Overcoming the challenges of solid bridging and constriction during Pd-catalyzed C-N bond formation in microreactors

Article abstract of DOI:10.1021/op100154d

We investigate the mechanisms that govern plugging in microreactors during Pd-catalyzed amination reactions. Both bridging and constriction were shown to be important mechanisms that lead to clogging in our system and greatly limited the utility of microsystems for this class of reactions. On the basis of these observations, several approaches were engineered to overcome the challenge of plugging and to enable the continuous-flow synthesis of a biarylamine. Bridging could be eliminated with acoustic irradiation while constriction was managed via fluid velocity and the prediction of growth rates.

Full text of DOI:10.1021/op100154d

Organic Process Research & Development 2010, 14, 1347–1357
Overcoming the Challenges of Solid Bridging and Constriction during Pd-Catalyzed
C-N Bond Formation in Microreactors
†
‡
†
,‡
,†
Ryan L. Hartman, John R. Naber, Nikolay Zaborenko, Stephen L. Buchwald,* and Klavs F. Jensen*
Department of Chemical Engineering, and Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts
AVenue, Cambridge, Massachusetts, U.S.A.
Abstract:
to overcome this challenge are necessary to apply microsystems
to a wider range of reaction discovery and process optimization.
A number of creative strategies have been employed to
handle solids in both microreactors and simple microfluidic
systems. While immobilization of the particles (e.g., catalyst
or solid-supported reagent) can minimize the risk of micro-
We investigate the mechanisms that govern plugging in microre-
actors during Pd-catalyzed amination reactions. Both bridging and
constriction were shown to be important mechanisms that lead to
clogging in our system and greatly limited the utility of microsys-
tems for this class of reactions. On the basis of these observations,
several approaches were engineered to overcome the challenge of
plugging and to enable the continuous-flow synthesis of a biary-
lamine. Bridging could be eliminated with acoustic irradiation
while constriction was managed via fluid velocity and the predic-
tion of growth rates.
17-20,27
channel clogging,
regenerating a packed bed necessitates
the removal and/or recharging of such immobilized solids. The
31-34
synthesis of nanoparticles
can mitigate clogging caused by
flow-induced bridging but can lead to retention and build-up
when the particles interact with the surfaces of the system.
Another approach is to use multiphase liquid-liquid flow to
prevent particles from interacting with the walls of the micro-
1
. Introduction
21,23,28
channels. Encapsulating particles in droplets
or establishing
The problem handling of solids in microfluidic systems has
30
annular-type flow not only limits the interactions between the
particles and walls but also constrains particle-to-particle
interactions that may lead to clogging. However, these special-
ized systems often require the use of solvents that can be
incompatible with certain reagents or can lead to changes in
the efficiency of many organic reactions. Noninvasive ap-
proaches such as flow focusing and filtration have proven to
gained considerable attention in recent years as many interesting
applications, from the manipulation of biological materials to
a wide range of organic chemistry reactions, involve hetero-
geneous mixtures. For example, many important reactions in
the synthesis of fine chemicals, including active pharmaceutical
ingredients (APIs), require more than one phase, be it gas and
liquid, liquid and liquid, or solid and liquid. Microreactors
have been widely applied in research to advance a deeper
understanding of the physical and chemical rate processes that
1
-3
(16) Jensen, K. F. In New AVenues to Efficient Chemical Synthesis:
Emerging Technologies; Seeberger, P. H., Blume, T., Eds.; Springer-
Verlag: Berlin, Heidelberg, 2007; p 5776.
(17) Haswell, S. J.; O’Sullivan, B.; Styring, P. Lab Chip 2001, 1, 164–
166.
4-16
govern reactions,
has been relatively limited.
to clogging in microsystems and the development of strategies
yet their use for reactions involving solids
17-33
Understanding why solids lead
(
18) Losey, M. W.; Schmidt, M. A.; Jensen, K. F. Ind. Eng. Chem. Res.
2
001, 40, 2555–2562.
(19) Baxendale, I. R.; Deeley, J.; Griffiths-Jones, C. M.; Ley, S. V.; Saaby,
*
To whom correspondence should be addressed. E-mail: kfjensen@mit.edu.
S.; Tranmer, G. K. Chem. Commun. 2006, 2566–2568.
Tel: +1 617 253-4589. Fax: +1 617 298-8992.
(20) Wiles, C.; Watts, P.; Haswell, S. J. Lab Chip 2007, 7, 322–330.
(21) Poe, S. L.; Cummings, M. A.; Haaf, M. R.; McQuade, D. T. Angew.
Chem., Int. Ed. 2006, 45, 1544–1548.
†
Department of Chemical Engineering.
Department of Chemistry.
‡
(
(
(
1) Roberge, D. M.; Ducry, L.; Bieler, N.; Cretton, P.; Zimmermann, B.
Chem. Eng. Technol. 2005, 28, 318–323.
(22) Kockmann, N.; Kastner, J.; Woias, P. Chem. Eng. J. 2008, 135, S110–
S116.
(23) Song, H.; Chen, D. L.; Ismagilov, R. F. Angew. Chem., Int. Ed. 2006,
45, 7336–7356.
(24) Horie, T.; Sumino, M.; Tanaka, T.; Matsushita, Y.; Ichimura, T.;
Yoshida, J. Org. Process Res. DeV. 2010, 14, 405–410.
(25) Conant, T.; Karim, A.; Rogers, S.; Samms, S.; Randolph, G.; Datye,
A. Chem. Eng. Sci. 2006, 61, 5678–5685.
(26) Li, W.; Pharn, H. H.; Nie, Z.; MacDonald, B.; Guenther, A.;
Kumacheva, E. J. Am. Chem. Soc. 2008, 130, 9935–9941.
(27) Kobayashi, J.; Mori, Y.; Okamoto, K.; Akiyama, R.; Ueno, M.;
Kitamori, T.; Kobayashi, S. Science 2004, 304, 1305–1308.
(28) Shestopalov, I.; Tice, J. D.; Ismagilov, R. F. Lab Chip 2004, 4, 316–
321.
(29) Trachsel, F.; Tidona, B.; Desportes, S.; Rudolf von Rohr, P. J.
