Comparison of large-scale routes to manufacture chiral exo-2-norbornyl thiourea

Article abstract of DOI:10.1021/op9002328

Two routes aimed at the manufacture of chiral exo-2-norbornyl thiourea (1) on large scale are described. The first approach involves five chemical steps and hinges on a classical resolution via diastereomeric salt formation. The synthesis utilizes amine 2 as the resolution handle. The second approach includes two chemical steps and a chiral chromatography of (±)-1. Despite the larger initial investment necessary to acquire the chiral stationary phase used in the chromatographic approach, the shorter reaction sequence and efficiency of the chromatographic separation make the second route a more attractive option for long-term applications.

Full text of DOI:10.1021/op9002328

Organic Process Research & Development 2010, 14, 133–141
Comparison of Large-Scale Routes to Manufacture Chiral exo-2-Norbornyl Thiourea
Seb Caille,*,† Jerome Boni,⊥ Geoffrey B. Cox,¶ Margaret M. Faul,† Pilar Franco,| Saab Khattabi,¶ Liane M. Klingensmith,†
Jay F. Larrow,† James K. Lee,¶ Michael J. Martinelli,† Larry M. Miller,‡ George A. Moniz,† Kenichi Sakai,O Jason S. Tedrow,†
and Karl B. Hansen*,†
Chemical Process Research and DeVelopment, Chemistry Research and DiscoVery, Amgen Inc., One Amgen Center DriVe,
Thousand Oaks, California 91320, and 360 Binney Street, Cambridge, Massachusetts 02142, NoVasep, Site Eiffel-81, BouleVard
de Moselle-BP50, 54340 Pompey, France, Chiral Technologies, 800 North FiVe Points Road, West Chester,
PennsylVania 19380, and Parc d’InnoVation, Gonthier d’Andernach BouleVard, 67400 Illkrich-Cedex, France
Abstract:
treatment of type 2 diabetes.5 The preparation of 1 in racemic
form has been reported.6 However, the development of a route
that could be utilized to prepare 1 in optically enriched form
was desired. This contribution describes two strategies for the
preparation of 1 with high optical purity. The first employs
resolution by diastereomeric salt formation (classical resolution).
The second achieves resolution by chiral chromatography. A
comparison of these two methods reveals that the chromato-
graphic approach is most effective due to the high efficiency
of the separation, the relative ease of access to the racemate,
and the fewer unit operations required. This report describes
development activities around these two approaches and
presents an evaluation of the two routes in terms of overall
efficiency.
Two routes aimed at the manufacture of chiral exo-2-norbornyl
thiourea (1) on large scale are described. The first approach
involves five chemical steps and hinges on a classical resolution
via diastereomeric salt formation. The synthesis utilizes amine 2
as the resolution handle. The second approach includes two
chemical steps and a chiral chromatography of (()-1. Despite the
larger initial investment necessary to acquire the chiral stationary
phase used in the chromatographic approach, the shorter reaction
sequence and efficiency of the chromatographic separation make
the second route a more attractive option for long-term applications.
1. Introduction
Retrosynthetic analyses of both approaches to 1 are shown
in Scheme 1. In order to access a moiety which can form a
chiral salt, the thiourea was envisioned as arising from S-(exo)-
2-aminonorbornane (2). Conversion of the amino group of 2
can be achieved via a two-step process involving reaction with
benzoyl isothiocyanate followed by hydrolysis of the benzoate
moiety. Norbornyl amine 2 can be accessed from norbornene
(3) via a Ritter reaction with CH3CN, followed by hydrolysis
of the resultant acetamide.7
During our development of a chiral assay for 1, conditions
developed on analytical scale suggested that preparative chiral
chromatography may be a viable method for resolution of (()-
1. This raised the possibility of an alternative approach that
would involve forming the thiourea directly via a Ritter reaction
with isothiocyanate,8 followed by ammonia addition and
chromatographic separation of the enantiomeric mixture. While
the chromatographic approach would require more specialized
equipment to effect the resolution than a traditional crystalliza-
tion, the relative ease of access to the racemic thiourea made it
worth considering. A description of the strengths and drawbacks
of these two approaches follows.
The efficient preparation of chiral non-racemic building
blocks continues to challenge synthetic chemists. Classical
resolution of racemic mixtures has been a method of choice1
to accomplish this goal. However, resolution by large-scale
chiral chromatography has developed into an attractive alterna-
tive.2 While resolution by diastereomeric salt formation has the
benefit of implementation in simple reactors and equipment,
large-scale chromatography can be scaled from a multi-kilogram
to a multi-ton process. The criteria for deciding which approach
is most attractive for large-scale applications lie in the overall
synthetic strategy.
Thiourea 1 is a starting material3 employed in the synthesis
of AMG 221, an inhibitor of 11ꢀ-hydroxysteroid dehydrogenase
type 1 (11ꢀ-HSD1) discovered at Amgen (Scheme 1).4 Inhibi-
tors of 11ꢀ-HSD1 are potential therapeutic agents for the
* Authors to whom correspondence may be sent. E-mail: scaille@amgen.com;
karl@amgen.com.
† Chemical Process Research and Development, Amgen.
‡ Chemistry Research and Discovery, Amgen.
⊥ Novasep.
¶ Chiral Technologies, West Chester, PA.
| Chiral Technologies, Illkrich-Cedex, France.
O Current Address: Toray Fine Chemicals, Technology Development Division
9-1, Oe-cho, Minato-ku, Nagoya, 455-8502, Japan.
