Mechanistic twist of the [8+2] cycloadditions of dienylisobenzofurans and dimethyl acetylenedicarboxylate: Stepwise [8+2] versus [4+2]/[1,5]-vinyl shift mechanisms revealed through a theoretical and experimental study

Article abstract of DOI:10.1021/ja072203u

Recently, it was reported that both dienylfurans and dienylisobenzofurans could react with dimethyl acetylenedicarboxylate (DMAD) to give [8+2] cycloadducts. Understanding these [8+2] reactions will aid the design of additional [8+2] reactions, which have

Full text of DOI:10.1021/ja072203u

Published on Web 08/14/2007
Mechanistic Twist of the [8+2] Cycloadditions of
Dienylisobenzofurans and Dimethyl Acetylenedicarboxylate:
Stepwise [8+2] versus [4+2]/[1,5]-Vinyl Shift Mechanisms
Revealed through a Theoretical and Experimental Study
Yuanyuan Chen,† Siyu Ye,† Lei Jiao,† Yong Liang,† Dilip K. Sinha-Mahapatra,‡
James W. Herndon,‡ and Zhi-Xiang Yu*,†
Contribution from the Beijing National Laboratory for Molecular Sciences (BNLMS), Key
Laboratory of Bioorganic Chemistry and Molecular Engineering of the Ministry of Education,
College of Chemistry, Peking UniVersity, Beijing 100871, People’s Republic of China, and
Department of Chemistry and Biochemistry, New Mexico State UniVersity,
MSC 3C, Las Cruces, New Mexico 88003
Received March 29, 2007; E-mail: yuzx@pku.edu.cn
Abstract: Recently, it was reported that both dienylfurans and dienylisobenzofurans could react with dimethyl
acetylenedicarboxylate (DMAD) to give [8+2] cycloadducts. Understanding these [8+2] reactions will aid
the design of additional [8+2] reactions, which have the potential for the synthesis of 10-membered and
larger carbocycles. The present Article is aimed to understand the detailed mechanisms of the originally
reported [8+2] cycloaddition reaction between dienylisobenzofurans and alkynes at the molecular level
through the joint forces of computation and experiment. Density functional theory calculations at the (U)-
B3LYP/6-31+G(d) level suggest that the concerted [8+2] pathway between dienylisobenzofurans and
alkynes is not favored. A stepwise reaction pathway involving formation of a zwitterionic intermediate for
the [8+2] reactions between dienylisobenzofurans that contain electron-donating methoxy groups present
in their diene moieties and DMAD has been predicted computationally. This pathway is in competition with
a Diels-Alder [4+2] reaction between the furan moieties of dienylisobenzofurans and DMAD. When there
is no electron-donating group present in the diene moieties of dienylisobenzofurans, the [8+2] reaction
occurs through an alternative mechanism involving a [4+2] reaction between the furan moiety of the tetraene
and DMAD, followed by a [1,5]-vinyl shift. This computationally predicted novel mechanism was supported
experimentally.
1. Introduction
2,4-diene and tetraenophiles such as tetracyanoethylene and
dimethyl azodicarboxylate4h (Scheme 1, reaction c). Develop-
ment of new [8+2] cycloadditions between geometrically
flexible tetraenes and tetraenophiles can result in new and highly
convergent synthetic approaches to 10-membered ring com-
pounds.5,6
In 2003, it was reported that dienylisobenzofurans can react
with dimethyl acetylenedicarboxylate (DMAD) to furnish [8+2]
adducts possessing the 11-oxabicyclo[6.2.1]undecane ring sys-
tem as the major products (Scheme 2, reaction a).7a It was later
noted that dienylfurans could also participate in the [8+2]
Orbital-symmetry allowed [8+2] cycloadditions between
tetraenes and tetraenophiles can in theory provide a straight-
forward approach for the synthesis of 10-membered ring
compounds.1-4 Before the year 2003, however, all of the
reported [8+2] cycloadditions were limited to geometrically
constrained tetraenes such as heptafulvenes, tropones, and
indolizines, in which the terminal carbons or heteroatoms at
positions 1 and 8 are rigidly held in close proximity (Scheme
1, reactions a and b). Consequently, only bicyclic or tricyclic
compounds instead of 10-membered ring compounds were
obtained in these [8+2] cycloadditions. The only reported [8+2]
cycloaddition employing geometrically flexible tetraenes in the
last century is the reaction between 1,6-dimethylene cyclohepta-
(4) For other selected [8+2] cycloadditions, see: (a) Liu, C. Y.; Mareda, J.;
Houk, K. N.; Fronczek, F. R. J. Am. Chem. Soc. 1983, 105, 6714. (b) Liu,
C. Y.; Ding, S. T. J. Org. Chem. 1992, 57, 4539. (c) Liu, C. Y.; Houk, K.
N. Tetrahedron Lett. 1987, 28, 1371. (d) Daub, J.; Hirmer, G.; Kakob, L.;
Maas, G.; Pickl, W.; Pirzer, E.; Rapp, K. M. Chem. Ber. 1985, 118, 1836.
(e) Bau¨mler, A.; Daub, J.; Pickl, W.; Rieger, W. Chem. Ber. 1985, 118,
1857. (f) Hayakawa, K.; Nishiyama, H.; Kanematsu, K. J. Org. Chem. 1985,
50, 512. (g) Kumar, K.; Kapur, A.; Ishar, M. P. S. Org. Lett. 2000, 2, 787.
(h) Farrant, G. C.; Feldmann, R. Tetrahedron Lett. 1970, 11, 4979. (i)
Komatsu, K.; Fujimori, M.; Okamoto, K. Tetrahedron 1977, 33, 2791. (j)
Pham, W.; Weissleder, R.; Tung, C. H. Tetrahetron Lett. 2002, 43, 19. (k)
Nair, V.; Anikumar, G.; Nandakumar, M. V.; Mathew, B.; Rath, N. P.
Tetrahedron Lett. 1997, 38, 6441. (l) Babaev, E. V.; Simonyan, V. V.;
Pasichnichenko, K. Y.; Nosova, V. M.; Kisin, A. V.; Jug, K. J. Org. Chem.
1999, 64, 9057.
† Peking University.
‡ New Mexico State University.
(1) The first reported [8+2] cycloaddition: Doering, W. v. E.; Wiley, D. W.
Tetrahedron 1960, 11, 183.
(2) (a) Woodward, R. B.; Hoffmann, R. The ConserVation of Orbital Symmetry;
Verlag Chemie/Academic Press: Weinheim, 1971. (b) Woodward, R. B.;
Hoffmann, R. Angew. Chem., Int. Ed. Engl. 1969, 8, 781.
(3) For a review of [8+2] cycloaddition, see: Nair, V.; Anilkumar, G. Synlett
1998, 950.
9
10.1021/ja072203u CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 10773-10784
10773

