Stereoselective process for a CCR3 antagonist

Article abstract of DOI:10.1021/op050202l

A convergent, multikilogram, stereoselective synthesis of 1 is described. A key fragment, (S)-3-(4-fluorobenzyl)piperidine (2) was synthesized from valerolactam in three steps using our recently discovered Ir-BDPP-catalyzed asymmetric hydrogenation. Anoth

Full text of DOI:10.1021/op050202l

Organic Process Research & Development 2006, 10, 262−271
Stereoselective Process for a CCR3 Antagonist
Tai-Yuen Yue,* Douglas D. McLeod, Kevin B. Albertson,‡ Steven R. Beck,† Joerg Deerberg, Joseph M. Fortunak,¶
William A. Nugent, Lilian A. Radesca,§ Liya Tang, and Cathie Dong Xiang#
Chemical Process Research and DeVelopment, Pharmaceutical Research Institute, Bristol-Myers Squibb Company,
One Squibb DriVe, New Brunswick, New Jersey 08903-0191, U.S.A.
Scheme 1. Retrosynthetic analysis
Abstract:
A convergent, multikilogram, stereoselective synthesis of 1 is
described. A key fragment, (S)-3-(4-fluorobenzyl)piperidine (2)
was synthesized from valerolactam in three steps using our
recently discovered Ir-BDPP-catalyzed asymmetric hydrogen-
ation. Another key fragment, (1R,2R)-2-(benzyloxycarbonyl-
amino)cyclohexanecarboxaldehyde (3) was synthesized from
meso-hexahydrophthalic anhydride in seven steps. The stereo-
chemistry was set in the first step of this sequence via a
quinidine-mediated desymmetrization of the meso-anhydride.
Coupling of the fragments 2 and 3 followed by deprotection
provided the penultimate 23. The active pharmaceutical ingre-
dient (API) free base 1 was obtained by treatment of 23 with
the aminothiazole fragment 4 under mild conditions.
20-kg pilot-plant manufacturing campaign are described
herein.
Introduction
Many APIs contain multiple stereogenic centers which
present synthetic challenges to process chemists. The ad-
vances in stereoselective synthesis and catalysis and their
applications are increasingly evident in the process develop-
ment of pharmaceuticals.1 An example is our target molecule
1,
Results and Discussion
Discovery Chemistry Synthesis. Our Discovery Chem-
istry colleagues employed a convergent synthesis for 1 from
three fragments, the enantiopure 3-(4-fluorobenzyl)piperidine
(2), enantiopure cyclohexane aminoaldehyde 3, and ami-
nothiazole fragment 4.2 Our process research efforts focused
on the development of efficient and scalable synthetic routes
to the two enantiopure fragments 2 and 3. The union of these
fragments could be achieved by a reductive amination
reaction between 2 and 3, followed by urea formation with
fragment 4 (Scheme 1).
Synthesis of Enantiopure Piperidine Fragment 2.
Several syntheses of the optically active piperidine 2 have
been reported.2-4 Reduction of N-Boc-3-(4-fluorobenzylidene)-
piperidine followed by deprotection gave the racemic pip-
eridine 2, which was then resolved by treatment with (R)-
mandelic acid in acetonitrile.3 Alternatively, hydrogenation
of the hydrochloride salt of (4-fluorophenyl)(pyridine-3-yl)-
a CCR3 antagonist, which has potential in the prevention of
inflammation in asthma and allergic rhinitis.2 A stereose-
lective synthesis and the scale-up issues associated with a
* To whom correspondence should be addressed. Telephone (732) 227-6857.
E-mail: tai.yue@bms.com.
‡ Current address: Merck and Co., Inc., WP14-1101, 770 Sumneytown Pike,
West Point, PA 19486-0004.
(2) De Lucca, G. V.; Kim, U. T.; Vargo, B. J.; Duncia, J. V.; Santella, J. B.;
Gardner, D. S.; Zheng, C.; Liauw, A.; Wang, Z.; Emmett, G.; Wacker, D.
A.; Welch, P. K.; Covington, M.; Stowell, N. C.; Wadman, E. A.; Das, A.
M.; Davies, P.; Yeleswaram, S.; Graden, D. M.; Solomon, K. A.; Newton,
R. C.; Trainor, G. L.; Decicco, C. P.; Ko, S. S. J. Med. Chem. 2005, 48,
2194-2211.
(3) Varnes, J. G.; Gardner, D. S., Santella, J. B.; Duncia, J. V.; Estrella, M.;
Watson, P. S.; Clark, C. M.; Ko, S. S.; Welch, P.; Covington, M.; Stowell,
N.; Wadman, E.; Davies, P.; Solomon, K.; Newton, R. C.; Trainor, G. L.,
Decicco, C. P.; Wacker, D. A. Bioorg. Med. Chem. Lett. 2004, 14, 1645-
1649.
† Current address: Roche Colorado, 2075 North 55th Street, Boulder, CO
80301.
¶ Current address: Department of Chemistry, Howard University, 525 College
St., NW Washington, DC 20059.
§ Current address: Alnylan Pharmaceuticals, 300 Third Street, Cambridge,
MA 02142.
# Current address: Pfizer Inc., 10777 Science Center Drive (CB3), San Diego,
CA 20059.
(1) (a) Blaser, H. U.; Spindler, F.; Studer, M. Appl. Catal., A. 2001, 221, 119-
143. (b) Hawkins, J. M.; Watson, T. J. N. Angew. Chem., Int. Ed. 2004, 43,
3224-3228. (c) Breuer, M.; Ditrich, K.; Habicher, T.; Hauer, B.; Kesseler,
M.; Stuermer, R.; Zelinski, T. Angew. Chem. Int. Ed. 2004, 43, 788-824.
(4) Emmett, G. C.; Cain, G. A.; Estrella, M. J.; Holler, E. R.; Piecara, J. S.;
Blum, A. M.; Mical, A. J.; Teleha, C. A.; Wacker, D. A. Synthesis 2005,
92-96.
262
•
Vol. 10, No. 2, 2006 / Organic Process Research & Development
10.1021/op050202l CCC: $33.50 © 2006 American Chemical Society
Published on Web 01/12/2006

Scheme 2. Synthesis of (S)-3-(4-fluorobenzyl)piperidine (2)
methanol using 5% PtO2 in ethanol gave (4-fluorophenyl)-
(piperidinyl-3-yl)methanol which was then treated with
trimethylsilyl iodide to afford the racemic piperidine 2, which
was resolved similarly.4
Hydrogenation of benzylidenelactam 6 was carried out
in methanol using 10% Pd/C under 55 psig pressure to give
racemic benzyllactam 7 in 90% yield. The benzyllactam 7
was converted to racemic piperidine 2 by reduction with
lithium aluminum hydride (LAH) in THF. Rochelle’s salt
solution was identified as the best quench for the LAH
reaction mixture in terms of ease of phase separation and
control of heat evolution. The racemic piperidine 2 was
extracted into ethyl acetate (solution yield 88%). Our
medicinal chemistry collegues had shown that mandelic acid
salt formation in acetonitrile provided optimal resolution for
racemic piperidine 2 after screening some chiral acids and
solvents.4 Following that procedure, a solvent switch from
ethyl acetate to acetonitrile was carried out to prepare for
the mandelic acid salt formation. Our studies on the
stoichiometry of mandelic acid and the isolation temperature
showed 0.7 equiv and isolation at 50 °C gave optimal product
purity and recovery. The de of the crude mandelic acid salt
of piperidine 2 was observed to be 84%, and was readily
upgraded to 97% by reslurry in acetonitrile at 50 °C. The
overall yield of enantiopure piperidine 2 from olefin 6 was
25%.
Asymmetric Hydrogenation of 6. Although the Pd/C
catalysed hydrogenation and chemical resolution sequence
provided a practical scale-up for piperidine 2, improvement
in the yield and efficiency was highly desirable. In our effort
to identify suitable catalysts for the asymmetric hydrogena-
tion of the benzylidenelactam 6, a 256-variation catalyst
screen was run.6 After the initial hits, the screen was refined
by focusing on additives and solvents. A cationic Ir-BDPP
complex was identified to have the best selectivity (90% ee)
under practical loading (0.2-1.0 mol %).6 Recrystallization
of the hydrogenation substrate 6 from IPA/water mixture was
routinely carried out to remove any traces of potential poisons
to ensure catalyst performance. Process development of this
new technology indicated the robustness of this reaction. The
Ir-BDPP catalyst was readily generated in situ from
[(COD)2Ir]BF4 and (2S,4S-BDPP). The reaction kinetics and
selectivity were almost identical for runs carried out in 1-L
laboratory reactor and a 400-L reactor under the same
reaction conditions (20 °C, 55 psig hydrogen pressure, 0.25
mol % catalyst loading, 10 volumes of 1:1 DCM/methanol).
