Interfacial reactions associated with ceramic ion transport membranes

SOLID

STATE Solid State Ionics 75 (1995) 157-165

Ionlo

Interfacial reactions associated with ceramic ion transport membranes B.C.H. Steele Department of Materials, Imperial College, London SW7 2BP, UK

Abstract Assuming ohmic behaviour for the relevant interfacial kinetics a simple equivalent circuit has been used to identify experimentally accessible parameters which may control the oxygen flux through a variety of technological devices. In particular the oxygen surface exchange coefficient (k cm SC’), which can be determined by isotopic exchange measurements is proportional to a characteristic electrode current density (& A cm-*) which determines the electrode resistance (RE &I2cm2) in solid-state electrochemical systems. For ceramic ion-conducting membranes a characteristic membrane thickness (L,) at which the changeover from bulk to surface control occurs is shown to be equal to D*/k where D* (cm* s- ’ ) is the oxygen self-diffusion coefficient in the oxide material. Attention is also drawn to correlations between D and k. It is noted, for example, that the ratio D/k often has a value around lo-* cm ( 100 pm) for most A02 fluorite and ABOs perovskite oxide materials, which implies that fabricating membranes less than 100 pm thick will not be advantageous unless the value of k can be specifically increased. Mechanisms responsible for correlations between D and k remain obscure and should be a fruitful area for further investigations. Finally, specific examples of materials selection for ceramic fuel cell operation over a wide range of temperatures (450-1000°C) are briefly surveyed. Keywords: Interfacial reactions; Ionic transport

1. Introduction

Many technological devices operating at elevated temperatures incorporate oxide components which are required to sustain high oxygen fluxes. Examples include mixed conductors for ceramic membranes designed to separate oxygen from air, oxide electrodes for ceramic fuel cells, electrolysers, oxygen pumps, and amperometric oxygen monitors. Similarly, oxygen electrolyte materials in these high-temperature electrochemical systems are also required to exhibit relatively high oxygen-ion conductivities. It should also be noted that high oxygen fluxes through these devices also imply large values for the oxygen

transport kinetics across the relevant gas-solid interfaces. A simple model has been proposed [ l-31 for the preliminary selection of materials that could satisfy specific design requirements. The author has found this approach also useful to identify those areas where further research is required to improve our understanding of the relevant rate-controlling mechanisms. This model incorporates interfacial (RE) and bulk (R,) resistive terms exhibiting ohmic behaviour, so that oxygen fluxes (current densities) through a particular device can be represented by the equivalent circuit depicted in Fig. 1. Depending upon the performance requirements for a device then specific values can be assigned to the current density (j,,, A

1)167-2738/95/$09.50 0 1995 Elsevier Science Publishers B.V. All rights reserved. XWI U167-2738(94)00182-O

B.C.H. Steele /Solid State Ionics 75 (1995) 157-165

158

the associated specific ionic conductivity. It can also be shown that RE equals (RT/zF) ( 1/j,), where R, T, z, F, have their usual meanings and j, represents a ‘characteristic’ electrode current density. This relationship between RE and j, can be derived following consideration of the overall interfacial reactions,

h-

'/2RE

‘/ZR,

1/202+Vg i I I

V2R,

I_______________ 77

j.

=

________________i

q/V+ +R,)

Fig. I. Schematic diagram and equivalent circuit indicating resistive terms controlling oxygen flux through ceramic ion conducting membranes.

cm-*) and allowed resistive voltage losses (9, V) across the ceramic oxide assembly. For example, in solid oxide fuel cell (SOFC) systems a target power density of 0.5 W cm-* for individual cell assemblies is often mentioned. If this is to be attained at 950°C with 85% fuel cell conversion this implies that the total resistive voltage losses (q) should be less than 0.2 V. For a power density about 0.5 W cm-2 (e.g. 0.7 V at 0.7 A cm-‘) it follows that the total cell resistance (R,+R,) should not exceed about 0.3 Q cm*. In subsequent discussions it has been assumed that RE=R0=0.15Rcm2. Depending upon the system under examination, obviously other values can be assigned to the parameters, R,, Ro, j, and ?Iproviding that the overall relationship is satisfied, namely:

