Chemical development of NBI-75043. Use of a flow reactor to circumvent a batch-limited metal#halogen exchange reaction

Article abstract of DOI:10.1021/op800071m

The discovery route and subsequent scale-up routes for NBI-75043 are presented. When traditional batch chemistry was found to limit the scale of a key reaction, a flow reactor was designed and optimized to provide an alternate method of production.

Full text of DOI:10.1021/op800071m

Organic Process Research & Development 2008, 12, 929–939
Chemical Development of NBI-75043. Use of a Flow Reactor to Circumvent a
Batch-Limited Metal-Halogen Exchange Reaction
Timothy D. Gross,*,† Shine Chou,† Daniel Bonneville,† Raymond S. Gross,† Peng Wang,† Onorato Campopiano,†
Michael A. Ouellette,† Scott E. Zook,† Jayachandra P. Reddy,† Wilna J. Moree,† Florence Jovic,† and Shubham Chopade‡
Chemical DeVelopment and Medicinal Chemistry, Neurocrine Biosciences, Inc., 12790 El Camino Real, San Diego, California,
92130, U.S.A., and Irix Pharmaceutical Inc., 101 Technology Place, Florence South Carolina, 29501, U.S.A.
Abstract:
The discovery route and subsequent scale-up routes for NBI-75043
are presented. When traditional batch chemistry was found to limit
the scale of a key reaction, a flow reactor was designed and
optimized to provide an alternate method of production.
Figure 1. NBI-75043.
Introduction
Four types of histamine receptors have been characterized
Scheme 1. Medicinal Chemistry Routea for NBI-75043
in humans (H1-H4).1 The H1 receptor is the primary receptor
involved in allergic rhinitis symptoms and motion sickness. H1
antagonists such as diphenhydramine hydrochloride block
histamine binding in peripheral tissues, thereby reducing
symptoms of allergic rhinitis.2 One side effect of diphenhy-
dramine results from its penetration of the blood-brain barrier
resulting in CNS depression. Accordingly, selective H1 an-
tagonists with good blood-brain barrier penetration are thought
of as likely candidates for next-generation sleep therapies.3 NBI-
75043 (Figure 1) is a highly selective and potent H1 antagonist
that was under evaluation for safety and efficacy in the treatment
of insomnia.
Background: Early lots of NBI-75043 were produced by
the Medicinal Chemistry department following the route shown
in Scheme 1. Deprotonation of benzo[b]thiophene 2 using
n-BuLi followed by alkylation with 2-N,N-dimethylaminoethyl-
1-bromide provided 3; subsequent bromination of this inter-
mediate afforded compound 4. After metal-halogen exchange,
the resulting 2-lithio derivative was treated with acetic anhydride
to produce the corresponding 2-acylbenzothiophene 5. Reaction
of 5 with in situ generated 2-lithiopyridine resulted in tertiary
alcohol 6 which in turn was dehydrated to alkene 7 and
subsequently hydrogenated to give 8 as a racemic mixture. The
a Reagents and conditions: (a) 1. n-BuLi, -78 °C, 2. 2-N,N-dimethyaminoethy-
1-bromide hydrobromide; (b) bromine, HOAc; (c) 1. n-BuLi, -78 °C, TMEDA,
2. Ac2O; (d) 2-pyridyllithium; (e) TFA, c. H2SO4; (f) H2, Pd/C, MeOH; (g)
D-tartaric acid; aq NaOH, EtOAC; L-tartaric acid.
desired product, 1, was obtained through resolution of this
racemate. Crystallization of the undesired isomer with D-tartaric
acid enriched the remaining solution with the desired isomer.
After neutralization and extraction, crystallization of the L-
tartaric acid salt produced the desired compound.
There were two areas of immediate concern for scale-up of
this route. The acetylbenzothiophene 5 was difficult to purify,
and the reaction of 5 with 2-lithiopyridine was run under high
dilution (∼0.04 M) conditions.4 The completion of the synthesis
† Neurocrine Biosciences, Inc.
‡ Irix Pharmaceutical Inc.
* Author to whom correspondence may be sent. E-mail: t.gross@earthlink.net.
(1) (a) Leurs, R.; Smit, M. J.; Timmerman, H. Pharmacol. Ther. 1995, 66,
413–463. (b) Lovenberg, T. W.; Roland, B. L.; Wilson, S. J.; Jiang,
X.; Pyati, J.; Huvar, A.; Jackson, M. R.; Erland, M. G. Mol. Pharmacol.
1999, 55, 1101–1107. (c) Schneider, E.; Rolli-Derkinderen, M.; Arok,
M.; Dy, M. Trends Immunol. 2002, 23, 255–263.
(2) Huang, Z. L.; Qu, W. M.; Li, W. D.; Mochizuki, T.; Eguchi, N.;
Wantanabe, T.; Urade, Y.; Hayaishi, O. Proc. Natl. Acad. Sci. U.S.A.
2001, 98, 9965–9970.
(3) (a) Kaneko, Y.; Shimada, K.; Saitou, K.; Sugimoto, Y.; Kamei, C.
Methods Find. Exp. Clin. Pharmacol. 2000, 22, 163–168. (b) Nicholson,
A. N.; Pascoe, P. A.; Turner, C.; Ganellin, C. R.; Greengrass, P. M.;
Casy, A. F.; Mercer, A. D. Br. J. Pharmacol. 1991, 104, 270–276. (c)
Timmerman, H. Therapeutic Index of Antihistamines; Church, M. K.,
Rihoux, J.-P., Eds., Hogrefe and Huber: New York, Toronto, Berne,
1992; pp 19-31.
(4) (a) Gardiner, J. M.; Crewe, P. D.; Smith, G. E.; Veal, K. T.; Pritchard,
R. G.; Warren, J. E. Org. Lett. 2003, 5 (4), 467–470. (b) Hari, Y.;
Obika, S.; Sakaki, M.; Morio, K.; Yamagata, Y.; Imanishi, T. Tetra-
hedron 2002, 58, 3051–3063. (c) Lima, P. G.; Sequeria, L. C.; Costa,
P. R. R. Tetrahedron Lett. 2001, 42, 3525–3527. (d) Parham, W. E.;
Piccirilli, R. M. J. Org. Chem. 1977, 42, 257–260.
