Asymmetric synthesis of the cholesteryl ester transfer protein inhibitor torcetrapib

Article abstract of DOI:10.1021/op060013i

Previously our group reported synthetic efforts used to synthesize kilogram quantities of the cholesteryl ester transfer protein (CETP) inhibitor torcetrapib, 1, via a mid-stage resolution. This account describes research conducted to develop an asymmetri

Full text of DOI:10.1021/op060013i

Organic Process Research & Development 2006, 10, 472−480
Asymmetric Synthesis of the Cholesteryl Ester Transfer Protein Inhibitor
Torcetrapib
David B. Damon, Robert W. Dugger, Stephen E. Hubbs, Jill M. Scott, and Robert W. Scott*,†
Pfizer Chemical Research and DeVelopment, Eastern Point Road, Groton, Connecticut 06340, U.S.A.
Scheme 1
Abstract:
Previously our group reported synthetic efforts used to syn-
thesize kilogram quantities of the cholesteryl ester transfer
protein (CETP) inhibitor torcetrapib, 1, via a mid-stage resolu-
tion. This account describes research conducted to develop an
asymmetric route to this clinical candidate suitable for long-
term manufacturing. The first asymmetric center is established
via coupling of (R)-3-aminopentanenitrile to a trifluoromethyl-
arene. After elaboration of the nitrile to a suitable precursor,
a key step in the synthesis is diastereoselective cyclization of
immonium ion 7 to provide the tetrahydroquinoline core. This
approach also permitted a streamlined sequence to complete
the synthesis of 1. Development of the process and synthetic
rationale are described.
Scheme 2
Introduction
Current therapies for the treatment of atherosclerosis focus
primarily on the reduction of low-density lipoprotein cho-
lesterol (LDL-C). Cholesteryl ester transfer protein (CETP)
is an agent responsible for the transfer of cholesterol from
high-density lipoprotein cholesterol (HDL-C) to LDL-C. A
compound that inhibits CETP may therefore be effective at
increasing the levels of HDL-C as a treatment of cardio-
vascular diseases. Torcetrapib (1) is a molecule currently
undergoing clinical trials as a CETP inhibitor, thereby
necessitating a synthetic route amenable to multikilogram-
scale synthesis.
tetrahydroquinoline core. The asymmetric route was based
on a key discovery during research to develop the cyclization
to produce 4 (Scheme 1). From the crude reaction mixture,
a byproduct (5) was isolated as a 1:1 mixture of diastereo-
mers. Resubjecting 5 to acidic cyclization conditions resulted
in clean formation of the desired racemic 4. The exclusive
formation of the desired cis orientation suggested that
assembly of a precursor with the C-2 center of defined
stereochemistry would provide the desired syn-tetrahydro-
quinoline upon cyclization. In development of a new
synthesis, two other factors were considered. First, the
continued use of N-vinylbenzylcarbamate 3 was undesirable,
due to varying quality of this material because of its low
stability. Second, we hoped to avoid the use of trifluoro-
methylaniline as a starting material due to safety issues with
long-term storage.2
The preceding paper outlined process research and scale-
up for the clinical candidate 1.1 The key step in the
production of asymmetric material was a diastereomeric salt
resolution. Although several multikilogram lots were pro-
duced by this resolution route, synthetic efficiency favored
development of an asymmetric route to synthesize the
We set out to synthesize an intermediate such as 7 and
verify that the absolute stereochemistry at the 2-carbon would
transfer to the C-4 center to form the tetrahydroquinoline
core 6 as a single enantiomer (Scheme 2). Retrosynthetically,
the acyl-immonium ion could be generated from the imide,3
* Author for correspondence. E-mail: bobscottorg@sbcglobal.net.
† Current address: Chemical Research and Development, Pfizer Global R&D,
10578 Science Center Drive, San Diego, CA 92121.
(1) Damon, D. B.; Dugger, R. W.; Magnus-Aryitey, G.; Ruggeri, R. B.;
Wester, R. T.; Tu, M. Abramov, Y. Org. Process Res. DeV. 2006, 10, 464-
471.
(2) Tickner, D.; Kasthurikrishnan Org. Process ReV. DeV. 2001, 5, 270-271.
(3) DeNinno, M. P.; Eller, C. Tetrahedron Lett. 1997, 38, 6545-6548.
472
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development
10.1021/op060013i CCC: $33.50 © 2006 American Chemical Society
Published on Web 03/16/2006

Scheme 3a
Scheme 4a
a Reagents and conditions: (a) i. BOC2O; ii. MsCl, TMEDA (86%); (b)
Bu4NCN (93%) or NaCN, Bu4NBr (74%); (c) MsOH, THF (81%).
which in turn could be formed via Pd-catalyzed condensation
of an enantiomerically pure amine and aryl halide.
a Reagents and conditions: (a) Na2CO3; (b) Pd(OAc)2, 2-(Cy)2P-2′-
(Me2N)biphenyl), PhB(OH)2, Cs2CO3; (c) H2SO4, H2O (77%); (d) ClCO2Bn,
LiOtBu (85%).
Results and Discussion
istry was not as clean or high yielding but did provide the
desired product in decent conversion at a low cost with
readily available reagents. Impurities formed in this step were
purged in the subsequent BOC-removal/salt formation step.
This procedure was transferred to pilot plant facilities to
produce 6 kg of 14. This key raw material represented a
useful cGMP starting material for the first campaign, and
further production research was not pursued.6 This allowed
us to focus our efforts on the remaining steps of the synthesis.
The next two-step sequence was coupling of chiral amine
15 to the trifluoroarene followed by hydrolysis of the nitrile
to produce amide 18 (Scheme 4). We chose the reaction
sequence in this order since the nitrile provided a nonbasic
and nonligating functionality, limiting potential problems in
the Pd-catalyzed amination. We examined many variables
for the Pd-mediated amination reaction of 15 with 16.7
Among the variables examined were different bases and base
combinations (NaOtBu, Cs2CO3, K2CO3, K3PO4); sol-
vents (toluene, CH3CN, THF, dioxane, DMF, DMAC,
DME); ligands [binap, (o-tol)3P, DPPE, DPPB, DPPF,
(biphenyl)P(Cy)2, (biphenyl)P(tBu)2, (dimethylaminobi-
phenyl)P(Cy)2]; catalyst activation protocols (in situ activa-
tion or preactivation using PhB(OH)2, NEt3, (iPr)2NH); and
Pd source [Pd2(dba)3 or Pd(OAc)2]. The extensive list of
permutations will not be reproduced here, but the optimal
conditions elucidated within our research time frame are
described below.
Although the chiral amine 14 was isolated as a salt,
cleaner reactions resulted from the use of the free amine 15.
