Carbon-supported Pt-Sn electrocatalysts for the anodic oxidation of H 2, CO, and H2/CO mixtures.: Part II: The structure-activity relationship

Article abstract of DOI:10.1016/j.jcat.2005.03.022

The catalytic activity of three different carbon-supported Pt-Sn catalysts for the anodic oxidation of hydrogen, carbon monoxide, and H₂/CO mixtures is correlated with the bimetallic microstructure established in Part I [V. Radmilovic, T.J. Richardson, S.J. Chen, P.N. Ross, Jr., J. Catal. 232 (2005) 199-209] of this study. These catalysts differ primarily by having differing amounts of Pt, Pt₃Sn, PtSn, and SnO₂ phase nanoparticles distributed on the carbon support. Further surface chemical characterization of the Pt₃Sn nanoparticles is provided by comparison of in situ vibrational spectra of CO adsorbed on the nanoparticles with CO adsorbed on Pt₃Sn(hkl) surfaces [V. Stamenkovic, M. Arenz, B.B. Blizanac, K.J.J. Mayrhofer, P.N. Ross, N.M. Markovic, Surf. Sci. 576 (2005) 145-157; V.R. Stamenkovic, M. Arenz, C.A, Lucas, M.E. Gallagher, P.N. Ross, N.M. Markovic, J. Am. Chem. Soc. 125 (2003) 2736-2745]. Qualitative measures of CO oxidation activity were also obtained from the FTIR spectra for the electrode potential at which CO₂ is first detected (appearance potential). Quantitative kinetic measurements using the carbon-supported catalysts were obtained with the thin-film rotating disk electrode (TF-RDE) method. The CO_ad stripping voltammetry clearly indicated that the 500N sample exhibits the lowest degree of alloying, showing a superposition of Pt₃Sn and Pt features, whereas the catalyst consisting entirely of stoichiometric Pt ₃Sn cubic phase (E270 sample) exhibits no features characteristic of pure Pt. None of the three carbon-supported Pt-Sn catalysts had the splitting of the CO stretching band of characteristic of CO_ad on Pt ₃Sn(111), the surface with the highest specific activity (turnover number) for CO and H₂/CO oxidation. All three catalysts had very high levels of activity for H₂ oxidation. The mass activity for CO and H₂/CO oxidation was proportional to the amount of Pt₃Sn phase in the catalyst. There is no clear evidence that either the PtSn phase or Pt/SnO₂ clusters contribute any significant activity for H ₂/CO or CO oxidation in these catalysts.

Full text of DOI:10.1016/j.jcat.2005.03.022

Journal of Catalysis 232 (2005) 402–410
www.elsevier.com/locate/jcat
Carbon-supported Pt–Sn electrocatalysts for the anodic oxidation
of H , CO, and H /CO mixtures.
2
2
Part II: The structure–activity relationship
1
∗
M. Arenz , V. Stamenkovic, B.B. Blizanac, K.J. Mayrhofer, N.M. Markovic, P.N. Ross
Materials Science Division, Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720, USA
Received 29 November 2004; revised 9 March 2005; accepted 15 March 2005
Available online 10 May 2005
Abstract
The catalytic activity of three different carbon-supported Pt–Sn catalysts for the anodic oxidation of hydrogen, carbon monoxide, and
H /CO mixtures is correlated with the bimetallic microstructure established in Part I [V. Radmilovic, T.J. Richardson, S.J. Chen, P.N. Ross,
2
Jr., J. Catal. 232 (2005) 199–209] of this study. These catalysts differ primarily by having differing amounts of Pt, Pt Sn, PtSn, and SnO
3
2
phase nanoparticles distributed on the carbon support. Further surface chemical characterization of the Pt Sn nanoparticles is provided by
3
comparison of in situ vibrational spectra of CO adsorbed on the nanoparticles with CO adsorbed on Pt Sn(hkl) surfaces [V. Stamenkovic, M.
3
Arenz, B.B. Blizanac, K.J.J. Mayrhofer, P.N. Ross, N.M. Markovic, Surf. Sci. 576 (2005) 145–157; V.R. Stamenkovic, M. Arenz, C.A, Lucas,
M.E. Gallagher, P.N. Ross, N.M. Markovic, J. Am. Chem. Soc. 125 (2003) 2736–2745]. Qualitative measures of CO oxidation activity were
also obtained from the FTIR spectra for the electrode potential at which CO is first detected (appearance potential). Quantitative kinetic
2
measurements using the carbon-supported catalysts were obtained with the thin-film rotating disk electrode (TF-RDE) method. The CO
ad
stripping voltammetry clearly indicated that the 500N sample exhibits the lowest degree of alloying, showing a superposition of Pt Sn and Pt
3
features, whereas the catalyst consisting entirely of stoichiometric Pt Sn cubic phase (E270 sample) exhibits no features characteristic of pure
3
Pt. None of the three carbon-supported Pt–Sn catalysts had the splitting of the C=O stretching band of characteristic of CO on Pt Sn(111),
ad
3
the surface with the highest specific activity (turnover number) for CO and H /CO oxidation. All three catalysts had very high levels of
2
activity for H oxidation. The mass activity for CO and H /CO oxidation was proportional to the amount of Pt Sn phase in the catalyst.
2
2
3
There is no clear evidence that either the PtSn phase or Pt/SnO clusters contribute any significant activity for H /CO or CO oxidation in
2
2
these catalysts.
