Preparation and electrochemical activities of Pt-Ti alloy PEFC electrocatalysts

Article abstract of DOI:10.1149/1.2756369

Titanium is one of the most stable materials, remaining undissolved under strongly acidic polymer electrolyte fuel cell (PEFC) conditions. Nanocrystalline Pt-Ti alloy electrocatalysts with a catalyst grain size of several nanometers have been successfully prepared by the colloidal and sol-gel method, characterized by X-ray diffraction, field-emission scanning electron microscope, scanning transmission electron microscope, energy-dispersive X-ray analysis, and X-ray photoelectron spectroscopy. Electrochemical activities were evaluated by current-voltage characteristics measurements using membrane electrode assemblies for humidified H2 fuels with or without 100 ppm CO, cyclic voltammetry, and CO -stripping voltammetry in 0.1 M HClO4. Pt-Ti alloy electrocatalysts exhibited higher electrochemical performance than pure Pt catalyst, i.e., higher activity for oxygen reduction reactions and improved CO tolerance.

Full text of DOI:10.1149/1.2756369

Journal of The Electrochemical Society, 154 ͑9͒ B969-B975 ͑2007͒
B969
0
013-4651/2007/154͑9͒/B969/7/$20.00 © The Electrochemical Society
Preparation and Electrochemical Activities of Pt–Ti Alloy
PEFC Electrocatalysts
a
a
a
a
b
a
Y. Kawasoe, S. Tanaka, T. Kuroki, H. Kusaba, K. Ito, Y. Teraoka, and
K. Sasaki
a,b,c, ,z
*
a
b
Interdisciplinary Graduate School of Engineering Sciences, Faculty of Engineering,
c
and Hydrogen Technology Research Center, Kyushu University, Fukuoka-shi, 819-0395 Fukuoka, Japan
Titanium is one of the most stable materials, remaining undissolved under strongly acidic polymer electrolyte fuel cell ͑PEFC͒
conditions. Nanocrystalline Pt–Ti alloy electrocatalysts with a catalyst grain size of several nanometers have been successfully
prepared by the colloidal and sol-gel method, characterized by X-ray diffraction, field-emission scanning electron microscope,
scanning transmission electron microscope, energy-dispersive X-ray analysis, and X-ray photoelectron spectroscopy. Electro-
chemical activities were evaluated by current-voltage characteristics measurements using membrane electrode assemblies for
humidified H2 fuels with or without 100 ppm CO, cyclic voltammetry, and CO-stripping voltammetry in 0.1 M HClO . Pt–Ti
4
alloy electrocatalysts exhibited higher electrochemical performance than pure Pt catalyst, i.e., higher activity for oxygen reduction
reactions and improved CO tolerance.
©
2007 The Electrochemical Society. ͓DOI: 10.1149/1.2756369͔ All rights reserved.
Manuscript submitted January 24, 2007; revised manuscript received May 18, 2007. Available electronically July 24, 2007.
1
3
Polymer electrolyte fuel cells ͑PEFCs͒ are considered to be
promising future power sources for electric vehicles, stationary co-
generation systems, as well as portable instruments because of their
environmental compatibility and high energy efficiency. However,
due to their low operational temperature and the strongly acidic
character of their electrolyte materials, noble metal, especially plati-
num, is used for electrocatalysts. These catalysts are usually dis-
persed as nanosized particles on conductive support like carbon
black to obtain a large surface area.
Pt, widely used as electrocatalysts, exhibits one of the highest
catalytic activities for the oxygen reduction reaction ͑ORR͒ among
pure metals. However, the ORR is relatively slow and the activation
overvoltage at the cathodes is also large, even with Pt. For higher
efficiency of PEFCs, development of electrocatalysts having ORR
activity much higher than pure Pt is desired.
showed high CO tolerance at room temperature in 0.1 M HClO4.
Mukerjee et al. revealed that Pt–Mo alloy showed even better CO
tolerance than Pt–Ru alloy.
1
4
While various Pt alloy catalysts have been extensively studied,
most transition metals in alloys are thermochemically stable as ions
1
5
and thus can dissolve in acid solution from the alloys. Because
catalyst particles on support materials are a few nanometers in di-
ameter, the dissolution of transition metals from the catalyst nano-
particles causes serious problems such as catalyst degradation and
electrolyte poisoning. Some Pt–Pt group metal ͑PGM͒ alloys are
expected to show superior durability in comparison with Pt-
transition metal alloys. For example, Pt–Ir alloy showed higher
ORR activity than pure Pt and it was examined as a cathode catalyst
1
6
for PEFCs. However, the use of Pt–PGM alloys is not desired due
to their limited availability.
Therefore, in order to develop alternative electrocatalysts using
widely available inexpensive metals stable under strongly acidic
PEFC conditions, we prepared Pt–Ti alloy electrocatalysts. Figure 1
shows a pH-potential equilibrium diagram, a so-called Pourbaix
Pure hydrogen is usually used as a PEFC fuel gas. However
hydrogen-containing gas generated by steam reforming or partial
oxidation of hydrocarbons or alcohols is often used as a practical
fuel, because the infrastructure for the production and storage of
pure hydrogen is not yet widely available. When reformed fuel gas
containing a small amount of CO is used, the anodic overvoltage
increases and the cell voltage becomes lower due to CO poisoning
1
7
diagram, of the Ti–H O system described against the standard
2
hydrogen electrode. The open circle indicates the cathodic condition
and the dotted circle the anodic condition in PEFCs. This figure
suggests that Ti is the stable material as an oxide in the practical
operation environment of PEFCs. The aim of this study is therefore
1
of Pt electrocatalysts. Pt–Ru alloy is known as a catalyst with a
2,3
high CO tolerance and is widely used as an electrocatalyst for
PEFC anodes. However, Ru is also a noble metal like Pt and the
natural resources are limited, so that alternative CO-tolerant electro-
catalysts instead of Pt–Ru alloy are also needed.
