Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells

Abstract

The conventional solid oxide cell is based on a Ni–YSZ support layer, placed on the fuel side of the cell, also known as the anode supported SOFC. An alternative design, based on a support of porous 3YSZ (3 mol.% Y2O3–doped ZrO2), placed on the oxygen electrode side of the cell, is proposed. Electronic conductivity in the 3YSZ support is obtained post sintering by infiltrating LSC (La0.6Sr0.4Co1.05O3). The potential advantages of the proposed design is a strong cell, due to the base of a strong ceramic material (3YSZ is a partially stabilized zirconia), and that the LSC infiltration of the support can be done simultaneously with forming the oxygen electrode, since some of the best performing oxygen electrodes are based on infiltrated LSC. The potential of the proposed structure was investigated by testing the mechanical and electrical properties of the support layer. Comparable strength properties to the conventional Ni/YSZ support were seen, and sufficient and fairly stable conductivity of LSC infiltrated 3YSZ was observed. The conductivity of 8–15 S cm–1 at 850 °C seen for over 600 h, corresponds to a serial resistance of less than 3.5 m Ω cm2 of a 300 μm thick support layer.

Full text

Development of a Novel Ceramic
SupportLayerforPlanarSolidOxide
Cells
T. Klemensø
1
*, D. Boccaccini
1
,K.Brodersen
1
,H.L.Frandsen
1
,P.V.Hendriksen
1
1
Department of Energy Conversion and Storage, Technical University ofDenmark, DK-4000 Roskilde, Denmark
Received May 31, 2013; accepted Dec ember 13, 2013; published online 䊏䊏䊏
1Introduction
Solid oxide cells are high temperature electrochemical
devices that can either be operated as solid oxide fuel cells
(SOFC) converting fuel to electricity, or operated as solid
oxide electrolysis cell (SOEC) converting electricity and H
2
O
and/or CO
2
to fuel (hydrogen or synthesis gas).
The electrochemically active layers (anode, cathode, elec-
trolyte) of the state of the art cell designs are thin, in the range
of 10 lm, which facilitates high power density at the opera-
tional temperature [1]. However, the thin active layers also
mean that the cell cannot support itself mechanically, and a
structural support layer is needed. The SOFC and SOEC tech-
nologies have for many years been focused on the Ni/YSZ
supported cell design. This design consists of a thick Ni/YSZ
support, a thin Ni/YSZ fuel electrode, a thin YSZ electrolyte,
and a thin YSZ/lanthanum strontium manganite (LSM) oxy-
gen electrode. In recent years, the LSM/YSZ oxygen electrode
is often exchanged with a mixed ionic and electronic conduc-
tor to increase performance and to lower the operational tem-
perature further. Some of the most active oxygen electrode
materials are Sr- and Fe-doped LaCoO
3
. However, these
materials are chemically reactive with the ZrO
2
-based electro-
lyte, and a barrier layer of doped ceria (e.g., Gd doped (GDC)
or Sm doped (SDC)) is needed next to the electrolyte to
ensure electrochemical reaction points (triple phase bound-
ary). The prevalent Ni/YSZ supported cell design is sketched
in Figure 1a (based on [2]).
Over the past years, several alternative supports and cell
designs have been investigated, since the Ni/YSZ support
suffers from drawbacks such as high raw material costs and
low redox tolerance [3]. The requirements to the cell support
component are: (i) to provide mechanical strength to the cell,
(ii) to be sufficiently conductive to transport the current from
cell to external system, and (iii) to be porous and provide suf-
ficient gas transport to the active layers. Among the alterna-
tive designs currently pursued by cell developers are metal
supported cells, LSM (oxygen electrode) supported cells, and
porous YSZ supported cells.
In the metal supported cells, the Ni/YSZ support is substi-
tuted with a porous stainless steel layer, which is potentially
cheaper, and possibly has advantageous mechanical proper-
ties. However, the introduction of metal introduces new chal-
Abstract
The conventional solid oxide cell is based on a Ni–YSZ sup-
port layer, placed on the fuel side of the cell, also known as
the anode supported SOFC. An alternative design, based on
a support of porous 3YSZ (3 mol.% Y
2
O
3
–doped ZrO
2
),
placed on the oxygen electrode side of the cell, is proposed.
Electronic conductivity in the 3YSZ support is obtained post
sintering by infiltrating LSC (La
0.6
Sr
0.4
Co
1.05
O
3
). The poten-
tial advantages of the proposed design is a strong cell, due
to the base of a strong ceramic material (3YSZ is a partially
stabilized zirconia), and that the LSC infiltration of the sup-
port can be done simultaneously with forming the oxygen
electrode, since some of the best performing oxygen electro-
des are based on infiltrated LSC. The potential of the pro-
posed structure was investigated by testing the mechanical
and electrical properties of the support layer. Comparable
strength properties to the conventional Ni/YSZ support
were seen, and sufficient and fairly stable conductivity of
LSC infiltrated 3YSZ was observed. The conductivity of 8–
15 S cm
–1
at 850 °C seen for over 600 h, corresponds to a
serial resistance of less than 3.5 m Xcm
2
of a 300 lm thick
support layer.
Keywords: Ceramics, Infiltration, Nanostructures, Oxygen
Electrode Supported Cell, Porous Materials, Solid Oxide
Fuel Cell
–
[
*
]Corresponding author, trkl@dtu.dk
FUEL CELLS 00, 0000, No. 0, 1–9 © 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1
ORIGINAL RESEARCH PAPER
DOI: 10.1002/fuce.201300121

Klemensø et al.: Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells
lenges with respect to incompatibility with the traditional
electrode materials, as well as the cell fabrication route [4].