Supercrit. Fluids 2009, 48, 146–153.
(30) Nagasawa, H.; Mae, K. Ind. Eng. Chem. Res. 2006, 45, 2179–2186.
(31) Takagi, M.; Maki, T.; Miyahara, M.; Mae, K. Chem. Eng. J. 2004,
101, 269–276.
(32) Khan, S. A.; Jensen, K. F. AdV. Mater. 2007, 19, 2556.
(33) Khan, S. A.; Gunther, A.; Schmidt, M. A.; Jensen, K. F. Langmuir
2004, 20, 8604–8611.
(34) Yen, B. K. H.; Gunther, A.; Schmidt, M. A.; Jensen, K. F.; Bawendi,
M. G. Angew. Chem., Int. Ed. 2005, 44, 5447–5451.
2) Carey, J. S.; Laffan, D.; Thomson, C.; Williams, M. T. Org. Biomol.
Chem. 2006, 4, 2337–2347.
3) Dugger, R. W.; Ragan, J. A.; Ripin, D. H. B. Org. Process Res. DeV.
2
005, 9, 253–258.
(
(
4) Jensen, K. F. Chem. Eng. Sci. 2001, 56, 293–303.
5) Fletcher, P. D. I.; Haswell, S. J.; Pombo-Villar, E.; Warrington, B. H.;
Watts, P.; Wong, S. Y. F.; Zhang, X. L. Tetrahedron 2002, 58, 4735–
4
757.
(
(
6) Jahnisch, K.; Hessel, V.; Lowe, H.; Baerns, M. Angew. Chem., Int.
Ed. 2004, 43, 406–446.
7) Pennemann, H.; Watts, P.; Haswell, S. J.; Hessel, V.; Lowe, H. Org.
Process Res. DeV. 2004, 8, 422–439.
(
(
8) Hessel, V.; Lowe, H. Chem. Eng. Technol. 2005, 28, 267–284.
9) Jensen, K. F. MRS Bull. 2006, 31, 101–107.
(
(
10) deMello, A. J. Nature 2006, 442, 394–402.
11) Mason, B. P.; Price, K. E.; Steinbacher, J. L.; Bogdan, A. R.; McQuade,
D. T. Chem. ReV. 2007, 107, 2300–2318.
(
(
(
(
12) Watts, P.; Wiles, C. Chem. Commun. 2007, 443–467.
13) Watts, P.; Wiles, C. Chem. Eng. Technol. 2007, 30, 329–333.
14) Hartman, R. L.; Jensen, K. F. Lab Chip 2009, 9, 2495–2507.
15) Ehrfeld, W.; Hessel, V.; Lowe, H. Microreactors: New Technology
for Modern Chemistry; Wiley-VCH: Weinheim, Germany, 2000.
1
0.1021/op100154d  2010 American Chemical Society
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1347
Published on Web 09/10/2010

be effective in separating particles based on hydrodynamic
flow microchemical synthesis of a biaryl amine. Consequently,
the present work establishes the groundwork for further
continuous-flow studies on both the mechanisms responsible
for clogging and synthetic applications of Pd-catalyzed cross-
coupling reactions.
35-38
conditions.
These conditions, however, limit reaction
screening applications and present challenges with systems that
have high particle concentrations or particles that have an
affinity for each other or the surfaces of the reactor. Electrical
potentials have also been used to sort particles in microfluidic
39,40
systems.
Yet another noninvasive approach is the application
2
. Reactor Design
of acoustic standing waves to order particles, such as the
2.1. Microfabrication and Materials. Microchannel de-
41-44
acoustic streaming of blood cells.
More recently, we have
vices were fabricated from a double-side-polished silicon wafer
and capped with a Pyrex wafer. As shown in Figure 1a, the
fabrication process involved several photolithography steps,
deep reactive ion etching of the silicon, and growth of a low-
stress silicon nitride coating (0.5 µm). Anodic bonding of the
Pyrex wafer capped the etched silicon-nitride-coated features
and completed the microfluidic device. The packaged device,
with a serpentine microchannel (0.4 × 0.4 × 875 mm) is shown
in Figure 1b. As will be discussed later, a device with a spiral
microchannel, of the same dimensions, was also fabricated in
the same manner (Figure 1c). The halo through-etch between
the cooled inlets and outlet and main channel (Figure 1b)
provided thermal separation between the two parts of the reactor
and allowed for operation with two separate temperature zones,
demonstrated the potential for using ultrasound to handle solids
during the course of a chemical reaction, while others have
combined ultrasound and multiphase flow to facilitate the flow
24
of solids during polymerization. In general, the strategy for
solids handling depends on the nature of the chemistry and
warrants a deeper understanding of the mechanisms that lead
to clogging.
The Pd-catalyzed C-N bond forming reaction is one of the
most important reactions of the past 15 years, finding applica-
tions in the production of APIs, natural products, and specialty
45-48
chemicals.
It is a versatile transformation that allows the
coupling of aryl and vinyl halides, triflates, nonaflates, and other
sulfonates with a wide variety of nitrogen nucleophiles. These
reactions, and many others used in fine chemical production,
form inorganic salt byproducts and are often run in nonpolar
solvents, which leads to precipitation.
34
each nearly isothermal, on a single chip. High purity poly-
tetrafluoroethylene (PFA) capillaries were also found to be
useful as model reactors (Figure 1d).
In the present work, we use the Pd-catalyzed amination
reaction to investigate the mechanisms that govern the clogging
2
.2. Acoustic Integration and Layout. Figure 2a illustrates
the equipment configuration used for the present study. Reagents
in 10 mL glass syringes were delivered to the microreactor (see
Figure 1B) by syringe pumps. A stainless-steel pressure
transducer was plumbed directly into one of the reagent lines
upstream to the mixing point. It was possible to constantly
monitor the pressure entering the reactor. Upon exiting the
reactor, the mixture passed through approximately 120 µL of
PFA tubing at ambient temperature before it was collected in a
vial that was under a constant 1.7 psig pressure of argon.
