Continuous flow production of thermally unstable intermediates in a microreactor with inline IR-analysis: Controlled Vilsmeier-Haack formylation of electron-rich arenes

Article abstract of DOI:10.1021/op2003437

The Vilsmeier-Haack formylation of aromatic compounds is a well-established process in organic synthesis, largely driven by the fact that the resulting aldehydes are generally useful intermediates for the synthesis of fine chemicals and pharmaceutical products. Industrial-scale production, however, is often hampered by laborious procedures requiring the use of hazardous chemicals to produce the highly reactive intermediates. In order to circumvent these issues, a flow chemistry approach was developed. This article describes the design and semiautomated optimization of the Vilsmeier-Haack formylation in continuous flow and subsequent scale-up to preparative volumes in an intrinsically safe manner.

Full text of DOI:10.1021/op2003437

Article
pubs.acs.org/OPRD
Continuous Flow Production of Thermally Unstable Intermediates in
a Microreactor with Inline IR-Analysis: Controlled Vilsmeier−Haack
Formylation of Electron-Rich Arenes
†
‡
Becker, Mariel
le M. E. Delville,‡
†
Sebastiaan (Bas) A. M. W. van den Broek, Jeroen R. Leliveld, Rene
́
̈
,†
†
,‡
Pieter J. Nieuwland,* Kaspar Koch, and Floris P. J. T. Rutjes*
†
FutureChemistry Holding BV, Toernooiveld 100, 6525 EC Nijmegen, The Netherlands
‡
Institute for Molecules and Materials, Radboud University Nijmegen, Heyendaalseweg 135, 6525 AJ Nijmegen, The Netherlands
S Supporting Information
*
ABSTRACT: The Vilsmeier−Haack formylation of aromatic compounds is a well-established process in organic synthesis,
largely driven by the fact that the resulting aldehydes are generally useful intermediates for the synthesis of fine chemicals and
pharmaceutical products. Industrial-scale production, however, is often hampered by laborious procedures requiring the use of
hazardous chemicals to produce the highly reactive intermediates. In order to circumvent these issues, a flow chemistry approach
was developed. This article describes the design and semiautomated optimization of the Vilsmeier−Haack formylation in
continuous flow and subsequent scale-up to preparative volumes in an intrinsically safe manner.
1
. INTRODUCTION
inline formation of the formylating agent followed by the
addition to the arene and hydrolysis to the aldehyde. The
combined process has been validated by successful application
to a series of substrates.
The methodologies developed throughout the years for
formylation of aromatics are as numerous as they are diverse,
indicating not only the intrinsic value of aromatic aldehydes as
synthetic intermediates but also the lack of generally applicable
routes to synthesize these aldehydes. Hence, a large variety of
relatively specific syntheses are well-established; however, they
2
. RESULTS AND DISCUSSION
We chose to investigate the formylation of pyrrole as a
representative case for VH formylation in a continuous flow
process (Scheme 1). In batch, 2-formylpyrrole (6) is prepared
1
−6
appear rather laborious.
often required, rendering scaling up problematic. The
Moreover, hazardous chemicals are
7
Vilsmeier−Haack (VH) reaction is nowadays frequently used
8
−12
13
Scheme 1. Vilsmeier−Haack formylation of pyrrole
for formylation of electron-rich arenes
and ketones, and
1
4
in addition is used in cyclohaloaddition, cyclization
1
5
16
reactions, and ring annulations. In the VH reaction a
chloroiminium ion is formed as the reactive species by mixing a
17
substituted amide with phosphorous oxychloride. The highly
electrophilic iminium ion then reacts with the arene, after which
basic hydrolysis with sodium hydroxide leads to the aldehyde.
