Synthesis and performance of sol-gel prepared Ni-YSZ cermet SOFC anodes

Article abstract of DOI:10.1149/1.1855875

Sol-gel derived (SD) NiO-YSZ particles were formed from metallic nitrate and butoxide precursors, and after heat-treatment at 1000°C, the NiO and yttria-stabilized zirconia (YSZ) average particle sizes were found to be less than 50 and 30 nm, respectively, as measured using X-ray diffraction and transmission electron microscopy. In both cases, small particles were observed as larger agglomerates, up to 100 nm in size. The SD powders were suspended in ethanol and sprayed onto YSZ disks, roughened with a thin layer of pure SD YSZ to improve active layer adhesion. Electrochemical evaluation (low- and high-field cyclic voltammetry and impedance spectroscopy) was carried out in a half-cell configuration at 800°C in a 3% H₂O/H₂ mixture. The exchange current density, obtained from all three methods, was found to agree only if a two-step hydrogen oxidation reaction was assumed, and the relatively low Tafel slope obtained suggests that the second electron-transfer step may be rate limiting. By comparison with ceramic grade (CG) Ni-YSZ, our results indicate an initial tenfold higher performance for the SD Ni-YSZ. However, a gradual loss of performance was observed for the SD electrodes that was not seen with the CG materials, possibly attributed to the thinness of the SD active layer.

Full text of DOI:10.1149/1.1855875

Journal of The Electrochemical Society, 152 ͑3͒ A645-A651 ͑2005͒
A645
0013-4651/2005/152͑3͒/A645/7/$7.00 © The Electrochemical Society, Inc.
Synthesis and Performance of Sol-Gel Prepared Ni-YSZ
Cermet SOFC Anodes
*
**,z
Peter G. Keech, Danielle E. Trifan, and Viola I. Birss
Department of Chemistry, University of Calgary, Calgary, Alberta, Canada
Sol-gel derived ͑SD͒ NiO-YSZ particles were formed from metallic nitrate and butoxide precursors, and after heat-treatment at
1000°C, the NiO and yttria-stabilized zirconia ͑YSZ͒ average particle sizes were found to be less than 50 and 30 nm, respectively,
as measured using X-ray diffraction and transmission electron microscopy. In both cases, small particles were observed as larger
agglomerates, up to 100 nm in size. The SD powders were suspended in ethanol and sprayed onto YSZ disks, roughened with a
thin layer of pure SD YSZ to improve active layer adhesion. Electrochemical evaluation ͑low- and high-field cyclic voltammetry
and impedance spectroscopy͒ was carried out in a half-cell configuration at 800°C in a 3% H2O/H2 mixture. The exchange current
density, obtained from all three methods, was found to agree only if a two-step hydrogen oxidation reaction was assumed, and the
relatively low Tafel slope obtained suggests that the second electron-transfer step may be rate limiting. By comparison with
ceramic grade ͑CG͒ Ni-YSZ, our results indicate an initial tenfold higher performance for the SD Ni-YSZ. However, a gradual loss
of performance was observed for the SD electrodes that was not seen with the CG materials, possibly attributed to the thinness of
the SD active layer.
© 2005 The Electrochemical Society. ͓DOI: 10.1149/1.1855875͔ All rights reserved.
Manuscript submitted July 8, 2004; revised manuscript received August 5, 2004. Available electronically February 1, 2005.
Solid oxide fuel cells ͑SOFCs͒ continue to garner interest for use
in stationary power systems due to their low SOx , NOx emissions
and high efficiencies.1 Composite cermet anodes, composed of Ni
and yttria-stabilized zirconia ͑YSZ͒, are widely used for prototype
SOFCs due to their low cost and high catalytic activity,2 despite
some reported problems associated with stability.3 Mechanical mix-
ing of ceramic grade NiO and YSZ powders is often selected as a
technique for fabrication of these anodes4 due to its simplicity. How-
ever, disadvantages of this approach include particle sintering,5
which decreases the catalytic area, and segregation of NiO and YSZ
particles, which can cause an uneven distribution of the Ni catalyst
following anode reduction.6 Consequently, recent strategies related
to SOFC anode synthesis have focused on the chemical coproduc-
tion of homogeneous mixtures of NiO and YSZ, thereby eliminating
these problems.6
Several of the recently investigated strategies for the formation
of homogeneous NiO-YSZ powders include self-assembly7 and the
use of sol-gel ͑SG͒ and sol-derived ͑SD͒ chemistry,6 due to the
relative simplicity and low cost of these synthetic approaches. Using
the SG method, chemical precursors of YSZ and NiO have been
combined in glycol and then dried to form a porous cake of NiO and
YSZ powders.8 While being a homogeneous mixture, the NiO and
YSZ powders produced using this approach are also nanoparticulate,
a promising characteristic for fuel cell electrocatalysts. In addition to
the increased surface area to mass ratio expected when particle size
is reduced, nanomaterials of this kind typically offer enhanced
physical properties when compared to their larger particle size coun-
terparts, e.g., ceramic-grade ͑CG͒ materials. These properties in-
clude strength, hardness, electrical resistivity, specific heat, etc.,9
and a decreasing tendency to sinter or undergo other catalyst deac-
tivation processes.
nificant during the early stages of testing, where more than a ten-
fold higher activity is observed for SD vs. CG anodes.
