Flash chemistry extensively optimized: High-temperature Swern-Moffatt oxidation in an automated microreactor platform

Article abstract of DOI:10.1002/asia.200900705

The generally accepted benefits of small lateral dimensions of microreactors (1 μm to 1 mm) enable a different way of performing synthetic chemistry: Extremely short contact times in the millisecond range can circumvent the need for performing highly exothermic and fast reactions at very low temperatures. In order to fully exploit this technology, such fast processes need to be redesigned and investigated for optimal reaction conditions, which can differ drastically from the ones traditionally applied. In a comprehensive study, we optimized the selective Swern-Moffatt oxidation of benzyl alcohol to benzaldehyde by varying five experimental parameters, including reaction time and temperature. Employing an ultrashort mixing and reaction time of only 32 ms, the optimal temperature was determined to be 70°C, approximately 150°C higher than in the conventional batch conditions. This remarkable difference shows both the potency of continuousflow chemistry as well as the urgency of a paradigm shift in reaction design for continuous-flow conditions.

Full text of DOI:10.1002/asia.200900705

DOI: 10.1002/asia.200900705
Flash Chemistry Extensively Optimized: High-Temperature Swern–Moffatt
Oxidation in an Automated Microreactor Platform
Pieter J. Nieuwland, Kaspar Koch, Noud van Harskamp, Ron Wehrens,
[
a]
Jan C. M. van Hest,* and Floris P. J. T. Rutjes*
Abstract: The generally accepted bene-
fits of small lateral dimensions of mi-
croreactors (1 mm to 1 mm) enable a
different way of performing synthetic
chemistry: Extremely short contact
times in the millisecond range can cir-
cumvent the need for performing
highly exothermic and fast reactions at
very low temperatures. In order to
fully exploit this technology, such fast
processes need to be redesigned and
investigated for optimal reaction condi-
tions, which can differ drastically from
the ones traditionally applied. In a
comprehensive study, we optimized the
selective Swern–Moffatt oxidation of
benzyl alcohol to benzaldehyde by
varying five experimental parameters,
including reaction time and tempera-
ture. Employing an ultrashort mixing
and reaction time of only 32 ms, the
optimal temperature was determined
to be 708C, approximately 1508C
higher than in the conventional batch
conditions. This remarkable difference
shows both the potency of continuous-
flow chemistry as well as the urgency
of a paradigm shift in reaction design
for continuous-flow conditions.
Keywords: experimental design
·
flash chemistry · flow chemistry ·
microreactors · oxidation
[9]
Introduction
Yoshida, opens up a whole new toolbox for the organic
chemist.
Microreactor technology is rapidly becoming a valuable tool
In order to fully exploit this toolbox, such fast processes
and reactions need to be redesigned and therefore reinvesti-
gated to identify optimal reaction conditions, which can
differ drastically from the ones traditionally applied. Cur-
rently, high flow-through microreactors are generally used
for reaction screening and optimization. Smaller devices,
[1–6]
in synthetic organic chemistry.
It is estimated that 20%
of all synthetic reactions in the pharmaceutical and fine
chemical industries could directly benefit in terms of yield,
selectivity, and efficiency from being carried out in micro-
[7,8]
reactors or, more generally, flow reactors.
The generally
À1
accepted benefits of small lateral dimensions of microreac-
tors (1 mm to 1 mm), such as efficient heat and mass trans-
fer, enable a different way of performing synthetic chemis-
try: Extremely short contact times in the millisecond range
can circumvent the need for performing highly exothermic
and fast reactions at very low temperatures. This concept of
process intensification, aptly termed ꢀflash chemistryꢁ by
however, with typical flow rates in the mLmin range and
hence requiring significantly smaller amounts of material,
are intrinsically considerably more attractive for this pur-
[10,11]
pose.
Once optimal reaction conditions have been deter-
mined on a small scale, flow chemistry allows a scaling-out
procedure due to the high level of reproducibility of the
process.
