Mechanism-guided design of flow systems for multicomponent reactions: Conversion of CO2 and olefins to cyclic carbonates

Article abstract of DOI:10.1039/c3sc53422g

A mechanism-guided design of a multi-step flow system enabled an efficient general process for the synthesis of cyclic carbonates from alkenes and CO ₂. The flow system proved to be an ideal platform for multicomponent reactions because it was straightforward to introduce reagents at specific stages without their interacting with each other or with reaction intermediates prone to destruction by them. This system exhibited superior reactivity, increased yield, and broader substrate scope relative to conventional batch conditions and suppressed the formation of undesired byproducts, such as, epoxides and 1,2-dibromoalkanes. The Royal Society of Chemistry 2014.

Full text of DOI:10.1039/c3sc53422g

Chemical
Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Mechanism-guided design of flow systems for
multicomponent reactions: conversion of CO2
and olefins to cyclic carbonates†
Cite this: Chem. Sci., 2014, 5, 1227
a
a
b
b
a
*
*
Jie Wu, Jennifer A. Kozak, Fritz Simeon, T. Alan Hatton and Timothy F. Jamison
A mechanism-guided design of a multi-step flow system enabled an efficient general process for the
synthesis of cyclic carbonates from alkenes and CO2. The flow system proved to be an ideal platform for
multicomponent reactions because it was straightforward to introduce reagents at specific stages
without their interacting with each other or with reaction intermediates prone to destruction by them.
This system exhibited superior reactivity, increased yield, and broader substrate scope relative to
conventional batch conditions and suppressed the formation of undesired byproducts, such as, epoxides
and 1,2-dibromoalkanes.
Received 13th December 2013
Accepted 16th January 2014
DOI: 10.1039/c3sc53422g
www.rsc.org/chemicalscience
afforded the desired product in approximately 5-fold higher
yield and without byproducts seen in the latter.
Introduction
Transformation of CO2 into useful organic chemicals is
attracting ever-increasing attention.4 The synthesis of ve-
membered cyclic carbonates, in particular, is a promising
pathway for CO2 chemical xation. These cyclic carbonates are
widely used as raw materials for the production of poly-
carbonates and polyurethanes, have important applications as
fuel additives, polar aprotic solvents and electrolytes in lithium-
ion batteries, and are important intermediates for pharmaceu-
tical/ne chemicals in biomedical applications.5 The introduc-
tion of CO2 as a feedstock avoids utilization of phosgene and
phosgene-generating reagents that are typically employed in
such transformations. In this regard, we have recently devel-
oped a bromine-catalyzed conversion of CO2 and epoxides to
carbonates in a continuous ow apparatus.6 Though efficient, a
more straightforward and economical approach would be the
direct production of cyclic carbonates starting from corre-
sponding olens. However, in contrast to extensive studies on
reactions with epoxides, few reports have documented the
direct synthesis of cyclic carbonates from olens, and most of
them entail a one-pot, sequential epoxidation/cycloaddition
pathway.7,8 Among these reported methods, we were attracted to
the recent elegant work from the Li group with the use of NBS
reagents.8a
Continuous ow methods have emerged as enabling technolo-
gies for development, optimization, and mechanistic investi-
gation of chemical transformations.1 Particularly well suited to
this format are gas–liquid reactions where an exceptionally high
interfacial contact between the two phases, enhanced safety,
and a small footprint compared to reactor output provide
signicant benets over corresponding batch conditions.2
Reagent incompatibility is a common problem in multicom-
ponent reactions (whether in batch or ow). The Yoshida group
has described some excellent examples that address this
problem in ow microreactors by “ash chemistry”.3 Herein we
demonstrate yet another means to solve the reagent incom-
patibility problem, utilization of a sequential ow reactor
system. The ease of sequential introduction of reagents in ow
(particularly gases, relative to batch) and precise control over
residence time and temperature allowed us to design optimal
multi-step continuous ow reactor systems based on a combi-
nation of mechanistic investigations, hypothesis-driven empir-
icism, and reaction parameter optimization. Moreover, side-by-
side comparisons of the optimized ow conditions and batch
experiments designed to emulate them revealed that the former
Herein we report a mechanism-guided design of a sequential
ow system for the xation of CO2 with olens as the starting
material, with superior results relative to those with obtained
under batch conditions. Through rational design and optimi-
zation of the ow system, we were able to avoid the low
conversion and reagent incompatibility problems observed in
our original ow design (Scheme 1).