(1) Jacques, J; Collet, A.; Wilen, S. H. Enantiomers, Racemates and
Resolutions; Wiley: New York, 1981.
(5) Kershaw, E. E.; Morton, N. M.; Dhillon, H.; Ramage, L.; Seckl, J. R.;
Flier, J. S. Diabetes 2005, 54, 1023–1031.
(2) Cox, G. PreparatiVe EnantioselectiVe Chromatography; Blackwell:
(6) Rasmussen, C. R.; Villani, F. J.; Weaner, L. E.; Reynolds, B. E.; Hood,
A. R.; Hecker, L. R.; Nortey, S. O.; Hanslin, A.; Coastanzo, M. J;
Powel, E. T.; Molinari, A. J. Synthesis 1988, 6, 456–459.
(7) Bobyleva, A. A.; Petrushchenkova, I. A.; Lukovsoya, E. V.; Pekhk,
T. I.; Dubitskaya, N. F.; Belikova, N. A. Zh. Org. Kim. 1989, 25,
1428.
London, 2005.
(3) Caille, S.; Cui, S.; Hwang, T-L; Wang, X.; Faul, M. M. J. Org. Chem.
2009, 74, 3833–3842.
(4) Henriksson, M. Homan, E.; Johansson, L.; Vallgarda, J.; Williams,
M.; Bercot, E.; Fotsch, C. H.; Li, A.; Cai, G.; Hungate, R. W.; Yuan,
C.; Tegley, C.; St. Jean, D.; Han, N.; Huang, Q.; Liu, Q.; Bartberger,
M. D.; Moniz, G. A.; Frizzle, M. J. Patent WO 2005116002, 2005.
(8) Diveley, W. R.; Buntin, G. A.; Lohr, A. D. J. Org. Chem. 1969, 34,
616–624.
10.1021/op9002328  2010 American Chemical Society
Published on Web 12/08/2009
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
133

Scheme 1. Approaches toward chiral norbornyl thiourea (1)
2. Results and Discussion
findings were attributed to the incorporation of 7 into the crystal
lattice of salt 6 as a solid solution. X-ray powder diffraction
(XRPD) spectra of salt 6 of various purity levels (>99.5/0.5
DR, 97.5/2.5 DR, and 90/10 DR) were measured and showed
no noticeable differences. This single set of signals displayed
in the XRPD spectra of salt 6 of various diastereomeric ratios
suggests that the minor isomer (7) in this mixture is not present
as a different polymorph and consequently supports the crystal-
lization of the mixture as a solid solution.9
2.1. Classical Resolution Approach. 2.1.1. Synthesis of
Amine 2 from Norbornene. The amine racemate (()-2 was
prepared in a two-step sequence via a Ritter reaction using
CH3CN and H2SO4 followed by hydrolysis of the resultant
acetamide intermediate [(()-4] (Scheme 2). The original
literature procedure to conduct this transformation involved
treatment of a relatively dilute (45 mL of CH3CN per g of 3)
solution of norbornene with 4 equivalents of H2SO4.7 Optimiza-
tion of the experimental conditions revealed that the Ritter
reaction could be carried out using a more concentrated CH3CN
solution (5 mL of CH3CN per g of 3) and 1.1 equivalents of
H2SO4. In order to control the strong exotherm generated, H2SO4
was added over 2 h (scale of 0.5 kg of 3) while maintaining
the temperature between -10 and 15 °C. Following the addition,
the mixture was warmed to 23 °C and stirred for an additional
2 h, resulting in complete consumption of 3.
The acetamide group of (()-4 could be hydrolyzed directly
without isolation. First, the excess CH3CN was removed by
distillation, water was added, and the dark mixture was refluxed.
An extended age time at reflux (30 h) was found to be necessary
to achieve complete hydrolysis of (()-4. Following treatment
with 10 M aqueous NaOH and extraction with MTBE, the crude
residue was vacuum distilled to generate (()-2. This procedure
was demonstrated using 150 kg of 3, affording (()-2 in 78%
overall yield.
2.1.2. Classical Resolution of Amine 2. A screen of 18
commercially available resolving agents was performed, and
(R)-(+)-N-(1-phenylethyl)phthalamic acid (5) was identified as
the most competent chiral acid for the classical resolution of
(()-2. The corresponding salts 6 (desired diastereomeric salt)
and 7 (undesired diastereomeric salt, amorphous glassy solid
with no defined melt) were prepared separately in >99.5/0.5
DR and their respective solubilities were measured in various
solvents (Table 1). The differences in solubility between 6 and
7 in a number of solvents were large enough to expect that the
resolution of (()-2 could be achieved via a single crystallization
to generate salt 6 of high DR (>98/2) and >40% yield. Despite
these solubility differences, resolution experiments carried out
in these solvents did not yield salt 6 of >75/25 DR. These
It was determined experimentally that the use of mixtures
of protic solvents (EtOH, IPA) and CH3CN was superior to
CH3CN alone at achieving the desired resolution. Experimental
conditions were surveyed to effect the resolution of (()-2 with
5 using mixtures of EtOH and CH3CN (Table 2). Comparison
of entries 1 and 2 with entry 3 (Table 2) reveals that the use of
0.5 or 0.75 equivalent of 5 in the resolution relative to the use
of 1 equivalent of 5 did not offer an advantage with regard to
the DR of resultant salt 6. However, the yields of 6 were lower
in the former instances than in the latter instance. All subsequent
screening was thus conducted with 1 equivalent of 5. Mixtures
of CH3CN and EtOH of various proportions were surveyed to
carry out the resolution (entries 3-6). A 6/1 mixture of CH3CN
and EtOH was found to afford the maximum yield of salt 6
(43%, entry 5) without negatively affecting the DR (90/10).
