Solving the structure of size-selected Pt nanocatalysts synthesized by inverse micelle encapsulation

Article abstract of DOI:10.1021/ja101997z

The structure, size, and shape of γ-Al₂O ₃-supported Pt nanoparticles (NPs) synthesized by inverse micelle encapsulation have been resolved via a synergistic combination of imaging and spectroscopic tools. It is shown that this synthesis method leads to 3D NP shapes even for subnanometer clusters, in contrast to the raft-like structures obtained for the same systems via traditional deposition-precipitation methods. Furthermore, a high degree of atomic ordering is observed for the micellar NPs in H₂ atmosphere at all sizes studied, possibly due to H-induced surface reconstruction in these high surface area clusters. Our findings demonstrate that the influence of NP/support interactions on NP structure can be diminished in favor of NP/adsorbate interactions when NP catalysts are prepared by micelle encapsulation methods.

Full text of DOI:10.1021/ja101997z

Published on Web 06/07/2010
Solving the Structure of Size-Selected Pt Nanocatalysts
Synthesized by Inverse Micelle Encapsulation
,
†,‡,§
†
†,§
†
Beatriz Roldan Cuenya,*
Jason R. Croy, Simon Mostafa, Farzad Behafarid,
|
|
|
⊥
,⊥
Long Li, Zhongfan Zhang, Judith C. Yang, Qi Wang, and Anatoly I. Frenkel*
Department of Physics, Nanoscience and Technology Center, and Department of CiVil, Construction
and EnVironmental Engineering, UniVersity of Central Florida, Orlando, Florida 32816, Department
of Mechanical Engineering and Materials Science, UniVersity of Pittsburgh, Pittsburgh,
PennsylVania 15261, and Department of Physics, YeshiVa UniVersity,
New York, New York 10016
Received March 9, 2010; E-mail: roldan@physics.ucf.edu; anatoly.frenkel@yu.edu
2 3
Abstract: The structure, size, and shape of γ-Al O -supported Pt nanoparticles (NPs) synthesized by inverse
micelle encapsulation have been resolved via a synergistic combination of imaging and spectroscopic tools.
It is shown that this synthesis method leads to 3D NP shapes even for subnanometer clusters, in contrast
to the raft-like structures obtained for the same systems via traditional deposition-precipitation methods.
2
Furthermore, a high degree of atomic ordering is observed for the micellar NPs in H atmosphere at all
sizes studied, possibly due to H-induced surface reconstruction in these high surface area clusters. Our
findings demonstrate that the influence of NP/support interactions on NP structure can be diminished in
favor of NP/adsorbate interactions when NP catalysts are prepared by micelle encapsulation methods.
4
1
. Introduction
Metal nanoparticles (NPs) possess unique properties as
steps and kinks present in their system. Komanicky et al.
described the role played by the NP shape on the oxygen
reduction reaction over Pt NPs prepared by electron beam
lithography. On the basis of the distinct electrocatalytic activities
observed for differently oriented NP arrays, the authors sug-
gested the preferential adsorption of O on {100} facets and its
2
subsequent diffusion to nearby {111} facets, where it could be
more efficiently reduced.
While knowledge of the average cluster size and size
distribution characteristics of a particular as-prepared sample
is necessary, this is not sufficient for the complete understanding
of catalyst reactivity. In fact, information on the most stable
cluster shapes under any given set of reaction conditions
temperature, pressure, chemical environment) is equally im-
portant. Mittendorfer et al. theoretically predicted drastic
changes in the morphology and equilibrium shape of Pd and
compared to their bulk counterparts, including enhanced chemi-
cal reactivity. It has recently been recognized that the NP shape
might play a crucial role in its catalytic performance. Specifi-
cally, it has been demonstrated that different crystallographic
facets stabilized on particles with different shapes may result
1
-5
in different reactivities and selectivities.
For example, the
rate of styrene oxidation over Ag nanocube NPs was found to
be 14 times higher than over nanoplates, and 4 times higher
than over nearly spherical NPs. These differences were attributed
to highly reactive {100} facets on the nanocubes, in contrast to
the {111} facets present on the nanoplates, and the mixture of
(
{
100} and {111} planes in the nanospheres. Similar findings
7
have been made for Pt nanocatalysts that are extensively used
6
5
in industrial applications. Tian et al. reported the superior
catalytic activity of tetrahexahedral Pt NPs with high-index
facets [{730},{210},{520}] for the electro-oxidation of formic
acid and ethanol as compared to NPs with spherical shapes.
This result was attributed to the increased number of atomic
Rh nanocrystals upon O
facets dominating under low O
rounding occurring at elevated O
2
adsorption, with closed-packed {111}
2
pressures, and nanocrystal
pressures due to its higher
adsorption energy on the initially less stable {110} open
surfaces. Grazing incidence X-ray diffraction and transmission
2
8
electron microscopy (TEM) experiments by Nolte et al. on Rh
†
Department of Physics, University of Central Florida.
Nanoscience and Technology Center, University of Central Florida.
Department of Civil, Construction and Environmental Engineering,
NPs supported on MgO(001) indicated an increase in the total
area of {100} facets at the expense of {111} surfaces upon
oxidation. Such shape changes were found to be reversible upon
‡
§
University of Central Florida.
|
University of Pittsburgh.
Yeshiva University.
subsequent NP reduction during CO oxidation. High-resolution
⊥
9,10
TEM images, obtained by Giorgio et al.
2 2
under H and O
atmospheres using an environmental TEM demonstrated the
faceting of TiO -supported Au NPs under H (truncated
(
(
(
(
1) Narayanan, R.; El-Sayed, M. A. Nano Lett. 2004, 4, 1343.
2) Narayanan, R.; El-Sayed, M. A. J. Phys. Chem. B 2004, 108, 5726.
3) Xu, R.; Wang, D.; Zhang, J.; Li, Y. Chem. Asian J. 2006, 1, 888.
4) Komanicky, V.; Iddir, H.; Chang, K. C.; Menzel, A.; Karapetrov, G.;
Hennessy, D.; Zapol, P.; You, H. J. Am. Chem. Soc. 2009, 131, 5732.
2
2
(
5) Tian, N.; Zhou, Z.-Y.; Sun, S.-G.; Ding, Y.; Wang, Z. L. Science 2007,
(7) Mittendorfer, F.; Seriani, N.; Dubay, O.; Kresse, G. Phys. ReV. B 2007,
76, 233413.
3
16, 732.
(
6) Vaarkamp, M.; Miller, J. T.; Modica, F. S.; Koningsberger, D. C. J.
Catal. 1996, 163, 294.
(8) Nolte, P.; Stierle, A.; Jin-Phillip, N. Y.; Kasper, N.; Schulli, T. U.;
Dosch, H. Science 2008, 321, 1654.
10.1021/ja101997z  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 8747–8756 9 8747

A R T I C L E S
Roldan Cuenya et al.
octahedron shape), and their rounding and dewetting under O
2
.
in the most intriguing nm-size range (<2 nm) that is least
investigated experimentally. These methods open up the pos-
sibility of characterizing well-defined sizes and shapes of
individual NPs by performing ensemble measurements on bulk
quantities of nearly identical NPs.
In this article, we demonstrate how extended XAFS (EXAFS)
spectroscopy can be used to deduce the cluster shape in Pt/γ-
In analogy to the previous observation of Rh clusters, the latter
shape changes also appeared to be reversible. These studies
demonstrate that chemisorption-induced morphological changes
in NPs need to be considered when models to explain catalytic
reactivity are proposed, since certain reaction environments
might lead to a decrease/increase in the relative area of the most
catalytically active surface facets/sites.
