the effects of redox-cycling and oxidation kinetics...
TRANSCRIPT
The effects of redox-cycling and oxidation kinetics on anode-supported microtubular SOFCs
H. Monzón and M. A. Laguna-Bercero
Instituto de Ciencia de Materiales de Aragón (ICMA), CSIC- Universidad de Zaragoza, C/ María
de Luna 3 E-50018, Spain
Abstract
Accidental reoxidation is one of the most critical issues of SOFC anodes, leading to
performance degradation or even producing catastrophic failure of the cell. In this work
several redox cycling experiments over microtubular SOFC cells are reported. Oxidation kinetic
parameters have been determined for this geometry by Thermogravimetric Analysis
(TGA).Oxidation kinetics is known to be diffusion limited at typical operation conditions and,
for this reason, strictly dependant on geometry. A linear relationship was found between
expansion kinetics and damage on SOFC anodes in a series of dilatometric experiments. A
microtubular Ni-YSZ/YSZ/LSM-YSZ cell was redox cycled several times at partial oxidation
conditions (10% oxidation) and performance degradation was measured by j-V
characterization and the loss was found to be approximately 2% in performance per cycle.
1- Introduction
Fuel cells represent an efficient and clean approach to transforming fuel into electric energy.
By eliminating the chemical-to-heat and heat-to-mechanical energy transformation steps a
high electric energy yield can be achieved. Solid oxide fuel cells (SOFC) are a type of fuel cell
where the electrolyte is an ion conducting ceramic oxide, commonly, yttria stabilized zirconia
(YSZ). These materials present an acceptable value of ion conductivity at working
temperatures between 800 and 1000°C.
Anode materials for SOFC are porous cermets composed of an electric conducting phase
interpenetrating an ionic conducting phase, commonly nickel-YSZ. This electrode operates in
contact with the fuel stream and thus nickel remains in the reduced state. Either fuel supply
interruption, seal leakage or system shutdown can produce the anode oxidation (Ni transforms
into NiO), which has considerably fast kinetics at high temperatures, being this one of the
major issues in SOFC anodes [1,2]. This oxidation will cause de Ni phase to expand over 60%
while the YSZ phase and electrolyte will oppose to this expansion. This situation will generate a
large amount of strained zones on both phases. Some of this strain will be accommodated by a
combination of elastic and pseudoplastic deformation, and some will be relaxed through
microcracking [3]. As the zirconia phase will act as a rigid backbone, its mechanical properties
are closely related to the redox stability of the material. When using an anode supported
configuration, this process is even more critical, as anode expansion can affect the thin
electrolyte layer. If cracking occurs in the electrolyte layer, the single cell will cease to work. If
the oxidation proceeds without electrolyte cracking, restoration of reducing atmosphere will
reduce again the anode cermet. Nevertheless, due to irreversibilities during the redox cycles,
damage accumulation can be expected in this electrode, as well as performance degradation
[4,5,6,7,8].
In order to minimize the impact of the reoxidation on this material some strategies have been
recently reported in the literature, such as softer operation conditions (i.e. lower
temperatures), increased porosity, modified microstructure or lower nickel contents
[9,10,11,12]. A very high degree of redox stability has been reported on anode materials
fabricated by nickel impregnation over porous zirconia. On the other hand these materials
present nickel coarsening problems on the long term operation [13].
Microtubular SOFC configuration has been widely studied and also present some advantages in
comparison with other SOFC designs [14,15]. One is their higher resistance to thermal cycles.
These properties could be linked to redox-cycling resistance. For example, any sudden fuel
supply interruption or seal leakage could be counteracted by rapidly cooling the system to
avoid a fast nickel reoxidation. Up to date little work has been done on redox-cycling using
anode-supported microtubular SOFCs. For example, Pusz et al. [16] studied Ni/8YSZ anodes
fabricated using fine and coarse particle size powders, then exposed them to several redox
cycles and finally compared them. They observed that anodes using fine-structured powders
showed better mechanical strength after exposure to three redox cycles than the coarse-
structured material. They believed that a better adhesion between the anode material and the
Ni current collector was achieved for the fine-structured powder. The redox tolerance of
Ni/YSZ anode supported microtubular SOFC was also studied by Dikwal et al. [17]. They
performed experiments at 800 °C in partial oxidation and reduction conditions and a total of
52 redox cycles and found an average degradation rate of 0.3% per cycle. They identified two
modes of degradation: (a) anode degradation due to micro-crack formation; and (b)
degradation due to delamination at the electrode–electrolyte interfaces.
