XPS and XAFS Pt L2,3-edge studies of dispersed metallic Pt and PtSn clusters on SiO2 obtained by organometallic synthesis: Structural and electronic characteristics

Article abstract of DOI:10.1021/jp030579y

In this work, well-defined silica-supported Pt and PtSn catalysts were prepared by surface organometallic reactions and were characterized by EXAFS-XANES at the Pt L_2,3 edge and XPS of the Pt 4f and Sn 3d levels. These catalysts were tested in the catalytic hydrogenation of cinnamaldehyde in the liquid phase. XPS results showed that in PtSn-OM* (not having butyl groups anchored on the surface), tin was present as Sn(0) adatoms "decorating" the metallic surface, isolating Pt atoms; in the PtSn-OM catalyst (Bu groups remain grafted on the surface), tin was found in the form of Sn(0) and Sn(II, IV) in similar proportions. EXAFS experiments do not provide evidence of the existence of PtSn alloys in any of these systems. In the bimetallic PtSn-BM systems, it was possible to observe by XPS that tin was found in the form of Sn(0) and Sn(II,IV). EXAFS experiments, in this case, allowed us to demonstrate the existence of a PtSn alloy diluting metallic Pt atoms. In tin-modified systems, when a fraction of ionic tin is present, the activation of the C=O group of the cinnamaldehyde is favored, increasing the selectivity to UOL. An increase in the number of Pt 5d holes measured by Pt-L_2,3 XANES could be indicating as a d → s, p rehybridization process in the PtSn 3D small nanoclusters present in PtSn-BM catalysts.

Full text of DOI:10.1021/jp030579y

J. Phys. Chem. B 2003, 107, 11441-11451
11441
XPS and XAFS Pt L2,3-Edge Studies of Dispersed Metallic Pt and PtSn Clusters on SiO2
Obtained by Organometallic Synthesis: Structural and Electronic Characteristics
Jose´ M. Ramallo-Lo´pez,† Gerardo F. Santori,‡,§ Lisandro Giovanetti,† Mo´nica L. Casella,‡
Osmar A. Ferretti,*,‡,§ and Fe´lix G. Requejo†,|
Instituto de F´ısica de La Plata (IFLP) (CONICET) and Departamento de F´ısica, Facultad de Ciencias Exactas,
UniVersidad Nacional de La Plata, 49 y 115 (1900) La Plata, Argentina, Centro de InVestigacio´n y Desarrollo
en Ciencias Aplicadas “Dr. Jorge Ronco” (CINDECA), Facultad de Ciencias Exactas, and
Departamento de Ingenier´ıa Qu´ımica, Facultad de Ingenier´ıa, UniVersidad Nacional de La Plata,
47 No. 257 - C.C. 59 - (1900) La Plata, Argentina, and Materials Science DiVision, Lawrence Berkeley
National Laboratory, Berkeley, California 94706
ReceiVed: May 8, 2003; In Final Form: July 23, 2003
In this work, well-defined silica-supported Pt and PtSn catalysts were prepared by surface organometallic
reactions and were characterized by EXAFS-XANES at the Pt L2,3 edge and XPS of the Pt 4f and Sn 3d
levels. These catalysts were tested in the catalytic hydrogenation of cinnamaldehyde in the liquid phase. XPS
results showed that in PtSn-OM* (not having butyl groups anchored on the surface), tin was present as
Sn(0) adatoms “decorating” the metallic surface, isolating Pt atoms; in the PtSn-OM catalyst (Bu groups
remain grafted on the surface), tin was found in the form of Sn(0) and Sn(II, IV) in similar proportions.
EXAFS experiments do not provide evidence of the existence of PtSn alloys in any of these systems. In the
bimetallic PtSn-BM systems, it was possible to observe by XPS that tin was found in the form of Sn(0) and
Sn(II, IV). EXAFS experiments, in this case, allowed us to demonstrate the existence of a PtSn alloy diluting
metallic Pt atoms. In tin-modified systems, when a fraction of ionic tin is present, the activation of the CdO
group of the cinnamaldehyde is favored, increasing the selectivity to UOL. An increase in the number of Pt
5d holes measured by Pt-L2,3 XANES could be indicating as a d f s, p rehybridization process in the PtSn
3D small nanoclusters present in PtSn-BM catalysts.
Introduction
catalytic activity. In addition to aspects traditionally analyzed
in monometallic catalysts (structure and electronic properties
of small supported metal clusters), the characterization of
bimetallic catalysts includes the nature elucidation of the
interaction between the metals at low dimensionalities. This has
already been discussed in detail by Rodr´ıguez and Goodman
on model surfaces,6 but systems with lower dimensionalities
are still an open field. Besides, in the case of PtSn catalysts
studied in this work, an adequate explanation of its functioning
is closely related to the Sn oxidation state. For instance, some
results found in the literature7 indicate that, for PtSn/SiO2
catalysts having particle sizes between 2 and 20 nm, Pt and Sn
atoms in the same particle do not necessarily present a uniform
distribution. A newly developed atomic-resolution in situ TEM8
could be helpful in elucidating the location and oxidation state
of the promoter under reaction conditions, so it would be
possible to explain its influence upon the catalytic activity.
Another technique that has been very useful in the study of
highly dispersive bimetallic systems is energy-dispersive X-ray
spectrometry, which gives detailed information about particle
sizes and their composition.9,10 X-ray photoelectron spectroscopy
(XPS) is another adequate technique for carrying out this type
of characterization.11
Bimetallic catalysts based on Pt modified by a second metal
(Sn, Pb, Ge) have been gaining importance, from their classical
utilization in the reforming process of naphthas up to their use
in reactions concerning environmental and fine chemistry.1,2
Obtaining well-defined active phases in their nature and
composition leads to systems with better global catalytic
properties, and for this reason the development and comprehen-
sion of controlled-preparation techniques of solid catalysts have
maintained permanent interest. One of these techniques is
surface organometallic chemistry on metals (SOMC/M), which
allows us to obtain bimetallic and organobimetallic catalysts
with very good activity, selectivity, and stability in hydrogena-
tion and dehydrogenation reactions and is useful in obtaining
synthesis gas.3-5
The state of the base monometallic catalyst, Pt/SiO2, in the
case of the present work plays a fundamental role in the
preparation of the bimetallic system. Thus, taking into account
the superficial character of the reaction between the organo-
metallic (SnBu4) and Pt/SiO2, it is important to find this last
one highly dispersed, so its characterization is more difficult
from conventional methods.
The knowledge of the structure of small bimetallic particles
will contribute to the understanding and improvement of their
XAFS spectroscopy is nowadays one of the most powerful
methods of local-order characterization in highly dispersed metal
catalysts.12,13 This method is especially useful for investigating
the environment of very small metal particles, which are very
difficult to study by other techniques. In many cases, information
can be obtained in situ during the different steps of preparation
such as calcination, reduction, and even during catalytic reaction,
* Corresponding author. E-mail: ferretti@quimica.unlp.edu.ar.
† Instituto de F´ısica de La Plata (IFLP) (CONICET) and Departamento
de F´ısica, Universidad Nacional de La Plata.
‡ Centro de Investigacio´n y Desarrollo en Ciencias Aplicadas “Dr. Jorge
Ronco” (CINDECA), Universidad Nacional de La Plata.
§ Departamento de Ingenier´ıa Qu´ımica, Universidad Nacional de La Plata.
| Lawrence Berkeley National Laboratory.
10.1021/jp030579y CCC: $25.00 © 2003 American Chemical Society
Published on Web 09/20/2003

11442 J. Phys. Chem. B, Vol. 107, No. 41, 2003
Ramallo-Lo´pez et al.
when the catalyst becomes inaccessible to traditional surface-
analysis techniques. Usually, it is very difficult to perform
structural investigations in systems with concentrations of
modifying components below 1%. It has been shown in the
literature that the XAFS methods might be used successfully
with very small concentrations of modifying components and
could provide information about the local geometry and
electronic properties of the atoms in very dilute systems.14
Extended X-ray absorption fine structure (EXAFS) spectroscopy
gives information about the surroundings of the absorber atom.