Experiments with acoustic irradiation were performed by
replacing the microreactor of Figure 1b with lengths of loosely
49
of microreactors. We will show that both bridging and
constriction are important mechanisms that lead to clogging and
serve to limit the application of microsystems to reactions that
form insoluble byproducts. We present several approaches to
overcome the challenge of plugging and thus enable continuous-
(
35) Di Carlo, D.; Irimia, D.; Tompkins, R. G.; Toner, M. P. Natl. Acad.
Sci. U.S.A. 2007, 104, 18892–18897.
(
(
36) Yamada, M.; Seki, M. Lab Chip 2005, 5, 1233–1239.
37) Kuntaegowdanahalli, S. S.; Bhagat, A. A. S.; Kumar, G.; Papautsky,
I. Lab Chip 2009, 9, 2973–2980.
(
(
38) Huang, L. R.; Cox, E. C.; Austin, R. H.; Sturm, J. C. Science 2004,
1
3
04, 987–990.
coiled PFA tubing ( /16 in. o.d., 500 or 1000 µm i.d.) (Figure
b). Upon exiting the bath, the reaction mixture was collected
under argon as described above.
39) Kralj, J. G.; Lis, M. T. W.; Schmidt, M. A.; Jensen, K. F. Anal. Chem.
2
2
006, 78, 5019–5025.
(
(
40) Pamme, N. Lab Chip 2007, 7, 1644–1659.
41) Laurell, T.; Petersson, F.; Nilsson, A. Chem. Soc. ReV. 2007, 36, 492–
5
06.
(
(
(
(
(
42) Nilsson, A.; Petersson, F.; Jonsson, H.; Laurell, T. Lab Chip 2004, 4,
31–135.
43) Petersson, F.; Aberg, L.; Sward-Nilsson, A. M.; Laurell, T. Anal. Chem.
007, 79, 5117–5123.
44) Petersson, F.; Nilsson, A.; Holm, C.; Jonsson, H.; Laurell, T. Lab Chip
005, 5, 20–22.
45) Surry, D. S.; Buchwald, S. L. Angew. Chem., Int. Ed. 2008, 47, 6338–
3. Results and Discussion
1
In this study of clogging on the microscale, we employed
the coupling of 4-chloroanisole and aniline, catalyzed by catalyst
supported by one of the bulky, electron-rich, biaryl phosphine
2
2
50
ligands, to form 4-methoxy-N-phenylaniline (Scheme 1).
6
361.
It can be seen in Scheme 1 that sodium chloride is formed
as a stoichiometric byproduct in this reaction. Inorganic salts
are generally insoluble in the solvents used to carry out this
reaction (e.g., 1,4-dioxane, toluene, THF). To complicate matters
further, typical batch conditions for these reactions employ high
concentrations (0.5-1.0 M) and the reactions are often complete
in minutes, resulting in fast rates of salt formation. This is a
problem that is synthetically interesting, yet challenging to carry
out under flow conditions; it served as a model to study solids
handling in microsystems.
46) Buchwald, S. L.; Jiang, L. In Metal-Catalyzed Cross-Coupling
Reactions; deMeijere, A, Diederich, F., Eds.; Wiley-VCH: Weinheim,
2
004; p 699.
(
(
(
47) Hartwig, J. F. Acc. Chem. Res. 2008, 41, 1534–1544.
48) Marion, N.; Nolan, S. P. Acc. Chem. Res. 2008, 41, 1440–1449.
49) For references pertaining to palladium-catalyzed amination reactions
in microflow see: Mauger, C.; Buisine, O.; Caravieilhes, S.; Mignani,
G. J. Organomet. Chem. 2005, 690, 3627–3629. Murphy, E. R.;
Martinelli, J. R.; Zaborenko, N.; Buchwald, S. L.; Jensen, K. F. Angew.
Chem., Int. Ed. 2007, 46, 1734–1737. Popa, D.; Marcos, R.; Sayalero,
S.; Vidal-Ferran, A.; Peric a` s, M. A. AdV. Synth. Catal. 2009, 351,
1
539–1556. Shore, G.; Morin, S.; Mallik, D.; Organ, M. G.
Chem.sEur. J. 2008, 14, 1351–1356. Bazinet, P.; McMullen, J. P.;
Naber, J. R.; Musacchio, A.; Jensen, K. F.; Buchwald, S. L.; Angew.
Chem., Int. Ed., submitted for publication. Naber, J. R.; Buchwald,
S. L. Angew. Chem., Int. Ed., submitted for publication.
(50) Huang, X.; Anderson, K. W.; Zim, D.; Jiang, L.; Klapars, A.;
Buchwald, S. L. J. Am. Chem. Soc. 2003, 125, 6653–6655.
1
348
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

Figure 1. (a) Microreactor fabrication process. (b) Silicon microreactor with serpentine channels used to study solids handling.
The image shows a compression chuck used to cool the inlets and outlet of each device. (c) Silicon microreactor designed to eliminate
the 180° turns. The device has a gradual spiral in and out to imitate a long straight channel. (d) PFA capillaries used as model
reactors (i.d. ) 500 and 1000 µm).
Figure 2. Experimental setup used to carry out solids handling experiments: (a) a silicon-based microreactor; (b) PFA capillary in
an ultrasonic bath.
Scheme 1. Pd-Catalyzed amination of an aryl chloride with
aniline
The sudden increase in pressure drop was followed by clogging
of the microreactor to the point that no flow occurred. As evident
in Figure 3a, the onset of clogging took place after less than 6
reactor volumes injected.
As shown in Figure 3b, clogging was caused by a white
solid that accumulated at the 180° turns within the microreactor.
Under increased magnification (Figure 3c) it can be seen that
the solid material was crystalline in nature and the sizes of the
particles found were on the same order as the channel
dimensions (e.g., 400 µm). Here, we define the crystal size as
the largest cross-section dimension of an individual particle.
When the size of flowing and stable particles is on the order of
the cross-sectional length of a channel, it has been shown that
3
.1. Clogging Mechanisms. The aforementioned reaction
was first carried out in a series of experiments using a silicon-
based microreactor to allow the process to be followed visually.