In the present paper we have performed similar redox cycling studies on anode supported
microtubular cells fabricated in our laboratory. Different findings concerning the relationship
between damage suffered by the samples and the kinetics of the deformation that takes place
will be discussed.
2- Materials and methods
The experiments were carried out in either bulk anode samples in the form of bars or in
tubular cells, depending on the type of experiment.
2.1- Anode sample preparation
Anode materials were fabricated using the powder composition as described elsewhere [18].
This powder composition consists of a mixture of commercial NiO (Alfa Aesar 99%), which was
previously ball-milled down to a final particle size of about 1 micron, 8YSZ (TZ-8YS: 8 mol%
Y2O3-stabilized ZrO2, 99.9% Tosoh, Japan) and corn starch as pore former. This composition
leads to a 50% porous anode with a 50% nickel and 50% zirconia phases on a solid volume
basis after sintering and the subsequent reduction.
Several cylindrical samples with this anode composition were fabricated via cold isostatic
pressing (CIP). Polyvinyl alcohol (PVA) was added as binder followed by pressing at 200 MPa
for 5 min using a latex mould. The resulting bar was then fired at 1350ºC for 2h. The bars were
then cut into 5 mm long and 4 mm in diameter cylindrical samples using a diamond saw. The
density of all samples was calculated geometrically and was found to be very close to the
theoretical values. This calculation is consistent with mercury intrusion porosimetry
measurements reported for similar samples [19]. All NiO-YSZ composite samples had a 65%
relative density, with deviations under 2% in all cases. This density leads to a 50% relative
density Ni-YSZ cermet.
2.2- Microtubular cell fabrication
Same composition was used to fabricate tubular green bodies by CIP, using latex mold and a
steel rod as die. These green tubes were pre-sintered at 950ºC during 4 hours. Electrolyte layer
was then deposited by wet basis spray coating, using 8YSZ (TZ-8Y: 8 mol% Y2O3-stabilized ZrO2,
99.9% Tosoh, Japan) in 2-propanol as coating agent; electrolyte thickness was controlled by
weight and was estimated to be around 20μm. Anode and electrolyte were co-sintered at
1350ºC for 2 hours. Two different materials were used as the cathode. For the first cell, porous
platinum ink (6082A, Metalor) was used, applied by brush coating, and sintered at 800 ºC for 1
hour. For the second cell a LSM-YSZ (Fuel cell materials) cathode layer was deposited by dip
coating and fired for 1.5 hours at 1250 ºC. Both cells were 12 cm long and had 3.5 mm outer
diameter; both cathodes were applied over one square centimeter surface due to measuring
equipment limitations.
2.3- Dilatometric measurements
The cylindrical samples were heated up to 800°C (typical operation temperature) in a
thermomechanical analyzer (Setaram, SETSYS 200, France). Performing redox cycles using this
setup allows the measurement of the sample elongation in real time while oxidation/reduction
proceeds. 5% hydrogen in Argon (5%H2/Ar) was initially used as reducing agent and thus
during the first heating, the initial NiO-YSZ precursor was reduced to the Ni-YSZ cermet. Once
the temperature was reached and a stable measurement was obtained, the inlet gas was
switched to synthetic air. The oxidation-reduction process was repeated for a total of three full
cycles, considering that the oxidation/reduction was completed when there was no sample
length variation.