2005 Elsevier Inc. All rights reserved.

Keywords: Pt–Sn alloys; CO oxidation; Hydrogen/CO mixtures; IR spectroscopy
1
. Introduction
lic systems (see, for example, Refs. [3–10]). The major-
ity of previous studies of Pt alloy catalysts in fuel cells
employed an unsupported polycrystalline form. Our previ-
ous studies with Pt3Sn(hkl) single-crystal electrodes have
shown that the electrochemical oxidation of carbon monox-
ide in acid electrolyte is a structure-sensitive reaction, in
which the (111) surface (25 at% Sn) has the highest activity
Bimetallic alloys are widely used as electrode materials
in fuel cells, especially for the electrooxidation of carbon
monoxide (CO) and/or hydrogen/carbon monoxide (H2/CO)
mixtures. Pt–Sn alloys are one of the most active bimetal-
[
3,7]. There have been surprisingly few studies of fuel cell
*
Corresponding author.
E-mail address: pnross@lbl.gov (P.N. Ross).
Present address: Department of Physical Chemistry I, Technical Uni-
reactions with carbon-supported Pt–Sn catalysts. Schmidt
et al. [8] used a commercially available carbon-supported
Pt–Sn (75 at% Pt) catalyst for the electrooxidation of H2/CO
1
versity of Munich.
0021-9517/$ – see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2005.03.022

M. Arenz et al. / Journal of Catalysis 232 (2005) 402–410
403
mixtures and found very similar activity to the Pt3Sn(111)
and/or (110) surfaces. In contrast to these studies of alloyed
Pt and Sn, there are also reports of a promotional effect for
H2/CO electrooxidation on nonalloyed SnOx phase(s) im-
pregnated into carbon-supported Pt [5]. Thus, one should ex-
pect very preparation-sensitive activity for carbon-supported
Pt–Sn catalysts in this reaction, and to understand this sensi-
tivity it is essential to make a detailed characterization of all
phases present, as we have done in Part I [1].
In the present work, we examine the catalytic activity
and the relationship of activity to the microstructure of the
catalysts. The mass activity (rate per gram of Pt) for the elec-
trooxidation of H2, CO, and H2/CO mixtures is established
by the thin-film RDE method. The specific activity (activity
per unit area of alloy) or turnover number (activity per active
site) is difficult to determine in the Pt–Sn alloy system, how-
ever. The usual methods of surface area determination by
integration of the charge for oxidation of either adsorbed hy-
drogen or adsorbed CO cannot be applied, because of strong
electronic modification of these states by the Sn [2,3,8]. We
present an alternative approach to determining the active site
density from the fine features in the vibrational spectra of
CO adsorbed on these high-surface-area catalysts and cor-
relating these features with those seen on the Pt3Sn(hkl)
single-crystal surfaces. We also examine the question of the
promotional role of Sn/SnO2 atoms, that is, whether Sn has
to be present in metallic form alloyed to Pt, and whether the
promotional role is a bifunctional or a ligand effect.
Isotemp Circulator) maintained the temperature of the elec-
◦
trolyte within ± 0.5 C. The reference electrode was a satu-
rated calomel electrode (SCE) separated by a closed bridge
from the reference compartment. All potentials, however,
were referenced to the potential of the reversible hydrogen
electrode (RHE) at the same temperature (calibrated from
the hydrogen oxidation reaction) in the same electrolyte.
The preparation of the catalyst layer on the surface of the
glassy carbon rotating disk electrode (RDE) has been de-
scribed previously [11]. In short, the catalyst was dispersed
ultrasonically in diluted Nafion solution (Aldrich), and 20 µl
of the suspension was pipetted onto a glassy carbon substrate
2
(0.283 cm geometrical surface area), leading to a Pt loading
2
of 14 µgPt/cm . After drying in an Ar atmosphere, the cata-
lyst film is attached to the substrate by a thin (< 0.2 µm) film
of recast Nafion. As shown previously, the electrochemical
response and the mass transport of the reactants is not influ-
enced by the Nafion thin film [12]. The thus prepared surface
was then transferred to the electrochemical cell protected
by a drop of ultrapure water and immersed under potential
control at 0.05 V in argon-saturated solution, and a cyclic
voltammogram was recorded. The upper potential limit was
set at 0.5 V versus RHE to avoid Sn dissolution. CO strip-
ping curves were recorded after the surface was saturated
with CO for at least 5 min at a potential of 0.05 V and sub-
sequently all CO was removed from the solution by purging
with argon for 30 min. Stripping curves are recorded with a
scan rate of 10 mV/s, and the upper potential limit was set at
1
.0 V to guarantee the complete oxidation of the CO adlayer.
In order to measure the continuous electrooxidation of pure
H2 or CO (COb) gas with the RDE, we equilibrated the elec-
trolyte for 5 min with the respective gas while holding the
potential at 0.05 V. For the H2/CO mixture, a longer 25-min
hold was used to reach the CO equilibrium coverage. Subse-
quently the potential was scanned positively at a sweep rate
of 1 mV/s while the electrode was rotated at 2500 rpm. We
expressed the catalytic activity is expressed as mass activ-
ity (i.e., mol/gPt) by dividing the current by nF (F : Faraday
constant; n: number of electrons involved in the reaction;
n = 2 for either H2 or CO oxidation).
2
. Experimental
2
.1. High-surface-area catalyst preparation
The Pt–Sn high-surface-area catalysts studied here (sam-
ples 500N and 900N) were supplied by the Johnson Matthey
Technology Center (Reading, UK), and the third one (de-
noted E270) by was supplied by E-Tek (Somerset, NJ, USA).
