Transition metal oxides as reconfigured fuel cell anode catalysts for improved CO tolerance: Polarization data

Article abstract of DOI:10.1149/1.1851057

We are developing fuel-cell-integrated approach for enhancing the effectiveness of air-bleed for CO tolerance of hydrogen and reformate polymer exchange membrane fuel cells (PEMFCs), called the reconfigured mode (RCA) [F. A. Uribe, J. A. Valerio, F. H. Garzon, and T. A. Zawodzinski, Electrochem. Solid-State Lett., 7, A376 (2004)]. It consists of a small modification to the backing cloth placed on the anode side of each membrane electrode assembly in a stack. A catalyst layer is placed on the gas-feed side of the cloth to catalyze oxidation of CO, utilizing the oxygen introduced in a small air-bleed. The purpose is to enhance the effectiveness of the air-bleed to achieve a high CO tolerance. We synthesized model RCA catalysts based on transition metal oxides, which were characterized using scanning electron microscopy, energy-dispersive spectroscopy, X-ray diffraction, thermogravimetric analysis, and Brunauer-Emmett-Teller techniques. Here we present performance data from polarization curves measured on fuel cells with air bleed using these model RCA catalysts. The results indicate that by using an RCA and air-bleed it is possible to raise the CO tolerance of a Pt/C anode to the same level as that of an anode with a Pt-Ru/C catalyst with similar platinum loading and without other means of CO mitigation. The RCA represents a built-in safety net for CO transients, which could be applied to PEMFCs destined to operate on a reformer-derived, hydrogen-rich fuel stream.

Full text of DOI:10.1149/1.1851057

Journal of The Electrochemical Society, 152 ͑2͒ A459-A466 ͑2005͒
A459
0013-4651/2005/152͑2͒/A459/8/$7.00 © The Electrochemical Society, Inc.
Transition Metal Oxides as Reconfigured Fuel Cell Anode
Catalysts for Improved CO Tolerance: Polarization Data
,z
Peter A. Adcock,a, Susan V. Pacheco, Kirsten M. Norman,
*
*
and Francisco A. Uribe
Los Alamos National Laboratory, Electronic and Electrochemical Materials and Devices Group,
Materials Science and Technology Division, Los Alamos, New Mexico 87545, USA
We are developing a fuel-cell-integrated approach for enhancing the effectiveness of air-bleed for CO tolerance of hydrogen and
reformate polymer exchange membrane fuel cells ͑PEMFCs͒, called the ‘‘reconfigured anode’’ ͑RCA͒ ͓F. A. Uribe, J. A. Valerio,
F. H. Garzon, and T. A. Zawodzinski, Electrochem. Solid-State Lett., 7, A376 ͑2004͔͒. It consists of a small modification to the
backing cloth placed on the anode side of each membrane electrode assembly in a stack. A catalyst layer is placed on the gas-feed
side of the cloth to catalyze oxidation of CO, utilizing the oxygen introduced in a small air-bleed. The purpose is to enhance the
effectiveness of the air-bleed to achieve a high CO tolerance. We synthesized model RCA catalysts based on transition metal
oxides, which were characterized using scanning electron microscopy, energy-dispersive spectroscopy, X-ray diffraction, thermo-
gravimetric analysis, and Brunauer-Emmett-Teller techniques. Here we present performance data from polarization curves mea-
sured on fuel cells with air bleed using these model RCA catalysts. The results indicate that by using an RCA and air-bleed it is
possible to raise the CO tolerance of a Pt/C anode to the same level as that of an anode with a Pt-Ru/C catalyst with similar
platinum loading and without other means of CO mitigation. The RCA represents a built-in safety net for CO transients, which
could be applied to PEMFCs destined to operate on a reformer-derived, hydrogen-rich fuel stream.
© 2005 The Electrochemical Society. ͓DOI: 10.1149/1.1851057͔ All rights reserved.
Manuscript submitted December 3, 2003; revised manuscript received July 19, 2004. Available electronically January 13, 2005.
During the transition to a hydrogen economy, early practical
polymer exchange membrane fuel cells ͑PEMFCs͒ for automotive
and large stationary applications are expected to operate on
reformate,2 obtained by reforming fuels such as gasoline, diesel,
natural gas, or methanol. Assuming cell operation below 100°C,
residual carbon monoxide is well known to poison the platinum
catalyst surface at levels as low as 10 ppm.2 For development of
PEMFC systems, the CO problem will not be eliminated until totally
durable PEMFC systems for operation at 130-160°C and/or eco-
nomical and green hydrogen production and storage are demon-
strated. Although there are major efforts channeled into these areas,
it remains of interest what to do about CO mitigation in proven,
low-temperature PEMFCs, both for Pt/C and for more tolerant al-
loys. Under these conditions, the case for on-board reforming of
gasoline for automobiles is currently less than compelling, chiefly
because of the need for several bulky stages of CO cleanup. In
gasoline reforming, relatively simple water gas shift reactors typi-
cally achieve a CO level of 5000-10000 ppm. Further reduction in
the CO level may require several stages of preferential oxidation
͑PROX͒. The complex processing needed to remove CO to
Ͻ10 ppm would have a high capital cost, volume, and weight pen-
alty. Hence, it would be desirable to simplify the required on-board
fuel processing. This, in turn, would be aided by raising the limit on
CO in reformate feed to fuel cells from 10 ppm by one to three
orders of magnitude.
In methanol reforming, the amount of CO produced is lower.
There seems to be a good prospect for development of a simple
steam reformer, producing a reformate containing less than 1000
ppm CO.3 Combining this general route with purification technolo-
gies in a single reactor has been reported to have a capability of
producing hydrogen of six nines purity.4 In the longer term, it is
expected that hydrogen will be produced at fueling stations by a
variety of processes, including reforming, and stored in liquid form
or as a highly pressurized gas, with on-vehicle storage as com-
pressed gas or a hydride. When hydrogen is produced by reforming
another fuel and stored, it may contain some CO, even though the
minimum quality standard under international consideration5 is ISO
type I grade A specified at 1 ppm maximum CO level.6
For stationary applications of PEMFCs operated on a reformate
even with CO cleanup, there remains a likelihood of cells receiving
transient high levels of CO during start-up or changes in operating
requirements.
