Development of multikilogram continuous flow cyclopropanation of N -benzylmaleimide through kinetic analysis

Article abstract of DOI:10.1021/op500263m

A convenient and high-yielding method for the synthesis of trans-(dioxo)-azabicyclo-[3.1.0]-hexane carboxylate, a key intermediate of complex molecules, is presented using a flow cyclopropanation process. From a detailed kinetic study, it is demonstrated that the reaction concentration, addition time of reagents, and initial mixing temperature are the critical parameters to minimize formation of byproducts and to get high reaction yield. Using a modular tubular flow reactor, the trans/cis-(dioxo)-azabicyclo-[3.1.0]-hexane carboxylate was obtained in 92% yield and isomerized under basic conditions to produce the trans-isomer in greater than 99% diastereomeric excess. Using this approach, the reaction yield was significantly increased compared to the batch process, and the robustness and reproducibility of this flow process was demonstrated for the synthesis of this key intermediate on a multikilogram scale.

Full text of DOI:10.1021/op500263m

Article
pubs.acs.org/OPRD
Development of Multikilogram Continuous Flow Cyclopropanation
of N‑Benzylmaleimide through Kinetic Analysis
Frederic G. Buono,* Magnus C. Eriksson,* Bing-Shiou Yang, Suresh R. Kapadia, Heewon Lee,
Jason Brazzillo, Jon C. Lorenz, Laurence Nummy, Carl A. Busacca, Nathan Yee, and Chris Senanayake
Department of Chemical Development, Boehringer Ingelheim Pharmaceuticals, Inc., 900 Ridgebury Road/P.O. Box 368, Ridgefield,
Connecticut 06877-0368, United States
ABSTRACT: A convenient and high-yielding method for the synthesis of trans-(dioxo)-azabicyclo-[3.1.0]-hexane carboxylate, a
key intermediate of complex molecules, is presented using a flow cyclopropanation process. From a detailed kinetic study, it is
demonstrated that the reaction concentration, addition time of reagents, and initial mixing temperature are the critical parameters
to minimize formation of byproducts and to get high reaction yield. Using a modular tubular flow reactor, the trans/cis-(dioxo)-
azabicyclo-[3.1.0]-hexane carboxylate was obtained in 92% yield and isomerized under basic conditions to produce the trans-
isomer in greater than 99% diastereomeric excess. Using this approach, the reaction yield was significantly increased compared to
the batch process, and the robustness and reproducibility of this flow process was demonstrated for the synthesis of this key
intermediate on a multikilogram scale.
INTRODUCTION
RESULTS AND DISCUSSION
■
■
In recent years, the synthesis of cyclopropane derivatives has
been intensively developed as these compounds are versatile
building blocks of complex molecules in pharmaceutical
research and development.1 Several methods exist for
converting alkenes to cyclopropane rings such as the addition
of carbene type reagents (e.g., Simmons−Smith reaction,2
metal-catalyzed addition of diazo esters,3 free carbene
addition4), sulfur ylide additions,5 intramolecular cyclization
reactions,6 and others.7 The addition of sulfonium ylides to
activated double-bonds was reported by Payne8 describing the
addition of sulfonium ylides to activated double-bonds and
particularly the practical synthesis of ethyl dimethylsulfurany-
lidene acetate (EDSA) and its reactivity with α,β-unsaturated
ketones, esters, aldehydes to readily afford the corresponding
cyclopropyl compounds. This methodology was applied to the
synthesis of cyclopropane-fused succinimides from N-substi-
tuted maleimides and various sulfonium ylide salts.9 In one
example, the synthesis of compound trans-(dioxo)-azabicyclo-
[3.1.0]-hexanecarboxylate 2, a key intermediate of several
natural products and complex active molecules,10 was reported
using EDSA and N-benzylmaleimide in benzene to afford a
solution of trans/cis-2, followed by an isomerization at elevated
temperature in the presence of base diazobicyclo[5.4.0]undec-
7-ene (DBU) to produce trans-2 on a multigram scale.9a More
recently, a mixture of diastereoisomers trans-2 and cis-2 was
obtained in 78% yield from the reaction between the N-
benzylmaleimide and the sulfonium salt diphenylsulfonium-
(methoxycarbonyl)-methylide with a base of N-phenyl-tris-
(dimethylamino) iminophosphorane (BEMP) in dichlorome-
thane.9c In this paper, we report a kinetic study of the reaction
between N-benzylmaleimide (1) and EDSA to develop and to
implement on multikilogram scale a flow process for the
synthesis of the cis/trans-(dioxo)-azabicyclo-[3.1.0]-hexanecar-
boxylate 2 followed by an isomerization in batch mode to
obtain the pure trans-2 in good yield (Scheme 1).
Following Payne’s procedure,8 the ylide ethyl-(dimethyl-
sulfuranylidene) acetate (EDSA) was prepared from dimethyl
sulfide and ethyl bromoacetate to form the carboethoxy-methyl
sulfonium halide that was then treated with an aqueous base to
give the isolated ylide EDSA in 65% overall yield on a 2.5 kg
scale (Scheme 2).
