Parallel mechanisms for the cycloaromatization of enyne allenes

Article abstract of DOI:10.1039/a904369a

The Myers-Saito cycloaromatization of enyne allenes is proposed to consist of two parallel mechanisms, one involving a biradical and the other with dipolar character. MCSCF calculations suggest that a nonplanar cyclic allene could be fairly close in enthalpy to the biradical, while the planar zwitterion originally proposed as a possible second intermediate is in fact a transition state for the interconversion of the two enantiomeric cyclic allenes. Competitive trapping experiments rule out the presence of a single intermediate and are consistent with the participation of parallel pathways. The reaction of (Z)-hepta-1,2,4-trien-6-yne in cyclopentadiene gave an inseparable mixture of two tetracyclic products whose structures were elucidated with 2-D NMR.

Full text of DOI:10.1039/a904369a

View Article Online / Journal Homepage / Table of Contents for this issue
Parallel mechanisms for the cycloaromatization of enyne allenes†
Thomas S. Hughes and Barry K. Carpenter*
Department of Chemistry and Chemical Biology, Cornell University, Ithaca, NY 14853-1301,
USA
Received (in Cambridge, UK) 1st June 1999, Accepted 30th July 1999
The Myers–Saito cycloaromatization of enyne allenes is proposed to consist of two parallel mechanisms, one
involving a biradical and the other with dipolar character. MCSCF calculations suggest that a nonplanar cyclic allene
could be fairly close in enthalpy to the biradical, while the planar zwitterion originally proposed as a possible second
intermediate is in fact a transition state for the interconversion of the two enantiomeric cyclic allenes. Competitive
trapping experiments rule out the presence of a single intermediate and are consistent with the participation of
parallel pathways. The reaction of (Z)-hepta-1,2,4-trien-6-yne in cyclopentadiene gave an inseparable mixture of two
tetracyclic products whose structures were elucidated with 2-D NMR.
Introduction
The mechanism of the Myers–Saito cycloaromatization of
enyne allenes is of interest because this reaction serves as a
model for the key reaction of the enyne cumulene moiety in
the antitumor agent neocarzinostatin.1 The reaction is also
notable for the mild conditions required to generate a reactive
α,3-didehydrotoluene biradical intermediate,2,3 and for the
dual modes of reactivity that have been attributed to that
intermediate.4 As the intermediate is presumably responsible
for the double hydrogen-atom abstraction that leads to the
double-stranded cleavage of DNA exhibited by these model
compounds,5 a fuller understanding of the mechanism of this
reaction seems relevant to the design of synthetic antitumor
agents.
When Myers et al. pyrolyzed parent compound 1 in various
trapping agents such as cyclohexa-1,4-diene (CHD), carbon
tetrachloride and methanol,4 products consistent with the
presence of a biradical intermediate were observed (Fig. 1).
In addition, trapping of the intermediates in methanol gave
benzyl methyl ether as the major product, which was attributed
to the reaction of methanol with a zwitterionic form of
α,3-didehydrotoluene. Given these modes of reactivity of the
intermediate, the overall mechanism for the formation of all
the products observed could involve either a single biradical
intermediate with some zwitterionic character or two distinct
intermediates, one biradical and one displaying zwitterionic
character.
Fig. 1 Single-intermediate mechanism for the formation of trapping
products from the pyrolysis of 1 in cyclohexa-1,4-diene and methanol.
behavior,6 but theory suggests that this should not be the case
for planar α,3-didehydrotoluene. In the formalism of Salem
and Rowland7 this biradical is heterosymmetric. Each of the
unpaired electrons is in an orbital that belongs to a different
representation of the symmetry group of the molecule. In such
species the mixing of biradical and zwitterionic electronic con-
figurations, as required in the single-intermediate mechanism, is
symmetry forbidden. The σ phenyl radical has AЈ symmetry
within Cs while the π benzylic radical has AЉ symmetry. Thus,
the σ1π1 biradical configuration has AЉ symmetry and the σ2π0
zwitterionic configuration has AЈ symmetry; mixing of these
electronic configurations is therefore symmetry forbidden.
Examination of solvent isotope effects on the product ratio
for the reaction of 1 in methanol allowed Myers et al. to rule
out cascade mechanisms involving initial formation of one
intermediate followed by irreversible decay to the second inter-
mediate. The rate constants for the disappearance of 1 were
found to be the same in cyclohexa-1,4-diene, a medium where
only biradical-like trapping products were observed, and in
CD3OH where only zwitterionic-like trapping products were
observed. This could imply that the paths to the biradical prod-
ucts and the zwitterionic products share a rate-determining
step, and thus arise from the same intermediate.
It is known that some biradicals can exhibit zwitterionic
† Calculated geometries and energies of intermediates are available as
supplementary data. For direct electronic access see http://www.rsc.org/
suppdata/p2/1999/2291, otherwise available from BLDSC (SUPPL.
NO. 57631, pp. 12) or the RSC Library. See Instructions for Authors
available via the RSC web page (http://www.rsc.org/authors).
J. Chem. Soc., Perkin Trans. 2, 1999, 2291–2298
This journal is © The Royal Society of Chemistry 1999
2291

View Article Online
Table 1 Calculated energies of α,3-didehydrotoluenesa
Method
2
3
7
E(3) Ϫ E(2)
E(7) Ϫ E(2)
E(3) Ϫ E(7)
RHF/6-31G*//RHF/6-31G*
Ϫ268.39111
Ϫ268.40015
Ϫ269.26436
Ϫ269.26967
Ϫ268.39112
Ϫ268.48638
Ϫ268.48762
Ϫ269.27586
—
Ϫ268.48177c
—
Ϫ268.40898
Ϫ268.40967
Ϫ269.30055
Ϫ269.29683
Ϫ268.44687
Ϫ268.53999
—
Ϫ269.32652
Ϫ269.34700
69.006e
59.9d
25.1d
48.7d
2.4d
11.2
6.0
22.7
17.0
35.0
33.6
RHFb/6-31G*//RHFa/6-31G*
MP2/6-31G*//MP2/6-31G*
Ϫ269.29952c
—
MP2b/6-31G*//MP2/6-31G*
CASSCF(2,2)/6-31G*//CASSCF(2,2)/6-31G*
CASSCF(8,8)/6-31G*//CASSCF(2,2)/6-31G*
CASSCF(8,8)/6-31G*//CASSCF(8,8)/6-31G*
CASMP2(8,8)/6-31G*//CASSCF(2,2)/6-31G*
CASPT2(8,8)/6-31G*//CASSCF(2,2)/6-31G*
ZPE
Ϫ268.46200
Ϫ268.54669
—
Ϫ269.33531
Ϫ269.35814
68.577f
44.5
37.8
9.5
4.2
37.3
0.7
5.5
7.0
1.1
31.8
0.4
67.920e
a Absolute energies given in hartrees, differences given in kcal molϪ1
energies were obtained by the unrestricted methods, UHF and UMP2. d These energy differences include a 3.0 kcal molϪ1 correction for the singlet–
triplet energy gap, see text. e Calculated from CASSCF(2,2)/6-31G* frequencies, reported in kcal molϪ1 f Calculated from RHF/6-31G* frequencies,
reported in kcal molϪ1
.
