Metal particle size and structure of the metal-support interface of carbon-supported platinum catalysts as determined with EXAFS spectroscopy

Article abstract of DOI:10.1021/jp049905f

The metal particle size and structure of the metal-support interface of platinum supported on Vulcan XC-72 (a commercial catalyst used in platinum fuel-cell electrodes) and on carbon nanofibers (CNF) have been determined with extended X-ray absorption fine structure spectroscopy (EXAFS). The CNF-supported Pt catalysts were synthesized using a homogeneous deposition precipitation (HDP) method. The amount of acidic oxygen groups on the CNF surface was modified by treatment in an inert atmosphere at different temperatures. The average first shell Pt-Pt coordination number (a??5.5) detected in Pt/CNF is much smaller than for Pt/ Vulcan XC-72 (a??8.2). The presence of oxygen-containing groups in the CNF support most probably leads to the stabilization of small Pt particles on the CNF support. A prominent interaction between the metal particles and the support atoms was detected on both kinds of catalysts, which confirms that the metal is in direct contact with the carbon support atoms. After reduction, a long metal-carbon distance around 2.62 Aì? was detected in both Pt/Vulcan XC-72 and Pt/CNF. After evacuation of Pt/CNF at higher temperatures, the distance between support and interfacial metal atoms decreased to 2.02 A. Therefore, the long metal-carbon support distance is ascribed to the presence of atomic chemisorbed hydrogen in the interface between the Pt particles and the carbon support. According to the number of interfacial Pt-C bonds (four), the platinum particles supported on CNF are proposed to be in contact with a prismatic surface of the carbon support, on which oxygen groups have more stable bonds with carbon atoms. Six Pt-C bonds could be detected in the metal-support interface of Pt/Vulcan XC-72 with an even longer carbon shell at 3.62 Aì?, indicating that the metal particles are located on a more carbon-rich surface. This supports a structural model in which the platinum metal particles are epitaxially grown on the (0001) basal surface plane of carbon graphite.

Full text of DOI:10.1021/jp049905f

J. Phys. Chem. B 2004, 108, 18509-18519
18509
Metal Particle Size and Structure of the Metal-Support Interface of Carbon-Supported
Platinum Catalysts as Determined with EXAFS Spectroscopy
Yihua Zhang,† Marjolein L. Toebes,† Ad van der Eerden,† William E. O’Grady,‡
Krijn P. de Jong,† and Diek C. Koningsberger*,†
Department of Inorganic Chemistry and Catalysis, Debye Institute, Utrecht UniVersity, Sorbonnelaan 16,
3584 CA, Utrecht, The Netherlands and Chemistry DiVision, NaVal Research Laboratory,
Washington, D.C. 20375
ReceiVed: January 8, 2004; In Final Form: July 12, 2004
The metal particle size and structure of the metal-support interface of platinum supported on Vulcan XC-72
(a commercial catalyst used in platinum fuel-cell electrodes) and on carbon nanofibers (CNF) have been
determined with extended X-ray absorption fine structure spectroscopy (EXAFS). The CNF-supported Pt
catalysts were synthesized using a homogeneous deposition precipitation (HDP) method. The amount of acidic
oxygen groups on the CNF surface was modified by treatment in an inert atmosphere at different temperatures.
The average first shell Pt-Pt coordination number (∼5.5) detected in Pt/CNF is much smaller than for Pt/
Vulcan XC-72 (∼8.2). The presence of oxygen-containing groups in the CNF support most probably leads
to the stabilization of small Pt particles on the CNF support. A prominent interaction between the metal
particles and the support atoms was detected on both kinds of catalysts, which confirms that the metal is in
direct contact with the carbon support atoms. After reduction, a long metal-carbon distance around 2.62 Å
was detected in both Pt/Vulcan XC-72 and Pt/CNF. After evacuation of Pt/CNF at higher temperatures, the
distance between support and interfacial metal atoms decreased to 2.02 Å. Therefore, the long metal-carbon
support distance is ascribed to the presence of atomic chemisorbed hydrogen in the interface between the Pt
particles and the carbon support. According to the number of interfacial Pt-C bonds (four), the platinum
particles supported on CNF are proposed to be in contact with a prismatic surface of the carbon support, on
which oxygen groups have more stable bonds with carbon atoms. Six Pt-C bonds could be detected in the
metal-support interface of Pt/Vulcan XC-72 with an even longer carbon shell at 3.62 Å, indicating that the
metal particles are located on a more carbon-rich surface. This supports a structural model in which the
platinum metal particles are epitaxially grown on the (0001) basal surface plane of carbon graphite.
1. Introduction
been extensively studied,20-22 a fundamental understanding of
the nature of the interactions between the carbon support and
Carbon as a support for small metal particles has been used
for many decades in electrochemistry. There is gaining interest
in Pt/C electrochemistry1-3 and catalysis.4-11 It has been shown
that precious metals supported on carbon nanofibers, which can
be optimized by modification of the surface groups, provide
highly active and selective catalysts in hydrogenation reac-
tions.12,13 It has been well established that the physicochemical
properties of the carbon support can have a dramatic effect on
the catalytic properties of the supported metal.14,15 The work
of Yacaman16 has shown that there is a strong interaction
between small Pt particles and a graphite surface. Furthermore,
it has been demonstrated that similar interactions occur when
high-surface-area platinum is supported on carbon black.17
Recent XAFS studies18,19 have revealed more structural details
about Pt fuel-cell electrodes. However, the structure of the
metal-support interface for carbon-supported Pt particles is not
as well studied as for metal particles supported on conventional
oxidic or zeolite supports. Whereas the influence of oxidic
(macroporous and crystalline zeolitic) supports on the catalytic
and electronic properties of the supported metal particles has
the dispersed metal is lacking. Therefore, it would be of great
importance to gain knowledge on the structure of the metal-
support interface in order to further understand the metal-
support interactions in carbon-supported metal catalysts.
EXAFS has been proven to be a unique tool for studying the
structure of the metal particles and metal-support interface in
highly dispersed supported metal catalysts. Detailed structural
information has been obtained for Rh/Al2O3, Rh/TiO2, Ir/MgO,
Ir/Al2O3, Pt/Al2O3, Pt/LTL, and Pt/Y zeolite.20,23-29 The EXAFS
results show metal-oxygen interactions characterized by both
short (∼2.10-2.20 Å) and long (∼2.5-2.8 Å) distances.26 By
combining the results of temperature-programmed desorption
(TPD) and XAFS experiments, Vaarkamp et al.27 proposed that
the long metal-oxygen distance is due to the presence of
hydrogen in the metal-support interface, because evacuation
leads to a shortening of this distance. Further proof for this
proposal was obtained from the XAFS experiments carried out
on Ir/Al2O3 catalysts.26 The long metal-oxygen distance was
shown to be reversible: Admission of hydrogen after the
evacuation treatment restored the originally observed long
metal-oxygen distance. Furthermore, the authors found that,
when the reduction temperature is raised from 300 to 450 °C,
the long metal-oxygen distance irreversibly disappeared; only
* To whom correspondence should be addressed. Phone: +31-30-
2536766. Fax: +31-30-2511027. E-mail: d.c.koningsberger@chem.uu.nl.
† Debye Institute.
‡ Naval Research Laboratory.
10.1021/jp049905f CCC: $27.50 © 2004 American Chemical Society
Published on Web 11/06/2004

18510 J. Phys. Chem. B, Vol. 108, No. 48, 2004
Zhang et al.