The 500N and 900N samples were used with an acetylene
black (Shawinigan) carbon support with a Pt/Sn stoichiome-
try of 1.23:1 and a metal loading of 23 wt%. They were made
from the same Pt–SnOx precursor used previously [5], but
differ in the thermal treatment they received to decompose
the precursor and promote alloying of the two metals (500
2
.3. Infrared spectroscopy
The detailed catalyst preparation for FTIR spectroscopy
of the HSAC is described in a separate paper [13]. In
short, the catalyst was dispersed ultrasonically in water, and
about 20 µl of the suspension was pipetted onto a heated
◦
or 900 C, respectively, in an atmosphere of N2). The E-Tek
sample, prepared by a proprietary method, was used with
Vulcan XC-72 (Cabot) carbon black, at a Pt/Sn stoichiome-
try of 3:1, and a metal loading of 20 wt%, with reduction in
◦
2
(
∼ 140 C) gold substrate (0.785 cm geometrical surface
area), cooled in an argon stream, and rinsed carefully with
ultrapure water to remove loosely bound catalyst particles.
No further attachment to the substrate was necessary; that
is, there was no Nafion film. For the in situ FTIR measure-
ments, a Nicolet Nexus 670 spectrometer was used, which
◦
H2 at 270 C as the final step in preparation.
2
.2. Electrochemical measurements
The electrochemical measurements were conducted in
was equipped with a liquid-N cooled MCT detector. All
2
a thermostatted standard three-compartment electrochem-
ical cell. A circulating constant-temperature bath (Fisher
IR measurements were performed in a spectroelectrochemi-
cal glass cell designed for the external reflection mode. The

404
M. Arenz et al. / Journal of Catalysis 232 (2005) 402–410
cell is coupled at its bottom with a CaF2 prism beveled at
◦
6
0 from the prism base. The experimental procedure for
the IR measurements followed that for the measurement of
the polarization curves. Before each experiment the solution
was saturated with argon. The electrode was immersed in
the electrolyte at a potential of 0.05 V, then CO was intro-
duced into the solution for at least 5 min, and subsequently
the sample was pressed against the prism. Starting at a poten-
tial of 0.05 V, the potential was scanned at a rate of 5 mV/s
in the positive direction while spectra consisting of a sin-
gle scan were continually recorded. The recording time was
thus reduced to ca. 1 s per spectrum. The resolution of the
−1
spectra was 4 cm , and p-polarized light was used. Ab-
sorbance spectra were calculated as the ratio −log(R/R0),
where R and R0 are the reflectance values corresponding to
the sample and reference spectra, respectively. To obtain the
COad spectra, the reference spectrum was recorded at 0.80 V,
where COad is completely oxidized. To follow the produc-
tion of CO2, we referenced the spectrum recorded at 0.05 V
to the spectra recorded at the potentials indicated. The po-
tential in the spectroelectrochemical cell was controlled by a
reversible hydrogen electrode (RHE).
3
. Results and discussion
Before introducing our results for the catalytic activity of
the Pt–Sn catalysts, we briefly summarize the basic charac-
terization of the different catalysts presented in the first part
of the paper [1].
3
.1. High-surface-area catalyst characterization
Three different carbon-supported Pt–Sn catalysts were
available. The composition, size, distribution, and morphol-
ogy of the phases present in three different carbon-supported
Pt–Sn catalysts were analyzed by high-resolution transmis-
sion electron microscopy (HRTEM), microchemical analy-
sis by X-ray emission spectroscopy (EDS), and X-ray dif-
fraction (XRD) [1]. The catalysts had nominal Pt/Sn atomic
ratios of 1:1 (500N and 900N samples) and 3:1 (E270
sample). The results showed that the 1:1 catalyst sample
Fig. 1. Typical TEM micrographs of the carbon-supported PtSn high surface
area catalysts with the corresponding histograms of the particle size distrib-
ution: E270, average particle size about 3–4 nm; 500N average particle size
about 4 nm, and 900N average particle size about 6 nm.
of the particle size distribution of the catalysts, see Fig. 1.
◦
The Pt Sn-type particles in the 500N and 900N are generally
3
heat-treated at 500 C (500N) contained a Pt-rich Pt3Sn-
highly faceted, with {111} and {200} facets, whereas those
type cubic phase (a0 = 0.3965 nm) and tetragonal SnO2.
◦
in the E270 catalyst generally had a smooth spheroidal out-
line [1]. Whereas the Pt3Sn particles in the E270 and 900N
catalysts were equiaxed, in the 500N catalyst the Pt3Sn par-
ticles had uniquely elongated (e.g., ellipsoidal) shapes. For
more details see Ref. [1].
