Process development and scale-up for (±)-reboxetine mesylate

Article abstract of DOI:10.1021/op7000063

Redevelopment of the commercial process for the synthesis of (±)-reboxetine methanesulfonate is described. An optimized and efficient process for the synthesis of (±)-reboxetine starting from cinnamyl alcohol was developed. The redeveloped process minimizes impurity formation and utilizes simplified processing to substantially improve process yield and throughput, and is suitable for the efficient synthesis of multiton quantities of reboxetine.

Full text of DOI:10.1021/op7000063

Organic Process Research & Development 2007, 11, 346−353
Process Development and Scale-up for (()-Reboxetine Mesylate
Kevin E. Henegar,* Cynthia T. Ball,† Carolyn M. Horvath, Keith D. Maisto,‡ and Sarah E. Mancini§
Pfizer Global Research and DeVelopment, 2800 Plymouth Road, Ann Arbor, Michigan 48105, U.S.A.
Abstract:
Process redevelopment was initiated, anticipating U.S.
Redevelopment of the commercial process for the synthesis of
(()-reboxetine methanesulfonate is described. An optimized and
efficient process for the synthesis of (()-reboxetine starting from
cinnamyl alcohol was developed. The redeveloped process
minimizes impurity formation and utilizes simplified processing
to substantially improve process yield and throughput, and is
suitable for the efficient synthesis of multiton quantities of
reboxetine.
approval of reboxetine and the need for a more efficient
process to prepare multiton amounts of reboxetine per year.
Since a commercial process for reboxetine had already been
registered, there were significant constraints on the extent
of process changes that were acceptable.
Results and Discussion
Epoxidation of cinnamyl alcohol3 was originally done
with monoperoxyphthalic acid. This reaction proceeded well,
but preparation of the reagent from phthalic anhydride and
hydrogen peroxide was time-consuming, and a large amount
of phthalic acid was formed as a byproduct, creating a
disposal issue. A variety of alternate reagents were examined
for this conversion, and ultimately it was found that
epoxidation with commercially available 40% peracetic acid
proceeded smoothly in methylene chloride in the presence
of sodium carbonate. The epoxide 3 is highly sensitive to
water and acidic conditions. The sulfuric acid stabilizer in
commercial peracetic acid was initially neutralized by
pretreatment with sodium carbonate, but later experiments
showed that this was unnecessary. An excess of solid sodium
carbonate was used as a base to neutralize the acetic acid
and also served to remove water from the reaction. The main
impurity formed in the reaction was phenylglycerol, formed
by hydrolysis of the epoxide. At the end of the reaction the
inorganic salts, a mixture of sodium carbonate, sodium
bicarbonate, and sodium acetate, were removed by filtration;
since these byproducts were water soluble and innocuous,
they were easily disposed of.
In the original process, the phenylglycidol solution was
concentrated to an oil and then reacted with 2-ethoxyphenol
in aqueous NaOH at 70 °C without solvent. After an aqueous
workup, the product was isolated by crystallization from
toluene. This process provided 4 in good purity, but the
epoxide-opening reaction was run under poorly controlled
conditions. The recovery of product from toluene was low
due to the high solubility of 4 in toluene, and the product
slurry filtered poorly. DSC experiments showed a low onset
temperature for the isolated phenylglycidol made with
peracetic acid, and accordingly the process was modified to
eliminate concentration to an oil. After filtration to remove
the inorganic salts, the solution of 3 in methylene chloride
was reacted with 2-ethoxyphenol and NaOH. Heating was
required to give a reasonable reaction rate, and even at 40
°C the reaction takes many hours to go to completion. To
shorten the cycle time methylene chloride was distilled as
Introduction
Reboxetine mesylate (1) is a selective norepinephrine
reuptake inhibitor, currently approved in over 60 countries
as an antidepressant, and is marketed under the trade names
Edronax, Norebox, Prolift, Vestra, and Integrex in Europe
and Latin America. The original commercial process closely
followed the originally published synthesis of reboxetine,
and is shown in Scheme 1.1,2 Cinnamyl alcohol 2 was
epoxidized with monoperoxyphthalic acid to give racemic
phenylglycidol 3, and the epoxide was reacted with 2-ethoxy-
phenol to give the diol 4. The primary alcohol of the diol
was protected as the p-nitrobenzoate ester 5, and then the
secondary alcohol was mesylated in situ, yielding 6. Base
treatment of 6 hydrolyzed the p-nitrobenzoate and formed
the epoxide 7. Reaction of the epoxide with ammonia yielded
the amino alcohol 8, isolated as the methanesulfonate salt.
Conversion of the amino alcohol to reboxetine was done by
chloroacetylation to form 9 followed by base treatment to
give the lactam 10, and finally reduction with Vitride and
salt formation yielded (()-reboxetine methanesulfonate.
This process suffers from poor yields and complicated
processing which leads to low process throughput. Key areas
for process improvement were as follows: (1) an improved
oxidation procedure for the conversion of cinnamyl alcohol
2 to the epoxide 3; (2) more controlled conditions for the
opening of the epoxide to form diol 4 and an improved
crystallization procedure for 4; (3) a more selective protection
protocol since acylation of the secondary alcohol produces
about 13% of the wrong diastereomer of 8, resulting in
diminished yields and a complicated isolation procedure for
8; and (4) minimizing formation of a dimeric impurity in
the final Vitride reduction of 10. This impurity was hard to
remove and substantially reduced the overall yield.
