The mechanism of epoxide carbonylation by [Lewis acid]+[Co(CO) 4]- catalysts

Article abstract of DOI:10.1021/ja061503t

A detailed mechanistic investigation of epoxide carbonylation by the catalyst [(salph)Al(THF)₂]⁺ [Co(CO)₄] ^- (1, salph = N,N′-o-phenylenebis(3,5-di-tert- butylsalicylideneimine), THF = tetrahydrofuran) is report

Full text of DOI:10.1021/ja061503t

Published on Web 07/14/2006
The Mechanism of Epoxide Carbonylation by
[Lewis Acid]+[Co(CO)4]- Catalysts
Tamara L. Church, Yutan D. Y. L. Getzler,† and Geoffrey W. Coates*
Contribution from the Department of Chemistry and Chemical Biology, Baker Laboratory,
Cornell UniVersity, Ithaca, New York 14853-1301
Received March 3, 2006; E-mail: gc39@cornell.edu
Abstract: A detailed mechanistic investigation of epoxide carbonylation by the catalyst [(salph)Al(THF)2]+
[Co(CO)4]- (1, salph ) N,N′-o-phenylenebis(3,5-di-tert-butylsalicylideneimine), THF ) tetrahydrofuran) is
reported. When the carbonylation of 1,2-epoxybutane (EB) to â-valerolactone is performed in 1,2-
dimethoxyethane solution, the reaction rate is independent of the epoxide concentration and the carbon
monoxide pressure but first order in 1. The rate of lactone formation varies considerably in different solvents
and depends primarily on the coordinating ability of the solvent. In mixtures of THF and cis/trans-2,5-
dimethyltetrahydrofuran, the reaction is first order in THF. From spectroscopic and kinetic data, the catalyst
resting state was assigned to be the neutral (â-aluminoxy)acylcobalt species (salph)AlOCH(Et)CH2COCo-
(CO)4 (3a), which was successfully trapped with isocyanates. As the formation of 3a from EB, CO, and 1
is rapid, lactone ring closing is rate-determining. The favorable impact of donating solvents was attributed
to the necessity of stabilizing the aluminum cation formed upon generation of the lactone.
Introduction
epoxides are easily obtained by well-established methods such
as hydrolytic kinetic resolution18 and asymmetric epoxida-
â-Lactones are an important class of heterocycles having a
wide range of applications in synthetic,1-3 natural product,2,4
and polymer5 chemistry. A number of synthetic methods
yielding â-lactones exist, and most can be described as
intramolecular substitutions, cycloadditions, ring expansions, or
ring contractions.1,3,4,6 Among these reaction types, two have
received considerable recent attention due to their ability to
produce enantiopure â-lactones. The discovery of chiral catalysts
(nucleophilic,7-9 Lewis acidic,10-14 or both15,16) has made net
[2+2] addition of ketenes and aldehydes a useful route to
stereopure â-lactones, and this topic has been reviewed.6,17 As
well, recent research efforts have begun to focus on catalytic
epoxide carbonylation, which is attractive because enantiopure
tion.19,20 The first contemporary report of epoxide carbonylation
directed toward the synthesis of â-lactones, by Drent and
Kragtwijk, described the use of Co2(CO)8 and 3-hydroxypyridine
to carbonylate epoxides (in particular propylene oxide, PO),
producing mixtures of lactone and oligoesters.21 Following this
discovery, Alper and co-workers reported increased activity and
selectivity for lactone using combinations of [PPN]+[Co(CO)4]-
and neutral Lewis acids (PPN ) bis(triphenylphosphine)-
iminium).22 Carbonylation of a number of substituted epoxides
by this catalyst system was demonstrated; however, turnover
frequencies were limited to <13 h-1. Subsequently, we reported
the bimetallic salt [(salph)Al(THF)2]+[Co(CO)4]- (1, salph )
N,N′-o-phenylenebis(3,5-di-tert-butylsalicylideneimine), THF )
tetrahydrofuran), which catalyzes the selective carbonylation of
epoxides to lactones at lower temperature and without the use
of solvent (Scheme 1).23 We have since reported additional
catalysts for this reaction, all of the general form [Lewis
acid]+[Co(CO)4]-,24-28 and have expanded their substrate scope
to include aziridines,24 lactones,29 oxetane,29 and a diverse array
† Present address: Department of Chemistry, Kenyon College, Gambier,
Ohio, 43022.
(1) Pommier, A.; Pons, J.-M. Synthesis 1993, 441-459.
(2) Wang, Y.; Tennyson, R. L.; Romo, D. Heterocycles 2004, 64, 605-658.
(3) Schneider, C. Angew. Chem., Int. Ed. 2002, 41, 744-746.
(4) Pommier, A.; Pons, J.-M. Synthesis 1995, 729-744.
(5) Mu¨ller, H.-M.; Seebach, D. Angew. Chem., Int. Ed. Engl. 1993, 32, 477-
502.
(6) Yang, H. W.; Romo, D. Tetrahedron 1999, 55, 6403-6434.
(7) Wynberg, H. Top. Stereochem. 1986, 16, 87-129.
(8) Song, C. E.; Ryu, T. H.; Roh, E. J.; Kim, I. O.; Ha, H.-J. Tetrahedron:
Asymmetry 1994, 5, 1215-1218.
(18) Tokunaga, M.; Larrow, J. F.; Kakiuchi, F.; Jacobsen, E. N. Science 1997,
277, 936-938.
(9) Calter, M. A. J. Org. Chem. 1996, 61, 8006-8007.
(10) Tamai, Y.; Yoshiwara, H.; Someya, M.; Fukumoto, J.; Miyano, S. J. Chem.
Soc., Chem. Commun. 1994, 2281-2282.
(19) Sharpless, K. B.; Woodward, S. S.; Finn, M. G. Pure Appl. Chem. 1983,
55, 1823-1836.
(20) Larrow, J. F.; Jacobsen, E. N. Topics Organomet. Chem. 2004, 6, 123-
152, and references therein.
(11) Dymock, B. W.; Kocienski, P. J.; Pons, J.-M. Chem. Commun. 1996, 1053-
1054.
(21) Drent, E.; Kragtwijk, E. (Shell Internationale Research Maatschappij, B.
V., Neth.). Eur. Pat. Appl. EP 577206; Chem. Abstr. 1994, 120, 191517c.
(22) Lee, J. T.; Thomas, P. J.; Alper, H. J. Org. Chem. 2001, 66, 5424-5426.
(23) Getzler, Y. D. Y. L.; Mahadevan, V.; Lobkovsky, E. B.; Coates, G. W. J.
Am. Chem. Soc. 2002, 124, 1174-1175.
(12) Yang, H. W.; Romo, D. Tetrahedron Lett. 1998, 39, 2877-2880.
(13) Evans, D. A.; Janey, J. M. Org. Lett. 2001, 3, 2125-2128.
(14) Nelson, S. G.; Zhu, C.; Shen, X. J. Am. Chem. Soc. 2004, 126, 14-15.
(15) Zhu, C.; Shen, X.; Nelson, S. G. J. Am. Chem. Soc. 2004, 126, 5352-
5353.
(24) Mahadevan, V.; Getzler, Y. D. Y. L.; Coates, G. W. Angew. Chem., Int.
Ed. 2002, 41, 2781-2784.
(25) Getzler, Y. D. Y. L.; Mahadevan, V.; Lobkovsky, E. B.; Coates, G. W.
Pure Appl. Chem. 2004, 76, 557-564.
(16) Calter, M. A.; Tretyak, O. A.; Flaschenreim, C. Org. Lett. 2005, 7, 1809-
1812.
(17) Orr, R. K.; Calter, M. A. Tetrahedron 2003, 59, 3545-3565.
9
10.1021/ja061503t CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 10125-10133
10125

A R T I C L E S
Church et al.