The sample from the same precursors heat-treated at 900 C
(
900N) consisted of stoichiometric hexagonal PtSn, a nearly
stoichiometric Pt3Sn-type cubic phase (a0 = 0.3988 nm),
and tetragonal SnO2. Neither pure tin nor pure platinum
particles were detected in any of the catalysts. The 3:1 cat-
◦
alyst sample reduced in H2 at 270 C (E270) was com-
3
.2. Catalytic activity for CO electrooxidation
posed entirely of the stoichiometric Pt3Sn cubic phase (a0 =
0
.3998 nm). The Pt3Sn-type particles are predominantly sin-
gle nanocrystals with mean diameters of ∼ 3–4 nm (E270),
The three Pt–Sn catalysts have significantly different Pt–
∼
4 nm (500N), and ∼ 5 nm (900N), and the PtSn particles
Sn coordinations, and thus one expects significantly different
catalytic properties if this coordination is important. In the
500N catalyst, the Pt and SnO2 precursors have only par-
in the latter sample were larger (taken together, this leads
to an average particle size of ∼ 6 nm). For the histograms

M. Arenz et al. / Journal of Catalysis 232 (2005) 402–410
405
one Pt-like, consistent with the microstructure of Pt3Sn-like
phases and Pt-like phases. It would appear from this result
that the clusters of Pt/SnO2 do not have synergistic proper-
ties for COad oxidation; that is, the Pt behaves like Pt. In
agreement with previous results for Pt3Sn bulk alloy elec-
trodes [7,14], although COad oxidation starts at very low
potentials, the CO adlayer can be only partially removed
at low potentials (< 0.5 V), even if several cycles are ap-
plied (not shown). To achieve complete COad stripping, that
is, complete restoration of the charge for hydrogen adsorp-
tion/desorption pseudocapacitance, the electrode potential
has to be at least 0.9 V, as in Fig. 2. The excursion to this very
positive potential, however, causes dissolution of Sn from
the surface and thus alters the surface irreversibly, as man-
ifested in a change in the cyclic voltammogram (for more
details see [7,14]). Thus determination of the specific ac-
tivity for Pt–Sn is not possible. For pure Pt the unit area
of catalyst, necessary for the determination of the specific
activity, can be established either by the Hupd (under po-
tential deposited hydrogen) or the CO stripping charge (the
turnover frequency (TOF)) is calculated by TOF = i/nFN,
where i is the measured current density per unit area, N is
the density of active sites, and n is the number of electrons
involved in the reaction (for more details see [15]). However,
for Pt–Sn catalysts neither method is suitable for establish-
ing either the number of active sites or the metallic area of
the catalyst, because on one hand in CO stripping, processes
such as Sn dissolution occur parallel to CO oxidation. On the
other hand, Hupd features are clearly visible only at the elec-
trodes scanned up to 1 V, that is, after there has been some
Sn dissolution (see below). Therefore, for our purposes here,
the CO stripping curves are used only as a “fingerprint” to
detect Pt-like characteristics in the catalysts (note that be-
fore the stripping curves were recorded the positive potential
limit was 0.5 V).
From Fig. 2 it is obvious that the 500N sample exhibits
the most pronounced Pt-like features. By comparison, a sam-
ple from the same precursor, treated at a higher temperature
Fig. 2. Comparison of CO stripping curves of a saturated CO adlayer in
◦
0
.5 M H SO solution purged with argon for 30 min at 0.05 V; T = 60 C,
2
4
scan rate 10 mV/s; dashed curves show second sweep; Pt loading of all cat-
2
alyst is 14 µg/cm : E270; 500N; 900N and pure Pt HSAC for comparison.
(
900N), shows a clearly reduced Pt-like COad oxidation peak
indicating more complete alloying, consistent with the mi-
crostructural analysis [1]. The COad stripping curve for the
900N is the most complex and appears to be composed of
multiple potential regions. Since this is the only catalyst with
some stoichiometric PtSn phase, it is possible that one of
these multiple regions corresponds to a unique oxidation po-
tential for COad on PtSn nanoparticles.
tially reacted to form a relatively small amount of Pt3Sn-like
alloy. All remaining SnO2 particles appear to be closely
associated with Pt-like particles, and it would be interest-
ing to determine whether a catalytically synergistic effect
comparable to that in Pt3Sn alloy exists for these Pt/SnO2
clusters. One of the consequences of the incomplete alloy-
ing process can be seen in the anodic stripping curves for
COad in comparison to that for a carbon-supported pure Pt
catalyst (HSAC), as shown in Fig. 2. For pure Pt HSAC, the
COad is oxidized in a relatively sharp stripping peak cen-
tered at ∼ 0.7 V (see Fig. 2). By comparison, on the E270
catalyst, which is composed entirely of Pt3Sn nanoparticles,
COad is oxidized over a broad potential region beginning
at potentials as low as 0.2 V with a peak at ca. 0.5 V. The
COad oxidation on the 500N catalyst appears to occur in
The polarization curves for H , H /2% CO, and CO ox-
2
2
idation recorded in the TF-RDE configuration are summa-
rized in Fig. 3. All measurements are performed in 0.5 M
◦
H SO solution at an electrolyte temperature of 60 C; the
2
4
catalytic activity is expressed in mass activity, that is, moles
reactant (hydrogen and CO) per unit mass of platinum. The
polarization curves were recorded at a scan rate of 1 mV/s,
at 2500 rpm, whereas the cyclic voltammograms (dashed
curves) were recorded without rotation at 50 mV/s in argon-
purged solution. The positive potential was limited to 0.5 V
(
at least) two distinct potential regions, one Pt3Sn-like and

406
M. Arenz et al. / Journal of Catalysis 232 (2005) 402–410
oxidation reaction, pronounced differences are observed for
continuous CO oxidation and the continuous oxidation of a
H2/2% CO mixture. As in the COad stripping curves, the on-
set of bulk CO oxidation is observed at approximately 0.2 V
on all three catalysts, and consequently at this potential the
oxidation of hydrogen starts in the case of the H2/CO mix-
ture. As we have discussed in detail in studies of Pt(hkl)
single crystals [16,17], in H2/CO mixtures H2 acts like a
“
current multiplier” for the CO oxidation current; that is,
if a Pt–COad site is freed because of CO oxidation, then
hydrogen oxidation starts at this site at a much higher rate
(
current). Consequently, the onset of CO oxidation is easier
to detect in H2/CO mixtures. For the Pt–Sn catalysts the or-
der for mass activity is E270 > 500N > 900N. At 0.3 V, the
CO oxidation rate is about two times higher for 500N versus
9
00N, and about two times higher for E270 versus 500N.
