●

Development of Mixed-Conducting Ceramic Membranesfor Hydrogen Separation *

J. Guan, S. E. Dorris, U. BalachandranEnergy Technology DivisionArgonne National Laboratory

Argonne, IL 60439

M. LiuSchool of Materials Science and Engineering

Georgia Institute of TechnologyAtlanta, GA 30332

April 1998

The submitted manwxipt has been created by the

University of Chicago ss Operator(”Argonn&) under

Contract No. W-3 1-1OP-ENG-38 with the U.S.

Department of Energy. The U.S. Government

retains for itself, snd others acting on its behalf, a

paid-up, nonexclusive, irrevocable worldwide

license in said srticle to reproduce, prepare

derivative works, distribute copies to the public,

and perform publicly and display publicly, by or on

behaif of the Government.

Paper to be presented at the 100th Annual Meeting and Exposition of the AmericanCeramic Society, Cincinnati, OH, May 3-6, 1998.

*Worlc supported by the U.S. Department of Energy, Federal Energy Technology Center,under Contract W-3 1-109-Eng-38.

DISCLAIMER

This report was prepared as an account of work sponsoredby an agency of the United States Government. Neither theUnited States Government nor any agency thereof, nor anyof their employees, make any warranty, express or implied,or assumes any legal liability or responsibility for theaccuracy, completeness, or usefulness of any information,apparatus, product, or process disclosed, or represents thatits use would not infringe privately owned rights. Referenceherein to any specific commercial product, process, orservice by trade name, trademark, manufacturer, orotherwise does not necessarily constitute or imply itsendorsement, recommendation, or favoring by the UnitedStates Government or any agency thereof. The views andopinions of authors expressed herein do not necessarilystate or reflect those of the United States Government orany agency thereof.

DISCLAIMER

Portions of this document may be illegiblein electronic image products. Images areproduced from the best available originaldocument.

DEVELOPMENT OF MIXED-CONDUCTING CERAMIC MEMBRANESFOR HYDROGEN SEPARATION

Jie Guan, Stephen E. Dorris, Uthamalingam BalachandranEnergy Technology Division, Argonne National Laboratory, Argonne, IL 60439

Meilin LiuSchool of Materials Science and Engineering, Georgia Institute of Technology,Atlanta, GA 30332

ABSTRACTSrCeOJ- and BaCeO~-based proton conductors have been prepared and

their transport properties have been investigated by impedance spectroscopy inconjunction with open circuit voltage and water vapor evolution measurements.BaCe0.gY020~.a exhibits the highest conductivity in a hydrogen-containingatmosphere; however, its electronic conductivity is not adequate for hydrogen ,separation in a nongalvanic mode. In an effort to enhance arnbipolar conductivityand improve interracial catalytic properties, BaCeO.gYO,zOMcermets have beenfabricated into membranes. The effects of ambipolar conductivity, membranethickness, and interracial resistance on permeation rates have been investigated. Inparticular, the significance of interracial resistance is emphasized.

Keyords

Mixed conductors, hydrogen separation, transport properties, perovskites

INTRODUCTIONMixed ionic-electronic conductors (MIECS) have been studied for

application in many solid-state electrochemical systems such as solid oxide fbelcells (SOFCS), chemical sensors, and membranes for gas separation [1-7]. Onewell-known class of MIECS consists of partially substituted perovskite-typeoxides such as CaZr03, SrCe03, and BaCe03, in which substitution of Ce or Zrby trivalent cations introduces oxygen vacancies and other charged defects and .leads to mixed conduction in atmospheres containing 02, Hz, and HZO vapor.

1

.

.

These materials exhibit significant proton conduction in a hydrogen-containingatmosphere and are used for various applications. CaZr03-based ceramics arebeing used as a hydrogen and humidity sensor [3]; BaCe03-based prototypeSOFCS are in development and significant progress has been made [4].

Meanwhile, dehydrogenation of hydrocarbons, electrolysis of steam, and pumpingof hydrogen have also been demonstrated [5].

Because oxygen ion/electron conductive ceramics are being studiedintensively as membranes for oxygen separation [6], it is immediately evident thatdense ceramic membranes fabricated born prototielectron-conducting ceramicscould also provide a simple and efficient means of separating hydrogen from gassteams and offer an alternative method of hydrogen recovery. To be suitable forhydrogen separation in a nongalvanic mode, the ceramic membranes must havesufficient protonic and electronic conductivities to achieve high permeability andselectivity to hydrogen. In addition, the materials must exhibit high catalyticactivity for oxidation and evolution of hydrogen at the solid/gas interfaces.

