International Journal of Hydrogen Energy Volume 32 Issue 16 2007 [Doi 10.1016%2Fj.ijhydene.2006.08.016] John S. Hardy; Edwin C. Thomsen; Nathan L. Canfield; Jarrod v. C -- Development

International Journal of Hydrogen Energy 32 (2007) 36313639www.elsevier.com/locate/ijhydene

Development of passive hydrogen separationmembranesmade fromCo-synthesized nanoscale cermet powders

John S. Hardy, Edwin C. Thomsen, Nathan L. Caneld, Jarrod V. Crum,K. Scott Weil, Larry R. Pederson

Pacic Northwest National Laboratory, 902 Battelle Blvd., Richland, WA 99352, USAAvailable online 22 September 2006

Abstract

A powder comprised of nickel oxide and proton-conducting Nd- and Zr-doped barium cerate with a particle size on the order of 10 nm hasbeen co-synthesized using the glycinenitrate combustion process. The two compositions are intimately mixed with no signicant elementalsubstitution between them after synthesis. To ensure complete reaction of the cerate components, the synthesized powder must be calcinedat 1000 C. Among the barium cerate compositions investigated, the 30% Zr- and 15% Nd-doped material exhibited the best combination ofchemical stability in CO2 and conductivity in hydrogen environments. At least 35 vol% Ni is required to achieve percolation in the composites.When sintering is carried out in an atmosphere which promotes reduction of nickel oxide to nickel metal, the result is a mixed electronic- andprotonic-conducting composite which has potential use as a hydrogen separation membrane. Composites with a relative density of 99.2% andnanoscale grains have been prepared by hot pressing. 2006 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.

Keywords: Cermet; Composite; Nanoscale; Synthesis; Proton; Hydrogen; Membrane; Gas separation; Ionic conduction; Cerate; Nickel; Percolation

1. Introduction

Much research has been devoted to promoting the hydrogeneconomy envisioned for the future of the United States andother nations worldwide [1,2]. One of the important enablingtechnologies for this vision is hydrogen separation. Efcientseparation of hydrogen from mixed gas streams would makelarge amounts of hydrogen that currently go unused availablefor use as fuel for clean energy generation. To this end, muchwork has been dedicated to the development of membranes withhigh levels of permeability and selectivity to hydrogen. Mem-brane concepts currently being pursued include mesoporous ce-ramic materials [35] that act as molecular sieves allowing onlymolecules smaller than the pores to pass through; metallic mem-branes [6] which allow atomic hydrogen to pass through theircrystal lattices; and proton conducting ceramic materials [711]that conduct protons under an applied electrical potential.

Corresponding author. Fax: +1 509 375 2186.E-mail address: [email protected] (J.S. Hardy).

0360-3199/$ - see front matter 2006 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.ijhydene.2006.08.016

Another concept which has been pioneered by Balachandranet al. [12,13] is the ceramic/metal (cermet) composite hydrogenpermeable membrane. Among the cermet concepts they haveinvestigated is one that consists of a hydrogen permeable alloytogether with a ceramic that contributes to improved mechani-cal properties. Another comprises a proton conducting ceramicand an electron conducting metal. They found that the cermetcontaining a hydrogen permeable alloy exhibited the higher hy-drogen ux and that the hydrogen ux through the membranecontaining a proton conducting ceramic was limited by surfacereaction kinetics.

In order for protons to be transported through a cermet mem-brane, hydrogen molecules must rst be dissociated and chargetransfer must take place. These reactions occur at the surface ofthe membrane and are concentrated at the triple phase bound-ary between the gas phase and the ceramic and metal phasesin cermet membranes utilizing a proton conducting ceramic.By decreasing the grain sizes in such a membrane, the triplephase boundary length should be greatly increased resultingin a higher rate at which surface reactions can be completed.

3632 J.S. Hardy et al. / International Journal of Hydrogen Energy 32 (2007) 36313639

This should decrease the interfacial polarization resistance andimprove the hydrogen ux through the membrane.