2 3
Al O nanocatalysts synthesized by inverse micelle encapsula-
Although significant progress in the in situ structural char-
acterization of nanomaterials has been made in the past
tion. In addition, by tuning NP synthesis parameters such as
the molecular weight of the encapsulating polymer or the metal/
polymer ratio, we were able to synthesize Pt NPs with
octahedron, truncated octahedron, and cuboctahedron shapes.
Nanoparticle shape determination in these samples is achieved
here through an integrated methodology that combines multiple-
scattering analysis of EXAFS data with results obtained by other
complementary techniques including scanning transmission
electron microscopy (STEM), scanning tunneling microscopy
(STM), and atomic force microscopy (AFM). In our XAFS
experiments, the sensitivity to shape and local structure was
greatly enhanced by carrying out the measurements under a
hydrogen atmosphere that partially lifts surface disorder. Our
measurements demonstrate that our NP preparation method
(micellar encapsulation) enables the achievement of predomi-
nantly high surface area 3D NPs in the nm-size range of interest
to most catalysis applications, instead of the conventional 2D
7
-16
decade,
most of the studies have been unable to quantify
geometrical properties of real-world nanocatalysts. The major
challenge toward the goal of precise three-dimensional char-
acterization is the requirement to discern many competing
influences on the NP structure and shape, most notably the
17
effects of the substrate, absorbate, size, and alloying elements
1
8
and their spatial distribution within the NP. While the studies
19
of size-selected clusters by advanced electron microscopy and
2
0
diffraction methods offer a high level of detail, and some of
these methods are now used in reactive environments and
variable temperatures, their spatial resolution is currently limited
to about 0.07 Å. X-ray absorption fine-structure (XAFS)
spectroscopy is a powerful alternative and/or complement to
imaging methods due to its sensitivity to the local atomic
environment (the bond lengths can be determined with the
accuracy of 0.01 Å or better), adaptability to a wide range of
nanocatalysts (ideally in the size range from subnanometer to
raft-like Pt and Ru NPs that are obtained on γ-Al
deposition-precipitation.
2
O
3
through
This suggests that the cluster-
support interactions that are thought to influence the dispersion
1
5,17
∼
5 nm), and in situ experimental conditions. However, one
2
6
limitation of this technique is its ensemble averaging of the local
structure over all absorbing atoms in the sample, rendering ill-
defined structural results for systems with broad size distribu-
2 3
and shape of nanoscale Pt catalysts on Al O may be tuned
by changing the sample preparation method.
2
1
tions. Only in the case of well-defined model NPs (in the size
range of 2 nm and larger) was it possible to explore the greatest
potential of this technique, using it to determine the size, shape,
structure, and surface orientation of a very few model systems
2. Experimental Methods
2.1. Sample Preparation. Size-selected Pt NPs were chemically
synthesized by inverse micelle encapsulation. The dissolution
of nonpolar/polar diblock copolymers [Poly(styrene)-block-poly-
1
6,22
made just for this purpose.
Recently, new approaches for
(
(
2vinylpyridine), Polymer Source Inc.] in a nonpolar solvent
toluene) leads to the formation of spherical nanocages known as
the fabrication of nanostructures have enabled us to produce
model catalyst samples with narrow NP size and shape
reverse micelles. Size-selected Pt NPs are formed when the former
2
3-25
distributions.
These synthesis methods are particularly
polymeric solution is loaded with H PtCl ·6H O. In these samples,
2
6
2
attractive for systematic investigations of the properties of NPs
the particle size was changed by using two polymers with different
head lengths (P2VP group), namely, PS(27700)-P2VP(4300) (sample
S1) and PS(16000)-P2VP(3500) (samples S2, S3, S4, S5), as well
as by modifying the metal-salt/polymer-head (P2VP) ratio (L). The
following L ratios were used in the present study: 0.05 (S2), 0.1
(
9) Giorgio, S.; Cabie, M.; Henry, C. R. Gold Bull. 2008, 41, 167.
(
10) Giorgio, S.; Sao Joao, S.; Nitsche, S.; Chaudenson, D.; Sitja, G.; Henry,
C. R. Ultramicroscopy 2006, 106, 503.
(
11) Wang, Q.; Hanson, J. C.; Frenkel, A. I. J. Chem. Phys. 2008, 129,
(
S3), 0.2 (S1, S4), and 0.4 (S5). Two sets of analogous samples
were prepared using these micellar-NP solutions; one set supported
on nanocrystalline γ-Al
powder (∼40 nm) for EXAFS and TEM
measurements/analysis (Table 1), and a second, analogous set
supported on single crystals [SiO /Si(100) and TiO (110)] for AFM
2
34502.
(
(
(
12) Helveg, S.; Hansen, P. L. Catal. Today 2006, 111, 68.
13) Newton, M. A. Chem. Soc. ReV. 2008, 37, 2644.
2 3
O
14) Newton, M. A.; Belver-Coldeira, C.; Martinez-Arias, A.; Fernandez-
Garcia, M. Nat. Mater. 2007, 6, 528.
2
2
(
(
(
(
(
15) Karim, A.; Prasad, V.; Mpourmpakis, G.; Lonergan, W. W.; Frenkel,
A. I.; Chen, J. G.; Vlachos, D. G. J. Am. Chem. Soc. 2009, 131, 12230.
16) Frenkel, A. I.; Hills, C. W.; Nuzzo, R. G. J. Phys. Chem. B 2001,
and STM measurement/analysis (Table 2). The second set was made
to obtain complementary height information since TEM only gives
information about the diameter of our particles, and our powder
samples are not AFM/STM compatible. Further sample preparation
details can be found in Tables 1, 2, and refs 23-25, 27. The
metal-polymeric solutions for the EXAFS/TEM samples were
1
05, 12689.
17) Sanchez, S. I.; Menard, L. D.; Bram, A.; Kang, J. H.; Small, M. W.;
Nuzzo, R. G.; Frenkel, A. I. J. Am. Chem. Soc. 2009, 131, 7040.
18) Knecht, M. R.; Weir, M. G.; Frenkel, A. I.; Crooks, R. M. Chem.
Mater. 2008, 20, 1019.
mixed with the γ-Al
S1) and 1 wt.% (S2, S3, S4, S5) Pt. The samples were stir-dried
at
2 3
O powder resulting in a loading of 2 wt %
19) Li, Z. Y.; Young, N. P.; Di Vece, M.; Palomba, S.; Palmer, R. E.;
Bleloch, A. L.; Curley, B. C.; Johnson, R. L.; Jiang, J.; Yuan, J. Nature
(
2
008, 451, 46.
in air at ∼60 °C for 24 h, and subsequently calcined in O
2
(
20) Huang, W. M.; Sun, R.; Tao, J.; Menard, L. D.; Nuzzo, R. G.; Zuo,
J. M. Nat. Mater. 2008, 7, 308.
375-425 °C for 24 h for the removal of the encapsulating polymer.
The total gas flow during calcination in a packed-bed reactor was
(
(
(
(
21) Frenkel, A. I. Z. Kristallogr. 2007, 222, 605.
5
2
0 mL/min, with 50-70% O balanced by He. The process for
22) Frenkel, A. I. J. Synchrotron Radiat. 1999, 6, 293.
23) Ono, L. K.; Sudfeld, D.; Roldan Cuenya, B. Surf. Sci. 2006, 600, 5041.
24) Naitabdi, A.; Ono, L. K.; Behafarid, F.; Roldan Cuenya, B. J. Phys.
Chem. C 2009, 113, 1433.
2
polymer removal from the NPs deposited on single crystals [SiO /
(26) Nellist, P. D.; Pennycook, S. J. Science 1996, 274, 413.
