50 10 mM pH ) 2.55 aqueous phosphate buffer/acetonitrile over
20 min, and then ramped to 20:80 10 mM pH ) 2.55 aqueous
phosphate buffer/acetonitrile over 10 min at 1.0 mL/min with a
Waters Xterra RP18 4.6 mm ×150 mm, 3.5 µm column, ambient
temperature, with PDA analysis at 215 nm. All reactions were run
under positive nitrogen flow. A Drager PAC III gas monitor,
equipped with an HCN sensor, was used throughout the process
development. The Robinson-Schopf reaction was scrubbed with
aqueous KOH and/or aqueous KOH/KMnO4.
vol to phenylglycinol) and warmed to a temperature of 60 °C;
1 L of water (2 volumes to phenylglycinol) was charged at a
rate that maintained the reaction solution temperature g50 °C
(45 min). The solution was cooled to 20 °C over 6 h using a
linear ramp, and the reactor contents were stirred an additional
8 h at 20 °C. Crystals began to form at a reaction solution
temperature of 44 °C. After 16 h, the reaction solution had
turned from pale-yellow to yellow, and a white, granular,
crystalline solid was present. The solids were collected and
washed with a 1:1 water/isopropanol solution (2 × 500 mL).
The resulting white, granular solid was dried at 45 °C for 14 h
to yield 411 g (53% yield, 98.34 LCAP purity, 99.6:0.4 dr (1/
1a)) of 1. KF analysis of 1 was 0.23 wt %. OVI analysis by
GC headspace revealed 1 contained 14 ppm dichloromethane
and 1090 ppm isopropanol. XRPD analysis showed a single-
crystal form was isolated. DSC analysis gave a sharp endotherm
of 79.16 °C with an onset temperature of 77.15 °C. 1H NMR
(CDCl3, 400 MHz) δ 7.34 (m, 5H), 4.25 (t, J ) 8.0 Hz, 1H),
4.12 (dd, J ) 9.8 Hz, 2.0 Hz, 1H), 3.89 (t, J ) 8.0 Hz, 1H),
3.83 (m, 1H), 3.73 (t, J ) 8.0 Hz, 1H), 2.11 (m, 1H), 1.95-1.67
(m, 4H), 1.59-1.52 (m, 1H). 13C NMR (CDCl3, 100 MHz) δ
137.0, 128.8 (2), 128.4, 127.8 (2), 115.8, 89.6, 72.8, 63.7, 47.2,
29.8, 27.8, 19.2. Calcd for C14H16N2O: C, 73.66; H, 7.06; N,
12.27. Found: C, 73.37; H, 7.03; N, 12.16.
(2-Cyano-6-phenyloxazolopiperidine, (3S,5R,8aS)-3-Phenyl-
hexahydro-2H-oxazolo[3,2-a]pyridine-5-carbonitrile), 1. A mix-
ture of distilled water (10 L), citric acid monohydrate (1991 g,
9.48 mol, 2.6 equiv), and (S)-(+)-phenylglycinol (500 g, 3.64
mol, 1.0 equiv.) was cooled to 5 °C, and glutaric dialdehyde
(1167 g, 5.83 mol, 1.6 equiv., 50% by weight in water) was
charged over 50 min at e5 °C. Dichloromethane (5 L) was
charged, and the temperature equilibrated e5 °C. A solution
of potassium cyanide (356 g, 5.47 mol, 1.5 equiv) in water (1
L) was charged over 25 min at e5 °C, and the contents were
stirred at 2-5 °C for 4 h. The temperature was maintained
below 5 °C to ensure any HCN formed during the reaction
remained in the liquid state (HCN bp ) 25.6 °C). The reaction
was quenched by charging dibasic potassium phosphate (508
g, 2.92 mol, 0.8 equiv) followed by aqueous 4.5 M potassium
hydroxide (1.2 kg of 90 wt % potassium hydroxide in 4.8 L
water). The potassium hydroxide solution was charged over 43
min at e15 °C. The pH of the reaction solution was monitored
during the last part of the base addition to ensure that the pH
did not exceed 8.13 When the pH of the reaction solution was
between 7 and 8, the phases were allowed to separate for 10
min. After removing the bottom organic phase, dichloromethane
(1.5 L) was charged and stirred for 10 min. The aqueous and
organic phases were allowed to separate for 10 min. The bottom
organic phase was removed, combined with the bulk organic
solution, and dried for 16 h with anhydrous sodium sulfate (500
g). LC analysis of the crude Robinson-Schopf condensation
reaction revealed a product purity of 1 at 46.2 LCAP with a dr
of 49.5:50.5 (1/1a). KF analysis after 12 h of drying returned
a water content of 0.28 wt %.