The batch procedure developed by Agrawal et al.9a where a
mixture of trans/cis-2 was synthesized in benzene via addition
of EDSA to a solution of N-benzylmaleimide (1) was the
starting point of our development. To develop a scalable
procedure, a more process-friendly solvent (e.g., toluene) was
used for laboratory development instead of benzene or
dichloroethane. On the gram scale, the reaction between 1
and EDSA (1.3 equiv) in toluene at reflux gave a 1/1 ratio of
trans/cis-2 in 72% assay yield, followed by the isomerization
with DBU to give the trans-2 isomer (d.r. > 99:1) in 47% yield
from N-benzylmaleimide.
While initial experiments following the Agrawal procedure9a
were successful for the cyclopropanation, the yields on gram
scale were generally 10−15% lower than reported (Table 1,
entry 1). Additionally, a lower reaction yield was obtained
(58%) when this reaction was scaled-up by 20-fold using a
similar protocol (entry 2). To get a better understanding of this
process, the reaction was studied in batch mode by varying
parameters such as reaction temperature, EDSA equivalent,
reaction concentration, and order of addition of reagents. In a
range from 1.1 to 1.3 equiv of EDSA, a similar reaction yield
was obtained (entries 3, 4), and no effect of the temperature
(100 °C vs 110 °C) was noticed at a 0.4 M concentration. By
adding EDSA slowly to 1, lower reaction yield was observed
Special Issue: Continuous Processes 14
Received: August 17, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op500263m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Synthesis of trans-(dioxo)-azabicyclo-[3.1.0]-hexanecarboxylate 2
Scheme 2. Synthesis of ethyl-(dimethyl-sulfuranylidene) acetate (EDSA)
Table 1. Study of the cyclopropanation reaction in batch mode
a
b
c
entry
scale (g)
concentration (M)
temperature (°C)
mode
equivalent EDSA
assay yield 2 (%)
1
1.8
36
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.2
0.2
0.2
0.2
110
A
A
A
A
1.2
1.2
1.3
1.1
1.1
1.1
1.1
1.2
1.2
1.2
1.3
71
58
72
71
64
33
44
77
75
76
79
2
110
3
1.8
1.8
1.8
1.8
1.8
1.8
1.8
18
100
4
100
5
100
B (2 h)
A
d
6
25 → 100
25 → 100
100
7
C (30 min)
8
A
9
100
B (2 h)
10
11
100
A
A
1.8
100
a
b
A concentration of 0.4 and 0.2 M corresponds respectively to 76 and 38 g of N-benzylmaleimide/mL of toluene when all reagents were added. A:
The ylide was added in one portion to N-benzylmaleimide solution 1 at the indicated temperature. B: The ylide was slowly added to 1 (time in
c
parentheses) at the indicated temperature. C: A solution of 1 was added slowly (time in parentheses) to EDSA at the indicated temperature. Assay
d
yield determined by H NMR using an internal standard. Ratio trans-2/cis-2: 1/1. Ratio trans-2/cis-2: 2/1.9a
1
(64%, entry 5). When the initial addition of reagents was done
at 25 °C and then the solution was heated to 100 °C, yields
were negatively impacted (entries 6, 7). By diluting the reaction
solution by a factor 2, a slightly better yield was obtained (entry
11), and the impact of the addition time of EDSA on reaction
yield was partially reduced (entries 8−10).
of 1 in toluene at 95 °C was recorded over time, and the
conversion based on heat detected vs reaction time was
determined. In this system, a fast mixing is assumed between
reagents, and when a solution of EDSA (1.3 equiv) was added
to a solution of 1 (0.16 M or 33 g/L of 1), an exothermic heat
flow was detected over 15 min (ΔH = 30.9 kcal/mol) with
most of the heat released in 2 min (conversion >80%) (Chart
1). To assess the importance of the reagent addition time,
EDSA was added in three portions (3 × 0.45 equiv), every 50
min, to the N-benzylmaleimide solution, and at the end of the
From these experiments, it is suggested that a fast addition of
reagent at high temperature is critical to get higher yield. An
instantaneous addition of reagent in batch mode on kilogram
scale in a pilot or manufacturing plant setting is difficult to
envision, without impacting the temperature of the reaction. In
addition, a perfect and a fast mixing of the reagents was
assumed at the scale studied, and the effects on incomplete
micromixing on scale may become more noticeable. However,
we noticed that the impact on the yield of a long addition time
of EDSA to the reaction solution can be minimized by diluting
the reaction solution, but that would not be economically
suitable for multikilogram scale-up. These results would suggest
that the development of a scalable robust procedure in batch
mode on multikilogram scale would present significant
challenges from a practical standpoint, and this led us to
consider developing this process using flow technology.
Reaction Kinetics Analysis. To develop this process, the
kinetics of reaction was studied using reaction calorimetry and
1H NMR. Using an isothermal Omnical Insight reaction
calorimeter, the exothermic heat flow of the reaction released
after the addition of EDSA to a thermally equilibrated solution
Chart 1. Heat flow (mW) and conversion vs time (min) for a
single EDSA injection process
B
dx.doi.org/10.1021/op500263m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
reaction, a lower reaction yield was obtained (81%) compared
to the single injection process (90%) (Chart 2).
the presence of many high mass species, confirming that most
of the remaining starting material 1 detected after the first
injection gave side products after addition of EDSA. In previous
studies,11,12 it was mentioned that lower yield may be explained
by a polymerization of maleimides, and adding the N-
substituted maleimides to a preheated solution of ylide could
minimize side reactions. However, this method of addition is
not practicable on scale as the rate of decomposition of EDSA
determined by FT-IR is fast in toluene at 95 °C (40%
decomposition in 1 h at a 0.2 M concentration), and this was
consistent with Payne’s observations8 (Scheme 3).