(
25) Naitabdi, A.; Behafarid, F.; Roldan Cuenya, B. Appl. Phys. Lett. 2009,
(27) Croy, J. R.; Mostafa, S.; Heinrich, H.; Roldan Cuenya, B. Catal. Lett.
2009, 131, 21.
9
4, 083102.
8
748 J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010

Solving the Structure of Size-Selected Pt Nanocatalysts
A R T I C L E S
a
Table 1. Description of Synthesis Parameters and Size (TEM) Information of Micellar Pt NPs Supported on Nanocrystalline γ-Al
2 3
O
synthesis details of the “as-prepared” Pt NPs/γ-Al TEM analysis after annealing + XAFS measurements in H
2
O
3
2
at 375 °C
sample name
polymer
L
annealing T (°C) [(% O
2
)]
TEM diameter (nm)
TEM weighted diameter (nm)
S1
S2
S3
S4
S5
PS(27700)-P2VP(4300)
PS(16000)-P2VP(3500)
PS(16000)-P2VP(3500)
PS(16000)-P2VP(3500)
PS(16000)-P2VP(3500)
0.2
0.05
0.1
0.2
0.4
375 [50]
375 [50]
425 [70]
400 [50]
375 [50]
1.0 (0.2)
1.0 (0.2)
1.0 (0.2)
0.9 (0.2)
1.7 (1.5)
1.2 (0.2)
1.1 (0.2)
1.2 (0.3)
1.0 (0.2)
5.7 (2.2)
a
The TEM values correspond to NP diameters. The volume-weighted TEM diameters are also indicated. The parameter “L” represents the metal-salt
to polymer ratio used in the NP synthesis.
Table 2. Height Analysis of Pt NPs Supported on SiO
2
/Si(100)
source (Al KR, 1486.6 eV, SPECS GmbH) operating at 250 W. A
flood gun was used during measurements to correct for sample
a
(
2
AFM) and TiO (110) (STM)
charging. All spectra were referenced to the Al-2s peak in Al
at 119.2 eV.
2 3
O
complementary size analysis
STM height (nm)
XAFS measurements were carried out at beamline X18B at the
National Synchrotron Light Source (NSLS) at Brookhaven National
Laboratory (Upton, NY). The data were collected in transmission
AFM height (nm)
after RT O -plasma (UHV)
after RT O
UHV annealing (°C)
2
-plasma +
analogous sample
2
S1
S2
S3
S4
1.2 (0.5)
1.8 (0.4) (500 °C)
0.9 (0.3) (1000 °C)
1.0 (0.3) (500 °C)
1.3 (0.3) (500 °C)
mode using the Pt L
pressing the Pt/γ-Al
3
edge. The XAFS samples were prepared by
powders into thin pellets which were
2 3
O
mounted in a sample cell that permitted sample heating via an
external PID controller, liquid nitrogen cooling, as well as the
continuous flow of gases during data acquisition and their online
1
.3 (0.4) (900 °C)
1.5 (0.4) (500 °C)
.5 (0.4) (700 °C)
S5
1.1 (0.4)
1
6
1
mass spectrometer analysis. A bulk Pt foil was measured in
transmission simultaneously with all samples for energy calibration
purposes. The EXAFS data were collected at RT in He (as-prepared)
a
The AFM and STM values correspond to the NP heights measured
-plasma treatment at RT (AFM),
and after subsequent annealing in UHV at 500°C (S1, S3, S4), 900°C
S4), and 1000°C (S2) (STM).
at RT after polymer removal by an O
2
2
and after annealing/reduction at 375 °C in 50% H balanced with
He for a total flow rate of 50 mL/min. EXAFS data from the
(
reduced samples where acquired at temperatures ranging from -100
2
to 375 °C in 50% H (50 mL/min, He balance) under the same gas
Si(100) and TiO
involved an O -plasma treatment in UHV at room temperature (RT)
4.0 × 10 mbar for 2 h).
.2. Sample Characterization. The AFM/STM NP solutions
were dip-coated onto SiO /Si(100) and TiO (110) substrates for
morphological characterization by ex situ atomic force microscopy
AFM, Digital Instruments-Nanoscope III) and in situ scanning
tunneling microscopy (STM, SPECS-Aarhus), respectively (Table
). The AFM images were taken after polymer removal in ultrahigh
vacuum (UHV) by an O -plasma treatment. The SiO /Si(100)
2
(110)] used in the AFM/STM measurements
flow conditions. Multiple scans (up to 6) were collected at each
temperature of interest and averaged in order to improve signal-
to-noise ratios.
2
-5
(
2
2
2
3. Data Analysis
(
3
.1. EXAFS. Data processing and analysis was done using
28
29
the Athena and Artemis software from the IFEFFIT package.
2
EXAFS data in k-space were obtained from raw absorption
coefficient data by subtracting a smooth, isolated-atom back-
2
2
substrate was chosen for AFM measurements because of its low
30
31
surface roughness (∼0.2 nm), while STM measurements require a
ground by the AUTOBK method. The FEFF6 code was used
to construct theoretical EXAFS signals that included single- and
multiple-scattering contributions from atomic shells through the
fourth nearest neighbors in the face-centered cubic (fcc) structure
of Pt. To minimize the temperature-dependent bond length
disorder contributions, such multiple-scattering analysis was
done only for the low temperature (-85 to -108 °C and RT)
data. The typical k-range used in the fitting was from 2.5 to 19
2
conducting substrate; therefore, TiO (110) was used. Furthermore,
after UHV treatments and subsequent air exposure, the TiO
substrate becomes too rough for reliable AFM measurements
considering our small particle sizes. Thus, AFM measurements were
2
not carried out on the TiO
measured in UHV after atomic oxygen exposure and annealing in
order to minimize artifacts arising from the enhanced TiO
roughness that resulted from the in situ O -plasma treatment as well
as to improve the tunneling conditions. The annealed TiO substrates
2
substrates. The STM samples were
2
2
-
1
Å , and the r-range from 1.7 to 5.7 Å. Thus, the total number
of relevant independent data points in the analysis (Nidp
2
)
are partially reduced and therefore more conducting, facilitating
thus the STM measurements.
TEM samples were prepared by making an ethanol suspension
of the Pt/γ-Al O powder, after EXAFS measurements, by placing
2 3
a few drops of this liquid onto an ultrathin C grid, and allowing
the sample to air-dry. High-angle annular dark-field (HAADF)
3
2
2∆k∆r/π) was of the order of 40, depending on the sample,
while the number of variables in the fit was much smaller,
namely, 13. To extract local structural information from the
reduced samples, we constructed theoretical EXAFS signals
from the contributions of the most important photoelectron
propagators (paths) for the fcc structure. They are the following:
single scattering (SS) paths to the first four nearest neighbors,
x f 1NN f x, x f 2NN f x, x f 3NN f x, x f 4NN f
2 3
images of the Pt/γ-Al O samples were acquired under scanning
mode within a JEM 2100F TEM, operated at 200 kV. The probe
size of the STEM is about 0.2 nm. The Pt NP diameters were
determined by measuring the full width at half-maximum of the
HAADF intensity profile across the individual Pt NPs (Table 1).
X-ray photoelectron spectroscopy (XPS) measurements were
carried out on each powder sample before the XAFS measurements
to monitor the removal of the encapsulating polymer (C-1s signal)
as well as any possible Cl residues (Cl-2p) from the NP synthesis
(28) Ravel, B.; Newville, M. J. Synchrotron Radiat. 2005, 12, 537.
(29) Newville, M. J. Synchrotron Radiat. 2001, 8, 322.
(
(
(
30) Newville, M.; Livins, P.; Yacoby, Y.; Rehr, J. J.; Stern, E. A. Phys.