Albeit that the intermediate chloroiminium ion can be readily
prepared, calorimetric studies have demonstrated that its
formation poses specific thermal hazards due to thermal
instability and can generate high and fast temperature rises
7
when heated, possibly resulting in a thermal runaway. This
requires active cooling, and the latter makes batchwise scale-up
in a three-step procedure, first allowing the chloroiminium ion
3 to be formed from DMF at low temperature, generally at 0
°C. In a second step, pyrrole is added at elevated temperatures
18
rather troublesome. We envisioned that a flow chemistry
approach would enable the VH formylation at industrial scale
with enhanced safety, because in a microreactor the reaction
takes place on a microliter scale, allowing excellent heat
dissipation and thereby rendering active cooling of the system
superfluous.
(
typically 60 °C), while in a third step the resulting iminium
adduct 5 is hydrolyzed to give product 6 upon heating in a
20
sodium acetate or sodium hydroxide solution.
In a continuous flow process, using FutureChemistry’s
Flowstart B-200, DMF (1, syringe A) and phosphorous
In 2010, Kim et al. briefly demonstrated the feasibility of VH
formylation of N,N-dimethylaniline as a proof for their newly
19
developed flexible polyimide (PI) film microreactor. On the
basis of these results, we decided to extensively investigate the
opportunities for a widely applicable flow method for
formylation of electron-rich arenes. Our approach involves
Special Issue: Continuous Processes 2012
Received: November 28, 2011
Published: January 24, 2012
©
2012 American Chemical Society
934
dx.doi.org/10.1021/op2003437 | Org. Process Res. Dev. 2012, 16, 934−938

Organic Process Research & Development
Article
Figure 1. Schematic drawing of the microreactor setup.
Figure 2. UV absorption spectra of the reaction species (A) and formation of the VH reagent in MeCN (B) at 304 nm.
2
3
Figure 3. (A) Zoom-in of the Flowchemistry setup with the FutureChemistry FlowStart Evo (B-400 and the Mettler Toledo Autochem FlowIR).
−1
−1
(
B) IR absorbance spectra showing the two main bands of interest at 769 cm (C−Cl) and 804 cm (P−O−C) changing as a function of residence
time.
oxychloride (POCl , syringe B) were mixed using a T-splitter
preparative-scale method. A schematic drawing of the micro-
reactor setup is shown in Figure 1. The reactor has an internal
volume of 92 μL, a channel width of 600 μm, a channel depth
of 500 μm, and an effective channel length of 360 mm. The
channel layout contains two additional mixing units (M) being
of the folding flow type. The reactor temperature was
controlled by Peltier elements and sensed by a Pt1000
temperature sensor.
3
(
T) and transferred into a coil (C1) to allow the chloroiminium
ion (3, Vilsmeier−Haack reagent) to be formed. An excess of
DMF was used to ensure full conversion of POCl and to retain
3
the intermediate in solution. The reactive intermediate 3 was
then delivered at a constant flow rate into the temperature-
controlled microreactor (MR) together with pyrrole and
toluene (internal standard, syringe C) for the formylation
reaction to proceed. Finally, the reaction was quenched with a
21
Initial experiments showed that it is critical to reach full
mixture of H O and ethanol (syringe Q). In contrast to known
conversion of POCl in the coil, because it reacts vigorously
2
3
procedures, as described in the previous section, the resulting
iminium ion 5 was hydrolyzed in a solution of NaOH (0.27 M)
and ethanol. Ethanol was added to ensure solubility of the
internal standard (toluene), which is not necessary for a
with pyrrole to form polymers, thereby clogging the reactor. An
atline spectrophotometric method was developed to determine
the reaction time for inflow formation of the chloroiminium ion
3 at room temperature. Measuring the reaction time without
9
35
dx.doi.org/10.1021/op2003437 | Org. Process Res. Dev. 2012, 16, 934−938

Organic Process Research & Development
Article
solvent proved to be impossible, since DMF shows UV-
absorbance around 300 nm; being used as a solvent, its UV
absorbance overlaps with that of the iminium ion (Figure 2A).