Scheme 1. Catalytic Carbonylation of Epoxides to Lactones by
[(salph)Al(THF)2]+[Co(CO)4]-, 1
Scheme 2. Preliminary Mechanism for Epoxide Carbonylation by
Catalysts of the Type [LnM]+[Co(CO)4]-
of functionalized and bicyclic epoxides.27,28 Recently, research-
ers collaborating at BASF and University of Ulm have carbo-
nylated PO to â-butyrolactone (â-BL) using catalyst mixtures
composed of readily available Lewis acids and cobalt carbonyl
sources, and used spectroscopic and theoretical studies to
examine the effect of catalyst precursors on the reaction.30-32
In a related study, Kim et al. have used Co2(CO)8 with a number
of ligated zinc salts to carbonylate ethylene oxide to methyl
â-hydroxypropionate in methanol solution.33
a Lewis acid must be present; reactions using the [Co(CO)4]-
ion with non Lewis acidic cations, such as [PPh4]+, [PPN]+,
[Cp2Co]+, or [nBu4N]+, were not successful.22,23,32 Interestingly,
lactone is produced in the presence of Na+,23 a very weak Lewis
acid.35 As Lewis acids catalyze the ring-opening of epoxides,36,37
they likewise have the potential to activate the epoxide substrate
for the carbonylation reaction.
When catalytic carbonylation of epoxides is attempted in the
absence of CO or with low pressures of CO, epoxide isomer-
ization to ketone (the Meinwald rearrangement38) is also
observed. Catalysis of this type of rearrangement by [Lewis
acid]+[Co(CO)4]- has been reported for the conversion of PO
to acetone using methanolic solutions of Co2(CO)8.39 These
solutions contain [Co(MeOH)6]2+ and [Co(CO)4]- (in a 1:2
ratio) as the primary cobalt species.40 Regiochemical evidence
is consistent with the ions [Co(MeOH)6]2+ and [Co(CO)4]-
effecting epoxide rearrangement via a (â-metalloxy)alkylcobalt
intermediate (Scheme 2, 2).39,41 This type of structure is also
accessible in our system, and 2 is believed to be on the reaction
path to lactone.
The preliminary mechanism we proposed for the carbonyla-
tion of epoxides by 1 and related catalysts (of the form [Lewis
acid]+[Co(CO)4]-)23-25,27,28 is based on these observations and
on stereochemical investigations (vide infra). The epoxide, once
activated by coordination to the Lewis acid, is ring-opened
through nucleophilic attack by [Co(CO)4]- to form intermediate
2. Migratory insertion of CO into the cobalt-alkyl bond forms
a cobalt acyl (intermediate 3) that subsequently undergoes an
intramolecular attack by the metal alkoxide to form a four-
membered lactone ring and release the catalyst.
The skeletal reaction sequence of (I) substrate activation, (II)
nucleophilic ring-opening, (III) CO insertion and uptake, and
(IV) product ring-closing is consistent with catalytic cycles
proposed for epoxide,23-25,27,28,30-32,34 aziridine,24,42 and lactone29
carbonylation in our and related systems. Alternatively, Stirling
We23-25,27-29 and others30-32,34 have discussed mechanistic
aspects of epoxide carbonylation using Lewis acid/[Co(CO)4]-
catalysts. The coordination environment around the cobalt center
has been studied in detail, as the attached carbonyl ligands
provide convenient and informative spectroscopic handles.
Specifically, in situ IR spectroscopy has been exploited for this
purpose.30-32 Conversely, the transformations taking place at
the Lewis acid component of the catalyst have received less
consideration in mechanistic studies. This is not surprising, as
many of the commercially available Lewis acids used with this
system (in particular aluminum alkyls) not only lack con-
venient spectroscopic handles, but also exhibit complex solution
behavior.
In an effort to develop improved catalysts for heterocycle
carbonylation, we sought a detailed mechanism of lactone
formation. Specifically, we aimed to elucidate the complete
catalytic cycle, identify the form of the catalyst during the
reaction, and uncover the rate-determining step. The discrete,
well-defined nature of our ligand-supported Lewis acid cations
proved advantageous in this endeavor. Herein, we present
kinetic, spectroscopic, labeling, and reactivity data that permit
the construction of a detailed mechanistic picture of epoxide
carbonylation by 1. We believe this study will set the stage for
similar investigations with related catalysts of the type [Lewis
acid]+[M(CO)n]- and with different substrate classes.
Results and Discussion
Analysis of the catalyst components necessary for epoxide
carbonylation, as well as the reaction products, has provided
an initial understanding of the course of this reaction. Notably,
(26) Coates, G. W.; Getzler, Y. D. Y. L.; Wolczanski, P.; Mahadevan, V. (Cornell
Research Foundation, Inc. USA). PCT Int. Pat. WO 2003050154; Chem.
Abstr. 2003, 139, 54560.
(27) Schmidt, J. A. R.; Mahadevan, V.; Getzler, Y. D. Y. L.; Coates, G. W.
Org. Lett. 2004, 6, 373-376.
(28) Schmidt, J. A. R.; Lobkovsky, E. B.; Coates, G. W. J. Am. Chem. Soc.
2005, 127, 11426-11435.
(35) Fukuzumi, S.; Ohkubo, K. Chem.-Eur. J. 2000, 6, 4532-4535.
(36) Parker, R. E.; Isaacs, N. S. Chem. ReV. 1959, 59, 737-799.
(37) Smith, M. B.; March, J. March’s AdVanced Organic Chemistry, 5th ed.;
John Wiley & Sons: New York, 2001.
(29) Getzler, Y. D. Y. L.; Kundnani, V.; Lobkovsky, E. B.; Coates, G. W. J.
Am. Chem. Soc. 2004, 126, 6842-6843.
(30) Molnar, F.; Luinstra, G. A.; Allmendinger, M.; Rieger, B. Chem.-Eur. J.
2003, 9, 1273-1280.
(38) Meinwald, J.; Labana, S. S.; Chadha, M. S. J. Am. Chem. Soc. 1963, 85,
582-585.
(31) Allmendinger, M.; Eberhardt, R.; Luinstra, G. A.; Molnar, F.; Rieger, B.
Z. Anorg. Allg. Chem. 2003, 629, 1347-1352.
(39) Eisenmann, J. L. J. Org. Chem. 1962, 27, 2706.
(40) Wender, I.; Sternberg, H. W.; Orchin, M. J. Am. Chem. Soc. 1955, 74,
1216-1219.
(32) Allmendinger, M.; Zintl, M.; Eberhardt, R.; Luinstra, G. A.; Molnar, F.;
Rieger, B. J. Organomet. Chem. 2004, 689, 971-979.
(33) Kim, H. K.; Bae, J. Y.; Lee, J. S.; Jeong, C. I.; Choi, D. K.; Kang, S. O.;
Cheong, M. Appl. Catal., A 2006, 301, 75-78.
(41) Prandi, J.; Namy, J. L.; Menoret, G.; Kagan, H. B. J. Organomet. Chem.
1985, 285, 449-460.
(34) Stirling, A.; Iannuzzi, M.; Parrinello, M.; Molnar, F.; Bernhart, V.; Luinstra,
G. A. Organometallics 2005, 24, 2533-2537.
(42) Davoli, P.; Moretti, I.; Prati, F.; Alper, H. J. Org. Chem. 1999, 64, 518-
521.
9
10126 J. AM. CHEM. SOC. VOL. 128, NO. 31, 2006

Epoxide Carbonylation by [Lewis Acid]+[Co(CO)4]-
A R T I C L E S
Figure 2. Effect of the concentration of catalyst 1 on the rate of
â-valerolactone formation from 1,2-epoxybutane (EB). Reactions performed
in 1,2-dimethoxyethane solution and monitored using in situ IR spectros-
copy. [EB]0:[1] ) 110 ( 1, PCO ) 300 ( 10 psi, T ) 25.0 ( 0.5 °C.
Figure 1. Plot of the formation of â-valerolactone from 1,2-epoxybutane
(EB) as a function of time. Reaction performed in 1,2-dimethoxyethane
solution and monitored using in situ IR spectroscopy. [1] ) 4.6 mM, [EB]0
) 0.51 M, PCO ) 300 ( 10 psi, T ) 25.0 ( 0.5 °C.
coincidental effects from the starting epoxide and product
lactone, the carbonylation was performed with â-VL present
initially ([â-VL]0 ) 0.32 M). Notably, this did not alter the
reaction rate; the time required for complete epoxide consump-
tion was unchanged ((5 min). Thus, neither epoxide nor lactone
concentration affect the rate of reaction, and epoxide binding
is eliminated as the rate-determining step.