The reaction solvent was chosen for its good solubility of
We envisioned that inexpensive and readily available
valerolactam could serve as a good starting material for the
multikilogram synthesis of piperidine 2. Valerolactam 5 can
be readily converted to the exocyclic olefin 6 using a
literature procedure.5 Benzylidenelactam 6 can be conve-
niently reduced to the benzyllactam 7 via catalytic hydro-
genation. The benzyllactam 7 can be converted to racemic
piperidine 2 by lithium aluminum hydride reduction. Chemi-
cal resolution by diastereomeric salt formation would then
furnish enantiopure piperidine 2. To improve the overall
efficiency of this process, asymmetric hydrogenation of olefin
6 was also pursued. A comparison of the chemical resolution
and asymmetric catalysis routes is discussed below (Scheme
2).
Racemic Hydrogenation of 6 and Resolution of 7.
Preparation of arylidenelactams by aldol condensation of
N-acetyllactams and benzaldehydes under basic conditions
(potassium tert-butoxide or sodium hydride) in modest yields
(20-45%) has been reported.5 The low yields were attributed
to an undesired competing aldol reaction of the acetyl group.5
Changing the protecting group to trifluoroacetyl eliminated
this undesired aldol reaction and perhaps also increased the
acidity of the R-proton. Preparation of 6 thus involved
protection of valerolactam with trifluoroacetic anhydride
(TFAA) in toluene.6 Removal of the trifluoroacetic acid
(TFA) byproduct was carried out by vacuum distillation, and
the TFA content was monitored by 19F NMR. The TFA-
protected valerolactam was mixed with 4-fluorobenzalde-
hyde, and the resultant mixture was charged to a 1.0 M
solution of potassium tert-butoxide in THF. The crude
product 6 was readily isolated by filtration after the reaction
mixture was quenched by addition of water. Reslurry
purification of the crude product in a mixture of isopropyl
acetate and heptane improved the purity from 70.3 to 99.1
wt % and afforded benzylidenelactam 6 in 54% isolated
yield.
(5) Bose, A. K.; Fahey, J. L.; Manhas, M. S. Tetrahedron 1974, 30, 3-9.
(6) For a detailed screening protocol, see the Supporting Information for Yue,
T.-Y.; Nugent, W. A. J. Am. Chem. Soc. 2002, 124, 13692-13693.
Vol. 10, No. 2, 2006 / Organic Process Research & Development
•
263

Scheme 3. Discovery chemistry synthesis of aminoaldehyde 3
Table 1. Comparison of the Pd/C and Ir-BDPP
been published by our Discovery Chemistry group (Scheme
3).2 The known amino alcohol 138c was protected as its
benzyl carbamate and was then subjected to Swern oxidation
conditions to furnish the aldehyde 3.2 Preparation of the
alcohol 13 from the readily available keto-ester 8 was
described by Corey et al.8c The moderate diastereoselectivity
(de ) 81%) in the formation of the amino ester 10 and the
need for chromatographic purification of intermediates 10
and 11 prompted us to consider an alternative synthesis.
A selective synthesis utilizing desymmetrization of a
meso-cyclic anhydride by ring-opening was envisioned.9 The
use of cinchona alkaloids as chiral mediators in these
reactions is well documented.10 This synthetic route, shown
in Scheme 4, was evaluated and found to be more efficient
in pilot-plant operations compared to the route outlined in
Scheme 3.
Using a procedure similar to that described by Bolm et
al,10c opening of the meso-hexahydrophthalic anhydride (14)
by ethanol in the presence of quinidine (1.3 equiv) furnished
enantiomerically enriched cis-hemi-ester 15 (87% ee) in good
conversion (99%). Quinidine was removed by extraction with
aqueous sulfuric acid, which allowed for recycling of this
alkaloid to reduce cost. The toluene solution containing cis-
hemi-ester 15 was used without further purification in the
epimerization step after removal of the residual water and
ethanol by distillation. Epimerization of the cis-hemi-ester
15 to its trans-isomer 16 was carried out at -15 °C using
potassium tert-amylate as the base. The reaction was
considered complete when equilibrium was reached at less
than 4% cis-hemi-ester 15. A multistage quench using glacial
acetic acid followed by aqueous hydrochloric acid eliminated
hydrolysis on workup. Isolation of the trans-ester 16 as its
(R)-R-methylbenzylamine salt 17 improved the ee from 87%
to greater than 99% in 65% yield over the three steps from
anhydride 14.
Hydrogenation Processes
10% Pd/C, 50% wet
Degussa type E101
Ir-BDPP
20:1
enantiomeric ratio
(as is from reduction)
reaction temperature
pressure
1:1
20 °C
55 psig
methanol
20 °C
55 psig
methanol/DCM
(1:1 v/v)
0.1 kg/L
14 h
solvent
concentration
reaction time
0.2 kg/L
6 h
overall yield from 6 to 2
optical purity of 2
(after salt formation)
25%
97% ee
79%
99% ee
the substrate 6. The catalyst was removed by filtration
through silica gel and Celite, and recovery was not attempted.
Crystallization of the benzyllactam 7 (88% ee) from a
number of solvents did not improve the enantiomeric purity.
Benzyllactam 7 was converted to the piperidine 2 (LAH in
THF/toluene) without any loss in enantiomeric purity. The
formation of the mandelic acid salt of piperidine 2 further
enhanced the ee of the derived free base to 99%, and the
salt was readily isolated by filtration.2,3,4 The overall yield
of enantiopure piperidine 2 from olefin 6 was 79%. Table 1
shows a comparison of the Pd/C and Ir-BDPP processes.
Synthesis of Enantiopure Aminoaldehyde Fragment
3. Enantiopure 2-aminocycloalkanecarboxylic acids and their
derivatives are common structural units in many natural
products and biologically active compounds. The synthesis
of this class of â-amino acids has attracted considerable
attention.7
The trans-hemi-ester 16 was liberated from the salt 17
by treatment with aqueous hydrochloric acid and extracted
into toluene. The toluene solution of 16 was distilled to a
water content of less than 0.1% to minimize the potential
for generation of hydrazoic acid by hydrolysis. The resultant
A synthesis of the aminoaldehyde 3 based on the
stereoselective reduction of a chiral enamino-ester 98 has
(9) For a review: Spivey, A. C.; Andrew, B. I. Angew. Chem., Int. Ed. 2001,
40, 3131.
(7) For a review: Fu¨lo¨p, F. Chem. ReV. 2001, 101, 2181.
(8) (a) Bartoli, G.; Cimarelli, C.; Marcantoni, E.; Palmieri, G.; Petrini, M. J.
Org. Chem. 1994, 59, 5328-5335. (b) Cimarelli, C.; Palmieri, G.; Bartoli,
G. Tetrahydron: Asymmetry 1994, 5, 1455-1458. (c) Hayashi, Y.; Rhode,
J. J.; Corey, E. J. J. Am. Chem. Soc. 1996, 118, 5502-5503
(10) (a) Hiratake, J.; Yamamoto, Y.; Oda, J., J. Chem. Soc., Chem. Commun.
1985, 1717. (b) Hiratake, J.; Inagaki, M.; Yamamoto, Y.; Oda, J. J. Chem.
Soc., Perkin Trans. 1 1987, 1053. (c) Bolm, C.; Gerlach, A.; Dinter, C. L.
Synlett 1999, 195.