This paper is focused on developing cermet membranes madefrom nanoscale precursor powders for which both the bariumcerate-based proton conducting ceramic and the oxide of nickelmetal are co-synthesized in a single combustion reaction. Sucha process eliminates the steps of synthesizing the two com-positions separately and attempting to combine them into ahomogeneous mixture. Additionally, starting with a nanoscalematerial improves the potential for sintering a hermetic mem-brane with ner grain sizes and therefore improved surface ki-netics and transport and mechanical properties. For example, inwork with cerium oxide, Zhou et al. [14,15] have shown that bydecreasing the average grain size from a few microns down to20 nm, the conductivity increases by an order of magnitudedue to a decrease in the enthalpy of oxygen vacancy formation.

A number of considerations will be addressed including theoptimum composition to be used for the membranes. Bar-ium cerate-based ceramics are among the most well-knownand the best high-temperature proton conductors, but they areknown to be chemically unstable in atmospheres containing car-bon dioxide at elevated temperatures. However, by substitutingzirconium for some of the cerium in the formulation, the sta-bility improves but unfortunately this improvement is accom-panied by a decline in conductivity [16,17]. Therefore, theminimum concentration of zirconium necessary to instill chem-ical stability must be determined. Furthermore, the conductivityof the cerate can be improved by doping with a limited con-centration of Nd. Steps will be taken to evaluate the effects ofNd-doping on chemical stability and conductivity. Since theelectronic conductivity of nickel is orders of magnitude higherthan the protonic conductivity of barium cerate, the concentra-tion of barium cerate in the composite will be maximized bylimiting the concentration of nickel in the composite to onlyslightly above its percolation threshold. Processing details willalso be examined.

2. Experimental

Nd-doped barium zirconate cerate and nickel oxide wereco-synthesized using the glycinenitrate combustion synthesistechnique [18] to form a two-phase cermet powder. This wasdone by mixing calculated ratios of the appropriate aqueousmetal nitrate salt solutions with enough glycine to produce astoichiometric combustion reaction. The batches were calcu-lated such that after synthesis, subsequent calcination, and NiOreduction the resulting powder would have the general compo-sition:

x vol% Ni : (100 x) vol% BaCe1mnZrmNdnO3.However, for the sake of simplication, the compositions arehereafter referred to as

XXNi-BCZNYYZZ,

where XX is a two-digit number representing the vol% Ni inthe composite, YY and ZZ are two-digit numbers representing

the percent of cerium for which zirconium and neodymium,respectively, are substituted in the barium cerate. For exam-ple, 25Ni-BCZN2010 will represent 25 vol% Ni : 75 vol%BaCe0.7Zr0.2Nd0.1O3.

The composite powders were imaged directly after co-synthesis using transmission electron microscopy (TEM) toevaluate the compositional distribution and primary particlesizes of the as-synthesized powder. Axsia software developedby Sandia National Laboratory was utilized to simultaneouslyanalyze and compare the energy dispersive X-ray (EDX) spec-tra associated with all of the pixels in a given image to mathe-matically extract the primary spectra, which can be combinedin different ratios to account for every individual compositespectrum collected in the image. A color-coded map was thengenerated for each extracted primary spectrum showing itscontribution to the composite spectrum for each pixel in theimage. Since each of the primary spectra is attributable toa chemical composition present in the sample, the result iseffectively a compositional map.

Differential thermal analysis and thermogravimetric analy-sis (DTA/TGA) were performed on the as-synthesized powder.The powder was then divided into small samples which wereeach calcined in a static air mufe furnace at different temper-atures including 600, 800 and 1000 C for 1 h. Phase analysiswas performed on the calcined powders using X-ray diffraction(XRD).

Single-phase barium cerate compositions doped with variouslevels of Zr and Nd were also synthesized using the glycinenitrate method as described above without including Ni-nitratesolution in the batch. The resulting powder was calcined at1000 C and pellets were pressed uniaxially at 112MPa andfurther compacted in a cold isostatic press at 176MPa. The pel-lets were sintered in a static air mufe furnace at 1600 C for2 h and the density was measured using theArchimedes methodto ensure complete sintering to at least 95% density. The pel-lets were pulverized and XRD was run on the resulting powderto conrm phase purity and calculate theoretical density. Thepowder was then subjected to DTA/TGA in pure CO2 whileheating at 20 C/min from room temperature to 1400 C Sub-sequent cooling was also carried out in CO2. Phase analysisby XRD was repeated after DTA/TGA measurements in CO2were complete.