The absorbance was therefore measured at low concentrations
of DMF, using acetonitrile as the solvent. Under these
conditions, UV absorbance at a wavelength of 304 nm showed
that formation of the chloroiminium ion 3 was completed
within 90 s (Figure 2B).
though the quench of the intermediate iminium ion could be
performed in the microreactor using a mixture of water and
ethanol, thus well defining the actual residence time for the
addition step, the resulting iminium product had to be
hydrolyzed by sodium hydroxide in water and ethanol in the
collection vials due to repetitive clogging of the outlet of the
microreactor. Experiments were performed in random order,
and the collected samples were analyzed using HPLC (Figure
5). Full conversion was observed with a reaction time of 180 s,
a temperature of 60 °C, and a molar ratio of 1.5, which we
considered optimal conditions. It may be clear from the model,
however, that various other sets of optimal parameters can also
be selected, depending on the demands one has to satisfy.
Now being able to reach a full conversion, different solvents
were investigated to reduce the use of DMF. The results are
presented in Table 1, showing that there is an evident
The main issue we encountered during the UV measure-
ments was the inability to perform real-time analysis of the
unstable intermediate without the use of additional solvents
(
e.g., MeCN). This problem was thought to be overcome by
22
connecting inline a Mettler Toledo FlowIR infrared flow cell
22
with the outlet of the microreactor. The data obtained from
this inline infrared analysis can provide insight not readily
available from other sources.
In order to identify spectral bands unique to the intermediate
species infrared spectra of both substrates DMF and POCl3
were recorded using the FlowIR system. Next, the formation of
the VH reagent was visualized by the relative intensity of a
Table 1. Solvent effect
a
entry
solvent
yield (%)
−1
1
2
3
4
5
6
DMF
>99
92
98
79
74
69
characteristic stretching vibration (C−Cl) at 769 cm (Figure
1,2-dichloroethane
acetonitrile
THF
3
), measured at fixed reaction times. The results show an
initially fast formation of the phosphonium salt 2 (green line,
Figure 4), as identified by a characteristic bending frequency
EtOAc
diglyme
a
Conditions: t = 180 s, T = 60 °C, MR = 1.5
r
difference in the conversion to 2-formylpyrrole (6) in different
solvents. The results indicate comparable reaction rates for
polar solvents such as DMF (entry 1) and acetonitrile (entry 3)
compared to the less polar THF, EtOAc, and diglyme (entries
4
−6). Even though the reaction rate is slightly lower in the
latter solvents, they clearly show additional value by being less
toxic.
Figure 4. Reaction times for the formation of the phosphonium salt 2
and the VH reagent.
Scaling up of the Vilsmeier−Haack formylation in batch
reactors generally poses thermal runaway threats due to the
highly exothermic formation of the reactive species. In flow
chemistry, scaling out is a one-to-one process in most cases,
eliminating the need for active cooling. Using optimal
conditions obtained from the optimization data, a preparative
synthesis of 2-formylpyrrole (6) resulted in the continuous
production at a 6.0 g/h rate (see: Experimental Section), using
the Uniqsis FlowSyn (FCUQ-1020) with a Uniqsis FlowSyn
(FCUQ-1020) equipped with a glass microreactor (0.65 mL)
containing folding flow type mixing units and a 2.4 mL stainless
steel coil reactor with an internal diameter of 1 mm. The
Vilsmeier−Haack reagent was formed in the glass microreactor
with a reaction time of 97.5 s, and the outflow was directly
pumped into the coil for the oxidation of the pyrrole to
−1
(
P−O−C, 804 cm ). The formation of the Vilsmeier−Haack
reagent 3 (blue line, Figure 4) shows a rapid increase between
0 and 100 s reaction time, which may be caused by increasing
5
chloride concentration due to decomposition of the dichlor-
ophosphate counterion of 3 (eq 1, Scheme 1). From these
results, it was concluded that the formation of the VH reagent
was completed within 90 s.
