Palladium-catalyzed [3+2] cycloaddition of carbon dioxide and trimethylenemethane under mild conditions

Article abstract of DOI:10.1021/ol7017246

Carbon dioxide undergoes a Pd-catalyzed [3+2] cycloaddition with trimethylenemethane (TMM) under mild conditions (1 atm, 75 °C, 30 min) to produce a γ-butyroiactone product in 63% yield, when the Pd-TMM complex is generated from 2-(acetoxymethyl)-3-(trimethylsilyl)propene. The reaction reported here is more rapid than the all-carbon [3+2] cycloaddition, and only the γ-butyrolactone is produced in a competition experiment. With substituted substrates, the reaction is completely regioselective, producing the product derived from the kinetic Pd-TMM complex.

Full text of DOI:10.1021/ol7017246

ORGANIC
LETTERS
2
007
Vol. 9, No. 19
817-3820
Palladium-Catalyzed [3
of Carbon Dioxide and
+2] Cycloaddition
3
Trimethylenemethane under Mild
Conditions
George E. Greco,* Brittany L. Gleason, Tiffany A. Lowery, Matthew J. Kier,
Lisa B. Hollander, Shoshanah A. Gibbs, and Amanda D. Worthy
Department of Chemistry, Goucher College, 1021 Dulaney Valley Road,
Baltimore, Maryland 21204
ggreco@goucher.edu
Received July 19, 2007
ABSTRACT
Carbon dioxide undergoes a Pd-catalyzed [3
to produce a -butyrolactone product in 63% yield, when the Pd
The reaction reported here is more rapid than the all-carbon [3
+
2] cycloaddition with trimethylenemethane (TMM) under mild conditions (1 atm, 75
TMM complex is generated from 2-(acetoxymethyl)-3-(trimethylsilyl)propene.
2] cycloaddition, and only the -butyrolactone is produced in a competition
°C, 30 min)
γ
−
+
γ
experiment. With substituted substrates, the reaction is completely regioselective, producing the product derived from the kinetic Pd
complex.
−TMM
The development of new methods for the utilization of
carbon dioxide is an area of considerable current interest
because carbon dioxide is both a greenhouse gas and a cheap
and abundant source of carbon atoms.1,2 Transition metals
the nickel-catalyzed carboxylation of 1,3-dienes with CO
2
10-12
and dialkylzinc reagents,
with CO to give R-pyrones.
and the cycloaddition of diynes
1
3,14
2
The drawback of many of
these methods is the requirement for high pressures of carbon
dioxide, thus limiting their synthetic utility. In addition, most
research into cycloaddition reactions has focused on the
have been shown to be effective at activating the relatively
,4
unreactive CO
2
molecule,3 and reactions in which a metal
complex catalyzes a transformation of carbon dioxide are
2
incorporation of CO into even sized rings since 2 carbon
5
especially appealing. Some recently reported examples
and 4 carbon components are readily available. Herein we
report a Pd-catalyzed [3+2] cycloaddition reaction between
carbon dioxide and a series of 2-trimethylsilylmethyl-2-
propenyl acetates to produce substituted γ-butyrolactones that
takes place efficiently under 1 atm of carbon dioxide (eq 1).
include the reaction of CO
2
with epoxides6 or aziridines,
-8
9
(1) Arakawa, H.; Aresta, M.; Armor, J. N.; Barteau, M. A.; Beckman,
E. J.; Bell, A. T.; Bercaw, J. E.; Creutz, C.; Dinjus, E.; Dixon, D. A.; Domen,
K.; Du Bois, D. L.; Eckert, J.; Fujita, E.; Gibson, D. H.; Goddard, W. A.;
Goodman, D. W.; Keller, J.; Kubas, G. J.; Kung, H. H.; Lyons, J. E.; Manzer,
L. E.; Marks, T. J.; Morokuma, K.; Nicholas, K. M.; Periana, R.; Que, L.;
Rostrup-Nielson, J.; Sachtler, W. M. H.; Schmidt, L. D.; Sen, A.; Somorjai,
G. A.; Stair, P. C.; Stults, B. R.; Tumas, W. Chem. ReV. 2001, 101, 953-
(8) Paddock, R. L.; Nguyen, S. T. J. Am. Chem. Soc. 2001, 123, 11498-
11499.