* Author for correspondence. E-mail: kevin.e.henegar@pfizer.com.
† Current address: Ben Venue Laboratories, 300 Northfield Rd., Bedford,
Ohio 44146.
‡ Current address: Retired.
§ Current address: Pfizer Global Manufacturing, Kalamazoo Michigan 49001.
(1) Melloni, P.; Della Torre, A.; Lazzari, E.; Mazzini, G.; Meroni, M.
Tetrahedron 1985, 41, 1393-1399.
(2) All compounds in this process are strictly racemic. For clarity, only a single
enantiomer is shown.
(3) Commercial cinnamyl alcohol was typically >99.5% trans stereoisomer.
The epoxidation proceeds with complete stereospecificity.
346
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development
10.1021/op7000063 CCC: $37.00 © 2007 American Chemical Society
Published on Web 03/23/2007

Scheme 1a
a Reagents and conditions: (a) monoperoxyphthalic acid, EtOAc; (b) 2-ethoxyphenol, NaOH; (c) p-nitrobenzoyl choloride, pyrdine; (d) MsCl, Et3N; (e) NaOH,
dioxane; (f) i. NH3; ii. CH3SO3H; (g) CICH2COCI, Et3N; (h) t-BuOK; (i) i. Vitride; ii. CH3SO3H.
the reaction proceeds, and the reaction was driven to
completion by heating at 60 °C.
original process, p-nitrobenzoate was used to protect the
primary alcohol of 3. Selectivity for this protection was poor,
and up to 13% of the secondary monobenzoate formed,
leading to the diastereomer, which complicated purification
and resulted in yield loss. In addition, a 20% excess of the
p-nitrobenzoyl chloride was used to drive the reaction to
completion which later formed about 20% of the bis-p-
nitrobenzoate ester, compounding the yield loss.
Optimization experiments showed no difference in reac-
tion rate or impurity levels for KOH and NaOH. A survey
of phase transfer catalysts showed that Bu4NCl, BnMe3NCl,
Bu3MeNCl, Aliquat 336, and BnNMe3OH gave comparable
rates of reaction and amounts of impurities present. Bu3-
MeNCl solution was chosen for convenience. Temperature
effects were studied in reactions at 50 °C, 60 °C, and 70 °C.
The reaction proceeds faster at higher temperatures, but more
impurities form, and thus, 60 °C was chosen as a compromise
between reaction time and impurity levels.
MTBE was found to be superior to toluene for the
crystallization, with better product recovery and better
filtering solids. Low water content was critical for high
recovery from MTBE. Operationally, it was found to be more
efficient to extract the product into toluene and then exchange
the solvent to MTBE for the crystallization than to extract
with MTBE, due to the lower water solubility and more
favorable water azeotrope in toluene vs that in MTBE.
Compound 4 was isolated in about 65% overall yield from
cinnamyl alcohol.4
Attempts were made to improve the selectivity of the
nitrobenzoylation. The intrinsic selectivity of the p-nitroben-
zoylation is almost irrelevant since the primary and secondary
esters rapidly equilibrate under the reaction conditions.
Lowering the reaction temperature did not alter the selectiv-
ity. A variety of bases were examined including triethy-
lamine, N-methylpiperdine, diethylaniline, N-methylmor-
pholine, pyridine, dimethylbenzylamine, and quinoline. Use
of weaker bases merely slowed the reaction rate but did not
improve selectivity. Some variation was observed in the
amount of bis-p-nitrobenzoate formed, but no remarkable
improvements were seen. Mesylation and cleavage of the
p-nitrobenzoate were facile. The monomesylate was then
converted to the epoxide 7 using aqueous sodium hydroxide
in dioxane. This reaction was very slow (18 h), and due to
the high boiling and high freezing points of dioxane (100-
102 °C and 10-12 °C, respectively), removal of the solvent
was both time-consuming and a safety concern since the
dioxane can freeze and damage the condenser if the jacket
temperature is too cold. On a production scale the distillation
of the dioxane took as much as 36 h per batch.
Selective protection of the primary alcohol in compound
4 is essential to generate the correct diastereomer of
compound 8 since formation of the primary monomesylate
results in the formation of the diastereomer of 8. In the
(4) The isolated diol was a single diastereomer (diastereomer <0.1%), and there
was no detectable regioisomer derived from reaction of 2-ethoxyphenol at
C-2 of the phenylglycidol.
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
347

Scheme 2a
a Reagents and conditions: (a) (CH3)3SiCl, Et3N; (b) CH3SO2Cl, Et3N; (c) HCl; (d) NaOH, PTC.
Scheme 3a
a Reagents and conditions: (a) NH3, CH3OH.
Table 1
Alternatives to the use of p-nitrobenzoate for protection
equiv of NH3
8:14 (area %)
of the primary alcohol were considered. Trialkylsilyl ethers
such as TBDMS or TIPS should accomplish high primary
25
76:24
88:12
91:9
vs secondary selectivity but are too expensive for use in this
application. Trimethylsilyl (TMS) protecting groups were
examined, although there is little precedent for high selectiv-
ity by TMS for primary over secondary alcohols. Typically,
trimethylsilylating reagents are used where the intent is
exhaustive hydroxy silylation. Selectivity has been seen for
reactions of secondary alcohols in different environments.5,6
Unexpectedly, it was found that the primary alcohol of the
compound 4 could be protected in very high regioselectivity
using trimethylsilyl chloride and triethylamine at low tem-
peratures.
