Real-time control of a catalytic solid in a fixed-bed reactor based on in situ spectroscopy

Article abstract of DOI:10.1002/anie.200700499

A reality check: Reactor control by online spectroscopy is illustrated for a Cr/Al₂O₃ alkane dehydrogenation catalyst and a vanadium phosphorus oxide catalyst used in the selective oxidation of n-butane to maleic anhydride. Rather than testing the product composition alone, the methodology monitors the state of the catalyst in the reactor, and changes the reaction conditions to keep the catalyst in its optimal state (see picture). (Figure Presented).

Full text of DOI:10.1002/anie.200700499

Communications
DOI: 10.1002/anie.200700499
Heterogeneous Catalysis
Real-Time Control of a Catalytic Solid in a Fixed-Bed Reactor Based
on In Situ Spectroscopy**
Simona M. Bennici, Bas M. Vogelaar, T. Alexander Nijhuis, and Bert M. Weckhuysen*
Large-scale reactors loaded with catalytic solids are employed
in a wide variety of chemical plants all over the world for the
production of transportation fuels, bulk and fine chemicals,
and pharmaceutical intermediates. These reactors are com-
monly controlled by monitoring the chemicals that are leaving
the catalytic process. Once the product composition deviates
from the desired one, process-control actions are taken to
correct for this change. Important parameters for running
most economically large-scale industrial reactors based on
heterogeneous catalysis for bulk chemicals production are the
reaction temperature and pressure, as well as the chemical
composition and physical properties of the reaction mixture.
Often such corrective scenarios are based on empirical
models, which are obtained from plant test runs.[1]
A better approach is to monitor online the physicochem-
ical phenomena of the catalyst material in the reactor by in
situ spectroscopy,[2–5] and if possible to develop an “expert
system” based on such spectroscopic measurements. Exam-
ples include coke formation (e.g., leading to active-site
blocking), catalyst reduction (e.g., over-reduction of the
catalyst surface), change of catalyst dispersion (e.g., metal-
oxide clustering), migration of the oxide phase into the oxide
support (e.g., metal-ion migration into the support oxide),
and change of phase (e.g., solid-phase transformation during
a redox reaction).[6–10] With such an expert system, it should be
possible to make corrective actions once a spectroscopically
visible change occurs in the catalyst material. This procedure
is faster than monitoring the off-specification product leaving
the reactor. In particular, for catalytic processes in which the
catalyst is operated with alternating reactive and regenerative
cycles, the use of in situ spectroscopy will have specific
advantages. Indeed, monitoring the catalyst allows the
possibility of dynamically tuning the cycles while the process
runs. In an industrial context, where the feedstock composi-
tion of a process is not always identical and process
disturbances are a reality, it would be especially advantageous
to adapt the cycle times to the actual state of the “catalyst
health” measured online with a robust in situ spectroscopic
tool.
The use of online spectroscopic techniques to obtain data
for process trend monitoring has already been applied in
polymer science,[11,12] whereas to the best of our knowledge
the only catalytic process in which in situ spectroscopy is
applied as a means of process control is the production of 1,3-
propanediol, which makes use of a homogeneous catalyst.[13]
The novelty of the present work is linked to the use of in situ
spectroscopy for the online control of different heteroge-
neous catalytic systems. Herein, we illustrate for the first time
the concept of reactor control by online spectroscopy, by
making use of two examples: an industrial-like supported
chromium oxide (Cr/Al2O3) catalyst for propane dehydrogen-
ation and an industrial-like vanadium phosphorus oxide
(VPO) catalyst for the selective oxidation of n-butane to
maleic anhydride.