(2)

The magnitude of the electrode resistive term, RE, will be determined by the rate-controlling step in reaction (2). In principle this rate-limiting process could be due to a Faradaic charge-transfer step (R,), or to mass transport limitations (R,), or chemical reaction kinetic control (RR). At elevated temperatures and low overpotential it is probable that the kinetics of the relevant chemical reaction will be controlling [ 41. However, whatever the rate-controlling mechanism, very similar expressions can be derived IS] for all the possible processes, providing the overpotential value is small, namely: R

T

=RTL

R

zFj,'

C

=RTL zFj,_

R

’

R

=RTk!, ZF_k P

The quantities jO, j,, jR denote the exchange current density, limiting current density, and reaction current density respectively. Clearly if the stoichiometric factor, or, and reaction order, p, are similar, then all the expressions are identical, and it is appropriate therefore to define a ‘characteristic’ electrode current density, j,, for the specific electrode structure under examination, i.e. R E =RTI

zF

j,=~~l(&+&).

+2e’$O,X.

J’,

’

(1)

It is also useful to note that the current density can be expressed in terms of molar fluxes, molecular oxygen volumes and ion fluxes. For example, to sustain a current density of 0.5 A cm -’ at 1000 K approximately 1 in lo6 oxygen molecules hitting the oxide should be adsorbed on the surface. Moreover the site turnover rate at the surface will be approximately 10’ s-‘. It should be emphasised that these requirements would represent relatively high values for heterogeneous catalytic processes. Furthermore it should be noted that R,, the ionic resistance, can be replaced by L/a, where L (cm) is the thickness of the oxide ceramic, and u (S cm-i )

2. Ceramic ion transport membranes separation

for oxygen

Mixed conducting perovskite oxides have been proposed for the separation of oxygen from air [ 6 1, and for the supply of oxygen in partial oxidation reactions [ 7 1. For the oxygen separation process very high values for the partial oxygen-ion conductivity have been reported [ 61 for selected perovskite compositions, (e.g. La,,2Sr0.8C00.sFe,,203-X), and so it is instructive to examine how the overall oxygen flux will be influenced by the surface exchange kinetics.

B.C.H. Steele /Solid State Ionics 75 (1995) 157-165

Making the appropriate substitutions in Eq. ( 1) one obtains, .

j=q/(Llo+RTlzFj,)

(3)

In this expression q can be replaced by (RT/zF) In and so Eq. (3) is similar to one of the expressions derived by Liu [ 8 1. If the interfacial resistance term, RTIzFj,, can be ignored then the expression becomes identical to the well-known relationship:

p&/p&,,

j,

=

RTZ

In

ZF L

RE=Ro

where E

=RTI

and

zF jE

R,=L

0’

The characteristic electrode current density (jE) can be shown [ 8,1,2] to be equal to zFk/ V,,, where k (cm s- ’ ) is the surface exchange coefficient which can be determined by isotopic exchange techniques [ 11,121 1 00

c z

5

5 _c

010

’

i,=s/(R,+Q

lj=Ol" Q

0

=

\

10~10 cm-’

\ \

= L/o

\

R,=015Rcm2

\

0 01

, I

Dam &=I/,RT.

Making the appropriate substitutions, it follows that: LV,,,RT Dam

p;;,

derived by Wagner [ 91 more than sixty years ago for parabolic metal oxidation kinetics. The influence of surface kinetics upon the oxygen flux using Eq. (3) is depicted in Fig. 2 for a specific value of RE (0.15 R cm’) and rl (O.lV). For large thicknesses ( 1000 urn) of the membrane the bulk resistance (L/o) dominates but as the thickness of the membrane decreases the surface kinetics becomes rate-limiting. The membrane thickness at which the change-over from bulk to surface control occurs is given by

R

and V, is the molar volume of oxygen in the oxide membrane. The relationship between the oxygen-ion conductivity, a, and self-diffusion coefficient, D*, is also given by the Nernst-Einstein equation:

RTV -m=_ ZF zFk

‘b2

159

““”

‘-““‘,

10

100

1000

Thiclaless (pm)

Fig. 2. Oxygen flux as a function of membrane thickness for selected operating conditions.

that is L=D*/k.

