UHV, electrochemical NMR, and electrochemical studies of platinum/ruthenium fuel cell catalysts

Article abstract of DOI:10.1021/jp020169u

It is well-known that platinum/ruthenium fuel cell catalysts show enhanced CO tolerance compared to pure platinum electrodes, but the reasons are still being debated. We have combined cyclic voltammetry (CV), temperature programmed desorption (TPD), electrochemical nuclear magnetic resonance, and radio active labeling to probe the origin of the ruthenium enhancement in Pt electrodes modified through Ru deposition. The results prove that the addition of ruthenium not only modifies the electronic structure of all the platinum atoms but also leads to the creation of a new form of adsorbed CO. This new form of CO may be ascribed to CO chemisorbed onto the Ru region of the electrode surface. TPD and CV results show that the binding of hydrogen is substantially modified due to the presence of Ru. Surprisingly though, TPD indicates that the binding energy of CO on platinum is only weakly affected. Therefore, the changes in the bond energy of CO due to the ligand effect only play a small role in enhancing CO tolerance. Instead, we find that the main effect of ruthenium is to activate water to form OH. Quantitative estimates based on the TPD data indicate that the bifunctional mechanism is about four times larger than the ligand effect.

Full text of DOI:10.1021/jp020169u

J. Phys. Chem. B 2002, 106, 9581-9589
9581
UHV, Electrochemical NMR, and Electrochemical Studies of Platinum/Ruthenium Fuel Cell
Catalysts
C. Lu, C. Rice, and R. I. Masel*
UniVersity of Illinois, Department of Chemical Engineering, Urbana, Illinois 61801
P. K. Babu, P. Waszczuk, H. S. Kim, E. Oldfield, and A. Wieckowski*
UniVersity of Illinois, Department of Chemistry, Urbana, Illinois 61801
ReceiVed: January 17, 2002; In Final Form: June 10, 2002
It is well-known that platinum/ruthenium fuel cell catalysts show enhanced CO tolerance compared to pure
platinum electrodes, but the reasons are still being debated. We have combined cyclic voltammetry (CV),
temperature programmed desorption (TPD), electrochemical nuclear magnetic resonance, and radio active
labeling to probe the origin of the ruthenium enhancement in Pt electrodes modified through Ru deposition.
The results prove that the addition of ruthenium not only modifies the electronic structure of all the platinum
atoms but also leads to the creation of a new form of adsorbed CO. This new form of CO may be ascribed
to CO chemisorbed onto the “Ru” region of the electrode surface. TPD and CV results show that the binding
of hydrogen is substantially modified due to the presence of Ru. Surprisingly though, TPD indicates that the
binding energy of CO on platinum is only weakly affected. Therefore, the changes in the bond energy of CO
due to the ligand effect only play a small role in enhancing CO tolerance. Instead, we find that the main
effect of ruthenium is to activate water to form OH. Quantitative estimates based on the TPD data indicate
that the bifunctional mechanism is about four times larger than the ligand effect.
Introduction
methanol) adsorbed on Pt electrode surfaces decorated with Ru
has shown that there is a significant electronic alteration for
platinum atoms near ruthenium islands. However, Pt atoms away
from Ru are only weakly affected.
Ruthenium addition is known to substantially enhance the
CO tolerance of platinum fuel cell anodes. The increased CO
tolerance allows fuel cells to run with reformates. It also
enhances the rate of direct methanol electrooxidation. The origin
of ruthenium enhancement is not completely understood. There
are two widely accepted theories for the enhancement: the
bifunctional mechanism1-5 and the ligand model.6-10 The
bifunctional mechanism holds that the main effect of the
ruthenium addition is to activate water molecules on the
ruthenium surface to yield Ru-OH
In this report, several surface techniques are combined to
probe the role of ruthenium in improving the CO tolerance of
platinum fuel cell catalysts. Cyclic voltammetry (CV) is used
to determine how the kinetics of CO and hydrogen removal
from the substrate change in the presence of ruthenium.
Electrochemical nuclear magnetic resonance (EC NMR) is used
to provide a measure of the electronic interactions between
adsorbed CO (from CO saturated media) and the metal substrate.
Radioactive labeling is used to measure the onset potential for
CO oxidation. Temperature programmed desorption (TPD) is
used to quantify the changes in binding energies of the various
species, and the activation energy of surface hydroxide forma-
tion. A comparison is made between the measurements carried
out on ruthenium modified Pt(110) single crystals and platinum
nanoparticles.
Ru0 + H2O f Ru-OH + e- + H+
(1)
CO then migrates along the anode surface to react with the Ru-
OH
Pt-CO + Ru-OH f CO2 + e- + H+ + Pt0 + Ru0 (2)
Pure ruthenium forms surface hydroxides at lower potentials
(>0.3 V vs RHE) than on platinum (0.6 V vs RHE),3,6,11,12 so
the removal of CO is much faster on a platinum/ruthenium
surface than on pure platinum.
