Cycloaddition of CO2 to epoxides over solid base catalysts

Article abstract of DOI:10.1006/jcat.2000.3145

The cycloaddition of CO₂ to various epoxides, i.e., ethylene oxide, epoxybutene, and epoxypropylbenzene over solid base catalysts (KX zeolite, Cs-loaded KX zeolite, Cs-doped alumina, and MgO) was performed in a batch autoclave reactor at 423 K and with excess CO₂. The occluded base sites on the Cs/KX were actually composed of occluded Cs and K species that could be removed by washing with water. The activity of the zeolite catalysts for ethylene oxide conversion to ethylene carbonate depended on the basicity of the sample, with the sample containing occluded alkali metal oxides being the most active. In addition, the site-time yields of ethylene carbonate formation, based on CO₂ adsorption capacity, over Cs/KX, Cs/Al₂O₃, and MgO were similar to each other and were within a factor of 4 of the site-time yield seen with the homogeneous catalyst [N(C₂H₅)₄Br. The rates of epoxypropylebenzene conversion over the solid base catalysts were much lower than the rates of ethylene oxide conversion, presumably due to steric hindrance of the bulky side group on the former. CO₂ addition to ethylene oxide was effectively catalyzed by Cs/KX, MgO, and Cs/Al₂O₃. Porosity and Lewis acidity influenced the reactivity of epoxybutene and epoxypropylbenzene more than that of ethylene oxide. Zeolites provided a unique reaction environment allowing water to play a beneficial part in base catalysis involving CO₂ cycloaddition reactions.

Full text of DOI:10.1006/jcat.2000.3145

Journal of Catalysis 199, 85–91 (2001)
doi:10.1006/jcat.2000.3145, available online at http://www.idealibrary.com on
Cycloaddition of CO to Epoxides over Solid Base Catalysts
2
1
Mai Tu and Robert J. Davis
Department of Chemical Engineering, University of Virginia, 102 Engineers Way, Charlottesville, Virginia 22904-4741
Received August 30, 2000; revised December 6, 2000; accepted December 6, 2000; published online February 26, 2001
been explored as possible catalysts for the reaction (4–6).
The cycloaddition of carbon dioxide to various epoxides (ethy- Interestingly, stereospecific addition of CO2 has been re-
lene oxide, epoxybutene, and epoxypropylbenzene) over solid base ported over the Mg-based solid oxides (4, 5).
catalysts (KX zeolite, Cs-loaded KX zeolite, Cs-doped alumina, and
MgO) was performed in a batch autoclave reactor at 423 K and
with excess CO2. Catalysts were characterized by elemental anal-
ysis, N2 physisorption, and CO2 adsorption microcalorimetry. The
combination of results from elemental analysis and microcalorime-
alkali metal cations (8–11). Stronger base sites can be incor-
try showed that the occluded base sites on the Cs/KX were actually
composed of occluded Cs and K species that could be removed by
washing with water. The activity of the zeolite catalysts for ethylene
oxide conversion to ethylene carbonate depended on the basicity of
Basic zeolites have attracted attention recently primar-
ily due to their environmentally benign character and their
potential use in fine chemical synthesis (7–9). The basicity
of zeolites can be developed by ion-exchange with heavy
porated in the zeolite pores by decomposing an occluded al-
kali metal salt such as cesium acetate (8–11). Basic zeolites
have been tested in various reactions such as toluene alky-
the sample, with the sample containing occluded alkali metal ox- lation with methanol, oxidative methylation of toluene with
ides being the most active. In addition, the site-time yields of ethy- methane, dehydrogenation of alchols, double bond isomer-
lene carbonate formation, based on CO2 adsorption capacity, over ization, and aldolcondensation (7–19). Despite these exten-
Cs/KX, Cs/Al2O3, and MgO were similar to each other and were
within a factor of 4 of the site-time yield seen with the homogeneous
catalyst [N(C2H5)4]Br. The rates of epoxypropylbenzene conversion
over the solid base catalysts were much lower than the rates of ethy-
lene oxide conversion, presumably due to steric hindrance of the
oxide with CO2 to make ethylene carbonate (6). The activ-
bulky side group on the former. Rate measurements with a mixture
of ethylene oxide and epoxypropylbenzene over Cs/KX and MgO
indicated that most of the basic sites on Cs/KX are located in the
zeolite micropores. A high catalytic activity of Cs/Al2O3 for all of
sive studies, the nature of occluded bases in zeolite pores
and the structure of the active sites are unknown (7, 9).