Next, a full optimization study
basis of reaction time (t = 10, 30, 60, 120, 180, 300 s),
temperature (T = 0, 20, 40, 60 °C), and molar ratio of POCl3
to the pyrrole substrate (MR = 1, 2, 3, 4). All experiments were
performed with a fixed reaction time of 90 s for the formation
of the chloroiminium ion 3 unless stated otherwise. Even
24−26
was designed on the
r
Figure 5. Univariate screening of reaction parameters. Fixed values were a reaction time of 180 s, a temperature of 20 °C, and a molar ratio of 1.5.
9
36
dx.doi.org/10.1021/op2003437 | Org. Process Res. Dev. 2012, 16, 934−938

Organic Process Research & Development
Article
Table 2. Preparative-scale experiments
a
b
c
Conditions: t = 180 s, T = 60 °C, MR = 1.5. Conditions: t = 180 s, T = 100 °C, MR = 1.5. Conditions: t = 360 s, T = 100 °C, MR = 1.5.
r
r
r
proceed. Quenching of the reaction was not required now since
the reaction was driven to full completion. The intermediate
iminium species was therefore continuously hydrolyzed in a
solution of NaOH (0.27 M) for 60 min and subsequently
iminium ion 3 was hydrolyzed in NaOH/EtOH/H O (1 mL,
2
0.27 M, 1:1, v/v) to ensure solubility of the internal standard.
Scale-Up Reaction. A scale-up experiment was performed
in a Uniqsis FlowSyn (FCUQ-1020) equipped with a glass
microreactor (0.65 mL) containing folding flow type mixing
units and a 2.4 mL stainless steel coil reactor with an internal
diameter of 1 mm. With flow A of 0.1 mL/min and flow B of
0.3 mL/min, a reaction time for formation of the VH reagent of
97.5 s was obtained. It was reacted in the coil with a solution of
pyrrole (4, 7.40 mL, 106.8 mmol) in DMF (32.60 mL). With a
total flow rate of 0.8 mL/min, a reaction time of 180 s was
obtained. After a stabilization time of 10 min, the product was
collected for 60 min. In contrast to the optimization, no inline
quench was used, because the reaction was driven to full
conversion now. Instead, the outflow of the reactor was directly
extracted with Et O to yield 5.98 g (98%). The yield obtained
2
was similar to that found in the optimization using a reactor
with an internal volume of 92 μL.
On the basis of the optimal conditions obtained with pyrrole,
a range of different substrates were screened (Table 2). The
conversion of the reactions was determined by HPLC and
confirmed by high resolution mass spectrometry. Isolated yields
are depicted in Table 2. It was concluded that amine-
substituted arenes (entries 2−5) are highly reactive towards
Vilsmeier−Haack formylation, all showing high or complete
conversion within 180 s. Di- and triformylation of triphenyl-
amine 12 proved to be less effective, resulting in extended
reaction times at elevated temperatures to reach full conversion.
Anisole (6, entry 1) appeared much less reactive than its amine
substituted derivative 8 since only trace amounts of product
were observed.
collected in a stirred solution of NaOH/H O (400 mL, 0.27
2
M). The product collected was extracted with Et O (3 × 200
2
mL). The organic layer was dried over MgSO , filtered, and
4
concentrated in vacuo (700 mbar, 40 °C) to yield 5.98 g of 2-
formylpyrrole (6, 98%) as a viscous oil which solidified upon
standing at −18 °C for 16 h. Purity (93%) was based on NMR
27
3. EXPERIMENTAL SECTION
measurements.
UV Analysis. To a UV vial containing acetonitrile (1 mL)
Reaction Optimization. A FutureChemistry FlowScreen
C-300) was used to perform the reaction parameter screening.
was added POCl (5 μL, 55 μmol) and DMF (5 μL, 65 μmol).
(
3
Four glass syringes with an internal volume of either 1 or 5 mL
were used, as indicated in Figure 1. Pumps 1 and 2 were loaded
The vial was stoppered, quickly shaken, and immediately
inserted into a UV/vis spectrometer. The spectrometer
measured the UV spectrum with 15 s intervals. In between
the measurements, the vial was taken out of the spectrometer
and shaken to homogenize the sample and promote the
reaction.
