Flow processing of microwave-assisted (heterogeneous) organic reactions

Article abstract of DOI:10.1021/op900257f

A commercially available continuous-flow reactor was adapted to run three organic reactions, e.g. two heterogeneous and one homogeneous mixture under microwave heating. The setup was operated either as a batch-loop reactor for running a biocatalyzed esterification of (R,S)-1-phenylethanol with vinyl acetate and the esterification of (S)-pyroglutamic acid with n-decanol (i.e., Laurydone process) or in a single pass for the aspirin synthesis as a homogeneous mixture. The tubular reactor has been characterized to perform a series of three equally sized, continuously operated stirred tank reactors on average. Although the (operational) costs of the microwave-heated tubular reactor are higher than conventionally heated processes in fine chemical operations, it was demonstrated that the experimental data can be used for process design. Plugging remains an important obstacle to be dealt with. However, benefits with respect to safety and scalability are expected to enable a fair compensation for the costs when implementing this novel technique.

Full text of DOI:10.1021/op900257f

Organic Process Research & Development 2010, 14, 351–361
Flow Processing of Microwave-Assisted (Heterogeneous) Organic Reactions
Mark H. C. L. Dressen,† Bastiaan H. P. van de Kruijs,† Jan Meuldijk,‡ Jef A. J. M. Vekemans,† and Lumbertus A. Hulshof*,†
EindhoVen UniVersity of Technology, Laboratory of Macromolecular and Organic Chemistry, Applied Organic Chemistry, and
EindhoVen UniVersity of Technology, Process DeVelopment Group, Den Dolech 2, 5612 AZ EindhoVen, The Netherlands
Abstract:
A commercially available continuous-flow reactor was adapted to
run three organic reactions, e.g. two heterogeneous and one
homogeneous mixture under microwave heating. The setup was
operated either as a batch-loop reactor for running a biocatalyzed
esterification of (R,S)-1-phenylethanol with vinyl acetate and the
esterification of (S)-pyroglutamic acid with n-decanol (i.e., Lau-
rydone process) or in a single pass for the aspirin synthesis as a
homogeneous mixture. The tubular reactor has been characterized
to perform a series of three equally sized, continuously operated
stirred tank reactors on average. Although the (operational) costs
of the microwave-heated tubular reactor are higher than conven-
tionally heated processes in fine chemical operations, it was
demonstrated that the experimental data can be used for process
design. Plugging remains an important obstacle to be dealt with.
However, benefits with respect to safety and scalability are
expected to enable a fair compensation for the costs when
implementing this novel technique.
Figure 1. FlowSynth of Milestone srl, Italy.
safety, and the investment costs in equipment, any supplemen-
tary costs of applying microwave technology have to be
compensated by the added value of improved reaction condi-
tions (e.g., microwave effect)4 or more effective downstream
processing.
In this study we demonstrate that, with our selected setup
(utilizing the FlowSynth tubular reactor of Milestone), a proper
starting point for direct comparison of microwave with con-
ventional heating has been created, allowing prediction of the
potential upscaling of microwave-assisted chemistry in continu-
ous operation. Three chemical reactions were selected to
illustrate how much effort and adaptation are required, particu-
larly with heterogeneous systems, to make continuous-flow
operations under microwave irradiation conditions viable.
1. Introduction
Today, much effort is being invested in an increasingly
competitive world to meet the requirements of the fine chemical
and pharmaceutical industry with respect to “first-time-right”
performance, a short-time-to-market, and avoidance of surprises
during process scale-up. A demand for larger quantities does
not imply replacing “mg” with “kg” in the recipe of the
chemical process.
Process intensification based on, e.g., microreactor technol-
ogy and microwave heating is actively pursued to achieve a
better position in the industrial scene. Increase of reaction rates
and improved selectivity, combined with the automation of
repetitive procedures, demonstrate the advantageous application
of these enabling techniques. One of the major weaknesses of
microwave-assisted chemistry is scaling-up. Since the penetra-
tion depth of microwaves is limited, the best chances for
increased production volumes lie in relatively small-scale
continuous operations, somewhat uncommon in the fine chemi-
cal industry.1-3
2. Experimental Characterization of the FlowSynth Reactor
Characterization of the flow reactor was necessary to allow
a comparison with standardized equipment, especially to clarify
the limitations of the flow reactor and to define the operational
window. A selection of reactor type and peripheral equipment
(for example, type of pump) had to be made on the basis of
the specification of the reaction conditions. At the start of our
study one type of microwave equipment, i.e. the FlowSynth of
Milestone srl, Italy (Figure 1),5,6 was commercially available
that involves a combination of a membrane pump, a tubular
reactor, and a microwave oven, all controlled by one computer.
To make microwave heating feasible on a larger scale this
technique should perform better than operation with conven-
tional heating. Besides aspects like productivity, selectivity,
(4) (a) Dressen, M. H. C. L.; Van de Kruijs, B. H. P.; Meuldijk, J.;
Vekemans, J. A. J. M.; Hulshof, L. A. Org. Process Res. DeV. 2007,
11, 865. (b) Dressen, M. H. C. L.; Stumpel, J.; Van de Kruijs, B. H. P.;
Meuldijk, J.; Vekemans, J. A. J. M.; Hulshof, L. A. Green Chem.
2009, 11, 60.
* To whom correspondence should be addressed. E-mail: L.A.Hulshof@tue.nl.
† Laboratory of Macromolecular and Organic Chemistry, Applied Organic
Chemistry.
‡ Process Development Group.
(1) Baxendale, I. R.; Pitts, M. R. Chem. Today 2006, 24, 41.
(2) Glasnov, T. N.; Kappe, C. O. Macromol. Rapid Commun. 2007, 28,
395.
(5) Bowman, M. D.; Holcomb, J. L.; Kormos, C. M.; Leadbeater, N. E.;
Williams, V. A. Org. Process Res. DeV. 2008, 12, 41.
(6) Moseley, J. D.; Lenden, P.; Lockwood, M.; Ruda, K.; Sherlock, J.-P.;
Thomson, A. D.; Gilday, J. P. Org. Process Res. DeV. 2008, 12, 30.
(3) Wiles, C.; Watts, P. Eur. J. Org. Chem. 2008, 1655.
