Trans-1,2-Disiloxybenzocyclobutene, an adequate partner for the auto-oxidation: EPR/spin trapping and theoretical studies

Article abstract of DOI:10.1039/c3cp55077j

The auto-oxidation of trans-1,2-disiloxybenzocyclobutene 1 was found to be very efficient, giving endo-peroxide 7 in quantitative yield. Each step of the mechanism of spin-forbidden addition of triplet oxygen O₂( ³Σ_g) was

Full text of DOI:10.1039/c3cp55077j

PCCP
View Article Online
View Journal | View Issue
PAPER
trans-1,2-Disiloxybenzocyclobutene, an adequate
partner for the auto-oxidation: EPR/spin trapping
and theoretical studies†
Cite this: Phys.Chem.Chem.Phys.,
2014, 16, 7513
Jean Drujon,a Raphael Rahmani,a Virginie Heran,a Romain Blanc,a
´
¨
Yannick Carissan,*a Beatrice Tuccio,b Laurent Commeiras*a and Jean-Luc Parrain*a
´
Received 2nd December 2013,
Accepted 13th February 2014
The auto-oxidation of trans-1,2-disiloxybenzocyclobutene 1 was found to be very efficient, giving endo-
peroxide 7 in quantitative yield. Each step of the mechanism of spin-forbidden addition of triplet oxygen
O2(3Sg) was studied by both EPR/spin trapping and theoretical studies.
DOI: 10.1039/c3cp55077j
www.rsc.org/pccp
Introduction
For several years now, the preparation and the reactivity of
g-alkylidenebutenolides have been studied in our laboratory in
order to exploit their intrinsic features.1 Among the directions
taken into account, we have used these building blocks to
design fused polycyclic spirolactones2 in a step-economical
fashion. Because cycloaddition reactions are among the most
useful processes for generating molecular complexity, inter- and
intramolecular [4+2]-cycloaddition reactions between trans-1,2-
disiloxybenzocyclobutenes 1 and d-substituted g-alkylidenebute-
nolides 2 yielded, under very mild conditions, these spirolactone
moieties 3 or naphthofuranone moieties 4 and, at least, gave an
access to deoxylambertellol B 53 and the CDEF ring system of
lactonamycinone 6 (Scheme 1).4 During these studies, we observed
that trans-1,2-disiloxybenzocyclobutene 1 was very reactive towards
the dioxygen, even without any activation, giving the corresponding
Scheme 1 Synthesis of CDEF ring system of lactonamycinone.
endo-peroxide 7 which has been detected during both preparation one has described this interesting reactivity.6 Moreover, to the
of 1 (reduction of diketone 8 followed by protection of the hydroxyl best of our knowledge, only one example, using drastic conditions
function) and in subsequent [4+2]-cycloaddition reaction. Particu- (T = 190–240 1C), was reported for the trapping of dioxygen by
larly, we found that the use of oxygen free solvents under strictly benzocyclobutene.7 In our case, the reaction was performed at
inert Ar atmosphere was a crucial point to obtain good yields in the lower temperature (40 1C) and within 6 hours. Reaction times
preparation of 3 or 4. Without paying attention to this last point, up can also dramatically decrease using a saturated atmosphere of
to 40% of stable endo-peroxide 7 was for example observed in the dioxygen (Scheme 2).
crude 1H NMR (Scheme 1).
This oxidation process is the major source of material
In spite of the numerous uses of disilyloxybenzocyclobutene degradation and the most important problem of the shelf-life
1 or derivatives by several groups in Diels–Alder reaction,5 no of organic compounds such as polymers, lipids or living
organisms. . . Actually, two processes have been assigned for
a
´
Aix Marseille Universite, Centrale Marseille, CNRS, iSm2 UMR 7313, 13397,
Marseille, France. E-mail: yannick.carissan@univ-amu.fr,
laurent.commeiras@univ-amu.fr, jl.parain@univ-amu.fr; Fax: +33-491-289-187;
Tel: +33-491-289-126
b
´
Aix Marseille Universite, CNRS, ICR, UMR 7273, 13397 Marseille, France
† Electronic supplementary information (ESI) available: Synthesis procedure,
1H and 13C NMR spectra, computational details, and the UV spectrum of 1 at
400 nm. See DOI: 10.1039/c3cp55077j
Scheme 2 Reaction of 1 with dioxygen.
This journal is ©the Owner Societies 2014
Phys. Chem. Chem. Phys., 2014, 16, 7513--7520 | 7513

View Article Online
Paper
PCCP
this transformation. The first one, which is a rapid process, an oxygen containing biradical molecule, 10. This latter triplet
involved singlet oxygen O2(1Sg) generally obtained via a photo- molecule should undergo an ISC due to spin orbit coupling and
catalytic activation; the second one, slower than the first one, should lead to the singlet product 7.10 We also suggested the
involved triplet oxygen O2(3Sg) without any activation, often possibility of a light activated pathway, which implies an ISC
called auto-oxidation or ‘‘dark’’ oxidation. In some cases, and between excited states of 9.
with a particular type of olefins, the auto-oxidation can occur
rapidly. The ease of benzocyclobutene 1 for auto-oxidation to
give endo-peroxide 7 fits into this pattern and led us to study the
mechanism of this reaction by both EPR/spin trapping and
computational approach.