aMassachusetts Institute of Technology, Department of Chemistry, 77 Massachusetts
Ave., Cambridge, Massachusetts 02139, USA. E-mail: t@mit.edu; Tel:
+1-617-253-2135
bMassachusetts Institute of Technology, Department of Chemical Engineering, 77
Massachusetts Ave., Cambridge, Massachusetts 02139, USA. E-mail: tahatton@mit.
edu; Tel: +1-617-253-4588
† Electronic supplementary information (ESI) available: Details of mechanistic
study, reaction optimization, experimental procedures and spectroscopic data
of all new compounds. See DOI: 10.1039/c3sc53422g
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1227–1231 | 1227

View Article Online
Chemical Science
Edge Article
Table 1 Control experiments to explore the reaction mechanism
Yield
Other
Entry Conditions
Conversiona of 1ba productsa
1
2
3
4
NBS + H2O
100%
26%
84%
50%
0%
0%
55%
0%
87% 3b
15% 4b
8% 4b
NBS + H2O + DBU
NBS + H2O + DBU + CO2
NBS + DMF + DBU + CO2
Scheme 1 Conversion of CO2 and olefins into cyclic carbonates in
flow.
45% 2b
a
Conversion and yield based on crude 1H NMR analysis using
trichloroethylene as an external standard.
Results and discussion
Li proposed a bromohydroxylation pathway (Scheme 2, path a)8a
for the conversion of olens to cyclic carbonates using stoi-
chiometric NBS and DBU,9 based on the transformation that
bromohydrins can be generated from olens and NBS in water.
However, when we treated styrene with NBS and DBU in water
under pressurized CO2, dibromide 2a, instead of bromohydrin
3a, was obtained as the byproduct. By reconsidering the
possible mechanism for this transformation, it was argued that
different intermediates, such as DBU activated CO2 (complex C,
path b),10 water induced bicarbonate anion B (path c), and the
epoxide intermediate 4a (path d) could also play a role in the
formation of the cyclic carbonate product 1a.
Control experiments were rst conducted in sealed tubes to
explore the reaction mechanism using vinylnaphthalene as the
model substrate. As shown in Table 1, full conversion was
achieved with NBS in water to deliver bromohydrin 3b in 87%
yield (entry 1). However, in the presence of DBU, the reaction
rate became exceedingly low (only 26% conversion aer 3 hours,
entry 2), indicating an interaction between DBU and NBS
(complex D, Scheme 2),11 which hindered the formation of the
reactive bromonium ion between NBS and olens. With the
assistance of CO2, the conversion increased to 84% in a 3 h
reaction period (entry 3), probably due to the formation of
complex C (ref. 10) that released NBS to react again with the
olen. In the absence of H2O, only the dibromide by-product 2b
was observed (entry 4). Further control experiments indicated
that the reaction seemed likely to proceed through DBU acti-
vation of CO2 via pathway b.12 However, it is highly possible that
all the pathways contributed simultaneously to the formation of
the nal cyclic carbonates during the reaction process.
With a better understanding of the reaction mechanism, we
turned our attention to the ow synthesis. A two-stream gas–
liquid continuous ow apparatus was rst constructed
(Scheme 3).6 A screen of water-miscible solvents suitable for use
in the ow system was carried out, with DMF emerging as the
best solvent to ensure a homogenous reaction solution with
retention of good reactivity. A temperature study with styrene in
a stainless steel tubing (SS-tubing) reactor (2 mL) indicated that
styrene oxide 4a was the major product at low temperature, and
the yield of carbonate 1a was higher at elevated temperatures.