10.1021/op900257f  2010 American Chemical Society
Published on Web 02/09/2010
Vol. 14, No. 2, 2010 / Organic Process Research & Development
•
351

Figure 3. Stirring shaft with three Weflon15 blades for the
tubular reactor of the FlowSynth.
vessel with microwave-transparent material is a simple solution
to improve the energy efficiency. Fortunately, a low heat
conductivity is also a good material property of polymer
materials such as Teflon or PEEK (poly(ether ether ketone))
which are commonly used to construct microwave-heated
reactors with a high chemical resistance or high mechanical
strength (e.g., for pressure), respectively.
The efficiencies of transferring microwave energy into heat
were measured in the FlowSynth reactor for three solvents (p-
xylene, demineralized water, and ethylene glycol) under dif-
ferent conditions. The Milestone FlowSynth was equipped with
a typical stirrer shown in Figure 3. The efficiency (η) was
defined by the ratio of the thermal energy required to heat up
the solvent (see eq 1) and the amount of electromagnetic energy
delivered by the microwave oven.
Figure 2. Energy efficiency of conventional heating (steam) and
microwave heating.10–13
The precommercial version of the FlowSynth was described
by the group of Ondruschka.7 This novelty has been patented.8,9
Energy efficiencies, operation, the flow and mixing charac-
teristics, and heat transfer were investigated.
φm · cp · (∆T)observed
microwave power used
2.1. Energy Efficiencies. To make a fair comparison
between energy efficiencies of conventional and microwave
heating the starting point needs to be identical, being gas as a
natural source of energy. Figure 2 illustrates the energy
efficiency for each transformation step. Different energy sources
(gas or electricity) imply different prices, based on the technol-
ogy necessary to generate this power.10-13 In short, microwave
heating loses at least 2 times more energy (43% over 22%, see
Figure 2) in its fore track, and at the end the energy bill is at
least a factor of 8 higher (0.088 € over 0.011 €)10 compared to
condensing steam heating for industrial applications. The
efficiency values given in Figure 2 are based on different
literature sources and experimental results.10–13 Altogether, the
energy costs for microwave heating will be higher than those
for conventional (steam) heating.
The efficiency of the transition from microwave energy to
the reaction mixtures is a rather uncertain value, depending on
scale, equipment type, and loss tangent. Hoogenboom et al.
reported that the losses on small scale are much higher than
for larger scales.14 Predominantly this is due to heat losses to
the environment, based on a relatively high surface-to-volume
ratio for small equipment. Thermal insulation of the reaction
η )
φm ) mass flow rate [kg · s-1]
cp ) heat capacity [J · kg-1 · K-1]
∆T ) difference in temperature, outlet - inlet [K]
η ) efficiency [%]
(1)
The experimental results demonstrate that the absorption of
microwaves increases in the order p-xylene < water < ethylene
glycol.16 This could be expected from the loss tangents for
ethylene glycol, water, and p-xylene at 20 °C being 1.35, 0.127,
and 0.0011, respectively.17 Operating in a turbulent flow regime
would improve the efficiency. Each solvent appears to have an
average maximum of η, which is not influenced by the
operational conditions. The error margin of the measured values
is in the range of 5-10%.
2.2. Reactor Type. At first sight the reactor in the Flow-
Synth looks like a tubular reactor, and the presence of stirrers
(Figure 3) may induce (additional) residence time distribution.
Measurement of the residence time distribution (RTD) allows
for characterization of the reactor.18 The theory of RTD was
first proposed by MacMullin and Weber,19 and worked out in
more detail by Danckwerts20 some years later. A tracer is
(7) Bierbaum, R.; Nu¨chter, M.; Ondruschka, B. Chem. Eng. Technol. 2005,
28, 427.
(8) Fagrell, M. Continuous-flow system with microwave heating, Personal
Chemistry, Patent number WO/2003/041856, 2003.
(9) Lautenschlager, W. Apparatus and method for treating chemical
substances in a microwave field. U.S. Patent 2005/0034972, 2005.
(10) Energy Tips - Steam, 2006, sheet #15.
(15) Weflon is Teflon impregnated with carbon. Indirect heating within a
microwave oven is obtained, which is obligate by introducing
microwave-transparent liquids and achieving elevated temperatures
(see also ref 27).
(11) Graus, W. H. J.; Voogt, M.; Worrell, E. Energy Policy 2007, 35, 3936.
(12) Nu¨chter, M.; Ondruschka, B.; Bonrath, W.; Gum, A. Green Chem.
2004, 6, 128.
(16) See Supporting Information.
(13) Assuming that electrical energy costs 0.15 € per kWh and natural gas,
0.65 € per m3 (energy content 33 MJ per m3) for the private sector
and 0.27 € per m3 for industry, the costs to generate one MJ of steam
on an industrial scale is 0.011 €, for a private boiler 0.046 € and for
one MJ of microwave heat with the average efficiency the costs are
0.088 €, including the indicated efficiencies.
(17) Physical constants are obtained from the Handbook of Chemistry and
Physics, 89th ed. (online edition); Lide, D. R., Ed.; CRC Press: Boca
Raton, FL, 2008-2009.
(18) Fogler, H. S. Elements of Chemical Reaction Engineering, 3rd ed.;
Prentice Hall: NJ, 1999.
(19) Macmullin, R. B.; Weber, M., Jr. Trans. Am. Inst. Chem. Eng. 1935,
31, 409.
(14) Hoogenboom, R.; Wilms, T. F. A.; Schubert, U. S. Polym. Prepr.
(Am. Chem. Soc., DiV. Polym. Chem.) 2008, 49, 930.
(20) Danckwerts, P. V. Chem. Eng. Sci. 1953, 2, 1.
352
•
Vol. 14, No. 2, 2010 / Organic Process Research & Development

σ2 ) ∞ (t - 〈τ〉)2 · E(t) dt
(4b)
(4c)
Table 1. Number of equally sized tanks in series for various
∫
0
flow rates and stirrer speeds in the Milestone FlowSynth
flow rate
stirrer speed reactors in series
〈τ〉2
entry (mL·min-1
)
(rpm)
(number)a
remark
n )
σ2
1
2
191
191
191
191
191
191
214
214
214
214
232
232
232
0
140
215
285
327
380
0
140
215
327
0
140
215
3.0
4.9
5.3
2.4
1.8
2.3
3.7
3.6
2.5
4.2
3.5
3.9
2.9
-
-
3
-
On average the FlowSynth is equivalent to approximately
three equally sized CSTRs in series. If the flow inside the reactor
is synchronized with the stirrer speed, the residence time
distribution is small. Then, the stirred tubular reactor tends to
perform more like a plug-flow reactor.