In this study, doped SrCe03 and BaCe03 were prepared and theirtransport properties were investigated. Dense BaCeo8Yo.203.5cermet membranes

exhibited high hydrogen permeation rates. The dependence of permeation rate onmembrane thickness and on interracial polarization was characterized.

MEMBR4NE SEPARATION MECHANISMSIn a hydrogen pump, hydrogen can be pumped by a voltage applied to the

ceramic conductor from the anode to the cathode side. The hydrogen evolutionrate is governed by Faraday’s law, and the current efficiency depends on thetransport properties of the conductor, such as the proton transference numberunder the operating conditions. Alternatively, hydrogen can be pumped byshorting the external circuit, where the driving force is the electrochemicalpotential difference across the proton conductor.

A simple and efilcient approach is to use a mixed conducting membranewithout electrodes. Under the conditions of ambipolar diffhsio~ the currentdensity due to the motion of protons can be expressed as [8]

(1)

2

where L is the thickness of the membrane, q is the interracial overpotential, EN is

the Nemst potential imposed across the membrane, and a,~~ is the arnbipolarconductivity of the membrane defined as

cJeaH+~amb=

CJe+CYH+‘if the interaction between electrons and protons is negligible.of hydrogen, N~2, is related to the proton current densit y by

(2)

The permeation flux

(3)

It is clear that the permeation flux of hydrogen through the MIEC membrane canbe improved by enhancing of arnbipolar conductivity and reducing of themembrane thickness and interracial overpotential.

The interracial overpotential q and arnbipolar conductivity ~~b can be

determined by measuring permeation rates through membranes with differentthicknesses. If Eqs. 1 and 3 are rewritten as [9]

(4)

then the interracial polarization resistance, RP,can be estimated from

(5)

while ambipolar conductivity can be determined as

oam~=

EXPERIMENTALSample Preparation

-1

i3(EN/2FNH2)

1~L “

-1

(6)

SrCe03-and BaCe03-based ceramic powders and pellets were prepared asdescribed in Refs. 10 and 11. Pt paste was applied to polished surfaces of thepellets and fued in air at 1200”C for 12 min to form porous electrodes beforeelectrochemical characterization. Cermet membranes were prepared by mixingmetallic powder with BaCeO.SYO.@s.~ After being ball-milled for 24 h, the

resulting powders were pressed at 200 MPa and then sintered at 1400”C for 5 h in4’% H2balance argon.

Impedance SpectroscopyThe ceramic pellets with two Pt electrodes were immersed in various

atmospheres at different temperatures (600-800°0. Total conductivity WaS .measured with an impedance analyzer (HP 4192A) in the frequency range of 13

3

4

MHz to 5 Hz under open-circuit conditions. Water vapor was introduced bybubbling desired gases through deionized water at room temperatures.

Ionic Transference NumbersTransference numbers were determinedly measuring either open-circuit

vohage (OCV) or water vapor evolution rates. OCVS of fuel cells andconcentration cells, measured by a high-impedance digital multimeter (HP 3446A),were used to estimate the ionic and electronic transference numbers. The ionictransference numbers so obtained maybe underestimated in some cases because ofinterracial polarization [12]. Evolution of water vapor by discharging an 02/H2

fhel cell was used to determine the ratio of proton and oxide ion transferencenumbers to the total ionic transference numbers. The gas inlets in the setup werepositioned as close as possible to the test sample. Gas flow rates were adjustedto 50-100 cm3/min. Water vapor concentration in both cathode and anode sideswas measured with a digital hygrometer (Fisher Scientific) under open- and short-circuit conditions.

Hydrogen Pumping and Water Vapor ElectrolysisA 2.09-mm-thick pellet of BaCeo.8Yo.203.6with three electrodes was

affixed to a test setup (Fig. 1) with a glass sealant. Hydrogen was pumped fromthe anodic to the cathodic side by applying a constant DC current. The anodicside was fed with 4°/0 H2 balance N2 at 86 cm3/min and the cathodic side wasswept by Ar at 43 crn3/min. The anodic overpotentials were directly determinedby measuring the potentials between the anode and the reference electrode. Inwater vapor electrolysis, Ar saturated with water vapor at room temperature wasfed to the anode side at 86 cm3/min, and the cathode side was also swept with Arat 43 cm3/min. Gas concentrations were analyzed with a gas chromatography (HP5830A) and a hygrometer.