10.1021/op800071m CCC: $40.75
Published on Web 06/05/2008
2008 American Chemical Society
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
929

Scheme 2. Avoiding the use of 2-pyridyllithium
Scheme 3. Initial scale-up routea
a Reagents and conditions: (a) MTBE, aq NaOH; (b) s-BuLi, TMEDA; 2-cyanopyridine, PhMe, -78 °C; 6 N HCl; (c) MeLi, PhMe, -78 °C; (d) TMS-Cl, NaI,
MeCN; (e) D-tartaric acid; aq NaOH, EtOAc; L-tartaric acid.
was considered straightforward with alcohol 6 being the last
dilution conditions and the resulting high workup volumes
required were avoided. Second, rather than repeating the steps
of dehydration and hydrogenation as in the earlier route, direct
removal of the hydroxyl group was performed by treatment with
in situ generated TMS-I.7
crystalline intermediate prior to isolation of the final salt.5
Results and Discussion:
Initial Scale-up Route. The decision was made to outsource
the synthesis of the hydrobromide salt of bromobenzothiophene
4, and this material was used as the starting point for future
syntheses. Taking advantage of the crystalline nature of alcohol
6, we decided to use this intermediate as a convergence point
for future routes.
Two major changes were implemented to the original route.
First, the order of addition of the pyridyl and methyl function-
alities to the benzothiophene was reversed. Where the earlier
route used the addition of a pyridyl anion to a methyl ketone,
we chose to add a methyl anion to a pyridyl ketone as shown
in Scheme 2 Cyanopyridine was added to the 2-lithiated
benzothiophene intermediate which, after workup with HCl,
hydrolyzed the intermediate imine resulting in ketone 9.
Addition of methyllithium to 9 produced the desired intermedi-
ate alcohol 6. By avoiding the use of 2-pyridyllithium6, high
Even though these improvements allowed us to scale up and
provide necessary quantities of NBI-75043 for preclinical
studies (Scheme 3), the approach still suffered from low-yielding
steps and intermediates that were difficult to purify. Chroma-
tography of ketone 9 failed to afford any purity improvement
due to instability of this intermediate. It was found that exposure
of ketone 9 to UV or visible light resulted in rapid degradation.
It was determined that the ketone should not be held for
extended times and converted to the alcohol as rapidly as
practical. Attempts at larger-scale halogen-metal exchange
reactions resulted in lower yields of ketone 9 with yields in the
22-28% range on 2.1-2.6 mol scales. It was clear additional
route improvements would be required going forward.
(6) Wang, X.; Rabbat, P.; O’Shea, P.; Tillyer, R.; Grabowski, E. J. J.;
Reider, P. J. Tetrahedron Lett. 2000, 41, 4335–4338.
(5) Gross, T. D.; Schaab, K.; Ouellette, M.; Zook, S.; Reddy, J. P.; Shurtleff,
A.; Sacaan, A. I.; Alebic-Kolbah, T.; Bozigian, H. Org. Process Res.
DeV. 2007, 11, 365–377.
(7) (a) Sakai, T.; Miyata, K.; Utaka, M.; Takeda, A. Tetrahedron Lett. 1987,
28 (33), 3817–3818. (b) Toudic, F.; Ple´, N.; Turck, A.; Que´guiner, G.
Tetrahedron 2002, 58, 283–293.
930
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development

Experiments were performed in order to explore the limits
of time and temperature that could be tolerated on larger scale.
The results from these experiments are summarized in Table
1. Following our small-scale procedure where addition of
compound 4 was controlled at or below -70 °C, ∼70%
of compound 9 was obtained along with ∼10% of des-bromo
compound 3 (entries 1, 2). Increasing the solvent volume and
adding compound 4 at or below -60 °C resulted in a dramatic
decrease in the amount of product formed and a corresponding
increase in the amount of des-bromo side product 3 (entry 3).
Reducing solvent volumes decreased these effects; however,
they remained unacceptably high (entry 4). Repeating the
starting conditions on a slightly larger scale (entry 5) resulted
in product and side product in the expected range.
The addition of compound 4 was quite exothermic, and it
was calculated that large-scale addition (5-6 kg) would require
∼4 h. Repeating the experimental conditions and slowly adding
compound 4 over 4 h resulted in an unacceptably low yield of
product and a high level of the des-bromo side product (entry
6). We carried out an experiment where the solvent volume
was doubled with the extended addition time (entry 7). It was
expected that a larger precooled solvent volume would provide
a larger heat sink and that we would be able to add compound
4 at a faster rate. This experiment resulted in a product:des-
bromo ratio similar to that of our small-scale results, and these
conditions were deemed acceptable with which to proceed to
larger scale. To aid cooling, both solutions of compound 4 and
2-cyanopyridine were precooled prior to charging to the reactor.
The kilo-scale batches showed a modest lowering in the yield
along with a slight increase in the amount of the side product
(entries 8, 9). While these conditions were sufficient to provide
material on a 6 kg scale, it was clear from the data that alternate
chemistry would be required for any larger-scale campaigns.
Final Scale-up Route. Due to the cryogenic conditions
required for the lithiation chemistry, we also explored using
Grignard chemistry to accomplish the conversion at more
moderate temperatures. The Grignard reagent formed from
compound 4 was found to be unreactive with 2-acetylpyridine
by itself or in the presence of zinc chloride. Deuterium quench
of the reagent failed to demonstrate any deuterium incorporation
while compound 4 was consumed. Attempts at Friedel-Crafts
reaction with benzothiophene 3 showed no reaction in the
presence of various Lewis and protic acids (CeCl3, H2SO4,
AlCl3, AgOTf, and Et3B/C6).
Figure 2. Quaternized product.
Second Scale-Up Route. During the evaluation of the
halogen-metal exchange reaction, different types of organo-
lithium reagents were explored. It was found that n-BuLi
performed acceptably on a small scale. In larger-scale reactions,
the quaternary ammonium salt 10 (Figure 2) was detected as a
new impurity. Compound 10 is formed as a result of the reaction
of n-butyl bromide (formed as a reaction byproduct during the
metal-halogen exchange) with the dimethyl amine side chain
of compound 9. The explanation for its formation is that, as
the scale increased from ∼50 mmol to ∼200 mmol, the reaction
and workup times increased correspondingly. The reaction
mixture was difficult to analyze, giving inconsistent results, and
impurity levels could only be determined after the workup. As
expected, use of s-BuLi for the halogen-metal exchange
resulted in a lower amount of the corresponding impurity. Use
of t-BuLi (2 equiv) showed significantly better results with no
quaternization observed and therefore was used in the manu-
facturing campaigns.
The halogen-metal exchange reaction was performed by
addition of a toluene solution of compound 4 to a cooled
solution of n-BuLi and TMEDA in toluene. Other solvents such
as THF were investigated; the best results were obtained using
toluene. The addition rate was controlled carefully to maintain
an internal temperature less than -70 °C in order to obtain the
best yields and consistent results. Because of the exothermic
nature of this addition, it took approximately 3 h from the start
of addition of compound 4 to the start of addition of 2-cyan-
opyridine on a 2 mol scale.