Any disturbance leads to more axial mixing, which has the
tendency to be more equivalent to a continuous stirred-tank
reactor (CSTR) for the flowing fluid. The presence of either a
shoulder or split-peak points to two or multiple routes which
lead to differences in the residence time (see Figure 4). Figure
5 illustrates the theoretically calculated residence time distribu-
tions (eq 4a) for a series of equally sized CSTRs with different
values of n. Comparing Figures 4 and 5 shows that the residence
time distribution in the Milestone FlowSynth is equivalent to
that in a series of three equally sized CSTRs.
4
split peak
shoulder
-
5
6
7
-
8
split peak
split peak
shoulder
split peak
shoulder
shoulder
9
10
11
12
13
a n calculated from tracer response curves using eqs 4a-4c.
injected in the reactor feed. Detection of the tracer is performed
at the outlet. A pulse or a step injection can be chosen to
introduce the tracer. In our setup the response on a tracer pulse
has been measured. These measurements give access to the exit-
age function E(t) which characterizes RTD. E(t) quantifies the
time-dependent fraction of the tracer. Equation 2 shows how
E(t) can be calculated from the response of a tracer pulse. The
flow rate (φν) is constant, the concentration C(t) of the tracer is
measured, and the exact amount of tracer (N0) is not directly
known. E(t) dt stands for the fraction tracer leaving the system
between t and t + dt after the injection. Note that for t f ∞ all
the tracer should have left the reactor, see eq 3.
φν.C(t)
C(t)
E(t) )
)
(2)
∞ C(t) dt
N0
∫
0
Figure 4. Plot of response of a pulse injection measured by
GC (in arbitrary units) changing in time (dimensionless; 〈τ〉 is
mean residence time in the whole system) for the FlowSynth
with Weflon impeller blades (see Figure 2). (a) Typical plot with
small disturbance [Table 1, entry 2]. (b) Large disturbance in
tubular reactor leading to enhanced axial mixing [Table 1, entry
8].
∞ E(t) dt ) 1
(3)
∫
0
Benzyl alcohol, as tracer in a stream of toluene, has been
measured by GC-FID. In this way there is a linear relation
between the measured GC-peak area and the actual concentra-
tion (C(t)). The exit-age function E(t) is used in a one-parameter
model for nonideal reactors implying the number of equally
sized CSTRs in series (n) which gives the same E(t) as measured
for the reactor considered, see eq 4a. The number of tanks (n)
is calculated from the dimensionless variance. The variance itself
is the square of the standard deviation (σ), and the mean
residence time is 〈τ〉 see eqs 4b and 4c.18
The RTD results are influenced when the reactor setup is
operated under different conditions. Table 1 summarizes the
number of equally sized tanks and the operational conditions
(i.e., flow rates and stirrer speeds) for the Milestone FlowSynth.
nn · tn-1 · e-n·t/〈τ〉
〈τ〉n · (n - 1)!
Figure 5. Theoretically calculated residence time distribution
with an increasing number of tanks (n ) 1, 2, 3, 5 and 20).21
E(t) )
(4a)
Vol. 14, No. 2, 2010 / Organic Process Research & Development
•
353

Table 2. Mixing regimes21
F · N · d2imp
Reimp
)
entry time constant
type of mixing
µ
(6)
(treaction
)
dimp ) impeller diameter [m]
N ) impeller speed [s-1]
1
2
3
10 min ·h
s·min
ms
independent of mixing (intrinsic kinetics)
macromixing (i.e., circulation)
micromixing (i.e., diffusion)
2.4. Heat Transfer. The Nusselt number (Nu) is the ratio
of the convective heat transfer to the conductive heat transfer.
Equations 7 and 8 give the definition of the Nusselt number
and an empirical dimensionless correlation from heat transfer
in stirred tanks, respectively.22 The overall heat flow rate from
the Weflon15 blades (φq) to the fluid in the reactor is estimated
by eq 9. Solvents with a small loss tangent can only absorb
microwave energy indirectly via the Weflon blades mounted
on the stirrer shaft in the tubular reactor. Efficient heat transport
prevents high temperatures of the heating surface (A ) 95 cm2)
of the Weflon blades (eq 9). With the assumption of direct and
fast absorption of microwaves by the Weflon blades the
temperature of these blades is at a constant value.
Table 3. Relationship between flow rate and Reynolds
numbera in the reactor coil at four temperatures
flow rate (mL ·min-1
)
temperature (°C)
53
115
167
214
250
25
50
31
40
50
60
69
89
110
132
100
128
158
191
128
165
204
245
149
192
238
286
75
100
a Reynolds number using eq 5; dt ) 1.75 × 10-2 m; p-xylene.
Table 4. Relationship of stirring speed and Reynolds
numbera in the Milestone FlowSynth at four temperatures
stirrer speed (rpm)
〈h〉 · L
〈Nu〉 )
temperature (°C)
60
195
322
380
425
λ
25
50
402
517
640
769
1306
1680
2080
2501
2191
2817
3488
4193
2546
3273
4054
4873
2854
3669
4544
5463
〈Nu〉 ) Nusselt number
〈h〉 ) heat transfer coefficient
75
100
L ) characteristic length (reactor diameter)
λ ) thermal conductivity
a Reynolds using eq 6; dimp ) 1.68·10-2 m; p-xylene.
(7)
〈Nu〉 ) cr · Re2/3 · Pr1/3 · Vi0.14
〈Nu〉 ) Nusselt number
2.3. Flow and Mixing. The Reynolds number, as a dimen-
sionless number, represents the ratio of inertia forces and the
viscous forces on the flowing fluid. The Reynolds number
characterizes in this way the flow regime in the reactor for a
certain experiment.
Mixing may be crucial not only to bring the reactants
together (mass transport) but also to remove/supply the heat
for temperature control during the reaction (heat transport).
