Identification of the active species generated from supported Pd catalysts in heck reactions: An in situ quick scanning EXAFS investigation

Article abstract of DOI:10.1021/ja108636u

Quick scanning extended X-ray absorption fine structure (QEXAFS) studies in the subsecond time scale have been performed to gain insight into the reaction mechanism of Heck-type C-C coupling reactions in the presence of supported Pd-based catalysts. Using a specially designed in situ EXAFS cell, both the solid catalyst and the liquid reaction mixture during the reaction of phenyl bromide (PhBr) with styrene were monitored. Soluble Pd species were only, but rapidly, detected in the liquid reaction phase once the reaction temperature of 150 °C was reached. At the same time, the conversion of PhBr started, and during the following "active phase" of the catalyst hardly any changes in the corresponding EXAFS and XANES spectra were observed. The present species could be identified as colloidal Pd⁰ clusters with a size of ～2 nm estimated from the corresponding EXAFS spectra. The QEXAFS mode not only allowed monitoring rapid changes in the second time scale but also permitted minimization of effects caused by the heterogeneity of the systems. When the reaction rate started to decrease, pronounced changes in the EXAFS spectra were observed, which were attributed to an increased formation of bromo-palladates ([PdBr₄]^2-, [Pd₂Br₆]^2-). In addition to the liquid-phase species, significant changes were observed for the solid catalyst that was also probed in situ during the reaction. The originally oxidized Pd catalyst was efficiently reduced upon heating. Additionally, growth of the supported Pd particles was observed by both EXAFS and STEM. The above results confirm the role of the soluble molecular Pd species as the catalytically active species and clarify their conjunction with the in situ formed Pd colloids. Furthermore, the investigation demonstrates the potential of the QEXAFS not only for monitoring rapid changes during catalysis but also for gaining deeper insight into the mechanism of such complex industrially important systems under relevant reaction conditions.

Full text of DOI:10.1021/ja108636u

ARTICLE
pubs.acs.org/JACS
Identification of the Active Species Generated from Supported Pd
Catalysts in Heck Reactions: An in situ Quick Scanning EXAFS
Investigation
†
‡
‡
†,§
,§,^
Sven Reimann, Jan St €o tzel, Ronald Frahm, Wolfgang Kleist, Jan-Dierk Grunwaldt,* and
,
†
Alfons Baiker*
†
Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, H €o nggerberg, HCI, CH-
093 Zurich, Switzerland
8
‡
Department C - Physics, University of Wuppertal, Gausstrasse 20, D-42097 Wuppertal, Germany
§
Institute for Catalysis Research and Technology (IKFT), Karlsruhe Institute of Technology (KIT), Herrmann-von-Helmholtz-Platz 1,
D-76344 Eggenstein-Leopoldshafen, Germany
^
Institute for Chemical Technology and Polymer Chemistry (ITCP), Karlsruhe Institute of Technology (KIT), Kaiserstr. 12, D-76128
Karlsruhe, Germany
S Supporting Information
b
ABSTRACT: Quick scanning extended X-ray absorption fine structure (QEXAFS)
studies in the subsecond time scale have been performed to gain insight into the reaction
mechanism of Heck-type C-C coupling reactions in the presence of supported Pd-based
catalysts. Using a specially designed in situ EXAFS cell, both the solid catalyst and the
liquid reaction mixture during the reaction of phenyl bromide (PhBr) with styrene were
monitored. Soluble Pd species were only, but rapidly, detected in the liquid reaction phase
once the reaction temperature of 150 ꢀC was reached. At the same time, the conversion of
PhBr started, and during the following “active phase” of the catalyst hardly any changes in the corresponding EXAFS and XANES
0
spectra were observed. The present species could be identified as colloidal Pd clusters with a size of ∼2 nm estimated from the
corresponding EXAFS spectra. The QEXAFS mode not only allowed monitoring rapid changes in the second time scale but also
permitted minimization of effects caused by the heterogeneity of the systems. When the reaction rate started to decrease,
pronounced changes in the EXAFS spectra were observed, which were attributed to an increased formation of bromo-palladates
2
-
2-
(
[PdBr4] , [Pd Br ] ). In addition to the liquid-phase species, significant changes were observed for the solid catalyst that was
2 6
also probed in situ during the reaction. The originally oxidized Pd catalyst was efficiently reduced upon heating. Additionally, growth
of the supported Pd particles was observed by both EXAFS and STEM. The above results confirm the role of the soluble
molecular Pd species as the catalytically active species and clarify their conjunction with the in situ formed Pd colloids. Furthermore,
the investigation demonstrates the potential of the QEXAFS not only for monitoring rapid changes during catalysis but also
for gaining deeper insight into the mechanism of such complex industrially important systems under relevant reaction
conditions.
’
INTRODUCTION
selectivity compared to their homogeneous counterparts, they
may only serve as precursors for an in situ formed homogeneous
catalyst.
In particular, the reaction mechanism of the heterogeneously
catalyzed, ligandless Heck coupling reaction is still a matter of
discussion and speculation. One of the most popular mechan-
isms describes a reaction pathway where the palladium is leached
into the liquid phase from a heterogeneous precursor or from Pd
C-C coupling reactions are a versatile tool in synthetic
1,2
chemistry and are of practical interest in industry. One of
the most thoroughly investigated is the Heck coupling reaction
3
-5
(Scheme 1).
It is usually catalyzed by homogeneous Pd
complexes, often with phosphine ligands, in the presence of a
base. The advantage of this reaction is its high tolerance to a wide
range of functional groups and the good availability of aryl
colloids that were intermediately formed from organometallic
5
bromides and chlorides. However, from a practical point of
6-8
precursors.
Consequently, the active species is suggested to
view, heterogeneous catalysts are often preferred to homoge-
neous ones because of their easier handling. Nonetheless, one
may be confronted with other challenges in the case of hetero-
geneous catalysts. Besides the generally lower activity and
be a homogeneous complex formed in situ from the leached Pd.
Received: September 24, 2010
Published: February 23, 2011
r 2011 American Chemical Society
3921
dx.doi.org/10.1021/ja108636u
|J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
Scheme 1. Heck Coupling of PhBr (1) with Styrene (2), Leading to the Formation of trans-Stilbene (3), cis-Stilbene (4),
Diphenylethylene (5), and Diphenyl (6)
In contrast, mechanisms assuming a purely heterogeneously
Catalytic Experiments. All the catalytic experiments were carried
out in a specially designed XAS batch cell. The cell possesses two beam
paths, which allows probing the liquid phase and the catalyst bed with
the solid catalyst separately. The details of this cell are described
9,10
catalyzed Heck coupling have been suggested as well.