(9) Miller, A. W.; Nguyen, S. T. Org. Lett. 2004, 6, 2301-2304.
(10) Takimoto, M.; Mori, M. J. Am. Chem. Soc. 2001, 123, 2895-2896.
(11) Takimoto, M.; Mori, M. J. Am. Chem. Soc. 2002, 124, 10008-
10009.
9
96.
(2) Sakakura, T.; Choi, J. C.; Yasuda, H. Chem. ReV. 2007, 107, 2365-
2
387.
(
3) Yin, X.; Moss, J. R. Coord. Chem. ReV. 1999, 181, 27-59.
(12) Takimoto, M.; Nakamura, Y.; Kimura, K.; Mori, M. J. Am. Chem.
Soc. 2004, 126, 5956-5957.
(4) Aresta, M.; Quaranta, E.; Tommasi, I.; Giannoccaro, P.; Ciccarese,
A. Gazz. Chim. Ital. 1995, 125, 509-538.
(13) Tekavek, T. N.; Arrif, A. M.; Louie, J. Tetrahedron 2004, 60, 7431-
(
(
(
5) Louie, J. Curr. Org. Chem. 2005, 9, 605-623.
7437.
6) Coates, G. W.; Moore, D. R. Angew. Chem., Int. Ed. 2004, 43.
7) Darensbourg, D. J. Chem. ReV. 2007, 107, 2388-2410.
(14) Louie, J.; Gibby, J. E.; Farnworth, M. V.; Tekavek, T. N. J. Am.
Chem. Soc. 2002, 124, 15188-15189.
1
0.1021/ol7017246 CCC: $37.00
© 2007 American Chemical Society
Published on Web 08/23/2007

Heating a THF solution of 2a under reflux overnight with
mol % of (PPh Pd under 1 atm of carbon dioxide results
in the smooth formation of γ-butyrolactone 1a in which the
carbon and one oxygen atom from CO have been incorpo-
8
3 4
)
2
rated into the 5-membered ring. The initial cycloadduct is
almost certainly compound 3a, but it is never detected; the
CdC double bond presumably undergoes a facile isomer-
ization to an endocyclic position in conjugation with the
carbonyl group under the reaction conditions. A possible
mechanism for this reaction is depicted in Scheme 1.
On the basis of NMR and GC/MS analysis of the crude
product mixture, other products that could be identified were
“higher order” cycloadducts in which more than one
The Pd-catalyzed [3+2] cycloaddition of methylenecy-
clopropane (MCP) with carbon dioxide to form γ-butyro-
lactones was initially reported by Inoue in 1979 and further
studied by Binger.16 While yields of compound 1a as high
as 80% can be obtained under the proper conditions, the
15
2
reaction only succeeds under CO pressures of 40 atm and
temperatures greater than 130 °C, thus severely limiting its
utility. It is also plagued by the fact that Pd catalyzes the
reaction of product 1a with more MCP, leading to higher
2
molecule of 1a was co-oligomerized with CO (Figure 1),
16
oligomers. It is believed that the mechanism of this reaction
involves oxidative addition of Pd(0) into the cyclopropane
3
ring to produce an η -Pd trimethylenemethane (TMM)
complex, which is an excellent all-carbon 1,3-dipole (see
Scheme 1 for a drawing of the Pd-TMM complex). The
Scheme 1. Possible Catalytic Cycle
2
Figure 1. Higher order cycloadducts of TMM and CO .
oligomers of compound 2a that do not contain CO
methallyl acetate (4) obtained through protodesilylation of
a.
These byproducts were not unexpected: the “higher order”
2
, and
2
cycloadducts were observed by Binger when methylenecy-
clopropane was used as the Pd-TMM precursor, and
compound 4 was present in many of Trost’s reactions.