Various Pt-based alloy catalysts have been investigated for PEFC
4,5
cathode catalysts besides non-Pt alternative catalyst materials
such as carbides and metal complex. To mention a few for PEFC
cathode catalysts, Toda et al. reported more than 10 times higher
activity of ORR for sputtered Pt alloys ͑with Ni, Co, and Fe͒ com-
6
pared to pure Pt in 0.1 M HClO solution at room temperature.
4
Mukerjee et al. found enhanced electrocatalytic activity of ORR for
some binary alloys between Pt and transition metals ͑e.g., Ni, Cr,
and Co͒ under PEFC conditions. They attributed the activity en-
hancement caused by alloying to the electronic interaction to form
Pt 5d band vacancies and the contraction in the Pt–Pt bond
7
distance.
It is also reported that for PEFC anode catalysts, Pt alloys such
8
-10
as Pt–Sn, Pt–Os, Pt–Mo,
and Pt-based composite catalysts with
11,12
oxides such as Pt/MoO and Pt/SnO ,
exhibit CO tolerance.
Igarashi et al. reported that sputtered Pt alloys ͑Pt–Ni, Co, Fe, Mo͒
x
x
Figure 1. pH-potential equilibrium ͑Pourbaix͒ diagram for the titanium-
water system at 80°C. The open circle shows cathodic condition and the
dotted circle shows anodic condition in PEFCs.
*
Electrochemical Society Active Member.
E-mail: sasaki@mech.kyushu-u.ac.jp
z
Downloaded on 2014-12-06 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B970
Journal of The Electrochemical Society, 154 ͑9͒ B969-B975 ͑2007͒
to prepare Pt–Ti alloy electrocatalysts for PEFCs and to analyze
their microstructure, electronic state, and electrochemical properties
under the cathodic and anodic conditions.
measurements to remove O dissolved in the HClO electrolyte so-
2 4
lution and to substitute the gas phase of the cell with N gas. Peaks
2
of hydrogen adsorption and desorption on the Pt surface could be
observed between 0.05 and 0.4 V. The CV measurements
were started after 200 cycles to remove any contamination
on the electrode. The coulombic charge of monolayer of Hupd ad-
Experimental
Electrocatalysts supported on carbon black ͑CB͒ were prepared
1
8-20
sorbed on Pt ͑Q ͒ between 0.05 and 0.4 V in CV was then obtained.
by colloidal
and sol-gel methods. Carbon black ͑Vulcan XC-72,
H
Cabot Corp.͒ was dispersed in distilled water and chloroplatinic acid
was added, which was then reduced by sodium hydrogen sulfite.
Colloidal Pt-oxide ͑hydroxide͒ was prepared by adding hydrogen
peroxide solution while adjusting the solution at pH 5 by NaOH. In
this dispersion, Ti oxide ͑hydroxide͒ was additionally prepared by
adding Ti͑OC H ͒ , which could be easily hydrated. Metal loading
The atomic density of the Pt polycrystalline surface is 1.31
1
5
−2
ϫ 10 cm and the coulombic charge of the hydrogen monolayer,
in which one hydrogen atom is adsorbed on one Pt atom,
−
2
is thus 210 ␮C cm . Electrochemical surface area ͑ECSA͒ was
2
therefore determined using the relation: ECSA ͑cm ͒
−
2
3
7 4
= QH ͑␮C͒/210 ͑␮C cm ͒.
was fixed to 20% for the catalysts. The initial atomic ratio of Pt and
Ti was set to 3:1, 1:1, or 1:3. Pt–Ti/C was heat-treated at tempera-
tures between 700 and 900°C in pure hydrogen for 2 h. Pt/C was
heat-treated at temperatures between 200 and 700°C in 5% H –N
RDE measurement for the ORR was performed at various rota-
tion speeds ͑ranging from 2700 to 900 rpm͒ after supplying O into
2
the electrolyte solution for 30 min. The electrode potential was
swept in the negative direction from 1.1 to 0.3 V at a scan rate of
2
2
−
1
for 2 h. With the notation of the catalyst samples, reducing tempera-
ture ͑°C͒ for the catalysts is shown in parenthesis throughout this
study, e.g., Pt3Ti1͑900͒ denotes the catalyst with the initial Pt-to-Ti
ratio of 3:1 heat-treated at 900°C in the hydrogen atmosphere. Com-
mercial Pt–Ru/C was also used for comparison, consisting of
Pt 30.4 wt % and Ru 23.6 wt %, supplied from Tanaka Kikinzoku
Kogyo.
Prepared catalysts were observed using field-emission scanning
electron microscope ͑FESEM, Hitachi S-5200͒ and scanning trans-
mission electron microscope ͑STEM, Hitachi HD-2200͒. Elemental
analysis was carried out by energy-dispersive X-ray analysis ͑EDX,
EDAX͒. In addition, crystallographic structure and electronic state
of the prepared catalysts were characterized by X-ray diffraction
10 mV s .