De-
pending on the reaction conditions, the formation of Pd colloids
and/or the redeposition of the Pd on a support can be observed,
and a lot of effort has been put into the control of this behavior in
the past. However, it is difficult to detect intermediate palladium
species and their dynamic behavior during the reaction based on
ex situ methods such as transmission electron microscopy
2
5
elsewhere. For a typical experiment, the 5 wt % Pd/Al
(
(
2
O
3
catalyst
168 mg) was placed in the catalyst bed of the in situ XAS cell, and NMP
1 mL) was carefully added. Then styrene (3.75 mmol, 0.390 g, 0.431
mL) and PhBr (2.5 mmol, 0.393 g, 0.26 mL) followed by a dispersion of
NaOAc 3H O (1.1 mmol, 150 mg) in NMP (2 mL) were added to the
3
2
(
TEM), ESI-MS, ICP-MS, or ICP-OES, which can only give a
11,12
XAS cell. After setting the desired pressure, the heating was started. The
conversion and the yield of the products were followed by GC using a
HP 6890 series gas chromatograph equipped with a HP 5 column (30 m
ꢀ 0.32 mm, i.d. 0.25 μm). Decane was used as internal standard for these
measurements. The products were identified by comparison with
authentic samples and GC-MS using a HP 6890 series gas chromato-
graph equipped with a HP 5-MS column (25 m ꢀ 0.20 mm, i.d. 0.33
μm) coupled to a HP 5973 mass spectrometer.
snapshot of the reaction mixture.
In addition, it has been
shown that very small concentrations of the catalytically active
species in solution are sufficient to catalyze the reaction but might
13
be difficult to detect using these techniques. In contrast,
spectroscopic in situ investigations allow following the operation
and changes of a catalyst in real time and thus provide crucial
14
information about the structural changes of the catalyst.
Nevertheless, only a few studies dealing with such in situ
QEXAFS Measurements. The QEXAFS measurements were
performed at the SuperXAS beamline at the Swiss Light Source (SLS,
Villigen, Switzerland). The ring was operated in top-up mode with 400
mA of 2.4 GeV electrons, which emitted polychromatic radiation in a
superbend magnet (2.9 T) with a critical energy of 11.9 keV. The beam
was collimated by a Rh-coated mirror and then monochromatized using
investigations for C-C coupling reactions have been published
1
5-17
to date.
EXAFS is a promising in situ technique in this case because it
allows selectively monitoring the Pd-containing species and
14,15,18,19
additionally resolving their specific structure.
It is
further possible to study very small particles and species with
short-range order in solution, which are decisive advantages
compared to other structure-resolving techniques like e.g. XRD
and NMR. Since changes in such reactions may happen in a quite
short time frame, the Quick EXAFS (QEXAFS) technique was
27
a channel-cut Si(311) crystal in the QEXAFS monochromator. The
crystal was continuously moved in a sinusoidal way with a peak-to-peak
amplitude of 0.60ꢀ corresponding to an energy range of about 1.7 keV at
the Pd K-edge at 24.35 keV (see ref 28, Figure 2, right side, fourth black
line from the bottom). A rotation speed of 1 Hz was chosen, yielding one
EXAFS spectrum each 500 ms. Another Rh-coated mirror behind the
QEXAFS monochromator was used to focus the beam on the sample.
The samples were measured in transmission mode with a 15-cm
ionization chamber filled with Ar in front of the sample and another
15-cm ionization chamber filled with Kr behind the sample. A third 30-
cm ionization chamber filled with Ar was used to measure a Pd metal foil
located between the second and a third chamber to obtain an absolute
calibration value for the energy scale of each spectrum. This value was
related to the relative position of the monochromator crystal which was
measured simultaneously to each data point with a fast optical angular
encoder system, yielding an accurate energy scale for the time-resolved
20-24
used to monitor the reaction.
Here, a time-resolved in-
vestigation on the subsecond scale was achieved, while main-
taining all the mentioned advantages of the EXAFS technique. In
addition, new tools for QEXAFS data analysis are presented and
applied to the data to meet the challenges of this experiment,
namely the low concentrations of Pd species and the complex
mixtures of them. With a combination of EXAFS and XANES
analysis techniques, the maximum of information was extracted
from the data. To investigate the leaching behavior, a special cell
with two beam paths for the X-ray beam was used, which allows
25
probing the catalyst bed and the liquid phase separately.
2
9
spectra. The ionization chambers were operated with a high voltage of
.5 kV, and the signals were amplified with Keithley 428 current
amplifiers and then measured with a Keithley KUSB3116 data acquisi-
1
’
EXPERIMENTAL SECTION
3
0
tion board using a fast data acquisition software. The acquired data
were processed mostly with homemade software, which is partly based
Materials. Phenyl bromide (1, PhBr, Fluka, >99.5%), styrene (2,
Acros, 99.5%), trans-stilbene (3, Acros, 99%), cis-stilbene (4, Fluka,
g95%), 1,1-diphenylethylene (5, Acros, 99%), diphenyl (6, Fluka, >
3
1
on the IFEFFIT package. All EXAFS fits were performed using
3
2
9
8%), decane (Acros, > 99%), N-methyl-2-pyrrolidone (NMP, Merck,
9.5%), and sodium acetate trihydrate (NaOAc 3H O, Merck, >99.5%)
(E40692)
catalyst was purchased from Engelhard. The BET surface of the catalyst
IFEFFIT, and the according path generation used FEFF6. The spectra
from the bottom window of the cell were treated with IFEFFIT
algorithms concerning pre-edge and postedge fitting, normalizing,
background fitting with AUTOBK and Fourier transformation (FT).
9
2
3
2 3
were used as received. The commercial 5 wt % Pd/Al O
2
-1
-1
3
is 130 m g according to the data sheet provided by the manufacturer.
The mean Pd particle size and the corresponding metal dispersion where
determined to be 3.3 nm and 0.36, respectively, as described else-
For the FT with E = 24355 eV, a k-range of 0-14 Å , a k -weighting,
0
1
-
and a Kaiser-Bessel window (dk = 0.2 Å ) were chosen. The EXAFS
fits of these spectra were all performed within an R-range of 1-3 Å,
except for the third Pd shell, which was fitted within 4.2-4.8 Å. The
26
where.
3
922
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
Figure 1. Schematic design of the reactor used for the QEXAFS in situ
study. The two beam paths allow separate monitoring of the supported
catalyst in the catalyst bed (lower window) and soluble species in the
liquid phase (upper window).
33,34
principal component analysis (PCA)
as well as the data treatment
for the spectra with very small edge jumps measured in the upper
35
window of the reaction cell were carried out with WinXAS. Here, the
AUTOBK algorithm did not yield reasonable results so that cubic spline
functions were applied for background fitting. The resulting χ(k) spectra
were Fourier transformed as described above. However, because of the
lower Pd concentration, a k-range of 2-10 Å- was chosen. EXAFS data
1
fitting was again performed within an R-range of 1-3 Å. A Pd metal foil
2
was fitted with each of the described data treatments yielding S
0
values,
which were used as reference to determine the coordination numbers.
The linear combination fits (LCF) were also carried out with homemade
scripts using the IFEFFIT code. For the fits a range of 24335-24385 eV
was chosen to cover the XANES region and thus to get results
complementary to the EXAFS fits.