Conversion of 9 to 6 was found to proceed easily with either
methyllithium or methylmagnesiumbromide. Work-up of the
magnesium salts resulting from the use of methylmagnesium-
bromide was problematic, and it was decided to use methyl-
lithium for this conversion.
The direct reductive dehydroxylation was found to be
unreliable for scale-up and was again replaced with a two-step
process. The alcohol was first dehydrated to olefin 7 by
treatment with methanesulfonic acid in dichloromethane. This
procedure was more convenient and productive when compared
with the original H2SO4/TFA procedure. Hydrogenation using
palladium catalysis proved inconsistent, presumably due to the
susceptibility to catalyst poisoning by either the sulfur or amine
groups in the molecule. After a screening effort, this reduction
was conveniently accomplished with hydrogen in the presence
of Adam’s catalyst. The final improvement for the second scale-
up route (Scheme 4) was the direct crystallization of the desired
isomer from the reaction mixture. At larger scale, this crystal-
lization proceeded without yield loss while omitting enrichment
by initial crystallization of the undesired isomer. This route was
utilized to produce the early requirements of a multikilo batch.
Attempts at adding 2-acetylpyridine to lithiated compound
4 failed to produce the desired product. Lithiation followed by
the addition of 2-acetylpyridine in the presence of catalytic
Ti(OiPr)4 formed the desired alcohol 6. Repeating these
conditions with ZnCl2 as the Lewis acid resulted in ∼50% of
des-bromo side product (3) and ∼50% of alcohol 6. 2-Methyl-
THF gave similar results to THF as expected. Use of toluene
as the solvent resulted in larger amounts of the des-bromo side
product in the presence of Ti(OiPr)4. Our best conditions (t-
BuLi, THF, catalytic Ti(Oi-Pr)4) provided a 67% yield of
alcohol 6.
With an alternate approach to alcohol 6 in hand, we sought
to improve the reduction of olefin 7. An obvious approach was
to attempt asymmetric hydrogenation with the expectation of
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
931

Scheme 4. Second scale-up routea
a Reagents and conditions: (a) MTBE, aq NaOH; (b) s-BuLi or t-BuLi, TMEDA; 2-cyanopyridine, PhMe, -78 °C; 6 N HCl; (c) MeLi, PhMe, -78 °C; (d) MsOH,
DCM; (e) PtO2, EtOH, 30 °C, H2 120 psi; (f) L-tartaric acid.
Table 1. Cyanopyridine addition
Flow Reactor Development: Small-scale experiments in-
dicated that both the halogen-metal exchange and anion
addition to acetylpyridine reactions were extremely rapid and
quite reproducible. Obvious limitations on larger scale would
be due to the exothermic nature of these two steps and the time
required for reagent addition to maintain low internal temper-
ature. Having chemistry that worked effectively only on small
scale, we chose to explore this methodology using a solution
afforded by engineering.8 Focusing on the final scale-up route,
we investigated the application of flow reactor technology to
our scale-limited lithiation sequence.
4
(wt)
9
3
yield
conditions
(%) (%) (%)
1
2
3
4
5
6
7
8
9
16 g
16 g
16 g
25 g
27 g
25 g
25 g
15 volumes THF -70 °C
repeat
20 volumes THF -60 °C
15 volumes THF -60 °C
15 volumes THF -70 °C
4 h addition
71
9
6
7
47
85
93
29 29
72
9.50 86
13 54
4 h addition double THF volume 72
8
56 15
57 21
5.7 kg batch 1
6.4 kg batch 2
90
A schematic of our first-generation flow reactor is shown in
Figure 3. The reactor was constructed using 1/4 in. stainless
steel tubing, 3 peristaltic pumps and 3/8 in.Teflon tubing. The
entire steel portion of the reactor was submerged in a dry ice/
acetone bath with Teflon lines running to and from the reactor.
A solution of t-BuLi and TMEDA in THF was prepared and
cooled in a -78 °C bath. This solution was combined via a
T-connection with a solution of compound 4 in THF with an
in-line thermocouple and through a static line mixer. As the
solution exited the mixer, it passed an in-line thermocouple and
into a coil of tubing (residence time control) to allow the mixture
to age briefly. A third stream of 2-acetylpyridine and Ti(OiPr)4
in THF was introduced Via a T-connection along with a
thermocouple. This mixture passed through a second static line
mixer and hold coil. As the solution exited the hold coil,
temperature was recorded prior to quenching in a stirred flask
of methanol. The flow rates of the pumps were calibrated in
order to determine approximate starting values (Figure 4). The
actual settings were to be determined experimentally both by
monitoring the temperature at various points in the reactor and
by periodic sampling of the reaction. Flow was adjusted until
the temperature remained in-range and stable which would
producing an enantio-enriched mixture of 8. Reaction screening
(a total of 45 experiments) using different Ru, Rh and Ir catalysts
was performed as a first pass. One combination resulted in
100% conversion with 78% ee of the undesired enantomer. The
reaction was further optimized using this ligand while efforts
went towards preparing the ligand with the opposite (desired)
chirality. Using purified (column chromatographed) olefin 7,
substrate to catalyst (S/C) ratios of 1000:1 could be achieved
while the ratio dropped to 100:1 when using starting material
isolated/purified through an extractive workup (as would be
preferred for scale-up). Running the reduction on a 5-g scale
with an S/C ratio of 100 using the ligand with the correct
chirality, full conversion was observed with an ee of 60% of
the desired isomer. A single crystallization with L-tartaric acid
improved the ee to 95% in 56% isolated yield (from olefin 7).
Along with these route improvements, a change was made in
the counterion for the final salt formation from L-tartrate to
benzenesulfonate. Use of L-tartaric acid was still required for
the enantio-enrichment. The final salt formation was ac-
complished with a neutralization of the L-tartrate salt and
formation of the benzenesulfonate salt. With the latest route
improvements, we were prepared to manufacture again with
our final scale-up route (Scheme 5). These improvements to
the route bought a short amount of time to afford us the
opportunity to generate the required quantities of API for
additional studies more reliably and productively. Given the
instability of lithiated compound 4, it was clear that additional
improvements to the preparation of alcohol 6 were needed for
the future.
(8) Recent publications citing the benefits of flow reactor technology
include: (a) Ehrfeld, W.; Hessel, V.; Lowe, H. Microreactors - New
Technology for Modern Chemistry; Wiley-VCH: Weinheim, 2000. (b)
Delhaye, L.; Stevens, C.; Merschaert, A.; Delbeke, P.; Brioˆne, W.;
Tilstam, U.; Borghese, A.; Geldhof, G.; Diker, K.; Dubois, A.; Barberis,
M.; Casarubios, L. Org. Process Res. DeV. 2007, 11, 1104–1111. (c)
Bogaert-Alvarez, R. J.; Demena, P.; Kodersha, G.; Polomski, R. E.;
Soundararajan, N.; Wang, S. S. Y. Org. Process Res. DeV. 2001, 5,
636–645. (d) Brechtelsbauer, C.; Ricard, F. Org Process Res. DeV. 2001,
5, 646–651. (e) Choe, J.; Kim, Y.; Song, K. H. Org. Process Res. DeV.