TABLE 1: Carbon-Supported Platinum Catalysts
operated at 200 kV. Samples were prepared by suspending the
fibers in ethanol under ultrasonic vibration. Some drops of the
thus-produced suspension were brought onto a holey carbon film
on a copper grid.
hydrogen chemisorption
Pt loading TEM
sample name
(wt %)
(nm)
Htotal/Pt
Hirr/Pt
Pt/Vulcan XC-72
Pt/CNF200
Pt/CNF500
10
3.4
3.4
2.0
1.4
1.5
Hydrogen chemisorption measurements were performed using
a Micromeritics ASAP 2010C. Before the chemisorption
measurements, each sample was dried in He at 120 °C for 1 h
and reduced in flowing H2 (flow rate ) 50 mL/min STP) at
200 °C for 2 h, heating rate 5 °C /min. After reduction, the
catalyst was degassed for 2 h at 10-1 Pa at the reduction
temperature in order to eliminate chemisorbed hydrogen and
water. The first (total amount of chemisorbed hydrogen) and
second (amount of reversible chemisorbed hydrogen) isotherms
were measured at 35 °C. The H/Pt ratios are based on the
adsorbed amounts at zero pressure found by extrapolation of
the linear part of the isotherm.
2.3. XAFS Spectroscopy. XAFS Data Collection. The X-ray
absorption data at the Pt L3 edge of the Pt/Vulcan XC-72 sample
were measured in the transmission mode on beam line X11-A
of the National Synchrotron Light Source (NSLS), Brookhaven
National Laboratory, U.S.A. The station was operated with a
double crystal monochromator, Si(111), detuned to 20% inten-
sity to minimize the presence of higher harmonics. The X-ray
absorption spectra (both Pt L3- and L2-edge) of the Pt/CNF
samples were collected at HASYLAB synchrotron (Beamline
X1.1) in Hamburg, Germany. The monochromator was detuned
to 50% of maximum intensity to avoid higher harmonics present
in the X-ray beam. All measurements were done in transmission
using ion chambers filled with a mixture of Ar and N2 to have
an X-ray absorbance of 20% in the first and 80% in the second
chamber.
0.79
0.74
0.37
0.33
a short metal-oxygen distance was observed in the presence
of hydrogen. These results led to the idea that the presence of
surface hydroxyl groups is a prerequisite for the movement of
hydrogen between the metal particle and the support, thereby
leading to the long metal-oxygen interfacial distance. However,
a detailed understanding of the related mechanism is still lacking.
In this paper, the particle size and the structure of the metal-
support interface of two different Pt/C catalysts has been
determined with EXAFS. First, a CNF-supported Pt catalyst with
different amounts of oxygen groups on the carbon surface was
investigated. The oxygen-containing groups have a strong
influence on the hydrogenation properties of this Pt/CNF
catalyst.30,31 The Pt particles were surprisingly stable, and TEM
showed Pt particles with sizes in the range 1.2-1.4 nm. As a
comparison, a commercial Pt/Vulcan XC-72 catalyst used in
fuel cell applications with metal particle sizes in the range 1.8-
2.0 nm was also studied. The changes in the metal particle and
metal-support interface structure induced by adsorption of
hydrogen have been discussed. The results of our XAFS study
on the Pt/CNF catalysts show clear evidence for an interfacial
platinum-carbon distance of around 2.65 Å due to the presence
of hydrogen in the platinum-carbon interface. The interfacial
Pt atoms of the CNF-supported Pt particles are proposed to be
in contact with a prismatic surface of the support, on which
oxygen groups have more stable bonds with carbon atoms.
However, the platinum atoms in the interface of Pt/Vulcan XC-
72 sample have a (111) orientation in contact with a (0001)
basal surface plane of the carbon.
The samples were pressed into self-supporting wafers (cal-
culated to have an absorbency of 2.5) and mounted in a
controlled-atmosphere cell operated at 1 atm.34 All EXAFS
experiments were carried out at liquid nitrogen temperatures.
The Pt/Vulcan XC-72 sample was treated in flowing hydrogen
at room temperature (RT) (indicated by REDRT), and the
EXAFS spectra were collected under 1 atm of hydrogen. The
sample was further evacuated for 1 h at RT (indicated by
VACRT) followed by EXAFS measurements maintaining a
vacuum better than 10-3 Pa.
2. Experimental Section
2.1. Synthesis of Carbon-Supported Platinum Catalysts.
Pt/Vulcan XC-72 is a commercial catalyst used in fuel cell
applications. The sample contains 10 wt % Pt on Vulcan XC-
72.
The Pt/CNF samples were in situ reduced at 200 °C for 1 h
in flowing hydrogen and cooled in hydrogen (treatment denoted
as RED), and the EXAFS data were measured under 1 atm of
hydrogen. Subsequently, the samples were evacuated at 50, 100,
150, and 200 °C for 30 min (treatments denoted as VAC50,
VAC100, VAC150, and VAC200, respectively). After each
evacuation, the EXAFS data were collected maintaining a
vacuum better than the 10-3 Pa sample.
The Pt L3-edge of Pt foil and [Rb2Pt(CN)4]‚(1.5H2O) were
also measured and used to obtain phase shifts and backscattering
amplitudes of the Pt-Pt and Pt-C absorber-backscatterer pairs,
respectively.
EXAFS Data Analysis. Extraction of the EXAFS data from
the measured absorption spectra was performed with the XDAP
code.35 Three scans were averaged, and the pre-edge was
subtracted using a modified Victoreen curve. The background
was subtracted employing cubic spline routines with a continu-
ously adjustable smooth parameter. Normalization was per-
formed by dividing the data by the height of the absorption
edge at 50 eV.
The synthesis of CNF has been reported elsewhere.32 Platinum
(3.4 wt %) was deposited on the fibers using an HDP method.
To an acidified suspension (pH ) 3) of 5 g CNF in 250 mL
demiwater heated to 90 °C, 0.41 g of urea (Acros) and 0.52 g
of Pt (NH3)4(NO3)2 were added under vigorous stirring. The
pH of the slurry was monitored continuously. After 18 h, the
loaded CNF were filtered and washed thoroughly with de-
mineralized water (conductivity), dried at 120 °C, and reduced
with H2 at 200 °C for 1 h (heating rate ) 5 °C /min).
The freshly reduced catalysts contained oxygen-containing
groups with O/C atomic ratio of 0.038 in the CNF support as
determined by XPS,33 further indicated by Pt/CNF200. To
change the concentration of oxygen-containing groups on the
CNF surface, samples of the freshly reduced catalyst were heat-
treated in a nitrogen atmosphere for 2 h to vary the amount of
oxygen-containing groups. In this study also, a catalyst treated
at 500 °C was investigated (O/C atomic ratio of 0.026,33 further
indicated as Pt/CNF500). The catalyst samples together with
their identification codes are listed in Table 1.