In a previous paper, we reported that alkali-modified ze-
olites were catalytically active in the reaction of ethylene
ity increased with the electropositivity of the countercation
and with incorporation of extraframework cesium oxide.
In the present work, we investigate the reactions of CO2
the epoxides studied suggests that proximity of surface Lewis acid with ethylene oxide, epoxybutene, and epoxypropylben-
sites to surface base sites is needed for the cycloaddition reaction. zene over a series of base catalysts, including X zeolites,
In the case of a zeolite catalyst, addition of a small amount of wa- Cs-loaded alumina, and magnesia. For comparison, pure
ter enhanced the rate ethylene oxide reaction without significant
Al2O3 and tetraethylammonium bromide were also stud-
ied. These reactions are illustrated in Fig. 1.
formation of glycol side products.
�c 2001 Academic Press
Key Words: zeolite, basic; expoxide, reaction of; carbon dioxide,
addition of; magnesia; alumina, Cs-doped; catalysis, base.
EXPERIMENTAL METHODS
INTRODUCTION
Potassium-exchanged zeolite X (KX) was prepared by
triply exchanging NaX (UOP, Lot No. 07483-36) with 1 M
aqueous solutions of KNO3 (Alfa, 99.0% ). The NaX was
slurried in the aqueous solution for 2 h during each ex-
change and then vacuum filtered. The exchanged sample
was then washed with distilled water until the filtrate was
neutral. The resultingmaterialwasdried overnight and then
Cyclic carbonates synthesized by the addition of CO2 to
epoxides are useful as highly polar solvents and reagents
in polymer synthesis (1–6). Many organic and inorganic
compounds, including onium halides, phosphines, amines,
and metal halides, catalyze the reaction of CO2 with epox-
ides under mild conditions to give high yields of cyclic car-
bonate (1–3). Solid bases such as MgO, calcined hydro-
talcite (Mg–Al mixed oxide), and basic zeolites have also
calcined in flowing air at 773 K for 5 h. The heating rate was
� 1
4
K min from room temperature to 773 K except that the
sample was held at 423 K for 1 h during the ramping period.
Cesium was loaded into the KX by incipient wetness im-
pregnation of aqueous solution of cesium acetate (Aldrich,
1
To whom correspondence should be addressed, rjd4f@virginia.edu.
85
0021-9517/01 $35.00
Copyright �c 2001 by Academic Press
All rights of reproduction in any form reserved.

8
6
TU AND DAVIS
Reactions of epoxides with CO2 were carried out at
23 K in a 50-ml high-pressure, stainless steel, stirred au-
4
toclave from Parr Instrument Co. To give approximately
the same base site loading in each run, about 50 to 150 mg
of catalyst was loaded into the reactor prior to introduc-
tion of about 2 to 2.5 g of epoxide (ethylene oxide, Messer,
9
2
9.9% ; epoxybutene, donated by Eastman Chemical Co.;
, 3-epoxypropylbenzene, Aldrich, 98% ). The reactor was
then pressurized with about 550 psig CO2 (BOC Gases,
9% ) at ambient temperature. Prior to reaction, the catalyst
9
was dried in flowing He at 623 K for 2 h and then immedi-
ately transferred to the reactor. After a reaction time of 3 h,
products were extracted with acetone. A weighed amount
of dodecane (Aldrich, 99+% ) was added as an external
standard to the product mixture, which was analyzed on a
Hewlett Packard 6890 GC/MS with a 50-m HP-1 column.
In a few experiments, catalysts were loaded with water by
exposing them to air at 100% relative humidity at ambient
temperature and pressure.
FIG. 1. Cycloaddition of carbon dioxide to various epoxides used in
this work.
RESULTS
Results from elemental analysis, N2 physisorption, and
CO2 chemisorption are summarized in Table 1. Each unit
cell of the Cs/KX(I) contains 80 potassium cations and
9
9.99% ) (10). After drying overnight, the sample was
calcined in flowing air at 773 K for 5 h with the same
temperature program as for the KX. The resulting sam-
ple is denoted as Cs/KX(I). Another cesium-loaded zeo-
lite X sample was prepared by identical procedures and is
denoted as Cs/KX(II). Some of the Cs/KX(I) sample was
added to distilled water, slurried for 2 h, and then filtered,
dried, and calcined. All the conditions were identical to
those described above for the KX. This washed sample is
referred to as Cs/KX(W). Elemental analysis of the zeolite
samples was performed by Southern Testing, NC. In addi-
tion, an alumina-supported cesium oxide (Cs/Al2O3) was
prepared by incipient wetness impregnation following the
procedures used for the preparation of the Cs/KX samples.