We originally developed this coupling using 1-bromo-4-
(trifluoromethyl)benzene, and upon optimization we were
able to switch to the much cheaper and readily available aryl
chloride 16. Typical catalyst loading to ensure complete
consumption is at the 0.5% Pd level, with 1.5 equiv of ligand
relative to Pd. For larger-scale reactions, the catalyst load
was increased by 50% from normal lab conditions to 0.75%
Pd and 1.12% ligand. A reaction temperature of 80 °C was
utilized, as higher temperatures led to a slight erosion of
To develop a scalable synthetic strategy around the
asymmetric â-amino acid functionality, we searched for
readily available compounds that contained the desired
stereocenter. A commercially available raw material that
incorporates the desired chiral amine center, the required
ethyl group, and a handle for further elaboration was (R)-
2-amino-1-butanol. Using standard transformations for one-
carbon homologation, we developed the following synthesis
for the chiral nitrile 14 (Scheme 3). Starting with 11,4 we
protected the amine followed by conversion of the alcohol
to a mesylate. TMEDA was used as the amine base, which
provided an insoluble amine‚HCl salt in the reaction solvent
(EtOAc) and was thus removed from the reaction by
crystallization. After workup, mesylate 12 was isolated as a
crystalline solid, providing a convenient purification point.
The first two steps could be further streamlined into one
reaction by the stepwise addition of reagents to a single
reactor.
Initial attempts to displace the mesylate of 12 with simple
cyanide sources (NaCN, KCN) led to disappointing results,
mainly due to lack of solubility of the cyanide salts and the
negative effects on the reaction profile upon prolonged
heating.5 In contrast, displacement of the mesylate with
Bu4NCN was very clean in EtOAc. To complete the
sequence, the nitrogen protecting group was removed under
acidic conditions. Due to the high water solubility of the
aminonitrile, the acid for deprotection of the BOC group was
chosen so as to avoid the need for aqueous workup. The
methanesulfonic acid salt of 14 was a crystalline solid, so
deprotection with methanesulfonic acid in THF provided the
salt via direct filtration from the reaction mixture.
Difficulty in sourcing and handling Bu4NCN and high
cost forced us to reevaluate this approach, and a facile
cyanide displacement utilizing readily available materials was
sought to develop this route into a large-scale process.
Several attempts to generate Bu4NCN “in situ” using Bu4NF/
TMSCN provided a reagent mixture that was significantly
less reactive than the purified commercial Bu4NCN. We were
able to achieve a compromise using NaCN in DMF contain-
ing Bu4NBr as a solubilizing counterion. The NaCN chem-
(6) In time, other vendors were identified to provide this key raw material using
variations of the described method or alternate technologies.
(7) For overviews of the literature coupling conditions, see: (a) Wolfe, J. P.;
Buchwald, S. L. J. Org. Chem. 2000, 65, 1144-1157. (b) Wolfe, J. P.;
Tomori, H.; Sadighi, J. P.; Yin, J.; Buchwald, S. L. J. Org. Chem. 2000,
65, 1158-1174.
(4) This material was purchased from Fisher Scientific.
(5) The main impurity is (R)-4-ethyloxazolidin-2-one formed from cyclization
of the BOC-protected mesylate.
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
473

enantiomeric purity.8 Unfortunately, the amination product
17 is an oil, eliminating the opportunity to purify at this stage
for large-scale manufacturing.
Scheme 5
With successful amination our focus was turned to
elaboration of the C-4 carbon to the desired imide. We
focused our approach on generation of the primary amide at
this carbon, which required hydrolysis of the nitrile to the
desired amide. Both acidic and basic hydrolysis conditions
were explored. Several acids were examined to hydrolyze
the nitrile (CH3SO3H, HCl, HBr, TFA/acetic acid) to the
amide, but varying amounts of byproducts or hydrolysis to
the carboxylic acid were observed. However, hydrolysis in
sulfuric acid (2.2 mL/g 17) with a small amount of water (4
equiv) present provided a robust procedure. Overreaction to
the acid was not detected even when the reaction was heated
overnight. Basic hydrolysis conditions were also screened
(with or without H2O2) but were not competitive with the
sulfuric acid hydrolysis method due to lower yield and
consistent byproduct resulting from overreaction to produce
the carboxylic acid.
To address the issue that intermediate 17 is an oil, we
successfully telescoped the Pd-mediated amination reaction
and hydrolysis together. After completion of the Pd-mediated
amination reaction, the toluene solution was filtered to
remove the salts, and the sulfuric acid/water mixture was
added to the filtrate. The bilayer was stirred at 35 °C
overnight to provide clean hydrolysis. The product 18 was
completely soluble in the sulfuric acid layer (presumably as
the sulfuric acid salt), and after the reaction was complete
the toluene layer was discarded. After neutralization and
extraction with diisopropyl ether (IPE), amide 18 was isolated
via crystallization by addition of cyclohexane as an anti-
solvent.9 This procedure routinely provided an 85% yield
over the two-step sequence under laboratory conditions.
When the reaction was performed on kilogram scale, the Pd-
catalyzed reaction was slightly slower than expected based
on our previous experience on smaller scale. The two bulk-
scale reactions had 6-7% of 16 remaining after 20 h,
whereas on multihundred-gram scale the reactions were
complete in this time frame. We propose that differences in
mixing due to the heterogeneous nature of the reaction led
to this lower rate for the multikilogram-scale reactions. The
high density of Cs2CO3 led to this material being agitated
only near the bottom of the reactor, and poor agitation in
laboratory experiments also exhibited slower reactivity. Upon
further scale-up, reactors with more efficient stirring could
be chosen to minimize this effect. The two-step yields for
two bulk runs were 79% (3.2 kg) and 74% (3.0 kg). The
main difference in yield is in recovery from the crystalliza-
tion. The second run had a higher concentration of IPE, and
therefore the mother liquor retained more product.
water.3 This approach required conversion of amide 18 to a
suitable imide (Scheme 4). This transformation turned out
to be more complicated than expected, as benzylchloro-
formate was reacted with the amide under a variety of basic
conditions (LDA, NaH, DMAP, KOtBu, and NaOtBu) in
different solvents. In each case, significant starting material
would remain after multiple additions of base/chloroformate,
leading to an unacceptably low overall yield. We discovered
a satisfying balance in base strength and counterion to
provide primarily the desired transformation using a solution
of lithium tert-butoxide in THF. When this base was added
to 18 in the presence of the chloroformate, a very rapid and
clean reaction produced the desired imide 19.
With the key imide in hand, we set out to explore the
reduction/cyclization protocol outlined in the retrosynthesis.
Initially, the reduction was effected with NaBH4 in MeOH
(Scheme 5). Large amounts of borohydride (∼15 equiv) in
multiple charges were used to consume the starting material.
