Catalytic coupling of epoxides and CO2 to cyclic carbonates by carbon nanotube-supported quaternary ammonium salts

Article abstract of DOI:10.1016/j.apcata.2014.09.034

Quaternary ammonium chlorides bound to multi-walled carbon nanotubes as a catalyst for coupling of CO₂ and epoxides to produce cyclic carbonates were explored. Reaction variables such as the epoxide structure, the length of alkyl substituents in the quaternary ammonium salts and the spacer chain on the catalytic performance were discussed. The yield of the cyclic carbonates varied between 7 and 89% after 6 h at 110 °C under low pressure (2 MPa of CO₂). The epoxide:catalyst mass ratio was 20-30, while 1 mmol g^-1 of the quaternary salt was grafted on the carbon nanotubes. A synergy between carboxyl moiety and ammonium moiety grafted on carbon nanotubes was found, and a strong impact of the length of the spacer group used for grafting of the quaternary ammonium salt on nanotubes was observed. The best performance was achieved with short (2 carbon atoms) and long (10 atoms) spacer groups, while a middle-sized spacer group (6 atoms) was not suitable. The length of the alkyl chain of the substituents of the ammonium salt (head group) had a low impact where ethyl and methyl groups performed better than butyl. The reactivity of epoxides was as follows: epichlorohydrin > propylene oxide > styrene oxide. Observations were rationalized by a mechanism where Br?nsted's sites on the surface of nanotubes play an important role during carboxylation of epoxides. The catalyst can easily be separated by filtration recycled without a significant decrease in the catalytic activity if dried properly between the runs.

Full text of DOI:10.1016/j.apcata.2014.09.034

Applied Catalysis A: General 488 (2014) 96–102
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Catalytic coupling of epoxides and CO2 to cyclic carbonates by carbon
nanotube-supported quaternary ammonium salts
Stefan Baja,∗, Tomasz Krawczyka, Katarzyna Jasiaka, Agnieszka Siewniaka,
Mirosława Pawlytab
a Department of Chemical Organic Technology and Petrochemistry, Faculty of Chemistry, Silesian University of Technology, ul. Krzywoustego 4,
44-100 Gliwice, Poland
b Institute of Engineering Materials and Biomaterials, Silesian University of Technology, Konarskiego 18a, 44-100 Gliwice, Poland
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 25 June 2014
Received in revised form 28 August 2014
Accepted 22 September 2014
Available online 2 October 2014
Quaternary ammonium chlorides bound to multi-walled carbon nanotubes as a catalyst for coupling of
CO2 and epoxides to produce cyclic carbonates were explored. Reaction variables such as the epoxide
structure, the length of alkyl substituents in the quaternary ammonium salts and the spacer chain on
the catalytic performance were discussed. The yield of the cyclic carbonates varied between 7 and 89%
after 6 h at 110 ◦C under low pressure (2 MPa of CO2). The epoxide:catalyst mass ratio was 20–30, while
1 mmol g−1 of the quaternary salt was grafted on the carbon nanotubes. A synergy between carboxyl
moiety and ammonium moiety grafted on carbon nanotubes was found, and a strong impact of the length
of the spacer group used for grafting of the quaternary ammonium salt on nanotubes was observed. The
best performance was achieved with short (2 carbon atoms) and long (10 atoms) spacer groups, while a
middle-sized spacer group (6 atoms) was not suitable. The length of the alkyl chain of the substituents
of the ammonium salt (head group) had a low impact where ethyl and methyl groups performed better
than butyl. The reactivity of epoxides was as follows: epichlorohydrin > propylene oxide > styrene oxide.
Observations were rationalized by a mechanism where Brønsted’s sites on the surface of nanotubes play
an important role during carboxylation of epoxides. The catalyst can easily be separated by filtration
recycled without a significant decrease in the catalytic activity if dried properly between the runs.
© 2014 Elsevier B.V. All rights reserved.
Keywords:
Carbon dioxide
Cyclic carbonates
Epoxides
Multi-walled carbon nanotubes
Quaternary ammonium salts.
1. Introduction
1943 by IG Farben [2]. In 2010, the annual output of cyclic car-
bonates reached 0.1 Mt [3]. These compounds are widely used
The issue of carbon dioxide utilization is particularly impor-
tant, especially in the view of toughening the regulations on the
discharge limits of greenhouse gases into the atmosphere. The
necessity of lowering carbon dioxide emission to the atmosphere
has brought about advancements in the research on using this feed-
stock for chemical syntheses. This is evidenced in the substantial
number of publications, which is constantly growing, or the num-
ber of subsidies granted to the research on the new methods of
CO2 management. Regardless of the environmental policy, CO2 is
a cheap, abundant and readily available C1 source for chemical
synthesis [1]. One of the possible uses of CO2 is in the cyclic car-
bonate synthesis, which enabled to eliminate the environmentally
unfriendly process of phosgenation of glycols. The cyclic carbon-
ate synthesis from carbon dioxide and epoxide was patented in
in producing polymers and polycarbonates. Ethylene and propyl-
ene carbonates are used as polar aprotic solvents. They dissolve
both small-molecule compounds and polymers, e.g., polyacryloni-
trile and polysuccinimide. Since propylene carbonate is not toxic,
it is used as a solvent for paint strippers, plastics and epoxy resins.