Figure 3a shows that the dimensionless pressure drop across
0
the microreactor, ∆P/∆P (normalized to the initial, solvent-
only, pressure drop), rapidly increased upon injecting the
reagents at 20 µL/min (τ ) 7 min) and 80 °C, even when the
concentration of aryl halide was only 0.1 M (concentrations
refer to the reaction mixture after the two streams are combined).
51-53
bridging takes place in micrometer-sized pore throats
and
54
fabricated geometries. The depiction in Figure 3d illustrates
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1349

Figure 3. Pd-catalyzed amination leading to bridging in microreactors. (a) Dimensionless pressure drop across a microreactor
during the injection of the reagents and catalyst (at 20 µL/min and 80 °C) shown in Scheme 1. (b) Photograph of the microreactor
illustrating the accumulation of white solid, primarily at 180° turns. (c) Expanded view of the microchannel. (d) Depiction of flow-
induced bridging between two flat plates.
that particles can plug by forming a bridge across a microchan-
nel. It is generally understood that particles can be retained when
the aspect ratio (e.g., the ratio of channel width to particle size,
D/a, is in the range 3-4. It should be noted, however, that
this type of bridging has been shown for stable particle
suspensions that do not exhibit attractive interactions or that
do not change size with time. Based on our experimental
observations, it is evident that bridging also takes place during
the amination reaction and is one of the mechanisms that leads
to clogging.
The clogged microreactor was replaced with a fresh
device and the reaction was repeated with an increased
injection rate of 70 µL/min (τ ) 2 min). Figure 4a shows
that the pressure drop once again increased to the point
of clogging with the injection of only a few reactor
volumes of the reaction. However, in this case the pressure
drop gradually increased for 4 reactor volumes before
suddenly rising and clogging.
in the bulk solution. The presence of constriction in
addition to bridging warrants more than one approach to
address the clogging issues caused by the insoluble
byproducts of the Pd-catalyzed amination.
51
3.2. Influence of Reactor Geometry. Particle-to-wall
interactions can take place in laminar flow when changes
in the velocity occur (e.g., fluid flow around turns) or when
the particle momentum is too large for particles to stay
55
on a streamline path (e.g., inertial impaction). To
minimize these scenarios, we investigated the use of
silicon microreactors with gradual turns (e.g., analogous
to coiled tubing), as shown in Figure 1c. The reactor
design was closely related to the serpentine microreactor
(
Figure 1b), retaining the same layout of ports and mixing
zone, an equal volume, and an identical thermal isolation
of the reaction zone from the ports. However, the nested
spiral layout of the heated channel provided a gradual
change in turning radius, with the sharpest turn having a
much larger radius (4.4 mm) than the hairpin turns (0.4
mm) of the serpentine layout. When the experiment
described in Figure 3 was repeated using the spiral reactor,
we observed the solids flowed for several reactor volumes
without the accumulation shown in Figure 3b. Neverthe-
less, the microreactor wall (from the top down) gradually
changed from transparent to white, and clogging eventu-
Further inspection of the microreactor (Figure 4b)
illustrates that the reactor walls changed color, implying
that material was deposited or grown on these surfaces.
5
5-57
Both deposition
and nucleation followed by growth
can lead to channel constriction (Figure 4c). A time-
dependent decrease in cross-sectional diameter will likely
bring about bridging in a system with particles suspended
(
(
51) Ramachandran, V.; Fogler, H. S. J. Fluid Mech. 1999, 385, 129–156.
52) Vitthal, S.; Sharma, M. M. J. Colloid Interface Sci. 1992, 153, 314–
(55) Ramachandran, V.; Fogler, H. S. Langmuir 1998, 14, 4435–4444.
3
36.
(56) Song, L. F.; Elimelech, M. J. Colloid Interface Sci. 1994, 167, 301–
(
(
53) Muecke, T. W. J. Petrol. Tech. 1979, 31, 144–150.
313.
54) Wyss, H. M.; Blair, D. L.; Morris, J. F.; Stone, H. A.; Weitz, D. A.
Phys. ReV. E Stat. Nonlin. Soft. Matter Phys. 2006, 74, 061402.
(57) Marshall, J. K.; Kitchener, J. A. J. Colloid Interface Sci. 1966, 22,
342–351.
1
350
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

Figure 4. Pd-catalyzed amination resulting in constriction in a microreactor. (a) Dimensionless pressure drop across a microreactor
during the injection of the reagents and catalyst (at 70 µL/min and 80 °C) shown in Scheme 1. (b) Photograph of the microreactor
during reaction. (c) Depiction of constriction when material grows or deposits on a channel wall.
ally took place. A video of these results can be found in
the Supporting Information.
when frequencies are in the range of megahertz and wavelengths
41,42,59,60
are on the order of the channel dimensions.
We wanted
3
.3. Acoustic Irradiation. It is commonly known that sound
to study whether these forces, with frequencies in the kilohertz
range, could be used to overcome the particle-to-particle
attractive interactions and hydrodynamic conditions that bring
about bridging in laminar flow (see the Supporting Information
for a description of the acoustic waveform applied in this study).
The reaction from Scheme 1 was carried out both with and
without acoustic irradiation by injecting the reagents and base
(0.1 M ArCl, τ ) 7 min) into a PFA capillary (1000 µm i.d.,
58
waves can impose a force on a system of particles. Thus, we
investigated the influence of acoustic waves on microreactor
clogging during Pd-catalyzed C-N bond forming reactions. In
the case of a standing wave, the amplitude of the primary
radiation force acting on a particle (in the radial direction)
r
flowing in a microchannel, F , has previously been described
and given as
41,42
2
40 µL) coiled in an ultrasonic bath filled with water at 80 °C.
2
2 3
2
π p r ꢀ
0 S
4πx
λ
Figure 5a shows that the microreactor clogged in the absence
of ultrasound, shown by the rapid increase in dimensionless
pressure drop. Repeating the experiment with acoustic irradia-
tion demonstrated that the presence of ultrasound prevented
clogging, as the pressure drop remained negligible for ap-
proximately 20 reactor volumes. Removing the acoustic ir-
radiation in the same experiment resulted in clogging of the
microreactor, as shown in Figure 5a, and subsequent investiga-
tions proved that reinitiating the ultrasonic bath did not unplug
the reactor.