Electron Microscopy Measurements. STEM and EDXS in-
vestigations were performed on a field emission transmission electron
s
microscope Tecnai 30F (FEI; SuperTwin lens with C = 1.2 mm),
operated at 300 kV. For the STEM measurements, the spent Pd/Al
2 3
O
catalyst was washed with water and dichloromethane to remove
inorganic and organic compounds, especially bromides, which could
disturb the STEM measurements substantially. For the STEM investiga-
tions of the liquid reaction phase, some droplets of the reaction mixture
were deposited on a holey carbon foil supported on a copper grid.
Figure 2. Development of the Pd K-edge absorption in the liquid phase,
depending on the temperature and time (a) (the inset shows the results
without applying the new median/averaging filter), and the correspond-
ing time-dependent EXAFS spectra of the liquid phase after the
stabilization of the edge jump without (b) and with a moving median
filter (c).
’
RESULTS
First, the in situ cell was optimized with respect to the catalytic
and spectroscopic conditions in order to get spectra with a
sufficient signal-to-noise ratio. The principle of the cell is shown
in Figure 1. The lower beam allows monitoring of the supported
catalyst, whereas the upper one with a longer path length permits
monitoring the liquid phase. In order to exclude reactions
catalyzed by the in situ cell made from stainless steel, a PEEK
inset was used. In a preliminary test, the reactor was filled with
phenyl bromide (PhBr, 1), styrene (2), and 7 mL of N-methyl-2-
pyrrolidone (NMP) under Ar. The corresponding measurement
through the liquid phase revealed the presence of Pd in the liquid
phase. However, the edge jump was rather small, and the quality
of the corresponding X-ray absorption spectra was not sufficient
for a detailed EXAFS analysis. Reduction of the solvent volume
from 7 to 3 mL led to better results. To further increase the
amount of leached Pd, Ar was replaced by air in accordance to
edge jump with those of reference samples with well-defined
concentrations.
At the end of the reaction, the maximum Pd concentration in
the liquid phase was determined to be 2.8 mM which equates to a
Pdliquid phase/substrate ratio of 0.3 mol %. This corresponds to
14% of the total amount of Pd in the system. Compared to other
catalysts, the Pd leaching in our system is within the typical
11,37,38
range.
Conversion and product yields that could be
obtained in the in situ experiments were also in the same range
as those reported in the literature under comparable reaction
conditions (vide infra). For all further experiments, we applied
these optimized reaction conditions. In a first step, the leaching of
Pd into the liquid phase was investigated by following the
development of the Pd K-edge in the liquid phase (upper
window). The results are shown in Figure 2. At RT, no Pd was
observed in the liquid phase. Also during the heating phase from
3
6
earlier reports, that led to a higher Pd leaching and subsequent
increase in the quality of the EXAFS spectra. The Pd concentra-
tion in the liquid phase was determined by comparison of the
3
923
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
RT to 150 ꢀC, no edge jump was visible, indicating that no Pd had
yet been leached from the Pd/Al O catalyst into the liquid
2
3
phase. The absence of any edge jump in this stage of the
experiments demonstrates that the heterogeneous catalyst re-
mains in the catalyst bed and is not transported into the upper
part of the reactor where the liquid phase is probed. We can
therefore conclude that all the Pd detected in the liquid phase is
actually leached Pd. At a reaction temperature of 150 ꢀC, the
sudden appearance of an edge jump at the Pd K-edge energy in
the liquid phase indicates the formation of leached Pd species
from the Pd/Al O catalyst: after 22 min of total reaction time,
2
3
the absorption increases to a significant level within one minute.
Thereafter, only a weak ascent of the edge jump is observable up
to 60 min reaction time before it maintains a constant level
during the remaining reaction time, indicating a constant Pd
concentration in the liquid phase (Figure 2a). The corresponding
normalized EXAFS spectra are shown in b and c of Figure 2.
In order to get a reasonable quality of the spectra, 400
individual spectra were averaged, which corresponds to a time
resolution of 3.3 min for the shown EXAFS spectra, while 100
spectra (time resolution: 50 s) were averaged to determine the
edge jump. The EXAFS spectra acquired during the initial
increase of the Pd concentration in the liquid phase were rather
distorted because of possible superheating of the liquid phase.
After another ∼20 min, the whole system was equilibrated, and
the spectra became clearer (Figure 2b). Since the distortions are
randomly distributed over the spectra, a new software tool for
QEXAFS data analysis with an implemented median filter was
designed to exploit the high time resolution here. The software
tool compares every absorption value in each spectrum with the
equivalent absorption values of the (in this case) next two
neighboring spectra. If the value is the smallest or highest of
the resulting three values, it is sorted out as runaway value and
replaced by the mean value, which is then considered for further
averaging. Thus, runaway values are effectively suppressed as
shown in Figure 2c where the distorted spectra are significantly
improved so that now all spectra during the active phase of the
catalyst can be considered for further data analysis. In addition,
the edge jump shown in Figure 2a was determined with filtered
spectra, while the results deduced from the unfiltered data are
shown in the inset for comparison. This simple example clearly
demonstrates the capabilities of this filter to suppress nonstatis-
tical noise, and to our knowledge, it is a novel approach to exploit
the high time resolution of QEXAFS data in catalysis and
chemical applications in general. One can easily imagine that it
is not only effective for heterogeneous solids or inhomogeneous
liquid solutions where such time-dependent inhomogeneities
occur, but also cancels out randomly appearing electrical dis-
turbances that might show up in the spectra. Prerequisites are a
high time resolution compared to the investigated processes and
a reproducible and exact data acquisition system.
Figure 3. (a) Yield of the main product trans-stilbene (3, 9) and the
observed side products cis-stilbene (4, b), 1,1-diphenylethylene (5, Δ),
and diphenyl (6, 0) over time as measured by GC. (b) Conversion of
PhBr (1, b) over time depending on the reaction temperature (9).
in the Pd concentration in the liquid phase. Therefore, it is most
likely that rather structural changes of the leached Pd took place.
To ensure that the actually measured Pd in the liquid phase is not
just Pd deposited on the reactor wall, the empty PEEK inset was
measured after the reaction. No edge jump was observed in this
case, demonstrating that no significant amount of Pd was
deposited on the inset wall during the reaction and the measured
Pd is dissolved in the liquid phase.
The corresponding yields of the formed products as deter-
mined by gas chromatography (GC) are shown in Figure 3a.
The main product is trans-stilbene (3), the formation of which is
accompanied by the usual side products, cis-stilbene (4), 1,1-
diphenylethylene (5), and diphenyl (6) (Scheme 1). The
formation of benzene was not observed under the applied
reaction conditions. The maximum conversion was determined
by the amount of the added base. After 90 min, a conversion of
85% relative to the base was achieved. The overall chemoselec-
tivity for the main product 3 was determined to be 89% at the end
of the reaction. For comparison, the conversion of PhBr (1) was
measured, too (Figure 3b). It followed the formation of the
products and reached a value of 35% at the end of the reaction.
