A large-scale asymmetric synthesis of (S)-cyclohexylphenyl glycolic acid

Article abstract of DOI:10.1016/j.tetasy.2003.08.036

A practical asymmetric synthesis of (S)-cyclohexylphenyl glycolic acid has been demonstrated at pilot scale. The key step is the asymmetric aldol reaction using the Seebach approach of self-replication of stereochemistry. (S)-Mandelic acid was converted i

Full text of DOI:10.1016/j.tetasy.2003.08.036

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 14 (2003) 3593–3600
A large-scale asymmetric synthesis of (S)-cyclohexylphenyl
glycolic acid
Xiping Su,* Nandkumar N. Bhongle,* Derek Pflum,† Hal Butler, Stephen A. Wald, Roger P. Bakale
and Chris H. Senanayake*,‡
Chemical Research and Development, Sepracor Inc., 84 Waterford Drive, Marlborough, MA 01752, USA
Received 15 July 2003; accepted 18 August 2003
Abstract—A practical asymmetric synthesis of (S)-cyclohexylphenyl glycolic acid has been demonstrated at pilot scale. The key
step is the asymmetric aldol reaction using the Seebach approach of self-replication of stereochemistry. (S)-Mandelic acid was
converted into (2S,5S)-3 in >99/1 diastereomeric ratio. The dioxolone (2S,5S)-3 was then subjected to aldol conditions to give the
aldol intermediate (2S,5R)-5 in 91% yield with >99.5/0.5 diastereomeric ratio. The aldol product was converted to the desired
hydroxy acid, (S)-CHPGA, in three steps with only one isolation in high yield (95%) and high purity.
© 2003 Elsevier Ltd. All rights reserved.
1. Introduction
tinence, but it suffers from side effects such as dry mouth,
tachycardia and mydriasis.2 (S)-Oxybutynin is currently
beingevaluatedinphaseIIIclinicaltrialsforthetreatment
of overactive bladder. As a continuation of our efforts
to develop cost-effective syntheses of (S)-oxybutynin,3,4
we have sought to identify an efficient asymmetric
synthesis of (S)-cyclohexylphenyl glycolic acid, [(S)-
CHPGA], the key intermediate to (S)-oxybutynin
(Scheme 1). Herein, we report a successful asymmetric
synthesis of (S)-CHPGA at pilot scale that is suitable for
large-scale manufacturing.
The incidence of urinary incontinence, the involuntary
loss of urine in sufficient quantities to be considered a
socialorhealthproblem, continuestogrowasthegeriatric
population increases. Urinary incontinence is caused by
hyper reactivity of the detrusor muscle and is widely
treated by muscarinic receptor antagonists, particularly
acting at the M3 subtype.1 Racemic oxybutynin (ditropan)
is a widely prescribed muscarinic receptor antagonist for
the treatment of urinary urgency, frequency and incon-
Scheme 1.
* Corresponding authors. Tel.: +1-508-357-7418; fax: +1-508-753-7808; e-mail: xiping.su@sepracor.com; nandkumar.bhongle@sepracor.com
† Present address: Pfizer Inc., 2800 Plymouth Road, Ann Arbor, MI 48105, USA.
‡ Present address: Boehringer-Ingelheim Pharmaceutical Inc., 900 Ridgebury Road, Ridgefield, CT 06877, USA.
0957-4166/$ - see front matter © 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2003.08.036

3594
X. Su et al. / Tetrahedron: Asymmetry 14 (2003) 3593–3600
Scheme 2. Synthesis of dioxolone 3.
2. Results and discussion
2.1. Diastereoselective dioxolone formation
Since the landmark work by Seebach et al.,5 the con-
cept of self-regeneration of stereocenters has found
widespread applications in organic synthesis, especially
in the creation of quaternary stereocenters. The require-
ment of pentane as the reaction solvent, however,
severely limited its use in large-scale production. For
the practical scale-up of the reaction, the highly volatile
and flammable solvent pentane4b has to be avoided. It
is widely perceived that solvents with higher boiling
points such as heptane or hexanes result in rather low
diastereoselectivity.6 The Merck group7 has used triiso-
propyl orthoformate in toluene to prepare an
orthoester of mandelic acid before the addition of
pivaldehyde to achieve high diastereoselectivity of the
desired dioxolone, but this approach adds an additional
step to the reaction sequence and also requires the use
of a relatively expensive reagent, triisopropyl orthofor-
mate. Our goal was to develop a practical method that
was both highly selective and economical.
Figure 1. Dependence of diastereomeric ratio on temperature
(cis/trans ratio refers to the diastereomeric ratios of the
slurry).
