Selective propene epoxidation on immobilized Au6-10 clusters: The effect of hydrogen and water on activity and selectivity

Article abstract of DOI:10.1002/anie.200804154

Epoxidation made easy: Subnanometer gold clusters immobilized on amorphous alumina result in a highly active and selective catalyst for propene epoxidation. The highest selectivity is found for gas mixtures involving oxygen and water, thus avoiding the us

Full text of DOI:10.1002/anie.200804154

Angewandte
Chemie
DOI: 10.1002/anie.200804154
Heterogeneous Catalysis
Selective Propene Epoxidation on Immobilized Au Clusters: The
6
–10
Effect of Hydrogen and Water on Activity and Selectivity**
Sungsik Lee, Luis M. Molina,* Marꢀa J. Lꢁpez, Julio A. Alonso, Bjørk Hammer, Byeongdu Lee,
Sꢂnke Seifert, Randall E. Winans, Jeffrey W. Elam, Michael J. Pellin, and Stefan Vajda
Propylene oxide (PO) is an important intermediate bulk
chemical that is used in the production of polyurethane and
polyols. This product is now commercially produced using
either the chlorohydrin or the hydroperoxide routes. Each of
selectivity (greater than 90%). Although those results make
the use of nanoscale gold very promising, several issues still
have to be clarified and improved to employ this catalyst in
practical applications. The selectivity of these particles is
extremely sensitive to their size and shape, with particles
smaller than 1.5–2.0 nm mainly producing propane and
particles larger than 4–5 nm assisting oxidation of propene
to CO and H O. The other limitation in this reaction is the
consumption of hydrogen, which should be as low as possible
or, if possible, be suppressed altogether for economical
reasons. As the stability of the catalysts should be improved
for practical applications, there is a need to study the
properties and detailed reaction mechanisms of these catalyst
systems, and to look for related new catalysts with improved
features.
[
1]
these routes has its own limitations, owing to the production
of undesired chlorinated byproducts or high cost of the H O2
2
reactant. A new possibility has arisen based on the experi-
ments performed by M. Haruta et al., wherein small gold
nanoparticles were used for the direct propene oxidation by
2
2
[
2]
an O /H mixture. This new catalyst, if supported on either
2
2
TiO or titanium silicalite zeolites, converts between 1–10%
2
propene (depending on the support) with a very high
[
*] Dr. L. M. Molina, Prof. M. J. Lꢀpez, Prof. J. A. Alonso
Departamento de Fꢁsica Teꢀrica, Atꢀmica y ꢂptica
Universidad de Valladolid, 47011 Valladolid (Spain)
Fax: (+34)983-423-013
Herein we present the results of an experimental and
theoretical study of the catalytic activity of soft-landed
subnanometer gold clusters (Au –Au ) for propene epoxida-
E-mail: lmolina@fta.uva.es
6
10
Dr. S. Lee, Dr. S. Vajda
Chemical Sciences and Engineering Division, Argonne National
Laboratory, 9700 South Cass Avenue, Argonne, IL 60439 (USA)
tion. Several studies have found a high catalytic activity for
[3]
similar systems in a number of reactions. Interestingly, for
small gold clusters, the activity does not seem to depend very
[
4]
Prof. B. Hammer
iNano and Department of Physics and Astronomy
University of Aarhus, Ny Munkegade, 8000 Aarhus C (Denmark)
strongly on the type of oxide support. Irreducible oxides are
as good as the reducible oxides as supports, as long as the
oxide surface contains defects that serve as traps and activate
Dr. B. Lee, Dr. S. Seifert, Dr. R. E. Winans
X-ray Sciences Division, Argonne National Laboratory
[5]
the catalysts. With this in mind, we designed a new type of
gold-cluster-based catalyst that is highly active and selective
for direct propene epoxidation. The work presented herein
comprises the following: 1) amorphous alumina films are
used as support instead of the usual titania-based oxides;
9700 South Cass Avenue, Argonne, IL 60439 (USA)
Dr. J. W. Elam
Energy Systems Division, Argonne National Laboratory
9700 South Cass Avenue, Argonne, IL 60439 (USA)
2
) subnanometer gold clusters are employed instead of larger
Dr. M. J. Pellin
Materials Sciences Division, Argonne National Laboratory
nanoparticles; and 3) water vapor can replace the expensive
and dangerous use of hydrogen in the gas mixture. Ab initio
DFT calculations comparing Au/TiO and Au /Al O support
the experimental results, assigning a higher activity to the
alumina-supported subnanometer gold clusters owing to
easier formation of reaction intermediates. The calculations
also confirm that different reaction mechanisms take place for
both types of catalysts.