The � -alumina was provided by Mager Scientific Inc. and
the magnesia (Grade 500A) was donated by Ube.
2
0 extra cesium cations, which translates to a loading of
about 2.5 Cs cations per supercage. Elemental analysis of
Cs/KX(W) revealed that only 65 potassium cations and
1
5 cesium cations remained in the unit cell. Washing the
Cs-loaded zeolite removed the extraframework alkalimetal
cations, since the total number of Cs and K cations was the
same as the Al content in the unit cell. However, only 25%
of the cesium atoms that were loaded into the sample were
eliminated. Evidently, 75% of the added cesium exchanged
with potassium ions already in the zeolite, thus allowing
some of the potassium to be washed out of the sample. The
high surface areas and micropore volumes of all zeolite
The BET surface areas and micropore volumes of the
samples were evaluated from N2 adsorption isotherms at
TABLE 1
Characterization of Zeolites, Cs/Al2O3, and MgO
7
7 K measured on a Coulter Omnisorp 100CX instrument.
Prior to measurement, a sample was outgassed overnight
at 573 K. The micropore volume of the zeolite samples is
reported as the liquid volume of N2 adsorbed at a relative
pressure of 0.3. The micropore volume of the Cs-loaded
alumina and magnesia was evaluated by a t-plot analysis of
the adsorption data. The magnesia sample was found to be
nonmicroporous.
The basicity of the samples was measured by CO2 ad- Cs/KX(W) Cs K Al Si112O384
sorption microcalorimetry (20, 21). The details of the mi- Cs/Al2O3
crocalorimeter used in this work are described in a previ-
BET
surface
area
CO2
Micropore adsorption
a
volume
capaciity
2
� 1
� 1
� 1
(�mol g )
Catalyst
KX
Cs/KX(I)
Cs/KX(II)
Composition
(m g
)
(ml g
)
—
Cs20K80Al80Si112O384
—
467
313
346
366
69
0.234
0.154
0.153
0.181
0.008
—
80
87
145
45
123
45
1
5
65
80
b
13.1 wt% Cs
MgO
—
42
ous paper (22). About 1.5 g of sample was loaded into the
microcalorimeter and evacuated at 773 K for 5 h before
cooling to 373 K. The adsorption was performed at 373 K.
a
Based on results from microcalorimetry; only sites stronger than
�
1
6
0 kJ mol were counted.
b
Based on t-plot analysis.

CYCLOADDITION OF CO2TO EPOXIDES
87
FIG. 2. Differential heats of adsorption as a function of CO2 uptake
FIG. 3. Differential heats of adsorption as a function of CO2 uptake
on basic zeolites.
on Cs/Al2O3 and MgO compared to Cs/KX.
samples suggest that incorporation of cesium did not ex-
tensively damage the crystalline framework. The Cs/Al2O3
and MgO catalysts had relatively low surface areas and no
micropores detected by t-plot analysis.
experiments with dry catalysts, monocyclic carbonate was
the only product detected by GC/MS. The catalytic activity
is reported in this work as a site-time yield, which was calcu-
lated from the molar product yield after 3 h and the CO2 ad-
sorption capacity measured by microcalorimetry (Table 1).
The CO2 adsorption capacity was determined somewhat
arbitrarily as the number of adsorption sites with 1Hads �
Figure 2 shows the differential heat of CO2 adsorp-
tion as a function of the loading on the zeolite samples.
Carbon dioxide adsorbed on KX with a 1Hads of about
�
1
� 1
�
1
7
0 kJ mol up to a loading of about 80 �mol g . Addi-
6
0 kJ mol . The reaction temperature and time were cho-
tion of cesium to the zeolites increased both the heat of
sen to give reasonable yields of product.
�
1
adsorption (to greater than 100 kJ mol ) and the maxi-
mum loading of CO2. The stronger basicity of the Cs/KX
can be attributed to the occluded alkali metal oxide species,
which may be either cesium or potassium oxide. The val-
ues of 1Hads on Cs/KX samples were similar to those re-
ported previously for Cs loaded into CsX using analogous
synthesis procedures (17). The CO2 adsorption sites on the
water-washed, cesium-containing zeolite, Cs/KX(W), were
Table 2 compares the reactivity results for synthesis
of 1,3-dioxolan-2-one (also called ethylene carbonate)
from ethylene oxide over basic zeolites. The site-time
yield of ethylene carbonate increased in the order of
KX < Cs/KX(W) < Cs/KX, consistent with the order of
the basicity of the samples determined by microcalori-
metry. This result suggests that the catalyst basicity plays
an important role in the addition reaction.