Reaction of 4 with trimethylsilyl chloride (TMSCl) and
triethylamine was very rapid even at -20 °C and was
essentially complete in <10 min at this temperature to give
11 (Scheme 2). Some silyl migration occurred with extended
processing times, and the selectivity deteriorated at higher
temperature or in contact with excess base. Accordingly, the
silylation was run as rapidly as possible while staying within
the temperature range, and the solution was not held before
proceeding with the mesylation. To further minimize silyl
migration, the contact time with triethylamine was mini-
mized; methanesulfonyl chloride was added first and then
the second portion of triethylamine. Hydrolysis of the TMS-
mesylate 12 yielded the monomesylate 13. The monosilyl
ether 11 and the silyl ether-mesylate 12 were isolated for
characterization but in normal processing were reacted in
situ. The selectivity of the silylation was conclusively
50
50 + saturated with NH3 gas
determined by conversion to the epoxide 7, where <0.5%
of the diastereomer was found by HPLC analysis.
Ethyl acetate was chosen as the solvent for the reaction
since this was solvent used in the original registered process.
Ethyl acetate was inert to the silylation and mesylation reac-
tions, and control experiments showed that transesterification
of 4 to yield the monoacetate was a slow process, even under
conditions of extended distillation. Some transesterification
can occur on production scale during distillation to remove
water before start of the TMSCl addition.
The mixture was quenched with hydrochloric acid.
Hydrolysis of the TMS ether was not instantaneous and
required about 30 min. When the TMS ether cleavage was
complete, the solution was washed with sodium bicarbonate
and sodium chloride solutions. The ethyl acetate was distilled
and replaced with toluene since ethyl acetate will react in
the next step.
Conversion of 13 to the epoxide 7 was found to proceed
smoothly under phase transfer conditions. As before,
BnMe3NCl was used as the catalyst, and the reaction was
run in toluene with aqueous NaOH as the base. The reaction
was rapid and essentially quantitative. These conditions
eliminate dioxane from the process.
The main side reaction in the ammonolysis was formation
of the secondary amine dimer 14 (Scheme 3). The amount
of dimer 14 formed was directly related to the amount of
ammonia used in the reaction, as shown in Table 1. Reaction
(5) Schneider, H. J.; Horning, R. Leibigs Ann. Chem. 1974, 1864-1871.
(6) Yankee E. W.; Axen, U.; Bundy, G. L. J. Am. Chem. Soc. 1974, 5865-
5876.
348
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

Scheme 4
Table 2
mode of addition
temperature did not have an effect on dimer formation;
however, at elevated temperatures the rate of reaction was
faster. The final optimized conditions use 45 equiv of
ammonium hydroxide to produce about 14% of dimer 14
with none detected after workup and isolation. 8 was isolated
as the methanesulfonate salt by crystallization from IPA. It
was also possible to isolate 8 as the free base, but isolation
as the mesylate was more reliable and gave better-handling
solids. Unexpectedly, 8 mesylate was found to be the
monohydrate.
equiv Vitride
area % 15
inverse
inverse
inverse
inverse
inverse
inverse
normal
2.5
3
3.5
4
4.5
5
5
1.02
0.63
0.64
0.48
0.28
0.06
0.43
and the structure was determined to have the dimeric
structure shown, as an indeterminate mixture of diastereo-
mers, by NMR and mass spectral data. This dimer is an
unqualified impurity with a specification of NMT 0.1%.
Amounts typically produced in the processing were about
0.5%, and this was removed from the process stream by
controlled pH extractions, causing loss of substantial amounts
of the reboxetine that could not readily be recovered. The
reduction was normally run with 2.4 equiv of Vitride;
although less should be required in theory, in practice this
was about the minimal amount required for complete
reduction. The reduction was initially rapid and then slowed
substantially, requiring several hours to go to completion
even though there was still a substantial amount of active
hydride present as evidenced by hydrogen evolution during
the quench.
A plausible mechanism for the formation of the dimer is
shown in Scheme 4. Key observations were that adding
Vitride to 10 gave dimer 15 regardless of Vitride equivalents
and that inverse addition of Vitride (2.4 equiv) also gave
high dimer. We felt that if the imine could be intercepted
by hydride before it could tautomerize to the enamine, then
dimer formation could be suppressed. In the key experiment,
inverse addition to 5 equiv of Vitride gave very low dimer
formation. The results from a definitive set of experiments
are shown in Table 2.
No significant changes were made to the conversion of 8
to 9 to minimize the regulatory impact of the proposed
process changes. The reaction was done in dimethyl carbon-
ate, which had been previously implemented as a replacement
for methylene chloride due to emission concerns. These
conditions were far from ideal because of the reactivity of
dimethyl carbonate and its high mp (2-4 °C), which severely
constrains reaction temperature. Conditions for this reaction
were optimized in the synthesis of (S,S)-reboxetine (subse-
quent article) to eliminate the formation of colored material
and impurities derived from reaction of 8 with dimethylcar-
bonate.