These results are in sharp contrast to the observations
made by the Merck group. While the high selectivities
achieved by the Merck group seemed to originate from
kinetic control, our results appear to favor a thermody-
namically controlled formation of the cis-diastereomer,
presumably due to the highly crystalline nature of cis-3.
In a control experiment, a mixture of the cis-3 and
trans-4 isomers was stirred in hexanes at 21°C in the
presence of 0.1 equiv. of methanesulfonic acid with the
diastereomeric ratio increasing with time (Fig. 2). Start-
ing with a ratio of 1.3/1 cis/trans, the diastereomeric
ratio increased rapidly to 25/1 in 1.3 h at 21°C and
approached steady state. Decreasing the temperature to
0°C resulted in a ratio of 42/1. On the other hand, in a
toluene solution with 6 mol% of p-toluenesulfonic acid
monohydrate (similar to the Merck conditions), a mix-
Since the dioxolone 3 is a highly crystalline product,
formed from (S)-mandelic acid in pentane, we reasoned
that the selective formation of the cis isomer could be
driven by selective crystallization (Scheme 2).8 If this is
the case, under equilibration conditions, crystallization
of a mixture of (2S,5S)-3 and (2R,5S)-4 should result in
products that are highly enriched in
a
single
diastereomer, even if the diastereomeric ratio is low in
solution.9 A similar crystallization driven selective for-
mation of imidazolidinones and ketimines has been
reported.10
Catalytic amounts of methanesulfonic acid were used
effectively to catalyze this formation as well as the
epimerization of the dioxolones 3 and 4 in hexanes. In
a typical experiment, a slurry of (S)-mandelic acid in
hexanes was allowed to reflux with 1.1 equiv. of
pivaldehyde and 0.3 equiv. of methanesulfonic acid.
The water generated in the reaction was separated with
a Dean–Stark trap. The samples of the slurry were
dissolved in THF and the diastereomeric ratios deter-
mined by HPLC analysis. At 70°C the diastereomeric
ratio was approximately 4/1. When the slurry was
allowed to slowly cool, the diastereomeric ratio
increased dramatically, and a 51/1 diastereomeric ratio
was obtained at 21°C and was increased to 91/1 when
the mixture was cooled to 0°C (Fig. 1).
Figure 2. Equilibration of dioxolones 3/4 with methanesul-
fonic acid (cis/trans ratio refers to the diastereomeric ratios of
the slurry).

X. Su et al. / Tetrahedron: Asymmetry 14 (2003) 3593–3600
3595
ture of 3 and 4 (1.3/1 ratio of 3/4 by HPLC) did not
by a Dean–Stark trap. This method, which was demon-
strated at a kilolaboratory scale, eliminated the partial
racemization seen at higher temperature and has
yielded comparable results as with the hexanes. How-
ever, when compared to the hexanes process, this pro-
cess modification increased the operational difficulty
associated with maintaining the system at constant
pressure and complicates the efficient scale-up of the
process.
give any measurable change in ratio at 20–28°C over 18
h.
The increase in the diastereomeric ratio of 3/4 at lower
temperatures is also closely related to the solubility of 3
in the specific hydrocarbon solvents tested. Figure 3
shows the solubility curve of 3 in heptane.11 At 69°C,
the solubility of 3 is 29.4 mg/mL in heptane while its
solubility decreases to <5 mg/mL at 0°C. The isomer
trans-4 has not been crystallized, as it is oil at ambient
temperature. It is also noteworthy that during the
formation of 3 from (S)-mandelic acid 2, the ratio of
3/4 remained <2/1 in the solution phase throughout the
cooling process from 68 to 0°C. This observation sup-
ports the hypothesis that selective formation of 3 is
driven by crystallization of 3 from the reaction mixture
and is in agreement with LeChatelier’s principle.
2.2. Telescoped process: aldol addition reaction
With a reliable method available for the highly selective
synthesis of 3, we then turned our attention to a
telescoped process without isolating dioxolone 3. In
order to solubilize 3 after quenching the reaction, THF
was added, and the mixture heated to 60°C. A clear
solution formed and the phases were separated. The
organic phase was azeotropically dried at reflux with a
Dean–Stark trap. This dried solution of 3 was then
cooled to about −78°C with dioxolone 3 partially crys-
tallizing to form a slurry in the mixed solvent of THF
and hexanes. A solution of LDA was then added to the
slurry at −70 to −78°C and the reaction mixture
allowed to warm to −50 to −58°C for 1 h to complete
the enolate formation. The mixture was then cooled to
about −78°C. Cyclohexanone was added between −70
and −78°C and after quenching and aqueous work-up,
the desired aldol intermediate 5 was found to be >98%
by HPLC analysis (Scheme 3). On the other hand, if the
enolate formation was kept below −70°C, the aldol
intermediate 5 was found to be contaminated with
10–20% of dioxolone 3, suggesting that the deprotona-
tion was incomplete at −70 to −78°C. This is not
surprising given the fact that both 3 and its lithium
enolate were not completely soluble in the reaction
solvent system below −40°C. When the enolization was
allowed to proceed above −40°C, the reaction yield was
decreased and highly variable (50–75%), due to signifi-
cant side reactions during the enolization reaction at
higher temperatures.
