Mixed phase Pt-Ru catalyst for direct methanol fuel cell anode by flame aerosol synthesis

Article abstract of DOI:10.1149/1.2109547

A spray-flame aerosol catalyzation technique was studied for producing Pt-Ru anode electrodes for the direct methanol fuel cell. Catalysts were produced as aerosol nanoparticles in a spray-flame reactor and deposited directly as a thin layer on the gas diffusion layer. The as-prepared catalyst was found to be a mixture of nanocrystalline, mostly unalloyed Ft and an amorphous phase mostly of Ru and to a lesser extent of Pt oxides on top of the crystalline phase. The flame-produced Pt1Ru1 demonstrated similar onset potential but ～60% higher activity compared to commercially available Pt1Ru1/Vulcan carbon. The kinetics of methanol oxidation on the mixed phase catalyst was also explored by electrochemical impedance spectroscopy.

Full text of DOI:10.1149/1.2109547

Journal of The Electrochemical Society, 152 ͑12͒ A2357-A2363 ͑2005͒
A2357
0013-4651/2005/152͑12͒/A2357/7/$7.00 © The Electrochemical Society, Inc.
Mixed Phase Pt-Ru Catalyst for Direct Methanol Fuel Cell
Anode by Flame Aerosol Synthesis
a
a,b
D. Chakraborty,a, H. Bischoff, I. Chorkendorff, and T. Johannessen
*
a,**,z
aInterdisciplinary Research Center for Catalysis (ICAT), Department of Chemical Engineering,
and bDepartment of Physics, Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark
A spray-flame aerosol catalyzation technique was studied for producing Pt-Ru anode electrodes for the direct methanol fuel cell.
Catalysts were produced as aerosol nanoparticles in a spray-flame reactor and deposited directly as a thin layer on the gas diffusion
layer. The as-prepared catalyst was found to be a mixture of nanocrystalline, mostly unalloyed Pt and an amorphous phase mostly
of Ru and to a lesser extent of Pt oxides on top of the crystalline phase. The flame-produced Pt1Ru1 demonstrated similar onset
potential but ϳ60% higher activity compared to commercially available Pt1Ru1/Vulcan carbon. The kinetics of methanol oxida-
tion on the mixed phase catalyst was also explored by electrochemical impedance spectroscopy.
© 2005 The Electrochemical Society. ͓DOI: 10.1149/1.2109547͔ All rights reserved.
Manuscript submitted January 20, 2005; revised manuscript received August 6, 2005. Available electronically October 26, 2005.
The direct methanol fuel cell ͑DMFC͒ uses methanol as fuel and
has significant advantages over hydrogen-fed proton exchange
membrane ͑PEM͒ fuel cells in mobile applications in terms of fuel
storage and supply.1 However, one of the main challenges of making
DMFCs commercially feasible is to improve the slow reaction ki-
netics of the methanol oxidation reaction at the anode. There is an
initial activation overpotential of some 0.2–0.55 V, depending on
the catalyst and operating conditions, required to oxidize methanol
anodically at an appreciable rate. At least some of this initiating
overpotential is kinetic in nature and could be reduced by suitable
catalysts at the anode.2 Another critical issue of DMFC is the thick-
ness of the anode catalyst layer. A thick catalyst layer increases the
ohmic resistance as well as mass-transfer resistance for methanol.
Therefore, to improve the DMFC anode performance it is necessary
to investigate new catalytic materials for methanol electro-oxidation
as well as alternative methods for catalyst preparation and mem-
brane electrode assembly ͑MEA͒ fabrication.
In acidic medium, only Pt or Pt-based catalysts have been found
to show acceptable activity and stability.3 Platinum dehydrogenates
methanol producing possibly adsorbed CO and other surface species
which potentially block the Pt sites. To regenerate the Pt sites for
further methanol oxidation at the DMFC operating range of interest
͑Ͻ0.5 V͒, Pt needs to be promoted by other metals. A promoter
could change the electronic structure of Pt so that surface-blocking
species are less strongly adsorbed, reducing the coverage and
thereby the blocking effect. This effect is known as the ligand
effect.4 A promoter could also activate water at lower potentials than
Pt to accelerate the oxidation and removal of surface-blocking spe-
cies. This is known as a bifunctional mechanism.5 Any enhancement
of performance due to the promoter is related to either one or both
of the two above-mentioned processes.6 Although several other ad-
ditives have been studied, and some ternary and quaternary alloys
have been shown to possess higher activity than Pt-Ru as DMFC
anode catalyst,7 Pt-Ru is still the most studied catalyst for DMFC
anodes.