The effect of channel dimensions on plugging was examined
by repeating the experiment described in Figure 5a with a
smaller capillary (500 µm i.d.). The reactor volume (240 µL)
and injection rate (τ ) 7 min) were held constant by increasing
the reactor length. As shown in Figure 5b, the dimensionless
pressure drop also remained small until the acoustic irradiation
was removed. In both cases, samples were collected, and yield
and conversion, relative to the internal standard, were deter-
mined by GC to be 100% and >95%, respectively. Given that
the reaction rate increases with concentration we investigated
whether acoustic-mediated transport would allow for uninter-
Fr ) -
· φ · sin
(1)
(
)
(
)
3
λ
where p
radius, ꢀ
0
is the acoustic pressure amplitude, r is the particle
is the compressibility of the solvent, and φ is a contrast
S
4
1,42
factor, described as
5
FP - 2FS
ꢀP
ꢀS
φ )
-
(2)
(
)
2
FP + FS
where F is density, and the subscripts P and S denote the particle
and solvent, respectively. One observes in eq 1 that the primary
force is strongly dependent on particle radius; thus, the acoustic
force weakens as the particle diameter is reduced. Furthermore,
the primary force is driven by differences in density and
compressibility, described by the contrast factor. The acoustic
pressure amplitude also influences the primary radiation force
and is dependent on the voltage and frequency under which
piezoceramic operation takes place. Such forces have been
exploited to order and separate particulates and biological matter
(
58) Mason, W. P. Physical Acoustics; Academic Press: New York, 1982.
Challis, R. E.; Povey, M. J. W.; Mather, M. L.; Holmes, A. K. Rep.
Prog. Phys. 2005, 68, 1541–1637. Gedanken, A. Ultrason. Sonochem.
(59) Coakley, W. T. Trends Biotechnol. 1997, 15, 506–511.
(60) Hawkes, J. J.; Barber, R. W.; Emerson, D. R.; Coakley, W. T. Lab
Chip 2004, 4, 446–452.
2
6
004, 11, 47–55. Spengler, J.; Jekel, M. Ultrasonics 2000, 38, 624–
28.
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1351

Figure 5. (a) Dimensionless pressure drop during the reaction (τ ) 7 min, 0.1 M ArCl) in a PFA capillary (1000 µm i.d., 240 µL)
at 80 °C. The figure illustrates that the presence of acoustic forces prevents the pressure drop from rapidly increasing. (b) Repeating
the same experiment with 500 µm PFA tubing revealed a similar result.
Figure 6. (a) Particle size estimation for the reaction in the presence of ultrasound. (b) Particle size estimation for the reaction in
the absence of ultrasound. The figures show that acoustic irradiation reduces the maximum effective particle size; the larger particles
are likely aggregates.
rupted flow when concentrations were closer to those of
traditional batch experiments (e.g., 1 M). However, we found
that the solubility limit of NaOt-Bu in 1,4-dioxane at room
temperature was approximately 1.0 M. Taking into account both
the stoichiometry of the reaction and the dilution resulting from
the combination of the base stream and the stream contain-
ing the remainder of the reagents, the maximum concentration
was determined to be 0.36 M in ArCl. Under these conditions
proximately 0.15 to 36 µm. The different loadings resulted in
different total particle concentrations (depending on the reaction
conversion); however, the relative particle size distributions were
equivalent in all cases.
In a separate set of experiments, we were able to flow the
reaction for several reactor volumes in the absence of ultrasound
without plugging. Samples were collected before plugging took
place, and particle size analysis revealed particles in the range
(τ ) 1 min), flow was possible in the presence of ultrasound,
0.15-112 µm, as shown in Figure 6b. The additional mode of
samples were collected, and the conversion and yield were
measured to be 100% and >95% by GC, relative to the internal
standard. These observations were made for both 500 and 1000
µm PFA microreactors. Nevertheless, exactly why the ultra-
sound was preventing clogging remained ambiguous.
To elucidate the role of acoustic irradiation on flow, particles
exiting the capillary (1000 µm i.d., τ ) 1 min) were analyzed.
The reaction mixture was collected into vials, under an argon
atmosphere, containing 1,4-dioxane and the samples were
analyzed using a laser diffraction analyzer. Reactions were run
with catalyst loadings increasing from 0 to 1.0 mol % to give
increasing conversions and yields without changing the flow
conditions or the reagent concentrations of the experiments. As
evident in Figure 6a, varying the catalyst loading (in separate
experiments) yielded particles ranging in diameter from ap-
particles observed in Figure 6b can be attributed to the
aggregation of the salt particles, and combination of the results
in Figure 6a,b demonstrates that ultrasonic forces can reduce
the effective maximum particle size, which in turn prevents
bridging. Thus, applying external forces such as ultrasonic
irradiation can influence and even overcome the particle-to-
particle interactions that can lead to clogging via bridging. In
general, mixing can influence the distributions of particle sizes
61
in microscale reactors and may be different from those
measured in batch vessels. As we have seen, the presence of
acoustics can lead to different particle sizes. One might,
therefore, consider the overall influence of acoustics and reactor
(61) Yen, B. K. H. Massachusetts Institute of Technology, 2007.
1
352
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

Figure 7. (a) PFA capillary showing the wall deposit (left side) after the reaction mixture exited the heated section exposed to
acoustics (right side). (b) Cross-section of the capillary illustrating the wall deposit. SEM micrographs of the wall deposit in the (c)
axial and (d) radial direction.
design (e.g., batch or tubular) on the expected particle size if it
is relevant to a particular process.
material gradually formed on the PFA capillary walls after
injecting approximately 40 reactor volumes. It should be noted,
however, that this material was observed to form upon the walls
of the tubing after the heated zone where the exposure to
ultrasound was limited. No wall deposits were observed to form
in the section of the reactor exposed to ultrasound. Parts a and
b of Figure 7 illustrate the axial and radial cross-sections,
respectively, of the capillary containing the wall deposits. The
capillary was filled with reaction solvent (1,4-dioxane) and was
cooled below the freezing point of the solvent (11.8 °C). Cuts
were immediately made with a razor to preserve the deposited
material on the capillary walls.