In many industrial catalytic processes, coke deposition is
the primary cause of catalyst deactivation.[6,14] Therefore, any
means of directly measuring the amount of coke on the
catalyst surface would be welcomed, especially for the
purpose of online control of a chemical process. In this
manner, one can keep the catalyst material in its optimal state
in terms of activity, selectivity, and stability. For this reason, as
a first example we have chosen a methodology to perform
real-time quantitative coke analysis on a catalytic solid by
using combined in situ Raman–UV/Vis spectroscopy with
online calibration. The studied process is based on the Catofin
propane dehydrogenation reaction over Cr/Al2O3 cata-
lyst.[15–17] In this process, the catalyst dehydrogenates propane
in a reactive cycle, during which coke is deposited on the
catalyst and some of the chromium is reduced. In a
regenerative cycle the chromium is reoxidized and the coke
is burned with air. Currently, for the industrial Catofin
process, the real state of the catalyst material is not taken into
consideration and the cycle times are based on the overall
performance of the unit. In this manner, gradual catalyst
deactivation is compensated by an increase of the reaction
temperature to maintain the desired conversion level. The
gradual recovery of the activity declines during the lifetime of
the catalyst as a result of migration of chromium (as Cr3+) into
[#]
[*] Dr. S. M. Bennici,[+] Dr. Ir. B. M. Vogelaar,[$] Dr. Ir. T. A. Nijhuis,
Prof. Dr. Ir. B. M. Weckhuysen
Inorganic Chemistry and Catalysis Group
Department of Chemistry
Utrecht University
P.O. Box 80083, 3508 TB Utrecht (The Netherlands)
Fax: (+31)30-251-1027
E-mail: b.m.weckhuysen@chem.uu.nl
[+] Current address:
IRCELYON, UMR 5256 CNRS/UniversitØ Lyon 1
2 avenue Albert Einstein, 69626 Villeurbanne Cedex (France)
[$] Current address:
Albemarle Catalysts
P.O. Box 37650, 1030 BE Amsterdam (The Netherlands)
[#] Current address:
Department of Chemical Engineering
Technische Universiteit Eindhoven
P.O. Box 513, 5600 MB Eindhoven (The Netherlands)
[**] The ASPECT (ACTS) and CW-VICI (NWO) funding schemes are
kindly acknowledged for financial support.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
5412
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 5412 –5416

Angewandte
Chemie
the alumina support, which occurs at the high temperatures
commonly found in the hot spots of the reactor bed. In
particular, the risk of permanent deactivation is very high
during the regeneration step, as the exothermic coke com-
bustion process may lead to hot spots in the catalyst bed.
An example of the controlled coke burn-off procedure is
illustrated in Figure 1, in which coke and propane conversion
In the developed methodology, the catalytic solid is
automatically regenerated when the amount of coke exceeds
a predefined level (determined from the intensity of the
Raman peak at 1580 cmÀ1, see insert in Figure 1).[15,16] The
regeneration cycle is stopped automatically when the coke
level drops below a given level. Furthermore, the oxygen
concentration is gradually increased in four stages during
regeneration, based on the actual level of coke. This results in
a controlled coke burn-off, which reduces the formation of
hot spots during catalyst regeneration and maintains the small
amount of coke that seems to have a positive effect on the
activation of the catalyst.[18] In the expert-system “control
algorithm” used in this experiment, the coke level on the
catalyst (“coke”) was the control variable. The first propane
dehydrogenation cycle (see Figure 1) was carried out at
5508C until the coke level exceeded a predefined limit (250 in
these experiments). Then, after a flushing step (1min,
20 mLminÀ1 He), the feed gas was switched to a diluted
oxygen stream (2% O2/He) to kinetically limit the rate of
coke burn-off until the coke level reached a second prede-
fined limit (100 in these experiments). The regeneration
procedure was prolonged at increasing oxygen concentrations
of 4, 10, and 20% to reach a lower coke limit in each stage (for
the subsequent three regeneration stages the coke limit was
fixed at 50, 20, and 10, respectively). After the last stage of
regeneration a new propane dehydrogenation cycle was
performed at increased temperature (> 10 up to 6008C) to
verify the robustness of the proposed methodology by
checking how it would behave in different conditions. Further
details on the developed methodology can be found in the
Supporting Information.