�
1
1
0 kJ mol stronger than those on KX, probably due to
The results from CO2 addition to ethylene oxide (EO),
epoxypropylbenzene (EPBz), and epoxybutene (EpB) on
various catalysts are reported in Table 3. The Cs/KX,
Cs/Al2O3, and MgO catalysts were active in the synthe-
sis of ethylene carbonate (D) with site-time yields ranging
the Cs cations that remained in the washed sample. The
�
1
adsorption sites with 1Hads greater than 90 kJ mol were
removed by washing a cesium-loaded zeolite, which con-
firmed that the strongest base sites were associated with
the occluded alkali metal oxides.
�
1
from 208 to 325 h . Although the total base site density
The basicity of the MgO and Cs/Al2O3 samples was also
determined by CO2 adsorption microcalorimetry and re-
sults are summarized in Fig. 3. The MgO had the strongest
�
1
TABLE 2
adsorption sites (160–170 kJ mol ), but the total uptake of
�
1
CO2 was less than 50 �mol g due to its low surface area.
The adsorption sites on Cs/Al2O3 were energetically simi-
lar to those on MgO, but the total adsorption capacity of
Catalytic Activity of Basic Zeolites in the Addition
a
of CO to Ethylene Oxide
2
�
1
1Hads of CO
Ethylene carbonate
yield (% )
Site-time yield
2
about 150 �mol g was much higher. Results from mi-
crocalorimetry revealed a significant difference between
adsorption sites of Cs/Al2O3 and cesium-loaded zeolites
�
1
b
� 1
Catalyst
KX
Cs/KX(I)
Cs/KX(W)
(kJ mol
)
(h
)
70
110
80
1.8
14
27
250
70
(
Cs/KX and Cs/CsX (17)), indicating a significant support
3.0
effect on the nature of cesium oxides.
The reaction of epoxide with CO2 prefers cyclization
rather than polymerization due to the thermodynamic sta-
bility of five-membered cyclic carbonate (1). In all of our
a
Reaction conditions: 423 K, initial CO2 pressure 550 psig at 293 K,
reaction time = 3 h.
b
Based on microcalorimetry.

8
8
TU AND DAVIS
TABLE 3
alumina. It should be noted that ethylene oxide was the
Addition of CO2 to Epoxides over Solid Base Catalystsa
most reactive epoxide, regardless of the catalyst used.
The low reactivity of the bulky epoxypropylbenzene over
Cs/KX(I) compared to that of ethylene oxide suggests that
diffusional limitations inside the zeolite micropores may
be limiting the mobility of the large reagent and/or prod-
uct near the basic sites. To test this hypothesis, the CO2
cycloaddition reaction was carried out with a mixture of
ethylene oxide and epoxypropylbenzene. The results are
shown in Table 4 for epoxide mixtures converted over
Cs/KX(I) and MgO. Over the nonmicroporous MgO cata-
lyst, the site-time yield of the bulky carbonate (BzD) was
about the same as that observed with only epoxypropyl-
benzene (Table 3) and the STY of ethylene carbonate in
the reaction mixture was slightly depressed by a factor of
Carbonate
yield (% )
Site-time yield
Catalyst
N(C2H5)4]Br
Reactantb
Productb
� 1
(h )
[
EO
EPBz
EpB
D
BzD
ED
74
41
87
812
239
345
Cs/KX(I)
CS/KX(II)
CS/Al2O3
EO
EPBz
D
BzD
14
3.1
250
15
EO
EpB
D
ED
19
0.3
208
3
EO
EPBz
EpB
D
BzD
ED
24
20
28
325
56
231
Al2O3
MgO
EpB
ED
0
0
1
.6 over the pure component run (Table 3). Over the mi-
EO
EPBz
EpB
D
BzD
ED
19
4.0
0.9
309
26
10
croporous Cs/KX(I), the rate of bulky carbonate formation
was unaffected by the presence of ethylene oxide whereas
the rate of ethylene carbonate formation was decreased by
a factor of 8 by the presence of bulky epoxypropylbenzene.