Examination of these deposits with scanning electron
microscopy illustrated the film thickness to range from 30 to
50 µm (Figure 6c). As shown in Figure 7d, the deposit was
composed of cauliflower-type clusters, which could be particle
aggregates. Further inspection of the film with SEM revealed
the presence of different crystals, highlighted in Figure 7d. X-ray
diffraction later confirmed the material to be a composite of
sodium chloride and an unknown organic crystal (see Support-
ing Information). Isolation and GCMS analysis of this unknown
substance revealed that it is at least, in part, reaction product.
Subsequent experiments proved that the deposited material
could be easily removed by flushing with reaction solvent
The experimental results shown in Figure 6a,b offer insight
on potential microreactor designs that could prevent bridging
for Pd-catalyzed aminations or other salt forming reactions.
From Figure 6a, the experiment in the presence of ultrasound,
the aspect ratio (D/a) between the largest particle and the
capillary (1000 µm i.d.) was 28. Without any ultrasound, Figure
6b, the system clogged even though the aspect ratio of 8.9 was
greater than the previously stated cutoff of 4 that has been
51
determined for stable colloids. Attractive particle interactions
likely contributed to clogging at this aspect ratio. Nevertheless,
these observations imply that microreactors with channels of
400 µm could potentially tolerate NaCl particles in the range
of 14 µm without bridging. It is possible, however, that particles
larger than 14 µm but smaller than 45 µm (i.e., aspect ratio of
8.9) will not bridge. However, these conservative approxima-
tions neglect the time-dependent change in microchannel
geometry caused by constriction via particle deposition or
growth.
3.4. Constriction via Deposition or Growth. As shown
in the microreactor of Figure 4, constriction can take place in
addition to bridging. Switching the reactor surface from silicon
nitride to fluoropolymer did not inhibit constriction, as white
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1353

particle aggregation took place), we can predict from Figure
a and a constriction rate of 3.2 µm/min that the bridging took
place when particles were in the range of 16 µm.
.5. Eliminating Bridging and Constriction. The rate at
6
3
which salt byproduct accumulates on the surface is influenced
by convection. To examine this influence, the experiment of
Figure 8 was repeated, and the total reactor volume increased
from 240 to 600 µL while maintaining a constant residence
time of 3 min. The reactor volume exiting the ultrasonic bath
was also held constant at 120 µL. One observes in Figure 9a
that increasing the flow rate from 80 µL/min (average velocity
V
avg ) 10.2 cm/min) to 200 µL/min (Vavg ) 25.5 cm/min)
decreased the constriction rate. The solid curved lines of Figure
a represent eq 4. Applying eqs 4 and 5 yields a constriction
Figure 8. Dimensionless pressure drop during the reaction (τ
3 min, 0.36 M ArCl) in a PFA capillary (1000 µm i.d., 240
µL) at 80 °C. The experimental data are plotted with the model
of eq 4, illustrating that constriction initially took place, followed
by bridging.
)
9
rate of approximately 1.8 µm/min upon increasing the flow rate
to 200 µL/min. These observations demonstrate that convective
forces alone can potentially overcome the particle-to-particle
or particle-to-wall interactions that lead to constriction. More-
over, the particles can interact mechanically with the deposit
and limit growth via abrasion. Partial evidence of this phenom-
enon can be seen in Figure 9b. Before bridging, the pressure
drop appeared to approach a plateau (see Figure 9b). Reducing
the residence time from 3 to 1 min eliminated constriction and
bridging altogether, despite the reaction being complete in all
cases. Consequently, the dimensionless pressure drop remained
approximately 1 even after injecting 76 reactor volumes.
3.6. Influence of Acoustics on Reaction. Ultrasound has
proven to be a useful tool in enhancing the rate of organic
followed by water. In this case, constriction could be managed
via periodic flushing schemes, especially when the deposition
or growth rate is predictable.
When constriction takes place before heating of the reactants
or after they have been cooled, the concentration of starting
material or the particles formed remains relatively constant. It
is under these conditions that the change in microchannel
geometry is a constant, R, and expressed as
dD
dt
)
-R
(3)
62,63
transformations,
including reactions catalyzed by pal-
A reduction in cross-section diameter is readily monitored by
measurement of the pressure drop. For laminar flow in a
cylindrical pipe, it can be shown that
64
ladium. It is generally understood that the energy emitted from
cavitations can contribute to enhanced reaction rates in batch-
scale systems that have temperature gradients. Recently, the
research work of others has elucidated that ultrasound enhances
mixing in microfluidic systems.
these temperature and mixing effects would impact the Pd-
catalyzed C-N bond formation reaction in capillary flow.
The coupling reaction was carried out in a PFA capillary at
80 °C and a residence time of 1 min. As shown in Figure 10,
increasing the catalyst loading from 0-1.0 mol % resulted in
increased conversion from starting material to the biaryl amine
product. Without any ultrasound, the yield was reduced for a
given catalyst loading. The data in Table 1 show the results of
a series of experiments performed to confirm the results in
Figure 10. Solution 1 (1.0 M NaOtBu in 1,4-dioxane) and
solution 2 (0.72 M ArCl, 0.86 M aniline, 0.14 M biphenyl, and
∆
P
1
)
(4)
(5)
(6)
4
60,65,66
∆P0
We questioned whether
(1 - γt*)
when
and
Rτ
D0
γ )
t
τ
t* )
0
where D is the initial channel diameter, D is the diameter at
time t, and τ is the residence time.