The Swern–Moffatt oxidation is potentially a highly rele-
vant reaction for industry, because this selective oxidation of
primary alcohols to aldehydes does not require heavy
metals. This reaction, however, is traditionally carried out at
temperatures around À788C, which limits its viability for
high-volume manufacturing. In the conventional procedure,
the activator trifluoroacetic anhydride (TFAA) is first
mixed with dimethylsulfoxide (DMSO) at À788C. After the
reactive sulfonium species 2 has been formed (Scheme 1),
reactant 1 (a primary or secondary alcohol) is added to ini-
tiate the oxidation. In the final step a tertiary amine base
[
a] P. J. Nieuwland, K. Koch, N. van Harskamp, R. Wehrens,
Prof. Dr. J. C. M. van Hest, Prof. Dr. F. P. J. T. Rutjes
Institute for Molecules and Materials
Radboud University Nijmegen
Heyendaalseweg 135, NL-6525 AJ Nijmegen (The Netherlands)
Fax : (+31)24-36-53393
E-mail: j.vanhest@science.ru.nl
f.rutjes@science.ru.nl
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/asia.200900705.
Chem. Asian J. 2010, 5, 799 – 805
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
799

FULL PAPERS
It was previously shown by Yoshida et al. that it is possi-
ble to perform the Swern–Moffatt oxidation in a microreac-
[12]
tor at room temperature. Using a continuous-flow setup,
and by drastically reducing the time between the addition of
TFAA and the alcohol (on the order of 100 ms) the reaction
at room temperature gave results comparable to convention-
al procedures. In addition, a broad study of Swern–Moffatt
oxidations with a range of primary and secondary alcohols
[13]
was performed by Van der Linden et al.,
demonstrating
that premixing DMSO with the alcohol prior to reacting it
with TFAA inside the microreactor led to similar or even
better aldehyde yields and selectivities than conventional
conditions. It can be reasoned that if the alcohol substrate is
present in the DMSO solution, the reactive intermediate
will react in situ with the alcohol so that one mixing step
can be eliminated.
Since these studies indicate that the traditional barriers
for the Swern–Moffatt oxidation have disappeared by apply-
ing microreactor technology, it is now of interest to find the
optimal reaction conditions for this reaction. Herein we de-
scribe a comprehensive study to systematically screen
Swern–Moffatt oxidation parameters using a microreactor
device (internal reactor volume 140 nL), which generates a
large amount of chemical information with only minute
amounts of starting compound. The oxidation of benzyl al-
cohol (1, R=Ph) to benzaldehyde was chosen as the model
reaction, because the intermediate alkoxysulfonium salt 3 is
most prone to undesired solvolytic attack.
Scheme 1. Proposed reaction scheme for the Swern–Moffatt oxidation of
primary alcohol 1 to aldehyde 4.
(
triethylamine or N,N-diisopropylethylamine (DIPEA)) is
added to terminate the reaction, creating the desired alde-
hyde or ketone, and to quench unreacted TFAA. The low
temperatures are required to prevent an important side re-
action: the Pummerer rearrangement. Both reaction inter-
mediates trifluoroacetoxy dimethylsulfonium salt 2 and the
alkoxy dimethylsulfonium salt 3 rearrange at higher temper-
atures, forming either the trifluoroacetyl ester 8 or thio-
methyl ether 6.
Since the effect of multiple parameters was investigated,
multivariate screening was employed instead of the more
commonly used univariate screening. Earlier examples of
employing multivariate screening using microreactor flow
[14]
chemistry were demonstrated by Yoshida et al. The bene-
fit of a multivariate approach includes the detection of pos-
sible dependency between parameters. Because multivariate
experiments tend to require a large number of experiments
when all possible combinations of settings would be
screened, experimental design methodology should be used.
