A highly active and recyclable catalytic system for CO2/propylene oxide copolymerization

Article abstract of DOI:10.1002/anie.200801852

(Figure Presented) Converting CO₂ into polymer: A catalytic system that produces a high molecular weight CO₂/propylene oxide copolymer with high activity and selectivity is disclosed. After filtration through silica gel, elution of the catalyst leaves a solid phase with a negligible metal residue (see picture). The catalyst can be reused without significant loss of performance.

Full text of DOI:10.1002/anie.200801852

Communications
DOI: 10.1002/anie.200801852
Copolymerization
A Highly Active and Recyclable Catalytic System for CO2/Propylene
Oxide Copolymerization**
Sujith S, Jae Ki Min, Jong Eon Seong, Sung Jea Na, and Bun Yeoul Lee*
Copolymerization of CO2 and epoxide catalyzed by metal
complexes has drawn much attention recently.[1] In the initial
report by Inoue and co-workers in 1969 the catalyst activity
was poor (turnover number (TON): 6; turnover frequency
(TOF): 0.12 hÀ1),[2] but by the early 2000s it had been
significantly improved by using binary catalytic systems of
[Co(salen)][3] or [Cr(salen)] complexes[4] with an onium salt or
base
(1;
salen = N,N’-bis(salicylidene)ethylenediamine,
[PPN]+ ClÀ = bis(triphenylphosphine)iminium chloride) or
b-diketiminato zinc complexes.[5] It was proposed that the
two components or the two zinc centers participated in the
propagation reaction, and therefore binding of the two metal
centers or the two components further improved catalysis to
the maximum reported TON and TOF of 14500 and 4600 hÀ1,
respectively.[6,7] These values, which were achieved with 2 in
the copolymerization of CO2 and propylene oxide, are still
low enough to warrant further improvement,[7] because low
activity means higher catalyst cost and higher levels of
catalyst-derived metal residue in the resin.[8] This metal
residue either colors the resin or causes toxicity. For the TON
of 14500 attained with 2, the residual cobalt level in the resin
reached 40 ppm unless it was removed.[9] Herein, we report a
highly active catalytic system for CO2/propylene oxide
copolymerization and an efficient recovery strategy of the
catalyst.
Scheme 1. Synthesis of catalysts. Reagents and conditions: a) Cl-
=
(CH2)3C(O)Cl/AlCl3 for A=H ; Br, then tBuLi (3 equiv)/[Cl(CH2)3]2C O
2
for A=(CH2)3Cl; b) H2/Pd on C; c) paraformaldehyde/MgCl2; d) NaI;
e) trans-1,2-diaminocyclohexane; f) Bu3N then AgBF4; g) Co(OAc)2
then O2/2,4-(NO2)2C6H3OH; h) 2,4-(NO2)2C6H3ONa.
chlorobutyryl chloride and subsequent hydrogenation
attached a (CH2)4Cl group in the para position. In a similar
manner, attack by 1,7-dichloroheptan-4-one on lithium 2-
alkyl-4-lithiophenolate yielded a tertiary benzylic alcohol,
and subsequent hydrogenation attached CH[(CH2)3Cl]2.
After formylation, the chloro substituent, that was not
susceptible to nucleophilic substitution with tributylamine,
was transformed into a more reactive iodo substituent.
Formation of the quaternary ammonium salt occurred in
nearly quantitative yields after generation of the salen-type
ligand. Because the halide ion interfered with the metalation
reaction, we replaced it with inert BF4À. Following metalation
by a routine method, an active 2,4-dinitrophenolate anion
replaced this BF4À ion.[3a]
We prepared various cobalt–salen complexes by varying
either the ortho substituent or the number of attached
quaternary ammonium groups (Scheme 1). Thus, the Frie-
del–Crafts acylation of ortho-substituted phenol using 4-
We screened the newly prepared complexes for CO2/
propylene oxide copolymerization under the following con-
ditions: [propylene oxide (PO)]/[catalyst (cat.)] = 25000,
808C, and CO2 pressure 2.0–1.7 MPa (Table 1). The polymer-
ization was terminated before 25% conversion was reached,
because after this time the solution became highly viscous,
thus making stirring and diffusion difficult. The identity of the
ortho substituent strongly influenced the activity. Contrary to
other reports,[3] the bulky tert-butyl group was not the best
choice in this study, and replacing it with a small methyl group
significantly enhanced TOF. The TOF also increased dramat-
[*] S. S, J. K. Min, J. E. Seong, S. J. Na, Prof. B. Y. Lee
Department of Molecular Science and Technology
Ajou University, Suwon 443-749 (Korea)
Fax: (+82)31-219-2394
E-mail: bunyeoul@ajou.ac.kr
Homepage: http://www.ajou.ac.kr/~polylab
[**] This work was supported a Korea Research Foundation Grant
funded by the Korean Government (MOEHRD, Basic Research
Promotion Fund; KRF-2005-015-C00233).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200801852.
7306
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 7306 –7309

