Capillary-based, serial-loading, parallel microreactor for catalyst screening

Article abstract of DOI:10.1021/ac051844+

Soluble metal-ligand complexes are useful as catalysts for many organic reactions. The large number of metals and ligands available suggests a combinatorial approach to catalyst discovery. Carrying out reactions in very small (microliter) volumes in capil

Full text of DOI:10.1021/ac051844+

Anal. Chem. 2006, 78, 1972-1979
Capillary-Based, Serial-Loading, Parallel
Microreactor for Catalyst Screening
Guoyue Shi, Feng Hong, Quansheng Liang, Hui Fang, Scott Nelson, and Stephen G. Weber*
Department of Chemistry, University of Pittsburgh, 219 Parkman Avenue, Pittsburgh, Pennsylvania 15260
Soluble metal-ligand complexes are useful as catalysts
for many organic reactions. The large number of metals
and ligands available suggests a combinatorial approach
to catalyst discovery. Carrying out reactions in very small
convenient pumping mechanism for fluidic manipulation, with
minimal hydrodynamic dispersion.5
,8,10-13
In the electrokinetically
driven system, electrodes are placed in the appropriate reservoirs
to which specific voltage sequences can be delivered under
automated computer control. Hydrodynamic pumping also exploits
(
microliter) volumes in capillaries has many advantages
in this regard, including material conservation, isolation
from the atmosphere, and ease of transport of species by
using pressure-induced flow. We have developed a capil-
lary reactor in which separate zones of catalyst and
reactants are combined and react. Zones are loaded
serially into the capillary reactor from an autosampler,
they react in parallel in the capillary reactor (at elevated
temperature) and are ejected serially and under computer
control for analysis by online GC. Offline analysis following
sample collection is also possible. The Stille cross-
coupling reaction has been the focus of our recent activity.
Known palladium-based precatalysts and phosphine or
arsine ligands were screened to validate the approach
taken here. The results largely agree with results obtained
by traditional organic synthesis, validating the method.
The throughput of the nonoptimized system is over two
conventional or microscale pumps to maneuver solutions around
the channel network.2
,14-17
Among the many materials, glass and
silica are the most popular materials to construct the microreac-
tors.
While most applications in this area have been directed toward
analysis, some have been directed toward synthesis. Several
1
8-28
excellent reviews written from different perspectives exist.
Researchers have investigated fundamental studies on mass
1
0,11,29,30
transport and fluid flow
as well as particular synthetic
applications,1
3,15,31-34
including the controlled generation of reactive
intermediates.3 In a novel application, Comer and Organ have
5,36
(10) Fletcher, P. D. I.; Haswell, S. J.; Zhang, X. Lab Chip 2001, 1, 115-121.
(11) Fletcher, P. D. I.; Haswell, S. J.; Zhang, X. Lab Chip 2002, 2, 102-112.
(12) Kohlheyer, D.; Besselink, G. A. J.; Lammertink, R. G. H.; Schlautmann, S.;
Unnikrishnan, S.; Schasfoort, R. B. M. Microfluidics Nanofluidics 2005, 1,
242-248.
(13) Skelton, V.; Haswell, S. J.; Styring, P.; Warrington, B.; Wong, S. Micro Total
Analysis Systems 2001, Proceedings mTAS 2001 Symposium, 5th, Monterey,
CA, October 21-25, 2001; pp 589-590.
5
-h reactions/h. For example, 40 5-h reactions examining
the effect of catalyst loading were performed in 2 9-h runs
requiring a total of 2 h of operator time.
(14) Fernandez-Suarez, M.; Wong, S. Y. F.; Warrington, B. H. Lab Chip 2002,
2, 170-174.
(
(
15) Haswell, S. J.; O’Sullivan, B.; Styring, P. Lab Chip 2001, 1, 164-166.
16) Kashid, M. N.; Gerlach, I.; Goetz, S.; Franzke, J.; Acker, J. F.; Platte, F.;
Agar, D. W.; Turek, S. Ind. Eng. Chem. Res. 2005, 44, 5003-5010.
In recent years, miniaturized total analytical systems (µTAS)
1-9
(17) Benito-Lopez, F.; Verboom, W.; Kakuta, M.; Gardeniers, J. G. E.; Egberink,
R. J. M.; Oosterbroek, E. R.; van den Berg, A.; Reinhoudt, D. N. Chem.
Commun. (Cambridge, U. K.) 2005, 2857-2859.
have been an area of great activity. Though many µTAS systems
involve separations, there are also many that do not. The latter
µTASs consist of reagents, microchannels, and an online or offline
detection system. Electrokinetic or hydrodynamic pumping gener-
ally motivates solution flow. Electroosmotic flow offers a very
(18) Cullen, C. J.; Wootton, R. C. R.; de Mello, A. J. Curr. Opin. Drug Discovery
Dev. 2004, 7, 798-806.
(
(
19) Doku, G. N.; Verboom, W.; Reinhoudt, D. N.; van den Berg, A. Tetrahedron
005, 61, 2733-2742.
20) Haswell, S. J. Micro Total Analysis Systems 2001, Proceedings mTAS 2001
2
*
Corresponding author: (tel) +1(412)624-8520; (fax) +1(412)624-1668;
Symposium, 5th, Monterey, CA, October 21-25, 2001; pp 637-639.