264
•
Vol. 10, No. 2, 2006 / Organic Process Research & Development

Scheme 4. Synthesis of aminoaldehyde 3 from meso-anhydride (14)
16 was subjected to Curtius rearrangement conditions similar
to those described by Mittendorf et al.11 The toluene solution
containing 16 was treated with triethylamine and heated to
85 °C. Diphenyl phosphoryl azide (DPPA) in toluene was
charged to this mixture at a rate to control the batch
temperature between 85 and 90 °C. Consumption of the
trans-hemi-ester 16 (over 98% as evident by HPLC) was
observed in 30 min, and benzyl alcohol was added to the
intermediate isocyanate to give the Cbz-protected amino-
ester 18 in 80% isolated yield. In this reaction only a small
amount (up to 0.5 area %) of the symmetrical urea 19 was
formed, which was removed by an 2-propanol/water recrys-
tallization to a very low level (0.1 area %).
Important Safety Information. With respect to the safety
issues of this reaction involving DPPA, we carried out
thermochemical studies which showed the heat flow of the
reaction between 16 and DPPA is addition controlled.
Analysis by React-IR showed that no buildup of the
potentially dangerous acylazide occurred under these reaction
conditions. The complete consumption of 6 and production
of one equivalent of nitrogen gas (with respect to DPPA)
under the dry reaction conditions (less than 0.1% water)
suggested that the hydrolysis of DPPA to hydrazoic acid by
residual water is slow. However, as the formation of
hydrazoic acid in this reaction was not thoroughly investi-
gated, we had taken measures to ensure low water content
in the reaction mixture in pilot-plant runs. Also see caution
in Experimental Section.
Amino-ester 18 was reduced to the aminoaldehyde 3 by
treatment with DIBAL-H in toluene/methylcyclohexane at
-90 °C. Formation of the over-reduction product, the amino
alcohol 20, was suppressed to 4.4% at this low temperature
as compared to 12.5% at -78 °C. The product isolation and
purity were improved by formation of the bisulfite adduct
21. Unreacted starting material 18 and the amino alcohol 20
were selectively removed in the mother liquors. After
liberation of the aminoaldehyde 3 from the bisulfite adduct
21 with aqueous potassium carbonate, the yield of the
aminoaldehyde 3 was 85%.
The overall yield from the anhydride 14 to the aminoal-
dehyde 3 was 44%. The desymmetrization reaction was
shown to be robust on scale-up and consistent in enantiose-
lectivity. The sequence involves four isolations, and the
crystalline intermediates 17, 18 , and 21 provided convenient
purification points.
Formation of the Penultimate (23) and API (1). With
the enantiopure benzylpiperidine 2-mandelate and aminoal-
dehyde 3 in hand, the emphasis of development work turned
to the coupling reactions to bring the fragments together,
while maintaining the overall chiral purity (Schemes 5, 6).
The reaction sequence followed the Discovery Chemistry
synthesis.2 The reductive amination conditions developed and
scaled utilized sodium triacetoxyborohydride (STAB) in
DMSO at 10-14 °C, with a slow addition of the substrates
to the STAB solution to control the reaction temperature.
No epimerization of the stereogenic centers was observed
under these conditions. The coupled product 22 was worked
up with aqueous NaOH and extracted into isopropyl acetate
(iPrOAc). It was readily deprotected by either of two
methods. Treatment of 22 with hydrobromic acid in acetic
acid/iPrOAc furnished the deprotected diamine 23 and was
isolated as its bis-hydrobromide salt. However, the byproduct
benzyl bromide is a lachrymator and poses hazards in
operations. Hydrogenolysis was used as a greener alternative.
An iPrOAc solution of 22 was hydrogenated in the presence
of palladium on carbon under 55 psig pressure for 4 h to
remove the Cbz-group to give the penultimate 23 (see
Scheme 5). The Pd catalyst was removed by filtration, and
the iPrOAc solution of the penultimate 23 was dried by
azeotropic distillation to set the stage for the final API
formation step.
The thiazolecarbamate fragment 4 was prepared in 78%
isolated yield by treatment of a suspension of aminothiazole
24 and diphenyl carbonate in THF with sodium hydride as
reported in the literature (Scheme 6).2 A suspension of
sodium hydride in mineral oil (60 wt %) was added in
(11) Mittendorf, J.; Benet-Buchholz, J.; Fey, P.; Mohrs, K.-H. Synthesis 2003,
1, 136.
Vol. 10, No. 2, 2006 / Organic Process Research & Development
•
265

Scheme 5. Formation of the penultimate 23
Scheme 6. Formation of the API
multiple portions to control heat and gas evolution. Thermal
conditions (e.g., 100 °C in DMSO, NMP, and toluene) and
other basic conditions (n-BuLi in THF, NaHMDS or LiH-
MDS in THF) were attempted; however, all gave low
conversion or complex reaction mixtures.
overall yield. The coupling of these fragments followed
standard chemistry. The API formation procedure was shown
to be robust in producing the desired solid form.
Experimental Section
The thiazole carbamate 4 was dissolved in THF and added
to the iPrOAc solution of the penultimate 23 at 40 °C. Upon
reaction completion, the THF/iPrOAc solvent mixture was
switched to IPA by distillation. The free base API 1 was
crystallized from a mixture of 2-propanol and water. The
isolated yield of the API 1 was 80% from 2 and 3.
The API free base 1 unexpectedly gave a new polymorph
on large scale (1000 L). The new polymorph was shown to
be higher melting by DSC. The low solubility of this new
form (18 vs 38 mg/mL) led to difficulties in making the
fumarate salt which was the desired solid form of this
compound. The free base 1 was dissolved in 1-methoxy-2-
propanol at 70 °C and a solution of fumaric acid (0.2 equiv
to the free base 1) in 1-methoxy-2-propanol was added while
maintaining the batch temperature at 70 °C. Seed crystals
were charged to this clear solution. Once the fumarate salt
had started to crystallize, the remaining fumaric acid (0.8
equiv) in 1-methoxy-2-propanol was added. The batch was
then cooled to 20 °C, and the product was collected by
filtration. This process gave an 86% yield of the final product
with higher than 99.5% purity. The dried final product was
analyzed by XRPD and DSC and was shown to be the
desired form.
Chiral HPLC analytical methods were developed using
authentic racemic samples. The Chiralpak AS and Chiralpak
OD columns used to determine enantiomeric purity of the
products were purchased from Chiral Technologies, Inc.,
Exton, PA. NMR spectra were obtained at 25 °C in CDCl3
except as indicated, and field strengths for the various nuclei
were as follows: 1H (400.1 MHz), 13C (100.6 MHz), 19F
(376.5 MHz), 31P (162.0 MHz). Coupling constants (J) are
given in Hertz.
3-(4-Fluorobenzylidene)-2-piperidone (6). A 30-gal
(120-L) glass-lined reactor was charged with valerolactam
(18.0 kg, 182 mol) and toluene (42.0 kg). Trifluoroacetic
anhydride (42.0 kg, 200 mol) was added below 25 °C over
30 min, and the mixture was stirred for an additional 60 min.
Volatiles were distilled off at 60-70 ° C at 140 Torr. Toluene
(50.0 kg) was added, and the mixture was further distilled
until approximately 27 L remained in the pot; this process
was repeated twice. To the residue was added tert-butyl
alcohol (18.8 kg) and 4-fluorobenzaldehyde (20.3 kg, 164
mol). The resultant solution was slowly transferred to a 100-
gal (400-L) glass-lined reactor containing potassium tert-
butoxide in tetrahydrofuran (122 kg of a 20% solution, 218
mol). The mixture was warmed to 60 °C for 1 h. Volatiles
were distilled at 40-50 °C at 250 Torr to a volume of
approximately 100 L. Water (180 L) was added to the slurry
at 40 °C, the mixture was cooled to 5 °C for 2 h, and the
solid was collected by filtration. The wet cake was washed
and reslurried with water (3 × 200 L) and dried in a tray
dryer (50 °C, 50 Torr) to afford crude 3-(4-fluoroben-
zylidene)-2-piperidone (32.6 kg); the purity by HPLC was
70.3 wt % (95.4 area %).
Conclusion
The target compound 1 was synthesized in multikilogram
quantities by a convergent synthesis from readily available
starting materials. The stereochemistry of the key piperidine
fragment 2 was established by our recently reported Ir-
BDPP-catalyzed asymmetric hydrogenation,6 and its prepara-
tion from valerolactam involved four steps with two isola-
tions in 43% overall yield. The preparation of another key
fragment 3 followed the well-known desymmetrization of a
meso-anhydride, which was shown to be practical and robust.