These reaction mixture experiments demonstrated that the
bulky epoxide or its reaction product blocks access of ethy-
lene oxide to the interior of the zeolite where most of the
catalytically active basic sites are located.
a
Reaction conditions: 423 K, initial CO2 pressure 550 psig at 293 K,
reaction time = 3 h.
b
EO, ethylene oxide; D, 1,3-dioxolan-2-one (ethylene carbonate);
EPBz, epoxypropylbenzene; BzD, 4-benzyl-1,3-dioxolan-2-one; EpB,
epoxybutene; ED, 4-ethenyl-1,3-dioxolan-2-one.
Since ethylene carbonate can also catalytically decom-
of Cs/KX(I) and Cs/KX(II) was different (see Table 1), pose to form ethylene oxide and CO2 (23), it is possible
the site-time yield for ethylene carbonate synthesis varied that ethylene carbonate formed in a mixed epoxide reac-
by only 17% , demonstrating the reproducibility of the re- tion (Table 4) participated in the reaction to form the bulky
action protocol. The homogeneous catalyst, [N(C2H5)4]Br, carbonate of epoxypropylbenzene. Therefore, we decom-
was about 2.5 to 4 times more active than all the solid base posed ethylene carbonate (Aldrich, 98% ) in the presence
catalysts for ethylene carbonate production. The reactivity of epoxypropylbenzene and 550 psig N2. The results in
of the larger epoxides was more complicated than that of Table 5 clearly show that decomposition of ethylene car-
ethylene oxide. For example, the cycloaddition of CO2 to bonate on the solid bases provided an alternative source of
bulky epoxypropylbenzene was very slow over Cs/KX(I) CO2 for the epoxypropylbenzene. The site-time yields of
�
1
with a site-time yield 16 times less than that measured with BzD were 30 and 13 h on the MgO and the Cs/KX(I),
the homogeneous catalyst, presumably due to the steric re- respectively, and were similar to those obtained with CO2
strictions around the basic sites located in the zeolite mi- as a carbonate source. For comparison, the site-time yield
�
1
cropores. The site-time yields for epoxypropylbenzene re- of BzD was only 4.3 h with the homogeneous catalyst.
�
1
action over nonmicroporous Cs/Al2O3 (56 h ) and MgO
In a previous study, we found that exposing the cesium-
26 h ) were greater than that over Cs/KX(I) (15 h ). containing zeolite X to a humid atmosphere before reaction
�
1
� 1
(
Epoxybutene (EpB), which is larger in size than ethylene
oxide but smaller than epoxypropylbenzene, wasalso tested
in the cycloaddition reaction. As seen in Table 3, the site-
TABLE 4
Addition of CO2 to Epoxide Mixtures over Cs/KX and MgOa
�
1
time yield of the EpB reaction over Cs/KX(II) (3 h ) was
more than 100 times lower than that found with the ho-
Carbonate
yield (% )
Site-time yield
� 1
�
1
Catalyst
Reactant b
Product b
(h
)
mogeneous catalyst (345 h ). The olefinic functionality
apparently plays a significant role in reactivity of EpB on
solid bases. Interestingly, the site-time yield of the EpB re-
Cs/KX(I)
EO and EPBz
D
BzD
3.5
5.6
31
14
�
1
action on Cs/Al2O3 (231 h ) was within a factor of 2 of
MgO
EO and EPBz
D
BzD
16
12
189
29
�
1
the homogeneous catalyst (345 h ) and 23 times greater
�
1
than that on MgO (10 h ). The surprisingly high activity
of the Cs/Al2O3 catalyst led us to check the reactivity of
the support alumina without added cesium. As illustrated
a
Reaction conditions: 423 K, initial CO2 pressure 550 psig at 293 K,
reaction time = 3 h.
b
EO, ethylene oxide; D, 1,3-dioxolan-2-one (ethylene carbonate);
in Table 3, no reaction of EpB was observed over pure EPBz, epoxypropylbenzene; BzD, 4-benzyl-1,3-dioxolan-2-one.

CYCLOADDITION OF CO2TO EPOXIDES
89
TABLE 5
sium cations prior to impregnation beyond ion-exchange
capacity.
Reaction of Ethylene Carbonate and Epoxypropyl Benzene
a
The reactivity results indicate that basicity and porosity
affect the rate of cycloaddition of carbon dioxide to epox-
ides. For example, the rate of ethylene carbonate forma-
tion over basic zeolites increased with increasing basicity
as evaluated by carbon dioxide adsorption microcalorime-
try. Clearly, the basic strength of a catalyst cannot be too
great or carbon dioxide will be adsorbed too strongly to
react. The catalysts used in this study are of appropriate
base strength that allows for both CO2 adsorption and re-
action.
on Various Catalysts
BzD yield b
(% )
Site-time yield
�
1
Catalyst
(h )
MgO
Cs/KX(I)
N(C2H5)4]Br
9.3
5.3
9.0
30
13
4.3
[
a
Reaction conditions: 423 K, initial CO2 pressure 550 psig at
2
93 K, reaction time = 3 h. Amount of ethylene carbonate was � 1.8
g for runs with solid catalysts and 2.9 g for the run with [N(C2H5)4]Br.
