Cyclization and Cleavage of Preoxetane Diradicals
A R T I C L E S
competitive simultaneous photocycloaddition between cis-1b
and BP, for which the results of the cis-1b photocycloaddition
were utilized (Table 1, systems 5 and 5′). Thus, only a moderate
trans diastereoselectivity of 35:65 is noted for 2b and little if
any selectivity (54:46) for 2b′ at -95 °C (Table 1, systems 6
and 6′, entry 1). With rising temperature, however, for the
regioisomer 2b (system 6) the diastereoselectivity increases
significantly to <2:98 in favor of the trans product, and only
under considerable thermal stress (T > 80 °C) are traces (ca.
5%) of the cis-2b oxetane found for trans-1b (Table 1, system
6, entries 11 and 12). For 2b′ (system 6′), surprisingly, the trans
diastereoselecitivity increases at first to a maximum value of
4:96 at -40 °C (entry 4), and then again decreases to 51:49 at
+110 °C (entry 12).
This complex temperature dependence of the diastereoselec-
tivity data (Table 1) in the Paterno`-Bu¨chi photocycloaddition
for the cis- and trans-configured cyclooctenes 1a and 1b with
benzophenone (BP) and benzoquinone (BQ) shall now be
rationalized in terms of a consistent mechanism, which encom-
passes all these experimental facts. Because the photochemical
behavior is quite similar for BQ and BP, our mechanistic
analysis will focus mainly on the latter.
of such triplet diradicals,19 the orbital-orientation rule defined
by Salem and Rowland applies,20 which states that a perpen-
dicular geometry of the 2p orbitals at the radical sites is optimal
for spin-orbit coupling.
We shall first consider the cis-configured (Scheme 2) and
the trans-configured (Scheme 3) cyclooctenes 1a,b separately
to rationalize the results mechanistically in terms of the pertinent
conformational changes at the reaction center,21 and subse-
quently compare the cis and trans substrates to point out
similarities and differences in the form of general trends (see
Mechanistic Comparison). Although in these schemes the olefin
configurations are dealt with separately, the olefin substitution
(R ) H, Me) is presented in an integrated manner. This allows
one to grasp better the complexities of the temperature-
dependent substituent effects on the triplet-diradical conforma-
tions derived from the cyclooctene diastereomers. Each scheme
is layed out in two major halves: The top half illustrates the
attack of the triplet-excited carbonyl partner at the C-2 carbon
atom of the cyclooctene, which leads to the major oxetane
regioisomer when R1 is different from H; the bottom half
features the corresponding C-1 attack to afford the minor oxetane
regioisomer. Of course, for the unsubstituted cyclooctenes (R1
) H) the two halves coincide in Schemes 2 and 3, since no
regioisomers are possible. The left-hand and the right-hand sides
of each scheme portray the important role of the syn and the
anti attacks (Figure 1) and their influence on the diastereose-
lectivity of the present [2 + 2] photocycloaddition.
The detailed mechanism in terms of the competitive syn and
anti approaches for the cycloadditions of the triplet-excited
benzophenone (BP) to the cis-configured cyclooctenes cis-1a,b
is shown in Scheme 2, and applies also for benzoquinone (BQ).
