S. M. Bonesi et al. / Tetrahedron Letters 55 (2014) 2932–2935
2933
amount of highly polar products. In order to understand the mech-
anism of the oxidative photocyclization, the photoreaction of Ph3N
in the presence of potential oxygen acceptor was investigated. Sul-
fides and phosphines were chosen as suitable probes (see
Scheme 3). The results are reported in Tables 1 and 2 and can be
summarized as follows. Sulfides were co-oxidized to sulfoxides at
a rate dependent on structure and conditions. The rate of the tri-
phenylamine oxidative cyclization increased to a various extent
(10–250%) and the rate ratio of sulfoxide versus carbazole forma-
tion was P1 (from ca. 1.1 to 2.5). In contrast, irradiation of triphen-
ylamine in the presence of either triphenylphosphine or tris-(2-
methylphenyl)phosphine led to the corresponding phosphine oxi-
des, but at the same time the formation of N-phenylcarbazole (1)
was completely inhibited. A precise comparison between the oxi-
dation rate of sulfides and phosphines was not feasible, since dif-
ferent irradiation conditions had to be adopted to avoid auto-
oxidation of phosphines, but it was clear that the photooxidation
of the latter substrates was at least five times faster than either
oxidation of sulfides or oxidative photooxidation of TPA alone.
These results pointed to the involvement of different intermediates
in the reaction of the two traps.
H
H
a
d
3Ph3N
.
.
N
Ph
3DHC
b
hν
H
H
C-
N+
Ph3N
DHC0
Ph
c
N
Ph
NPC
H4-
NPC
Scheme 1. Photochemical cyclization of triphenylamine (Ph3N).
below a combined experimental and computational investigation
about the photocyclization of triphenylamine in different solvents
and in the presence of oxidation-liable substrates.
Energy transfer to give singlet oxygen is energetically viable
only by quenching of 3Ph3N (see above)14 and not from 34, due
to the low energy gap of the latter from ground state 14, much be-
low that required for 1O2 formation (ca. 22 kcal molÀ1).15a If these
were the only effect by oxygen, it would lead to quenching of the
cyclization and oxygenation of the additives.15b,c The following evi-
dences suggested that this was not a singlet oxygen reaction, how-
ever: (1) thus, under bona fide 1O2 conditions, ring-substituted
triphenylphosphines are known to give phosphinates along with
phosphine oxides,16 contrary to what observed in the present case
(see Table 1); (2) as for sulfides, sulfones are known to always
accompany the sulfoxides, at least in polar aprotic and non-polar
solvents, and benzyl sulfides do undergo both sulfoxidation and
oxidative cleavage to benzaldehyde in protic media, again differ-
ently from what observed under the present conditions (see
Table 2).17 An alternative mechanism is that the diradical character
of dihydrocarbazole 34 makes viable a chemical quenching of the
intermediate by molecular oxygen (path e in Scheme 3). Actually,
the interaction of 34 with ground state (triplet) oxygen can result
in a singlet, a triplet, or a quintet. Computing the energy of all of
the possible combinations demonstrated that the most favored
path led to a chemically bonded adduct with triplet multiplicity
that was formed through an almost thermoneutral process
In our hands, irradiation of a nitrogen-flushed 1 Â 10À2
M
acetonitrile solution of Ph3N led to N-phenylcarbazole (1) as the
main product (53%) along with two tetrahydro-derivatives that
were spectroscopically identified (see Scheme 2) as N-phenyl-
1,2,7,8-tetrahydro- (2, 14%) and N-phenyl-1,2,3,4-tetrahydrocar-
bazole (3, 5%), containing respectively a doubly conjugated pyrrole
and an indole aromatic system. Compounds 1–3 obviously arose
from a dihydrocarbazole intermediate (4) the role of which was
assessed by means of a DFT computational study at the (U)B3LYP
level of theory, by using the integral equation formalism model
(IEFPCM) for the solvent (acetonitrile). The Pople’s basis set
6-311++G(2d,p) was adopted for all atoms (see ESI for further
details). This essentially confirmed the AM1-SCI results, though
with some differences in the electronic distribution. Thus, singlet
14 had a purely zwitterionic character (a mixture of two predomi-
nant mesoionic structures, as indicated in Scheme 2) and lay a
mere 3.4 kcal molÀ1 below the triplet 34 (of diradical character,
satisfactory represented by the single formula in Scheme 2). As
mentioned, rearrangement to 34 and ISC to 14 mainly led to retro-
cyclization to triphenylamine (path d). No intramolecular con-
certed path leading to an aromatic system was available and
reasonably intermolecular hydrogen transfer took place leading
to radicals 5 and 6 and to the observed products 1–3 from them.
The corresponding experiment with an oxygen-equilibrated
solution gave 1 as the only isolated product, along with a small
(D
G° = 0.24 kcal molÀ1, through a transition state that was located
at 6.7 kcal molÀ1,see Scheme 4). More precisely, taking into
account both the regiochemistry and the stereochemistry of such
H
H
H
H
H
H
.
H
c
a
3Ph3N
+
.
N
.
N
Ph
.
6
N
Ph
5
Ph
34
b
H
H
H
H
H
H
CH-
hν
C-
N+
Ph
H
N+
Ph
N
Ph
H
H
H
3
14
N
Ph
H
H
d
H
N
Ph
H
2
1, NPC
H
H
Ph3N
Scheme 2. Cyclization of Ph3N under O2-free conditions.