A proton-conducting fuel cell operating with hydrocarbon fuels

Article abstract of DOI:10.1002/anie.200801667

Low temperature, high voltage! Direct oxidation of hydrocarbon fuels, including methane, ethane, propane, and butane, occurs over Pt-free anodes in a proton-conducting fuel cell in the temperature range 100-300°C. A Mo ₂C-ZrO₂/C anode (30 mg Mo₂C-ZrO₂ per cm₂) yields power densities equal to those obtained from Pt/C anodes (1-7 mg Pt per cm₂) and generates higher open-circuit voltages than the Pt/C anodes. (Figure Presented)

Full text of DOI:10.1002/anie.200801667

Angewandte
Chemie
DOI: 10.1002/anie.200801667
Fuel Cells
A Proton-Conducting Fuel Cell Operating with Hydrocarbon Fuels
Pilwon Heo, Kenichi Ito, Atsuko Tomita, and Takashi Hibino*
It is commonly believed that hydrogen is the most ideal fuel
for fuel cells because it has the highest energy conversion
efficiency and the most environmentally friendly properties
among the various possible fuels. Nearly half of the chemical
energy of hydrogen and oxygen is actually converted into
electricity with only water as a product. However, well-to-
wheels (WTW)analyses are necessary to evaluate fuel-cell
systems since the energy and environmental benefits of
hydrogen as a fuel strongly depend on the choice of hydrogen
production pathways. Indeed, the benefits of fuel-cell vehicles
using hydrogen as fuel are not so great if hydrogen is
produced from fossil fuels.[1–3] This is mainly because of the
energy loss in steam reforming hydrocarbons at high temper-
atures. Additional energy losses result from hydrogen storage
and purification, a problem common to all hydrogen produc-
tion pathways.
high tolerance towards CO.[11] Another important character-
istic is that these fuel cells allow the use of transition-metal
carbides as alternative anode materials to platinum.[12] The
present results demonstrate that these characteristics can
successfully address the challenges encountered in SOFCs
and PEFCs. The PCFCs used here consisted of MxC/C (M =
Mo, W, Nb, V, and Ti, 30 mg MxC per cm2) j Sn0.9In0.1P2O7 j Pt/
C (0.6 mg Pt per cm2). X-ray diffraction (XRD) patterns of
these anode materials showed that each metal carbide existed
as a single phase (see Figure S1 in the Supporting Informa-
tion). Polarization measurements were carried out after
supplying the fuel chamber with a mixture of propane and
water vapor. The Mo2C/C anode exhibited the best perfor-
mance for propane oxidation (Figure 1). Mo2C has catalytic
Direct utilization of hydrocarbons in fuel cells can be
expected to significantly improve their WTW efficiency.[4,5]
Indeed, solid oxide fuel cells (SOFCs)that use methane show
higher WTW efficiencies ( ꢀ 50%)than polymer electrolyte
fuel cells (PEFCs)with gases reformed from methane
(ꢀ 30%).[6] However, SOFCs present a number of inherent
challenges for transport applications, such as low mechanical
strength, slow start-up time, and serious anode deteriora-
tion.[7] The first two challenges are crucial for fuel-cell
vehicles. The development of PEFCs to meet this challenge
has also been attempted, but some serious challenges remain,
including low open-circuit voltages (OCVs), poor cell perfor-
mance, and high Pt loadings.[8,9] This is because operating
temperatures below 1008C are too low to achieve sufficiently
high reaction activity at the anode.
Figure 1. Overpotentials of various anodes for propane oxidation at
3008C. The loadingof the carbides was 30 mgcm ꢁ2. A mixture of
propane (7 vol%) and water vapor (42%) in argon was supplied into
the anode chamber at 60 mLminꢁ1
.
