Generation of Pd Model Catalyst Nanoparticles by Spark Discharge

Generation of Pd Model Catalyst Nanoparticles by Spark Discharge

Maria E. Messing,*,† Rasmus Westerstrom,‡ Bengt O. Meuller,† Sara Blomberg,‡

Johan Gustafson,‡ Jesper N. Andersen,‡ Edvin Lundgren,‡ Richard van Rijn,§,| Olivier Balmes,§

Hendrik Bluhm,⊥ and Knut Deppert†

Solid State Physics, Lund UniVersity, Box 118, 221 00 Lund, Sweden, Synchrotron Radiation Research, LundUniVersity, Box 118, 221 00 Lund, Sweden, ESRF, B. P. 220, F-38043 Grenoble, France, Kamerlingh OnnesLaboratory, Leiden UniVersity, P.O. Box 9504, 2300 RA Leiden, The Netherlands, and Chemical SciencesDiVision, Lawrence Berkeley National Laboratory, Berkeley, California 94720

ReceiVed: February 15, 2010; ReVised Manuscript ReceiVed: April 17, 2010

We present a method to deposit Pd nanoparticles with a very small size distribution by an aerosol processonto oxide substrates for the creation of model systems in catalytic research. The Pd nanoparticles arecharacterized by transmission electron microscopy, scanning electron microscopy, X-ray photoelectronspectroscopy, and X-ray diffraction. We confirm the small size dispersion from the desired particle size, andwe show that the particle surface coverage can be highly controlled. Further, our measurements indicate thatan amorphous shell surrounding a crystalline core of the Pd particles may form during the particle synthesisand that the shell contains carbon.

1. Introduction

More than 80% of all chemical products are producedengaging catalysts, which are used to facilitate a specific reactionwithout being consumed. In most cases, the catalyst is in anotherphase than the reactants, for example, a solid catalyst andgaseous reactants, a situation referred to as heterogeneouscatalysis. Because of its practical importance, heterogeneouscatalysis has been studied for centuries. A catalyst used forindustrial purposes or for cleaning of exhaust gases is acomplicated system of materials usually consisting of aninsulating oxide support with dispersed metal nanoparticles ofthe active catalyst, as well as a wide range of additives topromote or poison specific reactions. Because of the materialcomplexity, atomic scale information on the inner workings ofthe catalyst is, at best, limited, and as a consequence, catalystdevelopment is, for a large part, based on a trial-and-errorapproach.

However, because of the need to obtain fundamental informa-tion on catalytic reaction pathways, model systems have beendeveloped. Studies of processes related to heterogeneouscatalysis under ultra-high-vacuum (UHV) conditions on well-defined single-crystal surfaces have been a major part of surfacescience for decades.1

In recent years, more complex material model systems havebeen developed by depositing pure metal or alloy nanoparticlesby molecular beam epitaxy (MBE)2,3 or by wet chemicalmethods4 on a thin oxide film formed on a conducting material.In this way, the electric conductivity of the sample is maintained,allowing for the use of electron-based surface analysis tech-niques under UHV conditions or even in the millibar range.4

Further, recent in situ experiments on MBE grown epitaxial Rh

particles on insulating MgO and Al2O3 substrates demonstratethe potential of surface X-ray diffraction (SXRD)5,6 for in situstudies of catalysts in a working environment.

Here, we present a different route to a model system: size-selected metal nanoparticles deposited onto any substrate byaerosol deposition. The system presented consists of Pd nano-particles with a diameter of 15 or 35 nm deposited on HF-etchedSiO2 or on Al2O3 substrates. The conducting SiOx is used forthe sake of characterization using X-ray photoelectron spec-troscopy (XPS), but the type of metal deposit or substrate canbe chosen almost arbitrarily. Our studies show that the particlescan be produced within a narrow size distribution and that theparticle coverage can be well-controlled. Both of these propertiesare desirable for comparative studies of catalytic properties. Forexample, when using different substrates or when addingadditives to control certain reaction pathways or hinder com-pound formation and sintering. Our studies also show that themain contaminant of the pristine particles is carbon, either inthe form of hydrocarbons or, more likely, contained in anamorphous shell around the crystalline Pd core of the particles.

* To whom correspondence should be addressed. E-mail:[email protected].

