A practical procedure for the resolution of (+)- and (-)-tramadol

Article abstract of DOI:10.1021/op000281v

A practical procedure for the efficient resolution of cis-tramadol [cis-2-(dimethylaminomethyl)-1-(3-methoxyphenyl)cyclohexanol] has been developed. This process was based on the observation that cis-tramadol free base selectively formed mandelic acid salts at different rates, affording a readily scalable kinetic resolution of each enantiomer. The key to the process was the observation that the resolving salt needed to be broken and re-formed to ultimately improve the optical purity. The mandalate salt of each cis-enantiomer was found to be > 99% optically pure after three cycles through the salt formation process. A sample of each mandelate salt enantiomer was successfully converted to the known, optically active hydrochloride salt.

Full text of DOI:10.1021/op000281v

Organic Process Research & Development 2000, 4, 291−294
Technical Notes
A Practical Procedure for the Resolution of (+)- and (-)-Tramadol
Zinovy Itov and Harold Meckler*
Chemical DeVelopment Department, Albany Molecular Research, Inc., 21 Corporate Circle, Albany New York 12203
Abstract:
A practical procedure for the efficient resolution of cis-tramadol
[
cis-2-(dimethylaminomethyl)-1-(3-methoxyphenyl)cyclohex-
anol] has been developed. This process was based on the
observation that cis-tramadol free base selectively formed
mandelic acid salts at different rates, affording a readily scalable
kinetic resolution of each enantiomer. The key to the process
was the observation that the resolving salt needed to be broken
and re-formed to ultimately improve the optical purity. The
mandalate salt of each cis-enantiomer was found to be >99%
optically pure after three cycles through the salt formation
process. A sample of each mandelate salt enantiomer was
successfully converted to the known, optically active hydro-
chloride salt.
Figure 1.
The synthesis and purification of the racemic material was
6
4
described in 1978. A later article described the resolution
of both enantiomers by the use of dibenzoyl-D-tartaric acid.
Enantiomers were isolated from the racemic salt by a tedious
fractional digestion, followed by fractional crystallization.
While an isolated yield was provided for the preparation of
the racemate salt, no yields were provided for the ultimate
isolation of the resolved salts and subsequent conversion back
to the respective hydrochlorides. Later, a brief description
was published for the resolution of cis-tramadol by fractional
Introduction
Tramadol [2-(dimethylaminomethyl)-1-(3-methoxyphen-
yl)cyclohexanol] is a mild, non-addictive, centrally acting
binary analgesic agent, introduced in Germany in the late
1
1
970s by Grunenthal. It was approved for use in the United
7
crystallization with mandelic acid, but no experimental
States in 1995 and is currently marketed as Ultram by Ortho-
McNeil Pharmaceuticals, Inc. While the marketed form is
procedures were provided. Our material requirements for the
resolved crystalline salt (the acid component was not
particularly specified or required) were to be >99% optically
pure (by chiral HPLC analysis) and to match the literature
data of both corresponding hydrochloride salt isomers.
2,3
the racemic hydrochloride salt of the cis-isomer (separated
from the minor, trans-isomer2 by crystallization of the
hydrochloride salt during the production process), it is known
that the (+)-enantiomer of the cis-isomer exhibits analgesic
,3
4
,5
activity 10-fold higher than that of the (-)-enantiomer.
Discussion
Recently we required a ready supply of both enantiomers of
cis-tramadol. While a chiral synthesis of each enantiomer
was a tantalizing challenge, the quickest method to produce
appreciable quantities of each enantiomer appeared to be the
classical resolution of the readily prepared racemate (Figure
In our hands, the method of Frankus and co-workers4
readily afforded the racemate salt with dibenzoyl-D-tartaric
acid. However, all attempts to separate the enantiomers by
sequential digestion of the solids in ethyl acetate (to dissolve
the (+)-enantiomer), followed by digestion of the remaining
solids in 2-propanol (to dissolve the (-)-enantiomer) failed.
Tramadol isolated by this procedure was, at best, only slightly
enriched in either enantiomer, and this method was judged
impractical for the preparation of the required quantities of
each resolved isomer.
1).
*
To whom correspondence should be addressed. Telephone: (518) 464-0279.