Experimental
SD powder synthesis.—Synthesis of NiO sol and SD
powders.—75.6 mmol Ni͑NO3)2•6H2O ͑Alfa Aesar, 99.9985%͒
was dissolved in 75 mL of ethanol and 10 mL of methanol for 4.5 h
in a round-bottom flask, as described elsewhere.11 The solution was
filtered and added to 160 mL of ethyl acetate to produce the NiO sol.
NiO SD powder was prepared by a two-part heat-treatment, as de-
scribed later. The sol was gradually heated, over 40 min, in air to
230°C, where it was held for 2 h and then allowed to cool over
several hours to room temperature. The resulting caked solid was
then ground to powder using a mortar and pestle and then heated
over 1.5 h in air to 1000°C, where it was held for 2 h before allow-
ing it to gradually cool to room temperature over several hours.
Synthesis of 8% YSZ sol and SD powders.—5.6 mmol of Y
͑NO3)3•6H2O ͑Alfa Aesar, 99.9%͒ were dissolved in 20 mL of iso-
propanol. This solution was added to a vessel containing 64 mmol of
Zr(n-BuO)4•BuOH ͑Alfa Aesar, 28% ZrO2 in butanol solution͒ dis-
solved in 76 mL of isopropanol. The synthesis is based on the
method of Kim et al., which involves the same starting materials
and solvents, but also includes acetic acid.12 The combined solution
was stirred for 18 h at room temperature under atmospheric condi-
tions to produce the YSZ sol. If exposed to the atmosphere for more
than 18 h, the sol undergoes gelation, but this process can be inhib-
ited through the addition of small quantities of nitric acid. The YSZ
SD powder was prepared following the same two-part heat-
treatment as described for the NiO SD powder, starting with the
acid-stabilized sol.
While offering much promise from a synthetic point of view, SD
Ni-YSZ anodes have undergone little in the way of performance
testing. Only the most rudimentary electrochemical tests10 have
been performed on any of these materials, and none have specifi-
cally targeted SOFC anode performance. This paper represents the
first detailed study of the synthesis and electrochemical performance
of Ni-YSZ SD anodes. We show that for our material, a mixture of
SD NiO and YSZ, with particle sizes ranging from 30 to 45 nm,
there is a significant performance improvement compared to CG
materials, as measured during electrochemical testing at 800°C in a
97:3 H2 :H2O atmosphere. This improved performance is most sig-
Synthesis of NiO-YSZ composite sol and SD powders.—The
composite sol was prepared by mixing the NiO and YSZ sols to-
gether in known molar ratios. With no further treatment, a gel
formed after 0.5-1 h in air. Addition of concentrated HNO3 was
found to increase time to gelation but does not eliminate it. NiO-
YSZ SD powder was prepared by the two-part heat-treatment de-
scribed.
Preparation of electrochemical test cells.—Preparation of YSZ
electrolytes.—0.5 mm YSZ electrolyte disks were prepared by
pressing 1 g quantities of 8% YSZ ceramic-grade ͑CG͒ powder
͑Tosoh, 99.7%͒ in a 25 mm X-ray fluorescence ͑XRF͒ die to 5000
psi. To improve the adhesion of the anode layer to these YSZ disks,
it was necessary to form a YSZ ‘‘anchoring layer’’. Consequently,
the YSZ disks were heated to ϳ80°C and sprayed with an air-
* Electrochemical Society Student Member.
** Electrochemical Society Active Member.
z E-mail: birss@ucalgary.ca
Downloaded on 2014-09-26 to IP 128.59.222.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A646
Journal of The Electrochemical Society, 152 ͑3͒ A645-A651 ͑2005͒
nebulized solution containing ϳ1 g of SD YSZ powder in 20 mL of
ethanol. These modified disks were then sintered in air at 1350°C for
4 h.
Application of NiO-YSZ anodes and other electrodes to YSZ
electrolyte.—1 g of 50:50 ͑mol͒ YSZ:NiO SD powders was sus-
pended in 20 mL of ethanol and air-nebulized and sprayed over a
circular area ͑0.40 cm2͒ on the masked YSZ disks. 1 g of pure SD
NiO in 20 mL of ethanol was subsequently sprayed on top of the
composite layer. During application of both of these layers, a sig-
nificant amount of this material did not affix to the electrolyte, mak-
ing quantification of the catalyst impossible at this point. Experi-
ments are ongoing to optimize the application of the SD powders
and to improve the thickness of the catalytic layer. The anode-coated
YSZ disk was then heated over 1 h to 1100°C, where it was held for
2 h before returning it to room temperature over several hours to
produce the working electrode ͑WE͒. Reduction of the NiO anode
material to Ni was carried out in an atmosphere of H2/3% H2O for
at least 2 h prior to electrochemical testing.
