Plasma-enhanced metal-organic chemical vapor deposition (PEMOCVD) of catalytic coatings for fuel cell reformers

138 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 33, NO. 1, FEBRUARY 2005

Plasma-Enhanced Metal–Organic Chemical VaporDeposition (PEMOCVD) of Catalytic Coatings for

Fuel Cell ReformersRomit Dhar, Patrick D. Pedrow, Senior Member, IEEE, KNona C. Liddell, Quentin Ming,

Trevor M. Moeller, Member, IEEE, and Mohamed A. Osman, Senior Member, IEEE

Abstract—Fuel cells have the potential to solve several majorchallenges in the global energy economy: dependence on petro-leum imports, degradation of air quality, and greenhouse gasemissions. Using catalyst-based reformer technology, hydrogenfor fuel cells can be derived from infrastructure fuels such asgasoline, diesel, and natural gas. Platinum is one catalyst that isknown to be very effective in hydrogen reformers. Reformer sizecan be reduced when there is more efficient catalyst loading ontothe substrate. In this experimental work, platinum was loadedonto -alumina coated substrates by plasma-polymerizationfollowed by heat treatment. Vapor from a platinum-containingorganic precursor was converted to plasma and deposited onto thesubstrate. The plasma-polymerized film was then calcined to driveoff organic material, leaving behind a catalyst-loaded substrate.The plasma-polymerized organic film and the final heat-treatedcatalyst-loaded substrate surface were characterized by scanningelectron microscopy (SEM) and impedance spectroscopy. Energydispersive spectroscopy (EDS) was used to detect the presence ofthe catalyst on the substrate.

Index Terms—Catalyst loading, fuel reformer, plasma-enhancedmetal–organic chemical vapor deposition (PEMOCVD), plasma-polymerized film, platinum.

I. INTRODUCTION

I N SOME hydrogen-based energy systems now under devel-opment [1], [2], fuel cells utilize hydrogen and oxygen to

produce electricity, while hydrogen reformers [3] produce hy-drogen from infrastructure fuels such as gasoline, diesel, andnatural gas. In typical hydrogen reformers, catalyst is dispersedthroughout a porous support where it is exposed to the reac-tants. Autothermal reforming (ATR) of natural gas and otherhydrocarbon fuels to hydrogen, carbon monoxide, and carbondioxide is commonly carried out on an industrial scale at about1000 C. Fig. 1 shows the major fuel processing steps for a hy-drogen reformer that might be used with a fuel cell [4], [5]. Sev-eral researchers have done extensive experiments, which showthat both nonnoble metals (like nickel, copper, iron, cobalt) and

Manuscript received November 12, 2003; revised August 20, 2004. This workwas supported by the Washington Technology Center (WTC) and InnovaTek,Inc. under WTC Project F03-A2.

R. Dhar, P. D. Pedrow, and M. A. Osman are with the School of ElectricalEngineering and Computer Science, Washington State University, Pullman, WA99164-2752 USA (e-mail: [email protected]).

K. C. Liddell is with the Department of Chemical Engineering, WashingtonState University, Pullman, WA 99164-2710 USA.

Q. Ming is with the InnovaTek, Inc., Richland, WA 99352 USA.T. M. Moeller is with the University of Tennessee Space Institute, Tullahoma,

TN 37388 USA.Digital Object Identifier 10.1109/TPS.2004.841620

Fig. 1. Basic fuel processing steps in hydrogen based fuel cell.

noble metals (palladium, platinum) bonded to alumina or zir-conia supports form good catalysts for fuel cell reformers. Theplatinum containing catalysts are quite resistant to sulfur, a con-taminant present in fossil fuels [6], [7]. If the fuel used in the re-former is an alkane , the chemical reactions include

A reformer based on microchannel technology [8] requires acatalyst dispersed throughout a porous support, while the sup-port must adhere to the substrate (the walls of the microchan-nels). Catalyst-loaded microchannel reactor applications havebeen described in the literature [9]–[11]. Microchannel reactorshave significant advantages over packed bed reactors, includingfaster heat and mass transfer which allow for process miniatur-ization without loss of throughput [8], [12].

The three most important components in a supported catalystare: 1) substrate, 2) support, and 3) catalyst metal. The substratehosts the support while the support hosts the catalyst attached toit. The supported catalyst can take the form of individual catalystatoms or catalyst aggregates with a variety of shapes and sizes.It is known that for some applications an ensemble of catalystatoms is most effective at promoting chemical reactions [13]. Inthose cases, small agglomerates of catalyst metal are more effec-tive than atomically dispersed catalyst. Some researchers havereported [13], [14] that an optimal agglomerate physical sizemight be in the range 1–2 nm (ensemble size about 15 to 120atoms, assuming spherical agglomerates.) Aggregates smallerthan this do not fulfill the ensemble requirement while aggre-gates larger than this have a significant number of inaccessibleinterior catalyst atoms.