b These calculations include the Onsager model for solvent effects. c These
.
.
The accessibility of the various singlet electronic states of
α,3-didehydrotoluene has been considered theoretically. MRCI
calculations have shown the vertical excitation energy of the
ion, cyclic allene and triplet biradical. The RHF/6-31G*
method found the planar zwitterion to be 11.2 kcal molϪ1 high-
er in energy than the cyclic allene. The zwitterion was found not
to be a local minimum on the potential energy surface; thus
when the Cs symmetry constraint was removed from the geom-
etry optimization, the cyclic allene was obtained. A frequency
calculation on the planar zwitterion revealed one imaginary
frequency; it is thus a transition state for the interconversion of
the cyclic allene enantiomers. The barrier height of 11.2 kcal
molϪ1 was reduced to 5.9 kcal molϪ1 when the Onsager model
was employed to capture the potentially large effect of solvent
on the relative energies of 3 and 7, but the planar zwitterion
remained a saddle point on the potential energy surface.
Second order Møller–Plesset calculations placed the energy
of the cyclic allene 0.6 kcal molϪ1 lower in energy than the
triplet biradical, while the planar zwitterion was 22.1 kcal molϪ1
higher than the triplet biradical. Squires et al. calculated the
singlet–triplet gap to be 3 kcal molϪ1 at the MCSCF(8,8)/cc-
pVDZ//MCSCF(8,8)/3-21G level of theory,10a,14 and an inclu-
sion of this correction puts the cyclic allene 2.4 kcal molϪ1
above the singlet biradical. Again, the energy of the zwitterion
was lowered by approximately 6 kcal molϪ1 relative to the cyclic
allene by the inclusion of the Onsager model for methanol in
the MP2 method.
singlet biradical to the lowest zwitterion to be 41 kcal molϪ1 8
.
Bob Squires performed elegant collision-induced dissociation
experiments to measure the absolute heat of formation of ben-
zynes9 and didehydrotoluenes, including 2.10 While calculating
the energies of the various electronic states of the species poten-
tially generated by these experiments, he found that the ground
state closed shell structure for α,3-didehydrotoluene was in
fact not planar at the RHF level of theory, but was instead a
puckered cyclic allene, existing as a pair of enantiomers.11 Simi-
lar intermediates have been reported, their structures proposed
on the basis of the products of their trapping.12,13 The cyclic
allene could give rise to the zwitterionic reactivity observed in
methanol. For instance, the central carbon of the strained
allene could be protonated by methanol, followed by nucleo-
philic attack of methanol on the resultant benzylic cation.
CASSCF(2,2),
CASSCF(8,8),
CASMP2(8,8),
and
CASPT2(8,8) single-point calculations were performed on
CASSCF(2,2) minimized geometries. The geometry of the
zwitterion was also minimized at the CASSCF(8,8) level since,
as expected, the CASSCF(2,2) method gave an energy and
geometry identical to those obtained in the RHF calculations,
indicating that the (2,2) active space was dominated by a single
configuration, namely the σ2π0 zwitterionic configuration. This
is expected for heterosymmetric biradicals because the mixing
of open-shell with closed-shell configurations is symmetry
forbidden. The resulting difference between the calculated
single-point energies of the CASSCF(2,2) geometry and the
CASSCF(8,8) geometry was small, being only 0.8 kcal molϪ1
for CASSCF(8,8) and 0.1 kcal molϪ1 for CASMP2(8,8).
The existence and nature of the putative second intermediate
have been investigated by theoretical and experimental
methods, and the results form the basis of this paper.
The zwitterion ranged from 37.3 to 44.5 kcal molϪ1 higher in
energy than the singlet biradical in the CAS calculations. The
cyclic allene, however, was 9.5 kcal molϪ1 higher than the singlet
biradical from calculations employing a 2 × 2 active space, and
only 4.2 kcal molϪ1 higher using the larger 8 × 8 active space.
However, inclusion of dynamic electron correlation with
CASMP2(8,8) and CASPT2(8,8) methods increased this differ-
ence to 5.5 and 7.0 kcal molϪ1, respectively. A zero-point energy
correction also raised the energy of 7 with respect to 2 by an
additional 1.1 kcal molϪ1. These results are consistent with
those obtained for cyclohexa-1,2-diene15 using HF and
MCSCF methods. The calculations suggest that the cyclic
allene is more likely to be energetically competitive with the
biradical than is the zwitterion, although none of the methods
Results and discussion
Computational results
To determine whether either the zwitterion or cyclic allene is
enthalpically competitive with the singlet biradical, calculations
were performed at various levels of theory including multi-
configurational complete active space (CAS) methods. Geo-
metric parameters of the minimized geometries are provided in
the supplementary material.† Energies of the biradical, planar
zwitterion and cyclic allene calculated using several methods
are listed in Table 1.