Table 2 gives an overview of the type of mixing relevant for
various time constants of reaction.22
The tubing and the reactor have different diameters. Depend-
ing on the flow rate, a certain reaction mixture or solvent
displays a laminar or turbulent pattern. If the Reynolds number
(Re, see eq 5) is over 3500 in the tube, the flow will develop
turbulence. In case of a turbulent flow the heat exchange from
the Weflon blades of the semi-coil-shaped stirrer inside the
reactor column towards the reaction mixture is much more
efficient than for a laminar flow. Table 3 gives an overview of
the conditions to convert a laminar flow in the reactor (without
stirring) into a turbulent regime. The influence of stirring on
the Reynolds number is summarized in Table 4 using eq 6.
cr ) stirrer constant
(8)
Re ) Reynolds number (rotation)
Pr ) Prandtl number
Vi ) viscosity constant (≈1)
φq ) 〈h〉 · A · (∆T)ln
φq ) heat supplied to the system [J · s-1]
〈h〉 ) overall heat transfer
coefficient [J · s-1 · m-2 · K-1]
A ) surface [m2]
(9)
(∆T)ln ) logarithmic mean
temperature difference [K]
(Tout - Tblade) - (Tin - Tblade
)
(∆T)ln )
Tout - Tblade
Tin - Tblade
ln
(
)
In case of p-xylene (λ ) 0.124 W · m-1 · K-1) and a
relatively low stirrer constant (cr ) 0.36; blade stirrer, no
baffles; dimp ) 30.5 mm; Dtube ) 34.3 mm)23 the Nusselt
number will range from 100 to 200 depending on
temperature (50-100 °C) and stirrer speed (200-400
rpm). The heat transfer coefficient (h), consequently, is
calculated to range from 360-725 W · m-2 · K-1. Typically
the surface of the blades (A ) 95 cm2) can reach a
temperature of 160 °C when an incoming stream of
p-xylene at room temperature is heated up to 120 °C for
an efficient microwave-induced power input of 277 W into
the system.
dt · u · F
Retube
)
µ
dt ) tube diameter [m]
(5)
u ) mean velocity [m · s-1]
F ) density [kg · m-3]
µ ) dynamic viscosity [Pa · s]
354
•
Vol. 14, No. 2, 2010 / Organic Process Research & Development

Figure 6. Various modifications of the original Milestone FlowSynth apparatus.
Scheme 1. Esterification of (R,S)-1-phenylethanol with vinyl
acetate in an enantiomerically selective reaction using
Novozym 435
3. Flow Chemistry: A Continuous Operation
Most studies with continuous-flow reactors are reported in
the field of organic chemistry and describe the application of
conventional heating; so-called flow chemistry.24-27 The use
of continuously operated reactors demands for relatively fast
reactions, otherwise flow rates have to become extremely low
or reactors become too large. Residence time is the key
parameter determining the degree of conversion. One way to
combine a high flow rate with complete conversion is to
completely recycle the reactor outlet, resulting in batch-loop
operation. In this way, the advantages of a tubular reactor and
a batch reactor are combined.
Scheme 2. Esterification of (S)-pyroglutamic acid with
n-decanol
As mentioned before the penetration depth (δp) of micro-
waves is limited to the range of centimetres depending on the
composition and physical properties of the reaction mixture.
The penetration depth is expressed as the distance at which the
energy is reduced to 37%, see eq 10.28 Therefore, the reactor
diameter for microwave heating equipment ranges between 5
to 100 mm.
of a cosmetic product, Laurydone)30-33 with each a set of two
experiments were executed in the FlowSynth (Schemes 1 and
2). Each set of experiments was compared with the correspond-
ing batch experiment. The Laurydone reaction is to the best of
our knowledge, the only fine chemical reaction performed on
an industrial scale similar to our approach.
ꢀ
√
λ0 ε
δp )
2πεꢀꢀ
δp ) penetration depth [m]
(10)
λ0 ) free space wavelength [m]
εꢀ ) dielectric constant
(21) Ullmann’s Encyclopedia of Industrial Chemistry Online, Wiley-VCH,
2008.
εꢀꢀ ) dielectric loss
(22) Atherton, J. H.; Carpenter, K. J. Process DeVelopment: Physicochem-
ical Concepts; Oxford University Press: Oxford, 1999.
(23) Zlokarnik, M. Stirring: Theory and Practice; Wiley-VCH: Weinheim,
2001; Chapter 7, Intensification of heat transfer by stirring.
(24) LaPorte, T. L.; Wang, C. Curr. Opin. Drug DiscoVery DeV. 2007, 10,
738.
Two chemical reactions were investigated in a batch-loop
setup and one reaction in a continuous operation.
(25) Anderson, N. G. Org. Process Res. DeV. 2001, 5, 613.
(26) Tundo, P. Continuous Flow Methods in Organic Chemistry; Ellis
Horwood: Chichester, 1991.
4. Performance for Batch-Loop Operation
Roughly, the setup for batch-loop operation that we have
used involves flow chemistry in a microwave oven. This
approach is basically an ideal situation to accommodate
reactions requiring different reaction times, to transfer a small-
scale batch reaction into a flow setup and to increase the
productivity by means of parallel circuits (i.e., scaling out). On
the basis of the experience gained in initial studies, the
FlowSynth reactor was adapted at various places to enable
processing of heterogeneous reactions. Our modifications are
marked with circles in Figure 6.27
(27) Dressen, M. H. C. L.; Van de Kruijs, B. H. P.; Meuldijk, J.; Vekemans,
J. A. J. M.; Hulshof, L. A. Org. Process Res. DeV. 2009, 13, 895.
(28) Risman, P. O. J. MicrowaVe Power Electromagn. Energy 1991, 26,
243.
(29) Howarth P.; Lockwood, M. TCE, 2004, June, 30.
(30) Rochas, J. F.; Bernard, J. P.; Lassalle, L. Ann. MicrowaVe Heating
Symposium 2002, conf. 37, 20–23.
(31) Jose, A.; Takeru, H. Pyroglutamic acid esters used as dermal
penetration enhancers for drugs. (Merck & Co Inc.). Patent number
EP227531, 1987.
(32) Thomae, K. (GmbH). Neue Pyrrolidoncarbonsa¨urealkylester und
Verfahren zu ihrer Herstellung. Patent number DE2102173, 1972.
(33) Lassalle, L. Proce´de´ de pre´paration d’un ester par chauffage, au moyen
d’un champ de micro-ondes d’un me´lange heteroge`ne d’acide et
d’alcohol. Patent number FR2833260, 2003.
Two chemical reactions (i.e., a biocatalyzed esterification
and the esterification of (S)-pyroglutamic acid in the production
Vol. 14, No. 2, 2010 / Organic Process Research & Development
•
355

4.1. Biocatalyzed Esterification. It was demonstrated that
the enantioselective esterification of commercially available
(R,S)-1-phenylethanol with vinyl acetate and Novozym 435 in
toluene is relatively fast (Scheme 1).34 Similar to the batch
experiments an excess of vinyl acetate was used for reactions
in the FlowSynth to enforce a high conversion within a
reasonable time (2-3 h). Although many reports claiming
positive microwave effects in biocatalyzed processes,35-41 we
were unable to observe any difference with conventional
heating.4a
The scale of the continuous-flow experiment (0.4 mol
1-phenylethanol) was a factor of 100 larger than the batch mode
reaction (4 mmol). The reaction mixture without biocatalyst
(Novozym 435 beads) was premixed in a 1 L round-bottomed
flask and magnetically stirred. After a short period this
homogeneous mixture was pumped through the loop. The
temperature in the top of the tubular reactor (T1) was set at 70
°C. The microwave power was automatically adjusted to
maintain the temperature at T1 of this set point, see Figure 7.