Lactamization to give 10 was done with potassium tert-
butoxide in IPA. This reaction was originally run at low
temperature ostensibly to avoid impurity formation. The main
impurities formed were the trivial product arising from
isopropoxide displacement of the chloride and a number of
cyclic and acyclic dimers. A detailed study of the reaction
showed that virtually anything that slows the reaction led to
elevated impurity formation and that, in fact, the reaction
ran best at high temperatures. The preferred conditions call
for very rapid addition of the tert-butoxide-IPA solution to
the amide, which caused very rapid cyclization to the lactam.
The primary issue in the final Vitride7 reduction was the
formation of the RRT 3.25 impurity 15. This was isolated,
Addition of a solution containing 1.2 equiv of aqueous
NaOH and allowing the exotherm from the quench to heat
the mixture to as high as 45-55 °C was sufficient to dissolve
(7) Vitride is sodium bis(2-methoxyethoxy)aluminum dihydride. NaAlH2(OCH2-
CH2OCH3)2, also known as Red-Al, and was purchased as a 65 wt %
solution in toluene.
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
349

Scheme 5a
a Reagents and conditions: (a) CH2CO3H, Na2CO3 CH2Cl2; (b) 2-ethoxyphenol NaOH; (c) i. TMSCl, Et3N. ii. MsCl, Et3N. iii. HCl; (d) NaOH, PTC, CH2Cl2; (e)
i. NH3. ii CH3SO3H; (f) ClClH2COCl, Et3n; (g) t-BuOK; (h) i. Vitride. ii. CH3SO3H.
1
all aluminum solids or gels and produced a clean phase break
in the subsequent extractions. Rewarming the mixture also
produced clean phase breaks if solids or gels formed upon
cooling. The toluene solution after the Vitride quench was
concentrated and the solvent exchanged to acetone. The
calculated amount of methanesulfonic acid was added,
causing precipitation of the final product, which was filtered
and dried.
run on Varian INOVA operating at 400 MHz for H and
100 MHz for 13C. Mass spectra were recorded on a
Micromass Platform LC mass spectrometer.
(2RS,3SR)-3-(2-Ethoxyphenoxy)-2-hydroxy-3-phenyl-
1-propanol (4). Sodium carbonate (420 kg, 3962 mol),
cinnamyl alcohol (375 kg, 2794 mol), and methylene chloride
(3700 L) were charged to an 8000 L reactor. The internal
temperature was adjusted to 15-20 °C. A solution of
peracetic acid (40% in acetic acid-water, 717 kg, 3773 mol)
was added over about 3 h, maintaining the temperature
between 20 and 25 °C. The mixture was stirred for 3 h, and
then a solution of sodium sulfite (300 kg dissolved in 2250
L water) was added, maintaining the temperature <30 °C.
At the end of the quench the temperature was adjusted to
30-35 °C. The phases were separated, and the aqueous phase
was extracted with 375 L of methylene chloride. The organic
phases were combined and added to a mixture of water (1530
L), 50% NaOH (240 kg, 3000 mol), tributylmethylammo-
nium chloride (50 kg), and 2-ethoxyphenol (575 kg, 4166
mol). The mixture was agitated and heated to 60 °C and the
methylene chloride distilled. The mixture was held at 60 °C
until reaction was complete (about 4 h) and then was cooled
to 25 °C. Toluene (2350 L) was added, and the phases were
separated. The aqueous phase was extracted with 1500 L of
toluene. The organic phases were combined and washed with
aqueous NaOH (2 × 60 kg 50% NaOH in 750 L of water)
and then with 750 L of water. The toluene solution was
distilled to an oil, and then MTBE (1425 L) was added, and
Conclusion
An efficient process was developed for the large-scale
synthesis of reboxetine methanesulfonate. Key improvements
eliminated impurity formation, leading to more efficient
processing and higher yields relative to the original registered
process. Overall chemical yield was increased from about
11% yield in the original process to about 25% in the
modified process, substantially increasing process throughput.
The optimized scheme for the final process is shown in
Scheme 5.
Experimental Section
Materials were obtained from commercial suppliers and
used without purification. Production-scale procedures were
run in 4000-L or 8000-L nominal volume, glass-lined reactors
equipped with typical utilities. Products were isolated on
agitated stainless steel pressure nutsches and dried with
positive pressure nitrogen. Melting points were run in open
tube capillaries and are uncorrected. IPA refers to 2-propanol;
MTBE refers to methyl-tert-butyl ether. NMR spectra were
350
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

the mixture was seeded and cooled to 0 °C to crystallize.
The solids were filtered and washed with 0 °C MTBE (2 ×
375 L) and then dried at 30 °C with nitrogen. Yield: 525
kg of 4 as a white solid (65% overall yield from cinnamyl
alcohol).
A sample of material from a smaller-scale run was
characterized: mp 71-74 °C. 1H NMR (399.76 MHz,
CDCl3) δ 0.09 (s, 9H), 1.50 (t, J ) 7.0 Hz, 3H), 3.25-3.29
(m, 1H), 3.39 (d, J ) 8 Hz, 1H), 3.63-3.70 (m, 1H), 3.87-
3.99 (m, 2H), 4.1 (q, J ) 7.0 Hz, 2H), 5.25 (d, J ) 4.1 Hz,
2H), 6.61-6.4 (m, 1H); 6.68-6.74 (m, 1H), 6.86-6.92 (m,
2H), 7.27-7.42 (m, 5H).13C NMR (100.53 MHz, CDCl3) δ
15.0. 62.5, 64.4, 86.3, 99.9, 112.8, 116.54, 121.0, 122.5,
126.6, 128.3, 128.9, 138.2, 147.3, 149.3. LRMS-APCI m/z
calcd for C17H21O4 (M + H)+: 289. Found: m/z ) 289 [M
+ 1]+. Anal. Calcd for C17H20O4: C, 70.81; H, 6.99.