We report herein the investigation of the direct oxidation
of hydrocarbon fuels, including methane, ethane, propane,
and butane, over Pt-free anodes in a proton-conducting fuel
cell (PCFC)in the temperature range 100–300 8C. In3+-doped
SnP2O7 shows high proton conductivities above 10ꢁ1 Scmꢁ1
over this temperature range (see the Supporting Informa-
tion).[10] This material has also been explored for use as an
electrolyte in intermediate-temperature fuel cells, which are
characterized by excellent thermal stability up to 3008C and a
characteristics similar to those of platinum-group metals, its
catalytic activity is explained by the change in the electron
density of the d band of Mo upon carburization: the
introduction of C atoms into the Mo metal lattice causes a
contraction of the d band, and thus enhancing the d-electron
density to the d-electron density levels of Pt.[13] However, the
performance of Mo2C/C was still somewhat inferior to that of
Pt/C (Figure 1); the overpotential value of Mo2C/C was
154 mV at 50 mAcmꢁ2, which was lower than those of Pt/C
with Pt loadings of 1.0 and 2.5 mgcmꢁ2, but higher than those
of Pt/C with Pt loadings of 4.0 and 7.0 mgcmꢁ2. Our previous
study showed that the hydrogen oxidation activity of Mo2C
[*] Dr. P. Heo, K. Ito, Prof. Dr. T. Hibino
Graduate School of Environmental Studies, Nagoya University
Furo-cho, Chikusa-ku, Nagoya 464-8601 (Japan)
Fax : (+81)52-789-4206
was improved by impregnating
a molybdenum salt,
(NH4)6Mo7O24·4H2O, with a zirconium salt, ZrCl2O·8H2O,
and then carburizing the impregnated sample in the same
manner as Mo2C.[12] Zirconium-modified Mo2C/C had a
similar effect on hydrocarbon oxidation (Figure 2a); the
overpotential for propane oxidation decreased with increas-
ing content of Zr species and reached a minimum at a
Mo2C:ZrO2 weight ratio of 0.8:0.2. The minimum over-
potential value was 60 mV at 50 mAcmꢁ2, which was approx-
E-mail: hibino@urban.env.nagoya-u.ac.jp
Dr. A. Tomita
National Institute of Advanced Industrial
Science and Technology (AIST)
Simosidami, Moriyama-ku, Nagoya 463-8560 (Japan)
Supportinginformation for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200801667.
Angew. Chem. Int. Ed. 2008, 47, 7841 –7844
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7841

Communications
imately one-third of the overpotential of untreated Mo2C/C.
The XRD profile of Zr-modified Mo2C/C revealed that the
added Zr species was present as ZrO2 (Figure 2b). In
addition, the Mo2C diffraction peaks did not shift upon the
addition of ZrO2 to the catalyst, which suggested that Zr was
not incorporated into the Mo2C crystalline lattice. The
Scherrer formula was applied to estimate the Mo2C particle
size, which was seen to attain a minimum at a Mo2C:ZrO2
weight ratio of 0.8:0.2 (Figure 2a); the particle sizes were
28 nm and 14 nm for Mo2C/C and Mo2C-ZrO2/C, respectively.
A similar difference was observed in the Mo2C particle size, as
revealed by transmission electron microscopy (TEM)meas-
urements of the two catalysts (Figure 2c); the TEM images
showed that the Mo2C particle sizes were smaller for Mo2C-
ZrO2/C than for Mo2C/C. These results indicate that the
Mo2C particles show a higher dispersion with ZrO2 than
without ZrO2. It is likely that excessive sintering of Mo2C
during carburization is inhibited in the presence of ZrO2,
which results in the significantly improved catalytic activity of
Mo2C towards propane oxidation.
Fuel-cell tests were performed between 100 and 3008C
using propane as fuel (Figure 3a). The OCV increased with
temperature until 2008C and thereafter slightly decreased
with temperature. The temperature dependence observed
above 2008C is consistent with the thermodynamic prediction
suggesting that the anode potential is no longer limited by
reaction kinetics under such conditions. However, the OCV
value of 862 mV at 2008C is lower than the theoretical value
of 1080 mV. The lower OCV value is due to the physical
leakage of gas through the electrolyte and partial electron-
hole conduction in the electrolyte, as reported in previous
studies.[10] The voltage drop was greatly reduced by elevating
the temperature (Figure 3a). The resulting peak power
density reached 53 mWcmꢁ2 at 3008C. There is not a large
difference in peak power density between the Mo2C-ZrO2/C
and Pt/C (7 mg Pt per cm2)anodes; the peak power density
for Pt/C at 3008C was 60 mWcmꢁ2 (see Figure S2 in the
Supporting Information). More importantly, the OCVs
observed for Mo2C-ZrO2/C are higher than those for Pt/C;
the OCV for Pt/C at 3008C is only 679 mV. This point will be
discussed later. These results demonstrate that Mo2C-ZrO2/C
is a promising anode catalyst for direct hydrocarbon fuel cells
operating at intermediate temperatures.