The Ligand model asserts that the main role of ruthenium is
to modify the electronic structure of platinum surface by
interacting with the conduction band of platinum. This modi-
fication is then presumed to induce a weakening of the Pt-CO
bond, so that less energy is required to oxidize the adsorbed
CO. Previous NMR investigation13 of CO (generated from
Experimental Section
The TPD experiments were preformed on a Pt(110) single
crystal, in a UHV chamber having a base pressure of 1 × 10-10
Torr. The apparatus and procedures have been described
elsewhere.14-17 The Pt (110) single crystal was cleaned by
heating in oxygen at 800 °C and sputtering with argon ions,
followed by annealing for 3 min at 1000 °C. Ru deposited on
the single crystals through spontaneous deposition when used
for UHV experiments always showed some carbon contamina-
tion that was difficult to remove. Therefore, chemical vapor
deposition (CVD) was used to decorate the Pt single crystals
* To whom correspondence should be addressed.
10.1021/jp020169u CCC: $22.00 © 2002 American Chemical Society
Published on Web 08/24/2002

9582 J. Phys. Chem. B, Vol. 106, No. 37, 2002
Lu et al.
Figure 1. (A) Cyclic voltammogram of a clean Pt(110) electrode at a sweep rate of 50 mV s-1 in 0.1 M HClO4 solution. Focusing on the hydrogen
adsorption/desorption region. (B) A series of 2 amu TPD spectra taken by exposing a clean Pt(110) surface to various amounts of H2 at 100 K and
then heating at 15 K/s.
with Ru for carrying out the UHV experiments by heating the
sample to 220 °C in the presence of 5 × 10-8 Torr of Ru3-
(CO)12. Ru3(CO)12 deposits a mixture of ruthenium and carbon,
and so the carbon was removed by heating the sample to 300
°C in 2 × 10-7 Torr of oxygen for 30 min. The sample was
then cooled to 100 K, and 5 Langmuirs (L) of hydrogen were
added to reduce the surface oxide. Finally, the sample was
flashed to 600 K to remove any residual hydrogen. A ruthenium
coverage of 0.25 monolayers was obtained in this way. The
sample was cooled to 110 K and dosed with hydrogen, CO, or
water by backfilling the chamber. The TPD measurements were
carried out at a heating rate of 15 K/s.
time (12 h) at low potentials, 0.9 monolayer CO coverage was
attained. CO stripping was performed at a scan rate of 10 mV/
min.
Technical details of the 13C EC NMR measurements have
been reported previously.13,20 The Pt black based catalysts were
prepared as above by spontaneous deposition of Ru and NMR
samples were made from these catalysts as follows: CO was
chemisorbed onto about 500 mg of catalyst from 13C (99%)
enriched CO solutions at an open circuit. After 1 h of adsorption,
excess CO was removed by purging with argon and rinsing with
0.1 M D2SO4. The nanoparticles were then transferred into an
NMR ampule containing a small (1-2 mL) portion of electrolyte
and then flame sealed. Spectra and relaxation rates were obtained
with a home-built solid-state NMR spectrometer, operated in
8.5 T magnetic field.
The radioactive labeling experiments21 were done under
potential control and slow scan-rate conditions, in a three-
electrode electrochemical cell. The bottom of the cell was a
radiation detector: a glass scintillation disk. The disk was
covered with 7.5 µm thick polyimide film, onto which a 500 Å
layer of gold was vacuum deposited. The working electrode
was prepared by placing platinum black or ruthenium decorated
Pt nanoparticles on top of the gold conducting support. CO is
adsorbed onto the surface from 0.01 M methanol and 0.1 M
H2SO4 solution at 0.2 V vs a standard hydrogen reference
electrode.
The Pt(110) single crystal used in the hydrogen adsorption/
desorption electrochemical measurements was initially flame
annealed and cooled in a nitrogen/hydrogen atmosphere.
Spontaneous deposition of ruthenium was carried out at 0.3 V
vs a RHE from RuCl3 dissolved in 0.1 M HClO4 to form 0.20
monolayers of Ru .18 The cyclic voltammograms were acquired
either in 0.1 M HClO4 or in 0.1 M H2SO4 at a scan rate of 50
mV/sec.
Commercial fuel cell grade platinum black nanoparticles
(Johnson Matthey) were modified by spontaneous deposition
of ruthenium onto the platinum substrate to yield from 0.14 to
0.52 monolayers of ruthenium, as estimated by inductively
coupled plasma spectroscopy. From scanning tunneling micro-
scopy of Pt single crystals decorated by Ru through spontaneous
deposition, we have found that Ru forms essentially monolayer
islands of about 3-5 nm.19 Though, there has not been any
direct evidence of the form in which Ru is deposited on Pt
nanoparticle electrode surfaces, we assume that the behavior is
not different from that of Pt single-crystal surfaces. Electro-
chemical measurements were preformed on catalysts prepared
by this technique. Current-time measurements of 0.5 M
methanol electrooxidation in 0.5 M H2SO4 were carried out at
0.3 V vs RHE with the current levels reaching steady-state
values within 10 h. Prior to the cyclic voltammetry measure-
ments, methanol was chemisorbed onto the catalyst surface to
form CO at 0.19V in 0.5 M H2SO4. After extended adsorption
Results: Adsorption/Desorption Studies
Hydrogen Desorption. Figure 1A,B compares the CV and
TPD spectra of hydrogen desorption from Pt(110). The CV
shows that hydrogen desorbs in two peaks at 0.086 and 0.150
V. The adsorption part of the CV consists of a peak at 0.084 V
and a sideband between 0.150 and 0.20 V (Figure 1A), in
addition to a peak near 0 V, associated with H2 evolution. These
data were taken in perchloric acid. When the same measure-
ments were done in sulfuric acid, the 0.150 V band moves to
0.160 V and the 0.086 V band is suppressed due to interaction
with the sulfate ions.