Found: C, 70.85; H, 7.07.
(91.2%) of the product as an oil that solidified on standing;
mp 80-82.5 °C. 1H NMR (400.13 MHz, CDCl3) δ 0.17 (s,
9H), 1.50 (t, J ) 6.8 Hz, 3H), 3.06 (s, 3H), 3.77 (dd, J )
11, 6 Hz, 1H), 4.00 (dd, J ) 11, 6 Hz, 1H), 4.10 (q, J ) 6.8
Hz, 2H), 5.07 (m, 1H), 5.51 (d, J ) 4.4 Hz, 1H), 6.75 (m,
2H), 6.91 (m, 2H), 7.2-7.49 (m, 5H). 13C NMR (100.62
MHz, CDCl3) δ 0.1, 15.66, 38.87, 61.57, 64.88, 79.90, 85.20,
113.97, 116.99, 121.32, 122.79, 128.26, 129.09, 129.14,
147.72, 149.95. LRMS-APCI m/z calcd for C21H31O6SSi (M
+ H)+: 439. Found: m/z ) 439 [M + 1]+.
(2RS,3SR)-3-(2-Ethoxyphenoxy)-2-mesyloxy-3-phenyl-
1-propanol (13). 3-(2-Ethoxyphenoxy)-2-hydroxy-3-phe-
nylpropanol (0.288 g, 1 mmol) and triethylamine (0.15 mL,
1.1 mmol) were dissolved in ethyl acetate (5 mL) and cooled
to -17 °C. Trimethylsilyl chloride (0.13 mL, 1.0 mmol)
dissolved in 2 mL of ethyl acetate was added over 10 min,
keeping the temperature below -15 °C. A white precipitate
formed during this addition. The mixture was stirred below
-15 °C for 15 min. Triethylamine (0.15 mL, 1.1 mmol) was
added, followed by methanesulfonyl chloride (0.085 mL, 1.1
mmol) dissolved in 2 mL of ethyl acetate, keeping the
temperature below -15 °C. The mixture was stirred below
-15 °C for 15 min. Hydrochloric acid (2M, 2 mL) was
added, and the mixture was allowed to warm to 20-25 °C
and stirred for 30 min. The phases were separated, and the
organic phase was washed with aqueous sodium chloride
solution (5 mL) and dried over sodium sulfate. The solution
was evaporated to yield 0.377 g of an oil. The oil was
chromatographed on silica (230-400 mesh), eluting with 1:1
hexane-ethyl acetate. The product-containing fractions were
concentrated to yield 0.33 g (91%) of the product as an oil
that solidified on standing; mp 83-86 °C. 1H NMR (400.13
MHz, CDCl3) δ 1.66 (t, J ) 8.2 Hz, 3H), 2.85 (s, 3H), 4.14-
4.35 (m, 4H), 5.12 (m, 1H), 5.52 (d, J ) 6.1 Hz, 1H), 6.8-
7.15 (m, 4H), 7.5-7.7 (m, 5H). 13C NMR (100.62 MHz,
CDCl3) δ 14.73, 37.80, 62.19, 64.27, 81.40, 84.04, 112.88,
117.19, 120.67, 122.86, 127.40, 128.77, 128.86, 146.40,
149.30. LRMS-APCI m/z calcd for C18H23O6S (M + H)+:
367. Found: m/z ) 367 [M + 1]+.
(1RS, 2SR) 3-Amino-1-(2-ethoxy-phenoxy)-1-phenyl-
propan-2-ol (8). 4 (250 kg, 868 mol), triethylamine (105
kg, 1037 mol), and ethyl acetate (2500 L) were charged to
an 8000 L reactor. The mixture was agitated and cooled to
-15 to -20 °C. Trimethylsilyl chloride (98 kg, 903 mol)
was added, maintaining the temperature between -15 and
-20 °C. The mixture was stirred for 15 min after completion
of the trimethylsilyl chloride addition. Methanesulfonyl
chloride (119 kg, 1039 mol) was added, and then triethy-
lamine (105 kg, 1037 mol) was added, maintaining the
temperature at -15 to -20 °C. The mixture was stirred for
15 min after completion of the triethylamine addition, and
then a solution of HCl (85 kg of concd HCl in 799 L of
water) was added. The mixture was heated to between 15
and 35 °C for 45 min, and then the phases were separated.