In the LSM (oxygen electrode) supported cell, the support
is placed on the oxygen side of the cell as illustrated in Fig-
ure 1b. The support in this design is typically based on a thick
LSM/YSZ layer, but supports containing materials such as
Pr
0.35
Nd
0.35
Sr
0.3
MnO
3–x
(PNSM) have also been investigated
[3, 5]. The LSM supported cell has the advantages of poten-
tially cheaper raw material (LSM), redox stability and
improved durability [3]. Also, it has been suggested by Ni et
al. [6] that gas diffusion limitations at high current densities
in SOEC mode, can be reduced by using an oxygen electrode
supported cell design. The fabrication of a LSM supported
cell is complex due to the co-sintering of the LSM/YSZ sup-
port and the electrolyte. To obtain a dense YSZ electrolyte,
typically sintering temperatures above 1,200 °C are needed
[3]. However, already above 1,000 °C reaction between LSM
and YSZ into insulating zirconates is seen to occur and have
detrimental effect on the electrochemical performance [7]. To
facilitate a gas tight electrolyte and minimize the interfacial
LSM/YSZ reaction, the co-sintering of LSM/YSZ supported
cells has been restricted to the sintering range 1,200–1,250 °C.
However, when co-sintering in the lower temperature range,
less gas tight cells were reported [5], and when co-sintering
in the higher temperature range, lower performance were
reported due to zirconate formation, LSM coarsening, and
low porosity [3, 5]. To avoid the high-temperature co-sinter-
ing step, alternative techniques, such as spin coating the YSZ
electrolyte layer, has also been attempted [8], but the power
density of the LSM supported cells is so far not competitive
to the Ni/YSZ supported [3, 9].
Designs based on a co-sintered porous support of YSZ into
which a percolating electronic conducting phase is intro-
duced post sintering, have also been investigated by some
groups [10–12]. In these designs, the support consists of a
porous 8YSZ backbone layer on the fuel side, into which Ni is
infiltrated, as described in Refs.
[10–12]. The half cell design is
sketched in Figure 1c, without
inclusion of the oxygen electrode
which is not specified in the refer-
ences. It has also been suggested
that the electronic percolation in
such structures could be intro-
duced by impregnation of Ni/
La
0.75
Sr
0.25
Cr
0.5
Fe
0.5
O
3–d
[13], or
by electrodeposition of, e.g., Cu
[14].
The function of the 8YSZ back-
bone supported cell design relies
on that enough and stable elec-
tronic percolation can be obtained
in the support. The infiltrated
(and nano sized) Ni is known to
be prone to coarsening, and high
degradation rates have been ob-
served, especially when the infiltrated Ni also acted as the
electrochemical active fuel electrode [15, 16]. This is a compli-
cating factor for the design sketched in Figure 1c.
A backbone supported cell design, where the backbone is
on the oxygen electrode side could also be imagined. Zhao et
al. [17] have reported on a design based on a thick porous
8YSZ support co-sintered with a thin porous YSZ layer and
the YSZ electrolyte. The co-sintered trilayer is screen-printed
with a CuO-SDC fuel electrode. The porous YSZ layers act as
support and oxygen electrode by post sintering being infil-
trated with LSF (La
0.8
Sr
0.2
FeO
3
). From the studies available in
literature, it is seems that the infiltrated structures relying on
electronic conduction through a perovskite phase (e.g., LSM
or LSF) are less prone to conductance degradation than the
structures relying on percolation through an infiltrated metal-
lic network as, e.g., Ni, though coarsening of the nano sized
perovskites is still an issue for the long-term performance of
the perovskite based electrodes, see, e.g., reviews [4, 18].
For the YSZ backbone based cell designs, infiltration with
LSC ((La,Sr)CoO
3
) in the support layer has not been consid-
ered. Infiltrated LSC has shown remarkable high perfor-
mance as oxygen electrode material when infiltrated into
backbones of doped ceria in combination with electrolytes of
doped ceria [19, 20]. However, when infiltrated into a YSZ
backbone and in combination with a YSZ electrolyte, the high
reactivity between LSC and YSZ results in detrimental zirco-
nate phases and performance loss [21, 22]. To be able to use
infiltrated LSC in YSZ electrode backbones, it was suggested
to infiltrate a dense doped ceria barrier layer as described in
Ref. [23], prior to infiltrating the LSC.
To facilitate the use of the highly active infiltrated LSC
electrode, together with the advantages of a co-sintered YSZ
backbone support, the authors suggest the cell design
sketched in Figure 1d. In this design, the porous YSZ back-
bone support is co-sintered with a porous doped ceria layer
(GDC), a doped ceria barrier layer, and the YSZ electrolyte at
Fig. 1 Sketch and comparison of four different cell configurations: (a) the conventional Ni/YSZ supported
cell [2], (b) the LSM supported cell as described in Ref. [3], (c) the 8YSZ backbone on the fuel side supported
cell [11], and (d) the suggested 3YSZ backbone on the oxygen side supported cell. The postfix (i) refers to
that the phase is being infiltrated post sintering.