CO stripping voltammetry was separately carried out after the
corresponding CV measurements. For the CO-stripping voltamme-
try, pure CO was supplied into the electrolyte solution for 10 min
under the fixed potential at 0.05 V and then CO dissolved in the
electrolyte solution was removed by supplying N2 into the solution
for 40 min. The current-potential cycle was obtained from 0.05 to
−
1
1.1 V at 10 mV s .
For the electrochemical characterization in membrane-
electrode assembly ͑MEA͒, the electrocatalyst slurry was prepared
by mixing catalyst, 5% Nafion solution ͑Aldrich͒, and
ethanol ͑catalyst/5% Nafion solution/ethanol = 0.1 g:2.33 mL:0.89
mL͒. The catalyst slurry was deposited by the spray-printing tech-
nique ͑Nordson͒ on hydrophobic carbon papers coated with poly-
͑tetrafluoroethylene͒ ͑PTFE, Electrochem͒ and then dried at room
temperature. A Nafion membrane ͑DuPont, Nafion 112͒ was sand-
wiched with the catalyst-deposited carbon papers and then hot-
pressed at 130°C, 13 MPa to obtain an MEA. Catalyst loading was
͑
͑
XRD, Rigaku RINT2200V͒ and X-ray photoelectron spectroscopy
XPS, Kratos Analytical ESCA-3400͒. XRD patterns for the cata-
lysts were obtained using a Cu K␣ X-ray source ͑40 kV, 40 mA͒.
XPS measurements were made using a Mg K␣ source ͑12 kV,
20 mA͒. Au was sputtered on the catalysts and binding energy was
0.2 mg cm− for the anodes and 0.4 mg cm for the cathodes,
2
−2
corrected using the Au 4f7/2 peak located at 84.0 eV.
−
2
Cyclic voltammetry ͑CV͒ and the rotating disk electrode ͑RDE͒
while catalyst loading was 0.6 mg cm for the counter electrodes
using commercial catalyst ͑Tanaka Kikinzoku Kogyo,
44.6 wt % Pt/C͒. The MEA prepared was fixed in a sample holder
kept at 80°C in an atmospheric pressure. Single-cell testing was
carried out using H fuel or 100 ppm CO–H fuel supplied to the
2
1,22
method
of the catalysts. Electrochemical measurements were carried out in a
0 mL three-electrode cell ͑Hokuto Denko, HR200͒. A commercial
were applied to evaluate electrochemical characteristics
5
glassy carbon ͑GC͒ electrode ͑Hokuto Denko, HR2-D1-GC5, 5 mm
2
2
2
in diameter, 0.196 cm ͒, Pt-wire electrode ͑Hokuto Denko͒, and an
anodes, while air was supplied to the cathodes. Both gases were
humidified at 79.5°C so that the air and the fuel gas were almost
fully saturated with water vapor at 80°C, the operational tempera-
ture of the MEAs. Ohmic resistance and nonohmic overvoltage were
separately evaluated by applying the current-interrupt technique.
Ag/AgCl electrode ͑saturated in KCl͒ were used as a working elec-
trode, a counter electrode, and a reference electrode, respectively.
The potential of the RDE was controlled by an automatic polariza-
tion system ͑Hokuto Denko, HSV-100͒. A 0.1 M HClO solution
4
͑
Sigma-Aldrich͒ was used as the electrolyte solution for the RDE
Results and Discussion
measurements. At first, the GC electrode was polished using a sus-
pension ͑Baikowski͒ containing Al O particles of 0.3 ␮m diameter
2
3
Characterization of catalysts.— The crystal structure of cata-
lysts prepared via the colloidal and sol-gel method was examined by
XRD. Figure 2 shows the XRD patterns of the Pt/C and Pt–Ti/C
catalysts prepared in the initial Pt/Ti atomic ratio of 3:1, 1:1, and
1:3, denoted as Pt3Ti, Pt1Ti1, and Pt1Ti3, respectively. The initial
ratio of Pt and Ti had been verified by inductively coupled plasma
atomic emission spectrometry ͑ICP-AES͒. Figure 2 indicates that
the XRD peaks of the Pt–Ti/C were shifted to higher angles than
those of the Pt/C. The lattice constants calculated from the XRD
peaks of ͑111͒, ͑200͒, ͑220͒, ͑311͒, and ͑222͒ are listed in Table I.
The lattice constant of Pt–Ti/C was smaller than that of Pt/C
͑3.921 Å͒, indicating the formation of solid solution between Pt and
Ti. The samples Pt3Ti1 and Pt1Ti1 are partly ordered to form the
to obtain the clean GC surface and then it was ultrasonically cleaned
in distilled water. The prepared catalysts ͑25 mg͒ were dispersed
ultrasonically in distilled water ͑35 mL͒ to obtain a catalyst disper-
−
1
sion ͑0.71 mgcatal mL ͒. Using a micropipette, 28 ␮L of the sus-
pension was put on the GC disk electrode, and then the solvent was
evaporated in vacuum at room temperature. Furthermore, 10 ␮L
of diluted Nafion solution ͑5% Nafion solution/ethanol/water
=
1009 ␮L:10 mL:39 mL͒ were put onto the GC disk electrode and
the RDE was dried at 60°C for 1 h in order to obtain a well-
dispersed thin catalyst layer on the GC disk of the RDE. After dry-
ing, the catalyst layer was coated by a thin Nafion film with a thick-
ness of 0.1 ␮m, based on the procedure reported by Watanabe et
2
2
al.
intermetallic compound as Pt Ti, exhibiting their superlattice lines.
3
In this study, the electrochemical active surface area was deter-
However, sample Pt1Ti3 revealed no superlattice line. In addition,
mined by the hydrogen underpotential deposition ͑Hupd͒ region ob-
tained in the CV measurements. All potential values in this paper are
referred to a reversible hydrogen electrode ͑RHE͒. In the CV mea-
surements, potential cycles were applied between 0.05 and 1.2 V at
the XRD pattern of Pt1Ti3 shows peaks from Ti O as excess Ti
3
5
exists, i.e., the presence of excess Ti led to the preferential formation
of Ti O in Pt1Ti3.