A R T I C L E S
Chen et al.
Scheme 1. [8+2] Reactions of Tetraenes with Tetraenophiles
mechanisms of Diels-Alder reactions, which occur via either
a concerted or a stepwise pathway.8 In addition to pathways A
and B, we propose a novel pathway C for the [8+2] cycload-
dition (Scheme 3). Pathway C begins with a [4+2] cycloaddition
of DMAD and the furan moiety of the tetraene to give a [4+2]
cycloadduct, which then isomerizes via a [1,5]-vinyl shift to
furnish the final [8+2] cycloadduct.9,10 If [8+2] cycloadditions
occur via pathway C and this could be verified experimentally,
it is likely that any bicyclo[2.2.1]heptene system, easily obtained
through Diels-Alder chemistry, could be transformed to a 10-
membered ring system after installation of the exocyclic diene
group. We envisioned that such a ring-enlargement strategy
could also be applied to the construction of other large ring
compounds.
With the above-mentioned mechanistic objectives in mind,10
especially to test whether pathway C is feasible or not, a
theoretical and experimental study of the [8+2] cycloadditions
of dienylisobenzofurans and DMAD has been performed. These
results show that pathway A is not favored intrinsically (see
discussions below). Pathways B and C are possible, and their
preference varies, depending on the substituents present in the
tetraene substrates. When there is no electron-donating group
(such as methoxy group) in the dienyl moiety of dienylisoben-
zofurans, the favored pathway for the [8+2] reaction is pathway
C via the [4+2]/[1,5]-vinyl shift mechanism. When an electron-
donating group such as the methoxy group is present in the
dienyl moiety of dienylisobenzofurans, pathway C is still favored
in gas phase but pathway B is very competitive in solution
because the methoxy group can stabilize the zwitterionic
transition states and intermediates involved in pathway B more
significantly than those stationary points in pathway C. In this
Article, detailed theoretical and experimental studies that explore
the whole scenario of mechanistic twists of [8+2] cycloadditions
between dienylisobenzofurans and DMAD are presented.
Scheme 2. [8+2] Reactions Reported by Herndon and
Co-workers
cycloadditions with DMAD (Scheme 2, reaction b).7b These
[8+2] cycloadditions provide a direct approach for the synthesis
of ring skeletons of eleutherobin, briarellins, and other natural
products that have anticancer activity.5,6 As compared to [8+2]
cycloadditions using geometrically constrained tetraenes, the
[8+2] cycloaddition reactions highlighted in Scheme 2 employ
flexible tetraenes in which the terminal C1 and C8 are not held
in close proximity. More importantly, 10-membered ring
compounds with an oxygen bridge can be readily synthesized
through these [8+2] cycloadditions. Understanding the mech-
anisms of these [8+2] cycloadditions will not only enhance
knowledge of [8+2] cycloaddition reactions and the chemistry
of pericyclic reactions, but also provide insights and guides for
the future design of new [8+2] and other higher order [m+n]
cycloadditions that have the potential application in the synthesis
of 10-membered or larger ring compounds.
2. Computational Methodologies
All of the calculations were performed with the Gaussian 03
program.11 The hybrid B3LYP functional12 in conjunction with the
6-31+G(d) basis set13 was applied for the optimization of all of the
stationary points in gas phase.14-17 A basis set that includes diffuse
functions for heavy atoms better describes the zwitterionic species in
these reactions. Singlet diradical transition states and intermediates were
(8) For studies of the mechanisms of Diels-Alder reactions, see: (a) Houk,
K. N.; Gonzalez, J.; Li, Y. Acc. Chem. Res. 1995, 28, 81. (b) Houk, K. N.;
Li, Y.; Evanseck, J. D. Angew. Chem., Int. Ed. Engl. 1992, 31, 682. (c)
Goldstein, E.; Beno, B.; Houk, K. N. J. Am. Chem. Soc. 1996, 118, 6036.
(d) Barriault, L.; Thomas, J. D. O.; Clement, R. J. Org. Chem. 2003, 68,
2317. (e) Rodriguez, D.; Navarro-Vazquez, A.; Castedo, L.; Dominguez,
D.; Saa, C. J. Am. Chem. Soc. 2001, 123, 9178. (f) Sakai, S. J. Phys. Chem.
A 2000, 104, 922. (g) Valley, N. A.; Wiest, O. J. Org. Chem. 2007, 72,
559. (h) Kong, S.; Evanseck, J. D. J. Am. Chem. Soc. 2000, 122, 10418.
(9) For theoretical studies of [1,5]-vinyl shift, see: (a) Kla¨ner, F.-G.; Ehrhardt,
R.; Bandmann, H.; Boese, R.; Bla¨ser, D.; Houk, K. N.; Beno, B. R. Chem.-
Eur. J. 1999, 5, 2119. (b) Alder, R. W.; Grimme, W. Tetrahedron 1981,
37, 1809. For reactions involving [1,5]-vinyl shift, see: (c) Semmelhack,
M. F.; Weller, H. N.; Foos, J. S. J. Am. Chem. Soc. 1977, 99, 292. (d)
Semmelhack, M. F.; Weller, H. N.; Clardy, J. J. Org. Chem. 1978, 43,
3791. (e) Fra´ter, G.; Mu¨ller, U. HelV. Chim. Acta 1988, 71, 808.
(10) There is another possible pathway D, which starts from [4+2] cycloaddition
of alkyne to the dienyl moiety of isobenzofuran, followed by [1,5]-vinyl
shift. Such pathway was proved to be very difficult because the computed
[1,5]-vinyl shift step requires an activation energy more than 73 kcal/mol.
See the Supporting Information for details.
Likely mechanisms for the formation of [8+2] cycloadducts
between dienylisobenzofuran and DMAD are depicted in
Scheme 3. These pathways include a concerted [8+2] cycload-
dition (pathway A) or a stepwise pathway B involving the
formation of a zwitterionic (or diradical) intermediate. These
two pathways can be easily proposed if one considers the
(5) For a review of 10-membered ring natural compounds, see: (a) Bernardelli,
P.; Paquette, L. Heterocycles 1998, 49, 531. The most important compound
in this class is the potential anticancer agent eleutherobin, see: (b) Lindel,
T.; Jensen, P. R.; Fenical, W.; Long, B. H.; Casazza, A. M.; Carboni, J.;
Fairchild, C. R. J. Am. Chem. Soc. 1997, 119, 8744. (c) Lindel, T. Angew.
Chem., Int. Ed. 1998, 37, 774.
(6) The synthesis of 10-membered ring compounds could also be achieved
via [6+4] cycloadditions, see: (a) Houk, K. N.; Woodward, R. B. J. Am.
Chem. Soc. 1970, 92, 4143. (b) Houk, K. N.; Woodward, R. B. J. Am.
Chem. Soc. 1970, 92, 4145. (c) Bhacca, N. S.; Luskus, L. J.; Houk, K. N.
Chem. Commun. 1971, 109. (d) For a recent theoretical study of [6+4]
cycloaddition, see: Leach, A. G.; Goldstein, E.; Houk, K. N. J. Am. Chem.
Soc. 2003, 125, 8330.
(11) Frisch, M. J.; et al. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford,
CT, 2004.
(12) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (b) Lee, C.; Yang, W.;
Parr, R. G. Phys. ReV. B 1988, 37, 785.
(13) Hehre, W. J.; Radom, L.; Schleyer, P. v. R.; Pople, J. A. Ab Initio Molecular
Orbital Theory; Wiley: New York, 1986.
(7) (a) Luo, Y.; Herndon, J. W.; Cervantes-Lee, F. J. Am. Chem. Soc. 2003,
125, 12720. (b) Zhang, L.; Wang, Y.; Buckingham, C.; Herndon, J. W.
Org. Lett. 2005, 7, 1665.
9
10774 J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007