Figure 3. Solubility of 3 in heptane at different temperatures.
Since this process proceeds at a higher temperature
compared to previously published conditions in pen-
tane, it is necessary to assess the potential for racemiza-
tion of dioxolone 3. A chiral HPLC method was
developed to separate each of the four stereoisomers of
the dioxolones.12 Under the reaction conditions above,
no detectable racemization occurred. On the other
hand, when heptane was used as a reaction solvent, the
reflux temperature rose to about 91°C with partial
racemization (about 2%/h) occurring. It is possible to
lower the reflux temperature of the heptane mixture by
applying a vacuum in the system while separating water
The aldol addition reaction proceeded smoothly at −70
to −78°C in 1–2 h. Quenching the reaction at a low
temperature proved critical in order to obtain both a
high yield and high diastereoselectivity of the aldol
intermediate 5. The reaction could be quenched with
either concentrated HCl or an acetic acid solution.
When the temperature of the reaction mixture was
allowed to rise above −40°C, decomposition of the
aldol reaction mixture began to occur and the
diastereomeric ratio of 5 was lowered. When the inter-
Scheme 3. Aldol addition reaction.

3596
X. Su et al. / Tetrahedron: Asymmetry 14 (2003) 3593–3600
nal temperature was maintained below −60°C, 5 could
be obtained with high diastereoselectivity (>97/3). After
quenching the reaction, the organic layer was separated
and washed with water and potassium bicarbonate
solution. A solvent switch to heptane was performed
and the desired product 5 was crystallized and isolated
in high yield (91%). This highly efficient process was
successfully scaled-up in kilolab and demonstrated at
pilot scale with 15 kg input of (S)-mandelic acid per
batch and the aldol intermediate 5 was isolated in
82–85% yield with high chemical purity and isomeric
purity (Table 1).
described previously. In order to select one of the two
routes for further development, both the sequences
were carried out on a 25 g scale and compared for the
chemical yield, product quality, volumetric productiv-
ity, convenience and potential for the development of a
telescoped process.
Although both of the routes are very comparable in
terms of yields, in general, Path A produced more
impurities. In addition, hydrogenation of the potassium
salt of 7 did not proceed at a favorable reaction rate,
and also generation of the free acid 7 and hydrogena-
tion was very cumbersome and inefficient, which would
lead to difficulties in telescoping the two steps. On the
other hand, Path B generated less impurities and the
crude reaction mixtures were conveniently carried for-
ward in further transformations. Thus, the Path B
process was selected for further development.
2.3. Conversion of aldol intermediate 5 to (S)-CHPGA
1
During our earlier process research evaluations, we
found that the aldol intermediate 5 could be converted
to (S)-CHPGA 1 via two alternative routes (Scheme
4).4b In the first route (Path A), 5 was subjected to an
elimination process with thionyl chloride/pyridine fol-
lowed by hydrolysis under KOH/MeOH/H2O condi-
tions to yield unsaturated (S)-CHPGA 7 that was
further converted to (S)-CHPGA 1 by hydrogenation
with 10% Pd–C. In an alternative route (Path B), the
last two steps in Path A were reversed, i.e. hydrogena-
tion was carried out after elimination to afford the
saturated dioxolone 8 that was further converted to
2.4. Elimination reaction of aldol intermediate 5
In our earlier work,4b the aldol intermediate 5 was
treated with thionyl chloride and pyridine in THF at
0–5°C over about 1 h followed by quenching and
washing the reaction with saturated ammonium chlo-
ride. Hydrogenation of the resulting reaction mixture
was problematic and typically resulted in very low or
no conversion most likely due to poisoning of the Pd–C
catalyst by sulfurous acid that is formed after hydroly-
(S)-CHPGA
1
under the hydrolysis conditions
Table 1. Data for isolated aldol intermediate 5
Scale
Yield (%)a
Total impurities (%)
SLIb (%)
Ee (%)
De (%)
125 g
0.7 kg
15 kg
91
87
85
0.1
0.7
<0.1
<0.1
0.7
<0.1
>99.9
99.8
>99.9
>99.9
>99.9
>99.9
a % yield is the overall yield of the aldol intermediate.
b Single largest impurity.