Note that with a pure Pt HSAC, there is no measurable cur-
rent for CO or H2/CO oxidation at any potential below 0.5 V.
In our study of the Pt3Sn(hkl) surfaces, the two surfaces
with 25 at% Sn, annealed Pt3Sn(111), and lightly sputtered
Pt3Sn(110), had superior activity compared with the sample
with 50 at% Sn on the surface (annealed Pt3Sn(110)) [2,3].
Therefore the low catalytic activity of the 900N sample is
most likely due to the higher content surface concentration
of Sn as a consequence of the high annealing temperature
◦
(
900 C). Our analysis in Part I [1] indeed showed that this
sample (900N) consists in part of stoichiometric PtSn parti-
cles. Furthermore, it may well be that even the stoichiometric
Pt3Sn particles have a 50% Sn surface concentration, de-
pending on the detailed surface structure of these particles.
An observation supporting this supposition is the fact that in
the cyclic voltammogram of the 900N sample (Fig. 3) Hupd
features are almost absent, as is also the case for the annealed
Pt3Sn(110) surface [2]. Note that if the potential is scanned
up to 1 V, Sn dissolves and subsequently Hupd can be seen
(see Fig. 2).
By contrast, the comparably mild preparation tempera-
◦
tures (270 C) for the E270 sample most likely prevent such
a segregation of Sn, and from the above results (Fig. 3) stoi-
chiometric Pt3Sn can be identified as the most active phase.
Another interesting result is the fact that the 500N sample
is quite active, that is, it has a mass activity similar to that
of the E270 sample, although in the CO stripping curves the
former showed the most pronounced “Pt-like” behavior. The
analysis in Part I showed that the 500N sample consists of
a Pt-rich Pt3Sn-type cubic phase and tetragonal SnO2; the
SnO2 particles in the sample appear to be closely associ-
ated with Pt-like particles, that is, unalloyed Pt. A catalyti-
cally synergistic interaction between unalloyed Pt and SnO2
might therefore occur in the 500N catalyst, as implied by
Fig. 3. Polarization curves for different Pt–Sn/C catalysts recorded in 0.5 M
H SO solution saturated with either pure hydrogen, a H /2% CO mix-
2
4
2
◦
ture, or pure CO; curves are recorded at 60 C with 1 mV/s and 2500 rpm;
◦
dashed curves: CV (a.u.) in argon purged solution at 60 C, scan rate
50 mV/s.
to avoid Sn dissolution. As shown in Fig. 3, the polariza-
tion curves of pure hydrogen are very similar, and no differ-
ences in mass activity are clearly seen; that is, all catalysts
are extremely active for H2 oxidation. Note, however, that
the diffusion-limited current density for the 900N sample
is slightly lower than for the other two samples, indicating
that the former sample is not completely evenly distributed
over the glassy carbon support. In contrast to the hydrogen
the reported promotional effect of nonalloyed SnO phase(s)
x
impregnated in carbon-supported Pt [5]. Here the relatively
high mass activity of the 500N sample at very low poten-
tial (i.e., below 0.3 V) further supports that proposition. In
the potential region between the onset of measurable current
with pure CO, near 0.2 V, and 0.3 V where the current be-

M. Arenz et al. / Journal of Catalysis 232 (2005) 402–410
407
comes significant, the current ratio between 500N and the
E270 exceeds the factor of 2 difference in Pt3Sn content in
the two catalysts.
3
.3. FTIR measurements
Before describing the FTIR spectra of CO adsorbed on
Pt–Sn nanoparticles, we briefly summarize our previous
FTIR results of CO adsorption on Pt3Sn(hkl) single-crystal
electrodes [2,3]. As we reviewed in the Introduction, the
electrochemical oxidation of carbon monoxide is a strongly
structure-sensitive reaction, for which the Pt3Sn(111) sur-
face (25 at% Sn) has the highest activity [3,7]. The high
activity on this surface is linked to the formation of a com-
pressed CO adlayer associated with the observation of a
splitting of the C–O stretching frequency of a-top bonded
CO. By comparison, on the lightly sputtered and the an-
nealed Pt3Sn(110) surface (25 and 50 at% Sn, respectively),
no compressed adlayer is formed and only a single peak can
be observed in the spectra. On these surfaces the oxidation of
solution-phase CO compared with Pt3Sn(111) is shifted by
∼
100 and ∼ 200 mV, respectively, to more positive poten-
tials. The C–O stretching frequency, however, is not affected
by the surface concentration of Sn [2]. For more details see
Refs. [2,3].