This sequence starting from meso-hexahydrophthalic anhy-
dride involved seven steps with four isolations and 44%
The crude product 6 (32.6 kg) was charged to a 100-gal
(400-L) glass-lined reactor with isopropyl acetate (80 kg).
Residual water was removed by distillation at atmospheric
pressure to a volume of 70 L. Heptane (56 kg) was added at
80 °C, and the slurry was cooled to 5 °C. Filtration, washing
266
•
Vol. 10, No. 2, 2006 / Organic Process Research & Development

with a mixture of isopropyl acetate (8 kg) and heptane (62
kg) at 5 °C, and drying (50 °C, 50 Torr) afforded purified
3-(4-fluorobenzylidene)-2-piperidone 6 (20.0 kg, 54% yield)
which by HPLC analysis has a purity of 99.1 wt % (99.6
area %). To ensure the material was free from catalyst
poisons, the majority of the product (19.7 kg) was further
crystallized by dissolution in 2-propanol (61.9 kg) and water
(20 L) at 75 °C. The mixture was seeded at 70 °C and cooled
to 30 °C, whereupon water (140 L) was added. After standing
overnight, the mixture was cooled to 2 °C, and the product
was collected by filtration, washed with 10% 2-propanol in
water (88 L), and dried (65 °C, 50 Torr). This afforded
crystalline 3-(4-fluorobenzylidene)-2-piperidone (6, 19.5 kg,
54% yield from valerolactam). Mp 170 °C. 19F NMR (376
MHz, CDCl3) δ 112.99. 1H NMR (400 MHz, CDCl3) δ 1.88
(m, 2H), 2.80 (m, 2H), 3.45 (m, 2H), 7.08 (m, 3H), 7.40
(m, 2H), 7.77 (s, 1H). 13C NMR (400 MHz, CDCl3) δ 23.2,
26.4, 42.2, 115.6 (d, J ) 21 Hz, C-3′), 129.5(C-3), 131.8
(d, J ) 8 Hz, C-2′), 132.1 (d, J ) 4 Hz, C-1′), 134.7
(olefinic), 162.5 (d, J ) 249 Hz, C-4′), 167.0 (C-2). Anal.
Calcd. for C12H12FNO: C, 70.23; H, 5.89; N, 6.82. Found:
C, 70.07; H, 5.86; N, 6.79.
kg, 10 wt % in THF, 133 mol) was added to the reaction
solution at such a rate that the temperature of the reaction
solution did not exceed 20 °C. Upon complete addition of
the LAH solution, the batch was heated to 30 °C and stirred
for 2 h. The batch was sampled and analyzed by HPLC and
shown to contain less than 1.0 area % of 7. To a 300-gal
glass-lined reactor was added Rochelle salt (sodium potas-
sium tartrate tetrahydrate, 112 kg, 397 mol) and water (358
L). The reaction mixture was quenched by slow addition to
the Rochelle’s salt solution. The transfer line was rinsed with
THF (25 kg). The quenched mixture was stirred for 15 min,
and the aqueous phase, containing lithium and aluminum
salts, was transferred to the 200-gal reactor for back
extraction. Ethyl acetate (189 kg) was added to the aqueous
layer, the mixture was agitated at room temperature and
allowed to settle. The aqueous layer was split away for
disposal. The ethyl acetate layer was transferred to the 300-
gal reactor and mixed with the THF layer. Additional water
separated and was removed for disposal. Distillation was used
to replace the THF with ethyl acetate. In the first batch, the
ethyl acetate/THF mixture was distilled at atmospheric
pressure until ca. 170 L of distillate was collected. Ethyl
acetate (146 kg) was added to the reactor and distilled again
until ca. 170 L of distillate was collected. A sample of the
bottoms showed 40% residual THF. Ethyl acetate (50 kg)
was charged to the reactor, and the distillation was repeated.
Following the distillation, the ethyl acetate solution was
washed twice with 10% brine (100 kg each wash). The
phases split readily. The solution of racemic 3-(4-fluoroben-
zyl)piperidine (2) in ethyl acetate was drummed, weighed,
and sampled. The solution yield of racemic 3-(4-fluoroben-
zyl)-piperidine (2) was 22.5 kg (88%) by HPLC assay.
The solution of racemic piperidine 2 (22.5 kg, 117 mol
contained) in ethyl acetate was charged to a 200-gal glass-
lined reactor. Atmospheric distillation was used to replace
the ethyl acetate with acetonitrile as the solvent. Residual
ethyl acetate in acetonitrile was shown to be less than 5 v/v%.
(R)-(-)-Mandelic acid (12.4 kg, 81.6 mol, 0.7 equiv based
on racemic piperidine 2 solution yield) and acetonitrile (137
kg) were charged to a 300-gal glass-lined reactor. The
racemic piperidine 2 solution and the mandelic acid solution
were separately heated to 78 °C. The racemic piperidine 2
solution was transferred to the mandelic acid solution,
keeping the temperature above 75 °C. Acetonitrile (20 kg)
was used to rinse the transfer line. The resulting mixture
was aged for 1 h at 78-80 °C, cooled to 65 °C over 3 h,
cooled to 50 °C over 3 h, and held at 50 °C for 2 h. The
slurry was sampled and filtered, and the wet cake was
analyzed for ee. The ee of the solid from the slurry was 84%.
The batch was filtered at 50 °C. The wet cake was washed
with acetonitrile (85 kg) at 50 °C and dried with nitrogen to
remove as much liquor as possible. The wet cake was
sampled and transferred to a 300-gal glass-lined reactor. The
ee of the wet cake was 91%. To the wet cake in the 300-gal
reactor was added acetonitrile (15 L per kg of contained
piperidine 2). The slurry was heated to 80 °C and held for
4 h with agitation. The solids did not all dissolve. The batch
was then cooled to 50 °C over 3 h. The slurry was filtered
Racemic 3-(4-Fluorobenzyl)-2-piperidone (7). A 50-gal,
stainless steel reactor was charged with 3-(4-fluoroben-
zylidene)-2-piperidone (6, 30.0 kg, 146 mol), 10% Pd on
carbon (3.12 kg), and methanol (118 kg). The reactor was
purged with nitrogen to a pressure of 30 psig and vented,
with the purge/vent cycle being carried out four times in
total. The reactor was then pressurized with hydrogen to 55
psig and 25 °C. After approximately 6 h reaction time, the
batch was sampled, analyzed by HPLC, and shown to contain
<1 area % of 6. The batch was filtered through a bag filter
to remove the catalyst. The filtrate was concentrated to a
volume of about 70 L by distillation under atmospheric
pressure at 65-66 °C. Toluene (104 kg) was charged to the
reactor and distilled under atmospheric pressure to remove
approximately 145 L of distillate. Toluene (77.8 kg) was
again charged and concentrated to collect approximately 80
L of distillate. The concentrate in the reactor was sampled
and shown to contain less than 2 vol % of methanol in
toluene. Heptane (152 kg) was charged to the toluene solution
at about 80 °C. The batch was cooled to 0 °C over 2 h and
then held at 0 °C for 1 h. The slurry was filtered in a 30-in.
diameter filter. The cake was washed with a mixture of
heptane (51.5 kg) and toluene (7.2 kg) at 0 °C. The wet cake
was dried in a tray dryer at 50 °C and 50 Torr to a constant
weight of 27.5 kg, and the yield of 7 was 90%. Mp 127 °C.
1
19F NMR (376 MHz, CDCl3) δ -117.63. H NMR (400
MHz, CDCl3) δ 1.43 (m, 1H), 1.57-1.89 (m, 3H), 2.50 (m,
1H), 2.71 (m, 1H), 3.29 (m, 3H), 6.01 (br s, 1H), 6.98 (m,
2H), 7.16 (m, 2H). 13C NMR (100 MHz, CDCl3) δ 20.2,
24.3, 35.6, 41.3, 41.8, 114.1 (d, J ) 21 Hz), 129.6, 134.4,
160.4 (d, J ) 241 Hz), 173.5.