Mishra and Rao [15] have described the synthesis of alu-mina and zirconia ceramic supports from polymer-based com-posites. In that work, ceramic-loaded polymer tape was exposedto heat treatment for the purpose of pyrolyzing the polymer,thus yielding the desired ceramic support. They reported thatmost organic material was pyrolyzed at temperatures less than

0093-3813/$20.00 © 2005 IEEE

DHAR et al.: PLASMA ENHANCED METAL–ORGANIC CHEMICAL VAPOR DEPOSITION 139

500 C. Additional heat treatment of the resulting ceramic sup-port took place at sintering temperatures as high as 1400 C.Plasma processing techniques have been applied to fabricationof supported catalysts [16]. Applications include fabrication ofultra fine metallic catalyst agglomerates and direct loading ofmetal catalyst onto the support. Liu et al. [16] claim that ad-vantages of plasma processing applied to catalyst preparationinclude reduced number of processing steps and a concomitantreduction in toxic effluent.

Metal–organic chemical vapor deposition (MOCVD) is awell established, versatile, and widely used method for pro-ducing metal thin films. Adding steady state RF plasma tothis process yields advantages in uniformity and repeatability.This work presents the fabrication of catalyst-loaded slabs asused in wall reactors by deposition of platinum on -aluminasupport using plasma-enhanced metal–organic chemical vapordeposition (PEMOCVD) as the initial processing step. Themain focus of our work was to demonstrate that PEMOCVDcould be used to load a surface with platinum. Optimization ofthe loaded catalyst remains for future work.

II. EXPERIMENTAL

A. Plasma Process

Plasma-enhanced chemical vapor deposition (PECVD) is amethod of depositing thin films from source gases or vaporsthat can be either fed into or generated within the reactor.Coupling of a radio frequency (RF) electric field (typically13.56 MHz) into the plasma reactor results in dissociation ofthe source precursor molecules. Free electrons present in thechamber are accelerated by the applied RF electric field andcollide with precursor molecules. Impact ionization by thesefree electrons yield avalanches that continue until steady-stateplasma is established. In this way, the precursor molecules areexcited to higher energy states, primarily by inelastic collisionswith the energetic electrons, and dissociate into a variety ofradicals, ions, atoms, and more electrons [17].

Radicals and atoms, generated in the plasma, travel to the sur-face of the growing film. On arrival, they are adsorbed onto thesurface where some then diffuse on the growing film surface andmake chemical bonds at favorable sites to form an amorphousfilm. Others are desorbed depending on their respective stickingcoefficients.

B. Plasma Reactor

Fig. 2 shows a schematic of the plasma reactor with subli-mator (not to scale). The plasma reactor consisted of a 10-cminside diameter Pyrex tube evacuated to a base pressure of

torr and surrounded by a 4-turn 13.56-MHz RF coil.Applying RF fields to platinum acetylacetonate, representedin this paper as Pt(acac) , precursor vapor emanating from theheated sublimator, generated an inductively coupled plasmaplume. Plasma processing took place directly in the precursorvapor plume without added carrier gas. The axis of the RF coilwas aligned with the axis of the vapor plume emanating fromthe heated sublimator crucible. The sublimator was positionedon the floor of the reactor such that the axis of the vapor plume

Fig. 2. Schematic diagram of the plasma reactor used for the experiments.

passed through the axis of the Pyrex tube. The cylindrical cru-cible had outside diameter and height equal to 1.2 and 0.7 cm,respectively. Distance from the top of the crucible to the sub-strate was 4.5 cm. The RF coil contained two upper windingsand two lower windings. Each winding of the RF coil was madewith the insulation-shrouded center copper conductor fromRG-8 coaxial cable. Each winding had a rectangular shape withside lengths 16 and 21 cm. The 16-cm sides were parallel to theaxis of the Pyrex tube. Distance between the two upper and thetwo lower windings was 15 cm with the upper winding 7.5 cmabove and the lower winding 7.5 cm below the center line ofthe Pyrex tube. The upper and lower RF coil windings wereconnected in series with orientations such that their on-axismagnetic fields were parallel and reinforcing.

Using dimensions given above and calculation techniques de-scribed by Haus and Melcher [18], the relationship between RFcurrent, , and the no-plasma magnetic flux density at theintersection of the RF coil axis and the Pyrex tube axis is givenby

(1)

where and represent units of Tesla and amperes, respec-tively, and the magnetic field at this location is aligned withthe axis of the RF coil. The no-plasma current in this workwas 1 A (peak) giving a no-plasma magnetic flux density at thecenter of the processing volume of 12.7 T (peak). RF currentwas measured with a Model 2877 current monitor manufacturedby Pearson Electronics, Inc. [19] and a Model 9350AL dig-ital oscilloscope manufactured by LeCroy [20]. The no-plasmavoltage at the RF coil was 410 V (peak). Power absorbed duringplasma processing was 300 W.

C. Sublimator Characteristics

In the temperature range 160–170 C, the precursor mole-cule Pt(acac) is known to undergo sublimation without thermalfragmentation of its molecules [21]. Fig. 3 shows the struc-ture and chemical composition of the precursor molecule. ThePt(acac) was used as received from the supplier [22]. In this

140 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 33, NO. 1, FEBRUARY 2005

Fig. 3. Structure of the Pt(acac) precursor.