Abstract in Dutch: De algemeen geaccepteerde voordelen
van de zijdelingse dimensies van microreactoren (1 mm tot
1
mm) maken een nieuwe manier van het uitvoeren van syn-
thetische reacties mogelijk: extreem korte contacttijden in
de orde van milliseconden kunnen de noodzaak van het uit-
voeren van zeer exotherme reacties bij lage temperaturen
omzeilen. Om deze technologie ten volle te benutten,
moeten dergelijke snelle processen opnieuw worden ontwor-
pen, en moet opnieuw worden gezocht naar optimale reac-
tiecondities, die nu drastisch kunnen verschillen van de con-
ventionele omstandigheden. In een uitgebreide studie
hebben we de selectieve Swern–Moffatt oxidatie van benzy-
lalcohol naar benzaldehyde geoptimaliseerd door vijf experi-
mentele parameters te variꢃren, waaronder reactietijd en
temperatuur. Door gebruik te maken van zeer korte meng-
en reactietijden, zoals 32 ms, is vastgesteld dat de optimale
reactietemperatuur 70 8C bedraagt, ongeveer 150 8C hoger
dan onder gebruikelijke condities. Dit opmerkelijke verschil
reflecteert enerzijds het potentieel van flowchemie, maar
toont anderzijds ook de noodzaak aan van een drastische
verandering in denkwijze bij het ontwerpen van continue
flowchemie.
[15]
In this paper, D-optimal designs were employed, based on
linear models containing up to cubic terms. Using linear re-
gression, nonsignificant terms were removed from the equa-
tion in a stepwise fashion, which eventually was refit using
only relevant terms. This method was used for visualization
and determination of the optimal reaction conditions. The
selected and simultaneously screened reaction parameters
are listed in Table 1. The yield of the aldehyde was chosen
as the actual goal of the optimization. In commercial chemi-
cal manufacturing, other goals for optimization are typically
space–time yield and overall production rate versus costs.
However, we felt that for simplicity reasons, yield would be
a more appropriate choice for demonstration of the meth-
odꢁs viability. After optimization, the optimal reaction con-
ditions were applied to a larger continuous-flow reactor to
validate the scalability.
800
www.chemasianj.org
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2010, 5, 799 – 805

High-Temperature Swern–Moffatt Oxidation in an Automated Microreactor Platform
Table 1. Overview of experimental parameters.
[
a]
Run
No. of
dimensions
No. of
experiments
Reaction time [s]
TFFA/substrate
stoichiometry
T [8C]
DMSO/substrate
stoichiometry
Substrate conc. [m]
1
2
3
2
3
5
126
0.2–20
0.04–3.55
0.04–3.55
1.0–9.6
1.0–8.0
1.0–8.5
23
25–70
25–70
5
5
0.5
0.5
0.15–0.25
[
b]
55
[
b]
180
2.5–10
[
a] In feeding liquid. [b] D-optimal selections.
Results and Discussion
periments were visualized by local interpolation and genera-
tion of a simple contour plot (Figure 2a). For the calculation
of the interpolation, the simple linear algorithm griddata in
MATLAB (MathWorks, R2007a) with default linear settings
was used. However, in order to locate the optimum value
for reaction yield, curve fitting was required. Results from
third-order two-dimensional curve fitting are shown in Fig-
ure 2b. It must be noted that regulating the stoichiometric
ratio was performed by varying the flow rates of the benzyl
alcohol/DIPEA/DMSO solution and the TFAA solution.
One can argue that owing to different flow rates also the
Microreactor Setup
Schematic representations of the microreactor setups are
shown in Figure 1. All parts within the dotted line consist of
one single glass chip with single-sided wet etched channels
of 55ꢄ120 mm. The mixing time could be held sufficiently
low even without further specific channel geometries: Even
at the highest flow rates the time required for diffusive
mixing in the straight mixing channels was kept well below
the total residence time, mainly owing to low viscosities and
fast diffusive properties of the small molecules. The channel
volumes determining the reaction time, designated as R1
and R2 in setups 1 and 2, respectively, are indicated in
Figure 1.
Run 1: Two-Parameter Optimization: Reaction Time and
TFAA Stoichiometry
In the first optimization run at room temperature, setup 1
was used (Figure 1). Reaction time and TFAA stoichiometry
relative to the alcohol substrate were simultaneously varied
in the range from 0.2 to 20 s and from 2 to 9, respectively,
resulting in two-dimensional plots. The results from 126 ex-
Figure 2. Yield of aldehyde 4 in run 1 at room temperature, shown as
contour plots: a) interpolated response, b) third-order curve fitting,
c) third-order curve fitting on selection of D-optimal design (n=30 ex-
periments).