Pt was sputtered in two locations on the opposite side of the WE
to serve as the counter and reference electrodes ͑CE and RE, respec-
tively͒. The 0.83 cm2 CE was positioned symmetrically opposite to
the WE, while the 0.13 cm2 RE was 2 mm from the CE, as per the
geometry recommended to avoid artifacts in electrochemical data.13
Preparation of ceramic grade NiO-YSZ composite
anodes.—44:56 NiO ͑Alpha Aesar 99%͒:8% YSZ ͑Tosoh͒ were
combined and ballmilled for 1 h in ethanol. This material was then
added to glycerol, and the resultant paste was painted directly onto
sintered YSZ disks in a circular region of 0.03 cm2 to form the WE.
Following heat-treatment at 1450°C for 4 h, a Pt CE and RE were
sputtered onto the YSZ, prior to a second heat-treatment at 1100°C
for 2 h.
Figure 1. Schematic of electrochemical test cell.
holder consisted of an alumina rod on which the test cell was placed.
The cell itself was covered by an alumina cap, which was held in
place by a spring-loaded Monel tension wire. The cap served to
press Pt mesh current collectors to the RE and CE, and a Ni mesh to
the WE. The current collectors were attached to Pt wires, which
were fed through the Al stand via quartz tubes to the potentiostat. A
closed-ended quartz tube attached to the Al stand isolated the test
atmosphere, and the entire apparatus was wrapped with a Chromel
shielding wire to minimize electrical noise from the tube furnace.
All standard electrochemical experiments were performed using
a Solartron 1286 potentiostat with a Solartron 1255 frequency gen-
erator, using Scribner Corrware and ZPlot software. Cyclic voltam-
metry ͑CV͒ scans ranged from Ϫ300 to 500 mV ͑uncompensated͒
vs. the RE, generally at 10 mV/s. Electrochemical impedance spec-
troscopy ͑EIS͒, using a perturbation AC voltage of 10 mV root-
mean-square ͑rms͒, was carried out over a frequency range of
106-10Ϫ1 Hz at both the open circuit potential ͑OCP͒ and at dc
biases of 100, 200, and 300 mV vs. the OCP. It has been observed
that inaccurate overpotential measurements can arise from even
small electrode asymmetries during three-electrode measurements
using thin electrolyte systems.14 However, these were minimized
using the guidelines proposed by Chan et al.13 The distance between
the edges of the RE and the CE was approximately four times the
thickness of the electrolyte, which is more than the recommended
factor of three. In addition, IR compensation was not performed in
our work, as this has been shown to give incorrect approximations in
asymmetric systems.13
Characterization of SD powders and thin films.—XRD
characterization.—X-ray diffraction ͑XRD͒ was performed on a
Rigaku multiflex diffractometer ͑Department of Geology & Geo-
physics, University of Calgary͒ equipped with a Cu target with an
operating voltage and current of 40 kV and 40 mA, respectively. The
scanning range was varied between 10 and 70° at a rate of 2°/min.
The average crystallite size was calculated from peak broadening
using Jade software ͑version 6.5͒.
Transmission
electron
microscopy
character-
ization.—Transmission electron microscopy ͑TEM͒ was employed
to characterize the NiO, YSZ and NiO-YSZ SD powders. Micro-
graphs were obtained on either a Hitachi H-7000 or a Philips/FEI
field emission TEM ͑Microscopy and Imaging Facility, University
of Calgary͒; the latter also performed energy-dispersive X-ray
͑EDX͒ analysis. Samples for analysis were obtained by sonicating
20 mg of the SD powder in 10 mL of ethanol, and then depositing
small aliquots onto carbon-coated Cu TEM grids.
SEM analysis.—Scanning electron microscopy ͑SEM͒ was per-
formed using a Philips/FEI ESEM, ͑Microscopy and Imaging Facil-
ity, University of Calgary͒, equipped with EDX capabilities.
Samples with just the anchoring YSZ layer, as well as those with
completed anodes produced from NiO-YSZ SD powders, were
carbon-coated and examined. Top-down analysis was performed on
both materials, while cross-sectional analysis was performed only
on the anode layer to determine its thickness.
Results and Discussion
Electrochemical performance testing of Ni-YSZ anodes.—All
electrochemical experiments were performed in a humidified ͑3%
water͒ hydrogen atmosphere at 800°C. The presence of H2 ensured
the reduction of NiO to Ni and served as the fuel for the anode
reaction, while water was added to allow its reduction at the CE in
these half-cell experiments.
