In-situ polarization-dependent total-reflection fluorescence XAFS studies on the structure transformation of Pt clusters on ?±-Al2O3(0001)

Article abstract of DOI:10.1021/jp970394p

Structures of Pt species derived from Pt₄(??-CH₃COO)₈ on an ?±-Al₂O₃(0001) single-crystal surface were studied by means of in-situ polarization-dependent total reflection fluorescence XAFS (EXAFS and XANES) techniques. The Pt₄ cluster framework was destroyed upon the deposition of Pt₄(??-CH₃COO)₈ by the reaction on the ?±-Al₂O₃ surface at room temperature. The isolated Pt species were converted to one-atomic layer thick Pt rafts with the Pt-Pt distance of 0.273 nm when the cluster was treated with H₂ at 373 K. The raftlike Pt clusters were stabilized by the formation of direct Pt-O-Al bonding with the ?±-Al₂O₃ surface. The raftlike Pt clusters were redispersed to isolated oxidized Pt species by the reaction with NO. They were also transferred to three-dimensional Pt particles by reduction with H₂ at 773 K.

Full text of DOI:10.1021/jp970394p

J. Phys. Chem. B 1997, 101, 5549-5556
5549
In-Situ Polarization-Dependent Total-Reflection Fluorescence XAFS Studies on the
Structure Transformation of Pt Clusters on r-Al2O3(0001)
Kiyotaka Asakura,† Wang-Jae Chun,‡ Masayuki Shirai,‡,§ Keiichi Tomishige,‡,| and
Yasuhiro Iwasawa*,|
Research Center for Spectrochemistry, Faculty of Science, The UniVersity of Tokyo, Hongo, Bunkyo-ku, Tokyo
113, Japan, Department of Chemistry, Graduate School of Science, The UniVersity of Tokyo, Hongo,
Bunkyo-ku, Tokyo 113, Japan, Institute for Chemical Reaction Science, Tohoku UniVersity, Katahira, Sendai
980-77, Japan, and Department of Applied Chemistry, Graduate School of Engineering, The UniVersity of
Tokyo, Hongo, Bunkyo-ku, Tokyo 113, Japan
ReceiVed: January 31, 1997; In Final Form: April 18, 1997X
Structures of Pt species derived from Pt4(µ-CH3COO)8 on an R-Al2O3(0001) single-crystal surface were studied
by means of in-situ polarization-dependent total reflection fluorescence XAFS (EXAFS and XANES)
techniques. The Pt4 cluster framework was destroyed upon the deposition of Pt4(µ-CH3COO)8 by the reaction
on the R-Al2O3 surface at room temperature. The isolated Pt species were converted to one-atomic layer
thick Pt rafts with the Pt-Pt distance of 0.273 nm when the cluster was treated with H2 at 373 K. The
raftlike Pt clusters were stabilized by the formation of direct Pt-O-Al bonding with the R-Al2O3 surface.
The raftlike Pt clusters were redispersed to isolated oxidized Pt species by the reaction with NO. They were
also transferred to three-dimensional Pt particles by reduction with H2 at 773 K.
1. Introduction
on the bondings parallel to the surface. When the polarization
vector is perpendicular to the surface (p-polarization), the
Industrial metal catalysts are usually used as metal particles
supported on high surface area metal oxide surfaces. The role
of support oxides is not only to disperse the metal particles but
also to modify the morphology, orientation, and electronic state
of the metal particles through metal-support interaction, which
are the most relevant factor to supported metal catalysis.
EXAFS (extended X-ray absorption fine structure) is regarded
as a powerful tool to gain the information about the morphology
of metal particles and metal-support bonding. The particle
morphology can be obtained by the analysis of the coordination
number of metal-metal bonding.1 The metal-support distance
can be determined by detecting the EXAFS signal of atoms
(mostly oxygen) at the support surface. However, it is not
straightforward to obtain such a precise type of structural
information from conventional EXAFS for powder catalyst
sample because of the following reasons. First, the coordination
number determined by EXAFS suffers from many factors like
the effects of thermal and static disorder or asymmetry of pair
distribution function.2,3 Second, the metal-support bond,
especially metal-oxygen bond with a small coordination
number, has a small contribution to whole EXAFS spectra and
is hindered by the metal-metal bond because the EXAFS
oscillation for oxygen does not continue to the high k-region.
The amplitude of EXAFS oscillation depends on the direction
of X-ray electric-field vector bꢀ.2,3 As a result the effective
coordination number, N* is expressed by eq 1:
information about the structure perpendicular to the surface, i.e.,
the metal-support bond is selectively obtained.
Since the conventional supported metal catalysts are used in
a powder form, the EXAFS oscillations are averaged over all
directions. On the other hand, when the flat oxide substrates
are used, the structural information parallel and perpendicular
to the surface can be separately obtained by the polarization-
dependent EXAFS measurements. However, a problem is to
measure the surface species of low concentrations. Concentra-
tions of monolayer metal on a flat surface are in the range 1014-
1015 atoms cm-2, which is 4-5 orders of magnitude lower than
concentrations of metals for conventional catalysts in a powder
form. Usually fluorescence yield detection is more preferable
for dilute samples than monitoring of direct absorption. In a
soft X-ray region, the fluorescence-detection mode successfully
gives the EXAFS signals with high quality4-7 and the polariza-
tion-dependent EXAFS studies determine the adsorption sites
of lighter elements such as C, O, S, Cl, and P.8,9 On the other
hand, most of the catalytically important elements have the
X-ray absorption edges higher than 4 keV. The X-ray with
such a high energy can penetrate deeply into the bulk, yielding
a large amount of scattering X-rays which make it difficult to
measure monolayer species supported on the flat substrate.
