tion of 1e, obtained from 1c by standard methods, for 2 h,
under the same conditions used for 1b, yielded 5c (6%) and
recovered starting material 1e (76%), exclusively. The
corresponding benzocycloheptene derivative was not detected
in the product mixture. Cyclopentenes 5a and 5c are the
products expected to result from conventional vinylcyclo-
propane rearrangement.1 However, the formation of 4a and
4c is surprising since, as far as we are aware, the photo-
chemical ring expansion of vinylcyclopropanes to benzocy-
cloheptenes has not been previously observed. The formation
this process.15 This SET process forms zwitterionic biradical
12 that then undergoes ring opening to produce dipolar
intermediate 13. Electrophilic addition to the aromatic ring
in 13 is quite plausible; this yields intermediate 11, the
precursor of 4c. In accord with this mechanistic proposal,
the formation of the five-membered compound 5a and the
seven-membered derivative 4c in the irradiation of 1b could
be due to a competition between the biradical path, yielding
5a and the SET route that affords 4c.
To test this mechanistic hypothesis, the ester 14 was
synthesized16 and irradiated under the same conditions used
for 1a. The presence of p-methoxy groups in the phenyl rings
should decrease the ionization potential of the alkene unit
and favor intervention of the intramolecular SET route.
Triplet-sensitized irradiation of 14 for 90 min yielded the
benzocycloheptene 15 (19%) and recovered starting material
(56%). The isomeric cyclopentene was not detected as a
product of the photoreaction of 14, a fact which can be
explained by the increased efficiency of the SET path leading
to the cycloheptene.
Scheme 2
At this point it was considered necessary to obtain
additional evidence that would demonstrate that in the
absence of electron-acceptor groups at C1, the ring expansion
to cycloheptenes would not take place. Thus, cyclopropane
1f17 was irradiated under the same conditions used for 1b,
for 5 h. In this instance, the only photoproduct obtained is
the cyclopentene 5d (10%).
of 4a, 4c, 5a, and 5c can be interpreted by the operation of
a conventional biradical mechanism (Scheme 2 for 1b). Thus,
excitation of 1b to its triplet excited state is followed by
homolytic rupture of the C1-C2 cyclopropane bond to
produce biradical 10. Radical recombination of 10 provides
11. Compound 4c results from the alternative cyclization
mode involving addition of the primary radical to one of
the phenyl rings affording the seven-membered-ring inter-
mediate 11. Hydrogen migration within 11 yields benzocy-
cloheptene 4c. Although this mechanism reasonably explains
the products formed in these reactions, it leaves in question
why the biradical intermediate 10 undergoes cyclization to
cycloheptenes in the reactions of 1a and 1b while this path
is not followed in reaction of 1e.
At this point, a possible alternative mechanism for the ring
expansion to benzocycloheptenes was considered. The ac-
etoxyimino and methoxycarbonyl groups present in 1a and
1b, respectively, could act as good electron acceptors for
single-electron transfer (SET) from the triplet excited state
of the diphenylvinyl chromophore (Scheme 3 for 1b).
Previous studies carried out by us support the feasibility of
The study was extended to cyclopropanes 1g and 1h. The
presence of a methyl in 1g and a phenyl in 1h at C2 of the
cyclopropane ring should facilitate the ring opening of the
cyclopropane. These two compounds were synthesized by a
route analogous to that used to prepare 1a, starting from 3b
and 3c, respectively.18 Irradiation of 1g, as a 7:3 mixture of
(Z:E) diastereoisomers,19 for 50 min, under the same condi-
tions used for 1a, followed by thermal elimination of AcOH,
yielded cycloheptene 4d (17%) and the nitrile 1i (53%) as a
3:1 mixture of (Z:E) diastereoisomers. This result shows that
(10) Spectral Data for 4c: 1H NMR (300 MHz, CDCl3) δ 2.42 (m,
1H), 2.61 (m, 3H), 2.99 (m, 1H), 3.69 (s, 3H), 6.61 (d, J ) 6.6 Hz, 1H),
6.98-7.82 (m, 9H); 13C NMR (63 MHz, CDCl3) δ 31.5, 39.7, 42.8, 59.7,
125.5-142.6, 175.3.
(11) Spectral Data for 5b: 1H NMR (300 MHz, CDCl3) δ 1.63 (s, 1
H), 2.48 (m, 2 H), 2.62 (m, 2 H), 4.26 (s, 2 H), 6.08 (s, 1 H), 7.17-7.29
(m, 10 H); 13C NMR (63 MHz, CDCl3) δ 32.2, 40.0, 61.3, 62.3, 126.0-
148.8.
(12) These yields are calculated from the mixture of 5a and 1b.
(13) For experimental details on the synthesis of 4c see the Supporting
Information.
(14) Hart, H.; Dunkenblum, E. J. Am. Chem. Soc. 1978, 100, 5141.
(15) (a) Armesto, D.; Ramos, A.; Ortiz, M. J.; Manchen˜o, M. J.; Mayoral,
E. P. Recl. TraV. Chim. Pays-Bas 1995, 114, 514. (b) Armesto, D.; Austin,
M. A.; Griffiths, O. J.; Horspool, W. M.; Carpintero, M. Chem. Commun.
1996, 2715.
(16) The synthesis is analogous to that used for 1b with the only
difference that diethyl bis(p-methoxyphenyl)methylphosphonate is used in
the Horner-Emmons step.
Scheme 3
(17) Zimmerman, H. E.; Samuel, C. J. J. Am. Chem. Soc. 1975, 97, 4025.
(18) (a) Compounds 3b and 3c were obtained by DIBALH reduction
(ref 18b) of the previously reported diethyl 2-methylcyclopropane-1,1-
dicarboxylate and diethyl 2-phenylcyclopropane-1,1-dicarboxylate, respec-
tively (ref 18c). (b) Davis, C. R.; Swenson, D. C.; Burton, D. J. J. Org.
Chem. 1993, 58, 6843. (c) Landor, S. R.; Punja, N. J. Chem. Soc. (C) 1967,
2495.
(19) The configurations of compounds 1g and 1h were established by
means of NOE difference measurements.
Org. Lett., Vol. 2, No. 2, 2000
185