Attention [ 13,1] has already been drawn to the importance of the ratio D/k in determining the influence of surface exchange kinetics upon oxygen fluxes through mixed conducting membranes. More recently Bouwmeester et al. [ 141 have provided further experimental evidence that oxygen fluxes through mixed conducting ion transport membranes are limited by the surface exchange kinetics. It has been noted by Kilner [ 151 that the quantity D/k often has a value around 1O-’ cm ( 100 urn) for most AO, fluorite and many ABOS perovskite oxide materials. This implies that fabricating membranes less than 100 urn thick will not be advantageous unless the value of k can be specifically increased. Procedures to bring about this increase could involve the deposition of surface-active species. Alternatively the surface kinetic limitations may be ameliorated by increasing the effective surface area using a graded porous surface layer of the same material. This latter approach has been shown to be effective by Thorogoodetal. [16]. The mechanisms responsible for correlations between D and k remain obscure but could be due to significant trapping of the oxygen vacancies by segregated dopants in near-surface regions involving the formation of defect complexes, e.g. (Vb;-Yzv)’ in stabilized zirconia electrolytes. It is certainly important to understand the basic mechanisms as such correlations have been reported elsewhere. For example Kleitz [ 41 has drawn attention to the correlation between RE and p (specific resistivity of electrolyte), and reports RJp equal to 3.5 X 1Ow2for a variety of

160

B.C. H. Steele /Solid State Ionics 75 (I 995) 15 7-165

electrode/electrolyte combinations when the value of poz equals 10 -4 atm at 780°C. It should be noted that:

&lp=REa= The value of close to the which would the electrode port through

RT V,,, D*(zF)’ z

~~~

m

D =

k.

3.5 x 10 -’ for D/k reported by Kleitz is Kilner correlation [ 131 value of lop2 suggest that under specific conditions resistance is dominated by charge transthe near-surface region.

3. Ion transport membranes in ceramic fuel cells

3.1. General

As mentioned in Section 1, high-performance fuel cells (powder density - 0.5 W cmm2) should have cell resistances lower than 0.3 Q cm2. It is appropriate, therefore, to assume a maximum value for L/aof 0.15 R cm2. Using the ionic conductivity-reciprocal temperature relationships depicted in Fig. 3 for selected ceramic electrolytes it becomes easy to calculate the maximum allowable thickness (right-hand ordinate) for a given conductivity value. It should be noted that 150 pm has been taken as the minimum practical thickness likely to be achieved for a self-supported membrane.

3.2. Membranes for operation above 950°C

The

well

known cubic zirconia electrolyte, has an ionic conductivity value of 1O- ’ S cm- ’ at 95O”C, and so self-supported ceramic electrolytes with a thickness around 150 pm have been successfully incorporated into planar SOFC modules by many groups around the world, and power densities approaching 0.5 W cm-’ have been achieved. However, operating at 950°C has problems associated with the lack of economic balance-of-plant equipment, or reliable bipolar plate materials, and there is a requirement to reduce the operating temperature to the intermediate range 700-900°C (for CH, fuel) and 450-500°C (for CHsOH fuel ). Zr0.85Y0.d1.925

3.3. Membranes for operation at 750-800°C Examination of Fig. 3 indicates that Zr,.,,Yo,,50 L.925has a conductivity about 2.5x 1O-2 S cm-’ at 800°C and for a value for L/a of 0.15 R cm2 the thickness of the electrolyte should be in the range 3540 pm. It is necessary, therefore, to fabricate supported thick film electrolytes, and Westinghouse SOFC technology provides an example of how this can be achieved for large modules (25- 100 kW) CeO.sGd,.,O,.,, electrolytes exhibit ionic conductivity values of 10-l S cm- ’ at 800°C and so self-supported ceramic electrolytes could be selected for electrochemical oxygen generators, provided the applied voltage does not exceed approximately 0.5 V. Higher voltages will produce an increasing electronic contribution to the total conductivity which will result in inefficiencies. Although offering higher conductivities the Bi203 solid-solution electrolytes can be irreversibly reduced by the application of voltages greater than about 0.6 V. Moreover, the BizO,-based electrolytes have very poor high-temperature mechanical properties and react with many of the obvious electrode materials [ 17 1. 3.4. Membranes for operation at 450-500°C Further examination of Fig. 3 reveals that the specific conductivity of Ce0.sGd0.,01.95 electrolytes has a value of 10e2 S cm-’ at 500°C and it is possible, therefore, to envisage using this material as a supported thick film ( - 15 pm) electrolyte structure. Moreover, the electronic conductivity under anodic conditions ( - 0.8 V) is much reduced at these lower temperatures and the associated additional losses can be tolerated. The additional development of metallic bipolar plates by Siemens/Plansee [ 18 ] ensures for the first time that individual cells can easily be connected together to assemble a stack operating in the temperature range 450-500°C.

4. Interfacial reactions 4. I. General As already mentioned in Section 2, the surface exchange coefficient, k (cm S-l), can be determined

161

B.C. H. Steele / Solid State Ionics 75 (I 995) 15 7- 165 900

800

700

400

500

600

T

300

('Cl

' -

1500pm

1SELF SUPPORTED / ELECTROLYTES

0.8

0.9

1

1.1

1.2

1.3

1.5

1.4

1.6

1.7

1.8

1000/T (K-') Fig. 3. Specific ionic conductivity values for selected oxide electrolytes as a function of reciprocal temperature.