ORIGINAL RESEARCH PAPER
2© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim FUEL CELLS 00, 0000, No. 0, 1–9
www.fuelcells.wiley-vch.de

Klemensø et al.: Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells
elevated temperatures (above 1,250 °C). Post sintering, the
porous support and porous GDC layers are infiltrated with
LSC, to form the active oxygen electrode (GDC/LSC) and to
form an electronic percolating network in the support (YSZ/
LSC). In this design, the reaction between the YSZ backbone
and the LSC is expected to be tolerable since the support only
acts as electronic conductor, and the YSZ/LSC interface does
not take part in the electrochemical reaction. In contrast, it is
speculated that the YSZ/LSC interfacial reaction could
improve the high-temperature stability of the infiltrated LSC.
One possible detrimental reaction that can occur in this
design is, similar to the conventional Ni/YSZ supported
design (cf. Figure 1a), the reaction between the YSZ and GDC
layer when co-sintering at high temperatures. GDC–YSZ
intermixing at the interface forms a lower ionic conducting
phase, which for a Ni/YSZ supported cell has been shown to
account for an increased cell resistance of 1.9 × 10
–1
Xcm
2
when sintering >1,300 °C [24]. The co-sintered cell design
should therefore preferably be co-sintered below 1,300 °C.
The proposed design may in addition have improved me-
chanical properties by using 3YSZ, as opposed to the earlier
reported 8YSZ, as the backbone material. The mechanical
properties of partially stabilized zirconia, such as 3YSZ,
include higher strength and especially introduces significant
toughness to the otherwise normally brittle ceramic materials.
The superior mechanical properties of 3YSZ compared to
other fuel cell related materials such as 8YSZ and doped
ceria, have been reviewed in Ref. [25], and the toughness
properties of 3YSZ have been extensively described and
reviewed in, e.g., [26–29]. Mechanical failures are detrimental
for fuel cells and stacks in operation, and it is therefore of
interest to have as strong and as tough as possible cells, to
improve the reliability during operation including tempera-
ture cycles [25].
In this paper, we show initial results to demonstrate the
potential of the proposed support layer. For this, we have fab-
ricated 3YSZ supports with open porosity ranging from ca.
1% to 54% and tested the mechanical properties by the ball on
ring method. The most infiltration suitable support structures
have been infiltrated with LSC, and the conductivity and sta-
bility have been examined by 4-terminal electrical measure-
ments for over 600 h at 850 °C. The first attempt on integrat-
ing the promising support structure into a half cell is also
reported.
2Experimental
Porous structures (backbones) of 3 mol.% yttria stabilized
zirconia (3YSZ) with variable open porosity (from <1% to
54%) were manufactured by tape casting. The tape casting
slurries were based on ethanol as solvent and contained in
addition to the 3YSZ powder (Tosoh Co.), poly(methyl meth-
acrylate) as pore former (PMMA, MR 7 G from Esprix ), and
an organic system based on [30], and the powders were used
as-delivered. In one tape, 10 wt.% of the 3YSZ was substi-
tuted with 10YSZ fibers from Zircar Zirconia, Inc. (Florida,
NY, USA). The layers were cast to a dried thickness of ca.
250 lm, and the layers were laminated to obtain thicker sam-
ples for handling and mechanical testing. Half cells were fab-
ricated by laminating the tape cast backbone layer with layers
of porous and dense Gd-doped ceria, and 8YSZ electrolyte
layer, also based on the organic system in Ref. [30] and pow-
ders from Tosoh Co. (doped zirconia) and Rhodia (doped
ceria).
Samples for mechanical testing were punched out of the
tapes and sintered at a temperature between 1,275 and
1,350 °C to a sintered diameter of ca. 25 mm and thickness ca.
300 lm. For each type of backbone, 29–40 samples were fabri-
cated and mechanically tested. The samples were tested with
the ball on ring method, and Weibull statistics were applied
to evaluate Weibull strength (r
0
) and the Weibull modulus
(m). In the set-up, the disc shaped specimens were supported
by a ring of balls and loaded by a piston through a ground
ball. The loading ball diameter was 2.78 mm and the ring di-
ameter 16 mm, see Ref. [31] for more details.
For conductivity measurements, sintered samples were cut
into 18 mm × 4 mm, and infiltrated with LSC40 precursor so-
lution based on water as solvent and metal-nitrates (1 M con-
centration), and La
0.6
Sr
0.4
Co
1.05
O
3
was subsequently formed
by heat treatment up to maximum 500 °C. The procedure and
all details were as described in Ref. [20]. The infiltrations
were repeated up to 12 times to obtain 10 vol.% LSC, where
percolation of the LSC40 phase can be expected to occur [12,
15]. The conductivity was tested on two samples: an infil-
trated sample where the sample surface was carefully pol-
ished before mounting in the conductivity set-up, to remove
excess LSC40 which was seen deposited onto the sample sur-
face, and a sample that was not polished.
A 4-electrode set-up was applied for the DC conductivity
measurements. Electrodes of platinum wire were mounted
onto the strip-shaped infiltrated sample by wrapping the wire
around the sample. The two current electrodes were placed at
the sample ends and platinum paste was applied to ensure
contact. The potential probes were positioned around the
middle of the sample with a separation of ca. 8–10 mm. The
conductivity was measured in air flow (50 mL min
–1
), at tem-
peratures up to 1,000 °C, and a durability study at 850 °C
was carried out. The resistance was logged with 5 min inter-
vals using a Keithley 580 lX m. The hardware defines a cur-
rent pulse that is enforced in both directions, and the poten-
tial between the probes is measured. The average resistance
for the opposite current directions was applied.