A series of representative FTIR spectra for Pt–Sn nano-
particles (E270 sample) obtained in CO-saturated 0.5 M
H2SO4 is shown in Fig. 4. The spectra consist of a main band
−1
−1
around 2060 cm and a smaller band around 1880 cm ,
which can be assigned, respectively, to a-top and bridge
bonded CO on the Pt sites [18]. The absorption bands are
broader than spectra obtained on Pt3Sn(hkl) bulk electrodes
Fig. 4. Series of potential-dependent FTIR spectra from the E270 sample
recorded in CO saturated 0.5 M H SO solution; the respective electrode
potential is indicated; for each spectra 50 scans were co-added, resulting in
a recording time of ca. 25 s. The left column of spectra represents evolution
2
4
[2,3] but similar to the respective bands for bands typical for
carbon-supported pure Pt catalyst [13]. The most important
result displayed in Fig. 4 is the fact that no peak splitting
of the a-top CO can be observed on the E270 sample (or
on either of the two other Pt–Sn catalysts). Consequently,
none of the Pt–Sn HSAC catalysts indicates the formation of
the compressed CO adlayer characteristic for the Pt3Sn(111)
electrode [3]. It would appear that the formation of this com-
pressed adlayer requires larger Pt3Sn(111) facets that are
formed on the nanoparticles in these carbon-supported cata-
lysts.
In order to compare the onset of CO oxidation on the
different Pt–Sn catalysts in a controlled way, we monitored
CO2 production while scanning the potential with a fixed
scan rate (5 mV/s), as in the electrochemical measurements,
while continuously recording spectra [19]. The results of
these measurements on the Pt–Sn HSAC are summarized in
Fig. 5, whereas in Fig. 6 results obtained for a pure Pt HSAC
control sample (E-Tek; metal loading 30 wt%) are displayed
separately reduce the complexity of the figures. The insets in
the figures show the (first) spectrum from the respective se-
ries recorded at 0.05 V. By analyzing the potential-dependent
peak area of CO2 in Fig. 5, one can see that CO oxidation
starts at the same potential on all Pt–Sn alloys, that is, un-
of CO , while corresponding CO spectra is given in the right column.
2
der these experimental conditions at ∼ 0.30 V. Interestingly,
on the Pt–Sn HSAC the onset of CO2 production is shifted
by only ∼ 50 mV relative to the pure Pt HSAC (see Fig. 6).
There are two phenomena that cause this shift to be smaller
than it is in practice, that is, in a fuel cell anode. One is the
integrating nature of the measurement of CO2 in solution
in the thin-layer geometry of the IR cell; that is, the rate of
CO2 production (the current) is proportional to the derivative
of the CO2 intensity, and the other is the unsteady, funda-
mentally time-dependent nature of CO oxidation on pure Pt
surfaces, as we have discussed in detail previously [16,17,
20]. The former phenomenon is illustrated by noting that
when the potential is held for a longer time (> 1 min), CO2
formation can be observed on the Pt–Sn nanoparticles at po-
tentials as low as 100 mV. However, in contrast to Pt–Sn
nanoparticles, on pure Pt nanoparticles no CO2 can be ob-
served when the potential is held at 100 mV for any extended
period. The latter phenomenon is seen in Figs. 5 and 6; in
contrast to Pt–Sn, the potential-dependent plot of the CO2
peak area for Pt shows two regions with different slopes. At

408
M. Arenz et al. / Journal of Catalysis 232 (2005) 402–410
Fig. 5. Potential-dependent band position and integrated intensities of the
CO bands of series of potential-dependent FTIR spectra of the different
2
PtSn HSAC; all spectra are recorded in CO saturated 0.5 M H SO solution
2
4
while scanning the potential with 5 mV/s starting at 50 mV; each spectra
Fig. 6. For comparison potential-dependent band position and integrated in-
tensities of the CO bands of a series of potential-dependent FTIR spectra
consists of 1 interferometer scan; the insert shows the first CO spectrum
ad
2
of each catalyst.
of a pure Pt HSAC; all spectra are recorded in CO saturated 0.5 M H SO
2
4
solution while scanning the potential with 5 mV/s starting at 50 mV;
each spectra consists of 1 interferometer scan; the insert shows the first
spectrum.
low potentials for Pt (0.3 < E < 0.65 V) the increase in the
peak area with potential (or time) is much smaller than at
high potentials, that is, at E > 0.65 V. In the low-potential
region (pre-ignition region), the CO oxidation reaction on
bulk Pt surfaces is time-dependent and eventually decreases
pear at potentials above 0.4 V, where sustained oxidation of
CO occurs on the Pt–Sn catalysts. The current interpretation
of the transition to a negative Stark tuning slope, dν/dE, on
Pt surfaces is that above the ignition potential, the steady-
state coverage by COad begins to decrease sharply with in-
creasing potential, and the dipole–dipole coupling is thereby
reduced, (down)shifting the frequency. Because the coverage
decreases rapidly with potential in this region, the down-
shift from reduced dipole coupling exceeds the up-shift from
the Stark effect, leading to a net negative dν/dE. What com-
plicates the detailed analysis of this process on pure Pt sur-
faces is the clustering of the remaining COad molecules as
OHad nucleates on the surface, which maintains stronger
local dipole–dipole coupling than would be expected with
randomly dispersed COad. Pt alloys like Pt–Ru and, here,
Pt–Sn show fundamentally different behavior with respect
to this phenomenon, where the Stark tuning slope becomes
negative concomitantly with the detection of CO2 dissolved
in solution. Most probably this fundamental difference in the
negative Stark tuning between pure Pt surfaces and the Pt al-
loy surfaces is due to the very different way in which OHad
nucleates on the surfaces. In the case of the alloys, preferen-
tially on the admetal sites, which are either randomly (Ru)
or periodically (Sn) distributed in the surface, but in pure Pt
(
to near zero) when the potential is held constant [20]. Only
above 0.7 V, where a sharp increase in the CO2 peak area
is observed with the Pt HSAC catalyst, is there a sustained
high rate of CO oxidation, leading us to characterize the po-
tential of 0.7 V and above as the ignition potential [20]. By
contrast, the CO2 peak area with the Pt–Sn catalysts in Fig. 5
shows a constant increase with the applied potential, and the
oxidation current remains steady if the potential is held at
any potential above ca. 0.4 V.