When the X-ray hits on the flat substrate with the incident angle
less than the critical angle δc, it undergoes a total reflection
and penetrates only a few nanometers into the bulk. Conse-
quently the scattering X-ray from the bulk is dramatically
reduced.10 After Heald et al. showed the possibility of the
detection of EXAFS signals from a monolayer Au film deposited
on a flat glass substrate by use of a total reflection fluorescence
technique,11 the total-reflection fluorescence EXAFS technique
has been applied to semiconducting materials, electrodes, and
corrosive surfaces.12 We have applied this technique to catalyti-
cally interesting systems.13-17 By analyzing the polarization
dependence of EXAFS amplitude, we have revealed the
orientation and growth mode of catalytic active species such as
N* ) 3 cos2θ
(1)
∑
j
j
where θj is an angle between the polarization vector bꢀ and the
bond vector brj. When the polarization vector is parallel to the
surface (s-polarization), EXAFS will provide the information
† Research Center for Spectrochemistry.
‡ Department of Chemistry.
§ Institute for Chemical Reaction Science.
| Department of Applied Chemistry.
X Abstract published in AdVance ACS Abstracts, June 15, 1997.
S1089-5647(97)00394-5 CCC: $14.00 © 1997 American Chemical Society

5550 J. Phys. Chem. B, Vol. 101, No. 28, 1997
Asakura et al.
SCHEME 1: Preparation Route of Pt Cluster on r-Al2O3(0001)
Co oxides and Cu oxides on flat inorganic oxide substrates as
model systems for supported powder catalysts.
degree (100°) of rotation of the sample around the X-ray light
axis enables us to measure the XAFS of the same sample in
different X-ray polarization vectors without exposure to air. The
fluorescence X-ray was detected by an NaI(Tl) scintillation
counter which can be set close to the sample surface. The XAFS
was measured at BL-7C of the Photon Factory in the National
Laboratory for High Energy Physics (KEK-PF) using Si (111)
sagittal focusing double-crystal monochromator.22 Higher
harmonics were removed by a double mirror made from fused
quartz. The energy resolution was about 2 eV at 9 keV. The
incident X-ray was monitored by a 5.5 cm ionization chamber
filled with N2 gas. Although it is better to measure Pt L1-edge
because eq 1 can be directly used for the analysis of the
polarization dependence of EXAFS amplitude, the edge height
of Pt L1 was so small that the EXAFS oscillation tolerable for
data analyses could not be obtained. Instead, we measured Pt
L3-edge spectra. Thus eq 1 should be modified as discussed
later. All measurements were carried out at room temperature.
One of the difficulty in the total-reflection fluorescence
technique is diffraction from the single-crystal substrate. We
can remove it by putting small pieces of lead plates at the
positions of the detector window on which the Bragg diffractions
hit. The other one is low intensity of the signals. The X-ray
beam was collimated onto the substrate surface using a pinhole
to eliminate undesirable scattering from the other part than
substrate surface. It took 12 h to obtain a whole spectrum. The
species A was measured under high-vacuum conditions. The
measurements of samples B and C were done in the presence
of H2 and the measurement of sample D was performed under
NO atmosphere.
XANES (X-ray absorption near edge structure) spectra give
information about electronic state and geometrical configuration
and also show similar polarization dependence. Thus the
combination of EXAFS and XANES, so-called XAFS (X-ray
absorption fine structure), will give clear insight into the
morphology, orientation, and metal-support interaction.
Moreover, the fluorescence XAFS allows us to carry out in-
situ measurements on the structure of catalysts in the working
conditions. We have constructed an in-situ chamber for
polarization-dependent total-reflection fluorescence EXAFS
measurement under various conditions from UHV (10-9 Pa) to
high pressure (105 Pa) and from low temperature (100 K) to
high temperature (800 K).18 In this paper this in-situ chamber
was employed for the measurement of the polarization-
dependence total-reflection fluorescence XAFS (hereinafter we
use abbreviation of PTRF-XAFS) for Pt species deposited on
R-Al2O3(0001) using a Pt4(µ-CH3COO)8 complex as a precursor.
The PTRF-XAFS technique demonstrated that isolated Pt
monomeric species and one-atomic layer thick Pt rafts were
formed on the R-Al2O3(0001) surface and reversibly transformed
between them depending on the treatment and the atmosphere.
2. Experimental Section
2.1. Preparation of Materials. The cluster [Pt4(µ-CH3-
COO)8] was prepared following the literature.19-21 An optical-
grade polished R-Al2O3(0001) crystal was purchased from Earth
Jewelry Co. The single crystal (10 × 30 × 1 mm3) was calcined
at 573 K for 2 h followed by evacuation at the same temperature
for 1 h. The preparation steps are summarized in Scheme 1. Pt
was deposited on the R-Al2O3(0001) surface by a dropwise
impregnation method using a CH2Cl2 solution of [Pt4(µ-CH3-
COO)8] in the in-situ chamber18 under a flow of high purity Ar
(99.999%). The loading was estimated to be 1.5 Pt4 nm-2 by
XPS, which corresponds to an atomic ratio of platinum:surface
oxygen ) 4:10. The solvent was removed by evacuation under
high vacuum at room temperature. The incipient supported
sample A was then reduced at 373 K with H, (13.3 kPa) to
form species B in the chamber. The species B was further
reduced to species C by H2 (13.3 kPa) at 673 K. Species B
was also exposed to NO (1.33 kPa) at room temperature in the
chamber (Species D).