Single determinant Hartree–Fock wavefunctions are inappro-
priate for the description of singlet biradicals, and so at this
level it is only possible to compare the energies of the zwitter-
2292
J. Chem. Soc., Perkin Trans. 2, 1999, 2291–2298

View Article Online
Table 2 B3LYP/6-31G* energies of parent and benzannelated system intermediates
2a
3
7
8a
9
10
Absolute E/hartrees
Ϫ270.22722
[0]
Ϫ270.15645
Ϫ270.21139
Ϫ423.87278
[0]
Ϫ423.79576
Ϫ423.84616
E relative to biradical/kcal molϪ1
44.4
9.9
48.3
16.7
a Triplet state.
predicts the near degeneracy that the experimental results
would require if the cyclic allene were responsible for formation
of the polar products.
It is possible that the effects of the very polar medium in
which this competition is observed, methanol, are significant
but are not captured in these MCSCF calculations. However,
given the magnitude and uniformity of the solvent correction
calculated at the RHF and MP2 levels, it seems unlikely that the
large MCSCF energy differences between 2 and 7 would be
mitigated by passive effects of the solvent.
Density functional calculations were performed on the triplet
biradical, zwitterion and cyclic allene as well as benzannelated
analogs. Calculated energies are given in Table 2. It was
expected that benzannelation would destabilize the cyclic allene
relative to the other intermediates, since the new ring could not
Fig. 2 Dependence of methanol product ratio on [CHD]. Ratios
determined by GC, line indicates linear least squares fit to data.
achieve aromaticity as it could in the case of the biradical and
zwitterion. In accord with the other theoretical models, B3LYP/
6-31G* calculations placed the zwitterion much higher in
energy than the biradical, 44.4 kcal molϪ1 in the parent system
and 48.3 kcal molϪ1 in the benzannelated system. Benzannel-
ation was found to selectively destabilize the cyclic allene (by 6.8
kcal molϪ1 with respect to the triplet biradical), as expected,
although the zwitterion was also somewhat destabilized (by 3.9
kcal molϪ1).
Competitive trapping results
Dilute methanol solutions of 1 containing varying concen-
trations of CHD were heated to 90 ЊC. The ratio of products
arising from the reaction of the enyne allene with methanol was
measured by GC as a function of the concentration of CHD.
The results are shown in Fig. 2. The ratio of products shows
a linear dependence on the concentration of CHD, with a
nonzero slope.
If the single intermediate scheme depicted above (Fig. 1)
were correct, then one would expect the ratio of the biradical-
derived products to zwitterion-derived products to be inde-
pendent of the concentration of the trapping agents present.
For the scheme in Fig. 1 this ratio would be given by eqn. (1).
Fig. 3 Dual-intermediate mechanism for the formation of trapping
products from the pyrolysis of 1 in cyclohexa-1,4-diene and methanol.
ratio is dependent on the concentration of methanol, but it is
assumed that in this two-component solvent system varying low
concentrations of CHD would leave the concentration of
methanol essentially invariant.
It is conceivable that 6 could be formed by a direct nucleo-
philic attack of methanol on 1, since this pathway would be
kinetically indistinguishable from that shown in Fig. 3 at high
methanol concentrations.
The dependence of the product ratio on the concentration of
CHD in the 0–0.2 M range does not seem explicable as a
solvent-polarity effect, since increasing the concentration of
CHD should decrease the solvent polarity and thereby disfavor
the polar pathway, whereas the opposite is observed.
When the concentration of CHD was increased beyond 0.2
M, the product ratio reached a maximum and then decreased.
[6] k4
(1)
=
[5] k3
On the other hand, if the mechanism involved two rapidly
interconverting intermediates (Fig. 3) the product ratio would
be given by eqn. (2).
[6] k4
=
k
Ϫ3(k1 ϩ k2) ϩ k1(k5[CH3OH] ϩ k6[CHD])
k3(k1 ϩ k2) ϩ k2(k4[CH3OH])
(2)
ͩ
ͪ
[5] k5
In this case, the ratio of the benzyl methyl ether to 2-
phenethanol would be linearly dependent on the concentration
of CHD, as is observed in the concentration range 0 ≤ [CHD] ≤
0.2 M. It is evident from this kinetic expression that the product
J. Chem. Soc., Perkin Trans. 2, 1999, 2291–2298
2293

View Article Online
was possible to assign the structures 14-syn and 14-anti to the
two 2:1 cycloadducts.
The formation of these cyclophanes was unexpected, but not
unprecedented. When 11 is heated, a biradical is generated
which undergoes sequential addition to two butadienes to
generate a cyclophane diadduct.18 The major products of
this reaction are cyclophane monoadducts, but presumably
the formation of such products in the case of reactive α,3-
didehydrotoluene would be sterically prohibited.
Fig. 4 Dependence of methanol product ratio on [CHD]. Ratios
determined by HPLC.
This is shown in Fig. 4. When 1 was heated to 90 ЊC in 1:1
benzene–methanol, a ratio of 6:5 of 1.97 0.02 was observed.
Thus in a less polar solvent system without CHD trapping, the
polar-pathway products decrease. On the basis of this control
experiment, it seems reasonable to attribute the decrease in 6:5
to the changing of the solvent polarity. It is true that increasing
CHD concentration from 0.2 to 1.0 M presumably involves a
very modest change in polarity, but only a small change in the
relative solvation energies of the two pathways would be
required for the ratios observed; changing from a product ratio
of 3.31 to 2.73 requires a change in the relative solvation ener-
gies of each pathway of only 0.14 kcal molϪ1. These results
imply that the second pathway involves some polar inter-
mediate or transition state, which at higher concentrations of
nonpolar components is disfavored.
The two 2:1 cyclopentadiene cycloadducts can be dis-
tinguished by their H NMRs. Molecular mechanics predicts
1
that one of the hydrogens attached to carbon 17 in the anti-
diadduct is placed over the aromatic ring, while no such con-
formational constraint is present in the syn-diadduct. The spin
1
system containing the H resonance at δ = 0.812 can thus be
Attempts to trap a cyclic allene
assigned to the anti-diadduct.