ReV. B 1993, 47, 14126.
31) Zabinsky, S. I.; Rehr, J. J.; Ankundinov, A.; Albers, R. C.; Eller, M. J.
Phys. ReV. B 1995, 52, 2995.
(
see Supplemental Figure 1). No contaminants were detected on
these samples after the calcination treatment in O described above.
The XPS data were collected in UHV using a monochromatic X-ray
2
32) Brillouin, L. Science and Information Theory; Academic Press: New
York, 1962.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010 8749

A R T I C L E S
Roldan Cuenya et al.
x (where x denotes the X-ray absorbing atom); double scattering
focusing (DSF) paths, including the time-reversed ones, to the
4
NN through the intervening 1NN, x f 1NN f 4NN f x;
triple-scattering focusing (TSF) paths corresponding to the same
group of neighbors, except for the forward scattering through
the intervening atom twice (x f 1NN f 4NN f 1NN f x);
and, finally, the triple-scattering focusing paths from the central
atom to its 1NN on each side (1NN and 1NN*), x f 1NN f
x f 1NN* f x. This path selection is similar to that described
in the previous works analyzing high order contributions to fcc
1
6,22,33
NPs.
Each path degeneracy was allowed to vary in the
fit to account for size effects that cause surface atoms to be
less coordinated than those in the particle interior. Constraints
between coordination numbers were set to follow ref 22. The
DSF and TSF paths to the 4NN had their mean square disorder
parameters constrained to be the same as that of the SS path to
3
4
the same neighbors. Finally, the nearest neighbor distances,
except for the 1NN, were constrained to follow the isotropic
bond distortion model, maintaining the overall fcc symmetry.
This assumption is validated later by the high crystalline order
obtained in the reduced NPs, Vide infra. The passive electron
2
reduction factor (S
0
) was fixed to 0.861, as obtained from
the multiple-scattering fitting analysis of the bulk Pt foil. The
correction to the photoelectron energy origin was varied in the
fit, and was kept the same for all paths. We have also included
the third cumulant of the 1NN pair distribution function. The
latter parameter was not found to affect the results of the fits
significantly (since we analyzed the data collected at relatively
low temperatures where that parameter is not significant), and
the final results are shown without accounting for the third
cumulant. The same model was applied to the analysis of the
bulk Pt data for calibration, and the results obtained were found
to be in good agreement with its known coordination numbers
and NN distances. For the as-prepared samples, we constructed
the fitting model by accounting for the Pt-O contribution that
is evident in the raw data. That contribution was modeled by
Figure 1. AFM images of micellar Pt NPs supported on SiO
2
/Si(100) taken
ex situ after polymer removal by an O -plasma treatment. The samples were
2
synthesized with the same NP solutions used for the EXAFS characteriza-
tion. Image (a) corresponds to NPs synthesized by encapsulation with
PS(16000)-P2VP(3500) (analogous to S5) and (b) on PS(27700)-P2VP(4300)
(
analogous to S1).
2 6
FEFF6 calculations done for the crystal structure of Na Pt(OH) .
includes octahedron, cuboctahedron, truncated octahedron, and
hexahedron shapes, as well as combinations of the above (mixed
shapes). In a second step, the symmetric shapes previously
described were truncated (layer-by-layer) following (111) and
In addition to Pt-O, only 1NN Pt-Pt contributions were
included in the analysis. The total number of relevant indepen-
dent data points was 13, which is much larger than the total
number of fitting variables (7).
(100) facets. In total, we have considered 131 symmetric shapes,
3
.2. Nanoparticle Shape Modeling. A MATLAB code was
1
643 shapes truncated along (111), and 1715 shapes truncated
written to obtain coordination numbers corresponding to dif-
ferent Pt NP shapes with fcc structure. A shape database was
generated starting from a bulk fcc structure that was subse-
quently truncated along two different planes, {111} (8 facets)
and {100} (6 facets). Higher Miller index facets have not been
considered, since they are energetically less favorable and have
not been observed in previous high resolution STM and TEM
along (100) facets. Hexahedron (cubic) shapes have not been
included in the subsequent analysis since such shapes were not
previously observed for analogously prepared but larger (∼3
nm) Pt NPs via STM. For each of the remaining shapes, the
NP height, diameter, total number of atoms, and average nearest
neighbor coordination numbers (N1-N4) were calculated and
compared to the experimental values obtained by EXAFS
3
5-39
studies of particles within the same size range.
Distinct
(
coordination numbers) and TEM (NP diameter). Some pre-
NP shapes were obtained by changing the distance of each of
the above facets to the center of the NP. The resultant database
liminary analysis was also done for the cubic shapes, but none
of them were found to lead to a better agreement with the
EXAFS coordination numbers than the noncubic shapes reported
in the subsequent sections.
(
(
33) Nashner, M. S.; Frenkel, A. I.; Adler, D. L.; Shapley, J. R.; Nuzzo,
R. G. J. Am. Chem. Soc. 1997, 119, 7760.
34) Shanthakumar, P.; Balasubramanian, M.; Pease, D.; Frenkel, A. I.;
Potrepka, D.; Budnick, J.; Hines, W. A.; Kraizman, V. Phys. ReV. B
4. Results
2
006, 74, 174106.
(
35) Dulub, O.; Hebenstreit, W.; Diebold, U. Phys. ReV. Lett. 2000, 84,
4
.1. Morphological and Structural Characterization (AFM,
3
646.
STM, STEM). Figure 1 shows AFM images of samples prepared
with NP solutions analogous to the XAFS samples S5 (a) and
(
36) Berko, A.; Szoko, J.; Solymosi, F. Surf. Sci. 2003, 532-535, 390.
(37) Ramachandran, A. S.; Anderson, S. L.; Datye, A. K. Ultramicroscopy
1
993, 51, 282.
38) Giorgio, S.; Henry, C. R.; Chapon, C. J. Cryst. Growth 1990, 100,
54.
39) Graoui, H.; Giorgio, S.; Henry, C. R. Surf. Sci. 1998, 417, 350.
S1 (b) but supported on SiO
of the polymeric shell by an in situ O
leaves the NPs highly oxidized (see XPS data of Supplemental
2
/Si(100) taken after the removal
(
(
2
-plasma at RT, which
2
8
750 J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010

Solving the Structure of Size-Selected Pt Nanocatalysts
A R T I C L E S
general trend, since its 2D nature is attributed at least partially
to a strong NP/support interaction (Pt/γ-Al O ) that might not
2 3
be present on the other two supports. The differences observed
for S1 are due to the fact that two different NP solutions (of
similar characteristics) were used to prepare each sample.
Histograms of the STM heights can be found in Supplemental
Figure 4.
Figure 3 displays examples of HAADF-STEM images obtained
for micellar Pt NPs supported on nanocrystalline γ-Al
HAADF images correspond to samples S1 (a) and 2 (b) and were
acquired after polymer removal by a prolonged annealing in O at
75 °C and subsequent in situ XAFS characterization after
reduction in H at 375 °C. Histograms of the Pt NP diameters as
2 3
O . The
2
3
2
measured from the HAADF images are included in Figure 3, panels
c (300 NP count) and d (602 NP count), respectively. Size
histograms extracted from HAADF measurements of the rest of
the samples discussed in this manuscript can be found in Supple-
mental Figure 5 and the average diameters in Table 1. With the
exception of sample S5, where coarsening seems to have occurred
during the polymer removal pretreatment leading to a certain
population of large NPs, the rest of the samples are characterized
by narrow NP size distributions with NPs smaller than 1.5 nm.