2.3. Data Analysis. The EXAFS spectra were calculated
directly from the ratio of the fluorescence from the sample to
the incident X-ray signal detected by an ionization chamber (If/
Io) without any correction for self absorption because the Pt
atoms were located only on the surface. The EXAFS oscillation
ø(k) was extracted from the spectra by a cubic spline method
and normalized by the edge height using EXAFS analysis
program REX. The energy dependence of the edge height was
taken into account using McMaster equation.23 The origin of
kinetic energy of photoelectron was temporarily taken in an
inflection point of the edge jump. The Fourier filtered data
were then analyzed by curve fitting analysis using eq 2:
Nj*Fj(kj)
2.2. PTRF-XAFS Measurements. The details of the in-
ø(k ) )
exp(-2σ2j k2j ) exp(-2rj/λj) sin(2kjrj +
φj(kj)) (2)
where kj, rj, σj, λj, Fj(k), and φj(k) are the wave-number of the
situ XAFS chamber was described elsewhere.18 Here we briefly
describe the in-situ XAFS chamber, in which the XAFS spectra
can be obtained under various conditions: from high vacuum
(1 × 10-9 Pa) to high pressure (1 × 105 Pa) and from low
temperature (100 K) to high temperature (800 K). A wide
∑
j
kr2j
j

Pt Clusters on R-Al2O3(0001)
J. Phys. Chem. B, Vol. 101, No. 28, 1997 5551
photoelectron, the interatomic distance, the Debye-Waller
factor, the mean free path of photoelectron, the backscattering
amplitude function, and phase shift function for the jth shell,
respectively. The Fj(k) and φj(k) for Pt-Pt are derived from
Pt foil. Those for Pt-O were calculated from feff v5.05 with
Pt-O ) 0.20 nm.24 Since the absolute value of the theoretically
derived amplitude is smaller by a factor of 0.7-0.9 than the
actual amplitude, it was corrected by using the coordination
number of 6 for Pt-O bonding in (NH4)4[H4PtMo7O24]. The
origin of photoelectron kinetic energy was adjusted using the
following relation 3 in the curve-fitting process:
Figure 1. The turnover number of formed CO2 per Pt4 cluster during
HCOOH decomposition reaction on the Pt species derived from Pt4(µ-
CH3COO)8/R-Al2O3 (9) and Pt4(µ-CH3COO)8/SiO2 (O) at 288 K.
k2j ) k2 - 2m∆Ej/p
(3)
and one may observe Pt-X bonding even though the Pt-X
bond direction is perpendicular to the X-ray polarization. In
this paper we use the terms, s-polarization and p-polarization,
which mean that the electric-field vectors are parallel to the
surface and perpendicular to the surface, respectively.
2.4. Catalytic Reaction. Catalytic activity for decomposi-
tion reaction of HCOOH on the R-Al2O3-supported Pt4(µ-CH3-
COO)8 powder was checked in a closed circulating system
connected to a gas chromatography (Shimadzu Co, GC-8A) with
a Porapack PS column. The pressure of HCOOH was 1.47 kPa
and the reaction temperature was 288 K.
where ∆Ej and m are the deviation of the origin of photoelectron
kinetic energy and the mass of electron, respectively.
Errors in curve-fitting analysis were estimated when ꢀv
2
defined in eq 4 exceeds 1 according to the report of Committee
on Standards and Criteria in X-ray Absorption Spectroscopy.25
ꢀν2 ) (Nidp/(Nidp - p))(1/Ndata
)
[(øobs(k) - øcal(k))2/σe2rror
]
∑
(4)
where p is a number of the used fitting parameters; Ndata is a
number of data points; øobs(k), øcal(k) and σerror are observed
and calculated ø(k) and the estimated error, respectively. In
this work we took into account only statistical errors. The fitting
parameters are Nidp is a number of freedom calculated by26
3. Results
3.1. Catalytic Performance. Figure 1 shows the activity
of Pt4(µ-CH3COO)8 dispersed on R-Al2O3 for the HCOOH
decomposition reaction at 288 K. HCOOH was selectively
decomposed to CO2 and H2 without any formation of CO and
H2O, contrasted to the HCOOH decomposition on Al2O3-
supported Pt particles prepared by a traditional impregnation
method using an aqueous solution, followed by the reduction
with H2 where both CO2 + H2 and CO + H2O were produced
with a much lower activity. For comparison, the activity of
Pt4(µ-CH3COO)8 supported on SiO2 is shown in Figure 1.29 As
has already been reported, the SiO2-supported Pt4(µ-CH3COO)8
showed a long induction period which became much longer than
that for DCOOH and DCOOD.29 The reaction on the SiO2-
supported Pt4(µ-CH3COO)8 terminated at ca. 10 min after the
reaction initiation because the Pt species was reduced to Pt metal
particles during the reaction. Unlike the SiO2-supported Pt
species, Pt species on R-Al2O3 did not show any induction
period. The initial rate for the formation of CO2 per Pt4
(turnover frequency) was 4.4 min-1 which is twice as large as
the maximum reaction rate for the Pt4(µ-CH3COO)8/SiO2.
