Practical asymmetric synthesis of an edivoxetine·HCl intermediate via an efficient diazotization process

Article abstract of DOI:10.1021/acs.oprd.5b00014

A convergent synthesis of (S)-(4-benzylmorpholin-2-yl)(morpholino)methanone methanesulfonate (1), a key regulatory starting material for edivoxetine·HCl, was developed at Eli Lilly & Company. This novel synthesis utilizes d-serine as the source of chirali

Full text of DOI:10.1021/acs.oprd.5b00014

Article
pubs.acs.org/OPRD
Practical Asymmetric Synthesis of an Edivoxetine·HCl Intermediate
via an Efficient Diazotization Process
Michael E. Kopach,* Perry C. Heath, Roger B. Scherer, Mark A. Pietz, Bret A. Astleford,
Mary Kay McCauley, Utpal K Singh, Sze Wing Wong, David M. Coppert, Mark S. Kerr,
and Peter G. Houghton
Small Molecule Design and Development, Eli Lilly and Company, Indianapolis, Indiana 46285, United States
Gary A. Rhodes
Technical Services & Manufacturing Science, Elanco External Manufacturing & Commercialization, Greenfield, Indiana 46140,
United States
Gregg A. Tharp
Manufacturing Technical Services, Eli Lilly and Company, Indianapolis, Indiana 46285, United States
S
* Supporting Information
ABSTRACT: A convergent synthesis of (S)-(4-benzylmorpholin-2-yl)(morpholino)methanone methanesulfonate (1), a key
regulatory starting material for edivoxetine·HCl, was developed at Eli Lilly & Company. This novel synthesis utilizes D-serine as
the source of chirality, which is preserved throughout the synthesis. Key features include the development of a scalable
diazotization process to produce (S)-epoxy acid 7, which was optimized to improve the process safety profile. The final (S)-
morpholino acid intermediate 11 was converted to the title compound using T3P with >99.9% purity in 75% yield. Life cycle
analysis of the new route revealed a 69% reduction in global warming potential (GWP) for solvent usage relative to the prior
route of manufacture.
an inefficient resolution, which is a primary contributor to the
high process mass intensity (PMI) of 219 realized for this
route.1a In addition, prior production campaigns utilized high
quantities of dichloromethane, which is a suspect carcinogen, in
the amide bond formation step. Finally a salt break and
reformation were required to produce the target API SM 1.
With these considerations in mind, we undertook the
development of an asymmetric synthesis of 1 (Scheme 3)
that addresses these shortcomings of the original route.
INTRODUCTION
■
One of the key regulatory active pharmaceutical ingredient
starting materials (API SMs) for the preparation of edivoxetine
HCl is (S)-(4-benzylmorpholin-2-yl)(morpholino)methanone
methanesulfonate (1) (Scheme 1). The original route to
produce 1 utilized 2-chloroacrylonitrile (2-CAN, 2) as the key
raw material (Scheme 2).1 One of the major drawbacks of this
synthesis was the inherent toxicity of 2-CAN, which is highly
toxic in all routes of administration, including oral, dermal, and
inhalation.2 In addition, 2-CAN is corrosive to both the skin
and mucous membranes.2 Another drawback of this synthesis is
RESULTS AND DISCUSSION
■
The key raw material for this synthesis is the chiral feedstock D-
serine, which is commercially available. It was anticipated that
D-serine could be readily converted to epoxide 7 via
diazotization, which is a well-known approach to produce
chiral epoxides and halohydrin intermediates.3 The stereo-
chemical course of these reactions proceeds with overall
retention of configuration. For the serine system, this is most
likely achieved by double inversion through a lactone
intermediate (Scheme 4). The proof that is generally cited
for the double inversion is the stereochemical outcome. An SN1
process to produce a carbocation is expected to be energetically
unfavorable (carbocation adjacent to the carbonyl). If a pure
Scheme 1. Edivoxetine·HCl retrosynthesis
Received: January 9, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/acs.oprd.5b00014
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 2. 2-Chloroacrylonitrile (2) route to produce compound 1
Scheme 3. D-Serine-based route for the preparation of compound 1
Scheme 4. Diazotization mechanistic pathway
SN1 process were in operation, some racemization due to the
carbocation instability would likely be expected.4 Invoking
Occam’s razor, we conclude that double inversion through the
lactone intermediate is the most likely explanation for the
preservation of the stereochemical integrity with overall
retention of configuration. However, on the basis of the
current data set we cannot definitively rule out a pure retention
or other related mechanisms.