We attempted to use ethanol as solvent to minimize reagent
decomposition and found that the desired aminal was formed
in only small amounts and the imide starting material mainly
converted to overreduced product 21. Interestingly, when we
performed the reaction in methanol on larger scale, desired
aminal 20 was the major product, but overreduction to 21
remained a significant problem.
We also examined the use of DIBAL-H3 and borane‚
DMS. Borane gave the desired aminal, but with a new
overreduction product 22 resulting from reduction of the
acylimide before fragmentation. DIBAL-H exhibited an
improved profile but still gave significant overreduction with
this system.
We decided to examine NaBH4 reduction combined with
acid activation, as described by Speckamp to mono-reduce
succinimides.10 Use of this methodology dramatically in-
creased the rate and selectivity of the reduction, but the
protocol was not ideal for large-scale production. Since the
addition of acid to the imine/borohydride mixture led to
significant evolution of hydrogen, careful addition of the acid
was warranted and would be a difficult operation on larger
Our strategy was to access the desired cyclization precur-
sor 7 via reduction of an imide followed by elimination of
(8) Comparison of reactions at 85 and 110 °C showed 3 times as much
enantiomer by chiral HPLC at the higher temperature. Typically, 0.2-0.6%
of enantiomer was detected at the end of reactions at 85 °C.
(9) Cyclohexane can be replaced with hexanes or heptane, but cyclohexane is
the solvent of choice based on the consistent physical characteristics of the
amide thus obtained.
(10) (a) Hubert, J. C.; Wijnberg, J. B. P. A.; Speckamp, W. N. Tetrahedron
1975, 31, 1437-1441. (b) Dijkink, J.; Speckamp, W. N. Tetrahedron Lett.
1975, 46, 4047-4050.
474
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development

Scheme 6a
Figure 1. Relative stability of in situ reduction intermediates.
Scheme 7a
a Reagents and conditions: (a) i. NaBH4, CaCl2 or MgCl2; ii. H+.
scale. We found that it took several additions of acid to drive
the reduction to completion, but using this activation the
reaction gave very little (<5%) overreduction as long as the
temperature was carefully controlled.
In hopes of developing a more scalable acid-catalyzed
procedure, we reasoned that Lewis acid should activate the
imide for reduction with less borohydride decomposition.
Calcium borohydride is a known mild reducing agent, with
greater ester reduction properties as compared to sodium
borohydride.11 The addition of CaCl2 to our imide-reduction
reaction caused a significant rate increase similar to that seen
with the addition of protic acid. Under the CaCl2/NaBH4
conditions, we did not observe the extensive borohydride
decomposition, which eliminated the need to add acid in
multiple portions. Using the Lewis acid-activated con-
ditions, nearly quantitative reduction of imide 19 was
achieved in a single operation, and this was readily tele-
scoped into the desired cyclization to produce 23 (Scheme
6). After aqueous workup with toluene as the organic solvent,
catalytic TsOH‚H2O was added to the organic layer, and the
solution was stirred at room temperature until cyclization
was complete (usually 10-30 min). We were only able
to detect the cis isomer in our reaction by 1H NMR
analysis, and the compound was >99% ee by chiral HPLC
analysis.
A concern we had for scale-up was that the reduction
required careful monitoring and was temperature sensitive.
Warmer temperatures or prolonged reaction times led to
significant overreduction products. We attributed the sig-
nificantly higher yield of the desired aminal with calcium-
ion activation in comparison to proton activation to its ability
to chelate the aminal oxygen of the product and the
carbamate carbonyl, providing a more stable intermediate.
Following this rationale, it was further reasoned that a
stronger chelating cation might provide a product with
enhanced stability toward overreduction relative to calcium
(Figure 1). This led us to the use of stoichiometric magne-
sium chloride, which provided a very clean reaction in ∼30
min to the desired aminal with less than 1% overreduction.
Remarkably, the product solution could be held at 0 °C for
several hours in the presence of excess NaBH4 with no
further overreduction. Under these conditions, the majority
of the product in the calcium chloride-activated reactions
was lost to overreduction. After NaBH4 reduction with MgCl2
activation, the reaction was simply quenched with aqueous
acid and the bilayer stirred for 3-4 h to effect the desired
cyclization.
a Reagents and conditions: (a) ClCO2CH3, LiOtBu (94%); (b) NaBH4, MgCl2;
(c) aq HCl (80%).
reported resolution route, but in asymmetric form (in the
resolution route, the benzylcarbamate protected material is
racemic).1 It would require four steps to complete the
synthesis from 23 to 1 via the previously reported chemis-
try: (1) conversion of the aniline nitrogen to the ethyl
carbamate, (2) removal of the benzyl carbamate, (3) reductive
alkylation of the amine with 3,5-bis(trifluoromethyl)benz-
aldehyde, (4) and conversion of the resultant amine to the
methyl carbamate. Attempts to further shorten the sequence
are described in the next sections.
Alternate Imide Strategy. The route described in the
previous section met the initial goals of the project in
developing an asymmetric synthesis and intersecting a
common intermediate in the synthetic sequence of the
resolution route. However, some inefficiency was present
as the developed route incorporated the benzylcarbamate-
protected tetrahydroquinoline (23). In the new asymmetric
route, the benzyl carbamate is not actually necessary for the
synthesis, and the extra steps to remove this group and
replace with one of the final nitrogen substituents is
inefficient. This provided us an opportunity to further shorten
the sequence by incorporation of the methyl carbamate as
the cyclization activator, which then can remain until the
end of the synthesis.
Utilizing the chemistry we developed for the benzylcar-
bamate 19, amide 18 was acylated with methylchloroformate
using lithium tert-butoxide as previously described to provide
imide 24 (Scheme 7). Upon reaction completion and extrac-
tive workup the organic layer is distilled and displaced by
cyclohexane to crystallize the product. The yield for the bulk
run to produce 7 kg of 24 was 94%.
The key reduction/cyclization sequence was run using the
previously described procedure. For added process safety and
ease of handling on large scale, NaBH4 was used as 11-mm
pellets.12 Imide 24 was mixed with NaBH4 in EtOH/H2O,
as there is virtually no background reaction of the substrate
With the successful synthesis of tetrahydroquinoline 23,
we have relayed a common intermediate in the previously
(12) This material was purchased from Aldrich Chemical Co. Due to lower
activity of the pellets, the surface was “preactivated” by stirring the pellets
in the solvent for 20 min at room temperature.
(11) Kollonitsch, J.; Fuchs, O.; Gabor, V. Nature 1955, 346.
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
475

Scheme 8
distillation provided crystalline 26. The bulk yield was 88%
to produce 5.6 kg of 26.