Figure 1. Real-time control of a Cr/Al2O3 propane dehydrogenation
catalyst by online in situ Raman–UV/Vis spectroscopy. Propane
conversion levels and amounts of coke are reported versus time on
stream t for different reaction cycles and temperatures. The control
system continues the reaction step until a given coke limit is reached
(a), after which the catalyst is automatically regenerated. Inset: in
situ Raman spectra indicative of increasing amounts of coke during a
propane dehydrogenation step.
To illustrate the beneficial effect of controlled coke burn-
off on the catalyst temperature, a series of experiments was
performed in which the reaction temperature was measured
online for “standard” and “controlled” activation procedures
(see Figure 2). Ten reaction–regeneration cycles are reported
for temperatures in the range of 550–6008C. The regeneration
levels are plotted versus time on stream, and the upper coke
level set-point is also indicated. A clear relation was found
between the presence of a certain amount of coke and the
maximum propane conversion of the catalyst. The reaction
and regeneration cycle times were automatically optimized as
a function of reaction temperature and deactivation or coking
rate: coke buildup appeared to be slower in the last two cycles
at 550 and 5608C, with a consequential increase of the
reaction cycle time. This slower kinetics in the formation of
coke could be related to the permanent catalyst deactivation
after exposure at high temperature. The approximately
20 minutes necessary to regenerate the catalyst in the
controlled coke burn-off step was slightly longer than the
cycle time currently used in the chemical industry. However,
the time during which propane could be dehydrogenated at
conversions > 20% was much longer than in industrial
practice (1to 3 h in dependence on the reaction temperature).
Note that in industrial reactors cycle times are typically
shorter because of heat consumption by the endothermic
reaction. The cycles performed at the highest temperatures
(590 and 6008C) presented some coke-level overshoots, which
are related to the delay between the flushing and regeneration
steps.
Figure 2. Beneficial effect of online spectroscopic control of a Cr/Al2O3
propane dehydrogenation catalyst. Catalyst temperature Tcat and coke
buildup during ten reaction–activation cycles, with alternating “stan-
dard” regeneration (fixed duration, 20% O2) and “controlled” regener-
ation (gradually increasing O2 concentration until the coke is
removed), are shown.
Angew. Chem. Int. Ed. 2007, 46, 5412 –5416ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5413

Communications
procedure was alternated between controlled burn-off (as in
Figure 1) and a standard regeneration procedure in air, which
is currently employed in industrial propane dehydrogenation
plants. In each case, at the beginning of the regeneration step
we noticed a sharp increase in the temperature of the catalyst
bed. During the controlled burn-off, the increment in this
temperature was three times smaller than in the standard
regeneration procedure (increments of 25 and 758C, respec-
tively). The lower temperature reached during regeneration
by controlled burn-off (Figure 2) proves the benefits of this
regeneration procedure; the possibility of reducing the
formation of hot spots in an industrial reactor, and thus
suppressing the irreversible deactivation of the catalyst (that
is, migration of Cr3+ into the Al2O3 support), could offer a
simple way to extend the catalyst lifetime.
In a second example, we explored the selective oxidation
of n-butane to maleic anhydride over an industrial-like VPO
catalyst. This reaction is presently the only industrial process
of light alkane oxidation on a large scale. The great
importance of this reaction and the unique chemical–physical
properties of the VPO-based catalysts make this system a hot
research topic.[19] To better understand and improve the
existing catalyst material and catalytic processes involved in
maleic anhydride production, different and complementary
physicochemical techniques have been used to provide
information on the bulk and surface properties of the VPO
system.[20–25] Among them, Raman spectroscopy demon-
strated its power in detecting the transformation of the
VPO precursor into the active VPO catalyst.[26–31] For this
reason, the second example is based on the use of in situ
Raman spectroscopy as a tool for the online control of the
catalyst performance of the VPO system.