2003, 7, 187–190.
932
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development

Scheme 5. Final scale-up routea
a Reagents and conditions: (a) MTBE, aq NaOH; (b) t-BuLi, TMEDA; 2-acetylpyridine, Ti(OiPr)4, THF, -78 °C; (c) MsOH, DCM; (d) [RuI2(p-cymene)]2, SL-
M004-2, MeOH, 30 °C, H2 260 psi; (e) L-tartaric acid; salt break; PhSO3H.
ature was still above our target of -65 °C. Closer examination
of the data revealed that the temperature at the end of the static
line mixer (temp 2, Table 2) remained constant, indicating that
the system had stabilized by the time the stream exited the
mixer. The initial variability in temperature was attributed to
both the heat of reaction and cooling of the room temperature
stream of compound 4 that was introduced. Thus, two 1/8 in.
cooling coils were added to the output of pumps 1 and 2 prior
to combining the streams (Figure 5). The smaller tubing
provided a large surface area to volume ratio which aided in
efficient cooling.
The results of the pump setting and thermal results from
the subsequent run are summarized in Table 3, and analyses of
in-process testing are summarized in Table 4. The modified
reactor was submerged in a -78 °C bath, and once the
temperature had stabilized (entry 1), the pumps were started
(entry 2). The reactor was run for 2 min, and then the reaction
mixture was sampled and the pumps were stopped (entries 3,
4). Analysis of the sample (Table 4, 206-A) showed a high
level of 2-acetylpyridine and its t-BuLi adduct, which indicated
over-addition from pump 3. The speed of pump 3 was reduced
and the process repeated (Table 3, entries 5-7), and the analysis
of the second sample (Table 4, 206-B) showed greatly reduced
levels of 2-acetylpyridine, and none of the t-BuLi adduct was
detected. The starting bromide 4 was detected, and the pump 2
flow rate was reduced; then the sequence was repeated. Analysis
Figure 3. First-generation flow reactor schematic.
of the third sample (Table 4, 206-C) showed alcohol 6 was
indicate steady-state conditions. The presence of bromide 4 in
present at the levels typical of small-scale batch runs and des-
the sample would indicate the rate of stream 1 addition was
bromo side product at or slightly below typical levels. In two
too fast, too slow a t-BuLi addition or inadequate resonance
time for the halogen-metal exchange. Excess 3,3-dimethyl-2-
runs with such a setup, we had equaled our best results observed
in small scale with batch reactors. We next sought to investigate
the limits of productivity possible for our reactor system.
We expected the halogen-metal exchange to be rapid, and
therefore the hold coil (hold coil 3) after the first static line
mixer (Figure 5) was removed, and the reactor system was run
using the best previous settings. Steady-state conditions were
achieved as evident by low and stable temperature readings at
all of the measured points. Analysis of in-process samples
showed consumption of starting bromide 4 and presence of
∼13% of product 6, ∼50% of des-bromo side product and
methyl-1-(2-pyridyl)-butan-2-ol 11 (the product of t-BuLi
addition to 2-acetylpyridine) would indicate too high a flow
rate of t-BuLi. The relative proportions of des-bromo side
product, the desired product, and 2-acetylpyridine would give
insight into the proper flow rate for the third pump.
The results of our initial run are shown in Table 2. Our
starting flow rates appeared too high as evidenced by the high
and variable temperatures observed at the initial mixing of the
streams. Slowing the flow rates both lowered the temperature
and the variability previously observed. However, the temper-
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
933

Figure 4. Pump calibration.
Table 2. First-generation flow reaction
temp 1
(streams 1, 2 mixed °C)
temp 2
(end of mixer °C)
temp 3
(Ac-pyr addition °C)
pump 1
setting(t-BuLi)
pump 2
setting( 4)
pump 3
setting(ac-pyr)
time
3:27
3:29
3:32
3:33
3:34
3:40
3:40
3:45
3:50
-15.6
-6.1
-76.0
-73.0
-73.0
-73.6
-73.8
-74.8
-76.7
-78.2
-78.4
-72.0
-72.6
-72.8
-68.2
-68.2
-67.2
-71.5
-76.6
-76.4
12.5
12.5
12.5
12.5
12.5
12.5
3.75
3.75
3.75
2
5
2
5
-7.5
2
5
-11.4
-12.9
-13.2
-48.1
-45.1
-57.1
2
5
2
5
2
5
2.5
2.5
2.5
2.5
2.5
2.5
∼26% of 11. Our interpretation of these results was that the
halogen-metal exchange had completed, but there was insuf-
ficient time for the elimination of t-BuBr; therefore, the
remaining t-BuLi reacted with 2-acetylpyridine, and the 2-
lithiobenzo[b]thiophene was quenched. This was clear evidence
that the hold coil was necessary for the elimination of t-BuBr,
and it was replaced. No further modifications were performed
on the reactor.
Previous experiments were run with pump 1 (supplying
t-BuLi) at or near its maximum speed. The speed of this pump
Figure 5. Second-generation flow reactor schematic.