To achieve a high level of CO tolerance, including tolerance to
brief transients far above steady-state levels, there are three main
categories of options. The first is to operate a high-temperature
PEMFC with a suitable, alternative membrane, for instance,
H3PO4-doped polybenzimidazole ͑PBI͒,7,8 which operates at about
200°C. The second is to use a conventional, Nafion-type membrane
at 80-100°C, with an alloyed platinum catalyst having increased
intrinsic tolerance to CO, such as Pt-Ru or Pt-Mo.9,10 A third option
is to use special measures at the cell operational stage. For instance,
operation with pulsed current11 has been suggested but would be
difficult for large-scale power supplies and has only been demon-
strated as an enhancement to the use of a Pt-Ru anode catalyst. For
a platinum-catalyzed anode ͑with no alloying elements͒ in a fuel cell
operated below 100°C, the only practical solution demonstrated is
the use of a small air-bleed to the anode gas stream. For instance, in
previous work at Los Alamos National Laboratory, an air-bleed ͑or
pure oxygen12-14͒ was added to the fuel stream to catalyze chemical
oxidation of CO, originally by the platinum anode catalyst itself.13
This method showed promise of extending the long-term tolerable
level of CO to several tens of ppm. Variations on this concept in-
clude the addition of hydrogen peroxide as a source of oxygen.15
In ongoing research into the air-bleed technique, it was found
more effective for the anode gas to make contact with a CO oxida-
tion catalyst before reaching the anode active layer.1 This has led to
the development, in our laboratory, of the reconfigured anode ͑RCA͒
system,1 which is shown schematically in Fig. 1. A thin layer of a
chemical catalyst is immobilized on the side of the gas distribution
cloth adjacent to the flow field plate. This is designed to perform a
low-temperature preferential oxidation using the air-bleed added to
the fuel stream as the anode gas is distributed to the fuel cells and
before it reaches the electrochemically active layer. Our laboratory
reported previously that oxides of metals having two stable, low
oxidation states showed promise for CO removal from hydrogen and
reformate streams, even at a temperature as low as 80°C.1
In this paper, we report new investigations on model catalysts:
carbon-supported oxides of the three transition metals, iron, cobalt,
and copper, which were prepared by thermal decomposition of the
acetate salts of the metals. These catalysts have been tested in larger
fuel cells ͑active area 50 cm2 rather than 5 cm2) with Nafion-based
membrane electrode assemblies ͑MEAs͒ having different anode and
* Electrochemical Society Active Member.
a Present address: Division of Natural Sciences, Campbellsville University, Camp-
bellsville, Kentucky 42718, USA.
z E-mail: paadcock@campbellsville.edu
Downloaded on 2015-06-17 to IP 193.0.65.67 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A460
Journal of The Electrochemical Society, 152 ͑2͒ A459-A466 ͑2005͒
Figure 1. Schematic diagram1 of a PEMFC with an RCA: ͑1͒ nonprecious-
metal-based chemical catalyst layer; ͑2͒ anode gas distribution cloth; ͑3͒
Pt-based anode electrocatalyst layer; ͑4͒ ionomer membrane; ͑5͒ Pt-based
cathode electrocatalyst layer; ͑6͒ cathode gas distribution cloth.
Figure 2. Schematic diagram showing two conditions necessary for ‘‘full
tolerance’’ attained using a reconfigured anode and air-bleed.
cathode combinations, as described in Table I. Type A represents
basic PEMFC technology. The second listed ͑type B͒ is a high-
tolerance anode with a high-performance cathode. Type C represents
an intermediate-tolerance anode combined with a high-performance
cathode. For evaluation of fuel cell performance, a practical defini-
tion of full tolerance is voltage losses no greater than 5% at a par-
ticular cell current in the presence of CO relative to its absence. For
this definition to be applicable to the reconfigured anode, the condi-
tions under investigation should be such that in the absence of a
reconfigured anode catalyst layer, the cell voltage drops significantly
in response to CO, even with air-bleed. This is shown schematically
in Fig. 2. In cells of type B as defined in Table I, full tolerance was
typically observed with air-bleed even with no RCA catalyst layer.
Therefore, it was impossible to measure a significant improvement
by adding the RCA catalyst. For the purpose of evaluating RCA
catalysts at relatively high CO levels ͑100-500 ppm͒, we found type
C cells ͑with lower anode loading of Pt-Ru/C͒ useful. Although
there is already literature data for effects of CO on PEMFCs without
and with air-bleed, we include in this paper a baseline study on the
three types of MEA described previously under our particular test
conditions. Future papers will be devoted to other studies using
these model catalysts, including endurance tests and investigations
aimed at elucidating the mechanism of RCA activity.
given period. After cooling, a spatula or a mortar and pestle was
used to produce a fine powder suitable for ink preparation. The
mixing ratios and firing temperatures, Tmax , for each model catalyst
are given in Table II.
Characterization.—The surface area of each catalyst powder
͑dried at 150°C under vacuum for 12 h͒ was determined with nitro-
gen at cryogenic temperatures using the Brunauer-Emmett-Teller
͑BET͒ method.
TGA measurements were made using a Perkin-Elmer TGA 7
controlled by a PC. Typical run conditions were from 25 to 800°C at
2°C/min under an atmosphere of either forming gas ͑6% H2 /Ar) or
cylinder air ͑20% O2 /N2).
X-ray powder diffraction measurements were made with a Si-
emens D5000 diffractometer fitted with a diffracted beam graphite
monochromator, using 40 kV accelerating voltage and 35 mA beam
current. A zero-background quartz sample holder was used to sup-
port the powder samples. Analysis of the X-ray diffraction ͑XRD͒
data was carried out using the SHADOW full profile refinement
package ͑Materials Data Corp.͒.
Each sample powder was mixed with iso-propanol, painted onto
a piece of mirror-flat silicon, and air-dried prior to energy-dispersive
spectroscopy ͑EDS͒ quantification and elemental mapping using an
EDAX instrument coupled to a Philips XL 20 SEM at an excitation
voltage of 10.0 kV.