(
e-mail) sweber@pitt.edu.
(21) Jas, G.; Kirschning, A. Chem.-Eur. J. 2003, 9, 5708-5723.
(22) Kikutani, Y.; Kitamori, T. Electrokinet. Phenom. 2004, 253-275.
(23) Roberge, D. M.; Ducry, L.; Bieler, N.; Cretton, P.; Zimmermann, B. Chem.
Eng. Technol. 2005, 28, 318-323.
(24) Schwalbe, T.; Autze, V.; Wille, G. Chimia 2002, 56, 636-646.
(25) Watts, P.; Haswell, S. J. Curr. Opin. Chem. Biol. 2003, 7, 380-387.
(26) Watts, P.; Haswell, S. J. Drug Discovery Today 2003, 8, 586-593.
(27) Watts, P.; Haswell, S. J. Chem. Eng. Technol. 2005, 28, 290-301.
(28) Zech, T.; Klein, J.; Schunk, S. A.; Johann, T.; Schueth, F.; Kleditzsch, S.;
Deutschmann, O. High-Throughput Anal. 2003, 491-523.
(29) Broadwell, I.; Fletcher, P. D. I.; Haswell, S. J.; McCreedy, T.; Zhang, X. Lab
Chip 2001, 1, 66-71.
(30) Benninger, R. K. P.; Hofmann, O.; McGinty, J.; Requejo-Isidro, J.; Munro,
I.; Neil, M. A. A.; de Mello, A. J.; French, P. M. W. Opt. Express 2005, 13,
6275-6285.
(31) Sands, M.; Haswell, S. J.; Kelly, S. M.; Skelton, V.; Morgan, D. O.; Styring,
P.; Warrington, B. Lab Chip 2001, 1, 64-65.
(
(
1) Kakuta, M.; Bessoth, F. G.; Manz, A. Chem. Rec. 2001, 1, 395-405.
2) Brivio, M.; Fokkens, R. H.; Verboom, W.; Reinhoudt, D. N.; Tas, N. R.;
Goedbloed, M.; van den Berg, A. Anal. Chem. 2002, 74, 3972-3976.
3) Culbertson, C. T.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1998, 70,
(
(
(
(
(
(
(
3
781-3789.
4) Hadd, A. G.; Raymond, D. E.; Halliwell, J. W.; Jacobson, S. C.; Ramsey, J.
M. Anal. Chem. 1997, 69, 3407-3412.
5) Jacobson, S. C.; McKnight, T. E.; Ramsey, J. M. Anal. Chem. 1999, 71,
4
455-4459.
6) Oleschuk, R. D.; Harrison, D. J. TrAC, Trends Anal. Chem. 2000, 19, 379-
88.
7) Colyer, C. L.; Tang, T.; Chiem, N.; Harrison, D. J. Electrophoresis 1997,
8, 1733-1741.
8) Salimi-Moosavi, H.; Tang, T.; Harrison, D. J. J. Am. Chem. Soc. 1997, 119,
716-8717.
3
1
8
9) Ramsey, J. M. Nat. Biotechnol. 1999, 17, 1061-1062.
1972 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006
10.1021/ac051844+ CCC: $33.50 © 2006 American Chemical Society
Published on Web 01/26/2006

mated a microreactor with microwave heating to give a 4-min
exposure of the reaction mixture to the radiation.37 In another
novel application, thermochromic liquid crystals reported on
is a hybrid between the simple combine-reagents-and-detect
approach common to most simple microfluidic platforms and the
far more instrumentally complex approach involving multiple
parallel flow streams. In the current instrument, reaction zones
are loaded serially, react in parallel, and are analyzed (or collected)
serially.
3
8
temperature in real time. Because of the low capacity of chip
devices, most reactions studied are relatively rapid. Nanoparticle
syntheses, which are difficult to control, are successfully carried
3
9-43
out in microreactors.
Microreactors have been applied to catalyst discovery.
In this work, we focus on the construction and evaluation of
this novel instrument. We have investigated catalysts for the Stille
44,45
4
6,47
Polymerization catalysts have been investigated
as have
reaction, which is one of the most important reactions leading to
48,49
15
53-55
catalysts for gas/liquid systems,
Kumada coupling, Baeyer-
the formation of new carbon-carbon bonds.
The reaction is
Villager oxidation, enamine formation,31 and expoxidation.51
In these microfluidic systems investigating organic reactions,
one reaction is underway at any time. Reagents and catalysts are
brought together and reactions occur in a continuous flow before
being detected. Parallel reactors are beginning to be developed
5
0
carried out in the presence of 1-2 mol % palladium catalyst. A
variety of palladium(II) or palladium(0) complexes with neutral
ligands were tested in the screening system we constructed. The
reaction products were analyzed by on-line gas chromatography
achieving quantitative information about catalyst activity. PdCl
2 3
(CH -
(see reviews cited above and ref 52).
CN) (2 mol %) + AsPh (6 mol %) is the best catalyst for the
2
3
We present here a powerful tool for high-throughput screening
Stille reaction in the group that we tested.
of slow reactions accelerated by homogeneous catalysts. Up to
0 reaction zones containing different catalysts can exist simul-
2
EXPERIMENTAL SECTION
taneously in the simple fused-silica capillary reactor. The zones
3 3
Chemicals and Materials. HPLC grade CH OH, THF, Bu -
are defined by natural hydrodynamic dispersion. The instrument
2
SnCHdCH , phenyl iodide (PhI), dodecane, and styrene were
purchased from Sigma (St.Louis, MO). Neutral red was from J.T.