Fax: (518) 464-0289. E-mail: hmeckler@albmolecular.com.
(1) (a) Flick, K.; Frankus, E. U.S. Patent 3,652,589, March 28, 1972; Chem.
Abstr. 1972, 76, 153321. (b) Flick, K.; Frankus, E. U.S. 3,830,934, August
2
0, 1974; Chem. Abstr. 1974, 82, 21817 (both to Grunenthal G.m.b.H.).
(
2) The biologically active isomer which was developed by Grunenthal was
While the mandelic acid procedure of Elsing and
1
4,6
originally described in the patent and journal citations as the trans-isomer.
As the drug was developed in the United States, this name was revised to
the cis-isomer.3
7
Blaschke successfully afforded partially resolved tramadol
after multiple fractional recrystallizations, these salts failed
to meet the desired specifications of >99% optical purity.
(
(
(
3) Physicians Desk Reference; Medical Economics Data: Oradell, NJ, 2000;
Vol. 54, pp 2218-2219.
4) Frankus, E. v.; Friedrichs, E.; Kim, S. M.; Osterloh, G. Arzneim.-Forsch./
Drug Res. 1978, 28 (I), 114-121.
5) Goeringer, K. E.; Logan, B. K.; Christian, G. D. J. Anal. Toxicol. 1997,
(6) Flick, K.; Frankus, E. v.; Friedrichs, E. Arzneim.-Forsch./Drug Res. 1978,
28 (I), 107-113.
(7) Elsing, B.; Blaschke, G. Arch. Pharm. (Weinheim, Ger.) 1991, 324, 719.
2
1, 529-537.
1
0.1021/op000281v CCC: $19.00 © 2000 American Chemical Society and The Royal Society of Chemistry
Vol. 4, No. 4, 2000 / Organic Process Research & Development
•
291
Published on Web 05/03/2000

Additionally, the isolated yields were very low (<5% yield
as the mandelate salt) and the necessity of ultimately
preparing multi 100-g quantities required the detailed
reevaluation of this method in an attempt to improve the
yield and purity of each of the separated enantiomers.
Early in the process of attempting to optimize the
fractional crystallization, it was noted that the enantiomers
of cis-tramadol free base formed mandelic acid salts at
different rates. When (R)-(-)-mandelic acid was used, the
solid product was enriched in the (-)-enantiomer. When (S)-
Table 1. Resolution of (-)-cis-tramadol
crystalli- (R)-(-)-mandelic
%
zation
acid (equiv)
recovery
HPLC purity
1
2
3
0.75
49.3
65.0
85.2
75.3% (-)-enantiomer
24.7% (+)-enantiomer
0.70
96.8% (-)-enantiomer
3.2% (+)-enantiomer
99.4% (-)-enantiomer
0.95
0.6% (+)-enantiomer
(
(
+)-mandelic acid was used, the solids were enriched in the
+)-enantiomer. The respective mother liquors were cor-
Table 2. Resolution of (+)-cis-tramadol
crystalli- (S)-(-)-mandelic
%
respondingly enriched in the opposite enantiomer (affording
the possibility of recycling and recovering the mother liquor
concentrates). When 1 mol equiv of (R)-(-)-mandelic acid
was used, the extremely rapid crystallization afforded an 87%
yield of the salt of which the -:+ ratio was 54.8:45.2. When
the theoretically optimal charge of mandelic acid was utilized
zation
acid (equiv)
recovery
HPLC purity
1
2
3
0.75
43.2
83.0
94.0
6% (-)-enantiomer
94% (+)-enantiomer
1.9% (-)-enantiomer
0.95
9
8.1% (+)-enantiomer
0.98
100% (+)-enantiomer
(no (-)-enantiomer
detected)
(
0.5 mol equiv of (R)-(-)-mandelic acid) the initial isolated
yield of crystalline salt was <15%. The solids isolated were
1.4% of the (-)-enantiomer of cis-tramadol. Attempts to
8
improve the isolated yield by conducting the resolution more
concentrated or to concentrate the filtrate following the first
crop isolation, afforded a solid of which the -:+ ratio was
about 60:40. A quick examination of the molar charge of
mandelic acid) afforded either salt in >99% optical purity
as determined by chiral HPLC analyses (Tables 1 and 2).9
To complete the comparison to the known data for
resolved tramadol, a small sample of each resolved enanti-
omer was converted to the free base after which the
respective hydrochloride salts were prepared. In both cases,
each salt compared very favorably to the literature data
(melting point and optical rotation).