Experimental
Catalyst preparation.—The acetate decomposition method
was based on a method reported previously.16 The preparation
involved mixing the acetate salt, Fe(ac)2 , Co(ac)2•4H2O, or
Cu(ac)2•H2O ͑all from Aldrich, 98ϩ%) where ac ϭ CH3COOϪ,
with water and carbon black first with a spatula, then sonicating,
followed by stirring with a stirrer bar overnight. The mixture was
slowly dried on a hot plate, gently evaporating the water. The dry
mixture in an evaporating basin was fired in a preset oven for a
Fuel cell tests.—MEAs of 50 cm2 active area were fabricated by
the decal method of pressing thin electrocatalyst layers onto Nafion
112.17 The electrocatalysts were from E-TEK ͑20% Pt/C and
Pt3Cr/C at 20% Pt͒ and Tanaka Kikinzoku Kogyo K.K. ͑TKK;
Pt2Ru3 /C at 54% Pt͒. Gas diffusion layers ͑also from E-TEK͒ were
ELAT single-sided V2.2 for the cathode and ELAT double-sided for
Table I. Specification of three types of MEA, all based on Nafion 112.
Anode
Cathode
Loading
Loading
Type
Catalyst
(mg Pt cmϪ2
)
Catalyst
(mg Pt¢mϪ2
)
A
B
C
20% Pt/C ͑E-TEK͒
Pt-Ru/C ͑54% Pt, TKK͒
Pt-Ru/C ͑54% Pt, TKK͒
0.2
0.2
0.1
20% Pt/C ͑E-TEK͒
Pt3Cr/C ͑20% Pt, E-TEK͒
Pt3Cr/C ͑20% Pt, E-TEK͒
0.2
0.4
0.4
a
a TKK ϭ Tanaka Kikinzoku Kogyo.
Downloaded on 2015-06-17 to IP 193.0.65.67 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 ͑2͒ A459-A466 ͑2005͒
A461
Table II. Details of preparation of each carbon-supported RCA
catalyst.
Metal
Fe
Co
Cu
Metal content ͑g͒
Carbon black ͑g͒
4.47
4.00
2.36
4.00
5.28
4.00
a
Tmax ͑°C͒
Time at Tmax ͑min͒
Recovered material ͑g͒
260 Ϯ 10
80
10.10
330 Ϯ 5
60
6.69
280 Ϯ 5
20
10.1
a VULCAN XC-72.
the anode. The standard methods of RCA ink preparation and RCA
layer preparation were reported previously.1 Standard RCA catalyst
loadings ͑dry ink͒ were in the range 1.0-1.5 mg cmϪ2. All fuel cell
tests were carried out at 80°C, at an anode stoichiometric ratio of 1.5
and a cathode air flow rate of 2100 sccm, with anode and cathode
backpressures of 30 psig. These cathode conditions help to eliminate
cell problems due to insufficient water removal in order to isolate
anode-related effects for quantitative measurement. The anode and
cathode feed gases, respectively, were humidified at 105 and 80°C.
Prior to humidification, synthetic reformate consisted of 40% H2 ,
35% N2 , and 25% CO2 on a molar basis.
Each combination of RCA layer and MEA was tested in a single
cell of 50 cm2 geometric active area. Tests were carried out on hy-
drogen and a 40% hydrogen (ϩ35% N2 and 25% CO2), synthetic
reformate with up to 500 ppm CO, using added air-bleeds of 6% or
less calculated on the basis of total moles of dry gas per unit time in
the anode input stream prior to air-bleed. The performance of test
cells was compared to that of standard cells, otherwise identical but
without the RCA modification of the anode gas distribution cloth.
Test conditions were always allowed to settle for at least 0.5 h be-
fore beginning to measure a polarization curve. CO tolerance was
evaluated from the net effect of CO and air-bleed on the cell voltage
at 0.5 A cmϪ2, for a cell operating on either hydrogen or simulated
40% hydrogen reformate fuel. For selected conditions, longer term
behavior, including life-test data, will be reported later.
Results and Discussion
Baseline fuel cell studies with air-bleed alone.—For a 99.9ϩ%
hydrogen feed stream, Fig. 3 shows single-cell polarization curves
for three levels of CO, in the absence and presence of a small air-
bleed, for three types of cell with no RCA catalyst on the anode
backing layer. Using a Pt/C anode ͑Fig. 3a͒, 2% air gave a little less
than full tolerance at 100 ppm CO and well below full tolerance at
200 ppm CO. For full tolerance, the levels of air-bleed required were
3 and 6%, respectively.
With a Pt-Ru/C anode (0.2 mg Pt cmϪ2; Fig. 3b͒, the anode
showed much-reduced susceptibility to poisoning by a given CO
concentration, as is already known from the literature.18 In the
present experiments, measurements at 0.5 A cmϪ2 indicated that full
tolerance to 100 ppm CO in hydrogen could be achieved with a 1%
air-bleed ͑curve not shown in Fig. 3b͒, and 200 ppm CO could be
tolerated with a 2% air-bleed. Therefore, under these conditions it
was impossible to determine whether there is an improvement by
adding RCA catalyst with air-bleed. However, when the Pt-Ru/C
loading on the anode was halved ͑Fig. 3c͒, 4% air-bleed was needed
for full tolerance to 100 ppm CO, and 5% air-bleed was needed for
full tolerance to 200 ppm CO. As seen in Fig. 3c, 2% air-bleed gave
well below full tolerance to 200 ppm CO.
Figure 3. Polarization curves for 99.9ϩ% hydrogen showing effects of CO
and air-bleed with no reconfigured anode layer: ͑a͒ Pt/C (0.2 mg Pt cmϪ2
)
anode; ͑b͒ Pt-Ru/C (0.2 mg Pt cmϪ2
)
anode; and ͑c͒ Pt-Ru/C
(0.1 mg Pt cmϪ2) anode.