2.2. Catalyst Characterization. The average diameter of the
platinum particles in Pt/Vulcan XC-72 was 1.8-2.0 nm as
determined by electron microscopy. The CNF-supported plati-
num catalysts were examined in a Philips Technic 20 FEG TEM
Data analysis was performed by multiple shell fitting in
R-space using the EXAFS data-analysis program XDAP, which
allows one to minimize the residuals between both the magni-

Carbon-Supported Platinum Catalysts
J. Phys. Chem. B, Vol. 108, No. 48, 2004 18511
TABLE 2:
(a) Parameters Used in the Calculations
for the Generation of Theoretical Refs
σ2 [Å2]
S0
Vr [eV] Vi [eV]
2
atom pair
potential
Pt-Pt
Pt-C
Dirac-Hara
Hedin-Linquist 0.00200 0.93
0.00234 0.82
-4.2
-10.4
3.0
3.0
(b) Best Fit Result for the Exptl Pt Foil and Rb2Pt(CN)4‚(1.5H2O)
Data Using the Theoretical Refsa
atom pair
Pt-Pt
Pt-C
N
R (Å)
∆σ (Å2)
E0 (eV)
11.7
3.9
2.77
1.98
-0.00004
0.00001
0.08
0.02
a Fits were performed in R-space, k3-weighted. Pt-foil: ∆k ) 2.5-
15 Å-1; ∆R ) 1.5-3.2 Å. Rb2Pt(CN)4‚(1.5 H2O): ∆k ) 2.5-12.9
Å-1; ∆R ) 1.0-2.0 Å.
tude and the imaginary part of the Fourier transforms of the
data and the fit. R-space fitting has significant advantages
compared to the usually applied fitting in k-space and is
extensively discussed in a recent paper by Koningsberger et
al.36 The difference file technique was applied together with
phase- and amplitude-corrected Fourier transforms to resolve
the different contributions in the EXAFS data. The difference
file technique allows one to optimize each individual contribu-
tion with respect to the other contributions present in the EXAFS
spectrum.36 Both high Z (Pt-Pt) and low Z (Pt-C) contributions
are present in the EXAFS data collected on carbon-supported
platinum. A k3 weighting emphasizes the high Z contributions
to the spectrum, while k1 weighting focuses on the low Z. In
this study, the EXAFS fits have been checked by applying k1
and k3 weightings in order to be certain that the results are the
same for all weightings.37
Theoretical phase shifts and backscattering amplitudes for
the Pt-Pt and Pt-C absorber-scattering pairs were used in
EXAFS data analysis, which were generated utilizing the FEFF7
code with the parameters listed in Table 2a. The theoretical
references were calibrated with the help of the experimental
data of Pt foil and [Rb2Pt(CN)4]‚(1.5H2O) employing the
procedure reported in ref 36. Table 2b gives the best fit result
for the experimental Pt foil and Rb2Pt(CN)4‚(1.5H2O) data using
the theoretical references. The fitted EXAFS coordination
parameters correspond with the crystallographic values for Pt
foil (N ) 12 and R ) 2.77 Å) and for Rb2Pt(CN)4‚(1.5H2O) (N
) 4 and R ) 1.99 Å).
Figure 1. TEM images of (part a) Pt/CNF200 and (part b) Pt/CNF500.
on CNF in this study are very thermally stable. Hydrogen
chemisorption shows a small decrease in Htotal/Pt from 0.79 on
Pt/CNF200 to 0.74 on Pt/CNF500.
EXAFS Data Analysis. The experimental EXAFS data (k1
weighted, ∆k ) 2-14 Å-1) of Pt/CNF200 and Pt/CNF500 are
plotted in Figure 2a and b, respectively. The signal-to-noise ratio
at k ) 4.4 Å-1 amounts to approximately 150 for both
Pt/CNF200 and Pt/CNF500, showing the excellent data quality.
The amplitude of oscillations decreased with the nodes shifted
to higher k values upon increasing the evacuation temperature.
These spectral changes are due to variations in both the structure
of the metal particle and the metal-support interface induced
by the desorption of chemisorbed hydrogen. It can clearly be
seen that the temperature dependence of the changes induced
by the desorption of hydrogen is different for both samples.
Figure 3a,b display the corresponding Fourier transforms (k1
weighted, ∆k ) 3-14 Å-1) of the EXAFS data of Pt/CNF200
and Pt/CNF500 for the different treatments (RED, VAC100,
and VAC200). Figure 3 shows that, with increasing evacuation
temperature, the spectra (both the magnitude and the imaginary
part of the Fourier transform) shift to lower values of R. The
amplitude of the contributions at R < 2 Å, which includes
AXAFS and Pt-C(O) contributions, is enhanced with the trend
in the opposite direction for the range ∼2-3 Å. Also, the Fourier
transforms show clearly that the temperature dependence of the
changes induced by desorption of hydrogen is different for both
samples.
3. Results
3.1. Pt/CNF. TEM and Hydrogen Chemisorption. The
physicochemical properties of carbon nanofibers employed in
this paper have been reported elsewhere.32 The CNF were of
the fishbone type, and an average fiber diameter of 25 nm was
derived from SEM results. The XPS results show that O/C
atomic ratios of 0.038 and 0.026 were determined in the freshly
reduced catalyst (Pt/CNF200) and the catalyst treated in inert
atmosphere at 500 °C (Pt/CNF500), respectively. The metal
loading of the CNF-supported platinum catalysts as determined
with inductively couple plasma (ICP) analysis amounts to 3.4
wt %. In Figure 1a,b, two TEM images of Pt/CNF200 and Pt/
CNF500 are shown. The platinum particles appear as dark dots
with a very narrow size distribution on the surface of the CNF.
Average particle sizes of 1.4 and 1.5 nm were obtained from
the TEM images for Pt/CNF200 and Pt/CNF500, respectively,
as concluded in Table 1. Pt/CNF500 presents a similar size
distribution to Pt/CNF200 without any treatment in inert
atmosphere. This implies that platinum nanoparticles supported

18512 J. Phys. Chem. B, Vol. 108, No. 48, 2004
Zhang et al.
Figure 3. The Fourier transforms (k1 weighted, ∆k ) 3-14 Å-1) of
(part a) Pt/CNF200 and (part b) Pt/CNF500. Solid lines, RED; dotted
lines, VAC100; and dashed line, VAC200.
Figure 2. The experimental EXAFS data (k1 weighted, ∆k ) 2-14
Å-1) of (part a) Pt/CNF200 and (part b) Pt/CNF500. Solid lines, RED;
dotted lines, VAC100; and dashed line, VAC200.
TABLE 3: Fit Parameters of EXAFS Spectraa and
Variances for Model Spectra of Pt/CNF200
Data analysis of the CNF-supported platinum catalysts was
performed by multiple shell fitting in R space in the range 1.6
< R < 3.2 Å. Pt and C/O backscatterers were identified using
the difference file technique. The EXAFS coordination param-
eters of Pt/CNF200 and Pt/CNF500 are given in Tables 3 and
4, respectively. The Pt-Pt first shell coordination numbers
around 5.5 point to extremely small Pt particles with average
particle sizes of about 1.1 nm. Because it is very difficult to
distinguish between C and O backscatterers,38 no attempt was
made in this study to separate the Pt-C and Pt-O contributions.
Figure 4 illustrates the comparison between the Fourier trans-
form of the experimental EXAFS and best fit of Pt/CNF200.
The results for Pt/CNF500 are of comparable quality.
The decrease in Pt-Pt coordination distances and increase
in Debye-Waller factors upon evacuation at increasing tem-
peratures are displayed in Figure 5a,b. After desorption of
hydrogen at low temperature (50 °C), the first metal shell
distance decreases with 0.06 Å compared to the sample in an
H2 atmosphere. After evacuation at 150 and 200 °C, a further
decrease of 0.01 Å was observed. Meanwhile, the Debye-
Waller factor increased from 1 × 10-3 to 7 × 10-3 Å2.
Figure 6a,b shows the Fourier transform of the difference
file of the Pt-C(O) contributions (raw data minus calculated
Pt-Pt contribution) and of the fit Pt-C(O) for both Pt/CNF200
and Pt/CNF500. Removal of the dominant Pt-Pt contributions
makes the changes in the interfacial structure more clearly
visible. The deviations between the Fourier transforms of the
difference file and the Fourier transforms of the calculated model
spectra at low values of R are due to the presence of atomic
k1 - variance
R
(Å,
∆σ2
∆E0
(%)
N
(10-3 Å2, (eV,
treatment scatter ((10%) (0.02 Å) (5%) (10%) im.
abs.