More details about the promotional effect of Sn are pro-
vided by the potential dependence of the C–O vibrational
frequency. At low potentials, the C–O stretching frequency
is the same on the Pt–Sn and pure Pt HSAC; for exam-
ple, at 50 mV the C–O frequency of a-top CO is about
−
1
2
061 ± 4 cm on all surfaces. By this measure, there is
no unambiguous weakening in the strength of the CO back-
bonding from the intermetallic bond between Pt and Sn (i.e.,
the ligand effect). The same linear shift of the C–O stretch
frequency with the applied potential, the so-called Stark-
shift [21,22], occurs with both Pt–Sn and pure Pt HSAC at
potentials below ca. 0.4 V. However, significant differences
in the dependence of νCO vs E between Pt and Pt–Sn do ap-

M. Arenz et al. / Journal of Catalysis 232 (2005) 402–410
409
surfaces the OHad follows a classical nucleation and growth
pattern. The negative Stark tuning slopes above 0.4 V on all
three Pt–Sn HSAC are remarkably similar. This observation
implies that the active sites in the three catalysts are essen-
tially the same, and that the differences in the overall rates
are due primarily to the variation in the surface concentra-
tion of active sites in the catalysts. Since the E270 catalyst
is the most active, and has only the Pt3Sn phase present, this
would in turn imply that the sites on the Pt3Sn surface are
the most active sites, which is in accordance to the discus-
sion and results related to Pt3Sn(hkl) properties described in
Section 3.2. There is no clear evidence that either the PtSn
phase or Pt/SnO2 clusters contribute any significant activity
for H2/CO or CO oxidation in these catalysts.
simply determined primarily by the formation of reactive
OH species at a much lower potential than on pure Pt sur-
faces.
4. Conclusions
The catalytic activity of three different carbon-supported
Pt–Sn catalysts toward the anodic oxidation of hydrogen,
carbon monoxide, and H2/CO mixtures is analyzed. Two of
the catalysts had nominal Pt/Sn atomic ratios of 1:1 (500N
and 900N samples), whereas one sample had an atomic ra-
tio of 3:1 (E270 sample). The characterization in Part I of
the paper showed that the E270 sample was composed en-
tirely of the stoichiometric Pt3Sn fcc alloy, with the particles
predominantly consisting of single nanocrystals with mean
diameters of ∼ 3–4 nm. By contrast, the 1:1 catalyst samples
possibly contained some pure Pt, a Pt-rich Pt3Sn-type cubic
phase, tetragonal SnO2 (500N), or stoichiometric hexagonal
PtSn, a nearly stoichiometric Pt3Sn-type cubic phase, and
tetragonal SnO2 (900N). The mean size of the Pt3Sn-type
particles in the latter samples is comparable to that in the
E270 sample, ∼ 4 and ∼ 5 nm for 500N and 900N, respec-
tively, whereas the PtSn particles of the 900N sample are
larger. These structural differences lead to different proper-
ties for CO stripping and for the electrooxidation of pure
CO and a H2/CO mixture. In the CO stripping experiments
all Pt–Sn catalysts exhibited CO oxidation at low poten-
tials (0.20 V). The sample that had the least alloying during
preparation (500N) exhibited the most pronounced “Pt-like”
features. For the continuous oxidation of CO and H2/CO
mixtures the E270 sample showed the highest mass activ-
ity, followed by the 500N and the 900N samples.
Vibrational spectroscopy of adsorbed CO by in situ FTIR
showed that none of the Pt–Sn catalysted exhibits the peak
splitting in the C–O stretching of a-top bonded CO unique to
the CO adlayer formed on a highly reactive Pt3Sn(111) sur-
face. Analysis of the potential-dependent CO coverage on
the active Pt sites and the potential-dependent C–O stretch-
ing frequencies shows very similar behavior for all three
catalysts. Based on the band positions, there was no evidence
of a ligand effect, that is, increased back-bonding. The pro-
motional effect of Sn in different Pt–Sn catalysts seems to
be simply determined primarily by the formation of reactive
OH species at a much lower potential than on pure Pt sur-
faces, that is, the bifunctional effect. The most active sites
for CO oxidation (at low potentials) are most probably pairs
of Pt–Sn atoms in the Pt3Sn alloy surface.
3
.4. Reaction mechanism
The bifunctional reaction mechanism of CO oxidation on
Pt alloys is usually described as a Langmuir–Hinshelwood
L-H)-type reaction, where the alloying component, here
(
Sn, preferentially adsorbs oxygenated species (simply rep-
resented as OHad) from dissociation of water. The respective
elementary steps are as follows:
Pt* + COdiss. = Pt–COad,
Sn* + H2O = Sn–OHad + H + e ,
SnO2* + H2O = SnO2–OHad + H + e ,
Pt–COad + Sn–OHad = CO2 + H + e ,
(1)
(2a)
(2b)
(3a)
(3b)
+
−
+
−
+
−
+
−
Pt–COad + SnO2–OHad = CO2 + H + e .