In the particular case of Pt, the study of the L3 absorption edge
measuring the transmission of X-rays provides information about
the kind, number, and distance of neighbors around Pt atoms.
XANES (X-ray absorption near-edge spectroscopy) gives
information about the electronic state of the absorbing element.
For platinum atoms, the study of L2 and L3 edges gives
information on the density of unoccupied 5d states, measuring
the transition of core electrons to this band. Using the method
described by Mansour et al., it is possible to determine the
variation in the density of unoccupied d states in a sample
containing Pt with respect to Pt foil.15
In this work, we present EXAFS-XANES at the Pt L2,3 edge
and XPS of Pt 4f- and Sn 3d-level studies performed on Pt and
PtSn catalysts supported on SiO2, with the aim of gaining
knowledge about the localization and state of the species of
the metallic phase for a better comprehension of the catalytic
behavior. Also, the modifications in the Pt electronic structure
and the effects they cause were studied. For this purpose, a test
reaction of applied interest was selected, such as the hydrogena-
tion of cinnamaldehyde.
was obtained for the sample previously reduced at 773 K for 4
h and then evacuated at the same temperature overnight.
Afterward, the sample was submitted to an evacuation at room
temperature for 2 h, and a second isotherm was measured in
the same manner. The difference between the two isotherms
extrapolated to zero pressure gave the amount of the irreversibly
adsorbed hydrogen and CO (H/Pt and CO/Pt).
XPS Characterization. XPS data were obtained with an
ESCA 750 Shimadzu spectrometer equipped with a hemispheri-
cal electron analyzer and a Mg KR (1253.6 eV) X-ray source.
Fresh samples were mounted onto a manipulator that allowed
the transfer from the preparation chamber into the analysis
chamber. PtSn-OM samples were dried, and PtSn-BM samples
were reduced in situ at 673 K for 1 h. The binding energy (BE)
of the C 1s peak at 284.6 eV was taken as an internal standard.
The intensities were estimated by calculating the integral of each
peak after subtracting the S-shaped background and fitting the
experimental peak to a Lorentzian/Gaussian mix of variable
proportion.
X-ray Absorption Spectroscopy. XAS cells are a glass ring,
about 1-cm thick, with sealed Kapton windows used as sample
holders. The samples were prepared as described below at the
desired temperature in a glass reactor connected to the XAS
cells. The powdered samples were transferred under Ar or H2
to the XAS cells, which were then sealed. In the case of
experiments performed in H2 atmospheres, for Pt/SiO2 and
PtSn-BM catalysts, the reduction step was conducted in H2 at
773 K, and PtSn-OM and PtSn-OM* catalysts were tested
after preparation at 363 and 423 K, respectively. In the case of
experiments performed in Ar atmospheres (Pt/SiO2, PtSn-BM),
catalysts were reduced in H2 at 773 K and then treated in Ar at
773 K for 2 h.
Experimental Section
X-ray absorption spectra were measured at the XAS beamline
at the LNLS - National Synchrotron Light Laboratory, Campi-
nas, Brazil. EXAFS and XANES spectra of the Pt L3 edge (11.6
keV) and L2 (13.3 keV) were recorded at room temperature
using a Si(220) single channel-cut crystal monochromator in
transmission mode and with two ion chambers as detectors.
EXAFS data were extracted from the measured absorption
spectra by standard methods.16,17 Raw data files were averaged,
and the pre-edge region was approximated by a modified
Victoreen curve. Normalization was completed by dividing by
the height of the absorption edge, and the background was
subtracted using cubic spline routines. The main contributions
to the spectra were isolated by Fourier transformation of the
final EXAFS functions. The analysis was performed on these
Fourier-filtered data.
Catalyst Preparation. A Degussa silica (Aerosil 200, 200
m2 g-1) was used as support. The silica was suspended in NH4-
OH(aq) under stirring prior to the addition of the [Pt(NH3)4]2+
solution, having a concentration so as to obtain 1% w/w Pt
exchanged on the silica. The solid was kept under stirring for
24 h at 298 K, and then the suspension was separated by
filtration under vacuum. The solid was repeatedly washed, dried
at 378 K, calcined in air at 773 K, and reduced in flowing H2
at the same temperature, leading to the monometallic Pt/SiO2
catalyst.
The tin-modified catalysts preparation consisted of the
reaction of a solution of SnBu4 in a paraffinic solvent (n-heptane
when the preparation was conducted at 363 K and n-decane
when the preparation temperature was at 423 K) with the
reduced Pt/SiO2 catalyst under flowing H2. After 4 h of reaction,
the liquid phase was separated, and the solid was repeatedly
washed with n-heptane and subsequently dried in Ar at 363 K.
The solids obtained after this procedure were identified as PtSn-
OM (having butyl groups grafted to the surface) or PtSn-OM*
(with no butyl groups grafted to the surface). The bimetallic
phases (PtSn-BM) were obtained by elimination of the organic
groups by activation of PtSn-OM catalysts in flowing H2 at
773 K for 2 h. The variation in SnBu4 concentration and the
amount of hydrocarbons evolved during the preparation reaction
were analyzed using a Varian 3400 CX gas chromatograph
equipped with a flame ionization detector, employing a 10%
OV-101 column (1/8 in. i.d., 0.5-m length) and a tricresyl
phosphate column (1/4 in., 6-m length), respectively. Platinum
and tin contents were determined by atomic absorption.
Hydrogen and CO Chemisorption. Hydrogen and CO
chemisorption were measured in a static volumetric apparatus
at room temperature. For each sample, a first adsorption isotherm
The absorption edge positions of a sample, defined as the
major inflection point in the edge, are a function of the oxidation
state of the element and shift to a higher energy as the oxidation
state of the absorbing element becomes higher than zero.18
At the Pt L3 edge, the allowed bound-state to bound-state
transition from 2p3/2 to vacant d states of the absorbing atom
gives rise to a resonance peak (white line) above the edge due
to vacancies in the 5d level of Pt. The line shape of the resonance
contains information on the type of final state involved in the
transition. Lytle et al.19 have shown that increases in the
resonance peak intensity with respect to pure element absorbers
correlate with the ionicity estimated from the number of d
electrons removed from the element by the formation of
chemical bonds. Lytle20 has also shown that the L3 resonance
peak areas for the third-row transition metals increased from
Au to Ta, which is also the sequence for increased electron
vacancies in the d band. Meitzner et al.21 have found similar

Characteristics of Pt and PtSn Clusters on SiO2
J. Phys. Chem. B, Vol. 107, No. 41, 2003 11443
TABLE 1: Composition and Denomination of the Studied
Catalysts
results for Re, Os, Ir, Pt, and Au. They also observed that the
resonance for the metal in a highly dispersed state was only
slightly more intense than it was for the bulk form of the metal,
indicating a small electron deficiency of the highly dispersed
form relative to the bulk. To quantify the differences in white-
line intensity between the catalyst and platinum foil, a variation
of the method described by Mansour et al.15 was used.22 After
subtracting the platinum foil data of the catalyst data, the
resulting curves were numerically integrated between -10 and
+14 eV for both the L2 (∆A2) and L3 (∆A3) edges using
entry
catalyst
Pt/SiO2
Sn/Pt
stoichiometry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
PtSn-OM*a
PtSn-OM*a
PtSn-OM*a
PtSn-OMa
PtSn-OMa
PtSn-OMb*
PtSn-OMb
PtSn-OMb
PtSn-BMa
PtSn-BMa
PtSn-BMa
PtSn-BMa
PtSn-BMb
0.06
0.12
0.20
0.30
0.40
0.40
0.70
1.41
0.06
0.20
0.30
0.40
0.70
PtSn0.06
PtSn0.12
PtSn0.2
Pt(SnBu1.6
Pt(SnBu1.8
PtSn0.4
)
)
0.3
0.4
Pt(SnBu1.6
Pt(SnBu2.7
PtSn0.06
)
)
0.7
1.4
∆A3σ3 + 1.11∆A2σ2
A3rσ3 + 1.11A2rσ2
PtSn0.2
PtSn0.3
PtSn0.4
PtSn0.7
fd )
(1)
The fractional change in the total number of unfilled states
in the d band of the sample compared to the number in the
platinum foil (fd) was calculated. The areas are normalized23
by multiplying by the X-ray absorption cross section at the edge
jump (σ). Values of 117.1 and 54.2 cm2 g-1 were used for the
absorption cross section at the platinum L3 and L2 edges,24
respectively. When the number of unfilled d states in the
reference material (hTr) is known, the number of unfilled d states
in the sample (hTs) can be calculated from
a Prepared at 90 °C. b Prepared at 150 °C.