3
5
The microstructure of the prepared catalysts was observed by
FESEM and STEM, and the average particle size ͑in diameter͒ was
−
1
a scan rate of 50 mV s . N gas was supplied for 10 min before CV
2
Downloaded on 2014-12-06 to IP 169.230.243.252 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, 154 ͑9͒ B969-B975 ͑2007͒
B971
Figure 3. STEM micrograph of Pt–Ti alloy electrocatalysts: ͑a͒ secondary
electron image and ͑b͒ transmission image.
XPS is a useful technique to analyze catalyst surfaces. However,
for analyzing nanosized support catalysts, size effect in XPS mea-
surements should be taken into account. Figure 5 summarizes bind-
ing energies of Pt 4f peak tops detected by XPS. With decreasing Pt
particle size ͑i.e., with decreasing heat-treatment temperature͒, bind-
ing energy became higher. The reason for this is generally explained
2
3
as being the size effects for ultrafine metal particles. It was there-
fore necessary to eliminate the size effects in order to discuss the
difference in materials characteristics. Therefore the XPS signal of
Pt͑700͒/C was compared to that of Pt–Ti/C, because Pt͑700͒/C had
almost the same particle size compared to Pt1Ti1͑900͒/C, as shown
in Table II. The Pt 4f XPS spectra of Pt͑700͒/C and Pt1Ti1͑900͒/C
are shown in Fig. 6. These doublet peaks were from metallic Pt but
not from oxide or chloride. In comparison with Pt͑700͒, the Pt 4f
peaks of Pt1Ti1͑900͒ were positively shifted. Figure 5 shows the
binding energy of all Pt–Ti alloys ca. 0.3 eV higher than that of
Figure 2. XRD patterns of prepared catalysts: ͑a͒ Pt͑200͒/C, ͑b͒
Pt3Ti1͑900͒/C, ͑c͒ Pt1Ti1͑900͒/C, and ͑d͒ Pt1Ti3͑900͒/C.
obtained, as listed in Table II. STEM images, shown in Fig. 3, con-
firm that Pt–Ti alloy particles were dispersed on the carbon support
and the particle size was ca. 5 nm diam. In addition, we can find
from Table II that grain growth of catalysts was depressed by alloy-
ing Pt with Ti, even though heat-treatment temperature was rela-
tively high, up to 900°C.
For supported alloy catalysts, their composition and uniformity
should be known. Therefore, elemental compositions of individual
alloy particles were analyzed using STEM-EDX. The content of Ti
in the Pt–Ti alloys was ca. 15 atom %, independent of the initial
Pt-to-Ti ratio of the samples. The composition was approximately
uniform. Figures 4b-d show the distribution of specific elements ͑Pt,
Ti, O͒ detected by STEM-EDX for the catalysts prepared, shown in
Fig. 4a. Excess Ti existed as oxides of ca. 100–350 nm diameter. Ti
oxides may be distributed as very thin film or in amorphous form,
which could be difficult to be detected by XRD.
Table I. Lattice constant of prepared catalysts, measured by
XRD.
Figure 4. ͑Color online͒ ͑a͒ STEM micrograph of Pt1Ti3/C electrocatalysts,
and the distribution of specific elements detected by STEM-EDX: ͑b͒ Pt, ͑c͒
Ti, and ͑d͒ O.
Catalyst
Lattice constant ͑Å͒
Pt͑200͒/C
3.921 ͑±0.0018͒
3.909 ͑±0.0013͒
3.905 ͑±0.0005͒
3.917 ͑±0.0006͒
Pt3Ti1͑900͒/C
Pt1Ti1͑900͒/C
Pt1Ti3͑900͒/C
Table II. Particle size of catalysts measured by using FESEM.
Catalyst
Pt
Pt3Ti1 Pt1Ti1 Pt3Ti1 PtRu a
Pt͑200͒ ͑450͒ Pt͑700͒ ͑900͒
͑900͒
͑900͒ ͑TKK͒
Particle
size
3.5 3.9 4.7 4.2
4.6
4.3 5.3
͑
nm͒
Figure 5. Binding energies of ͑a͒ Pt 4f7/2 and ͑b͒ Pt 4f5/2 photoelectron
spectra measured by XPS for various Pt-based catalysts.
a
TKKϭsupplied by Tanaka Kikinzoku Kogyo.
Downloaded on 2014-12-06 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B972
Journal of The Electrochemical Society, 154 ͑9͒ B969-B975 ͑2007͒
Figure 7. Tafel plots for the ORR on Pt/C catalysts for various heat-
treatment temperatures in O -saturated 0.1 M HClO at 25°C. The values of
jk at each potential were determined from the Koutecky-Levich plots. The
2
4
Figure 6. XPS spectra of Pt 4f doublets for Pt͑700͒ and Pt1Ti1͑900͒ cata-
lysts.
current density was normalized to ECSA measured by CV.
Pt͑700͒. This positive shift is similar to that reported for Pt–Ti al-
loys by Beard and Ross. This positive shift may indicate the in-
crease of Pt 5d vacancies by alloying with Ti.
Figure 8 shows Tafel plots of Pt and Pt–Ti alloy catalysts. Figure
24
9
shows j at 0.85, 0.80, and 0.75 V. All Pt–Ti alloys had higher j
k
k
than Pt, indicating improved ORR activity by alloying Pt with Ti.