Sodium bicarbonate solution (43 kg of NaHCO3 in 434 L of
water) was added to the organic phase. The mixture was
agitated, and then the phases were separated. The organic
phase was washed with a solution of NaCl (108 kg in 334 L
(2RS,3SR)-3-(2-Ethoxyphenoxy)-2-hydroxy-3-phenyl-
1-(trimethylsilyloxy)propane (11). 3-(2-Ethoxyphenoxy)-
2-hydroxy-3-phenylpropanol (1.44 g, 5 mmol) and triethy-
lamine (0.77 mL, 5.5 mmol) were dissolved in ethyl acetate
(15 mL) and cooled to -17 °C. Trimethylsilyl chloride (0.64
mL, 5.0 mmol) dissolved in 5 mL of ethyl acetate was added
over 10 min, keeping the temperature below -15 °C. A white
precipitate formed during this addition. The mixture was
stirred below -15 °C for 15 min, and the 20 mL of pentane
was added. The solids were removed by filtration, and the
filtrate was concentrated under vacuum to yield a cloudy
oil. The oil was chromatographed on silica (230-400 mesh),
eluting with 4:1 heptane-ethyl acetate. The product-contain-
ing fractions were combined and evaporated to yield 1.80 g
1
(88.5%) of the product as a clear, colorless oil. H NMR
(400.13 MHz, CDCl3) δ 0.09 (s, 9H), 1.47 (t, J ) 6.8 Hz,
3H), 2.82 (d, J ) 5.2 Hz, 1H), 3.80 (m, 3H), 4.0-4.11 (m,
4H), 5.08 (d, J ) 6.0 Hz, 1H), 6.76 (m, 2H), 7.2-7.45 (m,
5H).. 13C NMR (100.62 MHz, CDCl3) δ 0.0, 15.54, 63.34,
65.06, 75.22, 83.71, 114.28, 118.60, 121.51, 122.95, 127.84,
128.49, 128.84, 138.93, 148.34, 150.40. LRMS-APCI m/z
calcd for C20H29O4Si (M + H)+: 361. Found: m/z ) 361
[M + 1]+.
(2RS,3SR)-3-(2-Ethoxyphenoxy)-2-mesyloxy-3-phenyl-
1-(trimethylsilyloxy)propane (12). 3-(2-Ethoxyphenoxy)-
2-hydroxy-3-phenylpropanol (1.44 g, 5 mmol) and triethy-
lamine (0.77 mL, 5.5 mmol) were dissolved in ethyl acetate
(15 mL) and cooled to -17 °C. Trimethylsilyl chloride (0.64
mL, 5.0 mmol) dissolved in 5 mL of ethyl acetate was added
over 10 min, keeping the temperature below -15 °C. A white
precipitate formed during this addition. The mixture was
stirred below -15 °C for 15 min. Triethylamine (0.8 mL,
5.7 mmol) was added, followed by methanesulfonyl chloride
(0.46 mL, 6.0 mmol) dissolved in 5 mL of ethyl acetate,
keeping the temperature below -15 °C. The mixture was
stirred below -15 °C for 15 min, and the 20 mL of pentane
was added. The solids were removed by filtration, and the
filtrate was concentrated under vacuum to yield a cloudy
oil. The oil was chromatographed on silica (230-400 mesh),
eluting with 4:1 heptane-ethyl acetate. The product-contain-
ing fractions were combined and evaporated to yield 2.00 g
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
351

water). The organic phase was concentrated, and toluene (2
× 1506 kg) was added and distilled to a final volume of
1700 L. Water (470 L), 50% NaOH (312 kg), and tributyl-
methylammonium chloride (21.7 kg) were added, and the
mixture was stirred at 20-25 °C for about 2 h. The phases
were separated, and the aqueous phase was extracted with
376 kg of toluene. The toluene solutions were combined and
washed with NaCl solution (108 kg of NaCl in 334 L of
water). The organic phase was concentrated to a volume of
520 L. Methanol (2 × 2604 L) was added and distilled to a
final volume of 520 L.
A sample of epoxide prepared from a smaller scale run
was characterized. (2RS,SR)-[(2-Ethoxyphenoxy)phenyl-
methyl]oxirane (7): 1H NMR (400.13 MHz, CDCl3) δ 1.44
(t, J ) 7.0 Hz, 3H), 2.73 (dd, J ) 2.6, 5.0 Hz, 1H), 2.81
(dd, J ) 4.2, 5.0 Hz, 1H), 3.48 (m, 1H), 4.1 (q, J ) 7.0 Hz,
2H), 4.92 (d, J ) 6.0 Hz, 1H), 6.75-6.93 (m, 4H), 7.29-
7.48 (m, 5 H). 13C NMR (100.62 MHz, CDCl3) δ 15.22,
44.84, 55.31, 64.97, 76.99, 83.56, 114.65, 118.64, 121.18,
122.87, 127.05, 128.51, 128.77, 137.98, 138.00, 148.05,
150.22. LRMS-APCI m/z calcd for C17H19O3 (M + H)+:
271. Found: m/z ) 2 [M + 1]+. Anal. Calcd for C17H18O3:
C, 75.53; H, 6.71. Found: C, 75.32; H, 6.77.
H2O. Anal. calcd for C17H21NO3-CH3SO3H-H2O: C, 53.85;
H, 6.68; N, 3.49. Found: C, 53.81; H, 6.59; N, 3.42.
1
Free base: mp 107-110 °C. H NMR (400.13 MHz,
CDCl3) δ 1.49 (t, J ) 7.0 Hz, 3H), 2.56 (dd, J ) 13.0, 6.4
Hz, 1H), 2.66 (dd, J ) 13.0, 3.5 Hz, 1H), 3.94 (m, 1H), 4.1
(t, J ) 7.0 Hz, 2H), 4.75 (d, J ) 7.99 Hz, 1H), 4.58 (d, J )
8.2 Hz, 1H), 6.60-6.93 (m, 4H), 7.31-7.39 (m, 5H). 13C
NMR (100.53 MHz, CDCl3) δ 15.07, 43.45, 64.49, 76.60,
87.84, 113.20, 119.73, 121.07, 123.48, 127.50,128.54, 128.81,
138.99, 148.34, 150.25. LRMS-APCI m/z calcd for C17H21-
NO3 (M + H)+: 364. Found: m/z ) 364 [M + 1]+. Anal.