3.2. A Structure of Pt Species A. Figure 2A shows the
EXAFS oscillations parallel (s-polarization) and perpendicular
(p-polarization) to the R-Al2O3(0001) surface. The period and
amplitude of the EXAFS oscillations in both directions are
similar with each other. The corresponding Fourier transforms
over k ) 30-90 nm-1 give one peak around 0.17 nm (phase
shift: uncorrected) for both directions as shown in Figure 2B.
The original cluster Pt4(µ-CH3COO)8 has the Pt-Pt distance at
0.2495 nm which will give the Fourier transform peak around
0.20 nm (phase shift: uncorrected).29 However, there was no
peak in this region for both s and p-polarizations, indicating
that the Pt4 cluster framework was destroyed. Moreover, no
peak corresponding to Pt-Pt bond of Pt metal particles was
observed around 0.25 nm (phase shift: uncorrected) in the
Fourier transforms for both directions. Thus Pt atoms are
monoatomically dispersed on the R-Al2O3 surface by the
interaction with the substrate surface.
Nidp ) 2∆k∆r/π+2
(5)
where ∆k and ∆r are the ranges of the Fourier transformation
in r and k space. Normally p is 4(N,r,∆E,∆σ2) and Nidp is 6.6
for 1 shell fitting.
The eq 1 is no longer valid for L3-edge spectra because the
L3-edge EXAFS corresponds to the transition of angular
momentum from l ) 1 to l ) 0, 2. Therefore, the polarization
dependence of EXAFS is expressed by eq 6 for L2,3 edges,27
instead of eq 1.
Fj(k)
ø(k) )
exp(-2σ2j k2) exp(-2rj/λ) × {1/2(1 +
∑
kr2j
j
3 cos2θj)|M21| sin(2krj + φ2j) + 1/2|M01| sin(2krj + φ0j) +
2
2
M01M21(1 - 3 cos2 θj) sin(2krj + φ02j)}/(|M21| + 1/2|M01| )
2
2
(6)
where M01 and M21 are dipole moments for the transitions from
l ) 1 to l ) 0 and l ) 2, respectively. φ0j, φ2j, and φ02j are
phase shift functions for the transitions from l ) 1 to l ) 0,
from l ) 1 to l ) 2 and their cross term, respectively. Since
(M01/M21) is about 0.2, we adopted the following assumptions
for simplification in the present EXAFS analysis: (i) φ02j(k) )
φ2j(k), (ii) (M01/M21) ) 0.2, (iii) (M01/M21)2 is negligible small.
Then the eq 6 can be simplified to eq 7:28
Fj(k)
kr2j
ø(k ) ) N*
exp(-2σ2j k2j ) exp(-2rj/λj) sin(2kjrj +
∑
j
j
j
φ2j(kj) (7)
In this case N* is expressed as follows:
N*) (0.7 + 0.9 cos2θ )
(8)
∑
j
k
k
We carried out the curve fitting analysis on the inversely
Fourier transformed data of the first peaks filtered over 0.1-
Note that the polarization dependence for L3 edge is not perfect

5552 J. Phys. Chem. B, Vol. 101, No. 28, 1997
Asakura et al.
Figure 2. k2-weighted EXAFS oscillations for species A: (a) p-polarization, (b) s-polarization. (B): Fourier transforms of the k2-weighted EXAFS
oscillations for species A, (a) p-polarization, (b) s-polarization.
Figure 4. Model structure for species A.
TABLE 2: Calculated Effective Coordination Numbers (N*)
for Pt-O at Typical Three Adsorption Sites
coordination number
polarization
exptl
atop
bridge
3-fold
p
s
3.3 ( 0.5
3.0 ( 0.5
1.6
0.7
2.4
1.8
3.1
3.0
Figure 3. Best fitting results for inversely Fourier transformed data
of the species A: (a) p-polarization; (b) s-polarization. Solid line,
filtered data, broken line, calculated data.
expected N* values in both polarizations and their ratios for
the three Pt sites, only the 3-fold site model can reproduce the
observed data as shown in Table 2. The location of Pt atoms
are illustrated in Figure 4. The R-Al2O3(0001) has two types
of 3-fold hollow site which are distinguished by the presence
of the underneath Al atom. Here the site which has no Al atom
underneath the Pt is called as fcc site and the other site which
has an Al atom below is called as hcp site. In our previous
report Co atoms on R-Al2O3(0001) were demonstrated to occupy
the fcc hollow sites of the surface where no Al was located
under the Co atoms in the second surface layer of Al2O3 because
we did not observe the Co-Al interaction in the p-polarization
PTRF-EXAFS spectrum.16 Although the peak appeared around
0.3 nm in Figure 2B, it was difficult to assign it to Pt-Al
because the peak height is nearly in the noise level. In this
work, we can not determine whether the location of Pt is fcc or
hcp sites.