[8+
2] Cycloadditions of Dienylisobenzofurans
A R T I C L E S
Scheme 3. Three Possible Pathways for Dienylisobenzofuran-Alkyne [8+2] Cycloaddition
located with UB3LYP/6-31+G(d). Frequency calculations were per-
formed to confirm that each stationary point is either a minimum or a
transition structure. In cases where transition structures are not easily
confirmed by animation of their negative vibrations, IRC18 calculations
were used to confirm the connection between the reactant, product,
and transition state, which are given in the Supporting Information.
The wavefunctions of several key diradical and zwitterionic stationary
points in model reactions I and II have been computationally tested as
stable ones. The reported relative energies are free energies (∆G),
enthalpies (∆H, given in the Supporting Information), and zero-point
energy (ZPE)-corrected electronic energies (∆E0) in gas phase. Solvent
effects in benzene were computed by the CPCM model19 using the
gas-phase optimized structures (using keyword: RADII)UAHF). The
computed activation free energies in solution, referred to as ∆Gsol, were
calculated by adding the solvation energies to the computed gas-phase
relative free energies. The molecular orbital energies were computed
at the HF/6-31G(d) level based on the B3LYP/6-31+G(d) geometries
in gas phase. The spin density distribution of all of the diradical
stationary points is provided in the Supporting Information. Unless
otherwise specified, all discussed relative energies refer to the gas-
phase calculations.
twists caused by substituents in the dienyl moieties of dienyl-
isobenzofurans, model reactions I-IV shown in Scheme 4 have
been studied computationally. Experiments that test the theoreti-
cal predictions have also been conducted. First, the theoretical
study of the [8+2] reaction between dienylisobenzofuran and
acetylene will be presented to understand the inherent reaction
preferences of [8+2] over [4+2] in this parent system (section
3.1). In section 3.2, a theoretical study of model reaction II will
be presented to show the preference using tetraenophiles of
appropriate reactivities. After presenting this theoretical predic-
tion, we will then provide experimental evidence to support our
prediction. Furthermore, in this part, we have also obtained the
kinetic data to corroborate experimental and computational
activation parameters, showing that B3LYP/6-31+G(d) is very
suitable for the mechanistic investigation of the present [8+2]
cycloadditions.20-22 In section 3.3, we will present theoretical
studies of model reactions III and IV, demonstrating that the
presence of a methoxy group makes the stepwise zwitterionic
pathway B competitive with pathway C. This mechanism in
model reactions III and IV has also been tested and supported
experimentally.
3. Results and Discussion
To fully understand the mechanisms of the [8+2] reactions
between dienylisobenzofurans and DMAD and the mechanistic
3.1. Theoretical Study of Model Reaction I of Dienyl-
isobenzofuran and Acetylene. Figure 1 shows the DFT
computed energy surfaces for pathways B and C of model
(14) Spin contamination is well known for the calculations of singlet diradical
species. We have also computed the spin contamination for all singlet
diradical stationary points using the Yamaguchi-Houk spin projection
method15a (see the Supporting Information for details). It was found that
the computed activation energy without Yamaguchi-Houk correction for
the vinyl shift step of 19 to 21 is close to the experimentally measured
one, whereas the computed activation energy after Yamaguchi-Houk
correction is lower by about 4 kcal/mol than the experimentally measured
one. Due to this, the reported energies in this Article for all singlet diradical
stationary points have not been corrected using the Yamaguchi-Houk spin
projection method.15b
(15) (a) Yamaguchi, K.; Jensen, F.; Dorigo, A.; Houk, K. N. Chem. Phys. Lett.
1988, 149, 537. (b) For comments on the spin projection method, see:
Wittbrodt, J. M.; Schlegel, H. B. J. Chem. Phys. 1996, 105, 6574.
(16) We found that the UMP2 method is not suitable for the investigation of
[1,5]-vinyl shift reaction. This is because the UHF wavefunction for the
vinyl shift transition state is not stable; consequently, the UMP2 energy
based on this wavefunction is questionable.17
(19) (a) Tomasi, J.; Persico, M. Chem. ReV. 1994, 94, 2027. (b) Takano, Y.;
Houk, K. N. J. Chem. Theory Comput. 2005, 1, 70.
(20) For discussions of DFT calculations on pericyclic reactions, see: (a) Guner,
V. A.; Khuong, K. S.; Houk, K. N.; Chuma, A.; Pulay, P. J. Phys. Chem.
A 2004, 108, 2959. (b) Guner, V.; Khuong, K. S.; Leach, A. G.; Lee, P.
S.; Bartberger, M. D.; Houk, K. N. J. Phys. Chem. A 2003, 107, 11445.
(21) For recent applications of DFT calculations on mechanistic studies of
organic and organometallic reactions, see: (a) Niu, S.; Hall, M. B. Chem.
ReV. 2000, 100, 353. (b) O’Neil, L. L.; Wiest, O. J. Org. Chem. 2006, 71,
8926. (c) Zhong, G.; Chan, B.; Radom, L. J. Am. Chem. Soc. 2007, 129,
924. (d) Gutta, P.; Tantillo, D. J. J. Am. Chem. Soc. 2006, 128, 6172. (e)
Nova, A.; Ujaque, G.; Maseras, F.; Lledos, A.; Espinet, P. J. Am. Chem.
Soc. 2006, 128, 14571.
(22) For recent applications of DFT to study diradical species, see: (a) Yu,
Z.-X.; Caramella, P.; Houk, K. N. J. Am. Chem. Soc. 2003, 125, 15420.
(b) Jabbari, A.; Houk, K. N. Org. Lett. 2006, 8, 5975. (c) Bethke, S.; Hrovat,
D. A.; Borden, W. T.; Gleiter, R. J. Org. Chem. 2004, 69, 3294. (d)
Hoenigman, R. L.; Kato, S.; Bierbaum, V. M.; Borden, W. T. J. Am. Chem.
Soc. 2005, 127, 17772. (e) Nummela, J. A.; Carpenter, B. K. J. Am. Chem.
Soc. 2002, 124, 8512.
(17) Carsky, P.; Hubak, E. Theor. Chim. Acta 1991, 80, 407.
(18) (a) Fukui, K. J. Phys. Chem. 1970, 74, 4161. (b) Gonzalez, C.; Schlegel,
H. B. J. Chem. Phys. 1989, 90, 2154. (c) Gonzalez, C.; Schlegel, H. B. J.
Phys. Chem. 1990, 94, 5523.
9
J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007 10775