While the literature conditions were satisfactory for small-
scale work, significant modifications were required to improve
the process safety for scale-up. In this system, the addition of
aqueous sodium nitrite solution was of concern because of the
potential for heat generation and pressure buildup from
byproduct gases. Colorimetric measurements revealed a 290
kJ/mol exotherm for the sodium nitrite addition, which
translated to a potential adiabatic heat rise of 70 °C for this
step. The diazotization step was operated safely on scale by
B
DOI: 10.1021/acs.oprd.5b00014
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
using a carefully controlled addition rate of the sodium nitrite
solution and temperature control (<5 °C reaction temperature)
in combination with adequate venting. In addition, feeding the
sodium nitrite solution over at least 2 h minimized inadvanta-
geous foaming. Since the diazotization step is run in water,
there was no risk of a thermal runaway, as the maximum
exotherm associated with an accidental all-at-once feed of the
sodium nitrite solution would result in temperatures below the
boiling temperature of the reaction solvent (i.e., water). In
addition, the worst-case adiabiatic temperature rise (70 °C) was
also below the decomposition temperature of the reaction
mixture (121 °C onset, 360 J/g exotherm).
The next unit operation was extraction of the bromohydrin 6
from the aqueous mixture, which was initially accomplished
with several methyl tert-butyl ether (MTBE) extractions. These
were mass-inefficient, requiring ∼30 volumes (L/kg of 5) of
MTBE, which is a solvent of environmental concern.5 Several
potential replacement solvents were screened, and 2-methylte-
trahydrofuran (2-MeTHF) was found to be a superior
extraction solvent, allowing 90% partitioning of bromohydrin
6 into the organic layer with two extractions. From an
environmental standpoint, 2-MeTHF affords many benefits, as
it is a sustainable solvent derived from furfural with a global
warming potential (GWP) of 0.2, which is the lowest of all
commercial solvents.6 After the extractions are completed, the
volume is reduced to give a 25−40 wt % composition in order
to remove residual HBr, which is an important requirement
prior to the epoxidation step. On scale we found it beneficial to
add back 2-MeTHF and perform a second distillation to the
desired final concentration, which assured removal of residual
HBr. Bromohydrin 6 was found to lose potency after 1 week of
storage, so this intermediate was not campaigned but generated
in situ and processed directly to give chiral epoxide 7.
Differential scanning calorimetry (DSC) data were collected
for the concentrated 2-MeTHF solution of bromohydrin 6 (40
wt %) and revealed two significant exothermic events. The first
transition occurred at a 150 °C onset with an enthalpy of 561 J/
g. The second transition occurred at 242 °C with an enthalpy of
496 J/g. For these reasons, the maximum jacket temperatures
during the distillations were maintained at not more than 50
°C.
KBr content was then used directly in the next step of the
synthesis.
The other starting material for the convergent synthesis, 9,
was prepared by reacting N-benzylethanolamine (8) with
sulfuric acid in refluxing toluene. Water was removed by
azeotropic distillation during the reaction to drive the
conversion of 8 to sulfate intermediate 9. The reaction mixture
was then cooled, and IPA was added, after which the solids
were filtered and dried to produce 9 in 93% yield on a 20 kg
scale. This approach represents a significant improvement
relative to other approaches to produce the desired target.9
A methanolic suspension of 9 (2 volumes) was treated with
KOMe to deprotonate the zwitterion to produce 9a. The
methanolic solution of epoxide 7 was added to 9a, and the
mixture was vacuum-distilled to ∼10 volumes. The mixture was
then heated to 40−50 °C, where reactive crystallization of
intermediate 10 occurred. An important parameter is to
maintain at least a 10 volume composition in order to achieve
a stirrable mixture in the reaction vessel. After approximately 8
h the reaction mixture was cooled, and the product was filtered,
rinsed with 2 volumes of methanol, and dried to afford
intermediate 10 in 80% yield with 90% purity and >99.5% ee
on a pilot-plant scale.
A number of low-level impurities were formed in the
reaction, but the major impurity 12, resulting from displace-
ment of the sulfate group in the product 10 with 9a, was
produced in approximately 10% yield in the reaction mixture, of
which about 3% remained in the isolated product (Scheme 5).
Scheme 5. Impurities formed in the D-serine route
The 2-MeTHF solution of bromohydrin 6 was converted to
epoxide 7 by addition of ethanolic KOH at 0 °C. On pilot-plant
scale, epoxide 7 was isolated as a crystalline solid in 82% yield.