TABLE 1: Curve-Fitting Results for Pt Species A on
r-Al2O3(0001)
Pt-O
sample
N*
r/nm
∆σ2/nm
∆E/eV
p-polarization 3.3 ( 0.5 0.198 ( 0.003 0.001 ( 0.001 3 ( 2
s-polarization 3.0 ( 0.5 0.201 ( 0.003 0.001 ( 0.001 3 ( 2
0.22 nm. Figure 3 shows the curve-fitting results in p-
polarization and s-polarization. The structure parameters thus
obtained are summarized in Table 1. The Pt-O bonds are found
at about 0.200 ( 0.003 nm in both directions. The effective
coordination numbers (N*) for p-polarization and s-polarization
are 3.3 and 3.0, respectively (Table 1). The effective coordina-
tion numbers were calculated using eq 8 assuming typical
adsorption site models for Pt atoms on R-Al2O3(0001) in Table
2. We postulated three surface sites such as atop, bridge, and
3-fold sites on the unreconstructed R-Al2O3(0001) surface (O-O
distance ) 0.275 nm) as shown in Figure 4. Among the
3.3. Structure of Pt Species B. Figure 5A shows the
EXAFS oscillations for Pt species B prepared by the reduction

Pt Clusters on R-Al2O3(0001)
J. Phys. Chem. B, Vol. 101, No. 28, 1997 5553
Pt-Pt
TABLE 3: Best Fitting Results of the EXAFS Data for Species B
Pt-O
N
r/nm
∆σ2/nm
∆E/eV
N
r/nm
∆σ2/nm
∆E/eV
p-pol
s-pol
1.5 ( 1.0
0.220 ( 0.009
no
0.003 ( 0.001
no
5 ( 2
no
3 ( 0.8
6 ( 1
0.274 ( 0.003
0.273 ( 0.003
0.001 ( 0.001
0.002 ( 0.001
3 ( 2
3 ( 2
nob
a pol ) polarization. b no ) not observed.
Figure 5. (A): k2-weighted EXAFS oscillations of species B obtained
by the reduction of species A with hydrogen at 373 K, (a) p-polarization,
(b) s-polarization. (B): Fourier transforms of the k2-weighted EXAFS
oscillations for species B, (a) p-polarization, (b) s-polarization.
Figure 6. Best fitting results for the inversely Fourier transformed
data of the species B: (a) p-polarization, (b) s-polarization. Solid line,
observed data; broken line, calculated data.
of species A at 373 K with H2 (Scheme 1). The oscillation in
s-polarization lasts to the high k-region compared to that in
p-polarization. Figure 5B shows the Fourier transformation of
the EXAFS oscillations. The peak appears around 0.22 nm
(phase shift: uncorrected) in both directions. However, the peak
height for the Fourier transform of the p-polarization is 60% of
that of the s-polarization. In addition, the p-polarization has a
shoulder structure in the lower side of the main peak. We
carried out a curve fitting analysis assuming that the main peak
in s-polarization was due to Pt-Pt bonding. Since the shoulder
structure appears only in p-polarization and it is attributable to
Pt-O bonding, we carried out two-term fitting including Pt-
Pt and Pt-O. Since the number of fitting parameters in the
two-term fitting, 8, are near to the Nidp ) 9.6 (∆k ) 60 nm-1
and ∆r ) 0.2 nm), we fixed ∆E value to that obtained from
EXAFS for s-polarization. Consequently, we obtained the best
fitting depicted in Figure 6 and in Table 3. The larger error
for Pt-O arises from the smaller contribution from the Pt-O
than that from Pt-Pt and also the low signal to noise ratio of
the observed data.
The Pt-Pt distances are determined to be 0.273 ( 0.003 nm,
which is 0.005 nm shorter than that of Pt foil. The effective
coordination number (N*) of Pt-Pt for s-polarization is twice
as large as that for p-polarization. If the Pt particles have a
spherical shape, the N* would give an equal value for s- and
p-polarizations. Thus it is suggested that the anisotropic
structures such as thin layer, semispherical, and ellipsoid are
formed for Pt clusters on the Al2O3 surface. We calculated the
N* based on the models such as one atomic layer, two atomic
layer, and three atomic layer raft using eq 8. Figure 7 shows
the relation between N* for Pt-Pt and the number of the layer
as a function of the particle size (in diameter). The observed
N*’s, 3 for p-polarization and 6 for s-polarization, fit the
expected values only for the one atomic layer raft structure.
The other model, such as more than two layer, semisphere, and
ellipsoid do not entirely reproduce the observed data. Further-
more, the particle size (diameter) of the Pt raft can be estimated
Figure 7. Calculated effective coordination numbers of Pt-Pt for
s-polarization (a) and p-polarization (b) as a function of particle size
based on the one-layer model (solid line 1), the two-layer model (broken
line 2) and the three-layer model (dotted line 3). The observed data
are given in a broken line with error bars.
to be ca 1.5 ( 0.5 nm from the observed and calculated N*
values for the p- and s-polarization as shown in Figure 7.
For the one layer raft structure a strong Pt-support interaction
may be expected. In fact, we got the better fitting results for
the p-polarization EXAFS when Pt-O was assumed. The Pt-O
distance was determined as 0.220 ( 0.009 nm as shown in Table

5554 J. Phys. Chem. B, Vol. 101, No. 28, 1997
Asakura et al.
Figure 8. Model structure for species B.
Figure 10. (b) Fourier transforms of the k2-weighted EXAFS oscil-
lations for species D after the reaction of species B with NO: (a)
p-polarization, (b) s-polarization.
dicular to the surface which is caused by the Pt-O bonding as
was detected in EXAFS. On the other hand, The XANES
spectrum is s-polarization (Figure 9c) resembles the metallic
Pt particles reflecting metal-metal bonding. Thus these facts
also support the raftlike structure of Pt particles strongly
interacted with surface oxygen well corresponding to the
EXAFS results described in section 3.3.