Hydrogen Permeation through Cermet MembranesPolished cermet membranes with three different thicknesses (0.095, 0.114,

and 0.205 cm) were also affied with a glass sealant to one end of A1203 tubes, asillustrated in Fig. 2. The effective membrane area for permeation was 1.4 cm2. Asample gas (4°/0 H2balance Ar) was fed to one side at a rate of 86 cm3/min, while acarrier gas (Ar) was swept to the opposite side of the membrane at a rate of 43cm3/rnin. The gas concentrations were analyzed with a gas chromatography thatwas periodically calibrated with a standard gas. Gas leakage through the setup

was tested with helium as the feeding gas; typical leakage was < 10°/0of the totalpermeation flux.

\\\\\\\\\\\

Ar + H,2

#7//Yxx+?zx?77

Fig. 1 Schematic illustration of hydrogen pumping and water vapor electrolysis.

Inlet IGas MixtureContainingH2(uP to 100% H$

TO Hood

\GortiftexProtector *(Rated to 450%) — —

Furnaceup to 1000Oc

AluminaTube 3-

11t 1- — — ~e~=wle tt

1- Outlet II

Seal Cap (to hod)

1–

+ AluminaTube 2

— Membrane

I!Y

— Glass Seal GC

— AluminaTube 1

RGes Chromatography

4} OutletI (to hood)

JLlj

UnknownH2 Concentrationq r{’ J

Inlet II Sweep Gas AriN2

Fig. 2 Schematic illustration of setup for hydrogen separation.

5

RESULTS AND DISCUSS1ONConductivity

0,045 L., ,. ICC, >lO. -,1. .S .I-.--,.,,%,-%-j

+600”C ,.:-...

...-u--7oo0c .:”. ...

-.-0- 8000C .~%.:...

,... ...... .0

...

d~ .. B-s.\j

...0 b

.... H. .\....

/’ \...

.4. /. .\, .\

0“’ P’ .h0’

/

1In 4% hydrogen 0.....,..0.040 ...

,.”. ..

--,E0.035

.0

.go.030I

0.010

I

0.005 L--0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Y Dopant Level, x

Fig. 3 Conductivity of BaCel.XYXO~-3in 4% hydrogen at various temperatures.

BaCeO~-based materials exhibit higher total conductivities than SrCeO~-based materials in both hydrogen and oxygen [11]. Conductivity of BaCel-XYX03.&varied with the doping concentration x (Fig. 3). At a given temperature,conductivity measured in 4’%0H2, which is dominated by ionic conductio~increased monotonically with dopant concentration fi-om x = 0.05 to 0.2, thendecreased for an Y-concentration of 0.3, suggesting that the conductivity inhydrogen reached its maximum at 0.2< x <0.3. The increase in conductivitywith dopant concentration could be related to an increase in charge carrierconcentration through the reaction

Ce02Y*O3 + 2Y;e + 30; +V; (7)

in tie volubility limit. The decrease in conductivity for x >0.2 was probablycontributed to phase segregation when the volubility limit of Y was reached.

Ionic Transference NumbersThe proton transference number of Y-doped SrCeOJ and BaCe03 materials

was estimated to be about 0.8-0.95 from a hydrogen concentration cell in the

6

temperature range of 600-800”C. The ratios of proton and oxide ionnumbers to total ionic transference numbers were fiuther studied with

100% 02, Pt I SrCe0.g5Y0.0jOJ4I Pt, 4% H2 + Ar.

transferencea fhel cell of

(8)

~:;

r

:-:.W:::I 1 8 t ,

1

@m. 10%Gd-EaCa03 Pi ~~ ‘\.

c“..

\ \,-..

$ 0.6 - ~.~.

g ~..o./-v

s 0.4 - ./””zz

/-” *.A..-. OwObn. la%G&aacOo*[q ./ . . ..-.

0.2 - .-

- Oxidaion,lonY-aace03p’]

.......-...&........+--””+”fik&.fik&.

o 1.. ,,...._.--- t , 1

550 600 650 700 750 800 850

Temperature ~C)

Fig. 4 Ratios of proton and oxygen ion transference number to total ionictransference number in SrCeog~Yoo503.aand in 13aCe03-based materials [7].