For electrochemical testing, a special cell holder was designed to
fit into the horizontal tube furnaces used to heat to the 800°C test
environment; a schematic of the holder is shown in Fig. 1. The cell
NiO, YSZ, and NiO-YSZ SD powder characterization.— XRD
characterization.—The XRD results for SD NiO, 8% YSZ, and
NiO-8% YSZ powders, are shown as Fig. 2a-c, respectively. NiO
was identified by matching three peaks to the data library ͑Jade,
vers. 6.5͒, and the peak widths show the average crystallite size to
be 45 nm. Similarly, the formation of 8% YSZ was confirmed using
two well-resolved peaks, as well as three peaks containing shoul-
ders. To establish the average crystallite size of YSZ, shown to be 30
nm, the widths of the two most resolved peaks were used. For the
Downloaded on 2014-09-26 to IP 128.59.222.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑3͒ A645-A651 ͑2005͒
A647
synthesizing nanoparticulate compounds, it is necessary to compare
these data with TEM results to distinguish crystallite and particle
sizes.
TEM examination of NiO, YSZ, and NiO-YSZ SD
powders.—Figure 3a shows a TEM micrograph of the NiO SD pow-
der in which all of the particles are seen to be below 100 nm in size,
with 47% of the particles being smaller than 20 nm, and 95% less
than 50 nm in size ͑Fig. 3b͒. These results are consistent with the
average particle size seen by XRD of 45 nm. The particles generally
appear crystalline in nature, possessing relatively sharp, well-
defined edges. Further, the largest particles appear to be composed
of smaller ones, indicating a propensity for these materials to
agglomerate.
As can be seen in Fig. 3c, the SD YSZ particles are not as well
resolved spatially as NiO and instead appear to be agglomerates,
many micrometers in length, composed of nanoparticles. Although
the small particles appear noncrystalline in nature, with poorly de-
fined edges, their size is estimated to range between 2 and 10 nm.
Even for the larger dark agglomerates in Fig. 3c, where small
particles cannot be resolved, particle size is estimated to be below
50 nm; this value can be considered a maximum particle size. These
particle sizes compare well with the average crystallite size
measured by XRD of 29 nm, as they span from 2 to a maximum of
50 nm.
The TEM image of the NiO-YSZ composite material, shown in
Fig. 4a, clearly reveals the presence of nanoparticles. Significantly,
EDX spot analysis proved to be sensitive enough to differentiate
between them, in contrast to what was reported in a previous study.6
As examples, the EDX spectra of particles 1 and 2 are shown in Fig.
4b and c, respectively. Based on the EDX analysis of all of the
particles, 1 and 3-5 are NiO, while particle 2 is YSZ. The TEM
analysis also shows the retention of the morphological characteris-
tics of the individual phases, with the NiO component appearing
more crystalline than the YSZ.
Similar particle sizes have been observed, using both XRD and
TEM, for mixed SD powders of Ni and YSZ ͑originally Ͻ10 nm in
size͒ that have undergone heat-treatment and reduction at 650°C in
hydrogen, i.e., 30-50 and 30 nm, respectively.6 For similar SD ma-
terials that had undergone heat-treatment to 450°C, the particle size
was only 3 and 15 nm for YSZ and Ni, respectively, but these
increased to 8 and 25 nm, respectively, when heat-treated to 600°C.8
Conventional CG anodes are derived from NiO/YSZ, having sig-
nificantly larger particle size, e.g., ϳ2 ␮m for mechanically mixed
powders,15 although ballmilling can produce samples where large
particles show a fine structure below 50 nm,16 as measured using
XRD. Particle sizes of ϳ1 ␮m are obtained for anode materials
prepared using spray pyrolosis,17 and 250-800 nm when combustion
synthesis is employed.18 NiO/YSZ powders synthesized using
citrate/nitrate combustion are reported to exhibit crystallite sizes as
low as 11 nm for NiO and 7 nm for Ni prior to heat-treatment, but
the corresponding particle sizes have not been determined using
TEM.19
Improving NiO-YSZ/YSZ interfacial properties.—Several syn-
thetic strategies were employed during the initial preparation of SD
NiO-YSZ anode materials. Initially, sols and alcohol solutions con-
taining SD powder were directly aliquoted onto sintered YSZ disks
and then heat-treated. Alternatively, NiO-YSZ SD powders were
combined in glycerol and were screen-printed onto sintered YSZ
disks, followed by heat-treatment. However, the adhesion of the SD
anode materials and the relatively smooth YSZ electrolyte surface
proved to be poor, and the anode layers simply flaked off during, or
immediately after, heat-treatment.
Figure 2. XRD pattern for ͑a͒ NiO, ͑b͒ 8% YSZ, and ͑c͒ 1:1 NiO:8% YSZ
SD powders.
Consequently, a new strategy was adopted in which an interlayer
of SD YSZ was deposited on the YSZ disk prior to heat-treatment,
ostensibly serving to roughen the YSZ surface. This was based on
parallel successful efforts to similarly improve the interfacial prop-
erties of SG-derived lanthanum strontium manganite ͑LSM͒ cathode
layers on YSZ disk electrolytes.20 Anodes constructed with this in-
terlayer present were able to undergo heat-treatment, chemical re-
composite NiO-YSZ SD powder, average crystallite sizes were
found to be 49 nm for NiO and 35 nm for YSZ, although the latter
value was taken from a single peak to avoid overlapping peaks.
