that postulated above, but the chemistry results in photore-
duction. No case of displacement of a nucleofuge on the
aromatic ring from such an exciplex has been reported. The
triplet exciplexes dissociate rapidly and nongeminate radical
chemistry results. Moreover, aminium radical intermediates
undergo fast proton or hydrogen atom transfers that move
the electron hole from nitrogen to the R carbon.16 This would
prevent bonding at the amine nitrogen that occurs in the
photosubstitutions.
Our studies of base catalysis in meta Smiles photorear-
rangements17 showed that meta σ complexes live for tens to
hundreds of nanoseconds. This suggested that if meta σ com-
plexes occurred in the para photosubstitutions by amines,
they could be trapped by rapid proton-transfer reactions. Pro-
ton removal from the nitrogen of intramolecular photoadducts
that do not have a nucleofuge at the attack site results in
stable dihydrobenzene products.18 It seemed plausible, more-
over, that if meta σ complexes were intermediates in the
intermolecular photosubstutions, they could convert rapidly
to para σ complexes by an unprecedented but simple sig-
matropic rearrangement. In this case, we predicted for the
reaction of 4-nitroanisole with amines that base-dependent
photoformation of stable dihydrobenzene adducts would
compete with formation of the para photosubstitution prod-
ucts.
3.55, 3H, s; δ 2.65, 6H, s; methylene H not found. Product
5 was also identified by its NMR spectrum and by COSY,
which showed the required couplings: δ 7.07, 1H, d (J )
10.1 cps); δ 6.14, 1H, d (J ) 10.1 cps); δ 2.70, 6H, s;
methylene H not found. The structure of 5 was confirmed
by the coincidence of its spectrum with that of an analogous
adduct18 having an ethanolamino group in place of the
dimethylamino group of 5. Additional evidence for the
structures of 4 and 5 was obtained by extraction experiments.
When the photolyzed reaction solution was extracted with
CDCl3, all nonionic products except methanol were removed
to the CDCl3 phase, but the extraction did not change the
amounts of 4 and 5 in the aqueous phase. This confirmed
that 4 and 5 are ions.
The data in Scheme 1 show that 4 converts thermally to
5 plus the required 1 equiv of methanol. That the four NMR
signals assigned to 4 disappear together and the signals
assigned to 5 and methanol arise concurrently is strong
evidence for the assignments. The conversion of 4 to 5 plus
methanol was faster at pH 10.7 than at pH 11.6; at pH 10.7,
only 5 and not 4 could be detected after the photolysis,
meaning that the thermal hydrolysis was complete in the
photolysis time at that pH. At pH 12, the conversion was
not complete in 72 h, and at pH 12.5, 4 was stable.
The percent yield of photosubstitution product 2 (ca. 53%)
by direct NMR detection in the reaction of 1 in DMSO-
D2O containing 0.030 M dimethylamine (ca. 0.004 M [OH-])
agrees roughly with the yields of photosubstitution products
isolated from preparative reactions of 1 and aliphatic amines
in water.19 We found, however, that the ratio of 2 to 4 varied
with the hydroxide ion concentration in the photolysis mix-
ture as shown in Table 1. The data show that the yield of 4
increases with increasing hydroxide ion at the expense of 2.
Irradiation of 4-nitroanisole (1, 0.0060 M) in oxygen-free
40% DMSO-d6/D2O (v/v) containing 0.030 M dimethylamine
for 75 min at 0 °C in a borosilicate NMR tube with 300 nm
broadband light filtered through K2CrO4 solution caused 29%
of the starting material to disappear. The products and percent
yields are as shown in Scheme 1. Products 2, 3, 6, and
Scheme 1. Photoproduct Percent Yields from 4-Nitroanisole
and Dimethylamine
Table 1. Dependence on Hydroxide Ion Concentration of
Relative Yields of Photosubstitution Product vs Photoadduct
methanol were confirmed by enhancement of their NMR
signals with authentic samples. Product 4 was identified by
its NMR spectrum and by COSY, which showed the required
couplings: (shifts relative to terephthalate anion, δ 7.86) δ
6.57, 1H, d (J ) 10.3 cps); δ 5.72, 1H, d (J ) 10.3 cps); δ
The nitronate ions detected stem from meta attack by the
amine. If the SN(ET)Ar* mechanism operated as suggested
for dimethylamine,12 geminate radical recombination would
be expected at both the ortho and para positions. Ortho
(15) (a) Cohen, S. G.; Parola, A.; Parsons, G. H., Jr. Chem. ReV. 1973,
73, 141. (b) Inbar, S.; Linschitz, H.; Cohen, S. G. J. Am. Chem. Soc. 1981,
103, 1048.
(16) Chow, Y. L. React. Intermed. (Plenum) 1980, 1, 151.
(17) Wubbels, G. G.; Cotter, W. D.; Sanders, H.; Pope, C. J. Org. Chem.
1995, 60, 2960.
(18) Wubbels, G. G.; Winitz, S.; Whitaker, C. J. Org. Chem. 1990, 55,
631.
(19) Kronenberg, M. E.; van der Heyden, A.; Havinga, E. Recl. TraV.
Chim. Pays-Bas 1966, 85, 56.
1452
Org. Lett., Vol. 8, No. 7, 2006