The esterification was performed twice (experiments A and
B) in the FlowSynth. In experiment A the reaction mixture
should have been distributed evenly over the batch-loop setup.
In experiment B the catalyst was retained on purpose in the
microwave-heated tubular reactor (R1).
Figure 7. Setup for batch-loop operation for the esterification
of (R,S)-1-phenylethanol with vinyl acetate and the esterification
of (S)-pyroglutamic acid with n-decanol (Laurydone process):
R1 ) microwave-heated tubular reactor; R2 ) round-bottomed
flask with magnetic stirrer; P ) (membrane or gear) pump; C
) cooler with Archimedes’ screw; T1 ) temperature sensor
inside the upper part of the tubular reactor and T2
temperature sensor just downstream the cooler.
)
4.2. Experiment A. The flow rates were measured before
and after the addition of Novozym 435. In both cases the flow
rate was 175 mL ·min-1. Once the reaction setup reached a
steady state (in which the temperature was constant in time)
Novozym 435 was charged to flask R2. During the reaction
aliquots were taken just after point T2, but before vessel R2 (see
Figure 7). Figure 8 shows the time-conversion history. It is
noteworthy to mention that the amount of Novozym 435
appeared to be higher inside the reactor than in vessel R2 or
outside R1.
Figure 8. Esterification of (R,S)-1-phenylethanol with vinyl
acetate in toluene at 70 °C catalyzed by Novozym 435 in (a) an
oil-bath-heated (25 mL) batch reactor (4 mmol scale); (R)-1-
phenylethyl acetate (s2s) and (experiment A) in a microwave-
heated batch-loop-wise operated reactor setup shown in Figure
7 (0.4 mol scale) at Oν ) 0.175 L ·min-1; (R)-1-phenylethyl
acetate (s9s).
The liquid velocity in the tubing (νtub ) 0.15 m ·s-1) was
considerably higher compared to that in the tubular reactor (νreact
) 0.012 m ·s-1), which might explain settling of Novozym 435
in the column R1. The narrow exit after T2 can also lead to a
hold-up or plugging of the spherical particles of Novozym 435.
The temperature in vessel R2 has been monitored manually and
never exceeded 30 °C. The total slurry volume was 0.940 L
and the volume of the tubular reactor was 0.180 L. With a flow
rate of 0.175 L ·min-1 and a total residence time of 173 min,
the number of cycles was calculated to be 168 with an average
residence time of a little more than one minute per cycle in
column R1. Consequently, for this batch-loop experiment only
19%42 of the reaction mixture was exposed to the temperature
of 70 °C, while in the batch experiment, however, 100% of the
reaction mixture was subjected to that temperature. The uneven
distribution of Novozym 435 over the complete reaction mixture
hampers a quantitative comparison of the reaction rates for
thermal and microwave heating.
Note that comparison of batch experiments with respect to
both heating methods demonstrated the absence of a rate
enhancement effect by microwave heating.
4.3. Experiment B. The procedure applied in Experiment
A was repeated for the esterification in the continuous-flow
reactor operated in a batch-loop mode. Now, Novozym 435
was charged directly into the tubular reactor system, and the
reactor was put together to operate under flow conditions with
complete retention of the Novozym 435 inside the microwave
cavity, because the system was not stirred. During the reaction,
aliquots were taken after point T2, but before vessel R2. The
results are plotted in Figure 9. The conversion-time history in
Figure 9 is comparable with that of Figure 8. This observation
confirms that the majority of the solid particles in the experi-
mental setup for experiment A resides nearly completely in the
(34) Carrillo-Munoz, J.-R.; Bouvet, D.; Guibe´-Jampel, E.; Loupy, A.; Petit,
A. J. Org. Chem. 1996, 61, 7746.
(35) Yadav, G. D.; Lathi, P. S. Synth. Commun. 2005, 35, 1699.
(36) Yadav, G. D.; Sajgure, A. D. J. Chem. Technol. Biotechnol. 2007,
82, 964.
(37) Re´jasse, B.; Lamare, S.; Legoy, M.-D.; Besson, T. J. Enzym. Inhib.
Med. Chem. 2007, 22, 518.
(38) Pamies, O.; Ba¨ckvall, J.-E. Chem. ReV. 2003, 103, 3247.
(39) Huang, W.; Xia, Y.-X.; Gao, H.; Fang, Y.-J.; Wang, Y.; Fang, Y. J.
Mol. Catal. B: Enzym. 2005, 35, 113.
(40) Lin, G.; Lin, W.-Y. Tetrahedron Lett. 1998, 39, 4333.
(41) Kim, M.-J.; Chung, Y. I.; Choi, Y. K.; Lee, H. K.; Kim, D.; Park, J.
J. Am. Chem. Soc. 2003, 125, 11494.
(42) VR1/Vtot ) 0.19 (fractional residence time).
356
•
Vol. 14, No. 2, 2010 / Organic Process Research & Development

Figure 10. Concentration profile (CA) around and inside a catalyst
particle (with shell catalyst such as Novozym 435) with strong mass
transfer limitation (2) and with intrinsic kinetics (1).
Figure 9. Esterification of (R,S)-1-phenylethanol (4 mmol) with
vinyl acetate in toluene at 70 °C catalyzed by Novozym 435 in
(a) an oil bath-heated (25 mL) batch reactor; (R)-1-phenylethyl
acetate (s2s) and (experiment B) in a microwave-heated
batch-loop-wise operated reactor setup shown in Figure 7 (0.4
mol scale) with catalyst charged in the tubular reactor at Oν )
0.12 L ·min-1; (R)-1-phenylethyl acetate (s(s).
or conventionally heated. However, in Figure 11 (right)
overhead stirring has been used for the same experimental setup
as performed in Figure 11 (left), resulting in identical
conversion-time profiles. By efficient stirring the presence of
a temperature profile over the reaction mixture is minimized,
leading to a vanishing microwave-induced reaction rate
enhancement.
tubular reactor. As compared to the small-scale batch experi-
ment the concentration of Novozym 435 in the tubular reactor
is greater by a factor of 5.2. The total mean residence time per
pass has decreased with the same factor. Note that the effect of
the increased catalyst concentration and the decreased mean
residence time in the tubular reactor on the (R,S)-1-phenyle-
thanol conversion approximately cancel out (for plug flow the net
effect of catalyst concentration and residence time is exactly zero).
