ferrite-based perovskites as cathode materials for anode-supported solid oxide fuel cells: part ii....
TRANSCRIPT
w.elsevier.com/locate/ssi
Solid State Ionics 177 (2
Ferrite-based perovskites as cathode materials for anode-supported
solid oxide fuel cells
Part II. Influence of the CGO interlayer
Andreas Mai *, Vincent A.C. Haanappel, Frank Tietz, Detlev Stover
Institute for Materials and Processes in Energy Systems, Forschungszentrum Julich, 52425 Julich, Germany
Received 31 August 2005; accepted 8 December 2005
Abstract
It was shown in part I (Variation of the cathode composition) [A. Mai, V.A.C. Haanappel, S. Uhlenbruck, F. Tietz, D. Stover, Solid State Ionics
176 (2005), 1341], that an interlayer is needed between yttria-stabilised zirconia (YSZ) electrolytes and (La,Sr)(Co,Fe)O3�d (LSCF) cathodes in
order to prevent undesired chemical reactions between these materials. This interlayer makes it possible to benefit from the superior
electrochemical properties of the LSCF perovskites in solid oxide fuel cells (SOFCs). In this study the influence of a Ce0.8Gd0.2O2�d (CGO)
interlayer on the electrochemical performance of LSCF-type SOFCs is investigated in more detail. For screen-printed and sintered interlayers the
grain size as well as the sintering temperature affected the electrochemical performance. The use of a fine powder with a mean particle size of
d50=0.2 Am and sintered at 1250 -C resulted in the best performance. Furthermore, reactive sputtering resulted in dense CGO interlayers at low
deposition temperatures, which led to improved properties regarding diffusion inhibition and electrochemical performance. SOFCs with sputtered
interlayers gave power densities of up to 0.9 W/cm2 at 700 -C and 0.7 V, with H2+3% H2O as fuel gas (approx. 10% fuel utilisation).
D 2006 Elsevier B.V. All rights reserved.
Keywords: Solid oxide fuel cells; LSCF; Cerium oxide; Interlayer
1. Introduction
La1�x�zSrxCo1�yFeyO3�d (LSCF) perovskites are currently
used as cathode materials for solid oxide fuel cells (SOFCs) due
to their high electrocatalytic activity. It was shown in the first
part of this series of publications [1] that SOFCs with
La0.58Sr0.4Co0.2Fe0.8O3�d cathodes delivered about twice the
power density than SOFCs with lanthanum manganite based
(LSM/YSZ-composite) cathodes. Since the LSCF perovskites
are chemically incompatible with the Zr0.85Y0.15O1.925 (8YSZ)
electrolyte forming insulating phases at high temperatures the
use of an interlayer between cathode and electrolyte is
necessary [1]. A widely used material for this interlayer is
substituted ceria [2], for example Ce0.8Gd0.2O2�d (CGO). This
interlayer should be as dense as possible for maximum
efficiency as a diffusion barrier layer and for maximum
conductivity. To reduce the porosity of sintered interlayers, a
0167-2738/$ - see front matter D 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.ssi.2005.12.010
* Corresponding author. Tel.: +49 2461 61 3071; fax: +49 2461 61 2455.
E-mail address: [email protected] (A. Mai).
higher sintering temperature is feasible, but this will increase
the formation of a solid solution between YSZ and CGO at the
interface, having a lower ionic conductivity [3,4] than both
YSZ and CGO. As a result of these two contradicting influences
on the performance, most interlayers presented in literature are
porous [5–7]. This paper concentrates on the influence of the
processing parameters and the resulting properties of the
interlayer on the electrochemical performance of the SOFCs.
2. Experimental
The ceria compounds used in this study were commercially
available (Ce,Gd)O2�d powders manufactured by Treibacher,
Austria and Rhodia Electronics and Catalysis, France. The
powders were ground by ball milling. NiO/YSZ anode
substrates (50�50 mm2) were used for fabrication of the
SOFCs. The substrates were co-fired with an anode functional
layer (Ni/8YSZ) and an electrolyte (8YSZ) at 1400 -C for 5 h.
