillover on the carbon supports that can be generated atroom temperature. While TEM and H2 chemisorption deliveredthe same sequence of increasing Pd particle size, there was anappreciable difference in the values obtained (by up to a factorof 3). This apparent inconsistency is ascribed to the assumptionof an exclusive H2/Pd adsorption stoichiometry = 1/2. TEManalysis provides a valid mean particle size once it is estab-lished that the associated standard deviation is small and in-sensitive to additional particle counting. XRD line broadeninganalysis yielded mean particle sizes that were essentially equiv-alent for each supported catalyst (20–25 nm). Our catalytic re-sults demonstrate that hydrogen from Pd hydride contributes tothe catalytic HDC of CB under conditions where spillover andchemisorbed hydrogen are inactive. Pd hydride activity fell tozero within ca. 40 min on-stream due to hydride consumptionbut activity was fully recovered once the hydride was replen-ished during a reactivation step.

Acknowledgment

This work was supported in part by the National ScienceFoundation through Grant CTS-0218591.

References

[1] D.J. Suh, T.J. Park, S.K. Ihm, Carbon 31 (1993) 427.[2] C. Park, M.A. Keane, J. Colloid Interface Sci. 266 (2003) 183.


[3] C. Pham-Huu, N. Keller, G. Ehret, L.J. Charbonniere, R. Ziessel, M.J.Ledoux, J. Mol. Catal. A Chem. 170 (2001) 155.

[4] J.L. Benitez, G.D. Angel, React. Kinet. Catal. Lett. 70 (2000) 67.[5] F.J. Urbano, J.M. Marinas, J. Mol. Catal. A Chem. 173 (2001) 329.[6] N. Lingaiah, P.S.S. Prasad, P.K. Rao, F.J. Berry, L.E. Smart, Catal. Com-

mun. 3 (2002) 391.[7] M.A. Aramendia, V. Borau, I.M. Garcia, C. Jimenez, F. Lafont, A. Mari-

nas, J.M. Marinas, F.J. Urbano, J. Catal. 187 (1999) 392.[8] R. Gopinath, K.N. Rao, P.S.S. Prasad, S.S. Madhavendra, S. Narayanan,

G. Vivekanandan, J. Mol. Catal. A Chem. 181 (2002) 215.[9] P. Bodnariuk, B. Coq, G. Ferrat, F. Figueras, J. Catal. 116 (1989) 459.

[10] F.J. Berry, L.E. Smart, P.S.S. Prasad, N. Lingaiah, P.K. Rao, Appl. Catal.A Gen. 204 (2000) 191.

[11] N. Lingaiah, P.S.S. Prasad, P.K. Rao, L.E. Smart, F.J. Berry, Appl. Catal.A Gen. 213 (2001) 189.

[12] B. Coq, G. Ferrat, F. Figueras, J. Catal. 101 (1986) 434.[13] N. Lingaiah, M.A. Uddin, A. Muto, Y. Sakata, J. Chem. Soc. Chem. Com-

mun. 17 (1999) 1657.[14] S. Jujjuri, E. Ding, S.G. Shore, M.A. Keane, Appl. Organomet. Chem. 17

(2003) 493.[15] L. Prati, M. Rossi, Appl. Catal. B Environ. 23 (1999) 135.[16] B. Coq, G. Ferrat, F. Figueras, Coord. Chem. Rev. 178–180 (1998) 1753.[17] V.A. Yakovlev, V.V. Terskikh, V.I. Simagina, V.A. Likholobov, J. Mol.

Catal. A Chem. 153 (2000) 231.[18] L. Lassova, H.K. Lee, T.S.A. Hor, J. Mol. Catal. A Chem. 144 (1999) 397.[19] G.D. Angel, J.L. Benitez, J. Mol. Catal. A Chem. 165 (2001) 9.[20] M.A. Aramendia, V. Borau, I.M. Garcia, C. Jimenez, F. Lafont, A. Mari-

nas, J.M. Marinas, F.J. Urbano, J. Mol. Catal. A Chem. 184 (2002) 237.[21] M.A. Aramendia, R. Burch, I.M. Garcia, A. Marinas, J.M. Marinas,

B.W.L. Southward, F.J. Urbano, Appl. Catal. B Environ. 31 (2001) 163.[22] W. Juszczyk, A. Malinowski, Z. Karpinski, Appl. Catal. A Gen. 166

(1998) 311.[23] M.A. Aramendia, V. Borau, I.M. Garcia, C. Jimenez, J.M. Marinas, F.J.