2.4- Thermo-Gravimetric Analysis (TGA) of a microtubular sample
As oxidation kinetics of the Ni-YSZ cermet was found to be diffusion limited, as it will be
discussed in the results section, a TGA experiment was carried out to determine the kinetics on
a microtubular sample, as geometrical dependence is expected in the kinetics of a diffusion
limited process. For the present experiment, a 2 mm long tube was used. Similarly to the
dilatometric studies, the sample was heated up to 800°C in 5% H2/Ar atmosphere, then the
sample was purged under N2 atmosphere for 20 minutes and finally oxidized in air. During this
time sample weight was recorded as a measurement of the degree of oxidation (DoO) using a
TGA (TA instruments STD 29/60 simultaneous DTA-TGA).
2.5- Electrochemical measurements
j-V measurements for the microtubular cell were recorded using a VSP
Potentiostat/Galvanostat (Princeton Applied Research, Oak Ridge, US). Electrical connections
were made using Pt wires in a 4 tip measurement setup. A microtubular Ni-YSZ/YSZ/Pt cell was
employed for this experiment. The cell was heated up to the working temperature in a small
tubular furnace. 800°C was selected as the furnace set point temperature. The air temperature
close to the cell cathode was measured using an external thermocouple and was around
750°C. Although at these temperatures the performance of the cell is relatively poor due to
the low conductivity of the electrolyte and also to the low electronic conductivity of the LSM,
we have selected this temperature in order to compare with other existing results in the
literature and also to avoid a fast reoxidation of the Ni cermet. The performance at cell
temperature of 900°C for similar cells fabricated in our laboratory is about 800 mW cm-2
[18,19]. Pure hydrogen was humidified by bubbling through RT deionized water (97%H2-
3%H2O) and then fed to the anode side of the cell using a flow rate of 30 ml/min. The cathode
side was open to air atmosphere, as for low current regimes this has been proven enough, and
introduces a minor temperature fall compared to forcing an air flow towards the cathode.
2.6- Microstructural characterization
Scanning Electron Microscopy (SEM) experiments were performed under an accelerating
voltage of 20 kV using a JEOL 6400 SEM (JEOL, USA). Elemental
X-ray analysis were fitted with Oxford Instruments INCA energy dispersive
X-ray spectroscopy (EDS) analytical system. Some of the experiments were also performed
using a Merlin FE-SEM (Carl Zeiss).
3. Results
3.1- In situ Redox experiments in a dilatometer
The present experiment was performed over four different analogue samples, and each one
was cycled to a total of three redox cycles. As nickel oxidation takes place, bulk sample
expansion is recorded, as shown in figure 1. As previously mentioned, this bulk expansion
corresponds partially to elastic and pseudoplastic deformation and microcracking. Once the
steady state was reached, the carrier gas was switched back to 5%H2/Ar. At this time, NiO to Ni
reduction in the cermet leads to a small bulk contraction while some elongation is not
recovered. This non reversible cumulative redox deformation is an indicator of the redox
damage suffered by the material and is named in the literature as cumulative redox strain
(CRS) [20]. The oxidation-reduction process was repeated a total of three full cycles. As
observed in figure 1, CRS deviations for the different analyzed samples are important, and
there will be further discussed. The humps that can be observed after expansion finishes and
before reduction correspond to slight temperature changes caused by the gas change.
The present experiment also gives us useful information about the kinetics that rules the
oxidation/reduction process. As the length change is recorded in real time as
oxidation/reduction takes place, the elongation vs. time curve could be obtained. Assuming
that there is a relationship between degree of oxidation (DoO) and elongation, we can obtain
some information from the reaction kinetics. CRS can be calculated as shown in eq. 1 where L
is the instant sample length and L0, the initial sample length.
(1)
For each oxidation/reduction, normalized CRS can be calculated by dividing each CRS at each
moment by the maximum CRS on that cycle. The representation of this normalized CRS value
vs. time gives us useful information about deformation kinetics during oxidation/reduction.