The standard amount of EPBz (� 2 g) was used.
The porosity of the catalyst affected the conversion of
the sterically hindered epoxypropylbenzene. Rates of reac-
tions with both the pure component epoxides (Table 3) and
b
BzD, 4-benzyl-1,3-dioxolan-2-one.
promoted the production of ethylene carbonate from ethy- mixed epoxides (Table 4) are consistent with the majority
lene oxide (6). We have performed similar experimentswith of the active sites on the zeolites being located in the mi-
water-containing Cs/KX(I) and MgO and the results are re- cropores instead of on the external surface. For the mixed
ported in Table 6. Although the base site density of the wet epoxide experiments (Table 4) both epoxypropylbenzene
catalysts has not been measured, an influence of water on and its cyclic carbonate could retard the diffusion of ethy-
the reaction rate can be seen in the site-time yield based on lene oxide into the zeolite pores, leading to the severely
the dry catalyst base site density and the product selectiv- reduced production of ethylene carbonate. However, the
ity. The site-time yield of ethylene carbonate increased to bulky reactant and product negligibly affected the rate of
�
1
� 1
7
05 h over the wet Cs/KX(I) compared to 250 h under ethylene carbonate production over nonmicroporous MgO.
dry conditions. A trace amount of ethylene glycol was the These results provide clear evidence that most of the basic
only side product detected. The site-time yield of ethylene sites of zeolite catalysts containing occluded alkali metal
�
1
carbonate on wet MgO was reduced to 56 h compared to oxide species are within the micropore network and sug-
3
�
1
09 h on dry MgO because large amounts of mono-, di-, gest their use as shape-selective base catalysts.
and triethylene glycols were formed.
Shape-selective catalysis is important for synthesis of fine
chemicals and for processing of petroleum fractions. Most
of the applications of shape selectivity are based on re-
actions catalyzed by acidic zeolites. Recently, researchers
have started to explore the potential shape selectivity of
basic zeolites (8). Rodriguez et al. studied the reaction of
benzaldehyde with ethyl cyanoacetate over zeolites X and
Y containing cesium oxide prepared by decomposition of
occluded cesium acetate (24). Theyfound that Knoevenagel
condensation proceeded, but the product did not undergo
the subsequent Michael addition. The lack of further addi-
tion reactions may be due to inadequate base strength of the
catalyst or due to the limited space inside the zeolite cavi-
ties. Corma and Martin-Aranda also studied Knoevenagel
condensation reactions on basic zeolites and suggested that
the lack of Michael addition products is due to geometrical
constraints in the zeolite (12). In a related work, Ballini et al.
performed nitroaldolic condensation of aromatic aldehydes
with nitroalkanes over basic zeolites (19). The lower lev-
els of Michael addition products with larger reactants were
again attributed to steric restrictions in the zeolite pores. In
our work, we found a “reverse” reactant shape selectivity of
the Cs/KX catalyst since cofeeding bulky epoxypropylben-
zene with relatively small ethylene oxide almost totally in-
hibited the reaction ofthe smallmolecule. Thisismost likely
a consequence of the relatively low reaction temperatures
DISCUSSION
The results from elemental analysis and microcalorime-
try provided insights into the nature of the occluded species
in the basic zeolites. The basic species formed by decompo-
sition of occluded cesium acetate was easily removed by
washing in water, presumably through a hydrolysis pro-
cess. In addition, most of the cesium that was impregnated
as extraframework cations exchanged with potassium ions
already present in the zeolite. To obtain occluded cesium
species, a basic zeolite should be ion-exchanged with ce-
TABLE 6
Effect of Trace Water on the Addition of CO2 to Ethylene Oxidea
Ethylene carbonate Site-time yield
Catalystb
yield (% )
(h
� 1
)
By-productc
Cs/KX(I)
MgO
54
3.7
705
56
Trace amount EG
EG, DEG, TEG
a
Reaction conditions: 423 K, initial CO2 pressure 550 psig at 293 K,
reaction time = 3 h.
b
Exposed to 100% relative humidity prior to reaction.