If one compares the behavior of the edge jump and the EXAFS
spectra over time with the catalytic activity, one can observe a
good agreement between the conversion of PhBr, the formation
of the products, and the structural changes in the EXAFS spectra.
Emerging from Figure 3, the conversion of PhBr starts after 22
min. This is exactly the moment when leaching of Pd is indicated
The first spectrum shown in Figure 2c is measured directly
after the huge change of the edge jump described above, after a
reaction time around 25 min. It shows clear EXAFS features, and
the XANES region, more specifically the exact energy of the Pd
K-edge, indicates that the Pd is in a reduced metallic state. During
the next 50 min, the spectra look rather similar, indicating that no
substantial structural changes occur. Thereafter, the EXAFS
spectra show a unique behavior. After ∼75 min reaction time,
the XANES features disappear almost immediately, whereas the
EXAFS region still remains structured. However, the absorption
at the edge does not change, indicating that there was no change
3
924
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
Table 1. Coordination Numbers (CN) and Interatomic Dis-
tances (R) Derived from the Analysis of the FT Spectra of the
Liquid Phase
2
-3
2
reaction stage
shell
CN
σ /10
Å
R/Å
active phase:
Pd
Br
8.7 ( 1.1
2.7 ( 0.7
2.2 ( 1.2
10.0 ( 0.9
2.6 ( 1.8
5.4 ( 3.6
2.733 ( 0.005
2.423 ( 0.019
2.792 ( 0.028
inactive phase:
Pd
signals with a constant intensity at 1.8 and 2.5 Å during the active
phase up to about 75 min and one broad signal at 2.0 Å when the
system becomes inactive. The double feature during the active
phase could unambiguously be attributed to a single Pd neighbor.
This was rationalized by the facts that: (i) the amplitude ratio of
the two signals does not change with different k-weights as would
be expected for contributions of lighter backscattering atoms, (ii)
the signals can be reproduced by an EXAFS refinement using the
first Pd shell of bulk Pd, and (iii) the signals appear as well in the
EXAFS spectra of pure Pd metal foils if the k-range used for the
Fourier transformation is small, namely below 10-12 Å- for
1
Figure 4. Time-dependent Fourier transformed EXAFS spectra of the
liquid phase.
the upper k-limit depending on the chosen E . Similar peak
0
doublets for a single coordination caused by the resonating
(
peaked) behavior of the photoelectron backscattering ampli-
39,40
tude were also observed, for example in the case of Ag.
Thus,
the Fourier transformed spectra of the active phase were fitted
with a simple model of bulk Pd. For this purpose, 800 individual
spectra were averaged to increase the data quality and thus to get
as good fit results as possible. Since the observed active and
inactive states of the catalyst are stable over a long time interval
(
Figure 2c), the application of such an extensive averaging is
justified here. Consequently, the obtained fit results display a
reasonable agreement with the experimental data (Figure 5a).
The corresponding values for the coordination numbers and
interatomic distances are shown in Table 1.
On the basis of these fit results, one can conclude that most of
the leached Pd is present in the form of small Pd colloids with
average sizes of about 2 nm during the active phase. This can be
deduced from the coordination number of the first shell using the
4
1
diagrams for fcc lattice structures calculated by Jentys and
STEM investigations of the liquid phase, where Pd particles were
detected as well (Figure S1, Supporting Information). After the
reaction rate decreased significantly, the FT EXAFS spectra of
the liquid phase changed in favor of the broad backscattering
signal at 2.0 Å (Figure 4). Attempts to fit this peak with a single
Pd, O, or Br shell utterly failed. However, the structure can be
well reproduced with a combination of a Br and a Pd shell as
shown in Figure 5b. For this fit, the first Pd-Pd scattering path of
bulk Pd was used in combination with the first Pd-Br shell of
4
2
tetrabromo-palladate as determined by Makitova et al. The fit
results lead to nearly the same Pd-Pd distance as for the small Pd
colloids but with a substantially lower coordination number of
about 2.2 (Table 1). In relation to pure Pd, this would corre-
spond to about 20% bulk Pd or in relation to the colloids in the
active phase to about 30% Pd particles. The Pd-Br distance
Figure 5. EXAFS refinement fits for (a) active and (b) inactive phase of
the catalytic process measured in the liquid phase in R- and k-space.
by the increasing edge jump (Figure 2a). The conversion then
increases quickly at a constant rate until 45 min, and after 75 min,
almost no further formation of the reaction products is observed.
Interestingly, the decrease in the reaction rate is accompanied by
distinct changes in the EXAFS spectra.
2
-
agrees well with the expected value for [PdBr ] and even better
4
2
-
with the first Pd-Br path of [Pd Br ] , which might also be
2
6
involved here. The coordination number of about 2.7 corre-
2
-
2-
sponds roughly to about 70% of [PdBr4] /[Pd Br ] species.
2
6
In order to get more structural information about the Pd
species in the liquid phase and their structural changes, the
EXAFS spectra of the liquid phase were Fourier transformed
Thus, at this stage of the reaction, a mixture of about 20-30%
Pd colloids and accordingly 80-70% Pd-Br complexes
2
-
2-
([PdBr ] , [Pd Br ] ) are consistent with the measured
4
2
6
(Figure 4). As one can easily see, there are two well-separated
EXAFS signal. This superposition of Pd-containing species
3
925
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
Figure 8. Results of linear combination fits using Pd K-edge spectra of
pure Pd metal and PdO and the related fit errors of all spectra at the
catalyst bed (bottom part of the in situ cell) as a function of time.
Figure 6. Linear combination fit of spectra of the liquid phase at the end
of the reaction using Pd K-edge spectra of a pure Pd metal foil and solid
2
crystalline PdBr as references.
2
also interesting to see that the mean-square-displacement σ of
the Pd-Pd backscattering path is lower during the inactive
phase. Although the uncertainties for this value are comparably
high, this might be a hint that the Pd colloids might be
substantially larger at this point of the reaction compared to
the beginning of the leaching process. This is in accordance with
the observed Pd-particle growth of the solid catalyst (vide infra).
These results deduced from the above analysis of the Fourier
transformed EXAFS spectra could be confirmed by an alternative
approach. For this purpose, the corresponding normalized EX-
AFS spectrum of the solution at the end of the reaction was fitted
by a linear combination using the spectrum of pure Pd metal and
the spectrum of solid crystalline PdBr . It yielded 25% Pd metal
2
content versus 75% PdBr , which is in the same order of
2
magnitude as determined above (Figure 6). The structure of
the XANES region is qualitatively reproduced well. The devia-
tions may be attributed to the structural differences between the
crystalline PdBr and the presumably dissolved bromo-palla-
2
dates. However, it is obvious that the measured XANES spectra
can be reproduced only with mixtures of Pd metal/colloids and
bromo-palladates.
The special design of the cell made it also possible to monitor
the changes of the solid 5 wt % Pd/Al O catalyst during the
2
3
reaction by probing the lower window of the reactor. The
corresponding EXAFS spectra are shown in Figure 7a. Because
of the high Pd content in the catalyst bed, time resolutions up to
5
00 ms for each spectrum are feasible with the acquired data.