Since Squires’ calculations raised the possibility of the inter-
mediacy of a cyclic allene, several attempts were made to trap
this potential intermediate. Cyclic allenes are known to be very
reactive Diels–Alder dienophiles,16,17 and since it appeared that
neither a biradical or a zwitterion would give rise to any Diels–
Alder-like products, 1 was pyrolyzed in the presence of several
dienes. Reaction of 1 with anthracene, diphenylisobenzofuran
and tetraphenylcyclopentadienone gave complex product mix-
tures. The reaction of 1 with buta-1,3-diene and 2,3-dimethyl-
buta-1,3-diene gave a mixture of four products each. GC–MS
revealed that each of the four products in these two mixtures
had a mass corresponding to the addition of two equivalents of
diene to one equivalent of enyne allene. Very clean two com-
ponent product mixtures were obtained from the pyrolysis of 1
in the presence of cyclohexa-1,3-diene and cyclopentadiene.
GC–MS again revealed all of these products to have come from
the reaction of one molecule of 1 with two molecules of diene.
The two cyclopentadiene diadducts were purified as a mix-
ture, but all attempts to separate the compounds by chromato-
graphic means failed. To determine the structures of these
colorless oils, 1- and 2-D NMR experiments were performed.
These data are shown in Tables 3 and 4. The 1-D 1H NMR of
the mixture was complex, so the diadducts were prepared using
d6-cyclopentadiene, to determine which resonances arose from
the enyne allene-derived portion of the products. The six
hydrogens from 1 became hydrogens 10, 10Ј, 12, 13, 14 and 15
in both diadducts. The chemical shifts and coupling pattern
of these hydrogens indicated that the two diadducts were
α,3-disubstituted toluenes; this fragment is consistent with the
known reactivity of 1.
The formation of these cyclophane products explains the
product mixtures observed. Reaction with cyclopentadiene and
cyclohexa-1,3-diene gives two products, which correspond to
the syn- and anti-configurations of the two methano-bridges.
The four products obtained by trapping with butadiene
and 2,3-dimethylbutadiene correspond to the cis, cis-, cis,
trans-, trans, cis- and trans, trans-configurations of the two
double bonds in the macrocycle. Hydrogenation of the four-
component mixture of buta-1,3-diene 2:1 cycloadducts gave
two products. The initial product of this hydrogenation would
be expected to be [9]-m-cyclophane. It is known that strained
[8]-m-cyclophane can be hydrogenated under mild conditions to
the saturated hydrocarbon bicyclo[8.3.1]tetradecane.19 The two
hydrogenation products observed had masses corresponding to
the addition of two and five equivalents of hydrogen, which by
analogy are presumably [9]-m-cyclophane and bicyclo[9.3.1]-
pentadecane.
Determination of mechanism of diadduct formation
Given the unexpected products obtained from diene trapping
of 1, the question of the mechanism of their formation arises.
An obvious mechanism, as suggested above, involves the
sequential reaction of the biradical intermediate with two
equivalents of cyclopentadiene to close the macrocycle. How-
ever, if the cyclic allene were present and a Diels–Alder reaction
did occur with the diene, it is possible that bond homolysis in
the cycloadduct 12 as pictured in Fig. 5 could be com-
petitive with the 1,3-H-atom migration that would lead to a
stable aromatic product. (U)B3LYP/6-31G*//(U)HF/3-21G
calculations show the triplet biradical 13 resulting from such a
bond homolysis to be only 1.5 kcal molϪ1 higher in energy than
the Diels–Alder adduct 12, suggesting that such a pathway is
enthalpically feasible (assuming the singlet–triplet gap in the
biradical to be small).
Heteronuclear correlation NMR experiments were per-
formed, as these are excellent methods for resolving multiple
spin systems in mixtures. One-bond connectivities were deter-
mined by HMQC, and two- and three-bond connectivities were
determined using an HMBC pulse sequence. From these data, it
2294
J. Chem. Soc., Perkin Trans. 2, 1999, 2291–2298

View Article Online
Table 3 1- and 2-D NMR data for compound 14-syn
Atom
13C
1H (HMQC)
COSY (H,H)
HMBC (H→C)
HMBC (C→H)
1
2
3
4
5
6
7
8
9
10
10Ј
11
12
13
14
15
16
16Ј
17
17Ј
142.96
48.75
137.03
131.99
46.00
—
3.873
5.669
5.632–5.612
3.260–3.264
3.302
5.632–5.612
5.696–5.680
3.229–3.224
2.682
3.047
—
6.786
7.140
6.919
7.356
1.799
2.174–2.111
1.167
—
—
2,13,16,16Ј
4/7,9,14,15,16,16Ј
2,5,16
3,16,16Ј
2,4,16,16Ј
3,(10Ј)
6,16Ј,17
5,9,16
8,(10Ј)
7,17
6,10
10Ј,17
10,4/7
—
13,15
12,14
13,15
12,14
2,3,6,15,16Ј
2,3,5,16,17,17Ј
5,8,10,16Ј,17Ј
16Ј,17
1,3,4,5/9,15
5/9
(2,9,16,17)
3,4,6
2,5,16
2/3,10Ј,17Ј
5,8,16,16Ј
—
10,10Ј
2/3,4/7,10,10Ј
12,15
47.09
—
132.70
134.86
45.96
(2,9,16,17)
6,17
2,17
8,9,11,12,15
8,9,11,12,15,17
—
10,14,15
1,11
2,12,15
2,10,12,14
1,2,4,5,6
1,2,6
—
5
36.81
141.57
125.15
128.37
122.66
127.48
34.00
10,10Ј,13
10,10Ј,14,15
—
12,15
2,10,10Ј,12,14
4/7
28.70
4/7,8,9,10Ј
1.922
Table 4 1- and 2-D NMR data for compound 14-anti
Atom
13C
1H (HMQC)
COSY (H,H)
HMBC (H→C)
HMBC (C→H)
1
2
3
4
5
6
7
8
9
10
10Ј
11
12
13
14
15
16
16Ј
17
17Ј
146.13
49.41
130.61
136.65
44.53
—
3.628
5.796
6.233
3.316–3.319
3.287–3.295
5.810
5.803
3.242
2.576
3.171
—
6.826
7.153
6.903
7.046
1.234
—
—
2,13,16,16Ј
3,4,14,15
2,4,16
2,16
2,4,16
5,8,16,16Ј,17
—
10,10Ј
3,7,10,10Ј
12,15
3,4,16,16Ј
2,4,16
2,3,5,16,16Ј
4,6,16Ј,17Ј
5,9,10,16,17Ј
6,8
7,9
6,8,16Ј,17
10Ј,17
10,15
—
13,15
12,14
1,3,4,5,14,15
2,9,16,17
2,3,5,16
6
—
9
6,17
11
8,9,11,12,15
8,9,11,12,15,17
—
10,14,15
46.63
132.54
137.64
45.19
36.50
142.31
126.37
128.57
122.63
128.48
9,10,10Ј
10,10Ј,14,15
—
2,12,15
2,10,10Ј,12,14
3,4
1
13,15
2,12,15
2,10,12,14
1,3,4,5,6
1,6
—
6
10,12,14,17Ј
2,3,4,6,16Ј,17
2,4,5,9,16,17,17Ј
9,10,16,16Ј,17Ј
5,6,15,16Ј,17
37.46
28.16
2.130–2.200
0.812
1.799
3,8,10Ј
To determine which intermediate was responsible for the
formation of the 2:1 cycloadduct, a competitive trapping
experiment similar to that with CHD was performed. The ratio
of the methanol-trapped products was measured as a function
of the concentration of cyclopentadiene. The results, shown
in Fig. 6, were similar to those obtained for CHD. Since the
diadducts were formed at the expense of 5, this suggests that
the 2:1 cycloadducts arose from the biradical intermediate.