While these measured values are encouraging in terms of the goal of
Downloaded on 2014-09-26 to IP 128.59.222.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A648
Journal of The Electrochemical Society, 152 ͑3͒ A645-A651 ͑2005͒
Figure 4. ͑a͒ TEM image of 1:1 NiO:8% YSZ SD powder, and EDX data
for ͑b͒ NiO and ͑c͒ 8% YSZ particles.
application of the anchoring layer was successful. In addition, much
larger particles, up to several micrometers in diameter, are also seen,
indicating a tendency for some of the YSZ to agglomerate, either
during the spray deposition or heat-treatment steps.
SEM characterization of YSZ and NiO-YSZ SD anode
materials.—Figure 6a shows a low-resolution, top-down backscat-
tered SEM image of a SD anode, consisting of a layer of NiO on a
NiO-YSZ composite layer, deposited on a modified YSZ electrolyte.
As the image contrast indicates, the NiO layer is not entirely uni-
form in thickness or structure, as the underlying YSZ can be de-
tected in some locations. Top-down EDX analysis performed on the
NiO-coated YSZ anode showed a high YSZ signal at all locations.
This observation is consistent with the presence of either a highly
porous
Figure 3. ͑a͒ TEM image of NiO powder, ͑b͒ histogram of NiO particle
sizes, and ͑c͒ TEM image of 8% YSZ SD powder.
duction, and electrochemical performance testing, without any loss
of the catalyst layer or any evidence of interfacial problems.
Figure 5 shows a secondary electron SEM image of the SD YSZ
anchoring layer on a standard YSZ disk. Sub-micrometer-sized
granular surface features are readily observable, indicating that the
Downloaded on 2014-09-26 to IP 128.59.222.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑3͒ A645-A651 ͑2005͒
A649
Figure 5. SEM image of YSZ anchoring layer on YSZ electrolyte disk.
or a very thin NiO layer, so cross-sectional analysis was performed
Figure 7. Typical CV ͑a͒ of H2 oxidation at SD NiO-YSZ electrode on YSZ
disk electrolyte and ͑b͒ resultant Tafel plot at 10 mV/s in 97:3 H2 :H2O at
800°C.
as well. The cross-sectional images of the anode layer, shown in
Fig. 6b, reveals a highly porous NiO structure and demonstrates that
the ϳ5-10 ␮m layer is quite uniform in thickness.
Relatively little SEM analysis has been performed previously on
electrodes fabricated from SD and related materials. More typically,
materials have been characterized in their powder form using TEM
and XRD methods, rather than as thin-film electrodes. When a
citrate/nitrate combustion synthesis was employed, by mixing nitrate
precursors and combusting them in a citrate matrix, it produced
nanoscopic crystals of NiO and YSZ. SEM analysis of the ashes
formed during this process19 showed the presence of ordered nano-
particles at high magnification, although only rough micron sized
particles were seen at lower magnifications. In the case of ceramic
grade NiO-YSZ, studied previously, cross-sectional SEM
analysis21,22,23 showed a marked similarity to the images obtained in
the present investigation, revealing a distinct porous anode layer on
the much more dense electrolyte.
Electrochemical performance of SD Ni-YSZ anodes.—Typical
CV scans during hydrogen oxidation at anodic potentials up to 0.5 V
and water reduction between 0 and Ϫ0.3 V, at a SD Ni-YSZ anode
deposited on a YSZ sol modified YSZ disk, are shown in Fig. 7a.
These data were collected without IR compensation, which proved
unnecessary due to the low series resistance present under these
conditions. The CVs show approximately exponentially increasing
currents as a function of potential, consistent with activation-
controlled kinetics. At high positive potentials, the current appears
to level off somewhat, possibly as hydrogen oxidation becomes lim-
ited by mass transfer. Figure 7a also shows that there is a decrease in
performance of the Ni-YSZ SD anode with continued cycling. The
Figure 6. ͑a͒ Low-resolution SEM top-down view of NiO-YSZ electrode on
YSZ electrolyte disk and ͑b͒ high-resolution SEM image of cross section of
NiO-YSZ electrode on YSZ.
Downloaded on 2014-09-26 to IP 128.59.222.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A650
Journal of The Electrochemical Society, 152 ͑3͒ A645-A651 ͑2005͒
cause of this performance degradation is unknown but may be re-
lated to the relative thinness of the active layer; deposition of this
layer is still being optimized.