The remaining difference in reaction rate is partially related
to either a small temperature gradient in the microwave-heated
tubular reactor or to some mass transfer limitations inside the
reactor (without stirring). Without stirring (and with a laminar
flow) the global reaction rate can change as a result of a change
in (intraparticle) mass transport limitation (see Figure 10).43
This example of a biocatalytic esterification exemplifies a
direct correlation between batch and continuous operation.
Approximately the same final conversions are achieved for
comparable reaction times (i.e., for the batch-loop operation
reaction times are somewhat larger than for the small-scale batch
experiment, see Figure 9).
4.4. Esterification of (S)-Pyroglutamic Acid with n-
Decanol (Laurydone Process). At the moment, the only
example in the field of microwave-assisted fine chemical
applications is the Laurydone process, see Scheme 2.27,29 This
example illustrates the ability to execute microwave heating on
a large scale. Despite the successful large-scale operations with
microwave heating, the originators failed to demonstrate the
superiority of microwave heating technology over conventional
heating. A wrong conclusion about a probable microwave effect
could be drawn as a result of poor stirring (see Figure 11).
Figure 11 (left) depicts the results of two experiments in which
the reaction mixture was magnetically stirred, either microwave
All our batch reactions of the Laurydone synthesis have been
performed with a high concentration of substrate ((S)-pyro-
glutamic acid) in n-decanol. Such a solid weight percentage
(47 wt %) appeared to be impractical to handle in the
continuous-flow reactor. Reduction of the acid weight percent-
age to 10 wt % guaranteed a smooth flow for the initially
heterogeneous reaction mixture. Again, two different conditions
have been applied for this type of esterification (Table 5). It is
important to point out that the membrane pump in experiment
A was replaced by a better performing gear pump in experiment
B. Compared to a membrane pump, a gear pump is more
suitable for handling viscous reaction mixtures, like those
containing n-decanol. The flow scheme of the setup for
experiments A and B is schematically depicted in Figure 7.
4.5. Experiment A. After charging vessel R2 with reactants
and stirring with a mechanical impeller, the resulting solid/liquid
mixture was pumped through the tubular reactor (without
heating). This procedure enabled to run the flow system for
many hours without any complications.
Subsequently, heating was applied to the tubular reactor R1
(Figure 7, 150 °C). Cooling agent of 8 °C was supplied to the
jacket of cooler C. Cooling of vessel R2 was started by pumping
a coolant of 12 °C through the jacket. Shortly after starting up
the heating and cooling, plugging occurred in the cooler C. This
prevented any flow. Also stirring of the coil in the tubular reactor
R1 was not possible. A deposit of solid material on the walls
of cooler C was observed.
In the restarted experiment cooling was only applied to vessel
R2, the jacket of cooler C was not filled, and the jacket was
open to air. Under these operating conditions plugging occurred
again, and this could be attributed to the pump. Due to a larger
temperature difference over the pump-head section, settling
occurred. This settling hampered the flow through the pump
(43) The intra-particle diffusion limitation is determined with the Weisz-
Prater criterion (NW-P). Below 0.3 this limitation can be excluded. In
our case, it has been estimated to be 0.11; Weisz, P. B.; Prater, C. D.
AdV. Catal. 1954, 6, 143.
Vol. 14, No. 2, 2010 / Organic Process Research & Development
•
357

Figure 11. Esterification of pyroglutamic acid with n-decanol; initial experiments with (R,S)-pyroglutamic acid (7 mmol) at 145 °C
in a test tube with magnetic stirring (left) and (R,S)-pyroglutamic acid (97 mmol) at 145 °C in a 50 mL glass vessel with mechanical
stirring (pitched-blade impeller, four blades) (right) applying conventional heating (s9s) and microwave heating (s0s).
Table 5. Esterification of (S)-pyroglutamic acid with
n-decanol at different reaction conditions (experiments A
and B) in the batch-loop-wise operated setup shown in
Figure 7
entry
1
parameters
(S)-pyroglutamic acid
n-decanol
experiment A experiment B
69.1 g
0.54 mol
622 g
69.0 g
0.53 mol
622 g
3.93 mol
10
2
3.93 mol
10
3
4
5
6
7
wt % substrate
flow rate
pump
115 mL ·min-1 343 mL ·min-1
membrane
140 min
gear
178 min
total reaction time
mixing speed R2/suction medium/bottom low/top
line from R2
8
interruption:
plugging
manual
and ultimately the flow stopped. Therefore, a fair comparison
between the batch-loop operated, microwave-heated reactor
system and the conventionally heated, small-scale batch experi-
ments could not yet be made.
4.6. Experiment B. Experiment A was repeated, adopting
the previous procedure. Instead of a membrane pump, a gear
pump was applied in experiment B. Introduction of large
particles relative to the tube diameter led in this case also to
plugging in the suction line. During the first period this plugging
problem could be circumvented by stirring the contents of vessel
R2 slowly. Suction occurred via the top of vessel R2. In this
way the reaction mixture was always saturated in substrate,
although not completely homogeneously mixed.
The Laurydone process can be divided into three stages (see
Figure 12): stage I is the start-up, stage II is that of predominant
esterification, and stage III removes the reaction product water.
Stage I is not appropriate for making a comparison between
conventional and microwave heating.
Microwave heating is much faster than oil-bath heating. For
oil-bath heating it takes more than 10 min to reach the desired
temperature. For the flow experiments heating is, however,
almost instantaneous. In stage II the esterification is predomi-
nantly at a constant temperature, making this stage suitable for
a comparison. Stage III is dominated by the presence of water.
Water removal governs the reaction rate in stage III because it
shifts the equilibrium to the product side. Also the difference
in setup, i.e. batch/batch-loop, plays an important role in the
removal of water.
Figure 12. Esterification of (S)-pyroglutamic acid (10 wt %)
with n-decanol in (experiment B) a batch-loop-operated mi-
crowave-heated reactor setup, (a) conventionally heated batch
experiment at 150 °C and (b) conventionally heated batch
experiment at 130 °C.
For the batch experiment at 150 °C the conversion reached
a plateau of 60% after 90 min. Note that in the reaction in the
batch-loop setup a significant induction period can be observed.