not the way to improve catalyst performance for methanol electro-
oxidation.11,12 They observed that Pt0Ru0 has orders of magnitude
less electrocatalytically active for the reaction than the mixture of
metallic Pt and hydrous ruthenium oxides ͑RuOxHy͒. They have
also concluded that the ubiquitous way of confirming alloying from
the X-ray diffraction ͑XRD͒ pattern is not applicable for the high-
surface-area mixed-phase Pt-Ru methanol oxidation catalyst, and
so the mixtures of metallic Pt and amorphous RuOxHy have been
wrongly identified as Pt-Ru alloys. Also, it has been reported that
the record high performance of a DMFC at the Los Alamos National
Laboratory was obtained by using Pt-RuOx as the anode catalyst.13
The conventional methods of preparing fuel cell electrodes con-
sist of synthesis of catalysts, either supported or unsupported, and
applying the catalyst as a thin layer on the electrolyte membrane or
GDL, by painting or spraying.5,7-9,14,15 However, there have also
been a few attempts of directly loading the catalyst layer on the
membrane and/or on the gas diffusion layer ͑GDL͒ by sputter
deposition16-20 and electrochemically.21 Direct deposition could be
an important step for streamlining the MEA fabrication process de-
sirable for commercial production of fuel cells. Recently, we have
established a one-step method for preparation of the DMFC Pt-Ru
anode catalyst layer by using flame-spray pyrolysis. In this process,
after dissolving Pt and Ru precursors in appropriate solvents, the
solution is sprayed through a nozzle to produce micrometer-sized
droplets which burn out in a flame, resulting in metal atoms and/or
metal oxide molecules in the gas phase. Hereafter, clusters immedi-
ately nucleate and grow to nanosized particles by coagulation and/or
surface reaction and sintering.22 The mechanism of the particle for-
mation process in the gas phase ensures intimate mixing of the prod-
uct components.
As described in the Experimental section, the flame aerosol syn-
thesis method averts the tedious and cumbersome step of catalyst
ink painting in MEA preparation. Platinum and Pt-Ru with atomic
ratios of 3:1 and 1:1 ͑Pt:Ru͒ were prepared and deposited on the
GDL from the gas phase. In this article, the catalysts are designated
by writing the atomic ratio after the element, i.e., Pt3Ru1 means that
the nominal Pt to Ru atomic ratio of the deposit is 3:1. The deposits
have been characterized by XRD, X-ray photoelectron spectroscopy
͑XPS͒, scanning electron microscopy ͑SEM͒, X-ray energy-
dispersive spectroscopy ͑EDS͒, and transmission electron micros-
copy ͑TEM͒. After fabrication of the MEA, the catalysts have been
electrochemically tested for methanol electro-oxidation and the per-
formance has been compared with that of commercially available
E-TEK Pt1Ru1/Vulcan carbon anode. The methanol oxidation kinet-
ics has also been investigated by electrochemical impedance spec-
troscopy ͑EIS͒.
It is still a matter of debate whether Ru should be present in
the catalyst alloyed with Pt or as RuOx for the best catalytic per-
formance. Gottesfeld et al. have observed that Pt-Ru catalysts with
stronger metallic character exhibit higher electrocatalytic activity
for methanol oxidation than the more oxidic ones.8,9 However, they
concluded that the presence of RuOx is beneficial for protonic
conductivity within the catalyst layer.9 Sirk et al. also have observed
that the commercially available Johnson-Matthey Pt-Ru methanol
oxidation catalyst has higher activity when it is reduced in hydro-
gen compared to the activity shown by the as-received catalyst.10
Rolison et al. have hypothesized that the alloying of Pt with Ru is
Experimental
*
Electrochemical Society Student Member.
Electrochemical Society Active Member.
Catalyst preparation and deposition.— The metal precursors
used were ruthenium͑III͒ acetylacetonate ͓Ru͑acac͒3, Fluka, purity
**
z E-mail: tj@kt.dtu.dk
Downloaded on 2015-06-18 to IP 128.210.126.199 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A2358
Journal of The Electrochemical Society, 152 ͑12͒ A2357-A2363 ͑2005͒
Electrochemical experiments.— The flame-aerosol-produced an-
ode with a Pt1Ru1 catalyst layer and a total metal loading of
ϳ1.5 mg/cm2 was soaked with 0.1 mL of 5% Nafion in lower ali-
phatic alcohol solution ͑Aldrich Chemical Co.͒ and dried in open air
for 30 min and at 80°C for 1 h in an oven to remove the alcohol.
The anode from carbon-supported Pt–Ru ͑used for comparison͒ and
all the cathodes was prepared by the following procedure. Commer-
cially available PtRU/Vulcan carbon catalysts having 1:1 atomic ra-
tio and 10 wt % total metal loading was obtained from E-TEK, Inc.
Anode catalysts were applied to the MEAs at a total metal loading
of ϳ1.5 mg/cm2. The total metal loading of the cathode catalyst
was ϳ2 mg/cm2 ͑20% Pt/Vulcan carbon, E-TEK͒. The catalyst
samples were ultrasonically mixed for 20 min in a clean bottle with
0.1 mL of 5% Nafion in lower aliphatic alcohol solution and deion-
ized water. Isopropyl alcohol in water was added as a diluent to
form a well-mixed catalyst ink. The catalyst inks were then brushed
onto the GDL. Once all catalysts had been deposited, the electrodes
were dried for 30 min in air at room temperature and for 1 h in an
oven at 80°C. Similar cathodes were used with both flame-prepared
anode and conventionally made anode with E-TEK catalyst. To pre-
pare the MEAs, Nafion 112 polymer electrolyte membranes had
been sandwiched between anode and cathode electrodes and pressed
at ϳ140°C and 200 bars for 5 min. All the MEAs prepared had
circular electrodes with a diameter of 2 cm.