Scheme 4. Synthesis routes of (S)-1 from (2S,5R)-5.

X. Su et al. / Tetrahedron: Asymmetry 14 (2003) 3593–3600
Table 2. Data for elimination reaction
3597
vents such as methanol, ethanol, and isopropanol than
in non-polar solvents such as MTBE. When 10 vol% of
methanol was added to the MTBE solution of the
unsaturated dioxolone 6, and the hydrogenation was
conducted at 55°C, 10% Pd–C and 100 psi, H2, the
saturated dioxolone 8 formed with quantitative conver-
sion within 6 h. It was observed that the 50% wet
Degussa catalyst does not disperse well in a solution of
substrate and MTBE alone. After adding methanol, the
catalyst dispersion was markedly improved. The
efficient dispersion of the water wet catalyst by reduc-
ing the aggregation of the catalyst and increasing the
surface area available for the reaction is likely one of
the factors for the observed acceleration of the reaction
rate.
Scale
Conversion (%)
Total Impurities (%)
50 g
100 g
>99.9
>99.9
0.1
0.2
HPLC area %.
sis of excess thionyl chloride. The work-up procedure
was modified to remove sulfurous acid, and MTBE, a
less expensive and more hydrophobic solvent, was used
in place of THF to further improve the process reliabil-
ity. When the elimination reaction was conducted in
MTBE utilizing 1.3–1.4 equiv. of thionyl chloride and
2.1 equiv. of pyridine, the reaction conversion was
>99.9% after 1 h at 0–22°C. The by-products in the
reaction were identified as the pyridinium salts and
sulfurous acid. The pyridinium salts were completely
removed by water washes, and aqueous bicarbonate
washes were performed to remove the residual sul-
furous acid. The elimination process worked very well
with these modifications and produced the unsaturated
dioxolone 6 in good yields and high quality (Table 2).
The elimination process streams with these modifica-
tions consistently underwent clean hydrogenation in the
next step, indicating that telescoping the elimination
and hydrogenation processes was feasible.
2.6. Hydrolysis of saturated dioxolone 8 to (S)-
CHPGA 1
Sodium hydroxide was used for the hydrolysis in place
of potassium hydroxide used in the earlier version4b of
this method due to its lower molecular weight and the
higher water solubility of the by-product (NaCl versus
KCl), which facilitates the work-up and isolation and
affords higher reaction concentrations (increased volu-
metric productivity). The saturated dioxolone 8 was
suspended in a mixture of MeOH/H2O (1:1) containing
5 equiv. of NaOH and the resulting slurry refluxed at
70–73°C until a clear solution was obtained. After
about 2 h, HPLC analysis indicated >99% conversion
to CHPGA 1 which could be isolated with 99.96%
purity. Since an efficient solvent exchange from MTBE
following the hydrogenation was possible, MTBE was
removed by distillation after the addition of aq. NaOH
and methanol. Each of the three stages could be tele-
scoped into a through process from the aldol intermedi-
ate 5 to (S)-CHPGA 1.
2.5. Hydrogenation of unsaturated dioxolone 6
For the development of a telescoped process we first
needed to demonstrate that the hydrogenation of unsat-
urated dioxolone 6 obtained from the process stream
would work well at large scale. After an initial screen
for the catalyst type and catalyst loading, 10% Pd–C at
0.75 mol% loading was selected for development. An
elimination reaction mixture obtained using the work-
up conditions described above was hydrogenated at
ambient temperature under 50 psi H2 with 10% Pd–C
(50% wet Degussa catalyst). For safety reasons a 50%
water wet catalyst was selected and evaluated without
pretreatment for batch efficiency. The reaction was
judged complete by HPLC after 41 h with >99% con-
version and 99.8% purity (Table 3). Further optimiza-
tion was necessary to achieve the hydrogenation target
of 6–8 h reaction time. Increasing the H2 pressure to
100 psi did not improve the rate of the reaction.
However, when the hydrogenation was run at 100 psi
H2, at 50–55°C, the reaction was complete after 16 h.
This was an encouraging result, although the time
required for the reaction was still longer than desired.