934
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development

Table 3. Second-generation flow reactor, run 1
Table 7. Second-Generation Flow Reactor, Run 3
pump
temperature (°C)
pump
2
temperature (°C)
entry time
1
0
2
0
3
1
2
3
4
sample
entry time
1
3
1
2
3
4
sample
1
2
3
4
5
6
7
8
9
3:22
0
5
-78.7 -79.0 -77.0 -79.8
1
2
3
4
5
6
7
8
9
2:30 0 0
2:31 7 3
2:35
2:36 9 4
2:38
0
3
-78.2 -78.8 -77.8 -77.4
3:22 12.5 5
3:24
-71.9 -75.8 -62.8 -79.6 206-A
-53.7 -74.9 -61.7 -77.7 004-A
-41.8 -72.8 -70.9 -77.3 004-B
3:24
0
0
0
4
3:56 12.5 5 2.5 -78.3 -78.4 -77.8 -77.6
3:59
4:00
4:31
-70.0 -76.1 -76.4 -77.2 206-B
2:38 3 1.5 1.5
0
0
0
2:43 0 0
3:11
3:11 6 2
0
-78.2 -78.2 -77.4 -77.4
-78.2 -78.2 -77.7 -77.5
4:32 12.5 2.5 2.5
2
10 4:35
11 5:29
-69.3 -74.3 -77.2 -77.2 206-C
-78.2 -78.1 -77.5 -77.5
10 3:44
-78.2 -78.1 -77.3 -76.7 004-C
0
0
0
11 3:44 0 0
12 4:11 12 6
13 4:12 12 6
14 4:13
0
6
7
12 5:29 12.5 2.5 2.5
13 5:31
-78.2 -78.2 -77.7 -77.6
004-D
-63.7 -72.4 -48.8 -76.2
-75.1 -74.3 -77.3 -77.3
-71.9 -74.5 -77.3 -77.3
-73.3 -77.7 -77.3 -77.3
-73.6 -74.9 -77.2 -77.2
-73.4 -76.5 -77.3 -77.3
-73.9 -74.9 -77.3 -77.3
14 5:33
15 5:40
15 4:14 0 0
0
16 5:45
17 5:50
Table 8. Sample analysis from second-generation flow
18 5:52
0
0
0
reactor, run 3
HPLC area %
Table 4. Sample analysis from second-generation flow
reactor, run 1
sample
cpd 12
cpd 11
cpd 6
cpd 3
cpd 4
004-A
004-B
004-C
004-D
59.1
4.6
0
0
2.3
53.2
14.5
70.5
10.2
14.9
23.5
17.8
<1
0.0
0.0
1.6
HPLC area %
13.5
34.1
0
sample
cpd 12
cpd 11
cpd 6
cpd 3
cpd 4
206-A
206-B
206-C
49.3
5.9
0
11.3
0
18.6
33.7
66.2
11.5
13.9
16.5
0.0
28.6
0.0
7.3
8.8
12:
11:
6:
2-acetylpyridine
12/t-BuLi/adduct
product alcohol
des-brominated
12:
11:
6:
2-acetylpyridine
12/t-BuLi/adduct
product alcohol
des-brominated
3:
4:
bromobenzthiophene
3:
in unacceptable mixtures (Table 5, entries 1-5; Table 6,
samples 003-A, 003-B). Reducing the speed of pump 1 while
increasing the speed of pumps 2 and 3 gave a similar reaction
profile (Table 6, sample 003-C) at a higher temperature than
previous reactions (Table 5, entry 7). A much slower flow rate
was attempted (Table 5, entry 8), but at reduced speed, the
solution had a greater resonance time in the supply lines and
cavitations in pump 1 caused us to discontinue this run. Bubbles
that formed in the lines and pump head led to inconsistent
reagent supply to the reactor.
Our next set of experiments is summarized in Tables 7 and
8. Running the reactor with the pumps at approximately half
the flow rate of our best conditions, we observed an initial
temperature above -60 °C for the halogen-metal exchange.
Samples were taken, and we were surprised to find a high level
of product in one of them (Table 8, sample 004-B). We
suspected that the reaction might be tolerant of higher temper-
atures for a brief period of time, but this was our first such
observation. The pumps were run at slow speed (Table 7, entry
6) to try to minimize the measured temperature stream. Once
again, cavitations were observed in pump 1, and therefore the
reaction was suspended (Table 7, entry 7) while the tubing
leading from the cooled t-BuLi solution to the pump and from
the pump to the flow reactor was covered with pipe insulation.
Insulating the supply lines stabilized the solution, and no bubbles
were observed in the lines. After sampling (Table 8, sample
004-C) and final pump adjustments, we allowed the reactor to
reach steady state and took a final sample. Analysis of this
sample (Table 8, sample 004-D) exceeded all of our expecta-
4:
bromobenzthiophene
Table 5. Second-generation flow reactor, run 2
pump
2
temperature (°C)
entry time
1
3
1
2
3
4
sample
1
2
3
4
5
6
7
8
4:22 0 0
4:22 6 3
4:31
4:32 12 6
4:33
4:34 9 4.5 4.5
4:35
4:35 3 1.5 1.5
0
3
-78.2 -78.2 -78.1 -77.7
-74.6 -79.5 -74.9 -77.9 003-A
-48.6 -72.8 -60.0 -77.3 003-B
-64.8 -74.6 -68.3 -77.4 003-C
6
Table 6. Sample analysis from second-generation flow
reactor, run 2
HPLC area %
sample
cpd 12
cpd 11
cpd 6
cpd 3
cpd 4
003-A
003-B
003-C
10.3
25.1
5.6
0
0
0
26.7
44.9
63.0
13.8
16.7
20.8
35.5
7.6
5.6
12:
11:
6:
3:
4:
2-acetylpyridine
12/t-BuLi/adduct
product alcohol
des-brominated
bromobenzthiophene
was varied in the next experiment, and the results are shown in
Tables 5 and 6. Reducing the rate of pump 1 by 50% or
increasing the rates of pumps 2 and 3 by 100% both resulted
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
935

Table 9. Second-generation flow reactor, run 4
pump temperature (°C)
Table 11. Sample analysis from second-generation flow
reactor, run 4 corrected for 2-acetylpyridine
HPLC area %
entry time
1
2
3
1
2
3
4
sample
sample
cpd 11
cpd 6
cpd 3
cpd 4
1
2
3
4
5
6
7
8
9
10:54
10:55 12
11:00
11:00
11:36
11:37 12
11:41
0
0
6
0
7
-77.6 -78.3 -78.2 -76.9
053-D
053-E
n/a
n/a
92.6
92.7
7.4
7.3
0
0
-69.5 -76.0 -78.7 -76.0 053-A
-77.2 -77.3 -77.8 -77.4
0
0
6
0
6
0
6
0
7
-69.3 -76.8 -75.8 -77.4 053-B
-77.2 -77.3 -77.8 -77.4
11:41
12:25
0
10 12:25 12
11 12:30
12 12:35
13 12:40
14 12:45
-66.5 -74.9 -75.8 -77.3 053-C
-66.1 -74.3 -75.3 -77.4 053-D
-69.4 -73.8 -75.0 -77.4 053-E
-68.8 -73.5 -74.6 -77.3
15 12:45
0
0
0
Table 10. Sample analysis from second-generation flow
reactor, run 4
HPLC area %
sample
cpd 12
cpd 11
cpd 6
cpd 3
cpd 4
053-A
053-B
053-C
053-D
053-E
25.8
20.9
20.3
18
0
65.1
51.3
65.4
76.3
74.9
9.1
9.4
9.2
6.1
5.9
0
18.4
5.1
0
0
0.0
0.0
0
Figure 6. Actual flow reactor.
19.1
0
was converted to product, and there were no variations (within
experimental error) in the peak area of the final two samples,
again demonstrating steady-state conditions were achieved.