Detected by HP 6890 GC/MS. EG, ethylene glycol; DEG, di-ethylene
c
glycol; TEG, triethylene glycol.

9
0
TU AND DAVIS
that do not favor desorption of the bulky reactants and
products from the zeolite micropores, thus blocking access
to the basic sites.
The influence of CO2 availability at the active site is illus-
trated in Table 5. The decomposition of ethylene carbonate
was used to provide carbon dioxide for epoxypropylben-
zene. The site-time yields over MgO and Cs/KX(I) were
greater than the homogeneous catalyst and were similar to
the reactions in the presence of carbon dioxide. Evidently,
the virtual pressure of carbon dioxide on the solid bases is
quite high since reasonable site-time yields were observed.
The CO2 may even remain adsorbed on the solid bases until
it is reacted away. For the case involving the homogeneous
catalyst, a low site-time yield was probably due to the very
low partial pressure of CO2 in the reaction vessel due to
ethylene carbonate decomposition.
The reaction of epoxybutene requires special discussion
since it contains an additional olefinic functionality. It is un-
SCHEME 1. Postulated mechanism for cycloaddition of CO2 to an
epoxide on a basic metal oxide surface. M represents a metal cation (Lewis
known at this time why it is unreactive for cycloaddition of acid) and O represents a surface oxygen (Lewis base).
CO2 in the basic zeolites. This result requires further study.
Another interesting observation is that EpB was rapidly
acid cations becomes more important as the size of the re-
converted over Cs/Al2O3 but not over MgO. Since the heat
of CO2 adsorption on MgO is about the same as that on
Cs/Al2O3, the high catalytic activity of Cs/Al2O3 for EpB
conversion indicates that a factor other than basicity should
be considered.
acting epoxide increases. Indeed, Cs/Al2O3 was more active
than MgO for both epoxybutene and epoxypropyl benzene.
However, the large difference between the two catalysts for
EpB conversion is also likely due to the presence of the dou-
ble bond. Further investigations are needed to clarify this
point.
The beneficial effect of water observed for the Cs/KX
catalyst may now be understood in light of the mecha-
nism described in Scheme 1. Introduction of water into the
zeolite pores generated basic hydroxyl groups that are
highly mobile compared to the occluded alkali metal
species. Thus, the reaction proceeds more like a homoge-
neously catalyzed process in which the acid–base sites are
not constrained to a surface. Interestingly, the low level of
water in the pores and the limited space available for re-
action prevented the formation of glycol byproducts that
were seen on wet magnesia catalysts.
It has been reported that the calcined hydrotalcite (Mg–
Al mixed oxide) was more active than the MgO in the re-
action of CO2 with epoxides because of the coexistence
of Lewis acid sites on the former (5). These acid sites are
believed to stabilize the adsorbed epoxide prior to addi-
tion of carbon dioxide. Therefore, the proximity of Lewis
acid sites to basic sites capable of adsorbing carbon diox-
ide may be required to catalyze the cycloaddition reaction.
A possible reaction mechanism for the addition of CO2 to
epoxides over solid base catalysts, shown in Scheme 1, has
been suggested by Yamaguchi et al. (5). Carbon dioxide is
adsorbed on Lewis basic sites (i.e., O2 ) to form a surface
carbonate species. The strength and number of the surface
base sites is likely to be very important for the activation of
CO2. The epoxide is adsorbed on a neighboring Lewis acid
site, most likely an exposed cation. The carbonate surface
anion reacts with the less sterically hindered carbon atom
�
CONCLUSIONS
The cycloaddition of CO to various epoxides on a se-
2
of the adsorbed oxirane to generate the oxyanion, which ries of solid bases (KX, Cs/KX, Cs/Al O , and MgO) was
2
3
subsequently yields the cyclic carbonate product.
influenced by base strength, porosity, and Lewis acidity
For the basic zeolites, a large number of exchangeable of the catalysts. Decomposition of impregnated cesium
cations in the zeolite provide adequate coordination sites acetate precursor on zeolite KX produced a zeolite with
for the reactingepoxides. The case islikelyto be different on basic sites inside the zeolite micropores. The occluded moi-
the MgO and Cs/Al2O3 catalysts. Magnesia is a classic solid eties were composed of both Cs and K oxides due to ion-
base with rather poor Lewis acidity whereas Cs/Al2O3 likely exchange during the synthesis procedure. The heat of car-
exposes Lewis acidic aluminum cations in the vicinity of the bon dioxide adsorption on the Cs/KX zeolites was about
�
1
basic cesium “oxide” sites. The Cs/Al2O3 catalyst probably 50 kJ mol lower than that on either MgO or Cs/Al O .