Pt1Ti1͑900͒ exhibited the highest j . j of Pt1Ti1͑900͒ was about
k
k
ORR activity analyzed by RDE.— CV was applied to calculate
ECSA, and then the oxygen reduction activity of the catalyst was
5
0% higher at 0.8 V and twice higher at 0.75 V in comparison with
pure Pt. Tafel curves for smooth, polycrystalline Pt in diluted HClO4
examined by RDE in 0.1 M HClO . In general, electrode reactions
4
or H SO solutions can be fitted to two Tafel regions: a low-current-
2
4
occurred through ͑i͒ diffusion of reactants to the electrode surface,
ii͒ charge transfer on the electrode surface, and ͑iii͒ scattering of
density ͑LCD͒ region and a high-current-density ͑HCD͒ region.
Tafel curves were fitted and Tafel slopes were obtained. Table III
summarizes the Tafel slopes derived from Fig. 8. Tafel slopes be-
came smaller by alloying with Ti in both regions. A smaller Tafel
͑
products from the electrode surface. The slowest process is the rate-
controlling step of the overall electrode reactions. An activation pro-
cess can be rate-controlling if the rotation speed of the RDE is fast
enough. The current depends on the rotation rate, according to the
slope corresponds to a higher kinetic current density j at a given
k
potential and thus, improved ORR activity by alloying with Ti,
which leads to the difference in Tafel curves in the HCD region in
25
following Koutecky–Levich equation
1
1
1
1
1
=
+
=
+
͓1͔
I
Ik Ilim Ik 0.62nFAD2/3C␯−1/6␻1/2
where n is the number of electrons involved in the ORR per O
2
−
1
molecule, F, A, C, and D are Faraday constant ͑C mol ͒, geometri-
2
−3
cal surface area of the electrode ͑cm ͒, solubility ͑mol cm ͒, and
2
−1
the diffusion coefficient ͑cm s ͒ of O , respectively, x is kinetic
viscosity ͑cm s ͒, ␻ is the rotation rate ͑rad s ͒, and I is lim-
iting current ͑mA͒. I is the kinetic current and does not depend on
2
2
−1
−1
lim
k
rotation rate. Ik values for each catalyst and the potentials were
obtained by extrapolation of the Koutecky–Levich plots for ␻−
1/2
→
0. I was divided by ECSA calculated from CV curves.
k
−
2
Activation-controlled current density j ͑mA cm ͒ can be derived,
k
and Tafel plots of j were obtained.
k
There is a considerable number of reports concerning the effects
of particle size on the catalytic activity of Pt. Takasu et al. reported
that with decreasing Pt particle size, the specific activity decreases
2
3
and the Tafel slope increases by using Pt/GC model electrodes.
26
Watanabe et al. insisted that catalytic activity did not depend on
particle size. To examine the size effect of prepared catalysts, heat-
treatment temperature was varied, Pt particle size was controlled,
and ORR activity was evaluated by RDE. Figure 7 shows potential
vs j calculated from RDE measurements, indicating that Tafel slope
k
did not depend on Pt particle size for Pt/C heat-treated between 200
and 700°C ͑between 3.5 and 5 nm in diameter͒ under the conditions
in this study. In order to avoid the influence of particle size, Pt͑700͒
was used to compare Pt to Pt–Ti alloys for analyzing ORR activity.
Figure 8. Tafel plots for the ORR on Pt͑700͒, Pt3Ti1͑900͒, Pt1Ti1͑900͒,
and Pt1Ti3͑900͒ catalysts in O -saturated 0.1 M HClO at 25°C. The values
2
4
of j at each potential were determined from the Koutecky–Levich plots. The
k
current density was normalized to ECSA measured by CV.
Downloaded on 2014-12-06 to IP 169.230.243.252 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, 154 ͑9͒ B969-B975 ͑2007͒
B973
Figure 9. Kinetic current densities ͑j ͒ at 0.85, 0.80, and 0.75 V vs RHE in
k
0
2
.1 M HClO solution saturated with O . Sweep rate 10 mV/s, temperature
5°C. The current density was normalized to ECSA measured by CV.
4
2
27
Fig. 8. Paulus et al. studied Pt–Ni and Pt–Co alloys and reported a
similar decrease in Tafel slopes of supported alloy catalysts, which
was not observed for bulk alloys. This may also suggest the impor-
tance of both the difference of ORR activity between pure Pt and
Pt–Ti alloys and the influence of catalyst morphology.
CO tolerance analyzed by stripping voltammetry.— CO strip-
ping voltammetry was applied in 0.1 M HClO to electrochemically
4
evaluate CO tolerance of catalysts. The CO-stripping voltammo-
grams of prepared catalysts and commercial Pt–Ru/C ͑Tanaka
Kikinzoku Kogyo͒ are shown in Fig. 10. The potential of the
adsorbed-CO oxidation peak of Pt1Ti1͑900͒/C was 0.74 V, while
that of Pt͑200͒/C was 0.77 V. The potential of the adsorbed-CO
oxidation peak of Pt–Ru/C was 0.5 V, as Pt–Ru/C is highly
CO-tolerant. The CO-stripping peak potential for Pt–Ti/C showed a
negative shift of 0.03 V, however, this shift was smaller than that for
Pt–Ru/C ͑0.27 V͒. Pt–Ti alloys were thus considered to be slightly
CO-tolerant in comparison with pure Pt.
Figure 10. CO-stripping voltammograms of ͑a͒ Pt͑200͒/C, ͑b͒
Pt1Ti1͑900͒/C, and ͑c͒ commercial PtRu/C in 0.1 M HClO4 solution.