For a synthetic, 40% hydrogen reformate feed stream, Fig. 4
shows single-cell polarization curves for up to four levels of CO for
the three MEA configurations discussed previously, without an RCA
layer on the anode backing cloth. Data with an air-bleed are shown
for 100, 250, and 500 ppm CO. Data without air-bleed are shown for
no CO and 100 ppm CO. Using a Pt/C anode ͑Fig. 4a͒, 2% air-bleed
gave less than full tolerance to 100 ppm CO. About 4% air-bleed
͑result not shown in Fig. 4a͒ was needed for full tolerance. At a
higher CO level ͑250 ppm͒, data are shown for 4% air-bleed, but
even 6% air-bleed did not approach full tolerance. In all these cases
there is clearly much scope for an RCA layer to produce an im-
provement in the utilization of the air-bleed, and perhaps to reduce
the air-bleed necessary for full tolerance.
Downloaded on 2015-06-17 to IP 193.0.65.67 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A462
Journal of The Electrochemical Society, 152 ͑2͒ A459-A466 ͑2005͒
Figure 5. TGA data19 for preparation of each catalyst from acetate salt,
run in 20% O2 /N2 ͑reproduced with permission of the Materials Research
Society͒.
electrocatalyst. Under these conditions, only marginal benefits could
be obtained with an RCA layer. When the Pt-Ru/C loading on the
anode was halved ͑Fig. 4c͒, 2% air-bleed was insufficient for full
tolerance to 100 ppm CO at moderate to high current density. An
air-bleed of at least 3% was needed at this level of CO. At 250 ppm
CO, 4% air-bleed did not give full tolerance and the curve for 6%
air-bleed was similar. At 500 ppm CO, 6% air-bleed gave much less
than full tolerance. These results indicate scope for an RCA layer to
reduce the air-bleed necessary for full tolerance.
Composition and structure of RCA catalysts.—To gain insight
into the processes of thermal decomposition of acetates in air onto
carbon black, used to prepare the RCA catalysts, Fig. 5 presents the
thermogravimetric analysis ͑TGA͒ in cylinder air ͑20% O2 /N2) of
each of the starting acetates. On each trace, an arrow indicates the
firing temperature for the catalyst preparation. Cobalt and iron had
definite end states, Co3O4 and Fe2O3 , respectively. Copper acetate
was converted to Cu2O by about 255°C and oxidized to CuO as the
temperature increased above this.
The characteristics of each catalyst material according to BET,
EDS, XRD, and TGA are summarized in Table III. For comparison
purposes, the specific surface area of untreated XC-72 carbon black
Table III. Summary of RCA catalyst characteristics.
Element
Fe
Co
Cu
Amount in catalysta ͑wt %͒
44
137
35
116
52
62
Catalyst specific surface area
(m2 gϪ1
)
Distribution of element
Even
Clusters
Not
determined
Figure 4. Polarization curves for 40% hydrogen synthetic reformate for
various combinations of CO and air-bleed with no reconfigured anode layer:
͑a͒ Pt/C (0.2 mg Pt cmϪ2) anode; ͑b͒ Pt-Ru/C (0.2 mg Pt cmϪ2) anode; and
͑c͒ Pt-Ru/C (0.1 mg Pt cmϪ2) anode.
XRD
Major component
Crystallite size ͑nm͒
Minor component
Approximate proportion
͑metal atom %͒
Crystallite size ͑nm͒
TGA
CuO
32
Cu2O
30
Fe3O4
20
Fe2O3
30
Co3O4
12
CoO
10
9
10
36
For a Pt-Ru/C anode (0.2 mg Pt cmϪ2), Fig. 4b shows that 100
ppm CO in 40% hydrogen synthetic reformate was tolerated fully,
with a 2% air-bleed and no RCA layer. Even 500 ppm CO was
tolerated fully with a 6% air-bleed and no RCA layer. Hence, air-
bleed alone is much more beneficial for Pt-Ru/C than for a Pt/C
Mass loss ͑%͒
Temperature range ͑°C͒
14.2
250-510
12.3
225-355
11.4
170-370
a Calculated from starting materials and final mass.
Downloaded on 2015-06-17 to IP 193.0.65.67 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 ͑2͒ A459-A466 ͑2005͒
A463
Figure 6. XRD and TGA in 6% H2 /Ar ͑inset͒ of Vulcan XC-72-supported
cobalt oxide RCA catalyst19 ͑reproduced with permission of the Materials
Research Society͒.
was found by BET analysis to be 245 m2 gϪ1, with our instrument,
which measures within 5% of NIST standards. More details about
each catalyst have been given elsewhere.19
For the supported cobalt oxide catalyst ͑Fig. 6͒, the TGA in
forming gas showed a single mass loss of 12.3%, which is on the
low side for Co3O4 to Co based on 35.3% Co ͑calculated from the
data in Table II͒. XRD found mainly Co3O4 with about 10% CoO,
which is consistent with the TGA figure. The main component had a
crystallite size of 12 nm, and the minor component was of similar
size. From EDS elemental maps, there appeared to be some segre-
gation of the cobalt oxide into clusters.
For the supported iron oxide catalyst, the TGA in forming gas
showed two mass losses between 200 and 600°C, and a continuous
mass loss process from about 600 to 800°C. XRD showed that the
catalyst contains magnetite (Fe3O4) and hematite (Fe2O3). The
crystallite sizes were calculated as 19.5 and 8.8 nm, respectively.
The distribution was approximately 70% magnetite and 30% hema-
tite. EDS elemental maps showed the iron to be distributed over the
whole surface of the carbon. The two mass losses below 600°C are
attributed to reduction of Fe2O3 to Fe3O4 and reduction of Fe3O4 to
metallic iron. A model physical blend of carbon black, magnetite,
and hematite was also investigated by running a TGA in forming
gas. With this blend, there was no significant mass loss above
600°C. This suggests that for the actual catalyst, intimate contact
between the reduced catalyst species and the carbon black catalyzes
methanization of carbon above 600°C.
For the supported copper oxide catalyst, the TGA in forming gas
showed a mass loss of 11.4% between 170 and 370°C. XRD showed
a mixture of CuO ͑about 70%͒ and Cu2O ͑about 30%͒ along with
carbon and an impurity, which may be an iron oxide. Scanning
electron microscopy ͑SEM͒ showed large clusters of copper oxide
spread unevenly over flakes of agglomerated carbon black.