Fig. 4. Steady-state sublimator crucible temperature versus heater current.

Fig. 5. Transient crucible temperature demonstrating thermal time constant.Current of 1 A was applied at time t = 0. Thermal time constant was 6.4 min.

work, we used a sublimator, as shown in Fig. 2, to providePt(acac) vapor in which to conduct plasma-assisted materialsprocessing. The presublimation air pressure in the plasma re-actor was about torr, as provided by an oil diffusionpump.

Fig. 4 shows steady-state sublimator crucible temperature(as measured by a thermocouple at the floor of the crucible)versus dc current through the sublimator heater. Heater currentused during plasma-polymerization runs were in the range1.25–1.75 A, yielding temperatures at the floor of the subli-mator in the range 120 C–180 C.

Fig. 5 shows that the thermal time constant for the sublimatorwas about 6.4 min. Deposited organic film thickness profilessuggest the presence of a plume of Pt(acac) vapor emergingfrom the crucible with the vapor flux highly peaked on the axisof the crucible.

D. Preparation of Catalyst Support

The method used for wash coating the -alumina catalystsupport onto the aluminum substrates consists of the followingthree stages: 1) substrate pretreatment, 2) primer coating, and 3)gamma alumina coating.

1) Substrate Pretreatment: Aluminum slabs of 99.9% pu-rity with dimensions 9 101 1 mm were cleaned with ace-tone and deionized water. The aluminum slabs were exposed to

a fine grain silica sandblaster (30/40 mesh) for improving super-ficial roughness. After another cycle of cleaning with acetoneand deionized water, the slabs were then put into a furnace forhigh-temperature oxidation (600 C for 17 h), which enabledthe growth of a thin layer of alumina [23], [24].

2) Primer Coating: The use of primers to improve theadherence of the wash-coated support layer is described inthe literature [25]. In our case, since we wanted alumina to bethe only material deposited onto the substrates, we adopted aboehmite primer prepared by dispersing 10% (w/w) of alu-minum hydroxide powder in aqueous nitric acid solution of pH1.7. After continuous stirring for 4-h the pH of the suspensionincreased from 2.99 to 4.02, and a stable dispersion was ob-tained. The solution was further stirred for 15 h, and the pHincreased to 4.6. During primer deposition, the oxidized alu-minum substrates were dipped into the solution and then driedat room temperature. The slabs were then put into a furnace at600 C for one hour, which resulted in good adhesion of theprimer to the aluminum slab.

3) -Alumina Coating: A commercial -Al O sub-micrometer powder was dispersed in HNO aqueous so-lution with the following proportions: acid concentration,

g/l, and 299.7 g of Al O per literof the acid solution. The solution was stirred vigorously forabout 36 h, during which the pH changed from 1.00 to 1.74.The primer-coated slabs were immersed in the gamma aluminadispersion for 15 min and then dried at room temperature. Theywere then heated in the furnace at 280 C for 10–15 min.

E. Plasma Deposition

A crucible filled with Pt(acac) powder along with the subli-mator heater assembly was centered below the alumina-coatedaluminum slabs that were attached to a Plexiglas cowling.The aluminum slabs and sublimator were then positionedwithin the reactor tube and centered on the axis of the RFcoil. With ambient vacuum conditions ( torr) insidethe reactor tube, the sublimator heater was energized. The RFcoil produced plasma inside the reactor, without the additionof background gas, and the alumina-coated substrates wereprocessed in plasma generated from the Pt(acac) precursorvapors. The plasma spontaneously extinguished after the pre-cursor powder in the crucible was entirely consumed. Thisresulted in deposition of organic film on the substrates as shownin Fig. 6. Fig. 6(a) shows the scanning electron microscope(SEM) image of the texture of the organic film deposited on acover slip at 7000X magnification. Fig. 6(b) shows the crosssection of the organic film deposited on the cover slip. The filmthickness in this case was approximately 25 m. A quantityof 0.2 g of precursor powder required 21 min to completelysublimate at 140 C. Six crucible loadings were required for a25- m-thick plasma-polymerized Pt(acac) film.

F. Heat Treatment After Plasma Polymerization

After the plasma deposition, organic coated samples werethen heat treated in air in a calcination furnace to yield the de-sired catalyst loading while eliminating carbon and hydrogen.The temperature of the furnace was raised to 500 C at a rateof about 5 C/min and kept there for 4 h. The furnace was then

DHAR et al.: PLASMA ENHANCED METAL–ORGANIC CHEMICAL VAPOR DEPOSITION 141

Fig. 6. (a) SEM picture of plasma-polymerized organic film. (b) SEM picture of cross-section showing the organic film deposited on a cover slip.

cooled to ambient room temperature at a cooling rate of about3 C/min.