By comparison, the use of a 7/1 mixture of the two solvents
(entry 6) afforded salt 6 of a lower DR (86/14). The conditions
reported in entry 5 were selected to be applied on larger scale.
An overview of the resolution process to provide salt 6 of
>99.5/0.5 DR is presented in Scheme 3. First, the resolution of
(()-2 using 5 was conducted (Scheme 3). For this purpose, a
solution of (()-2 in CH3CN was added to 5 in EtOH at 60 °C,
the mixture was seeded with 6 (>99.5/0.5 DR, 1.5 wt %), cooled
to 50 °C, and aged for 12 h.10 The slurry was cooled to 23 °C
over 12 h. The total volume of solvents used was 24 L per kg
of (()-2. The yields ranged from 38% to 44%, and the
diastereomeric ratios spanned from 88/12 to 91/9 for the
resolution from (()-2.
In order to obtain material of >99.5/0.5 DR, two recrystal-
lizations of the initially generated salt (88/12 to 91/9 DR) were
necessary (Scheme 3). The procedure for these recrystallizations
involved dissolving 6 in EtOH at 60 °C, charging both CH3CN
(60 °C) and the seeds of 6 (>99.5/0.5 DR, 1.5 wt %), and
cooling to 50 °C. The remainder of the procedure for the
Scheme 2. Synthesis of amine (()-2 from 3 via a Ritter
reaction
(9) Kozma, D.; Tomor, K.; Novak, C.; Pokol, G.; Fogassy, E. J. Therm.
Anal. 1996, 46, 1613–1623.
(10) The 12 h cycling time improved the DR of resultant salt 6 for the
resolution and the subsequent recrystallizations.
134
•
Vol. 14, No. 1, 2010 / Organic Process Research & Development

Table 1. Solubility of salts 6 and 7 in various solventsa
entry
solvent
IPA
1-propanol
CH3CN
IPAC
solubility of 6 (mg/mL)
solubility of 7 (mg/mL)
1
2
3
4
5
6
24.7
68.5
1.0
1.1
8.7
11.3
>100
>100
>100
>100
>100
>100
THF
1,2-DME
a Solubilities measured at 23 °C.
Table 2. Resolution screen using chiral acid 5a
entry
equivalents of 5
CH3CN/EtOH
total volume of solvents
6 (%)
DR
1
2
3
4
5
6
0.5
0.75
1
4/1
4/1
4/1
5/1
6/1
7/1
25
25
25
24
24
24
21
27
36
40
43
46
90/10
92/8
90/10
91/9
90/10
86/14
1
1
1
a Reaction conditions: Total volume of solvents are reported in mL per g of (()-2 utilized. Isolated yields are reported. Scale of experiments span from 0.3 to 4 g of
(()-2.
Scheme 3. Resolution and recrystallizations to provide salt 6 of >99.5/0.5 DR
Table 3. Stability of 5 in EtOH and CH3CN at 60 °Ca
volume of
solvent
entry
solvent
8 (%)
10 (%)
11 (%)
Figure 1. Impurities generated in the resolution of (()-2 with
5.
1
2
EtOH
CH3CN
1.4
8.4
7
7
37
0
40
0
recrystallizations was identical to that used for the resolution
process. The total volume of solvents employed was 12.5 L
per kg of 6 in the first recrystallization and 10 L per kg of 6 in
the second. The yields varied from 56% to 66% using salt 6 of
88/12 to 91/9 DR as starting material.
a Reaction conditions: Volumes of solvents are reported in mL per g of 5
utilized. Scale of experiments is
3 g of 5. Percentages of side products are
measured by 1H NMR integration relative to 5.
yield, 90.3/9.7 DR) was 5.5% of 8, 2.1% of (()-9, 13.0% of
10, and 14.5% of 11.
Four major impurities were formed in the resolution of (()-2
with 5 (Figure 1) and isolated from the reaction mixture. The
Impurities 8, 10, and 11 were potentially generated in
significant amounts before addition of the CH3CN solution of
(()-2 to the EtOH solution of 5 at 60 °C. The preparation of
this charging step involved warming a solution of 5 in EtOH
to 60 °C. This took up to 1 h, and consequently the stability of
5 under these conditions had to be thoroughly understood. The
stability of 5 in EtOH and CH3CN at 60 °C was studied
separately (Table 3). Stirring of 5 in EtOH for 12 h at 60 °C
(entry 1) generated 37% of 10 and 40% of 11 relative to 5,
1
identity of these compounds was determined by H NMR
comparison with independently synthesized11 8 and (()-9 and
commercially available 10 and 11. The level of these impurities
relative to the total amount of both diastereomeric salts in a
resolution experiment performed to afford 100 g of salt 6 (38%
(11) Compound(()-9waspreparedaccordingtoareportedprocedure:Eddine,
A. C.; Daich, A.; Jilale, A.; Decroix, B. Heterocycles 1999, 51, 2907–
2915.
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
135

Scheme 4. Synthesis of thiourea 1 from salt 6
Scheme 5. Proposed mechanism for formation of byproduct 14
whereas the same experiment conducted in CH3CN (entry 2)
produced no 10 or 11. Considering that 10 and 11 were the
two principal impurities formed in the resolution experiment,
the result may be improved by warming 5 in CH3CN and (()-2
in EtOH instead of the opposite. This was first verified on small
scale and subsequently demonstrated using 238 kg of (()-2 to
afford 345 kg of 6 (42% yield, 90.1/9.9 DR). The levels of
impurities relative to the total amount of both diastereomeric
salts in this plant operation were 3.2% of 8, 3.1% of (()-9,
0.0% of 10, and 3.7% of 11. This protocol, providing a better
isolated yield of 6 (42% vs 38%) and impurity profile relative
to earlier results, represents our optimized procedure to carry
out the resolution of (()-2.