Blank experiments performed on silica via the above-
described preparation procedure did not evidence either a
variation in the concentration of tetra-n-butyltin between 298
and 423 K or a detectable concentration of tin in the solids.
The reaction between SnBu4 and Pt/SiO2 was followed by
GC analysis, measuring the variation in SnBu4 concentration
in the impregnating solution and the total butane amount evolved
per SnBu4 reacted. Results obtained show that a temperature
increase has a beneficial influence on both the SnBu4 reacted
with Pt/SiO2 and the reaction rate (eq 3). In fact, at 363 K, it is
possible to fix tin selectively onto platinum up to a Sn/Pt ratio
of approximately 0.40, whereas it increases to 1.40 when the
reaction temperature is 423 K. Table 1 shows results corre-
sponding to different platinum/tin catalysts prepared. Data
presented in Table 1 indicate that the cleavage of Sn-C bonds
and the elimination of butyl groups as butane depend on the
temperature. When reaction 3 is carried out at 363 K, there is
complete removal of butyl groups provided that the Sn/Pt atomic
ratio is kept under 0.2 (approximately 4BuH evolved per SnBu4
reacted), whereas for the plateau (Sn/Pt = 0.40) the amount of
butane eliminated per fixed tin is near 2.2, leading to an
organobimetallic-supported phase with a global stoichiometry
hTs ) (1 + fd) hTr
(2)
Samples of Pt/SiO2 and tin-modified platinum catalysts were
measured to determine the Pt surroundings. This enables us to
know if its neighborhood was modified by the incorporation of
tin so that we can study the interaction between Pt and Sn in
the tin-modified platinum catalysts to compare the number of
unfilled d states in both samples and to correlate all of these
properties with catalytic behavior.
EXAFS experiments were performed on Pt/SiO2, PtSn-OM,
and PtSn-BM samples to determine the nature, amount, and
distance of first neighbors of Pt atoms.
Catalytic Test. The hydrogenation of cinnamaldehyde (Al-
drich, 99+%) was carried out in a batch reactor, magnetically
stirred at 800 rpm, introducing the reactant in a toluene solution
(0.17 M). The temperature was kept at 313 K, and the pressure
was kept at 0.1 MPa of H2 during the experiment. The mass of
catalyst was 0.5 g. Samples of the reaction products were
analyzed using a Varian 3400 CX gas chromatograph equipped
with a capillary column (30 m × 0.53 in. i.d., DB Wax bonded
phase) and a FID detector.
of Pt(SnBu1.8)0.4/SiO2 (PtSn-OM-type phase). For higher tem-
peratures, for example, 423 K, reaction 3 proceeds with a
complete detachment of butyl groups as BuH, up to a Sn/Pt
ratio close to 0.40 leading to a phase with a stoichiometry of
PtSn0.4/SiO2, whereas for Sn/Pt = 1.40 an evolution of ap-
proximately 1.3 butane per fixed tin is observed and the
organobimetallic phase so obtained presents a global stoichi-
Before being submitted to the hydrogenation test, Pt/SiO2 and
PtSn-BM catalysts were pretreated under flowing H2, increasing
the temperature from ambient to 773 K, and holding it for 2 h.
PtSn-OM catalysts were prepared and tested in the same reactor
to prevent any contact with the atmosphere.
ometry of Pt(SnBu2.7)1.4/SiO2. When a PtSn-OM-type phase
is activated in hydrogen at 773 K, it is transformed into a
bimetallic PtSn/SiO2 catalyst, denoted PtSn-BM (reaction 4).
Results of hydrogen and carbon monoxide chemisorption
showed H(CO)/Pt values between 0.6 and 0.7 (Table 2), which
indicate a high dispersion of the metallic phase. An important
decrease in H(CO)/Pt values is observed for tin-modified
catalysts. In this case, the bimetallic phase chemisorbs gas up
to a ratio of approximately 0.20-0.25 H(CO)Pt, which might
be ascribed to an homogeneous and specific deposition of tin
over superficial platinum, without discarding the possibility that
electronic modifications generated by tin may contribute to this
effect. Previously published results of TEM measurements
showed that the particle-size distribution of both mono- and
bimetallic catalysts were very narrow, with a mean particle
diameter around 2.0 nm for Pt/SiO2 and with a slight increase
in the mean for PtSn-BM on the order of 0.5 nm.25 The
variation in particle size is in agreement with the results obtained
Results
Preparation, H2/CO Adsorption, and XPS Characteriza-
tion of Tin-Modified Platinum Catalysts. The two-step
anchoring process to prepare the PtSn-OM and PtSn-BM
catalysts can be represented by the following reaction scheme:
363-423 K: Pt/SiO2 + ySnBu4 + xy/2H2 f
Pt(SnBu4-x)y/SiO2 + xyBuH (3)
423-773 K: Pt(SnBu4-x)y/SiO2 + (4 - x)y/2H2 f
PtSny/SiO2 + (4 - x)yBuH (4)

11444 J. Phys. Chem. B, Vol. 107, No. 41, 2003
Ramallo-Lo´pez et al.
14
TABLE 2: H2/CO Chemisorption and XPS Results for PtSn-BM, PtSn-OM, and PtSn-OM* Systems
entry
1
6
7
13
catalyst
Pt/SiO2
PtSn-OM
PtSn-OM*
PtSn-BM
PtSn-BM
H/Pt
CO/Pt
binding energy (BE)
0.64
0.56
71.6
nd
nd
70.8
484.3
487.0
0.45
nd
nd
70.6
484.6
0.21
0.25
70.9
485.0
487.0
0.67
0.20
0.24
70.6
484.6
486.5
0.64
Pt 4f7/2
Sn(0) 3d5/2
Sn(II, IV) 3d5/2
Sn(0)/[Sn(0) + Sn(II, IV)]
1.00
by other authors for PtSn/γ-Al2O3 catalysts prepared using an
analogous methodology.26 The authors also assigned the de-
crease in the quantity of chemisorbed hydrogen to the selective
deposition of tin onto platinum.
Pt/SiO2, PtSn-OM, PtSn-OM*, and PtSn-BM samples
(entries 1, 6, 7, 13, and 14 in Table 1) were chosen for
characterization by XPS and/or EXAFS. The PtSn-OM catalyst
is a phase whose stoichiometry was determined as Pt[SnBu1.8]0.4/
SiO2; the PtSn-OM* catalyst consists of the phase having a
stoichiometry of PtSn0.4/SiO2. The PtSn-BM catalysts present
a stoichiometry of PtSnx/SiO2 (x ) 0.4 or 0.7) and are obtained
by activation in H2 at 773 K of the corresponding PtSn-OM
catalyst.
Table 2 shows XPS results giving the position of all of the
main photoelectron peaks after referencing them to the C1s BE
of 284.6 eV, and Figure 1 shows the XPS spectra of the Sn
3d5/2 level for the tin-modified catalysts. For all of the studied
catalysts in the region corresponding to Pt 4f7/2 (around 71 eV),
only one peak appears, indicating the complete reduction of
platinum. In considering this peak, an interesting aspect arises:
in the three systems modified by tin, a shift is observed in the
BE toward lower values of approximately 0.7-1 eV with respect
to Pt/SiO2. In relation to PtSn-OM and PtSn-BM catalysts,
two peaks are also observed around 485 and 487 eV, which
could be assigned to metallic tin (Sn(0)) and ionic tin (Sn(II,
IV)), respectively. For the PtSn-OM sample, Sn(0) and Sn(II,
IV) are found in similar proportions, whereas for the PtSn-
BM samples, approximately 70% is found as Sn(0) and the
remaining part is found as Sn(II, IV). The PtSn-OM* sample
presents different behavior in relation to the other tin-modified
catalysts, showing only one peak in the region of Sn 3d5/2 at
484.6 eV, which indicates that all tin is found as Sn(0).