Sweep rate 10 mV/s, temperature 25°C.
using the current-interrupt technique. Cell voltage with Pt1Ti3 was
almost the same as that with pure Pt, but cell voltage with Pt1Ti1 or
Pt3Ti1 was higher than that with pure Pt. In other words, Pt1Ti1 or
Pt3Ti1 showed higher ORR activities than pure Pt, consistent with
ORR activity analyzed by single-cell tests.— Figure 11 shows
the fuel cell performance of single cells with Pt–Ti alloy electrocata-
lysts used for cathodes and with Pt/C prepared by the colloidal
method for comparison. Commercial Pt/C was used for anode cata-
lysts. Pt loading was 0.4 mg cm−2 for the cathode and 0.6 mg cm
for the anode. The single cell was kept at 80°C in atmospheric
−2
the RDE results. Improved electrochemical performance by alloying
28,29
with Ti is also consistent with literature.
For Pt1Ti3, excess Ti
oxides existing as Ti O ͑see Fig. 2͒ and twice-larger catalyst layer
3
5
pressure. Single-cell testing was carried out using H fuel, while air
2
thickness might also affect ORR activities.
was supplied to the cathodes. Both gases were humidified at 79.5°C.
Figure 11 shows internal-resistance-corrected cell voltage vs current
density curves for which internal ohmic losses had been removed by
CO tolerance analyzed by single-cell tests.— The fuel cell per-
formance of single cells with the Pt–Ti alloy electrocatalysts used
for the anodes was also evaluated. In this case, the commercial Pt/C
−
2
was used for cathode catalysts. Pt loading was 0.6 mg cm for the
cathode and 0.2 mg cm− for the anode. The single cell was again
kept at 80°C in atmospheric pressure. The test was carried out using
Table III. Tafel slope for ORR on Pt and Pt–Ti alloy catalysts
2
measured in 0.1 M HClO at 25°C.
4
Tafel slope ͑mV dec−1
pure H gas or H gas containing 100 ppm CO, while air was sup-
plied to the cathodes. Both gases were humidified at 79.5°C. Cell
voltage vs current density curves of the cells with Pt–Ti alloy cath-
͒
2 2
Catalyst
LCD region
HCD region
ode catalysts and Pt anode catalysts for the pure H fuel are shown
in Fig. 12. Cell voltages with the Pt–Ti alloy catalysts were lower
Pt͑200͒
67
58
56
58
121
99
98
2
Pt3Ti1͑900͒
Pt1Ti1͑900͒
Pt1Ti3͑900͒
than those with Pt. For the pure H fuel, the cell voltage decreased
2
103
by alloying Pt with Ti.
Downloaded on 2014-12-06 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

B974
Journal of The Electrochemical Society, 154 ͑9͒ B969-B975 ͑2007͒
Figure 11. Cell voltage vs current density curves in H /air of the cell
2
Figure 13. Cell voltage vs current density curves in 100 ppm CO-H /air of
2
with cathode catalysts Pt͑200͒/C, Pt3Ti1͑900͒/C, Pt1Ti1͑900͒, and
Pt1Ti3͑900͒/C, operated at 80°C. The fuel and air were humidified at 80°C
in atmospheric pressure. Platinum loading of the cathodes was
the cell with anode catalysts Pt͑200͒/C, Pt3Ti1͑900͒/C, Pt1Ti1͑900͒,
Pt1Ti3͑900͒/C, and PtRu/C, operated at 80°C. The fuel and air were humidi-
fied at 80°C at atmospheric pressure. Platinum loading of the anodes was
−
2
0
.4 mg Pt cm
.
−2
0
.2 mg Pt cm .
However, the voltage of the cell with the Pt–Ti alloy catalysts
reason could be the difference in experimental conditions:
CO-stripping voltammetry was applied using pure CO at 25°C,
while I-V measurements in single-cell tests were performed using
100 ppm CO at 80°C. In addition, CO-stripping voltammetry re-
veals the potential for CO desorption, while I-V characteristics in-
clude various kinetic effects such as desorption, adsorption, shift
reaction, and oxygen crossover through electrolytes.
used for the anodes became higher than those with pure Pt when H2
gas containing 100 ppm CO was supplied as shown in Fig. 13. The
measurements were performed after cell voltage became stable, and
the minimum cell voltages were recorded because of the cell voltage
30
vibration phenomena by CO poisoning. Voltage losses with Pt–Ti
alloys for the H fuel containing 100 ppm CO were smaller than
2
those with Pt. Especially, voltage loss with Pt1Ti1͑900͒/C was eight
Mechanism of ORR activity enhancement.— The results with
times lower than that with Pt͑200͒/C at the current density of
single cells indicate that ORR activity of the prepared Pt–Ti alloy
−
2
2
00 mA cm , i.e., CO tolerance was actually improved by alloying
7
catalysts improved in comparison with Pt. Mukerjee et al. reported
with Ti. However, the clear CO tolerance of Pt–Ti/C specified by
the single-cell test seemed somehow inconsistent with the weak CO
tolerance specified by CO-stripping voltammetry using RDE. One
that a volcano relation between oxygen reduction current and the d
6
vacancy of Pt alloys was measured by XAS. Toda et al. observed a
positive binding energy ͑BE͒ shift of Pt 4f and Pt 4d spectra for
Pt–Ni, Pt–Co, and Pt–Fe alloys and concluded that the positive shift
was due to the increased 5d vacancy of Pt, which led to increased ␲
electron donation from O to the Pt surface, resulting in increased
2
O adsorption and weakening of the O–O bond. A positive BE shift
2
of Pt 4f for Pt–Ti alloys was also observed by XPS ͑see Fig. 5 and
6
͒, and thus the enhancement of ORR activity may also be due to
increased vacancy of Pt in Pt–Ti alloys.