Figure 1. Microreactor setups 1 and 2, with relevant reactor volumes designated ꢀR1ꢁ and ꢀR2’.
Chem. Asian J. 2010, 5, 799 – 805
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
801

J. C. M. van Hest, F. P. J. T. Rutjes et al.
FULL PAPERS
two reagent streams occupied different volumes in the mi-
croreactor. This could have an influence on the diffusion
time and hence the reaction efficiency. However, repeating
the experiment with different TFAA concentrations but
identical overall stoichiometric ratios yielded similar results
Run 3: Five-Parameter Optimization, Including DMSO
Stoichiometry, and Substrate Concentration
In the final part of the experiment, any possible influences
of DMSO stoichiometry relative to alcohol substrate and
the concentration of the alcohol substrate were also taken
into account, leading us to a five-dimensional optimization
run. A total number of 180 experiments were run. The re-
sults were again modelled following a cubic polynomial ap-
proach. Optimal values for all parameters were found, and
are listed in Table 2.
(
data not shown). Subsequently, a polynomial model was
prepared from a D-optimal selection of the experimental
data (30 experiments), resulting in a very similar plot (Fig-
ure 2c). This confirms the hypothesis that a D-optimal ex-
perimental design can be applied to perform multidimen-
sional reaction screening, thereby drastically decreasing the
required number of experiments.
Table 2. Overview of optimal reaction conditions.
Run 2: Three-Parameter Optimization: Reaction Time,
TFAA Stoichiometry, and Temperature
Temperature
Reaction time
Substrate concentration in reactor
TFAA stoichiometry
DMSO stoichiometry
708C
0.032 s
0.17m
6
In the next step, reaction temperature was investigated as a
third parameter. Conventionally, the Swern–Moffatt oxida-
tion is performed at low temperatures, typically around
À788C. While it was previously shown that the reaction
temperature could be raised to room temperature while re-
taining chemoselectivity, we aimed to increase the reaction
temperature even further and evaluate the reaction perfor-
mance in terms of yield and selectivity.
In Figure 3, the modelled aldehyde yield for experiment 2
is shown, clearly indicating optimal reaction parameters
near 0.5 s reaction time, a TFAA stoichiometry of 7, and a
temperature of 458C. For the linear interpolation the
MATLAB algorithm interp3 with default linear settings was
used. It must be noted that the optimum in reaction time in
this case is somewhat different than observed in the previ-
ous experiment. As shown in the slice plot, however, the re-
action yield is stable over a rather broad range of different
reaction times, from approximately 0.3 s up to several sec-
onds. Thus, it can be concluded that although fast mixing
and a short reaction time are required to prevent the reac-
tive intermediate 2 from decomposing, the alkoxysulfonium
salt 3 is stable on the timescale of a second, even at elevated
temperatures.
9
Residence time unit dimensions (diameterꢄlength)
Total substrate throughput
0.125ꢄ40 mm
0.50 gh
À1
These optimal settings were used to visualize the actual
model of the aldehyde yield (Figure 4, upper matrix for al-
dehyde yield). Each contour plot represents a two-dimen-
sional cross-section of the five-dimensional space. All other
parameters were fixed at the optimal settings.
It is clear that DMSO and TFAA stoichiometry have a
dramatic effect on the reaction rate. The same can be con-
cluded from the reaction time. The two other parameters,
overall reaction concentration and temperature, seem to
have much less effect on the reaction efficiency. Further-
more, significant parameter dependencies between TFAA
and DMSO stoichiometry on the one hand and TFAA stoi-
chiometry and reaction time on the other hand are ob-
served, demonstrating the need for simultaneous multidi-
mensional screening. Reaction time and temperature as
single parameters show that the actual optimal setting is on
the edge of the parametric domain chosen. This means that
Figure 3. Yield of aldehyde 4 in run 2, shown as a three-dimensional slice plot: locally interpolated data (left) and model fit (right).