DSC data revealed the onset of an exothermic (262 J/g)
decomposition at 130 °C. For this reason, the maximum drying
temperature for epoxide 7 was restricted to not more than 50
°C. As noted in the literature, the epoxide was isolated as a 1:1
mixture with potassium bromide.3 A known purification
method is an ethanolic recrystallization of epoxide 7, but this
procedure has low throughput due to the poor solubility of
both the epoxide and KBr in ethanol. Ultimately, we decided to
take advantage of the 6.5-fold-enhanced solubility of epoxide 7
versus KBr in methanol for the purification.7 It should be noted
that other literature references mentioned the use of methanol
for the preparation of 7, but the yield and purity from these
procedures were unclear.8 Since the optimal solvent in the next
processing step is methanol, we decided to telescope the
purification of 7 with the subsequent nucleophilic addition
reaction. We found that slurrying 7 in 10 volumes of methanol
at room temperature followed by removal of the majority of
KBr salts by filtration resulted in 90% recovery of 7 in the
filtrate. The filtered solution of epoxide 7 containing reduced
On a small scale, potassium tert-butoxide was effective at
promoting the reaction, but on a large scale a homogeneous
base was preferable for handling and operational ease. KOMe,
which is available as a 25 wt % methanolic solution, was
evaluated and found to be the best option. Sodium tert-
butoxide worked well in the reaction and produced the
analogous mixed salt 13. Unfortunately, the necessary methanol
desolvation of intermediate 13 proved to be problematic at
temperatures even as high as 60 °C. The residual methanol
C
DOI: 10.1021/acs.oprd.5b00014
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 6. Hofmann elimination pathway to produce impurity 18
level could not be reduced to <2 wt % for 13, whereas this
target could be achieved under the same drying conditions for
dipotassium salt 10. Since methanol is detrimental to the next
processing step, dipotassium salt 10 was selected as the
preferred salt form.
since the starting material was approximately 50% consumed.
The preferred concentration for the reaction was 15−20
volumes (L/kg of 10) in order to maintain good mixing, since
sulfate salts formed and precipitated from the solution as the
ring closure progressed. The optimum temperature for the ring
closure was 80 °C, with higher temperatures leading to
increased impurity formation and lower temperatures resulting
in excessive reaction times. On a pilot-plant scale, the ring
closure was performed in 16 volumes of tert-amyl alcohol with
2 equiv of 50% NaOH(aq) and the addition of 17 equiv of
water after 1−2 h of reaction time. Under these conditions, the
reaction was complete after 4.5 h at 80 °C.
A purification point in the reaction sequence prior to the
final isolation of 1 was essential in order to ensure that 1 having
a purity ≥99.5% could be routinely attained. We examined salts
of the morpholine acid as a possible crystalline intermediate
that would provide an appropriate purity upgrade. The p-TsOH
salt 11 was the only acid salt identified that was both crystalline
and possessed low solubility in water.10 The formation of the p-
TsOH salt and its isolation were incorporated into the ring-
closure step in the following manner. At the completion of the
ring closure, the suspension was cooled to 23 °C, and the pH
was adjusted to 5 with 5 N HCl. The aqueous layer was
extracted with n-heptane (1 volume) to remove a key impurity,
N,N′-dibenzylpiperazine (17) (Scheme 5). Since impurity 17
was not effectively rejected by the downstream purification,
removal at the tosylate salt stage became a critical parameter.
After the n-heptane extraction, a solution of p-TsOH
monohydrate (2 equiv) in deionized water (1 volume) was
added to the combined aqueous layers. The mixture was seeded
and cooled to 0 °C, and the product was filtered and dried to
produce tosylate 11. On a 50 gallon scale, the isolated yield of
11 was 56−57% over three batches, with achiral and chiral
purities of >99.9%.11
The ring closure of 10 under basic conditions to produce
unsalted morpholine acid was examined, and several bases were
screened. Potassium bases such as potassium tert-butoxide,
potassium tert-amylate, and potassium hydroxide in alcoholic
solvents resulted in complete conversion but also caused partial
racemization. Although up to an 8% yield of the undesired R
enantiomer could be rejected in the downstream crystallization
of 1, it was nevertheless preferential to avoid any racemization,
which increased during late stages of the cyclization and after
extended stirs. Sodium bases such as sodium tert-butoxide,
sodium tert-amylate, and sodium hydroxide in alcoholic
solvents all gave good conversion without racemization.
While alcoholic solvents were the most effective for cyclization,
the primary and secondary alcohols isobutanol and 2-butanol
produced substantial amounts of ether impurities 15 and 16,
respectively, from displacement of the sulfate group by the
corresponding alcohol (Scheme 5). Because this side reaction is
prevented with a tertiary alcohol, tert-amyl alcohol became the
preferred solvent for the cyclization reaction.
An additive crucial to the success of the ring-closure reaction
was water. On a small scale, 2 equiv of 50% NaOH(aq) was
effective in promoting the cyclization. However, on a larger
scale the cyclization reaction would stall under these conditions.
A key parameter governing the conversion and purity was
found to be the overall water content. The addition of more
water beyond the nominal amount brought in by the 50%
aqueous base improved the conversion but produced higher
levels of diol impurity 14, which suppressed the yield. We
found that the addition of at least 8 equiv of water after 1−2 h
of reaction time solved this problem. We surmised that on
larger scales mass transfer issues were prevalent and that large
particles of starting material were not efficiently broken up by
the mechanical agitation. The addition of water during the
course of the reaction assisted with solubilizing these large
particles, and adding the water late suppressed diol impurity 14
The primary focus in the development of the conversion of
11 to morpholine amide 1 was to develop an efficient process
that would prepare material having a purity of >99.5% with no
single impurity >0.1%. The peptide coupling reagent T3P was
of interest because of its known successes of minimizing
epimerization with sensitive substrates and overall mild
D
DOI: 10.1021/acs.oprd.5b00014
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 1. Life cycle analysis comparing the serine and 2-CAN processes.
processing conditions.12 Initial evaluation of T3P as a coupling
reagent produced levels of amide impurity 18 (>0.1%), likely
via a Hofmann elimination-type mechanism (Scheme 6).13 We
found that the formation of this impurity could be suppressed
(<0.05%) by maintaining the reaction mixture at less than 5 °C.