As the octahedral aluminum in 1 is coordinatively saturated,
the intimate mechanism of substitution (replacing THF with an
epoxide) at this center is tentatively assigned as dissociative.
This is consistent with the lack of order in epoxide, though a
rate-limiting dissociation of THF can be neither verified nor
excluded based solely on this observation.
and co-workers34 have proposed that another mechanism, in
which ring closing precedes CO uptake, may also be operable
under some conditions; this possibility will be discussed later.
The catalytic cycle as depicted in Scheme 2 served as the starting
point from which we conducted our mechanistic investigation.
(I) Activation of the Epoxide Substrate by Catalyst 1. We
undertook kinetic studies to determine the order of the carbo-
nylation reaction in each of its components (eq 1). Though the
d[â-VL]/dt ) k[EB]xPCOy[1]z
(1)
carbonylation of PO to â-BL has been the most studied reaction
of this type, we chose to examine the formation of â-valero-
lactone (â-VL) from 1,2-epoxybutane (EB) in 1,2-dimethoxy-
ethane (DME) solution. EB has a higher boiling point than PO,
which makes it easier to manipulate quantitatively at room
temperature. In situ IR spectroscopy43 has been used to study
the carbonylative formation â-lactones,30-32 polyesters,44,45 and
poly-â-peptoids46 and was used in this study to monitor the
emergence of the lactone carbonyl stretch (for â-VL in DME,
ν(CdO) ) 1827 cm-1). As the intensity of this band was found
to vary nonlinearly with the concentration of â-VL (particularly
at high concentration), a calibration curve was constructed from
31 independently prepared samples of â-VL in DME and fit to
a function such that the concentration of â-VL could be
calculated directly from the absorbance.47 Figure 1 shows a plot
of â-VL concentration versus time for a representative experi-
ment in which the pressure of CO was held constant while EB
was consumed. The reaction rate, d[â-VL]/dt, was unchanged
throughout the reaction, indicating that the rate of lactone
formation is independent of epoxide concentration (viz., x )
0). Thus, for a given CO pressure and concentration of 1, the
rate law is given by eq 2 . This suggests that EB does not enter
(II) Opening of the Epoxide Ring. The regio- and stereo-
chemical outcomes of the carbonylation of epoxides as catalyzed
by complex 1 provide insight regarding the ring-opening event.
In most cases, the carbonylation of a monosubstituted epoxide
by 1 yields the corresponding â-substituted â-lactone as the
exclusive lactone product, indicating that attack by [Co(CO)4]-
occurs at the less-substituted carbon atom.48 Further, carbony-
lation of enantiopure (R)-PO (Scheme 2, R ) (R)-Me, R′ ) H)
occurs with retention of stereochemistry, yielding (R)-â-BL.23
Finally, cis- and trans-2-butene oxide afford trans- and cis-
R,â-dimethylpropiolactone, respectively.24,25 Therefore, attack
of the [Co(CO)4]- ion on the epoxide occurs in an SN2
fashion,24,25 leading to inversion of stereochemistry at the car-
bon atom R to the inserted CO, and retention at the â-car-
bon. Similarly, SN2 pathways have been assigned for the
49
stoichiometric ring opening of epoxide by HCo(CO)4 and
Ph3SiCo(CO)4.50
The kinetics for the carbonylation of EB were also measured
while varying the catalyst concentration (Figure 2) to determine
the reaction order in 1. As the concentration of EB does not
affect the rate of carbonylation, this study was carried out while
maintaining a constant ratio of [EB]0/[1].51 From these experi-
ments, we found that the rate of â-VL formation depended
linearly on the concentration of catalyst 1, so the reaction is
d[â-VL]/dt ) k[EB]0PCOy[1]z
(2)
the catalytic cycle between the resting state and the transition
state of the rate-determining step. To rule out the possibility of
(48) Carbonylations of the substrates isobutylene oxide23 and styrene oxide
(Mahadevan, V.; Getzler, Y. D. Y. L.; Coates, G. W.; unpublished results)
yield some R-substituted-â-lactones, indicating that reaction of [Co(CO)4]-
with the more substituted carbon is possible for the special cases of aryl-
substituted or gem-disubstituted epoxides.
(43) Mettler Toledo Autochem Inc.: Columbia, MD.
(44) Allmendinger, M.; Eberhardt, R.; Luinstra, G.; Rieger, B. J. Am. Chem.
Soc. 2002, 124, 5646-5647.
(49) Kreisz, J.; Ungva´ry, F.; Sisak, A.; Marko´, L. J. Organomet. Chem. 1991,
417, 89-97.
(45) Allmendinger, M.; Molnar, F.; Zintl, M.; Luinstra, G. A.; Preishuber-Pflu¨gl,
P.; Rieger, B. Chem.-Eur. J. 2005, 11, 5327-5332.
(46) Darensbourg, D. J.; Phelps, A. L.; Le Gall, N.; Jia, L. J. Am. Chem. Soc.
2004, 126, 13808-13815.
(50) Kreisz, J.; Sisak, A.; Marko´, L.; Ungva´ry, F. J. Organomet. Chem. 1993,
451, 53-57.
(51) Consequentially, potential side reactions of 1 with undetected impurities
in the EB were not magnified at low catalyst concentration.
(47) See Supporting Information for details.
9
J. AM. CHEM. SOC. VOL. 128, NO. 31, 2006 10127

A R T I C L E S
Church et al.
exceptions cannot be prepared in the presence of donor
solvents.57 Conversely, aluminum alkoxides and siloxides sup-
ported by these ligands are five-coordinate at room temperature
in solution (C6D6 and CDCl3) and in the solid phase, even when
crystallized from donor solvents.56,58-64 Further, the binding of
4-(N,N-dimethylamino)pyridine, a strong donor, to the aluminum
alkoxide (TPP)AlOR has been shown to be much weaker than
to the related carbonate and carboxylate, (TPP)AlO2COR and
(TPP)AlO2CR (TPP ) tetraphenylporphyrin), respectively.60
Given this preference of salen(tBu)-supported aluminum alkox-
ides for pentacoordination, the conversion of the aluminum
center from a cation to a neutral alkoxide likely results in the
dissociation of the trans-coordinated solvent molecule. There-
fore, we formulate the aluminum-alkoxide/cobalt-alkyl species
that results from opening of EB as (salph)AlOCH(Et)CH2Co-
(CO)4 (Scheme 2, intermediate 2, LnM ) (salph)Al, R ) Et,
R′ ) H).
Figure 3. IR spectra of (a) catalyst 1 in 1,2-dimethoxyethane solution, (b)
immediately following addition of 1,2-epoxybutane and CO (t ) 0 min),
(c) t ) 1 min, and (d) t ) 5 min.
first-order in 1 (eq 3) . The reaction rate was unaltered by the
(III) Insertion of Carbon Monoxide into the Cobalt-Alkyl
Bond. The effect of carbon monoxide pressure on the reaction
rate was observed by in situ IR spectroscopy. From four runs
with [EB]0 ) 1.24 ( 0.01 M, PCO ) 300 ( 10 psi, [1] ) 11.6
( 0.1 mM, and T ) 25.0 ( 0.5 °C, the reaction rate was
calculated to be 6.2 ( 0.5 mM‚min-1. The same reaction was
repeated at a range of CO pressures from 220 to 1045 psi (see
Table S1, Supporting Information). In each case, the rate was
found to be within error of the rate determined for PCO ) 300
psi. Thus, the reaction rate is independent of CO pressure over
the range considered (eq 4).
d[â-VL]/dt ) k[EB]0PCOy[1]
(3)
addition of excess [(salph)Al(THF)2]+ (as the [BPh4]- salt) or
[Co(CO)4]- (as the [PPh4]+ salt). As the activated epoxide is
attacked by [Co(CO)4]- in an SN2 fashion, the absence of an
order in [Co(CO)4]- suggests that either (a) the epoxide ring
opening is not rate-limiting, or (b) following the dissociative
substitution of an epoxide for THF on the cation of 1, attack
by [Co(CO)4]- occurs within the ion pair [(salph)Al(THF)
(EB)]+[Co(CO)4]-, to the exclusion of other ions. These situa-
tions can be distinguished by the form of the cobalt in solution.