802
www.chemasianj.org
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2010, 5, 799 – 805

High-Temperature Swern–Moffatt Oxidation in an Automated Microreactor Platform
who reported reduced stability
of the trifluoroacetyl ester in
samples diluted with THF and
water for HPLC, we detected
no instability of this analyte
when dichloromethane was
used as the diluting solvent.
The difference can be explained
by the sample preparation used
by Van der Linden, resulting in
an aqueous basic solution in
which ester hydrolysis occurs
more readily. Because the tri-
fluoracetyl ester 8, presumably
resulting from the Pummerer
rearrangement, was detected at
significant levels, the ratio be-
tween the ester and the desired
aldehyde was chosen as a mea-
sure for reaction selectivity,
while the amount of aldehyde
being formed served as the
standard yield indicator.
In the bottom part of
Figure 4, formation of byprod-
uct 8 is visualized. In most of
the experimental domain, 8 was
formed in very low amounts,
typically less than 2%. At low
DMSO and high TFAA con-
centrations, however, the for-
mation of byproduct 8 steeply
increased. Interestingly, these
results indicate a pathway of
formation of byproduct 8 via
direct esterification of the alco-
hol by TFAA, possibly in com-
bination with esterification of
the alcohol by reaction with the
Pummerer rearrangement prod-
uct 7. It should be noted that
Van der Linden did not screen
the area of low DMSO and
high TFAA concentration, thus
avoiding direct esterification of
alcohol by TFAA.
Because of the integrated re-
Figure 4. Yields of aldehyde 4 (top) and formation of byproduct 8 (bottom) in run 3, shown as matrix contour
plots. Each contour plot is a cross-section of the model space. The set of optimal conditions is used as fixed actor design, both the addition
values for the remaining dimensions in each contour plot, shown as bold lines.
of the alcohol to the activated
sulfonium salt 2, as well as the
final deprotonation by the
somewhat higher aldehyde yields can be expected at even
lower reaction times and higher temperatures. This opens
opportunities for further investigations searching for the ab-
solute limit of reaction efficiency.
Very small amounts of thiomethyl ether 6 were observed,
approaching the detection limits. Unlike Van der Linden,
amine base, take place at the same temperature. Since we
already observed high yields using this design, further inves-
tigation into separating both reaction steps and individually
optimizing their temperatures was considered unnecessary
at this stage, but remains an interesting topic for future re-
search.
Chem. Asian J. 2010, 5, 799 – 805
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
803

J. C. M. van Hest, F. P. J. T. Rutjes et al.
FULL PAPERS
GC-FID equipped with a Quadrex 007 1701 column (length: 10 m, inter-
nal diameter: 0.1 mm, film thickness: 0.1 mm), using a temperature pro-
gram starting at 988C for 0.85 min with subsequent ballistic heating with
a set temperature of 2358C for 1.0 min, a linear flow rate of 1 ms , and
a split ratio of 750. An analysis cycle time of approximately 3 min was
used.
Preparative-Scale Continuous Reaction
The optimal conditions for the 140 nL reactor were trans-
ferred to a 500 nL internal volume microreactor in order to
conduct the same oxidation on a preparative scale. For this
purpose, a standard commercially available stainless steel
continuous-flow reactor with an internal diameter of 125 mm
was selected and the optimal settings from the screening ex-
periments were applied (Table 2). The reaction fluids were
À1
Microreactor setup: All syringes (Harvard apparatus; high-pressure sy-
ringe, 2 mL) mounted on a syringe pump (New Era; type NE-1000 or
NE-500) were connected to FEP tubing (1.59 mm OD, 254 mm ID). At
the end of each tubing, a special ꢀflat-bottom headless nutꢁ (Upchurch
Scientific; type: M 660) was mounted which pressed down onto a flat
bottom ferrule (Upchurch Scientific; type: M 650) to achieve a leak-free
fluid connection to the microreactor. The microreactor was placed in a
continuously pumped through the reactor for approximately
À1
2
h, with a substrate throughput of 0.5 gh . The aldehyde
[
17]
custom-designed chip holder
with threaded holes on the top side in
yield of the outflow was monitored at intervals and always
appeared greater than 96% with only traces of byproduct
based on GC analysis. This confirms that the initially identi-
fied optimal oxidation conditions can also be successfully
applied to a larger microreactor system, while aldehyde
yields compare favorably to those observed by Kawaguchi
et al. (75% yield at 208C) and Van der Linden (84% at
which the nuts were screwed. For temperature control, a custom-designed
heater (Peltier element) was used, which was slid into the microreactor
chip holder and contacted to the microreactorꢁs bottom side. A stainless
steel needle (UpChurch Scientific; type U 106 1/100” ID 1/16” OD,
custom prepared needle tip) was used as outlet.