Results and discussion
1. Spin trapping experiments11
As a possible scenario (Scheme 3), the following mechanistic In a preliminary experiment and to confirm the utility of the spin
pathway could be proposed. First of all, the disrotatory opening trapping technique, we verified that no EPR signal could be
of 1 could lead to the formation of o-quinodimethane 9. The spin- observed from a solution of 1 (in the absence of spin traps 11
forbidden addition of triplet oxygen O2(3Sg) to 9 would furnish or 12) in benzene, with or without dioxygen, whatever the reaction
triplet 10 and finally the singlet ground state endo-peroxide 7. time. Interestingly the two spin traps used in our study exhibit
It is worthy of note that the change in multiplicity of biradical different features. Spin adducts obtained after addition of carbon-
10 from a triplet to a singlet is necessary to give 7 and should be centered radicals on a nitroso trap12 are usually stable enough to
proven by calculation.
be EPR-observed while nitroso-derived adducts obtained from
As described in Scheme 4, conventional electron paramagnetic oxygen-centered radicals are almost impossible to detect due to
resonance (EPR) spectroscopy associated with the spin-trapping thermal and photochemical instability.13,14
technique is the tool of choice to study fleeting radical intermedi-
In a first step, a solution containing MNP 11 (0.2 mol LÀ1
)
ates. In this context, the fast addition of a radical to a diamagnetic and 1 (0.2 mol LÀ1) was prepared in benzene. After ca. 2 h at
spin-trap (usually a nitrone or a nitroso compound) leads to a long- 40 1C, oxygen was removed by argon bubbling before recording
lived paramagnetic spin adduct (a nitroxide) that can be easily the EPR spectrum of the medium. Its simulation performed
detected by EPR spectroscopy8 and can give much information using the WinSim software15 revealed the presence of three
about the addend structure. So, the spin trapping using the different paramagnetic species. The first (aN = 1.54 mT) and
commercially available traps (2-methyl-2-nitrosopropane (MNP) 11 second (aN = 2.63 mT) signals were also observed in blank tests
and the 31P-labelled nitrone diethyl(2-methyl-1-oxido-3,4-dihydro- performed in the absence of 1. According to data previously
2H-pyrrol-2-yl)phosphonate (DEPMPO) 12), followed by analysis by published,16 the above-mentioned signals were, respectively,
EPR spectroscopy should confirm the transient existence of the assigned to radicals 13 and 14 (Scheme 4) that correspond to
triplet biradical 10(3Sg) formed during the auto-oxidation process. MNP decomposition products formed upon light exposure of
To complete the experimental studies of the reaction, we also the spin trap. The third detected nitroxide 15, which was never
performed a complete theoretical study of the reaction pathway. observed in the absence of either 1 or O2, showed a six line EPR
In particular, we focused on the inter system crossing (ISC) spectrum (aN = 1.44 mT and aH = 0.13 mT). The spin adduct 15
which explains why O2(3Sg) could react with a singlet molecule was thus formed after reaction of dioxygen with benzocyclo-
with no activation, as normally required.9 In this study, we have butene 1, and obviously resulted from the trapping of a radical
shown that 9 exists and could react with O2(3Sg), which leads to centered on a secondary carbon, as indicated by the presence of
a small hyperfine coupling with one hydrogen nucleus.
In a second step, phosphorylated nitrone DEPMPO 12 (reported
to yield persistent spin adducts with oxygen-centered radicals)
assisted spin trapping assays were conducted.17 As for MNP
experiments, solutions missing either 1 or dioxygen were found
to be EPR silent. Heating (40 1C) an oxygenated benzene solution
containing 1 (0.1 mol LÀ1) and DEPMPO 12 (0.1 mol LÀ1) for
Scheme 3 Proposal mechanism.
10 min gave rise to the EPR spectrum given in Fig. 1, recorded
after dioxygen removal.
A signal including 24 lines due to hyperfine coupling reac-
tions with nitrogen, phosphorus, b-hydrogen and g-hydrogen
nuclei (aN = 1.27 mT, aP = 3.84 mT, aHb = 0.73 mT and aHg
=
0.18 mT) was recorded. The first integration of this signal was
checked and did not lead to a bell-shaped curve characteristic
of the presence of a strong electron exchange coupling. In
addition, the EPR lines were not broadened by an electron
spin–electron spin dipolar interaction. This clearly shows that
the EPR spectrum recorded corresponded to a monoradical
nitroxide. Though the existence of a long distance coupling
Scheme 4 Spin trapping of a radical Yꢀ using (a) the nitroso compound
MNP 11 or (b) the nitrone DEPMPO 12, and structures of the nitroxide spin
adducts formed.
7514 | Phys. Chem. Chem. Phys., 2014, 16, 7513--7520
This journal is © the Owner Societies 2014

View Article Online
PCCP
Paper
Fig. 1 Experimental EPR signal of 16 (black full lines) obtained in benzene
after 10 min of reaction at 40 1C between dioxygen and 1 (0.1 mol LÀ1) in
the presence of DEPMPO 12 (0.1 mol LÀ1). Dioxygen was removed by
argon bubbling before recording the spectrum in order to observe
narrower lines. The simulation (red lines) led to the following parameters:
aN = 1.27 mT, aP = 3.84 mT, aHb = 0.73 mT and aHg = 0.18 mT.