However, above 100 ꢀC, diol byproducts were observed,13 and
the ow rate became unsteady. Further optimization of the ow
condition revealed that the packed-bed reactor (packed with
stainless steel powder, 325 mesh, stream, 1.2 mL space volume)
was more efficient than the SS-tubing reactor, and the ow rate
became steady even at high temperatures probably due to better
mixing.14
Even though the two-stream ow technique exhibited
remarkable efficiency compared with conventional batch
conditions in the case of styrene,15 several limitations in this
system were identied. Most importantly, aliphatic olens
showed only sluggish reactivity under optimal conditions.
Moreover, nonpolar olens such as 1-octene and vinyl-
naphthalene are poorly soluble in the DMF–H2O co-solvent.
Extensive studies aimed at improving the reaction rate of
aliphatic olens, such as applying phase-transfer-reagents,
changing residence time and concentration, and utilizing
different brominating reagents and bases, proved fruitless.
Due to the poor solubility and incompatibility of starting
materials at high concentration (e.g. olens and NBS generated
dibromide in DMF),16 multi-stream ow systems (three-stream
and four-stream) were introduced.14 Even though the reaction
rate of aliphatic olens was not signicantly improved, it is
Scheme 2 Possible reaction pathways.
1228 | Chem. Sci., 2014, 5, 1227–1231
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
maintaining a steady ow. N2 was employed for back pressure
and a slow bleed was utilized to regulate the overall pressure of
the system. Aer steady state was achieved (ꢁ4 Â tR, 2 h), the
nal eluent stream was sampled using a 6-way valve, which
showed 75% yield of product 1c.
With the optimized ow conditions in hand, we investigated
the substrate scope (Scheme 4). A variety of aliphatic olens
were converted into the corresponding cyclic carbonates with
good yield. Interestingly, the ow conditions not only tolerated
functionalities such as ether, nitrile, silane and ester (1d to 1g),
but also allowed for conversion of even more challenging
substrates like ketone and alcohol that had potential to undergo
cyclization under basic conditions (1h, 1i). The reaction of
ethylene proceeded efficiently to furnish 1j in good yield, which
provides a successful example of the power of sequential as
opposed to simultaneous introduction of different gas reagents
into a continuous ow system. Aromatic carbonates were
generated effectively at a slightly lower temperature to avoid
diol formation. Electron-poor aromatics appeared to afford
Scheme
reactor.
3 Transformations in the two-stream gas–liquid flow
worth noting that the order of the gas stream can be easily
switched in the ow system, which is hard to achieve with
pressurized autoclaves.
To overcome the low conversion problem associated with
aliphatic olens, a different design was considered. Based on
the mechanistic study, it was concluded that DBU signicantly
decreased the reaction rate of the oxidative carboxylation due to
the interaction with NBS. It was hypothesized that if NBS and
DBU were introduced into the ow system at different stages
during the reaction, an enhanced reaction rate should be
achievable. Furthermore, the formation of bromohydrins in the
rst stage of the process17 could also avoid the dibromide
by-products observed in batch conditions (Scheme 2).
In this regard, we rst tested the sequential transformation
of 1-octene as the model substrate in the continuous ow
system. The use of acetone–H2O (1 : 1), which was the only
solvent mixture found to provide a homogenous solution in
both the bromohydroxylation and cyclization steps,18 precluded
the need to use phase-transfer-reagents. The addition of a
catalytic amount of NH4OAc facilitated conversion of olens to
bromohydrins.19 In order to achieve complete consumption of
starting olens, a temperature of 40 ꢀC was maintained during
the bromohydroxylation step. In the ow setup shown in
Scheme 4, aer the system reached the appropriate pressure
(130 psi of CO2), an acetone solution of 1-octene and NBS20 and
an aqueous NH4OAc solution were introduced by a Syrris Asia
pump. The organic and aqueous streams met at a T-mixer, and
were introduced into a PFA tubing reactor with a 30 minute
residence time. Then the CO2 stream was metered into the
system using a mass ow controller, which met the liquid ow
at a Y-mixer, and a 1 : 1 (v/v) liquid–gas slug ow stream was
observed at the outlet. The aqueous DBU solution was intro-
duced last and combined with the gas–liquid ow in a stainless
steel T-mixer at 100 ꢀC; in this way a possible epoxide formation
was avoided. The resulting gas–liquid segmented ow was
passed through a packed-bed reactor (lled with SS-powder) at
ꢀ
100 C for 10 minutes. The acetone–water co-solvent could be
heated at 100 ꢀC at 130 psi without vapor generation while Scheme 4 Sequential transformation in flow.a
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1227–1231 | 1229

View Article Online
Chemical Science
Edge Article
Scheme 6 Comparison between batch and flow sequential reactions.