X-ray Absorption Spectroscopy. XANES data were ob-
tained for Pt/SiO2 and PtSn-BM catalysts in H2 and in Ar after
evacuation and for a Pt foil reference at the Pt L2 and L3
absorption edges.
The normalized L2,3 XANES spectra of both samples in H2
and of the Pt foil reference are compared in Figures 2 and 3.
Neither the L2 nor the L3 absorption edge of the Pt/SiO2 sample
exhibits a substantial energy shift relative to that of the foil.
Difference spectra were obtained by subtracting the Pt foil
reference spectrum from that of the sample at each edge. The
L2 difference spectrum (Figure 2.a) contains a very weak
negative peak near the edge and a strong positive peak 8 eV
from the edge, whereas the L3 difference spectrum (Figure 2b)
contains a stronger negative feature after the edge and a similar
positive peak at 8 eV. These results for reduced Pt/SiO2 have
already been reported in the literature.22 Ramaker et al. have
Figure 1. XPS spectra of Sn 3d5/2 level for (a) PtSn-OM* (entry 7),
(b) PtSn-OM (entry 6), (c) PtSn-BM (entry 13), and (d) PtSn-BM
(entry 14).
Figure 2. Comparison between the absorption L2 (a) and L3 (b) edges
of Pt/SiO2 (entry 1) in H2 (dotted line) and bulk Pt (solid line). Dashed
lines show the difference spectra.

Characteristics of Pt and PtSn Clusters on SiO2
J. Phys. Chem. B, Vol. 107, No. 41, 2003 11445
Figure 4. Comparison between the absorption L2 (a) and L3 (b) edges
of Pt/SiO2 (entry 1) in Ar (dotted line) and bulk Pt (solid line). Dashed
lines show the difference spectra.
Figure 3. Comparison between the absorption L2 (a) and L3 (b) edges
of PtSnBM (entry 14) in H2 (dotted line) and bulk Pt (solid line). Dashed
lines show the difference spectra.
proposed that these peaks at 8 eV can be attributed to the
presence of chemisorbed hydrogen on Pt atoms.27 Structure-
related features are observed in each difference spectrum at 15-
40 eV. The negative sense of these features indicates the
presence in the spectrum of Pt foil but the absence in the spectra
of the samples. The negative feature at 15 eV present in both
L2 and L3 difference spectra resembles in position the maximum
of the first oscillation beyond the L3 white line of the Pt foil.
Hence, this negative feature can be attributed to a reduction in
the coordination number of Pt in going from the foil to the
clusters.28
The normalized Pt L2,3 spectra of the bimetallic catalyst in
H2 and a Pt foil reference are compared in Figure 3. The L3
absorption edge of this sample exhibits a shift of 0.8 eV to
higher energy related to the reference, but the L2 edge has no
energy shift and only a change in the edge shape. The L2
difference spectrum (Figure 3a) shows a positive broad feature
at 7 eV, but the L3 difference spectrum (Figure 3b) shows a
negative sharp peak near the edge and a positive peak at 7 eV.
Structure-related features appear again in the 15-40 eV region.
The negative feature at 15 eV found for the monometallic
catalyst is also present and is attributed again to a reduction of
the coordination number of Pt related to Pt foil.
Figure 5. Comparison between the absorption L2 (a) and L3 (b) edges
of PtSnBM (entry 14) in Ar (dotted line) and bulk Pt (solid line). Dashed
lines show the difference spectra.
feature at 7 eV, but the L3 difference spectrum (Figure 5b) shows
a negative sharp peak near the edge and a positive peak at 7
eV. Structure-related features appear again in the 15-40 eV
region.
Figures 6 and 7 show the k3-weighted Fourier transforms of
EXAFS L3-edge spectra of the catalyst in H2 and Ar atmo-
spheres, also known as the radial distribution function. The solid
line shows the amplitude and imaginary part of the Fourier
transforms. Dashed lines show the amplitude and the imaginary
part of the fitted EXAFS functions whose parameters are shown
in Table 3.
It has been already shown that the adsorption of H2 on Pt
crystallites on different supports has important effects on its
XANES spectra.27,29,30 Figures 4 and 5 show the normalized
L2,3 XANES spectra of both samples in Ar. Differences between
the reference and the catalyst near the edge are smaller than
those obtained in H2. The L2 difference spectrum for the
monometallic catalyst (Figure 4a) contains a very weak positive
feature at 5 eV, but the L3 difference spectrum (Figure 4b)
contains a weak negative feature before the edge and a similar
positive peak at 5 eV. As in the spectra obtained in the H2
atmosphere, structure-related features are observed in each
difference spectrum at 15-40 eV.
The normalized Pt L2,3 spectra of the bimetallic catalyst and
a Pt foil reference are compared in Figure 5. The L3 absorption
edge of this sample exhibits a shift of 0.6 eV to higher energy
related to the reference, but the L2 edge has no energy shift
and only a change in the edge shape. This might indicate that
interaction of Sn atoms with the 5d5/2 state differs from that of
the 5d3/2 state of platinum. Difference spectra between the
bimetallic catalyst and Pt foil reference are also shown in Figure
5. The L2 difference spectrum (Figure 5a) shows a positive broad
Figure 6a shows the Fourier transform of the Pt/SiO2 catalyst
in H2. Only one peak at 2.6 Å (without phase correction) is
observed, suggesting that all Pt is in a reduced state, forming
metallic particles. Figure 6b and c show Fourier transforms of
PtSn-OM and PtSn-OM*, respectively. Both show a similar
pattern with a broad contribution of scattering atoms between
1 and 3 Å, owing to at least two types of atoms. Figure 6d
shows the Fourier transform of the bimetallic PtSn-BM catalyst.
Two peaks at 2 and 2.6 Å (without phase corrections) are
observed, possibly showing the existence of bimetallic particles
in the catalyst.
Figure 7 shows the k3-weighted Fourier transforms of EXAFS
L3-edge spectra of samples in Ar. Figure 7a shows the Fourier
transform of the spectrum of Pt/SiO2, and Figure 7b shows the

11446 J. Phys. Chem. B, Vol. 107, No. 41, 2003
Ramallo-Lo´pez et al.
TABLE 3: EXAFS Analysisa
coordination distance
entry
catalyst
Pt/SiO2 [H2]
PtSn-OM
pair
number N
R (Å) σ2 (Å-2
2.73(1) 0.0064(2) 6.2(3)
)
E0 (eV)
1
6
Pt-Pt
8.3(1)
Pt-C
Pt-Pt
6(1)
6.7(6)
1.98(5) 0.025(2)
2.68(1) 0.011(1)
4.5(3)
4.9(3)
7
PtSn-OM*
Pt-C
Pt-Pt
4.4(5)
7.9(5)
2.0(1) 0.0084(5) 5.7(3)
2.67(2) 0.0096(3) 5.0(4)
14 PtSn-BM [H2] Pt-Sn
Pt-Pt
2.7(1)
1.8(1)
2.68(1) 0.0055(1) 5.57(4)
2.71(2) 0.0032(8) -1.82(8)
1
Pt/SiO2 [Ar]
Pt-Pt
8.5(1)
2.71(1) 0.0081(1) 5.6(3)
14 PtSn-BM [Ar] Pt-O
1.1(1)
2.7(1)
4.1(2)
1.99(1) 0.0039(2) 4.7(3)
2.65(1) 0.0092(4) 5.6(3)
2.71(1) 0.0074(3) 3.5(3)
Pt-Sn
Pt-Pt
a Bond distances, coordination numbers, Debye-Waller factors, and
the parameter E0.
Figure 6. Fourier transforms of EXAFS data of samples in H2: (a)
Pt/SiO2 (entry 1), (b) PtSnOM (entry 6), (c) PtSnOM* (entry 7), and
(d) PtSnBM (entry 14). Solid lines show the amplitude and imaginary
part of the Fourier transforms. Dashed lines show the corresponding
Fourier transforms of the fitted functions.