The microstructure of the sintered and tested samples was
characterized by scanning electron microscopy (SEM). Obser-
vations were made with a Hitachi TM1000 tabletop SEM, on
samples that were vacuum embedded in Epofix (Struers,
Denmark), ground and polished to 1 lm, and coated with
carbon to eliminate surface charging. The porosity and pore
size distribution were characterized by mercury intrusion
(Autopore IV 9500V1.05 from Micromeritics Instrument Cor-
poration, Norcross, GA).
ORIGINAL RESEARCH PAPER
FUEL CELLS 00, 0000, No. 0, 1–9 © 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 3
www.fuelcells.wiley-vch.de

Klemensø et al.: Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells
3 Results and Discussion
3.1 Microstructural Characterization
Porosity and pore size distribution are known to be essen-
tial for the mechanical properties, and several studies have
shown that the Weibull strength clearly decreases with
increased porosity, see, e.g., Ref. [31]. A high porosity is how-
ever needed in the backbone support since the electronic per-
colation must be obtained by infiltration. During infiltration,
the pores of the backbone structure are soaked with the pre-
cursor, and upon heat treatment, nano particles are formed
on the backbone surface [20]. To obtain electronic percolation
of the nano particles, several infiltration steps (>10) are typi-
cally needed, due to the possible precursor concentrations. To
minimize the time consuming infiltration steps, a high porosi-
ty, and a high pore volume:surface area ratio is desirable.
Porosities around 50% are typically applied for infiltration
based structures, see, e.g., Refs. [15, 20].
To obtain porous structures with highest possible pore
volume:surface area ratio, and with fewest possible notch-fea-
tures (for the mechanical properties), we focused on spherical
pore formers of PMMA (cf. Section 2). A
micrograph of the applied PMMA pore
former is shown in Figure 2a showing
spherical particles ranging from ca. 3 to
15 l, with most particles in the range of
5–10 lm. Examples of the resulting
microstructures are also shown, with the
dense structure containing no pore for-
mers and no measurable open porosity
(<1% according to the uncertainty of the
mercury intrusion porosimeter) in Fig-
ure 2b, and structures with measured
open porosity of 13% (Figure 2c and d),
and 46% (Figure 2e and f). The spherical
pore former shape is reflected in the sin-
tered structures, and the maximum pore
sizes are in the same range as the applied
pore former (5–10 lm). It can also be
seen that the porosity is evenly distribu-
ted across the tape casted layers.
From the micrographs, the microstruc-
tures appears to contain significant
closed porosity. Closed porosity is not
desirable for these structures, since it
cannot be infiltrated and the closed pores
can still act as mechanical weak points.
The fraction of closed porosity can be
evaluated from the bulk density (i.e., the
density including the volumes of the sol-
id, open, and closed pores), which is
measured with the mercury porosimeter
in the low pressure regime, where no
mercury is intruded. By comparing the
bulk density to the theoretical density of
the solid (i.e., 3YSZ), a measure of the
total porosity (open and closed) is obtained. The fraction of
closed porosity can then be deduced by subtracting the mea-
sured open porosity from the calculated total porosity. The
calculations are similar to Archimedes’ method, except the
bulk density and open porosity is measured by intruding
mercury instead of water. The measured open porosity (P
o
),
calculated closed porosity (P
c
), and average pore size based
on volume (d) are included in Table 1. It is seen that only
BB06 contains a relative significant amount of closed porosity
(i.e., half of the porosity), whereas the structures with higher
porosity contain mainly open porosity.
Most of the apparent closed porosity in the micrographs of
the high-porosity samples is due to the spherical nature of
the pores and the two-dimensional cross-section of the micro-
graph. This is also evident from the micrographs of the high-
porosity infiltrated structures, which show presence of infil-
trated material in almost all of the apparent closed pores (cf.
Figure 4b), illustrating that the pores must be connected and
percolating.
The open pore size distributions of the structures are
shown in Figure 3. Minor variations in pore size distribution
Fig. 2 SEM micrographs of the applied PMMA pore former (a), and the 3YSZ backbone layers
with different open porosity (shown in brackets); (b) open porosity < 1%, (c-d) open porosity 13%
and (e-f) open porosity 46%.
ORIGINAL RESEARCH PAPER
4© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim FUEL CELLS 00, 0000, No. 0, 1–9
www.fuelcells.wiley-vch.de

Klemensø et al.: Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells
can be observed between the samples with similar total open
porosity (BB07 and BB09 with respectively 46% and 47% open
porosity). These are ascribed to introducing a minor amount
of fibers in BB09 as described in the experimental section.
The samples with open porosity >46% appeared promising
for infiltration. Backbone structures of type BB07 was infil-
trated with 10 vol.% LSC, and this could be obtained after
10–12 infiltration cycles as measured by the weight gains. The
infiltrated LSC phase can be seen in Figure 4a and b as the
slightly darker phase which is not as densely sintered as the
YSZ backbone (also indicated by arrows), and it is seen to be
present both inside the pores of the porous structure (cf. Fig-
ure 4b), as well as deposited outside on the sample surface
(cf. Figure 4a). The irregular surface deposited LSC layer
could be removed by careful polishing (cf. Figure 4c). Due to
the low resolution of the tabletop microscope, the individual
nano particles cannot be discerned, but since the infiltration
procedure and solution was identical to the one described in
Ref. [20] similar nano structures can be assumed. The post
tested structures were also investigated and no evolution in
microstructure was detected within the resolution of the
tabletop microscope.
