Selectivity gains and energy savings for the industrial phenyl boronic acid process using micromixer/tubular reactors

Article abstract of DOI:10.1021/op0341768

An industrial production-scale process currently conducted at Clariant site/Frankfurt, termed Phenyl Boronic Acid Process from here on, was investigated at laboratory-scale using two micromixer/tubular reactor configurations, being equipped with either a glass interdigital (for details see (1) Ehrfeld, W.; Golbig, K.; Hessel, V.; Loewe, H.; Richter, T. Characterization of mixing in micromixers by a test reaction: single mixing units and mixer arrays. Ind. Eng. Chem. Res. 1999, 38, (3), 1075-1082; (2) Hessel, V.; Hardt, S.; Loewe, H.; Schoenfeld, F. Laminar mixing in different interdigital micromixers - Part I: Experimental characterization. AIChE J. 2003, 49, 566-577; (3) Hardt, S.; Schoenfeld, F. Laminar Mixing in Different Interdigital Micromixers - Part 2: Numerical Simulations. AIChE J. 2003, 49, 578-584; (4) Herweck, T.; Hardt, S.; Hessel, V.; Loewe, H.; Hofmann, C.; Weise, F.; Dietrich, T.; Freitag, A. Visualization of Flow Patterns and Chemical Synthesis in Transparent Micromixers. In Topical Conference Proceedings; IMRET 5, 5th International Conference on Microreaction Technology, AIChE Spring National Meeting; Matlosz, M., Ehrfeld, W., Baselt, J. P., Eds.; Springer-Verlag: Berlin, 2001; pp 215-229) or a steel split-recombine mixer (for details see Schoenfeld, F.; Hessel, V.; Hofmann, C. An Optimised Split-and-Recombine Micro Mixer with Uniform 'Chaotic' Mixing. Lab Chip 2004, 4, 65-69). The best yield of the microreactor investigations was 89%, exceeding the performance of the industrially employed stirred-tank process by nearly 25%. Moreover, the total amount of side and consecutive products was decreased from 10-15% to 5-10%. In addition to these yield improvements, the energy expenditure of the microreactor processing was also notably reduced, first of all because of carrying out the reaction at favorable ambient temperature. In contrast, the former batch process had to be carried out at cryogenic temperatures of -35°C. Even at temperatures as high as 50°C, a high selectivity was maintained when using the microreactor. As a further cause for energy savings, the higher purity of the product eliminated the need for distillation; in total therefore, only one heating-cooling cycle with reduced temperature difference was required for the microreactor process in contrast to the three cycles with large temperature differences used in conventional processing. Facing production and process liability issues, a specially made pilot-scale configuration was constructed entirely out of stainless steel components, among them a split-recombine caterpillar mixer having larger internal fluid channels than the interdigital glass mixer. The new configuration allowed one to perform scale-up studies at throughputs as high as 10 L/h.

Full text of DOI:10.1021/op0341768

Organic Process Research & Development 2004, 8, 511−523
Selectivity Gains and Energy Savings for the Industrial Phenyl Boronic Acid
Process Using Micromixer/Tubular Reactors
V. Hessel,*,† C. Hofmann,† H. Lo¨we,† A. Meudt,‡ S. Scherer,‡ F. Scho¨nfeld,† and B. Werner†
Institut fu¨r Mikrotechnik Mainz GmbH, Carl-Zeiss-Strasse 18-20, D-55129 Mainz, Germany, and Clariant,
D-65926 Frankfurt am Main, Germany
Abstract:
mixer. The new configuration allowed one to perform scale-up
studies at throughputs as high as 10 L/h.
An industrial production-scale process currently conducted at
Clariant site/Frankfurt, termed Phenyl Boronic Acid Process
from here on, was investigated at laboratory-scale using two
micromixer/tubular reactor configurations, being equipped with
either a glass interdigital (for details see (1) Ehrfeld, W.; Golbig,
K.; Hessel, V.; Lo1we, H.; Richter, T. Characterization of mixing
in micromixers by a test reaction: single mixing units and mixer
arrays. Ind. Eng. Chem. Res. 1999, 38, (3), 1075-1082; (2)
Hessel, V.; Hardt, S.; Lo1we, H.; Scho1nfeld, F. Laminar mixing
in different interdigital micromixers - Part I: Experimental
characterization. AIChE J. 2003, 49, 566-577; (3) Hardt, S.;
Scho1nfeld, F. Laminar Mixing in Different Interdigital Micro-
mixers - Part 2: Numerical Simulations. AIChE J. 2003, 49,
578-584; (4) Herweck, T.; Hardt, S.; Hessel, V.; Lo1we, H.;
Hofmann, C.; Weise, F.; Dietrich, T.; Freitag, A. Visualization
of Flow Patterns and Chemical Synthesis in Transparent
Micromixers. In Topical Conference Proceedings; IMRET 5, 5th
International Conference on Microreaction Technology, AIChE
Spring National Meeting; Matlosz, M., Ehrfeld, W., Baselt, J.
P., Eds.; Springer-Verlag: Berlin, 2001; pp 215-229) or a steel
split-recombine mixer (for details see Scho1nfeld, F.; Hessel, V.;
Hofmann, C. An Optimised Split-and-Recombine Micro Mixer
with Uniform ‘Chaotic’ Mixing. Lab Chip 2004, 4, 65-69). The
best yield of the microreactor investigations was 89%, exceeding
the performance of the industrially employed stirred-tank
process by nearly 25%. Moreover, the total amount of side and
consecutive products was decreased from 10-15% to 5-10%.
In addition to these yield improvements, the energy expenditure
of the microreactor processing was also notably reduced, first
of all because of carrying out the reaction at favorable ambient
temperature. In contrast, the former batch process had to be
carried out at cryogenic temperatures of -35 °C. Even at
temperatures as high as 50 °C, a high selectivity was maintained
when using the microreactor. As a further cause for energy
savings, the higher purity of the product eliminated the need
for distillation; in total therefore, only one heating-cooling cycle
with reduced temperature difference was required for the
microreactor process in contrast to the three cycles with large
temperature differences used in conventional processing. Facing
production and process liability issues, a specially made pilot-
scale configuration was constructed entirely out of stainless steel
components, among them a split-recombine caterpillar mixer
having larger internal fluid channels than the interdigital glass
1. Introduction
During the past years, aryl and alkyl boron compounds
have become very versatile synthetic building blocks which
can be converted, e.g. via Suzuki coupling, to a multitude
of economically very valuable fine chemicals in the fields
of pharmaceuticals and agrochemicals6 (see also the original
citations7 describing synthesis routes). Most prominent
among such species, aryl and alkyl boronic acids have
attained a virtually exponential increase in applications for
drug synthesis. In addition, diaryl boronic acids gain attention
due to their use as cocatalysts in olefin polymerisation or as
precursors for Suzuki couplings, by which both aryl moieties
can be transferred. Consequently, the conversion of aryl and
alkyl Grignard reagents to give aryl and alkyl boron
compounds is described by many publications. Given a strict
control of reaction conditions, which however need to be
optimised for each individual case, good yields can be
obtained. Due to the latter reason, a drawback of the present
processing routes is the large number of side products which
are formed in amounts that depend strongly on the applied
conditions.6 Following the hydrolysis of the reaction product
mixtures, homo-coupling reactions are found yielding the
corresponding biaryls and bialkyls. Moreover, boronic acids,
triaryl or trialkyl boranes, and tetraaryl or tetraalkyl boronates
(1) Ehrfeld, W.; Golbig, K.; Hessel, V.; Lo¨we, H.; Richter, T. Characterization
of mixing in micromixers by a test reaction: single mixing units and mixer
arrays. Ind. Eng. Chem. Res. 1999, 38, (3), 1075-1082.
