Development of a scaleable route for the production of cis-N-benzyl-3-methylamino-4-methylpiperidine

Article abstract of DOI:10.1021/op025599x

The synthesis of cis-N-benzyl-3-methylamino-4-methylpiperidine (5) via hydroboration of tetrahydropyridine 3 followed by oxidation and reductive amination was optimized and scaled up to produce 10-kg quantities of product. Three routes to 3 were identified, and the reduction of pyridinium salt 7 was selected as the most preferable to run on-scale. The hydroboration and oxidative workup were carefully studied to optimize throughput on that transformation, as was the reductive amination.

Full text of DOI:10.1021/op025599x

Organic Process Research & Development 2003, 7, 115−120
Development of a Scaleable Route for the Production of
cis-N-Benzyl-3-methylamino-4-methylpiperidine
David H. Brown Ripin,* Stefan Abele,† Weiling Cai, Todd Blumenkopf,§ Jeffrey M. Casavant,§ Jonathan L. Doty,§
Mark Flanagan,§ Christian Koecher,‡ Klaus W. Laue,‡ Keith McCarthy, Cliff Meltz, Mike Munchhoff,§ Kees Pouwer,|
Bharat Shah, Jianmin Sun,§ John Teixeira, Ton Vries,| David A. Whipple,§ and Glenn Wilcox
Chemical Research and DeVelopment, Pfizer Global Research DiVision, Pfizer Inc., Eastern Point Road,
Groton, Connecticut 06340, U.S.A.
Scheme 1a
Abstract:
The synthesis of cis-N-benzyl-3-methylamino-4-methylpiperidine
(5) via hydroboration of tetrahydropyridine 3 followed by
oxidation and reductive amination was optimized and scaled
up to produce 10-kg quantities of product. Three routes to 3
were identified, and the reduction of pyridinium salt 7 was
selected as the most preferable to run on-scale. The hydrobo-
ration and oxidative workup were carefully studied to optimize
throughput on that transformation, as was the reductive
amination.
Initial Synthesis
a Reaction conditions: a) BnNH2, HCHO, H2O, 60%; b) MeLi; c) SOCl2; d)
KOH, 45%, three steps; e) BH3‚THF; f) H2O2, 45%, two steps; g) Jones
Oxidation; h) MeNH2, NaHB(OAc)3, 70%, two steps.
cis-N-Benzyl-3-methylamino-4-methylpiperidine (5) was
identified as a useful intermediate in the synthesis of a
clinical drug candidate, necessitating its production on
kilogram scale. The original route used to produce this
compound relied on a hydroboration-oxidation sequence
starting from tetrahydropyridine 31 followed by oxidation
and reductive amination to install the methylamino func-
tionality.
Two routes to the tetrahydropyridine were used to make
initial lots of material (Scheme 1). Addition of methyllithium
to piperidone 2 followed by tertiary chloride formation and
elimination provided olefin 3 in 45% yield.1 Direct elimina-
tion of the tertiary alcohol was not successful under a variety
of conditions and necessitated the intermediacy of the
chloride. Alternatively, an aqueous Diels-Alder reaction
produced the material directly from isoprene (1) and was
the favored route to intermediate 3;2 however, the reaction
required 7-10 days to run to completion, and the product
(an oil) was difficult to separate from the excess benzylamine
in the reaction.3 Lewis acid catalysts such as Nb(OTf)3 failed
to improve the speed or efficiency of this reaction.4
From a scalability perspective, the Diels-Alder reaction
was sluggish and difficult to work up, the oxidation step of
the hydroboration was messy and sluggish, the Jones
oxidation5 needed replacing with a chromium-free oxidation,
and the diastereoselectivity of the reductive amination was
not reproducible. Nonchromatographic purifications were
needed for some of the intermediates as well as a way to
upgrade the diastereomeric excess (de) of the final product.
As an alternative, an asymmetric synthesis was explored.
A route was investigated using R-methylbenzylamine in place
of benzylamine in the Diels-Alder reaction. The hydro-
boration of the corresponding olefin did not proceed in high
diastereoselectivity, but the diastereomers could be separated
and one was crystalline. Unfortunately, the C-4 methyl-
bearing stereocenter was easily racemized during the oxida-
tion and reductive amination, thus obviating any advantage
of this route over the achiral variant. Attempts at doing a
Mitsunobu displacement on the alcohol led to ring-contracted
product.6
Route Modification To Develop a Scaleable Process
As an alternative to the methods described above for
making tetrahydropyridine 3, pipecoline (6) was benzylated
* Corresponding author. Fax: (860) 441-3630. E-mail: david_b_ripin@
groton.pfizer.com.