Cermet composite samples with the composition 25Ni-BCZN2010 were calcined at 1000 C for 1 h and pellets werepressed uniaxially and isostatically at the aforementioned pres-sures. In order to minimize grain growth, they were sinteredat 1250 C for 2 h which was the minimum sintering temper-ature required to achieve at least 90% density. Sintering ofone sample was performed in air, while another was sinteredin 3% hydrogen in argon to promote reduction of the NiOduring sintering. The sample sintered in air was reduced ina subsequent treatment at 1000 C in 3% hydrogen for 12 h.Archimedes density measurements and SEM were utilized tocompare the two samples.

Four-probe conductivity measurements were made in con-trolled atmospheres on bars from selected compositions of thesingle-phase barium cerates described above and of Ni/barium

J.S. Hardy et al. / International Journal of Hydrogen Energy 32 (2007) 36313639 3633

cerate composites. Samples were made from powders whichhad been calcined at 1000 C for 1 h in air and presseduniaxially and isostatically as described above into bars ap-proximately 4.6 cm1.6 cm0.5 cm in size. The single-phasecerate samples were sintered at 1600 C while composite sam-ples were sintered at the minimum temperature required toattain at least 90% density in an attempt to minimize graingrowth. This required sintering temperatures of between 1250and 1350 C. The sintered bars were then cut into sectionswith dimensions of approximately 3.4 cm 0.5 cm 0.4 cmand Pt leads were attached for four-probe conductivity tests.Single-phase barium cerate specimens were placed in a tubefurnace and measurements were taken as a function of tem-perature in 3% hydrogen in argon. For one measurement, the3% hydrogen was dry and for another, it was bubbled throughwater at room temperature to saturate it with 3% water va-por. Cermet composite samples with varying Ni-contents weremeasured isothermally at 1000 C in 3% hydrogen in argon.

Samples of 35Ni-BCZN3010 powder weighing 1.5 g werecalcined at 1000 C, pressed uniaxially and isostatically asabove, and sintered in 3% hydrogen in argon for 2 h at vari-ous temperatures. A 2.5 g sample was hot pressed in vacuumat 1100 C for 2 h at 48MPa. The relative density of each sam-ple was measured using the Archimedes method. A portion ofthe hot pressed sample was pulverized for XRD while anotherpiece was sectioned and polished for SEM imaging.

3. Results and discussion

The major impetus for co-synthesizing the composite pow-der using combustion methods was the potential for achievingparticle sizes on the nanoscale. One of the concerns accom-panying combustion co-synthesis of multiple compositions si-multaneously, however, is that the compositions may react toform a solid solution or possibly even an additional, unwantedphase. The TEM image in Fig. 1(a) shows a typical agglomer-ate from a sample of as-synthesized 35Ni-BCZN2010 ash. Inthis micrograph it can be seen that the agglomerate is madeup of primary particles ranging from 10 to 50 nm in size. Fur-ther analysis of the agglomerate performed using Axsia soft-ware determined that three fundamental EDX spectra could becombined in various ratios to largely account for the compositespectrum of every pixel in the image. These three spectra rep-resent three component chemical compositions that make upthe sample in the image. The rst component extracted by thesoftware was found to contain copper and carbon, which werenot present in the cermet powder, but made up the grid andmounting material used for preparing the powder specimens.The EDX spectra of the second and third extracted componentswere directly associated with the target cermet composition andare shown in Figs. 1(b) and (c), respectively. Together witheach primary spectrum is also a map in which every pixel iscolor coded to represent the fraction of that component whichcontributed to the composite EDX spectrum obtained for thatpixel. The spectrum in Fig. 1(b) corresponds to nickel oxide(NiO). The maximum intensity peak seen between 7 and 8KeVis the major peak for Ni. The spectrum in Fig. 1(c) contains

peaks corresponding to the elements (Ba, Ce, Zr, Nd, and O)which are the intended constituents of the proton conductingceramic component. However, the major peak for Ni was com-pletely absent from the spectrum of this component, indicatingthat there is no signicant elemental substitution between thetwo desired components. Furthermore, the compositional mapsof the small agglomerate in the micrograph indicate that an in-timate mixture of ne particles of the two components resultfrom combustion co-synthesis.