Typically, cyclic carbonates are used as solvents in electrolytes for
lithium-ion batteries. Alkylene carbonates are also utilized as mod-
ifiers of solid fast-crosslinking polyurethanes and polyurethane
glues. Moreover, these compounds may be used as intermediate
products in the fine chemicals synthesis, e.g., of polycarbonates or
polyurethanes. Cyclic carbonates have also found their use in phar-
maceutical and cosmetic industries as an additive for clay gellants
[4].
Major research effort in recent years has been devoted to the
development of a new catalytic system for the cyclic carbonates
synthesis from CO2 and various reagents, such as epoxides or
alkenes. Among others, quaternary onium salts and ionic liquids
were found to be one of the most effective ones. The current
∗
Corresponding author. Tel.: +48 32 2372973.
E-mail address: Stefan.Baj@polsl.pl (S. Baj).
http://dx.doi.org/10.1016/j.apcata.2014.09.034
0926-860X/© 2014 Elsevier B.V. All rights reserved.

S. Baj et al. / Applied Catalysis A: General 488 (2014) 96–102
97
O
O
O
HO(CH2)n+1N+(R1)3Cl-
DMF, pyridine
H2SO4, HNO3
5h, 50 °C
OH
OH
OH
Cl
OH
R1
SOCl2
R1
N
R1
Cl
O
n
O
O
O
OXCNT-(CH2)n+1(R1)3N+Cl-
CNT
OXCNT
Scheme 1. General procedure of the synthesis of CNT-ammonium salt catalysts.
research is focused on the improvement of the existing systems to
develop highly active catalytic agents that would be non-toxic and
environmentally friendly. Due to the potential cost reduction, it is
vital to apply immobilized catalysts, as they can be easily separated
from the post-reaction mixture [5].
2.2. Synthesis of quaternary ammonium salts
The method of hydroxyl functionalized quaternary salts has
been described elsewhere [10]. In typical procedure, tri-
a
ethylamine (0.08 mol), dry toluene (25 mL) and 2-chloroethanol
(0.08 mol) were stirred for 24 h at 70 ◦C under nitrogen atmosphere.
Upon completion of the reaction, the mixture was cooled to the
room temperature, filtered out and washed successively by anhy-
drous ether (3 × 15 mL) and dry acetonitrile (2 × 10 mL). Next, the
product was dried under vacuum (60 ◦C, 10 mbar, 24 h). The other
quaternary ammonium salts were obtained following a similar pro-
cedure (see supporting information).
Recently, Luo and co-workers have used a series of cross-linked
polydivinylbenzene polymers covalently grafted with hydroxyl,
carboxyl and amino functionalized ionic liquid as effective het-
erogeneous catalysts for the synthesis of cyclic carbonates in the
reaction of CO2 with epoxides. These catalysts showed excellent
reusability and stability [6]. Park and co-workers described the
application of a series of heterogeneous zinc-containing ionic liq-
uids on the silica gel as catalysts in the synthesis of cyclic carbonates
from epoxides and CO2. The incorporation of zinc ions in the immo-
bilized ionic liquids significantly enhanced the catalytic activity [7].
Dyson and co-workers [8] reported that cross-linked ionic poly-
mers based on styrene-functionalized imidazolium with chloride
are stable and can be easily recycled and reused.
Carbon nanotubes (CNTs), among other nanomaterials, have
been the object of extensive research in recent years. This arose
from the fact that they have numerous beneficial properties, such as
a substantial surface area, high mechanical and thermal resistance.
This material found a handful of application in the cyclic carbon-
ate synthesis. For example, 1-hydroxyethyl-3-methylimidazolium
halides immobilized on CNTs have high catalyst activity, and may
be recirculated five times without any substantial drop in activity.
The studies have shown that these catalysts exhibited significantly
enhanced catalytic activity in the CO2 and epoxides reaction in
comparison to conventional heterogeneous supports based on sil-
ica and polymers [9].
2.3. Immobilization of quaternary ammonium salts on CNTs
The synthesis was conducted according to Scheme 1. Com-
mercial CNT (5.3 g) were oxidized for 5 h with 320 mL of the
HNO3–H2SO4 mixture (1:3, v/v) in an ultrasonic bath at 50 ◦C.
The resulting black viscous suspension was diluted with water
(1500 mL), filtered through a 0.2 ␮m nylon filter and washed thor-
oughly with water, acetone, methanol and dried in an oven (100 ◦C).
The dry product (5.1 g) was mixed with an excess of thionyl chloride
(60 g) and stirred overnight at the room temperature. The excess of
SOCl2 was removed at reduced pressure to give 4.7 g of the prod-
uct. A sample of dry activated oxidized carbon nanotubes (OXCNTs)
(0.8 g) was dispersed in dry DMF (75 mL) and the excess of hydroxyl
functionalized quaternary salt was added (0.45 g). The reaction was
conducted in an ultrasonic bath at 50 ◦C for 1 h; then dry pyridine
(20 mL) was added and the reaction was continued overnight with
magnetic stirring at the room temperature. The suspension was
then filtered through a 0.2 ␮m nylon filter and thoroughly washed
with DMF, water, acetone, methanol and cyclohexane. Finally, it
was azeothropically dried for 3 h using cyclohexane. The yield of
the dry product was 0.7 g.