Platinum/Ruthenium Fuel Cell Catalysts
J. Phys. Chem. B, Vol. 106, No. 37, 2002 9583
Figure 2. (A) Cyclic voltammogram of a Pt(110)/Ru electrode (obtained through spontaneous deposition) at a sweep rate of 50 mV s-1 in 0.1 M
HClO4 solution. (B) A series of 2 amu TPD spectra taken by exposing Pt(110)/Ru (obtained through CVD) to various amounts of H2 at 100 K and
then heating at 15 K/s. At low exposures the hydrogen coverage is less that a full monolayer, while the coverage at the maximum exposure time
is in slight excess of a full monolayer.
At coverages below 1/2 of saturation, the TPD spectrum shows
a single peak, commonly called the â2 peak, at 300 K (Figure
1B).17 At higher coverages, a second peak, called the â1 peak
grows into the spectrum at 270 K.17 The â1 peak shifts to 230
K with increasing coverage. At coverages above 40% of
saturation, an additional peak commonly called the R peak is
seen at 180 K.17
Figure 2A,B shows cyclic voltammogram and TPD spectra
on ruthenium modified Pt(110) samples. The CV spectrum
shows a hydrogen desorption peak at 0.102 V and an adsorption
feature at 0.092 V (Figure 2A). There is also a hydrogen
desorption feature at 0.040 V and an H2 evolution feature at
0.0 V. A comparison of Figures 1A and 2A shows that the
addition of a quarter of a monolayer ruthenium decreases the
primary hydrogen desorption potential by about 10 mV. The
secondary desorption peak observed on the clean platinum
surface disappears with the addition of a quarter of a monolayer
ruthenium. The peaks are also broadened due to the increase in
surface heterogeneity. There is also a 30% decrease in the
amount of hydrogen adsorbed.
Figure 2B shows a series of TPD spectra for hydrogen
desorption from a ruthenium modified Pt(110) surface. Peaks
are seen at 140∼170 K and at 260 K. The peaks are broader
than that on the clean surface, similar to the behavior observed
for CV. The total amount of hydrogen adsorption decreases to
a little less than one-half when ruthenium is present on the
surface. A comparison of Figures 1B and 2B shows that the
binding of hydrogen is substantially modified by the presence
of ruthenium, proving that ruthenium has a significant effect
on the properties of platinum, even though pure ruthenium binds
hydrogen strongly.27
CO Binding. Figure 3A,B represents CO peaks on pure
platinum black probed via CO stripping CV and CO desorption
TPD. In Figure 3A, CO was adsorbed onto platinum black
nanoparticles to form 0.9 monolayers. CO is then removed from
the surface by slowly scanning the potential (10 mV/min). As
the potential becomes more anodic, water is activated on the
platinum surface, and CO is oxidized by, for example, the
following reaction
3B shows a series of TPD spectra for CO desorption from
Pt(110). At moderate coverages, only one peak is present at
515 K. At higher CO coverages, a second peak begins to form
at 415 K.
13C EC NMR spectrum of chemisorbed CO from the CO
saturated solution on platinum black nanoparticles in 0.1 M H2-
SO4 is shown in Figure 4A. A single resonance peak at 369
ppm is found as before in experiments where CO was generated
from methanol. For comparison, the 13C NMR spectrum of CO
adsorbed via methanol oxidation is given in Figure 4B. Because
about 500 mg of Pt nanoparticles are required for the NMR
sample preparation, methanol oxidation is found to produce CO
coverages in the region 0.7-0.8 monolayers only. On the other
hand, when CO is adsorbed from CO saturated solution, surface
coverage of about 0.9 monolayers can be achieved. The slight
increase in the width of the NMR spectrum for the CO
chemisorbed from the gas-phase arises due to the increased
dipolar interaction between the CO molecules as a consequence
of the higher CO coverage. The relaxation rates and their
temperature variation (see Table 1) are found to be identical
for both the cases, implying that CO adsorbed on Pt has the
same electronic properties irrespective of its source of genera-
tion.
Figure 5 shows how the presence of ruthenium alters the CO
stripping CV, TPD spectrum for CO desorption and 13C NMR
spectrum of adsorbed CO. Two CO stripping peaks are obtained
from slow scan cyclic voltammetry (10 mV/min) on ruthenium
modified platinum nanoparticles at 0.407 and 0.30 V (Figure
5A), corresponding to CO bound to platinum and ruthenium.
