High-throughput screening of catalysts by combining reaction and analysis

Article abstract of DOI:10.1002/anie.200701326

(Figure Presented) Drudgery minimized, efficiency maximized: By combining catalysis and separation in microcapillaries greater than 2 cm in length, it is possible to efficiently determine the reaction kinetics for entire libraries of substrates. This was demonstrated for hydrogenations over highly active Pd nanoparticles and ring-closing metatheses over the Grubbs 2nd generation catalyst. R: reagents, P: products.

Full text of DOI:10.1002/anie.200701326

Angewandte
Chemie
DOI: 10.1002/anie.200701326
Catalyst Screening
High-Throughput Screening of Catalysts by Combining Reaction and
Analysis**
Oliver Trapp,* Sven K. Weber, Sabrina Bauch, and Werner Hofstadt
Dedicated to Süd-Chemie on the occasion of its 150th anniversary
Discovering new catalysts is crucial to the development of
sustainable chemical processes for industrial applications as
well as for broadening the spectrum of synthetic method-
ologies and techniques in chemistry. The prerequisite for the
directed design of catalysts is understanding how the kinetics,
in other words, the activation barrier, in the mechanism of
reactions or for preparative synthesis to produce the target
compounds.
Typically, multiphase catalytic systems, which play a
predominant role in industrial processes, are difficult to
investigate because the interaction of the substrate with the
catalyst is controlled by the mass transfer between the
different phases, and therefore the apparent reaction rate is
composed of the inherent reaction rate and diffusion rates. To
reduce this effect, the interfacial area must be increased.
Microstructured reaction systems intrinsically have a high
specific interfacial area per volume, only dependent on the
radius of the reaction channels; that is, for capillaries with
inner diameters between 250 and 100 mm the specific
interfacial area per volume ranges from 16000 to
[
1]
catalysis is controlled by the structural parameters. To
identify rate-determining elementary steps and to develop
models, comprehensive experimental kinetic data of a broad
variety of substrates are necessary. Microfluidic devices
integrating chemical synthesis and analysis on the same
[
2–5]
chip
are one promising approach for parallelized high-
throughput (ht) kinetic measurements of catalysts with
[
6]
minute material consumption. A unique technique for
studying configurational changes in chiral compounds is
2
À3
40000 m m . Microfluidic systems are currently revolutio-
[
7]
[9–13]
dynamic chromatography combining molecular intercon-
version and analysis under precisely controlled conditions.
nizing chemical synthesis,
be more easily controlled, low operation volumes minimize
because physical processes can
[
8]
[14]
Recently, we reported a unified equation to directly access
reaction rate constants of first-order reactions of such experi-
ments. However, commonly used (micro)reactors, in which
the reaction, separation, and quantification of conversion are
performed consecutively, are limited to the study of single
reactions, because competing reactions lead to undefinable
reaction kinetics. Here we show for the hydrogenation over
highly active Pd nanoparticles and the ring-closing metathesis
over the Grubbs second generation catalyst that the synchro-
nous combination of catalysis and separation makes it
possible to efficiently perform ht reaction rate measurements
reagent consumption, and detection is integrated.
ever, there are still many challenges,
How-
[
15]
for example, the
control of mixing, because diffusion rates contribute to the
apparent reaction rates, incompatibility with standard ana-
lytical instruments, and interfacing with a mass spectrometer,
to mention only a few.
Our strategy unites synthesis and analysis by combining
catalytic activity and separation selectivity in the polymeric
stationary phase of a chromatographic separation capillary.
This concept makes it possible to control the selectivity and
contact time of the analytes with the catalyst. Whereas
reactions during chromatographic separations were often
(
147 reactions per hour) of substrate libraries. The catalytic
[
16–18]
systems for these multiphase reactions (gas–liquid–solid)
were prepared by embedding the catalysts in polysiloxanes,
which serve as both solvent and selective stationary separa-
tion phase. Furthermore this system can be used for cascade
considered to be serendipitous
appearing as overlapping peak profiles with plateau forma-
tions,
or interfering side effects,
[
19,20]
we have used this information to investigate the
reaction kinetics of a broad variety of substrates.