If the ring opening is fast, then [Co(CO)4]- should not be
present, whereas it should be observed in the event of slow
substitution within an ion pair. Using in situ IR spectroscopy,
we monitored the [Co(CO)4]- ion directly over the course of
the carbonylation (Figure 3). The IR spectrum of 1 in DME
(Figure 3a) shows a strong absorption at 1887 cm-1, assigned
to the ν(CO) of the [Co(CO)4]- ion (cf. 1885 cm-1 for solid-
state 123). Within one minute of the addition of EB and CO,
this band disappears completely (Figure 3b); the ν(CdO) for
lactone then appears at 1827 cm-1 (Figure 3c,d). Given the
immediate and total disappearance of [Co(CO)4]- from solution,
epoxide ring opening must be very fast on the time scale of the
overall reaction and, therefore, is not the rate-determining step.
Although [Co(CO)4]- is only a modest nucleophile,52,53 rapid
ring opening is not unexpected in this system; computational
studies found a low barrier to ring opening for epoxides activated
by strong Lewis acids30,34 and, in particular, by aluminum
cations.30
d[â-VL]/dt ) k[EB]0PCO0[1]
(4)
The insertion of CO into the cobalt-alkyl bond was also
observed spectroscopically. The IR spectra of tetracarbonyl-
cobalt alkyls and acyls have been extensively studied.65
Although the spectra for the alkyl and corresponding acyl species
are very similar in most cases, the latter type contains an
additional ν(CdO) absorbance associated with the cobalt acyl
group. The frequency of this stretch depends on the nature of
the alkyl/acyl group but is generally close to 1700 cm-1 65
. When
we added EB and CO to a solution of 1, we observed a broad
absorption, centered at 1715 cm-1 (Figure 3b-d), which
persisted in solution during the carbonylation. This band could
be assigned to the ν(CdO) of a cobalt acyl; similar assignments
(of bands at 1717 or 1710 cm-1) have been made in related
systems.30-32,45,46 However, catalysts of the type [Lewis
acid]+[Co(CO)4]- are known to produce small amounts of
ketone (isomeric to the starting epoxide) as byproducts of
epoxide carbonylation,22,23,30 and these are also expected to have
The cationic [(salph)Al(THF)(EB)]+ is converted to a neutral
aluminum alkoxide upon the ring opening of epoxide. In both
solid and solution phases, aluminum cations supported by
ligands of the salen(tBu) family (i.e., having the general form
N,N′-alkylene (or arylene) bis(3,5-di-tert-butylsalicylidene-
imine)) are almost always six-coordinate;54-56 the only reported
(57) Mun˜oz-Herna´ndez, M.-AÄ .; Sannigrahi, B.; Atwood, D. A. J. Am. Chem.
Soc. 1999, 121, 6747-6748.
(58) Mun˜oz-Herna´ndez, M.-AÄ .; Keizer, T. S.; Parkin, S.; Zhang, Y.; Atwood,
D. A. J. Chem. Cryst. 2000, 30, 219-222.
(59) Mun˜oz-Herna´ndez, M.-AÄ .; Keizer, T. S.; Wei, P.; Parkin, S.; Atwood, D.
A. Inorg. Chem. 2001, 40, 6782-6787.
(60) Chisholm, M. H.; Zhou, Z. J. Am. Chem. Soc. 2004, 126, 11030-11039.
(61) Chen, P.; Chisholm, M. H.; Gallucci, J. C.; Zhang, X.; Zhou, Z. Inorg.
Chem. 2005, 44, 2588-2595.
(52) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G. Principles and
Applications of Organotransition Metal Chemistry; University Science
Books: Sausalito, CA, 1987.
(53) Dessy, R. E.; Pohl, R. L.; King, R. B. J. Am. Chem. Soc. 1966, 88, 5121-
(62) Wang, Y.; Bhandari, S.; Parkin, S.; Atwood, D. A. J. Organomet. Chem.
2004, 689, 759-765.
5124.
(63) A six-coordinate salen(tBu)Al tosylate exists. Jegier, J. A.; Mun˜oz-
Herna´ndez, M.-AÄ .; Atwood, D. A. J. Chem. Soc., Dalton Trans. 1999,
2583-2588.
(54) Mun˜oz-Herna´ndez, M.-AÄ .; McKee, M. L.; Keizer, T. S.; Yearwood, B. C.;
Atwood, D. A. J. Chem. Soc., Dalton Trans. 2002, 410-414.
(55) Mun˜oz-Herna´ndez, M.-AÄ .; Parkin, S.; Yearwood, B.; Wei, P.; Atwood, D.
A. J. Chem. Cryst. 2000, 30, 215-218.
(56) Atwood, D. A.; Harvey, M. J. Chem. ReV. 2001, 2001, 37-52, and
references therein.
(64) The five-coordinate salen(tBu)AlOMe complex coordinates MeOH at -30
°C.58
(65) Kova´cs, I.; Ungva´ry, F. Coord. Chem. ReV. 1997, 161, 1-32, and references
therein.
9
10128 J. AM. CHEM. SOC. VOL. 128, NO. 31, 2006

Epoxide Carbonylation by [Lewis Acid]+[Co(CO)4]-
A R T I C L E S
Scheme 3. Equilibria between Alkyl- and Acyl-Cobalt Carbonyls
Scheme 4. Proposed Resting State 3a in the 1-Catalyzed
Carbonylation of 1,2-Epoxybutanea
ν(CdO) absorptions at ∼1715 cm-1 66
. To confirm that the peak
at 1715 cm-1 is due to a cobalt acyl, we repeated the
carbonylation reaction using 13CO. We expected this isotopic
substitution to affect the stretching frequency of cobalt-acyl
intermediates but not of ketones formed from isomerization of
epoxide. In fact, the IR spectra of the reaction run with 13CO
revealed multiple ν(CdO) peaks. First, a new band at 1783 cm-1
grew in during the reaction and was assigned to ν(13CdO) of
â-VL (cf. 1786 cm-1 predicted by Hooke’s Law66). In addition,
two smaller bands were present immediately after adding EB
and CO. One was similar to that observed in the spectra of the
unlabeled reaction, though less broad, and centered at 1719
cm-1; this is attributed to small amounts of 2-butanone
(confirmed to be present by GC of the final reaction mixture).
The other, a new peak, appeared at 1677 cm-1, which agrees
with the predicted ν(13CdO) of a cobalt acyl. Thus, in spectra
collected with unlabeled CO, both ketone and cobalt-acyl
moieties coincidentally absorb at ∼1715 cm-1. However, these
species can be distinguished using isotopically labeled CO. In
the present system, a persistent cobalt-acyl species appears in
solution immediately upon addition of epoxide and CO, indicat-
ing that carbon monoxide insertion is rapid.
The insertion of CO into alkylcobalt tetracarbonyl and related
alkylcobalt monophosphine tricarbonyl complexes has been
examined in detail for a variety of systems65,67 and is generally
thought to occur in two steps (Scheme 3, b and c).52,68 First, an
intramolecular CO insertion yields an unsaturated intermediate,
which is believed to be stabilized by η2-coordination of the
cobalt acyl.67,69
The rapidity of the CO insertion in this system is not
surprising, given the propensity of saturated alkylcobalt carbo-
nyls to insert CO. In the cases of primary R groups without
electron-withdrawing substituents, equilibrium a in Scheme 3
lies far to the acyl side.70 The rate of the reverse reaction has
been studied for R ) Me; release of CO from acetylcobalt
tetracarbonyl in an argon atmosphere is rapid.71 Further, as the
equilibrium strongly favors the acyl complex in that case,71 the
insertion of CO into the cobalt-alkyl bond is faster, particularly
under the conditions of our study (i.e., 220 psi e PCO e 1045
psi). Indeed, an NMR spectroscopic study of 13CO exchange
between MeCOCo(CO)4 and free CO under comparable pres-
sures found acyl formation to be significantly faster than alkyl
formation.72 It is the swift nature of CO insertion into the
cobalt-alkyl bond that allows lactone, rather than ketone, to
be produced in this system. At low pressures of CO, intermediate
a (a) Formation of 3a from [rac-(salph)AlOCH(Et)CH2Cl] (4) and
Na+[Co(CO)4]-. (b) Decay of 3a to â-VL. (c) Trapping of 3a with
isocyanates to form 1,3-oxazine-2,4-diones, OD. (d) Reaction of the model
aluminum alkoxide 4 with isocyanate.
