Catalytic double carbonylation of epoxides to succinic anhydrides: Catalyst discovery, reaction scope, and mechanism

Article abstract of DOI:10.1021/ja066901a

The first catalytic method for the efficient conversion of epoxides to succinic anhydrides via one-pot double carbonylation is reported. This reaction occurs in two stages: first, the epoxide is carbonylated to a β-lactone, and then the β-lactone is subsequently carbonylated to a succinic anhydride. This reaction is made possible by the bimetallic catalyst [(CITPP)Al(THF)₂]⁺[Co(CO)₄]^- (1; CITPP = meso-tetra(4-chlorophenyl)porphyrinato; THF = tetrahydrofuran), which is highly active and selective for both epoxide and lactone carbonylation, and by the identification of a solvent that facilitates both stages. The catalysis is compatible with substituted epoxides having aliphatic, aromatic, alkene, ether, ester, alcohol, nitrile, and amide functional groups. Disubstituted and enantiomerically pure anhydrides are synthesized from epoxides with excellent retention of stereochemical purity. The mechanism of epoxide double carbonylation with 1 was investigated by in situ IR spectroscopy, which reveals that the two carbonylation stages are sequential and non-overlapping, such that epoxide carbonylation goes to completion before any of the intermediate β-lactone is consumed. The rates of both epoxide and lactone carbonylation are independent of carbon monoxide pressure and are first-order in the concentration of 1. The stages differ in that the rate of epoxide carbonylation is independent of substrate concentration and first-order in donor solvent, whereas the rate of lactone carbonylation is first-order in lactone and inversely dependent on the concentration of donor solvent. The opposite solvent effects and substrate order for these two stages are rationalized in terms of different resting states and rate-determining steps for each carbonylation reaction.

Full text of DOI:10.1021/ja066901a

Published on Web 03/31/2007
Catalytic Double Carbonylation of Epoxides to Succinic
Anhydrides: Catalyst Discovery, Reaction Scope,
and Mechanism
John M. Rowley, Emil B. Lobkovsky, and Geoffrey W. Coates*
Contribution from the Department of Chemistry and Chemical Biology, Baker Laboratory,
Cornell UniVersity, Ithaca, New York 14853-1301
Received September 25, 2006; E-mail: gc39@cornell.edu
Abstract: The first catalytic method for the efficient conversion of epoxides to succinic anhydrides via one-
pot double carbonylation is reported. This reaction occurs in two stages: first, the epoxide is carbonylated
to a â-lactone, and then the â-lactone is subsequently carbonylated to a succinic anhydride. This reaction
is made possible by the bimetallic catalyst [(ClTPP)Al(THF)2]+[Co(CO)4]- (1; ClTPP ) meso-tetra(4-
chlorophenyl)porphyrinato; THF ) tetrahydrofuran), which is highly active and selective for both epoxide
and lactone carbonylation, and by the identification of a solvent that facilitates both stages. The catalysis
is compatible with substituted epoxides having aliphatic, aromatic, alkene, ether, ester, alcohol, nitrile, and
amide functional groups. Disubstituted and enantiomerically pure anhydrides are synthesized from epoxides
with excellent retention of stereochemical purity. The mechanism of epoxide double carbonylation with 1
was investigated by in situ IR spectroscopy, which reveals that the two carbonylation stages are sequential
and non-overlapping, such that epoxide carbonylation goes to completion before any of the intermediate
â-lactone is consumed. The rates of both epoxide and lactone carbonylation are independent of carbon
monoxide pressure and are first-order in the concentration of 1. The stages differ in that the rate of epoxide
carbonylation is independent of substrate concentration and first-order in donor solvent, whereas the rate
of lactone carbonylation is first-order in lactone and inversely dependent on the concentration of donor
solvent. The opposite solvent effects and substrate order for these two stages are rationalized in terms of
different resting states and rate-determining steps for each carbonylation reaction.
Introduction and Background
Substituted succinic anhydrides have previously been syn-
thesized by a number of methods,9 most often by the dehydration
of the corresponding diacid or from maleic anhydride via Diels-
Alder or Ene reactions.10 They have also been made by metal-
catalyzed carbonylation of alkynes,11 alkenoic acids,12 and
lactones;13,14 however, most of these catalytic reactions proceed
either in low yield, with significant side products, or without
demonstrating substrate generality or product stereochemical
purity. Thus, the development of more efficient and stereose-
lective syntheses remains an important goal.
Over the past 6 years, our group has developed a class of
well-defined bimetallic catalysts of the general type [Lewis
acid]+[Co(CO)4]- for the ring-expanding carbonylation of
strained heterocycles.14-22 Within this class of catalysts, we have
found that the identity of metal and ligand in the Lewis acid
component has a profound effect on the activity and substrate
scope of carbonylation. In particular, we recently reported that
Succinic anhydrides and their derivatives have many applica-
tions in organic and polymer chemistry.1 For example, their
copolymerization with epoxides2 or diols3 yields biodegradable
polyesters.4 Anhydrides are also useful synthetic intermediates,5
readily ring opened to diacids or other succinate derivatives;
some examples of succinates include biologically active natural
products,6 pharmaceuticals,7 and metalloprotease inhibitors.8
(1) Fumagalli, C. In Kirk-Othmer Encyclopedia of Chemical Technology;
Kroschwitz, J. I., Howe-Grant, M., Eds.; John Wiley & Sons: New York,
1997; Vol. 22, pp 1074-1102.
(2) Maeda, Y.; Nakayama, A.; Kawasaki, N.; Hayashi, K.; Aiba, S.; Yamamoto,
N. Polymer 1997, 38, 4719-4725.
(3) Ranucci, E.; Liu, Y.; Lindblad, M. S.; Albertsson, A. C. Macromol. Rapid
Commun. 2000, 21, 680-684.
(4) Okada, M. Prog. Polym. Sci. 2002, 27, 87-133.
(5) For example, see: (a) Bercot, E. A.; Rovis, T. J. Am. Chem. Soc. 2005,
127, 247-254. (b) Chen, Y.; McDaid, P.; Deng, L. Chem. ReV. 2003, 103,
2965-2983. (c) Stoncius, A.; Nahrwold, M.; Sewald, N. Synthesis 2005,
1829-1837.
(6) Wilkinson, R. A.; Strobel, G.; Stierle, A. J. Nat. Prod. 1999, 62, 358-
360.
(11) For example, see: Chiusoli, G. P.; Costa, M.; Cucchia, L.; Gabriele, B.;
Salerno, G.; Veltri, L. J. Mol. Catal. A: Chem. 2003, 204, 133-142 and
references therein. In related chemistry, alkynes have been carbonylated
to maleic anhydrides: Zargarian, D.; Alper, H. Organometallics 1991, 10,
2914-2921.
(12) For example, see: Osakada, K.; Doh, M. K.; Ozawa, F.; Yamamoto, A.
Organometallics 1990, 9, 2197-2198.
(13) Mori, Y.; Tsuji, J. Bull. Chem. Soc. Jpn. 1969, 42, 777-779.
(14) Getzler, Y. D. Y. L.; Kundnani, V.; Lobkovsky, E. B.; Coates, G. W.
J. Am. Chem. Soc. 2004, 126, 6842-6843.
(7) (a) Yamaguchi, T.; Yanagi, T.; Hokari, H.; Mukaiyama, Y.; Kamijo, T.;
Yamamoto, I. Chem. Pharm. Bull. 1998, 46, 337-340. (b) Bergmeier,
S. C.; Ismail, K. A.; Arason, K. M.; McKay, S.; Bryant, D. L.; McKay,
D. B. Bioorg. Med. Chem. Lett. 2004, 14, 3739-3742.
(8) Whittaker, M.; Floyd, C. D.; Brown, P.; Gearing, A. J. H. Chem. ReV.
1999, 99, 2735-2776.
(9) Arason, K. M.; Bergmeier, S. C. Org. Prep. Proced. Int. 2002, 34, 337-
366 and references therein.
(10) Mikami, K.; Shimizu, M. Chem. ReV. 1992, 92, 1021-1050.
9
4948
J. AM. CHEM. SOC. 2007, 129, 4948-4960
10.1021/ja066901a CCC: $37.00 © 2007 American Chemical Society