IR Analysis. A FutureChemistry FlowStart Evo (B-400) was
placed in line with the FlowIR (Mettler Toledo) to analyze the
formation of the VH reagent. Two glass syringes with an
with POCl and DMF, respectively, for the formation of the
3
Vilsmeier−Haack reagent. Pump 3 contained a solution of
pyrrole (6, 1.85 mL, 26.7 mmol) and toluene (1.91 mL,
internal standard) in DMF with a total volume of 10 mL. In
order to quench the reaction at the end of the microreactor,
ensuring well-defined reaction times, pump 4 contained a
solution of EtOH/H O (5 mL, 1:1, v/v). The intermediate
2
9
37
dx.doi.org/10.1021/op2003437 | Org. Process Res. Dev. 2012, 16, 934−938

Organic Process Research & Development
Article
internal volume of 5 mL were used, and pumps 1 and 2 were
(12) Marson, C. M.; Giles, P. R. Synthesis Using Vilsmeier Reagents;
CRC: Boca Raton, 1994.
loaded with POCl and DMF. A reactor with an internal
3
(
13) Marson, C. M. Tetrahedron 1992, 48, 3659−3726.
volume of 100 μL was connected to the FlowIR with a total
system volume of 120 μL, including the flow cell of the FlowIR.
Flows were fixed to corresponding reaction times, and IR
spectra were recorded continuously.
HPLC Analysis. Atline HPLC-analysis was performed, and
samples were loaded into the injector of a Shimadzu HPLC
system (Adsorbosphere C18 column (length: 100 mm, ID: 4.6
mm)) with a flow rate of 1.0 mL/min; acetonitrile/water: initial
(
14) (a) Fujisawa, T.; Lida, S.; Sato, T. Chem. Lett. 1984, 1173−1176.
(
(
b) Newman, M. S.; Sujeeth, P. K. J. Org. Chem. 1978, 43, 4367−4369.
c) Boeckman, R. K. Jr.; Ganem, B. Tetrahedron Lett. 1974, 15, 913−
17.
9
(
15) Majo, V. J.; Perumal, P. T. Tetrahedron Lett. 1997, 38, 6889−
6
892.
(16) Rao, M. S. C.; Rao, G. S. K. Indian J. Chem., Sec. B. 1988, 27,
213−216.
(
1
40/60); 4.0 min (40/60); 12.5 min (90/10); 13.0 (40/60);
7.5 (40/60), using a UV detector with analysis channels at 210
(17) Jugie, G.; Smith, J. A. S.; Martin, G. J. J. Chem. Soc., Perkin Trans.
2
1975, 925−928.
18) Dyer, U. C.; Henderson, D. A.; Mitchell, M. B.; Tiffin, P. D. Org.
Process Res. Dev. 2002, 6, 311−316.
19) Min, K.; Lee, T.-H.; Park, C. P.; Wu, Z.-Y.; Girault, H. H.; Ryu,
I.; Fukuyama, T.; Mukai, Y.; Kim, D. Angew. Chem., Int. Ed. 2010, 49,
063−7067. Unfortunately, this article does not describe any process
details.
(
and 254 nm. Accurate flow rates of the substrate in the
optimization experiments were calculated using toluene as an
internal standard and determined with our recently developed
(
28
flow marker methodology.
7
4. CONCLUSION
(20) Noland, W. E.; Lanzatella, N. P.; Sizova, E. P.; Venkatraman, L.;
Afanasyev, O. V. J. Heterocycl. Chem. 2009, 46, 503−534.
It has been demonstrated that the Vilsmeier−Haack
formylation can be readily performed in a continuous flow
microreactor system. A full optimization study was performed
(21) MacInnes, J. M.; Vikhansky, A.; Allen, R. W. K. Chem. Eng. Sci.
2
(
007, 62, 2718−2727.