(2) Hessel, V.; Hardt, S.; Lo¨we, H.; Scho¨nfeld, F. Laminar mixing in different
interdigital micromixers - Part I: Experimental characterization. AIChE J.
2003, 49, 566-577.
(3) Hardt, S.; Scho¨nfeld, F. Laminar Mixing in Different Interdigital Micro-
mixers - Part 2: Numerical Simulations. AIChE J. 2003, 49, 578-584.
(4) Herweck, T.; Hardt, S.; Hessel, V.; Lo¨we, H.; Hofmann, C.; Weise, F.;
Dietrich, T.; Freitag, A. Visualization of Flow Patterns and Chemical
Synthesis in Transparent Micromixers. In Topical Conference Proceedings;
IMRET 5, 5th International Conference on Microreaction Technology,
AIChE Spring National Meeting; Matlosz, M., Ehrfeld, W., Baselt, J. P.,
Eds.; Springer-Verlag: Berlin, 2001; pp 215-229.
(5) Scho¨nfeld, F.; Hessel, V.; Hofmann, C. An Optimised Split-and-Recombine
Micro Mixer with Uniform ‘Chaotic’ Mixing. Lab Chip 2004, 4, 65-69.
(6) Koch, M.; Wehle, D.; Scherer, S.; Forstinger, K.; Meudt, A.; Hessel, V.;
Werner, B.; Lo¨we, H. Verfahren zur Herstellung von Aryl- und Alkyl-Bor-
Verbindungen in Mikroreaktoren. EP 1285924 (20.07.02), 21.08.01, 2001.
(7) (a) Synthesis of phenyl boronic acid: Bean, F. R.; Johnson, J. R. Derivatives
of Phenylboric Acid, Their Preparation and Action Upon Bacteria. II
Hydroxyphenylboric Acids. J. Am. Chem. Soc. 1932, 54, 4415-4425. (b)
4-Ethoxyphenylboronic acid: Branch, G. E. K.; Yabroff, D. L.; Bettman,
B. The Dissociation Constants of the Chlorophenyl and Phenetyl Boric
Acids. J. Am. Chem. Soc. 1934, 56, 937-941.
* Author for correspondence. E-mail: hessel@imm-mainz.de.
† Institut fu¨r Mikrotechnik Mainz GmbH.
‡ Clariant.
10.1021/op0341768 CCC: $27.50 © 2004 American Chemical Society
Published on Web 03/04/2004
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
511

are formed. With the exception of the latter class of charged
compounds, expeditious purification procedures are thus
required, decreasing the yield and increasing the production
costs of the target product.
temperature on a desired level, even when large heats of
reaction are released.9-16 Moreover, by using micromixers
fast liquid mixing down to the millisecond range can be
achieved,2,3,17 thereby reducing the level of reaction with
concentrations that have not been equilibrated, i.e., where
an excess of one reactant may induce a secondary reaction.
In a similar way, microreactors have been recently success-
fully applied to many organic reactions, e.g. to the Suzuki
coupling,18 the Wittig reaction,19 the Stork enamine reac-
tion,20 and the fluorination of aromatics.21,22 Patent literature
shows an increasing number of references on industrial
microreactor-based chemical processes, containing experi-
mental reports, e.g., on the acid cleavage in the framework
of the Hock process, a chloroacetic acid process, and azo
pigment formation (see ref 9b for more details and original
citations).
If aryl or alkyl boronic esters are the target product, only
one equivalent of Grignard reagent needs to be reacted with
the boronic acid precursor.6 However, via di-, tri- and tetra-
fold substitutions by the metallo-organic reactant, biaryls/
bialkyls, boronic acids, boranes, and even boronates are
generated. If nonoptimal reaction conditions are run, the yield
can be decreased in a dramatic manner. In many cases,
purification of the crude products becomes a severe problem
as well. The same holds for the synthesis of boronic acids,
boranes, and boronates.
To decrease the amount of the before-mentioned side
products, operation at low temperature is mandatory to
prevent the primary products, obtained by the primary
elemental reaction, from decomposition into the free boronic
esters or halogenides.6 The latter species reacts with the
metallo-organic reactant, in competition with the noncon-
verted boronic-acid precursor, and thereby decreases the
yield. The same holds for higher arylated or alkylated
products.
Good reaction conditions are found certainly below -25
°C; however, ideal performance with low and zero side
products is produced only below -35 and -55 °C, respec-
tively.6 These temperatures cannot be achieved using simple
cooling methods or brine cooling; rather they demand large
energy expenditure and thus generate large costs. In conjunc-
tion with the need for an energy supply for refluxing during
the Grignard reagent generation and for distillation of the
crude product for removal of the solvent, several unwanted
cooling-heating cycles follow each other, until the purified
product is obtained by filtration or extraction at room
temperature. Consequently, an uneconomic process is as-
sociated with large manufacturing costs.
2. The Phenyl Boronic Ester Process
The Phenyl Boronic Ester Process is one process among
a class of similar reactions, all performing one or multiple
arylation or alkylation steps on a trivalent boron precursor
by use of a metallo-organic reactant. The boron precursor
may contain halide, C1-C5 alkoxy, N,N(C1-C5 alkylamino),
or C1-C5 alkyl-thio moieties. The metallo-organic reactant
is a magnesium-halide core substituted by a large variety
of moieties, among them linear and branched C1-C6 alkyl,
phenyl or substituted phenyl groups, hetero-aryls, and many
others.
The phenyl boronic ester process consists of a fast liquid
reaction involving contacting of two reactants, phenylmag-
nesium bromide and boronic acid trimethoxy ester, each
being dissolved in tetrahydofuran (THF) to yield a liquid
mixture. The arylation step is followed by hydrolysis to give
the final product phenyl boronic acid. This product and by-
Other drawbacks of the state-of-the-art processing arise
from the need for an excess of the expensive boronic ester
precursor.6 Even then, the reaction proceeds rather slowly,
since the Grignard reagent has to be added dropwise and
hence very slowly to remove heat fast enough to avoid hot
spots. As a result, processing time is considerably longer
than kinetics would predict. Thus, it is also usually preferred
to work with rather dilute solutions. All these measures
reduce the process performance.