The weight loss event between 800 and 1000 C in the TGAscan for 25Ni-BCZN2010 in Fig. 2 is typical of co-synthesizedNi/BCZN composites and indicates that the reaction of the cer-ate components is not carried to completion during combustionsynthesis, but requires additional calcination at a temperatureof at least 1000 C for complete reaction. This is veried bythe XRD patterns in Fig. 3 for 25Ni-BCZN2010 samples thatwere calcined for 1 h at various temperatures. Because bariumcarbonate is highly favorable in CO2 containing environmentssuch as air, the reversible barium cerate formation reaction inair can be written as follows:

BaCO3(s) + CeO2(s) BaCeO3(s) + CO2(g) .The uncalcined powder still contains signicant amounts ofbarium carbonate and cerium oxide which are the two mainreactants in the formation of the cerate. The products of thereaction are the cerate and carbon dioxide, which is releasedas a gas and is responsible for the weight loss observed inTGA. XRD of samples calcined at 600 and 800 C conrm thatthe cerate formation reactions do not make signicant progresstoward completion at calcination temperatures below 1000 C.However, after 1000 C calcination, cerate formation is virtu-ally complete. Fig. 4 shows the XRD patterns for compositescontaining BCZN2010 with various target Ni concentrations af-ter calcination at 1000 C for 1 h. It can be seen that Ni/BCZNcomposites with a wide range of target Ni concentrations canbe made by combustion co-synthesis followed by calcinationat 1000 C. The changing NiO concentration is reected in thechange in intensity of the major NiO peaks relative to the majorcerate peaks.

After calcination, the NiO must be reduced to electroni-cally conductive Ni metal. This chemical reduction is accompa-nied by a volume reduction and, therefore, the microstructuraleffects of NiO reduction were evaluated through SEM analy-sis and density measurements. The relative density of 25Ni-BCZN2010 when measured after sintering at 1250 C for 2 h inair was 91.2%. After subsequent reduction at 1000 C for 12 hin 3% hydrogen in argon, the relative density had decreasedto 77.8%. The drastic decrease in density can be attributed tothe widespread intergranular cracks seen in the SEM micro-graph of a polished cross-section in Fig. 5. The stresses result-ing from the volume reduction are evidently detrimental to asintered composite. However, sintering at 1250 C for 2 h ina reducing 3% hydrogen atmosphere resulted in a compositecontaining Ni metal that had a relative density of 90.8%, whichis nearly the same as the density of the sample sintered in airat the same temperature before it was reduced. By sinteringin the reducing gas, the stresses that caused cracking in the

3634 J.S. Hardy et al. / International Journal of Hydrogen Energy 32 (2007) 36313639

Fig. 1. TEM analysis of a typical agglomerate resulting from glycinenitrate co-synthesis of 35Ni-BCZN2010 including (a) a micrograph, (b) the EDXspectrum and compositional map for the rst component of the cermet composite extracted by Axsia software, and (c) the EDX spectrum and compositionalmap for the second component of the cermet composite extracted by Axsia software.

sample reduced after air sintering are avoided. This is becausethe volumetric shrinkage associated with reduction takes placewhile the furnace is ramping up to the nal sintering temper-ature. According to the Ellingham diagram, NiO reduction isprobably complete before 600 C in a 3% hydrogen in argonatmosphere. Therefore, when densication of the compact intoits nal microstructure occurs, the Ni is already in its desired,reduced state.

Once it had been established that combustion co-synthesiswas a viable method for making nanoscale cermet powders thatcould be formed into dense Ni/BCZN composites, the next stepwas to establish a desirable composition for the cermet. TGAof pulverized samples of single-phase 10% Nd-doped bariumzirconate cerate in owing CO2 showed increasing stabilitywith increasing Zr-content until no signicant barium carbonateformation was seen at 30% Zr-substitution. The mol% barium

J.S. Hardy et al. / International Journal of Hydrogen Energy 32 (2007) 36313639 3635

100

98

96

94

Wei

ght (%

)

0 200 400 600 800 1000 1200 1400 1600Temperature (C)

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

Tem

pera

ture

Diff

eren

ce (

C)

Fig. 2. DTA/TGA of as-synthesized 25Ni-BCZN2010 showing the weight lossevent that takes place between 800 and 1000 C in as-synthesized powders.