Two series of experiments were performed at a constant
temperature of 4008C, one devoted to control of the phase
changes occurring by alternating reaction (React, He/O2/
C4H10 = 77:21:2), oxidizing (Ox, He/O2 = 79:21), and reducing
(Red, He/C4H10 = 98:2) steps, and the other to real-time
control of the reversible transformation between (VO)2P2O7
and VOPO4 polymorphs, representative of five-valence
vanadium.[22,32–35] In Figure 3A and B, the Raman peak
intensity at 989 cmÀ1 and the maleic anhydride concentration
trends are reported as a function of the reaction time.
Examples of Raman spectra acquired under different reaction
conditions are reported in Figure 3C, while the Raman
spectra of the relatively pure phases present on the catalyst
are shown in Figure 3D.[26,28] The experiment was carried out
by alternating the reaction, oxidizing, and reducing conditions
for five React-Ox-React-Red cycles, as shown schematically
at the bottom of Figure 3, where the presence/absence of
butane and/or oxygen is reported. The increased/decreased
intensity of the 989 cmÀ1 Raman peak was clearly reversible
and, after each Ox or Red step, the maleic anhydride
concentration could be rapidly reestablished when the
catalyst was reexposed to React conditions.
Figure 3. Catalytic performances and related in situ Raman spectra of
a VPO catalyst as a function of feed gas composition. Raman
intensities I989 of the 989 cmÀ1 peak (A) and maleic anhydride (MA)
concentration (B) are shown as a function of time on stream during
alternating reaction, oxidizing, and reduction conditions (indicated at
the bottom). Examples of Raman spectra C) acquired 1) in the absence
of oxygen, 2) during reaction, and 3) in the absence of butane, and
D) of the pure-phase components acquired at 4008C. The 932 and
989 cmÀ1 peaks are indicated with arrows.
system: when the butane was removed from the reactant
mixture (Ox condition), as expected the maleic anhydride
concentration rapidly decreased to zero, while in the case of
the Red condition (no O2 present) the maleic anhydride
concentration diminished to a value of only 60 ppm. This
could be related to the fact that VPO operates as an oxidation
catalyst by a Mars–van Krevelen mechanism, that is, in the
absence of oxygen in the feed, the butane reacts to give maleic
anhydride by reacting with lattice oxygen.[36,37] This behavior
could also be deduced by observing the trend in maleic
anhydride concentration during the React steps (Figure 3). In
the React steps that followed the Ox treatments, the activity
of the VPO catalyst was immediately reestablished and
showed a constant trend, whereas after the Red treatments,
an increasing trend in the maleic anhydride concentration
could be observed that was related to the time request to
reoxidize the catalyst and replace the lattice oxygen, in
accordance with the reaction model presented by Gascón
et al.[38]
Examples of the characteristic Raman spectra acquired in
the different steps are reported in Figure 3C. The possibility
of acquiring real-time Raman spectra of the catalyst surface
and at the same time analyzing the reaction products
permitted observation of behavior peculiar to the VPO
The possibility of controlling the phase composition of the
VPO catalyst by online in situ Raman spectroscopy is
illustrated in Figure 4, which shows the intensity ratios of
5414
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 5412 –5416

Angewandte
Chemie
potential industrial impact of this new spectroscopic
approach.
Experimental Section
The first example dealing with propane dehydrogenation was
conducted with a flow setup similar to the one already described by
Nijhuis et al.,[15,16] at atmospheric pressure, in which Cr/Al2O3 catalyst
(0.3 g) was positioned in a quartz fix-bed reactor and tested in cycles
at different reaction temperatures (550–6008C). The gas stream
during the reaction step was composed of propane and He (2 and
20 mLminÀ1, respectively). After each cycle of reaction the catalyst
was regenerated in a flow of O2/He, in which the concentration of O2
was varied from 2 to 20% in relation to the control system already
described, or by standard regeneration in air. UV/Vis spectra of the
catalyst surface were collected (with a time resolution of 1s) by an
Avantes AvaSpec-4 analyzer with four high-temperature diffuse
reflectance probes (Ocean Optics/Micropack), focused at different
heights on the catalyst bed. At the same time, a Kaiser RXN Raman
spectrometer, equipped with a 532-nm diode laser and a noncontact
objective, was used to analyze the sample with a total recording time
of about 3 min (ten accumulations with a 3-s exposure time). The
reactor outlet stream was analyzed online using an Interscience
compact gas chromatography (GC) system, equipped with Porabond-
Q and Molsieve 5A columns, capable of performing analyses within
1.3 min.