Figure 8. Conversion of cinnamaldehyde as a function of time for
(0) Pt/SiO2 (entry 1), (O) PtSn-BM (entry 13), (4) PtSn-OM (entry
6), and ()) PtSn-OM* (entry 7). For experimental conditions, see the
text.
To analyze these data quantitatively, the main peaks were
isolated and fitted using standard procedures.31 Different
combination of shells were used to fit each spectrum. Theoretical
standards were generated by the FEFF program.32 In the fitting,
the bond distances, coordination numbers, Debye-Waller
factors, and E0 parameter for each atomic pair were allowed to
vary independently. The reduction factor S20 was obtained from
a Pt foil and was kept fixed for all samples. Results are shown
in Table 3. The monometallic sample shows similar results in
both atmospheres. In both cases, only a Pt-Pt shell was fitted,
indicating that the Pt on these samples is in metallic form. The
average coordination number of the Pt-Pt shell for these two
samples is smaller than 12, which is the average coordination
number for the Pt foil, indicating the small size of the crystals.
EXAFS spectra of PtSn-OM and PtSn-OM* were fitted
using two shellssone Pt-C shell and one Pt-Pt shell (Table
3). Even the fits are in agreement with experimental data; other
atoms could be present in the surroundings of some Pt atoms,
though because of its small contributions a confident fit with
more shells would not be possible. Some differences are found
between the fitted parameters of each sample. The number of
Pt first neighbors is smaller in the PtSn-OM sample, and the
number of C neighbor atoms is larger. The average coordination
number for Pt-Pt of both samples is smaller than that of the
monometallic sample, indicating the isolation of Pt particles due
to the presence of organotin fragments.
Figure 7. Fourier transforms of EXAFS data of samples in Ar. (a)
Pt/SiO2 (entry 1) and (b) PtSnBM (entry 14). Solid lines show the
amplitude and imaginary part of the Fourier transforms. Dashed lines
show the corresponding Fourier transforms of the fitted functions.
Fourier transform of the spectrum of bimetallic catalyst PtSn-
BM. Both Fourier transforms are very similar to those found in
the H2 atmosphere for both catalysts. Only one peak at 2.6 Å is
present in the Fourier transform corresponding to the mono-
metallic catalyst. In the case of the bimetallic catalyst, two peaks
are again present (one at 2.1 Å and the other at 2.7 Å).

Characteristics of Pt and PtSn Clusters on SiO2
J. Phys. Chem. B, Vol. 107, No. 41, 2003 11447
SCHEME 1: Reaction Pathways for the Hydrogenation of Cinnamaldehyde
TABLE 4: Selectivity to SAL, UOL, and SOL at 10%
Conversion for All of the Studied Catalystsa
the following sequence for SUOL: PtSn-OM > PtSn-BM .
PtSn-OM* . Pt/SiO2. It is very interesting to observe the Sn
effect on the reaction rate and on the selectivity to UOL. These
results are presented in Figures 10 and 11 as a function of the
Sn/Pt atomic ratio for the OM and BM series, respectively. For
all of the tin concentrations analyzed, the hydrogenation rate
suffers an increase with respect to that of the monometallic Pt/
SiO2 catalyst, reaching the maximum values for Sn/Pt between
0.2 and 0.4. Starting from Sn/Pt higher than 0.3, the SUOL is
markedly promoted by tin, reaching values between 80 and 90%.
In all cases, the catalysts of the OM series show SUOL values
higher than those of the BM series.
entry
catalyst
SSAL
%
SUOL
%
SSOL
%
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Pt/SiO2
66.15
37.55
31.70
29.10
23.00
20.58
40.00
12.77
10.56
49.35
47.22
34.39
23.24
12.89
25.88
62.45
68.30
68.00
73.80
77.80
54.00
87.23
89.44
47.20
50.87
64.00
74.04
87.11
7.96
PtSn-OM*
PtSn-OM*
PtSn-OM*
PtSn-OM
PtSn-OM
PtSn-OM*
PtSn-OM
PtSn-OM
PtSn-BM
PtSn-BM
PtSn-BM
PtSn-BM
PtSn-BM
2.79
3.20
1.62
6.00
3.45
1.91
1.59
2.72
a For experimental conditions, see the text.
The fits of PtSn-BM samples in both Ar and H2 atmospheres
show that there exist some differences in the formation of Pt
structures. The sample in H2 could be fitted using two shells
corresponding to the first shells around Pt in a PtSn alloy.33
The sample in Ar needed three shellssthe same two first shells
of Pt in PtSn-BM used in the fit of the sample in H2 and an
additional Pt-O shell. The other distinctive aspect of the fits is
that the coordination number for the Pt-Pt distance is higher
for the sample in Ar.
Hydrogenation of Cinnamaldehyde. In cinnamaldehyde,
besides the double CdC bond and the CdO group, the benzene
ring is susceptible to hydrogenation; hydrogenolysis reactions
can also occur. Under the operating conditions used in this work,
products that can be assigned to these reactions are not observed.
The only products detected at quantifiable levels are, according
to the pathways in Scheme 1, cinnamic alcohol (UOL),
phenylpropanal (SAL), and phenylpropanol (SOL).
Figure 9. Cinnamaldehyde hydrogenation. Selectivity to UOL as a
function of conversion for (0) Pt/SiO2 (entry 1), (O) PtSn-BM (entry
13), (4) PtSn-OM (entry 6), and ()) PtSn-OM* (entry 7). For
experimental conditions, see the text.
Figure 8 shows the conversion curves as a function of time
for Pt/SiO2, PtSn-BM, PtSn-OM, and PtSn-OM* systems
(entries 1, 6, 7, and 13). The shape of these curves is compatible
with a good level of stability of the active phases, and a reaction
order near zero with respect to cinnamaldehyde is in a good
agreement with results published in the literature on PtSn and
RuSn.34
Table 4 shows the catalytic results of the cinnamaldehyde
hydrogenation for these systems. In the case of the monometallic
catalyst, the major product is the saturated aldehyde; SUOL is
maintained at less than 30%. The tin addition produces a
noticeable change in the distribution of products, increasing SUOL
and decreasing SSAL. As can be seen in Figure 9, SUOL for tin-
modified catalysts is much higher than for monometallic
catalysts in the complete interval of conversions, according to
Figure 10. Hydrogenation of cinnamaldehyde for PtSnOM catalysts
as a function of the Sn/Pt atomic ratio. (O) Formation rate (mmol s-1
gPts-1) of UOL (estimated between conversion 0 and 10%) and (9)
selectivities to UOL at 40% conversion. For experimental conditions,
see the text.

11448 J. Phys. Chem. B, Vol. 107, No. 41, 2003
Ramallo-Lo´pez et al.
bulk because of the high proportion of surface atoms. This effect
is dependent on the size and shape of the metallic cluster. The
platinum particle size in Pt/SiO2 after reduction may be
estimated from coordination number N obtained for the Pt-Pt
shell according to the method proposed in the literature,36
assuming the spherical particle of the fcc package. From our
results, the diameter of the metallic particles is estimated to be
between 12 and 14 Å in both samples (approximately 500
atoms), which is compatible with a high level of dispersion as
shown by H/Pt values in Table 2. The Pt-Pt distance fitted for
these clusters is 2.73 Å for H2 and 2.71 Å for Ar; for the bulk
Pt, it is 2.77 Å. The Pt-Pt bond contraction that is found is
consistent with previous results.37 In effect, it has been reported
that nearest-neighbor distances determined from EXAFS mea-
surements of “bare” (i.e., without adsorbed gases) Pt clusters
show contractions of approximately 0.07 Å (2.5%) relative to
the bond length in bulk Pt. This shorter metal-metal bond
distance is in agreement with the known general trend observed
in metallic clusters. (See, for example, ref 38 for experimental
evidence and ref 39 for theoretical predictions.)
Figure 11. Hydrogenation of cinnamaldehyde for PtSnBM catalysts
as a function of the Sn/Pt atomic ratio. (O) Formation rate (mmol s-1
gPts-1) of UOL (estimated between conversion 0 and 10%) and (9)
selectivities to UOL at 40% conversion. For experimental conditions,
see the text.