Having demonstrated the feasibility of the [3+2] cycload-
dition with carbon dioxide as the 2-atom component, we
sought to optimize the reaction. In our initial experiments,
which employed typical conditions for [3+2] cycloaddition
reactions of 2a, the isolated yield of 1a was 25%. Variables
that we investigated included the concentration, temperature,
reaction time, stirring rate, solvent, and ligands around
palladium. We found that concentration was the factor most
responsible for higher order cycloadducts, and their formation
was suppressed at lower concentrations, until they were
barely detectable at a substrate concentration of 20 mM.
Since no further improvement was observed with additional
dilution, a concentration of 20 mM was employed in all
subsequent studies.
2
electron-rich end can attack the carbonyl carbon of CO ,
while the electron-poor end can accept electron density from
the oxygen. Pd-TMM complexes have been implicated as
1
7
intermediates in a wide variety of catalytic reactions, but
they are too unstable to isolate and characterize. As a result,
they must always be generated from an appropriate precursor
as an initial step in a catalytic cycle.
22
Another precursor to Pd-TMM complexes is com-
mercially available and readily prepared 2-(acetoxymethyl)-
3-(trimethylsilyl)propene (2a). This molecule has been
extensively used by Trost as a 3-atom partner in [3+2]
cycloadditions with both CdC and CdO double bonds as
the two-atom component.1
7-21
Compound 2a participates in
many of the same reactions as MCP, but the conditions are
generally milder, and product selectivity can be realized if
substituted versions of 2 are employed.
The suppression of non-CO
maximum exposure of the reaction solution to CO
the reaction was stirred rapidly under a CO atmosphere for
min at room temperature, then heated to 60 °C for 1.5 h,
a and 4 were the only products detected by GC. The isolated
2
containing products required
2
. When
2
(15) Inoue, Y.; Hibi, T.; Satake, M.; Hashimoto, H. J. Chem. Soc., Chem.
5
1
Commun. 1979, 982.
(
(
(
16) Binger, P.; Weintz, H. J. Chem. Ber. 1984, 117, 654-665.
17) Trost, B. M. Angew. Chem., Int. Ed. 1986, 25, 1-20.
yield of 1a under these conditions was 62% (Table 1, entry
). Using NMR spectroscopy, we determined the yield for
18) Trost, B. M.; Chan, D. M. T. J. Am. Chem. Soc. 1979, 101, 6429-
1
6
6
1
432.
(19) Trost, B. M.; Chan, D. M. T. J. Am. Chem. Soc. 1979, 101, 6432-
this reaction to be 69% by integrating the NMR signal for
the methyl groups of durene (1,2,4,5-tetramethylbenzene)
433.
(20) Trost, B. M.; Mignani, S. M.; Nanninga, T. N. J. Am. Chem. Soc.
986, 108, 6051-6053.
(21) Trost, B. M.; Stambuli, J. P.; Silverman, S. M.; Schworer, U. J.
(22) Trost, B. M.; Chan, D. M. T. J. Am. Chem. Soc. 1983, 105, 2315-
2325.
Am. Chem. Soc. 2006, 128, 13328-13329.
3818
Org. Lett., Vol. 9, No. 19, 2007

equiv of P(OiPr)
of 2a to 1a. No reaction was observed at all when the larger
ligands PCy and P(o-tol) were added to Pd (entries 13 and
4), and the bidentate phosphines dppe and BINAP were
equally unsuccessful (entries 15 and 16). Presumably, ligands
that are too large or coordinate to the metal too tightly are
3
relative to Pd, there is minimal conversion
Table 1. Optimization of the Pd-Catalyzed Cycloaddition of 2a
and Co
2
To Produce 1a
3
3
1
entry
catalyst
conditions
yield (%)
b
c
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
(Ph3P)4Pd
(Ph3P)4Pd
(Ph3P)4Pd
(Ph3P)4Pd
(Ph3P)4Pd
(Ph3P)4Pd
THF, 1.5 h, 60 °C 69, 62
a
THF, 3 h, 50 °C
45
2
unsuccessful because they do not permit both CO and the
a
b
c
DME, 0.5 h, 75 °C 72, 69, 55
substrate to approach the coordination sphere. Given that this
ligand screen sampled phosphines with a variety of steric
and electronic properties, and none were nearly as successful
as triphenylphosphine, no other ligands were screened.