Reactor Cleaning. The reactor used to perform the zinc
trifluoroaceate hydrate epimerization reaction was charged with
2 L of acetonitrile, and the contents were stirred for 15 min.
The resulting red solution was removed, and 2 L of 0.5 N
aqueous HCl was charged and stirred for 15 min. After
removing the aqueous HCl solution, the reactor was charged
with 1 L of methanol and stirred for 15 min. Removal of the
methanol wash solution gave a reactor flask free of visible
inorganic and organic impurities.
Acknowledgment
We are grateful to Zicheng Yang and Jing Yuan for
their analytical support. We thank Jo Ann Wilson, Sriram
Naganathan, and Denise Andersen for helpful discussions and
critical reviews of the manuscript.
The drying agent was filtered and washed with dichlo-
romethane (250 mL). The organic solution was degassed with
dry nitrogen for 10 min, and zinc trifluoroacetate hydrate (113
g, 0.36 mol, 10 mol %, MW calculations assume 1 mol of
water) charged. The contents were stirred at 20-25 °C until a
dr of g85:15 1/1a was achieved (16 h), resulting in a dark-red
solution of 1 (83.0 LCAP, 89.7:10.3 1/1a). The organic solution
was passed through a plug of silica gel (1.6 kg Merck grade
9385, 40-60 µm, 230-400 mesh silica gel, prewashed with
dichloromethane, 3.3 wt equiv to phenylglycinol) using gentle
vacuum. An additional 5 L of dichloromethane was passed
through the silica gel plug to yield a clear, colorless organic
solution that was concentrated to a transparent oil using a rotary
evaporator (30 °C bath temperature, 25-50 Torr pressure).19
The residual dichloromethane was chased with isopropanol (3
× 500 mL) to yield a crude, white solid.20 Analysis of the solid
revealed an 87.4 LCAP purity of 1 and 90.1:9.9 dr (1/1a). The
resulting crude solid was suspended in isopropanol (1.5 L, 3
Supporting Information Available
This material is available free of charge via the Internet at
Received for review August 12, 2010.
OP1002227
(19) Our kilolaboratory manufacturing used a rotary evaporator to exchange
dichloromethane for IPA for the sake of convenience; however, two larger-
scale manufacturing campaigns (based on 5 kg of phenyglycinol, ZnBr2 as
the epimerization reagent) at our CMO demonstrated that an atmospheric
pressure solvent exchange from dichloromethane to IPA does not adversely
affect the yield (54%, 50%) or quality (> 98% LCAP, > 99:1 dr) of the
finished product (1). The CMO noted the difficulty cleaning the reactors post-
zinc equilibration (a phenomenon we also experienced in our kilolaboratory)
which prompted us to search for the alternative epimerization catalysts.
(20) It was critical to remove as much dichloromethane as possible before
the isopropanol/water recrystallization. The high solubility of 1 in
dichloromethane (>35 mg/mL) can lead to depressed product yields
if dichloromethane is not effectively removed.
1478
•
Vol. 14, No. 6, 2010 / Organic Process Research & Development