4
.2. Structural, Electronic, and Chemical Characterization
(XANES, EXAFS). Figure 4 shows X-ray absorption near-edge
structure (XANES) spectra of the Pt L edge of the micellar Pt
NPs supported on γ-Al acquired (a) before and (b) after in
situ reduction in H at 375 °C. Reference data obtained on a Pt
2
Figure 2. (a and b) STM images of micellar Pt NPs on TiO (110) after
2
polymer removal by an O -plasma treatment and subsequent annealing in
3
UHV at 900 °C. The NP solution used to prepare this sample is similar to
that of sample S1. Image (c) shows the shape resolved STM image of a
faceted micellar NP prepared using the polymer PS(81000)-P2VP(14200).
Image (d) shows the facet orientations of an atomic model of a particle
constructed based on an fcc crystalline structure.
2
O
3
2
foil are also displayed. The resonance peak at the absorption
edge is known as the white line (WL) and it arises from 2p3/2
to 5d5/2, 5d3/2 transitions. The intensity of the WL increases with
decreasing d-band occupancy, and it is very strong for transition
metals with a partially filled d band (e.g., Pt-5d5/2 orbital). The
area under the WL can be used to extract information on
Figure 2). Narrow NP size distributions are observed for both
samples, with average NP heights of ∼1.2 and ∼1.1 nm,
respectively (see Supplemental Figure 3). The use of two
polymers with distinct tail lengths in the synthesis leads to two
different average interparticle distances, namely, ∼36 nm (S5)
and ∼43 nm (S1). Since in this NP size range ex situ AFM
characterization is challenging, our morphological analysis was
complemented by in situ STM measurements on similar NP
4
0
d-charge redistributions. The spectra in Figure 4a reveal the
strongly oxidized character of the as-prepared Pt NPs samples.
The WL intensities of the former samples are significantly higher
than those of the reduced samples, Figure 4b, due to the electron-
withdrawing nature of oxygen. This result is in agreement with
EXAFS data acquired for the same samples and will be discussed
in greater detail below. Furthermore, a size-dependence in the WL
intensity is observed, with the lowest value measured for the
largest NPs (S5, 1.7 ( 1.5 nm), followed by the 0.9-1.0 ( 0.2
nm NPs (S3, S4). Surprisingly, similarly sized NPs in S2 (1.0
2
solutions dip-coated on TiO (110). STM images were acquired
after polymer removal by atomic oxygen exposure and subse-
quent sample annealing in ultrahigh vacuum at 500 °C (30 min)
and 900 °C (20 min), which also reduces any oxides from the
plasma treatment (Supplemental Figure 2). Figure 2 displays
examples of STM images acquired on the micellar Pt NPs of
sample S1 after annealing at 900 °C. For a given sample, no
clear changes in the STM NP height were observed when
comparing samples annealed at 500, 700, 900, and 1000 °C,
indicating the lack of NP coarsening and/or atomic desorption.
In this sample, the average NP height at 900 °C was 1.8 ( 0.4
nm. We mention that, even though the AFM-measured (air
exposed) samples were not reduced (i.e., not annealed), and
contain an oxide shell as evidence by XPS (see Supplemental
Figure 2), good agreement is observed between the NP heights
obtained by AFM (oxidized) and STM (reduced), Table 2.
Therefore, we assume that the presence of oxides does not
influence the AFM size within the errors given. It is possible
that due to the single crystal supports being different than our
(
0.2 nm) showed a clearly higher WL intensity. The more
facile oxidation of small metal NPs can explain the differences
between samples S5 and S3, S4, but not the higher WL intensity
of S2. We will elaborate on this point in the discussion section.
2
After NP reduction in H , the WL intensity of all NP samples
is significantly decreased, reaching the bulk-Pt value. In
41-48
agreement with previous reports,
we also observe the shift
(
(
40) Prins, R.; Koningsberger, D. C. In X-ray Absorption. Principles,
Applications, Techniques of EXAFS, SEXAFS, and XANES; Prins, R.,
Ed.; Wiley: New York, 1988; Chapter 8.
41) Vaarkamp, M.; Miller, J. T.; Modica, F. S.; Lane, G. S.; Koningsberger,
D. C. Jpn. J. Appl. Phys. 1993, 32, 454.
(42) Lytle, F. W.; Greegor, R. B.; Marques, E. C.; Sandstrom, D. R.; Via,
G. H.; Sinfelt, J. H. J. Catal. 1985, 95, 546.
2 3
γ-Al O support the AFM/STM heights may not be representa-
(
(
43) Samant, M. G.; Boudart, M. J. Phys. Chem. B 1991, 95, 4070.
44) Ichikuni, N.; Iwasawa, Y. Catal. Lett. 1993, 20, 87.
tive of the height of the NPs analyzed by EXAFS. However,
because these two different methods (AFM/STM) give similar
results under differing conditions (unannealed/UHV-annealed
and on two different supports), we believe the measurements
(45) Allen, P. G.; Conradson, S. D.; Wilson, M. S.; Gottesfeld, S.; Raistrick,
I. D.; Valerio, J.; Lovato, M. Electrochim. Acta 1994, 39, 2415.
(
(
46) Asakura, K.; Kubota, T.; Chun, W. J.; Iwasawa, Y.; Ohtani, K.;
Fujikawa, T. J. Synchrotron Radiat. 1999, 6, 439.
47) Kubota, T.; Asakura, K.; Ichikuni, N.; Iwasawa, Y. Chem. Phys. Lett.
1996, 256, 445.
to be reasonable representatives for the 3D-shaped γ-Al
2 3
O -
supported NPs (S1, S3, S4, S5). S2 is an exception to this
J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010 8751

A R T I C L E S
Roldan Cuenya et al.
2 3 2
Figure 3. HAADF STEM images of micellar Pt NPs on γ-Al O : (a) S1, and (b) S2. The images were taken after polymer removal by annealing in O at
3
75 °C. Histograms of the corresponding NP diameter distributions are shown in panels c and d.
of the peak position to higher energies (relative to the bulk),
and the overall peak broadening for all NP samples. The origins
of these effects are subjects of intense debate in the litera-
4
3,44,46,49-55
ture.
The blue shift of the WL observed for the NP
samples relative to the Pt foil [inset in Figure 4b] is most
noticeable for S2 (+1.2 eV) but also present for the other
samples (+0.5 eV for S1). We note here that the setup at
beamline X18B along with our calibration procedures allow us
to discern relative energy shifts with a sensitivity better than
0
.1 eV.
Figure 5 displays k -weighted EXAFS data in k-space (a) and
r-space (b) acquired at RT in H for all reduced samples. A
2
2
comparison with the data obtained for the oxidized (as-prepared)
samples at RT in He is shown in the inset of Figure 5b. For
NPs of identical shape, the amplitudes of the EXAFS oscillations
in Figure 5a are related to distinct values of average NP size
and bond length disorder, with small (<5 nm) and most
disordered NPs showing significantly damped signals. For our
samples, S5 appears to contain large NPs (similar signal to the
(
(
48) Yoshitake, H.; Mochizuki, T.; Yamazaki, O.; Ota, K. J. Electroanal.
Chem. 1993, 361, 229.
49) Vaarkamp, M.; Modica, F. S.; Miller, J. T.; Koningsberger, D. C. J.
Catal. 1993, 144, 611.
(
(
50) Otten, M. M.; Clayton, M. J.; Lamb, H. H. J. Catal. 1994, 149, 211.
51) Vaarkamp, M.; Mojet, B. L.; Kappers, M. J.; Miller, J. T.; Konings-
berger, D. C. J. Phys. Chem. B 1995, 99, 16067.