Carbonylation of Epoxides to Succinic Anhydrides
A R T I C L E S
Scheme 1. One-Pot Double Carbonylation of Epoxides to
Succinic Anhydrides
have investigated the mechanism of the catalytic interaction of
[Lewis acid]+[Co(CO)4]- with other substrates.28 Thus, we
undertook a mechanistic investigation of double carbonylation
with the following goals: (1) to compare the mechanism of
epoxide to lactone carbonylation in this system with that of a
previously reported system,22 (2) to elucidate the mechanism
and rate-determining step of lactone carbonylation, and (3) to
understand the interplay between these reactions in double
carbonylation, specifically the important role that solvent plays
in these bimetallic [Lewis acid]+[Co(CO)4]- catalytic systems.
one such catalyst can carbonylate â-lactones to succinic
anhydrides in high yields while preserving stereochemical
purity.14 Given the many syntheses of enantiomerically pure
epoxides23,24 and the recent advances in epoxide carbonylation
to â-lactones, subsequent carbonylation of these lactones
constitutes a versatile two-step method for the stereoselective
synthesis of succinic anhydrides (Scheme 1). However, this
method would be far more synthetically useful if the two steps
could be consolidated, eliminating isolation and rigorous
purification of toxic lactone intermediates, saving time and
catalyst, and increasing overall yield. One-pot double carbo-
nylations are known;25 however, there are very few instances
with an epoxide as the substrate. A rare example is the
conversion of styrene oxide to the enol tautomer of 4,5-dihydro-
4-phenylfuran-2,3-dione.26 Our originally reported lactone car-
bonylation catalyst, [(salph)Al(THF)2]+[Co(CO)4]- (2; salph )
N,N′-bis(3,5-di-tert-butylsalicylidene)-1,2-phenylenediamine; THF
) tetrahydrofuran), produced trace amounts of succinic anhy-
drides directly from epoxides,14 but we were unable to obtain
anhydrides cleanly and in good yields with this or any of our
previously reported carbonylation catalysts (vide infra). Herein,
we report on [(ClTPP)Al(THF)2]+[Co(CO)4]- (1; ClTPP )
meso-tetra(4-chlorophenyl)porphyrinato), the first highly active
and selective catalyst for the one-pot double carbonylation of
epoxides to succinic anhydrides.
Table 1. Screening for One-Pot Double Carbonylation of 1-Butene
Oxidea
product distribution (%)
entry
catalyst
solvent
lactone
polymer
anhydride
1
2
3
4
5
6
2
2
3
3
1
1
neat
toluene
neat
toluene
neat
toluene
99
58b
93
67
78
7
33
22
99
a Reaction conditions: 1.0 mol % catalyst, [epoxide] ) 1.8 M in toluene
or neat, 850 psi CO, 6 h, 60 °C. Product distribution determined by 1H
NMR spectra of crude reaction. b Remainder is starting epoxide.
Chart 1. Epoxide Carbonylation Catalysts
In the course of optimizing the double carbonylation of
epoxides, we found that both carbonylation stages were rapid,
yet none of the lactone intermediate was consumed until all of
the epoxide had first been carbonylated, and that solvent greatly
impacted the rate of each step. We16,17,22 and others27 have
studied the mechanism of epoxide carbonylation, but no studies
Results and Discussion
(15) For a review, see: Church, T. L.; Getzler, Y. D. Y. L.; Byrne, C. M.;
Coates, G. W. Chem. Commun. 2007, 657-674.
Catalyst Discovery. Our initial attempts to effect clean one-
pot double carbonylation focused on the catalysts and conditions
we have previously reported for ring-expansive epoxide and
lactone carbonylation. The results of these studies are sum-
marized in Table 1, entries 1-4. Although [(salph)Al(THF)2]+-
[Co(CO)4]- (2, Chart 1) is the only catalyst reported for both
epoxide16 and â-lactone14 carbonylation, these two independent
reactions were found to be orthogonal, such that the reaction
conditions, particularly solvent (vide infra), which facilitated
the first carbonylation, severely limited the second, and vice
(16) Getzler, Y. D. Y. L.; Mahadevan, V.; Lobkovsky, E. B.; Coates, G. W.
J. Am. Chem. Soc. 2002, 124, 1174-1175.
(17) Mahadevan, V.; Getzler, Y. D. Y. L.; Coates, G. W. Angew. Chem., Int.
Ed. 2002, 41, 2781-2784.
(18) Getzler, Y. D. Y. L.; Mahadevan, V.; Lobkovsky, E. B.; Coates, G. W.
Pure Appl. Chem. 2004, 76, 557-564.
(19) Schmidt, J. A. R.; Mahadevan, V.; Getzler, Y. D. Y. L.; Coates, G. W.
Org. Lett. 2004, 6, 373-376.
(20) Schmidt, J. A. R.; Lobkovsky, E. B.; Coates, G. W. J. Am. Chem. Soc.
2005, 127, 11426-11435.
(21) Kramer, J. W.; Lobkovsky, E. B.; Coates, G. W. Org. Lett. 2006, 8, 3709-
3712.
(22) Church, T. L.; Getzler, Y. D. Y. L.; Coates, G. W. J. Am. Chem. Soc.
2006, 128, 10125-10133.
(23) (a) Yudin, A. K., Ed. Aziridines and Epoxides in Organic Synthesis; Wiley-
VCH: Weinheim, 2006 and references therein. (b) Gao, Y.; Hanson,
R. M.; Klunder, J. M.; Ko, S. Y.; Masamune, H.; Sharpless, K. B. J. Am.
Chem. Soc. 1987, 109, 5765-5780. (c) Shi, Y. Acc. Chem. Res. 2004, 37,
488-496.
(27) (a) Molnar, F.; Luinstra, G. A.; Allmendinger, M.; Rieger, B. Chem.-Eur.
J. 2003, 9, 1273-1280. (b) Allmendinger, M.; Eberhardt, R.; Luinstra,
G. A.; Molnar, F.; Rieger, B. Z. Anorg. Allg. Chem. 2003, 629, 1347-
1352. (c) Allmendinger, M.; Zintl, M.; Eberhardt, R.; Luinstra, G. A.;
Molnar, F.; Rieger, B. J. Organomet. Chem. 2004, 689, 971-979. (d)
Stirling, A.; Iannuzzi, M.; Parrinello, M.; Molnar, F.; Bernhart, V.; Luinstra,
G. A. Organometallics 2005, 24, 2533-2537.
(24) Schaus, S. E.; Brandes, B. D.; Larrow, J. F.; Tokunaga, M.; Hansen,
K. B.; Gould, A. E.; Furrow, M. E.; Jacobsen, E. N. J. Am. Chem. Soc.
2002, 124, 1307-1315.
(25) For a review, see: des Abbayes, H.; Salaun, J. Y. J. Chem. Soc., Dalton
Trans. 2003, 1041-1052.
(26) Alper, H.; Arzoumanian, H.; Petrignani, J. F.; Saldana-Maldonado, M.
J. Chem. Soc., Chem. Commun. 1985, 340-341.
(28) Mechanisms for lactone carbonylation have been proposed, but not studied
in detail.13,14
9
J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007 4949

A R T I C L E S
Rowley et al.
versa. For example, whereas lactone formation was facile in
neat epoxide (entry 1) or ether solvents, nonpolar conditions
that favor anhydride formation14 severely reduced epoxide
carbonylation (entry 2). [(OEP)Cr(THF)2]+[Co(CO)4]- (3, Chart
1; OEP ) octaethylporphyrinato),20 the most active catalyst
reported for epoxide carbonylation, was considered with the
hope that inhibition by nonpolar solvent would be mediated by
the exceptional activity of the catalyst. Under typical conditions
and catalyst loadings (<0.03 mol %), â-lactone was the only
product observed for this system. However, significantly
increasing the amount of catalyst to 1 mol % resulted in very
rapid lactone formation, followed by slower conversion of
lactone to low molecular weight poly(lactone) and a small
amount of anhydride (entry 3). Even under favorable (nonpolar
and diluted) conditions, polymer remains the major product
(entry 4). Given the failure of these previously reported catalysts,
conditions, and permutations thereof, it was clear that, to effect
clean anhydride formation, the efficient double carbonylation
of epoxides would require either a new catalyst, a new solvent,
or both.
selectivity along with rapid conversion to anhydride.31
Based upon mechanistic studies of epoxide carbonylation with
2,22 we expected that solvent would significantly impact the
rate of reaction, and therefore we attempted epoxide double
carbonylation in a range of solvents. Conversion to succinic
anhydride varied widely with solvent, as illustrated by a
representative selection of solvents in Figure 2. Hexane was a
poor solvent, yielding mostly polymer. Acyclic monofunctional
ether solvents such as diethyl ether were effective for double
carbonylation, though polymer remained a significant byproduct.
Moderately coordinating, cyclic, or multifunctional ether sol-
vents resulted in the cleanest reactions. In particular, the double
carbonylation was distinctively rapid in 1,4-dioxane, and this
became the solvent of choice for future reactions. Strongly
coordinating solvents such as acetonitrile severely inhibited
lactone carbonylation (vide infra).
In the course of investigating the variation of metal and ligand
in the cationic portion of our catalysts, we synthesized a
series29 of catalysts with porphyrin aluminum cations such
as 1 (Chart 1). As expected, the solid-state structure of 1 (Figure
1) had several key features in common with previously reported
Figure 1. Solid-state structure of 1. Hydrogen atoms omitted for clarity.
Displacement ellipsoids drawn at 40% probability.
Figure 2. Solvent screen for epoxide double carbonylation. Reaction
conditions: 2.0 mmol of epoxide in 1.0 mL of solvent, 0.20 mol % 1, 850
psi CO, 2 h, 90 °C. Product distribution determined by 1H NMR spectrum
of crude reaction mixture. Epoxide was completely converted to lactone
(not shown) in all of these solvents; subsequent conversion of lactone to
anhydride and polymer is shown.
bimetallic carbonylation catalysts:16,18,20,21 a well-separated
ion pair with a tetrahedral cobaltate and a pseudo-octahedral
cationic metal, which is coordinated by a tetradentate ligand
and two axial molecules of THF. Upon screening this new
complex, we discovered that it rapidly catalyzed both epoxide
and lactone carbonylation, though some low-molecular-weight
polymer30 was formed in the absence of solvent (Table 1, entry
5). Addition of solvent resulted in the first clean, one-pot
double carbonylation of epoxide to succinic anhydride (entry
6).
With an optimum solvent for epoxide double carbonylation,
we returned to screen a number of new and previously reported
catalysts of the general form [(ligand)M(THF)2]+[Co(CO)4]-.
These catalysts were also more active in 1,4-dioxane than other
solvents; however, none showed greater activity than 1.29
Variation of substituents in the para position of the porphyrin
phenyl rings had little effect on the reactivity of the resulting
catalysts;29 however, the p-Cl complex (1) was chosen for future
study, as Cl was an economical para substituent, and this
substitution imparted greater crystallinity to the isolated com-
Reaction Optimization. With the first highly active and
selective catalyst for one-pot double carbonylation in hand, we
sought to optimize the reaction. At 40 °C, anhydrides were
formed cleanly using CO pressures as low as 100 psi. Increasing
the temperature to 90 °C greatly increased the rate of reaction,
but at high temperature and low pressure or in the absence of
CO, 1 catalytically isomerizes epoxides to ketones (eq 1).16,21
Thus, higher pressures (>200 psi) were employed to maintain
(31) Further reactions used 850 psi CO to avoid significant loss of pressure due
to CO consumption. At this pressure, ketone was only formed during the
delay after mixing the epoxide with 1 and before pressurization with CO,
but once the system was saturated with CO, ketone formation ceased.
Rearrangement to ketone was minimized by keeping the epoxide and
catalyst solution cold until pressurization with CO (small scale, screening),16
or eliminated entirely by partially pressurizing the reactor with CO prior
to epoxide addition (large scale, kinetics). Ketone formation was also
observed when insufficient stirring hindered the effective dissolution of
CO. Good selectivity (<1% ketone) was retained at pressures as low as 40
psi if the reaction was kept at room temperature until epoxide was converted
to lactone, then heated to convert lactone to anhydride.
(29) See Supporting Information for details.
(30) By gel permeation chromatography, Mn ) 6000 g/mol. Porphyrin Al
alkoxides and carboxylates are known polymerize lactones to polyesters,
see: Aida, T.; Inoue, S. Acc. Chem. Res. 1996, 29, 39-48.
9
4950 J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007

Carbonylation of Epoxides to Succinic Anhydrides
A R T I C L E S
Table 2. Double Carbonylation of Functionalized Epoxidesa
a Unless specified otherwise, reactions run for 3 h at 850 psi CO and 90 °C in 1,4-dioxane (1.8 M epoxide). b Yield of anhydride determined by 1H NMR
spectroscopy of crude reaction mixture and verified by internal standard; corresponding lactone and ketone31 were the only other products detected. Unoptimized
isolated yields are in parentheses. c [epoxide] ) 0.5 M. d [epoxide] ) 1.0 M, 24 h. e [epoxide] ) 1.0 M at 50 °C for 12 h.
plex. In summary, under optimized conditions, 1 effects the
double carbonylation of an epoxide to succinic anhydride with
unprecedented activity, yield, and selectivity.
groups were also tolerated (entries 16 and 17), but styrene oxide
(36) was run at lower temperature for a longer time to avoid
thermal decarboxylation32 of the lactone intermediate.33 Sub-
strates with more than one epoxide can undergo multiple double
carbonylations; for example, 1,2,7,8-diepoxyoctane (38) was
quadruply carbonylated to give the bis(succinic anhydride), 39,
in good yield.
1,2-Disubstituted epoxides34 were carbonylated to the corre-
sponding succinic anhydrides with retention of relative stere-
ochemistry via a double inversion mechanism (vide infra), in
which the stereocenter adjacent to the inserted carbonyl is
inverted with each carbonylation. cis-Epoxides were carbony-
lated to cis-anhydrides via the trans-lactones, and trans-epoxides
were carbonylated to the trans-anhydrides via cis-lactones (Table
3). At elevated temperatures, some epimerization of the product
occurred in the presence of catalyst;14 however, lower temper-
atures and thus longer reaction times afforded either cis- or
trans-succinic anhydrides (entries 1 and 2) with good stereo-
chemical purity. The cis-anhydride 41 was not formed as easily
as the trans-anhydride 43 and had to be run at lower concentra-
tion to avoid poly(lactone) formation. Increasing steric bulk
decreased the rate of carbonylation, but relative stereochemical
purity was preserved (entries 3 and 4).35
Substrate Scope. The versatility of this reaction was
demonstrated with a variety of functionalized epoxides (Table
2). Substrate/catalyst ratios were optimized to achieve near
complete conversion of 1-2 mmol of epoxide in 3 h at 90 °C;
longer reaction times were used if more than 2 mol % catalyst
was required to complete in 3 h. Conversion to anhydride is
reported to indicate the high selectivity of the reaction. In most
cases, isolation is straightforward, only requiring removal of
solvent and catalyst. As shown in Table 2, the reaction was
most facile for ethylene oxide (4) but still rapid for epoxides
with pendent alkyl groups (entries 2-6). Substrates with ether
side-chains were less active, but high yields and selectivity were
retained (entries 7-9). Surprisingly, good conversion to anhy-
dride was obtained with a seemingly incompatible unprotected
alcohol (entry 10). The product, 23, was stable enough to isolate
and characterize by IR, 1H and 13C NMR spectroscopy, though
after a few days at room temperature it decomposed from a
white solid to a yellow oil. Ester functionality was well tolerated
(entry 11), except in the specific case of glycidyl esters, where
isomerization to γ-lactone20 was favored over double carbony-
lation. Anhydrides with nitrile and amide side-chains were also
synthesized (entries 12 and 13), although, in the presence of an
amide, carbonylation of the lactone intermediate was very slow,
presumably due to competitive coordination (vide infra) with
the Lewis acid. Epoxides with pendent alkenes exhibited varied
activity depending on chain length (entries 14 and 15). Aryl
(32) Minato, T.; Yamabe, S. J. Org. Chem. 1983, 48, 1479-1483.
(33) Unlike the other examples, styrene oxide carbonylates at the more hindered,
benzylic carbon to give R-phenyl â-propiolactone.
(34) Although 1,1-disubstituted epoxides readily formed â-lactones, they were
unreactive for further carbonylation; see Table 5 (vide infra).
(35) Epoxide 46 is carbonylated to 47 via a ∼1:1 mixture of two regioisomeric
â-lactones.
9
J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007 4951

A R T I C L E S
Rowley et al.
Table 3. Double Carbonylation of Disubstituted Epoxidesa
Figure 3. Plot of the concentrations of BBL and MSA during double
carbonylation of PO. Reaction performed in 1,4-dioxane and monitored by
in situ IR spectroscopy (νCdO BBL ) 1827 cm-1 and νCdO MSA ) 1787
cm-1). [PO]0 ) 1.0 M, [1] ) 4.0 mM, PCO ) 850 psi, T ) 40 °C.
carbonylated; only then is there an abrupt change in the reaction,
and the lactone intermediate is immediately consumed in a first-
order decay, producing succinic anhydride as the product of a
second carbonylation. This two-stage behavior was observed
under all of the conditions studied. Even in solvents where
lactone carbonylation is much faster than epoxide carbonylation,
the lactone intermediate is not consumed while it is formed;
the two stages remain sequential.
a All reactions run for 24 h at 850 psi CO and 50 °C in 1,4-dioxane (1.8
M epoxide). b Conversion to anhydride and relative stereochemistry deter-
1
mined from the H NMR spectroscopy of crude reaction mixture; lactone
and ketone31 were the only other products detected. c [epoxide] ) 1.0 M in
1,4-dioxane.
As a starting point, a general mechanistic scheme for double
carbonylation is proposed as a combination of the separate
mechanisms we have proposed for epoxide22 and lactone14
carbonylation (Scheme 2). The cycle is initiated by substitution
of epoxide for solvent at the Lewis acidic cation [(ClTPP)Al-
(S)2]+ (S ) solvent, substrate, or product), which binds and
activates the epoxide. Subsequent ring-opening nucleophilic
attack by [Co(CO)4]- occurs at the less-hindered carbon,
presumably with concomitant loss of solvent,22 generating a
formally neutral, five-coordinate aluminum alkoxide. This is
followed by a rapid migratory insertion of CO into the cobalt-
alkyl bond and uptake of another CO to form the catalyst resting
state (A). Recoordination of solvent to Al facilitates release of
the alkoxide and subsequent attack on the cobalt-acyl, producing
lactone and regenerating [(ClTPP)Al(S)2]+[Co(CO)4]- (B). In
an analogous second cycle, the aluminum cation coordinates
lactone, activating it for nucleophilic attack by [Co(CO)4]-,
which occurs at the â-carbon. The resulting neutral aluminum-
carboxylate/cobalt-alkyl complex readily undergoes migratory
Succinic anhydrides and succinate derivatives with excellent
enantiomeric purity were readily accessed via double carbony-
lation (Table 4). (S)-Methylsuccinic anhydride, (S)-7, was
rapidly obtained from (R)-propylene oxide, (R)-6, with a slight
loss of enantiomeric purity at 90 °C (entry 1), but at 50 °C,
greater than 99% enantiomeric excess is maintained through
clean inversion of the stereocenter (entry 2). Likewise, larger
alkyl and functionalized epoxides were carbonylated to yield
anhydrides with g97% ee (entries 3 and 4). Thus, double
carbonylation yields anhydrides as useful intermediates for
asymmetric synthesis.
Mechanism. The double carbonylation of propylene oxide
(PO) to methylsuccinic anhydride (MSA) was monitored by in
situ IR spectroscopy. A representative plot of the concentrations
of â-butyrolactone (BBL) and MSA as a function of time reveals
that the reaction proceeds in two distinct stages (Figure 3). First,
catalytic carbonylation of epoxide results in the linear production
of â-lactone, which continues until all of the epoxide is
Table 4. Double Carbonylation of Enantiomerically Pure Epoxidesa
1
a Unless specified otherwise, reactions run for 24 h at 850 psi CO in 1,4-dioxane (1.8 M epoxide). b Conversion to anhydride determined by H NMR
spectroscopy of crude reaction mixture; lactone and ketone31 were the only other products detected. c Determined by chiral GC or HPLC of crude product.
d Reaction run for 3 h. e Product recrystallized from ether, 87% isolated yield.
9
4952 J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007

Carbonylation of Epoxides to Succinic Anhydrides
A R T I C L E S
Scheme 2. Proposed Catalytic Cycle of Double Carbonylation
Figure 4. Plot of the formation of BBL as a function of time. Reaction
performed in 1,4-dioxane and monitored by in situ IR spectroscopy
(νCdO ) 1827 cm-1). [PO]0 ) 1.0 M, [1] ) 4.0 mM, PCO ) 850 psi, T )
40 °C.
insertion of CO and coordination of another CO to form an
aluminum-carboxylate/cobalt-acyl species. Ring-closing attack
of the carboxylate on the cobalt-acyl, which may be aided by
solvent coordination to aluminum, forms anhydride and
regenerates B.
The fact that each carbonylation occurs with separate and
non-overlapping rate-determining steps, and that the â-lactone
intermediate is readily synthesized, means that each can be
studied as an independent reaction. In this paper, we will first
describe mechanistic studies of epoxide carbonylation, then
investigate the mechanism of lactone carbonylation. In light of
these studies, we will discuss why lactone consumption does
not begin until epoxide carbonylation is complete. Last, we will
address the interplay of these two reactions, and how the separate
conditions necessary for efficient catalysis of each stage can
be fulfilled in one system, thus enabling one-pot double
carbonylation.
Epoxide Carbonylation. Our group has recently completed
an in-depth investigation of the mechanism of epoxide carbo-
nylation with 2, yielding a detailed picture of the essential steps
and important intermediates.22 There are three main criteria with
which to compare the present system to this previously reported
work: order in substrate and catalyst, solvent effects, and the
catalyst resting state. Although the present study differs in
temperature (40 vs 25 °C), ligand (ClTPP vs salph), and epoxide
(PO vs 1,2-epoxybutane), we found that the basic mechanistic
picture remained the same.
When the carbonylation of PO was monitored by in situ IR
spectroscopy, the rate of BBL formation was observed to be
constant over the course of the reaction (Figure 4), indicating
that the rate is zero-order in epoxide. The reaction was
performed using five CO pressures in the range 200-1000 psi,
and no effect on the rate or selectivity was observed.29 Next,
the reaction was repeated using a range of catalyst concentrations
(Figure 5), and the rate of reaction exhibited a first-order
Figure 5. Initial rates of the formation of BBL and MSA as a function of
the concentration of 1. Reactions performed in 1,4-dioxane and monitored
with in situ IR spectroscopy (νCdO BBL ) 1827 cm-1 and νCdO MSA )
1787 cm-1). [PO]0 ) 1.0 M, PCO ) 850 psi, T ) 40 °C.
dependence on the concentration of 1. These orders in substrate
and catalyst are the same those as reported for 2.22
Solvent effects on the rate of epoxide carbonylation also
follow a similar trend to those reported for 2.22 In general,
coordinating solvents, such as ethers, accelerate the reaction,
whereas the carbonylation is slow in non-coordinating solvents
such as hexane and toluene (vide infra). Strongly coordinating
solvents, such as acetonitrile, inhibit the reaction, and we suspect
that substitution of the epoxide for solvent may become rate
determining in this case. The effect of solvent donicity can be
quantified by examining the reaction in mixtures of THF and
2,5-dimethyltetrahydrofuran (DMTHF).22,36,37 These solvents
have similar properties, except that DMTHF is a much weaker
donor due to sterics.36,37 A plot of initial rate versus concentra-
tion of THF in DMTHF shows that the rate of epoxide
carbonylation is first-order in the concentration of THF (Figure
6). Because the primary difference between THF and DMTHF
as solvents is donicity, the order in THF effectively represents
the order in donicity or Lewis base.22,36 Additionally, the rate
laws are the same in both THF/DMTHF and 1,4-dioxane.
Addition of THF or other moderately donating solvents ac-
celerates the reaction in 1,4-dioxane, and dilution of 1,4-dioxane
or THF with toluene slows the reaction. Thus, the following
expression can be written for the carbonylation of PO by 1 in
(36) Wax, M. J.; Bergman, R. G. J. Am. Chem. Soc. 1981, 103, 7028-7030.
(37) Zhao, P.; Lucht, B. L.; Kenkre, S. L.; Collum, D. B. J. Org. Chem. 2004,
69, 242-249.
9
J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007 4953

A R T I C L E S
Rowley et al.
Figure 6. Effect of [THF] on the rate of carbonylation of PO and BBL in
mixtures of THF/DMTHF. [THF] ) 0 M corresponds to pure DMTHF,
and [THF] ) 12.3 M corresponds to pure THF solvent. Reactions monitored
with in situ IR spectroscopy (νCdO BBL ) 1827 cm-1 and νCdO MSA )
Figure 7. Plot of -ln[BBL] versus time for the carbonylation of BBL to
MSA. Reaction performed in 1,4-dioxane and monitored with in situ IR
spectroscopy (νCdO BBL ) 1827 cm-1). [1] ) 4.0 mM, [BBL]o ) 1 M,
PCO ) 850 psi, T ) 40 °C.
1787 cm-1). [PO]0 ) 1.0 M or [BBL]0 ) 1.0 M, [1] ) 2.0 mM, PCO
)
850 psi, T ) 40 °C. Red line calculated from linear fit of initial rate versus
charge into the porphyrin ligand stabilizes the rate-determining
transition structure as the aluminum center is converted from a
formally neutral alkoxide to a cation.
[THF]-1 29
.
1,4-dioxane, where rate1 ) d[BBL]/dt and S ) THF or 1,4-
dioxane:
Lactone Carbonylation. The carbonylation of â-butyro-
lactone to methylsuccinic anhydride was also studied by in situ
IR spectroscopy. There are two methods for generating lactone
to study its carbonylation, either by first synthesizing and
isolating the lactone or by beginning with epoxide and forming
the lactone in situ; both methods provide the same results. In
this way, we were able to determine a separate rate law for
lactone carbonylation.
rate1 [PO]0[CO]0[1]1[S]1
The catalyst resting state was also directly observed during
the reaction by in situ IR spectroscopy. In 1,4-dioxane solution,
the free [Co(CO)4]- ion of 1 displays a strong absorption at
1885 cm-1. Upon addition of epoxide and CO, this band
immediately disappears and is replaced by a weaker absorbance
at 1715 cm-1, which is assigned as the νCdO of the cobalt acyl
resting state.22 Furthermore, when the reaction is run in the
presence of 4-nitrophenylisocyanate, another product, 6-methyl-
3-(4-nitrophenyl)-1,3-oxazinane-2,4-dione, was formed in ad-
dition to BBL, indicating that the ring-opened aluminum-
alkoxide/cobalt-acyl resting state (A; Scheme 2) is sufficiently
long-lived to be trapped by an electrophilic isocyanate.22,38
From this evidence, we conclude that the mechanistic picture
developed for carbonylation of 1,2-epoxybutane with 2 in 1,2-
dimethoxyethane at 25 °C also applies to the carbonylation of
PO with 1 in 1,4-dioxane at 40 °C. Thus, we propose that the
reaction proceeds via rapid epoxide ring opening, CO insertion,
and uptake of CO to form the neutral catalyst resting state, an
aluminum alkoxide/cobalt acyl (A). The rate-determining step
is then subsequent solvent-assisted ring closing to lactone
(Scheme 3) and regeneration of the catalyst ion pair. Once the
Unlike the case for epoxide, BBL carbonylation displayed a
first-order dependence on substrate concentration, which is
evident from the first-order decay of BBL and corresponding
production of anhydride (Figure 3). A plot of -ln[BBL] versus
time during the carbonylation confirms that the reaction rate is
first-order in lactone concentration (Figure 7). When the reaction
was repeated, varying CO pressure from 100 to 1000 psi, there
was no change in the rate of reaction, and no side products were
observed.29 The reaction was also repeated with different
concentrations of 1, and a plot of initial rate versus catalyst
concentration (Figure 5) indicated that the rate of reaction is
first-order in the complex 1, [(ClTPP)Al(THF)2]+[Co(CO)4]-.
The resting state of catalyst during the reaction can be observed
by in situ IR spectroscopy. During epoxide carbonylation, the
catalyst resting state was a cobalt acyl (vide supra), but
immediately upon complete conversion of PO to BBL, the cobalt
acyl (νCdO ) 1715 cm-1) disappeared and the signal corre-
sponding to free cobaltate (νCdO ) 1885 cm-1) reappeared and
persisted, indicative of [Co(CO)4]- as the cobalt resting state
during lactone carbonylation (B; Scheme 2).
Scheme 3. Proposed Rate-Determining Step of Propylene Oxide
Carbonylation (RDS1)
As in epoxide carbonylation, solvent played a significant role
in lactone carbonylation. Coordinating solvents such as THF
severely inhibited lactone carbonylation. This reaction was the
fastest in toluene, a weakly coordinating solvent (vide infra).
The effect of donor solvent was again quantified by measuring
the initial reaction rate in mixtures of THF/DMTHF.22,36 A small
increase in [THF] resulted in a dramatic reduction in rate. A
plot of initial rate versus [THF]-1 in DMTHF indicated that
the reaction rate depends inversely on the concentration of THF
(Figure 6),29 exactly opposite the case seen for epoxide
carbonylation. Further, the orders in substrates and catalyst in
THF solvents were the same as in 1,4-dioxane, and mixtures
of THF in 1,4-dioxane resulted in a similar inverse dependence
aluminum cation and cobalt anion are re-formed, epoxide is
coordinated and ring opened, immediately returning to the
aluminum-alkoxide/cobalt-acyl resting state (A). As 1 and 2 are
proposed to have the same rate-determining step, we can
speculate on the origin of the much greater epoxide carbony-
lation activity of 1. We suspect that increased delocalization of
(38) Church, T. L.; Byrne, C. M.; Lobkovsky, E. B.; Coates, G. W. J. Am.
Chem. Soc., submitted.
9
4954 J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007

Carbonylation of Epoxides to Succinic Anhydrides
A R T I C L E S
Scheme 4. Proposed Rate-Determining Step of â-Butyrolactone
Carbonylation (RDS2)
Table 5. Effect of Substitution on Relative Rate of Lactone
Carbonylationa
entry
R
R
′
relative rate
1
2
3
4
H
Me
Et
H
H
H
14
3.5
1.0
0b
Me
Me
a Reactions performed in 1,4-dioxane and monitored by in situ IR
spectroscopy. [lactone]0 ) 1.0 M, [1] ) 4.0 mM, PCO ) 850 psi, T )
40 °C. b No anhydride detected.
on the concentration of donor solvent or Lewis base. Because
the order in THF represents an order in solvent donicity,22,36
the inverse dependence of rate on THF can thus be applied to
the mechanism in 1,4-dioxane and understood as an inverse
dependence on the concentration of donor solvent or Lewis base.
Thus, the following expression can be written, where rate2 )
d[MSA]/dt and S ) THF:
which is far removed from the â position of the lactone.
Additionally, coordination to a cationic aluminum center is
expected to be facile. Thus, we propose nucleophilic ring-
opening attack by an ion paired [Co(CO)4]- on lactone
coordinated to [(ClTPP)Al(S)n]+ as the rate-determining step,40
which is proceeded by a pre-rate-determining associative or
dissociative substitution of lactone for solvent.
rate2 [BBL]1[CO]0[1]1[S]-1
This proposal for the rate-determining step may be used to
interpret the greatly enhanced lactone carbonylation activity of
1 relative to 2. Porphyrin-aluminum cations are more Lewis
acidic than salph-aluminum cations;41 thus a lactone coordinated
to [(ClTPP)Al]+ should be more activated toward ring opening,
decreasing the rate-limiting energy barrier.
The inverse dependence of rate on solvent most likely arises
from a pre-rate-determining substitution of lactone for solvent
at the aluminum center. One possible mechanistic interpretation
for this rate law is a reversible substitution (either associative
or dissociative) of lactone for donor solvent coordinated to the
aluminum cation, followed by a rate-limiting nucleophilic attack
of [Co(CO)4]-, resulting in the ring-opening of lactone (Scheme
4). Another mechanistic interpretation, although unlikely, could
be the rate-limiting coordination of BBL after dissociation of
donor solvent. These two potential rate-limiting steps could be
distinguished if there was an order in either the aluminum or
the cobalt ion. To determine if there was an order in the
individual ions or the complex 1 as a whole, we independently
varied the concentration of both the aluminum and the cobalt
components as the salts [(ClTPP)Al(THF)2]+[BPh4]- and
[PPh4]+[Co(CO)4]-. The source of [(ClTPP)Al(THF)2]+ and
[Co(CO)4]- (from either the discrete complex 1 or independently
added as the respective [BPh4]- and [PPh4]+ salts) did not affect
the rate of reaction and indicated that the rate of ion pair
exchange is comparable to or faster than catalysis. Independently
increasing the concentration of either [(ClTPP)Al(THF)2]+ or
[Co(CO)4]- in this manner did not change the observed rate of
anhydride formation. Thus, the reaction rate had a first-order
dependence on the concentration of the entire complex, but no
dependence on the separate cation or anion components. This
suggests that the cation and anion may exist as an ion pair in
the ground state and is consistent with a rate-limiting attack of
a Lewis acid-bound lactone by [Co(CO)4]-.
Double Carbonylation. Although the catalytic cycles proceed
via analogous steps for the two substrates, the rate-determining
steps appear to be determined by the relative rates of ring
opening and closing for epoxide and lactone carbonylation. For
ring opening, epoxides have more ring strain than lactones, and
thus more driving force for ring opening. Furthermore, in the
case of PO, ring opening by [Co(CO)4]- occurs at the less
substituted carbon of the epoxide; thus, subsequent ring opening
of BBL must occur at the more hindered â-carbon of the lactone.
On the other hand, ring closing a five-membered anhydride is
energetically more feasible than closing a strained four-
membered lactone ring. Additionally, because porphyrin alu-
minum carboxylates are more labile than porphyrin aluminum
alkoxides,42 the necessary transfer of substrate electron density
from the aluminum to the cobalt acyl should be more facile for
anhydride ring closing than lactone ring closing.
The non-overlapping nature of the two stages of double
carbonylation (vide supra, Figure 3) can be understood in terms
of the two unique resting states of the catalyst. Although both
PO and BBL rapidly and reversibly coordinate the porphyrin
aluminum cation, we propose there is a much greater affinity
for the ring opening of the epoxide over the lactone. While the
ring opening of PO followed by CO insertion and uptake are
not rate determining, BBL ring opening is rate determining.
Thus, in the presence of both PO and BBL, 1 will rapidly,
selectively, and irreversibly react with PO to form the aluminum-
alkoxide/cobalt-acyl resting state (A; Scheme 2). Upon ring
closing of this species, lactone is formed and catalyst is
regenerated, followed by another rapid ring opening of PO and
In a further attempt to distinguish these two possibilities, we
repeated the lactone carbonylation, varying the substitution at
the â position of the lactone (Table 5). The relative magnitude
of the observed rates of carbonylation of unsubstituted, mono-
substituted, and disubstituted lactones is most consistent with a
rate-limiting SN2 attack. However, it is unlikely that the steric
hindrance at the â position would have such a large impact on
the rate of lactone coordination, particularly because lactone
coordination is expected to occur at the carbonyl oxygen,39
(40) Another explanation could be that the rate-determining step changes with
the substitution. However, this only explains these data if coordination also
occurs at the oxygen in the ring, which is unlikely.
(41) Chen, P.; Chisholm, M. H.; Gallucci, J. C.; Zhang, X.; Zhou, Z. Inorg.
Chem. 2005, 44, 2588-2595.
(42) Chisholm, M. H.; Zhou, Z. J. Am. Chem. Soc. 2004, 126, 11030-11039.
(39) Shimasaki, K.; Aida, T.; Inoue, S. Macromolecules 1987, 20, 3076-3080.
9
J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007 4955

A R T I C L E S
Rowley et al.
Figure 8. Comparison of simulated (solid lines) with experimental data
(points, same as Figure 3) for the double carbonylation of PO in 1,4-dioxane.
[PO]0 ) 1.0 M, [1] ) 4.0 mM, PCO ) 850 psi, T ) 40 °C.
Figure 9. Comparison of relative initial rates for PO (blue) and BBL (red)
carbonylation as a function of solvent. Reactions monitored with in situ IR
spectroscopy (νCdO ) 1827 cm-1). [PO]0 ) 1.0 M or [BBL]0 ) 1.0 M,
[1] ) 2.0 mM, PCO ) 850 psi, T ) 40 °C.
CO migration. Thus, in the presence of epoxide, the catalyst is
continuously returned to the epoxide carbonylation resting state,
A, thereby excluding ring opening of lactone with 1. Once all
of the epoxide has been consumed, the catalyst resting state
then becomes the ion pair, [(ClTPP)Al(S)2]+[Co(CO)4]- (B;
Scheme 2, S ) solvent, substrate, or product), as evidenced by
the abrupt shift in CO stretching frequency for the cobalt
carbonyl species, from a cobalt-acyl to free cobaltate, coincident
with complete conversion of PO. In this state, the catalyst is
free to react with lactone and produce anhydride.
(Figure 9). As expected from the proposed rate-determining step
of lactone carbonylation, poorly coordinating solvents (DMTHF,
DFB, and toluene) generally result in uninhibited lactone ring
opening, whereas the Lewis basic, coordinating solvents severely
slow anhydride formation, such that essentially no anhydride
is produced in THF.43 To find a solvent system that has
comparable rates for both carbonylations, a possible solution
might be to use a mixture of solvents, or a solvent with
intermediate donicity. However, simply varying the donicity of
a THF solvent system is futile, because the rate will always be
much slower for at least one stage of carbonylation (Figure 6).
1,4-Dioxane, however, is unique in that it facilitates both epoxide
and lactone carbonylation and is far more active than an
optimized mixture of the two solvent extremes.
To determine if the proposed mechanism and rate-determining
steps predict the observed kinetic behavior, in particular the non-
overlapping stages of increase and decay in lactone concentra-
tion, we calculated the concentrations of lactone and anhydride
predicted by our mechanism as a function of time. From the
proposed mechanism, we derived rate equations for the con-
centrations of the important species. Then from kinetic studies,
we extracted rate constants for the proposed rate-determining
steps, and estimated larger constants for the faster steps.29 Using
typical experimental conditions as initial concentrations, we
solved for the concentration of lactone and anhydride as a
function of time. The experimental data from Figure 3 were
overlaid with these calculated concentrations (Figure 8), and,
not surprisingly, the simulated and experimental initial rates
were essentially the same. What was interesting, however, was
that the simulated concentrations were in good agreement
throughout the reaction, and particularly that the second
carbonylation did not occur until the first stage was complete.
This simulation confirms that the proposed mechanism, par-
ticularly the different catalyst resting states and rate-determining
steps, is consistent with the observed two-stage kinetic behavior.
Solvent Effects. To further investigate the effect of solvent,
we repeated the double carbonylation of PO in a variety of
solvents and monitored the reaction by in situ IR spectroscopy.
For every solvent studied, the double carbonylation occurred
in two distinct and non-overlapping stages; however, the rate
of each stage varied greatly as a function of solvent. As seen in
Figure 9, the rate of epoxide carbonylation slowed with
increasing sterics and decreasing donicity36 of the substituted
THF solvents: THF, 2-methyltetrahydrofuran (MTHF), DMTHF.
Similar activity is seen with the moderately donating tetrahy-
dropyran (THP), 1,2-dimethoxyethane (DME), and dioxane, but
the reaction is very slow in poorly coordinating toluene and
difluorobenzene (DFB). For lactone carbonylation, the trend is
reversed, highlighting the difference between these two stages
Although it is clear that Lewis bases affect the rate of these
reactions, the origin of enhanced activity seen for both stages
in 1,4-dioxane cannot be explained simply in terms of a Lewis
base assisting or inhibiting the rate-determining steps. Instead,
1,4-dioxane is unique among the ether solvents studied in that
its symmetry results in a low dipole moment. The importance
of solvent polarity, separate from donicity, is dramatically
illustrated by comparing the rates of carbonylation in the isomers
of difluorobenzene (Figure 9). In most respects, ortho-, meta-,
and para-difluorobenzene (o-DFB, m-DFB, p-DFB) are similar
solvents; however, they vary widely in polarity (ꢀ ) 13, 5, and
2, respectively);44 thus differences in rate in these solvents may
be primarily attributed to the difference in polarity of each
isomer. Epoxide carbonylation was comparatively slow in all
three DFB isomers (as expected from their poor donicity), with
little change thus attributable to differences in polarity. In
agreement with our previous study,22 donicity appears to have
a greater influence than polarity on the rate of epoxide
carbonylation.
For lactone carbonylation, however, markedly different rates
are observed in each isomer of DFB. o-DFB, the most polar
isomer, is an extremely slow solvent for lactone carbonylation;
m-DFB is less polar and exhibits an increased rate; and p-DFB
(43) While this is unproductive for double carbonylation, it is synthetically very
useful for the rapid and exclusive production of the â-lactone intermediate,
because the reaction essentially stops after the first carbonylation. When
using 1 in THF, the carbonylation of 1,2-epoxybutane to â-valerolactone
is nearly twice as fast as the most active system reported to date,20 with a
TON of over 6000 in 6 h at 60 °C.
(44) Wohlfarth, C. In CRC Handbook of Chemistry and Physics, 76th ed.; Lide,
D. R., Ed.; CRC Press: New York, 1995; pp 6-149.
9
4956 J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007

Carbonylation of Epoxides to Succinic Anhydrides
A R T I C L E S
is nonpolar and the best isomer for rapid lactone carbonylation
(Figure 9). Based solely on their poor donicity, all three isomers
should be competent solvents for lactone carbonylation; instead,
we see a clear trend of increasing rate with decreasing solvent
polarity.
To test this effect of polarity with dioxane, we predicted that
1,3-dioxane (ꢀ ) 13.6),45 which is significantly more polar than
1,4-dioxane (ꢀ ) 2.2),44 would result in slow lactone carbony-
lation, comparable to the other more polar ether solvents. In
fact, this was exactly what we observed, confirming that polar
solvents inhibit lactone carbonylation (Figure 9).
opening SN2-type nucleophilic attack by [Co(CO)4]-. Although
the two stages have similar initial rates, ring opening is rapid
for PO and rate limiting for BBL. Thus, the catalyst ion pair
reacts irreversibly and selectively with epoxide, giving rise to
the observed two-stage behavior. The effect of solvent on each
stage was explained in terms of both donicity and polarity. Rates
of epoxide and lactone carbonylation display an opposite
dependence on donor solvent. This unique feature allows for
synthetic versatility, in which judicious choice of solvent
determines whether the reaction stops cleanly at â-lactone (THF)
or immediately continues on to succinic anhydride (1,4-dioxane).
Overall, the rate of epoxide carbonylation is primarily
dependent on solvent donicity, whereas the rate of lactone
carbonylation is strongly dependent on both solvent donicity
and polarity. The ability of 1,4-dioxane to function as a solvent
for both epoxide and lactone carbonylation may then be
explained in terms of its unique combination of these properties.
It retains enough Lewis basic character46 to assist in ring closing
to form lactone, while not significantly inhibiting the subsequent
ring opening of lactone. Additionally, its low polarity does not
retard epoxide carbonylation, but accelerates lactone carbony-
lation, due to a nonpolar environment, which may favor the
transition of catalyst from an ionic pair to a formally neutral
intermediate upon ring opening of lactone. Thus, endowed with
the proper combination of donicity and polarity, 1,4-dioxane
enables the rapid and efficient double carbonylation of epoxides.
Experimental Section
General Considerations. All manipulations of air- and/or water-
sensitive compounds were carried out under dry nitrogen using a Braun
Unilab drybox or standard Schlenk line techniques. NMR spectra were
recorded on a Varian Mercury spectrometer (1H NMR, 300 MHz; 13
C
NMR, 75 MHz) and referenced versus residual non-deuterated or
monoprotonated solvent shifts. Standard IR spectra were collected on
a Mattson RS-10500 Research Series FTIR. In situ IR data were
collected using a 100-mL Parr stainless steel high-pressure reactor
modified for use with a Mettler-Toledo ReactIR 4000 Reaction Analysis
System fitted with a Sentinel DiComp high-pressure probe, and analyzed
with ReactIR software version 2.21. Enantiomeric excesses were
measured using either GC (HP 6890 Series equipped with an Astec
R-cyclodex TFA chiral capillary column (250 µm × 60 m)), or HPLC
(Waters 515 HPLC pump, Waters 2410 refractive index detector, semi-
prep Regis Pirkle (S,S) Whelk-O 1 column (25 cm × 10 mm)). High-
resolution mass spectra were obtained using electron impact conditions
on a 70-VSE mass spectrometer by the Mass Spectrometry Laboratory,
School of Chemical Sciences, University of Illinois. X-ray crystal-
lographic data were collected using a Bruker X8 APEX II (Mo KR, λ
) 0.71073 Å) at 173(2) K, and frames were integrated with the Bruker
SAINT+ Program. Optimization of catalyst loading was performed in
custom designed and built six-well, stainless steel, high-pressure
reactors,14,19 which accommodated six 4- or 8-mL glass vials. All high-
pressure reactors were dried under vacuum at 90 °C prior to use.
Materials. Carbon monoxide (research grade) was purchased from
Matheson and used without further purification. 1,4-Dioxane, 1,3-
dioxane, tetrahydropyran, tetrahydrofuran, 2-methyltetrahydrofuran, 2,5-
dimethyltetrahydrofuran, diethyl ether, and 1,2-dimethoxyethane were
vacuum transferred from purple Na/benzophenone. Hexanes and toluene
were dried and deoxygenated on columns of alumina and Q5 copper,
respectively. Diethyl ether and methylene chloride were dried on
columns of alumina and degassed via repetitive freeze-pump-thaw
cycles (FPT). Difluorobenzene, ethyl acetate, acetone, and acetonitrile
were dried over 4 Å molecular sieves and vacuum transferred after
FPT. Epoxides were stirred for a few days over CaH2, degassed by
FPT, and vacuum distilled, except 10,11-epoxyundecan-1-ol (22) and
N,N-dimethyl-10,11-undecylamide (28), which were dried over CaH2,
filtered, and degassed by stirring under dynamic vacuum (0.1 Torr).
Ethylene oxide (4), propylene oxide (6), 1,2-epoxybutane (8), 1,2-
epoxyhexane (10), 1,2-epoxydodecane (12), n-butyl glycidyl ether (16),
tert-butyldimethylsilyl glycidyl ether (18), benzyl glycidyl ether (20),
1,2-epoxy-5-hexene (30), 1,2-epoxy-7-octene (32), (2,3-epoxypropyl)-
benzene (34), styrene oxide (36), 1,2,7,8-diepoxyoctane (38), â-pro-
piolactone, â-butyrolactone (BBL), 4-nitrophenylisocyanate, and meta-
chloroperoxybenzoic acid (mCPBA) were purchased from Aldrich. cis-
2,3-Epoxybutane (40) and trans-2,3-epoxybutane (42) were purchased
from GFS Chemicals. Iso-butylene oxide was purchased from TCI
America. Co2(CO)8 and octaethylporphyrin were purchased from Strem.
(R)-Propylene oxide ((R)-6),24 (S)-1,2-epoxyhexane ((S)-10),24 (R)-
benzyl glycidyl ether ((R)-20),24 10,11-epoxyundecan-1-ol (22),47 4,5-
Conclusion
We have reported the double carbonylation of epoxides to
give succinic anhydrides in a single synthetic step, using the
highly active bimetallic catalyst, 1. This reaction is compatible
with a variety of substituted epoxides with aliphatic, aromatic,
alkene, ether, ester, alcohol, nitrile, and amide functional groups.
Disubstituted and enantiomerically pure epoxides were doubly
carbonylated with excellent retention of stereochemical purity
via inversion of each of the two stereocenters. Given the ease
of catalyst synthesis (and its impending commercial availability),
coupled with the ability to convert readily available epoxides
to functionalized and stereochemically pure succinic anhydrides
in high yield on multigram scale, we believe this to be an
efficient and synthetically useful transformation.
The double carbonylation proceeds through two distinct and
non-overlapping stages: epoxide carbonylation to lactone, and
subsequent lactone carbonylation to anhydride. Although these
two reactions are proposed to follow analogous catalytic cycles,
the nature of the substrate determines which step within each
cycle will be rate determining. We investigated the mechanism
of double carbonylation of PO, and for each stage we determined
the rate law with respect to substrates, catalyst, and solvent.
The resting state of the catalyst during each stage was observed
by in situ IR spectroscopy. In the case of epoxide carbonylation,
rate1 [PO]0[CO]0[1]1[S]1 (S ) THF or 1,4-dioxane), which
is consistent with a rate-limiting, solvent-assisted ring closing
to lactone from an aluminum-alkoxide/cobalt-acyl resting state.
For lactone carbonylation, rate2 [BBL]1[CO]0[1]1[S]-1, which
is consistent with substitution of lactone for donor solvent at
the porphyrin aluminum cation, followed by a rate-limiting, ring-
(45) Walker, R.; Davidson, D. W. Can. J. Chem. 1959, 37, 492-495.
(46) Geerlings, J. D.; Varma, C. A. G. O.; van Hemert, M. C. J. Phys. Chem.
B 2000, 104, 56-64.
(47) Johnson, C. R.; Dutra, G. A. J. Am. Chem. Soc. 1973, 95, 7777-7782.
9
J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007 4957

A R T I C L E S
Rowley et al.
epoxypentyl butyrate (24),20 5,6-epoxyhexanenitrile (26),48 N,N-dimethyl-
10,11-undecylamide (28),49 trans-3,4-epoxyhexane (44),50 trans-2,3-
epoxyoctane (46),51 NaCo(CO)4,52 [(salph)Al(THF)2]+[Co(CO)4]- (2),16
[(OEP)Cr(THF)2]+[Co(CO)4]- (3),20 and [PPh4]+[Co(CO)4]- 53 were
synthesized as previously reported. All other materials were com-
mercially available and used as received.
General Procedure for the Small-Scale Carbonylation of Ep-
oxides. A six-well, stainless steel, high-pressure reactor was loaded
with six 4-mL glass vials and magnetic stir bars. In a nitrogen drybox,
an appropriate amount of 1 was weighed into each vial, then solvent
was added to each, with no effort made to dissolve the catalyst. Epoxide
was weighed into the vials, and the catalyst became soluble in the
reaction mixture. Each vial was cooled in the drybox freezer (-37 °C
for at least 5 min) to limit ketone formation,16 and then all were placed
in the reactor. The reactor was sealed and removed from the drybox,
immediately pressured to 850 psi with CO, stirred, and heated to the
appropriate temperature. After the indicated time, the reactor was placed
on dry ice, cooled to <0 °C, then slowly vented. Crude reaction mixture
Bis(tetrahydrofuran)-meso-tetra(4-chlorophenyl)porphyrinato Alu-
minum Tetracarbonyl Cobaltate, [Cl(TPP)Al(THF)2]+[Co(CO)4]-
(1). (Note: This and related catalysts may be commercially available
in 2007 from Sigma-Aldrich.) The following synthetic method is
analogous to that of related literature compounds.16-21,54a According
to the method of Adler,54b meso-tetra(4-chlorophenyl)-21H,23H-por-
phyrin (ClTPPH2) was easily synthesized from pyrrole and 4-chlo-
robenzaldehyde, and dried under vacuum overnight. All subsequent
manipulations were performed using strict air-free techniques, and all
reagents and solvents were dried and degassed prior to use. Using a
modified literature procedure,55 ClTPPH2 (3.44 g, 4.57 mmol) was
placed in a Schlenk tube equipped with a magnetic stir bar in a drybox.
Upon removal to the bench top, the dark purple ligand was dissolved
in 250 mL of CH2Cl2 to form a very dark red solution. Diethyl
aluminum chloride (5.0 mL, 5.0 mmol, 1.0 M in heptane, 1.1 equiv)
was added via syringe through a septum under flow of N2. Ethane
evolved from the reaction and was vented. The reaction was stirred at
room temperature for 3 h, then solvent was removed in vacuo. The
residual red-purple solid was dried under vacuum overnight and
identified by comparison of its 1H NMR spectrum (CDCl3, 300 MHz,
δ) 9.10 (s, 8H), 8.13 (broad d, 8H), 7.76 (m, 8H), to that of free ligand
(CDCl3, 300 MHz, δ) 8.92 (s, 8H), 7.88 (d, 8H, 3J ) 8.2 Hz), 7.49 (d,
8H, 3J ) 8.2 Hz), -2.19 (s, 2H, NH). Having reacted quantitatively, it
was used without further purification. The Schlenk tube was brought
into a drybox where NaCo(CO)4 (887 mg, 4.57 mmol) was added. Upon
removal to the Schlenk line, the solids were dissolved in THF (200
mL) and stirred overnight at room temperature. The very dark, red-
purple solution was concentrated to 100 mL in vacuo, and NaCl was
allowed to precipitate. The solution was then filtered and layered with
hexanes (200 mL). Slow diffusion over the course of a few days
afforded large, purple, X-ray quality crystals that were stable under N2
for over a year. (Intentional mixing of the layers results in rapid
precipitation of the complex as a crystalline powder with comparable
catalytic activity and selectivity.) The crystals were filtered, washed
1
from each vial was analyzed by H NMR spectroscopy in CDCl3.
General Procedure for the Large-Scale Carbonylation of Ep-
oxides. In a nitrogen drybox, a 100-mL Parr high-pressure reactor was
charged with the appropriate amount of 1 and solvent, then removed
from the drybox. The reactor was first pressured with 200 psi CO and
stirred for 10 min, then vented down to 20 psi without stirring. The
epoxide was injected via gastight syringe into the CO-filled reactor
through a septum at room temperature. (This procedure of presaturating
the solution with CO eliminated ketone formation.) The reactor was
then immediately pressured to 850 psi CO, followed by rapid stirring
and heating to 90 °C. After the appropriate time, the reactor was placed
on dry ice, cooled to <0 °C, and slowly vented.
General Procedure for in Situ IR Spectroscopy. In a nitrogen
drybox, a React IR modified reactor was charged with catalyst and
solvent, and substrate was drawn into a gastight syringe. After removal
from the drybox, the reactor was pressured to 850 psi with CO,
mechanically stirred, and heated to 40 °C. Once the system equilibrated,
a background spectrum was taken (16 scans). The reactor was vented
to 20 psi, and at time ) 0, substrate was injected through a septum,
and the reactor was immediately repressurized to 850 psi CO. IR spectra
were collected once per minute (16 scans/spectrum at 4 cm-1 resolu-
tion). Absorbances of BBL and MSA were measured at 1827 and 1787
cm-1, respectively, and corrected for overlap.
Isocyanate Trapping of Catalyst Resting State. In a nitrogen
drybox, an 8-mL glass vial equipped with a magnetic stir bar was
charged with 1 (10.4 mg, 9.5 µmol), 4-nitrophenylisocyanate (161 mg,
0.98 mmol), 1,4-dioxane (0.95 mL), and propylene oxide (55 mg, 0.95
mmol), and then placed in a six-well reactor. Upon removal from the
drybox, it was immediately pressured to 850 psi CO, stirred, and heated
to 40 °C. After 24 h, the reactor was cooled and vented. Analysis of
the crude reaction mixture by 1H NMR spectroscopy in CDCl3 indicated
complete conversion of epoxide to methyl succinic anhydride (48%),
â-butyrolactone (45%), and 3-(4-nitrophenyl)-6-methyl-1,3-oxazine-
2,4-dione (7%).
1
with hexanes, and dried in vacuo (4.33 g, 87% yield). H NMR (300
MHz, THF-d8, δ): 9.23 (s, 8H), 8.21 (m, 8H), 7.88 (m, 8H), 3.62
(m, 8H), 1.78 (m, 8H); IR (Nujol, NaCl) νCdO ) 1875 cm-1
.
Crystal data:29 monoclinic, space group P21/n, a ) 11.9296(6) Å,
b ) 22.6420(12) Å, c ) 20.4696(10) Å, R ) 90°, â ) 102.446(2)°,
γ ) 90°, V ) 5399.1(5) Å3; Z ) 4, formula weight ) 1092.63 for
C48H24AlCl4CoN4O4‚2C4H8O and density (calc.) ) 1.344 g/mL;
R ) 0.0593, Rw ) 0.1634 (I > 2σ(I)).
Representative Synthesis of Epoxide: Cyclohexyl Oxirane (14).
Prepared by a modified literature56 procedure: vinylcyclohexane (17.4
mL, 13.9 g, 126 mmol) was added dropwise to a cold solution of
mCPBA (36.4 g, 77%, 160 mmol) in CH2Cl2 (600 mL). The reaction
was stirred at 0 °C for 2 h, then at room temperature overnight. The
milky white suspension was extracted with 10% NaHSO3 (aq) (to remove
organic peroxides), then twice with saturated NaHCO3 (aq). The organic
layer was washed with saturated NaCl (aq), dried over Na2SO4, and
concentrated to an oil under reduced pressure. Vacuum distillation at
1 mmHg gave 14 as a clear, colorless liquid (13.1 g, 82%).
Bis(tetrahydrofuran)-meso-tetra(4-chlorophenyl)porphyrinato Alu-
minum Tetraphenyl Borate, [Cl(TPP)Al(THF)2]+[BPh4]-. The syn-
thetic procedure was identical to that for 1, except that NaBPh4 was
used in place of NaCo(CO)4. The structure was confirmed by X-ray
crystallography.29
(48) de Raadt, A.; Klempier, N.; Faber, K.; Griengl, H. J. Chem. Soc., Perkin
Trans. 1992, 137-140.
(49) Glushko, L. P.; Petukhova, E. I.; Klebanskii, E. O. Zh. Org. Khim. 1997,
33, 1527-1531.
Representative Synthesis of Lactone: â-Valerolactone. A 100-
mL Parr high-pressure reactor with mechanical stirrer was dried
overnight at 120 °C under vacuum. In a nitrogen drybox, the reactor
was charged with 1 (21 mg, 0.020 mmol), THF (16 mL, 200 mmol),
and 1,2-epoxybutane (8.6 mL, 100 mmol), then closed and removed
from the drybox. The reactor was pressured with 850 psi CO and stirred
at 60 °C for 6 h, by which time the pressure dropped to 400 psi. The
(50) Seebach, D.; Beck, A. K.; Mukhopadhyay, T.; Thomas, E. HelV. Chim.
Acta 1982, 65, 1101-1133.
(51) Mischitz, M.; Mirtl, C.; Saf, R.; Faber, K. Tetrahedron: Asymmetry 1996,
7, 2041-2046.
(52) Edgell, W. F.; Lyford, J. Inorg. Chem. 1970, 9, 1932-1933.
(53) Wei, C. H.; Bockman, T. M.; Kochi, J. K. J. Organomet. Chem. 1992,
428, 85-97.
(54) (a) Barbe, J. M.; Guilard, R. In The Porphyrin Handbook; Kadish, K. M.,
Smith, K. M., Guilard, R., Eds.; Academic Press: New York, 2000; Vol.
3, pp 225-229. (b) Adler, A. D.; Longo, F. R.; Finarelli, J. D.; Goldmacher,
J.; Assour, J.; Korsakoff, L. J. Org. Chem. 1967, 32, 476.
(55) Konishi, K.; Aida, T.; Inoue, S. J. Org. Chem. 1990, 55, 816-820.
(56) Vyvyan, J. R.; Meyer, J. A.; Meyer, K. D. J. Org. Chem. 2003, 68, 9144-
9147.
9
4958 J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007

Carbonylation of Epoxides to Succinic Anhydrides
A R T I C L E S
reactor was cooled and vented. Volatiles were removed in vacuo, and
the resulting oil was vacuum distilled to afford â-valerolactone (9.1 g,
91%).
IR (melt, NaCl) νCdO ) 1861, 1785 cm-1; mp ) 36 °C; HRMS (EI)
m/z calcd (C10H14O3) 182.0943, found 182.0945.
1
Benzyloxymethylsuccinic Anhydride (21). H NMR (300 MHz,
2
Representative Synthesis of Anhydride: Cyclohexylsuccinic
Anhydride (15). A 100-mL Parr high-pressure reactor with mechanical
stirrer was dried overnight at 120 °C under vacuum. In a nitrogen
drybox, the reactor was charged with 1 (262 mg, 0.240 mmol) and
1,4-dioxane (20 mL), then closed and removed from the drybox. The
reactor was pressured with 200 psi CO, stirred for 10 min, and then
vented down to 20 psi without stirring. Cyclohexyl oxirane (4.88 mL,
36.0 mmol) was injected via syringe into the CO-filled reactor at room
temperature, and then the reactor was immediately repressurized to 850
psi CO, followed by rapid stirring and heating to 90 °C. After 3 h, the
reactor was placed on dry ice, cooled to <0 °C, and slowly vented.
Volatiles were removed in vacuo, the resulting oil was vacuum distilled
to afford a clear oil that solidified on standing, then was recrystallized
from cold ether/hexanes to afford 15 (5.32 g, 81% isolated yield).
General Procedures for Anhydride Purification. Anhydrides were
obtained by rotary evaporation followed by bulb-to-bulb vacuum
distillation (9, 11, 15, 37, 45, 47) or sublimation (5, 7, 39, 41, 43) of
the crude reaction mixture. For higher boiling anhydrides, solvent and
any volatile ketones were removed from the crude reaction mixture in
vacuo. The catalyst residue was then removed from the resulting oil
by elution through a plug of silica gel (with 1:2 EtOAc:hexanes for
13, 17, 19, 21, 23, 27, 29, 31, 33, 35, and with 4:1 EtOAc:hexanes for
25). Subsequent concentration by rotary evaporation afforded the
product, which could be further purified by vacuum distillation or
recrystallization from cold ether/hexane, as appropriate. Anhydrides
were stable enough to be isolated and characterized, but reacted slowly
with moisture to give the corresponding diacids.
CDCl3, δ): 7.39-7.24 (m, 5H), 4.58 (d, 1H, J ) 12.1 Hz), 4.51 (d,
1H, 2J ) 12.1 Hz), 3.91 (dd, 1H, 2J ) 9.2, 3J ) 3.4 Hz), 3.63 (dd, 1H,
3
2J ) 9.2, J ) 3.1), 3.26 (m, 1H), 3.02 (m, 2H); 13C NMR (75 MHz,
CDCl3, δ): 172.66, 170.24, 137.08, 128.79, 128.32, 127.91, 73.69,
67.51, 42.27, 31.70; IR (melt, NaCl) νCdO ) 1858, 1782 cm-1; mp )
69-70 °C; HRMS (EI) m/z calcd (C12H12O4) 220.0736, found 220.0732.
Enantiomeric excess determined by chiral HPLC: solvent (4.5 mL/
min, 93:7 hexanes:isopropanol with 0.1% acetic acid). The (R)- and
(S)-enantiomers elute at 24.4 and 27.0 min, respectively. Absolute
configuration was assigned to (S)-21 on the basis of analogy to the
other stereochemically pure substrates.
9-Hydroxynonylsuccinic Anhydride (23). 1H NMR (300 MHz,
3
CDCl3, δ): 3.64 (t, 2H, J ) 6.6 Hz), 3.09 (m, 2H), 2.66 (m, 1H),
1.93 (m, 1H), 1.65 (m, 1H), 1.56 (m, 2H), 1.50-1.23 (m, 13H); 13C
NMR (75 MHz, CDCl3, δ): 174.06, 170.57, 62.84, 40.69, 34.08, 32.73,
30.90, 29.44, 29.36, 29.22, 29.08, 26.72, 25.73; IR (melt, NaCl) νO-H
) 3380, 1062, νCdO ) 1861, 1783 cm-1; mp ) 48 °C.
1
3-Butyroxypropylsuccinic Anhydride (25). H NMR (300 MHz,
3
CDCl3, δ): 4.05 (t, 2H, J ) 6.2 Hz), 3.11 (m, 2H), 2.64 (m, 1H),
2.23 (t, 2H, 3J ) 7.5 Hz), 1.96 (m, 1H), 1.72 (m, 3H), 1.58 (qt, 2H, 3J
) 7.4 Hz, 3J ) 7.5 Hz), 0.89 (t, 3H, 3J ) 7.4 Hz); 13C NMR (75 MHz,
CDCl3, δ): 173.82, 173.77, 170.31, 63.24, 40.44, 36.24, 34.28, 27.71,
26.65, 18.58, 13.85; IR (melt, NaCl) νCdO ) 1862, 1785, 1728 cm-1
;
mp ) 35 °C; HRMS (EI) m/z calcd (C11H16O5) 228.0998, found
228.0989.
3-Cyanopropylsuccinic Anhydride (27). 1H NMR (300 MHz,
3
CDCl3, δ): 3.14 (m, 2H), 2.68 (m, 1H), 2.42 (t, 2H, J ) 6.8 Hz),
Identification of Anhydrides. Product anhydrides were identified
2.00 (m, 1H), 1.92-1.67 (m, 3H); 13C NMR (75 MHz, CDCl3, δ):
173.62, 170.21, 119.40, 40.18, 34.33, 29.88, 23.04, 17.10; IR (neat,
NaCl) νC≡N ) 2247, νCdO ) 1862, 1784 cm-1; clear colorless oil.
HRMS (EI) m/z calcd (C8H9O3N-CO2) 123.0684, found 123.0684.
1
by comparison of their H NMR spectra to a commercially available
samples for 5 and 7 and by comparison to literature values for 9,14
13,14 17,14 19,14 31,14 35,57 37,58 41,58 43,58 and 45.59 Anhydrides 11,60
25,61 and 3962 have previously been reported; however, their 1H NMR
9-(Succinic anhydridyl)-N,N-dimethylnonanamide (29). 1H NMR
(300 MHz, CDCl3, δ): 3.09 (m, 2H), 3.00 (s, 3H), 2.93 (s, 3H), 2.66
(m, 1H), 2.30 (t, 2H, 3J ) 7.6 Hz), 1.92 (m, 1H), 1.62 (m, 3H), 1.45-
1.25 (m, 10H); 13C NMR (75 MHz, CDCl3, δ): 173.95, 173.15, 170.42,
40.56, 37.22, 35.24, 33.98, 33.18, 30.74, 29.23, 29.15, 28.99, 28.90,
26.59, 24.98; IR (melt, NaCl) νCdO ) 1861, 1782, 1642 cm-1; mp )
49 °C. HRMS (EI) m/z calcd (C15H25O4N-CO2) 239.1885, found
239.1883.
1
spectra are reported in DMSO-d6 or CCl4. Anhydrides for which H
NMR spectra have not been reported in CDCl3 are characterized below.
Methylsuccinic Anhydride (7). Enantiomeric excess determined by
chiral GC: inlet (T ) 250 °C, 32 psi, 260.0 mL/min, 100:1 split);
detector (250 °C, 35 mL/min H2 flow, 400 mL/min air flow); column
(32 psi, 1.5 mL/min); oven (150 °C, isothermal). The (R)- and (S)-
enantiomers elute at 6.9 and 7.2 min, respectively.
1
n-Butylsuccinic Anhydride (11). H NMR (300 MHz, CDCl3, δ):
1
Hex-5-enylsuccinic Anhydride (33). H NMR (300 MHz, CDCl3,
3.10 (m, 2H), 2.67 (m, 1H), 1.95 (m, 1H), 1.65 (m, 1H), 1.38 (m, 4H),
0.93 (m, 3H); 13C NMR (75 MHz, CDCl3, δ): 173.90, 170.34, 40.85,
34.28, 30.89, 29.00, 22.42, 13.97; IR (melt, NaCl) νCdO ) 1862, 1786
cm-1; mp ) 46 °C; HRMS (EI) m/z calcd (C8H12O3), 156.0786; found,
156.0790. Enantiomeric excess determined by chiral GC: inlet (T )
250 °C, 33 psi, 125.0 mL/min, 50:1 split); detector (250 °C, 35 mL/
min H2 flow, 400 mL/min air flow); column (33.0 psi, 1.5 mL/min);
oven (160 °C, isothermal). The (R)- and (S)-enantiomers elute at 11.3
and 12.2 min, respectively. Absolute configuration of (R)-11 was
confirmed by hydrolysis to (R)-n-butylsuccinic acid and comparison
of optical rotation ([R]25D +22.3° (c 1.5, H2O)) to literature values for
δ): 5.76 (ddt, 1H, 3J ) 6.6 Hz, 3J ) 10.2 Hz, 3J ) 17.1 Hz), 4.98 (m,
2H), 3.09 (m, 2H), 2.65 (m, 1H), 2.07 (m, 2H), 1.93 (m, 1H), 1.65 (m,
1H), 1.42 (m, 4H); 13C NMR (75 MHz, CDCl3, δ): 173.87, 170.34,
138.30, 115.17, 40.78, 34.21, 33.42, 30.91, 28.37, 26.24; IR (neat, NaCl)
νCdO ) 1862, 1786 cm-1; clear colorless oil; HRMS (EI) m/z calcd
(C10H14O3) 182.0943, found 182.0938.
1,4-Bis(succinic anhydridyl)butane (39). 1:1 mixture of the racemic
1
and meso stereoisomers. H NMR (300 MHz, acetone-d6, δ): 3.34
(dddd, 2H, 3J ) 5.3 Hz, 3J ) 6.3 Hz, 3J ) 8.8 Hz, 3J ) 9.9 Hz), 3.22
(dd, 2H, 3J ) 9.8 Hz, 2J ) 18.1 Hz), 2.86 (dd, 2H, 3J ) 6.3 Hz, 2J )
18.1 Hz), 1.95 (m, 2H), 1.77 (m, 2H), 1.54 (m, 4H); 13C NMR (75
MHz, acetone-d6, δ): 175.42, 171.77, 41.26, 34.53, 34.51, 30.71, 30.64,
27.10, 27.08; IR (melt, NaCl) νCdO ) 1857, 1778 cm-1; mp ) 105-
108 °C; HRMS (EI) m/z calcd (C12H14O6) 254.0790, found 254.0798.
(S)-n-butylsuccinic acid, [R]29 -21.5° (c 1.49, H2O).63
D
1
Cyclohexylsuccinic Anhydride (15). H NMR (300 MHz, CDCl3,
δ): 3.00 (m, 2H), 2.74 (m, 1H), 1.90 (m, 1H), 1.78 (m, 3H), 1.69 (m,
1H), 1.58 (m, 1H) 1.36-0.95 (m, 5H); 13C NMR (75 MHz, CDCl3,
δ): 173.14, 170.70, 46.50, 39.12, 31.36, 30.34, 28.47, 26.07, 25.88;
trans-3-Methyl-4-(n-pentyl)succinic Anhydride (47). 1H NMR (300
MHz, CDCl3, δ): 2.82 (dq, 1H, 3J ) 7.3 Hz, 3J ) 7.5 Hz), 2.69 (ddd,
3
3
3
(57) Qiao, C. H.; Marsh, E. N. G. J. Am. Chem. Soc. 2005, 127, 8608-8609.
(58) Crandall, J. K.; Schuster, T. J. Org. Chem. 1990, 55, 1973-1975.
(59) Bode, J.; Brockman, H. Chem. Ber. 1972, 105, 34-44.
(60) Bergmeier, S. C.; Ismail, K. A. Synthesis 2000, 1369-1371.
(61) Huang, N.-J.; Xu, L.-H.; Ye, M.-X.; Feng, J. Youji Huaxue 1991, 11, 174-
178.
1H, J ) 5.6 Hz, J ) 7.5 Hz, J ) 7.7 Hz), 1.88 (m, 1H), 1.66 (m,
3
3
1H), 1.51-1.25 (m, 6H), 1.40 (d, 3H, J ) 7.3 Hz), 0.87 (t, 3H, J )
6.9 Hz); 13C NMR (75 MHz, CDCl3, δ): 173.84, 173.00, 48.08, 41.27,
31.52, 29.98, 26.40, 22.46, 15.65, 14.06; IR (neat, NaCl) νCdO ) 1858,
1785 cm-1; clear colorless oil; HRMS (EI) m/z calcd (C10H16O3)
184.1009, found 184.1094.
(62) Bouvier, J. M.; Bruneau, C. M. Bull. Soc. Chim. Fr. 1975, 2195-2201.
(63) Zhou, B.; Xu, Y. J. Org. Chem. 1988, 53, 4419-4421.
9
J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007 4959

A R T I C L E S
Rowley et al.
Acknowledgment. We thank Dr. T. L. Church for valuable
discussion of mechanism, and Dr. J. A. R. Schmidt, J. W.
Kramer, and Dr. C. M. Byrne for providing substrates and
solvents. We also thank Prof. D. B. Collum, Dr. K. J. Singh,
and A. C. Hoepker for assistance with the simulation. G.W.C.
gratefully acknowledges a Packard Foundation Fellowship in
Science and Engineering and an Arnold and Mabel Beckman
Foundation Young Investigator Award, as well as funding from
the NSF (CHE-0243605), DOE (DE-FG02-05ER15687), and
Cornell Center for Materials Research supported by the NSF
MRSEC program (DMR-0250404).
Supporting Information Available: Screening and charac-
terization of additional catalysts, calibration curve for MSA
in 1,4-dioxane, rate dependence on PCO, plot of initial rate
versus [THF]-1, details of kinetics simulation for double
carbonylation, 1H and 13C NMR spectra of all new anhydrides,
and X-ray data for 1, [(ClTPP)Al(THF)2]+[BPh4]-, and
[(OEP)Al(THF)2]+[Co(CO)4]-. This material is available free
of charge via the Internet at http://pubs.acs.org.
JA066901A
9
4960 J. AM. CHEM. SOC. VOL. 129, NO. 16, 2007