9700 South Cass Avenue, Argonne, IL 60439 (USA)
Dr. S. Vajda
2
n
2
3
Center for Nanoscale Materials, Argonne National Laboratory
9700 South Cass Avenue, Argonne, IL 60439 (USA)
and
Department of Chemical Engineering, School of Engineering &
Applied Sciences, Yale University
9
Hillhouse Avenue, New Haven, CT 06520 (USA)
The fabrication of the supported gold model nanocatalysts
involves several steps. A thin three-monolayer (3ML) alu-
mina film was initially grown by atomic layer deposition
[
**] Work supported by Spanish MEC (MAT2005-06544-C03-01) and
JCyL (VA017A08) grants. L.M.M. acknowledges support from the
“
Ramon y Cajal” program. The work at Argonne National Laboratory
[
6]
was supported by the US Department of Energy, BES-Chemical
Sciences, BES-Materials Sciences, and BES-Scientific User Facilities
under Contract DE-AC-02-06CH11357 with UChicago Argonne, LLC,
Operator of Argonne National Laboratory. S.V. gratefully acknowl-
edges the support by the Air Force Office of Scientific Research.
B.H. acknowledges financial support from the Danish research
councils, FNU, DSF/NABIIT, and DCSC.
(ALD) on top of naturally oxidized silicon wafers, providing
a rough and amorphous support. Such morphology prevents
sintering of the subnanometer catalysts under reaction
conditions that usually lead to the loss of highly size-
dependent activity and selectivity. Earlier studies confirmed
the exceptional stability of the platinum clusters on ALD
[7]
alumina films. A distribution of cationic gold clusters in the
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200804154.
+
+
range Au6 to Au10 (hereafter referred to as Au ) were then
n
Angew. Chem. Int. Ed. 2009, 48, 1467 –1471
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1467

Communications
soft-landed on the alumina film, whilst still keeping surface
for C H /O /H O. Thus, we can conclude that under steady-
3 6 2 2
coverage below 0.03ML to prevent the clusters from sintering.
Finally, to fully prevent the clusters from sintering, an
additional ALD alumina layer (about 2ML thick) was
grown selectively to avoid the overcoating of the supported
state conditions, the presence of surface hydroxy groups at the
perimeter of the cluster is crucial. As the reaction proceeds,
the surface hydroxy groups are lost, and in absence of H or
2
H O are not replenished at the perimeter of the clusters.
2
Au clusters, leaving them firmly attached at the bottom of a
shallow well in the substrate. The reaction experiments were
carried out in a reactor by feeding various gas mixtures at
Different slopes below or above 1008C are found in all
reactivity plots, suggesting slightly different reaction mecha-
nisms with increasing temperature. Based on the above data,
the activation energy for the formation of propene oxide are
26–34 kJmol and 6–10 kJmol for the T< 1008C and T>
1008C regions, respectively. Acrolein production is strikingly
n
1
33 kPa (C H /O , C H /O /H , and C H /O /H O in helium),
3 6 2 3 6 2 2 3 6 2 2
À1
À1
and simultaneous X-ray scattering was used to monitor the
size of the gold clusters. No change in cluster size, monitored
by grazing-incidence small-angle X-ray scattering (GISAXS),
was observed over several hours during the experiments, thus
confirming that the immobilized gold clusters were resistant
to sintering. Sintering was found, however, in the absence of
the extra overlayer of alumina.