The backbone supported half cells suggested in Figure 1d
were attempted fabricated by laminating the preferred back-
bone support layer (type BB07) with additional cell layers. A
micrograph of the sintered half cell is shown in Figure 5, and
reveals a microstructure with a support layer suitable for
Table 1 Measured and calculated mechanical parameters for the 3YSZ support samples tested with the ball on ring method.
Sample name Sample data Mechanical data
Thickness (mm) P
o
(%) P
c
(%) d(lm) Number of samples r
0
(MPa) mV
eff
(mm
3
)r
0,corr
(MPa)
BB05 0.300 <1 – – 29 676 5.5 0.443 583
BB06 0.346 13 11 0.32 31 264 6.2 0.351 223
BB07 0.343 46 3 0.73 32 182 9.5 0.154 149
BB09 0.341 47 5 0.85 40 143 8.0 0.214 118
BB08 0.349 54 4 0.82 30 146 9.5 0.154 120
[34] – 30 – – 35 361 11.01 0.846 355
[34] – 30 – – 35 318 9.98 0.846 313
For comparison, data from ball on ring test with similar test geometry on Ni/YSZ supported half cells (support-, electrode-, and electrolyte-layers) are included, as
reported in [34]. For the half cell data, the mentioned porosity and mechanical properties refer to the support layer. The open porosity (P
o
), closed porosity (P
c
),
and average pore size (d) is based on mercury porosimetry measurements. The correlated Weibull strength (r
0,corr
) refers to an effective volume of 1 mm
3
.
Fig. 3 Pore size distribution of 3YSZ supports measured by mercury intru-
sion. Measured open porosity is indicated in the legend. All sample char-
acteristics are summarized in Table 1.
Fig. 4 SEM micrographs of porous backbone structure (type BB07) infil-
trated with 10 vol.% LSC. (a) and (b) show an unpolished sample, and (c)
shows the tested sample where the surface deposited LSC was removed by
polishing.
ORIGINAL RESEARCH PAPER
FUEL CELLS 00, 0000, No. 0, 1–9 © 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 5
www.fuelcells.wiley-vch.de

Klemensø et al.: Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells
infiltration. However, the four layered structure were not
shrinkage matched, and the samples suffered from severe
cracking and warping making further electrochemical cell
testing impossible. To obtain a large scale cell of this type,
which would be suitable for testing, would require extensive
processing optimization. The possible routes to improve the
shrinkage matching of the co-sintered multi-layered structure
could include tailoring the powder sinterability for some/all
of the layers (i.e., modification of particle size or inclusion of
sintering additives), modifying or substituting the organic
components in the tape cast slurry, and/or optimizing the
sintering profile, e.g., by detailed dilatometry studies. This
extensive optimization study was considered outside the
scope of the initial conceptual work reported here.
3.2 Mechanical Properties
Supports with variable porosity and pore size distribution
were mechanically tested using the ball on ring method. The
calculated Weibull parameters, Weibull strength (r
0
), Weibull
modulus (m), effective volume (V
eff
), and the Weibull
strength corrected to an effective volume of 1 mm
3
(r
0,corr
),
for the different samples are summarized in Table 1. The cal-
culations were done with the equations derived in Ref. [32],
and the effective volume was calculated according to the
method presented by Frandsen [submitted for publication].
As expected, a clear decrease in Weibull strength was
observed with increasing porosity. A maximum Weibull
strength of 676 MPa was observed for the denser sample
(BB05), 182 MPa for the sample with 46% open porosity
(BB07), and 146 MPa for the sample with highest porosity
(BB08 with 54%). Based on this, the BB07 was evaluated as
the most suitable support structure, since it shows reasonable
strength, combined with sufficient porosity for infiltration.
For the samples with similar porosities (46% and 47% for
BB07 and BB09), a significantly lower Weibull strength was
observed for the 47% porous sample. This sample also dis-
played a coarser pore size distribution (cf. Figure 3) and big-
ger average pore diameter (cf. Table 1). Bigger pores corre-
spond to bigger flaws, and the reflected lower measured
Weibull strength is therefore not surprising, and is in accor-
dance with the mechanical laws for brittle ceramics. The coar-
ser pore size distribution of BB09 is believed to be related to
the presence of fibers affecting the slurry dispersion, since no
special care was taken to stabilize the fibers in the processing.
A possible beneficial effect of the introduced fibers on the
flexural strength could therefore not be concluded from this
study, and would require samples with similar pore size dis-
tribution or alternatively toughness measurements.
Another trend with increasing porosity, is the increase in
Weibull modulus, going from 5.5 for BB05 (<1% open porosi-
ty) to 9.5 for BB07 and BB08 (with 46–54% porosity). The
trend with lower scattering as the Weibull strength is
decreased is not fully understood, but commonly observed
for ceramics, see, e.g., Ref. [33].
To evaluate the mechanical properties of the porous 3YSZ
supports, the data should be compared with data on the con-
ventional Ni/YSZ supported cell (after reduction of the Ni)
and obtained using similar test methodology. Few data of this
type is available to the authors’ knowledge, but in [34] ball on
ring data on reduced Ni/YSZ supported half cells (i.e., a mul-
tilayer of Ni/YSZ support and Ni/YSZ electrode and YSZ
electrolyte) are reported. The half cell data are included in
Table 1 together with the reported open porosity. The test
volume is not included directly in the [34] study, however in
the compilation of [25] it is reported to be 0.846 mm
3
(with a
support ring of diameter of 16 mm, and ball diameter of 2.74
or 2.90 mm).