(3) The unreacted benzylamine could be removed by treatment of the worked-
up reaction mixture with acetic anhydride, extraction of the desired product
into acidic water, and then neutralization of the aqueous phase to recover
the product.
(4) (a) Kobayashi, S.; Hachiya, I.; Araki, M.; Ishitani, H. Tetrahedron Lett.
1993, 34, 3755-3758. (b) Kobayashi, S.; Araki, M.; Ishitani, H.; Satoshi,
N.; Hachiya, I. Synlett. 1995, 233-234. (c) Kobayashi, S.; Hachiya, I.;
Takahori, T.; Araki, M.; Ishitani, H. Tetrahedron Lett. 1992, 33, 6815-
6818. (d) Yu, L.; Chen, D.; Wang, P. G. Tetrahedron Lett. 1996, 37, 2169-
2172.
† Corresponding author at CarboGen. Fax: 41 62 836 4810. E-mail:
stefanabele@carbogen.com.
§ Discovery Research, Pfizer Global Research Division, Pfizer Inc., Eastern
Point Road, Groton, Connecticut 06340, U.S.A.
‡ CarboGen Laboratories (Aarau) AG, Schachenallee 29, CH-5001 Aarau,
Switzerland.
| Syncom B.V., P.O. Box 2253, 9704 CE Groningen, The Netherlands.
Bioprocess Research and Development, Pfizer Global Research Division,
Pfizer Inc., Eastern Point Road, Groton, Connecticut 06340, U.S.A.
(1) Iorio, M. A.; Ciuffa, P.; Damia, G. Tetrahedron 1970, 26, 5519-5527.
(2) Larsen, S. D.; Grieco, P. A. J. Am. Chem. Soc. 1985, 107, 1768-1769.
(5) Bowers, A.; Halsall, T. G.; Jones, E. R. H.; Lemin, A. J. J. Chem. Soc.
1953, 2548-2560.
10.1021/op025599x CCC: $25.00 © 2003 American Chemical Society
Published on Web 12/02/2002
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
115

Scheme 2a
equivalent of BH3 cleanly complexes the amine and does
not react any further9 as evidenced by the vinylic hydrogen
in the NMR of the complex. In lots of starting olefin that
contained higher amounts of the HB(OH)2 complex from
the previous reaction, higher conversion was observed,
presumably due to the availability of more borane after the
initial complexation of the amine. The yield could be
improved either by using 2.2 equiv of BH3, or by first
forming a complex with 1.2 equiv of BF3 etherate followed
by hydroboration with 1.4 equiv of BH3 (Scheme 3). Both
procedures gave >85% yield, with the procedure utilizing
BF3 generating less hydrogen gas during the work-up as
compared to the reaction with excess borane. As some
diorganoborane intermediates were observed (vide infra), it
seems likely that the amount of borane used could be further
reduced.
To minimize the amounts of hydrogen liberated in a
quench in the presence of an oxidizing agent, an initial
nonoxidative quench of any B-H bonds remaining prior to
oxidation of the B-C bonds was investigated. This procedure
also effected decomplexation of the amine-borane complex.
A methanol/1 N HCl quench with calcium chloride added
(pH ) 2) was utilized. The calcium chloride was added to
trap any fluoride liberated from the quench of BF3 in the
case where BF3 was used to form the amine-borane
complex. It is interesting to note that this quench does not
destroy all of the B-H bonds in the mixture as evidenced
by hydrogen evolution during the final salt formation of the
product (vide infra). This was further supported by the MS
data of one of the intermediates observed after the hydrolysis
which appears to be of structure RB(H)OH (MW ) 217).
Heating the mixture did not push the B-H quench any
further, although it did convert the initially formed ester of
the boronic acid (R2BOMe) to the corresponding acid R2-
BOH (which had no apparent effect on the oxidation).
We have previously reported that oxidation of the
resulting alkylboranes under basic conditions led to sluggish
and messy reactions.10 This could be avoided by using an
Oxone oxidation of the alkylborane intermediate. As previ-
a Reaction conditions: a) BnCl, acetone, 55 °C, 73%; b) NaBH4, EtOH, 15
°C, 73%.
with benzyl chloride to produce the pyridinium salt, which
was then reduced with sodium borohydride7 (Scheme 2). The
generation of the benzylpyridinium salt was run in acetone
at 55°C with addition of benzylchoride to the heated solution
of pipecoline. Benzyl chloride was used as the limiting
reagent to minimize the amounts of the lachrymator present
in the isolated product, which was recovered in 73% yield
and >99% purity by filtration (1 crop, 87.7 kg).