† Solid State Physics, Lund University.‡ Synchrotron Radiation Research, Lund University.§ ESRF.| Leiden University.⊥ Lawrence Berkeley National Laboratory.

Figure 1. Schematic of the aerosol generator used to produce palladiumparticles by spark discharge between two electrodes.

J. Phys. Chem. C 2010, 114, 9257–9263 9257

10.1021/jp101390a 2010 American Chemical SocietyPublished on Web 05/03/2010

2. Experimental Methods

To produce palladium model catalyst particles, a com-mercially available aerosol generator, Palas, model GFG 1000,primarily constructed for carbon soot particle production wasused. The mechanism behind particle formation is based onspark discharge between two conducting electrodes positionedin the middle of a polymer chamber with their flat ends separatedby a distance of 2 mm (Figure 1). A 20 nF capacitor is connectedto one of the electrodes and charged by a high-voltage supplywith an adjustable output current.7 When the breakdown voltageof 2 kV is reached, the capacitor discharges instantaneously ina spark across the electrode gap. The local temperature of thespark reaches approximately 20 000-30 000 K,8 leading toevaporation of electrode material. Primary particles are formedby homogeneous nucleation of the supersaturated vapor and aresubsequently transported toward the aerosol outlet by a streamof carrier gas focused between the electrodes. Primary particlesare small particles, typically with a diameter between 1 nm anda few nm, with a homogeneous atomic structure. Further growthof primary particles by condensation and coagulation results inthe production of highly charged, nanometer-sized agglomerateparticles.9 To ensure a constant particle production, that is, aconstant breakdown voltage, the distance between the electrodesis maintained by an electrical motor.

Two parameters can be varied in order to affect particleproduction of this apparatus. The spark discharge frequency canbe set between 0 and 300 Hz, and the carrier flow rate can beset between 2 and 8 L/min. In this investigation, particles weregenerated at spark discharge frequencies of 30, 60, 120, 180,240, and 300 Hz and flow rates of 3.4, 3.9, 4.4, 4.9, 5.4, and5.9 L/min. Furthermore, the carrier gas type is known to affectparticle production.10 In the original setup, an argon-air mixtureis used as the carrier gas, but to comply with cleanlinessrequirements, the carrier gas was replaced with ultrapurenitrogen. In addition, the cylindrical carbon electrodes used inthe original setup were replaced by high-purity palladium rods(99.99%) with diameters of 3 mm, mounted to cylindricalstainless steel holders with diameters of 6 mm in order to fitthe apparatus.

The spark generator was connected to an aerosol nanoparticlesystem setup (Figure 2) in order to enable size distributionmeasurements, reshaping of the agglomerate particles intocompact particles, and controlling the deposition of particles.A �-emitting 63Ni source11 was used as a neutralizer in order toachieve a reproducible and known charge distribution on theagglomerate particles before size selection in a differentialmobility analyzer (DMA), labeled DMA 1 in Figure 2. TheDMA, a standard instrument in aerosol science, classifiescharged particles according to their mobility inside an electricfield.12 This mobility is roughly inversely proportional to the

particle diameter. Following size selection, the agglomerateparticles could be reshaped into more compact particles insidea compaction tube furnace (route 2 in Figure 2). Alternatively,knowing that a majority of the particles that pass the DMA carryone single charge, each in the size range applied here,11 particleconcentration measurements could be directly performed usingan electrometer (route 1 in Figure 2). By stepwise scanning thevoltage of DMA 1 and measuring the resulting particleconcentration, size distribution measurements of the agglomerateparticles were obtained. To size select and measure particleconcentrations of the reshaped particles, a second DMA, labeledDMA 2 in Figure 2, was scanned in a similar fashion. Thecompaction behavior of the particles was examined by scanningthe reshaping temperature and measuring the peak value of thesize distributions for each temperature.13

Depositions of particles for further characterization were donedirectly onto Si substrates as well as lacey carbon film Cu TEMgrids, positioned inside an electrostatic precipitator (ESP) (Figure2). The ESP focuses charged particles onto a collector electrode14

and allows for a high-efficiency deposition of particles. Thesetup used allows for deposition of particles with diameters ofup to 100 nm onto a spot of about 1-3 cm in diameter.Transmission electron microscopy (TEM) (JEOL, model 3000F)operated at 300 kV and equipped with a field emission gun andan X-ray energy-dispersive spectrometer (XEDS) together withscanning electron microscopy (SEM) (FEI, model Nova Nanolab600) was used for morphological, structural, and chemicalinvestigations.