Herein, the aim of the research was to develop an effective
method of the cyclic alkylene carbonate synthesis from carbon
dioxide and selected epoxides with the use of catalysts based on
the quaternary ammonium salts immobilized on CNT. The scope
of the research included the synthesis of catalysts and immobiliza-
tion, as well as performing a reaction of obtaining cyclic carbonates
using these immobilized catalysts in order to select an effective and
stable system suitable for practical application.
2.4. Synthesis of cyclic carbonates
All the reactions were carried out in
a 100 mL stainless
steel pressure reactor (Mettler Toledo) with a mechanical stirrer
(pitched-blade) immersed in a solid-state thermostat (EasyMaxTM
102) with an automatic temperature control system. The pressure
reactor was also equipped with temperature and pressure sensors
and a rupture disk. In a typical procedure, propylene oxide (15 mL,
0.21 mol) and the appropriate amount of the catalyst (Table 1 and
Table 3) were added into the reactor. The reactor was purged with
carbon dioxide twice and then heated to 110 ◦C. CO2 was charged
into the reactor from a reservoir tank to a chosen constant pres-
sure (2 MPa). The reaction progress was monitored by means of
CO2 consumption rate. After the reaction was completed, the reac-
tor was cooled to the ambient temperature, and the excess CO2 was
vented. The catalyst system was separated from the reaction mix-
ture by simple filtration. Samples from the organic layer were taken,
dissolved in CH2Cl2 and analyzed by gas chromatography (Perkin
Elmer GC, Clarus 500) equipped with a capillary column (SPBTM-5:
2. Experimental
2.1. Materials
Carbon dioxide (99.5%, from SIAD) was used without further
purification. Octane (>99%), (chloromethyl)ethylene carbonate,
propylene oxide (>99.5%), 2-chloroethanol (99%), 6-chloro-1-
hexanol (>95%), 10-chloro-1-decanol (90%), trimethylamine (>99%)
CNT (6–9 nm × 5 ␮m) were purchased from Sigma Aldrich.
Epichlorohydrin (>99%) was obtained from Fluka. Propylene
carbonate (99.5%), styrene oxide (>97%), triethylamine (99%), trib-
utylamine (99%), (2-hydroxyethyl)trimethylammonium chloride
(99%) were bought from Acros Organics. Other chemicals were of
reagent grade supplied by local manufacturers.

98
S. Baj et al. / Applied Catalysis A: General 488 (2014) 96–102
Table 1
Synthesis of a cyclic carbonate in the presence of OXCNT and ammonium catalyst from propylene oxide and CO2.
No.
System
OXCNT loading (g)
Quaternary ammonium
salt loading (mmol)
CO2 conversion (%)
Selectivitya (%)
1
2
3
4
5
6
7
8
No catalyst
OXCNT
Choline chloride
Trimethyl-(6-hydroxyhexyl)ammonium chloride
Trimethyl-(10-hydroxydecyl)ammonium chloride
OXCNT + choline chloride
Choline chloride + 1 mmol CH3COOH
Choline chloride + 10 mmol CH3COOH
0
0.65
0
0
0
0.65
0
0
0
0
1
1
1
1
1
1
0
–
–
Trace
16
15
18
35
37
38
95
95
95
96
96
95
Reaction conditions: epoxide (0.214 mol, 15 mL); 110 ◦C; 2 MPa of CO2; 500 rpm; reaction time 6 h.
a
Analysed by GC.
30 m × 0.25 mm × 0.25 ␮m film thickness). 1-Octane was used as
the absence of a catalyst. Interestingly, the addition of OXCNTs
increased slightly the yield of the cyclic carbonate but the concen-
tration of the product remained at trace level. More pronounced
effect had the presence of a quaternary ammonium salts (entries
3–5) when the conversion of CO2 reached 15–18%. No significant
effect of the structure of the ammonium salt was observed (entries
3–5). In contrast, the presence of both the quaternary ammonium
catalyst and oxidized CNTs yielded a 35% conversion of CO2, and
equal selectivity. Those preliminary results encouraged us to inves-
tigate the possibility of immobilizing an ammonium salt on the
CNT surface chemically to obtain a recyclable catalyst of high activ-
ity. To confirm whether the CNT itself or carboxylic acid moieties
an internal standard.
3. Results and discussion
The investigation of the possibility of application of the quater-
nary ammonium salts immobilized on OXCNTs was initiated by a
set of experiments in which a non-immobilized quaternary ammo-
nium salt and OXCNTs were investigated as potential catalysts for
the cyclic carbonate synthesis. The results are presented in Table 1.
From the control runs presented in Table 1, it was clear that
the reaction of CO2 with propylene oxide did not proceed in
Fig. 1. TEM and SEM characterization of catalysts and intermediate materials. (A, B) CNT; (C, D) OXCNT; (E–I) OXCNT-(CH2)n+1(R1)3N+Cl−; (E, G) R1 = Et, n = 1; (F) R1 = Bu,
n = 9; (H, I) R1 = Me, n = 5.

S. Baj et al. / Applied Catalysis A: General 488 (2014) 96–102
99
Table 2
be identified while comparing images A, C, E, F and I. Also higher
resolution images G and H suggests more loosely packed bundles
when n = 9 group was used.
Structural features and elemental composition of CNT-supported ammonium chlo-
rides OXCNT-(CH2)n+1(R1)3N+Cl−
.