The pyridinium salt (7) was cleanly reduced to the 3,4-
dehydropiperidine (3) using sodium borohydride in ethanol.
Originally, the sodium borohydride was charged to the
ethanol as pellets, followed by addition of the pyridinium
salt. The reduction required 2.5-3.5 equiv of sodium
borohydride to go to completion, but the amount of boro-
hydride used could be reduced by more than 2-fold (to 1.25
equiv) by adding the solid to a premixed solution of the salt.
The reaction was run in ethanol over 7 h (40-kg scale) while
maintaining the internal reaction temperature between 10 and
15 °C. The lower amount of borohydride also reduced the
amount of hydrogen generated in the workup of the reaction,
which was quenched with water. After removing most of
the ethanol from the solution in vacuo, the pH of the solution
was adjusted to ∼pH 8 with sodium hydroxide, and the
product was extracted with methyl tert-butyl ether (MTBE)
to provide 73% yield of the desired product in 91% purity
(HPLC) along with 3.6% of the HB(OH)2 complex (molec-
ular weight (MW) of 3 + 46).
The hydroboration of 3 was achieved using borane-THF,
either as a commercially available solution in THF, or
generated in situ from BF3‚etherate and sodium borohydride.8
When the amount of borane used in the hydroboration was
reduced to 1.2 equiv, the yield of the hydroboration began
to vary from lot-to-lot. It was observed that the first
Scheme 3
116
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development

Scheme 4a
ously noted, this procedure has the liability that large volumes
are required to dissolve all of the Oxone required for the
oxidation. A catalytic (10%) amount of Oxone along with
hydrogen peroxide resulted in only 60% conversion of the
alkylborane to the corresponding alcohol. This could have
been a result of the pH of the resulting solution, as the Oxone
mixture contains a stoichiometric amount of an acidic salt
(KHSO4), and a substoichiometric quantity of the reagent
was used. Using hydrogen peroxide under acidic conditions
(pH 2) did provide rapid oxidation of the alkylborane
intermediate without requiring the extended reaction times
at elevated temperature. This protocol required significantly
lower volumes of solvent to execute than did the Oxone
procedure. Other protocols for the oxidation of the B-C bond
led to lower yields and numerous byproducts.11
a Reaction conditions: a) SO3‚pyridine, DMSO, Et3N; b) MeNH2, NaH-
B(OAc)3, toluene, EtOH, THF, HOAc; c) HCl, EtOH, EtOAc.
and the flexibility to charge more reagent if the reaction does
not proceed to completion. Oxidation of alcohol 4 to ketone
8 was accomplished without first breaking the tosylate salt.15
After oxidation, the pH of the workup had to be carefully
controlled to maximize product recovery. When the aqueous
layer pH was kept around 10 with 25% ammonium hydrox-
ide, a 93% yield of product was extracted into toluene (HPLC
purity of 87%, with 6% of the methylthiomethyl ether as
the major contaminant16). Interestingly, the sulfate ester of
the cis-alcohol (cis-4) was identified in the aqueous phase
after workup, indicating a divergent reactivity pattern for this
diastereomer.17 The product from this step was carried into
the reductive amination crude as a toluene solution. Unre-
acted alcohol was always observed as a major impurity
(1-5%) from the oxidation, but this material readily purged
in the downstream chemistry (vide infra). After workup,
nitrogen was sparged through the organic phase containing
the ketone via an immersed tube at 50 °C for 1 h, and the
dimethyl sulfide was trapped in a bleach scrubber.18
A tosylate salt (4‚TsOH) was identified that allowed
isolation of the alcohol in high purity. Interestingly, formation
of the salt from product derived from the acidic hydrogen
peroxide oxidation generated a significant amount of hy-
drogen (∼1 mol equiv), while the material from the Oxone
oxidation evolved only small amounts of hydrogen in the
same salt-forming procedure. The salt is made by addition
of 1.1 equiv of p-toluenesulfonic acid monohydrate to the
product in acetone with seed crystal added to the acid
solution. Product thus obtained reproducibly had purity
higher than 99% in a yield of 80-90% for the transformation
of the olefin to the alcohol (four runs). The material isolated
from the hydroboration sequence was between 3.5:1 to 4.7:1
trans:cis.12
The Jones oxidation5 originally reported1 could be re-
placed with a Swern13 oxidation using either oxalyl chloride
or SO3‚pyridine14 as the activating agent (Scheme 4). On
scale, the SO3‚pyridine oxidation was selected due to the
noncryogenic reaction temperature, easier to handle reagents,
Several parameters were evaluated in an effort to optimize
the selectivity and reproducibility of the reductive amination.
