Bifunctional porphyrin catalysts for the synthesis of cyclic carbonates from epoxides and CO2: Structural optimization and mechanistic study

Article abstract of DOI:10.1021/ja507665a

We prepared bifunctional Mg^II porphyrin catalysts 1 for the solvent-free synthesis of cyclic carbonates from epoxides and CO₂. The activities of 1d, 1h, and 1i, which have Br^-, Cl^-, and I^- counteranions, respectively, increased in the order 1i < 1h < 1d. Catalysts 1d and 1j-m, which bear four tetraalkylammonium bromide groups with different alkyl chain lengths, showed comparable but slightly different activities. Based on the excellent catalyst 1d, we synthesized Mg^II porphyrin 1o with eight tetraalkylammonium bromide groups, which showed even higher catalytic activity (turnover number, 138,000; turnover frequency, 19,000 h^-1). The catalytic mechanism was studied by using 1d. The yields were nearly constant at initial CO₂ pressures in the 1-6 MPa range, suggesting that CO₂ was not involved in the rate-determining step in this pressure range. No reaction proceeded in supercritical CO₂, probably because the epoxide (into which the catalyst dissolved) dissolved in and was diluted by the supercritical CO₂. Experiments with ¹⁸O-labeled CO₂ and D-labeled epoxide suggested that the catalytic cycle involved initial nucleophilic attack of Br^- on the less hindered side of the epoxide to generate an oxyanion, which underwent CO₂ insertion to afford a CO₂ adduct; subsequent intramolecular ring closure formed the cyclic carbonate and regenerated the catalyst. Density functional theory calculations gave results consistent with the experimental results, revealing that the quaternary ammonium cation underwent conformational changes that stabilized various anionic species generated during the catalytic cycle. The high activity of 1d and 1o was due to the cooperative action of the Mg^II and Br^- and a conformational change (induced-fit) of the quaternary ammonium cation.

Full text of DOI:10.1021/ja507665a

Article
pubs.acs.org/JACS
Bifunctional Porphyrin Catalysts for the Synthesis of Cyclic
Carbonates from Epoxides and CO2: Structural Optimization and
Mechanistic Study
Tadashi Ema,*,† Yuki Miyazaki,† Junta Shimonishi,† Chihiro Maeda,† and Jun-ya Hasegawa*,‡
†Division of Chemistry and Biotechnology, Graduate School of Natural Science and Technology, Okayama University, Tsushima,
Okayama 700-8530, Japan
‡Catalysis Research Center, Hokkaido University, Kita 21, Nishi 10, Kita-ku, Sapporo 001-0021, Japan
S
* Supporting Information
ABSTRACT: We prepared bifunctional MgII porphyrin
catalysts 1 for the solvent-free synthesis of cyclic carbonates
from epoxides and CO2. The activities of 1d, 1h, and 1i, which
have Br−, Cl−, and I− counteranions, respectively, increased in
the order 1i < 1h < 1d. Catalysts 1d and 1j−m, which bear
four tetraalkylammonium bromide groups with different alkyl
chain lengths, showed comparable but slightly different
activities. Based on the excellent catalyst 1d, we synthesized
MgII porphyrin 1o with eight tetraalkylammonium bromide
groups, which showed even higher catalytic activity (turnover
number, 138,000; turnover frequency, 19,000 h−1). The
catalytic mechanism was studied by using 1d. The yields were nearly constant at initial CO2 pressures in the 1−6 MPa
range, suggesting that CO2 was not involved in the rate-determining step in this pressure range. No reaction proceeded in
supercritical CO2, probably because the epoxide (into which the catalyst dissolved) dissolved in and was diluted by the
supercritical CO2. Experiments with 18O-labeled CO2 and D-labeled epoxide suggested that the catalytic cycle involved initial
nucleophilic attack of Br− on the less hindered side of the epoxide to generate an oxyanion, which underwent CO2 insertion to
afford a CO2 adduct; subsequent intramolecular ring closure formed the cyclic carbonate and regenerated the catalyst. Density
functional theory calculations gave results consistent with the experimental results, revealing that the quaternary ammonium
cation underwent conformational changes that stabilized various anionic species generated during the catalytic cycle. The high
activity of 1d and 1o was due to the cooperative action of the MgII and Br− and a conformational change (induced-fit) of the
quaternary ammonium cation.
cooperative effects of the catalytic functional groups.3b,e,4d,5b,h,9
INTRODUCTION
■
In previous work, we designed and synthesized bifunctional
Carbon dioxide (CO2) is an inexpensive, nontoxic, and
abundant carbon source for organic synthesis, and CO2 has
porphyrin catalysts 1a−g (Figure 1),7 which showed high
been converted into various valuable chemicals.1−7 For
catalytic activity derived from the cooperative action of the
nucleophilic halide anion (X−) and the Lewis acid metal center
example, cyclic carbonates, which are used as raw materials
for polycarbonates, as electrolytes in lithium-ion secondary
batteries, and as polar aprotic solvents, have been synthesized
(M). The organocatalytic moiety is attached to the meta
position of the benzene ring in 1 to secure the coordination
space of epoxide. The advantage of this catalytic motif is that
from CO2 and epoxides with high atom efficiency. Various
catalysts for this reaction have been developed,4−7 including
strict adjustment of the spacer length is unnecessary because
the X− anion moves freely to attack the epoxide at an
metal complexes (e.g., metal salen complexes and metal-
loporphyrins) and organocatalysts (e.g., quaternary ammonium
salts and N-heterocyclic carbenes).
Among these catalysts, metalloporphyrins show relatively
appropriate angle.7 Metalloporphyrins 1c and 1d with
quaternary tributylammonium bromide groups were more
active than 1a and 1b with quaternary triphenylphosphonium
bromide groups, and MgII porphyrin 1d was more active than
ZnII porphyrin 1c. With 2a as a substrate, 1d showed the
highest catalytic activity, with a turnover number (TON) of
103,000 and a turnover frequency (TOF) of 12,000 h−1 under
high catalytic activity.6 Various metal ions can be incorporated
into porphyrins, and the rigidity of the porphyrin framework
permits control of the spatial arrangement of the functional
groups attached to it.8 Therefore, the catalytic activity of
metalloporphyrins can be improved further by suitable
functionalization. Bifunctional or trifunctional catalysts show
high catalytic activity for various reactions owing to the
Received: July 28, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
RESULTS AND DISCUSSION
Structural Optimization by Tuning the Pendent
■
Nucleophilic Group. We optimized the structure of catalysts
1 by synthesizing a series of new derivatives (1h−m, Figure 1)
and evaluating their catalytic activity for 2a (Scheme 1). First,
the effect of the X− anion on the catalytic activity was examined
(Table 1). The catalytic activities of 1d, 1h, and 1i, which bear
a
Table 1. Effect of Counter Anion on Catalytic Activity
b
entry
catalyst
yield (%)
TON
c
1
1d
99
48
7
19,800
9,600
1,400
3,900
3,800
1,000
14
c
2
1h
1i
c
3
d
4
Mg(TPP) + TBAB
Mg(TPP) + TBAC
Mg(TPP) + TBAI
TBAB
78
75
20
71
99
55
d
5
d
6
e
7
e
8
TBAC
20
e
9
TBAI
11
Figure 1. (a) Cooperative activation of epoxide with a bifunctional
catalyst. (b) Bifunctional catalysts 1.