2 can be diverted to ketone. A plausible mechanism for this
rearrangement has been proposed,41 though it is supported
only by regiochemical data. In it, â-H elimination occurs from
a (â-metalloxy)cobaltalkyl (such as 2) to form a vinyloxy metal
species and HCo(CO)4. Protonation of the vinyloxy group and
subsequent tautomerization yields the ketone and regenerates
the catalyst.
Calculations suggest that during catalytic epoxide carbony-
lation, the alkoxide oxygen, still bound to a Lewis acid, may
stabilize the RCOCo(CO)3 intermediate through the reversible
formation of a 2-metalla-3-oxo-furan species.34 Though this
process was calculated to be competitive with CO uptake when
BF3 was used as a Lewis acid, the steric bulk of the salph ligand
would likely preclude its formation in our system. The same
calculations found that lactone ring closing could actually
precede CO uptake when a soft Lewis acid (such as B(CH3)3)
was used, but that this pathway is very high in energy (∆Hq ≈
35 kcal/mol) for the harder, more Lewis acidic BF3. The
aluminum cation used in our system is a hard Lewis acid,73,74
so this pathway is not expected to be significant.
The Resting State of Catalyst 1. We have shown that steps
I-III in Scheme 2 are fast on the time scale of the catalysis, as
(1) neither epoxide nor CO is represented in the rate equation,
(2) in situ IR spectroscopy indicates that [Co(CO)4]- is not ob-
served in solution during lactone formation, and (3) the ν(CdO)
for cobalt acyl appears immediately upon CO addition. There-
fore, so long as both epoxide and CO remain in the reaction
mixture, 1 exists not as the ion pair [(salph)Al(solvent)x]+
[Co(CO)4]- but as the neutral aluminum-alkoxide/cobalt-acyl
3a, as depicted in Scheme 4 (cf. Scheme 2, intermediate 3, where
LnM ) (salph)Al, R ) Et, R′ ) H).
(66) Pavia, D. L.; Lampman, G. M.; Kriz, G. S. Introduction to Spectroscopy,
2nd ed.; Harcourt Brace College Publishers: New York, 1996.
(67) Ford, P. C.; Massick, S. Coord. Chem. ReV. 2002, 226, 39-49.
(68) Elschenbroich, C.; Salzer, A. Organometallics: A Concise Introduction,
2nd ed.; VCH Publishers Inc.: New York, 1992.
(69) Sweany, R. L. Organometallics 1989, 8, 175-179.
(70) Galamb, V.; Pa´lyi, G. Coord. Chem. ReV. 1984, 59, 203-238, and
references therein.
Evidence that 3a could be an intermediate in the carbonylation
of epoxides by 1 was obtained from its independent synthe-
(71) Ungva´ry, F.; Marko´, L. Inorg. Chim. Acta 1994, 227, 211-213.
(72) Roe, D. C. Organometallics 1987, 6, 942-946.
(73) Pearson, R. G. Inorg. Chem. 1988, 27, 734-740.
(74) Pearson, R. G. Inorg. Chim. Acta 1995, 240, 93-98.
9
J. AM. CHEM. SOC. VOL. 128, NO. 31, 2006 10129

A R T I C L E S
Church et al.
sis. When a model aluminum-alkoxide, [rac-(salph)AlOCH-
(Et)CH2Cl] (4), was combined with Na+[Co(CO)4]- in DME
solution and under a pressure of 300 psi of carbon monoxide,
there was no reaction at room temperature. Heating the reaction
mixture to 80 °C resulted in the appearance of â-VL in solution.
We propose that 3a is formed through a slow substitution of
[Co(CO)4]- onto the alkyl chloride of 4 and subsequent CO
insertion into the new cobalt-alkyl bond (Scheme 4a). The
formation of â-VL then occurs by ring closing (Scheme 4b).
Unfortunately, we were unable to measure the rate for this
process, as the substitution reaction was rate-determining.
However, assuming that the substitution reaction produced 3a,
the ensuing â-VL production supports that 3a is on the reaction
path to lactone.
Further, as we suspected that 3a is the resting state in the
catalytic cycle, we attempted to intercept it via reaction of the
aluminum-alkoxide terminus with electrophiles. We used
isocyanates for this purpose because they do not participate in
side reactions with the catalyst, the epoxide, CO, or the lac-
tone product. We predicted that the aluminum alkoxide of 3a
would react with isocyanate to form an aluminum carbamate,
which could then attack the cobalt-acyl moiety to form a
six-membered 1,3-oxazine-2,4-dione (OD, Scheme 4c). When
the carbonylation of EB in DME was run in the presence of
4-Me-C6H4NCO ([ArNCO] ) 0.50 M, [1] ) 0.0115 M, [EB]
) 0.55 M), â-VL was the only product detected in the crude
reaction mixture; the isocyanate was unchanged. However, when
the more electrophilic 4-F-C6H4NCO was used under identical
conditions, the products were 93% â-VL and 7% OD (in this
case 6-ethyl-3-(4-fluorophenyl)-1,3-oxazine-2,4-dione). That
intermediate 3a can be diverted during the carbonylation reaction
suggests that it has a significant lifetime in solution, consistent
with its assignment as the resting state.
(IV) Closing of the Lactone Ring. With 3a as the cata-
lyst resting state, the rate-determining step of the reaction
must occur during the lactone ring-closing or dissociation events.
To gain further insight into these processes, we examined the
effect of solvent on the catalysis. Although epoxide carbony-
lation by 1 can be run in neat substrate,23 the rate law was
measured in DME, and the reaction was also run in a num-
ber of other solvents. The rate of carbonylation was found to
vary widely depending on the reaction solvent. It was gen-
erally fastest in moderately polar solvents, such as ethers (at
25 °C, the reaction is much faster in DME than in neat epox-
ide). In nonpolar or weakly polar solvents, such as hexanes or
toluene, catalyst 1 was insoluble, and the overall reaction was
slowed. Although solutions of 1 are generally red, addition of
the catalyst to very polar, strongly donating, acetonitrile re-
sulted in a yellow solution. Carbonylation in this solvent was
slow and first-order in epoxide, indicative of a change in
mechanism or in rate-determining step compared to the DME
case. We suspect that one or both THF molecules of 1 are
displaced by CH3CN, and that epoxide must displace these
strong bases prior to carbonylation. Finally, the use of halo-
methane solvents resulted in a severe reduction of catalyst
activity, presumably due to reaction of [Co(CO)4]- with these
compounds.75
Figure 4. Comparison of the rate of carbonylation of 1,2-epoxybutane (EB)
by catalyst 1 in various solvents. ꢀ ) dielectric constant.76 [1] ) 11.6 (
0.1 mM, [EB]0 ) 1.25 ( 0.01 M, PCO ) 300 ( 10 psi, T ) 25.0 ( 0.5 °C.
THF ) tetrahydrofuran, DGM ) diglyme, DME ) 1,2-dimethoxyethane,
THP ) tetrahydropyran, MTHF ) 2-methyltetrahydrofuran, o-DFB ) 1,2-
difluorobenzene.