Scheme 5 EPR/Spin trapping experiments.
carbon could react much faster with dioxygen, yielding a peroxyl
radical, than with DEPMPO. These kinetic reasons could explain
why no carbon-centered radical adduct of DEPMPO was EPR-
detected in our experiments. This study brings evidence that the
auto-oxidation reaction of 1 could likely follow a radical pathway
(Scheme 5). Although EPR alone is not sufficient to establish
precisely the mechanism involved, the presence of free radical
intermediates was undoubtedly proven. A MNP adduct of a
radical centered on a secondary carbon was EPR-observed, while
a peroxyl radical was detected using the nitrone DEPMPO 12.
Both adducts could be formed after trapping the triplet inter-
mediate 10, the biradical adducts initially formed probably
evolved to the monoradical nitroxide species detected.20
At this stage, though the spin trapping experiments were
consistent with the proposed mechanism, some points remain
unclear as for example the very high yield of the process. So in
order to acquire supplemental information, theoretical studies
were undertaken to confirm the proposal mechanism, the EPR/spin
trapping results and, more particularly, the change in multi-
plicity of biradical 10 from a triplet to a singlet ground state.
Each step is detailed in the following paragraphs.
with a g-hydrogen atom is rather uncommon, it is usually
observed in the EPR spectra of hydroperoxyl or alkylperoxyl
radical adducts of five-membered cyclic nitrones. After a compar-
ison of the results thus obtained with DEPMPO 12 with literature
data, a likely hypothesis is to consider that this species corre-
sponds to an alkylperoxyl radical adduct 16 of DEPMPO.17c,18 This
implies that, after the addition of the peroxy radical function of 10
to DEPMPO, the carbon-centered radical moiety rapidly evolved,
but we are not able to describe the exact structure of the mono-
radical thus formed since EPR spectroscopy can only bring
structural information in the vicinity of the paramagnetic center.
At this stage, one could envisage the possibility that the oxygen-
centered radical trapped by DEPMPO could result from the
decomposition of 7. To discard this hypothesis, control spin
trapping experiments were performed starting from endoperoxide
7. An oxygenated solution containing DEPMPO (0.1 mol LÀ1) and
7 (0.1 mol LÀ1) was prepared in benzene. An EPR spectrum
recorded after ca. 1 h (and after oxygen removal) showed no
signal. The same experiment was performed in the presence of
trace metal ions (added by letting the solution go through an
iron needle). Whether the reaction was performed in the
presence of oxygen or not, a 12-line EPR signal was system-
atically recorded. Comparing the hyperfine coupling constant
2. Theoretical study
values determined after its simulation (aN = 1.24 mT, aP
=
2.1. Computational details. The study of the reaction path-
4.70 mT, and aHb = 0.77 mT) to previously published data led to way was carried out by determining the stationary points and
assign the signal to a sterically hindered alkoxyl radical spin related energy differences. In order to reduce the computational
adduct. These results proved two points. (i) In the presence of time, TBS groups were modelled by TMS. The geometrical
metal traces, a ring opening could be observed in endoperoxide structure optimizations were performed using density functional
7, which yielded an alkoxyl radical; this reaction did not require theory (DFT) with the PBE021 functional. The def2-TZVP basis
the presence of dioxygen. (ii) The DEPMPO spin adduct set22,23 was used on all atoms. The integration grid used was m3
detected starting from 1 did not come from the decomposition as defined in the TURBOMOLE program package and the energy
of the oxidation product 7 eventually present in the medium.
threshold was set to 1dÀ6 atomic units. Unless otherwise speci-
All the experiments described were repeated twice, and fied, all DFT singlet densities (respectively triplet) were obtained
identical results were obtained. It is also worthy of mention in a restricted (respectively unrestricted) formalism. All minima
that, in general, the radical trapping by a nitrone is a kinetically on the energy potential surface were characterized by frequency
slow reaction, all the more so since the radical is sterically calculations, all of them being positive.
hindered, while the reactions performed with nitroso spin traps
All transition states exhibit one and only one negative
are usually more rapid.19 Thus, a radical centered on a secondary frequency, which is along the reaction coordinate. The excited
This journal is ©the Owner Societies 2014
Phys. Chem. Chem. Phys., 2014, 16, 7513--7520 | 7515

View Article Online
Paper
PCCP
state calculations along the opening of the C4 ring in 1 and for ring of TMS-1 is a kinetically relevant step of the reaction
the UV spectrum were done using the CC224 method with the (Fig. 7). The activation energy of this reaction, calculated as the
aug-cc-pVDZ basis sets25 and related auxiliary basis sets.26 energy difference between TS1–9 and TMS-1, is 26.8 kcal molÀ1
.