Figure 1a,b shows the turnover frequency (TOF) for
propene oxide and acrolein coproduct formation at various
compositions of the gas mixture, as a function of temperature
and of reaction time. In all cases, a large activity is found,
although qualitative differences are found in the stability. The
different, and for the propene/O feed there is a high activity,
2
which actually increases with time at 2008C. With co-fed H2,
much less acrolein was formed, and already at 1008C the
production levels off, indicating that excess oxygen in the
surface promotes acrolein, whereas the surface hydroxy
groups favor propene oxide formation. The most interesting
results are found if water vapor is co-fed, for which formation
of acrolein is almost completely suppressed.
Plotting the temperature-dependent ratio of propene
oxide versus acrolein (Figure 1c), an estimation of the
catalyst selectivity is obtained. Without H or H O, we find
propene/O mixture was initially very active at 2008C, but its
2
2
2
activity decreases steadily with time. Adding either H or H O
a 1:2 ratio, whereas selectivity slightly improves with H , and
2
2
2
reverses this trend, and the activity slowly increases with time
we obtain ratios between 1:1 and 2:1 depending on the
temperature. The most spectacular results are found for
water, for which up to 14:1 ratios are observed at 2008C. This
makes propene/O /H O an optimal mixture for practical
2
2
applications of this type of catalyst: the slightly smaller
activity for propene oxide production is well compensated by
the high selectivity. Copper catalysts also produce propene
[
8]
oxide without hydrogen, although with poorer selectivity.
To understand the activity of these new Au /Al O model
n
2
3
catalysts for propene epoxidation and the underlying mech-
anisms, we performed ab initio DFT simulations of the
adsorption of C H on a model of the alumina-supported
3
6
small gold nanoclusters. Such supported gold nanoclusters are
formed by taking a large slab of the a-Al O (0001) surface,
2
3
excavating a hole in it (keeping the slab at 2:3 stoichiometry),
and placing a Au7 cluster inside, followed by complete
relaxation of the structure (Figure 2a). The hole-induced
disorder results in significant reconstruction, making the
support chemically close to amorphous alumina. In addition,
we compare the results obtained with analogous simulations
for a model of TiO -supported gold nanoparticles, which is the
2
[
2]
prototypical gold catalyst for propene epoxidation. TiO2-
supported gold nanoparticles are formed by supporting a one-
dimensional gold rod on an anatase-TiO (101) slab (Fig-
2
ure 2b). This approach was motivated by the results of Nijhuis
[
9]
et al., which provide evidence for the active interfacial gold-
[
10]
oxide region (also active for CO oxidation)
that is
responsible for the epoxidation reaction between propene
and substrate oxygen.
For a more realistic modeling we added a sizable number
of hydroxy groups (up to 14), which are likely to adsorb on the
alumina surface in presence of O /H or O /H O. Table 1
Figure 1. a) Propene oxide formation as a function of temperature
2
2
2
2
(
left) and time (right, monitored at 2008C) for various compositions of
the gas feed. b) Formation of acrolein as a function of temperature
left) and time (right, monitored at 2008C) for various compositions of
the gas feed. c) Temperature-dependent ratio of propylene oxide to
acrolein.
C H /O , * C H /O /H , ~ C H /O /H O.
shows the energetics for the sequential dissociative adsorp-
tion of up to seven water molecules on the Au /Al O model
(
7
2
3
catalyst. Hydroxy groups are easily formed with binding
energies of 1–2 eV for the first four pairs and 0.5–1.0 eV for
&
3
6
2
3
6
2
2
3
6
2
2
1
468
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 1467 –1471

Angewandte
Chemie
of Au /Al O (Figure 2c,d), the change is almost barrierless,
7
2
3
with a highly stable final state (1.5–1.8 eV binding, depending
on the site around Au ). This conformer, which bridges an
7
oxygen surface anion and a gold atom, resembles the metal-
lacycle conformers already found as key intermediates for
[
13]
ethene epoxidation on silver catalysts. For the formation of
the metallacycle, reactive non-hydroxylated oxygen anions
around the gold cluster must exist. Therefore, the removal of
the hydroxy group by the interaction with adsorbing O or
2
H O becomes essential. The removal of the hydroxy groups is
2
also aided by the decrease in hydroxy group stability at high
coverage found above. We also analyzed the interaction
between gold-adsorbed propene and coadsorbed hydroxy
groups, which could give rise to the C H OH intermediates,
3
6
but found that there is instead a strong repulsion. In contrast,
Au/TiO is much more inert to the formation of metallacyclic
2
intermediates (Figure 2g), and although such metallacycles
are metastable, they are slightly unstable towards desorption
to gas-phase propene.