The goal is, therefore, to develop a process using the same
raw materials as for the conventional batch process which
operates at temperatures exceeding -30 °C and at favorably
high reactant concentrations to give high yields of pure
products with low contents of side products.
(9) (a) Ehrfeld, W.; Hessel, V.; Lo¨we, H. Microreactors; Wiley-VCH: Wein-
heim, 2000. (b) Hessel, V.; Hardt, S., Lo¨we, H. Chemical Micro Process
Engineering; Wiley-VCH: Weinheim, 2004.
(10) Hessel, V.; Lo¨we, H. Micro Chemical Engineering: Components - Plant
Concepts - User Acceptance: Part I. Chem. Eng. Technol. 2003, 26, 13-
24.
(11) Ja¨hnisch, K.; Hessel, V.; Lo¨we, H.; Baerns, M. Chemistry in Microstructured
Reactors. Angew. Chem., Int. Ed. 2004, 43, (4), 406-446.
(12) Jensen, K. F. Microreaction Engineering - Is Small Better? Chem. Eng.
Sci. 2001, 56, 293-303.
(13) Jensen, K. F. Smaller, Faster Chemistry. Nature 1998, 393, 735-736.
(14) Gavriilidis, A.; Angeli, P.; Cao, E.; Yeong, K. K.; Wan, Y. S. S. Technology
and Application of Microengineered Reactors. Trans. Inst. Chem. Eng. 2002,
80/A, 3-30.
(15) Fletcher, P. D. I.; Haswell, S. J.; Pombo-Villar, E.; Warrington, B. H.; Watts,
P.; Wong, S. Y. F.; Zhang, X. Microreactors: Principle and Applications
in Organic Synthesis. Tetrahedron 2002, 58, 4735-4757.
(16) Haswell, S. T.; Watts, P. Green Chemistry: Synthesis in Microreactors.
Green Chem. 2003, 5, 240-249.
(17) Lo¨we, H.; Ehrfeld, W.; Hessel, V.; Richter, T.; Schiewe, J. Micromixing
Technology. In Topical Conference Proceedings; IMRET 4, 4th International
Conference on Microreaction Technology, AIChE Spring National Meeting,
March 5-9, 2000, Atlanta, GA; American Institute of Chemical Engi-
neers: New York, NY, 2000; pp 31-47.
(18) Skelton, V.; Greenway, G. M.; Haswell, S. J.; Styring, P.; Morgan, D. O.
Micro Reactor Synthesis: Synthesis of Cyanobiphenyls Using a Modified
Suzuki Coupling of an Aryl Halide and Aryl Boronic Acid. Microreaction
Technology. In Topical Conference Proceedings; IMRET 5, 5th International
Conference on Microreaction Technology, AIChE Spring National Meeting;
Matlosz, M., Ehrfeld, W., Baselt, J. P., Eds.; Springer-Verlag: Berlin, 2001;
pp 235-242.
For this purpose, a micromixer/tube reactor configuration
can be used.6,8 Continuous microchannel processing therein
allows exact control over residence time andkeeps the
(8) Hessel, V.; Lo¨we, H.; Hofmann, C.; Scho¨nfeld, F.; Wehle, D.; Werner, B.
Process Development of a Fast Reaction of Industrial Importance Using a
Caterpillar Micromixer/Tubular Reactor Setup. In Topical Conference
Proceedings; IMRET 6, 6th International Conference on Microreaction
Technology, AIChE Spring National Meeting, March 11-14, 2002, New
Orleans, LA; American Institute of Chemical Engineers: New York, NY,
2002; pp 39-54.
512
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Scheme 1. Series of elementary reactions of the Phenyl Boronic Acid Process and subsequent hydrolysis
products are abbreviated as follows (as given in Scheme 1):
yield of the target product and which generate purification
problems. The two reactants, phenylmagnesium bromide R1
and boronic acid trimethoxy ester R2, form a charged
intermediate I1, phenyltrimethoxyboronium salt. This reacts
to a second intermediate I2, the phenyl boronic acid
dimethoxy ester. This intermediate is then hydrolysed to give
the target product phenyl boronic acid P1.
A number of additional reactions decrease the selectivity
and yield. First, the reactants phenylmagnesium bromide and
boronic acid trimethoxy ester are labile to moisture. Hy-
drolysis results in the formation of the side products, reactants
benzene S1 and boric acid S5. By more complex pathways,
not depicted in detail in Scheme 1, three other known side
products are generated, phenol (S2), chlorobenzene (S3), and
biphenyl (S4). Second, diphenyl boronic acid methoxy ester,
P1: phenyl boronic acid
(product 1)
(reactant 1)
(reactant 2)
R1: phenylmagnesium bromide
R2: boronic acid trimethyl ester
I1: phenyl trimethoxy boronium magnesium bromide
(intermediate 1)
(intermediate 2)
I3: diphenyl dimethoxy boronium magnesium bromide
(intermediate 3)
I4: triphenyl methoxy boronium magnesium bromide
(intermediate 4)
I2: phenyl boronic ester dimethyl ester
S1: benzene
(side product 1)
(side product 2)
S2: phenol
(19) Skelton, V.; Greenway, G. M.; Haswell, S. J.; Styring, P.; Morgan, D. O.;
Warrington, B. H.; Wong, S. Micro Reaction Technology: Synthetic
Chemical Optimisation Methodology of Wittig Synthesis Enabling a Semi-
Automated Microreactor for Combinatorial Screening. In Topical Conference
Proceedings; IMRET 4, 4th International Conference on Microreaction
Technology, AIChE Spring National Meeting, March 5-9, 2000, Atlanta,
GA; American Institute of Chemical Engineers: New York, NY, 2000; pp
78-88.
(20) Sands, M.; Haswell, S. J.; Kelly, S. M.; Skelton, V.; Morgan, D.; Styring,
P.; Warrington, B. The investigation of an equilibrium dependent reaction
for the formation of enamines in a microchemical system. Lab Chip 2001,
1, 64-65.
S3: chlorobenzene
S4: biphenyl
(side product 3)
(side product 4)
S5: boric acid
(side product 5)
C1: diphenyl boronic acid
C2: triphenylborane
(consecutive product 1)
(consecutive product 2)
C1E: diphenyl borinic acid methyl ester
(ester of consecutive product 1)
(21) Chambers, R. D.; Spink, R. C. H. Microreactors for elemental fluorine.
Chemical Communications 1999, (10), 883-884.
(22) Ja¨hnisch, K.; Baerns, M.; Hessel, V.; Ehrfeld, W.; Haverkamp, W.; Lo¨we,
H.; Wille, C.; Guber, A. Direct fluorination of toluene using elemental
fluorine in gas/liquid microreactors. J. Fluorine Chem. 2000, 105, (1), 117-
128.