2
3
has an acid–base character analogous to that of the Mg–Al Nevertheless, CO addition to ethylene oxide was effec-
2
mixed oxides mentioned above. The accessibility of Lewis tively catalyzed by Cs/KX, MgO, and Cs/Al O . Porosity
2
3

CYCLOADDITION OF CO2TO EPOXIDES
91
and Lewis acidity appeared to influence the reactivity of
epoxybutene (an epoxide containing a carbon–carbon dou-
ble bond) and epoxypropylbenzene (a bulky epoxide) more
than that of ethylene oxide. Finally, zeolites appear to pro-
vide a unique reaction environment that allows water to
6. Doskocil, E. J., Bordawekar, S. V., Kaye, B. G., and Davis, R. J., J. Phys.
Chem. B 103, 6277 (1999).
7. Tanabe, K., Misono, M., Ono, Y., and Hattori, H., “New Solid Acids
and Bases.” Elsevier, Amsterdam, 1989.
8
9
. Hattori, H., Chem. Rev. 95, 537 (1995).
. Barthomeuf, D., Catal. Rev.-Sci. Eng. 38, 521 (1996).
playa beneficialpart in base catalysisinvolvingcarbon diox- 10. Hathaway, P. E., and Davis, M. E., J. Catal. 116, 263 (1989).
1
1
1
1. Hathaway, P. E., and Davis, M. E., J. Catal. 116, 279 (1989).
2. Corma, A., and Martin-Aranda., R. M., J. Catal. 130, 130 (1991).
3. Di Cosimo, J. I., Diez, V. K., and Apesteguia, C. R., Appl. Catal. A
ide cycloaddition reactions.
ACKNOWLEDGMENTS
1
37, 149 (1996).
4. Wieland, W. S., Davis, R. J., and Garces, J. M., J. Catal. 173, 490
1998).
5. Corma, A., Iborra, S., Miquel, S., and Primo, J., J. Catal. 173, 315
1998).
6. Doskocil, E. J., and Davis, R. J., J. Catal. 188, 353 (1999).
1
1
1
This work was supported by the Department of Energy—Basic Energy
Sciences (DEFG0295ER14549.) Eastman Chemical Co. is acknowledged
for kindly donating epoxybutene.
(
(
REFERENCES
17. Bordawekar, S. V., and Davis, R. J., J. Catal. 189, 79 (2000).
18. Moreau, C., Durand, R., Roux, A., and Tichit, D., Appl. Catal. A 193,
1
2
. Darensbourg, D. J., and Holtcamp, M. W., Coord. Chem. Rev. 153, 155
1996).
. Lichtenwalter, M., and Cooper, J. F., U.S. Patent 2,773,070, 1956,
assigned to Jefferson Chemical Co.; McClellan, P. P., and U.S. Patent
257 (2000).
(
19. Ballini, R., Bigi, F., Gogni, E., Maggi, R., and Sartori, G., J. Catal. 191,
348 (2000).
20. Cardona-Martinez, N., and Dumesic, J. A., Adv. Catal. 38, 149 (1992).
21. Auroux, A., and Gervasini, A., J. Phys. Chem. 94, 6371 (1990).
2
,873,282, 1959, assigned to Jefferson Chemical Co.; McMullen, C. H.,
Nelson, J. R., Ream, B. C., and Sims, J. A., U.K. Patent 2,011,402A, 22. Bordawekar, S. V., Dosckocil, E. J., and Davis, R. J., Langmuir 14,
1
978, assigned to Union Carbide Corp.
1734 (1998).
3
4
. Kihara, N. H., and Endo, T., J. Org. Chem. 58, 6198 (1993).
. Yano, T., Matsui, H., Koike, T., Ishiguro, H., Fujihara, H., Yoshihara,
M., and Maeshima, T., Chem. Commun. 1129 (1997).
. Yamaguchi, K., Ebitani, K., Yoshida, T., Yoshida, H., and Kaneda, K.,
J. Am. Chem. Soc. 121, 4526 (1999).
23. Shapiro, A. L., Levin, S. Z., and Chekohvskaya, V. P., Zh. Org. Khim.
4(12), 2065 (1968).
24. Rodriguez, I., Cambon, H., Brunel, D., Lasperas, M., and Geneste, P.,
in “Heterogeneous Catalysis and Fine Chemicals III” (M. Guisnet,
et al., Eds.), p. 623. Elsevier, Amsterdam, 1993.
5