(S)-3-(4-Fluorobenzyl)piperidine (2) Mandelate. To a
200-gal glass-lined reactor were charged racemic 3-(4-
fluorobenzyl)-2-piperidone (7, 27.5 kg, 133 mol) and THF
(254 kg). The resulting solution was cooled to 15-20 °C
with stirring. A solution of lithium aluminum hydride (50.3
Vol. 10, No. 2, 2006 / Organic Process Research & Development
•
267

in a Nutsche at 50 °C and washed with acetonitrile (45 kg)
at 50 °C. The product was dried at 50 °C and 50 Torr in a
tray dryer. The yield of 2-mandelic acid salt was 12.8 kg
(32%), and the ee was 97%. Chiral HPLC method, Chiralpak
AS (250 mm × 4.6 mm, 10-µm particle size); 1.0 mL/min,
30 °C; 265 nm; mobile phase, 95% (v/v) heptane, 4.5% (v/
v) ethanol, 0.25% (v/v) trifluoroacetic acid, 0.25% (v/v)
n-hexylamine. The retention times for the desired (S)-2
enantiomer and the undesired (R)-2 enantiomer are 10.1 and
12.8 min, respectively. Mp 172 °C. 19F NMR (376 MHz,
solution yield). Chiral HPLC analysis (as described above)
indicated that the ee was 90%.
A solution of (S)-3-(4-fluorobenzyl)piperidine (2, 12.2 kg
contained, estimated by HPLC assay, 63.1 mol at 90% ee)
in 120 L of toluene was heated to 70 °C. To this hot solution
was added a hot solution (at 70 °C) of (R)-(-)-mandelic acid
(10.1 kg, 66.4 mol) in acetonitrile (50 L). The resulting
mixture was stirred for 1 h at 70 °C, cooled to 25 °C over
2 h, and then held for 2 h. The slurry was filtered, washed
with acetonitrile, and dried in a vacuum-dryer. The yield of
2-mandelate was 21.1 kg (97%). Chiral HPLC analysis
indicated that the ee was 99%.
2-(Phenoxycarbonyl)amino-4-methyl-5-acetylthiazole
(4). The two solids, 2-amino-4-methyl-5-acetylthiazole (24,
10.5 kg, 67.2 mol) and diphenyl carbonate (14.5 kg, 67.7
mol) were each charged to a 100-gal glass-lined reactor,
followed by pump charging of THF (178 kg). The mixture
was heated to 30 °C and mixed for an additional 15 min.
Sodium hydride (60% dispersion in mineral oil, 3.5 kg, 87.5
mol) was added in 15 portions through a charge lock under
argon. The liberated hydrogen gas from the reaction was
vented to a flame-arrestor. After 2 h aging at 30 °C, the
reaction mass was sampled for reaction completion. A
solution of phenol (1.3 kg, 13.8 mol) in THF (5 kg) was
then pressure transferred from a pressure cylinder to the 100-
gal reactor. After aging for another 30 min, the reaction mass
was cooled to 15 °C and filtered through a bag filter and a
cartridge filter in series to remove residue particles, followed
by concentrating to 25% of initial volume in a glass-lined
reactor at 125 Torr pressure and 25 °C.
To a clean 100-gal glass-lined reactor was added purified
water (240 L). The concentrate was transferred into the 100-
gal reactor, while maintaining the pot temperature at 25 °C.
Solids started to form in the reactor. An aqueous solution of
6 N HCl was then added to the slurry, followed by a 1 N
HCl solution to fine-tune the pH of the aqueous solution to
between 4 and 7. The resulting slurry was held at 25 °C for
1 h and filtered on a 36-in. Nutsche installed with a filter
cloth. The wet cake was washed with 120 L of purified water
and dried in a pan-dryer for 12 h at 40 °C under 50 Torr.
The crude product was transferred back to the 100-gal
glass-lined reactor for reslurrying by charging a mixture of
140 kg of n-heptane and 10 kg of THF (5 vol % THF in
n-heptane). After aging at 25 °C for 1 h at 60 rpm, the slurry
was sampled for solid purity and then filtered on the 36-in.
Nutsche installed with a new filter cloth. The wet cake was
washed with a solution of 5% THF in n-heptane and dried
in a pan-dryer for 12 h at 40 °C under 50 Torr. The yield
was 13.5 kg (78%). Mp 189 °C. 1H NMR (400 MHz, CDCl3)
δ 2.49 (s, 3H), 2.75 (s, 3H), 7.20 (m, 2H), 7.29 (m, 1H),
7.41 (m, 2H). 13C NMR (100 MHz, CDCl3) δ 18.0, 30.7,
121.4, 125.7, 126.8, 129.9, 150.0, 152.3, 155.5, 162.7, 190.6.
HRMS Calcd for C13H12N2O3S: 277.0647 [M + 1]; found:
277.0652 [M + 1]. Anal. Calcd for C13H12N2O3S: C, 56.51;
H, 4.38; N, 10.14; found: C, 56.45; H, 4.24; N, 10.06.
1
DMSO-d6) δ -117.50. H NMR (400 MHz, DMSO-d6) δ
1.08 (m, 1H), 1.48 (m, 1H), 1.62 (m, 2H), 1.84 (m, 1H),
2.45 (m, 5H), 2.62 (m, 1H), 2.96 (d, J ) 13 Hz, 1H), 3.08
(d, J ) 13 Hz, 1H), 7.17 (m, 7H), 7.37 (m, 2H).13C NMR
(100 MHz, DMSO-d6) δ 22.1, 28.5, 35.2, 38.6, 43.4 47.8,
73.9, 115.2 (d, J ) 21 Hz, C-3′), 126.5, 126.6, 127.8, 131.0
(d, J ) 8 Hz, C-2′), 135.4 (d, J ) 3 Hz, C-1′), 143.9, 161.1
(d, J ) 240 Hz, C-4′), 175.9. Anal. Calcd for C20H24FNO3:
C, 69.55; H, 7.00; N, 4.06; found: C, 69.53; H, 7.00; N,
3.99.
(S)-3-(4-Fluorobenzyl)-2-piperidone (7) by Asymmetric
Hydrogenation. A 100-gal (400-L) glass-lined reactor was
charged with recrystallized 3-(4-fluorobenzylidene)-2-pip-
eridone (6, 19.5 kg, 95.0 mol), methanol (75 kg), and
dichloromethane (126 kg). (2S,4S)-(-)-2,4-Bis(diphenylphos-
phino)pentane (97 g, 0.25 mol %) and bis(1,5-cyclooctadi-
ene)iridium(I) tetrafluoroborate (109 g, 0.25 mol %) was
added to the stirred solution under nitrogen. The reactor was
pressured to 55 psig hydrogen and stirred for 13 h at which
time no starting material remained. The solvent was replaced
with toluene by successive vacuum distillation and toluene
charges, and the resultant solution was filtered through a bed
of charcoal (Darco G-60) and silica gel 60. The resultant
solution contained (S)-7 (19.1 kg, 97% solution yield, 88%
ee). Chiral HPLC analysis, Chiralpak AS (250 mm × 46
mm; 10-µm particle size); 1.0 mL/min, 40 °C; 210 nm;
mobile phase, 60% (v/v) heptane, 39.2% (v/v) 2-propanol,
0.4% (v/v) trifluoroacetic acid, 0.4% (v/v) n-hexylamine;
retention times of the desired (S)-7 enantiomer and the
undesired (R)-7 enantiomer are 15.6 and 10.6 min, respec-
1
tively. 19F NMR (376 MHz, CDCl3) δ -117.63. H NMR
(400 MHz, CDCl3) δ 1.43 (m, 1H), 1.57-1.89 (m, 3H), 2.50
(m, 1H), 2.71 (m, 1H), 3.29 (m, 3H), 6.01 (br s, 1H), 6.98
(m, 2H), 7.16 (m, 2H). 13C NMR (100 MHz, CDCl3) δ 20.2,
24.3, 35.6, 41.3, 41.8, 114.1 (d, J ) 21 Hz), 129.6, 134.4,
160.4 (d, J ) 241 Hz), 173.5.
(S)-3-(4-Fluorobenzyl)piperidine (2) Mandelate. A so-
lution of (S)-3-(4-fluorobenzyl)-2-piperidone (7, 6.4 kg, 30.8
mol) in 120 L of toluene was cooled to 10 °C. Lithium
aluminum hydride bis(tetrahydrofuran) solution in toluene
(1.0 M, 27.8 kg, 31.8 L, 31.8 mol) was added at a
temperature below 15 °C. The batch was heated to 40 °C
and stirred for 3 h. The batch was quenched by slow addition
to a Rochelle salt solution (16.4 kg in 90 L of water) and
stirred for 15 min. The organic phase was washed with 10%
brine followed by water. The toluene solution was concen-
trated by vacuum distillation to a volume of about 60 L.