Calcd for C17H21NO3: C, 71.06; H, 7.37; N, 4.87. Found:
C, 70.95; H, 7.47; N, 4.81.
Reboxetine Methanesulfonate (1). 8 methanesulfonate
monohydrate (125 kg, 298 mol) and 1534 L of dimethyl
carbonate were charged to a 4000 L reactor. Triethylamine
(103 kg, 1018 mol) was added, and the contents were cooled
to 4-10 °C. A solution of chloroacetyl chloride (30 kg, 265.5
mol) in 44 L of dimethyl carbonate was added, maintaining
the temp between 4 and 10 °C. After this addition was
completed, chloroacetyl chloride (30 kg, 265.5 mol) in 47 L
of dimethylcarbonate was added, maintaining the temperature
between 4 and 10 °C. After 30 min, 1250 L of water was
added, and the mixture was stirred for 15 min and then
allowed to settle. The aqueous phase was removed, and the
organic phase was washed with an aqueous solution of NaCl
(2 × (25 kg of NaCl dissolved in 1225 L of water)). The
organic phase was concentrated under vacuum to a volume
of 500 L.
A sample of the amide was characterized. 2-Chloro-N-
[3-(2-ethoxyphenoxy)-2-hydroxy-3-phenylpropyl]-aceta-
mide (9): mp 115-117.5 °C. 1H NMR (400.13 MHz, CDCl3)
δ 1.48 (t, J ) 7.0 Hz, 3H), 3.16-3.31 (m, 2H), 3.97 (s,
2H), 4.05-4.12 (m, 3H), 4.31 (s, 1H), 4.58 (d, J ) 8.2 Hz,
1H), 6.56-6.93 (m, 4H), 7.29-7.48 (m, 5 H). 13C NMR
(100.62 MHz, CDCl3) δ 15.06, 41.21, 42.83, 64.49, 73.88,
87.77, 113.16, 120.33, 121.14, 124.04, 127.55, 129.05,
129.10, 137.87, 147.90, 150.32, 166.10. LRMS-APCI m/z
calcd for C19H23ClNO4 (M + H)+: 364. Found: m/z ) 364
[M + 1]+. Anal. Calcd for C19H22ClNO4: C, 62.72; H, 6.09;
Cl, 9.74; N, 3.85. Found: C, 62.66; H, 6.02; Cl, 9.56; N,
3.83.
IPA (813 L) and tert-butanol (48 kg) were charged to a
4000 L reactor and held at about 20 °C. Potassium isopro-
poxide (319.4 kg of a 20% solution) in IPA was added. The
IPA solution was then transferred to the slurry of 9,
maintaining the temp >20 °C. After 1 h, an 8% solution of
HCl in water was added to a pH between 4 and 7.2 (about
164 L). The mixture was then distilled under vacuum to a
volume <1000 L and then cooled to <45 °C. Toluene (376
L) and water (384 L) were added. The phases were separated,
and the aqueous phase was extracted with 2 × 376 L of
toluene. The toluene phases were combined and washed with
a solution of 5 kg of NaCl in 1105 L of water. The phases
were separated, and the toluene solution was washed with
10% NaCl (2 × 419 kg). The toluene solution was then
distilled under vacuum and the final volume adjusted to 600
L with toluene.
The epoxide solution was diluted with 2344 L of
methanol, and 2347 kg of 28% aqueous ammonia was added.
The mixture was heated to 40 °C for 3 h. The mixture was
cooled to 15-25 °C, and 1938 L of methylene chloride was
added. The aqueous phase was extracted with methylene
chloride (2 × 868 L). The combined methylene chloride
phases were distilled to a volume of about 2600 L. Methylene
chloride (1562 L) was added and the mixture again distilled
to 2600 L. Methylene chloride (868 L) and water (2170 L)
were added. The mixture was agitated, and the aqueous phase
was separated and discarded. A solution of 103 kg of concd
HCl in 2170 L of water was added. The aqueous phase was
separated, and the organic phase was extracted with 2170 L
of water. The combined aqueous phases were extracted with
400 L of methylene chloride. Methylene chloride (1250 L)
and 86.7 kg of 50% NaOH were added to the aqueous phase.
The phases were separated, and the aqueous phase was
extracted with 625 L of methylene chloride. The methylene
chloride solutions were combined and distilled under vacuum.
