Classical and dynamic resolution of 1-amino-3-methyl-1,3,4,5- tetrahydrobenzo[d]azepin-2-one

Article abstract of DOI:10.1016/j.tetasy.2005.10.037

Two efficient production processes of enantioenriched 1-amino-3-methyl-1,3, 4,5-tetrahydro-benzo[d]azepin-2-one 1 were achieved using the readily available starting materials. A key step in the methodologies is a classical resolution or a dynamic resolution that provides excellent chemical (>80%) yields and enantiomeric excesses (>99.8% ee). The classical resolution was developed on a preparative scale while the dynamic resolution was implemented on a pilot plant scale.

Full text of DOI:10.1016/j.tetasy.2005.10.037

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 16 (2005) 3814–3819
Classical and dynamic resolution of
-amino-3-methyl-1,3,4,5-tetrahydrobenzo[d]azepin-2-one
1
a,
a
a
b
*
David Mitchell, Lynne A. Hay, Thomas M. Koenig, Stacey McDaniel,
b
b
Jeffrey S. Nissen and James E. Audia
a
Chemical Product R&D, Lilly Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis,
IN 46285, USA
b
Discovery Chemistry Research, Lilly Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis,
IN 46285, USA
Received 12 September 2005; accepted 19 October 2005
Abstract—Two efficient production processes of enantioenriched 1-amino-3-methyl-1,3,4,5-tetrahydro-benzo[d]azepin-2-one 1 were
achieved using the readily available starting materials. A key step in the methodologies is a classical resolution or a dynamic reso-
lution that provides excellent chemical (>80%) yields and enantiomeric excesses (>99.8% ee). The classical resolution was developed
on a preparative scale while the dynamic resolution was implemented on a pilot plant scale.
ꢀ
2005 Elsevier Ltd. All rights reserved.
1. Introduction
Use of conformationally constrained building blocks in
H N
2
1
–3
drug design has gained popularity in recent years. As
a conformationally restricted phenyl glycine analog, a
structure–activity relationship (SAR) investigation
required preparation of (S)-1-amino-3-methyl-1,3,4,5-
tetrahydro-benzo[d]azepin-2-one 1 as a pure enantiomer
O
N
Me
(S)-1
Figure 1.
(
Fig. 1). In the early stages of development, 1 was pre-
pared and utilized as a racemate. This racemate utiliza-
tion created problems for the overall synthesis efficiency
from a practical perspective. The chromatographic
separation of targets prepared from 1 led to an overall
low yield that contributed to a limitation for scale up.
Although other separation techniques such as chro-
2
. Results and discussion
2
.1. Classical resolution
Preparation of 1 as a racemate (RS)-1 is outlined in
Scheme 1. Starting with phenethyl bromide 2, the N-
4
,5
matographic moving bed separations or conglomerate
7
6
separations may have been very useful, we focused on a
methyl moiety was introduced by a direct aminolysis uti-
lizing methylamine in a THF–water mixture. Greater
than 12 equiv of methylamine were required to circum-
vent further reactions of 3 with the alkyl bromide starting
material. Compound 3 was acylated with chloroacetyl
dynamic resolution process that offered higher efficiency
by converting the racemate to a single non-racemic
chiral entity.
Herein, we report our efforts to prepare (S)-1 by classi-
cal resolution. The information generated by the classi-
cal resolution studies was then applied to a dynamic
resolution process.
8
chloride under Schotten-Baumann conditions to give 4.
9
A Friedel–Crafts alkylation of 4 provided the cyclic ben-
zoazepine 5. Reaction of benzoazepine 5 with a base such
as lithium hexamethyldisilazane (HMDS) in THF
followed by quenching with isoamyl nitrite gave hydrox-
ylamine 6. Racemic (RS)-1 was obtained by hydrogena-
tion using 10% palladium on carbon in a mixture of
*
Corresponding author. Tel.: +1 317 276 1384; fax: +1 317 276
507; e-mail: dmit@lilly.com
4
0957-4166/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2005.10.037

D. Mitchell et al. / Tetrahedron: Asymmetry 16 (2005) 3814–3819
3815
Recycling of (R)-1, that is converting (R)-1 to (RS)-1
and further additional resolution was also evaluated.
Heating (R)-1 in triethylamine for 18 h provided (RS)-
b
a
Cl
1
. Further development indicated that the process was
N
O
catalytic in triethylamine. For example, 10% molar of
triethylamine and a full equivalent of (S)-1 in methanol
at 60 ꢁC (reflux) for 18 h provided racemic (RS)-1 (0.3%
ee) in >98% yield. Therefore, a process was developed
whereby the classical resolution coupled with the recy-
cling methodology provided a >50% theoretical yield
of (S)-1Æddta.
