Scheme 2. This very large range of reactivities means
that the less reactive compounds would likely be consid-
ered to be unreactive unless one were specifically looking
for phototranspositions.
The low efficiency of the leakage processes indicates that
the quantum yield of phototransposition for the methyl-
benzotrifluorides is 5-10 times more efficient than that
reported for m-xylene (0.01).9
The fact that the major photochemical process is
conversion within the triads shown in Scheme 2 indicates
that the mechanism does indeed proceed through two
triads of bicyclic diradicals that are trifluoromethyl-
substituted at C6. One triad 9a -c allows interconversion
of 8-2,6, 8-2,3, and 8-3,4; the other, 10a -c, allows
interconversion of 8-3,5, 8-2,5, and 8-2,4. The same
mechanistic proposal has been advanced by us previously
for the dimethylbenzonitriles4 where the cyano-substi-
tuted carbon is at C6 (as in Scheme 1 for the mono-
methylbenzonitriles) and in the gas-phase phototrans-
position reactions of the dimethylpyridines where the
nitrogen of the pyridine served the role of the migratory
atom.6
We have examined the phototransposition reactions of
the xylenes in acetonitrile as solvent3 in more detail than
previous reports.7,9 The relative reactivity of the three
isomers is 1:1.4:0.3 for ortho/meta/para and the primary
product ratios are meta/para ) 11:1 for ortho, ortho/para
) 6.0:1 for meta and meta/ortho > 25:1 for para. Thus
the most reactive isomer is the meta one with the ortho
one reacting somewhat more slowly. Therefore, as ex-
pected and indicated in Scheme 2, three of the leakage
processes involve reaction of a meta xylene isomer, i.e.,
8-2,6, 8-3,5, and 8-2,4; the fourth involves reaction of an
ortho xylene derivative, namely 8-3,4.
The efficiency of these reactions must be determined
by two quite different factors. The first is how efficiently
the excited-state undergoes the required geometric changes
to convert it to the bicyclic diradicals 9 and 10. If the
barrier for this process is high, then other nonproductive
excited-state processes (fluorescence, internal conversion,
intersystem crossing) likely dominate. The second re-
quired step is migration of the trifluoromethyl-substi-
tuted carbon around the periphery of the other five
allowing interconversion of the three isomeric intermedi-
ates in the two sets 9 and 10. If the barrier to the latter
steps is high then, even if the intermediates 9 and 10
are formed, no reaction will be observed. The process will
just be one mode of internal conversion. With the limited
evidence available, determining whether one or the other
of these two possibilities is the more important one is
not posssible. However, a tentative proposal is possible
on the basis of the results for photolysis of the isomers
of 8 in TFE.
All six of the isomers react in TFE to give products
that clearly result from alcohol addition (GC/MS, molec-
ular ion m/z ) 274). No phototranspositions were ob-
served. All gave complex mixtures, with 8-2,3 being
typical with four major products accounting for around
65% of the yield and more than a dozen others constitut-
ing the rest. Presumably, these addition products are
similar to those shown in Scheme 1 for the methylben-
zonitriles but we were unable to characterize them
further. Even simply evaporating the solvent gave samples
whose 1H NMR spectra showed no assignable signals but
rather broad bands extending over ranges of δ 1-3 and
δ 4.5-4.0 suggesting oligomers or polymers. We had
previously observed the same difficulties characterizing
TFE addition products for the three monomethylbenzo-
trifluorides and trifluoromethylbenzene itself.3
However, the photoreactions in TFE did provide evi-
dence on their relative reactivity. As in the phototrans-
position reactions, the least reactive isomers were again
8-3,4 and 8-2,5. This suggests that formation of the
bicyclic diradical intermediates 9 or 10 is slow in both
solvents. After this formation, they either isomerize
(acetonitrile) or are trapped (TFE).
Because of the symmetry of the intermediate/transition
state for the migration of two of the intermediates,
namely 9c and 10c, some phototranspositions will be
invisible. For instance, both 8-3,4 and 8-2,5 can undergo
transposition reactions that switch a trifluoromethyl-
substituted carbon with a neighboring hydrogen-sub-
situted carbon in a process that does not result in
formation of a new isomer. Therefore, the efficiency of
reactivity for the two compounds could be greatly under-
estimated.
As indicated in Table 1, “leakage” occurs from one triad
to the other. This is shown graphically in Figure 1 where
formation of 8-2,3 from 8-2,4 occurs with a yield of 18%
at low conversion. This is the highest yield of the leakage
process, the others being under 10%. Because 8-2,3 is the
most reactive isomer of the set of six, it very rapidly
undergoes its own photochemistry so that 8-3,4 is formed.
These leakage pathways, shown in Scheme 2, all involve
phototransposition of a methyl-substituted carbon. This
is not unexpected because the substrates 8 can be
regarded as trifluoromethyl-substituted xylenes and
historically, the xylenes were among the first substrates
where phototranspositions were observed.7 In fact, this
discovery lead to the proposal that substituted benz-
valenes were important intermediates in aromatic pho-
totransposition reactions and eventually to the photo-
chemical preparations of benzvalene from benzene itself.8
Th e P h otoch em istr y of 4-Meth ylben zotr iflu or id e-
d 2 (5-d 2). This compound was synthesized starting from
(6) Pavlik, J . W.; Kebede, N.; Thompson, M.; Day, A. C.; Barltrop,
J . A. J . Am. Chem. Soc. 1999, 121, 5666-5673.
(7) Wiltzbach, K. E.; Kaplan, L. J . Am. Chem. Soc. 1964, 86, 2307-
2308.
(8) Wilzbach, K. E.; Ritscher, J . S.; Kaplan, L. J . Am. Chem. Soc.
1967, 89, 1031-1032.
(9) Anderson, D. J . Phys. Chem. 1970, 74, 1686-1690.
9486 J . Org. Chem., Vol. 67, No. 26, 2002