III. ANALYSIS

A. Profilometry

Profilometry studies were done on organic films plasma-de-posited onto microscope cover slips in the plane of the alu-minum slab substrates in the plasma reactor. The profilometer,manufactured by SPN Technology Inc., was a contact surfaceprofilometer and can profile only lines (rather than areas). Thestylus tip was spherical with a tip radius of 12.5 m. With theedge of a thin ceramic slab, fine trenches were made through theorganic film, which was on the microscope cover slip. Care wastaken not to scratch through the glass of the cover slip. One ofthe cover slips was located in the plane of the substrates near theaxis of the plasma plume while the other was located in the planeof the substrates about 6 cm from the axis of the plasma plume.Figs. 7 and 8 show profilometer scans for these two locationswhere the nominal film thicknesses were 20 and 40 m, respec-tively. Figs. 7 and 8 demonstrate that the organic film was notuniform along the aluminum slab substrates because the plasmaplume was densest on the plume axis.

As shown in the Appendix, the idealized relation betweenpostcalcination areal platinum loading, , and precalcinationPt(acac) organic film thickness, , is given by

(2)

where and are idealized areal platinum loading and organicfilm thickness, respectively. is expressed with units g/cmwhen carries units of cm. Using (2), the idealized postcalcina-tion loading that would result from these two organic film thick-nesses, 20 and 40 m, are 2.336 and 4.672 mg/cm , respectively.Loss of volatile species during plasma-polymerization and lossof platinum during calcination will make the actual areal plat-inum loading vary from these idealized values.

Fig. 7. Profilometer scan on the cover slip with “thin” organic film of 20 �m.

Fig. 8. Profilometer scan on the cover slip with “thick” organic film of 40 �m.

B. Surface Morphology

Surface morphology was studied using a SEM. Typicalresults are shown in Fig. 9. Microstructures of the aluminasupport structure are shown in Figs. 9(a) and (b). Presenceof spheres with diameters about 3 m result from the gammaalumina powder suspension adhered to the primer coatedslabs. Fig. 9(c) shows a SEM micrograph of the postcalcinedplatinum loaded aluminum slabs. Applying energy disper-sive spectroscopy (EDS) analysis to this postcalcined surface

142 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 33, NO. 1, FEBRUARY 2005

Fig. 9 (a) SEM picture of the alumina support surface before film deposition. (b) SEM picture of the support at higher magnification. (c) SEM of thepostcalcination, platinum-loaded aluminum slab.

Fig. 10. Microprobe EDS results from the postcalcination platinum-loadedaluminum slabs. Spectrum shows platinum peaks detected and theircorresponding energies.

resulted in the X-ray spectrum shown in Fig. 10. The EDSspectrum clearly shows the presence of platinum.

C. Impedance Spectroscopy

Fig. 11, shows the impedance spectroscopy cell that was fab-ricated after plasma-polymerized Pt(acac) was deposited ontothe Pt/SiO substrate. Input impedance, , which is the inverseof input admittance, , was measured as a function of fre-quency. Relative permittivity (dielectric constant) and electricalconductivity for the organic film were then obtained as functionsof frequency using the reduction techniques described below.Changes in process parameters can be evaluated by their influ-ence on film relative permittivity and electrical conductivity.

Solid circles in Fig. 12 show the measured impedance spec-trum for cell geometry described by mm and

m. This spectrum was measured using anapplied voltage of 50 mV rms with an HP 4192 LF ImpedanceAnalyzer. In Fig. 12, the minus sign associated with the verticalaxis shows that, as functions of time, sinusoidal current wasleading the applied sinusoidal voltage (capacitive behavior).The open squares in Fig. 12 are from the equivalent circuitshown in Fig. 13. The and values 35.8 M and 212 pF,respectively, shown in Fig. 13 were selected to make the modelcurve pass through the point labeled “600 Hz” in Fig. 12. Note

Fig. 11. Impedance spectroscopy cell used to measure the dielectric constantand electrical conductivity of the organic film.

Fig. 12. Impedance spectrum for plasma-polymerized Pt(acac) . Solid circlesare measured values while the open squares are from the model shown in Fig.13 where parameters R and C were selected to give perfect agreement withexperiment at the point labeled “600 Hz.”

that at a frequency other than 600 Hz (for example, considerthe 1-kHz frequency), the vertical coordinate from the modelnearly matched the observed value; however, for the horizontal

DHAR et al.: PLASMA ENHANCED METAL–ORGANIC CHEMICAL VAPOR DEPOSITION 143

Fig. 13. Equivalent circuit that was used to produce the open squares in Fig. 12.

(resistive) coordinate, the model consistently predicted valuestoo small when compared to measurements.

Equations describing the model points in Fig. 12 are foundby considering the effective series impedance of the paralleland circuit elements in Fig. 13

(3)

(4)

Fig. 12 shows that the model in Fig. 13 is marginal at predictingthe observed impedance spectrum. A natural extension to thisequivalent circuit is to allow the resistor and capacitor in themodel to be functions of frequency, and , with thevalues at each frequency given by

(5)

(6)

where the admittance is the inverse of the measuredimpedance . Plots of and are shown inFigs. 14 and 15, respectively, where frequency in units of Hz isused for the horizontal axis: .