Br
NHMe
Me
2
3
4
c
d
e
HO
H N
2
N
Knowing that an enantioenriched mixture of (R)-1
could be racemized fairly readily, a dynamic resolu-
N
N
O
O
N
12–14
O
tion
procedure that would combine the resolution
Me
Me
Me
and the racemization of (R)-1 into one step was investi-
gated. The methodology would ultimately convert the
racemate into a single isomer as the diastereomeric salt.
Using this technique, the theoretical yield of the resolu-
tion step would increase two-fold, from 50% to 100% of
(
RS)-1
5
6
Scheme 1. Preparation of (RS)-1 from phenethyl bromide. Conditions
and reagents: (a) CH
chloride, NaHCO , CH
isoamyl nitrite, LiHMDS, THF, 84%; (e) H
2%.
3
NH
2
, H
, 95%; (c) AlCl
, 10% Pd/C, EtOH/HCl,
2
O/THF (9:1), 95%; (b) chloroacetyl
3
2
Cl
2
3
, 2-Cl , 71%; (d)
2 6 4
C H
(
S)-1. This approach relies on the ability to preferen-
2
tially crystallize the desired diastereomeric salt, prevent-
ing this entity from being subjected to racemization,
while simultaneously keeping the undesired isomer in
solution so that it can be racemized. Therefore, identifi-
cation of an appropriate resolving agent and solvent
system combination that provides the selective precipita-
tion at a temperature high enough to allow racemization
was required. This process, a dynamic resolution, is well
8
dtta
.
a
H N
H N
2
(
RS)-1
2
+
N
N
O
O
12
documented in the literature.
Me
Me
(
S)-1.dtta
(R)-1
Although it was demonstrated that a base, such as tri-
ethylamine, racemizes (R)-1 in refluxing methanol, we
needed to find another means to do this because an
added amine base for racemization has the potential to
compete with the chiral acid in salt formation. Litera-
ture methods for dynamic resolutions utilize catalytic
amounts of non-enenolized aldehydes, which form the
imine (Schiff bases) with the amine. The modification
enhances the acidity of the a proton, promoting racemi-
zation under much milder conditions. Therefore, an
appropriate aldehyde that would promote facile racemi-
zation of the amine at a temperature as close to ambient
as possible (to enhance precipitation of the desired salt)
was required.
Scheme 2. Classical resolution of (RS)-1 from di-p-toluoyl-L-tartaric
acid (L-dtta). Conditions and reagents: (a) L-dtta, CH OH, 25%.
3
ethanol–hydrochloric acid in 44% overall yield from 2
Scheme 2).
(
With limited material available for practical develop-
ment, a brief evaluation of potential chiral resolv-
ing acids indicated that di-p-toluoyl-L-tartaric acid
1
0,11
(
L-dtta)
gave increased isomeric purity. Optimiza-
tion of this classical resolution included methanol as
solvent, 1 equiv of acid and the amine at reflux. Upon
cooling, the salt was stirred for 18 h at ambient temper-
ature then filtered to obtain 1 as the L-dtta salt in 96%
chemical yield (48% yield overall) and 95% ee.
2.2. Dynamic resolution
We began the investigation into the dynamic resolution
technique by using our classical resolution data as a
starting point. In that study, we had screened several
resolving agents in various solvent combinations to
determine which formed solid salts in high yield and dia-
stereomeric excess. In addition to L-dtta and (R)-10-
camphorsulfonic acid, (R)-(ꢀ)-mandelic acid, was also
investigated. Tartaric acid tended to form semi-solids
with (RS)-1 under the conditions tried. Of these acids,
(R)-(ꢀ)-mandelic acid formed salts in the highest yields
and diastereomeric excess upon process optimization.
Since L-dtta is a di acid, we considered using half an
equivalent of the acid. This approach would allow for
less reagent and increase the efficiency of the overall pro-
cess. When 0.5 equiv of the diacid L-dtta was utilized,
the result was an amorphous solid that was difficult to
handle. Analysis of the solid indicated high enrichment
of (S)-1Ædtta; however, isolation of the material was
not practical for preparative scale (>100 g). Therefore,
we continued to use a full equivalent of the resolving
agent, thus avoiding product isolation problems. The
L-dtta salt was liberated using a basic extractive proce-
dure. For example, the salt is stirred at room tempera-
An aldehyde, salicylaldehyde, was also added to the salt
formation screen in an attempt to investigate racemiza-
tion of the soluble enantiomer via Schiff base formation.
ture in 1.0 N NaOH and extracted with CH Cl2 to
2
provide (S)-1 in high recovery and purity.