X-ray diffraction (XRD) measurements of the samples werecarried out in a combined UHV/high-pressure flow reactor15

using a photon energy of 18 keV at beamline ID0316 at theEuropean Synchrotron Radiation Facility (ESRF). The X-rayphotoelectron spectroscopy measurements where performedusing the ambient pressure photoelectron spectrometer at theMolecular Science beamline 11.0.217 at the Advanced LightSource (ALS) using a photon energy of 525 eV.

A single-crystal Pd(100) was used as a reference for the bulkPd 3d5/2 binding energy. The single-crystal Pd(100) surface wascleaned by cycles of Ar+ sputtering and subsequent annealingbetween 100 and 700 °C, keeping the crystal in 10-7 mbar ofO2. After the cleaning procedure, no contaminants, such ascarbon, could be detected. The binding energy was calibratedto the Fermi level, which, in the case of the pristine Pd particlesamples, could easily be detected.

3. Results and Discussion

3.1. On-Line and Microscopy Characterization of the PdParticles. Particle production by the spark discharge methodwas found to be a robust and simple way of producing palladiumparticles that can be used as a realistic model for catalysts. By

Figure 2. Schematic of the aerosol nanoparticle system setup.

9258 J. Phys. Chem. C, Vol. 114, No. 20, 2010 Messing et al.

varying the spark discharge frequency and carrier flow rate, theparticle number concentration and the diameter were dramati-cally affected. An increase of spark discharge frequency resultsin an increase of particle number concentration and a shift tolarger particle diameters. This is in good agreement withparticles of other materials generated by spark discharge.7,10,18

Because a higher spark discharge frequency corresponds to anincrease in the number of sparks per second that evaporateselectrode material, a higher number concentration of particlescan be reached. Furthermore, an increase in evaporated particlematerial leads to a higher coagulation rate and hence fastergrowth of particles, resulting in particles with a larger diameter.On the other hand, an increase of carrier flow rate leads to adecrease of particle number concentration and a slight shift tosmaller particle diameters, which is in agreement with observa-tions by Tabrizi et al.10 This is explained by the much shortertime available for coagulation with increased carrier flow rate.

In Figure 3a,b, the particle concentration versus mobilitydiameter is displayed for the different spark discharge frequen-cies at a carrier flow rate of 3.4 and 5.9 L/min, respectively.The mobility diameter is the diameter given by the DMAmeasurements and does not necessarily correspond to thegeometric diameter of a particle; this would only be true forspherical particles. For nonspherical particles, as measured inthe scans presented here, it will allow, however, a comparisonof the different production conditions. For the carrier flow rateof 3.4 L/min, the peak value of the number concentrationincreased from 4.8 × 105 to 1.8 × 106 cm-3 when increasingthe spark discharge frequency from 30 to 300 Hz. At the carrierflow rate of 5.9 L/min, the same increase in spark dischargefrequency resulted in an increased peak value of the particlenumber concentration from 2.3 × 105 to 8.7 × 105 cm-3.Because the spark discharge frequency of 300 Hz combinedwith the carrier flow rate of 3.4 L/min gave the highest particleyield, those parameters were used for all particles produced forfurther investigations.

The increase of spark discharge frequency from 30 to 300Hz also leads to a shift of peak particle diameter from 17 to 33nm and from 10 to 20 nm for the carrier flow rates of 3.4 and5.9 L/min, respectively. In agreement with measurements ofgold particles produced by the same particle generation setupwith nitrogen as the carrier gas,19 an increased carrier flow ratewas observed to result in a decreased number concentration ofparticles for all spark discharge frequencies used. However,contradictory to the measurements of gold particle diameter thatwas almost unaffected by the same change of carrier flow rate,

a shift to smaller peak particle diameters was observed withincreasing carrier flow rate for the palladium particles. Thisobservation would need further investigations before the reasoncould be explained.

From morphological investigations by high-resolution TEM(HRTEM), the as-produced agglomerate particles were foundto consist of primary particles connected into a chainlikestructure (Figure 4a). No significant difference of morphologyor primary particle size was observed between particles producedat different carrier flow rates and/or spark discharge frequencies.The primary particles had diameters ranging from approximately2 to 5 nm. In addition, XEDS measurements confirmed thatthe particles were actually palladium particles, within the generaldetection limit for XEDS of <1 atom %.