The total process actually consists of a number of
elemental reactions which also give rise to side products,
especially to higher arylated boron species which reduce the
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
513

Figure 1. Flow sheet of the laboratory-scale microreactor configuration of the Phenyl Boronic Acid Process, equipped with an
interdigital micromixer.
C1E (the Ester of C1), and triphenyl borane, C2, are gener-
ated as consecutive products from I2 by reaction with one
or two equivalents phenylmagnesium bromide, respectively.
The intermediates I1 and I3, however, precipitate in the
course of the reaction. These species are dissolved by
subsequent reaction to the product P1 and the consecutive
product C1 (according to Scheme 1).
A low content of diphenyl boronic acid C1 is, in
particular, important for an efficient product separation of
phenyl boronic acid. C1 and all the side products remain in
solution after the end of the process and cannot be separated
simply by filtration.
separate reactant solutions were dosed into the flask under
the stirring conditions described above.
The industrial batch production plant cannot be described
here, as this relates to undisclosed internal know-how of
Clariant.
(b) Laboratory-Scale Configuration. Two reservoirs con-
tain the anhydrous THF solutions of the reactants. A third
reservoir contains pure anhydrous THF for purging of the
processing line and, possibly, for on-line dilution of the two
reactant streams (see Figure 1). Double-piston pumps (KP
2000, Desaga Sarstedt-Gruppe, Wiesloch, Germany) with
flow-indicated control fed the solutions from the reservoir
to a mixer/tubular reactor. A triangular interdigital micro-
mixer made of a specialty glass (see Table 1) was applied
for most of the experiments, which were carried out at IMM.
In one case a caterpillar minimixer was used; in a few further
cases, follow-up investigations were performed at the Clariant
site with a steel slit interdigital micromixer (see the Results
and Discussion section). The use of a conventional tube of
mesoscale diameter (2.4 mm i.d.), rather than a microchannel
reaction device such as a microheat exchanger, was manda-
tory as precipitation inevitably occurs during the reaction.
The whole reaction unit, mixer and tube, was submerged
into a water heating bath which is part of a thermostat sensing
and controlling the temperature. The product is collected by
pouring the reactant stream out of the tube into a beaker.
(c) Pilot-Scale Configuration. The pilot-scale configura-
tion was developed with aim to have a scaled-out version of
the laboratory configuration. Ideally, one would have used
purely the numbering-up concept for scale-up, i.e., the
repetition of many parallel interdigital units and the same
number of tubes attached. However encountering the known
fouling problems and the accordingly severely limited
operation time of the individual interdigital mixer units, it
3. Experimental Section
Recording of Sreamlines (Flow-Pattern and Fouling
Visualizations). Mixing was recorded by using digital re-
cording with an optical microscope Stemi 2000-C, Zeiss Go¨t-
tingen, Germany (magnification 13× ... 100×) equipped with
a video camera 3CCD Color Video Camera Power HAD,
DX-950P, Sony, Tokyo, Japan (resolution 768 × 576 × 24).
Apparatus. (a) Industrial Laboratory and Production
Batch Configurations. The laboratory batch configuration at
the Clariant site was based on a stirred four-necked glass
2-L flask (with the German standard ground joints 2 × 14.5;
1 × 29; 1 × 45) processing a liquid volume of about 1.5 L.
The impeller had a crescent-shaped Teflon flat blade (blade
length and width 10 and 2 cm, respectively) and a clearance
(distance to flask bottom) of 1 mm (no baffles were applied).
The impeller rotational speed was 60-150 rpm. The glass
flask was cooled, if required, and the gas chamber in the
flask above the stirred solution was purged with nitrogen.
About 10-20% of the total volume of the solvent was filled
into the flask before starting the reaction. From then, two
514
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Figure 2. Flow sheet of the laboratory-scale microreactor configuration of the Phenyl Boronic Acid Process.
Table 1. Specifications and dimensions of the glass triangular interdigital micromixer
triangular interdigital micromixer:
characteristic properties
specifications
mixer construction material
Foturan, a commercial specialty glass
made by Schott Desag AG (Gru¨nenplan, Germany) and
licensed to mgt mikroglas technik Mainz AG (Mainz, Germany).
SiO2 ) 75-85%, Li2O ) 7-11%, K2O ) 3-6%,
Al2O3 ) 3-6%, Na2O ) 1-2%, ZnO ) 0-2%,
Sb2O3 ) 0.3%, Ag2O ) 0.1%, CeO2 ) 0.015%.
60 µm; 150 µm; 30 µm
mixer feed channel width and depth;
fin walls separating the feed channels
total number of feed channels
triangular chamber: initial width; focused width;
depth; focusing length; mixing length;
focusing angle; focusing factor
2 × 15
3.25 mm; 500 µm; 150 µm; 8 mm; 19.4 mm; 20°; 6
device outer dimensions: length, width, and
thickness (bonded glass plates only, no frame)
76 mm; 26 mm; 2.3 mm
stood to reason that compromises had to be made. Following
a procedure which turned out to be successful for a Merck
microreactor-based production plant,23 the micromixer was
replaced by a minimixer of larger internal dimensions. A
so-called caterpillar mixer was chosen, since this device relies
on a split-recombine principle, which on ideal flow condi-
tions (low Re number) gives multilamination patterns, i.e.,
preserves the fluid dynamics utilised by the interdigital
micromixer.5 Only recently, it was found that deviations
occur from this ideal due to imperfections of the mixer
geometry and that under the conditions described in this
section secondary-flow patterns may superpose the multil-
amination pattern.5 From former studies concerning calcium
carbonate precipitation, which is an instantaneous forced
precipitation, it was known that this mixer has a better
performance for fouling-sensitive processes than interdigital
micromixers.24
Due to improved construction (all components being
solidly mounted on a rig) and the choice of the hard steel
material (instead of glass and polymer) the pilot-scale
configuration is more robust and more user-friendly. By
means of a 5/2-way valve four tubular reactors acting as
delay loops can be alternately set into operation. In addition,
one of the tubes was replaced by another, yielding finally a
choice of five tubes which were operated. These tubes differ
in inner diameter (0.7, 3.8, 4.8, 9.3, and 21.2 mm; all at a
length of 1000 mm) and hence residence time. The whole
reaction unit, including mixer and the four tubes, are
submerged into a cylinder completely filled with a cooling/
heating medium attached to a thermostat (see Figure 2).
Microchannel Devices. (a) Slit and Triangular Inter-
digital Micromixers. As the micromixer, initially a triangular
interdigital glass device was applied (see Table 1) which was
(24) Schenk, R.; Donnet, M.; Hessel, V.; Hofmann, C.; Jongen, N.; Lo¨we, H.
Suitability of Various Types of Micromixers for the Forced Precipitation
of Calcium Carbonate.In Topical Conference Proceedings; IMRET 5, 5th
International Conference on Microreaction Technology, AIChE Spring
National Meeting; Matlosz, M., Ehrfeld, W., Baselt, J. P., Eds.; Springer-
Verlag: Berlin, 2001; pp 489-498.