In general, hydrogenations proceed faster in polar sol-
The key by-products formed during the hydrolysis reac-
tion and work-up are sodium chloride, pivaldehyde,
neopentyl alcohol and pivalic acid. Pivaldehyde, gener-
ated during hydrolysis, further undergoes a Canniz-
zarro reaction to yield neopentyl alcohol and the
sodium salt of pivalic acid. The level of NaCl generated
during the neutralization of the hydrolysis reaction can
have a dramatic effect on the (S)-CHPGA 1 crystalliza-
tion process quality. Therefore, it is desirable to have a
minimal amount of NaCl present during the crystalliza-
tion and a study utilizing <5 equiv. of NaOH was
conducted. A portion of the results is summarized in
Table 4. As expected, the reaction was slower with
fewer equiv. of NaOH, but alternatives are currently
Table 3. Hydrogenation of unsaturated dioxolone 6 with 0.75 mol% of 10% Pd–C catalyst
Conditions
Solvent
Time (h)
Conversion (%)
Total impurities (%)
50 psi/22°C
100 psi/22°C
100 psi/55°C
100 psi/55°C
MTBE
MTBE
MTBE
MTBE/MeOH (10:1)
41
45
16
6
>99.9
>99.9
>99.9
>99.9
0.2
0.4
0.3
0.3
Reaction scale: 25 g.

3598
X. Su et al. / Tetrahedron: Asymmetry 14 (2003) 3593–3600
Table 4. Hydrolysis of saturated dioxolone 8
the elimination reaction. Methanol was added to the
reaction mixture to make the MTBE/MeOH ratio 10:1,
followed by the addition of 0.75 mol% of 10% Pd–C
Degussa catalyst (50% water wet). A slow warming of
the reaction to 40°C under a controlled feed of hydro-
gen (<100 psi)13 was required during the initiation of
hydrogenation. The reaction temperature was allowed
to rise to 50–55°C. Once the initial exothermic reaction
was consumed, the hydrogen feed was adjusted to 100
psi and the reaction temperature maintained at 50–
55°C. The reaction was complete after 6 h and the
saturated dioxolone 8 produced in high yield and
purity. Conversion was nearly quantitative with >99.7%
of purity of 8 with <0.3% impurities. Similar results
were obtained at 1 kg and 10 kg scales (Table 6).
Scale
Equiv. of NaOH
Reaction time (h)
25 g
25 g
25 g
5.0
1.5
2.0
4.0
6.5
5.5
under investigation. The hydrolysis reaction was com-
plete after 5.5 and 6.5 h with 2.0 and 1.5 equiv. of
NaOH respectively. HPLC analysis of the reaction
mixture after completion of the reaction indicated a
similar impurity profile for each of these reaction
conditions.
2.7. Demonstration of the telescoped process
Table 6. Data for hydrogenation reaction
After successful laboratory execution of each of these
reactions to demonstrate both high conversion and
purity, the research effort was focused on development
of a telescoped process (Scheme 5) first at 150 g and
then at pilot plant scale. Since the impurity profile of a
telescoped process (no isolated intermediates) would be
expected to be different than that of a process with
stepwise isolations, it was decided to optimize the
robustness of the isolation/crystallization process of
(S)-CHPGA 1 in correlation with development of the
telescoped reactions.
Scale
Conversion (%)
Total impurities (%)
150 g
150 g
1 kg
>99.9
>99.9
>99.9
>99.9
0.20
0.15
0.20
0.30
10 kg
2.10. Hydrolysis of saturated dioxolone 8
The reaction mixtures obtained after hydrogenation
were filtered through celite to remove the catalyst, after
which 2 equiv. of aq. NaOH (3 M) were added and
MTBE removed by distillation. Methanol was added to
the resulting slurry to obtain a 1:1 ratio of methanol/
water and the reaction refluxed over approximately 5–6
h to complete the dioxolone hydrolysis. HPLC analysis
of the reaction indicated nearly quantitative conversion
with formation of <2% in-process impurities and was
independent of scale.
2.8. Elimination reaction of aldol intermediate 5
When the elimination reaction was scaled up to 150 g
using the conditions described earlier, the unsaturated
dioxolone 6 was formed with >99.9% conversion and
>99.7% purity (HPLC analysis). The elimination reac-
tion worked very well at 1 kg and 10 kg scales (Table
5).
Table 5. Data for elimination reaction
2.11. Work-up and isolation of (S)-CHPGA 1
Scale
Conversion (%)
Total impurities (%)
During the laboratory investigation phase of this work,
it was determined that both pivaldehyde and neopentyl
alcohol could not be completely removed by extracting
the reaction mixture with heptane. Since the reaction
solvent is a 1:1 mixture of methanol/water, and pivalde-
hyde and neopentyl alcohol are also soluble in
methanol, they could not be efficiently extracted with
heptane. Using alternative extraction solvents such as
MTBE proved to be unsuccessful. MTBE was also
found to be partially miscible with the reaction mixture.