Scheme 5 Long term flow at the steady state.
higher yield than electron-rich ones (1k to 1m). Notably, some
heteroaryl olens were also amenable substrates for our ow
reactors (1n to 1p). No epoxide or dibromide by-product was
observed in these cases.
Conclusions
Overall, an efficient ow synthesis of cyclic carbonates starting
directly from olens and CO2 has been achieved (Scheme 7).
The ow synthesis was integrated into a hydroxybromination–
carboxylation two-step sequential transformation, which
represents a successful example of a multi-stage gas–liquid
continuous ow process. Specically, incompatible reagents
(NBS and DBU) were introduced easily without interacting with
each other, thus signicantly enhancing the reaction rate,
especially for aliphatic olens. These sequential ow systems
enable optimization of individual steps and allow numerous
experiments to be conducted at various residence times and
temperatures aer the initial loading of reagents, highlighting
the ease with which these operating conditions can be adjusted
in real time. Compared to conventional batch conditions, the
ow systems can be readily operated with a broad range of
substrates, with enhanced reaction rates and product yields by
avoiding the formation of epoxide or dibromide by-products.
Other merits of this gas–liquid ow system include a packed-
bed reactor used for carboxylation, which enabled efficient
mixing of the phases while stabilizing the ow patterns, i.e.,
ensuring steady ow. Acetone was applied as the co-solvent with
water to achieve a homogenous liquid solution at elevated
To challenge the newly developed ow conditions further,
internal olens with increasing steric bulk were chosen as
reaction partners. Cyclopentene was a good candidate for this
transformation and product 1q was obtained in high yield. The
reaction of cyclohexene proceeded at a slower rate to give the
desired product 1r in approximately 20% yield along with the
trans-diol by-product. With increased CO2 pressure and a lower
amount of H2O, the yield of 1r was improved to 49%. Impor-
tantly, trans-3-hexene afforded the desired carbonate 1t stereo-
specically with high yield. Moreover, even though two different
bromohydrins were formed with differentially substituted
alkenes, both were converted to the same cyclic product (1u).
Cis-3-hexene was transformed to carbonate 1v in a similar yield,
with the observation of partial isomerization. It should be noted
that this is the rst report of a direct synthesis of cyclic
carbonates from acyclic disubstituted olens, which were
widely considered to be difficult substrates for epoxide cyclo-
addition.21 Gem-disubstituted olens were also amenable to
this transformation, delivering product 1s in a moderate yield.
In cases where the product yields were not high, diol byproducts
were normally encountered.
Based on data presented in Scheme 4 and the residence time
investigated, products could be synthesized at a rate of
approximately 0.6 mmol hÀ1. To test this, experiments were
repeated using the optimal conditions, and the product stream
was collected for 8 hours at steady state. As shown in Scheme 5,
the isolated product yields were comparable to those in
Scheme 4.
Thus, both aromatic and aliphatic olens as well as mono-
substituted and disubstituted olens were converted to cyclic
carbonates in the continuous ow system within 40 minutes,
with full conversion and moderate to good yield. On balance,
our sequential transformation in the ow reactor compared very
favourably with reported conditions in batch (which required
>3 h for aromatic olens, >5 h for aliphatic olens, and in which
dibromide byproducts were detected).8a Batch reactions of the
sequential transformation were also conducted in pressurized
autoclaves (Scheme 6). In all cases, the products were accom-
panied by a large amount of epoxides, which were not observed
under the ow condition.22
Scheme 7 Function of each reactor feature.