Fig. 3. XRD patterns of combustion co-synthesized 25Ni-BCZN2010 aftercalcination at various temperatures for 1 h in air.

Fig. 4. XRD patterns of combustion co-synthesized powders containing various concentrations of Ni after calcination at 1000 C for 1 h.

that degraded from cerate to carbonate for each compositionis shown in the chart in Fig. 6(a). TGA of samples with 5%and 15% Nd-doping on the Ce-site also showed that stabilitywas achieved with 30% Zr-doping. This, however, was not thecase for the composition containing 20% Nd. Fig. 6(b) showsTGA traces of samples containing 10% Nd with varying Zr-content. The decrease in the weight gain associated with bar-ium carbonate formation is evident with increasing Zr-doping.Fig. 6(c) shows the results of XRD analysis on samples con-taining 10% Nd and varying levels of Zr after TGA in CO2.The samples with 30% and 35% Zr were both found to containonly the cerate material. However, as the level of Zr-doping de-creases, foreign peaks become increasingly prominent. Theseadditional peaks arise from the increasing presence of bariumcerate and cerium oxide. Fig. 6(d) is a TGA plot showing sam-ples containing 30% Zr with various levels of Nd. BCZN1510and BCZN2010 are also included to provide a frame of refer-ence for comparing the stability of these samples with thosefrom the plot in Fig. 6(b). While BCZN3005, BCZN3010, andBCZN3015 show no signicant carbonate formation, 30% Zr-doping is no longer sufcient to provide stability in CO2 forBCZN3020. Fig. 7 is a plot of four-probe conductivity for chem-ically stable BCZN3010 and BCZN3015. This plot indicatesthat total conductivity in both dry and moist 3% hydrogen inargon increases with Nd-doping in these cerate compositionsthat were found to be stable in CO2.

In samples red at the minimum temperature required toachieve 90% relative density, it was found that 35 vol% Ni isrequired to achieve percolation in compositions containing twodifferent cerate compositions. Fig. 8 shows the total conduc-tivity at 1000 C for samples containing reduced Ni. The sharpincrease in conductivity seen at 35% Ni is due to the presenceof continuous pathways for electronic conductivity through thenickel phase which were not present at lower Ni concentrations.Below 35% Ni, since continuous Ni pathways did not exist, theonly conductivity measured was the contribution of the ceratewhich is orders of magnitude lower than that of Ni.

3636 J.S. Hardy et al. / International Journal of Hydrogen Energy 32 (2007) 36313639

Fig. 5. SEM micrograph of a polished cross-section of 25Ni-BCZN2010 whichhad been sintered in air at 1250 C for 2 h then exposed to 3% hydrogen inargon at 1000 C for 12 h to reduce the NiO to Ni.

Inte

nsity

(Cou

nts)

20 30 40 502-Theta ()

60 70 80

(a)

0

5

10

15

20

25

30

35

0 10 20 30 40% Zr substitution

mo

l% B

a co

nver

ted

to C

arbo

nate

10% Nd Substitution5% Nd Substitution15% Nd Substitution20% Nd Substitution

(b)0 200 400 600 800 1000 1200 1400

100.00100.50101.00101.50102.00102.50103.00103.50104.00

TG /%

[1.1][2]

30% Zr

25% Zr

15% Zr

10% Zr

Temperature /C

[3.1][4]

(c)

353025201510

% Zr

(d)200 400 600 800 1000 1200 1400

Temperature /C

100.0

100.5

101.0

101.5

102.0

102.5

103.0

TG /%

[1]

[2][3]

[4][5][6]

1510

2010

3020

3005

3010

3015

Fig. 6. The results of TGA analysis on pulverized-doped barium cerate samples in a pure CO2 atmosphere, including (a) a plot of the mol% barium thatdegrades from cerate to carbonate as a function of Zr-doping, (b) TGA and (c) XRD scans of compositions with 10% Nd-doping and varying Zr-content, and(d) TGA scans of compositions with 30% Zr-doping and varying Nd-content.