After the reduction of the species B at 673 K, the XANES
spectra for s- and p-polarizations were found to be similar to
each other (Figure 9e,f). The structures A, B, and C in the
XANES spectra of Figures 9e,f appeared at the same positions
as those for Pt foil, which suggests the formation of Pt particles.
No polarization dependence of the XANES spectra indicates
the growth of three-dimensional Pt particles on the R-Al2O3-
(0001) surface after the 673 K reduction.
3.5. Structure of Species D after the Exposure of B to
NO. When species B was exposed to NO at room temperature,
the Pt-Pt bond was cleaved. As shown in Figure 10b, no peak
due to Pt-Pt bond appeared in the Fourier transform for
s-polarization in which the Pt-Pt peak in the sample B had
been observed before NO adsorption. The Pt raft was destroyed
to monomer species by the interaction with NO. The peak
observed in the Fourier transform (Figure 10b) was analyzed
to be due to Pt-O at 0.205 nm. In the p-polarization spectrum,
we were not able to find definite EXAFS oscillation and no
peak above noise level was observed in Figure 10a. The
EXAFS oscillation for p polarization might be composed of
platinum-oxygen (support) and Pt-NO (adsorbed). We infer
that the EXAFS oscillation might smear out because of the
interference of the different interatomic distances between
platinum-oxygen (support) and platinum-adsorbates. Pt-Pt
bond was cleaved in the species D after the exposure to NO
and Pt monomer adsorbed with NO was created on the Al2O3
surface though the further study is necessary to draw the definite
structure for the species D.
Figure 9. XANES spectra at Pt L3-edge for species A (a, b), species
B (c, d), and species C (e, f). Parts a, c, e are for s-polarization and
parts b, d, f are for p-polarization.
3. The model structure for Pt rafts interacted with the oxygen
layer of the R-Al2O3(0001) surface is illustrated in Figure 8.
3.4. Structure of Pt Species Reduced after 673 K. When
the Pt clusters were reduced at 673 K, the Pt atoms were
aggregated to 3-dimensional particles. Figure 9 shows the
polarization-dependent XANES spectra at the L3 edge of Pt
species on R-Al2O3(0001). The L3-edge peak intensity reflects
the vacancy of the 5 d state,30-32 i.e., stronger white line peak
suggests more electron deficiency in d state. The species A
exhibits a strong edge peak for both s- and p-polarizations, which
indicates that the Pt atoms possess positive charge induced by
the Pt-O-Al bonding as shown in Figure 4.
The species B formed by the reduction at 373 K shows
different XANES spectra between s- and p-polarizations. The
XANES feature of species B for s-polarization (Figure 9c) is
similar to that of the supported small Pt metal particles adsorbed
by hydrogen, the feature of which was characterized by the
broadening of the white line peak to the higher energy with
decreasing the edge height induced by the adsorption of
hydrogen.30,33-40 On the other hand, the XANES spectrum in
p-polarization shows a strong white line peak30,31 indicating the
presence of the more vacancy in the d electron states perpen-
4. Discussion
4.1. Structure of Species A and Its Catalytic Activities.
The Pt/R-Al2O3 and the Pt/SiO2 catalysts derived from Pt4(µ-
CH3COO)8 as precursor show high activity for the selective
dehydrogenation of HCOOH described above. While the Pt/
SiO2 has a long induction period, the Pt/R-Al2O3 shows no
induction period (Figure 1). The active species on Pt/SiO2 was
found to be Pt monomers which were bound to the SiO2 surface

Pt Clusters on R-Al2O3(0001)
J. Phys. Chem. B, Vol. 101, No. 28, 1997 5555
The structural transformation of the Pt4 clusters on R-Al2O3-
(0001) is summarized in Figure 11. The Pt4 cluster framework
is destroyed to Pt monomers which occupy the 3-fold hollow
sites of the oxygen atoms of R-Al2O3(0001) surface, making
Pt-O bonds at the distance of 0.200 ( 0.003 nm. The isolated
Pt atoms are reduced to one-atomic layer thick raft cluster with
H2 at 373 K. The Pt-Pt and Pt-O bondings are observed at
0.273 nm and 0.220 nm, respectively. The Pt rafts are
redispersed to Pt monomers by the exposure of NO, or
agglomerated to three-dimensional particles by the 673 K
reduction.
Figure 11. Structural transformation of the Pt species on R-Al2O3-
(0001).
Acknowledgment. The authors would like to express thanks
to Prof. M. Nomura in PF and Dr. T. Yokoyama in the Unversity
of Tokyo for their kind technical supports. The work was
carried out under the approval of PF advisory committee
(Proposals 90142, 92G174). This work has been supported by
CREST (Core Research for Evolutional Science and Technol-
ogy) of Japan Science and Technology Corporation (JST).
making Pt-O bonds.29 During the induction period the Pt4(µ-
CH3COO)8 cluster framework on SiO2 was decomposed to
monomers via a Pt dimer structure with the Pt-Pt distance at
0.250 nm by the exposure to HCOOH. The present EXAFS
analysis reveals that Pt monomers are readily formed with 0.200
( 0.003 nm on 3-fold site of R-Al2O3(0001) surface after the
deposition of Pt4(µ-CH3COO)8 cluster. Since the Pt monomer
structure has already been present before the catalytic reaction,
there is no induction period in the reaction. Such a different
behavior of Pt4(µ-CH3COO)8 clusters on SiO2 and Al2O3 can
be compared to the different structural transformation of Ru
clusters and Rh dimers on SiO2 and Al2O3,40,41 where the metal-
metal bonds of the cluster and the dimer were observed on SiO2
surface though the atomically dispersed species were present
on the Al2O3.