RED
Pt
C(O)L
C(O)S
5.2
1.6
0.7
5.6
0.9
1.1
5.5
0.6
1.4
5.6
1.5
5.4
2.0
2.77
2.65
2.02
2.71
2.59
2.01
2.70
2.51
2.00
2.70
2.02
2.69
2.02
1.1
2.2
2.0
3.9
4.1
3.8
5.3
5.0
4.5
5.5
6.6
6.6
6.7
2.1 0.39 0.14
-0.3
1.2
VAC50 Pt
C(O)L
C(O)S
VAC100 Pt
3.0 0.10 0.04
2.6
5.3
4.0 0.12 0.09
6.3
5.4
3.4 0.28 0.15
1.9
3.8 0.19 0.11
1.5
C(O)L
C(O)S
VAC150 Pt
C(O)S
VAC200 Pt
C(O)S
a∆k ) 3-14 Å-1; ∆R ) 1.6-3.2 Å.
XAFS. After reduction (RED) for both samples, a long distance
Pt-C(O)L (around 2.65 Å) was detected together with a very
small contribution of a much shorter distance Pt-C(O)S (around
2.02 Å). When the evacuation temperature is increased, the
contribution of the long distance decreases with a simultaneous
increase of the contribution of the short distance. The disap-
pearance of the long Pt-C(O)L distance occurs at a lower
evacuation temperature for the Pt/CNF500 than for the Pt/
CNF200 sample.
The long Pt-C(O)L distance of approximately 2.60 Å present
after reduction is in accordance with previous reports on EXAFS
studies over zeolite and/or alumina-supported metal particles

Carbon-Supported Platinum Catalysts
J. Phys. Chem. B, Vol. 108, No. 48, 2004 18513
TABLE 4: Fit Parameters of EXAFS Spectraa and
Variances for Model Spectra of Pt/CNF500
k1 - variance
R
(Å,
∆σ2
∆E0
(%)
N
(10-3 Å2, (eV,
treatment scatter ((10%) (0.02 Å) (5%) (10%) im.
abs.
RED
Pt
C(O)L
C(O)S
5.5
2.0
0.3
5.4
0.7
1.6
5.4
2.1
5.3
2.2
5.2
2.3
2.76
2.62
2.01
2.70
2.55
2.02
2.69
2.02
2.68
2.02
2.68
2.02
1.6
1.4
1.3
5.7
2.3
1.8
6.6
2.3
7.0
2.7
7.1
3.1
1.3
1.5
3.2
3.5
5.3
3.8
4.2
2.3
3.7
1.6
3.9
1.6
0.58 0.10
VAC50 Pt
C(O)L
C(O)S
VAC100 Pt
C(O)S
VAC150 Pt
C(O)S
VAC200 Pt
C(O)S
0.34 0.12
0.35 0.15
0.21 0.08
0.21 0.07
a∆k ) 3-14 Å-1; ∆R ) 1.6-3.2 Å.
Figure 5. (part a) Pt-Pt coordination distance and (part b) Debye-
Waller factor as a function of sample treatment. Pt/CNF200, solid line
with closed squares (9), Pt/CNF500, dotted line with open squares (0).
As comparison data of an acidic Pt/Cl-Al2O3, line with closed triangles
(2); and basic Pt/K-Al2O3, dashed-dotted line with open triangles
(4).46 The first points on the left refer to RED.
Figure 4. The Fourier transforms (k1 weighted, ∆k ) 3-14 Å-1) of
the experimental spectra (solid line) and total fit (dotted line) of Pt/
CNF200 after reduction (RED).
and is due to the presence of interfacial hydrogen.27,28,39 It has
to be noted that the total Pt-C(O) coordination number stays
constant around 2.0 in Pt/CNF200 and 2.3 in Pt/CNF500.
3.2. Pt/Vulcan XC-72. EXAFS Data Analysis. In Figure 7a,b,
the raw EXAFS data and the corresponding Fourier transform
spectra are displayed, characterizing the sample Pt/Vulcan XC-
72 after reduction at RT (REDRT) and after evacuation at RT
(VACRT). The EXAFS data have an excellent quality (signal-
to-noise ratio of about 90 at k ) 4 Å-1). The differences for
2.5 < R < 3 Å between the Fourier transforms of REDRT and
VACRT (Figure 7b) are caused by the removal of chemisorbed
hydrogen and are mostly due to an increase in the Debye-
Waller factor of the first shell Pt-Pt coordination of the VACRT
sample.
Comparable results are obtained for the fit of the EXAFS data
after evacuation at RT (VACRT).
Figure 9a,b displays the Fourier transforms (k1 weighted, ∆k
) 3.5-9.5 Å-1) of the total Pt-C difference file (raw EXAFS
data minus fitted Pt-Pt contributions [Pt-Pt1 + Pt-Pt2] and
calculated model (Pt-C1) + (Pt-C2) contributions. It can be
clearly seen from the difference file that additional scatterers
must be included. The coordination numbers for both shells are
around 2.3 on REDRT and 2.6 on VACRT. However, hardly
any difference in the structure of the interface between the Pt
particles and the carbon support is observed after evacuation at
RT.
The results of the EXAFS data analysis are given in Table
5. The Pt-Pt coordination numbers around 8.2 points to metal
particles of about 1.6 nm, much larger than for the Pt/CNF
samples. After evacuation at RT, a larger Debye-Waller factor
was obtained for the Pt-Pt first shell contribution. Even for
these larger Pt particles an interfacial Pt-C coordination could
be detected with a first and second Pt-C shell distance of ∼2.6
Å and ∼3.6 Å, respectively. The analysis of especially the
second Pt-C shell could only be reliably performed by including
in the fit procedure the second Pt-Pt shell, because this shell
has a considerable overlap with the second Pt-C shell. Figure
4. Discussion
4.1. Pt Particle Size. The EXAFS first Pt-Pt shell coordina-
tion number is about 5.5 in Pt/CNF and 8.2 in Pt/Vulcan XC-
72. By using a computer program40 to model spherical particles,
it can be calculated that the EXAFS coordination numbers
correspond to small Pt particles with sizes around 1 nm in Pt/
CNF and 1.6 nm in Pt/Vulcan XC-72. Larger Pt metal particle
sizes were determined by TEM of around 1.5 nm in Pt/CNF
and 2 nm in Pt/Vulcan XC-72, respectively. It has been
demonstrated before that the determination of metal particle size
using EXAFS coordination numbers is more accurate than the
results from TEM if the average metal particle size is smaller
than 2 nm. This due to the fact that TEM may not detect the
8 gives the Fourier transforms (k1 weighted, ∆k ) 3.5-14 Å-1
)
of the experimental data of the REDRT sample and of the
calculated total fit using the parameters listed in Table 5.

18514 J. Phys. Chem. B, Vol. 108, No. 48, 2004
Zhang et al.