a
Conditions: 2a (1.00 g, 10.0 mmol), catalyst (indicated below), CO2
b
1
(1.7 MPa), 120 °C, 3 h, in a 30 mL autoclave. Determined by H
NMR using 2-methoxynaphthalene as an internal standard. Catalyst 1
(0.005 mol %). Mg(TPP) (0.02 mol %), quaternary ammonium salts
(0.08 mol %). Quaternary ammonium salts (5 mol %).
c
solvent-free conditions (Scheme 1). In contrast, a binary
catalytic system composed of (5,10,15,20-tetraphenyl-
porphyrinato)magnesium(II) (Mg(TPP)) and tetrabutylam-
monium bromide (TBAB) showed much lower catalytic activity
(TON = 5,000) under otherwise identical reaction conditions.
These results clearly indicate the importance of cooperation
between the two functional groups in the same molecule.
Comparison of catalysts 1d−f, which have different alkyl chain
(linker) lengths, revealed that the catalytic activities decreased
slightly in the order 1d > 1e ≈ 1f (C6 > C8 ≈ C4). Catalyst 1g,
which has the alkylpyridinium bromide groups, showed much
lower activity because of the poor nucleophilicity of the Br− ion
of the tight ion pair and the low solubility of 1g in epoxide 2a.7
d
e
Br−, Cl−, and I− ions, respectively, increased in the order 1i <
1h < 1d (I− < Cl− < Br−; entries 1−3). In general, Cl− has the
highest nucleophilicity in aprotic solvents, whereas I− has the
highest leaving ability.11 Our results indicate that both the
nucleophilicity and the leaving ability of the X− anion are
important for the catalytic activity of 1, although the former
seems to be more important because 1h is more active than 1i.
For comparison, we also evaluated the catalytic activities of
TBAB, tetrabutylammonium chloride (TBAC), and tetrabuty-
lammonium iodide (TBAI) both in the presence and in the
absence of Mg(TPP) (entries 4−9). In the binary catalytic
systems, there was little or no difference in catalytic activity
between TBAB and TBAC (entries 4 and 5). However, when
the quaternary ammonium salts were used alone, TBAC
showed the highest catalytic activity (compare entries 7−9).
These results can be explained as follows: in the absence of the
Lewis acid, nucleophilic attack of Br− on the epoxide was
slower than that of Cl− because Br− is intrinsically less
nucleophilic than Cl−; but in the presence of Mg(TPP), the
ring-opening of the epoxide by Br− was more effectively
accelerated by the metal coordination. The ring-opening
activity of 1d was further enhanced by structure-specific
synergy between the Lewis acid (MgII) and base (Br−), which
was revealed by DFT calculations as described later.
Next we tuned the tetraalkylammonium cation moiety, which
can be expected to affect the nucleophilicity of the X− anion
(Table 2). Although bifunctional catalysts 1k−m were only
slightly less active than 1d, 1j showed somewhat lower activity.
The less bulky triethylammonium cation in this catalyst may
have formed a tighter ion pair with Br−, thus decreasing its
nucleophilicity. This result is similar to the previously reported
results indicating that catalyst 1g, bearing the alkylpyridinium
bromide groups, showed low activity because the alkylpyr-
idinium cation is less bulky than tetraalkylammonium cations.7
To evaluate the advantage conferred by the Br− anion of 1d,
we examined 1n, which has the diethylamino groups instead of
quaternary ammonium salts. We anticipated that 1n would
Scheme 1. Synthesis of Cyclic Carbonate 3 from Epoxide 2
and CO2
Here, we studied bifunctional catalysts 1 in more detail. We
conducted further structural optimization by synthesizing 1h−
m, which have different ion pairs, and 1n, which has the
diethylamino groups. We investigated the reaction mechanism
by changing the CO2 pressure (concentration) and by carrying
out the reaction with 18O-labeled CO2 and D-labeled epoxide.
Density functional theory (DFT) calculations were conducted
to gain additional insight into the catalytic mechanism. These
mechanistic studies have revealed the molecular behavior of
highly active bifunctional catalysts 1 cooperatively using the
flexible organocatalytic group and the Lewis acidic metal center
under solvent-free conditions. The present study is also useful
for the design of new catalysts as well as the understanding of
congener catalysts.10
B
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
the reaction proceeded under solvent-free conditions. The
autoclave was charged with 2a, 1d (0.005 or 0.002 mol %), and
CO2, and then the mixture was heated with stirring at 100 °C
for 3 h (Figure 3). The yields remained nearly constant at initial
Table 2. Effect of Counter Cation on Catalytic Activity
b
entry
catalyst
yield (%)
TON
1
2
3
4
5
1d
1j
68
56
67
66
65
34,000
28,000
33,500
33,000
32,500
1k
1l
1m
a
Conditions: 2a (1.00 g, 10.0 mmol), catalyst 1 (0.002 mol %), CO2
b
1
(1.7 MPa), 120 °C, 3 h, in a 30 mL autoclave. Determined by H
NMR using 2-methoxynaphthalene as an internal standard.
show lower activity than 1d or 1j because the nucleophile (the
amino group) of 1n must be connected to the substrate via a
covalent bond with some strain during catalysis. Indeed, the 1n-
catalyzed reaction of 2a (0.005 mol %, 1.7 MPa CO2, 120 °C, 3
h) gave 3a in only 15% yield (compared to 99% for 1d and 79%
for 1j under otherwise identical conditions). This result clearly
demonstrates the superiority of the tetraalkylammonium
bromide (ion pair) over the amino group as an organocatalytic
site.
Figure 3. Effect of CO2 pressure. Conditions: 2a (1.00 g, 10.0 mmol),
catalyst 1d (red and blue: 0.005 mol %, magenta: 0.002 mol %),
additive 3a (red and magenta: 0 mol %, blue: 20 mol %), CO2 (initial
pressure indicated above), 100 °C, 3 h, in a 35 mL stainless steel
autoclave with a sapphire window. The yield was determined by H
NMR using 2-methoxynaphthalene as an internal standard.
Based on all the results shown above, we designed and
synthesized MgII porphyrin 1o with eight tetraalkylammonium
bromide groups (Figure 2). Although we were concerned about
1
CO2 pressures in the range of 1−6 MPa, suggesting that CO2
was not involved in the rate-determining step in this pressure
range. Interestingly, almost no reaction proceeded in super-
critical CO2 (scCO2) at above 7 MPa of initial CO2 pressure.