4.2. Raftlike Structure of the Species B. Raftlike structures
of supported metal particles have been proposed for Ru/SiO2,
Os/SiO2, Ru-Os/SiO2, and Ir-Pt/Al2O3 by transmission electron
microscopy (TEM) and EXAFS and Ru-Cu particles on SiO2
by TEM and EXAFS.1,42-45 Yates et al. also reported a raft
structure for Rh particles on Al2O3 by means of TEM,
chemisorption, and IR.46 The raftlike structures have also been
demonstrated by observing the direct metal-oxygen(support)
bonding.47-51 The PTRF-EXAFS can determine more directly
the raft structure by comparing the EXAFS oscillations in two
directions, perpendicular, and parallel to the surface. The PTRF-
EXAFS showed the presence of metal-oxygen (support) bond
which was difficult to detect by conventional EXAFS because
a small contribution of the metal-oxygen bonding as compared
to a large contribution of the metal-metal bonding except for
very small metal cluster systems. Koningsberger et al. used a
difference file method to remove the large contribution of
metal-metal bonding.51 They reported the metal-oxygen
distances in the supported noble metal powder catalysts which
were present around 0.25-0.27 nm when the sample was
reduced at low temperatures and in the presence of H2, while
the Pt-O bond length decreased to ca. 0.22 nm when the sample
was evacuated or reduced at high temperatures (>723 K).52 In
the present work on the R-Al2O3(0001) single-crystal surface
we found the Pt-O distance at 0.220 nm even reduced with H2
at low temperature (373 K) and in the presence of H2 which
indicated the formation of the covalent bonds between Pt atoms
and surface oxygen atoms. Such a strong Pt-O interaction
stabilizes the Pt raftlike structure.
References and Notes
(1) Greegor, R. B.; Lytle, F. W. J. Catal. 1980, 63, 476.
(2) Koningsberger, D. C.; Prins, R. X-ray Absorption, Principles,
Applications, Techniques of EXAFS, SEXAFS, and XANES; John Wiley &
Sons: New York, 1988; Vol. 92.
(3) Iwasawa, Y. X-ray Absorption Fine Structure for Catalysts and
Surfaces; World Scientific: Singapore, 1996; Vol. 2.
(4) Stohr, J.; Kollin, E. B.; Fischer, D. A.; Hastings, J. B.; Zaera, F.;
Sette, F. Phys. ReV. Lett. 1985, 55, 1468.
(5) Fischer, D. A.; Hastings, J. B.; Zaera, F.; Stohr, J.; Sette, F. Nucl.
Instrum. Methods Phys. Res. 1986, A246, 561.
(6) Funabashi, M.; Ohta, T.; Yokoyama, T.; Kitajima, Y.; Kuroda, H.
ReV. Sci. Instrum. 1989, 60, 2505.
(7) Funabashi, M.; Yokoyama, T.; Takata, Y.; Ohta, T.; Kuroda, H.
Surf. Sci. 1990, 242, 59.
(8) Stohr, J. X-ray Absorption Principles, Applications, Techniques of
EXAFS, SEXAFS, and XANES; Prins, R., Koningsberger, D. C., Eds.; A
John Wiley & Sons: New York, 1988; p 443.
(9) Ohta, T. X-ray Absorption Fine Structure for Catalysts and
Surfaces; Y. Iwasawa, Ed.; World Scientific: Singapore, 1996; Vol. 2, p
354.
(10) Parratt, L. G. Phys. ReV. 1954, 95, 359.
(11) Heald, M.; Keller, E.; Stern, E. A. Phys. Lett. 1984, 103A, 155.
(12) Shirai, M.; Iwasawa, Y. XAFS Application to Catalysis and Surface;
Y. Iwasawa, Ed.; Elsevier: Amsterdam, 1996; p 332.
(13) Shirai, M.; Asakura, K.; Iwasawa, Y. Catal. Lett. 1992, 15, 247.
(14) Shirai, M.; Asakura, K.; Iwasawa, Y. Catal. Lett. 1992, 1037.
(15) Shirai, M.; Asakura, K.; Iwasawa, Y. Jpn. J. Appl. Phys., Part 1
1993, 32-2, 413.
(16) Shirai, M.; Inoue, T.; Onishi, H.; Asakura, K.; Iwasawa, Y. J. Catal.
1994, 145, 159.
(17) Shirai, M.; Asakura, K.; Iwasawa, Y. Catal. Lett. 1994, 26, 229.
(18) Shirai, M.; Nomura, M.; Asakura, K.; Iwasawa, Y. ReV. Sci.
Instrum. 1995, 66, 5493.
(19) Yamaguchi, T.; Sasaki, Y.; Nagasawa, A.; Ito, T.; Koga, N.;
Morokuma, K. Inorg. Chem. 1989, 28, 4311.
(20) Yamaguchi, T. Doctoral Thesis, Tohoku University, 1989.