(R)-(-)-mandelic acid to tramadol free base in a fixed
volume of acetate solvent indicated that when a 70-75%
mol % charge was used, the white solid isolated was
significantly enriched in one enantiomer at a maximized
isolated yield of about 50%. Resolution by differential salt
solubilities is not a new concept, but the specific undercharge
of the chiral resolving acid to maximize the resolution was
a concept we were unable to locate in the chemical literature.
However, recently this concept was elegantly demonstrated
Conclusions
The observation that the two enantiomers of tramadol
selectively form optically active salts of mandelic acid has
afforded a process to prepare each of the separated enanti-
omers. This process was successfully scaled up to prepare
100+ g of each enantiomer in >99% optical purity. A sample
of each enantiomer was converted to the known optically
active hydrochloride salts (via the free base) with excellent
agreement with the published data.
in the efficient resolution of (()-threo-methylphenidate with
8
(
R)-(-)-binaphthyl-2,2′-diyl hydrogen phosphate. With cis-
tramadol free base, this generalized procedure worked
surprisingly well. The first crystallization (40-50% absolute
yield) typically afforded a 75:25 to 94:6 mixture of each
enantiomer.
Attempts to improve the optical purity of the first
crystallization products to acceptable purity levels by simple
recrystallizations, fractional recrystallizations, or selective
solvent digestions were only marginally successful, as the
enantiomer ratio remained virtually unchanged. When the
partially resolved salt was converted to the free base and
the salt reformed with a charge of fresh mandelic acid which
matched the % of the desired enantiomer present in the
sample, the crystallized product was typically >95% in
optical purity of the desired enantiomer. Since the (+) and
Experimental Section
All non-aqueous reactions were performed under a dry
nitrogen atmosphere. Racemic cis-tramadol hydrochloride
was prepared following the literature procedure.1 Reagents
and solvents were obtained from commercial sources and
used as received. Proton magnetic resonance spectra were
obtained on a Bruker AC 300 MHz NMR, using either
tetramethylsilane or chloroform as an internal reference.
Infrared spectra were obtained as KBr pellets on a Perkin-
Elmer Spectrum 1000 Infrared Spectrophotometer. Mass
spectra analyses were performed on a Shimadzu QP-5000
GC/MS (CI mass spectrometry). Melting points were ob-
tained using a Perkin-Elmer model DSC-4 differential
scanning calorimeter. HPLC analyses were performed on a
Waters 600 E HPLC system equipped with a Waters 440
UV detector and Hewlett-Packard 3396A integrator.
,6
(-) enantiomers of mandelic acid are each readily com-
mercially available and relatively inexpensive, it was a quick
effort to show that this procedure worked equally well to
produce either enantiomer of cis-tramadol from the racemic
free base. One more repetition of the free base and salt re-
formation sequence (again using the diminished mol % of
(8) Prashad, M.; Hu, B.; Repi e` , O.; Blacklock, T. J.; Giannousis, P. P. Org.
Process Res. DeVel. 2000, 4, 55-59.
(9) Elsing, B.; Blaschke, G. J. Chromatogr. 1993, 612, 223-230.
292
•
Vol. 4, No. 4, 2000 / Organic Process Research & Development

Chiral HPLC Method for the Analysis of cis-Trama-
dol. The chiral HPLC analysis of cis-tramadol was carried
out using a modified version of the method of Elsing and
acid was extracted with diethyl ether (3 × 300 mL). The
aqueous solution was neutralized with solid sodium bicar-
bonate (34.2 g) and then extracted with methylene chloride
(3 × 300 mL). The combined organic extracts were dried
over sodium sulfate, filtered, and the solvents were removed
under reduced pressure to afford 46.5 g (0.18 mol) of free
base as a thick oil. This oil was dissolved, with vigorous
stirring, in ethyl acetate (200 mL) and a warm solution of
(R)-(-)-mandelic acid (18.8 g, 0.12 mol, 0.7 equiv) in ethyl
acetate (50 mL) was added in one portion. Crystallization
occurred very quickly and the resulting slurry was stirred at
room-temperature overnight. The resulting solids were col-
lected by vacuum filtration, washed with cold ethyl acetate
(20 mL), and dried to a constant weight to afford a 65%
recovery of cis-tramadol (R)-(-)-mandelate salt (48.6 g).
