Scheme 1
not needed when operating below the solvent boiling point. The
entire system is monitored and controlled by computer interface,
wherein data can also be collected (Figure 1).
The FlowSYNTH can operate at temperatures up to 230 °C
and pressures up to 30 bar (435 psi).8 The pump can operate at
flow rates between 10 and 200 mL/min i.e. up to one column
volume per minute or ∼12 L/h. This means the FlowSYNTH
can process significant quantities of materials per day. Assuming
24 h operation, ∼300 L could be processed per day, which for
a typical 10 wt %/volume pharmaceutical process would
represent nearly 30 kg of material processed. In principle, the
FlowSYNTH can process slurries when operating below the
solvent boiling point. However, in practice, the back pressure
regulator is required when operating near or above the solvent
boiling point, and when other volatile reaction components are
present. In these cases, slurries cannot be processed because
the back pressure regulator contains a frit which is easily
blocked by solids (or by solids forming during the reaction).
In our initial evaluation,7 we studied the unimolecular
homogeneous Newman-Kwart rearrangement (NKR).9 This
established the baseline performance for this reactor and also
provided a comparison to other larger-scale microwave
reactors.4a We have now expanded the scope of this work by
investigating more challenging reactions. For this purpose, we
chose the following reactions: an ortho Claisen rearrangement,
an acid-catalysed benzofuran formation, an alkylation reaction,
a Heck reaction, and a nucleophilic aromatic substitution (SNAr)
reaction (making six in total with the NKR). These reactions
covered a range of physical parameters, particularly the
important aspects of second-order and heterogeneous reactions,
and all were related to previous AstraZeneca projects to ensure
they were of relevance to potential pharmaceutical manufacture.
They had also been studied at AstraZeneca in an unrelated stop-
flow microwave reactor,10 which provided a useful comparison.
Each reaction is discussed separately below. In addition, some
energy consumption data is also presented.
was then introduced, but samples were not taken until one
column volume had passed (200 mL, 13.3 min), due to slight
dilution from some unavoidable back-mixing in the reaction
chamber despite assumed plug flow characteristics. Analysis
at 10-min intervals over 40 min showed stable and complete
reaction conversion of O-allyl ether 2 to C-allyl ether product
3, with no degradation to 1-naphthol (1), which can occur under
over-harsh conditions. Complete conversion was a benefit of
the slightly longer reaction time, although a few minor impuri-
ties were also detected, presumably from over-reaction, but the
overall quality was very similar to that from the previous
result.10 The flow rate was very stable, with no blockages
occurring during the >1 h run, and the production rate was 3.0
mol/h at these dilutions.
Naphthofuran Formation. The bulk of the reaction mixture
from the previous reaction (crude 3) was used directly in the
formation of naphthofuran (4) (Scheme 1), which had required
only 10 min at 100 °C in a large-scale microwave batch
reactor.12 Two volumes of formic acid (based on input 2) were
added to the crude reaction mixture containing 3. Unfortunately,
this resulted in the formation of two phases, almost certainly
due to the presence of the DCB, since formic acid and neat 3
had been a single phase in all previous preparations. Although
dual feed of two liquid phases is in principle possible in the
FlowSYNTH, separation of the phases in the vertical column
was likely to occur. Separation of these phases was thought to
have been the cause of the unsuccessful direct conversion of 2
into 4 in the stop-flow reactor preparation.10 An alternative
solution to the problem was found in this case by adding one
volume of N,N-dimethylacetamide (DMA) (relative to 2) to the
reaction mixture of crude 3, which then formed a single phase
with DCB and formic acid (Caution: heat of mixing noted on
1 L scale).
Results and Discussion
Ortho Claisen Rearrangement. We started with the simple,
high-temperature, unimolecular ortho Claisen rearrangement11
of 2 to 3 (Scheme 1) using the conditions previously em-
ployed.10 Thus, the reaction mixture was diluted 2:1 wt/volume
with 1,2-dichlorobenzene (DCB) even though this was probably
not needed for mobility in the FlowSYNTH. The reaction
settings required 195 °C for 12 min for 95% conversion, which
equated to a flow rate of 16.7 mL/min. After tuning, the actual
pump settings of 40% with a 20% stroke rate were found to
give a flow rate of 15.0 mL/min, which equated to a residence
time of 13.3 min, slightly longer than required, but well within
the scope of the previously tested conditions.10 The exit
temperature on the chiller unit was set to 80 °C to mimic the
previous study.
A residence time of 10 min required a flow rate of 20 mL/
min. After priming of the system with a formic acid/DMA
solution and heating to 100 °C, one column volume was allowed
to pass before samples were taken at 10-min intervals over a
40-min period. Once again, stable and complete conversion of
starting material 3 to product 4 was observed. An impurity of
up to 16% was detected, but it should be noted that the reaction
mixture had not been purified at the previous stage, and this
The entire system was flushed with fresh DCB for 30 min
whilst the chamber was heated to 195 °C. The reaction mixture
(9) Lloyd-Jones, G. C.; Moseley, J. D.; J. S. Renny, J. S. Synthesis 2008,
661–689.
(10) Moseley, J. D.; Woodman, E. K. Org. Process Res. DeV. 2008, 12,
967–981.
(11) Castro, A. M. M. Chem. ReV. 2004, 104, 2939–3002.
(12) Marafie, J. A.; Moseley, J. D. Unpublished results.
Vol. 14, No. 4, 2010 / Organic Process Research & Development
•
927