12:
11:
6:
2-acetylpyridine
12/t-BuLi/adduct
product alcohol
des-brominated
3:
Conclusions
4:
bromobenzthiophene
In the route development of NBI-75043, we identified and
developed a novel Ti(OiPr)4-catalyzed addition of a metalated
species to 2-acetylpyridine to afford a key intermediate and
simultaneously avoided unstable intermediates. As the exother-
mic nature of the key halogen-metal exchange reaction limited
batch size, we demonstrated the use of continuous flow reactor
technology to successfully execute this chemistry. In two
experimental runs with a custom-built reactor (Figure 6), we
were able to equal the best results from batch-wise runs. In
two additional experimental runs, we achieved higher conver-
sion than had been previously observed and confirmed these
results in a subsequent run. In the process, we have demon-
strated the ease and practicality of this valuable tool in process
chemistry.
tions. We observed the highest level of product of any reaction
we had run. The reaction was repeated in a final set of
experiments (Tables 9, 10) to verify the observations. Thus,
after cooling, the reactor was run repeating our best conditions
and was sampled (Table 9, entries 1-3). Analysis of this sample
(Table 10, sample 053-A) found the same level of product, less
debromination, and a slightly elevated level of 2-acetylpyridine
when compared to the previous run. The speed of pump 3 (2-
acetylpridine supply) was slightly reduced, the reactor run and
sampled (Table 9 entries 6-7). This sample was found to
contain a reduction in the amount of product formed, a slight
reduction in the amount of 2-acetylpyridine, the same level of
des-bromination, and a large amount of starting bromide (Table
10, sample 053-B). It was clear that a small adjustment to one
pump had a large effect on the entire system. The initial pump
settings were used and the reactor run continuously with periodic
sampling (Table 10, samples 053-C to 053-E) until the feed
reagents were exhausted (Table 9, entries 10-14). In sample
053-C, a small amount of starting bromide was observed;
however, as the reactor reached steady state, the starting bromide
was no longer detected. The last two samples (Table 10; samples
053-D, 053-E) showed a high level of product and a low level
of des-bromination. 2-Acetylpyridine was used in excess in the
batch process to ensure consumption of the 2-lithio species; it
was therefore known from experience that it would purge during
isolation of the product. The sampling data was reprocessed
without integrating the 2-acetylpyridine peak, and the results
are shown in Table 11. More than 92% of the starting material
Experimental Section
Compound 9 [2-(2-dimethylamino-ethyl)-benzo[b]thiophen-
3-yl]-pyridin-2-yl-methanone. To 100 g (277 mmol) of [2-(3-
bromo-benzo[b]-thiophen-2-yl)ethyl]dimethylamine HBr (com-
pound 4 HBr) in 500 mL of water was added 300 mL of 1N
sodium hydroxide solution. The mixture was extracted twice
with 1 L of MTBE. The combined organic phases were dried
and concentrated in Vacuo affording 72 g of freebase. The free
base was diluted with 500 mL of anhydrous toluene and charged
to a 3 L three-neck flask equipped with a mechanical stirrer,
thermocouple, addition funnel, and nitrogen inlet. To the
solution was charged 38.5 mL (257 mmol, 1 equiv) of TMEDA,
and the mixture was chilled with a dry ice-acetone bath. To
the mixture was carefully charged 193 mL (270 mmol, 1.05
936
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development

equiv) of s-BuLi in cyclohexane dropwise, keeping the internal
temperature <-60 °C. After the addition, which took ap-
proximately 30 min, the mixture was stirred at -78 °C for 15
min. To the solution was charged 29.4 g (283 mmol, 1.1 equiv)
of 2-cyanopyridine in 50 mL of anhydrous toluene. During the
addition the internal temperature was kept <60 °C and took
approximately 15 min. The mixture was stirred at -78 °C for
30 min, and the reaction was complete as evidenced by HPLC.
The reaction was quenched by the careful dropwise addition
of a 100 mL 6 N HCl/200 mL methanol solution. During the
addition of the first 150 mL, the internal temperature was kept
<-50 °C at which time the solution turned orange. The final
150 mL was added keeping the internal temperature <-40 °C.
The mixture was allowed to warm to 0 °C. The mixture was
transferred to a separatory funnel and the organic phase
separated. The organic phase was extracted with 400 mL of 1
N HCl. The combined aqueous phases were stirred at ambient
temperature until no imine was observed by HPLC. Additional
6 N HCl may be required to accelerate the hydrolysis. Following
complete hydrolysis of the imine, the mixture was brought to
pH 10 with the careful addition of 50% sodium hydroxide. The
mixture was extracted 3 times with 600 mL of ethyl acetate.
The combined organic phases were dried over magnesium
sulfate and concentrated in Vacuo. This afforded 78 g of a dark
oil, which was approximately 80% pure. The major impurity
was the debrominated starting material. The product was used
without further purification for the next step.
Compound 6 1-[2-(2-dimethylamino-ethyl)-benzo[b]-
thiophen-3-yl]-1-pyridin-2-yl-ethanol. MeLi Addition to Com-
pound 9. In a 2 L three-neck Morton flask equipped with a
mechanical stirrer, thermocouple, and addition funnel with a
nitrogen inlet was charged 75 g (242 mmol) of ketone 9 in 500
mL of anhydrous toluene. The mixture was chilled using a -78
°C bath and stirred. To the solution was charged dropwise 160
mL (254 mmol, 1.05 equiv) of methyllithium in cyclohexane.
The addition was kept at a rate to keep the reaction temperature
<-60 °C and took approximately 30 min. The mixture was
stirred for an additional 15 min and judged complete by HPLC.
The reaction was quenched with a solution of 200 mL 6 N
HCl/400 mL methanol, keeping the reaction temperature <-45
°C during the addition. The mixture was allowed to warm to
ambient temperature and the organic phase separated. The
organic phase was extracted with 500 mL of 1 N HCl. The
combined aqueous phases were neutralized with 50% sodium
hydroxide solution and extracted three times with 600 mL of
ethyl acetate. The combined organic phases were dried over
magnesium sulfate and concentrated. The crude product was
diluted with 600 mL of warm acetonitrile and seeded with
authentic product. The mixture was stirred at ambient temper-
ature for 3 h and then chilled in an ice bath. The slurry was
filtered using a Bu¨chner funnel, and the filter cake was washed
with cold acetonitrile. After drying in vacuo at 45 °C this
afforded 46 g of alcohol 10 in a 59% yield as a light-tan powder.
2-Acetylpyridine Addition to Compound 4. TMDEA (3.8 kg,
33 mol) was dissolved in THF (8 kg) and dried over activated
molecular sieves (1 kg). The solution was filtered and transferred
to a 400 L Hastelloy reactor followed by a THF (2 kg) rinse.
Additional THF (132 kg) was charged and agitation started.