To investigate the origin of the large solvent dependence of
the reaction, we repeated the carbonylation in a number of
moderately polar solvents, primarily ethers (Figure 4). Though
complete rate laws were not obtained in each solvent, we
confirmed that the reaction was zero-order in epoxide for each
solvent, as it was in DME. Additionally, isocyanate was added
to the slowest and fastest carbonylations to intercept the catalyst
resting state in these solvents. In 1,2-difluorobenzene (o-DFB,
slowest), the carbonylation of EB in the presence of 4-Me-
C6H4NCO ([ArNCO] ) 0.58 M, [1] ) 0.0116 M, [EB] ) 0.63
M) yielded 83% â-VL and 17% OD (cf. 100% â-VL in DME,
vide supra). Because the reaction is slow in o-DFB, 3a is longer-
lived and can be intercepted with the less electrophilic isocya-
nate (Ar ) 4-Me-C6H4), whereas the stronger electrophile
4-F-C6H4NCO was required in DME. In THF (fastest), how-
ever, no OD was detected with either 4-Me-C6H4NCO or
4-F-C6H4NCO present. We believe that this result is likely
due to much faster lactone formation than isocyanate in-
sertion in this solvent, and not to a change in resting state
for the catalyst. To test this hypothesis, we reacted a model
compound, [rac-(salph)AlOCH(Et)CH2Cl] (4), with 4-F-
1
C6H4NCO in THF-d8 and monitored the reaction by H NMR
spectroscopy (Scheme 4d). Under the conditions employed ([4]
) 0.0116 M, [4-F-C6H4NCO] ) 0.67 M, T ) 25 °C), the
reaction was found to proceed quite slowly, with an initial rate
(-d[4]/dt) of 4.3 × 10-6 M/min. Based on this rate and the
rate of epoxide carbonylation in THF, only 0.03% OD is
expected when epoxide carbonylation is run in the presence of
4-F-C6H4NCO, using THF as a solvent.47 Thus, the failure to
observe OD in carbonylations run in THF is attributed to the
inability of isocyanate insertion to compete with rapid lactone
formation.
With the presumption that epoxide carbonylation occurs via
the same mechanism in each of the solvents shown in Figure
4, we continued to analyze the effect of solvent on the rate of
carbonylation. Though moderately polar solvents generally
accelerate the carbonylation, the disparity in rates we observed
cannot be explained solely on polarity. For example, although
(75) Lai, C.-K.; Feighery, W. G.; Zhen, Y.; Atwood, J. D. Inorg. Chem. 1989,
28, 3929-3930.
(76) Wohlfarth, C. In CRC Handbook of Chemistry and Physics, 76th ed.; Lide,
D. R., Ed.; CRC Press: New York, 1995; pp 6-159.
9
10130 J. AM. CHEM. SOC. VOL. 128, NO. 31, 2006

Epoxide Carbonylation by [Lewis Acid]+[Co(CO)4]-
A R T I C L E S
Instead, we propose that the coordination of a Lewis base to
the aluminum center is necessary to facilitate the transformation
from the neutral aluminum alkoxide, which exists in the ground
state, to an aluminum cation. Whereas the aluminum alkoxide
is stable as a five-coordinate species, the cation is not; addition
of a Lewis base is required for stabilization (vide supra).
Coordination of a Lewis base to an unsaturated aluminum cation
is likely favorable and rapid; thus, the presence of solvent in
the rate law indicates that the coordination event must occur
immediately before, or concurrently with, lactone ring closing.
Although we are unable to distinguish between these cases, the
former evokes the ability of neutral, nitrogen-containing donors
to accelerate the catalytic copolymerization of epoxides and CO2
by LAlX complexes (L ) porphyrin, X ) Cl, OR). In this case,
the Lewis base has been proposed to facilitate CO2 insertion
by weakening the Al-O bond of the aluminum-alkoxide
intermediates.60,79
Figure 5. Effect of tetrahydrofuran (THF) concentration on the rate of
carbonylation of 1,2-epoxybutane (EB) to â-valerolactone, in mixtures of
THF/2,5-dimethylTHF. Reaction was monitored by in situ IR spectroscopy,
and initial rates taken for 5 to 15% conversion. [1] ) 11.5 ( 0.1 mM,
[EB] ) 1.27 ( 0.01 M, PCO ) 300 ( 10 psi, T ) 25.0 ( 0.5 °C.
The importance of stabilizing the cationic aluminum accounts
for the strong influence of solvent on the rate of this reaction,
even among solvents having similar polarities. Coordinating
solvents, such as THF or other unhindered ethers, successfully
stabilize the aluminum in this form, so the reaction is fast. Polar
solvents that are weak donors (such as MTHF, DMTHF, or
o-DFB) stabilize the aluminum cation less effectively, and result
in slower carbonylation. Very polar solvents, such as CH3CN,
likely coordinate to the aluminum strongly and inhibit epoxide
binding. In the absence of solvent, the reaction is first-order in
epoxide. In this case, an equivalent of epoxide or lactone, or a
THF molecule from 1, must solvate the aluminum cation,
resulting in an order in epoxide. Presumably, epoxide or lactone
could fulfill this role even when a donor solvent is available.
However, neither epoxide nor lactone is represented in the rate
equation for the solvated reaction. Therefore, the donor sol-
vents examined are considerably better at stabilizing the
aluminum cation, rendering the effects of epoxide and lac-
tone negligible (i.e., ksolvent[donor solvent] . kepoxide[epoxide],
THF, DME, and diglyme (DGM) have very similar dielectric
constants ꢀ,76 the reaction rate using THF as a solvent is nearly
twice that using DGM, and nearly three times that using DME.
Further, the reaction is slower in 2-methyltetrahydrofuran
(MTHF) than in tetrahydropyran (THP), though the former has
a significantly higher dielectric constant. When polar but
noncoordinating o-DFB (ꢀ ) 13.476) was used, the reaction was
extremely slow, reaching only 52% conversion after 28 h (cf.
100% conversion after 90 min in THF). Thus, the polarity of
the solvent used for the carbonylation reaction is not a consistent
predictor of the relative rate of catalysis.
The large rate difference between reactions conducted in THF
and MTHF (Figure 4) suggests that solvent Lewis basicity has
a significant impact on the reaction, as the increased steric bulk
about the O atom of MTHF makes it a weaker donor. To further
probe the effect of solvent basicity on the rate of carbonylation,
we examined the reaction in mixtures of THF and the more
sterically encumbered DMTHF (DMTHF ) 2,5-dimethyltet-
rahydrofuran, 56% cis, 44% trans), which allow the concentra-
tion of THF to be varied with minimum impact on polarity.77
We varied the THF concentration from 3.02 to 12.3 M (neat
THF)78 and found that the rate of carbonylation exhibited a first-
order dependence on THF (Figure 5). Additionally, when the
carbonylation was performed in a 1:1 mixture of DME and
DMTHF, â-VL formation was ∼3.6 times slower than in DME.
Thus, the rate law formulated in eq 4 is incomplete; a solvent
term is necessary (eq 5).
k
lactone[lactone]).
A modified catalytic cycle, which accounts for the new
mechanistic data, is shown in Scheme 5. The first four steps
(epoxide binding, ring opening, CO insertion, and CO uptake)
occur immediately upon addition of epoxide and CO to the
catalyst solution, forming the ground-state structure 3b (Scheme
5; cf. Scheme 2, intermediate 3, where LnM ) (salph)Al). This
species is comprised of a neutral, five-coordinate aluminum
alkoxide and a cobalt acyl. The intramolecular nucleophilic
attack of the aluminum alkoxide of 3b on its cobalt acyl yields
the four-membered lactone ring. As the lactone is formed, the
neutral aluminum alkoxide becomes an aluminum cation, and
the [Co(CO)4]- ion is released. The aluminum cation is highly
electrophilic and must be stabilized by the coordination of a
Lewis base, either before or during ring closing. This coordina-
tion process is analogous to the microscopic reverse of epoxide
ring opening; in that case, a donor molecule is released when
the aluminum alkoxide is formed. The combination of ring
closing and Lewis base uptake is rate-determining; thus, the
role of solvent in this system is critical.
d[â-VL]/dt ) k[EB]0PCO0[1][solvent]
(5)
The appearance of solvent in the rate law for catalysis implies
that a solvent molecule must enter the catalytic cycle between
the resting state and rate-limiting transition state. This could
be explained if a slow, solvent-assisted displacement of the
product lactone from the aluminum center was necessary before
another epoxide could be activated. This possibility is discarded,
however, because the reaction rate is independent of the
concentration of lactone and because [Co(CO)4]- (regenerated
upon lactone ring closing) is not detected in solution during
the reaction.