The MEAs were placed between two graphite bipolar plates ͑Si-
gracet͒ which had machined flow channels with serpentine flow pat-
tern, capable of accommodating electrodes of maximum 3.14 cm2 in
a homemade single cell ͑designed by IRD Fuel Cells A/S͒. The sides
of the stainless steel current collectors, which were not contacting
the bipolar plates, had serpentine channels. The channels were sepa-
rated from the water-filled chambers by Nafion membranes. Gas
streams were humidified by passing the gases through the channels.
When methanol was fed at the anode, the water chamber at the
anode side was emptied. For all the experiments reported in the
article where methanol was fed to the anode, the concentration of
the solution was 1 mol/L and the flow rate was 1 mL/min. For cell
polarization measurements, the cathode was fed with humidified O2
at a cylinder back pressure of 2 bars and a flow rate of 80 sccm. For
anode polarization and EIS, the cathode was fed with humidified
H2 at a rate of 5 sccm and functioned as a dynamic hydrogen elec-
trode ͑DHE͒.8,28-30 The cathode was also the counter electrode. The
anode was fed with methanol and functioned as working electrode.
The scans for polarization measurements were performed from 0 to
0.8 V vs DHE at 1 mV/s. For EIS, the frequency range was from
1 kHz to 50 mHz and the measurements were done at different dc
potentials vs DHE. The amplitude of the sinusoidal voltage signal
used was 10 mV. The conditioning of the MEAs was started by
flowing overnight humidified H2 and humidified Ar at the anode and
cathode, respectively, at room temperature. Then the cell was run as
a H2/O2 fuel cell at 90°C for 1 h with humidified H2 at the anode.8
After that H2 at the anode was replaced with 1.0 M methanol and
the cell voltage was cycled continuously for 1 h between open cir-
cuit to 0.8 V at a scan rate of 50 mV/s to finish the conditioning.
Before taking any measurement, the cell was cycled between 0 and
0.8 V at 50 mV/s until reproducibility was established. All the elec-
trochemical measurements were performed with the potentiostat
PARSTAT 2273.
ജ97%͔ and platinum͑II͒ acetylacetonate ͓Pt͑acac͒2, Aldrich Chemi-
cal Co., 97% purity͔. The solvent was a 4:1 volume ratio mixture of
isooctane ͑Fluka, purity 99%͒ and tetrahydrofuran ͑Sigma, purity
99%͒. In all the stock solutions, the concentration of Pt͑acac͒ was
2
kept constant while the concentration of Ru͑acac͒ was varied ac-
3
cording to the required catalyst composition. The following solu-
tions were prepared: 40 mg of Pt͑acac͒ in 100 mL of the solvent
2
for Pt, 40 mg of Pt͑acac͒ and Ru͑acac͒ each with 100 mL of the
2
3
solvent for Pt1Ru1, and 40 mg of Pt͑acac͒ and 13.3 mg of
2
Ru͑acac͒3 with 100 mL of the solvent for Pt3Ru1. The solution was
pumped by using a syringe pump ͑IVAC P6000͒ through a gas-
assisted nozzle to spray the precursor solution to the flame zone as
small droplets. The nozzle consisted of a capillary tube o.d. 0.9 mm
͑i.d. 0.6 mm͒, which is situated in an opening of 1.4-mm diam,
creating an annular space of 0.9 mm2. The design of the nozzle is
similar to the one used by Mädler et al.23 The precursor solution was
passed through the capillary tube at 30 mL/h and oxygen, as both
the dispersion and oxidation gas, was introduced through the annu-
lus at 2.2 L/min. The spray aerosol was ignited by eight hydrogen-
fed supporting flames made by horizontal injection through eight
equidistant holes drilled in a hollow metal ring placed around the
aerosol jet. Cold air, provided from a water-cooled quenching ring,
was used to quench the aerosol coming out of the flame.24 The
quenching helps decrease the average particle size by suddenly low-
ering the temperature and thus decreasing the rate of sintering. The
product particles were collected directly on the GDL ͑Toray
TGPH090͒ by withdrawing the nanoparticle aerosol from the flame
through it by using a gas-ejector vacuum pump ͑PIAB Classic͒.
Here, the GDL was, in principle, applied as a catalyst nanoparticle
filter.
Physical and chemical characterization.— The XRD patterns
were obtained from the catalyzed GDLs with a Philips diffracto-
meter with a Cu K␣ radiation source and a nickel filter. The scans
were performed over a 2␪ range of 20–95° with a scan step of 0.05°
and acquisition time of 4 s. The crystalline sizes and peak angles
were determined using Xfit25 with a fundamental parameter fitting
approach and using the entire XRD pattern ͑20–95°͒. An external Si
standard was used to correct the sample displacement errors.15 The
Si powder ͑Aldrich͒ was sprinkled carefully on the catalyzed layer
before taking the XRD pattern. The lattice parameters for the nano-
crystalline phase were calculated from the line position of Pt͑111͒
by using the following formula
= ͱ3␭/͑2 sin ␪͒
afcc
͓1͔
where ␭ = 1.54056 Å, the wavelength of radiation used, and ␪ is the
Bragg angle for the measured peak.