However, since no changes in the structure of the corresponding
EXAFS spectra were detected within less than one minute, the
data quality for an optimized EXAFS analysis could be improved
by averaging 100 spectra (time resolution: 50 s) without losing
any information about the comparable slow dynamics of the
system.
At the beginning of the reaction, the catalyst is in a partly
oxidized state as indicated by the strong whiteline at about 24350
eV. A linear combination fit using the spectra of metallic Pd and
PdO as references in the XANES region of 24335-24385 eV
revealed an initial fraction of about 70% PdO and only about 30%
metallic Pd (Figure 8). This is reasonable, as the catalyst was
2 3
Figure 7. (a) Normalized EXAFS spectra of the 5 wt % Pd/Al O
catalyst as a function of time. (b) Fourier transformed EXAFS spectra of
1
6
the 5 wt % Pd/Al O catalyst as a function of time.
stored under air and used without any pretreatment. During
2
3
the heating phase from RT to 150 ꢀC, the 5 wt % Pd/Al O
2
3
may also explain the vanishing XANES structure (in particular
the two features between 24360 and 24390 eV), which would not
be explainable with the presence of only one single species. It is
catalyst was reduced as indicated by the continuously decreasing
intensity of the whiteline in the EXAFS spectra. The reduction of
the catalyst was also evidenced by linear combination fits, which
3
926
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
Figure 10. Results of the PCA with normalized EXAFS spectra of the
catalyst bed (bottom part) for different time regions and the entire reaction.
linear combination fit results since 70% PdO and 30% Pd
correspond to a Pd/O ratio of 1.4.
2
The mean-square-displacement σ of the Pd-Pd scattering
path increases linearly as it could be expected because of the
applied temperature ramp, while the one for Pd-O is difficult to
investigate properly as the decreasing amount of oxygen leads to
increasing fit errors. The distance between Pd and O coincides
Figure 9. EXAFS refinement fit results for all spectra at the catalyst bed
(bottom part) as a function of time. The first region was fitted with a Pd
and an O shell, while only the first Pd shell of Pd metal was considered
for the second region (displayed in dark red). In the inset, structural
models for small Pd clusters are displayed (see text for more details).
4
3,44
perfectly with the one found in the literature for PdO.
The
Pd-Pd distance is very close to the literature value of 2.75 Å for
45
pure Pd metal and slightly below the one known for PdO.
These deviations are within the expected limits and can be
attributed to the small size of the investigated particles. To make
sure that there is no unexpected or unexplained additional
intermediate state during the first 35 min of the reaction, we
applied principal component analysis (PCA) to the normalized
EXAFS spectra of this time section as shown in Figure 10. The
results suggest only two components, which coincides with the
model of a straight reduction process from PdO to metallic Pd.
After this initial reduction of the catalyst, only the first shell of
bulk Pd was used as a fit model for the remaining spectra since
oxygen content could not be detected any longer. The results are
shown in Figure 9 (dark red quadrangles). They are consistent
with the fit results for the first 25 min as it can be seen from the 5
min overlap region of the two fits (Figure 9). Apart from the fit
model, the same data pretreatment and fit parameters were
chosen as for the first 25 min in order to maintain consistency
of the results. During the observed plateau after 25 min, the
average number of atoms in the Pd particles was derived from the
EXAFS first shell data presented in Figure 9 and amounts to
about 150 atoms using the theoretically calculated diagrams
revealed a decrease of the PdO amount to about 5% after 25 min
and to 0% after 50 min. However, the results of the linear
combination fit have to be interpreted carefully because of the
increasing fit error after 15 min reaction time (Figure 8). On the
basis of the EXAFS results (vide infra), the higher fit error can be
assigned to the small size of the Pd particles (∼2 nm), leading to
less enhanced EXAFS features which are reproduced better by a
linear combination fit with a low amount of PdO. Thus, one
could expect the catalyst to be already completely reduced after
2
5 min. The analysis of the Fourier transformed EXAFS spectra
of the solid catalyst at the beginning of the reaction reveals two
signals at distances of 1.55 Å and 2.45 Å as shown in Figure 7b.
The signal at 1.55 Å can be assigned to an oxygen neighbor, and
the one at 2.45 Å, to neighboring Pd atoms. The signal at 1.55 Å
decreases in intensity during the heating phase and vanishes after
2
5 min, which coincides with reaching the reaction temperature
of 150 ꢀC. The signal at 2.45 Å gains in intensity during the
heating period and then reaches a plateau for another 18 min
before it increases again after 43 (= 25 þ 18) min. The decrease
of the signal at 1.55 Å and simultaneous increase of the signal at
41,46,47
2
.45 Å are in good agreement with the reduction of PdO to Pd as
provided by Jentys.
Interestingly, this value is close to
it was already suggested on the basis of the results from linear
combination fit and the decreasing whiteline in the absorption
spectra. For a more quantitative analysis we fitted the spectra
during the first 25 min with a refinement of the first two
calculated scattering paths of PdO using the lattice structure as
determined by Waser. It is evident that the behavior of the
determined coordination numbers of Pd and O in this period is
consistent with the assumed reduction process of Pd (Figure 9).
Whereas the coordination number for Pd increases over time
from 3 to 8, that of oxygen decreases in the same time interval
from 2 to zero. This yields a Pd/O ratio for the initial state of 1.5
which agrees well—within the typical EXAFS errors—with the
the magic number of 147 atoms, which would be the exact
number for a cuboctahedron or a regular icosahedron with four
4
8
atoms in each edge as has been shown by Mackay. Both
geometries are visualized in Figure 9. Fitting the coordination
number of the third Pd shell in combination with the first shell is
an appropriate way to investigate the particle shape as was also
43
41
explained by Jentys. Applying this for the spectra acquired
during the plateau, a value of ∼13.6 is obtained for the
coordination number of the third shell (Figure S3, Supporting
Information) leading to a N /N ratio of about 1.6 ( 0.3. This
3
1
further supports a spherical shape of the Pd particles consisting of
about 150 atoms and supports the conclusion derived from the
3
927
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
analysis of N for the presence of cuboctahedral or regular
Scheme 2. Schematic Representation of the Catalytic Cycle
of the Heck Reaction with Two Different Reaction Pathways,
A and B
1
icosahedral particles. Thus, we suggest that these kinds of
particles with a size of ∼2 nm in diameter are the predominant
Pd-containing species at this stage of the reaction. This is also in
perfect agreement with the scanning transmission electron
microscopic (STEM) picture of the solid catalyst ex situ reduced
in flowing hydrogen where particles between 2-3 nm were
2
6
observed.
