the central σ bond and the two p-orbitals (i.e., æ1 ) æ4 ) 0°
and cos æ1 ) cos æ4 ) 1).1 Inspection of Table 1 reveals
that none of the four biradicals generated in the present study
has this ideal geometry, but that the biradical derived from
ketone 1 is close, with cos æ1 lying between 77% and 93%
of the ideal value and cos æ4 at 87% of maximum.18 This is
significant, as ketone 1 is the only one of the four ketones
to undergo type II cleavage. The biradicals derived from
ketones 3 and 4, in which overlap is very good on one side
and poor on the other, prefer cyclization to cleavage. The
behavior of the biradical derived from ketone 2 is unique.
In solution, and presumably in the anhydrous solid state as
well, it undergoes reverse hydrogen atom transfer to regener-
ate starting material. Only when the solid state photolysis is
carried out in a water suspension is a cyclization-derived
product observed. We discuss this result further in the next
paragraph, but note at this point that the failure of biradical
2 to cleave is consistent with its poor geometry for this
process, which is the worst of the four compounds studied.
Overall, therefore, we conclude that while perfect orbital
overlap is not required for cleavage, it should at least be
very good and should involve both p-orbitals. The data
clearly indicate that excellent overlap on one side and poor
overlap on the other does not lead to cleavage.
values well under 3.40 Å. We must also take into consid-
eration the directionality of the p-orbitals on C1 and C4, i.e.,
how well these orbitals overlap. Maximum overlap occurs
when the p-orbital on C1 is pointing directly at the top lobe
of the orbital on C4, i.e., when the C1 orbital is parallel to
the C2-C4 vector. The crystal structure data reveal that
ketone 1 deviates from this ideal geometry by 52-69° 18 and
that the corresponding deviation from ideality in ketones 2,
3, and 4 is 59°, 39°, and 32°, respectively. Thus, biradical
1, which has the most favorable geometry for cleavage, has
the least favorable geometry for cyclization, and it is hardly
surprising that cleavage dominates completely in this case.
Biradical 2 has a geometry that is imperfect for both cleavage
and cyclization, and this is probably a major factor in the
preference of this species for reverse hydrogen transfer in
solution and the anhydrous solid state. The high strain energy
of the cyclobutanol in this case (7) is undoubtedly a
contributing factor as well.
In summary, the photochemical reactivity displayed by the
benzoyladamantane system is exquisitely sensitive to the
geometry of the intermediate 1,4-hydroxybiradicals, with
cleavage predominating when the radical-containing orbitals
overlap poorly with one another and well with the central
carbon-carbon bond (ketone 1). In contrast, Yang photo-
cyclization takes over for biradicals that are reasonably well
aligned for closure and poorly aligned for cleavage (ketones
3 and 4), and in perhaps the most interesting case of all,
reverse hydrogen transfer dominates for the biradical in
which both cleavage and cyclization are geometrically
disadvantaged (ketone 2).
Having dealt with the structural features that favor 1,4-
hydroxybiradical cleavage, it is equally important to focus
on the geometric factors that favor cyclization. Intuitively,
it makes sense that Yang photocyclization will be favored
when the radical-containing carbon atoms C1 and C4 are close
to one another, probably e 3.40 Å, which is the sum of the
van der Waals radii for two carbon atoms. Table 1 reveals
that the 1,4-biradicals generated in this study all have D
Acknowledgment is made to the donors of the Petroleum
Research Fund, administered by the American Chemical
Society, for partial support of this research. Financial support
by the Natural Sciences and Engineering Research Council
of Canada is also gratefully acknowledged.
(18) This analysis is based on identifying which of the two possible
γ-hydrogen atoms is abstracted. This is clear in the case of ketones 2-4,
where one hydrogen is much closer to the ketone oxygen than the other,
but for compound 1, the two hydrogens are nearly equidistant (2.33 Å as
opposed to 2.36 Å). Because it is not clear which hydrogen is abstracted,
both sets of data are presented.
OL9912377
80
Org. Lett., Vol. 2, No. 1, 2000