Epoxide 2a dissolved in scCO2, whereas catalyst 1d did not.12
When the initial CO2 pressure was 6−6.5 MPa, CO2 was
partially liquid at room temperature but changed to scCO2
upon heating to 100 °C; during the heating period, 3a
accumulated at the bottom of the autoclave, and 1d dissolved in
the 3a, which appeared to promote the reaction. In contrast,
when 0.25−5.5 MPa CO2 was heated to 100 °C, CO2 remained
as a gas, substrate 2a remained as a liquid, and 1d dissolved in
it. These results suggest that dissolution of 1d in the epoxide or
the cyclic carbonate was the key to the reaction. In view of
these results, we decided to use 3a (20 mol %) as an additive
for the solubilization of 1d, and we carried out the reaction in
scCO2 (Figure 3). At a CO2 pressure of >7 MPa CO2, 1d and
3a dissolved completely in the scCO2,12 but no reaction
proceeded. Dilution of the epoxide and 1d in scCO2 likely
contributed to the drop in reaction rate.13
Figure 2. Bifunctional catalyst 1o.
the poor solubility of 1o in epoxide 2a because of the eight ion
pairs and higher molecular symmetry, fortunately, 1o was
soluble in 2a and exhibited even higher catalytic activity than
1d (Table 3, entries 1−4). The maximum TOF and TON for
a
Table 3. Comparison of Catalytic Activities of 1d and 1o
Catalytic Cycle. To determine the order of reaction with
respect to catalyst 1d, we examined the dependence of the
conversion on catalyst loading. The reactions of 2a with CO2
(1.7 MPa) were conducted with varying amounts of 1d at 120
°C for 1 h. The results are shown in Figure 4. Clearly, the
conversion of 2a at 1 h, which is proportional to the reaction
rate, is proportional to the amount (concentration) of 1d.
Therefore, the kinetic order with respect to 1d is 1, and one
catalyst molecule is involved in the rate-determining step.
To investigate the reaction mechanism in detail, we used
18O-labeled CO2 in the 1d-catalyzed reaction of styrene oxide
(2b) (Scheme 2). The Br− ion can attack either on the less
hindered side of 2b (path B, β attack) or on the more hindered
side of 2b (path A, α attack), giving two regioisomeric cyclic
carbonates 3b, which can be converted to benzaldehyde. Mass
spectrometric analysis of the benzaldehyde indicated that the
signal ratio of the products of path B and path A was 99:1 (see
the Supporting Information). This high β selectivity for 2b is
b
entry catalyst loading (mol %) time (h) yield (%)
TON
1
2
3
4
5
1d
1o
1d
1o
1o
0.003
0.003
1
36
57
82
95
69
12,000
19,000
1
0.0008
0.0008
0.0005
24
24
24
103,000
119,000
138,000
a
Conditions: 2a (1.00 g, 10.0 mmol), catalyst 1 (indicated above),
CO2 (1.7 MPa), 120 °C, in a 30 mL autoclave. Determined by H
NMR using 2-methoxynaphthalene as an internal standard.
b
1
1o reached 19,000 h−1 and 138,000, respectively (entries 2 and
5), which are the highest values for the reaction of epoxides
with CO2 into cyclic carbonates to the best of our knowledge.
Effect of CO2 Pressure. We investigated the effect of CO2
pressure on the 1d-catalyzed reaction in a stainless steel
autoclave with a sapphire window so that we could observe how
C
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Next, we determined whether the epoxide or CO2 was
initially attacked by the Br− ion, by investigating the reaction of
trans-deuterated epoxide 2c (Scheme 3).15 Attack of the Br−
Scheme 3. (a) Investigation of the Reaction Mechanism by
Using Deuterated Epoxide trans-2c and (b) Two Possible
Catalytic Cycles via the Nucleophilic Attack of Epoxide
(path B) or CO2 (path C) by the Br− Ion
Figure 4. Determination of the reaction order with respect to catalyst
1d. The experiments were done in triplicate.
Scheme 2. Investigation of the Reaction Mechanism by
a
Using 18O-Labeled CO2
ion on the epoxide (path B) would afford trans-3c (retention of
configuration) via two inversions, whereas attack of the Br− ion
on CO2 (path C) would afford cis-3c (inversion of
1
configuration). Analysis of the product mixture by H NMR
indicated that trans-3c and cis-3c were produced in a ratio of
99:1 (see the Supporting Information).
On the basis of the results of these mechanistic studies, we
propose the catalytic cycle shown in Scheme 4. After
coordination of the epoxide with the metal atom, nucleophilic
attack by the Br− ion on the less hindered side of the epoxide
Scheme 4. A Proposed Catalytic Cycle
a
(a) Synthesis and analysis of 18O-labeled cyclic carbonate 3b. (b)
Two possible catalytic cycles: nucleophilic attack by the Br− ion on the
less hindered side of 2b (path B, β attack) and more hindered side of
2b (path A, α attack).
surprising because Darensbourg, Lu, and co-workers have
reported that the control of regioselectivity is difficult in the
copolymerization of 2b with CO2.3c,d In addition, Kleij, Bo, and
co-workers have reported that the α pathway is predominant in
the conversion of 2b into 3b with a Zn(salphen)/TBAI
catalytic system.4l It is believed that in most catalytic systems,
terminal epoxides with electron-donating substituents favor β
attack, while those with electron-withdrawing substituents, such
as 2b, undergo α attack but with low selectivity. Therefore, the
high β selectivity of 1d toward 2b, which has been revealed by
the isotope experiment with 18O-labeled CO2 (Scheme 2), is
unusual and significant.14
D
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
affords an alkoxide anion via a ring-opening reaction,16 which is
followed by CO2 insertion and ring closure to form the cyclic
carbonate and regenerate 1d.
DFT Calculations. To gain deeper insight into the reaction
mechanism, we performed DFT calculations along the
proposed reaction pathway (Scheme 4). To investigate the
effect of the X− anion on catalytic activity, we used
monosubstituted porphyrins 1d′, 1h′, and 1i′ (Figure 5) as
anion-binding site. As represented by Figure 7a, the X− anion in
reactant complex R is stabilized by the positive charges
distributed over the QA cation and is poised to attack epoxide
2d. Nucleophilic attack by the X− anion gives intermediate I1
via transition state I1_TS (Figure 7b,c). The activation energies
(Ea) are 13.5, 16.1, and 20.2 kcal/mol for the reactions with
1d′, 1h′, and 1i′, respectively, which suggests that the
nucleophilicity of the X− anion in 1 increases in the order I−
< Cl− < Br− (Figure 6). Note that the usual order of the
nucleophilicities of Br− and Cl− is reversed in this case; the
nucleophilicity of “naked” X− anions in polar aprotic solvents
generally increases with basicity: I− < Br− < Cl−.11,17
The X− anion in 1 is not “naked” but “solvated by the
tethered QA cation” so that the nucleophilicity of the X− anion
can be affected by the tethered QA cation. Therefore, the
reversal in nucleophilicity occurred probably because the QA
cation in 1d′ stabilizes Br− with the best complementarity
throughout the nucleophilic attack, especially in the transition
state; see the concave anion-binding site that is complementary
to the Br− ion (Figure 7b). This stabilization can be deduced by
comparing the shortest distances between the QA cation and
the X− anions in transition state I1_TS. The shortest distance
between Br− and H(QA) in 1d′ is 2.67 Å (Figure 7b), whereas
that between Cl− and H(QA) in 1h′ is 2.69 Å and that between
I− and H(QA) in 1i′ is 3.48 Å (Supporting Information).