Values for ranged from 38 k to 35.8 M , and valuesfor ranged from 200 to 216 pF. Comparing with the circuitmodel in Fig. 13, it is seen that the frequency independent natureof is reasonable; however, must be modeled as having astrong dependence on frequency. Assuming that the capacitorin the model can be represented by

(7)

and using a planar model for the capacitor, we can calculate therelative permittivity of the organic film from the equation

(8)

This gives where the uncertainty in wasderived using well-known techniques [26], and using the un-certainties cited above for , , and . This value of isconsistent with the range 2.5–5.23 given by Yasuda for typicalplasma-polymerized materials [27]. The planar resistor model

Fig. 14. ResistanceR from (5) shown, as a function of frequency f .

Fig. 15. Capacitance C from (6) shown, as a function of frequency f .

Fig. 16. Conductivity as a function of frequency. Solid circles are plotted from(9) with resistor values from Fig. 14. Solid straight line is a least squares fit tothe data points and its form is given in (11).

can be used to convert the resistance data in Fig. 14 to conduc-tivity,

(9)

Conductivity computed from this equation is shown as a func-tion of frequency by the solid circles in Fig. 16. In this frequencyrange, conductivity was within the range toS/m, corresponding numerically to the dc conductivities of insu-lators such as some types of marble and mica, respectively [28].

In this model, conductivity increases markedly with fre-quency. This frequency dependence of conductivity is consistentwith results reported by others [29], [30]. They measured theconductivity of plasma-polymerized and conventionally-poly-merized films and reported that the frequency dependence ofconductivity was consistent with charge transport via hopping(between localized states). They also give a functional depen-dence

(10)

144 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 33, NO. 1, FEBRUARY 2005

Fig. 17. Measured impedance spectrum (solid circles) compared to the RCmodel (open squares) that includes a frequency-dependent resistor that modelscharge hopping between localized states.

where , and are constants while . Thestraight line in Fig. 16 is a least squares linear fit to the datapoints (log–log format) with the equation for the straight linegiven by

(11)

These values of and are consistent with values found inthe literature [29], [30]. Using (9) and (11), the resistor in Fig.13 can now be modeled as frequency-dependent

(12)

The refined model for interpreting the impedance spectrum nowconsists of substituting (7) and (12) into (3) and (4). Resultsare shown in Fig. 17. Similar to Fig. 12, the vertical coordi-nates from the model nearly matches the observed values butan improvement appears because for the horizontal coordinatesin Fig. 17, the model more nearly predicts the observed values.Agreement between model and measurement is still not perfectand another improvement to the hopping model [29] would beto consider , and functions of frequency in (10). Afterthat modification, closer agreement between model and experi-ment (if needed) could be obtained by adding additional circuitelements to account for nonhopping phenomena such as inter-face effects [31]–[33]; however, the refinements listed here arebeyond the scope of the present work.

IV. CONCLUSION

Organic film deposited with PEMOCVD was calcined andyielded platinum adhering to the alumina substrate. PowderedPt(acac) was found to be a suitable platinum-containing pre-cursor organic molecule with sublimation temperatures in therange 120–180 C. Applied RF fields in a 13.56-MHz induc-tively coupled plasma reactor elevated the precursor vapor to aplasma state suitable for deposition of the plasma-polymerizedorganic film. Plasma processing was successful without addi-tion of a carrier gas. The peak calcination temperature of 500 Cyielded platinum dispersed on the alumina surface. Energy dis-persive spectroscopy (EDS) confirmed the presence of platinumon the processed alumina surface. Impedance spectroscopy ap-plied to the precalcination organic film showed that relative elec-tric permittivity was 2.49, and that conduction in the organicfilm involved charge carrier migration via a hopping mechanismwith conductivity in the range to S/m as

Fig. 18. Idealized model that relates organic film thickness to catalyst arealsurface loading. (a) Precalcination conditions. (b) Postcalcination conditions.

frequency varied from 600 Hz to 1 MHz. These electrical prop-erties were consistent with other plasma-polymerized organicfilms described in the literature.

APPENDIX

An idealized relationship between organic film thickness andcatalyst areal loading in mass per unit area can be obtained byconsidering Figs. 18(a) and (b). If we assume that the plasma-polymerized organic film consists of an assembly of precursormolecules (ignoring volatile chemical species removed from theplasma state by the vacuum pump) then the mass of catalystassociated with Fig. 18(a) can be written as

(13)

where , and are catalystmass contained in the organic film in Fig. 18(a), area, organicfilm thickness, mass density of the organic film, molecularweight of the catalyst atom, and molecular weight of thecatalyst-containing precursor molecule, respectively. An as-sumption inherent in this model is that the elemental ratios inthe organic film are the same as in the precursor. Fig. 18(b)represents the postcalcination state of Fig. 18(a). In Fig. 18(b),the organic film has been removed but agglomerates of catalystmetal remain on the surface of the substrate and no platinumis volatilized during calcination. If we assume a mean arealcatalyst loading then the mass of catalyst associated with Fig.18(b) can be written as