IPA (2 × 1736 L) was added and distilled to a volume of
1736 L. Methanesulfonic acid (68.6 kg, 599 mol) was added,
and the mixture was stirred for about 4 h, then cooled to 0
°C, and stirred for 1 h. The solids were filtered and washed
with 868 L of IPA cooled to 0 °C. The solids were dried
with 60 °C nitrogen to yield 229 kg of 8 methanesulfonate
1
monohydrate (63% overall from 4) mp 122-125 °C. H
NMR (400.13 MHz, CDCl3) δ 2.30 (s, 3H), 2.67-2.83 (m,
2H), 3.26 (br s, 2H), 4.00-4.08 (m, 3H), 5.25 (d, J ) 5.1
Hz, 1H), 5.82 (br s, 1H), 6.68-6.94 (m, 4H) 7.29-7.48 (m,
5 H), 7.74 (br s, 3H). 13C NMR (100.62 MHz, CD3OD) δ
14.03, 38.27, 42.08, 64.50, 70.46, 84.15, 113.46, 117.02,
120.79, 122.43, 127.24, 128.31, 128.40, 137.48, 147.18,
149.34. LRMS-APCI m/z calcd for C17H22NO3 (M + H)+:
364. Found: m/z ) 364 [M + 1]+. KF calcd for C17H21-
NO3-CH3SO3H-H2O: 4.49 wt % H2O; found: 4.48 wt %
352
•
Vol. 11, No. 3, 2007 / Organic Process Research & Development

A sample of the lactam was characterized as a colorless
glass. 6-[(2-Ethoxyphenoxy)phenylmethyl]-3-morpholi-
none (10): 1H NMR (400.13 MHz, CDCl3) δ 1.41 (t, J )
7.0 Hz, 3H), 3.03-3.07 (m, 1H), 3.27 (dd, J ) 11.7, 11.3
Hz, 1H), 4.04 (q, J ) 7.0 Hz, 2H), 4.15-4.18 (m, 1H), 4.21
(d, J ) 12.9 Hz, 1H), 4.27 (d, J ) 12.9 Hz, 1H), 5.22 (d, J
) 6.3 Hz, 1H), 6.7-6.9 (m, 4H), 7.07 (br s, 1H), 7.26-
7.39 (m, 5H).13C NMR (100.62 MHz, CDCl3) δ 15.19, 42.98,
64.75, 67.89, 75.82, 76.96, 82.22, 114.32, 118.96, 121.06,
123.07, 127.53, 128.69, 137.08, 147.55, 150.26, 169.37.
LRMS-APCI m/z calcd for C19H22NO4 (M + H)+: 328.
Found: m/z ) 328 [M + 1]+. Anal. Calcd for C19H21NO4:
C, 69.71; H, 6.47; N, 4.28. Found: C, 69.32; H, 6.62; N,
4.22.
Two hundred liters of toluene and 450 kg of Vitride were
charged. The solution was cooled to <10 °C, and the solution
of 10 in toluene was added to the Vitride, maintaining the
temp <10 °C with rapid agitation. The solution was stirred
for 15 min after completion of the addition. A solution of
NaOH (53 kg 15% solution) and 42 L of water were slowly
added to the Vitride solution, maintaining the temperature
<55 °C. When the quench was complete, 456 kg of 15%
NaOH and 456 L of water were added. The mixture was
stirred for 15 min, and then the aqueous phase was removed.
The toluene solution was washed with 5% Na2CO3 solution
(3 × 122 L), 1500 L of water was added, and the pH was
adjusted to 3 to 3.5 with 8% HCl (105 kg). The aqueous
phase was separated, and 1617 kg toluene was added. The
pH was adjusted to 10-12 with 10% NaOH (70 L), and the
toluene phase was removed and washed with 122 kg of 5%
Na2CO3 solution. The toluene phase was washed with 440
L of water and then distilled to an oil; 250 L of acetone was
added. The solution was clarified by recirculation through a
carbon filter for 30 min. Methanesulfonic acid (13.03 L) was
added, and the slurry was stirred at 0-5 °C for 1 h. The
solids were filtered and washed with 400 L of acetone cooled
to 0 °C. The solids were dried to yield 78.7 kg (62% overall
from 8) of reboxetine methanesulfonate (1). Mp 148.4-149.1
°C (lit. mp 145-146 °C).1 1H NMR (400.13 MHz, CDCl3)
δ 1.43 (t, J ) 7.0 Hz, 3H), 2.71 (s, 3H), 2.9-3.1 (m, 2H),
3.29-3.35 (m, 2H), 3.89-3.95 (m, 1H), 4.02-4.08 (m, 3H),
4.24-4.28 (m, 1H), 5.13 (d, J ) 4.3 Hz), 6.66-6.90 (m,
4H), 7.26-7.39 (m, 5H). 13C NMR (100.62 MHz, CDCl3)
δ 15.15, 39.50, 43.16, 44.68, 64.07, 64.56, 75.88, 82.2,
113.85, 119.0, 120.94, 123.27, 127.51, 128.62, 136.82,
147.36, 150.17. LRMS-APCI m/z calcd for C19H23NO3 (M
+ H)+: 314. Found: m/z ) 314 [M + 1]+. Anal. Calcd for
C19H23NO3-CH3SO3H: C, 58.66; H, 6.65; N, 3.42. Found:
C, 58.83; H, 6.70; N, 3.30.
Acknowledgment
We thank Ronald J. VanderRoest, Joseph P. Burke,
Klinton D. Swope, Gregory A. Kalvin, Declan Hurley, and
Guy Maineult for technical assistance in the scale-up of this
process. William C. Schinzer, Mike Gangwer, and Joel T.
Coburn provided expert analytical support. Isolation and
structure determination of the RRT 3.25 dimer was done by
D. Borghi, M.C. Colombo, and L. Franzoi.
Received for review January 8, 2007.
OP7000063
Vol. 11, No. 3, 2007 / Organic Process Research & Development
•
353