Because the ionic radii of the X− anions increase in the order
Cl− < Br− < I−, the shortest distance between Br− and H(QA)
(2.67 Å) strongly suggests that tight and attractive interactions
between Br− and the QA cation stabilize the transition state.
Figure 5. Model catalysts for DFT calculations.
models for catalyst 1d, 1h, and 1i, respectively, and we used
epoxide 2d as a substrate (Scheme 1). The potential energy
profiles calculated for the reactions catalyzed by 1d′, 1h′, and
1i′ are shown in Figure 6. The intermediate and transition-state
structures in the 1d′-catalyzed reaction are shown in Figure 7,
and those for 1h′ and 1i′ are given in the Supporting
Information.
Interestingly, the quaternary ammonium (QA) cation carries
the positive charges on the H atoms, but not on the central N
atom (Figure 7), and the positively charged H atoms form an
Figure 6. Potential energy profiles for the 1d′, 1h′, and 1i′-catalyzed reactions of 2d with CO2. Computations were performed at the B3LYP/6-31G*
level with the self-consistent reaction field (SCRF) method (Et2O). The relative energies based on reactant R (stretched form) are given in kcal/mol.
The energy of CO2 is included in the former steps where CO2 does not appear explicitly. The stretched and bent forms designated here can be seen
in Figure 7.
E
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 7. Optimized structures of (a) reactant R (stretched form), (b) transition state I1_TS, (c) I1 (stretched form), (d) I1 (bent form), (e)
transition state I2_TS, (f) I2, (g) transition state P_TS, and (h) product P in the 1d′-catalyzed reaction of 2d with CO2.
The same tendency can be seen for the distances between X−
and H(QA) of the linker moiety, which is a part of the concave
anion-binding site: 2.99 Å (Figure 7b), 3.11 Å, and 3.84 Å
(Supporting Information) for 1d′, 1h′, and 1i′, respectively.
These observations strongly suggest that the linker length is
important for the transition-state stabilization, which was
observed experimentally, 1d (C6) > 1e (C8) ≈ 1f (C4),
although the difference was not so critical probably because of
the flexible alkyl chains.7 Furthermore, these attractive
interactions between the QA cation and the Br atom appear
to be more or less retained in intermediate I1 (stretched form)
(Figure 7c, 2.82 and 3.09 Å).
It is also interesting to compare the conformational changes
of the QA cations in 1d′, 1h′, and 1i′ in going from R to I1
(stretched form). We used the N(QA)−O(2d) distance to
estimate the conformational change of the QA cation. In the
case of 1d′, the N(QA)−O(2d) distance changes from 9.30 Å
in R to 8.54 Å in the stretched form of I1; that is, the N(QA)
atom in 1d′ gets closer to 2d by 0.76 Å. The corresponding
values in 1h′ and 1i′ are 1.40 and 1.11 Å, respectively. Clearly,
1d′ experiences the smallest conformational change. The
F
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
smallest conformational change observed for 1d′ is a result of
the best preorganization leading to the ring-opening reaction
guided by the multipoint interactions described above. For
more discussion of the origin of the reversed order of the
nucleophilicities of Br− and Cl− in 1 see the Supporting
Information.17
The charge on the Br atom changes from −0.88 in R to
−0.20 in the stretched form of I1, whereas that on the O atom
of 2d changes from −0.62 in R to −0.92 in the stretched form
of I1; this shift in negative charge from the Br atom to the O
atom is facilitated by the partial emergence of the positively
charged Mg atom from the porphyrin cavity (Figure 7a−c).18
This shift in negative charge triggers the conversion of the
stretched form of I1 into a bent form in which the QA cation is
closer to the negatively charged O atom of 2d that has been
ring-opened (2.31 Å; Figure 7d). To maximize these attractive
interactions, the methyl group of 2d, which is directed upward
in the stretched form of I1 (Figure 7c), points away from the
QA cation in the bent form of I1 (Figure 7d).
After CO2 has been weakly bound to give complex I1···CO2
(not shown in Figure 7), the C−O bond forms via transition
state I2_TS, in which a negatively charged O atom of CO2 is
stabilized by the QA cation (Figure 7e). This reaction has a
very small activation barrier (Ea = 0.6−0.7 kcal/mol for 1d′ and
1h′ and 2.8 kcal/mol for 1i′; Figure 6). Formation of the C−O
bond and coordination of the resulting carbonate anion to the
Mg atom lead to intermediate I2, which is stabilized by multiple
interactions between the carbonate anion and the QA cation in
a bent conformation (Figure 7f). The QA cation interacts with
the two negatively charged O atoms, with the shortest
H(QA)−O(carbonate) distance being 2.18 Å. Electrostatic
stabilization is clearly maximized by the conformational change
of the QA cation. This CO2 insertion reaction is very
exothermic (ΔH° ≈ −20 kcal/mol; Figure 6).19
The catalytic cycle is completed by an intramolecular SN2
reaction that occurs via transition state P_TS and gives product
P (Figure 7g,h). In transition state P_TS, the QA cation
interacts tightly with the two negatively charged O atoms of the
carbonate rather than with the leaving Br atom. The Ea values
for the ring-closing reaction increase in the order 1i′ < 1d′ <
1h′, and the ΔH° (absolute) values show the opposite trend
(Figure 6). These trends are related to the strength of the C−X
bonds. The average bond dissociation energies (D°) for the C−
Cl, C−Br and C−I bonds are reported to be 95, 67, and 50
kcal/mol, respectively.20 The Ea and ΔH° values for 1i′ are the
smallest and the most negative, respectively, because the C−I
bond is the weakest of the C−X bonds. In contrast, 1h′ shows
the largest Ea value and the least negative ΔH° value because
the C−Cl bond is the strongest of the C−X bonds. The X−
anion regenerated in P interacts strongly with the QA cation.
Replacement of cyclic carbonate 3d with epoxide 2d and a
conformational change gives a stretched form of R.21
prediction is supported by the fact that the yields were almost
constant at 1−6 MPa CO2 pressure (Figure 3). The final step is
the ring closure of stable intermediate I2. Because the Ea values
for transition state P_TS are relatively large (Figure 6), this
step contributes to the whole reaction rate to some degree. To
the best of our knowledge, the Ea value calculated for 1d′ (13.5
kcal/mol) is the smallest, which accords with the fact that 1d
and 1o are the most active catalysts so far reported.
CONCLUSIONS
■
Chemical fixation of CO2 is becoming increasingly important.