After this plateau, or after 43 min, respectively, the signal at
.45 Å starts to increase again. Since the Fourier transformed
2
spectrum does not indicate the formation of a new species, this
increase was attributed to particle growth. Extensive fitting of the
EXAFS data indeed revealed a particle growth to average sizes far
above 5 nm, which constitutes the upper limit for accurate
particle-size investigations with EXAFS and is clearly exceeded
here. On the other hand, the determined coordination numbers
are still below the value of 12 for bulk Pd, showing that metallic
Pd is still partly present in small particles at the end of the
2
reaction. The mean-square-displacement σ of the Pd-Pd path
decreases because of the higher lattice order that goes along with
the particle growth. Only slight changes in the Pd-Pd distance
are observable, as expected, since there is no reason for a
significant change here. STEM pictures taken of the catalyst at
the end of the reaction confirm the suggested growth of the Pd
particles (Figure S2, Supporting Information). However, the
growth seems to be quite inhomogeneous. Besides particles with
a diameter of about 5-8 nm, also bigger particles with diameters
of more than 20 nm can be found, which is consistent with the
EXAFS results. Again, to exclude the possibility that other
processes than particle growth caused the changes in the EXAFS
spectra, a PCA was applied to the normalized spectra of the time
range from 35 min to the end of the reaction (Figure 10). Since
only two components were detected, the second one being rather
small, one can assume that the particle growth is indeed the main
process occurring in this stage of the reaction.
PhBr in NMP, which showed the same leaching behavior for Pd
and the original reaction mixture (Figure S4, Supporting In-
formation). One could expect NMP to promote the leaching by
II
stabilizing the leached Pd species. The catalytic cycle then
continues according to the well-known homogeneous mechan-
II
ism via the coordination of an olefin to the Pd center, followed
by olefin insertion and β-hydride-abstraction (Scheme 2, cycle
5,7
A). Since at the end of the catalytic cycle the Pd is coordina-
tively undersaturated, it is prone to aggregation and consecutive
formation of Pd clusters as well as, depending on the reaction
7,12
conditions, Pd black.
It has been speculated that the in situ
formation of colloids depends on the activity toward the oxida-
tive addition of the aryl halide applied in the reaction. If the latter
is high enough, the formation of colloids is prevented as it is the
case for PhI, the most active aryl halide in the Heck reaction
(
Scheme 2, cycle A). However, when the less active PhBr is used,
’
DISCUSSION
it was speculated that the slower oxidative addition of PhBr to the
underligated Pd species favors the formation of Pd colloids.
0
The lively discussion about the active palladium species in
7
,8,38
Therefore one can also propose an alternative cycle where the
present Pd colloids do not only serve as Pd reservoir but are
Heck reactions
evidence the actual difficulty of its determi-
nation and identification in Heck and related C-C coupling
reactions due to the inhomogeneous reaction mixture and the
low concentration of Pd species in solution. The present study
shows that X-ray absorption spectroscopy in terms of XANES
and EXAFS combined with a specially designed in situ cell is a
powerful tool for in situ investigation in complex matrices like
industrially relevant reaction mixtures. The application of QEX-
AFS with fast single measurements for catching short intermedi-
ates and eliminating time-dependent inhomogeneities during the
course of the reaction provides additional insight into the
reaction mechanism.
5
2,53
directly involved in the catalytic cycle (Scheme 2, cycle B).
In
this case, the formation of the active catalyst is part of the catalytic
cycle, since the active catalyst is reduced to colloids at the end of
each catalytic cycle and has to be reformed at the beginning of a
new catalytic cycle by the interaction of PhBr with the surface of
the Pd colloids.
Our results show that during the active phase of the catalysis,
mainly Pd colloids are observed, and therefore support that the
second cycle B may significantly contribute to the product
formation. As the conversion was limited by the amount of the
added base, PhBr and styrene were always present in huge excess
in the reaction mixture. The fact that the amount of homoge-
neous Pd complexes increased after 75 min, when the reaction is
almost stopped, and became observable by EXAFS with decreas-
ing reaction rate, indicates that indeed the leaching and the
subsequent formation of homogeneous Pd species are quite
efficient under the applied reaction conditions. Therefore, one
can assume that leaching also takes part during the active phase of
the reaction. The hypothesized presence of homogeneous spe-
cies during the active phase of the catalyst is further supported by
the observed formation of diphenyl, which is formed under the
The currently dominating opinion in the literature for high-
temperature Heck reactions considers the active species in the
heterogeneous Heck coupling to be a homogeneous complex,
which forms upon leaching from in situ formed colloids
(
produced from a homogeneous precursor like Pd(OAc) or
2
7,8,11,37
from a supported Pd catalyst).
In the course of the
reaction, the interaction of the aryl halide with the catalyst
surface causes leaching of Pd into the liquid phase and the
generation of phenyl-bromo-palladate as the active homoge-
4
9-51
neous catalyst.
The leaching ability of PhBr was confirmed
by a control experiment under typical reaction conditions with
3
928
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
II
54,55
inclusion of homogeneous Pd species.
This agrees well with
observed growth of the particle of the supported Pd catalyst.
When the reaction proceeds with a constant rate the particle size
remains constant, and no conclusion concerning a dissolution/
redeposition process of Pd at this stage of the reaction can be
drawn. However it has been found that the presence of a solid
catalyst can favor the redeposition of leached Pd. Under the
assumption that continuous leaching is occurring (see above)
together with the observation, that the Pd concentration in the
liquid phase stays constant we suggest at this stage of the reaction
besides an equilibrium between the colloids and the leached Pd
also an equilibrium between the latter and the supported Pd
particles.
the literature, attributing the activity of such catalytic systems to
7,8
in situ formed homogeneous species. The fact that only Pd
colloids are observed during the active phase of the catalytic
reaction implies that in this case - besides the leaching process
57
-
an additional process leads to the reformation of Pd colloids
and therefore keeps the amount of homogeneous species in the
liquid phase below the observable level for EXAFS. Since the
formation of observable quantities of homogeneous Pd species
just appears when the formation of products fades out, it is
reasonable that one of the steps involved in the formation of the
product is involved in the formation of the colloids under the
given reactions conditions as well. As PhBr and styrene are
present in excess, the formation of an olefin complex is still
possible. Therefore, it seems reasonable that the base is taking
part in the key step initiating the formation of the colloids, since it
is involved in the reaction step that leads to the formation of the
product. This again supports our assumption made above, that
the second catalytic cycle B contributes essentially to the forma-
tion of the products under the applied reaction conditions.
Our findings are in good agreement with newer kinetic
investigations. Although most of the catalytic steps have been
considered to be rate-determining steps in the past years, it is
now accepted that in the case of aryl iodides the rate-determining
’
CONCLUSIONS
The in situ formation of Pd colloids under the reaction
conditions was observed by using a specially designed in situ
EXAFS cell and by a novel QEXAFS analysis approach. The use
of the QEXAFS technique does not only allow catching inter-
mediate species in the subsecond scale but also to use the huge
amount of data generated by the fast data acquisition hardware
for eliminating inhomogeneities in the reaction mixture in the
course of the reaction. In fact, changes in the concentration and
XANES structure of Pd in the solution were occurring within
seconds and sometimes minutes and hours. This allowed aver-
aging over several spectra while the intermediate states were
equilibrated and allowed for appropriate EXAFS analysis. Ran-
dom distortions in the spectra that occurred especially during the
course of the reaction because of fluctuations in liquid sample
systems were successfully eliminated by applying a median filter.