Fig. 9 shows a plot of the relative density as a function ofsintering temperature for 35Ni-BCZN3010 that was sinteredfor 2 h in a 3% hydrogen in argon atmosphere. It can be seenthat the 90% relative density target is not met until 1350 C.However, hot pressing in vacuum at 1100 C for 2 h at 48MPaproduced a relative density of 99.2% was achieved. Based onthe width of the peaks in the XRD scan shown in Fig. 10 whichwas collected from a pulverized piece of the hot pressed sample,Jade (v6.1, Materials Data, Inc., Livermore, CA) XRD analysissoftware estimated the average crystallite size of the Ni to bebetween 31 and 36 nm and that of the cerate to be less than90 nm. In the backscattered electron SEM micrographs of thehot pressed sample in Fig. 11, it can be seen that the sampleconsists of a matrix of BCZN (lighter contrast) with Ni grains(darker contrast) ranging from approximately 1m down tothe nanoscale. This range of grain sizes is conrmed in thesecondary electron image shown in Fig. 12 that was acquiredfrom a fracture surface of the hot pressed sample. However,there were some areas that contained large clusters that were

J.S. Hardy et al. / International Journal of Hydrogen Energy 32 (2007) 36313639 3637

-5

-4

-3

-2

-1

0

1

2

3

0.0008 0.001 0.0012 0.0014 0.0016 0.00181/T (1/K)

Ln (S

igma*

T)

BCZN - 3015 in wet 3% H2BCZN - 3015 in dry 3% H2BCZN - 3010 in wet 3% H2BCZN - 3010 in dry 3% H2

Fig. 7. Four-probe conductivity of BCZN3010 and BCZN3015 in dry andmoist 3% hydrogen in argon.

0

20

40

60

80

100

120

140

160

180

15 20 25 30 35 40 45vol% Ni

Cond

uctiv

ity (S

/cm) a

fter r

educ

ing N

i

BCZN-2010 composite

BCZN-3010 composite

Fig. 8. Four-probe conductivity of Ni/BCZN composites at 1000 C in 3%hydrogen in argon as a function of Ni-content.

78

80

82

84

86

88

90

92

1250C/2hr 1300C/2h 1350C/2hSintering Conditions

Rel

ativ

e De

nsity

(%)

Fig. 9. Relative density as a function of sintering temperature for35Ni-BCZN3010 samples red for 2 h in 3% hydrogen in argon.

rich in Ni and apparently underwent a greater degree of sinter-ing shrinkage than the more homogeneously mixed surround-ing regions resulting in defects like those seen in the SEM

Fig. 10. XRD of a pulverized specimen of 35Ni-BCZN3010 hot pressed invacuum at 1100 C for 2 h at 48MPa.

Fig. 11. Backscattered electron SEM micrographs acquired at different mag-nications of 35Ni-BCZN3010 hot pressed in vacuum at 1100 C for 2 h at48MPa.

micrograph in Fig. 13. Such clusters must form during the com-bustion synthesis process and make up a small fraction of thepowder produced. Efforts must be made to prevent formation

3638 J.S. Hardy et al. / International Journal of Hydrogen Energy 32 (2007) 36313639

Fig. 12. Secondary electron SEM image of a fracture surface of35Ni-BCZN3010 that was hot pressed in vacuum at 1100 C for 2 h at 48MPa.

Fig. 13. Backscattered electron SEM image of 35Ni-BCZN3010 that was hotpressed in vacuum at 1100 C for 2 h at 48MPa showing defects caused bylarge Ni-rich clusters that shrank more during sintering than the surroundingmatrix.

of these clusters during synthesis or lter them out of the nalproduct.

4. Conclusions

Combustion co-synthesis of a cermet composite powder con-taining doped barium cerate and nickel oxide has been found toproduce an intimate mixture of particles with diameters in therange of 1050 nm and no appreciable substitution of elementsfrom either target material into the other. Post-synthesis calci-nation at 1000 C should be performed to ensure complete reac-tion of the cerate components. Composite powders with a widerange of target Ni concentrations can be made by this techniquewith good phase purity after calcination. When the NiO is re-duced to Ni, the resulting material is a mixed conducting com-posite containing electronic conducting Ni metal and protonconducting doped barium cerate which has potential utility inhydrogen separation membranes. It was found that NiO can be