Here, we describe the structural optimization and detailed
mechanistic analysis of highly active bifunctional catalysts 1,
which have a metal center and quaternary ammonium groups.
Structural optimization of the organocatalytic moiety revealed
that the combination of a tributylammonium cation and a Br−
ion was the best choice. A key to the solvent-free reactions was
the solubility of the catalyst in the epoxide substrate or the
cyclic carbonate product. The high catalytic activity of 1d
(TON = 103,000 and TOF = 12,000 h−1) and 1o (TON =
138,000 and TOF = 19,000 h−1) is achieved by means of
synergistic cooperation between a Lewis acid (MgII) and a base
(Br−) without self-quenching. On the basis of the results of
experiments using 18O-labeled CO2 and D-labeled epoxide, we
have proposed a mechanism for the reaction. The proposed
reaction mechanism was supported by the results of DFT
calculations. The activation energies (Ea) estimated for the rate-
determining step were in good agreement with the
experimentally observed trend for the catalytic activities of
1d, 1h, and 1i. 1d and 1o are the most active catalysts so far
reported, which has been supported by the Ea value calculated
for 1d′ (13.5 kcal/mol), which is the smallest value. The
cooperative effect of the two functional groups in 1 is the
principal origin of the catalytic power. Although 1 was
originally designed to activate the epoxide by orienting it in a
near-attack conformation (Figure 1a), DFT calculations
revealed that the flexible QA cation stabilized various anionic
species generated during the catalytic cycle in an induced-fit
manner, which is reminiscent of enzymatic catalysis. Synergistic
multiple interactions accompanied by conformational changes
play a pivotal role in bifunctional catalysts 1 orchestrating metal
catalysis and organocatalysis.
EXPERIMENTAL SECTION
■
Synthesis of Catalysts. Bifunctional porphyrins 1h−o were newly
synthesized according to the previously reported methods.7,22 The
detailed procedures are given in the Supporting Information.
Effect of CO2 Pressure. A 35 mL stainless steel autoclave with a
sapphire window was charged with epoxide 2a (1.00 g, 10.0 mmol), 3a
(0 or 20 mol %), 1d (1.05 mg, 0.501 μmol, 0.005 mol % or 0.420 mg,
0.201 μmol, 0.002 mol %), and then CO2. The mixture was heated
with stirring at 100 °C for 3 h. The reactor was cooled in an ice bath
for 30 min, and excess CO2 was released carefully. The crude product
was dissolved in Et2O, and the solution was concentrated. The NMR
yield was determined by using 2-methoxynaphthalene as an internal
standard.
Because the highest energy transition state is generally the
rate-determining step,11 Figure 6 indicates that the rate-
determining step involves transition state I1_TS in all cases.
On the basis of the Ea values for I1_TS, we can predict that the
catalytic activities of the three catalysts should increase in the
order 1i′ (20.2 kcal/mol) < 1h′ (16.1 kcal/mol) < 1d′ (13.5
kcal/mol). This prediction agrees qualitatively with the yields
shown in Table 1. Figure 6 also suggests that the CO2 insertion
reaction is not the rate-determining step because the Ea value
for transition state I2_TS is very small in all cases, as described
above, and because I2_TS is lower in energy than I1_TS. This
Synthesis of 18O-Labeled Cyclic Carbonate 3b. A 30 mL
stainless steel autoclave was charged with styrene oxide (2b) (2.40 g,
20.0 mmol), 1d (2.10 mg, 1.00 μmol, 0.005 mol %), and then 18O-
labeled CO2 (0.75 MPa). The mixture was stirred at 120 °C for 24 h.
The reactor was cooled in an ice bath for 30 min, and excess CO2 was
released carefully. Purification by silica gel column chromatography
(hexane/EtOAc (3:1)) gave 18O-labeled 3b as a colorless oil (1.87 g,
55%). 1H NMR (CDCl3, 600 MHz) δ 4.35 (t, J = 8.2 Hz, 1H), 4.80 (t,
J = 8.2 Hz, 1H), 5.68 (t, J = 8.2 Hz, 1H), 7.36−7.38 (m, 2H), 7.42−
G
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
7.47 (m, 3H); 13C NMR (CDCl3, 150 MHz) δ 71.1, 78.0, 125.8,
129.2, 129.7, 135.8, 154.8; MS (EI) calcd for C9H8O18O2: 168.1;
found: 168.1 (M+).
measured at the SC-NMR Laboratory of Okayama University.
Computations were partly carried out at RCCS (Okazaki,
Japan) and ACCMS (Kyoto University). We thank Prof. A.
Osuka (Kyoto University) for ESI-MS measurements.
Conversion of 18O-Labeled 3b into Benzaldehyde. A mixture
of 18O-labeled 3b (934 mg, 5.56 mmol) and 2 M NaOH (11 mL) was
stirred at room temperature for 2 h. After neutralization, the product
was extracted with EtOAc and washed with brine. The mixture was
dried over Na2SO4 and concentrated. Purification by silica gel column
chromatography (hexane/EtOAc (1:2)) gave the product as a white
solid (699 mg, 90%). MS (EI) calcd for C8H10O18O: 140.1; found:
140.1 (M+). A solution of 18O-labeled 1-phenyl-1,2-ethanediol (210
mg, 1.50 mmol) in MeOH (5 mL) was stirred at 40 °C for 10 min
under Ar. NaIO4 (353 mg, 1.65 mmol) and 0.5 M H2SO4 (18 mL)
were added, and the mixture was stirred at 40 °C for 20 min. The
product was extracted with Et2O. The mixture was dried over Na2SO4
and concentrated. Purification by distillation (5 mmHg, 60 °C) gave
benzaldehyde (134 mg, 84%), which was subjected to mass
spectrometry. MS (EI) calcd for C7H6O: 106.0; found: 106.0 (M+);
MS (EI) calcd for C7H618O: 108.0; found: 108.0 (M+).
REFERENCES
■
(1) Reviews: (a) Sakakura, T.; Choi, J.-C.; Yasuda, H. Chem. Rev.
2007, 107, 2365−2387. (b) Darensbourg, D. J. Chem. Rev. 2007, 107,
2388−2410. (c) Sakakura, T.; Kohno, K. Chem. Commun. 2009,
1312−1330. (d) North, M.; Pasquale, R.; Young, C. Green Chem.
2010, 12, 1514−1539. (e) Riduan, S. N.; Zhang, Y. Dalton Trans.
2010, 39, 3347−3357. (f) Klaus, S.; Lehenmeier, M. W.; Anderson, C.
E.; Rieger, B. Coord. Chem. Rev. 2011, 255, 1460−1479. (g) Cokoja,
M.; Bruckmeier, C.; Rieger, B.; Herrmann, W. A.; Kuhn, F. E. Angew.
̈
Chem., Int. Ed. 2011, 50, 8510−8537. (h) Tsuji, Y.; Fujihara, T. Chem.
Commun. 2012, 48, 9956−9964. (i) Darensbourg, D. J.; Wilson, S. J.