The third possibility is shown on Figure 2e, in which
propene binds to the substrate through two covalent CÀO
Figure 2. Side and top (smaller inset) views of the relaxed structure for
the non-hydroxylated a) Au /Al O and b) Au/TiO model catalyst.
7
2
3
2
bonds, resembling the bidentate propoxy species recently
Most relevant relaxed structures and binding energies (with respect to
gas-phase propene) for propene adsorption at c–e) Au /Al O and
found during propene epoxidation at Au/TiO by Nijhuis
2
7
2
3
[9]
et al. The stability of the propoxy species is as high as the
f–h) Au/TiO . Al green, Au yellow, H white, O red, C gray, Ti light gray.
2
gold-oxide bridging mode, thus demonstrating that in a small
cluster, such as Au , the CÀAu bonds can be as stable as the
7
CÀO bonds with the oxide. As for the Au/TiO system,
2
Table 1: Cumulative binding energies (in eV, with respect to water vapor)
for dissociative adsorption of up to n water molecules on Au /Al O ,
Figure 2h shows the most stable metallacycle conformer that
forms on the oxide surface, with propene bridging a titanium
cation and an oxygen anion (a configuration analogous to the
one in Figure 2e is around 0.5 eV less stable). Again, propene
adsorbs in a metastable fashion with endothermic binding,
confirming the much more inert character of the Au/TiO2
system relative to the Au /Al O system, which traps propene
7
2
3
where E
b
2
(n)=nE(H O,g)+E(Al
2
3
O )ÀE(nH
2 2 3
O/Al O ).
n
1
2
3
4
5
6
7
Eb(n)
2.75
3.57
4.65
6.90
7.56
8.47
8.99
7
2
3
more efficiently by easy formation of metallacycle propene
higher coverage, confirming the high reactivity of the surface
towards saturation of the dangling bonds. In contrast, the
oxide intermediates. At this point, it is interesting to mention
[
11]
[14]
that several very recent experiments
suggest that the
TiO surface is much less reactive; analogous tests provide
presence of very small Au clusters could account for the
2
n
dissociative binding energies of only 0.5–0.6 eV for the first
observed activity in nanoparticle-supported gold catalysts on
certain reactions. We have analyzed the stability of the
conformer in Figure 2g at a TiO -supported Au cluster and
pair of H O molecules. Therefore, a slab with 12 surface
2
hydroxy groups was chosen as the final model for Au /Al O ,
7
2
3
2
7
which corresponds to the situation in which the adsorption of
the hydroxy groups begins to saturate (Figure 2c–e). For Au/
find an only slightly larger binding energy (0.1–0.3 eV,
depending on the coverage by the hydroxy groups; see the
Supporting Information), which highlights the qualitative
difference between alumina and titania supports.
TiO , the clean system was chosen, as tests in the presence of a
2
single hydroxy group yielded similar propene/propene oxide
binding energies.
Structures and energetics for the final reaction step, that
is, the formation of propene oxide, are shown in Figure 3.