Figure 4. XANES region [µ(E) versus E] of (a) oxidized “as-prepared”
micellar Pt/γ-Al samples at RT in He, and (b) reduced samples at RT
in H measured after annealing in H at 375 °C. Similar data from a Pt foil
are also displayed. The inset in panel b shows a positive shift in the energy
of the Pt L edge of Pt NPs as compared to the bulk Pt reference.
(
(
(
(
52) Ankudinov, A. L.; Rehr, J. J.; Low, J. J.; Bare, S. R. J. Synchrotron
Radiat. 2001, 8, 578.
2 3
O
2
2
53) Ankudinov, A. L.; Rehr, J. J.; Low, J. J.; Bare, S. R. Top. Catal. 2002,
1
8, 3.
3
54) Soldatov, A. V.; Della Longa, S.; Bianconi, A. Solid State Commun.
993, 85, 863.
55) Matsuura, A. T.; Fujikawa, T.; Kuroda, H. J. Phys. Soc. Jpn. 1983,
2, 3275.
1
bulk Pt foil), while the EXAFS data from sample S2 indicate
the smallest and/or most atomically disordered NPs among all
5
8
752 J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010

Solving the Structure of Size-Selected Pt Nanocatalysts
A R T I C L E S
5. Discussion
As was mentioned in the previous section, the unusually high
WL intensity of S2 as compared to samples S3 and S4 [Figure
a, as-prepared samples] cannot be understood simply based
4
on the stronger degree of oxidation of smaller NPs, since all
three samples were found to contain relatively monodispersed
∼
1 nm (diameter) NPs according to TEM. Instead, we attribute
the enhanced density of unoccupied 5d states in S2 to a distinct
shape of the NPs in this sample, namely, a flatter geometry with
the same apparent TEM diameter but an enhanced NP/support
contact area. Indeed, such model is consistent with the enhanced
WL intensity, since XANES data contain ensemble-averaged
information about the density of unoccupied Pt-5d states. Thus,
the greater the fraction of Pt atoms with a large 5d-hole density
(
i.e., those which are oxidized, for example via support
interaction), the larger the WL intensity. This model is also
consistent with the TEM-obtained average NP diameter. One
possible explanation for the WL enhancement of the Pt atoms
at the NP-substrate interface is the large charge transfer from
5
8
Pt to defects in Al
2
O
3
.
The inset in Figure 5b, displaying
r-space EXAFS data from the as-prepared oxidized samples,
also reveals the highest Pt-O signal for sample S2, consistent
with the XANES observation.
A comparison of the r-space plots shown in Figure 5b for
the reduced micellar Pt NPs to that of a bulk Pt foil revealed
that despite their small size (∼1 nm for most samples), all of
our NPs have good crystalline fcc structure, in contrast with
previous literature reports on similarly sized NPs prepared by
1
7
2
deposition-precipitation methods. This difference could be
attributed to several factors: (i) the reverse micelle encapsulation
method leads to more ordered structures, (ii) the prolonged
annealing times (24 h) used during the pretreatment of the
micellar NP samples (before XAFS characterization) lead to
an enhanced long-range order, and (iii) the shape of most of
Figure 5. k -weighted EXAFS data in k-space (a) and r-space (b) for all
reduced samples in H
data from the as-prepared samples measured in He at RT. Fourier transform
parameters are as follows: the k-range is from 2 to 16 Å , the Hanning
window sills are ∆k ) 2 Å
2
at room temperature. The inset in panel b shows
-
1
-
1
.
samples. The same trend is observed in the r-space data of
Figure 5b. We point out here that, as discussed previously, the
coordination numbers of the 1NNs do not provide conclusive
information on the NP size, since many clusters of different
shapes and sizes may have the same 1NN coordination
our NPs (S1, 3, 4 and 5) is 3D, instead of raft-like (2D) as has
been commonly reported for Pt/γ-Al
strain from the larger NP/support contact area is expected to
lead to more disordered structures. In addition, it should be noted
that our XAFS measurements were carried out in a H
atmosphere, and that H has been reported to induce faceting
17
2
O
3
.
In the latter case,
2
56
number. Furthermore, visual examination of the data does not
allow one to easily discriminate among the individual effects
of size, shape, and atomic disorder. In this work, we use
multiple-scattering fitting analysis to extract first through fourth
coordination numbers for the quantitative analysis of NP shapes.
Details in this respect will be given in the next section. The
inset in Figure 5b reveals a strong Pt-O peak in the as-prepared
2
9,10
17
on NPs.
EXAFS results from Pt/C and Pt/γ-Al
and H flows, that H relieves surface-induced strain. Wang and
Sanchez et al. also suggested, by comparing
2
O
3
measured under He
2
2
59
Johnson have interpreted those results theoretically by at-
tributing the increased order in the clusters to removal of the
inherent {100}-to-{111} shear instability, stabilizing the trun-
cated cuboctahedral shape when the particles are passivated by
hydrogen. If the micellar NPs are 3D in nature, a higher surface
(
oxidized) samples measured in He. Interestingly, almost all of
the as-prepared samples (with the exception of sample S2)
display a well-ordered fcc structure. The highest Pt-O signal
and highest atomic disorder were observed for sample S2. We
would like to highlight that PtO cannot be distinguished from
based on EXAFS signals of the first shell, since the Pt-O
distances for both species are about 2 Å (2.04 Å for PtO and
.99 Å for ꢀ-PtO
can conclude, by analysis of the coordination numbers of Pt-O
and Pt-Pt, that the “as-prepared” Pt/γ-Al samples contained
a maximum of ∼65% PtOx and ∼35% metallic Pt (data obtained
from S2 reflecting the maximum relative content of Pt-O
species among all of our samples).
2
area is available for H absorption, and therefore, more ordered
faceted NPs will be present. Interestingly, sample S2 shows the
least resemblance to the Pt foil, and since the diameter of the
NPs in this sample is nearly identical to that in sample S1,
qualitatively, a more anisotropic shape might be inferred for
sample S2.
The EXAFS data from Figure 5 (RT measurements) as well
as low-temperature (-85 to -100 °C) data have been fit by
multiple scattering analysis. The excellent quality of the fits
obtained for two representative samples (S1 and S2) is
demonstrated in Figure 6. From these fits, coordination numbers
PtO
2
5
7
1
2
). Nevertheless, from our EXAFS data, we
2 3
O
(
(
56) Glasner, D.; Frenkel, A. I. AIP Conf. Proc. 2007, 882, 746.
(58) Kwak, J. H.; Hu, J.; Mei, D.; Yi, C.-W.; Kim, D. H.; Peden, C. H. F.;
Allard, L. F.; Szanyi, J. Science 2009, 325, 1670.
57) Seriani, N.; Jin, Z.; Pompe, W.; Ciacchi, L. C. Phys. ReV. B 2007, 76,
1
55421.
(59) Wang, L. L.; Johnson, D. D. J. Am. Chem. Soc. 2007, 129, 3658.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010 8753

A R T I C L E S
Roldan Cuenya et al.
from our model database, no shape was found to be an exact
match with the experimental data. This is a result of our
consideration of an extensive set of coordination numbers
(N1-N4) when looking for matching shapes (see Supplemental
Figure 5 for additional details). Only after taking into consid-
eration “relaxed” error bars of the experimental coordination
numbers were six shapes identified as reasonable candidates
having closely matching coordination numbers with the EXAFS
data. For each N value of each sample, we calculate the
difference between the model and the measured (EXAFS)
coordination number, normalized by the experimental error, and
the maximum value for each sample is the allowed relaxation
of the error bars. The best fit will be the one with the lowest
value of the allowed relaxation having a diameter in the size
range of the TEM measurements. However, of the six different
shapes identified for S1, only two were in agreement with the
TEM-measured diameter, and the final shape chosen (Figure
7
) was the one showing the best agreement with the EXAFS-
derived coordination numbers. Similar results were obtained for
samples S2-S4, each having similar narrow size distributions
as in S1. Additionally, since the coordination numbers extracted
from EXAFS data are averaged over all NPs in a particular
sample, volume-weighted average size (i.e., TEM-obtained
diameter) distributions need to be calculated using the number
of atoms in each NP (or volume) as a weighting factor. For
samples with narrow size distributions (S1, S2, S3, and S4),
the volume-weighted average size is only slightly larger than
that of the standard distribution, but in the case of sample S5,
where a broad NP size distribution is present, the average NP
size was found to be significantly higher, explaining thus the
large coordination numbers obtained by EXAFS.