3.2- Electrochemical measurements
3.2.1-Full redox cycle
A microtubular Ni-YSZ/YSZ/Pt cell was measured at 800°C (furnace temperature). An open
circuit voltage of around 1.084 V was measured, which is very close to the theoretical one of
1.132 V predicted by the Nernst equation. Once stable conditions were reached, the anode
input gas was switched to laboratory air using the same flow rates. The OCV was now recorded
while oxidation is taking place: a fast fall in the OCV was first recorded followed by a plateau at
about 650 mV, which corresponds to the coexistence of Ni and NiO phases [21]. After this
linear regime, the OCV fell down to zero, indicating that all Ni has been oxidized to NiO. After
this, hydrogen flow was restored but an important hydrogen leakage was detected and the
initial OCV was not further recovered. Post-mortem analysis of the cell revealed severe
electrolyte cracking all over the cell, as observed in figure 2, indicating that one full redox cycle
is catastrophic for this type of anode supported cells with a very thin electrolyte. Micro-
cracking is clearly observed in figures 2b and 2c, especially near the YSZ electrolyte, as
previously reported by other authors [2].
Similar experiments where catastrophic failure was not reached have been also reported [22],
but their cells presented severe performance degradation (over 50% fall in power output).
3.2.2-Partial oxidation cycles
After the full redox cycle experiment, partial oxidation cycles up to a certain extent (about
22%) were proposed. For this test, a Ni-YSZ/YSZ/LSM-YSZ microtubular cell was employed. This
experiment was designed to study the structural degradation caused in the Ni-YSZ anode. For
this reason, operation conditions of this test were chosen to avoid catastrophic failure and the
cell performance was not optimized, which would have required a higher operation
temperature. The oxidation extent was controlled by time, using the kinetic data obtained
from the TGA experiment. Figure 3 shows oxidation vs. time recorded in the TGA experiment
for a tubular sample. As described in the corresponding section, a purge was introduced in the
system before the oxidation in air, during this purge some oxidation occurred due to some
content in oxygen in the purging gas used ( about 10-13 bar of O2). For this reason the oxidation
vs. time curve showed in figure 3 starts at time 0 at approximately 6% oxidation. During 100
seconds the oxidation occurred up to a 28% extent, subtracting the initial extent, this yields an
oxidation increase of approximately 22%.
Similar working conditions were used for the present experiment. A single j-V measurement
was recorded, and the presence of hysteresis indicates an electrode activation phenomenon
[23]. Prior to any further j-V measurements, the cell was held at a constant current of 70 mA
cm-2 during 30 min in order to fully activate the LSM-YSZ cathode. Partial oxidation
experiments were finally carried out using the following procedure. Firstly, the inlet chamber
(anode site) was purged using Argon. Purge is complete when a constant OCV measurement
was obtained. For this working conditions this measurement is about 734 mV and corresponds
to a state were no hydrogen is left in the anode side and Ni acts as electron donor while
oxidizing to NiO. Once this state is reached, anode gas input is switched to synthetic air (20%
O2, 80% N2) at a flow rate of 30 ml/min. This flow is kept for 100 seconds, which corresponds
to a 22% oxidation according to kinetic data obtained from previous experiments. After this
oxidation period, the anode site was again purged using Ar for 30 seconds, and H2 atmosphere
was finally restored. Initial OCV values were recovered after a period of about 30 seconds. This
partial redox cycle was repeated a total of nine times, performing polarization measurements
after each single cycle. A thermal cycle was also performed for inspection between cycles 5
and 6. j-V measurement after cycle 6 is not shown, as the LSM-YSZ electrode was not fully
activated and the polarization values do not follow the trend. Prior to cycle 7, the cell
electrodes were fully activated and the subsequent cycles followed the initial trend as
observed in figure 4. The presence of increased scattering data for cycles 8 and 9 could be an
indication of small damage in the cell (by the presence of microcracks in the anode and/or the
electrolyte). Another possibility could be a defect in an electrical contact, as the data seems to
be noisier when increasing the current density. For that reason we decided to conclude the
experiment at this stage.