In general, NaB(OAc)3H was found to be a more selective
reagent than NaBH4. On a laboratory scale, a cis:trans ratio
of 97:03 was obtained when 1.3 equiv of NaB(OAc)3H was
added to the imine, formed by treating the ketone with
methylamine (2 equiv, 2 M solution in THF) and acetic acid
(6) Fukuyama, T.; Jow, C.-K.; Cheung, M. Tetrahedron Lett. 1995, 36, 6373-
6374. On attempting to invert the alcohol stereocenter using N-methyl-p-
nitrobenzenesulfonamide under Mitsunobu conditions, the five-membered
ring product was isolated. This was presumably formed via displacement
of the activated alcohol by the piperidine nitrogen followed by opening of
the aziridine at the less hindered position. This instability also made the
epoxide intermediate that would be derived from compound 3 unattractive
synthetically.
(15) Water in the starting alcohol was found to have a large effect on the yield
of the oxidation. For example, while material with 0.1% water content
provided a 99% conversion in the oxidation, material with 2% water only
went to 42% conversion under the same conditions. The free base could
be azeotropically dried more reproducibly than the salt. The salt could be
free-based in toluene with aqueous sodium hydroxide followed by distil-
lation of toluene and additional drying by stripping with two more additions
of toluene. This procedure gave 0.1-0.2% water levels in the bulk
reproducibly. For tosylate salt lots with sufficiently low water levels, no
free-basing was performed.
(7) (a) Oediger, H.; Joop, N. Justus Liebigs Ann. Chem. 1972, 764, 21-27.
(b) Oediger, H.; Moeller, F. German Patent DE 2101997, 1971. (c) Massiot,
G. Bull. Soc. Chim. Belg. 1990, 99, 717-728. (d) Keay, J. G. In
ComprehensiVe Organic Synthesis; Trost, B. M., Ed.; Pergamon: New
York, 1991; Vol. 8, Chapter 3.6.
(16) This structural assignment is based on 1HNMR and GC/MS (M ) 265).
(17) Structure assigned by 1HNMR coupling constants. This difference in
reactivity pathway might be attributed to intramolecular conjugate base
assistance of the piperidine nitrogen on the axially disposed alcohol in the
cis-alcohol. The possibility that this sulfate ester is produced by inversion
of an activated intermediate from the trans-alcohol is discounted on the
basis of the results discussed in ref 6. This finding seems to be distinct
from the faster oxidation of axial alcohols (of cyclohexanols) as compared
to equatorial alcohols observed in the literature in that it is a difference in
reaction mechanism as opposed to rate. (a) Eliel, E. L.; Schroeter, S. H.;
Brett, T. J.; Biros, F. J. Richter, J.-C. J. Am. Chem. Soc. 1967, 89, S 1966,
88, 3327-3334. (b) Mueller, P.; Perlberger, J.-C. J. Am. Chem. Soc. 1976,
98, 8407-8413.
(8) (a) Brown, H. C. Tetrahedron 1961, 12, 117-138. (b) Brown, H. C.; Rao,
B. C. S. J. Org. Chem. 1957, 22, 1135-1136.
(9) Hutchins, R. O.; Learn, K.; Nazer, B.; Pytlewski, D.; Pelter, A. Org. Prep.
Proced. Int. 1984, 16, 337-372.
(10) Ripin, D. H. B.; Cai, W.; Brenek, S. J. Tetrahedron Lett. 2000, 41, 5817-
5819.
(11) The workup with alkaline hydrogen peroxide required refluxing for several
hours to consume the alkylborane. Also tried were trimethylamine N-oxide,
MCPBA, and sodium perborate.
(12) Based on 1H NMR analysis.
(13) Mancuso, A. J.; Swern, D. Synthesis 1981, 165-185.
(14) Parikh, J. R.; Doering, W. von E. J. Am. Chem. Soc. 1967, 89, 5505-
5507.
(18) Liu, C.; Ng, J. S.; Behling, J. R.; Yen, C. H.; Campbell, A. L.; Fuzail, K.
S.; Yonan, E. E.; Mehrotra, D. V. Org. Process Res. DeV. 1997, 1, 45-54.