Figure 6. (part a) The Fourier transforms (k1 weighted, ∆k ) 3-11 Å-1, Pt-C phase-corrected) of the Pt-C(O) difference file (raw spectra minus
fitted Pt-Pt contribution, solid line) and fitted Pt-C(O)L + Pt-C(O)S contributions (dotted line) of Pt/CNF200. Treatments are RED, VAC50,
VAC100, VAC150, and VAC200, respectively. (part b) The Fourier transforms (k1 weighted, ∆k ) 3-11 Å-1, Pt-C phase-corrected) of the
Pt-C(O) difference file (raw spectra minus fitted Pt-Pt contribution, solid line) and fitted Pt-C(O)L + Pt-C(O)S contributions (dotted line) of
Pt/CNF500. Treatments are RED, VAC50, VAC100, VAC150, and VAC200, respectively.
very small metal particles present in the small size end of the
particle size distribution curve.40
It is obvious that the presence of oxygen-containing groups
leads to a stabilization of the Pt particles. The average particle
size for Pt/CNF is not only much smaller than for Pt/Vulcan
XC-72, but the Pt-Pt first shell coordination numbers of 5.5
in Pt/CNF500 and 5.2 in Pt/CNF200 indicate that no sintering
of the Pt particles occurs even after treatment at 500 °C in

Carbon-Supported Platinum Catalysts
J. Phys. Chem. B, Vol. 108, No. 48, 2004 18515
Figure 8. The Fourier transforms (k1 weighted, ∆k ) 3.5-14 Å-1) of
the experimental spectra (solid line) and total fit (dotted line) of REDRT.
Figure 7. (part a) Experimental EXAFS data and (b) corresponding
Fourier transforms (k1 weighted, ∆k ) 3-14 Å-1) of Pt/Vulcan XC-
72 (solid line, REDRT; dotted line, VACRT).
TABLE 5: Fit Parameters of EXAFS Spectraa and
Variances for Model Spectra of Pt/Vulcan XC-72
k1 - variance
R
(Å,
∆σ2
∆E0
(%)
N
(10-3 Å2, (eV,
treatment scatter ((10%) (0.02 Å) (5%) (10%) im.
abs.
REDRT Pt
8.4
2.3
2.3
5.0
8.2
2.5
2.7
5.1
2.76
2.62
3.62
3.90
2.76
2.58
3.62
3.90
1.1
7.0
9.0
1.7
1.6
8.0
10.0
2.5
-1.6 0.34 0.09
C
C
Pt
-0.2
-7.9
-5.6
VACRT Pt
-2.0 0.27 0.08
C
C
Pt
4.9
-9.4
-6.2
a∆k ) 3.5-14 Å-1; ∆R ) 1.6-4.0 Å.
nitrogen atmosphere. This has been demonstrated to be crucial
for the application of the nanosized Pt metal particles for a
reaction under severe catalytic conditions.41
Figure 9. The Fourier transforms (k1 weighted, ∆k ) 3.5-9.5 Å-1
Pt-C phase corrected) of Pt-C difference file (raw spectra minus fitted
Pt-Pt contributions, solid line) and fitted (Pt-C1) + (Pt-C2) contribu-
tions (dotted line) of (part a) REDRT and (part b) VACRT.
,
4.2. Pt-C Contribution. After reduction, a long Pt-C(O)L
distance around 2.6 Å is observed for both Pt/CNF and
Pt/Vulcan XC-72 catalysts. With increasing evacuation tem-
perature, this long Pt-C(O)L distance disappears for the Pt/
CNF200 and Pt/CNF500 samples. It has to be noted that the
response on the evacuation temperature is different for both
samples: The long Pt-C(O)L distance disappears at a lower
evacuation temperature for Pt/CNF500 (TVAC ) 100 °C) than
for Pt/CNF200 (TVAC ) 150 °C). For the Pt/Vulcan XC-72
sample, it was not possible to follow the disappearance of the
long Pt-C(O)L distance, because evacuation at higher temper-
atures led to a severe sintering of the Pt particles, making the
detection of this distance impossible. However, after the
evacuation of Pt/Vulcan XC-72 at RT, the long Pt-C(O)L
distance could still be detected, also implying that evacuation
at high temperature is needed to remove this distance. After
evacuation at higher temperatures, only a short Pt-C(O)S
distance is detected around 2.02 Å with approximately the same
total Pt-C(O) coordination number.
As already mentioned in the Results section, it is almost
impossible to distinguish with EXAFS between C and O as the
backscatterer. However, the fact that the same kind of interaction

18516 J. Phys. Chem. B, Vol. 108, No. 48, 2004
Zhang et al.
Pt surface atoms, which are coordinatively unsaturated, and an
increase of Pt atoms in contact with the carbon support, (ii) a
decrease in the Pt-Pt coordination number, and (iii) an increase
in the Pt-Pt Debye-Waller factor.
4.3. Detectability with EXAFS of Interfacial Metal-
Support Contributions. The detection with EXAFS of support
neighbors of metal atoms present in the metal-support interface
is dependent on the metal particle size and morphology, the
structure of the support surface plane, and the quality of the
EXAFS experimental data. The interfacial metal coordination
number as detected with EXAFS is fractional. The fraction of
metal atoms present in the metal-support interface decreases
with increasing metal particle size, finally leading to an
undetectable contribution of the metal-support interface. For
fcc noble metal particles dispersed on high surface area γ-Al2O3
supports with EXAFS noise levels of 0.001, the EXAFS signal
arising from the metal-support interface becomes undetectable
for particles with metal-metal coordination numbers larger than
8.5-9. The Pt-Pt and Pt-OL coordination number found for
Pt/γ-Al2O3 catalysts after reduction at 300 °C is 4.8 and 1.2,
respectively.28 The (fractional) EXAFS Pt-OL coordination of
1.2 translates to a real Pt-OL coordination number of 3,
implying that Pt in the metal-support interface is 3-fold
coordinated by oxygen atoms of the support. However, in this
study for the Pt/CNF samples with a slightly higher average
Pt-Pt coordination around 5.5, the (fractional) Pt-OL coordina-
tion number is in the range 1.6-2.0, whereas for the much larger
Pt particles on Vulcan XC-72 (Pt-Pt around 8.5), an even higher
(fractional) Pt-CL coordination number around 2.5 is detected.
This can only be explained by a different structure of the metal-
support interface for both types of carbon-supported Pt particles
leading to a larger number of neighboring carbon atoms. This
increases the detectability of the contribution of the metal-
support interface for larger Pt metal particles.
Figure 10. The schematic illustration of the changes in interfacial
structure between the metal and the support induced by evacuation.
is detected in the metal-support interface of both types of
carbon supports, combined with the information that no
significant amount of oxygen-containing groups are present on
the Vulcan XC-72 support, may lead to the conclusion that C
is the neighboring atom in the metal-support interface.
The conventional view of the platinum-carbon support
interface would suggest that the Pt atoms are in direct contact
with the carbon support. If this would be the case under all
experimental conditions, the Pt-C bond distance would be a
maximum of 2.09 Å (atomic radius of Pt in Pt metal is 1.38 Å
and C in graphite is 0.71 Å). Only a similarly short distance is
found experimentally after evacuation: 2.02 Å. After reduction,
a Pt-C distance is observed around 2.6 Å. If the value of the
Pt radius is subtracted from the observed value after reduction,
a difference of 1.22 Å is obtained, which, in fact, is much closer
to the bond distance in a C-H bond (1.1 Å). The distance of
2.6 Å detected after reduction is consistent with metal-support
distances found in other precious metal catalysts having oxidic
(macroporous flat and crystalline zeolitic) supports. As described
in detail in the Introduction section, the metal-support distance
for these systems shortens to about 2.1-2.2 Å after evacuation
at higher temperatures and reappears again after admission of
hydrogen. A similar behavior is now observed for the Pt-C
coordination in this study. Therefore, we infer that the long
metal-support distance is a general characteristic feature of
interfaces consisting of hydrogen between interfacial metal
atoms and support neighbors such as carbon atoms of carbon
supports and oxygen anions of metal oxide supports, including
zeolites. It is most logical to assume that close to each other
multiple Pt-C distances exist at the interface of Pt and carbon,
all contributing to the XAFS signal, while various amounts of
interfacial hydrogen is left upon evacuation at increasing
temperatures. The changes in the interfacial structure between
the metal and the CNF support caused by the existence of
hydrogen in the interfacial area are schematically shown in
Figure 10. From experience in hydrogenation catalysis at
elevated pressure and temperature, it appears that Pt/CNF
catalysts are not prone to sintering at the prevailing condi-
tions.13,30,31 Despite the long Pt-C/O distance, the metal-
support interaction is still significant and clearly prevents metal
sintering. The precise nature of the Pt-H-C/O interface is not
fully known and deserves further experimental and theoretical
work.