The resultant solution contained (S)-2 (5.0 kg, 25.9 mol, 84%
268
•
Vol. 10, No. 2, 2006 / Organic Process Research & Development

(1R,2R)-Cyclohexane-1,2-carboxylic Acid Monoethyl
Ester (R)-(+)-r-Methylbenzylamine Salt (17) via cis-
Hemi-ester (15) and trans-Hemi-ester (16). To a 50-gal
glass-lined reactor was charged toluene (32.6 kg), followed
by anhydrous (water content < 400 ppm) ethanol (12.6 kg,
274 mol). The mixture was cooled to and maintained at -20
°C. To a 300-gal glass-lined reactor was charged (+)-
quinidine (89.3 kg, 275 mol), followed by toluene (242.5
kg) and meso-hexahydrophthalic anhydride (14, premelted
at 50 °C, 35.0 kg, 227 mol). The resulting slurry was cooled
to -20 °C. The ethanol/toluene mixture from the 50-gal
reactor was transferred to that slurry at a rate (approximately
0.2 L/min) to maintain batch temperature in the 300-gal
reactor below -15 °C. The transfer time was about 2.5 h.
After an additional 2 h aging at -15 °C, HPLC analysis
showed 99% conversion. To quench the reaction, the batch
was warmed to room temperature and charged with 230 kg
of 3 M sulfuric acid solution. After extraction and phase
separation, the organic layer was washed twice with 100 L
of purified water and then distilled at 35-40 °C and 35-50
Torr to approximately 178 L. GC analysis and Karl Fischer
titration showed completion of distillation: less than 0.5 v/v
% ethanol and less than 400 ppm water content in the cis-
hemi-ester 15 toluene solution. The batch was cooled to -20
°C and used directly in the next step. The ee of the cis-
hemi-ester 15 was determined by chiral HPLC to be 87%.
Chiral HPLC method: Chiralcel OJ-R (150 mm × 4.6 mm;
5-µm particle size); 0.5 mL/min; 20 °C; 210 nm; mobile
phase: 70% 0.1% phosphoric acid in water-30% acetoni-
trile. Retention times of the desired (1R,2S)-15 enantiomer
and undesired (1S,2R)-15 enantiomer were 8.4 and 9.9 min,
1.20 (t, J ) 7 Hz, 3H), 1.35 (m, 4H), 1.81 (m, 2H), 2.11 (m,
2H), 2.61 (m, 2H), 4.14 (q, J ) 7 Hz, 2H). 13C NMR (100
MHz, CDCl3) δ 14.2, 25.3, 25.4, 28.96, 29.04, 44.78, 44.83,
60.7, 175.1, 181.5.
A 300-gal glass-lined reactor containing trans-hemi-ester
16 toluene concentrate was charged with MTBE (470.0 kg).
The batch was heated to 40 °C. (R)-(+)-(R)-Methylbenzyl-
amine (25.6 kg, 211 mol) was added. After seeding at 40
°C with 420 g of the salt 17 suspended in n-heptane (3 L),
the batch was cooled to 33-34 °C at 0.2 °C/min ramping
and was aged for 30 min at this temperature. The batch was
then heated to 42-43 °C and held for 2 h. It was again cooled
to 20 °C at a rate of 0.2 °C/min and held for 12 h. After the
holding, n-heptane (100 kg) was charged at 20 °C. The batch
was cooled to -15 to -20 °C at the same cooling rate. The
batch was harvested on a centrifuge. About 50% of recovered
mother liquors was recycled to the vessel to thin out the
slurry. The wet cake was washed with 150 kg of cold
n-heptane (-10 °C) and dried on a tray dryer at 50 °C under
50 Torr. The yield of 17 was 19.5 kg (65% from anhydride
14). The ee was determined by chiral HPLC to be higher
than 99%. Chiral HPLC method: the same method as
described for the cis-hemi-ester 15. The retention times for
the desired (1R,2R)-17 enantiomer and the undesired (1S,2S)-
17 enantiomer were 8.5 and 7.3 min, respectively. Mp Form
I 99 °C (MTBE, n-heptane), mp Form II 107 °C (MTBE,
1
n-heptane). H NMR (400 MHz, CDCl3) δ 0.90-1.28 (m,
7H), 1.50 (d, J ) 7 Hz, 3H), 1.57 (m, 1H), 1.70 (m, 2H),
1.89 (m, 1H), 2.14 (m, 1H), 2.34 (m, 1H), 3.96 (m, 1H),
4.02 (m, 1H), 4.22 (q, J ) 7 Hz), 1H), 7.27 (m, 1H), 7.32
(m, 2H), 7.42 (m, 2H), 8.18 (br, 3H).13C NMR (100 MHz,
CDCl3) δ 14.4, 22.4, 25.7, 25.9, 29.3, 29.5, 46.0, 47.0, 51.1,
60.0, 126.7, 128.0, 128.9, 141.6, 176.6, 181.3. HRMS Calcd
for the acid 16, C10H16O4: 199.0970 [M - 1]; found:
199.0973 [M - 1]. Anal. Calcd for the salt 17, C18H27NO4:
C, 67.26; H, 8.47; N, 4.36; found: C, 67.34; H, 8.61; N,
4.38.
Ethyl(1R, 2R)-2-(Benzyloxycarbonylamino)cyclohexane-
carboxylate (18). CAUTION! Although diphenyl phospho-
ryl azide (DPPA) is widely utilized in multikilogram
quantities,11 prudent safety practices for its use are clearly
indicated. DPPA itself is thermally stable and is often purifed
by vacuum distillation (137 °C, 0.2 Torr).12 However,
adventitious moisture might hydrolyze the reagent, forming
volatile (bp 37 °C) anhydrous hydrazoic acid. At least in
principle, HN3 could collect in a cool region of the reactor
and present an explosion hazard. A scrutiny of the literature
did not reveal any incidents based on this eventuality.
However, several sensible measures were taken to circumvent
any potential problems. (1) Karl Fischer titration was used
to monitor water levels, and these were kept below 0.1%.
(Others have recommended still lower water levels.13) (2)
Excess triethylamine was employed. Any traces of strongly
acidic hydrazoic acid should be retained in solution as the
triethylammonium salt. (3) Cooling to the reactor condensor
1
respectively. H NMR (400 MHz, CDCl3) δ 1.25 (t, J ) 7
Hz, 3H), 1.38-1.62 (m, 4H), 1.78 (m, 2H), 2.02 (m, 2H),
2.83 (m, 2H), 4.12 (q, J ) 7 Hz, 2H).
To a 100-gal glass-lined reactor was charged 146.1 kg of
a toluene solution containing 25 wt % potassium tert-amylate
(contained 36.5 kg, 289 mol), followed by toluene (392.0
kg). The mixture was cooled to -20 °C. The toluene solution
of cis-hemi-ester 15 was added at -10 to -15 °C, at a rate
so that the temperature of reaction did not exceed -10 °C.
The addition took approximately 3 h. After an additional 1
h aging, the reaction reached 96% conversion to the trans-
hemi-ester 16 based on HPLC analysis. To quench the
reaction, glacial acetic acid (45.2 kg) was added over 1 h
while keeping the batch temperature below -5 °C. Hydro-
chloric acid solution (3 N, 144.0 kg) was charged to the
vessel, followed by 33.0 L of purified water, while maintain-
ing the pot temperature below 10 °C. After extraction and
phase separation, the organic layer was washed once with
180 L of purified water and then distilled at 35-45 °C under
35-50 Torr until approximately 60 L was left in the reactor.
An additional amount of toluene (50 kg each) was charged
twice, and each time the batch was distilled to approximately
60 L. The distillation reached end-point when the acetic acid
and amyl alcohol contents in the pot were below 4.0 and
0.8 v/v %, respectively, by GC analysis. The toluene solution
of trans-hemi-ester 16 was used directly in the next step
without further purification. 1H NMR (400 MHz, CDCl3) δ
(12) Shi, E.; Pei, C. Synth. Commun. 2005, 35, 669-673.