The final step alkylation was examined under a variety
of base (various metal alkoxides and hydroxides, carbonates,
and amines), solvent (THF, DMAC, CH2Cl2, toluene, DMSO,
DME, IPE, MTBE, alcohols, and hexanes, or combinations
of the above), and stoichiometry conditions. The optimal base
was tert-butoxide; however, the use of excess base also leads
to a stilbene byproduct [(E)-1,2-bis(3,5-bis(trifluoromethyl)-
phenyl)ethane, 29], in some cases to the extent of several
percent.15 The formation is minimized with the weaker
hydroxide bases, or by the use of stoichiometric tert-
butoxide. The best conditions elucidated were the use of
potassium tert-butoxide in toluene or CH2Cl2. Employing
stoichiometric base minimized the formation of stilbene 29
to nearly undetectable levels at the expense of ∼5%
unreacted 26.
After the reaction is complete, excess alkylating agent
27 was quenched with DABCO. The nucleophilic amine was
efficient at converting the excess alkylating agent to the
quaternary salt 28. The quaternary salt was very stable and
not readily hydrolyzed by acidic or basic water under the
time frame required for large-scale processing, allowing
efficient removal via aqueous HCl extractions. After extrac-
tions, the IPE organic layer was displaced by distillation with
denatured EtOH16 and reduced in volume to crystallize 1.
Filtration was followed by further concentration and reseed-
ing to obtain a second crop. Both crops had no detectable
alkylating agent (27 present at <10 ppm). The bulk run
provided 6.0 kg from two crops for a 73% yield, with purities
>99%. No enantiomer or diastereomer was detected by
HPLC analysis.
Last-Step Alkylation Purge Tests. Since this material
was designated for clinical use, extra attention was given to
the use of an alkylating agent in the final step, and the
subsequent removal is described in the last section. We have
addressed the fate of the excess alkylating agent: the
DABCO quaternary salt is washed out in the aqueous layer.
We also demonstrated the purge of the alkylating agent in
the final crystallization: a sample of 1 produced by the
alkylation route was spiked with 30 000 ppm of 27 in EtOH.
The solution was then subjected to the standard crystallization
procedure, and the resultant product was isolated free from
detectable bromide 27 (<10 ppm). This purge experiment
further confirms that we can control the level of the benzyl
halide in the final steps: the fate of the material under normal
reaction conditions is the quaternary salt that is washed out
with the water, and additional purge is provided by the
crystallization procedure.
with NaBH4 without Lewis acid catalysis under the reaction
conditions (-10 to 0 °C). To this mixture, a solution of aq
MgCl2 was then added. Upon reaction completion, the
solution was quenched into an aq HCl/citric acid/CH2Cl2
mixture and stirred for ∼2 h at room temperature to effect
the acid-catalyzed cyclization to produce 25. The citric acid
was added to complex Mg salts; without citric acid severe
emulsions were encountered. After further extractions, the
product was crystallized from the CH2Cl2 by displacement
with hexanes. The bulk yield to produce 5.3 kg of 25 was
80%. Note that the diastereoselectivity for this cyclization
is comparable to that seen in the benzyl carbamate case, with
no trans isomer detected by NMR analysis.
Development of the Endgame. With a reliable synthesis
of the chiral tetrahydroquinoline core, we set out to develop
the chemistry for the final two transformations. To complete
the synthesis, the N-1 position must be converted to the ethyl
carbamate and the C-4-nitrogen must be alkylated with the
required 3,5-bis(trifluoromethyl)benzyl group. One drawback
to the approach is the use of an alkylating agent in the final
step.13 In developing this chemistry, careful attention was
given to the levels of alkylating agent present as an impurity
in the final product.
Using a procedure analogous to that previously described,1
ethyl chloroformate was added to a solution of 25 in the
presence of pyridine (Scheme 8). During prior bulk cam-
paigns, multiple charges of chloroformate were often used
to drive the reaction to completion. During our research on
this step, slow bubbling was noticed during the reaction; as
a result, we proposed that reagent decomposition could
explain the stalled reactions.14 It was visually determined
that this bubbling is significant only at temperatures above
∼15 °C. Thus, by conducting the reaction at 0 °C with careful
temperature monitoring, the reaction could be driven to
completion without additional charges of chloroformate.
Upon reaction completion, acidic workup followed by solvent
(15) This impurity arises from deprotonation at the benzyl center of 27, which
displaces the bromide of a second molecule of 27. The intermediate can
then undergo base-catalyzed elimination to provide 29. Formation of this
material must be minimized as it is difficult to purge in the final
crystallization.
(13) To circumvent a final-step alkylation, we also examined the alternate
permutation: alkylation of 25 first with bis(trifluoromethyl)benzyl bromide
followed by conversion of the tetrahydroquinoline nitrogen to the ethyl
carbamate. Although the major product from this sequence was the desired
compound, the alkylation step was not completely selective for the desired
nitrogen, leading to significant impurities.
(14) A reasonable possibility is that the acyl pyridinium is formed, displacing
chloride. The chloride then attacks the ethyl group, forming ethyl chloride
and CO2, and releasing pyridine.
(16) Anhydrous ethanol denatured with 0.5% toluene was used.
476
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development

Conclusion
The work presented describes a scalable asymmetric
synthesis of compound 1. The six-step synthesis from chirally
pure amine 14 was demonstrated on multikilogram scale,
providing the title compound in 37% overall yield.
and concentrated to afford a solid (65.22 g). A portion of
the solid (61.60 g) was transferred to a flask equipped with
an overhead stirrer. Hexanes (186 mL) was added, and the
flask was heated to 65 °C. After all the solids were in
solution, the mixture was cooled to ambient temperature and
stirred overnight. The resulting solids were isolated by
filtration to afford 52.32 g (74%) of 13. Note: for reaction
on larger scale, the water washes were replaced with 2% aq
K2CO3, and the extractions were performed with EtOAc in
Experimental Section
All materials were purchased from commercial suppliers
and used without further purification. All reactions were
conducted under an atmosphere of nitrogen unless noted
otherwise.
1
place of IPE: mp 62.4-63.3. H NMR (300 MHz, d6-
DMSO) δ 0.84 (t, 3, J ) 7.4), 1.40 (s, 9), 1.37-1.52 (m, 2),
2.53 (dd, 1, J ) 7.4, 17.0), 2.67 (dd, 1, J ) 5.2, 16.8), 3.48-
3.62 (m, 1), 7.00 (br d, 1, J ) 8.0). 13C NMR (75 MHz,
d6-DMSO) δ 10.1, 22.6, 26.5, 28.1, 48.7, 77.8, 118.5, 155.2.
Anal. Calcd for C10H18N2O2: C, 60.58; H, 9.15; N, 14.13.
Found: C, 60.40; H, 9.17; N, 14.12.