There are also two peaks at around 480 and 400 K in TPD
(Figure 5B), and two peaks at 359 and 159 ppm can be obtained
through standard deconvolution of the 13C NMR spectrum
(Figure 5C). A comparison of these results to those in Figures
3 and 4 show that there are two key effects due to ruthenium
addition: (i) a small shift in the peak positions for all of the
spectra and (ii) formation of a weakly bound form of CO that
can be associated with CO adsorbed on or near Ru islands.
It is interesting to quantify these two key effects due to the
presence of Ru. An Arrhenius plot on the initial part of the
TPD curves shows a 2 kcal/mol decrease in the binding energy
of CO compared to the clean surface. There is also a 10 ppm
shift in the 13C NMR of CO bound to Pt atoms away from Ru
islands (remote Pt). Thus, it appears that all the CO molecules
Pt-CO + Pt-OH f CO2 + 2Pt0 + e- + H+
(3)
Figure 3A shows only one CO stripping peak at 0.522 V. Figure

9584 J. Phys. Chem. B, Vol. 106, No. 37, 2002
Lu et al.
Figure 4. (A) 13C NMR spectrum of chemisorbed CO from the CO
saturated solution on Pt-black nanoparticles. (B) 13C NMR spectrum
of chemisorbed CO on Pt-black from methanol oxidation.
TABLE 1: Comparison of the 13C NMR Spectrum of CO
Produced from CO and Methanol on a Nanocrystalline
Platinum Catalyst.
CO (from MeOH)
CO (from CO gas)
peak (ppm)
369
59
0.6
7.0
369
57
0.6
7.3
T1T (s.K)
D5σ(Ef) (Ry.atom)-1
D
2π*( Ef) (Ry.atom)-1
in a fuel cell because CO from the Ru phase is oxidized at
potentials at or below those typically used in a fuel cell.
Figure 6 shows how the CO coverage varies with exposure,
during CO adsorption on Pt(110) single crystals in UHV. On
the clean surface, there is an initial rapid rise in the coverage,
and then the coverage levels off. Similar behavior is seen on
the ruthenium modified Pt(110) sample except that the saturation
coverage is less than half of that on the clean surface.
Interestingly, the initial parts of the curves are identical on the
two samples suggesting that again the sticking probability is
identical on the two samples.
Figure 7A,B illustrates the effect of ruthenium on the back-
bonding of CO (adsorbed from CO saturated solutions) as
measured by 13C NMR. A single NMR peak is seen on the clean
platinum surface at 369 ppm (Figure 7A). As the coverage of
ruthenium is increased a second peak begins to emerge near
150 ppm.13 Observation of two peaks for the NMR spectra is
consistent with the two peaks found for the CO stripping CV
of Ru covered Pt electrode surfaces. Therefore, we can ascribe
the two-peak structure observed for CV and NMR to the same
origin viz., formation of two different forms of adsorbed CO.
Thus, for the deconvolution of the 13C NMR spectra, we kept
the ratio of the areas of the two peaks obtained from the CV,
as a constraint. Relaxation measurements were carried out at
these two peak positions to estimate the Fermi level local density
of states (Ef-LDOS). The 150 ppm peak continues to grow with
subsequent increase in the amount of deposited Ru, whereas
the high field peak shift decreases from 370 to 350 ppm. Using
the two-band models developed earlier and the correlations
between 13C chemical shift and the clean metal surface
Ef-LDOS,20 it was possible to obtain the 2π* Ef-LDOS at the
orbital of chemisorbed CO as well as that the total Ef-LDOS at
the clean metal surface. Variation of these Ef-LDOS values with
Ru coverage are shown in Figure 7B. For the weakly bound
form of CO, there is about 70% reduction in the metal surface
Ef-LDOS and about 30% reduction in the 2π* Ef-LDOS (open
Figure 3. (A) CO stripping voltammogram on Pt black nanoparticles,
at a slow scan rate of 10 mV/min in 0.5 M H2SO4, after 12 h of
methanol oxidation to form CO at 0.19 V vs RHE. (B) A series of 28
amu TPD spectra taken by exposing Pt(110) to various amounts of
CO at 100 K and then heating at 15 K/s.
bound to platinum are affected by ruthenium. However, the
effect is smaller than that found in the hydrogen TPD spectra
in Figure 2. Evidently, ruthenium is not changing the binding
of CO to remote Pt to a great extent, even though there is a
substantial effect on the binding of hydrogen.
For the weakly bound form of CO, TPD shows that a peak
appears at 400 K for the ruthenium covered electrode surfaces.
The peak in the NMR spectrum corresponding to this CO has
a smaller Knight shift, proving that this CO has a weaker
interaction with the metal surface than those on the clean surface.