R1
n
C (%)
H (%)
N (%)
1.43
N (mmol g−1
)
FT-IR analysis confirmed successful immobilization of the qua-
ternary ammonium salts onto carbon nanotubes. Fig. 2 shows the
FT-IR spectra of the homogeneous catalyst, oxidized CNTs and
immobilized catalyst. The characteristic peaks of oxidized CNTs
around 1617 and 1385 cm−1 was observed. They can be ascribed to
C=O and C-O stretching of carboxyl group. Free quaternary ammo-
nium salt exhibited the characteristic bands which occurred in
the region of 2869–2934 cm−1. This stretching vibrations of C-H
bond are specific for the molecules having a tertiary N-CH3 group.
Moreover, the spectra contained peak appearing at 1055 cm−1 cor-
responding to the C-N stretch. The peak appeared in the region
of 3347 cm−1 is ascribed to H2O that was not removed during
drying of the samples. In the case of immobilized ammonium
salt, characteristic peaks of support (1635, 1538 cm−1) and quater-
nary ammonium salt (1199, 2861, 2934 cm−1) confirmed successful
immobilization.
Bu
Et
Me
Me
Et
Bu
Me
Et
1
1
1
9
9
9
5
5
5
83.54
83.1
>0.09
0.13
0.55
0.33
0.15
1.08
>0.09
0.96
0.99
1.02
1.18
1.69
1.42
0.93
0.85
0.94
0.56
0.96
1.65
2.36
1.99
1.30
1.19
1.32
0.79
1.34
75.72
89.67
83.67
74.34
84.7
84.15
83.64
Bu
display catalytic activity we added acetic acid to the reaction sys-
tem instead of OXCNT (entries 7–8). Results were similar as with
OXCNTs. Therefore, carboxylic moieties play a dominant catalytic
role in the process.
3.1. Catalysts synthesis and characterization
The results of the TGA analysis (shown in Fig. 3) of catalysts,
OXCNTs and CNTs can be clearly distinguished. CNTs themselves
are thermally stable and only small peaks (53 and 85 ◦C) which
most likely corresponds to the desorption of water or other sub-
stances. Significant decomposition of this material is visible only
above 400 ◦C. OXCNTs were more susceptible to decomposition,
and additional peaks between 200 and 400 ◦C appeared. In the case
of catalysts, the degradation occurred in the same range of tem-
perature. In all cases, 2–3 peaks were visible at approximately 200
and 300 ◦C. Exact position of those peaks depended on the struc-
ture of alkyl substituents. If weight loss that occurred between 200
and 400 ◦C was used to calculate the nitrogen content, the results
were in good agreement with those obtained by means of elemental
analysis.
Catalysts were synthesized according to Scheme 1. In total, nine
different substances were obtained and characterized by elemental
analysis (Table 2), SEM and TEM (Fig. 1), IR (Fig. 2), and TGA (Fig. 3).
From Fig. 1, it is clear that oxidation of CNTs did not substantially
change the length of the tubes. Probably functional groups were
only introduced randomly on the surface where structural defects
facilitate the oxidation. Further functionalization with ammonium
salts did not alter the microscopic structure of the material. How-
ever, small differences in arrangement of bundles of nanotubes may
3.2. Catalysts performance
Among possible factors that affect the yield of cyclic carbonate
synthesis using immobilized ammonium catalysts are the structure
of alkyl substituents of ammonium nitrogen, halide type and the
epoxide structure. In the case of halides, the views on that matter
are divided.
In a series of recently published papers, where tetraalkylammo-
nium halides were considered as a possible factor in carboxylation
of epoxides, it was demonstrated that the impact of the halide type
strongly depends on a particular reaction system. The reason is
that the opening of the epoxy ring occurs through the nucleophilic
attack of a halide anion on the less-hindered carbon atom of epox-
ide. Therefore, the nucleophilicity of the halide is very important to
facilitate this step while its leaving ability is crucial during the last
step when cyclic carbonate is formed. However, experimental data
vary and different orders of reactivity of halides were reported.
In the case of ionic liquids immobilized on the carboxymethyl
cellulose (CMIL), it was found that propylene carbonate yield
increased in the order of CMIL-I (98.6%) > CMIL-Br (81.7%) > CMIL-
Cl (73.2%) [11]. Similar order was reported for imidazolium ionic
liquids immobilized on chitosan (CS) where the yield of propyl-
ene carbonate increased with the increase in nucleophilicity of
the anion (CS-EMImBr 96% > CS-EMImCl 74%) [12]. More profound
effect was observed when quaternary phosphonium salts immobi-
lized on silica was used for the same reaction. In that case chloride
resulted in 29% yield comparing with 100% for iodide [13]. Other
authors reported that fluctuations of catalytic activity of different
halides was in a narrow range of 1–15% [9,14,15] or no apparent
influence was visible [9,12,16,17].
Fig. 2. IR spectra of OXCNT, free ammonium salt and one of the final products.
Interestingly, it was found that in certain instances the yield of
carbonates did not rise according to the increase of the catalyst’s
Fig. 3. Representative DTG curves of catalysts and intermediate materials.

100
S. Baj et al. / Applied Catalysis A: General 488 (2014) 96–102
Table 3
Effect of the structure of a catalyst and type of epoxide on the cyclic carbonate formation.