A second minor impurity was also observed, namely,
dipropylamide 19 (also formed by Hofmann degradation),
but the levels of this impurity were quite low (<0.02%)
compared with 18 (Scheme 5). Alternative conditions where
the carboxyl group was activated with pivaloyl chloride worked
well when trimethylamine was used in place of diisopropyle-
thylamine (DIEA), but this activation method produced >0.1%
diethylamide impurity 20, which was more difficult to control
(Scheme 5).14
Ultimately toluene/DIEA was the best solvent/base
combination for the amide formation. The reaction compo-
nents were combined, and the solution was cooled to 0 °C,
after which 1.2 equiv of T3P was slowly added, keeping the
reaction temperature below 5 °C.15 At the completion of the
acylation reaction, the organic layer was washed twice with 0.5
M aqueous sodium carbonate to remove residual p-TsOH. The
organic solvent layer was vacuum-distilled to give a final
concentration of 7 volumes. Isopropyl alcohol (1 volume) and
methanesulfonic acid (1 equiv) were slowly added. The
resulting suspension was stirred for 1 h at 0 °C, and the
product was filtered, rinsed with isopropyl alcohol, and dried to
produce the title compound 1 in 85% yield in the laboratory.
On a pilot-plant scale 13 kg of intermediate 11 was converted
into 9.3 kg of 1 in two batches in 73% average yield with
>99.9% achiral purity and >99.9% ee.
Life Cycle Analysis. When routes are compared, useful
metrics to consider are PMI and GWP (kg of CO2/kg of
product).16 PMI is readily calculated as the sum of the total
material inputs divided by total kilograms of product produced,
and this metric has emerged in recent years as a preferred life
cycle analysis (LCA) metric in the pharmaceutical industry.16
PMI comparison between the D-serine and 2-CAN routes
revealed a 14.5% reduction for the serine process (Figure 1).
GWP contributions from solvent usage were calculated using
Ecoinvent LCA data.17 Overall, the GWP from solvent usage is
reduced 69% by the serine route. The largest GWP contributor
in the new route is tert-amyl alcohol (43%). Future
modifications to the serine route that would reduce the GWP
further include reduction of tert-amyl alcohol use and further
incorporation of 2-MeTHF, which has a GWP of 0.2.18 It is
noteworthy that the solvents used in the serine process are all
considered pharmaceutically acceptable solvents, whereas the 2-
CAN process uses the suspected carcinogen dichloromethane
as one of the primary process solvents. While it was difficult to
access the GWP data directly for the key raw materials D-serine
and 2-CAN, assessment of the routes to manufacture indicates
a clear environmental advantage for the D-serine raw material.
For example, D-serine is likely produced directly from D,L-serine
by an enzymatic process that selectively degrades the
L
isomer.19 In contrast, 2-CAN is produced by chlorination of
acrylonitrile followed by dehydrohalogenation, both of which
are highly hazardous unit operations.20 The overall yields of 1
obtained by the two processes are similar; 27.5% for the D-
serine process versus 22% for the 2-CAN process. However, the
2-CAN process has been run for eight API manufacturing
E
DOI: 10.1021/acs.oprd.5b00014
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
campaigns with the process yields optimized. On the other
hand, the D-serine process has greater opportunities for further
yield improvement (25−35% estimated), which could translate
into significant long-term environmental and economic gains.
For these reasons, and because of the inherent hazards
associated with 2-chloroacrylonitrile, we expect that the
serine-based approach will completely replace the 2-CAN
route as the preferred route for commercial manufacture of
intermediate 1.
were agitated until complete dissolution was achieved. The
potassium hydroxide solution was then charged into the reactor
containing the chilled stage 1 solution. The internal temper-
ature of the reactor containing the stage 1 solution was
maintained between −5 and 0 °C during the addition of the
potassium hydroxide solution. The reaction mixture, which was
now a suspension, was stirred for an additional hour at −5 to 5
°C. The suspension was filtered, washed with 2-MeTHF (29
L), and dried in vacuo at 40 °C to produce 14.7 kg of
compound 7 as a white solid (82% yield, 48.1% purity).21 1H
NMR (400 MHz, D2O) δ 2.64−2.66 (m, 1H), 2.79−2.82 (m,
1H), 3.22−3.24 (m, 1H).