During activation-controlled, the Butler-Volmer analysis is appli-
cable. A similar analysis has been performed during oxygen reduc-
tion on LSM cathodes24 using the familiar equation
i ϭ i0 e␣ ␩F/RT Ϫ eϪ␣ ␩F/RT
͒
͓1͔
a
c
͑
In this equation, R, T, and F have their usual thermodynamic mean-
ing, i is the measured current density, i0 is the exchange current
density, ␩ is the overpotential, and ␣a and ␣c are the anodic and
cathodic charge-transfer coefficients, respectively. Under the condi-
tions of activation control, both the low-field ͑LF, low overpoten-
tials͒ and high-field ͑HF, high overpotentials͒ regions can be used
for kinetic analysis.
At the high overpotentials associated with the HF region, the
reverse electrochemical reaction is assumed to be occurring at a
negligible rate. As a result, this simplifying assumption results in
Eq. 2, yielding a linear Tafel plot ͑␩ vs. log i)
␣aF␩
log i ϭ
ϩ log i0
͓2͔
RT
If it is assumed that the first electron-transfer step of a reaction
mechanism is rate determining, the high-field region is expected to
commence at about Ϯ100 mV at room temperature ͑with Ͻ1% er-
ror͒, while at 800°C, this increases to Ϯ400 mV. However, these
onset potentials are lowered when rate-limiting steps occur later in a
sequence of reaction steps. For these SD Ni-YSZ anodes, the Tafel
region appears to commence at an overpotential of 100 mV, as evi-
denced in Fig. 7b, suggesting that a later step in the mechanism is
rate-determining.
The i0 values, obtained by extrapolation of the Tafel plot ͑Eq. 3͒
to zero overpotential, were found to decrease from 390 to 270
␮A/cm2 over the analysis time of Fig. 7. This may suggest that an
unusual mechanistic change occurs at higher currents ͑overpoten-
tials͒, or possibly that an inhibiting effect of electrode porosity is
being experienced.25 Interestingly, there is also a slight decrease in
the Tafel slope derived from Fig. 7b, which drops from 310 to 285
mV/dec of current over the time of 4 h and 24 scans. To compare,
for a mechanism with an initial electron transfer as the rate-limiting
step, the Tafel slope would be expected to be 427 mV/dec of current
at 800°C. A Tafel slope of 213 mV/dec would be expected for a
reaction with a rapid first electron transfer followed by a rate-limited
chemical step, while a Tafel slope of 142 mV/dec would be expected
for a reaction with a rapid electron transfer, followed by a second
electron transfer. The measured Tafel slope does not agree with any
of these options, and this irregular result is discussed further.
In the LF region there is little perturbation from the OCP, and as
␩ is small, this allows the valid linearization of the Butler-Volmer
equation, consistent with the linear i/E relationship seen on either
side of 0 V derived from Fig. 7a
Figure 8. Nyquist data for H2 oxidation in 97:3 H2 :H2O at 800°C at SD
NiO-YSZ on YSZ disk electrolyte, all collected at the OCP, following 5 min
of polarization at ͑ᮀ͒ 0, ͑᭿͒ 100, ͑᭺͒ 200, and ͑᭹͒ 300 mV.
hydrogen oxidation at SD Ni-YSZ and were found to decrease from
830 to 585 ␮A/cm2 over the series of scans shown in Fig. 7a. A
comparison of the LF i0 values with those obtained from HF shows
that LF i0 values are larger, by a little more than a factor of two,
than those determined from HF analysis. Therefore, the LF i0 values
were recalculated using n/ of 2, giving values of 415 ␮A/cm2,
v
decreasing to 290 ␮A/cm2, which compare well with the HF i0
values given previously.
Mechanistically, the assumption that n/ is 2 can be easily ex-
v
plained for the hydrogen oxidation reaction that occurs at the anode.
As there are only two electrons transferred in total, n and v must be
2 and 1, respectively. This interpretation is consistent with the fol-
lowing mechanism
H2 ϩ O2Ϫ ꢀ OHϪ ϩ Hads ϩ eϪ
fast͒
͑
͓4͔
H
ads ϩ OHϪ → H2O ϩ eϪ
slow͒
͑
For this mechanism, the measured Tafel slope is expected to be
142 mV/dec of current at 800°C. The experimental slopes of 285
mV/dec, at their lowest, are approximately double the predicted
value, perhaps related to the porosity25 of the anode layer or indica-
tive of a more complex reaction mechanism involving the adsorp-
tion of intermediates.27
The EIS data also show a decrease in performance over the time
of these experiments. Figure 8 shows a series of Nyquist plots, col-
lected at the OCP, where an increasing diameter of the arc is indica-
tive of reduced performance. Scan 1 was collected prior to any
electrochemical testing, whereas scans 2- to 4 were taken following
the CVs shown in Fig. 7 and after brief periods of polarizations at
100, 200, and 300 mV. Notably, even when a series of EIS data sets
were collected only at the OCP without intermittent CVs, there was
still a tendency for a decrease in performance, typically between 4
and 9% per scan. The cause for this loss of performance is still
under investigation but appears to be partly due to the thin active
layer on these electrodes, which may be less stable than desired.