PtSn-OM and PtSn-OM* samples show a more complicated
radial distribution function. At least two different scatterer atoms
must be present to obtain such a result. Considering the way
the catalysts are prepared, we use Pt-C and a Pt-Pt shell to
perform the fits. Before the reduction process, a solvent fraction
remains adsorbed on the samples, so the appearance of C in
the vicinity of Pt atoms is natural. We do not expect the
appearance of oxygen because samples are not exposed to an
oxidizing atmosphere and the initial state (the monometallic Pt/
SiO2 catalyst) does not present such kinds of interactions.
Anyway, experimental evidence is not enough to discard the
presence of the very low contents of other scatterers, which
could be as nearest-neighbor atoms.
Figure 12. Comparison of the radial distribution functions of the Pt
foil (dashed line) and the Pt/SiO2 catalyst in Ar. Arrows show the
positions of the second, third, and fourth coordination shells. The Pt
foil Fourier amplitude was divided by 2 for scaling purposes
Results show that the coordination number for the Pt-Pt shell
is smaller in both samples than that of the monometallic sample.
The coordination numbers for the Pt-C shells are higher for
PtSn-OM. This is an expected result because of the presence
of PtSnBux in this sample. However, considering that the same
starting material was used for both samples and that temperatures
during preparation were kept below 423 K, no change in the
size of Pt particles is expected. Because of this, the diminution
of the Pt-Pt coordination number should be an artifact of the
fitting caused by the presence of the Pt-C shells and the
inhomogeneity of the structure.
Discussion
EXAFS experiments of Pt/SiO2 and tin-modified platinum
catalysts demonstrate that the Sn addition has strong effects on
Pt structures. This is clearly evident just from the comparison
of the Fourier transforms of EXAFS spectra of both samples.
The analysis of the parameters obtained from the fit of these
spectra let us obtain important information on the structures
present in each catalyst.
The PtSn-BM catalyst has a similar radial distribution
function in both atmospheres. However, some differences are
observed from the fitted parameters. The existence of a Pt-Sn
alloy is evident from the results. Two shells are enough to obtain
a very good fit for the sample in H2 consisting of Pt-Pt and
Pt-Sn shells. When the H2 is evacuated and the sample is kept
in an inert atmosphere of Ar, a Pt-O shell appears, and an
important increase in the coordination number of the Pt-Pt shell
is observed. We will first analyze results in H2 to explain this
behavior.
The monometallic catalyst shows a similar radial distribution
function in both atmospheres, H2 and Ar. In both samples, the
presence of only one type of scatterer atom for Pt in their radial
distribution functions is observed. They are perfectly fitted,
proposing only a Pt-Pt shell. No evidence of a Pt-O shell is
present, which would appear as a peak at a distance of 1.7 Å.35
All Pt is in a reduced state forming metallic particles, in
agreement with the XPS results shown in Table 2. Practically
no difference is observed in the average coordination number
fitted. Figure 12 shows a comparison between the Fourier
transform of the spectra of Pt/SiO2 in Ar and a Pt foil. The
amplitude of the Pt foil Fourier transform has been divided by
2 in the figure so that spectra can be more easily compared.
The arrows indicate the positions of higher Pt shells. It is clearly
observed that second, third, and even fourth Pt shells are present
in the catalyst, indicating that Pt crystallites have 3D fcc
structures. The fitted average coordination numbers are lower
than that of bulk Pt, indicating the small size of Pt particles.
When small metal clusters are examined by EXAFS, the average
coordination number is smaller than the one observed in the
It has been reported that PtSn and Pt3Sn are the most abundant
alloys in Pt-Sn bimetallic systems.40 For this reason, we will
consider these two alloys in our analysis. In Pt3Sn, Pt atoms
have 12 first-neighbor Pt atoms and 4 first-neighbor Sn atoms
at a distance of 2.828 Å.41 If the structure present in our catalyst
was this type of alloy, then one would expect a ratio of Pt and
Sn atoms in the first coordination of 3 to 1. This is clearly not
the case from our fits (Table 3). Moreover, the distances are
not equal for the two types of atoms as in the alloy. Thus, we
would not expect our bimetallic particles to have the structure

Characteristics of Pt and PtSn Clusters on SiO2
J. Phys. Chem. B, Vol. 107, No. 41, 2003 11449
TABLE 5: Intensity of Pt L2 and L3 X-ray Absorption
of this alloy. In contrast, in the PtSn alloy, Pt has 2 Pt atoms at
2.72 Å and 6 Sn atoms at 2.73 Å. In this case, one would expect
a Pt/Sn coordination ratio of 1 to 3. This is closer to the ratio
found in our case (Table 3). However, the distances for each
coordination shell are inverted. That is, Sn atoms are closer than
Pt atoms in our sample, contrary to what happens in the alloy,
though differences are small. This can be understood in terms
of the coexistence of two different phases in the catalyst: a
PtSn alloy and some unalloyed Pt particles. The unalloyed Pt
phase would contribute to the Pt-Pt shell increasing both its
coordination number and its distance. This would explain why
the ratio between coordination numbers for the Pt-Pt and Pt-
Sn shells is not 1 to 3 as expected for the PtSn alloy. In addition,
Borgna et al. have shown that the presence of the peak at 2.15
Å found in the radial distribution function is not only a
fingerprint for Pt-Sn interactions but also evidence of the
existence of unalloyed Pt.42
a
Edges on Pt/SiO2 and PtSn/SiO2 in Ar and in H2
entry
sample
∆A2
∆A3
fd
hTs
1
14
1
Pt/SiO2 in Ar
PtSn-BM in Ar
Pt/SiO2 in H2
-0.04
1.12
0.62
0.09
0.01
0.46
0.30
0.017
0.08
0.116
0.17
1.627
1.728
1.785
1.872
14
PtSn-BM in H2
1.66
a Their corresponding fractional change in the total number of
unfilled states in the d band compared to the number in the platinum
foil (fd) and the number of unfilled d states in the samples (hTs)
number of d holes for the monometallic catalyst, with the change
observed previously being caused only by the H2 adsorption.
For the PtSn-BM catalyst in Ar, there is an increase in the
number of d holes relative to the number in the Pt foil (Table
5), but this increase is smaller than that found in H2.
A 5d hole density on metal atoms covered with hydrogen
higher than the electron density in the bulk metal has been
published for Pt/γ-Al2O3,46 Pt/H-LTL,47 and Pt/SiO215 catalysts,
but no previous calculation for the bimetallic PtSn catalyst has
been reported. However, Stagg et al.48 found for PtSn/ZrO2
catalysts that the L3 white-line intensity was much lower than
that for the Pt foil, which would indicate a lower number of d
holes than in the Pt foil. The same result was found by Roma´n-
Mart´ınez et al.49 for PtSn/C catalysts. Nevertheless, they have
not investigated the L2 edge, so a conclusion on the number of
d holes is not possible on the basis of only their results. Jeon et
al.50 showed that Pt atoms in a PtSn alloy have a 5d electron
population smaller than that of the atoms in metallic Pt. This is
consistent with our results, considering the appearance of a PtSn
alloy in the bimetallic catalyst. Rodr´ıguez et al. have made ab
initio self-consistent-field calculations that predict a small shift
of s and p electrons from Sn toward Pt in 2D systems.51 The
formation of a Pt-Sn bond leads to an increase in the Pt (6s,
6p) electron population and a decrease in the Pt (5d) population,
as shown in our results. This phenomenon is equivalent to a d
f s, p rehybridization that moves electrons around Pt into the
Pt-Sn bond. Our results would show that a similar rehybrid-
ization process could also be taking place in PtSn 3D small
nanoclusters. In this context, the binding energy shifting (of
about 1 eV) to higher values at the Pt 4f level observed in XPS
results for bimetallic samples with respect to those for the
monometallic samples would originate in the electronic transfer
from the Pt 4f level to the Pt-Sn bond.