Chart 2. Heat flow (mW) and conversion vs time (min) for
successive EDSA injection processes
To get a better understanding of the kinetic profile and the
formation of byproducts, this reaction was studied by 1H NMR.
In Chart 4, the characteristic signals for further analysis are for
1: 5.70 ppm (olefinic protons) and 4.33 ppm (benzylic
protons), and for 2: 4.45 ppm (cis) and 4.22 ppm (trans)
(benzylic protons) and 3.62 ppm (cis) and 3.68 ppm (trans)
1
(−OCH2CH3). In Chart 5, H NMR of experiments with
EDSA added to 1 in toluene-d8 at 25 and 100 °C and analyzed
after 5 min of reaction time are shown in the range of 2.5−6.0
ppm. At 100 °C, with 0.5 equiv of EDSA, 37% of trans/cis (t/c)-
2 was observed with 50% of remaining 1. When added at 25 °C,
lower amounts of 1 (39%) and 2 (24%) were quantified, and
several small peaks in the range of benzylic protons chemical
shift (4.0−5.0 ppm) were observed, which may correspond to
the presence of byproducts. This was more pronounced when
1.25 equiv of EDSA was added at 25 °C (Chart 5d and Chart 6
for details), with only 38% yield of a 2:1 mixture of trans/cis-
2.13 The reaction assay yield was unchanged, and the 1H NMR
spectra was similar after 60 min at 25 °C. By increasing the
temperature to 100 °C, a small additional amount of product
was produced (e.g., 45% yield), and the formation of multiple
new species in the range of 4.0−5.0 ppm was coupled with the
disappearance of peak at 4.45 ppm and a broad signal at 4.1−
4.2 ppm. The other signals observed in the 3.8−3.9 ppm range,
correspond to the decomposition of excess ylide over time.
This experiment suggests that a portion of 1 was consumed via
alternative reactions pathways, such as polymerization side-
reactions11 if the reagents are mixed initially at low temper-
ature.
Using the same experimental protocol, the solution
concentration was studied in the range of 33−66 g/L
(equivalent to 0.17−0.36 M), and reaction yields for each
addition mode were compared (Chart 3). For a single addition
Chart 3. Assay yield (%) vs reaction concentration [M] at 95
°C
To further understand the identity of the impurities, the
cyclopropanation was conducted at 25 °C by adding 1.3 equiv
of EDSA in one portion to a stirred solution of 1 in toluene.
After 2 days, the crude reaction mixture was concentrated and
chromatographed on silica gel using 20% EtOAc/hexanes.
Despite our efforts to purify the byproducts, the fractions
contained mixtures of several compounds. The analysis by
electrospray ionization in positive mode revealed species with
(m/z + 1) = 360.14, 446.17, 532.21, and 461.16, and plausible
structures corresponding to these masses are shown in Figure 1.
The accepted mechanism for the cyclopropanation involves
the addition of EDSA to 1, leading to a betaine as shown in
Figure 2.8 Intramolecular ring-closing with concomitant
elimination of dimethylsulfide leads to a mixture of trans/cis-
mode, the yield was impacted by the concentration (86% at
0.17 M vs 68% at 0.36 M). For a sequential addition mode of
EDSA, reaction yields are lower, and this difference was less
pronounced at higher dilution as we previously observed in
Table 1. The composition of the reaction solution was
quantified by HPLC between each injection of EDSA for
different reaction concentrations (Table 2). At higher reaction
solution concentrations (0.36 M) and after the second addition
of EDSA, only 52% product was observed with 4% of remaining
1. No productive intermediates leading to the trans/cis-
products 2 were assigned by HPLC. Analysis by LC-MS of
the crude mixture at the end of reaction (55% yield) showed
a
Table 2. Composition of reaction solution in successive addition mode (3 × 0.45 equiv EDSA)
Injection 1
Injection 2
Injection 3
concentration N-benzylmaleimide
residual 1 (%)
yield of 2 (%)
residual 1 (%)
yield of 2 (%)
residual 1 (%)
yield of 2 (%)
0.36 M
0.16 M
54.5
64.5
33.5
35.5
4
52
72
0
0
55.1
81.4
13.5
a
The yields in % were determined by HPLC assay.
C
dx.doi.org/10.1021/op500263m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 3. Products of EDSA decomposition isolated by Payne8
1
Chart 4. H-NMR reference spectra in toluene-d8
Chart 6. Cyclopropanation reaction in toluene-d8 at time
intervals
a
a
The yields of 2 are based on N-benzylmaleimide (1).
proposed reaction mechanism fits with the experimental
observations where reactions at room temperature give a low
yield of 2 (i.e., slow intramolecular ring-closing and more
intermolecular side-reactions) and reactions at 100 °C give a
high yield (i.e., fast intramolecular ring-closing and less
intermolecular side-reactions). A competitive reaction rate for
the productive pathway vs impurity formation with a different
kinetic order for each side-reaction may explain the impact of
the reaction concentration on the reaction yield.14 This analysis
also demonstrated that a perfect mixing of reagents at high
temperature is necessary to ensure a high yield and to minimize
the formation of unproductive species, and this suggested that
the development of a scalable robust procedure in batch mode
on multikilogram scale would present significant challenges
from a practical standpoint.