4- Discussion
4.1- Ni-YSZ oxidation kinetics
Reduction and oxidation kinetics for Ni-YSZ cermets have been widely studied in the literature
[20,24,25 and references there in]. It is believed that high temperature nickel oxidation follows
a two-step mechanism. First, a nickel oxide layer is formed on the outer surface of the particle;
this process follows a logarithmic law, as specific volume of NiO is greater than that of metallic
Ni cracks often appear on this step. Once the NiO layer reaches a thickness of about 0.1 μm,
the second step takes place. This step consists of two competing processes: the inner diffusion
of O2- through the NiO and cracks, and the outer diffusion of Ni2+ ions. This second process is
faster at 800°C and tends to dominate at these temperatures, though the balance between
these two processes controls the microstructural modification that the cermet undergoes [27].
As bulk deformation is caused by nickel oxidation, it is expected to find a relationship between
deformation kinetics and oxidation kinetics. It was found that deformation kinetics follow a
parabolic law, as oxidation does in the mechanism proposed formerly. From the four different
experiments carried out in the dilatometer, the parabolic kinetic constant kp can be obtained
as the squared slope of the lineal part of the normalized CRS vs time1/2 plot, as seen in figure
5.a, for all analyzed samples the parabolic interval corresponds to a normalized CRS of
between 20 and 75%. Significant dispersion is observed for the different analyzed samples; this
effect could be associated to the diffusion limitation of the process, as samples were
fabricated via CIP using calibrated latex molds, but their geometries could present either small
deviations from perfect cylindrical bodies or fabrication defects. Another remarkable result
from these experiments is the evolution observed in the kp values of each sample after the
different redox cycles. Figure 6 shows this trend of the acceleration of the deformation kinetics
as the different cycles take place. This effect could be again explained by the diffusion
limitation, as redox cycling modifies the microstructure, generating additional porosity and
probably more specific surface. Additionally, by plotting the total CRS reached by each sample
as a function of kp in the first oxidation, a lineal relationship was found (figure 7). This
relationship apparently is indicating that the faster the oxidation process takes place the more
damage is resulting in the cermets, and it prevents pseudoplastic deformation from relaxing
the generated stresses. It is well known that one of the factors which have a major influence
on the redox damage on these materials is the temperature at which oxidation takes place. In
these experiments we have tested several samples at the same temperature and obtained
different deformation kinetics, which were associated to differences in the microstructure, and
we have seen that these kinetics lead to different amounts of accumulated redox damage. We
have also performed SEM studies of all the samples in order to see any microstructural
differences between samples. No apparent differences were observed between the four
analyzed samples, although careful image analysis is under way in order to confirm this. In
addition, small differences could be observed between the sample at the initial stage and the
sample after 3 redox cycles, as observed in figure 8. The presence of bigger cracks and also the
bigger pores which seem to have been formed during the cycles will contribute to faster
oxidation kinetics. We can conclude that damage taken by Ni-YSZ cermets is proportional to
the oxidation kinetics, and probably most of the influence of temperature over damage can be
explained through faster kinetics.
4.2 - NiO-YSZ reduction kinetics
For NiO reduction kinetics, a close relationship was found between the measured contraction
kinetics and the reduction mechanism. NiO reduction follows linear kinetics, which means that
it is limited by a superficial adsorption step. Two additional stages can be seen in this reaction
kinetics, the first takes place at low conversion and corresponds to an induction period, and
the second one is observed at high conversion rates where kinetics are parabolic-like, probably
because water vapor removal is inhibited by the cermet densification and thus a diffusion
limited regime is taking place. The deformation kinetic constant for the lineal part was
obtained from the slope of the normalized CRS vs. time plot as shown in figure 5.b. A lineal
behavior was observed for values of CRS between 0.3 and 0.8. Values for the kinetic constant
obtained in these experiments (average obtained value of 3,68 min-1) show no evolution
through the redox cycling and are in good agreement with reduction kinetic values present in
the literature [28]. This result indicates that contraction is lineal with reduction, and that this
process is reaction rate limited and not diffusion limited as it was observed for the oxidation
process, so microstructural modification has a minor effect on it.