(23) Krummradt, H.; Kopp, U.; Stoldt, J. Experiences with the Use of
Microreactors in Organic Synthesis. In Topical Conference Proceedings;
IMRET 3, 3rd International Conference on Microreaction Technology,
AIChE Spring National Meeting; Ehrfeld, W., Ed.; Springer-Verlag: Berlin,
2000; pp 181-186.
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
515

Table 2. Specifications and dimensions of the steel slit interdigital micromixer
slit interdigital micromixer:
characteristic properties
specifications
mixer construction material
housing: stainless steel (1.4571; SS 316 Ti)
mixer inlay: nickel on copper base
40 µm; 300 µm; 30 µm
mixer feed channel width and depth;
fin width (separating the channels)
total number of feed channels
slit depth in steel housing
tubing attached to slit: diameter
device outer dimensions: diameter and height
2 × 25
60 µm
500 µm
20 mm; 16.5 mm
fluids results in a repeated splitting, reshaping, and recom-
bination, thereby decreasing the lamellae width and generat-
ing a multilamellae pattern.
The caterpillar minimixer is a commercial product of
IMM (CPMM-R1200). Recently, a simulation study showed
that at high volumetric flow rates (and consequently high
Re numbers) secondary-flow patterns become active and de-
viations from the ideal multilamination pattern occur due to
shear and friction forces.5 As a consequence, a considerably
improved design has been proposed and realised (a second-
generation caterpillar minimixer) which is especially suited
for mixing of highly viscous solutions,5 which was, however,
not used for the synthesis described here (see Table 3).
Experimental Protocols. (1) ConVentional Batch Syn-
thesis. THF solutions (c ) 0.5 mol/L) of the reactants
phenylmagnesium bromide and boronic acid trimethoxy ester
are dosed simultaneously in a round flask which is filled
with solvent (10-20% of the total volume), held at a
temperature of 20 °C. The speed of dosing is chosen in such
a way that the temperature is kept at 20 ( 5 °C. For a total
volume of 1.5 L, for example, 22 min are needed. A total
yield of all products (including side products) amounts to
83% (see Table 4). By HPLC analysis, a content of diphenyl
boronic acid of 13.8% and a content of triphenyl borane of
1.1% is determined.
(2) Laboratory-Scale Continuous Microreactor Synthesis.
Anhydrous THF solutions (c ) 0.5 mol/L, puriss. p.a. >
99.5%, Aldrich no. 87368, Taufkirchen/Germany) of the
reactants phenylmagnesium bromide (2 mol/L in THF,
Aldrich no. 22,444-8, Taufkirchen/Germany) and boronic
acid trimethyl ester (99.5%, Aldrich no. 44,399-9, Taufkirchen/
Germany) were filled in vessels (of a volume of typically 1
L), stored under dry nitrogen, and cooled there to reaction
temperature. From a third reservoir, the whole laboratory-
scale apparatus was flushed with anhydrous THF. By use of
double piston pumps, the reactant solutions were fed from
the reservoirs into a triangular interdigital micromixer which
was connected to a PTFE tube (i.d.: 2 mm) of 70 cm length.
Both the mixer and the tube were inserted into a cooling/
heating bath filled with water. The reaction solutions were
poured into a stirred beaker filled with water with a pH of
4.5 (50 mg water; 30-60 s insertion of reaction product
stream, time being dependent on volume flow, and weighing
of that charge; addition of 1.5 g concentrated H2SO4). THF
is removed from the solvent mixture by smooth distillation
under reduced pressure at 100 mbar. After cooling to 0 °C,
Figure 3. Identical multilamination flow patterns obtained at
low (100 mL/h, bottom image) and high (2000 mL/h, top image)
volumetric flow rate in the glass interdigital micromixer. The
imaging is achieved by using aqueous test solutions, one of them
containing acid blue as dye.
jointly developed by the Institut fu¨r Mikrotechnik Mainz
GmbH AG (IMM), Mainz/Germany and mgt mikroglas
technik Mainz, Mainz/Germany and is meanwhile a com-
mercial
pro-
duct of mgt mikroglas technik Mainz.2,17 This mixer relies
only on a multilamination flow pattern, undisturbed by eddies
and wakes (see Figure 3).17 This was regarded to be crucial
so as to keep fouling as low as possible. By means of geo-
metric focusing of the multilamellae stream, fast mixing is
achieved. The initial lamellae width of 90 µm is focused by
a factor of 6-15 µm. Glass was chosen as construction ma-
terial to enable simple visual inspection of the extent of foul-
ing within the device. This mixer is known to achieve liquid
mixing within about 100 ms for a mixing degree of 99%,
assuming water-like systems at room temperature and a dif-
fusion constant typical for organic molecules (10-9 m2/s).2,3
In later follow-up studies at the Clariant site, it was desired
to change the material from glass to steel. Accordingly, the
off-the-shelf product standard slit interdigital micromixer
(SSIMM) was chosen (see Table 2).1
(b) Caterpillar Minimixer. For the minimixer, a split-
recombine stainless steel mixer (the caterpillar mixer) was
used (see Figure 4)5 (see also ref 24). The name of this mixer
is derived from its characteristic 3D geometry consisting of
ascending and descending steps on the top and the bottom
of a minichannel. Their impact on the flow of two inserted
516
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Figure 4. Schematic of the caterpillar minimixer and SEM images of parts of the microchannel structure.
Table 3. Specifications and dimensions of the steel caterpillar minimixer
caterpillar minimixer:
characteristic properties
specifications
mixer construction material
stainless steel 1.4571 (SS 316 Ti)
number of plates needed to form minimixer channel
minimixer channel: initial width; maximum width
minimixer channel (both plates): initial depth; maximum depth
microstructure in one plate: initial depth; maximum depth
minimixer stage: length
2
1200 µm; 2400 µm
1200 µm; 1700 µm
600 µm; 850 µm
2400 µm
number of mixing stages
8
total length of caterpillar minimixer
19.2 mm
device outer dimensions
50 mm × 50 mm × 10 mm
Table 4. Yields obtained for the target product phenyl boronic acid and of the consecutive product diphenyl boronic acid of the
Phenyl Boronic Acid Process using a laboratory-scale and an industrial batch processing
T
[°C]
c
τ
Ycrude product
HPLC a/a purity
[%]
Ypure product
Ydiphenyl boronic acid
[%] HPLC a/a
reactor type
[mol/L]
[min]
[%]
[%]
laboratory-scale batch (1.5 L)
industrial batch
20
20
0.5
2.0
22
83.0
85.1
70.6
65.0
13.8
a
a
a
a
-
-
-
-
a Not disclosed.
the product is precipitated, isolated by filtration, and dried.
HPLC analysis was then performed.