Baker Chemical Co. (Phillipsburg, NJ). All the ligands including
(
(
(
(
32) Watts, P.; Wiles, C.; Haswell, S. J.; Pombo-Villar, E.; Styring, P. Chem.
Commun. (Cambridge, U. K.) 2001, 990-991.
33) Watts, P.; Wiles, C.; Haswell, S. J.; Pombo-Villar, E. Lab Chip 2002, 2, 141-
AsPh
purchased from Sigma. Pd
Pd[(C P] , Pd[(C P]
were purchased from Strem Chemicals (Newburyport, MA).
PdCl (CH CN) was prepared according to ref 56. Nitrogen, argon,
3
, PPh
3
, (2-furyl)
3
P, (4-FC
dba
, PdCl
6
H
4
)
3
P, and (4-ClC
(dba ) dibenzylidine acetone),
[(C P] , and Pd(OAC)
2
6 4 3
H ) P were also
1
44.
34) Wiles, C.; Watts, P.; Haswell, S. J.; Pombo-Villar, E. Tetrahedron 2003, 59,
0173-10179.
35) Fortt, R.; Wootton, R. C. R.; de Mello, A. J. Micro Total Analysis Systems
002, Proceedings of the mTAS 2002 Symposium, 6th, Nara, Japan, November
-7, 2002; Vol. 2, pp 850-852.
36) Fortt, R.; Wootton, R. C. R.; de Mello, A. J. Org. Process Res. Dev. 2003, 7,
62-768.
2
3
6
H
5
)
3
4
6
H
5
)
3
4
2
6
H
5
)
3
2
1
2
3
2
3
2
and compressed air were obtained from Valley National Gases
Inc. (Washington, PA)
(
7
(
(
(
37) Comer, E.; Organ, M. G. J. Am. Chem. Soc. 2005, 127, 8160-8167.
38) Iles, A.; Fortt, R.; de Mello, A. J. Lab Chip 2005, 5, 540-544.
39) Chan, E. M.; Alivisatos, A. P.; Mathies, R. A. J. Am. Chem. Soc. 2005, 127,
Instrumentation. Syringe pumps were purchased from Har-
vard Apparatus Inc. (Holliston, MA). The Waters M-45 pump was
from Waters Corp. (Milford, MA). The HP 1050 autosampler were
purchased from Agilent (Palo Alto, CA). VICI six-port injector
1
3854-13861.
(
40) Yen, B. K. H.; Gunther, A.; Schmidt, M. A.; Jensen, K. F.; Bawendi, M. G.
Angew. Chem., Int. Ed. 2005, 44, 5447-5451.
(Model E60) and VICI 10-port valve (model EPCA-CE) were
(
(
41) Wagner, J.; Koehler, J. M. Nano Lett. 2005, 5, 685-691.
42) Edel, J. B.; Fortt, R.; de Mello, J. C.; de Mello, A. J. Micro Total Analysis
Systems 2002, Proceedings of the mTAS 2002 Symposium, 6th, Nara, Japan,
November 3-7, 2002; Vol. 2, pp 772-774.
43) Edel, J. B.; Krishnadasan, S.; Cao-Romero, J. T.; Vilar-Compte, R.; de Mello,
J. C.; de Mello, A. J. Transducers ′03, International Conference on Solid-
State Sensors, Actuators and Microsystems, Digest of Technical Papers, 12th,
Boston, MA, June 8-12, 2003; Vol. 2, pp 1730-1733.
44) Keil, F. J. Chem. Eng. Sci. 2004, 59, 5473-5478.
45) Snyder, D. A.; Noti, C.; Seeberger, P. H.; Schael, F.; Bieber, T.; Rimmel, G.;
Ehrfeld, W. Helv. Chim. Acta 2005, 88, 1-9.
46) Potyrailo, R. A.; Lemmon, J. P.; Leib, T. K. Anal. Chem. 2003, 75, 4676-
purchased from Valco Instruments Co, Inc. (Houston, TX). The
Focus GC was purchased from Thermo-Electron. It has a single
column (RTX-5, 7 m × 0.32 mm (0.25-µm thick phase)) and
detector (FID). A USB 2000 optical fiber UV-visible absorbance
detector was purchased from Ocean Optics. Inc. (Dunedin, FL).
A pump (M6) for injection into the GC was purchased from
Intelligent Motion systems, Inc. The heater for the organic
reactions was constructed locally. The temperature controller for
the heater was from Minco Products Inc. (Minneapolis, MN). The
fused-silica capillary with 75-µm i.d., 360-µm o.d. that was used
as the microreactor was purchased from Polymicro Technologies,
L.L.C (Phoenix, AZ).
(
(
(
(
4
681.
(
47) Boussie, T. R.; Diamond, G. M.; Goh, C.; Hall, K. A.; LaPointe, A. M.; Leclerc,
M.; Lund, C.; Murphy, V.; Shoemaker, J. A. W.; Tracht, U.; Turner, H.;
Zhang, J.; Uno, T.; Rosen, R. K.; Stevens, J. C. J. Am. Chem. Soc. 2003,
1
25, 4306-4317.
GC Analysis. The initial temperature of the oven was 85 °C
and held for 0.8 min. The temperature was increased from 85 to
(
48) De Bellefon, C.; Tanchoux, N.; Caravieilhes, S.; Grenoullet, P.; Hessel, V.
Angew. Chem., Int. Ed. 2000, 39, 3442-3445.
(
49) Pennemann, H.; Hessel, V.; Kost, H. J.; Loewe, H.; de Bellefon, C.; Pestre,
N.; Lamouille, T.; Grenouillet, P. Better Processes for Bigger Profits, Interna-
tional Conference on Process Intensification for the Chemical Industry, 5th,
Maastricht, The Netherlands, October 13-15, 2003; pp 137-147.
50) Mikami, K.; Islam, M. N.; Yamanaka, M.; Itoh, Y.; Shinoda, M.; Kudo, K.
Tetrahedron Lett. 2004, 45, 3681-3683.
2
00 °C at 100 °C/min. Two minutes was allowed for cooling.
Typically, flowsplitting was used to inject 10% of the 1.0-µL loop
contents.
Construction of the Screening System for Catalyst Librar-
ies. The designed screening setup is shown in Scheme 1. The
(
(
(
51) Wan, Y. S. S.; Chau, J. L. H.; Yeung, K. L.; Gavriilidis, A. J. Catal. 2004,
2
23, 241-249.
52) Swenson, R. E.; DeWitt, S. H.; Lin, J.; Hamilton, T.; Emerich, C. Innovation
and Perspectives in Solid Phase Synthesis & Combinatorial Libraries: Peptides,
Proteins and Nucleic Acids- -Small Molecule Organic Chemistry Diversity,
Collected Papers, International Symposium, 6th, York, U.K., Aug. 31-Sept.
(53) Farina, V. Comprehensive Organometallic Chemistry, 2 ed.; Pergamon:
Oxford, 1995.
(54) Stille, J. K. Angew. Chem., Int. Ed. Engl. 1986, 25, 508.
(55) Farina, V.; Krishnan, B. J. Am. Chem. Soc. 1991, 113, 9585-9595.
(56) Mitchell, T. N. Synthesis 1992, 803.
4
, 1999, 2001; pp 113-118.
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006 1973

Scheme 1. Schematic Diagram of the Screening
by online GC. Where applicable, we prefer the latter approach.
The online injection process is triggered by the absorbance
detector. A 10-port valve enables the loading of one loop while
the contents of the other loop are chromatographed. The contents
of the loop are pushed into the GC with a low-pressure, refillable
syringe pump (M6).
a
System
To check the screening system, confirm how many zones we
can put in the capillary microreactor, and define optimum
experimental conditions, the system created zones of a dye, 1.0
3
mM neutral red in methanol. CH OH was pumped by SP2 in this
case. The wavelength of the optical fiber absorbance detector was
fixed at 530 nm. Flow rates of SP1 and SP2 were optimized to
maximize reproducibility and minimize band spreading. Param-
eters taken into consideration in the timing sequence are the
volumes of the microinjector’s loop (1 µL) and the capillary
microreactor (i.d. 75 µm, length, 6.7 m, the volume of the
microreactor is ∼30 µL), the microinjector loop loading time, and
the microinjector injection time. To use the same flow rate for
SP2 in both the loading phase and the detection phase, the GC
run time and the loading/injection times were made the same
(∼5 min).
a
The loading section consists of a syringe pump (SP1) for the
reagents and another for the driving flow (SP2). Catalysts are
combined with reagents at equal flow rates in the loop of a six-
port injector by an autosampler. Under computer control, when
the autosampler has completed its cycle, the loop’s contents are
pushed into the reactor. The reactor is heated. The detection
section consists of a UV-visible absorbance detector for finding
the reaction zones. Under computer control, the zones are
alternately pushed into one loop in the 2-loop, 10-port microin-
jector by SP2 and then injected into the capillary GC by SP3.
The injection by SP3 initiates the GC program for quantative
analysis of the product of the reaction.
Screening Experiment. The Stille reaction was carried out
system includes three sections: loading, reaction, and GC analysis.
Fused-silica capillary joined by homemade low-volume con-
nectors
in anhydrous THF at 50 °C. Pd
PdCl [(C P] , Pd(OAC) , and PdCl
the precatalysts and AsPh , PPh
ClC
2
dba
3
, Pd[(C
6
H
5
)
3
P]
CN)
P, (4-FC
4
, Pd[(C
were chosen as
P, and (4-
6 5 3 4
H ) P] ,
2
H )
6 5 3
2
2
2
(CH
3
2
5
7-60
comprise the fluidic part of the microreactor. Each
3
3
, (2-furyl)
3
6 4 3
H )
of the three sections carries out that function as follows. The “load”
section creates a small zone of solvent containing catalysts and
reagents and moves the zone into the reactor. Several zones are
moved serially into the reactor. The reaction occurs at controlled
temperature in the reactor. After reacting in parallel in the capillary
reactor (at elevated temperature) for a specific time, the samples
are ejected serially and analyzed by online GC under the control
of a computer. Each of these three stages is described in more
detail below.