This result does not rule out the presence of 7, as it is possible
that the Diels–Alder reaction of 7 with cyclopentadiene was not
competitive with the radical addition of 2 to the diene. The
straight-line dependence of the product ratio provided further
evidence that a single pathway cannot be responsible for the
formation of all of the observed products.
Benzannelated ꢀ,3-didehydrotoluene
1-Ethynyl-2-(propa-1,2-dienyl)benzene (16) was synthesized as
depicted in Fig. 7 and heated to 90 ЊC in methanol to examine
the distribution of products. 2-Methylnaphthalene, 2-(meth-
oxymethyl)naphthalene and 2-(naphthalen-2-yl)ethanol were
observed in 1.5, 14 and 24% yields respectively. 17 and 18 arise
from the biradical intermediate 8, while 19 arises from the polar
pathway. The ratio of biradical derived to polar products was
0.55, compared to ~3 for the parent system.
This reduction of products from the polar pathway was
qualitatively consistent with the DFT energies calculated for
the various intermediates. However, since benzannelation raised
the calculated energies of both the cyclic allene and zwitterion
J. Chem. Soc., Perkin Trans. 2, 1999, 2291–2298
2295

View Article Online
mined by the relative barrier heights of the two pathways which
are undetermined.
Conclusions
The competitive trapping experiments with both CHD and
cyclopentadiene are incompatible with a single intermediate
mechanism and suggest that there are two parallel pathways for
the cycloaromatization of 1. While it seems fairly certain that
one pathway involves the biradical 2, the nature of the inter-
mediate involved in the second pathway is not clear.
The cyclic allene 7 is the more likely candidate for the second
intermediate based on the calculated energies for 7 and 3. It is
reasonable to assume that there could be some solvent stabiliz-
ation of the zwitterion in methanol that would not be captured
by the MCSCF calculations, but the 30–50 kcal molϪ1 differ-
ence in energy seems large enough to outweigh the potential
solvation energy. If the cyclic allene was present however, it is
not obvious why it was not possible to trap it with dienes, as this
Diels–Alder reaction is fairly efficient. The predicted product
ratio for the reaction of 16 in methanol was also not in accord
with that observed.
Fig. 5 Mechanisms for the formation of cyclopentadiene diadducts.
The zwitterion could be invoked as the second intermediate.
The solvent effects on the product ratio in more nonpolar reac-
tion media support a more polar intermediate, but the calcu-
lated energies seem to indicate that 3 would not be energetically
competitive with the biradical. There are also polar intermedi-
ates potentially involved in the formation of 6 from 7 which
could give rise to the observed solvent effects without invoking
3. The best candidate thus appears to be the cyclic allene, but
the evidence is not overwhelming. There is also the possibility
that the polar pathway involves a more direct participation of
the methanol, such as a direct nucleophilic attack of methanol
on 1.
A further issue needs to be resolved if two parallel mechan-
isms are to be proposed. If the rate determining step for the
formation of products from each pathway is the same, as
Myers’ results seem to indicate, when does branching occur?
Calculations8,21 and experimental results22 support an early
transition state for the cyclization of 1. MCSCF calculations
indicated that the transition state still was on the closed-shell
electronic surface. It is possible that a crossing to the open-shell
electronic surface occurs after the transition state, and this
serves as the branching point for the reaction. Thus, the branch-
ing would be post-rate determining, and therefore consistent
with the rates observed.
Fig. 6 Dependence of methanol product ratio on [cyclopentadiene].
Ratios determined by HPLC, line indicates linear least squares fit to
data.
Computational details
All ab initio calculations were performed using the GAUS-
SIAN9823 package, save the CASPT2 and CASMP2 calcu-
lations which were performed using the MOLCAS24 and
GAMESS25 software packages, respectively. Geometries were
minimized using the 6-31G* basis set26 at the RHF, MP2 and
B3LYP levels, and using a complete active space MCSCF pro-
cedure. Calculations on biradicals at the MP2 and B3LYP levels
employed the unrestricted methods and were performed on the
triplet state. The zwitterion was constrained to Cs symmetry in
all cases. The MCSCF geometry optimization for the planar
closed-shell zwitterion (1AЈ) was performed using two different
active spaces. The smaller included just the non-bonding σ
orbital and the non-bonding π orbital, while the larger included
these two non-bonding orbitals as well as the three bonding π
and three antibonding π orbitals. The active space employed for
the MCSCF geometry optimization of the cyclic allene
included the non-bonding σ and π orbitals. The CASSCF(2,2)
geometry used for the planar singlet biradical (1AЉ) was that
obtained by Squires.10a Single-point calculations at the RHF
and MP2 levels were also performed with the Onsager solvation
model, which surrounds a spherical cavity containing the solute
Fig.