(3) PreparatiVe Continuous Microreactor Synthesis. For
isolation of the product phenyl boronic acid in quantitative
amounts, the laboratory-scale configuration was used; how-
ever, the interdigital micromixer was exchanged against a
caterpillar minimixer. As such, much longer operational times
in the range of several 10s of minutes could be achieved
without plugging. The minimixer was connected to a PTFE
tube of 1/4′′ diameter with residence times equal to the ones
reported above. This was achieved by setting the volumetric
flow rate higher to 10 L/h with the additional benefit of
having advanced mixing quality due to higher flow velocities.
The other process parameters, concentration and temperature,
were chosen equal to those given in the protocol (2).
The reaction was carried out at four temperatures, -12,
0, 22, and 50 °C, and at four total volumetric flow rates,
200, 500, 1000, and 1500 mL/h. Since a constant length of
the PTFE tube was used, the variation of the volumetric flow
rate changed the residence time too. Thereby, the following
residence times were achieved: 200 mL/h: 40 s, 500 mL/h:
16 s, 1000 mL/h: 8 s, 1500 mL/h: 5 s. The yields of the
product and of the side and consecutive products were
determined by HPLC. The operation of the laboratory-scale
configuration was limited to a few minutes for each run, since
plugging by precipitation of the charged tetravalent inter-
mediates occurred.
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
517

Figure 5. Precipitate formation composed of intermediate products of the Phenyl Boronic Acid Process along the interfaces between
the multilaminated lamellae (Figure 5 is actually composed of two separately taken photos).
(4) Pilot-Scale Continuous Microreactor Synthesis. Op-
eration was carried out in a scaled-out pilot-scale configu-
ration with a steel caterpillar minimixer being connected to
four stainless steel tubes via a 5/2-way valve. The pilot-scale
synthesis was conducted at residence times of 5, 10, 26, and
120 s at temperatures of 10 and 20 °C. In a second set of
experiments, residence times were set to 1, 5, 10, 26 s at 30
and 40 °C. The concentration of the THF solution was 0.5
mol/L, equal to that in the protocols (2) and (3).
Thereby, it was possible to operate the configuration for
about 15 min. Thereafter, fouling was clearly visible in the
interdigital mixer by white stripes composed of precipitates
along the interfaces between the liquid lamellae (see Figure
5, see also ref 24). The 15-min processing time turned out
to be long enough to gather reliable data on process yield.
For reasons of fouling, the postprocessing step, the
hydrolysis of the product mixture, was not carried out using
microfluidic devices; it was experimentally found that the
introduction of the particle-containing solutions into a further
set of tiny feed channels, e.g. of a second micromixer,
inevitably resulted in clogging. In this context, it has to be
emphasized that based on the experience of the authors care
has to be taken when making general conclusions about the
impact of particle-containing solutions on fouling in mi-
crodevices.24 Naturally, the result is strongly dependent on
type, concentration, and size of the particles as well as
geometry and material of the microdevice.
Results of Laboratory-Scale Experiments. Table 5 presents
all results obtained with the laboratory-scale configuration
using the interdigital micromixer. The best result obtained
amounts to 83% at a reaction temperature of 22 °C and a
total volumetric flow rate of 1000 mL/h, which is equivalent
to 8 s operation. This exceeds the performance of the
formerly used batch production process by nearly 20%,
although about half of all measurements differ by less than
5% and only two of eleven measurements differ by more
than 10% from the maximum yield. Therefore, it can be
stated that within the operation range investigated, from -12
to 50 °C and from 500 to 1500 mL/h the yield is relatively
constant. In absolute terms the yield ranges from 65.5 to
83%. This is in contrast to the findings for the conventional
batch operation, as outlined in the Introduction section.
Although in general the yields turned out to be relatively
constant, there are exceptions, and it is also worthwhile to
discuss the small variations of the yield as a function of the
process parameters. Accordingly, in the following, the results
first are represented as yield vs temperature plots at a given
residence time to facilitate discussion of the correspond-
ingdependency. Thereafter, representation is done as yield
vs volumetric flow rate plots at a given temperature.
Temperature Variation. The yield vs temperature curves
are characterized by a nearly constant behavior, actually
4. Results and Discussion
Experiments Using a Laboratory-Scale Batch: Pro-
tocol (1). The results of the laboratory-scale and industrial
batch processing, both being performed at the Clariant site,
are given in Table 4. The yields of pure product for both
apparatuses are 71 and 65%, respectively. Reducing the
reactor volume and the concentration of the reactants leads
to increased yield of pure phenyl boronic acid. In both cases,
significant amounts of side and consecutive products are
generated, leading to large expenditure for purification by
distillation and filtration/extraction.
The beneficial effect of reducing the flask volume is a
result of the better stirring of smaller volumes and hence
the correspondingly enhanced mass and heat transfer. Albeit
the type of laboratory batch stirring chosen (for details see
Experimental Section) is typical for industrial process
development of such organic reactions, it has to be admitted
that probably still better mixing is achievable when using
more advanced and specialised batch-mixing equipment, i.e.,
it cannot be excluded that slightly better yields can be found
for batch operation.
Experiments Using the Laboratory-Scale Configura-
tion with the Interdigital Micromixers: Protocol (2).
Setting Fouling to an Acceptable LeVel. Typically, 2 mol/L
solutions of the reagents phenylmagnesium bromide and
boronic acid trimethoxy ester were applied for the Phenyl
Boronic Acid Process in the past, when being performed on
an industrial level. However, operation with such highly
concentrated solutions in the triangular interdigital mixer/
tubular reactor resulted in intense fouling and even clogging.
Therefore, it was attempted to dilute the solution to a
concentration which can be properly handled. It turned out
that a 0.5 mol/L reactant solution did fulfill this criterion.
518
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Figure 6. Yield of products as a function of temperature for the laboratory configuration with the interdigital glass micromixer.
Table 5. Yields obtained for the target product and side/consecutive products of the Phenyl Boronic Acid Process using the
laboratory-scale configuration with the triangular glass micromixer
temperature
flow rate
[mL/h]
-12 °C
0 °C
22 °C
50 °C
200
500
1000
1500
500
1500
500
1000
1500
500
1000
1500
P1
C1
S2
S1
S3
S4
51.9
2.1
10.2
0.0
0.0
0.0
72.0
2.7
0.9
1.0
0.0
0.0
80.7
1.2
7.9
5.2
0.0
0.9
65.5
1.3
9.7
1.9
0.0
0.0
80.6
1.3
6
2.4
1.9
0.9
78.4
1.3
4.8
1.7
0
79.6
1.9
5.7
1.5
8.9
0.7
83.2
1.4
5.8
0.9
1.9
1.2
78.7
1.0
5.0
1.3
1.3
0.8
77.4
3.6
7.7
0.4
1.9
1.2
82.1
1.7
7.0
0.7
1.9
1.2
80.1
1.2
6.1
1.6
1.3
0.9
0.8
showing a small increase in yield from -12 to 0 °C and
being identical at all volumetric flow rates. By flow imaging
using aqueous model solutions it is known that indeed
stacked fluid layers are found for volumetric flow rates
ranging from 5 to 1500 mL/h (see Figure 3 and ref 2).