1
Chart 5. H NMR spectra of reactions at 25 and 100 °C in
a
toluene-d8
Development of Flow Process. Using Uniqsis microflow
screening system,15 two streams (Stream A: N-benzylmaleimide
in toluene and Stream B: EDSA in toluene−total concen-
tration: 0.20 M) were mixed in microchip (1.6 mL) followed by
an Hastelloy coil tube (5 mL) both heated at the desired
temperature and controlled pressure (10 bar) (Scheme 4). At
95 °C, with 1.2 equiv of EDSA and 2 min residence time, the
product trans/cis-2 was obtained in 75% yield (entry 1, Table
3). At 105 °C, a higher yield was obtained (82%) (entry 2).
The impact of excess EDSA amount was noted, and a higher
yield was obtained with 1.3 equiv of EDSA (88% yield, entry 4).
By increasing the temperature above the boiling point of
toluene (120 °C), the yield was increased to 93% (entry 6), and
at least 1.2 equiv of EDSA was needed to get an optimal yield.
When the residence time was reduced to 45 s, lower yield was
observed due to an incomplete conversion of intermediate to
product (entry 8).
a
The yields of t/c (trans/cis)-2 are based on N-benzylmaleimide (1).
2. It could be envisioned that the proposed structures A−C
above could form if the betaine reacted intermolecularly with
one or multiple EDSA as a nucleophile, followed by ring-
closing. Alternatively, the betaine could react with an additional
molecule of 1 to form D as well as polymeric species. The
Flow Scale-Up. To get higher throughput and to ensure
our delivery on kilogram scale, a system similar to the Uniqsis
system was made in-house. A modular reactor was built with a
jacketed T-mixer (5 mL) and a static mixer (15 mL) connected
in series to a jacketed tubular (tube-in-tube) reactor with a total
D
dx.doi.org/10.1021/op500263m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 1. Proposed structures of impurities detected by electrospray ionization analysis.
Table 3. Parameter flow optimization with the Uniqsis flow
reactor
b
d
entry
a
temperature
equiv of
EDSA
conversion
(%)
assay yield
(%)
no.
(°C)
1
2
3
4
5
6
7
8
95
105
105
105
120
120
120
120
1.2
100
100
100
100
100
100
100
100
75
82
78
88
80
92
93
1.2
1.05
1.3
1.05
1.2
1.3
c
1.3
74
a
Reaction concentration: 0.20 M on a 5 g scale of N-benzylmaleimide.
Conversion calculated on N-benzylmaleimide remaining and for a
b
c
d
residence time of 2 min. For a residence time of 45 s. trans/cis-2
obtained: 1:1.
Figure 2. Proposed reaction mechanism for the cyclopropanation of 1
by EDSA.
of trans/cis-2 was obtained; while with 1.3 equiv of EDSA, the
yield was 94% (Table 4). During this stage of development, a
total amount of 2.3 kg of 1/1 mixture of trans/cis-2 was
generated with 81% average yield. To check the process
robustness on larger scale and longer process time, we ran this
flow process continuously for 7 h at an average flow rate of 450
g product/h (equivalent to 2 min residence time and for a
concentration of 50 g of 1/L). During this run, a total of 3.3 kg
of trans/cis-2 (ratio: 1/1) was generated in 92% yield. In
conclusion, the kinetic analysis showed that a combination of
parameters such as temperature and reaction concentration
impacted the reaction yield. The benefits of running the
reaction at high temperature and under pressure using flow
technology (e.g., plug-flow reactor) were demonstrated, and a
greener and more economical process was developed.
volume of 310 mL. A back pressure regulator (from 20 to 100
psi) was installed at the end of the process stream to ensure
consistent residence time for the reaction mixture (Scheme 5).
In order to quickly heat the reaction solution during the mixing,
the feed line for the N-benzylmaleimide solution was preheated
with a heat exchanger.
The flow process parameters were further optimized based
on the preliminary results obtained from the Uniqsis system.
We processed five batches from 100 g up to 780 g of 1 with a
fixed residence time of 3 min at 120 °C and varying the amount
of EDSA in each batch. In order to get higher throughput, the
reaction was done at higher reaction concentration (0.27 M or
50 g/L) compared to what was previously screened using the
Uniqsis system. With 1.15 equiv of EDSA, only 75% assay yield
Scheme 4. Flow configuration of the Uniqsis system for the cyclopropanation of N-benzylmaleimide
E
dx.doi.org/10.1021/op500263m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 5. Modular setup for flow scale-up
Table 4. Effect of ratio of EDSA/1 on reaction yield (%)
CONCLUSION
■
b
A convenient and high yielding process for the synthesis of
trans-(dioxo)azabicyclo-[3.1.0]-hexane carboxylate 2 was de-
scribed using a flow cyclopropanation process followed by an
isomerization with DBU. Using a systematic kinetic study, the
impacts of reaction concentration, addition time of reagents,
and initial temperature were assessed to improve the reaction
yield. Using a modular tubular flow reactor, a total of 5.6 kg of
trans/cis-(dioxo)azabicyclo-[3.1.0]-hexane carboxylate 2 was
obtained with an average of 85% yield (92% yield for our
optimized run on 3.3 kg scale) and then isomerized to pure
trans-2. With this approach, the reaction cost was considerably
reduced on scale due to higher reaction yield compared to
batch process, and the robustness of this flow process was
demonstrated for the synthesis of trans-(dioxo)azabicyclo-
[3.1.0]-hexane carboxylate 2 on a multikilogram scale.
experiment
ratio
EDSA/1
quantified amount of trans/cis-2 assay yield
a
no.