Gilson 223) was used to dispense all samples during reaction screening.
The pumps, robot, and temperature controller were automatically con-
trolled with custom-designed software program (developed by
A sample robot
(
a
2
08C). Furthermore, the observation that this particular re-
Fraunhofer IMS, Duisburg, Germany).
action is easily scaled up to higher diameter tubing is in line
with the findings of Van der Linden, indicating that mixing
efficiency is not a limitation up to a certain tubing diameter,
even for these ultrafast reactions.
Microreactor: The actual microreactor was fabricated from borosilicate
glass by Micronit Microfluidics BV, Enschede, The Netherlands (HF
etched). Chip dimensions: length 45 mm, width 15 mm, height 2.2 mm.
Channel dimensions: width 120 mm, depth 55 mm, total length 26 or
1
0
320 mm, depending on desired residence time. Reaction volumes were
.14 or 7.02 mL, respectively.
Runs 1 and 2 using setup 1: The first syringe was loaded with liquid A
containing benzyl alcohol 1 (R=Ph; 1.35 g, 12.5 mmol), DMSO (4.88 g,
Conclusions
6
2.5 mmol), and 1-bromo-3,5-dimethylbenzene (2.04 g, 11.0 mmol, inter-
We have shown that it is possible to employ an automated
microreactor platform to optimize a very fast and exother-
mic reaction. Five factors (temperature, substrate concentra-
tion, stoichiometries of two reagents, and reaction time)
were investigated simultaneously in continuous-flow micro-
reactors for optimization of the selective oxidation of benzyl
alcohol to benzaldehyde. Employing a very short mixing
and reaction time of only 32 ms, the optimal reaction tem-
perature was found to be 708C, approximately 1508C higher
than under conventional batch conditions. This remarkable
difference shows both the potency of continuous-flow
chemistry as well as the urgency of a paradigm shift in the
design of chemical reactions when carried out under contin-
uous-flow conditions.
The optimal conditions were also applied to a larger mi-
croreactor system to synthesize the aldehyde product on a
preparative scale. In conclusion, the oxidation could be per-
formed at around 96% conversion in a continuous-flow mi-
croreactor, both on a small and a preparative scale, which
clearly underlines the potential of flow chemistry in organic
synthesis. Furthermore, efficient multivariate screening is re-
quired when dependency between multiple parameters af-
fects reaction efficiency.
nal standard) dissolved in dichloromethane (total volume 25 mL). The
second syringe was loaded with liquid B containing TFAA (5.25 g,
2
5 mmol) and 1,2-dichlorobenzene (1.95 g, 13.3 mmol, internal standard)
dissolved in dichloromethane (total volume 25 mL). The third syringe
was filled with DIPEA (liquid C, neat). Liquid D was prepared by dis-
solving 1-bromonaphthalene (0.1% v/v, internal standard) in dichlorome-
thane. Syringes with liquids A to C were then connected to the micro-
reactor system. Of each reaction mixture, 20 mL was collected in 500 mL
of liquid D. Owing to the varying flow rates, sampling times differed for
every experiment. All reaction conditions were randomized. All samples
were analyzed with GC. Retention times were 0.77, 0.81, 0.91, 1.10, 1.17,
and 1.77 min for benzaldehyde 4, TFA ester 8, 1,2-dichlorobenzene,
benzyl alcohol 1, 1-bromo-3,5-dimethylbenzene, and 1-bromonaphtha-
lene, respectively.