The comparison of cell performance with various hydro-
carbon fuels was conducted at 3008C (Figure 3b). Although
the OCV ranged between 790 and 850 mV for all the
hydrocarbon fuels, the voltage drop depended quite strongly
on the hydrocarbon species; the voltage drop decreased in the
order butane < propane < ethane < methane. This corre-
sponded to the anode overpotential order observed at the
same temperature (see Figure S3 in the Supporting Informa-
tion). Thus, the present fuel cell yielded its highest peak
power density of 55 mWcmꢁ2 with butane as the fuel. This
suggests that higher hydrocarbon fuels are more desirable for
the operation of this fuel cell, which is suited to transport
applications since such fuels are easily stored.
Figure 2. a) Particle size of Mo2C and overpotential of Mo2C–ZrO2/C
for propane oxidation as a function of ZrO2 content in Mo2C–ZrO2/C.
The overpotential of the Mo2C–ZrO2/C anode was measured under the
same conditions as in Figure 1. b) XRD patterns of Mo2C–ZrO2/C with
different proportions of ZrO2. c) TEM micrographs of Mo2C/C and
Mo2C–ZrO2/C (Mo2C:ZrO2 =0.8:0.2).
To better understand the reaction of hydrocarbons at the
anode, the products in the outlet gas from the fuel-cell anode
at 3008C were analyzed using an online gas chromatograph.
7842 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 7841 –7844

Angewandte
Chemie
Figure 4. a) Product concentration for propane oxidation as a function
of current density over a Mo2C–ZrO2/C anode at 3008C. b) Cyclic-
voltammetry (CV) profiles of Pt/C and Mo2C–ZrO2/C anodes in a
mixture of propane and water vapor (solid line) and argon (dotted
line) at 3008C. The reference electrode was attached beside the
counter electrode on the surface of the electrolyte and exposed to pure
hydrogen.
Figure 3. Cell voltage and power density versus current density for a
fuel cell with a Mo2C-ZrO2/C anode (Mo2C:ZrO2 =0.8:0.2). a) T=100–
3008C, fuel=propane (7 vol%) and water vapor (42 vol%), flow
rate=60 mLminꢁ1, total pressure=1 atm. b) T=3008C, fuel= meth-
ane (7 vol%) and water vapor (14 vol%), ethane (7 vol%) and water
vapor (28 vol%), propane (7 vol%) and water (42 vol%) vapor, and
butane (7 vol%) and water vapor (56 vol%), flow rate=60 mLminꢁ1
total pressure=1 atm.
,
The rate-determining step in the overall oxidation of the
hydrocarbon to CO2 is thought to be Equations (4)and (5),
wherein the oxygen-containing species (OH)ads, formed
through the oxidation of water vapor, reacts with (C1)ads to
form CO2. This mechanism is similar to that reported for
methanol or CO oxidation in the presence of water.[15–17]
Cyclic voltammetry (CV)measurements allow us to
understand the potential regions corresponding to Equa-
tions (4)and (5.) For the Pt/C anode, the processes in
Equations (4)and (5)occurred at E = 0.54 V in the positive-
direction scan and the reaction in Equation (5)occurred, as
before, at E = 0.51 V in the negative-direction scan (Fig-
ure 4b). This means that the reaction in Equation (5) was
inhibited by the strong adsorption of (OH)ads on the Pt surface
at high potentials during the scan, and then restarted by the
reduction of the strongly adsorbed (OH)ads upon scan
direction reversal. Meanwhile, the CV profiles of the Mo2C-
ZrO2/C anode showed the oxidation peak of the reactions in
No product was observed at the OCV for any of the
hydrocarbons tested and CO2 was the only constituent of
the product gas during the cell discharge. The CO2 concen-
tration increased linearly with current density and
approached the theoretical value of CO2 concentration for
100% faradic efficiency (Figure 4a for propane and Figure S4
in the Supporting Information for other hydrocarbons). These
results demonstrate that the hydrocarbons are converted
directly into CO2, with a current efficiency of approximately
100%, according to the oxidation reaction shown in Equa-
tion (1).