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
117

(1 equiv, T < 30 °C, 30 min). Under the same reaction
conditions, a cis:trans ratio of 93:07 could be obtained using
a EtOH/THF solvent mixture. The order of addition did not
have a significant effect on the diastereoslectivity; addition
of a solution of the imine to a suspension of the reducing
reagent or adding a solution of methylamine and acetic acid
to the ketone and NaB(OAc)3H, at 0-5 °C did not influence
the selectivity.
resolution, without isolation of the free base. The resolution
has to be seeded with salt of high ee. If seeding is not done,
a low ee salt is initially recovered (∼60% ee), and the ee
has to be upgraded with successive reworks. The procedure
is to free-base using 2.04 equiv of 2 N NaOH in i-PrOH/
MeOH(3:2), add the tartaric acid (0.5 equiv), reflux, cool
with seeding at 72 °C, and slurry prior to isolation. The pH
of the solution after addition of the NaOH was 10; a previous
run where 2.11 equiv of NaOH was used went to pH 13 and
resulted in a lower isolated yield (37% vs 44%). Using this
resolution, material of 99.2% ee was obtained. This procedure
resolves, but does not effectively purge, the trans isomer from
the reductive amination, and it is therefore important to
upgrade the de at the bis-HCl salt stage. Reworks of the
tartrate salt do not significantly improve the de or ee of the
material beyond what is initially isolated after seeding.
On-scale, sourcing issues prompted us to use an 8 M
solution of methylamine in EtOH and dilute it with 3 vols
of THF. While 2 equiv of methylamine was found to be
sufficient for complete conversion on lab scale, 4 equiv of
methylamine with 1 equiv of HOAc was used on-scale to
ensure a rapid and clean conversion to the imine. The
obtained solution of the imine was added to a suspension of
NaB(OAc)3H in THF (2.3 vols) at 10 °C internal temperature.
Online IR data indicated that the imine is quite stable in
solution at 20 °C for several hours, but when vacuum was
applied, a significant amount of ketone was regenerated.
Partial vacuum had been used for transfer of the imine
solution to a feed tank for the reduction, and this feed tank
had been put under vacuum prior to transfer. To solve the
problems on-scale, extra methylamine was used to generate
the imine (4.1 equiv vs 4.0 equiv), and all transfers and
inerting were performed under nitrogen pressure as opposed
to partial vacuum. These changes resulted in a clean product
solution (86% purity, 86:14 cis:trans) in toluene, with <4%
of alcohol 4 and 5-8% of the methylthiomethyl ether as
the major contaminants. This material was purified via
formation of the HCl salt.
The bis-HCl salt (5‚2HCl) was identified as an efficient
purge of the undesired trans isomer of the product, as well
as being a nicely stable form of the material. The salt is
formed by adding HCl in ethanol to the crude toluene
solution from the reductive amination. The initially formed
salt precipitates with a diastereomeric ratio of 91:9. The
diastereomeric purity of this salt can be upgraded by slurry
in ethanol, recrystallization from methanol, or recrystalliza-
tion from methanol/ethyl acetate. All three methods gave
comparable cis:trans ratios (98.6:1.4 to 98.9:1.1), with the
ethanol repulp yielding much more (91% recovery) than
either of the recrystallizations (79-83% for two crops). The
isolated yield over the last three steps (oxidation, reductive
amination, and salt formation) was 47-63%. A total of 34.2
kg of material was made via this process.
Conclusions
In conclusion, a synthesis of 9 has been developed and
scaled to produce 19.4 kg (10.3 kgA) of resolved material
in 7.2-11.4% overall yield (two batches), 97.8% purity
(1.9% trans diastereomer), and 99.4% ee.
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. All reactors were glass-lined steel vessels. Reac-
tions were monitored for completion by removing a small
sample from the reaction mixture and analyzing the sample
by TLC and GC or HPLC. HPLC analyses were performed
using a Waters Symmetry Shield RP18 column (4.6 mm ×
75 mm, 3.5 µm) and an acetonitrile/water (0.5% trifluoro-
acetic acid) mobile phase. GC analyses were performed on
1
a DBWaxEtr (15 m × 0.25 mm i.d. × 0.25 µm film). H-
spectroscopy was performed at 400 MHz on a Bruker-
Spectrospin Avance 400 MHz. The LOD (loss on drying)
was determined by concentrating a weighed sample from
the bulk material.