(14)

where and are catalyst mass contained on the surfaceof the substrate in Fig. 18(b) and areal catalyst loading, respec-tively. Assuming that no catalyst atoms are lost during calcina-tion, we can set and obtain an idealized expressionfor as

(15)

We now specialize (15) to our work with the Pt(acac) mole-cule shown in Fig. 3

(16)

where 195, 4, 16, 10, 12, 14, and 1 represent atomic massof platinum in atomic mass units (amu), number of oxygenatoms, atomic mass of oxygen in amu, number of carbon atoms,

DHAR et al.: PLASMA ENHANCED METAL–ORGANIC CHEMICAL VAPOR DEPOSITION 145

Fig. 19. Predicted relationship between areal platinum loading, L , andthickness of the plasma-polymerized Pt(acac) organic film d. This is a plot of(17).

atomic mass of carbon in amu, number of hydrogen atoms, andatomic mass of hydrogen in amu, respectively. From Postonand Reisman [34], has a value of 2.336 g/cm for solidphase Pt(acac) . Equation (15) yields the predicted relationshipbetween areal platinum mass loading and thickness of theplasma-polymerized Pt(acac) film as

(17)

A plot of (17) is shown in Fig. 19. The range of values(0–100 m) was selected because it corresponds to values( – mg/cm ) typical in the fuel cell/reformer literature[35]. Values of used in this work are in the range 20–40

m corresponding to ideal catalyst areal loading values of2.336–4.672 mg/cm , respectively.

ACKNOWLEDGMENT

The authors thank Dr. P. Irving for her support and encourage-ment. C. Davitt, with the Electron Microscopy Center, Wash-ington State University, Pullman, assisted with SEM diagnos-tics. M. Fuller, with the Center for Materials Research, Wash-ington State University, Pullman, provided profilometry scans.

REFERENCES

[1] J. M. Ogden, M. M. Steinbugler, and T. G. Kreutz, “A comparison ofhydrogen, methanol, and gasoline as fuels for fuel cell vehicles: Im-plications for vehicle design and infrastructure development,” J. PowerSources, vol. 79, pp. 143–168, 1999.

[2] L. Barreto, A. Makihira, and K. Riahi, “The hydrogen economy in the21st century: A sustainable development scenario,” Int. J. HydrogenEnerg., vol. 28, pp. 267–284, 2003.

[3] S. Springmann, G. Friedrich, M. Himmen, M. Sommer, and G. Eigen-berger, “Isothermal kinetic measurements for hydrogen production fromhydrocarbon fuels using a novel kinetic reactor concept,” Appl. Catal-ysis A, General, vol. 235, pp. 101–111, 2002.

[4] J. M. Zalc and D. G. Loffler, “Fuel processing for PEM fuel cells: Trans-port and kinetic issues of system design,” J. Power Sources, vol. 111, pp.58–64, 2002.

[5] J. M. Zalc, V. Sokolovskii, and D. G. Loffler, “Are noble metal-basedwater—Gas shift catalysts practical for automotive fuel processing?,” J.Catalysis, vol. 206, pp. 169–171, 2002.

[6] E. Romero-Pascual, A. Larrea, A. Monzon, and R. D. Gonzalez,“Thermal stability of Pt/Al O catalysts prepared by sol-gel,” J. SolidState Chem., vol. 168, pp. 343–353, 2002.

[7] E. A. Ticianelli, J. G. Beery, and S. Srinivasan, “Dependence of perfor-mance of solid polymer electrolyte fuel cells with low platinum loadingon morphologic characteristics of the electrodes,” J. Appl. Electrochem.,vol. 21, pp. 597–605, 1991.

[8] A. Y. Tonkovich, J. L. Zilka, M. J. LaMont, Y. Wang, and R. S. Wegeng,“Microchannel reactors for fuel processing applications. I. Water gasshift reactor,” Chem. Eng. Sci., vol. 54, pp. 2947–2951, 1999.

[9] M. T. Janicke, H. Kestenbaum, U. Hagendorf, F. Schuth, M. Fichtner,and K. Schubert, “The controlled oxidation of hydrogen from an explo-sive mixture of gases using a microstructured reactor/heat exchanger andPt/Al O catalyst,” J. Catalysis, vol. 191, pp. 282–293, 2000.

[10] K. Haas-Santo, M. Fichtner, and K. Schubert, “Preparation of mi-crostructure compatible porous supports by sol-gel synthesis forcatalyst coatings,” Appl. Catalysis A, General, vol. 220, pp. 79–92,2001.

[11] P. Claus, D. Honicke, and T. Zech, “Miniaturization of screening devicesfor the combinatorial development of heterogeneous catalysts,” Catal-ysis Today, vol. 67, pp. 319–339, 2001.

[12] S. Tschudin, T. Shido, R. Prins, and A. Wokaun, “Characterizationof catalysts used in wall reactors for the catalytic dehydrogenation ofmethylcyclohexane,” J. Catalysis, vol. 181, pp. 113–123, 1999.