The SEM and EDS experiments were done by LEO SUPRA50.
For EDS, 32 different nonoverlapping areas with a diameter of
1 ␮m on the Pt1Ru1 catalyst layer were investigated with an
electron-beam of 20 keV accelerating voltage for elemental infor-
mation and the average was taken as the catalyst layer composition.
With this accelerating voltage, the electron could penetrate to a
depth on the order of 1.0 ␮m.26 The TEM images were obtained
with JEOL 3000F. The sample was prepared by scraping off the
deposits from the GDL and suspending in ethanol. This suspension
was added as a droplet on the holey carbon grid.
XPS of the catalyzed face of GDLs for various samples were
recorded on a Phi 550 spectrometer by employing a Mg K␣ source
operating at 200 W. All the XPS spectra were obtained with pass
energy of 50 eV for wide scans and 25 eV for individual elements.
The samples were introduced into the spectrometer using a separate
fast-entry chamber. The ratio between Pt and Ru atoms at the sur-
face layer, ͓Pt͔surf/͓Ru͔surf, were estimated from the following
equation27
Results and Discussion
The SEM image shows the morphology of the deposits on the
GDL ͑Fig. 1͒. The deposit seems to be porous and fluffy. This is
possibly the result of the primary particles sticking together rather
than sintering to form bigger single particles. The porous nature of
the deposits is thought to be advantageous for transport of materials
in and out of the agglomerates. The TEM image shows the highly
crystalline metallic particles on the carbon grid ͑Fig. 2͒. The lattice
spacing of 2.23 and 2 Å representing ͑111͒ and ͑002͒ planes, respec-
tively, of the face-centered cubic ͑fcc͒ crystal of Pt are observed. No
crystalline phase of either Ru or RuO2 could be detected from the
͓Pt͔surf/͓Ru͔ = ͑IPt/SPt͒/͑IRu/SRu
͒
͓2͔
surf
where Si is the atomic sensitivity factor and Ii is the area under the
core-level peak considered.
Downloaded on 2015-06-18 to IP 128.210.126.199 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 ͑12͒ A2357-A2363 ͑2005͒
A2359
Figure 1. SEM image of the flame-produced Pt-Ru deposited on GDL.
d-spacing of the crystallites. Figure 3 shows the XRD patterns of the
flame-synthesized Pt, Pt3Ru1, and Pt1Ru1 catalyzed layers on GDL.
For all the patterns shown, the five characteristic peaks of the fcc
crystalline structure of Pt, namely, the ͑111͒, ͑200͒, ͑220͒, ͑311͒, and
͑222͒ could be seen. However, no significant peak shift was ob-
served with increasing Ru content in the sample. The lattice param-
eters ͑afcc͒ show an insignificant gradual decrease with increasing
Ru content compared to the arc-melted Pt-Ru alloy reported by
Gasteiger et al.31 ͑Fig. 4͒. For the arc-melted alloys, the lattice pa-
rameter decreases linearly with increasing Ru content, consistent
with the smaller size of Ru atoms compared to Pt. Also shown in the
figure are the reported afcc values for the Johnson-Matthey unsup-
ported Pt-Ru catalyst15 and the high-surface-area unsupported Pt-Ru
catalyst prepared by rapid borohydride reduction of aqueous metal
salts.7 The afcc values for high-surface-area Pt1Ru1 is higher than
that of arc-melted Pt-Ru. However, the decrease in afcc from pure Pt
to Pt1Ru1 for flame-synthesized catalyst is almost an order of mag-
nitude less than that obtained for commercial Johnson-Matthey
Pt1Ru1,15 implying that in our catalyst the nanocrystalline phase
consists mostly of Pt. The absence of any other crystalline phase
leads us to envisage the flame-prepared catalysts as a mixture of a
crystalline phase consisting predominately of Pt and one or more
amorphous phases consisting of mainly Ru as metallic or oxidic
states. The particle size of the crystalline phase of Pt1Ru1, as deter-
Figure 3. XRD patterns for catalyzed layers of Pt, Pt3Ru1, and Pt1Ru1 on
GDL. Si powder sprinkled on the catalyst layers served as an external stan-
dard for correcting sample displacement errors. The spectra were collected
between the 2␪ values of 20 and 95°; however, a zoomed-in window of
25–95° is shown.
mined from XRD peak broadening, is 10.3 0.3 nm. The peak at
ϳ54° is a carbon peak coming from the TGPH 090 support and it
has been confirmed by performing XRD on a blank TGPH 090
͑figure not shown͒.
Overall X-ray photoemission spectrum of the TGPH 090 carbon
support and different catalysts are shown in Fig. 5. Peaks of Pt 4f, Pt
Figure 2. TEM image of the flame-produced nanocomposites on holey car-
bon grid. The insert shows the crystalline plane of a nanoparticle. Note that
the scale bar is for the main figure and not the insert.
Figure 4. Comparison of lattice parameters of the fcc phase of flame-
synthesized catalysts with that of the literature reported high-surface-area
catalysts ͑Ref. 7 and Ref. 15͒ and arc-melted Pt-Ru alloys ͑Ref. 31͒.