CnH2nþ2 þ 2 n H2O ! ð6n þ 2ÞHþ þ ð6n þ 2Þeꢁ þ n CO2
ð1Þ
It has been reported that the oxidation product of a
hydrocarbon on Pt in acidic solutions is CO2, for which the
reaction mechanism shown in Equations (2)–(5) is proposed
(ads = adsorbed).[9,14]
Equations (4)and (5)at
E = 0.15 V during the positive-
direction scan (Figure 4b). This potential value is much lower
than that observed for the Pt/C anode, which is consistent
with the difference in OCV between the two anodes. Another
important result is that the Mo2C–ZrO2/C anode has no peak
potential corresponding to Equation (5)during the negative
potential scan, in contrast to the Pt/C anode, which is
indicative of a weak interaction between (OH)ads and the
surface sites of the catalyst. Thus, it is assumed that the
ðCnH2nþ2Þads ! ðCnH2nþ1Þads þ Hþ þ eꢁ
ðCnH2nþ1Þads ! ðC1Þads þ ðCnꢁ1Þads þ x Hþ þ x eꢁ
2 H2O ! 2ðOHÞads þ 2 Hþ þ 2 eꢁ
ð2Þ
ð3Þ
ð4Þ
ð5Þ
ðC1Þads þ 2ðOHÞads ! CO2 þ 2 Hþ þ 2 eꢁ
Angew. Chem. Int. Ed. 2008, 47, 7841 –7844
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7843

Communications
and the transient response of the anode potential was recorded using
an oscilloscope. The potential components of the fast drop and slow
decay were assigned to the IR (I: current density, R: electrolyte
resistance)drop and overpotential, respectively. The overpotential
was thus estimated by eliminating the IR drop from the anode
potential. The cell performance was measured by using the four-probe
method, wherein the ohmic resistance of the lead wires was negligible.
The outlet gas from the fuel chamber was analyzed using online gas
chromatography. The gas concentrations were obtained after the
steady-state composition of the reaction was attained.
turnover rate of the reaction in Equation (5)on the Mo C–
ZrO2/C anode is considerably higher than on the Pt/C anode.
2
The present results demonstrate that the direct oxidation
of hydrocarbons no longer needs to be carried out above
3008C. The operation of the fuel cell at 3008C can reduce the
start-up time as well as prevent carbon deposition on the
anode surface. The formation of carbon from propane over a
Ni/Al2O3 catalyst, one of the most active catalysts for coking,
begins at 4308C.[18] Another important result is that Mo2C can
be used in anodes for hydrocarbon oxidation. This is
significant for the conversion of hydrocarbon fuels because
Mo2C shows high tolerance toward sulfidation[19] as well as
carbon deposition.[20] However, the power densities obtained
in this study are substantially lower than those reported for
SOFCs.[1–3] One reason for this is the large thickness (1 mm)
of the electrolyte used; additional work needs to be done to
prepare thinner electrolyte films.
CV measurements: Another three-electrode arrangement was
used with Au and Pt/C electrodes supplied with pure hydrogen as
reference and counter electrodes, respectively (see Figure S6 in the
Supporting Information). The CV profiles were collected between 0
and 1 V vs. reversible hydrogen electrode (RHE)at a scan rate of
10 mVsꢁ1
.
Received: April 9, 2008
Revised: August 12, 2008
Keywords: catalysis · fuel cells · hydrocarbon fuels ·
.
molybdenum · zirconium
Experimental Section
Metal carbide anodes, MxC/C (M = Mo, W, Nb, V, and Ti), were
synthesized as follows. The corresponding metal salts (e.g.,
(NH4)6Mo7O24·4H2O)were impregnated in a carbon support (Black
Pearl, BET surface area: 1365 m2 gꢁ1). The impregnated sample was
reduced in a mixture of 10 vol% H2 in Ar at 5008C for 2 h and then
carburized in a mixture of 20 vol% CH4 in H2 at a temperature
between 700 and 12008C depending on the material (e.g., T= 7008C
for Mo2C)for 3 h. The loading of all the carbides was 30 mgcm ꢁ2. Pt/C
anodes (30 wt% Pt/C, 1–7 mg Pt per cm2, Black Pearl)were used for
comparison. A Pt/C cathode (10 wt% Pt/C, 0.6 mg Pt per cm2)was
purchased from E-TEK, Inc. The microstructure of the catalysts was
analyzed using X-ray diffraction (XRD, Shimadzu XRD-600)and
transmission electron microscopy (TEM, Hitachi H-800).