Methanesulfonic Acid (R)-2-tert-Butoxycarbonylamino-
butyl Ester (12). Run 1. BOC anhydride (515.9 g, 2.364
mol) in ethyl acetate (400 mL) was added to a solution of
R-(-)-2-amino-1-butanol (11, 200.66 g, 2.251 mol) in ethyl
acetate (1500 mL) via an addition funnel. The reaction
mixture was stirred for approximately 30 min. TMEDA (360
mL, 2.39 mol) was added, and the reaction mixture was
cooled to approximately 10 °C. Methanesulfonyl chloride
(184.7 mL, 2.386 mol) was added to the reaction mixture
over a 30-min period. After stirring for 1 h, the reaction
mixture was filtered to remove the TMEDA salts, and the
filtrate was collected.
Run 2. BOC anhydride (514.5 g, 2.357 mol) in ethyl
acetate (400 mL) was added to a solution of R-(-)-2-amino-
1-butanol (200.12 g, 2.245 mol) in ethyl acetate (1100 mL)
via an addition funnel. The reaction mixture was stirred for
approximately 30 min. TMEDA (359.1 mL, 2.379 mol) was
added, and the reaction mixture was cooled to approximately
10 °C. Methanesulfonyl chloride (184.1 mL, 2.379 mol) was
added to the reaction mixture over a 30-min period. After
stirring for 1 h, the reaction mixture was combined with the
filtrate from Run 1 and the combined mixture was filtered.
The solids were washed with ethyl acetate (400 mL), which
was collected with the reaction filtrate. Hexanes (12 L) was
added to the combined filtrates, and the mixture was cooled
in an ice/water bath. After 2.5 h the solids were isolated by
filtration, washed with hexanes (2 L), and dried under
vacuum to afford 971.57 g (81%) of 12: mp 89.9-90 (sub).
1H NMR (300 MHz, d6-DMSO) δ 0.86 (t, 3, J ) 7.4), 1.25-
1.57 (m, 2), 1.39 (s, 9), 3.15 (s, 3), 3.49-3.63 (m, 1), 4.04
(dd, 1, J ) 6.3, 10.0), 4.10 (dd, 1, J ) 5.3, 10.0), 6.85 (br
d, 1, J ) 8.3). 13C NMR (75 MHz, d6-DMSO) δ 10.0, 23.4,
28.1, 36.6, 50.8, 71.0, 77.8, 155.4. Anal. Calcd for
C10H21NO5S: C, 44.93; H, 7.92; N, 5.24. Found: C, 45.21;
H, 8.02; N, 5.12.
(R)-3-Aminopentanenitrile Methanesulfonic Acid Salt
(14). Methanesulfonic acid (48 mL, 0.74 mol) was added to
a solution of 13 (52.32 g, 0.2639 mol) in THF (530 mL).
The reaction mixture was heated to 40 °C for 30 min. The
temperature was raised to 45 °C, and the reaction mixture
was stirred for 1 h. The temperature was raised again to 65
°C, and the reaction mixture was stirred for 5 h. The solution
was allowed to cool to room temperature, and upon cooling
the product crystallized. The resulting solids were isolated
by filtration to afford 41.53 g (81%) of the title compound:
1
mp 125.1-126.0. H NMR (300 MHz, d6-DMSO) δ 0.95
(t, 3, J ) 7.5), 1.60-1.75 (m, 2), 2.39 (s, 3), 2.89 (dd, 1, J
) 6.2, 17.4), 2.97 (dd, 1, J ) 5.5, 17.4), 3.35-3.45 (m, 1),
8.16 (br s, 3). 13C NMR (75 MHz, d6-DMSO) δ 9.3, 20.4,
25.0, 39.8, 48.3, 117.2. Anal. Calcd for C6H14N2O3S: C,
37.10; H, 7.26; N, 14.42. Found: C, 37.34; H, 7.27; N, 14.61.
(3R)-3-(4-Trifluoromethylphenylamino)pentane-
nitrile (17). To a clean, dry 100-L glass reactor was charged
14 (3000 g, 15.44 mol), sodium carbonate (2.8 kg, 26.4 mol),
and methylene chloride (21 L). The heterogeneous mixture
was stirred well for at least 2 h. The mixture was filtered
and rinsed with methylene chloride (3 × 2 L). The resulting
filtrate was placed in a clean, dry, and nitrogen gas-purged
50-L glass reactor. The methylene chloride was removed by
distillation until the internal temperature reached 50-53 °C
to provide the free-based amine as a thin oil. The reactor
was then cooled to room temperature and charged with
toluene (20 L), chloro-4-(trifluoromethyl)benzene (4200 g,
23.26 mol), and cesium carbonate (7500 g, 23.02 mol). The
solution was sparged with nitrogen gas for 1 h. Near the
time of completion of the sparging, fresh catalyst solution
was prepared by charging a 2-L round-bottom flask, equipped
with stir bar and flushed with nitrogen gas, with 2-dicyclo-
hexylphosphino-2′-(N,N-dimethylamino)biphenyl (68 g, 0.17
mol), phenylboronic acid (28 g, 0.23 g), and THF (1.2 L)
followed by palladium acetate (26 g, 0.12 mol). The catalyst
solution was stirred at room temperature under nitrogen
atmosphere for 15 min. The catalyst solution was added to
the 50-L reactor with a cannula (excluding air). The mixture
was heated to 79 °C internal temperature under nitrogen
atmosphere for 16 h. The reaction solution was cooled to
room temperature and filtered through Celite. The solids were
(R)-N-tert-Butyloxycarbonyl-3-aminopentanenitrile (13).
NaCN (24.05 g, 0.4907 mol) was added to DMF (500 mL),
and the mixture was stirred at 35 °C for 30 min. Tetra-
butylammonium bromide (120.59 g, 0.3741 mol) was added,
and the reaction mixture was stirred at 35 °C for 2 h. 12
(101.23 g, 0.3787 mol) was added, and the reaction mixture
was stirred at 35 °C overnight. The mixture was then
partitioned between water (2 L) and diisopropyl ether (IPE,
1 L). The aqueous layer was extracted with IPE (1 L). The
combined organic layers were extracted sequentially with
water and a saturated solution of sodium chloride in water.
The organic layer was dried over magnesium sulfate, filtered,
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
477

rinsed with toluene (3 × 2 L), and the filtrate was collected.
All filtrates were combined to afford a crude solution of 17.