From the CO stripping CV, it is clear that the CO adsorbed on
the Ru phase is easier to oxidize than the CO on Pt phase of
the Pt/Ru electrode. However, the former peak is not important

Platinum/Ruthenium Fuel Cell Catalysts
J. Phys. Chem. B, Vol. 106, No. 37, 2002 9585
Figure 6. Comparison of CO adsorption kinetics on Pt(110) and
Pt(110)/Ru vs exposure time by TPD methods. The coverages are
calculated from integration of the TPD peak area normalized to the
saturation coverage.
ruthenium content. A maximum enhancement in oxidation
current by a factor of about 20 over that on pure platinum is
found for the catalyst with 0.35 monolayers of Ru (Figure 8).
The catalytic activity appears to level off at ca. 0.5 monolayers
of ruthenium and a decrease in activity is observed for higher
ruthenium coverages.22
TPD of 18O. At this point, we have shown that there is about
a 0.2 V decrease in the electro-catalytic potential to oxidize the
strongly bound CO from the platinum surface, but TPD and
NMR results show that the binding of strongly bound CO on
platinum is only moderately affected by ruthenium. Therefore,
we need to consider how ruthenium is making it easier to oxidize
the CO when it is not strongly affecting its binding. One
possibility is that ruthenium is activating water or hydroxyl on
the surface which can increase the CO oxidation rate which is
commonly referred to as the bifunctional mechanism. Figure 9
shows TPD results involving hydroxyls on Pt(110) and Pt(110)/
Ru surfaces from ref 17. 18O2 is adsorbed on the surface before
normal H216O is added. Mass 20 amu spectrum shows two
peaks at 160 and 205 K. The peak at lower temperature results
from the 18O exchange process through OH as the intermediate23
Figure 5. (A) CO stripping voltammogram on Ru modified Pt black
nanoparticles (0.35 monolayers of Ru), at a slow scan rate of 10 mV/
min in 0.5 M H2SO4, after 12 h of methanol oxidation to form CO at
0.19 V vs RHE. (B) A series of 28 amu TPD spectra taken by exposing
Pt(110)/Ru (0.25 monolayers of Ru) to various amounts of CO at 100
K and then heating at 15K/s. (C) Spin-echo mapped 13C EC NMR
spectra of chemisorbed 13C labeled CO on Ru modified Pt black
nanoparticles (0.35 monolayers of Ru) in 0.5 M H2SO4, formed from
12 h methanol oxidation at 0.19 V vs RHE.
18O + H2Of 18OH +OH
18OH + OH f H218O +O
(4)
(5)
symbols in Figure 7B). However, for Pt atoms away from Ru,
the changes in Ef-LDOS are comparatively less.
This reaction is water desorption rate limited.24 The second peak
involves hydroxyl formation and recombination. Oxygen react-
ing with hydrogen in the background yields hydroxyls, which
then recombine to produce water.23 This peak shows up by itself
when water is not present. On Pt(110)/Ru surface, similar
experiments are done with twice the amount of water dosage
which gives approximately the same amount of adsorbed water
as in the case of clean platinum surface. Only one peak at 160
K appears whose area is even larger than the sum of the two
peaks on Pt(110). This indicates that the peak at 205 K on
Pt(110) is now shifted below 160 K and the process of 18O
exchanging into water is enhanced. These results prove that
Results: Kinetic Measurements
Current-Time Transients. Next, we wish to examine the
effect of ruthenium on the kinetics of methanol oxidation on
platinum anodes. Figure 8 compares the methanol electrooxi-
dation activity on platinum black nanoparticles modified with
various amounts of Ru. On pure platinum nanoparticles, the
methanol electrooxidation current at 0.3 V vs RHE decays
immediately, due to CO passivation. In contrast, significant
currents are observed on the ruthenium modified surfaces. The
steady-state methanol oxidation current density also varies with

9586 J. Phys. Chem. B, Vol. 106, No. 37, 2002
Lu et al.
Figure 8. Current-time transients at 0.3 V vs RHE on both pure Pt
black and Ru modified Pt black electrodes in 0.5 M methanol solution.
The Ru coverage was varied; 0.00, 0.14, 0.35, and 0.52 monolayers of
Ru.
electrode surface. However, clear similarities between voltam-
metry and TPD were found in this work, showing that the
desorption is controlled through steps preceding both the
recombination or hydrogen atom ionization. The hydrogen
desorption peak as proton, seen at 0.15 V in CV, and the
desorption peak as a H2 molecule, observed at 300 K in TPD
(Figure 1) are absent on the ruthenium modified surface.
Ruthenium deposition on Pt(110) is reducing the hydrogen
coverage and binding energy of hydrogen both in an electro-
chemical cell and UHV. This contributes to simplification of
the fine structure of the CV curves and TPD spectra (Figure 2,
cf. with Figure 1). There is, therefore, a clear correspondence
between the data obtained in these two dissimilar environments.
On this basis, we may interpret the data from UHV and
electrochemical cell jointly.