This salt was a mixture of 96.8% of the (-)-enantiomer and
3.2% of the (+)-enantiomer (by chiral HPLC analysis).
This salt was combined with another sample of the same
enantiomeric purity (89.5 g total) and the mixture was
dissolved in water (1 L). The solution was acidified (pH ≈
2) with concentrated hydrochloric acid (25 mL), and the
mandelic acid was washed out with diethyl ether (3 × 500
mL). The aqueous solution was neutralized with solid sodium
bicarbonate (34.4 g) and then extracted with methylene
chloride (3 × 400 mL). The combined organic extracts were
dried over sodium sulfate and filtered, and the solvents were
removed under reduced pressure to afford 54.5 g (0.21 mol)
of free base as a thick oil. This oil [98.1% (-)-enantiomer]
was dissolved, with vigorous stirring, in ethyl acetate (50
mL), and a warm solution of R-(-)-mandelic acid (29.9 g,
0.20 mol, 0.95 equiv) in ethyl acetate (50 mL) was added in
one portion. Crystallization occurred very quickly, and the
resulting slurry was stirred at room temperature overnight.
The resulting solids were collected by vacuum filtration,
washed with cold diethyl ether (40 mL), and dried to a
constant weight to afford a 85% recovery of (-)-cis-tramadol
(R)-(-)-mandelate salt (74.6 g). This salt was composed of
99.4% of the (-)-enantiomer and 0.6% of the (+)-enantiomer
9
Baschke. A sample (about 1 mg of any salt or free base in
5
(
2
mL of 2-propanol) was eluted on a Chiralpak AD column
4.6 mm × 25 cm) with an isocratic mobile phase of hexane/
-propanol/diethylamine (97.5:2.5:0.01) at 1 mL/minute. The
system was monitored by UV at 280 nm. The (+)-cis
enantiomer eluted at 7.0 min and the (-)-cis enantiomer
eluted at 11.4 min.
Preparation of cis-Tramadol Free Base. To a vigorously
stirred mixture of saturated aqueous sodium bicarbonate
solution (2.0 L) and solid sodium carbonate (50 g, 0.47 mol)
was added, portionwise, racemic cis-tramadol hydrochloride
(
254.4 g, 0.85 mol). Once the solids had dissolved, the
aqueous solution was extracted with methylene chloride (3
1.2 L). The combined organic extracts were washed with
×
saturated aqueous sodium chloride solution (2 × 2.0 L), dried
over anhydrous sodium sulfate, clarified, and concentrated
under reduced pressure to afford the free base with 98.7%
recovery as a viscous oil (221.4 g, 0.84 mol). This material
was used without purification in the following resolution
procedures.
Preparation of (-)-cis-Tramadol (R)-(-)-Mandelate.