4.3- Damage and degradation in microtubular cells
As exposed in the experimental methods section, these devices were not found to be able to
withstand a single full redox cycle, as bulk anode expansion causes irreversible electrolyte
cracking. On the other hand, these cells present a very good thermal shock resistance [18],
suggesting that some damage can be prevented by rapid cooling in the event of an accidental
oxidation.
Some more quantitative information can be obtained from the partial oxidation experiments
performed. From the polarization curves measured after each cycle, two parameters were
used for comparative purposes: the maximum power output generated and the internal
resistance of the cell. Both parameters show a small progressive degradation as cycles
accumulate. The data represented in figure 9 shows a power output decay of about 2% for
each cycle. This degradation could be associated with a loss of
electrical contact inside the anode as the microstructure is modified during
the redox cycling, which could increase the electrical resistance.
The results exposed in this paper show that permanent performance degradation occurs in
these devices even when exposed to short oxidations. These results also validate the
assumption of damage accumulation, as the degradation measured after each cycle was found
to be approximately constant.
The microstructure of this cell was also examined by SEM. Figure 10.a shows the three layers
present in the cell, from left to right: cathode, electrolyte and anode. The observation
conditions used allow compositional contrast between YSZ and nickel, YSZ is seen dark gray,
while nickel is almost white [23]. No damage was found on the electrolyte and no
delamination was observed between electrode-electrolyte interfaces. This supports the idea of
the observed performance degradation being caused by anode microstructural degradation.
Figure 10.b shows the anode microstructure, were porosity can be observed in black in two
different geometries: 10 μm pores generated by pore former and submicron porosity
generated during reduction. This degradation can be qualitatively observed as a slight
modification in the geometry of the submicron porosity as well as an overall volume increase
of this phase.
5- Conclusions
The present experiments showed that anode supported microtubular geometry presents a
poor resistance towards redox cycling, as oxidation kinetics are relatively fast due to the small
anode thickness in a diffusion limited regime. Additionally the thin electrolyte layer will crack if
anode expansion reaches a critical level, as seen in the full redox cycle experiment. As
microtubular configuration allows rapid thermal cycling, coupled redox and thermal
experiments are suggested in order to reduce the damage. Additionally, it has been proven
that even partial oxidation cycles cause irreversible performance degradation, which is
accumulated from cycle to cycle. For these particular operation conditions and cell setup
tested, this performance degradation was around 2% power output lost after every 10%
oxidation cycle.
5 – Acknowledgements
The authors thank grants MAT2009-14324-C02-01, financed by the Spanish Government
(Ministerio de Ciencia e Innovacion) and Feder program of the European Community, and
GALC-2009/09, financed by the Aragon Government and La Caixa Foundation, for funding the
project. The use of Servicio de Microscopia Electronica (University of Zaragoza) is also
acknowledged. M.A.L.-B. finally thanks the JAE-program (CSIC) for financial support.
References
[1] S.C. Singhal, K. Kendall (Eds.), High Temperature SOFCs: Fundamentals, Design and
Applications, Elsevier, 2003
[2] S. J. Skinner and M. A. Laguna-Bercero, Advanced Inorganic Materials for Solid Oxide Fuel
Cells, in Energy Materials, Duncan W. Bruce, Richard Walton and Dermot O'Hare (Eds.),Wiley,
2010.
[3] Pihlatie M.H., Frandsen H.L., Kaiser A., Mogensen M. Continuum
mechanics simulations of NiO/Ni-YSZ composites during reduction and
re-oxidation. J. Power Sources 2010;195:2677-2690
[4] M. Ettler et al. Durability of Ni anodes during reoxidation cycles. Journal of Power Sources
2010;195:5452-5467.
[5] D. Sarantaridis and A. Atkinson. Redox cycling of Ni-based solid oxide fuel cell anodes: A
review. Fuel Cells 2007;7:246-258.
[6] T. Klemensø, C. Chung, P.H. Larsen and M. Mogensen. The mechanism behind redox
instability of anodes in high-temperature SOFCs. J.Electrochem. Soc. 2005; 152:A2186–A2192.