To transform the agglomerates into compact spherical par-ticles, a reshaping step at elevated temperatures was performed.The tandem DMA setup was used to measure the reshapingbehavior of the particles that had passed the compaction furnace.Particles with a preselected diameter, obtained by DMA 1, of25 and 40 nm were reshaped, and their peak particle diameter,measured by DMA 2, was plotted as a function of reshapingtemperature (Figure 5). From this graph, the compactiontemperature, TC, that is, the temperature where no moreshrinking of particle diameter is observed, was found to beapproximately 500 °C (773 K) and 550 °C (823 K) for theparticles with preselected diameters of 25 and 40 nm, respec-tively. This is in good agreement with the observations byKarlsson et al.,13 stating that the TC for metal particles shouldcorrespond to between one-third and one-half of the bulk meltingpoint (1828 K) and that smaller particles have a slightly lowerTC than larger particles. Furthermore, it was found that the finaldiameters of the particles after complete shrinkage were 15 and20 nm from agglomerates of initially 25 and 40 nm (Figure 5).

Although no further shrinkage of particle diameter occursabove the compaction temperature, the reshaping continues athigher temperatures with internal rearrangement of the particlestructure. To further investigate this, as well as morphologicaltransformations during compaction, HRTEM investigations ofparticles reshaped at 300, 600, 900, and 1150 °C wereperformed. After reshaping at 300 °C, the particles still proveto have a chainlike structure (Figure 4b) similar to that of theagglomerate particles. However, the primary particles havecoalesced together and are probably more tightly bound to eachother, making it more difficult to distinguish every primaryparticle. At 600 °C, the particles are fully compact andpolycrystalline with some amorphous regions and not yet

Figure 3. Size distributions of the agglomerate particles generated at carrier flow rates of (a) 3.4 and (b) 5.9 L/min, for spark discharge frequenciesbetween 30 and 300 Hz, measured by DMA 1.

Generation of Pd Catalyst NPs by Spark Discharge J. Phys. Chem. C, Vol. 114, No. 20, 2010 9259

completely spherical (Figure 4c). Particles reshaped at 900 and1150 °C (Figure 4d) exhibit very similar structures, indicatinga complete reshaping already below 900 °C. At these temper-atures, the particles are spherical with a single-crystalline coresurrounded by an amorphous shell. It is not clear whether theshell is actually formed during particle generation or duringtransport, in air, to the TEM. Although palladium most oftenhas the fcc crystal structure, palladium nanoparticles with anamorphous structure have been reported.20,21 A possible explana-tion could be the formation of a carbonaceous palladium species,which would also be consistent with the XPS and XRDmeasurements presented below. A similar amorphous carbon-aceous shell has also been observed in the case of Pt nanopar-

ticles.22 In addition, amorphous regions are also found in theagglomerate particles, further indicating that the amorphous shellis formed during particle generation in the polymer chamberrather than during transport in air. It has been observed that thepolymer chamber of the spark generator can lead to contamina-tion of particles.23 Measurements of lattice fringes and itsassociated fast Fourier transform (FFT) from HRTEM imagesof the crystalline core part confirm that the particles are indeedfcc structured palladium surrounded by an amorphous shell.

3.2. Size Distribution Measurements of the Pd Particles.For standard investigations of catalytic properties, typically byXRD and XPS, a rather high surface coverage of particles ofabout 5-15% is required. For particles with diameters of 15and 35 nm, a 10% surface coverage corresponds to 566 and104 particles/µm2, respectively. At the optimized conditions (aspark discharge frequency of 300 Hz, carrier flow rate of 3.4L/min, and reshaping temperature of 800 °C), that is, the highestyield of particles, deposition times of roughly 6 and 22 h wereused to obtain a 10% surface coverage for 15 and 35 nm sizedparticles, respectively. Even during such extended depositiondurations, the particle concentration and diameter remainedalmost constant, indicating the extremely controllable, constant,and simple production of particles by spark discharge. The uppergeneration limit, with respect to particle diameter, is roughly40 nm due to the narrow size distribution of agglomerateparticles. Because the agglomerate particle concentration almostdrops to zero above 80 nm, large enough agglomerates toproduce larger reshaped particles do not exist with the equipmentused. One advantage of this narrow agglomerate distributionincludes almost complete avoidance of doubly and triply chargedparticles that attenuate the control of the size distribution,24 whendepositing particles larger than around 25 nm. Although the

Figure 4. TEM micrographs of (a) agglomerate palladium particles and palladium particles reshaped at (b) 300, (c) 600, and (d) 1150 °C.