Run 3 using setup 2: The first syringe was loaded with liquid A contain-
ing benzyl alcohol 1 (R=Ph; 2.70 g, 25.0 mmol) and 1-bromo-3,5-dime-
thylbenzene (2.04 g, 11.0 mmol, internal standard) dissolved in dichloro-
methane (total volume 25 mL). The second syringe was loaded with liq-
uid B containing DMSO (9.76 g, 125 mmol) and 1,3,5-trimethylbenzene
(
1.30 g, 10.8 mmol, internal standard) dissolved in dichloromethane (total
volume 25 mL). The third syringe was loaded with liquid C containing
TFAA (21.0 g, 100 mmol) and 1,2-dichlorobenzene (1.95 g, 13.3 mmol, in-
ternal standard) dissolved in dichloromethane (total volume 25 mL). The
fourth syringe was loaded with liquid D containing 1,3-dimethylnaphtha-
lene (1.47 g, 9.43 mmol, internal standard) dissolved in dichloromethane
(
total volume 25 mL). The fifth syringe was filled with DIPEA (liquid E,
neat). Liquid F was prepared by dissolving 1-bromonaphthalene (0.1%
v/v, internal standard) in dichloromethane. Syringes with solutions A to E
were then connected to the microreactor system. Of each reaction mix-
ture, 20 mL was collected in 500 mL of liquid F. Owing to the varying flow
rates, sampling times differed for every experiment. All reaction condi-
tions were randomized. All samples were analyzed with GC. Retention
times were 0.63, 0.77, 0.81, 0.91, 1.10, 1.17, 1.63, and 1.77 min for 1,3,5-tri-
methylbenzene, benzaldehyde 4, TFA ester 8, 1,2-dichlorobenzene,
benzyl alcohol 1, 1-bromo-3,5-dimethylbenzene, 1,3-dimethylnaphthalene,
and 1-bromonaphthalene, respectively.
Experimental Section
GC analysis: All GC analyses were performed off-line. The effluent of
the microreactor was diluted using dichloromethane marked with an in-
ternal standard in order to constantly monitor flow rates as previously
[
16]
demonstrated.
GC analysis was performed on a Shimadzu GC 2010
804
www.chemasianj.org
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Asian J. 2010, 5, 799 – 805

High-Temperature Swern–Moffatt Oxidation in an Automated Microreactor Platform
Reaction in the larger-scale continuous-flow system: A stainless steel
tubing microreactor (IDEX, Oak Harbor WA, internal diameter 125 mm,
internal volume 0.50 mL between two mixers) was used in combination
with two commercially available T-junctions (IDEX, Oak Harbor WA)
acting as mixers, analogous to microreactor setup 1. The reactor was sub-
merged in an oil bath and the three inlets of the T-junctions were con-
nected to the syringes. The following solutions were prepared: Liquid A:
benzyl alcohol 1 (R=Ph; 3.68 g, 34.1 mmol), DMSO (23.9 g, 306 mmol),
and 1,3,5-trimethylbenzene (3.04 g, 25.3 mmol, internal standard) dis-
solved in dichloromethane (total volume 100 mL). Liquid B: TFAA
[4] A. R. Bogdan, S. L. Poe, D. C. Kubis, S. J. Broadwater, D. T.
McQuade, Angew. Chem. 2009, 121, 8699–8702; Angew. Chem. Int.
Ed. 2009, 48, 8547–8550.
[5] P. Watts, C. Wiles, Chem. Commun. 2007, 443–467.
[6] E. V. Rebrov, E. A. Klinger, A. Berenguer-Murcia, E. M. Sulman,
J. C. Schouten, Org. Process Res. Dev. 2009, 13, 991–998.
[7] D. M. Roberge, L. Ducry, N. Bieler, P. Cretton, B. Zimmermann,
Chem. Eng. Technol. 2005, 28, 318–323.
[8] N. Kockmann, M. Gottsponer, B. Zimmermann, D. M. Roberge,
Chem. Eur. J. 2008, 14, 7470–7477.