[7] J. Malzbender, E. Wessel, R.W. Steinbrech and L. Singheiser. Reduction and re-oxidation of
anodes for solid oxide fuel cells. Solid State Ion. 2005;176:2201-2203.
[8] D. Waldbillig, A. Wood and D.G. Ivey. Electrochemical and microstructural characterization
of the redox tolerance of solid oxide fuel cell anodes. J.Power Sourc. 2005;145: 206–215.
[9] D. Fouquet, A.C. Müller, A. Weber and E. lvers-Tiffte. Kinetics of Oxidation and Reduction of
Ni/YSZ Cermets. Solid State Ion. 2003; 8.
[10] David Waldbillig, Anthony Wood, Douglas G. Ivey. Thermal analysis of the cyclic reduction
and oxidation behaviour. Solid State Ion.2005:176;847–859.
[11] Pihlatie M., Ramos T., Kaiser A. Testing and improving the redox
stability of Ni-based solid oxide fuel cells. J. Power Sources
2009;193:322-330
[12] Faes A., Fuerbringer J. M., Mohamedi D., Hessler-Wyser A., Caboche G.
and Van herle J. Design of experiment approach applied to reducing and
oxidizing tolerance of anode supported solid oxide fuel cell. Part I:
Microstructure Optimization. J. Power Sources 2011;196:7058-7069
[13] Trine Klemensø, Karl Thydén, Ming Chen and Hsiang-Jen Wang. Stability of Ni–yttria
stabilized zirconia anodes based on Ni-impregnation. J.Power Sourc., 2010; 195.
[14] T. Suzuki et al. Micro Fuel Cells: Principles and Applications. Elsevier 2009.
[15] K. S. Howe, G. J. Thompson and K. Kendall. Micro-tubular solid oxide fuel cells and
stacks. J.Power Sourc. 2011;196:1677–1686.
[16] J. Pusz, A. Smirnova, A.Mohammadi and N. M. Sammes. Fracture strength of micro-
tubular solid oxide fuel cell anode in redox cycling experiments. J.Power Sourc.2007;163:900–
906.
[17] C.M. Dikwal, W. Bujalski and K. Kendall. Characterization of the electrochemical
performance of micro-tubular SOFC in partial reduction and oxidation conditions. J.Power
Sourc. 2008;181;267–273.
[18] R. Campana, R.I. Merino, A. Larrea, I. Villarreal and V.M. Orera. Fabrication,
electrochemical characterization and thermal cycling of anode. J.Power Sourc.2009;192:120–
125.
[19] R. Campana, A. Larrea, R. I. Merino, I. Villarreal, V. M. Orera, SOFC-minitubulares basadas
en YSZ, Bol. Soc. Esp. Ceram. V. 2008; 47: 189-195.
[19] M. A. Laguna-Bercero, R. Campana, A. Larrea, J. A. Kilner, V. M. Orera, Performance and
aging of microtubular YSZ-based Solid Oxide Regenerative Fuel Cells. Fuel Cells 2011; 11; 116-
123.
[20] M. Pihlatie, A. Kaiser and M. Mogensen. Redox stability of SOFC: Thermal analysis of Ni–
YSZ composites. Solid State Ion. 2009;180:1100–1112.
[21] D. Sarantaridis, R. A. Rudkin and A. Atkinson. On the Redox Cycling of Anode‐Suported
SOFCs: Mechanical Properties and Damage. ECS Trans. 2007; 7:1491‐1499.
[22] C. M . Dikwal. Cycling Studies of Micro-tubular Solid. PhD Thesis, Birmingham, 2009
[23] S. McIntosh, S. B. Adler, J. M. Vohs, R. J. Gorte. Electrochem. Solid State Lett. 2004; 7;
A111-A114.
[24] N.M. Tikekar, T.J. Armstrong, A.V. Virkar, J.Electrochem. Soc. 2006;153; A654–A663.
[25] M.A. Laguna-Bercero, A. Larrea, J.I. Peña, R.I. Merino, V.M. Orera, J. European Ceram. Soc.