Green Chem. 2012, 14, 2665−2671. (j) Omae, I. Coord. Chem. Rev.
2012, 256, 1384−1405. (k) Kielland, N.; Whiteoak, C. J.; Kleij, A. W.
Adv. Synth. Catal. 2013, 355, 2115−2138. (l) Maeda, C.; Miyazaki, Y.;
Ema, T. Catal. Sci. Technol. 2014, 4, 1482−1497.
Synthesis of trans-Deuterated Cyclic Carbonate 3c. A 30 mL
stainless steel autoclave was charged with trans-deuterated epoxide 2c
(984 mg, 9.73 mmol),15 1d (1.05 mg, 0.501 μmol, 0.005 mol %), and
then CO2 (1.5 MPa). The mixture was heated with stirring at 120 °C
for 6 h. The reactor was cooled in an ice bath for 30 min, and excess
CO2 was released carefully. Purification by silica gel column
chromatography (hexane/EtOAc (3:1)) gave trans-3c as a colorless
(2) Various transformations of CO2: (a) Kayaki, Y.; Yamamoto, M.;
Ikariya, T. Angew. Chem., Int. Ed. 2009, 48, 4194−4197. (b) Riduan, S.
N.; Zhang, Y.; Ying, J. Y. Angew. Chem., Int. Ed. 2009, 48, 3322−3325.
(c) Zhang, W.-Z.; Li, W.-J.; Zhang, X.; Zhou, H.; Lu, X.-B. Org. Lett.
2010, 12, 4748−4751. (d) Gu, L.; Zhang, Y. J. Am. Chem. Soc. 2010,
132, 914−915. (e) Berkefeld, A.; Piers, W. E.; Parvez, M. J. Am. Chem.
Soc. 2010, 132, 10660−10661. (f) Mizuno, H.; Takaya, J.; Iwasawa, N.
J. Am. Chem. Soc. 2011, 133, 1251−1253. (g) Ohmiya, H.; Tanabe, M.;
Sawamura, M. Org. Lett. 2011, 13, 1086−1088. (h) Inamoto, K.;
Asano, N.; Nakamura, Y.; Yonemoto, M.; Kondo, Y. Org. Lett. 2012,
14, 2622−2625. (i) Fujihara, T.; Nogi, K.; Xu, T.; Terao, J.; Tsuji, Y. J.
Am. Chem. Soc. 2012, 134, 9106−9109. (j) Dai, Y.; Feng, X.; Wang, B.;
He, R.; Bao, M. J. Organomet. Chem. 2012, 696, 4309−4314. (k) Mita,
T.; Michigami, K.; Sato, Y. Org. Lett. 2012, 14, 3462−3465. (l) Mita,
T.; Sugawara, M.; Hasegawa, H.; Sato, Y. J. Org. Chem. 2012, 77,
2159−2168. (m) Ishida, N.; Shimamoto, Y.; Murakami, M. Angew.
1
oil (1.02 g, 72%). H NMR (CDCl3, 600 MHz) δ 0.93 (t, J = 7.2 Hz,
3H), 1.32−1.49 (m, 4H), 1.66−1.72 (m, 1H), 1.79−1.85 (m, 1H),
4.06 (d, J = 7.0 Hz, 1H), 4.70 (q, J = 7.0 Hz, 1H); 13C NMR (CDCl3,
150 MHz) δ 13.5, 22.0, 26.2, 33.2, 68.9 (t, JCD = 23.7 Hz), 76.9, 155.0;
IR (neat) 2962, 2936, 2870, 1794, 1543, 1466, 1366, 1304, 1177, 1061,
775, 714 cm−1.
Computational Details. DFT calculations were performed along
the proposed reaction pathway (Scheme 4). Monosubstituted
porphyrins 1d′, 1h′, and 1i′ (Figure 5) were employed as model
catalysts for 1d, 1h, and 1i, respectively, while epoxide 2d was
employed as a substrate (Scheme 1). Computations were performed at
the B3LYP23/6-31G*24 level. The self-consistent reaction field
(SCRF) method with the polarizable continuum model (PCM)25
was adopted to take into account of the solvation effect. The PCM
parameters for Et2O, which is expected to have a dielectric constant
similar to that of epoxide, was used.26 The harmonic frequency and
normal mode were calculated to verify the transition-state structures.
Natural population analysis (NPA)27 was done to calculate the natural
atomic charges. Computations were performed with Gaussian 09
program.28
́
Chem., Int. Ed. 2012, 51, 11750−11752. (n) Leon, T.; Correa, A.;
Martin, R. J. Am. Chem. Soc. 2013, 135, 1221−1224. (o) Ishida, T.;
Kikuchi, S.; Tsubo, T.; Yamada, T. Org. Lett. 2013, 15, 848−851.
(p) Li, S.; Miao, B.; Yuan, W.; Ma, S. Org. Lett. 2013, 15, 977−979.
(q) Yu, B.; Zhang, H.; Zhao, Y.; Chen, S.; Xu, J.; Huang, C.; Liu, Z.
Green Chem. 2013, 15, 95−99. (r) Ueno, A.; Kayaki, Y.; Ikariya, T.
Green Chem. 2013, 15, 425−430. (s) Correa, A.; Leon
́
, T.; Martin, R. J.
Am. Chem. Soc. 2014, 136, 1062−1069.
(3) Polycarbonate synthesis from epoxides and CO2: (a) Nakano, K.;
Hashimoto, S.; Nozaki, K. Chem. Sci. 2010, 1, 369−373. (b) Vagin, S.
I.; Reichardt, R.; Klaus, S.; Rieger, B. J. Am. Chem. Soc. 2010, 132,
14367−14369. (c) Wu, G.-P.; Wei, S.-H.; Lu, X.-B.; Ren, W.-M.;
Darensbourg, D. J. Macromolecules 2010, 43, 9202−9204. (d) Wu, G.-
P.; Wei, S.-H.; Ren, W.-M.; Lu, X.-B.; Li, B.; Zu, Y.-P.; Darensbourg,
D. J. Energy Environ. Sci. 2011, 4, 5084−5092. (e) Wu, G.-P.; Wei, S.-
H.; Ren, W.-M.; Lu, X.-B.; Xu, T.-Q.; Darensbourg, D. J. J. Am. Chem.
Soc. 2011, 133, 15191−15199. (f) Li, H.; Niu, Y. Appl. Organometal.
Chem. 2011, 25, 424−428. (g) Nakano, K.; Kobayashi, K.; Nozaki, K.
J. Am. Chem. Soc. 2011, 133, 10720−10723. (h) Kim, J. G.; Cowman,
C. D.; LaPointe, A. M.; Wiesner, U.; Coates, G. W. Macromolecules
2011, 44, 1110−1113. (i) Widger, P. C. B.; Ahmed, S. M.; Coates, G.