4.4. Structure of the Metal-Support Interface in Pt/CNF
and Pt/Vulcan XC-72. The structure of layered graphite gives
basically two kinds of surfaces: prismatic (edge) surfaces and
basal plane surfaces. The structure of the metal-support
interface on carbon-supported metal particles depends on the
morphology of the surface on which the metal atoms are
anchored. By assuming a half-spherical shape of the platinum
particles sticking with a particular fcc orientation, it is possible
to derive an expression for the fraction (f) of the number of
interfacial platinum atoms: f ) 4R/d, with R the atomic radius
of Pt and d the diameter of the platinum particle.23 The fraction
is calculated to be 0.35, using d ) 16 Å with R ) 1.38 Å on
Pt/Vulcan-XC72, while it is 0.55, using d ) 10 Å with R )
1.38 Å on Pt/CNF200.
By using the EXAFS Pt-CL coordination number of 2.3
(REDRT) for both the first and second carbon shell at 2.62 and
3.62 Å, respectively, a real interfacial coordination number
around 6.6 can be calculated for both coordination shells. A
model with six nearest and six second nearest Pt-CL neighbors
is possible for the Pt growing epitaxially on the basal (0001)
plane of carbon as shown in Figure 11. The C-C distance of
1.415 Å is very close to the Pt-Pt radius of 1.38 Å, making an
epitaxial interface not unlikely. Using this assumption, we find
that the structure of the metal-support interface can be further
analyzed. With the experimental value of 2.62 Å for the Pt-
CL first shell distance and the C-C bond distance from graphite
1.415 Å, it can be calculated that the Pt atom sits 2.205 Å above
the center of the basal plane. Calculating now the distance to
the second carbon neighbor, we obtain a value of 3.59 Å, which
is 0.03 Å lower than the observed distance of 3.62 Å. For
As already mentioned on Pt/Vulcan XC-72, hardly any
difference in the Pt-C contributions was found between REDRT
and VACRT. However, after evacuation at RT, a small decrease
in the Pt-Pt coordination number, a small increase in the Pt-
Pt Debye-Waller factor, and small increases in the Pt-C
coordination numbers of both carbon neighbor shells are
observed. This can be explained by a slight morphology change
of the Pt particles after desorption of hydrogen at RT. Flattening
out of the Pt particles on the support surface can compensate
the increase in Pt surface free energy because of the removal
of chemisorbed hydrogen. This leads to (i) a lower amount of

Carbon-Supported Platinum Catalysts
J. Phys. Chem. B, Vol. 108, No. 48, 2004 18517
Figure 11. Structural model of Pt supported on Vulcan XC-72; hollow (O) and solid circles (B) refer to the first and second carbon shell, respectively.
EXAFS studies, the limits of accuracy in the bond distance can
be normally calculated to be (0.01 Å,36 showing that the
calculated value is close to the experimental value. This indicates
that the EXAFS coordination parameters are in agreement with
the proposed model for the structure of the metal-support
interface. The results obtained in an earlier EXAFS study of
our group42 on the formation of the metal-support interface
by controlled heating of the precursor [Pt(NH3)4](OH)2 impreg-
nated on γ-Al2O3 in flowing hydrogen fully support the findings
in this study. The EXAFS data obtained after reduction at 180
°C are consistent with a structural model in which small Pt
particles grow with a (111) epitaxy on the (111) surface plane
of the high surface area γ-Al2O3 support. Also, in this case,
two metal-support oxygen distances have been detected at 2.70
and 3.84 Å. Just like for the Pt carbon interface as studied here,
the second shell interfacial distance is a consequence of the
existence of a first shell interfacial distance at 2.70 Å, caused
by the presence of hydrogen in the interface between the Pt
metal particle and the support. Using the (111) epitaxial model,
we can calculate a second shell Pt-O distance of 3.86 Å, which
is almost equal to the value found in the EXAFS analysis.
Compared with Vulcan-XC72, the carbon nanofibers used
in this study have very different physicochemical properties (i.e.,
a fishbone structure, higher surface area, and higher purity).
According to the already-calculated fraction of Pt atoms in the
metal-support interface for the Pt/CNF samples, one interfacial
Pt atom would only feel about four carbon atoms. This suggests
that Pt particles are not located on the basal (0001) support
surface, which is highly symmetrically ordered and will give
more carbon neighbors. Also, it is not likely that the metal
particles are intercalated between two graphite layers, because
the particle size around 10 Å is much larger than the graphite
layer distance of 3.4 Å. Because the introduction of the oxygen-
containing groups seems to be essential for the synthesis of much
smaller Pt particles than on Vulcan, the possible adsorption sites
of oxygen-containing groups on the CNF surface can give some
hints about the interfacial structure between the Pt particles and
the CNF support.43 To investigate the interaction of atomic
oxygen with graphite surfaces, using first-principles total energy
calculations, Incze et al.44 proposed that a zigzag edge surface
of carbon graphite is the most reactive one, followed by armchair
and basal plane surfaces. Oxygen atoms absorbed on the sites
of a zigzag surface are more stable. It is also well-known that
the chemistry and morphology of the prismatic surfaces of
graphite play a major role for the graphite anode performance
in lithium ion cells (e.g., in chemical and electrochemical
reactivity, interaction with the solid electrolyte interphase SEI,
etc.).45 By placing the Pt particles on the zigzag or armchair
surfaces, the interfacial Pt atoms would have only four carbon
neighbors (see Figure 12). This is in excellent agreement with
the EXAFS results. This would also lead to an explanation of
the role which the oxygen-containing groups play in stabilizing
the Pt particles on the CNF support. More research is needed
to fully explain the mechanism of this stabilizing process.
4.5. Influence of Particle Size, Hydrogen Chemisorption,
and Carbon Support Properties on the Electronic Structure
of the Supported Pt Particles. In going from bulk to very small
metal particles, the electronic structure of the metal atoms is
changing because of the dehybridization of the spd metal
orbitals.46 The dehybridization results in increased electron
density (less delocalization) between the atoms in the small
particle, which in turn results in a contraction of the interatomic
distance. At the same time, caused by the loss of neighboring
atoms, the phonon spectrum softens, causing a broader potential
well. This, in turn, leads to a larger disorder. The Pt structural
parameters obtained by the EXAFS analysis after evacuation
are fully consistent with the theoretical and experimental results
presented in the literature.46 The first shell Pt-Pt coordination
as obtained for the large Pt particles supported on Vulcan-XC72
(VACRT) has a distance close to the Pt-Pt distance in bulk Pt
(2.76 vs 2.77 Å) with a slightly larger Debye-Waller factor
(0.0016 Å-1). The much smaller Pt particles supported on CNF
(VAC200) show a considerable contraction (Pt/CNF200, 2.69
Å; Pt/CNF500, 2.68 Å) with a considerable increase in the
Debye-Waller factor (Pt/CNF200, 0.0066 Å-1; Pt/CNF500,
0.0071 Å-1).