The jacket was set to -85 °C. While the reactor was cooling,
a 50 L round-bottom flask was charged with compound 4 (6
kg, 21 mol) and THF (22 kg). The flask was placed under a
nitrogen atmosphere, agitated, and cooled to <-65 °C. In a
second flask, THF (10 kg) and 2-acetylpyridine (2.8 kg, 23 mol)
were mixed. To the 400 L reactor was charged t-BuLi (11.9
kg in heptane, 25 wt %, 46 mol), maintaining an internal
temperature <-70 °C. The solution of compound 4 was
charged to the reactor while maintaining an internal temperature
<-70 °C. The reaction was monitored by HPLC for the
disappearance of compound 4. Once complete, Ti(Oi-Pr)4 (0.4
kg, 1.4 mol) was charged to the 2-acetylpyridine solution; the
solution was agitated and charged to the 400 L reactor while
maintaining an internal temperature <-70 °C. The reaction was
sampled, and the amount of 2-acetylpyridine was found to be
>1%. The reaction was resampled and judged complete as the
difference in area % for compound 3 in the two samples was
less than 2%. The reaction was quenched by the addition of
methanol (2.4 kg) while maintaining an internal temperature
<-70 °C. The reactor was warmed to 0 °C, and then water
(48 kg) and isopropyl acetate (44 kg) were charged while
maintaining an internal temperature <20 °C. Agitation was
stopped, and the layers were split. The aqueous layer was
extracted again with isopropyl acetate (44 kg). The combined
organic layers were washed with water (3 × 48 kg) until the
pH of the wash was found to be <9. The organic layer was
concentrated by distillation until 192 kg of distillate was
collected. Acetonitrile (48 kg) was charged and the reactor
concentrated to ∼20 L while maintaining an internal temper-
ature <45 °C. This process was repeated, and then additional
acetonitrile (48 kg) was charged, and the reactor was agitated
for 2 h at 20 °C. The contents were filtered through a Bu¨chner
funnel with Whatmann #5 filter paper and blown dry with
nitrogen for 20 min. The filter cake was transferred to a tray
dryer and dried at 45 °C until constant weight was obtained to
produce 3.35 kg, 49%; 99.6% HPLC purity.
Compound
8
Dimethyl-{2-[3-(1-pyridin-2-yl-ethyl)-
benzo[b]thiophen-2-yl]-ethyl}-amine. Direct Dehydroxylation
Route: In a 500 mL round-bottom flask, alcohol 6 (46 g, 141
mmol, 1.0 equiv) was slurried in 200 mL of anhydrous
acetonitrile. To the mixture was charged sodium iodide (42.3
g, 282 mmol, 2 equiv) followed by trimethylsilyl chloride (36
mL, 282 mmol, 2 equiv). The mixture was stirred at ambient
temperature for 1 h and the precipitate collected by filtration
using a Bu¨chner funnel. The filter cake was split into two 37 g
lots and diluted with 100 mL of acetonitrile in a 350 mL
pressure flask. To the slurry was charged sodium iodide (27.9
g, 186 mmol) and trimethylsilyl chloride (23.6 mL, 186 mmol).
To the mixture was added 1 drop of water, and the pressure
vessel was sealed and heated in an oil bath at 95 °C for 3 h.
The mixture was cooled and the reaction checked for complete-
ness by HPLC. If starting material remained, the reaction was
heated until complete. The completed reaction mixture was
poured into a 500 mL Erlenmeyer flask and treated with an
aqueous solution of sodium thiosulfate until the solution was
light yellow. The solution was neutralized with 50% sodium
hydroxide solution and extracted with 500 mL of ethyl acetate.
The organic phase was dried over anhydrous MgSO4 and
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
937

concentrated in Vacuo. The crude product was chromatographed
using a Biotage 75 eluting with 95:5 dichloromethane/methanol.
When pure product was observed, the eluent was changed to
9:1 dichloromethane/methanol. The pure fractions were com-
bined and concentrated, affording 17.9 g of compound 8 in a
41% yield.
filtered and washed with 200 mL of ethanol. The filter cake
was diluted with 350 mL of ethanol and heated until the solution
was homogeneous. The mixture was seeded and stirred at
ambient temperature for 8 h. The precipitate was collected by
filtration using a Bu¨chner funnel, and the filter cake was washed
with 200 mL of ethanol. The filter cake was dried in Vacuo at
45 °C for 36 h (constant weight), affording 50.0 g of NBI-
75043 in a 69% enantiomeric yield. Chiral purity determined
by chiral HPLC was 95.6% ee with an overall HPLC purity of
98.7%.
One-Step Resolution. Compound 8 (2.25 kg, 7.2 mol) was
dissolved in absolute ethanol (7 kg) in a small vessel under
nitrogen. The solution was charged to a 400 L GL reactor under
nitrogen followed by a rinse of absolute ethanol (3 kg). Agitation
was started and the reactor heated to 75 °C (internal tempera-
ture). A solution of L-tartaric acid (1.1 kg, 7.3 mol, 1.01 equiv)
in absolute ethanol (10 kg) was charged to the reactor. The
solution was stirred until a clear solution was obtained. With
gentle agitation, the reactor was cooled to 55 °C at a rate of 15
°C/h. A suspension of NBI-75043 L-tartrate seed crystals (2-5
g) in absolute ethanol (0.05 kg) charged. The reactor was cooled
to 45 °C over one hour; additional seeds were charged if the
first charge dissolved. The reactor was cooled to 20 °C at a
rate of 15 °C/h and then agitated at 20 °C for 4 h. The resulting
crystals were filtered through a Nutsche filter. The cake was
washed twice with absolute ethanol (5 kg). The cake was dried
in a vacuum oven at 45 °C to constant weight to give 1.5 kg,
3.3 mol, 45% yield, 99.7% HPLC purity.
Achiral Hydrogenation. A solution of compound 7 (3.4 kg,
11.0 mol) in absolute ethanol (20 kg) was charged to a glass
lined (GL) 200 L reactor under a nitrogen blanket followed by
a slurry of of Darco G60 charcoal (0.7 kg) in absolute ethanol
(10 kg). The reactor was agitated and heated to 50 °C for 2 h.
The reactor cooled to 25 °C, and the contents were filtered
through a bag filter followed by a 0.45 µm polish filter to
remove trace contaminants prior to hydrogenation. The cake
was washed with absolute ethanol (15-20 kg). The filtrate and
wash were charged to a 200 L pressure reactor under nitrogen.