22) For more detailed information on the FlowIR, see www.mt.
com/FlowIR.
23) For information on the FlowStart Evo, see www.futurechemistry.
com.
involving reaction time (t ), temperature (T) and molar ratio
r
(
MR) as process parameters. Optimal conditions were obtained
(
at a reaction time of 180 s at 60 °C with a molar ration of 1.5
and DMF as a solvent. Other less toxic solvents also showed
remarkably high conversions within 180 s. Under the optimized
conditions, scale-up of the reaction has been successfully
realized with a continuous production of 2-formylpyrrole of
(24) Nieuwland, P. J.; Segers, R.; Koch, K.; van Hest, J. C. M.; Rutjes,
F. P. J. T. Org. Process Res. Dev. 2011, 15, 783−787.
(25) Nieuwland, P. J.; Koch, K.; van Harskamp, N.; Wehrens, R.; van
Hest, J. C. M.; Rutjes, F. P. J. T. Chem. Asian J. 2010, 10, 799−805.
(26) Delville, M. M. E.; Nieuwland, P. J.; Janssen, P.; Koch, K.; van
5
.98 g/h. The newly developed process was also shown to be
Hest, J. C. M.; Rutjes, F. P. J. T. Chem. Eng. J. 2010, 167, 556−559.
more widely applicable by the successful formylation of several
amine-substituted arenes.
1
(
27) H NMR: δ 10.75 (br s, 1H), 9.48 (s, 1H), 7.18 (td, J = 1.3, 2.5
Hz, 1H), 7.01 (dd, J = 1.4, 3.9 Hz, 1H), 6.35 (dd, J = 2.5, 3.9 Hz, 1H).
1
H NMR data are in agreement with literature: Choy, J.; Jaime-
ASSOCIATED CONTENT
■
Figueroa, S.; Jiang, L.; Wagner, P. Synth. Commun. 2008, 38, 3840−
3853.
(28) Nieuwland, P. J.; Koch, K.; van Hest, J. C. M.; Rutjes, F. P. J. T.
Open Chem. Eng. J. 2010, 4, 61−67.
*
S
Supporting Information
More detailed information on GC retention times and reaction
data. This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
■
*
Corresponding Author
P.Nieuwland@futurechemistry.com; F.Rutjes@science.ru.nl
ACKNOWLEDGMENTS
■
Dr. Jon Goode from Mettler Toledo Autochem is kindly
acknowledged for his fruitful collaboration in the loan of an
inline IR-monitoring system and the data analysis.
REFERENCES
■
(
(
(
1) Gattermann, L.; Koch, J. A. Ber. 1897, 30, 1622−1624.
2) Vilsmeier, A.; Haack, A. Ber. 1927, 60, 119−122.
3) Rahm, A.; Guilhemat, R.; Pereyre, M. Synth. Commun. 1982, 12,
4
(
(
(
(
(
85−487.
4) Duff, J. C.; Bills, E. J. J. Chem. Soc. 1932, 1987−1988.
5) Reimer, K.; Tiemann, F. Ber. 1876, 9, 824−828.
6) Rieche, A.; Gross, H.; Hoft, E. Chem. Ber. 1960, 91, 88−91.
7) Bollyn, M. Org. Process Res. Dev. 2005, 9, 982−996.
8) Meth-Cohn, O.; Stanforth, S. P. In Comprehensive Organic
Synthesis; Trost, B. M.; Fleming, I., Eds.; Pergamon: Oxford, 1991; Vol.
2
(
(
, pp 777−794.
9) Lai, G.; Xiu, R.; Santos, J.; Mintz, E. A. Synlett 1997, 1275−1278.
10) Perumal, P. T.; Amaresh, R. R. Tetrahedron 1999, 55, 8083−
8094.
(
11) Pedras, M. S. C.; Zaharia, I. L. Org. Lett. 2001, 3, 1213−1216.
9
38
dx.doi.org/10.1021/op2003437 | Org. Process Res. Dev. 2012, 16, 934−938