Along this opening, the molecule keeps a symmetry axis of The reaction energy i.e., the energy difference between TMS-9 and
order 2. Thus, all calculations along this pathway were done in TMS-1, is 10.6 kcal molÀ1. The C–C distances in the C6 ring (Fig. 2)
the C2 point symmetry group. Configuration Interaction with show a loss of aromaticity in TMS-9: two of them are short
Single and Double excitations (CISD)27,28 calculations were C3–C4 and C5–C6 (1.35 Å), and four of them are long, around
performed with the def2-TZVP basis set on PBE0/def2-TZVP 1.45 Å, and can be considered as single bonds (C–C bond
optimized geometries. Eight electrons were correlated. DFT distances are 1.54 Å in hexane). This explains the destabiliza-
energies contain the zero point energy corrections. This ZPE tion of TMS-9 with respect to TMS-1.
correction does not influence drastically the energy differences
At room temperature, as was already mentioned by
and could have been ignored. Thus, the CISD calculations were Danishefsky and Choy,5e,33 benzocyclobutene 1, exhibits a light
left uncorrected. DFT optimizations and CC2 calculations were yellow colour. It is worthy of note that the yellow colour
carried out using the TURBOMOLE 6.3 program package29 and disappears at low temperature. This thermochromism could
CISD calculations were performed by the use of the MOLPRO be explained by an inspection of the potential energy surfaces
program package.30 Basis sets for MOLPRO were obtained via of the ground and the first excited states of the molecule along
the EMSL website.31 Geometries are provided as ESI.† A scaling the reaction coordinate corresponding to the opening of the C4
factor of 0.959332 was used to compute thermodynamical ring (Fig. 3). At low temperature, the crossing of the barrier is
properties.
not possible and certainly only 1 is present in the medium. As
2.2. Disrotatory opening of the C4 ring: proof of existence can be seen in Fig. 3, the energy difference between the ground
of 9 by UV/visible analysis. The disrotatory opening of the C4 and the first excited states is roughly 5.19 eV.
Such a radiation is in the far ultraviolet (239 nm). When the
temperature is high enough to cross the barrier, TMS-9 is
formed and absorbs at around 400 nm in a broad range as
the UV/visible spectrum shows. This corresponds to violet. The
complementary colour of violet, yellow, is then observed.
We have performed the UV/visible spectrum of the reactant.
We want to determine whether 1 exists alone or in equilibrium
with 9. Furthermore, as the reaction with O2 is easy, we compare
the experimental UV/visible spectrum to the ones computed for
TMS-1, TMS-9 and TMS-7 (Fig. 4). In this spectrum, a shoulder at
330 nm and a weak absorption at 400 nm are visible. According
to calculations (Table 1 and Fig. 4), only TMS-9 is likely to
absorb at wavelengths larger than 300 nm at two wavelengths
compatible with experiment (324 nm for 330 nm and 394 nm
for 400 nm). Furthermore, we have detected phosphorescence
Fig. 2 Optimized geometries of 1 and 9. Indicated bond distances are in
Angstrom.
Fig. 3 Evolution of the CC2/aug-cc-pVDZ energy of the ground state 11A (solid line/cross symbols), the first triplet excited state 13B (dashed line/filled
circles) and the two first singlet excited states (solid lines/empty circles 21B and empty triangles 21A) along the breaking of the C1–C8 bond. Arrows in solid
line styles at 290 pm indicate the allowed electronic transitions at l1 = 324 nm and l2 = 394 nm which appear in the UV/visible spectrum and explain the
colour of the compound. Arrow in dashed line indicates the emission transition at l3 = 701 nm observed in the phosphorescence spectrum. Left picture: 1
at À20 1C. Right picture: 1 at r.t.
7516 | Phys. Chem. Chem. Phys., 2014, 16, 7513--7520
This journal is © the Owner Societies 2014

View Article Online
PCCP
Paper
Fig. 5 Emission spectrum of 1 at 312 nm. Emission at 350 nm is the
fluorescence of the state populated by the excitation. The band at 624 nm
is the harmonic of the 312 nm excitation. The phosphorescence emission
band at 700 nm corresponds to the l3 CC2/aug-cc-pVDZ value of 701 nm.
Intensity of the calculated wavelength is arbitrary.
Fig. 4 UV/visible spectrum of 1 at two different concentrations (bold line:
1 mol lÀ1, thin line; 0.33 mol lÀ1). CC2/aug-cc-pVDZ calculations allowed
transitions are reported as vertical lines with arbitrary intensities. The
calculated absorptions of TMS-1 and TMS-7 are in dashed lines. The
calculated absorptions of TMS-9 are in solid lines. The wavelengths
presented in Fig. 3 and 6 are reported (l1 = 324 nm, l2 = 394 nm).
Table 1 First excitation energies and related wavelengths for TMS-1,
TMS-7 and TMS-9
(2S+1)Symmetry
E (ev)
l (nm)
TMS-1
3B
4.34
5.02
5.19
6.10
6.27
6.34
286
247
239
203
198
195
3A
1A
1B
1B
1A
TMS-7
3B
4.35
4.97
5.17
5.82
6.18
6.22
285
250
240
213
201
199
3A
1A
1B
1A
Fig. 6 Explanation of the electronic spectra of TMS-9 calculated at the
CC2/aug-cc-pVDZ level of theory. Inter system crossing is likely to occur
between S2 and T2. Solid lines are used for singlet states and allowed
transitions (fluorescence). Dashed lines are used for triplet states and
forbidden transitions (phosphorescence).
1B
TMS-9
3B
1.77
3.14
3.82
3.84
4.22
4.33
701
394
324
323
294
286
1B
3A
1A
1A
(630–730 nm emission). This gives a range for l4 of 602–693 nm. It
should be emphasized that neither experimental nor theoretical
evaluations of l4 take into account the geometry relaxation. They
should then be considered as lower bound values. Thus, they
1B
in the range 630–730 nm with an optimal excitation wavelength of appear to be in very good agreement. The broad emission at
312 nm (Fig. 5). This can be understood by a detailed inspection of around 700 nm is the overlap of two transitions, the fluorescence
the spectrum of the excited states of TMS-9 (Fig. 6).34 The 312 nm from T2 to T1 and the phosphorescence from T1 to S0. We have
excitation corresponds to the S0–S2 transition. This transition is shown that 9 exists despite its disfavoured electronic structure
computed to be at 324 nm and is attributed to the shoulder at and the fact that it was not detected by spin trapping experiments.