In turn, the experiment at -12 °C is characterized by a
maximum in the yield at a volumetric flow rate of 1000 mL/
h. The decrease in yield at the highest volumetric flow rate
of 1500 mL/h can be explained by setting a residence time
too short for this low temperature. On the contrary, the
decrease in yield for low volumetric flow rates, and corre-
sponding long residence times, is currently not well under-
stood. Since this decrease is not accompanied by a rise in
side- or hydrolysis products, one currently has to rely on
the working hypothesis that a new side product possibly was
formed which is not detected by means of HPLC.
With the exception of the measurements made at the
lowest temperature (-12 °C), the yields of the consecutive
and side products C1, S1, and S3 decrease with increasing
volumetric flow rate, i.e., shorter residence time. The yield
of the two other side products S2 and S4 is nearly constant.
As it is a common phenomenon that side reactions have
higher activation energies than the main reaction, so their
relative contribution is expected to increase with increasing
temperature. Consequently, this means that control over
residence time at elevated temperature is a proper way of
reducing the amount of side products, which is confirmed
by these experiments. At the Clariant site, a later, more
detailed investigation on the dependency of the formation
of the side product diphenyl boronic acid on the residence
time was made, using the glass slit interdigital micromixer
(see Table 6).
from there either a slightly constant or decreasing profile
up to 50 °C (see Figure 6). Such a constant relationship is
unusual for conventionally processed, strongly exothermic
reactions such as the Phenyl Boronic Acid Process, typically
showing a remarkable decrease in yield or selectivity when
being carried out at high temperature. Similar favorable
operation for such exothermic reactions using microreactors
at high temperature was reported, e.g. for a metallo-organic
synthesis23 and for the hydrolysis of benzal chloride.4
Since variation of volumetric flow rate is equivalent to
changing residence time for the given configuration with a
constant tube length, it is to be expected that the maximum
of the yield may be shifted to a higher temperature with
increasing volumetric flow rate. Indeed, this is found at 500
mL/h (0 °C) and 1000 mL/h (22 °C), while a maximum is
not observed in the temperature range investigated up to 50
°C when setting the volumetric flow rate to 1500 mL/h.
For a given residence time, the impact of temperature
variation on the yield of side and consecutive products is
notably low. This is confirmed by a second set of experi-
ments at the Clariant site, which were performed after the
initial laboratory-scale processing at IMM, not shown here
(see ref 6).
Residence Time Variation. A striking feature of the
laboratory configuration investigations is the nearly constant
yield of P1 as a function of volumetric flow rate for operation
at the two higher temperatures at 22 and 50 °C (see Figure
7). Obviously, in these two cases even the shortest residence
time is long enough to ensure completion of reaction.
Moreover, this shows that mixing sensitivity is not pro-
nounced. The latter probably is due to the focused multi-
lamination flow pattern generated by the interdigital mixer,
Concentration Variation. Doubling of the concentration
to 1.0 mol/L shows that selectivity is decreased, i.e., the
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
519

Figure 7. Yield of products as a function of volumetric flow rate, respectively residence time for the laboratory configuration with
the triangular interdigital glass micromixer.
Table 6. Yields at various residence times obtained for the
target product phenyl boronic acid and of the consecutive
product diphenyl boronic acid of the Phenyl Boronic Acid
Process using the laboratory-scale configuration with the slit
interdigital glass micromixer
volumetric flow rate of 2 L/h, however, resulted in a
significant reduction of process performance. The decrease
in yield to only 65% demonstrates that the caterpillar
minimixer was operating in a flow range which probably
caused nonoptimal mixing.
Both isolated preparative product samples had a high
purity of 98-99%. The yield of the consecutive product C1
(diphenyl boronic acid) was found to be 1-2%, the other
impurities (S1-S5) only ranging in total below 0.1%. Again,
similar findings were made in later investigations at the
Clariant site (reaching purities up to 99.2%), demonstrating
also that the impact of process temperature on the purity of
the product is negligible, i.e., at a favourable large temper-
ature range highly pure products can be achieved.
Although only two values for the yield were determined,
both results point to a large impact of volumetric flow rate
on the product yield for the caterpillar mixer. This finding
is consistent with results obtained previously which analysed
the corresponding dependency of the mixing quality. Results
obtained with a competitive reaction system confirmed a
steep increase of mixing quality with volumetric flow rate
in the range of 100-5000 mL/h.24
It is further known through CFD (computational fluid
dynamics) simulation that nonideal flow splitting and
recombination in the caterpillar mixer at low volumetric flow
rates is the origin of this behaviour.5 In turn, at high
volumetric flow rates a more intense mixing is observed
which has characteristics of turbulent flow (see Figure 8).
Experiments Using the Pilot-scale Configuration with
the Caterpillar Minimixer: Protocol (4). Table 9 lists all
experiments made with the pilot-scale configuration using
T
τ
Yisolated product HPLC a/a purity Ydiphenyl boronicacid
[°C] [s]
[%]
[%]
[%] HPLC a/a
20
20
20 120
20 180
5
60
94.9
93.9
95.7
94.8
98.9
99.1
98.9
98.5
0.6
0.8
1.1
1.3
content of diphenyl boronic acid is decreased as well. This
effect is, however, small compared to the difference in
selectivity between any of the two microreactor experiments
given in Table 7 (using the glass slit mixer) as compared to
that of batch processing.
Experiments Using a Laboratory Configuration with
the Caterpillar Minimixer: Protocol (3). Corroboration
of HPLC Results by PreparatiVe Isolation of Product Phenyl
Boronic Acid. To substantiate the analytical results gained,
the product P1 was isolated by preparative means. However,
since the time of operation of the laboratory configuration
with the triangular interdigital micromixer was limited to
about 30 min, it was not possible to obtain the desired
quantitative amount of the product phenyl boronic acid which
was set to 15 g.
Using this set of process parameters and equipment, the
highest yield of 89% measured during the complete inves-
tigations was found (see Table 8). By preparative isolation
of about 30 g product this result basically could be confirmed.
A preparative yield of 81% was evaluated. Using a lower
520
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

Table 7. Yields at two concentrations of the reactant solutions obtained for the target product phenyl boronic acid and of the
consecutive product diphenyl boronic acid of the Phenyl Boronic Acid Process using the laboratory-scale configuration with the
slit interdigital glass micromixer
T
[°C]
c
τ
[s]
Ycrude product
HPLC
a/a purity [%]
Ypure product
Ydiphenyl boronic acid
[%] HPLC a/a
[mol/L]
[%]
[%]
20
20
0.5
1.0
5
5
94.9
93.7
98.9
98.4
93.9
92.2
0.6
1.1
Table 8. Yields obtained for the target product and side/
consecutive products of the Phenyl Boronic Acid Process
using the laboratory configuration with the caterpillar steel
minimixer
however, did not exceed that value. The best result was
achieved at 10 °C und 10 s.