by isotopic exchange measurements [ 10,11,191. At Imperial College [ 111 we prefer to employ ‘so/‘60 isotopic exchange together with a determination of ‘80/‘60 diffusion profile in the solid oxide (IEDP) technique. These measurements produce an unambiguous value for the oxygen self-diffusion coefficient, 08, and a value for the surface exchange coefficient, k, which is specific to the particular surface under examination. Typical data [ 1 ] for Lao.&ro.&In03-, and Lao.sCao.4Coo.*Feo.203-x are reproduced in Figs. 4 and 5. It should also be emphasized that the kinetic data values will be a function of the oxide stoichiometry which is controlled by the imposed oxygen partial pressure or overpotential as shown in Fig. 6 for three perovskite compositions. In the temperature range 800-900°C the oxygen vacancy diffusion coefficient, D,, typically has values of 10p6- 1O-’ cm2 s - ‘. It follows that the large changes in the oxygen self-diffusion coefficient, 08, reported in Figs. 4 and 5 can be attributed to major differences in the oxygen vacancy

‘3 8

c

oar 0

1

L

3

2 Depth

5

6

!~l@~rm

Fig. 4. ‘*O/‘6O isotopic diffusion anneal profile in La,,6,Mn,,3509_, annealed for 270 s at 900°C.

concentration, C,, exhibited by these materials (Fig. 6) at high oxygen partial pressures, i.e. COD?,= C, D,, where Co is the concentration of oxygen ions.

B.C.H. Steele /Solid State Ionics 75 (1995) 157-165

162

10-5

Temperature ("Cl 600 500 700 I I I

800 I IO) ti&sl

\ 10-6

C= ceramic - 10-1 O= dispersedon ceramicelectrolyte - 6-8 rn2.a+ I

‘“&*

t(D) \

3

,_

10'2, 5

10-?

10-6

10-q

lo-'0

Fig. 5. 1*O/‘6O isotopic diffusion anneal profile in L~I&~.~Co,,8Feo.203_Xannealed for 400 s at 800°C. po,

lo+

ibar) ,O%

I

0.9

I

1.0

I

I

1.1 1.2 1000/T(K)

I

I

1.3

Fig. 7. Compilation of oxygen surface exchange coeffkient values for selected oxides as a function of reciprocal temperature.

-!I02 -0

03

Q 1-L 12

-\

L_._~JLL_

13

1s8

06

n4

/

J

,3)_,;, _J

_~

02

‘_

‘_

E(V) ip02rer,s,ar 1

Fig. 6. Composition-oxygen partial pressure isotherms for Lao.b5Sr0.35Mn03+,,k.$b5MnO~-,, and La0.+%.3Co03-, at

1000°C. 4.2. Surface

exchange coef$cient values

Selected values for the surface exchange coeffcient, k [ 10-l 2,19,20] determined by isotopic exchange experiments are compiled in Fig. 7 together with derived values for the associated characteristic electrode current density, jE. These values were all obtained at high oxygen partial pressures and so are applicable to cathodic conditions in SOFC systems. Examination of Fig. 7 reveals a number of interesting features. The solid electrolyte Zr,,,5Y0.,501.925 has a low surface exchange coefficient, and the deposition of porous Pt electrodes only increases the value of k by about three orders of magnitude. This value

is still below exchange coefticient values measured for bare Bi,,6Er0.403 and Ceo.sYo.,0,.95 electrolyte surfaces. In fact, the magnitude of the exchange coefficient for these electrolytes is not influenced by the presence of noble metal (Pt, Au, Ag ) electrodes [ 1O], implying that the electrolyte surface is catalytically active for the dissociative adsorption of oxygen molecules. To increase the oxygen exchange kinetics for zirconia-based electrolytes it is necessary to use oxide electro-catalysts. Data for bulk Lao.sSro.sMn03 samples are included in Fig. 7 and attention is drawn to the fact that the magnitude of the surface exchange coefficient is comparable with values obtained for Bi,.6Er0.403 and Ceo.sYo.r0,.95 ceramic electrolytes. However, when fine particles of this oxide are dispersed on zirconia electrolyte surfaces to produce porous electrode structures incorporating a high proportion of three-phase boundary regions, then the exchange coefficient increases to around 1Om6cm s- ’ at 800°C and 10e5 cm s- ’ at 950 “C. However, at 800°C the value of k is about an order of magnitude too low and significant cathodic polarisation occurs at this temperature for high current densities.

B.C.H. Steele /Solid Statelonics 75 (1995) 157-165