The above-mentioned CGO powders were screen-printed and
sintered for 3 h on top of the electrolytes, resulting in layer
thicknesses of approximately 7 Am. The varied processing
006) 2103 – 2107
ww
a)
b)
c)
d) 10 m
5 m
5 m
5 m
Fig. 1. CGO interlayers sintered at 1300 -C for 3 h, made from different
powders: a) CGO from Treibacher Auermet ground to a d50 of 0.9 Am, b)
ground to a d50 of 0.22 Am, c) CGO from Rhodia ground to a d50 of 0.4 Am, d)
CGO mixed with cobalt oxide as sintering aid (note the different magnifica-
tion), with partial delamination of the interlayer.
Fig. 2. Current–voltage curves of SOFCs at 750 -C with La0.58Sr0.4Co0.2Fe0.8O3�d cathodes (cathode type I) and different interlayers: layers made from coarse
(d50=0.9 Am) and fine (d50=0.2Am) CGO and CGOwith cobalt oxide as sintering
aid, all sintered at 1300 -C, as well as a layer made by reactive sputtering.
A. Mai et al. / Solid State Ionics 177 (2006) 2103–21072104
parameters were the grain size of the powders and the sintering
temperature. Additionally, cobalt oxide was used as a sintering
aid for the interlayer. CGO interlayers were also deposited by
reactive sputtering similar to the deposition of 8YSZ described
by Wanzenberg et al. [8]. For this, metallic sputtering targets
were used, while the sample temperature and the oxygen partial
pressure in the reactor chamber were optimised, resulting in
dense layers with thicknesses of 3 to 5 Am. On top of the
interlayers, La0.58Sr0.4Co0.2Fe0.8O3�d cathodes were deposited,
as described in part I [1]. The interlayers were analysed by SEM
scanning electron microscopy (LEO 1530 (Gemini), Zeiss
Ultra55) equipped with an energy dispersive X-ray (EDX)
analysis system. Electrochemical measurements of the cell
performance and their analysis were conducted as described in
Ref. [1]. For some of the SOFCs the nickel oxide was reduced to
nickel with a different anode reduction procedure [9] (named
hereafter ‘‘reduction procedure II’’), while for the others, the
same procedure as in the first part [1] was used (‘‘reduction
procedure type I’’).
3. Results and discussion
3.1. Variation of the powder characteristics (particle size,
supplier)
Fig. 1a–c show CGO interlayers on YSZ electrolytes
sintered at 1300 -C, made with three different powders: CGO
from Treibacher, ground to an average particle size (d50) of 0.9
Am (Fig. 1a), to a d50 of 0.2 Am (Fig. 1b) and CGO from Rhodia
(Fig. 1c) ground to a d50 of 0.4 Am. According to the
expectation, the layer made from the coarser powder (Fig. 1a)
shows a coarser microstructure. The two layers made from the
finer powders (Fig. 1b and c), but received from different
suppliers, have a similar microstructure. All three pictures show
powder particles that are very well sintered together, providing
well-established conductivity paths. Fig. 2 shows I–V-curves
measured at 750 -C of SOFCs made with the coarse (d50=0.9
Am) and the fine (0.2 Am) powders (both originally from
Treibacher). Table 1 lists the corresponding current densities
and area specific resistances in the first two rows. The interlayer
made with the finer powder results in a slightly higher
performance of the SOFC in comparison to the SOFC with
Table 1
Current densities (A/cm2, at 700 mV) and area-specific resistances (mVIcm2) of SOFCs with La0.58Sr0.4Co0.2Fe0.8O3�d cathodes and various CGO interlayers
Temperature
(-C)
Variation of the powder properties
(sintering at 1300 -C)