R1
n
Epoxide
Catalyst
loading (g)
CO2
conversion
(%)
Yield of
carbonatea
(%)
Selectivitya
(%)
k1 × 106
TOFc (mol
epoxide
k1N × 106
b
(s−1
)
(mmol
mol N−1 s−1
)
N−1 s−1
)
a
Me
Et
Bu
Me
Et
Bu
Me
Et
1
1
1
9
9
9
5
5
5
Styrene oxide
0.57
0.66
0.46
0.77
0.58
0.70
0.53
0.55
0.62
23
68
53
86
17
21
7
22
66
50
81
16
20
6.8
12
15
96
98
96
94
97
98
97
92
96
20.3
66.7
24.9
110.6
14.4
8.9
3.4
5.5
13.4
0.0023
0.0083
0.0107
0.0073
0.0030
0.0034
0.0014
0.0039
0.0025
21.1
85.6
53.0
101.0
26.7
15.0
6.80
17.7
22.6
Propylene oxide
Epichlorohydrin
Epichlorohydrin
Styrene oxide
Propylene oxide
Propylene oxide
Epichlorohydrin
Styrene oxide
13
16
Bu
Reaction conditions: epoxide (0.214 mol); 110 ◦C; 2 MPa of CO2; 500 rpm; reaction time 6 h.
a
Analysed by GC.
Measured within the period of 3000–7000s.
Averaged over the entire 6 h.
b
c
anion nucleophilicity [18]. In one case, it was rationalized by lower
stability of an iodide comparing with a bromide under employed
reaction conditions [14]. In 2008, an article was published regarding
the reaction of carbon dioxide with butyl glycidyl ether (BGE). This
article showed that steric hindrance can have stronger influence on
the course of the reaction of obtaining cyclic carbonates than the
nucleophilicity of anions. In connection with this, the conversion
of BGE increased in the order of Cl− > Br− > I− [19].
Some authors reported that cyclic carbonates can be obtained
in similar yields regardless of the catalyst’s halide type if reaction
conditions are changed to offset different reactivity [20]. For exam-
ple, catalytic activity of Zn3[Co(CN)6]2/TBAB (bromide) system at
120 ◦C was the same as in the case of Zn3[Co(CN)6]2/TBAC (chloride)
at 140 ◦C. In both cases, the yield of styrene carbonate amounted
in to 97% [16]. A similar phenomenon was described by Zhao et al.
[21].
Another view on the subject was presented by other authors
who attributed variation in catalytic activity to physical proper-
ties of halides. Shim et al. [22] claimed that the yield of styrene
carbonate depend on the melting point of quaternary ammonium
salt they applied. Zhou et al. [23] noticed that HBetCl (1-carboxy-
N,N,N-trimethylmethanaminium chloride) showed higher catalytic
activity than HBetBr, which can be explained by the difference in
Fig. 4. Progress of the cyclic carbonate synthesis from epoxides followed by the
consumption of CO2. Reaction conditions: epoxide (0.214 mol); catalyst (≈0.5 g);
110 ◦C; 2 MPa; 500 rpm. ECH, epichlorohydrin; PO, propylene oxide; SO, styrene
oxide.
Table 4
Impact of the investigated factor on the carbonate yield. Reaction conditions: cata-
lyst loading, 0.46-0.77 g; epoxide,0.214 mol; 110 ◦C; 2 MPa of CO2; 500 rpm; reaction
time 6 h (see Table 3).
a
Factor
Factor value
Averages of k1N × 106 (s−1 mmol N−1
)
the catalyst solubility in the reaction mixture. HBetBr precipitated
from the product, and HBetCl not.
R1
Me
Et
Bu
1
5
9
43
43
30
53
16
48
36
23
57
At this stage of our research, we selected ammonium chlorides
as they are cheaper comparing with bromides or iodides, and we
focused only on the length of alkyl substituents of the ammonium
salt (head group) and the length of a spacer between the CNT func-
tional site and the ammonium nitrogen. As the reported reactivity
of epoxides towards CO2 varies and depends on the reaction system
[9,12,18,24], we included this factor in our research. Each experi-
mental variable was investigated at three levels.
n
Epoxide
Propylene oxide
Styrene oxide
epichlorohydrin
a
k1N, reaction rate per mmol of the ammonium nitrogen present in the reaction
system. Estimated uncertainty of k1N × 106 = 6 (s−1 mmol N−1).
With three factors that could be set at three levels, the total of
a 27-reaction system was possible. Investigations were performed
with various catalysts according to the orthogonal array of experi-
ments presented in Table 3 in order to limit the number of required
runs. Instead of 27, only 9 runs were performed to investigate three
factors: the length of the spacer group (n), the size of alkyl groups
of the quaternary ammonium salt (R1) (Scheme 1) and the epoxide
type.
In all cases, the reaction followed the first-order kinetics at least
for the first 3 h. Within this period, the relationship between the
reaction time and the CO2 consumption was linear on the loga-
rithmic scale as showed in Fig. 4. The exception was a short initial
period, which could be attributed to the formation of a dispersion
of the catalyst in the epoxide. For the same reason, in one case
the reaction rate increased during the 6th hour (curve 8, Fig. 4).