To put this strategy into practice we focused on hydro-
genations over Pd nanoparticles embedded in an inert
polydimethylsiloxane matrix without any interfering protect-
ing shell (for example, tetraalkylammonium salts as surfac-
[
*] Dr. O. Trapp, S. K. Weber, S. Bauch, W. Hofstadt
Department of Heterogeneous Catalysis
Max-Planck-Institut für Kohlenforschung
Kaiser-Wilhelm-Platz 1, 45470 Mülheim an der Ruhr (Germany)
Fax: (+49)208-306-2995
E-mail: trapp@mpi-muelheim.mpg.de
Homepage: http://www.mpi-muelheim.mpg.de/kofo/institut/
arbeitsbereiche/trapp/trapp_e.html
[
21]
tants). We prepared a methylvinylsiloxane–dimethylsilox-
ane copolymer (4.5% Si(O)(CH )(CH=CH ) groups) to
3
2
coordinate Pd ions to the vinyl groups. Hydridomethylsilox-
ane–dimethylsiloxane copolymer (25.7% Si(O)(CH )H
3
2+
0
groups) was added to this mixture to reduce Pd to Pd and
to crosslink with the other copolymer in a Pd-catalyzed
[
**] Generous financial support by the MPI für Kohlenforschung, the
DFG (Emmy Noether program (TR 542/3), and the Cusanuswerk
hydrosilylation reaction to form
Figure 1).
a stabilizing matrix
(
fellowship to S.K.W.) is gratefully acknowledged. We thank H. J.
(
Bongard, A. Dreier, B. Spliethoff, and Dr. B. Tesche for TEM, SEM,
and EDX measurements.
1
H NMR measurements confirmed that the proportion of
Si(O)(CH )H groups decreased to 2.6% (starting from 16.1%
in the mixture of the two polymers). The Pd nanoparticles
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
3
Angew. Chem. Int. Ed. 2007, 46, 7307 –7310
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7307

Communications
were quantified by flame ionization detection (FID) and
identified by MS. Substrate libraries consisting of 22 unsatu-
rated compounds (alkenes, alkynes, aromatic hydrocarbons)
and functionalized compounds (nitro compounds, aldehydes,
ketones) to investigate the chemoselectivity were simulta-
neously injected onto this column configuration at different
temperatures and gas flows to vary the reaction time and to
obtain temperature-dependent kinetic data. We observed
extraordinary fast hydrogenations leading to complete con-
version of the substrates even for Pd nanoparticle capillaries
only 5 cm in length and at low reaction temperatures (608C).
Therefore, to achieve incomplete conversions for the kinetic
measurements all experiments were performed with a 2-cm-
long capillary (Figure 2, Table 1) and with reaction times
between 20 ms and 1 s. For some compounds, for example,
trans-cinnamaldehyde, methylbicyclo[2.2.1]hept-5-ene-2-car-
boxylate, dimethylacetylenedicarboxylate, and styrene, we
obtained 100% conversion for Pd nanoparticles with the
highest activity even under these mild conditions.
Figure 1. Preparation of highly active Pd nanoparticles embedded in a
polysiloxane matrix.
were spherical and crystalline with a narrow size distribution
of 3.2 nm Æ 0.7 nm (determined by TEM; see the Supporting
Information). Surprisingly, we observed that the morphology
of the Pd nanoparticles strongly depends on the ratio of the
two polysiloxanes. TEM characterization revealed that treat-
ment with hydrogen at elevated temperatures (180–2008C)
led to larger amorphous nanoparticles (3.6 Æ 1.6 nm; see the
Supporting Information). The resulting viscous brownish gray
polymers were then applied as a thin film onto the inner
surface of fused-silica capillaries and heated to 1808C at a rate
From temperature- and flow-dependent conversion meas-
urements for each compound we obtained data sets which
were put into kinetic models based on a Langmuir–Hinshel-
[
22]
wood mechanism to determine reaction rate constants k
°
and activation parameters (Gibbs activation energy DG ,
°
°
activation enthalpy DH , and activation entropy DS ). We
achieved a very good agreement when we applied first-order
reaction kinetics with respect to the substrates.