2005; 25; 1455–1462.
[26] D. Waldbillig, A. Wood, D.G. Ivey, Solid State Ion. 2005; 176; 847–859.
[27] A. Faes. RedOx Stability of Anode Supported Solid Oxide Fuel Cells. PhD Thesis, Lausanne,
2011.
[28] K. Thydén, Y.L. Liu and J.B. Bilde-Sørensen. Microstructural characterization of SOFC Ni–
YSZ anode composites. : Solid State Ion. 2008;178:1984–1989.
Figure captions
Figure 1: Elongation recorded during redox cycling for samples: 1 (a), 2(b), 3(c), 4(d).
Figure 2: Post mortem inspection of the full reoxidized cell. (a): Overal view of severe
electrolyte cracking, (b): SEM micrograph showing how electrolyte cracks are penetrating
through the whole electrolyte, (c): Wide cracking cause the platinum cathode layer to crack
and delaminate.
Figure 3: Oxidation vs. time for a tubular sample recorded in a TGA experiment.
Figure 4: Electrochemical measurements of the cell after each partial redox cycle.
(a):Polarization curves, (b): Power output curves. For clarity means only some cycles have been
represented, though we have observed that all of them follow the same trend.
Figure 5: Fit of dilatometry measurements for (a) oxidation of sample 1, (b) reduction of
sample 2.
Figure 6: Evolution of kinetic constants of the four samples as a function
of the number of oxidation.
Figure 7: Oxidation kinetics - Cumulative redox strain (CRS) relationship.
Figure 8: Observed anode samples at initial state (a,b) and after 3 redox cycles (c,d).
Figure 9: Evolution of internal resistance and maximum power in redox cycled cell.
Figure 10: FE-SEM micrographs of partial redox cycled sample showing an overview of the
cathode, electrolyte and anode (a) and detail of anode microstructure (b).
Figure 1.a
Figure 1.b
Figure 1.c
Figure 1.d
Figure 2.a
Figure 2.b
Figure 2.c
Figure 3
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
0 100 200 300 400 500 600
Do
O [
%]
Time [s]
Figure 4.a
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
0 20 40 60 80 100 120
E (V
)
j(mAcm-2)
1
3
5
9
Figure 4.b
0
10
20
30
40
50
60
0 20 40 60 80 100 120
Po
we
r d
en
sity
(m
Wcm
-2)
j (mAcm-2)
1
3
5
9
Figure 5.a
y = 0.2904x - 0.609 R² = 0.9993
y = 0.4571x - 0.6931 R² = 0.9992
y = 0.552x - 0.8453 R² = 0.9948
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 1 2 3 4 5
no
rmal
ized
CR
S
time1/2[min1/2]
1st oxidation
2nd oxidation
3rd oxidation
Figure 5.b
y = -0.0391x + 0.8107 R² = 0.9994
y = -0.0421x + 0.805 R² = 0.9988
y = -0.0415x + 0.7978 R² = 0.9962
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 2 4 6 8 10 12 14
no
rmal
ized
CR
S
time [minutes]
2nd reduction
3rd reduction
1st reduction
Figure 6
0.0000
0.0500
0.1000
0.1500
0.2000
0.2500
0.3000
0.3500
1 2 3
kp [
min
-1]
Oxidation number
1
2
3
4
Figure 7
y = 0.1337x - 0.014 R² = 0.9727
0
0.005
0.01
0.015
0.02
0.025
0.03
0.15 0.17 0.19 0.21 0.23 0.25 0.27 0.29 0.31
CR
S
Kp 1st oxidation
Figure 8.a
Figure 8.b
Figure 8.c
Figure 8.d
Figure 9
40.00
42.00
44.00
46.00
48.00
50.00
52.00
54.00
56.00
58.00
4
4.5
5
5.5
6
6.5
7
0 2 4 6 8 10
max
po
we
r o
utp
ut
(mW
cm-2
)
Inte
rnal
re
sist
ance
[Ωcm2]
Redox cycles
Figure 10.a
Figure 10.b