Layer applied by
sputtering deposition
Variation of the sintering temperature (d50=0.2 Am)
d50=0.9 Am d50=0.2 Am With cobalt-oxide 1200 -C 1250 -C 1275 -C 1300 -C 1325 -C
Anode reduction type I Anode reduction type II
Current density (A/cm2, 700 mV)
800 1.38*+ 1.76*T0.08 1.40*T0.04 1.81*T0.08 2.10*T0.13 2.30*T0.07 2.20*T0.15 2.11*T0.05 1.95*+
750 1.13+ 1.43*T0.08 1.15T0.01 1.68*T0.03 1.59*T0.11 1.80*T0.06 1.73*T0.13 1.65*T0.05 1.56*+
700 0.78+ 0.99T0.06 0.80T0.05 1.30*T0.01 0.99T0.08 1.20T0.05 1.10T0.07 1.13T0.04 0.99+
650 0.48+ 0.58T0.03 0.46T0.03 0.86T0.01 0.47T0.07 0.65T0.01 0.60T0.03 0.58T0.02 0.60+
Area-specific resistance (mXIcm2)
800 195+ 179 222 165 150 136 147 148 163+
750 239+ 219 271 173 189 167 173 186 200+
700 326+ 297 380 196 267 215 232 244 277+
650 531+ 522 707 271 531 427 473 480 526+
+Results from only one cell measurement;
*Current densities above 1.25 A/cm2 were linearly extrapolated to 0.7 V from the last three measurement points (see Fig. 3).
A. Mai et al. / Solid State Ionics 177 (2006) 2103–2107 2105
the coarser interlayer. This is probably the effect of a better
contact between the CGO particles due to the higher sintering
activity of the finer particles.
3.2. Cobalt oxide as sintering aid
To achieve a better densification of the CGO interlayer
during sintering, cobalt oxide as sintering aid for CGO, as
proposed by Kleinlogel et al. [10], was used. Therefore, fine
CGO powder (d50=0.2 Am) was impregnated with a cobalt-
nitrate (Co(NO3)2I6H2O, Sigma-Aldrich) solution and dried at
100 -C. The amount of cobalt-nitrate solution was chosen to
result in 2 wt% of cobalt oxide. The impregnated powder was
then processed to a paste and screen-printed on top of the
electrolyte. The nitrate decomposes during the sintering process,
which leads to a thin layer of cobalt oxide on the CGO grains.
As shown in Fig. 1d, the cobalt-containing interlayer has a
significantly denser structure than the corresponding interlayer
without cobalt (Fig. 1b). However, a large number of cracks have
evolved resulting in a partial delamination of the interlayer.
Lower sintering temperatures (1200 -C) resulted in a slightly
higher porosity, but cracks were still present. The cracking is
probably caused by stresses due to the increased shrinkage of the
interlayer on top of the non-shrinking, nearly dense electrolyte
(sintered at 1400 -C, before). This mismatch could explain, why
it is possible to get dense, crack-free pellets at sintering
temperatures as low as 1000 -C [10], while it is much more
complicated to get dense, crack-free CGO interlayers on pre-
sintered 8YSZ. Fig. 2 and Table 1 show the electrochemical data
frommeasurements of single cells with a cobalt-containing CGO
interlayer sintered at 1300 -C. The performance is lower than for
the corresponding cells without sintering aid, which is certainly
due to the presence of cracks in the interlayer.
3.3. Interlayers applied by reactive sputtering
As mentioned above, it would be desirable to deposit a
dense CGO diffusion barrier layer on the 8YSZ electrolyte and
to minimize the formation of a solid solution between the two
materials at the interface. The maximum temperature without
interaction at the interface is 1200 -C [11]. A possibility to
achieve dense layers with thicknesses of up to 5 Am at a sample
temperature of 700 -C within a few hours is the deposition of
CGO by reactive sputtering, as described for 8YSZ in Ref. [8].