330 nm in the UV spectrum. As S2 and T2 are quasi degenerated From this point we consider that it is the reactive species with O2.
3
(3.82 vs. 3.84 eV), an inter system crossing (ISC) can occur between
2.3. Reaction of 9 with SgO2. Two reaction pathways are
them. In a second step fluorescence occurs from T2 to T1. This possible: a chemical and a photochemical one. Both imply an
latter transition is computed to be at l4 = 603 nm and can be inter system crossing (ISC), i.e. a jump from the triplet to the
estimated experimentally by the energy difference between the singlet potential energy surface, which is forbidden. Along the
energy of the S2 state (at 330 nm) and the energy of the T1 state chemical pathway, ISC is likely to occur once O2 binds to 9.
This journal is ©the Owner Societies 2014
Phys. Chem. Chem. Phys., 2014, 16, 7513--7520 | 7517

View Article Online
Paper
PCCP
Fig. 8 CISD energy profile of the triplet and first two singlet states for the
endo-peroxide TMS-7 formation.
Because spin contamination is an issue with DFT and to
Fig. 7 Complete chemical reaction pathway. Energies relative to the
separated reactants TMS-1 and O2(3Sg) are in kcal molÀ1 at the PBE0/ make sure that the low singlet state energy is not an artefact,
def2-TZVP level of theory. Unrestricted and restricted Kohn–Sham form-
CISD/def2-TZVP//PBE0/def2-TZVP calculations were performed
along the geometrical deformation from TMS-10 to TMS-7
alism are compared because of spin contamination issues (see text).
(Fig. 8). These are free of spin contamination. On each geometry,
The photochemical pathway starts from the first excited triplet
state T1 of 9, which binds directly O2. In the following we
calculate the two pathways.
the ground state and the first singlet and triplet excited states were
computed. The first point of Fig. 8 confirms the quasi degeneracy of
the singlet and triplet states whereas the first excited singlet state is
38 kcal molÀ1 higher. These results are in agreement with the DFT
2.3.a. Chemical pathway. The activation energy for the results and are completely reliable in terms of spin contamination.
O2(3Sg) approach is calculated to be 4.0 kcal molÀ1 (Fig. 7). Along the C–O bond formation, the triplet is destabilized by Pauli
The exothermic reaction (9.0 kcal molÀ1 from TMS-9) leads to repulsion whereas the closed shell singlet state is stabilized until it
the biradical TMS-10 that is a stable triplet molecule. The becomes the ground state. For this process to occur, a small
reaction cannot lead to TMS-7 on the triplet energy surface. activation barrier, 4.8 kcal molÀ1, was found at the CISD level.
Yet, experiment shows undeniably that 7 is the product of the
At the DFT level of theory TMS-10 is not a minimum on the
reaction and that 10 exists in the medium as a biradical species. singlet energy surface: geometry optimizations converge to
To go from 10 to 7 the system must undergo an ISC.
the singlet endo-peroxide form TMS-7 (activation barrier =
To invoke such a phenomenon the necessary conditions are 0 kcal molÀ1 from TMS-10) when dioxygen is added. The
that the triplet and singlet states of 10 have similar and quasi- formation of TMS-7 in a singlet state is favoured energetically
degenerated geometries. We shall now show that these condi- (reaction energy = À51.8 kcal molÀ1).
tions are fulfilled. We expect the singlet and triplet states to be
The DFT results confirmed by the CISD calculations show
quasi degenerated, as the atoms that formally carry a radical that the triplet state of TMS-10 is quasi degenerated with the singlet
are spatially far from one another. On the optimized triplet TMS-10. Thus, if there is enough energy available to cross the
geometry of TMS-10, a singlet state density was computed in cyclobutene opening barrier (from 1 to 9) a chemical equilibrium
both unrestricted and restricted Kohn–Sham formalism, takes place between 1 and 10 as soon as dioxygen is available.
respectively, denoted UKS and RKS.35
Experimentally, the energy is brought into the system by heating up
The restricted calculation leads to a very high energy of at 40 1C. Thus, 10 exists as a triplet state. As we have shown, the
28.0 kcal molÀ1. This indicates that the RKS calculation converges singlet and triplet states of 10 are quasi degenerated with similar
to the first excited state of the molecule. The ground state could geometries. The conditions for the inter system crossing are fulfilled.
be modelled only via an unrestricted calculation, as the radical The system undergoes a formally forbidden triplet singlet transition,
separation cannot be modelled with the doubly occupied orbital. which leads irreversibly to 7 in an almost barrierless process.
The energy cost to flip one electron from alpha to beta is the
The rate constants for the unimolecular and bimolecular
difference in energy between the triplet and the UKS singlet. It reactions are calculated using the transition state theory.37 For
is found to be about 0.1 kcal molÀ1. This energy difference is the ring opening, from TMS-1 to TMS-9, a rate constant at
much lower than the precision of any DFT method. Our 25 1C, k1 is 4.2 Â 10À9 sÀ1 (k is found to be 1.6 Â 10À2
s
À1).