(3R)-3-(4-Trifluoromethylphenylamino)pentanoic Acid
Amide (18). Aqueous acid (8.2 L concentrated sulfuric acid
and 1.1 L water premixed and cooled to 35 °C or less) was
added to the crude toluene solution of 17 from the above
procedure. The resulting bilayer was stirred well and heated
to 35 °C for 17 h. The lower aqueous layer was collected
and the upper toluene layer discarded. The aqueous layer
was quenched with aq NaOH (95 L water and 10.7 kg
NaOH) and IPE (40 L). After extraction and removal of the
aqueous layer, the organic layer was extracted with saturated
aqueous NaHCO3 (10 L). The layers were separated and the
organic phase concentrated by distillation to a volume of 19
L. The solution was cooled to room temperature and seeded
with 18 and allowed to granulate for 3 h while stirring. To
the heterogeneous mixture was added cyclohexane (38 L),
and the mixture granulated for an additional 11 h. The solids
were filtered, rinsed with cyclohexane (4 L), and dried under
vacuum at 40 °C to provide 3173 g (79%) of 18. A repeat
of the above procedure provided 3021 g (75%) for an average
yield over two runs of 77%. 1H NMR (400 MHz, CDCl3) δ
0.98 (t, 3, J ) 7.5), 1.60-1.76 (m, 2), 2.45 (d, 2, J ) 5.8),
3.73-3.80 (m, 1), 5.53 (br s, 1), 5.63 (br s, 1), 6.65 (d, 2, J
) 8.7), 7.39 (d, 2, J ) 8.7). 13C NMR (100 MHz, CDCl3) δ
10.7, 27.8, 40.0, 51.9, 112.6, 118.9 (q, J ) 32.7), 125.2 (q,
J ) 271.0), 126.9 (q, J ) 3.8), 150.2, 174.3. Anal. Calcd
for C12H15F3N2O: C, 55.38; H, 5.81; N, 10.76. Found: C,
55.25; H, 5.98; N, 10.59.
(3R)-[3-(4-Trifluoromethylphenylamino)pentanoyl]-
carbamic Acid Benzyl Ester (19). To a clean, dry flask
was charged 18 (20.11 g, 77.27 mmol) and IPE (100 mL),
and the mixture was cooled to -12 °C. Benzyl chloroformate
(13.25 mL, 92.8 mmol) was then added followed by the slow
addition of 1.0 M lithium tert-butoxide in THF solution
(185.5 mL, 185.5 mmol). Lithium tert-butoxide solution was
added at such a rate that the internal temperature remained
below 0 °C. This addition took 1 h. Fifteen minutes after
the completion of base addition, the reaction was quenched
by adding the mixture to IPE (100 mL) and 1.5 M HCl (130
mL). The phases were separated, and the organic layer was
washed with sat. aqueous NaCl solution (130 mL). The
phases were separated, and the organic layer was dried
(MgSO4), filtered, and concentrated under partial vacuum
(at 40 °C) to a total volume of 100 mL. Additional IPE (200
mL) was added, and the solution was again concentrated
under partial vacuum (at 40 °C) to a total volume of 100
mL. After cooling, the solution was seeded with 19 and
allowed to stir at room temperature overnight. The remaining
solvent was displaced with cyclohexane using partial vacuum
distillation (45 °C bath, 200 mL followed by 100 mL). The
resultant slurry was cooled and stirred for 40 min and filtered,
and the solids were dried to provide 25.8714 g (85%) of 19:
mp 100.6-101.4. 1H NMR (400 MHz d6-acetone) δ 0.96 (t,
3, J ) 7.5), 1.57-1.75 (m, 2), 2.87 (dd, 1, J ) 6.6, 16.2),
2.97 (dd, 1, J ) 6.2, 16.2), 3.94-4.00 (m, 1), 5.16 (s, 2),
5.50 (br s, 1), 6.75 (d, 2, J ) 5.7), 7.33-7.43 (m, 7), 9.52
(br s, 1). 13C NMR (100 MHz CDCl3) δ 10.7, 28.1, 40.3,
51.5, 68.3, 112.5, 118.9 (q, J ) 32.3), 125.2 (q, J ) 269.9),
126.9 (q, J ) 3.8), 128.6, 128.98, 129.04, 135.1, 150.1,
152.1, 173.5. Anal. Calcd for C20H21F3N2O3: C, 60.91; H,
5.37; N, 7.10. Found: C, 60.96; H, 5.22; N, 7.07.
(2R,4S)-(2-Ethyl-6-trifluoromethyl-1,2,3,4-tetrahydro-
quinolin-4-yl)carbamic Acid Benzyl Ester (23). A clean,
dry flask was charged with 19 (11.51 g, 29.18 mmol) and
95% ethanol (80 mL), and the solution was cooled in an
ice/acetone bath (-12 °C). NaBH4 (0.773 g, 20.4 mmol) was
then added to the solution. The internal temperature of the
reaction was -11.5 °C. To the reaction flask was slowly
added a solution of MgCl2‚6H2O (6.23 g, 30.6 mmol, in 13
mL of H2O). The internal temperature was maintained below
-5 °C by adjusting the addition rate. The total addition time
was 15 min. Once all of the magnesium chloride solution
was added, the solution temperature was raised to 0 °C and
stirred for 30 min. The reaction was then quenched by
addition of the reaction solution to a mixture of CH2Cl2 (115
mL), 1 N HCl (115 mL), and citric acid (14.02 g, 72.97
mmol). The resulting bilayer was stirred at room temperature.
After 3.75 h, the phases were separated. Water (58 mL) and
citric acid (8.41 g, 43.77 mmol) were added to the organic
layer, and the mixture was stirred at room temperature for
45 min. The phases were separated, and G-60 Darco activated
charcoal (1.52 g) was added to the organic layer. After
stirring for 45 min, the solution was filtered through Celite
and washed with CH2Cl2 (2 × 15 mL). The filtrate was then
displaced with hexanes by distillation under atmospheric
pressure (used ∼350 mL hexanes) and concentration of the
mixture to a total volume of 230 mL. The mixture was stirred
at room temperature for 14 h and filtered, and the solids
were dried to provide 9.0872 g (82%) of 23: mp 154.0-
155.2. 1H NMR (400 MHz d6-acetone) δ 1.00 (t, 3, J ) 7.5),
1.51-1.69 (m, 3), 2.17-2.26 (m, 1), 3.46-3.54 (m, 1), 4.96
(ddd, 1, J ) 5.4, 9.5, 11.6), 5.14 (d, 1, J ) 12.9), 5.20 (d,
1, J ) 12.9), 5.66 (br s, 1), 6.65 (d, 1, J ) 8.3), 6.71 (br d,
1, J ) 9.1), 7.20 (dd, 1, J ) 1.9, 8.9), 7.30-7.43 (m, 6). 13
C
NMR (100 MHz CDCl3) δ 9.9, 29.2, 35.3, 48.2, 52.4, 67.3,
113.7, 118.9 (q, J ) 32.7), 121.4, 124.1 (q, J ) 3.8), 125.1
(q, J ) 270.6), 125.7 (q, J ) 3.8), 128.4, 128.5, 128.9, 136.6,
147.7, 156.7. Anal. Calcd for C20H21F3N2O2: C, 63.48; H,
5.59; N, 7.40. Found: C, 63.08; H, 5.50; N, 7.46.