CONCLUSIONS
■
A second-generation asymmetric synthesis to produce 1, the
API starting material for edivoxetine·HCl, was developed and
piloted to produce the target compound with >99.9% ee in
27.5% overall yield from D-serine. This novel approach features
the development of a scalable diazotization reaction and highly
chemoselective transformations to cyclize and install the
morpholine ring. The D-serine process replaces the prior
production process, which utilized highly hazardous 2-
chloroacrylonitrile and an inefficient resolution operated on a
tonnage scale.
2-(Benzylamino)ethyl Hydrogen Sulfate (9). N-Benzy-
lethanolamine (14.0 kg, 92.6 mol) and toluene (84.6 L) were
charged into a 50 gallon reactor. The agitated mixture was
inerted with nitrogen and heated to 35−40 °C. Sulfuric acid
(96%, 9.74 kg, 95.37 mol) was added over 20 min, leading to an
exotherm of 54 °C. The mixture was heated to reflux, and water
was removed by azeotropic distillation to decrease the water
level to <100 ppm. The mixture was cooled to 20−30 °C, and
isopropyl alcohol (33.5 L) was added. The mixture was cooled
to 10−20 °C, stirred for 30 min, and filtered, and the wet cake
washed with a solution prepared from toluene (26.6 L) and
isopropyl alcohol (15.4 L). The solids were dried in vacuo at 40
°C to afford 19.71 kg of compound 9 in 93% yield with >99%
EXPERIMENTAL SECTION
■
Reaction solvents and reagents were used as purchased, and no
special precautions were used to further dry them, unless
otherwise noted. Reactions were carried out under a dry
1
purity. H NMR (400 MHz, DMSO-d6) δ 3.11 (s, 2H), 4.00−
4.03 (m, 2H), 4.15 (s, 2H), 7.40−7.48 (m, 5H), 8.90 (s, 2H);
13C NMR (75.5 MHz, DMSO-d6) δ 132.2, 130.5, 129.4, 129.1,
61.7, 50.5, 46.7; MS (ESI) m/z 232.0627 (232.0638 calcd for
C9H14NO4, MH), mp (DSC) 234 °C.
1
nitrogen atmosphere. H NMR (300 or 400 MHz) and 13C
NMR (75.5 MHz) spectra were recorded in CDCl3 unless
otherwise specified. Achiral and chiral HPLC area % purities are
reported, unless otherwise noted. The achiral and chiral HPLC
methods used to analyze compounds 1 and 11 are described in
the Supporting Information.
Potassium (S)-3-(Benzyl(2-(sulfonatooxy)ethyl)-
amino)-2-hydroxypropanoate (10). Stage 1: Purification
of Epoxide 7. Potassium (S)-(−)-2,3-epoxypropanoate (46.6%,
12.68 kg, 46.85 mol) and methanol (59 L) were charged into a
50 gallon reactor. The mixture was agitated for 2 h at 20−30 °C
and then filtered. The filtered solution of purified epoxide 7 was
used in the next stage. The solution was analyzed by
quantitative NMR spectroscopy to measure the amount of 7
(42.6 mol, 91% recovery).
Potassium (S)-(−)-2,3-Epoxypropanoate (Potassium
(S)-Oxirane-2-carboxylate, 7). Stage 1: Formation of
Bromohydrin 6. D-Serine (7.2 kg, 68.5 mol) and potassium
bromide (23.4 kg, 196.6 mol) were dissolved in water (24.8 L)
in a 50 gallon reactor. Hydrobromic acid (48%, 25.3 kg, 150
mol) was added with agitation, and the reaction mixture was
subsequently cooled to an internal temperature of 0 °C. A
solution of sodium nitrite (5.9 kg, 85.5 mol) in water (11.2 L)
was added to the above solution at a controlled rate, such that
the internal temperature remained <5 °C. The reaction vessel
was well-vented during the addition to prevent pressure
buildup. The reaction mixture was then agitated for 1 h, during
which it was repeatedly pressurized to 20 psia and vented to
purge byproduct gases. The mixture was warmed to 20−25 °C,
and the aqueous layer was extracted three times sequentially
with 2-MeTHF (43, 29, and 14.5 L). The 2-MeTHF layers
were combined and vacuum-distilled (1.5 psia, 20−25 °C
internal temperature) to a final volume of 29 L. 2-MeTHF (39
L) was added to the reaction vessel, and the vacuum distillation
was repeated to a final volume of 29 L to ensure that residual
hydrobromic acid was removed. This solution of bromohydrin
6 was used directly in the next stage.