Injection. Samples of catalysts loaded into sampling vials are
placed onto an autosampler tray. The autosampler samples 20 µL
from the vial and pumps the sample zone out. This sample zone
passes through the loop of a microinjector. A loop volume of either
6
H
)
4 3
P were selected as the ligands. Various mole equivalents
of palladium-based precatalysts and ligands were mixed in 6 mL
of THF to form the catalyst for the Stille reaction. A 100-µL aliquot
of each catalyst solution was placed in a 2-mL glass autosampler
vial. The catalyst vials were set on the tray of the autosampler.
3 2
Reagents PhI (75 µL) and Bu SnCHdCH (250 µL) were mixed
in 3 mL containing 24 µL of dodecane as the GC internal standard.
The reactants were loaded in a syringe and driven by SP1. SP2
was turned off, and the flow was stopped after all the samples
were loaded into the capillary microreactor. The heater was turned
on, and the Stille reaction was carried out in the microreactor for
varying times at 50 °C. After reaction, SP2 was turned on and the
reaction zones were analyzed as described above.
Calibration Curve and the Yields of the Stille Reaction. A
calibration curve was prepared from standard solutions of styrene
before every run. A stock solution containing 77 µL of styrene,
1
1
µL or 750 nL is used. The flow rate of the autosampler pump is
5 µL/min. The reagents are pumped with a syringe pump 1 (SP1)
and are combined with the catalyst at an equal flow rate. Thus,
the loop contains both catalyst and reagents. The loop contents
are injected under computer control into the reactor. The contents
of the loop are pushed with THF from a syringe pump (SP2)
operating at typically 250 nL/min. Combining the catalyst and
reagents before the loop creates distinct zones containing reagents
and catalysts with pure solvent from SP2 between them.
Reaction. The reactor consists of a 75-µm-i.d., 6.7-m-long piece
of fused-silica tubing. When all the zones are loaded, the
temperature is increased to the reaction temperature and the flow
from SP2 is stopped for a defined time.
2
4 µL of docecane, and 3.224 mL of THF was diluted in THF to
obtain standards. Reaction yields are based on styrene to dodecane
peak area ratios.
RESULTS AND DISCUSSION
Reactor Components. Injection. We have employed a stan-
dard HPLC autoinjector for sample management. This apparatus,
although specified to have the capability to inject microliter
volumes, normally performs those injections into a system in
which a 50- or 100-µL injection would be acceptable. Further, low
carryover is an important specification. Thus, the contents of the
autosampler syringe in normal operating conditions are essentially
flushed into an HPLC. The shape of the concentration-time profile
of a solute emanating from the autosampler can be determined
by using only the autosampler and the absorbance detector with
the solute neutral red. The concentration-time profile of the
autosampler operating normally does not have a plateau, even
when injecting 60 µL of neutral red solution. Without a plateau, it
Detection. Detection can be done with the Ocean Optics
absorbance detector, by offline GC/MS on collected fractions, or
(
57) Beisler, A. T.; Sahlin, E.; Schaefer, K. E.; Weber, S. G. Anal. Chem. 2004,
6, 639-645.
58) Sahlin, E.; ter Halle, A.; Schaefer, K.; Horn, J.; Then, M.; Weber, S. G. Anal.
7
(
Chem. 2003, 75, 1031-1036.
(
(
59) Sahlin, E.; Weber, S. G. J. Chromatogr., A 2002, 972, 283-287.
60) Sahlin, E.; Beisler, A. T.; Woltman, S. J.; Weber, S. G. Anal. Chem. 2002,
7
4, 4566-4569.
1974 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

Figure 1. Absorbance response of 1 mM neutral red at 530 nm. Loop volume is 1.0 µL. (A) SP1, 15µL/min; SP2, 1.0 µL/min. (B) SP1,
5µL/min; SP2, 0.5 µL/min. (C) SP1, 15µL/min; SP2, 0.25 µL/min.
1
is difficult to get reproducible quantities into the loop injector.
Consequently, we altered the functioning of the autosampler. After
the autosampler syringe aspirates the sample, the sample is not
flushed out by the autosampler pump, but by the syringe itselfs
the plunger of the syringe goes down, ejecting the sample. This
can be achieved by changing the autosampler working program.
With this modification, a 20-µL sample is enough to create a
significant plateau in the concentration-time profile. With suitable
timing, a loop microinjector can be reproducibly loaded.
are the increased pressure required to move liquid through the
system (for longer and smaller diameter reactors) and the
increased time required to load the larger number of samples (for
longer reactors).
Detection. Online detection is preferred over offline. Both have
been used. In early investigations, we collected reaction zones
for offline GC/MS in order to positively identify the product,
styrene. Makeup flow is required for collection in vials. None of
these preliminary data are reported here. For online detection,
we automated the injection of reaction zones into the capillary
GC column. To accomplish this, a 10-port injector with dual
injection loops and a refillable syringe pump are required.
Reactor. The reactor is simply a fused-silica capillary under
temperature control. The reactor length and inside diameter have
not been optimized. Briefly, smaller diameters allow higher fluid
6
1
velocities without appreciable band spreading. Longer reactors
can contain more zones. Countervailing effects of these changes
(61) Probstein, R. F. Physicochemical Hydrodynamics: An Introduction, 2nd ed.;
John Wiley & Sons: New York, 2003.