This probably originates from the slow dissolution rate of the
acid in n-decanol. As the rate of esterification for relatively low
conversions obeys a second-order rate where the rate is k2 *
[acid]* [alcohol], low deviation from the saturation concentration
of the acid in the alcohol leads to lower reaction rates than in
the saturated situation which is achieved in batch experiments
(a) and (b). The conversion-time history resulting from the
microwave-heated run in the continuous-flow setup operating
in a batch-loop mode approaches the 60% conversion level
approximately linearly. The batch-loop setup clearly has another
influence on the removal rate of water than the batch operation.
In the Laurydone synthesis comparing conversion rates of batch
and flow processing is more complicated than in the cases of
biocatalysis and racemization. A much higher temperature is
measured outside the tubular reactor R1 than in the other cases.
As a consequence, direct correction with the fractional residence
time in the tubular reactor is hampered due to a significant
continuation of the esterification outside the tubular reactor R1.
Therefore, to make an estimation for the rate in R1, a
correction has been applied with a combination of the fractional
358
•
Vol. 14, No. 2, 2010 / Organic Process Research & Development

Table 6. Conversion rates (interval of 13 and 42% conversion) of the esterification of (S)-pyroglutamic acid under different
reaction conditions
overall reaction rate
fractional residence
timea
relative reaction rate
entry
conditions
[∆conv ·min-1
]
[∆conv ·min-1
]
1
2
3
batch (a), 150 °C
1.14
0.51
0.70
1
1.14
batch (b), 130 °C
1
0.51
experiment B, with correction
0.26
2.68, 1.27b
a Fractional residence time in the reactor compared to the total reaction time [based on the ratio of the volume of the microwave-heated tubular reactor (R1, 180 mL) per
1
total reaction volume (700 mL); overall rate ) VR /Vtot · rate (MW, R1). b [Real rate (MW, R )] ) VR /Vtot · ([overall rate] - VR /Vtot ·[rate, (batch, 130 °C)]); (R2, 520 mL).
1
1
1
Scheme 3. Synthesis of aspirin: O-acetylation of salicylic
acid
residence time and the reaction rate outside the tubular reactor
R1 based on batch (b), see entry 2, Table 6. After correction of
the overall reaction rates, Table 6 demonstrates almost identical
behaviour for batch ((a), 150 °C) or batch-loop operation by
microwave heating. This is, however, only valid for stage II in
the Laurydone process. The stages I and III are controlled by
the heating rate, the dissolution rate of the acid, and the
experimental setup, respectively.
Figure 13. Conversion profile of the aspirin synthesis in acetic
acid (1 M) with 2 mol equiv of acetic anhydride at 120 °C in a
continuous-flow experiment. Volumes of the product stream
during 3 min process time.
5. Performance of the Continuous-Flow Reactor
The heterogeneous processes studied so far in the FlowSynth
demonstrated relatively long reaction times demanding for
multiple passes through the microwave-heated tubular reactor
to reach an acceptable conversion. To get a more complete
picture of the performance of the FlowSynth also a homoge-
neous reaction has been selected. The step to homogeneous
continuous-flow processing has also been reported without
further process research involving a fast reaction.44,45
5.1. Aspirin Synthesis. The one-step synthesis of aspirin
allows investigation of an intrinsically fast, homogeneous
reaction in a potentially single-pass operation (Scheme 3).46
Thus, this aspirin synthesis may be executed as a real continuous
process.
(Figure 13). The mean residence time (390 ( 39 s) was
determined by measuring the volume of the product stream
during 3 min process time. The measurement was repeated
20 times. Also the salicylic acid conversion (57 ( 4%)
was measured in these samples, see Figure 13.
The actual product stream shows a delay which is similar
to the average residence time (the sum of volume of tubing
and cooling unit is approximately also 180 mL). The relatively
longer residence times of fractions 9 and 13 are witnessed as
relatively higher conversions at fractions 11 and 15, respectively.
Batch experimentation using either microwave or conventional
heating with an identical reaction time (390 s) resulted in closely
related conversions of 66% and 64%, respectively.
Starting with a recipe based on a high concentration
(25 wt %; 2.3 mol · L-1) of salicylic acid in acetic acid, at
which the substrate was completely dissolved at 50 °C,
instantaneously led to plugging. Lowering the concentra-
tion of salicylic acid of course improved the handling in
a continuous-flow mode, but reaction rates were negatively
affected. Therefore, a switch from a gear pump to a
membrane pump was necessary to ensure a stable and
efficient reaction flow through the tubular reactor. After
the first 150 mL of the product stream volume (〈τ 〉 ≈ 1)
(R1 ) 180 mL) operation was at steady state. So, residence
time and conversion can be correlated at multiple points
With the assumption that the O-acetylation of salicylic
acid obeys apparently first-order kinetics in the first part
of the reaction, the reaction constant (k) can be estimated
(see eq 11). Together with the model representing the
tubular reactor R1 as a series of CSTRs, the conversion
(xA) related to residence time can be calculated for the
continuous-flow process (eq 11). The total volume (Vtotal
)
of the tubular reactor R1 is 180 mL, and the flow rate
(φν) is 0.46 mL · s-1. When the number (n) of equally sized
tanks in series is 3 or 2, then the theoretical conversion
would be 60 or 57%, respectively; which perfectly agrees
with the experimentally observed conversion (57%).
(44) Bowman, M. D.; Holcomb, J. L.; Kormos, C. M.; Leadbeater, N. E.;
Williams, V. A. Org. Process. Res. DeV. 2008, 12, 41.
(45) Moseley, J. D.; Lenden, D.; Lockwood, M.; Ruda, K.; Sherlock, J.-
P.; Thomson, A. D.; Gilday, J. P. Org. Process Res. DeV 2008, 12,
30.
n
1
xA ) 1 -
Vtotal
1 + k ·
(
)
(11)
n · φν
(46) Drevina, V. M.; Nesterov, V. M. Khim.-Farm. Zh. 1976, 10, 1546.
(47) The k value is obtained from unpublished batch-wise experimentation.
k ) 2.7 · 10-3 s-1[120 oC] (see ref 47)
Vol. 14, No. 2, 2010 / Organic Process Research & Development
•
359

The translation from batch to continuous processing
appears to be straightforward, although more knowledge
(e.g., on the internal temperature of tubular reactor R1) is
required to allow a rational prediction. However, a longer
residence time is not feasible with the current setup, while
complete conversion demands a very long mean residence
time (∼1 h). Presumably, this might be circumvented by
the addition of a suitable catalyst. Nevertheless, the batch-
loop operation allows high conversions in a short time
with a proper temperature control.
mixture; see each specific procedure. Tetradecane was
used as an internal standard.