3816
D. Mitchell et al. / Tetrahedron: Asymmetry 16 (2005) 3814–3819
For example, using mandelic acid as the resolving agent
provided 98.1% ee (S)-1 in 60% chemical yield using 10
OH
HCl
OH
.
.
%
molar salicylaldehyde in isopropyl acetate and 2-pro-
panol at 50 ꢁC. The filtrate from this reaction showed
.1% ee, clearly indicating that racemization was occur-
a
b
H N
2
O
H N
2
0
N
N
O
.
S)-1 (R)-mandelic acid
O
12
ring. We then turned our attention to improving the
chemical yield of the process primarily through choice
of aldehyde and solvent ratios.
(
Me
Me
Scheme 3. Dynamic resolution of (RS)-1 with mandelic acid 11.
Reagents and conditions: (a) (R)-(ꢀ)-mandelic acid, 5-nitrosalicyl
By reacting (R)-1 (77% ee) in 2-propanol with various
aldehydes and at various temperatures, we were able
to determine which aldehydes were the most promising
for racemization (Table 1). These results show that 5-
nitrosalicylaldehyde and 2,4-dichlorosalicylaldehyde
readily racemize the amine between 22 and 55 ꢁC. Inter-
estingly, heating the amine (R)-1 in 2-propanol at reflux
for 21 h causes significant racemization with no alde-
hyde present. It was found that this resolution is best
conducted at 40–50 ꢁC to keep the diastereomeric excess
of the precipitating salt high, as well as to speed the race-
mization process.
2 2
aldehyde, 83%; (b) 1.0 M NaOH, CH Cl .
Table 2. Conversion of chiral salt to free amine
11,
(% ee)
Base
pH
T
Recovery 1,
(ꢁC)
(%)
(% ee)
9
9
7.6
8.4
NaOH (1.0 N)
NaOH (1.0 N)
NaOH (1.0 N)
13–14 22
13–14
13–14 40 (1 h)
78
82
2
42
69
91.9
84.9
71.4
61.7
48.0
0
98.4
98.4
98.4
NaHCO
NaHCO
3
(satd)
(1.0 N) 8–9
9–10
22
22
3
Ultimately, the procedure developed utilized slightly less
than an equivalent of (R)-(ꢀ)-mandelic acid as the
resolving agent dissolved in a 7:3 isopropyl acetate/
OH
2-propanol mixture at 50 ꢁC for 14 h, with 5 mol %
5-nitrosalicylaldehyde. This method provided the man-
HCl
OH
.
2
.
a
H N
O
H2N
delate salt of (R)-1 in 83% yield and (R)-1 in 98.4% ee
on a preparative scale [30 g of (RS)-1] (Scheme 3).
Because making the free-base from the di-p-toluoyl-L-
tartrate salt was so facile, we did not anticipate a
problem with obtaining the free-base from (S)-1Æ(R)-
mandelic acid. However, using 1.0 M NaOH and
CH Cl to free-base the mandelate salt provided mate-
N
Me
N
Me
O
O
.
(S)-1.mandelic acid
(S)-1 HCl
Scheme 4. Direct conversion to the HCl salt. Conditions and reagents:
a) HCl (anhydrous)/EtOAc, or HCl (concentrated)/EtOAc, 92%.
2
2
(
rial with significantly lower enantiomeric excess (Scheme
). Therefore, we investigated the conditions for produc-
3
ing the free-base (S)-1Æ(R)-mandelic acid with regard to
the ee of (R)-1 (Table 2).
no improvement. Therefore, this (S)-1Æ mandelic acid
salt was converted directly into the HCl salt, (S)-1ÆHCl
(Scheme 4).
It is suspected that a small amount of the Schiff base is
still present in the diastereomeric salt as it is filtered
from the reaction mixture, which contributes to the
rapid racemization of the amine upon contact with base.
Unfortunately, this imine is not simple to remove or
hydrolyze, as the mandelate salt is very soluble in water,
making an aqueous wash of the solids impractical. Even
adding 3% water to the dynamic resolution lowers the
yield from ꢁ85% to ꢁ70%. Reslurrying the salt in iso-
propyl acetate prior to generating the free-base gave
This transformation can be performed with anhydrous
HCl/EtOAc or concentrated HCl/EtOAc, both methods
providing (S)-1ÆHCl in 92% yield and no change in enan-
tiomeric excess. Not only does this method provide
material in higher enantiomeric excess than the basic
work-up, it also provides an additional purification of
this intermediate, as the crystalline HCl salt is filtered
from the EtOAc/water mixture. Any remaining Schiff
base is likely to be cleaved during the salt formation
and the aldehyde is removed to provide uncontaminated
Table 1. Comparison of non-enenolizable aldehyde in dynamic
(
S)-1ÆHCl. Additionally, this salt is much easier to han-
a
resolution
dle than the free base, which is typically a viscous oil.