Figure 5. Evolution of particle mobility diameter plotted againstreshaping temperature of the compaction furnace. Particles withdiameters of 40 and 25 nm were preselected and then reshaped between25 and 800 °C.


majority of the particles carry one single charge after passingthe neutralizer, a small fraction of the particles might carry two(doubly charged) or three (triply charged) charges. The mobilitydiameter of a particle with additional charges will appeardifferent (smaller) than the mobility diameter of an equally sizedparticle that carries a single charge upon size selection by theDMA and hence attenuate the control of the size distribution.

The lower generation limit of particles is around 5 nm becauseof the difficulties to charge very small particles24 that hencecannot be size-selected by the DMA nor deposited by the ESP.It should be noted, however, that the deposition times neededto achieve a reasonable surface coverage of particles around orslightly above 5 nm might be extremely extended. Becausenanoparticles with a size below approximately 5 nm are knownto have properties considerably different from those of largerparticles, the lower diameter limit is a drawback of the currentsetup. However, with a slightly different design of the system,this problem may be overcome.

Figure 6a,b shows a typical SEM image after deposition of15 (35) nm particles with a concentration of 93 (503) particles/µm2 on a Si(111) wafer with a thin conducting SiOx film. Thecontrollability of particle concentration and diameter, as wellas homogeneity, was determined by recording several SEMimages acquired at different positions on the sample. By doingimage analysis of the SEM images, the individual particle areascould be determined, and the resulting distribution is shown inFigure 6c,d. The fitted Gaussian in Figure 6c,d corresponds tothe area distribution of single 15 (35) nm particles. The seconddistribution at a higher particle area is for the 15 nm particles,a result of a combination of a small fraction of doubly chargedparticles and the random deposition of particles on the substrate.For the 35 nm particles, the second peak is solely a result ofthe random deposition of particles on the substrate. A doublycharged particle would have to be 51 nm in order to have thesame mobility diameter as that of the 35 nm particles, andbecause the maximum particle size is around 40 nm with the

equipment used, such large particles are not produced. At theserelatively high particle coverages, there is, however, a probabilitythat two (or more) particles touch each other, effectivelydoubling (or tripling and so on) the observed particle area.Because of the finite resolution of the SEM images and variationin contrast between particles and substrate, it is difficult todetermine the precise values for the area of the individualparticles. Depending on the threshold value determining whetheror not a pixel in the SEM image belongs to the particle, or thesubstrate, the area distribution may be shifted slightly. Takingthis into account and calculating the particle diameters fromthe area (assuming a spherical shape), we find a particle diameterof 15 ( 1 nm (35 ( 3 nm). The image analysis confirms theexpected narrow size distribution.25 Regardless of particle size,the diameter deviation is less than 5%, providing particles witha very small deviation from the desired particle size. In addition,the observed particle coverage is within 10% of the expectedvalue, resulting in a good control of particle density as well.

3.3. X-ray Photoelectron Spectroscopy Measurements ofthe Pd Particles. Turning to our XPS studies, we show typicalsurvey scans from the 35 and 15 nm Pd particles on SiOx

samples as well as from a SiOx substrate without any depositedPd particles in Figure 7a. Here, it should be noted that the scanfrom the SiOx sample without Pd particles is recorded using aphoton energy of 735 eV, whereas the scans from the Pd samplescorrespond to a photon energy of 525 eV, changing the probingdepth. Nevertheless, it can be seen that the thickness of the SiOx

layer varies between the samples and that the main contaminantin all cases is carbon. The XPS study also shows that the SiOx

substrate is contaminated by carbon, which most likely originatesfrom hydrocarbons in the air.26

XPS spectra corresponding to the binding energy region ofC 1s is shown in Figure 7b. The spectra from the sample withoutPd particles show a main component at around 285.3 eV, whichwould correspond to hydrocarbons.27 The spectra from the 15and 35 nm Pd samples show a similar emission in the C 1s

Figure 6. (a, b) SEM images of samples with 15 and 35 nm Pd particles on SiOx. (c, d) The distribution of the individual particle area obtainedfrom several SEM images (177 and 962 nm2 correspond to the area of circles with diameters of 15 and 35 nm, respectively).


region, although a small shift toward lower binding energiescan be detected, which could possibly indicate the formationof a graphite species.27 Because 90% of the surface correspondsto SiOx, and because it is well-known that the main contaminantof Si wafers exposed to air is carbon from hydrocarbons,26 weassume that the majority of the carbon contamination is locatedat the SiOx substrate.