It is seen that the support type BB07 has similar Weibull
modulus (ca. 10) as the Ni/3YSZ support of the half cells. In
contrast, the correlated Weibull strength of the porous back-
bone (BB07) is considerably lower than the Ni/YSZ support.
The more than halved correlated Weibull strength of the
BB07 backbone sample (149 MPa), compared to the Ni/YSZ
supports (313 or 355 MPa), can be ascribed to the higher open
porosity (46% vs. 30%).
The decrease in Weibull strength as a function of porosity
can often be described by an exponential expression, see, e.g.,
Ref. [32]. In Figure 6, the Weibull strength as a function of
open porosity is illustrated for the measured samples, and
the trend is also indicated for these samples. For comparison,
the data reported in Ref. [31] for NiO–3YSZ obtained with
similar ball on ring setup, and the data in Refs. [25, 34] for
Ni–3YSZ, are included with Weibull strengths correlated to a
test volume of 1 mm
3
. It can be seen that the strength of the
porous YSZ backbones is comparable to the standard Ni/YSZ
supports when taking the higher porosity into account.
For a full evaluation of the mechanical properties of the
porous 3YSZ support and the introduction of fibers, tough-
ness measurements should be included. To obtain these data
more advanced experiments are required, see, e.g., Ref. [35]
for a method to obtain toughness data. Work is in progress to
obtain these characteristics and will be reported in a future
publication.
The infiltration is believed to influence the mechanical
properties to a minor degree. It has been reported how infil-
trated material may improve the adhesion between co-sin-
tered layers, and to some degree be able to prevent delamina-
Fig. 5 SEM micrograph of a sintered half cell with a porous YSZ support
(type BB07) on the oxygen side.
ORIGINAL RESEARCH PAPER
6© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim FUEL CELLS 00, 0000, No. 0, 1–9
www.fuelcells.wiley-vch.de

Klemensø et al.: Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells
tion [36]. However, the mechanical properties of the layer
itself are not believed to be significantly changed by the low-
temperature process of infiltration of material. In state of the
art cell fabrication, the layers applied post sintering (e.g.,
screen printing of the oxygen electrode layer), are known to
have minor effect of the stresses and mechanical properties of
the cell [25].
3.3 Conductivity
The electronic conductivity of the backbone supports (type
BB07) infiltrated with 10 vol.% LSC was tested in a four ter-
minal setup. Two samples were tested, where the first was
mounted in the set-up directly after infiltration, and the other
was polished on all surfaces before mounting, removing most
of the excess LSC deposited on the sample surfaces, cf. Fig-
ure 4a and c. After mounting the sample, the temperature
was gradually increased up to ca. 1,000 °C, kept at >1,000 °C
for 2 h, before the temperature was lowered again to 850 °C
for the stability study. The test profile was chosen to simulate
a typical cell test, where the cell set-up is initially sealed at
high temperatures, before the temperature is again lowered
to the operation conditions. The logged (non-equilibrium)
conductivity as a function of temperature is summarized in
Figure 7 for the unpolished sample (solid lines), and the pol-
ished sample (dashed lines). Part of the curves show the
warm-up (black solid symbols), the cool down to 850 °C (gray
symbols), and the cool down upon termination of experiment
with the unpolished sample (open triangles).
For both samples, the conductivity was seen to increase
with temperature during the warm-up to ca. 900 °C (black
symbols). LSC is known to have metallic conduction proper-
ties, i.e., increased conductivity at lower temperatures. The
opposite trend during warm-up can be ascribed to the LSC
phase being not completely formed, the formation of second-
ary phases, and sintering of the precursor phases. From the
LSC precursor used in the present study, the formation of
LSC has been reported to be complex with the associated for-
mation of several secondary phases [20]. Non-equilibrium
effects during the ramping may also affect the slope of the
curve.
Above ca. 900 °C, a sudden decrease in conductivity is ob-
served. This is believed related to the LSC nano particle sin-
tering starting to be dominant. After ca. 2 h at 1,000 °C, the
sample is cooled to 850 °C, where the durability study is car-
ried out. For both samples, increased conductivity upon low-
ering the temperature is now observed (gray symbols),
reflecting metallic behavior. After ending the durability study,
the test is shutdown and the sample is again cooled, while show-
ing slightly increased conductivity(open symbols).
The result of the stability tests at 850 °C is shown in Fig-
ure 8. After reaching 850 °C, the conductivity stabilizes with-
in a 2–3 h at respectively 16 S cm
–1
(black curve, infiltrated
sample without polishing) and 8.5 S cm
–1
(gray curve, pol-
ished sample).
The conductivity values are comparable with the ones
reported in Ref. [37], where the identical LSC phase was infil-
trated into a GDC backbone. In Ref. [37], a conductivity
between 2 and 5 S cm
–1
was reported for a GDC backbone
infiltrated with 12 vol.% LSC after exposure to 600–900 °C.
When compared to this study, it appears that the interfacial
reaction between the infiltrated LSC and the YSZ backbone
does not have a detrimental effect on the conductivity of the
LSC phase.