[13] G. Jacobs, F. Ghadiali, A. Pisanu, A. Borgna, W. E. Alvarez, and D.E. Resasco, “Characterization of the morphology of Pt clusters incorpo-rated in a KL zeolite by vapor phase and incipient wetness impregnation.Influence of Pt particle morphology on aromatization activity and deac-tivation,” Appl. Catalysis A, General, vol. 188, pp. 79–98, 1999.

[14] T. Frelink, W. Visscher, and J. van Veen, “Particle size effect of carbon-supported platinum catalysts for the electrooxidation of methanol,” J.Electroanalytical Chem., vol. 382, pp. 65–72, 1995.

[15] R. Mishra and K. J. Rao, “Thermal and morphological studies of bi-nary and ternary composites of poly(vinylalcohol) with Alumina andZirconia,” Ceramics Int., vol. 26, pp. 371–378, 2000.

[16] C. Liu, G. P. Vissokov, and B. Jang, “Catalyst preparation using plasmatechnologies,” Catalysis Today, vol. 72, pp. 173–184, 2002.

[17] M. A. Lieberman and A. J. Lichtenberg, Principles of Plasma Dis-charges and Materials Processing. New York: Wiley, 1994.

[18] H. A. Haus and J. R. Melcher, Electromagnetic Fields and En-ergy. Englewood Cliffs, NJ: Prentice-Hall, 1989, pp. 320–324.

[19] Model 2877 200 Mhz Current Monitor. Pearson Electronics, Inc., PaloAlto, CA. [Online]. Available: http://www.pearsonelectronics.com

[20] LeCroy 9350AL Digitizing 500 MHz Oscilloscope. LeCroy Corpora-tion, Chestnut Ridge, NY. [Online]. Available: http://www.lecroy.com

[21] S. Barison, M. Fabrizio, G. Carta, G. Rossetto, P. Zanella, D. Barreca,and E. Tondello, “Nanocrystalline Pt thin films obtained via metal or-ganic chemical vapor deposition on quartz and CaF substrates: An in-vestigation of their chemico-physical properties,” Thin Solid Films, vol.405, pp. 81–86, 2002.

[22] Platinum (II) Acetylacetonate, Catalog Number 78-1400.Strem Chemicals, Inc., Newburyport, MA. [Online]. Available:http://www.strem.com

[23] M. Valentini, G. Groppi, C. Cristiani, M. Levi, and E. Tronconi, “Thedeposition of -Al O layers on ceramic and metallic supports forthe preparation of structured catalysts,” Catalysis Today, vol. 69, pp.307–312, 2001.

[24] S. Zhao, J. Zhang, D. Weng, and X. Wu, “A method to form well-ad-hered -Al O layers on FeCrAl metallic supports,” Surface CoatingsTechnol., vol. 167, pp. 97–105, 2003.

[25] M. F. M. Zwinkels, Ed., Proc. 6th Int. Symp. Sci. Bases Prep. Heteroge-neous Catalysts, Amsterdam, The Netherlands: Elsevier, 1995.

[26] H. W. Coleman and W. G. Steele, Experimentation and UncertaintyAnalysis for Engineers, 2nd ed. New York: Wiley, 1999, vol. 49.

[27] H. Yasuda, “Electrical properties of plasma-polymerized thin organicfilms,” in Plasma Polymerization. New York: Academic, 1985, ch. 11,pp. 370–417.

[28] E. Baguley, Reference Data for Radio Engineers, 6th ed. New York:Howard W. Sams Co., 1975, ch. 4, pp. 23–24.

[29] F. U. Z. Chowdhury and A. H. Bhuiyan, “Dielectric properties ofplasma-polymerized diphenyl thin films,” Thin Solid Films, vol. 370,pp. 78–84, 2000.

[30] R. I. Mohamed, “AC conductivity and dielectric constant of poly(vinylalcohol) doped with MnSO ,” J. Phys. Chem. Solids, vol. 61, pp.1357–1361, 2000.

[31] M. Meier, S. Karg, and W. Riess, “Light-emitting diodes based onpoly-p-phenylene-vinylene: II. Impedance spectroscopy,” J. Appl.Phys., vol. 82, no. 4, pp. 1961–1966, 1997.

[32] G. D. Sharma, M. Roy, D. Saxena, and M. S. Roy, “Investigation ofschottky barrier of poly(phenyl azo methane thiophene) using current-voltage and impedance spectroscopy,” Mater. Sci. Eng., vol. B79, pp.146–153, 2001.

[33] J. R. MacDonald, Impedance Spectroscopy. New York: Wiley, 1987.[34] S. Poston and A. Reisman, “Density determination of silver neode-

canoate, tungsten hexacarbonyl, and a series of metal acetylacetonatesand hexafluoroacetylacetonates,” J. Electron. Mater., vol. 18, no. 1, pp.79–84, 1989.