In the C-2 attack (Scheme 2, top) from the syn (left) as well as
the anti (right) side, the initial conformations of the resulting
triplet-diradical conformers A and B reflect the cis configuration
of the olefin 1. Therefore, from these conformations the cis-
oxetanes are produced after intersystem crossing (ISC) to the
singlet state and subsequent ring closure. However, on closer
inspection of the intervening triplet-diradical conformations A
and B in Scheme 2, unfavorable gauche interactions become
apparent due to the initially cis-configured eight-membered ring,
which may induce CC-bond rotation to the C and D conformers
to avoid this steric strain. Whereas for the syn approach also
Mechanistic Analysis
The results in Table 1 clearly demonstrate that the configu-
ration (cis versus trans) and substitution (R1 ) H versus R1 )
Me) of the cyclooctenes 1a,b are decisive for the stereocontrol
in the [2 + 2] photocycloaddition with the symmetrical carbonyl
partners BP and BQ. In particular, the photocycloaddition of
the two unsubstituted cyclooctenes cis-1a and trans-1a with the
ketones BP and BQ (Table 1, systems 1-4), as well as the
methyl-substituted cis-1b with BP (Table 1, systems 5 and 5′),
follow a similar temperature dependence in the diastereoselective
oxetane formation. At low temperature (below -80 °C), the
diastereoselectivity is very high with preservation of the initial
cis or trans configuration of the substrate, and a gradual rise of
the temperature leads to a significant diminution (systems 2, 5,
and 5′) or even inversion (systems 1 and 3) of the diastereo-
selectivity. In contrast, the trans-1b cyclooctene (systems 6 and
6′) manifests its special status in that it displays an unprec-
edented temperature-dependent stereochemical behavior, that
much cis-oxetane product (extensive loss of configuration) is
observed already at very low temperature (-95 °C). As the
temperature is raised, the amount of trans-oxetanes increases
and becomes essentially the exclusive cycloadduct for both
regioisomers 2b and 2b′. Furthermore, the minor regioisomer
2b′ attains a maximum of trans-oxetane 2b′ at about -40 °C.
To account for the temperature dependence of the diastereo-
selectivity observed in these Paterno`-Bu¨chi reactions, we
propose the competitive syn and anti approaches of the triplet-
excited carbonyl compounds toward the olefinic substrate to
afford the respective intermediary preoxetane triplet diradicals
(Figure 1).18 Their initial conformations are defined by the syn
versus anti type of approach. The subsequent transformations
of these diradicals through competitive conformational changes
and intersystem crossing (ISC), followed by cyclization or
cleavage, allow for the rationalization of the trends in the
diastereoselectivity. For the intersystem-crossing (ISC) process
(19) Griesbeck, A. G.; Mauder, H.; Stadtmu¨ller, S. Acc. Chem. Res. 1994, 27,
70.
(20) Salem, L.; Rowland, C. Angew. Chem., Int. Ed. Engl. 1972, 11, 92.
(21) The following clarifications need to be made about our conformational
analysis: We have conducted PM3-UHF calculations on the triplet-diradical
intermediates, but the flexibility of the eight-membered ring allows the
CH2 units of the cyclooctyl ring to be arranged in numerous essentially
isoenergetic (all within less than 0.5 kcal/mol) conformations such that no
definite energy minima may be reliably computed. However, this also
implies that these conformational changes do not contribute significantly
to the energetics of the overall cycloaddition process and, expectedly, do
not provide any mechanistically useful information. In view of this
difficulty, by inspection of molecular models, we have in our analysis
concentrated on the conformational details in the direct vicinity of the
reaction center, defined by the two olefinic carbon atoms involved in the
cycloaddition, rather than on the periphery of the cyclooctyl ring system.
Our intentions are more clearly illustrated in the detailed structures of Figure
2, in which the remaining six CH2 units of the eight-membered ring have
been intentionally left out, and attention has been focused on the
conformational features with their steric implications of the two CH2
substituents at the reaction center. For simplicity, since Schemes 2 and 3
are already quite complex, we have drawn the remaining cyclooctyl ring
as planar structures with no conformational preferences. On the basis of
our computational results and inspection of models, we contend that the
conformational changes of the remaining cyclooctyl ring do not significantly
influence our mechanistic conclusions.
(18) (a) Freilich, S. C.; Peters, K. S. J. Am. Chem. Soc. 1985, 107, 3819. (b)
Freilich, S. C.; Peters, K. S. J. Am. Chem. Soc. 1981, 103, 6255.
9
J. AM. CHEM. SOC. VOL. 124, NO. 14, 2002 3603
Adam, Waldemar