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006 1975

Figure 3. Chromatogram of standard solution (a) styrene, (b)
iodobenzene, and (c) dodecane (internal standard). Column: RTX-
5
, 7 m × 0.32 mm (0.25-µm thick phase). Initial: temperature, 85
°
°
C; for 0.8 min, column temperature increased at 100 °C/min to 200
C and held for 0.00 min.
empirically to ensure good separation. Because the volume of the
capillary microreactor is ∼30 µL, the time required for the first
zone to flow from the injector to the detector is about 30, 60, and
Figure 2. Absorbance responses of 1 mM neutral red in the capillary
microreactor for various residence times.
Scheme 2. Scheme of the GC Analyzera
1
20 min at 1.0, 0.5 and 0.25 µL/min, respectively. The time
required to see the last zone is about twice that. Figure 1 shows
that 15 (12, 9) samples can coexist in the microreactor at 0.25
µL/min (0.5, 1.0 µL/min). Apparently, the higher flow rate permits
more zones per time (9/1.0 ) 9 zones/h at 1.0 µL/min compared
to 15/4 ) 3.5 at 0.25 µL/min). However, in practice, there is a
lengthy reaction time. For the Stille reaction, it is typically 5 h or
longer. When this time is added to the time for eluting the last
peak, the number of zones per hour is actually a maximum at 0.5
µL/min. In considering that 5 h would be the shortest time
typically used, 0.25 µL/min was used as the flow rate for SP2.
In Figure 1, the first peak in each trace has a lower absorbance
than the others. This appears not to be an inherent aspect of the
instrument, as we never see such an effect when we carry out
organic reactions. We suppose that there is some adsorptive loss
of the dye at the beginning of each experiment. While this does
not detract from the usefulness of the dye for simple optimization,
it is a warning that adsorption could potentially be an issue in
studies of chemical reactions.
a
Ten-port valve is used here as a double-loop injector. The
injector’s output is connected to GC’s injection zone by a short
capillary. A computer, equipped with locally developed software,
monitors the signal from the UV detector. If there is a new zone
passing by, the computer will send a command to switch the
injector and then start M-6 pump to pump the sample into the
GC for a predefined time period, followed by a trigger signal to
start the GC. In the meantime, the next reaction zone is filling
the other loop.
Reaction Time. The zone broadening is primarily from hydro-
dynamic dispersion. If pure molecular diffusion were important,
we would see the band spreading increase with reaction time.
We kept zones in the capillary for 1, 3, 10, and 24 h, and then the
UV signals were recorded by the optical fiber UV detector. Figure
2 shows that the four UV signals are almost identical. So, the
residence time in the reactor can be dictated by the chemistry,
and diffusional broadening for up to 24 h is not important.
GC Analyzer. Scheme 2 shows the GC analyzer, which includes
the absorbance detector, M-6 pump, 10-port valve, and GC. When
the reactions in the microreactor are finished, the samples are
pushed out of the microreactor by SP2. The absorbance detector
monitors the capillary, finding zones using user-defined absor-
bances. We define the start of a zone as A ) 0.8 (absorbance
ascending) and the end as A ) 0.2 (absorbance descending). Zone
detection alerts the injector to switch positions so that the just-
detected zone is captured. The pump connected to the 10-port
valve (SP3) is, at the same time, instructed to inject the contents
of the previously filled 1-µL loop into the GC. Finally, also at the
same time, the GC is triggered to start its temperature program.
Parameter Adjustment. SP1 (Refer to Scheme 1). The
reactants are pumped into the system from SP1. The flow rate
should be high because band spreading in this section of the
system is not a problem, but the time taken to load the reactor
should be a minimum. This flow rate is equal to the flow rate
from the autosampler. On the other hand, the pressure cannot
be too high. The optimal flow rate of the autosampler and SP1 is
1
5 µL/min.
SP2 and Injector Timing. The time between reaction zones
controls the overlap of the zones in the reactor. Two times, both
controllable, dictate the time between zones. One is the time
required by the autosampler to pull sample into the syringe and
eject it. This is essentially the time required to load the loop. The
other is the time needed to empty the loop. SP2 controls the time
required to push the loop’s contents into the reactor. It also
controls the velocity of the zones in the reactor, and thus the band-
spreading and the time needed to fill and empty the reactor. Figure
1
shows the UV responses at 530 nm from injections of neutral
red. For each flow rate, the sum of the two times was set
1976 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

Figure 4. Absorbance response of the products of Stille reactions. Catalyst: Pd compound (2%) + 8.2 mg of AsPh3 (4%) + 3 mL of THF.
Reactant: 75 µL of PhI + 250 µL of Bu3SnCHdCH2 + 3 mL of THF. Loop volume, 1.0 µL; SP1, 15 µL/min; SP2, 0.25 µL/min; injection time,
4.2 min.
To accomplish this, specific software was written to control the
absorbance detector, pump, valve, and GC.
Stille Reaction. The reaction is shown below
catalyst
R-X + R′SnR3′′
8 R-R′ + XSnR3′′
R, R′ ) aryl, vinyl, allyl, X ) Br, I
(reaction 1)
The mechanism of the palladium-catalyzed Stille reaction has been
widely investigated. The original mechanism proposed by Stille,54
in the case of monodentate ligand L, included four steps: oxidative
addition, transmetalation, trans/cis isomerization, and reductive
elimination involving saturated 16-electron aryl-Pd(II) complexes,
6
2
all ligated by two phosphine ligands. The catalyst is the key to
the reaction.