Downloaded on 2015-06-18 to IP 128.210.126.199 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A2360
Journal of The Electrochemical Society, 152 ͑12͒ A2357-A2363 ͑2005͒
Figure 5. Overall XPS spectra of the TGPH 090 GDL, Pt, and Pt1Ru1
catalyst layer on GDL.
Figure 7. The XPS spectra for Ru 3p3/2 core region of Pt1Ru1 catalyst layer
on GDL. The vertical lines are the values reported in literature for the bind-
ing energies of different oxidation states of Ru.
4d, C 1s, and O1s are observed for the Pt catalyst layer. All the
peaks present in the Pt catalyst layer could also be seen for Pt1Ru1.
The peaks of Ru 3d overlapped with C 1s, and Ru 3p peaks are also
observed. In Fig. 6, the Pt 4f core-level spectrum of the Pt and
Pt1Ru1 catalyst layers on GDL are shown. The vertical lines at 71.0,
72.4, and 73.8 eV are the binding energies reported for the 4f7/2
peak of Pt, Pt͑OH͒2, and PtO, respectively.32 Because of the lack of
information about the exact nature of the compounds formed, no
deconvolution and curve-fitting has been done. However, from Fig.
6, it may be deduced that oxidized Pt is present in relatively large
amount in the Pt1Ru1 sample compared to the Pt sample. A signifi-
cant enhancement of the amount of Pt-O species with the presence
of Ru has been reported in the literature for Pt-Ru DMFC anode
catalyst. The observation has been attributed to the increase of the
coverage of Pt-Ru surfaces by oxy-species due to Ru.3,33-35 How-
ever, no crystalline Pt-oxide phase in our catalyst is detected from
XRD, possibly indicating that Pt oxides are present mostly at the
surface in an amorphous phase. Figure 7 shows the Ru 3p3/2 peak
for the Pt1R1 sample. The vertical lines at 462 and 466 eV are the
binding energies reported for metallic Ru and rutile RuO2.36 It could
be deduced that the bulk of the Ru exists at oxidation states between
metallic Ru and Ru+4. The surface concentration of Pt and Ru esti-
mated from the intensities of Pt 4f and Ru 3p3/2 ͑atomic sensitivity
factors of 4.4 and 1.30, respectively37͒ peaks by using Eq. 2 gives a
Pt atomic fraction of 0.48. However, the average of the Pt atomic
fraction determined from EDS is 0.58 0.03. Considering the fact
that XPS signals mainly come from the surface and EDS signals
primarily come from the bulk, the difference between the Pt atomic
fraction determined from XPS and EDS could be rationalized by
postulating a thin amorphous surface layer consisting of Pt-Ru ox-
ides on top of the nanocrystalline Pt fcc phase.
Figure 8 shows the Tafel slopes ͑b͒ for methanol oxidation by
Pt1Ru1 at different temperatures. The overpotential data is corrected
for iR losses while the ohmic resistances at different temperatures
have been determined from ac impedance experiments. The Tafel
slope decreases from 135 mV/d at 25°C to 95 mV/d at 90°C. The
decrease in Tafel slope with increase of temperature has been re-
ported in literature for PtRu/C when the catalyst was incorporated in
the MEA with Nafion as electrolyte.38 Each of the Nyquist plots at
80°C from the impedance spectroscopy of methanol oxidation at
different dc potentials shows a capacitive loop in the first quadrant
and another pseudoinductive loop in the forth quadrant ͑Fig. 9͒. The
capacitive loop arises from the charge-transfer resistance of metha-
nol oxidation, and the x-axis intercept of the loop represents the
magnitude of charge-transfer resistance,29,39-41 whereas it has been
postulated that the presence of the pseudoinductive loop is an indi-
cation that the oxidative removal of COads is the rate-determining
step for methanol oxidation on Pt-Ru at DMFC anodes.30,42 The
improvement in the methanol oxidation kinetics with increasing po-
tential is probably because at higher potentials, Pt sites, beside Ru,
supply active oxygen-containing species for COads oxidation.43
Methanol electro-oxidation on Pt-Ru is an activated process and
Figure 6. The XPS spectra for the Pt 4f core region of Pt and Pt1Ru1
catalyst layers on GDL. The vertical lines are the literature reported values
for the binding energies of different oxidation states of Pt.
Downloaded on 2015-06-18 to IP 128.210.126.199 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 ͑12͒ A2357-A2363 ͑2005͒
A2361
Figure 8. Overpotential vs log͑i͒ at different temperatures for methanol
oxidation: 1 mL/min, 1.0 M CH3OH at the anode, and 5 mL/min humidified
H2 at the cathode. Anode: ϳ1.5 mg/cm2 Pt1Ru1. Cathode: ϳ2 mg/cm2 20%
Pt/Vulcan carbon ͑E-TEK͒.