When the two electrodes are set at equal potentials, the current that passesthrough the electrolyte is approximately ionic, while electronic current is ‘negligible. Accordingly, the ratios of ( t~+/t iO~and to2_/ tiOn) can be determined as

follows:

and

where WCand w~ are the rates

tH+ wc—=— (9a)

ti~” W~‘to2- wa—=— (9b)t. Wt ‘Ion

ti~n= tH++ to2_, (9C)

of water vapor evolution tiom the cathode and

anode sides, respectively, and w~is the total rate of water vapor evolution Iiom

both sides or the total rate of water evolution calculated from the short-circuitcurrents (when Faraday’s law is applied). The rates of water vapor evolution dueto the electrochemical process were calculated fi-om the flow rates of carrier gas

7

and humidity measured in both sides of the cell under open- and short-circuitconditions. The rate of water vapor evolution from the anode side (hydrogen side)was much slower than that from the cathode side (oxygen side), indicating that

~ proton transport dominated the ionic transport. The total rate of water vaporevolution observed on both sides was in very good agreement with the ratecalculated born short-circuit currents. As shown in Fig. 4, the ionic transferencenumbers are dominated by protons, even at 800°C. Unlike BaCeO~-basedmaterials [13], SrCeo,g5Yo,0503.8did not exhibit a conduction transition fi-om

proton to oxide ion as temperature increased from 600-800”C. In the temperaturerange studied, proton transference numbers were much higher than oxide iontransference numbers.

Hydrogen Pumping and Water Vapor Electrolysis

0.7. 1 { 1 t

-.’..-

0.6 - .......

--... --.-Theoretical.. .

.. ~0.5 - + 600”C ,..

...~800”C -..

...0.4 ..-

0.3 -

0.2 -

[..

... ‘1

0 t“””‘ I t I t i

o 20 40 60 80 100

Applied Current Deneity(mlUcm2)

Fig. 5 Hydrogen pumping rates as a fi.mction of current density and temperatures.

Shown in Fig. 5 are the hydrogen evolution rates as a fimction of appliedc&ents and temperatures. The theoretical values calculated from Faraday’s laware also shown for reference. When the applied current was low (<22 mA/cm2),the current efficiency was close to unity, which was in good agreement with thehigh proton transference numbers of the materials (0.8-0.95). Current efficiencydropped significantly at higher current densities, and this was more evident at hightemperature (800”C) than at low temperature (600°C).

8

L

Fig. 6

,.....’

...,.$

. . . . . .. . . Theoretical,,-

.. ... .

~600”C ,..

~ 600°C...

,.-. ..

..,”. .

.,”’,..,.,..””

..-,,.

0 1“4”’ 1 1 # i 10 20 40 60 60 100

Ap@iect CurrentDensity(mNcm2)

Hydrogen evolution rates during water vapor electrolysis.

Very dry hydrogen (dew point < -40”C) was obtained by electrolyzingwater vapor. As shown in Fig. 6, the current efficiencies were lower than thoseobserved in hydrogen pumping. The low current efficiencies suggested thatoxygen ion and electron conduction was significant in water vapor electrolysis.Electron holes could have been formed in a reaction of the oxygen vacancies withthe oxyge~ which may remains locally at the anodic intetiace with a relativelyhigher pressure as a by-product of the water vapor electrolysis:

~02+V~+O; +2h”. (lo)

Hydrogen Permeation through Cermet MembranesFigure 7 shows net permeation flux through three cermet membranes.

Hydrogen permeation rates increased with temperature and were approximatelyproportional to the inverse of membranes thickness (0.095 to 0.205 cm) at high

temperature (800”C), with a slope close to (E~cr,~~)/2F. This suggests that therate-limiting step is the bulk properties for samples with thicknesses behveen0.095 and 0.205 cm. However, at low temperatures (<650”C), the flux datashowed significant deviation from this relationship, suggesting that interracialresistance, rather than bulk properties, is significant at low temperatures.

,

9

‘.oo~550 600 650 700 750 800 650

Temperature (“C )

Fig. 7 Temperature and thickness dependence of hydrogen permeation

through BaCeo8Yo,20q.6cermet membranes.