W. Macromolecules 2011, 44, 5666−5670. (j) Nakano, K.; Kobayashi,
K.; Ohkawara, T.; Imoto, H.; Nozaki, K. J. Am. Chem. Soc. 2013, 135,
8456−8459. (k) Saini, P. K.; Romain, C.; Williams, C. K. Chem.
Commun. 2014, 50, 4164−4167.
ASSOCIATED CONTENT
■
S
* Supporting Information
Synthesis of 1h−o, 1H and 13C NMR spectra, isotope
experiments, determination of binding constants, computa-
tional details, and complete ref 28. This material is available free
of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Authors
ema@cc.okayama-u.ac.jp
hasegawa@cat.hokudai.ac.jp
Notes
The authors declare no competing financial interest.
(4) Recent examples of metal complex catalysts: (a) Song, J.; Zhang,
Z.; Hu, S.; Wu, T.; Jiang, T.; Han, B. Green Chem. 2009, 11, 1031−
1036. (b) Qiao, K.; Ono, F.; Bao, Q.; Tomida, D.; Yokoyama, C. J.
ACKNOWLEDGMENTS
■
This work was supported by ENEOS Hydrogen Trust Fund,
Adaptable and Seamless Technology Transfer Program through
Target-Driven R&D (A-STEP, Exploratory Research) from
Japan Science and Technology Agency (JST), and Okayama
Foundation for Science and Technology. NMR spectra were
Mol. Catal. A: Chem. 2009, 303, 30−34. (c) Ulusoy, M.; C
etinkaya, B. Appl. Organometal. Chem. 2009, 23, 68−74.
(d) Melendez, J.; North, M.; Villuendas, P. Chem. Commun. 2009,
̧etinkaya, E.;
Ç
́
2577−2579. (e) Clegg, W.; Harrington, R. W.; North, M.; Pasquale, R.
H
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Chem.Eur. J. 2010, 16, 6828−6843. (f) Decortes, A.; Belmonte, M.
M.; Benet-Buchholz, J.; Kleij, A. W. Chem. Commun. 2010, 46, 4580−
(11) For example, see: Smith, J. G. Organic Chemistry, 3rd ed.;
McGraw-Hill: Singapore, 2011.
(12) (a) Chrastil, J. J. Phys. Chem. 1982, 86, 3016−3021.
(b) Kawanami, H.; Sasaki, A.; Matsui, K.; Ikushima, Y. Chem.
Commun. 2003, 896−897.
(13) There is another possibility that 1d was decomposed by scCO2.
To test this possibility, only 1d was subjected to scCO2 (8 MPa CO2
initial pressure) at 100 °C for 1 h. After releasing the CO2, 2a and CO2
(2 MPa) were added, and the mixture was heated with stirring at 100
°C for 3 h. As a result, 3a was obtained in 66% yield. Therefore, this
possibility was ruled out.
(14) Interestingly, DFT calculations strongly suggested that the
regioselective ring-opening of 2d is under kinetic control depending
on the relative stability of the first transition state, whereas that of 2b is
under thermodynamic control depending on the relative stability of
the first reaction intermediate, both of which favor the β pathway
(path B). See the Supporting Information.
́
4582. (g) Melendez, J.; North, M.; Villuendas, P.; Young, C. Dalton
Trans. 2011, 40, 3885−3902. (h) Ramidi, P.; Sullivan, S. Z.; Gartia, Y.;
Munshi, P.; Griffin, W. O.; Darsey, J. A.; Biswas, A.; Shaikh, A. U.;
Ghosh, A. Ind. Eng. Chem. Res. 2011, 50, 7800−7807. (i) Chen, F.; Li,
X.; Wang, B.; Xu, T.; Chen, S.-L.; Liu, P.; Hu, C. Chem.Eur. J. 2012,
18, 9870−9876. (j) Whiteoak, C. J.; Kielland, N.; Laserna, V.;
Escudero-Adan
135, 1228−1231. (k) Escar
Martin, E.; Escudero-Adan
, E. C.; Kleij, A. W. Chem.Eur. J. 2013, 19,
2641−2648. (l) Castro-Gomez, F.; Salassa, G.; Kleij, A. W.; Bo, C.
́
, E. C.; Martin, E.; Kleij, A. W. J. Am. Chem. Soc. 2013,
́
cega-Bobadilla, M. V.; Belmonte, M. M.;
́
́
Chem.Eur. J. 2013, 19, 6289−6298. (m) Babu, H. V.; Muralidharan,
K. Dalton Trans. 2013, 42, 1238−1248. (n) Kim, J.; Kim, S.-N.; Jang,
H.-G.; Seo, G.; Ahn, W.-S. Appl. Catal. A: General 2013, 453, 175−
180. (o) Zalomaeva, O. V.; Chibiryaev, A. M.; Kovalenko, K. A.;
Kholdeeva, O. A.; Balzhinimaev, B. S.; Fedin, V. P. J. Catal. 2013, 298,
179−185. (p) Kim, S. H.; Ahn, D.; Go, M. J.; Park, M. H.; Kim, M.;
Lee, J.; Kim, Y. Organometallics 2014, 33, 2770−2775. (q) Castro-
(15) Zhou, H.; Zhang, W.-Z.; Liu, C.-H.; Qu, J.-P.; Lu, X.-B. J. Org.
Chem. 2008, 73, 8039−8044.
(16) When a mixture of epoxide 2a and catalyst 1d (0.005 mol%)
was heated at 120 °C for 3 h under Ar (1.5 MPa), no reaction took
place with 2a recovered (90% yield), which suggests that the initial
ring-opening step is reversible. DFT calculations indicated that the
reverse reaction of the epoxide ring-opening has a very small barrier
(6.5 kcal/mol for 2d), supporting reversibility in the initial step.
(17) DFT calculations on the ring-opening reactions of 2d with a
single catalyst TBAX (TBAC, TBAB, and TBAI) were also carried out
in the same way, and the optimized structures are given in the
Supporting Information. The activation energies for TBAC, TBAB,
and TBAI were calculated to be 22.3, 26.5, and 27.2 kcal/mol,
respectively, predicting the relative catalytic activity as follows: TBAI <
TBAB < TBAC. In fact, this agrees with the experimental results
(Table 1). Therefore, the reversal of the nucleophilicity of the halide
anion in 1 (Cl− < Br−) results from something specific to the
bifunctional catalyst 1 as discussed in the main text and the Supporting
Information. For the DFT calculations on similar systems, see: Wang,
J.-Q.; Dong, K.; Cheng, W.-G.; Sun, J.; Zhang, S.-J. Catal. Sci. Technol.
2012, 2, 1480−1484.
́
Osma, J. A.; Alonso-Moreno, C.; Lara-Sanchez, A.; Martinez, J.; North,
M.; Otero, A. Catal. Sci. Technol. 2014, 4, 1674−1684.
(5) Recent examples of organocatalysts: (a) Sakai, T.; Tsutsumi, Y.;
Ema, T. Green Chem. 2008, 10, 337−341. (b) Sun, J.; Zhang, S.;
Cheng, W.; Ren, J. Tetrahedron Lett. 2008, 49, 3588−3591.