Furthermore, this method allows for the facile recovery
of the chiral resolving agent, which can be collected
from the filtrate of (S)-1ÆHCl.
Aldehyde
% ee at
% ee at % ee at
2
2 ꢁC, 21 h 55 ꢁC, 6 h 83 ꢁC, 17 h
5
3
-Nitrosalicylaldehyde
,5-Dichlorosalicylaldehyde
0.2
2.4
—
0.0
—
—
Salicylaldehyde
,4-Dichlorobenzaldehyde
None
65
73
—
14
60
—
5.4
12
11
Assignment of the configuration of the stereogenic cen-
2
ter was determined by subsequent reaction of 1 with a
1
5
peptide residue and by X-ray structure evaluation.
a
The single X-ray crystal structure of (S)-1ÆHCl (Fig. 2)
also confirms the stereochemistry. In this structure, the
benzylic stereogenic center has the (S)-configuration.
Reaction conditions: The chiral amine (R)-1 (77% ee) was dissolved in
-propanol with 10 mol % aldehyde and stirred at the designated
temperature; % ee determined by CE analysis.
2

D. Mitchell et al. / Tetrahedron: Asymmetry 16 (2005) 3814–3819
3817
was separated and the aqueous portion further extracted
with MTBE (95 mL). Water (189 mL) was added to the
combined extracts and the pH adjusted to 6–6.5 with
HCl (5 N). The aqueous portion was isolated and
diluted with MTBE (94.5 mL) before adjusting the pH
to 10–10.5 with sodium hydroxide (5 N). The resulting
organic portion was separated and the aqueous portion
extracted with MTBE (94.5 mL). The combined extracts
2
containing 3 were used as a MTBE solution in the next
step.
4.1.2. 2-Chloro-N-methyl-N-(2-phenylethyl)acetamide 4.
A solution of sodium bicarbonate (9.2 g, 110 mol) in
water (100 mL) was added to a MTBE (95 mL) solution
of compound 3 cooled at 0–10 ꢁC. Chloroacetyl chloride
Figure 2. X-ray crystal structure of (S)-1ÆHCl.
(
11.04 g, 0.096 mol) dissolved in MTBE (17 mL) was
added to the solution of 3 dropwise. After stirring for
3. Conclusion
1
h at 10 ꢁC, the reaction was complete (based on HPLC
analysis). Extractive work-up using MTBE as solvent
Although significant improvements to the classical reso-
lution of (RS)-1 with L-dtta were accomplished, it was a
short-term solution for the facile preparation of (S)-1. A
mild method to enantiomeric-enriched material was
observed during the classical resolution study. Even
recycling the undesired isomer from the classical resolu-
tion through a mild racemization process and subse-
quent resolution was not ideal as it adds several
operation steps.
followed by concentration provided compound 4 as an
oil (14.52 g, 67.1% from 2). H NMR (300 MHz,
1
DMSO-d ) d 2.73–2.87 (m, 2H), 2.87+2.95 (s+s, 3H),
6
3
5
3
1
.47–3.54 (m, 2H), 4.22+4.32 (s+s, 2H), 7.20–7.31 (m,
13
H); C NMR (75 MHz, DMSO-d ) d 32.70, 33.32,
6
3.78, 35.18, 49.31, 50.87, 126.10, 126.37, 128.29,
28.38, 128.60, 128.85, 138.49, 138.91, 165.52, 165.60;
MS m/z (rel intensity) 212.10 (100%, M+H).
4
.1.3. 3-Methyl-1H,4H,5H-benzo[d]azaperhydroepin-2-
The ideal method to convert racemic (RS)-1 to the pure
isomer by use of a dynamic resolution or classical reso-
lution was successfully developed. This method provides
the HCl salt of (S)-1 in high yields and excellent enantio-
meric excess, while saving reagents, time, effort and dou-
bling the theoretical yield of the classical resolution.