Significance of Interracial Polarization

‘ooo~

rates

1 1.2 1.4 1.6

Log(k n@fcm 2,

Fig. 8 Anodic overpotentials as a function of appliedhydrogen pumping.

t .6 2

current density during

As shown in Fig. 8, significant anodic overpotentials were observed inhydrogen pumping, which consumed about 50-70% of the applied potentialsunder the pumping conditions studied. Interracial polarization resistances ofcerrnet membranes were investigated through dependence of the gas permeationrate on membrane thickness, as implied by Eq. 5. The intetiacial polarizationresistance declined dramatically with an activation energy of about 1.03 eV whentemperature was increased flom 600 to 800°C (Fig. 9). At high temperature(800°C), bulk resistance is much greater than interglacialresistance when sarnpIethickness is >0.095 cm, at lower temperatures, however, interglacial resistance islikely predominant. Thus, reducing the membrane thickness will enhance thepermeation flux significantly only if the bulk transport properties are moreimportant than interracial polarization.

.f-C

-0.4

-0.6

-0.8

-1.0

-1.2

-1.4

-1.6

~\\

\‘9

\\

\o\

\\

\\‘n

\\

\b

1 t 1 I 1

0.90 0.95 1.00 t .05 1.10 1.15 1.20

low-r (K-’)

Fig. 9 Interracial resistance of BaCeo.8Y0.203.3cermet membrane (1.4 cm2 in ~ea)

as a function of temperature.

CONCLUSIONSSrCe03- and BaCeOJ-based materials are good proton conductors in

hydrogen-containing atmospheres. Hydrogen was pumped though BaCe0.8Y0.zOJ.6by a DC current; the current efficiency decreased with an increase in temperatureand applied current density. Significant interracial polarization was observed andfurther reduced the energy efficiency in hydrogen pumping and steam electrolysis.

11

BaCeo8Y020J.a cermets can be used as MIEC membranes for hydrogen separation

in a nongalvanic mode. Intefiacial resistances were significant at lowtemperatures. Further reduction of membrane thickness may not improve thepermeation rate effectively if interracial polarization is more significant than bulkresistance.

ACKNOWLEDGMENTSThis work has been supported by the U.S. Department of Energy, Federal

Energy Technology Center, under Contract W-3 1-109-Eng-38

REFERENCES[1]. N.Q. Minh and T. Takahashi, “Science and Technology of Ceramic FuelCells,” Elsevier Science, Amsterdam, 1995.[2]. H.J.M. Bouwmeester and A.J. Burggraa~ kc Fundamentals of InorganicMembrane Science and Technology, A.J. Burggraaf and L. Cot, eds., ElsevierScience, Amsterdam, 435 (1996).[3]. N. Fukatsu, N. Kurita, T. Yaj& K. Koide, and T. Ohashi, J. Alloys andComp., 231,706-12 (1995).[4]. N. Taniguchi, E. Yasumoto, T. Garno, J. Electrochem. Sot., 143,1886 (1996).[5]. J. Langgu~ R. Dittmeyer, H. Hofbnn, and G. Tomandl, Chemie IngenieurTechn~ 69,3197 (1997).[6]. U. Balachandr~ J.T. Duselq S.M. Sweeney, R.El. Poeppel, R.L. Mieville,P.S. Maiya, M.S. Kleefisch, S. Pei, T.P. Kobylinski, C.A. Udovich, and A.C.Bose, Am. Ceram. Sot. Bull., 74,71 (1995).[7]. H. Iwahara, T. Yajima, and H. Uchi@ Solid State Ionics, 70/71, 267 (1994).[8]. M. Liu, Proc. 1st Int. Symp. on Ionic and Mixed Conducting Ceramics, T. A.Ramanarayana and H. L. Tuner, eds., The Electrochemical Society, Bennington,NJ, Vol. 91-12>215 (1991).[9]. M. Liu and A. Joshi; 1st Int. Symp. on Ionic and Mixed ConductingCeramics, T. A. Ramanarayana and H. L. Tuner, eds., The ElectrochemicalSociety, Pennington, NJ, Vol. 91-12,231 (1991).[lOj. J. Guan, S. E. Dorris, U. Balachandran, and M. Liu, Solid State Ionics, 100,45 (1997).[11]. J. Guan, S. E. Dorris, U. Balachandran, and M. Liu, J. Electrochem. Sot.,145,1780 (1998).[12]. M. Liu and H. Hu, J. Electrochem. Sot. 143, L109 (1996).[13]. N. Bonanos, K. S. Knight, and B. Ellis, Solid State Ionics, 79,161 (1995).