À1
conclusion is that these two states of TMS-10, in the geometry The equilibrium constant of this reaction is 0.3 Â 10À6. At
of the triplet can be considered as degenerate. The singlet wave 40 1C, the kinetic constant k1 is ten times larger than the kinetic
function is spin contaminated as expected.36
constant at 25 1C. The rate constant of the bimolecular reaction
7518 | Phys. Chem. Chem. Phys., 2014, 16, 7513--7520
This journal is © the Owner Societies 2014

View Article Online
PCCP
Paper
which leads from TMS-9 to TMS-10 is k2 = 1.3 Â 10À3
s
À1. The experiments. 1H NMR (300 MHz, C6D6) d 0.18 (6H, s, 2 Â CH3),
opposite unimolecular rate constant is kÀ2 = 5.5 Â 10À5 sÀ1. At 0.20 (6H, s, 2 Â CH3), 0.98 (18H, s, 6 Â CH3), 6.16 (2H, s, 2 Â
25 1C, the concentration of 9 is small yet non-zero. If O2(3Sg) is CH), 7.07–7.09 (2H, m, 2 Â CHAr), 7.27–7.29 (2H, m, 2 Â CHAr);
present in the medium, 10 is generated immediately which 13C NMR (75 MHz, C6D6) d À4.8 (2 Â CH3), À3.9 (2 Â CH3), 18.3
leads to the formation of 7. The formation of 9 occurs until the (2 Â C), 25 9 (6 Â CH3), 96.4 (2 Â CH), 125.4 (2 Â CHAr), 128.4
reactant is consumed.
(2 Â CHAr), 136.2 (2 Â CAr); HRMS m/z (ESI) calcd for
C
20H40NO4Si2 [M+NH4]+: 414.2490, found 414.2489.
2.3.b. Photochemical pathway. As can be seen in Fig. 6, in the S1
state the molecule is not likely to undergo an inter system crossing
because the triplet state T1 is energetically too far from S1. This is
supported by the emission spectrum with an excitation wavelength
of 400 nm in which only fluorescence is observed (emission
spectrum at 400 nm is provided as ESI†). Thus the excited electronic
state, which undergoes an ISC, is higher in energy. As was shown in
the analysis of the UV/visible spectrum of 9, the T1 state can be
populated after an excitation from the ground state S0 to S2, an ISC
from S2 to T2 (these two states being degenerated) and a fluores-
cence from T2 to T1. Even if unlikely, such ISC can occur in organic
molecules as shown in the evaluation of elements of the spin–orbit-
coupling matrix by Marian et al.38 and Russo et al.39 Once in the T1
state, 9 can react with O2(3Sg) and leads to 7 with no activation
energy. It is likely to be extremely fast, thus the only kinetically
relevant step is the opening of the C4 ring of 9.
2.4. Conclusion on the reaction between 1 and O2. We have
computed two reaction pathways for the reaction to lead to 7
from 1 and O2. Both pathways are likely to occur once the C4
ring of 1 opens to give 9. To distinguish the two postulated
mechanisms, the oxidation reaction of 1 was carried out in the
dark and inside an amberized NMR tube. After 5 hours at 40 1C
in saturated O2 C6D6 medium, the major product observed is
endo-peroxide 7 with a conversion rate of 75%. This experimental
evidence allows us to conclude that the chemical pathway is
certainly the one which occurred.
EPR/spin trapping experiments
MNP was purchased from Sigma-Aldrich while DEPMPO was
synthesized and purified according to methods previously
described.14,40 The spin adducts were produced by mixing 1
(0.1–0.2 mol LÀ1) and the spin trap (0.2 mol LÀ1 MNP or
0.1 mol LÀ1 DEPMPO) in benzene. The system was allowed to
react for 10 minÀ2 h, and the sample was deoxygenated by argon
bubbling before EPR analysis. The reaction medium was then
transferred into a glass pipette closed with a septum for EPR
analysis. Blank tests were also performed in the absence of O2, of
1 or of the spin trap. EPR spectra were recorded at room tempera-
ture (20–22 1C) on a continuous wave X-band Bruker EMX spectro-
meter. The following conditions were used: modulation frequency,
100 kHz; non saturating microwave power, 20 mW; modulation
amplitude, 0.1–0.15 mT; receiver gain, 105–106; time constant,
1.28–655 ms; scan time, 60–180 s. The hyperfine coupling
constants aX were determined after simulation of the experi-
mental EPR spectra using the WinSim2002 software.15,41
Notes and references
1 (a) S. Inack-Ngi, K. Cherry, V. Heran, L. Commeiras, J.-L.
Parrain, A. Duchene, M. Abarbri and J. Thibonnet, Chem.–Eur. J.,
2011, 17, 13692; (b) S. Inack-Ngi, R. Rahmani, L. Commeiras,
G. Chouraqui, J. Thibonnet, A. Duchene, M. Abarbri and
J.-L. Parrain, Adv. Synth. Catal., 2009, 351, 779; (c) A. Duchene,
J. Thibonnet, J.-L. Parrain, E. Anselmi and M. Abarbri, Synthesis,
2007, 597; (d) S. Rousset, M. Abarbri, J. Thibonnet, J.-L. Parrain
and A. Duchene, Tetrahedron Lett., 2003, 44, 7633; (e) S. Rousset,
J. Thibonnet, M. Abarbri, A. Duchene and J.-L. Parrain, Synlett,
2000, 260; ( f ) S. Rousset, M. Abarbri, J. Thibonnet, A. Duchene
and J.-L. Parrain, Org. Lett., 1999, 1, 701.
2 (a) A. Bartoli, F. Rodier, L. Commeiras, J.-L. Parrain and
G. Chouraqui, Nat. Prod. Rep., 2011, 28, 763; (b) A. Bartoli,
G. Chouraqui and J.-L. Parrain, Org. Lett., 2012, 14, 122;
(c) F. Rodier, M. Rajzmann, J.-L. Parrain, G. Chouraqui and
L. Commeiras, Chem.–Eur. J., 2013, 19, 2467.
Conclusions
This work elucidated for the first time each step of the mechanism
of spin-forbidden addition of triplet oxygen O2(3Sg) to trans-1,2-
disiloxybenzocyclobutene 1 to give the corresponding endo-peroxide
7 by using both EPR/spin trapping and theoretical studies. First of all
EPR/spin trapping proved that the auto-oxidation reaction of 1 could
likely follow a radical pathway and the presence of free radical
intermediates was undoubtedly proven. Theoretical studies have
confirmed EPR/Spin trapping experiments and allow us to under-
stand the formation of singlet state 7. The required inter-crossing
system for the forbidden reaction can take place through a chemical
pathway (vs. a photochemical pathway) and between 10 and 7.
3 R. Blanc, V. Heran, R. Rahmani, L. Commeiras and J.-L. Parrain,
Org. Biomol. Chem., 2010, 8, 5490.
4 S. Dubois, F. Rodier, R. Blanc, R. Rahmani, V. Heran,
J. Thibonnet, L. Commeiras and J.-L. Parrain, Org. Biomol.
Chem., 2012, 10, 4712.
Experimental section
Synthesis of compound 7
To a stirred solution of 1 (30 mg, 0.82 mmol) in C6D6 (1 mL) was
bubbled O2. The solution was then heated at 40 1C until the
disappearance of the starting material (followed by 1H NMR,
3 hours). Because of the instability of 7 during the purification
step on silica gel, the crude reaction in C6D6 was used for NMR
5 (a) K. Yamamoto, M. F. Hentemann, J. G. Allen and
S. J. Danishefsky, Chem.–Eur. J., 2003, 9, 3242; (b) J. Barluenga,
F. Aznar and M. A. Palomero, Chem.–Eur. J., 2002, 8, 4149;
(c) J. Barluenga, M. A. Fernandez-Rodriguez and E. Aguilar,
Org. Lett., 2002, 4, 3659; (d) J. G. Allen and S. J. Danishefsky,
This journal is ©the Owner Societies 2014
Phys. Chem. Chem. Phys., 2014, 16, 7513--7520 | 7519

View Article Online
Paper
J. Am. Chem. Soc., 2001, 123, 351; (e) M. F. Hentemann, J. G. Allen
PCCP
¨
K. Hald, J. Chem. Phys., 2002, 116, 5401; ( f ) C. Hattig,
¨
and S. J. Danishefsky, Angew. Chem., Int. Ed., 2000, 39, 1937;
( f ) J. G. Allen, M. F. Hentemann and S. J. Danishefsky, J. Am.
Chem. Soc., 2000, 122, 571.
6 All supporting information (ref. 5) neither mentioned the
use of degassed solvent.
A. Hellweg and A. Kohn, Phys. Chem. Chem. Phys., 2006,
8, 1159.
25 (a) T. H. Dunning Jr., J. Chem. Phys., 1989, 90, 1007–1023;
(b) D. E. Woon and T. H. Dunning, Jr., J. Chem. Phys., 1993,
98, 1358.
¨
¨
7 W. R. Roth, T. Ebbrecht and A. Beitat, Chem. Ber., 1988, 26 F. Weigend, A. Kohn and C. Hattig, J. Chem. Phys., 2001,
121, 1357.
116, 3175.
8 C. Evans, Aldrichimica Acta, 1979, 12, 23.
9 B. F. Minaev, N. A. Murugan and H. Ågren, Int. J. Quantum
Chem., 2013, 113, 1847–1867.
27 P. J. Knowles and H. J. Werner, Chem. Phys. Lett., 1988,
145, 514.
28 H. J. Werner and P. J. Knowles, J. Chem. Phys., 1988,
89, 5803.
10 (a) B. F. Minaev and H. Ågren, THEOCHEM, 1998, 434,
193–206; (b) B. F. Minaev and H. Ågren, EPA Newsl., 1999, 29 TURBOMOLE V6.3 2011, a development of University of
65, 7–37.
11 See ESI† for EPR spectrum.
12 Nitroso spin traps have an important advantage over
Karlsruhe and Forschungszentrum Karlsruhe GmbH,
1989–2007, TURBOMOLE GmbH, since 2007; available from
http://www.turbomole.com.
nitrones in that the addend is directly bound to the nitro- 30 MOLPRO, version 2008.1, a package of ab initio programs,
gen, thereby yielding additional hyperfine couplings. There-
fore, they are the traps of choice to differentiate radicals
centered on a primary, a secondary or a tertiary carbon.
13 M. Perkins and B. Roberts, J. Chem. Soc., Perkin Trans. 1,
1974, 297.