Figure 4. Real-time control of a VPO catalyst by online in situ Raman
spectroscopy. Intensity ratios of the 989 and 932 cmÀ1 Raman peaks
(A) and maleic anhydride (MA) production (B) are shown as a
function of time on stream. The control system alternates reaction and
lean oxygen conditions (indicated at the top) until the given Iox/Ired
limit is reached (a).
In the second example, the home-made VOHPO4·0.5H2O
catalyst precursor was prepared in organic medium[26] and activated
in situ (reaction conditions: 4008C, total flow: 30 mLminÀ1, molar
ratio: He/O2/C4H10 = 77:21:2, 24 h) to form the activated VPO
catalyst (mainly (VO)2P2O7). The setup was the same as that
described for the first example (see also the Supporting Information).
However, the analysis of the product gases was performed within
170 s with an online Interscience compact GC system equipped with
three capillary columns (Rtx-1.5u, PorbondU, and PorbondQ) and
two different detectors (flame ionization and thermal conductivity
detectors). All the pipes, starting from the exit of the reactor to the
end of the catalytic line, were constantly heated at 1008C to avoid
condensation of the maleic anhydride. Ten Raman spectra were
accumulated within 6 s exposure time for a total spectral recording
time of about 5 min. A second series of experiments was performed
on 0.3 g of catalyst, by alternating reaction condition (molar ratio:
He/O2/C4H10 = 77:21:2) and lean oxygen steps (molar ratio: He/O2/
C4H10 = 77:3:2).
the Raman bands at 989 and 932 cmÀ1 (Iox/Ired, related to V5+
and V4+ phases, respectively)[39] for alternating reaction
conditions, as shown schematically for the oxygen concen-
tration as a function of time on stream. The lean oxygen steps
were maintained until an arbitrarily chosen Iox/Ired value of
0.19 was reached. Once this lower-limit value was obtained,
the lean oxygen step was automatically terminated and the
reaction condition restored. The first three cycles were
affected by the changing of phases as a result of catalyst
activation that, after 24 hours, was not yet complete,[40] as
demonstrated by the global increasing trend in the maleic
anhydride concentration (Figure 4B). After ten days of
activation (4008C, 30 mLminÀ1, He/O2/C4H10 = 77:21:2) the
transformation towards (VO)2P2O7 was complete, and the Iox/
Ired ratio and maleic anhydride concentration in the reaction
steps showed a constant trend. The benefit of this online
control of phase composition can be applied to prolong the
life of the VPO catalyst, thus avoiding in time any dramatic
phase changes that could not only decrease the catalyst
performance, but also result in permanent catalyst deactiva-
tion, as explained above for the Cr/Al2O3 catalyst system.
On the basis of both examples it is clear that in situ
spectroscopy of a catalytic solid can be used to control online
the amount of a specific catalyst material as well as that of
coke deposits during a reaction process. Based on this input
the catalyst performance and its long-term stability can be
automatically controlled. We expect that the developed
methodology will be of general use for developing “catalyst
diagnostics” strategies, in cases where catalyst deactivation
processes are related to the blocking of active sites by coke, as
well as by the formation of catalyst phases that contain less
active and/or selective active sites. More specifically, based on
the promising results for the two examples discussed, catalyst
users and producers have now decided to perform test
experiments in pilot-scale reactors, thus illustrating the
Received: February 4, 2007
Revised: March 12, 2007
Published online: June 6, 2007
Keywords: heterogeneous catalysis · process control ·
.