Moreover, the results show that even for long residence
times (about 4 times longer as for the laboratory configu-
ration) and tubes of large inner diameter, no significant
decrease of yield was observed at the two lowest tempera-
tures applied, 10 and 20 °C. In a broader perspective, this
also elegantly shows that hybrid systems, composed of
microfluidic components and conventional equipment, are
suitable solutions for process intensification of, at least,
certain fast organic reactions.
flow rate [mL/h]
mass of isolated crystals [g]
yield [%]
P1 Content [%]
C1 Content [%]
2000
26.29
65
98.9
∼1.0
<0.1
10000
43.83
81
98.2
∼1.7
total content of S1, S2, S3, S4 [%]
<0.1
Only one experiment with the pilot-scale configuration
showed a considerable decrease in yield. At a temperature
of 40 °C and a residence time of 10 s a decrease of yield to
22% was observed. The steel tube employed for this
experiment had an inner diameter of 4.8 mm which roughly
is equivalent to the channel diameter of the caterpillar mixer,
however, not to its internal microstructures. The results get
even worse when a still larger tube (i.d.: 9.3 mm) is used.
Plugging was observed when operating at a temperature of
40 °C despite the flow guidance through a large tube.
Plugging occurred neither at 30 °C with a residence time of
26 s nor at 20 °C and 120 s.
Generally, a much larger dependency of yield on tem-
perature was observed for the experiments with pilot-scale
configuration as compared to the laboratory configuration,
probably for reasons of worse mass- and heat transfer (see
Figure 9).
The yield of the side and consecutive products was small
when using the pilot-scale configuration, in agreement with
the experiments using the laboratory configuration with the
caterpillar mixer.
Figure 8. CFD simulation of streamlines in the caterpillar
minimixer: 1 m/s; 5.2 L/h; Re ) 1200.
the caterpillar minimixer. The maximum yield of 89%
obtained with the laboratory configuration using the caterpil-
lar mixer could be confirmed also with the pilot-scale
configuration equipped with the same device (see Table 9).
The corresponding experiments, carried out at 10, 20, 30,
and 40 °C with residence times of 1, 5, 10, 26, and 120 s,
5. Conclusions
The replacement of a former batch process by a continu-
ous-flow process with microfluidic devices results in a
notable process intensification for the Phenyl Boronic Acid
Table 9. Yields obtained for the target product and side/consecutive products of the Phenyl Boronic Acid Process using the
pilot-scale configuration with the caterpillar steel minimixer
temperature
10 °C
10 26
20 °C
10
30 °C
10
40 °C
residence time [s]
6
120
6
26
120
1
6
26
1
6
10
P1
C1
S2
S1
S3
S4
85.8 89.2 85.8 86.1 90-95a 90-95a 75.8 90-95* 71.1 76.7 77.0 75.9 79.0 56.6 22.4
0.0
4.9
0.0
0.0
0.0
0.0
4.9
0.0
0.0
0.0
0.0
5.8
0.0
0.0
0.0
0.0
5.6
0.0
0.0
0.0
0.0
5.6
0.0
0.0
0.0
0.0
7.0
0.0
0.0
0.0
0.0
5.5
0.0
0.0
0.0
0.0
5.9
0.0
0.0
0.0
0.0
6.3
0.0
0.0
0.0
0.0
5.1
0.0
0.0
0.0
0.0
5.1
0.0
0.0
0.0
0.0
4.8
0.0
0.0
0.0
0.0
5.2
0.0
0.0
0.0
0.0
5.5
0.0
0.0
0.0
0.0
6.1
0.0
0.0
0.0
a Above 90% yield the accuracy of the HPLC measurements was lower, i.e., indicating typically too high yields resulting in dissonant overall balances of the
product and by-products, probably due to insufficient calibration in this region. These data were corrected to the range shown to give reasonable overall balances.
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
521

Figure 9. Yield of products as a function of volumetric flow rate, respectively residence time for the pilot-scale configuration with
the caterpillar steel minimixer.
Process with regard to product yield, reduction of side/
consecutive products, and energy expenditure.
The best result of the investigations was an HPLC yield
of 89% (preparatively confirmed: 81%). This is nearly 25%
higher as compared to the established Phenyl Boronic Acid
Process, when being industrially performed in a batch-wise
manner, and about 18% more than the laboratory-scale batch.
The side- and consecutive products are reduced to levels
below 1%, thereby increasing selectivity and facilitating
separation by elimination of the distillation step. As a result,
a product of higher purity (∼99%) is achieved.
Instead of having three heating-cooling cycles, the
microreactor operation needs only one heating-cooling cycle
with considerably reduced temperature differences; the latter
basically because of increasing the reaction temperature from
cryogenic to ambient (see Figure 10). Using the laboratory
configuration with the triangular micromixer even a tem-
perature of 50 °C could be used, if desired, without any
notable decrease in yield. In a broader view, this is another
confirmation that microdevices allow a much higher vari-
Figure 10. Temperature levels due to the need of heating and
cooling for Phenyl Boronic Acid Process. (a) Conventional batch
process; (b) novel microreactor process.
ability regarding process temperature as compared to con-
ventional reactors.
distinct advantages and drawbacks. In some cases, such as
here, they are thereby complementary.
The small dependency of the yield on either the volu-
metric flow rate or on the temperature when using the
laboratory-scale configuration is advantageous with regard
to process control. A small deviation of these parameters
during operation, for some reason, will not cause a loss of
the process benefits. However, this positive feature is lost
to a larger extent when using the pilot-scale configuration.
This shows that the micro- and mini-approach have their own
Using the minimixer already at the laboratory stage,
favourably high throughputs of 10 L/h could be achieved
which facilitated preparative confirmation of the analytical
results. In addition, thereby a fast transfer of the results to
scale-up investigations using the pilot-scale configuration was
achieved already at present. A throughput of 10 L/h enables
a production of about 6 kg of phenyl boronic acid per day.
With pumps of higher-pumping capacity as used in the
522
•
Vol. 8, No. 3, 2004 / Organic Process Research & Development

investigations described herein, a throughput of 100 L/h is
achievable with the caterpillar mixer, equivalent to a produc-
tion of 20 tons of product per year.
Last but not least, the results show that fouling-sensitive
reactions and microreactors are by no means a contradiction
if optimal internal dimensions are chosen appropriately, if
sufficient hydrodynamic information is available, and if
compromises concerning setting of process parameters are
acceptable. This positive interplay of circumstances certainly
has its severe limits, but presently the possibilities of using
microfluidic devices for such a processing, albeit far from
being optimal at present, seem by far not to be exploited.
This technique displays great potential for a broader ap-
plication.
Received for review November 28, 2003.
OP0341768
Vol. 8, No. 3, 2004 / Organic Process Research & Development
•
523