To obtain i0 from the EIS data, the charge-transfer resistance was
first obtained by fitting the data to an equivalent circuit ͑EC͒. In the
present work, a simple R1-R2/CPE circuit, shown in the inset of Fig.
8, was used. Also shown as another inset to Fig. 8 is the good fit
F n
i/␩ ϭ
i0
͓3͔
RT
v
where n is the total number of electrons transferred in the overall
reaction, and is the stoichiometric coefficient, the number of times
the rate-determining step occurs. The n/ term in Eq. 3 stems from
the relationship, n/ ϭ ␣ ϩ ␣ , which is used when the ␣ and ␣
values are unknown or when the reaction mechanism is presumed to
v
v
v
a
c
a
c
be multistep in nature.26
Often, n/ is assumed to be 1 for the calculation of i0 . Using
v
this assumption at room temperature, the LF region should span
Ϯ10 mV overpotential, allowing for an error of less than 1%, while
at the elevated working temperature of these experiments ͑800°C͒,
the LF region will extend between Ϫ40 and 40 mV. Using the as-
sumption that n/ ϭ 1, LF i values were calculated from Eq. 3 for
v
0
Downloaded on 2014-09-26 to IP 128.59.222.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑3͒ A645-A651 ͑2005͒
A651
Table I. Comparison of electrochemical performancea of SD vs.
CG Ni-YSZ anodes.
methods, was found to agree only if a two-step reaction was as-
sumed, with the slow step occurring only once in the reaction of one
molecule of hydrogen. Further, the relatively low Tafel slope ob-
tained suggests that a second electron transfer step may be rate
limiting.
By comparison with CG Ni-YSZ, our results indicate an initial
tenfold higher hydrogen oxidation activity for the SD Ni-YSZ. Even
so, there is good agreement between the inferred reaction mecha-
nism at the SD and CG electrodes. However, a gradual loss of per-
formance was observed for the SD electrodes that was not seen with
the CG materials, possibly attributed to the thinness of the SD active
layer. To overcome this loss of activity, fabrication of the SD anode
will be further optimized in near future work.
LF i0 ͑␮A/cm2͒ HF i0 ͑␮A/cm2͒ EIS i0 ͑␮A/cm2͒
Tafel slope
SD
CG
415-290
40-60
390-270
35-60
415-245
40-57
310-285
195-210
a In first 3 h of electrochemical evaluation.
obtained from one set of data to the selected EC. R1 is due to series
resistors, such as that originating from the electrolyte between the
WE and RE, contacts, cables, etc. R2 yields i0 and the values ob-
tained from R2 were similar in both magnitude and trends to those
calculated from the LF analysis above. For n/ ϭ 2, the initial i
v
0
Acknowledgments
value is 415 ␮A/cm2, dropping to 245 ␮A/cm2 at the end of the
experiment.
We gratefully acknowledge the overall financial support of the
Natural Sciences and Engineering Research Council of Canada
͑NSERC͒, as well as scholarship support for P.G.K. from NSERC
and the Alberta Ingenuity Fund. All imaging was performed at the
Microscopy and Imaging Facility University of Calgary, and we
thank R. Humphrey and Dr. J. Liu for their assistance with these
analyses.
Using the same electrochemical techniques and cell design, ex-
periments have also been carried out on CG Ni-YSZ composite
anodes for comparative purposes ͑Table I͒. Their i0 values increased
during the course of testing, although the final values are still well
below those measured for our as-yet unoptimized SD materials. The
LF i0 values were calculated using n/ ϭ 2, as calculations using
v
The University of Calgary assisted in meeting the publication costs of
this article.
n/ ϭ 1 again showed a twofold disagreement between the LF and
v
HF i0 values, similar to the SD materials. As can be seen from Table
I, the CG material has a lower Tafel slope than the SD material,
closer to what is expected based on the suggested mechanism of
Eq. 4. Overall, however, an approximate ten-fold improvement in
the H2 oxidation activity is seen at our SD materials compared to the
CG anodes during initial testing, although the differences became
somewhat smaller with time of operation.
References
1. S. C. Singhal, Solid State Ionics, 135, 305 ͑2000͒.
2. N. Q. Minh, J. Am. Ceram. Soc., 76, 563 ͑1993͒.
3. O. A. Marina, N. L. Canfield, and J. W. Stevenson, Solid State Ionics, 149, 21
͑2002͒.
4. J.-H. Lee, H. Moon, H.-W. Lee, J. Kim, J.-D. Kim, and K.-H. Yoon, Solid State
Ionics, 148, 15 ͑2002͒.
5. R. Vaben, D. Simwonis, and D. Stover, J. Mater. Sci., 36, 147 ͑2001͒.
6. G.-Q. Shao, H. Cai, J.-R. Xie, X.-L. Duan, B.-L. Wu, R.-Z. Yan, and J.-K. Guo,
Mater. Lett., 57, 3287 ͑2003͒.