Figure 7. Effects on CO tolerance of various reconfigured anode layers
compared to no such layer, for cells with air-bleed operating on 99.9ϩ%
hydrogen: ͑a͒ low CO levels and ͑b͒ higher CO levels. Specifications of
MEA types A and C are given in Table I. CO tolerance measured by cell
voltage difference at 0.5 A cmϪ2 from neat hydrogen. Dotted horizontal line
indicates maximum voltage loss meeting definition of ‘‘full tolerance.’’ Ab-
sence of a bar means no data to report for that particular condition.
200 ppm CO ϩ 4% air. The supported cobalt oxide catalyst again
produced a smaller improvement than the iron-based catalyst.
Tests were carried out in cells with a Pt-Ru/C anode at a reduced
loading ͑0.1 mg Pt cmϪ2) and a Pt3Cr/C cathode ͑Type C in Table
I͒. In hydrogen containing 500 ppm CO ͑Fig. 7b͒, the combination
of 6% air-bleed with supported iron oxide RCA catalyst gave full
tolerance. Other conditions for full tolerance were 100 ppm CO
ϩ 3% air ͑Fig. 7a͒, 200 ppm CO ϩ 4% air ͑Fig. 7b͒. The sup-
ported copper oxide catalyst gave similar results at the lower CO
levels but did not allow full tolerance at 500 ppm CO. With this type
of MEA, the supported cobalt oxide catalyst performed as well as
iron oxide at 100 or 200 ppm CO. No data is currently available for
500 ppm CO.
Cell performance on hydrogen feed with a reconfigured anode
and air-bleed.—For 100 ppm CO, Fig. 7 compares the performance
of several different RCA layers using air-bleed to the performance
with no RCA layer. In cells with Pt/C electrocatalyst layers
(0.2 mg Pt cmϪ2) for both the anode and the cathode ͑type A in
Table I͒, the supported iron oxide RCA catalyst showed full toler-
ance to 100 ppm CO, even at 2% air-bleed, and appeared particu-
larly promising in this test. Under the same conditions, the sup-
ported cobalt oxide catalyst showed only marginally increased CO
tolerance with 2% or 4% air-bleed. At higher CO levels ͑Fig. 7b͒,
the supported iron oxide catalyst continued to perform extremely
well, with full tolerance even to 500 ppm CO with 6% air, as well as
Downloaded on 2015-06-17 to IP 193.0.65.67 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A464
Journal of The Electrochemical Society, 152 ͑2͒ A459-A466 ͑2005͒
Table IV. Summary of conditions for full tolerance in hydrogen.
Minimum air-bleed ͑%͒ for full tolerance to CO in hydrogen
Anode
Pt/C at 0.2 mg Pt cmϪ2
Pt-Ru/C at 0.1 mg Pt cmϪ2
CO ͑ppm͒
100
RCA
Catalyst
100
3
Ͻ2
3
200
6
Ͻ4
6
500
200
500
None
4
3
5
4
Ͻ4
Ͼ6
6
Ͼ6
6
Ͼ6
Ͼ6
FeOx /C
CoOx /C
CuOx /C
NDa
ND
3
ND
ND
4
ϳ3
a ND ϭ Not determined.
catalyst. In contrast, at 6% air-bleed with a Pt/C anode, copper oxide
gave a result marginally better than iron oxide and cobalt oxide had
no benefit.
Figure 8b shows the net effect of very high levels of CO ͑250
and 500 ppm͒ and air-bleed for each catalyst. In cells having MEAs
with a Pt-Ru/C anode electrocatalyst at low loading ͑series C͒, the
iron oxide and copper oxide RCA catalysts displayed similar activ-
ity. In reformate containing 250 ppm CO, 6% air-bleed gave full
tolerance with any of the three supported oxide RCA catalysts. This
is remarkable because the CO:H2 ratio was equivalent to that for
625 ppm CO in hydrogen. Not surprisingly, full tolerance has not
been achieved at 500 ppm CO in a 40% hydrogen reformate. The
best result is the one for supported copper oxide in Fig. 8b, for
which at 0.5 A cmϪ2, the remaining voltage loss due to CO is about
15%. This CO level represents a CO:H2 ratio equivalent to that for
1250 ppm CO in hydrogen. For cells with Pt/C anode electrocata-
lyst, no reliable, quantitative data was obtained in this range. Pre-
liminary experiments had suggested the possibility of complete tol-
erance for cobalt-based RCA catalysts and 4-6% air-bleed,20 but this
was not borne out in further tests.
Discussion.—Table IV shows the effect of each type of RCA
catalyst layer on the minimum air-bleed required for full tolerance at
three different CO levels in 99.9ϩ% hydrogen for two types of
anode. For a Pt/C anode, the iron oxide RCA catalyst gave a signifi-
cant decrease in air requirements. For Pt-Ru/C at 0.1 mg Pt cmϪ2
,
air-bleed could be reduced by about 1% for each RCA catalyst, at
least for 100 or 200 ppm CO. In contrast, at 500 ppm CO the iron
oxide RCA catalyst was the only one to achieve full tolerance with
6% air-bleed, the maximum level of air-bleed tested. Table V shows
the effect of each type of RCA catalyst layer on the minimum air-
bleed required for full tolerance at three different CO levels in 40%
hydrogen reformate for two types of anode. With Pt-Ru/C ͑0.1 mg
Pt cmϪ2), the iron oxide and copper oxide RCA catalysts gave a
Figure 8. Effects on CO tolerance of various reconfigured anode layers
compared to no such layer for cells with air-bleed operating on 40% hydro-
gen synthetic reformate: ͑a͒ low CO level and ͑b͒ higher CO levels. Speci-
fications of MEA types A and C are given in Table I. CO tolerance measured
by cell voltage difference at 0.5 A cmϪ2 from synthetic reformate with no
CO. Dotted horizontal line indicates maximum voltage loss meeting defini-
tion of ‘‘full tolerance.’’ Absence of a bar means no data to report for that
particular condition.
Table V. Summary of conditions for full tolerance in 40%-
hydrogen reformate.
Minimum air-bleed ͑%͒ for full tolerance to CO in hydrogen
Anode
Pt/C at 0.2 mg Pt cmϪ2
Pt-Ru/C at 0.1 mg Pt cmϪ2
CO ͑ppm͒
500 100
Performance with 40% hydrogen, synthetic reformate
feed.—Figure 8a shows the effects of a combination of 100 ppm CO
and various levels of air-bleed on similar cells incorporating each
individual catalyst in the backing layer. For a Pt/C anode, cobalt
oxide RCA catalyst was not beneficial but iron oxide was beneficial.