In Figure 7c, we show the core-level spectra from the Pd3d5/2 core-level binding energy region, the bottom spectra beinga reference spectra from single-crystal Pd(100). This spectrumcan be decomposed by the use of a single component, althougha surface component would be expected.28,29 The reasons forthe absence of a surface component could be possible hydrogencontamination, which would shift the surface componentunderneath the bulk component. Nevertheless, the single-crystalspectrum is useful to compare to the Pd 3d5/2 spectra from thePd nanoparticle samples.

It is immediately clear from Figure 7c that the peakcorresponding to emission from the Pd bulk at a binding energyof 335 eV is surprisingly weak. There is, however, a significantcomponent with a core-level shift (CLS) of 0.52 eV towardhigher binding energy, which could originate from Pd atomscoordinated to one or two chemisorbed oxygen on the surface

of the particles. We consider this unlikely with respect to thestrength of the peak and the high kinetic energy of thephotoelectrons, making the measurement bulk-sensitive. It ismuch more likely that this component originates from Pd atomscoordinated to carbon either on the surface of the Pd particlesor contained in the amorphous shell surrounding the crystallineparticle, as observed by TEM.

Further, by comparing the present measurements to XPSresults from thin oxide films and bulk oxides on Pd singlecrystals, it is immediately clear that the particles are notoxidized. Characteristic for thin surface oxides are Pd atomscoordinated to four oxygen atoms, resulting in a CLS of roughly1.3 eV.28-31 The XPS results also display a weak and broadcomponent on the high binding energy side centered at a bindingenergy of 1.3 eV. We attribute this peak to a small amount ofPd atoms coordinated to oxygen atoms at the SiOx interface ormixed within the SiOx layer. A full study of the cleaning, theoxidation, and the reduction of the Pd particles will be presentedin a different report.

3.4. X-ray Diffraction Measurements of the Pd Particles.Pd/SiOx and Pd/Al2O3 samples were also characterized by XRDmeasurements. Because of the Pd particles being randomlydistributed on the substrate, powder diffraction is used.32 Theresulting 2θ scans are shown in Figure 8 for 35 nm Pd on SiOx

and on Al2O3. The 2θ scan from the pristine 35 nm Pd on SiOx

is shown at the bottom of Figure 8a. Obviously, the diffractionfrom the Pd(111) planes and the (200) planes in the particlescan easily be detected, confirming their fcc crystalline nature.A peak in the Pd/SiOx 2θ scans at approximately 19° is assignedto powder diffraction from the Ta clips clamping the Si wafer.Comparing to tabulated XRD values for Pd, we find anexpansion of the Pd lattice of approximately 2%, most likelydue to the presence of carbon in the Pd particles. This isconfirmed by exposing the Pd particles to oxygen and therebyremoving the carbon. A 2θ after oxygen is shown in the top ofFigure 8, in which a component corresponding to PdO can alsobe seen. A zoom-in on the (111) diffraction peak is shown inFigure 8b, in which a shift of 0.4 degrees toward higher 2θangles can be seen, corresponding to a contraction of the Pdlattice after oxygen exposure. The strength of the technique inthat a nonconducting substrate, such as Al2O3, can be used isillustrated in Figure 8c. In this case, the scan was made out-of-plane to avoid the X-ray beam hitting the Ta clips. Further-more, the diffraction peaks are observed to be broader in thePd/SiOx than from the Pd/Al2O3, which is explained by thedifference in acceptance angle of the detector when performingin-plane or out-of plane scans. However, also in the Pd/Al2O3

case, an expansion of the Pd lattice in the pristine samples dueto carbon is observed.

Finally, the absence of a diffraction peak at 2θ ) 14.9° showsthat there is no presence of the PdO bulk oxide in the pristinesamples, in agreement with the XPS observations.