18518 J. Phys. Chem. B, Vol. 108, No. 48, 2004
Zhang et al.
Figure 12. Structural model of Pt supported on CNF.
Hydrogen chemisorption is showing the reverse effect by
going from large to small Pt particles. This can be explained
by the fact that hydrogen chemisorption leads to an increase in
the amount of neighboring atoms for the surface Pt. Hydrogen
is now sharing Pt valence electron density in the Pt-H bond,20,22
making the neighboring Pt-Pt bonds weaker. In EXAFS, this
effect is averaged over the whole particle, effectively leading
to an observed larger Pt-Pt distance. The first shell Pt-Pt
coordination in Pt/Vulcan-XC72 after hydrogen chemisorption
(REDRT) is not affected by hydrogen chemisorption (2.76 Å);
however, a small decrease in disorder is observed (0.0011 vs
0.0016 Å2). The effects are much larger for the smaller Pt
particles supported on CNF. Upon hydrogen chemisorption, the
Pt-Pt distance relaxes back close to the value of Pt foil (Pt/
CNF200, 2.77 vs 2.69 Å; Pt/CNF500, 2.76 vs 2.68 Å) with
also a significant decrease in disorder (Pt/CNF200, 0.0011 vs
0.0066 Å2; Pt/CNF500, 0.0016 vs 0.0071 Å2). These results are
amounts of oxygen-containing groups,13,30 whereas the same
activity trends are observed for the hydrogenation of benzal-
dehyde to benzyl alcohol48,49 using Pt or Pd particles supported
on oxides with a high electron richness (more basic) of the
support oxygen atoms.
Indications that the oxygen-containing groups in the CNF
support influence the electronic structure of Pt can be found in
Figure 5. In Figure 5a,b, the Pt-Pt bond distance and the
Debye-Waller factor are plotted as a function of the evacuation
treatment carried out at different temperatures. All data were
measured at liquid nitrogen temperature. As a comparison, the
same kind of data are also shown for acidic Pt/Cl-Al2O3 and
basic Pt/K-Al2O3 catalysts.50 When increasing the evacuation
temperature, a decrease in Pt-Pt bond distance is observed with
the effect highest for the basic oxidic support. An increase in
the Debye-Waller factor is found also with the largest effect
for the basic oxidic support. It can be seen in Figure 5 that the
same trends are observed for the CNF samples (effect largest
for the Pt/CNF500 sample); however, the effect of the oxygen
groups in the CNF supports are smaller. It has been proven
elsewhere21 that the nature of the metal-support interaction
involves a charge rearrangement from the Pt atoms in the
metal-support interface to the center of the Pt cluster when
the support oxygen atoms have a higher electron richness (more
basic). This results in an opposite trend as induced by chemi-
sorption of hydrogen: A smaller equilibrium Pt-Pt distance
and a broadened potential well, resulting in a larger structural
disorder. This explains the different response on evacuation for
the Pt/Cl-Al2O3 and Pt/K-Al2O3 samples and may indicate
that oxygen-containing groups in CNF supports also influence
the electronic properties of supported Pt particles. More research
is needed to further verify the effect of the oxygen-containing
groups of the CNF support on the catalytic and electronic
properties of the Pt particles.
47
in complete agreement with previous reports for Pt/γ-Al2O3
and Ir/γ-Al2O3.26
The third important factor which can influence the electronic
structure of the metal particles are the properties of the support.
The influence of the electron richness (depending on the acid/
base properties) of the support oxygen atoms of oxidic
(macroporous flat and zeolitic) supports on the catalytic and
electronic properties of the supported metal particles has been
extensively studied.20-22 Preliminary catalytic experiments on
the Pt/CNF samples show an increase in activity for Pt-catalyzed
CdC double bond hydrogenation (cyclohexene and toluene
hydrogenation) for CNF supports with increasing amounts of
oxygen-containing groups. The same trend is observed for Pt
particles supported on oxidic supports, which have a low
electron richness (acidic) on the support oxygen atoms. There-
fore, the CNF200 support can be considered to be more acidic
than the CNF500. In this model, the oxygen-containing groups
in the CNF support withdraw electron density from the carbon
atoms, resulting in a lower electron richness of the support
carbon atoms. In analogy with oxidic supports, this may affect
the catalytic and electronic properties of the Pt particles.
However, the already-mentioned catalytic effects are small for
the CNF-supported samples. Larger effects are observed for
oxide-supported Pt particles, where the differences in electron
richness of the support oxygen atoms are larger. Moreover, it
has to be mentioned that a reverse support effect is found for
the hydrogenation of reactants having CdO double bonds. The
activity for hydrogenation of cinnemaldehyde to cinnamyl
alcohol is higher for Pt and Ru supported on CNF with low
5. Conclusion
Very small Pt particles with diameters around 1 nm have been
prepared on CNF supports, which have controllable amounts
of oxygen-containing groups. The Pt particles have high
thermostability. In contrast, much larger Pt particles with sizes
around 1.6 nm have been detected for Pt/Vulcan XC-72, a
commercial fuel cell electrode. After reduction and in the
presence of hydrogen, a long Pt-C distance around 2.6 Å is
observed in the interface between the Pt particles and the carbon
support. Upon evacuation at higher temperatures on Pt/CNF,

Carbon-Supported Platinum Catalysts
J. Phys. Chem. B, Vol. 108, No. 48, 2004 18519
(14) Coloma, F.; Sepu´lveda-Escribano, A.; Rodr´ıguez-Reinoso, F. Appl.
Catal., A 1995. 123, L1.
(15) Bachiller-Baeza, B.; Guerrero-Ruiz, A.; Rodr´ıguez-Ramos, I. Appl.
Catal., A 2000, 192, 289.
(16) Yacaman, M. J.; Dominguez, E.; Manuel, J. J. Catal. 1980, 64,
213.
(17) Bett, J.; Kinoshita, K.; Routsis, K.; Stonehart, P. J. Catal. 1973,
29, 160.
the long distance Pt-C contribution decreased and eventually
disappeared. A much shorter distance at about 2.02 Å can be
observed in a vacuum. Because the same kind of behavior has
been observed in the literature for Pt particles supported on high
surface area oxides, the origin of the long Pt-C distance has
been attributed to the presence of hydrogen in the interface
between metal and support atoms.
(18) Lampitt, R. A.; Carrette, L. P. L.; Hogarth, M. P.; Russell, A. E. J.
Electroanal. Chem. 1999, 460, 80.
A higher Pt-C coordination number was detected in the
metal-support interface of Pt/Vulcan XC-72. A structural model
is presented in which the metal cluster interacts with the carbon-
rich surface of the basal (0001) plane of carbon graphite.
However, on CNF-supported Pt samples, only about four carbon
neighbors were detected per interfacial Pt atom. In this case,
an interface with a zigzag or armchair edge surface is in
agreement with the EXAFS coordination parameters. It has been
argued that modification of the carbon surface with oxygen
groups is crucial for the deposition of small metal particles.
The EXAFS data analysis shows a marked contraction of the
Pt-Pt coordination distance and an increase in the Debye-
Waller factor during evacuation at different temperatures for
the Pt/CNF samples. The response on the evacuation treatment
of the Pt/CNF200 and Pt/CNF500 samples is compared with
that of a Pt/Cl-Al2O3 and Pt/K-Al2O3 catalyst, respectively.