For comparison of the catalyst’s performance, the interval between
3000 and 7000s was used. The corresponding slopes of kinetic
curves are given in Table 3. As the quantity of the ammonium salt
grafted on CNTs was different for each catalyst, we decided to com-
pare not directly the yield of the carbonate, but the rate of the
reaction divided by the amount of the ammonium nitrogen present
during the reaction (k1N). For the same reason, the quantity of the
catalyst was not maintained strictly constant. The results are given
in Table 4.
Each of nine main runs conducted according to the orthogonal
array allowed to establish the main effect of three experimental
variables simply by averaging the reaction rates of the carbonates
obtained for a given setting (Table 3). In other words, based on

S. Baj et al. / Applied Catalysis A: General 488 (2014) 96–102
101
the nine results, nine averages could be calculated and compared
(Table 4).
According to Table 4, the impact of the head group on the reac-
tion rate was small. Between ethyl and methyl groups, there was no
difference, whilst R1 = Bu gave lower yield compared with the other
two. A more important factor was the epoxide type as the variation
between the possible rates was higher. The reactivity of epoxides
was as follows: epichlorohydrin > propylene oxide > styrene oxide.
Compared with other published data, the observed differences
were higher especially between epichlorohydrin and propylene
oxide. Often it was reported that the reactivity of propylene oxide
is higher than epichlorohydrin [12,16,24–31] but other teams
observed the same order of reactivity as presented here [32,33].
The different behaviour of epoxides in various systems is probably
a consequence of different structures of reaction sites. The pres-
ence of an electron-withdrawing chlorine atom in epichlorohydrin
results in reduced electron density of the oxygen atom in epoxide
[34] [35]. This probably disfavours the coordination of epoxide by
carboxylic acid but facilitates the reaction of epoxide with halide
ion as C1 carbon is more positively charged. In our case, the latter
process is likely to be important as we used ammonium chlo-
rides which are often reported to be less reactive comparing with
bromides or iodides. The other factor that may contribute to the
reactivity of epoxides is the size of molecules. In our case, the het-
erogeneity of the catalyst disfavours more bulky substrates such
as styrene oxide which is less reactive comparing with propylene
oxide or epichlorohydrin.
Because in our experiment, only nine runs were performed and
nine averages were calculated, there are no degrees of freedom
left to calculate the significance of results. The precision of k1N
measurements is high (RSD of linear regression < 0.5%). However,
using the data obtained during recycle of a catalyst, we estimated
the potential variations between runs as high as 15%. Taking into
account all results from Table 3, it means that the differences of
averages higher than 6 units of k1N 106 can be treated as significant.
Based on the obtained results, it is clear that in our system
the main factor that influences the yield of carbonate was the
length of the spacer group. The average rates were 16, 48 and 53
×10−6 s−1 mmol N−1 for hexyl, ethyl and dodecyl chains, respec-
tively. Clearly, the middle-sized spacer was the worst performer,
while the shortest was similar to the longest one.
Such observation, counterintuitive at first sight (see Table 1,
entries 3–5), could be rationalized on the basis of the mechanism as
presented in Scheme 2. The mechanism was derived from a propo-
sition given in previous publications [24,27,36].
Initially (Scheme 2, step A), the epoxide molecule is adsorbed
at a Brønsted’s site (COOH) on the surface of CNT [37]. Based on
the elementary analysis, the number of carboxylic acid sites on
an oxidized CNT that was used in this study was approximately
5 mmol g−1. Further functionalization affords the material contain-
ing 1 mmol g−1 of ammonium groups; therefore, a large number
of carboxylic acid sites randomly distributed on the surface are
available for the coordination of the epoxide.
As short (n = 1) and long (n = 9) spacer groups are more effec-
tive in the process compared with the middle-sized one (n = 5),
but only when grafted on the CNT (entries 3–5 in Table 1) sur-
face, it means that the chain length plays an important role during
the rate-determining step. It is likely that the distance between
positively charged nitrogen and carboxylic acid moiety decreases
with the increase in the chain length as longer chains are more
flexible. However, long alkyl chain may inhibit the approach of
epoxide to the CNT surface because of steric hindrance. This effect
is probably negligible for short spacer group and offset by flexi-
bility of the longest one. Comparable effect was reported recently
for amino alcohols potassium iodide catalytic system [38]. Other
authors reported irregular dependence of carbonate yield on the
Scheme 2. Mechanism of the synthesis of cyclic carbonates catalysed by OXCNT-
(CH2)n(R1)3N+Cl−
.
alkyl substituents in the case of imidazolium ionic liquid catalysts
[9,39,40]. However, in that research the distance between the solid
support and reaction centre was not varied.
Further steps (Scheme 2) include nucleophilic attack of chlo-
ride to the epoxide (B) with the subsequent ring opening, addition
of CO2 (C), formation of an alkylcarboante anion stabilized by an
ammonium cation and, finally, intramolecular substitution lead-
ing to abstraction of chloride and regeneration of the catalyst and
cyclic carbonate desorption from the site (D). In comparison with
the chloride attack and coordination of the epoxide, those steps are
not likely to be affected by the structure of the ammonium salt,
which supports our observation of the dominant role of the chain
length.
The mechanism also allowed to explain the observed impact of
the remaining two factors. A small effect of the head group indicates
that the epoxide–nitrogen distance is comparable to two C-C bond
lengths during the initial step. For that reason we did not observe
any difference in activity between catalysts containing Et or Me
groups. In the case of larger Bu substituents, the epoxide approach
to the carboxylic group is probably more difficult because of the
overall steric hindrance around the reaction centre.