A R T I C L E S
Chen et al.
Figure 1. Computed energy surfaces of pathways B and C of model reaction I between tetraene 1 and acetylene at the (U)B3LYP/6-31+G(d) level.
Scheme 4. The Four Model Reactions Studied by DFT Calculations
reaction I. The DFT optimized transition states involved in both
pathways are given in Figure 2 (the discussed atom numbering
is also given in this Figure).
level of theory and at the Hartree-Fock theory with various
basis sets (see Supporting Information for this concerted [8+2]
TS). However, at the B3LYP/6-31+G(d) level, such a concerted
[8+2] transition state could not be located. Using other DFT
methods (such as BP86 and MPW1K) and the MP2 method,
such a concerted [8+2] transition structure could not be obtained
either. The failure of locating a concerted [8+2] cycloaddition
at the B3LYP/6-31+G(d) level and others could be due to the
possibility that such a concerted process is energetically
disfavored as compared to the stepwise diradical pathway B.
Calculations indicated that the concerted [8+2] transition
structure at the B3LYP/6-31+G(d)//B3LYP/6-31G(d) level is
higher in energy than diradical C-C bond formation transition
The parent tetraene of dienylisobenzofuran 1 has two
conformers, 1-t and 1-c (t and c denote the s-trans and s-cis
configurations of the butadienyl moiety in 1). Calculations
indicate that 1-t is the ground-state conformer and is more stable
than 1-c by 5.4 kcal/mol in terms of free energy. However,
conformer 1-c, in which the terminal carbons C1 and C8 are
geometrically close with a distance of 4.25 Å, is expected to
be the reacting conformer for a concerted [8+2] cycloaddition
through pathway A. We can locate a concerted [8+2] transition
structure between 1-c and acetylene at the B3LYP/6-31G(d)
9
10776 J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007

[8+
2] Cycloadditions of Dienylisobenzofurans
A R T I C L E S
cataloged as a normal-electron demand Diels-Alder cycload-
dition because the energy gap of HOMO1-LUMOacelylene is
smaller than the energy gap of HOMOacetylene-LUMO1 (see
Figure 3).23,24 In addition, the charge transfer from 1 to acetylene
in TS6 is 0.014 electrons in terms of Mulliken charge, further
demonstrating the character of normal electron demand Diels-
Alder reaction for TS6. The FMO of 1-c is also given in Figure
3. We can understand why a concerted [8+2] is not favored in
view of the fact that the orbital coefficient of C8 is smaller
than both C1 and C4 in the HOMO of 1-c, suggesting that C8
is not reactive in cycloaddition as compared to C1 and C4 in 1.
This is also consistent with the high reactivity of the furan
moieties in isobenzofurans.25
The parent [8+2] reaction of pathway C could stop at the
Diels-Alder reaction if the ensuing [1,5]-vinyl shift is difficult.
Calculations show that the [1,5]-vinyl shift in pathway C starts
from an s-trans to s-cis interconversion step, transforming 7-t
to 7-c, which is the reacting conformer for [1,5]-vinyl shift. The
energy required to undergo cis/trans isomerization is negligible
when compared to those required for the [4+2] and the [1,5]-
vinyl shift.26
We could locate a concerted, closed-shell singlet [1,5]-vinyl
shift transition structure using the restricted B3LYP/6-31+G-
(d) method. However, it was found that a concerted open-shell
singlet diradical transition structure TS8 located at the UB3LYP/
6-31+G(d) level is more stable by 3.5 kcal/mol than the closed-
shell singlet vinyl shift transition structure.16,17 The diradical
TS8 with a computed S2 of 0.51 has a long breaking bond
(1.91 Å) and a short forming C-C bond (1.57 Å). This is a
very late transition state because the forming bond is almost
formed. The overall process from 7-t f 7-c f 5 requires an
activation free energy of 30.5 kcal/mol.
The rate-determining step in pathway C is the [4+2]
cycloaddition, indicating that the activation free energy required
for pathway C is 32.8 kcal/mol. Comparing this to the activation
free energy of 34.1 kcal/mol required for pathway B, we can
conclude pathway B is disfavored by 1.3 kcal/mol in terms of
activation free energy.
Figure 2. Computed geometries of transition states involved in pathways
B and C of model reaction I. Distances are in angstroms.
structure TS2 (see below) of pathway B by 7.8 kcal/mol,
indicating that concerted [8+2] is inherently disfavored as
compared to pathway B of the parent [8+2] cycloaddition.
The first step in pathway B is the formation of a single C-C
bond between C1 and C9 via a diradical transition structure
TS2, in which the forming C-C bond distance is 1.94 Å. The
alkyne moiety in this transition structure is pointing away from
the benzofuran ring with the dihedral angle of C10-C9-C1-
C2 of 112.9°. The computed S2 for TS2 is 0.33. This step
requires an activation free energy of 32.7 kcal/mol and an
activation energy of 24.5 kcal/mol. The formed diradical
intermediate 3 with a computed S2 of 1.05 is higher in energy
than the reactants by 22.2 kcal/mol. The second step is a one-
step ring-closure reaction via TS4, which is higher in free energy
than the diradical intermediate 3 by 11.9 kcal/mol and has a
computed S2 of 0.81. Formation of the [8+2] cycloadduct 5
is exergonic by 24.0 kcal/mol. Although TS2 and TS4 are close
in terms of electronic energy (24.5 vs 24.2 kcal/mol), the latter
is higher than the former by 1.4 kcal/mol in terms of free energy,
suggesting that in gas phase, the rate-determining transition state
in the stepwise diradical pathway B is the ring-closure TS4 and
the overall activation free energy of pathway B is 34.1 kcal/
mol.
Pathway C of model reaction I starts from a Diels-Alder
reaction between 1-t and acetylene via TS6. Transition structure
TS6 has almost the same free energy as that of TS2 of pathway
B. The [4+2] transition structure TS6 is concerted but asyn-
chronous with the two forming C-C bonds of 2.09 (C1-C9)
and 2.42 Å (C4-C10), respectively. IRC calculations indicate
that TS6 leads to formation of 7 without involvement of a
zwitterionic intermediate. The [4+2] cycloadduct 7 also has two
conformers, 7-c and 7-t, with the former being higher in energy
than the latter by 4.3 kcal/mol. The [4+2] step in pathway C
requires an activation free energy and an activation energy of
32.8 and 23.9 kcal/mol, respectively, and is exergonic by 5.9
kcal/mol. The [4+2] cycloaddition step in pathway C can be
The above theoretical analysis revealed that model reaction
I between 1-t and acetylene has two competitive pathways, B
and C. The stepwise diradical pathway B can lead to the
formation of a [8+2] cycloadduct, whereas stepwise [4+2]/
[1,5]-vinyl shift pathway C can lead to the formation of a [4+2]
cycloadduct, which can then isomerize to the [8+2] cycloadduct.
(23) FMO theory: (a) Fleming, I. Frontier Orbitals and Organic Chemical
Reactions; Wiley & Sons: New York, 1996. (b) Dewar, M. J. S. The
Molecular Orbital Theory for Organic Chemistry; McGraw-Hill: New
York, 1969. (c) Zimmerman, H. E. Acc. Chem. Res. 1971, 4, 272. (d)
Sustmann, R. Tetrahedron Lett. 1971, 12, 2721. (e) Sustmann, R.; Schubert,
R. Angew. Chem., Int. Ed. Engl. 1972, 11, 840. (f) Sustmann, R. Pure
Appl. Chem. 1974, 40, 569. (g) Houk, K. N. Acc. Chem. Res. 1975, 8, 361.
(h) Houk, K. N. J. Am. Chem. Soc. 1973, 95, 4092.
(24) For recent reviews and researches on normal and inverse electron-demanded
Diels-Alder reactions, see: (a) ref 23d. (b) Sauer, J.; Sustmann, R. Angew.
Chem., Int. Ed. Engl. 1980, 19, 779. (c) Yu, Z.-X.; Dang, Q.; Wu, Y.-D.
J. Org. Chem. 2001, 66, 6029. (d) Yu, Z.-X.; Dang, Q.; Wu, Y.-D. J. Org.
Chem. 2005, 70, 998. (e) Dai, M.; Sarlah, D.; Yu, M.; Danishefsky, S. J.;
Jones, G. O.; Houk, K. N. J. Am. Chem. Soc. 2007, 129, 645. (f) Gomez-
Bengoa, E.; Helm, M. D.; Plant, A.; Harrity, J. P. A. J. Am. Chem. Soc.
2007, 129, 2691.
(25) For references of isobenzofurans, see: (a) Wiersum, U. E.; Mijs, W. J. J.
Chem. Soc., Chem. Commun. 1972, 347. (b) Berson, J. A. J. Am. Chem.
Soc. 1953, 75, 1240. (c) Rio, G.; Scholl, M.-J. J. Chem. Soc., Chem.
Commun. 1975, 474. (d) Faragher, R.; Gilchrist, T. L. J. Chem. Soc., Perkin
Trans. 1 1976, 336. (e) Tobia, D.; Rickborn, B. J. Org. Chem. 1987, 52,
2611.
(26) For a study of the cis-trans interconversion of butadiene, see: Squillacote,
M. E.; Liang, F. J. Org. Chem. 2005, 70, 6564.
9
J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007 10777