Additionally, this method allows for the facile recovery
of the homochiral mandelic acid that may be recycled.
one 5. Aluminum chloride (23.60, 0.177 mol) was
added to a solution of compound 4 in 1,2-dichlorobenz-
ene (105 mL) which exothermed from 20 to 25 ꢁC. The
resulting reaction mixture was heated to 165 ꢁC for 2 h
and then cooled (0–5 ꢁC) before adding the reaction
mixture (in small portions) to a cooled (0–5 ꢁC) HCl
solution (150 mL, 1 N). The organic portion was sepa-
rated and the aqueous portion extracted with CH Cl
2
2
(45 mL). HCl (150 mL, 1 N) was used to wash the com-
bined extracts followed by a saturated sodium bicarbon-
ate (150 mL) wash. The organics were then placed on a
silica gel pad (75 g, 150 mL CH Cl ) and eluted first
4. Experimental
2
2
4
.1. General
with CH Cl followed by ethyl acetate. All fractions
2
2
containing compound 5 were combined and concen-
All solvents and reagents were used as obtained from
commercial sources and used without further purifica-
tion. Melting points were measured on a Thomas Hoo-
ver Capillary Melting Point apparatus and are
trated to provide 8.08 g, (65.2%) of product: mp
1
1
3
24.4 ꢁC; H NMR (300 MHz, DMSO-d ) d 2.87 (s,
6
H); 3.04 (t, J = 5.8 Hz, 2H); 3.69 (t, J = 5.8, 2H),
3.82 (s, 2H); 7.13–7.48 (m, 4H). C NMR (75 MHz,
1
3
1
uncorrected. H NMR spectra were recorded at
DMSO-d ) d 31.0, 33.4, 42.2, 47.2, 126.0, 126.4, 130.1,
6
3
00 MHz on a Bruker-300 spectrometer in a DMSO-d6
130.3, 132.1, 136.4, 170.8; MS m/z (rel intensity)
solution with tetramethylsilane as an internal standard.
1
3
176.10 (100, M+H).
C NMR spectra were recorded at 75 MHz on a Bru-
ker-300 spectrometer in a DMSO-d solution. Chiral
6
analysis was performed using an Applied Biosystems
4.1.4. 1-(Hydroxyimino)-3-methyl-4H,5H-benzo[d]aza-
perhydroepin-2-one 6. A THF (6.13 L) solution of
compound 5 (613 g, 3.5 mol) was cooled to 0 ꢁC and iso-
amyl nitrite (550 g, 4.55 mol) was added dropwise. To
the resulting mixture was added lithium bis(trimethyl-
silyl) amide (4.5 L, 4.5 mol, 1 N THF) at a rate such that
the temperature remained <10 ꢁC. After addition, the
reaction was allowed to stir at room temperature for
2–3 h while monitoring for the reaction progress by
HPLC. Upon completion of the reaction, the mixture
was cooled to 0 ꢁC, and the pH adjusted from 12 to
2
70A-HT Capillary Electrophoresis System.
4
.1.1. Methyl-2-phenylethylamine 3. To a THF (20 mL)
and methylamine (189 mL, 120.5 g, 2.2 mol, 40% in
water) solution at room temperature was added a solu-
tion of 2-phenylethyl bromide 2 (18.9 g, 0.102 mol) in
THF (80 mL) over a 1 h period allowing the tempera-
ture to exotherm (>30 ꢁC). After 4 h, the reaction was
complete as judged by HPLC analysis and the mixture
diluted with MTBE (189 mL). The organic portion

3818
D. Mitchell et al. / Tetrahedron: Asymmetry 16 (2005) 3814–3819
2
–3 using HCl (2 N). The resulting precipitate was stir-
135.53, 143.59, 164.73, 167.59, 167.90; Anal. Calcd for
C H N O : C, 64.57; H, 5.59; N, 4.86; O, 24.97.
red for 12–16 h before isolation by filtration and drying
3
1
32
2
9
to provide 5 as a crystalline solid (633 g, 89%):
Found: C, 64.72; H, 5.58; N, 4.72.
1
mp > 230 ꢁC; H NMR (300 MHz, DMSO-d ) d 2.94
6
(
2
s, 3H), 3.06 (t, J = 5.12 Hz, 2H), 3.61 (t, J = 5.12 Hz,
4.1.7.
(1S)-1-Amino-3-methyl-1H,4H,5H-benzo[d]aza-
1
3
H), 7.2–7.4 (m, 4H), 11.6 (s, 1H);
C NMR
perhydroepin-2-one (S)-1. Compound 1 was obtained
from mandelate salt (S)-1Æmandelic acid (11.83 g,
20.5 mmol) by dissolving (S)-1Æmandelic acid in NaOH
(45 mL, 1 N) and extracting with methylene chloride
(3 · 25 mL). The combined organic layers were washed
with NaOH (35 mL 1.0 N) then brine solution
(
75 MHz, DMSO-d ) d 31.04, 31.09, 32.33, 46.29,
6
4
1
1
6.48, 125.22, 126.21, 126.57, 128.42, 129.04, 129.23,
29.64, 130.30, 130.46, 130.50, 136.44, 137.38, 152.26,
53.03, 164.72, 165.66; MS m/z (rel intensity) 205.10
(
100, M+H); Anal. Calcd for C H N O : C, 64.69;
11 12 2 2
H, 5.92; N, 13.71. Found: C, 64.86; H, 6.01; N, 13.64.