Figure 10. Methanol oxidation overpotential vs current density at different
temperatures. Anode: ϳ1.5 mg/cm2 Pt1Ru1. Cathode: ϳ2 mg/cm2 20% Pt/
Vulcan carbon ͑E-TEK͒.
forth quadrant loops when the temperature is increased from
50 to 80°C at 0.5 V ͑Fig. 11͒. This could be an indication of lower
charge-transfer resistance of methanol oxidation caused by the im-
proved COads oxidation kinetics. An interesting feature in the plot is
the temperature dependence of the ohmic resistance, which is given
by the interception of the high-frequency curve with the x axis. The
ohmic resistance decreases from 1.34 ⍀ cm2 at 50°C to 0.96 ⍀ cm2
and 0.89 ⍀ cm2 at 80 and 90°C, respectively. Arrhenius plots of
current density ͑i͒ obtained from potentials below 0.5 V are given in
depends heavily on temperature.3 Figure 10 shows the dependence
of methanol oxidation on temperature for flame-synthesized Pt-Ru.
The onset potential varies from ϳ400 mV at 25°C to ϳ250 mV at
90°C. The decrease in the initial activation overpotential with tem-
perature may be because of enhanced hydroxyl species formation on
Pt at higher temperature,43 providing more OHads-like species for
the removal of COads or other surface-blocking species. The Nyquist
plots at 0.5 V vs DHE show a huge shrinkage of both the first and
Figure 9. Nyquist plots showing the effect of anode potential vs DHE on the
methanol oxidation kinetics at the DMFC anode as determined by impedance
spectroscopy. Anode: ϳ1.5 mg/cm2 Pt1Ru1. Cathode: ϳ2 mg/cm2 20% Pt/
Vulcan carbon ͑E-TEK͒.
Figure 11. Nyquist plots showing the effect of temperature on methanol
oxidation kinetics as determined by impedance spectroscopy at a dc voltage
of 0.5 V vs DHE. Anode: ϳ1.5 mg/cm2 Pt1Ru1. Cathode: ϳ2 mg/cm2 20%
Pt/Vulcan carbon ͑E-TEK͒.
Downloaded on 2015-06-18 to IP 128.210.126.199 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

A2362
Journal of The Electrochemical Society, 152 ͑12͒ A2357-A2363 ͑2005͒
Figure 12. Arrhenius plot at different overpotentials. Anode: ϳ1.5 mg/cm2
Pt1Ru1. Cathode: ϳ2 mg/cm2 20% Pt/Vulcan carbon ͑E-TEK͒.
Fig. 12. The activation energy is 49 kJ/mol at 0.3 V, 53 kJ/mol at
0.4, and 37.6 kJ/mol at 0.5 V. The value at 0.5 V is slightly higher
than the 30 kJ/mol reported for PtRu/C obtained at 0.5 V in a genu-
ine fuel cell environment as in this article.44 However, the average
of the three activation energies on our catalyst ͑47 kJ/mol͒ is
smaller than the average value ͑58 kJ/mol͒ below 0.5 V reported on
45
Ru decorated Pt͑111͒ and polycrystalline Pt-Ru ͑60 kJ/mol͒.3,43
A
comparison of performance between flame-synthesized
Pt1Ru1 and 1:1 PtRu/C ͑10% metal loading͒ is shown in Fig. 13.
The supported Pt-Ru catalyst has been chosen for comparison be-
cause we wanted to observe the difference between a well-alloyed
metallic Pt-Ru catalyst and the flame-synthesized catalyst. The un-
supported Pt-Ru has been reported to have lesser degree of alloying
and a significant presence of RuO2 compared to carbon-supported
Pt-Ru from E-TEK.46 The potentials in Fig. 13a have been corrected
for iR losses. Even though the onset potentials are similar, the MEA
prepared from the flame-synthesized catalyst performs better than
the E-TEK catalyst with similar catalyst loading for all higher cur-
rent densities. At 400 mV, the current density for Pt1Ru1 is ϳ1.6
times higher than the current density obtained from the MEA with
anode made from E-TEK catalyst. For the sake of completeness, we
also present the comparison of performance when the fuel cell is
producing power, i.e., when O2 is fed at the cathode and CH3OH at
the anode ͑Fig. 13b͒. Again, the performance is better for the MEA
that has flame-synthesized Pt1Ru1 anode. It is worthwhile to men-
tion that in this mode of operation, the data obtained could be af-
fected by oxygen reduction kinetics at the cathode and methanol
crossover from anode to the cathode, which could be different for
two different catalyst layers at the anode.
One has to be careful during comparison of the performance data
for catalysts having different physical and chemical properties and
prepared by different methods. Besides the difference in the catalytic
properties of the active elements, other factors could also affect the
performance. The E-TEK Pt1Ru1/C catalyst has a particle size of
2–3 nm, whereas the flame-synthesized Pt1Ru1catalyst has an aver-
age particle size of ϳ10 nm. Because of smaller particle size, the
E-TEK catalyst has ϳ10 times higher specific metal surface area
than flame-prepared Pt1Ru1. However, the electrochemically active
surface could well be smaller for the anode prepared from the
E-TEK catalyst, resulting in the lower current density. Moreover,
Figure 13. DMFC anode ͑a͒ and cell ͑b͒ current density–voltage curves,
comparing performance of flame-synthesized Pt1Ru1 to that of commercially
available E-TEK Pt1Ru1/C as anode catalyst at 90°C. Anode loading:
ϳ1.5 mg/cm2. Cathode: ϳ2 mg/cm2 20% Pt/Vulcan carbon ͑E-TEK͒.
the anode catalyst layer prepared from carbon-supported catalyst
could have a higher thickness than the unsupported flame-prepared
anode, and as a consequence, mass-transfer resistance could be
higher in the carbon-supported anode. However, at a low current
density, i.e., just above the onset potential, mass-transfer resistance
should not have a significant contribution. At an overpotential of
280 mV, the current density obtained from the MEA with anode
prepared by E-TEK catalyst is 9 mA/cm2, whereas that for MEA
with anode fabricated from our method is 14 mA/cm2, i.e., ϳ1.6
times higher.