2.1.3. Synthesis of Thiourea 1 from Salt 6. The synthesis
of chiral building block 1 was completed by conversion
of salt 6 into thiourea 1. A two-step procedure was chosen
whereby acylthiourea intermediate 13 was obtained from
6 by treatment with benzoyl isothiocyanate (12)6 and
subsequently solvolyzed to yield 1 (Scheme 4). To
eliminate the need for a separate salt break step, chiral 6
was employed directly in this process. Suspension of 6
in CH2Cl2 was followed by addition of Et3N and 12 to
afford intermediate 13 in >95% purity. The main side
product of the reaction was benzamide 14, presumably
arising from competitive attack of amine (S)-2 at the
carbonyl group rather than the isothiocyanate carbon of
12 (Scheme 5).12 Benzamide 14 is not easily rejected in
the subsequent processing steps. Consequently, its forma-
tion had to be minimized and controlled at this step. At
ambient temperature the levels of 14 were of approxi-
mately 10%. Maintaining the reaction temperature below
0 °C minimized the formation of this impurity to less than
2%. A simple basic work up performed after the formation
of 13 allowed the removal of chiral acid 5 and the crude
process streams could be telescoped into the solvolysis
by solvent exchange from CH2Cl2 to MeOH. Treatment
of 13 with aqueous NaOH in MeOH afforded thiourea 1,
methyl benzoate and benzoic acid (presumably as sodium
benzoate). Crude 1 precipitated from the reaction mixture
and additional water was charged in order to minimize
losses to the liquors (typically 2-3%). Benzamide 14
(0.5-1%) and methyl benzoate (8-10%) were the prin-
cipal contaminates from the MeOH/water isolation. A
suspension and slurry of crude 1 in IPAC effectively
rejected these impurities with minimal losses of 1 (<5%).
This protocol was demonstrated using 123 kg of 6 to
afford 44 kg of 1 (82 wt % adjusted yield, 99.1% ee,
Scheme 4).
2.2. Chromatographic Approach. 2.2.1. Preparation of
(()-1. In order to facilitate the chromatographic separation of
the enantiomers on production scale, a simple and efficient
means to prepare (()-1 was required. The development of an
inexpensive process to prepare the racemate would alleviate
the need to obtain a near quantitative yield of resolved material
and thus reduce the cost of the chromatographic resolution.
Additionally, (()-1 had to be prepared in sufficiently high purity
to prevent contamination of the chiral stationary phase.
The existing route to prepare (()-1 involved inexpensive
starting material and reagents.8 In the first step, norbornene was
reacted with KNCS in the presence of H2SO4. Following
aqueous work up, isothiocyanate intermediate 15 (Scheme 6)
was purified and reacted with ammonia to afford crystalline 1.
Our goal was to improve the synthesis of (()-1 using a
minimum number of unit operations and a simple purification
of the racemate.
The existing procedure was developed into an efficient
process with a few modifications. The isothiocyanate
intermediate (15, Scheme 6) was formed by reaction of
norbornene with NH4NCS (in the presence of H2SO4) in
a toluene/water mixture at 40 °C. Use of NH4NCS instead
of KNCS allowed for complete removal of the inorganic
(12) (a) Becher, J.; Frandsen, E. J. Tetrahedron 1977, 33 (3), 341–343.
(b) Zhang, Y.; Wei, T.; Lu, J. Synth. Commun. 1998, 28 (17), 3243–
3248.
136
•
Vol. 14, No. 1, 2010 / Organic Process Research & Development

Scheme 6. Synthesis of (()-1
Table 4. Solubility of (()-1 in different solventsa
entry
solvent
solubility of racemic 1 (mg/mL)
1
2
3
4
5
6
7
8
9
water
<1
35
21
12
10
38
5
MeOH
EtOH
IPA
CH3CN
THF
EtOAc
CH2Cl2
Heptane
3
<1
a Measurements at 23 °C.
salts during the work up. Potassium salts were determined
to be major impurities in the final product when KNCS
was utilized in the process. Assay yields of 85% were
obtained for the toluene solution and the major impurity
1
present was thiocyanate 16, as judged by H NMR. This
impurity reacted further in the subsequent step and the
species thus generated was rejected in the final isolation.
Isothiocyanate (()-15 was easily converted to (()-1 by
treatment of the toluene solution with aqueous NH4OH
at 50 °C for 20 h. The process afforded (()-1 in 77%
isolated yield over two steps.
Figure 2. Analytical separation of (()-1.
2.2.2. Chiral Stationary Phase Screening and Physiochemi-
cal Data. Two parameters had to be settled upon in order to
develop a scalable process for the chromatographic resolution
of (()-1. First, a solvent system in which the racemate had
high solubility had to be identified. Second, a chiral stationary
phase effecting an efficient enantiomeric separation (selectivity,
R) of the mixture with a retention factor (k′)13 of between 2
and 5 had to be identified. Solubility studies using the racemate
revealed that high (>20 mg/mL) solubility could be achieved
in three common process solvents: MeOH, EtOH, and THF
(Table 4).