On the basis of computational and experimental studies of
(79) It should be noted that this differs considerably from the roles played by
nitrogen-containing donors in epoxide/CO2 copolymerizations catalyzed by
(salen)CrX complexes. Darensbourg, D. J.; Mackiewicz, R. J. Am. Chem.
Soc. 2005, 127, 14026-14038.
(77) Wax, M. J.; Bergman, R. G. J. Am. Chem. Soc. 1981, 103, 7028-7030.
(78) The reaction rate was not measured in DMTHF. Due to lower solubility,
an 11.6 mM solution of 1 cannot be prepared in DMTHF.
9
J. AM. CHEM. SOC. VOL. 128, NO. 31, 2006 10131

A R T I C L E S
Church et al.
[rac-(salph)AlOCH(Et)CH2Cl] and Na+[Co(CO)4]-, decay to
â-VL is observed.
Scheme 5. Expanded Mechanistic Cycle for the 1-Catalyzed
Carbonylation of Epoxidesa
Kinetic and solvent studies of the catalytic carbonylation of
1,2-epoxybutane by 1 also indicate that the rate-determining
step in the cycle is generation of lactone product from 3b.
Moreover, this process requires the coordination of a Lewis base
to the aluminum center. Though epoxide can perform this
function, donating solvents such as THF, DME, and DGM do
so more effectively, and result in increased rates of epoxide
carbonylation.
In general, epoxide carbonylation catalysts of the form [Lewis
acid]+[Co(CO)4]- are highly active. As all of these complexes
facilitate very rapid epoxide ring opening, resulting in stable
metal alkoxides, stabilization of the cationic form of the Lewis
acid provides for faster catalyst turnover (as long as epoxide
activation is not compromised). In the present system, we found
that donor solvents are effective for this purpose.80 We
hypothesize that through the use of a ligand that more readily
delocalizes charge on aluminum, the rate of epoxide carbony-
lation can be increased.
As this is the first experimental study to examine each step
of a catalytic epoxide carbonylation, we believe that the insight
gained will be of use in comprehending related catalysts (i.e.,
of the form [Lewis acid]+[M(CO)n]-), and the carbonylation
of other substrates. All catalysts for this reaction must be able
to perform each of the four basic steps shown in Scheme 2.
Rational improvement of a carbonylation catalyst entails an
appreciation of the challenges at each step. In seeking a strategy
for elucidating the mechanism of catalytic epoxide carbonyla-
tion, we found it useful to (1) establish the reaction order in
the substrates and catalyst (as well as its components), (2)
examine the effect of solvent of the reaction, (3) study the IR
spectrum of the catalyst, and (4) attempt to intercept the catalyst
resting state with an isocyanate. Upon the basis of the results
of these mechanistic probes, we have pinpointed the resting state
and rate-determining step of the cycle operative for the case of
catalyst 1. We anticipate that this approach will be valuable for
the study of related catalysts as well and will aid in the creation
of improved catalysts. Investigations into new catalysts are
currently underway in our laboratory, as are mechanistic
explorations of the carbonylation of other substrate types.
a L ) Lewis base (solvent, epoxide, lactone). Neutral aluminum centers
are shown in blue, cationic aluminum centers are in red. rds ) Rate-
determining step.
epoxide carbonylation using [Co(CO)4]- in concert with some
simple Lewis acids, Molnar et al. noted that the relative energetic
barriers for epoxide ring opening and lactone ring closing depend
strongly on the strength of the Lewis acid.30 Not surprisingly,
strong Lewis acids result in facile ring opening but energetically
demanding ring closing; the converse is true for weaker Lewis
acids. An aluminum cation such as [(salph)Al(THF)2]+ is a very
strong Lewis acid, and as such would be expected to result in
fast epoxide ring opening, but slow lactone formation. The high
activity of 1 as an epoxide carbonylation catalyst is attributable
to the presence of weak, highly labile Lewis bases, which may
be solvent or substrate molecules. In reversibly binding to the
aluminum center, they modulate its Lewis acidity, resulting in
both rapid epoxide opening and rapid lactone closing.
Experimental Procedures
General Considerations. All air- and/or water-sensitive reactions
were carried out under dry nitrogen using an MBraun UniLab drybox
or standard Schlenk line techniques. All syringes were gastight and
were dried overnight in a glassware oven prior to use. NMR spectra
were recorded on Varian Mercury 300 (1H, 300 MHz; 13C, 75 MHz),
Bruker ASX 300 (1H, 300 MHz; 13C, 75 MHz), or Varian INOVA-
400 (1H, 400 MHz; 13C, 100 MHz) spectrometers and referenced versus
residual solvent shifts. Gas chromatography (GC) analyses were
performed on an HP 6890 Series GC equipped with an HP 19091J-
413 phenylmethylsilicone column (0.32 mm × 30 m). All in situ IR
spectroscopic reactions were carried out using a 100-mL, stainless steel,
high-pressure Parr reactor modified for use with a Mettler-Toledo
Conclusions
Compound 1 is known to efficiently carbonylate epoxides to
lactones.23 In this study, we have presented kinetic and
spectroscopic data demonstrating that 1 reacts rapidly with
epoxide and CO to form the (â-metalloxy)acylcobalt species
3b. Addition of an aryl isocyanate to the carbonylation system
traps 3b, forming a 1,3-oxazine-2,4-dione derivative, which also
supports the assignment of 3b as the catalyst resting state. When
3a (cf. 3b, R ) Et, R′ ) H) is synthesized independently from
(80) Molnar et al. found that Co2(CO)8/AlMe3 also rapidly catalyzes epoxide
carbonylation.30 They propose that this mixture forms [AlMe2‚diglyme]+
in diglyme solution. On the basis of DFT calculations, they attribute its
ability to carbonylate epoxides to lactones to the availability of a stabilizing
interaction between the Co and an H atom of the lactone in the transition
state. Although we cannot rule out the participation of such an interaction
in our system, the experimental evidence in our study suggests that the
participation of solvent allows 1 to effect rapid lactone ring closing.
9
10132 J. AM. CHEM. SOC. VOL. 128, NO. 31, 2006

Epoxide Carbonylation by [Lewis Acid]+[Co(CO)4]-
A R T I C L E S
ReactIR 4000 Reaction Analysis System fitted with a Sentinel DiComp
High-Pressure Probe. Data were acquired and analyzed using ReactIR
software version 2.21.
mg, 8.11 µmol) was dissolved in THF-d8 (0.70 mL, 8.6 mmol), and
0.60 mL of the solution was transferred into a predried NMR tube
modified with a Kontes valve. The NMR tube was then charged with
4-F-C6H4NCO (46 µL, 0.40 mmol) via syringe, and the sample was
sealed. The sample was immediately transported to the INOVA 400
MHz NMR spectrometer (equilibrated at 25 °C), and spectra were
recorded every 2 min. The integral, I, of the imine resonance of the
starting material (δ ) 9.22 ppm) was used to monitor the progress of
the reaction. The initial rate was evaluated over the period t0 f t such
that It ) 0.95I0.
1,2-Dimethoxyethane (DME), diglyme (DGM), 2-methyltetrahydro-
furan (MTHF), and 2,5-dimethyltetrahydrofuran (DMTHF) were re-
fluxed with Na/benzophenone for a minimum of 16 h and then distilled
under nitrogen and degassed via freeze-pump-thaw cycles. Tetrahy-
drofuran was dried over columns of alumina81 and degassed via freeze-
pump-thaw cycles. 1,2-Difluorobenzene was stirred over P2O5 for 8
days and then vacuum transferred and degassed via freeze-pump-
thaw cycles. Benzene-d6 and tetrahydrofuran-d8 were dried over
Na/benzophenone, vacuum transferred, and sparged with dry nitrogen.