c
PhH, 20 h, 60 °C
63
b
a
a
a
PhMe, 24 h, 75 °C 31
DMF, 18 h, 55 °C 19
Pd(OAc)2 + 4 PPh3
Pd(OAc)2 + 3 PPh3
Pd(OAc)2 + 2 PPh3
Pd(OAc)2 + 3 P(OiPr)3 THF, 18 h, 55 °C
Pd(OAc)2 + 5 P(OiPr)3 THF, 18 h, 50 °C
Pd(OAc)2 + 8 P(OiPr)3 THF, 21 h, 50 °C
Pd(OAc)2 + 4 PCy3
Pd(OAc)2 + 4 P(o-tol)3 THF, 72 h, 55 °C
Pd(OAc)2 + 2 dppe
Pd(OAc)2 + 2 BINAP
THF,18 h, 55 °C
THF, 18 h, 55 °C
THF, 3.5 h, 55 °C
42
46
On the basis of the conditions we screened, the optimal
procedure for substrate 2a consists of using 8 mol % of
a
0
a
1
1
1
1
1
1
1
0
Pd(PPh
0 mM in DME as the solvent.
Several control experiments were carried out to demon-
3 4
) as the catalyst with a substrate concentration of
a
a
minimal
minimal
2
a
THF, 18 h, 55 °C
0
0
a
strate that the use of 2a as a Pd-TMM precursor is required
for the formation of 1a under such mild conditions. Specif-
ically, 2a was recovered unchanged after a THF solution
a
THF, 48 h, 55 °C
THF, 18 h, 55 °C
minimal
a
0
a
GC yield. b NMR yield. c Isolated yield
2
was heated overnight under CO in the absence of palladium.
Interestingly, the formation of byproduct 4 also requires
palladium, since none of it was formed in the absence of
palladium. Furthermore, when diphenylmethylenecyclopro-
pane, a substrate known to undergo the [3+2] reaction with
CO under high temperatures and pressures, was synthe-
2
sized and subjected to the mild reaction conditions, there
was no reaction. In another experiment, a solution of
against the signal for the methyl group of 1a. Presumably,
carrying out the reaction at a temperature below the boiling
1
6
point of the solvent allows a greater partial pressure of CO
above the solution, and an increased concentration of CO
2
2
in solution. However, we also observed that if the temper-
ature is reduced too much, the amount of undesired 4
increases at the expense of 1a. In a reaction run at 50 °C,
the GC yield of 1a determined with mesitylene as an internal
standard and correcting for the response factor of mesitylene
was 45% (entry 2).
substrate 2a was stirred with Pd(PPh
of under CO . Complete consumption of 2a in the absence
of CO requires 2 h instead of 30 min; the products are
3 4
) under nitrogen instead
2
2
various sized oligomers of TMM.
Of particular interest is the fact that the [3+2] cycload-
The need to keep the temperature close to, but below the
boiling point of THF proved inconvenient, so we screened
several higher boiling solvents. Similar yields were obtained
in DME at 75 °C; at this temperature the reaction is complete
in 30 min (entry 3). Benzene or toluene could be used as
solvents (entries 4 and 5), but the rate is much slowers
requiring overnight reaction at 75 °C to go to completion,
and the yield is reduced in toluene. DMF is also an acceptable
solvent, but once again, the GC yield is lower (entry 6).