Angewandte
Chemie
Table 1: CO2/propylene oxide copolymerization.[a]
entry 12). The molecular weight
distribution was narrow (Mw/Mn =
1.18). When polymerization was
carried out for an additional
30 min (90 min total) at [PO]/
[cat.] = 100000, the TON increased
to 27000 (27% conversion), but
with concomitant formation of
cyclic carbonate (92% selectivity;
Table 1, entry 9). We attributed the
lowering of the selectivity to a
possible back-bite mechanism in
the highly viscous polymerization
solution at a later stage in the run.
At [PO]/[cat.] = 150000, we did not
observe CO2 consumption until 3 h
after CO2 injection. After that, the
polymerization started and the
TOF (12400) was also reduced
(Table 1, entry 10).
[c]
Entry
[PO]/[cat.]
(cat.)
t [h]
TOF [hÀ1
]
TON
Selectivity[b]
Mn ”10À3
Mw/Mn
1
2
3
4
5
25000 (3)
25000 (4)
25000 (5)
25000 (6)
25000 (7)
25000 (8)
50000 (8)
100000 (8)
100000 (8)
150000 (8)
25000 (2)
2000 (1)
6.0
1.0
1.0
2.0
0.67
0.25
0.50
1.0
1.5
2.5
270
3100
3000
1300
7900
26000
26000
22000
18000
12400
3300
1600
3100
3000
2500
5300
70
95
89
42
11
36
38
76
114
208
285
225
175
71
1.20
1.49
1.52
1.34
1.32
1.29
1.20
1.18
1.28
1.20
1.25
1.01
84
>99
>99
>99
>99
92
96
94
97
6
7
6400
13000
22000
27000
31000
3300
8[d]
9[d]
10[d]
11[e]
12[f]
1.0
0.70
1400
980
26
[a] Polymerization conditions: propylene oxide (10.0 g, 172 mmol), CO2 pressure 2.0–1.7 MPa, 808C.
[b] Selectivity for the polycarbonate over the cyclic carbonate in % units. [c] Molecular weight
distribution: Mw =weight-average molecular weight; Mn =number-average molecular weight. [d] An
induction time (ca. 30 min for entries 8 and 9, ca. 3 h for entry 10) was observed, which was excluded in
calculating TOF. [e] Data from reference [7]. [f] Data from reference [3a].
Even at a TON as high as
22000, which corresponds to
38 (kgpolymer)(gCo)À1,
the
cobalt level in the isolated polymer
ically as more ammonium salt units were attached. By
applying these principles, we obtained the best catalytic
performance with 8 (Scheme 1), for which about 25%
conversion was obtained in 15 min at [PO]/[cat.] = 25000
(Table 1, entry 6). This result corresponds to TOF =
26000 hÀ1 or 2.7 ” 106 (gpolymer)molÀ1 hÀ1, one order of
magnitude higher than the values obtained with 1 and 2
(1400 and 3300 hÀ1, respectively; Table 1, entries 11 and 12).
Another advantage of 8 was its high selectivity for
copolymer formation over cyclic carbonate (> 99%). In
contrast, 2–6 exhibited selectivities below 95%, and the
binary [Co(salen)]/PPNCl system (1) produced exclusively
cyclic carbonate under the same conditions of [PO]/[cat.] =
25000 and 808C.[7] Our results also demonstrated that binding
the quaternary ammonium salts to the salen unit crucially
affected activity and selectivity, because the catalytic system
of 5/2[NBu4]+[2,4-(NO2)2C6H3O]À exhibited significantly
lower activity (TOF, 4300 hÀ1) and selectivity (75%) than 8
reached 26 ppm unless the catalyst was separated; the
residual cobalt complex made the isolated polymer a light