1,2-Epoxybutane was purchased from Aldrich, stirred over CaH2, then
distilled under nitrogen and degassed. Carbon monoxide (Research
Grade) was purchased from Matheson and used as received. Labeled
13CO was purchased from Cambridge Isotope Laboratories, Inc., with
the aid of a Research Grant, and was also used as received. The
complexes [(salph)Al(THF)2]+[Co(CO)4]- (1),23 [PPh4]+[Co(CO)4]-,82
and Na+[Co(CO)4]-83 were prepared as previously reported. [(salph)-
Al(THF)2]+[BPh4]- was synthesized according to the published pro-
cedure for [(R,R)-salcy(tBu)Al(THF)2]+[BPh4]- (salcy ) N,N′-cyclo-
hexylenebis(3,5-di-tert-butylsalicylideneimine),52,84 and [rac-(salph)-
AlOCH(Et)CH2Cl] (4) was formed using the procedure for the related
[((R,R)-salen(tBu))AlOCH(S-Me)CH2Cl].61 â-Valerolactone was pre-
pared from 1,2-epoxybutane and CO using catalyst 1, washed through
Celite with diethyl ether, and distilled. It was then stirred for 1 week
with 4 Å molecular sieves, vacuum transferred onto CaH2, stirred for
1 week, and vacuum transferred. 4-Methylphenyl isocyanate and
4-fluorophenyl isocyanate were purchased from Aldrich, distilled from
P2O5, and degassed.
Representative Epoxide Carbonylation in the Presence of Iso-
cyanate. A Parr reactor, with or without ReactIR modification, was
dried and prepared as described above, with the addition of a third
syringe containing 4-fluorophenyl isocyanate. The catalyst solution (1,
11.5 mM in DME, 10 mL) was injected into the reactor first, followed
by 4-fluorophenyl isocyanate (0.57 mL, 5.0 mmol) and finally the
epoxide (0.50 mL, 5.8 mmol). The reactor was immediately pressured
with CO (300 psi). After 5h, the reactor was vented, and the reaction
1
mixture was analyzed by H NMR spectroscopy.
1,3-Oxazine-2,4-diones. Upon completion of the reactions run with
isocyanates, the reactor was vented, and the reaction mixture was
transferred to a flask. The solvent was removed in vacuo, and the
resulting mixture was washed with hexanes to remove â-VL. The
1
remaining solid was purified as detailed below and analyzed by H
and 13C NMR spectroscopy.
6-Ethyl-3-(4-methylphenyl)-1,3-oxazine-2,4-dione was purified by
crystallization from CH2Cl2/hexanes. The melting point, elemental
analysis, and partial IR spectrum of this compound have been reported.85
1H NMR (CDCl3, 300 MHz) δ 1.11 (3H, t, J ) 7.5 Hz), 1.87 (2H, m),
2.39 (3H, s), 2.82 (1H, dd, 2J ) 17.1 Hz, 3J ) 11.1 Hz), 2.92 (1H, dd,
3
3
2J ) 17.1 Hz, J ) 3.9 Hz), 4.59 (1H, dtd, J ) 11.1, 7.5, 3.9 Hz),
7.07 (2H, pseudo-doublet), 7.26 (2H, pseudo-doublet). 13C NMR
(CDCl3) δ 9.23, 21.48, 27.57, 36.74, 75.81, 128.09, 128.30, 130.22,
130.37, 131.28, 139.52 149.13, 168.55.
Representative Procedure for In Situ Infrared Experiments. The
assembled Parr reactor was dried at 90 °C under vacuum overnight
and allowed to cool. It was then sealed under vacuum and brought
into the glovebox. A solution of 1 (0.107 g, 0.122 mmol) in DME
(9.09 g, 101 mmol) was prepared in a scintillation vial, and 10.0 mL
were drawn into a glass syringe. The needle was inserted through a
septum covering the injection port of the reactor. 1,2-Epoxybutane (1.09
mL, 11.6 mmol) was drawn into another syringe, which was also
inserted through the septum. The reactor was removed from the
glovebox and connected to the React-IR. A heating jacket was added,
and the heater was set to 25 °C. After ∼5 min, a background spectrum
(16 scans) was recorded. Following this, IR spectra (16 scans/spectrum
at a gain of 2 and a resolution of 4 cm-1) were recorded every minute.
After the first spectrum was recorded, the catalyst solution was injected
into the reactor, and the solution was stirred until a temperature of 25
°C was reached. The epoxide was then added to the solution, and the
system was immediately pressurized with CO to 300 psi. The formation
of the product was recorded via the emergence of the carbonyl stretch
of lactone. The reaction was allowed to proceed until no more lactone
was being formed (i.e., the lactone absorbance was constant). The CO
was then vented, and the crude reaction mixture was transferred to a
scintillation vial. Product analysis was confirmed by GC and, in some
6-Ethyl-3-(4-fluorophenyl)-1,3-oxazine-2,4-dione was purified by
1
column chromatography (2:3 EtOAc:hexanes, Rf ) 0.30). H NMR
3
(CDCl3, 300 MHz) δ 1.11 (3 H, t, J ) 7.5 Hz), 1.86 (2 H, m), 2.82
2
3
2
(1 H, dd, J ) 17.1 Hz, J ) 11.1 Hz), 2.92 (1 H, dd, J ) 17.1 Hz,
3J ) 3.6 Hz), 4.59 (1 H, m), 7.15 (2 H, m), 7.168 (2 H, br s). 13C
NMR δ 9.20, 27.54, 36.68, 75.87, 116.49, 116.80, 130.20, 130.32,
151.55, 161.08, 164.37, 168.38. HRMS (EI) m/z calcd (C12H12NO3F)
237.08012, found 237.07632.
Acknowledgment. We thank the National Science Founda-
tion (CHE-0243605) and Department of Energy (DE-FG02-
05ER15687) for funding this work. We are grateful to Cam-
bridge Isotope Laboratories, Inc. for a research grant, which
aided in the purchase of 13CO. T.L.C. thanks the Natural
Sciences and Engineering Research Council (NSERC) Canada
for a postgraduate fellowship. We are indebted to Prof. B.
Carpenter and Ms. A. Litovitz for useful discussions and to Dr.
I. Keresztes for performing the HRMS analysis of 6-ethyl-3-
(4-fluorophenyl)-1,3-oxazine-2,4-dione.
1
cases, by H NMR spectroscopy.
Procedure for In Situ Infrared Observation of Reactions in THF/
DMTHF. When carried out in THF/DMTHF mixtures, the carbony-
lation reaction was performed as described above; however, IR spectra
(8 scans/spectrum, at a gain of 2 and resolution of 8 cm-1) were
collected every 8.3 s, to collect more data during the early stages of
the reaction.
Supporting Information Available: Calibration curve relating
Absν(CdO) to the concentration of â-VL in DME solution, table
of values of reaction rates at different values of PCO, initial
kinetics of reaction between [rac-(salph)AlOCH(Et)CH2Cl] (4)
and 4-F-C6H4NCO, and prediction of [â-VL]/[OD] ratio in a
carbonylation run in THF with 4-F-C6H4NCO present. This
material is available free of charge via the Internet at
http://pubs.acs.org.
Reaction of [rac-(salph)AlOCH(Et)CH2Cl] (4) with 4-Fluoro-
phenyl Isocyanate. In a glovebox, [rac-(salph)AlOCH(Et)CH2Cl] (5.46
(81) Glass Contour Systems: Laguna Beach, CA.
(82) Wei, C.-H.; Bockman, T. M.; Kochi, J. K. J. Organomet. Chem. 1992,
428, 85-97.
(83) Edgell, W. F.; Lyford, J. Inorg. Chem. 1970, 9, 1923-1933.
(84) Sahasrabudhe, S.; Yearwood, B.; Atwood, D. A. Inorg. Synth. 2004, 34,
14-20.
JA061503T
(85) Saitkulova, F. G.; Semenov, V. I.; Lapkin, I. I. Khim. Elementoorg. Soedin.,
Gork’ii 1982, 22-26.
9
J. AM. CHEM. SOC. VOL. 128, NO. 31, 2006 10133