(R)-1-Phenylethyl Acetate. A 1 L round-bottomed flask
was charged with a solution of (R,S)-1-phenylethanol (48
g, 0.4 mol) and vinyl acetate (172.2 g, 2 mol) in toluene
(700 L). The reaction mixture was circulated via a
membrane pump with a flow of 175 mL · min-1 through
the continuous-flow reactor of Milestone equipped with
a Weflon stirrer and was heated to 70 °C (T1). On top of
the reactor cooling was applied to maintain T2 below 30
°C. Subsequently, Novozym 435 (5.0 g) was added to the
flask (experiment A) or to the tubular reactor (experiment
B, including a temporary hold of the flow), and the set
temperature was maintained for a period of 180 min. The
total reaction volume was 940 mL. Average power
usage during reaction was 226 W (experiment A) and
236 W (experiment B). During the reaction, aliquots
were taken from the liquid phase and measured by GC.
After 180 min the mixture was collected, resampled, and
cooled.
6. Concluding Remarks
The transfer of microwave energy into heat, directly
or indirectly via the Weflon blades, allows the use of
several reaction mixtures in terms of polarity or loss
tangent. The combination of stirring and heating gives rise
to an efficient energy transfer. The residence time
distribution of the CFR (Milestone FlowSynth) with a
single pass results in a performance equivalent to that of
a series of three equally sized CSTRs.
Laurydone [Experiment A]. A double-walled reactor was
charged with (S)-pyroglutamic acid (69.1 g, 0.54 mol) and
n-decanol (622 g, 3.93 mol). The reaction mixture was circulated
via a membrane pump with a flow rate of 115 mL ·min-1
through the continuous-flow reactor of Milestone equipped with
a Weflon stirrer and was heated to 150 °C (T1). On top of the
reactor no cooling was applied. The double-walled reactor was
thermostatic at 60 °C at the walls. The reaction time was 140
min, and the total reaction volume was 700 mL. The average
power usage during reaction was 294 W. During the reaction,
aliquots (∼50 mg) were taken, quenched with dichloromethane
(3 mL), and measured by 1H NMR (CD3OD, 200 MHz) typical
signals δ (ppm) 4.14 (t, 2H, CH2-OCO, product), 3.53 (t, 2H,
CH2-OH, decanol).
Laurydone [Experiment B]. A double-walled reactor was
charged with (S)-pyroglutamic acid (69.0 g, 0.53 mol) and
n-decanol (622 g, 3.93 mol). The reaction mixture was circulated
via a gear pump with a flow rate of 343 mL ·min-1 through
the continuous-flow reactor of Milestone equipped with a
Weflon stirrer and was heated to 150 °C (T1). On top of the
reactor no cooling was applied. The double-walled reactor was
not thermostatic at the walls. The reaction time was 178 min,
and the total reaction volume was 700 mL. The average power
usage during reaction was 444 W. During the reaction, aliquots
(∼50 mg) were taken, quenched with dichloromethane (3 mL),
and measured by 1H NMR (CD3OD, 200 MHz) typical signals
δ (ppm) 4.14 (t, 2H, CH2-OCO, product), 3.53 (t, 2H,
CH2-OH, decanol).
Recalculating the operational costs is necessary to decide
whether microwave heating is still beneficial if larger production
scales are planned. In view of the higher operational costs and
apart from the safety issue the chances to incorporate microwave
heating as a competitive technique seem to decrease with
increasing scale.
As demonstrated in our experiments, plugging as a
result of solid reactants/products is a point of attention in
further process design for solid-liquid organic reactions.
Redesigning the equipment, involving Teflon-coating and
thermostatic tracing, may overcome fouling and plugging
in the present study. Also redesigning the present Flow-
Synth setup allows coping with issues such as solids and
slurry handling.
From a technical point of view scaling up heteroge-
neous reactions by microwave irradiation is feasible. Not
only is it possible to rapidly heat the reaction mixture,
but the experimentally observed batch data of the
conversion-time history can also be used for process
design. Direct temperature control and the opportunity to
quench the reaction easily are highly advantageous. High
production rates can be achieved for batch-wise operation
by numbering up the microwave-heated reactor units.
7. Experimental Section
The continuous-flow reactor was supplied by Milestone srl,
Italy. The tubular reactor could either be equipped with a Teflon
or a Weflon stirrer. The original setup included a membrane
pump of Alldos GmbH (now known under the name Grundfos);
type 281-9, 6-1004; 100 bar; 50 Hz. The gear pump was
supplied by the manufacturer Gather Industries, type 320;
PM8060 L.
General Methods. Chiral gas chromatography (GC) was
performed on a Shimadzu 6C-17A GC equipped with a
Chrompack Chirasil-DEX CG (DF ) 0.25) column, and
a FID. Injection temperature was set at 250 °C and the
detection temperature at 300 °C. Temperature programs
were used to optimize the analysis for each reaction
O-Acetyl Salicylic Acid. A round-bottomed flask was
charged with salicylic acid (207 g, 1.5 mol), acetic acid (1228
g, 1.12 L), and acetic anhydride (307 g, 3.0 mol). After 15 min
of mixing (almost complete dissolution), the reaction mixture
was pumped via a membrane pump with a flow rate of 27
mL·min-1 through the continuous-flow reactor of Milestone
equipped with a Weflon stirrer and was heated to 120 °C (T1).
On top of the reactor, cooling was applied. The residence time
was 6.5 min. The average power usage during reaction was
253 W. During the run, the reaction mixture was collected in
360
•
Vol. 14, No. 2, 2010 / Organic Process Research & Development

fractions, and aliquots were taken and measured by 1H NMR
(CD3OD, 200 MHz) typical signals δ (ppm) 8.01 (d, 1H,
CH-C-COOH, aspirin), 7.84 (d, 1H, CH-C-COOH, sali-
cylic acid).
and Syncom B.V., Groningen, The Netherlands, for their
support.
Supporting Information Available
This material is available free of charge via the Internet at
http://pubs.acs.org.
Acknowledgment
We are grateful to SenterNovem for funding this research.
We thank DSM Pharma Chemicals, Geleen, The Netherlands,
Received for review October 1, 2009.
OP900257F
Vol. 14, No. 2, 2010 / Organic Process Research & Development
•
361