A complete screen of chiral stationary phases (CSP)
and solvent systems was carried out. During this study, a
total of 78 CSPs and 330 chromatographic methods were
evaluated. Of these, two method-stationary phase combi-
nations showed promise: (1) Chiral Technologies T101,
MeOH mobile phase and (2) Chiral Technologies CSP-2,
MeOH mobile phase. The analytical separation using both
conditions is shown in Figure 2. In order to select the
optimal CSP for further optimization, loading studies were
performed with the use of an analytical column (4.6 mm
× 250 mm). These studies showed that while the separa-
tion was larger for CSP-2, T-101 exhibited superior
loading capacity, resulting in an approximately 2-fold
increase in productivity14 (Figure 3). Further development
of the method focused on the T101 CSP.
Additional screening of the resolution using T-101 CSP
with alternative mobile phases on analytical scale revealed
that the separation could be further optimized (Table 5).
These experiments showed that the addition of 40%
CH3CN to the mobile phase increased selectivity15 from
2.57 to 3.51 (entry 1 vs entry 6). Despite this increase in
selectivity, an overall decrease in productivity was
expected in such a case due to the reduced solubility of
the racemate in CH3CN/MeOH mixtures. This diminished
solubility coupled with the increased complexity of
recycling a binary solvent mixture led to the choice of
pure MeOH as solvent system for the separation.
In order to model the separation, adsorption isotherm
parameters were determined via a series of analytical and
overloaded injections.16 The adsorption isotherm shape used to
fit the experimental data was based on a modified competitive
Langmuir model. The parameters17 for the adsorption isotherms
are:
(14) The productivity is defined as the amount of desired enantiomer
recovered from the column/amount of chiral stationary phase in the
column/time from beginning of first peak to end of second peak. The
units are normalized to kg of enantiomer/kg of CSP/24 h.
(15) Selectivity is defined as the ratio between the k′ constant of the second
peak and the k′ constant of the first peak (k′ second peak/k′ first peak).
(16) Nicoud, R. M.; Seidel-Morgenstern, A. Isol. Purif. 1996, 2, 165.
(13) k′ is defined as k′ ) (rt- t0)/t0, where rt is the retention time of S-1
(time from injection to the peak maximum) and t0 is the dead time of
the column.
j
(17) Ci is the concentration of the species in the liquid phase and Ci is the
concentration of the species in the solid phase.
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
137

Table 5. Optimization of eluent composition
Rt1
Rt2
selectivity
entry
solvent mixtures (v/v)
(min) (min)
(R)
1
2
3
4
5
6
7
8
MeOH
3.63
3.69
3.45
3.56
3.36
3.28
3.37
5.24
4.61
4.68
3.79
4.25
4.14
3.98
4.08
7.33
2.57
2.43
1.86
2.21
3.17
3.51
3.00
1.93
MeOH/EtOH (4/1)
MeOH/1-propanol (4/1)
MeOH/IPA (4/1)
MeOH/CH3CN (4/1)
MeOH/CH3CN (3/2)
MeOH/CH3CN (2/3)
MeOH/H2O (4/1)
Table 6. Comparison of SMB and Varicol processes
parameter
SMB
1
varicol
1
column diameter
(cm)
column length
(cm)
number of columns
racemate concentration
(g/L)
13.8
15
6
35
5
35
feed flow rate
(mL/min)
1.32
33.25
3.37
5.24
1.60
35.5
4.08
7.33
daily production
(g enantiomer/day)
specific productivity
(g enantiomer/g CSP/day)
eluent composition
(mL/g enantiomer)
Figure 3. Loading studies and estimation of productivity.
Figure 4. Internal concentration profile.
evaluated using the HELP18-CHROM modeling software with
the physiochemical data described for both a simulated moving
bed (SMB) and a Varicol19 process. For this study the SMB
and Varicol processes were optimized for 99% purity and yield.
Identical robustness margins were applied for the two options.
Comparisons of some of the critical parameters for both
processes are summarized in Table 6. The Varicol process
resulted in a 15% increase in specific productivity as well as a
reduction in eluent composition relative to a standard SMB
process.
The pressure drop was measured in order to determine the
Darcy law coefficient: ∆P/L (bar/cm) ) 0.98 u (cm/s). The
column efficiency was calculated by measuring the height
equivalent to theoretical plate (HETP) at two flow rates. The
following Van Deemter equation was obtained: HETP (cm) )
4.9dp (cm) + 0.41u (cm/sec). Using these parameters, modeling
of the system was undertaken to establish the critical factors
necessary to determine productivity on large scale.
(18) HELP is a proprietary chromatographic modeling software developed
by Novasep.
(19) Varicol is an advanced, multicolumn chromatographic process devel-
oped and patented by Novasep.
2.2.3. DeVelopment of SMB/Varicol Process. The efficiency
of a potential large-scale chromatographic resolution was
138
•
Vol. 14, No. 1, 2010 / Organic Process Research & Development

Table 7. Operating conditions at manufacturing scalea
the chromatographic resolution, recycling of the solvent
on production scale would be highly efficient and straight-
forward due to the use of MeOH as a single solvent. These
factors make the chromatographic approach a more
attractive option for large-scale manufacture of 1.