A R T I C L E S
Chen et al.
Figure 3. Frontier molecular orbitals of 1, acetylene, and DMAD. The orbital coefficients are referred to the 2p part of each heavy atom. Atom numbering
is referred to that in Figure 2.
The dominant pathway for the [8+2] cycloaddition of model
reaction I is pathway C because it is preferred over pathway B
by 1.3 kcal/mol in terms of activation free energy in gas
phase.
locate a concerted [8+2] transition structure only led to locating
a [4+2] cycloaddition transition structure, further demonstrating
that a direct concerted [8+2] cycloaddition is disfavored as
compared to a [4+2] cycloaddition.
It is interesting to point out that the formation of the [8+2]
cycloadduct is thermodynamically preferred over the formation
of the [4+2] cycloadduct because the [8+2] cycloadduct is more
stable than the [4+2] cycloadduct by 18.1 kcal/mol. This
thermodynamic preference of [8+2] cycloadduct is the driving
force for the [1,5]-vinyl shift. The lower energy of [8+2]
cycloadduct as compared to the [4+2] cycloadduct can be
attributed to less ring strain and more conjugation in the [8+2]
cycloadduct relative to the [4+2] cycloadduct.
As compared to the gas-phase energy surfaces of model
reaction I, in solution, the stepwise diradical pathway B and
the [4+2] step of pathway C are disfavored by 2.2 and 1.7 kcal/
mol, respectively (see Figure 1). In contrast, the vinyl shift step
in solution is easier by 2.0 kcal/mol with respect to that
rearrangement in gas phase (28.5 vs 30.5 kcal/mol). Neverthe-
less, in solution, pathway C is still favored over pathway B for
model reaction I.
3.2. Theoretical Study of Model Reaction II and the
Experimental Test of the Theoretical Prediction. We now
turn our attention to model reaction II between 1 and DMAD
to investigate its reaction mechanism and energetic differences
with respect to those in model reaction I. Figure 4 shows the
computed energy surfaces of pathways B and C. The optimized
geometries of the transition states involved are given in Figure
5. Calculations show that pathway A is not feasible, similar to
model reaction I. All attempts to use other DFT functionals in
conjunction with either 6-31G(d) or 6-31+G(d) basis set to
In contrast to model reaction I, where pathway B is a stepwise
diradical pathway, pathway B here is through a stepwise
zwitterionic route. The first step of pathway B is the formation
of a zwitterionic intermediate 10 via transition structure TS9,
which has the forming bond distance of C1-C9 of 2.06 Å. In
TS9, the C9 atom must point away from the furan ring with
the dihedral angle of C10-C9-C1-C4 of 179.1°, otherwise,
only concerted [4+2] transition structure TS13 of pathway C,
where both C9 and C10 are above the furan moiety of 1, can
be located. The formation of 10 requires an activation free
energy of 33.5 kcal/mol and an activation energy of 21.4 kcal/
mol in gas phase. The following ring-closure step to form the
[8+2] cycloadduct is very easy and occurs with an activation
free energy of 2.9 kcal/mol. All attempts to locate a ring-closure
transition structure for the formation of a [4+2] cycloadduct
14-t from intermediate 10 (instead of a [8+2] cycloadduct 12)
were unsuccessful, suggesting 14-t is formed only through a
Diels-Alder reaction (see discussions below). As compared to
model reaction I, pathway B of model reaction II has the C1-
C9 bond formation as the rate-determining step with an
activation free energy of 33.5 kcal/mol, which is 0.6 kcal/mol
lower than the stepwise diradical pathway B of model reaction
I. However, pathway B of model reaction II is still disfavored
as compared to pathway C, which has an activation free energy
of 30.5 kcal/mol (see below). This suggests that if the [8+2]
cycloadduct could be obtained in model reaction II, the reaction
would take place through pathway C.
9
10778 J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007