Another reason for the ORR activity enhancement may be the
change in the Pt–Pt nearest-neighbor bond distance. Jalan and
3
1
Taylor reported ORR activity was increased about three times by
alloying Pt with Ti, Cr, Fe, Co, and Ni, and they suggested the Pt–Pt
distance was shortened by alloying, which resulted in weakening of
the O–O bond. The lattice constant of the prepared catalysts de-
creased by alloying with Ti ͑Table I͒, and Pt1Ti1 showed the highest
ORR activity at the smallest lattice constant among the Pt–Ti alloy
catalysts. Therefore, the contraction of Pt–Pt distance was also re-
lated to the ORR activity improvement for the prepared Pt–Ti al-
loys.
The ORR activity enhancement leads to higher cell voltages us-
ing the Pt–Ti alloy electrocatalysts, as experimentally verified in this
study. However, because Pt loading for the electrodes was fixed in
this study, the thickness of electrocatalyst layers increased with in-
creasing Ti content. This could be one of the reasons for a decrease
in cell voltage using the Pt1Ti3 electrocatalyst with the twice-
thicker electrocatalyst layer in which more sluggish mass transport
may be expected. This thickness effect could also explain the de-
crease in cell voltage in case the Pt–Ti alloy catalysts were used as
Figure 12. Cell voltage vs current density curves in H /air of the cell with
2
anode catalysts Pt͑200͒/C, Pt3Ti1͑900͒/C, Pt1Ti1͑900͒, Pt1Ti3͑900͒/C,
and PtRu/C, operated at 80°C. The fuel and air were humidified at 80°C at
−
2
atmospheric pressure. Platinum loading of the anodes was 0.2 mg Pt cm
.
Downloaded on 2014-12-06 to IP 169.230.243.252 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, 154 ͑9͒ B969-B975 ͑2007͒
B975
anode catalysts for ͑pure͒ H2 fuels, if no enhancement effect of
hydrogen oxidation reactions by alloying with Ti exists.
vealing that Pt–Ti alloys exhibited higher ORR activity and slight
CO tolerance compared to pure Pt under PEFC conditions. XPS
results indicated the formation of additional d vacancies in Pt by
alloying with Ti, which could be one of the reasons for Pt–Ti alloy
electrocatalysts exhibiting higher ORR activity and CO tolerance
compared to pure Pt.
Mechanism of CO tolerance enhancement.— The
prepared
Pt–Ti alloy electrocatalysts were more CO-tolerant compared to
pure Pt. It was reported that methanol oxidation on the Pt surfaces
was promoted by the presence of an oxide ͑Nb O , Ta O , ZrO ,
2
5
2
5
2
1
2
TiO2͒. It was considered that oxides act as a source of OH species.
In addition, Ioroi et al. prepared the mixed Pt–MoO catalysts,
Kyushu University assisted in meeting the publication costs of this
article.
11
x
exhibiting CO tolerance equal to the Pt–Ru alloy. They suggested a
water-gas shift reaction ͑WGS͒ promoted CO oxidation by coexist-
ing Pt and MoO . While excess TiO might affect the CO tolerance
References
1
2
. R. A. Lemons, J. Power Sources, 29, 367 ͑1990͒.
. M. Watanabe and S. Motoo, J. Electroanal. Chem. Interfacial Electrochem., 60,
275 ͑1975͒.
3. H. F. Oetjen, V. M. Schmidt, U. Stimming, and F. Trila, J. Electrochem. Soc., 143,
838 ͑1996͒.
. S. Mukerjee and S. Srinivasan, J. Electroanal. Chem., 357, 201 ͑1993͒.
. E. Antolini, Mater. Chem. Phys., 78, 563 ͑2003͒.
. T. Toda, H. Igarashi, H. Uchida, and M. Watanabe, J. Electrochem. Soc., 146, 3750
͑1999͒.
x
x
of our prepared catalysts because a part of Ti existed as TiO ͑see
x
11
Fig. 2 and 4͒, the enhanced CO tolerance was mainly caused by
alloying, because Pt1Ti1 exhibited the highest CO tolerance while
Pt1Ti3 exhibited almost the same CO tolerance as pure Pt, which
was revealed by the single-cell performance tests.
3
4
5
6
13
Igarashi et al. examined hydrogen oxidation activity for more
than 40 kinds of alloy catalysts in 100 ppm CO containing hydrogen
fuels. They revealed that CO coverage remained low in
CO-containing fuels for Pt–Fe, Pt–Ni, Pt–Co, and Pt–Mo alloy, as
well as conventional Pt–Ru, and high hydrogen oxidation activity
could be maintained. In addition, electronic states of the alloy sur-
faces after the electrochemical measurements were examined by
XPS, which revealed that the second constituent ͑besides Ru͒ was
dissolved from the alloy surfaces, and Pt skin layers with increased
7
8
. S. Mukerjee, S. Srinivasan, M. P. Soriaga, and J. McBreen, J. Electrochem. Soc.,
1
42, 1409 ͑1995͒.
. K. Wang, H. A. Gasteiger, N. M. Markovic, and P. N. Ross, Jr., Electrochim. Acta,
1, 2578 ͑1997͒.
9. B. N. Grgur, G. Zhuang, N. M. Markovic, and P. N. Ross, Jr., J. Phys. Chem. B,
01, 3910 ͑1997͒.