column
diameter
(cm)
year production
entry (metric ton of 1/year))
daily production
(kg enantiomer/day)
1
2
3
5
10
20
20
30
45
15
35
78
4. Experimental Section
GC and HPLC Methods. GC method for (()-2: Agilent
HP-5 (30 m × 0.32 mm × 0.25 µm), 1 µL injection, 1.5 mL/
min, initial temperature 35 °C, temperature ramp 10 °C/min to
100 °C, FID detector, (()-2 at 11.73 min, 3 at 7.78 min. Chiral
GC method for 6 (analyzed as trifluoroacetamide derivative):
Sample preparation is as follows. A sample of 6 (10 mg) was
suspended in CH2Cl2 (1 mg/mL) and treated with aqueous 5
M NaOH (5 mL). The biphasic mixture was shaken and the
layers were separated. The CH2Cl2 layer was treated with 50
µL of trifluoroacetic anhydride, the resultant solution was shaken
and used for injection. Astec Beta TA (30 m × 0.25 mm ×
0.25 µm), 1 µL injection, 4.0 mL/min, constant temperature
95 °C, FID detector, derivative of R-2 at 12.1 min, derivative
of S-2 at 13.3 min. HPLC method for (S)-1 or (()-1: XBridge
C18 (100 mm × 3.0 mm × 3.5 µm), 3.0 µL injection, 0.7 mL/
min, 30 °C, A: 0.1% HClO4 in H2O, B: CH3CN, t ) 0 min:
15% B, t ) 5 min: 60% B, t ) 12 min: 60% B, S-1 or (()-1
at 4.12 min, (()-15 at 9.66 min. Chiral HPLC method for
(S)-1: Chiralcel AD-H (250 mm × 4.6 mm × 5 µm), 10 µL
injection, 1.0 mL/min, 25 °C, 90% hexane/10% EtOH, isocratic;
S-1 at 9.92 min, R-1 at 12.45 min.
a Column length: 15 cm. Number of columns: 5. Working days/year: 250-300.
Eluent composition: 370 mL/g enantiomer.
A proof of concept study was performed utilizing 10 g
of (()-1. For this study specifications of >98% optical
purity and >95% yield were set. Using 10 mm × 150 mm
columns, 118 cycles were performed on a five-column
µVaricol system.20 The desired enantiomer (raffinate, S-1)
was collected at 99% optical purity and 95% yield. The
specific productivity was 0.90 kg of enantiomer/kg of CSP/
day, consistent with values obtained in the modeling
exercise (Vide supra, Table 6). The internal concentration
profile under these conditions is shown in Figure 4. The
proof of concept study showed the Varicol process to be
robust, meeting the purity and yield specifications.
The proof of concept study coupled with the modeling
exercise allowed for the operating conditions at the proposed
manufacturing scale to be determined. As can be seen in Table
7, large quantities of the optically pure material could easily
be accessed using the Varicol process. Despite the large amount
of eluent needed for the resolution, recycling of MeOH is
straightforward. This allows the overall process to be practical
from an operational and cost perspective.
Amine (()-2. A mechanically stirred solution of 3 (150 kg,
1596 mol) in CH3CN (588 kg, 750 L) was kept under N2 and
cooled to -15 °C. H2SO4 (96% purity, 179 kg, 1755 mol, 1.1
equiv) was added over 2.5 h (batch temperature -5 to 15 °C).
The solution was warmed to 23 °C and stirred for 1.5 h. The
reaction mixture was concentrated under reduced pressure (20
mmHg) to remove the solvent. Water (265 L) was added and
the resultant solution was refluxed for 30 h. The reaction mixture
was cooled to 23 °C. Aqueous 10 M NaOH (665 kg, 500 L)
was added (temperature e40 °C and pH of resultant mixture
∼13). The aqueous mixture was extracted with MTBE (1 ×
270 kg and 2 × 135 kg). The organic phases were combined
and washed with brine (50 kg). The organic phases were
concentrated under reduced pressure to remove MTBE (500
mmHg, 50 °C). The residual material was fractionally distilled
under reduced pressure (80 mmHg, boiling point 87-91 °C)
to yield (()-2 as an oil (138 kg, 78%, 1245 mol, 98.9 wt %,
3. Conclusion
The results obtained for these resolution routes allow
for the two techniques to be compared in terms of
efficiency of the resolution as well as overall synthetic
approach. While both routes can deliver resolved 1 of high
optical purity, they differ substantially with regards to the
methods utilized. In the classical resolution approach
(14.5% overall yield), 1 was prepared in five chemical
steps from norbornene (3) via amine 2, an intermediate
that incorporates the necessary resolution handle. The
resolution step required two recrystallizations of the
phthalamic acid salt (6) initially generated in order to
afford material of sufficiently high DR. This is most
probably due to diastereomeric salts 6 and 7 crystallizing
as a solid solution, since the thermodynamic solubility of
the individual salts suggested the development of a
considerably more efficient resolution to be possible.
These two recrystallizations required a large number of
unit operations and resulted in the isolation of salt 6 of
>99.5/0.5 DR in relatively low overall yield (23%).
In contrast, in the chromatographic process (36.5%
overall yield) (()-1 was prepared from 3 in two chemical
steps and resolved with a high level of efficiency. The
chiral stationary phase “resolving agent” is recyclable and
has years of lifetime under the resolution conditions.
Despite the large amount of solvent required to carry out
1
99.2 GCAP). H NMR (400 MHz, CDCl3) δ 2.73 (dd, J )
3.2, 7.6 Hz, 1H), 2.14 (br s, 1H), 1.84-1.88 (m, 1H), 1.56 (ddd,
J ) 4.0, 7.6, 12.8 Hz, 1H), 1.30-1.44 (m, 3H), 0.93-1.25 (m,
6H); 13C NMR (125 MHz, CDCl3) δ 54.5, 44.9, 41.8, 35.5,
33.8, 27.8, 26.3; IR (neat): 3360, 2930, 2868, 1606, 1452, 1354,
1309, 1095, 943, 806 cm-1; Exact Mass [C7H13N + 1]+:
calculated ) 112.1121, measured ) 112.1117.