Equation 4 can be applied to approximate a constriction rate,
R, or to determine the clogging mechanism. When constriction
takes place, one would expect the dimensionless pressure drop
to gradually increase before plugging, as was the case in Figure
3.6 mM XPhos precatalyst in 1,4-dioxane) were injected (120
1
µL/min each, τ ) 1 min) into a PFA capillary ( /16 in. o.d.,
000 µm i.d., 240 µL heated, 360 µL total) submerged in an
1
ultrasonic bath filled with water at 80 °C. While sonicating,
the reaction was run for 5 min to achieve steady state, at which
point sample 1 was collected (5 reactor volumes). The sonication
4. Repeating an experiment in a 1000 µm capillary (240 µL
and τ ) 3 min) in the presence of acoustics and expanding the
y-axis revealed an important observation. As can be seen in
Figure 8, an abrupt increase took place after injecting 57 reactor
volumes, while the approximation of eq 4 predicts the pressure
increase at a later time. These results imply that constriction of
the microreactor diameter eventually resulted in bridging. From
eq 5, the constriction rate was estimated to be 3.2 µm/min.
Assuming that the particles had stopped growing (i.e., only
(
62) Thompson, L. H.; Doraiswamy, L. K. Ind. Eng. Chem. Res. 1999, 38,
1215–1249.
(
(
63) Mason, T. J. Chem. Soc. ReV. 1997, 26, 443–451.
64) Barge, A.; Tagliapietra, S.; Tei, L.; Cintas, P.; Cravotto, G. Curr. Org.
Chem. 2008, 12, 1588–1612.
(65) Johansson, L.; Johansson, S.; Nikolajeff, F.; Thorslund, S. Lab Chip
2
009, 9, 297–304.
(66) Yang, Z.; Goto, H.; Matsumoto, M.; Maeda, R. Electrophoresis 2000,
21, 116–119.
1
354
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

Figure 9. (a) Influence of changing the average flow velocity from 10.2 to 25.5 cm/min on the dimensionless pressure drop. (b)
Reducing the residence time from 3 to 1 min further illustrates that increasing the velocity eliminates significant constriction and
bridging.
scale systems. Unstable gas bubbles, smaller than the channel
cross-section, were observed in the presence of ultrasound,
which may be cavitations that enhance mixing. Further evidence
supporting these observations can be seen in Figure S5 and a
video (see Supporting Information).
The preparation of 4-methoxy-N-phenylaniline represented
a case study for understanding how to handle salt byproducts
in microsystems during Pd-catalyzed reactions. Figure 11
summarizes a decision roadmap for understanding solids
handling while completing the reaction of Scheme 1 and under
the conditions presented in this study. As shown, a simple
algorithm can be followed to identify the mechanism of
microreactor plugging. The choice of reactor design can further
be elucidated depending upon the established mechanism. It is
Figure 10. Influence of ultrasound on product yield for
important to note that microscale tubular reactors have been
the focus of this investigation. Other flow reactor designs, such
as continuous stirred tank reactors (CSTR’s), have been applied
different catalyst loadings.
Table 1. Yields for a series of experiments (0.5 mol %
Catalyst, τ ) 1 min) testing the influence of ultrasound on
reaction yield
67
to successfully handle solids. We would not expect clogging
in any continuous-flow reactor design when aspect ratios meet
the criteria outlined and constriction is managed. In CSTR’s,
however, transfer tubing inlets and outlets may present the
potential for bridging and constriction similar to our findings,
which could be addressed via the guidelines described in Figure
11.
exp
yield w/ ultrasounda
yield w/o ultrasounda
1
2
3
4
86
78
b
b
80
77
86
85
84 ( 3
80
83
73
77 ( 3
74
average
isolated
c
On the basis of the data reported in Figure 9b and the
residence times investigated, approximately 2.6 g/h of product
could be synthesized, whereas synthesis in microflow under
these conditions was previously not possible. Just as importantly,
the working principles could be extended to other Pd-catalyzed
C-N bond formation reactions. Consequently, the laboratory-
scale production of fine chemicals prepared with these reactions
is possible. Furthermore, microchemical systems may be applied
to and the optimization of reaction conditions for individual
reactions and complex synthetic pathways. Finally, the ability
to handle salt byproducts in microsystems offers the opportunity
to use these systems to discover the reaction kinetics and
mechanisms of important transformations.
a
GC yield relative to internal standard. b Reaction performed in reverse order.
c
Combined isolated yield for experments 1-4.
was stopped, and the reaction was run for 4 min to reach a
new steady state, at which point sample 2 was collected (5
reactor volumes). The injection was stopped, and the reaction
was flushed sequentially with dioxane, water, acetone, and
dioxane to remove all reagents, salt byproducts, water, and
acetone, respectively. The system was then ready for the next
experiment. This process was repeated a total of four times,
with the second and fourth experiments being performed with
sample 1 being the reaction with no sonication and sample 2
being the reaction during sonication. The 8 samples were
analyzed by GC and were subsequently combined and isolated
4
. Concluding Remarks
We have found that both bridging and constriction take place
(
see Table 1). These results support that the acoustic irradiation
during Pd-catalyzed amination reactions in microreactors,
may have enhanced temperature, mixing, or both. Given that
temperature gradients are less significant in microscale systems,
one might expect an even more pronounced influence in larger-
(
67) Braden, T. M.; Gonzalez, M. A.; Jines, A. R.; Johnson, M. D.; Sun,
W. Eli Lilly and Co.: US, 2009.
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1355

Figure 11. Decision algorithm for reactor design and solids handling of the reaction in Scheme 1.
leading to severe plugging. The use of acoustic irradiation
reduced the maximum effective particle size of the salt
byproduct and thus prevented bridging in PFA capillary-based
reactors. It is likely that the magnitude of the acoustic forces
exceeded the hydrodynamic and particle-to-particle attractive
forces that led to bridging. Slurries of particles, the largest of
which had aspect ratios of 28, flowed without severe plugging
when acoustic irradiation was applied. In a system that did not
include ultrasound, plugging was observed and the aspect ratio
of the largest particles was estimated to be 8.9. Constriction
was observed to be quite severe in silicon nitride coated devices,
and switching to a fluoropolymer capillary did not eliminate
the problem. A composite of NaCl and product was formed on
the surfaces that were not submerged in the ultrasonic bath and
exposed to acoustic irradiation. When the flow rate was
increased, the pressure increased at a slower rate. Fitting this
data to a constriction model showed a correlation between the
rate of wall deposition and flow rate and suggested that a further
increase of the flow rate could potentially eliminate clogging
due to constriction. Moreover, the presence of acoustic irradia-
tion appeared to enhance the reaction rate, which may be the
result of a temperature increase based on the energy emitted
from cavitation and/or a result of enhanced mixing. Based on
the results reported herein, general guidelines can be prescribed
to handle salt byproducts during reactions in microsystems. It
is important to limit the particle sizes to aspect ratios of ∼9
either by removing them from the reactor before they grow to
that size or by imposing an external force such as acoustic
irradiation. The filtration of solvents and reagent mixtures can
also be important in eliminating potential bridging. Additionally,
increasing the flow velocity can help to mitigate constriction.