(20 mL), and dried over anhydrous MgSO . Removal
4
of solvent under vacuum provided (S)-1 (3.38 g, 87%
2
D
5
4
.1.5. 1-Amino-3-methyl-1H,4H,5H-benzo[d]azaperhy-
yield, 93.2% ee): mp 98–99 ꢁC; ½aꢂ ¼ þ54:9 (c 0.5,
1
droepin-2-one (RS)-1. An ethanol (525 mL) solution
of compound 6 (35 g, 0.171 mol) was added to a glass
lined autoclave along with palladium on carbon (10%,
MeOH), H NMR (300 MHz, DMSO-d ) d 2.06 (br s,
2H), 2.90 (s, 3H), 3.09 (t, J = 5.1 Hz, 2H), 3.34 (dt,
J = 15.0, J = 4.8, 1H), 4.1–4.3 (m, 1H), 5.27(s, 1H),
7.0–7.2 (m, 3H), 7.70–7.73 (m, 1H);
6
1
3
3
.5 g) as a dilute HCl (concentrated, 17.5 g in 17 mL
C NMR
water) slurry. The resulting mixture was hydrogenated
at 50 ꢁC and 250 psi of hydrogen until the reaction
was completed (24 h). The reaction mixture was filtered
(75 MHz, DMSO-d ) d 31.00, 34.16, 46.64, 52.19,
6
124.82, 125.67, 126.44, 129.80, 135.28, 138.19, 173.67;
MS m/z (rel intensity) 191.11 (100, M+H); Anal. Calcd
for C H N O: C, 69.45; H, 7.42; N, 14.73. Found: C,
ꢂ
over a pad of Celite using ethanol as solvent and the
1
1
14
2
filtrate concentrated to 90 mL. Water (350 mL) was
added to the concentrate and the resulting solution fur-
ther concentrated to 200 mL. The pH of the mixture was
then adjusted to 11–11.5 with sodium hydroxide (1 N)
followed by extraction with dichloromethane
68.99; H, 7.46; N, 14.47.
4.1.8. Racemization of (1R)-1-amino-3-methyl-1H,
4H,5H-benzo[d]azaperhydroepin-2-one. (R)-1 (6.293 g,
33.1 mmol, 64.5% ee) was heated (50 ꢁC) in methanol
(30 mL) to form a solution. Triethylamine (0.46 mL,
3.31 mmol, 10 mol %) was added to the solution and
the mixture heated to reflux for 16 h then cooled to
ambient temperature. The mixture was then concen-
trated under vacuum to obtain 7 (6.2 g, 98.4% yield,
0.3% ee).
(
350 mL). The organic portion was separated and the
aqueous portion extracted with CH Cl (175 mL). The
2
2
combined extracts were concentrated to provide com-
pound (RS)-1(27.3 g, 84%). This material was used with-
1
out purification in the next step: mp = 69–70 ꢁC; H
NMR (300 MHz, DMSO-d ) d 2.06 (br s, 2H), 2.90 (s,
6
3
H), 3.09 (t, J = 5.1 Hz, 2H), 3.34 (dt, J = 15.0,
J = 4.8, 1H), 4.1–4.3 (m, 1H), 5.27 (s, 1H), 7.0–7.2 (m,
4.1.9.
(1S)-1-Amino-3-methyl-1H,4H,5H-benzo[d]aza-
1
3
3
H), 7.70–7.73 (m, 1H); C NMR (75 MHz, DMSO-
perhydroepin-2-one hydrochloride (S)-1ÆHCl. A solu-
tion of compound (RS)-1 (27.7 g, 0.146 mol) and
isopropyl acetate (100 mL) heated to 45 ꢁC was added
to 2-propanol (150 mL) and (ꢀ)-mandelic acid (21.72 g,
0.143 mol) solution at 45 ꢁC. The combined solutions
were stirred for 3–4 h at 45 ꢁC before the addition of
d6) d 31.00, 34.16, 46.64, 52.19, 124.82, 125.67, 126.44,
1
1
6
1
29.80, 135.28, 138.19, 173.67; MS m/z (rel intensity)
91.11 (100, M+H); Anal. Calcd for C H N O: C,
1
1
14
2
9.45; H, 7.42; N, 14.73. Found: C, 69.95; H, 7.21; N,
4.41.