A stirred biphasic mixture of tramadol free base (66.0 g,
0.25 mol) in a mixture of isopropyl acetate (300 mL) and
ethyl acetate (200 mL) was heated at 70 °C until a clear
solution was obtained. A solution of (R)-(-)-mandelic acid
(28.6 g, 0.188 mol, 0.75 equiv) in ethyl acetate (100 mL)
was prepared with heating at 40 °C. The mandelic acid
solution was added to the free base solution, and the resulting
mixture was stirred at 35 °C until crystallization began. The
crystallizing mixture was allowed to cool to room temper-
ature and stirred overnight. The resulting slurry was cooled
to -5 °C, with stirring, for 3.5 h, and the solids were
collected by vacuum filtration. The filter cake was washed
with isopropyl acetate (2 × 50 mL) and diethyl ether (2 ×
50 mL) and dried to a constant weight to afford 72.2 g of a
25°C
white crystalline solid. Chiral HPLC analysis of the solid
showed a 40:60 mixture of (+)- and (-)-cis-tramadol
enantiomers. Chiral HPLC analysis of the filtrate showed
an 88.7:11.3 mixture of the (+)- and (-)-cis-tramadol
enantiomers. The solid crystalline material was mostly
dissolved in boiling ethyl acetate (1.2 L), and the solution
was clarified and cooled to room temperature. The solution
was then concentrated to a volume of about 500 mL and
cooled to 0 °C with stirring for about 4 h. The resulting
crystals were collected by vacuum filtration, washed with
cold ethyl acetate (50 mL), and dried to a constant weight
to afford a 49.3% yield (57.3 g) of the (-)-mandelate salt
of cis-tramadol as white crystalline solid. This solid contained
(by chiral HPLC analysis), mp ) 153-155 °C, [R]
D
)
1
-57.2° (c ) 1.02, methanol), H NMR, δ (CD OD): 1.40-
3
1.90 (m, 8H), 2.20 (m, 1H), 2.55 (s, 6H), 2.75-2.95 (m,
1H), 3.30 (m, 1H), 3.80 (s, 3H), 4.85 (s, 1H), 4.90 (s, 3H,
exchanges with D
4H) and 7.45 (m, 2H) ppm. IR (KBr, cm ): 3400 (ν, OH,
S), 3237 (ν, H-bonding, S, br), 2926 (ν, CH , CH , S), 1616
(ν, carboxylate, VS), 1580 (ν, aromatic C ) C, M), 1351
(δ, OH, M); 1253, 1216 (ν, C-O-CH , S). Anal. Calcd for
: C, 69.37; H, 8.01; N, 3.37. Found: C, 69.23;
2
O), 6.80 (d, 1H), 7.10 (m, 2H), 7.25 (m,
-
1
3
2
3
24 5
C H33NO
H, 7.82; N, 3.33.
A small sample of this mandelate salt was converted to
the free base and then to the known (-)-cis-tramadol
4
75.3% of the (-)- and 24.7% of the (+)-enantiomer (by
hydrochloride salt following literature procedures. The
chiral HPLC analysis).
melting point and optical rotation were comparable to the
2
6°C
This sample of partially resolved (-)-cis-tramadol was
combined with several additional samples of the same 3:1
mixture of enantiomers (75.0 g total) and was dissolved in
water (500 mL). The solution was acidified (pH ≈ 2) with
concentrated hydrochloric acid (15 mL) and the mandelic
literature values: mp ) 172-174 °C, [R]
) 1.0, water), [lit: mp ) 169-170 °C, [R]
) 1.0, water)].
D
) -28.8° (c
) -28° (c
4
20°C
D
Preparation of (+)-cis-Tramadol (S)-(+)-Mandelate.
A stirred biphasic mixture of racemic cis-tramadol free base
Vol. 4, No. 4, 2000 / Organic Process Research & Development
•
293

(
7
(
66.0 g, 0.25 mol) in ethyl acetate (200 mL) was heated at
0 °C until a clear solution was obtained. A solution of (S)-
+)-mandelic acid (28.6 g, 0.188 mol, 0.75 equiv) in ethyl
ether (3 × 300 mL). The aqueous solution was neutralized
with solid sodium bicarbonate (24 g) and extracted with
methylene chloride (3 × 300 mL). The combined organic
extracts were dried over sodium sulfate, filtered, and the
solvents were removed under reduced pressure to afford 33.4
g (0.13 mol) of free base as a thick oil. This oil was
dissolved, with vigorous stirring, in ethyl acetate (20 mL)
and a warm solution of (S)-(+)-mandelic acid (19.3 g, 0.12
mol, 0.98 equiv) in ethyl acetate (55 mL) was added in one
portion. Crystallization occurred very quickly and the result-
ing slurry was stirred at room temperature for 7 h. The solids
were collected by vacuum filtration, washed with cold ethyl
acetate (25 mL), and dried to a constant weight to afford a
94% recovery of (+)-cis-tramadol S-(+)-mandelate salt
(48.5 g). This salt was entirely the (+)-enantiomer by chiral
acetate (150 mL) was prepared with warming at 40 °C. The
mandelic acid solution was added in one portion to the free
base solution and the resulting mixture was stirred at about
4
0 °C until crystallization began. The crystallizing mixture
was allowed to cool to room temperature and stirred
overnight. The resulting slurry was cooled to -5 °C and
stirred for 4.5 h, at which point the solids were collected by
vacuum filtration. The filter cake was washed with isopropyl
acetate (50 mL) and diethyl ether (50 mL) and was dried to
a constant weight to afford 43.2% yield (50.3 g) of a white
crystalline solid. Chiral HPLC analysis of the solid showed
a 94:6 mixture of the (+)- and (-)-cis-tramadol enantiomers.
