SCHEME 3. Evid en ce Disfa vor in g a n Op en -Ch a in
Ra d ica l (P a th A)
SCHEME 4. Discr im in a tin g Ra d ica l Br om in e
Atom (P a th B) or Br om id e An ion Med ia ted
Op en in g (P a th C)
to give an open chain radical precedes propagation with
NBS (or Br2), the intermediate radical 16 would be
expected to cyclize onto the tethered radicalophile in
6-endo-trig or even 8-endo-trig fashion, leading to 17 (or
isomers). Molecular modeling studies on the fragmented
acyclic radical intermediate provide no reasoning why
such a cyclization should not occur as either of the radical
centers are in close proximity to the â-position of the
acrylate. On the other hand, if radical fragmentation does
not occur prior to bromine atom or anion attack on the
respective intermediate benzylic radical or cation (paths
B and C), acyclic substitution products are to be expected.
The reaction of substrate 14 proceeded without incident
to give a single product readily identified as the allylic
bromide 15 (55.2%, 25.3% recovered 14, 80.5% mass
balance). No evidence for any cyclized products was
observed; trace polar impurities observed were likely
produced through slow hydrolysis of 15. This result
provides strong evidence that under these conditions the
initial benzylic radical does not immediately fragment;
or at least such fragmentation is considerably slower than
the major opening process. The major reaction pathway
therefore proceeds through nucleophilic bromine atom or
anion opening of the cyclic radical or cation (paths B or
C).
Reaction pathways B and C are mechanistically very
similar and difficult to unravel, the major difference being
whether a bromine atom (propagation with NBS) or
bromide anion attacks the activated intermediate. We
devised the following experiment to illuminate the subtle
difference in reactivity expected in this process for a
neutral radical or a polar anion attack on the intermedi-
ate. It is well-known that radicals add preferentially to
the â-position of enones;7 however, since they are not
subjected to the same electronic constraints as the
nucleophilic addition of an anion, they may also add to
the R-position of an enone in certain cases where this is
mechanistically feasible.9 Test substrate 18 was prepared
(see the Supporting Information) and subjected to the
standard reaction, Scheme 4. Given that the propensity
for SN2′-like opening has already been demonstrated,
epimeric bromides 21 would be expected (attack at C-2′)
if the reaction proceeds through path B. In contrast, if
benzylic bromination occurs first following path C and
to be of synthetic value,7 we decided to clarify aspects of
the mechanism of this reaction to illuminate its overall
scope.8
We chose to develop mechanistic probes for the reaction
employing substrates derived from the 2,4-benzilidene
derivative of D-erythrose, themselves readily available
from D-glucose. We first determined that the regioselec-
tivity of the fragmentation was completely reversed by
the simple expedient of attaching an olefin allylic to the
more hindered position on the acetal, Scheme 3. For
example, when 12 was reacted under standard conditions
(NBS, BPO(cat.), PhCl, 70 °C, 3 h), the allylic bromide
13 was isolated as the single (Z)-isomer. Careful workup
and purification gave 13 in 64.2% isolated yield along
with 21.5% recovered 12 allowing for 85.7% overall mass
balance. No other products were observed; the balance
of the material is likely lost to polymerization of starting
material and/or product, which is not unexpected. To our
knowledge, this is the first time that the regiochemistry
of the fragmentation has been altered in this fashion. The
reversal of regioselectivity in the opening step mediated
by the olefin is likely due to a lowering of the energy of
the now allylic σ* orbital of the O3-C4 bond on the
intermediate radical or cation. Product 13 appears to be
the result of a concerted bromine radical or bromine
anion addition at C2′ on the intermediate benzylic radical
or cation respectively in SN2′-like fashion. The fact that
a single olefin is obtained is evidence against an open
chain allylic radical (analogous to 7, path A) through
which both allylic isomers as well as possible E and Z
configurational isomers would be expected. This result
does not discriminate between the other two substitution
pathways outlined (Scheme 2), however.
More definitive evidence against path A was obtained
through the second experiment (14 to 15) shown in
Scheme 3. The acryloyl-substituted olefin 14 was pre-
pared and subjected to our standard fragmentation
protocol. If fragmentation of the initial benzylic radical
(7) (a) Hands, S.; Pattenden, G. Contemp. Org. Synth. 1997, 196-
215. (b) Curran, D. P. Aldrichim. Acta 2000, 33, 104-110.
(8) Alternative methods for both oxidative and Lewis acid mediated
fragmentation of benzilidene acetals have been reported recently;
see: (a) Chen, Y.; Wang, P. G. Tetrahedron Lett. 2001, 42, 4955-4958.
(b) Harada, T.; Sekiguchi, K.; Nakamura, T.; Suzuki, J .; Oku, A. Org.
Lett. 2001, 3, 3309-3312.
(9) For an example of radical addition at the R-position of an R,â-
unsaturated ester in a cyclic carbohydrate dervivative, see: Araki, Y.;
Endo, T.; Arai, Y.; Tanji, M.; Ishido, Y. Tetrahedron Lett. 1989, 30,
2829-2832.
564 J . Org. Chem., Vol. 69, No. 2, 2004