Owing to the differences for propene adsorption, different
pathways are explored for either Au/TiO or Au /Al O . For
The results for propene adsorption on both catalysts (Au7/
Al O and Au/TiO ) are shown in Figure 2c–e and Figure 2 f–
2
3
2
h, respectively. Three possibilities were found: 1) the adsorp-
tion of gas-phase propene onto gold by interaction of the
intact double C=C bond and a low-coordinate gold atom
2
7
2
3
Au/TiO , direct surface oxygen removal is not feasible and
2
calls for other sources of reactive oxygen; as extensive
simulations of the coadsorption and reaction of H /O at the
(
Figure 2c,f). The strong binding (0.81 and 1.09 eV) suggests
2
2
an easy adsorption, which is in agreement with experiments
by Ajo et al. (Au/TiO ) and by Nijhuis et al. (Au/SiO )
Au/TiO catalyst show that COOH radicals easily form after
2
reaction of O with surface hydroxy groups (formed after H
2
2
,
2
2
[
12]
[15]
showing that propene binds to gold by p bonding.
Fig-
dissociation at the gold nanoparticle),
which can be
ure 2d,g shows the diffusion to the gold/oxide interface,
which is probably the second step of the reaction and involves
the replacement of the double C=C bond by a single CÀC
bond to form the two new covalent CÀsurface bonds. In case
perceived as oxidizing agents. Figure 3a shows that after a
low 0.5 eVenergy barrier, the peroxo radical dissociates and a
strongly bound (1.59 eV) propene oxide metallacycle is
formed, which can be identified as the bidentate propoxy
Angew. Chem. Int. Ed. 2009, 48, 1467 –1471
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 1469

Communications
peroxo COOH radicals are needed for the formation of C H O
3
6
metallacycle intermediates, whereas for Al O -supported
2
3
gold clusters they can form directly after C H adsorption.
3
6
Such differences can be related to the replacement of
hydrogen by water in the Al O -supported catalysts; for Au/
2
3
TiO , we find the dissociation of hydrogen at the gold
2
nanoparticle to be an essential part of the process, both for the
formation of the surface hydroxy group and for promoting the
adsorption of molecular oxygen with subsequent formation of
reactive peroxo radicals. In the case of Au /Al O , oxygen
n
2
3
takes care of “healing” the vacancies (providing either
reactive oxygen adatoms, adsorbed hydroxys, or desorbing
water, depending on the neighboring hydroxy concentration).
Finally, in both cases the presence of neighboring hydroxy
groups is probably important for the process of detachment of
[9]
propene oxide metallacycles from the substrate. We expect
the calculated energy barriers to be lowered upon inclusion of
more coadsorbates, such as water, thus bringing the barriers
closer to the measured activation energies (see above). The
results shown in Figure 1a (propene/O2 atmosphere) can
probably be interpreted as arising from a situation in which
the hydroxy-terminated surface is transformed into an
oxygen-terminated situation, leading to an accumulation of
adsorbed metallacycles. These metallacycles are difficult to
desorb and block the active catalytic sites.
In summary, the results from the experimental and
theoretical study are unexpected and surprising in several
ways. First, we demonstrate that our model catalysts based on
subnanometer gold clusters are active for the partial oxida-
tion of propene (other similar model catalysts are active for
[
5]
CO oxidation). Our work thus complements existing reports
[
1,2]
on similar activity of larger gold nanoparticles.
our knowledge, this is the first catalyt for the C H /O system
Second, to
3
6
2
which uses alumina as support, instead of the commonly
applied titania that was considered essential for the produc-
tion of COH/COOH radicals, which are believed to be needed
for the promotion of the partial oxidation step on gold/titania.
The results from the simulations confirm the viability of this
system, assigning an even more active character to the Aun/
Figure 3. a) Structures and reaction energetics for the formation of
propene oxide from coadsorbed propene (Figure 2g) and peroxo
COOH radicals at Au/TiO catalysts. b) Reaction energetics for propene
2
oxide desorption from the anatase TiO (101) surface. c) Reaction
2
energetics for propene oxide formation at the Au /Al O model,
7
2
3
involving abstraction of substrate oxygen. Insets with rotated views are
included for transition and final states. The color code is the same as
in Figure 2.