[8+
2] Cycloadditions of Dienylisobenzofurans
A R T I C L E S
Figure 4. Potential energy surface of pathways B and C of model reaction II between 1 and DMAD computed at the (U)B3LYP/6-31+G(d) level.
Subsequent s-trans to s-cis interconversion from 14-t to 14-c
and a vinyl shift process takes place with an activation free
energy of 30.4 kcal/mol via TS15. This isomerization step is
exergonic by 9.2 kcal/mol. The vinyl shift transition structure
in model reaction II is also a very late transition structure with
distances of 1.99 and 1.57 Å for the breaking and forming bonds,
respectively.
In solution, pathway B has an activation free energy of 36.6
kcal/mol, 3.1 kcal/mol higher than that in gas phase. In solution,
the [4+2] step of pathway C also becomes more difficult by
3.4 kcal/mol in terms of activation free energy. Similar to model
reaction I, the [1,5]-vinyl shift in solution is easier by 1.4 kcal/
mol as compared to that in gas phase. Figure 3 shows that even
in solution, pathway C is preferred over pathway B.
If model reaction II could occur via pathway C, the [1,5]-
vinyl shift would be the rate-determining step. This is due to
the fact that the activation free energy of the [1,5]-vinyl shift is
higher by 3.5 kcal/mol than that of the [4+2] cycloaddition (30.4
vs 26.9 kcal/mol) in gas phase. In solution, the vinyl shift
requires an activation free energy of 29.0 kcal/mol, 1.3 kcal/
mol lower than the [4+2] step with an activation energy of 30.3
Figure 5. Computed geometries of transition states involved in pathway
kcal/mol. It is well known that computationally calculated
activation free energies for bi- or trimolecular reactions in
solution are overestimated by a few kcal/mol.28 In this regard,
in solution, the [1,5]-vinyl shift is still more difficult than the
bimolecular [4+2] process. Because the formation of [4+2]
cycloadduct is exergonic and the following [1,5]-vinyl shift
requires higher activation free energy, we hypothesized that
model reaction II could stop at the formation of the [4+2]
cycloadduct, if the reaction temperature is decreased or the
reaction time is shortened. If we could obtain the [4+2]
C of model reaction II. Distances are in angstroms.
Pathway C of model reaction II starts from an exergonic (by
7.5 kcal/mol) [4+2] cycloaddition with an activation free energy
of 26.9 kcal/mol and an activation energy of 13.8 kcal/mol via
TS13. As compared to the Diels-Alder reaction step in pathway
C of model reaction I, the activation free energy of the [4+2]
step in model reaction II is lower by 5.9 kcal/mol. This can be
well rationalized by FMO theory because the LUMO of DMAD
is lower than that of acetylene by 3.32 eV (Figure 3). In TS13,
the charge transfer from 1-t to DMAD is 0.217 electrons in
terms of Mulliken charge. Transition structure TS13 is more
asynchronous because the forming C-C bond is 2.06 Å while
the other forming C-C bond is hardly formed at all with a
distance of 2.94 Å. This [4+2] process can be regarded as a
two-stage process because the formation of one bond is more
advanced than the other without involvement of a zwitterionic
intermediate.27
(27) (a) Dewar, M. J. S.; Olivella, S.; Stewart, J. J. P. J. Am. Chem. Soc. 1986,
108, 5771-5779. (b) Domingo, L. R.; Picher, M. T.; Arroyo, P.; Sa´ez, J.
A. J. Org. Chem. 2006, 71, 9319.
(28) For discussions of entropy overestimation in bimolecular reactions in
aqueous solution, see: (a) Strajbl, M.; Sham, Y. Y.; Villa, J.; Chu, Z.-T.;
Warshel, A. J. Phys. Chem. B 2000, 104, 4578. (b) Hermans, J.; Wang, L.
J. Am. Chem. Soc. 1997, 119, 2707. (c) Amzel, L. M. Proteins 1997, 28,
144. (d) Yu, Z.-X.; Houk, K. N. J. Am. Chem. Soc. 2003, 125, 13825. (e)
Xia, Y.; Liang, Y.; Chen, Y.; Wang, M.; Jiao, L.; Huang, F.; Liu, S.; Li,
Y.; Yu, Z.-X. J. Am. Chem. Soc. 2007, 129, 3470.
9
J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007 10779

A R T I C L E S
Chen et al.
Scheme 5. Experiments To Isolate [4+2] Cycloadduct and Test [4+2] Cycloadduct’s Isomerization
Scheme 6. Crossover Experiment
cycloadduct, we then would be able to monitor the [1,5]-vinyl
shift process experimentally to support our computational
prediction.
Isomerization of 17 in the presence of 10 equiv of DMAD-
d6 (CD3O2C-CtC-CO2CD3) at 85 °C in deuterated benzene
was examined (Scheme 6). After this reaction system was heated
for 8 h, only the non-deuterated 18 was observed without any
incorporation of deuterated DMAD. This assignment was based
on the ratio of hydrogens on ester group of 18 as compared to
the other hydrogens of 18 according to the signals in the H
NMR spectrum. Experimental observation of no formation of
any crossover product 18′ clearly demonstrated that the isomer-
To test our hypothesis, we focused on the reaction between
16 and DMAD shown in Scheme 5. In the previous report,7a
the reaction was conducted in toluene at 85 °C for 3.5 h and
only [8+2] cycloadduct 18 was obtained. We speculated that
by running the [8+2] reaction between 16 and DMAD at lower
temperature or heating this reaction in a shorter time, we could
isolate the [4+2] cycloadduct 17. Monitoring the isomerization
of 17 to 18 would then give direct evidence to either support
or disprove our theoretical prediction in Figure 4. Therefore,
we followed the previous procedure to synthesize substrate 16
and then ran the reaction between 16 and DMAD in toluene at
85 °C. To our excitement, after a 50-min reaction period, a
mixture of [4+2] and [8+2] cycloadducts 17 and 18 in a ratio
of 3.3:1 was obtained with a total yield of 59% (Scheme 5).
Figure 6 shows how 17 gradually isomerized to 18 upon
1
1
heating at 80 °C in deuterated benzene by H NMR spectros-
copy. Formation of 18 became obvious after 30 min of heating
17, when the ratio of 18/17 was 1:5. The ratio of 18/17 increased
after further heating. After about 3.5 h, the major substance in
the NMR tube was 18 with the ratio of 18/17 of 10:1. This
experiment suggests that 17 can be transformed to 18, but cannot
guarantee that this transformation occurs via an intramolecular
[1,5]-vinyl shift. This is due to the possibility that the generation
of 18 from 17 could arguably take place through a two-step
process, starting from the retro-Diels-Alder reaction of 17 to
generate dienylisobenzofuran 16 and DMAD, which then react
with one another through a concerted or stepwise [8+2]
cycloaddition process to give the [8+2] cycloadduct 18. To rule
out this possibility, it is necessary to prove that the isomerization
of 17 to 18 is an intramolecular process instead of an
intermolecular process.
Figure 6. Monitoring the vinyl shift of 17 to 18 by 1H NMR in C6D6. The
atom labeling is given in Scheme 5.
9
10780 J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007