7. M. Mamak, N. Coombs, and G. Ozin, Adv. Mater. (Weinheim, Ger.), 12, 198
͑2000͒.
8. M. Mamak, N. Coombs, and G. Ozin, Chem. Mater., 13, 3564 ͑2001͒.
9. S. T. Aruna and K. S. Rajam, Scr. Mater., 48, 507 ͑2003͒.
10. M. Mamak, N. Coombs, and G. Ozin, Chem. Commun. (Cambridge), 20, 2300
͑2002͒.
11. F. H. Moser and N. R. Lynam, U.S. Pat. 4,959,247 ͑1990͒.
12. S.-G. Kim, S. W. Nam, S.-P. Yoon, S.-H. Hyun, J. Han, T.-H. Lim, and S.-A. Hong,
J. Mater. Sci., 39, 2683 ͑2004͒.
13. S. H. Chan, X. J. Chen, and K. A. Khor, J. Appl. Electrochem., 31, 1163 ͑2001͒.
14. J. Winkler, P. V. Hendriksen, N. Bonanos, and M. Mogensen, J. Electrochem. Soc.,
145, 1184 ͑1998͒.
15. D. Simwonis, F. Tietz, and D. Stover, Solid State Ionics, 132, 241 ͑2000͒.
16. R. Wilkenhoener, R. Vaben, H. P. Buchkremer, and D. Stover, J. Mater. Sci., 34,
257 ͑1999͒.
Conclusions
The successful synthesis and electrochemical characterization of
SD NiO-YSZ particles demonstrates the excellent potential of these
materials as catalysts for SOFC applications. SD NiO and YSZ were
easily and rapidly produced from metallic nitrate and butoxide pre-
cursors, forming a dispersed nanoparticulate sol in the case of NiO,
or a gel in the case of YSZ. If desired, YSZ gel formation can be
inhibited using acid, so that the nanoparticulate sol remains. All
these materials are easily dried and can withstand heat-treatments of
at least 1000°C while retaining small particle sizes.
After heat-treatment at 1000°C, the NiO and YSZ average par-
ticle sizes were found to be below 50 and 30 nm, respectively, as
measured using XRD and TEM. In each case there appeared to be
some agglomeration of particles to produce larger structures, al-
though small particles could generally be seen within the large ag-
glomerates. Structurally, NiO particles appeared to have well-
defined edges when compared to YSZ particles, which looked less
crystalline in appearance.
Spray coat application of the YSZ SD-powder to pressed YSZ
disks prior to disk firing was found to enhance the adhesion of the
NiO-YSZ composite powders, likely by roughening the YSZ disk
surface. While not yet optimized, these anode layers are between 5
and 10 ␮m in thickness, although the relative thickness of the NiO-
YSZ composite active layer compared to the NiO current collecting
layer has not yet been determined.
Both low- and high-field CV, as well as impedance spectroscopy,
were employed in a half-cell configuration to assess the electro-
chemical performance of these SD anode layers at 800°C in 97:3
H2 :H2O. The exchange current density, obtained from all three
17. T. Fukui, S. Ohara, M. Naito, and K. Nogi, J. Power Sources, 110, 91 ͑2002͒.
18. S. T. Aruna, M. Muthuraman, and K. C. Patil, Solid State Ionics, 111, 45 ͑1998͒.
`
19. M. Marinsek, K. Zupan, and J. Maeek, J. Power Sources, 106, 178 ͑2002͒.
20. J. Fournier, J. Liu, and V. I. Birss, J. Electrochem. Soc., In preparation.
21. F. Tietz, F. J. Dias, D. Simwonis, and D. Stover, J. Eur. Ceram. Soc., 20, 1025
͑2000͒.
22. S. Primdahl and M. Mogensen, J. Appl. Electrochem., 30, 24 ͑2000͒.
23. M. Guillodo, P. Vernoux, and J. Fouletier, Solid State Ionics, 127, 99 ͑2000͒.
24. A. C. Co, S. J. Xia, and V. I. Birss, in Solid Oxide Fuel Cells, S. C. Singhal and M.
Dokiya, Editors, PV 2003-7, p. 478, The Electrochemical Society Proceedings
Series, Pennington, NJ ͑2003͒.
25. R. De Levie, in Advances in Electrochemistry and Electrochemical Engineering,
Vol. 6, P. Delahay, Editor, p. 329, Interscience Publishers/John Wiley & Sons, New
York ͑1967͒.
26. J. O’M Bockris and A. K. N. Reddy, Modern Electrochemistry, 2A Fundamentals of
Electronics, 2nd ed., p. 1053, Kluwer Academic/Plenum Publishers, New York
͑2000͒.
27. C. F. Zinola, A. M. Castro Luna, W. E. Triaca, and A. J. Arvia, Electrochim. Acta,
11Õ12, 1627 ͑1994͒.
Downloaded on 2014-09-26 to IP 128.59.222.12 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).