The observed order of epoxides reactivity reflects the increas-
ing positive charge on the carbon that is attacked by the chloride
anion and the steric effects of the R2 group (epoxide). Compared
with another publication devoted to CNT-supported catalyst for the
cyclic carbonates synthesis from epoxides, we observed a higher
difference of reactivity between epoxides, especially between
styrene oxide comparing with propylene oxide [17]. It is therefore
likely that the quaternary ammonium salts are more sensitive to
the reagent’s structure compared with imidazolium ionic liquids.
Probably, the presence of CNTs favours only parallel orientation of
aromatic ring-CNT (due to ␲–␲ interaction), which may hamper
the attack of the chloride ion on the epoxide comparing with non-
aromatic epoxides. This effect might be diminished by the presence
of the imidazolium ring. Important role of CNT support in differen-
tiating the reactivity of different epoxides was confirmed by two
experiments analogous to entry 3 of Table 1. If styrene oxide and
epichlorohydrin were used instead of propylene oxide with choline
chloride as a catalyst observed conversions of CO2 were 14 and
16%, respectively (95 and 96% selectivity). Therefore, the observed

102
S. Baj et al. / Applied Catalysis A: General 488 (2014) 96–102
Acknowledgement
This work was supported by the Polish National Science Centre
(NCN); Decision No.: 2011/03/B/ST8/06178. Co-author of this paper
Katarzyna Jasiak is a scholar under the project “DoktoRIS - scholar-
ship program for innovative Silesia” co-financed by European Union
under the European Social Fund.
Appendix A. Supplementary data
Supplementary material related to this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.apcata.
2014.09.034.
References
Fig. 5. Catalytic activity of the reused immobilized catalyst (n = 1 and R1 = Et). Reac-
tion conditions: propylene oxide (0.214 mol); catalyst (≈0.6 g); 110 ◦C; 2 MPa; 6 h;
500 rpm.
[1] T. Sakakura, J.-C. Choi, H. Yasuda, Chem. Rev. 107 (2007) 2365–2387.
[2] Ger. Patent 740366 (1943).
[3] M. North, R. Pasquale, C. Young, Green Chem. 12 (2010) 1514–1539.
[4] A.-A.G. Shaikh, S. Sivaram, Chem. Rev. 96 (1996) 951–976.
[5] D.-W. Kim, R. Roshan, J. Tharun, A. Cherian, D.-W. Park, Korean J. Chem. Eng. 30
(2013) 1973–1984.
[6] W.-L. Dai, B. Jin, S.-L. Luo, S.-F. Yin, X.-B. Luo, C.-T. Au, J. CO2 Util. 3–4 (2013)
7–13.
[7] M.-I. Kim, S.-J. Choi, D.-W. Kim, D.-W. Park, J. Ind. Eng. Chem. 20 (2014)
differences of reactivity were lower comparing with immobilized
ammonium salts.
3.3. Reuse of the catalyst
3102–3107.
[8] S. Ghazali-Esfahani, H. Song, E. Pa˘unescu, F.D. Bobbink, H. Liu, Z. Fei, G.
Laurenczy, M. Bagherzadeh, N. Yan, P.J. Dyson, Green Chem. 15 (2013)
1584–1589.
[9] L. Han, H. Li, S.-J. Choi, M.-S. Park, S.-M. Lee, Y.-J. Kim, D.-W. Park, Appl. Catal.
Gen. 429–430 (2012) 67–72.
[10] J. Sun, S. Zhang, W. Cheng, J. Ren, Tetrahedron Lett. 49 (2008) 3588–3591.
[11] K.R. Roshan, G. Mathai, J. Kim, J. Tharun, G.-A. Park, D.-W. Park, Green Chem. 14
(2012) 2933–2940.
[12] J. Sun, J. Wang, W. Cheng, J. Zhang, X. Li, S. Zhang, Y. She, Green Chem. 14 (2012)
654–660.
[13] T. Takahashi, T. Watahiki, S. Kitazume, H. Yasuda, T. Sakakura, Chem. Commun.
(2006) 1664–1666.
[14] J.-Q. Wang, D.-L. Kong, J.-Y. Chen, F. Cai, L.-N. He, J. Mol. Catal. Chem. 249 (2006)
143–148.
[15] R.A. Watile, K.M. Deshmukh, K.P. Dhake, B.M. Bhanage, Catal. Sci. Technol. 2
(2012) 1051–1055.
[16] M.M. Dharman, J.-I. Yu, J.-Y. Ahn, D.-W. Park, Green Chem. 11 (2009)
1754–1757.
[17] H.-J. Choi, M. Selvaraj, D.-W. Park, Chem. Eng. Sci. 100 (2013) 242–248.
[18] J. Sun, J. Ren, S. Zhang, W. Cheng, Tetrahedron Lett. 50 (2009) 423–426.
[19] H.-Y. Ju, M.-D. Manju, K.-H. Kim, S.-W. Park, D.-W. Park, J. Ind. Eng. Chem. 14
(2008) 157–160.
[20] K. Kossev, N. Koseva, K. Troev, J. Mol. Catal. Chem. 194 (2003) 29–37.