(gram)
(%)
1
1.15
1.2
765
780
75
80
84
92
94
81
2
3
1.24
1.28
1.3
200
4
445
5
110
average
1.2
2300
a
For a reaction concentration of 0.27 M (50 g of N-benzylmaleimide/
L) and a residence time of 3 min. Quantified by HPLC ratio cis/trans-
2: 1:1.
b
Isomerization to trans-(Dioxo)azabicyclo-[3.1.0]-hex-
ane Carboxylate-2. The isomerization of the solution of
trans/cis-2 was conducted as a separate step in batch using a
catalytic amount of DBU.9a It was not feasible to consider a
telescoped cyclopropanation/isomerization under flow con-
ditions, due to a long reaction time. Experimentally, after an
aqueous workup of the crude toluene solution of trans/cis-2, the
solvent was distilled and replaced with xylene for the
isomerization reaction. After addition of DBU, the mixture
was heated for 16 h at 140 °C to afford a 96:4 ratio of trans/cis-
2. After workup and crystallization from ethanol, 4.3 kg of pure
trans-2 (d.r. > 99:1) was obtained in 72% yield.
EXPERIMENTAL SECTION
■
General Information. All of the compounds described
herein were previously published and 1H NMR spectra were in
agreement with literature description.8,9 Reaction calorimetry
was carried out in an Insight Omnical Super CRC reaction
calorimeter. The system operates as a differential scanning
calorimeter by comparing the heat released or consumed in a
sample vessel compared to a reference compartment. The
sample vessel is a 16 mL septum-cap vial equipped with a
stirring bar. High-resolution mass spectral data were acquired
using an Agilent LC/MSD TOF (time-of-flight) mass
spectrometer. A sample solution was delivered to the mass
spectrometer by flow injection with an HPLC system and
analyzed in an electrospray positive ionization mode. LC−MS
analysis was performed on an Agilent HP 1100 series
chromatography (Zorbax Eclipse XDB-C18) attached to a
Waters ZQ2000 mass spectrometer with ESCi ionization
source in ESI mode. Elution was carried out at a flow rate of
1.5 mL min−1 using a reverse-phase gradient of acetonitrile and
water containing 0.2% HClO4 and 0.1% H3PO4. Retention time
(Rt) is given in minutes to the nearest 0.1 min, and the m/z
value is reported to the nearest mass unit (m.u.). Laboratory
scale flow experiments were performed using a FlowSyn
(Uniqsis) equipped with two pumps, micromixer (1.6 mL), and
Scheme 7. Synthesis of trans-(dioxo)-azabicyclo-[3.1.0]-
hexane carboxylate-2
F
dx.doi.org/10.1021/op500263m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
a Hastelloy tube coil (5 mL). For the large scale unit, the two
feed streams proceeded through a jacketed T-mixer (5 mL) and
a static mixer (15 mL) connected in series to a jacketed tubular
(tube-in-tube) reactor with a total volume of 310 mL.
mixture of 2. A sample was purified for characterization
purposes on a Combiflash 40 g silica gel cartridge using 0−40%
EtOAc/hexanes to give pure trans- and cis-2. (1S,5R,6R)-3-
Benzyl-2,4-dioxo-3-aza-bicyclo-[3.1.0]-hexane-6-carboxylic acid
1
Representative Procedure for EDSA Scale-up Syn-
thesis. Ethoxycarbonylmethyl-dimethyl-sulfonium Bro-
mide. A solution of ethyl bromoacetate (9.0 kg, 52.8 mol, 1.0
equiv) in acetone (19 L) was stirred under nitrogen, and
dimethyl sulfide (4.06 kg, 64.7 mol, 1.23 equiv) was added in
one portion. The resulting clear solution was stirred at 20−25
°C, and a slurry of ethoxycarbonylmethyl-dimethyl-sulfonium
bromide (25 g, 0.11 mol) in acetone (0.5 L) was added as
seeds. A thin suspension began to form after 15 min, and the
mixture was stirred for 48 h at 20−25 °C to afford a thick white
slurry. The slurry was filtered on a 10″ Sparkler filter fitted with
a filter cloth, and the cake was washed with three consecutive
portions of acetone (8 L). The solids were dried in a vacuum
oven at 22 °C for 12 h to afford 10.4 kg (86%) sulfonium salt as
a white solid.
ethyl ester (trans-2) H NMR (500 MHz, CDCl3); 7.26−7.33
(m, 5H), 4.51 (s, 2H), 4.18 (q, J = 7.2 Hz, 2H), 2.86 (d, J = 2.8
Hz, 2H), 2.27 (t, J = 2.8 Hz, 1H), 1.27 (t, J = 7.1 Hz, 3H). 13
C
NMR (125 MHz, CDCl3); 172.2, 167.4, 135.7, 128.9, 128.7,
128.3, 62.3, 42.2, 32.3, 26.5, 14.2. (1S,5R,6S)-3-Benzyl-2,4-
dioxo-3-aza-bicyclo-[3.1.0]-hexane-6-carboxylic acid ethyl ester
(cis-2) 1H NMR (500 MHz, CDCl3); 7.24−7.39 (m, 5H), 4.49
(s, 2H), 4.04 (q, J = 7.2 Hz, 2H), 2.79 (d, J = 8.2 Hz, 2H), 2.53
(t, J = 8.3 Hz, 1H), 1.20 (t, J = 7.1 Hz, 3H). 13C NMR (125
MHz, CDCl3); 171.4, 167.2, 136.0, 129.1, 128.7, 128.0, 62.2,
42.9, 32.0, 26.9, 14.1.