[8+
2] Cycloadditions of Dienylisobenzofurans
A R T I C L E S
pathway C, a simple experimental test can determine whether
compounds 24a-d were formed via [1,5]-vinyl shift isomer-
izations from compounds 23a-d. Therefore, we synthesized
and isolated four [4+2] cycloadducts 23a-d according to the
previous procedures.7a To our surprise, heating these [4+2]
cycloadducts 23a-d in dioxane at 85 °C did not lead to
formation of 24a-d. Heating compounds 23a-d longer or
under higher temperatures (in toluene at 110 °C) only resulted
in decomposition. These experiments demonstrated that [1,5]-
vinyl shifts are difficult for 23a-d, and the formation of 24a-d
is not through pathway C.
Why is there a mechanistic difference for the [8+2] reactions
of 22a-d versus 16 toward DMAD? After comparing the
structural differences between 22a-d and 16, we hypothesized
that the different mechanisms for the formation of [8+2]
cycloadducts might be attributed to the methoxy group in 22a-
d. In model reaction II, pathway B is disfavored with respect
to pathway C. However, we reasoned that in the cases of 22a-
d, the methoxy group in the diene moiety could stabilize the
zwitterionic transition structures in pathway B to a greater extent
than the transition structures of pathway C, making pathway B
competitive with pathway C. In this scenario, the final [8+2]
cycloadducts could result from either pathway B or C. However,
if the [1,5]-vinyl shift is difficult and pathway C would stop at
the [4+2] step, the final products would consist of both [8+2]
cycloadducts (via pathway B) and [4+2] cycloadducts (via the
Diels-Alder reaction of pathway C). To better explain the
experimental results in Scheme 7, we studied model reactions
III and IV to investigate the influence of the methoxy group on
the reaction outcome. We will first discuss model reaction III
and the reaction of 22a and DMAD. We will then discuss model
reaction IV and the reactions of 22b-d and DMAD to explore
whether the cyclic ring in the terminal alkene of the tetraenes
has additional influence on the reaction mechanism. The
computed energy surfaces for pathways B and C of model
reaction III are given in Figure 9.
Figure 7. Arrhenius plot for the [1,5]-vinyl shift of 17 to 18.
ization of 17 to 18 takes place in an intramolecular fashion and
occurs via a [1,5]-vinyl shift process.
The above theoretical and experimental evidence strongly
supports that the formation of [8+2] cycloadduct in this case
is through a stepwise mechanism via [4+2]/[1,5]-vinyl shift
process. The discovery of this newly proposed mechanism
suggests that 10-membered ring compounds could be synthe-
sized via such a ring enlargement strategy9 to complement the
one-pot [8+2] strategy highlighted in Scheme 2.
Finally, we wanted to test the accuracy of the DFT method
used in this study. The best way to do this is to compare the
DFT computed activation barrier with the experimentally
measured one for the isomerization of 17 to 18. The kinetics of
this isomerization process has been quantitatively studied by
1
means of H NMR through heating the [4+2] product 17 in
toluene-d8 (details are given in the Supporting Information). The
kinetics of the [1,5]-vinyl shift process were investigated at four
different temperatures, 75.0, 85.0, 95.0, and 105.0 °C, respec-
tively. Activation parameters k ) 9.2 × 1011 exp(-23.0/RT)
min-1 were obtained from the Arrhenius plot (Figure 7). From
the Arrhenius plot, the activation energy of the [1,5]-vinyl shift
is estimated to be 23.0 ( 2.3 kcal/mol. The measured ∆Gq,
∆Hq, and ∆Sq values of this isomerization are 24.2 kcal/mol,
22.3 kcal/mol, and -6.2 cal/(mol‚K), respectively. With this
experimentally measured activation energy in hand, we com-
puted the [1,5]-vinyl shift process from 19 to 21 depicted in
Figure 8. This system is similar to the experimental one (17 to
18) except the phenyl group in 17 is substituted by a hydrogen
atom. To our delight, the computed activation free energy from
19 to 21 in benzene is 24.0 kcal/mol, which is in great agreement
with the experimental activation energy, suggesting that DFT
is quite good in predicting the activation parameters for the
present [8+2] reaction (Figure 8). The computed ∆Hq (21.8
kcal/mol) and ∆Sq (-11.7 cal/(mol‚K)) values in gas phase are
also close to those measured experimentally in toluene.
Pathways B and C of model reaction III are similar to those
of model reaction II, and the geometries of the stationary points
in these pathways will not be discussed in detail. One major
difference of pathway B in model reactions II and III is the
activation free energies of the first C-C bond formation step,
which is 33.5 kcal/mol for model reaction II and 25.1 kcal/mol
for model reaction III, confirming our hypothesis about the
stabilizing influence of the methoxy group. Formation of
intermediate 26 in model reaction III is less endergonic than
the formation of 10 in model reaction II by 6.6 kcal/mol, further
confirming that the methoxy group is very effective at stabilizing
the zwitterionic species. The ring closure step for model reaction
III is also more facile via TS27 with an activation free energy
of 3.2 kcal/mol.
In pathway C, the [4+2] cycloaddition between the furan
moiety of 1-OMe and DMAD is easier than the corresponding
cycloaddition between furan moiety of 1 and DMAD (22.8 vs
26.9 kcal/mol in terms of activation free energy). The higher
[4+2] reactivity of 1-OMe as compared to 1 is due to the higher
HOMO energy of the methoxy-containing tetraenes (-6.39 vs
-6.46 eV) based on FMO theory. The activation free energy
of the [1,5]-vinyl shift of 30-t to 28 in model reaction III is
lower by 3.0 kcal/mol than that of the vinyl shift converting
14-t to 12 in model reaction II (27.4 vs 30.4 kcal/mol). The
3.3. Model Reactions III and IV: Methoxy Group Is
Critical To Switch Pathway C to Pathway B. In the original
report of the [8+2] cycloaddition reactions,7a it was observed
that the reactions of 22a-d with DMAD can give both [4+2]
cycloadducts 23a-d and [8+2] cycloadducts 24a-d, and the
ratio of 23a-d/24a-d depends on the substituents in 22a-d
(Scheme 7). To test whether these [8+2] reactions occur via
9
J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007 10781

A R T I C L E S
Chen et al.
Figure 8. Relative energies for the isomerization of 19 to 21 computed at the (U)B3LYP/6-31+G(d) level.
Scheme 7. Previously Reported [8+2] Reactions
easier [1,5]-vinyl shift of 30-t to 28 is due to additional
stabilization of the diradical isomerization transition state by
the methoxy group because radical species are usually stabilized
by heteroatoms.29 Pathway C of model reaction III would stop
at the [4+2] cycloaddition because the rate-determining step is
the [1,5]-vinyl shift.
In solution, pathway B has an activation free energy of 28.4
kcal/mol, 3.3 kcal/mol higher than that in gas phase. In solution,
the [4+2] step of pathway C becomes also difficult by 4.2 kcal/
mol in terms of activation free energy. Similar to model reactions
I and II, the [1,5]-vinyl shift in solution is easier by 1.8 kcal/
mol as compared to that in gas phase. Figure 9 shows that even
in solution, pathway C is preferred over pathway B.
How can the experimental observation that a reaction similar
to model reaction III stops at the [4+2] cycloaddition step be
rationalized? Even though the methoxy group can reduce the
activation free energy of the [1,5]-vinyl shift by 3 kcal/mol both
in gas phase (27.4 vs 30.4 kcal/mol) and in solution (25.6 vs
29.0 kcal/mol), this isomerization of 30-t to 28 of model reaction
III in solution is still about 1.4 kcal/mol higher than that of the
isomerization of 17 to 18, which is experimentally observed
and has a measured activation energy of 24.2 kcal/mol. The
model isomerization of 30-t to 28 is very similar to the
isomerization of 23a to 24a, suggesting that the isomerization
of 23a to 24a also requires an activation energy close to 25.6
kcal/mol. Therefore, the activation free energy of 23a to 24a
in solution is estimated to be about 1.4 kcal/mol higher than
that of the isomerization of 17 to 18. The isomerization of 17
to 18 can finish at 85 °C in 3.5 h; however, the isomerization
of 23a to 24a should take longer or at higher temperature.
Heating 23a to a higher temperature causes decomposition,
which may occur through pathways with activation energies
lower than [1,5]-vinyl shift, to become competitive with
isomerization. This is the reason why [1,5]-vinyl shift does not
happen for 23a.
Now let us focus on explaining why the reactions between
22a and DMAD gave both [4+2] and [8+2] cycloadducts.
Model reaction III shows that pathway B can give an [8+2]
cycloadduct and pathway C can give a [4+2] cycloadduct, which
does not isomerize to the [8+2] cycloadduct. The ratio of [8+2]/
(29) For discussions of stabilization of radicals and diradicals, see: (a) Baldwin,
J. E. Chem. ReV. 2003, 103, 1197. (b) Togo, H. AdVanced Free Radical
Reactions for Organic Synthesis; Elsevier Press: Amsterdam, 2004.
9
10782 J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007