0. R. Liu, H. Iddir, Q. Fan, G. Hou, A. Bo, K. L. Ley, E. S. Smotkin, Y.-E. Sung, H.
Kim, S. Thomas, and A. Wiekowski, J. Phys. Chem. B, 104, 3518 ͑2000͒.
1. T. Ioroi, K. Yasuda, Z. Siroma, N. Fujiwara, and Y. Miyazaki, J. Electrochem. Soc.,
4
1
1
1
150, A1225 ͑2003͒.
1
3
1
1
2. A. Hammett, Catal. Today, 38, 445 ͑1997͒.
3. H. Igarashi, T. Fujino, Y. Zhu, H. Uchida, and M. Watanabe, Phys. Chem. Chem.
d vacancies were formed. However, in their study, binding energy
͑
BE͒ of Pt 4f for their Pt–Ti alloy was 0.3 eV negatively shifted,
Phys., 3, 306 ͑2001͒.
and Pt–Ti alloy did not show enhanced CO tolerance. This result
seemed inconsistent with our study, and one of the reasons may be a
difference between the bulk alloy and supported alloy and/or a dif-
ference in the catalyst preparation procedures. The BE of Pt 4f for
Pt–Ti alloy catalysts was shifted to the positive direction compared
with pure Pt, as shown in Fig. 5 and 6. This positive shift may be
associated with the formation of d vacancies contributing to the
enhanced CO tolerance. In addition, if Ti atoms existed on the alloy
1
1
4. R. C. Urian, A. F. Gullá, and S. Mukerjee, J. Electroanal. Chem., 554-555, 307
͑2003͒.
5. M. Watanabe, K. Tsurumi, T. Mizukami, T. Nakamura, and P. Stonehart, J. Elec-
trochem. Soc., 141, 2659 ͑1994͒.
6. T. Ioroi and K. Yasuda, J. Electrochem. Soc., 152, A1917 ͑2005͒.
7. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, Pergamon
Press, New York ͑1974͒.
8. H. G. Petrow and R. Allen, U.S. Pat. 4,044,193 ͑1977͒.
9. M. Watanabe, M. Uchida, and S. Motoo, J. Electroanal. Chem. Interfacial Electro-
chem., 229, 395 ͑1987͒.
1
1
1
1
surfaces, they might have bonded with oxygen as TiO , one of the
2
20. M. K. Ravikumar and A. K. Shukla, J. Electrochem. Soc., 143, 2601 ͑1996͒.
21. D. T. Sawyer, A. Sobkowiak, and J. L. Roberts, Jr., Electrochemistry for Chemists,
most stable compounds under PEFC conditions. It may be possible
2
nd ed., John Wiley & Sons, New York ͑1995͒; W. Vielstich, A. Lamm, and H. A.
that CO tolerance was enhanced by CO oxidation on Pt neighboring
2,3
Gasteiger, Handbook of Fuel Cells: Fundamentals, Technology and Applications,
Vol. 2, Electrocatalysis, John Wiley & Sons, West Sussex ͑2003͒; T. J. Schmidt, H.
A. Gasteiger, and R. J. Behm, J. Electrochem. Soc., 146, 1296 ͑1999͒.
2. M. Watanabe, H. Igarashi, and K. Yoshioka, Electrochim. Acta, 40, 329 ͑1995͒.
3. Y. Takasu, N. Ohashi, X.-G. Zhang, Y. Murakami, H. Minagawa, S. Sato, and K.
Yahikozawa, Electrochim. Acta, 41, 1595 ͑1996͒.
Ti via, e.g., the bifunctional mechanism.
Conclusion
2
2
Pt–Ti alloy electrocatalysts supported on carbon black for PEFCs
were prepared by the combined colloidal and sol-gel method. The
oxygen reduction reaction ͑ORR͒ activity and CO tolerance of the
prepared catalysts were examined using a rotating disk electrode
2
2
4. B. C. Beard and P. N. Ross, Jr., J. Electrochem. Soc., 133, 1839 ͑1986͒.
5. A. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamentals and Ap-
plications, p. 291, Wiley, New York ͑1980͒.
2
2
6. M. Watanabe, S. Saegusa, and P. Stonehart, J. Electroanal. Chem. Interfacial Elec-
trochem., 261, 375 ͑1989͒.
7. U. A. Paulus, A. Wokaun, G. G. Scherer, T. J. Schmidt, V. Stamenkovic, N. M.
͑
RDE͒. Pt–Ti alloy catalyst particles were ca. 4.5 nm in diameter
and homogeneously dispersed on carbon black. ORR activity of
Pt–Ti alloys was higher than pure Pt. ORR activity of the Pt–Ti
alloy, with the initial atomic ratio of Pt and Ti set to 1:1, was en-
hanced about 1.5 times in comparison with pure Pt at 0.8 V.
CO-stripping voltammetry revealed that the potential of the CO oxi-
dation peak was slightly shifted to the negative potential by alloying
with Ti. Prepared catalysts were evaluated in single-cell tests, re-
Markovic, and P. N. Ross, Electrochim. Acta, 47, 3787 ͑2002͒.
28. J. Shim, C.-R. Lee, H.-K. Lee, J.-S. Lee, and E. J. Cairns, J. Power Sources, 102,
72 ͑2001͒.
1
2
3
9. L. Xiong and A. Manthiram, Electrochim. Acta, 49, 4163 ͑2004͒.
0. H. Fukumoto, H. Maeda, and K. Mitsuda, Electrochemistry (Tokyo, Jpn.), 68, 794
͑2000͒.
31. V. Jalan and E. J. Taylor, J. Electrochem. Soc., 130, 2299 ͑1983͒.
Downloaded on 2014-12-06 to IP 169.230.243.252 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).