À1
of 0.5 Kmin under hydrogen flow to provide a permanently
The high activity of the Pd nanoparticles was corrobo-
rated by the activation parameters, which show low activation
bonded polymer. The SEM image of the capillary and the
SEM and EDX patterning of Si and Pd show the fused-silica
microcapillary with the Pd/polysiloxane surface coating
°
°
enthalpies DH and negative activation entropies DS ,
corresponding to a restraint transition state. It is important
to note that our experimental setup has the advantage of
precise temperature control and that diffusion processes can
be quantified and experimentally controlled by the polysil-
oxane used. Because we could inject simultaneously large
substrate libraries (22 different substrates, see the Supporting
Information), we achieved experimentally a throughput of
5880 reactions in 40 h. This corresponds to the determination
of a complete set of activation parameters (see the Supporting
Information) in less than 2 h. This efficiency allowed us to
screen 40 systematically varied Pd nanoparticle samples and
(
1
Figure 2). The Pd loading was extremely low, only 0.73 ”
À12
0
mol per cm of capillary (corresponding to approxi-
mately 27 billion Pd nanoparticles per cm with 1600 to
1
700 Pd atoms per particle).
On-column catalysis was performed by coupling this Pd
nanoparticle microcapillary between a pre-separation capil-
lary (1 m) and a separation column (25 m). The purpose of the
pre-separation column is to thermally equilibrate the reac-
tants and to spatially separate the substrates of the injected
compound library, which enables ht kinetic investigations
because of the absence of competing reactions. Hydrogen gas
was used as the reactive carrier gas. Substrates and products
°
to correlate the activation energy DG with the structure of
the particles. The highest activity was observed for amor-
Figure 2. On-column hydrogenation over highly active Pd nanoparticles. Capillaries only 2 cm long coated with Pd nanoparticles stabilized in a
polysiloxane matrix were used as the reactor. SEM and Si/Pd EDX (16 h) measurements show the coating of the fused silica microcapillaries
(
i.d. 250 mm, film thickness 0.25 mm).
7308
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 7307 –7310

Angewandte
Chemie
Table 1: Selected results of the on-column hydrogenations over highly active Pd nanoparticles.
were quantified and identified by
[
a]
°[b]
°
°
[c]
FID and MS detection (Figure 3).
We injected a substrate library
Substrate
Product
k
[
DG
DH
[kJmol
DS
r
À1
À1
À1
À1
À1
s
]
[kJmol
]
]
[JK mol
]
(s.d.)
of 12 different compounds simulta-
neously for ring-closing metathesis
onto the catalytically active sepa-
ration column. In these experi-
ments we obtained elution profiles
characterized by a plateau forma-
tion between the substrate and
product (see Figure 3 and the Sup-
porting Information). These elu-
tion profiles were analyzed by the
unified equation (see the Sup-
porting Information) to obtain
reaction rate constants (Table 2).
The dissociation and chromato-
graphic removal of the tricyclohex-
3
0.1
À126
Æ3
0.997
(0.057)
1
4
94
67.8
Æ0.5
25.2
À152
0.996
(0.052)
2
6
70.4
71.4
Æ0.5
Æ4
27.2
À148
Æ6
0.996
(0.046)
3
4
Æ0.7
56.0
À94
Æ2
0.998
(0.020)
82.3
73.4
75.3
Æ1.0
[
8]
37.5
À121
Æ3
0.999
(0.025)
4
2
4
3
Æ0.6
38.3
À124
Æ7
0.985
(0.116)
Æ1.5
ylphosphine ligand (PCy ) could
3
°
[
a] Reaction rate constant at 1208C. [b] Gibbs activation energy DG at 258C. [c] Correlation factor r and
not be detected; this can be
explained by the fact that the
phosphine ligand dissolves in the
polysiloxane and can stabilize the
residual standard deviation of the linear regression of the Eyring plot.
[
25]
phous nanoparticles stabilized by a highly crosslinked poly-
siloxane made from a stoichiometric mixture of the two
polymers. By continuous injection of a single substrate, for
example, cyclohex-2-enone, onto these columns we could also
complexafter the reaction.
It is remarkable that the
catalyst was stable over a wide temperature range (up to
1508C) without any detectable degradation of the catalytic
activity or leaching.
À1
preparatively hydrogenate in amounts of about 20 mgh (see
Activation parameters were obtained from temperature-
dependent measurements: for example, for N,N-diallyltri-
the Supporting Information).