Urbano, Appl. Catal. B Environ. 20 (1999) 101.[24] J.E. Benson, H.S. Hwang, M. Boudart, J. Catal. 30 (1973) 146.[25] N.K. Nag, J. Phys. Chem. B 105 (2001) 5945.[26] G. Neri, M.G. Musolino, C. Milone, D. Pietropaolo, S. Galvagno, Appl.

Catal. A Gen. 208 (2001) 307.[27] N. Krishnankutty, J. Li, M.A. Vannice, Appl. Catal. A Gen. 173 (1998)

137.[28] N. Krishnankutty, M.A. Vannice, J. Catal. 155 (1995) 312.[29] M. Bonarowska, B. Burda, W. Juszczyk, J. Pielaszek, Z. Kowalczyk,

Z. Karpinski, Appl. Catal. B Environ. 35 (2001) 13.[30] M. Bonarowska, J. Pielaszek, V.A. Semikolenov, Z. Karpinski, J. Catal.

209 (2002) 528.[31] A. Gampine, D.P. Eyman, J. Catal. 179 (1998) 315.[32] A.L. Bonivardi, M.A. Baltanas, J. Catal. 138 (1992) 500.[33] G.M. Tonetto, D.E. Damiani, J. Mol. Catal. A Chem. 202 (2003) 289.[34] P.S.S. Prasad, N. Lingaiah, S. Chandrasekhar, K.S.R. Rao, P.K. Rao, K.V.

Raghavan, F.J. Berry, L.E. Smart, Catal. Lett. 66 (2000) 201.[35] F. Pinna, F. Menegazzo, M. Signoretto, P. Canton, G. Fagherazzi, N. Per-

nicone, Appl. Catal. A Gen. 219 (2001) 195.[36] J.K. Murthy, S.C. Shekar, V.S. Kumar, K.S.R. Rao, Catal. Commun. 3

(2002) 145.[37] L.M. Gomez-Sainero, A. Cortes, X.L. Seoane, A. Arcoya, Ind. Eng.

Chem. Res. 39 (2000) 2849.[38] L.M. Gomez-Sainero, X.L. Seoane, J.L.G. Fierro, A. Arcoya, J. Catal. 209

(2002) 279.[39] P. Chou, M.A. Vannice, J. Catal. 104 (1987) 1.[40] D.B. Williams, C.B. Carter, Transmission Electron Microscopy, vol. I,

Plenum, New York, 1996.[41] R.J. Matyi, L.H. Schwartz, J.B. Butt, Catal. Rev. Sci. Eng. 29 (1987) 41.[42] J. Liu, Microsci. Microanal. 10 (2004) 55.[43] T. Ouchaib, B. Moraweck, J. Massardier, A. Renouprez, Catal. Today 7

(1990) 191.[44] C. Park, M.A. Keane, Catal. Commun. 2 (2001) 171.

[45] C. Park, M.A. Keane, Chem. Phys. Chem. 2 (2001) 733.[46] W. Palczewska, Adv. Catal. 24 (1975) 245.[47] JCPDS-ICDD, PCPDFWIN, Version 2.2, June 2001.[48] G. Tavoularis, M.A. Keane, J. Chem. Technol. Biotechnol. 74 (1999) 60.[49] E.-J. Shin, A. Spiller, G. Tavoularis, M.A. Keane, Phys. Chem. Chem.

Phys. 1 (1999) 3173.[50] E. Rocchini, M. Vicario, J. Llorca, C. Leitenburg, G. Dolcetti, A. Trova-

relli, J. Catal. 211 (2002) 407.[51] C.B. Wang, H.K. Lin, C.M. Ho, J. Mol. Catal. A Chem. 180 (2002) 285.[52] F. Frusteri, F. Arena, A. Parmaliana, N. Mondello, N. Giordano, React.

Kinet. Catal. Lett. 51 (1993) 331.[53] S.C. Shekar, J.K. Murthy, P.K. Rao, K.S.R. Rao, Catal. Commun. 4 (2003)

39.[54] A. Malinowski, W. Juszczyk, M. Bonarowska, J. Pielaszek, Z. Karpinski,

J. Catal. 177 (1998) 153.[55] G. Garcia, J.R. Vargas, M.A. Valenzuela, M. Rebollar, D. Acosta, Mat.

Res. Soc. Symp. 549 (1999) 237.[56] R.W. Wunder, J.W. Cobes, J. Phillips, Langmuir 9 (1993) 984.[57] F. Benseradj, F. Sadi, M. Chater, Appl. Catal. A Gen. 228 (2002) 135.[58] T. Fujitani, E. Echigoya, Nippon Kagaku Kaishi 4 (1991) 261.[59] J. Batista, A. Pintar, D. Mandrino, M. Jenko, V. Martin, Appl. Catal. A