(c) Motokura, K.; Itagaki, S.; Iwasawa, Y.; Miyaji, A.; Baba, T. Green
Chem. 2009, 11, 1876−1880. (d) Tsutsumi, Y.; Yamakawa, K.;
Yoshida, M.; Ema, T.; Sakai, T. Org. Lett. 2010, 12, 5728−5731.
(e) Prasetyanto, E. A.; Ansari, M. B.; Min, B.-H.; Park, S.-E. Catal.
Today 2010, 158, 252−257. (f) Zhou, H.; Wang, Y.-M.; Zhang, W.-Z.;
Qu, J.-P.; Lu, X.-B. Green Chem. 2011, 13, 644−650. (g) Liang, S.; Liu,
H.; Jiang, T.; Song, J.; Yang, G.; Han, B. Chem. Commun. 2011, 47,
2131−2133. (h) Han, L.; Choi, H.-J.; Choi, S.-J.; Liu, B.; Park, D.-W.
Green Chem. 2011, 13, 1023−1028. (i) Aoyagi, N.; Furusho, Y.; Endo,
T. Chem. Lett. 2012, 41, 240−241. (j) Yang, Z.-Z.; Zhao, Y.-N.; He, L.-
N.; Gao, J.; Yin, Z.-S. Green Chem. 2012, 14, 519−527. (k) Song, Q.-
W.; He, L.-N.; Wang, J.-Q.; Yasuda, H.; Sakakura, T. Green Chem.
2013, 15, 110−115. (l) Zhao, Y.; Yao, C.; Chen, G.; Yuan, Q. Green
Chem. 2013, 15, 446−452. (m) Aoyagi, N.; Furusho, Y.; Endo, T. J.
Polym. Sci. A: Polym. Chem. 2013, 51, 1230−1242. (n) Cheng, W.;
Chen, X.; Sun, J.; Wang, J.; Zhang, S. Catal. Today 2013, 200, 117−
124. (o) Chatelet, B.; Joucla, L.; Dutasta, J.-P.; Martinez, A.; Szeto, K.
C.; Dufaud, V. J. Am. Chem. Soc. 2013, 135, 5348−5351. (p) Wang, Y.-
B.; Sun, D.-S.; Zhou, H.; Zhang, W.-Z.; Lu, X.-B. Green Chem. 2014,
16, 2266−2272.
(18) In the absence of the axial ligand, the Mg atom is located just
inside the porphyrin cavity with an atomic charge of +1.43.
(19) The energy trough for intermediate I2 may suggest the existence
of a potential pathway to the formation of the alternating copolymer.
However, even when a mixture of 2d (0.29 g, 5.0 mmol), CO2 (3.0
MPa), and 1d (0.05 mol%) was stirred at ambient temperature (30
°C) for 24 h, 3d was exclusively produced in 80% without any
formation of polycarbonate. This result indicates that the structure of
1d is highly tuned for the formation of cyclic carbonate.
(20) CRC Handbook of Chemistry and Physics, 75th ed.; Lide, D. R.,
Ed.; CRC: Boca Raton, FL, 1994.
(21) Judging from the poor coordinative abilities of 2 and 3
estimated from binding constants, we consider that product 3 on the
catalyst can be easily replaced with substrate 2. See the Supporting
Information.
(6) (a) Aida, T.; Inoue, S. J. Am. Chem. Soc. 1983, 105, 1304−1309.
(b) Kruper, W. J.; Dellar, D. V. J. Org. Chem. 1995, 60, 725−727.
(c) Paddock, R. L.; Hiyama, Y.; McKay, J. M.; Nguyen, S. T.
Tetrahedron Lett. 2004, 45, 2023−2026. (d) Srivastava, R.; Bennur, T.
H.; Srinivas, D. J. Mol. Catal. A: Chem. 2005, 226, 199−205. (e) Jin, L.;
Jing, H.; Chang, T.; Bu, X.; Wang, L.; Liu, Z. J. Mol. Catal. A: Chem.
2007, 261, 262−266. (f) Bai, D.; Wang, Q.; Song, Y.; Li, B.; Jing, H.
Catal. Commun. 2011, 12, 684−688. (g) Ahmadi, F.; Tangestaninejad,
S.; Moghadam, M.; Mirkhani, V.; Mohammadpoor-Baltork, I.;
Khosropour, A. R. Inorg. Chem. Commun. 2011, 14, 1489−1493.
(h) Wang, M.; She, Y.; Zhou, X.; Ji, H. Chin. J. Chem. Eng. 2011, 19,
446−451. (i) Anderson, C. E.; Vagin, S. I.; Xia, W.; Jin, H.; Rieger, B.
Macromolecules 2012, 45, 6840−6849. (j) Wu, W.; Qin, Y.; Wang, X.;
Wang, F. J. Polym. Sci. A: Polym. Chem. 2013, 51, 493−498.
(7) Ema, T.; Miyazaki, Y.; Koyama, S.; Yano, Y.; Sakai, T. Chem.
Commun. 2012, 48, 4489−4491.
(22) Kubat
́ ́ ́
, P.; Lang, K.; Anzenbacher, P., Jr.; Jursíkova, K.; Kral, V.;
Ehrenberg, B. J. Chem. Soc., Perkin Trans. 1 2000, 933−941.
(23) (a) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785−
789. (b) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652.
(24) (a) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972,
56, 2257−2261. (b) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta
1973, 28, 213−222.
̌
(25) (a) Miertus, S.; Scrocco, E.; Tomasi, J. Chem. Phys. 1981, 55,
(8) Mizutani, T.; Ema, T.; Tomita, T.; Kuroda, Y.; Ogoshi, H. J. Am.
Chem. Soc. 1994, 116, 4240−4250.
117−129. (b) Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 106,
5151−5158. (c) Cammi, R.; Mennucci, B.; Tomasi, J. J. Phys. Chem. A
2000, 104, 5631−5637.
(9) (a) Konsler, R. G.; Karl, J.; Jacobsen, E. N. J. Am. Chem. Soc.
1998, 120, 10780−10781. (b) Breinbauer, R.; Jacobsen, E. N. Angew.
Chem., Int. Ed. 2000, 39, 3604−3607. (c) Ema, T.; Tanida, D.;
Matsukawa, T.; Sakai, T. Chem. Commun. 2008, 957−959.
(26) We confirmed that the polarity of the solvent had only a small
impact on the energy profile. See the Supporting Information.
(27) (a) Reed, A. E.; Weinhold, F. J. Chem. Phys. 1983, 78, 4066−
4073. (b) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys.
1985, 83, 735−746.
(10) Ema, T.; Miyazaki, Y.; Taniguchi, T.; Takada, J. Green Chem.
2013, 15, 2485−2492.
I
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(28) Frisch, M. J.; et al. Gaussian 09, Revision B.1; Gaussian, Inc.:
Wallingford, CT, 2009.
J
dx.doi.org/10.1021/ja507665a | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX