10.1002/anie.201703334
Angewandte Chemie International Edition
COMMUNICATION
transfer process should be considered as an alternative
mechanism.[13] Considering that we did not observe any overlap
in the emission of the photocatalysts 3b (λmax= 453 nm) and the
absorption of the nitrocyclopropane 1a (λmax= 229 nm) (see S.I.),
the energy transfer process from 3b* to the substrate should
proceed through a Dexter type mechanism (instead of a Forster
pathway). Such mechanism implies that triplet spin state in
substrate is induced by short distance interaction with triplet
excited state of photocatalyst. Indeed, the calculated values of
singlet-triplet gap in 1a and the emission energy of photocatalyst
3a are in good agreement (see below). In a quest for evidence of
an energy transfer mechanism, the cyclopropane 8 was irradiated
with UV light (λmax= 365 nm), observing the expansion reaction
with a moderate conversion (23%) in 24h. Although a value of the
quantum yield lower than one does not exclude a possible radical
chain mechanism, we have measured the Φ of the reaction of 1a
in the presence of the photocatalyst 3b (see S.I.). The observed
value (Φ= 0.05) could suggest that a radical chain propagation is
not taking place.
enables the final observed 5-membered ring isomer (2a) whereas
that from 3IIb would lead to the 5-membered ring 1IVb that is not
observed experimentally. Finally, 3IIc is an unproductive pathway
due to the disposition of the aldehyde group. Product formation
through pathways a and b requires the conformational change
from 3II to another biradical 3III in order to properly orient the
reacting groups before ring closure occurs. This ring closure
implies a spin crossing from the triplet to the singlet state that can
occur either before or after II III interconversion. In order to
understand the reaction mechanism, we explored the reactivity of
1a as a representative example (for cyclopentene 5a, see S.I.),
considering both the triplet and open shell singlet states for all
biradical intermediates.
3
3
Figure 2 shows the Gibbs energy profiles for pathways a, b
and c of 1a. The three possible C-C cleavages are kinetically and
thermodynamically favored. Energy barriers are low (9.3-11.5
kcal mol-1) and the reaction G ranges from –15 to –30 kcal mol -
3
1. Noteworthy, the most stable biradical corresponds to IIb. In
these species, the radical character lies at the two carbons that
are able to better stabilize the unpaired electrons: the tertiary
carbon presenting the stabilizing CHO group and the benzyl
carbon (See S.I. for details). However, this cleavage pathway
would lead to a ring expansion product 1IVb that is not
experimentally observed (right-Figure 2), thereby indicating that
other factors determine the observed reactivity. In particular, the
evolution of intermediates 3IIa and 3IIb differ significantly. That is,
although for both intermediates the Gibbs energy barrier for the
biradical rearrangement (3II 3III)is low, this process is exergonic
for path a and endergonic for path b. This suggests that the
Considering that changes in R1, R2 and CHO/CH=EWG did
not suppress the observed reactivity, we hypothesized that the
nitro group must play a key role in the photocatalytic activation of
cyclopropane. Indeed, when the nitro group is substituted by an
ester or a ketone (10 and 11, see Figure 1), the reaction under
visible light photocatalytic conditions did not take place, indicating
that the presence of the nitro group is essential for the ring-
expansion process. With the aim of gaining a deeper insight into
this point, the singlet-triplet Gibbs energy gap was calculated for
1a and other related structures where nitro group was substituted
by different functional groups (Figure 1). Interestingly, the lowest
singlet-triplet gaps are computed for molecules enclosing the nitro
group. In fact, such energetic difference between the spin states
of the organic reagent, reasonably matches the reported emission
energy for photocatalyst 3b (~60 kcal/mol).[12] Certainly, when the
nitro group is substituted by COPh or two esters, the gap
substantially increases (>70 kcal/mol), hindering the
photocatalytic process, which is in agreement with the absence of
reactivity of 10 and 11. The two SOMO orbitals of the triplet state
are localized on the NO2 group, in agreement with the role of such
a fragment on lowering the excitation energy of the reagent.
3
3
intermediate IIIa would be a more long-lived species than IIa
while the opposite is expected for pathway b. Moreover, open-
shell singlet state optimizations of intermediates 1IIa and 1IIIa yield
to two minima with very similar geometries and energetics to
those of the triplet states.1IIaevolves to reactants in an essentially
barrierless process, whereas 1IIIa leads to the final product
observed, after overcoming a low energy barrier of 3.6 kcal mol-1.
This suggests that spin crossing probably occurs after biradical
rearrangement; that is, at geometries close to 3IIIa. On the other
hand, all attempts to locate open-shell singlet state structures for
IIb and IIIb collapsed to reactants, suggesting that spin crossing
always leads to the initial reactants.[14] Thus, spin crossing in
pathway b appears to be unproductive, which is in agreement with
1
the non-observation of IVb as resulting product[15] and the low
80
75
70
65
60
value of the quantum yield.
The calculations suggest that the key point is that the
biradical intermediates of pathway a can evolve to products,
whereas those of pathway b can only return to reactants. This
seems to originate from three key points: i) The biradical
rearrangement is only exergonic for pathway a. This favors the
species that leads to the final product and suggests that it is more
likely that spin crossing occurs after biradical rearrangement. ii)
There is almost no overlap between the carbon p orbitals in both
species (IIa and IIIa) involved in the formation of dihydrofurane in
pathway a, in contrast to the overlap observed in IIb. Furthermore,
the S2 expectation value obtained for open shell singlets is around
1. Indeed, this is observed for IIa and IIIa, whereas in the case of
IIb the calculated value is 0.86 (Figure 3). Consequently, the latter
intermediate exhibits a smaller biradical character. Therefore,
after spin-crossing in pathway b the reactant is easily recovered
55
50
O
CO2Me
O
O
O
H
H
iPr
H
CO2Et
H
iPr
Ph
H
H
H
H
Ph
Ph
COPh
Ph
NO2
CO2Et
Ph
NO2
NO2
1a
1g
4a
10
11
Figure 1. SOMO orbitals calculated for triplet state of 1a (left) and singlet-triplet
energy gap in a range of cyclopropane reagents (right).
The first step is the opening of the cyclopropane ring that
leads to the formation of a first biradical 3IIa (right-Figure 2). Three
C-C bond breaks are possible (paths a, b or c), which give three
different isomers (3IIa, 3IIb and 3IIc). Ring expansion from 3IIa
This article is protected by copyright. All rights reserved.
Luis-Barrera, Javier