Phthalamic Salt 6. (R)-(+)-N-(1-Phenylethyl)phthalamic
acid (5) (578 kg, 2149 mol) was dissolved in CH3CN
(3827 kg, 4907 L) at 60 °C. Amine (()-2 (238 kg, 2144
mol) was dissolved in EtOH (645 kg, 816 L) at 60 °C
and the resultant solution was added to the CH3CN
solution of 5 (60 °C) over 10 min. The mixture was seeded
with 6 (12 kg, 99.7/0.3 DR) and cooled to 50 °C over
(20) The total pressure drop generated had an average value of 27 bar.
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
139

0.5 h. The heterogeneous mixture was stirred at 50 °C
for 12 h and the following cooling ramp to 23 °C was
applied: 50 to 40 °C (3 h), 40 °C (3 h), 40 to 30 °C (1.5
h), 30 °C (1.5 h), 30 to 23 °C (1.5 h), 23 °C (2 h). The
mixture was filtered and the solid was rinsed with CH3CN
(400 kg, 513 L). Salt 6 was vacuum-dried at 30 °C for
12 h (345 kg, 41.7%, 908 mol, 100 LCAP, 90.1/9.9 DR).
Salt 6 (345 kg, 908 mol, 90.1/9.8 DR) was dissolved in
EtOH (688 kg, 871 L) at 60 °C. CH3CN (2679 kg, 3434
L, 60 °C) was added over 0.3 h and the solution was
seeded with 6 (5.1 kg, 99.7/0.3 DR). The mixture was
cooled to 50 °C over 0.5 h. The heterogeneous mixture
was stirred at 50 °C for 12 h and the following cooling
ramp to 23 °C was applied: 50 to 40 °C (3 h), 40 °C (3
h), 40 to 30 °C (1 h), 30 °C (1 h), 30 to 23 °C (1 h), 23
°C (2 h). The mixture was filtered, and the solid was rinsed
with CH3CN (343 kg, 440 L). Salt 6 was vacuum-dried
at 30 °C for 12 h (239 kg, 28.4%, 629 mol, 100 LCAP,
98.8/1.2 DR). Salt 6 (239 kg, 629 mol, 98.8/1.2 DR) was
dissolved in EtOH (377 kg, 477 L) at 60 °C. CH3CN (1486
kg, 1906 L, 60 °C) was added over 0.3 h and the solution
was seeded with 6 (3.5 kg, 99.7/0.3 DR). The mixture
was cooled to 50 °C over 0.5 h. The heterogeneous
mixture was stirred at 50 °C for 12 h, and the following
cooling ramp to 23 °C was applied: 50 to 40 °C (3 h), 40
°C (3 h), 40 to 30 °C (1 h), 30 °C (1 h), 30 to 23 °C (1
h), 23 °C (2 h). The mixture was filtered and the solid
was rinsed with CH3CN (178 kg, 230 L). Salt 6 was
vacuum-dried at 30 °C for 12 h (198 kg, 23.2%, 521 mol,
100 LCAP, 99.8/0.2 DR). Mp 159-161 °C; 1H NMR (400
MHz, CD3OD) δ 7.51-7.64 (m, 2H), 7.28-7.48 (m, 6H),
7.20-7.27 (m, 1H), 5.17 (q, J ) 8.0 Hz, 1H), 3.00 (dd,
J ) 4.0, 8.0 Hz, 1H), 2.25-2.33 (m, 2H), 1.65-1.74 (m,
1H), 1.45-1.63 (m, 6H), 1.33-1.43 (m, 1H), 1.07-1.27
(m, 3H); 13C NMR (125 MHz, CD3OD) δ 176.6, 171.2,
145.5, 141.0, 135.3, 135.3, 131.1, 129.6, 129.4, 129.2,
128.0, 127.2, 54.9, 50.9, 41.9, 38.2, 37.4, 35.5, 28.7, 27.7,
23.0; IR (neat): 3277, 2967, 2937, 2627, 1638, 1544, 1510,
1380, 1329, 702 cm-1. A single-crystal X-ray analysis of
6 was obtained, and the CIF for this analysis is included
in the Supporting Information. Diastereomer 7 was
synthesized from R-2 and 5. It is an amorphous glassy
solid (no defined melt, XRPD available in Supporting
g, 82%). An analytically pure sample was obtained for
characterization using chromatography (silica gel, 10%
MeOH/CH2Cl2). Mp 148-152 °C; H NMR (400 MHz,
1
CD3OD) δ 7.91-7.97 (m, 1H), 7.54-7.61 (m, 1H),
7.47-7.54 (m, 1H), 7.37-7.42 (m, 1H), 3.78 (dd, J )
4.0, 8.0 Hz, 1H), 2.35-2.42 (m, 1H), 2.27 (br s, 1H),
1.70-1.80 (m, 1H), 1.44-1.62 (m, 4H), 1.14-1.35 (m,
3H); 13C NMR (125 MHz, CD3OD) δ 172.3, 169.7, 140.1,
133.0, 131.3, 131.2, 130.5, 129.1, 55.0, 43.2, 39.7, 37.1,
36.4, 29.6, 27.7; IR (neat): 3258, 2953, 1700, 1632, 1548,
1292 cm-1; Exact Mass [C15H17NO3 + 1]+: calculated )
260.1826, measured ) 260.1821.