Such filter will generally be a very effective way to improve data
quality in certain regions with random distortions. This shows
that even unknown and much diluted systems can be investigated
effectively with QEXAFS, since it is possible to decide afterward
how many spectra are used for averaging and thus how much
effective acquisition time is used for each spectrum.
12
step is the cleavage of the C-H bond. In the case of the aryl
bromide, the situation is different. Although it was believed that
the oxidative addition of the aryl bromide is the rate-determining
step, it was recently shown that in the case of the high-
temperature ligandless Heck reactions, actually the dissolution
of the Pd from a heterogeneous precursor, i.e., colloidal or
supported Pd, is the rate-determining step. The dissolution of
Pd from soluble Pd colloids is more efficient and therefore most
7
likely the dominating process. The oxidative addition of PhBr to
53
the Pd precursor itself is fast as well.
This fits well in the picture as it implies that Pd colloids are the
dominating species just before the rate-determining steps and
therefore are prone to accumulation (Scheme 2, cycle B).
In accordance with earlier hypotheses, the leaching starts as
soon as the supported catalyst is entirely reduced, which is
speculated to be a prerequisite for the oxidative addition of the
The techniques described above made it possible to follow the
leaching of Pd from a heterogeneous catalyst in situ. A more in-
depth analysis of the in situ spectra of the liquid phase revealed
the presence of both Pd colloids and molecular Pd complexes like
0
11,56
2-
2-
aryl halide as it can only take place with Pd species.
The
[Pd Br ] or [PdBr4] that were formed from leached Pd in
2 6
in situ reduction of the supported Pd catalyst by the reaction
mixture was also observed in other cases, the solvent most likely
being the reducing agent although it cannot be excluded that
other reagents present in the reaction mixture may support the
reduction of the catalyst. Nevertheless, this indicates that, in
general, the applied reaction conditions also possess some
reducing capabilities despite their corrosive character that is
due to the presence of PhBr. Therefore, the reaction mixture
may promote both the leaching of Pd as well as the in situ
formation of Pd colloids. Together with the formation of the
product, this will finally determine the rate of the Pd agglomera-
tion and dissolution. Since the composition of the reaction
mixture changes over time, these rates will also change depending
on the reaction time.
The observed growth of the Pd particles on the solid catalyst
illustrates nicely the hypothesized dissolution/redeposition pro-
cess of Pd at the end of the reaction. As soon as the reaction rate
decreases, the concentration of homogeneous Pd species in the
solution increases because of the decomposition of the colloids.
Consequently, the equilibrium of the dissolution/redeposition
process is shifted in favor of the Pd redeposition leading to the
the liquid phase. During the active phase of the catalysis, mainly
Pd colloids were observed, whereas after full conversion (relative
to the base) the fraction of Pd complexes increased substantially
as the colloids are decomposed partly. This implies the following
conclusions for the reaction mechanism: (i) the reaction is
catalyzed by molecular Pd complexes which are formed in situ,
(ii) the rate-determining step of the Heck coupling reaction with
PhBr is most likely the dissolution of Pd from soluble Pd colloids
and therefore (iii) these colloids are directly involved in the
catalytic cycle and do not only serve as a reservoir when
nonactivated substrates like PhBr are used (Scheme 2, cycle B).
The special design of the reactor made it also possible to track
the changes of the heterogeneous catalyst in the catalyst bed. The
analysis of the corresponding XANES spectra showed a reduc-
tion of the Pd, indicating the potential reducing capabilities of the
reaction mixture. Furthermore, careful fitting of the EXAFS
spectra of the catalyst bed allowed following the changes in the
Pd particle size. During the active phase of the catalysis, a
constant particle size and concentration were observed, most
likely indicating a continuous leaching and redeposition process
of Pd. The study indicates the high potential of QEXAFS to
16
3
929
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930

Journal of the American Chemical Society
ARTICLE
elucidate complex reaction dynamics in industrially important
reaction mixtures, not only interesting for C-C couplings with
other aryl halides but also for related reaction systems.
(20) Frahm, R. Nucl. Instrum. Methods Phys. Res., Sect. A 1988,
270, 578–581.
(
(
21) Frahm, R. Rev. Sci. Instrum. 1989, 60, 2515–2518.
22) L €u tzenkirchen-Hecht, D.; Grunwaldt, J.-D.; Richwin, M.; Grie-
sebock, B.; Baiker, A.; Frahm, R. Phys. Scr. 2005, T115, 831–833.
’
ASSOCIATED CONTENT
(23) Grunwaldt, J.-D.; Beier, M.; Kimmerle, B.; Baiker, A.;
Nachtegaal, M.; Griesebock, B.; L €u tzenkirchen-Hecht, D.; St €o tzel, J.;
Frahm, R. Phys. Chem. Chem. Phys. 2009, 11, 8779–8789.
S
Supporting Information. Additional scanning transmis-
b
sion electron microscopy (STEM) images of the supported
catalyst after reaction (Figures S1 and S2), the fitting results
for the third shell of the supported Pd particles (Figure S3), and
the edge jump of the Pd K-edge in the liquid phase after heating
to 150 ꢀC when only PhBr (1) was present in the reaction
mixture (Figure S4). This material is available free of charge via
the Internet at http://pubs.acs.org.
(24) Clausen, B. S.; Topsøe, H.; Frahm, R. Adv. Catal. 1998, 42, 315.
(25) Grunwaldt, J.-D.; Ramin, M.; Rohr, M.; Michailovski, A.;
Patzke, G. R.; Baiker, A. Rev. Sci. Instrum. 2005, 76, 054104.
(26) Reimann, S.; Mallat, T.; Baiker, A. J. Catal. 2007, 252, 30–38.
(27) Frahm, R.; Richwin, M.; L €u tzenkirchen-Hecht, D. Phys. Scr.
2005, T115, 974–976.
(
28) St €o tzel, J.; L €u tzenkirchen-Hecht, D.; Frahm, R. Rev. Sci. Instrum.
010, 81, 073109.
29) St €o tzel, J.; L €u tzenkirchen-Hecht, D.; Fonda, E.; De Oliveira, N.;
Briois, V.; Frahm, R. Rev. Sci. Instrum. 2008, 79.
30) St €o tzel, J.; L €u tzenkirchen-Hecht, D.; Frahm, R. J. Synchrotron
Rad. 2011, in print, doi:10.1107/S0909049510051897.
2
(
’
AUTHOR INFORMATION
(
Corresponding Authors
grunwaldt@kit.edu; baiker@chem.ethz.ch
(
(
(
31) Newville, M. J. Synchrotron Rad. 2001, 8, 322–324.
32) Rehr, J. J.; Albers, R. C. Rev. Mod. Phys. 2000, 72, 621–654.