For the copper oxide catalyst, clear benefits were shown previously
at a smaller scale.1 For a Pt-Ru/C anode ͑0.1 mg Pt cmϪ2), at 2% or
4% air-bleed, the best RCA catalyst was the iron oxide followed by
copper oxide and then cobalt oxide, which was better than no RCA
RCA
Catalyst
100
200
200
500
None
4
3
4
3
2
4
2
Ͼ6
Ͼ6
Ͼ6
Ͼ6
Ͼ6
ND
Ͼ6
6
6
Ͼ6
Ͼ6
Ͼ6
Ͼ6
FeOx /C
CoOx /C
CuOx /C
NDa
ND
ND
Ͼ6
a ND ϭ Not determined.
Downloaded on 2015-06-17 to IP 193.0.65.67 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 ͑2͒ A459-A466 ͑2005͒
A465
significant decrease in air requirements at 100 ppm CO. The air-
bleed could be reduced by about 1%. At 250 ppm CO, the iron oxide
and cobalt oxide RCA catalysts achieved full tolerance for the Pt-
Ru/C ͑0.1 mg Pt cmϪ2) anode with 6% air-bleed. With Pt/C electro-
catalyst (0.2 mg cmϪ2), the iron oxide RCA catalyst was effective at
up to 100 ppm CO.
According to the results presented, the optimal specification of
anode and reconfigured anode layer would depend strongly on the
fuel stream hydrogen content, the incoming CO level, and the
amount of air-bleed allowable. A low air-bleed ͑2% maximum͒ is
targeted by the U.S. Department of Energy ͑DOE͒ and auto manu-
facturers. For a 2% air-bleed added to either 99.9ϩ% or 40% hy-
drogen, the supported iron oxide catalyst gave the best RCA perfor-
mance, with about 100 ppm CO being the maximum level for full
tolerance. In 99.9ϩ% hydrogen using up to 6% air-bleed, up to 500
ppm CO could be tolerated using the iron oxide RCA catalyst. For
40% hydrogen, even at 6% air-bleed with an RCA layer, 250 ppm or
more CO poisoned cells with Pt/C anodes ͑with 0.2 mg Pt cmϪ2),
but MEAs with Pt-Ru/C at 0.1 mg Pt cmϪ2 gave better performance.
A 6% air-bleed with any of the three RCA catalysts gave full toler-
ance to 250 ppm CO.
A cell system with a reconfigured anode and Pt/C anode electro-
catalyst would be cheaper than cells with high levels of both plati-
num and another precious metal as the anode electrocatalyst. Ac-
cording to the literature, catalytic activity for hydrogen oxidation is
also maximized by using unalloyed Pt electrocatalyst rather than a
Pt alloy.10 Using a reconfigured anode and a Pt/C electrocatalyst, at
least 100 ppm CO in reformate can be tolerated for a time scale of
1-2 h under steady-state conditions, a performance, according to the
results in Fig. 4, equivalent to using a Pt-Ru/C electrocatalyst
(0.2 mg Pt cmϪ2) and no other form of CO mitigation. The system
with the reconfigured anode can certainly withstand transient CO
levels of the order of 100 ppm during continuous operation at gen-
erally lower CO. This may make it attractive for applications on fuel
derived from reforming, in combination with an additional CO
cleanup process easily able to give an intermediate CO output. Such
methods might include pressure swing adsorption ͑PSA͒ and use of
a hydrogen separation membrane ͑HSM͒, either of which would be
more economical for an intermediate CO output than, e.g., 1 ppm
CO, because of tradeoffs between purity and both recovery and re-
actor size. Even when used for production of purified hydrogen,
such an impurity cleanup process may not operate at its rated CO
output 100% of the time, especially during start-up or changes in
load. Therefore the transient safety net behavior of the RCA layer
could remain important.
electrocatalysts on anode potential23 ͑or chemical potential of spe-
cies at the anode, strongly influenced by the level of air-bleed͒
and/or (ii) changes in catalysis of either the CO air oxidation or the
reverse water gas shift reaction2,14,24 at the RCA catalyst. Other
speculations on the need for electronic contact with the fuel cell
have been given elsewhere.20 Investigation of these effects is be-
yond the scope of the present paper. Their determination will in-
volve a range of new studies of RCA catalyst behavior inside and
outside of a fuel cell environment.