Fig. 3a shows such a CGO layer with a columnar structure.
After sintering the cathode on top of this layer, small pores
have evolved in the layer due to sintering processes (Fig. 3b) as
it was also found in YSZ layers [8]. Despite these pores, no
strontium enrichment could be detected at the interface (Fig.
3c). This is in contrast to investigations of SOFCs with sintered
CGO layers, where the formation of SrZrO3 at the interface
CGO/8YSZ was clearly visible (see Fig. 4b of part I [1] and
Ref. [12]) after sintering of the cathode, being the result of Sr
diffusion through the interlayer. The results of the electro-
chemical characterisation (Fig. 2) clearly show an improvement
of the electrochemical performance for cells with a sputtered
interlayer in comparison to cells with sintered interlayers. To
estimate the effect of the decreased porosity on the area specific
resistance (ASR) of the cell, the area specific resistance of a
porous layer and of a dense layer was estimated. For a porous
layer as shown in Fig. 1b, this results in 41 mV cm2 at 700 -C(with a porosity of 33%, measured by image analysis, an
estimated tortuosity of 2, and a thickness of 6.5 Am), while a
dense layer as shown in Fig. 3a would have an ASR of 8.5 mV
cm2 at 700 -C (thickness 4 Am, bulk conductivity of
Ce0.8Gd0.2O2�d: 0.047 S/cm at 700 -C [13]). The difference
between these two calculated ASRs (32 mV cm2) is much
smaller than the difference (99 mV cm2) between the measured
ASRs of the corresponding cells (297 and 196 mV cm2,
respectively). The reason for the improved performance of the
cell with the sputtered CGO is therefore probably a decreased
resistance of the interface CGO-YSZ which is caused on the
one hand by less SrZrO3 formation due to an improved Sr-
blocking of the dense CGO layer and on the other hand by a
minimization of the formation of a solid solution between the
two materials at the interface.
A. Mai et al. / Solid State Ionics 177 (2006) 2103–21072106
3.4. Optimization of the sintering temperature
Together with the aforementioned experiments with sintered
interlayers, three different sintering temperatures were tested:
1200, 1300 and 1350 -C (all for 3 h, with 1300 -C resulting in
the best performance). Due to the large influence found, the
sintering temperature of the interlayer was varied again with a
smaller step size. For these SOFCs, an optimised anode
reduction procedure [9] was followed (reduction procedure
a)
b)
c)
Sr Ce Zr
1 m
1 m
Fig. 3. CGO interlayers deposited by reactive sputtering deposition, a) layer
with columnar structure directly after deposition, b) similar layer after sintering
of the cathode (1080 -C, 3 h) on top of the interlayer with c) EDX line scans for
strontium, cerium and zirconium taken along the black line.
Fig. 4. a) Current densities (A/cm2) at 700 mV of SOFCs CGO (d50=0.2 Aminterlayers sintered at various temperatures (Current densities larger than
1.25 A/cm2 are calculated from a linear approximation of the linear part in the
I–V-curves to 0.7 V, as illustrated in the I–V-curves b) Current–voltage curves
of one of the SOFCs with a CGO interlayer sintered at 1250 -C.
)
type II). Therefore, a direct comparison with the electrochem-
ical results shown above is difficult. Fig. 4a shows the current
densities measured at 0.7 V with cells containing CGO
interlayers (d50=0.3 Am, Treibacher) that were sintered at
temperatures from 1200 to 1325 -C for 3 h. The highest power
densities were obtained for the cells with a sintering
temperature of the interlayer of 1250 -C. Fig. 4b shows
exemplarily the corresponding I–V-curves for one of these
cells. The decrease of the power density both with sintering
temperatures above 1250 -C is probably the result of the
increased formation of the solid solution between YSZ and
CGO, while the decrease of the power density at lower
sintering temperatures is due to the higher porosity and lower
adherence of the interlayer.