(13) Thompson, A. S.; Hartner, F. W., Jr.; Grabowski, E. J. J. Org. Synth. 1998,
75, 31-36.
Vol. 10, No. 2, 2006 / Organic Process Research & Development
•
269

was not turned on to maximize the likelihood of HN3 vapor
being flushed from the system.
kg, 57.3 mol) was dissolved in toluene (167 kg) and
methylcyclohexane (27.0 kg). The mixture was cooled to
-90 to -96 °C. A 20 wt % solution of diisobutylaluminum
hydride (DIBAL-H) (90.1 kg, 18.0 kg contained 127 mol)
in hexanes was added at a rate that kept the temperature
below -90 °C. After the addition the mixture was held for
an additional 30 min, and a solution of 2-propanol (13.8 kg)
and n-heptane (23.0 kg) was added to quench the excess
DIBAL-H. The mixture was then transferred to a 1000-L
reactor containing a solution of citric acid (54.3 kg, 258 mol)
and water (112 L), keeping the temperature below 30 °C.
The phases were cut, and the organic was washed with 5%
aqueous sodium bicarbonate (147 kg) and 10% brine (150
kg). 2-Propanol (27.5 kg) was added to the organic solution.
To the organic solution was added sodium bisulfite (11.9
kg, 62.6 mol) in water (24.5 L), and solids separated. The
aldehyde 3-bisulfite adduct was isolated in a centrifuge. Mp
To a 100-gal vessel were charged (1R,2R)-cyclohexane-
1,2-carboxylic acid monoethyl ester, (R)-(+)-R-methylben-
zyl-amine salt (17, 55.0 kg, 172 mol), water (77.0 L), and
toluene (119 kg). To the above suspension was added a 37%
aqueous solution of HCl (22.0 kg). The phases were cut,
and the toluene solution was distilled to a volume of about
120 L to remove water to under 1000 ppm. The solution
was cooled to 20 °C, and triethylamine (26.0 kg, 257 mol)
was added. The mixture was heated to 85 °C. A solution of
diphenylphosphoryl azide (DPPA, 45.4 kg, 165 mol) in
toluene (44.0 kg) was added to the mixture at a rate to
maintain the temperature at 83-87 °C. The reaction was
stirred for 30 min after the addition was complete and after
a sample showed the reaction was complete (>98% conver-
sion by HPLC, analysed as N-benzyl urea derivative of the
isocyanate). To the above mixture was added benzyl alcohol
(17.7 kg, 164 mol), and the mixture was heated to reflux at
110 °C for 5 h. When the reaction was complete, it was
cooled to 20-25 °C and washed with water (50 L) and then
twice with 10% brine (50 L). The toluene solution was
solvent switched first to 2-propanol and then to n-heptane
by vacuum distillation (to an end-point of less than 2% IPA
in the n-heptane in the pot). The mixture was cooled to 50
°C, seeded with 200 g of 18 suspended in n-heptane (300
mL), and then cooled to 20 °C at 0.2 °C/min. After 8 h at
this temperature the mixture was cooled further to -5 °C
and filtered on a 36-in. Nutsche. The cake was washed with
n-heptane and dried in a vacuum oven. The yield of 18 was
44.7 kg (80%). The crude product contained 0.5 area % of
the symmetrical urea impurity 19, and had a purity of 96.6
area % and 96.5 wt %.
1
141 °C, H NMR (400 MHz, DMSO-d6) δ 0.95-1.25 (m,
4H), 1.59 (t, J ) 12 Hz, 2H), 1.84 (m, 2H), 2.19 (d, J ) 13
Hz, 1H), 3.28 (m, 1H), 4.17 (d, J ) 5 Hz, 1H), 4.90 (d, J )
6 Hz, 1H), 4.99 (q, J ) 13 Hz, 2H), 7.10 (d, J ) 9 Hz, 1H),
7.35 (m, 5H).13C NMR (100 MHz, DMSO-d6) δ 23.9, 25.4,
25.5, 34.0, 43.5, 51.0, 65.4, 81.9, 127.97, 128.02, 128.7,
137.7, 156.1. The wet cake of the aldehyde 3-bisulfite
adduct was charged back to the 1000-L reactor for acid
treatment without further purification. Water (140 L), MTBE
(90.9 kg), and 50% aqueous potassium carbonate solution
(40.8 kg) were charged in that order to the 1000-L reactor.
The mixture was stirred until the solids dissolved, and then
the aqueous phase was removed. The organic phase was
washed twice with 15% aqueous brine (180 kg) The organic
phase was solvent switched to n-heptane (final volume about
88 L) by vacuum distillation. The product aldehyde 3
crystallized out and was filtered. The solid was dried in a
vacuum oven at 30 °C, and the yield of 3 was 12.75 kg
The dried crude product 18 (37.0 kg) was charged to a
200-gal glass-lined reactor, followed by IPA (87.1 kg). The
solid dissolved after mixing at 40 °C for 15 min. After an
additional 15-min hold, the batch was cooled to 22 °C
without ramping. Crystals formed in the batch without
seeding. Water (167 L) was charged to the reactor over a 2
h period (about 1.4 kg/min). After a 30-min hold, the solid
was filtered in two drops on a 36-in. diameter, glass-lined
Nutsche and washed with a mixture of IPA (62.0 kg) and
water (130 kg). The washed cake was dried in a tray dryer.
The level of symmetrical urea impurity 19 was lowered from
0.5 area % to 0.1 area %. The yield of pure 18 was 32.0 kg
1
(85%). Mp 67 °C. H NMR (400 MHz, CDCl3) δ 1.14-
1.53 (m, 4H), 1.68-1.83 (m, 3H), 2.00 (m, 1H), 2.15 (m,
1H), 3.82 (m, 1H), 4.98-5.13 (m, 3H), 7.26-7.38 (m, 5H),
9.58 (d, J ) 4 Hz, 1H).13C NMR (100 MHz, CDCl3) δ 24.0,
24.7, 25.6, 32.5, 49.6, 56.6, 67.0, 128.26, 128.34, 128.7,
136.5, 155.9, 203.7. HRMS Calcd for C15H19NO3: 262.1443
[M + 1]; found 262.1436 [M + 1]. Anal. Calcd for: C15H19-
NO3: C, 68.94; H, 7.33; N, 5.36; found: C, 68.92; H, 7.39;
N, 5.23.
N-(5-Acetyl-4-methyl-1,3-thiazol-2-yl)-N′-((1R,2S)-2-
{[3(S)-(4-fluorobenzyl)piperidinyl]methyl}cyclohexyl)-
urea (1). In a 100-gal glass reactor were warmed sodium
triacetoxyborohydride (18.5 kg, 87.3 mol) and DMSO (71
kg) to 40 °C until homogeneous. The batch was cooled to
12 °C, and (S)-3-(4-fluorobenzyl)piperidine 2-(R)-mandelate
salt (20.2 kg, 58.5 mol) was added. In a separate reactor,
(1R,2R)-2-(benzyloxycarbonylamino)cyclohexanecarboxal-
dehyde (3, 15.3 kg, 58.5 mol) and DMSO (33 kg) were
mixed to give a homogeneous solution. The aldehyde 3
solution was added over 2 h to the piperidine 2 solution while
keeping the temperature at 10-14 °C. After the addition the
reaction was stirred for an additional hour and then quenched
with 6 N aqueous HCl solution (8.4 kg) at 20-24 °C. To
1
(87%). Mp 81 °C. H NMR (400 MHz, CDCl3) δ 1.12-
1.28 (m, 2H), 1.18 (t, J ) 7 Hz, 3H), 1.39 (m, 1H), 1.56 (m,
1H), 1.72 (m, 1H), 1.93 (m, 1H), 2.08 (m, 1H), 2.23 (m,
1H), 3.71 (m, 1H), 4.08 (q, J ) 7 Hz, 2H), 4.84 (br, 1H),
5.06 (s, 2H), 7.25-7.38 (m, 5H).13C NMR (100.6 MHz,
CDCl3) δ 14.3, 24.6, 24.8, 28.8, 33.2, 50.1, 51.9, 60.7, 66.7,
128.2, 128.7, 136.8, 155.6, 174.1. HRMS Calcd for C17H23-
NO4: 306.1705 [M + 1]; found: 306.1704 [M + 1]. Anal.