146 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 33, NO. 1, FEBRUARY 2005

[35] P. S. Kauranen, E. Skou, and J. Munk, “Kinetics of methanol oxida-tion on carbon-supported Pt and Pt+ Ru catalysts,” J. ElectroanalyticalChem., vol. 404, pp. 1–13, 1996.

Romit Dhar recieved the B.Tech. degree in energyengineering from the Department of Electrical Engi-neering, Indian Institute of Technology, Kharagpur,India, in 2001 and the M.S. degree in electrical engi-neering from Washington State University, Pullman,in 2003 where he is currently working toward thePh.D. degree in materials science at the Center forMaterials Research, Washington State University.

His research interests include piezoelectric sensorsand devices, thin films, and plasma processing.

Patrick D. Pedrow (SM’96) received the B.S. de-gree in electrical engineering from the University ofIdaho, Moscow, in 1975, the Masters of Engineeringdegree in electric power engineering from Rensse-laer Polytechnic Institute, Troy, NY, in 1976, theM.S. degree in physics from Marquette University,Milwaukee, WI, in 1981, and the Ph.D. degreein electrical engineering from Cornell University,Ithaca, NY, in 1985.

From 1976 to 1981, he was with the Mc-Graw-Edison Company where he conducted

research and development on electric power circuit breakers. He is presentlyan Associate Professor in the School of Electrical Engineering and ComputerScience at Washington State University, Pullman. His research interests are inplasma-assisted materials processing, including plasma source developmentand the deposition and evaluation of thin plasma-polymerized films.

Dr. Pedrow a Member of the American Physical Society, a Member of TauBeta Pi, and a registered professional engineer in the state of Wisconsin. He hasserved on the Executive Committee of the IEEE Nuclear and Plasma SciencesSociety Plasma Science and Applications Committee.

KNona C. Liddell received the B.S. degree cum laude and with distinction inchemistry from the University of Washington, Seattle, in 1971, the M.S. degreein chemistry from the University of Arizona, Tucson, in 1973, and the Ph.D.degree in chemical engineering from Iowa State University, Ames, in 1979.

She is presently a Professor in the School of Chemical Engineering and Bio-engineering, Washington State University, Pullman. Her research interests are indeposition and characterization of thin films and multiphase reactor modeling.

Dr. Liddell is a Member of American Institute of Chemical Engineers, TheElectrochemical Society, and The Minerals, Metals, and Materials Society.

Quentin Ming received the M.S. degree in chemicalengineering from Dalian University of Technology,Dalian, China in 1992, and the Ph.D. degree in chem-ical engineering from the University of Houston,Houston, TX, in 1999.

From 1999 to 2000, he was with Blasch PrecisionCeramics as a Staff Scientist working on the develop-ment of advanced processing of material used in solidoxide fuel cells. In 2000, he joined InnovaTek, wherehe is currently a Senior Scientist working to developfuel reforming system for fuel cell application. His

research interests include catalytic reaction and separation, and advanced ce-ramic materials.

Trevor M. Moeller (M’92) received the B.S. degreein mechanical engineering from Rose-HulmanInstitute of Technology, Terre Haute, IN, in 1991,and the M.S. and Ph.D. degrees, all in mechanicalengineering from the University of Tennessee,Knoxville, in 1993 and 1998, respectively.

He is a Research Associate Professor at the Uni-versity of Tennessee Space Institute (UTSI). Beforejoining the UTSI faculty, he directed many of thetechnology development efforts for InnovaTek, Inc.,Richland, WA. His efforts at InnovaTek included the

research of plasma surface treatment for biological and chemical warfare agentdecontamination, microchannel combustion, and aerosol collector technology.He is currently interested in the modeling and testing of plasmas with anemphasis in electric propulsion. He has more than 18 papers in journals andconference proceedings and has received 1 patent.

Dr. Moeller is a Member of American Society of Mechanical Engineers, TauBeta Pi, and Pi Mu Epsilon. He is currently serving as Secretary of the TennesseeSection of American Institute of Aeronautics and Astronautics.

Mohamed A. Osman (S’82–M’84–SM’96) re-ceived the B.S. degree (with honors) in physics fromthe University of Khartoum, Khartoum, Sudan, in1977, the M.S. degree in physics from University ofArizona, Tucson, in 1980, the M.S. degree in elec-trical and computer engineering from the Universityof Massachusetts, Amherst, in 1982, and the Ph.D.degree in physics from Arizona State University,Tempe, in 1986.

From 1986 to 1989, he was with the Scientific Re-search Associates where he conducted research on

transport in semiconductors and device simulations. He is presently a Professorat the School of Electrical Engineering and Computer Science, Washington StateUniversity, Pullman. His research interests include transport in semiconductors,polymers, carbon nanotubes, self-assembly, and device simulations.

Dr. Osman is a Member of the American Physical Society. He serves on theprogram Committee of the SPIE Symposium on Ultrafast Phenomena in Semi-conductors and Nanomaterials, and cochair of the Electrical Track of the IEEESoutheast Conference 2005, and was on the technical program of the 2003 IEEEUGIM Symposium. He was the recipient of NASA Group Achievement Awardin 2000.