H.-J. Werner, P. J. Knowles, R. Lindh, F. R. Manby,
M. Schutz et al., see http://www.molpro.net.
31 K. L. Schuchardt, B. T. Didier, T. Elsethagen, L. S. Sun,
V. Gurumoorthi, J. Chase, J. Li and T. L. Windus, J. Chem.
Inf. Model., 2007, 47, 1045.
14 E. Janzen, D. Lindsay and D. Haire, J. Am. Chem. Soc., 1990, 32 Y. Tantirungrotechai, K. Phanasant, S. Roddecha,
112, 8279.
P. Surawatanawong, V. Sutthikhum and J. Limtrakul,
15 The free WinSim program is available at the following
THEOCHEM, 2006, 760, 189.
address: www.niehs.nih.gov/research/resources/software/ 33 (a) W. Choy and H. Yang, J. Org. Chem., 1988, 53, 5796;
tox-pharm/tools/index.cfm.
16 (a) T. Wajer, T. Mackor, T. de Boer and J. van Voorst,
(b) M. F. Hentemann, J. G. Allen and S. J. Danishefsky,
Angew. Chem., Int. Ed., 2000, 39, 1937.
Tetrahedron, 1967, 23, 4021; (b) K. Makino, N. Suzuki, 34 Electronic states are labelled using both spectroscopic
F. Moriya, S. Rokushida and H. Hatano, Anal. Lett., 1980,
13, 311; (c) G. R. Buettner, Free Radical Biol. Med., 1987, 3, 259.
(2S+1)G or Jablonski notation (S: singlet, T: triplet, index: 0
for ground state and larger or equal to 1 for excited states).
As TMS-9 has a C2 symmetry values A and B are possible for
the irreducible representations G according to the symmetry
of the state.
´
17 (a) C. Frejaville, H. Karoui, B. Tuccio, F. Le Moigne,
M. Culcasi, S. Pietri, R. Lauricella and P. Tordo, J. Med. Chem.,
´
1995, 38, 258; (b) B. Tuccio, R. Lauricella, C. Frejaville,
J.-C. Bouteiller and P. Tordo, J. Chem. Soc., Perkin Trans. 2, 35 IUPAC, Compendium of Chemical Terminology, compiled by
1995, 295; (c) S. Dikalov, P. Tordo, A. Motten and R. Mason,
Free Radical Biol. Med., 2003, 37, 705.
A. D. McNaught and A. Wilkinson, Blackwell Scientific
Publications, Oxford, 2nd edn (the ‘‘Gold Book’’), 1997,
XML on-line corrected version: http://goldbook.iupac.org
(2006-) created by M. Nic, J. Jirat, B. Kosata; updates
compiled by A. Jenkins, ISBN 0-9678550-9-8, DOI: 10.1351/
goldbook.
´
18 (a) S. Barbati, J.-L. Clement, G. Olive, V. Roubaud, B. Tuccio
and P. Tordo, in NATO ASI Series, Serie A* Life Sciences, ed.
F. Minisci and Kluwer Academic Publishers, Dordrecht,
1997, vol. 27, p. 39; (b) H. Karoui, F. Chalier, J.-P. Finet
and P. Tordo, Org. Biomol. Chem., 2011, 9, 2473.
36 J. A. Pople, P. M. W. Gill and N. C. Handy, Int. J. Quantum
Chem., 1995, 56, 303.
19 T. K. Kemp, Prog. React. Kinet. Mech., 1999, 24, 287.
´
`
20 M. Triquigneaux, L. Charles, C. Andre-Barres and B. Tuccio, 37 J. I. Steinfeld, J. S. Francisco and W. L. Hase, Chemical
Org. Biomol. Chem., 2010, 8, 1361.
Kinetics Dynamics, Prentice Hall, Upper Saddle River, New
21 C. Adamo and V. J. Barone, Chem. Phys., 1999, 110, 6158.
Jersey, 1998.
22 F. Weigend, M. Haser, H. Patzelt and R. Ahlrichs, Chem. 38 J. Tachen, H. Gilka and C. M. Marian, Phys. Chem. Chem.
Phys. Lett., 1998, 294, 143. Phys., 2007, 9, 5209–5221.
23 F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 39 (a) A. D. Quartarrolo, S. G. Chiodo and N. Russo, J. Chem.
7, 3297.
Theory Comput., 2010, 6, 3176–3189; (b) A. D. Quartarrolo,
S. G. Chiodo and N. Russo, J. Comput. Chem., 2012, 33,
1091–1100.
24 (a) O. Christiansen, H. Koch and P. Jørgensen, Chem. Phys.
¨
Lett., 1995, 243, 409; (b) C. Hattig and F. Weigend, J. Chem.
¨
¨
´
´
Phys., 2000, 113, 5154; (c) C. Hattig and A. Kohn, J. Chem. 40 S. Barbati, J.-L. Clement, C. Frejaville, J.-C. Bouteiller,
¨
Phys., 2002, 117, 6939; (d) C. Hattig and K. Hald, Phys.
P. Tordo and J.-C. Yadan, Synthesis, 1999, 2036.
¨
¨
Chem. Chem. Phys., 2002, 4, 2111; (e) C. Hattig, A. Kohn and 41 D. R. Duling, J. Magn. Reson., Ser. B, 1994, 104, 105.
7520 | Phys. Chem. Chem. Phys., 2014, 16, 7513--7520
This journal is © the Owner Societies 2014