One solution was to distill methanol and pivaldehyde
(bp 74°C/730 mm). Thus, after the completion of the
150 g
150 g
1 kg
>99.9
>99.9
>99.9
>99.9
0.24
0.20
0.18
0.20
10 kg
2.9. Hydrogenation of unsaturated dioxolone 6 reaction
stream
The hydrogenation reactions were conducted directly
on the reaction mixtures obtained after the work-up of
Scheme 5. Telescope process for the synthesis of (S)-1 from (2S,5R)-5.

X. Su et al. / Tetrahedron: Asymmetry 14 (2003) 3593–3600
3599
reaction, methanol and pivaldehyde were distilled at
atmospheric pressure until the reaction temperature
reached 95°C. The reaction was then cooled to room
temperature and washed with heptane. Methanol and
water were added back to the reaction mixture followed
by neutralization with about 3 equiv. of 6 M HCl. The
resulting slurry was stirred at room temperature,
filtered and the wet cake washed twice with water, and
then heptane to provide (S)-CHPGA after drying in
94–95% yield and >99.9% chemical and enantiomeric
purity. The hydrolysis worked very well at kilo and
pilot plant scales (Table 7).
water by Karl–Fisher titration (internal temperature
about 66°C). It was then distilled to remove about 125
mL of solvent and cooled to −78°C with stirring (crys-
tals precipitated from the solution upon cooling).
In a separate flask, LDA was made from diisopropyl-
amine (138 mL, 0.986 mol, 1.2 equiv.) in THF (100) by
the addition of n-BuLi (2.43 M in hexanes, 372 mL,
0.90 mol, 1.1 equiv.) at 0 to −10°C. The LDA solution
was then added to the above slurry maintaining the
internal temperature below −70°C. The mixture was
then stirred at −70 to −75°C for 2 h, warmed to −55°C
and stirred between −50 and −58°C for 2 h to ensure
the complete formation of the enolate. The mixture was
then cooled to −78°C and a solution of cyclohexanone
(93.7 mL, 0.90 mol, 1.1 equiv.) in heptane (95 mL)
added slowly to maintain the internal temperature
below −70°C. After the addition of cyclohexanone, the
mixture was stirred at −70 to −78°C for 2 h. The
reaction was quenched by adding concentrated HCl
(177 mL, 2.5 equiv.) slowly so that the internal temper-
ature was maintained below −70°C for the first third
and below −50°C for the rest of the HCl addition. The
mixture was then allowed to warm to 10°C after which
water (125 mL) was added. The aqueous layer was
removed and the organic layer washed consecutively
with water (250 mL), 20% KHCO3 (125 g) and water
(125 mL). It was distilled to about 600 mL under
reduced pressure, while maintaining the internal tem-
perature <60°C. Heptane (625 mL) was added and
distilled to about 600 mL. Another portion of heptane
(625 mL) was added and distilled to about 600 mL
again. The solvent in the pot contained <1% THF as
judged by GC. Heptane (500 mL) was added and the
mixture cooled to 0°C and stirred for 2 h. The slurry
was then filtered and the cake washed with heptane
(125 mL). The solid was dried under high vacuum at
35–45°C for 5 h to give the title compound as a white
crystalline product (238.0 g, 91.0%).
In conclusion, a practical and scalable asymmetric syn-
thesis of (S)-CHPGA has been developed starting from
(S)-mandelic acid. This efficient five-step process gives
(S)-CHPGA in 86% overall yield with only two isola-
tions. This process has been successfully performed at
pilot scale (15 kg inputs of mandelic acid) to provide
(S)-CHPGA 1 with very high chemical and enan-
tiomeric purity.
3. Experimental
3.1. cis-(2S,5S)-2-(tert-Butyl)-5-phenyl-1,3-dioxolan-4-
one (S,S)-3
To a suspension of (S)-mandelic acid (125 g, 0.822 mol)
in hexanes (625 mL) was added pivaldehyde (98.2 mL,
0.904 mol, 1.1 equiv.), followed by the addition of
methanesulfonic acid (16.0 mL, 0.25 mol, 0.3 equiv.)
under argon with stirring. The internal temperature
rose to 26°C from 19°C. The mixture was then heated
to reflux with the separation of water by a Dean–Stark
trap. When about 11 mL of water was collected, the
internal temperature rose to about 71°C. The mixture
was then distilled to remove about 125 mL of solvent.