N-Benzyl-4-methyl-pyridinium Chloride (7). A 640-L
reactor was charged with 4-picoline (6) (54.8 kg, 586 mol),
followed by acetone (150 L) at 18 °C. The solution was
heated to 52 °C, and a solution of benzyl chloride (69.8 kg,
551 mol, 0.95 equiv20) in acetone (53 L) was added over a
period of 4 h and 30 min at 50-52 °C. The suspension was
aged for 16 h at 55 °C. It was then cooled to 10 °C over a
period of 90 min prior to filtration over a pressure filter dryer.
The filter cake was washed with acetone (2 × 54 L) and
dried on the filter bed at 41 °C to give 7 (82.99 kg, 69%) as
slightly pink crystalline solid. Purity: 100% a/a (HPLC);
LOD ) 0%.
N-Benzyl-4-methyl-1,2,5,6-tetrahydropyridine (3). A
400-L reactor was charged with 7 (40.65 kg, 185 mol) and
EtOH (185 L). NaBH4 (8.75 kg, 231 mol, 1.25 equiv) was
added to the solution via a solid dosing system over a period
of 7 h and 10 min at 10-19 °C. The suspension was stirred
Two salts have been identified that cleanly resolve the
racemic amine, the (S)-(+)-Andeno acid19 (phencyphos,
(S)-(+)-2-hydroxy-5,5-dimethyl-4-phenyl-1,3,2-dioxyphos-
phorinane-2-oxide) salt and the di-p-toluoyl-L-tartaric acid
salt. The Andeno acid is not commercially available for a
reasonable cost; consequently, the tartaric acid salt was
utilized on-scale. The salt that resolves nicely is a 2:1
amine:acid salt, meaning that in theory, 0.25 equiv of the
acid are needed for complete resolution. An unusual aspect
of this resolution is the fact that the free-basing step to
neutralize the bis-HCl salt can be combined with the
(19) (a) Wijnberg, H.; Ten Hoeve, W. EP180276, 1986. (b) Ten Hoeve, W.;
(20) The use of 0.95 equiv of BnCl is to minimize contamination of workers
during workup. Essentially no excess is present after the reaction.
Wijnberg, H. J. Org. Chem. 1985, 50, 4508-4514.
118
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development

for 10 h at 15 °C. IPC (HPLC) showed >96% conversion.
The quench was effected by addition of water (112 L) over
a period of 3 h at 8-10 °C. Celite (8.03 kg) was charged
into the reactor to achieve a more granular solid. The stirrer
was turned off to allow for sedimentation overnight at 14
°C. The upper clear phase was siphoned. The lower phase
(suspension) was filtered and washed with EtOH (3 × 13
L). The filtrate (288 L) was concentrated to ca. 133 L. The
pH of the mixture was adjusted to 8 by addition of 1 N NaOH
(2 L), the layers were separated, and the aqueous phase was
extracted with TBME (2 × 50 L). The aqueous phase was
mixed with 1 N NaOH (10 L) and extracted with the last
organic phase. The combined organic phases were washed
with water (60 L) and saturated aqueous NaCl solution (50
L) and concentrated at 35 °C to give ca. 60 L of orange oil.
To remove the remaining water the oil was stripped with
THF (2 × 30 L) at 39 °C to give 3 (30.35 kg, 73%) as a
yellow oil. Purity: 94.8% a/a (HPLC);21 GC headspace:
0.64% EtOH, 11.1% THF; Karl Fischer: 0.51% H2O.