(3R)-[3-(4-Trifluoromethylphenylamino)pentanoyl]-
carbamic Acid Methyl Ester (24). To a clean, dry 100-L
glass reactor was charged 18 (6094 g, 23.42 mol), isopropyl
ether (30 L), and methylchloroformate (2.7 kg, 29 mol). The
resulting slurry was cooled to 2 °C and the reactor jacket
set at -8 °C. The reactor was then charged with lithium
tert-butoxide solution (18-20% in THF, 24.6 kg, ∼58 mol)
at such a rate as to maintain the internal temperature below
10 °C and preferably at a temperature of about 5 °C. Ten
minutes after addition of base was complete, the reaction
was quenched by the addition of 1.5 M HCl (36 L). The
aqueous layer was removed, and the organic phase was
extracted with saturated aq NaCl solution (10 L). The
aqueous layer was removed, and the organic phase was
concentrated by distillation under vacuum and at a temper-
ature of about 50 °C until the volume was reduced to about
478
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development

24 L. Cyclohexane (48 L) was added to the reaction vessel,
and distillation was again repeated at an internal temperature
of 45-50 °C under vacuum until the volume of solution in
the vessel was reduced to 24 L. A second portion of
cyclohexane (48 L) was added to the reaction vessel, and
distillation was again repeated at an internal temperature of
45-50 °C under vacuum until the volume of solution in the
vessel was reduced to 24 L. While the temperature was held
at 50 °C, the solution was seeded with 24 and allowed to
granulate while stirring 2 h. The solution was then cooled
slowly (over 1.5 h) to room temperature and allowed to
granulate while stirring for 15 h. The mixture was filtered.
The resulting solids were rinsed with cyclohexane (10 L)
and dried under vacuum at 40 °C to afford 7504 g of solids
that contained 7 wt % of residual cyclohexane; further drying
was not necessary for continuation of the process. This
provided a theoretical recovery of 6979 g (94%) of 24;
analytical data is presented for a solvent-free sample: mp
142.3-142.4. 1H NMR (400 MHz, d6-acetone) δ 0.96 (t, 3,
J ) 7.4), 1.55-1.75 (m, 2), 2.86 (dd, 1, J ) 6.6, 16.2), 2.96
(dd, 1, J ) 6.2, 16.2), 3.69 (s, 3), 3.92-3.99 (m, 1), 5.49
(br d, 1, J ) 8.7), 6.76 (d, 2, J ) 8.7), 7.37 (d, 2, J ) 8.7),
9.42 (br s, 1). 13C NMR (100 MHz, CDCl3) δ 10.6, 28.1,
40.2, 51.5, 53.4, 112.5, 119.0 (q, J ) 32.70), 125.2 (q, J )
270.2), 126.9 (q, J ) 3.8), 150.1, 152.7, 173.4. Anal. Calcd
for C14H17F3N2O3: C, 52.83; H, 5.38; N, 8.80. Found: C,
52.71; H, 5.37; N, 8.80.
(2R,4S)-(2-Ethyl-6-trifluoromethyl-1,2,3,4-tetrahydro-
quinolin-4-yl)carbamic Acid Methyl Ester (25). To a clean,
dry 100-L glass reactor was charged 24 (7474 g of a lot that
contains 7% solvent, theoretical 6951 g, 21.84 mol) followed
by EtOH (46 L, denatured with 0.5% toluene) and water
(2.35 L). NaBH4 (620 g, 11-mm pellets, 16.4 mol) was added
to the solution in one portion. Some off-gassing of hydrogen
occurred; nitrogen gas purging was maintained. The mixture
was stirred at room temperature for 20 min and then cooled
to -10 °C. A solution of 3.3 M aq MgCl2 solution (4.68 kg
of MgCl2‚6H2O, 23.0 mol in 7 L of water) was added at
such a rate that the internal temperature did not exceed -5
°C. Once addition was complete, the reaction solution was
warmed to 0 °C for 45 min. The reaction was quenched by
transferring the reaction mixture to a 200-L reactor containing
methylene chloride (70 L), and 1 M HCl/citric acid solution
(5.8 L of concentrated HCl, 64 L of water, and 10.5 kg of
citric acid). The headspace of the reactor was purged with
nitrogen gas since the quench liberates some H2. This bilayer
was stirred at room temperature for 2 h. The phases were
separated, and the lower organic product layer was removed.
After aqueous layer removal, the organic phase was returned
to the reaction vessel and extracted with an aqueous citric
acid solution (6.3 kg of citric acid, 34 L of water). The
mixture was stirred for 1 h and allowed to settle overnight.
The layers were separated, and to the organic phase was
added Darco activated carbon (G-60 grade, 700 g), and the
solution was stirred for 30 min. The mixture was filtered
through Celite, and the carbon was rinsed twice with
methylene chloride (14 and 8 L). The filtrate was distilled
while periodically adding hexanes so as to displace the
methylene chloride with hexanes to a total final volume of
70 L (the internal volume was kept at approximately 65-
70 L, total hexanes used ) 112 L). Product crystallized
during the displacement. Once a stable distillation temper-
ature was reached, the solution was cooled and granulated
at room temperature for 10 h. The solids were filtered off,
rinsed with hexanes (14 L), and dried at 40 °C under vacuum
to provide 5291 g (80%) of 25: mp 139.0-140.5. 1H NMR
(400 MHz, d6-acetone) δ 1.00 (t, 3, J ) 7.5), 1.51-1.67
(m, 3), 2.19 (ddd, 1, J ) 2.9, 5.4, 12.4), 3.44-3.53 (m, 1),
3.67 (s, 3), 4.89-4.96 (m, 1), 5.66 (br s, 1), 6.56 (br d, 1, J
) 8.7), 6.65 (d, 1, J ) 8.7), 7.20 (d, 1, J ) 8.7), 7.30 (br s,
1). 13C NMR (100 MHz, CDCl3) δ 9.9, 29.2, 35.5, 48.1,
52.4, 52.6, 113.7, 118.9 (q, J ) 33.1), 121.4, 124.1 (q, J )
3.8), 125.1 (q, J ) 270.6), 125.7 (q, J ) 3.8), 147.7, 157.3.
Anal. Calcd for C14H17F3N2O2: C, 55.62; H, 5.67; N, 9.27.
Found: C, 55.68; H, 5.78; N, 9.31.
(2R,4S)-2-Ethyl-4-methoxycarbonylamino-6-trifluoro-
methyl-3,4-dihydro-2H-quinoline-1-carboxylic Acid Ethyl
Ester (26). To a clean, dry 100-L glass reactor was charged
25 (5191 g, 17.17 mol), methylene chloride (21 L), and
pyridine (4.16 L, 51.4 mol). The reaction vessel was cooled
to -10 °C. Ethyl chloroformate (4.10 L, 42.9 mol) was
slowly added at such a rate that the internal temperature did
not exceed -5 °C (jacket temperature was set at -25 °C to
absorb exotherm). The reaction solution was brought to 0
°C and held for 20 h. The reaction was quenched by adding
to a mixture of IPE (36 L), CH2Cl2 (6.2 L), and 1.5 M HCl
solution (52 L). The resulting phases were separated, and
the organic layer was extracted with 1 M NaOH solution
(15 L). The resulting phases were separated, and the organic
layer was extracted with a sat. aq NaCl solution (15 L). The
resulting phases were separated, and the organic layer was
concentrated by distillation to a volume of 40 L. Crystal-
lization initiated at the lower volume. The CH2Cl2 was
displaced with IPE by distilling the mixture and periodically
adding IPE to maintain a constant volume at ∼40 L until a
distillation temperature of 68 °C was maintained (46 L total
IPE used). The mixture was cooled and allowed to granulate
with stirring at room temperature for 19 h. The solids were
filtered, rinsed with IPE (8 L), and dried under vacuum at
40 °C to provide 5668 g (88%) of 26: mp 157.3-157.6. 1H
NMR (400 MHz, d6-acetone) δ 0.84 (t, 3, J ) 7.5), 1.26 (t,
3, J ) 7.0), 1.44-1.73 (m, 3), 2.59 (ddd, 1, J ) 4.6, 8.3,
12.9), 3.67 (s, 3), 4.14-4.28 (m, 2), 4.46-4.54 (m, 1), 4.66-
4.74 (m, 1), 6.82 (br d, 1, J ) 9.1), 7.53 (s, 1), 7.58 (d, 1,
J ) 8.3), 7.69 (d, 1, J ) 8.3). 13C NMR (100 MHz, CDCl3)
δ 9.9, 14.6, 28.5, 38.1, 46.9, 52.6, 53.7, 62.4, 120.8 (q, J )
3.4), 124.32 (q, J ) 271.7), 124.36 (q, J ) 3.4), 126.38,
126.46 (q, J ) 32.7), 134.7, 139.7, 154.7, 156.9. Anal. Calcd
for C17H21F3N2O4: C, 54.54; H, 5.65; N, 7.48. Found: C,
54.50; H, 5.68; N, 7.55.
(2R,4S)-4-[(3,5-Bis(trifluoromethyl)benzyl)methoxy-
carbonylamino]-2-ethyl-6-trifluoromethyl-3,4-dihydro-2H-
quinoline-1-carboxylic Acid Ethyl Ester (1). To a clean,
dry 100-L glass reactor was charged 26 (5175 g, 13.82 mol),
CH2Cl2 (20 L), and potassium tert-butoxide (1551 g, 13.82
mol) at room temperature. The mixture was stirred for 5 min.
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
479

3,5-Bis(trifluoromethyl)benzylbromide (3.50 L, 19.1 mol)
was added to the mixture in one portion. The internal
temperature was maintained between 20 and 25 °C for 1.5
h. After 2.3 h of reaction time, an additional charge of
potassium tert-butoxide (46.10 g, 0.41 mol) was added. After
a total reaction time of 4.5 h, the reaction was quenched.
1,4-Diazabicyclo[2.2.2]octane (DABCO, 918 g, 8.18 mol)
was added to the reaction solution, and the mixture was
stirred 1 h. IPE (40 L) and 0.5 M HCl (30 L) were added to
the reaction mixture. The resulting organic and aqueous
phases were separated, and the organic layer was extracted
with 0.5 M HCl (2 × 30 L). The resulting organic and
aqueous phases were then separated, and the organic layer
was extracted with sat. aq NaCl (15 L), and the layers were
separated. Anhydrous magnesium sulfate (3.5 kg) was added
to the organic layer, and the mixture was stirred for 30 min.
The mixture was then filtered (0.5 µm filter) into a 50-L
glass reactor with IPE wash (8 L) in two portions. The filtrate
was concentrated under vacuum to a total volume of 12 L
(jacket temp 45 °C, max internal temp 35 °C). EtOH (25 L,
denatured with 0.5% toluene) was added to the oil, and the
solution was concentrated under vacuum to a volume of 12
L. To the solution was added EtOH (15 L, with 0.5%
toluene), and the solution was again concentrated under
vacuum to a volume of 12 L. The solution was cooled to
room temperature and seeded with 1 (3 g). The solution was
granulated for 38 h and filtered, and the solids were rinsed
with EtOH (4 L + 2 L, each with 0.5% toluene). The solids
were dried under vacuum (no heat) to provide 4610 g (55%)
of the title compound 1. The mother liquor from the above
filtration was concentrated under vacuum (solution temp )
62 °C) to a final volume of 6 L and cooled to 38 °C. The
solution was seeded with 1 (0.5 g) and allowed to cool and
granulate while stirring for 19 h. The mixture was filtered,
and the solids were rinsed with EtOH (2.5 L containing 0.5%
toluene). The resulting cake was dried under vacuum (no
heat) to provide 1422 g (17%) of the title compound as the
second crop. Combined recovery of 1 was 6032 g (73%).
The final material was identical to previously prepared
material1 by chiral and achiral HPLC analysis. 1H NMR (600
MHz, CDCl3, 55 °C) δ 0.75 (br s, 3), 1.29 (t, 3, J ) 7.1),
1.40-1.47 (m, 2), 1.65-1.69 (m, 1), 2.26 (br s, 1), 3.80 (br
s, 3), 4.17-4.29 (m, 3), 4.32-4.37 (m, 1), 5.2 (br s, 2), 7.13
(s, 1), 7.51 (br d, 1, J ) 8.3), 7.58 (br d, 1, J ) 8.3), 7.74
(br s, 2), 7.80 (s, 1). 13C NMR (150 MHz, CDCl3, 55 °C) δ
9.5, 14.6, 29.4, 37.1, 47.2, 53.7 (2 C), 54.7, 62.5, 119.8,
121.7, 123.5 (q, J ) 273), 124.3 (q, J ) 272), 126.7, 127.1,
127.6, 132.5 (q, J ) 33.4), 133.8, 140.6, 141.9, 154.7, 157.5.
Anal. Calcd for C26H25F9N2O4: C, 52.01; H, 4.20; N, 4.67.
Found: C, 51.97; H, 4.08; N, 4.55.
Acknowledgment
We thank Steve Buchwald, Robert Singer, Tamim Braish,
and Steve Ley for helpful discussions concerning this
research. We thank Kyle Leeman for development of a chiral
HPLC analysis for 17 and John Larrman and group for
analysis of 3,5-bis(CF3)benzyl bromide samples.
Received for review January 13, 2006.
OP060013I
480
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development