4. Conclusions

To understand the complex behavior of catalysts, differenttypes of model systems are often investigated. In this article,we have presented a new route to a model system consisting ofnanometer-sized palladium particles produced by an aerosolmethod. From SEM, TEM, XPS, and XRD measurements, theas-produced palladium particles were found to consist of acrystalline core surrounded by a carbonaceous amorphous shell.The carbon is believed to be incorporated during particleproduction and can be removed by oxidation by O2 of theparticles at elevated temperatures. The present study shows that

Figure 7. XPS investigations of Pd/SiOx samples. (a) Survey scansfrom 400 eV from 15 and 35 nm Pd/SiOx particles as well as from aSiOx without Pd. (b) C 1s spectra from 15 and 35 nm Pd/SiOx particlesand from a SiOx without Pd. (c) Pd 3d5/2 spectra of 15 and 35 nmPd/SiOx particles from a single-crystal Pd(100) surface used as areference.


aerosol deposition is a useful method for deposition of nano-particles on virtually any substrate to be used as a model systemfor studies of catalytic properties. The described drawbacks ofthe method, such as contamination from the polymer chamberand particle size limitations, are setup-related and may easilybe overcome by a better design. By this method, a well-definedsize of the particles can be chosen and a suitable coverage canbe selected, determined by the technique to be used ininvestigations of the catalytic properties.

Acknowledgment. This work was performed within theNanometer Structure Consortium at Lund University andsupported by the Swedish Research Council, the CrawfordFoundation, the Knut and Alice Wallenberg Foundation, theSwedish Energy Agency, and the Anna and Edwin BergerFoundation. The authors gratefully acknowledge the support ofthe ESRF and ALS staff.

References and Notes

(1) lmbihl, R.; Ertl, G. Chem. ReV. 1995, 95, 697–733.(2) Schalow, T.; Brant, B.; Laurin, M.; Schauermann, S.; Libuda, J.;

Freund, H.-J. J. Catal. 2006, 242, 58–70.(3) Schalow, T.; Brant, B.; Laurin, M.; Schauermann, S.; Guimond,

S.; Kuhlenbeck, H.; Libuda, J.; Freund, H.-J. Surf. Sci. 2006, 600, 2528–2542.

(4) Tao, F.; Grass, M. E.; Zhang, Y.; Butcher, D. R.; Renzas, J. R.;Liu, Z.; Chung, J. Y.; Mun, B. S.; Salmeron, M.; Somorjai, G. A. Science2008, 322, 932–934.

(5) Nolte, P.; Stierle, A.; Kasper, N.; Jin-Phillipp, N. Y.; Reichert, H.;Ruhm, A.; Okasinski, J.; Dosch, H.; Schoder, S. Phys. ReV. B 2008, 77,115444.

(6) Nolte, P.; Stierle, A.; Jin-Phillipp, N. Y.; Kasper, N.; Schulli, T. U.;Dosch, H. Science 2008, 321, 1654–1658.

(7) Helsper, C.; Molter, W.; Loffler, F.; Wadenpohl, C.; Kaufmann,S.; Wenninger, G. Atmos. EnViron. 1993, 27A, 1271–1275.

(8) Reinmann, R.; Akram, M. J. Phys. D: Appl. Phys. 1997, 30, 1125–1134.

(9) Schwyn, S.; Garwin, E.; Schmidt-Ott, A. J. Aerosol Sci. 1988, 19,639–642.

(10) Tabrizi, N. S.; Ullmann, M.; Vons, V. A.; Lafont, U.; Schmidt-Ott, A. J. Nanopart. Res. 2009, 11, 315–332.

(11) Wiedensohler, A. J. Aerosol Sci. 1988, 19, 387–389.

(12) Knutson, E. O.; Whitby, K. T. J. Aerosol Sci. 1975, 6, 443–451.(13) Karlsson, M. N. A.; Deppert, K.; Karlsson, L. S.; Magnusson, M. H.;

Malm, J.-O.; Srinivasan, N. S. J. Nanopart. Res. 2005, 7, 43–49.(14) Deppert, K.; Schmidt, F.; Krinke, T.; Dixkens, J.; Fissan, H. J.