yellowish color. When the light yellow polymerization
solution prepared with 8 was passed through a short pad of
silica gel (230–400 mesh, Merck), the top layer of the pad
trapped the colored catalyst, thus yielding a colorless polymer
solution (Scheme 2). Importantly, this simple separation was
possible because the quaternary ammonium salts were bound
to the salen unit. When the polymerization solution prepared
with an unbound binary catalytic system [Co(salen)]/
PPN+ClÀ (1) was filtered through the silica-gel pad, the
[Co(salen)] complex passed through it, which gave the filtrate
a deep red color. It was proposed that the 2,4-dinitropheno-
late anion attacking the coordinated epoxide initiated poly-
mer-chain growth. If this scenario was the case, after
polymerization the 2,4-dinitrophenolate anion would
become a carbonate anion or alkoxide anion attaching a
polymer chain, either of which might be protonated on a silica
surface. This protonation allowed it to pass through the silica-
gel pad while leaving the cobalt complex on the silica surface.
Afterward, we were able to recover the catalyst from the
silica-gel pad and reuse it. Nothing was extracted when the
collected composite of silica gel and salen–cobalt complex
was dispersed in CH3OH, but the color of the solution rapidly
became reddish brown when excess NaBF4 was added to the
dispersion (Scheme 2). A possible salt exchange reaction
(TOF, 26000 hÀ1
; selectivity, > 99%). All copolymers
obtained with 3–8 were strictly alternating copolymers that
did not contain any ether linkages.
To increase the TON, we should be able to reach the same
level of conversion (ca. 25%) at a higher [PO]/[cat.] ratio by
running the polymerization for a longer time. Doubling the
[PO]/[cat.] ratio from 25000 to 50000 did not reduce the
catalytic performance of 8 (TOF, 26000 hÀ1), and we could
double the TON (13000) by doubling the polymerization time
from 15 to 30 min (Table 1, entry 7). The system was also
À
between the surface siloxy anion and the inert BF4 might
have solubilized the complex in the CH3OH phase. We did not
highly active even at [PO]/[cat.] = 100000 (TOF, 22000 hÀ1
;
observe 2,4-dinitrophenolate signals in the 1H NMR spectrum
À
TON, 22000), under which conditions 2 completely lost its
activity. We observed some induction time (ca. 30 min) in this
extreme condition. Through realizing such a high TON as
22000, we obtained a very high molecular weight copolymer
(Mn = 285000), one order of magnitude higher than that
obtained with the binary system 1 (Mn = 26000; Table 1,
of the recovered BF4 salt of the cobalt complex; further-
more, the white color of the recovered silica gel indicated that
it did not contain 2,4-dinitrophenolate anions, which would
yield a yellowish color. These observations implied that all
five 2,4-dinitrophenolate anions originally present in 8 were
used up during polymerization. Treatment of the extracted
Angew. Chem. Int. Ed. 2008, 47, 7306 –7309ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7307

Communications
Table 2: CO2/propylene oxide copolymerization on the 90 g scale using
recovered catalyst 8.[a]
Run t [min]
TON
Selectivity Mn ”10À3 Mw/Mn Recovery
yield [%]
1st
2nd
3rd
4th
5th
65
85
85
80
90
21000
21000
21000
21000
20000
>99
98
97
98
97
296
172
176
190
210
1.19
1.41
1.34
1.22
1.21
88
89
87
85
86
[a] Polymerization conditions: propylene oxide (250 g), [PO]/[cat.]=
100000, 70–778C, CO2 pressure 2.0 MPa.
To isolate the copolymer, the propylene oxide was simply
removed from the filtrate by distillation. The distillate was
pure propylene oxide, which could also be reused for
subsequent polymerization after drying with CaH2. The
isolated polymer was white, and inductively coupled plasma
mass analysis indicated that it contained negligible amounts
of cobalt residue (1–2 ppm).
In conclusion, we have presented a highly active catalytic
system (TON > 20000; TOF > 20000 hÀ1) for CO2/propylene
oxide copolymerization to selectively produce a very high
molecular weight copolymer (Mn = 300000). The catalyst can
be readily separated by filtration through a short pad of silica
gel to yield a resin containing negligible amounts of metal
residue (1–2 ppm). The separated catalyst can be reused
without significant loss of catalytic performance. All these
features will encourage the commercialization of these CO2/
propylene oxide copolymers.
Received: April 21, 2008
Published online: August 11, 2008
Scheme 2. Strategy for catalyst separation and recovery.
Keywords: carbon dioxide · catalyst recycling · cobalt ·
.
copolymerization · homogeneous catalysis
complex with two equivalents of 2,4-dinitrophenol and excess
sodium 2,4-dinitrophenolate in CH2Cl2 afforded a recovered
complex, with a 1H NMR spectrum (see the Supporting
Information) and catalytic performance (TOF = 24000 hÀ1
;
[1] For reviews, see: a) D. J. Darensbourg, Chem. Rev. 2007, 107,
2388; b) G. W. Coates, D. R. Moore, Angew. Chem. 2004, 116,
6784; Angew. Chem. Int. Ed. 2004, 43, 6618.
[2] a) S. Inoue, H. Koinuma, T. Tsuruta, J. Polym. Sci. Part B 1969, 7,
287; b) S. Inoue, H. Koinuma, T. Tsuruta, Makromol. Chem. 1969,
130, 210.
[3] a) X.-B. Lu, L. Shi, Y.-M. Wang, R. Zhang, Y.-J. Zhang, X.-J.
Peng, Z.-C. Zhang, B. Li, J. Am. Chem. Soc. 2006, 128, 1664;
b) C. T. Cohen, T. Chu, G. W. Coates, J. Am. Chem. Soc. 2005, 127,
10869; c) Z. Qin, C. M. Thomas, S. Lee, G. W. Coates, Angew.
Chem. 2003, 115, 5642; Angew. Chem. Int. Ed. 2003, 42, 5484.
[4] a) D. J. Darensbourg, A. L. Phelps, N. L. Gall, L. Jia, Acc. Chem.
Res. 2004, 37, 836; b) D. J. Darensbourg, P. Bottarelli, J. R.
Andreatta, Macromolecules 2007, 40, 7727.
selectivity > 99% at [PO]/[cat.] = 50000) that were almost
identical to those of fresh, virgin 8. We presume that the
À
extract is a dihydroxycobaltate complex bearing three BF4
groups (Scheme 2), because two equivalents of 2,4-dinitro-
phenol were required to convert the extract back to an active
catalyst.
In the scale-up reaction, the copolymer could be prepared
on the 90 g scale at [PO]/[cat.] = 100000 (95 mg 8/250 g PO)
with almost the same catalytic performance as that attained in
a small-scale reaction (first run in Table 2). Again, we
repeatedly used the recovered catalyst without significant
loss in catalytic activity. In the repeated polymerizations,
some reductions of the selectivity from > 99% to 97–98%
and of the molecular weights from 300000 to about 200000
were observed. The recovery yields were 85–89%; we
attributed some of this loss to the necessarily incomplete
transfer of the polymerization solution from the reactor to the
filtration apparatus. We supplemented the lost catalyst with
fresh material for each subsequent batch.
[5] D. R. Moore, M. Cheng, E. B. Lobkovsky, G. W. Coates, J. Am.
Chem. Soc. 2003, 125, 11911.
[6] a) B. Y. Lee, H. Y. Kwon, S. Y. Lee, S. J. Na, S.-i. Han, H. Yun, H.
Lee, Y.-W. Park, J. Am. Chem. Soc. 2005, 127, 3031; b) T. Bok, H.
Yun, B. Y. Lee, Inorg. Chem. 2006, 45, 4228; c) K. Nakano, T.
Kamada, K. Nozaki, Angew. Chem. 2006, 118, 7432; Angew.
Chem. Int. Ed. 2006, 45, 7274.
[7] E. K. Noh, S. J. Na, S. S, S.-W. Kim, B. Y. Lee, J. Am. Chem. Soc.
2007, 129, 8082.
7308
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 7306 –7309

Angewandte
Chemie
[8] Extensive studies have been performed on the catalyst separation
for atom-transfer radical polymerization: a) S. C. Hong, K.
Matyjaszewski, Macromolecules 2002, 35, 7592; b) D. E. Berg-
breiter, P. N. Hamilton, N. M. Koshti, J. Am. Chem. Soc. 2007, 129,
10666; c) Z. Shen, Y. Chen, H. Frey, S.-E. Stiriba, Macromolecules
2006, 39, 2092; d) S. Faucher, P. Okrutny, S. Zhu, Macromolecules
2006, 39, 3; e) J. V. Nguyen, C. W. Jones, J. Catal. 2005, 232, 276.
[9] Recently, others have reported a catalytic recovery strategy using
a biphasic system for CO2/cyclohexene oxide copolymerization
catalyzed by [Cr(salen)]/PPNCl, but some catalyst was lost into
the polymer phase, thus yielding a polymer containing 39ppm Cr:
C. Hongfa, J. Tian, J. Andreatta, D. J. Darensbourg, D. E.
Bergbreiter, Chem. Commun. 2008, 975.
Angew. Chem. Int. Ed. 2008, 47, 7306 –7309ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 7309