[8+
2] Cycloadditions of Dienylisobenzofurans
A R T I C L E S
Figure 9. Potential energy surface of pathways B and C of model reaction III between 1-OMe and DMAD computed at the (U)B3LYP/6-31+G(d) level.
Figure 10. Potential energy surface of pathways B and C of the model reaction IV between 1-OMe-C and DMAD computed at the (U)B3LYP/6-31+G(d)
level.
[4+2] would be determined by the relative activation free
energies of pathways B and C. In gas phase, the formed products
would be dominated by the [4+2] cycloadduct because pathway
B to give [8+2] cycloadduct requires an activation free energy
that is about 2.3 kcal/mol higher than the [4+2] transition
structure TS29. However, in solution, the percentage of [8+2]
cycloadduct will increase because the zwitterionic TS25 and
intermediate 26 will be stabilized more significantly by the
solvent interaction as compared to the [4+2] transition state.
TS25 has a dipole moment of 6.12 debye, while TS29 has a
dipole moment of 5.40 debye. Calculations show the preference
of pathway C over pathway B is decreased to about 1.4-1.0
kcal/mol in benzene and other solvents such as dioxane, ether,
THF, and methanol (see Supporting Information for details).
These computed results suggest that both [4+2] and [8+2]
cycloadducts in a ratio ranging from 10:1 to 6:1 can be obtained.
This computationally predicted ratio for model reaction III in
solution is reasonably comparable to the experimental measured
ratio of 1:3 for the reaction of 22a with DMAD (Scheme 7).
Because the experimental systems contain a phenyl substituent
at the furan moiety that is not present in the computational
models, the DFT computed [8+2]/[4+2] ratio will be slightly
different from the experimentally measured one.
Most of the [8+2] reactions reported previously have a cyclic
structure in the terminal alkenes of the tetraenes (see 22b-
d).7a To investigate the influence of its presence on the [8+2]
reactions, model reaction IV has also been computed. The
energy surfaces of pathways B and C are given in Figure 10,
and the computed geometries of stationary points involved are
given in the Supporting Information.
9
J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007 10783

A R T I C L E S
Chen et al.
Figure 10 shows that the mechanism of 1-OMe-C is similar
to that of model reaction III. Let us first discuss why [1,5]-
vinyl shifts for 23b-d occur with difficulty. The computed
isomerization of 37 to 35 in solution requires an activation free
energy of 27.8 kcal/mol, 2.2 kcal/mol higher than the isomer-
ization of 30-t to 28 in model reaction III. This suggests that
isomerization of 37 to 35 is even more difficult than other
competitive pathways of the [4+2] cycloadducts. Consequently,
we could not observe a [1,5]-vinyl shift for 23b-d, which has
estimated activation free energies for isomerization of about 28.9
kcal/mol in gas phase and 27.8 kcal/mol in solution, similar to
the rearrangement of 37 to 35 in Figure 10.
Therefore, model reaction IV will give both [4+2] cycload-
duct 37 and [8+2] cycloadduct 35. In gas phase, pathway C is
favored by 2.9 kcal/mol, suggesting that the dominant product
of model reaction IV is [4+2] cycloadduct 37 and that the [8+2]
cycloadduct is almost negligible. However, the computed
preference of 37 over 35 is reduced by around 1.5 kcal/mol in
various solvents such as dioxane, benzene, methanol, ether, and
THF. This predicts that the 37:35 ratio should be around 10:1.
Although the predicted values do not indicate that the [8+2]
cycloadduct should be major as was observed experimentally,
the calculations do suggest that the formations of 37 and 35
are competitive processes. Because the calculations were
conducted on systems where R1 and R2 ) H, it is reasonable
that subtle structural changes could alter the [8+2]/[4+2] ratio,
as was observed experimentally. For example, it is likely that
sterically different substituent groups (e.g., TMS vs n-butyl)
would have very different solvation energies. Comparison of
the energy surfaces of model reactions III and IV indicates that
the cyclic substituents in the tetraenes do not significantly alter
the reaction scenario. The only difference is the [1,5]-vinyl shift
in model reaction IV is more difficult.
favored. The stepwise pathway B via formation of either a
diradical or a zwitterionic intermediate followed by ring closure
can occur for the [8+2] cycloadditions. However, this pathway
is not favored as compared to pathway C, which starts from a
concerted, asynchronous [4+2] reaction and a diradical [1,5]-
vinyl shift. Experiments have been performed to confirm that
one previously reported [8+2] cycloaddition occurs through
pathway C. When an electron-donating group is present in the
dienyl moieties of tetraenes, pathway B is still not favored in
gas phase but can compete in solution because the zwitterionic
transition state in pathway B can be stabilized much more
significantly than the transition structures in pathway C. These
two pathways can compete with one another to furnish both
[8+2] and [4+2] cycloadducts. In several cases, the [1,5]-vinyl
shifts are difficult for the [4+2] cycloadducts shown in Scheme
7. The confirmation of pathway C in this study suggests a new
strategy could be used to synthesize 10-membered or larger ring
compounds. The unveiling of the stepwise mechanism for the
present [8+2] cycloaddition implies that, even though many
[8+2] and other high order [m+n] cycloadditions for the
synthesis of large ring systems follow the Woodward-Hoff-
mann orbital symmetry rules, these reactions could also adopt
stepwise mechanisms because the long distance between the
terminal atoms of the reactants does not allow a concerted
mechanism. Further study of other [8+2] cycloadditions and
application of vinyl shift for ring-enlargement reactions9 are
underway.
Acknowledgment. Z.-X.Y. and his group are indebted for
generous financial support from a new faculty grant from Peking
University, the Natural Science Foundation of China (9800445,
0240203, and 20672005), and the Scientific Research Founda-
tion for the Returned Overseas Chinese Scholars, State Educa-
tion Ministry of China. J.W.H. thanks the SCORE Program of
NIH. Z.-X.Y. also thanks Prof. K. N. Houk of the University
of California at Los Angeles (UCLA) for his many insightful
suggestions, support, and encouragement of this project, and
the theoretical and synthetic organic chemistry group at PKU.
Ms. J. R. Luft from UCLA is highly appreciated for her help in
improving our writing of this paper.
From the theoretical studies of model reactions III and IV,
together with the isomerization experiments, we can conclude
that in the [8+2] reactions involving tetraenes 22a-d with a
methoxy group in the dienyl moieties, [8+2] cycloadducts 24
are formed via pathway B, and [4+2] cycloadducts 23 are
formed via direct [4+2] Diels-Alder cycloadditions. The [4+2]
cycloadducts 23 have difficulty undergoing [1,5]-vinyl shifts
due to the higher activation free energies of these processes
than those of various unknown decomposition pathways.
Supporting Information Available: Computational and ex-
perimental details, the Cartesian coordinates, energies of all
computed stationary points, characterization of 17, and synthesis
procedures of compounds 23a-d. This material is available free
of charge via the Internet at http://pubs.acs.org.
4. Conclusions
In conclusion, a combined theoretical and experimental study
of [8+2] cycloadditions of dienylisobenzofurans and DMAD
showed that the concerted, one-step [8+2] cycloaddition is not
JA072203U
9
10784 J. AM. CHEM. SOC. VOL. 129, NO. 35, 2007