A significantly higher initial conductivity is seen for the
sample that was not surface polished before testing. This may
be ascribed to the surface deposited LSC providing a surface
layer of pure LSC with higher conductivity (due to the lower
Fig. 6 Measured Weibull strength determined by ball on ring experiments
as a function of the open porosity of the YSZ samples (BB05, BB06, BB07,
BB08). For comparison, data from ball on ring measurements on NiO/
3YSZ [31], and the Ni/3YSZ data from Ref. [34] are included. The Wei-
bull strengths are correlated to V
eff
=1mm
3
.
Fig. 7 Measured conductivity as a function of temperature during test
warm-up (black symbols), cool-down to durability test temperature of
850 °C (gray symbols), and during test shut-down (open symbols). The
unpolished sample is represented by solid lines, and the polished sample
by the dashed lines.
ORIGINAL RESEARCH PAPER
FUEL CELLS 00, 0000, No. 0, 1–9 © 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 7
www.fuelcells.wiley-vch.de

Klemensø et al.: Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells
porosity) as seen in the micrographs in Figure 4a. The con-
ductivity of dense LSC pellets at 850 °C has been reported to
be in the range of 1,000–2,000 S cm
–1
(depending on Sr-
dopant level), and for La
0.7
Sr
0.3
CoO
3
it is reported to be ca.
1,650 S cm
–1
[38]. A micron thin surface layer is therefore
believed to have a significant effect on the total conductivity.
As a rough approximation, the unpolished sample could be
modeled as a three-layer system consisting of the 0.343 lm
thick LSC infiltrated 3YSZ support with conductivity
8.5 S cm
–1
(as the polished sample), which is sandwiched
between two ca. 1 lm thin surface layers of dense LSC with
conductivity of 1,650 S cm
–1
. If the three layers are assumed
to be parallel connected, the total conductivity of the three-
layer system is calculated to be in the range of 18 S cm
–1
, i.e.,
in good agreement with the measured unpolished sample
(16 S cm
–1
).
The surface deposited layer may be speculated to have a
useful role and act as a possible current collection layer, when
stacking the cells. However, to evaluate the conductivity of
the porous support layer, the polished sample is believed to
be more representative.
For both samples, the conductivity is sufficiently high for
the serial resistance of the support to be small when looking
at the cell resistance contributions. A conductivity of 16 or
8.5 S cm
–1
of a 300 lm support, corresponds to serial resis-
tances of respectively below 2 and 3.5 m Xcm
2
. For compari-
son, a 10 lm thick YSZ electrolyte at 850 °C will contribute
with ca. 5 m Xcm
2
[1].
The samples displayed remarkable stability despite the
relatively high temperature of 850 °C. After more than 600 h
of test, the conductivity was still >8 S cm
–1
. The observed sta-
bility is suggested to be due the reaction between LSC and
the YSZ backbone known to occur above 650 °C [21, 22]. The
interfacial zirconates will form a poorly conducting layer
between the LSC and the YSZ, however in the support struc-
ture, where no electrochemical mechanisms are taking place,
this is apparently not too detrimental for the electronic con-
ductivity of the remaining percolated LSC coating. Over time
however, if the zirconates phase will continue to grow, the
associated Sr-depletion of LSC phase may result in decreased
conductivity of the LSC phase. To determine whether this is
an issue, more long term durability studies or accelerated
tests are needed.
4Conclusion
SOC support layers of porous 3YSZ with porosity suitable
for infiltration was fabricated. The porous structures showed
reasonable mechanical properties, with a Weibull strength of
182 MPa for a 46% porous support. The Weibull strengths
were not directly superior to the conventional, and less por-
ous, Ni/YSZ supported cells. However, when taking the
higher porosity into account, the mechanical properties were
comparable. It was demonstrated possible to introduce suffi-
cient electronic conductivity into the support structures by
infiltration of LSC. Initial conductivity values of 8–16 S cm
–1
were obtained, corresponding to a serial resistance contribu-
tion of a 300 lm support below 3.5 m Xcm
2
. The conductiv-
ity was almost stable for over 600 h, and the 3YSZ support
layer for an oxygen electrode supported SOFC, is therefore
considered a promising component. To test the full concept
of the backbone supported cell, more work is needed, espe-
cially on developing the multilayered structure.
Acknowledgements
The work was financially supported by the Department of
Energy Conversion and Storage. The authors gratefully ac-
knowledge M. Nielsen and H. Paulsen (DTU Energy Conver-
sion) for technical assistance in sample fabrication, and the
Department of Wind Energy for technical help with the me-
chanical tests.
References
[1] N. Q. Minh, T. Takahashi, Science and Technology of Cera-
mic Fuel Cells, Elsevier, Amsterdam, The Netherlands,
1995, pp. 69.
[2] N. Christiansen, H. Holm-Larsen, S. Primdahl, M. Wan-
del, S. Ramousse, A. Hagen, ECS Trans. 2011,35, 71.
[3] M. Chen, J.-L. Luo, K. T. Chuang, A. R. Sanger, Electro-
chim. Acta 2012,63, 277.
[4] M. C. Tucker, J. Power Sources 2010,195, 4570.
[5] C. Bernuy-Lopez, R. Knibbe, Z. He, X. Mao, A. Hauch,
K. A. Nielsen, J. Power Sources 2011,196, 4396.
[6] M. Ni, M. K. H. Leung, D. Y. C. Leung, J. Power Sources
2006,163, 460.
[7] T. Klemensø, C. Chatzichristodoulou, J. Nielsen, F. Boz-
za, K. Thyden, R. Kiebach, S. Ramousse, Solid State Io-
nics 2012,224, 21.