The mixture was stirred and cooled to 35°C in about 1
h and HPLC analysis of a sample of the slurry showed
about 0.7% (area percent) of (S)-mandelic acid and
diastereomeric ratio of 30/1. It was then stirred at room
temperature for 2 h, cooled to 0°C and stirred for 1 h.
The mixture was quenched at 0°C by adding 20%
KHCO3 solution (250 g), maintaining the internal tem-
perature below 5°C. After the addition of THF (500
mL), the mixture was heated to 60°C and the solid
dissolved. The aqueous layer was removed and HPLC
analysis of the organic layer showed 101/1 syn/anti. The
organic solution was refluxed with a Dean–Stark trap
to remove water until the organic layer showed <0.1%
3.2. (S)-Cyclohexylphenyl glycolic acid 1
To a slurry of (2S,5R)-2-tert-butyl-5-(1-hydroxycyclo-
hexyl)-5-phenyl-[1,3]dioxolan-4-one (150 g, 0.47 mol) in
900 mL MTBE was added thionyl chloride (73.4 g, 0.62
mol) and pyridine (77.34 g, 0.98 mol) sequentially at
0–20°C. The reaction was stirred for 1.5 h. After
quenching the reaction with water (600 mL), the
aqueous layer and organic layer were separated. The
organic layer was washed with water (600 mL), 20% aq.
KHCO3 (600 mL) and water (450 mL) successively.
Table 7. Data for isolated (S)-CHPGA 1
Scale
Yield (%)a
Assay (%)b
Total impurities (%)
SLIc
ee (%)
150 g
150 g
1 kg
104 g (94.2)
105 g (95.2)
0.685 kg (93.1)
6.72 kg (92.1)
99.98
99.90
99.93
99.90
0.02
0.10
0.07
0.10
0.02
0.03
0.07
0.05
>99.9
>99.9
>99.9
>99.9
10 kg
a % yield is the overall yield for the three steps.
b Purity determined by titration assay.
c Single largest impurity.

3600
X. Su et al. / Tetrahedron: Asymmetry 14 (2003) 3593–3600
After adding 90 mL methanol, the organic layer was
subjected to hydrogenation for 6 h at 100 psi/55°C
using 10% Pd–C (0.75 mol%, 50% wet Degussa type)
catalyst. The reaction mixture was filtered through
celite to remove the catalyst. A solution of 38 g (0.95
mol) of NaOH in 300 mL of water was added to the
MTBE solution and MTBE distilled. 300 mL of
methanol was added to the resulting slurry and the
heterogeneous reaction heated to reflux for approxi-
mately 6 h. After distillation of volatiles, the reaction
was washed with heptane (600 mL). Methanol (240
mL) and water (960 mL) were added and the reaction
mixture was cooled to 12°C, and 240 mL of 6 M HCl
was added while maintaining the temperature below
20°C. The resulting slurry was stirred at room tempera-
ture for 1 h. The slurry was then filtered and the cake
was washed with water (2×300 mL), heptane (2×300
mL), and dried to yield (S)-CHPGA as white solid (104
g, 94.2% yield, >99.9% enantiomeric excess). The ee was
determined by HPLC (Chiralpak AS, mobile phase
95% hexane/5% IPA/0.1% TFA). (S)-1 elutes at
approximately 9.24 min. (R)-1 elutes at approximately
Optically Active Cyclohexylphenyl Glycolate Esters’; (c)
Bakale, R. P.; Lopez, J. L.; McConville, F. X.; Vanden-
bossche, C. P.; Senanayake, C. H. US Patent 6,140,529,
Oct. 31, 2000, ‘Synthesis Of Optically Active Cyclo-
hexylphenyl Glycolate Esters’; (d) Masumoto, S.; Suzuki,
M.; Kanai, M.; Shibasaki, M. Tetrahedron Lett. 2002, 43,
8647; (e) Gupta, P.; Fernandes, R. A.; Kumar, P. Tetra-
hedron Lett. 2003, 44, 4231.
4. (a) Senanayake, C. H.; Fang, K.; Grover, P. T.; Bakale,
R. P.; Vandenbossche, C. P.; Wald, S. A. Tetrahedron
Lett. 1999, 40, 819; (b) Grover, P. T.; Bhongle, N. N.;
Wald, S. A.; Senanayake, C. H. J. Org. Chem. 2000, 65,
6283 and references cited therein.