Representative Procedure for Cyclopropanation Flow
Screening. A solution of N-benzyl-maleimide (5 g, 27 mmol,
Stream A) in toluene (37 mL) (concentration: 0.74 M), and a
solution of EDSA (6 g, 33 mmol, Stream B) in toluene (75 mL)
(concentration: 0.44 M) were pumped using a Uniqsis Flowsyn
with a total flow rate of 2−6 mL/min (for a residence time
from 1 to 3 min) through a successive micromixer (1.6 mL
volume) and 5 mL stainless steel coil tubing heated at desired
temperature (105−120 °C) and equipped at the end with a
back pressure regulator (e.g., 10 bar). The exiting flow stream
was recovered into a vial cooled down to 25 °C and analyzed by
HPLC for assay yield.
Typical Procedure for Flow-Scale-up Cyclopropana-
tion. N-Benzylmaleimide (2.45 kg, 13.5 mol) was dissolved in
14.6 L of toluene in a reactor and transferred to a 20 L bottle
(Stream A). A solution of 7.78 kg of ylide (32.4 wt %, 16.9 mol,
1.25 equiv) was mixed with toluene (22.5 L) in a stirred reactor
and transferred to two 20 L bottles (Stream B). Two Encynova
pumps were connected to modular reactors of a successive
jacketed Swagelok 1/4″ T-mixer, Koflo 316 stainless steel static
mixer (5 mL) and an Exergy jacketed tubular tube-in-tube (310
mL) connected to a chiller. The preheating of the maleimide
stream was made by passing through an Exergy shell and tube
heat exchanger with internal volume of ∼10 mL. The system
was operated at 5 bar with a total flow rate of 165 mL/min
(Encynova pumps A flow rate: 60 mL/min and Encynova
pump B flow rate: 105 mL/min) (for 2 min residence time)
through the reactor system at the desired temperature (internal
temperature: 120 °C, chiller temperature: 140 °C). The
reaction mixture stream was collected in 20 L bottles and
analyzed by HPLC for assay yield determination. Based on
these calculations, 3.3 kg (12.04 mol) of trans/cis-2 were
obtained with an assay yield of 92%.
Typical Procedure for Reaction Calorimetry. A 16 mL
vial containing a solution of N-benzylmaleimide (281 mg, 1.5
mmol) in toluene (4 mL) was placed in the Omnical Insight
RT-10 calorimeter, at 95 °C, and stirred until thermal
equilibrium was reached (ca. 45 min). Another vial containing
4 mL of toluene was placed in the reference port. The reaction
was initiated by injecting simultaneously a solution of EDSA
(237 mg, 1.6 mmol) in toluene (1 mL) to the solution of N-
benzylmaleimide and toluene (1 mL) to the sample reference,
respectively. An exothermic heat flow signal was observed, and
when the heat flow signal came back to the baseline, the vial
was removed, and the reaction solution was analyzed by HPLC
to determine the assay yield.
Ethyl-(dimethyl-sulfuranylidene) Acetate (EDSA). A
solution of ethoxycarbonylmethyl-dimethyl-sulfonium bromide
(5.16 kg, 22.1 mol) in dichloromethane (31 L) was cooled to
0−5 °C, and a mixed base solution of a saturated solution of
potassium carbonate in water (12 kg, ∼4 equiv, 10.2 L) and a
solution of 12.5 M sodium hydroxide (0.61 L, 1 equiv) was
added. The batch temperature increased temporarily to ∼8 °C.
The mixture was stirred vigorously at 0−5 °C for 10 min and
then warmed to 20−25 °C over 30 min and stirred at that
temperature for 10 min. The stirring was stopped, and after 15
min, the lower aqueous layer was removed (Note: the layers are
inverted due to the high concentration of salts in the aqueous
phase so the dichloromethane is the upper layer). The organic
layer was filtered through a pad of Celite (1.1 kg), the pad
rinsed with dichloromethane (5 L), and the rinse combined
with the filtrate. The reactor was rinsed with water (6 L)
followed by dichloromethane (5 L) and then charged with
anhydrous potassium carbonate (0.8 kg). The combined
organic filtrate was charged to the reactor and the mixture
stirred vigorously at 20−25 °C for 1.5 h and then filtered to
remove potassium carbonate. The reactor was cleaned and the
organic layer concentrated by vacuum distillation at 20−25 °C
jacket temperature to remove about 15 L of dichloromethane.