2
Figure 6. Fourier transform magnitudes of the k -weighted EXAFS data
and multiple-scattering fit for the sample S1 measured at room temperature
(
a) and sample S2 measured at -85 °C (b).
of, and distances to, the first through the fourth nearest neighbor
shells have been extracted, Table 3, and used to gain information
on the NP shape by comparison with model fcc cluster shapes,
Figure 7. In Table 3, N1, N2, N3, and N4 represent the
coordination numbers of Pt-Pt coordination shells that cor-
respond to the first through the fourth nearest neighbors in the
fcc structure. The distances (r) to each of these neighboring
For sample S5, a significant number of large (coarsened) NPs
were detected by TEM (see Supplemental Figure 6), leading to
a broad volume-weighted NP size distribution and large
coordination numbers in EXAFS. Therefore, for this sample, a
large number of shapes were found to fit the EXAFS data
(within the error margins), and the model shape shown in Figure
7a is displayed only for illustration purposes. For the rest of
the samples, the discrimination between different model shapes
was much more conclusive, as discussed above, and the
following NP shapes were inferred from modeling: an octahe-
dron (S1, S4), cuboctahedron (S3), and a flat (2-layers high)
16
shells are also shown in Table 3. Frenkel et al. have previously
used the sequence of nearest neighbor atomic distances and
coordination numbers extracted from EXAFS measurements to
correlate different models for cluster shapes with average cluster
diameters obtained independently. In the present study, the
synergistic combination of the EXAFS multiple scattering
analysis, NP size information obtained by TEM, and calculations
of model clusters shapes has revealed the 3D cluster shape of
NPs in samples S1, S3, S4 and S5, and 2D raft-like NP shapes
for sample S2. The final NP shape determination is based on
the best agreement between EXAFS-obtained coordination
numbers and those derived for a particular model fcc Pt cluster
shape that would be in the size range indicated by our TEM
data. However, it should be mentioned that such analysis is only
meaningful if samples with narrow size distributions are
available (S1, S2, S3, S4), since only then will it be possible to
reduce the number of model cluster shapes that agree with the
EXAFS coordination numbers as well as TEM NP diameters.
In addition, the NP size should be small (<ca. 3 nm), since a
large degeneracy in shapes is otherwise obtained when coor-
dination numbers close to the bulk fcc metal are considered.
Even if all of these requirements are met, additional compro-
mises might be required, for example, ones that compensate
for strong distortions expected in the nanometer-scale clusters
1
5,61
truncated octahedron (S2).
The most common NP shapes,
observed by our group via STM for annealed micellar NPs
supported on TiO (110), were used as reference to discard
2
unstable cluster shapes consisting of nonenergetically favorable
facets. Our data analysis indicates the possibility of creating
NPs with distinct shapes after annealing (∼400 °C) using the
inverse micelle encapsulation method by changing the size of
the micellar core (different molecular weight of the P2VP block),
as well as by tuning the metal-salt to P2VP ratio (L). In
particular, our only 2D shape (S2) was obtained for the lowest
loading of the micellar nanocage (L ) 0.05) considered. More
importantly, our results demonstrate that the micelle encapsula-
tion synthesis method allows one to overcome strong NP/support
interactions that would normally prevent the stabilization of
high-surface area 3D cluster shapes. For NPs in the present size
range (∼1 nm), such 3D cluster shapes might display distinct
catalytic activity and selectivity as compared to 2D, raft-like
Pt/γ-Al O NPs prepared by deposition-precipitation methods
6
0
due to surface and substrate-induced strain. For example,
sample S1 has a size of 1.0 ( 0.2 nm (TEM diameter), and
2
3
due to the larger number of facets in 3D clusters.
(
60) Yevick, A.; Frenkel, A. I. Phys. ReV. B 2010, 81, 115451.
754 J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010
(61) Chang, C. M.; Chou, M. Y. Phys. ReV. Lett. 2004, 93, 133401.
8

Solving the Structure of Size-Selected Pt Nanocatalysts
A R T I C L E S
Table 3. Coordination Numbers (N) and Distances of Each Coordination Shell to the Absorbing Atom (r) Obtained from the Multiple
a
2 3
Scattering Analysis of EXAFS Data Acquired on Micellar Pt NPs Supported on γ-Al O
coordination numbers of the first through the fourth nearest neighbor shells, N, and the distance to each shell, r (Å)
N1
r1
N2
r2
N3
r3
N4
r4
sample
diameter (nm) TEM/model
S1
EXAFS
9.1 (0.5)
2.8 (1.0)
3.910(0.009)
3.1
9.8 (2.3)
4.800(0.006)
11.3
7.8 (2.0)
5.54(0.01)
5.2
1.2 (0.2)
2
.763(0.004)
Model
8.5
1.3
S2
EXAFS
7.3 (0.5)
1.3 (0.7)
3.918(0.004)
2.1
5.1 (1.8)
4.799(0.005)
7.0
4.4 (1.1)
5.541(0.006)
3.1
1.1 (0.2)
2
.754(0.006)
Model
7.0
1.5
S3
EXAFS
8.7 (0.6)
3.4 (1.7)
3.918(0.004)
3.3
6.9 (2.3)
4.799(0.005)
9.6
5.6 (1.4)
5.541(0.006)
4.2
1.2 (0.3)
2
.758(0.006)
Model
7.8
1.3
S4
EXAFS
10.1 (0.6)
5.6 (2.0)
3.918(0.004)
3.6
11.8 (3.2)
4.798(0.005)
13.4
8.3 (2.2)
5.541(0.006)
6.2
1.0 (0.2)
2
.760(0.005)
Model
9.1
1.5
S5
EXAFS
10.9 (1.0)
3.9 (5.6)
3.919(0.006)
5.0
14.1 (4.5)
4.800(0.007)
18.5
8.5 (2.7)
5.542(0.009)
9.2
5.7 (2.2)
2
.748(0.008)
Model
10.6
3.8
Pt bulk
EXAFS
12.0 (0.4)
5.0 (1.2)
3.923(0.003)
6
24 (4.0)
4.805(0.003)
24
10.8 (1.6)
5.548(0.004)
12
thin foil
2
.767(0.003)
Model
12
a
The samples underwent an initial annealing as given in Table 1 plus in situ annealing in H at 375 °C during XAFS measurements. The fits were
2
carried out on EXAFS spectra measured at low temperature (-85 to -108°C) for samples S2, S3, S4, and S5 and at room temperature for S1.
Coordination numbers obtained theoretically for relevant model cluster shapes displayed in Figure 7 are also shown. The model shapes have been
selected based on three criteria: (i) their agreement with the coordination numbers extracted by EXAFS, (ii) their agreement with the NP diameters
obtained by TEM, and (iii) their resemblance to real cluster shapes observed by STM on similarly synthesized Pt NPs after annealing (see example in
Figure 2). The experimental diameters displayed in the EXAFS rows correspond to the average diameters obtained from TEM measurements after
volume-weighting the particle diameter histograms. Uncertainties are shown in parentheses.