[21] Y. Zhao, J.-S. Tian, X.-H. Qi, Z.-N. Han, Y.-Y. Zhuang, L.-N. He, J. Mol. Catal. Chem.
271 (2007) 284–289.
[22] J.-J. Shim, D. Kim, S.R. Choon, Bull. Korean Chem. Soc. 27 (2006)
744–746.
[23] Y. Zhou, S. Hu, X. Ma, S. Liang, T. Jiang, B. Han, J. Mol. Catal. Chem. 284 (2008)
52–57.
A series of cycles of catalyst’s reuse was performed with the cat-
alyst of n = 1 and R1 = Et using propylene oxide as a model epoxide.
After correcting the results for slight losses of the catalyst dur-
ing its separation from the reaction mixture, it was clear that the
catalyst retains its activity (Fig. 5). Nevertheless, it was crucial to
carefully dry the sample between the runs. Typically the catalyst
was azeotropically dried for 1 h with cyclohexane. After evapora-
tion of the solvent, the solid was dried over NaOH under vacuum
(50 mbar). In the case of the third cycle, the drying period was
only 15 min, which appeared to be not enough to remove water.
If no drying was performed and the catalyst was used immedi-
ately after filtering only, the yield of epoxide was approximately
halved. The observation was surprising as water was demonstrated
as a possible solvent for the cyclic carbonate synthesis [18]. In our
system, water probably competes with the epoxide for Brønsted’s
sites on the nanotube’s surface (Scheme 2, step A) and decreases
the likelihood of the catalyst–epoxide contact. In order to con-
firm our assumption, we deliberately added water to the reaction
system using 0.35 g of OXCNT-(CH2)10(CH3)3N + Cl− and propylene
oxide (0.214 mol) as a model system. With the addition of 0.018 g of
water, the carbonate yield was 12% (97% selectivity) after 6 h while
0.18 g of water in the system diminished the yield to 3% only (88%
selectivity).
[24] D. Bai, H. Jing, G. Wang, Appl. Organomet. Chem. 26 (2012) 600–603.
[25] L. Han, H.-J. Choi, S.-J. Choi, B. Liu, D.-W. Park, Green Chem. 13 (2011)
1023–1028.
[26] T. Chang, H. Jing, L. Jin, W. Qiu, J. Mol. Catal. Chem. 264 (2007) 241–247.
[27] T. Cao, L. Sun, Y. Shi, L. Hua, R. Zhang, L. Guo, W. Zhu, Z. Hou, Chin. J. Catal. 33
(2012) 416–424.
4. Conclusions
[28] D. Anselmo, V. Bocokic´, A. Decortes, E.C. Escudero-Adán, J. Benet-Buchholz,
J.N.H. Reek, A.W. Kleij, Polyhedron 32 (2012) 49–53.
[29] J. Tharun, D.W. Kim, R. Roshan, Y. Hwang, D.-W. Park, Catal. Commun. 31 (2013)
62–65.
[30] Y. Du, J.-Q. Wang, J.-Y. Chen, F. Cai, J.-S. Tian, D.-L. Kong, L.-N. He, Tetrahedron
Lett. 47 (2006) 1271–1275.
[31] J.-Q. Wang, X.-D. Yue, F. Cai, L.-N. He, Catal. Commun. 8 (2007) 167–172.
[32] Y.-Y. Zhang, L. Chen, S.-F. Yin, S.-L. Luo, C.-T. Au, Catal. Lett. 142 (2012)
1376–1381.
[33] Y. Xiong, F. Bai, Z. Cui, N. Guo, R. Wang, J. Chem. 2013 (2013) 1–9.
[34] D. Bai, G. Nian, G. Wang, Z. Wang, Appl. Organomet. Chem. 27 (2013) 184–187.
[35] T. Sakai, Y. Tsutsumi, T. Ema, Green Chem. 10 (2008) 337–341.
[36] F. Li, L. Xiao, C. Xia, B. Hu, Tetrahedron Lett. 45 (2004) 8307–8310.
[37] L.-F. Xiao, D.-W. Lv, D. Su, W. Wu, H.-F. Li, J. Clean. Prod. 67 (2014)
285–290.
In summary, quaternary ammonium chlorides covalently bound
to CNTs are proved to be efficient and recyclable heterogeneous
catalysts for the solvent-free synthesis of cyclic carbonates under
mild conditions. The nanotubes are not only a solid support that
facilitates the reuse of the catalyst, but carboxylic moieties present
on CNT surface also participate in the reaction presumably by
coordination of an epoxide. The synergistic effect is clearly visi-
ble comparing the results for free, immobilized ammonium salts
and for carboxylic acid-catalysed systems. The activity of the cata-
lyst is fully retained upon reuse; however, it should be thoroughly
dried between the runs. Strong impact of alkyl chain that links
the surface of carbon nanotube with ammonium nitrogen indi-
cates that steric effects play an important role in the activity of the
catalyst.
[38] T. Werner, N. Tenhumberg, J. CO2 Util. 7 (2014) 39–45.
[39] X. Zheng, S. Luo, L. Zhang, J.-P. Cheng, Green Chem. 11 (2009) 455–458.
[40] H. Kawanami, A. Sasaki, K. Matsui, Y. Ikushima, Chem. Commun. 9 (2003)
896–897.