1230 | Chem. Sci., 2014, 5, 1227–1231
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
temperature to avoid the use of phase-transfer reagents, thus
moving toward even greener processes.
M. North and M. Omedes-Pujol, Tetrahedron Lett., 2009, 50,
4452–4454; (e) J. Sun, L. Liang, J. Sun, Y. Jiang, K. Lin,
X. Xu and R. Wang, Catal. Surv. Asia, 2011, 15, 49–54.
8 There are only sporadic reports of direct synthesizing cyclic
carbonates from aliphatic olens and CO2, associated with
problems such as low yield, low conversion, requiring
excess reagents and high pressure of CO2: (a) N. Eghbali
and C.-J. Li, Green Chem., 2007, 9, 213–215; (b) D. Bai and
H. Jing, Green Chem., 2010, 12, 39–41; (c) F. Chen, T. Dong,
T. Xu, X. Li and C. Hu, Green Chem., 2011, 13, 2518–
2524.
Acknowledgements
We gratefully acknowledge nancial support from the DOE (DE-
FOA-0000253) and the Siemens Corporation. We thank Li Li
(MIT) for obtaining mass spectrometric data.
Notes and references
1 Recent reviews on continuous ow chemistry: (a)
9 Conversion of olens to cyclic carbonates using a catalytic
amount of bromide ion together with excess aqueous H2O2
failed in the ow system due to decomposition of H2O2
towards stainless steel: S. S. Lin and M. D. Gurol, Environ.
Sci. Technol., 1998, 32, 1417–1423.
¨
¨
K. Jahnisch, V. Hessel, H. Lowe and M. Baerns, Angew.
Chem., Int. Ed., 2004, 43, 406–446; (b) B. P. Mason,
K. E. Price, J. L. Steinbacher, A. R. Bogdan and
D. T. McQuade, Chem. Rev., 2007, 107, 2300–2318; (c)
S. V. Ley and I. R. Baxendale, Proc. Bozen Symp., Systems 10 It has been reported that DBU activated CO2 in H2O to form a
Chem., 2008, 65–85; (d) R. L. Hartman and K. F. Jensen,
Lab Chip, 2009, 9, 2495–2507; (e) D. Webb and
T. F. Jamison, Chem. Sci., 2010, 1, 675–680; (f)
R. L. Hartman, J. P. McMullen and K. F. Jensen, Angew.
Chem., Int. Ed., 2011, 50, 7502–7519; (g) N. G. Anderson,
Org. Process Res. Dev., 2012, 16, 852–869; (h) J. Yoshida,
bicarbonate salt of DBU instead of a zwitterionic adduct,
which exhibited good reactivity for various CO2-xation
reactions: (a) D. J. Heldebrant, P. G. Jessop, C. A. Thomas,
C. A. Eckert and C. L. Liotta, J. Org. Chem., 2005, 70, 5335–
5338; (b) M. Yoshida, Y. Komatsuzaki and M. Ihara, Org.
Lett., 2008, 10, 2083–2086.
Y. Takahashi and A. Nagaki, Chem. Commun., 2013, 49, 11 The structure of complex D was not fully determined at this
9896–9904.
2 For select recent examples, see: (a) B. K. H. Yen, A. Gunther,
stage. For control experiments exploring the interaction
between NBS and DBU, see ESI.†
¨
M. A. Schmidt, K. F. Jensen and M. G. Bawendi, Angew. 12 For further control experiments to elucidate the mechanism,
Chem., Int. Ed., 2005, 44, 5447–5451; (b) M. T. Rahman,
see ESI.†
T. Fukuyama, N. Kamata, M. Sato and I. Ryu, Chem. 13 Cyclic carbonates are known to be hydrolyzed under basic
Commun., 2006, 2236–2238; (c) E. R. Murphy,
J. R. Martinelli, N. Zaborenko, S. L. Buchwald and
K. F. Jensen, Angew. Chem., Int. Ed., 2007, 46, 1734–1737;
condition to give diols: D. J. Cross, J. A. Kenny, I. Houson,
L. Campbell, T. Walsgrove and M. Wills, Tetrahedron:
Asymmetry, 2001, 12, 1801–1806.