Ceramic samples of the mixed conductor La,,8Sr0.2C003_, exhibit encouraging values for k of about lo-’ cm s-’ above 800°C and when these are used in conjunction with zirconia electrolyte, the very high current densities can be sustained with minimal cathodic overpotential [ 2 11. Whilst excellent electro-catalytic performance can be obtained in the laboratory for La0.8Sr0.2Co03_-xfor small samples, it is difficult to reproduce this behaviour for large technological samples because the processing routes usually produce high-resistance interfacial reaction products, e.g. La2Zr207, SrZrO,, and the large thermal expansion mismatch between the zirconia electrolyte and the electrode produces cracking and spalling when the electrolyte/electrode structure is cycled. It was emphasized earlier (Section 3.4. ) that ceriabased electrolytes such as Ceo.sGdo.,O,.ss could be incorporated in electrochemical systems designed to operate at intermediate temperatures (450-750°C). Ceria-based electrolytes not only exhibit higher oxygen-ion conductivities under these conditions but possess further advantages. For example, ceria solid solutions typically exhibit thermal expansion coefficients around 12.5 x lob6 K-’ (approximately 20% higher than that of Zr0.ssY0,150,.925).These values are compatible with ferritic stainless metallic components, and also selected La-Co-O electrode compositions, e.g. Lao.sSro.4Coo,2Feo.80~_~(thermal expansion value 14 x 10e6 K- ’ ). Moreover these electrode compositions appear also to be chemically stable in contact with ceria-based electrolytes [22] as the pyrochlore compound LazCe,O, does not exist. Although the incorporation of Fe3+ lowers the concentration of anion vacancies, samples of Lao.sSro.4Coo.2Feo.803_~still exhibit relatively high values of D* and k as shown in Fig. 8. As expected, therefore, electrode resistances associated with Ce,,Gd,, 0, .ss/Lao.sSr,.,Coo.,Feo.803 structures are extremely low as shown in Fig. 9. These values were obtained with symmetrical electrodes [ 231 using impedance spectroscopy, and it is evident that the electrode resistivities are lower than the target figure ofO.l5Rcm*at lOOOK. 4.3. Surface exchange mechanisms The assumption of a relationship between k and j, and thus RE has been useful in developing alternative

163

cathode materials such as Lao.sSro.4Coo.2Feo.sO~. However these phenomenological relationships provide little information about the mechanisms of surface exchange processes and the associated oxygen electrode reactions. For Zr(Y)Oz_, electrolytes the nature of the metallic electrode influences the value measured for the electrode resistance. Silver provides the best electrode kinetic behaviour followed by platinum, then gold. The isotopic exchange data for bare zirconia electrolyte surfaces (Fig. 7) indicate very slow kinetics and noble metals are required for the initial dissociative adsorption step. The slow kinetics for ZrOz electrolyte surfaces are undoubtedly associated with the relatively small concentration of electronic charge carriers in this material [ 23 1. It is well known [ 2426] that the electrode kinetics of Zr02 electrolytes can be considerably improved by the incorporation of catalytically active cations in the near-surface region and presumably this enhanced activity is associated with an increase in the concentration of electronic charge carriers. Less attention has been given to CeOz-based electrolytes, and available data are confusing probably due to the presence of segregated impurities at interfaces. It should be noted, however, that the isotopic exchange kinetics for CeOa electrolytes are relatively fast (Fig. 7) probably due to the ability of the Ce4+/ Ce3+ redox reaction to provide sufficient electronic charge carriers. It is probable, therefore, that the type of electrode material will have less influence on the electrode resistance than in the case for Zr02 electrolytes. It is also well known that ceria particles used in three-way automative catalysts function as an oxygen reservoir [ 27,281, and participation of the lattice oxygen in oxide catalysts is well documented in the Max-Van Krevelen mechanism [ 28 ] for partial oxidation reactions. A heuristic model [ 31 appears to be emerging in which the surface oxygen exchange process, and probably the electrolyte/electrode resistivities, are controlled by a reservoir of electroactive oxygen species. This oxygen storage capacity comprises a near-surface region and so both diffusive (D) and surface exchange (k) processes are required to provide the electroactive oxygen species for the Faradaic reaction. To improve the relevant electrode kinetics it will be necessary to elucidate how to influence the