A model has been discussed in which the oxygen-containing
groups of the CNF support can have an influence on the
electronic properties of the supported Pt particles.
(19) Maniguet, S.; Matthew, R. J.; Russell, A. E. J. Phys. Chem. B 2000,
104, 1998.
(20) Mojet, B. L.; Miller, J. T.; Ramaker, D. E.; Koningsberger, D. C.
J. Catal. 1999, 186, 373.
(21) Ramaker, D. E.; de Graaf, J.; van Veen, J. A. R.; Koningsberger,
D. C. J. Catal. 2001, 203, 7.
(22) Koningsberger, D. C.; Oudenhuijzen, M. K.; de Graaf, J.; van
Bokhoven, J. A.; Ramaker, D. E. J. Catal. 2003, 216, 178.
(23) Van Zon, J. D.; Koningsberger, D. C.; van Blik, H. J.; Sayers, D.
E. J. Chem. Phys. 1985, 82, 5742.
(24) Martens, H. A.; Prins, R.; Zandbergen, H.; Koningsberger, D. C.
J. Phys. Chem. 1988, 92, 1903.
(25) van Zon, F. B. M.; Maloney, S. D.; Gates, B. C.; Koningsberger,
D. C. J. Phys. Chem. 1993, 115, 10317.
(26) Koningsberger, D. C.; van Zon, F. B. M.; Vaarkamp, M.; Mun˜oz-
Paez, A. X-ray Absorption Fine Structure for Catalysts and Surfaces;
Iwasawa, Y., Ed.; Tokyo, 1995; p 257.
(27) Vaarkamp, M.; Modica, F. S.; Miller, J. T.; Koningsberger, D. C.
J. Catal. 1993, 144, 611.
(28) Vaarkamp, M.; Miller, J. T.; Modica, F. S.; Koningsberger, D. C.
J. Catal. 1996, 163, 294.
(29) de Graaf, J.; van Dillen, A. J.; de Jong, K. P.; Koningsberger, D.
C. J. Catal. 2001, 203, 307.
(30) Toebes, M. L.; Zhang, Y.; Ha´jek, J.; Bitter, J. H.; van Dillen, A.
J.; Murzin, D. Y.; Koningsberger, D. C.; de Jong, K. P. J. Catal. 2004,
226, 215.
(31) Toebes, M. L.; Nijhuis, T. A.; Ha´jek, J.; Bitter, J. H.; van Dillen,
A. J.; Murzin, D. Y.; de Jong, K. P. Submitted for publication.
(32) Toebes, M. L.; Bitter, J. H.; van Dillen, A. J.; de Jong, K. P. Catal.
Today 2002, 76, 33.
(33) Toebes, M. L.; van Heeswijk, J. M. P.; Bitter, J. H.; van Dillen, A.
J.; de Jong, K. P. Carbon 2004, 42, 307.
(34) Vaarkamp, M.; Mojet, B. L.; Modica, F. S.; Miller, J. T.;
Koningsberger, D. C. J. Phys. Chem. 1995, 99, 16067.
(35) Vaarkamp, M.; Linders, J. C.; Koningsberger, D. C. Physica B 1995,
208/209, 159.
Acknowledgment. The authors thank C. van der Spek for
performing the TEM measurements. The synchrotron labora-
tories of HASYLAB (Hamburg, Germany) and NSLS (Brook-
haven) are acknowledged for the allocated beamtime. This
research project is supported by the Council for Chemical
Sciences of The Netherlands Organization for Scientific Re-
search with financial aid from The Netherlands Technology
Foundation (CW/STW 349-5357).
(36) Koningsberger, D. C.; Mojet, B. L.; van Dorssen, G. E.; Ramaker,
D. E. Top. Catal. 2000, 10, 143.
References and Notes
(37) Ros, T. G.; Keller, D. E.; van Dillen, A. J.; Geus, J. W.;
Koningsberger, D. C. J. Catal. 2002, 211, 85.
(1) Amine, K.; Yasuda, K.; Takenaka, H. Ann. Chim. (Paris) 1998,
23, 331.
(38) Pandya, K. I.; Koningsberger, D. C. Physica B 1989, 158, 386.
(39) Koningsberger, D. C.; Vaarkamp, M. Physica B 1995, 209, 633.
(40) de Graaf, J. Ph.D. Thesis, Utrecht University, Utrecht, The
Netherlands, 2001.
(41) Joo, S. H.; Choi, S. J.; Oh, I.; Kwak, J.; Terasaki, O.; Ryoo, R.
Nature 2001, 412, 169.
(42) Mun˜oz-Paez, A.; Koningsberger, D. C. J. Phys. Chem. 1995, 99,
4193.
(43) Ros, T. G. Ph.D. Thesis, Utrecht University, Utrecht, The Neth-
erlands, 2002.
(2) Pattabiraman, R. Appl. Catal., A 1997, 153, 9.
(3) Torre, T.; Arico`, A. S.; Alderucci, V.; Antonucci, V.; Giordano,
N. Appl. Catal., A 1994, 114, 257.
(4) Raro´g-Pilecka, W.; Szmigiel, D.; Kowalczyk, Z.; Jodzis, S.;
Zielinski, J. J. Catal. 2003, 218, 465.
(5) Pe´rez-Cadenas, A. F.; Moreno-Castilla, C.; Maldonado-Ho´dar, F.
J.; Fierro, J. L. G. J. Catal. 2003, 217, 30.
(6) Bessel, C. A.; Laubernds, K.; Rodriguez, N. M.; Baker, R. T. K. J.
Phys. Chem. B 2001, 105, 1115.
(7) Pham-Huu, C.; Keller, N.; Charbonniere, L. J.; Ziessel, R.; Ledoux,
M. J. Chem. Commun. 2000, 1871.
(8) Baker, R. T. K.; Laubernds, K.; Wootsch, A.; Paa´l, Z. J. Catal.
2000, 193, 165.
(44) Incze, A.; Pasturel, A.; Chatillon, C. Surf. Sci. 2003, 537, 55.
(45) Winter, M.; Novak, P.; Monnier, A. J. Electrochem. Soc. 1998,
145, 428.
(46) Delley, B.; Ellis, D. E.; Freeman, A. J.; Baerends, E. J.; Post, D.
Phys. ReV. B 1983, 27, 2132.
(9) Mitchell, P. C. H.; Ramirez-Cuesta, A. J.; Parker, S. F.; Tomkinson,
J. J. Mol. Struct. 2003, 651-653, 781.
(47) Oudenhuijzen, M. K.; Bitter, J. H.; Koningsberger, D. C. J. Phys.
Chem. B 2001, 105, 4616.
(10) de Jong, K. P.; Geus, J. W. Catal. ReV.sSci. Eng. 2000, 42, 481.
(11) Serp, P.; Corrias, M.; Kalck, P. Appl. Catal., A 2003, 253, 337.
(12) Mojet, B. L.; Hoogenraad, M. S.; van Dillen, A. J.; Geus, J. W.;
Koningsberger, D. C. J. Chem. Soc., Faraday Trans. 1997, 93, 4371.
(13) Toebes, M. L.; Prinsloo, F. F.; Bitter, J. H.; van Dillen, A. J.; de
Jong, K. P. J. Catal. 2003, 214, 78.
(48) Vannice, M. A.; Poondi, D. J. Catal. 1997, 169, 166.
(49) Vannice, M. A.; Poondi, D. J. Catal. 1998, 178, 386.
(50) Oudenhuijzen, M. K. Ph.D. Thesis, Utrecht University, Utrecht,
The Netherlands, 2002.