Stage 2: Epoxide Opening. Compound 9 (9.85 kg, 42.6
mol) and methanol (19.7 L) were charged into a 50 gallon
reactor. Potassium methoxide (25 wt % in methanol, 11.95 kg,
42.59 mol) was added while maintaining the temperature below
35 °C. Then the epoxide solution from stage 1 was added to
the reactor. The volume was reduced to 49 L by vacuum
distillation (4 psia, 30−40 °C). After the completion of the
distillation, the mixture was agitated at 50 °C until at least 80%
conversion had been achieved (ca. 8 h). Once the target
conversion was achieved, the reactor contents were cooled to
20−25 °C, and the thick suspension was diluted with 49 L of
methanol to aid with the isolation. The product was filtered,
and the solids were washed with 20 L of methanol and dried in
vacuo at 60 °C to afford 13.4 kg of compound 10 in 80% yield
1
with 90% purity and 99.5% ee. H NMR (400 MHz, D2O) δ
2.63−2.87 (m, 4H), 3.64−3.67 (m, 1H), 3.75−3.78 (m, 1H),
4.02−4.05 (m, 3 H), 7.22−7.33 (m, 5H); 13C NMR (75.5
MHz, D2O) δ 137.6, 129.8, 128.5, 127.5, 70.1, 66.2, 57.8, 57.7,
51.8; MS (ESI) m/z 320.0788 (320.0798 calcd for
C12H18NO7S, MH).
(S)-4-Benzylmorpholine-2-carboxylic Acid p-Toluene-
sulfonate (11). Compound 10 (7.9 kg, 90.5% potency, 18.0
mol) and tert-amyl alcohol (95 kg) were charged into a 50
gallon reactor. Sodium hydroxide (50 wt %, 3 kg, 37.5 mol of
Stage 2: Formation of Epoxide 7. The stirred solution of
bromohydrin 6 from the first stage was diluted with ethanol
(denatured with 5% methanol) (14.4 L) and cooled to an
internal temperature of −5 °C. Potassium hydroxide (8.5 kg,
128.7 mol) was charged into a separate vessel containing
ethanol (denatured with 5% methanol) (43 L), which was
agitated and cooled at 0 °C. At the completion of the charge,
the internal temperature was raised to 20 °C, and the contents
F
DOI: 10.1021/acs.oprd.5b00014
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
base, 83.3 mol of water) was added while the mixture was
agitated. The mixture was heated to an internal temperature of
80 °C, and after 45−90 min, water (4 kg, 222 mol) was added.
After the addition of the water, the mixture was agitated at 80
°C until reaction completion (≤2% starting material remaining,
ca. 3 h). The reaction mixture was cooled to 30 °C, and water
(35.7 kg) was added. The mixture was agitated, and the pH was
adjusted to 5.0 using HCl (37%, ca. 3.6 kg). The layers were
separated, and the aqueous layer was charged into a separate 50
gallon reactor. n-Heptane (5.4 kg) and water (23.7 kg) were
added to the organic layer, and the mixture was agitated for 0.5
h at 30 °C. The layers were separated, and the aqueous phase
was combined with the first aqueous layer. The combined
aqueous layers were cooled to an internal temperature of 25 °C,
and a solution of p-toluenesulfonic acid monohydrate (96%, 8.0
kg, 40.38 mol) dissolved in water (6.2 kg) was slowly added to
form the acid salt. The suspension was cooled to 5 °C, agitated
for 1 h, and then filtered. The filtered solids were washed with
water (15.8 kg) and dried at 20 mm vacuum at a jacket set
point of 80 °C to afford 4.4 kg of compound 11 in 56% yield
ASSOCIATED CONTENT
■
S
* Supporting Information
Chiral and achiral HPLC methods for compounds 1 and 11.
This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: kopach_michael@lilly.com.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors acknowledge the management at Eli Lilly and
Company for past and present support of the edivoxetine
project. In addition, we acknowledge the contributions of
Evonik Industries for the use of their facilities and personnel to
implement pilot-plant runs for the D-serine process. Finally, we
acknowledge the contributions of Prof. A. I. Meyers, one of the
founding fathers of asymmetric synthesis.
1
with >99.9% HPLC purity and >99.9% ee. H NMR (400
MHz, DMSO-d6) δ 7.51−7.40 (m, 7H), 7.10 (d, J = 7.9 Hz,
2H), 4.37 (m, 3H), 4.02 (d, J = 11.3 Hz, 1H), 3.73 (m, 1H),
3.44 (d, J = 11.4 Hz, 1H), 3.10−3.25 (m, 3H), 2.26 (s, 3H);
13C NMR (75.5 MHz, DMSO-d6) δ 168.8, 145.6, 138.4, 131.7,
130.2, 129.5, 129.3, 128.6, 125.9, 71.6, 63.3, 59.8, 51.5, 50.5,
21.2; MS (ESI) m/z 222.1112 (222.1125 calcd for C12H16NO3,
MH); mp (DSC) 212 °C.
REFERENCES
■
(1) (a) Kopach, M. E.; Singh, U. K.; Kobierski, M. E.; Trankle, W. G.;
Murray, M. M.; Pietz, M. A.; Forst, M. B.; Stephenson, G. A.;
Mancuso, V.; Giard, T.; Vanmarsenille, M.; DeFrance, T. Org. Process
Res. Dev. 2009, 13, 209. (b) Kopach, M. E.; Roberts, D. J.; Johnson, M.