B.C.H. Steele /Solid State Ionics 75 (1995) 157-165

164

Fig. 8. Values of D and k for Lao&a0 ,Fe,&oa20,

3.5

as a function of reciprocal temperature.

1

0.7

0.8

0.9

1

1.2

1.1

1.3

1.4

1.5

IOOOrr (l/K)

Fig. 9. Values of electrode resistivity as a function of reciprocal temperature for ~.6Sr~,4Feo.sCo~.zO~_a/Zr~.85Yo.,501.925 and ~.sC~.4Feo.8Coo.z0,-a/Cc~,~G~.,0,.~~ cathode/electrolyte assemblies.

rate of supply reservoir.

of electroactive

species

from

the

The author wishes to thank present colleagues in the Centre for Technical Ceramic for many useful discussions and for providing experimental data. Financial support from the Science and Engineering Research Council, Commission of the European Communities, Club of Industrial Affiliates for the development of Ceramic Electrochemical Reactors and NED0 International Joint Research Grant is also gratefully acknowledged.

References [l] B.C.H. Steele, Mater. Sci. Eng. B13 (1992) 79. [2] B.C.H. Steele, in: High Temperature Electrochemical Behaviour of Fast Ion and Mixed Conductors, ed. F.W. Paulson, J.J. Bentzen, T. Jacobsen, E. Skou and M.J.L. 0stergArd (Risa National Laboratory, Denmark, 1993) p. 423. [3] B.C.H. Steele, Proc. 3rd Grove Fuel Cell Symposium, 28 Oct.-l Sept. 1993, J. Power Sources 49 (1994) 1. [4] M. Kleitz, T. Kloidt and L. Dessemond, in: High Temperature Electrochemical Behaviour of Fast Ion and Mixed Conductors, ed. F.W. Pot&en, J.J. Bentzen, T. Jacobsen, E. Skou and M.J.L. 0sterg;ird (Rise National Laboratory, Denmark, 1993) p. 89.

B.C.H. Steele I Solid State Ionics 75 (1995) 157-165

[ 51 K.J. Vetter, Electrochemical Kinetics (Academic Press, New York, 1967) p. 339. [6] Y. Teraoka, H.M. Zhang, K. Okamoto and N. Yamazoe, Mater. Res. Bull. 23 ( 1988) 5 1. [7] T.J. Mazenec, T.L. Cableand J.G. Frye Jr., Solid State Ionics 53-56(1992)111. [ 81 M. Liu, in: Ionic and Mixed Conducting Ceramics, Vol. 9 I12, ed. R.A. Ramanarayan and H.L. Tuller (Electrochemical Society, New Jersey, 199 I ) p. 95. [9] C. Wagner, Z. Phys. Chem. B21 (1933) 25. ] 10) B.C.H. Steele, J.A. Kilner, P.F. Dennis, A. McHale. M. Van Hemert and A.J. Burggraaf, Solid State Ionics 18/ 19 ( 1986) 1038. [ 111J.A. Kilner, L. Ilkov and B.C.H. Steele, Solid State Ionics 12 (1984) 89. [ 12 ] B.A. Boukamp, I.C. Vinke, J.J. DeVries and A.J. Burggraaf, Solid State Ionics 32/33 (1989) 918. [ 131 R.J. Chater, S. Carter, J.A. Kilner and B.C.H. Steele, Solid State Ionics 53-56 (1992) 859. [ 141 H.J.M. Bouwmeester, H. Kruidhof and A.J. Burggraaf, Solid State Ionics 72 ( 1994) 185. [IS] J.A. Kilner, in: 2nd Int. Symp. on Ionic and Mixed Conducting Ceramics, eds. T.A. Ramanarayanan, W.L. Worrell and H.L. Tuller, Vol. 94-12 (Electrochemical Society, New Jersey, 1994) p. 174. [ 161 R.M. Thorogood, R. Srinivasan, T.F. Yee and M.P. Drake, US Patent 5,240,480 (31 August 1993).

16.5

[ 171 M. Dennis-Dumelie, G. Nowogrocki and J.C. Boivin, Br. Ceram. Proc. 43 ( 1988) 15 1. [ IS] E. Ivers-Tiffee, W. Wersing, M. Schiessl and H. Greiner, Ber. Bunsenges. Phys. Chem. 94 ( 1990) 978. [ 191 E.Kh Kurumchin and M.V. Perfiiiev, Solid State Ionics 42 (1990) 129. [20] S. Carter, A. Selcuk, R.J. Chater, J. Kajda, J.A. Kilner and B.C.H. Steele, Solid State Ionics 53/56 ( 1992) 597. [21] Y. Takeda, R. Kanno, M. Noda, Y. Tomida and 0. Yamamoto, J. Electrochem. Sot. 134 (1987) 2656. [22] C.C. Chen, M.M. Nasrallah and H.U. Anderson, in: Proc. 3rd Int. Symp. on Solid Oxide Fuel Cells, ed. S.C. Singhal and H. Iwahara, Vol. 93-4 (Electrochemical Society, New Jersey, 1993) p. 252. [23] B.C.H. Steele, in: Proc. 3rd Grove Fuel Cell Symp., J. Power Sources49 (1994) l-14. [24] H.H. Mobius and B. Rohland, US Patent No. 3,377,203 (9 April 1968). [ 251 H. Tannenberger and P. Kovacs, German Patent DE 177 I 829 B2 (18 July 1968). [26] B.A. Van Hassel and A.J. Burggraaf, Appl. Phys. A49 ( 1989) 33. [27]G.S.ZafirisandR.T.Gorte, J.Catal. 139 (1993) 561. [28] B.C. Gates, J.R. Katzer and G.C.A. Schuit, The Chemistry of Catalytic Processes (McGraw Hill, New York, 1979) p. 344.