The results in this investigation provide a consistent picture
about the effects of ruthenium on hydrogen and CO adsorption
and oxidation on platinum. First, the data show a strong
modification in the electronic structure of platinum when
ruthenium is added. Hydrogen TPD reveals that the strongly
bound state of H2 at 300 K disappears. CV shows that one
hydrogen desorption peak at higher potential (0.16 V) disap-
pears. NMR data reveal that the Ef-LDOS of all the adsorbed
CO decreases with increasing ruthenium coverage. The H2 peak
seen at 0.15 V in CV, and 300 K in TPD (Figure 1) are absent
on the ruthenium modified surface. If some of the platinum
atoms were unaffected by ruthenium, one would expect to see
these peaks on the ruthenium modified surface too, but Figure
2 shows no evidence for that. Therefore, addition of ruthenium
modifies the electronic structure of the entire platinum surface.
Figure 7. (A) Spin-echo mapped 13C EC NMR spectra of chemisorbed
13C labeled CO on Ru modified Pt black nanoparticles in 0.5 M H2-
SO4. Ru coverages correspond to 0.00, 0.14, 0.35, and 0.52 monolayers.
(B) The metal surface Ef-LDOS was calculated from the 13C NMR
chemical shift for the strongly bound CO (open symbols) and the weakly
bound CO (closed symbols). For chemisorbed 13CO, 2π* Ef-LDOS
variation is shown for strongly bound (solid symbols) and weakly bound
(open symbols) forms as a function of the Ru coverage.
Surprisingly, the changes in the electronic structure of
platinum have only a mild effect on the binding energy of CO.
TPD shows that the desorption temperatures of CO on Pt(110)/
Ru are about 400 and 480 K, compared to 415 and 515 K on a
clean Pt(110) surface. This corresponds to a change of about 2
kcal/mol in the binding energy of CO. The NMR peak of
strongly bound CO shifts by only 10-20 ppm. By comparison,
hydrogen TPD shows a 100 K change in desorption temperature.
These observations suggest that the binding of strongly bound
CO is only marginally affected by ruthenium, even though the
hydrogen TPD and CV data prove that the whole surface is
water is much more active on a ruthenium modified Pt(110)
surface than on clean platinum.
Discussion: Adsorption
Conceptually, electrochemical and thermal activation causing
electrooxidation of chemisorbed species may be basically
different. The difference may arise because two hydrogen atoms
need to recombine to a H2 molecule before desorbing to UHV,
whereas atomic hydrogen is ionized to proton to leave the

Platinum/Ruthenium Fuel Cell Catalysts
J. Phys. Chem. B, Vol. 106, No. 37, 2002 9587
Figure 10. (A) Comparison of CO desorption kinetics at Pt Black
and Ru decorated Pt nanoparticle catalysts using radioactive labeling
and the slow-scan, positive-going electrode polarization. Catalyst
loading is 0.4 mg cm-2. (B) Comparison of CO desorption kinetics
(also to the clean electrolyte) at Pt Black and Pt/Ru decorated
nanoparticle catalyst using radioactive labeling and potential step
experiment, from 0.1 to 0.43 V. (The arrow indicates the potential step.)
Under some conditions, the CO might also desorb from the
platinum surface, especially at low potentials and high temper-
ature.26
Figure 10A compares radioactive labeling data of 14CO
oxidation on platinum black and the platinum/ruthenium elec-
trode obtained under slow-scan polarization conditions. CO was
formed on the catalyst surface from 0.01 M methanol (14C
labeled) in 0.1 M H2SO4 at 0.1 V vs SHE. Once a constant CO
coverage, corresponding to surface concentration of 7.5 × 1014
molecules cm-2 was obtained, methanol was washed off from
the solution, the surface concentration was re-checked, and the
positive-going scan of 0.1 mV/s was applied. The data show
that within most of the hydrogen region, there is no CO
oxidation from the surface, and neither is there any significant
change in the coverage between pure Pt and Ru decorated Pt.
The CO oxidation does not begin until 0.3 V on platinum black,
as seen by the drop in count rates (Figure 10A). However, on
platinum/ruthenium the CO oxidation is initiated at 0.1 V more
negative potential showing that there is about a 0.1 V reduction
in the energy to oxidize CO from the Pt/Ru surface. The Γ vs
E branch is negatively shifted with the deposited Ru coverage
up to the maximum value of 0.2 V, confirming data in the
previous study.21 Similar separations were observed on the
Pt(111) electrode decorated by ruthenium regardless of whether
the CO adlayer was created from methanol or a gaseous CO
precursor.
Figure 9. (A) 18 and 20 amu TPD spectra taken by adsorbing about
0.5 × 1015 molecules/cm2 of 18O then adding about 1 × 1015 molecules/
cm2 of H216O onto Pt(110) before flashing in a background containing
1 × 10-10 Torr of hydrogen with a rate of 15 K/s. (B) Repeat the same
experiment as in A on Pt(110)/Ru
being affected by ruthenium. The TPD experiments in Figure
6 confirm that the initial CO adsorption is unaffected by the
presence of Ru, suggesting a similarity in sticking probability
on both Pt and Pt/Ru surfaces at the beginning of the process.
As the surface population of CO grows, the adsorption rate
decreases.
Discussion: Kinetics Measurements
Next, we wish to focus on the role of ruthenium in promoting
CO oxidation from the platinum surface. Recall that previous
investigators1-7,9-12,25 have found that platinum-ruthenium
catalysts are much more CO tolerant than pure platinum
catalysts. It is thought that the main mode of CO oxidation is
a simple process:
Further support for the strong effect of surface ruthenium on
the CO oxidation process is provided by the data in Figure 10B.