Calcd For C17H23NO4: C, 66.86; H, 7.59; N, 4.59; found:
C, 66.89; H, 7.68; N, 4.53.
(1R,2R)-2-(Benzyloxycarbonylamino)cyclohexanecar-
boxaldehyde (3). In a 600-L reactor, ethyl (1R,2R)-2-
(benzoloxycarbonylamino)cyclohexanecarboxylate (18, 17.5
270
•
Vol. 10, No. 2, 2006 / Organic Process Research & Development

the quenched reaction mass were sequentially added iPrOAc
(59 kg), water (67 kg), and 15% aqueous sodium hydroxide
solution (50 kg). The mandelic acid was removed in the
aqueous solution. The product was extracted into the iPrOAc,
which was washed with 10% brine (40 kg) and water (45
kg). This iPrOAc solution containing (1R,2S)-1-(benzyloxy-
carbonyl)amino-2-{[3(S)-(4-fluorobenzyl)piperidinyl]methyl}-
cyclohexane (22) was carried forward into the next step
without further purification.
(d, J ) 7 Hz, 2H), 2.49 (3H, s), 2.56 (d, J ) 9 Hz, 1H),
2.65 (3H, s), 2.91 (d, J ) 11 Hz, 1H), 3.29 (dt, J ) 4, 11
Hz, 1H), 6.79 (t, J ) 9 Hz, 2H), 6.96-7.03 (m, 2H). 13C
NMR (100 MHz, CDCl3) δ 18.1, 24.5, 25.1, 25.7, 30.4, 30.5,
30.7, 33.3, 38.6, 38.8, 40.0, 54.0, 57.7, 60.8, 64.8, 115.0 (d,
J ) 21 Hz), 123.552, 130.0 (d, J ) 8 Hz), 135.3 (d, J ) 3
Hz), 154.318,161.3 (d, J ) 242 Hz) 162.578, 190.488.
HRMS Calcd for C26H35FN4O2S: 487.2543 [M - 1];
found: 487.2533 [M - 1]. Anal. Calcd for C26H35FN4O2S:
C, 64.17; H, 7.25; N, 11.51; found: C, 64.21; H, 7.28; N,
11.53.
To a 100-gal glass reactor was charged palladium on
carbon (5% Pd, 50% wet, 6.3 kg) followed by the iPrOAc
solution of 22 and iPrOAc (120 kg). The reactor was inerted
and then pressurized with 55 psig hydrogen and stirred for
4 h at 25 °C. When the reaction was complete, the catalyst
was filtered through a 1-µm bag filter followed by 0.5-µm
and 0.2-µm cartridge filters. The resulting solution was
distilled to a volume of about 100 L and a water content of
less than 500 ppm. The penultimate, (1R,2S)-1-amino-2-
{[3(S)-(4-fluorobenzyl)piperidinyl]methyl}cyclohexane (23),
was used in the next step without further purification. 19F
N-(5-Acetyl-4-methyl-1,3-thiazol-2-yl)-N′-((1R,2S)-2-
{[3-(S)-(4-fluorobenzyl)-piperidinyl]methyl}cyclohexyl)-
urea, Fumarate Salt (1-Fumarate). In a 50-gal glass reactor
the API free base 1 (11.0 kg, 22.6 mol) was charged followed
by 1-methoxy-2-propanol (147.1 kg). The mixture was heated
to 70 °C, stirred for 30 min, and filtered through a 0.45-µm
filter into a 100-gal reactor. To the 50-gal reactor were
charged fumaric acid (3.1 kg, 26.7 mol) and 1-methoxy-2-
propanol (45.8 kg). The mixture was heated to 60 °C, and
the first 20% was added to the API free base 1 solution.
The batch was seeded with 1 wt % of seeds suspended in
1-methoxy-2-propanol. Then the remainder (80%) of the
fumaric acid solution was added to the batch. The batch was
cooled to 20 °C over 6 h and was held at that temperature
for 2 h. The solids were filtered and dried in a vacuum oven
at 50 °C and under 50 Torr. The yield of the dried product
1-fumarate was 11.7 kg (86%). Mp 209 °C. 19F NMR (376
MHz, DMSO-d6), δ -117.91. 1H NMR (400 MHz, DMSO-
d6) δ 0.95-1.03 (m, 3H), 1.21 (m, 3H), 1.43-1.75 (m, 6H),
1.87 (m, 3H), 2.15 (t, J ) 12 Hz, 1H), 2.34 (m, 1H), 2.42
(s, 3H), 2.51 (s, 3H), 2.53 (m, 2H), 2.74 (m, 2H), 3.10 (br,
2H), 3.24 (br, 1H), 6.58 (s, 2H), 7.02 (m, 2H), 7.16 (m, 2H).
13C NMR (100 MHz, DMSO-d6) δ 18.5, 23.2, 24.9, 28.5,
30.2, 30.7, 33.3, 36.2, 38.5, 38.8, 52.2, 54.1, 56.4, 57.9, 61.3,
105.0 (d, J ) 21 Hz), 124.5, 130.9 (d, J ) 8 Hz), 135.2,
135.6, 154.0, 155.1, 161.1 (d, J ) 240 Hz), 162.5, 168.0,
190.4. Anal. Calcd for: C30H39FN4O6S, C, 59.78; H, 6.52;
N, 9.30; found: C, 59.72; H, 6.70; N, 9.10.
1
NMR (376 MHz, CDCl3) δ -118.22. H NMR (400 MHz,
CDCl3) δ 0.84 (m, 2H), 1.15 (m, 3H), 1.36 (m, 1H), 1.46
(m, 1H), 1.65-1.85 (m, 7H), 2.08 (dd, J ) 4, 12 Hz, 1H),
2.28 (m, 3H), 2.44 (m, 3H), 2.67 (d, J ) 10 Hz, 1H), 3.01
(d, J ) 11 Hz, 1H), 6.96 (t, J ) 9 Hz, 2H), 7.08 (m, 2H).13C
NMR (100 MHz, CDCl3) δ 25.3, 25.5, 26.2, 30.8, 31.2, 36.6,
38.9, 40.7, 41.2, 54.4, 57.3, 62.1, 67.0, 115.3 (d, J ) 21
Hz), 130.7 (d, J ) 8 Hz), 136.5 (d, J ) 3 Hz), 161.6 (d, J
) 242 Hz).
The solution containing the penultimate 23 (17.5 kg
contained by HPLC assay, 57.5 mol) from the previous step
was heated to 40 °C, and a solution of 2-(phenoxycarbonyl)-
amino-4-methyl-5-acetylthiazole (4, 16.2 kg, 58.6 mol) in
THF (105 kg) was added in portions while monitoring
reaction completion. When 95% of the thiazole 4 had been
added, less than 1.0 area % of 23 was present by HPLC,
and the reaction was considered complete. The reaction
solvent was switched to IPA through distillations to a volume
of about 114 L (with THF and iPrOAc level <1 v/v % in
IPA by GC analysis). Water (88 L) was added as an anti-
solvent, and the mixture cooled to 0 °C. The crude product
was filtered. It was dissolved in a mixture of IPA (95 kg)
and water (15 kg) at 80 °C. The solution was then cooled to
55 °C and held for 1 h and cooled further to 30 °C. Water
(120 kg) was added, and the mixture was cooled to 0 °C.
The solids were filtered and dried in a vacuum oven at 50
°C and under 50 Torr. The yield of the API free base 1 was
22.8 kg (80% from 2 and 3). Mp Form I 181 °C, Form II
Acknowledgment
We are grateful to Dr. Pat N. Confalone for helpful
discussions. Thanks are due to Dr. Percy H. Carter and Dr.
Richard H. Mueller for useful comments on the manuscript.
We thank Mr. Michael Peddicord and Dr. John Castoro for
assistance with mass spectrometry. Thanks are also due to
Mr. Thomas Smyser and Ms. Wei Wu for skilled technical
assistance.
1
199 °C.19F NMR (376 MHz, CDCl3) δ -117.58. H NMR
Received for review October 17, 2005.
OP050202L
(400 MHz, CDCl3) δ 0.80-1.48 (7H, m), 1.55-1.79 (8H,
m), 2.08 (dd, J ) 3, 13 Hz, 1H), 2.25-2.34 (2H, m), 2.38
Vol. 10, No. 2, 2006 / Organic Process Research & Development
•
271