°
À1
°
To demonstrate the generality of our approach and the
broad range of reaction kinetics that can be investigated, we
fluoroacetamide DG (298 K) = 84.1 kJmol , DH = (15.5 Æ
À1 ° À1 À1
0.9) kJmol , DS = (À230 Æ 8) JK mol . These parame-
[
23,24]
[26]
focused on ring-closing metathesis
(RCM) catalyzed by
ters corroborate recently reported theoretical calculations
the Grubbs 2nd generation catalyst. For the preparation of
the catalytically active microcolumns we dissolved the
Grubbs 2nd generation catalyst in dimethylpolysiloxane
performed in an effort to explain the high activity of the
Grubbs 2nd generation catalyst. Also in these experiments by
the simultaneous injection of an substrate library we achieved
an extraordinary high throughput in the determination of
reaction rate constants (36 rate constants per hour). The
polarity of the dimethylpolysiloxane used here corresponds to
a nonpolar solvent. Tuning the polarity of stationary phases
with functional groups should make it possible to investigate
solvent effects in continuous polarity steps.
(
GESE30) and coated microcapillaries (10 m) with a film
thickness of 1 mm under strict exclusion of oxygen. The
À9
catalyst loading was only 1.6 mg (1.9 ” 10 mol) per meter of
capillary. The on-column catalysis was performed by coupling
this column with a pre-separation column 1 m in length.
Helium was used as the inert carrier gas. Eluted compounds
Figure 3. On-column metathesis over the Grubbs 2nd generation catalyst. In these experiments catalytic activity and separation selectivity is
combined in a single 10-m-long capillary by dissolving the catalyst in the stationary separation phase. The elution profiles obtained were
characterized by a pronounced conversion profile from the subsrate to the product. Kinetic analysis was performed directly with the unified
equation from Ref. [8].
Angew. Chem. Int. Ed. 2007, 46, 7307 –7310
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 7309

Communications
Table 2: Selected results of the on-column ring-closing metathesis.
approach the advantage of heterogeneous catalysis—the
ready separation of catalyst and reaction product, which is
normally a problem for nanopartical and homogeneous
catalysts—is also conserved.
[
a]
[b]
°
Substrate
Product
T
C
k
DG
À3 À1
À1
[8C]
[%]
[10
s
]
[kJmol
]
These results impressively illustrate the (r)evolution of
the chemistꢀs toolkit from flasks and beakers to miniaturized
reactors with highly selective catalysts and inherently com-
bined separation techniques. These reactors mimic biological
systems in which reactions take place efficiently at mini-
aturized interfaces in continuous-flow reactors. Therefore, the
development of chemical microplants for the production of
fine chemicals can be envisaged as a natural step in terms of
energy efficiency and environmental impact.
1
10.0 39.0 2.2
50.0 97.3 3.4
114.1
1
1
124.9
5
0.0 62.5 8.6
89.8
20.0 59.5 7.7
113.1
Received: March 26, 2007
Revised: May 21, 2007
Published online: July 31, 2007
9
0.0 51.0 4.9
105.6
Keywords: homogeneous catalysis · hydrogenation ·
metathesis · multiphase reactions · nanoparticles
[
a] Conversion C. [b] Reaction rate constant k. Conditions: 10-m-long
.
microcapillary (i.d. 250 mm, film thickness 1 mm), He as the inert carrier
gas.
[
[
1] C. T. Campbell, Nature 2004, 432, 282 – 283.
2] J. Knight, Nature 2002, 418, 474 – 475.
Because these catalytically active separation capillaries
are easy to prepare and handle we used them in a modular
design for two-step cascade reactions. We coupled a 80-cm-
long column, coated with the dissolved Grubbs 2nd gener-
ation catalyst, and a 10-cm-long Pd nanoparticle column,
followed by a separation column for product analysis (cf.
Figure 4). We used hydrogen as a carrier gas, but the
experiment could also be performed with helium in the first
column section and adding hydrogen in the second section.
We demonstrated that the metathesis of N,N-diallyltrifluoro-
acetamide followed by on-column hydrogenation is possible
in less than 6 min with an overall yield of 49% (see Figure 4
and the Supporting Information).