M-CO + M-OH f CO2 + 2M + e- + H+
(6)

9588 J. Phys. Chem. B, Vol. 106, No. 37, 2002
Lu et al.
The beginning of the experiment was the same as reported
above, and the constant surface concentration of CO was
measured at 0.1V vs SHE in the clean supporting acid
electrolyte. Next, instead of applying the slow-scan polarization,
the potential was stepped to 0.43 V. It is shown in Figure 10B
that the oxidative CO desorption is much faster on Pt/Ru than
on Pt, and that the process is incomplete on pure platinum.
Clearly, there exists a significant difference in the stability of
CO on these two surfaces. Thus, the data in Figure 10 also show
that platinum/ruthenium is much more CO tolerant than pure
platinum.
seem to be larger than the ligand effects. According to our data,
of the 4-6 kcal/mol (170-260 meV) reduction in the potential
for CO oxidation, only about 1 kcal/mol (40 meV) is associated
with the ligand effect, whereas 3-5 kcal/mol (130 to 200 meV)
are associated with the bifunctional mechanism. Thus, the effect
of the bifunctional mechanism is about 4 times larger than the
ligand effect.
Conclusions
This investigation explored the effect of ruthenium on the
CO tolerance of platinum fuel cell anodes. We find that
ruthenium has a major effect on the electronic structure of
platinum. The Ef-LDOS of 13CO at Pt sites near Ru decreases
substantially while the binding energy of H2 shows a significant
drop. CO oxidation occurs at a potential 200 mV lower on
platinum/ruthenium than on platinum. Surprisingly, ruthenium
only has a small effect on the binding of CO to platinum atoms
that are away from Ru. The initial sticking probabilities are
identical for platinum and platinum/ruthenium surfaces. TPD
indicates that the binding energy of the strongly bound form of
CO is changed by only 2 kcal/mol and the NMR peaks shift by
only 10-20 ppm.13
Analysis of our TPD data indicates that, of the 200mV
reduction, about 40 meV is associated with the ligand effect
and 130 to 200 meV to the bifunctional mechanism. Thus, the
bifunctional mechanism is approximately 4 times larger than
the ligand effect for the enhancement of CO oxidation from
the platinum catalyst.
Finally, we can use the TPD data to quantify the contributions
of the ligand effect and the bifunctional mechanism on the
enhancement of reaction 6. TPD results yield only about 2 kcal/
mol reduction in the binding energy of strongly bound CO and
13C NMR shows a maximum of 20 ppm peak shift for strongly
bound CO. According to the Polanyi relationship27 a 2 kcal
change in the binding energy of the CO on the rate should only
produce about a 40 mV reduction in the activation barrier for
reaction 6.17 That is insufficient to explain the 200 mV reduction
in potential observed in Figure 10. Therefore, the ligand effect
alone is insufficient to explain the reduction in the potential
for CO oxidation. However, although the ligand effect is overall
small, there is clear evidence from XAS28 and NMR13 data that
there is a sharing of electrons between platinum and ruthenium.
This sharing of electrons reduces the 2π* Ef-LDOS on the C-O
bond in the CO chemisorbed state on Pt/Ru electrodes.
As explained earlier, the data in Figure 9 show that an
additional role of ruthenium is to activate water. One can make
a quantitative estimate of this effect by comparing the activation
barrier for OH recombination on a clean surface to that on a
ruthenium covered one. Assuming a preexponential factor of
1013/sec, we have calculated an activation barrier of 12 kcal/
mol for OH recombination on the clean surface. It is harder to
get an accurate value on a ruthenium covered surface. The OH
recombination peak is at 160 K on the ruthenium covered
surface. Water itself desorbs at 160 K, so the 160 K peak must
be desorption limited but not reaction limited. Therefore, one
cannot use TPD to accurately measure the activation barrier
for OH recombination on the ruthenium covered surface. Still,
the barrier for OH recombination on the ruthenium covered
surface must be less than the activation barrier for desorption
of water, 9 kcal/mol, because water desorption is rate determin-
ing. Further, the barrier for OH recombination on the ruthenium
covered surface must be at least 7 kcal/mol or else we would
have observed water formation during dosing at 100 K. Thus,
the presence of ruthenium has reduced the barrier for OH
recombination from 12 kcal/mol to 7-9 kcal/mol.17 This result
shows that the bifunctional mechanism is active on the
ruthenium covered platinum surface producing a 3-5 kcal/mol
reduction in the activation barrier for hydroxyl recombination.
According to the Polanyi relationship, a 3-5 kcal/mol reduction
in the activation barrier for hydroxyl recombination should
translate into a 130 to 220 mV reduction in the activation barrier
to oxidize CO.
Acknowledgment. This material is based upon work sup-
ported by the Department of Energy under Grant No. DEGF-
02-99ER14993. P. Waszczuk acknowledges support by the
Department of Energy Grant No. DOE LANL 30167010139.