Comparisons with literature systems.—The traditional PROX
process at 200°C or higher typically uses Pt/␥-Al2O3 as selective
CO oxidation catalyst.25 An alternative, Au/␣-Fe2O3 ,26 has been
investigated for several years for operation in the low-temperature
range around 80°C. Another type of catalyst known for some years
was Pt on a zeolite support.27 These catalysts and their recent vari-
ants all contained precious metals in the form of gold or a platinum
group metal. Our laboratory reported previously that the use of tran-
sition metal oxide catalysts ͑like CuO and Fe2O3) in a PROX reac-
tion at around 80°C was unprecedented.1,16 In fact, the first reported
precious metal free, PROX catalysts for use outside a fuel cell at
around 100°C appear to be the recently reported copper oxide/ceria
͑or zirconia͒ materials.28,29 Significantly, Engelhard21,30 has reported
on development of copper-containing catalysts, which also con-
tained platinum group metal promoters. We note also that cobalt and
iron or their oxides have been reported as cocatalysts with Pt for this
reaction at low temperature.23,31 There has been some interest in
copper, cobalt, and iron as potentially suitable elements going back
several years.25 A major difference between our catalysts and those
studied previously for the PROX reaction is the electronically con-
ductive carbon support material, chosen because high electronic
conductivity is essential in the reconfigured anode geometry
͑Fig. 1͒.1
In terms of carrying out the PROX reaction within a fuel cell
assembly, a similar system has been investigated by Rohland and
Plzak.32 It has been reported that in a cell with Pt-Ru anodes using
ϳ5% air ͑1% oxygen͒ combined with a ‘‘Au catalyst on Fe2O3
support,’’ the tolerable CO level in hydrogen was between 250 and
1000 ppm.32 Other systems of similar configuration have also been
suggested or investigated by Wilkinson et al.33 and Haug et al.10
These devices used a precious metal ͑Pt or Ru, respectively͒ as the
catalyst for CO oxidation. In the ‘‘ruthenium filter,’’10 instead of
placing ruthenium in the anode electrocatalyst, the ruthenium was
placed as a CO filter layer between the gas diffusion layer and the
anode catalytic layer. With mainly ionic contact to the anode elec-
trocatalyst, this is a different configuration from that in Fig. 1. The
best results were obtained when oxygen was added at a concentra-
tion of at least 1% of the hydrogen concentration. A platinum anode
could tolerate less than 100 ppm CO in 99.9ϩ% hydrogen, even
with 2% oxygen addition. The results in 40% hydrogen reformate
were inferior to this. Similarly, researchers at Johnson-Matthey have
investigated a ‘‘bilayer anode.’’34 In this bilayer, the first Pt-Mo
pretreats the stream to remove most of the CO so that Pt-Ru can
tolerate the remainder, while the latter catalyst performs most of the
hydrogen electrocatalysis. The bilayer anode worked well at 100
ppm CO. At 400 ppm, there was 50 mV loss. At high CO levels, the
increases in the loss were gradual compared to published data for
other anode eletrocatalysts.34
Besides freedom from a requirement for precious metals, addi-
tional advantages of the reconfigured anode technique for maximi-
zation of CO tolerance and smoothing of transient spikes include the
following:1 (i) it works with the proven MEA material Nafion in its
preferred temperature range ͑80-100°C͒ and (ii) it does not signifi-
cantly increase the dimensions or mass of a fuel cell stack or re-
former hardware.
Ideal catalysts would function by facilitating the reaction be-
tween CO and oxygen at 80°C,1 but the water gas shift reaction
could possibly also play a positive role. The tests reported herein
have shown excellent results on humidified hydrogen, if the sup-
ported iron oxide catalyst with mixed oxidation states is used in an
RCA layer. As noted previously, the existence of multiple valence
states is common to elements having active oxides for reconfigured
anode layers.1,16 According to other authors, compounds such as
Fe2O3 and Fe3O4 can support active oxygen sites, which facilitate
oxidation of CO and can be easily replenished from the oxygen in
the air-bleed.21,22
Note on stability of RCA chemical catalysts.—The present and
earlier1 studies on nonoptimized, transition metal oxide catalysts
demonstrate the utility of the reconfigured anode using model cata-
lysts. Iron-oxide-promoted, alumina-supported platinum has been
reported by Engelhard Corporation23 as an efficacious and economi-
cal catalyst for CO removal in an anode fuel stream prior to humidi-
fication. Notwithstanding this, there is a potentially serious concern
about the long-term stability of compounds such as iron oxides in a
humidified stream within a fuel cell. Further study is required to
The complex differences observed between the different electro-
catalysts, RCA catalysts, and air-bleed levels highlight the gaps in
existing knowledge of the mechanism of improvement of CO toler-
ance in the presence of RCA catalyst and air-bleed. The differences
may be due to such factors as (i) a dependence of the behavior of
Downloaded on 2015-06-17 to IP 193.0.65.67 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A466
Journal of The Electrochemical Society, 152 ͑2͒ A459-A466 ͑2005͒
Los Alamos National Laboratory assisted in meeting the publication
costs of this article.
identify practical catalysts, with both high intrinsic activity and high
stability, to maximize both performance and lifetime under operat-
ing fuel cell conditions.
References
Further work.—Apart from endurance tests, our current and
pending work on the reconfigured anode scheme is concentrated on
understanding the mechanism of catalytic activity. From the results
presented thus far, it remains unclear whether there can be a contri-
bution from the water gas shift pathway, or if the activity is influ-
enced by the electrochemical potential induced at the catalyst sur-
face by its placement in the fuel cell.
The materials studied so far are model catalysts, the long-term
stability of which has not been demonstrated in a PEMFC. Other
materials such as perovskites or spinels, or the recently reported
doped ceria and zirconia materials,28,29 offer greater likelihood of
high stability combined with catalytic activity. Ultimately, we aim to
maximize the durability of low-temperature PEMFCs using Pt/C
anode electrocatalysts exposed to moderate levels of CO. Results of
further studies will be reported in subsequent papers.
1. F. A. Uribe, J. A. Valerio, F. H. Garzon, and T. A. Zawodzinski, Electrochem.
Solid-State Lett., 7, A376 ͑2004͒.
2. S. Gottesfeld and T. A. Zawodzinski, in Advances in Electrochemical Science and
Engineering, Vol. 5, R. C. Alkire, H. Gerischer, D. M. Kolb, and C. W. Tobias,
Editors, p. 195, Wiley VCH, Weinheim, Germany ͑1997͒.
3. J. Agrell, H. Birgersson, and M. Boutonnet, J. Power Sources, 106, 249 ͑2002͒.
4. J. Han, I.-S. Kim, and K.-S. Choi, J. Power Sources, 86, 223 ͑2000͒.
5. Draft ECE Compressed Gaseous Hydrogen Regulation, Rev. 12, p. 9, UNECE
GRPE Informal Group: Hydrogen/Fuel Cell Vehicles ͑2003͒.
6. Hydrogen fuel-Product specification, ISO 14687͑E͒ (1999) ϩ Cor1 (2001).
7. Y. Ma, A. Schechter, J. S. Wainwright, and R. F. Savinell, Abstract 55, The Elec-
trochemical Society Meeting Abstracts, Vol. 2001-2, San Francisco, CA, Sept 2-7,
2001.
8. E. W. Choe, B. Krishnamurthy, L. Xiao, H. Zhang, G. Calundann, and B. C.
Benicewicz, in Proton Conducting Membrane Fuel Cells III, J. W. Van Zee, M.
Murthy, T. Fuller, and S. Gottesfeld, Editors, The Electrochemical Society Proceed-
ings Series, Pennington, NJ, In press.
9. S. Mukerjee, S. J. Lee, E. A. Ticianelli, J. McBreen, B. N. Grgur, N. M. Markovic,
P. N. Ross, J. R. Gialombardo, and E. S. De Castro, Electrochem. Solid-State Lett.,
2, 12 ͑1999͒.