The presence of Sn in the surface thus allows for the con-
tinuous oxidation of the CO at a potential where OHad does
not form on Pt. The relatively large amounts of unalloyed
SnO2 on two of the catalysts studied here suggest that, as
was proposed previously [5], not only Sn atoms but also
SnO2 (or Sn hydro-oxides) adjacent to Pt may provide a bi-
functional effect as well, as shown by reaction schemes (3a)
and (3b). Support for this supposition may also be found in
the gas-phase oxidation of CO at room temperature on Pt
supported on SnO2 [23,24]. We recall that the chemical na-
ture of the oxygenated species is still unknown, that is, it
can be either OHad or oxide. However, the rate of CO oxi-
dation at low potential appears to scale with the amount of
Pt3Sn phase present in the catalyst, implying that reaction
(
3b) does not contribute in any significant way to the total
activity of these Pt–Sn catalysts.
In addition to the bifunctional effect, some theories have
proposed a ligand effect [25,26], whereby the Sn lowers the
adsorption energy of COad on neighboring Pt atoms, thus in-
creasing the rate of COad surface diffusion and increasing
the collision frequency with OHad. However, we could not
find any support for the ligand effect based on our FTIR re-
sults. Although the exact nature of the oxygenated species
associated with the active surface is not known, the promo-
tional effect of Sn in different Pt–Sn catalysts seems to be
Acknowledgments
This work was supported by the Director, Office of Sci-
ence, Office of Basic Energy Sciences, Division of Materials
Sciences, U.S. Department of Energy, under contract DE-
AC03-76SF00098. We thank Dr. David Thompsett from the

410
M. Arenz et al. / Journal of Catalysis 232 (2005) 402–410
Johnson Matthey Technology Center and E-Tek Corporation
for supplying the Pt–Sn catalyst samples. M.A. is grateful
for a Feodor Lynen fellowship from the German Alexander
von Humboldt foundation. K.M. acknowledges the Austrian
BMBWK for a scholarship.
[10] B.E. Hayden, M.E. Rendall, O. South, J. Am. Chem. Soc. 125 (2003)
738–7742.
7
[
[
[
[
11] T.J. Schmidt, H.A. Gasteiger, G.D. Stab, P.M. Urban, D.M. Kolb,
R.J. Behm, J. Electrochem. Soc. 145 (1998) 2354–2358.
12] T.J. Schmidt, H.A. Gasteiger, R.J. Behm, J. Electrochem. Soc. 146
(1999) 1296–1304.
13] V. Stamenkovic, M. Arenz, P.N. Ross Jr., N.M. Markovic, J. Phys.
Chem. B (2004), submitted for publication.
14] K. Wang, H.A. Gasteiger, N.M. Markovic, P.N. Ross, Electrochim.
Acta 41 (1996) 2587–2593.
References
[
[
15] N.M. Markovic, P.N. Ross, CatTech 4 (2000) 110–126.
16] N.M. Markovic, B.N. Grgur, C.A. Lucas, P.N. Ross, J. Phys. Chem.
B 103 (1999) 487–495.
[
[
[
[
[
[
1] V. Radmilovic, T.J. Richardson, S.J. Chen, P.N. Ross Jr., J. Catal. 232
(2005) 199–209.
2] V. Stamenkovic, M. Arenz, B.B. Blizanac, K.J.J. Mayrhofer, P.N.
Ross, N.M. Markovic, Surf. Sci. 576 (2005) 145–157.
3] V.R. Stamenkovic, M. Arenz, C.A. Lucas, M.E. Gallagher, P.N. Ross,
N.M. Markovic, J. Am. Chem. Soc. 125 (2003) 2736–2745.
[17] N.M. Markovic, C.A. Lucas, B.N. Grgur, P.N. Ross, J. Phys. Chem.
B 103 (1999) 9616–9623.
[
[
18] S.C. Chang, M.J. Weaver, Surf. Sci. 238 (1990) 142–162.
19] V. Stamenkovic, K.C. Chou, G.A. Somorjai, P.N. Ross Jr., N.M. Mar-
kovic, J. Phys. Chem. B 109 (2005) 678–680.
4] H.A. Gasteiger, N.M. Markovic, P.N. Ross, J. Phys. Chem. 99 (1995)
8
945–8949.
5] E.M. Crabb, R. Marshall, D. Thompsett, J. Electrochem. Soc. 147
2000) 4440–4447.
6] A.C. Boucher, N. Alonso-Vante, F. Dassenoy, W. Vogel, Langmuir 19
2003) 10885–10891.
[20] N.M. Markovic, J. Ross, Surf. Sci. Rep. 45 (2002) 117–229.
[21] D.M. Bishop, J. Chem. Phys. 98 (1993) 3179–3184.
[22] D.K. Lambert, Electrochim. Acta 41 (1996) 623–630.
[23] S.D. Gardner, G.B. Hoflund, B.T. Upchurch, D.R. Schryer, E.J. Kielin,
J. Schryer, J. Catal. 129 (1991) 114–120.
(
(
[
[
7] H.A. Gasteiger, N.M. Markovic, P.N. Ross, Catal. Lett. 36 (1996) 1–8.
8] T.J. Schmidt, H.A. Gasteiger, R.J. Behm, J. New Mater. Electrochem.
Syst. 2 (1999) 27–32.
[24] D.S. Stark, M.R. Harris, J. Phys. E Sci. Instrum. 16 (1983) 492–496.
[25] T.E. Shubina, M.T.M. Koper, Electrochim. Acta 47 (2002) 3621–3628.
[26] P. Liu, A. Logadottir, J.K. Norskov, Electrochim. Acta 48 (2003)
3731–3742.
[9] S. Tillmann, G. Samjeske, K.A. Friedrich, H. Baltruschat, Elec-
trochim. Acta 49 (2003) 73–83.