Any opinions, findings, and conclusions or recommendations
expressed in this publication are those of the authors and do
not necessarily reflect the views of The Department of Energy.
References and Notes
(1) Hamnett, A., Mechanism of Methanol Electro-Oxidation, in
Interfacial Electrochemistry: Theory, Experiment, and Applications; Wieck-
owski, A. Ed.; Marcel Dekker: New York, 1999; 843-884.
(2) Watanabe, M.; Motoo, S. J. Electroanal. Chem. 1975, 60,
283.
(3) Ticanelli, E.; Beery, J. G.; Paffett, M. T.; Gottesfeld, S. J.
Electroanal. Chem. 1989, 258, 61.
(4) Tremiliosi, G.; Kim, H.; Chrzanowski, W.; Wieckowski, A.;
Grzybowska, B.; Kulesza, P. J. Electroanal. Chem. 1999, 467, 143.
(5) Davies, J. C.; Hayden, B. E.; Pegg, D. J. Surf. Sci. 2000, 467, 118.
(6) Goodenough, J. B.; Hamnett, A.; Manoharan, R.; Kennedy, B. J.;
Weeks, S. A. J. Electroanal. Chem. 1988, 240, 133.
(7) Iwasita, T.; Nart, F. C.; Vielstich, W. Ber. Bunsenegs. Phys. Chem.
1990, 94, 1030.
(8) Krausa, M.; Vielstich, W. J. Electroanal. Chem. 1994, 379, 307.
(9) Frelink, T.; Visscher, W.; Vanveen, J. A. R. Surf. Sci. 1995, 335,
353.
(10) Frelink, T., W. Visscher, and J. A. R. Vanveen, Langmuir, 1996,
12, 3702.
(11) Hadzi-Jordanov, S.; Angerstein-Kozlowska, H.; Vukovic, M.;
Conway, B. E.; J. Phys. Chem. 1977, 81, 2271.
(12) Gasteiger, H. A.; Markovic, N.; Ross, P. N.; Cairns, E. J. J. Phys.
Chem. 1993, 97, 12 020.
(13) Tong, Y. Y.; Kim, H. S.; Babu, P. K.; Waszczuk, P.; Wieckowski,
A.; Oldfield, E. J. Am. Chem. Soc. 2002, 124, 468.
(14) Wang, J.; Masel, R. I. J. Am. Chem. Soc. 199, 113, 5850.
(15) Wang, J.; Masel, R. I. Surf. Sci. 1991, 243, 199.
(16) Wang, J.; Masel, R. I. J. Catal. 1991, 126, 519.
(17) Lu, C.; Masel, R. I. J. Phys. Chem. B 2001, 105, 9793.
(18) Chrzanowski, W.; Wieckowski, A. Langmuir 1997,13, 5974.
(19) Crown, A.; Wieckowski, A. Phys. Chem. Chem. Phys. 2001, 3,
3290.
Notice that the total reduction in the potential for CO
oxidation inferred from TPD (170 to 260 mV) agree quite well
with the 200 mV reduction observed in ref 22 and in this study
at high Ru coverage. Consequently, it does appear that the TPD
measurements are reproducing the changes seen in the electro-
chemical experiment. From the TPD data, we can conclude that
although both the “ligand model” and the “bifunctional mech-
anism” seem to be active on our sample the bifunctional effects

Platinum/Ruthenium Fuel Cell Catalysts
J. Phys. Chem. B, Vol. 106, No. 37, 2002 9589
(20) Tong, Y. Y.; Rice, C.; Godbout, N.; Wieckowski, A.; Oldfield, E.
J. Am. Chem. Soc. 1999, 121, 2996.
(25) Koper, M. T. M.; Lukkien, J. J.; Jansen, A. P. J.; van Santen, R.
A.; J. Phys. Chem. B 1999, 103, 5522.
(21) Waszczuk, P.; Wieckowski, A.; Zelenay, P.; Gottesfeld, S.;
Coutanceau, C.; Leger, J. M.; Lamy, C.; J. Electroanal. Chem. 2001, 511,
55.
(22) Waszczuk, P.; Solla-Gullon, J.; Kim, H. S.; Tong, Y. Y.; Montiel,
V.; Aldaz, A.; Wieckowski, A. J. Catal. 2001, 203, 1.
(23) Bendorf, C.; Nobl, C.; Madey, T. E. Surf. Sci. 1984, 138, 292.
(24) Capitano, A. T.; Gabelnick, A. M.; Gland, J. L. Surf. Sci. 1999,
419, 104.
(26) Pedersen, M. O.; Helveg, S.; Ruban, A.; Stensgaard, I.; Laegsgaard,
E.; Norksov, J. K.; Besenbacher, F. Surf. Sci. 1999, 426, 395.
(27) Masel, R. I. Chemical Kinetics and Catalysis; Wiley: New York,
2001.
(28) McBreen, J.; Mukerjee, S. In Interfacial Electrochemistry: Theory,
Experiment, and Applications; Wieckowski, A., Ed.; Marcel Dekker: New
York, 1999, p 895.