7
Synthesis and pyrolysis of benzannelated enyne allene.
Reagents and conditions: a) TMSCCH, CuI, Pd(PPh3)2Cl2, NEt3, Et2O;
b) HCCMgBr, Et2O; c) K2CO3, CH3OH; d) DEAD, PPh3, o-nitro-
benzenesulfonylhydrazine, THF;20 e) CH3OH, 90 ЊC.
relative to the triplet biradical, this result does not distinguish
between those two potential intermediates. A change in the
product ratio by a factor of 5.5 corresponds to a difference in
free energy of 1.2 kcal molϪ1 at 90 ЊC. This is much lower than
the ∆Es predicted by the DFT calculations for the intermedi-
ates, which are 6.8 kcal molϪ1 for the cyclic allene and 3.9 kcal
molϪ1 for the zwitterion, but the product ratio is also deter-
2296
J. Chem. Soc., Perkin Trans. 2, 1999, 2291–2298

View Article Online
with a continuum of constant dielectric.27 MCSCF energies for
both the planar biradical and planar zwitterion were calculated
with an active space consisting of the σ and π non-bonding, the
three bonding π and three antibonding π orbitals. The active
space for the cyclic allene consisted of the π and π* orbitals.
Single point energies were also calculated using the CAS pro-
cedure with a second order perturbative correction. Frequency
calculations were performed at the RHF level for the zwitterion
using the RHF minimized geometry, and at the CASSCF(2,2)
level for the singlet biradical and cyclic allene. Molecular
mechanics were performed using the MMX force field as
implemented in PCModel.28
stirred solution was allowed to warm to room temperature over
8 hours, at which time it was concentrated in vacuo to thick
orange oil which was eluted through a column of silica with
pentane. The pentane solution was concentrated to give 50 mg
(46.3% yield) of colorless oil. IR: 3289.7 (s), 3061.8 (w), 2927.8
(w), 2101.4 (w), 1938.5 (s), 1718.4 (m), 1480.5 (w), 1444.5 (m),
1060.5 (w), 855.7 (m), 758.4 (m). 1H NMR (CDCl3, 300 MHz)
δ: 7.47 (1H, d, J = 8.1 Hz), 7.45 (1H, d, J = 7.5 Hz), 7.28 (1H,
d × d, J = 8.1, 7.5 Hz), 7.14 (1H, d × d, J = 8.1, 7.5 Hz), 6.74
(1H, t, J = 7.0 Hz), 5.17 (2H, d, J = 7.0 Hz), 3.32 (1H, s). 13C
NMR (CDCl3, 75 MHz) δ: 79.1, 82.1, 92.1, 126.7, 126.8, 129.2,
133.2 (the four quaternary carbons were not observed).
General procedure for pyrolyses
Experimental
Solutions were sealed under vacuum (0.015 torr) in thick-
walled glass tubes after deoxygenation by three to five
freeze–pump–thaw deoxygenations. All reactions were run in a
thermostatted water bath for 8–16 hours, at which time no start-
ing material was detected. After cooling, the tubes were scored
and their contents analyzed.
General
(Z)-Hepta-1,2,4-trien-6-yne,4 1-(2-trimethylsilylethynylphenyl)-
prop-2-yn-1-ol,29 and o-nitrobenzenesulfonylhydrazine30 were
prepared as described in the literature. All reagents were used
as received. Tetrahydrofuran was distilled from potassium
benzophenone ketyl, methanol was distilled from magnesium
methoxide and benzene was distilled from calcium hydride.
IR spectra were acquired with a Nicolet Impact 410 FT-IR
Competitive trapping experiments
3.55 mmolar solutions of 1 in methanol with various concen-
trations of cyclohexa-1,4-diene or cyclopenta-1,3-diene were
sealed and heated to 90 ЊC for 24 hours. The product mixtures
were quantified using m-xylene as an internal standard, using
either analytical GC or HPLC.
1
spectrometer on neat samples on KBr plates. 1-D H NMR
spectra were acquired at 200 MHz on a Varian XL 200 spec-
trometer or at 300 MHz on a Bruker AF-300 spectrometer. 13
C
NMR spectra were acquired at 75 MHz on a Bruker AF-300
spectrometer. 2-D COSY, HMQC and HMBC spectra were
acquired on a Varian Unity 500 spectrometer. All spectra were
recorded in CDCl3 using TMS as a chemical shift standard. Gas
chromatographs were recorded on a Hewlett-Packard 5880 GC
with a flame ionization detector and a 15 m × 0.25 mm RTX-5
(5% phenylmethylpolysiloxane) fused silica capillary column.
GC–mass spectra were acquired on a Hewlett-Packard 5890
GC with a Hewlett-Packard 5970 Series mass-selective detector
and a 30 m × 0.25 mm DB-5 (5% phenylmethylpolysiloxane)
fused silica capillary column. HPLC chromatograms were
obtained on a Hewlett-Packard Series 1050 HPLC with a
MWD detector and equipped with a reverse-phase HP 79916
OD Opt.574 Hypersil ODS 5 µm 200 × 4.6 mm column.
(2R,5R,6R,9S)-Tetracyclo[9.3.1.12,5.16,9]heptadeca-1(15),3,7,
11,13-pentaene and enantiomer (14-anti) and (2S,5S,6R,9S)-
tetracyclo[9.3.1.12,5.16,9]heptadeca-1(15),3,7,11,13-pentaene and
enantiomer (14-syn). 0.100 ml of a 70 mmolar solution of 1 in
C6D6 was mixed with 0.100 ml of cyclopenta-1,3-diene and
heated to 80 ЊC for 10 hours. 14 (syn) and 14 (anti) were
produced in a 1:1 ratio.