[1] M. Wang, J. Power Sources 2002, 112, 307.
[2] M. Wietschel, U. Hasenauer, A. de Groot, EnergyPolicy 2006,
34, 1284.
[3] A. M. Svensson, S. Møller-Holst, R. Glöckner, O. Maurstad,
Energy 2007, 32, 437.
[4] E. P. Murray, T. Tsai, S. A. Barnett, Nature 1999, 400, 649.
[5] S. Park, J. M. Vohs, R. J. Gorte, Nature 2000, 404, 265.
[6] Z. Zhan, S. A. Barnett, Science 2005, 308, 844.
[7] B. C. H. Steele, A. Heinzel, Nature 2001, 414, 345.
[8] O. Savadogo, F. J. Rodriguez, J. New Mater. Electrochem. Syst.
2001, 4, 93.
[9] W. S. Li, D. S. Lu, J. L. Luo, K. T. Chuang, J. Power Sources 2005,
145, 376.
[10] M. Nagao, A. Takeuchi, P. Heo, T. Hibino, M. Sano, A. Tomita,
Electrochem. Solid-State Lett. 2006, 9, A105.
[11] P. Heo, H. Shibata, M. Nagao, T. Hibino, M. Sano, J. Electro-
chem. Soc. 2006, 153, A897.
[12] P. Heo, M. Nagao, M. Sano, T. Hibino, J. Electrochem. Soc. 2007,
154, B53.
[13] J. S. Choi, G. Bugli, G. Djega-Mariadassou, J. Catal. 2000, 193,
238.
Sn0.9In0.1P2O7 was prepared as reported previously.[10] SnO2 and
In2O3 powders were mixed with H3PO4 (85%)and deionized water
and stirred at 3008C until the mixture formed a highly viscous paste.
The pastes were calcined in an alumina pot at 6508C for 2.5 h and
then ground with a mortar and pestle. Finally, the compound powders
were pressed into pellets under a pressure of 2 ” 103 kgcmꢁ2
.
Electrode and cell performance measurements: The anode and
cathode (area: 0.5 cm2)were attached to opposite sides of the
electrolyte (thickness: 1.0 mm; see Figure S5 in the Supporting
Information). Two gas chambers were set up by placing the cell
assembly between two alumina tubes. A mixture of hydrocarbon
(methane, ethane, propane, or butane)and water vapor was supplied
to the fuel chamber at a flow rate of 60 mLminꢁ1 at ambient pressure.
The pressure ratio of water to hydrocarbon was maintained at the
stoichiometric composition in Equation (1): 7 vol% methane and
14 vol% water vapor; 7 vol% ethane and 28 vol% water vapor;
7 vol% propane and 42 vol% water vapor; 7 vol% butane and
56 vol% water vapor. Unhumidified air was supplied to the air
chamber at a flow rate of 30 mLminꢁ1. The anodic polarization
properties were measured by the current interruption method using a
current pulse generator. For this measurement, an Au reference
electrode was attached on the side surface of the electrolyte and
exposed to ambient atmosphere. A constant current with a periodic
off-pulse signal of 100 ms was applied between the anode and cathode,
[14] B. Brummer, J. I. Ford, M. J. Turner, J. Phys. Chem. 1965, 69,
3424.
[15] H. A. Gasteiger, N. Markovic, P. N. Ross, E. J. Cairns, J. Phys.
´
Chem. 1993, 97, 12020.
[16] T. Iwasita, H. Hoster, A. John-Anacker, W. F. Lin and W.
Vielstich, Langmuir 2000, 16, 522.
[17] J. S. Spendelow, P. K. Babu, A. Wieckowski, Curr. Opin. Solid
State Mater. Sci. 2005, 9, 37.
[18] S. Natesakhawat, R. B. Watson, X. Wang, U. S. Ozkan, J. Catal.
2005, 234, 496.
[19] D. J. Sajkowski, S. T. Oyama, Appl. Catal. A 1996, 134, 339.
[20] D. C. LaMont, A. J. Gilligan, A. R. S. Darujati, A. S. Chellappa,
W. J. Thomson, Appl. Catal. A 2003, 255, 239.
7844 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 7841 –7844