2] cycloaddition regioisomer 6 (3.5:1). It was found that 5
was actually derived from the initially formed [5 + 2]
regioisomer 7, as irradiation of 4 for just 10 min gave a
mixture of 5-7. Subsequent irradiation of isolated regio-
isomer 7 gave 5 in good yield (Scheme 2).
maleimides studied. For example, 4 undergoes complete [5
+ 2] cycloaddition in under 10 min, whereas 1 (R ) Me)
requires 2.5 h of irradiation for complete conversion. This
may be explained by the fact that the key absorption for 1
(R ) Me) is in general a weak band observed in the UV
spectra at around 300 nm, which is probably excited by the
relatively weak 302 and 313 nm emissions from the medium-
pressure mercury source used in this study. The same absorp-
tion in 4 is shifted to a maximum at around 330 nm with a
continuum >370 nm. As this lies close to the most powerful
Hg lamp emission at 365 nm, it is therefore likely that the
excited state of 4 is populated much more efficiently than
1. It is also likely that the weak emission at 334 nm
contributes.
Scheme 2. Methoxymaleimide [5 + 2]/Norrish II Cascadea
We then set out to explore the scope of this reaction by
synthesizing a range of alkoxy maleimide derivatives and
investigating their photocycloaddition reactions (Table 1).
The photosubstrates were prepared by either Mitsunobu
alkylation of monomethoxy maleimide7 or by alkoxylation
of the N-alkylated dichloromaleimides.8 Generally, the
various maleimides displayed the same trend with oxetane
formation predominating. Excellent stereocontrol was ob-
served in the formation of the cycloadducts where up to four
contiguous centers could be formed in a single irradiation
(e.g., entry 3).
The five-membered alkene substrate 10 (entry 7) proved
to be very interesting, as no oxetane product was isolated.
With care, the [5 + 2] enol 11 could be isolated in up to
60% yield. It is likely that the ketone carbonyl in the initially
formed [5 + 2] cycloadduct 12 undergoes photoenolization
faster than Norrish-Yang oxetane formation.9 This in
essence protects the ketone carbonyl from undergoing further
photoreactions. This photoenolization may occur through a
diradical intermediate such as 13, where 1,5-hydrogen atom
abstraction leads to enol formation instead of the oxetane
(Scheme 3).
a Conditions: (a) hν, Pyrex, MeCN, 40 min; (b) hν, Pyrex,
MeCN, 10 min; (c) hν, Pyrex, MeCN, 2 h.
It is most likely that 5 is formed from 7 by the Norrish-
Yang sequence described in Scheme 2. Further excitation
of 7 results in the formation of 8, which undergoes 1,5-hy-
drogen atom abstraction (Norrish II) from the methoxy group
to give the diradical 9, which then undergoes recombination
to 5. A related reaction was observed by Feigenbaum5 for
methoxy-substituted cholestenones. Intermolecular [2 + 2]
photocycloaddition onto methoxy-substituted chromones, fol-
lowed by oxetane formation, has also been reported by Mal
and Venkateswaran.6 There are several noteworthy features
concerning this overall sequence. First, recombination of 9
is highly stereoselective (>99% de), presumably a result of
the methylene radical approaching the R-hydroxy radical
from the least hindered, convex face of the azulene ring sys-
tem. The methoxy group in 4 is clearly exerting a powerful
directing effect on the initial [5 + 2] cycloaddition, resulting
in the predominance of regioisomer 7 and ultimately 5. In
the case of 4, cycloaddition could proceed via the two pos-
sible amide bonds “a” and “b”. Cycloaddition via bond a
leads to the major regioisomer 7, whereas reaction through
bond b leads to the minor isomer 6. As carbonyl b can be
considered a vinylogous carbamate, then the cycloaddition
selectivity is likely to be a result of absorption differences
between these two carbonyls, although detailed modeling is
likely to be required to explain these differences. Mazzocchi
also observed regioselective alkene insertion reactions with
methoxy phthalimide, although it is interesting to note that
in a vinylogous sense, the regioselectivity obtained was oppo-
site that of the methoxy maleimides in this study.1d,e The
initial [5 + 2] reaction of 4 is much faster than previous
Scheme 3. Proposed Mechanism for Photoenolization
When the six-membered alkene substrate 14 was irradi-
ated, only the oxetane 15 and the regioisomer 16 were
obtained, and no sign of the enol was observed. However,
(5) Feigenbaum, A.; Pete, J. P. Tetrahedron Lett. 1972, 13, 2767.
(6) (a) Mal, J.; Venkateswaran, R. V. J. Org. Chem. 1998, 63, 3855. (b)
See also: Wender, P. A.; Rawlins, D. B. Tetrahedron 1992, 48, 7033.
(7) (a) Nicolaus, R. A.; Nicoletti, R. Rend. Accad. Sci. Fis. Mat., Naples
1959, 26, 148. (b) Walker, M. A. J. Org. Chem. 1995, 60, 5352.
(8) Lynch, D. M.; Crovetti, A. J. J. Heterocycl. Chem. 1972, 9, 1027.
(9) For reference to photoenolization during Norrish-Yang cyclization,
see: Wagner, P. J. In CRC Handbook of Organic Photochemistry and
Photobiology, 2nd ed.; CRC Press: Boca Raton, FL, 2004, Chapter 58, pp
1-70.
1482
Org. Lett., Vol. 6, No. 9, 2004