2
dition between 2a and CO is more rapid than the cycload-
dition between 2a and an alkene, which suggests that CO2
is involved in the rate-limiting step of the mechanism. In a
competition experiment, an equimolar ratio of 2a and
benzylideneacetone (an alkene typically employed by Trost
as a 2-atom component in the [3+2] cycloaddition)22 was
heated to 50 °C under a CO atmosphere in the presence of
2
the palladium catalyst. The reaction is slower than in the
absence of benzylideneacetone: 6 h are required for complete
To vary the phosphine ligands, we generated a catalyst in
situ by mixing palladium(II) acetate with the desired phos-
phine. We validated the in situ catalyst generation by carrying
out experiments with triphenylphosphine and comparing the
consumption of 2a, but most importantly, the ratio of CO
2
cycloadduct 1a to the all carbon cycloadduct (5) is >9:1 by
GC, and most of the benzylideneacetone was not consumed
(eq 2).
results to those obtained with (Ph
of PPh relative to Pd, the yields are slightly lower than with
Ph P) Pd, but the reaction is clearly successful (entries 7
and 8). In contrast, if only 2 equiv of PPh are added, there
3 4
P) Pd. With 3 or 4 equiv
3
(
3
4
3
is no conversion of substrate 2a, and palladium black is
observed in the flask (entry 9).
Trost reported that triisopropyl phosphite is another
successful ligand in Pd-catalyzed all-carbon [3+2] cycload-
dition reactions.23 However, the same is not true for the
reaction with carbon dioxide (entries 10-12). Even with 8
To examine the scope and regioselectivity of the reaction,
we synthesized methylated substrates 2b, 2c, and 2d as
described in the Supporting Information. All three methylated
24
24
(
23) Trost, B. M.; Nanninga, T. N.; Satoh, T. J. Am. Chem. Soc. 1985,
(24) Trost, B. M.; Chan, D. M. T. J. Am. Chem. Soc. 1981, 103, 5972-
5974.
1
07, 721-723.
Org. Lett., Vol. 9, No. 19, 2007
3819

substrates participate in the cycloaddition as shown in Table
. However, the reaction conditions for the cycloaddition
2
Scheme 2. Kinetic Pd-TMM Complexes
Table 2. Substrates for Pd-Catalyzed Cycloaddition
is faster than isomerization. This is in contrast with the all-
carbon reaction where isomerization to the thermodynamic
Pd-TMM complex (ii) with the methyl group on the anionic
2
4
carbon precedes cycloaddition, but consistent with a
a
NMR yield. b Isolated yield.
recently reported [3+3] cycloaddition involving compounds
2
5
2
b and 2c. The formation of product 1b from either 2b or
2
d suggests that complex i is formed from each of these
substrates, and unlike in the case of the all-carbon [3+2],
this intermediate is competent to undergo the cycloaddition
with CO , albeit more slowly than complex ii does.
2
In summary, we have developed an efficient Pd-catalyzed
were not individually optimized for each substituted sub-
strate.
Unlike in the case of the all-carbon [3+2] cycloaddition,24
the product depends on the structure of the starting material.
Specifically, the product obtained from reaction of 2b or 2d
contains the methyl group at the 5-position of the furanone
ring (1b), whereas reaction of substrate 2c gives a product
with the methyl group at the 3-position of the ring (1c).
The rate of the reaction also depends on the structure of
the starting material. Reactions starting with 2c are faster
than those starting with 2a, while reactions starting with 2b
or 2d are slower (but still not as slow as the all-carbon
[3+2] cycloaddition of 2-(trimethylsilylmethyl)allyl acetates
with carbon dioxide to produce γ-butyrolactones under mild
conditions (1 atm of CO , 75 °C, 30 min). Further investiga-
2
tions of the scope and mechanism of this reaction are in
progress.
Acknowledgment. We thank the donors of the American
Chemical Society Petroleum Research Fund and Goucher
College for financial support of this research.
[3+2]). On the basis of the product regiochemistry and the
relative rate observations, we believe that in all cases the
product is formed from attack of the kinetic Pd-TMM
complex (i or ii) on CO (Scheme 2), and that cycloaddition
2
Supporting Information Available: Discussion of sub-
strate synthesis, and full experimental details. This material
is available free of charge via the Internet at http://pubs.acs.org.
(25) Shintani, R.; Hayashi, T. J. Am. Chem. Soc. 2006, 128, 6330-6331.
OL7017246
3820
Org. Lett., Vol. 9, No. 19, 2007