A slurry of PtO2 (0.07 kg) in absolute ethanol (10.0 kg) was
charged to the reactor. With gentle agitation, the reactor was
thoroughly flushed with nitrogen followed by hydrogen. The
reactor was pressurized to 120 psig hydrogen and heated to
45-50 °C internal temperature. The agitation was increased to
the best gas mix mode, the temperature was maintained at
45-50 °C and the hydrogen pressure at 120 psig. Samples were
taken every 6 h and checked (HPLC). The end point was
reached when compound 7 was <0.5% by HPLC analysis. The
hydrogenation could take more than 70 h. The agitation was
slowed, the reactor cooled to 25 °C, the hydrogen pressure
released, and the reactor thoroughly flushed with nitrogen. The
contents were filtered through a bag filter, and the filtrate was
discharged to a 200 L reactor. The pressure reactor was flushed
with a small amount of absolute ethanol and the flush drained
to the same reactor. Charcoal (3.3 kg) and silica gel (7.0 kg)
were charged to the 200 L reactor, and the contents were
agitated at 30 °C for 2 h.
The contents of the 200 L reactor were filtered through a
bed of silica gel (7 kg) in a Nutsche filter, and the cake was
rinsed with absolute ethanol until no more product eluted (as
determined by HPLC or TLC). The filtrate was concentrated
by vacuum distillation at 40 °C to dryness to afford 2.9 kg,
9.34 mol, 85% yield of compound 8 with >98% HPLC purity.
Compound 1 (NBI 75043) Dimethyl-{2-[3-((R)1-pyridin-
2-yl-ethyl)-benzo[b]thiophen-2-yl]-ethyl}-amine. Two-Step
Resolution. To a 2-L round-bottom flask was charged 97 g (312
mmol) of compound 8 followed by 312 mL of ethanol. The
mixture was heated to near reflux, and a hot solution of 46.9 g
(312 mmol, 1 equiv) of D-tartaric acid in 312 mL of ethanol
was added. The mixture was seeded and stirred at ambient
temperature for 4 h. The precipitate was filtered, and the filter
cake was washed with ethanol. The mother liquor was
concentrated in Vacuo and the semisolid diluted with 500 mL
of water. The mixture was brought to pH 10 with 50% sodium
hydroxide and extracted twice with 500 mL of ethyl acetate.
The combined organic phases were concentrated in Vacuo,
affording 54 g of oil. The oil was diluted with 175 mL of ethanol
and heated to near reflux. To the mixture was charged a solution
of 26.1 g (174 mmol, 1 equiv) of L-tartaric acid in 175 mL of
hot ethanol. The hot solution was filtered through Celite, seeded,
and stirred at ambient temperature for 10 h. The solid was
Compound
7
Dimethyl-{2-[3-(1-pyridin-2-yl-vinyl)-
benzo[b]thiophen-2-yl]-ethyl}-amine. To a 200 L GL reactor
was charged 3.5 kg of compound 6 and DCM (10 kg). Agitation
was started. Methanesulfonic acid (10.3 kg, 10.3 equiv) was
added gradually at such a rate as to allow a gentle reflux. The
mixture was stirred at reflux for 5 h and monitored by HPLC.
The reaction was judged complete when the starting material
was <0.5 area % by HPLC analysis. The reaction was cooled
to room temperature, and then DCM (70 kg) was charged.
NaOH (64 kg, 2 M) was charged while maintaining the
temperature <30 °C until the pH of the aqueous layer measured
>9. Agitation was stopped, and the phases were split. The
aqueous layer was extracted with DCM (23 kg); the emulsion
was kept with the aqueous layer. The combined organic layers
were washed with 15% brine (58 kg). The organic phase was
dried over anhydrous Na2SO4, filtered, and concentrated under
vacuum at ∼45 °C to yield 3.4 kg (100% yield) of compound
7 as a dark amber oil. HPLC purity: 94.3%.
Flow Reactor. Pumps. Masterflex Console Drive, model
7520-60; head model 77390-00
Static Line Mixers. Koflo Corporation, Stratos tube mixer,
series 250; -0.250 in. O.D., 7 in. length, 21 elements; 11.5 in.
length, 34 elements.
Reactor Construction. See Figures 5 and 6 for schematic
and photo, respectively. The reactor was constructed from 316
stainless steel 1/8 in. and 1/4 in. tubing with Swagelock fittings
and tees to connect the parts; 1/8 in. stainless steel thermo-
couples were used to monitor temperature at various points in
the reactor. Reagent streams were plumbed with 3/8 in. PTFE
tubing from the supply flasks to the pumps and from the pumps
938
•
Vol. 12, No. 5, 2008 / Organic Process Research & Development

to the reactor. Two of the pumps were connected to two coils
of 1/8 in. tubing with a diameter of ∼3 in. and a volume of 7
mL (hold coils no. 1 and no. 2). These coils were connected
with a T-fitting to a short length of 1/8 in. tubing (∼3 in.) and
then to a second T-fitting containing a thermocouple (no. 1).
At this T-fitting, the O.D. was changed to 1/4 in. and the fitting
was connected to the 7 in. static line mixer and then to another
T-fitting with a thermocouple (no. 2). Next was a 1/4 in. O.D.
hold coil (no. 3) with a diameter of ∼5 in. and a volume of 39
mL. At the end of the coil no. 3 was a T-fitting with tubing
from the third pump and a T-fitting with a thermocouple (no.
3). Attached to the T-fitting was the second (11.5 in.) static
line mixer which was connected to a 1/4 in. O.D. hold coil
(no. 4) with a volume of 39 mL. A final T-fitting contained a
thermocouple (no. 4) and was connected to 3/8 in. PTFE tubing
running to the quench vessel.
mol) was dissolved in THF (200 mL) and connected to pump
no. 2. In a third flask, THF (200 mL), 2-acetylpyridine (21.7
mL, 0.19 mol), and Ti(OiPr)4 (3.9 mL, 0.013 mol) were mixed
and connected to pump no. 3.
The flow reactor was placed in a large container (the bottom
18 in. of a 30-gal plastic drum was used) and covered with
CO2(s) and acetone. While the reactor was cooling, the outflow
tubing was placed in a stirred container of methanol (500 mL).
Once the temperature reading stabilized, the pump speeds were
set and the pumps switched on. The reactions were sampled,
and the pumps were switched on and off and adjusted as shown
in Tables 3, 5, 7, and 9.
Acknowledgment
We gratefully acknowledge Kevin Schaab and Art Shurtleff
for superb analytical support and Anita Schnyder and Ulrkie
Nettekoven for chiral hydrogenation work.
Reagent Preparation. A dry flask was charged with TMEDA
(40 mL, 0.27 mol) and THF (250 mL). The flask was placed
under a nitrogen atmosphere and cooled in a -78 °C bath.
t-BuLi (200 mL, 0.34 mol. 1.7 M) was added slowly, maintain-
ing an internal temperature <-65 °C. This flask was connected
to pump no. 1. In a second dry flask, compound 4 (50 g, 0.18
Received for review March 26, 2008.
OP800071M
Vol. 12, No. 5, 2008 / Organic Process Research & Development
•
939