[8] H. Nagamoto, Z. H. Cai, Electrochem. Soc. Proc. Sixth Int.
Symp. SOFC,1999, 99-19, pp. 163.
Fig. 8 Conductivity stability tests at 850 °C in air flow (50 mL min
–1
) for
the unpolished (black) and polished sample (gray).
ORIGINAL RESEARCH PAPER
8© 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim FUEL CELLS 00, 0000, No. 0, 1–9
www.fuelcells.wiley-vch.de

Klemensø et al.: Development of a Novel Ceramic Support Layer for Planar Solid Oxide Cells
[9] S. M. Beresnev, O. F. Bobrenok, B. L. Kuzin, N. M. Bog-
danovich, A. A. Kurteeva, D. A. Osinkin, G. K. Vdovin,
D. I. Bronin, Russian J. Electrochem. 2012,48, 969.
[10] A. Torabi, A. R. Hanifi, T. H. Etsell, P. Sarkar, J. Electro-
chem. Soc. 2012,159, B201.
[11] A. R. Hanifi, A. Torabi, T. H. Etsell, L. Yamarte, P. Sar-
kar, Solid State Ionics 2011,192, 368.
[12] E. Hardjo, D. Monder, K. Karan, ECS Trans. 2011,35,
1823.
[13] X. Zhu, Z. Lü, B. Wei, Y. Zhang, X. Huang, W. Su, Int. J.
Hydrogen Energy 2010,35, 6897.
[14] S.-W. Jung, J. M. Vohs, R. J. Gorte, J. Electrochem. Soc.
2007,154, B1270.
[15] T. Klemensø, K. Thydén, M. Chen, H.-J. Wang, J. Power
Sources 2010,195, 7295.
[16] M. C. Tucker, G. Y. Lau, C. P. Jacobsen, L. C. De Jonghe,
S. J. Visco, J. Power Sources 2008,175, 447.
[17] L. Zhao, X. Ye, Z. Zhan, J. Power Sources 2011,196, 6201.
[18] S. P. Jiang, Int. J. Hydrogen Energy 2012,37, 449.
[19] F. Zhao, L. Zhang, Z. Jiang, C. Xia, F. Chen, J. Alloys
Compounds 2009,487, 781.
[20] A. Samson, M. Søgaard, R. Knibbe, N. Bonanos, J. Elec-
trochem. Soc. 2011,158, B650.
[21] M. Sase, D. Ueno, K. Yashiro, A. Kaimai, T. Kawada,
J. Mizusaki, J. Phys. Chem. Solids 2005,66, 343.
[22] H.-S. Noh, J.-W. Son, H. Lee, J.-S. Park, H.-W. Lee, J.-H.
Lee, Fuel Cells 2010,10, 1057.
[23] R. Küngas, F. Bidrawn, J. M. Vohs, R. J. Gorte, Electro-
chem. Solid-State Lett. 2010,13, B87.
[24] R. Knibbe, J. Hjelm, M. Menon, N. Pryds, M. Søgaard,
H. J. Wang, K. Neufeld, J. Am. Ceram. Soc. 2010,93,
2877.
[25] A. Nakajo, J. Kuebler, A. Faes, U. F. Vogt, H. J. Schind-
ler, L.-K. Chiang, S. Modena, J. Van herle, T. Hocker,
Ceram. Int. 2012,38, 3907.
[26] R. Stevens in Engineered Materials Handbook,Vol. 4 (Ed.
S. J. Schneider) ASM International, Materials Park, OH,
USA, 1991, pp. 775.
[27] J. Chevalier, L. Gremillard, A. V. Virkar, D. R. Clarke,
J. Am. Ceram. Soc. 2009,92, 1901.
[28] R. H. J. Hannink, P. M. Kelly, B. C. Muddle, J. Am. Cer-
am. Soc. 2000,83, 461.
[29] J. R. Kelly, I. Denry, Dental Mater. 2008,24, 289.
[30] H. Raeder, C. Simon, T. Chartier, H. L. Toftegaard,
J. Eur. Ceram. Soc. 1994,13, 485.
[31] H. L. Frandsen, T. Ramos, A. Faes, M. Pihlatie, K. Bro-
dersen, J. Eur. Ceram. Soc. 2012,32, 1041.
[32] H. L. Frandsen, Mech. Mater. 2012,55, 33.
[33] X. Fan, E. D. Case, F. Ren, Y. Shu, M. J. Baumann,
J. Mech. Behav. Biomed. Mater. 2012,8, 21.
[34] A. Faes, H. L. Frandsen, M. Pihlatie, A. Kaiser, D. R.
Goldstein, J. Fuel Cell Sci. Technol. 2010,7,1.
[35] S. Goutianos, H. L. Frandsen, B. F. Sørensen, J. Eur. Cer-
am. Soc. 2010,30, 3173.
[36] K. Chen, N. Ai, S. P. Jiang, J. Electrochem. Soc. 2010,157,
P89.
[37] A. J. Samson, P. Hjalmarsson, M. Søgaard, J. Hjelm,
N. Bonanos, J. Power Sources 2012,216, 124.
[38] A. Mineshige, M. Kobune, S. Fujii, Z. Ogumi, M. Inaba,
T. Yao, K. Kikuchi, J. Solid State Chem. 1999,142, 374.
______________________
ORIGINAL RESEARCH PAPER
FUEL CELLS 00, 0000, No. 0, 1–9 © 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 9
www.fuelcells.wiley-vch.de