(d) A. Polyzos, M. O'Brien, T. P. Petersen, I. R. Baxendale 14 For condition optimization, see ESI.†
and S. V. Ley, Angew. Chem., Int. Ed., 2011, 50, 1190–1193.
3 For recent excellent examples that address the reagent
15 See ESI for studies on comparing the two-stream ow with
conventional batch conditions.†
incompatibility problem by ash chemistry in ow 16 C. Caristi, G. Cimino, A. Ferlazzo, M. Gattuso and M. Parisi,
microreactors, see: (a) A. Nagaki, H. Kim and J. Yoshida, Tetrahedron Lett., 1983, 24, 2685–2688.
Angew. Chem., Int. Ed., 2008, 47, 7833; (b) A. Nagaki, 17 Several reports have demonstrated the feasibility of
H. Kim and J. Yoshida, Angew. Chem., Int. Ed., 2009, 48,
8063–8065; (c) H. Kim, A. Nagaki and J. Yoshida, Nat.
Commun., 2011, 2, 264; (d) A. Nagaki, C. Matsuo, S. Kim,
K. Saito, A. Miyazaki and J. Yoshida, Angew. Chem., Int. Ed.,
2012, 51, 3245–3248.
4 Selected recent reviews, see: (a) T. Sakakura, J. Choi and
H. Yasuda, Chem. Rev., 2007, 107, 2365–2387; (b) I. Omae,
Coord. Chem. Rev., 2012, 256, 1384–1405.
converting bromohydrins to carbonates with CO2 using
ammonium salts, ionic liquids, or poly(ethyleneglycol): (a)
C. Venturello and R. D'Aloisio, Synthesis, 1985, 33; (b)
J. S. Yadav, B. V. S. Reddy, G. Baishya, S. J. Harshavardhan,
J. Chary and M. K. Gupta, Tetrahedron Lett., 2005, 46,
3569–3572; (c) J. Wang, L. He, X. Dou and F. Wu, Aust.
J. Chem., 2009, 62, 917–920.
18 Other solvent–water mixtures evaluated included THF, DMF,
dioxane, and MeCN with water.
5 Selected reviews: (a) J. H. Clements, Ind. Eng. Chem. Res.,
2003, 42, 663–674; (b) D. J. Darensbourg, Chem. Rev., 2007, 19 B. Das, K. Venkateswarlu, K. Damodar and K. Suneel, J. Mol.
107, 2388–2410. Catal. A: Chem., 2007, 17–21.
6 J. A. Kozak, J. Wu, X. Su, F. Simeon, T. A. Hatton and 20 Unlike in DMF, olens and NBS did not react with each other
T. F. Jamison, J. Am. Chem. Soc., 2013, 135, 18497–
18501.
7 For selected examples, see: (a) J. Sun, S. Fujita, F. Zhao,
M. Hasegawa and M. Arai, J. Catal., 2005, 398–405; (b)
in acetone.
´
21 C. J. Whiteoak, N. Kielland, V. Laserna, E. C. Escudero-Adan,
E. Martin and A. W. Kleij, J. Am. Chem. Soc., 2013, 135, 1228–
1231.
F. Ono, K. Qiao, D. Tomida and C. Yokoyama, Appl. Catal., 22 For control experiments exploring the conversion of
A, 2007, 107–113; (c) J.-L. Wang, J.-Q. Wang, L.-N. He,
X.-Y. Dou and F. Wu, Green Chem., 2008, 10, 1218–1223; (d)
epoxides to cyclic carbonates under ow conditions and
related discussion, see ESI.†
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1227–1231 | 1231