(21) Yamaguchi, T.; Sasaki, Y.; Ito, T. J. Am. Chem. Soc. 1990, 112,
4038.
(22) Nomura, M.; Koyama, A.; Sakurai, M. KEK Report 1991, 91-1,
1.
(23) Asakura, K. X-ray Absorption Fine Structure for Catalysts and
Surfaces; Iwasawa, Y., Ed.; World Scientific: Singapore, 1996; Vol. 2; p
33.
(24) Rehr, J. J.; Mustre de Leon, J.; Zabinsky, S. I.; Albers, R. C. J.
Am. Chem. Soc. 1991, 113, 5135.
(25) Lytle, F. W. Sayers, D. E.; Sterns, E. A. Physica B 1988, 158,
701. Report on the International Workshops on Standards and Criteria in
XAFS, York. Proceedings XAFS 6 1990, 751.
5. Conclusions
(26) Stern, E. Phys. ReV. B 1993, 48, 9825.
In this paper, we have reported the polarization-dependent
total-reflection fluorescence XAFS (EXAFS and XANES)
studies on the Pt cluster on R-Al2O3(0001). This technique gives
clearer information on adsorption sites of metal atoms, metal-
support, and metal-metal bondings, and anisotropic surface
structures.
(27) Teo, B. K.; Lee, P. A. J. Am. Chem. Soc. 1979, 101, 2815.
(28) Comin, F.; Citrin, P. H.; Eisenberger, P.; Rowe, J. E. Phys. ReV. B
1982, 26, 7060.
(29) Chun, W. J.; Tomishige, K.; Hamakado, M.; Iwasawa, Y.; Asakura,
K. J. CChem. Soc., Faraday Trans. 1995, 91, 4161.
(30) Lytle, F. W.; Wei, P. S. P.; Greegor, R. B.; Via, G. H.; Sinfelt, J.
H. J. Chem. Phys. 1979, 70, 4849.

5556 J. Phys. Chem. B, Vol. 101, No. 28, 1997
Asakura et al.
(31) Horsley, J. A. J. Chem. Phys. 1982, 76, 1451.
(32) Mansour, A. N.; Cook, J. W.; Sayers, D. E. J. Phys. Chem. 1984,
88, 2330.
(33) Samant, M. G.; Boudart, M. J. Phys. Chem. 1991, 95, 4070.
(34) McHugh, B. J.; Larsen, G.; Haller, G. L. J. Phys. Chem. 1990, 94,
8621.
(35) Vaarkamp, M.; Miller, J. T.; Modica, F. S.; Lane, G. S.; Konings-
berger, D. C. Proc. Int. Cong. Catal. 1992, 10, 809.
(36) Vaarkamp, M.; Modica, F. S.; Miller, J. T.; Koningsberger, D. C.
J. Catal. 1993, 144, 6111.
(37) Vaarkamp, M.; Miller, J. T.; Modica, F. S.; Koningsberger, D. C.
Jpn. J. Appl. Phys. 1993, 32-2, 454.
(38) Ichikuni, N.; Iwasawa, Y. Catal. Lett. 1993, 20, 87.
(39) Kubota, T.; Asakura, K.; Ichikuni, N.; Iwasawa, Y. Chem. Phys.
Lett. 1996, 256, 445.
(40) (a) Asackura, K.; Kubota, T.; Ichikuni, N.; Iwasawa, Y. In
Proceedings of the 11th International Congress on Catalysts; Elsevier
Science B.V.: Baltimore, MD, 1996; p 911. (b) Asakura, K.; Iwasawa, Y.
J. Chem. Soc., Faraday Trans. 1 1988, 84, 2445.
(41) Bando, K. K.; Asakura, K.; Arakawa, H.; Isobe, K.; Iwasawa, Y.
J. Phys. Chem. 1996, 100, 13636.
(42) Prestridge, E. B.; Via, G. H.; Sinfelt, J. H. J. Catal. 1977, 50, 115.
(43) Lytle, F. W.; Via, G. H.; Sinfelt, J. H. J. Chem. Phys. 1977, 67,
3831.
(44) Via, G. H.; Sinfelt, J. H.; Lytle, F. W. J. Chem. Phys. 1979, 71,
690.
(45) Sinfelt, J. H.; Via, G. H.; Lytle, F. W. J. Chem. Phys. 1982, 76,
2779.
(46) Yates, D. J. C.; Murrel, L. L.; Prestridge, E. B. J. Catal. 1979, 57,
41.
(47) Fung, A. S.; Tooley, P. A.; Kelly, M. J.; Koningsberger, D. C.;
Gates, B. C. J. Phys. Chem. 1991, 95, 225.
(48) Van Zon, F. B. M.; Maloney, S. D.; Gates, B. C.; Koningsberger,
D. C. J. Am. Chem. Soc. 1993, 115, 10317.
(49) Asakura, K.; Yamada, M.; Iwasawa, Y.; Kuroda, H. Chem. Lett.
1985, 511.
(50) Asakura, K.; Iwasawa, Y. J. Chem. Soc., Faraday Trans. 1990,
86, 2657.
(51) van Zon, J. B. A. D.; Koningsberger, D. C.; van’t Blik, H. F. J.;
Prins, R.; Sayers, D. E. J. Chem. Phys. 1984, 80, 3914.
(52) Koningsberger, D. C.; Gates, B. C. Catal. Lett. 1992, 14, 271.