4. Conclusions
The investigations have shown that the structural properties
of the ceria interlayer strongly influence the electrochemical
performance of SOFCs with LSCF cathodes. Several
approaches were made to maximise the density of the interlayer
and at the same time minimize the formation of a solid solution
A. Mai et al. / Solid State Ionics 177 (2006) 2103–2107 2107
between CGO and 8YSZ. The experiments have shown that a
fine powder (d50=0.2 Am) leads to a better performance and
that for these layers 1250 -C is the optimum sintering
temperature. It was confirmed that cobalt oxide acts as a
sintering aid for CGO, but cracks evolved, probably due to the
high shrinkage of the layer relative to the underlying dense
electrolyte. It was further shown that reactive sputtering is a
method to deposit dense CGO layers on the electrolyte,
resulting in an improved performance of the SOFCs. An
estimation on the influence of the layers’ porosity on the ASR
leads to the conclusion, that the improved performance is not
only a result of an increase in conductivity of the layer due to a
lower porosity, but mainly a result of the improved buffering
properties of a dense CGO layer, inhibiting SrZrO3 formation,
and avoiding the formation of a solid solution between CGO
and 8YSZ due to the low deposition temperature. The tested
SOFCs reached power densities of up to 0.9 W/cm2 at 700 -C.
Acknowledgements
The authors thank the staff, namely G. Blah, M. Kampel, S.
Heinz, and G. Klein of the IWV department at Forschungszen-
trum Julich for processing the anode substrates and electrolyte
layers, for the SEM investigations, D. Sebold and for
performing the electrochemical measurements, C. Tropartz,
B. Rowekamp, and H. Wesemeyer. Thanks are also given to P.
Panjan (Institute Jo”ef Stefan, Ljubijana) for reactive sputter-
ing. The work was carried out in the networking project
‘‘Renewable Energies’’ under contract no. 01SF0039 and
financial support from the German Federal Ministry of Science
and Education (BMBF) is gratefully acknowledged.
References
[1] A. Mai, V.A.C. Haanappel, S. Uhlenbruck, F. Tietz, D. Stover, Solid State
Ionics 176 (2005) 1341.
[2] H. Uchida, S. Arisaka, M. Watanabe, Electrochem. Solid-State Lett. 2
(1999) 428.
[3] A. Tsoga, A. Gupta, A. Naoumidis, P. Nikolopoulos, Acta Mater. 48
(2000) 4709.
[4] X.-D. Zhou, B. Scarfino, H.U. Anderson, Solid State Ionics 175 (2004)
19.
[5] S.P. Simner, J.F. Bonnett, N.L. Canfield, K.D. Meinhardt, J.P. Shelton,
V.L. Sprenkle, J.W. Stevenson, J. Power Sources 113 (2003) 1.
[6] T.L. Nguyen, K. Kobayashi, T. Honda, et al., Solid State Ionics 174
(2004) 163.
[7] M. Shiono, K. Kobayashi, T. Lan Nguyen, K. Hosoda, T. Kato, K. Ota, M.
Dokiya, Solid State Ionics 170 (2004) 1.
[8] E. Wanzenberg, F. Tietz, P. Panjan, D. Stover, Solid State Ionics 159
(2003) 1.
[9] V.A.C. Haanappel, A. Mai, J. Mertens, Solid State Ionics (in press).
[10] C. Kleinlogel, L.J. Gauckler, Solid State Ionics 135 (2000) 567.
[11] A. Tsoga, A. Naoumidis, A. Gupta, D. Stover, in: Materials Science
Forum, vol. 308–311, Trans Tech Publications, Switzerland, 1999,
pp. 794–799.
[12] A. Mai, M. Becker, W. Assenmacher, F. Tietz, E. Ivers-Tiffee, D. Stover,
W. Mader, Solid State Ionics (submitted for publication).
[13] B.C.H. Steele, Solid State Ionics 129 (2000) 95.