Thiourea S-1. A solution of 6 (123.2 kg, 324.2 mol)
and Et3N (65.8 kg, 651.5 mol) in CH2Cl2 (1634 kg, 1241
L) under N2 was cooled to -10 °C. Benzoyl isothiocyanate
(12, 56 kg, 344 mol) was charged over 45 min (-10 and
-5 °C). After an additional 15 min, reaction completion
was measured (GC) and H2O (618 kg, 618 L) was added.
After separation of the layers, the organic phase was
washed with aqueous NaOH (614 kg, 4% w/w) and
aqueous HCl (615 kg, 7.3% w/w). The organic phase was
filtered. Most of the CH2Cl2 was distilled under reduced
pressure, MeOH (426 kg, 540 L) was charged, and the
solvent exchange was completed by additional distillation
of CH2Cl2. H2O (60 kg, 60 L) was charged to the thick
slurry. Aqueous NaOH (86 kg, 30% w/w) was charged
over 20 min (27 °C). H2O (1664 kg, 1664 L) was finally
added over 1 h. The suspension was cooled to 14 °C,
filtered, and the solids were washed twice with H2O (2 ×
30 L). The cake was dried at 45 °C under a N2 sweep for
3 h. The cake was transferred back in the reactor and IPAC
(177 kg, 203 L) was charged. The slurry was stirred for
3 h, cooled to 0 °C and filtered. The cake was rinsed with
IPAC (3 × 5 L) and dried at 45 °C under a N2 sweep for
3 h to afford S-1 (44.1 kg, 80.0% yield, 259.4 mol, 99 wt
1
%, 99.1 LCAP, 99.4% ee). Mp 180-182 °C; H NMR
(400 MHz, CD3OD) δ 3.97 (br s, 0.55H), 2.28 (s, 2H),
1.74-1.83 (m, 1H), 1.10-1.60 (m, 7H), 13C NMR (100
MHz, CDCl3, some signals doubled) δ 183.8, 180.3, 59.2,
58.0, 43.7, 43.0, 40.8, 40.3, 37.2, 36.4, 29.3, 27.5; IR
(neat): 3225, 2954, 1613, 1558, 1453, 716 cm-1; Exact
Mass [C8H14N2S + H]+: calculated ) 171.0956, measured
) 171.0945.
1
Information). H NMR (400 MHz, CD3OD) δ 7.54-7.64
Thiourea (()-1. To a mixture of norbornene (188.3 g,
2 mol), toluene (942 mL), water (122 mL) and NH4NCS
(159.9 g, 2.1 mol) was added 12 M aqueous H2SO4 (165
mL, 2 mol) over 25 min. The reaction mixture was stirred
at 40 °C for 3 h. The transformation was measured to be
96% complete (HPLC). IPA (188 mL) was charged and
the mixture was cooled to 22 °C. The phases were
separated and the organic layer was determined to contain
259 g of (()-15 (1.69 mol, 85% assay yield).
(m, 2H), 7.30-7.47 (m, 6H), 7.20-7.26 (m, 1H), 5.17
(q, J ) 8.0 Hz, 1H), 3.03 (dd, J ) 4.0, 8.0 Hz, 1H),
2.27-2.35 (m, 2H), 1.67-1.71 (m, 1H), 1.47-1.64 (m,
6H), 1.36-1.44 (m, 1H), 1.10-1.28 (m, 3H); 13C NMR
(125 MHz, CD3OD) δ 176.8, 171.3, 145.6, 141.0, 135.6,
135.3, 131.1, 129.6, 129.4, 129.3, 128.0, 127.3, 55.0, 51.0,
42.0, 38.3, 37.5, 35.5, 28.7, 27.7, 23.0; IR (neat): 2958,
2873, 1630, 1545, 1448, 1371, 698 cm-1
.
Amide (()-9. To a solution of phthalic anhydride (10
g, 0.067 mol) in IPA (40 mL) was added (()-2 (7.4 g,
0.067 mol) via syringe over 5 min. The mixture was stirred
for 3 h and water was added (120 mL). The mixture was
stirred for 2 h and filtered. The solid was rinsed with water
(30 mL) and dried on filter to afford amide (()-9 (14.2
Aqueous NH4OH (28-30%, 296 mL, 2.13 mol) was
charged to this solution and the temperature was increased
to 50 °C. After 20 h, the transformation was 99% complete
(HPLC). The mixture was cooled to 22 °C, held for 1 h,
and filtered through a M porosity sintered glass filter frit.
The wet cake was washed with toluene (2 × 435 mL)
140
•
Vol. 14, No. 1, 2010 / Organic Process Research & Development

and H2O (435 mL). The white solid was vacuum-dried at
45 °C (100 Torr) under a N2 sweep for 72 h to afford
(()-1 (261 g, 77% yield, 1.53 mol, 103 wt %, 100 LCAP).
port. We thank Richard Staples for performing the single-
crystal X-ray analysis included.
Supporting Information Available
1
Copies of H and 13C NMR spectra. XRPD analyses.
Acknowledgment
Crystallographic information file (CIF). This material is avail-
able free of charge via the Internet at http://pubs.acs.org.
We thank Brenda Burke for help with the preparation
of this manuscript. We thank Peter Grandsard for guidance
pertaining to the chromatographic separation of (()-1. We
thank Gary Guo, Steven Wu, Tiffany Correll, Anette
Gruesshaber-Muth, and Kelly Nadeau for analytical sup-
Received for review September 1, 2009.
OP9002328
Vol. 14, No. 1, 2010 / Organic Process Research & Development
•
141