Toluene (6 L) was added and the distillation continued at 20−
25 °C jacket temperature and <60 Torr until no more
dichloromethane was distilled. The solution (7.78 kg) was
transferred into a container and stored at 2−8 °C. Analysis by
1H NMR assay indicated 2.51 kg sulfur ylide (EDSA) (75%
yield) and 11.5 wt % dichloromethane in addition to toluene.
Representative Procedure for Batch Process Cyclo-
propanation. A solution of N-benzylmaleimide (17.9 g; 95.6
mmol) in toluene (230 mL) was stirred and heated to 100 °C
in a 3-neck flask under argon equipped with argon inlet, reflux
condenser, and addition funnel. A solution of EDSA (17.0 g;
114.7 mmol; 1.2 equiv) in toluene (230 mL) was added to the
flask in one portion via the addition funnel (took 30−60 s) to
give a light red solution. The internal temperature decreased
from 100 to ca. 80 °C upon fast addition of the room-
temperature solution. The resulting solution slowly heated back
up to ∼100 °C over about 20 min and was stirred at 100 °C for
about 3 h. The color changed to light yellow during the heating.
The reaction mixture was concentrated under vacuum to a light
1
red solid (27.5 g). Assay by H NMR using diethylfumarate as
Representative Procedure for NMR Experiments. (a)
internal standard indicated 75% yield of a 50:50 trans/cis
Sulfur ylide addition at 95 °C: EDSA (247 mg; 1.45 mmol; 1.3
G
dx.doi.org/10.1021/op500263m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
equiv) was added in one portion to a solution of N-
benzylmaleimide (210 mg; 1.12 mmol) in toluene-d8 (4.0
mL) at 95 °C. The solution was stirred for 1 h and cooled to
room temperature, and p-xylene (140 μL) was added to the
cake washed with cold (2−8 °C) ethyl alcohol (7 L) followed
by heptane (7 L) and then dried in a vacuum oven at 35 °C for
16 h to afford pure trans-2 as a tan solid in 72% (4.26 kg) yield.
1
AUTHOR INFORMATION
Corresponding Authors
*E-mail: frederic.buono@boehringer-ingelheim.com.
*E-mail: magnus.eriksson@boehringer-ingelheim.com.
solution as internal standard to assay the amount of 2 by H
■
NMR (82.9%). (b) Sulfur ylide addition at 25 °C: EDSA (247
mg; 1.45 mmol; 1.3 equiv) was added in one portion to a
solution of N-benzylmaleimide (210 mg; 1.12 mmol) in
toluene-d8 (4.0 mL). The solution was stirred at 25 °C for 2
h, p-xylene (140 μL) was added to the solution as internal
Notes
The authors declare no competing financial interest.
1
standard to assay the amount of 2 by H NMR (36.5% yield).
Typical Procedure for DBU-Catalyzed Isomerization.
1,8-Diazabicyclo[5.4.0]undec-7-ene (6.24 mL; 40.9 mmol; 0.25
equiv) was added in one portion to a solution of 3-benzyl-2,4-
dioxo-3-aza-bicyclo-[3.1.0]-hexane-6-carboxylic acid ethyl ester
(44.7 g; 163.7 mmol; 1 equiv) in xylene (200 mL) at room
temperature. The resulting red mixture was stirred and heated
in a 145 °C oil bath for 18 h and then cooled to room
temperature and analyzed by HPLC. The ratio of trans/cis-2
was 95:5. The reaction mixture was diluted with toluene (300
mL), and the organic layer was washed with 1 M aq. HCl (250
mL) followed by water (2 × 250 mL). The brown organic layer
was filtered through a pad of Celite and concentrated at 40 °C/
15 Torr to a dark brown oil, which was azeotroped with
heptane and subsequently solidified to afford 43.5 g of brown
solid. The resulting solid was heated to reflux in ethanol (350
mL) to obtain a dark solution that was stirred and cooled to
room temperature. Water (100 mL) was added slowly over 30
min. The suspension was stirred for 2 h and filtered on a
medium Buchner frit. The cake was rinsed with a mixture of
ethanol (50 mL) and water (50 mL) and dried on the frit for 2
h and then in a vacuum oven at 40 °C overnight to afford 29.5
g (66%) pure [(1S,5R,6R)-3-benzyl-2,4-dioxo-3-aza-bicyclo-
[3.1.0]-hexane-6-carboxylic acid ethyl ester (trans-2)] as a tan
solid.
Scale-up of DBU-Catalyzed Isomerization to
[(1S,5R,6R)-3-Benzyl-2,4-dioxo-3-aza-bicyclo-[3.1.0]-hex-
ane-6-carboxylic Acid Ethyl Ester (trans-2). The trans/cis-2
solution in toluene (6 kg, 22 mol of 2 into 91.6 kg toluene)
from the flow process was charged to a jacketed kilo-lab reactor.