14 (syn): MS: m/z 222 (Mϩ, 13%), 157 (14), 156 (100), 155
(48), 153 (10), 142 (12), 141 (66), 129 (12), 128 (30), 115 (26). 1H
NMR (CDCl3, 500 MHz) δ: 7.35 (1H, s), 7.14 (1H, d × d,
J = 5.9, 5.5 Hz), 6.92 (1H, d, J = 6.7 Hz), 6.79 (1H, d, J = 5.5
Hz), 5.67–5.70 (2H, m), 5.61–5.63 (2H, m), 3.87 (1H, d × d,
J = 5.8, 2.0 Hz), 3.30 (1H, br s), 3.26–3.27 (1H, m), 3.22–3.23
(1H, m), 3.05 (1H, d × d, J = 12.6, 1.4 Hz), 2.68 (1H, d × d,
J = 12.6, 5.1 Hz), 2.11–2.17 (1H, m), 1.80 (1H, d × m, J = 11.0
Hz), 1.92 (1H, d × d × d, J = 11.8, 8.6, 8.7 Hz), 1.17 (1H,
d × d × d, J = 11.8, 2.8, 2.7 Hz).
1-(2-Ethynylphenyl)prop-2-yn-1-ol (15)
0.14 g (1.013 mmol, 1.54 equiv.) of potassium carbonate was
added to a solution of 0.15 g (0.657 mmol, 1.00 equiv.) of 14 in
25 ml of methanol. After stirring at room temperature for 20
minutes, TLC indicated the absence of starting material. The
reaction mixture was extracted with three 25 ml portions of
ethyl ether. The combined organic layers were dried over
sodium sulfate and concentrated in vacuo to give 0.09 g (90%
yield) of orange oil which was purified by chromatography on
silica using 10% ethyl acetate in hexanes. IR: 3533.85 (br),
3396.4 (s), 3283.8 (s), 3067.8 (w), 2894.4 (w), 2118.4 (w), 2105.3
(w), 1480.5 (s), 1447.5 (m), 1274.1 (m), 1019.6 (s), 951.7 (s),
760.3 (s), 657.4 (m). 1H NMR (CDCl3, 300 MHz) δ: 7.72 (1H, d,
J = 7.4 Hz), 7.52 (1H, d, J = 7.5 Hz), 7.38–7.43 (1H, m), 7.28–
7.32 (1H, m), 7.52 (1H, d, J = 7.5 Hz), 5.89 (1H, d, J = 2.2 Hz),
3.39 (1H, s), 2.95 (1H, br s), 2.65 (1H, d, J = 2.2 Hz. 13C NMR
(CDCl3, 75 MHz) δ: 63.0, 75.1, 75.7, 81.1, 83.1, 120.6, 126.9,
128.7, 129.8, 133.5, 142.5.
14 (anti): MS: m/z 222 (Mϩ, 13%), 157 (14), 156 (100), 155
(41), 153 (10), 142 (11), 141 (59), 129 (11), 128 (33), 115 (26). 1H
NMR (CDCl3, 500 MHz) δ: 7.15 (1H, d × d, J = 5.9, 5.5 Hz),
7.05 (1H, s), 6.90 (1H, d, J = 6.7 Hz), 6.83 (1H, d, J = 5.9 Hz),
6.23 (1H, d × d, J = 5.4, 2.0 Hz), 5.84 (2H, br s), 5.80 (1H,
d × m, J = 5.4 Hz ), 3.63 (1H, d × d, J = 6.3, 2.0 Hz), 3.31–3.32
(1H, m), 3.29–3.30 (1H, m), 3.24 (1H, br s), 3.17 (1H, d,
J = 12.6 Hz), 2.58 (1H, d × d, J = 12.6, 7.1 Hz), 2.13–2.20 (1H,
m), 1.80 (1H, m), 1.23 (1H, d, J = 11.0 Hz), 0.81 (1H, d × d × d,
J = 11.8, 2.4, 2.3 Hz). The d12-compounds were prepared in the
same fashion, using d6-cyclopenta-1,3-diene that was prepared
by stirring cyclopenta-1,3-diene (obtained by high-temperature
distillation of dicyclopentadiene) with five aliquots of a solu-
1
tion of sodium deuteride and DMSO. H NMR analysis of
the Diels–Alder cycloadduct of the deuterated cyclopenta-1,3-
diene with N-methyltriazolinedione indicated that the diene
had 80% deuterium incorporation.
1-Ethynyl-2-(propa-1,2-dienyl)benzene (16)
0.18 g of DEAD (1.037 mmol, 1.35 equiv.) were added drop-
wise via syringe to an ice-cooled solution of 0.26 g PPh3 (1.00
mmol, 1.3 equiv.) in 10 ml THF. After stirring for 10 minutes, a
solution of 0.12 g of 15 (0.768 mmol, 1.0 equiv.) in 5 ml THF
was added via syringe. After an additional 10 minutes of stir-
ring, a solution of 0.22 g o-nitrobenzenesulfonylhydrazine (1.00
mmol, 1.3 equiv.) in 5 ml THF was added via syringe. The
Pyrolysis of 1 in buta-1,3-diene
Approximately 0.5 ml of buta-1,3-diene was condensed into a
tube containing 0.100 ml of a 70 mmolar solution of 1 in C6D6.
The tube was sealed and heated to 80 ЊC for 16 hours. GC–MS
revealed four products with a mass of 198. This mixture was
J. Chem. Soc., Perkin Trans. 2, 1999, 2291–2298
2297

View Article Online
Wierschke, J. J. Nash and R. R. Squires, J. Am. Chem. Soc., 1993,
115, 12 611.
concentrated in vacuo to remove unreacted buta-1,3-diene, and
the residue dissolved in methanol. The methanolic solution was
hydrogenated (40 lbs inϪ2, 2.76 × 10Ϫ3 N mϪ2) over palladium
on carbon for 24 hours. The products of this reaction were
analyzed by GC–MS, and were produced in a 60:40 ratio. The
MS of the major product: MS: m/z 202 (Mϩ, 43%), 145 (26),
131 (67), 118 (58), 117 (57), 115 (29), 106 (28), 105 (60), 104
(100), 103 (29), 78 (28), 77 (26). MS of the minor product: m/z
208 (Mϩ, 14%), 109 (11), 97 (31), 96 (29), 95 (29), 83 (24), 82
(49), 81 (79), 79 (14), 69 (30), 68 (14), 67 (71).