The strategies outlined here can be generally applied to
other catalytic processes, and we found that they can be
utilized in the comprehensive kinetic characterization of
catalysts and materials. Furthermore it can be envisioned that
catalytic capillaries could be useful for selective transforma-
tions in analytical applications and for structure elucidation.
Moreover, for a preparative scale-up only a stack of reactor
capillaries is necessary to increase the productivity. In this
[3] D. Janasek, J. Franzke, A. Manz, Nature 2006, 442, 374 – 380.
[4] D. Belder, M. Ludwig, L.-W. Wang, M. T. Reetz, Angew. Chem.
2006, 118, 2523 – 2526; Angew. Chem. Int. Ed. 2006, 45, 2463 –
2466.
[
[
[
[
5] S. J. Haswell, Nature 2006, 441, 705.
6] S. J. Haswell, P. Watts, Green Chem. 2003, 5, 240 – 249.
7] O. Trapp, G. Schoetz, V. Schurig, Chirality 2001, 13, 403 – 414.
8] O. Trapp, Anal. Chem. 2006, 78, 189 – 198.
[9] C. de Bellefon, N. Tanchoux, S. Caravieilhes, P. Grenouillet, V.
Hessel, Angew. Chem. 2000, 112, 3584 – 3587; Angew. Chem. Int.
Ed. 2000, 39, 3442 – 3445.
[
10] J. Kobayashi, Y. Mori, K. Okamoto, R. Akiyama, M. Ueno, T.
Kitamori, S. Kobayashi, Science 2004, 304, 1305 – 1308.
11] P. Watts, S. J. Haswell, Chem. Soc. Rev. 2005, 34, 235 – 246.
12] Y. Uozumi, Y. M. A. Yamada, T. Beppu, N. Fukuyama, M. Ueno,
T. Kitamori, J. Am. Chem. Soc. 2006, 128, 15994 – 15995.
[
[
[13] Micro Instrumentation (Eds.: M. V. Koch, K. M. Van den
Bussche, R. W. Chrisman), Wiley-VCH, Weinheim, 2007.
[14] A. J. deMello, Nature 2006, 442, 394 – 402.
[15] G. M. Whitesides, Nature 2006, 442, 368 – 373.
[16] E. Gil-Av, Y. Herzberg-Minzly, Chem. Commun. 1961, 316.
[17] R. Thede, E. Below, D. Haberland, S. H. Langer, Chromatogra-
phia 1997, 45, 149 – 154.
[
[
[
[
[
[
[
18] N. A. Katsanos, R. Thede, F. Roubani-Kalantzopou-
lou, J. Chromatogr. A 1998, 795, 133 – 184.
19] T. D. Vu, A. Seidel-Morgenstern, S. Grüner, A.
Kienle, Ind. Eng. Chem. Res. 2005, 44, 9565 – 9574.
20] D. W. Bassett, H. W. Habgood, J. Phys. Chem. 1960,
64, 769 – 773.
21] M. T. Reetz, W. Helbig, S. A. Quaiser, U. Stimming,
N. Breuer, R. Vogel, Science 1995, 267, 367 – 369.
22] A. Schmidt, R. Schomꢁcker, Ind. Eng. Chem. Res.
2007, 46, 1677 – 1681.
23] R. H. Grubbs, Angew. Chem. 2006, 118, 3845 – 3850;
Angew. Chem. Int. Ed. 2006, 45, 3760 – 3765.
24] A. Fürstner, Angew. Chem. 2000, 112, 3140 – 3172;
Angew. Chem. Int. Ed. 2000, 39, 3012 – 3043.
Figure 4. Modular design for a two-step on-column cascade reaction. N-trifluoro-
acetylpyrrolidine was synthesized within 6 min by coupling a 80 cm long meta-
[25] M. S. Sanford, J. A. Love, R. H. Grubbs, J. Am.
Chem. Soc. 2001, 123, 6543 – 6554.
thesis column and a 10 cm long Pd nanoparticle hydrogenation column. H was
used as the reactive carrier gas.
[26] B. F. Straub, Angew. Chem. 2005, 117, 6129 – 6132;
Angew. Chem. Int. Ed. 2005, 44, 5974 – 5978.
2
7310
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 7307 –7310