5
-nitrosalicylaldehyde (1.22 g, 7.3 mmol). The resulting
4
.1.6.
(1S)-1-Amino-3-methyl-1H,4H,5H-benzo[d]aza-
reaction mixture was then stirred at 45 ꢁC for 12–14 h
and cooled to ambient temperature. After cooling to
room temperature, the resulting crystalline mandelate
salt was collected by filtration and the crystals washed
with ethyl acetate (2 · 50 mL). The mandelate salt of
(S)-1 was then added to ethyl acetate (250 mL) and
heated to 50 ꢁC. At that temperature, HCl (concen-
trated, 22.7 g, 0.233 mol) was added to the heteroge-
neous mixture and the resulting mixture stirred for 3–
4 h at 50 ꢁC to ensure a solid to solid salt exchange from
mandelate to hydrochloride. The mixture was then
cooled to ambient temperature and the resulting HCl
salt was collected by filtration, washed with ethyl acetate
(2 · 50 mL) and dried in the vacuum oven to provide
(S)-1ÆHCl (23.7 g, 61.4%, 97.9% ee): mp = >230 ꢁC;
perhydroepin-2-one di-p-toluoyl-L-tartrate (S)-1Ædtta.
mixture of (RS)-1 (1.5 g, 8.12 mmol) in MeOH
15 mL) was heated to form a solution (50 ꢁC, 30 min).
In another flask, di-p-toluoyl-L-tartaric acid (3.12 g,
.08 mmol) was dissolved in MeOH (45 mL) and added
to the warm solution of (RS)-1. After refluxing for
0 min, the solution was allowed to cool to ambient tem-
A
(
8
3
perature and stirred at that temperature for 16 h. The
resulting crystals that formed were collected by filtration
and rinsed with cold methanol (10 mL) to provide (S)-
1
Ædtta (2.24 g) in 96% yield, 94.7% ee: mp 196–197 ꢁC.
25
D
½
aꢂ ¼ ꢀ71. (c 10, MeOH); IR (KBr) 3450, 3251, 2949,
1
7
712, 1683, 1612, 1516, 1407, 1269, 1179, 1111, 1021,
47, 695 cm ; H NMR (300 MHz, DMSO-d ) d 2.36
ꢀ1 1
6
2
5
1
(
4
(
s, 6H), 2.89 (s, 3H), 3.2–3.0 (m, 1H), 3.3–3.2 (m, 2H),
.2–4.15 (m, 1H), 5.15 (s, 2H), 5.85 (s, 1H), 7.35–7.0
m, 4H), 7.29 (d, J = 8.05 Hz, 4H), 7.81 (d, J = 8.4 Hz,
½aꢂ ¼ þ54:9 (c 0.5, MeOH); H NMR (300 MHz,
D
DMSO-d ) d 2.90 (s, 3H), 3.01–3.20 (m, 1H), 3.20–3.4
6
(m, 2H), 4.15–4.3 (m, 1H), 5.94 (s, 1H), 7.2–7.4 (m,
1
3
13
4
H), 10.5–8.5 (br s, 2H); C NMR (75 MHz, DMSO-
4H), 9.15 (br s, 3H); C NMR (75 MHz, DMSO-d ) d
6
d6) d 21.07, 30.46, 34.40, 47.40, 52.02, 71.55, 123.27,
26.08, 126.80, 127.80, 129.11, 129.21, 130.21, 131.35,
30.46, 34.48, 47.49, 52.15, 123.71, 126.09, 127.99,
130.26, 130.40, 135.66, 166.73; MS m/z (rel intensity)
1

D. Mitchell et al. / Tetrahedron: Asymmetry 16 (2005) 3814–3819
3819
1
5
1
91.11 (100, M+H); Anal. Calcd for C H ClN O: C,
8.27; H, 6.66; N, 12.35. Found: C, 57.90; H, 6.48; N,
2.15.
chemistry scale-up and to Dr. Gregory Stephenson for
X-ray structure determination.
1
1
13
2
References
4
.1.10. (1S)-1-Amino-3-methyl-1H,4H,5H-benzo[d]aza-
perhydroepin-2-one (ꢀ)-mandelic acid salt. The racemic
amine 7 (31.9 g, 168 mmol) was slurried in 302 mL iso-
propyl acetate and heated to 45 ꢁC (mixture was still a
slurry at this temperature). In a separate flask, (R)-
1
2
3
. The Practice of Medicinal Chemistry; Wermuth, C. G.,
Ed.; Academic Press: San Diego, 1996.