Raman spectroscopy · supported catalysts · UV/Vis spectroscopy
[1] J. S. Lin, S. S. Jang, S. S. Shieh, M. Subramaniam, Ind. Eng.
Chem. Res. 1999, 38, 4700.
[2] Examples of relevant textbooks are: a) In Situ Spectroscopy in
Heterogeneous Catalysis (Ed.: J. F. Haw), Wiley-VCH, Wein-
heim, 2002; b) In Situ Spectroscopy of Catalysts (Ed.: B. M.
Weckhuysen), American Scientific, Stevenson Ranch, 2004;
c) J. W. Niemansverdriet, Spectroscopy in Catalysis: An Intro-
duction, VCH, Weinheim, 1993.
[3] For examples of relevant review papers, see: a) M. Hunger, J.
Weitkamp, Angew. Chem. 2001, 113, 3040; Angew. Chem. Int.
Ed. 2001, 40, 2954; b) B. M. Weckhuysen, Chem. Commun. 2002,
97; c) B. M. Weckhuysen, Phys. Chem. Chem. Phys. 2003, 5,
4351; d) H. Topsøe, J. Catal. 2003, 216, 155; e) M. A. Banares,
Angew. Chem. Int. Ed. 2007, 46, 5412 –5416ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5415

Communications
Catal. Today 2005, 100, 71; f) I. E. Wachs, Catal. Today 2005, 100,
79.
[21] M. P. Casaletto, G. Mattogno, M. A. Massucci, Appl. Surf. Sci.
2003, 211, 216.
[4] G. Mestl, J. Mol. Catal. A 2000, 158, 45.
[22] M. A. Carreon, V. V. Guliants, Catal. Today 2003, 78, 303.
[23] L. OꢀMahony, T. Curtin, D. Zemlyanov, M. Mihov, B. K.
Hodnett, J. Catal. 2004, 227, 270.
[24] M. Hävecker, A. Knop-Gericke, H. Bluhm, E. Kleimenov, R. W.
Mayer, M. Fait, R. Schlögl, Appl. Surf. Sci. 2004, 230, 272.
[25] T. Giornelli, A. Löfberg, E. Bordes-Richard, Thin Solid Films
2005, 479, 64.
[5] a) S. Bordiga, A. Damin, F. Bonino, G. Ricchiardi, C. Lamberti,
A. Zecchina, Angew. Chem. 2002, 114, 4928; Angew. Chem. Int.
Ed. 2002, 41, 4734; b) E. Groppo, C. Prestipino, F. Cesano, F.
Bonino, S. Bordiga, C. Lamberti, P. C. Thune, J. W. Niemants-
verdriet, A. Zecchina, J. Catal. 2005, 230, 98.
[6] E. H. Wolf, A. Alfani, Catal. Rev. Sci. Eng. 1982, 24, 329.
[7] C. H. Bartholomew, Appl. Catal. A 2001, 202, 1 7.
[8] M. L. Occelli, J. P. Olivier, A. Auroux, J. Catal. 2002, 209, 385.
[9] I. Rossetti, E. Bencini, L. Trentini, L. Forni, Appl. Catal. A 2005,
292, 118.
[10] J. Aßmann, D. Crihan, M. Knapp, E. Lundgren, E. Loffler, M.
Muhler, V. Narkhede, H. Over, M. Schmid, A. P. Seitsonen, P.
Varga, Angew. Chem. 2005, 117, 939; Angew. Chem. Int. Ed.
2005, 44, 917.
[11] P. D. Coates, S. E. Barnes, M. G. Sibley, E. C. Brown, H. G. M.
Edwards, I. J. Scowen, Polymer 2003, 44, 5937.
[12] A. Tuchbreiter, J. Marquardt, B. Kappler, J. Honerkamp, M. O.