Al O interface than to the Au/TiO one. Third, and more
2
3
2
important, for Al O -supported gold subnanometer clusters
2
3
the expensive and dangerous hydrogen can be replaced by
abundant and safe water vapor, which is also an efficient
method to maintain the hydroxy equilibrium at the surface. It
will be a challenging task to scale up the production of size-
selected clusters by other, more conventional methods, but
there are very encouraging experimental efforts demonstrat-
[
9]
species. The last reaction step involves desorption of these
species; Figure 3b shows that for TiO (the gold particle is
omitted now, as test calculations show that it does not modify
the C H O–TiO interaction), this desorption is the most
2
3
6
2
[16]
difficult stage of the reaction, taking a moderately high
barrier of 1.5 eV to close the metallacycle. The same holds for
Au /Al O (Figure 3c), with a 1.6 eV barrier for substrate
ing the feasibility of mass production of supported clusters.
7
2
3
oxygen abstraction. The process involves a gradual detach-
Experimental Section
ment of the oxygen atom by breaking each of the three OÀAl
The beam of gold clusters is produced in a laser ablation source and a
+
surface bonds: one when the metallacycle is formed, the
second in the transition state, and the third during the
formation of the second CÀO bond in C H O. As the process
distribution of {Au6–10
}
ions is selected by a mass-spectrometer-
[
17]
quadrupole deflector assembly for deposition. Flux and surface
coverage upon soft landing are monitored with a picoammeter. The
size of the supported clusters is verified by synchrotron grazing
3
6
leaves an oxygen vacancy behind, the catalytic cycle will
finally be closed by healing it with oxygen.
[7,17]
incidence small angle X-ray scattering (GISAXS)
before, during,
and after the reaction, confirming no aggregation of the subnano-
meter clusters (see Supporting Information). The studies are per-
formed in a flow reactor at 133 kPa pressure and 30 sccm gas flow.
Comparing the mechanisms for the two systems, some
interesting conclusions can be drawn. For Au/TiO , reactive
2
1
470
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 1467 –1471

Angewandte
Chemie
The reactants used are gases at 1% concentration in helium, their
ratio is kept 2:1 for C H /O and 2:1:1 for C H /O /H and C H /O /
Lee, M. J. Pellin, S. Seifert, G. Y. Tikhonov, N. A. Tomczyk, Top.
Catal. 2006, 39, 145 – 149.
3
6
2
3
6
2
2
3
6
2
H O. Products are detected on a mass spectrometer (Pfeiffer).
[8] O. P. H. Vaughan, G. Kyriakou, N. Macleod, M. Tikhov, R. M.
Lambert, J. Catal. 2005, 236, 401 – 404.
2
Turnover frequencies (TOF) are calculated using calibrated gas
mixtures and the count of deposited clusters. Activation energies are
estimated from the temperature dependence of the TOFs. DFT
[9] T. A. Nijhuis, T. Visser, B. M. Weckhuysen, Angew. Chem. 2005,
117, 1139 – 1142; Angew. Chem. Int. Ed. 2005, 44, 1115 – 1118.
[10] L. M. Molina, B. Hammer, Appl. Catal. A 2005, 291, 21 – 31.
[11] S. P. Adiga, P. Zapol, L. A. Curtiss, Phys. Rev. B 2006, 74, 064204;
S. P. Adiga, P. Zapol, L. A. Curtiss, J. Phys. Chem. C 2007, 111,
7422 – 7429.
[
18]
[19]
simulations used a plane-waves basis, ultrasoft pseudopotentials
scalar-relativistic for gold to take into account relativistic effects),
(
[
20]
and the Perdew-Wang-91 GGA functional for exchange-correla-
tion effects. Transition states and energy barriers are evaluated with a
[
21]
constrained minimization technique.
[12] H. M. Ajo, V. A. Bondzie, C. T. Campbell, Catal. Lett. 2002, 78,
359 – 368; T. A. Nijhuis, E. Sacaliuc, A. M. Beale, A. M. J.
van der Eerden, J. C. Schouten, B. M. Weckhuysen, J. Catal.
2008, 258, 256 – 264.