D.; McClary-Groh, J.; Adler, J. J.; Schafer, J. P.; Kobierski, M. E.;
Trankle, W. G. Green Chem. 2012, 14, 1524. (c) King, F. D.; Martin, R.
T. Tetrahedron Lett. 1991, 32, 2281. (d) Cases-Thomas, M. J.; Masters,
J. J.; Walter, M. W.; Campbell, G.; Haughton, L.; Gallagher, P. T.;
Dobson, D. R.; Mancuso, V.; Bonnier, B.; Giard, T.; DeFrance, T.;
Vanmarsenille, M.; Ledgard, A.; White, C.; Ouwerkerk-Mahdevan, S.;
Brunelle, F. J.; Dezutter, N. A.; Herbots, C. A.; Lienard, J. Y.; Findley,
J.; Hayhurst, L.; Boot, J.; Thompson, L. K.; Hemrick-Luecke, S. Bioorg.
Med. Chem. Lett. 2006, 16, 2022.
(2) (a) Gizhlaryan, M. S.; Khechumov, S. A. Toxikologische
Charakterisierung von α-Chloracrylnitril (German translation from
the Russian). Gig. Tr. Prof. Zabol. Heft 2, 1983, 53. (b) Saillenfait, A.
M.; Bonnet, P.; Guenier, J. P.; De Ceaurriz, J. Fundam. Appl. Toxicol.
1993, 20, 365. (c) Eastman Kodak Company, Laboratory of Industrial
Medicine Toxicity report-2-chloroacrylonitrile Report (1971) NTIS/
OTS 0555346. (d) BG Chemie, Toxicological Evaluation of 2-
chloroacrylonitrile. http://www.bgrci.de/fileadmin/BGRCI/
Downloads/DL_Praevention/Fachwissen/Gefahrstoffe/
TOXIKOLOGISCHE_BEWERTUNGEN/Bewertungen/
ToxBew038-E.pdf (accessed Jan 9, 2015).
(S)-(4-Benzylmorpholin-2-yl)(morpholino)methanone
Methanesulfonate (1). PTSA salt 11 (6.5 kg, 16.52 mol) and
toluene (65 L) were charged into a 50 gallon reactor that was
kept under a nitrogen atmosphere. N,N-Diisopropylethylamine
(4.3 kg, 33.26 mol) was added to the mixture, which was
agitated for 0.5 h at 25 °C. The reactor was cooled to 0 °C, and
1-propanephosphonic acid cyclic anhydride (T3P, 52.1 wt % in
toluene, 12.7 kg, 20.79 mol) was added. The reaction mixture
was stirred for 1 h at 0 °C, and then morpholine (2.2 kg, 25.25
mol) was charged into the mixture at a rate such that the
reaction temperature remained at ≤5 °C. The reaction mixture
was stirred for 1.5 h at 0 °C and then quenched by the addition
of a chilled (0 to 5 °C) aqueous sodium carbonate solution (0.5
M, 40 L). The mixture was agitated while being warmed to 25
°C, and then the layers were separated. The organic layer was
washed two more times at 25 °C with aqueous sodium
carbonate (0.5 M, 2 × 40 L). The organic layer was reduced to
a volume of 40 L by vacuum distillation (0.5 psia, 50 to 55 °C).
Isopropyl alcohol (6.3 L) was added to the organic layer, and
the solution was cooled to 0 °C. A solution of methanesulfonic
acid (1.6 kg, 16.64 mol) dissolved in isopropyl alcohol (4.35 L)
was added to the reaction mixture over 20 min to form the acid
salt. The suspension of crystals that formed was stirred for 1 h
at 0 °C and filtered, and the filtered solids were washed with
isopropyl alcohol (12.5 L) and dried under full vacuum at a
jacket set point of 80 °C to afford 4.77 kg of compound 1 in
75% yield with >99.9% HPLC purity and >99.9% ee. 1H NMR
(400 MHz, CDCl3) δ 7.54−7.49 (m, 2H), 7.47−7.41 (m, 3H),
5.02 (dd, J = 2.4 Hz, J = 10.8 Hz, 1H), 4.36 (ddd, J = 2.2 Hz, J
= 13 Hz, J = 13 Hz, 1H), 4.26−4.18 (m, 2H), 4.01 (dd, J = 3.4
Hz, J = 13 Hz, 1H), 3.71−3.38 (m, 10H), 3.17−3.09 (m, 1H),
2.91−2.83 (m, 4H); 13C NMR (75.5 MHz, CDCl3) δ 164.5,
131.3, 130.5, 129.5, 127.2, 69.6, 66.8, 66.5, 63.5, 61.7, 52.2,
50.8, 46.1, 42.4, 39.5; MS (ESI) m/z 291.1687 (291.1703 calcd
for C16H23N2O3, MH); mp (DSC) 176 °C.