XPS data suggest that a significant fraction of Pt and most of the
Ru is present as oxides in as-prepared Pt1Ru1 catalyst. However,
during conditioning, when the catalyst was exposed to H2 and sub-
sequently to CH3OH at 90°C, the metal oxides could be reduced
significantly to metallic state.8 This left us with several possible
Downloaded on 2015-06-18 to IP 128.210.126.199 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 ͑12͒ A2357-A2363 ͑2005͒
A2363
12. D. R. Rolison, P. L. Hagans, K. E. Swider, and J. W. Long, Langmuir, 15, 774
͑1999͒.
speculations about the state of the catalyst after conditioning. Even
though there is no bulk alloying, the metals could be partially al-
loyed at the surface layer. The so-called surface decoration45,47 of
Ru on Pt particles could also be possible. However, with the current
results and considering the complexity of the material, we are not in
a position to make any definite conclusion about the physical and
chemical state of the catalyst at the operating conditions.
13. S. J. C. Cleghorn, X. Ren, T. E. Springer, M. S. Wilson, C. Zawodzinski, T. A.
Zawodzinski, and S. Gottesfeld, Int. J. Hydrogen Energy, 22, 1137 ͑1997͒.
14. W. D. King, J. D. Corn, O. J. Murphy, D. L. Boxall, E. A. Kenik, K. C. Kwiat-
kowski, S. R. Stock, and C. M. Lukehart, J. Phys. Chem. B, 107, 5467 ͑2003͒.
15. R. R. Diaz-Morales, R. Liu, E. Fachini, G. Chen, C. U. Segre, A. Martinez, C.
Cabrera, and E. S. Smotkin, J. Electrochem. Soc., 151, A1314 ͑2004͒.
16. S. Hirano, J. Kin, and S. Srinivasan, Electrochim. Acta, 42, 1587 ͑1997͒.
17. S. Y. Cha and W. M. Lee, J. Electrochem. Soc., 146, 4055 ͑1999͒.
18. R. O’Hayre, S.-J. Lee, S.-W. Cha, and F. B. Prinz, J. Power Sources, 109, 483
͑2002͒.
Conclusion
We have successfully demonstrated a one-step catalyzation
method for preparation of the DMFC anodes. The as-prepared nano-
composites are a mixture of fcc, mostly unalloyed Pt crystalline
phase, and an amorphous surface layer, postulated to be a mixture
of Pt and Ru oxides. Comparison with commercially available
10%Pt–Ru/C E-TEK catalyst demonstrated that even though the
onset potential for methanol oxidation at 90°C on both catalysts are
similar ͑ϳ250 mV͒, the flame-prepared Pt1Ru1 has 60% higher ac-
tivity at 0.4 V. However, the exact nature of the physical and chemi-
cal state of the flame-synthesized mixed-phase nanocomposite dur-
ing the methanol oxidation in a fuel cell is yet to be established and
is a subject of our ongoing research.
19. C. K. Witham, W. Chun, T. I. Valdez, and S. R. Narayanan, Electrochem. Solid-
State Lett., 3, 497 ͑2000͒.
20. Y. K. Xiu and N. Nakagawa, J. Electrochem. Soc., 151, A1483 ͑2004͒.
21. E. J. Taylor, E. B. Anderson, and N. R. K. Vilambi, J. Electrochem. Soc., 139, L45
͑1992͒.
22. S. E. Pratsinis, Prog. Energy Combust. Sci., 24, 197 ͑1998͒.
23. L. Mädler, H. K. Kammler, R. Mueller, and S. E. Pratsinis, J. Aerosol Sci., 33, 369
͑2002͒.
24. J. P. Hansen, J. R. Jensen, H. Livbjerg, and T. Johannessen, AIChE J., 47, 2413
͑2001͒.
25. R. W. Cheary and A. A. Coelho, Engineering and Science Research Laboratory,
Warrington, England ͑http://www.ccp14.ac.uk/tutorial/xfit-95/xfit.htm͒ ͑1996͒.
26. C. W. Oatley, Scanning Electron Microscopy, p. 187, Cambridge University Press,
Cambridge, U.K. ͑1972͒.
27. C. D. Wagner, W. M. Riggs, L. E. Davis, and J. F. Moulder, Handbook of X-Ray
Photoelectron Spectroscopy, G. E. Muilenberg, Editor, p. 22, Perkin-Elmer Corpo-
ration, Eden Prairie, MN ͑1978͒.
Acknowledgments
This work was supported by PSO ͑project 5357͒. The authors
wish to thank IRD Fuel Cells A/S for assistance with the project
during startup regarding materials and the design of the fuel cell test
unit. The scientific and technical support for the EDS analysis of the
DG JRC Institute for Energy, Microstructural Analysis Centre, is
gratefully acknowledged.