Otherwise, an estimation of the constriction rate can be useful
in determining how often solvents need be injected to remove
deposits.
Although the handling of solids in microsystems offers new
opportunities for chemical reactions that were previously
difficult or not attainable, moving forward is not without
challenges. Many of the routes to APIs involve solids, which
can participate as reactants, products, and catalysts, in addition
to salt byproducts. Strategies for handling other types of solids
will prove useful for developing efficient flow-based syntheses.
Furthermore, the need to operate in both the micro- and
macroscale reactors warrants a deeper understanding of solids
handling. Additionally, deposition and growth on peripheral
equipment, reactors, instruments, and transfer tubing surfaces
is an important consideration that needs to be addressed when
flowing salts are suspended in organic solvents. Strategies and
techniques to remove such deposits will find utility on all scales.
5
. Experimental Details
5.1. Microreactor Integration and Control. Standard
machining techniques were employed to produce a compression
chuck (316 Stainless Steel) with standard #10-32 coned-bottom
fluidic ports for each microreactor. The microreactor inlets and
outlet were compressed in this chuck (see Figure 1b) using
Kalrez o-rings for chemical compatibility, and the chuck was
cooled (to 20 °C) using a Thermo Scientific NESLAB RTE-7
refrigerating bath. A Kapton flexible heater (KHLV series, 28
V) and an Omega CN9000 series PID controller were used to
control the temperature in the heated section of the microchan-
1
4
nel. All fluidic connections were made using / in.-28 PTFE
1
fittings and PFA tubing ( / in. o.d., 500 µm i.d., IDEX Corp.).
16
The pressure transducer was excited using a 10 V power supply,
1
356
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development

and the output was recorded using a National Instruments USB
219 data acquisition card. Images were captured with a high-
Analysis of particle sizes exiting the reactors was made
possible with a Malvern Mastersizer 2000 laser diffractometer
fitted with a Hydro 2000 µP cell. In all cases, samples were
collected under argon and diluted by a volumetric factor of 2
with 1,4-dioxane. The analysis was performed by injecting
samples ranging from 0.2 to 0.8 mL into the 14.4 mL reservoir
of the diffractometer cell that was filled with 1,4-dioxane.
5.2. Workup and Yields. Samples that were collected under
argon were diluted with equal volumes of ethyl acetate and
water and mixed vigorously. The organic phase was separated,
filtered through a short plug of silica gel, and analyzed by GC.
Yield and conversion were determined on the basis of the peak
area, relative to the internal standard. In two examples, the water
phase was extracted with ethyl acetate a total of three times
and the organic phases were combined and concentrated. The
crude material was then purified by column chromatography
(Biotage Isolera, 25 g SNAP column, hexanes and 0-20% ethyl
acetate). Isolated yields were found to be in excellent agreement
with the GC yields. See the Supporting Information for reaction
details.
9
speed color CCD camera (JAI CV-S3200 series). In the case
of capillary PFA reactors, the tubing was submerged in a VWR
ultrasonic bath (Model 50HT) filled with deionized water. The
bath temperature was monitored via a thermocouple and
maintained with a Waage immersion heater controlled by a
J-KEM Scientific Gemini PID controller.
5.2. Reagents and Analytical Data. All experiments were
carried out using reagent grade solvents, and all solutions were
prepared under argon atmospheres. XPhos and the XPhos
50,68
precatalyst were prepared according to literature procedures.
Reaction solutions were prepared in screw-cap, oven-dried
volumetric flasks. For the clogging experiments, two solutions
were prepared. The first solution contained the base (NaOt-
Bu) and was prepared in 1,4-dioxane. The second solution
contained the aryl chloride (4-chloroanisole), amine (aniline),
internal standard (biphenyl), and catalyst (XPhos precatalyst)
and was also prepared in 1,4-dioxane. Reagents that were solids
(biphenyl, XPhos precatalyst, and NaOt-Bu) were added to the
volumetric flasks that were then evacuated and refilled with
argon. This process was repeated a total of 3 times. Liquid
reagents were added by syringe, and the solutions were made
up to the desired volume with 1,4-dioxane. The base solution
was then taken up into a syringe and filtered through a PTFE
membrane filter (0.4 µm porosity) into a second flask that had
previously been oven-dried and purged with argon. The two
solutions were then loaded into syringes and fitted in the syringe
pumps for the experiments.
Acknowledgment
This work has been funded by the Novartis-MIT Center for
Continuous Manufacturing.
Supporting Information Available
Discussion of the acoustic waveform with related figures.
Discussion of the X-ray diffraction and GCMS analysis of the
wall deposit with related figures. Discussion of the influence
of acoustics on the reaction by particle analysis and a related
figure. Discussion of the general procedure for Pd-catalyzed
amination reactions with related NMR spectra. Video showing
the gradual clogging of the microreactor. This material is
available free of charge via the Internet at http://pubs.acs.org.
1
13
All compounds were characterized by H NMR, C NMR,
1
IR spectroscopy, and elemental analysis. Copies of the H and
1
3
C spectra can be found at the end of the Supporting
13
Information. All C NMR spectra were obtained with 1H
decoupling. All GC analyses were performed on a Agilent 6890
gas chromatograph with an FID detector using a J & W DB-1
column (10 m, 0.1 mm i.d.).
Received for review June 2, 2010.
OP100154D
(
68) Biscoe, M. R.; Fors, B. P.; Buchwald, S. L. J. Am. Chem. Soc. 2008,
30, 6686–6687.
1
Vol. 14, No. 6, 2010 / Organic Process Research & Development
•
1357