Gen. 206 (2001) 113.[60] C.-W. Chou, S.-J. Chu, H.-J. Chiang, C.-Y. Huang, C.-J. Lee, S.-R. Sheen,

T.P. Perng, C.-T. Yeh, J. Phys. Chem. B 105 (2001) 9113.[61] E.J.A.X. van de Sandt, A. Wiersma, M. Makkee, H. van Bekkum, J.A.

Moulijn, Appl. Catal. A Gen. 155 (1997) 59.[62] M. Boudart, H.S. Hwang, J. Catal. 39 (1975) 44.[63] X.-G. Zhao, Q. Lin, W.-D. Xiao, Appl. Catal. A Gen. 284 (2005) 253.[64] L. Jinjun, J. Zheng, H. Zhengping, X. Xiuyan, Z. Yahui, J. Mol. Catal. A

Chem. 225 (2005) 173.[65] S. Velu, M.P. Kapoor, S. Inagaki, K. Suzuki, Appl. Catal. A Gen. 245

(2003) 317.[66] A.L.D. Ramos, D.A.G. Aranda, M. Schmal, Stud. Surf. Sci. Catal. 138

(2001) 291.[67] R. Kramer, M. Andre, J. Catal. 58 (1979) 287.[68] E.-J. Shin, M.A. Keane, Ind. Eng. Chem. Res. 39 (2000) 883.[69] E.-J. Shin, A. Spiller, G. Tavoularis, M.A. Keane, Phys. Chem. Chem.

Phys. 1 (1999) 3173.[70] J.T. Miller, B.L. Meyers, F.S. Modica, G.S. Lane, M. Vaarkamp, D.C.

Koningsberger, J. Catal. 143 (1993) 395.[71] Z.X. Cheng, S.B. Yuan, J.W. Fan, Q.M. Zhu, M.S. Zhen, Stud. Surf. Sci.

Catal. 112 (1997) 261.[72] M.C. Roman-Martinez, D. Cazorla-Amoros, A. Linares-Solano, C.S.M.

Lecea, Carbon 31 (1993) 895.[73] A.D. Lueking, R.T. Yang, Appl. Catal. A Gen. 265 (2004) 259.[74] P.A. Sermon, G.C. Bond, Catal. Rev. 8 (1973) 211.[75] F. Boccuzzi, A. Chiorino, M. Manzoli, D. Andreeva, T. Tabakova, J. Catal.

188 (1999) 176.[76] J.G. Santiesteban, D.C. Calabro, W.S. Borghard, C.D. Chang, J.C. Vartuli,

Y.P. Tsao, M.A. Natal Santiago, R.D. Bastian, J. Catal. 183 (1999) 314.[77] D.C. Calabro, J.C. Vartuli, J.G. Santiesteban, Top. Catal. 18 (2002) 231.[78] M. Stoica, N.I. Ionescu, React. Kinet. Catal. Lett. 68 (1999) 319.[79] X.S. Li, W.Z. Li, Y.X. Chen, H.L. Wang, Catal. Lett. 32 (1995) 31.[80] S.T. Srinivas, P.K. Rao, J. Catal. 148 (1994) 470.[81] J.M. Guil, A.P. Masia, A.R. Paniego, J.M.T. Menayo, Thermochim. Acta

312 (1998) 115.[82] S.I. Reut, G.L. Kamalov, Kinet. Catal. 37 (1996) 870.[83] C. Amorim, G. Yuan, P.M. Patterson, M.A. Keane, J. Catal. 234 (2005)

268.[84] R. Louw, P. Mulder, J. Environ. Sci. Health 25 (1990) 555.[85] I.W.C.E. Arends, W.R. Ophorst, R. Louw, P. Mulder, Carbon 34 (1996)

581.[86] R.J. Rennard, R.J. Kokes, J. Phys. Chem. 70 (1966) 2543.[87] M. Morkel, G. Rupprechter, H.-J. Freund, Surf. Sci. 588 (2005) L209.[88] G. Rupprechter, G.A. Somorjai, Catal. Lett. 48 (1997) 17.[89] Z. Karpinski, Adv. Catal. 37 (1990) 45.

palladium supported on structured and nonstructured carbon: a consideration of pd particle size and...

Documents