Table 4. First Nearest Neighbor (N1) Coordination Numbers of the
As-Prepared (oxidized) Micellar Samples Measured in He at RT
Together with Their Reduced Counterparts and a Reference Pt
a
2
Foil Measured in H at RT
samples
path
N1
R (Å)
Pt bulk
Pt-Pt
Pt-Pt
Pt-O
Pt-Pt
Pt-Pt
Pt-O
Pt-Pt
Pt-Pt
Pt-O
Pt-Pt
Pt-Pt
Pt-O
Pt-Pt
Pt-Pt
Pt-O
Pt-Pt
12.1 (0.2)
8.4 (1.2)
0.8 (0.4)
8.6 (0.2)
4.2 (1.5)
1.4 (0.5)
7.5 (0.3)
8.4 (1.4)
1.5 (0.5)
9.0 (0.3)
10.3 (1.2)
0.8 (0.3)
9.7 (0.4)
10.8 (1.0)
-
2.765 (0.004)
2.763 (0.005)
1.97 (0.02)
2.761 (0.005)
2.76 (0.01)
2.02 (0.02)
2.741 (0.006)
2.764 (0.006)
2.01 (0.02)
2.746 (0.005)
2.764 (0.004)
1.98 (0.02)
2.756 (0.006)
2.764 (0.003)
-
S1
S2
S3
S4
S5
as prepared (in He)
reduced (in H
2
)
as prepared (in He)
reduced (in H
2
)
as prepared (in He)
reduced (in H
2
)
as prepared (in He)
Figure 7. Model of the NP shapes that resulted in the best agreement with
the coordination numbers obtained from the fits to the EXAFS data. The
shapes are representative of the following samples: (a) S1, (b) S2, (c) S3,
reduced (in H
2
)
as prepared (in He)
(
d) S4, and (e) S5. Since a narrow size distribution was not obtained for
reduced (in H
2
)
10.4 (0.4)
2.767 (0.006)
sample S5, the shape shown in panel e only reflects the volume-weighted
size distribution for that sample, since large NPs in that sample will provide
the most significant contribution to the observed EXAFS spectra.
a
Interatomic distances (R) measured for both types of samples are
also shown.
62
rameters to be varied according to the correlated Debye model.
The EXAFS data of the as-prepared oxidized samples [inset
in Figure 5b] were also fit (only Pt-O and 1NN Pt-Pt
contributions were considered) and a comparison with the
corresponding reduced samples is shown in Table 4. The as-
prepared samples were measured only at room temperature,
while the reduced samples were also measured between -85
and -108 °C. Nevertheless, a comparison of the structure of
the oxidized versus the reduced samples requires that both types
of samples are analyzed at the same temperature. To minimize
uncertainties in the determination of the coordination numbers
of the reduced samples at room temperature, we have applied
a multiple-data fit procedure where data at different temperatures
were analyzed concurrently by constraining the disorder pa-
As a result, the room temperature data were analyzed with higher
accuracy than if processed independently. As can be seen in
Table 4, the Pt-Pt coordination numbers in the samples
containing the smallest NPs (e.g., S2, where this effect can be
best seen outside the error bars) were found to increase after
sample reduction, indicating that two different Pt species
coexisted in these samples before reduction, Pt-oxide and Pt-
21
metal, since no coarsening (that would also have led to higher
coordination numbers) was detected for these samples by TEM
upon annealing in H
2
. In fact, the sample temperature during
the H pretreatment in the EXAFS cell was always equal to or
2
(62) Sevillano, E.; Meuth, h.; Rehr, J. J. Phys. ReV. B 1979, 20, 4908.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010 8755

A R T I C L E S
Roldan Cuenya et al.
smaller than the maximum annealing temperature used for the
The present study highlights the importance of integrated
analysis methodology involving multiple complementary tech-
sample preparation (polymer removal in O below 400 °C) that
2
6
3
leads to the “as-prepared” samples. Interestingly, the direct
comparison of two samples with analogous average diameter
niques, and also the uniqueness of EXAFS, combined with
cluster shape modeling, for the extraction of information on
cluster shapes in size-selected NP samples. Such insight is
crucial for the understanding of structure-reactivity relation-
ships in real-world (3D) nanocatalysts in the ∼1 nm size range,
since such analysis is outside the resolution range of most
complementary techniques (such as STM and AFM for 3D
clusters) and constitute a serious challenge for state-of-the-art
high-resolution TEM microscopes.
(
S1 and S2) reveals that sample S2 has a significantly higher
percentage of Pt-O bonding with respect to Pt-Pt bonding
before reduction as compared to sample S1. A larger difference
in the Pt-Pt coordination numbers before and after reduction
is also observed for S2 as compared to S1. These trends suggest
the presence of interfacial Pt-oxides in this sample, validating
its presumably 2D shape. Therefore, the fact that all of our as-
deposited samples are strongly oxidized helps us to gain insight
into the NP/support interface as well as NP shape.
Acknowledgment. The authors would like to acknowledge the
excellent beamline support provided by Dr. Nebojsa Marinkovic.
This work has been made possible thanks to the financial support
of the Office of Basic Energy Sciences of the US Department of
Energy under grants DE-FG02-08ER15995 (B.R.C.) and DE-FG02-
6
. Conclusions
The structure (size and shape) of small (<2 nm) size-selected
micellar Pt NPs synthesized by inverse micelle encapsulation
has been resolved by a combination of microscopic (TEM,
AFM, STM), and spectroscopic (XANES and EXAFS) tools,
all integrated within a self-consistent geometric modeling
method.
0
3ER15476 (J.C.Y. and A.I.F.). Travel support from the National
Synchrotron Light Source at Brookhaven National Laboratory
Faculty-Student Research Support Program) and DOE’s Synchro-
(
tron Catalysis Consortium (DE-FG02-05ER15688) are greatly
appreciated. Use of NSLS was supported by DOE-BES (DE-AC02-
Our work demonstrates that the NP preparation method plays
a key role in the final shape of the synthesized NPs. Following
our synthetic approach, NP/support interactions that commonly
lead to the stabilization of 2D-raft-like cluster shapes in systems
98CH10866). NFCF at the University of Pittsburgh is acknowledged
for the use of JEM 2100F.
Supporting Information Available: XPS data from the C-1s,
Cl-2p, Pt-4d and Pt-4f core level regions of Pt NPs supported
2 3
such as Pt/γ-Al O can be overcome, resulting in faceted 3D
NPs that are ideal candidates for catalysis applications. Fur-
thermore, our EXAFS data evidence the high degree of atomic
ordering that can be achieved in very small NPs (∼1 nm) using
micelle encapsulation synthetic routes as compared to traditional
deposition-precipitation methods. This effect might be in part
due to the annealing pretreatment that our micellar samples
on γ-Al
distributions obtained by AFM [on SiO
TiO (110)] and of the diameter distributions obtained by TEM
on γ-Al ], and additional details on the selection of the model
2
O
3
and TiO
2
(110), histograms of the Pt NP height
2
/Si(100)] and STM [on
2
[
2 3
O
shapes. This material is available free of charge via the Internet
underwent, but also to their interaction with H
EXAFS measurements. H is known to induce NP faceting, and
due to their enhanced surface area, our 3D NPs should be more
prone to reconstructing under H as compared to traditional 2D
Pt/γAl clusters.
2
during the
at http://pubs.acs.org.
2
JA101997Z
2
2
O
3
(63) Billinge, S. J. L.; Levin, I. Science 2007, 316, 561.
8
756 J. AM. CHEM. SOC. 9 VOL. 132, NO. 25, 2010