In the current work, the Stille reaction was performed within
the capillary microreactor of the screening system. PhI reacts with
3
Bu SnCHdCH
2
with the help of a catalyst.
catalyst
PhI + Bu SnCHdCH
8 PhCHdCH + Bu SnI
2 3
3
2
Figure 5. Chromatograms of the products of the Stille reaction by
online GC. (A) Catalyst: PdCl2(CH3CN)2 (2%) + 8.2 mg of AsPh3
(4%) + 3 mL of THF. Reactant: 75 mL of PhI + 250 mL of Bu3-
SnCHdCH2 + 3 mL of THF. (B) Catalyst: PdCl2[(C6H5)3P]2 (2%) +
8.2 mg of AsPh3 (4%)+ 3 mL of THF. Reactant: 75 mL of PhI + 250
mL of Bu3SnCHdCH2 + 3 mL of THF.
The microreactor uses GC to calculate reaction yield. Figure 3
shows that standards of styrene (0.47 min) and PhI (0.75 min)
are well separated. The calibration curve for styrene/dodecane
2
is linear (r ) 0.997) in the range from 0 to 100% yield.
Screening the Catalyst Libraries of the Stille Reaction.
Precatalysts. Six Pd precatalysts and 5 ligands were selected for
5
5,63-67
the Stille reaction.
For screening the precatalysts with a
known good ligand, the autosampler was loaded with 2 mol %
precatalyst and 4 mol % AsPh
3
as the ligand.55 Each catalyst was
(
(
62) Burgess, K. Chem. Ind. 2001, 189-190.
63) Espinet, P.; Echavarren, A. M. Angew. Chem., Int. Ed. 2004, 43, 4704-
loaded twice (total 12 reactions). Figure 4 shows the 12 zones as
seen by the absorbance detector. We note that several of the zones
have a rather high absorbance at or near the stray light limit of
the instrument. As we do not use the absorbance quantitatively,
this not a problem. In Figure 5, A and B are two chromatograms
from the analysis of the Stille reaction by online GC with 2 mol %
4
734.
64) Su, W.; Urgaonkar, S.; McLaughlin, P. A.; Verkade, J. G. J. Am. Chem. Soc.
004, 126, 16433-16439.
65) Su, W.; Urgaonkar, S.; Verkade, J. G. Org. Lett. 2004, 6, 1421-1424.
66) Scrivanti, A.; Matteoli, U.; Beghetto, V.; Antonarol, S.; Crociani, B. Tetra-
hedron 2002, 58, 6881-6886.
(
2
(
(
(
67) Littke, A. F.; Schwarz, L.; Fu, G. C. J. Am. Chem. Soc. 2002, 124, 6343-
6
348.
2
PdCl (CH
3 2 6 5 3 4 3
CN) and Pd[(C H ) P] , (each with 4 mol % AsPh ). It
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006 1977

Table 1. Yields of the Stille Reaction with Various Precatalystsa
precatalyst (2 mol %)
yield (%)
precatalyst (2 mol %)
yield (%)
Pd2dba3
Pd[(C6H5)3P]4
38.0 ( 0.7
PdCl2[(C6H5)3P]2
15.9 ( 0.7
PdCl2(C6H5CN)2
43.8 ( 0.7
Pd (OAC)2
23.0 ( 1.5
b
49.2 ( 0.9
PdCl2(CH3CN)2
50.7 ( 1.1
a
Reactants: 75 µL of PhI + 250 µL of Bu3SnCHdCH2 + 24 µL of dodecane + 3 mL of THF. Catalysts: Pd compound (2%) + 8.2 mg of AsPh3
b
(
4%) + 3 mL of THF, 50 °C for 5 h in capillary microreactor. Mean ( SD, n ) 3.
Table 2. Yields of Stille Reaction with Various
Ligandsa
ligands
AsPh
3
PPh
3
(2-furyl)
3
P
(4-FC
6
H
4
)
3
P
(4-ClC
6 4 3
H ) P
yield (%) 51.7 ( 0.7b 28.5 ( 0.5 40.0 ( 0.7 21.1 ( 0.7
15.5 ( 0.4
a
Reactants: 75 µL of PhI + 250 µL of Bu3SnCHdCH2 + 24 µL of
dodecane + 3 mL of THF. Catalysts: PdCl2(CH3CN)2 (2 mol %) +
b
ligand (4 mol %) + 3 mL of THF, 50 °C for 5 h. Mean ( SD, n ) 3.
particles were deposited from solvent if 3 and 4 mol % PdCl
CN) were used for the Stille reaction, only the yields of styrene
with 0.5, 1. and 2 mol % PdCl (CH CN) are listed in Figure 6. It
2 3
(CH -
2
2
3
2
shows that generally all the catalytic activities are increased as
the concentration of precatalyst increases.
Figure 6. Yields of the Stille reaction with different mole equivalents
of precatalysts and 4 mol % AsPh3.