Results obtained from the characterization of Pt/SiO2 phases
and from the systems modified by Sn permit us to interpret the
catalytic behavior of these phases in the selective hydrogenation
of cinnamaldehyde. As previously mentioned, in all catalysts
studied, the Pt is found to be completely reduced. The modified
systems with Sn named PtSn-OM and PtSn-OM* (used in
an H2 atmosphere at temperatures below 423 K) do not present
PtSn alloys. In the case of PtSn-OM*, which does not present
Bu groups anchored on the surface or Sn(II, IV) (according to
XPS results), Pt atoms would be isolated by Sn(0) adatoms
“decorating” the metallic surface. In the PtSn-OM catalyst, Bu
groups remain grafted on the surface, with Sn in the form of
Sn(0) and Sn(II, IV) in similar proportions. The image of the
active phase would be represented by Pt atoms isolated by
PtSnBux islets, as represented in Scheme 2a. In bimetallic
systems PtSn-BM, it is possible to observe that a part of the
Pt is alloyed with the metallic Sn (PtSn, according to EXAFS
results) and that the other part of the Pt is alloyed with metallic
Pt atoms isolated from such an alloy. There also exists a Sn
percentage of ionic nature (20-30%), probably placed in the
When the bimetallic is sealed in Ar, some changes in the
alloy structure are found. The most noticeable aspects are the
appearance of a Pt-O shell and the increase in the Pt-Pt
coordination number. Both of them are related. It has been
shown for NiSn bimetallic catalysts that when they are exposed
to an inert gas the metal with a lower sublimation point is
segregated from the alloy.43,44 In our case, Sn would be
segregated from the PtSn alloy, increasing the size of the Pt
particles and thus increasing the Pt-Pt average coordination
number and making more important the interaction with the
SiO2 support, which could be the explanation of the Pt-O shell.
These interactions may be present in the sample in H2, but
because of its small contribution to the total signal, it cannot
be fitted confidently.
The coexistence of the PtSn alloy and an unalloyed metallic
Pt phase can be confirmed by considering the number of Pt
and Sn atoms in the catalyst. Thus, there are more Pt atoms
than Sn atoms available to form the alloy (Sn(0)/Pt ) 0.5 from
XPS results presented in Table 2). Because the stoichiometry
of the alloy is PtSn, the rest of the Pt atoms should be in a
metallic state because no other type of scatterer is found in our
EXAFS results, indicating that there are no other Pt phases
present. Besides, XPS results indicate that platinum was
completely reduced in this sample.
XANES data show that the addition of Sn has important
effects on the electronic structure of Pt. This is not surprising
because the EXAFS results show the existence of a PtSn alloy
in the bimetallic sample. To quantify this effect, we estimated
changes in the d band of Pt. Calculations of the number of holes
in the d band of platinum clusters and bimetallic clusters were
performed following a variation of the method proposed by
Mansour et al.15 Edges of samples and the Pt foil were aligned,
and their differences were numerically integrated from -10 to
14 eV. The fractional change in the total number of unfilled
states in the d band of the sample compared to the number in
the platinum foil (fd) was calculated using eq 1. The number of
unfilled d states in the sample (hTs) was calculated from eq 2,
using 1.6 as the number of unfilled 5d states in platinum.45
Results are summarized in Table 5. It was found that the average
platinum atoms in both the monometallic and the bimetallic
samples in H2 have fewer 5d electrons than bulk platinum atoms,
with the number of holes in the PtSn-BM sample being higher
than that in the Pt/SiO2 sample. Because the chemisorption of
hydrogen has an important effect on the electron properties of
Pt particles,27,30 we measured the same samples in Ar to separate
the effect of the tin addition from the one produced by hydrogen.
Results in Ar show that there is practically no change in the

11450 J. Phys. Chem. B, Vol. 107, No. 41, 2003
Ramallo-Lo´pez et al.
SCHEME 2: Representations of Catalytic Surfaces for
(a) PtSnOM and PtSnOM* and (b) PtSnBM
explain the higher yields of UOL reached in these catalysts. In
PtSn-BM catalysts, the presence of Pt(0) and Sn(II, IV) has
also been determined, but in this case the interaction would be
less efficient than for the PtSn-OM catalystprobably because
of the location of the ionic tin.
Conclusions
Well-defined silica-supported Pt and PtSn catalysts were
prepared by surface organometallic reactions and were charac-
terized by XPS and XAFS techniques and the catalytic
hydrogenation of cinnamaldehyde. The main conclusions can
be summarized as follows:
XPS results show that in PtSn-OM* (not having Bu groups
anchored on the surface) tin is present as Sn(0) adatoms
“decorating” the metallic surface, isolating Pt atoms; in the
PtSn-OM catalyst (Bu groups remain grafted on the surface),
tin is found in the form of Sn(0) and Sn(II, IV) in similar
proportions. EXAFS experiments do not evidence the existence
of PtSn alloys in any of these systems.
In bimetallic systems PtSn-BM, it is possible to observe by
XPS that tin is found in the form of Sn(0) and Sn(II, IV).
EXAFS experiments, in this case, allow us to demonstrate the
existence of a PtSn alloy diluting metallic Pt atoms.
A d f s, p rehybridization process could be taking place in
PtSn 3D small nanoclusters, leading to an increase in the number
of Pt 5d holes measured by Pt L2,3 XANES.
In tin-modified systems, especially PtSn-OM and PtSn-
BM where a fraction of ionic tin is present, the activation of
the CdO group of the cinnamaldehyde is favored, increasing
the selectivity to UOL.
metal-support interface. The image of the catalytic surface for
PtSn-BM can be represented by Scheme 2b.
In the hydrogenation of cinnamaldehyde, as observed in
Figure 8 and in Table 4, the hydrogenation rate of the CdO
group leading to UOL formation shows a noticeable increase
in catalysts modified with tin with respect to Pt/SiO2. The effect
of site isolation generated by tin favors the presence of species
leading to UOL, mainly of types η1-(O) and η2-(C,O), according
to the classical scheme of adsorption proposed in the literature
for R,â unsaturates aldehydes,52 which is clearly favorable to
the selectivity to UOL. However, the increase in the reaction
rate seems rather controversial, taking into account the diminu-
tion in the quantity of chemisorbed hydrogen, at least when
comparing PtSn-BM and Pt/SiO2. This indicates that the
chemisorbed hydrogen is enough, even in tin-modified catalysts,
and that the chemisorption of hydrogen is not the rate-controlling
step. The increase in the reaction rate has to be explained by a
modification in the electronic nature of the active site, which
agrees with our XANES results of an electronic transfer from
Pt to the PtSn bond. In this same sense, recently published results
regarding crotonaldehyde hydrogenation with SnPt(111) and Pt-
(111) model catalysts showed an increase in the hydrogenation
rate of a platinum-tin alloy with respect to platinum, which is
explained by means of a reduction in the density of states at
EF. This is in agreement with scanning tunneling microscopy
studies that give direct evidence of the mentioned electronic
transfer. A lower activation energy for this reaction using a SnPt
alloy with respect to Pt is also reported.53,54
Acknowledgment. This work has been sponsored by the
Consejo Nacional de Investigaciones Cient´ıficas y Te´cnicas
(CONICET, Argentina), the Agencia Nacional de Promocio´n
Cient´ıfica y Te´cnica (PICT no. 14-04378, Argentina), the
Fundacio´n Antorchas (Argentina), and the LNLS (Brazil) under
project XAS 802/01.
References and Notes
(1) Jacobson, R. L.; Kludsdahl, M. E.; McCoy, C. S.; Davis, R. W.
Proc. - Am. Pet. Inst., DiV. Refin. 1969, 49, 504.
(2) Sinfelt, J. Bimetallic Catalysts, DiscoVery, Concepts and Applica-
tions; Wiley: New York, 1983.
(3) Santori, G.; Casella, M.; Ferretti, O. J. Mol. Catal. A: Chem. 2002,
186, 223.
(4) Siri, G. J.; Casella, M. L.; Santori, G. F.; Ferretti, O. A. Ind. Eng.
Chem. Res. 1997, 36, 4821.
(5) Nichio, N.; Casella, M.; Santori, G.; Ponzi, E.; Ferretti, O. Catal.
Today 2000, 62, 231.
(6) Rodr´ıguez, J. A.; Goodman, D. W. Science 1992, 257, 897.
(7) Chojnacki, T. P.; Schmidt, L. D. J. Catal. 1991, 129, 473.