11 M. Balci and W. M. Jones, J. Am. Chem. Soc., 1980, 102, 7607;
M. Balci and W. M. Jones, J. Am. Chem. Soc., 1981, 103, 2874.
12 R. P. Johnson, Chem. Rev., 1989, 89, 1111; G. Wittig and P. Fritze,
Angew. Chem., Int. Ed. Engl., 1966, 5, 846.
13 M. Christl, M. Braun and G. Müller, Angew. Chem., Int. Ed. Engl.,
1992, 31, 473; H. Hopf, H. Berger, G. Zimmermann, U. Nüchter,
P. G. Jones and I. Dix, Angew. Chem., Int. Ed. Engl., 1997, 36,
1187.
14 For an experimental estimate of the singlet–triplet gap, see C. F.
Logan, J. C. Ma and P. Chen, J. Am. Chem. Soc., 1994, 116, 2137.
15 R. O. Angus, Jr., M. W. Schmidt and R. P. Johnson, J. Am. Chem.
Soc., 1985, 107, 532.
Pyrolysis of 1 in 2,3-dimethylbuta-1,3-diene
16 W. R. Moore and W. R. Moser, J. Org. Chem., 1970, 35, 908.
17 A. T. Bottini and L. L. Hilton, Tetrahedron, 1975, 31, 1997.
18 T. Tsuji, T. Shibata, Y. Hienuki and S. Nishida, J. Am. Chem. Soc.,
1978, 100, 1806.
0.100 ml of a 70 mmolar solution of 1 in C6D6 was mixed with
0.50 ml of the diene and sealed in a tube. The solution was
heated to 80 ЊC for 10 hours. Four products with the same MS
pattern were distinguished by GC–MS, which accounted for
7%, 40%, 16% and 38% of the product mixture. MS: m/z 254
(Mϩ, 71%), 211 (20), 170 (47), 157 (100), 155 (30), 143 (49), 142
(33), 141 (25), 129 (37), 128 (22).
19 K.-L. Noble, H. Hopf and L. Ernst, Chem. Ber., 1984, 117, 455.
20 A. G. Myers and B. Zheng, J. Am. Chem. Soc., 1996, 118, 4492.
21 N. Koga and K. Morokuma, J. Am. Chem. Soc., 1990, 113, 1907.
22 P. G. Dopico and M. G. Finn, Tetrahedron, 1999, 55, 29.
23 GAUSSIAN98, Revision A.3, M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G.
Zakrzewski, J. A. Montgomery, Jr., R. E. Stratmann, J. C. Burant,
S. Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C.
Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi,
B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski,
G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick,
A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslowski,
J. V. Ortiz, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz,
I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith,
M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, C. Gonzalez,
M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong,
J. L. Andres, C. Gonzalez, M. Head-Gordon, E. S. Replogle and
J. A. Pople, Gaussian, Inc., Pittsburgh, PA, 1998.
24 MOLCAS Version 4.1. K. Andersson, M. R. A. Blomberg, M. P.
Fülscher, G. Karlström, R. Lindh, P.-Å. Malmqvist, P. Neogrády,
J. Olsen, B. O. Roos, A. J. Sadlej, M. Schütz, L. Seijo, L. Serrano-
Andrés, P. E. M. Siegbahn and P.-O. Widmark, Lund University,
Sweden (1997).
25 GAMESS: M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T.
Elbert, M. S. Gordon, J. H. Jensen, S. Koseki, N. Matsunaga, K. A.
Nguyen, S. J. Su, T. L. Windus, M. Dupuis and J. A. Montgomery,
J. Comput. Chem., 1993, 14, 1347.
Acknowledgements
We wish to thank Dr David Hrovat and Professor Weston T.
Borden for performing the CASPT2 calculations. We would
also like to thank Dr Cathy Lester for her assistance in obtain-
ing the 2-D NMR data. We gratefully acknowledge support for
this work by the Petroleum Research Fund through grant
number 31117-AC4.
References
1 A. G. Myers, Tetrahedron Lett., 1987, 28, 4493.
2 R. Nagata, H. Yamanaka, E. Okazaki and I. Saito, Tetrahedron
Lett., 1989, 30, 4995.
3 A. G. Myers, E. Y. Kuo and N. S. Finney, J. Am. Chem. Soc., 1989,
111, 8057.
4 A. G. Myers, P. S. Dragovich and E. Y. Kuo, J. Am. Chem. Soc.,
1992, 114, 9369.
5 A. G. Myers, P. J. Proteau and T. M. Handel, J. Am. Chem. Soc.,
1988, 110, 7212; A. G. Myers and P. J. Proteau, J. Am. Chem. Soc.,
1989, 111, 1146.
6 For representative examples: A. B. Padias and H. K. Hall, Jr., J. Org.
Chem., 1987, 52, 4536; M. B. Zimmt, C. Doubleday, Jr. and N. J.
Turro, Chem. Phys. Lett., 1987, 134, 549.
7 L. Salem and C. Rowland, Angew. Chem., Int. Ed. Engl., 1972, 11,
92.
8 B. Engels and M. Hanrath, J. Am. Chem. Soc., 1998, 120, 6356.
9 P. G. Wenthold, J. A. Paulino and R. R. Squires, J. Am. Chem. Soc.,
1991, 113, 1845; P. G. Wenthold and R. R. Squires, J. Am. Chem.
Soc., 1994, 116, 6961.
26 P. C. Hariharan and J. A. Pople, Theor. Chim. Acta, 1973, 28, 213.
27 M. W. Wong, M. J. Frisch and K. B. Wiberg, J. Am. Chem. Soc.,
1991, 113, 4776.
28 PCModel: Serena Software, Box 3076, Bloomington, IN 47402-
3076.
29 J. Suufert, E. Abraham, S. Raeppel and R. Brückner, Liebigs Ann.,
1996, 447.
30 A. T. Dann and W. J. Davies, J. Chem. Soc., 1929, 1050.
10 (a) P. G. Wenthold, S. G. Wierschke, J. J. Nash and R. R. Squires,
J. Am. Chem. Soc., 1994, 116, 7378; (b) P. G. Wenthold, S. G.
Paper 9/04369A
2298
J. Chem. Soc., Perkin Trans. 2, 1999, 2291–2298