Conclusions
10. A. T. Haug, R. E. White, J. W. Weidner, and W. Huang, J. Electrochem. Soc., 149,
A862 ͑2002͒.
11. L. P. L. Carrette, K. A. Friedrich, M. Huber, and U. Stimming, Phys. Chem. Chem.
Phys., 3, 320 ͑2001͒.
The RCA system allows CO tolerance to be improved. For in-
stance, the CO tolerance of Nafion MEAs at 80°C with Pt/C anode
catalyst can be raised to a level similar to that normally achieved
with alloyed electrocatalysts, such as Pt-Ru/C. The RCA system is
characterized by relatively inexpensive CO removal catalysts
painted in a thin layer onto the flow-field side of the anode gas
distribution cloth. Ideal catalysts would function by facilitating het-
erogeneous catalysis of the reaction between CO and oxygen ͑from
anode air-bleed͒ at 80°C, but elucidation of the actual mechanism
requires further investigation using an expanded range of measure-
ment techniques.
12. S. Gottesfeld, U.S. Pat. 4,910,099 ͑1990͒.
13. S. Gottesfeld and J. Pafford, J. Electrochem. Soc., 135, 2651 ͑1988͒.
14. M. S. Wilson, T. E. Springer, T. A. Zawodzinski, and S. Gottesfeld, in Proceeding
of the 28th Intersociety Energy Conversion Engineering Conference, Vol. 1, p.
1203, American Chemical Society, Washington, DC ͑1993͒.
15. J. Divisek, H. F. Oetjen, V. Peinecke, V. M. Schmidt, and U. Stimming, Electro-
chim. Acta, 43, 3811 ͑1998͒.
16. F. A. Uribe and T. A. Zawodzinski, U.S. Pat. Appl. 20020119363.
17. M. Wilson, U.S. Pat. 5,211,984 ͑1983͒; M. Wilson, J. Valerio, and S. Gottesfeld,
Electrochim. Acta, 40, 355 ͑1995͒.
With a 2% limit on air-bleed, a reconfigured anode can withstand
transient CO levels in either hydrogen or reformate approaching 100
ppm during continuous operation at generally much lower CO. By
combining the RCA layer with an anode of higher intrinsic tolerance
to CO ͑modeled in this work by Pt-Ru/C at 0.1 mg Pt cmϪ2), the
transient CO level that can be tolerated is increased to above 100
ppm. In addition, increasing the air-bleed to 6% raises the maximum
tolerable CO level by a factor of 2-3.
In summary, the use of reconfigured anodes and a small air-bleed
is attractive for application as a safety net with Nafion-type PEM-
FCs operated below 100°C on reformer-derived fuel, and in terms of
the active center, iron oxides are at least as promising as copper
oxides. Further work on evaluation and development is in progress.
18. H. F. Oetjen, V. M. Scmidt, U. Stimming, and F. Trila, J. Electrochem. Soc., 143,
3838 ͑1996͒.
19. P. A. Adcock, E. L. Brosha, F. H. Garzon, and F. A. Uribe, Mater. Res. Soc. Symp.
Proc., 756, 365 ͑2003͒.
20. P. A. Adcock, S. Pacheco, E. Brosha, T. A. Zawodzinski, and F. A. Uribe, in Proton
Conducting Membrane Fuel Cells III, J. W. Van Zee, M. Murthy, T. Fuller, and S.
Gottesfeld, Editors, The Electrochemical Society Proceedings Series, Pennington,
NJ, In press.
21. X. Liu, O. Korotkikh, and R. Farrauto, Appl. Catal., A, 226, 293 ͑2002͒.
22. M. M. Schubert, S. Hackenberg, A. C. van Veen, M. Muhler, V. Plzak, and R. J.
Behm, J. Catal., 197, 113 ͑2001͒.
23. S. Mukerjee and R. C. Urian, Electrochim. Acta, 47, 3219 ͑2002͒.
24. F. A. de Bruijn, D. C. Papageorgopoulos, E. F. Sitters, and G. J. M. Janssen, J.
Power Sources, 110, 117 ͑2002͒.
25. S. H. Oh and R. M. Sinkevitch, J. Catal., 142, 254 ͑1993͒.
26. M. J. Kahlich, H. A. Gasteiger, and R. J. Behm, J. New Mater. Electrochem. Syst.,
1, 39 ͑1998͒.
27. M. Watanabe, H. Uchida, H. Igarashi, and M. Suzuki, Chem. Lett., 1995, 21.
28. G. Sedmak, S. Hocevar, and J. Levec, J. Catal., 213, 135 ͑2003͒.
29. P. Ratnasamy, D. Srinivas, C. V. V. Satyanarayana, P. Manikandan, R. S. Sethil
Kumaran, M. Sachin, and V. N. Shetti, J. Catal., 221, 455 ͑2004͒.
30. L. Shore, W. F. Ruettinger, and R. J. Farrauto, U.S. Pat. Appl. 2002131915 ͑2002͒.
31. J. Yan, J. Ma, P. Cao, and P. Li, Catal. Lett., 93, 55 ͑2004͒.
32. B. Rohland and V. Plzak, J. Power Sources, 84, 183 ͑1999͒.
33. D. P. Wilkinson, H. H. Voss, J. Dudley, G. J. Lamont, and V. Basura, U.S. Pat.
5,482,680 ͑1996͒.
Acknowledgments
We gratefully acknowledge funding from the U.S. DOE Office of
Hydrogen, Fuel Cells, and Infrastructure Technologies, and Los Ala-
mos National Laboratory for meeting publication costs. We also
thank Professor Tom Zawodzinski for helpful discussions, Tommy
Rockward and Judith Valerio for their participation in the prepara-
tion and testing of fuel cells, and Fernando H. Garzon, Eric L.
Brosha, and Rangachary Mukundan for advice and assistance with
the catalyst characterization.
34. S. Ball and D. Thompsett, Abstract FF5.1, p. 697, Abstracts of the Materials Re-
search Society Fall 2002 Meeting, Warrendale, PA ͑2002͒.
Downloaded on 2015-06-17 to IP 193.0.65.67 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).