(8) Hansen, T. W.; Wagner, J. B.; Hansen, P. L.; Dahl, S.; Topsoe,
H.; Jacobsen, C. J. H. Science 2001, 294, 1508.
(9) Prestvik, R.; Totdal, B.; Lyman, C. E.; Holmen, A. J. Catal. 1998,
176, 246.
(10) Lakis, R. E.; Lyman, C. E.; Stenger, H. G. J. Catal. 1995, 154,
261.
(11) Coloma, F.; Sepu´lveda-Escribano, A.; Garcia Fierro, J. L.; Rod-
riguez-Reynoso, F. Appl. Catal., A 1996, 136, 231.
(12) Bazin, D. Top. Catal. 2002, 18, 79.
Besides the existence of “Lewis acid sites”, the presence of
ionic tin promotes a better polarization of the CdO group,
favoring the H2 attack on it and, in this way, leading to a higher
hydrogenation rate and higher selectivity for the unsaturated
alcohol. Thus, the presence of ionic tin seems to have a positive
effect on the increase in the UOL formation rate for the PtSn-
OM and PtSn-BM systems compared to that for PtSn-OM*,
in which ionic tin is not detected.
(13) Alexeev, O.; Gates, B. C. Top. Catal. 2000, 10, 273.
(14) Marcus, M. In EXAFS Spectroscopy: Techniques and Applications;
Teo, B. K., Joy, D. C., Eds.; Plenum Press: New York, 1981; p 181.
(15) Mansour, A. N.; Cook, J. W.; Sayers, D. E. J. Phys. Chem. 1984,
88, 2330.
(16) Cook, J. W., Jr.; Sayers, D. E. J. Appl. Phys. 1981, 52, 5024.
(17) van Zon, J. B. A. D.; Koningsberger, D. C.; van’t Blik, H. F. J.;
Sayers, D. E. J. Chem. Phys. 1985, 12, 5742.
EXAFS results also contribute to extract some important
conclusions with regard to the structure of the active phase.
Taking into account that Sn(0) does not possess intrinsically
hydrogenating properties, it seems reasonable to accept that PtSn
alloys are not active in hydrogenation reactions and participate
only in the isolation of the electronically modified Pt atoms.
η2-type adsorption is favored in PtSn-OM systems, which
present Pt(0) atoms interacting with Sn(II, IV). This could
(18) Lee, P. A.; Citrin, P. H.; Eisenberger, P.; Kincaid, B. M. ReV. Mod.
Phys. 1981, 53, 769.

Characteristics of Pt and PtSn Clusters on SiO2
J. Phys. Chem. B, Vol. 107, No. 41, 2003 11451
(19) Lytle, F. W.; Wei, P. S. P.; Greegor, R. B.; Via, G. H.; Sinfelt, J.
H. J. Chem. Phys. 1979, 70, 4849.
(20) Lytle, F. W. J. Catal. 1976, 43, 376.
(21) Meitzner, G.; Via, G. H.; Lytle, F. W.; Sinfelt, J. H. J. Phys. Chem.
1992, 96, 4960.
(22) Reifsnyder, S. N.; Otten, M. M.; Sayers, D. E.; Lamb, H. H. J.
Phys. Chem. B 1997, 101, 4972.
(23) Brown, M.; Peierls, R. E.; Stern, E. A. Phys. ReV. B 1977, 15,
738.
(24) McMaster, W. H.; Kerr Del Grande, N.; Hubell, J. H. Compilation
of X-ray Cross Sections; National Technical Information Service: Spring-
field, VA.
(25) Santori, G. F.; Casella, M. L.; Siri, G. J.; Adu´riz, H. R.; Ferretti,
O. A. Appl. Catal. A: Gen. 2000, 197, 141.
(26) Merlen, E.; Beccat, P.; Bertolini, G. R.; Deliche`re, P.; Zanier, N.;
Didillon, B. J. Catal. 1996, 159, 178.
(27) Ramaker, D. E.; Mojet, B. L.; Garriga Oostenbrink, M. T.; Miller,
J. T.; Koningsberger, D. C. Phys. Chem. Chem. Phys. 1999, 1, 2293.
(28) Samant, M. G.; Boudart, M. J. Phys. Chem. 1991, 95, 4070.
(29) Ankudinov, A. L.; Rehr, J. J.; Low, J.; Bare, S. R. Phys. ReV. Lett.
2001, 86, 1642.
(30) Asakura, K.; Kubota, T.; Chun, W.; Iwaqsawa, Y.; Otan, K.;
Fujikawa, T. J. Synchrotron Radiat. 1999, 6, 439.
(31) Koningsberger, D. C.; Mojet, B. L.; van Dorssen, G. E.; Ramaker,
D. E. Top. Catal. 2000, 10, 143.
(32) Zabinsky, S. I.; Rehr, J. J.; Ankudinov, A.; Albers, R. C.; Eller,
M. J. Phys. ReV. B 1995, 52, 2995.
(33) Harris, I. R.; Norman, M.; Bryant, A. W. J. Less-Common Met.
1968, 16, 427.
(34) Tronconi, E.; Crisafully, L.; Galvagno, S.; Donato, A.; Neri, G.;
Pietropaolo, R. Ind. Eng. Chem. Res. 1990, 29, 1766.
(35) Gutierrez, L. B.; Ramallo-Lo´pez, J. M.; Irusta, S.; Miro´, E. E.;
Requejo, F. G. J. Phys. Chem. B 2001, 105, 9514.
(36) Greegor, R. B.; Lytle, F. W. J. Catal. 1980, 63, 476.
(37) Khodakov, A.; Barbouth, N.; Oudar, J.; Villain, F.; Bazin, D.;
Dexpert, H.; Schulz, P. J. Phys. Chem. 1997, 101, 776.
(38) Klimenkov, M.; Nepijko, S.; Kuhlenbeck, H.; Ba¨umer, M.; Schlo¨gl,
R.; Freund, H. J. Surf. Sci. 1997, 391, 27.
(39) Ankudinov, A. L.; Rehr, J. J.; Low, J. J.; Bare, S. J. Chem. Phys.
2002, 116, 1911.
(40) Meitzner, G.; Via, G. H.; Lytle, F. W.; Fung, S. C.; Sinfelt, J. H.
J. Phys. Chem. 1988, 92, 2925.
(41) Ellner, M. J. Less-Common Met. 1981, 78, 21.
(42) Borgna, A.; Stagg, S. M.; Resasco, D. E. J. Phys. Chem. B 1998,
102, 5077.
(43) Ferretti, O. A.; Bettega de Pauli, L. C.; Candy, J. P.; Mabillon, G.;
Bournonville, J. P. Stud. Surf. Sci. Catal. 1987, 31, 713.
(44) Ferretti, O. A.; Casella, M. L. Lat. Appl. Res. 1995, 25, 125.
(45) Ankudinov, A. L.; Nesvizhskii, A. I.; Rehr, J. J. Synchrotron Radiat.
2001, 8, 92.
(46) Vaarkamp, M. Pd.D. Thesis, Eindhoven University of Technology,
Eindhoven, The Netherlands, 1993.
(47) Vaarkamp, M.; Mojet, B. L.; Kappers, M. J.; Miller, J. T.;
Koningsberger, D. C. J. Phys. Chem. 1995, 99, 16067.
(48) Stagg, S. M.; Romeo, E.; Padro, C.; Resasco, D. E. J. Catal. 1998,
178, 137.
(49) Roma´n-Mart´ınez, M. C.; Cazorla-Amoro´s, D.; Yamashita, H.; de
Miguel, S.; Scelza, O. A. Langmuir 2000, 16, 1123.
(50) Jeon, Y.; Chen, J.; Croft, M. Phys. ReV. B 1994, 50, 6555.
(51) Rodriguez, J. A.; Chaturvedi, S.; Jirsak, T.; Hrbek, J. J. Chem. Phys.
1998, 109, 4052.
(52) Claus P. Top. Catal. 1998, 5, 51.
(53) Batzill, M.; Beck, D. E.; Koel, B. E. Surf. Sci. 2000, 466, L821.
(54) Jerdev, D. I.; Olivas, A.; Koel, B. E. J. Catal. 2002, 205, 278.