33) Malinowski, E. R. Factor Analysis in Chemistry, 3rd ed.: Wiley-
’
ACKNOWLEDGMENT
Interscience: New York, 2002.
34) Jackson, J. E. A user’s guide to principal components; John Wiley
Financial support by the Swiss National Science Foundation is
(
kindly acknowledged. We thank Dr. Maarten Nachtegaal for his
assistance during XAS measurements and the SLS at PSI
Switzerland) for providing beam time. Electron microscopy
&
Sons: New York, 1991.
(35) Ressler, T. J. Synchrotron Rad. 1998, 5, 118–122.
(36) Pr €o ckl, S. S.; Kleist, W.; K €o hler, K. Tetrahedron 2005,
(
investigations were carried out by Dr. Frank Krumeich at the
Electron Microscopy ETH Zurich (EMEZ). Finally, we thank
Prof. Klaus K €o hler (TU M €u nchen) for stimulating discussions,
especially in the beginning of the study.
61, 9855–9859.
37) Biffis, A.; Zecca, M.; Basato, M. J. Mol. Catal. A: Chem. 2001,
73, 249–274.
38) Phan, N. T. S.; Van Der Sluys, M.; Jones, C. W. Adv. Synth.
Catal. 2006, 348, 609–679.
39) Dalba, G.; Fornasini, P.; Rocca, F.; Mobilio, S. Phys. Rev. B
(
1
(
(
’
REFERENCES
1
990, 41, 9668–9675.
(
1) Blaser, H.-U.; Indolese, A.; Naud, F.; Nettekoven, U.; Schnyder,
(40) L €u tzenkirchen-Hecht, D.; Strehblow, H.-H. Surf. Interface Anal.
2006, 38, 686–690.
A. Adv. Synth. Catal. 2004, 346, 1583–1598.
(
(
(
(
2) Torborg, C.; Beller, M. Adv. Synth. Catal. 2009, 351, 3027–3043.
3) Heck, R. F.; Nolley, J. P. J. Org. Chem. 1972, 37, 2320–2322.
4) Mizoroki, T.; Mori, K.; Ozaki, A. Bull. Chem. Soc. Jpn. 1971, 44, 581.
5) Oestreich, M., Ed. The Mizoroki-Heck Reaction; John Wiley &
(41) Jentys, A. Phys. Chem. Chem. Phys. 1999, 1, 4059–4063.
(42) Makitova, D. D.; Korableva, L. G.; Kraschoka, O. N.; Lavren-
t’ev, I. P.; Atovmyan, L. O.; Khidekel, M. L. Koord. Khim. 1986,
12, 1262–1269.
(43) Waser, J.; Levy, H. A.; Peterson, S. W. Acta Crystallogr. 1953,
6, 661–663.
(44) Grunwaldt, J.-D.; Caravati, M.; Baiker, A. J. Phys. Chem. B 2006,
110, 9916–9922.
(45) Kittel, C.; McEuen, P. Introduction to Solid State Physics; John
Wiley & Sons: New York, 2005.
(46) Clausen, B. S.; Nørskov, J. K. Top. Catal. 2000, 10, 221–230.
(47) Grunwaldt, J.-D.; Molenbroek, A. M.; Topsøe, N. Y.; Topsøe,
H.; Clausen, B. S. J. Catal. 2000, 194, 452–460.
(48) Mackay, A. L. Acta Crystallogr. 1962, 15, 916–918.
(49) Schmidt, A. F.; Mametova, L. V. Kinet. Catal. 1996,
37, 406–408.
Sons: Chichester, 2009.
(
(
(
(
6) Biffis, A.; Zecca, M.; Basato, M. Eur. J. Inorg. Chem. 2001, 1131–1133.
7) de Vries, J. G. Dalton Trans. 2006, 421–429.
8) Astruc, D. Inorg. Chem. 2007, 46, 1884–1894.
9) Kaneda, K.; Higuchi, M.; Imanaka, T. J. Mol. Catal. 1990,
6
9
4
2
1
3, L33–L36.
10) Augustine, R. L.; Oleary, S. T. J. Mol. Catal. A: Chem. 1995,
5, 277–285.
11) K €o hler, K.; Kleist, W.; Pr €o ckl, S. S. Inorg. Chem. 2007,
6, 1876–1883.
12) Schmidt, A. F.; Al Halaiqa, A.; Smirnov, V. V. Synlett
006, 2861–2878.
13) Kleist, W.; Pr €o ckl, S. S.; K €o hler, K. Catal. Lett. 2008,
25, 197–200.
14) Michailowski, A.; Grunwaldt, J.-D.; Baiker, A.; Kiebach, R.;
Bensch, W.; Patzke, G. R. Angew. Chem. 2005, 44, 5463.
15) Evans, J.; O’Neill, L.; Kambhampati, V. L.; Rayner, G.; Turin, S.;
Genge, A.; Dent, A. J.; Neisius, T. J. Chem. Soc., Dalton Trans.
002, 2207–2212.
16) Reimann, S.; Grunwaldt, J.-D.; Mallat, T.; Baiker, A. Chem. Eur.
J. 2010, 16, 9658–9668.
17) Ellis, P. J.; S., F. I. J.; Hackett, S. F. J.; Wilson, K.; Lee, A. F.
Angew. Chem., Int. Ed. 2010, 49, 1820–1824.
18) Aschwanden, L.; Mallat, T.; Grunwaldt, J.-D.; Krumeich, F.;
Baiker, A. J. Mol. Catal. A: Chem. 2009, 300, 111.
19) Rohr, M.; Grunwaldt, J.-D.; Baiker, A. J. Mol. Catal. A: Chem.
005, 266, 253–257.
(
(
(
(
(50) Reetz, M. T.; Westermann, E. Angew. Chem., Int. Ed. 2000,
39, 165–168.
(
(51) Gniewek, A.; Trzeciak, A. M.; Ziolkowski, J. J.; Kepinski, L.;
Wrzyszcz, J.; Tylus, W. J. Catal. 2005, 229, 332–343.
(52) Schmidt, A. F.; Khalaika, A. Kinet. Catal. 1996, 39, 803–809.
(53) Schmidt, A. F.; Al-Halaiqa, A.; Smirnov, V. V. Kinet. Catal.
2007, 48, 716–727.
(54) McGuinness, D. S.; Cavell, K. J.; Skelton, B. W.; White, A. H.
Organometallics 1999, 18, 1596–1605.
(55) Schmidt, A. F.; Smirnov, V. V.; Al-Halaiga, A. Kinet. Catal. 2007,
48, 390–397.
(56) K €o hler, K.; Pr €o ckl, S. S.; Kleist, W. Curr. Org. Chem. 2006,
10, 1585–1601.
(
2
(
(
(
(
(57) Huang, L.; Wong, P. K.; Tan, J.; Ang, T. P.; Wang, Z. J. Phys.
Chem. C 2009, 113, 10120–10130.
2
3
930
dx.doi.org/10.1021/ja108636u |J. Am. Chem. Soc. 2011, 133, 3921–3930