(3) (a) Petit, Y.; Larcheveque, M. In Organic Syntheses; Wiley &
Sons: New York, 2004; Collect. Vol. X, p 401. (b) Koppenhoefer, B.;
Schurig, V. In Organic Syntheses, Wiley & Sons: New York, 1993;
Collect. Vol. VIII, p 119.
(4) Bunegar, M. J.; Dyer, U. C.; Green, A. P.; Gott, G. G.; Jaggs, C.
M.; Lock, C. J.; Mead, B. J. V.; Spearing, W. R.; Tiffin, P. D.;
Tremayne, N.; Woods, M. Org. Process Res. Dev. 1998, 2, 334.
(5) Prat, D.; Hayler, J.; Wells, A. Green Chem. 2014, 16, 4546.
(6) Savelski, M.; Slater, S. C.; Aycock, D.; Comanita, B.; Prescott, S.;
Shifflette, J. Life Cycle Analysis of Biobased Pennakem’s ecoMeTHF:
Green Solvents for the Pharmaceutical and Fine Chemical Industry.
Presented at the 16th Annual Green Chemistry and Engineering
Conference, June 18−20, 2012. https://pharmahub.org/resources/
585/download/PennAKem_MeTHF_LCA_-_Green_Chemistry_
and_Engineering_Conference_2012.pps (accessed Jan 9, 2015).
(7) Solubility data of epoxide 7 in methanol: 110 mg/mL at 22 °C;
146 mg/mL at 40 °C. Solubility of KBr in methanol: 17 mg/mL at 20
°C; 18.0 mg/mL at 40 °C; 22 mg/mL at 60 °C. See: Zeitlin, S. Ukr.
Khim. (Russ. Ed.) 1925, 1, 581.
G
DOI: 10.1021/acs.oprd.5b00014
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(8) (a) Stenlake, J. B.; Dhar, N. C.; Haddow, J.; McDonald, I. M.;
Maehr, R. B.; Wastila, W. B. Eur. J. Med. Chem. 1992, 27, 463.
(b) Hirschbein, B. L.; Whitesides, G. M. J. Am. Chem. Soc. 1982, 104,
4458. (c) Bihovsky, R. J. Org. Chem. 1992, 57, 1029. (d) Grosjean, F.;
Huche, M.; Larcheveque, M.; Legendre, J. J.; Petit, Y. Tetrahedron
1994, 50, 9325.
(9) Buckley, B. R.; Patel, A. P.; Wijayantha, K. G. U. J. Org. Chem.
2013, 78, 1289.
(10) Other acids that were used to prepare acid salts of 8a were
trifluoroacetic acid, hydrochloric acid, methanesulfonic acid, maleic
acid, and (+)-10-camphorsulfonic acid.
(11) Three batches of compound 11 were prepared with a total
weight of 13.3 kg.
(12) Dunetz, J. R.; Xiang, Y.; Baldwin, A.; Ringling, J. Org. Lett. 2011,
13, 5048.
(13) Cope, A. C; Trumbull, E. R. Org. React. 1960, 11, 317−493.
(14) Diethylamine contamination of the triethylamine reagent cannot
be ruled out as the source of diethylamide impurity 20.
(15) T3P as a 50% ethyl acetate solution is the most commonly
available formulation. However, T3P is now commercially available in
several solvents, and the toluene composition was selected for
purposes of process simplification.
(16) Jimenez-Gonzalez, C.; Ponder, C. S.; Broxterman, Q. B.;
Manley, J. B. Org. Process Res. Dev. 2011, 15, 912.
(17) (a) Wernet, G.; Conradt, S.; Isenring, H. P.; Jiminez- Gonzalez,
C.; Hungerbuhler, K. Int. J. Life Cycle Assess. 2010, 15, 294.
(b) Raymond, M. J.; Slater, C. S.; Savelski, M. J. Green Chem. 2010,
12, 1826.
(18) Rowan University GWP for 2-MeTHF was used to supplement
Ecoinvent data (see ref 6).
(19) May, O.; Bucholz, S.; Drauz, K.; Schwarm, M.; Turner, R. J.;
Fotheringham, G. Mutants for the Preparation of D-Amino Acids. U.S.
Patent 20060228787 A1.
(20) (a) Matsuda, H.; Hashimoto, H.; Sadashima, C. Preparation of
2-Cyanopiperazine. Japanese Patent JP2001354662, A1. (b) Li, W.; Li,
J.; Wan, Z.; Wu, J.; Massetski, W. Org. Lett. 2007, 9, 4607.
(21) The potency was determined by quantitative 1H NMR
spectroscopy (300 MHz, D2O). Sodium acetate was used as the
internal standard.
H
DOI: 10.1021/acs.oprd.5b00014
Org. Process Res. Dev. XXXX, XXX, XXX−XXX