5. (a) Seebach, D.; Naef, R.; Calderari, G. Tetrahedron
1984, 40, 1313; (b) Seebach, D.; Sting, A. R.; Hoffmann,
M. Angew. Chem., Int. Ed. Engl. 1996, 35, 2708; (c)
Seebach, D.; Naef, R. Helv. Chim. Acta 1981, 64, 2704;
(d) Seebach, D.; Renaud, P. Helv. Chim. Acta 1985, 68,
2342; (e) Seebach, D.; Muller, S.; Gysel, U.; Zimmer-
mann, J. Helv. Chim. Acta 1985, 68, 2342.
6. Although both hexane and heptane are classified as
flammability 1 solvents (composition of flammable mix-
ture is 1–7 wt% in air), heptane is generally considered to
be a better solvent for scale-up because hexane has a
higher tendency to build up static charge (conductivity @
23°C: heptane 200 pS/m; hexane 10 pS/m). See:
McConville. F.X. The Pilot Plant Real Book: A Unique
Handbook for the Chemical Process Industry; FXM Engi-
neering and Design, 2002; Chapter 6.
1
6.36 min. H NMR (CDCl3) l 1.01–1.76 (m, 10H), 2.17
(m, 1H), 5.20 (bs, 1H), 7.23 (t, J=7.7 Hz, 1H), 7.33 (t,
J=7.7 Hz, 2H), 7.61 (d, J=7.7 Hz, 2H). 13C NMR l
25.57, 26.27, 26.42, 27.52, 81.15, 126.10, 127.85, 128.24,
140.03, 180.97. MS (m/e) 234 (M+). Anal. calcd for
C14H18O3: C, 71.77; H, 7.74. Found: C, 71.86; H, 7.77.
7. Mase, T.; Houpis, I. N.; Akao, A.; Dorzoitis, I.; Emer-
son, K.; Hoang, T.; Iida, T.; Itoh, T.; Kamei, K.; Kato,
S.; Kato, Y.; Kawasaki, M.; Lang, F.; Lee, J.; Lynch, J.;
Maligres, P.; Molina, A.; Nemoto, T.; Okada, S.;
Reamer, R.; Song, J. Z.; Tschaen, D.; Wada, T.; Zewge,
D.; Volante, R. P.; Reider, P. J.; Tomimoto, K. J. Org.
Chem. 2001, 66, 6775.
8. (a) Chapel, N.; Greiner, A.; Ortholand, J. S. Tetrahedron
Lett. 1991, 32, 1441; (b) Frater, G.; Moller, U.; Gonther,
W. Tetrahedron Lett. 1981, 22, 4221.
Acknowledgements
Mr. Robert Prytko is acknowledged for his help in the
scale-up of the process in the kilolab and Mr Hui Li for
analytical support. The authors acknowledge Dr. Paul
T. Grover for helpful discussions during the develop-
ment of this process and Dr. Seth Ribe for helpful
suggestions while preparing this manuscript.
9. (a) Salomaa, P.; Sallinen, K. Acta Chem. Scand. 1965, 19,
1054; (b) Farines, M.; Soulier, J. Bull. Soc. Chem. Fr.
1970, 332.
10. (a) Yee, N. K. Org. Lett. 2000, 2, 2781; (b) Kosˇmrlj, J.;
Weigel, L. O.; Evans, D. A.; Downey, C. W.; Wu, J. J.
Am. Chem. Soc. 2003, 125, 3208.
References
1. (a) Wein, A. J. Exp. Opin. Invest. Drugs 2001, 10, 65; (b)
Owens, R. G.; Karram, M. M. Drug Safety 1998, 123.
2. Yarker, Y. E.; Goa, K. L.; Fitton, A. Drug Aging 1995,
6, 243.
11. The solubility curve of dioxolone 3 in hexanes is expected
to be similar to that obtained in heptane.
12. HPLC conditions: column: Regis Whelk-O-1(R,R), 5
mm×25 cm×4.6 mm; mobile phase: hexanes/IPA 97/3,
ambient temperature at 1 mL/min; UV detection at 210
nm. Retention times: (2R,5S)-4, 5.95 min; (2S,5R)-4, 8.47
min; of (2S,5S)-3, 9.10 min; of (2R,5R)-3, 12.51 min.
13. Hazard analyses were completed to determine the reac-
tion profile and better define the scale-up conditions.
3. (a) Bakale, R. P.; Lopez, J. L.; McConville, F. X.;
Vandenbossche, C. P.; Senanayake, C. H. US Patent
6,090,971, July 18, 2000, ‘Resolution Process For Cyclo-
hexylphenyl Glycolic Acid’; (b) Bakale, R. P.; Lopez, J.
L.; McConville, F. X.; Vandenbossche, C. P.;
Senanayake, C. H. US Patent 5,973,182, Oct. 26, 1999,
‘Carbonate Intermediates Useful In The Preparation Of