reduced during sintering in a reducing gas such as 3% hydrogenin argon. Reduction during sintering avoids intergranular cracksthat form in samples that are reduced after sintering due tostresses that develop when the chemical reduction of NiO to Nibegets a signicant reduction in its volume. In the barium cer-ate, a minimum of 30% Zr-doping is required to achieve chem-ical stability in high-temperature CO2 environments, when thecomposition contains between 5% and 15% Nd. In single-phasecerate samples with 30% Zr that were found to exhibit stabilityin CO2, it was found that increasing Nd-content improves elec-trical conductivity up to at least 15% Nd. Samples containing20% Nd were not chemically stable in CO2 when 30% Zr wasadded, so further work will be done to determine whether higherlevels of Zr-doping will improve stability in 20% Nd-dopedcerates. In cermet composite samples sintered at the minimumtemperature at which greater than 90% density was achieved,35 vol% Ni was required to form a percolating network of Nithroughout the composite. While sintering at 1350 C for 2 h inan atmosphere of 3% hydrogen in argon was required to achieveabove 90% relative density in 35Ni-BCZN3010, hot pressingat 1100 C for 2 h at 48MPa produced 99.2% relative densityin the same composition. Powder XRD indicated a crystallitesize of 3136 nm for the Ni phase and less than 90 nm forthe cerate phase in the hot pressed sample. This is conrmedin SEM images, which reveal nanoscale Ni grains in a ceratematrix.

References

[1] President Bush. State of the union address to the joint session of congress;January 28, 2003.

[2] Analysis of effectiveness of the DOE hydrogen program. HydrogenTechnical Advisory Panel; 1998.

[3] Lin YS. Microporous and dense inorganic membranes: current statusand prospective. Sep Purif Technol 2001;25:3955.

[4] Breck WD. Zeolite molecular sieves. New York: Wiley; 1974 p. 636.[5] Bischoff BL, Judkins RR, Armstrong TR. Microporous inorganic

membranes for hydrogen purication. Presented at the DOE HydrogenSeparations Workshop, Arlington, VA; September 2004.

[6] Edlund DJ. Hydrogen separation and purication using dense metallicmembranes. Presented at the DOE Hydrogen Separations Workshop,Arlington, VA; September 2004.

[7] Sammells T. Ionically conducting membranes for hydrogen productionand separation. Presented at the DOE Hydrogen Separations Workshop,Arlington, VA; September 2004.

[8] Alberti G, Casciola M. Solid state protonic conductors, present mainapplications and future prospects. Solid State Ionic 2001;145(14):316.

[9] Norby T, Larring Y. Mixed hydrogen ion-electronic conductorsfor hydrogen permeable membranes. Solid State Ionics 2000;136:13948.

[10] Norby T. Solid-state protonic conductors: principles, properties, progressand prospects. Solid State Ionics 1999;125(14):111.

[11] Kreuer KD. Proton-conducting oxides. Ann Rev Mater Res 2003;33:33359.

[12] Balachandran U, Lee TH, Wang S, Picciolo JJ, Dusek JT, Dorris SE.Dense cermet membranes for hydrogen separation. Am Chem Soc FuelChem Div 2003;48(1):134 (preparation paper).

[13] Zhang G, Dorris SE, Balachandran U, Liu M. Interfacial Resistances ofNi-BCY mixed-conducting membranes for hydrogen separation. SolidState Ionics 2003;159:12134.

[14] Zhou X-D, Huebner W, Anderson HU. Size effect on the electronicproperties of doped and undoped CeO2. Diffusion and defect

J.S. Hardy et al. / International Journal of Hydrogen Energy 32 (2007) 36313639 3639

data-solid state data, pt. A: defect and diffusion forum; 2005, p. 24244,27789.

[15] Zhou X-D, Huebner W, Anderson HU. Unpublished data.[16] Wienstrer S, Wiemhfer H-D. Investigation of the inuence of

zirconium substitution on the properties of neodymium-doped bariumcerates. Solid State Ionics 1997;101103:11137.

[17] Ryu KH, Haile SM. Chemical stability and proton conductivity of dopedBaCeO3-BaZrO3 solid solutions. Solid State Ionics 1999;125:35567.

[18] Chick LA, Pederson LR, Maupin GD, Bates JL, Thomas LE, ExarhosGJ. Glycine-nitrate combustion synthesis of oxide ceramic powders.Mater Lett 1990;10:6.

Development of passive hydrogen separation membranes made from Co-synthesized nanoscale cermet powdersIntroductionExperimentalResults and discussionConclusionsReferences