Aerosol Sci. 1996, 27, S151–152.(15) van Rijn, R.; Ackermann, M. D.; Balmes, O.; Dufrane, T.; Geluk,

A.; Gonzalez, H.; Isem, H.; de Kuyper, E.; Petit, L.; Sole, V. A.; Wermeille,D.; Felici, R.; Frenken, J. W. M. ReV. Sci. Instrum. 2010, 81, 014101.

(16) Balmes, O.; van Rijn, R.; Wermeille, D.; Resta, A.; Petit, L.; Isern,H.; Dufrane, T.; Felici, R. Catal. Today 2009, 145, 220–226.

(17) Ogletree, D. F.; Bluhm, H.; Lebedev, G.; Fadley, C. S.; Hussain,Z.; Salmeron, M. ReV. Sci. Instrum. 2002, 73, 3872–3877.

(18) Evans, D. E.; Harrison, R. M.; Ayres, J. G. Aerosol Sci. Technol.2003, 37, 975–987.

(19) Messing, M. E.; Dick, K. A.; Wallenberg, L. R.; Deppert, K. GoldBull. 2009, 42, 20–26.

(20) Tehuacanero, S.; Herrera, R.; Avalos, M.; Jose Yacaman, M. ActaMetall. Mater. 1992, 40, 1663–1674.

(21) Lu, W.; Wang, B.; Wang, K.; Wang, X.; Hou, J. G. Langmuir 2003,19, 5887–5891.

(22) Paal, Z.; Schlogl, R. Surf. Sci. 2009, 603, 1793–1801.(23) Roth, C.; Ferron, G. A.; Karg, E.; Lentner, B.; Schumann, G.;

Takenaka, S.; Heyder, J. Aerosol Sci. Technol. 2004, 38, 228–235.(24) Hinds, W. C. Aerosol Technology: Properties, BehaViour, and

Measurements of Airborn Particles; Wiley and Sons: New York, 1982.(25) Karlsson, L. S.; Deppert, K.; Malm, J.-O. J. Nanopart. Res. 2006,

8, 971–980.(26) Gobel, U.; Wesemann, M.; Bensch, W.; Schlogl, R. Fresenius’ J.

Anal. Chem. 1992, 343, 582–592.(27) Moulder, J. F.; Stickle, W. F.; Sobol, P. E.; Bomben, K. D.

Handbook of X-ray Photoelectron Spectroscopy; Chastain, J., King, R. C.,Jr., Eds.; Physical Electronics, Inc.: Eden Prairie, MN, 1995.

(28) Todorova, M.; Lundgren, E.; Blum, V.; Mikkelsen, A.; Gray, S.;Gustafson, J.; Borg, M.; Rogal, J.; Reuter, K.; Andersen, J. N.; Scheffler,M. Surf. Sci. 2003, 541, 101–112.

(29) Kostelnık, P.; Seriani, N.; Kresse, G.; Mikkelsen, A.; Lundgren,E.; Blum, V.; Sikola, T.; Varga, P.; Schmid, M. Surf. Sci. 2007, 601, 1574–1581.

(30) Lundgren, E.; Kresse, G.; Klein, C.; Borg, M.; Andersen, J. N.;De Santis, M.; Gauthier, Y.; Konvicka, C.; Schmid, M.; Varga, P. Phys.ReV. Lett. 2002, 88, 246103.

(31) Westerstrom, R.; Gustafson, J.; Resta, A.; Mikkelsen, A.; Andersen,J. N.; Lundgren, E.; Seriani, N.; Mittendorfer, F.; Schmid, M.; Klikovits,J.; Varga, P.; Ackermann, M. D.; Frenken, J. W. M.; Kasper, N.; Stierle,A. Phys. ReV. B 2007, 76, 155410.

(32) Schlogl, R. AdV. Catal. 2009, 52, 273–338.

JP101390A

Figure 8. (a) 2θ scans from pristine (bottom) and oxygen-exposed 35 nm Pd/SiOx. (b) A zoom-in on the Pd(111) difraction from the pristine andoxygen-exposed 35 nm Pd/SiOx showing a shift in the 2θ angle of 0.4 degrees. (c) A 2θ scan from the pristine 35 nm Pd/Al2O3 Pd/SiOx. All XRDmeasurements indicate an expansion of the pristine Pd lattice of approximately 2%.


Generation of Pd Model Catalyst Nanoparticles by Spark Discharge

Documents