Spectral data was consistent with that reported in the
literature.1
N-Benzyl-3-hydroxy-4-methyl-piperidinium Toluene-
4-sulfonate (4‚TsOH). A 640-L reactor was charged with
3 (27.78 kg, 123 mol) and THF (62 L). After cooling the
solution to 5 °C, BF3‚OEt2 (22.4 kg, 158 mol, 1.3 equiv)
was added over 26 min while keeping the temperature
between 5 and 10 °C. Next, BH3‚THF (1 M in THF, 171 L,
1.4 equiv) was added to the off-white suspension over a
period of 3 h and 10 min, keeping the temperature between
8 and 13 °C. The solution was heated to 20 °C and stirred
for 14.5 h. According to IPC (HPLC) the starting material
was fully consumed. The solution was cooled to 7 °C, and
MeOH (37 L) was slowly added over 2 h, keeping the
temperature between 7 and 12 °C. The reaction mixture was
warmed to 19 °C over 1 h and stirred at this temperature for
30 min; 10% CaCl2 in 0.2 N HCl solution (40 L) was added
over 17 min. The resulting solution had a pH of 2. The
reaction mixture was stirred for 1 at 20 °C and then for 1 h
at 41 °C. Finally, 135 L of solvent were distilled off at 40
°C. The reaction mixture was cooled to 7 °C, and H2O2
(17.5%, 28.4 kg, 146 mol, 1.2 equiv) was added over 2 h,
keeping the temperature between 7 and 13 °C. The reaction
mixture was stirred for 68 h at 19-22 °C. For workup, 10%
Na2SO3 solution (13 L) and TBME (50 L) were added at 19
°C. The pH was adjusted to 12 by addition of 30% NaOH
solution (37 L) while keeping the temperature between 9
and 18 °C. The layers were separated, and the aqueous layer
was extracted twice with TBME (2 × 50 L). The combined
organic layers were washed with water (25 L) and aqueous
saturated NaCl solution (25 L); 155 L of solvent was distilled
off at 45 °C. In a 100-L glass-lined reactor p-toluenesulfonic
acid monohydrate (25.5 kg, 134 mol, 1.1 equiv) was
dissolved in acetone (61 L). Seed crystals (0.1 g) in acetone
(0.5 L) were added to this solution prior to addition to the
free base. This solution was added at 44-47 °C to the
solution of the free base in the 600-L vessel over 50 min
(N2 flow to dilute H2: 8.4 m3/h). The white suspension was
stirred for 1.5 h at 5 to -2 °C before the product was filtered
and washed with acetone (2 × 18 L). The solid was dried at
40 °C to afford 4 (41.37 kg, 88%) as a white, fine crystalline
solid. Purity: 99.6% a/a (GC); Karl Fischer: <0.18%;
TGA: <0.3% ashes. Spectral data was consistent with that
reported in the literature for the free base.1
N-Benzyl-4-methyl-piperidin-3-one (8). A 200-L scrub-
ber was charged with 13-15% bleach (60 L) and water (60
L). A 640-L vessel was charged with SO3‚pyridine (51.47
kg, 323.1 mol, 3.0 equiv). DMSO (169 L) was added, and it
was heated slowly at 33 °C IT.22 After a solution was
obtained, it was cooled to 25 °C prior to be transferred into
the feed tank of the vessel. The TsOH salt 4 (40.9 kg, 107.7
mol) was added into the vessel and suspended in DMSO
(50 L). After the addition of Et3N (62 L, 43.8 mol, 4.0 equiv),
the SO3‚pyridine solution in DMSO was added to the two-
phase mixture in the vessel at such a rate as to keep the IT
below 25 °C.23 After 1 h stirring at 22 °C IPC showed 92%
conversion. The mixture was cooled to 10 °C prior to
quenching with water (182 L) over a period of 40 min at
such a rate as to keep IT below 17 °C; 25% NH3 solution
(16 L) was added during 8 min. IPC showed a pH of 10.
After phase separation, the aqueous phase (ca. 445 L) was
extracted with three portions of toluene (3 × 60 L) while
controlling the pH of the aqueous phase after each extrac-
tion: the first aqueous phase had a pH of 7, and 25% NH3
solution (6 L) was added to reach pH 10. The second aqueous
phase had a pH of 8, and 25% NH3 solution (6 L) was added
to reach pH 10. The combined organic phases (ca. 240 L)
were extracted with water (61 L), and the bright-orange
solution was heated at 40-50 °C jacket temperature during
1 h while blowing a nitrogen stream into the solution via an
immersing tube.24 Then, toluene (170 L) was stripped off at
50 °C to afford 8 (50.84 kg, 93%) as an orange solution in
toluene. Purity: 87.4% a/a (GC); LOD ) 54%; the toluene
solution was processed directly in the next step. Spectral data
was consistent with that reported in the literature.1
(N-Benzyl-4-methyl-piperidin-3-yl)methyl Amine (5).
A 640-L reactor was charged with the toluene solution (103
kg 23%, 23.5 kg 100% 7, 116 mol) from a second run of
the previous step. A solution of methylamine in EtOH (33%
in EtOH, 62 L, 496 mol, 4.3 equiv) was added over a period
of 17 min at 17-18 °C, followed by the addition of HOAc
(6.8 L, 116 mol) during 25 min at 18-24 °C. Online IR
showed complete consumption of the ketone7 after 54 min.
This orange solution was transferred into a feed tank which
had been evacuated beforehand. After the reactor was washed
with THF (40 L), NaBH4 (5.74 kg, 150.8 mol, 1.3 equiv)
was added, followed by THF (60 L). To this suspension was
added HOAc (38 L, 684 mol, 5.9 equiv) under vigorous
stirring over a period of 2.5 h such as to keep the temperature
(22) The maximum jacket temperature has to be set at 40 °C due to a Tonset of
59 °C (687 kJ/kg) for a 22% SO3‚pyridine solution in DMSO. The obtained
solution should be stirred no longer than 1-2 h at 30-35 °C.