(
42.8 g, 204 mmol) dissolved in dichloromethane (total volume 100 mL).
[9] a) J. Yoshida, Chem. Commun. 2005, 4509–4516; b) J. Yoshida, A.
Nagaki, T. Yamada, Chem. Eur. J. 2008, 14, 7450–7459.
[10] H. Pennemann, P. Watts, S. J. Haswell, V. Hessel, H. Lowe, Org.
Process Res. Dev. 2004, 8, 422–439.
[11] D. M. Ratner, E. R. Murphy, M. Jhunjhunwala, D. A. Snyder, K. F.
Jensen, P. H. Seeberger, Chem. Commun. 2005, 578–580.
[12] T. Kawaguchi, H. Miyata, K. Ataka, M. Kazuhiro, J. Yoshida,
Angew. Chem. 2005, 117, 2465–2468; Angew. Chem. Int. Ed. 2005,
44, 2413–2416.
Liquid C: DIPEA, neat. The flow rates of pumps A, B, and C were set to
4
time of 0.032 s. After stabilizing the system for 1 min, the outflow was
collected for 127 min. Subsequently, the effluent collected was worked up
by the following procedure: The mixture (~150 mL) was diluted with di-
chloromethane (200 mL) and washed with 1m HCl (2ꢄ200 mL) and
À1
50, 450, and 300 mLmin , respectively, corresponding to a total reaction
4
brine (150 mL). The organic phase was dried on MgSO and concentrated
in vacuum. The residue was purified by standard flash chromatography
using 3% (v/v) diethyl ether in pentane as eluent and concentrated in
vacuum to yield 1.70 g of benzaldehyde (4, R=Ph) as a colorless liquid.
[13] J. J. M. van der Linden, P. W. Hilberink, C. M. P. Kronenburg, G. J.
Kemperman, Org. Process Res. Dev. 2008, 12, 911–920.
[
14] a) H. Usutani, Y. Tomida, A. Nagaki, H. Okamoto, T. Nokami, J.
Yoshida, J. Am. Chem. Soc. 2007, 129, 3046; b) A. Nagaki, Y.
Tomida, H. Usutani, H. Kim, N. Takabayashi, T. Nokami, H. Oka-
moto, J. Yoshida, Chem. Asian J. 2007, 2, 1513; c) A. Nagaki, H.
Kim, J. Yoshida, Angew. Chem. 2008, 120, 7951; Angew. Chem. Int.
Ed. 2008, 47, 7833.
Acknowledgements
This research was carried out with financial aid from EUREGIO Rhine-
Waal (Interreg IIIA). M. van der Linden, P. Hilberink, Dr. C. M. P. Kro-
nenburg, and Dr. G. Kemperman (Schering-Plough, Oss, the Nether-
lands) are kindly acknowledged for valuable discussions and fruitful col-
laboration.
[
[
[
15] P. F. de Aguiar, B. Bourguignon, M. S. Khots, D. L. Massart, R.
Phan-Than-Luu, Chemom. Intell. Lab. Syst. 1995, 30, 199–210.
16] P. J. Nieuwland, K. Koch, J. C. M. v. Hest, F. P. J. T. Rutjes, The
Open Chem. Eng. J. 2008, submitted.
17] H. C. Trieu, J. Slotkowski, R. Klieber, J. C. M. v. Hest, F. P. J. T.
Rutjes, K. Koch, P. J. Nieuwland, P. Wiebe, Chip holder, fluidic
system and chip holder system. PCT/EP/2006/010299, 2006.
[
1] R. L. Hartman, K. F. Jensen, Lab Chip 2009, 9, 2495–2507.
2] A. Palmieri, S. V. Ley, K. Hammond, A. Polyzos, I. R. Baxendale,
Tetrahedron Lett. 2009, 50, 3287–3289.
[
Received: December 9, 2009
[
3] K. Geyer, T. Gustafsson, P. H. Seeberger, Synlett 2009, 2382–2391.
Published online: February 17, 2010
Chem. Asian J. 2010, 5, 799 – 805
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemasianj.org
805