28. C. He, Z. Qi, M. Hollett, and A. Kaufman, Electrochem. Solid-State Lett., 5, A181
͑2002͒.
29. J. T. Mueller and P. M. Urban, J. Power Sources, 75, 139 ͑1998͒.
30. J. T. Müller, P. M. Urban, and W. F. Holderich, J. Power Sources, 84, 157 ͑1999͒.
31. H. A. Gasteiger, P. N. Ross, Jr., and E. J. Cairns, Surf. Sci., 293, 67 ͑1993͒.
32. M. C. Denis, G. Lalande, D. Guay, J. P. Dodelet, and R. Schulz, J. Appl. Electro-
chem., 29, 951 ͑1999͒.
33. A. S. Aricó, P. Cretì, H. Kim, R. Mantegna, N. Giordano, and V. Antonucci,
Technical University of Denmark assisted in meeting the publication
costs of this article.
J. Electrochem. Soc., 143, 3950 ͑1996͒.
34. B. J. Kennedy and A. Hamnett, J. Electroanal. Chem. Interfacial Electrochem.,
283, 271 ͑1990͒.
References
35. J. B. Goodenough, A. Hamnett, B. J. Kennedy, and S. A. Weeks, Electrochim.
Acta, 32, 1233 ͑1987͒.
1. B. D. McNicol, D. A. J. Rand, and K. R. Williams, J. Power Sources, 83, 15
͑1999͒.
36. C. Bock, M. A. Blakely, and B. MacDougall, Electrochim. Acta, 50, 2401 ͑2005͒.
37. Practical Surface Analysis by Auger and X-Ray Photoelectron Spectroscopy, D.
Briggs and M. P. Seah, Editors, p. 513, John Wiley & Sons, New York ͑1983͒.
38. G. Wu, L. Li, and B. Q. Xu, Electrochim. Acta, 50, 1 ͑2004͒.
39. M. Ciureanu, S. D. Mikhailenko, and S. Kaliaguine, Catal. Today, 82, 195 ͑2003͒.
40. I.-M. Hsing, X. Wang, and Y.-J. Leng, J. Electrochem. Soc., 149, A615 ͑2002͒.
41. W. Sugimoto, K. Aoyama, T. Kawaguchi, Y. Murakami, and Y. Takasu, J. Elec-
troanal. Chem., 576, 215 ͑2005͒.
2. G. T. Burstein, C. J. Barnett, A. R. Kucernak, and K. R. Williams, Catal. Today,
38, 425 ͑1997͒.
3. A. Hamnett, Catal. Today, 38, 445 ͑1997͒.
4. T. Frelink, W. Visscher, and J. A. R. van Veen, Surf. Sci., 335, 353 ͑1995͒.
5. M. Watanabe and S. Motoo, J. Electroanal. Chem. Interfacial Electrochem., 60,
267 ͑1975͒.
6. T. Kawaguchi, W. Sugimoto, Y. Murakami, and Y. Takasu, Electrochem. Commun.,
6, 480 ͑2004͒.
42. W. Sugimoto, K. Aoyama, T. Kawaguchi, Y. Murakami, and Y. Takasu, J. Elec-
troanal. Chem., 576, 215 ͑2004͒.
43. H. A. Gasteiger, N. Markovic, P. N. Ross, Jr., and E. J. Cairns, J. Electrochem.
Soc., 141, 1795 ͑1994͒.
7. B. Gurau, R. Viswanathan, R. Liu, T. J. Lafrenz, K. L. Ley, E. S. Smotkin, E.
Reddington, A. Sapienza, B. C. Chan, T. E. Mallouk, and S. Sarangapani, J. Phys.
Chem., 102, 9997 ͑1998͒.
44. B. Gurau and E. S. Smotkin, J. Power Sources, 112, 339 ͑2002͒.
45. G. Tremiliosi-Filho, H. Kim, W. Chrzanowski, A. Wieckowski, B. Grzybowska,
and P. Kulesza, J. Electroanal. Chem., 467, 143 ͑1999͒.
8. H. N. Dinh, X. Ren, F. H. Garzon, P. Zelenay, and S. Gottesfeld, J. Electroanal.
Chem., 491, 222 ͑2000͒.
9. S. C. Thomas, X. Ren, and S. Gottesfeld, J. Electrochem. Soc., 146, 4354 ͑1999͒.
10. A. H. C. Sirk, J. M. Hill, S. K. Y. Kung, and V. I. Birss, J. Phys. Chem. B, 108, 689
͑2004͒.
46. A. S. Aricó, A. K. Shukla, K. M. El Khatib, P. Cretì, and V. Antonucci, J. Appl.
Electrochem., 29, 671 ͑1999͒.
11. J. W. Long, R. M. Stroud, K. E. Swider-Lyons, and D. R. Rolison, J. Phys. Chem.
B, 104, 9772 ͑2000͒.
47. P. Waszczuk, J. Solla-Gullon, H.-S. Kim, Y. Y. Tong, V. Montiel, A. Aldaz, and A.
Wieckowski, J. Catal., 203, 1 ͑2001͒.
Downloaded on 2015-06-18 to IP 128.210.126.199 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).