Chiral HPLC analysis of the filtrate showed a 4.3:95.7
mixture of the (+)- and (-)-cis-tramadol enantiomers.
This sample of partially resolved (+)-cis-tramadol was
combined with two additional small samples of a 95:5
mixture of enantiomers (54.8. g total), and the mixture was
dissolved in water (600 mL). The solution was acidified (pH
2
5°C
HPLC analysis, mp ) 153-155 °C, [R]
D
) +57.8° (c
1
) 1.07, methanol), H NMR, δ (CD
3
OD): 1.40-1.90 (m,
8H), 2.20 (m, 1H), 2.55 (s, 6H), 2.75-2.95 (m, 1H), 3.30
(m, 1H), 3.80 (s, 3H), 4.85 (s, 1H), 4.90(s, 3H), 6.80 (d,
1H), 7.10 (m, 2H), 7.25 (m, 4H) and 7.45 (m, 2H) ppm. IR
-
1
(KBr, cm ): 3400 (ν, OH, S), 3237 (ν, H-bonding, S, br),
2927 (ν, CH , CH , S), 1615 (ν, carboxylate, VS), 1580 (ν,
aromatic C ) C, M), 1351 (δ, OH, M), 1253, 1216 (ν,
C-O-CH , S). Anal. Calcd for C24 : C, 69.37; H,
≈
2) with concentrated hydrochloric acid (15 mL), and the
3
2
mandelic acid was extracted with diethyl ether (3 × 300 mL).
The aqueous solution was neutralized with solid sodium
bicarbonate (34 g) and then extracted with methylene
chloride (3 × 300 mL). The combined organic extracts were
dried over sodium sulfate and filtered, and the solvents were
removed under reduced pressure to afford 40.5 g (0.15 mol)
of free base as a thick oil. This oil was dissolved, with
vigorous stirring, in ethyl acetate (30 mL), and a warm
solution of (S)-(+)-mandelic acid (22.2 g, 0.146 mol, 0.95
equiv) in ethyl acetate (60 mL) was added in one portion.
Crystallization occurred very quickly, and the resulting slurry
was stirred at room temperature overnight. The solid was
collected by vacuum filtration, washed with cold ethyl acetate
3
H33NO
5
8.01; N, 3.37. Found: C, 69.42; H, 7.87; N, 3.34.
A small sample of this mandelate salt was converted to
the free base and then to the known (+)-cis tramadol
4
hydrochloride salt following literature procedures. The
melting point and optical rotation were comparable to the
2
6 °C
literature values: mp ) 172-175 °C, [R]
D
) +27.8° (c
) +27.5°
4
20°C
) 1.0, water), [lit.: mp ) 169-170 °C, [R]
(c ) 1.0, water)].
D
Acknowledgment
We thank Zeeland Chemical, Inc. (Zeeland Michigan) for
their generous gift of (R)-(-)-mandelic acid and (S)-(+)-
mandelic acid. We also thank Drs. Donald Kuhla, R. Jason
Herr, and Michael Sherrod for assistance with the preparation
of this manuscript.
(20 mL), and dried to a constant weight to afford a 83%
recovery of cis-tramadol (S)-(+)-mandelate salt (51.6 g). This
salt was 98.1% (+)-enantiomer and 1.9% (-)-enantiomer
(
HPLC).
This salt was dissolved in water (600 mL) and the solution
was acidified (pH ≈ 2) with concentrated hydrochloric acid
15 mL). The mandelic acid was washed out with diethyl
Received for review February 10, 2000.
OP000281V
(
294
•
Vol. 4, No. 4, 2000 / Organic Process Research & Development