. Drug Design for Neuroscience; Kozikowski, A., Ed.;
Raven Press: New York, 1993.
. Silverman, R. B. The Organic Chemistry of Drug Design
and Drug Action; Academic Press: San Diego, 1992.
(
ꢀ)-D-mandelic acid (25.0 g, 164 mmol) was heated in
1
29 mL 2-propanol until a solution formed. While
warm, this solution was added to the amine/Isopropyl
acetate slurry. (This mixture provides a 13.5 vol. [mL
per gram of (RS)-1] of 7:3 isopropyl acetate/2-propa-
nol). A solution formed upon addition of the MA/2-
propanol mixture, but a precipitate quickly formed
thereafter. The mixture was stirred at 45 ꢁC for ꢁ3 h.
4. Huthmann, E.; Juza, M. J. Chromatogr. A 2005, 1092, 24–
35.
5. Molnar, Z.; Nagy, M.; Aranyi, A.; Hanak, L.; Argyelan,
J.; Pencz, I.; Szanya, T. J. Chromatogr. A 2005, 1075, 77–
8
6.
6
7
. Grabowski, E. J. J. Chirality 2005, 17, S249–S259.
. Repke, D. B.; Ferguson, W. J.; Bates, D. K. Tetrahedron
Lett. 1979, 43, 4183–4184.
5
-Nitrosalicylaldehyde (2-hydroxy-5-nitrobenzaldehyde)
(
1.40 g, 8.38 mmol, 5 mol %) was added to the warm
solution and the mixture was stirred at 45 ꢁC overnight
14 h). The slurry was cooled to room temperature and
8
2 3
. The Shotten-Baumann conditions of MTBE and Na CO ,
provided a much facile reaction work-up compared to a
pure organic solvent reaction.
(
stirred for 2 h. The solids were filtered, rinsing with cold
Isopropyl acetate (70 mL). The solid was dried in the
vacuum oven at 40 ꢁC to obtain 46.62 g (82.9% yield)
of (S)-1Æmandelic acid in 98.4% ee.
9. High temperature is required for the Friedel–Crafts
alkylation to take place.
10. For example: Acs, M.; Szili, T.; Fogassy, E. Tetrahedron
Lett. 1991, 32, 7325–7328.
1
1. For Example: Schaus, J. M.; Huser, D. L.; Titus, R. D.
Synth. Commun. 1990, 20, 3553–3562.
4
.1.11. (1S)-1-Amino-3-methyl-1H,4H,5H-benzo[d]aza-
1
2. Reider, P. J.; Davis, P.; Hughes, D. L.; Grabowski, E. J. J.
J. Org. Chem. 1997, 52, 957–958.
perhydroepin-2-one hydrochloride (S)-1ÆHCl. The (ꢀ)-
mandelic acid salt of (S)-1 (2.42 g, 7.06 mmol, 98.4%
ee) was slurried in EtOAc (25 mL) at room temperature.
Concentrated HCl (1.1 mL, ꢁ11.2 mmol) was added
and the mixture was heated to 50 ꢁC with vigorous stir-
ring for 3.5 h. The slurry was cooled to room tempera-
ture and filtered, rinsing the solids with 10 mL MTBE
to obtain (S)-1ÆHCl 1.48 g (92.5% yield) of the purified
HCl salt in 97.9% ee.
13. Colson, P.-J.; Przybyla, C. A.; Wise, B. E.; Babiak, K. A.;
Seaney, L. M.; Korte, D. E. Tetrahedron: Asymmetry
1
998, 9, 2587–2593.
1
1
4. Ebbers, E. J.; Ariaans, G. J. A.; Houbiers, J. P. M.;
Bruggink, A.; Zwanenburg, B. Tetrahedron 1997, 53,
9
417–9476.
5. Single crystals were mounted on a thin glass fiber and
immersed in a stream of nitrogen at 100 K. Data were
˚
collected using a MoK radiation source (k = 0.71073 A)
a
and a P4 diffractometer equipped with a SMART 1000
CCD area detector. Cell refinement and data reduction
were performed using the SAINT program. Systematic
conditions suggested the space group selections. The
structures were solved by direct methods. All atomic
parameters were independently refined. The space group
choice was confirmed by successful convergence of
the full-matrix least-squares refinement on F . Crystallo-
graphic data have been deposited with the Cambridge
Crystallographic Data Centre: CCDC 287358.
Acknowledgements
We appreciate the efforts of Mr. Duane Pierson and Mr.
Paul Dodson who conducted the CE analysis of the
many lots of chiral and racemic amine that this work
generated. We are also grateful to Mr. Guy Hansen,
Michael A. Carr and Rick Miller for the process
2