Experimental Optimization of Throughput. At present, we are
trying to make small adjustments in the screening system to
increase throughput while maintaining the 6.7-m capillary mi-
croreactor and the 5-h reaction time. We have decreased the
volume of the loop from 1.0 to 0.75 µL to narrow the reaction
zones. Figure 7 is the UV response of 20 reaction zones in the
capillary microreactor. The total time for this set of reactions is
∼9 h (2-h injection and detection phases plus 5-h reaction time).
This represents a throughput of 2.2 5-h reactions/h.
is obvious that the styrene response of Figure 5A is higher than
that of Figure 5B. Accordingly, as one of the reactants, the
remaining amount of PhI in Figure 5B is higher than that of Figure
5
A. This means the PdCl
than Pd[(C P] . Table 1 shows the yields from the six
precatalysts. The reactions with Pd dba and PdCl (CH CN)
AsPh
yield ∼50% styrene The yield is only 14% with the catalyst
Pd[(C P] /AsPh . PdCl (CH CN) will be used as the pre-
2 3 2
(CH CN) has more catalytic activity
6
H
5
)
3
4
2
3
2
3
2
/
3
Ligands. Using the same procedures as for the precatalysts
except using the higher throughput conditions described above,
H
6 5
)
3
4
3
2
3
2
catalyst in screening ligands (discussed below).
the ligands, including AsPh
(4-ClC P were screened in the capillary microreactor. Table
2 compares the yields from the five ligands using one precatalyst
(2 mol % PdCl (CH CN) ). AsPh and the furyl phosphine show
significant catalytic activity. The effect of various mole equivalents
3 3 3 6 4 3
, PPh , (2-furyl) P, (4-FC H ) P, and
The amount of the catalyst is also important to the reaction
6 4 3
H )
yield. Amounts of 0.5, 1, 2, 3, and 4 mol % PdCl
mol % AsPh were selected as the catalyst, and all the reactions
were performed in the capillary microreactor. Because small
2 3 2
(CH CN) and 4
3
2
3
2
3
Figure 7. UV response of the products of 20 Stille reactions. Catalyst: Pd compound (2 mol %) + 8.2 mg of AsPh3 (6 mol %) + 3 mL of THF.
Reactant: 75 µL of PhI + 250 µL of Bu3SnCHdCH2 + 3 mL of THF. Loop volume, 0.75 µL; SP1, 15 µL/min; SP2, 0.25 µL/min; stop time, 1.0
min; injection time, 3.0 min.
1978 Analytical Chemistry, Vol. 78, No. 6, March 15, 2006

system. Note that Figure 8 displays statistical error bars from n
4 or 8. This plot consists of 40 determinations of reaction yield.
)
This represents two 9-h runs on the instrument requiring no more
than 2 h of operator time in total.
Reaction Time. The reaction time is also an important factor
for this obviously slow organic reaction. Various reaction times
were used in the microreactor (Figure 9). When the reaction time
is in the range of 1-3 h, the yield increases slowly from 36 to
4
2%. But the yield increases very quickly from 42 to 63% when
the reaction time is 5 h. The yield seems to plateau after 5 h.
This reaction time is also similar to that found by the traditional
method.5
3-55,66
Reproducibility. Ten reactions were carried out at the optimal
catalyst and reaction time. The yields are 61.99 ( 0.36% (mean (
SEM, n ) 10). The reproducibility is remarkable.
Figure 8. Yields of the Stille reaction with different mole equivalents
of AsPh3 and 2 mol % equiv PdCl2(CH3CN)2.
CONCLUSION
Microliter volumes of reagents and catalysts can be brought
together and reacted in parallel for a specified time. Online GC
analyzes the components of the reaction zones. The Stille reaction
was studied by the micro total analytical screening system. Six
precatalysts and five ligands with different mole equivalents were
screened in the capillary microreactor at the same time. PdCl
2 3
(CH -
CN) (2 mol %) + 8.2 mg of AsPh (6 mol %) was selected as the
2
3
optimum catalyst for the Stille reaction. It is a high-throughput
and efficient, enviromentally friendly method and can shorten the
development time from laboratory to commercial production. This
system with capillary microreactor technology promises to yield
a wide range for combinatorial synthesis, screening of catalysts,
and reaction kinetics.
Figure 9. Effect of reaction time to the Stille reaction in the capillary
microreactor. Two mole % of PdCl2(CH3CN)2 + 6 mol % AsPh3.
ACKNOWLEDGMENT
This work is supported by NIH P50 GM067082. We thank
Chris van Thielsburg (Valco) and Dr. Flavio Bedini (Thermo-
Electron) for valuable discussions on the automated injection.
of AsPh
styrene is only ∼35% with 2 mol % AsPh
CN) increases to 6 mol %, and
. It increases to ∼65% as AsPh
then the yield reaches a plateau up to 12 mol %. So, in this work,
the catalyst composed of 2 mol % PdCl (CH CN) and 6 mol %
AsPh is the most active. This stoichiometry was also obtained
by traditional synthesis,
3
for the Stille reaction is shown in Figure 8. The yield of
3
and 2 mol % PdCl (CH
2
3
-
2
3
Received for review October 14, 2005. Accepted
December 27, 2005.
2
3
2
3
5
5,66
proving the validity of the screening
AC051844+
Analytical Chemistry, Vol. 78, No. 6, March 15, 2006 1979