(23) During the reaction (1 h and 45 min), a slight stream of nitrogen, via one
feed tank, over the reaction mixture was maintained to chase dimethyl
sulfide.
(21) Purity accounts for the sum of the two signals: product at 5.5 min (91.2%
a/a) and HB(OH)2 complex (3.6% a/a) at 5.1 min. This proceeding is based
on the fact that the latter has been shown to be a valuable starting material.
(24) This is to chase the dimethyl sulfide from the organic solution into the
scrubber. For a 573 mol run of an “industrial” Swern oxidation see ref 18.
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
119

below 13 °C. The liberated hydrogen was diluted with a
nitrogen stream after the reactor (4.9 m3 N2/h). This was
maintained during the following reduction step. The imine
solution was added from the feed tank to the suspension in
the reactor over 2 h so as to keep the temperature below 16
°C and to control foaming. The mixture was stirred for 13 h
at 2 °C. IPC showed complete conversion to the amine
(>99%). Water (201 L) was added over 40 min keeping an
IT of 5-8 °C. Subsequently, the pH of the solution was
adjusted to 11 by adding 30% NaOH (35 L). The aqueous
layer (ca. 310 L) was extracted with three portions of toluene
(3 × 80 L).25 All organic phases were combined (ca. 423 L)
and washed with water (80 L). The organic phase was
concentrated at 50 °C to yield 5 (33.93 kg, 92%) as an orange
solution in toluene. Purity: 86.1% a/a sum of cis- and trans-5
(GC); LOD ) 20%; dr 86:14. Rf (freebase) ) 0.24 (9:1 CH2-
Cl2-MeOH). 1H NMR (freebase) (400 MHz, DMSO, 70 °C)
δ 7.28 (m, 5H), 3.47 (d, J ) 13.27, 1H), 3.40 (d, J ) 13.27,
1H), 2.51 (m, 2H), 2.33 (m, 1H), 2.20 (s, 3H), 2.09 (m, 2H),
1.64 (m, 1H), 1.39 (m, 2H), 0.85 (d, J ) 7.05, 3H). 13C
NMR (300 MHz, CD3OD) δ 131.5, 130.3, 129.2, 128.7, 60.8,
55.9, 46.8, 46.4, 30.7, 27.5, 26.3, 9.5.
mol) was added over 45 min keeping the IT at 3-10 °C.
Then 100 L of solvent was stripped off at 55 °C. The red
suspension was cooled to 30 °C, and AcOEt (215 L) was
added; 210 L of solvent was stripped off at 50 °C. This was
repeated with a second portion of AcOEt (215 L). Acetone
(111 L) was added at 35 °C over 15 min. The suspension
was cooled to -1 °C over 7 h and then filtered and washed
with acetone (55 L) to give the crude product (35.77 kg,
LOD 20%, dr 91:9). This crude was slurried three times with
AcOEt (the crudes showed dr 95.8:4.2, 97.2:2.8) to yield
5‚2 HCl (27.38 kg, 19.44 kg corrected, 62%) as off-white
powder. Purity: 99.8% a/a (GC); LOD 29%; dr 97.3:2.7;
mp 260-263 °C. Anal. Calcd for C14H24Cl2N2: C, 57.73;
H, 8.31; N, 9.62. Found: C, 57.32; H, 8.31; N, 9.48.
Acknowledgment
We acknowledge Ernie Pollard for developing and run-
ning assays for the chemistry described, (mass spec, analyti-
cal, other support) and Volker Wesseling (CarboGen) for
developing and running the analytical methods applied in
the development and in the pilot plant (IPC and assay).
Technical assistance of Dario Menia and Barbara Bianchi
(CarboGen) is gratefully acknowledged. Jim Phillips is
acknowledged for collecting some analytical data.
(N-Benzyl-4-methyl-piperidin-3-yl)methylamine Dihy-
drogenchloride (5‚2 HCl). A 640-L reactor was charged
with 5 (33.93 kg solution in toluene, 69%, 107 mol, dr
86:14) and EtOH (120 L). At 3 °C, 32% HCl (25 L, 249
Received for review October 11, 2002.
OP025599X
(25) Some solid precipitated in the interphase and was separated in the aqueous
phase, as this proved not to be the product.
120
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development