Water (30 L) was added, the mixture stirred for 20 min, and
then organic and aqueous phases were separated. The organic
layer was concentrated by vacuum distillation at ∼50 °C/<60
Torr until about 10 L remained in the reactor. p-Xylene (25 L)
was added and the vacuum distillation continued until about 10
L remained. The batch was cooled to 20−25 °C, and additional
p-xylene (11 L) was added, followed by 1,8-diazabicyclo[5.4.0]-
undec-7-ene (0.88 kg; 0.25 equiv). The stirred mixture was
heated at reflux (∼142 °C) under nitrogen for 16 h. HPLC
analysis of a sample indicated a 95:5 ratio of trans/cis-2. The
batch was cooled to 50−60 °C, diluted with toluene (37 L),
and further cooled to 20−25 °C. Aqueous HCl (2M, 19 L) was
added, the solution was stirred for 10 min and the organic and
aqueous phases were separated. The organic layer was washed
with water (2 × 19 L), filtered through a pad of Celite (1 kg),
and the reactor and filter rinsed with toluene (5 L). The
combined organic layer was transferred back to the cleaned
reactor and concentrated by vacuum distillation at ∼50 °C/<30
Torr until about 10 L remained in the reactor. Ethyl alcohol
(25 L) was added over 30 min, keeping the batch temperature
at 50−60 °C. The batch was stirred and cooled to about 45 °C,
5 g of trans-2 was added as seeds, and the batch cooled to about
20 °C and stirred overnight. The slurry was filtered and the
REFERENCES
■
(1) (a) Chen, D. Y.-K.; Power, R. H.; Richard, J.-A. Chem. Soc. Rev.
2012, 41, 4631−4632. (b) Cao, W.; Zhang, H.; Chen, J.; Deng, H.;
Shao, M.; Lei, L.; Qia, J.; Zhu, Y. Tetrahedron 2008, 64, 6670−6674.
(c) Tang, P.; Qing, Y. Synthesis 2012, 44 (19), 2969−2984.
(2) Simmons, H. E.; Smith, R. D. J. Am. Chem. Soc. 1958, 5323.
(3) Brookhart, M.; Studabaker, W. B. Chem. Rev. 1987, 87 (2), 411−
432.
(4) Fedorynski, M. Chem. Rev. 2003, 103 (4), 1099−1132.
́
(5) Li, A.-H.; Dai, L.-X.; Aggarwal, V. K. Chem. Rev. 1997, 97 (6),
2341−2372.
(6) (a) Carson, C. A.; Kerr, M. A. Chem. Soc. Rev. 2009, 38, 3051.
(b) Rubin, M.; Rubina, M.; Gevorgyan, V. Chem. Rev. 2007, 107, 3117.
(7) Wessjohann, L. A.; Brandt, W.; Thiemann, T. Chem. Rev. 2003,
103 (4), 1625−1648.
(8) Payne, G. B. J. Org. Chem. 1967, 32 (11), 3351−3355.
(9) (a) Agrawal, K. C.; Srivastava, S.; Singh, V. S.; Dikshit, D. K. Ind.
J. Chem. 2004, 43B, 873−875. (b) Oikawa, M.; Sasaki, S.; Sakai, M.;
Ishikawa, Y.; Sakai, R. Eur. J. Org. Chem. 2012, 5789−5802. (c) Biagetti,
M.; Leslie, C. P.; Mazzali, A.; Seri, C.; Pizzi, D. A.; Bentley, J.; Genski,
T.; Di Fabio, R.; Zonzini, L.; Caberlotto, L. Bioorg. Med. Chem. Lett.
2010, No. 20, 4741−4744. (d) For a review on the topic: Sun, X. L.;
Tang, Y. Acc. Chem. Res. 2008, 41, 937−948.
(10) (a) Nair, A. G.; Duca, J. S.; Dwyer, M. P.; Kozlowski, J. A.;
Rosenblum, S. B. PCT Int. Appl. (2012), WO 2012125661 A1
20120920. (b) Berliner, M. A.; Dubant, S. P. A.; Makowski, T.; Ng, K.;
Sitter, B.; Wager, C.; Zhang, Y. Org. Process Res. Dev. 2011, 15 (5),
1052−1062. (c) Zhang, L.; Rogers, B. N.; Duplantier, A. J.; McHardy,
S. F.; Efremov, I.; Berke, H.; Qian, W.; Zhang, A. Q.; Maklad, N.;
Candler, J.; Doran, A. C.; Lazzaro, J. T., Jr.; Ganong, A. H. Bioorg. Med.
Chem. Lett. 2008, No. 18, 5493−5496.
(11) Polymerization of N-substituted maleimides is known, see
(a) Onimura, K.; Tsutsumi, H.; Oishi, T. Macromolecules 1998, 31,
5171−76. (b) Hodge, P.; Ezzatollah, K. E.; Naim, A. A. Polym.
Commun. 1986, 27, 322−23.
(12) In reference 23 of Eur. J. Org. Chem. 2012, 5789−5802, the
authors mentioned a polymerization of maleimides as a side reaction,
but no clear experimental evidence was provided.
(13) The preference for thermodynamically more stable trans-2 when
the ylide is added at 25 °C was noted previously by Agrawal et al.9a
This observation is supported by MM2 energy minimization
calculations that indicate that trans-2 is lower in energy than cis-2 by
3.6 kcal/mol.
(14) The authors acknowledge referees for helpful discussions and
comments. Further kinetic study is under progress and will be
presented in a further communication.
(15) Bagley, M. C.; Fusillo, V.; Jenkins, R. L.; Lubinu, M. C.; Mason,
C. Org. Biomol. Chem. 2010, 8, 2245. For further details on the Uniqsis
system, see: http://www.uniqsis.com.
H
dx.doi.org/10.1021/op500263m | Org. Process Res. Dev. XXXX, XXX, XXX−XXX