Kristen, R. Mülhaupt, Macromol. Rapid Commun. 2003, 24, 47.
[13] P. R. Weider, J. B. Powell, K. D. Allen, J. Knifton, Examination
of the Hydroformylation of Ethylene Oxide by In Situ IR and the
Development of a Commercial Route to 1,3-Propanediol, pre-
sented at the 1st Conference on Operando Spectroscopy,
Lunteren (The Netherlands), 2003.
[26] F. Ben Abdelouahab, R. Olier, N. Guilhaume, F. Lefebvre, J. C.
Volta, J. Catal. 1992, 134, 151.
[27] H. Knözinger, Catal. Today 1996, 32, 71 .
[28] Z. Y. Xue, G. L. Schrader, J. Phys. Chem. B 1999, 103, 9459.
[29] F. J. Cabello Sanchez, R. P. K. Wells, C. Rhodes, J. K. Bartley,
C. J. Kiely, G. J. Hutchings, Phys. Chem. Chem. Phys. 2001, 3,
4122.
[30] N. Topsøe, M. Anstrom, J. A. Dumesic, Catal. Lett. 2001, 76, 1 .
[31] M. V. Martꢁnez-Huerta, X. Gao, H. Tian, I. E. Wachs, J. L. G.
Fierro, M. A. Baꢂares, Catal. Today 2006, 118, 279.
[32] R. Mallada, S. Sajip, C. J. Kiely, M. Menꢃndez, J. Santamarꢁa, J.
Catal. 2000, 196, 1 .
[33] M. Conte, G. Budroni, J. K. Bartley, S. H. Taylor, A. F. Carley, A.
Schmidt, D. M. Murphy, F. Girgsdies, T. Ressler, R. Schlögl, G. J.
Hutchings, Science 2006, 313, 1270.
[34] X. Huang, B. Chen, B. Liu, P. L. Silveston, C. Li, Catal. Today
2002, 74, 121.
[14] J. R. Rostrup-Nielsen, Catal. Today 1997, 37, 225.
[15] T. A. Nijhuis, S. J. Tinnemans, T. Visser, B. M. Weckhuysen,
Phys. Chem. Chem. Phys. 2003, 5, 4361.
[16] S. J. Tinnemans, M. H. F. Kox, T. A. Nijhuis, T. Visser, B. M.
Weckhuysen, Phys. Chem. Chem. Phys. 2005, 7, 211.
[17] M. M. Bhasin, J. H. McCain, B. V. Vora, T. Imai, P. R. Pujadó,
Appl. Catal. A 2001, 221, 397.
[18] S. T. Korhonen, S. M. K. Airaksinen, A. O. I. Krause, Catal.
Today 2006, 112, 37.
[19] N. Ballarini, F. Cavani, C. Cortelli, S. Ligi, F. Pierelli, F. Trifirò, C.
Fumagalli, G. Mazzoni, T. Monti, Top. Catal. 2006, 38, 147.
[20] G. W. Coulston, E. A. Thompson, N. Herron, J. Catal. 1996, 163,
122.
[35] D. Wang, H. H. Kung, M. A. Barteau, Appl. Catal. A 2000, 201,
203.
[36] M. J. Lorences, G. S. Patience, R. Cenni, F. Dꢁez, J. Coca, Catal.
Today 2006, 112, 45.
[37] M. Ruitenbeek, A. J. van Dillen, F. M. F. de Groot, I. E. Wachs,
J. W. Geus, D. C. Koningsberger, Top. Catal. 2000, 10, 241.
[38] J. Gascón, R. Valenciano, C. Tꢃllez, J. Herguido, M. Mꢃnendez,
Chem. Eng. Sci. 2006, 61, 6385.
[39] M. Abon, J. M. Herrmann, J. C. Volta, Catal. Today 2001, 71, 121.
[40] V. V. Guliants, J. B. Benziger, S. Sundaresan, N. Yao, Catal. Lett.
1995, 32, 379.
5416
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 5412 –5416