[13] M.-L. Bocquet, D. Loffreda, J. Am. Chem. Soc. 2005, 127,
Received: August 22, 2008
Revised: October 29, 2008
Published online: January 16, 2009
17207 – 17215; S. Linic, M. A. Barteau, J. Am. Chem. Soc. 2003,
1
25, 4034 – 4035; D. Torres, N. Lꢂpez, F. Illas, R. M. Lambert,
Keywords: cluster compounds · density functional calculations ·
epoxidation · gold · propene
.
Angew. Chem. 2007, 119, 2101 – 2104; Angew. Chem. Int. Ed.
007, 46, 2055 – 2058.
2
[
14] A. A. Herzing, C. J. Kiely, A. F. Carley, P. Landon, G. J.
Hutchings, Science 2008, 321, 1331 – 1335; M. Turner, V. B.
Golovko, O. P. H. Vaughan, P. Abdulkin, A. Berenguer-Murcia,
M. S. Tikhov, B. F. G. Johnson, R. M. Lambert, Nature 2008, 454,
981 – 983.
[
1] T. A. Nijhuis, M. Makkee, J. A. Mouljin, B. M. Weckhuysen, Ind.
Eng. Chem. Res. 2006, 45, 3447 – 3459.
[
[
2] T. Hayashi, K. Tanaka, M. Haruta, J. Catal. 1998, 178, 566 – 575.
3] P. Pyykkꢀ, Angew. Chem. 2004, 116, 4512 – 4557; Angew. Chem.
Int. Ed. 2004, 43, 4412 – 4456.
[15] L. M. Molina, M. J. Lꢂpez, J. A. Alonso, B. Hammer, unpub-
lished results.
[
4] M. M. Schubert, S. Hackenberg, A. C. van Veen, M. Muhler, V.
Plzak, R. J. Behm, J. Catal. 2001, 197, 113 – 122.
5] A. Sanchez, S. Abbet, U. Heiz, W. D. Schneider, H. Hꢁkkinen,
R. N. Barnett, U. Landman, J. Phys. Chem. A 1999, 103, 9573 –
[16] B. C. Gates, Chem. Rev. 1995, 95, 511 – 522; A. M. Argo, J. F.
Odzak, F. S. Lai, B. C. Gates, Nature 2002, 415, 623 – 626.
[17] S. Vajda, R. E. Winans, J. W. Elam, B. Lee, M. J. Pellin, S. Seifert,
G. Y. Tikhonov, N. A. Tomczyk, Top. Catal. 2006, 39, 161 – 166.
[18] M. C. Payne, M. P. Teter, D. C. Allan, T. A. Arias, J. D. Joanno-
poulos, Rev. Mod. Phys. 1992, 64, 1045 – 1097.
[19] D. Vanderbilt, Phys. Rev. B 1990, 41, 7892 – 7895.
[20] J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R.
Pederson, D. J. Singh, C. Fiolhais, Phys. Rev. B 1992, 46, 6671 –
6687.
[
9578; S. Lee, C. Y. Fan, T. P. Wu, S. L. Anderson, J. Am. Chem.
Soc. 2004, 126, 5682 – 5683; G. Pacchioni, S. Sicolo, C. Di Va-
lentin, M. Chiesa, E. Giamello, J. Am. Chem. Soc. 2008, 130,
8690 – 8695.
[
[
6] J. W. Elam, S. M. George, Chem. Mater. 2003, 15, 1020 – 1028.
7] R. E. Winans, S. Vajda, B. Lee, S. J. Riley, S. Seifert, G. Y.
Tikhonov, N. A. Tomczyk, J. Chem. Phys. B 2004, 108, 18105 –
[21] A. Alavi, P. Hu. T. Deutsch, P. L. Silvestrelli, J. Hutter, Phys. Rev.
Lett. 1998, 80, 3650 – 3653.
18107; R. E. Winans, S. Vajda, G. E. Ballentine, J. W. Elam, B.
Angew. Chem. Int. Ed. 2009, 48, 1467 –1471
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1471